Sequences in the 5 and 3 R Elements of Human Immunodeficiency Virus Type 1 Critical for Efficient Reverse Transcription

JOURNAL OF VIROLOGY, Sept. 2000, p. 8324 8334 Vol. 74, No. 18 0022-538X/00/$04.00 0 Copyright 2000, American Society for Microbiology. All Rights Reserved. Sequences in the 5 and 3 R Elements of Human Immunodeficiency Virus Type 1 Critical for Efficient Reverse Transcription YUKI OHI AND JARED L. CLEVER* Department of Microbiology, University of Texas Health Science Center at San Antonio, San Antonio, Texas 78229-3900 Received 31 March 2000/Accepted 18 June 2000 The genome of human immunodeficiency virus type 1 (HIV-1) contains two direct repeats (R) of 97 nucleotides at each end. These elements are of critical importance during the first-strand transfer of reverse transcription, during which the minus-strand strong-stop DNA ( sssdna) is transferred from the 5 end to the 3 end of the genomic RNA. This transfer is critical for the synthesis of the full-length minus-strand cdna. These repeats also contain a variety of other functional domains involved in many aspects of the viral life cycle. In this study, we have introduced a series of mutations into the 5, the 3, or both R sequences designed to avoid these other functional domains. Using a single-round infectivity assay, we determined the ability of these mutants to undergo the various steps of reverse transcription utilizing a semiquantitative PCR analysis. We find that mutations within the first 10 nucleotides of either the 5 or the 3 R sequence resulted in virions that were markedly defective for reverse transcription in infected cells. These mutations potentially introduce mismatches between the full-length sssdna and 3 acceptor R. Even mutations that would create relatively small mismatches, as little as 3 bp, resulted in inefficient reverse transcription. In contrast, virions containing identically mutated R elements were not defective for reverse transcription or infectivity. Using an endogenous reverse transcription assay with disrupted virus, we show that virions harboring the 5 or the 3 R mutations were not intrinsically defective for DNA synthesis. Similarly sized mismatches slightly further downstream in either the 5, the 3, or both R sequences were not detrimental to continued reverse transcription in infected cells. These data are consistent with the idea that certain mismatches within 10 nucleotides downstream of the U3-R junction in HIV-1 cause defects in the stability of the cdna before or during the first-strand transfer of reverse transcription leading to the rapid disappearance of the sssdna in infected cells. These data also suggest that the great majority of first-strand transfers in HIV-1 occur after the copying of virtually the entire 5 R. Downloaded from http://jvi.asm.org/ Retroviruses harbor two direct repeat sequences (R) at the 5 and 3 ends of their genomic RNA (for a review, see reference 13). These repeats are necessary for directing the minusstrand strong-stop cdna ( sssdna) from the 5 end of the viral genome, close to where reverse transcription initiates, to the 3 R during the synthesis of the full-length minus-strand cdna copy (for a review, see reference 48). This is known as the first-strand transfer of reverse transcription. It has been shown that the complementarity of the sssdna and the 3 R is important for directing the first-strand jump (12, 13, 26, 47, 48). It is not entirely clear if this complementarity is the only factor which directs the first-strand transfer. It has recently been reported that both complementarity-dependent and complementarity-independent mechanisms guide the firststrand transfer during reverse transcription in Moloney murine leukemia virus (MMLV) (49). It is also unclear whether the terminal complementarity between the growing 3 end of the cdna and the acceptor template or complementarity behind this region is more important for the first-strand switch or if both contribute (49). The length of the R sequence varies considerably between different retroviruses (for a review, see reference 13). The R is as short as 16 nucleotides in mouse mammary tumor virus and is up to about 250 nucleotides long in the human T-cell leukemia and bovine leukemia virus retroviruses (13). In the human immunodeficiency virus type 1 (HIV-1), the R elements * Corresponding author. Mailing address: Department of Microbiology, University of Texas Health Science Center, San Antonio, TX 78229-3900. Phone: (210) 567-3935. Fax: (210) 567-6612. E-mail: cleverj @uthscsa.edu. are each 97 nucleotides long. These facts suggest that very short R sequences can efficiently direct the first-strand jump. In fact, it has been reported that R sequences much shorter than wild-type sequences function efficiently during the firststrand transfer of HIV-1 reverse transcription. This is based on observing the replication characteristics of HIV-1 virions harboring deletions in the 3 acceptor R (5). The R sequences of HIV-1 contain many overlapping functional domains. This fact has made it difficult to functionally dissect the importance of R sequences during the first-strand switch using an infectious viral system. In addition to being important during the first-strand transfer, the R sequences of HIV-1 fold into two important RNA stem-loop structures termed the poly(a) and TAR hairpins (Fig. 1A) (4). The poly(a) stem-loop contains the polyadenylation signal which is exclusively utilized in only the 3 R. It has recently been reported that the structure and stability of this hairpin constitute one factor that directs poly(a) site selection to the 3 R (19, 31, 32). In addition, it has been shown that the 5 copy of the poly(a) hairpin is an integral part of the HIV-1 packaging signal (9, 20, 39). Mutations that disrupt the folding of this element are detrimental to proper RNA encapsidation, while compensatory mutations restore packaging to wild-type levels (9, 20). The other RNA element located in the R, the TAR hairpin, performs several critical functions during the viral life cycle (13). The 5 TAR element is the well-established binding site for the viral transcriptional-transactivator protein, Tat. Tat binds to a bulge near the top of this hairpin and recruits several host proteins that in turn lead to the hyperphosphorylation of RNA polymerase II causing it to become more processive and to synthesize full-length primary tran- on January 28, 2019 by guest 8324

VOL. 74, 2000 REVERSE TRANSCRIPTION IN HIV-1 R SEQUENCES 8325 Downloaded from http://jvi.asm.org/ FIG. 1. (A) Diagram of the RNA secondary structures located at the 5 and 3 termini of the HIV-1 genome in the R sequences. The TAR and the poly(a) hairpins (pa) are indicated. The poly(a) tail is shown at the 3 end. (B) Diagram of the various TAR mutants used in this study and comparison to wild-type TAR (WT). Mutated nucleotides are highlighted in black. The ds-1, ms-1, ds-2, ms-2, and ms-3 mutations have been described before (9). Previously, they were introduced into only the 5 TAR element (9). (C) Relative positions of the primer pairs in the HIV-1 genome used in the semiquantitative PCR analysis of newly synthesized viral DNA. Abbreviations: SS, strong stop; SJ, minus-strand jump; FL, full length; PBS, primer binding site; SD, major 5 splice donor. The diagram is not drawn to scale. on January 28, 2019 by guest scripts. The binding sites for these factors are located near the top of the TAR element (for a review, see reference 15). The 5 TAR stem-loop has also been shown to be part of the HIV-1 packaging signal (9, 29, 39). Mutations that disrupt the lower portion of the 5 TAR stem cause severe defects in proper genomic RNA encapsidation, while compensatory mutations restore packaging to wild-type levels (9, 29). These data indicate that the structure of the 5 TAR element is necessary for the fidelity of genomic RNA encapsidation, while the primary sequence is not. Mutations in the 5 TAR hairpin have also been reported to cause defects in the initiation of reverse transcription (9, 28). These defects in initiation were not attributable to defects in RNA packaging. However, it could be argued that these apparent defects in the initiation of reverse transcription resulted from mismatches between the mutant sssdna and the wild-type 3 R elements. These mismatches could result in inefficiencies in the first-strand transfer and thus could indirectly cause the sssdna to be rapidly degraded leading to the conclusion that initiation of reverse transcription was reduced. In order to address some of these questions, we have introduced a series of mutations into the 5, the 3, or both R

8326 OHI AND CLEVER J. VIROL. sequences in an HIV-1 viral clone. We have designed these mutations to avoid the other functional domains, described above, that lie within these regions. The ability of the mutants to undergo the various steps of reverse transcription during one round of viral replication was determined using a semiquantitative PCR assay. We found that certain mutations in either the 5 or the 3 R sequence caused marked defects in viral infectivity that were attributable to inefficient reverse transcription. These mutations would cause mismatches between the full-length sssdna and the 3 acceptor R. In contrast, these viruses were able to undergo reverse transcription in detergent-disrupted virions during endogenous reverse transcription assays as efficiently as wild-type virus. This indicates that these virions were not intrinsically defective for DNA synthesis. Other R mutations, farther downstream, did not affect viral infectivity or interfere with reverse transcription efficiency in infected cells. Our results suggest that, in HIV-1, the terminal complementarity between the sssdna and 3 acceptor R sequences is critically important for the stability of the minus-strand cdna and thus profoundly affects the efficiency of reverse transcription. MATERIALS AND METHODS Cell culture. Human osteosarcoma (HOS), 293T, and COS-7 cells were cultured in Dulbecco s modified Eagle medium containing 4.5 g of glucose/liter, 100 U of penicillin G/ml, 0.1 mg of streptomycin sulfate/ml, and 10% fetal calf serum at 37 C in 5% CO 2. Plasmid construction. All mutations were introduced into the previously described HIV-gpt vector (gift from N. Landau and D. Littman) (35, 40). The amphotropic murine leukemia virus (AMLV) Env expression vector has also been previously described (35, 40). The 5 TAR hairpin mutations ms-4, -5, and -6 were created by oligonucleotide-directed mutagenesis within the BspEI-KasI (nucleotides 309 to 637) fragment of HIV-1 subcloned into pbluescript II KS (pbs/ks ; Stratagene), after which DNAs were sequenced in order to confirm the mutations. This fragment was then subcloned back into the HIV-gpt vector through a multistep subcloning process. The other 5 TAR hairpin mutations have been previously described (9). HIV-gpt constructs containing mutations in only the 3 TAR element were created by swapping the BspEI-HindIII (nucleotides 309 to 531) fragment from the 5 repeat containing the desired mutation with the 3 repeat sequence. HIV-gpt constructs harboring mutations in both the 5 and the 3 repeat elements were constructed by swapping a unique 2.7-kb BamHI-XbaI fragment containing the 3 TAR mutation into the HIV-gpt construct harboring the desired 5 TAR mutation. Constructs for in vitro transcription of antisense riboprobes used in the RNase protection assays were made by subcloning the KpnI-ClaI fragment of wild-type or mutant HIV-gpt into pbs/ KS cut with the same enzymes, as described before (9). Prior to in vitro transcription with T7 RNA polymerase, plasmids were linearized with BspEI. Radiolabeled transcripts were prepared exactly as described previously (8, 10). Virus production and infectivity assays. All virions used in these studies consisted of HIV-1 core particles (strain HXB2) pseudotyped with the AMLV Env protein. Viral stocks were prepared from transient calcium phosphate cotransfection of 293T cells, exactly as before (9 11). Infectivity assays with HOS cells were performed in duplicate by using serial dilutions of the viral supernatants as previously described (9 11). Virus quantitation and exogenous reverse transcriptase assays. The concentration of viral antigen (p24) in the stocks was determined using an enzyme immunoassay as recommended by the manufacturer (Coulter-Immunotech) and as previously described (9 11). Reverse transcriptase assays were performed in duplicate on virions pelleted from 0.5 ml of viral stocks at 25,000 g for 2 h at 4 C, as described before (9 11). RNase protection assays. Viral stocks (12 ml) were layered onto a 5-ml, 20% sucrose cushion (in phosphate-buffered saline [PBS]) and centrifuged at 57,771 g in an S-20/20 rotor (Sorvall) for 1.5 h at 4 C. Viral pellets were resuspended in 0.1 ml of PBS, and an aliquot was removed in order to determine the p24 concentration as described above. Virion and cytoplasmic RNAs were extracted exactly as described before (9 11). Viral and cytoplasmic RNA preparations were treated with 1.0 U of RQ1 RNase-free DNase (Promega) and 10 U of RNase inhibitor in 0.1 ml for 30 min at 37 C followed by treatment with phenol-chloroform and ethanol precipitation to remove any plasmid DNA contamination. Amounts of viral RNAs were quantitated using an RNase protection assay as recommended by the manufacturer (RPA III; Ambion). For virionderived RNAs, the amount of RNA equivalent to 100 ng of pelleted p24 was annealed to an excess of 32 P-labeled riboprobe (10 5 cpm, 200 pg). For cytoplasmic RNAs, approximately 1/20 of the RNA isolated from one T75 flask of 293T cells was used. The protected fragments were electrophoresed on denaturing 5% polyacrylamide 8 M urea sequencing gels and subjected to autoradiography. Radioactivity in the various bands was quantitated using a Molecular Dynamics PhosphorImager. Semiquantitative PCR analysis. Viral supernatants containing either 50 or 500 ng of p24 were brought to a final volume of 4 ml using fresh media. After addition of MgCl 2 (5 mm final concentration) and 100 U of RNase-free DNase I, supernatants were incubated at 24 C for 30 min. After addition of 8 g of Polybrene per ml, the DNase-treated supernatants were split into two samples. The reverse transcriptase inhibitor AZT (zidovudine) was added to one-half of the supernatants, to a final concentration of 10 M. COS-7 cell monolayers grown to about 50% confluence on 10-cm 2 plates were infected with 2 ml of DNase-treated viral supernatants containing either 25 or 250 ng of p24. Those plates of cells infected with virus in the presence of 10 M AZT had been pretreated with the same drug concentration for 3 h prior to infection. After a 90-min infection at 37 C, cell monolayers were extensively washed with PBS. An additional 10 ml of medium was added (with or without AZT [10 m]), and cells were cultured for about 20 more hours. After extensive washing with PBS, cells were briefly trypsinized and pelleted. Total-cell lysates were prepared by a previously published procedure (9, 14). Briefly, cells were disrupted by the addition of lysis buffer (100 mm KCl, 20 mm Tris-HCl [ph 8.4], 0.2% Nonidet P-40, 500 g of proteinase K per ml) and then incubated at 60 C for 2 h, followed by 15 min at 95 C. Serial dilutions of the lysates were then assayed for the presence of the cellular CC chemokine receptor-5 gene (CCR5) to assure that approximately equal amounts of nucleic acids were present in each sample. A previously described hot PCR-based procedure was used (9, 28, 52). Lysates were diluted in 10-fold increments, and 5 l of each was used in the PCRs. The reaction contents were essentially as previously described, except that 50 ng of the unlabeled oligonucleotide (5 -ATGGATTATCAAGTGTCAAGT-3 ; sense) and 25 ng of the 32 P-labeled oligonucleotide (5 -GCAGGAGGCGGGCTGCA ATTT-3 ; antisense), which hybridized to the CCR5 gene, were added to each reaction mixture. Thirty amplification cycles consisting of 93 C for 1 min and 65 C for 2 min were used, and reaction products were separated on 5% polyacrylamide gels in 1 Tris-borate-EDTA buffer. The CCR5 PCR product was 100 bp in length. Gels were visualized by autoradiography and quantitated using a Molecular Dynamics PhosphorImager. Identical reaction conditions were used for hot PCR of viral DNAs. The 10 3 -fold dilutions of the cellular lysates were used in the PCRs for cells infected with 25 ng of p24 (10 4 -fold for cells infected with 250 ng of p24), because it was found that the viral DNA products fell within the linear range of the standard curves. These oligonucleotide pairs, which hybridized to HIV-1 (HXB2), were used to amplify strong-stop (5 -ATCTG AGCCTGGGAGCTCTCT-3 [sense]; 5 -ACTGCTAGAGATTTTCCACACT GA-3 [antisense]), minus-strand jump (5 -CTTTCCGCTGGGGACTTTCCA- 3 [sense]; 5 -GAGAGCTCCCAGGCTCAGATCTGG-3 [antisense]), and fulllength DNAs (5 -TGTGCCCGTCTGTTGTGTGACTCT-3 [sense]; 5 -TCCTG CGTCGAGAGAGCTCCTCTGG-3 [antisense]). Reaction products were visualized and quantitated as described above. The sizes of the PCR products were 162 bp for strong-stop DNA, 141 bp for minus-strand jump DNA, and 138 bp for full-length DNA. In order to amplify strong-stop DNA produced from HIV-gpt harboring the TAR ms-4, -5, and -6 mutations, a different sense primer, 5 -CT GCTTAAGCCTCAATAAAGC-3, that did not overlap with the point mutations was used. This produced a PCR product 123 bp in length. Endogenous reverse transcriptase assays. The endogenous reverse transcriptase reactions were performed essentially by a previously described procedure (38). Viral stocks (12 ml) were brought to 5 mm MgCl 2 and treated with 400 U of RNase-free DNase I for 30 min at 37 C. DNase I-treated supernatants were layered onto 5-ml, 20% sucrose cushions in TEN buffer (100 mm NaCl, 1 mm EDTA, 10 mm Tris-HCl [ph 8.0]) and centrifuged at 57,771 g in an S-20/20 rotor (Sorvall) for 2 h at 4 C. The viral pellet was then resuspended in 0.1 ml of ice-cold TEN buffer, and the amount of viral p24 antigen was determined using an enzyme-linked immunosorbent assay (ELISA) as described above. Reactions were performed in 30 l, and reaction mixtures contained 10 ng of p24 in 0.01% Triton X-100, 50 mm NaCl, 50 mm Tris-HCl (ph 8.0), 10 mm dithiothreitol, 5 mm MgCl 2, and 100 M (each) datp, dgtp, dctp, and dttp. As negative controls for synthesis of the reverse transcription products during the endogenous reaction, parallel reactions were performed without the addition of dttp. After incubation of the reaction mixtures at 37 C for 2 h, 270 l of stop buffer (50 g of proteinase K/ml, 20 g of yeast RNA/ml, 1.5 mm EDTA [ph 8.0]) was added and the incubation continued at 60 C for 1 h. Proteinase K was inactivated by incubation of reaction mixtures at 95 C for 15 min. Hot PCR was performed on 5- l aliquots of the reaction mixtures as described above. RESULTS We have introduced a series of mutations into the R sequences of the HIV-1 viral construct HIV-gpt. This consists of a full-length provirus (strain HXB2) into which a selectable marker gene (gpt) has been inserted in the place of env sequences. We have pseudotyped our HIV-1 core particles with the AMLV envelope protein (35, 40). These virions are capable of undergoing one round of replication after infection of susceptible cells. The extent of DNA synthesis in infected cells

VOL. 74, 2000 REVERSE TRANSCRIPTION IN HIV-1 R SEQUENCES 8327 was determined using a semiquantitative PCR assay that amplifies early, intermediate, or late products of reverse transcription (52). Infectivities of the various mutants were determined by their ability to stably transduce the marker gene-producing colonies during drug selection. The ability of selected mutants to undergo intravirion reverse transcription was determined using an endogenous reverse transcriptase assay in which nucleotides and small amounts of detergent are incubated with sucrose-purified virion particles (38). DNA products were then amplified from disrupted virions using semiquantitative PCR. Certain 5 or 3 R mutations reduce reverse transcription efficiency in infected cells. It has been previously reported that certain mutations in the 5 TAR element result in apparent defects in reverse transcription initiation that were not attributable to defects in RNA packaging (ms-1, ms-2, and ms-3) (9). These mutations could, however, introduce mismatches between the mutant sssdna and wild-type 3 R during the first-strand transfer of reverse transcription. To determine if the putative mismatches were causing these DNA synthesis defects, we introduced these three mutations (Fig. 1B) into the 3 R(3 TAR ms-1, ms-2, and ms-3) or created double mutations in both the 5 and the 3 R sequences of HIV-gpt (5-3 TAR ms-1, ms-2, and ms-3). These viral constructs were cotransfected into 293T cells along with the AMLV expression vectors as previously described (9 11). Supernatants were collected 48 h posttransfection and assayed for the viral capsid protein p24, as well as for exogenous reverse transcriptase activity. All of these mutants produced levels of p24, with associated reverse transcriptase activity, which were similar to that produced by wild-type HIV-gpt (data not shown). Viral supernatants containing equivalent amounts of p24 were treated with DNase I to remove any potentially contaminating plasmid DNA. These supernatants were then used to infect COS-7 cell cultures in either the presence or absence of AZT (10 M). The AZT controls served to show that the DNA products we were observing were synthesized post-cell entry and were not being made inside virion particles prior to infection, which can occur in HIV-1. After 90 min at 37 C, cell monolayers were extensively washed and refed with media with or without AZT and then harvested about 20 h postinfection. Appropriate dilutions of total-cell lysates which contained approximately equal amounts of the cellular gene CCR5 were prepared. These lysates were then used in hot PCRs to amplify the various products of reverse transcription (Fig. 1C). As previously reported, a virus containing the ms-1 mutation in only the 5 copy of TAR accumulated markedly reduced amounts of all the products of reverse transcription in infected cells (Fig. 2A to C) (9). A virus harboring this same mutation in only the 3 copy of TAR (3 TAR ms-1) had a similar phenotype, accumulating reduced amounts of all reverse transcription products in cells (Fig. 2A to C). In sharp contrast, a double mutant containing the ms-1 mutation in both the 5 and 3 R elements (5-3 TAR ms-1) produced wild-type levels of DNA in infected cells (Fig. 2A to C). Phosphorimaging the gels indicated that the single-mutant-infected (5 TAR ms-1 or 3 TAR ms-1) cellular lysates contained at least 10-fold-less DNA products than either the wild-type or the 5-3 TAR ms-1 lysates. Virions harboring either the ms-2 or ms-3 mutation in only the 5 R also accumulated reduced amounts of reverse transcription products, compared to the wild type, at 20 h postinfection, as previously described (Fig. 3A and B) (9). Virions containing these identical mutations in only the 3 R, 3 TAR ms-2 and 3 TAR ms-3, accumulated reduced amounts of reverse transcription products, at about the same level as the 5 TAR mutants (Fig. 3A and B). In contrast, lysates from cells infected with the double mutants 5-3 TAR ms-2 and ms-3 FIG. 2. Semiquantitative PCR analysis of the efficiency of reverse transcription by the 5, the 3, and the 5-3 ms-1 mutant virions. Equal amounts of DNase I-treated viral supernatants (containing equivalent amounts of p24) were used to infect cell monolayers as described in Materials and Methods. One-half of the cells were treated with 10 M reverse transcriptase inhibitor AZT. Total-cell lysates, harvested 20 h postinfection, were assayed for the presence of strongstop (A), minus-strand jump (B), or full-length (C) viral DNAs. PCR standards are shown for reaction mixtures that contained 10, 50, 250, and 2,500 copies of viral DNA in an HIV-gpt vector. To verify that approximately equal amounts of host cell-derived nucleic acids were present in the samples, PCR was performed using a primer pair that amplifies the cellular gene CCR5 (D). PCR standards for CCR5 were generated from reaction mixtures containing 10-fold serial dilutions of cell lysate, from undiluted (0) to a 10 4 dilution. The 10-fold dilution of lysate was used to amplify CCR5 in the various samples. Mock, control plates of cells were incubated with DNase I-treated supernatants from mock-transfected 293T cells and then processed in parallel with the other samples. each harbored essentially wild-type levels of reverse transcription products (Fig. 3A and B). Phosphorimager analysis indicated that the single mutants, in either the 5 or the 3 R, produced at least 10-fold-less reverse transcription products than either the wild-type or double-mutant virions. The infectivities of these mutants were then determined based on their abilities to stably transduce the gpt gene into human HOS cells. Infected-cell cultures were grown for approximately 18 days under selection before visible colonies were quantitated. As previously reported, the 5 TAR ms-1 and ms-2 virions were about 100-fold less infectious than wild-type HIV-gpt, while the 5 TAR ms-3 mutant had about a 15-fold reduction in infectivity (Fig. 4) (9). Similar reductions in infectivities were observed for the 3 TAR ms-1, ms-2, and ms-3 TAR mutants (Fig. 4). In contrast, the infectivities of all three of the 5-3 TAR double mutants were greater than those of the single mutants and were only slightly less than that of wild-type virions (Fig. 4). Therefore, the infectivities correlated well with the results obtained using the semiquantitative PCR analysis, although the semiquantitative PCR appeared to underestimate the extent of the infectivity defect. These results are consistent with the idea that certain mismatches between the sssdna and the 3 R interfere with reverse transcription, most likely during the first-strand transfer, since these defects are not observed with virions containing identically mutated repeats (Fig. 5A and B).

8328 OHI AND CLEVER Some virions containing mismatched 5ⴕ and 3ⴕ R sequences can efficiently undergo reverse transcription. Previous work had indicated that mismatches in the 3 part of the R sequences, in the poly(a) hairpin region, did not significantly reduce viral infectivity (9). Based on this previous work and the above observations we wanted to test the hypothesis that it was the proximity of the mismatches to the U3-R junction which was causing the observed defects in reverse transcription (Fig. 5A and B). We created two double-mutant HIV-1 virions harboring the ms-1 or ms-2 mutations in the 5 copy of TAR while harboring mutations that would disrupt the base pairing in the 3 TAR stem, ds-1 and ds-2 (Fig. 1B). These new TAR double mutants were named 5 ms-1/3 ds-1 and 5 ms-2/3 ds-2 (Fig. 5D). These combinations of mutations would have the effect of moving potential mismatches farther away from the U3-R junction as diagrammed in Fig. 5D. The 5 ms-1/3 ds-1 construct would produce a 5-bp mismatch between the sssdna and 3 R sequence that is 51 bp 3 of the U3-R junction. The 5 ms-2/3 ds-2 viruses would produce a 9-bp mismatch that is 47 bp from this junction. We also introduced these two disruptive mutations into only the 3 TAR element of HIV-gpt, creating constructs named 3 TAR ds-1 and ds-2. The 3 TAR ds-1 mutant would produce a 5-bp mismatch, while the 3 TAR ds-2 mutant would form a 9-bp mismatch, that would be within 4 bp of the U3-R junction for each mutant (Fig. 5C). These viral constructs were cotransfected into 293T cells along with the AMLV env expression vectors to produce infectious stocks containing pseudotyped virions. The concentration of the viral capsid antigen p24 was determined for each stock, as was the pelletable reverse transcriptase activity associated with these virions. All of these mutants produced levels of p24, with associated reverse transcriptase activities, which were similar to those of wild-type HIV-gpt (data not shown). Because it was previously shown that the two disruptive mutations near the bottom of the 5 TAR stem resulted in virions that were defective for proper genomic RNA encapsidation, we needed to determine their effects on RNA packaging when present in only the 3 TAR element (9). Wild-type HIV-gpt, 3 TAR ds-1, and 3 TAR ds-2 virions were partially purified by being pelleted through 20% sucrose cushions. RNA extracted from known quantities of virions was subjected to the previously described quantitative RNase protection assay (9 11). This analysis, which detects both genomic and subgenomic HIV-1 RNAs, revealed that 3 TAR ds-1 and 3 TAR ds-2 virions packaged wild-type amounts of genomic RNA and properly excluded spliced RNAs, similarly to wild-type HIVgpt (Fig. 6, upper). Therefore disrupting the base pairing near the bottom of the 3 TAR hairpin does not affect HIV-1 RNA encapsidation, in contrast to what was found for the 5 TAR element (9). Based on these results, we would not expect 5 ms-1/3 ds-1 or 5 ms-2/3 ds-2 to have defects in RNA packaging either. We then determined the abilities of these mutants to undergo efficient reverse transcription in infected-cell cultures using the semiquantitative PCR assay. DNase I-treated stocks were used to infect COS cell cultures as described above. After 90 min at 37 C, cell monolayers were extensively washed and refed with media with or without AZT and then harvested about 20 h postinfection. Appropriate dilutions of total-cell lysates, which contained approximately equal amounts of the cellular gene CCR5, were prepared (Fig. 6C). These lysates were then used in hot PCRs to amplify the various products of reverse transcription. Lysates from both the 3 TAR ds-1- and 3 TAR ds-2-infected cells contained significantly less DNA products than lysates from wild-type HIV-gpt-infected cells (Fig. 6A and B). Phosphorimaging the gels indicated that the 3 TAR ds-1 and ds-2 lysates contained at least 5- to 10-foldless DNA products than the wild-type lysates. Double mutants 5 ms-1/3 ds-1 and 5 ms-2/3 ds-2 produced wild-type amounts of DNA products in infected cells at about 20 h postinfection, indicating that there was no obvious defect in reverse transcription in either mutant (Fig. 6A and B). The infectivities of the 3 TAR ds-1 and 5 TAR ms-1/3 ds-1 FIG. 4. Infectivities of the ms-1, ms-2, and ms-3 series mutants compared to the wild-type HIV-gpt. Infectivities of the constructs are expressed as the gpt CFU per microgram of the viral capsid protein p24 on HOS cells. The standard deviations from duplicate infection assays are indicated. FIG. 3. Semiquantitative PCR analysis of the efficiency of reverse transcription by the 5, the 3, and the 5-3 ms-2 and ms-3 mutant virions. Infections were performed as described for Fig. 2. Total-cell lysates, harvested 20 h postinfection, were assayed for the presence of strong-stop (A) or full-length (B) viral DNAs. To verify that approximately equal amounts of host-cell derived nucleic acids were present in the samples, PCR was performed using a primer pair that amplifies the cellular gene CCR5 (C), as for Fig. 2. J. VIROL.

VOL. 74, 2000 REVERSE TRANSCRIPTION IN HIV-1 R SEQUENCES 8329 FIG. 5. Diagram of the first-strand transfer of reverse transcription with some of the R mutants used in this study. Potential mismatches between the sssdna and 3 acceptor R sequences are indicated as bulges. The 5 TAR ms-1, ms-2, and ms-3 virions would have two mismatched regions (A), as would the 3 TAR ms-1, ms-2, and ms-3 mutants (B). The 3 TAR ds-1 and ds-2 virions would have one mismatched region within five nucleotides of the U3-R junction (C). The double mutants 5 ms-1/3 ds-1 and 5 ms-2/3 ds-2 would also contain only one mismatched region that would be 51 and 47 bp from the U3-R junction, respectively (D). Thick lines, RNA; thin lines, newly synthesized DNA. The relative positions of the U3, R, U5, and primer binding site (PBS) elements are indicated. The primer trna lys is shown as a curved line (trna), and point mutations are denoted with an X. The diagram is not drawn to scale. virions were determined using the colony formation assay on HOS cell cultures. The 3 TAR ds-1 virions had about a 10- fold reduction in infectivity compared to the parental HIV-gpt virions (Fig. 7). In contrast, the double mutant 5 ms-1/3 ds-1 had an essentially wild-type infectivity (Fig. 7). Therefore, these data are consistent with the idea that it is the proximity of the mismatches to the U3-R junction, and not the destabilization of the overall base pairing between the sssdna and 3 acceptor, that causes the defects in reverse transcription. This is based on the fact that the 3 TAR ds-1 and ds-2 mutants have the same number of base pair mismatches as the double 5 ms-1/3 ds-1 and 5 ms-2/3 ds-2 mutants; however, only the double R mutants are able to efficiently undergo reverse transcription (Fig. 5C and D). We can also conclude that certain mutations farther than 47 bp from the U3-R junction do not cause defects in reverse transcription efficiency or in stable provirus formation. Mutations farther than 10 nucleotides downstream from the U3-R junction do not cause defective reverse transcription in infected cells and define a region of R that is sensitive to mismatches. In order to define the region of R in which mismatches cause defects in reverse transcription, we introduced a series of additional mutations into the 5, the 3, or both R elements. We called these new mutations, which introduce altered stem sequences into the 5 and 3 TAR elements, ms-4, ms-5, and ms-6 (Fig. 1B). We introduced this type of mutation, which maintains 5 TAR stem base pairing, so as not to cause defects in genomic RNA encapsidation (9, 29). Those constructs in which the mutations were introduced into only the 5 R were called 5 TAR ms-4, ms-5, and ms-6. Those constructs harboring the mutations in only the 3 R were named 3 TAR ms-4, ms-5, and ms-6, while the double mutants containing the altered sequences in both repeats were called 5-3 TAR ms-4, ms-5, and ms-6. These viral constructs were transfected into 293T cells along with the AMLV envelope expression vector, as described above. After about 48 h, supernatants were harvested and filtered through 0.45- mpore-size filters and assayed for p24 concentrations by enzymelinked immunosorbent assay as described above. Infectious supernatants containing equal amounts of p24 were treated with DNase I and then used to infect COS cell cultures, in either the presence or absence of AZT (10 M), as before. Total-cell lysates were prepared about 20 h postinfection, and appropriate dilutions were assayed for the presence of approximately equal amounts of host cell CCR5 by hot semiquantitative PCR (Fig. 8B [lower two panels] and C). Lysates were then used in the semiquantitative PCR assay with primers specific for either strong-stop or nearly full-length HIV-1 DNA (Fig. 8). We initially characterized the 5 TAR ms-4, ms-5, and ms-6 mutants with primers specific for strong-stop and fulllength DNAs (Fig. 8A and B, upper panel). None of these single 5 R mutants appeared to have reduced amounts of reverse transcription products in host cell lysates compared to wild-type HIV-gpt (Fig. 8, upper panel). We next tested all

8330 OHI AND CLEVER J. VIROL. FIG. 6. (Upper) Quantitative RNase protection assay. Cytoplasmic (cyto.) or virion-derived RNAs were annealed to an excess of radiolabeled riboprobe and treated with single-strand-specific RNases; protected fragments were then separated on denaturing polyacrylamide gels. The positions and sizes (in nucleotides) of the genomic and spliced fragments are indicated to the left, while the positions of molecular weight markers (nucleotides) are shown on the right. Phosphorimaging the genomic fragments from virions revealed that 3 TAR ds-1 contained 160%, and 3 TAR ds-2 contained 115%, of the levels of genomic RNA in wild-type HIV-gpt virions. (Lower) Semiquantitative PCR analysis of the efficiency of reverse transcription by the 3 TAR ds-1 and ds-2 single-mutant virions and the 5 ms-1/3 ds-1 and 5 ms-2/3 ds-2 double-mutant virions. Infections were performed as for Fig. 2. Total-cell lysates, harvested 20 h postinfection, were assayed for the presence of strong-stop (A) or full-length (B) viral DNAs. To verify that approximately equal amounts of host cell-derived nucleic acids were present in the samples, PCR was performed using a primer pair that amplifies the cellular gene CCR5 (C), as for Fig. 2. grown for approximately 18 days before being fixed, stained, and counted. All nine mutants had approximately wild-type infectivities, confirming that these mutants were not defective for reverse transcription (Fig. 9). Therefore, certain mutations greater than 10 bp downstream from the U3-R junction do not interfere with reverse transcription or viral infectivity. The inability of mutants to efficiently undergo reverse transcription in infected cells is not an intrinsic property of the virion particles. Our results suggest that mismatches between the sssdna and 3 R sequences near the U3-R boundary cause defects in reverse transcription during infection. While these mutations likely cause defects before or at the time of the first-strand transfer, it is possible that they somehow interfere with the initiation of reverse transcription in infected cells. We have not detected wild-type levels of strong-stop DNA in lysates from cells infected with the 5 or the 3 ms-1 to ms-3 single R mutants even at earlier times postinfection (data not shown). Therefore, if the sssdnas are being synthesized at normal levels by these single mutants, they are being rapidly degraded. This putative degradation of mismatched sssdna to the 3 repeat RNA might be occurring through the actions of cellular nucleases. If this were true, these defects in reverse transcription would not be expected to occur during the endogenous reverse transcriptase reaction, since no cellular nucleases would be present. The ability of these mutants to undergo reverse transcription was determined using a previously described semiquantitative endogenous reverse transcriptase assay that utilizes virions which have been disrupted with small amounts of detergent (38). Viral stocks containing wild-type HIV-gpt or the 5, the 3, or the 5-3 ms-1, ms-2, and ms-3 virions were treated with DNase I to remove extravirion DNA. Viral particles in these supernatants were then purified by centrifugation through 20% sucrose. After p24 capsid concentrations were determined, equal amounts of virions were used in the endogenous reaction without the addition of exogenous primers or templates. As a control for the synthesis of viral DNA during the endogenous reaction, identical incubations were performed without the addition of deoxynucleotide dttp. DNA products in the reactions were then quantitated by using the above-described semiquantitative PCR analysis, with a primer pair that could amplify strong-stop DNAs (Fig. 10, S.S.). In contrast to what was found for infections, all nine of these mutants produced approximately wild-type levels of strong-stop DNA in the endogenous reaction (Fig. 10, S.S.). Therefore, the ability of these R mutants to initiate reverse transcription is not inher- nine of the 5, 3, and double R mutants (Fig. 8, lower two panels). None of the six single (5 TAR ms-4 to ms-6 and 3 TAR ms-4 to ms-6) or 3 double (5-3 TAR ms-4 to ms-6) R mutants appeared to have consistently reduced amounts of full-length DNA in infected-cell lysates compared to wild-type HIV-gpt (Fig. 8A, lower two panels). To confirm these observations and to determine if this proviral DNA could be stably maintained in infected cells, the infectivities of these mutants were determined using the colony formation assay. Equal amounts of p24 were used to infect HOS cell cultures. After 24 h, infected cells were placed into drug selection media and FIG. 7. Infectivities of the 3 TAR ds-1 and 5 TAR ms-1/3 ds-1 mutants compared to that of the wild-type HIV-gpt. Infectivities of the constructs are expressed as the gpt CFU per microgram of the viral capsid protein p24 on HOS cells. The standard deviations from duplicate infection assays are indicated.

VOL. 74, 2000 REVERSE TRANSCRIPTION IN HIV-1 R SEQUENCES 8331 with the idea that the strong-stop DNAs are being degraded in host cells as a result of certain mismatches between the sssdna and 3 R sequences that are located near the U3-R boundary. Alternatively, the virion-disrupted endogenous re- FIG. 8. (Upper panel) Semiquantitative PCR of the efficiency of reverse transcription by the 5 TAR ms-4, ms-5, and ms-6 mutant virions. (Lower two panels) Semiquantitative PCR analysis of the efficiency of reverse transcription by the 5, the 3, and the 5-3 ms-4, ms-5, and ms-6 mutant virions. Infections were performed as for Fig. 2. Total-cell lysates, harvested 20 h postinfection, were assayed for the presence of strong-stop (upper panel, A) and full-length (upper panel, B; lower two panels, A) viral DNAs. To verify that approximately equal amounts of host-cell derived nucleic acids were present in the samples, PCR was performed using a primer pair that amplifies the cellular gene CCR5 (upper panel, C; lower two panels, B), as for Fig. 2. ently reduced in disrupted virions. This also suggests that the amount of primer trna incorporated into these virions is not significantly altered and is not the cause of their reduced ability to synthesize DNA in infected cells. The ability of the ms-3 series virions to synthesize reverse transcription products after the second-strand transfer was also determined by using a primer pair that amplifies nearly full-length DNAs. This analysis revealed that ms-3 mutant virions were as efficient as wild-type HIV-gpt in synthesizing late DNA products (Fig. 10, F.L.). Therefore, these 5 or 3 R mutant virions are not inherently defective for DNA synthesis, but the phenotype appears during infection of host cells. These results are consistent FIG. 10. Semiquantitative PCR analysis of endogenous reverse transcriptase reactions using purified 5, 3, or 5-3 ms-1, ms-2, and ms-3 virions. Viral stocks (12 ml) were treated with DNase I before pseudotyped virions were pelleted through 20% sucrose. Pellets were resuspended in 100 l of TEN buffer (Materials and Methods). Aliquots containing 10 ng of p24 were incubated in endogenous reaction mixtures containing 100 M concentrations of each deoxynucleotide with or without dttp. As an additional control, reactions were performed on pelleted mock-transfected supernatants that were treated exactly as the virion-containing samples (Mock). After the endogenous reaction, virions were digested with proteinase K. To detect newly synthesized viral DNAs, aliquots were amplified using the hot semiquantitative PCR assay, as described in Materials and Methods. Samples were assayed for the presence of strong-stop DNA products (S.S.). The ms-3 series mutants were also assayed for nearly full-length DNAs (F.L.). PCR standards are shown for reaction mixtures that contained 10, 50, 250, 2,500, and 5,000 copies of viral DNA in an HIV-gpt vector. FIG. 9. Infectivities of the ms-4, ms-5, and ms-6 series mutants compared to that of the wild-type HIV-gpt. Infectivities of the constructs are expressed as the gpt CFU per microgram of the viral capsid protein p24 on HOS cells. The standard deviations from duplicate infection assays are indicated.

8332 OHI AND CLEVER J. VIROL. verse transcription reaction is not reflecting what is happening in infected cells. DISCUSSION We have utilized a single-round replication system to study the effects of mutations in the HIV-1 terminal repeats on reverse transcription efficiency. We provide genetic evidence indicating that mutations which introduce mismatches of 3 bp or more within 10 nucleotides downstream of the HIV-1 U3-R border, between the sssdna and the 3 repeat RNA sequences, result in viruses that are unable to efficiently undergo reverse transcription in infected cells. We detected greatly reduced levels of even the initial products of reverse transcription, the sssdnas, in lysates from cells infected with these mutants. However, these same mutations do not cause defects in initial DNA synthesis in disrupted virions during an endogenous reverse transcription reaction. These data indicate that reverse transcription is not intrinsically defective in these mutants, at least in the presence of detergent. They are consistent with the idea that sssdnas which have misalignments near the polymerization site at the 3 R, near the U3-R junction, are rapidly degraded in infected cells. These data further indicate that, during an infection, most HIV-1 virions synthesize fulllength or near-full-length sssdnas that have copied the entire 5 R and that are then transferred to the 3 R acceptor leading to full-length minus-strand cdna, in agreement with previous reports (33). Otherwise, we would not have observed defective reverse transcription with most of the mutants used in this study which introduce near-terminal mismatches between full-length sssdna and the 3 R. Premature jumping would not lead to mismatches with most of the mutations used here. These results also have implications for the positions of important contacts between the reverse transcriptase enzyme and primer template strands during the first-strand transfer reaction. In vitro studies, with synthetic donor and acceptor molecules and purified reverse transcriptase, have been used to examine aspects of strand switching in many retroviral systems (1 3, 6, 16, 22 25, 27, 42, 51). For HIV-1, it has been shown that the first-strand transfer is greatly facilitated by the nucleocapsid (NC) protein, which is known to have nucleic acid chaperonelike activities (7, 16, 21, 27, 30, 36, 41, 51). A study in which mismatches were introduced between the donor DNA and acceptor RNA molecules revealed that misalignments at the 3 polymerization site of the donor were especially detrimental to continued reverse transcription. HIV-1 reverse transcriptase was able to extend primers containing 5-bp terminal mismatches at about 3 to 5% of the efficiency of wild-type controls, depending on the concentration of NC protein (36). Others have also observed that reverse transcriptase is able to extend mismatched primer termini in vitro (36, 43, 44, 53). Studies have examined the effects of mutations in R sequences on replication and provirus formation using several retroviral systems. A study using replication-competent MMLV showed that the 5 repeat element was not always completely copied before it was translocated to the 3 acceptor (37). This led to the eventual loss of insertion and deletion mutations introduced into either the 5 or the 3 R and caused the eventual outgrowth of wild-type virus. Virions harboring these same mutations in both R elements stably maintained them (37). A spleen necrosis virus (SNV) single-round replication system was used to determine the frequency of premature first-strand transfers (45). By placing a genetic marker into the middle of the 5 or 3 SNV R, it was determined that about 10% of progeny virions resulted from premature strand transfers which occurred before the entire 5 R was copied (45). Utilizing point mutations near the U3-R border, another group determined the sites of the first-strand transfers in an MMLVbased system and showed that premature transfers occurred in 1 to 2% of DNA synthesized during reverse transcription (34). Studies with HIV-1 have also revealed that the first-strand transfer occurs prematurely in a small number of cases (33). Our data are in general agreement with these studies, because the infectivities of our mutants, which introduced near-terminal mismatches between the sssdna and 3 R, were reduced by between 90 and 99% compared to that of the wild type. This suggests that in 1 to 10% of viruses, a premature first-strand transfer occurred before the mutations were copied, thus introducing no mismatches. Alternatively, our results may indicate that in 1 to 10% of viruses, reverse transcription was able to bypass the nearly terminal mismatches in some manner. A study using an infectious clone of HIV-1 in which the 3 R was progressively truncated from its 3 end showed that much shorter acceptors, as short as 30 nucleotides, could work efficiently during the first-strand transfer (5). These mutations did not generally introduce mismatches near the U3-R border. More recently a series of viruses with deletion and substitution mutations spanning the 3 U3-R region were used to examine the first-strand transfer in MMLV by means of a single-round replication system (49). Small and large deletions in the 3 R sequences slightly downstream of the U3-R junction reduced viral titers between one- and fivefold. A virus harboring a mutation altering the first five bases of the 3 R had a fivefold titer reduction compared to the wild type. Fairly large deletion mutations that spanned the U3-R junction, in contrast, reduced titers between 25- and 200-fold. Based on these and other data, the authors concluded that complementarity-independent mechanisms at the U3-R junction were sufficient to direct the first-strand transfer in MMLV (49). Taking these results together with our results, it seems that the first-strand transfer in HIV-1 has significant differences with that in MMLV. Reverse transcription in HIV-1 appears to be more sensitive to small mutations near the U3-R border, significantly reducing viral infectivities. A 3-bp mutation reduced titers by 10-fold, while a 4-bp mutation reduced viral infectivity by 100- fold in our HIV-1 system. Therefore, small mismatches between the sssdna and the 3 acceptor are quite deleterious to efficient reverse transcription in HIV-1. Although our study did not exhaustively address the contribution(s) of the TAR hairpin structures to the first-strand transfer of reverse transcription, we can make several conclusions. Two mutants used in this study introduced disruptions into the lower stem of the 3 TAR hairpin, 5 ms-1/3 ds-1 and 5 ms-2/3 ds-2. Both of these double mutants appeared to undergo efficient reverse transcription (Fig. 6), and the 5 TAR ms-1/3 ds-1 mutant had wild-type infectivity (Fig. 7). We conclude that these two disruptions, of four and eight stem base pairs, did not significantly reduce reverse transcription efficiency. Therefore, our results agree with a recent study that found no evidence that TAR structure had a significant effect on the mechanism of reverse transcription (18). These data agree well with previous studies that have examined the effects of terminal or near-terminal misalignments between the primer trna lys and the primer binding site of HIV-1 (17, 46, 50). Previous studies have shown that the first six nucleotides of the primer binding site are sufficient for primer trna binding and initiation of reverse transcription in HIV-1 mutants (50). When two- to four-nucleotide insertion and deletion mutations were introduced into the HIV-1 primer binding site, those that were closest to the 3 trna polymerization site had the most severe defects in viral replication (17).