Cloning and Characterization of a Potential Transcriptional Activator of Human γ-Globin Genes

(MEL)se

h laythat

l liversamefitting

t rom eacho teosomes d threec id systema es in theN eral fetalt gene mayi w

Yang et al. Blood Cells, Molecules, and Diseases (2001)27(1) February: 1–15

doi:10.1006/bcmd.2000.0344, available online at http://www.idealibrary.com on

Cloning and Characterization of a Potential TranscriptionalActivator of Human g-Globin GenesSubmitted 11/13/00(Communicated by G. Stamatoyannopoulos, M.D., 11/15/00)

Yi Yang,1 Zhijun Duan,2 Eva Skarpidi,2 Qiliang Li,2 Thalia Papayannopoulou,1

and George Stamatoyannopoulos2

ABSTRACT: Hybrids produced by fusing human fetal erythroblasts (HFE) with mouse erythroleukemiacells initially produce predominantly or exclusively humang-globin and switch to humanb globin expression atime in culture advances. One explanation for the initially predominant expression ofg-globin gene in thes

ybrids is the presence of trans-acting factors that activateg-globin gene transcription. We used differential dispof hybrids before and after theg to b switch as well as fetal liver and adult erythroblasts to identify cDNAscould be candidates for potentialg gene activators. Identically sized amplicons which were present in fetaerythroblasts and in the hybrids expressing onlyg-globin but were absent in the adult erythroblasts and in thehybrids after they had switched tob globin expression were cloned and sequenced. Fifty pairs of cDNAshese criteria were chosen for further analysis. The sequences of the two members of 48 pairs differed fther, revealing the low efficiency of this experimental approach. One clone pair coded for human proubunit X. The second pair coded for a protein containing an acidic domain in the N-terminus anonsecutive CDC10/SW16/ankyrin repeats in the C-terminus. Transactivation assays in the yeast hybrnd transient transfection assays in COS cells showed that a potent trans-activating domain resid-terminus of this protein. Northern blot and RT-PCR assays showed that this gene is expressed in sev

issues but not in adult tissues. Stable transfection assays provided evidence that the product of thisncrease the level ofg mRNA in HFE3 MEL cell hybrids that undergo theg to b switch, suggesting that this negene encodes a protein that may function asg gene activator. © 2001 Academic Press

Key Words:differential mRNA display; somatic hybrids; humang-globin gene; transcriptional activator.

res-sentctsR(3,)cti-ct

ha-n

ing

p tein( rs,F cti-v on-

in,ins (9,he.

T oth-e the

opoulo ashington,on.edugton 9

INTRODUCTION

The switches of human globin gene expsion during development presumably reprechanges in the transcriptional milieu that interawith specific motifs of globin genes and the LC(1). The cloning (2) and the targeted disruption4) of the erythroid Krupple-like factor (EKLFstrongly suggest that EKLF is essential for avation of b-globin gene. EKLF fails to interawith the g gene promoter and activateg geneexpression (5). Little is known about the mecnism of silencing of theg-globin gene other tha

Correspondence and reprint requests to: George StamatoyannSeattle, WA 98195. Fax: 206-543-3050. E-mail: [email protected] Division of Hematology, University of Washington, Seattle, Washin
2 Division of Medical Genetics, University of Washington, Seattle, Washing
1

that it is autonomous (1, 6, 7) and that silencsequences are located in the upstreamg gene

romoter. Three factors, the stage selector proSSP), and two CACCC box binding factoKLF and FKLF-2 have been proposed to aate theg gene. The SSP is a heterodimer c

sisting of CP1 and an erythroid specific proteNFE4 (1, 8). FKLF and FKLF-2 act as globgene activators in transient expression assay10) by interacting with the CACCC boxes of tg, e and, to a much lesser degree,b globin genes

hese two factors were cloned under the hypsis that EKLF-type factors may modulate

s, Division of Medical Genetics, Box 357720, University of W.8195.
ton 98195.
1079-9796/01 $35.00Copyright© 2001 by Academic Press

All rights of reproduction in any form reserved

s by

flls,lon-rsglo-of

anrinero-ithinralhisd

t t thel ve

ente

f tot ridsw -

tin-

fi sc on-t n-

mo-ith

theins

s asuttheultro-tect

heirthisen-A

fh

M

C

b-t b-t inI ulte anb em /mlh ec ered usedf singM a5 eth ellsw ads of

ca-A

as(12).-40byere

a .A and4 hed NAf andr im-e ndc I).A eu

D

thed T7p rsion

Blood Cells, Molecules, and Diseases (2001)27(1) February: 1–15 Yang et al.doi:10.1006/bcmd.2000.0344, available online at http://www.idealibrary.com on

expression of embryonic and fetal globin geneinteracting with thee and theg gene CACCCboxes (9, 10).

In parallel to our cloning of FKLF-type ofactors from embryonic and fetal erythroid cewe have pursued an alternative approach for cing potential g gene activators or coactivatofrom erythroid tissues that express fetal hemobin. We have previously shown that fusionmurine erythroleukemia (MEL) cells with humcells produce hybrids, which express the muas well as the human globins (11). Hybrids pduced by fusing human fetal erythroblasts wMEL cells synthesize initially fetal hemogloband switch to adult globin formation after sevemonths in culture (11). One explanation of tphenotype is that the initial highg expression anhe subsequent switch are controlled in cis, aevel of g locus chromatin (11). An alternatihypothesis is that the transcriptional environmthat is responsible for the highg expression in thetal liver erythroid cells might be transferred inhe hybrids and it is present in those hybhich express high levels ofg mRNA; subse

quently this transcriptional environment is exguished, triggering theg to b switch. In therst case theg to b switch in the hybrids iontrolled in cis. In the second case it is crolled in trans, perhaps by factors that are ecoded by genes located in the human chrosomal component transferred into MEL cells wthe cell fusion.

If trans-acting proteins are responsible forg-globin expression in the hybrids, these proteshould be present in human fetal erythroblastwell as in theg-globin expressing hybrids bshould be absent (or present in low levels) inb globin expressing hybrids and in the aderythroblasts. If the mRNA coding for these pteins are not very rare, it may be possible to detheir presence in the fetal erythroid cells and tabsence in the adult erythroblasts. To testhypothesis we applied the technique of differtial mRNA display (12) to compare the RNspecies present in a human fetal erythroid3 MELhybrid before or after theg to b switch to those o

uman fetal and human adult erythroblasts. 2

2

ATERIALS AND METHODS

ells

Human fetal erythroblasts (HFE) were oained by culturing human fetal liver cells (oained from 59-day-old fetuses) overnightMDM, 10% FCS and 10 u/ml Epo. Human adrythroblasts were obtained by culturing humone marrow CD341 cells in methylcellulosedia containing 30% FCS, 10% BSA, 50 ng-SCF, 50 u/ml IL-3 and 10 u/ml Epo. BFUolonies were lifted from the plates, the cells wispersed and used in our assays. The hybrid

or these studies has been produced by fuEL cells with liver-derived erythroblasts from2-day-old fetus. A pool of cells from the tim

his hybrid was synthesizing mainlyg-globinave been cryopreserved. Another pool of cere cryopreserved when this hybrid hwitched tob globin expression. The two pools

cells were expanded and used for RNA purifition and the application of the differential mRNdisplay.

Differential mRNA Display

The method used for differential display wessentially as described by Liang and PardeeCytoplasmic RNA was prepared from the NPlysate of fresh cells and converted to cDNAdT12–18 and reverse transcriptase. cDNAs w

mplified using T12VN and 59 arbitrary primerfter PCR, the samples were displayed on 6.5% polyacrylamide DNA sequencing gels. Tifferentially displayed bands were excised. D

ragments were recovered from the gel sliceseamplified by PCR using the appropriate prrs. Reamplified products were purified aloned into the pGEM-T vector (Promega, Wll routine DNA recombinant work was donsing standard techniques (13).

NA Sequence Analysis

DNA sequencing was performed byideoxy chain termination method (14) usingrimer and SP6 primer and the Sequenase ve
DNA sequencing kit, (USB, OH). Sequences

the

bya-

-%,

0in

r, adonedithin

s3-

edasinto

nd

ted,

a I

tictheidt ofro-po-o-n in-onon,BS

Lu-rase

asu thep sou sionp wn-s ree hy-b alf-sh )a d

ridncemeinglls)Mo

c ino ces

R

asi NAw ithp 2nde pro-t

-t izedD elso byP

I

ga-t tri-



were compared with GenBank database usingFASTA (15) and the BLAST (16) programs.

Northern Blot Hybridization

Northern blot hybridization was performedthe method of Pelle and Murphy (17). Hybridiztion was performed overnight after 4 h prehybridization at 42°C in hybridization solution (50formamide, 43 SSPE, 23 Denhardt’s solution

.1% SDS, 100 mg/ml tRNA, 125 mg/mlpoly(A)). The membranes were washed twice23 SSC, 0.1% SDS, twice in 0.13 SSC, 0.1%SDS at 68°C and exposed to X-ray film.

Yeast Transactivation Assays

To clone FGIF into a yeast expression vectoNcoI (CCATGG) site was created at the start coof FGIF by PCR. The cDNA fragment was cloninto pAS2-1 vector (Clontech, CA) in-frame wthe GAL4 DNA binding domain. A yeast straY153 (MATa leu2-3,112, ura3-52, trp1-901, hiD200, ade2-101, gal4Dgal80DURA3::GAL-lacZ,LYS2::GAL-HIS3) which contains an integratLacZ reporter gene controlled by GAL UAS wused. The expression plasmid was transformedY153 yeast cells by the LiAc method (18) amaintained by selection on Trp2 medium. Individ-ual Trp1 colonies were maintained in a Trp2 liquidmedium for 2 days. Yeast cells were harvesresuspended in lysis buffer and assayed forb-galctivity using the Luminescentb-gal detection kit I

(Clontech, CA).

Transient Transfection Assays

FGIF cDNA was cloned into an eukaryoexpression vector pSG424 in-frame withGAL4 DNA binding domain. A reporter plasmpG5BCAT which contains five tandem repeaGAL4 binding sequence and E1b minimal pmoter was cotransfected into COS cells by lifection (Life technologies, MD). An RSV prmoter-driven luciferase gene was used as aternal control to normalize for transfectiefficiency. Forty-eight hours after transfectithe cells were collected, washed twice with P
and lysed. CAT activity was measured using a f
3

CAT enzyme assay system (Promega. WI).ciferase activity was assayed using a lucifeassay system (Promega, WI).

Stable Transfections

A mLCR-b globin promoter cassette (19) wsed to replace the CMV promoter fromcDNA3 (Invitrogen, CA) vector which was alsed as a control plasmid. The FGIF expreslasmid was constructed by placing FGIF dotream of themLCR-b globin promoter to assurythroid-specific high-level expression in therid cells. Plasmids were transfected into hwitched or completely switched HFE3 MELybrids by lipofection (Life Technologies, MDnd selected with 700mg/ml G418. Repeate

enrichment in chromosome 11 containing hybcells was achieved by indirect immune adhere(panning) as described (11). At various tipoints, recently panned cell aliquots (containhuman chromosome 11 in over 90% of the cewere induced in medium containing 3 mHMBA and 10mM hemin for 3 days in order tollect cells for RNA analysis and for four daysrder to collect cells for immunofluorescentudies.

NA Analysis and Quantitation

Total cellular RNA from the induced cells wsolated as described (11). Human globin mRas analyzed by RNAse protection assay wrobes that gave protected fragments from thexon of the respected gene (the size of the

ected bands is in parentheses): pT7b(205) andpT7Ag(170). Mousea globin mRNA was ana-lyzed with a pT7a(128) probe from exon I. Anisense probes were synthesized from linearNA templates using T7 polymerase. The levf human and murine mRNA were quantitatedhosphorImager analysis.

mmunofluorescence Staining

Induced cells were collected by centrifuion, washed and fixed in methanol. Cytocen
uge smears of fixed cells were stained with spe-

-

s

tstiale

t-l mt om-i led

ofyth-ichaststherid,vingwithyth-lon-s.Aetalg-hy-en-mridem-

er asbaseAse,

romse-. Inewen-

ands ad toairsne

rine

othetaledersnce

nce

elandbridr

a useo hy-b redt s

al

m5er

Aofpennce

up-t thisrod-nce

se-res-

peatu-I6

therionf),e

p enef ,


cific anti-humang- andb-globin monoclonal antibodies as described (11).

RESULTS

Identification of Differentially Displayed mRNA

Differentially expressed cDNA fragmenwere identified by using the method of differenmRNA display. The mRNAs of a hybrid at a timit expressed predominantlyg mRNA (subsequeny called the “fetal” hybrid) and the mRNAs frohe same hybrid after it had switched to prednantly b mRNA synthesis (subsequently calthe “adult” hybrid) were compared to thosefetal human erythroblasts and adult human erroblasts. Amplified fragments of equal size, whwere present in both the human fetal erythrobland the fetal hybrid but were absent in bothhuman adult erythroblasts and the adult hybwere cloned and sequenced. Fragments hadifferent sequences were discarded. Those,identical sequences in both the human fetal erroblasts and the fetal hybrid, were used for cing full-length cDNA and for functional studie

Fifty pairs which had identically sized cDNfragments in the fetal erythroblasts and the fhybrid lanes but differently sized cDNA framents in the adult erythroblast and the adultbrid were selected for further analysis. The idtically sized cDNA fragments were cloned frohuman fetal erythroid cells and the fetal hyband sequenced. The cDNA sequences of the mbers of each pair were compared to each othwell as with sequences in the GenBank datausing the FASTA and the Blast program.shown in Table 1, in spite of their identity in sizin 48 pairs the sequences of the fragment fhuman fetal erythroblasts differed from thequences of the fragment from the fetal hybrid36 such pairs both cDNAs of the pair were n(i.e., their sequences were not present in GBank). In seven pairs (pairs 3, 5, 8, 22, 34, 42,44 of Table 1), one member of the pair wanovel cDNA sequence while the other belongean already identified human gene. In five p(pairs 7, 15, 19, 30, and 46 of Table 1), o
member of the pair was a novel sequence while
4

the other belonged to a previously cloned mugene.

Two pairs showed identical sequences in bthe human fetal erythroblasts and in the fhybrid. Both members of pair 28 of Table 1 codfor human proteosome subunit X. Both membof pair 6 of Table 1 coded for the same sequewhich did not have homology with any sequedeposited in GenBank.

Isolation and Sequence Analysis of the Full-Length FGIF cDNA

The cDNA of pair number 6 by being novand present in cells expressing fetal globinabsent in adult erythroblasts and the adult hycould potentially be coding for ag gene activato

nd therefore it was investigated further. Becaf results obtained with transfection of adultrids (see later), this new cDNA was conside

o potentially increaseg expression in the hybridand it was preliminarily called FGIF (for fetglobin increasing factor).

The full-length FGIF cDNA was isolated froa human fetal liver cDNA library. The most9fragment was cloned from human fetal livmRNA by 59-RACE using the Marathon cDNamplification kit (Clontech, CA). Inspectionthe sequence (Fig. 1) revealed a potential oreading frame of 239 amino acids. The preseof three in-frame stop codons immediatelystream of predicted methionine suggested thaopen reading frame is the major translation puct. Analysis of the deduced amino acid sequeshowed no significant homology with anyquence in protein databases except for the pence of three ankyrin repeats (Fig. 2). This remotif was first identified in two cell cycle-reglatory proteins of yeast, CDC10 (21) and SW(22), and has also been found in several oproteins which play roles in cell determinatand cell cycle control, including theb subunit oheteromeric DNA-binding protein GABP (23the NfkB/KBF-1 transcription factor (24), th

rotein encoded by the sex-determining gem-1 (25),lin-12 (26), glp-1 (27) ofC. elegans
the Notch (28) of Drosophila and the product of

L

L

L

L

Ln

L

r

FL

FL

FL

FL

FL

L

FL



TABLE 1

Results of Sequencing of Pairs of cDNA Fragments Cloned by Differential Display

1FL New 8FL Human MREII homolog MRNAAccession No. Hsu37359

16FL New

1H New * different from 1FL 8H New * different from 8FL 16H New * different from 16F2FL New 9FL New 17FL New2H New * different from 2FL 9H New * different from 9FL 17H New * different from 17F3FL H. sapiensSOD-2 gene for

manganese superoxide dismutaseAccession No. HSMNSODR

10FL New 18FL New

3H New * different from 3FL 10H New * different from 10FL 18H New * different from 18F4FL New 11FL New 19FL New4H New * different from 4FL 11H New * different from 11FL 19H Mouse mRNA for

nonmuscle tropomycin 5Accession No. mmtropm5

5FL H. sapienspartial cDNA sequenceAccession No. N52610

12FL New 20FL New

5H New * different from 5FL 12H New * different from 12 FL 20H New * different from 20F6FL New 13FL New 21FL New6H New SAME AS 6FL 13H New * different from 13FL 21H New * different from 21F7FL New 14FL New 22FL Human DNA sequence o

chromosome XAccession No. hsu212c1

7H Mouse embryonal carcinoma F9 cellcDNA

Accession No. Mus62a03* different from 7FL

14H New * different from 14FL 22H New * different from 22F

15FL New 23FL New15H Mus domesticushybrophobic

Protein mRNAAccession No. Musmthypa

23H New * different

24FL New 33FL New 42FL Human ALAS2 mRNA foD-aminolevulinatesynthesase (erythroid)

Accession No. hsalas2r24H New * different from 24FL 33H New * different from 33FL 42H New * different from 4225FL New 34FL Human replication protein A 70-kDa

subunit mRNAAccession No. humrpa 70kd

43FL New

25H New * different from 25FL 34H New * different from 34FL 43H New * different from 4326FL New 35FL New 44FL Human metastasis-

associated mRNAAccession No. hsu35113

26H New * different from 26FL 35H New * different from 35FL 44H New * different from 4427FL New 36FL New 45FL New27H New * different from 27FL 36H New * different from 36FL 45H New * different from 4528FL Human mRNA for proteasome

subunit XAccession No. Humpsx1

37FL New 46FL New

28H Human mRNA for proteasomesubunit X

Accession No. Humpsx1Same AS 28FL

37H New * different from 37FL 46H Mouse endogenousretroviral intracisternalA-partical mRNA

Accession No. musiapa29FL New 38FL New 47FL New29H New * different from 29FL 38H New * different from 38FL 47H New * different from 4730FL New 39FL New 48FL New30H Mus musculusautoantigen La (ss-B)

mRNAAccession No. muslassb

39H New * different from 39FL 48H New * different from 48F

31FL New 40FL New 49FL New31H New * different from 31FL 40H New * different from 40FL 49H New * different from 4932FL New 41FL New 50FL New
32H New * different from 32FL 41H New * different from 41FL 50H New * different from 50FL
5

rN-

es.ac-

thethaene

e ofan

theper-as-

ers5urR isig.edheig.ter-A

. Inn oftaltoersed

he threei consensusa


mouse mammary tumor geneint-3 (29). Anothefeature of FGIF is that the sequence at theterminus is rich in acidic amino acid residuMany transcriptional factors contain an acidictivating domain.

We also cloned the mouse homologue ofhuman FGIF gene. Sequence analysis showedthe sequences of human and murine FGIF gare more than 90% identical in DNA sequencthe protein coding region. The murine and humproteins are 88% identical.

Tissue Distribution of FGIF

To examine the expression pattern ofFGIF gene in human and mouse tissues, weformed Northern blot analysis and RT-PCR

FIG. 1. Nucleotide sequence of FGIF cDNA andn-frame stop codons are in italicized typeface. The thmino acids in these repeats are in boldface.

says. The location in the FGIF cDNA of the

6

t

probes for the Northern blot assay and the primfor RT-PCR are shown in Fig. 3A. Both a9probe and a 39 probe were used to confirm oresults. The sequence of the primer for RT-PCidentical in human and mouse. As shown in F3B, FGIF is widely expressed in all examinhuman fetal tissues (transcript A of Fig. 3B). Thybridization signal marked as transcript B in FB perhaps derives from alternative splicing, alnative polyadenylation or from another mRNspecies having high homology with the probethe mouse, we detected high level expressioFGIF in fetal liver, fetal brain and day 13 felung (Fig. 3C). At day 16, we were still abledetect high-level expression of FGIF in fetal livwhile expression in brain and lung was decrea

educed amino acid sequence of the FGIF protein. TDC10/SWI6/ankyrin repeats are underlined and the

the dree C

significantly compared to that of day 13. In the

d intoge.

rip-53,on-althegormd in

y-m ty

the

l 6428-se-y.ans-tedon

an-sed

idaseL4ade-ceiche

/ankysin-C10/

s. Ide



adult mouse, no FGIF expression was detecteliver, brain or lung. Also, we were not abledetect FGIF in blood at any developmental sta

FGIF Is a Transcriptional Activator

To test whether FGIF acts as a transctional activator, we used a yeast strain Y1which contains the LacZ reporter gene ctrolled under the GAL4 UAS and a minimpromoter. An expression plasmid in whichFGIF was fused with the GAL4 DNA bindindomain was constructed and used to transfY153 yeast cells. Transformants were platea Trp2 medium. Individual Trp1 colonieswere analyzed using theb-galactosidase enz

atic activity assay.b-galactosidase activiwas normalized for protein concentration in

FIG. 2. Consensus alignment of the CDC10/SWI6bearing proteins. (A) Alignment of the three CDC10/SSWI6/ankyrin consensus sequence in different protein

yeast lysate. As shown in Fig. 4A, the whole

7

FGIF sequence, or the truncated N-terminaamino acid sequence, or the N-terminal 1amino-acid or the C-terminal 73-amino-acidquence increasedb-galactosidase activitThese results suggested that FGIF is a tractivator and the activating domain is locawithin the N-terminal 64 amino acids regiwhich is rich in acidic amino acids.

To determine whether FGIF can activate trscription in a mammalian cell system, we utransfections of COS cells. A reporter plasmcarrying the chloramphenicol acetyltransfer(CAT) gene was used. In this plasmid, five GAbinding sites were inserted upstream of thenovirus E1b TATA box (30). The FGIF sequenwas placed into the vector pSG424 (30), in whthe GAL4 DNA binding domain is under th

rin repeat sequence of FGIF and other CDC10/SW16ankyrin motifs found in FGIF. (B) Alignment of the CDntical amino acid residues are shadowed.

/ankyWI6/

control of the SV40 early promoter. Transient

OSole

atingted

The 51 codings bp to 735b

df lot. Twom me froma mRNAe smal trol.


transfection assays were then performed in Ccells. As shown in Fig. 4B, presence of the wh

FIG. 3. Tissue distribution of FGIF expression. (A36-bp sequence contains the 59-UTR region. The sequequence. The sequence from 137 to 329 bp codes fop codes for the three CDC/SW16/ankyrin repeats. Th9

and theClaI site. The 39 probe used for Northern blot isor RT-PCR are shown by the two bold arrows. (B) FGajor transcripts were detected, transcript A, with explternative splicing or may represent cross-hybridizatxpression in murine tissues by RT-PCR. The murine

FGIF sequence increased CAT activity 22-fold,

8

suggesting that FGIF possesses trans-activactivity in a mammalian cell assay. The trunca

ematic representation of the structure of FGIF cDNA.9from 137 bp to 852 bp corresponds to the protein-

N-terminal acidic domain while the sequence from 439used for Northern blot is located between the 59 terminus

ed between theEcoRV site and the 39 terminus. Primers useRNA expression in human fetal tissues by Northern b

size of about 1.3 kb, and transcript B, which may cof the probe with another homologous gene. (C) FGIFl ribosomal protein 14 gene mRNA was used as a con

) Schencer thee 5probelocatIF mectedion o

N-terminal 64 amino acid sequence increased

lyveids)edn to

g manf -p ing

IFe earc e

F thep ro-m wass tea m.T

heseer-

-ub-x-tion,

a d-i res -

ay. Filledb orter genee s representt



CAT activity 76-fold, indicating that the highacidic N-terminus of FGIF (with 18 negaticharged amino acids in the first 64 amino accontains an activating domain. Similarly chargregions of other proteins have also been showbe effective transcriptional activators (31).

FGIF Increasesg Gene Expression in StablyTransfected, Nearly Switched HFE3 MELHybrids

To test whether FGIF can activateg-globinene expression, we used transfections of hu

etal erythroid (HFE)3 MEL hybrids. The hyothesis was that if FGIF is capable of influenc

g gene transcription, transfection with an FGxpression plasmid in hybrids which were nompletion of theirg to b switch would delay th

FIG. 4. (A) FGIF fused with Gal4 binding domain acoxes represent the Gal4 binding domain and open bxpression in transiently transfected COS cells. The fi

he FGIF domains.

switch or increase the level ofg gene expression. fi

9

or these experiments the CMV promoter ofcDNA3 vector (which contains an SV40 poter driving the neomycin resistance gene)

ubstituted by amLCR-b globin promoter cassetnd the FGIF cDNA was cloned downstreahus, FGIF is under the control of themLCR-b

globin promoter and should be expressed in thybrids. The following experiments were pformed:

1. An HFE 3 MEL cell hybrid at an advanced stage of switching was obtained by scloning of a switching hybrid. Human globin epression was assessed after chemical inducby immunofluorescence staining with anti-g and

nti-b human globin specific monoclonal antiboes. Only 4.2% of the cells of this hybrid wetained with the anti-g antibody (Table 2), con

s reporter gene expression in yeast transactivation assrepresent the FGIF domains. (B) FGIF activates rep

oxes represent the pSG424 vector and the open boxe

tivateoxes

lled b

rming that the hybrid was near the end of itsg to

rebyIFan

Nase. As

rolheterntedhens-

nuc-ixithols

r 25ase17,%e-atsix

e %.

seda beae ana hy-b , ah

lyas-terto18

12,, anin-anse

p thecfl ondt allg thatn y-b B),i dw in

s-trol(

six

d


b switch. Pools of cells from this hybrid wetransfected with the FGIF and control plasmidslipofection. 2.5 weeks after transfection, the FGand the control pools were induced and humglobin gene expression was assessed by Rprotection and by immunofluorescent stainingshown in Table 2, the frequency of theg positivecells in the hybrid transfected with the contplasmid was 3.9% while it was 10.8% in thybrid transfected with the FGIF plasmid. Afinduction with HMBA and hemin, the perceg/g 1 b mRNA was 6.6 in the untransfecthybrid, 11.6 in the hybrid transfected with tcontrol plasmid; and 41.3 in the hybrid trafected with the FGIF plasmid.

2. To test whether the higherg expression ithe hybrid transfected with FGIF was reprodible, one hybrid cell pool was divided in ssubpools, which were individually lipofected wthe FGIF plasmid. The six transfected subpoand the untransfected control were cultured foweeks. Globin gene expression studies by RNprotection assay were done in weeks 2, 11,and 25. As shown in Figs. 5A and 5B, theg/g 1 b mRNA of the untransfected control dclined from a level of 18.9% at week 2 to 2.7%week 17 and 4.1% at week 25. In contrast, theFGIF transfected hybrids sustainedg-globin geneexpression so that after 25 weeks in culture, thg/g 1 b mRNA values ranged from 14 to 46%

3. Since a single untransfected pool was us a control in the previous experiment, it couldrgued that the observed loss ofg-globin genexpression over time in culture was due toccelerated rate of switching inherent to therid used as control. To test this possibility

TABLE 2

g-Positive Cells andg/g1 b mRNA Ratios in Hybrids,Transfected Hybrids, and Controls

Immunofluorescent %g/g1 b mRNA

% g1 cells %b1 Uninduced Induce

Control plasmid 3.9 76.8 25.1 11.6FGIF plasmid 10.8 64.5 35.1 41.3

ybrid expressing mainly humanb-globin was

10

divided into two pools which were individuallipofected with either the experimental FGIF plmid or with the control plasmid. Two days aflipofection each lipofected pool was divided insix subpools which were cultured under G4selection for a total of 21 weeks. At 2, 5, 9,and 21 weeks after the stable transfectionaliquot of cells of the growing hybrids wasduced for 3 days by HMBA and hemin and humg andb globin mRNA was quantitated by RNa

rotection assay. As shown in Fig. 6A, inontrol hybrids the percentg/g 1 b mRNA ratiosuctuated around a value of 10% from the seco the twelfth week. However, by week 21lobin gene expression was extinguished soo globin mRNA was detectable. Of the six hrids transfected with the FGIF cassette (Fig. 6

n three the %g/g 1 b mRNA ratios declineith time and by week 21 of culture all glob

FIG. 5. (A) gmRNA expression in six hybrids tranfected with FGIF plasmid (solid circles) and one con(open circles). (B) Mean and standard deviation of %g/g 1 b) mRNA levels. Open bar: control. Solid bar: the
FGIF transfected hybrids.

theere. In

eks2%

ed

ionids,ive.ithas

n ase ofc sf

GIFc lyt thes llss d

asdi-leterolsiume

cedionffer-d

b l hy-b

D

thes n.T orks talh miac e-m heys ).T ish llsf andc itchw thep hy-b e

theh s as edw theM eh ns-f hatpe

y-sA.

three



gene expression has been extinguished. Inremaining three FGIF transfected hybrids thwas no extinction of globin gene expressiontwo of these hybrids the percentg/g 1 b ratioremained about 15% throughout the first 12 weof culture, and by week 21, it was 14 and 2respectively. The level ofg mRNA in the thirdhybrid initially declined, but after week 9 startincreasing so that by week 21 the % of (g/g 1 b)mRNA ratio was 12%. Because of the extinctof globin gene expression in the control hybrthe results of this experiment were inconclusHowever, the fact that in the three hybrids wactive humanb locus theg gene expression h

ot turned off by week 21 of culture but it wssentially identical to that in the first weekulture suggest that theg gene in the three hybrid

FIG. 6. g/(g 1 b) mRNA ratios in human fetal erthroid 3 MEL cell hybrids. (A) Data in six control hybridand in B results in six hybrids transfected with FGIF cDNNotice that three of these hybrids have elevatedg/g 1 bmRNA ratios at 21 weeks in culture. Globin (g or b) mRNAwas not measurable in the six controls and the otherFGIF-transfected hybrids.

ailed to undergo the expectedg to b switch.

11

4. We subsequently examined whether Fould alter the rate of switching if it was stabransfected in a hybrid at an earlier stage ofwitch. A hybrid in which 38.5% of the cetained positively forg-globin and 45.7% staine

positive for b-globin was used. The hybrid wdivided into three pools. Two of these were invidually lipofected with the FGIF plasmid whithe third with the control plasmid. Two days aflipofection each pool was divided in 2 subpoand cultured in the presence of selective med(G418 at 700mg/ml) for a total of 20 weeks. Oncevery 4 weeks an aliquot of cells was induwith HMBA/hemin and globin gene expressassessed by RNAse protection assay. No dience in the rate ofg to b switch was observe

etween the FGIF transfected and the controrid cell pools.

ISCUSSION

In this study we used a novel approach inearch of potential inducers ofg gene expressiohis approach is based on our previous whowing that hybrids produced by fusing feuman erythroblasts with murine erythroleukeells initially produce almost exclusively fetal hoglobin and after 20 to 40 weeks in culture t

witch to exclusiveb globin gene expression (11he rate of switch in each of these hybridsighly reproducible. Thus when an aliquot of ce

rom such a hybrid is frozen soon after fusionultured at a later time, it displays a rate of swhich is essentially superimposable to that ofarent hybrid. The mechanism whereby theserids undergo theirg to b switch is unclear. Th

switch could be controlledin cis if the chromatinof theb locus of the human chromosome 11 of

ybrid contains a structure that determinewitching program that is faithfully executhen the chromosome 11 is transferred intoEL cells. Trans control could imply that thuman chromosomal component that is tra

erred into the MEL cells contains a locus troduces trans elements which supportg genexpression. In that case, theg to b globin gene

switch in these hybrids could reflect the extinction

ofue,-avesent

i -

asedandasge.eenthe

insts.romm-tald ofec-s-allyetalandialint-urichts,rityat

pre-bersse-thesene.hichtor.(i)N-a-nalP-asainsam-

icD-lianthatent

lextionts

or.eenodes,

re-I6/in

es-ins

thatithedi-ctthis

ingco-NA-

tiesvelc-

i o-g orm yi c-i atf tionc e

the. In

t ati byd yi re

eteds-

hat,-n


of expression from this locus. If the hypothesistrans control of switching in these hybrids is trtrans-acting factor(s) that activateg gene expression should be present in the hybrids that hexclusiveg gene expression and should be abn the hybrids that have exclusiveb gene expression.

The experimental system we used was bon this hypothesis. A single hybrid was usedcell aliquots were collected when the hybrid win its fetal stage and when it was in its adult staThe putative trans factors should have bpresent not only in the fetal hybrid but also infetal-origin erythroblasts and absent not onlythe adult hybrid but also in the adult erythroblaTherefore we collected such erythroblasts ffetal and adult erythroid cell cultures and copared their mRNA composition to that of the feand adult hybrids. As a screening, the methodifferential mRNA display was used. Our exptation was that the putativeg gene inducing tranacting elements will be present among identicsized amplicons present in fetal liver and the fhybrid and absent in the adult erythroblaststhe adult hybrids. The yield of the differentmRNA display methods was however disappoing. Thus, 96% (48 of 50) of the pairs fitting oselection criteria were composed of clones whalthough had identically sized cDNA fragmenthey had different DNA sequences. The majoof the cloned cDNA belonged to genes whichthe time of sequencing were considered to resent not previously identified genes. The memof only two of the 50 pairs we cloned andquenced had identical sequences and one ofhad a sequence of a previously unidentified g

The new gene encodes a novel protein whas the characteristics of a transcriptional facThis factor contains two functional domains.An acidic activation domain located in theterminal portion of the molecule. Acidic activtion domains are present in various transcriptiofactors, such as GAL4 (32), GCN4 (33, 34), A1/Jun (35, 36), YY1 (37), and VP16 (38). It hbeen proposed that the acidic activation domfacilitate transcription initiation by interacting inrelatively nonspecific manner with a general co
ponent of the transcription initiation complex, w
12

e

such as TFIID or pol II itself (39). Some acidactivation domains function by stabilizing TFIIpromoter interactions in yeast and mammacells (40). Recently, it has been reportedacidic activation domains mediate the recruitmof the SWI/SNF chromatin remodeling compto a target promoter and stimulate transcrip(41). (ii) Three CDC10/SWI6/ankyrin repeawhich contribute almost half of the new factThe CDC10/SWI6/ankyrin repeats have bfound in diverse species such as yeast, nematinsects and mammals, indicating that thesepeats are of functional importance. CDC10/SWankyrin motifs have been shown to be involvedprotein-protein interactions (42–44). The prence of CDC10/SWI6/ankyrin repeats in proteof so diverse species raise the possibilitythese proteins fulfill their roles by interacting wother factors and these interactions may be mated by the CDC10/SWI6/ankyrin motif. The fathat the deduced amino acid sequence offactor does not show any potential DNA binddomain may indicate that it functions as aactivator rather than a sequence-specific, Dbinding factor.

To test whether this factor has the properof a fetal hemoglobin inducer, we used a noassay ofin vivo g gene activation, i.e., the capaty of a potential activator to induce fetal hemlobin in our hybrids. A transcriptional factay affect theg to b switch either negatively, b

nhibiting g gene silencing or positively by indung g gene transcription. It is therefore likely thactors that increase fetal hemoglobin producan act on the proximalg gene promoter, or on th

distal promoter or may affect the interaction ofg gene promoter with the locus control regionhe HFE3 MEL cell hybrid system factors thncreaseg gene expression may function eitherelaying theg to b switch in the hybrids or b

ncreasingg-globin expression in hybrids that anear the end of their switch or by reactivatingggene expression in hybrids that have compltheir switch. The FGIF cDNA was stably tranfected into our hybrids under the expectation tif the factor inducedg gene transcription or inhibited silencing in that system,g gene expressio

ill continue in these hybrids beyond the ex-

ofec-andtor-

p hirde IF

The-

g-

se

vELfec-usns ton

e emwb chrom

thisto

a rry-i eI LFa ve

emrip-.we

ing.

f

nt tivityo lsoo -

rumpect

ay

c theh

99a

) He-dPerl-35–

oidto

er,iesisF

95)roid

cti-

the

g, J.,an

. M.is

biq-tally

, G.i-.

, G.l

ene.

nno-bin-

dis-the



pected time of switch of these hybrids. Threethe four experiments we did fulfilled these exptations. In the experiments shown in Table 2Fig. 5, hybrids transfected with the new faccontinued to have significantly higherg gene ex

ression compared to the controls. In the txperimentg gene expression continued in FG

transfected hybrids after 21 weeks in culture.positive results in thesein vivo experiments sug

est that the new factor can increaseg-globingene expression either by inducingg gene transcription or by delaying the process ofg geneilencing in these hybrids.

There are very few assays for assessing thinivo effects of factors that are potentialg gene

activators. Stable transfections of K562 or Hcells can provide information whether a transtant can induce transcription of the endogenoggenes but the relevance of these observatiothe in vivo g gene control in primary huma

rythroid cells is unclear. The hybrid cell syste have used is closer to thein vivo situationecause expression of the genes of a humanosome that undergoes the developmentalg to b

switch is assessed. From this point of viewassay is closer to thein vivosituation compared

ny other assay except for transgenic mice cang the whole humanb locus. As mentioned in thntroduction, three potential genes, NFE4, FKnd FKLF2, that activateg gene expression ha

been described, but this assay has not beenployed before for testing any of these transctional factors forin vivo activation of g genesTherefore, definitive evidence that the assayhave used functions as predicted is still lackWe have previously found that HFE3 MEL cellhybrids growing continuously in the presence oaaminobutyric acid display a very prolongedg to bswitch (45), indicating thatg gene transcription ihese hybrids can be sustained through the acf factors that affect chromatin acetylation. Aur previous work has shown thatg gene expres

sion in the hybrids can be modulated by sefactors (45). It is reasonable, therefore, to exthat the forced expression of factors that maffect negatively or positively theg to b switchan influence accordingly the phenotype of
ybrids.
13

-

-

ACKNOWLEDGMENTS

This study was supported by NIH Grants HL208nd DK30852.

REFERENCES

1. Stamatoyannopoulos, G., and Grosveld, F. (2001moglobin switching.In The Molecular Basis of BlooDiseases (Stamatoyannopoulos, G., Majerus, P.,mutter, R., and Varmus, H., Eds.), 3rd ed., pp. 1182. Saunders, New York.

2. Miller, I. J., and Bieker, J. J. (1993) A novel, erythrcell-specific murine transcription factor that bindsthe CACCC element and is related to the Kru¨ppelfamily of nuclear proteins.Mol. Cell. Biol. 13, 2776–2786.

3. Nuez, B., Michalovich, D., Bygrave, A., PloemachR., and Grosveld, F. (1995) Defective haematopoin fetal liver resulting from inactivation of the EKLgene.Nature375,316–318.

4. Perkins, A. C., Sharpe, A. H., and Orkin, S. H. (19Lethal-thalassaemia in mice lacking the erythCACCC-transcription factor EKLF.Nature375,318–322.

5. Asano, H., and Stamatoyannopoulos, G. (1998) Avation ofb-globin promoter by erythroid Kru¨ppel-likefactor.Mol. Cell. Biol. 18, 102–109.

6. Dillon, N., and Grosveld, F. (1991) Humang-globingenes silenced independently of other genes inb-globin locus.Nature350,252–254.

7. Stamatoyannopoulos, G., Josephson, B., Zhanand Li, Q. (1993) Developmental regulation of humg-globin genes in transgenic mice.Mol. Cell. Biol.13,7636–7644.

8. Jane, S. M., Nienhuis, A. W., and Cunningham, J(1995) Hemoglobin switching in man and chickenmediated by a heteromeric complex between the uuitous transcription factor CP2 and a developmenspecific protein.EMBO J.14, 97–105.

9. Asano, H., Li, X. S., and Stamatoyannopoulos(1999) FKLF, a novel Kru¨ppel-like factor that actvates human embryonic and fetalb-like globin genesMol. Cell. Biol. 19, 3571–3579.

10. Asano, H., Li, X. S., and Stamatoyannopoulos(2000) FKLF-2: A novel Kru¨ppel-like transcriptionafactor that activates globin and other lineage gBlood 95, 3578–3584.

11. Papayannopoulou, T., Brice, M., and Stamatoyapoulos, G. (1986) Analysis of human hemogloswitching in MEL 3 human fetal erythroid cell hybrids.Cell 46, 469–476.

12. Liang, P., and Pardee, A. B. (1992) Differentialplay of eukaryotic messenger RNA by means of
polymerase chain reaction.Science257,967–971.

89)d.

NA

ved

.,rch

rid-ns.

1 A.ted

1 N.,enicbin

2 N.,C.,ngb-

, P.on-

’

on-gu-

ton,ofro-

2 ve,y,NA1

2 90)ningtrol

2 Thes-h-

2 lin-onsins.

2 ta-fromductGF-

2 alla-: A

am-

3 otenttein.

3 ec-in,

3 ran-f a

oticr to

88)hortN4

T.,co-uc-tor

rateast.

95)the

ght,ns-

arly

nalcific

e-astm-

M.


13. Sambrook, J., Fritsch, E. F., and Maniatis, T. (19Molecular Cloning: A Laboratory Manual, 2nd eCold Spring Harbor Laboratory Press, New York.

14. Sanger, F., Nicklen, S., and Sorge, J. A. (1977) Dsequencing with chain-terminating inhibitors.Proc.Natl. Acad. Sci. USA74, 5463–5467.

15. Pearson, W. R., and Lipman, D. J. (1988) Improtools for biological sequence comparison.Proc. Natl.Acad. Sci. USA85, 2444–2448.

16. Altschul, S. F., Gish, W., Miller, W., Myers, E. Wand Lipman, D. J. (1990) Basic local alignment seatool. J. Mol. Biol. 215,403–410.

17. Pelle, R., and Murphy, N. B. (1993) Northern hybization: rapid and simple electrophoretic conditioNucleic Acids Res.21, 2783–2784.

8. Ito, H., Fukuda, Y., Murata, K., and Kimura,(1983) Transformation of intact yeast cells treawith alkali cations.J. Bacteriol.153,163–168.

9. Li, Q., Clegg, C., Peterson, K., Shaw, S., Raich,and Stamatoyannopoulos, G. (1997) Binary transgmouse model for studying the trans control of glogene switching: Evidence that GATA-1 is anin vivorepressor of humane gene expression.Proc. Natl.Acad. Sci. USA94, 2444–2448.

0. Akiyama, K., Yokota, K., Kagawa, S., Shimbara,Tamura, T., Akioka, H., Nothwang, H. G., Noda,Tanaka, K., and Ichihara, A. (1994) cDNA cloniand interferong down-regulation of proteasomal suunits X and Y.Science265,1231–1234.

21. Aves, S. J., Durkacz, B. W., Carr, A., and Nurse(1985) Cloning, sequencing and transcriptional ctrol of the Schizosaccharomyces pombecdc10 ‘startgene.EMBO J.4, 457–463.

22. Breeden, L., and Nasmyth, K. (1987) Cell cycle ctrol of the yeast HO gene: cis- and trans-acting relators.Nature48, 389–397.

23. LaMarco, K., Thompson, C. C., Byers, B. P., WalE. M., and McKnight, S. L. (1991) IdentificationEts- and Notch-related subunits in GA binding ptein. Science253,789–792.

4. Kieran, M., Blank, V., Logeat, F., VandekerckhoJ., Lottspeich, F., Bail, O. L., Urban, M. B., KourilskP., Baeuerle, P. A., and Israel, A. (1990) The Dbinding subunit of NF-kB is identical to factor KBFand homologous to the rel oncogene product.Cell 62,1007–1018.

5. Spence, A. M., Coulson, A., and Hodgkin, J. (19The product of fem-1, a nematode sex-determigene, contains a motif found in cell cycle conproteins and receptors for cell-cell interaction.Cell 60,981–990.

6. Yochem, J., Weston, K., and Greenwald, I. (1988)Caenorhabditis eleganslin-12 gene encodes a tranmembrane protein with overall similarity to Drosop
ila Notch.Nature335,547–550.
14

7. Yochem, J., and Greenwald, I. (1989) glp-1 and12, genes implicated in distinct cell–cell interactiin C. elegans,encode similar transmembrane proteCell 58, 553–563.

8. Wharton, K. A., Johansen, K. M., Xu, T., and Arvanis-Tsakonas, S. (1985) Nucleotide sequencethe neurogenic locus Notch implies a gene prothat shares homology with proteins containing Elike repeats.Cell 43, 567–581.

9. Robbins, J., Blondel, B. J., Gallahan, D., and Chan, R. (1992) Mouse mammary tumor gene int-3member of the notch gene family transforms mmary epithelial cells.J. Virol. 66, 2594–2599.

0. Fields, S., and Jang, S. K. (1990) Presence of a ptranscription activating sequence in the p53 proScience249,1046–1049.

1. Hope, I. A., and Struhl, K. (1986) Functional disstion of a eukaryotic transcriptional activator proteGCN4 of yeast.Cell 46, 885–894.

2. Brent, R., and Ptashne, M. (1985) A eukaryotic tscriptional activator bearing the DNA specificity oprokaryotic repressor.Cell 43, 729–736.

33. Hope, I. A., and Struhl, K. (1987) GCN4, a eukarytranscriptional activator protein, binds as a dimetarget DNA.EMBO J.6, 2781–2784.

34. Hope, I. A., Mahadevan, S., and Struhl, K. (19Structural and functional characterization of the sacidic transcriptional activation region of yeast GCprotein.Nature333,635–640.

35. Bohmann, D., Bos, T. J., Admon, A., Nishimura,Vogt, P. K., and Tjian, R. (1987) Human proto-ongene c-jun encodes a DNA binding protein with strtural and functional properties of transcription facAP-1. Science238,1386–1392.

36. Struhl, K. (1988) The JUN oncoprotein, a vertebtranscription factor, activates transcription in yeNature332,649–650.

37. Bushmeyer, S., Park, K., and Atchison, M. L. (19Characterization of functional domains withinmultifunctional transcription factor, YY1.J. Biol.Chem.270,30213–30220.

38. Triezenberg, S. J., Kingsburg, R. C., and McKniS. L. (1988) Functional dissection of VP16, the traactivator of herpes simplex virus immediate egene expression.Genes Dev.2, 718–729.

39. Mitchell, P. J., and Tjian, R. (1989) Transcriptioregulation in mammalian cells by sequence-speDNA binding proteins.Science245,371–378.

40. Horikoshi, M., Carey, M. F., Kakidani, H., and Roder, R. G. (1988) Mechanism of action of a yeactivator: Direct effect of GAL4 derivatives on mamalian TFIID-promoter interactions.Cell 54, 665–669.

41. Armstrong, J. A, Bieker, J. J., and Emerson, B.
(1998) A SWI/SNF-related chromatin remodeling

an-

Then

withon-

nding

H

H.,ainsdon.

Pa-odu-



complex, E-RC-1, is required for tissue-specific trscriptional regulation by EKLFin vitro. Cell 95, 93–104.

42. Siegmund, R. F., and Nasmyth, K. A. (1996)Saccharomyces cerevisiaestart-specific transcriptiofactor swi4 interacts though the ankyrin repeatsthe mitotic clb2/cdc28 kinase and though its cserved carboxyl terminus with swi6.Mol. Cell. Biol.16, 2647–2655.

43. Klein, E. S., Simmons, D. M., Swanson, L. W., aRosenfeld, M. G. (1993) Tissue-specific RNA splic
generates an ankyrin-like domain that affects the
15

dimerization and DNA-binding properties of a bHLprotein.Genes Dev.7, 55–71.

44. Sawa, C., Goto, M., Suzuki, F., Watanabe,Sawada, J., and Handa, H. (1996) Functional domof transcription factor hGABPb1/E4TF1-53 requirefor nuclear localization and transcription activatiNucleic Acids Res.24, 4954–4961.

45. Zitnik, G., Li, Q., Stamatoyannopoulos, G., andpayannopoulou, Th. (1993) Serum factors can mlate the developmental clock ofg- to b-globin geneswitching in somatic cell hybrids.Mol. Cell. Biol. 13,
4844–4851.

Documents

Cloning and Characterization of a Potential Transcriptional Activator of Human γ-Globin Genes