The K65R Mutant Reverse Transcriptase of HIV-1 Cross-resistant to 2’,3’-Dideoxycytidine, 2’,3’-Dideoxy=3’-thiacytidine, and 2’,3’-Dideoxyinosine Shows Reduced Sensitivity to Specific Dideoxynucleoside Triphosphate Inhibitors in Vitro*

The K65R mutation in human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) encodes cross-resistance to 2’,3’-dideoxycytidine (ddC), 2’,3‘-dideoxy- 3’4hiacytidine (3TC), and 2’,3’-dideoxyinosine (ddI). We characterized the in vitro sensitivities of recombinant wild type (wt) and K65R mutant RT to dideoxynucleoside triphosphate (ddNTP) inhibitors, using a variety of primer-templates. With poly(rA)-oligo(dT), the K65R mutant showed slight increases in Ki for ddTTP and 3’- azido,3’-deoxythymidine triphosphate (AZTTP) compared to wt RT, but neither wt nor K65R RT was inhibited by ddCTP or ddATP. With poly(r1)-oligo(dC), the K65R mutant showed a 2-fold increase in K,,, for dCTP and a 20-fold increase in Ki for ddCTP

nucleoside analogs, once intracellularly metabolized to their triphosphate forms, are believed to inhibit HIV-1 RT by acting both as chain terminators of the nascent DNA chain and as competitive inhibitors with respect to the natural deoxynucleoside triphosphate (dNTP) substrates . Although such drugs reduce viral load in HIV-l-infected individuals, their use has unfortunately resulted in the appearance of drug-resistant strains of HIV-1 (reviewed by Richman (1993)). In all cases to date, this resistance correlates with mutations in HIV-1 RT (Larder and St. Clair et al., 1991;Fitzgibbon et al., 1992;Gu et al., 1992Gu et al., , 1994. The molecular mechanisms of the antiviral drug resistance of HIV-1 are not yet clear. The RT in lysates of HIV-1 strains that show more than 100-fold resistance to AZT in cell culture has identical sensitivity to AZTTP in vitro compared to RT in lysates of wild type virus (Wainberg et al., 1990;Lacey et al., 1992). Martin and co-workers (1993) have used purified recombinant RT to demonstrate that the L74V mutation, responsible for single-drug resistance to ddI, may result in altered substratehnhibitor recognition by the enzyme. Our recent studies have implied that a subdomain of HIV-1 RT, in which five mutations correlated with nucleoside analog drug resistance occur, may be important for the correct binding of the dNTP complementary to the cognate template base residue (Wu et al., 1993).
We and others have shown that cross-resistance to ddC, 3TC, and ddI results from a K65R mutation in the viral RT Zhang et al., 1994). To elucidate the mechanism of resistance caused by this mutation, we used site-directed mutagenesis to introduce the K65R substitution into vectors allowing expression of both p66 and p51 subunits of HIV-1 RT. We report here that purified recombinant ~5 1 .~6 6 K65R mutant RT shows significantly decreased in vitro sensitivity to ddCTP, 3TCTP, and ddATP (the intracellular active form of ddI) but little or no change in sensitivity to ddTTP, AZTTP, or ddGTP, compared to wt enzyme. MATERIALS AND METHODS AZTTP was purchased from Moravek Biochemicals. 3TCTP was a generous gift from Glaxo Group Research (Greenford, United Kingdom). Restriction enzymes were obtained from Boehringer Mannheim. The MEGAscript" T7 polymerase transcription kit was a product of Ambion (Austin, TX). The expression vector pKK223-3, ultrapure ~T P s , rNTPs, and ddNTPs, and the homopolymeric primer-templates (PR) poly(rA)-oligo(dT),,,, and poly(rC)-oligo(dG),,,, were purchased from Pharmacia Biotech Inc. The PPT poly(rl)-oligo(dC),,,, was prepared from poly(r1) and oligo(dC),,,, (Pharmacia) as described (Martin et al., 1993). DNA oligonucleotides were synthesized by GSD (Toronto, Canada). Heteropolymeric template-primer for RT RNA-dependent DNA polymerase activity was prepared from RNA transcripts of plasmid pHIV-PBS and a synthetic 18-nucleotide DNA primer, as described Wild type. Reported as nmol dNMP incorporatedl30 midpg p51/p66 RT heterodimer. e In assays with heteropolymeric template/primer, the listed substrate dNTP concentration was varied while the concentrations of the other three dNTPs were held constant at 5 p~ each. (Arts et al., 1994). L3H1-and [CU-~~PI~NTPS were products of Amersham Corp. and DuPont.
Cloning of Wild o p e and Mutated pRT66 and pRT51-The HN-1 proviral clone HXB2 was obtained from the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, National Institutes of Health, courtesy of Drs. B. Hahn and G. Shaw. We used the polymerase chain reaction to amplify the complete RT coding sequence of HXB2 (nucleotides 2549-4228) using RTeu (5'-CTGAA'I"l'C(ATG)CCCATT-AGCCCTATTGAG-3') and RTed (5'-CTAAGC('ITACTA)TAGTACTT-TCCTGATTCGAG-3'), as forward and reverse primers, respectively. The amplified segment was digested with EcoRI and HindIII, restriction sequences built into the primers (highlighted in bold), and the resulting fragment was cloned into EcoRIIHindIII-digested expression vector pKK223-3. The initiation and stop codons (delineated by parentheses in the listed sequences) required for the appropriate protein translation were introduced immediately upstream and downstream, respectively, of the RT coding region, giving rise to construct pRT66. When RTeu was paired with reverse primer RT5ld (5'-CTAAGC(T-TACTA)GAAGG'MTCTGCTCCTAC-3'), only the first 440 amino acids of RT corresponding to those of the p51 subunit (nucleotides 2549-3868 of HXB2D) were amplified. This fragment was cloned into pKK223-3 as described for pRT66 to produce pRT51. K65R mutant constructs were made in the same way, except that the RT coding region was amplified from the HXB2-derived HIV-1 molecular clone plasmid pHIVpol65 that contains the K65R mutation . All constructs were cloned and sequenced to verify the correct nucleotide sequences.
Expression and Isolation of Wild Dpe and Mutant pRT66 and pRT5I"Escherichia coli JM109 was transformed with either wt or K65R mutant plasmid forms of pRT66 and pRT51. Transformed bacteria were grown overnight at 37 "C in LB medium containing 100 pg/ml ampicillin, then induced with isopropyl-1-thio-P-D-galactopyi-anoside (1 mM) for 4 h. Bacterial lysates were prepared using lysozyme/DNase I treatment as described (Sambrook et al., 1989). Lysates containing approximately equal amounts of the expressed pRT66 and pRT51 RT (determined by intensity of Coomassie staining of the individual subunits resolved by SDS-polyacrylamide gel electrophoresis) were mixed and incubated at 37 "C to allow formation of p51/p66 heterodimers. The resulting ~5 1 .~6 6 RT heterodimers were purified by a rapid high performance liquid chromatographic method involving sequential chromatographic steps with Q (strong anion) and S (strong cation) chemistries, similar to that previously described (Wu et al., 1993). The enzyme preparations were >95% pure by SDS-polyacrylamide gel electrophoresis analysis, and the specific activities of the isolated RT were comparable to those reported for other recombinant preparations (Muller et al., 1989;Clark et al., 1990;Deibel et al., 1990) (see Table I).
Preparation of Heteropolymeric Primer-Template-Plasmid pHN-PBS (Arts et al., 1994) was linearized with AccI and used to prepare RNA transcripts with the MEGAscriptTM transcription system according to manufacturer's directions. These 497-nucleotide RNA transcripts comprise the complete U5 region, the primer binding sequence, and a portion of the gag gene of HN-1. The 18-nucleotide DNA oligomer dPR (5'-GTCCCTGWCGGGCGCCA-3') is complementary to the HIV-1 primer binding sequence (Ratner et al., 1985). Primer dPR and the pHIV-PBS RNA transcript were mixed in a 1:l molar ratio in 50 m~ Tris (pH 7.8, 25 "C) containing 60 m M KC1, heated at 95 "C for 2 min, then allowed to cool slowly to room temperature. The resulting PPT yields a product of 149 nucleotides upon reverse transcription, corresponding to the (-)-strong stop DNA product of HIV (Arts et al., 1994).
Enzyme Assays-RT RNA-dependent DNA polymerase activity was measured as described (Wu et al., 1993), but with variable concentrations of dNTP in the assay. With the heteropolymeric Pl", the concentration of dCTP, dATP, dTTP, or dGTP was vaned while the concentrations of the other three dNTP substrates was maintained at 5 JIM each. In assays of ddNTP inhibition, reactions included saturating concentrations of both P/T and dNTP and variable concentrations of ddNTP. Kinetic parameters of the recombinant RT enzymes were calculated as described previously (Wu et al., 1993). Inhibition constants were calculated by fitting the data to the equation K,,,,, = K,(1 + [SI/K,) (Segel, 1975).

RESULTS
Cloning and Expression of wt and K65R RT-The p51 and p66 subunits of wt and K65R mutant RT were overexpressed separately in E. coli JM109. The ~5 1 1~6 6 heterodimeric forms of wt or K65R RT were formed by mixing aliquots of the p51 and p66 lysates prior to purification, as described under "Materials and Methods." High performance liquid chromatographic size exclusion analysis under conditions promoting dissociation of RT homodimers, but not of RT heterodimers (Muller et al., 1989, 19911, showed that both wt and K65R RT purified by our method were ~5 1 .~6 6 heterodimers and not merely a mixture of p51 and p66 subunits (data not shown).
RNA-dependent DNA Polymerase Activity of wt and K65R RT in the Absence of Inhibitors-A variety of homopolymeric and heteropolymeric Pi" were used to compare the RNA-dependent DNA polymerase activity of wt and K65R RT in the absence of ddNTP inhibitors. The V values for the K65R mutant were similar to those obtained with wt RT, with each of the PIT employed (Table I). No differences in K, for dTTP and dGTP were noted between the K65R mutant and wt enzymes. In contrast, the K65R mutant showed 2-3-fold increases in K, for both dATP and dCTP relative to the wt enzyme (Table I), with both homopolymeric and heteropolymeric PIT.
Inhibition of wt and K65R RT by Dideoxynucleoside IFiphosphate Inhibitors-Several ddNTPs were assessed for inhibition of wt and K65R RT (Table 11). With homopolymeric Pi", inhibition was noted only with ddNTPs that were homologous to the dNTP substrate complementary to the template strand. For example, with poly(r1)-oligo(dC) as Pi" and dCTP as substrate, only ddCTP and 3TCTP inhibited RT activity, whereas no inhibition by ddATP, ddTTP, or AZTTP was seen, even at concentrations of 50-100 p~ (Table 11). Similarly, with poly(rA)oligo(dT1 and dTTP, inhibition was noted with dd! "P and AZTTP, but not with ddCTP or ddATP. With the heteropolymeric PIT, all of the ddNTP inhibitors were able to reduce RT polymerase activity.
The K65R mutant enzyme showed approximately 20-fold increases in the K, for ddCTP and 3TCTP inhibition of dCMP incorporation into poly(r1)-oligo(dC) compared to wt RT (Table The highest concentration of inhibitor used in these inhibition studies was 50 w.
The concentration of each of the dNTPs was 5 p.

11). Similar decreases in sensitivity of the K65R mutant were
noted for inhibition by ddCTP (Fig. U), 3TCTP (Fig. lB), and ddATP ( Fig. 1C) of dCMP incorporation into the heteropolymeric PA'. The sensitivity of the K65R mutant to inhibition by AZTTP was less affected, with about 4-6-fold increases in K, relative to wt RT with either homopolymeric or heteropolymeric PPT (Table 11; Fig. W ) . Interestingly, the K65R mutant and wt enzymes had essentially equal sensitivities to inhibition by ddGTP and ddTTP of dCMP incorporation into the heteropolymeric P/T (Fig. 1, D and E ) . DISCUSSION The K65R mutation of HIV-1 RT occurs within an IKKK motif conserved in other DNA polymerases (Hizi et al., 1989;Boyer et al., 1992). Mutations correlated with resistance to antiviral drugs have been described at sites 65,67,69, 70, and 74 (Larder et al., 1989;St. Clair et al., 1991;Gu et al., 1994). These residues are in the "fingers" subdomain of HIV-1 RT (Kohlstaedt et al., 1992;Jacobo-Molina et al., 1993) and may be important for enzyme-P/T interaction (Kohlstaedt et al., 1992) and for the binding of dNTP substrate complementary to the template base during formation of the RT-PA'-dNTP ternary complex (Wu et al., 1993).
It was reported previously that the RNA-dependent DNA polymerase activity of K65R RT is less than 5% of that associated with wt enzyme (Boyer et al., 1992). We did not observe any reduction in activity of recombinant K65R RT, with both wt and K65R mutant RT having similar V values for RNA-dependent DNA synthesis determined with several PA' (Table I). Fully processed HIV-1 RT is a heterodimer consisting of 66-and 51-kDa subunits, the latter derived by proteolytic cleavage of the 66-kDa subunit by the HIV-1 protease (Rey, 1984;Hoffman et al., 1985). Although both the ~6 6 .~6 6 homodimer and ~5 1 .~6 6 heterodimer are functionally active, only the heterodimer is present in mature HIV-1 virions (Veronese et al., 1986). The differences between our results and those of Boyer et al. might be attributed to the fact that our studies were performed with the more biologically relevant ~5 1 .~6 6 heterodimer, while Boyer et al. (1992) used a ~6 6 .~6 6 homodimer. However, we have found that the p66 form of our recombinant K65R RT has activity comparable to that of the wt p66 (data not shown).
We used a variety of PA' in the present studies. The apparent K, values for each of the dNTPs with both wt and mutant RT were lower when measured with heteropolymeric PIT than with homopolymeric PA'. This may be due to differences in RT conformation in the RT-PPT complex with different primer-templates. We have noted significant changes in non-nucleoside inhibitor and monoclonal antibody binding to RT when the enzyme is complexed to different PPT. ' The K,,, and V values for our recombinant wild type RT-catalyzed incorporation of dTMP and dCMP into RNA templates were comparable to previously published values (Martin et al., 1993;Reardon et al., 1990).
No significant differences were observed in the K,,, for dTTP and dGTP, whereas 2-3-fold increases in K,,, were seen for dCTP and dATP with the mutant enzyme relative to wt RT. Nonetheless, both wt and K65R RT had similar V values no matter which dNTP was used, implying that the K65R mutation results in a selective alteration in recognition for dCTP and dATP, but not for d l T P or dGTP.
The "catalytic efficiency" of the enzyme was assessed from V/K ratios (Table I). With dCTP as substrate and poly(r1)oligo(dC) as PA', conditions which promote distributive DNA synthesis (Martin et al., 1993), the K65R mutant showed a 3-fold lower V/K compared to wt RT. However, no significant differences in V/K ratio was noted in processive DNA synthesis between the wt and K65R RT with dTTP as substrate and poly(rA)-oligo(dT) as PA'. Our findings differ from those reported with the ddI-resistant L74V mutant (Martin et al., 1993), which showed a lower VIK ratio than wild type RT for processive synthesis using dTTP and poly(rA)-oligo(dT), but no difference in V/K for incorporation of dCMP into poly(r1)oligo(dC). The K65R mutant RT showed a 20-fold increase compared to wt enzyme in the value of K, for ddCTP and 3TCTP inhibition of dCMP incorporation into poly(r1)-oligo(dC) (Table 11). Similar increases in K, were noted for inhibition by each of ddCTP, BTCTP, and ddATP of dCMP incorporation into heteropolymeric PIT (Fig. 1, A, B , and C ) . The ratio K,/K,,,, used to normalize for changes in substrate recognition by the enzyme (Segel, 1975) was increased approximately 10-fold with the K65R mutant. These findings are consistent with the 10-fold increases in IC, for ddC, 3TC, and ddI inhibition of replication of HIV-1 containing the K65R mutation . The parallel results obtained for ddC/3TC/ddI with in vivo drug sensitivity assays and for ddCTP/3TCTPIddAW with in vitro R. S enzyme studies suggest that the K65R mutation results in changes in substratelinhibitor recognition by RT. This is entirely consistent with our previous observations that the binding of these compounds is competitively inhibited by a monoclonal antibody that binds t o residues 65-73 of RT (Wu et al., 1993).
Interestingly, ddATP, ddTTP, and AZTTP were unable to inhibit dCMP incorporation into poly(r1)-oligo(dC) catalyzed by either the K65R mutant or wt RT, even at concentrations up to 100 p~. Similarly, when poly(rA1-oligo(dT1 and dTTP were used as primer-template and substrate, neither ddCTP nor ddATP was able to inhibit wt and K65R mutant RT (Table 11). However, ddCTP, ddATP, ddTTP, ddGTP, 3TCTP, and AZTTP were each effective inhibitors of wt RT-catalyzed dCMP incorporation into heteropolymeric PPT. The inhibitors ddCTP, 3TCTP, and ddATP were significantly less effective against the K65R mutant RT, whereas ddTTP, AZTTP, and ddGTP gave essentially similar inhibition of wt and mutant enzymes. Assays employing heteropolymeric PPT were carried out in the presence of saturating levels of the other three dNTPs. Chain termination may therefore be more important than competitive inhibition in ddCl3TClddI-mediated inhibition of HIV replication.
Although neither the VIK ratio nor K,,, differed significantly between wt and K65R RT for dTTP with both homopolymeric and heteropolymeric PPT, the K, and KJK, values for inhibition of K65R RT by ddlTP and AZTTP were increased approximately 6-fold, using a homopolymeric PPT. This implies some decrease in ddTTP and AZTTP recognition by the K65R RT.
With the heteropolymeric P/T, although the Ki and KJK, values for AZTTP were 4-6-fold higher with the K65R RT compared to wt, no differences were noted in the same parameters for ddTTP. We cannot yet explain the increased K, for both dd'R'P and AZTTP with the K65R mutant in the absence of corresponding changes of K, for dTTP using homopolymeric PR. In addition, the increased inhibition by ddTTP of the K65R mutant when assayed with the heteropolymeric PPT compared to the homopolymeric poly(rA)-oligo(dT) is puzzling but may be related to the effect of P/T structure on RT conformation. Moreover, HIV-1 variants containing the K65R mutation do not express decreased sensitivity to AZT in infected cells Zhang et al., 1994). Interestingly, similar findings were obtained with the ddI-resistant L74V mutant (St. Clair et al., 1991;Martin et al., 1993). In contrast, RT in lysates of HIV strains showing high level resistance to AZT in infected cells (Wainberg et al., 1990) or recombinant RT containing mutations responsible for AZT resistance (Lacey et al., 1992) do not exhibit altered sensitivity to AZTTP in vitro. Thus, although two mutations (D67N and K70R) associated with resistance to AZT occur in the same region of RT as those associated with resistance to ddI and ddC, altered recognition of substrate and inhibitor does not appear to be a factor in AZT resistance.
In summary, our data suggest that cross-resistance to ddC, 3TC, and ddI conferred by the K65R mutation in HIV-1 RT is due to selective alterations in substrate and inhibitor recognition by the enzyme. This mutation occurs in a region of RT shown to figure in dNTP binding. Studies of the changes in RT structure resulting in decreased inhibitor recognition by the K65R mutant are in progress.