JOURNAL OF VIROLOGY,0022-538X/97/$04.0010

Nov. 1997, p. 8657–8665 Vol. 71, No. 11

Copyright © 1997, American Society for Microbiology

Divergent Transcriptional Regulation among ExpandingHuman Immunodeficiency Virus Type 1 Subtypes

MONTY A. MONTANO,1 VLADIMIR A. NOVITSKY,1 JASON T. BLACKARD,1

NANCY L. CHO,1 DAVID A. KATZENSTEIN,2 AND MAX ESSEX1*

Harvard AIDS Institute, Department of Immunology and Infectious Diseases, Harvard Schoolof Public Health, Boston, Massachusetts 02115,1 and Division of Infectious

Diseases, Department of Medicine, Stanford University Schoolof Medicine, Stanford, California 943052

Received 1 May 1997/Accepted 6 August 1997

The current AIDS pandemic represents the uneven spread of multiple genetically related subtypes (A to J)of human immunodeficiency virus type 1 (HIV-1). Notably, HIV-1 E in southeast Asia and HIV-1 C in sub-Saharan Africa are expanding faster and are likely of greater global significance than the HIV-1 B subtypeprevalent in the United States and Europe. While many studies have focused on genetic variation amongstructural genes, we chose to conduct a comparative analysis of the long terminal repeats of HIV-1 E and HIV-1C isolates and report subtype-specific differences in enhancer copy numbers and sequences, as well as divergentactivation in response to the cellular transcriptional activators Rel-p65 and NFATc and viral Tat. This studyis the first to identify functional distinctions in promoter architecture between HIV-1 subtypes and raises thepossibility that regulatory divergence among the subtypes of HIV-1 has occurred. Divergent transcriptionalregulation may explain some of the epidemiologically observed differences in transmission and pathogenesisand underscores the need for further comparative analysis of HIV-1 regulation.

Although we know that human immunodeficiency virus type2 (HIV-2) is different from HIV-1 in both virulence and trans-mission efficiency (12, 19), the extent to which such differencesoccur between subtypes of HIV-1 remains unclear. Epidemio-logic analysis of the spread of these viruses indicates that whileHIV-2 is largely restricted to west Africa, HIV-1 has a globaland expanding distribution (3). Several subtypes of HIV-1 (Ato J) have been described based on phylogenetic analysis of gagand env gene sequences (16). The HIV-1 subtypes have spreadunevenly, such that HIV-1 B is predominantly associated withthe epidemics linked to injection drug users and homosexualcontact in the West while subtypes such as HIV-1 E and HIV-1C have spread by heterosexual contact in Thailand, India, andsub-Saharan Africa (3, 15, 39). A possible explanation for thismay be that Langerhans’ cells, a likely target in heterosexualtransmission, provide for more efficient replication of HIV-1 Ethan HIV-1 B (36).

Recent studies of HIV-1 B isolates indicate that selectionamong genetic variants of HIV-1 env can occur during sexualtransmission (34, 42) and that such variation may underlie thedifferential utilization of chemokine receptors by T-cell tropic(6) and monocyte/macrophage tropic (1) strains of HIV-1within and, perhaps, between subtypes. However, we reasonedthat successful HIV replication also includes postentry events,such as viral transcription, replication, and viral turnover, suchthat extant genetic sequences in the regulatory region ofHIV-1, the long terminal repeat (LTR), may also impact theability of different HIVs to replicate efficiently through differ-ential recruitment of cellular transcription factors. Differentialreplication efficiency and activation have been partially ad-dressed for HIV-2 compared to HIV-1 B, and have been ex-amined in longitudinal and cross-sectional studies within

HIV-1 B (8, 17, 21, 31, 33, 35, 37), but have not yet beenaddressed among distinct subtypes of HIV-1.

Transcription studies have previously identified the impor-tant immunoregulatory determinants NF-kB p50:p65 andNF-AT and their cognate cis-acting enhancer sequences withinthe HIV-1 LTR (8). Importantly, however, the identification ofthese cis-acting sequences, and that of a variety of others, hasbeen based on the architecture of the first described viralHIV-1 B sequence isolated from T cells (23, 33, 35, 38). Sub-sequent studies have revealed that regulators such as NF-kBp50:p65 and NF-AT represent members of large gene familieswith complex functions in a broad range of cell types (17, 24).Given that HIV-1 has an unusual propensity for genetic diver-sity and complex expression in a variety of cell types, thereemerges a rich potential for alternative regulatory modes toarise, if advantageous (10, 21, 31).

Analysis of the current worldwide distribution of HIV-1subtypes concluded that HIV-1 E and HIV-1 C are the mostprevalent HIVs in the world (3). HIV-1 B although prevalentin the West and present in Africa and Asia accounts for only asmall number of infections in the latter sites. For this reason,we decided to compare potential LTR and/or regulatory dif-ferences between subtypes E and C by characterizing LTRsfrom infected individuals representing distinct areas experienc-ing expanding epidemics, namely, southeast Asia (Thailand,HIV-1 E) and sub-Saharan Africa (Zimbabwe, HIV-1 C).Therefore, LTR-specific primers were designed based on iden-tified consensus regions across A, B, D, and O subtypes in theLos Alamos database (22) (no E or C LTR sequences wereavailable) and were used to generate primary PCR productsfrom proviral DNA that were then sequenced directly orcloned and sequenced by using an ABI automated DNA se-quencer and dye terminator chemistry.

MATERIALS AND METHODS

DNA sequencing and analysis. For PCR analysis, DNA was extracted with aQiagen Blood Kit (Chatsworth, Calif.) from either uncultured or minimally (less

* Corresponding author. Mailing address: Harvard AIDS Institute,Harvard School of Public Health, 651 Huntington Ave., Boston, MA02115.

8657

on May 31, 2015 by guest

http://jvi.asm.org/

Dow

nloaded from

than three passages) cocultured peripheral blood mononuclear cells from do-nors. PCRs were carried out in a total volume of 100 ml, containing 100 ng to 1mg of genomic DNA, by using GeneAmp Kits (Perkin-Elmer Cetus, Foster City,Calif.). Samples were overlaid with 50 ml of mineral oil and then subjected to 30to 39 amplification cycles consisting of a denaturing step (94°C, 30 s), a primer-annealing step (50°C, 30 s), and a primer extension step (72°C, 45 s). Oligonu-cleotide primers were designed based on an analysis of conserved regions withinthe LTR across the subtypes represented in the Los Alamos National LaboratoryHIV Database (A, B, D, and O) (22). The upstream primer for the LTR wasPH59KPN. Downstream primers for the LTR were PH39LTR and PH56. PCRproducts were sequenced directly or were isolated from agarose gels by usingSpin-X columns (Costar, Cambridge, Mass.) and were cloned with the PCR2.1TA cloning system (Invitrogen, San Diego, Calif.). Sequencing was conductedwith fluorescent dye terminators and a 373A Applied Biosystems automatedsequencer. Sequences were aligned by using the neighbor-joining algorithm withClustal W(1.6) (9), gap-stripped with corrections for multiple substitutions, andbootstrap analyzed with 1,000 bootstrap resamplings. Phylograms were gener-ated with NJ-plot (9). Similar analyses and results were obtained using PHYLIPand PAUP (data not shown). A computer algorithm, MatInspector, was used insome cases to identify degenerate enhancer and promoter elements (28). Thesequences for the DNA oligonucleotides were as follows: PH59KPN, 59-CTCAGG TAC CTT TAA GAC C-39; PH39LTR, 59-CTG AGG GAT CTC TAGTTA CCA GAG-39; PH56, 59-ATTGAGGCTTAAGCAGTGGG-39; TFIIDconsensus TATA box (Santa Cruz Biotechnology, Santa Cruz, Calif.), 59-GCAGAGCATATAAAATGAGGTAGGA-39; E-TATA1, 59-CAG ATG CTG CATAAA AGC AGC CGC T-39; E-TATA2, 59-AGC GGC TGC TTT TAT GCAGCA TCT G-39; B-TATA1, 59-CAG ATC CTG CAT ATA AGC AGC CGCT-39; B-TATA2, 59-AGC GGC TGC TTA TAT GCA GGA TCT G-39; E-TATAUp1, 59-CCA GAG TAC TAT AAA GAC TGC TGA C-39; E-TATAUp2, 59-GTC AGC AGT CTT TAT AGT ACT CTG G-39; B-TATAUp1,59-CCG GAG TAC TTC AAG AAC TGC TGA C-39; B-TATAUp2, 59-GTCAGC AGT TCT TGA AGT ACT CCG G-39; KB1, 59-TAC AAG GGA CTTTCC GCT GG-39; KB2, 59-CCA GCG GAA AGT CCC TTG TA-39; E-KB1,59-CTA ACT AGG ACT TCC GCT GG-39; E-KB2, 59-CCA GCG GAA GTCCTA GTT AG-39; C-KB1, 59-CAA CTG GGG CGT TCC AGG AG-39; andC-KB2, 59-CTC CTG GAA CGC CCC AGT GG-39. The sequences for theRNA oligonucleotides were as follows: Tar-B, GAC CAG AUC UGA GCCUGG GAG CUC UCU GGC-39; Tar-E, GAC CAG GUC GAG CCC GGGAGG UCC CUG GC-39; and Tar-del, GAC CAG AUC UGA GCC CGG GCUCUG GC-39.

Electrophoretic mobility shift assays. DNA binding assays using purified pro-tein or nuclear extracts from unstimulated Jurkat T cells (used in TATA boxoligonucleotide competitions) or phorbol-myristate-acetate (PMA)-stimulatedJurkat T cells (used in NF-kB oligonucleotide competitions) were carried out bystandard methods and as previously described (20). Briefly, 5 mg of nuclearextract or 100 ng of purified Tat protein was incubated with radiolabeled oligo-nucleotide (or preincubated for 15 min with cold competitor oligonucleotides)followed by an additional 15-min incubation. DNA-protein or RNA-proteincomplexes were resolved by native polyacrylamide (6%) gel electrophoresis andexposed from 2 h to overnight by using Bio-MAX AR film.

Reporter gene construction and transient-transfection assays. PCR productswere digested with KpnI and HindIII and directly cloned into pbgal-Basic (Clon-tech, Palo Alto, Calif.). The pHIV-lacZ plasmids contain the KpnI-HindIII frag-ment of the HIV-1 LTR (2450 to 160) from representative clones of HIV-1subtypes B, C, and E. Plasmid transfection of target cells was conducted by eithercalcium phosphate transfection (293 cells) or electroporation (Jurkat T cells).Target plasmids were transfected with 100 ng of genomic DNA/105 293 cells or10 mg of genomic DNA/107 Jurkat T cells. Expression vectors Rel-p65, Rel-p52,and NFATc were transfected with 1 mg in 293 cells or 10 mg in Jurkat cells. Tatwas transfected with 100 ng in 293 cells or 10 mg in Jurkat cells. NFATc stimu-lation of Jurkat T cells required PMA (10 ng/ml) and Ionomycin (10 nM). Alltransfections were done in duplicate at least three times, and average values(6 standard errors of the mean) from representative transfections are shown.Cells were assayed for b-galactosidase activity by using the methylumbelliferyl-b-glucuronide assay (20). Fluorescence was determined with a Fluoroskan platereader (Flow Laboratories, Helsinki, Finland).

Heteroduplex mobility assays. PCR products (approximately 550 bp) fromproviral DNA, or relevant clones, were used in a modified heteroduplex assay (5,25) as follows. Briefly, 5 ml of second-round PCR products (;100 to 250 ng ofDNA) of target DNA was mixed with 5 ml of subtype reference (or H2O) and0.5 ml of 103 heteroduplex annealing buffer for a final concentration of 50 mMNaCl-5 mM Tris (pH 7.8)-1 mM EDTA in MicroAmp reaction tubes (Perkin-Elmer). Heteroduplexes were formed by heating the mixture to 94°C for 2 minand cooling to 4°C for 10 min by using an MJR cycling machine. A total of 0.3 mlfrom each mixture was applied directly onto an upturned sample applicatorwhich was placed in the PhastSystem (Pharmacia LKB Biotechnology AB, Upp-sala, Sweden) and electrophoresis was carried out. Separation of homoduplexesand heteroduplexes was carried out by using PhastGel DNA buffer strips with thefollowing conditions: loading, 100 V; 7 mA; and 1 W at 15°C for 5 V z h; andrunning, 300 V; 10 mA; and 2 W at 15°C for a total of 150 to 200 V z h. Stainingwas conducted with the PhastSystem DNA Silver Kit.

Nucleotide sequence accession numbers. Nucleotide sequences were submit-ted to GenBank (accession no. AF023426 to AF023441).

RESULTS

Phylogenetic and heteroduplex analyses indicate subtypedivergence within the LTR. Phylogenetic analysis of proviralDNA obtained from individuals with HIV-1 infections in theUnited States (subtype B), Thailand (subtype E), and Zimba-bwe (subtype C) was conducted (Fig. 1a). This analysis indi-cated that LTR sequences clustered by subtype (note bootstrapvalues) and were congruent with subtype classifications basedon the env or tat genes (data not shown). The LTR has there-fore diverged between these subtypes and may be useful as anadditional marker for subtype assignment and for the detectionof intergenotypic recombinants. Intrasubtype pairwise distancecomparisons were performed and indicated that, consistentwith a recent expansion, subtype E displayed a relatively mod-est distribution (range of 3 to 6%) relative to subtype C iso-lates, which were broadly distributed (range of 5 to 12%). TheB-subtype isolates displayed an intermediate distribution.

To investigate LTR quasispecies complexity, heteroduplexmobility assays were conducted. The primary PCR products ob-tained for direct sequencing (quasispecies) were tested againsta panel of clones subsequently established for each subtype.Quasispecies obtained from individuals infected with the samesubtype of virus displayed heteroduplex bands with faster mo-bility, while concomitantly displaying slower-mobility heterodu-plex band patterns with different subtypes (Fig. 1b). Consistentwith Fig. 1a, heteroduplex analysis confirmed LTR variationbetween subtypes and supports the potential for functionaldivergence.

HIV-1 subtypes E and C contain divergent promoter archi-tectures. Subtype-related variation within and between HIV-1B, E, and C was observed in many regulatory sequences, no-tably NF-kB and NF-AT, but also in USF, TCF-1a, the TATAbox, and the TAR region (Fig. 2a and b). The NF-kB andNF-AT sites have been correlated with activation of HIV-1 bycytokines and mitogens and are likely to play a critical role inthe recruitment of Rel-related transcription factors that aug-ment HIV-1 expression (14, 23, 24, 33). The USF and TATAbox influence basal HIV expression and assembly of the preini-tiation complex, respectively (8). The HIV-1 B regulatory siteswere highly conserved, with the single exception of a dupli-cated region upstream of the NF-kB site in one individual. Thisduplication has been previously observed (41). Direct se-quence products from 44 independent sequence reactions (sev-en individuals) confirmed that the C-subtype isolates consis-tently contained a minimum of three NF-kB enhancer sites(four sites in one individual) (Fig. 3a). All E-subtype LTRs,representing direct sequence products from 38 independentsequence reactions (six individuals), contained one canonicalNF-kB site and one variant NF-kB site. All B-subtype LTRscontained two invariant NF-kB sites in 34 independent se-quence reactions (seven individuals). All HIV-1 E isolates alsocontained consistent changes in the TAR region, including anucleotide deletion creating a 2-nucleotide bulge which mayinfluence Tat activity (4, 7). Also observed within HIV-1 Esubtypes, and potentially relevant to Tat function (2, 26) andNF-kB function (13, 40), were sequence changes in the highlyconserved TATA box as well as the presence of an additionalputative upstream TATA box.

Interestingly, analysis of sequences surrounding the up-stream TATA box (2140) indicated a homology with the ca-nonical TATA (230)-initiator region of both HIV-1 E and B(Fig. 3c), as well as with the flanking regions. However, clari-

8658 MONTANO ET AL. J. VIROL.

on May 31, 2015 by guest

http://jvi.asm.org/

Dow

nloaded from

FIG. 1. Phylogenetic and heteroduplex analyses indicating LTR divergence among HIV-1 subtypes based on LTR DNA sequences (500 nucleotides, entire U3through TAR [2455 to 146, relative to transcription start]). (a) Unrooted tree based on neighbor-joining algorithm with 1,000 bootstrap resamplings. Included are theconsensus sequences from subtype B primary isolates obtained from homosexual men at the Fenway Community Health Center in Boston, Mass. (B-2, B-19, B-3, B-32,B-B017, and B-56) and from subtype E primary isolates obtained from heterosexual men in northern Thailand (Mahidol University) (E-17, E-18, E-10, E-11, and E-15).The U.S. and Thai samples have been described in detail previously (36, 38). Subtype C primary isolates (C-A, C-B, C-D, C-E, C-G, and C-H) were obtained fromheterosexual men in Zimbabwe without a history of travel outside of the region. Also included are the reference Los Alamos database subtype consensus sequencesfor the LTR (B, A, D, and O), two subtype A sequences from Senegal, and molecular clones cm240 (4) and cm2220 (32). (b) Heteroduplex panel. Selected quasispecieswere obtained from individuals infected with either HIV-1 subtype B (isolates B017, 19, and 56), subtype E (isolates 10, 11, and 18), or subtype C (isolates D, G, andA). Subtype viruses were duplexed with heterologous clones from an LTR panel containing B-subtype viral clones (HXB2, B-2.1, and B-3.3), E-subtype viral clones(E-15.2, E-16.2, and E-17.1), and C-subtype viral clones (C-F.1, C-H.1, and C-G.3ii). These clones are identified at the top of the panel as B (1, 2, 3), E (1, 2, 3), andC (1, 2, 3), respectively. Quasispecies displayed relatively homoduplex band patterns (faster mobility) with clones from the same subtype (boxed in figure) whileconcomitantly displaying heteroduplex band patterns (slower mobility) from the alternate subtypes. Homoduplexes with self are indicated with an asterisk.

VOL. 71, 1997 DIVERGENT TRANSCRIPTIONAL REGULATION 8659

on May 31, 2015 by guest

http://jvi.asm.org/

Dow

nloaded from

8660

on May 31, 2015 by guest

http://jvi.asm.org/

Dow

nloaded from

fication of the functional significance of this upstream TATAbox awaits genetic analysis. In addition, analysis of the NF-ATsites revealed that 9 of 11 mutations were shared among sub-type E and C isolates, relative to HIV-1 B (Fig. 3d).

DNA and RNA binding assays reveal a divergent phenotypeassociated with enhancer and promoter elements within sub-

type E and C isolates. To functionally assess the in vitro bind-ing activity of altered NF-kB, TATA, and TAR sequences, gelshift assays were performed with nuclear extracts from JurkatT cells stimulated with PMA, a potent inducer of both Relheterodimer NF-kB p50:p65 nuclear translocation and HIV-1transcriptional activation. Oligonucleotides were synthesized

FIG. 2. (a) Alignment of LTR sequences and comparative LTR regulatory landscape in subtypes B, E, and C. Enhancer and promoter elements reported toinfluence viral gene expression are boxed. Deviations from the B-subtype consensus are indicated. Sequence variations within NF-AT, USF, TCF1a, TATA, and TARas well as copy number changes in the NF-kB enhancer are evident. Also included are the recently reported E clone, cm240 (4), and the recently reported C clone,c2220 (32). (b) Diagram summarizing sequence data shown in panel a. Solid circles indicate conserved mutations.

VOL. 71, 1997 DIVERGENT TRANSCRIPTIONAL REGULATION 8661

on May 31, 2015 by guest

http://jvi.asm.org/

Dow

nloaded from

8662

on May 31, 2015 by guest

http://jvi.asm.org/

Dow

nloaded from

based on the NF-kB consensus within each subtype. Compe-tition assays with oligonucleotides derived from the alteredupstream E-subtype kB site, TAGGACTTCC, indicated a dra-matically reduced capacity to compete for NF-kB p50:p65binding (Fig. 3b, lanes 2 and 3); this result is consistent with theexistence of a single functional kB site within the HIV-1 Esubtype. Alternatively, all isolates of HIV-1 C contained a thirdfunctional kB site, GGGGCGTTCC, in addition to two con-sensus kB sites, which efficiently competed with a consensusoligonucleotide for NF-kB p50:p65 binding (Fig. 3b, lanes 4and 5). By contrast, an additional polymorphic site resemblinga kB motif in one HIV-1 C isolate (C-G), GGGGTTTTAC,failed to compete (Fig. 3b, lanes 6 and 7). Within the corepromoter, nucleotide changes were also observed in the highlyconserved TATA box, which functions to position RNA syn-thesis through the recruitment of the preinitiation factorTATA binding protein (TBP), an integral part of the TFIIDcomplex. The altered TATA motif CATAAAA (T3A) wasobserved in five of the six HIV-1 E isolates. Although unusualfor HIV-1, this TATA polymorphism can be observed in manysimian immunodeficiency virus isolates obtained from natu-rally infected African green monkeys (11) and resembles theloss of function in murine leukemia virus TATA previouslyobserved in mutagenesis studies of a laboratory-adapted mo-lecular clone of HIV-1 B (26). Nevertheless, this alteredTATA sequence competed with an apparently equivalent effi-ciency for TBP-TFIID binding as did the wild-type TATAsequence from HIV-1 B (Fig. 3b, compare lanes 9 and 10 withlane 8). Computer analysis predicted the presence of an up-stream additional TATA box (2140), ACTATAAA, within theHIV-1 LTR of all E-subtype isolates. In DNA binding assayswith extracts from unstimulated Jurkat T cells, oligonucleo-tides derived from the putative additional TATA box in HIV-1E (2140) competed for TBP-TFIID binding over a corre-sponding region in HIV-1 B (Fig. 3b, compare lane 12 with 11),albeit with a reduced efficiency compared to the primaryTATA box (230) of subtypes E and B (Fig. 3b, compare lane12 with lanes 9 and 10). The observed binding efficiencies ofthe respective (230) and (2140) TATA sequences perhapsindicate a role for flanking sequences surrounding the TATAbox in influencing the overall binding activity of TBP-TFIID.In fact, mutagenized TATA and surrounding sequences havebeen correlated with altered HIV-1 activation by viral TAT (2,26).

Binding assays were conducted to test the potential alteredcapacity of Tat protein to recognize HIV-1 E TAR (Tar-E)compared to the reference HIV-1 B TAR (Tar-B). RNA wassynthesized, 118 to 144, corresponding to Tar-B, Tar-E, or acontrol deletion construct (Tar-del). Competition assays indi-cated that both Tar-B (Fig. 3b, lane 14) and Tar-E (lane 15)were efficient at competing for binding to HIV-1 B Tat,whereas the control Tar-del (lane 16) did not effectively com-

pete. Therefore, no obvious defect in HIV-1 B Tat binding wasobserved. Determination of whether homologous Tat fromHIV-1 E displays a similar phenotype or, perhaps, has under-gone compensatory changes altering Tat binding activity tohomologous TAR sequences awaits cloning and functionalcharacterization (research in preparation).

Differential transcriptional activation of HIV-1 E and HIV-1C in response to Rel-p65, NFATc, and viral Tat. We speculatedthat NF-kB, NF-AT, and TAR might have altered functions inHIV-1 E and HIV-1 C and chose to assess potential functionalchanges directly by cotransfection assays. Proviral amplicons ofthe LTR from each subtype were cloned into the KpnI-HindIIIsite of pBgal-Basic, allowing for lacZ reporter gene measure-ments in cotransfection assays with 293 cells or in Jurkat T cellswith expression vectors for the NF-kB binding factors Rel-p65and Rel-p52, as well as NFATc, and HIV-1 B Tat. Transfec-tions with an expression vector for Rel-p65 indicated a corre-lated activity of NF-kB enhancer copy number with Rel-p65transcriptional induction such that the HIV-1 E subtypeweakly transactivated in comparison with HIV-1 C subtypeLTRs, which contained three or four kB sites and were acti-vated more efficiently than either HIV-1 E or HIV-1 B (Fig. 4aand d). Though each subtype consistently differed in responseto Rel-p65, we did not observe a similar trend in response toRel-p52 (Fig. 4a). This may indicate different criteria for p52function or may represent compensation by other low-affinityp52 sites which are not efficiently utilized by p65 (18, 20).

Transfections with an expression vector for NFATc indi-cated a consistently reduced activity (Fig. 4b and d) by both theC- and E-subtype LTRs compared to HIV-1 B. The activationby NFATc in 293 cells did not require exogenous stimuli,whereas NFATc activation in Jurkat T cells required the use ofboth PMA and Ionomycin (17, 33). Since PMA influences bothNFATc and NF-kB activation, a suboptimal concentration ofPMA (10 ng/ml), which had no detectable effect on LTR ac-tivation, was chosen to minimize the contribution of PMA-induced NF-kB activity. At higher concentrations of PMA (100ng/ml) HIV-1 LTR activation was correlated with kB copynumber in Jurkat T cells (data not shown). NF-AT bindingincludes a large family of regulatory proteins that have differ-ent cell type distributions (17). Therefore, differential recog-nition among members of the NF-AT family that partitionHIV-1 E and C from HIV-1 B may have occurred, and thispossibility awaits further study. However, the reduced NFATcactivity may not be due to lack of binding activity, since co-transfections with a Tat expression vector resulted in NFATc-specific transcriptional interference (data not shown). Addi-tionally, since NFATc activity is associated with AP-1 (24), theconfounding role of these factors in NF-AT activity remains tobe determined. Recently, NFATc activation of an HIV-1 Bmolecular clone was reported which indicated a critical role forthe core kB enhancer (14). Since the core enhancer, as well as

FIG. 3. Sequence and biochemical analyses of divergent HIV-1 E and HIV-1 C LTR enhancer and promoter elements. (a) Computer identification of kB sequenceswithin HIV-1 E and HIV-1 C isolates. MatInspector (28) was used with a consensus similarity cutoff score of 0.9 to identify NF-kB enhancer sites. All HIV-1 E isolatescontained a single copy, while all HIV-1 C isolates contained a third site (duplicated in one individual). The sequence position, relative to the start of U3, is shownas are the core and matrix similarity (simil.) scores. (1), analysis of plus strand. (b) Gel shift analysis of variant kB, TATA, and TAR sequences. NF-kB enhancerbinding activity is shown in lanes 1 to 3 and 4 to 7. The upstream kB site in the E subtype, TAGGACTTCC (lanes 2 and 3), failed to compete for binding to aradiolabeled consensus kB site (lane 1). The third kB site in all C subtype isolates, GGGGCGTTCC, (lanes 4 and 5) competed efficiently, and a polymorphic siteresembling a kB motif, GGGGTTTTAC, failed to compete (lanes 6 and 7). NS, non-specific complexes (20). TATA box binding activity is shown in lanes 8 to 12. BothHIV-1 B (lane 9) and HIV-1 E (lane 10) TATA boxes (230) competed for binding to a radiolabeled consensus TATA oligonucleotide (lane 8). The putative upstreamTATA box in subtype E (2140) also competed, with reduced affinity, for binding to the TATA box (lane 12), while the cognate upstream site in subtype B (lane 11)displayed no competitive binding (compare lanes 11 and 12 with lane 8). TAR binding activity is shown in lanes 13 to 16. HIV-1 B Tat binding activity to radiolabeledTar-B (lane 13) was competed for by both Tar-B (lane 14) and the variant Tar-E (lane 15) but not by a Tar-del control (lane 16). Arrows indicate specific complexes.(c) Alignment of HIV-1 TATA sequences. The region surrounding the HIV-1 E putative upstream TATA box (2140) with binding activity for TBP (panel B) is alignedwith the canonical TATA boxes (230) in HIV-1 B and HIV-1 E. (d) Alignment of NF-AT sequences. Deviations from the HIV-1 B consensus in the region overlappingthe upstream NF-AT site (2298 to 2258) in C- and E-subtype isolates are indicated. Shared nucleotide changes (9 of 11) among HIV-1 E and HIV-1 C (in comparisonwith HIV-1 B) are marked with an asterisk.

VOL. 71, 1997 DIVERGENT TRANSCRIPTIONAL REGULATION 8663

on May 31, 2015 by guest

http://jvi.asm.org/

Dow

nloaded from

the NF-AT sites and others, have changed in distinct ways bothin HIV-1 E and HIV-1 C, there are likely to be complexenhancer interactions which underlie HIV-1 transcriptionalactivity.

The HIV-1 E TAR stem loop structure is predicted to con-tain a 2-nucleotide bulge and an energetically weakened hair-pin (4, 7). We observed similar Tat activation of HIV-1 E and

an HIV-1 B control in 293 cells (Fig. 4b), whereas HIV-1 Eactivation was clearly reduced in Jurkat T cells (Fig. 4d) with amolecular expression clone of HIV-1 B Tat. The reduced ac-tivity of HIV-1 E in Jurkat T cells is consistent with previousreports showing that mutagenized TATA motifs influenceTAT function (2, 26). By contrast, Tat activation of HIV-1 Cwas reproducibly lower than that of either HIV-1 B or HIV-1E in 293 cells (Fig. 4b) but was comparable to that of theHIV-1 B control in Jurkat T cells (Fig. 4d). The observedsubtype differences in activation between 293 cells and JurkatT cells may be due in part to the presence of E1a in 293 cells.Nevertheless, these results suggest that Tat regulation is sen-sitive to cell type and promoter context, and further clarifica-tion awaits a comparative analysis using representative clonesof Tat from HIV-1 E and HIV-1 C subtypes (research inpreparation). Cotransfection assays with Rel-p65 augmentedTat activation of HIV-1 B and HIV-1 C LTRs but not that ofHIV-1 E (Fig. 4c), consistent with the reduced p65 activity inHIV-1E (Fig. 4a and d).

DISCUSSIONWe have described distinct regulatory landscapes in subtypes

HIV-1 E and HIV-1 C that differ with subtype HIV-1 B in bothcopy number and exact nucleotide sequence of key enhancerand promoter elements. Notably, functional changes implicat-ing the activities of the NF-kB enhancer, NF-AT, USF, theTATA box, and the TAR region were observed. The observednatural variation in response to Rel-p65, NFATc, and TATsupports accumulating evidence from epidemiological andtransmission studies that a model for HIV-1 gene expressionbased only on the HIV-1 B subtype may not encompass therange of regulatory expression and pathogenic potential ofexpanding subtypes such as HIV-1 E and C.

Recent studies suggest that recombinant viruses may ac-count for up to 10% of all HIV-1 genomes, including HIV-1 Ewhich dominates in Asia (29, 30). While emphasis has beenplaced on identifying and characterizing such recombinantswith respect to structural genes, regulatory elements could playan equally important role in the genetic divergence and patho-genic potential of HIV-1 subtypes. The E genotype has beenrecently described as being an intergenotypic recombinant be-tween A and E, with most of the genome being derived fromsubtype A but with gp120, an external portion of gp41, and asegment of the LTR being derived from subtype E (4, 7, 29).Therefore, an intriguing hypothesis is that coselection for bothstructural and regulatory regions may have occurred.

HIV-1 promoter variation is qualitatively different from se-quence variation in structural genes in that there is no proteinproduct and presumably no direct immune selection pressure.Therefore, dissociating the relative roles of drift and selectionis complicated. Nevertheless, whether the observed LTR vari-ation is due to founder effects within subtypes or regions orstochastic effects on virus replication or is the result of selec-tion remains to be determined by further genetic and func-tional analysis. There is no compelling reason to suspect thatregulatory gene expression will necessarily remain linked tosubtype or be confined to a given geographic region. Selectivepressures related to transmission or pathogenesis may result inthe emergence of recombinant viruses with altered regulatoryregions, in addition to the observed intersubtype recombina-tion in env and gag genes.

Taken together, these results underscore the need to de-velop an appropriate model for the control of HIV-1 geneexpression that is relevant to those subtypes such as HIV-1 Eand HIV-1 C that are spreading most rapidly (3). As signalingcascades are better delineated, particularly in cell types rele-

FIG. 4. Differential transactivation of HIV-1 E and C LTR clones by co-transfection of Rel-p65, Rel-p52, NFATc, and Tat in 293 cells (a, b, and c) andJurkat T cells (d). (a) Transcriptional activity with Rel-p65 and Rel-p52. Up-regulated expression of HIV-1 C-subtype LTRs was observed, but minimalinduction was observed in the HIV-1 E-subtype LTRs relative to an HIV-1 Bcontrol, consistent with binding data shown in Fig. 3b. (b) NFATc activity wasreduced in both HIV-1 E- and HIV-1 C-subtype LTRs relative to the HIV-1B-subtype LTR, consistent with the sequence changes shown in Fig. 3d. Tat(HIV-1 B) activity was essentially at the wild-type level with HIV-1 E LTRtargets; HIV-1 C LTR activation was consistently lower relative to an HIV-1 BLTR reference. (c) Rel-p65 augmented Tat activation of HIV-1 B and HIV-1 CLTRs, but HIV-1 E displayed minimal increased activity, consistent with resultsshown in panel a. (d) In Jurkat T cells, transcriptional activity with p65 wasupregulated in HIV-1 C and reduced in HIV-1 E relative to an HIV-1 B control,consistent with results shown in panel a. NFATc activity in the presence of PMAand Ionomycin (1 P/I) was reduced in both HIV-1 E and HIV-1 C relative to anHIV-1 B control, consistent with results obtained with 293 cells shown in panelb. Tat activation of HIV-1 E was consistently reduced compared to that of HIV-1B and HIV-1 C, in contrast with results obtained with 293 cells (see text). TargetLTRlacZ plasmids E-cl.1, E-cl.2, E-cl.3, C-cl.1, C-cl.2, and C-cl.3 are represen-tative consensus clones from HIV-1 E-subtype isolates E-17, E-15, and E-18 andHIV-1 C-subtype isolates C-G, C-H, and C-E, respectively. The reference cloneused (B-ref) is a consensus clone from isolate B-19 and is similar to the labora-tory-adapted molecular clone ARV-2 lacZ (20).

8664 MONTANO ET AL. J. VIROL.

on May 31, 2015 by guest

http://jvi.asm.org/

Dow

nloaded from

vant to heterosexual transmission (e.g., mucosal transmission)(27, 36), perhaps a more predictive model for HIV gene con-trol in these subtypes will be developed. Since HIV-1 E andHIV-1 C appear to have different LTR configurations and havediverged in distinct ways from the more familiar HIV-1 B-regulated phenotype in T cells, there may be no single modelthat can adequately explain the regulated expression of emerg-ing HIV-1 subtypes.

ACKNOWLEDGMENTS

This study was supported in part by grants CA 39805 and AI 07387from the National Institutes of Health and by training grant 5 D43TW0004 from the Fogarty International Center, National Institutes ofHealth. V. Novitsky was sponsored in part by the Educational Com-mission for Foreign Medical Graduates.

We acknowledge discussions with P. Kanki, T. H. Lee, and G. P.Nolan and thank K. MacArthur for editorial assistance.

REFERENCES

1. Alkhatib, G., C. C. Broder, and E. A. Berger. 1996. Cell type-specific fusioncofactors determine human immunodeficiency virus type 1 tropism for T-celllines versus primary macrophages. J. Virol. 70:5487–5494.

2. Berkhout, B., and K. T. Jeang. 1992. Functional roles for the TATA pro-moter and enhancers in basal and Tat-induced expression of the humanimmunodeficiency virus type 1 long terminal repeat. J. Virol. 66:139–149.

3. Burke, D. S., and F. E. McCutchan. 1996. Global distribution of humanimmuno-deficiency virus type-1 clades, p. 119–126. In V. T. DeVita, Jr., S.Hellman, and S. A. Rosenberg (ed.), AIDS: biology, diagnostics, treatment,and prevention, 4th Ed. Lippencott-Raven Publishers, New York, N.Y.

4. Carr, J. K., M. O. Salminen, C. Koch, D. Gotte, A. W. Artenstein, P. A.Hegerich, D. St. Louis, D. S. Burke, and F. E. McCutchan. 1996. Full-lengthsequence and mosaic structure of a human immunodeficiency virus type 1isolate from Thailand. J. Virol. 70:5935–5943.

5. Delwart, E. L., H. W. Sheppard, B. D. Walker, J. Goudsmit, and J. I.Mullins. 1994. Human immunodeficiency virus type 1 evolution in vivotracked by DNA heteroduplex mobility assays. J. Virol. 68:6672–6683.

6. Feng, Y., C. C. Broder, P. E. Kennedy, and E. A. Berger. 1996. HIV-1 entrycofactor: functional cDNA cloning of a seven transmembrane, G protein-coupled receptor. Science 272:872–877.

7. Gao, F., D. L. Robertson, S. G. Morrison, H. Hui, S. Craig, J. Decker, P. N.Fultz, M. Girard, G. M. Shaw, B. H. Hahn, and P. M. Sharp. 1996. Theheterosexual human immunodeficiency virus type 1 epidemic in Thailand iscaused by an intersubtype (A/E) recombinant of African origin. J. Virol.70:7013–7029.

8. Gaynor, R. 1992. Cellular transcription factors involved in the regulation ofHIV-1 gene expression. AIDS 6:347–363.

9. Higgins, D. G., J. D. Thompson, and T. J. Gibson. 1996. Using CLUSTALfor multiple sequence alignments. Methods Enzymol. 266:383–402.

10. Hirsch, V. M., G. Dapolito, R. Goeken, and B. J. Campbell. 1995. Phylogenyand natural history of the primate lentiviruses, SIV and HIV. Curr. Opin.Genet. Dev. 5:798–806.

11. Jin, M. J., H. Hui, D. L. Robertson, M. C. Muller, F. Barre-Sinoussi, V. M.Hirsch, J. S. Allan, G. M. Shaw, P. M. Sharp, and B. H. Hahn. 1994. Mosaicgenome structure of simian immunodeficiency virus from west African greenmonkeys. EMBO J. 13:2935–2947.

12. Kanki, P. J., K. Travers, S. MBoup, C.-C. Hsieh, R. G. Marlink, A. Gueye-NDiaye, T. Siby, I. Thior, M. Hernandez Avila, J.-L. Sankale, I. NDoye, andM. E. Essex. 1994. Slower heterosexual spread of HIV-2 than HIV-1. Lancet343:943–946.

13. Kerr, L. D., L. J. Ransone, P. Wamsley, M. J. Schmitt, T. G. Boyer, Q. Zhou,A. J. Berk, and I. M. Verma. 1993. Association between proto-oncoproteinRel and TATA-binding protein mediates transcriptional activation by NF-kappa B. Nature 365:412–419.

14. Kinoshita, S., L. Su, M. Amano, L. A. Timmerman, H. Kaneshima and G. P.Nolan. 1997. The T cell activation factor NF-ATc positively regulates HIV-1replication and gene expression in T cells. Immunity 6:235–244.

15. Kunanusont, C., H. M. Fay, J. K. Kreiss, S. Rerks-Ngarm, P. Phanuphak, S.Raktham, C.-P. Pau, and N. L. Young. 1995. HIV-1 subtypes and male-to-female transmission in Thailand. Lancet 345:1078–1083.

16. Louwagie, J., W. Janssens, J. Mascola, L. Heyndrickx, P. Hegerich, G.Groen, F. E. McCutchan, and D. S. Burke. 1995. Genetic diversity of theenvelope glycoprotein from human immunodeficiency virus type 1 isolates ofAfrican origin. J. Virol. 69:263–271.

17. Luo, C., E. Burgeon, and A. Rao. 1996. Mechanisms of transactivation bynuclear factor of activated T cells-1. J. Exp. Med. 184:141–147.

18. Mallardo, M., E. Dragonetti, F. Baldassarre, C. Ambrosino, G. Scala, and I.Quinto. 1996. An NF-kappaB site in the 59-untranslated leader region of the

human immunodeficiency virus type 1 enhances the viral expression in re-sponse to NF-kappaB-activating stimuli. J. Biol. Chem. 271:20820–20827.

19. Marlink, R., P. Kanki, I. Thior, K. Travers, G. Eisen, T. Siby, I. Traore, C.-C.Hsieh, M. C. Dia, E.-H. Gueye, J. Hellinger, A. Gueye-Ndiaye, J.-L. Sankale,I. Ndoye, S. MBoup, and M. Essex. 1994. Reduced rate of disease develop-ment after HIV-2 infection as compared to HIV-1. Science 265:1587–1590.

20. Montano, M. A., K. Kripke, C. D. Norina, P. Achacoso, L. A. Herzenberg,A. L. Roy, and G. P. Nolan. 1996. NF-kB homodimer binding within theHIV-1 initiator region and interactions with TFII-I. Proc. Natl. Acad. Sci.USA 93:12376–12381.

21. Myers, G., and B. Korber. 1994. The future of human immunodeficiencyvirus, p. 211–232. In S. S. Morse (ed.), Evolutionary biology of viruses. RavenPress, New York, N.Y.

22. Myers, G., B. Korber, S. Wain-Hobson, R. Smith, and G. N. Pavlakis. 1996.Human retroviruses and AIDS: a compilation and analysis of nucleic andamino acid sequences. Los Alamos National Laboratory, Los Alamos, N.M.

23. Nabel, G., and D. Baltimore. 1987. An inducible transcription factor acti-vates expression of human immunodeficiency virus in T cells. Nature 326:711–713.

24. Nolan, G. P. 1994. NFAT-AP-1 and Rel-bZIP: hybrid vigor and bindingunder the influence. Cell 77:795–798.

25. Novitsky, V., C. Arnold, and J. P. Clewley. 1996. Heteroduplex mobility assayfor subtyping HIV-1: improved methodology and comparison with phyloge-netic analysis of sequence data. J. Virol. Methods 59:61–72.

26. Olsen, H. S., and C. A. Rosen. 1992. Contribution of the TATA motif toTat-mediated transcriptional activation of human immunodeficiency virusgene expression. J. Virol. 66:5594–5597.

27. Pope, M., M. G. Betjes, N. Romani, H. Hirmand, P. U. Cameron, L. Hoff-man, S. Gezelter, G. Schuler, and R. M. Steinman. 1994. Conjugates ofdendritic cells and memory T lymphocytes from skin facilitate productiveinfection with HIV-1. Cell 78:389–398.

28. Quandt, K., K. Frech, H. Karas, E. Wingender, and T. Werner. 1995. MatIndand MatInspector—new fast and versatile tools for detection of consensusmatches in nucleotide sequence data. Nucleic Acids Res. 23:4878–4884.

29. Robertson, D. L., P. M. Sharp, F. E. McCutchan, and B. H. Hahn. 1995.Recombination in HIV-1. Nature 374:124–126. (Letter.)

30. Robertson, D. L., B. H. Hahn, and P. M. Sharp. 1995. Recombination inAIDS viruses. J. Mol. Evol. 40:249–259.

31. Ross, E. K., A. J. Buckler-White, A. B. Rabson, G. Englund, and M. A.Martin. 1991. Contribution of NF-kB and Sp1 binding motifs to the repli-cative capacity of human immunodeficiency virus type 1: distinct patterns ofviral growth are determined by T-cell types. J. Virol. 65:4350–4358.

32. Salminen, M. O., B. Johansson, A. Sonnerborg, S. Ayehunie, D. Gotte, P.Leinikki, D. S. Burke, and F. E. McCutchan. 1996. Full-length sequence ofan Ethiopian human immunodeficiency type 1 (HIV-1) isolate of geneticsubtype C. AIDS Res. Hum. Retrovirus 12:1329–1339.

33. Shaw, J. P., P. J. Utz, D. B. Durand, J. J. Toole, E. A. Emmel, and G. R.Crabtree. 1988. Identification of a putative regulator of early T cell activationgenes. Science 241:202–205.

34. Shioda, T., J. A. Levy, and C. Cheng-Mayer. 1991. Macrophage and Tcell-line tropisms of HIV-1 are determined by specific regions of the enve-lope gp120 gene. Nature 349:167–169.

35. Sodroski, J., C. Rosen, F. Wong-Staal, S. Z. Salahuddin, M. Popovic, S.Arya, R. C. Gallo, and W. A. Haseltine. 1985. Trans-acting transcriptionalregulation of human T-cell leukemia virus type III long terminal repeat.Science 227:171–173.

36. Soto-Ramirez, L. E., B. Renjifo, M. F. McLane, R. Marlink, C. O’Hara, R.Sutthent, C. Wasi, P. Vithayasai, V. Vithayasai, C. Apichartpiyakul, P.Auewarakul, V. Pena Cruz, D.-S. Chui, R. Osathanondh, K. Mayer, T. H.Lee, and M. Essex. 1996. HIV-1 Langerhans’ cell tropism associated withheterosexual transmission of HIV. Science 271:1291–1293.

37. Starcich, B., L. Ratner, S. F. Josephs, T. Okamoto, R. C. Gallo, and F.Wong-Staal. 1985. Characterization of long terminal repeat sequences ofHTLV-III. Science 227:538–540.

38. Wang, W.-K., M. Essex, M. F. McLane, K. H. Mayer, C. C. Hsieh, H. G.Brumblay, G. Seage, and T. H. Lee. 1996. Pattern of gp120 sequence diver-gence linked to a lack of clinical progression in human immunodeficiencyvirus type 1 infection. Proc. Natl. Acad. Sci. USA 93:6693–6697.

39. Weniger, B. G., Y. Takebe, C.-Y. Ou, and S. Yamazaki. 1994. The molecularepidemiology of HIV in Asia. AIDS 8:13S–28S.

40. Xu, X., C. Prorock, H. Ishikawa, E. Maldonado, Y. Ito, and C. Gelinas. 1993.Functional interaction of the v-Rel and c-Rel oncoproteins with the TATA-binding protein and association with transcription factor IIB. Mol. Cell. Biol.13:6733–6741.

41. Zhang, L., Y. Huang, H. Yuan, B. Chen, J. Ip, and D. Ho. 1997. Identificationof a replication-competent pathogenic human immunodeficiency virus type 1with a duplication in the TCF-1alpha region but lacking NF-kB binding sites.J. Virol. 72:1651–1656.

42. Zhu, T., N. Wang, A. Carr, D. S. Nam, R. Moor-Jankowski, D. A. Cooper,and D. D. Ho. 1996. Genetic characterization of human immunodeficiencyvirus type 1 in blood and genital secretions: evidence for viral compartmen-talization and selection during sexual transmission. J. Virol. 70:3098–3107.

VOL. 71, 1997 DIVERGENT TRANSCRIPTIONAL REGULATION 8665

on May 31, 2015 by guest

http://jvi.asm.org/

Dow

nloaded from