Comparative Fitness of Multi-Dideoxynucleoside-Resistant Human Immunodeficiency Virus Type 1 (HIV-1) in an In Vitro Competitive HIV-1 Replication Assay

Abstract

We examined whether human immunodeficiency virus type 1 (HIV-1) fitness was altered upon the acquisition of a set or subset of five mutations (A62V, V75I, F77L, F116Y, and Q151M) in the pol gene, which confers resistance to multiple dideoxynucleosides (MDR), as well as the zidovudine resistance-associated mutation T215Y, using a competitive HIV-1 replication assay in a setting of an HXB2D genetic background. Target H9 cells were exposed to a 50:50 mixture of paired infectious molecular clones, and HIV-1 in the culture supernatant was transmitted to new cultures every 7 to 10 days. The polymerase-encoding region of the virus was sequenced at various time points, and the relative proportion of the two viral populations was determined. In the absence of drugs, the comparative order for replicative fitness was HIV-1_{62/75/77/116/151} > HIV-1_77/116/151 > HIV-1₁₅₁ > wild-type HIV-1 (HIV-1_wt) > HIV-1_{75/77/116/151} > HIV-1_151/215 > HIV-1₂₁₅. In the presence of zidovudine or didanosine, the order was HIV-1_{62/75/77/116/151} > HIV-1_77/116/151 > HIV-1_{75/77/116/151} > HIV-1₁₅₁ > HIV-1₂₁₅. HIV-1_215S(TCC), a putative intermediate infectious clone for HIV-1₂₁₅, replicated comparably to HIV-1_wt, while two putative intermediates for HIV-1₁₅₁ [HIV-1_151L(CTG) and HIV-1_151K(AAG)] replicated much less efficiently than HIV-1_wt and HIV-1₁₅₁, suggesting that for HIV-1₁₅₁ to develop, two base substitutions are likely to occur concurrently or within a short interval. These data may illustrate the molecular basis by which HIV-1₁₅₁ emerges much less frequently than HIV-1₂₁₅. The present data also demonstrate that several MDR HIV-1 variants are more fit than HIV-1_wt in the absence of drugs and that resistance-associated mutations and drug pressure are critical variates for HIV-1 fitness.

Reverse transcriptase (RT), a key enzyme in the replication cycle of human immunodeficiency virus type 1 (HIV-1), has become a major target for antiviral chemotherapy (3, 13, 20). Growing bodies of clinical data have shown that drugs targeting RT can increase CD4 lymphocyte counts, reduce the incidence of opportunistic infections, and prolong survival in HIV-1-infected patients. To achieve greater virus suppression and delay or circumvent the development of drug-resistant HIV-1 variants, aggressive combination chemotherapy using multiple antiviral drugs, e.g., two RT inhibitors plus one or two protease inhibitors, has been employed (4, 20). The benefits of antiviral drug therapy, however, are limited (2, 6, 11, 14, 15, 23, 26) by the emergence of drug-resistant HIV-1 variants (6). Moreover, the appearance of multi-dideoxynucleoside-resistant (MDR) HIV-1 has complicated combination chemotherapy using nucleoside RT inhibitors in certain patients (24, 26, 28). Recently, we and others have reported that in patients receiving long-term combination chemotherapy with zidovudine plus zalcitabine or zidovudine plus didanosine, HIV-1 can develop a set of five MDR mutations in the pol gene, Ala62→Val (A62V), V75I, F77L, F116Y, and Q151M, which confers viral resistance to multiple dideoxynucleosides (12, 24, 26, 28).

To define the consequences of such multiple amino acid substitutions in RT, the biochemical properties of RT mutants and the replication kinetics of HIV-1 have been studied by several groups. However, such data mostly provide limited information or are controversial (10, 19, 24), because the replication rates of viruses were determined primarily in independent cultures using p24 Gag protein production as the endpoint, thereby preventing a direct comparison of relative viral fitness.

To address this issue, we used a competitive HIV-1 replication assay (CHRA) with HIV-1 variants carrying single or multiple MDR mutations in the presence and absence of drug pressure. The data show that this assay provides a clearer understanding of virus fitness as represented by replication rates and degrees of drug resistance. This test system may be of use in predicting the genotypic and phenotypic changes of HIV-1 in the course of HIV-1 infection and antiviral therapy.

MATERIALS AND METHODS

Reagents and cells.

Zidovudine and didanosine were purchased from Sigma (St. Louis, Mo.) and Calbiochem (San Diego, Calif.), respectively. H9 and MT-2 cells were grown in an RPMI 1640-based culture medium supplemented with 10% fetal calf serum (HyClone, Logan, Utah), 2 mM l-glutamine, penicillin (50 U/ml), and streptomycin (50 μg/ml). HeLa CD4-LTR/β-gal cells (13), kindly provided by M. Emerman through the AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health, were propagated in Dulbecco modified Eagle medium supplemented with 10% fetal calf serum hygromycin B (0.1 μg/ml), and geneticin (0.2 μg/ml).

Generation of infectious HIV-1 clones.

Infectious molecular HIV-1 clones with mutations of interest were constructed by using the pHXB2RIP7-based plasmid, pSUM9, as previously described (19, 27). An HIV-1 clone carrying the wild-type RT is designated HIV-1_wt. HIV-1 clones carrying the amino acid substitution Q151M (CAG→ATG) in the RT-encoding region, two amino acid substitutions (Q151M and T215Y), three amino acid substitutions (F77L, F116Y, and Q151M), four amino acid substitutions (V75I, F77L, F116Y, and Q151M), five amino acid substitutions (A62V, V75I, F77L, F116Y, and Q151M), and the zidovudine resistance-associated mutation (T215Y) are designated HIV-1₁₅₁, HIV-1_151/215, HIV-1_77/116/151, HIV-1_{75/77/116/151}, HIV-1_{62/75/77/116/151}, and HIV-1₂₁₅, respectively. Degrees of drug sensitivity of these infectious clones have been previously published (18, 19, 27). Two possible intermediate infectious clones for HIV-1₁₅₁ carrying Q151L (CAG→CTG) and Q151K (CAG→AAG) substitutions were designated HIV-1_151L and HIV-1_151K, and those for HIV-1₂₁₅ carrying T215N and T215S were designated HIV-1_215N and HIV-1_215S, respectively. Determination of the nucleotide sequences of infectious clones obtained following transfection and propagation confirmed that each had the intended mutations.

Viral titration.

To determine virus titers, MT-2 cells (2,000 cells/well) in 96-well flat-bottomed microtiter culture plates (Costar, Cambridge, Mass.) were exposed to each infectious clone which had been prepared in H9 cells and serially diluted. MT-2 cultures were examined for cytopathic effect on day 7 of culture, and the 50% tissue infectious dose (TCID₅₀) was determined by the method of Reed and Muench (22). All titration assays were performed in six replicates.

CHRA.

Two titrated infectious clones to be examined in the CHRA were combined and added to freshly prepared H9 cells (3 × 10⁵) in the presence or absence of various concentrations of zidovudine or didanosine. To ensure that the two infectious clones to be compared were of approximately equal infectivity, a fixed amount (200 TCID₅₀) of one infectious clone was combined with three different amounts (100, 200, and 300 TCID₅₀) of the other infectious clone. On day 1, one-third of infected H9 cells were harvested and washed twice with phosphate-buffered saline, and cellular DNA was purified by using Instagene Matrix (Bio-Rad, Hercules, Calif.) according to the manufacturer’s protocol. Purified DNA was subjected to nested PCR and sequencing as described below. The HIV-1 coculture which best approximated a 50:50 mixture on day 1 was further propagated, and the remaining cultures were discarded. Every 7 to 10 days, the cell-free supernatant of virus coculture (1 ml) was transmitted to new uninfected H9 cells (1.5 × 10⁵ in 1 ml), 8 ml of culture medium was added on the following day, and half of the culture medium was replenished with an equal amount of fresh medium every 3 to 4 days. The cells harvested at the end of each passage were subjected to direct DNA sequencing, and a viral population change was determined. Each time the cell-free supernatant was harvested, the viability of H9 cells producing the virus was >90%. The viability of uninfected H9 cells used was always ∼100%. To ensure that each infectious clone preparation contained no unexpected base or amino acid substitutions, we passaged each clone in H9 cells as described above for 6 weeks (six passages), harvested the cells, and examined the nucleotide sequence of integrated HIV-1 genome. The persistence of the original amino acid substitution(s) was confirmed for all infectious clones used in this work.

Determination of nucleotide sequences of HIV-1.

Nucleotide sequences of the RT-encoding region were determined as previously described (12). Briefly, each purified DNA was first subjected to nested PCR. The first round of PCR consisted of 35 cycles with a 55°C annealing temperature and used primers SA009 (5′TTT AAA TTT TCC CAT TAG CCC TAT-3′) and SA015 (5′-ACT CCA TGT ACT GGT TCT TTT AGA-3′), which generated a fragment including codons 1 through 272 of RT. First-round PCR products (1 μl) were used directly in the second round of 25 cycles at a 55°C annealing temperature, with primers 881MF (5′-TGT AAA ACG ACG GCC AGT CCC GGG ATG GAT GGC CCA AAA GTT AAA CAA-3′) and 891MR (5′-CAG GAA ACA GCT ATG ACC GCT AGC CCA ATT CAA TTT TCC CAC TAA-3′), which included the M13 forward and M13 reverse standard primers, respectively. This generated a fragment which spanned codons 16 through 266 of RT, containing an M13 tail at each end.

Second-round PCR products were purified with PCR Select III columns (5 Prime→3 Prime, Inc., Boulder, Colo.) and sequenced directly, using both M13 forward and M13 reverse dye-labeled primers on a model 373 automated DNA sequencer (Perkin-Elmer Corp., Foster City, Calif.). Following electrophoresis, sample files were processed with Factura software (Perkin-Elmer). The forward and reverse electropherograms for each sample were aligned by using AutoAssembler software (Perkin-Elmer). At the relevant nucleotide positions, for example, the peak heights of the nucleotide bases representing the wild-type and mutant coding sequences were compared.

Determinations of individual nucleotide mixture ratios.

Determinations of individual nucleotide mixture ratios represent estimations based on relative peak heights as previously described (12). For example, equal peak heights for both wild-type and mutant bases would indicate a 50:50 ratio, and a peak at twice the height of the other would indicate a 67:33 ratio. Such analysis was conducted solely by one analyst to avoid variations in scoring. Each relative peak height defined was an estimation by the sequence analyst, who categorized the peak height of the major peak as equal to or 1.25-, 1.5-, 2-, 2.5-, 3-, or 4-fold higher than the other peak. Such categorizations corresponded to approximately a 50, 55, 60, 67, 70, 75, or 80% representation of the major peak within the mixture; mixtures composed of less than 20 to 25% of the minor population were not reliably determined due to background peaks and were not used. The determination of mixture ratios was made on low-background sequence data sets with relatively even peak heights, both of which are the advantages of the FS-dye primer methodology (Perkin-Elmer); with this method the same two primers, the M13 forward primer for the forward nucleotide sequence and the M13 reverse primer for the reverse nucleotide sequence, were used throughout the project, thereby greatly reducing primer variations and their influence on sequencing results. Moreover, nucleotide mixtures were determined based on the average of the forward and reverse results, and we found that there was general concordance between the sequences of the two strands. For cocultures involving clones which contained more than one mutation, we determined the ratios based on the average mixtures of each mutation. All approximations of viral proportions were made in a blinded and nonordered manner.

Determination of infectivity and cytopathic effect of infectious clones.

DNA (1 μg) from each molecular clone was transfected into Cos-7 cells by using LipofectAmine as specified by the manufacturer (Life Technologies, Gaithersburg, Md.). Infectious virions were harvested on days 1 to 4 and were subjected to determination of infectivity and cytopathic effect in MAGI and MT-2 cells as described above. All experiments were performed in triplicate. In the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay, each infectious clone was 2× diluted and inoculated to MT-2 cells (2,000 cells/well) in 96-well flat-bottomed microtiter culture plates (Costar), and the cytopathic effect was determined on day 7 in culture. All MTT assays were performed in triplicate.

MAGI assay.

Replication rates of different infectious clones were assessed by MAGI assay (13). Briefly, target cells (HeLa CD4-LTR/β gal; 10⁴ cells/well) were plated in 96-well flat-bottomed microtiter culture plates. On the following day, the medium was aspirated, cells were exposed to various infectious clones, and the cell culture was replenished with fresh medium. Forty-eight hours after viral exposure, the total number of blue cells in each well was determined as previously described (13, 19). All MAGI assays were performed in triplicate.

RESULTS

CHRA in the absence of drugs.

When examined in the conventional HIV-1 propagation assay, no difference was detected among the replication rates of HIV-1 clones with or without the MDR mutations in the absence of drugs (19). It is possible, however, that such an assay fails to detect subtle differences in viral replication kinetics. In the present study, postulating that CHRA can directly compare replication rates between any two clones, we first compared the replication rate of HIV-1_wt to those of various infectious clones carrying MDR mutations in CHRA in the absence of drugs over passages.

We found that three infectious MDR clones (HIV-1₁₅₁, HIV-1_77/116/151, and HIV-1_{62/75/77/116/151}) outgrew HIV-1_wt by passages 2 to 6 (Fig. 1a to c). HIV-1_wt, however, outgrew three other clones (HIV-1₂₁₅, HIV-1_151/215, and HIV-1_{75/77/116/151}) by passage 4 (Fig. 1d to f). All MDR clones were also paired with each other and examined in the absence of drugs. As illustrated in Fig. 2, HIV-1_{62/75/77/116/151} outgrew all clones tested (Fig. 2a to c). Notably, HIV-1_{75/77/116/151} had the lowest replicative ability among the four MDR infectious clones examined. We also noted that HIV-1₂₁₅ had the least replicative ability among all clones tested, including HIV-1_wt (Fig. 1 and 2). Taken together, the comparative order of replication rates in the absence of drugs is HIV-1_{62/75/77/116/151} > HIV-1_77/116/151 > HIV-1₁₅₁ > HIV-1_wt > HIV-1_{75/77/116/151} > HIV-1_151/215 > HIV-1₂₁₅ (Table 1).

FIG. 1.

CHRA of HIV-1_wt versus mutant HIV-1 clones in the absence of drugs. Data were generated based on relative peak heights in electropherograms produced from direct DNA sequencing of the HIV-1 genome at various passages. All approximations of percent population were made in a blinded and nonordered manner in two or more independent assays. Representative data are shown. In one experiment, two infectious clones (HIV-1_wt and HIV-1_215N(AAC)) were combined at three different ratios, 70:30, 50:50, and 30:70, producing the same result: that HIV-1_wt outgrows HIV-1_215N(AAC) in the absence of drugs (panel g).

FIG. 2.

CHRA of mutant HIV-1 clones in the absence of drugs. Data were generated based on relative peak heights in electropherograms produced from direct DNA sequencing of the HIV-1 genome at various passages. All approximations of percent population were made in a blinded and nonordered manner in two or more independent assays. Representative data are shown.

TABLE 1.

Comparative orders for replicative fitness of MDR and zidovudine-resistant HIV-1 clones

Drug	Order (greatest to lowest fitness)
None	HIV-162/75/77/116/151 > HIV-177/116/151 > HIV-1151 > HIV-1wt > HIV-175/77/116/151
Zidovudine (0.5 and 3 μM)	HIV-162/75/77/116/151 > HIV-177/116/151 > HIV-175/77/116/151 > HIV-1151 > HIV-1wt
Didanosine (2 and 10 μM)	HIV-162/75/77/116/151 > HIV-177/116/151 > HIV-175/77/116/151 > HIV-1151 > HIV-1wt
None	HIV-1215S(TCC) ≥ HIV-1wt > HIV-1215Y(TAC) > HIV-1215N(AAC)
Zidovudine (0.05 μM)	HIV-1wt > HIV-1215S(TCC) ≈ HIV-1215Y(TAC) > HIV-1215N(AAC)

CHRA in the presence of drugs.

Drug-resistant HIV-1 variants, in theory, should replicate faster than HIV-1 with a lesser degree of resistance than wild-type HIV-1 in the presence of drugs. Indeed, in the conventional HIV-1 propagation assay, infectious MDR clones exhibited different replication profiles in the presence of zidovudine or didanosine, although they replicated comparably in the absence of drugs as described above (19). Employing CHRA, we paired and examined infectious MDR clones in the presence of multiple concentrations of zidovudine or didanosine. In the presence of 0.5 μM zidovudine or 2 μM didanosine, HIV-1₁₅₁, possessing the lowest level of resistance to zidovudine and didanosine among the MDR variants tested (19) (see the legend to Fig. 3), was outgrown by the other three infectious clones (HIV-1_77/116/151, HIV-1_{75/77/116/151}, and HIV-1_{62/75/77/116/151}) (Fig. 3a to f). In the presence of 2 μM zidovudine or 10 μM didanosine, however, the domination occurred earlier, indicating that the replication rates of HIV-1 variants are more affected by greater drug pressure. At the end of the CHRA, the RT-encoding region of each HIV-1 mixture spanning codons 6 to 247 was sequenced, and no mutations other than the intended mutations were identified throughout this work.

FIG. 3.

CHRA of mutant HIV-1 clones in the presence of low (0.5 μM zidovudine and 2 μM didanosine; open symbols) and high (2 μM zidovudine and 10 μM didanosine; closed symbols) drug concentrations. The 50% infective concentrations of zidovudine against HIV-1_wt, HIV-1₁₅₁, HIV-1_77/116/151, HIV-1_{75/77/116/151}, and HIV-1_{62/75/77/116/151} were 0.043, 0.67, 4.8, 18, and 12 μM, respectively; those of didanosine were 2.1, 8.8, 18, 63, and 19 μM, respectively (19).

Comparative replicative fitness and drug resistance of HIV-1_{75/77/116/151} and HIV-1_{62/75/77/116/151}.

HIV-1_77/116/151 and HIV-1_{62/75/77/116/151}, both possessing a moderate level of resistance to zidovudine and didanosine (19) (see the legend to Fig. 3), outgrew HIV-1₁₅₁ and HIV-1_{75/77/116/151} (Fig. 3a to d and g to j) when tested in the presence of drugs. Although HIV-1_{75/77/116/151} has the highest level of drug resistance among the MDR variants, in the presence of zidovudine in the conventional propagation assay it replicated more slowly than the other three MDR variants (P < 0.01, P < 0.05, and P < 0.01 compared to HIV-1_{62/75/77/116/151}, HIV-1_77/116/151, and HIV-1₁₅₁, respectively) (19). However, it was possible that the drug concentrations used were not high enough to give sufficient drug pressure to HIV-1_{62/75/77/116/151}, and HIV-1_{75/77/116/151} could not outgrow HIV-1_{62/75/77/116/151}, which is less resistant. Therefore, higher concentrations of zidovudine (20 μM) and didanosine (40 μM), which are not clinically attained in patient plasma, were used. As shown in Fig. 4a and b, HIV-1_{62/75/77/116/151} still outgrew HIV-1_{75/77/116/151}. These data suggest that HIV-1_{62/75/77/116/151} replicates at higher levels than HIV-1_{75/77/116/151} through the addition of A62V despite being less drug resistant. It is also possible that the drug resistance level per se does not determine replication rates, as assessed in CHRA for certain variants.

FIG. 4.

CHRA of HIV-1_{75/77/116/151}, HIV-1_{62/75/77/116/151}, HIV-1₁₅₁, and HIV-1₂₁₅. HIV-1_{75/77/116/151} and HIV-1_{62/75/77/116/151} were examined in CHRA under high selective drug pressures, 20 μM zidovudine (a) and 40 μM didanosine (b). HIV-1₁₅₁ and HIV-1₂₁₅ were examined in the absence (c) and presence of 1 (d) and 2 (e) μM zidovudine.

HIV-1₁₅₁ outgrows HIV-1₂₁₅ with or without zidovudine.

It has been noted that HIV-1 carrying Q151M and its associated substitutions seldom acquires zidovudine-associated mutations, particularly the T215Y substitution (12, 24, 26–28). As assessed in a conventional viral propagation assay, there was no significant difference in the replication rates among HIV-1_wt, HIV-1₂₁₅, HIV-1₁₅₁, and HIV-1_151/215 (19). Therefore, we conducted CHRA for these four infectious clones in the absence of drugs (Fig. 1d and e; Fig. 2g and h). The comparative order of replicative fitness was HIV-1₁₅₁ > HIV-1_wt > HIV-1_151/215 > HIV-1₂₁₅. We also noted that when tested in the presence of zidovudine (1 and 2 μM), HIV-1₁₅₁ invariably outgrew HIV-1₂₁₅ (Fig. 4d and e).

Possible intermediate HIV-1 variants for HIV-1₁₅₁ replicate poorly.

The codon change for Q151M substitution (CAG→ATG) (27, 28) is presumed to occur via either a CTG (Leu) or AAG (Lys) intermediate. To determine whether possible intermediate HIV-1 variants had favorable replicative fitness over HIV-1_wt, two infectious clones, HIV-1_151L(CTG) and HIV-1_151K(AAG), were generated. These two variants proved to replicate poorly compared to HIV-1_wt and HIV-1₁₅₁ in MAGI and MT-2 cells (Fig. 5). In contrast, two possible intermediate variants for HIV-1 carrying the T215Y substitution (ACC→TAC) (18, 19), HIV-1_215N(AAC) and HIV-1_215S(TCC), replicated comparably to HIV-1_wt (Fig. 5) in MAGI assay. However, as examined in CHRA, HIV-1_215N(AAC) was readily outgrown by HIV-1_wt in the absence of drugs (Fig. 1g and Fig. 6b). We also found that HIV-1_215S(TCC) replicated comparably to HIV-1_wt in the absence of zidovudine, although in the presence of zidovudine (0.05 μM), HIV-1_wt outgrew HIV-1_215S(TCC) (Fig. 6).

FIG. 5.

Replication profiles of two HIV-1_T215Y intermediates and two HIV-1_Q151M intermediates. The infectivity and cytopathic effect of HIV-1_T215Y and its two intermediates, HIV-1_T215N HIV-1_T215S, and of HIV-1_Q151M and its two intermediates, HIV-1_Q151K and HIV-1_Q151L, were examined in MAGI cells and MT-2 cells. Cos-7 cells were transfected with the same amount of each DNA preparation (1 μg); the infectivities of culture supernatants harvested on days 1 to 4 were determined in MAGI assays (a), and the cytopathic effect against MT-2 cells was determined by MTT assays (b).

FIG. 6.

CHRA of intermediate infectious clones for HIV-1₂₁₅ in the presence (closed symbols) or absence (open symbols) of 0.05 μM zidovudine.

DISCUSSION

In a CHRA, the relative fitness of two clones may be directly compared since two virus populations in culture compete with each other until one clone outgrows the other (8, 9). In contrast, the conventional HIV-1 culture assay, in which p24^gag protein is often used as an endpoint, does not accurately define small differences in the replication rates among HIV-1 isolates.

In the present work using CHRA, we found that in the absence of drugs, three MDR HIV-1 clones (HIV-1₁₅₁, HIV-1_77/116/151, and HIV-1_{62/75/77/116/151}) exhibited greater fitness than HIV-1_wt, although there were no discernible differences among these infectious clones in conventional virologic or enzymatic assays (19, 29, 30). There are only few studies documenting that mutant viruses have a replicative advantage over wild-type viruses when examined in the absence of drugs. A mutant polyomavirus selected in Friend erythroleukemic cells was reported to display a cis-acting growth advantage over the parental strain when tested in a growth competition assay (5). The growth advantage of this mutant virus was observed only in Friend erythroleukemic cells, not in other cells, indicating a cell-specific cis advantage associated with modifications in the viral regulatory region. Caliendo et al. reported described that HIV-1 carrying four zidovudine-associated mutations (HIV-1_{67/70/215/219}) outgrew wild-type HIV-1 when target peripheral blood mononuclear cells were exposed to HIV-1, stimulated with phytohemagglutinin and interleukin-2 on day 10 after infection, and cultured in the absence of drugs (1). Schmit et al. (24) have examined replication rates of three clinical HIV-1 isolates (cHIV-1), wild-type HIV-1, and two clinical MDR HIV-1 isolates corresponding to HIV-1_{75/77/116/151} and HIV-1_{62/75/77/116/151} in our study. They reported that an in vitro drug-free coculture of cHIV-1_wt and cHIV-1_{75/77/116/151} resulted in a loss of the latter and that a coculture of cHIV-1_wt and cHIV-1_{62/75/77/116/151} resulted in a loss of the former by 5 to 10 weeks in culture, in agreement with our present data obtained by CHRA. Natural selection dictates that wild-type HIV-1, in general, should have a replication advantage over essentially any other HIV-1 variants in vivo. Indeed, HIV-1, collectively known as quasispecies, can produce any possible variants in vivo, but natural drug-resistant variants are generally not seen in drug-naive individuals. Although several studies have shown that antiretrovirus therapy-naive patients can harbor zidovudine resistance-associated mutations, such observations appear to be related to primary infection with zidovudine-resistant HIV-1 (6, 7). The replication advantage of several MDR mutant HIV-1 clones over HIV-1_wt observed in this work is unlikely to be associated with the use of an HXB2D genetic background, albeit such is possible, since the observations by Caliendo et al. and Schmit et al. were made with different genetic viral backgrounds (1, 24).

It should be noted that while the order for replicative fitness in the presence of drugs remained the same when the drug concentrations were increased (Table 1), the diversion of two replication curves in the CHRA tended to occur at earlier passages (Fig. 3). It is also worth noting that HIV-1_{75/77/116/151}, the infectious clone most resistant to dideoxynucleosides among MDR HIV-1 infectious clones (19, 27, 30), was outgrown by HIV-1₁₅₁ and HIV-1_wt in the absence of drugs (Fig. 1). In the presence of drugs, however, HIV-1_{75/77/116/151} readily outgrew HIV-1₁₅₁ (Fig. 3), suggesting that under greater drug pressure, the replication advantage of one mutant HIV-1 over another becomes more evident. It should be noted, however, that even in the presence of high concentrations of drugs, HIV-1_{75/77/116/151} unexpectedly exhibited a lower replication rate than HIV-1_{62/75/77/116/151} (Fig. 4). There might be some unidentified environment(s) or factor(s) which affects the replication profile of HIV-1.

Interestingly, HIV-1₁₅₁, which emerges first during combination chemotherapy (24, 25, 27, 28), had a growth advantage over HIV-1₂₁₅ both in the presence and in the absence of drugs (Fig. 4). Nevertheless, HIV-1₁₅₁ has not been identified in RT-inhibitor-naive patients with HIV-1 infection, and its emergence during antiviral therapy is much less frequent than the emergence of HIV-1₂₁₅ (12, 27, 28). One possible explanation is that two base changes (CAG→ATG) are required for the acquisition of the Q151M substitution. Thus, emergence of the Q151M substitution would be relatively rare compared to a resistance-related substitution requiring a single nucleotide substitution. We also found that both possible intermediate viruses [HIV-1_151L(CTG) and HIV-1_151K(AAG)] replicated poorly, further reducing the likelihood of the subsequent base substitution occurring. If an intermediate for HIV-1₁₅₁ has very poor replicative ability, the virus might have to undergo two base changes concurrently or within a short interval, both relatively rare events. Such an assumption appears to be consistent with observations that HIV-1 carrying Q151M typically emerges in patients only after long-term therapy with multiple dideoxynucleosides (10, 12, 24, 26–28). It is also possible, although unlikely, that the virus undergoes other sequential base changes which involve more than two base changes to bypass the two intermediates examined in this work to acquire the Q151M substitution. As for the predominance of HIV-1₂₁₅ in clinical settings, it is possible that the T215Y substitution can be readily acquired in the presence of other preceding zidovudine resistance-related substitutions, such as M41L and K70R (17). In contrast, no amino acid substitutions that precede the emergence of HIV-1₁₅₁ have been identified (12), which may also hinder the development of HIV-1₁₅₁. It is also possible that a particular genetic background may promote the development of T215Y and preclude Q151M, or vice versa, although such a genetic background has not yet been identified (12).

The zidovudine resistance-related T215Y substitution also requires two base changes but develops much more frequently and quickly than does Q151M (12, 24, 25, 27, 28). It is possible that the T215Y substitution (ACC→TAC) is more prevalent because it confers greater fitness on HIV-1 compared to Q151M. The present data, however, demonstrate that the replication rate of HIV-1₂₁₅ is lower than those of HIV-1_wt (Fig. 1 and reference 8) and HIV-1₁₅₁ (Fig. 4). In this respect, two possible intermediate clones for HIV-1₂₁₅ [HIV-1_215S(TCC) and HIV-1_215N(AAC)] were sensitive to zidovudine (data not shown), in agreement with a previous report by Lacey and Larder (16). However, one clone, HIV-1_215S(TCC), replicated faster than HIV-1₂₁₅ and comparably to HIV-1_wt in the absence of zidovudine (Fig. 6). These data suggest that, in contrast to HIV-1₁₅₁, HIV-1₂₁₅ likely develops via these intermediates: HIV-1_215S(TCC) develops first and acquires another base change.

It should be noted that homologous recombination represents one mechanism for HIV-1 to acquire drug resistance (21, 31) and that coculture of HIV-1 isolates may possibly produce various recombinant (hybrid) forms of HIV-1. Indeed, in the present study, when HIV-1_wt and HIV-1_{75/77/116/151} were examined in CHRA, a new variant, HIV-1_116/151, emerged in the culture and outgrew HIV-1_wt after the fourth passage in one of three such assays conducted. This new variant presumably arose through genetic recombination and was apparently more fit than the two parental infectious clones. Although such a homologous recombination event was noted in only one of 75 CHRAs performed, the lack of detection of recombination with this method does not exclude its occurrence. The recombination may also occur during culture and/or PCR. During PCR, artifactual recombination tends to occur in the latter cycles (after 25 cycles) (31); therefore, PCR-related recombination may not have significant effects on the data presented here; however, caution should be used in interpreting the data obtained with CHRA.