Identification of c fos responsive elements downstream of TAR in the long terminal repeat of human immunodeficiency virus type 1

Identification of c-fos-responsive elements

downstream of TAR in the long terminal repeat

of human immunodeficiency virus type-1.

K A Roebuck, … , D A Brenner, M F Kagnoff

J Clin Invest. 1993;92(3):1336-1348. https://doi.org/10.1172/JCI116707.

Activation of HIV-1 requires the binding of host cell transcription factors to cis elements in the proviral long terminal repeat (LTR). This study identifies c-fos-responsive sequence motifs in the U5 transcribed noncoding leader sequences downstream of the viral transactivator responsive (TAR) element. These DNA sequence motifs are the most downstream regulatory elements described thus far in the HIV-1 LTR. Functional studies, using human colon epithelial cell lines, demonstrate that the downstream elements are transactivated by expression of the c-fos protooncogene and can transmit PMA and TNF alpha activation signals to the viral LTR. Moreover, the c-fos-responsive elements mediate HIV-1 LTR transcription independent of Tat and the NF kappa B-binding enhancer element. Nuclear extracts of colon epithelial cells form distinct gel mobility shift complexes with the c-fos-responsive elements. These complexes comigrate with a gel shift complex formed on a classical CRE oligonucleotide and are competed by CRE oligonucleotides. These data indicate that the HIV-1 LTR contains previously unrecognized functional DNA cis-regulatory elements downstream of TAR in the transcribed noncoding 5' leader sequence and suggest that early response genes such as c-fos play a role in the activation of HIV-1 gene

expression.

Research Article

Find the latest version:

http://jci.me/116707/pdf

Identification of

c-fos-responsive

Elements Downstream of

TAR in

the

Long

Terminal Repeat

of

Human

Immunodeficiency

Virus

Type-1

Kenneth A. Roebuck, DavidA.Brenner,and Martin F.Kagnoff

Department ofMedicine, University of California, SanDiego,LaJolla, California92093-0623

Abstract

Activation ofHIV-1 requires the binding ofhost cell transcrip-tion factorstocis elements in the proviral long terminalrepeat

(LTR). This study identifies

c-fos-responsive

sequencemotifs in the U5transcribed noncodingleadersequencesdownstream of the viraltransactivator responsive (TAR) element. These DNAsequencemotifsarethemostdownstream regulatory

ele-mentsdescribed thus far in theHIV-1 LTR.Functionalstudies,

using human colonepithelial celllines, demonstrate that the downstreamelements aretransactivated byexpression of the

c-fos

protooncogene andcantransmitPMA andTNFa activa-tionsignalstotheviral LTR.Moreover, the

c-fos-responsive

elementsmediate HIV-1LTRtranscriptionindependentof Tat and the

NFiB-binding

enhancer element. Nuclearextractsof colonepithelial cells form distinct gelmobilityshiftcomplexes

with the

c-fos-responsive

elements. These complexes

comi-gratewithagel shift complex formedonaclassical CRE

oligo-nucleotide andarecompeted byCREoligonucleotides.These dataindicate that the HIV-1 LTR contains

previously

unrecog-nized functional DNAcis-regulatoryelements downstream of TAR inthetranscribednoncoding5' leadersequenceand sug-gestthatearlyresponse genessuchas

_c-fos

playarole in the activation ofHIV-1 gene expression. (J. Clin. Invest. 1993.

92:1336-1348.)Key words:oncogenes *transcriptionfactors .

epithelialcells *activator

protein-i

* untranslated 5' leader

Introduction

HIV-1 is the etiological agent ofAIDS (1, 2). After HIV-1

integrates into the host genome as latent proviral DNA, a

threshold burst ofgenomic length transcriptionisrequiredto

complete theviral life cycle and produce progeny virus (3). Thisinitial transcription oftheHIV- 1 genomeis controlledby

the5'longterminalrepeat

(LTR)'

andisdependentuponhost

cellulartranscriptionfactorsbindingto aseries of DNA

regula-Address correspondence andreprintrequeststoMartin F. Kagnoff,

M.D.,Department ofMedicine, 0623D, UniversityofCalifornia,San

Diego,9500GilmanDrive,LaJolla,CA 92093. K.A.Roebuck's

pres-entaddressis the Department ofImmunology/Microbiology,

Rush-Presbyterian-St.Luke's MedicalCenter,1653 WestCongress Parkway, Chicago,IL60612. D. A. Brenner's present address is theDepartment

ofMedicine, University of North Carolina, CB #7080, Room 326,

Burnett-WomackBuilding, Chapel Hill,NC27599.

Receivedfor publication 7December 1992 andinrevisedform14

April1993.

tory elements in the LTR promoter(4).Transcriptional stud-ies of the HIV-1 genomeindicate that the HIV-l LTR

pro-motercanbe subdivided into threefunctionaldomains: abasal corepromoter,which is comprised of thetranscriptional start

site, aTATAbox, and three tandem Spl binding sites

(GC-boxes);anupstream enhancer element containingtwo

adja-cent _binding sites forthe inducible transcriptional activator

nuclear factorkappaB(NFKB);andafarupstream modula-tory region (4, 5). The HIV-l LTR promoter also contains

several recognition sites for nucleic acid binding proteins

downstreamofthetranscriptionalstartsite(6). However,with theexceptionofthe transactivatorresponsive (TAR)element,

which,upontranscriptionintoaspecific RNAstructure,binds

the viraltransactivator protein Tat, the importance of these downstreamsequence motifs ortheir role inHIV-1 gene

ex-pressionhasnotbeendemonstrated.

HIV- 1 isassociatedwithdisease oftheimmune,nervous, and digestive systems( 1, 7, 8). Infected epithelial and other

cell types in thegastrointestinaltract arepotential reservoirs of

HIV- 1 and maycontributetothe pathophysiologyofdiarrheal syndromes in HIV-l-infected individuals (9-17). Recently,

wereported differencesamongthree human colonic epithelial

celllines, SW620,HT29,andT84,with respecttotheir expres-sionofthe CD4/HIV-1 receptor,HIV-l infectivityand replica-tion, and activation ofthe HIV- 1 LTR byprotein kinase C

(PKC)activators( 18). Thus, transcriptionalactivation ofthe HIV-1 LTR was stimulated by phorbol esters (phorbol

12-myristate 13-acetate [PMA]) and theproinflammatory cyto-kineTNFain theHIV-1infectible SW620and HT29 celllines,

butnotin thenoninfectibleT84 colon epithelial cell line( 18). Although mutational analysis indicated that increased

tran-scriptional activity of the HIV- 1 LTR in SW620 and HT29 cellswasdue,atleast in part,toactivationoftheNFKB-binding

enhancer elements ( 18), the same agonists are also knownto

activate other sequence-specific DNAbindingproteins,

includ-ingthe transcription factor activator protein- 1 (AP- 1 )( 19).

AP- 1 isamember ofaclass of DNAbindingproteins

en-coded by a multigene family of nuclear protooncogenes that mediate cellular responses to growthfactors, regulatory

cyto-kines, and tumor-promoting PMA via an AP-l binding site,

termed a TRE (forTPA/PMA-responsive element) (19-23).

The AP- 1 bindingsite or TRE is recognized by dimericprotein

complexes composed of Jun homodimers or Jun/Fos

hetero-dimers ( 19, 24). The Fos component of AP-1 does not itself

1.Abbreviationsusedinthis paper: AP-1, activatorprotein-l; ATF, activating transcription factor; CAT, chloramphenicol

acetyltransfer-ase; CRE, cAMP responsiveelement; CREB, CREbindingprotein;

DSE, downstream sequence element; LTR, long terminal repeat; NFKB, nuclear factorkappa B; PIE,PMAinhibitory element;TAR, transactivator responsive; TPA,12-0-tetradecanoylphorbol-I3-acetate

TRE,TPA/PMAresponsiveelement. 1336 K A.Roebuck,D. A.Brenner, and M. F.Kagnoff

J.Clin. Invest.

©TheAmericanSocietyforClinicalInvestigation,Inc.

0021-9738/93/09/1336/13 $2.00

bind the TRE because of an inability to form homodimers, but inthe presence ofJun,Fos forms aFos/Junheterodimer com-plex of greater stability than Jun/Jun homodimers (25-27). SeveralTRE-likemotifs,someofwhich bindc-Fos-containing complexes, havebeenidentified in theupstream modulatory region of the HIV- ILTR(24, 28). However,afunctionalrole

for c-Fos orAP-1 in eitherpositive ornegative regulation of

HIV-1 geneexpressionhas notbeendemonstrated.

ThecAMP-responsive element (CRE) binding protein/ac-tivating transcription factor (CREB/ATF) family of cellular

transcription factorsis aclosely related class of transcription

factorswith members thatarestructurallysimilar to members

of theJun/Fos family.LikeJun/Fos family members, CREB/ ATFfamily memberscan also interact with each other to form

dimers thatrecognizetheCRE,asequencesimilarto theTPA/ PMAresponsive element (TRE) (29-31).Recentstudies have

shown thatmembers of theJun/Fos familycancombinewith

certainmembers ofthe CREB/ATF family,via their leucine

zipper dimerization domains, to form cross-family

hetero-dimers that preferentially bind to CRE-like motifs (32-35).

Forexample,c-Fos andc-Jun can cross-combine with several

differentATFtranscriptionfactors (36). Theabilityto form

suchcross-familydimers increases therepertoireofprotein

fac-tors availableforgeneregulation.

In this study, we identify functional TRE/CRE-like

se-quence motifs withintheU5 regionofthe HIV-lLTR.

Expres-sionstudiesinhumancolonepithelialcelllines indicate that

TRE /CRE-like downstreamsequence elements(DSE)can me-diatetranscriptionofthe HIV-1LTR in the absence of a

func-tionalHIV- 1 enhancer elementandindependentof Tat

trans-activation. These DSE sitesaretransactivatedbythec-fos

nu-clear protooncogene and can transmit PMA and TNFa

activation signalsto the HIV- 1 LTRin HIV-1 infectible,but

notinnoninfectible, colonepithelial celllines. DNAbinding studies show that nuclearextracts from colon epithelial cells

form c-Fos-containing gel shift complexes with the TRE/

CRE-likeDSE thatarealsorecognized byclassical CRE mo-tifs.

Methods

Plasmids and synthetic deoxyoligonucleotides.The HIV- 1 -91/ +232 LTRchloramphenicol acetyltransferase(CAT) reporterplasmidwas previouslydescribed(37). It consists of91 bp ofHIV-ILTRsequences upstream of thetranscriptionalinitiation site (+1)and232 bp of LTR sequences downstreamoftranscription, extendinginto the U5region

of the LTR,linkedtoaCAT reporter gene. The -91/+232ASLTR CATplasmidwasderived from the -91/+232 construction by

excis-ing aninternal 183 bpSstIfragment from nucleotide position +39 to +222. The-91/+232M6 LTR CAT plasmid was constructed in two stepsbyusingcassettemutagenesis. In thefirststep, an interim muta-tional plasmid (pBS-HIV-SstIM) was constructed by cloning a mutant

oligonucleotidecassetteinto a pBluescript II vector (Stratagene, Inc., SanDiego, CA) toyieldaplasmidcontaining LTR wild-type sequences from +39 to +222 except for several point mutations around + 160 and +95. The base pair changes (see Fig. 3 A) introduced into the two motifs were designed to create two new restriction enzyme recognition

sites,aBssHIIsite at+160 and a NotI site at +95, as well as disrupt the nucleotide sequence homologies to TRE and CRE motifs in those re-gions. These mutations also do not appreciably alter the predicted RNAsecondary structure of the transcribed noncoding 5' leader se-quence. Mutations in pBS-HIV-SstIM were identified by restriction

digestions withBssHIIandNotIand were verified by

dideoxynucleo-tide sequencing.For thesecond step of the mutagenesis, a 183-bpSstI

restriction fragment from pBS-HIV-SstIM was cloned back into its

originalHIV- I context toobtainthe -91 / +232M6 LTR CAT reporter plasmid (see Fig. 3 A). The correct transcriptional orientation of the clonedfragmentin the -91 /+232M6 plasmid was determined by

re-striction analysisandconfirmed functionallywith a Tat cotransfection assay(seeFig.3 B).

The expression plasmids RSV-c-jun, SV-c-fos, SV-fosB, RSV-junB, andRSV-NFlhave been described previously (20, 28, 38, 39). TheSV-tat expression plasmid, provided by A. B. Rabson (Center for Advanced Biotechnology and Medicine, Piscataway, NJ), contains the

firstexonoftheHIV-l tat gene driven by theSV40 early promoter

(40). The HIV-1 firefly luciferase reporter gene plasmid

(HIV-l-LUC), provided by A. Siddiqui (University of Colorado, Denver), contains complete LTR promoter sequences upstream of +80 (41). The TREluciferase reporter gene plasmid (TRE/AP-1-LUC), pro-vided by M. G. Rosenfeld (University of California, San Diego), con-sistsoftwopairs ofconsensus TREmotifsupstream of a rat prolactin minimal promoter containing a TATA box (see Fig. 5 A) (42). Plas-mids used as negative controls (e.g.,RSV-fgal, SV-fgal,and RSV-neo) were described previously (43).

Deoxyoligonucleotides were synthesized using a Milligen DNA syn-thesizer(MilliporeCorp., Novato, CA) and purified as specified by the manufacturer. TheSplandNFloligonucleotides were described

previ-ously (39). Otheroligonucleotidesused in these studies are listed in Table I. Toproduce DNA gelshiftprobes, oligonucleotides were made double stranded byannealingthesingle-strandedcomplementary oligo-nucleotides and end labeling with["32P]ATP(3,000 Ci/mmol) and T4polynucleotidekinase.

Cellculture, transienttransfections, and reportergeneassays. Hu-mancolonepithelial cell lines SW620, HT29, and T84 were obtained,

maintained,andtransfected as previously described ( 18). For

stimula-tion,cells were treated 16-20 haftertransfection with optimal concen-trationsofhumanrecombinant TNFa (20 ng/ml) (Genentech Inc., South SanFrancisco, CA) or PMA (20 ng/ml) (Sigma Chemical Co., St.Louis,MO) for 16 to 20 h. CAT (EC 2.3.1.28) and firefly luciferase

TableI.Synthetic Deoxyoligonucleotides UsedtoDetectDNABindingActivities

Oligo* Source Position$ Sequence

+95 HIVDSE-1 +84/+105 GCCTTGAGTGCTTCAAGTAGTG

+160 HIVDSE-2 +149/+170 GACCCTTTTAGTCAGTGTGGAA

+160M HIVDSE-2 +149/+170 GACCCTTGGGGCGGGTGTGGAA TRE Collagenase -79/-58 TAAAGCATGAGTCAGACACCTC CRE Somatostatin -58/-37 CCTTGGCTGACGTCAGAGAGAG PIE Collagen -732/-707 GCCACACACTGAGTTCACCTAGGT

* The nontemplate strand is shown. t Relative to the transcriptional initiation site (+ 1). § Mutated sequences and sequences homologous to

TREandCRE consensus motifs are underlined. TRE, TPA/PMA-responsive element; CRE, cAMP-responsive element; PIE, phorbol ester

inhibitoryelement.

U3

NF-icB SP-1 TFIID TAT

R

DSE-1 C

91 +1 TM| 95 + HIV-1 -91/+232 LTR CAT

B

RSV-neo SV-cfos O PMATNF1l O PMATNFa1

F

4 _;

_*

₁₁

₁

AC

YoCON.1.0 2.5 1.0 2.6 22.0 10.4

LANE 1 2 3 4 5 6

C

30-

.0-ui

oc

zI-50

;Z

-I

0

I-IL.

0z wZ

20-

10-(EC 1.13.12.7) activityassays wereperformedonequivalentamounts

ofprotein from cellularextractsand quantitatedasdescribed previ-U 5 ously( 18, 39, 43).Forcomparisonsbetweenexperiments,relative

ex-USE-2

pression was measured _{by normalizing reporter gene activity}tothe

CAT or luciferase activity produced by thesame reporter genecoupled CAT to thecontrolexpressionplasmidsRSV-flgalorRSV-neo as indicated

inthe figures. RSV and SV bothactasstrong promotersin allthreecell

0160

+Sst232

lines, as assayed in control cellstransfected with RSV-luciferase or

REPORTER GENE SV-luciferase.

DNA binding studies. Electrophoretic mobilityshiftassayswere

performedasdescribedbySmealetal. (26). Briefly, purified

bacte-rially expressed trpE-c-Junfusionprotein (10ng/,gl)(26)ornuclear

extracts(5-7ggprotein/l)preparedfromcolonepithelialcellsbythe method ofShapiroetal.(44)wereincubated with50,000cpm(- 0.5

ng) of various 32P-end-labeledHIV-l LTR DNAordouble-stranded

synthetic deoxyoligonucleotide probes (Table I)for30minat 40C ina

20-Ml reactionvolumecontaining 12%glycerol, 12 mMHepes-NaOH (pH 7.9),60 mMKCL,5 mM MgCl2,4mMTris-Cl(pH 7.9),0.6

mM EDTA (pH 7.9), and 0.6 mM DTT. Forcompetition experi-ments, 10 or 50 ng of unlabeled deoxyoligonucleotide competitor

(- 20- and 100-foldexcess competitor, respectively) wasaddedto

bindingreactions before addition ofextract. Inexperiments using

c-Fosantibody, antibody wasaddedto thebindingreactions 10min beforethe addition of the DNAprobe.Protein-DNAcomplexeswere

resolvedin5% nativepolyacrylamide gels preelectrophoresedfor1to2

h at 4°C in 0.25X TBE buffer(22.5 mM Tris-borate and 0.5 mM

EDTA,pH 8.3).Gelsweredried andexposed overnighttox-rayfilm

(Eastman Kodak Co., Rochester, NY) with an intensifying screen

at-700C. Results

c-Fos transactivates theHIV-J LTR in colonepithelialcells

treated with PMA or TNFa. Previous results indicated that TNFaand PMA activate the HIV- I LTRthroughthe NFKB-binding sites inHIV-1 infectible, butnotinnoninfectible, hu-mancolonepithelial cells ( 18 ). However, forafull transcrip-tional response, itappeared that the NFKB binding sites,

al-thoughnecessary, werenotsufficientand that LTRsequences in additionto the HIV-1 enhanceralso contributed tothese

agonistresponses(18).Sinceactivatorsof PKCcaninduce,in

0 PMA TNFa

Figure1. Effectofc-fos expression onHIV-I LTR-mediatedgene

transcriptionin PMA-orTNFa-activatedSW620colonepithelial

cells.(A)Structure of the HIV-1 -91/+232LTRCATconstruction.

Bindingsites(solid boxes) for known viral ( TAT) and cellular factors (NFKB, SP], and TFIID)areindicatedinrelationtotheU3,R, and

U5viral LTRregions. Thetwonewlyidentifiedc-fos-responsive

DSEat+95(DSE-1 )and +160(DSE-2)arealsoindicated. The

po-sitionofthetwoSstI sitesat+39 and +222usedtoconstructa

dele-tionmutationplasmidareindicated.(B) Transient transfections of

the-91/ +232LTRCATconstructinSW620 cells. 5 Mgof the -91/

+232LTRreporterplasmidwascotransfected with5 Mgofeither the control RSV-neoexpressionvector(lanes 1-3)orSV-c-fos(lanes

4-6). 16-20haftertransfection,somecellsweretreated with 20ng/

mlofPMA(lanes2and5)orTNFa(lanes 3 and 6), whereas other cellswereuntreated(lanes Iand4). CATassays wereperformed

us-ing equivalentamountsofproteinasdescribedpreviously (18).For

eachtransfection,thepercentconversion(% CON.), calculated by dividingthe "4Ccountsconverted intoacetylated chloramphenicol product (AC) by the total 14Ccounts(AC+CM) multiplied by 100, isindicated belowthecorrespondinglane.(C) Quantitation of the c-fostransactivationresponseinSW620cells.In fourindependent

transfectionexperiments, SW620cellswerecotransfected witheither

ac-fosexpressionvector(open bars)or acontrolRSV-flgalplasmid (stippled bars) togetherwiththechimeric -91/+232LTRCAT

re-portergene.Cellswerestimulatedwith PMAorTNFaor

unstimu-latedasindicated.Heightofthe barsrepresentthemeanfold increase of-91/+232LTR-directedCATactivity normalizedtotheCAT

activityof theunstimulatedRSV-,Bgal expression plasmidcontrolplus

the standarderrorofthemean(SEM).Incontrolexperiments,the RSV and SVpromoterswereshowntobecomparablyactivein

SW620cells. 1338 K.A.Roebuck,D.A.Brenner,and M.F.Kagnoff

A

addition to NFKB/rel, cellular factors in the Jun/Fos family of transcription factors, we asked whether AP-l proteins play a

role intranscriptional activation of the HIV- 1 LTR. Several AP- I bindingsites had been identifiedbefore, upstream of the NFKBbinding sites, but a functional role for these motifs has

thus farnot been demonstrated (24, 28, 45).

Wedescribe herein that therealso arepreviously

unrecog-nizedAP-1-likebindingsites in the HIV-1 LTR downstream of the promoter in the transcribed noncoding 5' leader se-quence. Atleasttwodistinct regions,one centeredat

nucleo-tideposition+95 and the other at+160 in the 5' untranslated leadersequencesdownstream ofTAR,exhibitahigh degreeof sequencesimilarity toclassical TRE and CREmotifs(Fig. 1 A). To assesswhether thesenewly recognizedDSEmight

con-tributetoactivationofthe HIV- 1 LTR,weperformedtransient

cotransfection experiments witha truncated HIV-l LTR

re-porter gene plasmid, -91/+232 LTR CAT, which contains HIV- 1 LTRsequencesfrom nucleotideposition-91 to +232

(relativetothe viraltranscription initiationsite) coupledto a

CATreporter gene(Fig. 1A).Unlike LTR CAT constructions

that contain upstream AP- 1 binding sites and a functional HIV- 1 enhancer element with two NFKB-binding sites, the

-91/+232 LTRCATplasmid hasonly the 3'NFKBbinding site intactand,comparedwithaconstruction with both

NFKB-bindingsites (18), is onlyweakly expressedwhen cotransfected

into SW620 cellswith a control RSV-neo expression plasmid

(Fig. 1 B, lane 1). Consistent withourprevious results, this enhancer mutant construction exhibited little or no response to the PKC activators PMA and TNFa, respectively (Fig. 1 B,

lanes1-3).

Asshownin Fig. 1, we first tested whetherc-Fos, amajor

componentofAP- 1 complexes, could stimulate HIV- 1 LTR genetranscription in the absence ofanintactHIV- 1 enhancer andupstream AP- 1 bindingsites.Forthese studies, the -91/

+232reporterplasmidwascotransfected into theHIV- 1 infec-tible colon epithelial cell line SW620, together withan

expres-sionvectorthat produces c-Fos. When the -91/+232reporter

plasmidwascotransfected withthec-fos expression plasmid,

HIV- 1 LTR reporter geneactivity increased (Fig. 1 B, lane4 compared with lane 1).Moreover,inc-fos-cotransfected cells, the-91/+232LTR reporter genebecamehighly responsiveto

bothPMA andTNFa stimulation (Fig. 1 B, lanes 5 and 6).

c-Fos transactivation ofthe HIV-1 LTR reporter gene was

quantitatedin several independent transfection experiments.

Asshown in Fig. 1 C,c-fos transfection increasedHIV- I LTR

activation fourfoldovercontrols

(RSV-flgal

plasmid).In con-trast, there was a mean 10-20-fold activation ofthe HIV-1 LTR when thec-fos-cotransfectedcells werestimulatedwith PMAor TNFa. The data shown in Fig. 1 demonstrate that an

intact HIV- 1 enhancer andupstream AP- 1 binding sites are

HIV-1 -914/232 LTR CAT REPORTER

RSV-neo

SV-cfos

SV-cfos+RSV-cjun

SV-fosB

OPMATNFa

I

O

PMATNFal

I O

FMATa

I O PMATNFa

S...,@

.@e*@@

1 2 3 4 5 6 7 8 9 10 11 12

HIV-1 -91/+232 A S LTR CAT REPORTER

SV-cfos

RSV-neo SV-cfos +RSV-cjun

f

_o

_PMATNFal

F

_o

_PMATNFa1

0

PMATNFxl

SV-fosB

I

O PMA

ThNF&

_.

S....

...9@@

13 14 15 16 17 18 19 20 21 22 23 24

Figure 2.Adeletion in the -91 / +232 LTR CAT

plasmidabrogatesc-fostransactivation ofthe

HIV- I LTR.Transfectionsand CAT assayswere performedasdescribedin Methods and the leg-endofFig. 1.The toppanel (lanes1-12)shows

resultsusingthe -91 / +232construction

trans-fected into SW620 cells. The bottompanel

(lanes 13-24) shows parallel results with the

-91/+232ASdeletionalmutation,whichis identical instructure tothe-91/+232 plasmid

except for the deletion of sequences between the

twoSstI sitesat+39 and +222(seeFig. 1A). Expression plasmidsthat werecotransfected and

agonist(PMA, 20 ng/ml and TNFa, 20ng/ml) treatments areindicated above each lane.

notrequired for c-Fostransactivation ofHIV- 1 gene expres-sion and indicate the c-Fos response ismediatedby LTR se-quences downstream ofthe HIV-1enhancer.Infurthercontrol

experiments, using additional LTR constructions orproviral

plasmids, itwasshownthat c-Fostransactivationofthe HIV- 1 LTR canalsooccurin thepresenceofafunctionalHIV- 1 en-hancerelement, which contains both NFKBbinding sites(data notshown).

The HIV-J LTR contains c- Fos-responsive elements downstream ofTAR. Theobservation that c-Fos, in concert with PMA or TNFa, stimulated HIV-1 LTR-mediated CAT

activity suggested that the HIV-1 LTR contained a TRE or CRE element downstream of the HIV-1 enhancer. We first

tested whether a deletion of the TRE/CRE-like LTR se-quencesencompassingthedownstreamTRE/CRE-like motifs

centered atnucleotidepositions +95and+160 would alter the c-Fos transactivation effect documented in Fig. 1. The

con-struction used hadadeletion betweentwoSstIrestriction sites from +39 to +222 in the -91/+232 LTR CATconstruction.

Incotransfection experiments (Fig.2,top), thewild-type-91 /

+232construction,asindicated previously (Fig. 1),was

tran-sactivated byc-FosinSW620 cells.Incontrast,c-Fos

transacti-vation of asimilar constructionlackingthe downstreamTRE/

CRE-like motifs (-91/+232AS) was significantly reduced (Fig.2, compare lanes 4-6 and 16-18). Thedownstream

dele-tionhad nodetectableeffectonLTR-mediated CAT activity in the control RSV-neo plasmid cotransfections(compare lanes 1-3 and 13-15). These results indicatethat LTR sequences

downstream ofTAR, between nucleotide positions +39 and +222, are required for c-Fostransactivation of the HIV-1 re-porter gene construct.In further experiments, cotransfectionof

the c-Junexpressing plasmid together with c-Fos did not

re-storethe transactivation response in cells containing the -91 / +232AS construct and, in fact, it attenuated the transcriptional response stimulated by TNFa and PMA in the c-Fos-trans-fected cells containing thewild-type -91/+232construct (Fig. 2, compare lanes 7-9 and 19-21). FosB, another memberof

the fos gene family (27), was unable to significantly stimulate LTR-ediated CAT activity in SW620 cells (see Fig. 2, lanes 10-12 and 22-24).

Pointmutations inthe+160and+95TRE/CRE-like

mo-tifs abrogatec-Fos transactivation of HIV-I gene expression.

To demonstrate whether theTRE/CRE-like motifs at +160 and +95 were important for c-Fos transactivation ofthe HIV- 1 LTR, weintroduced cluster point mutations into these DSE. Unlike the wild-type reporter gene (Fig. 3B, lanes 3 and 4), when the -91/+232M6construction containing mutationsin the +95 and +160 DSE motifs was transfected into SW620 cells, no detectable transactivation of the HIV- 1 LTR was ob-served with cotransfection ofa c-Fos-expressing plasmid in either untreated or PMA-treated cells (Fig. 3B,lanes 7 and 8). In contrast, a cotransfected Tat expression plasmid strongly

stimulated the -91/+232M6 reporter gene construction in PMA-treated(Fig. 3, lane 9)and untreatedSW620 colon epi-thelial cells (data not shown). The cluster mutations had little or noeffect on the basal level ofgene activity (compare lanes I

A

NotI BaaNII

-91/+232M6 +87 TTGGCGGCCGCAA +99....+153 CTTGCGCGCAGT +164

*** *****

-91/+232WT +87 TTGAGTGCTTCAA +99... .+153 CTTTTAGTCAGT +164

+95 DSZ-1 +160 DSZ-2

B

-91/+232M6

N.01'~~~~

SV-Bgal

SV-CfOS

SV-Bgal

SV-cfos SV-tat

IOPA1PMA

MA

11

FOPA1

PM

A

MA

IIPMAI

1 2 3 4 5 6 7 8 9

Figure 3. Effect ofclusterpointmutations in the HIV-1 DSE motifs onc-fostransactivation of theHIV-1 LTR. Transfections and CAT assays wereperformedasdescribed in Methods and thelegendofFig. 1.(A) Mutationsintroduced

into the-91/+232WTreporterconstruct at

DSE-1 and DSE-2 bycassettemutagenesisare asdescribed in Methods. Asterisks denote spe-cific basepair changesfrom the wild type that

create aNotI siteat+95andaBssHIIsiteat +160 in the -91 /+232M6 reporterplasmid. (B) Expressionresultsfromexperimentsinwhich the-91/+232WT (lanes 1-4)or-91/+232M6

reporterconstructs(lanes5-9)weretransfected

intoSW620 cells. TheplasmidsSV-c-fos(lanes 3,4, 7,and8), SV-,Bgal (lanes1,2, 5,and6),

andSV-tat(lane9)cotransfectedwiththe LTR reporter genesareindicated.CATactivity de-rivedfromPMA-treated cellsorunstimulated cells(0)is shown.

1340 K A.Roebuck,D. A.Brenner,andM.F. Kagnoff

and 2 with lanes 5 and 6).Together,these results demonstrate

that, althoughnotnecessary for Tattransactivation,theTRE/

CRE-like DSE motifs are required for c-Fos transactivation of the HIV- 1 LTR in colonepithelialcells.However,whether one orboth of the motifs is required is not known, as each site was notmutatedindividually.

Differentialc-Fos transactivation in three colon epithelial cell lines.To assess whetherthe markedincrease in CAT activ-ity stimulated by c-Fos and PMA or TNFa treatment was spe-cific for theCD4'SW620 cell line or reflected a more general phenomenon, weconducted paralleltransfectionexperiments with two additional human colon epithelial cell lines

previ-ously characterized in ourlaboratory, HT29 and T84 (18). The HT29 cells usedinthese studies do not express the CD4 cell surface glycoprotein but are, nevertheless, infectible by HIV- 1 viaanon-CD4-mediated pathwayinvolvinggalactosyl

ceramide or acloselyrelated molecule( 18, 46). As shown in

Fig. 4, therewasstrongtranscriptionalactivation of the -91 / +232LTRCAT template when HT29 cellswerecotransfected withthe c-Fosexpression plasmid and stimulated withPMA or TNFa.T84 cells,awell-differentiated colonic epithelialline of cryptorigin,also lack CD4 butare notinfectiblebyHIV- 1 and

donotactivateHIV- 1 LTRtranscriptionin responsetoTNFa or PMA stimulation (18). T84 cells cotransfected with the

-91 /+232LTRCATtemplateandac-Fosexpression vector

and stimulated with PMA or TNFaproduced onlyamodest

increase in CAT activity(Fig. 4). The markedlylower activa-tion response in cotransfected T84 cells compared with the HIV- 1-infectiblecelllines SW620 and HT29suggeststhat acti-vationofthe HIV- 1 LTRin colonepithelialcellsislimitedto

A

HIV-1 LTR

2xTRE 2xTRE _PRC

B

IRE/AP1 LUC

-PMA +PMA

HIV-1

LTR LUC

-PMA +PMA

SW620

RLU (foldincremoe

315 1512 (4.8X)

4236 51295 (12X)

T84 RLU (fold increaWe

4959 29228 (5.9X)

148136 146392 (1X)

Figure 5. Phorbol ester-stimulatedtransactivationresponseinT84

and SW620 cells transfectedwithaTRE/API-LUCorHIV-1

LTR-LUCreportergeneplasmid. (A)Structure of the luciferasereporter

plasmids.HIV-1 LTR-LUCcontained LTRsequencesupstreamof

+80.TRE/API-LUCcontained fourconsensusTRE motifs

up-streamofaratprolactinminimalpromoter.(B) Expressionresults

ofSW620(left) and T84 (right) transfected withaTRE/API-LUC

or anHIV-1 LTR-LUCreportergeneplasmidasdescribed in Meth-ods andtreated withPMA(20 ng/ml)or noPMA. Dataarethe meanof three transfection experimentsandareexpressedasrelative

light unitsper 10susing equivalentamountsof protein from cell

extracts.The fold increase inresponsetoPMA stimulation is indi-cated inparentheses.Thegreaterluciferaseactivityobserved in T84 cells reflects theirhighertransfectionefficiency.

-911+232 LTR CA

+ RSV0 gal PMA

-TNF

0

+ SV-cFOS PMA TNF

+RSV-cJUN PIA TNF

+RSV-JUNB PMA

+RSV-NF-1 PIA TNF

Figure4.Effects of

diatedgenetranscril

(left), HT29 (center

transfected with5 with 5ggoftheexpi treatment(open baj

TNFa(stippled bar.

sentsCATactivitye

thelegendofFig. 1

periment.Similarre

periments.

colonic celltypescapable of mediating PMAor TNFa-stimu-SW620 HT29 T84 latedactivationsignalstothe HIV-l LTR.

Incontrast tothe c-Fos expression plasmid, littleor no

in-creaseinCATreportergeneactivitywasnoted after PMAor TNFa stimulation inanyof the colon epithelial cell lines

co-transfected with c-Jun, junB,orNFl expression plasmidsor

with the control

RSV-f3gal

expressionvector.However,

RSV-c-jun, RSV-junB and SV-c-fos plasmidswerefunctional in the

three humancelllines, since in controls they readily transacti-vatedareportergenecontaining multipleconsensusTRE mo-tifs_(datanot_shown).

T84 cellscanmediateaPMA transactivationresponsetoa

TRE-dependent reportergenebutnottothe HIV-1 LTR. To determine whether the inability of T84 cellsto mediate PMA activationsignalstothe HIV-1 LTRwaspromoterspecificor

representedamoregeneral defect inasignal transduction

path-] X 1 1t I t

Iway,

wetested theability of another reportergene

construc-6 15 24 6 15 24 6 15 24 tion, TRE/AP1-LUC, to be activated byPMA in T84 cells.

Thisconstruction (Fig. 5 A) contains multipleconsensusTRE

motifs upstream ofa rat prolactin minimal promoter (42)

variousexpressionplasmidsonHIV-1 LTR-me- linkedtothefirefly luciferasegeneandis transcriptionally

re-ption in threecolonepithelialcelllines. SW620 sponsivetoPMAin SW620 cells (Fig.5 B).Similarly, inT84

r), and T84(right)colonepithelial cellswere co- cells transfected with the TRE/AP1-LUC reporter gene, a

gof the-91/+232 CATreporterplasmid together nearly sixfold stimulation ofluciferase activitywas noted in ression vectorsindicated. Cells either received no

responsetoPMAstimulation (Fig. 5 B). Thesedataindicate

Ts),rs),orwerestimulated with PMA(solid bars)or _{thatT84 cells, likeSW620 cells, are capable of mediatingPKC}

s)asindicatedinFig.1.Lengthof thebarsrepre- _{n n}

-xpressedaspercentconversionascalculated in activation signals to trgger gene activation, presumably by

ac-[.Thesedataarefromasinglerepresentativeex- tivatingTRE-bindingJun/Fostranscriptionfactors.

esultswereobtained intwo tothree repeatedex- In contrastto theTRE/API-LUCplasmid,whenanHIV-I LTR-LUCreportergeneconstruct(Fig. 5 A) containing

com-c-fos-responsiveElementsintheHIV-J LongTerminalRepeat 1341 LUCIFERASE

_I

plete LTR sequences upstreamof+80but lacking a fully

tran-scribed noncoding 5' leader sequence was transfected into SW620 orT84cells, astrong PMA response wasdetected in SW620 cells, butnot in T84 cells underidentical conditions (Fig.5B). PMA-treated T84 cellswerealso unabletoactivate a full-length LTR CAT construct that included downstream

LTR sequences tonucleotide position +232 or a full-length proviral clone(data notshown). These dataareconsistent with

our previous results, indicatingthat T84 cells are unable to

mediate PMA activation ofthe HIV-l LTR(18). Although

T84 cellscantransmitPMAactivation signalsto AP- 1-depen-dentreporter genes,theyareapparently deficient in the intra-cellularsignaling that is requiredtotrigger the activation ofthe HIV-l LTRby PMA.

Downstream sequencemotifs weakly bind recombinant c-Junprotein. Havingdocumented theimportance ofthe DSE

motifs in c-Fos transactivation ofthe HIV-1 LTR, we next

assessed their AP-l binding properties. Because c-Fos itself doesnotbind DNA,weperformed gel mobility shift

experi-mentswith purified recombinant trpE-c-Jun fusion protein, which, as amajorcomponentofAP-1 complexes,was

previ-ously showntobind withhigh affinitytoaclassical TRE motif (26). WhenincreasingamountsoftrpE-c-Junwereincubated with an HIV-1 LTR restriction fragment (+80/+222)

con-tainingthe DSEmotifs,aspecific gelshiftcomplexwasformed

(Fig.6 A, lane 5).Thisc-Jun-DNA complexmigratednear a

TRE deoxyoligonucleotide c-Jun binding complex, but dis-played lessbinding activity(compareFig.6 A, lanes1and5). Toassesswhether c-Jun couldbindtothe HIV-l DSEsites,we

synthesized a series ofshort double-stranded deoxyoligonu-cleotides containingthe

TRE/CRE-like

motifs (Table I). In thegelmobility shiftassaysshown inFig.6 B,an oligonucleo-tidecontainingthe +160 DSE-2 motif formed specific

com-plexes with trpE-c-Jun(Fig.6 B, lanes2and3)thatcomigrated

with complexesformedonthecollagenaseTRE

oligonucleo-tide control(Fig.6 B, lane1),albeit with lessbinding activity.

In contrast, an oligonucleotide witha mutationin the +160

motif(+ 160m) formed no detectable trpE-c-Jun complexes (Fig. 6 B, lanes 4 and 5). The +95 DSE-1 oligonucleotide formed only weakly detectable complexes with trpE-c-Jun (Fig. 6 B, lanes 6 and7).Theseresults indicatethat, although

c-Juncanrecognizeand bind the downstream+160 and +95 DSEmotifs, itdoessoless

efficiently

thanto aclassicalTRE

oligonucleotide.

Tomoreaccuratelyassesstherelativebinding affinities of

thevariousoligonucleotides forc-Jun,increasingamountsof

unlabeled +160, +160m and +95 DSEoligonucleotideswere

competedfor trpE-c-Junbindingwith the labeledcollagenase

TREoligonucleotide (Fig.6 C).Inaccordance with their ap-parentbindingefficiencies documentedinFig. 6 B, the+160 DSE-2 oligonucleotide competed with theTRE oligonucleo-tide for c-Junbinding(Fig.6 C,lanes 5 and6) better than the +95

DSE-l

oligonucleotide (lanes 9 and 10), orthe +160m DSE-2oligonucleotide(lanes7and8). However, the

homolo-gouscollagenaseTREoligonucleotidewas asubstantially

bet-tercompetitorthan anyoftheHIV- 1LTRderived

oligonucleo-tides(Fig.6 C, lanes1-3).

T84cell nuclear extracts produce a distinct gel shift pattern

ofCREmotifbinding activity. To identify the DNA binding activitiespresentin colon epithelialcells that recognize

HIV-l

LTR sequence

motifs,

weprepared cell-free nuclear extracts

(44)

fromT84, SW620,and HT29cells,incubatedthemwith

labeled oligonucleotide sequence motifs, and resolved DNA

binding complexes by gel electrophoresis (Fig. 7). Fig. 7 A shows that specific gel shift complexes form with oligonucleo-tides containing SpI (lanes 1-3) and NFl binding sites (lanes 4-6). Spl andNFl are common transcription factors present

inawide variety ofcell types and their binding sites arefound inmanycellular and viralpromoters, includingHIV-1(4, 39);

their similar DNA-binding activitiesamong thethree cell lines provide, therefore, anindicationof the quality of the colonic

cell nuclearextracts.

Because thedownstream motifs in theHIV-l LTR closely

resemble consensus TRE and CRE motifs, we tested each cell extract for DNA binding activity to oligonucleotides with a

classical TRE(20, 21 )orCRE (30) motif(Table I). Although

classical TRE and CRE motifsare nearlyidenticalin primary sequence,thecollagenase TREmotif isa strong binding site for

members ofthe Jun/Fos family, whereas the somatostatin

CRE motif isapreferential binding site for CREB/ATFfamily

members (20, 21, 30). The gel shiftdatashown in Fig. 7 B

indicate thateachofthe threecolonepithelialcell lines contain separateCRE (lanes 1-3) andTRE(lanes4-6) binding

activi-tiesthat donotcrosscompete(lanes 2 and 6).Although the

individual CRE andTREcomplexeseachexhibitasimilar gel

migrationpatternwith allthree cell nuclear extracts, the

rela-tiveDNAbinding activity ofoneofthe twomajorcomplexes

formed with theCRE oligonucleotidewas differentwith the

nuclearextractofT84 compared withSW620 and HT29 cells.

Thefaster-migrating CREcomplex with T84 cell nuclear ex-tract wasmarkedly reduced compared withthe

slower-migrat-ing CRE complex and compared with the CRE complexes

formedwithidenticalamountsof SW620and HT29 cell

nu-clearextract(Fig.7B, lanes 1-3).Together, these data indicate

that theDNAbinding activitiesin colonepithelialcells are very

similar; theonenotableexception beingtheCRE binding activ-ity in T84 cells, whichappears to bedeficientin the

faster-mi-gratingCRE gelshift complex.

Downstreamsequence elements form novel c-fos-contain-ing complexes with colonic epithelial cell proteins. To further

assesstheDNAbinding activity oftheTRE/CRE-likemotifs

in the HIV-l LTR, we performed gel mobility shift

experi-ments including oligonucleotides containing the +160 and

+95 DSE motifs (Table I). Theresultsin Fig. 8show that a

specificDNAbinding activitycanbe detected withthe + 160

and+95 DSEoligonucleotidesin nuclearextractsfromeachof the colonepithelial cell lines(Fig. 8A, lanes I and5). The

DNAbindingactivity detected with the T84cell nuclear ex-tractwas, however, weak compared with the relative binding

activitygenerated by identicalamountsof SW620orHT29 cell nuclearextracts. When unlabeled oligonucleotideswere used tocompetefor complex formation,the unlabeled collagenase TREoligonucleotidedid notinterfere withthe formation of either the +95or+160 DSE gel shift complexes(Fig.8 A, lanes

2and 6). Incontrast, thesomatostatin CRE oligonucleotide blocked the formation of these gelshift complexes (Fig. 8A,

lanes 3 and 7) to a degree similarto orbetter than that of homologous competition withthe +160 or +95

oligonucleo-tide (lanes4and8). These results indicatethat the DNA bind-ing proteinsincolonicepithelial cellsthatrecognizethe HIV- I LTRDSE alsorecognizeaclassical CREmotif.

Using the SW620 nuclear extract, the gel shift complex formed on the +160 DSE-2 oligonucleotide clearly migrates fasterin thegelthanthecomplex formedon the collagenase

A

Probe LTR +801.222

protein 9 0 1 3 9 (uI)

B

OLIGO PROBE TRE +160 +160M +95

JUN PROTEIN (ui)FI9W1 13 91 13 91 13 91

'p

1 2 3 4 5

Figure6. TrpE-c-Jun bindingtoHIV-1 LTR DNAfragments and

synthetic deoxyoligonucleotides. (A)Agel-purified fragmentof the HIV-l LTRfrom +80to +222wasgenerated by digestion of the

HIV-1 LTR withHindIll and SstI. This LTRfragmentwas

32P-end-labeledatthe HindIII restriction site(+80), incubatedwith

increas-ingamountsofTrpE-c-Jun (0, 1, 3,or9til),and assessed forbinding

inagel mobilityshiftassay(lanes 2-5).Asapositive control,a

22-merTREoligonucleotide containingacanonical TRErecognition

site(Table I)wasincluded(lane 1).Themigrationof the c-Jun-LTR

complexisseeninlane 5. Unbound LTRfragment migratedatthe bottom of thefigurewhereas the unbound 22-mer TRE

oligonucleo-tide isnotshown.(B) TrpE-c-Jun bindingtosynthetic

oligonucleo-tidescontainingDSE sites. The 32P-end-labeledoligonucleotidesfrom the downstreamsequencesinthe HIV- I LTR(shownin Table I and indicated above eachlane)wereincubated with3or9 oftrpE-c-Jun

proteinasdescribedinMethodsandassessed for bindingingel

mo-bilityshiftassays.The 22-mercollagenaseTREoligonucleotide (lane 1)wasincluded forcomparisonas apositive control. Migration of free probe and c-Jun-DNA complexesareindicatedtothe rightby

arrows.Ofnote,insometrpE-c-Jun preparations the Escherichiacoli

trpE moietyiscleaved, producingtwodifferent-sizedbinding species

andthereby generatingtwodistinct gel shift complexes. (C)

Compe-tition oftrpE-c-Jun bindingtoTREbyHIV- IDSE oligonucleotides.

The32P-end-labeledTREoligonucleotidewasincubated witha

con-stant amountof trpE-c-Jun and competedasindicatedabove each lane with either0, 10,or50ngof unlabeledTRE (lanes 1-3),+160

(lanes 4-6),+160m (lanes7and 8),or+95 oligonucleotides(lanes

9and10). Migrationofthe free unbound TRE probe(bottomarrow) and the c-Jun-TREoligonucleotide complexes(top arrow)are

indi-cated.

TRE oligonucleotide (Fig. 8B,lanesIand2).Incontrast,the

+160 DSE-2 complex comigrated with the faster migrating complex formed on the somatostatin CRE oligonucleotide

(Fig.8 B, lane 7), which, interestingly, wasthesamecomplex

astheonepreferentiallyreduced whenusing T84cellextracts

FREE PROBE 1 2 3 4 5 6 7

C

COMPETITOR DNA (ng)

TRE OLIGONUCLEOTIDE PROBE

TRE +160 +160M +95 I0 10 5011 0 10

-5-0l

F

0750

r-10-5-01

0TRE

PROBE

1 2 3 4 5 6 7 8 9 10

(Fig.7B). Again,thecollagenase TRE oligonucleotidedidnot

compete for formation of the+160 DSE-2complex (Fig.8B,

lane 3)nordidaTRE/CRE-related oligonucleotide, phorbol esterinhibitoryelement(PIE) (Table I),derived from the

col-lagengenepromoter (lane 5). Incontrast, competitionwith

c-fos-responsive Elements in theHIV-1LongTerminalRepeat 1343

*- _cJUN

COMPLEXES

.0

.,A

-N

IL -& .:.

imlkt ', A ML

.41F--A

f I

1 2

CRI

10 TREC

NF1

Figure 7. Comparative binding activities of co-lonepithelialcell nu-clearextracts toDNA

sequencemotifs.A con-stant amountof nuclear

extract from SW620

(top), HT29(middle), i andT84(bottom) colon

epithelialcell lineswere

3 4 5 6 incubated with the

32P-end-labeled

oligonucle-otideprobes (- 0.5ng)

asindicated.

Protein-DNAcomplexes were

E TRE resolved by gel electro-51 rO TRECREI phoresis. (A) Gel shift

complexes formed with

SpI (lanes1-3)and NFI oligonucleotides I I _ (lanes4-6)in the

ab-sence(lane1)or

pres-enceof unlabeled

Spl

(lanes2 and5)orNFI competitor oligonucleo-tide (lanes3 and6). (B)Gelshift complexes formedwithCRE

(lanes 1-3) and TRE

oligonucleotides (lanes 4-6)in theabsence

(lane

1)

orpresenceof

unlabeled TRE(lanes

2and5)orCRE

com-petitoroligonucleotide

(lanes3and6).

Migra-tionof the unbound free

probeis notshown.

Competition, for un-knownreasons, some-times increased the

for-mationofnonspecific

complexes(lanes 2, 3, 3 4 5 6 5,and6).

the somatostatin CRE motifreducedformation of the+160 DSE gel shift complex (Fig. 8B, lane 4) againto an extent

similartothatseenwith thehomologous

oligonucleotide (lane

6). Identical resultswereobtained with the +95 DSE- 1 oligonu-cleotide (datanotshown).

To determine whether c-Foswas acomponentof the CRE-like DSE gel shift complex,ac-Fosantibody (1 8C3) thatwas

previously showntospecifically interfere with c-Fos-contain-ing complexes (47)wasincluded in the binding reactions

be-fore adding DNA probe (Fig. 8 C). Since the c-Fos gene is induced by serum, for these gel shift experiments SW620 serum-starvedextracts were usedto assesswhether the DSE complexwould be affected. As shown inFig. 8 C, thec-Fos

antibodyblocked theformationoftheCRE-like DSE complex (lane 2, complex I). Interestingly, using serum-starved ex-tracts, a newcomplexwasalso formed thatmigratedslower in the gel than theCRE-like DSE complex (Fig. 8 C, lane 1,

com-plexII). Thisnewcomplexwasalsosomewhatdiminished by the anti-c-Fos antibody (lane 2, complex II). c-Fos antibody

wasspecific for thecomigrating complexshared by the DSE and the CRE,asshown in Fig. 8 C, lanes 3 and 4,sinceitwas

abletoblockonly the fastermigratingCREcomplex (lane 4). Thus,although the HIV-l DSEmotifscanformmorethanone

gel shiftcomplex, the faster-migrating complexappears tobea

c-Fos-containing CRE-like binding complex.

Discussion

We describe DSE in the transcribed noncoding 5'leader se-quencesof the HIV-1 LTR thatarec-Fosresponsiveandare

differentiallyactivated in three human colon carcinoma

epithe-lial celllines. OneorbothoftheDSE,shown inrelationtothe HIV- I LTR in Fig. 1A andatthenucleotide level inFig.3 A, mediate c-Fos transactivation ofthe HIV-1 LTR in colonic epithelial cells stimulated with PMAorTNFa. Thesearethe

mostdownstreamcis-regulatoryelements described thus far in theHIV-1 LTR.Thesefindingssuggestthatpreviously

unrec-ognizedsequenceelements that reside downstreamofTARin the transcribednoncoding 5'leadersequence maycontribute

tothe activation andregulationofHIV- 1 geneexpression.

Role ofthe c-Fos-responsive DSE sites intheactivationof theHIV-1 LTR. Thefirst downstreamelement, DSE-1, is lo-catedimmediatelydownstreamofTARatnucleotide position

+95 andappears tobe composedoftwopartially overlapping TRE/CRE-like sites. DSE-2,the other HIV-1 c-Fos-respon-sive element,islocated atposition +160 and hasperfect

se-quenceidentitytothefunctional AP-l binding sitesfound in the72-bprepeatsofSV40and theenhancerregion ofpolyoma virus(TTAGTCAG)(48).The DSE-2site isalsoidenticalto twoof the three

PMA-responsive

elementsrecentlyidentified in theHIV-1 polgene,whicharethoughttocomprisea

poten-tialHIV- 1 intragenicenhancer element(49).Althoughthe in vivo role ofthe DSE elements remainstobedemonstrated,it is noteworthy that Jun/Fos expression plasmids can stimulate

newviralproduction fromatransfectedHIV-l proviral

plas-mid clone in ahighly transfectible human colonic

epithelial

cellline(Roebuck, K.A.,

unpublished

results).

The c-Fos-responsive DSE sites may comprise a

down-stream LTRenhancerthatfunctions independently of, orin

concertwith, theupstreamNFKB-bindingenhancertoactivate HIV-l transcription. An additional inducible enhancer

ele-mentin the viralLTR mayprovideamechanismtobroaden the viral response toextracellular stimuli and activate

tran-scription undera widervarety of cellularconditions. Indeed, functional redundancy isacommonfeature of viralaswellas

A

OLIGO PROBES co

OL

spi

IOS (5Ong) Io SP1 NF110 _{SP1 NF1}

I

.W.

SW620

HT29

T84

B

OLIGO PROBES

COMPETITOR OLIGOS (50ng)

SW620

HT29

Sb

T84

1 2

A

OLIGO PROBES

+160

+95

B

OLIGO PROBES

TRE

+160

COMPETITOR 0 TRE CRE+1601' 0 TRE CRE +95 1

OLIGOS (Song)

COMPETITOR

OLIGOS (50ng)

SW620

0 0 TRECRE PIE

+160

0

6.

"W

S

~~~~~~~1wY,

=1

W

,. , a f. | _

HT29

i

_

LI .

I.

T84

x*U;1

I"

F~

A

40-1 2 3 4 5 6

I

1

2

3

4

5

6

7

6 Q

E E

o

W 0

U.

U-C-III

I

1 2 3 4

8

Figure 8.Characterizationof novelTRE/CRE-likemotifbinding ac-tivitiesincolonepithelialcell nuclear extracts.32P-end-labeledHIV-l

DSE-l and -2oligonucleotidesshown in Table I(-0.5ng)were

as-sayed forbinding activityasdescribedin Methods and thelegendof

Fig.7.(A) Gelshiftcomplexesformedwith +160(lanes 1-4)and +95 DSEmotifoligonucleotides (lanes 5-8) withSW620(top),

HT29(middle),and T84(bottom)colonepithelialcell extracts. The

+160 and +95 gelshiftcomplexes were not competed (lanes 1 and

5)orcompetedwith 50 ngofunlabeledoligonucleotideasindicated

above each lane.Migration ofthe unboundfree probeisnot shown.

(B)Comparison ofCRE(lane 7), TRE (lane1),andHIV-l DSE

oligonucleotidecomplexes (lanes2-6)formed with SW620 cell nu-clear extracts.The+160 DSE-2complexwaseither noncompetedor

competed with50 ngofthe unlabeledoligonucleotidesshown in Ta-ble Iasindicatedabove each lane. Threemajor distinguishable

com-plexesareindicated to theright byarrows.Migration ofthe unbound

free probe is indicatedatthe bottomrightby an arrow. Note that the

+160 DSE-2 gel shift complexmigrates differentlythan the TRE

complex (lane 1)andcomigrateswith thefaster-migratingCRE

complex (lane 7). (C)Comparisonof the gel shift complexes formed on the+160 DSE-2(lane1)andCRE(lane 3) with serum-starved SW620 nuclearextract.Bindingreactions were done in the absence (lanes 1 and3) or in the presence of ac-fosmonoclonal antibody

(lanes2and4)asindicated above each lane.c-fosantibody blocks theformation of theCRE-likeDSE complex (lane 2, complex I) and thefaster-migratingCREcomplex (lane 4, complex I). Note that extracts from serum-starved cells also have a slower migrating com-plex(complex II) on the DSE motif, which also is partiallyblocked byc-fosantibody.

c-fos-responsive ElementsintheHIV-1 LongTerminal Repeat 1345

CRE

C

1

A.

4

.sI -i

I

i

cellularregulatory regions and isapotentialmechanismto in-crease the viral cell host range.Afunctional enhancer

down-streamof the viral

transcription

initiation site is also consistent with reports indicating the upstream NFKB-binding enhancer

was notessentialforHIV-l infectivity(50). Inparticular,

en-hancer-bound NFKB in vitro (51) or nuclear-translocated NFKB invivo (52)was notsufficienttoactivate HIV-1 gene

transcription. Recently,DNA sequenceswithinthetranscribed noncoding leader ofHIV-l andHIV-2have beenreportedto

play a positive and negative role, respectively, in basal gene

expression (53, 54). Wenotethat the DSE sequencesare

tran-scribed into RNAasistheTat-responsive element. However,

c-Fos,unlike Tat, actsthrough DNAelements in association with otherleucine zipper proteins. Nonetheless, wehavenot

formallyruledoutthepossibilitythat c-FosmaybindtoRNA andactivate LTRtranscription, perhapsina mannersimilarto

the Tat and TAR interaction.

Afunctional TRE/CRE-likesequencemotif in the HIV-I LTR implies that members ofthe Jun/Fos and/or CREB/

ATFfamilies of transcriptional activators play a role in the

regulation ofHIV-1 geneexpression.These

transcription

fac-tors are highly regulated. Their genes are

differentially

ex-pressed(29, 55,56) andtheirproducts havedifferent biological properties(25, 38, 57), whichcan confernegative aswellas

positive

transcriptional

controloftargetgenes

( 19,

22,

38,

58,

59). Recently, Jun/Fos and/orCREB/ATFhave been

impli-cated in the transcription of other

retroviruses,

including

HTLV-1 (60), human foamy virus (61 ), bovine leukemia virus-1 (62), and feline

immunodeficiency

virus-l (63). In

addition, previously recognized HIV-l LTR transactivator

proteinsencodedbyothervirusescanactivategeneexpression

throughfunctionalTRE and CREmotifs(e.g., the productof

theEpstein-Barr early immediategeneBZLF1 [64],which

en-codesaleucine zipper factor similartoc-Fos, andtheXprotein

of hepatitisBvirus[41],whichcanform complexes withATF

proteins[65 ]).

Intracellular signaling pathways that activate the c-Fos-re-sponsiveDSEsites appeartodiffer among human colonic epi-thelialcells. As an immediate early gene, c-Fos plays a crucial

role in theintracellular transduction ofextracellularstimuli; its activation, along with

c-jun,

isimportant forentryintothe cell

cycle(23). Ourresultssuggestthat c-Fosinduction mightbean

important triggerin the activationofHIV- I geneexpression.A

transfectedc-FosexpressionvectorstimulatedHIV-1 LTR-de-pendentgenetranscriptioninconcertwith activators ofPKC in HIV-l-infectible colonepithelial cell lines, butnotin the

noninfectiblecolonepithelialcellline T84.However,T84 cells

did mediate PMA-activated intracellular

signals

to a reporter gene construction containing multiple TREmotifs. Thus, it

appearsthat thec-Fos-responsiveDSEmotifsareactivatedin

HIV-l-infectible colon epithelial cells by acellularsignaling pathway differentfrom thatactivatingclassical Jun/Fos-bind-ingTREelements. Activation of PKCmayinducecell

type-specific posttranslational modifications of c-Fosdirectlyorit

may differentially regulate a dimerization partner of c-Fos, such as a member of the CREB/ATF family (55, 66, 67). Alternatively,itispossiblethat PKC coulddifferentially

influ-enceother regulatoryproteinsinthe threecelllines,for

exam-ple, thoseinvolvedin the modulation of c-Fos and/or

CREB/

ATFbinding activities, includingIP-1, aninhibitorprotein of Fos/Jun that is regulated by extracellular signals (68) or CREM and Ref-l, whichmodulate CREB and

AP-l

binding activities, respectively (66, 69).

HIV-JDSEmotifsappeartodifferfrom classicalTRE

mo-tifs.

Althoughthe DSE motifs in theHIV-l LTR haveahigh

degree

of sequencesimilaritytoclassical TRE motifs and

func-tional AP-1

binding

sites,ourdata suggest that these

motifs,

perhaps because of their context, do not function astypical

AP-l

binding

TREsites.First, recombinant c-Junbound the

HIV-l

DSE sites

quite

poorly comparedwiththe collagenase

TREmotif.

Second,

nuclearprotein complexes formedonthe

HIV-l

DSE motifs

migrated

ingels differently fromthe TRE motifcomplexandwere notblockedby competition withthe TRE sequenceoligonucleotide. Third,in transienttransfection experiments, c-Jun did not synergize with c-Fos to activate

HIV- I LTRtranscription,butratherattenuatedthe c-Fos trans-activation response. Finally, T84 colon epithelialcells could

not mediate PMA or TNFaactivation signals tothe HIV-l DSE motifsto produce efficient c-Fos transactivation of the

HIV-l LTR,but couldreadily mediatePMAactivationsignals

tothecollagenaseTRE motiftotransactivate AP-l-dependent

reporter geneexpression.Collectively, these data indicate that

DSE-l andDSE-2areprobably interactingwithtranscription

factorsconsistingof components otherthan,orin addition to, those of theAP-l /Jun/Fos familytomediate HIV-l gene

ex-pression.

Differential

c-Fos transactivation of SW620, HT29, and

T84 cells correlates with anovel CRE-like binding activity in colonepithelialcells. The ability of c-Fostotransactivate

ex-pression of the HIV-1 LTR in human colon epithelial cells correlated with a CRE-like binding activity. In the T84 cell extracts,oneoftheCREcomplexes was weakly detected

com-paredwith its counterpart in the HT29 and SW620 cell ex-tracts.This latterCRE complexcomigrated with those formed

ontheDSE motifs. Inaddition,thecomigratingCREandDSE

complexes wereblocked when incubated with an anti-c-Fos monoclonal antibody ( 1 8G3), shownpreviouslytospecifically interfere withc-Fos-containinggel shift complexes (47). The secondCREcomplex, which does not comigrate with the DSE

complex,wasunaffectedbythe antibody. Thus, c-Fos isa

com-ponentofboth theCRE and DSE complexes. This is consistent with studiesdemonstrating that,as aheterodimer, c-Fos binds

toCREaswellasTRE (25, 33). It also suggests that the c-Fos transactivation of the HIV- 1 LTR is most likely due to a direct interaction of c-Fos with the DSE motifs.

Although cotransfections with the c-Fos expression plas-mid resulted in high levels oftranscriptional activity from a

construction containing the HIV-l DSE motifs, different

com-binations of other members of the Jun/Fos family failedto

stimulate comparable levels of HIV-1 LTR expression in PMA-orTNFa-stimulated colon epithelial cells. Because Jun formsamorestable complex as a heterodimer with Fos thanas

aJun-Jun homodimer, in many cell systems c-Jun cotrans-fected with c-Fos stimulates higher levels of gene expression than c-jun alone ( 19, 25, 27). However, in the colonic epithe-lial cells, cotransfected c-Jun reduced expression of the HIV-l LTR stimulated by c-Fos transactivation. Because c-Fos

re-quires a partner to bind DNA asaheterodimer (19),it is

con-ceivable in our studies that c-Jundisplaced,byadirect compe-tition or squelching mechanism, the protein dimerization partnerthat combines with c-Fostoform theHIV-1DSE acti-vation complex. Thisputative dimerization partner of c-Fos might beamember of theCREB/ATFfamilythat cross-com-bines with c-Fos to form a hybrid dimer with an altered

se-quence binding specificity in which CRE-like, rather than

TRE-like, sequences are preferred (32-34). This possibilityis

consistentwithourDNA

binding

studies

demonstrating

selec-tivecompetitionof_{the DSE motifs with classical CRE motifs.}

However,

wenotethattwomembersoftheCREB/ATF

_family

tested inour _{system, CREB} and ATF-2, likewise attenuated c-Fos

_{transactivation,}

_indicating

_{that c-Fosmaydimerize with}

other yet unknowntranscriptionfactorstoactivate the HIV-1

LTR

_(Roebuck,

K. _A., unpublisheddata). In MHC class II

gene promoters, c-Fos has been shown to dimerize with an

unknownproteinfactor in

binding

tothe nonconsensus TRE

X2box

(70).

Itis also possible that c-Fosinteractswith

non-AP-1

/CREB/ATF proteins,

such as those from the

_C/EBP

andNF-IL6_{basic-leucine zipper family}

_(71).

Acknowledaments

WethankC.Lu,S.

Lundgren,

M._Arce,S._Kim,G._Martinez,J._Han, and T. _Mchezajifortechnical_assistance;T. Smealfor

_providing

_the

trpE-c-Jun protein;K. A._Jones,A.

_Siddiqui,

M.G.

_Rosenfeld,

_{A. B.}

Rabson,M. C.Nehls,andM.Karin for

_{providing plasmids;}

_{and R. A.}

RippeforthePIEoligonucleotide.We alsothankE.

Morzycka-Wrob-lewska for criticalreadingof the

manuscript

before submission. ThisresearchwassupportedbyNational InstitutesofHealthgrant

DK-40582,agrant fromthe

_University

ofCalifornia_{Universitywide} AIDS Research_Program

_(UARP),

anda

Fellowship

toK. A._Roebuck

fromtheUARP.

References

1.Greene,W. C. 1991. The molecularbiologyof human_{immunodeficiency} virustype1infection.N._Engl.J.Med.324:308-317.

2.Vaishnav,Y. N.,and F. Wong-Staal. 1991.The_{biochemistryof AIDS.} Annu. Rev.Biochem. 60:557-630.

3.Varmus,H. 1988.Regulationof_HIV-1and HTLVgeneexpression.Genes Dev.2:1055-1062.

4.Gaynor,R. B. 1992. Cellular_{transcription}factors involved_{in the} regula-tion ofHIV-Igeneexpression.AIDS(Phila.).6:347-363.

5.Nabel, G.,andD.Baltimore. 1987. Aninducibletranscriptionfactor

acti-vates_expressionof human_{immunodeficiency}virusin T-cells. Nature(Lond.). 326:711-713.

6.Jones,K.A.,P. A.Luciw,andN._Duchange.1988. Structural_arrangement oftranscriptioncontroldomainswithin the5'-untranslated leaderregionsof the

HIV-IandHIV-2promoters.GenesDev._2:1101-1111.

7.Price,R.W.,B.Brew,J.Sidtis,M._Rosenblum,A.C.Scheck,and P._Cleary. 1988. Thebrainin AIDS: centralnervous_{systemHIV-1 infectionandAIDS}

dementiacomplex.Science(Wash.DC).239:586-592.

8. Rogers, V. D.,and M. F.Kagnoff. 1988._{Acquiredimmunodeficiency} syndromeanddiseaseof thegastrointestinaltract.Immunol._AllergyClin.N. Am. 8:451-467.

9.Nelson,J.A.,C. A._Wiley,C._{Reynolds-Kohler,}C. E.Reese,W.

Margaret-ten, and J. A.Levy. 1988. Human

_{immunodeficiency}

virus detectedin bowel

epitheliumfrom_patientswithgastrointestinalsymptoms. Lancet.i:259-262. 10.Levy,J.A.,W.Margaretten,and J.Nelson. 1989.Detection ofHIV in enterochromaffincells in the rectalmucosaofanAIDSpatient.Am. J. Gastroen-terol. 84:787-789.

11.Fox,C.H.,D._Kotler,A._Tierney,C. S._Wilson,andA.S. Fauci. 1989. Detection of HIV in RNA in thelaminapropriaofpatientswith AIDS and

gastrointestinaldisease. J._{Infect.Dis. 159:467-471.}

12.Moyer,M.P.,R. I.Huot,A.Ramirez,Jr.,S._Joe,M. S.Meltzer,and H. E.

Gendelman.1990.Infection of human_{gastrointestinal}cellsbyHIV-1. AIDSRes. Hum.Retroviruses._6:1409-1415.

13.Heise, C.,S.Dandekar,P.Kumar,R._Duplanteir,R. M.Donovan,and C. H. Halsted. 1991.Human

_{immunodeficiency}

virusinfectionof_enterocytes andmononuclearcells in human_jejunalmucosa._{Gastroenterology.}

100:1521-1527.

14.Amerongen,H.M.,R.Welzin,C. M._Farnet,P._Michetti,W.A.Haseltine,

and M. R. Neutra. 1991.Transepithelialtransportof HIV-I_byintestinalM cells:

a mechanism _{for transmission of AIDS. J. AcquiredImmuneDefic.} Syndr. 4:760-765.

15.Fantini,J.,N.Yahi,S.Baghdiguian,and J.-C.Chermann.1992.Human colonepithelialcells_productivelyinfectedwithhuman_{immunodeficiencyvirus} showimpaireddifferentiation_{and altered secretion. J. Virol. 66:580-585.}

16.Phillips,D.M.,and A.S.Bourinbaiar.1992.MechanismsofHIVspread from_lymphocytesto_{epithelia.Virology. 186:261-273.}

17.Gill,J.M.,L. R._Sutherland,D.L._Church,and theUniversityofCalgary

Gastrointestinal/HIVstudy group. 1992.Gastrointestinaltissue cultures for HIV inHIV-infected/AIDSpatients. AIDS(Phila.). 6:553-556.

18. Omary, M. B., D. A. Brenner, L. Y.deGrandpre,K. A.Roebuck,D. D. Richman,and M. F._Kagnoff.1991. Humanimmunodeficiencyvirus-Iinfection andexpressionin human colonic cells:infectionand expression in CD4positive and_CD4negative cell lines. AIDS (Phila.).5:275-281.

19. Karin, M. 1991. TheAPIcomplex and its role intranscriptionalcontrol by protein kinase C. In TheHormonal Control Regulationof Gene

Transcrip-tion. P. Cohen and J. G. Foulkes,editors. ElsevierScience,Amsterdam,The Netherlands.235-253.

20. Angel, P., M.Imagawa,R. Chiu, B.Stein,R. J.Imbra,H. J.Rahmsdorf,C. Jonat, P. Herrlich, and M. Karin. 1987. Phorbolester-induciblegenescontaina

common cis element recognized byaTPA-modulatedtrans-actingfactor. Cell. 49:729-739.

21.Brenner,D. A., M. O'Hare, P.Angel,M.Chojkier,and M. Karin. 1989. Prolonged activationof Jun andcollagenasegenesbyTNFa. Nature

_(Lond.).

337:661-663.

22.Castellazzi,M., G. Spyrou, N. La Vista, J.-P.Dangy,F.Piu,M.Yaniv,

and G. Brun. 1991.Overexpressionofc-jun, junB,orjunDaffects cellgrowth

differently.Proc.Natl.Acad. Sci. USA.88:8890-8894.

23. Kovary, K., and R. Bravo. 1991. Thejunand fosproteinfamiliesareboth required for cell cycleprogressioninfibroblasts.Mol. Cell. Biol._9:4466-4472.

24. Franza, B. R., F. J. Rauscher, S. F.Josephs,and T. Curran. 1988. The fos complex and fos related antigensrecognizesequenceelements that contain API binding sites. Science (Wash. DC).239:1150-1153.

25. Ryseck, R. P., and R. Bravo. 1991.c-jun, JunB,and JunD differ in their binding affinities to AP- 1 and CRE consensussequences:effects of Fosproteins. Oncogene.6:533-542.

26. Smeal, T., P. Angel, J. Meek, and M. Karin. 1989. Differentrequirements for the formation of Jun:Jun and Jun:Fos complexes. Genes Dev.3:2091-2100. 27. Zerial, M., L. Toschi, R. P.Ryseck,M.Schuermann,R.Muller,and R. Bravo. 1989. The product ofanovel growth factor activatedgene,fosB,interacts with JUN proteinsenhancingtheir DNA_{binding activity.}EMBO(Eur.Mol.

Biol._{Organ.) J.}8:805-813.

28.Spandidos,D. A., M.Yiagnisis,and A. Pintzas. 1989. Human immunode-ficiency virus long terminal repeat responds totransformationbythemutantT24

H-rasloncogene and it contains multiple AP-1bindingTPA-inducible

consen-sussequenceelements.AnticancerRes.9:383-386.

29. Flint, K. J., and N. C. Jones. 1991. Differential regulation of three members of the ATF/CREB family ofDNA-binding proteins. Oncogene. 6:2019-2026.

30.Montminy,M. R., K. A.Sevarino,J. A.Wagner,G.Mandel,and R. H. Goodman.1986.Identificationofacyclic-AMP-responsiveelement within the ratsomatostatingene.Proc. Natl._Acad.Sci. USA.83:6682-6686.

31. Sakurai, A., T. Maekawa, T. Sudo, S.Ishii,and A. Kishimoto. 1991. Phosphorylation of cAMP response_{element-binding} protein, CRE-BPl, by

cAMP-dependentprotein kinase and protein kinase C.Biochem.Biophys.Res. Commun. 181:629-635.

32. Benbrook, D. M., and N. C. Jones. 1990.Heterodimer formationbetween CREB and JUNproteins. Oncogene. 5:295-302.

33. Hai, T., and T. Curran. 1991.Cross-familydimerizationoftranscription

factors Fos/Jun and ATF/CREBalters DNAbinding specificity.Proc. Natl. Acad.Sci. USA.88:3720-3724.

34. Ivashkiv, L. B., H.-C. Liou, C. J.Kara,W. W.Lamph,I.Verma,and L. H. Glimcher. 1990.mXBP/CRE-BP2andc-junformacomplexwhich bindstothe cyclic AMP, but not to the 1_{2-O-Tetradecanoylphorbol-13-Acetate,response} ele-ment. Mol. Cell. Biol. 10:1609-1621.

35.Macgregor,P. F., C. Abate, and T. Curran. 1990. Directcloningofleucine zipper proteins: Jun bindscooperativelyto the CRE with

_CRE-BPl.

_Oncogene. 5:451458.

36.Hoeffler,J. P., J. W.Lustbader,and C. Y. Chen. 1991.Identificationof multiple nuclear factors that interact withcyclicadenosine_{3',5'-monophosphate} responseelement-bindingprotein and_{activating transcription}factor-2by

pro-tein-protein interactions.Mol.Endocrinol. 5:256-266.

37. Dinter, H., R. Chiu, M.Imagawa,M.Karin,and K. A. Jones. 1987. In vitroactivationof_{the HIV- 1 enhancer in extracts from cells treated withphorbol} ester tumor promoter. EMBO (Eur.Mol.Biol.Organ.)J.6:4067-4071.

38. Chiu, R., P. Angel, and M. Karin. 1989.jun-Bdiffers in itsbiological propertiesfrom and isanegativeregulatorofc-jun.Cell.59:979-986.

39.Nehls,M. C., R. A.Rippe,L.Veloz,and D. A.Brenner.1991. Transcrip-tional factors nuclear factor 1 and Sp 1 interact with the murinecollagenal(I)

promoter.Mol.Cell.Biol. 11:4065-4073.

40. Ross, E. K., A. J.Buckler-White,A. B.Rabson,G.Englund,and M. A. Martin. 1991.Contributionof NFKB and_{Splbinding motifs}tothereplicative

capacity of human immunodeficiencyvirustype 1: distinct patterns of viral growth aredeterminedbyT-celltypes. _J.Virol.65:4350-4358.

41. Siddiqui, A., R. Gaynor, A. Srinivasan, J. Mapoles,andR. W. Farr. 1989. Trans-activation_{of viral enhancers including the long terminal repeatofthe} humanimmunodeficiencyvirus by the hepatitis B virus X protein. Virology.

169:479-484.

42. Nelson, C., V. R. Albert, H. P. Elsholtz, L.I.Lu,andM. G. Rosenfeld.