Structured treatment interruptions in highly treatment-experienced HIV-1-infected individuals are not recommended because of the increased risk of clinical progression.1 In contrast, partial drug withdrawal has been proposed a maintenance strategy for individuals experiencing virological failure and who have no further therapeutic options.2-4
Studies of total drug discontinuation have provided an opportunity to explore the adaptive evolution of HIV-1 after antiretroviral selective pressure. Wild-type viral variants have been shown to re-emerge soon after discontinuation of therapy5-7; when pharmacological selective pressure is removed, wild-type quasispecies are expected to outgrow drug resistance mutants because they have higher fitness in a drug-free environment.6,8,9 The origin of the rebounding virus is controversial, as withdrawal of antiretroviral therapy in patients with therapeutic failure has been shown to facilitate the resurgence of an archival viral population rather than to generate back mutations of drug-resistant forms.9 In addition, the presence of minority memory multidrug resistance genomes reminiscent of earlier dominant viral variants affects the evolution of the virus during drug interruption,10 thus generating the hypothesis that minority viral genomes could again become dominant variants. Nevertheless, the most widely reported theory is based on the emergence of viral variants from the latent reservoir, implying stochastic reactivation of different clones from long-lived latently infected cells.11-14 These results indicated the existence of viral reservoirs that lead to the emergence of old viral variants. However, viral recombination with ancestral viral genomes11 and reversal of resistance mutations by backward point mutagenesis14 should not be excluded.
In contrast, studies on viral evolution under partial drug discontinuation strategies are scarce. Loss of resistance after enfuvirtide discontinuation has been suggested to be due to ongoing viral evolution (and back mutation).12 In addition, after a switch from reverse transcriptase (RT) inhibitors to protease inhibitors (PIs), drug-associated mutations rapidly disappear from the RT gene.13 After discontinuation of PI in highly treatment-experienced patients, RT resistance mutations are maintained during long-term follow-up, whereas PI resistance mutations wane, thus suggesting a beneficial role of this strategy for in vivo replication capacity.3,4 Additionally, the sudden disappearance of PI resistance mutations has been observed in some patients, resulting in total replacement of the mutant by the wild-type virus, rather than a gradual reversion of individual mutations.3 Nevertheless, the origin of these new viral variants has yet to be elucidated.
Besides the influence of antiretroviral therapy on viral outcome, pressure from the immune system also shapes HIV-1 evolution. Specifically, neutralizing antibody responses have been detected in acute infection, although they lead to extensive variation in the envelope gene, resulting in complete replacement of neutralization-sensitive viral variants.15,16 In addition, the increase in neutralizing activity is marginal and does not lead to viral suppression after total treatment interruption.17 However, neutralizing antibody responses have not been previously evaluated in studies on partial drug discontinuation.
The primary objective of the present study was to investigate the origin of partially sensitive viral variants after discontinuation of PI. To do so, we determined the pattern of evolution of viral populations sequenced over a period of 10 years at different time points before and after partial treatment interruption. We also evaluated the potential contribution of phenotypic changes, such as replication capacity, drug susceptibility, and neutralizing antibodies, to the evolution of the new viral variants.
MATERIALS AND METHODS
Samples from highly treatment-experienced HIV-1-infected patients were retrospectively collected from a previous randomized controlled study that evaluated the safety and immunological outcome of a transient PI-interruption, N(t) RT inhibitors (RTI)-only regimen versus a genotype-guided salvage therapy in subjects awaiting for additional treatment options to use against multiply resistance viruses.3 In the study arm, 4 subjects with persistent virologic failure showed a reversion to wild type of all drug resistance mutations within the protease (PR)-coding region, whereas 5 subjects maintained viruses with a similar pattern of resistance mutations within the PR gene. For the present study, we chose 2 subjects in whose viruses the resistance mutations reverted to wild type after discontinuation of PI, and 1 subject in whose viruses only 1 of 8 PR resistance mutations reverted to wild type. Selection was based on the availability of stored ancestral samples.
Population-Based Sequencing of HIV-1 pol and env
Viral RNA was extracted from plasma samples at weeks 0, 12, 24, and 48 after initiation of the previous study3 using the QIAamp Viral RNA kit (Qiagen, Barcelona, Spain). Ancestral samples from each subject, 1 per year when available, were also included. The HIV-1 env was amplified using reverse transcriptase polymerase chain reaction (RT-PCR) (SuperScript III One-Step RT-PCR System with Platinum Taq High Fidelity, Invitrogen, Barcelona, Spain) with the primer 6858U20 and 7833L22 (nucleotides 6858-6877 and 7812-7833, respectively) of the HIVHXB2 numbering system. A nested polymerase chain reaction (PCR) (Platinum Taq DNA Polymerase High Fidelity; Invitrogen, Barcelona, Spain) was then carried out with primers 6957U20 and 7672L20 (nucleotides 6957-6976 and 7653-7672, respectively). The PR and RT coding regions were amplified by RT-PCR with the primers 1633U23 and 4461L25 (nucleotides 1633-1656 and 4436-4461 respectively). Subsequently, a nested PCR was performed with the primers 1811U24 and 4335L25 (nucleotides 1811-1835 and 4310-4335 respectively). Population-based sequencing of env, PR, and RT was performed using the Big-Dye Terminator Cycle Sequencing Kit and the ABI 3100 sequence analyzer (Applied Biosystems, Barcelona, Spain). The GenBank accession numbers for the HIV-1 envelope and polymerase sequences used in this analysis were from HQ882077 to HQ882174.
Four different trees were constructed for each subject. Population sequences were modified by treating incongruent bases as gaps, and DNA sequences were translated into protein sequences. A protein alignment was performed for each data set using MUSCLE.18 DNA alignments were obtained by codon concatenation according to the aforementioned protein alignment. DNA multiple sequence alignment was used for phylogenetic reconstruction: first, we fit the best evolutionary model using jModeltest v1.0,19 and then we performed phylogenetic reconstruction using PHYML v 3.020 with the best molecular evolutionary model found in jModeltest. A total of 500 bootstrap replicates were used to test the branch support with PHYML. To remove the influence of convergent evolution at antiretroviral drug resistance mutations on the phylogenetic analysis, we excluded from the alignment codons associated with major resistance in PR and RT. Additionally, we also performed a third-base positions only analysis.21
The DNA multiple sequence alignment was tested for recombination events using the RDP322,23 and GARD program24 from the datamonkey.org.25 In addition, the SimPlot analysis was tested to compare the recombinant form related to its parental.26 We tested for positive selection using the programs SLAC and FEL27 from datamonkey.org to detect sites under positive selection. Modifications of the default parameters are indicated in the supplementary material (http://links.lww.com/QAI/A147).
Drug Susceptibility Assays
Drug susceptibility assays were performed using the PhenoSense HIV system (Monogram Biosciences, South San Francisco, CA).28 This assay is based on the use of a modified HIV-1 vector derived from the NL4-3 molecular clone, which contains an insert derived from amplification of plasma samples including the entire PR-coding region and the first 915 nucleotides of the RT-coding region. Fold change drug susceptibility was determined as the ratio between the luciferase activity obtained for each virus and the activity of the wild-type reference HIV-1NL4-3 strain.
Replication Capacity Assays
Replication capacity was measured using a modified version of the PhenoSense drug susceptibility assay.29 The relative replication capacity of the virus was determined by measuring the amount of luciferase activity produced 72 hours after infection in the absence of drug. Replication capacity is expressed as the percentage of luciferase activity produced by the vectors containing patient-derived gag-pol sequences compared with the percentage of luciferase activity from vectors containing the HIV-1NL4-3gag-pol reference sequences (100%).
Neutralizing Antibody Responses
Neutralizing antibody titers were determined using the Phenosense HIV Neutralizing Antibody Assay, which measures the inhibition of a recombinant virus in a single replication cycle assay.15 HIV-env DNA (gp160) from the plasma of infected patients was amplified by PCR and cloned within the pCXAS expression vector. Recombinant viruses pseudotyped with patient virus envelope proteins were harvested and incubated with serial-fold dilutions of heat-inactivated patient plasma samples (antibody). Virus infectivity was determined by measuring the luciferase activity expressed in infected cells. Titers were calculated as the reciprocal of the plasma dilution conferring 50% inhibition (IC50).
Plasma viremia, CD4+ T-cell counts, and antiretroviral regimens are shown in the Supplemental Digital Content 1 (see Figure S1, http://links.lww.com/QAI/A147). Population-based sequencing of the viral genome showed that the baseline viruses of all 3 subjects harbored resistance mutations in the PR and RT, which conferred resistance to all of the drugs included in their last antiretroviral regimen. We chose 2 subjects of 4 in whom viral PR sequences reverted to wild type after partial drug interruption (subjects 1 and 2), and 1 of 5 subjects in whom viral PR did not revert to wild type, despite withdrawal of PI and persistent virologic failure (subject 3). The evolution of primary resistance mutations within the PR and RT-coding regions after discontinuation of PI is depicted for each patient in Figures 1-3.
Only phylogenetic analyses of the key sequences are shown in Figures 1-3. Phylogenetic analyses of all tested sequences are shown in the Supplemental Digital Content (see Figure S2, http://links.lww.com/QAI/A147). For subject 1, the phylogenies corresponding to the PR-RT-coding region showed that newly generated viruses replicating after partial treatment interruption (weeks 24 and 48) were distantly related to its immediate temporal ancestor, suggesting that new viral variants did not evolve from baseline viruses (Fig. 1). When the PR-coding region was evaluated alone, samples from weeks 24 and 48 after interruption formed a significant cluster with an ancestral sequence from 1994. However, the phylogenies of RT and the envelope showed a statistically significant cluster of samples at baseline and at weeks 12, 24, and 48 (Fig. 1). From the drug resistance standpoint, baseline and week-12 samples were genotypically resistant to RT and PIs. In contrast, weeks 24 and 48 samples, along with the year-94c sample were fully sensitive to PI although maintained resistance to RTI. These results suggest that acquisition of the wild-type PR after discontinuation of PI could occur via recombination of the PR with an ancestral viral genome shaping a genotype highly adapted to the new antiretroviral regimen.
The pattern topology of subject 2 (Fig. 2) was slightly different because the phylogenetic trees of the PR-RT-coding region, PR-coding region, and RT-coding region showed the same topology. At weeks 12 and 24 after partial drug interruption, samples were closely related to an ancestral sample from 1996, with a statistically significant cluster in all 3 phylogenetic trees. However, the phylogeny of env showed a cluster of samples at baseline, weeks 12 and 24 (Fig. 2). In addition, the baseline sample was largely resistant to both RT and PIs. In contrast, weeks 12, 24, and 48, along with the year-96 sample, were fully sensitive to PI although maintained resistance to RTI. These results indicate that the acquisition of wild-type PR after PIs discontinuation could have resulted from recombination of at least the PR- and RT-coding regions with an ancestral viral genome also shaping a genotype highly adapted to the new antiretroviral regimen.
The phylogenies of the PR-RT, PR alone, RT alone, and env- from subject 3 were similar (Fig. 3) because they showed a unique statistically significant cluster of samples at baseline and at weeks 12 and 24. After discontinuation of PI, neither baseline samples nor weeks 12 and 24 samples formed a separated cluster with any ancestral viral genome (Fig. 3). However, the allocation of a week-48 sample at an interior position in the PR, RT, and PR-RT trees suggests that a recombination event happened late in the study.
The same topology was seen in all subjects when drug resistance mutations, according to the Stanford Genotype Resistance Interpretation Algorithm (version 6.0.1), were excluded from the analysis, or after restricting the analysis to the third-base position in the 300 codons sequenced (data not shown), suggesting that drug resistance mutations were not driving the phylogenetic observations.
To test the strength of the recombination events shown in the phylogenetic reconstructions, we performed a recombination analysis using the RDP3 and GARD programs. We detected in subject 1 a significant recombination event in the PR at weeks 24 and 48 (P < 0.001) (see Figure S3; Table, Supplemental Digital Content 1, http://links.lww.com/QAI/A147) The PR region (35-276) was shown to be recombinant, and the putative parental sequences detected were from week 12 and an ancestral sequence from 1994 (Year-94c). This result was fully supported by the SimPlot analysis (Fig. 4). Overall, new recombinant genotypes were fully resistance to RTI although maintained their genotypic susceptibility to PI.
Even when the phylogenetic tree showed that a recombination event could have happened in subject 2 by exchanging at least the PR and RT-coding regions, the analysis was not able to detect any recombination break point. This could be due to the absence of the flanking region of the PR and RT in sequences, however, the SimPlot analysis supported the observation that week 12 and 24 were the result of recombination between the baseline and the ancestral sample dated from 1996 (Fig. 4).
The inner position of samples from week 48 in the PR, RT, and PR-RT trees from subject 3 could indicate a recombination event. The analysis shows a significant recombination event in the week-48 sample for the PR-RT region in 4 of 8 algorithms (see Figure S4; Table, Supplemental Digital Content 1, http://links.lww.com/QAI/A147). The region that was detected as recombinant (37-863) spans the whole PR and a large part of the sequenced RT, and the putative parental sequences detected were the week-24 sequence and an ancestral sequence not included in the analysis.
No recombination breakpoint was detected within the envelope gene in any of the study subjects.
To understand why viral variants recombined with ancestral viral genomes in response to changes in antiretroviral pressure, drug susceptibility assays were performed with the baseline sample, one sample obtained after discontinuation of PI (contemporaneous sample), and the ancestral sample, which formed a cluster in the phylogenetic tree with the contemporaneous sample. The results for subject 1 showed that fold change in drug susceptibility to RTI was almost identical between ancestral (1994c) and baseline samples, with an increase in zidovudine resistance of contemporaneous sample (week 24). However, ancestral and contemporaneous samples were fully sensitive to PI although the baseline sample was fully resistant (Fig. 5A). These data provide an explanation for the only recombination of the PR-coding region in subject 1.
Results from subject 2 showed that RT resistance was higher in the ancestral sample (year-1996) than baseline sample, being similar to the contemporaneous sample (week 12). Resistance to PI revealed that contemporaneous and ancestral samples were fully sensitive to all 5 drugs. In contrast, the baseline sample was highly resistant to PI. Full recombination of the PR- and RT-coding regions after discontinuation would explain the re-emergence of phenotypic drug susceptibility to RTI and PI in this subject (Fig. 5A).
Because no recombination event was apparent in the phylogenetic analysis in subject 3, the most ancestral sample available was selected for the phenotypic assay. The ancestral virus (year 1999) showed a higher grade of resistance to lamivudine than the baseline virus. Resistance to PI revealed that baseline and contemporaneous (week 24) viruses were fully resistant to all 5 drugs. In contrast, the ancestral virus was mildly resistant to PI (Fig. 5A). Based on the fold change, drug susceptibility to PI and RTI in this subject could benefit from recombination with the ancestral genome.
Replication Capacity Assay
To explain recombination events in response to treatment changes, replication capacity assays in absence of drugs were performed with ancestral baseline and contemporaneous samples. The results from subjects 1 and 2 for the replication capacity were identical. Ancestral and contemporaneous samples showed higher replication capacity than baseline samples (Fig. 5B). These results could explain why recombination with ancestral viral genomes is beneficial for the fitness of the new viral variant emerging after the partial treatment interruption. However subject 3 showed a different result; the ancestral sample had a higher replication capacity than the baseline sample and the contemporaneous sample (Fig. 5B).
To explain why ancestral viral variants do not re-emerge as a whole virus rather than recombined in response to a partial treatment interruption, the virus-specific humoral immune response in each subject was evaluated for ancestral baseline and contemporaneous virus. Based on the neutralization of viral envelope pools by contemporaneous plasma samples, antibody titers were low (inhibitory concentration of <100-fold dilution) for contemporaneous viruses and higher against viruses from the ancestral sample for subjects 1 and 2. However, subject 3 had similar antibody titers against both ancestral and contemporaneous viruses (Fig. 5C). This might explain why ancestral viral genomes in subjects 1 and 2 do not re-emerge as a whole virus and why viral recombination was observed.
We analyzed the origin of partially drug-sensitive viral quasispecies, which re-emerge after switching antiretroviral treatment to a PI-sparing regimen. Several ancestral plasma samples for each patient were collected during a 10-year period before interruption and for 48 weeks thereafter. Phylogenetic studies were performed to ascertain the origin of the new viral variants. Various hypotheses have been postulated on the source of emerging variants in treatment interruption strategies.9,14,30-36 New viral variants may be the result of back mutations, emergence of minority genomes, products of viral recombination with drug-sensitive quasispecies, or re-emergence of latently infected cells.
Single genome analysis is considered to be the more rigorous approach to characterize viral quasispecies because minimizes the frequency of PCR-based recombination events, which have been shown to be between 1 and 7%.37,38 However, we decided to explore first a population-based sequencing strategy optimizing the PCR reaction conditions to reduce PCR-based recombination.39 The fact that the observed recombination events occurred in the same genetic location in consecutive longitudinal samples suggested the robustness of using population-based sequencing in this study.
Thus, our phylogenetic and recombination analysis strongly supported the hypothesis that new partially sensitive viral variants that re-emerge after partial drug interruption are the result of viral recombination events with ancestral viral genomes. We discard the possibility of resurgence of minority populations with the newly desired genotype because we detected recombination breakpoints and parental sequences in the recombination test matching with the phylogenetic trees, being unlikely to detect them if minority viral variants were outgrowing. Therefore, viral recombination would indicate an independent evolution of the PR-RT- and envelope-coding regions under antiretroviral pressure. The viral proteins PR and RT were under pharmacological pressure, and recombination of such genes with ancestral genomes conferred a phenotypic advantage, measured as replication capacity and susceptibility to the current antiretroviral regimen. However, the viral protein envelope was under constant humoral selective pressure, and, consequently, evolution of this protein differed from that of the drug-targeted enzymes, the acquisition of an ancestral envelope being unlikely. These results are consistent with those of previous studies, which showed that selection acting at the pol region has no significant effect on the evolution of the envelope gene due to the possibility of recombination between the 2 distant HIV-1 genomic regions.40,41 Moreover, several studies support in vivo viral recombination,41-43 and some have showed that the emergence of new viral strains resulted from recombination between distinct HIV-1 subtypes.44,45 In addition, in vivo viral recombination has been shown to contribute to the diversity of viral quasispecies, thus supporting the idea that recombination may be critical to adaptive evolution under constantly changing selective pressures, whether exerted by the immune system or antiretroviral therapy,46 with recombination as an important mechanism for virus evolution in vivo.
Evolution of drug resistance mutations is characterized by severe fitness loss when the virus is tested in vitro in the absence of drugs. This can be partially overcome by compensatory mutations or other adaptive changes that restore replication capacity.47,48 Therefore, when drug pressure is removed, wild-type viral variants with higher replication capacity are expected to outgrow the resistant ones. In the present study, we observed a resurgence of viral genomes with wild-type viral PR after partial drug discontinuation. Our results show the significance of resistance mutations and viral replication driving in vivo viral evolution throughout recombination events.
In addition, autologous neutralizing antibody titers are generally high and sufficiently potent to neutralize most infectious viruses in vitro. Moreover, selection of escape mutations whose kinetics is comparable with that of antiretroviral therapy in vivo has been reported.16 This observation is consistent with our data, showing effective neutralization of ancestral viruses by contemporaneous antibodies in plasma and supporting the need for viral recombination of the envelope gene.
Although it seems that partially susceptible viral variants are generated as a result of recombination with ancestral viral variants, the origin of the reservoirs of these ancestral variants remains unclear. Possible explanations for this phenomenon are as follows: the ancestral variants are found in latent cells whose stochastic activation could lead to active viral replication followed by viral recombination; the ancestral variants in the viral quasispecies that replicate at any time before the partial interruption continue to replicate at low levels despite the presence of drugs, with subsequent recombination events in response to selective forces; and pharmacological sanctuary sites where the wild-type virus is able to replicate may also play a relevant role.
Overall, resurgence of viral variants after partial drug discontinuation is the result of viral recombination with partially drug-sensitive ancestral viral genomes, an observation that is supported by the improved viral fitness and decreased drug susceptibility of the newly generated viruses. These data also indicate the extreme long-term conservation of all viral variants and advise that drug susceptibility, viral fitness, and neutralizing antibodies are major driving forces for the evolution of virus populations through viral recombination.
1. Kaufmann DE, Lichterfeld M, Altfeld M, et al. Limited durability of viral control following treated acute HIV infection. PLoS Med
2. Abadi J, Sprecher E, Rosenberg MG, et al. Partial treatment interruption of protease inhibitor-based highly active antiretroviral therapy regimens in HIV-infected children. J Acquir Immune Defic Syndr
3. Bonjoch A, Buzon MJ, Llibre JM, et al. Transient treatment exclusively containing nucleoside analogue reverse transcriptase inhibitors in highly antiretroviral-experienced patients preserves viral benefit when a fully active therapy was initiated. HIV Clin Trials
4. Deeks SG, Hoh R, Neilands TB, et al. Interruption of treatment with individual therapeutic drug classes in adults with multidrug-resistant HIV-1 infection. J Infect Dis
5. García F, Plana M, Vidal C, et al. Dynamics of viral load rebound and immunological changes after stopping effective antiretroviral therapy. AIDS
6. Harrigan PR, Whaley M, Montaner JS. Rate of HIV-1 RNA rebound upon stopping antiretroviral therapy. AIDS
7. Ruiz L, Martinez-Picado J, Romeu J, et al. Structured treatment interruption in chronically HIV-1 infected patients after long-term viral suppression. AIDS
8. Devereux HL, Emery VC, Johnson MA, et al. Replicative fitness in vivo of HIV-1 variants with multiple drug resistance-associated mutations. J Med Virol
9. Kijak GH, Simon V, Balfe P, et al. Origin of human immunodeficiency virus type 1 quasispecies emerging after antiretroviral treatment interruption in patients with therapeutic failure. J Virol
10. Briones C, de Vicente A, Molina-París C, et al. Minority memory genomes can influence the evolution of HIV-1 quasispecies in vivo. Gene
11. Bello G, Casado C, García S, et al. Co-existence of recent and ancestral nucleotide sequences in viral quasispecies of human immunodeficiency virus type 1 patients. J Gen Virol
12. Kitchen CM, Lu J, Suchard MA, et al. Continued evolution in gp41 after interruption of enfuvirtide in subjects with advanced HIV type 1 disease. AIDS Res Hum Retroviruses
13. Verhofstede C, Wanzeele FV, Van Der Gucht B, et al. Interruption of reverse transcriptase inhibitors or a switch from reverse transcriptase to protease inhibitors resulted in a fast reappearance of virus strains with a reverse transcriptase inhibitor-sensitive genotype. AIDS
14. Yerly S, Rakik A, De Loes SK, et al. Switch to unusual amino acids at codon 215 of the human immunodeficiency virus type 1 reverse transcriptase gene in seroconvertors infected with zidovudine-resistant variants. J Virol
15. Richman DD, Wrin T, Little SJ, et al. Rapid evolution of the neutralizing antibody response to HIV type 1 infection. Proc Natl Acad Sci U S A
16. Wei X, Decker JM, Wang S, et al. Antibody neutralization and escape by HIV-1. Nature
17. Trkola A, Kuster H, Leemann C, et al. Humoral immunity to HIV-1: kinetics of antibody responses in chronic infection reflects capacity of immune system to improve viral set point. Blood
18. Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res
19. Posada D. jModelTest: phylogenetic model averaging. Mol Biol Evol
20. Guindon S, Gascuel O. A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol
21. Lewis F, Hughes GJ, Rambaut A, et al. Episodic sexual transmission of HIV revealed by molecular phylodynamics. PLoS Med
22. Martin D, Rybicki E. RDP: detection of recombination amongst aligned sequences. Bioinformatics
23. Martin DP, Williamson C, Posada D. RDP2: recombination detection and analysis from sequence alignments. Bioinformatics
24. Kosakovsky Pond SL, Posada D, Gravenor MB, et al. Automated phylogenetic detection of recombination using a genetic algorithm. Mol Biol Evol
25. Pond SL, Frost SD. Datamonkey: rapid detection of selective pressure on individual sites of codon alignments. Bioinformatics
26. Lole KS, Bollinger RC, Paranjape RS, et al. Full-length human immunodeficiency virus type 1 genomes from subtype C-infected seroconverters in India, with evidence of intersubtype recombination. J Virol
27. Kosakovsky Pond SL, Frost SD. Not so different after all: a comparison of methods for detecting amino acid sites under selection. Mol Biol Evol
28. Petropoulos CJ, Parkin NT, Limoli KL, et al. A novel phenotypic drug susceptibility assay for human immunodeficiency virus type 1. Antimicrob Agents Chemother
29. Campbell TB, Schneider K, Wrin T, et al. Relationship between in vitro human immunodeficiency virus type 1 replication rate and virus load in plasma. J Virol
30. Charpentier C, Dwyer DE, Mammano F, et al. Role of minority populations of human immunodeficiency virus type 1 in the evolution of viral resistance to protease inhibitors. J Virol
31. Chun TW, Davey RT, Ostrowski M, et al. Relationship between pre-existing viral reservoirs and the re-emergence of plasma viremia after discontinuation of highly active anti-retroviral therapy. Nat Med
32. Imamichi H, Crandall KA, Natarajan V, et al. Human immunodeficiency virus type 1 quasi species that rebound after discontinuation of highly active antiretroviral therapy are similar to the viral quasi species present before initiation of therapy. J Infect Dis
33. Joos B, Fischer M, Kuster H, et al. HIV rebounds from latently infected cells, rather than from continuing low-level replication. Proc Natl Acad Sci U S A
34. Martinez-Picado J, Frost SD, Izquierdo N, et al. Viral evolution during structured treatment interruptions in chronically human immunodeficiency virus-infected individuals. J Virol
35. Noë A, Plum J, Verhofstede C. The latent HIV-1 reservoir in patients undergoing HAART: an archive of pre-HAART drug resistance. J Antimicrob Chemother
36. Verhofstede C, Noë A, Demecheleer E, et al. Drug-resistant variants that evolve during nonsuppressive therapy persist in HIV-1-infected peripheral blood mononuclear cells after long-term highly active antiretroviral therapy. J Acquir Immune Defic Syndr
37. Judo MS, Wedel AB, Wilson C. Stimulation and suppression of PCR-mediated recombination. Nucleic Acids Res
38. Meyerhans A, Vartanian JP, Wain-Hobson S. DNA recombination during PCR. Nucleic Acids Res
39. Fang G, Zhu G, Burger H, Keithly JS, Weiser B. Minimizing DNA recombination during long RT-PCR. J Virol Methods
40. Brown AJ, Cleland A. Independent evolution of the env and pol genes of HIV-1 during zidovudine therapy. AIDS
41. Shi B, Kitchen C, Weiser B, et al. Evolution and recombination of genes encoding HIV-1 drug resistance and tropism during antiretroviral therapy. Virology
42. Mild M, Esbjörnsson J, Fenyö EM, et al. Frequent intrapatient recombination between human immunodeficiency virus type 1 R5 and X4 envelopes: implications for coreceptor switch. J Virol
43. van Rij RP, Worobey M, Visser JA, et al. Evolution of R5 and X4 human immunodeficiency virus type 1 gag sequences in vivo: evidence for recombination. Virology
44. Leitner T, Escanilla D, Marquina S, et al. Biological and molecular characterization of subtype D, G, and A/D recombinant HIV-1 transmissions in Sweden. Virology
45. Sabino EC, Shpaer EG, Morgado MG, et al. Identification of human immunodeficiency virus type 1 envelope genes recombinant between subtypes B and F in two epidemiologically linked individuals from Brazil. J Virol
46. Charpentier C, Nora T, Tenaillon O, et al. Extensive recombination among human immunodeficiency virus type 1 quasispecies makes an important contribution to viral diversity in individual patients. J Virol
47. Buzón MJ, Dalmau J, Puertas MC, et al. The HIV-1 integrase genotype strongly predicts raltegravir susceptibility but not viral fitness of primary virus isolates. AIDS
48. Martinez-Picado J, Martínez MA. HIV-1 reverse transcriptase inhibitor resistance mutations and fitness: a view from the clinic and ex vivo. Virus Res