Universidad Autónoma de Madrid Facultad de ciencias

Departamento de Biología Molecular

BASE MOLECULAR DE LA ESPECIFICIDAD PEPTÍDICA DE

HLA-B27

JOSÉ RAMÓN LAMAS LÓPEZ Madrid, 1999

Universidad Autónoma de Madrid Facultad de ciencias

Departamento de Biología Molecular

Base molecular de la especificidad peptídica de HLA-B27

Memoria para optar al grado de Doctor en Ciencias presentada por: José Ramón Lamas López Director: Dr. José Antonio López de Castro Álvarez Profesor de investigación del C.S.I.C Centro de Biología Molecular "Severo Ochoa"

Madrid, Junio de 1999.

ABREVIATURAS

APC Células presentadoras de antígeno.

AR Artritis reactiva

Aua Ácido 11-amino undecanoico.

EA Espondilitis anquilosante.

EBNA Antígeno nuclear del EBV. EBV Virus de Epstein-Barr.

β2m β2-microglobulina.

BCR Receptor de células B.

CD Cluster of Differentiation.

C-terminal Extremo Carboxilo.

CTL Linfocito T citotóxico.

HB (R)-3-Hidroxibutirato. HLA Antígenos Leucocitarios Humanos.

Kb Kilobase.

kDa Kilodalton.

LMP Proteína latente de membrana del EBV

MHC Complejo principal de histocompatibilidad.

N-terminal Extremo amino.

PX Posición X del péptido.

PΩ Posición C-terminal.

RE Retículo Endoplásmico.

TAP Transportador asociado con la presentación de antígeno.

TCR Receptor de célula T.

+H3N

COO-

C

H

H

GLICINA(Gly)G

+H3N

COO-

C

CH2

H

C

NH2O

ASPARAGINA(Asn)N

+H3N

COO-

C

CH2

H

CH2

C

NH2O

GLUTAMINA(Gln)Q

S

CH3

+H3N

COO-

C

CH2

H

CH2

METIONINA(Met)M

SH

+H3N

COO-

C

CH2

H

CISTEÍNA(Cys)C

+H3N

COO-

C

C

H

CH3

CH2

CH3H

ISOLEUCINA(Ile)I

+H3N

COO-

C

CH2

H

CH3H3C

CH

LEUCINA(Leu)L

+H3N

COO-

C

CH

H

CH3H3C

VALINA(Val)V

+H3N

COO-

C

CH3

H

ALANINA(Ala)A

COO-

C H

CH2

CH2H2C

PROLINA(Pro)P

Aminoácidos apolares alifáticos

Aminoácidos polares (sin carga)

OH

+H3N

COO-

C

C

H

CH3

H

TREONINA(Thr)T

OH

+H3N

COO-

C

C

H

H

H

SERINA(Ser)S

OH

+H3N

COO-

C

CH2

H

TIROSINA(Tyr)Y

+H3N

COO-

C

CH2

H

FENILALANINA(Phe)F

+H3N

COO-

C

CH2

H

NH

TRIPTÓFANO(Trp)W

Aminoácidos apolares aromáticos

Aminoácidos polares (con carga)

+H3N

COO-

C

CH2

H

CH2

C

O-O

Ác. GLUTÁMICO(Glu)E

+H3N

COO-

C

CH2

H

C

O-O

Ác. ASPÁRTICO(Asp)D

+H3N+H3N

+H3N H

COO-

C

CH2

CH2

CH2

NH

C=NH

NH2

2+

ARGININA(Arg)R

COO-

C

CH2

H

CH2

CH2

NH 3+

LISINA(Lys)K

CH2

NH

COO-

C

CH2

H

+HN

HISTIDINA(His)H

+H2N

ÍNDICE -i-

I. INTRODUCCIÓN. ...................................................................................................................................................................................... 1

I.1. Características generales del sistema inmunitario. .................................................................................................... 3

I.2. Organización genómica y estructural del MHC humano. ...................................................................................... 4

I.2.1. Organización de los genes del MHC de clase I y clase II. ...................................................................... 4

I.3. Estructura de las proteínas de clase I. ............................................................................................................................. 6

I.3.1. La cadena pesada y la Beta-2 microglobulina (β2m). . .............................................................................. 6

I.3.2. Los péptidos. ................................................................................................................................................................ 8

I.4. Procesamiento de antígeno. Formación, transporte y presentación de los complejos péptido-MHC de clase I. .......................................................................................................................................................... 9

I.4.1. Origen y procesamiento de los péptidos antigénicos. ................................................................................ 9

I.4.2. Translocación al Reticulo Endoplásmico. ...................................................................................................... 10

I.4.3. Biosíntesis, Ensamblaje y expresión en membrana de las moléculas de clase I. ........................... 11

I.5. TCR y reconocimiento de los complejos MHC-péptido. ........................................................................................... 11

I.5.1. Características estructurales del TCR y reconocimiento de los complejos MHC-péptido. ....... 11

I.6. HLA-B27. ...................................................................................................................................................................................... 13

I.6.1. Subcavidad A. ............................................................................................................................................................. 13

I.6.2. Subcavidad B. ............................................................................................................................................................. 13

I.6.3. Subcavidades C y F. ................................................................................................................................................ 14

I.6.4. Subcavidades D y E. ................................................................................................................................................ 15

I.7. HLA-B27 y espondiloartropatías. ....................................................................................................................................... 15

I.7.1. Posible papel patogénico de la presentación de péptidos por HLA-B27. ......................................... 16

I.7.2. Distribución étnica y asociación a enfermedad de los subtipos de HLA-B27. ................................ 17

II. OBJETIVOS. ............................................................................................................................................................................................... 21

III. MATERIALES Y MÉTODOS. .......................................................................................................................................................... 25

III.1. Líneas celulares. ..................................................................................................................................................................................27

III.2. Anticuerpos monoclonales (mAb). ............................................................................................................................................27

III.3. Síntesis, purificación y cuantificación de péptidos. ..........................................................................................................27

II.3.1. Síntesis y purificación de péptidos y análogos no peptídicos ........................................................................27

III.3.2. Cuantificación. ...................................................................................................................................................................28

III.4. Ensayo de unión de péptidos. .......................................................................................................................................................28

III.4.1. Análisis por citometría de flujo. ................................................................................................................................29

III.4.2. Cálculo de la unión de péptidos. ...............................................................................................................................29

IV. RESULTADOS. ........................................................................................................................................................................................ 31

IV.1 Efecto del polimorfismo de HLA-B27 sobre la especificidad de unión de péptidos. . .......................................33

IV.1.1. Unión de péptidos a HLA-B*2705, B*2704 y B*2706. Modulación de la especificidad por el residuo peptídico C-terminal. ........................................................................................................................33

IV.1.1.1. Efectos de la pérdidas y ganancias de residuos cargados en las subcavidades E/C/F. ..................................................................................................................................34

ÍNDICE -ii-

IV.1.2. Unión de péptidos a B*2701 y B*2702. Efecto del polimorfismo sobre la especicificidad de los residuos en P2. ........................................................................................................ 37

IV.1.3. Unión de péptidos a B*2703:Papel del polimorfismo de la subcavidad A. ........................................40

IV.1.3.1. Propiedades dinámicas vs afinidad de unión de péptidos. ................................................. 40

IV.1.3.2. Fluctuaciones atómicas, áreas accesibles y no accesibles. ............................................... 41

IV.1.3.3. Analisis cualitativo y cuantitativo de los puentes de hidrógeno. ..................................... 42

IV.2. Relación entre la unión de péptidos y la selección de epítopos virales por células T. ............................. 43

IV.2.1. Unión de péptidos virales a diferentes subtipos deHLA-B27, y su relación con la inmunogenicidad. ....................................................................................................................... 43

IV.2.2. Reconocimiento de péptidos de EBV por CTLs restringidos por HLA-B27. .................. 44

IV.2.3. El motivo Arg2, anclaje principal de los péptidos unidos a HLA-B27, no es esencial para mantener la capacidad antigénica del péptido. .............................................. 44

IV.3. Modulación de la especificidad en las posiciones de anclaje P1, P3 y PΩ, por el polimorfismo de HLA-B27. ............................................................................................................................................. 47

IV.3.1. Especificidad de B*2705, B*2704 y B*2706 por los residuos en P1, P3 y P9. .............. 47

IV.3.2. La unión de un péptido es el resultado de la contribución aditiva de varios residuos de anclaje. .................................................................................................................................. 48

IV.3.3. Distribución de los residuos P1, P3 y PΩ. entre los ligandos naturales de B*2705. . 52

IV.4. Unión de análogos no peptídicos a HLA-B27. .......................................................................................................... 55

IV.4.1. Reemplazamiento de la parte central de epítopos naturales con un espaciador monofuncional. .............................................................................................................. 55

IV.4.2. Reemplazamiento de la parte central de epítopos naturales con espaciadores bifuncionales. ............................................................................................................. 56

IV.4.3. Modelado molecular de la unión de análogos con espaciadores no peptídicos a B*2705. ........................................................................................................................... 56

V. DISCUSIÓN. ................................................................................................................................................................................................. 59

V.1. Polimorfismo de HLA-B27 y solapamiento de repertorios peptídicos entre subtipos. .............................. 61

V.1.1. Influencia del polimorfismo de la cavidad C/F sobre la especificidad por el extremo C-terminal del péptido. Diferencias entre B*2704 y B*2706. ............................. 61

V.1.2. El polimorfismo de la cavidad C/F determina parcialmente la especificidad de la subcavidad B. . ................................................................................................................................ 63

V.1.3. Análisis del efecto del polimorfismo de la subcavidad A sobre la unión de péptidos. ........................................................................................................................................................ 65

V.2. Relación entre la unión de péptidos y la selección de epítopos por células T. .............................................. 67

V.3. La especificidad de péptidos por los subtipos de HLA-B27 es modulada en múltiples posiciones de anclaje. ............................................................................................................................................................ 70

V.4. Unión de análogos no peptídicos a HLA-B27. ............................................................................................................ 72

Resumen y discusión general. ...................................................................................................................................................... 73 VI. CONCLUSIONES. .................................................................................................................................................................................. 75

VII. REFERENCIAS. .................................................................................................................................................................................... 79

VIII. ANEXOS. ................................................................................................................................................................................................. 91

-1-

I. INTRODUCCIÓN

INTRODUCCIÓN -3-

I. INTRODUCCIÓN I.1. CARACTERÍSTICAS GENERALES DEL SISTEMA INMUNITARIO.

l sistema inmunitario es el encargado de proteger a un organismo discriminando entre

lo que le pertenece y lo extraño a él. Cuando esta función falla y lo propio no es

reconocido como tal, se desencadena una respuesta autoinmune.

Funcionalmente, se distinguen dos mecanismos inmunitarios perfectamente coordinados.

El sistema inmune innato o no adaptativo, genera respuestas rápidas e inespecíficas dirigidas

hacia características generales del agente patógeno, creando una primera barrera defensiva tanto

a nivel físico y bioquímico, a cargo de la piel, mucosas, secreciones y diferentes factores

solubles, como celular por células fagocíticas (polimorfonucleares neutrófilos, monocitos y

macrófagos), células citotóxicas NK (Natural Killer) o células secretoras (eosinófilos, basófilos

y mastocitos).

E

El sistema inmune específico o adaptativo a diferencia del anterior, está adaptado a la

naturaleza del patógeno, a su estrategia invasora y posee memoria, lo que le permite generar una

respuesta más potente y rápida ante una reexposición al antígeno que provocó la respuesta

inicial.

Los mecanismos efectores de la respuesta adaptativa dependen de receptores expresados en

los linfocitos B y T que reconocen antígenos de diferentes características. Los receptores de los

linfocitos B o BCR (B Cell Receptor) reconocen el antígeno en su estado nativo, activando la

diferenciación y maduración de los linfocitos B para secretar las inmunoglobulinas (Ig) o

anticuerpos, una versión soluble del receptor que mantiene su especificidad original.

Los receptores de los linfocitos T o TCR (T Cell Receptor), reconocen fragmentos

peptídicos derivados del procesamiento del antígeno, unidos a moléculas del MHC (Major

Histocompatibility Complex). Para ello requieren la presencia de células presentadoras

encargadas de procesar el antígeno y de presentarlo en la superficie celular para su

reconocimiento por los receptores antigénicos de los linfocitos T. Basándose en su estructura y

función se distinguen dos tipos de moléculas del MHC: de clase I y de clase II.

INTRODUCCIÓN -4-

Los linfocitos T se distribuyen en dos subpoblaciones dependiendo del tipo de proteína

correceptora, CD4 o CD8, que expresan en su membrana. Estos correceptores en general no se

expresan juntos en un mismo linfocito T diferenciado, y la presencia de uno u otro determina

diferentes patrones de restricción por MHC (Zinkernagel y Doherty, 1974). Los linfocitos T

citotóxicos (CD8+) o CTLs (Citotoxic T Lymphocytes), son células efectoras líticas, que

reconocen el péptido presentado por moléculas del MHC de clase I, se dice que están

restringidas por el MHC de clase I. Las células Th (helper) o reguladoras (CD4+), están

restringidas por el MHC de clase II. El reconocimiento de péptidos unidos a estas moléculas

inicia una serie de procesos esenciales para desencadenar una respuesta inmune, puesto que de

ello depende tanto la diferenciación de los linfocitos B en células plasmáticas secretoras de

anticuerpos, como la activación y diferenciación de los linfocitos T CD8+ en células efectoras.

I.2. ORGANIZACIÓN GENÓMICA Y ESTRUCTURAL DEL MHC HUMANO.

Las moléculas del MHC, denominadas en humanos HLA (Human Leukocyte Antigens),

son codificadas por un conjunto de genes agrupados en una zona de aproximadamente 3.500 Kb,

en el brazo corto del cromosoma 6 (Francke y Pellegrino, 1977; Dunham et al., 1987) donde se

diferencian tres regiones: de clase I, II y III.

I.2.1. Organización de los genes del MHC de clase I y clase II.

La región de clase I abarca una extensión de 1600 Kb orientada hacia el extremo

telomérico, donde se emplazan los loci codificantes de las cadenas α o cadenas pesadas de los

antígenos de clase Ia o clásicos (HLA-A, -B y -C) y los de clase Ib o no clásicos (HLA-E, F, G)

(Figura 1).

Los genes que codifican las cadenas pesadas de los antígenos de clase Ia, son muy

polimórficos (Parham et al., 1995) y se expresan en la mayoría de las células somáticas de forma

codominante y a niveles de expresión variables, aunque típicamente elevados en las células de

linaje hematopoyético. Por el contrario, los genes codificantes de las moléculas de clase Ib, son

poco polimórficos y su expresión tisular está más limitada (Wei y Orr, 1990). Estructuralmente

son semejantes a los antígenos de clase I clásicos pero reconocen antígenos de diferente

naturaleza. Entre estos antígenos de clase Ib, se incluyen además, las moléculas CD1 (Melián et

al., 1996) aunque son codificadas por el cromosoma 1, fuera del MHC. Recientemente se ha

identificado una familia de cinco genes ubicados en las regiones de clase I y III denominada

-5- INTRODUCCIÓN

MIC (MHC class I Chain related) (Bahram et al., 1994), que codifican cadenas pesadas con

secuencias muy divergentes de las otras cadenas de clase I codificadas en el MHC.

Orientada hacia el extremo centromérico, la región de clase II tiene una amplitud de 900

Kb, donde se ubican los loci HLA-DP, -DQ y -DR que codifican a las cadenas α y β de las

moléculas de clase II. Estas moléculas se expresan exclusivamente en macrófagos, células

dendríticas y células B activadas, conocidas como células presentadoras de antígeno

“profesionales” o APCs (Antigen-Presenting Cells). En esta región se localizan además otros

genes que codifican a varias proteínas implicadas en el procesamiento, transporte y presentación

de antígeno como TAP1 y TAP2 (Spies et al., 1990; Trowsdale et al., 1990) la Tapasina

(Herberg et al., 1998) y las subunidades del proteosoma LMP2 y LMP7 (Large Multifunctional

Proteasome). (Kelly et al., 1991; Glynne et al., 1991).

Entre las regiones de clase I y clase II, se sitúa la región de clase III. Esta región de unas

1000 Kb, comprende a un grupo de genes denominados en conjunto genes de clase III, que

codifican entre otras, a las proteínas de los factores del complemento C2, C4, B y F y los factores

de necrosis tumoral TNFα y TNFβ (Tumor Necrosis Factor α y β).

DRTNFα TNFβ

Factores delcomplemento

A GB CClase II

β

DP DM DQ

αβαβ αβ αβLM

PT

AP

Tap

asin

a

Clase III

6

E F

MIC A MIC EMIC DMIC CMIC B

Clase I

Figura 1: Mapa genético del complejo principal de histocompatibilidad humano, ubicado en el brazo corto del cromosoma 6.

INTRODUCCIÓN -6-

I.3. ESTRUCTURA DE LAS PROTEÍNAS DE CLASE I.

Las moléculas HLA de clase I, son glicoproteínas de membrana formadas por la

asociación no covalente de tres componentes: una cadena pesada, HC (Heavy Chain) o

cadena α, de 45 kDa, codificada en el MHC, una cadena ligera de 12 kDa denominada Beta-2

microglobulina (β2m) (Ploegh et al., 1981), codificada en el cromosoma 15 (Goodfellow et al.,

1975) y un péptido antigénico de entre ocho y once aminoácidos. Sus funciónes inmunológicas

son la de actuar como receptoras y presentadoras de péptidos al TCR y la de ser reconocidas por

receptores de células NK (Ljunggren y Kärre, 1990a; Moretta et al., 1994; Phillips et al., 1996)

I.3.1. La cadena pesada y la Beta-2 microglobulina (β2m).

Los genes que codifican la cadena pesada de las moléculas de clase I se distribuyen en

ocho exones de diferente longitud separados por siete intrones (Jordan et al., 1985). Cada exón

codifica fundamentalmente un dominio de la proteína. El exón 1, situado en el extremo 5’ del

gen, codifica una porción no traducida de 18 pares de nucleótidos y un péptido señal de 24

aminoácidos. Los exones 2 y 3 (270 pares de nucleótidos) y 4 (276 pares de nucleótidos)

codifican los dominios extracelulares de la cadena pesada, α1, α2, y α3 respectivamente. El

exón 5 (117 pares de nucleótidos) codifica la región transmembrana y los residuos que la

flanquean. Los exones 6, 7 y 8 (33, 48 y 400 pares de nucleótidos respectivamente) codifican la

región intracitoplásmica y la región 3’ no traducida.

La cadena pesada resultante, consta de aproximadamente 340 aminoácidos estructurados

en tres regiones bien definidas: una región amino terminal extracelular de 274 aminoácidos,

formada por los dominios α1, α2 y α3 de aproximadamente 90 aminoácidos cada uno, una

región transmembrana hidrofóbica de 25 aminoácidos y una intracitoplásmica hidrofílica de unos

30 aminoácidos (Ploegh et al., 1981).

Mediante técnicas de difracción de rayos X, ha podido determinarse la similar

organización tridimensional de estos dominios en distintas moléculas de clase I, humanas y

murinas (Bjorkman et al., 1987; Garrett et al., 1989; Saper et al., 1991; Madden et al., 1991;

Madden et al., 1992; Fremont et al., 1992; Zhang et al., 1992; Young et al., 1994; Smith et al.,

1996b; Smith et al., 1996c). Los dominios α1 y α2, que son estructuralmente idénticos, constan

cada uno de una lámina β antiparalela formada por cuatro cadenas polipeptídicas y de una larga

región en α-hélice (Figura 2). Tras el plegamiento de ambos dominios las dos estructuras en α-

-7- INTRODUCCIÓN

hélice coronan una base constituida por ocho

láminas β-plegadas, delimitando una hendidura,

denominada surco de unión del péptido, de

aproximadamente 25Å de largo por 10Å de ancho.

El polimorfismo de los antígenos de clase I,

se concentra principalmente en los residuos

situados en el interior y las zonas próximas a este

surco, donde se distinguen seis subcavidades

(Saper et al., 1991) denominadas A-F. Su forma,

tamaño y polaridad, está determinada por los

residuos polimórficos que las constituyen, y que

generan diferentes especificidades por los péptidos

unidos (Figura 2).

En el residuo Asn86 del dominio α1 se situa

un carbohidrato de 3,3 kDa y un puente disulfuro

intradominio en α2 conecta las Cisteínas 101 y

164.

El dominio α3, que está muy conservado

(Parham et al., 1988), es estructuralmente

homólogo a la β2m, y a los dominios constantes de

las inmunoglobulinas (Orr et al., 1979). Un puente

disulfuro conecta los residuos Cys203 y Cys259 y

una corta α-hélice adicional entre los residuos

177-181, conecta este dominio con el α2.

Contigua al dominio α3 se encuentra la región transmembrana encargada del anclaje de la

molécula a la membrana celular.

Figura 2: (Arriba, vista de perfil), diagrama esquemático que representa la estructura tridimensional adoptada por los diferentes dominios de una molécula de clase I. No aparecen representadas ni la región transmembrana ni el fragmento que une los dominios extracelulares a la superficie celular. En la figura inferior (vista apicalmente) se detalla la localización de las diferentes subcavidades distribuidas a ambos lados del surco de unión del péptido.

AB C

D EF

La β2-microglobulina, de la que sólo se conoce un alelo en humanos, consta de 99

aminoácidos muy conservados entre diferentes especies. Carece de dominio intracitoplásmico y

se mantiene asociada a la región extracitoplásmica de la cadena pesada interaccionando con sus

tres dominios, lo que determina en gran medida la conformación y estabilidad de ésta (Krangel et

al., 1979; Seong et al., 1988).

INTRODUCCIÓN -8-

La β2m y el dominio α3 se sitúan por debajo de la lámina β de α1/α2 dejando una cavidad

entre ellos donde el correceptor CD8 de la célula T interacciona con la molécula de clase I

(Salter et al., 1990, Gao et al., 1997).

I.3.2. Los péptidos.

Los péptidos unidos a las moléculas de clase I, funcionan como una parte integral de las

mismas permitiendo su plegamiento y expresión estable en la membrana celular, y participando

en la selección de repertorios de células T CD8+ (Abe et al., 1992). La cinética de unión entre un

péptido y la molécula de clase I, está gobernada por las constantes de asociación y disociación, a

su vez determinantes de su afinidad y estabilidad. Esta última parece ser el factor determinante

de la inmunogenicidad de los péptidos (Brooks et al.1998).

Desde un punto de vista estructural, la unión de un péptido está determinada por dos

características básicas: una es su tamaño, generalmente comprendido entre 8 y 11 aminoácidos

(Jardetzky et al., 1991; Falk et al., 1991) y la otra, es la conservación en determinadas posiciones

de residuos de anclaje cuya localización y contribución varían en función del polimorfismo de la

molécula de clase I a la que se unen. La presencia de un único aminoácido o aminoácidos con

cadenas laterales similares en una posición de la secuencia del péptido, define los motivos de

anclaje primarios, frecuentemente P2 y el residuo C-terminal (PΩ). Otras posiciones de la

secuencia del péptido como P1, P3, y P7, son menos restrictivas respecto al aminoácido que las

ocupa, y definen los motivos de anclaje secundarios que contribuyen adicionalmente a la unión

del péptido.

La unión tiene lugar principalmente entre los extremos N-terminal y C-terminal con los

residuos que conforman las subcavidades A y F respectivamente, adoptando una conformación

extendida (Madden et al., 1991), y arqueandose en su parte central (Guo et al., 1992), aunque en

péptidos más largos puede sobresalir un extremo (Collins et al., 1994). Otras posiciones de la

cadena principal del péptido (P2, P3, P7), interaccionan con el resto de las subcavidades (Garrett

et al., 1989). En la parte central, que es la más accesible al contacto con el TCR, las cadenas

laterales de los residuos P4, P5, P6 y P8 del péptido, pueden adoptar distintas orientaciones en

función de su secuencia (Madden et al., 1993).

-9- INTRODUCCIÓN

I.4. PROCESAMIENTO DE ANTÍGENO. FORMACIÓN, TRANSPORTE Y PRESENTACIÓN DE LOS COMPLEJOS PÉPTIDO-MHC DE CLASE I.

I.4.1. Origen y procesamiento de los péptidos antigénicos.

Los linfocitos T, reconocen los antígenos en forma de péptidos unidos a las moléculas del

MHC. El origen de la proteína antigénica determina tanto la vía de procesamiento, como el

ulterior mecanismo de respuesta inmunitaria. Así, las moléculas del MHC de clase II presentan

principalmente péptidos derivados de proteínas de origen exógeno, que son endocitadas por

APCs y degradadas en el endosoma, en condiciones de pH ácido. Las moléculas de clase I

presentan principalmente péptidos procedentes de la degradación en el citosol de proteínas

propias o de secuencias señal (Wei y Cresswell, 1992) por un mecanismo independiente de la vía

endocítica o lisosómica (Morrison et al., 1986).

La degradación de proteínas citosólicas es un proceso estrechamente regulado para evitar

la degradación indiscriminada de proteínas propias. Las formas de “etiquetar” la futura

degradación proteica son variadas; unas se basan en características de la propia proteína, como la

identidad del aminoácido N-terminal (Townsend et al., 1988; Varshavsky, 1992) o la presencia

de secuencias de aminoácidos, como las secuencias "PEST" ricas en Pro, Gln, Ser y Thr (Rogers

et al., 1986) o las denominadas “destruction boxes” (Glotzer et al., 1991) que determinan la

degradación específica en un determinado momento del ciclo celular. En otras ocasiones, las

proteínas que deben ser degradadas, son ubiquitinadas enzimáticamente como paso previo a su

degradación específica. La mayor parte de la actividad proteolítica del citosol es realizada por el

proteosoma, un complejo proteico multicatalítico con actividad proteasa y dependiente de ATP,

encargado del reciclaje de proteínas propias dentro de la célula (Rock et al., 1994, Fentenay et

al., 1995).

Existen dos variantes funcionales del proteosoma: El proteosoma 20S de 700 kDa, está

formado por cuatro anillos superpuestos, de siete subunidades cada uno (Löwe et al., 1995; Groll

et al., 1997). Los anillos exteriores poseen subunidades α cuya función es esencialmente

estructural y reguladora. Los dos anillos internos poseen subunidades β de función catalítica.

Entre los componentes del proteosoma se encuentran las subunidades constitutivas X, Y(δ)

y Z. Tras la inducción por interferón-γ (IFN-γ), éstas son sustituidas por las subunidades

funcionales LMP7, LMP2 y MECL-1 respectivamente (Kelly et al., 1991; Glynne et al., 1991;

Groettrup et al., 1996). Los estudios realizados sobre la función de las subunidades inducibles

son poco concluyentes; así, mientras unos sugieren que en la presentación de antígeno no son

INTRODUCCIÓN -10-

necesarias (Arnold et al., 1992; Momburg et al., 1992; Yewdell et al., 1994), otros indican que

en su ausencia la presentación es menor (Cerundolo et al., 1995). En otros estudios estas

subunidades parecen alterar el patrón de corte del proteosoma (Gaczynska et al., 1993;

Gaczynska et al., 1994; Driscoll et al., 1993), y mientras que LMP7 incrementa la ruptura

catalítica tras residuos hidrofóbicos y básicos, LMP2 reduce la capacidad de corte tras residuos

ácidos. Por último, en un estudio reciente (Eleuteri et al., 1997) LMP7 favorece alguna de las

diferentes actividades catalíticas del proteosoma. En concreto, disminuye la actividad de tipo

quimotripsina, de corte tras residuos hidrofobicos, y aumenta la preferencia de corte tras

aminoácidos aromáticos o con cadenas laterales ramificadas.

La otra variante del proteosoma, encargada de la ruptura de proteínas ubiquitinadas, es la

26S de 1500 kDa. Su núcleo catalítico está formado por el proteosoma 20S asociado a proteínas

adicionales que regulan su actividad (Pamer y Cresswell, 1998).

Aunque hay evidencias tanto de la existencia de un recorte o “trimming” por

endopeptidasas dentro del RE (Retículo Endoplásmico) (Snyder et al., 1994; Hughes et al.,

1996), como de la retrotranslocación al citoplasma previamente a su recorte (Roelse et al., 1994),

se desconoce la importancia que tiene esta variante generadora de péptidos.

I.4.2. Translocación al Retículo Endoplásmico.

Una vez generados en el citosol, los péptidos pasan al RE donde se unen a las moléculas de

clase I nacientes, para formar complejos péptido/MHC de clase I estables que son translocados

hasta la superficie celular.

La demostración de que la expresión de las moléculas de clase I era recuperada en células

mutantes tras añadir péptido exógeno (Townsend et al., 1989), y de que la expresión era

recuperable transfectando dos genes localizados en la subregión de clase II (Spies et al., 1991),

sugerían el importante papel del péptido en la estabilización de la molécula de clase I, y la

implicación de un transportador responsable de la trasnslocación de péptidos al RE. El

responsable resultó ser una proteína heterodimérica, expresada en la membrana del RE,

dependiente de ATP y perteneciente a la familia de los ABC (ATP Binding Cassette)

(Androlewicz et al., 1994) denominada TAP (Transporter associated with Antigen Processing).

El TAP está constituido por la asociación de los monómeros TAP1 y TAP2. Ambos poseen una

región N-terminal con múltiples dominios transmembrana, y un dominio C-terminal de unión a

ATP.

-11- INTRODUCCIÓN

TAP muestra preferencia por péptidos de 9-10 aminoácidos, el tamaño canónico de los

péptidos unidos a las moléculas de clase I. La identidad del aminoácido C-terminal parece

importante en la selectividad del TAP que en humanos une preferentemente péptidos con

residuos C-terminales hidrofóbicos y básicos y no acepta Prolina ni Glicina (Momburg et al.,

1994). El tamaño de los péptidos translocados por TAP, oscila entre 8 y 13 aminoácidos aunque

pueden ser mayores (Androlewicz et al., 1994; Schumacher et al., 1994). La translocación de los

péptidos al RE, a diferencia de la unión de péptidos a TAP, es un proceso dependiente de ATP.

I.4.3. Biosíntesis, Ensamblaje y expresión en membrana de las moléculas de clase I.

Tras la biosíntesis de las proteínas de clase I, en el correcto plegamiento y ensamblaje de

HLA intervienen secuencialmente diferentes chaperonas del RE rugoso; éstas actúan como un

control de calidad, que impide la asociación de las subunidades plegadas incorrectamente. Las

cadenas pesadas de las moléculas de clase I formadas de novo, son retenidas en primera instancia

por la Calnexina, una chaperona de membrana de 88 kDa (Jackson et al., 1994; Rajagopalan et

al., 1994) que las mantiene parcialmente plegadas hasta que se une la β2m. Una vez formado el

heterodímero Cadena α/β2m, se disocia de la Calnexina y se une a otra chaperona similar,

aunque no insertada en la membrana, la Calreticulina (Sadasivan et al., 1996) y a la Tapasina

(Sadasivan et al., 1996; Ortmann et al., 1997), una glicoproteína transmembranal de 48 kDa. La

Tapasina mantiene asociado al heterodímero parcialmente plegado con la subunidad TAP1 en

espera de que un péptido translocado se una. La unión del péptido induce un cambio de

conformación que estabiliza el complejo trimérico Péptido/Cadena pesada/β2m y desestabiliza la

union con TAP, lo que provoca la disociación del complejo. El complejo péptido/MHC de clase I

es posteriormente transportado, pasando por el TGN (Trans Golgi Network) y por un sistema de

vesículas, hacia la superficie celular.

I.5. TCR Y RECONOCIMIENTO DE LOS COMPLEJOS MHC-PÉPTIDO. I.5.1. Características estructurales del TCR y reconocimiento de los complejos MHC-péptido.

El receptor de antígeno de los linfocitos T maduros, está compuesto por heterodímeros

clonotípicos α:β, el más común en los linfocitos humanos de sangre periférica, o γ:δ. Las

subunidades están unidas por puentes disulfuro, formando una estructura muy similar a la del

fragmento Fab de una inmunoglobulina. Cada uno de los monómeros se estructura en dos

dominios: uno distal muy variable, responsable de la unión a los complejos MHC/Péptido, y otro

INTRODUCCIÓN -12-

proximal constante. Ambos dominios tienen unido un carbohidrato. Éstos heterodímeros se

asocian no-covalentemente al complejo CD3, que está formado por tres polipéptidos: γ, δ y ε;

asociados a dos subunidades independientes, las cadenas ζ y η que forman homodímeros ζζ o

heterodímeros ζη unidos por puentes disulfuro.

CD3 participa en el proceso de transducción de señales al interior celular tras el

reconocimiento por el TCR de un complejo MHC/Péptido (Weiss y Littman, 1994) y en el

mantenimiento de la expresión del TCR en la superficie celular.

Los genes codificantes del TCR se asemejan a los de las inmunoglobulinas, tanto en su

secuencia, como en la forma de ensamblarse, mediante reordenamientos de los segmentos

génicos que codifican las regiones variables (V), de diversidad (D), de unión (J) (Joining) y

constantes (C). La cadena α se origina a partir de reordenamientos de segmentos génicos V-J

mientras que la cadena β surge de dos reordenamientos sucesivos V-D-J para generar exones

funcionales. Ambos segmentos reordenados se unen durante la maduración del linfocito en el

timo, a sus correspondientes regiones C. Tanto en las cadenas TCRα como TCRβ existen cuatro

regiones hipervariables, tres de las cuales forman bucles denominados CDRs (Complementarity-

Determining Regions) CDR1, CDR2 y CDR3 que participan directamente en el reconocimiento

del complejo MHC/Péptido.

Recientemente se ha podido resolver, mediante cristalografía de rayos X, la topología del

reconocimiento por el TCR del complejo MHC-péptido (Garboczi et al., 1996; García et al.,

1996). Los resultados obtenidos se ajustan en gran medida a predicciones indirectas previas,

basadas en el reconocimiento de análogos peptídicos (Jorgensen et al., 1992) y mutantes de

moléculas del MHC (Sun et al., 1995).

En el modelo cristalográfíco, El TCR cubre la superficie del péptido y parte de la molécula

del MHC transversalmente, de manera que el reconocimiento del complejo MHC-péptido es

realizado por los CDRs. Cada uno de los CDRs reconoce una parte diferente del complejo. Las

regiones CDR3α y CDR3β se sitúan sobre la parte central del péptido y las regiones CDR1α y

CDR1β sobre los extremos del péptido N- y C-terminales respectivamente. Las regiones CDR2α

y CDR2β interaccionan directamente con los dominios α2 y α1 respectivamente de la molécula

del MHC. Este modo de interaccion entre el TCR y los complejos MHC-péptido parece ser

general.

-13- INTRODUCCIÓN

I.6. HLA-B27.

El antígeno HLA-B27 es una de las moléculas de clase I humanas mejor estudiadas debido

a su fuerte asociación con las espondiloartropatías. El 95% de los pacientes que sufren

espondilitis anquilosante (EA), y entre el 60-80% de los que sufren artritis reactiva (AR), son

HLA-B27,

Hasta el momento se han identificado quince subtipos (Tabla 1) cuyo polimorfismo se

localiza fundamentalmente en el sitio de unión del péptido. La estructura cristalográfica del

subtipo principal, HLA-B*2705, ha permitido el análisis detallado de las interacciones de esta

molécula con péptidos. A continuación se detalla la estructura de las subcavidades del sitio de

unión de péptido y la naturaleza de las interacciones entre éstas y las cadenas laterales de los

péptidos unidos.

I.6.1. Subcavidad A.

Esta subcavidad es idéntica en todos los

subtipos de HLA-B27 a excepción de B*2703

que presenta el cambio Y59H.

El extremo N-terminal del péptido se aloja

en esta subcavidad formando una red pentagonal

de puentes de hidrógeno con los residuos

conservados Tyr7, Tyr59 y Tyr 171, y a través de

una molécula de agua intermedia, con los

residuos Glu63 y Glu45 (Figura 3). El oxígeno

del grupo carbonilo de P1 establece un puente de

hidrógeno adicional con el residuo Tyr 159. La

presencia de His59 en B*2703 provoca una

disrupción de esta red induciendo un reforzamiento de las interacciones en la subcavidad B

(Villadangos et al., 1995)

α2

α1 Lámina-β

Tyr7

Tyr171

Tyr59P1

P2P3

Tyr159

Subcavidad A

Figura 3: Red de puentes de hidrógeno que se establecen en la subcavidad A con el extremo N-terminal del péptido. (Madden et al., 1992).

I.6.2. Subcavidad B.

Esta subcavidad está conservada entre los subtipos de HLA-B27 y es diferente en otras

moléculas de clase I a excepción de B73 (Parham et al., 1994; Vilches et al., 1994b). Está

constituida, entre otros residuos, por His9, Thr24, Glu45, Cys67 y Tyr99 y acomoda la cadena

INTRODUCCIÓN -14-

lateral del residuo P2 del péptido (Figura 4), que

en HLA-B27 es Arg y constituye el anclaje

peptídico principal (Jardetzky et al., 1991;

Rammensee et al., 1995).

En el modelo cristalográfico, el grupo

guanidinio de la Arg2, establece cinco puentes

de hidrógeno, tres de ellos directamente con

Glu45 y Thr24 y dos indirectos con His9 y Tyr99

por mediación de una molécula de agua. El

grupo sulfhidrilo de la Cys67 se sitúa

exactamente sobre el plano del grupo guanidinio.

El Glu45 es un residuo crítico en la especificidad

de la subcavidad B por Arg2 (Villadangos et. al., 1995).

Glu45

Cys67

His9

Tyr99

P2Arg

PEPTIDO

Thr24

Subcavidad B

Figura 4: Interacciones que tienen lugar entre la cadena lateral de la Arginina y residuos de la subcavidad B. (Madden et al., 1992)

La Lys70, un residuo exclusivo de HLA-B27 presente en casi todos los subtipos, se

localiza en las proximidades de esta subcavidad. Sin embargo su cadena lateral se orienta hacia

fuera de la subcavidad B porque establece un puente salino con el Asp74. La mutación Tyr74

presente en B*2701 impide la formación de dicho puente salino relajando la conformación de la

Lys70, lo que afecta a la especificidad de la subcavidad B y favorece la presencia de Gln en P2

(García et al., 1997c / Anexo 2).

I.6.3. Subcavidades C y F.

Las subcavidades C y F forman en HLA-B27 una única cavidad denominada C/F, donde el

extremo C-terminal del péptido establece una amplia red de puentes de hidrógeno (Figura 5). La

mayor parte de las diferencias entre los subtipos de HLA-B27 se concentra en esta cavidad, lo

que sugiere una tendencia evolutiva del polimorfismo de HLA-B27 centrado en modular las

especificidades por el extremo extremo C-terminal del péptido. En HLA-B27 esta posición

constituye junto con la posición 2 otro anclaje primario.

En el modelo cristalográfico, el residuo P9, enlaza directamente con Tyr84 y Thr143 e

indirectamente, a través de dos moléculas de agua, con Thr80 y Lys 146.

-15- INTRODUCCIÓN

Las cadenas laterales de los residuos

Asp77 y Asp116, favorecen la unión de residuos

C-terminales básicos (Guo et al., 1992; Silver et

al., 1992). Sin embargo, en HLA-B27, esta

situación no impide acomodar residuos

hidrofóbicos debido, en parte a la influencia de

las cadenas laterales apolares de la Leu81,

Tyr123 y Thr143, y al hecho de que los residuos

C-terminales pueden adoptar distintas

orientaciones y contactar en la subcavidad C/F

según su naturaleza química.

Asp77

Trp147

Thr80 Tyr84

Thr143Lys146

PÉPTIDOP8

P9

α1

α2

P7

Subcavidad C/F

Figura 5: Interacciones entre el extremo C-terminal del péptido y los residuos de la cavidad C/F de B*2705. Los residuos polimorficos aparecen subrayados. (Madden et al., 1992).

I.6.4. Subcavidades D y E.

Las posiciones P3 y P7 constituyen anclajes secundarios de los péptidos que une HLA-

B27. La cadena lateral del residuo P3 interacciona en la subcavidad D. Ésta contiene las cadenas

laterales aromáticas Tyr99 y Tyr159 y la alifática Leu156 que determinan su naturaleza

predominantemente apolar. En B*2705 la subcavidad D, muestra preferencia por residuos con

cadenas laterales voluminosas hidrofóbicas. La cadena lateral polar de His114, se sitúa

lateralmente a esta subcavidad pudiendo aceptar la cadena lateral de residuos polares. Esta

preferencia por el residuo P3 puede variar en los subtipos B*2706, B*2707 y B*2711 donde la

posición 114 está ocupada por Asp o Asn.

La subcavidad E aloja la cadena lateral del residuo P7. En HLA-A2 (Madden et al., 1993)

la cadena lateral de este residuo muestra una gran variabilidad conformacional dependiente de la

secuencia completa del péptido que le permite, según el caso, interaccionar con la molécula de

clase I o ser accesible al TCR. En la región intermedia del péptido los contactos con la molécula

de HLA-B27 son escasos y las cadenas laterales de los residuos P4, P5, P6 y P8 son en general

accesibles al contacto con el TCR.

I.7. HLA-B27 Y ESPONDILOARTROPATÍAS.

Con el término espondiloartropatías seronegativas, se integra a cinco subcategorías

clínicas: la espondilitis anquilosante (EA), la artritis reactiva (AR), la artritis psoriásica, la artritis

asociada a la enfermedad inflamatoria intestinal y las espondiloartropatías indiferenciadas. Estas

INTRODUCCIÓN -16-

enfermedades comparten entre otras características, la ausencia de factores reumatoides y una

predisposición genética asociada a la presencia de HLA-B27. Sus manifestaciones clínicas se

presentan como un conjunto de desordenes reumáticos y extraarticulares que solapan en muchos

de sus síntomas. La espondilitis anquilosante (EA), considerada como la manifestación más pura

de espondiloartropatía, es una enfermedad crónica que afecta principalmente al esqueleto axial.

La expresión de HLA-B27 no es una condición necesaria ni suficiente para desarrollar una

espondiloartropatía, pero sí supone un factor de riesgo, sobre todo en el caso de la EA

(Brewerton et al., 1973; Schlosstein et al., 1973) y en menor medida en la AR (Brewerton et al.,

1974). Evidencias indirectas sugieren además que otros genes pueden incrementar la

susceptibilidad a padecer enfermedad (Brown y Wordsworth, 1997).

La importancia que juegan los factores ambientales en el desarrollo de la EA, se

desconocen; pero en el caso de la AR, está bien documentada la implicación de bacterias

intracelulares obligadas o facultativas responsables de las infecciones de los epitelios urogenital

(Chlamidia trachomatis), entérico (Yersinia, Sigella, Salmonella o Campylobacter) y respiratorio

(Chlamidia pneumoniae). Los estudios realizados en ratas y ratones transgénicos refuerzan la

estrecha relación existente entre la expresión de HLA-B27, la flora bacteriana y el desarrollo de

artritis.

I.7.1. Posible papel patogénico de la presentación de péptidos por HLA-B27. El mecanismo de asociación de HLA-B27 a las espondiloartropatías se desconoce. Entre las

diversas hipótesis que intentan explicar el papel patogénico de HLA-B27, tal vez la que está

apoyada por evidencias más convincentes es la teoría del péptido artritogénico (Benjamin y

Parham, 1990). Esta hipótesis postula que el evento patogénico primario en las

espondiloartropatías sería una estimulación antigénica externa (por ejemplo una infección

bacteriana) de una respuesta celular citotóxica contra un péptido restringido por HLA-B27. Estos

CTLs activados exógenamente reaccionarían cruzadamente con algún péptido propio presentado

por HLA-B27, lo que desencadenaría una reacción autoinmune. Las evidencia a favor de esta

hipótesis son más o menos circunstanciales.

(i) En humanos, ha sido demostrada la existencia de CTLs autorreactivos contra péptidos

del colágeno de tipo II en el líquido sinovial de pacientes con AR, (Gao et al., 1994), y de CTLs

específicos de péptidos bacterianos restringidos por HLA-B27 en pacientes con AR y EA

(Hermann et al., 1993, Duchman et al., 1996. Ugrinovic et al., 1997).

-17- INTRODUCCIÓN

(ii) En ratas transgénicas, la expresión de un elevado número de copias de HLA-B27, es

suficiente para el desarrollo de una enfermedad inflamatoria espontánea, con muchas

características similares a las espondiloartropatías humanas. El papel de las bacterias en este

modelo es evidente puesto que la enfermedad no se desarrolla en condiciones libres de gérmenes

(Hammer et al., 1990; Taurog et al., 1994; Breban et al., 1996). Recientemente también se ha

demostrado que la alteración del repertorio peptídico de HLA-B27 induce la ausencia de artritis

periférica en ratas transgénicas (Zhou et al., 1998). Esta es la evidencia más directa de un papel

crítico de los péptidos de HLA-B27 en la patogenia de la enfermedad asociada con este antígeno.

(iii) La sociación diferencial de los subtipos de HLA-B27 a EA. Este tema se desarrolla

en detalle en el siguiente apartado.

I.7.2. Distribución étnica y asociación a enfermedad de los subtipos de HLA-B27.

Se conocen hasta el momento quince subtipos de HLA-B27 que difieren entre uno y siete

residuos ubicados principalmente en el sitio de unión del péptido (Figura 6). Probablemente

todos ellos han derivado por diferentes mecanismos evolutivos de B*2705 (López-Larrea et al.,

1996). Sus distribuciónes étnicas y su asociación a enfermedad son variables (Tabla 1).

Entre las poblaciones euro-caucasoides el subtipo mayoritario es B*2705, presente en el

90% de los individuos B27+. Prácticamente el 10% restante son B*2702. A su vez este es el

subtipo predominante en judíos, donde representa el 60% de los individuos B27+ (González-

Roces et al., 1994). B*2709 es frecuente entre la población Sarda. B*2701, B*2708 y B*2710

son subtipos aparentemente minoritarios.

B*2703 detectado inicialmente entre población negra de Norteamérica es originario de

poblaciones del oeste africano.

B*2704 y B*2706 se encuentran en poblaciones asiáticas, donde el primero es mayoritario.

B*2707 es menos frecuente pero se encuentra también en poblaciones del sudeste asiático.

B*2705, B*2702, B*2704(López-Larrea et al., 1995; Nasution et al., 1997; Ren et al.,

1997) y B*2707 están claramente asociados a enfermedad. B*2706 y B*2709 (D’Amato et al.,

1995) no están asociados a enfermadad en poblaciones donde otros subtipos si lo están aunque se

han descrito dos casos de individuos B*2706 en China.

B*2703 se ha encontrado en algún individuo con EA (González-Roces et al., 1997), sin

embargo esta enfermedad es muy rara entre la poblacion del oeste de África, donde predomina

este subtipo, incluso entre los individuos B*2705 de la misma población, lo que sugiere que

INTRODUCCIÓN -18-

probablemente están implicados otros factores genéticos protectores (Brown et al., 1997).

Recientemente se ha encontrado co-segregación de EA con B*2708 en un estudio familiar

(Armas et al., 1999).

Los subtipos B*2701, B*2710, B*2711, B*2712. son subtipos poco frecuentes

encontrados en individuos aislados, lo que impide un tratamiento estadístico adecuado para

determinar su asociación a enfermedad. B*2713, B*2714 y B*2715 han sido descritos

recientemente como nuevos subtipos. El primero es una variante cuya mutación aparece en la

secuencia señal del exón 1 (Tabla 1).

α1 α2 α3

Subcavidad A

C/F

C/F

C/F

C/F

E C/F

D

D/E

C/F

E

Subtipo Asoc. a EA

59

69

70

71

74

77

80

81

82

83

97

113

114

116

131

152

211

B*2705 (a) + Y A K A D D T L L R N Y H D S V A B*2701 (b) ¿ ? − − − − Y N − A − − − − − − − − B*2702 (c) + − − − − − N I A − − − − − − − − − B*2703 (d) + H − − − − − − − − − − − − − − − − B*2704 (e) + − − − − − S − − − − − − − − − E G B*2706 (f) − − − − − − S − − − − − − D Y − E G B*2707 (g) + − − − − − − − − − S H N Y R − − B*2708 (h) + − − − − − S N − R G − − − − − − − B*2709 (i) − − − − − − − − − − − − − − H − − − B*2710 (j) ¿ ? − − − − − − − − − − − − − − − E n.d B*2711 (k) ¿ ? − − − − − S − − − − S H N Y R − − B*2712 (l) ¿ ? − T N T − S N − R G − − − − − − n.d B*2713 (m) ¿ ? − − − − − − − − − − − − − − − − − B*2714 (n) # ¿ ? − − − − − − − − − − − − − − − − B*2715 (ñ) # ¿ ? Tabla 1: Residuos polimórficos entre los subtipos de HLA-B27.

Los guiones indican identidad con el aminoácido presente en B*2705. (n.d; no determinado). Referencias: a: (Ezquerra et al., 1985; Moses et al., 1995). b: (Rojo et al., 1987a; Choo et al., 1986). c: (Vega et al., 1985a; Moses et al., 1995). d: (Rojo et al., 1987b; Choo et al., 1988). e: (Vega et al., 1985b; Rudwaleit et al., 1996). f: (Vega et al., 1986; Vilches et al., 1994a). g: (Choo et al., 1991). h: (Hildebrand et al., 1994). i: (Del Porto et al., 1994). j: (Fernández-Viña et al., 1996). k: (Hasegawa et al., 1997). l: (Balas et al., 1998). m: (Seurynck y Baxter-Lowe, 1998). n: (Hurley,C./ Nº acceso.AF072763-64, presenta los cambios L95W; N97T; V103L). ñ: (Van den Berg-Loonen E.M./ Nº acceso Y16637-38. Secuencia no publicada). # www.anthonynolan.com.

-19- INTRODUCCIÓN

Figura 6: Distribución de las posiciones polimórficas en los dominios α1 y α2 de la molécula de clase I HLA-

B27.

59

114

8180 82 837774

71

70

69

116

152113

131

97

-21-

II. OBJETIVOS

-25-

III. MATERIALES Y MÉTODOS

MATERIALES Y MÉTODOS -27-

III. MATERIALES Y MÉTODOS

III.1. LÍNEAS

as células RMA-S pertenecen a una línea celular mutante derivada del linfoma de células

T de ratón RBL-5 (H-2

CELULARES.

b) inducido por el virus de Rauscher (Ljunggren y Kärre, 1985).

Estas células, son defectivas en proteínas TAP2 y muestran una baja expresión en

superficie de antígenos de clase I a 37ºC. Cuando se cultivan a temperaturas inferiores a 30ºC la

expresión en membrana se incrementa, y expresan moléculas inestables de MHC de clase I vacías

de péptido (Ljunggren et al., 1990b). La estabilidad, en estas condiciones, puede ser recuperada

añadiendo péptido exógeno. En esta tesis se han utilizado diferentes transfectantes de subtipos y

mutantes de HLA-B27 clonados previamente en el laboratorio.

L

III.2. ANTICUERPOS MONOCLONALES (mAb). El anticuerpo ME1 (isotipo IgG1) (anti-HLA-B27+B7+B22) (Ellis et al., 1982), reconoce un

epítopo conformacional en el dominio α1 (El-Zaatari et al., 1990) que no es polimórfico entre los

subtipos de HLA-B27, y que no se altera tras la unión del péptido (Smith et al., 1996a). Se obtuvo

a partir de sobrenadante de cultivo del hibridoma secretor de dicho anticuerpo. Una vez verificada

su correcta actividad, el sobrenadante se filtró y almacenó a –20ºC con 0,02% azida sódica.

Como segundo anticuerpo, en citometría de flujo, se utilizó una dilución 1/100 de un mAb

comercial fluoresceinado (FITC, Fluorescein Isothiocianate-conjugated rabbit F(ab’)2-anti-mouse

IgG [H+L]) (Southern Biotechnology, Birmingham, Ala).

III.3. SÍNTESIS PURIFICACIÓN Y CUANTIFICACIÓN DE PÉPTIDOS. III.3.1. Síntesis y purificación de péptidos y análogos no peptídicos. Los péptidos se sintetizaron en fase sólida mediante un sintetizador múltiple de péptidos

AMS 422 utilizando la química F-moc. Tras la síntesis, se purificaron en HPLC de fase reversa

midiendo su absorbancia a 210 y 280 nm., empleando un equipo Waters LC 625 provisto de una

columna µ-Bondapack C18 (Waters) de 300 x 7,8mm. Se utilizó un gradiente de concentración

creciente de acetonitrilo a un flujo de 2ml/min durante 60 minutos. con: (A): H2O +TFA 0,1%

MATERIALES Y MÉTODOS -28-

(ácido trifluoroacético) y (B): acetonitrilo + TFA 0,1%. El gradiente, fue el siguiente: 0-5 minutos

con 100% de (A), 5-40 min hasta un 40% de (B), de 40-45 hasta 60% de (B) de 45-50 hasta el

100% de (B). La pureza de los picos purificados se confirmó mediante HPLC analítico en fase

reversa utilizando una columna Nova-Pack® C18 (Waters) 3,9 x 150mm a un flujo de 0,5ml/min y

el mismo gradiente de acetonitrilo-TFA. La correcta masa molecular se determinó por

espectrometría de masas mediante un espectrómetro VG Auto Spec.

Los análogos no peptídicos utilizados en esta tesis, fueron sintetizados en el Laboratory for

Organic Chermistry, Swiss Federal Institute of Technology. Zürich (Switzerland). Su

cuantificación y control de pureza, se analizó igual que el resto de péptidos.

III.3.2. Cuantificación. La cuantificación de los péptidos y su correcta composición, se determinó mediante un

proceso previo de hidrólisis con ácido clorhídrico en atmósfera reductora de Nitrógeno, en

presencia de un control estandard de norleucina, durante 24h a 110ºC. El análisis posterior se llevó

a cabo mediante un analizador de aminoácidos (Beckman, Palo Alto, CA). Los péptidos se

conservaron en agua Milli-Q a -70ºC a una concentración final 1mM.

III.4. ENSAYO DE UNIÓN DE PÉPTIDOS. El ensayo de unión de péptidos se realizó mediante un procedimiento cuantitativo (Galocha

et al., 1996 / Anexo 2), basado en la estabilización, por péptidos añadidos exógenamente, de las

moléculas vacías de HLA-B27 expresadas a 26ºC.

Los transfectantes se incubaron a 26ºC durante 18-24 h, a una concentración de 106

células/ml en placas estériles de 96 pocillos de fondo en V (NUNC) en 100µl de RPMI 1640,

25mM HEPES y 10% de FBS (Suero bovino fetal) en ausencia de péptido. Transcurrida la

incubación a 26ºC, las placas se lavaron dos veces con PBS estéril y se les añadió el péptido

diluido en medio RPMI 1640, 25mM HEPES sin suero, para obtener concentraciones finales

comprendidas entre 10-4M hasta 10-9M. Posteriormente las células se incubaron durante una hora

a 26ºC y, tras este periodo, a 37ºC durante 2h o 4h dependiendo del subtipo: B*2701, B*2702,

B*2703, B*2704 y los mutantes Y74, A81, D114 y E152 se incubaron durante 2h. B*2705,

B*2706, S77, Y116, D114Y116 se incubaron durante 4h. Estos tiempos se eligieron de forma

que la expresión de HLA-B27 era claramente superior a la expresión basal a 37ºC, pero en los

que en presencia de péptido la disociación y pérdida de expresión en superficie predominase

-29- MATERIALES Y MÉTODOS

sobre la asociación e inducción de la expresión.

III.4.1. Análisis por citometría de flujo. Las células se lavaron dos veces con PBS para eliminar cualquier traza de péptido en el

sobrenadante y se incubaron nuevamente durante 30 minutos a 4ºC con 30-50 µl de anticuerpo

monoclonal ME1. Tras este período, se lavaron otras dos veces con PBS estéril y se añadieron de

30-50µl del segundo anticuerpo monoclonal fluoresceinado, dejando transcurrir otros 30 minutos a

4ºC. Transcurrida la segunda incubación, se lavaron dos veces con PBS estéril. y se fijaron con

100µl de paraformaldehído previamente a su análisis en un citómetro de flujo EPICS Profile II

Coulter donde se midió la fluorescencia lineal para un total de 5000. células.

III.4.2. Cálculo de la unión de péptidos. La media de la fluorescencia lineal asociada a la expresión de HLA-B27 se representó en

función de las diferentes concentraciones de péptido. El análisis cuantitativo y comparativo de la

unión de los diferentes péptidos se realizó mediante el programa Origin MicroCal Software Inc.

Tras calcular la concentración molar de péptido a la cual se producía el 50% de la fluorescencia

máxima (C50). La concentración molar requerida de cada péptido para alcanzar el valor de

fluorescencia C50 del péptido utilizado como referencia o control, se denominó EC50. La unión

relativa se definió como la relación molar entre los valores EC50 de los péptidos comparados.

Los intervalos de afinidad se escogieron en función de los valores de afinidad obtenidos en

los ensayos de unión con los péptidos presentados in vivo. Así las afinidades se definieron como

sigue: (Afinidad alta: <5µΜ.), (Afinidad intermedia: 6µM-50µM ) y (Afinidad baja: >50µM). La

afinidad de péptidos mayor de 100µM no se midió en este ensayo, y refleja una unión marginal o

nula.

-31-

IV. RESULTADOS

RESULTADOS -33-

IV. RESULTADOS IV.1. EFECTO DEL POLIMORFISMO DE HLA-B27 SOBRE LA ESPECIFICIDAD DE UNIÓN DE

PÉPTIDOS.

on la finalidad de determinar las bases moleculares que determinan el solapamiento de

repertorios peptídicos entre subtipos de HLA-B27, el estudio se centró en el análisis de

la especificidad peptídica de: (i) B*2704 y B*2706, (ii) B*2701 y (iii), B*2703.

B*2704 y B*2706 están relacionados estructuralmente pero desigualmente asociados a

enfermedad. B*2701 y B*2702 difieren de B*2705 en mutaciones distintas pero localizadas en la

misma región de la molécula. El polimorfismo de B*2703, afecta a una interacción conservada con

el extremo peptídico N-terminal.

C Las diferencias entre estos subtipos se analizaron comparando las afinidades de unión in vitro

de ligandos naturales sintéticos de B*2705 (05.Pi) (Jardetzky et al., 1991) y B*2702 (02.Pi)

(Rötzschke et al., 1994). Las bases moleculares de estas diferencias se analizaran con mutantes

que mimetizaban los cambios puntuales entre los subtipos, y mediante análogos peptídicos a los

que se cambia en los residuos principales de anclaje PΩ o P2.

IV.1.1. Unión de péptidos a HLA-B*2705, B*2704 y B*2706. Modulación de la especificidad

por el residuo peptídico C-terminal. (Anexo 1) B*2704 y B*2706, son los dos subtipos HLA-B27 restringidos a poblaciones asiáticas, muy

diferentes antigénicamente de B*2705 (López et al., 1994). B*2706 no está asociado a enfermedad

en poblaciones donde B*2704 si lo está. B*2704 difiere de B*2705 en dos cambios: D77S y

V152E. B*2706 tiene además dos cambios adicionales a los de B*2704, H114D y D116Y. Ambos

subtipos tienen además el cambio A211G, que por su localización, en el dominio α3, no afecta a la

unión del péptido.

Tanto los ligandos (05.Pi) como (02.Pi), se unieron a B*2705 con alta afinidad (EC50 ≤5µM),

(Tabla 2), lo que sugiere que el repertorio de B*2705 podría englobar una parte del repertorio

natural de B*2702.

El estudio con los análogos peptídicos en P2 mostró un descenso de la afinidad de unión,

RESULTADOS -34-

indicando que este anclaje era esencial para mantener la buena unión del péptido, sin embargo a

diferencia de lo que ocurre con los análogos de Ala2, el residuo Gln2 funcionaba como un residuo

subóptimo permitiendo una unión significativa, y por lo tanto compatible con la presencia de este

residuo entre los ligandos naturales de B*2705 (Tabla 3A).

La contribición del residuo P9 fue variable entre los diferentes análogos. Así, Lys en 05.P2

era mejor que en 05.P6 mientras que Tyr era equivalente a Arg y Leu en 02.P4 y 02.P6, y también

a Ala en 02.P4 pero no en 02.P6 (Tabla 3B). Esto indica que la tolerancia de determinados residuos

está influida por el resto de la secuencia del péptido. El estudio de la contribución de los diferentes

residuos debe realizarse con análogos que eviten, en lo posible, la interferencia de otras posiciones

(ver Anexo 5).

B*2704, unió eficientemente péptidos con residuo C-terminal alifático (Leu) o aromático

(Tyr) (EC50≤5µM) y peor aquellos con residuo básico (Tabla 2). Los análogos con Gln2 de

péptidos con residuos C-terminales alifáticos (Leu) se unieron a este subtipo mejor que a B*2705,

indicando que el anclaje C-terminal hidrofóbico es más fuerte que en B*2705. En cambio en

05.P6Q2 no se unió, lo que significa que la unión de Lys C-terminal es más débil que en B*2705

(Tabla 3B). La unión de los análogos de 02.P4 y 02.P6 con L9 (Tabla 3B) fue superior a la de los

péptidos correspondientes, indicando que Leu es mejor residuo C-terminal que Tyr para la unión a

B*2704.

B*2706 unió eficientemente péptidos con residuos C-terminales apolares (Leu, Phe), pero no

aquellos con Tyr o residuos básicos (Tabla 2). La unión relativa de los análogos con Gln2, que fue

buena para péptidos con Leu y Phe y mala para Lys o Tyr, confirmó estas preferencias (Tabla 3A)

La unión de análogos con L9 fue claramente superior que en los péptidos con Tyr C-terminal

(02.P4, 02.P6) lo que confirma las preferencias de B*2706 por Leu respecto a Tyr.

Estos resultados indican que B*2704 y B*2706 difieren de B*2705 sobre todo en su peor

tolerancia por residuos C-terminales básicos. Además la diferencia fundamental entre B*2704 y

B*2706, reside en la menor aceptación por este último de residuos con Tyr C-terminal.

IV.1.1.1. Efecto de las pérdidas y ganancias de residuos cargados en las subcavidades E/C/F. El uso de mutantes que reproducen los cambios en B*2705 y B*2706, permitió el análisis del

papel que juegan las posiciones polimórficas que diferencian a estos subtipos. Estas mutaciones

determinan las propiedades electrostáticas de las subcavidades donde interacciona el péptido.

La eliminación de una carga negativa en la cavidad C/F, como ocurre con los mutantes S77 y

-35- RESULTADOS

Y116 no impidió la unión eficiente de los ligandos naturales, independientemente del residuo C-

terminal (Tabla 2). Además, en general, los péptidos con residuos C-terminales apolares se unieron

mejor a los dos mutantes que a B*2705. Una característica importante del mutante Y116 fue el

aumento notable de su preferencia por Leu como residuo C-terminal respecto a Tyr, como se

apreció con los análogos de 02.P4 y 02.P6.(Tabla 3B).

D114 y E152, al contrario que las mutaciones anteriores, suponen la ganancia de una carga

negativa. La unión de los ligandos naturales a estos mutantes mostró un empeoramiento general de

la unión relativa de los péptidos respecto a la unión a B*2705 (Tabla 2).

La doble mutación D114Y116, supone un balance neto de cargas neutro. Este mutante unió

bien los péptidos con residuos C-terminales tanto básicos, como alifáticos o aromáticos, y en

general las afinidades de unión al doble mutante se mostraron similares a las del mutante Y116,

indicando el efecto compensador de esta mutación sobre los efectos adversos de la mutación D114.

PÉPTIDO SECUENCIA B*2705 B*2704 B*2706 S77 Y116 E152 D114 DY

05.P2 RRIKEIVKK 0,8 10 100 0,5 0,1 6 60 0,1 05.P6 GRIDKPILK 2 20 70 1 1 20 5 0,3

05.P5 RRSKEITVR 2 10 30 1 3 20 40 3 05.P8 KRFEGLTQR 2 20 40 0,6 2 >100 90 2

05.P10 RRISGVDRY 3 5 40 0,8 5 >100 40 4 02.P6 KRGILTLKY 4 2 40 0,3 5 30 10 2

02.P2 GRLTKHTKF 4 4 4 0,9 1 5 8 0,4 02.P3 RRFVNVVPTF 2 2 1 0,5 0,4 3 4 0,5

05.P1 RRYQKSTEL 1 1 0,5 1 0,1 5 20 0,2

Tabla 2: Unión de ligandos naturales de B*2705 y B*2702, a subtipos y mutantes de HLA-B27. La unión a cada variante de HLA-B27 está expresada en valores de EC50 (µM) que indica la concentración requerida de péptido para obtener la mitad de la fluorescencia máxima (C50), asociada a HLA-B27 unido al péptido de máxima afinidad entre los péptidos de las series 05.Pi y 02.Pi. Los péptidos con EC50 ≤ 5µM se considera que tienen alta afinidad por ser este el rango en el que se unen los péptidos naturales presentados por B*2702 y B*2705 respectivamente. EC50 ≥ 50 µM indica baja afinidad, 5 µM ≤ EC50 ≤ 50 µM afinidad intermedia, EC50≥100µM, carencia de unión.

RESULTADOS -36-

Tabla 3A

Péptido Secuencia B*2705 B*2704 B*2706 S77 Y116 E152 D114 DY

05.P1 RRYQKSTEL 1 (1) 1 (2) 1 (0,1) 1 (2) 1 (0,05) 1 (3) 1 (20) 1 (0,4) 05.P1Q2 -Q------L 10 2 1 0,5 6 7 5 2,5 05.P1A2 -A------L 90 4 5 3 60 - - 7,5

05.P4 RRWLPAGDA 1 (5) 1 (7) 1 (9) 1 (1) 1 (2) 1 (30) 1 (20) 1 (3) 05.P4Q2 -Q------A - - - 40 20 - 2 3 05.P4A2 -A------A - - 8 60 20 - 2 13

05.P6 GRIDKPILK 1 (0,9) 1 (12) 1 (50) 1 (1) 1 (1) 1 (26) 1 (4) 1 (0,1) 05.P6Q2 -Q------K 11 - - 30 6 - 10 2 05.P6A2 -A------K - - - - - - - - 02.P2 GRLTKHTKF 1 (4) 1 (4) 1 (2) 1(0,9) 1 (0,6) 1 (6) 1 (6) 1 (0,2) 02.P2Q2 -Q------F 18 5 4 6 5 - - 2,5 02.P2A2 -A------F - - 40 8 17 - - 20 02.P4 KRYKSIVKY 1 (3) 1 (6) 1 (2) 1 (0,5) 1 (1) 1 (10) 1 (8) 1 (1) 02.P4Q2 -Q------Y 20 5 25 10 30 4 - 30 02.P4A2 -A------Y - - - - - - - - 02.P6 KRGILTLKY 1 (1) 1 (5) 1 (40) 1 (0,2) 1 (3) 1 (50) 1 (20) 1 (5) 02.P6Q2 -Q------Y - 10 - 40 13 - - - 02.P6A2 -A------Y - - - - - - - -

Tabla 3B

Péptido Secuencia B*2705 B*2704 B*2706 Y116 D114

05.P6 GRIDKPILK 1 (0,9) 1 (12) 1 (50) 05.P6A9 -R------A 8 0,6 0,04 02.P2 GRLTKHTKF 1 (4) 1 (4) 1 (2) 02.P2R9 -R------R 1 5 30 02.P4 KRYKSIVKY 1 (3) 1 (6) 1 (2) 1 (1) 1 (8) 02.P4R9 -R------R 1 1 0,4 5 0,4 02.P4L9 -R------L 0,7 0,3 0,2 0,01 0,4 02.P4A9 -R------A 1 1 0,4 0,04 1 02.P6 KRGILTLKY 1 (1) 1 (5) 1 (40) 1 (3) 1 (20) 02.P6R9 -R------R 1 2 - 3 0,4 02.P6L9 -R------L 0,9 0,6 0,02 0,002 0,3 02.P6A9 -R------A 20 1 0,08 0,2 -

La unión relativa está expresada como el cociente entre el EC50 del correspondiente análogo y el C50 del ligando natural. Los valores de C50 (µM) del péptido sin cambios aparecen indicados entre paréntesis. Los guiones indican no unión del péptido.

Tabla 3: Efecto de los análogos en P2 (Tabla 3A) y P9 (Tabla 3B) en la unión relativa a

subtipos y mutantes.

-37- RESULTADOS

IV.1.2. Unión de péptidos a B*2701 y B*2702. Efecto del polimorfismo sobre la especificidad de los residuos en P2. (Anexo 2)

En este estudio, siguiendo una estrategia similar a la del apartado anterior, se determinó la

especificidad de unión de péptidos in vitro a B*2701 y B*2702. B*2701 es un subtipo poco

frecuente, cuya asociación a enfermedad se desconoce; se diferencia de B*2705 en los cambios

D74Y, D77N y L81A. El primero es un cambio único entre los subtipos conocidos de HLA-B27.

B*2702, que sí está asociado a enfermedad, se diferencia de B*2705 en los siguientes cambios:

D77N, T80I y L81A.

La unión de los péptidos 05.Pi y 02.Pi mostró que tanto B*2701 como B*2702 presentaban

preferencias similares por residuos C-terminales alifáticos (Leu) y aromáticos (Phe y Tyr). Los

residuos C-terminales básicos estaban desfavorecidos en la unión a B*2701 pero B*2702

también unía péptidos con Arg9 (05.P5, 05.P8) con afinidades compatibles con su posible

presentación in vivo (Tabla 4). Estos resultados sugieren que B*2701 y B*2702 comparten con

B*2705 una parte del repertorio peptídico, que incluye sobre todo residuos C-terminales

alifáticos o aromáticos.

Los análogos con Gln2 de los péptidos con residuos C-terminales básicos se unieron mal a

B*2702, pero esos mismos análogos de péptidos con Leu, Tyr o Phe C-terminal (05.P1Q2,

02.P4Q2, 02.P6Q2) se unieron de forma similar a los péptidos naturales, mostrando una mejor

unión relativa a B*2702 que a B*2705 (Tabla 5A). Estos resultados sugieren que los residuos

alifáticos y aromáticos, están más favorecidos en B*2702 que en B*2705, y que los péptidos con

Gln2 podrían formar parte del repertorio natural de B*2702, aunque no han sido aún

encontrados. Todos los análogos con Gln2, independientemente del residuo C-terminal, se

unieron a B*2701 con una eficiencia similar a la del péptido natural, indicando que las

contribuciónes de la Arg2 y de la Gln2 en B*2701 son equiparables.

La preferencia, tanto de B*2701 como de B*2702 por Leu, Phe y Tyr C-terminal, se

confirmó mediante los análogos con cambios en esta posición. La unión relativa de los análogos

con Leu C-terminal (02.P4L9 y 02.P6L9, originalmente Tyr9) mostraron una eficiencia de unión

similar, indicando que tanto Tyr como Leu son igualmente adecuados en ambos subtipos. Sin

embargo, la peor unión a B*2701 de los análogos con Ala9 de esos mismos péptidos, sugiere que

los residuos C-terminales Tyr y Leu contribuyen más fuertemente a la unión en B*2701 que en

B*2702 (Tabla 5B).

En general la unión de los péptidos naturales al mutante Y74 fue buena (Tabla 4), por

RESULTADOS -38-

tanto, la mala unión de algunos péptidos a B*2701 no era debida a esta mutación. Por el

contrario, la unión de los mismos péptidos al mutante A81 fue bastante peor, indicando que esta

mutación es probablemente responsable de la deficiente unión de muchos de los péptidos a

B*2701. El hecho de que algunos de los péptidos se unían bien a B*2702, que también posee el

cambio A81, sugiere la existencia de efectos compensadores de otros cambios, posiblemente I80,

en este subtipo (Tabla 4).

Al igual que con B*2701, los análogos con Gln2 se unieron bien al mutante Y74,

indicando que la preferencia de B*2701 por Gln2 es debida al efecto de esta mutación (Tabla

5A). Debido a que la posición Y74 se localiza fuera de la subcavidad B, este resultado indica que

la especificidad de esta subcavidad es modulada por polimorfismo existente fuera de ella.

Confirmando los resultados de unión in vitro, la secuenciación de los respectivos pooles de

péptidos demostró la presencia, tanto de Arg2 como de Gln2 entre los péptidos presentados in

vivo por B*2701 y por el mutante Y74 (Anexo 2).

Péptido Secuencia B*2705 B*2702 B*2701 Y74 A81

05.P1 RRYQKSTEL 1 8 5 0,7 10 05.P2 RRIKEIVKK 0,8 20 >100 0,8 9 05.P3 RRVKEVVKK 2 50 >100 2 20 05.P11 ARLFGIRAK 5 >100 >100 3 80 05.P6 GRIDKPILK 2 20 >100 1 20 05.P5 RRSKEITVR 2 20 >100 1 80 05.P7 FRYNGLIHR 4 6 40 0,6 20 05.P8 KRFEGLTQR 2 4 30 0,4 30 05.P10 RRISGVDRY 3 8 4 2 40 05.P4 RRWLPAGDA 4 6 >100 3 10 05.P9 RRFTRPEH 20 40 >100 20 70 02.P3 RRFVNVVPTF 2 1 0,9 0,5 2 02.P2 GRLTKHTKF 4 2 10 3 9 02.P4 KRYKSIVKY 4 2 1 1 20 02.P6 KRGILTLKY 4 1 0,6 3 7

Tabla 4: Unión de ligandos naturales de B*2705 y B*2702 a subtipos y mutantes de HLA-B27. Se indican valores de EC50 (µM).( ver pie de la tabla 2)

-39- RESULTADOS

Péptido Secuencia B*2705 B*2702 B*2701 Y74 A81 05.P1 RRYQKSTEL 1 (1) 1 (4) 1 (5) 1 (1) 1 (12) 05.P1Q2 -Q------L 10 1 0,8 0,8 - 05.P1A2 -A------L 90 15 1,4 20 - 05.P6 GRIDKPILK 1 (0,9) 1 (10) 1 (2) 1 (20) 05.P6Q2 -Q------K 11 - 2 4 05.P6A2 -A------K - - - - 05.P7 FRYNGLIHR 1 (2) 1 (5) 1 (20) 1(21) 1 (24) 05.P7Q2 -Q------R 5 8 1,5 2 3 02.P2 GRLTKHTKF 1 (4) 1 (4) 1 (6) 1(0,8) 1 (9) 02.P2Q2 -Q------F 18 7,5 0,5 2,5 - 02.P2A2 -A------F - - - - - 02.P3 RRFVNVVPTF 1 (2) 1 (1) 1 (2) 1(1) 1 (3) 02.P3Q2 -Q-------F 3 3 1 1 10 02.P3A2 -A-------F 35 20 2,5 20 - 02.P4 KRYKSIVKY 1 (3) 1 (4) 1 (2) 1 (1) 1 (10) 02.P4Q2 -Q------Y 20 1 0,4 0,7 9 02.P4A2 -A------Y - 15 2,5 20 - 02.P6 KRGILTLKY 1 (1) 1 (1) 1 (0,7) 1 (1) 1 (15) 02.P6Q2 -Q------Y - 2 0,9 2 - 02.P6A2 -A------Y - 80 29 - -

Péptido Secuencia B*2705 B*2702 B*2701 Y74 A81 05.P6 GRIDKPILK 1 (0,9) 1 (10) 1 (2) 1 (20) 05.P6A9 -R------A 8 0,4 2 1,5 02.P2 GRLTKHTKF 1 (4) 1 (4) 1 (6) 1(0,8) 1 (9) 02.P2R9 -R------R 1 15 - 2,5 1 02.P3 RRFVNVVPTF 1 (2) 1 (1) 1 (2) 1(1) 1 (3) 02.P3R10 -R-------R 2 4 35 2 3 02.P3A10 -R-------A 2 2 3,5 3 7 02.P4 KRYKSIVKY 1 (3) 1 (4) 1 (2) 1 (1) 1 (10) 02.P4R9 -R------R 1 12,5 10 1 1 02.P4L9 -R------L 0,7 1 1,5 0,6 0,3 02.P4A9 -R------A 1 2,5 40 1 1 02.P6 KRGILTLKY 1 (1) 1 (1) 1 (0,7) 1 (1) 1 (15) 02.P6R9 -R------R 1 20 - 2 1,3 02.P6L9 -R------L 0,9 1 1,3 0,7 0,2 02.P6A9 -R------A 20 10 - 10 -

Tabla 5: Unión relativa de varios análogos peptídicos con cambios en las posiciones P2 (Tabla 5A) y P9 (Tabla 5B) a los B*2705, B*2702 y B*2701 y a los mutantes Y74 y A81. La unión relativa está expresada como el cociente entre el EC50 del correspondiente análogo y el C50 del ligando natural. Entre paréntesis aparecen Los valores de C50 (µM) del péptido sin cambios. Los guiones indican no unión del péptido.

Tabla 5A

Tabla 5B

RESULTADOS -40-

IV.1.3. Unión de péptidos a B*2703. Papel del polimorfismo de la subcavidad A. (Anexo 3)

B*2705 y B*2703 se diferencian exclusivamente en el cambio Y59H, localizado en la

subcavidad A. Este residuo está implicado en la estabilización del extremo N-terminal del

péptido, y está conservado entre los subtipos de HLA-B27 a excepción de B*2703.

Para determinar los efectos de esta

mutación en la interacción con péptido,

ésta se estudió bajo dos puntos de vista. En

primer lugar, mediante un ensayo de unión

in vitro se determinaron las eficiencias de

unión a B*2705 y B*2703 de tres péptidos:

(1) RRYQKSTEL, un ligando natural de

ambos subtipos (Jardetzky et al, 1991;

Boisgérault et al, 1996.) y dos análogos del

mismo, (2) ARYQKSTEL y (3)

RQYQKSTEL (Tabla 6).

Unión relativaPéptido Secuencia B*2703 B*2705

(1) RRYQKSTEL 1 (2.10-6) 1 (2.10-6) (2) ARYQKSTEL 1,5 2 (3) RQYQKSTEL >>100 10

Tabla 6: Afinidades de unión de tres péptidos a los

subtipos B*2705 y B*2703. Los valores correspondientes al C50 (µM) del péptido (1), que aparece entre paréntesis, indican la concentración en la cuál se alcanza la mitad de la máxima fluorescencia, determinada mediante citofluorimetría. La unión relativa de los análogos, indica el exceso molar de péptido necesario para alcanzar el valor de C50

Los resultados indican que el cambio

de Arg por Ala en P1 tiene un efecto limitado y cualitativamente similar sobre la unión a ambos

subtipos. Sin embargo, el cambio Arg por Gln en P2 tiene un efecto mucho más dramático sobre

la unión a B*2703 que a B*2705. Esto muestra que el cambio Y59H en B*2703, además de los

efectos en la subcavidad A afecta a las interacciones en la subcavidad B, aumentando la

preferencia por Arg2 con relación a Gln2.

Los efectos de los cambios peptídicos sobre su unión a ambos subtipos, se analizaron

mediante una simulación de dinámica molecular basada en la estructura cristalográfica de HLA-

B27. Esta técnica permite predecir varios parámetros moleculares (distancias entre ligando y el

centro de masas de la proteína, fluctuaciones atómicas, áreas accesibles y no accesibles del

ligando, frecuencias de formación de enlaces de hidrógeno) que ayudan a interpretar las

características de la unión a HLA-B27 observadas experimentalmente (Rognan et al., 1994).

IV.1.3.1. Propiedades dinámicas vs afinidad de unión de péptidos. Mediante la determinación de las variaciones en las distancias entre el péptido y un centro

de masas teórico de la molécula de HLA-B27, puede obtenerse una medida indirecta de la

-41- RESULTADOS

movilidad del péptido y, por tanto, de su estabilidad. Definiendo una serie de distancias: d1:

Distancia proteína-péptido; d2: proteína-(P1, P2, P3), d3: proteína-(P4, P5, P6, P7 y P8); d4:

Distancia entre la subcavidad A y el residuo P1, d5: Distancia subcavidad B-residuo P2; d6:

Distancia subcavidad D-residuo P3 y d7: Distancia subcavidad F-residuo P9, la modelización

permite determinar lo que sucede en la interacción de los diferentes péptidos a ambos subtipos.

Cuantitativamente, la parte central del péptido expuesta al TCR (residuos P4-P8) sufrió un

incremento de las distancias, similar en los tres péptidos unidos a ambos subtipos (Anexo 3,

Tabla 2: d3). El análisis de las distancias d4-d7 entre los anclajes y sus subcavidades

complementarias A, B, D y F mostró que el péptido (3), tanto unido a B*2703 como a B*2705

experimentaba una repulsión localizada en el residuo P9 (d7), muy alejada del lugar de la

mutación en la subcavidad A, así como una mayor distancia d5 para el complejo RQYQKSTEL-

B*2703. La distancia d4, era dependiente del residuo presente en P1, aunque para un mismo P1

siempre fue mayor en el subtipo B*2703. Por último, las distancias d6 fueron más variables

demostrando la flexibilidad de ésta (Anexo 3, Tabla2).

Estos datos indican que la mutación Tyr→His59 desestabiliza la subcavidad A, de modo

que en estas circunstancias, un mal anclaje en P2 causa una inestabilidad global del péptido en

varias posiciones, y de forma especial en su extremo C-terminal.

IV.1.3.2. Fluctuaciones atómicas, áreas accesibles y no accesibles. Como era de esperar, la parte central del péptido expuesta al TCR mostró mayores

fluctuaciones que las posiciones de anclaje del péptido. En concordancia con los valores de

afinidad de unión in vitro, la comparación de las movilidades atómicas de los péptidos (1) y (2)

unidos a ambos subtipos no mostró diferencias significativas, pero la flexibilidad del péptido (3)

se incrementó, especialmente en el complejo de menor afinidad RQYQKSTEL-B*2703 (Anexo 3:

Figura 3). La magnitud de las fluctuaciones en el residuo P9, indican que cuando el anclaje en

P2 es débil, la inestabilidad se distribuye a otras posiciones de anclaje incluída, de forma

prominente, la posición C-terminal.

En consonancia con el resultado anterior, el análisis de las áreas accesibles frente a las no

accesibles, nuevamente mostró que la mayor accesibilidad de P9, se daba en el complejo

RQYQKSTEL-B*2703 (Anexo 3: Figura 4). Sin embargo, a pesar de la flexibilidad que mostró la

Gln2 (Anexo 3: Figura 3), este residuo permaneció inaccesible sugiriendo que, aunque la Gln2

no es por si misma un buen residuo de anclaje, aún conserva interacciones en la subcavidad B.

RESULTADOS -42-

IV.1.3.3. Análisis cualitativo y cuantitativo de los puentes de hidrógeno La estabilidad de los puentes de hidrógeno establecidos entre HLA-B27 y el péptido, se

determinó mediante el cálculo de la frecuencia de formación de dichos enlaces. Se consideró

como un puente de hidrógeno fuerte aquél con una frecuencia mayor del 50%, medio cuando la

frecuencia estaba entre el 25% y el 50%, y débil si era inferior al 25%. Con estas premisas, el

número, distribución y frecuencia de los puentes de hidrógeno permitió distinguir cualitativa y

cuantitativamente, los péptidos (1) RRYQKSTEL, (2) ARYQKSTEL y (3) RQYQKSTEL (Anexo 3:

Figura 5). La modelización predijo un número de 25 enlaces para los péptidos (1) y (2) unidos a

ambos subtipos, y de la mitad cuando el péptido unido era (3). Asimismo, la distribución de los

enlaces débiles y fuertes se correspondió con las afinidades de unión in vitro, de forma que los

complejos formados por los péptidos (1) y (2) mostraron un número similar de enlaces fuertes en

ambos subtipos frente al reducido número de enlaces medios o fuertes encontrados para el

péptido (3). Estos resultados indican que la afinidad de unión de un péptido está directamente

relacionada con la cantidad y calidad de los enlaces de hidrógeno que establece el péptido con

diferentes residuos de la molécula de clase I.

-43- RESULTADOS

IV.2. RELACIÓN ENTRE LA UNIÓN DE PÉPTIDOS Y LA SELECCIÓN DE EPÍTOPOS VIRALES POR CÉLULAS T. (Anexo 4)

El polimorfismo de las moléculas de clase I tiene una función clave en el reconocimiento

por las células T de los complejos MHC-Péptido, puesto que determina la presentación o no de

un péptido antigénico, modula su afinidad y estabilidad de unión, y puede alterar la

conformación final del epítopo reconocible por el TCR. En este estudio se trató de determinar el

papel que juega el polimorfismo de HLA-B27, en la presentación y en el reconocimiento por

células T de péptidos derivados del virus de Epstein-Barr (EBV), que son inmunogénicos sólo en

el contexto de determinados subtipos de HLA-B27.

IV.2.1. Unión de péptidos virales a diferentes subtipos de HLA-B27, y su relación con la

inmunogenicidad.

Se estudió la unión de tres péptidos derivados del EBV, restringidos por diferentes

subtipos de HLA-B27. EBNA3C (258-266) (Brooks et al., 1993) es restringido por los subtipos

B*2705 / B*2702 / B*2704; LMP2 (236-244) es restringido por B*2704 (Brooks et al., 1993) y

EBNA3B (243-253) es restringido por B*2702 (Brooks et al., 1998) (Tabla 7). Aunque todos se

unieron a los tres subtipos, EBNA3B se unió mejor a B*2705 que a B*2702, y LMP2 se unió con

igual afinidad a B*2705 y B*2704. Por tanto, no existe una correlación entre la eficiencia de

unión in vitro de un péptido viral a un subtipo de HLA-B27, y su inmunogenicidad en el

contexto de dicho subtipo. La unión relativa tampoco se correspondió con sus patrones de

restricción; así en el caso de LMP2 la unión a B*2702 fue 30 veces mejor que la unión de

EBNA3B y 7,5 veces mejor que la unión de EBNA3C a B*2705 (Tabla 7). Estos resultados

indican que la unión y la inmunogenicidad de un péptido, están determinadas por factores

distintos al de su afinidad por el subtipo HLA que actúa como elemento de restricción.

Los valores entre paréntesis corresponden al C50 µM del péptido que mejor unión presentaba (LMP2). El resto de valores corresponde al exceso molar necesario de péptido al necesario para alcanzar el valor de C50 .

Péptido Restricción Secuencia B*2705 B*2702 B*2704

LMP2 B*2704 RRRWRRLTV 1 (0,4 µM) 1 (1 µM) 1 (0,4 µM)

EBNA3C B*2705 / 02 / 04 RRIYDLIEL 7,5 3 12,5 EBNA3B B*2702 RRARSLSAERY 15 30 50

Tabla 7: Unión de péptidos de EBV a tres subtipos de HLA-B27 que actúan como

elementos de restricción.

RESULTADOS -44-

La unión de los tres péptidos virales a otros subtipos de HLA-B27 (B*2705, B*2703 y

B*2706) fue buena en general (Anexo 4, Tabla 3), lo que indica que el polimorfismo entre los

subtipos B*2701-B*2706, no afecta críticamente a la unión de los péptidos estudiados. Sin

embargo, para LMP2 las diferencias de afinidad por diferentes subtipos se hicieron más patentes

mediante el uso de análogos con Ala2 (Anexo 4, Tabla4). La unión relativa de LMP2A2 fue más

baja para B*2705 / 03 / 01 que para su elemento de restricción (B*2704) y para B*2706 y

B*2702. Este resultado confirma que la inmunogenicidad de LMP2, exclusivamente en B*2704

no está relacionada con su mayor afinidad por este subtipo.

IV.2.2. Reconocimiento de péptidos de EBV por CTLs restringidos por HLA-B27. En estos experimentos se analizó la capacidad de CTLs específicos de péptido, y

restringidos por un subtipo determinado, para reconocer dicho péptido en el contexto de otros

subtipos. La capacidad de los tres péptidos de EBV, para ser reconocidos por diferentes CTLs

restringidos por B*2705, B*2704 y B*2702 se resume en la Tabla 8.

De 10 clones restringidos por B*2705 específicos de EBNA3C, tres reconocieron al péptido

también en el contexto de B*2702. De igual forma, cuatro de los cinco clones restringidos por

B*2702 contra el mismo péptido lo reconocieron en el contexto de B*2705, pero ninguno de los

quince lo reconocieron unido a B*2704. Los clones restringidos por B*2702 específicos de

EBNA3B, también reconocían al péptido, unido tanto a B*2702 como a B*2705 pero no a

B*2704. De los cuatro clones restringidos por B*2704 específicos de LMP2, ninguno reconoció

al péptido unido ni a B*2705 ni a B*2702. Estos resultados muestran que B*2705 y B*2702,

pueden actuar como elementos de restricción equivalentes, al menos para algunos CTLs. Sin

embargo, B*2704 muestra un grado mucho mayor grado de disparidad funcional.

IV.2.3. El motivo Arg2, anclaje principal de los péptidos unidos a HLA-B27, no es esencial

para mantener la capacidad antigénica del péptido. En estos experimentos se analizó el papel del residuo principal de anclaje de los péptidos

en HLA-B27, en mantener la capacidad antigénica del epítopo viral.

El análisis realizado con el péptido EBNA3C y sus análogos con Gln2 y Ala2 mostró

diferencias en la afinidad de unión a los subtipos que actuan como elementos de restricción para

este péptido, en función del residuo presente en P2 y del subtipo (Figura 7A). No obstante, los

CTLs restringidos por B*2705 reconocieron tanto al péptido como a sus análogos. Los CTLs

-45- RESULTADOS

restringidos por B*2702 reconocieron al análogo con Ala2 mucho peor.

El mismo análisis realizado con los análogos del péptido EBNA3B y sus análogos en P2

mostró una unión similar del análogo con Gln2 a B*2702, sin embargo el análogo con Ala2

aunque se unió bastante peor, fue reconocido con igual eficacia por los CTLs restringidos por

B*2702 específicos de EBNA3B. (Figura 7B).

Estos resultados mostraron que el residuo de anclaje principal Arg2 es un requisito

fundamental para obtener una buena unión los péptidos a HLA-B27, pero no es esencial para

mantener la estructura antigénica del péptido.

RESULTADOS -46-

Clon Donante Célula Diana Nº clones

B*2705 B*2702 B*2704 EBNA3C

RTc10 B*2705 65 3 2 7 RTc37 B*2705 51 60 1 3

Total 10 EBNA3C

LYc39 B*2702 9 55 0 1 Kor c69 B*2702 44 53 0 4

Total 5 EBNA3B

LYc40 B*2702 72 76 0 NWc20 B*2702 50 64 0 Total 4

LMP2

DHc21 B*2704 0 3 35 4 Total 4

Figura 7:

Unión y reconocimiento de dos péptidos virales y sus análogos en P2 por CTLs específicos de péptido. A/ Arriba; unión del péptido EBNA3C y sus análogos, a los tres subtipos que actúan como sus elementos de restricción. Debajo, a la izquierda, lísis de una línea celular linfoblastoide autóloga, por un CTL específico del péptido, provenientes de un donante B*2702+(LYc25), y sensibilizada a varias concentraciones con los péptidos anteriores. Relación Efector : Diana=4:1. A la derecha un experimento similar en el que diana y efector (Alc12) eran de un individuo B*2705.+ Relación E:D=2:1.

B/ Unión del péptido EBNA3B y sus análogos al subtipo que actúa como su elemento de restricción, (B*2702). A la izquierda, se muestra la lisis de una línea celular linfoblastoide autóloga por el clon restringido y específico de péptido LYc29. Relación E:D = 4:1.

PÉPTIDO SECUENCIA B*2702 B*2705 B*2704

EBNA3C

EBNA3CA2

EBNA3CQ2

RRIYDLIEL

-A-------

-Q-------

3

−

20

3

−

100

5

−

−

Concentración de péptido (M)

LY c25

10-11 10-10 10-9 10-8 10-7 10-6 10-5

0

20

40

60

80 EBNA 3C EBNA 3CA2 EBNA 3CQ2 SIN PÉPTIDO

0

20

40

60

80

100

10-11 10-10 10-9 10-8 10-7 10-6 10-5

AL c12

% d

e lis

is e

spec

ífica

Concentración de péptido (M)

SIN PÉPTIDO

10-11 10-10 10-9 10-8 10-7 10-6 10-5

0

10

20

30

40

50LY c29 EBNA 3B

EBNA 3BA2 EBNA 3BQ2

% d

e lis

is e

spec

ífica

PÉPTIDO

EBNA3BEBNA3BA2EBNA3BQ2

SECUENCIA

RRARSLSAERY

-A---------

-Q---------

B*2702

20100 20

A

B

Tabla 8: Reconocimiento por CTLs restringidos por HLA-B27 de péptidos derivados de EBV. Los valores indican el % de lisis específica a una relación Efector:Diana de 4:1. Las células diana se incubaron durante 30 minutos con una concentración de péptido 10-6M. En todos los casos el fondo fue inferior al 5%. En la columna derecha aparece el número de clones que tenían el mismo patrón de reactividad. En negrita y para mayor claridad, se marcan los casos en los que la reactividad es claramente positiva.

-47- RESULTADOS

IV.3. MODULACIÓN DE LA ESPECIFICIDAD EN LAS POSICIONES DE ANCLAJE P1, P3 Y PΩ, POR EL POLIMORFISMO DE HLA-B27. (Anexo 5)

IV.3.1. Especificidad de B*2705, B*2704 y B*2706 por los residuos en P1, P3 y P9.

Para determinar la contribución de las posiciones de anclaje secundario P1 y P3 y del

anclaje primario P9, a la unión de péptidos a HLA-B27, se utilizaron tres series de nonámeros de

poli-Alanina en los que conservando el anclaje principal Arg2, cada una de las tres posiciones

fue sustituida por diferentes aminoácidos. La unión de cada análogo (EC50), se cuantificó con

base a la unión del péptido control RRYQKSTEL, un ligando natural de B*2705, B*2704 y

B*2706. La contribución relativa de cada residuo, se calculó como la unión relativa del análogo

correspondiente respecto al péptido ARAAAAAAA (ARA7) (Figs. 8A, 8B y 8C). Este péptido, se

unió con más afinidad a B*2704 y B*2706 (10 y 9 µM respectivamente) que a B*2705 (30 µM)

lo que significa que el esqueleto de poli-Alanina interacciona más fuertemente en los dos

primeros.

Residuo P1:

El residuo mas favorecido en B*2705 fue R (EC50 de RRA7= 9 µM), cuya contribución al

anclaje respecto a Ala1 fue aproximadamente 3. K, H, G, I, M y los residuos aromáticos fueron

aproximadamente equivalentes a Ala y los residuos ácidos, polares, y L y V redujeron

claramente la afinidad.

En general, la especificidad de B*2704 por residuos en P1 fue similar a B*2705 en lo

referente al efecto negativo de los residuos ácidos, polares y alifáticos. Una diferencia fue la

mayor preferencia de este subtipo por G (EC50 de GRA7= 6 µM) que por R.

B*2706 se diferenció claramente de B*2704, en que sólo D estaba muy desfavorecido. La

contribución a la unión de los residuos básicos R, K, H, los polares, Q, G y Y fue mayor en

B*2706 que en los otros dos subtipos. Estos resultados indican que B*2705, B*2704 y B*2706 a

pesar de tener una misma subcavidad A, muestran diferencias en sus preferencias por el residuo

P1.

Residuo P3:

Los residuos en esta posición mostraron un amplio rango de afinidades de unión a B*2705.

El más favorecido fue W (EC50 =5µM), y en general se apreció una preferencia por los residuos

apolares. H, N, y Y fueron equivalentes a Ala y los residuos ácidos, T, Q, G y P estaban

desfavorecidos.

RESULTADOS -48-

B*2704 se asemejó a B*2705 en general. Las principales diferencias con B*2705 fueron

que Ala no era peor que otros residuos alifáticos de mayor tamaño y que N era peor que Ala en

B*2704, pero no en B*2705.

La especificidad de B*2706 por los residuos en P3 mostró importantes diferencias con

B*2704. Así, H, R, y los residuos polares estaban especialmente favorecidos respecto a Ala en

B*2706, y G, P y los residuos ácidos estaban menos desfavorecidos. Además, los residuos

apolares alifáticos y los aromáticos eran más adecuados que Ala, pero Y estaba menos

favorecida que muchos otros residuos, entre los que se incluían los aromáticos. Ala por si misma

estaba entre los mejores residuos de B*2704, pero era peor que muchos otros en B*2706.

Estos resultados indican que B*2704 y B*2706 difieren entre si, y también de B*2705, en las

preferencias por el residuo P3.

Residuo P9:

El número de aminoácidos que se estudió en esta posición se limitó a los residuos básicos

y apolares, por ser este el tipo de residuos C-terminales mayoritariamente presentado por

B*2705. Adicionalmente se incluyó Pro, un residuo C-terminal presente entre los péptidos de

HLA-B73, que muestra la misma preferencia que HLA-B27 por Arg2.

Como era esperado, tanto los residuos básicos como los apolares se unieron en general

bien a B*2705, pero no de forma equivalente. Entre los residuos aromáticos Y estaba favorecido,

pero no F ni W. Asimismo, P estaba desfavorecida.

En B*2704, F, W y P también estaban desfavorecidos, además de los residuos básicos y Y.

Por último, en B*2706, y al igual que en B*2704, ni los residuos básicos, ni Y, W o P eran

adecuados, pero B*2706 mostró una mayor preferencia que B*2704 por los residuos alifáticos y

F. Estos resultados indican que una de las principales diferencias entre la especificidad de

B*2704 y B*2706 por los residuos C-terminales, radica en la mejor aceptación de F por B*2706.

IV.3.2. La unión de un péptido es el resultado de la contribución aditiva de varios residuos de anclaje.

Para determinar si la afinidad de unión de un péptido es simplemente la suma de las

contribuciones individuales de cada uno de los residuos de anclaje, o está influida por efectos

cooperativos más complejos, se analizó la unión de dos ligandos naturales RRYQKSTEL y

KRYKSIVKY y de distintos análogos de poli-Alanina en los que se conservaron una o varias de

las posiciones P1, P3, P7 y P9. Para calcular el EC50 de todos los análogos se utilizó como

-49- RESULTADOS

referencia el péptido RRYQKSTEL

Primeramente, se midió en qué medida la introducción de esas posiciones de anclaje

incrementaba o hacía decrecer la unión respecto a ARA7. Después se calculó la relación entre la

unión relativa de cada ligando, o análogo portador de multiples posiciones de anclaje del ligando

natural, y la suma de las uniones relativas de cada análogo de poli-Alanina, con uno sólo de los

residuos de anclaje. Si la contribución de los residuos individuales es aditiva, esta relación debe

ser 1. Si existen efectos interactivos entre los residuos de anclaje, esta relación se desviará de

forma significativa de ese valor. Para tener en cuenta el error experimental, los valores de esa

relación entre 1,5 (1,5:1) y 0,67 (1:1,5) se consideró que implicaban contribución aditiva.

Como se aprecia en la (Figura 8D_a), el EC50 de la mayor parte de los péptidos portadores

de varios anclajes del péptido RRYQKSTEL, (P1, P3, P7, P9) es resultado, en cuatro de cinco

casos, del valor aditivo de sus correspondientes análogos portadores de un simple residuo. De

manera similar el EC50 de los análogos con múltiples posiciones de anclaje del péptido

KRYKSIVKY es, en cuatro de cinco casos, el resultado de la contribución de los análogos con

uno o pocos residuos de anclaje (Figura 8D_b), con un único caso (KRAAAAAAY, relación 0,6)

que mostró una ligera desviación del rango 0,67-1,5.

Estos resultados indican que en general, la eficiencia de unión de un péptido es

simplemente una función aditiva de la contribución de los anclajes individuales; no obstante los

efectos interactivos mútuos de las cadenas laterales pueden en ocasiones afectar a la unión.

El papel que juegan las posiciones P4, P5, P6 y P8, se determinó a partir de la relacion

EC50(ligando natural)/EC50(análogoP1-P2-P3-P4-P9) que mostró unos valores de 10/15=0,67 y

7,5/4,3=1,7 para RRYQKSTEL y KRYKSIVKY respectivamente. Este resultado muestra que su

contribución a la unión es variable y dependiente del péptido.

RESULTADOS -50-

Leyendas para las figuras 8A,8B,8C y 8D. (Ver página siguiente). Figuras 8A, 8B y 8C. Unión relativa de diferentes análogos de poli-Alanina a B*2705, B*2704 y B*2706.: Cada análogo aparece representado en el eje de abscisas por el código de una letra del residuo introducido en la secuencia del péptido ARA7 en posición P1, P3 o P9. El péptido de referencia RRYQKSTEL (EC50: 8A=3 µM; 8B=4 µM; 8C=0,3 µM ) está marcado con un asterisco (*). La unión relativa de cada análogo (eje de ordenadas, en escala logarítmica) indica la relación entre el EC50 de ARA7 (8A=30 µM; 8B=10 µM; 8C=9 µM) y el del análogo correspondiente. Debido a que la concentración máxima que se utilizó de péptido fue 100 µM, la unión relativa menor que 0.3, que indica carencia de unión en este ensayo (EC50>100 µM), no puede medirse. En la gráfica se le asignó un valor de 0,25 por motivos de representación.

Figura 8D. Relación entre la unión de los péptidos (a) RRYQKSTEL y (b) KRYKSIVKY. con distintos análogos en las posiciones de anclaje: Las barras negras indican la unión relativa respecto al análogo ARA7, expresada como la relación molar entre el EC50 de ARA7 y el de cada péptido (abscisas). En los residuos cuyo efecto fue detrimental respecto a Ala (T7, K1), el descenso de la unión relativa se expresó como: 1-unión relativa. Las barras blancas indican el valor aditivo de las uniones relativas de los análogos de poli-Ala portadoras de sustituciones sencillas en P1, P3, P7 o P9. El efecto de V7 no se analizó de forma separada; en su lugar se utilizó el análogo ARAAAAVAY. Teniendo en cuenta el error experimental en la determinación de los valores de EC50, los valores representados por las barras negras y blancas se consideraron iguales, cuando su relación se encontraba entre 0.67 (1:1.5) y 1.5 (1.5:1). Los valores que excedieron esos margenes aparecen marcados con un asterísco (*).

-51- RESULTADOS

B*2704

0,1

1

10

K H R D E S T N Q G A V I L M F Y W *

0,1

1

10

H R D E S T N Q G A V I L M F W P *

0,1

1

10

K R A V I L M F Y W P *

bARAAAAVAY

ARYAAAAAAARAAAAAAY

KRAAAAAAAKRYKSIVKY

KRYAAAVAY

KRYAAAAAA

KRAAAAAAY

ARYAAAAAY

ARYAAAVAY

0 2 4 6 8 10 12 14 16

a

0 2 4 6 8 10 12 14 16

ARYAAAAAAARAAAATAAARAAAAAAL

RRAAAAAAARRYQKSTEL

RRYAAATAL

ARYAAATAL

ARAAAATAL

ARYAAAAAL

ARYAAATAA

P1

P3

P9

P1

P3

P9

0,1

1

10

100

K H R D E S T N Q G A V I L M F Y W *

0,1

1

10

100

H R D E S T N Q G A V I L M F Y W P *

0,1

1

10

100

K R A V I L M F Y W P *

B*2706

0,1

1

10

K R A V I L M F Y W P *

0,1

1

10

H R D E S T N Q G A V I L M F Y W P *

0,1

1

10

K H R D E S T N Q G A V I L M F Y W *

B*2705 A B

C D

Y

*

*

RESULTADOS -52-

IV.3.3. Distribución de los residuos P1, P3 y PΩ entre los ligandos naturales de B*2705.

Conocida la contribución al anclaje de cada residuo, se procedió al estudio comparativo de

su distribución entre 54 ligandos naturales de B*2705 (Tabla 10), con el propósito de determinar

las reglas que rigen el uso de residuos en las posiciones de anclaje, por el repertorio peptídico

natural de este subtipo. En función de la eficacia de unión de cada uno de los residuos, respecto

al péptido de referencia ARA7, se establecieron cinco niveles, siendo el nivel 1 el de los residuos

más favorecidos. (Tabla 9).

Rango Nivel P1 P3 PΩ <10 µΜ 1 86% ( 44/51 )

11-20 µΜ 2 44% ( 24/54 ) 85% ( 45/53 ) -

21-40 µΜ 3 (ARA7) 41-80 µΜ 4 14% ( 7/51 )

>80 µΜ 5 48% ( 26/54 ) 15% ( 8/53 )

- Tabla 9: Distribución porcentual de los residuos P1, P3 y P9 entre los ligandos naturales de B*2705.

Los resultados indican que las posiciones P3 y P9, utilizan residuos favorables (niveles 1 y

2) en el 85% de los casos, pero pueden acomodar también residuos desfavorables. La posición

P1, es más permisiva y acomoda residuos desfavorables (niveles 4 y 5) en el 48% de los casos.

Adicionalmente se siguen las siguientes normas de uso:

Entre los, nonámeros y decámeros con PΩ subóptimo (niveles 3 a 5) (cinco nonámeros y

dos decámeros) todos tienen un anclaje óptimo en P3 (nivel 1), independientemente del residuo

presente en P1. De igual modo, todos los péptidos de estos tamaños, con un anclaje subóptimo

en P3 (cinco nonámeros y tres decámeros) tienen un PΩ bueno (nivel 1), independientemente del

residuo presente en P1. Este resultado indica que la presencia de un anclaje subóptimo en P3 o

en P9, siempre se compensa con un buen anclaje en la otra posicion. Dieciocho de los 30

péptidos con un mal residuo de anclaje en P1 (niveles de 3 o >3), tienen tanto en P3 como en P9

un buen residuo de anclaje (niveles 1 y 2) y todos, como mínimo, un buen anclaje al menos en

una de estas dos posiciones. Siete de los péptidos, que muestran un mal anclaje en P1 y P3 o P1

y P9 tienen en el residuo restante uno de nivel 1. Por lo tanto, un mal residuo de anclaje en P1

requiere un residuo óptimo al menos en una de las otras dos posiciones.

Entre los péptidos de tamaño no-canónico (octámeros, undecámeros y dodecámeros), las

-53- RESULTADOS

tres posiciones estudiadas mostraron buenos anclajes (niveles 1 y 2), lo que sugiere que este tipo

de ligandos, es más restrictivo en el tipo de residuo que puede ocupar sus posiciones P1, P3 y

PΩ, aunque el número de ligandos conocidos de estos tamaños es aún limitado.

Estas reglas, estudiadas en B*2705 tambien se cumplen con los escasos ligandos conocidos de

B*2704 y B*2706 (Anexo 5, Tabla 2).

RESULTADOS -54-

Secuencia Valoración del residuo Referencia. Secuencia Valoración del

residuo Referencia.

Octámeros Decámeros

RRFFPYYV 1-1----1 [1] KRFEETGQEL 4-1------1 -Este estudio- RRFTRPEH 1-1----X [2] NRFAGFGIGL 5-1------1 -Este estudio-

RRQDILDLWI 1-5------1 HIV (*) Nonámeros GRFNGQFKTY 4-1------1 [7]

RRYDRKQSGY 1-2------1 [6] ARLQTALLV 3-1-----1 -Este estudiob GRWPGSSLYY 4-1------1 [6] RRYQKSTEL 1-2-----1 [2] GRKTGQAPGY 4-X------1 [6] SRTPYHVNL 4-4-----1 -Este estudio- GRILSGVVTK 4-2------1 [6] RRLPIFSRL 1-1-----1 [3] RKGGNNKLIK 1-5------1 [6] GRHGVFLEL 4-2-----1 -Este estudio- LRDNIQGITK 5-5------1 -Este estudio- RRIYDLIEL 1-2-----1 [4] (*) KRWIILGLNK 4-1------1 HIVd (*) RRYPDAVYL 1-2-----1 [5] (*) RRFVNVVPTF 1-1------4 -Este estudio- GRFGSGMNM 4-1-----1 [6] KRWQAIYKQF 4-1------4 [6] GRTFIQPNM 4-4-----1 [6] RRIKEIVKKH 1-2------X [3] LRFQSSAVM 5-1-----1 [6] RRSKEITVR 1-3-----1 [2] Undecámeros FRYNGLIHR 3-2-----1 [2] KRFEGLTQR 4-1-----1 [2] RRYLENGKETL 1-2-------1 [7] HRAQVIYTR 3-3-----1 [6] RRMGPPVGGHR 1-1-------1 [3] SRYWAIRTR 4-2-----1 [2] (*) WRLGSSDILNY 2-1-------1 -Este estudio- RRFMPYYVY 1-1-----1 [3] SRVKLILEY 4-2-----1 -Este estudio- Dodecámeros RRFFPYYVY 1-1-----1 [1, 7] RRVLVQVSY 1-2-----1 [6] RRFVNVVPTFGK 1-1--------1 [6] RRISGVDRY 1-2-----1 [2, 3]c RRIKEIVKK 1-2-----1 [2] RRVKEVVKK 1-2-----1 [2] ARLFGIRAK 3-1-----1 [2,3] GRIDKPILK 4-2-----1 [2] GRFEGTSTK 4-1-----1 [6] GRAFVTIGK 4-3-----1 HIV (*) IRLRPGGKK 4-1-----1 HIV (*) RRWLPAGDA 1-1-----3 [2] GRLTKHTKF 4-1-----4 [3, 7] KRFKEANNF 4-1-----4 -Este estudio-RRFGDKLNF 1-1-----4 [3] KRFSFKKSF 4-1-----4 [3] TRYPILAGH 5-2-----X [3] KRVVINKDT 4-2-----X [8] (*)

(b) Descrito previamente como octámero (ARLQTALL) por secuenciación de Edman [Rötzschke et al., 1994]. También encontrado como nonámero en B*2709 (ARLQTALLV) [Fiorillo et al., 1997]. (c) Descrito también como nonámero en B*2701 [García et al., 1997c] y como decámero (RRISGVDRYY) en B*2703 [Boisgérault et al., 1996 ] y B*2710 [García et al., 1998]. (d) Existe también una variante natural de este péptido con el cambio L6M. [1] Paradela et al., 1998. [2] Jardetzky et al., 1991. [3] Rötzschke et al., 1994. [4] Brooks et al., 1993. [5] van Binnendijk et al., 1993. [6] Fiorillo et al., 1997. [7] Boisgérault et al., 1996. [8] Ugrinovic et al., 1997.

Tabla 10: Ligandos naturales de B*2705 y valoración de los residuos P1, P3 y PΩ. a

(a) Los péptidos marcados con un asterisco son de origen bacteriano o viral (*). El resto pertenecen a proteínas endógenas celulares. En negrita se indican los valores asignados a cada residuo. X: valoración no analizada. Los péptidos derivados del HIV se obtyuvieron del HIV Molecular Immunology Database of Los Angeles National Laboratory. (http://hiv-web.lanl.gov/).

-55- RESULTADOS

IV.4. UNIÓN DE ANÁLOGOS NO PEPTÍDICOS A HLA-B27. (Anexo 7)

Como una primera aproximación al posible uso de ligandos no-peptídicos como

antagonistas en la respuesta inmune mediada por CTLs, en este estudio se analizaron las

propiedades de unión de varios de estos

ligandos a HLA-B27. Para ello se alteró la

región central, que comprende los residuos P4-

P8, en varios ligandos naturales de B*2705

(Tabla 11). Se pretendía con ello, conservar

las propiedades de unión del péptido natural a

la molécula de clase I, y simultaneamente

alterar su región central, reconocida por el

TCR.

En los análogos no peptídicos utilizados,

la parte central fue sustituida por espaciadores

no peptídicos de diferente naturaleza química

y longitud (Figura 9). El diseño de estos ligandos se realizó por técnicas de simulación de

dinámica molecular, y su unión a HLA-B27 se analizó mediante un ensayo de estabilización de

péptidos in vitro. Asimismo, la eficiencia de unión a B*2705, medida por este ensayo, se

compararó con el parámetro de desnaturalización térmica (Tm) obtenido mediante dicroísmo

circular.

Figura 9: Estructura química de los tres espaciadores no peptídicos utilizados en los análogos. Aua: ácido 11-amino undecanoico. HB: (R)-3-Hidroxibutirato.

Aua: 11-Amino undecanoico.H2N OH

O OOHB3

O OO

OOOO

O OOOHB4

IV.4.1. Reemplazamiento de la parte central de epítopos naturales con un espaciador

monofuncional.

El ácido 11-amino undecanoico (Aua) es un espaciador monofuncional, que únicamente

enlaza los residuos P3 y P9, pero no establece interacciónes con la molécula de clase I por

carecer de cadenas laterales. En los cuatro análogos analizados con este espaciador, se apreció

una pérdida general de afinidad respecto al péptido natural, tanto en el ensayo de unión in vitro

como en su estabilidad térmica. Sin embargo las diferencias observadas por este último criterio

fueron más acusadas (Tabla 11).

La unión de los análogos de -Aua- a B*2704, fue muy ineficiente cuando el residuo P9 era

básico, pero apenas se alteró cuando dicho residuo era apolar. Este resultado indica la

importancia de un buen anclaje en P9 para la unión del análogo con este tipo de espaciador, y el

RESULTADOS -56-

papel decisivo de las interacciones secundarias entre el péptido y B*2704, cuando P9 es un mal

anclaje para este subtipo.

IV.4.2. Reemplazamiento de la parte central de epítopos naturales con espaciadores

bifuncionales.

En base a los resultados anteriores, se ensayaron dos tipos de espaciadores bifuncionales

formados por oligómeros de (R)-3-hidroxibutirato (HB), uno trimérico -HB3- y otro tetramérico

-HB4. Estos espaciadores, además de enlazar P3 y P9 forman polímeros químicamente estables, y

pueden adoptar plegamientos conformacionalmente similares a los de los péptidos en estado

libre (Plattner et al., 1993). Adicionalmente, poseen grupos metilo capaces de interaccionar en

las subcavidades del sitio de unión del péptido. Con estos espaciadores se construyeron los

correspondientes análogos del ligando de HLA-B27 QRLKEAAEK, y una variante con Ala1,

ARLKEAAEK, con objeto de paliar posibles problemas en el ensayo de unión derivados de la

eventual ciclación de la Gln N-terminal.

Ambos análogos mostraron eficiencias de unión in vitro muy diferentes (Tabla 11). El

análogo (12) con -HB4- se unió 15 veces mejor que el (11) con -HB3-. La presencia del

espaciador -HB4- siempre mejoró la unión respecto al péptido natural (7). Sin embargo, los

valores de Tm que mostraron los análogos estudiados fueron similares (Péptidos 11 y 12, Tabla

11). Estos resultados indican que la introducción de grupos -HB4- en la parte central del

espaciador, mejora la unión a B*2705 considerablemente, pero no afecta de forma significativa a

su estabilidad térmica.

IV.4.3. Modelado molecular de la unión de análogos con espaciadores no peptídicos a

B*2705. Mediante la modelización, se predicen diferencias claras entre el péptido natural

QRLKEAAEK y los análogos con espaciador no peptídico (Anexo 7, Figuras 2 y 5). En primer

lugar, el número de contactos establecidos por el análogo con -Aua- era menor, lo que explica la

menor afinidad de los análogos con este espaciador respecto al péptido natural. La menor

longitud del análogo con -HB3- impide la interaccion óptima con las subcavidades A y F, lo que

no sucede con el análogo con -HB4- cuya mayor longitud le permitía incrementar el número de

interacciones en ambos extremos, que son suplementadas por los contactos adicionales de los

grupos metilo. Por tanto, la modelización teórica de las interacciones con HLA-B27 confirmó, al

-57- RESULTADOS

menos cualitativamente, las diferencias experimentales y proporcionó una posible interpretación

molecular de estos resultados.

Secuencia EC50 (a)

Péptido P1-P2-P3 -Espaciador- P9 B*2705 B*2704

Tm (ºC)(b)

1(c) RRR WRRLT V 1,2 0,8 52,8 ± 0,7 2 RRR -Aua- V 4 1 42,9 ± 0,3

3(d) SRY WAIRT R 3 1,4 46,3 ± 0,5 4 SRY -Aua- R 8,6 100 39,5 ± 0,2

5(e) GRA FVTIG K 1,8 6,4 61,9 ± 0,3 6 GRA -Aua- K 7 >100 48,1 ± 0,4

7(f) QRL KEAAE K 10 62,8 ± 0,7 8 QRL -Aua- K 46,5 ± 0,2 9 QRL -HB3- K 40 10 QRL -HB4- K 2,5

11 ARL -HB3- K 20 63,2 ± 0,6 12 ARL -HB4- K 1,6 62,1 ± 0,7

Tabla 11: Valores de EC50 y Tm de los péptidos naturales y sus análogos no peptídicos.

(a) EC50: Concentración de ligando a la cuál la fluorescencia de HLA-B27 en las células RMA-S, es la mitad de la fluorescencia máxima obtenida con el péptido natural. (b) Tm: Punto medio del valor térmico de desnaturalización del heterodímero formado por la cadena pesada de B*2705, la β2m y el ligando.

(c) Proteína latente de membrana del EBV (236-244), (Brooks et al., 1993). (d) Nucleoproteína del virus Influenza A (383-391), (Huet et al., 1990). (e) Glicoproteína 120 del virus del SIDA (314-322), (Jardetzky et al., 1991). (f) Proteína DnaK de E. coli (260-268), (Rognan et al., 1995).

-59-

V. DISCUSIÓN

DISCUSIÓN -61-

V. DISCUSIÓN V.1. POLIMORFISMO DE HLA-B27 Y SOLAPAMIENTO DE REPERTORIOS PEPTÍDICOS

ENTRE SUBTIPOS. V.1.1. Influencia del polimorfismo de HLA-B27 sobre la especificidad por el residuo C-

terminal del péptido. Diferencias entre B*2704 y B*2706.

a teoría del péptido artritogénico implica que los subtipos asociados a enfermedad,

deben de ser capaces de presentar un mismo péptido unido selectivamente a los

subtipos asociados a enfermedad. Por esta razón, es particularmente importante

conocer el grado de solapamiento de los repertorios peptídicos unidos a los diferentes subtipos

de HLA-B27, y las diferencias entre los repertorios de los subtipos asociados diferencialmente a

enfermedad. Contrariamente a lo que ocurre en la subcavidad B, cuya estructura es común entre

los subtipos de HLA-B27 y está optimizada para unir péptidos con Arg2, la mayor parte del

polimorfismo de esta molécula se concentra en las subcavidades C y F, lo que se traduce en

distintas especificidades por el residuo C-terminal.

L

B*2705 une péptidos con residuos C-terminales básicos, alifáticos o aromáticos. Muchos

de los péptidos presentados por B*2702 se unen in vitro y probablemente también in vivo a

B*2705 (García et al., 1997b). Este solapamiento de repertorios peptídicos entre ambos subtipos

concuerda cualitativamentecon los patrones de reactividad cruzada de CTLs alorreactivos

(López et al., 1994). Por tanto B*2705 y B*2702 son un ejemplo de dos subtipos asociados a

EA, con un grado importante de solapamiento de sus repertorios peptídicos. Puesto que B*2702

no une péptidos con residuos C-terminales básicos, éstos probablemente no están implicados en

la patogénesis de la EA.

A pesar de las similitudes estructurales entre B*2704 y B*2706, sólo el primero está

asociado a enfermedad (López-Larrea et al., 1995), y sus especificidades por el residuo C-

terminal muestran marcadas diferencias.

B*2704 une preferentemente in vitro, péptidos con residuos C-terminales alifáticos y

aromáticos, similares a los presentados por B*2702 in vivo. Algunos péptidos con residuos C-

terminales básicos también se unen a B*2704 in vitro, sin embargo tales péptidos no se han

DISCUSIÓN -62-

detectado entre los ligandos naturales de este subtipo.

B*2706 por su parte, discrimina en mayor medida que B*2704 entre los residuos C-

terminales polares y apolares, lo que se concreta en una mayor preferencia que B*2704 por Leu

y Phe y menor por residuos básicos y Tyr. Ello sugiere que el repertorio de péptidos que B*2706

comparte con otros subtipos, incluye principalmente a los péptidos con Leu y Phe C-terminales.

Las diferencias de especificidad observadas in vitro entre B*2704 y B*2706 se

correlacionan con las que muestran in vivo, como se deriva de la secuenciación de los pooles de

péptidos eluídos de ambos subtipos (García et al., 1997b).

El aumento de la preferencia por residuos C-terminales apolares tras la eliminación de los

dos residuos ácidos Asp77 y Asp116, explica las especificidades C-terminales de B*2704 y

B*2706. No obstante, la ausencia de uno sólo de estos residuos en B*2704, no es suficiente para

impedir totalmente la unión de péptidos con residuos C-terminales básicos.

La preferencia que muestra B*2706 por Leu y Phe, y la unión ineficiente de Tyr C-

terminal, es probablemente debida al cambio D→Y116. El papel de esta posición ha sido

analizada en dos estudios previos en los que se analizaban mutaciones distintas, pero en ambos

tanto Leu como Phe eran los dos residuos C-terminales más adecuados, cuando D116 era

mutado. El cambio de Asp por Phe disminuía la aceptación de residuos C-terminales básicos

(Parker et al., 1994). Esta especificidad se explicaría por la incapacidad de Phe para formar

puentes de hidrógeno. En el otro estudio (Fiorillo et al., 1995), B*2709 cuyo residuo 116 es His,

era incapaz de unir péptidos de poli-Alanina con Arg o Tyr C-terminal pero sí podía unir Lys.

Puesto que B*2709 tampoco se asocia a EA, este estudio junto con los de B*2706 sugiere que el

residuo tiene un papel clave en determinar la susceptibilidad a EA.

La introducción de residuos ácidos en las posiciones 114 y 152, no altera la especificidad

de la subcavidad C/F debido a su localización, pero afectan notablemente a la unión in vitro de

los péptidos estudiados. Estos resultados sugieren que es posible alterar la especificidad por

péptidos unidos sin alterar las preferencias por los anclajes primarios, debido probablemente a la

modulación de las interacciones con residuos de anclaje secundarios (por ejemplo P7).

En el caso concreto de E152 (B*2710), este subtipo es muy distinto antigénicamente de

B*2705, puesto que es el que presenta una menor reactividad cruzada (Calvo et al., 1990). Sin

embargo, el análisis del repertorio peptídico natural de B*2710 reveló una sorprendente

homología con el de HLA-B27 (García et al., 1998 / Anexo 8). Por tanto el efecto de la mutación

E152 es mucho mayor sobre el reconocimiento por los TCR que sobre la especificidad de unión

-63- DISCUSIÓN

de péptidos. Estudios de modelización sugieren que ésto es debido a que el residuo 152 en HLA-

B27 es accesible al TCR (García et al., 1998 / Anexo 8; Figura 5).

Cuando se compara la unión de muchos de los péptidos a B*2704, S77 y E152, se observa

que la mutación S77 compensa el efecto negativo del residuo E152. De igual modo, las

afinidades de unión similares a Y116 y al doble mutante D114Y116 muestran una compensación

de la mutación Y116 sobre la mutación D114. Estos efectos compensatorios que modulan la

unión de péptidos, se correlacionan con resultados previos en los que se observaron efectos

compensatorios de las mutaciones en el reconocimiento por CTLs alorreactivos (López et al.,

1992; Villadangos et al., 1994).

Estos efectos compensadores, podrían suponer una ventaja evolutiva en el polimorfismo de

HLA-B27. Debido a que dos mutaciones generadas mediante conversión génica, un mecanismo

evolutivo frecuente en este sistema (López de Castro, 1989; Parham et al., 1995), tienen un

efecto menos disruptivo que una mutación puntual.

La ausencia de Asp116, y la menor capacidad para unir péptidos con C-terminal Tyr, son

características que comparten B*2706 y B*2709 y que diferencian a estos subtipos de otros

asociados a enfermedad: B*2705, B*2702 y B*2704. Esta correlación podría sugerir que un

posible péptido artritogénico debería tener Tyr lo que restringiría en gran medida el número de

candidatos. B*2707, aunque tampoco presenta el motivo Tyr C-terminal (Tieng et al., 1997) sí

está asociado a enfermedad. Sin embargo, es posible que este subtipo pudiera presentar péptidos

con Tyr C-terminal en niveles no detectables por secuenciación de Edman.

V.1.2. El polimorfismo en la cavidad C/F determina parcialmente la especificidad de la

subcavidad B. Los resultados derivados del estudio realizado con B*2701, B*2702 y los mutantes que

mimetizan sus cambios, todos localizados en la cavidad C/F, pusieron de manifiesto la

preferencia de B*2701 por péptidos con residuo Leu, Tyr y Phe C-terminal de forma similar a

B*2702 (Rötzschke et al., 1994).

Estas preferencias se explican por la mayor hidrofobicidad de la cavidad C/F de ambos

subtipos, respecto a B*2705. Adicionalmente, muchos de los péptidos secuenciados de B*2701

son compartidos por otros subtipos asociados a enfermedad, lo que es compatible con una

posible asociación de B*2701 a EA. De hecho uno de los pocos individuos tipados como

B*2701+, padecía esta enfermedad. La principal diferencia entre B*2701 y B*2702 reside en las

DISCUSIÓN -64-

preferencias por el residuo en P2. Este residuo, que está conservado entre los subtipos de HLA-

B27, es Arg2; la Gln2 aunque compatible con la unión de péptidos in vitro a varios subtipos, es

un residuo sobóptimo (García et al., 1997a; Villadangos et al., 1995; Galocha et al., 1996 /

Anexo 1; Parker et al., 1994; Fukazawa et al., 1994; Raghavan et al., 1996). Sin embargo,

B*2701 es capaz de unir in vitro y presentar in vivo péptidos tanto con Gln2 como con Arg2 con

eficiencia similar. Esta característica singular de B*2701 radica en el cambio D→Y74, como se

deduce de la presencia de Gln2 y Arg2 entre los péptidos que unen in vivo tanto B*2701 como

en el mutante Y74 (Anexo 2).

El efecto que ejerce la mutación Y74, situada en la subcavidad C/F, sobre la especificidad

de la subcavidad B, tiene su explicación en el residuo Lys70. Este residuo, casi exclusivo de

HLA-B27, se localiza próximo a la subcavidad B, pero no interviene en las interacciones con el

residuo P2. En B*2705, su cadena lateral se orienta hacia el exterior de la subcavidad B

(Madden et al, 1992) estableciendo un puente salino con el residuo Asp74. En B*2701, la

mutación Y74 impide la formación de ese enlace permitiendo la reorientación de la cadena

lateral de la Lys70 hacia la subcavidad B, donde puede interaccionar con residuos peptídicos de

Gln2.

Estudios recientes realizados en nuestro laboratorio (Krebs et al. 1999 / en revisión), han

confirmado experimentalmente el papel crítico de la Lys70 en la especificidad de B*2701 por

Gln2. En estos estudios, la mutación de Lys70 a Ala70 en B*2701, revierte la especificidad a

Arg2 exclusivamente. Adicionalmente, los estudios de modelización molecular sugieren que la

especificidad dual de B*2701 por Arg2 y Gln2 se debe a la capacidad de Lys70, en este subtipo,

para adoptar dos estados rotaméricos distintos dependiendo del residuo P2. Como se ha dicho, si

P2 es Gln, Lys70 se orienta hacia la subcavidad B para interaccionar con Gln2; si P2 es Arg,

Lys70 se reorienta en una conformación diferente hacia fuera de la subcavidad B, preservando el

mismo modo de interacción de la Arg2 en esta subcavidad, que en B*2705.

La mala unión general de muchos de los péptidos al mutante A81 se explica porque la

ausencia de Leu81 impide las interacciones con los residuos C-terminales apolares o con la

porción alifática de los básicos (Madden et al., 1992). Finalmente, la mejor unión de algunos

péptidos a B*2701 o B*2702 que al mutante A81, sugiere la existencia de efectos

compensatorios de los otros cambios.

V.1.3. Análisis del efecto del polimorfismo de la subcavidad A sobre la unión de péptidos.

-65- DISCUSIÓN

HLA-B*2703 se diferencia de B*2705 en un solo aminoácido, Y59H, ubicado en la

subcavidad A. Esta mutación es la responsable de que B*2703 presente un subconjunto de los

péptidos presentados por B*2705 (López et al, 1994; Villadangos et al, 1994) lo que hace

interesante el estudio del efecto que tiene esta mutación en la unión de péptidos.

En la estructura cristalográfica de HLA-B*2705 (Madden et al, 1992), las cadenas laterales

de los aminoácidos Tyr7 y Tyr171 establecen puentes de hidrógeno con el extremo N-terminal

del péptido. Ambas cadenas interaccionan adicionalmente, por mediación de una molécula de

agua, con las cadenas laterales de los residuos cercanos Tyr59, Glu45 y Tyr171. (Figura 6 /

Anexo 3)

En B*2703, se produce la distorsión de esa red de enlaces. La molécula de agua desaparece

y el extremo N-terminal del péptido se une a la His59, en lugar de a Tyr7. Como consecuencia,

se pierden el enlace directo con Tyr171 y cinco interacciones mediadas por la molécula de agua.

La desaparición de la molécula de agua que enlaza la subcavidad A con Glu45 en la subcavidad

B, probablemente refuerza la interacción de este residuo con Arg2.

En el caso en que P2 es Gln, los enlaces en la subcavidad B de B*2705 se debilitan,

principalmente con Thr24 y Glu45 y se generan nuevos reordenamientos conformacionales en

las interacciones con P1. Gln2 es más tolerado en B*2705, debido a que Tyr59 aún puede unirse

a Tyr7. Adicionalmente el enlace Cα−Ν del péptido unido, sufre una rotación que establece un

nuevo puente de hidrógeno con Glu63 sin perder el enlace con Tyr7. La cadena lateral de Gln2

establece además puentes de hidrógeno con Glu45 y Tyr99 reforzando las interacciones en la

subcavidad B. Estas interacciones no se producen en B*2703, como explica la baja afinidad de

péptidos con Gln2 a este subtipo (Figuras 7 y 8 / Anexo 3).

Tanto en B*2705 como en B*2703, la presencia de Gln2 induce una desestabilización del

extremo C-terminal. Esta situación ya detectada en otros péptidos (Rognan et al., 1994), sugiere

la posible función estabilizadora de la región N-terminal para el resto del péptido.

En los péptidos analizados, el residuo de Arginina en P1 es accesible a la formación de dos

puentes salinos con Glu63 y Glu163, sin embargo esto no supone una ventaja significativa sobre

la unión in vitro del análogo con Alanina. La cadena lateral de Alanina puede interaccionar con

residuos apolares conservados de la subcavidad A como (Met5 y Trp167), lo que explicaría la

similar afinidad de unión de péptidos tales como RRYQKSTEL y ARYQKSTEL. No obstante, hay

estudios que indican la preferencia de B*2703 por residuos básicos (Colbert et al., 1994) y la

DISCUSIÓN -66-

presencia mayoritaria de éstos entre los péptidos que une in vivo (Boisgérault et al., 1996;

Griffin et al., 1997) que no pueden explicarse mediante este modelo, aunque es posible la

formación de puentes salinos entre varios rotámeros de la cadena lateral básica de P1 y los

residuos cargados negativamente cercanos a la subcavidad A.

En conclusión, el efecto del cambio Y59→H presente en B*2703 tiene varios efectos

simultáneos que consisten en: (i) una disrupción de la red de puentes de hidrógeno que se

establecen en la subcavidad A, (ii) reordenamientos de las interacciones en la subcavidad B y

(iii) debilitamiento general de las interacciones que se producen con el extremo C-terminal del

péptido. Estos efectos pueden afectar en gran medida al conjunto de péptidos presentados por

B*2703 y hacer la unión de éstos más dependiente de la presencia de residuos básicos en P1.

-67- DISCUSIÓN

V.2. RELACIÓN ENTRE LA UNIÓN DE PÉPTIDOS Y LA SELECCIÓN DE EPÍTOPOS POR CÉLULAS T.

Los resultados de este estudio muestran: (i) la moderada influencia del polimorfismo en la

unión de péptidos de EBV restringidos por HLA-B27, (ii) la inexistencia de correlación entre la

unión promiscua de los péptidos virales a varios subtipos y su antigenicidad y/o

inmunogenicidad en el contexto de un subtipo particular, (iii) el motivo de anclaje Arg2, no es

necesario para mantener la estructura antigénica del péptido.

La moderada incidencia que muestra el polimorfismo entre los subtipos B*2701-B*2706,

en la unión de péptidos de EBV, que poseen residuos C-terminales alifáticos o aromáticos, es

consistente con el predominio de este tipo de residuos entre los ligandos naturales de los seis

subtipos (García et al., 1997a; Jardetzky et al., 1991; Boisgérault et al., 1996; Rötzschke et al.,

1994; García et al., 1997b). En general, no existe una correlación entre la unión a un subtipo y su

inmunogenicidad en el contexto del mismo. Este es el caso del péptido EBNA3B, restringido por

B*2702 y de LMP2, restringido por B*2704, que muestran una buena unión a todos los subtipos,

en algunos casos mejor incluso que al subtipo por el que están restringidos.

A pesar de que el desarrollo de una respuesta mediada por CTLs requiere habitualmente

una alta afinidad al elemento de restricción (Sette et al., 1994), en el estudio presente no se

aprecia una correlación directa entre la afinidad de unión a un subtipo y su inmunogenicidad en

el contexto del mismo. La inmunogenicidad parece correlacionarse mejor con la estabilidad del

complejo MHC-péptido (Van der Burg et al., 1996; Levitsky et al., 1996). En un trabajo reciente

(Brooks et al., 1998), EBNA3B, cuya unión in vitro es tan buena a B*2705 como a su elemento

de restricción B*2702, mostró una unión más estable a B*2702 que a B*2705 que se

correlacionó con la ausencia de reconocimiento del péptido en el contexto de B*2705. Por lo

tanto, las diferencias de estabilidad, no evidentes en el ensayo in vitro, pueden ser las

responsables de la inmunogenicidad de un péptido en el contexto de un subtipo y no de otros.

Aunque múltiples factores contribuyen a limitar el número de péptidos inmunogénicos en

una respuesta antiviral, tales como la afinidad, la acción de los proteosomas, el transporte

mediado por TAP, el repertorio de células T y la supresión de la respuesta de células T por otros

péptidos inmunodominantes (Deng et al., 1997); en este estudio los tres péptidos virales se unían

significativamente a todos los subtipos y además eran inmunogénicos, al menos en el contexto

de un subtipo, reduciendo las posibles variables, exclusivamente al de la capacidad de los

repertorios de células T para reconocer la estructura del complejo péptido-MHC.

DISCUSIÓN -68-

El hecho de que la estructura del complejo péptido-MHC depende del subtipo, es evidente

en la ausencia de correlación entre la unión de un péptido a varios subtipos, y su reconocimiento

en el contexto sólo de algunos. Así, los CTLs restringidos por B*2704 para el péptido LMP2 no

reconocían al péptido en el contexto de B*2702 ni de B*2705, confirmando observaciones

previas (Brooks et al., 1993); los CTLs restringidos por B*2705 o B*2702 no reconocían al

péptido EBNA3C unido a B*2704; los CTLs restringidos por B*2705 o B*2702 no reconocían al

péptido en el contexto de B*2704, y la mayoría de los efectores contra B*2705 no reaccionaban

de forma cruzada con B*2702.

El reconocimiento por CTLs activados de un péptido restringido por un subtipo, implica un

cambio en la estructura del epítopo tras la unión del péptido a diferentes subtipos, bien debido a

un cambio conformacional del propio epítopo, o bien a que el polimorfismo entre subtipos altera

la estructura del complejo o la interacción con el TCR. Esta última posibilidad es probablemente

la responsable de la ausencia de reactividad cruzada entre B*2704 y B*2705 o B*2702. La razón

de esto es que B*2704 difiere de los otros dos subtipos en el cambio E152→V. Esta mutación

que no parece afectar mucho a la presentación in vivo de los mismos péptidos naturales que

B*2705, ni altera su conformación, tiene sin embargo efectos drásticos en el reconocimiento por

el TCR (García et al., 1998 / Anexo 8).

Por lo tanto, el polimorfismo de HLA-B27 modula el reconocimiento por células T

mediante factores adicionales a la simple afinidad de unión de péptidos. Las diferencias mayores

entre subtipos residen en su diferente modulacion de la inmunogenicidad y antigenicidad de los

péptidos unidos, más que en su diferente especificidad de unión de péptidos.

La modulación de la antigenicidad e inmunogenicidad de un péptido en el contexto de

diferentes subtipos, tiene claras implicaciones en la patogenia de HLA-B27. La hipótesis que

sugiere la unión selectiva de un péptido artritogénico a los subtipos asociados a enfermedad tiene

ahora otra alternativa adicional, consistente en que dicho péptido artritogénico pueda unirse a

varios subtipos, pero sólo ser relevante en el contexto de algunos.

En este estudio, péptidos que carecían del motivo Arg2 podían ser reconocidos por CTLs

restringidos por el péptido con Arg2, lo que implica que su estructura antigénica reconocida no

era alterada tras eliminar un motivo de anclaje principal. Evidencias adicionales como la

reacción cruzada de CTLs con péptidos carentes de motivos canónicos (Malarkannan et al.,

1996), la secuenciación de péptidos carentes de Arg2 en B*2701 (García et al., 1997c / Anexo 2)

y en B*2705 de ratas transgénicas (Simmons et al., 1997) y la demostración de que los péptidos

-69- DISCUSIÓN

con Gln2 se unen in vitro a muchos subtipos de HLA-B27, sugiere que estos podrían constituir

una fracción menor del conjunto total de péptidos en subtipos diferentes a B*2701. Por tanto, la

búsqueda de posibles péptidos artritogénicos, en particular autopéptidos, no debería basarse

exclusivamente en los motivos canónicos. Péptidos con motivos no canónicos, debido a su baja

afinidad y escasa presentación in vivo, podrían eludir la autotolerancia mejor que los péptidos

con Arg2 y convertirse en dianas de CTLs autorreactivos generados tras una infección

artritogénica.

DISCUSIÓN -70-

V.3. LA ESPECIFICIDAD PEPTÍDICA DE LOS SUBTIPOS DE HLA-B27 ES MODULADA EN MULTIPLES POSICIONES DE ANCLAJE.

La variabilidad de los residuos en P1, P3 y PΩ determina en gran medida las propiedades de

unión del péptido, pero la contribución de cada una de las posiciones está jerarquizada según el

orden P9>P3>P1. Adicionalmente, el hecho de que muchos de los residuos sean inapropiados en

cualquiera de estas tres posiciones indica que la unión del péptido depende tanto de los efectos

positivos como de los negativos.

La exploración de los residuos presentes entre los ligandos naturales de B*2705 revela un

predominio de los residuos más adecuados en las posiciones P3 y P9; no obstante, la presencia de

un residuo subóptimo en una de estas posiciones es tolerada, siempre y cuando se compense con la

presencia en la otra posición de anclaje de un residuo óptimo. En el análisis de la distribución de

residuos entre los ligandos naturales de B*2705 no se aprecian excepciones a esta regla, sugiriendo

que si se excluyen los factores implicados en el procesamiento y transporte, aún insuficientemente

caracterizados (van Endert et al., 1995; Uebel et al., 1997; Daniel et al., 1998; Peh et al., 1998), la

limitación del número de ligandos naturales se basa en este tipo de restricciones.

El residuo P1 es el más permisivo de los tres, y su contribución a la unión es menor, pero esta

permisividad está condicionada por la presencia de un buen anclaje en P3 o P9. Teniendo en cuenta

que la contribución aditiva de las tres posiciones, junto con la posición P2 (Arg2), da cuenta de la

mayor parte de la afinidad del péptido natural, la predicción de epítopos naturales puede reducirse

al estudio de los residuos P1, P3 y P9 obviando el papel de los residuos de las posiciones centrales,

cuya menor contribución permite su sustitución por espaciadores no peptídicos que no alteran en

gran medida su afinidad (Rognan et al., 1995; Krebs et al., 1998 / Anexo 6; Poenaru et al., en

prensa / Anexo 7). A pesar de que no se conocen muchos ligandos naturales de tamaños distintos a

los canónicos, de nueve o diez residuos, la unión de ligandos con tamaños no canónicos sólo parece

posible cuando presentan buenos anclajes en las tres posiciones P1, P3 y PΩ.

Un aspecto relevante de la comparación entre B*2705 y B*2704 reside en las diferentes

especificidades por PΩ. En concordancia con su ausencia in vivo (García et al., 1997b) y su mala

unión in vitro (Tanigaki et al., 1994; Galocha et al., 1996 / Anexo 1), los residuos PΩ básicos no

están favorecidos en B*2704. Adicionalmente, los residuos C-terminales Y y F aunque no están

favorecidos, son presentados de forma natural por B*2704 (García et al., 1997b) probablemente

debido a que la estabilización por los otras posiciones de anclaje es suficiente para permitir la unión

de ligandos con estos residuos. Por lo tanto, el solapamiento de repertorios peptídicos de B*2705 y

-71- DISCUSIÓN

B*2704, probablemente consiste sobre todo en péptidos con residuos C-terminales alifáticos, y

aquellos con Y y F que además poseen en P3 residuos apolares.

Para entender la base molecular de su diferente asociación a enfermedad, es fundamental

analizar las diferencias entre B*2704 y B*2706. Las diferencias de B*2706 con B*2704 se

localizan en posiciones que pueden influir directamente en la interacción con los residuos P3 y P9

de los péptidos unidos. Los resultados obtenidos concuerdan plenamente con los estudios previos

de secuenciación que muestran la mayor preferencia de B*2706 por residuos C-terminales

alifáticos y F, así como la ausencia de Y en PΩ (García et al, 1997b). Además, en los estudios in

vitro, los residuos alifáticos están más favorecidos que Y en B*2706, pero no en B*2704 (Galocha

et al., 1996 / Anexo 1). La ausencia de Y se explicaría por el mejor ajuste en B*2706 de los

residuos alifáticos voluminosos y F que de Y, de forma que los péptidos con Y competirían en

desventaja para unirse in vivo a B*2706.

B*2704 y B*2706, muestran además otras diferencias en la preferencias por P3 y P1. Lo más

destacado es la distinta especificidad por el residuo P3, especialmente una mejor aceptación de R,

N y Q por B*2706 y la peor de A respecto a otros residuos. Estas diferencias son probablemente

debidas a la presencia del residuo D114 en lugar de H114. Las diferencias en el residuo P1, entre

subtipos que comparten una subcavidad A común, sugiere la existencia de efectos a larga distancia

de posiciones polimórficas alejadas de esta subcavidad.

La diferente especificidad de B*2704 y B*2706 en múltiples posiciones de anclaje implica

que su asociación diferencial a enfermedad puede no correlacionarse exclusivamente con la

incapacidad de B*2706 para unir Tyr C-terminal, sino también por la modulación de la

especificidad por los residuos en P1 y sobre todo en P3. Por lo tanto B*2704 y B*2706, aunque

comparten ligandos comunes pueden diferir en los repertorios de péptidos que presentan y muchos

péptidos con Arg2 y residuo C-terminal compatible con la unión tanto a B*2704 como a B*2706

pueden sin embargo unirse exclusivamente a un subtipo.

DISCUSIÓN -72-

V.4. UNIÓN DE ANÁLOGOS NO PEPTÍDICOS A HLA-B27. La supresión de la reactividad por CTLs, puede lograrse mediante la utilización de

ligandos modificados que al reaccionar con el TCR induzcan anergia, un fenómeno llamado

antagonismo. Debido a los numerosos detalles conocidos de las interacciones entre un péptido y

la molécula del MHC, el reemplazamiento en un péptido antigénico de la parte central

reconocida por el TCR, por espaciadores orgánicos no peptídicos que conserven los residuos de

anclaje a la molécula de clase I, es posiblemente una buena estrategia para diseñar ligandos que

funcionen como antagonistas de péptidos antigénicos. La ventaja de tales compuestos sobre los

antagonistas peptídicos, para su utilización in vivo, reside en su alta resistencia a la acción de

proteasas y sus mejores propiedades farmacocinéticas (Ishioka et al., 1994).

La modificación de la porción central del péptido con espaciadores no peptídicos, debe

mejorar en lo posible tanto su afinidad de unión como su estabilidad.

En los análogos estudiados con espaciador de -Aua- se observa una disminución de la

afinidad, respecto al péptido natural, dependiente del subtipo al que se une. Su naturaleza

monofuncional, permite la unión covalente entre los residuos P3 y P9 pero no el establecimiento

de contactos adicionales en las subcavidades centrales de la molécula de clase I. La ausencia de

grupos funcionales hace a estos análogos muy dependientes del buen anclaje en P9 que queda

claramente reflejada en los valores de baja afinidad que muestran los análogos con P9 básico

unidos a B*2704.

Los espaciadores bifuncionales formados por trímeros y tetrámeros de -HB-, además de

servir de enlace entre P3 y P9, poseen grupos funcionales metilo capaces de establecer contactos

adicionales en el surco de unión del péptido. Sin embargo, los contactos sólo son posibles de

forma óptima cuando la longitud del espaciador es apropiada, permitiendo el contacto de ambos

extremos del análogo con las subcavidades A y F; en este sentido el análogo con -HB3- no

cumple este requisito, una característica que lo diferencia claramente del análogo con el

espaciador de -HB4-, que si lo cumple.

El incremento de afinidad medido in vitro, es debido al anclaje adicional que proporcionan

los dos grupos metilo del espaciador -HB4-, éstos interaccionan con las subcavidades C y E. El

mayor número de interacciones determina una unión de este análogo superior a la del ligando

natural. En este sentido la utilización del estereoisómero (R) no es trivial, puesto que la

conformación estereoisomérica (R o S) de los sustituyentes es importante para mejorar la

afinidad (Poenaru et al., en prensa /Anexo 7).

-73- DISCUSIÓN

En este estudio se observaron discrepancias en los valores comparativos de afinidad entre

ligandos de HLA-B27 y sus análogos no peptídicos dependiendo de que se midiera su unión in

vitro (EC50) o su estabilidad térmica (Tm). Estas discrepancias se explican porque el ensayo de

unión in vitro, está muy influído por la cinética de asociación del péptido a la molécula de MHC,

mientras que la Tm es una medida de la estabilidad del complejo y se relaciona directamente con

la cinética de disociación del péptido.

RESUMEN Y DISCUSIÓN GENERAL:

Si el papel patogénico de HLA-B27 reside en su función presentadora de péptidos, el

conocimiento de las pautas que determinan el solapamiento de repertorios peptídicos entre

subtipos, así como el reconocimiento de dichos péptidos en función del subtipo al que están

unidos, son cuestiones fundamentales.

En esta tesis se han definido algunos aspectos importantes de la modulación del repertorio

peptídico por el polimorfismo de HLA-B27. Se ha analizado la relación entre la unión de

péptidos y su inmunogenicidad y antigenicidad, y se han estudiado algunos aspectos de la unión

de ligandos no peptídicos a HLA-B27.

El polimorfismo de HLA-B27 modula la unión de péptidos a tres niveles:

En primer lugar, el polimorfismo de las posiciones localizadas en la subcavidad C/F afecta

directamente a la especificidad por el residuo C-terminal, que es un anclaje principal a HLA-

B27. En consecuencia, limita o impide la aceptación de residuos C-terminales básicos en

múltiples subtipos. Además introduce una importante diferencia entre B*2704 y B*2706, dos

subtipos asociados diferencialmente a EA, en cuanto a que restringe la aceptación de Y C-

terminal en este último subtipo.

En segundo lugar, algunas posiciones polimórficas localizadas en una determinada

subcavidad modulan la especifidad por residuos peptídcos que interaccionan en subcavidades

distintas. En esta tesis se han analizado dos efectos contrapuestos. El poliformismo de B*2703,

localizado en la subcavidad A, fortalece la interacción de Arg2 en la subcavidad B y afecta a la

estabilización del extremo C-terminal del péptido. Por otra parte, la mutación D74Y en B*2701,

localizada cerca de la subcavidad C/F, favorece la interacción de Gln2 en la subcavidad B por un

efecto indirecto mediado por la Lys70.

DISCUSIÓN -74-

En tercer lugar, el polimorfismo de B27 modula la especifidad en posiciones secundarias

de anclaje. Esto es particularmente relevante, ya que indica que la asociación diferencial de

B*2704 y B*2706 a EA, no se correlaciona solamente con la aceptación de Y C-terminal, sino

con una modulación más compleja que incluye, adicionalmente, otras posiciones de anclaje. Las

interrelaciones entre los repertorios peptídicos de los dos subtipos se hacen más complejas por el

hecho de que las diversas posiciones de anclaje pueden tolerar en mayor o menor medida

residuos desfavorecidos, que son compesados por la presencia de buenos anclajes en otras

posiciones.

El estudio efectuado con péptidos virales pone de manifiesto que más allá de las

diferencias y similitudes en la especifidad de unión de péptidos, las diferencias funcionales entre

subtipos dependen de la modulación adicional que ejerce el polimorfismo sobre la

inmunogenicidad y antigenicidad de los péptidos unidos. En este estudio se demuestra que no

existe una correlación entre la eficiencia con la que un péptido se une HLA-B27, y su capacidad

para estimular una respuesta inmune o de ser reconcido por CTLs activados. Estos datos están de

acuerdo con los conceptos de que la estabilidad del complejo MHC-péptido es el determinante

crítico de la inmunogenicidad, y con que un CTL activado puede reconocer un péptido unido con

muy baja afinidad a MHC si la conformación del epítopo está conservada.

Finalmente, se han explorado las propiedades de unión a HLA-B27 de ligandos no

peptídicos en los que los residuos P4-P8 fueron sustituidos por varios espaciadores orgánicos. Se

ha demostrado que estos ligandos se unen a HLA-B27 con una afinidades que pueden ser

superiores a la del péptido natural, dependiendo de la naturaleza química del espaciador. Estos

estudios abren una vía a un ulterior análisis sobre el posible uso de estos compuestos en la

modulación de la respuesta citotóxica restringida por HLA-B27.

-75-

VI. CONCLUSIONES

CONCLUSIONES -77-

VI. CONCLUSIONES

Los subtipos B*2704 y B*2706, asociados diferencialmente a enfermedad, difieren en la menor preferencia de B*2706 por Tyr C-terminal y en su especificidad por residuos de anclaje secundario, en particular en P3.

B*2701 es el único subtipo conocido de HLA-B27 que une de forma significativa péptidos con Gln2 in vivo. La mutación Y74 es la responsable de esta característica y su efecto es indirecto a través de la Lys70.

Por lo tanto un residuo polimórfico (D74Y) puede ejercer efectos sobre la especificidad por residuos del péptido que interaccionan en regiones alejadas (cavidad B). Estos efectos están mediados a través de un residuo conservado (Lys70).

El cambio Y59→H en B*2703 tiene varios efectos simultáneos (i) ruptura de la red de puentes de hidrógeno en la subcavidad A, (ii) reordenamiento de las interacciones en la subcavidad B y (iii) debilitamiento general de las interacciones con el extremo C-terminal del péptido. Estos efectos hacen a B*2703 más dependiente del anclaje en P1.

La unión promiscua de los péptidos virales a HLA-B27 no se correlaciona con su antigenicidad e inmunogenicidad en el contexto de subtipos particulares. El polimorfismo de HLA-B27, probablemente influye en la inmunogenicidad peptídica modulando la estabilidad más que la afinidad. El motivo canónico Arg2 no es necesario para mantener la estructura antigénica de los epítopos peptídicos analizados.

Las restricciones al número de ligandos naturales de B*2705 se basan en gran medida en la coexistencia compensada de residuos óptimos y subóptimos en las posiciones de anclaje P1, y sobre todo P3 y P9. Éstas posiciones muestran una contribución jerárquica y aditiva a la afinidad del péptido. Entre los ligandos naturales de longitud no canónica es necesaria la presencia de buenos anclajes en las tres posiciones estudiadas.

Los análogos de péptidos antigénicos modificados en su región central mediante espaciadores no peptídicos de tipo bifuncional pueden mejorar la afinidad de unión a la molécula de clase I.

-79-

VII. REFERENCIAS

REFERENCIAS -81-

VII. REFERENCIAS

A Abe, R., Ishida, Y., Yui, K., Katsumata, M., Chused, T.M. (1992). T cell receptor-mediated recognition of self-ligand induces signaling in immature thymocytes before negative selection. J. Exp. Med. 176: 459-68. Androlewicz, M.J., Cresswell, P. (1994). Human transporters associated with antigen processing possess a promiscuous peptide-binding site (TAP1 and TAP2). Immunity. 1: 7-14. Armas, J.B., González, S., Martínez-Borra, J., Laranjeira, F., Ribeiro, E., Correia, J., Ferreira, M.L., Toste, M., López-Vázquez, A., López-Larrea, C. 1999. Susceptibility to ankylosing spondylitis is independent of the Bw4 and Bw6 epitopes of HLA-B27 alleles. Tissue Antigens. 53: 237-243. Arnold, D., Driscoll, P., Androlewicz, M., Hughes, E., Cresswell, P., Spies, T. (1992). Proteasome subunits encoded in the MHC are not generally required for the processing of peptides bound by MHC class I molecules. Nature. 360: 171-173.

B Bahram, S., Bresnaham, M., Geraghty, D.E., Spies, T. (1994). A second lineage of mammalian major histocompatibility complex class I genes. Proc. Natl. Acad. Sci. U. S. A.. 91: 6259-6263. Balas, A., Santos, S., García-Sánchez, F., Lillo, R., Merino, J.L., Vicario, J.L. (1998). Complete coding sequence of HLA-B*2712: a serologic B27-negative antigen associated to Bw6. Tissue Antigens. 4: 394-397. Benjamin, R. Parham, P. (1990). Guilt by association: HLA-B27 and ankylosing spondylitis. Immunol.Today. 11: 137-142. Bjorkman, P.J., Saper, M.A., Samraoui, B., Bennett, W.S., Strominger, J.L., Wiley, D.C. (1987). Structure of the human class I

histocompatibility antigen, HLA-A2. Nature. 329: 506-512. Boisgérault, F., Tieng, V., Stolzenberg, M.C., Dulphy, N., Khalil, I., Tamouza, R., Charron, D., Toubert, A. (1996). Differences in endogenous peptides presented by HLA-B*2705 and B*2703 allelic variants: implications for susceptibility to spondylarthropathies. J. Clin. Invest. 98: 2764-2770. Breban, M., Fernandez-Sueiro, J.L., Richardson, J.A., Hadavand, R.R., Maika, S.D., Hammer, R.E.,Taurog, J.D. (1996). T cells, but not thymic exposure to HLA-B27, are required for the inflammatory disease of HLA-B27 transgenic rats. J. Immunol. 156: 794-803. Brewerton, D.A., Caffrey M, Hart, F.D., James, D.C.O., Nichols, A., Sturrock, R.D. (1973). Ankylosing spondylitis and HL-A27. Lancet. 1: 904-907. Brewerton, D.A., Caffrey M, Hart, F.D., James, D.C.O., Nichols, A., Sturrock, R.D. (1974). Reiter’s disease and HL-A27. Lancet. 2: 996-998. Brooks, J.M., Murray, R.J., Thomas, W.A., Kurilla, M.G., Rickinson, A.B. (1993). Different HLA-B27 subtypes present the same immunodominant Epstein-Barr virus peptide. J. Exp. Med. 178: 879-887. Brooks, J.M., Colbert, R.A., Mear, J.P., Leese, A.M. Rickinson, A.B. (1998). HLA-B27 subtype polymorphism and CTL epitope choice: studies with EBV peptides link immunogenicity with stability of the B27-peptide complex. J. Immunol. 161: 5252-5259. Brown, M., Jepson, A., Young, A., Whittle, H., Greenwood, B., Wordsworth, P. (1997). Spondyloarthritis in West Africa: evidence for a non-B27 protective effect. Ann. Rheum. Dis. 56: 68-70. Brown, M., Wordsworth, P. (1997). Predisposing factors to spondyloarthropathies. Curr. Opin. Rheumatol. 9: 308-314.

C Calvo, V., Rojo, S., López, D., Galocha, B., López de Castro, J.A. (1990). Structure and diversity of HLA-B27specific T cell epitopes. Analysis with site-directed mutants mimicking HLA-B27 subtype polymorphism. J. Immunol. 144: 4038-4045.

REFERENCIAS -82-

Cerundolo, V., Kelly, A., Elliot, T., Trowsdale, J., Townsend, A. (1995). Genes encoded in the major histocompatibility complex affecting the generation of peptides for TAP transport. Eur. J. Immunol. 25: 554-562. Choo, S.Y., Antonelli, P., Nisperos, B., Nepom, G.T., Hansen, J.A. (1986). Six variants of HLA-B27 identified by isoelectric focusing. Immunogenetics. 23: 24-29. Choo, S.Y., St. John, T., Orr, H.T., Hansen, J.A. (1988). Molecular analysis of the variant alloantigen HLA-B27d (HLA-B*2703) identifies a unique single amino acid substitution. Hum. Immunol. 21: 209-219. Choo, S.Y., Fan, L.A., Hansen, J.A. (1991) A novel HLA-B27 allele maps B27 allospecificity to the region around position 70 in the alpha 1 domain. J. Immunol. 147: 174-80. Colbert, R.A., Rowland Jones, S.L., McMichael, A.J., Frelinger, J.A. (1994). Differences in peptide presentation between B27 subtypes: the importance of the P1 side chain in maintaining high affinity peptide binding to B*2703. Immunity. 1: 121-130. Collins, E.J., Garboczi, D.N., Wiley, D.C. (1994). Three-dimensional structure of a peptide extending from one end of a class I MHC binding site. Nature. 371: 626-629.

D Daniel, S., Brusic, V., Caillat-Zucman, S., Petrovsky, N., Harrison, L., Riganelli, D., Sinigaglia, F., Gallazzi, F., Hammer, J., Van Endert, P.M. (1998). Relationship between peptide selectivities of human transporters associated with antigen processing and HLA class I molecules. J.Immunol. 161: 617-624. D'Amato, M., Fiorillo, M.T., Carcassi, C., Mathieu, A., Zuccarelli, A., Bitti, P.P., Tosi, R., Sorrentino, R. (1995). Relevance of residue 116 of HLA-B27 in determining susceptibility to ankylosing spondylitis. Eur. J. Immunol. 25: 3199-3201. Del Porto, P., D'Amato, M., Fiorillo, M.T., Tuosto, L., Piccolella, E., Sorrentino, R. (1994). Identification of a novel HLA-B27 subtype by restriction analysis of a cytotoxic gamma delta T cell clone. J. Immunol. 153: 3093-3100.

Deng, Y., Yewdell, J.W., Eisenlohr, L.C., Bennink, J.R. (1997). MHC affinity, peptide liberation, T cell repertoire, and immunodominance all contribute to the paucity of MHC class I-restricted peptides recognized by antiviral CTL. J. Immunol. 158: 1507-1515. Driscoll, J., Brown, M.G., Finley, D., Monaqco, J.J. (1993). MHC-linked LMP gene products specifically alter peptidase activities of the proteasome. Nature. 365: 262-264. Duchmann, R., May, E., Ackermann, B., Goergen, B., Meyer zum Buschenfelde, K.H., Marker-Hermann, E. (1996). HLA-B27-restricted cytotoxic T lymphocyte responses to arthritogenic enterobacteria or self-antigens are dominated by closely related TCRBV gene segments. A study in patients with reactive arthritis. Scand. J. Immunol. 43: 101-108. Dunham, I., Sargent, C.A., Trowsdale, J., Campbell, R.D. (1987). Molecular mapping of the human Major Histocompatibility Complex by pulsed-field gel electrophoresis. Proc. Natl. Acad. Sci. U. S. A. 84: 7237-7241.

E Eleuteri, A.M., Kohanski, R.A., Cardozo, C., Orlowski, M. (1997). Bovine spleen multicatalytic proteinase complex (proteasome). Replacement of X, Y and Z subunits by LMP2, LMP7 and MECL1 and changes in properties and specificity. J. Biol. Chem. 272: 11, 824-831. El-Zaatari, F.A., Sams, K.C., Taurog, J.D. (1990). In vitro mutagenesis of HLA-B27. Amino acid substitutions at position 67 disrupt anti-B27 monoclonal antibody binding in direct relation to the size of the substituted chain. J. Immunol. 144: 1512-1517. Ellis, S.A., Taylor, C., McMichael, A. (1982). Recognition of HLA-B27 and related antigens by a monoclonal antibody. Hum. Immunol. 5: 49-59. Ezquerra, A., Bragado, R., Vega, M.A., Strominger, J.L., Woody, J., López de Castro, J.A. (1985). Primary structure of papain-solubilized human histocompatibility antigen HLA-B27. Biochemistry. 24: 1733-1741.

F Falk, K., Rötzschke, O., Stevanovic, S., Jung, G.,

-83- REFERENCIAS

Rammensee, H.G. (1991). Allele-specific motifs revealed by sequencing of self-peptides eluted from MHC molecules. Nature. 351: 290-296. Fentenay, G., Standaert, R.F., Lane, W.S., Choi, S., Corey, E.J. Schreiber, S.L. (1995) Inhibition of proteasome activities and subunit specific amino-terminal threonine modification by lactacystin. Science. 268: 726-731. Fernández-Viña, M.A., Lázaro, A.M., Nulf, C.J., Stastny, P. (1996). Nucleotide sequence of novel subtypes of HLA-B27, B55 and B57. 22nd Meeting of the American Society for Histocompatibility and Immunogenetics. San Diego, October 11-15 1996. Hum. Immunol. 49: 43. Fiorillo, M.T., Greco, G., Sorrentino, R. (1995). The Asp116-His116 substitution in a novel HLA-B27 subtype influences the acceptance of the peptide C-terminal anchor. Immunogenetics. 4: 38-41. Fiorillo, M.T., Meadows, L., D'Amato, M., Shabanowitz, J., Hunt, D.F., Apella, E., Sorrentino, R. (1997). Susceptibility to ankylosing spondylitis correlates with the C-terminal residue of peptides presented by various HLA-B27 subtypes. Eur. J. Immunol. 27: 368-373. Francke, U., Pellegrino, M.A. (1977). Assignement of the Major Histocompatibility Complex to a region of the short arm of human chromosome 6. Proc. Natl. Acad. Sci. U. S. A. 74: 1147-1151. Fremont, D.H., Matsumura, M., Stura, E.A., Peterson, P.A., Wilson, I.A. (1992). Crystal structures of two viral peptides in complex with murine MHC class I H-2Kb. Science. 257: 919-927. Fukazawa, T., Wang, J., Huang, F., Wen, J., Tyan, D., Williams, K.M., Raybourne, R.B., Yu, D.T. (1994). Testing the importance of each residue in a HLA-B27-binding peptide using monoclonal antibodies. J. Immunol. 152: 1190-1196.

G Gaczynska, M., Rock, K.L., Goldberg, A.L. (1993). Gamma-interferon and expression of MHC genes regulate peptide hydrolysis by proteasomes. Nature. 365: 264-267. Gaczynska, M., Rock, K.L., Spies, T., Goldberg, A.L. (1994). Peptidase activities of proteasomes are differentially regulated by the major histocompatibility

complex-encoded genes for LMP2 and LMP7. Proc. Natl. Acad. Sci. U. S. A. 91: 9213-9217. Galocha, B., Lamas, J.R., Villadangos, J.A., Albar, J.P., López de Castro, J.A. (1996). Binding of peptides naturally presented by HLA-B27 to the differentially disease-associated B*2704 and B*2706 subtypes, and to mutants mimicking their polymorphism. Tissue antigens. 48: 509-518. Gao, X.M., Wordsworth, P., McMichael, A.J., Kyaw, M.M., Seifert, M., Rees, D., Dougan, G. (1996). Homocysteine modification of HLA antigens and its immunological consequences. Eur. J. Immunol. 26: 1443-1450. Gao, G.F., Tormo, J., Gerth, U.C., Wyer, J.R., McMichael, A.J., Stuart, D.I., Bell, J.I., Jones, E.Y., Jakobsen, B.K. (1997). Crystal structure of the complex between human CD8 alpha (alpha) and HLA-A2. Nature. 387: 630-634. Garboczi, D.N., Ghosh, P., Utz, U., Fan, Q. R., Biddison, W.E., Wiley, D.C. (1996). Structure of the complex between human T-cell receptor, viral peptide and HLA-A2. Nature. 384: 134-141. García, K.C., Degano, M., Stanfield, R.L., Brunmark, A., Jackson, M.R., Peterson, P.A., Teyton, L., Wilson, I.A. (1996). An T cell receptor structure at 2.5 Å and its orientation in the TCR-MHC complex. Science. 274: 209-219. García, F., Marina, A., Albar, J.P., López de Castro, J.A. (1997a). HLA-B27 presents a peptide from a polymorphic region of its own molecule with homology to proteins from arthritogenic bacteria. Tissue antigens 49: 23-28. García, F., Marina, A., López de Castro, J.A. (1997b). Lack of carboxiyl-terminal tyrosine distinguishes the B*2706-bound peptide repertoire from those of B*2704 and other HLA-B27 subtypes associated with ankylosing spondylitis. Tissue Antigens. 49: 215-221. García, F., Galocha, B., Villadangos, J.A, Lamas, J.R., Albar, J.P., Marina, A., López de Castro, J.A. (1997c). HLA-B27 (B*2701) specificity for peptides lacking Arg2 is determined by polymorphism outside the B pocket. Tissue Antigens. 49: 580-587. García, F., Rognan, D., Lamas, J.R., Marina, A., López de Castro, J.A. (1998). An HLA-B27 polymorphism (B*2710) that is critical for T-cell recognition has limited effects on peptide specificity. Tissue Antigens.51: 1-10.

REFERENCIAS -84-

Garret, T.P.J., Saper, M.A., Bjorkman, P.J., Strominger, J.L., Wiley, D.C. (1989). Specificity pockets for the side chains of peptide antigens in HLA-Aw68. Nature. 342: 692-696. Glotzer, M., Murray, A., Kirschner, M. (1991) Cyclin is degraded by the ubiquitin pathway. Nature. 349: 132-138. Glynne, R., Powis, S.H., Beck, S., Kelly, A., Kerr, L.A., Trowsdale, J. (1991). A proteasome related gene between the two ABC transporter loci in the classII region of the human MHC. Nature. 353: 357-360. González-Roces, S., Brautbar, C., Peña, M., Domínguez, O., Coto, E., Álvarez, V., Segal, R., López-Larrea, C. (1994). Molecular analysis of HLA-B27 haplotypes in Caucasoids. Frequencies of B27-Cw in Jewish and Spanish populations. Hum. Immunol. 41:127-134. González-Roces, S., Álvarez, M.V., González, S., Dieye, A., Makni, H., Woodfield, D.G., Housan, L., Konenkov, V., Abbadi, M.C., Grunnet, N., Coto, E., López-Larrea, C. (1997). HLA-B27 polymorphism and worlwide susceptibility to ankylosing spondylitis. Tissue Antigens. 49, 116-123. Goodfellow, P.N., Jones, E.A., van Heiningen, V., Salomon, E., Bobrow, M., Miggiano, V., Bodmer, W.F. (1975). The β2m gene is on chromosome 15 and not in the HLA region. Nature. 254: 267-269. Griffin TA, Yuan J, Friede T, Stevanovic S, Ariyoshi K, Rowland-Jones SL, Rammensee HG, Colbert RA. (1997). Naturally occurring A pocket polymorphism in HLA-B*2703 increases the dependence on an accessory anchor residue at P1 for optimal binding of nonamer peptides. J.Immunol. 159:4887-4897.

Groettrup, M., Kraft, R., Kostka, S., Standera, S., Stohwasser, R., Kloetzel, P.M. (1996). A third interferon-gamma-induced subunit exchange in the 20S proteasome. Eur. J. Immunol. 26: 863-869. Groll, M., Ditzel, L., Löwe, J., Stock, D., Bochtler, M., Bartunik, H.D., Huber, R. (1997). Structure from 20S proteasome from yeast at 2.4Å resolution Nature. 386: 463-471. Guo, H.C., Jardetzky, T.S., Garrett, T.P.J., Lane, W.S., Strominger, J.L., Wiley, D.C. (1992). Different lenght peptides bind to HLA-Aw68 similarly at their ends but bulge out in the middle. Nature. 360: 364-366.

H Hammer, R.E., Maika, S.D., Richardson, J.A., Tang, J.P., Taurog, J.D. (1990). Spontaneous inflammatory disease in transgenic ratsexpressing HLA-B27 and human 2m: an animal model of HLA-B27-associated human disorders. Cell. 63:1099-1112. Hasegawa, T., Ogawa, A., Sugahara, Y., Moriyama, Y., Nanzai, H., Tawara, K., Tokunaga, K., Juji, T., Kondo, S. (1997). A novel HLA-B27 allele (B*2711) encoding an antigen reacting with both B27- and B40-specific antisera. Tissue Antigens. 49: 649-652. Herberg, J.A., Sgouros, J., Jones, T., Copeman, J., Humphray, S.J., Sheer, D., Cresswell, P., Trowsdale, J. (1998) Genomic analysis of the Tapasin gene, located close to the TAP loci in the MHC. Eur. J. Immunol. 28: 459-467. Hermann, E., Yu, D.T., Meyer zum Buschenfelde, K.H., Fleischer, B. (1993) HLA-B27-restricted CD8 T cells derived from sinovial fluids of patients with reactive arthritis and ankylosing spondylitis. Lancet. 342: 646-650. Hildebrand, W.H., Domena, J.D., Shen, S.Y., Marsh, S.G., Bunce, M., Guttridge, M.G., Darke, C., Parham, P. (1994). The HLA-B7Qui antigen is encoded by a new subtype of HLA-B27 (B*2708). Tissue Antigens. 44: 47-51. Huet, S., Nixon, D.F., Rothbard, J.B., Townsend, A., Ellis, S.A., McMichael, A.J. (1990) Structural homologies between two HLA B27-restricted peptides suggest residues important for interaction with HLA B27. Int. Immunol. 2: 311-316. Hughes, E.A., Ortmann, B., Surman, M., Cresswell, P. (1996). The proteinase inhibitor, N-actyl-L-leucyl-leucyl-L-norleucinal, decreases the pool of major histocompatibility complex class I-binding peptides and inhibits peptide trimming in the endoplasmic reticulum. J. Exp. Med. 183: 1569-1578.

I Ishioka, G.Y., Adorini, L., Guery, J.C., Gaeta, F.C., LaFond, R., Alexander, J., Powell, M.F., Sette, A., Grey, H.M. (1994). Failure to demonstrate long-lived MHC saturation both in vitro and in vivo. Implications for therapeutic potential of MHC-

-85- REFERENCIAS

blocking peptides. J Immunol. 152: 4310-4319.

J Jackson, M.R., Cohen-Doyle, M.F., Peterson, P.A., Williams, D.B. (1994). Regulation of MHC class I transport by the molecular chaperone, calnexin (p88, IP90). Science. 263: 384-387. Jardetzky, T.S., Lane, W.S., Robinson, R.A., Madden, D.R., Wiley, D.C. (1991). Identification of self peptides bound to purified HLA-B27. Nature. 353: 326-329. Jorgensen, J.L., Reay, P.A., Ehrich, E.W., Davis, M.M. (1992). Molecular components of T-cell recognition. Annu. Rev. Immunol. 10: 835-873. Jordan, B.R., Caillol, D., Damotte, M., Delovitch, T., Ferrier, P., Kahnperles, B., Kourilsky, F., Layet, C., Le Boutiller, P., Lemonnier, F.A., Malissen, M., N’Guyen, C., Sire, J., Sodoyer, R., Strachan, T., Trucy, J. (1985). HLA class I genes: from structure to expression, serology and function. Immunol. Rev. 84: 74-92.

K Krangel, M.S., Orr, H.T., Strominger, J.L. (1979). Assembly and maduration of HLA-A and HLA-B antigens in vivo. Cell. 18: 979-991. Kelly, A., Powis, S.H., Glynne, R., Beck, S. Trowsdale, J. (1991). Second proteasome-related gene in the human MHC class II region. Nature. 353: 667-668. Krebs, S., Lamas, J.R., Poenaru, S., Folkers, G., López de Castro, J.A., Seebach, D., Rognan, D. (1998). Substituting nonpeptidic spacers for the T Cell Receptor-binding part of class I Major Histocompatibility Complex-binding peptides. J.Biol. Chem. 273: 19072-19079. Krebs, S., Rognan, D., López de Castro, J.A. (1999). Long range effects in protein-ligand interactions mediate peptide specificity in the human major histocompatibility antigen HLA-B27 (B*2701). (en revisión)

L Levitsky, V., Zhang, Q.J., Levitskaya, J., Masucci,

M.G. (1996). The life span of major histocompatibility complex-peptide complexes influences the efficiency of presentation and immunogenicity of two class I-restricted cytotoxic T lymphocyte epitopes in the Epstein-Barr virus nuclear antigen. 4. J. Exp. Med.183: 915-926. Ljunggren, H.G., Kärre, K. (1985). Host resistance directed selectively against H-2-deficient lymphoma variants. Analysis of the mechanism. J. Exp. Med. 162: 1745-1759. Ljunggren, H.G., Kärre, K. (1990a). In search of the “missing self”: MHC molecules and NK cell recognition. Immunol. Today. 11: 237-242. Ljunggren, H.G., Stam, N.J., Öhlen, C., Neefjes, J.J., Höglund, P., Heemels, M.T., Bastin, J., Schumacher, T.N.M., Townsend, A., Kärre, K., Ploegh, H.L. (1990b). Empty MHC class I molecules come out in the cold. Nature. 346: 476-480. López de Castro, J.A. (1989) HLA-B27 and HLA-A2 subtypes: evolution and function. Immunol Today. 10: 239-246. López, D., García Hoyo, R., López de Castro, J.A. (1994). Clonal analysis of alloreactive T cell responses against the closely related B*2705 and B*2703 subtypes. Implications for HLA-B27 association to spondyloarthropathy. J. Immunol. 152: 5557-5571. López-Larrea, C., Sujirachato, K., Mehra, N.K., Chiewsilp, P., Isarangkura, D., Kanga, U., Domínguez, O., Coto, E., Peña, M., Setién, F., González-Roces, S. (1995). HLA-B27 subtypes in Asian patients with ankylosing spondylitis. Evidence for new associations. Tissue Antigens. 45: 169-176. López-Larrea, C., González-Roces, S., Álvarez, V. (1996). HLA-B27 structure, function, and disease association. Curr. Opin. Rheumatol. 8: 296-308. Löwe, J., Stock, D., Jap, B., Zwickl, P., Baumeister, W., Huber, R. (1995). Crystal structure of the 20S proteosome from the archaeon T. acidophilum at 3.4Å resolution. Science. 268: 533-539.

M Madden, D.R., Gorga, J.C., Strominger, J.L., Wiley, D.C. (1991). The structure of HLA-B27 reveals nonamer self-peptides bound in an extended conformation. Nature 353: 321-325.

REFERENCIAS -86-

Madden, D.R., Gorga, J.C., Strominger, J.L., Wiley, D.C. (1992). The three-dimensional structure of HLA-B27 at 2.1Å resolution suggests a general mechanism for tight peptide binding to MHC. Cell. 70: 1035-1048. Madden, D.R., Garboczi, D.N., Wiley, C. (1993). The antigenic identity of peptide-MHC complexes: A comparison of the conformations of five viral peptides presented by HLA-A2. Cell. 75: 693-708. Malarkannan, S., González, F., Nguyen, V., Adair, G., Shastri, N. (1996). Alloreactive CD8+ T cells can recognize unusual, rare, and unique processed peptide/MHC complexes. J. Immunol. 157: 4464-4473. Melián, A., Beckman, E.M., Porcelli, S.A., Brenner, M.B. (1996) Antigen presentation by CD1 and MHC-encoded class I-like molecules. Curr. Opin. Immunol. 8: 82-88. Momburg, F., Ortiz-Navarrete, V., Neefjes, J. Goulmy, E., van de Wal, Y., Spits, H., Powis, S.J., Butcher, G.W., Howard, J.C., Walden, P., Hämmmerling, G.J. (1992). The proteasome subunits encoded by the major histocompatibility complex are not essential for antigen presentation. Nature. 360: 174-177. Momburg, F., Roelse, J., Howard, J.C. Butcher, G.W., Hammerling, G.J., Neefjes, J.J. (1994). Selectivity of MHC-encoded peptides transporters from human, mouse and rat Nature. 367: 648-651. Moretta, A., Vitale, M., Sivori, S., Bottino, C., Morelli, L., Augugliaro, R., Brbaresi, M., Pende, D., Ciccone, E., López-Botet, M. (1994). Human natural killer cell receptors for HLA-class I molecules. Evidence that the Kp43 (CD94) molecule functions as receptor for HLA-B alleles. J.Exp. Med. 180: 545-555. Morrison, L.A., Lukacher, A.E., Braciale, V.L., Fan, D.P., Braciale, T.J. (1986). Difference in antigen presentation to MHC classI- and class II-restricted influenza virus-specific cytolytic T lymphocyte clones. J. Exp. Med. 163: 903-921. Moses, J.H., Marsh, S.G.E., Arnett, K.L., Adams, E.J., Bodmer, J.G., Parham, P. (1995). On the nucleotide sequences of B*2702 and B*2705. Tissue Antigens. 45: 50-53.

N Nasution, A.R., Mardjuadi, A., Kunmartini, S., Suryadhana, N.G., Setyohadi, B., Sudarsono, D., Lardy, N.M.,Feltkamp, T.E.W. (1997). HLA-B27 subtypes positively and negatively associated with spondylarthropathy. J. Rheumatol. 24: 1111-1114.

O Orr, H.T., Lancet, D., Robb, R.J., López de Castro, J.A. Strominger, J.L. (1979). The heavy chain of human histocompatibility antigen HLA-B7 contains an immunoglobulin-like region. Nature. 282: 266-279. Ortmann, B., Copeman, J., Lehner, P.J., Sadasivan, B., Herbert, J.A., Grandea, A.G., Ridell, S.R., Tampé, R., Spies, T., Trowsdale, J., Cresswell, P. (1997). A critical role for Tapasin in the assembly and function of multimeric MHC class I-TAP complexes. Science. 277: 1306-1309.

P Pamer, E., Cresswell, P. (1998) Mechanisms of MHC class I-restricted antigen processing. Annu. Rev. Immunol. 16: 323-358. Paradela, A., García-Peydró, M., Vázquez, J., Rognan, D., López de Castro, J.A. (1998). The same natural ligand is involved in allorecognition of multiple HLA-B27 subtypes by a single T cell clone: Role of peptide and the MHC molecule in alloreactivity. J. Immunol. 161: 5481-5490. Parham, P., Lomen, C.E., Lawlor, D.A., Ways, J.P., Holmes, N., Coppin, H.L., Salter, R.D., Wan, A.M., Ennis, P.D. (1988). Nature of polimorphism in HLA-A, B, and C molecules. Proc. Natl. Acad. Sci. U. S. A. 85: 4005-4009. Parham, P., Arnett, K.L., Adams, E.J., Barber, L.D., Domena, J.D., Stewart, D., Hildebrand, W.H., Little, A.M. (1994). The HLA-B73 antigen has a most unusual structure that defines a second lineage of HLA-B alleles. Tissue Antigens. 43: 302-313. Parham, P., Adams, E.J., Arnett, K.L. (1995). The origins of HLA-A,B,C polymorphism. Immunol. Rev. 143: 141-180. Parker, K.C., Biddison, W.E., Coligan, J.E. (1994).

-87- REFERENCIAS

Pocket mutations of HLA-B27 show that anchor residues act cumulatively to stabilize peptide binding. Biochemistry. 33: 7736-7743. Peh, C.A., Burrows, S.R., Barnden, M., Khanna, R., Cresswell, P., Moss, D.J., McCluskey, J. (1998). HLA-B27-restricted antigen presentation in the absence of tapasin reveals polymorphism in mechanisms of HLA class I peptide loading. Immunity 8: 531-542.

Phillips, J.H., Chang, C.W., Mattson, J., Gumperz, J.E., Parham, P., Lanier, L.L. (1996). CD94 and a novel associated protein (94AP) form a NK cell receptor involved in the recognitionof HLA-A, HLA-B and HLA-C allotypes. Immunity. 5: 163-172. Plattner, D.A., Brunner, A., Dobler, M., Müller, H.M., Petter, W., Zbinden, P., Seebach, D. (1993). Cyclic oligomers of (R)-3-hydroxybutanoic acid: Preparation and structural aspects. Helv. Chim. Acta. 76: 2004-2033. Ploegh, H.L., Orr, H.T., Strominger, J.L. (1981). Major Histocompatibility Antigens: the human (HLA-A,B,C) and murine (H2-K, H2-D) class I molecules. Cell. 24: 287-299. Poenaru, S., Lamas, J.R., Folkers, G., López de Castro, J.A., Seebach, D., Rognan, D. (1998). Nonapeptide analogues containing (R)-3-hydroxybutanoate and -homoalanine oligomers: synthesis and binding affinity to a class I MHC protein. (J. Med. Chem. en prensa).

R Raghavan, M., Lebron, J.A., Johnson, J.L., Bjorkman, P.J. (1996). Extended repertoire of permissible peptide ligands for HLA-B*2702. Protein Sci. 5: 2080-2088. Rajagopalan, S., Brenner, M.B. (1994). Calnexin retains unassembled major histocompatibility complex class I free heavy chains in the endoplasmic reticulum. J. Exp. Med. 180: 407-412. Rammensee, H.G., Friede, T., Stevanoviic, S. (1995). MHC ligands and peptide motifs: first listing. Immunogenetics. 41: 178-228. Ren, E.C., Koh, W.H., Sim, D., Boey, M.L., Wee, G.B., Chan, S.H. (1997). Possible protective role of HLA-B*2706 for ankylosing spondylitis. Tissue Antigens. 49: 67-69.

Rock, K.L., Gramm, C., Rothstein, L., Clark, K., Stein, R., Dick, L.R., Hwang, D., Goldberg, A.L. (1994). Inhibitors of the proteosome block the degradation of most cell proteins and the generation of peptides presented on MHC clas I molecules. Cell. 78: 761-777. Roelse, J. Gromme, M., Momburg, F., Hammerling, G.J., Neefjes, J. (1994). Trimming of TAP-translocated peptides in the endoplasmic reticulum and in the cytosol during recycling. J. Exp. Med. 180: 1591-1597. Rogers, S., Wells, R., Rechsteiner, M. (1986) Amino acid sequences common to rapidly degraded proteins: the PEST hypothesis. Science. 234: 364-368. Rognan, D., Scapozza, L., Folkers,G., Daser, A. (1994). Molecular dinamics simulation of MHC-peptide complexes as a tool for predicting potential T cell epitopes. Biochemistry 38: 11476-11485. Rognan, D., Scapozza, L., Folkers,G., Daser, A. (1995). Rational design of nonaturalpeptides as high-affinity ligands for the HLA-B*2705 human leucocyte antigen. Proc. Natl. Acad. Sci. U. S. A. 92: 753-757. Rojo, S., Aparicio, P., Hansen, J.A., Choo, S.Y., López de Castro, J.A. (1987a). Structural analysis of an HLA-B27 functional variant, B27d, detected in American blacks. J.Immunol. 139: 3396-3401. Rojo, S., Aparicio, P., Choo, S.Y., Hansen, J.A., López de Castro, J.A. (1987b). Structural analysis of an HLA-B27 population variant, B27f. Multiple patterns of amino acid changes within a single polypeptide segment generate polymorphism in HLA-B27. J. Immunol. 139: 831-836. Rötzschke, O., Falk, K., Stevanovic, S., Gnau, V., Jung, G., Rammensee, H.G. (1994). Dominant aromatic/aliphatic C-terminal anchor in HLA-B*2702 and B*2705 peptide motifs. Immunogenetics. 39, 74-77. Rudwaleit, M., Bowness, P., Wordsworth, P. (1996). The nucleotide sequence of HLA-B*2704 reveals a new amino acid substitution in exon 4 which is also present in HLA-B*2706. Immunogenetics. 43: 160-162.

S Sadasivan, B., Lehner, P.J., Ortman, B., Spies, T., Cresswell, P. (1996). Roles for Calreticulin and a novel glycoprotein Tapasin, in the interaction of MHC

REFERENCIAS -88-

molecules with TAP. Immunity. 5: 103-114. Salter, R.S., Benjamin, R.J., Wesley, P.K., Buxton, S.E., Clayberger, C., Krensky, A.M., Norment, A.M., Littman, D.R., Parham, P. (1990). A binding site for the T-cell co-receptor CD8 on the 3 domain of HLA-A2. Nature. 345: 41-46. Saper, M.A., Bjorkman, P.J., Wiley, D.C. (1991). Refined structure of the human histocompatibility antigen HLA-A2 at 2.6Å resolution. J. Mol. Biol. 219: 277-319. Seong, R.H., Clayberger, C.A., Krensky, A.M., Parnes, J.R. (1988). Rescue of Daudi cell HLA expression by transfection of the mouse beta-2-microglobulin gene. J. Exp. Med. 167: 288-299. Sette, A., Vitiello, A., Reherman, B., Fowler, P., Nayersina, R., Kast, W.A., Melief, C.J., Oseroff, C., Yuan, L., Ruppert, J., Sidney, J., del Guericio, M. F., Southwood, S., Kubo, R.T., Chesnut, R.W., Grey, H.M., Chisari, F.V. (1994). The relationship between class I binding affinity and immunogenicity of potential cytotoxic T cell epitopes. J. Immunol. 153: 5586-5592. Seurynck, K., Baxter-Lowe, L.A. (1998). Novel polymorphism detected in exon 1 of HLA-B*2713. Tissue Antigens. 52: 187-189. Schlosstein, L., Terasaki, P.I., Bluestone, R., Pearson, C.M. (1973). High association of an HL-A antigen, W27, with ankylosing spondylits. N. Eng. J. Med. 288: 704-706. Schumacher, T.N.M., Kantesaria, D., Serreze, D.V., Roopenian, D.C., Ploegh, H.L. (1994). Transporters from H-2b, H-2d, H-2s, H-2k and H-2g7 (NOD/Lt) haplotype translocate similar peptides. Proc. Natl. Acad. Sci. U. S. A. 91: 4-8. Silver, M.L., Guo, H.C., Strominger, J.L., Wiley, D.C. (1992) Atomic structure of a human MHC molecule presenting an influenza virus peptide. Nature. 360: 367-369. Simmons, W.A., Summerfield, S.G., Roopenian, D.C., Slaughter,C.A., Zuberi, A.R.; Gaskell, S.J., Bordoli,R.S., Hoyes, J.; Moomaw, C.R., Colbert, R.A., Leong, L.Y.W., Butcher, G.W.; Hammer, R.E., Taurog, J.D. (1997).Novel HY peptide antigen presented by HLA-B27. J. Immunol. 159: 2750-2759. Smith, K.D., Mace, B.E., Valenzuela, A., Vigna, J.L., McCutcheon, J.A., Barbosa, J.A., Huczko, E., Engelhard, V.H., Lutz, C.T. (1996a). Probing HLA-B7 conformational shifts induced by peptide-binding

groove mutations and bound peptide with anti-HLA monoclonal antibodies. J Immunol. 157: 2470-2478. Smith, K.J., Reid, S.W., Harlos, K., McMichael, A.J., Stuart, D.I., Bell, J.I., Jones, E.Y. (1996b). Bound water structure and polymorphic amino acids act together to allow the binding of different peptides to MHC class I HLA-B53. Immunity. 4: 215-228. Smith, K.J., Reid, S.W., Stuart, D.I., McMichael, A.J., Jones, E.Y., Bell, J.I. (1996c). An altered position of the 2 helix of MHC class I is revealed by the crystal structure of HLA-B*3501. Immunity. 4: 203-213. Snyder, H.L., Yewdell, J.W., Bennik, J.R. (1994). Trimming of antigenic peptides in an early secretory compartment. J. Exp. Med.180: 2389-2394. Spies, T., Bresnahan M., Bahram, S., Arnold, D., Blanck, G., Mellins, E., Pious, D., DeMars, R. (1990). A gene in the human histompatibility complex class II region controlling the class I antigen presentation pathway. Nature. 348: 744-747. Spies, T., DeMars, R. (1991) Restored expression of major histocompatibility class I molecules by gene transfer of a putative peptide transporter. Nature. 351: 323-324. Sun, R., Shepherd, S.E., Geier, S.S., Thomson, C.T., Sheil, J.M., Nathenson, S.G. (1995). Evidence that the antigen receptors of cytotoxic T lymphocytes interact with a common recognition pattern on the H-2Kb molecule. Immunity. 3:573-582.

T Tanigaki, N., Fruci, D., Vigneti, E., Starace, G., Rovero, P., Londei, M., Butler, R.H., Tosi, R. (1994). The peptide binding specificity of HLA-B27 subtypes. Immunogenetics. 40: 192-198. Taurog, J.D., Richardson, J.A., Croft, J.T., Simmons, W.A., Zhou, M., Fernández Sueiro, J.L., Balish, E., Hammer, R.E. (1994). The germfree state prevents development of gut and joint inflammatory disease in HLA-B27 transgenic rats. J. Exp. Med. 180: 2359-2364. Tieng, V., Dulphy, N., Boisgérault, F., Tamouza, R., Charron, D., Toubert, A. (1997). HLA-B27 peptide motif: Tyr C-terminal anchor is not shared by all disease associted subtypes. Immunogenetics. 47:103-105.

-89- REFERENCIAS

Townsend, A., Bastin, J., Gould, K., Brownlee, G., Andrew, M., Coupar, B., Boyle, D., Chan, S., Smith, G. (1988). Defective presentation to class I- restricted cytotoxic T lymphocytes in vaccinia-infected cells is overcome by enhanced degradation of antigen. J. Exp. Med. 168: 1211-24. Townsend, A., Öhlen, C., Bastin, J., Ljunggren, H.G., Foster, L., Kärre, K. (1989). Association of class I major histocompatibility heavy and light chains induced by viral peptides. Nature. 340: 443-448. Trowsdale, J., Hanson, I., Mockridge, I., Beck, S., Townsend, A., Kelly, A. (1990). Sequences encoded in the class II region of the MHC related to the “ABC” superfamily transporters. Nature. 348: 741-744.

U Uebel, S., Kraas, W., Kienle, S., Wiesmuller, K.H., Jung, G., Tampé, R. (1997). Recognition principle of the TAP transporter disclosed by combinatorial peptide libraries. Proc.Natl.Acad.Sci.U.S.A. 94: 8976-8981.

Ugrinovic, S., Mertz, A., Wu, P., Braun, J., Sieper, J. (1997). A single nonamer from the Yersinia 60 Kd heat shock protein is the target of HLA-B27 restricted CTL response in Yersinia induced reactive arthritis. J. Immunol. 159: 5715-5723.

V van Binnendijk, R.S., Versteeg-van Oosten, J.P., Poelen, M.C., Brugghe, H.F., Hoogerhout, P., Osterhaus, A.D., UytdeHaag, F.G. (1993). Human HLA class I- and HLA class II-restricted cloned cytotoxic T lymphocytes identify a cluster of epitopes on the measles virus fusion protein. J. Virol. 67: 2276-2284.

van der Burg, S.H., Visseren, M.J., Brandt, R.M., Kast, W.M., Melief, C.J. (1996). Immunogenicity of peptides bound to MHC class Imolecules depends on the MHC-peptide complex stability. J. Immunol. 156: 3308-3314.

van Endert, P.M., Riganelli, D., Greco, G., Fleischhauer, K., Sidney, J., Sette, A., Bach, J.F. (1995). The peptide-binding motif for the human transporter associated with antigen processing. J.Exp.Med. 182: 1883-1895.

Varshavsky, A. (1992). The N-end rule. Cell. 69: 725-735.

Vega, M.A., Ezquerra, A., Rojo, S., Aparicio, P., Bragado, R., López de Castro, J.A. (1985a). Structural analysis of an HLA-B27 functional variant: identification of residues that contribute to the specificity of recognition by cytolytic T lymphocytes. Proc Natl Acad Sci U S A. 82: 7394-7398. Vega, M.A., Wallace, L., Rojo, S., Bragado, R., Aparicio, P., López de Castro, J.A. (1985b). Delineation of functional sites in HLA-B27 antigens. Molecular analysis of HLA-B27 variant Wewak I defined by cytolytic T lymphocytes. J Immunol. 135: 3323-3332. Vega, M.A., Bragado, R., Ivanyi, P., Peláez, J.L., López de Castro, J.A. (1986). Molecular analysis of a functional subtype of HLA-B27. A possible evolutionary pathway for HLA-B27 polymorphism. J Immunol. 137: 3557-3565.

Vilches, C., de Pablo, R., Kreisler, M. (1994a) Nucleotide sequence of HLA-B*2706. Immunogenetics. 39:219. Vilches, C., de Pablo, R., Herrero, M.J., Moreno, M.E., Kreisler, M. (1994b). HLA-B73: an atypical HLA-B molecule carrying a Bw6-epitope motif variant and a B pocket identical to HLA-B27. Immunogenetics. 40: 166. Villadangos, J.A., Galocha, B., García, Hoyo, R., López, D., García, F., López de Castro, J.A. (1994). Structure of HLA-B27-specific T cell epitopes. Antigen presentation in B*2703 is limited mostly to a subset of the antigenic determinants on B*2705. Eur.J. Immunol. 24: 2548-2555. Villadangos, J.A., Galocha, B., García, F., Albar, J.P., López de Castro, J.A. (1995). Modulation of peptide binding by HLA-B27 polymorphism in pockets A and B, and peptide specificity of B*2703. Eur. J. Immunol. 25: 2370-2377.

W Wei, M.L., Cresswell, P. (1992). Molecules in an antigen-processing mutant contain signal sequence-derived peptides. Nature. 356: 443-446. Wei, X., Orr, H.T. (1990). Differential expression of HLA-E, HLA-F and HLA-G transcripts in human tissue. Hum. Immunol. 29: 131-142.

REFERENCIAS -90-

Weiss, A., Littman, D.R. (1994). Signal transduction by Lymphocyte antigen receptors. Cell. 76: 263-274.

Y Yewdell, J., Lapham, C., Bacik, I., Spies, T., Bennink, J. (1994). MHC-encoded proteasoma subunits LMP-2 and LMP-7 are not required for efficient antigen presentation. J. Immunol. 152: 1163-1170. Young, A.C., Zhang, W., Sacchettini, J.C., Nathenson, S.G. (1994). The three-dimensional structure of H-2Db at 2.4Å resolution: implications for antigen-determinant selection. Cell. 76: 39-50.

Z Zhang, W., Young, A.C., Imarai, M., Nathenson, S.G., Sacchettini, J.C. (1992). Crystal structure of the MHC class I H-2Kb molecule containing a single viral peptide: implications for peptide binding and T cell receptor recognition. Proc. Natl. Acad. Sci. U. S. A. 89: 8403-8407. Zhou, M., Sayad, A., Simmons, W.A., Jones, R.C., Maika, S.D., Satumtira, N., Dorris, M.L., Gaskell, S.J., Bordoli, R.S., Sartor R.B., Slaughter, C.A., Richardson, J.A., Hammer, R.E., Taurog, J.D. (1998). The specificity of peptides bound to human histocompatibility leukocyte antigen (HLA)-B27 influences the prevalence of arthritis in HLA-B27 transgenic rats. J. Exp. Med. 18: 8877-886. Zinkernagel, R.M., Doherty, P.C. (1974) Immunological surveillance against altered self components by sensitised T lymphocytes in lymphocytic choriomeningitis. Nature. 251: 547-548.

-91-

VIII. ANEXOS

ANEXOS -93-

VIII. ANEXOS

-Anexo 1-

Galocha B., Lamas, J.R., Villadangos J.A., Albar J.P., López de Castro J.A. (1996). Binding of peptides naturally presented by HLA-B27 to the differentially disease-associated B*2704 and B*2706 subtypes, and to mutants mimicking their polymorphism. Tissue Antigens. 48: 509-518.

-Anexo 5- Lamas, J.R., Paradela, A., Roncal, F., López de Castro, J.A. (1999). The peptide specificity of HLA-B27 subtypes differentially associated to ankylosing spondylitis is modulated at multiple anchor positions. (aceptado en Arthritis and Rheumatism).

-Anexo 2- García, F., Galocha, B., Villadangos, J.A, Lamas, J.R., Albar, J.P., Marina, A., López de Castro, J.A. (1997). HLA-B27 (B*2701) specificity for peptides lacking Arg2 is determined by polymorphism outside the B pocket. Tissue Antigens. 49: 580-587.

-Anexo 6-

Krebs, S., Lamas, J.R., Poenaru, S., Folkers, G., López de Castro, J.A., Seebach, D., Rognan, D. (1998). Substituting nonpeptidic spacers for the T Cell Receptor-binding part of class I Major Histocompatibility Complex-binding peptides. J.Biol. Chem. 273: 19072-19079.

-Anexo 3- Rognan, D., Krebs, S., Kuonen, O., Lamas, J.R., López de Castro, J.A., Folkers, G. (1997). Fine specificity of antigen for two clas I major histocompatibility protein alleles (B*2705 and B*2703) differing in one amino acid. J.Comput. Aid. Mol. Des. 11: 463-478.

-Anexo 7- Poenaru, S., Lamas, J.R., Folkers, G., López de Castro, J.A., Seebach, D., Rognan, D. (1998). Nonapeptide analogues containing (R)-3-hydroxybutanoate and β-homoalanine oligomers: synthesis and binding affinity to a class I MHC protein. (enviado a J. Med. Chem).

-Anexo 4- Lamas, J.R., Brooks, J.M. Galocha, B. Rickinson, A.B., López de Castro, J.A. (1998). Relationship between peptide binding and T-cell epitope selection: a study with subtypes of HLA-B27. International Immunology. 10: 259-266.

-Anexo 8- García, F., Rognan, D., Lamas, J.R., Marina, A., López de Castro, J.A. (1998). An HLA-B27 polymorphism (B*2710) that is critical for T-cell recognition has limited effects on peptide specificity. Tissue Antigens. 51:1-9.

ANEXO -1-

ANEXO -2-

ANEXO -3-

Journal of Computer-Aided Molecular Design, 11 (1997) 463–478. 463

*Dedicated to Prof. D. Seebach on the occasion of his 60th birthday.**To whom correspondence should be addressed.

KLUWER/ESCOM© 1997 Kluwer Academic Publishers. Printed in The Netherlands.

J-CAMD 410

Fine specificity of antigen binding to two class I major histocompatibilityproteins (B*2705 and B*2703) differing in a single amino acid residue*

Didier Rognana,**, Stefan Krebsa, Oliver Kuonena, José R. Lamasb,José A. López de Castrob and Gerd Folkersa

aDepartment of Pharmacy, Swiss Federal Institute of Technology, Winterthurerstrasse 190, CH-8057 Zürich, SwitzerlandbCentro de Biologia Molecular Severo Ochoa, Universidad Autonoma de Madrid, Facultad de Ciencas, E-28049 Madrid, Spain

Received 10 March 1997Accepted 26 May 1997

Keywords: MHC; HLA-B27; Drug design; Molecular dynamics simulations

Summary

Starting from the X-ray structure of a class I major histocompatibility complex (MHC)-encoded protein(HLA-B*2705), a naturally presented self-nonapeptide and two synthetic analogues were simulated inthe binding groove of two human leukocyte antigen (HLA) alleles (B*2703 and B*2705) differing in asingle amino acid residue. After 200 ps molecular dynamics simulations of the solvated HLA–peptidepairs, some molecular properties of the complexes (distances between ligand and protein center ofmasses, atomic fluctuations, buried versus accessible surface areas, hydrogen-bond frequencies) allowa clear discrimination of potent from weak MHC binders. The binding specificity of the three nonapep-tides for the two HLA alleles could be explained by the disruption of one hydrogen-bonding networkin the binding pocket of the HLA-B*2705 protein where the single mutation occurs. Rearrangementsof interactions in the B pocket, which binds the side chain of peptidic residue 2, and a weakening ofinteractions involving the C-terminal end of the peptide also took place. In addition, extension of thepeptide backbone using a β-Ala analogue did not abolish binding to any of the two HLA-B27 subtypes,but increased the selectivity for B*2703, as expected from the larger peptide binding groove in thissubtype. A better understanding of the atomic details involved in peptide selection by closely relatedHLA alleles is of crucial importance for unraveling the molecular features linking particular HLA allelesto autoimmune diseases, and for the identification of antigenic peptides triggering such pathologies.

Introduction

Major histocompatibility complex (MHC)-encodedclass I proteins play a major role in the immune surveil-lance of intracellular pathogens by presenting antigenicpeptides to cytotoxic T-lymphocytes (CTLs) at the surfaceof infected cells [1]. In the last decade, tremendous re-search efforts have been made to delineate the molecularaspects of antigen presentation to CTLs. Three majorbreakthroughs in this field were the description of thefirst class I MHC crystal structure [2], the identificationof allele-specific motifs for naturally bound peptides [3],and the prominent role played by MHC-encoded TAPheterodimers for antigen transport [4]. Starting from thesethree major observations, experimental data on antigen

presentation to MHC proteins are presently accumulatingat an incredible pace. The huge amount of moleculardetails have made human leukocyte antigen (HLA) pro-teins very interesting drug design targets for two reasons.First, it is likely that the three-dimensional (3D) structuresof all class I and class II alleles are very similar, and thatthe MHC binding mode is rather conserved for most of thepresented peptides, within each HLA class [5,6]. Reliable3D pictures of MHC–peptide complexes are now accessibleand can be interpreted with respect to biological data [7–10]. Second, the expression of certain MHC alleles is asso-ciated with either resistance or susceptibility to humanimmunological diseases like ankylosing spondylitis [11],diabetes [12], rheumatoid arthritis [13] or malaria [14].

One of the strongest linkages known to date between

464

the expression of one class I HLA allele and susceptibilityto a pathology is that of HLA-B27 to inflammatory dis-eases of the joints called spondyloarthropathies [15,16].For example, more than 95% of patients suffering fromankylosing spondylitis are HLA-B27 positive, while thisMHC type is only expressed in 7% of the healthy popula-tion [11]. Interestingly, among the 11 HLA-B27 alleles(HLA-B*2701 to HLA-B*2711) reported to date, at leasttwo are not associated with spondyloarthropathies: HLA-B*2706 [17,18] and HLA-B*2709 [19]. All subtypes differamong each other only at a few residues, mostly locatedin the peptide binding groove.

Deciphering the molecular parameters responsible forthe binding of antigenic peptides to the different HLA-B27 subtypes is an absolute prerequisite for better under-standing their differential association with susceptibilityto spondyloarthropathy, and to identify the sequence ofpotential arthritogenic peptides [15] that may trigger thedisease. Sequencing of peptides naturally presented by thedifferent alleles [20–26] shows that Arg at position 2 (P2)is a conserved motif for peptides binding to all subtypes.Gln2 is an additional motif among B*2701-bound pep-tides [25], but is a suboptimal residue (not found in vivo)for other subtypes [27–30]. Two other anchoring positionscould be disclosed from the frequency of occurrence ofamino acids in peptides naturally bound to B*2705. Hy-drophobic amino acids are preferred at P3, and hydro-phobic as well as positively charged residues can be foundat P9. Other positions are more variable and probablyindicate a less important binding role. These observationsare compatible with the crystal structure of one allele(HLA-B*2705) in complex with a peptide pool [31,32]. Asfor all other class I alleles crystallized to date, boundligands are mainly nonapeptides strongly hydrogen-bonded in a sequence-independent manner at their N-and C-termini to both ends of the MHC binding groove.The central part of the peptide (from P4 to P8) bulges outof the binding groove (Fig. 1) [32]. Some peptide sidechains (P2, P3 and P9) are responsible for the allele speci-ficity of the recognition process, by binding to comple-mentary pockets of the MHC. Notably, the conservedArg2 of B27-bound peptides is perfectly centered in apolar subsite (pocket B) composed of MHC polymorphicresidues (Tyr7, His9, Thr24, Glu45, Cys67). HLA-B*2705 isthe only B27 subtype that has been crystallized up tonow. Furthermore, although peptide binding motifs havebeen described for various HLA-B27 subtypes, the exactpeptide repertoire selected by each allele is still underinvestigation, and the exact contribution of individualamino acids at the nine peptide positions to the varioussubtypes remains largely unknown.

HLA-B*2703 contains a single point mutation(Tyr59→His) in a subsite (pocket A) responsible for thebinding of the peptide N-terminal residue. HLA-B*2703probably selects a subset of peptides presented by the

common HLA-B*2705 allele [33,34]. Positively chargedresidues (Lys, Arg and His) at P1 have been proposed tobe one of the main characteristics of this subset commonto both subtypes as the replacement of Ser1 by Arg re-stored the binding of a B*2705-restricted viral epitope toHLA-B*2703 [35]. In addition, basic residues were pre-dominant among self-peptides naturally bound to B*2703[23]. Additional interactions given by basic side chainscould compensate for the weaker interaction of the pep-tide N-terminus to B*2703 pocket A. However, this is nota stringent requirement because amino acids other thanArg/Lys at P1 (Ala for example) are compatible with agood binding to this allele [28].

To rationalize these data, molecular dynamics (MD)simulations of both alleles in complex with three differentpeptides were undertaken to delineate similarities and dif-ferences in MHC binding, which may explain specificityvariations.

Materials and Methods

Coordinates setupStarting coordinates were taken from the crystal struc-

ture of HLA-B*2705, solved at 2.1 Å resolution [32] anddeposited in the Brookhaven Protein Databank [36] withthe entry 1HSA. In order to save computational time,only the antigen-binding α1–α2 domains were taken intoaccount in the study. This approximation was previouslyshown not to alter the accuracy of MD simulations[9,37,38] because only limited interactions exist betweenthe α1-α2 part and the other two domains (α3 and β2m)that do not significantly contact antigenic peptides in thebinding groove. This observation has been validated bythe crystal structure of one class I MHC protein lackingthe membrane-proximal α3 domain, for which a con-served 3D fold of the α1–α2 antigen-binding domain wasreported [39]. The C-terminal residue of the α2 domain(Thr182) was protected by an N-methyl group to avoidunrealistic electrostatic interactions. The HLA-B*2703subtype was obtained from the HLA-B*2705 crystalstructure by mutating Tyr59 into His, without changingthe direction of the side chain. All MHC-bound nonamerswere built from the peptide (ARAAAAAAA) modeled inthe original crystal structure [32] by substituting the cor-responding residue for alanine without altering the direc-tion of the side chains. Six crystal water molecules wereexplicitly taken into account, as they are located in thepeptide binding cleft and bridge the binding of the pep-tide to the protein X-ray structure. Polar hydrogen atomswere then added and the complexes were centered in a 7.5Å thick shell of TIP3P water molecules [40] without posi-tional constraint on solvent atoms. Any water atom closerthan 1.75 Å to any solute atom was discarded, so thatapproximately 1500 water molecules were added to eachMHC–peptide binary complex.

465

Parametrization of the N-terminal b-alanine monomerThe N-terminal β-alanine was parametrized for

AMBER 4.0 [41] using a previously described procedure[42]. Briefly, atomic coordinates for β-alanine were ob-tained from the SYBYL biopolymer dictionary [43] andoptimized by semiempirical quantum mechanics (MOPAC6.0) using the PM3 Hamiltonian [44]. Potential-derivedatomic charges were then computed on this geometry bya single-point SCF calculation at the MNDO level [45].As new atom types were not defined, nonexisting forceconstants have been assigned according to availableAMBER values for closely related bond, angle and di-hedral types. They were incorporated into the parm91parameter set.

Molecular mechanics and dynamics simulationsAll computations were performed on a CRAY J90

using the AMBER 4.0 program [41] and the united-atomrepresentation of the parm91 parameter set. Since explicitwater molecules were taken into account, a dielectricconstant of 1 was used for all calculations. To avoidsplitting dipoles, nonbonded interactions were calculatedwithin a residue-based cutoff of 10 Å. The solvent atomswere first relaxed by 1000 steps of steepest descent energyminimization, the solute being held fixed. The solvatedcomplex was then fully minimized by 1000 steps of steep-est descent, followed by a conjugate gradient minimiza-tion procedure until the rms gradient of the potentialenergy was less than 0.25 kcal mol−1 Å−1. The minimizedcoordinates were thereafter used as a starting point for anMD simulation at constant temperature. Initial velocitieswere taken from a Maxwellian distribution at 50 K andan integration step of 2 fs was used. The system wasprogressively heated from 50 to 297 K during the firstpicosecond, the temperature being held at 297 K for therest of the simulation by coupling the system to a heatbath [46] using a temperature coupling constant of 0.05ps. All bond lengths were constrained to their equilibriumvalues using the SHAKE algorithm [47] with a bondlength tolerance of 2.5 × 10−4 Å. Coordinates, energies andvelocities were collected and saved every 250 steps (0.5 ps)for 200 ps. The analyses of MD trajectories were achievedusing in-house routines and the CARNAL module ofAMBER [41].

Calculation of electrostatic interaction energiesElectrostatic free energies were computed by solving

the linear form of the Poisson–Boltzmann equation usingthe finite-difference method [48,49] of the DelPhi program[50,51]. Peptides, MHC proteins and MHC–peptide com-plexes were centered in three-dimensional boxes withresolutions of 2.0, 1.20 and 1.10 grid points per Å, re-spectively. For each calculation, 90% of the box was filledwith the corresponding molecule. Atomic radii andcharges were taken from the AMBER 4.0 united-atom

parameter set [41]. Inner and outer dielectrics were as-signed values of 2.0 and 1.0 (vacuum) or 80 (water envi-ronment). An ionic strength of 0.145 M and an ion exclu-sion radius (Stern layer) of 2.0 Å were used according topreviously reported solvent calculations [51]. A proberadius of 1.8 Å was utilized for computing the surface atwhich the electrostatic potential was extrapolated.

Peptide synthesisPeptides 1 and 3 were synthesized as previously de-

scribed [28]. Peptides 2 and 4 were synthesized by auto-mated, multiple solid-phase peptide synthesis with a robotsystem (Syro, MultiSynTech, Bochum, Germany) usingan Fmoc/tBu strategy. For side-chain protection, Tyr(tert-butyl), Ser(tert-butyl), Thr(tert-butyl), Glu(tert-butyl),Gln(trityl), Arg(2,2,3,5,5-pentamethyl-chromansulfonyl)and Lys(tert-butyloxy-carbonyl) were used. The N-ter-minal residues were obtained using single couplings withdiisopropylcarbodiimide/1-hydroxy-benzotriazole activa-tion, 10-fold excess and a coupling time of 1 h on the 2-chlorotritylchloride resin. The peptides were cleaved withtrifluoroacetic acid/thioanisole/thiocresol (20:1:1) within3 h, collected by centrifugation and lyophilized fromwater.

They were then purified by reversed-phase HPLC(Merck-Hitachi, Darmstadt, Germany) on a nucleosil 5µM/C18 column (125 × 3 mm) at a flow rate of 600 µl/min.The absorbance was measured at 220 nm. The solventsystem used consisted of 0.1% trifluoroacetic acid inwater (A) and 0.1% trifluoroacetic acid in acetonitrile (B).A linear gradient from 10 to 60% B in 30 min was ap-plied. The peptides were purified to homogeneity by asecond HPLC on a versapack 10 µm C18 column (300 ×7.8 mm) at a flow rate of 2 ml/min with the same buffersystem, and a linear gradient from 0 to 40% B in 35 minfollowed by a 40–60% B linear gradient for 20 min. Fur-thermore, the peptides were analyzed by ion spray massspectrometry on a triple-quadrupole mass spectrometerAPI III with a mass range of m/z = 10–2400 equippedwith an ion spray interface (Sciex, Thornhill, ON,Canada), and quantified by amino acid analysis using a6300 amino acid analyser (Beckman, Palo Alto, CA,U.S.A.).

Peptide binding assayThe quantitative assay used has been described previ-

ously [29]. Briefly, RMA-S transfectants expressingB*2705 or B*2703 were used. These are murine cells withimpaired TAP-mediated peptide transport and low sur-face expression of (empty) class I MHC molecules, whichcan be induced at 26 °C [52] and stabilized at the cellsurface through the binding of exogenously added pep-tides. These cells were incubated at 26 °C for 24 h. Afterthis, they were incubated for 1 h at 26 °C with 10−4–10−9

M peptides, transferred to 37 °C and collected for flow

466

microcytometry (FMC) analysis with the ME1 mAb

P1 P2

P3

P4

P5

P6

P7

P8 P9

A

B

C

DE

F

Fig. 1. Orientation of a canonical nonapeptide (ball-and-stick model) in the binding groove of HLA-B27 (cyan ribbons) [32]. Peptide positionsare labeled at the Cα atoms from 1 (P1) to 9 (P9). MHC specificity pockets (A to F) are displayed according to the usual nomenclature [72]. Thefigure has been obtained with the MOLSCRIPT [73] and Raster3D [74] programs.

TABLE 1RELATIVE BINDING OF THREE NONAPEPTIDES TO TWOHLA-B27 SUBTYPES

Peptide Sequence Relative bindinga

B*2703 B*2705

1 RRYQKSTELb 1 (2×10−6) 1 (2×10−6)2 ARYQKSTEL 1.5 23 RQYQKSTEL >>100 10

a Data are expressed as the molar excess of peptide analogue, relativeto the wild-type peptide 1, at which HLA-B27 fluorescence (meas-ured by FMC analysis with an anti-B27 monoclonal antibody) onRMA-S cells was half the maximum obtained with peptide 1. Themolar concentration of peptide 1 at 50% maximum fluorescence(C50) is given in parentheses.

b Human histone H3 peptide: a self-peptide, naturally bound to HLA-B*2705 [20] and to B*2703 [23]. Dominant anchor residues are dis-played in boldface.

(IgG1, specific for HLA-B27, B7 and B22) [53] after 4 hfor B*2705, or after 2 h for B*2703. The determinantrecognized by ME1 is not affected by bound peptides orby HLA-B27 polymorphism (data not shown). The bind-ing of a given peptide was measured as its C50. This is itsmolar concentration at 50% of the fluorescence obtainedwith that peptide at 10−4 M. Peptides with C50 ≤ 5 µMwere considered to bind with high affinity, as these werethe values obtained for most of the natural B27-boundligands. C50 values between 5 and 50 µM were consideredto reflect intermediate affinity. C50 ≥ 50 µM indicated lowaffinity. Peptides having C50>100 µM were assumed notto bind. The binding of peptide analogues was measuredas the concentration of the peptide analogue required toobtain the fluorescence value at the C50 of the unchangedpeptide. This was designated as EC50. Relative bindingwas expressed as the ratio between the EC50 of the pep-tide analogue and the C50 of the corresponding unchangedpeptide.

Results and Discussion

A naturally bound nonapeptide (1) from the humanhistone H3 protein was taken as reference for its equalbinding to both subtypes (Table 1). The Ala1 analogue (2)was chosen for its unexpected high affinity for both sub-types [28,35]. Finally, the Gln2 analogue 3 of the human

histone peptide was also studied, in order to investigatethe influence of pocket B–P2 interactions that may alsoparticipate in the peptide discrimination [28].

Dynamical properties of MHC–peptide complexesMonitoring instantaneous rms deviations (rmsd’s) of

protein atoms from the starting structure is usually per-formed to ascertain the reliability of MD simulations [54].For both alleles complexed to peptides 1–3, similar rmsdvalues were observed (about 1.7 Å for backbone atoms;

467

data not shown), indicating that protein distortion upon

0 50 100 150 2007.0

7.5

8.0

8.5

9.0

9.5

10.0 1a, 1b2a, 2b3a, 3b

Dis

tanc

e, Å

Time, psFig. 2. Time course of the distance (in Å) between the center of mass of peptides 1–3 and that of the host HLA-B27 subtype (a: B*2703; b:B*2705).

ligand binding cannot account for the different bindingaffinities of these peptides. Stable rmsd values (1–1.5 Å)were also observed for bound peptides. Interestingly, theyremain in the same range as that observed for differentpeptides co-crystallized with the same MHC host protein[55,56]. One exception concerns the weakest binder (pep-tide 3 to HLA-B*2703), for which larger distortions (upto 2.5 Å) were observed and still increasing after 200 pssimulation.

To follow a possible dissociation of peptides 1–3 fromthe two HLA-B27 alleles, the distance between ligand andprotein center of mass (cmass) was monitored. For bind-ing peptides (1, 2), whatever the B27 subtype, this inter-molecular distance increases during the warm-up phasefrom 8 to 8.5 Å and remains constant for the rest of thesimulation (Fig. 2). The weak binder (peptide 3) exhibitsa slight but continuous increase of the intermoleculardistance, suggesting a partial dissociation of the ligandfrom the binding groove. This phenomenon was signifi-cantly enhanced for the less stable complex (3a: peptide3 in complex with HLA-B*2703). However, localizationof the dissociating peptide amino acids is not possiblewith this analysis. For example, TcR-binding residuesmay bulge even more out of the peptide binding site andinduce similar shifts in the intermolecular distance be-tween center of masses.

Therefore, this analysis was extended by computing thecmass of bound peptide substructures (MHC anchors: P1,P2, P3, P9; TcR anchors: P4, P5, P6, P7, P8) for eachMD conformation and the distance relating it to the

protein cmass (Table 2). An examination of d2 and d3intermolecular distances for the weak binder 3 clearlyshows that MHC-anchoring amino acids only are pro-gressively expelled from the binding groove. The dissocia-tion is more significant when compound 3 is complexedto HLA-B*2703, which relates well with the observedbinding data (Table 1). The TcR-binding region (P4–P8)is similarly bulging out (at least quantitatively) of thebinding cleft for all the six studied complexes (d3 dis-tance, Table 2). An even more precise analysis has beendone by evaluating the inter-cmass distance betweenindividual MHC anchor residues and their complemen-tary pocket (A, B, D and F; d4–d7 distances, Table 2).Surprisingly, the repulsion noticed for peptide 3 in com-plex with both alleles seems to be located at theP9–pocket F interaction level (d7 = 7.3 Å), far from whereprotein and peptide single-point mutations occur. Interest-ingly, the critical d5 distance, illustrating the strength ofthe most important interaction between the invariant Arg2

and pocket B, is higher (5.1 ± 0.4 Å) for the less stablecomplex (3a) than for other pairs (Table 2). The differentd4 distances reported here, especially for peptides 1 and3 and peptide 2, are related to the size of the correspon-ding P1 side chain (Arg versus Ala). It indicates that theArg1 side chain is pointing away from its binding subsite(pocket A) and therefore induces higher d4 inter-cmassdistances. However, for a given P1 side chain, higher d4distances are always observed for HLA-B*2703. Thisindicates that the Tyr59→His mutation found in the latterallele is detrimental for a stable and strong binding of P1side chains to pocket A of the protein. The highest stan-

468

dard deviations are found for the P3–pocket D interac-

TABLE 2DISTANCE BETWEEN PROTEIN AND PEPTIDE CENTER OF MASSES

Distance (Å) Peptide

1a 1b 2a 2b 3a 3b

d1 08.2 ± 0.3 08.5 ± 0.3 08.3 ± 0.3 08.2 ± 0.4 09.3 ± 0.9 09.1 ± 0.7

d2 05.9 ± 0.3 05.9 ± 0.3 05.5 ± 0.3 05.6 ± 0.4 07.3 ± 0.5 06.6 ± 0.5

d3 11.5 ± 0.4 12.1 ± 0.5 12.3 ± 0.5 10.7 ± 0.5 12.0 ± 0.6 12.3 ± 0.6

d4 03.6 ± 0.3 02.9 ± 0.4 02.3 ± 0.3 01.5 ± 0.5 03.4 ± 0.4 02.8 ± 0.3

d5 04.7 ± 0.2 04.7 ± 0.3 04.7 ± 0.3 04.4 ± 0.3 05.1 ± 0.4 04.7 ± 0.3

d6 04.9 ± 0.6 04.8 ± 0.5 05.3 ± 0.5 05.0 ± 0.7 04.6 ± 0.5 05.2 ± 0.6

d7 02.5 ± 0.4 02.6 ± 0.3 02.9 ± 0.3 03.4 ± 0.4 07.3 ± 0.6 04.8 ± 0.5

d1: protein–peptide; d2: protein–MHC anchors (P1–P3, P9); d3: protein–TcR anchors (P4–P8); d4: pocket A–P1; d5: pocket B–P2; d6: pocketD–P3; d7: pocket F–P9.

P1 P2 P3 P4 P5 P6 P7 P8 P9

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

1a, 1b2a, 2b3a, 3b

rmsf

,Å

PnFig. 3. Rms atomic fluctuations of peptides 1–3 when bound to an HLA-B27 protein (a: B*2703; b: B*2705). Pn represents the peptide position(from 1 to 9).

tion (d6, Table 2). The 0.6 Å variations, observed in suchnonbonded distances, correspond to the alternative estab-lishment of strong and weak hydrophobic contacts, whichmay be explained by the topology of the binding groove.Pocket D is a hydrophobic subsite open to the centralpart of the binding cleft, and partially filled by hydro-phobic side chains [20]. The significantly higher varianceof the d6 distance (Table 2) suggests that this interactionis the most flexible one and that different local conforma-tions at P3 are compatible with a good occupancy ofpocket D. Retrospectively, it explains why the P3–pocketD interaction may be important for an optimal peptidebinding to HLA-B*2705. Bulky hydrophobic nonnaturalside chains (α- and β-naphthylalanine, cyclohexylalanine)have been shown to significantly enhance binding to the

HLA-B*2705 protein [38,57]. This positive effect probablyresults from two correlated components: (i) the existenceof additional contacts to pocket D, a pure enthalpic ef-fect; and (ii) a reduced flexibility of the MHC-bound P3side chain, a favorable entropic effect.

Atomic fluctuationsAtomic fluctuations of MHC-bound peptides were com-

puted from mean conformations, time-averaged over thelast 50 ps (Fig. 3). As expected, MHC anchors are muchless flexible than the TcR-binding middle part. If atomicmobility of the bound ligand is considered, it is not poss-ible to depict real differences in the binding of peptides 1and 2 to the two subtypes. However, the C-terminal an-chor residue clearly tends to be more flexible when thecorresponding ligand does not strongly bind to the B27

469

subtype. The highest flexibilities are observed for the

P1 P2 P3 P4 P5 P6 P7 P8 P90.0

0.5

1.0

1.5

2.0

1a, 1b2a, 2b3a, 3b

Acc

essi

ble

/B

urie

d

PnFig. 4. Ratio of accessible to buried surface area for relaxed MD time-averaged conformations. Surface areas were calculated with the MS program[75] using a 1.4 Å probe radius. High ratios were truncated to a value of 2.0. Pn represents the peptide position (from 1 to 9).

1a 1b 2a 2b 3a 3b0

5

10

15

20

25

30

H-b

onds

Num

ber

Complex

>50%

25-50%

Fig. 5. MHC–peptide H-bonding frequency for peptides 1–3 in complex with HLA-B27 subtype (a:B*2703; b: B*2705). H-bonds have beengeometrically defined by an acceptor (A) to donor (D) distance less than 3.25 Å and a D-H..A angle greater than 120°. Interactions werestatistically monitored throughout the simulations for a total of 400 conformations per MHC–peptide complex. Two categories of H-bonds weredefined: strong ones with frequencies higher than 50% and medium ones with occurrences between 25 and 50%.

weakest binder (peptide 3 to B*2703), especially at theimportant anchoring positions (P2, P3 and P9). It is logi-cal to find that the mutation at P2 (for peptide 3)

weakens the interactions to MHC pocket B that has beendesigned to accommodate an arginine side chain [32].However, the P2 amino acid is more flexible when thecorresponding peptide is complexed to HLA-B*2703

470

(compare 3a and 3b, Fig. 3). Again, the most striking

Fig. 6. Crystal structure of HLA-B*2705 [32]. The view is focused on MHC pocket A (Met5, Tyr7, Tyr59, Glu63, Tyr159, Trp167, Tyr171) and pocketB (His9, Thr24, Glu45, Tyr99) side chains interacting via H-bonds (direct H-bonds: green broken lines; water-mediated H-bonds: yellow broken lines)with the P1–P2 positions of a bound peptide. The following color coding has been used: carbon, white (protein) or orange (peptide); nitrogen,blue; oxygen, red; sulfur, yellow. Bound water molecules are shown as cyan balls. Arrows indicate the direction of the H-bonds (from the donorto the acceptor). HLA-B*2703 was obtained by mutating Tyr59 into His (SYBYL Biopolymer module) [43]. The side-chain χ2 dihedral was justmodified here in order to bring the Nε atom as close as possible to the bound water molecule. However, whatever the rotamer chosen, a directH-bond to a water molecule is not possible. The interatomic distance between the peptide N-terminus and Tyr59 (OH) or His59 (Nε2) atoms is 4.31and 5.35 Å, respectively. Figures 6–8 have been obtained by using the rendering program Raster 3D [74].

differences are not observed at the variable amino acidsof the peptides (P1, P2) but at the C-terminal anchor(P9), a feature already noticed for other MHC–peptidecomplexes [9]. One may hypothesize that the N-terminaltripeptide (P1-P2-P3) determines the stability of peptide–MHC interactions over the whole length of the bindinggroove by controlling the conformational space accessibleto the bulging middle part and, consequently, the bindingcapacity of the following C-terminus. However, atomicflexibilities of the P1–P3 and P4–P8 parts are not interre-lated. It is also possible that these differences are relatedto the short time scales (200 ps) used for simulating thecomplexes and that much longer simulations are neededto see significant molecular differences at positions P1and P2 of the peptide ligand. Nanosecond MD simula-tions of macromolecules are nowadays feasible [58,59],but still remain unrealistic as a structure–activity relation-ship tool for comparing a series of ligands in their pro-tein-bound state.

Accessible versus buried surface areasWhether peptide flexibility correlates with dissociation

from the binding cleft was addressed by looking at access-

ible and buried surface areas of each HLA-bound peptideresidue (Fig. 4). Only position 9 of peptide 3 in complexwith B*2703 was much more accessible than the others.Otherwise, the main anchor P2 was similarly buried what-ever the peptide and the host HLA protein. This meansthat a partial dissociation was only observed for oneposition (P9) in one complex (3a) and that the previouslyreported higher flexibility of Gln2 for the same peptide–MHC complex did not correspond with a release of theP2 side chain from pocket B. It may however influencethe quality and frequency of hydrogen bonds between theGln2 side chain and pocket B of both B27 alleles. Theinfluence of secondary anchor positions is more difficultto ascertain. P3 is similarly buried for all MHC–peptidecomplexes. P6 (Ser) and P7 (Thr) positions are probablyaccessory anchor positions that marginally bind, in someconformations, to the central part of the peptide bindinggroove (pockets C/E). This observation is not incompat-ible with the high atomic fluctuations of P6-P7 aminoacids (Fig. 3), as their side chains are directed towards thebinding groove but without reaching its floor. P4 (Gln),P5 (Lys) and P8 (Glu) residues are potential candidatesfor TcR recognition because of their concomitant atomicflexibility and surface accessibility.

471

Qualitative and quantitative analysis of peptide–MHC H-

Fig. 7. MD model of complex 3a, focused on MHC pockets A–B (see the legend to Fig. 6). A mean conformation was averaged from the last 100conformers and submitted to 500 steps of steepest descent, followed by 1500 steps of conjugate-gradient energy relaxation.

Fig. 8. MD model of complex 3b, focused on MHC pockets A–B (see the legend to Fig. 6). The mean conformation was obtained as describedin Fig. 7.

bondsReporting the number of MHC–peptide H-bonds as

well as their frequencies during the simulation allows a

clear distinction between peptides 1 and 2 and the Gln2

analogue (peptide 3, Fig. 5). The stability of protein–peptide H-bonds was assessed by computing the frequencyof occurrence of the interaction throughout the MD tra-

472

jectory (400 conformations). A frequency higher than 50%

TABLE 3MHC–PEPTIDE H-BONDS WITH A FREQUENCY HIGHERTHAN 50%

Pn atom HLA-B27 1a 1b 2a 2b 3a 3b

P1 N Tyr7 (OH) 1 1 1His59 (NE2) 1 1 1Glu45 (OE2) 3Glu63 (OE1) 1 1 1Glu63 (OE2) 1 1Glu163 (OE2) 3

NE Glu163 (OE2) 1 × × 1NH2 Glu63 (OE1) × × 3

Glu63 (OE2) × × 3O Tyr159 (OH) 1 1 1 1

P2 N Glu63 (OE1) 1 1Glu63 (OE2) 1 1

NE Glu45 (OE2) 3 × ×

Glu63 (OE1) 3 × ×

OE1 Tyr99 (OH) × × × × 3NE2 His9 (NE2) × × × × 3

Glu45 (OE2) × × × × 3NH1 His9 (NE2) 1 1 1 × ×

Thr24 (OG1) 3 × ×

NH2 Thr24 (OG1) 1 1 1 × ×

Glu45 (OE1) 1 1 1 × ×

Glu45 (OE2) 1 1 1 × ×

Glu63 (OE1) 3 × ×

P3 N Tyr99 (OH) 1 1 1 1P6 OG Ala69 (O) 1

Thr73 (OG1) 1P8 O Trp147 (NE1) 1 1 1

OE1 Lys146 (OE1) 1P9 N Asp77 (OD1) 1 1 1 1 1

OXT Tyr84 (OH) 1Thr143 (OG1) 1 1 1Lys146 (NZ) 1 1 1

Empty boxes indicate interactions that are common to at least twocomplexes, whereas filled boxes represent unique MHC–peptide hy-drogen bonds. The absence of a specific side chain is featured by across.

was chosen to characterize strong H-bonds. Medium in-teractions were assigned a frequency between 25 and 50%.

About 25 H-bonds have been identified for peptides 1and 2 in complex with B*2703 and B*2705 while 50% lesscould be found for the Gln2 analogue with the two sub-types (Fig. 5). The distribution of strong and medium H-bonds correlates well with the binding potency of thepeptide. A similar number of strong H-bonds were found

for complexes 1a, 1b, 2a and 2b, consistent with the simi-lar binding efficiencies of peptides 1 and 2 to both sub-types. On the other hand, a reduced number of mediumand/or strong H-bonds (peptide 3 in complex with thetwo alleles) correlates with the decreased binding of thispeptide. The weakest binding potency (peptide 3 toB*2703) could effectively be qualitatively and quantitat-ively related to the distribution of intermolecular H-bonds. Not only the number but also the quality of theMHC–ligand interactions correlates well with the bindingpotency.

To accurately localize the interactions that may explainpeptide specificity variations, all H-bonds with frequencieshigher than 50% were identified for the six complexes(Table 3). The first noticeable difference between the twoHLA-B27 alleles is the H-bonding network between theMHC residues involved in binding to the peptide P1position. In HLA-B*2705, two amino acid side chains areH-bonded to the peptide N-terminus (Tyr7/Tyr171 in thecrystal structure, Tyr7/Glu63 in the MD models) (Table 4,Figs. 6–8). Both side chains are fixed by a subtle water-relayed H-bond network involving proximal MHC sidechains (Tyr59, Glu45, Tyr171). Tyr59 is directly bound toTyr171, and indirectly to Tyr7, Glu45 and Glu63. The singlepoint mutation occurring for HLA-B*2703 (Tyr59→His)perturbs this network. The bound water molecule disap-pears and the peptide N-terminus binds to His59 and nomore to Tyr7 (Table 3, Fig. 7). The consequence on theH-bond balance is a loss of one direct MHC–MHC inter-action (His59 cannot interact with Tyr171) and five water-mediated interactions for complex 3a (Fig. 7). The result-ing conformational change may be well accommodated asfar as P2 is strongly bound to pocket B (His9, Thr24,Glu45) and the resulting H-bonds are strong enough tomaintain the peptide in the binding groove (P2=Arg). IfP2 is not an arginine (peptide 3), the resulting interactionto pocket B (Thr24 and Glu45 notably) is much weakerand the conformational rearrangement at P1 is important(see the three new H-bonds for the P1 position in com-plex 3a, Fig. 7). The Arg to Gln change at the P2 posi-tion of the bound peptide is better tolerated by HLA-B*2705 (complex 3b, Fig. 8) as the Tyr59 side chain is stillable to fix the position of Tyr7. During the MD simula-tion, the N-terminal Cα-N bond of the bound peptide hasrotated to gain a new H-bond to Glu63. However, it isstill bound to Tyr7 as in the reference structure. Import-antly, the Gln2 side chain is bound to Glu45 and Tyr99,thus providing additional interactions to pocket B whencompared to complex 3a (Fig. 8). The quality of theinteraction between Gln2 and pocket B is, however, muchinferior to that observed for peptide analogues 1 and 2bearing an optimal Arg residue (Table 4), thus explainingthe reduced binding affinity of peptide 3 for HLA-B*2705.

For the set of peptides studied here, the advantage of

473

Arg over Ala at P1 could be quantified by the gain of

P1 P2 P3 P4 P5 P6 P7 P8 P90.0

0.5

1.0

1.5

2.02a2b4a4b

Acc

essi

ble/

buri

ed

PnFig. 9. Accessible versus buried surface areas of peptides 2 and 4 in complex with B*2703 (a) and B*2705 (b) alleles (see the legend to Fig. 4).

2a 2b 4a 4b0

5

10

15

20

25

30

H-B

onds

Complex

> 50%25- 50%

Fig. 10. Intermolecular hydrogen bonds for peptides 2 and 4 in complex with B*2703 (a) and B*2705 (b) alleles (see the legend to Fig. 5).

two water accessible salt bridges to Glu63/Glu163 (Table 3).However, this does not correspond to a higher bindingaffinity of the Arg analogue when compared to the Ala1

peptide. An Ala side chain is much easier to desolvateand optimally interacts with conserved apolar residues ofpocket A (Met5, Trp167), thus explaining a rather similarbinding affinity of peptides 1 and 2 to both subtypes.

However, the present model cannot fully explain recentdata, indicating that basic residues are overrepresented atthe P1 position of B*2703-bound natural ligands [23].From a purely statistical point of view, various rotamersof basic P1 side chains could develop a salt bridge with atleast three negatively charged amino acids located at therim of pocket A (Glu58, Glu63, Glu163), and thus stabilizethe MHC–peptide complex.

474

MHC–peptide interaction energies

TABLE 4MHC–PEPTIDE INTERACTION ENERGIES CALCULATED FROM THE LINEAR POISSON–BOLTZMANN EQUATION ANDAMBER FORCE-FIELD CALCULATIONS

Peptide ∆G0coul

a ∆G0reac

b ∆G0elec

c ∆Helecd ∆Hvdw

e ∆Gtotf ∆Htot

g

1a −499 487 −12 −87 −90 −−102 −−1771b −542 498 −44 −79 −97 −−141 −−1762a −359 340 −19 −72 −77 0−−96 −−1492b −389 335 −54 −75 −80 −−134 −−1553a −306 334 −28 −44 −68 0−−40 −−1123b −371 380 −09 −62 −67 0−−58 −−129

a ∆G0coul: Coulombic component of MHC–peptide electrostatic interaction energy (charge–charge, charge–dipole, dipole–dipole interactions). ∆G0

coul

= ∆G0coul(P–L) − ∆G0

coul(P) − ∆G0coul(L) [50], where P–L describes the protein–ligand complex, P the protein and L the ligand.

b ∆G0reac: corrected self-reaction field component of MHC–peptide electrostatic interaction energies (energy required to transfer a molecule from

a continuum dielectric (vacuum) to another (water). ∆G0reac = ∆G0

reac(P–L) − ∆G0reac(P) − ∆G0

reac(L). As the contribution of the protein–ligand complexes(∆G0

reac(P–L)) and of the isolated protein (∆G0reac(P)) could be omitted from the calculation without affecting the reliability of the results [61], this

component corresponds here to the free energy of peptide desolvation (−∆G0reac(L)).

c ∆G0elec: total electrostatic interaction energy (∆G0

coul + ∆G0reac).

d ∆Helec: AMBER electrostatic interaction energy (ε = 4rij).e ∆Hvdw: AMBER van der Waals interaction energy. ∆Hvdw = A/r12 − B/r6, where r is the distance between atom pairs and A and B are atom-type-

dependent parameters.f ∆Gtot: total interaction energy (∆G0

elec + ∆Hvdw).g DHtot: total AMBER interaction enthalpy (∆Helec + ∆Hvdw).

TABLE 5INFLUENCE OF A P1 β-AMINO ACID ON THE HLA-B27SUBTYPE SELECTIVITY OF A MODIFIED HLA-B27LIGAND

Peptide Sequence EC50 (µM)a

number B*2703 B*2705

(P1-RYQKSTEL)2 P1=Ala 3.0 04.04 P1=Balb 7.5 20

a Concentration of the peptide (in µM) at which HLA-B27 fluor-escence (measured by FMC analysis with an anti-B27 monoclonalantibody) on RMA-S cells was half the maximum obtained with thewild-type peptide (peptide 1, Table 1).

b Bal: β-alanine (H2N-CH2-CH2-CO).

Interaction energies were extrapolated for all six ener-gy-minimized time-averaged conformations (Table 4) bysumming up the van der Waals nonbonded interactionenergy (calculated with the AMBER 4.0 force field) andthe electrostatic component (calculated by solving thelinear form of the Poisson–Boltzmann equation), as re-cently described [60]. As both ligands and protein struc-tures are very similar for all complexes, distortion ener-gies as well as translational/rotational entropy losses uponbinding were neglected here. Moreover, the self-reactionfield energy component of the electrostatic interactionenergy was limited to the contribution of the isolatedpeptide (free energy of desolvation) and calculated fromthe bound-peptide coordinates extracted from the MHC–peptide binary complexes. It has recently been shown thatneglecting the protein contribution to the self-reactionfield energy is indeed possible and does not alter thereliability of the obtained results [61].

Our computational protocol is able to properly rankthe binding of the three peptides 1–3 to both MHC al-leles. Peptide 3 clearly interacts much weaker than theother two peptides 1 and 2, whatever the MHC allele.The weakest interaction energy was observed for bindingof 3 to B*2703, and is thus in agreement with bindingdata (Table 1). Force-field interaction enthalpies (calcu-lated by summing up both AMBER van der Waals andelectrostatic components, using a dielectric permittivity of4rij) were much less related to the observed binding data,as peptide 2 was always disfavoured with regard to pep-tide 1 (Table 4). Notably, taking into account the peptidedesolvation energy by the continuum eletrostatics methodpermits to compensate for the weakest Coulombic interac-tions provided by peptide 2 (P1=Ala) to both alleles, with

respect to that observed for the Arg1 analogue. It may benoticed that the free electrostatic interaction energies(∆G0

elec, Table 4) computed by the continuum electrostaticsmethod were also in rather good qualitative agreementwith the binding data reported in Table 1. Hence, thethree peptides are highly polar and interact mainly via H-bonds and salt bridges.

For an even more realistic ranking of highly polar li-gands than those presented here, free energy perturbation[62] is probably the method of choice. Unfortunately, theenormous amount of CPU time that would be necessaryfor this computation precludes its systematic use in fastscreening of a set of congeneric molecules. Synthesis andin vitro binding assays in this case provided a faster andexperimentally determined answer.

MD simulation of MHC–peptide complexes couldrelate observed binding potencies and allele specificity tosimple molecular criteria (inter-cmass distances, atomicfluctuations, accessible surface areas, distribution and

475

location of intermolecular H-bonds). A single point muta-

Bal1

Arg2

Tyr3

Gln4 Lys5

Ser6

Thr7Glu8

Leu9

Tyr84

Asp77Thr80

Leu81Leu95

Lys146

Trp147

Val152

Thr143Tyr123

Asp116

His114

Leu156Leu160

Tyr159

Glu163Trp167Tyr171

His59Glu63

Glu45

Thr24His9

Tyr99

Tyr7

Fig. 11. Energy-minimized time-averaged MD model of complex 4a. The MHC protein backbone is displayed as a solid cyan tube, with peptide-inter-acting side chains. The bound peptide 4 is represented by sticks. The following color coding has been used: carbon, white (protein) or green (peptide);nitrogen, blue; oxygen, red; sulfur, yellow. Bound water molecules are shown as cyan balls. Yellow broken lines indicate MHC–peptide H-bonds.

Bal1

Arg2

Tyr3

Gln4Lys5

Ser6Thr7

Glu8

Leu9

Thr80

Leu81Asp77

Leu95

Trp147Lys146

Thr143

Ty123Asp116

His114Val152

Leu156

Leu160

Tyr159

Tyr99

Tyr7

His9

Thr24

Glu163

Trp167Ty171

Tyr59

Glu63

Glu45

Tyr84

Fig. 12. Energy-minimized time-averaged MD model of complex 4b (see the legend to Fig. 11).

tion in the HLA binding groove is sufficient to break anH-bond network in the vicinity of the peptide N-terminus.As previously suggested on the basis of peptide binding

analyses [28], this minor change strengthens even morethe binding role of the dominant anchor P2 side chain(Arg) for one allele (HLA-B*2703) and explains whychanging P2 to Gln has more detrimental effects in pep-

476

tide binding for HLA-B*2703 than for HLA-B*2705,where compensatory stabilization of the MHC–peptidecomplex is still possible by H-bonded MHC side chains.Interestingly, the most spectacular consequence of pro-tein/peptide variability affects an area far away (10 Å)from the location of the point mutations. It concerns thestability of the interaction between the peptidic C-ter-minal residue and its complementary pocket F, which hasrecently been shown to play a decisive role in linkingparticular B27 alleles to spondyloarthropathies [24,26].

Protein-based designRelating the structure of B27 subtypes to the sequence

of their naturally bound peptides is a crucial step inidentifying potential immunodominant epitopes that maydiscriminate alleles and confer susceptibility or resistanceto autoimmune diseases. One striking feature concerns thesingle point mutation (Tyr59His) distinguishing HLA-B*2705 from HLA-B*2703, which is unique among HLAproteins. It is believed that B*2703 selects a subset of thepeptides presented by HLA-B*2705 [34]. Recent studieshave identified some peptides that are naturally presentedby both subtypes, and at least one natural B*2705 ligand(the undecamer RRYLENGKETL) is not presented byB*2703 [23,63]. The only difference between both allelesconcerns the position 59 located in pocket A which inter-acts with the N-terminal amino acid of the bound peptide(Fig. 6). As pocket A is slightly wider for HLA-B*2703,extending the peptide backbone towards His59 by replacingthe natural P1 residue by a β-amino acid should theoreti-cally allow a better discrimination of both alleles. Thisstructural change should be much better accommodated byHLA-B*2703 (H-bond between His59 and the N-terminusof the P1 β-amino acid) than by HLA-B*2705, for whicha steric clash with the Tyr59 side chain may be expected.

Starting from the self-peptide 1 (RRYQKSTEL) nat-urally presented by HLA-B*2705 [20] and B*2703 [23],Ala and Bal (β-alanine) were substituted for the naturalArg at P1 (Table 5). The Ala1 peptide analogue was heretaken as a reference for its strong binding to both sub-types. The two ligands were synthesized and tested fortheir binding to B*2703 and B*2705. As expected fromthe topology of the binding cleft, only the Bal analoguecould discriminate between the two subtypes, with a bet-ter binding to HLA-B*2703 (Table 5).

To rationalize the experimental binding data, the non-natural ligand 4 was simulated a posteriori, in complexwith both alleles, using exactly the same conditions asthose employed for simulating the natural MHC–peptidecomplexes (see the section Computational procedures).Using two of the previously described molecular parame-ters (accessible versus buried surface area of the boundligand, intermolecular H-bonds) as quality control of thecomplex stability, peptide 4 was indeed found to be muchbetter accommodated by B*2703 than by B*2705 (Figs.

9 and 10). Interestingly, the weak binding of the β-Alapeptide to B*2705 could also be related to a partial disso-ciation of the C-terminus from its complementary pocketF (Fig. 9), far away from the peptide mutation site. Thepresent data, in agreement with previous MD simulationsof different MHC–peptide complexes [9], suggest that theexpulsion of the C-terminus from pocket F could be thevery first event in the dissociation of weak binding pep-tides from class I MHC binding grooves. The modified P1position is, however, significantly more buried when thehost protein is the B*2703 allele (Fig. 9). A qualitativeand quantitative analysis of intermolecular hydrogenbonds also supports the reported binding data. A total of15 H-bonds could be depicted for complex 4b (peptide 4in complex with B*2705), whereas 26 interactions havebeen found for complex 4a (peptide 4 in complex withB*2703, Fig. 10). However, this analysis was unable toexplain the reduced affinity of the β-Ala compound forB*2703, when compared to that of the natural Ala ana-logue 2 (Table 4). The slight differences seen in the epi-tope stabilization assay are certainly too subtle for theshort MD runs reported here. They probably result from:(i) the absence of a side chain at position P1 of ligand 4;and (ii) a weaker binding contribution of the bulgingP4–P8 part, for which higher atomic fluctuations (datanot shown) and less nonbonded contacts (see the highsolvent accessibility of the P5 and P8 residues for ligand4, Fig. 9) have been noticed. Energy-minimized time-averaged conformations of both complexes (Figs. 11 and12) clearly depict significant differences in the MHCpocket A (Tyr59, Trp167, Tyr171), which deviates dramati-cally from the starting crystal coordinates for B*2705only (rmsd values from all pocket A atoms of 1.5 and 2.5Å for B*2703 and B*2705, respectively). The major con-formational alterations upon Bal1 binding were observedfor the Tyr59-Trp167-Tyr171 triad (rmsd values of 1.7 and2.7 Å for B*2703 and B*2705, respectively)sl. As pre-dicted, the Tyr59 side chain was shifted away from thepeptide N-terminus and is now interacting via a watermolecule with the β-alanine terminal ammonium (Fig.12). In contrast, the β-amino acid can directly interactwith the larger pocket A of B*2703 through three H-bonds to His59 (mutated position), Glu63 and Glu163 (Fig.11). Another significant difference in the binding of pep-tide 4 to both alleles concerns the C-terminal amino acid,which has nearly lost, upon binding to B*2705, all H-bonds to the polar side chains of pocket F (Tyr84, Thr143,Lys146; compare Figs. 11 and 12). The incorporation of aβ-amino acid at P1 has modified the above described H-bond network between MHC side chains and the peptideN-terminus. The bound water molecule located in pocketA of HLA-B*2705 (recall Fig. 6) has either disappeared(B*2703, Fig. 11) or has been shifted towards the extremeleft end of the binding groove (B*2705, Fig. 12).

More importantly, these results show that the incor-

477

poration of a β-amino acid in the peptide sequence doesnot abrogate binding to HLA-B27 subtypes. Peptide 4 isone of the very first ligands for which the backbonemodification of an anchor residue does not abolish classI MHC binding. Up to now, only a retro-inverso (NH-CO instead of CO-NH) and a reduced peptide bond(CH2-NH instead of CO-NH) pseudopeptide analogue ofan HLA-A2-binding peptide have been proposed as suc-cessful P1 modifications [64,65]. However, a β-amino acidat P1 presents the advantage to preserve the backbonedirection of the peptide ligand and the H-bonding capac-ity of the first peptide bond (to Glu63 and Tyr159), so thatless 3D conformational changes of the MHC binding cleftare necessary to accommodate the modified ligand. Therecently described X-ray structure of an MHC–peptide–TcR ternary complex [66] suggests that the latter featuremay be of particular importance for a proper recognitionof the MHC–ligand pair by a TcR. Furthermore, it opensthe door to the incorporation of β-amino acids at otheranchor positions, notably P2, P3 and P9. Potential TcR-binding amino acids have already been replaced by vari-ous organic spacers without affecting the binding of thecorresponding ligand to class I MHC proteins [38,67,68].The present design study demonstrates that substitutinga β-amino acid for a natural residue is a further solutionfor designing high-affinity MHC ligands with improvedstability and pharmacokinetic properties. This is an abso-lute prerequisite for the therapeutic use of MHC ligandseither as MHC blockers [69] or as T-cell receptor antag-onists [70].

Conclusions

MD simulations have been used in the present study asa tool for explaining peculiar structure–activity relation-ships at the level of the protein–ligand interaction com-plexes. The current study is not aimed at quantitativelyranking MHC ligands and predicting their binding affin-ities. For that purpose, free energy calculations usingmuch longer equilibration and conformational samplingwould be necessary. More simply, dynamical properties ofthe modeled complexes can be qualitatively well relatedto known binding data. Notably, monitoring protein–ligand intramolecular distances, the atomic mobility ofthe bound ligands, the ratio of accessible versus buriedsurface areas, the history and the quality of peptide–protein H-bonds allow a clear discrimination of high-affinity from weak-binding peptides. This computationalapproach, based on the qualitative analysis of short MDtrajectories, has already been used to succesfully predictthe bound conformation of a natural HLA-A2-restrictedepitope [37] prior to X-ray structure determination [55],to identify T-cell epitopes from the primary structure ofpotentially interesting proteins [9] and to design high-affinity nonnatural ligands [38,71]. Herewith, we propose

its application to the rationalization of peptide specificityfor closely related HLA alleles and the design of non-natural ligands with increased specificity for one HLA-B27 subtype. Identifying the molecular rules, fine tuningpeptide selection by HLA alleles is a crucial step forbetter understanding the peptide–HLA interactions thatmay confer either susceptibility or resistance to immuno-logical diseases associated with particular HLA alleles.

Acknowledgments

D.R. wishes to thank the computational center of theETH Zürich for generous allocation of computer time onthe CRAY J90 and PARAGON machines. This work wassupported by the Schweizerischer Nationalfonds zurFörderung der wissenschaftlichen Forschung (Project No.31-45504.95) and by Grant SAF 94-0891 from the PlanNacional de I+D to J.A.L.C. J.R.L. is a fellow of theBasque Government.

References

1 Heemels, M.T. and Ploegh, H.L., Annu. Rev. Biochem., 64 (1995)643.

2 Bjorkman, P.J., Saper, M.A., Samraoui, B., Bennet, W.S.,Strominger, J.L. and Wiley, D.C., Nature, 329 (1987) 506.

3 Falk, K., Rötzschke, O., Stevanovic, S., Jung, G. and Rammen-see, H.-G., Nature, 351 (1991) 290.

4 Spies, T., Bresnahan, M., Bahram, S., Arnold, D., Blank, G.,Mellins, E., Pious, D. and DeMars, R., Nature, 348 (1990) 744.

5 Stern, L.J. and Wiley, D.C., Structure, 2 (1994) 245.6 Madden, D.R., Annu. Rev. Immunol., 13 (1995) 587.7 Corr, M., Boyd, L.F., Frankel, S.R., Kozlowski, S., Padlan, E.A.

and Margulies, D.H., J. Exp. Med., 176 (1992) 1681.8 Huczo, E.L., Bodnar, W.M., Benjamin, D., Sakaguchi, K., Zhu,

N.Z., Shabanowitz, J., Henderson, R.A., Appella, E., Hunt, D.F.and Engelhard, D., J. Immunol., 151 (1993) 2572.

9 Rognan, D., Scapozza, L., Folkers, G. and Daser, A., Biochemis-try, 33 (1994) 11476.

10 Chelvanayagam, G., Jakobsen, I.B., Gao, X. and Easteal, S.,Protein Eng., 9 (1996) 1151.

11 Brewerton, D.A., Hart, F.D., Nicholls, A. and Sturrock, R.D.,Lancet 2, (1973) 994.

12 Todd, J.A., Bell, J.I. and McDevitt, H.O., Nature, 329 (1987) 599.13 Wordsworth, B.P., Lanchbury, J.S., Sakkas, L.I., Welsh, K.I.,

Panayi, G.S. and Bell, J.I., Proc. Natl. Acad. Sci. USA, 86 (1989)10049.

14 Hill, A.V.S., Elvin, J., Willis, A.C., Aidoo, M., Allsopp, C.E.M.,Gotch, F.M., Gao, X.M., Takiguchi, M., Greenwood, B.M.,Townsend, A.R.M., McMichael, A.J. and Whittle, H.C., Nature,360 (1992) 434.

15 Benjamin, R. and Parham, P., Immunol. Today, 11 (1990) 137.16 Kingsley, G. and Sieper, J., Immunol. Today, 14 (1993) 387.17 López-Larrea, C., Sujirachato, K., Mehra, N.K., Chiewsilp, P.,

Isarangkura, D., Kanga, O., Dominguez, O., Coto, E., Peña, M.,Setién, F. and Gonzales-Roces, S., Tissue Antigens, 45 (1995) 169.

18 Nasution, A.R., Mardjuadi, A., Kunmartini, S., Suryadhana,N.G., Setyohadi, B., Sudarsono, D., Lardy, N.M. and Feltkamp,T.E.W., J. Rheumatol., 24 (1997) 1111.

478

19 D’Amato, M., Fiorillo, M.T., Carcassi, C., Mathieu, A.,Zuccarelli, A., Bitti, P.P., Tosi, R. and Sorrentino, R., Eur. J.Immunol., 25 (1995) 3199.

20 Jardetzky, T.S., Lane, W.S., Robinson, R.A., Madden, D.R. andWiley, D.C., Nature, 353 (1991) 326.

21 Rojo, S., Garcia, F., Villadangos, J.A. and López de Castro, J.A.,J. Exp. Med., 177 (1993) 613.

22 Rötzchke, O., Falk, K., Stevanovic, S., Gnau, V., Jung, G. andRammensee, H.G., Immunogenetics, 39 (1994) 74.

23 Boisgérault, F., Tieng, V., Stolzenberg, M.C., Dulphy, N., Khalil,I., Tamouza, R., Charron, D. and Toubert, A., J. Clin. Invest., 98(1996) 2764.

24 Garcia, F., Marina, A. and López de Castro, J.A., TissueAntigens, 49 (1997) 215.

25 Garcia, F., Galocha, B., Villadangos, J.A., Lamas, J.R., Albar,J.P., Marina, A. and López de Castro, J.A., Tissue Antigens, 49(1997) 580.

26 Fiorillo, M.T., Meadows, L., D’Amato, M., Shabanowitz, J.,Hunt, D.F., Apella, E. and Sorrentino, R., Eur. J. Immunol., 27(1997) 368.

27 Parker, K.C., Biddison, W.E. and Coligan, J.E., Biochemistry, 33(1994) 7736.

28 Villadangos, J.A., Galocha, B., Garcia, F., Albar, J.P. and Lópezde Castro, J.A., Eur. J. Immunol., 25 (1995) 2370.

29 Galocha, B., Lamas, J.R., Villadangos, J.A., Albar, J.P. andLópez de Castro, J.A., Tissue Antigens, 48 (1996) 509.

30 Raghavan, M., Lebron, J., Johnson, J. and Bjorkman, P., ProteinSci., 5 (1996) 2080.

31 Madden, D.R., Gorga, J.C., Strominger, J.L. and Wiley, D.C.,Nature, 353 (1991) 321.

32 Madden, D.R., Gorga, J.C., Strominger, J.L. and Wiley, D.C.,Cell, 70 (1992) 1035.

33 López, D., García-Hoyo, R. and López de Castro, J.A., J.Immunol., 152 (1994) 5557.

34 Villadangos, J.A., Galocha, B., García-Hoyo, R., López, D., Garcia,F. and López de Castro, J.A., Eur. J. Immunol., 24 (1994) 2548.

35 Colbert, R.A., Rowland-Jones, S.L., McMichael, A.J. andFrelinger, J.A., Immunity, 1 (1994) 121.

36 Bernstein, F.C., Koetzle, T.F., Williams, G.J.B., Meyer Jr., E.F.,Brice, M.D., Rodgers, J.M., Kennard, O., Shimanouchi, T. andTasumi, M., J. Mol. Biol., 112 (1977) 535.

37 Rognan, D., Zimmermann, N., Jung, G. and Folkers, G., Eur. J.Biochem., 208 (1992) 101.

38 Rognan, D., Scapozza, L., Folkers, G. and Daser, A., Proc. Natl.Acad. Sci. USA, 92 (1995) 753.

39 Collins, E.J., Garboczi, D.N., Karpusas, M.N. and Wiley, D.C.,Proc. Natl. Acad. Sci. USA, 92 (1995) 1218.

40 Jorgensen, W.L., Chandrasekhar, J., Madura, J.D., Impey, R.W.and Klein, M.L., J. Chem. Phys., 70 (1983) 926.

41 Pearlman, D.A., Case, D.A., Caldwell, J.C., Seibel, G.L., Singh,U.C., Weiner, P. and Kollman, P.A., AMBER 4.0, University ofCalifornia, San Francisco, CA, U.S.A., 1992.

42 Besler, B.H., Merz, K.M. and Kollman, P.A., J. Comput. Chem.,11 (1990) 431.

43 SYBYL, release 6.2, TRIPOS Associates Inc., St. Louis, MO,U.S.A.

44 Stewart, J.J.P., J. Comput.-Aided Mol. Design, 4 (1990) 1.45 Dewar, M.J.S. and Thiel, W., J. Am. Chem. Soc., 99 (1977) 4899.46 Ryckaert, J.P., Cicotti, G. and Berendsen, H.J.C., J. Comput.

Phys., 23 (1977) 327.

47 Berendsen, H.J.C., Postma, J.P.M., van Gunsteren, W.F., DiNola,A. and Haak, J.R., J. Chem. Phys., 81 (1984) 3684.

48 Klapper, I., Hagstrom, R., Fine, R., Sharp, K. and Honig, B.,Proteins Struct. Funct. Genet., (1986) 47.

49 Warwicker, J. and Watson, H.C., J. Mol. Biol., 157 (1982) 671.50 Gilson, M.K. and Honig, B., Proteins Struct. Funct. Genet., 4

(1988) 7.51 Gilson, M.K., Sharp, K. and Honig, B., J. Comput. Chem., 9

(1988) 327.52 Ljunggren, H.G., Stam, N.J., Ohlen, C., Neefjes, J.J., Hoglund,

P., Heemels, M.T., Bastin, J., Schumacher, T.N., Townsend, A.,Karre, K. and Ploegh, H.L., Nature, 46 (1990) 476.

53 Ellis, S.A., Taylor, C. and McMichael, A., Hum. Immunol., 5(1982) 49.

54 van Gunsteren, W.F. and Berendsen, H.J.C., Angew. Chem., Int.Ed. Engl., 29 (1990) 992.

55 Madden, D.R., Garboczi, D.N. and Wiley, D.C., Cell, 75 (1993)693.

56 Reid, S.W., McAdam, S., Smith, K.J., Klenerman, P.,O’Callaghan, C.A., Harlos, K., Jakobsen, B.K., McMichael, A.J.,Bell, J.I., Stuart, D.I. and Jones, E.Y., J. Exp. Med., 184 (1996)2279.

57 Rovero, P., Vigano, S., Pegorado, S., Revoltella, R., Riganelli, D.,Fruci, D., Greco, G., Butler, R. and Tanigaki, N., J. Pept. Sci., 1(1995) 266.

58 Brunne, R.M., Berndt, K.D., Güntert, P., Wüthrich, K. and vanGunsteren, W.F., Proteins Struct. Funct. Genet., 23 (1995) 49.

59 Fox, T. and Kollman, P.A., Proteins Struct. Funct. Genet., 25(1996) 315.

60 Gallego, J., Ortiz, A.R., de Pascual-Teresa, B. and Gago, F., J.Comput.-Aided Mol. Design, 11 (1997) 114.

61 Taylor, N.R. and von Itzstein, M., J. Comput.-Aided Mol.Design, 10 (1996) 233.

62 Bash, P.A., Singh, U.C., Langridge, R. and Kollman, P.A.,Science, 236 (1987) 564.

63 Garcia, F., Marina, A., Albar, J.P. and López de Castro, J.A.,Tissue Antigens, 49 (1997) 23.

64 Guichard, G., Calbo, S., Muller, S., Kourilsky, P., Briand, J.-P.and Abastado, J.-P., J. Biol. Chem., 270 (1995) 26057.

65 Guichard, G., Connan, F., Graff, R., Ostankovitch, M., Muller,S., Guillet, J.-G., Choppin, F. and Briand, J.-P., J. Med. Chem.,39 (1996) 2030.

66 Garboczi, D.N., Ghosh, P., Utz, U., Fan, Q.R., Biddison, W.E.and Wiley, D.C., Nature, 384 (1996) 134.

67 Weiss, G.A., Collins, E.J., Garboczi, D.N., Wiley, D.C. andSchreiber, S.L., Chem. Biol., 2 (1995) 401.

68 Bouvier, M. and Wiley, D.C., Proc. Natl. Acad. Sci. USA, 93(1996) 4583.

69 Adorini, L., Muller, S., Cardinaux, F., Lehmann, P.V., Falcioni,F. and Nagy, Z.A., Nature, 334 (1988) 623.

70 De Magistris, M.T., Alexander, J., Coggeshall, M., Altman, A.,Gaeta, F.C.A., Grey, H.M. and Sette, A., Cell, 68 (1992) 625.

71 Rognan, D., Habilitationsschrift, ETH Zürich, Zürich, Switzer-land, 1997.

72 Saper, M.A., Bjorkman, P.J. and Wiley, D.C., J. Mol. Biol., 219(1991) 277.

74 Kraulis, P.J., J. Appl. Crystallogr., 24 (1991) 846.75 Merritt, E.A. and Murphy, M.E.P., Acta Crystallogr., D50 (1994)

869.76 Connolly, M.J., J. Appl. Crystallogr., 16 (1983) 548.

ANEXO -4-

ANEXO -5-

ARTHRITIS & RHEUMATISMVol. 42, No. 9, September 1999, pp 1975–1985© 1999, American College of Rheumatology

MODULATION AT MULTIPLE ANCHOR POSITIONS OF THEPEPTIDE SPECIFICITY OF HLA–B27 SUBTYPES DIFFERENTIALLY

ASSOCIATED WITH ANKYLOSING SPONDYLITIS

JOSE R. LAMAS, ALBERTO PARADELA, FERNANDO RONCAL, and JOSE A. LOPEZ DE CASTRO

Objective. To investigate the rules governing pep-tide binding to HLA–B*2705, and to B*2704 andB*2706, which are 2 subtypes differentially associatedwith ankylosing spondylitis.

Methods. Poly-Ala analogs carrying the HLA–B27motif Arg-2, and substitutions at anchor positions P1,P3, or P, were used to determine a binding score foreach residue at each position. Binding was assessed in aquantitative epitope stabilization assay, where the cellsurface expression of HLA–B27 was measured by flowcytometry as a function of peptide concentration.

Results. Peptide anchor residues contributed ad-ditively to B*2705 binding. About 15% of the naturalB*2705 ligands used a deficient P3 or P anchor, butnever both, indicating that detrimental anchoring at oneof these positions is always compensated by a goodanchor at the other one. About 50% of the B*2705ligands used suboptimal P1 residues. However, this wascompensated with optimal P3 and/or P anchoring.Peptides that were longer than decamers used goodanchor residues at the 3 positions, suggesting morestringent binding requirements. B*2704 and B*2706differed in their residue specificity at P1, P3, and P.The rules derived for B*2705 also applied to the knownligands of these 2 subtypes.

Conclusion. The B*2705, B*2704, and B*2706

peptide repertoires are limited by the allowed residuecombinations described in this study. The differentialassociation of B*2704 and B*2706 with spondylarthro-pathy correlates with differences in their peptide spec-ificity at multiple anchor positions. However, it is nowpossible to predict the peptide features that determinethis differential binding to both subtypes.

HLA class I proteins bind endogenous peptidesand present them at the cell surface for recognition bycytotoxic T lymphocytes. These peptides, generally rang-ing in size from 8 to 12 amino acids but with a predom-inance of nonamers and decamers, bind to the class Imolecule through interactions involving the peptidemain chain, both peptide ends, and various peptide sidechains. These interact in cavities or pockets of thepeptide-binding site, which are formed by amino acidresidues of the class I molecule. The structure of thesepockets is modulated by HLA polymorphism, whichaffects the size and polarity of the pockets, and thisdetermines the peptide-binding specificity of each class Iallotype (1).

Typically, the peptide repertoire bound to a givenclass I molecule shows limited diversity at some peptidepositions (P) (2). These so-called main anchor residuesinteract with the HLA molecule and have a significantcontribution to peptide affinity. However, only 30% ofthe nonamers carrying the right main anchor residueswill actually bind a given class I molecule (3). This is dueto the significant influence of auxiliary anchor positions.Thus, the peptide specificity of an HLA class I moleculecannot be understood without knowing the suitability ofdifferent amino acid residues at these auxiliary positions.This is also essential for the prediction of putativeligands.

The peptide-binding specificity of HLA–B27 hasreceived much attention because of the strong associa-tion of this molecule with ankylosing spondylitis (AS)

Supported by grant no. SAF97/0182 from the Plan Nacionalde ID, grant no. PM95-002 from the Spanish Ministry of Education,and an institutional grant from the Fundacion Ramon Areces to theCentro de Biologıa Molecular Severo Ochoa. Dr. Lamas is a fellow ofthe Basque Government.

Jose R. Lamas, Alberto Paradela, Jose A. Lopez de Castro,PhD: Consejo Superior de Investigaciones Cientıficas, and Univer-sidad Autonoma de Madrid, Madrid, Spain; Fernando Roncal:Pharmacia-Consejo Superior de Investigaciones Cientıficas, CentroNacional de Biotecnologıa, Madrid, Spain.

Address reprint requests to Jose A. Lopez de Castro, PhD,Centro de Biologıa Molecular Severo Ochoa, Universidad Autonomade Madrid, Facultad de Ciencias, Cantoblanco, 28049 Madrid, Spain.

Submitted for publication February 22, 1999; accepted inrevised form May 4, 1999.

1975

and reactive arthritis (4–6). Among other mechanisms,it has been proposed that an autoimmune responsetriggered by bacterial infections against a self peptidepresented by HLA–B27 could be a primary pathogeneticevent (7). This hypothesis is supported by recent studiesin transgenic rats (8), and by the differential associationof HLA–B27 subtypes with AS. Multiple HLA–B27subtypes, including those that are predominant in whites(B*2705) and Asians (B*2704), are associated with thisdisease (9). However, B*2709 (10) and B*2706 (11–13)are less or not associated with AS. Thus, the peptidespecificity of HLA–B27 and its alteration by subtypepolymorphism becomes highly relevant to understandthe differential association of closely related subtypeswith AS. Presumably, an arthritogenic peptide should beselectively presented by the various disease-associatedsubtypes to autoimmune T cells that are putativelyinvolved in the pathogenesis of spondylarthropathies.

The main anchor positions of HLA–B27–boundpeptides are P2 and the C-terminal residue (P). Aux-iliary anchor positions include P1, P3, and P7 (14,15).Arg-2 is the main anchor motif for all HLA–B27 sub-types, and the overwhelming majority of natural HLA–B27 ligands have this residue. C-terminal residues aremore variable, being basic, aliphatic, and aromatic inB*2705, B*2703, and B*2710 (14,16–19), and aliphatic/aromatic with subtype variability in B*2701, B*2702,B*2704, B*2706, B*2707, and B*2709 (16,20–23). Se-quence studies of subtype-bound peptide repertoireshave revealed the main anchor motifs and the residuesoccurring naturally at other peptide positions, but pro-vide little information about the relative suitability ofresidues at a given position. Very few studies, all limitedto B*2705 and B*2703 (18,24–27), have addressed therole of auxiliary anchor residues in determining thepeptide specificity of HLA–B27.

In this study, we have systematically explored thespecificity of B*2705, B*2704, and B*2706 for P1, P3,and P residues using series of poly-Ala nonameranalogs carrying Arg-2. The following issues were ad-dressed. First, we evaluated the relative suitability ofresidues at each of the 3 positions for binding to B*2705.This allowed us to rank the contribution of differentresidues at the various positions, providing a novelunderstanding of the peptide specificity of the “proto-type” HLA–B27 molecule. Second, we demonstratedthat binding of a given ligand to B*2705 is usually anadditive function of the contribution of the differentanchor residues, implying that interactive effects amongamino acid residues at different peptide positions do notplay, in general, a significant role. This allows a reliable

prediction of putative HLA–B27 ligands. Third, wedemonstrated that B*2704 and B*2706 differ in theirresidue specificity at the 3 positions, P1, P3, and P.This provides a novel understanding of the functionaldifferences of the 2 subtypes that are differentiallyassociated with AS.

MATERIALS AND METHODS

Synthetic peptides. Peptides were synthesized in anAMS 422 Multiple Peptide Synthesizer (Abimed, Langelfeld,Germany) using Fmoc chemistry, and purified by reverse-phase high-performance liquid chromatography. The correctcomposition and quantitation of the peptides was determinedby amino acid analysis, as previously described (28). Peptideswere stored as stock solutions at 4°C in water.

Cell lines and monoclonal antibodies (mAb). RMA-Sis a mutant cell line of the Rauscher virus–induced murine Tcell lymphoma RBL-5 (H-2b), which has impaired transporter-associated antigen-processing (TAP)–mediated peptide trans-port (29,30) and low surface expression of class I majorhistocompatibility complex antigens that can be induced at26°C (31). RMA-S transfectant cells expressing B*2705,B*2704, or B*2706 plus human 2-microglobulin have beenpreviously described (32,33). The expression levels of B*2705and B*2706 at 26°C were the same, and that of B*2704 was30% lower. These transfectants were grown in RPMI 1640medium containing 25 mM HEPES buffer and 10% fetal calfserum (FCS; all from Life Technologies, Paisley, UK).HMy2.C1R (C1R) is a human lymphoblastoid cell line that haslow expression of its endogenous class I antigens. Transfec-tants of these cells, with high expression of B*2705, werecultured in Dulbecco’s modified Eagle’s medium with 7.5%heat-inactivated FCS. The mAb ME1 (-HLA–B27 B7 B22) (34) was used as undiluted culture supernatant.

Peptide-binding assay. The quantitative epitope stabi-lization assay used has been previously described (32). Briefly,RMA-S transfectant cells were incubated at 26°C for 24 hoursin 96-well plates in RPMI 1640, 25 mM HEPES buffer, and10% FCS. After incubation, plates were washed with sterilephosphate buffered saline, and peptides diluted in RPMI/HEPES medium without FCS were added at a final concen-tration ranging from 10–4 to 10–9M. Cells were then incubatedfor 1 hour at 26°C, transferred at 37°C, and collected for flowcytometric analysis after 4 hours. This time point was chosenbecause there was a significant difference between the HLA–B27–associated fluorescence in the presence of a peptideligand relative to its absence. In addition, peptides differing intheir dissociation rates are more easily distinguished at thistime point. For B*2704, whose peptide-induced stabilizationafter this time was too low, cells were collected after 2 hours at37°C. At this time point, the difference between the HLA–B27–associated fluorescence in the presence or in the absenceof a suitable ligand was comparable with that for B*2705 orB*2706 at 4 hours.

HLA–B27–associated fluorescence on RMA-S trans-fectant cells was plotted as a function of peptide concentration.Binding was calculated as follows. First, a natural HLA–B27ligand (RRYQKSTEL) was chosen as reference, and its molar

1976 LAMAS ET AL

concentration at 50% of the maximum fluorescence obtainedwith that peptide (C50) was calculated. Second, the molarconcentration of each other peptide required to obtain thefluorescence value at the C50 of the reference peptide wasfound by interpolation, and this was designated as the EC50.EC50 values of 5 M indicated high affinity, since thesevalues were generally obtained with natural HLA–B27 ligands.EC50 values 5 M and 50 M were considered to reflectintermediate affinity. EC50 values 50 M indicated lowaffinity. Peptides with an EC50 100 M were considered notto bind, since their affinity was below the detection limits ofthis assay (32). The binding-promoting effect of substitutionsat P1, P3, or P9 was calculated as the binding of the corre-sponding poly-Ala peptide carrying each substitution plusArg-2, relative to the ARAAAAAAA (ARA7) peptide. Thiswas expressed as the ratio between the EC50 of ARA7 and thatof each analog. All calculations were carried out with theOrigin program (MicroCal Software, Northampton, MA).Flow cytometric analysis was carried out as previously de-scribed (28).

Isolation and sequencing of B*2705-bound peptides.Natural B*2705 ligands were isolated by acid extraction ofimmunoaffinity-purified B*2705 from C1R transfectant celllysates as previously described (35). Peptides were sequencedby quadrupole ion trap mass spectrometry using a nanosprayinterface as described elsewhere (36). Assignment of residueswith the same mass (for example, I/L, Q/K) was done on thebasis of unambiguous matching with known human sequencesin the protein database.

RESULTS

Specificity of B*2705 for peptide anchor resi-dues. The role of the P1, P2, and P anchor positions inbinding to B*2705 was analyzed with 3 series of poly-Alanonamers carrying Arg-2 and most of the amino acidresidues at P1 or P3. A more restricted series, mainlyincluding basic, aliphatic, and aromatic residues, wasused for P (P9). The contribution of each residue wasmeasured as the ratio between the EC50 of the corre-sponding poly-Ala analog and that of ARA7. Thispeptide bound to B*2705 with an EC50 of 30 M,reflecting the contribution of the peptide main chain,Arg-2, and Ala side chains.

P1. The most favored residue at position P1 wasArg, so that the RRA7 peptide bound 3-fold betterthan ARA7 (Figure 1). H and aromatic (Y, W, F)residues were roughly equivalent to Ala (relative binding0.7–1.5). A few residues, including K, G, I, and M, wereslightly less favored than Ala (relative binding 0.6).Finally, acidic (D, E), polar (S, T, N, Q), and somealiphatic (L, V) residues were detrimental (relativebinding 0.4).

P3. The binding efficiency of P3 poly-Ala analogsspanned a wide range (Figure 1). The most favored

residue was W (relative binding 6), followed by F andsome aliphatic (L, M) residues (relative binding 3 orhigher). Other aliphatic (V, I), some polar (N, S), H, andY residues were similar to Ala (relative binding 0.7–1.5).A number of chemically diverse residues, including Tand especially R, and acidic (D, E), Q, G, and P residueswere detrimental.

Figure 1. Relative binding of poly-Ala peptide analogs to B*2705.Each analog (x-axis) is represented by the 1-letter code of the aminoacid residue introduced at P1, P3, or P9 into the ARAAAAAAA(ARA7) sequence. The reference peptide RRYQKSTEL (EC50 of 3M) is designated by an asterisk. Binding of each analog was measuredas described in Materials and Methods and its relative binding (y-axis,logarithmic scale) was expressed as the ratio between the EC50 ofARA7 (30 M) and that of the analog. Since the maximum amount ofpeptide tested was 100 M, relative binding lower than 0.3, whichindicates lack of binding in this assay (EC50 100 M), could not bemeasured. For representation purposes only, a value of 0.25 wasassigned to these analogs. Data are the mean of at least 2 independentexperiments.

PEPTIDE SPECIFICITY OF B*2705, B*2704, AND B*2706 1977

P9. Since P9 is usually a basic or nonpolar residueamong B*2705-bound peptides (Table 1), the analysis ofthe contribution of this position to B*2705 binding wasrestricted to these residues. Pro was also tested, since itis a C-terminal motif among HLA–B73–bound peptideswhich, as in HLA–B27, also have Arg-2 (37). As ex-pected, K, R, aliphatic, and Y residues were stronglyfavored. In contrast, F and, even more, W and Pro weredisfavored (Figure 1).

Additive contribution of anchor residues to pep-tide binding. In the next set of experiments, we analyzedwhether binding of a natural B*2705 ligand could beexplained by the additive contribution of anchor resi-dues or whether this binding was a more complexfunction involving interactive effects. Thus, we tested the

binding of 2 natural ligands of B*2705 and of a series ofpoly-Ala analogs carrying 1 or more of the anchorresidues of each ligand. We first measured how muchthe introduction of these anchor residues increased ordecreased binding relative to ARA7. We then calculatedthe ratio between the relative binding of each ligand, oranalog carrying multiple anchor residues of that ligand,and the additive value of the relative binding of poly-Alaanalogs carrying single anchor residues. If the contribu-tion of individual residues is additive, this ratio shouldbe 1. If interactive effects among anchor residues play arole, this ratio should deviate significantly from 1. Twonatural B*2705 ligands and their corresponding analogswere tested: RRYQKSTEL and KRYKSIVKY. Thefirst one was used as the reference peptide to calculate

Table 1. Natural ligands of B*2705 and binding scores of their P1, P3, and P residues*

* Except for the peptides of viral or bacterial origin, all other peptides are from endogenous proteins of the cell. Residues that were scored are shownin boldface type in the sequences. The binding scores are rated 1–5, from high to low, on the basis of the binding efficiency of the correspondingpoly-Ala analogs, as follows: 1 EC50 10 M; 2 EC50 11–20 M; 3 EC50 21–40 M; 4 EC50 41–80 M; 5 EC50 80 M. X scorenot determined. The reference numbers for previously reported ligands are given. Human immunodeficiency virus (HIV)–derived peptides wereobtained from the HIV Molecular Immunology Database of Los Angeles National Laboratory (http://hiv-web.lanl.gov/).† Previously reported as an octamer (ARLQTALL) based on Edman sequencing (see ref. 16). Also found as a nonamer (ARLQTALLV) in B*2709(see ref. 23).‡ Peptide of viral or bacterial origin.§ Reported as a nonamer also in B*2701 (see ref. 20) and as a decamer (RRISGVDRYY) in B*2703 (see ref. 17) and B*2710 (see ref. 19).¶ A natural variant of this peptide with the L6M change is also known.

Ligand Binding score Ref. number

OctamersRRFFPYYV 1-1----1 35RRFTRPEH 1-1----X 14

NonamersARLQTALLV 3-1-----1 This study†RRYQKSTEL 1-2-----1 14SRTPYHVNL 4-4-----1 This studyRRLPIFSRL 1-1-----1 16GRHGVFLEL 4-2-----1 This studyRRIYDLIEL 1-2-----1 44‡RRYPDAVYL 1-2-----1 45‡GRFGSGMNM 4-1-----1 23GRTFIQPNM 4-4-----1 23LRFQSSAVM 5-1-----1 23RRSKEITVR 1-3-----1 14FRYNGLIHR 3-2-----1 14KRFEGLTQR 4-1-----1 14HRAQVIYTR 3-3-----1 23SRYWAIRTR 4-2-----1 14‡RRFMPYYVY 1-1-----1 16SRVKLILEY 4-2-----1 This studyRRFFPYYVY 1-1-----1 17,35RRVLVQVSY 1-2-----1 23RRISGVDRY 1-2-----1 14,16§RRIKEIVKK 1-2-----1 14RRVKEVVKK 1-2-----1 14ARLFGIRAK 3-1-----1 14,16GRIDKPILK 4-2-----1 14GRFEGTSTK 4-1-----1 23GRAFVTIGK 4-3-----1 HIV‡

Ligand Binding score Ref. number

IRLRPGGKK 4-1-----1 HIV‡RRWLPAGDA 1-1-----3 14GRLTKHTKF 4-1-----4 16,17KRFKEANNF 4-1-----4 This studyRRFGDKLNF 1-1-----4 16KRFSFKKSF 4-1-----4 16TRYPILAGH 5-2-----X 16KRVVINKDT 4-2-----X 46‡

DecamersKRFEETGQEL 4-1------1 This studyNRFAGFGIGL 5-1------1 This studyRRQDILDLWI 1-5------1 HIV‡GRFNGQFKTY 4-1------1 17RRYDRKQSGY 1-2------1 23GRWPGSSLYY 4-1------1 23GRKTGQAPGY 4-X------1 23GRILSGVVTK 4-2------1 23RKGGNNKLIK 1-5------1 23LRDNIQGITK 5-5------1 This studyKRWIILGLNK 4-1------1 HIV‡¶RRFVNVVPTF 1-1------4 This studyKRWQAIYKQF 4-1------4 23RRIKEIVKKH 1-2------X 16

UndecamersRRYLENGKETL 1-2-------1 17RRMGPPVGGHR 1-1-------1 16WRLGSSDILNY 2-1-------1 This study

DodecamersRRFVNVVPTFGK 1-1--------1 23

1978 LAMAS ET AL

the EC50 for all other peptides. To account for experi-mental error in the determination of EC50 values, ratiosbetween 1.5 (1.5:1) and 0.67 (1:1.5) were considered toreflect the additive contribution of anchor residues. Thisrange was chosen because it was similar to the differ-ences observed among EC50 values obtained in individ-ual experiments when binding of a given peptide toB*2705 was repeatedly measured.

As shown in Figure 2A, binding of most of thepeptide analogs carrying multiple anchor residues ofRRYQKSTEL (P1, P3, P7, P9) was accounted for, in 4of 5 cases, by the additive contribution of the corre-sponding analogs carrying single anchor residues. Simi-larly (Figure 2B), binding of KRYKSIVKY analogs wasaccounted for, in 4 of 5 cases, by the additive contribu-tion of analogs carrying single or a smaller number ofanchor residues, with 1 case (KRAAAAAAY, ratio 0.6)showing a small deviation from the 0.67–1.5 range.

These results indicate that, in general, the bind-ing efficiency of a given peptide is a simple additivefunction of the contribution of individual anchor resi-dues. However, mutual effects among peptide sidechains may occasionally affect binding.

The joint contribution of the P4–P6 and P8residues was inferred from the ratio between the relativebinding of each peptide and of its corresponding analogcarrying the P1, P2, P3, P7, and P9 residues. Thus, theratio between the binding of RRYQKSTEL relative toARA7 (10-fold) and that of the RRYAAATAL analog(15-fold) was 0.67, indicating little contribution of P4–P6and P8 to binding of the natural ligand (Figure 2A). Inthe second example (Figure 2B), the ratio between thebinding of KRYKSIVKY relative to ARA7 (7.5-fold)and that of the KRYAAAVAY analog (4.3-fold) was1.74, which is slightly outside the 0.67–1.5 range, andtherefore compatible with some contribution of P4–P6and P8 to binding of this natural ligand.

Rules determining usage of anchor residuesamong natural HLA–B27 ligands. After observing theeffect of individual residues at P1, P3, and P onbinding and finding that their contribution was additive,it was possible to address the question of whethernatural B*2705 ligands used only suitable anchor resi-dues or used detrimental ones at these 3 positions. Toexamine this, we started with a database of 54 naturalB*2705 ligands of known sequence, including 10 re-ported for the first time here, consisting of 2 octamers,34 nonamers, 14 decamers, 3 undecamers, and 1 dodec-amer (Table 1). The P1, P3, and P residue of eachligand was assigned a score ranging 1–5 according to thebinding efficiency of the corresponding poly-Ala analog

carrying this residue. Scores 1 and 2 were assigned toresidues whose corresponding poly-Ala analogs boundbetter than ARA7 (EC50 30 M). Score 1 was an EC50 of10 M, and score 2 was an EC50 of 11–20 M. Score 3was assigned to residues whose effect was similar to Ala(EC50 21–40 M). Scores 4 and 5 corresponded toresidues that were less suitable than Ala at the corre-

Figure 2. Relationship between binding of peptides carrying anchorresidues of A, RRYQKSTEL or B, KRYKSIVKY. Solid bars indicatebinding relative to ARAAAAAAA (ARA7), expressed as the molarratio between the EC50 of ARA7 and that of each peptide (x-axis). Forsubstitutions that were detrimental relative to A (T7, K1), the decreaseon binding was expressed as 1 minus the relative binding. Open barsindicate the additive value of the relative binding of poly-Ala analogscarrying single substitutions at P1, P3, P7, or P9. The effect of V7 wasnot analyzed separately. Instead, the ARAAAAVAY analog was used.To account for experimental error in the determination of EC50 values,the data indicated by the solid and open bars were considered equalwhen their ratio was between 0.67 (1:1.5) and 1.5 (1.5:1). Ratio valuesoutside this range are marked with an asterisk. Data are the mean ofat least 2 independent experiments.

PEPTIDE SPECIFICITY OF B*2705, B*2704, AND B*2706 1979

sponding position (score 4 EC50 41–80 M; score 5 EC5080 M).

As shown in Table 1, 44 of the 51 peptides (86%)scored at P had optimal anchor residues (score 1), andonly 7 (14%) had P residues similar or worse than Ala(scores 3 and 4). Similarly, good anchor residues werelargely predominant at P3: 45 of 53 peptides (85%) werescored 1 (51%) or 2 (34%) at this position, and 8peptides (15%) showed P3 residues with a score equal toor worse than that of Ala. In contrast, P1 residues witha score of 1 or 2 occurred in only 24 of 54 peptides(44%), and 26 (48%) had residues less suitable than Ala.This indicates that P1 is more permissive than P3 or P.

Two additional points are worth noting. First, the11-mer and 12-mer peptides had good anchors (score 1or 2) at the 3 positions. This suggests that the peptiderepertoire that deviates from the canonical size of classI ligands (8–10 residues) has a more stringent residuespecificity at P1, P3, and P. Second, all of the peptideswith P residues scoring 3 or higher (5 nonamers and 2decamers) had an optimal (score 1) P3 residue, irrespec-tive of P1. Conversely, P3 anchors scoring 3 or higher,which were observed in 5 nonamers and 3 decamers,always occurred with an optimal (score 1) P residue,also irrespective of P1. These results indicate that defi-cient anchoring at either P3 or P is always compen-sated with an optimal anchor at the other position.

Finally, of the 30 peptides with P1 residuesscoring 3 or higher, 18 (60%) had residues with a scoreof 1 or 2 at both P3 and P, and all had this in at least1 of these 2 positions. Seven peptides had detrimental(score 4 or 5) residues at P1 and at either P3 or P, butin all these cases, the other position had an optimal(score 1) residue. Thus, detrimental P1 residues requireat least an optimal P3 or P anchor.

These rules derived for B*2705 also applied tothe few known natural ligands of B*2704 and B*2706(Table 2) and are likely to apply generally to other classI proteins.

Modulation of P1, P3, and P specificity byB*2704 and B*2706 polymorphism. In these experi-ments, we addressed the effect of B*2704 and B*2706polymorphism on residue selection at these 3 peptidepositions. ARA7 bound to B*2704 and B*2706 moreefficiently (EC50 10 M and 9 M, respectively) than toB*2705. This probably reflects a stronger interaction ofAla residues at 1 or more anchor positions.

P1. The effect of the P1 residue on binding toB*2704 (Figure 3) was similar to that for B*2705, butwith some differences. For instance, G rather than R wasthe most favored residue for B*2704, but its effect on

increasing binding relative to Ala was smaller than theeffect of R1 on B*2705. In addition, G and K wereslightly detrimental for binding to B*2705 (Figure 1), butnot for B*2704 binding. Otherwise, the pattern of P1residues that were equivalent to Ala (H, and aromaticresidues) or detrimental (acidic, polar, and aliphatic)was similar to that for B*2705.

The effect of P1 substitutions on B*2706 binding(Figure 3) was different than that for B*2704, mainly inthat only D was strongly detrimental, whereas E, polar(S, T, N), and aliphatic (V, I, L, M) residues were similarto Ala, and Q was favored, also in contrast to B*2705.

These results indicate that HLA–B27 subtypeswith an identical A pocket have nonidentical residuespecificities.

P3. B*2704 was similar to B*2705 in its accep-tance of H and aliphatic (V, I, L, M) and aromatic (F, Y,W) residues at P3, and in that acidic (D, E), polar (S, T,N, Q), G, and P residues were disfavored (Figure 4). Adifference was that, in contrast to B*2705 (Figure 1),Ala-3 was no worse than bulkier nonpolar residues forbinding to B*2704. In addition, N was less suitable thanAla in B*2704 binding, but not in B*2705 binding.

The specificity of B*2706 for P3 residues showedimportant differences in comparison with B*2704 (Fig-

Table 2. Natural ligands of B*2704 and B*2706 and binding scoresof their P1, P3, and P residues*

Ligand Binding score Ref. number

B*2704RRFFPYYV 1-1----1 36RRYQKSTEL 1-1-----1 21RRIYDLIEL 1-1-----1 44†RRRWRRLTV 1-2-----1 44†GRLTKHTKF 1-1-----4 21QRKKAYADF 3-X-----4 21GRFNGQFKTY 1-1------3 21RRYLENGKETL 1-1-------1 47

B*2706RRLRNHMAV 1-1-----1 21IRHNKDRKV 2-1-----1 21RRHWGGNVL 1-1-----1 21RRYQKSTEL 1-1-----1 21QRKKAYADF 1-X-----2 21RRYLENGKETL 1-1-------1 47

* Except for peptides of viral origin, all other peptides are fromendogenous proteins of the cell. Residues that were scored are shownin boldface type in the sequences. Scores were calculated on the basisof the binding efficiency of the corresponding poly-Ala analogs asdescribed in the text. X score not determined. Binding scores forB*2704 ligands are assigned as for B*2705 (see footnote to Table 1).Scores for B*2706 ligands were assigned as follows: 1 EC50 5 M;2 EC50 6–10 M; 3 EC50 11–20 M; 4 EC50 21–40 M; 5 EC50 40 M.† Peptide of viral origin.

1980 LAMAS ET AL

ure 4). For instance, H, R, and polar residues weresignificantly favored relative to Ala in B*2706, and G, P,and acidic residues were less detrimental. In addition,nonpolar aliphatic and aromatic residues were moresuitable than Ala, but Y was less favored than manyother residues, including other aromatic ones. Ala itselfwas among the best residues in B*2704, but worse thanmany others in B*2706.

These results indicate that B*2704 and B*2706polymorphisms affect P3 specificity, so that these 2subtypes differ from B*2705 and between each other intheir residue preferences at this position.

P9. B*2704 differed from B*2705 in the detri-mental effect of basic and Y residues at P9 (Figure 5). Inaddition, although aliphatic residues were favored onB*2704, this was not significantly above the effect of Ala.

This subtype was similar to B*2705 in the detrimentaleffect of F, W, and P residues.

B*2706 differed from B*2704 in that F was notdetrimental relative to Ala or to other aromatic residues(Figure 5), and in its much stronger preference for bulkyaliphatic residues than for Ala. As in B*2704, Y, W, P,and basic residues were detrimental. However, the det-rimental effect of P was smaller on B*2706, and that ofK was larger, in comparison with B*2704.

DISCUSSION

The strategy used in this study to analyze thepeptide specificity of HLA–B27 was an epitope stabili-zation assay, which measures peptide binding to “emp-ty” HLA–B27 molecules expressed on the surface ofTAP-deficient cells. No in vitro binding assay fullyreproduces peptide loading in vivo, since this is a highlyorganized and incompletely known process requiringphysical association of the TAP transporter, severalchaperones, and the HLA molecule. However, naturalB*2705 ligands consistently bind with high affinity in our

Figure 4. Relative binding of poly-Ala P3 analogs to B*2704 andB*2706. Each analog (x-axis) is represented by the 1-letter code of theamino acid residue introduced at P3 into the ARAAAAAAA (ARA7)sequence. The RRYQKSTEL peptide is designated by an asterisk.Peptide binding is expressed as described in Figure 3. Data are themean of at least 2 independent experiments.

Figure 3. Relative binding of poly-Ala P1 analogs to B*2704 andB*2706. Each analog (x-axis) is represented by the 1-letter code of theamino acid residue introduced at P1 into the ARAAAAAAA (ARA7)sequence. The RRYQKSTEL peptide is designated by an asterisk.Binding of each analog was measured as described in Materials andMethods and its relative binding (y-axis, logarithmic scale) was ex-pressed as the ratio between the EC50 of ARA7 (10 M for B*2704and 9 M for B*2706) and that of the analog. Since the maximumamount of peptide tested was 100 M, relative binding values lowerthan 0.1, which indicate lack of binding in this assay (EC50 100 M),could not be measured. Data are the mean of at least 2 independentexperiments.

PEPTIDE SPECIFICITY OF B*2705, B*2704, AND B*2706 1981

assay (32), suggesting that the peptide specificity ofHLA–B27 as measured in the present study closelyreflects its specificity in vivo. Poly-Ala analogs arecommonly used to compare the contribution of differentamino acid residues at single positions, since the poly-Ala backbone provides a uniform background to whichbinding of peptide analogs can be related. Although theAla side chain may have a contribution to binding atsome peptide positions, its effects are minimized due toits small size and neutral chemical character.

The data concerning peptide binding to B*2705demonstrate that residue variability at P1, P3, or P cansignificantly affect binding. However, on the basis of ourobservations regarding the effect of the best residue ateach of these positions on increasing binding relative toARA7, the importance of these positions can be rankedas P9 P3 P1. Thus, K9 increased binding by a factorof 8, W3 by a factor of 6, and R1 by a factor of 3. Thatmultiple residues were detrimental emphasizes that pep-tide binding is determined by both positive and negativeeffects at individual positions. In particular, detrimental

residues can substantially reduce peptide affinity (1).Thus, the unsuitability of Pro-9 for binding to B*2705 isa critical difference in comparison with HLA–B73,which is an antigen that has the same B pocket asHLA–B27 and also binds peptides with Arg-2, but hasPro as a prominent C-terminal motif (37).

Our study is consistent with a previous one (25)that also used poly-Ala peptide analogs to test the roleof P3 and P9 on B*2705 binding. In particular, itconfirms the suitability of basic and aliphatic residues atP9 and of nonpolar aromatic residues at P3. However,the reported suitability of F9 and relatively low accep-tance of Y9 is in contrast to our results. This differencemight be related to the fact that, in the previous study, arefolding assay was used to assess binding and thepoly-Ala analogs used were different, since, besides R2,they had residues other than Ala at several anchorpositions.

The examination of residues present among nat-ural HLA–B27 ligands reveals that the most suitableresidues are predominant at P3 and P. However, thereis a significant allowance (15%) for suboptimal resi-dues at each of these positions, which is always compen-sated with the presence of an optimal residue at theother position. Because there are virtually no exceptionsto this pattern among the known B*2705 ligands, thisemerges as a significant constraint to the B*2705 peptiderepertoire in applying it to the prediction of naturalligands, in addition to putative additional constraintsimposed by proteasomal cleavage and peptide transport(38–41). That P1 was significantly more permissive forsuboptimal residues among B*2705 ligands supports thesmaller contribution of this position to peptide binding.However, this permissivity was limited by the require-ment of a good anchor residue at P3 and/or P9.

That these 3 positions generally contributed in anadditive way to binding and, together with R2, restoredmuch of the affinity of natural ligands, strongly suggeststhat a reliable prediction of HLA–B27 ligands can bedone on the basis of P1, P3, and P only. In furthersupport of this, nonpeptidic ligands with significantaffinity for an HLA class I molecule can be obtained bykeeping the P1–P3 and P residues of a given peptideligand and substituting organic spacers for P4–P8(42,43). Although the limited number of known ligandslarger than 10-mers makes it somewhat risky to derivegeneral rules, our data might suggest that peptides thatare suboptimal in size (11-mers, 12-mers) bind in vivoonly if they have good anchors at the 3 positions, P1, P3,and P.

A second issue addressed herein was the compar-

Figure 5. Relative binding of poly-Ala P9 analogs to B*2704 andB*2706. Each analog (x-axis) is represented by the 1-letter code of theamino acid residue introduced at P9 into the ARAAAAAAA (ARA7)sequence. The RRYQKSTEL peptide is designated by an asterisk.Peptide binding is expressed as described in Figure 3. Data are themean of at least 2 independent experiments.

1982 LAMAS ET AL

ison of the peptide specificities of B*2705 and B*2704.These 2 subtypes are both associated with AS (9) anddiffer in only 3 amino acid residues: D77S, V152E, andA211G. Aside from rather slight differences in residuespecificity at P1 and P3, these 2 subtypes differ mainly atP. The detrimental effect of basic P residues isconsistent with the absence of these motifs amongB*2704-bound peptides (21), and is consistent withfindings in previous binding studies (32). In spite of thedetrimental effect of C-terminal Y and F for binding toB*2704, these residues are C-terminal motifs forB*2704-bound peptides (21). However, as in B*2705, adetrimental C-terminal residue can occur among B*2704ligands if they have optimal P3 residues. On the basis ofour results, it is likely that the B*2705 and B*2704peptide repertoires overlap to a significant extent. Theoverlapping repertoire probably consists mainly of pep-tides with C-terminal aliphatic residues, and peptideswith C-terminal Y or F plus nonpolar P3 residues.

The peptide-binding differences between B*2704and B*2706 are particularly relevant because, in contrastto the former subtype, B*2706 is less or not associatedwith AS in various populations (11–13). B*2706 differsfrom B*2704 in 2 amino acid residues, H114D andD116Y, both of which are located in the -pleated sheetfloor of the peptide-binding site and can affect theinteraction with at least the P3 and P residues of boundpeptides. Previous sequencing studies revealed that amajor difference between B*2706 and B*2704 was theabsence of Y as a C-terminal motif in B*2706 (21). Invitro binding studies also have shown that aliphaticC-terminal residues were much better than Y for B*2706binding, and that this difference was smaller for B*2704binding (32). In agreement with these previous data, ourresults now confirm the high suitability of nonpolaraliphatic residues for B*2706 binding. In addition, theabsence of Y as a C-terminal motif for this subtype isexplained because bulky aliphatic and F residues bindmore strongly, relative to Y, than in B*2704. Therefore,peptides with C-terminal Y will compete less advanta-geously in vivo for binding to B*2706.

B*2704 and B*2706 showed additional differ-ences in their P1 and P3 residue specificity, which havenot been revealed by previous studies. Most noteworthywere the specificity differences at P3, especially thoseconcerning the better acceptance of R and polar resi-dues in B*2706, and the worse suitability of A and Yrelative to many other residues. These differences areprobably due mainly to the presence of D114 in B*2706,instead of H114 in B*2704 and other subtypes. P1differences between B*2704 and B*2706 were surprising

because both subtypes have an identical A pocket, whichis the site where P1 binds. This suggests that long-rangeeffects of distant polymorphic residues in HLA–B27modulate its interaction with N-terminal peptide resi-dues. Long-range effects on B pocket specificity havebeen previously observed in B*2701 (20).

The different residue specificity of B*2704 andB*2706 at multiple anchor positions implies that theirdifferential association with AS correlates not only withtheir differential acceptance of C-terminal Y, but alsowith a more complex modulation of their peptide rep-ertoires, affecting at least also P1 and P3 residues.Therefore, B*2704 and B*2706 may differ significantlyin their peptide repertoires, although they share com-mon ligands (21). The actual extent of their overlap andthe type of peptides that are bound in vivo to only 1 ofthese subtypes will require a more extensive analysis oftheir natural ligands.

In conclusion, the results in this study revealedthe preferences of B*2705 for P1, P3, and P residuesand some major rules governing residue usage in thesepositions among natural ligands. This will allow a mean-ingful prediction of B*2705 ligands based on P1, P2, P3,and P. Thus, from any protein putatively involved indisease-related T cell responses, it is now possible toselect a rather limited number of peptides fulfilling theserules to test their role as HLA–B27–restricted antigens.In addition, our results show that 2 subtypes differen-tially associated with AS differ in their residue specificityat multiple anchor positions, suggesting that many pep-tides having R2 and C-terminal motifs common toB*2704 and B*2706 may nevertheless bind with differentefficiency or to only 1 subtype. Since this point is crucialfor assessing the nature of putative arthritogenic pep-tides, correlations between the binding score of differentresidues at each position and their actual usage amongnatural ligands should be determined, as was done forB*2705 in this study. However, this requires a moreextensive database of natural B*2704 and B*2706 li-gands, of which very few are yet known.

This study has defined some major peptide fea-tures that shape HLA–B27–bound peptide repertoires,and has described how these features are modulated bydisease-related subtype polymorphism. The peptide-binding specificity of HLA–B27 is probably a criticalfeature for its linkage to spondylarthropathy, but thepathogenetic role of this antigen is unlikely to be ex-plained solely by its peptide-binding properties. Thecritical question as to which HLA–B27–bound peptidesmay become target antigens of autoimmune T cellresponses in disease pathogenesis remains unanswered.

PEPTIDE SPECIFICITY OF B*2705, B*2704, AND B*2706 1983

ACKNOWLEDGMENTS

We thank Jesus Vazquez and Anabel Marina (Depart-ment of Protein Chemistry, CBMSO) for help in mass spec-trometry, Juan Pablo Albar (Centro Nacional de Biotecnolo-gıa, Madrid, Spain) and Francisco Gavilanes (UniversidadComplutense de Madrid) for help in peptide chemistry, andManuel Ramos (CBMSO) for his contributions to the databaseof HLA–B27 ligands.

REFERENCES

1. Madden DR. The three-dimensional structure of peptide-MHCcomplexes. Annu Rev Immunol 1995;13:587–622.

2. Rammensee HG, Friede T, Stevanoviic S. MHC ligands andpeptide motifs: first listing. Immunogenetics 1995;41:178–228.

3. Ruppert J, Sidney J, Celis E, Kubo RT, Grey HM, Sette A.Prominent role of secondary anchor residues in peptide binding toHLA-A2.1 molecules. Cell 1993;74:929–37.

4. Brewerton DA, Hart FD, Nicholls A, Caffrey M, James DC,Sturrock RD. Ankylosing spondylitis and HL-A 27. Lancet 1973;1:904–7.

5. Schlosstein L, Terasaki PI, Bluestone R, Pearson CM. Highassociation of an HL-A antigen, W27, with ankylosing spondylitis.N Engl J Med 1973;288:704–6.

6. Kingsley G, Sieper J. Current perspectives in reactive arthritis.Immunol Today 1993;14:387–91.

7. Benjamin R, Parham P. Guilt by association: HLA-B27 andankylosing spondylitis. Immunol Today 1990;11:137–42.

8. Zhou M, Sayad A, Simmons WA, Jones RC, Maika SD, SatumtiraN, et al. The specificity of peptides bound to human histocompat-ibility leukocyte antigen (HLA)-B27 influences the prevalence ofarthritis in HLA-B27 transgenic rats. J Exp Med 1998;188:877–86.

9. Breur-Vriesendorp BS, Dekker Saeys AJ, Ivanyi P. Distribution ofHLA-B27 subtypes in patients with ankylosing spondylitis: thedisease is associated with a common determinant of the variousB27 molecules. Ann Rheum Dis 1987;46:353–6.

10. D’Amato M, Fiorillo MT, Carcassi C, Mathieu A, Zuccarelli A,Bitti PP, et al. Relevance of residue 116 of HLA-B27 in determin-ing susceptibility to ankylosing spondylitis. Eur J Immunol 1995;25:3199–201.

11. Lopez-Larrea C, Sujirachato K, Mehra NK, Chiewsilp P,Isarangkura D, Kanga U, et al. HLA-B27 subtypes in Asianpatients with ankylosing spondylitis: evidence for new associations.Tissue Antigens 1995;45:169–76.

12. Nasution AR, Mardjuadi A, Kunmartini S, Suryadhana NG,Setyohadi B, Sudarsono D, et al. HLA-B27 subtypes positively andnegatively associated with spondyloarthropathy. J Rheumatol1997;24:1111–4.

13. Ren EC, Koh WH, Sim D, Boey ML, Wee GB, Chan SH. Possibleprotective role of HLA-B*2706 for ankylosing spondylitis. TissueAntigens 1997;49:67–9.

14. Jardetzky TS, Lane WS, Robinson RA, Madden DR, Wiley DC.Identification of self peptides bound to purified HLA-B27. Nature1991;353:326–9.

15. Madden DR, Gorga JC, Strominger JL, Wiley DC. The three-dimensional structure of HLA-B27 at 2.1 A resolution suggests ageneral mechanism for tight peptide binding to MHC. Cell1992;70:1035–48.

16. Rotzschke O, Falk K, Stevanovic S, Gnau V, Jung G, RammenseeHG. Dominant aromatic/aliphatic C-terminal anchor in HLA-B*2702 and B*2705 peptide motifs. Immunogenetics 1994;39:74–7.

17. Boisgerault F, Tieng V, Stolzenberg MC, Dulphy N, Khalil I,Tamouza R, et al. Differences in endogenous peptides presentedby HLA-B*2705 and B*2703 allelic variants: implications for

susceptibility to spondylarthropathies. J Clin Invest 1996;98:2764–70.

18. Griffin TA, Yuan J, Friede T, Stevanovic S, Ariyoshi K, Rowland-Jones SL, et al. Naturally occurring A pocket polymorphism inHLA-B*2703 increases the dependence on an accessory anchorresidue at P1 for optimal binding of nonamer peptides. J Immunol1997;159:4887–97.

19. Garcia F, Rognan D, Lamas JR, Marina A, Lopez de Castro JA.An HLA-B27 polymorphism (B*2710) that is critical for T-cellrecognition has limited effects on peptide specificity. TissueAntigens 1998;58:1–9.

20. Garcia F, Galocha B, Villadangos JA, Lamas JR, Albar JP, MarinaA, et al. HLA-B27 (B*2701) specificity for peptides lacking Arg2is determined by polymorphism outside the B pocket. TissueAntigens 1997;49:580–7.

21. Garcia F, Marina A, Lopez de Castro JA. Lack of carboxyl-terminal tyrosine distinguishes the B*2706-bound peptide reper-toire from those of B*2704 and other HLA-B27 subtypes associ-ated to ankylosing spondylitis. Tissue Antigens 1997;49:215–21.

22. Tieng V, Dulphy N, Boisgerault F, Tamouza R, Charron D,Toubert A. HLA-B*2707 peptide motif: Tyr C-terminal anchor isnot shared by all disease-associated subtypes. Immunogenetics1997;47:103–5.

23. Fiorillo MT, Meadows L, D’Amato M, Shabanowitz J, Hunt DF,Apella E, et al. Susceptibility to ankylosing spondylitis correlateswith the C-terminal residue of peptides presented by variousHLA-B27 subtypes. Eur J Immunol 1997;27:368–73.

24. Rovero P, Riganelli D, Fruci D, Vigano S, Pegoraro S, RevoltellaR, et al. The importance of secondary anchor residue motifs ofHLA class I proteins: a chemometric approach. Mol Immunol1994;31:549–54.

25. Fruci D, Greco G, Vigneti E, Tanigaki N, Butler RH, Tosi R. Thepeptide-binding specificity of HLA-B27 subtype (B*2705) ana-lyzed by the use of polyalanine model peptides. Hum Immunol1994;41:34–8.

26. Wen J, Wang J, Kuipers JG, Huang F, Williams KM, RaybourneRB, et al. Analysis of HLA-B*2705 peptide motif, using T2 cellsand monoclonal antibody ME1. Immunogenetics 1994;39:444–6.

27. Colbert RA, Rowland Jones SL, McMichael AJ, Frelinger JA.Differences in peptide presentation between B27 subtypes: theimportance of the P1 side chain in maintaining high affinitypeptide binding to B*2703. Immunity 1994;1:121–30.

28. Villadangos JA, Galocha B, Garcia F, Albar JP, Lopez de CastroJA. Modulation of peptide binding by HLA-B27 polymorphism inpockets A and B, and peptide specificity of B*2703. Eur J Immu-nol 1995;25:2370–7.

29. Ljunggren HG, Karre K. Host resistance directed selectivelyagainst H-2-deficient lymphoma variants: analysis of the mecha-nism. J Exp Med 1985;162:1745–59.

30. Townsend A, Ohlen C, Bastin J, Ljunggren HG, Foster L, KarreK. Association of class I major histocompatibility heavy and lightchains induced by viral peptides. Nature 1989;340:443–8.

31. Ljunggren HG, Stam NJ, Ohlen C, Neefjes JJ, Hoglund P,Heemels MT, et al. Empty MHC class I molecules come out in thecold. Nature 1990;346:476–80.

32. Galocha B, Lamas JR, Villadangos JA, Albar JP, Lopez de CastroJA. Binding of peptides naturally presented by HLA-B27 to thedifferentially disease-associated B*2704 and B*2706 subtypes, andto mutants mimicking their polymorphism. Tissue Antigens 1996;48:509–18.

33. Villadangos JA, Galocha B, Lopez de Castro JA. Unusual topol-ogy of an HLA-B27 allospecific T cell epitope lacking peptidespecificity. J Immunol 1994;152:2317–23.

34. Ellis SA, Taylor C, McMichael A. Recognition of HLA-B27 andrelated antigens by a monoclonal antibody. Hum Immunol 1982;5:49–59.

35. Paradela A, Garcia-Peydro M, Vazquez J, Rognan D, Lopez de

1984 LAMAS ET AL

Castro JA. The same natural ligand is involved in allorecognitionof multiple HLA-B27 subtypes by a single T cell clone: role ofpeptide and the MHC molecule in alloreactivity. J Immunol1998;161:5481–90.

36. Yague J, Vazquez J, Lopez de Castro JA. A single amino acid changemakes the peptide specificity of B*3910 unrelated to B*3901 andcloser to a group of HLA-B proteins including the malaria-protectingallotype HLA-B53. Tissue Antigens 1998;52:416–21.

37. Barber LD, Percival L, Parham P. Characterization of the peptide-binding specificity of HLA-B*7301. Tissue Antigens 1996;47:472–7.

38. Van Endert PM, Riganelli D, Greco G, Fleischhauer K, Sidney J,Sette A, et al. The peptide-binding motif for the human trans-porter associated with antigen processing. J Exp Med 1995;182:1883–95.

39. Uebel S, Kraas W, Kienle S, Wiesmuller KH, Jung G, Tampe R.Recognition principle of the TAP transporter disclosed by combina-torial peptide libraries. Proc Natl Acad Sci U S A 1997;94:8976–81.

40. Daniel S, Brusic V, Caillat-Zucman S, Petrovsky N, Harrison L,Riganelli D, et al. Relationship between peptide selectivities ofhuman transporters associated with antigen processing and HLAclass I molecules. J Immunol 1998;161:617–24.

41. Peh CA, Burrows SR, Barnden M, Khanna R, Cresswell P, MossDJ, et al. HLA-B27-restricted antigen presentation in the absenceof tapasin reveals polymorphism in mechanisms of HLA class Ipeptide loading. Immunity 1998;8:531–42.

42. Rognan D, Scapozza L, Folkers G, Daser A. Rational design ofnonnatural peptides as high-affinity ligands for the HLA-B*2705human leukocyte antigen. Proc Natl Acad Sci U S A 1995;92:753–7.

43. Krebs S, Lamas JR, Poenaru S, Folkers G, Lopez de Castro JA,Seebach D, et al. Substituting nonpeptidic spacers for the T cellreceptor-binding part of class I major histocompatibility complex-binding peptides. J Biol Chem 1998;273:19072–9.

44. Brooks JM, Murray RJ, Thomas WA, Kurilla MG, Rickinson AB.Different HLA-B27 subtypes present the same immunodominantEpstein-Barr virus peptide. J Exp Med 1993;178:879–87.

45. Van Binnendijk RS, Versteeg-van Oosten JP, Poelen MC, Brug-ghe HF, Hoogerhout P, Osterhaus AD, et al. Human HLA class I-and HLA class II-restricted cloned cytotoxic T lymphocytes iden-tify a cluster of epitopes on the measles virus fusion protein.J Virol 1993;67:2276–84.

46. Ugrinovic S, Mertz A, Wu P, Braun J, Sieper J. A single nonamerfrom the Yersinia 60Kd heat shock protein is the target ofHLA-B27 restricted CTL response in Yersinia-induced reactivearthritis. J Immunol 1997;159:5715–23.

47. Garcia F, Marina A, Albar JP, Lopez de Castro JA. HLA-B27presents a peptide from a polymorphic region of its own moleculewith homology to proteins from arthritogenic bacteria. TissueAntigens 1997;49:23–8.

PEPTIDE SPECIFICITY OF B*2705, B*2704, AND B*2706 1985

ANEXO -6-

Substituting Nonpeptidic Spacers for the T Cell Receptor-bindingPart of Class I Major Histocompatibility Complex-binding Peptides*

(Received for publication, December 23, 1997, and in revised form, April 21, 1998)

Stefan Krebs‡, Jose, R. Lamas§, Sorana Poenaru¶, Gerd Folkers‡, Jose A. Lopez de Castro§,Dieter Seebach¶, and Didier Rognan‡

From the ‡Department of Pharmacy, Swiss Federal Institute of Technology, Winterthurerstrasse 190, CH-8057 Zurich,Switzerland, the ¶Laboratory for Organic Chemistry, Swiss Federal Institute of Technology, Universitatstrasse 16, CH-8092 Zurich, Switzerland, and the §Centro de Biologia Molecular “Severo Ochoa,” Universidad Autonoma de Madrid,Facultad de Ciencas, E-28049 Madrid, Spain

X-ray diffraction studies as well as structure-activityrelationships indicate that the central part of class Imajor histocompatibility complex (MHC)-binding non-apeptides represents the main interaction site for a Tcell receptor. In order to rationally manipulate T cellepitopes, three nonpeptidic spacers have been designedfrom the x-ray structure of a MHC-peptide complex andsubstituted for the T cell receptor-binding part of sev-eral antigenic peptides. The binding of the modifiedepitopes to the human leukocyte antigen-B*2705 proteinwas studied by an in vitro stabilization assay, and thethermal stability of all complexes was examined by cir-cular dichroism spectroscopy. Depending on theirchemical nature and length, the introduced spacers maybe classified into two categories. Monofunctional spac-ers (11-amino undecanoate, (R)-3-hydroxybutyrate tri-mer) simply link two anchoring peptide positions (P3and P9) but loosely contact the MHC binding groove andthus decrease more or less the affinity of the alteredepitopes to human leukocyte antigen-B*2705. A bifunc-tional spacer ((R)-3-hydroxybutyrate tetramer) not onlybridges the two distant anchoring amino acids but alsostrongly interacts with the binding cleft and leads to a5-fold increase in binding to the MHC protein. To ourknowledge, this is the first report of a nonpeptidic mod-ification of T-cell receptor binding residues that signif-icantly enhances the binding of altered peptide ligandsto their host MHC protein. The presented modified li-gands constitute interesting tools for perturbing the Tcell response to the parent antigenic peptide.

Class I MHC1 molecules are highly polymorphic proteinsthat play a key role in immune surveillance by presentingforeign peptides to cytotoxic T lymphocytes (1). The molecu-lar mechanisms of peptide selection have been characterizedby x-ray diffraction studies of several MHC proteins in com-plex with either a peptide pool or single ligands (2). Peptides,

generally nonamers, tightly bind to conserved MHC residuesin a sequence-independent manner at their N and C termini(3), whereas the central part of the bound peptide bulges outof the binding groove (4). Peptide specificity is governed bythe position and chemical nature of some anchoring sidechains (often P2, P3, and P9) that bind to MHC polymorphicpockets (5, 6). Complementary to x-ray structure determina-tions, sequencing self-peptides naturally bound to MHC pro-teins allows the determination of peptide binding motifs (7, 8)and thus the identification of conserved amino acids respon-sible for MHC binding (named dominant anchors, generallyat positions P2 and P9) and more variable residues hypoth-esized to account for TcR recognition (usually in the centralpart of the peptide sequence, from P4 to P8). Peptide muta-tion (9, 10) as well as recently determined x-ray structures of TcRs in complex with a MHC-peptide (11, 12) unambigu-ously support this assumption. Since some class I MHC alle-les are associated with either susceptibility or resistance tohuman diseases (13–15), altering TcR contact residues of T cellepitopes has been proposed for designing altered peptide li-gands with TcR antagonist properties (16), leading to in vivo Tcell anergy (17). However, natural peptides cannot be easilyused as immunosuppressors because of poor enzymatic stabil-ity and pharmacokinetic properties (18). Herewith, we describethe substitution of nonpeptidic moieties for the TcR contactamino acids of several T cell epitopes naturally presented bythe class I MHC protein B*2705, which is strongly linked tosevere inflammatory diseases like ankylosing spondylitis (13)or reactive arthritis (19). Some reports in which a similarstrategy has been followed (20–22) show that the altered pep-tide ligands still form stable complexes with their host MHCprotein but often present a reduced affinity relative to theparent peptide. The present study describes a novel oligomericspacer able not only to link two MHC anchoring positions (P3and P9) but also to significantly improve binding to the restrict-ing class I MHC protein.

EXPERIMENTAL PROCEDURES

Computer-assisted Ligand Design—Molecular mechanics and dy-namics calculations were carried out using the AMBER 4.1 package(23), using the parm94 parameter set (24) and an all-atom force fieldrepresentation. Force field parameters for the ester group were takenfrom the literature (25). Atomic charges for the Aua and HB monomerswere calculated using the GAUSSIAN 94 package (26) and the HF/6–31G* basis set by fitting atom-centered charges to an ab initio electro-static potential, using the RESP method (27) according to a previouslydescribed procedure (28). Atomic charges for both new monomers arelisted in Table I.

Initial coordinates for the MHC-ligand complexes were obtained fromthe x-ray structure of HLA-B*2705 (3) as described previously (20, 29).The spacers were substituted for the natural pentapeptide sequenceusing the SYBYL modeling package (TRIPOS Association, Inc., St.

* This work was supported by the Schweizerischer Nationalfonds zurForderung der wissenschaftlichen Forschung (Project 31-45504.95);Ministry of Education Grant PM95–002 and Plan Nacional de I DGrant SAF97–0182 (to J.A.L.C.); and an institutional grant of theFundacion Ramon Areces to the Centro de Biologia Molecular “SeveroOchoa.” The costs of publication of this article were defrayed in part bythe payment of page charges. This article must therefore be herebymarked “advertisement” in accordance with 18 U.S.C. Section 1734solely to indicate this fact.

To whom correspondence should be addressed. Fax: 41-1-635-68-84;E-mail: didier@pharma.ethz.ch.

1 The abbreviations used are: MHC, major histocompatibility com-plex; HLA, human leukocyte antigen; TcR, T cell receptor; Pn, peptideposition n; Fmoc, N-(9-fluorenyl)methoxycarbonyl; Aua, 11-amino un-decanoate; HB, (R)-3-hydroxybutyrate.

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 273, No. 30, Issue of July 24, pp. 19072–19079, 1998© 1998 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

This paper is available on line at http://www.jbc.org19072

Louis, MO). From a starting fully extended conformation, dihedralangles of the main chain between P3 and P9 were modified by hand inorder to reproduce a correct trans geometry for the newly introducedamide or ester bonds. The ligand was first relaxed by 500 steps ofconjugate gradient energy minimization while maintaining the proteinfixed. It was then submitted to a 100-ps simulated annealing protocol inorder to sample the broadest conformational space accessible. Startingwith random velocities assigned at a temperature of 1000 K, the ligandwas first coupled for 50 ps to a heat bath at 1000 K using a relativelyweak temperature coupling constant (0.2 ps) and then linearly cooleddown to 50 K for the next 50 ps while was strengthened to a value of0.05 ps. During these 100 ps, no protein atoms were allowed to move.The last conformer was then solvated in a 10-Å-thick TIP3P water shell.Energy minimization of the ligand, of the MHC-ligand complex, fol-lowed by 200-ps molecular dynamics simulation of the fully solvatedMHC-ligand pair was performed as previously reported (20).

Synthesis of the Modified Peptides—Ligands 1–8 (Table II) wereobtained by automated solid-phase peptide synthesis using a Fmoc/tert-butyl protecting strategy. Chain elongation was performed by a robotsystem (Syro Multi-Syn-Tech, Bochum, Germany) with a subsequentmanual deprotection and analysis. Fmoc-protected amino acids werecoupled to the diisopropylcarbodiimide-activated carboxyl terminus in10-fold excess using 1-hydroxybenzotriazole as a coupling reagent. Thefinal peptide was simultaneously cleaved from the resin and depro-tected by the addition of trifluoroacetic acid with thiocresole and thio-anisole as scavengers. The peptides were precipitated and washed withice-cold ether and further lyophilized from water. Natural as well asnonnatural peptides were analyzed by reverse phase high performanceliquid chromatography (Merck-Hitachi, Darmstadt, Germany) on anucleosil 5, C-18 column (125 3 mm) at a flow rate of 600 l/min.Absorbance was measured at 220 nm. The solvent system consisted of0.1% trifluoroacetic acid in water (buffer A) and 0.1% trifluoroaceticacid in acetonitrile (buffer B). A linear gradient from 10 to 60% B in 30min was applied. Furthermore, peptides were analyzed by ion spraymass spectrometry on a triple quadrupole mass spectrometer, APII III,with a mass range of m/z 10–2400 equipped with an ion spray interface(Sciex, Thornhill, Canada). The mass spectrometer was operated inpositive ion mode under conditions of unit mass resolution for alldeterminations.

The synthesis of ligands 9–12 will be reported elsewhere.2

Epitope Stabilization Assay—The quantitative assay used was de-scribed previously (30). Briefly, RMA-S transfectants expressingB*2705 or B*2704 were used. These are murine cells with impairedTAP-mediated peptide transport and low surface expression of (empty)class I MHC molecules, which can be induced at 26 °C (31) and stabi-lized at the cell surface through binding of exogenously added ligands.These cells were incubated at 26 °C for 24 h. After this, they wereincubated 1 h at 26 °C with 104 to 109 M peptides, transferred to37 °C, and collected for flow microcytometry analysis with the ME1monoclonal antibody (IgG1, specific for HLA-B27, -B7, and -B22) (32)after 4 h for B*2705 or after 2 h for B*2704. The determinant recognizedby ME1 is not affected by bound peptides or by polymorphism in thesetwo subtypes (data not shown). Binding of a given ligand was measuredas its C50. This is its molar concentration at 50% of the fluorescenceobtained with that ligand at 104 M. Ligands with C50 5 M wereconsidered to bind with high affinity, since these were the valuesobtained for most of the natural B27-bound peptides. C50 values be-tween 5 and 50 M were considered to reflect intermediate affinity. C50

50 M indicated low affinity. Binding of peptide analogs was meas-ured as the concentration of the peptide analog required to obtain thefluorescence value at the C50 of the unchanged peptide. This was des-ignated as EC50. Relative binding was the ratio between the EC50 of thepeptide analog and the C50 of the corresponding unchanged peptide.

HLA-B*2705 Expression and Purification—A cDNA encoding for hu-man 2-microglobulin (gift of Dr. C. Vilches, Clinica Puerta de Hierro,Madrid) was cloned into a pGex vector (Amersham Pharmacia Biotech),yielding a fusion protein with glutathione S-transferase. Escherichiacoli cells transformed with this pGex vector were grown under vigorousshaking in LB broth for 24 h at 25 °C after induction with isopropyl-1-thio--D-galactopyranoside. Cells were frozen at 70 °C, thawed, sus-pended in TBS (20 mM Tris, 150 mM NaCl, pH 8.0), and lysed by theaddition of lysozyme and brief sonication. The crude extract was passedover a glutathione-agarose column (Sigma), and after extensive wash-

2 D. Seebach, S. Poenaru, G. Folkers, and D. Rognan, manuscript inpreparation.

TABLE IRestrained electrostatic potential-derived point charges, calculated using the RESP method (26) from GAUSSIAN 94 HF/6–31G* electrostatic

potentials

a Ab initio derived electrostatic potentials have been calculated for protected monomers (Ac-Aua-NMe, Ac-HB-OMe) and atomic charges of theisolated monomers (Aua and HB) adjusted to neutrality using Lagrange constraints, as previously described (28). Charges for equivalent hydrogenatoms are only mentioned once.

b 11-aminoundecanoate.c (R)-3-hydroxybutyrate.

Nonnatural HLA-B27 Ligands 19073

ing with TBS, the 2-microglobulin was eluted by thrombin cleavage asa single band at 11 kDa (SDS-polyacrylamide gel electrophoresis).

The HLA-B*2705 heavy chain was affinity-purified under denatur-ing conditions as a His6 fusion protein. The expression vector wasobtained by subcloning the cDNA encoding for the first extracellular274 amino acids (gift of Dr. K. C. Parker, National Institutes of Health,Bethesda) into the polycloning site of the oligohistidine vector pQE30(Quiagen) with the restriction endonucleases BamHI and HindIII. Theheavy chain was expressed in E. coli at 35 °C for 2 h after inductionwith isopropyl-1-thio--D-galactopyranoside. Longer expression timesled to an increase of immature or degraded heavy chains. Inclusionbodies were prepared using a standard procedure (33) and solubilized in8 M urea, 20 mM Tris, 150 mM NaCl at pH 8.0. Purification on anickel-nitriloacetate-agarose column led to the HLA-B*2705 heavychain with two minor impurities of lower molecular weights consistingof truncated heavy chains.

Folding of the MHC Protein upon Ligand Binding—Reconstitution ofthe heavy chain-2-microglobulin-ligand heterotrimer was achieved bydialysis (cellulose ester tubings, 500-Da cut-off) of a solution containing0.15 mg/ml heavy chain, 0.1 mg/ml 2-microglobulin, and 0.1 mg/mlpeptide ligand, using 5 mM glutathione to establish reducing conditionsin 6 M urea against TBS. The solution was sparged with nitrogen toprevent premature formation of disulfide bridges and oxidation of freeCys67 in the B*2705 heavy chain. After 36–48 h at 10 °C, the mixturewas concentrated to 500 l in a Centripep ultrafiltration unit (Amicon-Grace Ltd.). The folded heterotrimer was purified by gel filtration on asuperdex 75 column (Amersham Pharmacia Biotech) with UV detectionat 280 nm. The chromatogram showed three major peaks at 9-, 11.5-,and 14-ml elution volume corresponding to heavy chain aggregates,refolded complex, and excess 2-microglobulin, respectively. The overallyield of the fully reconstituted heterotrimer varied around 5%. Theheterotrimer peak was collected, concentrated in a Centricon 30 ultra-filtration unit (Amicon-Grace), and immediately subjected to thermaldenaturation.

Monitoring the Thermal Stability of MHC-Ligand Complexes by CDSpectroscopy—All CD measurements were done on a Jasco J-720 polar-imeter with a water-jacketed 1-mm sample cell connected to a comput-er-interfaced Neslab 111 circulating water bath. Temperature controlwas achieved by measuring the circulating water immediately after thesample cell. The thermal denaturation profiles were recorded at 218 nmin 10 mM Tris, 150 mM NaCl (pH 8.0) with the Jasco TEMPSCANsoftware using 0.1 °C increments at a heating rate of 30 °C/h. Sampleconcentrations were determined photometrically and held at 0.2 mg/ml.Different scan rates did not affect the Tm value of B*2705 in complex

with a reference peptide (GRAFVTIGK; compare Ref. 34 and Table II).Three denaturation curves from independent refolding preparationswere averaged, after conversion to molar ellipticity values. The curveswere reduced to 70 data points by replacing each of the 10 neighboringpoints with their mean value. By assuming a two-state equilibrium (35),data were fitted by a nonlinear least-squares routine with the programOrigin 2.9 (MicroCal Software, Inc.) to the following equations.

T u f u/1 expx (Eq. 1)

x Hm/R1/T 1/Tm Cp/RTm/T 1 lnT/Tm (Eq. 2)

The measured ellipticity () is given as a function of the temperature(T) with the enthalpy (Hm), heat capacity upon unfolding (Cp), andthe midpoint temperature of unfolding (Tm) being the fitting parame-ters. Initial estimates for Hm were obtained by plotting ln K versus 1/T(van’t Hoff plot) in the transition region. Cp was assumed to be tem-perature-independent (36), and initial values were estimated from theprimary sequence (37). The linear base-line functions of the unfoldedand folded states u and f were determined as linear regressions of thepre- and post-transitional regions. The enthalpy change at the midpointof unfolding (Hm) was determined by the least-squares fit of theunfolding curve to Equation 1. Because Cp estimates obtained by thisapproach are not very accurate and the Hm values are largely influ-enced by the observed deviations from a two-state model, a directextrapolation from the midpoint of unfolding to obtain Gunfolding at25 °C was not taken into consideration.

RESULTS

Replacing a Pentapeptide with a Polymethylene Spacer inFour Unrelated Natural Epitopes—For mimicking the se-quence of the central pentapeptide part (P4–P8) of MHC-boundnonapeptides, any nonpeptidic fragment needs first to repro-duce as closely as possible the conformation of this bulging partand second to allow the same intermolecular distance betweenthe neighboring anchoring positions (P3 and P9) that arelinked by the new spacer. The key distance between C- atomsof P3 and P9 positions is 16.6 Å in the x-ray structure ofHLA-B*2705 complexed by a nonapeptide model (3). The samedistance can be easily obtained after linking a polymethylenechain (Aua, Fig. 1) to P3 and P9 residues by simple amidebonds. The 11-amino undecanoate fragment was then chosen

TABLE IIBinding of natural and altered T cell epitopes

LigandSequence EC50

a

Tmb

P1 P2 P3 Spacer P9 B*2705 B*2704

M °C

1 R R R WRRLT Vc 1.2 0.8 52.8 0.72 Auad 4.0 1.0 42.9 0.33 S R Y WAIRT Re 3.0 5.4 46.3 0.54 Aua 8.6 100.0 39.5 0.25 G R A FVTIG Kf 1.8 6.4 61.9 0.36 Aua 7.0 100.0 48.1 0.47 Q R L KEAAE Kg 10.0 62.8 0.78 Aua 46.5 0.29 (HB)3

h 40.010 (HB)4 2.511 A (HB)3 20.0 63.2 0.612 (HB)4 1.6 62.1 0.7

a Concentration of ligand at which HLA-B27 fluorescence (measured by flow microcytometry analysis with an anti-B27 monoclonal antibody) onRMA-S cells was half the maximum obtained with the wild type peptide (30).

b Melting temperature: midpoint of the thermal unfolding of the B*2705 heavy chain. Tm values are means of three denaturation experimentsperformed on independently reconstituted heavy chain-2-microglobulin-ligand heterotrimers. S.D. values have been obtained by fitting theobtained curves to a two-state model as previously described (35).

c Epstein-Barr virus latent membrane protein (236–244) (49).

d Aua: 11-amino undecanoate.

e Influenza A nucleoprotein (383–391) (50).f Human immunodeficiency virus 1 glycoprotein 120 (314–322) (39).g E. coli DnaK protein (260–268) (20).

h HB: (R)-3-hydroxybutyrate.

Nonnatural HLA-B27 Ligands19074

for its optimal length in an extended conformation and theabsence of any substituents, which should allow a conforma-tional flexibility sufficient for a proper fit into the bindinggroove. To check the independence of the proposed modificationon the parent epitope sequence, the Aua spacer was introducedin four unrelated sequences of natural epitopes, known to bindto B*2705 (Table II). The question of whether the new ligandswere able to remain tightly bound in the peptide binding cleftlike the natural nonapeptides was addressed by molecular dy-namics simulations of the solvated complexes (Table II). Thecomputational protocol used has been previously shown to ex-plain the binding potency of several HLA-B27-binding peptides(38) and to predict the high affinity of designed peptide ana-logues (20, 29). Energy-minimized conformations show that theproposed bridging has modified neither the overall conforma-tion of the bound ligands nor the intermolecular distance be-tween P3 and P9 C- atoms (Fig. 2). Moreover, the chemicalsubstitutions were compatible with the conservation of themain interactions between the altered peptides and B*2705,especially the electrostatic interactions provided by the twocharged termini and the arginine found at position 2 of B*2705-binding peptides (39, 40).

In order to experimentally validate the proposed model, thefour B*2705-restricted T cell epitopes and their modified ana-logues (Table II) were synthesized and then tested for theirbinding to B*2705, in an in vitro epitope stabilization assay(30). Replacing the central pentapeptide sequence by the un-substituted Aua fragment led in all cases to a slight decrease inB*2705 stabilization (Table II), which was also reflected by alesser thermal stability of the resulting complexes monitoredby CD spectroscopy (Fig. 3, A–D). The temperature shift in themidpoint of unfolding depends on the sequence of the referencepeptide but varies from 7 to 14 °C (Table II). Whereas theeffect of the Aua spacer is similar in both assays, there seemsto be no clear correlation between the EC50 scores obtainedfrom the epitope stabilization assay and the Tm values calcu-lated from the thermal denaturation experiments. The Tm val-

ues reported here are fairly similar to those found for B*2705-binding peptides by other groups (21, 34, 41). The higheststability against temperature was found for complexes withpeptides 5 and 7, which all present a lysine at P9. Less favoredamino acids at P9 are Val (peptides 1 and 2) and especially Arg(peptides 3 and 4), which gives by far the least stable complexeswith B*2705 (peptides 1 and 3, respectively).

Binding of peptides 1–6 to a closely related HLA-B27 sub-type (B*2704) was also examined by the same in vitro stabili-zation assay. B*2704 differs from B*2705 by two amino acidchanges in the peptide binding groove (Asp77 to Ser; Val152 toGlu), which influence its peptide specificity, relative to B*2705(42). In contrast to B*2705, substitution of Aua spacers forP4–P8 dramatically decreases binding to B*2704 in our epitopestabilization assay (Table II) when the last anchoring position(P9) is a basic amino acid (Lys, Arg). If P9 is an apolar residue(Val, peptide 2), no real change in B*2704 binding was noticed.

Substituting 3-Hydroxybutyrate Oligomers for the P4–P8 Se-quence of a Natural Peptide—The decreased binding of the Auaanalogues to B*2705 is probably due to the nonfunctionalizednature of the introduced spacer and the lack of interactionsbetween the unsubstituted Aua moiety and the central part ofthe binding groove. Thus, a rational improvement in terms ofbinding affinity would be to ramify the spacing moiety in orderto reach one of the two central pockets (pockets C and E) of thepeptide binding groove that face the spacer fragment. The(R)-3-hydroxybutyrate (HB) monomer was selected for threemain reasons: (i) polymers of HB are chemically stable (43); (ii)they adopt conformations whose folding in the free state resem-bles that found for peptides (44); and (iii) the methyl substitu-ent is large enough to fit into pockets C and E. Thus, a trimer(three units) and a tetramer (four units) of HB were substitutedfor the P4–P8 sequence of one natural peptide (polyesterpep-tides 9 and 10; Table II), since they should optimally span thekey distance between the two anchor positions (P3 and P9) tobridge (Fig. 1).

In order to circumvent cyclization of the N-terminal gluta-mine (45) that would prevent binding of the peptidic N termi-nus in the A pocket of B*2705, the Ala1 analogue was alsosynthesized in the nonnatural series (polyesterpeptides 11 and12; Table II).

The modified ligands 9–12 have totally different bindingaffinities in the in vitro stabilization assay (Table II), the tet-ramer-containing compounds (ligands 10 and 12) being about15 times more potent that the trimeric analogues (ligands 9and 11). Furthermore, a HB tetramer segment leads to a sig-nificant enhancement of the binding to B*2705 relative to thenatural pentapeptide sequence. Again, the differences observedbetween natural and polyesterpeptides in the in vitro stabili-zation assay are not reflected by the thermal denaturationexperiments, performed only for ligands 11 and 12. Both com-pounds promote a similarly high stability of the resultingMHC-ligand pair with Tm values of 62–63 °C (Fig. 4, Table II)analogous to that found for the parent peptide 7, and charac-teristic of high affinity ligands (41).

Molecular Modeling of the Altered Peptides in Complex withB*2705—A rationale for the (de)stabilizing effects of the threespacers presently studied is proposed by the molecular dynam-ics time-averaged conformations of a reference peptide (QR-LKEAAEK; peptide 7) and its analogues (peptides 8–10). Bylooking at all close nonbonded contacts between any peptideresidue and its protein neighboring atoms, the three spacers(Aua, HB trimer, and HB tetramer) can be easily distinguished(Fig. 5). The Aua spacer provides fewer contacts to the MHCbinding groove than the pentameric P4–P8 sequence of theparent peptide 7. This could explain the decreased binding

FIG. 1. Chemical structure of the modified peptide analogues.

Nonnatural HLA-B27 Ligands 19075

affinity of Aua-containing peptides to B*2705. The detrimentaleffect of the HB trimer can be explained by the weakening ofthe interactions between both terminal residues (PN and PC)and their respective pockets (A and F). The better complemen-tarity of the HB tetramer to the B*2705 binding cleft is prob-ably related to the following factors: (i) the additional interac-tions provided by two methyl groups of the tetrameric spaceritself and (ii) a higher number of nonbonded contacts of allother MHC anchors (PN, P2, P3, and PC).

The total buried surface area of the modified ligands 8–10has been maintained when compared with that of the parentpeptide 7 (about 650 Å2; data not shown). However, the totalaccessibility of the ligands in their bound state is different. It isreduced by 20% for HB analogues with respect to the naturalepitope 7 (from 500 to 400 Å2). The Aua compound 8, althoughslightly less potent, has a much lower accessible surface area(250 Å2) due to the lack of substituents in the spacing area.

DISCUSSION

Replacing the central TcR-binding residues of MHC classI-bound peptides (P4–P8) by nonpeptidic moieties has beenreported previously (20, 21). Herewith, we propose to rational-ize the effect of three novel spacers on binding to the HLA-B*2705 protein. The simplest spacer (Aua) is a single poly-methylene chain linking the P3 and P9 positions by amidebonds. In accordance with a previous report studying the effectof non- amino acids (20), the Aua spacer does not impairbinding to B*2705. Only a moderate decrease in relative bind-ing to B*2705 was observed in an epitope stabilization assay,performed for four unrelated modified peptides (Table II). How-ever, the effect of this modification on the thermal stability ofthe resulting MHC-ligand pair was more significant (Fig. 3,A–D). Depending on the peptide in which the Aua moiety wasintroduced, the midpoint of unfolding of the B*2705 heavychain (Tm) was lowered by 7–14 °C. The corresponding free

FIG. 2. Close-up into the binding groove of HLA-B*2705 (orange surface) in complex with peptides 7–10 (Table II). These structuresrepresent energy-minimized conformations obtained from the x-ray structure of HLA-B*2705 in complex with a model peptide (Protein Data Bankentry 1hsa) under a previously described protocol (20, 29). The heavy chain backbone atoms of HLA-B*2705 have first been fitted together in thefour complexes, and the protein atoms are not shown for the sake of clarity. Since protein distortion upon energy minimization of the resultingcomplexes is minimal, the MHC protein is here represented by a unique molecular surface independent of the bound ligand. The color coding isas follows: blue, nitrogen; red, oxygen; white, carbon atoms of ligand 7; cyan, carbon atoms of ligand 8; green, carbon atoms of ligand 9; yellow,carbon atoms of ligand 10. The arrows indicate two methyl substituents of the HB tetramer interacting with the central pockets C/E of the protein.The figure has been prepared using the program GRASP (51).

Nonnatural HLA-B27 Ligands19076

energy change in unfolding Gunfolding at the midpoint ofunfolding, derived from the CD spectra (22), varies from 0.8to 1.3 kcal/mol. Since unfolding of the heavy chain shouldfollow release of the ligand, this observation supports a fasterdissociation of the modified peptides with respect to the parentepitope, as recently illustrated in a homogeneous series ofH-2Kd-binding nonapeptides (46). However, the present studysuggests that extrapolating peptide binding differences fromTm values is not allowed for unrelated sequences. For the set of4 T cell epitopes presently studied, EC50 values cannot be

related to melting temperatures calculated by CD spectroscopy.A likely explanation for this is that binding, as measured inepitope stabilization assays, is significantly influenced by theassociation rate of the peptide, whereas CD measurementsrelate only to the dissociation rates. The highest thermal sta-bilities were obtained for the B*2705 protein in complex with

FIG. 3. Thermal denaturation, monitored by CD spectroscopy at 218 nm, of HLA-B*2705 loaded with ligands 1 and 2 (A), ligands3 and 4 (B), ligands 5 and 6 (C), and ligands 7 and 8 (D). The arrows indicate the midpoint of unfolding (Tm) of the B*2705 heavy chain.

FIG. 4. Thermal denaturation, monitored by CD spectroscopyat 218 nm, of HLA-B*2705 loaded with ligands 7, 11, and 12.

FIG. 5. Nonbonded interactions between the HLA-B*2705 pro-tein and ligands 7–10, measured on energy-minimized time-av-eraged conformations obtained after 200-ps Molecular Dynam-ics simulations of the corresponding solvated complexes.Protein-ligand contacts are recorded for interaction distances up to 4 Å.

Nonnatural HLA-B27 Ligands 19077

peptide ligands bearing a Lys at P9. This makes sense, sinceLys is the P9 residue most complementary to its binding pocketF. Its side chain forms a buried salt bridge with Asp116, locatedat the bottom of the pocket. The predominance of the enthalpiccontribution to peptide dissociation would thus be compatiblewith the lower Tm values observed with peptides having anamino acid (Val, Arg) for which the interaction with pocket F isweaker. It also corroborates previous computational simula-tions, suggesting that peptide dissociation first occurs at the Cterminus (20, 29, 38).

Interestingly, the effect of the Aua spacer is subtype-depend-ent, since differences between the natural and the Aua peptidesin binding to B*2704 were much more significant (Table II).B*2704 basically differs from the B*2705 allele by its weakpropensity to present peptides with basic P9 amino acids andits improved suitability for nonpolar P9 residues (42). Thus, thedeleterious effect of the Aua spacer is amplified for peptidesbearing a weak anchoring amino acid at P9 (peptides 4 and 6;Table II) and decreased for peptides with nonpolar P9 residues(peptide 2). The Aua group can be considered as a monofunc-tional spacer, since it simply provides the covalent linkagebetween two neighboring anchor positions (P3 and P9). There-fore, it has the same effect on HLA binding as previouslyreported spacing moieties like oligomers of 4-aminobutyrate or6-aminohexanoate (20) or substituted phenanthridines forwhich a similar thermal destabilization (Tm of 12 °C) hasbeen reported (21). However, a modification of TcR-bindingamino acids that also enhances the binding affinity for the hostMHC protein is possible. We describe here the first bifunctionalspacer that provides additional interactions to the bindinggroove. The tetramer of HB, introduced between P3 and P9,significantly enhances binding to B*2705 (Table II). The bene-ficial effect of the (HB)4 spacer is attributed to two of its methylsubstituents that reach the central pockets C/E of the bindingcleft (Fig. 2). Since the global binding mode of the modifiedpeptide has not been altered, the direct consequence of thisreplacement is an enhanced number of nonbonded contactswith the protein (Fig. 5). Again, discrepancies are observed forthat series of compounds (ligands 7–12) between EC50 valuesand melting temperatures derived from CD experiments on thereconstituted complexes (Fig. 4). Tm values calculated for thetetrameric and trimeric HB analogues are nearly identical,whereas a 12–16-fold decreased binding was observed aftershortening the length of the spacing area by one HB unit. TheTm values of a series of MHC-peptide complexes have recentlybeen directly related to experimental equilibrium dissociationconstants, KD (46). Thus, the higher affinity observed for the(HB)4 compounds relative to the parental peptide and the tri-meric analogues could be due to faster on-rate kinetics. Alter-natively, since the correlation proposed by Morgan et al. (46)takes into account a series of highly related nonapeptides, itmay not be valid for altered ligands lacking a canonical non-apeptide structure. Importantly, the present study demon-strates that CD denaturing curves cannot be used alone toexplain differences in binding of altered peptide ligands to aclass I MHC protein. This is of crucial importance in any designeffort aimed at enhancing binding affinities by increasing theon-rate kinetics of the designed molecule. It should be notedthat two CD denaturation curves (peptides 7 and 12, Fig. 4)slightly deviate from the expected two-state model by present-ing an additional transition at a temperature (45 °C) corre-sponding to the unfolding of peptide-free heavy chain (47).Such deviations from an ideal two-state model have alreadybeen observed (34) but remain difficult to explain at the mo-lecular level.

Our data demonstrate that B*2705-restricted epitopes may

be easily modified by introducing simple nonpeptidic elementsin their central part without drastic changes in binding to theirrestriction MHC proteins. Two conditions seem to be necessaryfor these modifications: (i) the last amino acid (PC) should be astrong anchor, and (ii) the parent epitope should not contain adominant anchor position between the P4 and P8 positions.Since this is the case for a majority of class I MHC peptidebinding motifs (8), such chemical manipulations should be fea-sible for many antigenic peptides binding to class I MHC pro-teins. Class II MHC-binding peptides that utilize nearly allpeptidic bonds to interact with their host MHC protein (48)must be excluded from these epitope modifications.

The altered ligands reported in this study constitute a fur-ther step toward obtaining full nonpeptide ligands for class IMHC proteins. They represent interesting tools for altering theresponse of B*2705-restricted T cells to naturally occurringantigenic peptides and for designing novel synthetic vaccines.

Acknowledgment—We thank the calculation center of the ETH Zur-ich for allocation of computer time on the CRAY J90 and PARAGONsupercomputers.

REFERENCES

1. Heemels, M. T., and Ploegh, H. L. (1995) Annu. Rev. Biochem. 64, 643–6912. Madden, D. R. (1995) Annu. Rev. Immunol. 13, 587–6823. Madden, D. R., Gorga, J. C., Strominger, J. L., and Wiley, D. C. (1992) Cell 70,

1035–10484. Guo, H. C., Jardetzky, T. S., Garrett, T. P. J., Lane, W. S., Strominger, J. L.,

and Wiley, D. C. (1992) Nature 360, 364–3665. Saper, M. A., Bjorkman, P. J., and Wiley, D. C. (1991) J. Mol. Biol. 219,

277–3196. Guo, H. C., Madden, D. R., Silver, M. L., Jardetzky, T. S., Gorga, J. C.,

Strominger, J. L., and Wiley, D. C. (1993) Proc. Natl. Acad. Sci. U. S. A. 90,8053–8057

7. Falk, K., Rotzschke, O., Stevanovic, S., Jung, G., and Rammensee, H.-G. (1991)Nature 351, 290–296

8. Rammensee, H.-G., Friede, T., and Stevanovic, S. (1995) Immunogenetics 41,178–228

9. Jameson, S. C., Carbone, F. R., and Bevan, M. J. (1993) J. Exp. Med. 177,1541–1550

10. Stryhn, A., Andersen, P. S., Pedersen, L. O., Svejgaard, A., Holm, A., Thorpe,C. J., Fugger, L., Buus, S., and Engberg, J. (1996) Proc. Natl. Acad. Sci.U. S. A. 93, 10338–10342

11. Garcia, K. C., Degano, M., Stanfield, R. L., Brunmark, A., Jackson, M. R.,Peterson, P. A., Teyton, L., and Wilson, I. A. (1996) Science 274, 209–219

12. Garboczi, D. N., Ghosh, P., Utz, U., Fan, Q. R., Biddison, W. E., and Wiley,D. C. (1996) Nature 384, 134–141

13. Brewerton, D. A., Hart, F. D., Nicholls, A., and Sturrock, R. D. (1973) Lancet2, 994–996

14. Benjamin, R., and Parham, P. (1990) Immunol. Today 11, 137–14215. Hill, A. V. S., Elvin, J., Willis, A. C., Aidoo, M., Allsopp, C. E. M., Gotch, F. M.,

Gao, X. M., Takiguchi, M., Greenwood, B. M., Townsend, A. R. M.,McMichael, A. J., and Whittle, H. C. (1992) Nature 360, 434–439

16. De Magistris, M. T., Alexander, J., Coggeshall, M., Altman, A., Gaeta, F. C. A.,Grey, H. M., and Sette, A. (1992) Cell 68, 625–634

17. Sloan-Lancaster, J., Evavold, B. D., and Allen, P. M. (1993) Nature 363,156–159

18. Ishioka, G. Y., Adorini, L., Guery, J.-C., Gaeta, F. C. A., LaFond, R., Alexander,J., Powell, M. F., Sette, A., and Grey, H. M. (1994) J. Immunol. 152,4311–4319

19. Kingsley, G., and Sieper, J. (1993) Immunol. Today 14, 387–39120. Rognan, D., Scapozza, L., Folkers, G. and Daser, A. (1995) Proc. Natl. Acad.

Sci. U. S. A. 92, 753–75721. Weiss, G. A., Collins, E. J., Garboczi, D. N., Wiley, D. C., and Schreiber, S. L.

(1995) Chem. Biol. 2, 401–40722. Bouvier, M., and Wiley, D. C. (1996) Proc. Natl. Acad. Sci. U. S. A. 93,

4583–458823. Pearlman, D. A., Case, D. A., Caldwell, J. C., Ross, W. S., Cheatham, T. E.,

Ferguson, D. M., Seibel, G. L., Singh, U. C., Weiner, P. K., and Kollman,P. A. (1995) AMBER 4.1, University of California, San Francisco

24. Cornell, W. D., Cieplak, P., Bayly, C. I., Gould, I. R., Merz, K. M., Jr., Ferguson,D. M., Spellmeyer, D. M., Fox, T., Caldwell, J. W., and Kollman, P. E. (1995)J. Am. Chem. Soc. 117, 5179–5197

25. Fox, T., Scanlan, T. S., and Kollman, P. A. (1997) J. Am. Chem. Soc. 119,11571–11577

26. Frisch, M. J., Trucks, G. W., Schlegel, H. B., Gill, P. M. W., Johnson, B. G.,Robb, M. A., Cheeseman, J. R., Keith, T. A., Petersson, G. A., Montgomery,J. A., Raghavachari, K., Al-Laham, M. A., Zakrzewski, V. G., Ortiz, J. V.,Foresman, J. B., Peng, C. Y., Ayala, P. A., Wong, M. W., Andres, J. L.,Replogle, E. S., Gomperts, R., Martin, R. L., Fox, D. J., Binkley, J. S.,Defrees, D. J., Baker, J., Stewart, J. P., Head-Gordon, M., Gonzalez, C., andPople, J. A. (1995) Gaussian 94, Revision C.3, Gaussian, Inc., Pittsburgh,PA

27. Bayly, C. I., Cieplak, P., Cornell, W. D., and Kollman, P. A. (1993) J. Phys.Chem. 97, 10269–10280

28. Cieplak, P., Cornell, W. D., Bayly, C., and Kollman, P. A. (1995) J. Comput.

Nonnatural HLA-B27 Ligands19078

Chem. 16, 1357–137729. Rognan, D., Krebs, S., Kuonen, O., Lamas, J. R., Lopez de Castro, J. A., and

Folkers, G. (1997) J. Comput. Aided Mol. Des. 11, 463–47830. Galocha, B., Lamas, J. R., Villadangos, J. A., Albar, J. P., and Lopez de Castro,

J. A. (1996) Tissue Antigens 48, 509–51831. Ljunggren, H. G., Stam, N. J., Ohlen, C., Neefjes, J. J., Hoglund, P., Heemels,

M. T., Bastin, J., Schumacher, T. N., Townsend, A., Karre, K., and Ploegh,H. L. (1990) Nature 346, 476–480

32. Ellis, S. A., Taylor, C., and McMichael, A. (1982) Hum. Immunol. 5, 49–5933. Nagai, K., and Thogersen, H. C. (1987) Methods Enzymol. 131, 266–28034. Weiss, G. A., Valentekovich, R. J., Collins, E. J., Garboczi, D. N., Lane, W. S.,

Schreiber, S. L., and Wiley, D. C. (1996) Proc. Natl. Acad. Sci. U. S. A. 93,10945–10948

35. Bouvier, M., and Wiley, D. C. (1994) Science 265, 398–40236. Privalov, P. L., and Gill, S. J. (1988) Adv. Protein Chem. 39, 191–23437. Myers, J. K., Pace, C. N., and Scholtz, J. M. (1995) Protein Sci. 4, 2138–214838. Rognan, D., Scapozza, L., Folkers, G., and Daser, A. (1994) Biochemistry 33,

11476–1148539. Jardetzky, T. S., Lane, W. S., Robinson, R. A., Madden, D. R., and Wiley, D. C.

(1991) Nature 353, 326–32940. Rotzschke, O., Falk, K., Stevanovic, S., Gnau, V., Jung, G., and Rammensee,

H. G. (1994) Immunogenetics 39, 74–77

41. Reich, Z., Altman, J. D., Boniface, J. J., Lyons, D. S., Kozono, H., Ogg, G.,Morgan, C., and Davis, M. M. (1997) Proc. Natl. Acad. Sci. U. S. A. 94,2495–2500

42. Garcia, F., Marina, A., and Lopez de Castro, J. A. (1996) Tissue Antigens 49,215–221

43. Muller, H.-M., and Seebach, D. (1993) Angew. Chem. Int. Ed. Engl. 32,477–502

44. Plattner, D. A., Brunner, A., Dobler, M., Muller, H.-M., Petter, W., Zbinden, P.,and Seebach, D. (1993) Helv. Chim. Acta 76, 2004–2033

45. Fields, G. B., and Noble, R. I. (1990) Int. J. Peptide Protein Res. 35, 161–21446. Morgan, C. S., Holton, J. M., Olafson, B. D., Bjorkman, P. J., and Mayo, S. L.

(1997) Protein Sci. 6, 1771–177347. Fahnestock, M. L., Johnson, J. L., Feldman, R. M. R., Tsomides, T. J., Mayer,

J., Narhi, L. O., and Bjorkman, P. J. (1994) Biochemistry 33, 8149–815848. Stern, L. J., Brown, J. H., Jardetzsky, T. S., Gorga, J. C., Urban, R. G.,

Strominger, J. L., and Wiley, D. C. (1994) Nature 368, 215–22149. Brooks, J. M., Murray, R. J., Thomas, W. A., Kurilla, M. G., and Rickinson,

A. B. (1993) J. Exp. Med. 178, 897–88750. Huet, S., Nixon, D. F., Rothbard, J. B., Townsend, A., Ellis, S. A., and

McMichael, A. J. (1990) Int. Immunol. 2, 311–31651. Nicholls, A., Sharp, K. A., and Honig, B. (1991) Proteins Struct. Funct. Genet.

11, 281–296

Nonnatural HLA-B27 Ligands 19079

ANEXO -7-

Articles

Nonapeptide Analogues Containing (R)-3-Hydroxybutanoate and -HomoalanineOligomers: Synthesis and Binding Affinity to a Class I MajorHistocompatibility Complex Protein

Sorana Poenaru,‡,§ Jose R. Lamas,† Gerd Folkers,| Jose A. Lopez de Castro,† Dieter Seebach,*,‡ andDidier Rognan*,|

Laboratory for Organic Chemistry, Swiss Federal Institute of Technology, Universitatstrasse 16, CH-8092 Zurich, Switzerland,Centro de Biologia Molecular ”Severo Ochoa”, Facultad de Ciencias, Universidad Autonoma de Madrid,E-28049 Madrid, Spain, and Department of Pharmacy, Swiss Federal Institute of Technology, Winterthurerstrasse 190,CH-8057 Zurich, Switzerland

Received December 4, 1998

Crystal structures of antigenic peptides bound to class I MHC proteins suggest that chemicalmodifications of the central part of the bound peptide should not alter binding affinity to theMHC restriction protein but could perturb the T-cell response to the parent epitope. In oureffort in designing nonpeptidic high-affinity ligands for class I MHC proteins, oligomers of(R)-3-hydroxybutanoate and(or) -homoalanine have been substituted for the central part of aHLA-B27-restricted T-cell epitope of viral origin. The affinity of six modified peptides to theB*2705 allele was determined by an in vitro stabilization assay. Four out of the six designedanalogues presented an affinity similar to that of the parent peptide. Two compounds, sharingthe same stereochemistry (R,R,S,S) at the four stereogenic centers of the nonpeptidic spacer,bound to B*2705 with a 5-6-fold decreased affinity. Although the chiral spacers do not stronglyinteract with the protein active site, there are configurations which are not accepted by theMHC binding groove, probably because of improper orientation of some lateral substituents inthe bound state and different conformational behavior in the free state. However wedemonstrate that -amino acids can be incorporated in the sequence of viral T-cell epitopeswithout impairing MHC binding. The presented structure-activity relationships open the doorto the rational design of peptide-based vaccines and of nonnatural T-cell receptor antagonistsaimed at blocking peptide-specific T-cell responses in MHC-associated autoimmune diseases.

Introduction

Class I major histocompatibility complex (MHC)-encoded proteins play a key role in the intracellularimmune surveillance by selectively binding to intracel-lular peptide antigens and presenting them at the cellsurface to T-cell receptors (TCRs) of cytotoxic T-lym-phocytes (CTL).1 Due to the genetically encoded dis-crepancy between the limited number of class I alleles(about six) expressed by each individual and the infinitenumber of potential antigenic peptides (usually non-amers), class I MHC molecules must bind diverse setsof foreign peptides with a broad specificity but a highaffinity. Numerous structural data on class I MHC-peptide complexes are nowadays available at the three-dimensional level2 and provide an explanation for thatparadigm. The 27 reported X-ray structures (for ninedifferent class I MHC molecules) illustrate a peptide-

independent recognition in which both terminal endsof the peptide backbone are tightly bonded to conservedresidues of the MHC binding groove. Allele specificityis ensured by the interaction of anchoring side chains,3,4

usually at positions P2, P3 (Pn standing for position n),and the C-terminus with polymorphic pockets5 of thehost MHC protein. The central part of the boundpeptides (from positions 4 to 8) generally zigzags6 orbulges7 out of the binding groove and thus allowsvariation in the length of the bound peptides (from 8 to11 amino acids). Systematic peptide mutation8 andX-ray structure of MHC-peptide-TCR ternary com-plexes9-12 show that this central part whose conforma-tion is not complementary to that of the MHC proteinis the major contact area for R TCRs that trigger theT-cell response to the foreign peptide.

The tight association observed between MHC expres-sion and susceptibility or resistance to autoimmunedisorders led us to consider class I MHC proteins asparticularly interesting targets for the selective immu-notherapy of autoimmune diseases. At least two waysof shunting the T-cell response to autoantigens usingsmall-molecular-weight molecules have been proposed.The first one involving MHC blockade by a high-affinity

* To whom correspondence should be addressed. D. Seebach: fax,+41.1.632 11 44; e-mail, seebach@org.chem.ethz.ch. D. Rognan: fax,+41.1.635 68 84; e-mail, didier@pharma.ethz.ch.

‡ Laboratory for Organic Chemistry, SFIT.† Universidad Autonoma de Madrid.| Department of Pharmacy, SFIT.§ Present address: Department of Chemistry, University of Cali-

fornia, Berkeley, CA 94720-1460.

2318 J. Med. Chem. 1999, 42, 2318-2331

10.1021/jm981123l CCC: $18.00 © 1999 American Chemical SocietyPublished on Web 06/08/1999

competitor13 is unlikely as MHC-bound peptides at thecell surface are almost impossible to displace.14 The onlyway to overcome this drawback would be to supply thepeptide competitor in liposomes15 or as lipopeptides16

into the endoplasmic reticulum where assembly of theclass I MHC-peptide complexes takes place. The secondinhibiting pathway relying on TCR antagonism17 sug-gests that the presentation of the epitope to autoreactiveT-cells would be antagonized by a modified peptideanalogue. This approach is much more promising sinceonly a few TCRs at the surface of CTLs need to betargeted,18 whereas MHC blockade requires saturationof all MHC binding sites at the cell surface.19 Twoprerequisites are however necessary for designing TCRantagonists: (i) a good affinity to the MHC-restrictionprotein, (ii) a fast dissociation of the correspondingMHC-ligand complex to the TCR.20,21 The few TCRantagonists known to date are all peptide analogues forwhich one TCR-anchoring amino acid has been mu-tated.22 Unfortunately, the poor stability and pharma-cokinetic properties inherent to their peptidic naturepreclude their general use as immunosuppressors. Thus,there is a need for designing high-affinity nonpeptideligands for class I MHC proteins. Rather few variationsaround the canonical nonapeptide structure have beendescribed up to date.23 Peptides bearing unnatural L-or D-R-amino acids at MHC-anchoring positions,24-29

reduced peptide bond pseudopeptides,30 retroinversoanalogues,31 poly-N-acylated amines,32 or incorporationof a -homoglycine residue at the peptide N-terminus33

have been reported. An alternative strategy we initiated3 years ago is to replace the central TCR-binding aminoacids by various nonpeptidic spacers: oligomers ofaminoalkanoates,24,34 phenanthridine derivatives,35 orpoly(ethylene glycol) loops.36 All these chemical modi-fications led to ligands that could associate with class IMHC proteins but always with a slight decrease inbinding affinity when compared to that of the parentpeptides.

We recently described the replacement of a naturalpentameric peptide sequence (from positions P4 to P8)by (R)-3-hydroxybutanoate (R-HB) oligomers in HLA-B27-binding nonapeptides37 while enhancing 5-fold thebinding affinity for the MHC restriction protein.34

However, the partial hydrolysis of oligo-HB ester bonds,observed during the synthesis, suggests that theseanalogues should have very poor in vivo pharmacoki-netic properties because of their high sensitivity toesterases and peptidases. Recent reports on the remark-able enzymatic stability of -peptides38,39 led us toconsider oligomers of -amino acids as potential sur-rogates for the TCR-binding residues of class I MHC-binding peptides. Since -homoalanine (-HAla) is anisostere of HB, binding to HLA-B27 should thus beretained in light of our previous results on poly(esterpeptides) (PEPs).34 However, the low solubility of pro-tected -HAla oligomers in any solvent40 could be adrawback to the synthesis and biological evaluation ofthese compounds. To increase the solubility in water ofcompounds containing four -HAla units,41 a positivelycharged peptide epitope from the HIV-1 gp120 protein(G314RAFVTIGK322, one-letter amino acid code), knownto bind well to HLA-B*2705,34 was chosen as templatefor the reported chemical modifications (Table 1). Fur-

thermore, combining -HAla and HB oligomers shouldenhance the solubility of the resulting spacers inchlorinated solvents42 and thus facilitate their syntheticaccess. Most of the designed peptide analogues bindindeed with a high affinity to a class I MHC protein(HLA-B*2705 allele), whose expression is associatedwith susceptibility to severe autoimmune diseases.43

Results and DiscussionChemistry. Synthesis of the derivative 11 was

achieved using a fragment-type coupling strategy(Scheme 1). Boc-protected -homoalanine38,44 and benzyl3-hydroxybutanoate (2)45 were coupled using DCC andDMAP as activating reagents46 to give 3. Deprotectionof the amino group under acidic conditions gave theamino ester 4, whereas hydrogenolysis of the benzylester gave acid 5. Coupling of 4 with the commerciallyavailable Boc-protected alanine under traditional HOBt/EDC peptide conditions47-49 gave the fully protectedderivative 6 whose benzyl ester group was cleaved byH2 (Pd/C) to yield acid 7. 1H NMR measurements ledto assignment of all signals to the corresponding protonsof the R-amino acid, of the HB unit, and of the -HAlamoiety.

To obtain the second fragment 9 with free amino andprotected carboxy group, the acid 5 was coupled withH-Lys(Z)-OBn, using the HOBt/EDC strategy to give 8(80%), treatment of which with a saturated HCl/dioxanesolution yielded the corresponding HCl salt 9. Com-pound 10, consisting of six building blocks, was then

Scheme 1a

a (a) DCC, DMAP, CH2Cl2; (b) HCl/ether (satd); (c) H2, Pd/C,MeOH; (d) Boc-Ala-OH, HOBt, EDC, Et3N, CH2Cl2; (e) H2, Pd/C,MeOH; (f) HCl•H-Lys(Z)-OBn, HOBt, EDC, Et3N, CH2Cl2; (g) HCl/dioxane (satd); (h) HOBt, EDC, DIEA, CH2Cl2; (i) TFA/CH2Cl2,1:1.

Nonapeptide Analogue Binding Affinity to MHC Protein Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 2319

obtained in 90% yield by coupling of 7 with 9.50

Subsequent cleavage of the Boc protecting group led tothe amino ester 11 which was used for the next couplingwith arginine, without further purification (Scheme 4).

The derivative 20, with HB and -HAla incorporatedin different sequence (Scheme 2), was synthesized by

application of the same fragment coupling strategy. Thefragment 15 was synthesized in a linear fashion,coupling first Boc--HAla-OH (1) with H-Lys(Z)-OBn togive dipeptide 12 (85%) which was deprotected to givethe HCl salt 13 (HCl in dioxane). The amino functional-ity of 13, set free in situ by the base present in thereaction mixture, was coupled with 1 to give the fullyprotected compound 14 (74% from 12). The Boc groupof 14 was cleaved to yield 15. The fragment 17 wasobtained by ester formation between the hydroxy dimer1637,51 and Fmoc-protected alanine (using DCC, DMAP),and cleavage of the tert-butyl ester group (50% TFA)gave the desired compound 18. It should be noted thatby using a small amount of DMAP (0.05 equiv), noracemization was observed upon coupling.37 The aminofunctionality, generated by in situ deprotonation of theHCl salt 15, was coupled with the acid group of 18(HOBt/EDC) to give the fully protected compound 19in 92% yield. The Fmoc protecting group was thencleaved (20% diethylamine in DMF52) to give the aminoester 20. It is noteworthy that the backbone of com-pound 19 varies from that of 10 only by the respectivepositions of HB and -HAla in the sequence. Howevertheir respective solubilities in organic solvents are verydifferent. Compound 10 is highly soluble in chlorinated

Scheme 2a

a (a) HCl•H-Lys(Z)-OBn, HOBt, EDC, Et3N; (b) HCl/dioxane(satd); (c) 1, HOBt, EDC, DIEA, DMF; (d) HCl/dioxane (satd); (e)Fmoc-Ala-OH, DCC, DMAP, CH2Cl2; (f) TFA/CH2Cl2, 1:1; (g)HOBt, EDC, DIEA, CH2Cl2/DMF, 1:1; (h) Et2NH/DMF, 1:4.

Scheme 3a

a (a) HCl•H-Lys(Z)-OBn, HOBt, EDC, DIEA, DMF; (b) TFA, 10min; (c) Boc-Ala-OH, HOBt, EDC, DIEA, DMF; (d) TFA, 10 min.

Scheme 4a

a (a) Boc-Arg(NO2)-OH, HOBt, EDC, DIEA, DMF; (b) TFA, 10min; (c) Boc-Gly-OH, HOBt, EDC, DIEA, DMF; (d) H2, Pd/BaSO4,TFE/AcOH, 4:1; (e) TFA, 10 min.

2320 Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 Poenaru et al.

solvents such as CH2Cl2, whereas 19 is poorly solubleeven in DMF.

The configurational isomers (diastereoisomers) 21a-dcontaining four -HAla units (Scheme 3) have beenpreviously prepared, starting from (S)- and (R)-Boc-Ala-OH and using a fragment coupling strategy.40 Theseoligomers are difficult to handle because of their lowsolubility in any solvent tested so far. For example, the1H NMR spectra of protected -HAla tetramers can onlybe measured using dimethyl-d6 sulfoxide as solvent andshow rather broad signals. During the synthesis, largeramounts of DMF were necessary to dissolve the -HAlaoligomers (c ) 0.05 M), as compared to those requiredin R-peptide synthesis, and reaction times were conse-quently longer. Therefore, the yields of the reactions arenot specified, since no purification is possible before thelast deprotection step, due to the poor solubility of thisclass of compounds.

The acids 21 were coupled with H-Lys(Z)-OBn usingHOBt/EDC, to provide the fully protected derivatives,the Boc protecting groups of which were cleaved (con-centrated TFA) to give the TFA salts 22 which wereprecipitated with ether and dried in high vacuum beforethe next reaction step: deprotonation and reaction withBoc-Ala-OH to yield, after another deprotection step, thecorresponding TFA salts 23.

During the cleavage of the Boc group, the Z protectinggroup of the lysine side chain was partially cleaved,giving rise to a byproduct (e5%) which has beendetected by mass spectrometry. Such debenzylationshave been previously reported by Merrifield et al.53

However, considering the much harsher conditions usedby the Merrifield group, the observed loss of the Z group,in our case, was surprising. Due to solubility problems,it was impossible to purify the intermediates at thisstage. Thus, we have carried the byproduct all alongthe following synthetic steps, with the consequence thatsome additional impurities were formed. After the lastdeprotection step, preparative HPLC purification stillgave the desired pure compounds (27-29). The aminofunctionalities of the derivatives 11, 20, and 23a-d(Scheme 4) were allowed to react with Boc-Arg(NO2)-OH,54 using the same coupling procedure as for theother coupling steps. Again, the Boc groups of the fullyprotected derivatives were cleaved with TFA to yieldcompounds 24, 25, and 26a-d which, in turn, werecoupled with Boc-Gly-OH to give the fully protectednonapeptides analogues. The NO2 and Z protectinggroups, as well as the benzyl ester group, were thensimultaneously removed by hydrogenation, using Pd/

BaSO4 as catalyst.56 Subsequent treatment with TFAled to cleavage of the Boc groups to give, after HPLCpurification, the desired nonapeptide analogues 27, 28,and 29a-d which were used for binding assays.

Binding Affinity to HLA-B*2705. The bindingaffinities of the modified peptides clearly show that thechirality of the spacer is important for recognition ofthe B*2705 protein. Compounds in which the chiralbuilding block linked with the C-terminus (PC) has (R)-configuration (27, 29a-b) were all potent ligands with

Table 1. Binding and Analytical Properties of Ligands 27, 28, and 29a-d

compd sequence C50a (µmol) HPLC (tR, min)b MSc (M + 1)+

HIV gp120d Gly-Arg-Ala-[Phe-Val-Thr-Ile-Gly]-Lys 2.8A68P1e Glu-Val-Ala-Pro-Pro-Glu-Tyr-His-Arg nbf

27 Gly-Arg-Ala-[(S--HAla-R-HB)2]-Lysg 2.8 18.7 773.328 Gly-Arg-Ala-[(R-HB)2-(S--HAla)2]-Lys 17.0 19.5 774.129a Gly-Arg-Ala-[(R--HAla)4]-Lys 2.8 16.4 771.929b Gly-Arg-Ala-[(S--HAla-R--HAla)2]-Lys 4.8 15.4 772.629c Gly-Arg-Ala-[(S--HAla)4]-Lys 6.0 15.3 771.929d Gly-Arg-Ala-[(R--HAla)2-(S--HAla)2]-Lys 30.0 15.4 772.0

a Concentration of ligand at which HLA-B*2705 fluorescence (measured by FMC analysis with an anti-B27 monoclonal antibody) onRMA-S cells was half the maximum obtained with that compound (see Experimental Section). b HPLC purification using a gradient of A(0.1% trifluoroacetic acid in water) and B (acetonitrile): 0-100% B, 60 min. c MALDI-TOF spectra, recorded on a Bruker Biflex instrument(Bruker-Franzen Analytik, Bremen, Germany) in linear mode. d HIV-1 glycoprotein 120 (314-322). e Self-peptide eluted from the HLA-A68 allotype.8 f No detectable binding at 10-4 M. g -HAla, -homoalanine; HB, 3-hydroxybutanoate.

Figure 1. Dynamic properties of complexes of B*2705 withtwo modified peptides (29b, 29d) and the reference HIV-1gp120 (314-322) peptide. (A) Intermolecular hydrogen-bondfrequencies, recorded for the whole 500-ps trajectory of HLA-B27-ligand complexes over 1000 conformations. Frequenciesbetween 25% and 50% and higher than 50% are displayed aswhite and gray columns, respectively. (B) Buried surface areasof HLA-B27-bound ligands 29b (white columns) and 29d (graycolumns) calculated on energy-minimized time-averaged con-formations. Surface areas were calculated using the MSprogram74 with a 1.4-Å radius probe.

Nonapeptide Analogue Binding Affinity to MHC Protein Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 2321

2322 Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 Poenaru et al.

binding affinities similar to that of the parent HIV-1gp120 peptide (Table 1). This observation is in agree-ment with our previous report on PEPs for which apenultimate (R)-HB unit was proposed to interact withthe MHC binding groove.34 Analogues bearing a moietyof (S)-configuration next to PC were less active, espe-cially compounds 28 and 29d sharing the same sequenceof (R,R,S,S)-configuration of the spacing oligomers(Table 1). However, an (S)-chiral spacer attached to PCdoes not necessarily prevent binding (see compound 29c,Table 1). Furthermore, it seems that certain configura-tions of the four spacing monomers are detrimental tobinding. Thus, the two weakest binders (28, 29d) sharethe same (R,R,S,S)-configuration at positions 4 to 7.Replacing ester by amide groups in the spacer (cf. 27with 29b and 28 with 29d) did not affect binding forboth high-affinity and low-affinity ligands. This resultis in agreement with the available X-ray structure56 ofa B*2705-nonapeptide complex, showing weak contri-butions of peptide bonds, located between the P4 andP8 positions, to the binding of a nonapeptide to HLA-B27.

Apart from binding potencies, it should be noticedthat HB-containing compounds 27 and 28 are probablystill sensitive to esterases, although stability studies onthese compounds have not been performed yet. We alsoexpect that the replacement of HB oligomers by -aminoacids in analogues 29a-d enhances the resistance ofthe modified peptides to enzymatic degradation.

Molecular Modeling of B*2705-Ligand Com-plexes. To find a rational explanation for the weakbinding of compounds with (R,R,S,S)-configuration ofthe chiral spacer, a 500-ps molecular dynamics (MD)study of the binary complexes between B*2705 andthree ligands was undertaken. Compound 29b waschosen as representative of the high-affinity peptides,whereas 29d was selected as representative of weakbinding ligands. The parent HIV peptide was selectedas reference. The trajectory of the three solvated com-plexes was stable after 350 ps, with rms deviations ofthe protein backbone from the starting X-ray coordi-nates of ca. 1.5 Å (data not shown). We previously usedatomic fluctuations of the bound ligands, as a criterion,for discriminating high-affinity from low-affinity pep-tides.24,33,57 In the present case, they were very similarfor ligands 29b and 29d. Thus, subtle differences mustcause the very different binding affinities of the twomodified peptides. In fact, recording the frequency ofthe MHC-ligand hydrogen bonds allows to distinguishthe two modified peptides. High-affinity ligands (HIVgp120, 29b) present many more hydrogen bonds to theHLA-B27 binding groove than the weak binding com-pound 29d (Figure 1A). Strong H-bonds with a fre-quency higher than 50% were remarkably identical inboth cases, but medium H-bonds (with frequenciesbetween 25% and 50%) are significantly in favor of 29b.A very similar pattern has already been observed for aset of four natural peptides binding to two closelyrelated HLA-B27 alleles.33 The major differences be-

tween the two nonnatural complexes could be correlatedwith the H-bond-donating strength of the N-terminus,well-known to significantly contribute to the bindingfree energy of nonapeptides to class I MHC proteins.58

The buried surface areas of each monomer of theprotein-bound ligand were also very similar with theexception of two residues P5, and PC (Figure 1B). P5corresponds to the second unit of the spacer (R-HAlain both cases). Depending on its environment in thesequence of the modified peptide, it is more or lessdeeply buried in the HLA-B27 binding groove. Withcompound 29b, the P5 position is significantly deeperinside the groove than with compound 29d (compareFigure 2A,B). However, this feature is unlikely to inducenearly a 10-fold difference in the binding affinity of thecorresponding ligands. The C-terminal amino acid (Lys)also shows a better fit in the case of the high-affinityligand (Figure 1B). As the C-terminal residue is animportant anchor to B*2705,56 this structural differenceshould also contribute to the improved binding of 29bversus 29d.

However, the computed properties of the two ana-logues bound to their target protein can only explain apart of the experimentally determined difference ofbinding affinities. The modeling study presented heretakes into account only enthalpic contributions to thebinding of each ligand to HLA-B*2705. As desolvationenergies and rotational/translational entropy lossesupon binding (assuming a conserved binding mode ofthe two compounds) should be very similar due to thestructural analogy of all modified peptides listed inTable 1, the 10-fold decreased binding of two analogues(28, 29d), having the same configuration, may be dueto different association rates and different conforma-tional populations in the free state. This feature hasalready been experimentally described for two relatedPEPs,34 for which the length of the polyester spacervaries. Hydrophobic -peptides are known to haveconformations strongly depending on the chirality oftheir residues and on the nature of their side chains.38,59

The (R,R,S,S)-configuration of four chiral monomers inlow-affinity ligands might lead to a conformational spacearrangement in the free state that is different from thatof high-affinity compounds (27, 28, 29a-c). The higher“strain energy” necessary to bring ligands 28 and 29dfrom the free to the bound state may partially contributeto the weaker binding of these two analogues. Unfor-tunately, simulating the free ligands, although compu-tationally easier, is very risky because they adopt nostable secondary structure, as concluded from their CDor NMR spectra.

ConclusionReplacing the central amino acids of class I MHC-

binding peptides by (R)-3-hydroxybutanoate and(or)-homoalanine oligomers leads to still high-affinityligands. Up to now, -amino acids have hardly been usedin medicinal chemistry. Some natural -amino acids(taurine, -aminobutyric acid, -aminoisobutyric acid)

Figure 2. Three-dimensional structure of HLA-B*2705 in complex with 29b (A) and 29d (B). Peptide positions are labeled atthe CR atom from 1 (P1) to 8 (P8). The backbone trace of the MHC antigen-binding domain (R1, R2) of the B*2705 protein isrepresented as bands (R helices), arrows ( strands), and tubes (coil). Altered peptides are displayed by a ball-and-stick model. Awhite arrow indicates the position of the second -ΗAla unit in both chiral spacers. The figure has been prepared using theMOLSCRIPT75 and RASTER3D76 programs.

Nonapeptide Analogue Binding Affinity to MHC Protein Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 2323

have been reported as agonists of the inhibitory glycinereceptor.60 Substituted -amino acids have also beendescribed as fibrinogen receptor GIIb/IIIa antagonists,61

-lactamase inhibitors,62 µ-opioid receptor agonists,63 orenkephalin-degrading enzyme inhibitors.64 Further-more, various -amino acids are found in naturalantibiotics, fungicides, and antineoplastic compounds.65

However, to the best of our knowledge, this is the veryfirst report of biologically active molecules containing-amino acid oligomers. The present study demonstratesthat -amino acids are valuable tools, indeed, fordesigning peptidomimetics of bioactive peptides. Bycontrast to most R-amino acid surrogates, the H-bondingproperties of backbone atoms, the backbone direction,or the side chain directionality might be similar innatural and -peptides, at the condition that the -pep-tide can adopt the biologically active conformation of itsnatural R-peptide analogue. Thus, if all side chains arenot mandatory for biological activity, -amino acids and,more generally, -peptides undoubtedly represent newpromising tools in medicinal chemistry. In the specialcase of class I MHC ligands, one might imagine to use-amino acids for replacing MHC anchors and(or) TCRcontact residues. Such altered peptides may thus leadto peptide-based vaccines and TCR antagonists, whichwould be stable to all common peptidases tested so far,including pronase, 20S proteasome, and proteinase K.

Experimental SectionAbbreviations: (R)--homoalanine (R--HAla), (S)--ho-

moalanine (S--HAla), dicyclohexylcarbodiimide (DCC), diiso-propylethylamine (DIEA), 4-(dimethylamino)pyridine (DMAP),dimethylformamide (DMF), N-(3-(dimethylamino)propyl)-N′-ethylcarbodiimide hydrochloride (EDC), 1-hydroxy-1H-benzo-triazole (HOBt), trifluoroacetic acid (TFA), trifluoroethanol(TFE), (R)-3-hydroxybutanoate (R-HB).

Chemistry. Dichloromethane (CH2Cl2) was dried over 4-Åmolecular sieves. Solvents for chromatography and workupprocedures were distilled from Sikkon (anhydrous CaSO4,Fluka). Triethylamine (Et3N) was distilled from CaH2 andstored over KOH. Amino acid derivatives were purchased fromBachem or Senn. All other reagents were used as received fromFluka.

1H (300-MHz) and 13C (75-MHz) NMR spectra were recordedon a Varian Gemini 300 spectrometer and are reported in ppmon the δ scale from TMS. Coupling constants are reported inhertz (Hz). FAB-MS spectra were obtained with a HitachiPerkin-Elmer RHU-6M using a 3-nitrobenzyl alcohol (3-NOBA) matrix. Elemental analyses were performed by theMicroanalytical Laboratory of the Laboratorium fur Orga-nische Chemie, ETH-Zurich (only analyses above 0.4% weregiven).

Chromatography generally refers to flash silica gel 60 (Fluka40-63 mm) and TLC (Merck Kieselgel 60 F254 plates), detec-tion with UV and ninhydrin. HPLC analyses were carried outon a C18 analytical column on a Knauer HPLC system (pumptype 64, EuroChrom 2000 integration package, degaser, UVdetector (variable-wavelength monitor)) using a linear gradientof (A) 0.1% CF3COOH in H2O and (B) MeCN at a flow rate of1 mL/min with UV detection at 220 nm. HPLC purificationwas carried out on a C8 preparative column on a KnauerHPLC system (pump type 64, programmer 50, UV detector(variable-wavelength monitor)) using a gradient of (A) 0.1%CF3COOH in H2O and (B) MeCN at a flow rate of 4 mL/minwith UV detection at 214 nm. Retention times (tR) are givenin min.

General Procedure A: Amino Acid Coupling. The freeamine or the appropriate salt (1 equiv) was dissolved in CH2-Cl2 or 50% CH2Cl2/DMF (0.1 M) under argon and cooled to 0°C. The reaction mixture was treated with a base (Et3N or

DIEA, 3 equiv). HOBt (1.25 equiv), the acid (1 equiv), and EDC(1.25 equiv) were then successively added. The reactionmixture was allowed to warm to room temperature and thenstirred for 18 h. The mixture was diluted with CH2Cl2 andwashed with 1 N HCl, saturated NaHCO3 solution, and brine.The organic layer was dried over anhydrous MgSO4, filtrated,and concentrated. The resulting residue was purified on silicagel to afford the pure product.

General Procedure B: Amino Acid Coupling. The freeamine or the appropriate salt (1 equiv) was dissolved in DMF(0.15 M) under argon and cooled to 0 °C. The reaction mixturewas treated with DIEA (3 equiv). HOBt (1.25 equiv), the acid(1 equiv), and EDC (1.25 equiv) were then successively added.The product was precipitated by the addition of a saturatedNaHCO3 solution. The precipitate was washed several timeswith saturated NaHCO3, 1 M KHSO4 solutions and H2O anddried 24 h under high vacuum to give the crude product whichwas utilized in the next step without further purification.

General Procedure C: Boc Cleavage Using a HClSolution. Under argon and at 0 °C, the Boc-protected com-pound was dissolved in a saturated HCl/(EtO2 or dioxane)solution. The mixture was stirred for 15 min to 1 h and thenevaporated. The resulting HCl salt was precipitated in ether,dried under high vacuum, and used for the next step withoutfurther purification.

General Procedure D: Boc Cleavage Using a TFASolution. Under argon and at 0 °C, the Boc-protected com-pound was dissolved in a TFA/CH2Cl2 (50-100%) solution. Themixture was stirred for 10 min to 1 h and then evaporated.The resulting TFA salt was precipitated in Et2O, dried underhigh vacuum, and used for the next step without furtherpurification.

General Procedure E: Final Deprotection. The fullyprotected compound was dissolved in TFE/CH3COOH (3/1),and a catalytic amount of 10% Pd/BaSO4 was added. Theapparatus was evacuated and flushed three times with H2, andthe mixture was stirred under an atmosphere of H2 for ca. 15h. The mixture was then filtrated though Celite, concentrated,and precipitated from Et2O. The resulting yellow-white CH3-COOH salt was then treated with concentrated TFA. After 15min, the crude product was precipitated from Et2O andpurified by HPLC.

Boc-S--HAla-R-HB-OBn (3). To a solution of the (R)-3-hydroxybutanoic benzyl ester (2)45 (1.90 g, 9.8 mmol) in CH2-Cl2 (30 mL) was added a solution of the acid 138 (2.00 g, 9.8mmol) in CH2Cl2 (30 mL) under argon, cooled to -5 °C. DCC(2.12 g, 10.3 mmol) and DMAP (0.09 g, 0.49 mmol) were added.The resulting mixture was allowed to warm to room temper-ature and then stirred for 24 h. The mixture was diluted withEt2O and washed with 1 N HCl, saturated NaHCO3 solution,and brine. The organic layer was dried over anhydrous MgSO4,filtrated, and concentrated. The residue was purified on silicagel (20% Et2O/pentane) and gave compound 3 (3.25 g, 88%)as a colorless oil. 1H NMR (300 MHz, CDCl3): δ 7.37-7.33(m, 5H ar), 5.36-5.26 (m, 1H, CHO), 5.14 (AB, J ) 12.1, 1H,OCH2Ph), 5.09 (AB, J ) 12.1, 1H, OCH2Ph), 5.00-4.90 (m,1H, NH), 4.10-3.96 (m, 1H, CHN), 2.66 (dd, ABX, J ) 7.5,15.6, 1H, CH2CHO), 2.54 (dd, ABX, J ) 5.3, 15.6, 1H, CH2-CHO), 2.43 (dd, ABX, J ) 5.3, 14.9, 1H, CH2CHN), 2.37 (dd,ABX, J ) 5.9, 14.9, 1H, CH2CHN), 1.43 (s, 9H, tBu), 1.29 (d,J ) 6.2, 3H, Me of HB), 1.17 (d, J ) 6.5, 3H, Me of -HAla).13C NMR (75 MHz, CDCl3): δ 171.00, 170.42, 155.36, 135.93,128.84, 128.60, 67.58, 66.65, 43.60, 40.77, 28.46, 20.31, 19.96.FAB-MS: m/z 759 26, (2M + 1)+, 308 64, (M + 1)+, 280(100).

HCl•H-S--HAla-R-HB-OBn (4). According to generalprocedure C, compound 3 (227 mg, 0.6 mmol) was treated witha saturated HCl/Et2O solution (6 mL). The resulting HCl salt4 was obtained as a white precipitate and used in the nextcoupling step without further purification.

Boc-S--HAla-R-HB-OH (5). The benzyl-protected com-pound 3 (900 mg, 2.9 mmol) was dissolved in MeOH (20 mL);catalytic amounts of 10% Pd/C (90 mg) and acetic acid (0.1

2324 Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 Poenaru et al.

mL) were added. The apparatus was evacuated and flushedthree times with H2, and the mixture was stirred under anatmosphere of H2 for ca. 8 h. Subsequent filtration thoughCelite and concentration under reduced pressure yielded theacid 5 (558 mg, 84%) as a colorless oil which was identified byNMR and used for the next coupling step without purification.

Boc-Ala-S--HAla-R-HB-OBn (6). According to generalprocedure A, to a solution in CH2Cl2 (26 mL) of HCl salt 4 (1equiv, 2.61 mmol) was added Et3N (1.45 mL, 10.4 mmol). HOBt(440 mg, 3.3 mmol), Boc-Ala-OH (4.94 mg, 2.61 mmol), andthen EDC (623 mg, 3.3 mmol) were successively added to thereaction. The residue was purified by recrystallization (Et2O/pentane, 2/5) to give compound 6 (994 mg, 85%) as a whitesolid. 1H NMR (300 MHz, CDCl3): δ 7.35-7.326 (m, 5H ar),6.70 (br d, J ) 6.8, 1H, NH), 5.33-5.27 (m, 1H, CHO), 5.12 9(s, 2H, OCH2Ph), 5.12-5.06 (m, 1H, NH), 4.36-4.22 (m, 1H,CHN), 4.18-4.06 (m, 1H, CHN), 2.74-2.66 (m, 1H, CH2CHO),2.58 (dd, ABX, J ) 5.0, 15.57, 1H, CH2CHO), 2.42 (d, J ) 5.3,2H, CH2CHN), 1.44 (s, 9H, tBu), 1.34-1.29 (m, 6H, 2 Me), 1.17(d, J ) 6.8, 3H, Me). 13C NMR (75 MHz, CDCl3): δ 172.12,170.87, 135.80, 128.86, 128.67, 128.62, 67.87, 66.78, 42.02,40.69, 40.33, 28.38, 19.98, 19.66, 18.64. FAB-MS: m/z 901 11,(2M + 1)+, 451 100, (M + 1)+9, 351 (48).

Boc-Ala-S--HAla-R-HB-OH (7). The benzyl-protectedcompound 6 (675 mg, 1.5 mmol) was dissolved in MeOH (10mL); catalytic amounts of 10% Pd/C (70 mg) and acetic acid(0.1 mL) were added. The apparatus was evacuated andflushed three times with H2, and the mixture was stirred underan atmosphere of H2 for ca. 8 h. Subsequent filtration thoughCelite and concentration under reduced pressure yielded theacid 7 in almost quantitative yield as a colorless oil which wasused for the next coupling step without purification.

Boc-S--HAla-R-HB-Lys(Z)-OBn (8). According to generalprocedure A, to a solution in CH2Cl2 (15 mL) of HCl‚H-Lys-(Z)-OBn (784 mg, 1.9 mmol) was added Et3N (1.08 mL, 7.7mmol). HOBt (326 mg, 2.4 mmol), compound 5 (558 mg, 1.9mmol), and then EDC (460 mg, 2.4 mmol) were successivelyadded to the reaction. The residue was purified on silica gel(50% Et2O/pentane) to give compound 8 (964 mg, 78%) as awhite solid. 1H NMR (300 MHz, CDCl3): δ 7.38-7.29 (m, 10Har), 6.96-6.90 (m, 1H, NH), 5.27-5.17 (m, 1H, CHO), 5.22-5.09 (m, 2H, OCH2Ph), 5.08 (s, 2H, OCH2Ph), 4.97-4.94 (m,1H, NH), 4.90-4.86 (m, 1H, NH), 4.66-4.58 (m, 1H, CHN),4.16-4.05 (m, 1H, CHN), 3.16-3.09 (m, 2H, CH2NHZ), 2.54-2.27 (m, 4H, CH2CHN, CH2CHO), 1.90-1.60 (m, 4H, 2 CH2),1.50-1.28 (m, 2H, CH2), 1.40 (s, 9H, tBu), 1.30 (d, J ) 6.2,3H, Me), 1.13 (d, J ) 6.8, 3H, Me). 13C NMR (75 MHz,CDCl3): δ 172.57, 170.84, 169.74, 156.83,155.65, 136.87,135.69, 128.86, 128.76, 128.71, 128.55, 111.19, 79.61, 68.19,67.16, 66.74, 52.19, 43.84, 42.30, 41.96, 40.61, 31.71, 29.46,28.43, 22.42, 20.90, 19.46. FAB-MS: m/z 642 12, (M + 1)+,542 (100).

HCl•H-S--HAla-R-HB-Lys(Z)-OBn (9). According to gen-eral procedure C, compound 8 (712 mg, 1.1 mmol) was treatedwith a saturated HCl/dioxane solution (10 mL). After 15 min,the reaction was completed and the solvent was evaporated.The resulting HCl salt 9 was obtained in almost quantitativeyield as a white precipitate and used in the next coupling stepwithout further purification.

Boc-Ala-(S--HAla-R-HB)2-Lys(Z)-OBn (10). According togeneral procedure A, to a solution in CH2Cl2 (15 mL) of theHCl salt 9 (1 equiv, 1.5 mmol) was added DIEA (1.0 mL, 6.0mmol). HOBt (253 mg, 1.8 mmol), compound 7 (1 equiv, 1.5mmol), and then EDC (358 mg, 1.8 mmol) were successivelyadded to the reaction. The residue was purified on silica gel(ethyl acetate/hexane, 9/1) to give compound 10 (884 mg, 90%)as a white solid foam. 1H NMR (300 MHz, CDCl3): δ 7.36-7.30 (m, 10H ar), 7.10-7.00 (m, 3H, NH), 5.34 (d, J ) 7.5, 1H,NH), 5.30-5.10 (m, 3H, CHO, NH), 5.21-5.09 (m, 2H, OCH2-Ph), 5.07 (s, 2H, OCH2Ph), 4.60-4.53 (m, 1H, CHN), 4.44-4.31 (m, 2H, CHN), 4.17-4.08 (m, 1H, CHN), 3.16-3.08 (m,2H, CH2NHZ), 2.55-2.29 (m, 8H, CH2CHN, CH2CHO), 1.88-1.60 (m, 2H, CH2), 1.58-1.22 (m, 4H, CH2), 1.42 (s, 9H, tBu),1.31 (d, J ) 6.8, 3H, Me), 1.29 (d, J ) 6.2, 3H, Me), 1.25 (d, J

) 6.2, 3H, Me), 1.17 (d, J ) 6.8, 3H, Me), 1.15 (d, J ) 6.5, 3H,Me). 13C NMR (75 MHz, CDCl3): δ 172.49, 170.79, 170.08,169.58, 156.98, 136.84, 135.64, 128.86, 128.78, 128.73, 128.54,128.37, 128.29, 68.61, 68.27, 67.19, 66.74, 52.39, 42.47, 42.21,41.24, 40.40, 31.42, 29.40, 28.41, 22.36, 20.06, 19.85, 19.71,18.81. FAB-MS: m/z 906 6, (M + Na)+, 884 32, (M + 1)+,784 (100). Anal. (C45H65N5O13•H2O) C, H, N.

TFA•H-Ala-(S--HAla-R-HB)2-Lys(Z)-OBn (11). Accord-ing to general procedure D, compound 10 (654 mg, 0.74 mmol)was treated with a CH2Cl2/TFA (1:1) solution (6 mL). After30 min, the reaction was completed and the solvent wasevaporated. The TFA resulting salt 11 was obtained in almostquantitative yield as a white precipitate (from Et2O) and usedin the next coupling step without further purification.

Boc-S--HAla-Lys(Z)-OBn (12). According to general pro-cedure A, to a solution in DMF (50 mL) of the HCl salt of Lys-(Z)-OBn (2.00 g, 4.9 mmol) was added Et3N (2.04 mL, 14.7mmol). HOBt (0.83 g, 6.1 mmol), the acid Boc--HAla-OH (1)(1.00 g, 4.9 mmol), and then EDC (1.17 g, 6.1 mmol) weresuccessively added to the reaction. The residue was purifiedby recrystallization (ethyl acetate/hexane, 20/1) to give com-pound 12 (2.30 g, 85%) as a white solid. 1H NMR (300 MHz,CDCl3): δ 7.37-7.26 (m, 10H ar), 6.48-6.36 (m, 1H, NH),5.22-5.10 (m, 3H, OCH2Ph, NH), 5.09 (s, 2H, OCH2Ph), 4.95-4.87 (m, 1H, NH), 4.62-4.56 (m, 1H, CHN), 4.00-3.90 (m, 1H,CHN), 3.17-3.10 (m, 2H, CH2NHZ), 2.46-2.32 (m, 2H, CH2-CHN), 1.90-1.60 (m, 2H, CH2), 1.50-1.25 (m, 4H, CH2), 1.42(s, 9H, tBu), 1.16 (d, J ) 6.8, 3H, Me). 13C NMR (75 MHz,CDCl3): δ 172.44, 156.90, 136.85, 135.56, 128.89, 128.78,128.65, 128.37, 67.29, 66.74, 52.01, 44.18, 42.52, 40.41, 31.61,29.30, 28.46, 22.15, 20.60. FAB-MS: m/z 556 28, (M + 1)+,456 (100).

HCl•H-S--HAla-Lys(Z)-OBn (13). According to generalprocedure C, compound 12 (1.00 g, 1.8 mmol) was treated witha saturated HCl/dioxane solution (20 mL). After 30 min, thereaction was completed and the solvent was evaporated. Theresulting HCl salt 13 was obtained in almost quantitative yieldas a white precipitate (from Et2O) and used in the nextcoupling step without further purification.

Boc-(S--HAla)2-Lys(Z)-OBn (14). According to generalprocedure A, to a solution in DMF (5 mL) of the HCl salt 13(1 equiv, 1.8 mmol) was added DIEA (0.92 mL, 5.4 mmol).HOBt (304 mg, 2.2 mmol), the acid Boc--HAla-OH (1) (365mg, 1.8 mmol), and then EDC (430 mg, 2.2 mmol) weresuccessively added to the reaction. The residue was purifiedby recrystallization (CH3Cl/hexane, 20/1) to give compound 14(850 mg, 74%) as a white solid. 1H NMR (300 MHz, CDCl3):δ 7.38-7.26 (m, 10H ar), 6.68-6.65 (m, 2H, NH), 5.23-5.09(m, 4H, OCH2Ph, NH), 5.09 (s, 2H, OCH2Ph), 4.58-4.50 (m,1H, CHN), 4.30-4.20 (m, 1H, CHN), 4.00-3.90 (m, 1H, CHN),3.18-3.10 (m, 2H, CH2NHZ), 2.42-2.22 (m, 4H, CH2CHN),1.90-1.62 (m, 2H, CH2), 1.52-1.25 (m, 4H, CH2), 1.42 (s, 9H,tBu), 1.19 (d, J ) 6.5, 3H, Me), 1.15 (d, J ) 6.8, 3H, Me). 13CNMR (75 MHz, CDCl3): δ 172.44, 155.78, 128.88, 128.78,128.62, 128.37, 67.30, 66.74, 52.27, 43.10, 40.29, 31.29, 28.56,28.48, 22.21, 20.24. FAB-MS: m/z 663 22, (M + Na)+, 64138, (M + 1)+, 541 (100).

HCl•H-(S--HAla)2-Lys(Z)-OBn (15). According to generalprocedure C, compound 15 (712 mg, 1.1 mmol) was treatedwith a saturated HCl/dioxane solution (10 mL). After 30 min,the reaction was completed and the solvent was evaporated.The HCl resulting salt 15 was obtained in almost quantitativeyield as a white precipitate (from Et2O) and used in the nextcoupling step without further purification.

Fmoc-Ala-(R-HB)2-OtBu (17). To a solution of the hydroxyderivative 1651 (1 equiv, 4.4 mmol) in CH2Cl2 (40 mL) wasadded Fmoc-Ala-OH (1.45 g, 4.4 mmol) under argon, and themixture was cooled to - 5 °C. DCC (0.95 g, 4.62 mmol) andDMAP (0.04 g, 0.22 mmol) were added, and the resultingmixture was allow to warm to room temperature and thenstirred for 24 h. The mixture was diluted with Et2O andwashed with 1 N HCl, saturated NaHCO3 solution, and brine.The organic layer was dried over anhydrous MgSO4, filtrated,and concentrated. The residue was purified on silica gel (Et2O/

Nonapeptide Analogue Binding Affinity to MHC Protein Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 2325

pentane, 2/3) to give compound 17 (700 mg, 30%) as a whitefoam. 1H NMR (300 MHz, CDCl3): δ 7.77-7.75 (m, 2H ar),7.61-7.59 (m, 2H ar), 7.42-7.37 (m, 2H ar), 7.34-7.26 (m,2H ar), 6.49 (d, J ) 7.2, NH), 5.38-5.22 (m, 2H, CHO), 4.44-4.32 (m, 3H, CHN, CH2 of Fmoc), 4.25-4.20 (m, 1H, CH ofFmoc), 2.69-2.37 (m, 4H, CH2CHO), 1.43 (s, 9H, tBu), 1.30(d, J ) 6.2, 3H, Me), 1.29 (d, J ) 6.2, 3H, Me), 1.26 (d, J )6.5, 3H, Me). 13C NMR (75 MHz, CDCl3): δ 172.44, 169.69,169.37, 155.91, 144.23, 144.12, 141.58, 127.94, 125.33, 120.20,81.10, 68.68, 68.21, 67.09, 49.83, 47.27, 40.05, 40.88, 28.09,19.80, 19.67, 18.67. FAB-MS: m/z 540 6, (M + 1)+, 484 (100).

Fmoc-Ala-(R-HB)2-OH (18). According to general proce-dure C, compound 17 (600 mg, 1.1 mmol) was treated with aTFA/CH2Cl2 (1:1) solution (6 mL). After 30 min, the reactionwas completed and the solvent was evaporated. The acid 18was obtained in almost quantitative yield as a yellow oil andused in the next coupling step without further purification.

Fmoc-Ala-(R-HB)2-(S--HAla)2-Lys(Z)-OBn (19). Accord-ing to general procedure A, to a solution in CH2Cl2/DMF (1:1,12 mL) of the HCl salt 15 (1 equiv, 1.10 mmol) was addedDIEA (0.75 mL, 4.40 mmol). HOBt (185 mg, 1.37 mmol), theacid 18 (1 equiv, 1.10 mmol), and then EDC (262 mg, 1.37mmol) were successively added to the reaction. After 15 hreaction, some of the product precipitated in the reactionmixture. The CH2Cl2 was then evaporated, and the resultingoil was precipitated in a saturated NaHCO3 solution. Theprecipitate was washed several times with saturated NaHCO3,KHSO4 (1 N) solutions and finally with H2O and dried 15 hunder high vacuum. The presence of compound 19 wasconfirmed by 1H and 13C NMR and MS spectra. Compound 19was used without further purification. 1H NMR (300 MHz,DMSO-d6): δ 8.28-8.22 (m, 1H, NH), 7.30-7.94 (m, 1H, NH),7.80-7.74 (m, 1H, NH), 7.75-7.65 (m, 2H ar), 7.42-7.24 (m,16H ar), 7.22-7.16 (m, 1H, NH), 5.23-5.02 (m, 3H, CHO, NH),5.208 (s, 2H, OCH2Ph), 5.96 (s, 2H, OCH2Ph), 4.32-4.14 (m,2H, CHN), 4.10-4.22 (m, 2H, CHN), 2.96-2.87 (m, 2H, CH2-NHZ), 2.40-2.00 (m, 8H, CH2CHN, CH2CHO), 1.70-1.51 (m,2H, CH32), 1.40-1.10 (m, 4H, 2 CH2), 1.23 (d, J ) 6.8, 3H,Me), 1.13 (d, J ) 5.0, 3H, Me), 1.04 (d, J ) 5.9, 3H, Me), 1.03-0.98 (m, 2 Me). 13C NMR (75 MHz, DMSO-d6): δ 172.28,170.52, 169.11, 168.09, 167.99, 156.26, 155.99, 144.02, 140.90,137.43, 136.14, 128.55, 128.47, 128.15, 127.94, 127.84, 127.77,127.21, 125.38, 120.25, 68.34, 67.77, 65.83, 65.64, 65.12, 63.44,51.88, 49.42, 46.58, 44.25, 42.18, 42.05, 41.82, 41.64, 30.38,28.86, 23.18, 22.55, 19.82, 19.75, 19.51, 19.38, 19.15. FAB-MS: m/z 1028 23, (M + Na)+, 1006 41, (M + 1)+, 809 (84),713 (100).

H-Ala-(R-HB)2-(S--HAla)2-Lys(Z)-OBn (20). The Fmoc-protected compound 19 (520 mg, 0.52 mmol) was dissolved inDMF/Et2NH (9:1, 4 mL) under argon and cooled to 0 °C. Themixture was stirred for 1-2 h, and concentration underreduced pressure yielded the crude amine 20 which wasidentified by NMR and used without further purification.

TFA•H-(R--HAla)4-Lys(Z)-OBn (22a). According to gen-eral procedure B, to a solution in DMF (11 mL) of HCl•H-Lys-(Z)-OBn (442 mg, 1.09 mmol) was added DIEA (0.56 mL, 3.27mmol). HOBt (184 mg, 1.36 mmol), the acid Boc-(-HAla)4-OH (21a)40 (1 equiv, 1.09 mmol), and then EDC (260 mg, 1.36mmol) were successively added to the reaction. The residuewas dried under high vacuum and used without furtherpurification as it is not soluble in any flash chromatographyor HPLC solvent. The presence of the desired Boc-(R--HAla)4-Lys(Z)-OBn (723 mg, 82% crude) was confirmed by 1H and 13CNMR and MS spectra. Further treatment with TFA (3.6 mL),according to the general procedure D, gave the TFA salt 22a,which was used without further purification.

TFA•H-(S--HAla-R--HAla)2-Lys(Z)-OBn (22b). Accord-ing to general procedure B, to a solution in DMF (2 mL) ofHCl•H-Lys(Z)-OBn (62 mg, 0.15 mmol) was added DIEA (0.08mL, 0.45 mmol). HOBt (26 mg, 0.19 mmol), the acid Boc-(S--HAla-R--HAla)2-OH (21b)40 (1 equiv, 0.15 mmol), and thenEDC (36 mg, 0.19 mmol) were successively added to thereaction. The resulting precipitate was dried under highvacuum and used without further purification as it is not

soluble in any flash chromatography or HPLC solvent. Thepresence of the desired Boc-(S--HAla-R--HAla)2-Lys(Z)-OBn(11 mg, 90% crude) was confirmed by 1H and 13C NMR andMS spectra. Further treatment with TFA (0.55 mL), accordingto the general procedure D, gave the TFA salt 22b, which wasused without further purification.

TFA•H-(S--HAla)4-Lys(Z)-OBn (22c). According to gen-eral procedure B, to a solution in DMF (15 mL) of HCl•H-Lys-(Z)-OBn (515 mg, 1.27 mmol) was added DIEA (0.65 mL, 3.81mmol). HOBt (214 mg, 1.59 mmol), the acid Boc-(S--HAla)4-OH (21c)40 (1 equiv, 1.27 mmol), and then EDC (304 mg, 1.59mmol) were successively added to the reaction. The resultingprecipitate was dried under high vacuum and used withoutfurther purification as it is not soluble in any flash chroma-tography or HPLC solvent. The presence of the desired Boc-(S--HAla)4-Lys(Z)-OBn (819.1 mg, 79% crude) was confirmedby 1H and 13C NMR and MS spectra. Further treatment withTFA (4 mL), according to the general procedure D, gave theTFA salt 22c, which was used without further purification.

TFA•H-(R--HAla)2-(S--HAla)2-Lys(Z)-OBn (22d). Ac-cording to general procedure B, to a solution in DMF (11 mL)of HCl•H-Lys(Z)-OBn (442 mg, 1.09 mmol) was added DIEA(0.56 mL, 3.27 mmol). HOBt (184 mg, 1.36 mmol), the acidBoc-(R--HAla)2-(S--HAla)2-Lys(Z)-OH (21d)40 (1 equiv, 1.09mmol), and then EDC (260 mg, 1.36 mmol) were successivelyadded to the reaction. The resulting precipitate was driedunder high vacuum and used without further purification asit is not soluble in any flash chromatography or HPLC solvent.The presence of the desired Boc-(R--HAla)2-(S--HAla)2-Lys-(Z)-OBn (796 mg, 90% crude) was confirmed by 1H and 13CNMR and MS spectra. Further treatment with TFA (3.9 mL),according to the general procedure D, gave the TFA salt 22d,which was used without further purification.

TFA•H-Ala-(R--HAla)4-Lys(Z)-OBn (23a). According togeneral procedure B, to a solution in DMF (7 mL) of the TFAsalt 22a (1 equiv, 0.89 mmol) was added DIEA (0.61 mL, 3.56mmol). HOBt (150 mg, 1.11 mmol), Boc-Ala-OH (202 mg, 1.06mmol), and then EDC (212 mg, 1.1 mmol) were successivelyadded to the reaction. The resulting precipitate was driedunder high vacuum and used without further purification asit is not soluble in any flash chromatography or HPLC solvent.The presence of the desired Boc-Ala-(R--HAla)4-Lys(Z)-OBn(680 mg, 87% crude) was confirmed by 1H and 13C NMR andMS spectra. Further treatment with TFA (3.1 mL), accordingto the general procedure D, gave the TFA salt 23a, which wasused without further purification.

TFA•H-Ala-(S--HAla-R--HAla)2-Lys(Z)-OBn (23b). Ac-cording to general procedure B, to a solution in DMF (2 mL)of the TFA salt 22b (1 equiv, 0.135 mmol) was added DIEA(0.092 mL, 0.540 mmol). HOBt (23 mg, 0.169 mmol), Boc-Ala-OH (31 mg, 0.162 mmol), and then EDC (32 mg, 0.169 mmol)were successively added to the reaction. The resulting pre-cipitate was dried under high vacuum and used withoutfurther purification as it is not soluble in any flash chroma-tography or HPLC solvent. The presence of the desired Boc-Ala-(S--HAla-R--HAla)2-Lys(Z)-OBn (73 mg, 61% crude) wasconfirmed by 1H and 13C NMR and MS spectra. Furthertreatment with TFA (0.33 mL), according to the generalprocedure D, gave the TFA salt 23b, which was used withoutfurther purification.

TFA•H-Ala-(S--HAla)4-Lys(Z)-OBn (23c). According togeneral procedure B, to a solution in DMF (10 mL) of the TFAsalt 22c (1 equiv, 1.00 mmol) was added DIEA (0.68 mL, 4.00mmol). HOBt (169 mg, 1.25 mmol), Boc-Ala-OH (227 mg, 1.20mmol), and then EDC (239 mg, 1.25 mmol) were successivelyadded to the reaction. The resulting precipitate was driedunder high vacuum and used without further purification asit is not soluble in any flash chromatography or HPLC solvent.The presence of the desired Boc-Ala-(S--HAla)4-Lys(Z)-OBn(762 mg, 86% crude) was confirmed by 1H and 13C NMR andMS spectra. Further treatment with TFA (3.5 mL), accordingto the general procedure D, gave the TFA salt 23c, which wasused without further purification.

TFA•H-Ala-(R--HAla)2-(S--HAla)2-Lys(Z)-OBn (23d).

2326 Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 Poenaru et al.

According to general procedure B, to a solution in DMF (10mL) of the TFA salt 22d (770 mg, 0.93 mmol) was added DIEA(0.64 mL, 3.72 mmol). HOBt (157 mg, 1.16 mmol), Boc-Ala-OH (211 mg, 1.12 mmol), and then EDC (215 mg, 1.16 mmol)were successively added to the reaction. The resulting pre-cipitate was dried under high vacuum and used withoutfurther purification as it is not soluble in any flash chroma-tography or HPLC solvent. The presence of the desired Boc-Ala-(R--HAla)2-(S--HAla)2-Lys(Z)-OBn (490 mg, 60% crude)was confirmed by 1H and 13C NMR and MS spectra. Furthertreatment with TFA (2.2 mL), according to the generalprocedure D, gave the TFA salt 23d, which was used withoutfurther purification.

TFA•H-Arg(NO2)-Ala-(S--HAla-R-HB)2-Lys(Z)-OBn (24).According to general procedure A, to a solution in CH2Cl2/DMF(4:3, 8 mL) of the TFA salt 11 (1 equiv, 0.74 mmol) was addedDIEA (0.38 mL, 2.22 mmol). HOBt (125 mg, 0.92 mmol), Boc-Arg(NO2)-OH (259 mg, 0.81 mmol), and then EDC (176 mg,0.92 mmol) were successively added to the reaction. Theresulting residue was purified on silica gel (CH2Cl2/MeOH, 9/1)to give compound Boc-Arg(NO2)-Ala-(S--HAla-R-HB)2-Lys(Z)-OBn (650 mg, 82%) as a fine white powder. 1H NMR (300 MHz,CDCl3): δ 8.45-8.30 (m, 1H, NH), 7.80-7.60 (m, 3H, NH),7.8-7.28 (m, 10H ar), 7.20-7.12 (m, 3H, NH), 5.70-5.60 (m,1H, NH), 5.34-5.26 (m, 1H, NH), 5.26-5.06 (m, 4H, CHO,OCH2Ph), 5.07 (s, 2H, OCH2Ph), 4.60-4.50 (m, 1H, CHN),4.46-4.2 (m, 4H, CHN), 3.36-3.24 (m, 2H, CH2NHC), 3.16-3.07 (m, 2H, CH2NHZ), 2.53-2.33 (m, 8H, CH2CHN, CH2-CHO), 1.90-1.60 (m, 6H, CH2), 1.54-1.10 (m, 4H, CH2) 1.41(s, 9H, tBu), 1.34 (d, J ) 6.8, 3H, Me), 1.28 (d, J ) 6.2, 3H,Me), 1.24 (d, J ) 6.5, 3H, Me), 1.20-1.16 (m, 6H, Me). 13CNMR (75 MHz, CDCl3): δ 172.61, 170.35, 170.27, 169.77,157.09, 136.82, 135.61, 128.87, 128.80, 128.54, 128.37, 128.23,80.46, 68.68, 68.40, 67.22, 65.96, 52.48, 49.54, 42.48, 42.16,40.46, 31.36, 29.41, 28.39, 24.85, 22.41, 20.02, 19.88, 19.74,19.66, 18.19. FAB-MS: m/z 1107 60, (M + Na)+, 1085 100,(M + 1)+, 985 (22), 713 (7). Further treatment with TFA (5mL), according to the general procedure D, gave the TFA salt24, which was used without further purification.

TFA•H-Arg(NO2)-Ala-(R-HB)2-(S--HAla)2-Lys(Z)-OBn(25). According to general procedure B, to a solution in DMF(6 mL) of the TFA salt 20 (1 equiv, 0.52 mmol) was addedDIEA (0.26 mL, 1.55 mmol). HOBt (87 mg, 0.65 mmol), Boc-Arg(NO2)-OH (198 mg, 0.62 mmol), and then EDC (123 mg,0.65 mmol) were successively added to the reaction. Theresulting precipitate was dried under high vacuum and usedwithout further purification as it is not soluble in any flashchromatography or HPLC solvent. The presence of the desiredBoc-Arg(NO2)-Ala-(R-HB)2-(S--HAla)2-Lys(Z)-OBn (409 mg,77% crude) was confirmed by 1H and 13C NMR and MS spectra.Further treatment with TFA (1.4 mL), according to the generalprocedure D, gave the TFA salt 25, which was used withoutfurther purification.

TFA•H-Arg(NO2)-Ala-(R--HAla)4-Lys(Z)-OBn (26a). Ac-cording to general procedure B, to a solution in DMF (9 mL)of the TFA salt 23a (1 equiv, 0.77 mmol) was added DIEA (0.53mL, 3.08 mmol). HOBt (130 mg, 0.96 mmol), Boc-Arg(NO2)-OH (295 mg, 0.92 mmol) and then EDC (183 mg, 0.96 mmol)were successively added to the reaction. The resulting pre-cipitate was dried under high vacuum and used withoutfurther purification as it is not soluble in any solvent to bepurified. The presence of the desired Boc-Arg(NO2)-Ala-(R--HAla)4-Lys(Z)-OBn (757 mg, 81% crude) was confirmed by 1Hand 13C NMR and MS spectra. Further treatment with TFA(2.7 mL), according to the general procedure D, gave the TFAsalt 26a, which was used without further purification.

TFA•H-Arg(NO2)-(S--HAla-R--HAla)2-Lys(Z)-OBn (26b).According to general procedure B, to a solution in DMF (9 mL)of the TFA salt 23b (1 equiv, 0.83 mmol) was added DIEA(0.057 mL, 0.33 mmol). HOBt (14 mg, 0.104 mmol), Boc-Arg-(NO2)-OH (32 mg, 0.099 mmol), and then EDC (20 mg, 0.104mmol) were successively added to the reaction. The resultingprecipitate was dried under high vacuum and used withoutfurther purification as it is not soluble in any flash chroma-

tography or HPLC solvent. The presence of the desired Boc-Arg(NO2)-(S--HAla-R--HAla)2-Lys(Z)-OBn (74 mg) was con-firmed by FAB-MS. Further treatment with TFA (0.5 mL),according to the general procedure D, gave the TFA salt 26b,which was used without further purification.

TFA•H-Arg(NO2)-Ala-(S--HAla)4-Lys(Z)-OBn (26c). Ac-cording to general procedure B, to a solution in DMF (9 mL)of the TFA salt 23c (1 equiv, 0.86 mmol) was added DIEA (0.59mL, 3.44 mmol). HOBt (145 mg, 1.07 mmol), Boc-Arg(NO2)-OH (329 mg, 1.03 mmol), and then EDC (205 mg, 1.07 mmol)were successively added to the reaction. The resulting pre-cipitate was dried under high vacuum and used withoutfurther purification as it is not soluble in any flash chroma-tography or HPLC solvent. The presence of the desired Boc-Arg(NO2)-Ala-(S--HAla)4-Lys(Z)-OBn (860 mg, 82% crude)was confirmed by 1H and 13C NMR and MS spectra. Furthertreatment with TFA (3.1 mL), according to the generalprocedure D, gave the TFA salt 26c, which was used withoutfurther purification.

TFA•H-Arg(NO2)-Ala-(R--HAla)2-(S--HAla)2-Lys(Z)-OBn (26d). According to general procedure B, to a solutionin DMF (9 mL) of the TFA salt 23d (1 equiv, 0.45 mmol) wasadded DIEA (0.31 mL, 1.81 mmol). HOBt (76 mg, 0.56 mmol),Boc-Arg(NO2)-OH (172 mg, 0.54 mmol), and then EDC (107mg, 0.56 mmol) were successively added to the reaction. Theresulting precipitate was dried under high vacuum and usedwithout further purification as it is not soluble in any flashchromatography or HPLC solvent. The presence of the desiredBoc-Arg(NO2)-Ala-(R--HAla)2-(S--HAla)2-Lys(Z)-OBn (460 mg,84% crude) was confirmed by 1H and 13C NMR and MS spectra.Further treatment with TFA (2.2 mL), according to the generalprocedure D, gave the TFA salt 26d, which was used withoutfurther purification.

H-Gly-Arg-Ala-(S--HAla-R-HB)2-Lys-OH (27). Accordingto general procedure B, to a solution in DMF (6 mL) of theTFA salt 24 (1 equiv, 0.50 mmol) was added DIEA (0.26 mL,1.51 mmol). HOBt (85 mg, 0.63 mmol), Boc-Gly-OH (97 mg,0.55 mmol), and then EDC (124 mg, 0.63 mmol) were succes-sively added to the reaction. The resulting precipitate wasdried under high vacuum and used without further purificationas it is not soluble in any solvent to be purified. The presenceof the desired Boc-Gly-Arg(NO2)-Ala-(S--HAla-R-HB)2-Lys-(Z)-OBn (447 mg, 68% crude) as a fine yellow-white powderwas confirmed by 1H and 13C NMR and MS spectra.

According to general procedure E, Boc-Gly-Arg(NO2)-Ala-(S--HAla-R-HB)2-Lys(Z)-OBn (300 mg, 0.26 mmol) was dis-solved in TFE/CH3COOH (3:1, 4 mL) and hydrogenated in thepresence of Pd/BaSO4 (10%, 60 mg). The resulting precipitatewas purified by HPLC C8 (5-40% B, 30 min), tR 12.5 min, togive after lyophilization the pure compound 27 in about 40%yield. 1H NMR (300 MHz, D2O): δ 5.24-5.10 (m, 2H, CHO),4.34-4.28 (m, 2H, CHN), 4.22-4.15 (m, 3H, CHN), 3.84 (s,2H, CH2N), 3.19 (t, J ) 6.8, 2H, CH2NHC), 3.00-2.95 (m, 2H,CH2NH2), 2.58-2.42 (m, 8H, CH2CHO, CH2CHN), 1.92-1.58(m, 8H, CH2), 1.50-1.40 (m, 2H, CH2), 1.32 (d, J ) 7.5, 3H,Me), 1.26 (d, J ) 6.2, 3H, Me), 1.25 (d, J ) 6.2, 3H, Me), 1.18-1.13 (m, 6H, Me). FAB-MS: m/z 811 10, (M + K)+, 795 32,(M + Na)+, 773 100, (M + 1)+. Purity by analytical HPLC(0-100% B, 60 min, tR 18.7 min) >99%.

H-Gly-Arg-Ala-(R-HB)2-(S--HAla)2-Lys-OH (28). Accord-ing to general procedure D, to a solution in DMF (5 mL) ofthe TFA salt 25 (1 equiv, 0.37 mmol) was added DIEA (0.19mL, 1.11 mmol). HOBt (62 mg, 0.46 mmol), Boc-Gly-OH (77mg, 0.44 mmol), and then EDC (88 mg, 0.46 mmol) weresuccessively added to the reaction. The resulting precipitatewas dried under high vacuum and used without furtherpurification as it is not soluble in any flash chromatographyor HPLC solvent. The presence of the desired Boc-Gly-Arg-(NO2)-Ala-(R-HB)2-(S--HAla)2-Lys(Z)-OBn (252 mg, 60% crude)was confirmed by 1H and 13C NMR and MS spectra.

According to general procedure E, Boc-Gly-Arg(NO2)-Ala-(R-HB)2-(S--HAla)2-Lys(Z)-OBn (200 mg, 0.17 mmol) wasdissolved in TFE/CH3COOH (3:1, 3.5 mL) and hydrogenatedin the presence of Pd/BaSO4 (10%, 40 mg). The resulting

Nonapeptide Analogue Binding Affinity to MHC Protein Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 2327

precipitate was purified by HPLC (10-40% B, 30 min), tR 6.2min, to give after lyophilization the pure compound 28 in about25% yield. 1H NMR (300 MHz, D2O): δ 5.22-5.06 (m, 2H,CHO), 4.28-4.20 (m, 3H, CHN), 4.16-4.04 (m, 2H, CHN), 3.76(s, 2H, CH2N), 3.16-3.10 (m, 2H, CH2NHC), 2.93-2.86 (m,2H, CH2NH2), 2.65-2.49 (m, 2H, CH2CHO), 2.43-2.32 (m, 2H,CH2CHO), 2.36 (d, J ) 7.2, 2H, CH2CHN), 2.26 (d, J ) 7.2,2H, CH2CHN), 1.85-1.53 (m, 8H, CH2), 1.41-1.28 (m, 2H,CH2), 1.35 (d, J ) 7.2, 3H, Me), 1.20-1.14 (m, 6H, Me), 1.08-1.03 (m, 6H, Me). FAB-MS: m/z 811 12, (M + K)+, 795 23,(M + Na)+, 773 100, (M + 1)+. Purity by analytical HPLC(0-100% B, 60 min, tR 19.5 min) >99%.

H-Gly-Arg-Ala-(R--HAla)4-Lys-OH (29a). According togeneral procedure B, to a solution in DMF (7 mL) of the TFAsalt 26a (1 equiv, 0.68 mmol) was added DIEA (0.47 mL, 2.71mmol). HOBt (115 mg, 0.85 mmol), Boc-Gly-OH (143 mg, 0.82mmol), and then EDC (163 mg, 0.85 mmol) were successivelyadded to the reaction. The precipitate was dried under highvacuum and used without further purification as it is notsoluble in any flash chromatography or HPLC solvent. Thepresence of the desired Boc-Gly-Arg(NO2)-Ala-(R--HAla)4-Lys-(Z)-OBn (622 mg, 80% crude) was confirmed by 1H and 13CNMR and MS spectra.

According to general procedure E, Boc-Gly-Arg(NO2)-Ala-(R--HAla)4-Lys(Z)-OBn (200 mg, 0.18 mmol) was dissolvedin TFE/CH3COOH (3:1, 3.5 mL) and hydrogenated in thepresence of Pd/BaSO4 (10%, 40 mg). The resulting precipitatewas purified by HPLC (5-40% B, 30 min), tR 17.20 min, togive after lyophilization the pure compound 29a in about 25%yield. 1H NMR (300 MHz, D2O): δ 4.32-4.24 (m, 2H, CHN),4.21-4.07 (m, 5H, CHN), 3.85-3.80 (m, 2H, CH2N), 3.20-3.14(m, 2H, CH2NHC), 2.98-2.92 (m, 2H, CH2NH2), 2.50-2.25 (m,8H, CH2CHN), 1.90-1.58 (m, 8H, CH2), 1.47-1.36 (m, 2H,CH2), 1.32 (d, J ) 7.2, 3H, Me), 1.14-1.09 (m, 12H, Me). 13CNMR (75 MHz, D2O): δ 178.57, 176.08, 175.29, 56.37, 55.39,52.77, 46.22, 46.11, 45.20, 43.36, 43.15, 41.95, 32.66, 31.13,28.98, 27.05, 24.86, 22.00, 19.43. FAB-MS: m/z 1542 9, (2M+2)+, 771 100, (M + 1)+. Purity by analytical HPLC (0-100% B, 60 min, tR 16.4 min) >99%.

H-Gly-Arg-Ala-(S--HAla-R--HAla)2-Lys-OH (29b). Ac-cording to general procedure B, to a solution in DMF (2 mL)of the TFA salt 26b (1 equiv, 0.068 mmol) was added DIEA(0.046 mL, 0.28 mmol). HOBt (12 mg, 0.085 mmol), Boc-Gly-OH (14 mg, 0.082 mmol), and then EDC (16 mg, 0.085 mmol)were successively added to the reaction. The resulting pre-cipitate was dried under high vacuum and used withoutfurther purification as it is not soluble in any flash chroma-tography or HPLC solvent. The presence of the desired Boc-Gly-Arg(NO2)-(S--HAla-R--HAla)2-Lys(Z)-OBn (70 mg) wasconfirmed by FAB-MS.

According to general procedure E, Boc-Gly-Arg(NO2)-(S--HAla-R--HAla)2-Lys(Z)-OBn (70 mg, 0.06 mmol) was dis-solved in TFE/CH3COOH (3:1, 1 mL) and hydrogenated in thepresence of Pd/BaSO4 (10%, 10 mg). The resulting precipitatewas purified by HPLC (5-40% A, 30 min), tR 8.6 min, to giveafter lyophilization the pure compound 29b in about 25% yield.1H NMR (300 MHz, D2O): δ 4.26-4.18 (m, 2H, CHN), 4.18-4.04 (m, 5H, CHN), 3.76-3.73 (m, 2H, CH2N), 3.12-3.07 (m,2H, CH2NHC), 2.91-2.85 (m, 2H, CH2NH2), 2.40-2.32 (m, 2H,CH2CHN), 2.30-2.22 (m, 6H, CH2CHN), 1.85-1.50 (m, 8H,CH2), 1.40-1.30 (m, 2H, CH2), 1.26-1.22 (m, 3H, Me), 1.09-1.02 (m, 12H, Me). FAB-MS: m/z 771 (86, [M + 1]+). Purityby analytical HPLC (0-100% B, 60 min, tR 15.4 min) >99%.

H-Gly-Arg-Ala-(S--HAla)4-Lys-OH (29c). According togeneral procedure B, to a solution in DMF (8 mL) of the TFAsalt 26c (1 equiv, 0.77 mmol) was added DIEA (0.53 mL, 3.1mmol). HOBt (131 mg, 0.97 mmol), Boc-Gly-OH (163 mg, 0.93mmol), and then EDC (185 mg, 0.97 mmol) were successivelyadded to the reaction. The resulting precipitate was driedunder high vacuum and used without further purification asit is not soluble in any flash chromatography or HPLC solvent.The presence of the desired Boc-Gly-Arg(NO2)-Ala-(S--HAla)4-Lys(Z)-OBn (719 mg, 82% crude) was confirmed by 1H and 13CNMR and MS spectra.

According to general procedure E, Boc-Gly-Arg(NO2)-Ala-(S--HAla)4-Lys(Z)-OBn (340 mg, 0.30 mmol) was dissolved inTFE/CH3COOH (3:1, 4 mL) and hydrogenated in the presenceof Pd/BaSO4 (10%, 60 mg). The resulting precipitate waspurified by HPLC (10-40% B, 30 min), tR 6.8 min, to give afterlyophilization the pure compound 29c in about 30% yield. 1HNMR (300 MHz, D2O): δ 4.34-4.27 (m, 2H, CHN), 4.24-4.09(m, 5H, CHN), 3.84-3.81 (m, 2H, CH2N), 3.21-3.15 (m, 2H,CH2NHC), 2.99-2.93 (m, 2H, CH2NH2), 2.45-2.41 (m, 2H,CH2C9HN), 2.40-2.26 (m, 6H, CH2CHN), 1.92-1.57 (m, 8H,CH2), 1.47-1.35 (m, 2H, CH2), 1.31 (d, J ) 7.2, 3H, Me), 1.15-1.09 (m, 12H, Me). 13C NMR (75 MHz, D2O): δ 178.59, 176.58,176.50, 176.08, 175.30, 170.09, 159.82, 56.22, 55.30, 52.71,46.25, 45.22, 44.78, 43.37, 43.13, 42.00, 32.78, 31.13, 28.99,27.02, 24.85, 22.16, 21.97, 19.50. FAB-MS: m/z 1542 17, (2M+ 2)+, 771 100, (M + 1)+. Purity by analytical HPLC (0-100% B, 60 min, tR 15.3 min) >99%.

H-Gly-Arg-Ala-(R--HAla)2-(S--HAla)2-Lys-OH (29d).According to general procedure B, to a solution in DMF (5 mL)of the TFA salt 26d (1 equiv, 0.42 mmol) was added DIEA(0.29 mL, 1.68 mmol. HOBt (71 mg, 0.52 mmol), Boc-Gly-OH(88 mg, 0.50 mmol), and then EDC (100 mg, 0.52 mmol) weresuccessively added to the reaction. The resulting precipitatewas dried under high vacuum and used without furtherpurification as it is not soluble in any flash chromatographyor HPLC solvent. The presence of the desired Boc-Gly-Arg-(NO2)-Ala-(R--HAla)2-(S--HAla)2-Lys(Z)-OBn (333 mg, 70%crude) was confirmed by 1H and 13C NMR and MS spectra.

According to general procedure E, Boc-Gly-Arg(NO2)-Ala-(R--HAla)2-(S--HAla)2-Lys(Z)-OBn (150 mg, 0.13 mmol) wasdissolved in TFE/CH3COOH (3:1, 3 mL) and hydrogenated inthe presence of Pd/BaSO4 (10%, 25 mg). The resulting pre-cipitate was purified by HPLC (2-40% B, 30 min), tR 20.5 min,to give after lyophilization the pure compound 29d in about20% yield. 1H NMR (300 MHz, D2O): δ 4.36-4.28 (m, 2H,CHN), 4.27-4.12 (m, 5H, CHN), 3.87-3.83 (m, 2H, CH2N),3.23-3.18 (m, 2H, CH2NHC), 3.01-2.96 (m, 2H, CH2NH2),2.346-2.43 (m, 2H, CH2CHN), 2.41-2.31 (m, 6H, CH2CHN),1.92-1.60 (m, 8H, CH2), 1.50-1.40 (m, 2H, CH2), 1.35 (d, J )7.5, 3H, Me), 1.17-1.11 (m, 12H, Me). FAB-MS: m/z 793 15,(M + Na)+, 771 100, (M + 1)+. Purity by analytical HPLC(0-100% B, 60 min, tR 15.4 min) >80%.

Molecular Dynamics Simulations. Molecular mechanicsand dynamics calculations were realized using the AMBER5.0 package66 using the parm96 parameter set and an all-atomforce-field representation.67 Force-field parameters for the esterbonds were taken from the literature.68 Atomic charges for thenew monomers (R-HB, S--HAla, R--HAla) were calculatedusing the GAUSSIAN94 package69 and the HF/6-31G* basisset, by fitting atom-centered charges to an ab initio electro-static potential, using the RESP method.70 Initial coordinatesfor the MHC-ligand complexes were obtained from the X-raystructure of HLA-B*270556 (Protein Data Bank code 1hsa) aspreviously described.33,34 The spacers were substituted for thenatural pentapeptide sequence using the SYBYL 6.3 modelingpackage (TRIPOS Assoc., Inc., St. Louis, MO). From a startingfully extended conformation, dihedral angles of the main chainbetween P3 and P9 were modified in order to reproduce acorrect trans geometry for the newly introduced amide or esterbonds. The ligand was first relaxed by 1000 steps of conjugategradient energy minimization while maintaining the proteinfixed. It was then submitted to a 100-ps Simulated annealing(SA) protocol in order to sample the broadest possible confor-mational space. Starting with random velocities assigned ata temperature of 1000 K, the peptide was first coupled to aheat bath at 1000 K using a temperature coupling constantTτ of 0.2 ps and then linearly cooled to 50 K for the next 50 pswhile strengthening Tτ to a value of 0.05 ps. During these 100ps, no protein atom was allowed to move. As the simulatedannealing was performed in vacuo, a distance-dependentdielectric function (ε ) 4r) was used. A twin cutoff (10.0, 15.0Å) was used to calculate nonbonded electrostatic interactionsat every minimization step and every nonbonded pair listupdate (10 steps), respectively.

2328 Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 Poenaru et al.

From the last SA conformer, 13 counterions (9 Na+ and 4Cl- ions) were then placed at electrostatic minima to neutralizethe protein, using the CION routine of AMBER.66 It was thensolvated in a 10-Å thick TIP3P water shell. After the solventwas minimized by 1000 steps of steepest descent, the solvent(water and counterions) was equilibrated by 25-ps MD at 300K. The solvent was minimized again, and the fully solvatedcomplex was finally relaxed by 1000 steps of steepest descent.The obtained coordinates were then used as a starting pointfor a 500-ps MD simulation at 300 K. To avoid large driftsfrom the protein crystal structure, a weak positional harmonicconstraint of 0.05 kcal‚mol-1‚Å-1 was applied to backboneatoms of B*2705. As the solvent was implicitly taken intoaccount, a constant dielectric function (ε ) 1) was utilized. Forthe whole trajectory, the same twin cutoff (10-15 Å) was usedfor calculating nonbonded interactions, and the nonbonded pairlist was updated every 10 steps. The SHAKE algorithm wasused on hydrogens with a tolerance of 0.00025 Å, a time stepof 2 fs, and Berendsen temperature coupling with separatecoupling of solute and solvent atoms to the heat bath.Coordinates, velocities, and energies were saved every 0.5 ps.All computations were done using the parallel version ofAMBER5.0 implemented on a CRAY J90 cluster and anINTEL paragon machine. The analyses of molecular dynamicstrajectories were achieved using in-house routines and theCARNAL module of AMBER.66

Epitope Stabilization Assay. The quantitative assay usedwas previously described.71 Briefly, RMA-S transfectantsexpressing B*2705 were used. These are murine cells withimpaired TAP-mediated peptide transport and low surfaceexpression of (empty) class I MHC molecules, which can beinduced at 26 °C72 and stabilized at the cell surface throughbinding of exogenously added ligands. These cells were incu-bated at 26 °C for 24 h. After this time they were incubatedfor 1 h at 26 °C with 10-4-10-9 M peptides, transferred to 37°C, and collected after 4 h for flow microcytometry (FMC)analysis with the ME1 mAb (IgG1, specific for HLA-B27, -B7,and -B22).73 The determinant recognized by ME1 is notaffected by bound peptides (data not shown). Binding of a givenligand was measured as its C50. This is its molar concentrationat 50% of the fluorescence obtained with that ligand at 10-4

M. Ligands with C50 e 5 µM were considered to bind with highaffinity, as these were the values obtained for most of thenatural B27-bound peptides. C50 values between 5 and 50 µMwere considered to reflect intermediate affinity. C50 g 50 µMindicated low affinity.

Acknowledgment. This work is supported by theSchweizerischer Nationalfonds zur Forderung der Wis-senschaftlichen Forschung (Project No. 31-45504.95)and by Grant SAF97/0182 from the Spanish PlanNacional de I+D to J.A.L.C. D.S. thanks NovartisPharma (Basel) for continuous financial support to hisgroup and A. K. Beck, J. Schreiber, and S. Sigrist forprocessing the manuscript. D.R. thanks the calculationcenter of the ETH-Zurich for allocation of computer timeon the CRAY J90 and PARAGON supercomputers.

References(1) Heemels, M. T.; Ploegh, H. L. Generation, translocation and

presentation of MHC class I-restricted peptides. Annu. Rev.Biochem. 1995, 64, 643-691.

(2) Batalia, M. A.; Collins, E. J. Peptide binding by class I and classII MHC molecules. Biopolymers 1997, 43, 281-302.

(3) Falk, K.; Rotzschke, O.; Stevanovic, S.; Jung, G.; Rammensee,H.-G. Allele-specific motifs revealed by sequencing of self-peptides eluted from MHC molecules. Nature 1991, 351, 290-296.

(4) Rammensee, H. G.; Friede, T.; Stevanovic, S. MHC ligands andpeptide motifs: first listing. Immunogenetics 1995, 41, 178-228.

(5) Saper, M. A.; Bjorkman, P. J.; Wiley, D. C. Refined structure ofthe human histocompatibility antigen HLA-A2 at 2.6 Å resolu-tion. J. Mol. Biol. 1991, 219, 277-319.

(6) Madden, D. R.; Garboczi, D. N.; Wiley, D. C. The antigenicidentity of peptide/MHC complexes: a comparison of the con-formations of five viral peptides presented by HLA-A2. Cell 1993,75, 693-708.

(7) Guo, H. C.; Jardetzky, T. S.; Garrett, T. P. J.; Lane, W. S.;Strominger, J. L.; Wiley, D. C. Different length peptides bind toHLA-Aw68 similarly at their ends but bulge out in the middle.Nature 1992, 360, 364-366.

(8) Stryhn, A.; Andersen, P. S.; Pedersen, L. O.; Svejgaard, A.; Holm,A.; Thorpe, C. J.; Fugger, L.; Buus, S.; Engberg, J. Shared finespecificity between T-cell receptors and an antibody recognizinga peptide/major histocompatibility class I complex. Proc. Natl.Acad. Sci. U.S.A. 1996, 93, 10338-10342.

(9) Garcia, K. C.; Degano, M.; Stanfield, R. L.; Brunmark, A.;Jackson, M. R.; Peterson, P. A.; Teyton, L.; Wilson, I. A. An RT cell receptor structure at 2.5 Å and its orientation in the TCR-MHC complex. Science 1996, 274, 209-219.

(10) Garboczi, D. N.; Ghosh, P.; Utz, U.; Fan, Q. R.; Biddison, W. E.;Wiley, D. C. Structure of the complex between human T-cellreceptor, viral peptide and HLA-A2. Nature 1996, 384, 134-141.

(11) Garcia, K. C.; Degano, M.; Pease, L. R.; Huang, M.; Peterson,P. A.; Teyton, L.; Wilson, I. A. Structural basis of plasticity in Tcell receptor recognition of a self-peptide-MHC antigen. Science1998, 279, 1666-1672.

(12) Ding, Y. H.; Smith, K. J.; Garboczi, D. N.; Utz, U.; Biddison, W.E.; Wiley, D. C. Two human T cell receptors binding in a similardiagonal mode to the HLA-A2/Tax peptide complex using dif-ferent TCR amino acids. Immunity 1998, 8, 403-411.

(13) Adorini, L.; Muller, S.; Cardinaux, F.; Lehmann, P. V.; Falcioni,F.; Nagy, Z. A. In vivo competition between self-peptides. Nature1988, 334, 623-625.

(14) Townsend, A.; Ohlen, C.; Bastin, J.; Ljunggren, H. G.; Foster,L.; Karre, K. Association of class I major histocompatibilityheavy and light chains induced by viral peptides. Nature 1989,340, 443-448.

(15) Collins, D. S.; Findlay, K.; Harding, C. V. J. Processing ofexogenous liposome-encapsulated antigens in vivo generatesclass I MHC-restricted T cell responses. J. Immunol. 1992, 148,3336-3341.

(16) Deres, K.; Schild, H.; Wiesmuller, K. H.; Jung, G.; Ramensee,H. G. In vivo priming of virus-specific cytotoxic T lymphocyteswith synthetic lipopeptide vaccine. Nature 1989, 342, 561-563.

(17) De Magistris, M. T.; Alexander, J.; Coggeshall, M.; Altman, A.;Gaeta, F. C. A.; Grey, H. M.; Sette, A. Antigen analog-majorhistocompatibility complexes act as antagonists of the T cellreceptor. Cell 1992, 68, 625-634.

(18) Valitutti, S.; Muller, S.; Cella, M.; Padovan, E.; Lanzavecchia,A. Serial triggering of many T-cell receptors by a few peptide-MHC complexes. Nature 1995, 375, 149-151

(19) Ishioka, G. Y.; Adorini, L.; Guery, J.-C.; Gaeta, F. C. A.; LaFond,R.; Alexander, J.; Powell, M. F.; Sette, A.; Grey, H. M. Failureto demonstrate long-lived MHC saturation both in vitro and invivo. J. Immunol. 1994, 152, 4311-4319.

(20) Alam, S. M.; Travers, P. J.; Wung, J. L.; Nasholds, W.; Redpath,S.; Jameson, S. C.; Gascoigne, N. R. J. T-cell receptor affinityand thymocyte positive selection. Nature 1996, 381, 616-620.

(21) Lyons, D. S.; Lieberman, S. A.; Hampl, J.; Boniface, J. J.; Chien,Y.-H.; Berd, J.; Davis, M. M. A TCR binds to antagonist ligandswith lower affinities and faster dissociation rates than toagonists. Immunity 1996, 5, 53-61.

(22) Collins, E. J.; Frelinger, J. A. Altered peptide ligand design:altering immune responses to class I MHC/peptide complexes.Immunol. Rev. 1998, 163, 151-160.

(23) Krebs, S.; Rognan, D. From peptides to peptidomimetics: Designof nonpeptide ligands for Major Histocompatibility Proteins.Pharm. Helv. Acta 1998, 73, 173-181.

(24) Rognan, D.; Scapozza, L.; Folkers, G.; Daser, A. Rational designof nonnatural peptides as high-affinity ligands for the HLA-B*2705 human leukocyte antigen. Proc. Natl. Acad. Sci. U.S.A.1995, 92, 753-757.

(25) Rovero, P.; Vigano, S.; Pegorado, S.; Revoltella, R.; Riganelli,D.; Fruci, D.; Greco, G.; Butler, R. H.; Tanigaki, N. Augmentationof the affinity of HLA class I-binding peptides lacking primaryanchor residues by manipulation of the secondary anchorresidues. J. Pept. Sci. 1995, 1, 266-273.

(26) Raghavan, M.; Lebron, J. A.; Johnson, J. L.; Bjorkman, P. A.Extended repertoire of permissible peptide ligands for HLA-B*2702. Protein Sci. 1996, 5, 2080-2088.

(27) Weiss, G. A.; Valentekovich, R. J.; Collins, E. J.; Garboczi, D.N.; Lane, W. S.; Schreiber, S. L.; Wiley, D. C. Covalent HLA-B27/peptide complex induced by specific recognition of anaziridine mimic of arginine. Proc. Natl. Acad. Sci. U.S.A. 1996,93, 10945-10948.

(28) Krebs, S.; Folkers, G.; Rognan, D. Binding of rationally designednonnatural peptides to the human leukocyte antigen HLA-B*2705. J. Pept. Sci. 1998, 4, 378-388.

Nonapeptide Analogue Binding Affinity to MHC Protein Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 2329

(29) Bianco, A.; Zabel, C.; Walden, P.; Jung, G. N-hydroxy-amideanalogues of MHC class I peptide ligands with nanomolarbinding affinities. J. Pept. Sci. 1998, 4, 471-478.

(30) Guichard, G.; Calbo, S.; Muller, S.; Kourilsky, P.; Briand, J.-P.;Abastado, J.-P. Efficient binding of reduced peptide bondpseudopeptides to major histocompatibility complex class Imolecule. J. Biol. Chem. 1995, 270, 26057-26059.

(31) Guichard, G.; Connan, F.; Graff, R.; Ostankovitch, M.; Muller,S.; Guillet, J.-G.; Choppin F.; Briand, J.-P. Partially modifiedretro-inverso pseudopeptides as nonnatural ligands for the classI histocompatibility molecule HLA-A2. J. Med. Chem. 1996, 39,2030-2039.

(32) Bianco, A.; Brock, C.; Zabel, C.; Walk, T.; Walden, P.; Jung, G.New synthetic non-peptide ligands for classical major histocom-patibility complex class I molecules. J. Biol. Chem. 1998, 273,28759-28765.

(33) Rognan, D.; Krebs, S.; Kuonen, O.; Lamas, J. R.; Lopez de Castro,J. A.; Folkers, G. Fine specificity of antigen binding to two classI major histocompatibility proteins (B*2705 and B*2703) differ-ing in a single amino acid residue. J. Comput.-Aided Mol. Des.1997, 11, 463-478.

(34) Krebs, S.; Lamas, J. R.; Poenaru, S.; Folkers, G.; Lopez de Castro,J. A.; Seebach, D.; Rognan, D. Substituting organic spacers forthe T-cell receptor binding part of class I MHC-restrictedpeptides. J. Biol. Chem. 1998, 273, 19072-19079.

(35) Weiss, G. A.; Collins, E. J.; Garboczi, D. N.; Wiley, D. C.;Schreiber, S. L. A tricyclic ring system replaces the variableregions of peptides presented by three alleles of human MHCclass I molecules. Chem. Biol. 1995, 2, 401-407.

(36) Bouvier, M.; Wiley, D. C. Antigenic peptides containing largePEG loops designed to extend out of the HLA-A2 binding siteform stable complexes with class I major histocompatibilitycomplex molecules. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 4583-4588.

(37) Seebach, D.; Poenaru, S.; Folkers, G.; Rognan, D. Synthesis ofOligo(3-hydroxybutanoate) (OHB) containing peptides with highbinding affinity to a class I MHC protein. Helv. Chim. Acta 1998,81, 1191-1200.

(38) Seebach, D.; Overhand, M.; Kuhnle, F. N. M.; Martinoni, B.;Oberer, L.; Hommel, U.; Widmer, H. -Peptides: Synthesis byArndt-Eistert homologation with concomitant peptide coupling.Structure determination by NMR and CD spectroscopy and byX-ray crystallography. Helical secondary structure of a -hexapep-tide in solution and its stability towards pepsin. Helv. Chim.Acta 1996, 79, 913-941.

(39) Hintermann, T.; Seebach, D. The Biological Stability of -Pep-tides: No Interactions between R- and -Peptidic Structures?Chimia 1997, 51, 244-247.

(40) Matthews, J. L.; Overhand, M.; Kuhnle, F. N. M.; Ciceri, P. E.;Seebach, D. -Peptides: Oligo--homoalanines - The amideanalogue of poly(3-hydroxybutanoate). Liebigs Ann. Chem. 1997,1371-1379.

(41) Guichard, G.; Abele, S.; Seebach, D. Preparation of N-Fmoc-protected -2- and -3-amino acids and their use as buildingblocks for the solid-phase synthesis of -peptides. Helv. Chim.Acta 1998, 81, 187-206.

(42) Holmes, P. A. In Developments in crystalline polymers-2; Bassett,D. C., Ed.; 1988.

(43) Lopez de Castro, J. A. The pathogenic role of HLA-B27 in chronicarthritis. Curr. Opin. Immunol. 1998, 10, 59-66.

(44) Podlech, J.; Seebach, D. On the preparation of -amino acidsfrom R-amino acids using the Arndt-Eistert reaction: Scope,limitations and stereoselectivity. Application to carbohydratepeptidation. Stereoselective R-alkylations of some -amino acids.Liebigs Ann. Chem. 1995, 1217-1229.

(45) Lengweiler, U. D.; Fritz, M. G.; Seebach, D. Monodisperse linearand cyclic oligo[(R)-3-hydroxybutanoates] containing up to 128monomeric units. Helv. Chim. Acta 1996, 79, 670-701.

(46) Neise, B.; Stelich, W. Einfaches Verfahren von Carbonsauren.Angew. Chem. 1978, 90, 556-557; Angew. Chem., Int. Ed. Engl.1978, 17, 522.

(47) Sheehan, J. A.; Cruickshank, P. A.; Boshart, J. L. A ConvenientSynthesis of Water-Soluble Carbodiimides. J. Org. Chem. 1961,26, 2525-2529.

(48) Konig, W.; Geiger, R. Eine neue Methode zur Synthese vonPeptiden: Aktivierung der Carboxylgruppe mit Dicyclohexyl-carbodiimid unter Zusatz von 1-Hydroxy-benzotriazolen. Chem.Ber. 1970, 103, 788-798.

(49) Konig, W.; Geiger, R. N-hydroxyverbindungen als Katalysatorenfur die Aminolyse aktivierter Ester. Chem. Ber. 1973, 106, 3626-3635.

(50) Compound 10 could also be obtained using an alternativesynthetic route: coupling of 4 to 5 to give the tetramer Boc-(-HAla-HB)2-OBn. Cleavage of the benzyl ester group in order tocouple the lysine and subsequent Boc deprotection allow couplingof the N-terminal alanine to give the desired compound.

(51) Plattner, D. A.; Brunner A.; Dobler, M.; Muller, H.-M.; Petter,W.; Zbinden, P.; Seebach, D. Cyclic oligomers of (R)-3-hydrox-ybutanoic acid: Preparation and structural aspects. Helv. Chim.Acta 1993, 76, 2004-2033.

(52) Bodansky, M.; Deshmane, S. S.; Martinez, J. Side reactions inpeptide synthesis. 11. Possible removal of the 9-fluorenylmethyl-oxycarbonyl group by the amino components during coupling.J. Org. Chem. 1979, 44, 1622-1625.

(53) Mitchell, A. R.; Merrifield, R. B. Occurrence of N-alkylationduring the acidolytic cleavage of urethane protecting groups. J.Org. Chem. 1976, 41, 2015-2019.

(54) The use of a nitro group for protecting arginine during thesynthesis of short peptides, using N-Boc-protected amino acids,seems to be the best choice, as this protecting group is stableunder acidic conditions (such as TFA) and can be cleaved bycatalytic hydrogenation.

(55) Doulut, S.; Lugrin, D.; Vecchini, F.; Aumelas, A.; Martinez, J.Reduced peptide bond pseudopeptide analogues of neurotensin.Pept. Res. 1992, 5, 30-38.

(56) Madden, D. R.; Gorga, J. C.; Strominger, J. L.; Wiley, D. C. Thethree-dimensional structure of HLA-B27 at 2.1 Å resolutionsuggests a general mechanism for tight peptide binding to MHC.Cell 1992, 70, 1035-1048.

(57) Rognan, D.; Scapozza, L.; Folkers, G.; Daser, A. Moleculardynamics simulation of MHC-peptide complexes as a tool forpredicting potential T cell epitopes. Biochemistry 1994, 33,11476-11485.

(58) Bouvier, M.; Wiley, D. C. Importance of peptide amino andcarboxy termini to the stability of MHC class I molecules. Science1994, 265, 398-402.

(59) Seebach, D.; Ciceri, P. E.; Overland, P. M.; Jaun, B.; Rigo, D.;Oberer, L.; Hommel, U.; Amstutz, R.; Widmer, H. Probing thehelical secondary structure of short-chain -peptides. Helv.Chim. Acta 1996, 79, 2043-2066.

(60) Schmieden, V.; Betz, H. Pharmacology of the inhibitory glycinereceptor: Agonist and Antagonist actions of amino acids andpiperidine carboxylic acid compounds. Mol. Pharmacol. 1995,48, 919-927.

(61) (a) Zablocki, J. A.; Tjoeng, F. S.; Bovy, P. R.; Miyano, M.;Garland, R. B.; Williams, K.; Schretzman, L.; Zupec, M. E.; Rico,J. G.; Lindmark, R. J.; Toth, M. V.; McMackins, D. E.; Adams,S. P.; Panzer-Knodle, S. G.; Nicholson, N. S.; Taite, B. B.;Salyers, A. K.; King, L. W.; Campion, J. G.; Feigen, L. P. A novelseries of orally active antiplatelet agents. Bioorg. Med. Chem.1995, 3, 539-551. (b) Hutchinson, J. H.; Cook, J. J.; Brashear,K. M.; Breslin, M. J.; Glass, J. D.; Gould, R. J.; Halczenko, W.;Holahan, M. A.; Lynch, R. J.; Sitko, G. R.; Stranieri, M. T.;Hartman, G. D. Non-peptide glycoprotein IIb/IIIa antagonists.11. Design and in vivo evaluation of 3,4-dihydro-1(1H)-isoqui-nolone-based antagonists and ethyl ester prodrugs. J. Med.Chem. 1996, 39, 4583-4591. (c) Kottirsch, G.; Zerwes, H.-G.;Cook, N. S.; Tapparelli, C. Beta-amino acid derivative of orallyactive non-peptide fibrinogen receptor antagonists. Bioorg. Med.Chem. Lett. 1997, 6, 727-732. (d) Hayashi, Y.; Katada, J.;Harada, T.; Tachiki, A.; Iijima, K.; Takiguchi, Y.; Muramatsu,M.; Miyazaki, H.; Asari, T.; Okazaki, T.; Sato, Y.; Yasuda, E.;Yano, M.; Uno, I.; Ojima, I. GIIb/IIIa integrin antagonists withthe new conformational restriction unit, trisubstituted -aminoacid derivatives, and a substituted benzamidine structure. J.Med. Chem. 1998, 41, 2345-2360.

(62) Bateson, J. H.; Gasson, B. C.; Khushi, T.; Neale, J. E.; Payne,D. J.; Tolson, D. A.; Walker, G. The synthesis and serine-lactamase inhibitory activity of some phosphonamidate ana-logues of dipeptides. Bioorg. Med. Chem. Lett. 1994, 4, 1667-1672.

(63) Yamazaki, T.; Probstl, A.; Schiller, P. W.; Goodman, M. Biologicaland conformational studies of [Val4]morphiceptin and [D-Val4]-morphiceptin analogues incorporating cis-2-aminocyclopentanecarboxylic acid as a peptidomimetic for proline. Int. J. Pept.Protein Res. 1991, 37, 364-381.

(64) Hernandez, J. F.; Soleilhac, J. M.; Roques, B. P.; Fournie-Zaluski,M. C. Retro-inverso concept applied to the complete inhibitorsof enkephalin-degrading enzymes. J. Med. Chem. 1988, 31,1825-1831.

(65) (a) Winkelman, G.; Allgaier, H.; Lupp, R.; Jung, G. Iturin A (1)- A new long chain iturin A possessing an unusual high contentof C16-beta-amino acids. J. Antibiot. 1983, 36, 1451-1457. (b)Kajiyama, S. I.; Kanzaki, H.; Kawazu, K.; Kobayashi, A. Nosto-fungicidine, an antifungal lipopeptide from the field-grownterrestrial blue-green alga Nostoc commune. Tetrahedron Lett.1998, 39, 3737-3740. (c) Sone, H.; Nemoto, T.; Ishiwata, H.;Ojika, M.; Yamada, K. Isolation, structure and synthesis ofDolastin, D. A cytotoxic cyclic depsipeptide from the sea hareDolabella auriculum. Tetrahedron Lett. 1993, 34, 8449-8452.

2330 Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 Poenaru et al.

(66) Case, D. A.; Pearlman, D. A.; Caldwell, J. W.; Cheatham, T. E.,III; Ross, W. S.; Simmerling, C. L.; Darden, T. A.; Merz, K. M.;Stanton, R. V.; Cheng, A. L.; Vincent, J. J.; Crowley, M.;Ferguson, D. M.; Radmer, R. J.; Seibel, G. L.; Singh, U. C.;Weiner, P. K.; Kollman, P. A. AMBER5.0; University of Cali-fornia: San Francisco, CA, 1997.

(67) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K.M., Jr.; Ferguson, D. M.; Spellmeyer, D. M.; Fox, T.; Caldwell,J. W.; Kollman, P. A. A second generation force field for thesimulation of proteins, nucleic acids and organic molecules. J.Am. Chem. Soc. 1995, 117, 5179-5197.

(68) Fox, T.; Scanlan, T. S.; Kollman, P. A. Ligand binding in thecatalytic antibody 17E8. A free energy perturbation study. J.Am. Chem. Soc. 1997, 119, 11571-11577.

(69) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T. A.;Petersson, G. A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski, V. G.; Ortiz, J. V.; Foresman, J. B.;Peng, C. Y.; Ayala, P. A.; Wong, M. W.; Andres, J. L.; Replogle,E. S.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Binkley, J. S.;Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-Gordon, M.;Gonzalez, C.; Pople, J. A. Gaussian 94, revison C.3; GaussianInc.: Pittsburgh, PA, 1995.

(70) Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A. A well-behaved electrostatic potential based method using chargerestraints for determining atom-centered charges: The RESPmodel. J. Phys. Chem. 1993, 97, 10269-10280.

(71) Galocha, B.; Lamas, J. R.; Villadangos, J. A.; Albar, J. P.; Lopezde Castro, J. A. Binding of peptides naturally presented by HLA-B27 to the differentially disease-associated B*2704 and B*2706subtypes, and to mutants mimicking their polymorphism. TissueAntigens 1996, 48, 509-518.

(72) Ljunggren, H. G.; Stam, N. J.; Ohlen, C.; Neefjes, J. J.; Hoglund,P.; Heemels, M. T.; Bastin, J.; Schumacher, T. N.; Townsend,A.; Karre, K.; Ploegh, H. L. Empty class I molecules comes outin the cold. Nature 1990, 46, 476-480.

(73) Ellis, S. A.; Taylor, C.; McMichael, A. Recognition of HLA-B27and related antigens by a monoclonal antibody. Hum. Immunol.1985, 5, 49-59.

(74) Connolly, M. J. Analytical molecular surface calculation. J. Appl.Crystallogr. 1983, 16, 548-558.

(75) Kraulis, P. J. MOLSCRIPT: a program to produce both detailedand schematic plots of protein structures. J. Appl. Crystallogr.1991, 24, 946-950.

(76) Merritt, E. A.; Murphy, M. E. P. Raster3D Version 2.0. Aprogram for photorealistic molecular graphics. Acta Crystallogr.1994, D50, 869-873.

JM981123L

Nonapeptide Analogue Binding Affinity to MHC Protein Journal of Medicinal Chemistry, 1999, Vol. 42, No. 13 2331

ANEXO -8-