INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Retroviral populations display extensive genetic variability (9, 45). Variation in retroviral populations is generated because of mutations introduced into their genomes during replication by mammalian DNA polymerases, RNA polymerase II, and virally encoded reverse transcriptase (RT). It has been observed in vivo that approximately 32% of the mutations in retroviral replication are generated during the plus-strand DNA synthesis of reverse transcription (31). Therefore, it is likely that error-prone replication by RT is a major contributor to the variation generated in retroviral populations. The innate ability of RT to be less processive than most polymerases and/or the lack of a 3'-5' exonuclease activity contributes to the error-prone nature of retroviral DNA synthesis (5, 40, 45).

Several amino acid motifs present in different retroviral RTs have been identified that may exert an impact on the fidelity of DNA synthesis during reverse transcription. These include residues of the Tyr-X-Asp-Asp (YXDD) motif, the deoxyribonucleoside triphosphate (dNTP) binding site, the -helix H of the thumb domain, the conserved Leu-Pro-Gln-Gly (LPQG) motif, the finger domain, and specific amino acids such as E89 and F160 in human immunodeficiency virus type 1 (HIV-1) RT (1, 3, 4, 10, 13, 16-18, 29, 35, 42, 50). The mechanism by which fidelity is maintained may encompass features in RT that involve the local geometry of the active site, the proper positioning of the template-primer complex, or the global effects attributed to different conformational states of the protein (8, 18).

It was previously shown that residues 110 to 116 in HIV-1 RT are located in the active site of RT and have a moderate to high solvent accessibility, suggesting that they may play a role in dNTP binding (22, 44). Furthermore, extensive mutational analysis of residues Y115 and Q151 in HIV-1 RT revealed that substitution mutations at these sites could have drastic effects on substrate dNTP binding (35, 42) as well as resistance to nucleoside analogs (21, 46). Resistance to nucleoside analogs has also been observed with position 116 as well (48). Additional evidence for the importance of this stretch of residues in dNTP binding and catalysis was provided by the ability of the D113 and A114 mutants of HIV-1 RT that exhibit a decrease in the sensitivity to phosphonoformic acid, a pyrophosphate analog (33). It was hypothesized that these residues, in addition to the highly conserved R72, play an important role during pyrophosphate exchange (33, 43). Finally, mutations at K65 have implicated this residue in template-primer complex binding that may ultimately have effects on fidelity (15, 44).

Recently, the structure of a ternary complex of HIV-1 RT catalytically trapped with both template-primer and dTTP substrates was elucidated, which identified the residues that directly contacted the incoming dNTP (20). These included residues K65, R72, D113, A114, Y115, and Q151 of HIV-1 RT. The K65 and R72 side chains as well as the D113 and A114 main chain amide nitrogens (NH) bind to the , , and  phosphates of the incoming dTTP through hydrogen bonding interactions. In addition, it has been illustrated that the side chains of R72, Y115, and Q151 interact with the base of the dTTP. Finally, the main chain NH of residue Y115 was also shown to be in close proximity to the ribose moiety of the dTTP (20).

Since substrate dTTP and the aforementioned amino acids are in close proximity to each other, it is reasonable to hypothesize that substitutions at these positions may alter the binding of the incoming dNTP or the positioning of the template-primer complex or modify the local geometry of the substrate binding pocket (18). This alteration in turn can lead to changes in fidelity. This hypothesis is supported by evidence from mutations introduced at positions K65 and Q151 of recombinant HIV-1 RT, which exhibited decreased mispair extension in vitro as well as resistance to a number of dideoxynucleoside analogs such as 2',3'-dideoxy-3'-thiacytidine, 3'-azido-3'-deoxythymidine, 2',3'-dideoxycytidsine, and 2',3'-dideoxyinosine (14, 18, 47). Finally, a number of mutations introduced at residue Y115 of HIV-1 RT were shown to affect the frequency of misinsertions and mispair extensions (35, 36).

Primary sequence and crystal structure data can be utilized to identify amino acids in murine leukemia virus (MLV) RT that are equivalent to those in HIV-1 RT. Sequence comparisons of several RTs illustrate the presence of highly conserved amino acid motifs (39). The partial structure of the fingers and palm subdomains of MLV RT has been solved (13, 38). Comparison of the fingers and palm subdomains of HIV-1 and MLV RTs reveal similar tertiary structures despite having only ~25% amino acid sequence identity (13, 19). Furthermore, MLV RT possesses several of the structural motifs present in HIV-1 RT, including the conserved YXDD, Thr-Val-Leu-Asp (TVLD), and LPQG motifs (13, 39). Sequence alignment of common amino acid motifs within the two RTs and a comparison of the crystal structures reveal MLV RT residues that are homologous to those in HIV-1 RT (13, 19). Such comparisons strongly suggest that residues K103, R110, D153, A154, F155, and Q190 of MLV RT are equivalent to the dTTP binding residues K65, R72, D113, A114, Y115, and Q151 in HIV-1 RT, respectively. Furthermore, K103 of MLV RT has been observed to interact with the incoming dNTP, and substitutions at R110 affect processivity (2, 8). Recent studies have provided direct evidence that the F155V mutant of MLV RT is able to incorporate rNTPs during polymerization (12). It has been postulated that F155 of MLV RT may be involved in the selection of dNTPs over ribonucleoside triphosphates (rNTPs) due to steric hindrance between the 2' OH of rNTP and the F155 side chain (12).

In this study, we sought to determine whether alterations in the structure of MLV RT can affect the fidelity of reverse transcription in vivo. Specifically, we examined whether mutations at residues in the vicinity of the dNTP binding pocket as well as residues that directly contact the substrate dNTP could alter the fidelity of reverse transcription in vivo. We assessed the effects of these mutations on replication, polymerase activity, and fidelity by comparing the mutant MLV RTs to wild-type RT.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids and retroviral vectors. The construction of the MLV-based retroviral vector pGA-1 was previously described (26). The vector pGA-1 possesses cis-acting elements needed for viral replication. In addition, the vector expresses the bacterial -galactosidase gene (lacZ) and the neomycin phosphotransferase gene (neo) from a single RNA transcript initiating from the viral promoter. An internal ribosomal entry site of encephalomyocarditis virus is used to ensure the translation of neo (23). Plasmid pLGPS expresses the MLV gag and pol genes from a truncated long terminal repeat (LTR) promoter (37). Plasmid pRMBNB, which was derived from pLGPS, contains three unique restriction sites near the dNTP-binding site of MLV RT. These three restriction sites were generated by the introduction of silent substitutions to facilitate further mutagenesis. The plasmid pSV-A-MLV-env, which expresses the amphotropic MLV envelope gene from the LTR promoter and simian virus 40 enhancer, was obtained from the NIH AIDS Research and Reference Reagent Program (32). Plasmid pBSpac encodes the puromycin N-acetyltransferase gene and therefore confers resistance to puromycin (49). Plasmid pSV3.6 encodes the  subunit of the murine Na⁺,K⁺-ATPase gene and confers resistance to ouabain (28).

Generation of dNTP-binding site MLV RT mutants. A detailed description of the mutagenic oligonucleotides and strategies used to generate as well as identify mutants is available upon request. Briefly, pRMBNB was generated by two rounds of site-directed mutagenesis using a Chameleon kit (Stratagene) with mutagenic primers and pLGPS as the template. The plasmid pRMBNB was identified by the presence of three unique restriction sites, Bst1107I, NsiI, and BstB1. To ensure the absence of any undesired mutations, the BclI-to-SalI DNA fragment of pRMBNB was subcloned into pLGPS. In addition, the Q190E, Q190H, and Q190N mutants were generated in a similar manner using a Chameleon kit with either pLGPS or pRMBNB as the templates and the appropriate mutagenic primers. The Q190E mutant was generated utilizing pLGPS as the template and screened by the creation of a new Bsu36I site. The BclI-to-SalI DNA fragment of Q190E was subcloned into pLGPS to ensure the absence of any undesired mutations. The Q190E mutant was also generated in pRMBNB and screened by the creation of a new Bsu36I site. Both the Q190H and Q190N mutants were generated using the Chameleon kit with Q190E as the template and screened by the loss of the new Bsu36I site. The NsiI to SalI DNA fragments of Q190H and Q190N were subcloned into pRMBNB and sequenced to ensure the absence of any undesired mutations.

Cells, transfections, and infections. D17 dog osteosarcoma cells were transfected with MLV-based vectors or expression constructs and infected with MLV followed by selection for resistance to ouabain or G418 (a neomycin analog) as previously described (25, 26). The D17-based ANGIE P cells were maintained, transfected, and selected for drug resistance in a similar manner (17).

Assay for determining in vivo fidelity. The ANGIE P cells were plated at a density of 2 × 10⁵ cells per 60-mm-diameter dish and 24 h later were cotransfected with wild-type or mutated pLGPS and pSV3.6. The transfected cells were selected for resistance to ouabain (10⁷ M), and then resistant colonies were pooled, expanded, and plated at a density of 5 × 10⁶ cells per 100-mm-diameter dish. After 48 h, the culture medium containing GA-1 virus was harvested and used to infect D17 target cells plated at a density of 2 × 10⁵ cells per 60-mm-diameter dish. Infected D17 cells were selected for resistance to G418 (400 µg/ml) and stained with 5-bromo-4-chloro-indolyl--D-galactopyranoside (X-Gal) as previously described (17). The observation that the mutant frequency of the wild-type RT was very consistent in 20 or more experiments indicated that any reinfection of the virus producer cells that may have occurred during expansion of the producer cells did not significantly affect the frequencies of lacZ inactivation.

Virus preparation, RT assays, and Western blotting. Virus isolation, concentration, RT assays, and Western blotting were performed as previously described (17). Briefly, helper cells containing different pLGPS constructs were plated at 5 × 10⁶ cells per 100-mm-diameter dish, viruses were collected 2 days later and centrifuged at 25,000 rpm for 90 min in an SW41 rotor (Beckman) at 4°C. Viral pellets were resuspended in phosphate-buffered saline and virus was stored at 80°C.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Determination of in vivo fidelity. Construction and characterization of the ANGIE P cell line have been previously described (17). Briefly, the ANGIE P cell line expresses the MLV-based retroviral vector pGA-1 and the MLV envelope expression construct pSV-A-MLV-env (Fig. 1). The MLV gag/pol expression construct pLGPS was subjected to site-directed mutagenesis at residues constituting the dNTP-binding site of MLV RT. Subsequently, wild-type or mutated pLGPS was separately introduced into the ANGIE P cells by cotransfection with pSV3.6. Ouabain-resistant colonies were pooled and expanded; virus was harvested from these pools, serially diluted, and used to infect D17 target cells. Infected D17 cells resistant to G418 were selected and stained with X-Gal, and the frequency of lacZ inactivation was determined by dividing the number of white colonies by the total number of colonies (blue plus white colonies). The frequency of lacZ inactivation provided a measure of the inactivating mutations (white phenotype) introduced into the lacZ gene during one cycle of retroviral replication, and the virus titers were used to determine the effect of mutations on the efficiency of virus replication.

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Structures of MLV-based constructs and rapid in vivo assay to identify structural determinants in MLV RT that are important for fidelity. The ANGIE P encapsidating cell line is a D17-based cell line expressing both the MLV-based vector pGA-1 and pSV-A-MLV-env. The pGA-1 vector, which contains LTRs and all cis-acting elements of MLV, transcribes the Escherichia coli lacZ and neo genes from the LTR promoter. The internal ribosomal entry site (IRES) of encephalomyocarditis virus is used to express neo. pSV-A-MLV-env expresses the amphotropic MLV envelope from a truncated MLV LTR (LTR) and the simian virus 40 (SV40) promoter-enhancer. Transfection of the wild-type or mutated MLV gag/pol construct pLGPS into the ANGIE P cell line allows for expression of the MLV gag and pol from the LTR (step 1). Infectious viral particles are harvested (step 2) and used to infect D17 cells (step 3). G418r infected colonies are selected (step 4), resistant clones are stained with X-Gal (step 5), and the frequency at which lacZ is inactivated is determined (step 6).

Identification of the dNTP-binding site in MLV RT. Comparison of both HIV-1 and MLV RT revealed the dNTP-binding site in MLV RT. Amino acid sequence alignment of HIV-1 and MLV RTs revealed the highly conserved TVLD and LPQG motifs (Fig. 2A). This sequence comparison suggested that residues K103, R110, D153, A154, F155, and Q190 of MLV RT are equivalent to residues K65, R72, D113, A114, Y115, and Q151 of HIV-1 RT, respectively. Analysis of the crystal structures of unliganded HIV-1 RT (not bound to DNA or substrate) and MLV RT in the context of the position and orientation of these residues also revealed similarities between the two RTs (Fig. 2B and C). In addition, calculated bond distances between residues constituting the dNTP-binding sites of both RTs in the absence of ligands (13, 19) showed that most distances were within 3 Å, the length of a hydrogen bond (data not shown). The only distances greater than 3 Å involved the R110 position. These lengths were determined by the RasMol program and were characterized as an average bond distance calculated from a number of measurements between the various atoms of the side chain pairs inspected. The equivalence of these MLV RT residues with those in HIV-1 RT is further supported by previous studies (2, 8, 13, 39).

View larger version (24K): [in this window] [in a new window]   FIG. 2.   The dNTP-binding sites of MLV and HIV-1 RTs. (A) Primary sequence homology of the MLV and HIV-1 RT dNTP-binding sites. HIV-1 and MLV RTs were aligned through their TVLD, LPQG, and YXDD amino acid motifs. Residues of the dNTP-binding site of MLV RT were determined by further alignment with residues proposed to constitute the dNTP-binding site of HIV-1 RT. Two dots represent residues that are identical, whereas one dot represents conservative changes between the two RTs. Boldface lettering represents residues constituting the dNTP-binding sites in both HIV-1 and MLV RTs. (B) Structure of HIV-1 RT dNTP-binding site (modified from Huang et al. [20]). The structure shown contains amino acid positions K65, R72, D113, A114, Y115, and Q151 interacting with the dTTP substrate. Italic lettering represents the equivalent residues in MLV RT constituting the dNTP-binding site. Components of the dTTP substrate are denoted as B (base), R (ribose sugar), and P (phosphate). Dashed lines represent interactions between the amino acid residues of RT and the dTTP substrate. mc, main chain interactions between the substrate and the RT. (C) Comparison of structures of MLV and HIV-1 RTs constituting the dNTP-binding sites (13, 19). The backbone of HIV-1 RT (amino acid sequences 63 to 76, 106 to 121, and 148 to 154) was superimposed on the homologous structure in MLV RT (amino acid sequences 101 to 114, 146 to 161, and 187 to 193). Free HIV-1 RT (not bound to DNA or substrate dNTP) is represented in gray; MLV RT is represented in black. Side chains of residues creating the dNTP-binding sites of MLV and HIV-1 RTs are illustrated as wire frames. The structures were superimposed using the RasMol Program to maximize the overlaps between the peptide backbones of the HIV-1 and MLV RTs.

Mutational analysis of the MLV RT dNTP-binding site and effects on fidelity. MLV RT was mutated at six different residues hypothesized to contact the incoming dNTP. Several conserved and nonconserved substitutions were introduced at residues K103, R110, D153, A154, F155, and Q190 of MLV RT. The effects of these various mutations on the frequency of lacZ inactivation were determined and compared with the frequency of inactivation obtained with wild-type MLV RT in parallel experiments (Table 1). Utilization of the wild-type pLGPS construct resulted in the inactivation of lacZ with a frequency of 5.4% ± 0.23% (mean ± standard error [13 independent experiments]) during one cycle of retroviral replication. These results were comparable to results obtained previously (17).

Effects of mutations at the F156 position of MLV RT on fidelity. We hypothesized that the F156 residue of MLV RT played an indirect role in substrate binding because of its proximity to other residues that constitute the dNTP-binding site. Specifically, analysis of the MLV RT crystal structure suggested that the bulky hydrophobic nature of F156 might play a role in the proper positioning of Q190, which is important for dNTP binding. Therefore, we investigated the effects of seven mutations (F156I, F156L, F156M, F156Q, F156V, F156W, and F156Y) of this residue in MLV RT on the frequency of lacZ inactivation and viral replication (Table 2). As expected, substitutions at the F156 position were tolerated to a greater extent than residues that directly contacted the incoming dNTP and the effects on fidelity could be determined for four of the seven F156 mutants. The frequency with which lacZ was inactivated by mutants F156L and F156M increased by 1.6- and 1.7-fold relative to the wild-type RT (5.4% ± 0.25% [obtained from eight independent experiments]). These changes were shown to be statistically different from the frequency of lacZ inactivation by the wild-type RT (P < 0.001). Both the F156Y and F156W mutants of MLV RT displayed a slight increase in fidelity, as evidenced by the frequencies of lacZ inactivation, of 4.3% ± 0.27% and 4.3% ± 0.15%, respectively, compared to wild-type MLV RT (P < 0.05). All of these mutants displayed reductions in viral titers relative to the wild-type RT (8.1 × 10⁴ CFU/ml) that ranged from 22- to 500-fold. Finally, the F156I, F156Q, and F156V mutants displayed reductions in virus titers greater than 10,000-fold, and relative changes in fidelity for these mutants could not be determined (Table 2).

Effects of mutations at residues flanking the MLV RT dNTP-binding site on fidelity. Regions of MLV RT (positions 147 to 152 and 157 to 161) flanking the residues of the dNTP-binding site were also subjected to mutagenesis, and the effects of these mutations on the fidelity of reverse transcription and viral replication were determined. The data obtained are summarized in Table 3. The frequencies of lacZ inactivation by wild-type MLV RT obtained from six independent experiments were again highly reproducible (5.3% ± 0.26%). The T147A, L151F, K152A, C157A, and H161A mutants exhibited frequencies of lacZ inactivation that were 1.2- to 2.4-fold higher than the wild-type RT (P < 0.04). On the other hand, the R159A mutant was found to be statistically similar to wild-type RT (P > 0.05). The L151F and the T147A mutants displayed an 8- and 50-fold reduction in viral titers, respectively. In contrast, the K152A, C157A, and H161A mutants did not exhibit substantial changes in viral titers. Finally, viral replication of the V148D, L149W, and D150E mutants was reduced by >10,000-fold, and the frequency of lacZ inactivation for these mutants could not be determined (Table 3).

RT activities of the dNTP-binding site mutants. Viruses generated from either wild-type or mutated pLGPS were harvested, concentrated by ultracentrifugation, and analyzed by Western blotting (data not shown). Western blots using an anti-MLV capsid antibody were quantified to estimate the amount of capsid protein present in the viral preparations and to ensure that equivalent amounts of viral proteins were used for determination of RT activities (summarized in Table 4).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we compared the MLV and HIV-1 RT structures to identify residues of MLV RT that contact the dNTP substrate. Mutational analysis of these residues in MLV RT strongly suggests that they are critical for proper polymerase function and viral replication. The location and orientation of MLV RT residues D153, A154, F155, and Q190 are essentially identical to those of HIV-1 RT residues D113, A114, Y115, and Q151, respectively (Fig. 2A and C) (13, 19). In addition, the distances between the side chains of these residues, as well as the MLV RT residue K103, are very similar to those of the equivalent residues of HIV-1 RT (within 3 Å). However, the MLV RT residue R110 appears in some cases to be as much as 4.0 Å farther away and as much as 6.0 Å closer to the other dNTP-binding site residues in comparison to the homologous distances in HIV-1 RT. It is possible that the MLV RT finger domain that contains the K103 and R110 residues undergoes a larger movement upon binding the dNTP substrate than does the HIV-1 RT. It has been proposed that the R110 residue is involved in the conformational changes that occur during polymerization (8). Alternatively, their location may be altered in the MLV RT crystal structure, since crystallized protein lacks the thumb, connection, and RNase H domains of MLV RT. It is to be noted that all mutations at the R110 position also resulted in very low RT activities, which was consistent with previous observations (8).

Mutations of residues constituting the dNTP-binding site of MLV RT resulted in statistically significant decreases in fidelity of up to 2.8-fold relative to the wild-type MLV RT. It is remarkable that in some cases, reductions in viral titers of >1,000-fold resulted in only minor alterations in fidelity. One possible explanation is that these mutations severely affected the efficiency of specific steps in reverse transcription. For example, mutations that reduced the efficiency of minus-strand DNA synthesis initiation would be expected to have a strong effect on viral titer but not necessarily have a great effect on RT fidelity. Another possible explanation is that the mutations greatly increased the frequency of specific types of mutations. A 10-fold increase in the frequency of a transition substitution is expected to increase the overall frequency of lacZ inactivation by only about 2.5-fold (it is estimated that approximately 16% of all mutations are one transition type because 80% of the mutations are substitutions, 80% of the substitutions are transitions, and there are four different transition types). This hypothesis could be verified by characterizing the nature of mutations generated by mutant RTs. Finally, it is possible that RTs have evolved to minimize the effects of mutations on their fidelity and that it is not possible to drastically alter the accuracy of DNA synthesis during viral replication by introducing single amino acid substitutions. However, a more extensive mutational analysis of RT is needed to verify this hypothesis. Depending on the nature and frequency of the mutations generated by mutant RTs, it is possible that some of the RT mutants will alter the spectrum of viral variants generated in viral populations. However, it is unlikely that a 2.8-fold decrease in the fidelity of RT will have a large effect on the evolution of large HIV-1 populations that rapidly undergo multiple rounds of replication in infected patients (9). Under these conditions, the selective forces are likely to have a greater impact on the viral population than small changes in the mutation rate of the virus.

Most of the substitutions introduced at the F155 position of MLV RT resulted in severe reductions in viral titers (>10,000-fold). The results obtained with the F155M, F155I, and F155V mutants were consistent with previous observations (11). In addition, the result that the F155Y mutant did not significantly affect viral titer was also consistent with previous observation (11). The same study previously reported that the F155W mutant was not viable and the RT activity could not be detected 10 days after transfection of viral DNA. In our study, the F155W mutant did produce a low viral titer (2,500-fold less than the wild type) and displayed very low RT activity. These slightly different results are likely to be due to the fact that we used a single cycle assay in which extremely low levels of viral replication could be detected.

The D153A, A154S, and F155W substitutions of MLV RT increased the frequencies of lacZ inactivation by 1.6-, 1.3-, and 2.8-fold, respectively (Table 1). All three homologous residues in HIV-1 RT have been observed to form hydrogen bonds (D113 and A114) or other interactions (Y115) with the incoming dTTP substrate (20), and mutations in at least the A114 and Y115 residues have been implicated in alterations of the affinity for the dNTP substrate (34-36). Interestingly, coordination of the triphosphate moiety of the dTTP substrate by the D113 and A114 residues of HIV-1 RT is believed to be mediated by the main chain amino (NH) groups of these residues, and the side chains may not have a direct effect on dNTP binding (20). This lack of side chain involvement might account for the fact that several substitutions of the D153 position of MLV RT could complete replication (equivalent to D113 in HIV-1 RT). The D153F and D153R substitutions of MLV RT with amino acids containing large side chains displayed drastic reductions in titers (>10,000-fold) and RT activities (30- to 50-fold), whereas the D153A, D153C, D153Q, and D153S substitutions of MLV RT with smaller side chains had less severe effects on viral replication and RT activity (Table 1). The data suggest that amino acid side chains containing more than two carbons alter the active site in the polymerase in a way that is incompatible with efficient catalysis.

In this study, the F155W mutant of MLV RT exhibited the largest increase in the frequency of lacZ inactivation (2.8-fold). The effect on fidelity is in agreement with previous observations that mutations of the equivalent Y115 residue of HIV-1 RT resulted in alterations in in vitro processivity, frequency of misinsertions, and mispair extensions (18, 35, 36). In addition, the reduction of RT activity associated with the F155 mutants, including F155W (Table 4), may be due to a decrease in the affinity for substrate dNTPs that may have drastically impeded catalysis as previously proposed for the F115 position of HIV-1 RT (36). The aromatic side chains of the HIV-1 RT F115 and MLV RT F155 were shown to play a role in the selection of dNTPs over rNTPs (6, 12). Our results also support the importance of a bulky side chain at this position, since only the F155W and F155Y mutants generated detectable virus titers (Table 1). The F155V mutant of MLV RT was observed to incorporate UTP (12). The F155V mutant in our study exhibited an 18-fold reduction in RT activity, and the incorporation of rNTPs into the proviral DNA in vivo may account for the >10,000-fold reductions in titer (Table 4).

The majority of substitutions introduced at the K103, R110, and Q190 positions in MLV RT were characterized by defects in viral replication (Tables 1 and 4). Several mutants of the R72 and Q151 positions of HIV-1 RT that are equivalent to the R110 and Q190 of MLV RT were previously shown to result in drastic reductions in processivity as well as nucleotidyltransferase activity (27, 42, 43). Furthermore, similar observations were made for several substitutions of the R110 residue and the Q190N substitution of MLV RT (8, 24). Therefore, our results support previous studies indicating that these residues are critical for catalysis (8, 18, 30, 42-44). The K65A and Q151N mutants of HIV-1 RT were previously observed to be more accurate than the wild-type HIV-1 RT in vitro (18). However, the equivalent Q190N mutant of MLV RT reduced viral titers by >10,000-fold, and its effect on in vivo fidelity could not be determined. The Q190M mutant of MLV RT exhibited no change in fidelity, viral replication, or RT activity (Tables 1 and 4), which was in agreement with previously published in vitro results obtained with the equivalent Q151M mutant of HIV-1 RT (18).

Our results indicate that the size of the substituted amino acid side chain at the F156 position of MLV RT can affect fidelity, viral replication, and RT activity (Tables 2 and 4). The side chain of the equivalent F116 position of HIV-1 RT is believed to form a part of the pocket that accommodates the 3' OH of the dNTP substrate (20). In addition, analysis of MLV RT crystal structure suggests that the F156 side chain might also be important for correct positioning of the Q190 side chain so that it can properly contact the dNTP substrate (13). Small side chains (F156I, F156Q, and F156V) exhibited no detectable levels of viral titer, whereas mutants possessing larger side chains (F156L, F156M, F156W, and F156Y) permitted viral replication (Table 2). It is possible that substitutions with amino acids possessing smaller side chains disrupt interactions between the F156 and Q190 residues in MLV RT, leading to a defect in reverse transcription.

Finally, as expected from previous analyses of the D110 position of HIV-1 RT, which is involved in the coordination of the metal cation, the equivalent D150 residue of MLV RT was highly conserved and did not tolerate the D-to-E substitution (Table 3) (13, 33, 34). Substitutions of the V148 and L149 positions also resulted in undetectable viral titers, perhaps because of their proximity to D150. It is interesting to note that the L151F substitution C terminal to D150 was tolerated and resulted in a 2.4-fold decrease in fidelity (Table 3).

The results of these studies suggest that amino acids that directly contact the dNTP substrate can have an affect on the efficiency of viral replication as well as fidelity of reverse transcription. In general, the substitution mutations of residues that directly contacted the substrate dNTP resulted in drastic reductions in viral titers. Therefore, future studies will be targeted to residues that are in close proximity to the residues that contact the substrate. These changes may be more likely to alter the dNTP-binding site and affect processivity and fidelity without affecting the essential residues that contact the dNTP. In addition, experiments to determine the spectrum of mutations generated by specific mutants of RT that led to the largest increases in the overall mutation rate will be performed.