J. Mol. Biol. (1996) 263, 1–7

COMMUNICATION

Effect of the Position of TAR on TranscriptionalActivation by HIV-1 Tat in Vivo

Stephanie Wright 1* and Craig Luccarini 2

Efficient expression of the human immunodeficiency virus (HIV) genome1Department of Biochemistryrequires the viral-encoded transactivator Tat. Tat interacts with the highlyand Molecular Biologystructured trans-activation-response (TAR) RNA that is found at the 5' endUniversity of Leedsof all viral transcripts, and mediates the formation of transcriptionLeeds, LS2 9JT, Englandcomplexes that are capable of elongation through the entire length of the2Wellcome/CRC Institute of viral genome. By placing TAR immediately downstream from the P2

Cancer and Development promoter of the mouse c-myc gene, we have previously shown that TatBiology, Tennis Court Road can also direct transcriptional elongation through potential sites ofCambridge, CB2 1QR premature termination within c-myc in transfected HeLa cells. We nowEngland demonstrate that Tat can activate c-myc transcription when TAR is

positioned internally within the c-myc transcript at distances up to 353 ntdownstream from the P2 promoter. We show that Tat can also activatetranscription from the c-myc P1 promoter, which is located 165 ntupstream from P2 in these hybrid gene constructs. These novel findingsshow that Tat can activate transcription in vivo when TAR is positionedat distances up to 518 nt downstream from the site of transcriptionalinitiation. The ability of TAR to mediate Tat-activated transcription overdistances greater than previously appreciated has important implicationsfor the mechanism of action of Tat.

7 1996 Academic Press Limited

*Corresponding author Keywords: elongation; HIV; myc; Tat; transcription

The regulation of transcriptional elongation playsa major role in the expression of eukaryotic genes(reviewed by Bentley, 1995; Kerppola & Kane, 1991;Spencer & Groudine, 1990; Wright, 1993). Prema-ture transcriptional termination was originallydemonstrated near the 5' ends of genes usingnuclear run-off transcription analysis in isolatedmammalian nuclei; examples include c-myc(Bentley & Groudine, 1986), c-fos (Collart et al.,1991), and the HIV genome (Kao et al., 1987). Inmany situations, changes in steady-state levels ofRNA correlate with fluctuations in the degree oftranscriptional elongation past the 5' end of thegene, rather than with changes in transcriptionalinitiation (e.g Bentley & Groudine, 1986). Shortprematurely terminated transcripts of cellulargenes are unstable in mammalian cells, and cannotbe detected in steady-state cytoplasmic mRNA;such RNAs are, however, stable in the Xenopusoocyte, where they may be analysed uponexpression of cloned microinjected genes (Bentley& Groudine, (1988). In contrast, prematurelyterminated HIV transcripts are highly structuredand may be readily detected in the cytoplasmic

RNA (Kao et al., 1987). Sites of prematuretranscriptional termination within the mouse c-mycgene have been mapped near the 3' end of exon 1(Bentley & Groudine, 1988; Nepveu et al., 1987;Wright & Bishop, 1989), and between the twopromotors, P1 and P2, of the gene (Meulia et al.,1992; Wright et al., 1991).

Efficient transcriptional elongation through theintegrated HIV genome requires the viral-encodedtransactivator TAT, which acts via interaction withthe highly structured trans-activation-response(TAR) RNA that is found at the 5' end of all viraltranscripts (reviewed by Cullen, 1991; Frankel,1992; Karn, 1991; Karn & Graeble, 1992); in theabsence of Tat, transcription initiating from theviral LTR terminates prematurely within severalhundred nucleotides and only short RNAs arerecovered (Kao et al., 1987). The regulation oftranscriptional elongation by Tat has been demon-strated both in transfected human cells (Kao et al.,1987) and in reconstituted in vitro transcriptionreactions (Graeble et al., 1993; Kato et al., 1992;Lapsia et al., 1989, 1993; Marciniak & Sharp, 1991).

The similarities between transcriptional control

0022–2836/96/410001–07 $25.00/0 7 1996 Academic Press Limited

Distance-dependence of Tat Action2

in cellular genes and in HIV suggests that Tatmight be representative of a class of cellularprotein that activates gene expression by enhanc-ing the efficiency of transcriptional elongationand/or by overcoming premature termination. Insupport of this model, we have shown that Tat candirect transcriptional elongation through sites ofpremature termination within the c-myc genewhen the cis-acting HIV TAR element is insertedimmediately downstream from the c-myc P2promoter (Wright et al., 1994). It has also beenshown that Tat-modified transcription complexescan read through artificial terminators in vitro(Graeble et al., 1993), and that Tat can directtranscriptional readthrough of specific terminationsites at the 3' ends of snRNA genes when TAR ispositioned immediately downstream from thecorresponding promoter (Ratnasabapathy et al.,1990).

Efficient activation of the HIV LTR by Tat intransient expression assays has been reported torequire the precise positioning of TAR immediately3' to the site of transcriptional initiation (Selby et al.,1989). It has therefore been suggested thatactivation requires the interaction of Tat withtranscription components present only in thosecomplexes located close to the promoter, or that Tatselectively stimulates transcriptional initiation by apolymerase with specific elongation properties.Alternatively, however, the effect of the position ofTAR in this system may merely reflect theextremely low processivity of transcription initiat-ing from the HIV LTR in the absence of Tat; thetarget TAR RNA would not be transcribed ifpositioned far downstream from the transcriptionstart site. If this latter possibility were true, wereasoned that Tat-activation of a c-myc/TARhybrid gene would require a less precise position-ing of TAR, since ‘‘unactivated’’ transcription from

Figure 1. The effect of Tat on transcription from the P1and P2 promoters of a hybrid mouse c-myc-P2/TAR genein transfected HeLa cells. A, representation of the hybridmouse c-myc-P2/TAR gene. HIV-1SF2 TAR sequenceswere inserted immediately downstream from the P2promoter of a mouse c-myc genomic clone. The plasmidcontained a 10.5 kb KpnI fragment that included themouse c-myc coding sequences together with 5 kb and2 kb of 5' and 3' flanking sequences, respectively; detailsof this construction have been described (Wright et al.,1994). The plasmid contained the bacterial aminogly-coside phosphotransferase (AGPT) gene that confersneomycin resistance (neor) upon stably transfectedmammalian cells. Filled boxes denote c-myc exonsequences, hatched boxes denote 5' and 3' flankingsequences. The positions of the P1 and P2 promoters ofthe c-myc gene are indicated. B, Levels of mouse c-myctranscript were measured in HeLa transfectants thatcontained the hybrid c-myc-P2/TAR gene. Transcriptswere analysed by ribonuclease protection mapping usingthe uniformly labelled probes that spanned the P1 andP2/TAR regions of the hybrid gene (probe A) or the P1region (probe B) as diagrammed beneath the gel. Thepresence of wild-type Tat (SV-TAT) or mutant Tat(SV-TATzx) in the transfectants is indicated at the top ofthe gel (lanes 1 to 3). Lane 4 shows hybridisation of theprobe to non-transfected HeLa RNA, and lane 5 shows aportion of the RNase protection probe that had not beenhybridised or digested with RNase. Positions ofprotected bands corresponding to transcripts initiating

from the P1 and P2 promoters are indicated. Markers (m)were end-labelled MspI fragments from the plasmidpBR322. Bands corresponding to P1 and P2-initiatedtranscripts were visible in lanes 1 and 3 upon longerexposure of the gel. C, Levels of AGPT (neor) transcriptwere measured in HeLa transfectants by ribonucleaseprotection analysis. RNA samples were as in B and areindicated at the top of the gel.

Plasmids were stably introduced into HeLa cells bycalcium phosphate transfection (Wigler et al. 1979) andselection with neomycin (1 mg/ml). Cotransfectionscontained 5 mg of c-myc-P2/TAR together with 20mg ofplasmid expressing wild-type or mutant Tat driven bythe SV40 promoter (SV-TAT or SV-TATzx respectively;Kao et al. 1987). For each transfection, RNA was isolatedfrom a pool derived from approximately 200 indepen-dent colonies using methods described previously(Wright & Bishop, 1989). RNase protections were carriedout as described above; hybridisations contained 5mg oftotal RNA and were carried out at 52°C (Wright &Bishop, 1989). The plasmid used to generate the probeswere created by cloning the indicated restrictionfragments into the vector pSP72 (Promega Biotech).

Distance-dependence of Tat Action 3

the c-myc promoter is more processive than fromthe HIV LTR, i.e. the target TAR RNA would betranscribed when positioned at distances furtherdownstream from the c-myc promoter. We there-fore attempted to distinguish between theseimportant mechanistic alternatives by examiningTat-mediated activation of hybrid gene constructsin which TAR is positioned at varying distancesdownstream from the c-myc promoter.

Tat activates transcription from both P1 and P2promoters of a hybrid c-myc/TAR gene

HIV TAR sequences were inserted immediatelydownstream from the site of transcriptionalinitiation at the P2 promoter of a mouse c-mycgenomic clone (Figure 1A) described by Wrightet al. (1994); this generated a hybrid genedesignated c-myc-P2/TAR (previously designatedc-myc/TAR by Wright et al. (1994). c-myc-P2/TARwas stably transfected into HeLa cells in thepresence or absence of plasmids expressingwild-type Tat (SV-TAT) or mutant Tat (SV-TATzx).Mutant Tat was inactive as a result of a frameshiftmutation introduced at +21 (Kao et al., 1987). Apool comprising approximately 200 colonies wasgenerated for each transfection. RNA was isolatedand the expression of the transfected c-myc-P2/TAR gene was determined by RNase protectionanalysis.

As shown previously, expression of the hybridc-myc-P2/TAR gene was increased 20 to 50-fold bythe presence of Tat when measured by hybridis-ation to a probe derived from exon 1 of the gene.Mutant Tat did not activate c-myc-P2/TAR, and Tathad no effect on expression of a c-myc gene that didnot contain TAR sequences (not shown; See Wrightet al., 1994).

As an initial investigation into the effect of theposition of TAR on activation of a hybridc-myc/TAR gene by TAT, we determined whetherTat also regulated transcription from the upstreamc-myc promoter, P1, which is located 165 ntupstream from TAR in the hybrid c-myc-P2/TARgene (Figure 1A).

The expression of the c-myc-P2/TAR hybridgene was analysed in RNase protections byhybridisation to a probe that spanned the P1 andP2/TAR regions of the gene (Figure 1B, probe A).Unexpectedly, levels of both P1 and P2 initiatedtranscripts from the c-myc-P2/TAR hybrid genewere increased by the presence of Tat (lanes 1 and2); mutant Tat had no effect on either transcript(lane 3). Tat caused a similar activation of both P1and P2-initiated transcripts (20 to 50-fold). Theabsolute level of P1-transcripts represented ap-proximately 5% of the total c-myc mRNA derivedfrom the c-myc-P2/TAR gene, in accordance withthe normal ratio of P1 to P2-initiated c-myctranscripts in transfected HeLa cells. A similar levelof activation of P1-transcripts by Tat was observedusing an RNase protection probe that covered the5' flanking sequences and P1 promoter region only

of the construct (Fig 1B, probe B). All transfectantsexpressed similar levels of transcript from theAGPT (neor) gene, which was present on thec-myc-P2/TAR plasmid (Figure 1C).

The activation of the P1 promoter of thec-myc-P2/TAR hybrid gene by Tat was surprisingin view of the fact that P1 is located 165 nt upstreamfrom the TAR region in this construct, and thatTat-dependent activation of HIV requires theprecise positioning of TAR immediately down-stream from the site of transcriptional initiation.Activation of the P1 promoter could reflect anon-specific effect resulting from an increase intranscription from the P2 promoter, or couldrepresent a genuine Tat-dependent activation viathe TAR element located 165 nt downstream.

Tat activates transcription of a hybridc-myc/TAR gene when TAR is positioned atdistances up to 353 nt downstream from theP2 promoter

The activation of both P1 and P2 promoters inc-myc-P2/TAR suggested that Tat may regulatetranscription when TAR is located far downstreamfrom the site of transcriptional initiation in c-myc.We therefore inserted TAR at various positionsdownstream from the P2 promoter of the mousec-myc gene, and analysed the ability of Tat toactivate transcription from both the P1 andP2-promoters in transfected HeLa cells.

TAR was inserted into a mouse c-myc genomicclone in either the sense (A) or antisense (B)orientation at various distances downstream fromthe P2 promoter (Figure 2a). Constructs weredesignated according to the distance between theP2 promoter and TAR insert (nt), and theorientation of TAR insert (A or B). These hybridgenes were stably transfected into HeLa cells, eachin the presence or absence of plasmids thatexpressed wild-type Tat (SV-TAT) or mutant Tat(SV-TATzx). The effect of Tat on expression of thevarious hybrid c-myc/TAR genes was analysed byRNase protection mapping.

Figure 2B shows an analysis of the expression ofhybrid c-myc genes in which TAR is positioned ineither the sense (A) or antisense (B) orientation atthe HaeIII site located 173 nt downstream from theP2 promoter. RNA from transfected cells wasanalysed by hybridization to an RNase protectionprobe that spanned the P1, P2 and TAR regions ofthe construct. Tat activated the expression fromboth P1 and P2-promoters when TAR waspositioned in the sense orientation within c-myc(c-myc-173/TAR(A); lanes 1 and 2); mutant Tat hadno effect on c-myc-173/TAR(A) expression (lane 3),and Tat did not affect the expression of the hybridc-myc/TAR gene into which TAR was inserted inthe antisense orientation (c-myc-173/TAR(B); lanes5 to 7). The ratio of P1 to P2-initiated transcriptsfrom c-myc-173/TAR was reproducibly higherthan from other hybrid c-myc/TAR genes; we donot have a clear explanation of this observation.

Distance-dependence of Tat Action4

Figure 2(a–d) legend opposite

Distance-dependence of Tat Action 5

Expression of a c-myc gene into which TAR wasinserted 353 nt downstream from the P2 promotershowed a similar induction by Tat. Again, Tatactivated both P1 and P2-promoters, and mutantTat had no effect (Figure 2C, lanes 1 to 3). c-mycconstructs in which TAR was inserted in theantisense orientation did not respond to Tat (Fig-ure 2C, lanes 5 to 7). Tat can therefore activatec-myc transcription when TAR is positioned up to353 nt downstream from the P2 promoter, and canactivate the P1 promoter that is located 518 ntupstream from TAR in these constructs. Activationof both P1 and P2-promoters by Tat was two- tofour-fold less efficient when TAR was positioned353 nt downstream from P2, as compared withbeing immediately juxtaposed to the site oftranscriptional initiation at the P2 promoter.

When TAR was positioned 721 nt or 1060 ntdownstream from the c-myc P2-promoter, Tat wasunable to activate transcription from either the P1or P2 promoters (Figure 2D).

Our data show that Tat may activate transcrip-tion of a hybrid c-myc/TAR gene when TARis positioned up to 518 nt downstream from thesite of transcriptional initiation (i.e. the distancebetween P1 and TAR in c-myc-353/TAR(A)). Thisis in contrast to the original results of Selby et al.(1989), who showed that activation of HIV by Tatrequired the positioning of TAR immediatelydownstream from the promoter; insertion of only88 nt between the transcription start site and TARin this system resulted in a 5.5-fold reductionactivation by Tat. More recent evidence, however,suggests that TAR may function at distances furtherdownstream from the HIV promoter than originallyappreciated. Thus, Braddock et al. (1994) haveshown that an intact TAR element was functionalin vivo when placed downstream from a defectiveTAR in a construct driven by the HIV LTR; TARwas functional when placed 90 nt downstream

from the start site of transcription, although theprecise distance requirements were not examined.In extensive experiments aimed at determining theeffect of the position of TAR on transcriptionalactivation of the HIV LTR by Tat in vitro, Churcheret al. (1995) have shown that a mutated TAR in thenormal wild-type position could be rescued by awild-type TAR element placed up to 200 ntdownstream from the transcription start site. In theexperiments described here, the effect of theposition of TAR was examined in the context of aheterologous promoter using an in vivo system; theeffects were thus independent of other regulatoryelements in the HIV LTR or of properties of in vitrotranscription extracts.

The magnitude of the transcriptional activationby Tat was greater in the c-myc/TAR hybrid genesin vivo than that observed for HIV LTR in vitro; TARwas also functional when placed at greaterdistances downstream from the start site inc-myc/TAR as compared with HIV (Rittner et al.,1995; Churcher et al., 1995). These differences mayreflect differences in elongation properties betweenthe in vivo systems and in vitro extracts. The fact thatTAR is functional at a greater distance from thepromoter in c-myc/TAR as compared with HIVmay reflect the fact that ‘‘unactivated’’ transcriptionfrom the c-myc promoter is more processive thanthat from HIV; synthesis of the target TAR RNAand recruitment of Tat would thus be possible atgreater distances from the promoter in c-myc/TARas compared with HIV.

Different models for the mechanism of action ofTat have been proposed. Previous suggestions thatTAR must be positioned immediately adjacent tothe transcription start site were interpreted to implythat Tat contacted transcription complexes locatedat the promoter, with the elongation properties oftranscription thus being determined at or near thesite of initiation. Tat was presumed to interact with

Figure 2. Tat activates transcription of hybrid c-myc/TAR genes when TAR is positioned at distances up to 353 ntdownstream from the c-myc P2 promoter. A, A representation of the hybrid c-myc/TAR genes. The 67 nt HIV-1SF2 TARelement was inserted into the mouse c-myc genomic clone (described for Figure 1A) at the indicated restriction enzymesites in exon 1 (HaeIII or XhoI) or intron I (MscI to HcII). Insertions were in either the sense (A) or antisense (B)orientation with respect to c-myc transcription. Constructs were designated according to the distance between the P2promoter and TAR insert (nt), and the orientation of the TAR insert (A or B). Filled boxes denote c-myc exon 1sequences, stippled boxes denote TAR sequences. B, The effect of Tat on expression of the hybrid c-myc-173/TARgenes. Levels of mouse c-myc transcript were measured in HeLa transfectants that contained the hybridc-myc-173/TAR (A) or (B) genes, as indicated at the top of the gel. Transcripts were analysed by ribonuclease protectionanalysis using a uniformly labelled probe that detected transcription from both P1 and P2 promoters as diagrammed.The presence of wild-type Tat (SV-TAT) or mutant Tat (SV-TATzx) in the transfectants is indicated at the top of thegel (lanes 1 to 3 and 5 to 7). Lanes 4 and 8 show hybridisation of the probes to non-transfected HeLa RNA, and lanes9 and 10 show portions of the probes that had not been hybridised or digested with RNase. The positions of protectedbands corresponding to transcripts initiating from the P1 and P2 promoters are indicated. Markers (m) wereend-labelled MspI fragments from the plasmid pBR322. Bands corresponding to P1 an P2-initiated transcripts werevisible in lanes 1 and 3 upon longer exposure of the gel. C, The effect of Tat on expression of the hybrid c-myc-353/TARgenes. The expression of the c-myc-353/TAR(A) and (B) genes was analysed as for B using the RNase protection probesdiagrammed beneath the gel. D, The effect of Tat on expression of the hybrid c-myc-721/TAR and c-myc-1060/TARgenes. The expression of the c-myc-721/TAR and c-myc-1060/TAR genes was analysed as for B using an RNaseprotection probe that detected transcription across mouse c-myc exon 1 as diagrammed beneath the gel.

Transfection of HeLa cells and ribonuclease protection mapping was as described for Figure 1, except thathybridisations contained 40 mg of RNA. Gels in B and C were exposed for 48 hours; gels in D were exposed for sevendays.

Distance-dependence of Tat Action6

transcription complexes at the promoter by the‘‘looping back’’ of the RNA to which it is attached viaTAR, and would thus interact with complexes otherthan on the nascent transcript to which it is attached.According to this model, Tat would be required toloop back over distances greater than 500 nt in orderto activate the P1 promoter of the c-myc-353/TARhybrid gene described here; we feel that this ismechanistically unlikely. Alternatively, it has beensuggested that Tat is introduced into the transcrip-tion complex during its transit through TAR(Churcher et al., 1995), thus resulting in the modifica-tion of the transcription complex to an ‘‘elongationcompetent’’ form; in this situation, Tat would mostlikely affect only the transcription complexes on thenascent transcript to which it is attached via TAR.The distance from a given promoter within whichTAR may mediate Tat-activation would depend onthe processivity of transcription from the promoterin the absence of Tat; a gene would be responsiveto Tat only if TAR is positioned at a location with-in which elongation-incompetent complexes arefound. Most elongation-incompetent transcriptioncomplexes detach from a gene within a certaindistance from the start site, with complexes locatedfurther downstream being competent for elongationthrough the remainder of the gene and not furtherresponsive to Tat introduced via a downstream TARelement. In the case of c-myc, most elongation-in-competent complexes have detached within 600 ntfrom the P2 promoter, and the introduction of Tat at721 nt (c-myc-721/TAR) therefore has no effect ontranscription. Our data support the model wherebyTat regulates transcription via introduction into thetranscription complex during its transit throughTAR. It should be noted that TAR may, however,serve multiple functions in addition to the deliveryof Tat: these include the pausing of polymerase at thehighly structured stem-loop region, and theintroduction of factors other than Tat (Sheline et al.,1991; Wu et al., 1991).

AcknowledgementsThis work was supported by Project grants SP2194/

0101 and SP2194/0201 from the Cancer ResearchCampaign. We thank Matija Peterlin and Sohail Qureshifor many useful discussions and comments on themanuscript.

ReferencesBentley, D. L. (1995). Regulation of transcriptional

elongating by RNA polymerase II. Curr. Opin. Genet.Dev. 5, 210–216.

Bentley, D. L. & Groudine, M. (1986). A block toelongation is largely responsible for decreasedtranscription of c-myc in differentiated HL60 cells.Nature, 321, 702–706.

Bentley, D. L. & Groudine, M. (1988). Sequencerequirements for premature termination of transcrip-tion in the human c-myc gene. Cell, 53, 245–256.

Braddock, M., Powell, R., Sutton, J., Kingsman, A. J. &Kingsman, S. M. (1994). Orientation-specific ciscomplementation by bulge- and loop-mutatedhuman immunodeficiency virus type 1 TAR RNAs.J. Virol. 68, 8396–8400.

Churcher, M. J., Lowe, A. D., Gait, M. J. & Karn, J. (1995).Human immunodeficiency virus type-1TAR RNAelement is functional when displaced downstream ofthe start of transcription. Proc. Natl Acad. Sci. USA,92, 2408–2412.

Collart, M. A., Tourkine, N., Belin, D., Vassalli, P.,Jeanteur, P. & Blanchard, J. M. (1991). C-fos genetranscription in murine macrophages is modulatedby a calcium dependent block to elongation in intron1. Mol. Cell. Biol. 22, 2826–2831.

Cullen, B. R. (1991). Regulation of HIV-1 gene expression.FASEB J. 5, 2361–2368.

Frankel, A. D. (1992). Activation of HIV transcription byTat. Curr. Opin. Genet. Dev. 2, 293–298.

Graeble, M. A., Churcher, M. J., Lowe, A. D., Gait, M. J.& Karn, J. (1993). Human immunodeficiency virustype 1 transactivator protein, tat, stimulates tran-scriptional read-through of distal terminator se-quences in vitro. Proc. Natl. Acad. Sci. USA, 90,6184–6188.

Kao, S. Y., Calman, A. F., Luciw, P. A. & Peterlin, B. M.(1987). Anti-termination of transcription within thelong terminal repeat of HIV-1 by tat gene product.Nature, 330, 489–493.

Karn, J. (1991). Control of human immunodeficiencyvirus replication by the tat, rev, nef and proteasegenes. Curr. Opin. Immunol. 3, 526–536.

Karn, J. & Graeble, M. A. (1992). New insights into themechanism of HIV-1 transactivation. Trends. Genet. 8,365–8.

Kato, H., Sumimoto, H., Pognonec, P., Chen, C.-H.,Rosen, C. A. & Roeder, R. G. (1992). HIV-1 Tatacts as a processivity factor in vitro in conjunctionwith cellular elongation factors. Genes. Dev. 6,655–666.

Kerppola, T. K. & Kane, C. M. (1991). RNA polymerase:regulation of transcript elongation and termination.FASEB J. 5, 2833–2842.

Lapsia, M. F., Rice, A. P. & Mathews, M. B. (1989). HIV-1Tat protein increases transcriptional initiation andstabilizes elongation. Cell, 59, 283–292.

Lapsia, M. F., Wendel, P. & Mathews, M. B. (1993). HIV-1Tat overcomes inefficient transcriptional elongationin vitro. J. Mol. Biol. 232, 732–746.

Marciniak, R. A. & Sharp, P. A. (1991). HIV-1 Tat proteinpromotes formation of more processive elongationcomplexes. EMBO J. 10, 4189–4196.

Meulia, T., Krumm, A., Spencer, C. & Groudine, M.(1992). Sequences in the human c-myc P2 promoteraffect the elongation and premature termination oftranscripts initiated from the upstream P1 promoter.Mol. Cell. Biol. 12, 4590–4600.

Nepveu, A., Marcu, K. B., Skoultchi, A. I. & Lachman,H. M. (1987). Contributions of transcriptional andpost-transcriptional mechanisms to the regulation ofc-myc expression in mouse erythroleukemia cells.Genes Dev. 1, 938–945.

Ratnasabapathy, R., Sheldon, M., Johal, L. & Hernandez,N. (1990). The HIV-1 long terminal repeat containsan unusual element that induces the synthesis ofshort RNAs from various mRNA and snRNApromoters. Genes Dev. 4, 2061–2074.

Rittner, K., Churcher, M. J., Gait, M. J. & Karn, J. (1995).The human immunodeficiency virus long terminal

Distance-dependence of Tat Action 7

repeat includes a specialised initiator element whichis required for Tat-responsive transcription. J. Mol.Biol. 248, 562–580.

Selby, M. J., Bain, E. S., Luciw, P. A. & Peterlin, B. M.(1989). Structure, sequence, and position of thestem-loop in tar determine transcriptionalelongation by tat through the HIV-1 long terminalrepeat. Genes Dev. 3, 547–558.

Sheline, C. T., Milocco, L. H. & Jones, K. A. (1991). Twodistinct nuclear transcription factors recognise loopand bulge residues of the HIV-1 TAR RNA hairpin.Genes Dev. 5, 2508–2520.

Spencer, C. A. & Groudine, M. (1990). Transcriptionelongation and eukaryotic gene regulation. Onco-gene, 5, 777–785.

Wigler, M., Sweet, R., Sim. G. K., Wold, B., Pellicer, A.,Lacy, E., Maniatis, T., Silverstein, S. & Axel, R. (1979).Transformation of mammalian cells with genes fromprocaryotes and eucaryotes. Cell, 16, 777-785.

Wu., F., Garcia, J. & Gaynor, R. (1991). Tat regulatesbinding of the human immunodeficiency virustrans-activating region RNA loop-binding proteinTRP-185 Genes Dev. 5, 2128–2140.

Wright, S. (1993). Regulation of eukaryotic geneexpression by transcriptional attenuation. Mol. Biol.Cell, 4, 661–668.

Wright, S. & Bishop, J. M. (1989). DNA sequences thatmediate attenuation of transcription from the mouseprotooncogene myc. Proc. Natl Acad. Sci. USA, 86,505–509.

Wright, S., Mirels, L. F., Calayag, M. C. & Bishop, J. M.(1991). Premature termination of transcription fromthe P1 promoter of the mouse c-myc gene. Proc. NatlAcad. Sci. USA, 88, 11383–11387.

Wright, S., Lu, X. & Peterlin, B. M. (1994). Humanimmunodificiency virus type 1 Tat directs transcrip-tion through attenuation sites within the mousec-myc gene. J. Mol. Biol. 243, 568–573.

Edited by J. Karn

(Received 8 May 1996; received in revised form 2 August 1996; accepted 8 August 1996)