Functional characterization of chimeric reverse transcriptases with polypeptide subunits of highly divergent HIV-1 group M and O strains.

Human immunodeficiency virus (HIV)-1 strains have been divided into three groups: main (M), outlier (O), and non-M non-O (N). Biochemical analyses of HIV-1 reverse transcriptase (RT) have been performed predominantly with enzymes derived from HIV-1 group M:subtype B laboratory strains. This study was designed to optimize the expression and to characterize the enzymatic properties of HIV-1 group O RTs as well as chimeric RTs composed of group M and O p66 and p51 subunits. The DNA-dependent DNA polymerase activity on a short heteropolymeric template-primer was similar with all enzymes, i.e. the HIV-1 group O and M and chimeric RTs. Our data revealed that the 51-kDa subunit in the chimeric heterodimer p66(M:B)/p51(O) confers increased heterodimer stability and partial resistance to non-nucleoside RT inhibitors. Chimeric RTs (p66(M:B)/p51(O) and p66(O)/p51(M:B)) were unable to initiate reverse transcription from tRNA(3)(Lys) using HIV-1 group O or group M:subtype B RNA templates. In contrast, HIV-1 group O and M RTs supported (-)-strand DNA synthesis from tRNA(3)(Lys) hybridized to any of their corresponding HIV-1 RNA templates. HIV-2 RT could not initiate reverse transcription on tRNA(3)(Lys)-primed HIV-1 genomic RNA. These findings suggest that the initiation event is conserved between HIV-1 groups, but not HIV types.

Synthesis of double-stranded DNA from the retroviral RNA genome is an essential step for replication of human immunodeficiency virus (HIV) 1 -1. This step is catalyzed by the viral reverse transcriptase (RT), a heterodimer composed of a 66-kDa (p66) and a 51-kDa (p51) subunit. Subdomains within each subunit are termed "fingers," "palm," "thumb," and "connection" based on x-ray crystal structure and standard subdomain classification in various polymerases (1,2). The heterodimer is generated after the viral protease cleaves one of the RNase H domains of an asymmetric p66/p66 homodimer (3). The DNAbinding cleft and polymerase active-site residues are exposed in the "open" conformation of p66. On the other hand, the p51 subunit is closely folded, with its catalytic residues occupying an internal position in the molecule (1, 4 -6). In addition to providing structural support to the p66 subunit, p51 appears to be required for loading of the p66 subunit onto the templateprimer (7) and for maintaining the proper p66 conformation during initiation of reverse transcription from primer tRNA 3 Lys (8 -11). It has also been reported that the p66/p51 heterodimer as compared with the p66/p66 homodimer displays higher processivity and strand displacement activity during DNAdependent DNA polymerase activity (12,13). Thus, the process of dimerization and subsequent maturation into the p66/p51 heterodimer is essential for a fully functional RT and constitutes a target for therapeutic intervention.
Peptides mimicking the dimer interface of p66 and p51 have been used to prevent heterodimer formation (14,15). In addition, certain non-nucleoside RT inhibitors (NNRTIs), such as [1-[2Ј,5Ј-bis-O-(tert-butyldimethylsilyl)-␤-D-ribofuranosyl]-N 3ethylthymine]-3Ј-spiro-5Љ-(4Љ-amino-1Љ,2Љ-oxathiole-2Љ,2Љdioxide), are known to produce a modest, but significant destabilizing effect on the heterodimer (16). However, the molecular determinants of heterodimer stability are not known. Singleamino acid substitutions in HIV-1 RT, such as W229A, G231A, L234A, and L289K, have been shown to inhibit heterodimer association (17)(18)(19). These substitutions mediate their effects through the 66-kDa subunit, but were not found at the dimer interface of p66 and p51, suggesting an indirect conformational effect on dimer stability. Further confirmation of the role of Leu-234 in dimerization was obtained using the yeast twohybrid system (20). Using this approach, it appears that the fingers and palm subdomains of p66 are dispensable for p51 * This work was supported in part by the United States-Spain Joint Commission for Scientific and Technological Cooperation under Project 99162 and by an institutional grant from the Fundación Ramón Areces (to the Centro de Biología Molecular "Severo Ochoa"). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF068947 and AF378083.
§ To whom correspondence should be addressed. Tel. interaction. Deletion of 25 or 26 amino acids at the C terminus of p51 prevented dimerization, as demonstrated in reconstitution experiments with wild-type p66 and using the two-hybrid system (8,20).
Phylogenetic analysis of HIV-1 sequences obtained throughout the world has identified three genetically distinct groups designated as major (M), outlier (O), and non-M non-O (N). Nucleotide sequence diversity between these groups can be as high as 35% in the envelope-coding region. This is considerably more than the 10 -20% sequence diversity separating HIV-1 group M subtypes (A-J), but comparable to the 40 -50% nucleotide sequence diversity separating HIV-1 and HIV-2. With the exception of some sporadic infections in Western Europe and the United States, the prevalence of the highly divergent, rare HIV-1 group O is greatest in central Africa (i.e. Cameroon, Gabon, and Nigeria) (reviewed in Ref. 21). In a preliminary study, we reported on the expression and characterization of an HIV-1 group O RT (ESP1 isolate) (22 (24). Genomic regions encoding the 66-kDa subunit of the RT were PCR-amplified from patient samples, cut with the NcoI and EcoRI restriction enzymes, and then cloned into the pRT6 or pTrcHisB (Invitrogen) expression vector (22,24). The pRT6 clones were used for large-scale purification of p66, whereas constructs made with pTrcHisB were used for screening the expression levels of RT variants. The RT clones were derived from blood samples obtained from patient ESP1 in January 1996 (clones 1-4 and 21-24), April 1996 (clones 14 -20), and August 1996 (clones 25-28 and 55-66); from patient ESP2 in January 1996 (clones 10 -13 and 36 -54) and May 1996 (clones 5-9); and from patient ESP3 in January 1996 (clones 29 -35) (24). To obtain an expression vector for p51 with the histidine tag for reconstitution and purification, the coding region of the 51-kDa subunit of HIV-1 RT was cloned in the expression vector pTrcHisB. Briefly, primers RTO7 and RTO8 (22) were used to PCR-amplify the p51-coding region from the pRT6-derived plasmid (see above) and to introduce the NheI and EcoRI endonuclease restriction sites for subsequent cloning of the p51 region into the pTrcHisB plasmid.
Bacterial clones (E. coli DH5␣) transformed with the pTrcHisB group O clones (expressing p66) were grown overnight at 37°C in Luria broth medium containing 50 g/ml ampicillin. Approximately 150 l of grown cultures were used to inoculate 3 ml of freshly prepared medium. Isopropyl-␤-D-thiogalactopyranoside (0.8 mM) was added when the culture reached an absorbance of 0.8 -1.0 at 600 nm. Following a 1-h incubation, cells were harvested, and pellets were washed with 400 l of 25 mM Tris-HCl (pH 8.0) containing 10 mM EDTA and 50 mM glucose and then resuspended in 400 l of lysis buffer (10 mM Tris-HCl (pH 8.0) containing 100 mM NaCl, 1 mM EDTA, 5 mM dithiothreitol (DTT), and 5% Triton X-100). Aliquots of this solution were used for SDS-polyacrylamide gel electrophoresis and for the determination of RT activity.
Analysis of RT Activity in Crude Bacterial Lysates-DNA polymerization assays were carried out by mixing 20 l of the obtained lysate and 80 l of 62.5 mM Tris-HCl (pH 8.0) containing 25 mM NaCl, 12.5 mM MgCl 2 , 2 mM DTT, 0.5 g/ml poly(rC)/oligo(dG) [12][13][14][15][16][17][18] , and 10 M [␣-32 P]dGTP. The specific activity of the nucleoside triphosphate used in the assays was 0.3-0.5 Ci/nmol. Samples were incubated for 30 min at 37°C, and reactions were stopped by adding 50 l of 0.5 M EDTA. After acid-insoluble precipitation (25), samples were spotted on a Biodyne A transfer membrane (Millipore Corp.), and the amount of incorporated [ 32 P]dGMP was quantitated by phosphorimaging with a Fuji BAS 1500 scanner. The RT activity of each RT variant-producing clone was determined by substracting the average RT activity of the background clones (bacteria containing the expression vector without the HIV-1 RT-coding region). The polymerase activity of each clone was then compared with that of the previously characterized HIV-1 ESP1 group O RT clone (22). Expression and Purification of HIV-1 RT-Purification of HIV-1 RT variants was carried out after independent expression of each subunit, i.e. p66 and p51 (26). The 51-kDa subunit contained an extension of 14 amino acid residues at its N terminus, which included six consecutive histidine residues to facilitate purification by metal chelate affinity chromatography. The purification procedure starts by harvesting the cells from cultures expressing p66 together with those from cultures expressing p51. This process allows for the reconstitution of chimeric heterodimers, i.e. the 66-kDa subunit of the BH10 clone (a subtype B HIV-1) with the 51-kDa subunit of an HIV-1 group O isolate, or vice versa. Purity of enzymes was assessed by SDS-polyacrylamide gel electrophoresis. RT concentrations were determined using the Bio-Rad protein assay. The specific activity of the purified enzyme was estimated in DNA polymerase assays carried out in 50 l of 50 mM Tris-HCl (pH 8.0) containing 20 mM NaCl, 10 mM MgCl 2 , 8 mM DTT, 5 M dTTP (3-5 Ci/ml [ 3 H]dTTP), and 0.5 M poly(rA)/oligo(dT) [12][13][14][15][16][17][18] (concentration expressed as 3Ј-hydroxyl primer termini) as previously described (22,27). One activity unit is the amount of enzyme that incorporates 1 nmol of [ 3 H]dTMP into acid-insoluble products in 10 min at 37°C. HIV-2 heterodimeric RT was kindly provided by Drs. Roger Goody and Brigitta Wöhrl (Max-Planck-Institut fü r Molekulare Physiologie, Dortmund, Germany).
Kinetic Studies-Single-nucleotide incorporation assays were performed with template-primer D2-47/PG5-25 as previously described (27,28). Briefly, primer PG5-25 (5Ј-CCAGAATGCTGGTAGGGCTATA-CAT-3Ј) was 5Ј end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase (Promega) and then annealed to template D2-47 (5Ј-GGGATTA-AATAAAATAGTAAGAATGTATAGCCCTACCAGCATTCTGG-3Ј) in 150 mM NaCl and 150 mM magnesium acetate at a molar ratio of 1:1. Assays were carried out at 37°C in 10 l of 50 mM Hepes (pH 7.0) containing 15 mM NaCl, 15 mM magnesium aspartate, 130 mM potassium acetate, 1 mM DTT, and 5% polyethylene glycol 6000. The template-primer concentration was 30 nM, and the active enzyme concentration was 3-5 nM. After a 10-min incubation of enzyme with template-primer, reactions were initiated by the addition of dTTP at various concentrations and then stopped after 20 s at 37°C by adding 8 l of 10 mM EDTA in loading buffer containing 90% formamide. The products were analyzed on urea-20% polyacrylamide gels and quantitated by phosphorimaging (25). Elongation measurements were fitted to the Michaelis-Menten equation to determine the corresponding k cat and K m values.
Stability of the RT Heterodimer in the Presence of Urea-Denaturation isotherms were obtained by preincubating RT with different concentrations of urea (up to 4 M) for 10 min at 37°C in 10 l of 50 mM Hepes (pH 7.0) containing 15 mM NaCl, 15 mM magnesium aspartate, 130 mM potassium acetate, 1 mM DTT, and 5% polyethylene glycol 6000. The polymerase reaction was initiated by adding the D2-47/PG5-25 template-primer and dTTP in a 10-l solution of 100 mM Hepes (pH 7.0), 30 mM NaCl, 30 mM magnesium aspartate, 130 mM potassium acetate, 1 mM DTT, and 5% polyethylene glycol 6000. The final concentrations of template-primer and dTTP in these assays were 30 nM and 50 M, respectively. After incubating the samples for 30 s at 37°C, reactions were stopped with EDTA/formamide, and products were analyzed as described above. The obtained amount of primer extension products was plotted versus the urea concentration, and the data were fitted to a logistic curve using the program CurveExpert Version 1.34 (Microsoft Corp.) to determine the concentration of urea at the midpoint of the denaturation isotherm.
Cloning of the HIV-1 Group O Primer-binding Site RNA Expression Vector-Proviral DNA was extracted directly from lysed peripheral blood mononuclear cells (24). A 409-base pair long terminal repeat fragment spanning the repeat, unique 5Ј (U5), and primer-binding site (pbs) regions was PCR-amplified using a set of nested oligonucleotides primers. Briefly, primers s-163 (5Ј-GAAGAGGCAGAAAGACTAGGAG-3Ј) and a-880 (5Ј-CCTAATTTGTTCCCATGCATCC-3Ј) were used in the first external amplification. Products (5 l) of the external amplification were reamplified as previously described (24) using nested primers s-489 (5Ј-GCGCGCAGATCTGGGTCTCGGTTAGAGGACC-3Ј) and a-854 (5Ј-GCGCGCCTCGAGCTTCCTGTCAACACAGACGC-3Ј), which include BglII and XhoI endonuclease restriction sites (underlined), respectively. Nested PCR products were purified (QIAquick PCR purification kit, QIAGEN Inc.), digested with BglII and XhoI, and then cloned into the pSP72 vector (Promega), cut with the same enzymes. The long terminal repeat clone used in the experiments described below was sequenced using the ABI Prism BigDye terminator cycle sequencing ready reaction kit (PerkinElmer Life Sciences).
RNA-dependent DNA Synthesis of (Ϫ)-Strand Strong-stop DNA-RNA-dependent DNA polymerase activity of wild-type and chimeric RTs was measured using either a DNA or a tRNA 3 Lys primer annealed to HIV-1 group O or group M:subtype B RNA templates containing the pbs, U5, and repeat regions (HIV-1 O pbs RNA and HIV-1 M:B pbs RNA, respectively). In addition to these RNA templates, primers were also  annealed to the HIV-1 M:B pbs⌬A RNA template that lacked the A-rich sequence upstream of the primer-binding site. We have previously described the transcription and purification of these HIV-1 RNA templates (10). The 18-nucleotide DNA primer was 5Ј end-labeled using T4 polynucleotide kinase and [␥-32 P]ATP as described (32). Synthetic (or unmodified) tRNA 3 Lys was transcribed from a DNA cassette and internally labeled with T7 polymerase, ribonucleotides, [␣-32 P]UTP, and a complementary 18-nucleotide DNA primer (10). Primers were annealed to RNA templates prior to addition of HIV-1 RT (57 nM), the appropriate Mg 2ϩ -containing buffer, and dNTPs (10,32). Reaction mixtures were incubated at 37°C for 5 and 80 min and then quenched with formamide loading buffer. (Ϫ)-Strand DNA synthesis products were separated on an 8% denaturing polyacrylamide gel, visualized by autoradiography, and quantitated by phosphorimaging.
Computer Analysis of Crystal Structures-Coordinates of crystal structures used in this study were taken from the Brookhaven Protein Data Bank. The viewing program INSIGHT II Version 98.0 (Molecular Simulations Inc., San Diego, CA) was used to analyze the three-dimensional structures. Residues involved in interactions between p66 and p51 were identified using the program WHAT IF (33). For this purpose, the accessibility of all atoms in the p66/p51 heterodimer was determined, as well as in the p66 and p51 subunits considered individually. Residues in p66 (or p51) contributing to the p66/p51 dimer interface would be those that are not accessible in the p66/p51 heterodimer, but are exposed to the solvent when the p66 (or p51) subunit is considered individually. Coordinates from crystal structures were taken from Protein Data Bank files and correspond to unliganded HIV-1 RT (codes 1DLO, 1HMV, and 1RTJ), RT complexed with non-nucleoside RT inhibitors (codes 1BQM, 1BQN, 1HNI, 1HNV, 1KLM, 1REV, 1RT1, 1RT2, 1RT3, 1RT4, 1RT5, 1RT6, 1RT7, 1RTH, 1RTI, 1TVR, 1UWB, 1VRT, 1VRU, and 3HVT), RT complexed with double-stranded DNA (codes 1C9R and 2HMI), and a covalently trapped catalytic complex of HIV-1 RT containing a DNA template-primer and dTTP (code 1RTD).  (24). Five RT clones (designated 42, 44, 46, 49, and 54 and all derived from the January 1996 sample of patient ESP2) consistently displayed the highest level of RT activity in bacterial lysates containing the p66 subunit (Fig. 1). RNA-dependent DNA polymerase activity was determined using poly(rC)/oligo(dG) [12][13][14][15][16][17][18] and [␣-32 P]dGTP as substrates. All five RTs from this ESP2 sample had nearly identical amino acid sequence in the polymerase domain, and we selected clone 46 as the group O RT prototype in the biochemical studies described in this work. The Gen-Bank TM /EBI Data Bank accession number of the ESP2-46 RT sequence is AF068947.

Expression Levels and Purification of Chimeric
Although ESP2 (clone 46) RT shared 98.4% identity with the previously characterized ESP1 RT (22), the levels of RT activity in culture supernatants expressing the 66-kDa subunit of clone ESP2-46 were 10 -20% higher than those of ESP1 (Fig. 1). The variations in amino acid sequence between the two group O RT clones (T196A, L228M, I276V, D348N, L372I, W398R, E415D, M435R, and N470D; amino acids in ESP1 versus ESP2- 46) have not been reported as having major effects on DNA polymerase or RNase H activity. However, the increased RT activity of clone ESP2-46 RT over ESP1 RT in bacterial lysates was not related to RT expression or purification yields. The recovery of pure p66 O /p51 O RT was quite similar for both clones. We did observe a relatively high yield (280 g/liter) of the chimeric RT composed of the p66 subunit of ESP2-46 and the p51 subunit of the M:B laboratory strain BH10 (p66 O /p51 M:B ). In contrast, 4 -6-fold less of the chimeric p66 M:B /p51 O and natural p66 M:B / p51 M:B RTs was obtained from the same bacterial clones. All purified RTs were found to be at least 95% pure and to have the correct subunit composition (Fig. 1). It is important to note that all subsequent experiments were performed with natural or chimeric heterodimers containing subunits from the group O ESP2-46 clone or the group M:subtype B BH10 clone.
Functional Characterization of Chimeric RTs and Sensitivity to Non-nucleoside Inhibitors-All four recombinant heterodimers were active in RNA-and DNA-dependent DNA polymerase assays. The specific activity of each enzyme was determined using poly(rA)/oligo(dT) 12 (Fig. 2). This resistance was observed regardless if poly(rA)/oligo(dT) [12][13][14][15][16][17][18] and dTTP or heteropolymeric templateprimer and dNTPs were employed as substrates. Contribution of the group O p51 subunit to NNRTI resistance in the p66 M:B /p51 O chimera was also observed in single-nucleotide incorporation assays using the 47/25-mer template-primer (Table II). Inhibition constants for nevirapine and the TSAO derivatives TSAO-T and TSAO-m 3 T were 4 -6 times higher for the p66 M:B /p51 O chimera than for p66 M:B /p51 M:B .

Characterization of Chimeric HIV-1 Reverse Transcriptases
p66 M:B /p51 O was impaired in its ability to catalyze (Ϫ)-strand DNA synthesis from tRNA 3 Lys . Moreover, the p66 O /p51 M:B RT was unable to initiate distributive synthesis from tRNA 3 Lys annealed to any of the three templates. Although the natural p66 O /p51 O HIV-1 RT catalyzed less synthesis from the DNA primer than p66 M:B /p51 M:B , the relative extension from a tRNA versus DNA primer was similar for both enzymes with any of the three templates used in these assays (Fig. 4D). The ratio of tRNA-to DNA-primed (Ϫ)-strand DNA synthesis catalyzed by the chimeric p66 M:B /p51 O RT was at least 5-10 times lower than that shown by the natural heterodimers. These findings suggest that subtle structural changes in the chimeric heterodimers versus the natural heterodimers could have an impact on proper tRNA/RNA template binding and initiation of reverse transcription.
In a previous study, we provided evidence that other lentiviral RTs could not initiate reverse transcription from tRNA 3 Lys annealed to a wild-type HIV-1 RNA template (37). Even though the tRNA 3 Lys isoacceptor species acts as the primer for all lentiviruses, the A-rich U5 inverted repeat loop/tRNA anti-codon loop interaction in HIV-1 group M:subtype B blocks initiation by non-HIV-1 RTs (37). HIV-1 groups M and O share only 65-75% sequence identity in the long terminal repeat and the RT-coding region (Fig. 5A). Differences in nucleotide sequence between HIV-2 and HIV-1 groups O and M were observed upand downstream of the pbs (Fig. 5A). Computer predictions of RNA secondary structures imply that HIV-1 group O RNA, but not HIV-2 RNA, has a conserved U5 inverted repeat stem/Arich loop similar to that characterized in HIV-1 M:B RNA (Fig.  5B). Interestingly, both HIV-1 group O and M:B RTs were capable of initiating (Ϫ)-strand DNA synthesis on either group O or M:B RNA templates using an oligonucleotide DNA or a tRNA 3 Lys primer (Fig. 4). HIV-2 RT was an active polymerase from the DNA primer on all three RNA templates. However, this enzyme showed only weak activity using tRNA 3 Lys annealed to the HIV-1 M:B RNA lacking the A-rich sequence and could not utilize the tRNA 3 Lys primer annealed to the wild-type group O or M:B pbs RNA templates (Fig. 6). In contrast to the divergent evolution of HIV-2, it appears that HIV-1 groups O and M share a common mechanism for the initiation of reverse transcription. O and HIV-2 can infect humans and cause AIDS, although the prevalence of those viruses is low and mostly restricted to West Africa. Structural analyses of the NNRTI-binding site of HIV-1 RT and genotypic analyses of nevirapine-resistant M:B isolates suggest that residues such as Cys-181 in the p66 subunit are likely to be responsible for the intrinsic HIV-1 group O resistance to NNRTIs. However, the contribution of the p51 subunit to NNRTI resistance by group O isolates has not been determined.
Expression and purification of chimeric group O/group M:subtype B RT heterodimers (i.e. p66 M:B /p51 O and p66 O / p51 M:B ) were rather efficient, and purification yields were similar or better than for the natural heterodimers (i.e. p66 M:B / p51 M:B and p66 O /p51 O ). As shown in Fig. 7, residues involved in p66/p51 interactions are highly conserved in the RTs of HIV-1 groups O and M, as well as in HIV-2. However, attempts to reconstitute chimeric RTs from separately expressed or coexpressed HIV-1 and HIV-2 subunits resulted in very inefficient and unstable heterodimer formation (40,43). Although structural incompatibility between the different RT subunits could have affected the stability of HIV-1/HIV-2 RT heterodimers, low levels of expression and/or higher sensitivity to degradation by bacterial proteases may have also contributed to the poor recovery of the reconstituted chimeras. In contrast, other groups have obtained active chimeric heterodimers composed of HIV-1 and feline immunodeficiency virus (FIV) RT subunits (i.e. p66 HIV-1 /p51 FIV and p66 FIV /p51 HIV-1 ) (44). This was somewhat surprising considering the relatively high number of amino acid sequence differences between FIV and HIV-1 (Ͻ70% sequence identity) and that the FIV RT lacks the Trprich sequence found in the connection subdomain of HIV-1 RT (see residues 401-410 in Fig. 7). This region is found at the dimer interface in crystal structures of HIV-1 RT (6) and constitutes a potential target for peptide inhibitors of RT dimerization (15). Taken together, these results suggest that formation of chimeric heterodimers is mainly affected by structural compatibility, rather than by sequence identity at the dimer interface.
RT inhibition experiments showed that the chimeric p66 O / p51 M:B RT and the homologous p66 O /p51 O RT were resistant to nevirapine, TSAO-T, and TSAO-m 3 T. Unlike nevirapine, TSAO derivatives have been shown to destabilize the p66/p51 HIV-1 RT heterodimer (16) through binding to regions in the p51 subunit involved in RT dimerization. Chimeric heterodimers consisting of the p66 subunit of HIV-1 subtype B (clone BH10) and the p51 subunit of group O (ESP2 clone 46) were partially resistant to nevirapine and TSAO derivatives. However, increases in inhibition constants were similar when the chimeric p66 M:B /p51 O RT was used in the assay in the place of the homologous p66 M:B /p51 M:B RT. This result implies that differences in amino acid sequence between the ESP2 and BH10 RTs may also encode NNRTI resistance through the ESP2 p51 subunit rather than solely through the ESP2 p66 subunit. The substitution of Lys for Glu-138 confers genotypic resistance to TSAO derivatives in group M:subtype B isolates (45). However, phenotypic resistance is restricted to the catalytically inactive p51 subunit in the heterodimer (46). The RT-coding region of HIV-1 group O does not encode Lys-138 or any other mutations conferring NNRTI resistance through the p51 subunit. Currently, this partial resistance to nevirapine and TSAO derivatives mediated by the group O p51 subunit in the p66 M:B /p51 O heterodimer cannot be attributed to any specific amino acid residue.
Stability of the heterodimeric RTs was determined by comparing the DNA polymerase activities of enzymes pretreated with increasing concentrations of urea. Previous studies revealed that homologous HIV-2 RT heterodimers are more stable than those of HIV-1 RT upon exposure to organic solvents (14,47). However, the contribution of each subunit to stability was not measured. In our study, HIV-1 group O RT heterodimers were significantly more stable than M:B RT heterodimers. Furthermore, our analysis of chimeric RTs predicts that the molecular determinants of greater heterodimer stability lie within the p51 subunit rather than in the catalytically active p66 subunit. The role of p51 in heterodimer stability is supported by (i) studies showing reduced stability of heterodimers containing p51 with C-terminal deletions (8,20) and (ii) structural analyses showing a greater contribution of p51 to the dimer interface via an interaction between the p51 subunit (i.e. fingers, connection, and thumb subdomains) with a more  7. Structural analysis of contact surfaces between the p66 and p51 subunits of HIV-1 subtype B RT. A, the contribution of each residue of p66 or p51 to the contact surface between both subunits of the p66/p51 heterodimer was determined by measuring the loss of accessibility of each subunit upon its interaction with the other, using the program WHAT IF (33). The represented values correspond to the average obtained from 26 crystal structures of HIV-1 RT since unliganded RTs and RTs complexed with inhibitors or double-stranded DNA showed similar patterns of surface accessibility. B, shown is the sequence variation at positions of p66 and p51 with the highest accessibility. The residues indicated are those with an accessibility score higher than 10. Asterisks indicate amino acids with a score above 20. C, presented is a ribbon representation of the HIV-1 RT heterodimer showing amino acids clusters involved in p66/p51 interactions. Residues of p66 (white ribbon) and p51 (yellow ribbon) are shown in cyan and magenta, respectively, using a space-filling representation. The positions indicated correspond to residues with accessibility scores above 10. Coordinates have been obtained from Protein Data Bank code 1HMV (42). limited region in the p66 subunit (i.e. palm subdomain, connection subdomain, and RNase H domain) (Fig. 7). The partial resistance to NNRTIs shown by the chimeric p66 M:B /p51 O RT suggests that intersubunit interactions favor a p51 O -induced conformation over a p66 M:B -induced conformation in this chimeric heterodimer. The p51 subunit of HIV-1 group O RT may also cause a significant perturbation in the NNRTI-binding site of p66 M:B /p51 O RT, probably by altering interactions between the fingers subdomain of p51 and the palm subdomain of p66.
We have previously shown that initiation of reverse transcription in HIV-1 group M:subtype B can be significantly impaired by single-amino acid substitutions (48). Although natural group M:subtype B and group O RTs share only 79% sequence identity (22), both enzymes are able to catalyze (Ϫ)-strand DNA synthesis from a short oligonucleotide DNA and tRNA 3 Lys annealed to either an HIV-1 group O or M:Bderived RNA template. In contrast, other lentiviral RTs, including HIV-2 and simian immunodeficiency virus heterodimeric RTs, cannot efficiently catalyze (Ϫ)-strand DNA synthesis from tRNA 3 Lys annealed to HIV-1 RNA (37). These lentiviral non-HIV-1 RTs can, however, (i) catalyze (Ϫ)-strand DNA synthesis from short oligonucleotide RNA or DNA primers and (ii) initiate reverse transcription from tRNA 3 Lys annealed to FIV or equine infectious anemia virus RNA (37). The A-rich loop in the HIV-1 pbs RNA interacts with the U-rich tRNA 3 Lys anti-codon loop, leading to a bimolecular complex that may block other non-HIV-1 lentiviruses from utilizing primer tRNA 3 Lys annealed to HIV-1 pbs RNA (35,36). A disruption of this tRNA loop⅐HIV-1 RNA loop complex by deleting the A-rich region in the HIV-1 genomic RNA (HIV-1 M:B pbs⌬A RNA) also permits tRNA 3 Lys utilization by other lentiviral RTs. In this study, HIV-2 RT could initiate (Ϫ)-strand DNA synthesis only when tRNA 3 Lys was annealed to an HIV-1 RNA template lacking an A-rich sequence upstream of the pbs. Computer predictions of HIV-1 group M:subtype B, HIV-1 group O, and HIV-2 pbs RNA secondary structures suggest that the A-rich sequence may be accessible to the tRNA anti-codon loop in both HIV-1 group M and O RNAs, but not in HIV-2 RNA. In addition, Arts et al. (48) demonstrated that three basic residues (i.e. Lys-249, Arg-307, and Lys-311) in the p66 thumb subdomain, opposite to the template-primer-binding cleft, are involved in binding to the tRNA anti-codon loop. RT variants with either Gln or Glu at any of these positions have reduced affinity for tRNA 3 Lys , alter interactions with the tRNA anti-codon loop, and support significantly less tRNA-primed (Ϫ)-strand DNA synthesis. These three residues are conserved in all HIV-1 group O and M isolates. Interestingly, HIV-2 RT, which has a serine at position 311, cannot initiate DNA synthesis from a tRNA 3 Lys ⅐HIV-1 pbs RNA complex, i.e. a result analogous to that obtained with the p66 K311E/Q /p51 HIV-1 M:B enzyme (48). These findings unveil the coevolution of RNA and RT elements involved in the initiation of reverse transcription in HIV-1 groups O and M. This is remarkable considering that this conservation of function does not exist between HIV-1 and HIV-2, even though there is a comparable nucleotide sequence diversity between HIV-1 groups (30 -35% nucleotide sequence diversity; O versus M) and HIV types (1 versus 2).
Subtle conformational differences influence inhibitor sensitivity and heterodimer stability in the chimeric HIV-1 group M/group O RTs. However, these differences between the natural and chimeric group O and M:B RTs do not affect DNA-or RNA-dependent DNA polymerase activity using a nonspecific DNA⅐DNA template-primer as substrate. Use of a more complex and endogenous reverse transcription assay has revealed possible structural abnormalities in the chimeric group M/group O RTs. Although the polymerase activities of p66 M:B / p51 M:B , p66 O /p51 O , and p66 M:B /p51 O RTs were similar, the p66 O /p51 M:B RT showed only distributive synthesis using the same 18-nucleotide DNA primer⅐HIV-1 pbs RNA template as substrate. Even the p66 M:B /p51 O RT, which had wild-type activity, from the DNA primer was severely impaired in its ability to use tRNA 3 Lys on the HIV-1 group O and M:B RNA templates. In contrast to the inability of wild-type HIV-2 RT to prime from tRNA 3 Lys on an HIV-1 RNA template (Fig. 3) (37), this defect with the M:B/O chimeras was not intrinsic to the natural HIV-1 group O RT, which could utilize tRNA 3 Lys as a primer on both the HIV-1 group O and M:B RNA templates (Fig. 4). Instead, it appears that the reconstitution of these intergroup chimeras resulted in enzymes that were competent polymerases, but lacked the exact structural architecture to interact with tRNA 3 Lys . Utilization of tRNA 3 Lys on the HIV-1 RNA template is, however, a characteristic of both parental natural group M and O heterodimers. The importance of an exact p51 framework in the heterodimer is highlighted by studies showing that short deletions (13 amino acids) of the p51 C terminus can have drastic effects on tRNA 3 Lys binding and initiation of (Ϫ)-strand DNA synthesis without affecting DNA polymerase activity (8). Interestingly, slightly longer deletions of the p51 C terminus prevent dimerization and recovery of reconstituted heterodimeric RT (8,20).
In summary, we found that heterodimeric RTs composed of highly divergent HIV-1 subtype B and group O RT subunits are functional during elongation of reverse transcription, with the p51 subunit contributing to heterodimer stability and inhibitor sensitivity. The mechanism of initiation of reverse transcription appears to be conserved in HIV-1 group M:subtype B and group O. On the other hand, chimeric RT heterodimers catalyzed little or no (Ϫ)-strand DNA synthesis, suggesting that specific p66/p51 interactions are critical for complex, retrovirus-specific RT functions.