Fitness Comparison of Thymidine Analog Resistance Pathways in Human Immunodeficiency Virus Type 1

Abstract

Resistance to zidovudine (ZDV) results from thymidine analog resistance mutations (TAMs) at human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) codons 41, 67, 70, 210, 215, and 219. Two mutations are possible at codon 215: Y or F. Whereas T215Y occurs alone or with M41L and L210W (TAM-1 pattern), T215F rarely occurs with these mutations or by itself; it is found instead with D67N, K70R, and K219Q (TAM-2 pattern). The L210W mutation most often occurs with M41L and T215Y and rarely occurs with the T215F or TAM-2 mutation. To explain these associations, TAMs were introduced into HIV-1_Hxb2 by site-directed mutagenesis and expressed in recombinant viruses. Viral replication kinetics, relative fitness, and infectivity were tested in the absence or presence of ZDV. Viruses carrying the 215Y mutation showed faster replication kinetics and greater relative fitness than did T215F mutants in the absence or presence of ZDV. In addition, T215Y mutants showed greater infectivity than did wild-type HIV-1 over a range of ZDV concentrations, but T215F mutants had only a modest advantage over the wild-type virus. Whereas introduction of L210W improved the relative fitness of an M41L/T215Y mutant in the presence of ZDV, introduction of this mutation into a D67N/K70R/K219Q background resulted in decreased relative fitness in the presence or absence of drug. By contrast, introduction of T215F into the D67N/K70R/K219Q background increased viral fitness in the presence of ZDV. These results help explain why T215Y but not T215F usually emerges as the first major TAM, as well as the clustering of L210W with TAM-1 mutations and T215F with TAM-2 mutations.

Regimens including the nucleoside analog zidovudine (ZDV) have contributed to reductions in AIDS-related morbidity and mortality and to the prevention of mother-to-child transmission of human immunodeficiency virus type 1 (HIV-1) (5, 8, 9, 31). However, the high prevalence of antiretroviral drug resistance among treatment-experienced patients and the increasing transmission of drug-resistant virus may limit the clinical benefits of ZDV-containing regimens for many patients (23, 34).

Resistance to ZDV results from the sequential accumulation of thymidine analog resistance mutations (TAMs) at reverse transcriptase (RT) codons 41, 67, 70, 210, 215, and 219 (2, 13, 22). The K70R mutation is usually the first mutational change in HIV-1 RT to emerge during ZDV therapy (2). Although this mutation produces only a modest (eightfold) decrease in ZDV susceptibility in molecular clones of HIV-1 (21), variants of HIV-1 carrying the K70R mutation are selected rapidly in vivo after initiation of ZDV monotherapy (7, 20). Subsequently, variants with mutations at codon 215 and 41 emerge and replace the K70R mutants. Two mutations are possible at codon 215 (Y or F), both of which involve double-nucleotide mutations (ACC to TAC [Y] or TTC [F]). The combined presence of M41L and T215Y confers a 60-fold increase in the 50% inhibitory concentration for ZDV and a 1.8-fold increase in the risk of disease progression and/or death (2, 16, 18). Continued evolution leads to the accumulation of mutations at codons 67 and 210 (13). The combined presence of three to six TAMs results in high-level (>500-fold) ZDV resistance and confers cross-resistance to other nucleoside RT inhibitors (35).

Data from several studies suggest that TAMs are found in two distinct clusters. Mutations that occur together with T215Y (including M41L, L210W, and sometimes D67N) constitute the TAM-1 cluster; mutations that occur together with K70R (including D67N, T215F, and K219Q) constitute the TAM-2 cluster (10, 26, 36). The division of TAMs into two distinct clusters has important clinical significance: ZDV-resistant viruses carrying TAM-1 mutations usually are cross-resistant to didanosine and tenofovir, whereas viruses carrying TAM-2 mutations usually remain susceptible to those drugs (27, 30). The T215Y mutation may be found by itself or in combination with M41L and L210W, but T215F rarely occurs alone or with the M41L and L210W mutations. The L210W mutation, which generally occurs in combination with M41L and T215Y, rarely occurs with the T215F or other TAM-2 mutations (37). To explore the virologic basis for this clustering, we compared the relative replicative fitness and infectivity of HIV-1 recombinants carrying various combinations of TAMs in the absence and presence of ZDV.

(These data were presented in part at the following meetings: (i) 11th Conference on Retroviruses and Opportunistic Infections, 10 to 14 February 2004, San Francisco, Calif. [abstr. 638]; and (ii) 13th International HIV-1 Drug Resistance Workshop, 9 to 11 June 2004, Tenerife, Canary Islands, Spain [abstr. 59].)

MATERIALS AND METHODS

Reagents and cells.

Zidovudine was purchased from Sigma (St. Louis, Missouri). MT-2 cells were grown in R-10 medium (RPMI 1640 [Cellgro, Herndon, Virginia] supplemented with 10% fetal calf serum, 2 mM l-glutamine, penicillin [100 U/ml], and streptomycin [100 μg/ml]). HeLa CD4-LTR/β-gal cells, kindly provided by M. Emerman through the AIDS Research and Reference Reagent Program, were propagated in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal calf serum, hygromycin B (0.2 mg/ml), and Geneticin (0.2 mg/ml).

Construction of recombinant marker viruses.

A 3,314-bp fragment encompassing the entire coding sequence of HIV-1 protease and reverse transcriptase (PR-RT) was amplified by PCR from proviral clone pNL4-3 (corresponding to nucleotides 1998 to 5312 of the HIV-1 NL4-3 sequence [http://hiv-web.lanl.gov]) and cloned into vector pGEM-T Easy vector (Promega, Madison, Wisconsin) to generate plasmid pGEM-PR-RT. Subsequently, 1266 nucleotides of the PR-RT-coding sequence (corresponding to nucleotides 2243 to 3509 in the HIV-1 NL4-3 sequence) were deleted, and a unique BstEII restriction enzyme site was introduced at the deletion junction by PCR-based site-directed mutagenesis using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, California). The PR-RT gene carrying this deletion was then cloned into pNL4-3 to yield pHIVΔPR.RTBstEII. A segment of the Salmonella enterica serovar Typhimurium histidinol dehydrogenase gene (hisD) or the green fluorescent protein gene (GFP) was then introduced into the XhoI site in nef to serve as a sequence tag, yielding plasmid pHIVΔPR.RTBstEIInef-hisD or pHIVΔPR.RTBstEIInef-GFP, respectively. The hisD gene fragment was amplified with primers 5′-CCGCTCGAGCGATATTCTGGAAAGCAATGCCAG-3′ and 5′-CTGGTCTCGAGCAGGTCAGAAAAAATACGTTC-3′, and the GFP fragment was amplified with primers 5′-CCGCTCGAGATGGTGAGCAAGGGCGAGGAGC-3′ and 5′-CCGCACGTGTCTTCTGCTTGTCGGCCATG-3′.

Site-directed mutagenesis.

Mutations at RT codons 41, 67, 70, 210, 215, and 219 were introduced into the pGEM-PR-RT plasmid using the QuikChange site-directed mutagenesis kit (Stratagene). The presence of mutant sequences was confirmed by automated sequencing of the final plasmid clone with an ABI 377 automated sequencer.

Generation of recombinant marker viruses.

Infectious recombinant marker viruses carrying the desired mutations in RT were generated as described previously (24). The RT-coding sequence was amplified from proviral DNA of infected cells at the end of virus culture and analyzed by automated DNA sequencing to verify the presence of the correct alleles at codons 41, 67, 70, 210, 215, and 219. Susceptibility of HIV-1 recombinants to ZDV was determined by a standardized assay modified for use with MT-2 cells (15).

Viral replication kinetics.

Inoculation of 500 tissue culture infectious doses was used to infect 5 × 10⁵ MT-2 cells (multiplicity of infection [MOI] of 0.001). After incubation for 2 h at 37°C, cells were washed twice with phosphate-buffered saline (PBS) and resuspended in R-10 medium at 7.5 × 10⁴ cells/ml. Triplicate 2-ml cultures were tested in the absence and presence of ZDV (0.20 μM). Viral replication was quantified by measuring HIV-1 p24 antigen production in cell-free culture supernatants. Experiments were performed three or four times, at least once with recombinants carrying the hisD marker and at least once with the GFP marker, and the results were averaged.

Growth competition assays.

Recombinant marker viruses of interest carrying the hisD or GFP sequence tag were mixed together at a ratio of 50:50, 80:20, or 20:80 and inoculated onto 1.5 × 10⁶ MT-2 cells suspended in 300 μl of R-10 medium to yield an MOI of 0.001. After incubation at 37°C for 2 h, cells were washed twice with phosphate-buffered saline (PBS), resuspended in 10 ml of R-10 medium at a concentration of 0.15 × 10⁶ cells/ml in 25-cm² tissue culture flasks, and reincubated (day 0). Cultures were passaged by inoculating 200 μl of supernatant onto 10 × 10⁶ fresh MT-2 cells every 3 or 4 days. The proportions of the two competing variants were estimated by quantifying GFP and hisD sequences present in culture supernatants on days 1, 4, 7, 11, and 14 using real-time reverse transcriptase PCR with an ABI PRISM 7000 sequence detection system (Applied Biosystems, Foster City, California). Primers GFP86F (5′-GACGTAAACGGCCACAAGTTC-3′) and GFP180R (5′-TTGCCGGTGGTGCAGAT-3′) and the probe 5′-FAM-CAAGCTGACCCTGAAGT-TAMRA-3′ (FAM is 6-carboxyfluorescein; TAMRA is tetramethylrhodamine) were used to detect the GFP fragment, and primers HisD13F (5′-TCGAGCGATATTCTGGCAAA-3′) and HisD71R (5′ CGGAAAGTACGGTCGCTACCT-3′) and the probe sequence 5′-VIC-CCTTGCCTGGCATTG-TAMRA-3′ were used to detect the hisD fragment. Viral RNA was extracted from culture supernatants using the QIAGEN kit and treated with RNase-free DNase (QIAGEN, Valencia, California). Parameters for the real-time PCR were described previously (24), except that the initial reverse transcription reaction was performed at 50°C for 30 min. Quantitative real-time reverse transcriptase PCR was performed in triplicate for each sample. Control experiments in which two wild-type recombinants tagged with the hisD and GFP markers were competed against each other showed no change in relative proportions of the two viruses over time (data not shown). Similar results were obtained in control experiments using T215Y mutants tagged with these two markers, indicating that the sequence tags had no significant effect on the relative fitness of the recombinants. For each pair of viruses tested, reciprocal growth competition assays were conducted in which the RT genes of interest were linked to hisD and GFP or vice versa; in each case, similar results were obtained. Therefore, data shown represent the means ± standard deviations (SD) of reciprocal experiments.

Estimation of viral fitness.

Quantitative estimates of relative viral fitness were calculated as described previously (33). For each pair of recombinant viruses tested, the final ratio of the two viruses was determined by quantitative real-time PCR at day 11 as described above. The relative fitness (w) of each virus was obtained from the average of the results of three independent growth competition assays (inoculated at ratios of 80:20, 50:50, and 20:80) (33). The fitness difference (W_D) was estimated by the ratio of the relative fitness values (W_D = W_M/W_L), where W_M is the more-fit virus and W_L the less-fit virus in the growth competition assay. Growth competition assays that compared two wild-type recombinants carrying the hisD and GFP markers gave a mean fitness difference (±SD) of 1.0 ± 0.02-fold. Therefore, we considered fitness differences greater than 1.1 to be significant.

Infectivity and fitness profile assays.

Viral infectivity was determined on MAGI cells (19) (obtained from the NIH AIDS Research and Reference Reagent Program) by a modification of the method of Mammano et al. (25). Cells were plated in DMEM in 24-well plates at 4 × 10⁴ cells/well the day before infection. After the medium was removed, virus stocks at an MOI of 0.05 were adsorbed onto cells in triplicate wells for 2 h. Subsequently, DMEM was added to a final volume of 1 ml/well, and cultures were incubated at 37°C. After 3 days, the medium was removed, cells were washed with PBS and lysed in 500 μl of lysis buffer (5 mM MgCl₂, 0.1% NP-40 in PBS), and β-galactosidase activity was assayed by hydrolysis of chlorophenol red-β-d-galactopyranoside (CPRG) as described previously (25). Infectivity was determined from the mean CPRG units (optical density at 570 nm) in the triplicate wells and expressed as a percentage of wild-type infectivity. To determine the replicative advantage of mutant viruses as a function of ZDV concentration, we calculated the ratio of mutant to wild-type infectivity (in CPRG units) for each mutant at ZDV concentrations ranging from 0.001 to 10 μM. Three independent experiments were performed for each mutant; the values plotted represent the mean infectivity ratios ± 95% confidence intervals obtained at each drug concentration.

RESULTS

Replication kinetics of T215Y and T215F mutants.

We compared the rate of p24 antigen production in the absence and presence of ZDV by HIV-1 recombinants that carried wild-type or mutant alleles at RT codon 215 (Fig. 1). In the absence of ZDV, replication of the T215Y mutant was slightly slower than that of the wild-type virus. By contrast, the T215F mutant had a substantially lower replication rate. In the presence of 0.20 μM ZDV, the T215Y mutant showed markedly faster replication than the wild-type virus did. Although the T215F mutant replicated somewhat faster than the wild type in the presence of ZDV did, its replication rate was substantially lower than that of the T215Y mutant.

FIG. 1.

Replication kinetics of wild-type (WT) or T215Y (215Y) or T215F (215F) recombinant HIV-1 in the absence (A) or presence (B) of 0.20 μM ZDV. The data shown are the means ± SD (error bars) of three or four independent experiments.

Relative fitness of T215Y versus T215F mutants.

The relative replicative fitness of recombinant viruses carrying the T215Y or T215F mutation alone or with different combinations of TAMs was tested in growth competition assays. In these experiments, the T215F recombinant was substantially less fit than T215Y both in the absence and presence of 2.0 μM ZDV (Fig. 2). The relative fitness differences for T215Y versus T215F were 3.7-fold in the absence of ZDV and 5.3-fold in the presence of ZDV (Table 1). Likewise, the M41L/T215F and M41L/L210W/T215F mutants were less fit than the M41L/T215Y and M41L/L210W/T215Y mutants, respectively (data not shown).

FIG. 2.

Growth competition assays between recombinant virus carrying T215Y (215Y) or T215F (215F) in the absence (A to C) and presence (D to F) of 2.0 μM ZDV. The T215Y and T215F recombinants were inoculated at a ratio of 80:20 (A and D), 50:50 (B and E), and 20:80 (C and F), respectively. The data shown are the means ± SD (error bars) of two independent experiments.

TABLE 1.

Relative fitness values of HIV-1 recombinants carrying thymidine analog resistance mutations from growth competition assays in the absence or presence of zidovudine

RT mutant tested in growth competition assaysa	ZDVb	Production of individual recombinant relative to initial inoculum (f0/i0) for virus inoculated at the following ratio:c			Relative fitnessd	Fitness difference (fold)e
		80:20	50:50	20:80
T215Y	−	1.15	1.32	3.29	1.92	3.7
T215F		0.44	0.63	0.49	0.52
T215Y	+	1.23	1.60	3.11	1.98	5.3
T215F		0.19	0.41	0.53	0.38
M41L/L210W/T215Y	−	0.47	0.78	0.79	0.68
M41L/T215Y		1.20	1.22	1.69	1.37	2.0
M41L/L210W/T215Y	+	1.81	1.71	1.20	1.57	3.6
M41L/T215Y		0.80	0.29	0.20	0.43
D67N/K70R/T215F/K219Q	−	0.37	1.75	0.43	0.82
D67N/K70R/K219Q		0.38	0.31	3.49	1.55	1.9
D67N/K70R/T215F/K219Q	+	2.04	1.35	1.09	1.50	2.2
D67N/K70R/K219Q		0.72	0.67	0.66	0.69
D67N/K70R/K219Q	−	1.17	1.69	1.74	1.54	3.1
D67N/K70R/L210W/K219Q		0.32	0.41	0.75	0.49
D67N/K70R/K219Q	+	1.18	1.61	2.53	1.78	4.2
D67N/K70R/L210W/K219Q		0.27	0.39	0.62	0.43
D67N/K70R/T215F/K219Q	−	0.17	0.55	0.79	0.50
D67N/K70R/L210W/K219Q		1.24	1.52	1.94	1.57	3.1
D67N/K70R/T215F/K219Q	+	1.22	1.39	2.61	1.74	3.2
D67N/K70R/L210W/K219Q		0.15	0.65	0.83	0.54

^aGrowth competition assays to compare the relative fitness of the indicated pairs of HIV-1 recombinants were performed with three dual infections. Viruses were inoculated at three different ratios (80:20, 50:50, and 20:80) as described in Materials and Methods.

^bAssays were performed in the absence (−) or presence (+) of 2 μM zidovudine.

^cSingle-virus production (ws) was determined as the amount of each virus produced from the dual infection at day 11 (f₀) divided by the proportion of that virus in the initial inoculum (i₀) as described previously (33). Values shown represent means of at least two independent experiments.

^dRelative fitness (w) is the average of the three single-virus production values (ws) determined at each inoculum.

^eFitness difference (W_D) between HIV-1 recombinants in the growth competition assays was estimated as the ratio of the relative fitness values (W_D = W_M/W_L), where W_M and W_L represent the fitness of the most-fit and least-fit recombinants, respectively, in each set of pairwise growth competition assays (33). Growth competition assays between isogenic recombinants differing only in their sequence tags gave mean fitness differeneces of 1.0 ± 0.02. Therefore, fitness differences greater than 1.1-fold were considered significant.

Fitness profiles of T215Y versus T215F mutants.

To determine the relative replication advantage of recombinant viruses carrying the T215Y or T215F mutation over a range of ZDV concentrations, we determined the fitness profiles for both mutants. Figure 3 shows the fitness profiles of the T215Y and T215F mutants over a range of ZDV concentrations compared to the profile of the wild-type virus and to each other. These profiles show that the T215Y mutant had a twofold or greater replicative advantage over the wild-type at ZDV concentrations ranging from 0.1 to 5 μM (Fig. 3A). Results of these comparisons may be less reliable at higher drug concentrations because so little wild-type virus is produced that the denominator approaches zero and the ratios become unstable. By contrast, the T215F mutant showed only a modest advantage compared to the wild type, with a trend toward worse fitness than the wild type at lower ZDV concentrations (Fig. 3B). Direct comparison of the T215Y and T215F mutants suggested an advantage for the 215Y mutant, but the difference was statistically significant at only a single ZDV concentration (0.5 μM).

FIG. 3.

Fitness profiles of recombinants carrying ZDV resistance mutations at codon 215. The infectivity ratio of the T215Y (A) or T215F (B) mutant to the wild-type virus or the ratio of the T215Y mutant to the T215F mutant (C) (measured as relative β-galactosidase activity) was determined in the presence of 0.001, 0.01, 0.05, 0.1, 0.5, 1.0, 5.0, and 10.0 μM ZDV. The data shown are the means ± 95% confidence limits (error bars) of three independent experiments.

Effect of L210W on viral fitness in a TAM-1 background.

Introduction of the L210W mutation into an M41L/T215Y background produced lower fitness of the resulting recombinants in the absence of ZDV (Fig. 4A). However, when tested in the presence of 2.0 μM ZDV, the M41L/L210W/T215Y mutant was fitter than the M41L/T215Y mutant (Fig. 4B). Estimated fitness differences based on the results of these growth competition experiments showed that in the absence of ZDV, the M41L/T215Y mutant had a 2.0-fold relative fitness advantage over the M41L/L210W/T215Y mutant, whereas in the in the presence of ZDV, the relative fitness was reversed, and the M41L/L210W/T215Y mutant had a relative fitness difference of 3.6-fold compared to the M41L/T215Y mutant (Table 1).

FIG. 4.

Growth competition assays comparing the fitness of recombinants carrying M41L/T215Y (41L/215Y) and M41L/L210W/T215Y (41L/210W/215Y) mutations inoculated at a ratio of 50:50 in the absence (A) or presence (B) of 2 μM ZDV. The data shown are the means ± SD (error bars) of two independent experiments.

Effects of L210W and T215F on viral fitness in a TAM-2 background.

To determine why T215F, but not L210W, occurs together with the TAM-2 cluster of mutations, we compared the relative replicative fitness of recombinant viruses carrying the D67N/K70R/K219Q mutations with L210W or T215F in growth competition assays. Introduction of T215F into a D67N/K70R/K219Q backbone resulted in a virus with lower fitness in the absence of ZDV but greater fitness in the presence of drug (Fig. 5 and Table 1). By contrast, the L210W mutation into a D67N/K70R/K219Q background produced lower fitness of the resulting recombinant (D67N/K70R/L210W/K219Q) in the absence and presence of ZDV (Fig. 5 and Table 1). Although the D67N/K70R/T215F/K219Q recombinant was less fit than D67N/K70R/L210W/K219Q in the absence of ZDV, it showed greater fitness in the presence of 2.0 μM ZDV (Fig. 5 and Table 1).

FIG. 5.

Growth competition assays comparing the fitness of recombinants carrying the L210W or T215F mutation in a TAM-2 background. Panels show the results of competition assays between D67N/K70R/K219Q versus D67N/K70R/T215F/K219Q (A and B), D67N/K70R/K219Q versus D67N/K70R/L210W/K219Q (C and D), and D67N/K70R/T215F/K219Q versus D67N/K70R/L210W/K219Q (E and F) mutants inoculated at a ratio of 50:50 in the absence (A, C, and E) or presence (B, D, and F) of 2 μM ZDV. The data shown are the means ± SD (error bars) of at least two independent experiments.

DISCUSSION

We compared the replication kinetics, relative fitness, and fitness profiles of HIV-1 recombinants carrying different TAMs in the absence and presence of ZDV. Our results demonstrated that the T215Y recombinant had faster replication kinetics and greater fitness than the T215F mutants in the absence and presence of ZDV. Whereas fitness profiles for the T215Y mutant showed a replication advantage of more than twofold over ZDV concentrations ranging from 0.1 to 5 μM, the T215F mutant showed a more modest advantage, with a trend towards poorer infectivity relative to wild-type virus at lower ZDV concentrations. Introduction of the L210W mutation into an M41L/T215Y (TAM-1) backbone increased viral fitness in the presence of ZDV but resulted in decreased fitness when introduced into a TAM-2 backbone (D67N/K70R/K219Q). By contrast, introduction of T215F into a TAM-2 backbone increased relative fitness in the presence of ZDV but resulted in decreased viral fitness in a TAM-1 backbone. Direct comparison through growth competition assays showed that a D67N/K70R/T215F/K219Q mutant was fitter in the presence of ZDV than a D67N/K70R/L210W/K219Q mutant was. Thus, it appears that the common association of the L210W and T215F mutations with TAM-1 and TAM-2 clusters, respectively, can be explained in part on the basis of the replicative advantage these mutations confer within the appropriate genetic context.

Our findings extend results of a previous study in which recombinant viruses carrying ZDV resistance mutations were competed against wild-type virus (12). In that study, mutants carrying the T215Y mutation with or without M41L were less fit than wild-type virus in the absence of drug but had a fitness advantage at ZDV concentrations of 0.01 μM or greater. However, that study did not explore the effect of the T215F mutation on viral fitness. Our results also agree with an earlier study that demonstrated adverse effects of the L210W mutation on viral fitness in the absence of ZDV (13). However, in contrast to that study, we found that the M41L/L210W/T215Y mutant was fitter than a M41L/T215Y mutant in the presence of ZDV. A possible explanation for these discrepant results could be differences in the ZDV concentration tested—whereas the earlier study used a concentration of 5 μM ZDV, we performed the growth competition assays in the presence of 2 μM ZDV. The choice of a lower concentration of ZDV for use in the current study was based on the results of 50% inhibitory concentration determinations (data not shown). Another possible explanation is that we expressed the mutant Hxb2 RTs in NL4-3 recombinants, whereas in the earlier study the RTs were expressed in an Hxb2 background. Given the close sequence homology of Hxb2 and NL4-3 pol genes, it seems unlikely that this factor accounts for the differences in our results.

Initial studies reporting the occurrence of L210W as a ZDV resistance mutation suggested that it was of marginal significance and observed relatively infrequently in clinical samples (13). Subsequent studies have shown that the L210W mutation commonly occurs together with M41L and T215Y and contributes significantly to ZDV resistance (14, 37). Moreover, didanosine and tenofovir are significantly less active against viruses that carry the 41L/210W/215Y combination (27, 30). Likewise, we have found L210W to be present in HIV-1 sequences from approximately 50% of highly treatment-experienced patients tested in our laboratory at the time of screening for entry into salvage therapy trials (M. Marcial and D. R. Kuritzkes, unpublished observations). Results of the current study showing a fitness advantage of the M41L/L210W/T215Y mutant in the presence of ZDV are consistent with the results of clinical studies that demonstrate the importance of the TAM-1 cluster. It would be interesting to explore the fitness of M41L/L210W/T215Y mutants in the presence of other antiretroviral drugs to which this cluster of mutations confers cross-resistance.

Our results provide additional insights that help explain the ordered appearance of ZDV resistance mutations. Previous work showed that the single point mutation at codon 70 (K→R) emerges first but confers relatively low-level ZDV resistance (2). Subsequently, the T215Y mutation emerges in different viral genomes and these viruses overgrow the K70R mutants. Because the T215Y substitution requires a double-nucleotide mutation (ACC to TAC), it would be expected to occur at a much lower frequency than that of the K70R point mutation, thus explaining its later emergence. Continued replication of partially resistant T215Y mutants in the setting of nonsuppressive regimens eventually results in the accumulation of the M41L, L210W, and D67N mutations. Alternatively, the D67N mutation arises in viruses already carrying a K70R mutation, with eventual accumulation of K219Q, K219E, or K219R substitution and the T215F substitution (which, like T215Y, requires a double mutation [ACC to TTC]). The poor replication kinetics and lower fitness of T215F compared to T215Y along with the modest advantage over the wild-type virus in the presence of ZDV most likely explain why this mutation rarely occurs as an initial ZDV resistance mutation or in association with M41L.

Biochemical studies show that the T215Y and T215F mutations confer resistance to ZDV by accelerating the rate at which the terminal azidothymidine monophosphate is removed through phosphorolytic cleavage, thereby allowing DNA polymerization to proceed (1, 28, 29). The aromatic side chains of Y or F at position 215 are proposed on the basis of modeling studies to enhance ATP binding through stabilizing interactions with the adenine ring of ATP, the physiologically relevant pyrophosphate donor for intracellular nucleotide excision (3). Given the structural similarities of the two amino acid side chains, the fitness difference between T215Y and T215F mutants is unexpected. A possible explanation is suggested by the RT crystal structure. Molecular modeling based on published coordinates (4) suggests that the side chain of the amino acid at position 215 is exposed to the aqueous environment at the enzyme surface (Z. Hu and D. R. Kuritzkes, unpublished observations). Whereas the hydroxyl group on a tyrosine side chain at position 215 could still interact favorably with the aqueous solvent, substitution of the hydrophobic aromatic side chain of a phenylalanine at this position potentially could have a destabilizing effect on the enzyme. Additional biochemical and structural studies are needed to explore this hypothesis.

Although our results are consistent with ZDV resistance patterns observed in the clinic, in vitro conditions cannot reproduce the full array of selective pressures to which HIV-1 is subjected in a patient receiving ZDV-containing antiretroviral therapy. For example, the host HLA haplotype can affect the rate and pattern of emergence of drug resistance mutations because of overlap between key drug resistance loci and epitopes recognized by cytotoxic T lymphocytes (11, 17). It is also possible that differences in the intracellular concentrations of nucleoside analog triphosphates and the deoxynucleoside triphosphate pool result in different selective pressures on RT (6). In addition, viral subtype and polymorphisms in RT that are not associated with drug resistance could influence the emergence of specific TAMs through mutational interactions (32).

In conclusion, HIV-1 recombinants carrying the T215Y mutation for thymidine analog resistance have a clear advantage over T215F mutants with respect to relative replicative fitness and infectivity. Differences in the fitness profiles of these two mutants relative to the wild-type virus in the presence of ZDV likely explain the relative frequency with which T215Y and T215F mutants are observed early in the evolution of ZDV-resistant viruses. Likewise, the differential effects of the L210W and T215F mutations on viral fitness in different TAM backbones help explain the uncommon occurrence of the L210W mutation in combination with T215F in clinical isolates. It will be interesting to determine whether similar fitness differences are observed in RT backbones from other (non-B) viral subtypes and to determine the influence of the RT backbone on the pattern of emergence of TAMs.