Dynamics of the �2-Adrenergic G-Protein CoupledReceptor Revealed by Hydrogen-DeuteriumExchange

Xi Zhang,† Ellen Y. T. Chien,‡ Michael J. Chalmers,† Bruce D. Pascal,† Jovylyn Gatchalian,‡

Raymond C. Stevens,‡ and Patrick R. Griffin*,†

Department of Molecular Therapeutics, The Scripps Research Institute, Jupiter, Florida 33458, and Department ofMolecular Biology, The Scripps Research Institute, La Jolla, California 92037

To examine the molecular details of ligand activation ofG-protein coupled receptors (GPCRs), emphasis has beenplaced on structure determination of these receptors withstabilizing ligands. Here we present the methodology forreceptor dynamics characterization of the GPCR human�2 adrenergic receptor bound to the inverse agonistcarazolol using the technique of amide hydrogen/deuterium exchange coupled with mass spectrometry(HDX MS). The HDX MS profile of receptor bound tocarazolol is consistent with thermal parameter obser-vations in the crystal structure and provides additionalinformation in highly dynamic regions of the receptorand chemical modifications demonstrating the highlycomplementary nature of the techniques. After opti-mization of HDX experimental conditions for thismembrane protein, better than 89% sequence coveragewas obtained for the receptor. The methodology pre-sented paves the way for future analysis of �2AR boundto pharmacologically distinct ligands as well as analysisof other GPCR family members.

G-protein coupled receptors (GPCRs) are molecular targetsof many therapeutic drugs used in the treatment of a wide rangeof disorders and diseases.1-3 The prototypic GPCR �2-adrenergicreceptor (�2AR) plays a critical role in converting externalstimuli into intracellular signals that regulate many cellularprocesses. �2AR is expressed predominantly in smooth muscle,skeletal muscle, and liver and has been shown to be involvedin regulation of muscle relaxation, glycogenolysis, gluconeo-genesis, and potassium uptake.4 Drugs that modulate �2AR

have been developed to treat diseases such as asthma.1,5 Typicalof members of the GPCR superfamily, human �2AR contains 413amino acids composed of seven transmembrane helices, anextracellular N-terminus, and an intracellular C-terminus.1–3 Thereceptor is activated upon binding of endogenous (catechola-mines) or synthetic ligands that modulate receptor-G-proteininteraction. Specifically, in the resting state, �2AR associates withG-protein via its intracellular third loop and C-terminus andupon ligand binding, conformational changes likely facilitatethe release of G-protein subunits that initiate a subsequentsignaling cascade.4 Understanding the structure and confor-mational dynamics of �2AR is essential to understanding themechanism of action of natural and synthetic ligands. Thisstructural information will aid in the design and developmentof functionally selective GPCR modulators.

Recent successes in high-resolution X-ray crystallography haverevealed the structures of human �2AR in the inverse agonistbound form.6-8 These high-resolution structures were obtainedfor the ligand bound form with the insertion of the T4-lysozymedomain replacing a highly dynamic third intracellular loop.6–8

While these inactive structures provided significant insight intothe structure of GPCRs as well as the molecular details of ligandinteraction, questions remain about conformational dynamics,active state structure, and the role of dynamic regions of thereceptor. For instance, the N- and C-termini of the receptor, whichare subjected to extensive post-translational modifications and areexpected to play critical roles in the interaction with signalingmolecules, remain largely unresolved by crystallography.6–8 Mostproteins are globular in nature and are dynamic molecules thatundergo constant flexing and motion. Changes in protein confor-mation and dynamics are important to molecular recognition andthese changes occur during many key biological processes suchas ligand binding to receptors. Thus, techniques to study proteinstructure and dynamics are critical to the understanding ofbiological processes.

* To whom correspondence should be addressed. Patrick R. Griffin, Ph.D.,The Scripps Research Institute, Scripps Florida, 130 Scripps Way No. 2A2, Jupiter,FL 33458. Phone: (561) 228-2200. Fax: (561) 228-3081. E-mail: pgriffin@scripps.edu.

† The Scripps Research Institute, Jupiter, Florida.‡ The Scripps Research Institute, La Jolla, California.

(1) Hanson, M. A.; Stevens, R. C. Structure (Cambridge, MA, U. S.) 2009, 17,8–14.

(2) Kroeze, W. K.; Sheffler, D., J.; Roth, B. L. J. Cell Sci. 2003, 116, 4867–4869.

(3) Conn, P. J.; Christopoulos, A.; Lindsley, C. W. Nat. Rev. Drug Discovery2009, 8, 41–54.

(4) Golan, D. E., Tashjian, A. H., Jr., Armstrong, E. J., Armstrong, A. W.Principles of Pharmacology: The Pathophysiologic Basis of Drug Therapy, 2nded.; Lippincott Williams & Wilkins: Baltimore, MD, 2008.

(5) Penn, R. B. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 2095–2096.(6) Rosenbaum, D. M.; Cherezov, V.; Hanson, M. A.; Rasmussen, S. G.; Thian,

F. S.; Kobilka, T. S.; Choi, H. J.; Yao, X. J.; Weis, W. I.; Stevens, R. C.;Kobilka, B. K. Science 2007, 318, 1266–1273.

(7) Cherezov, V.; Rosenbaum, D. M.; Hanson, M. A.; Rasmussen, S. G.; Thian,F. S.; Kobilka, T. S.; Choi, H. J.; Kuhn, P.; Weis, W. I.; Kobilka, B. K.;Stevens, R. C. Science 2007, 318, 1258–1265.

(8) Hanson, M. A.; Cherezov, V.; Griffith, M. T.; Roth, C. B.; Jaakola, V.-P.;Chien, E. Y. T.; Velasquez, J.; Kuhn, P.; Stevens, R. C. Structure (Cambridge,MA, U. S.) 2008, 16, 897–905.

Anal. Chem. 2010, 82, 1100–1108

10.1021/ac902484p 2010 American Chemical Society1100 Analytical Chemistry, Vol. 82, No. 3, February 1, 2010Published on Web 01/08/2010

Amide hydrogen/deuterium exchange (HDX) can be used toprobe protein dynamics and is an ideal complement to otherstructural techniques such as X-ray crystallography.9-11 HDX hasincreasingly demonstrated its exceptional capability to study thefine dynamic structural changes of large complicated proteinsystems.12,13 The exchange rate of an amide hydrogen is influ-enced significantly by hydrogen bonding and exchange kineticsof an amide hydrogen can be highly reflective of its locations insecondary and tertiary structures.10,14 The methods of choice formeasuring solution phase HDX are nuclear magnetic resonance(NMR) and mass spectrometry.15,16 While NMR offers singleamide resolution, the technique requires large amounts of proteinand the analysis can be limited to relatively small proteins. Thecoupling of HDX with mass spectrometry affords high sensitivityrequiring a very low amount of protein with virtually no limit tosize or complexity of the protein and is tolerant of impurities.However, to date, HDX studies have primarily focused on solubleproteins and reports on membrane proteins are very limited17,18

with no published comprehensive HDX studies of a GPCR. Thelack of HDX reports on membrane proteins is likely due to thetechnical challenges of analysis of membrane proteins.

HDX MS involves proteolytic digestion of the target proteinwith acid stable proteases under so-called “slow exchange”conditions (low pH and low temperature) to minimize loss ofdeuterium during digestion and subsequent mass analysis. Un-fortunately, like all GPCRs, �2AR contains seven highly hydro-phobic transmembrane helices, is very sensitive to pH, and isprone to aggregation in detergent solubilized solution.19 Thesestructural features limit accessibility of protease leading to lowdigestion efficiency in aqueous solution conditions optimizedfor HDX studies. Further limiting digestion efficiency underHDX conditions are short digestion times and use of im-mobilized protease columns. Purification procedures for isolat-ing GPCRs require the presence of detergent to maintain nativestructure, solubility, and functional activity,20 and purifiedGPCRsoftenhaveboundphospholipids,sterols,anddetergents.6–8

Problematic for HDX MS, many commonly used detergents caninterfere with mass spectrometry analysis. Because of their highlypolar or charged head groups, detergents can cause ion suppres-sion during the electrospray ionization (ESI) or matrix-assistedlaser desorption/ionization (MALDI) process, although nonionic

detergents have been shown to be the least detrimental.21 Thenonionic detergent n-dodecyl-�-D-maltoside (dodecyl maltoside,DDM) has been found to be applicable to most membraneproteins, not only in efficient solubilization and maintaining activitybut also in crystallization.20,22-24 To further complicate HDX MSanalysis GPCRs such as �2AR are potentially heavily post-translational modified with glycosylation, cysteine palmitoyla-tion, phosphorylation, disulfide bonds, and ubiquitination.8,25-27

Through careful optimization of a wide range of experimentalvariables, significantly improved digestion efficiency can beobtained under HDX compatible conditions resulting in excellentsequence coverage. These optimized conditions enabled compre-hensive HDX analysis of �2AR bound to the inverse agonistcarazolol. This study provides insights into structural dynamicsof �2AR and paves the way for comparative HDX studies of�2AR bound with various functional ligands as well as HDXstudies of other GPCRs.

EXPERIMENTAL SECTIONReagents and Chemicals. HPLC grade H2O, D2O (99.9%),

acetonitrile, formic acid, iso-propanol, tris-(carboxyethyl)-phos-phine hydrochloride (TCEP), iodoacetamide, Tris, NaCl,NaH2PO4, glycerol and anti-FLAG antibody, and horse heartmyoglobin were purchased from Sigma-Aldrich (St. Louis,MO). Trifluoroacetic acid (TFA, Sequanal grade) was obtainedfrom Pierce (Rockford, IL). High purity n-dodecyl-�-D-maltoside(DDM) and cholesterol hemisuccinate (CHS) were purchasedfrom Anatrace Inc. (Maumee, OH). PNGase F at 500 000 unit/mL was purchased from New England BioLabs Inc. (Ipswich,MA). The preprepared 10% Bis-Tris gels used in sodiumdodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE)for Coomassie stain and Western blot were purchased fromInvitrogen (Carlsbad, CA). The porcine pepsin-immobilizedPOROS 20 AL beads (particle size 20 µm) used to packimmobilized pepsin columns were purchased from AppliedBiosystems (Foster City, CA).

Expression and Purification of �2AR. Human �2AR con-taining an N-terminal FLAG tag and a C-terminal 10xHis tagwith E122W and N187E mutations and shortened 3IL andC-terminus (amino acid nos. 245-259 and 349-413 weredeleted) was expressed in Sf9 insect cells and purified by usingimmobilized metal affinity chromatography (IMAC) with 50 µMcarazolol present. The protein was treated with 1 mg/mLiodoacetamide prior to solubilization to block the solvent-exposed free cysteine residues.8 The final purified protein(typically 40-50 µM) with carazolol-bound and was kept in thebuffer containing 50 mM Hepes (pH 7.5), 150 mM NaCl, 10%(v/v) glycerol, 0.05% (m/v) DDM, 0.01% (m/v) CHS and 50

(9) Englander, S. W. J. Am. Soc. Mass Spectrom. 2006, 17, 1481–1489.(10) Katta, V.; Chait, B. T.; Carr, S. Rapid Commun. Mass Spectrom. 1991, 5,

214–217.(11) Yan, X.; Zhang, H.; Watson, J.; Schimerlik, M. I.; Deinzer, M. L. Protein

Sci. 2002, 11, 2113–2124.(12) Bruning, J. B.; Chalmers, M. J.; Prasad, S.; Busby, S. A.; Kamenecka, T. M.;

He, Y.; Nettles, K. W.; Griffin, P. R. Structure (Cambridge, MA, U. S.) 2007,15, 1258–1271.

(13) Chalmers, M. J.; Busby, S. A.; Pascal, B. D.; He, Y.; Hendrickson, C. L.;Marshall, A. G.; Griffin, P. R. Anal. Chem. 2006, 78, 1005–1014.

(14) Cravello, L.; Lascoux, D.; Forest, E. Rapid Commun. Mass Spectrom. 2003,17, 2387–2393.

(15) Shimada, I. In Nuclear Magnetic Resonance of Biological Macromolecules,Part C; Academic Press: San Diego, CA, 2005; Vol. 394, p 483.

(16) Engen, J. R.; Smith, D. L. Anal. Chem. 2001, 73, 256A–265A.(17) Busenlehner, L. S.; Salomonsson, L.; Brzezinski, P.; Armstrong, R. N. Proc.

Natl. Acad. Sci. U.S.A. 2006, 103, 15398–15403.(18) Busenlehner, L. S.; Branden, G.; Namslauer, I.; Brzezinski, P.; Armstrong,

R. N. Biochemistry 2008, 47, 73–83.(19) White, S. H.; Ladokhin, A. S.; Jayasinghe, S.; Hristova, K. J. Biol. Chem.

2001, 276, 32395–32398.(20) Garavito, R. M.; Ferguson-Miller, S. J. Biol. Chem. 2001, 276, 32403–32406.

(21) Cristoni, S.; Bernardi, L. R. Mass Spectrom. Rev. 2003, 22, 369–406.(22) Thompson, D. A.; Suarez-Villafane, M.; Ferguson-Miller, S. Biophys. J. 1982,

37, 285–293.(23) le Maire, M.; Champeil, P.; Møller, J. V. Biochim. Biophys. Acta, Biomembr.

2000, 1508, 86–111.(24) Qin, L.; Hiser, C.; Mulichak, A.; Garavito, R. M.; Ferguson-Miller, S. Proc.

Natl. Acad. Sci. U.S.A. 2006, 103, 16117–16122.(25) Marchese, A.; Benovic, J. L. In Methods in Molecular Biology; Humana Press:

Totowa, NJ, 2004; Vol. 259, pp 299-305.(26) Trester-Zedlitz, M.; Burlingame, A.; Kobilka, B.; Von Zastrow, M. Biochem-

istry 2005, 44, 6133–6143.(27) Li, J.; Edwards, P. C.; Burghammer, M.; Villa, C.; Schertler, G. F. X. J. Mol.

Biol. 2004, 343, 1409–1438.

1101Analytical Chemistry, Vol. 82, No. 3, February 1, 2010

µM carazolol. The protein concentration was determined withthe bicinchroninic acid (BCA) protein assay kit (Pierce,Rockford, IL) using bovine serum albumin as a reference.Purity of the protein sample was followed using SDS-PAGEand analytical size exclusion chromatography. All sequencenumbers cited in the current study referred to the �2AR wildtype sequence number shown in Figure S1 in the SupportingInformation.

Sequence Coverage of �2AR. Unless specified otherwise, thepurified protein was diluted to 15 µM for mass spectrometryexperiments using the buffer composed of 50 mM Hepes (pH 7.5),150 mM NaCl, 2% (v/v) glycerol, 0.02% (m/v) DDM (∼3.3 criticalmicelle concentration (cmc)), 0.01% (m/v) CHS, and H2O. Thebuffer for HDX on-exchange studies was the same compositionexcept H2O was replaced with D2O (99.9%). A total of 4 µL of�2AR protein was mixed by an automated system (describedelsewhere13) with 16 µL of the H2O buffer and then 30 µL ofquench solution containing a number of different compositions(all at final pH of 2.2-2.5 prior to pepsin digestion) and digestedeither online by passing through an immobilized pepsin-coupledcolumn (2 mm i.d. × 20 mm) or offline by incubation withpepsin-coupled bead slurry under room temperature (27.8 ±0.5 °C at the bench). The digest was then separated by a Agilent(Santa Clara, CA) high-performance liquid chromatography(HPLC) equipped with a C8 trap (2 mm i.d., Thermo FisherScientific, San Jose, CA) and C4 column (2 mm i.d. × 50 mm,Thermo Fisher Scientific, San Jose, CA) using a gradient of45 min, which started with 5% B and linearly increased to 100%B over the next 45 min, unless specified otherwise. The mobilephase for the online immobilized pepsin-coupled column is H2Owith 0.1% (v/v) TFA and the flow rate is 50 or 200 µL/min.For HPLC, buffer A is H2O containing 0.3% (v/v) formic acid,buffer B is acetonitrile/water at 4:1 (v/v) containing 0.3% (v/v) formic acid, and the flow rate is 50 or 200 µL/min. The ESIMS and collision-induced dissociation (CID) tandem massspectrometry (MS/MS) spectra of the peptides eluted fromHPLC were acquired in positive ion mode on a quadrupolelinear ion trap mass spectrometer equipped with an ESI source(Thermo Fisher Scientific, San Jose, CA) at capillary temper-ature of 225 °C and spray voltage of 3.5 kV. Mass spectra wereacquired for m/z 300-2000 using a maximum injection timeof 75 ms and target value of 3 × 104. Tandem mass spectrawere obtained using data-dependent acquisition with dynamicexclusion where the top five most abundant ions in each scanwere selected and subjected to CID. MS/MS spectra wereacquired under wide band activation at an activation Q valueof 0.25, activation time of 30 ms, normalized collision energyof 35, isolation width of 2.0, target number of 1 × 104, andmaximum injection time of 100 ms unless specified otherwise.Each scan was the average of 3 microscans under normal scanmode in both MS and MS/MS.

All MS/MS spectra were analyzed using the MASCOT pro-gram. Searches were run without protein modifications and withmodifications such as carboxamidemethyl cysteine, methionineoxidation, phosphorylation, and deglycosylation. Results from allsearches against multiple sets of MS/MS data were combined.Peptides included in the peptide set used for HDX had a MASCOTscore of 20 or greater. The MS/MS MASCOT search was also

performed against a decoy (reverse) sequence and ambiguousidentifications were ruled out. The MS/MS spectra of all of thepeptide ions from the MASCOT search were further manuallyinspected and only those verifiable are used in the coverage.

HDX Analysis of �2AR. For HDX MS experiments, 4 µL of�2AR protein was mixed with 16 µL of the D2O buffer (finalD2O content was 80%) and incubated at various time intervalsat 4 °C, before being added to 30 µL of quench solutioncontaining 15 mM TCEP, 0.1 M NaH2PO4 (HCl, pH 2.4), and0.02% (m/v) DDM. Deuterated protein was then digestedonline using the same procedure described above. HPLC(Eksigent, Dublin CA) was performed with a 9.5 min gradientat flow rate of 50 µL/min, and all of the pepsin column, trap,and analytical columns applied to HDX-MS experiments werethe same as in the MS/MS experiments except that the internaldiameters were all decreased to 1 mm. The gradient startedwith 5% B, increased to 15% B over 0.1 min, increased to 50%B within 5.4 min, then quickly ramped to 98% within 0.5 min,held for 1.5 min, and dropped back to 5% within 0.1 min. MSwas acquired in the range of m/z 300-2000 for 8 min in positiveion mode on an orbi-trap mass spectrometer (Thermo FisherScientific, San Jose, CA) equipped with an ESI source operatedat capillary temperature of 225 °C and spray voltage of 3.5 kV.Protein and quench solutions, pepsin column, trap, and HPLCcolumns were all kept at 1.0 °C unless specified otherwise. HDXincubation was performed at 4.0 °C for six different time points:10, 30, 60, 300, 900, and 3600 s. The experiments wereperformed at random order, and between every two injections,blank injections were made to remove any potential proteincarryover from previous runs. Two additional incubations (12± 1 and 15 h) were performed in 92% D2O buffer in sealedcentrifuge tubes also under 4 °C. The t0 data were acquiredusing the H2O buffer. Data for each on-exchange time pointwere obtained in four replicates (four individually preparedsamples, not four injections of the same sample).

All HDX data were normalized to 100% D2O content (80% Don-exchange buffer for the first 6 time points and 92% D on-exchange buffer for the 12 ± 1 and 15 h time points), correctedfor an estimated average deuterium recovery of 70%, andprocessed using the software HD Desktop.28 Percent deuteriumincorporation was plotted versus incubation time in log scale,and the data points were connected through a second ordersmoothing polynomial. All on-exchange plots used in theanalysis of �2AR are shown in Figure S2 in the SupportingInformation. The HDX profile of carazolol-bound �2AR wasmapped to its own amino acid sequence and also to the crystalstructure of human �2AR T4L monomer which is carazolol-bound and starts with W40 (PDB: 2RH1).8

Deglycosylation of �2AR through PNGase F Digestion.The purified �2AR was incubated with PNGase F at 4 or 37 °Cfor 1 h (PNGase F final concentration at 50 unit/µL) prior tomass spectrometry experiments (MS/MS and HDX t0 MS).The effect of PNGase F digestion was monitored by SDS-PAGEand Western blot. HDX was performed on the protein followingPNGase F treatment and separation using fast protein liquidchromatography (FPLC) at 4 °C.

(28) Pascal, B. D.; Chalmers, M. J.; Busby, S. A.; Griffin, P. R. J. Am. Soc. MassSpectrom. 2009, 20, 601–610.

1102 Analytical Chemistry, Vol. 82, No. 3, February 1, 2010

Ta

ble

1.E

xp

eri

me

nta

lV

ari

ab

les

Pro

be

dto

Imp

rove

the

Se

qu

en

ce

Co

vera

ge

of

� 2A

R

[�2A

R]

(µM

)

peps

inon

line/

offli

ne;

dige

stio

nte

mp

urea

_TFA

TC

EP

NaH

2PO

4pH

2.40

HC

l/T

FAD

DM

(m/v

)(%

)

MS,

MS/

MS

µsca

n;m

axim

umin

ject

ion

time

reje

ctlis

t(5

m/z

valu

es)

HPL

Cflo

wra

te(µ

L/m

in)

MA

SCO

T(>

20)

cove

rage

(%)

CC

Clo

opco

vere

d?

24.8

onlin

e;R

Ta

3M

_1%

defa

ultb

200

34no

24.8

onlin

e;R

Ta

4M

_1%

defa

ultb

200

51no

24.8

onlin

e;R

Ta

5M

_1%

defa

ultb

200

48no

24.8

onlin

e;R

Ta

6M

_1%

defa

ultb

200

45no

24.8

onlin

e;R

Ta

7M

_1%

defa

ultb

200

44no

24.8

onlin

e;R

Ta

8M

_1%

defa

ultb

200

42no

49.6

offli

ne;3

0m

in37

°C3

M_1

%de

faul

tb20

045

no

16.5

onlin

e;R

Ta

4M

_1%

MS

1µs

can

5021

no16

.5on

line;

RT

a4

M0.

5M

MS

1µs

can

500

no16

.5on

line;

RT

a4

M_1

%0.

02M

S1

µsca

n50

27no

16.5

onlin

e;R

Ta

0.1

MN

aH2P

O4

HC

l0.

02M

S1

µsca

n50

45no

16.5

onlin

e;R

Ta

4M

_1%

defa

ultb

5021

no16

.5on

line;

RT

a4

M_1

%1

µsca

n_20

0m

s50

20no

16.5

onlin

e;R

Ta

4M

_1%

defa

ultb

yes

200

25no

16.5

onlin

e;R

Ta

4M

0.5

Mde

faul

tbye

s20

027

yes

16.5

onlin

e;R

Ta

4M

_1%

0.02

defa

ultb

yes

200

35no

16.5

onlin

e;R

Ta

0.1

MN

aH2P

O4

HC

l0.

02de

faul

tbye

s20

046

no16

.5on

line;

RT

a0.

1M

NaH

2PO

4T

FA0.

02de

faul

tbye

s20

042

no24

.8on

line;

RT

a0.

1M

NaH

2PO

4H

Cl

0.02

defa

ultb

200

41no

24.8

onlin

e;R

Ta

0.1

MN

aH2P

O4

HC

l0.

02de

faul

tbye

s20

048

no24

.8of

fline

;30

min

37°C

0.1

MN

aH2P

O4

HC

l0.

02de

faul

tbye

s20

055

no

16.5

onlin

e;R

Ta

0.2

M0.

1M

NaH

2PO

4H

Cl

0.02

defa

ultb

yes

200

37ye

s16

.5on

line;

RT

a0.

1M

NaH

2PO

4H

Cl

0.02

defa

ultb

yes

200

55ye

s16

.5on

line;

RT

a0.

2M

0.02

defa

ultb

yes

200

46ye

s16

.5on

line;

RT

a0.

1M

NaH

2PO

4H

Cl

0.02

defa

ultb

yes

200

36no

16.5

onlin

e;R

Ta

2m

M0.

1M

NaH

2PO

4H

Cl

0.02

defa

ultb

yes

200

50on

lyon

eC

16.5

onlin

e;R

Ta

5m

M0.

1M

NaH

2PO

4H

Cl

0.02

defa

ultb

yes

200

51ye

s16

.5on

line;

RT

a10

mM

0.1

MN

aH2P

O4

HC

l0.

02de

faul

tbye

s20

058

yes

16.5

onlin

e;R

Ta

20m

M0.

1M

NaH

2PO

4H

Cl

0.02

defa

ultb

yes

200

49ye

s16

.5on

line;

RT

a0.

1M

NaH

2PO

4H

Cl

0.02

defa

ultb

yes

200

34no

[�2A

R]

(µM

)

peps

inon

line/

offli

ne;

dige

stio

nte

mp

urea

_TFA

TC

EP

(mM

)

NaH

2PO

4pH

2.40

HC

l/T

FA

DD

M(m

/v)

(%)

MS,

MS/

MS

µsca

n;m

axim

umin

ject

ion

time

reje

ctlis

t(5

m/z

valu

es)

HPL

Cflo

wra

te(µ

L/m

in)

MA

SCO

T(>

20)

cove

rage

(%)

CC

Clo

opco

vere

d?

sear

chw

ithca

rbox

amid

ylm

ethy

lCys

(Cap

-C)

sear

chw

ithC

ap-C

and

degl

ycos

ylat

ion

15on

line;

RT

a4

M_1

%de

faul

tbye

s50

35no

ne47

%C

CD

cove

red

52%

CC

D15

onlin

e;R

Ta

100.

1M

NaH

2PO

4H

Cl

0.02

defa

ultb

yes

5065

only

one

C71

%C

CC

cove

red

73%

CC

Cco

vere

d15

onlin

e;R

Ta

200.

1M

NaH

2PO

4H

Cl

0.02

defa

ultb

yes

5064

yes

69%

CC

Cco

vere

d70

%C

CC

cove

red

aR

Tw

as27

.8±

0.5

°C.b

defa

ult

repr

esen

tsM

S,3

µsca

n_75

ms

and

MS/

MS,

3µs

can_

100

ms.

1103Analytical Chemistry, Vol. 82, No. 3, February 1, 2010

SDS-PAGE and Western Blot of �2AR. �2AR proteins withand without PNGase F treatment were mixed with NuPAGELDS sample buffer and loaded onto 10% Bis-Tris gels andstained with SimplyBue SafeStain (Invitrogen) or transferredto a membrane for Western blot using anti-FLAG antibody at1:1000 (v:v) dilution.

RESULTS AND DISCUSSIONOptimization of Protein Sequence Coverage: Detergent,

Reductant, and HPLC Parameters. To obtain a comprehensiveHDX profile of �2AR, procedures for analysis were progressivelyoptimized directed by insights gained from the crystal struc-ture. Consideration was also made for putative post-translationalmodifications and considering the physical characteristics ofmembrane proteins. Our initial experiments used a quenchsolution of 3 M urea and 1% TFA; however, very limitedsequence coverage was obtained for �2AR (∼25%). In an effortto improve sequence coverage various experimental parameterswere systematically optimized. Parameters optimized includedprotein concentration, various quench buffers (different com-binations of denaturing and reducing reagents at variousconcentrations), temperature and length of digestion, HPLCconditions, MS acquisition settings (number of microscans,maximum injection time), and the application of a staticexclusion ion list including several recurrent low m/z nonpep-tide ions for the MS/MS experiments (see Table 1).

During optimization of HDX experimental parameters, variousprotein concentrations were considered. As expected, 25 µM �2ARafforded higher sequence coverage than 16 µM, indicating thatthe digestion efficiency is not limited by the digestion capacityof the pepsin column (Table 1). With the use of the typical HDXquench solutions composed primarily of high concentrations ofurea and 1% (v/v) TFA,12,13 poor sequence coverage was obtainedfor all concentrations of urea tested (3-8 M) with 4 M ureaproviding the highest coverage. Also, without reductant, nopeptides were observed from regions in proximity to the disulfidebonds. Although urea is usually considered to denature proteinsmaking them more susceptible to proteases, the reagent waseither not efficient at denaturing the membrane protein �2AR orwas unable to prevent aggregation at the low pH conditions ofthe digestion step.29 Use of quench solutions composed of 0.1M NaH2PO4 (pH 2.4)/0.02% DDM consistently afforded highercoverage than quench buffers containing urea.

The crystal structure of �2AR revealed two disulfide bondsin �2AR (corresponding to C106-C191 and C184-C190 in�2AR) and palmitoylation of C341. Out of the total 11 cysteineresidues, three cysteine residues (C77, C265, and C327) weredetermined to be present predominantly carboxamidomethy-lated as a result of prior iodoacetamide treatment (Table S1 inthe Supporting Information). Peptides containing unmodifiedC265 and C327 were also detected but at low abundance.Consistent with the crystal structure, C265 was noted to bemodified with an acetamide group in the electron density maps.These results suggested that these three cysteines are likelyreduced and are readily accessible to the iodoacetamide treatment.

Peptides containing cysteines that are involved in disulfide bondswill only be detected after exposure of protein to reductant. Toimprove coverage in proximity of disulfide bonds, various reducingconditions were considered. The reductant TCEP has previouslybeen applied in HDX studies to reduce disulfide bonds underacidic conditions and is typically used at high concentrations (1M).30 However, high concentrations of TCEP were shown hereto yield poor peptide sequence coverage (Table 1). Upon inves-tigation, high concentrations of TCEP (>200 mM) resulted in massspectra dominated by clusters of peaks formed by adducts ofTCEP throughout most of the chromatographic gradient (data notshown) thus interfering with peptide identification. The interfer-ence results from overlapping ions, suppression of ionization, andfilling the ion trap with chemical noise. In addition, the highlyionic nature of TCEP might have caused the membrane proteinto aggregate or precipitate prior to peptic digestion. Although thesequence coverage was poor, the region of I182-F193 (IN-CYAEETCCDF) in the extracellular loop of �2AR, previouslyshown in crystal structures to contain one and a half disulfidebonds (C106 with C191 and C184 with C190)7,8 was observedonly when TCEP was used, confirming that this region is disulfidebonded in the protein construct used here. Upon further investiga-tion of the �2AR crystal structures (PDB: 3D4S and 2RH1)indicate that this region is in the solvent exposed extracellularloop (Figure 3 and Figure S1 in the Supporting Information),suggesting that perhaps low concentrations of TCEP might besufficient to reduce this disulfide bond. Thus, addition of TCEPto the quench solution (0.1 M NaH2PO4 (HCl, pH 2.40) 0.02%(m/v) DDM) was tested at various concentrations (0, 2, 5, 10,20, 200, and 500 mM). Results displayed in Table 1 demonstratethat incorporation of TCEP at concentrations as low as 5 mM levelis sufficient to reduce the disulfide bonds in �2AR and asconcentrations of TCEP reached 20 mM, peptide sequencecoverage started to decrease. Optimal TCEP concentration wasdetermined to be in the range of 10-15 mM, and 15 mM wasused in the HDX studies presented here. Unfortunately,

(29) Lim, W. K.; Rosgen, J.; Englander, S. W. Proc. Natl. Acad. Sci. U.S.A. 2009,106, 2595–2600.

(30) Woods, V. L., Jr.; Hamuro, Y. J. Cellular Biochem. Suppl. 2001, 37, 89–98.

Figure 1. Western blot analysis of carazolol-bound �2AR afterPNGase F incubation at 4 °C for 1 h, probed by using anti-FLAG (A),following separation by SDS-PAGE gel 10% Bis-Tris stained bySimplyBlue Safestain (B).

1104 Analytical Chemistry, Vol. 82, No. 3, February 1, 2010

peptides containing C116 and C285 were not observed underall conditions employed.

The use of DDM at 0.02% m/v (∼3 cmc) in quench solutionconsistently afforded improved sequence coverage under allconditions suggesting that keeping the membrane protein solubleand stable prior to and during digestion is important to obtaininghigh sequence coverage. The use of DDM in the quench buffermay also aid the solubilization of extremely hydrophobic peptidesgenerated during peptic digestion. Interestingly, DDM eluted nearthe end of the HPLC gradient as a single sharp peak, indicatingits high purity, and resulted in minimal interference of �2ARpeptides. DDM formed predominantly protonated dimeric ionswith minimal protonated monomer or cationic adduct formationobserved. These results suggest that, in contrast to proteasedigestion of soluble proteins, digestion of highly hydrophobicmembrane proteins requires careful maintenance of proteinsolubility in addition to efficient denaturation. While denatur-ation involves alteration of the protein tertiary structure, partialor complete protein aggregation may occur resulting in poordigestion efficiency by limiting the access of the protease tocleavage sites.

Several mass-spectrometry-compatible detergents that involvecleaving the detergent into smaller moieties prior to massspectrometry analysis have recently been applied to improvingprotein analysis.31,32 However, some types of the cleavabledetergents are ionic and have been shown to cause precipitationof purified membrane proteins. In addition, the time required tocleave these detergents introduces an undesirable time windowfor back exchange. Each membrane protein is unique, and itsstructure, stability, and activity are highly sensitive to the proper-

ties of a detergent.20 For structural studies including HDX, it isessential that the detergent used during purification and analysisdoes not alter the protein’s structure and stability. The evidencethat DDM, a detergent proven to maximally maintain the structureand activity of �2AR,6–8 can significantly improve the digestionefficiency without causing appreciable interference in the massspectrometer, makes the use of cleavable detergents unnecessaryfor HDX studies of �2AR.

Effects of temperature and length of digestion time wereoptimized by comparing sequence coverage and deuteriumretention. Briefly, online digestion using immobilized pepsin-coupled column at room temperature was compared to offlinedigestion using a pepsin-coupled beads slurry at 37 °C for 30 minand online digestion at 1 °C and room temperature (27 °C) usingt0 MS (Table 1). Not unexpectedly, these results demonstrate thatdigestion for longer time at higher temperature yields highersequence coverage (offline 37 °C for 30 min vs online at roomtemperature for less than 1 min). However, it is important to notethat these on-line digestion conditions had minimal impact ondeuterium retention. With the use of myoglobin as a control, noappreciable difference in deuterium recovery was observed whenthe pepsin-coupled column was held at 1 or 22 °C.

Both C4 and C8 analytical columns were chosen in conjunctionwith a C8 trap column to ensure the recovery of the hydrophobiccomponents within the gradient used for HDX.33 With the use ofthe C8 trap column and C4 or C8 column, peptides were elutedwithin the 9.5 min HPLC gradient. Importantly, under theseconditions DDM eluted as a narrow well-defined peak just priorto the end of the gradient. In addition, intact diacylglycerolphos-pholipids isolated along with �2AR eluted after the DDM peak.Nearly all of the peptides are each found only at a single

(31) Chen, E. I.; McClatchy, D.; Park, S. K.; Yates, J. R., III Anal. Chem. 2008,80, 8694–8701.

(32) Chen, E. I.; Cociorva, D.; Norris, J. L.; Yates, J. R., III J. Proteome Res. 2007,6, 2529–2538.

(33) Roos, M.; Soskic, V.; Poznanovic, S.; Godovac-Zimmermann, J. J. Biol. Chem.1998, 273, 924–931.

Figure 2. Peptide sequence map showing the MS/MS and MS verified coverage of carazolol-bound �2AR obtained under experimental conditionsused in HDX experiments: 15 µM deglycosylated �2AR digested online using the immobilized pepsin column at room temperature (22 °C) and thequench solution containing 0.1 M NaPi_HCl (pH 2.4) 0.02% (m/v) DDM and 10-15 mM TCEP. The pink boxes indicate sites of N-glycosylation. Theorange boxes indicate a cysteine residue (see Table SI in the Supporting Information). The green box indicates the site of protein palmitoylation.

1105Analytical Chemistry, Vol. 82, No. 3, February 1, 2010

retention time range, suggesting that aggregation of proteinor peptide in the eluted solution was minimal. This suggestedthat inclusion of detergents in the HPLC mobile phase is notrequired.

In initial MS/MS analysis of �2AR under HDX conditions,sequence coverage of helix 1 and helix 3 was minimal. Theseregions of the protein are very hydrophobic and containmultiple pepsin cleavage sites,34 suggesting that under HDXconditions this region would only be represented by severalsmall hydrophobic peptides. In addition to pepsin treatment,other acid stable proteases (fungal XIII and XVIII) were testedbut failed to improve sequence coverage in this region. Thus,in an effort to improve detection of these peptides, the MS/MS experiment was run with an extended 120 min gradient.This method improved sequence coverage to 89% (Figure 2).Changing the MS/MS selection from the top five ions to the top12 ions during the 120 min gradient did not result in any furtherimprovement in sequence coverage, suggesting that the extendedgradient afforded efficient separation of the peptic peptides.

Optimization of Protein Sequence Coverage: Mass Spec-trometer Settings. Application of a precursor ion reject listcontaining five recurrent abundant precursor ions in data-depend-ent acquisition improved the coverage of identified peptides (Table1). Blocking these abundant nonpeptide ions allowed the lessabundant peptide ions to be selected for MS/MS analysis. Thereject list was applied in addition to the dynamic exclusion.Increasing the maximum ion injection time for MS/MS experi-ments while keeping the target number unchanged can potentiallyenhance the fragment ion abundance in the MS/MS spectrumby allowing the precursor ions more time to fill the trap to reachthe target number before they are fragmented and scanned outto the detector. However, increasing the maximum injection timefor MS/MS acquisition from 100 to 200 ms did not improvesequence coverage, suggesting that failing to identify peptidesfrom the uncovered sequence is not due to their low fragmentationefficiency or low precursor ion abundance. Peptides not observedwere either not formed efficiently during digestion and/or theirabundance was not sufficient to trigger MS/MS. It is also possiblethat the peptides are formed but cannot be trapped with the HPLCcolumns and chromatography conditions employed here. The

combined optimal HDX experimental conditions were determinedto use a quench buffer of 0.1 M NaH2PO4 (HCl, pH 2.40) 0.02%(m/v) DDM with offline digestion at 37 °C for 30 min of 25µM �2AR.

Optimization of Protein Sequence Coverage: ProteinModifications. On the basis of the homology with seven othertransmembrane-helix GPCRs,27,33,35 glycosylation and phospho-rylation of the N-terminus of �2AR was expected. These modi-fications have been reported in human �2AR using biochemicaland proteomic approaches, and the phosphorylation was foundto be regulated by ligand binding.26 Initial MS/MS studies of�2AR failed to detect peptic peptides from the N-terminus andtransmembrane helix 1, even though this region of the proteinis extracellular and should be accessible to protease. As thisregion of the protein contains two putative glycosylation sites,N6 and N15,26,27,33,35,36 the protein was treated with PNGase Fto remove N-glycosylation of asparagines. Western blot analysisof �2AR following 1 h of PNGase F treatment (Figure 1) clearlydemonstrates effective removal of sugars. No difference wasobserved between the 4 and 37 °C treatments. MS/MS analysisof �2AR following PNGase F treatment lead to the identificationof peptides originating from the N-terminus (residues 1-19),confirming that �2AR was heavily glycosylated.

Further post-translational modifications were considered in aneffort to improve sequence coverage. Analysis of the crystalstructure of �2AR shows that C334 is palmitoylated. However,in all MS and MS/MS runs under HDX conditions, no pepticpeptides were detected that corresponded to C334 as eitherunmodified or palmitoylated even though the palmitoyl groupwas observed in the crystal structure. Thus, this region of thereceptor was undetected in the HDX experiments describedbelow. Lastly, �2AR contains several phosphorylation sites26

although the �2AR protein material made in sf9 cells are notexpected to be phosphorylated under the existing conditions.MASCOT analysis of all MS/MS spectra confirmed the absenceof phosphorylated residues. Finally, MASCOT analysis of allMS/MS spectra where methionine was considered as sulfoxide

(34) Hamuro, Y.; Coales, S. J.; Molnar, K. S.; Tuske, S. J.; Morrow, J. A. RapidCommun. Mass Spectrom. 2008, 22, 1041–1046.

(35) Papac, D.; Thornburg, K.; Bullesbach, E.; Crouch, R.; Knapp, D. J. Biol.Chem. 1992, 267, 16889–16894.

(36) Pepinsky, R. B.; Zeng, C.; Wen, D.; Rayhorn, P.; Baker, D. P.; Williams,K. P.; Bixler, S. A.; Ambrose, C. M.; Garber, E. A.; Miatkowski, K.; Taylor,F. R.; Wang, E. A.; Galdes, A. J. Biol. Chem. 1998, 273, 14037–14045.

Figure 3. HDX profile of carazolol-bound �2AR mapped to a model derived from the 3-D crystal structure of �2AR T4L (PDB: 2RH1) after anexchange time of 30 s (A), 300 s (B), and 15 h (C). Two views are represented for each time point. Each pair corresponds to rotation through180°. The structure model was derived from 2RH1 with removal of T4L amino acids to reflect the construct used for HDX studies. All nonproteinmolecules were removed with the exception of carazolol. The gray cartoons represent the sequences not presented in the HDX profile.

1106 Analytical Chemistry, Vol. 82, No. 3, February 1, 2010

or sulfone did not result in an improvement of �2AR sequencecoverage, suggesting that the methionine oxidation is minimal.

HDX Profile of �2AR Bound to Carazolol. To visualize thedynamics of �2AR, HDX data obtained at 30 s, 300 s, and 15 h

Figure 4. HDX profile of deglycosylated �2AR mapped to its sequence after exchange time of 10 s, 30 s, 60 s, 300 s, 900 s, 3600 s, 12 h, and15 h and discarding the first two amino acid residues at the N-terminus of each peptide (A), and graphs of %D as a function of exchange timein log scale for selected peptide ions shown in panel A (B). The error bars in panel B were plotted as the standard deviation of four independentreplicates, although in some cases the error bars are too small to be visible.

1107Analytical Chemistry, Vol. 82, No. 3, February 1, 2010

were overlaid onto a structure model derived from the crystalstructure of �2AR T4L bound to carazolol (PDB: 2RH1) asshown in Figure 3. The structure model was derived from PDB:2RH1 with removal of T4L amino acids to reflect the constructused for HDX studies. All nonprotein molecules were removedwith the exception of bound carazolol. A comprehensive HDXprofile of carazolol-bound �2AR is presented in Figure 4A.Deuterium incorporation versus time (log scale) for severalrepresentative peptides are shown in Figure 4B (see Figure S2 inthe Supporting Information for all plots). As illustrated in Figure3A, significant protection to exchange was observed for alltransmembrane helices throughout the entire time course moni-tored (15 h). The highest degree of protection to exchange wasdetected toward the core of these helices, while the linkagesbetween helices displayed moderate exchange kinetics. Peptidesrepresentative of the third cytoplasmic loop at the C-terminus ofhelix 5 (3rd intracellular loop) demonstrated the most rapidexchange kinetics. This is the region of �2AR that was replacedwith T4L to facilitate crystallization and also the loop regionknown to interact with G-R. Peptides representative of theN-terminus of the receptor also exhibited rapid exchangekinetics. The short helical loop between helix 4 and helix 5(2nd extracellular domain) displayed an intermediate exchangerate that is faster than the transmembrane helices but slowerthan the intrahelical linkages. This observation is interestingsince the rhodopsin second extracellular loop region coversthe access to the retinal binding site in the rhodopsin structurewhere retinal undergoes a cis to trans isomerization instead ofretinal coming on and off.37 In contrast to the �2AR structureand the HDX data, the second extracellular loop that containsan ordered short helix may be a lid that is capable of movingover and away from the ligand binding site in differentconformational states.

CONCLUSIONSPreviously we have described a robust HDX platform that

provides comprehensive and reproducible analysis of proteins

such as nuclear receptors, enzymes, and matrix metallopro-teases.13 Using this HDX platform, we uncovered a novel mech-anism for partial agonist activation of the nuclear receptor PPARγ,unique binding modes of natural and synthetic estrogens on ERRand ER�, and substrate binding site on MMP1.13,38-41 Here wedemonstrate the ability to determine which regions of the integralmembrane �2-adrenergic G-protein coupled receptor are moredynamic than other regions of the membrane protein. Althoughfor membrane proteins, the regions of flexibility might seemobvious, the results presented here highlight a number ofdynamic secondary elements being observed in relatively smallextra- or intracellular loops indicating the functional importanceof these regions. While analysis of �2AR with previouslyoptimized conditions resulted in less than ideal sequencecoverage, optimization of several experimental parameters andby consideration of the sequence and structure of the receptor,sequence coverage of �2AR was significantly improved whilemaintaining HDX compatible conditions. Following optimiza-tion, excellent sequence coverage was obtained spanning theentire �2AR structure with few exceptions. Importantly, se-quence coverage was obtained for regions that were unresolvedin the crystal structures, and these highly dynamic regions ofthe receptor are likely important to receptor function such asligand recognition and receptor signaling. The improvedmethodology presented here enabled effective and compre-hensive HDX characterization of human �2AR bound to theinverse agonist carazolol. A great deal of insight on GPCRdynamics will come from differential HDX analysis of differentligands bound such as comparison of agonist, antagonists, andinverse agonists and observing the changes in dynamicsrelative to the varying functional states of the receptor. Themethodology presented paves the way for future analysis ofother GPCR family members.

ACKNOWLEDGMENTThe work is supported by GM084041 (X.Z., P.R.G., M.J.C.,

B.D.P) and NIH Roadmap Initiative Grant P50 GM073197 (E.Y.T.C.,R.C.S.).

SUPPORTING INFORMATION AVAILABLEAdditional information as noted in text. This material is

available free of charge via the Internet at http://pubs.acs.org.

Received for review November 1, 2009. AcceptedDecember 16, 2009.

AC902484P

(37) Palczewski, K.; Kumasaka, T.; Hori, T.; Behnke, C. A.; Motoshima, H.; Fox,B. A.; Le Trong, I.; Teller, D. C.; Okada, T.; Stenkamp, R. E.; Yamamoto,M.; Miyano, M. Science 2000, 289, 739–745.

(38) Dai, S. Y.; Burris, T. P.; Dodge, J. A.; Montrose-Rafizadeh, C.; Wang, Y.;Pascal, B. D.; Chalmers, M. J.; Griffin, P. R. Biochemistry 2009, 48, 9668–9676.

(39) Dai, S. Y.; Chalmers, M. J.; Bruning, J.; Bramlett, K. S.; Osborne, H. E.;Montrose-Rafizadeh, C.; Barr, R. J.; Wang, Y.; Wang, M.; Burris, T. P.;Dodge, J. A.; Griffin, P. R. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 7171–7176.

(40) Avvaru, B. S.; Busby, S. A.; Chalmers, M. J.; Griffin, P. R.; Venkatakrishnan,B.; Agbandje-McKenna, M.; Silverman, D. N.; McKenna, R. Biochemistry2009, 48, 7365–7372.

(41) Lauer-Fields, J. L.; Chalmers, M. J.; Busby, S. A.; Minond, D.; Griffin, P. R.;Fields, G. B. J. Biol. Chem. 2009, 284, 24017–24024.

1108 Analytical Chemistry, Vol. 82, No. 3, February 1, 2010