Evolution of HIV-1 in a patient population failing multiple-drug therapy

Authors

  • Shaolin Hong,

    1. Duke Comprehensive Cancer Center, Duke University Medical Center, Durham, North Carolina 27710, USA
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Jingjiang Cao,

    1. Department of Dermatology and Venereology, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022, China
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Ya-ting Tu

    1. Department of Dermatology and Venereology, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022, China
    Search for more papers by this author

Correspondence
Ya-ting Tu, Department of Dermatology and Venereology, Union Hospital, Tongji Medical College, Wuhan 430022, China.
Tel: +86 27 85733882; fax: +86 27 83693500; email: yatingtu@yahoo.com.cn

ABSTRACT

To understand the evolutionary pathway of the multi-drug-resistant virus HIV-1 under drug-induced selection pressure, plasma from seven patients from baseline to different intervals post-treatment failure were used in RT-PCR protocols. Multiple clones were sequenced for each time point. Drug-resistant mutations were detected in five patients. Phylogenetic analysis showed that at different time points, viral sequences clustered separately and formed independent lineages. Genetic diversity decreased from 1.59 to 0.55, whereas non-synonymous/synonymous mutation ratios increased from 0.067 to 0.118, respectively. These data suggest that the virus population changed dynamically and clustered in a time point-specific manner whereas genetic diversity decreased consistently.

List of Abbreviations: 
ABC

abacavir

AZT

zidovudine

HAART

highly active antiretroviral therapy

MDR

multiple-drug resistance

3TC

lamivudine

Failure of HAART is often accompanied by the emergence of drug resistance mutations (1, 2). Analysis of protease (PR) and reverse transcriptase (RT) genes from failed HAART patients shows genetic changes are the foundation of MDR (3, 4). Due to the high genetic diversity of HIV-1 sequences, variants harboring partial resistance to antiretroviral therapy will expedite HIV evolution during HAART (5, 6). Results describing virus diversity changes have been varied at best (7, 8). As such, it is imperative to thoroughly investigate the viral populations existing before treatment and after-treatment failure in order to understand how drug selection can actively shape this diversity and the relationship between genetic diversity and clinical outcome. Indeed, this pursuit may help further elaborate HIV population evolutionary pathways and assist in refining treatment paradigms to maximize anti-retrovirus therapy (ART) options.

In this retrospective study, we used clone sequencing to investigate the evolution of resistance in patients from a clinical cohort during 2001–2003 at Duke University. All patient materials were collected under approval of the Duke University Institutional Review Board. Treatment-naïve HIV-infected subjects were pre-stratified according to entry HIV RNA levels (< or > 100 000 copies/ml) for lamivudine (3TC, Epivir; GSK, USA), abacavir (ABC, Ziagen; GSK, USA) and zidovudine (AZT, Retrovir; GSK, RTP, NC, USA) therapy. From 291 subjects in this cohort, seven patient samples meeting the following criteria were evaluated in the present study: (i) All patients had an HIV-1 subtype B infection; (ii) The patient had begun treatment with RT inhibitor ABC for the first time and had been on continuous treatment with RT inhibitors over the time of evaluation; (iii) In the course of 24 weeks, patients had successfully decreased their viral load to a level of <400 copies/ml. However, after said 24 weeks of treatment, all subjects did not have their viral loads suppressed to an undetectable level. The mean viral load in plasma prior to treatment with RT inhibitors was 4.85 of log10 copies/ml (Table 1), and the mean CD4+-T-cell count was 267 cells/μl.

Table 1.  Drug-resistance mutations detected in reverse transcriptase sequences from seven patients treated with ABC, 3TC and AZT
PatientWeeks after treatmentViral load (log10 copies/ml)No. clonesDrug resistance mutations
P104.528 
963.446M41L(46), M184V(46), L210W(10), T215Y(46)
P205.435 
963.338M184V(38)
P304.540 
723.139M41L(27), M184V(39), T215Y(39)
P405.234 
323.145M184V(45)
963.949D67N(49), K70R(49), M184V(49), T215F(49), K219E(49)
P504.946M184V(1)
363.717 
883.846M184V(46)
P605.736 
285.648 
365.515 
P705.311 
443.610 
805.59 

To obtain baseline viral genetic population data, RT-PCR was used to amplify the partial pol gene, approximately 1.5 kb including all identified drug-resistance mutations in both PR and RT genes. To ensure the sequences represented the viral quasispecies population in vivo (9), five independent first-round amplification reactions were carried out using the PR5 outer primer (5′ TTCAATTGT GGCAAAGAAGGGCA 3′) and the RT3 outer primer (5′ TTGGCCTTGCCCCTGCTTCTGTA 3′). The second-round PCR used the PR5 inner primer (5′ CCAAAAATTGCAGGGCCCTAGGA 3′) and the RT3 inner primer (5′ CACTCCATGTAC CGGTTCTTTTAG 3′). The PCR products were cloned into pSTBlue-1 vectors by using a Perfectly Blunt Cloning kit (Novagen, Madison, WI, USA). Fifty clones from each patient (10 clones from each individual PCR amplification, i.e.) were sequenced.

The raw sequence data were imported into the Sequencher program (Gene Codes Corp., Ann Arbor, MI, USA) for complete and contiguous sequence assembling and editing. Amino acid sequences were deduced from nucleotide sequences and aligned together by using Clustal W and multiple aligned sequence editor (MASE). The genetic diversity and non-synonymous/synonymous rates (dN/dS) were calculated to evaluate the selection pressure on the viral population (10). The DNADIST program implemented in PHYLIP 3.5 was used to calculate genetic diversity for sequences from resistant viruses and assess their divergence from baseline viruses. These nucleotide sequence alignments were also used to study phylogenetic relationships and synonymous/non-synonymous ratios for the viral population in each patient. The topology and reliability of phylogenetic trees were evaluated with 1000 bootstrap replicates.

Nine to 49 clones were obtained from different time points for each of the seven patients. In comparison to the previously reported IAS-USA Mutations Figures (11), five patients, P1, P2, P3, P4 and P5, had single or multiple-drug-resistance mutations whereas patients P6 and P7 had no detectable mutations (Table 1). Reconstruction of all sequences revealed seven distinct phylogenetic lineages, or sequence subsets, of HIV-1. Needless to say, this finding confirmed that no cross-contamination between samples during the PCR step and sequencing reaction took place (data not shown). To examine the phylogenetic relationship between the derived strains, we carried out individual exploratory tree analyses over multiple time points for all cases. Viral sequence clustering relationships are shown in Figure 1. Many distinctive clusters were accompanied by high bootstrap values (>80%), suggesting that these sequences were sufficiently different from each other to form independent lineages.

Figure 1.

Exploratory trees showing genetic relationships between HIV clones derived from different time points subjected to highly active antiretroviral therapy. Time points are specified by the symbol in each patient map. Asterisks at the nodes indicate that the percentage of the bootstrap value was above 80% with which the adjacent cluster is supported. Phylogenetic analysis showed that viral sequences at different time points clustered separately in patients P1, P2, P3, P4 and P5. For patient P5, a pre-existing minor population harbors a drug resistance mutation (circled) and developed into the dominant population.

Five patients with single or multiple-drug-resistance mutations developed clones clustered in a time point-specific manner. Viral grouping within the different lineages (time points) was consistent but segregated in respect to other populations, suggesting genetic differentiation and restructuring of HIV-1 populations through antiretroviral drug selection. For the last two patients, P6 and P7, no drug resistance mutation in the RT region was identified despite HAART treatment failure. Viral populations did not show cluster differences in the reconstructed tree. Although there was an observable tendency by populations to segregate at different time points, baseline clones still possessed the most diverse population. With continued antiretroviral drug inhibition, the populations in the first five patients showed increasingly tight clustering and, finally, the least genetic diversity at the endpoint.

The genotypic assay is capable of missing some minor populations of drug-resistance variants (12), especially in cases in which clinically important reductions were small or associated with impaired replication fitness (13). Alternatively, early minority populations have been shown to engage in drug resistance development to protease inhibitor according to several reports (7, 14). However, no genetic evidence has directly demonstrated the evolutionary pathway of these populations. Patient P5 in our study had one clone from 46 baseline clones bearing mutation M184V early on. At week 36, 17 clones were sequenced with no drug resistance mutation detected. At week 88 after therapy, 46 clones were sequenced with M184V mutation in all clones. Phylogenetic tree analysis revealed that the mutant clone was genetically distant from the majority of baseline populations, but close to the same populations after therapy (Fig. 1). The clone was so genetically similar to post-treatment viruses that, effectively, it could not be separated from later populations phylogenetically. This finding provided the phylogenetic evidence that the final drug-resistant population was evolutionarily derived from a minor population at baseline due to strong selection by multiple drugs.

As the RT region had been shown to possess a high degree of variation and be implicated in playing a role in drug resistance, an HIV quasispecies was analyzed by assessing the genetic diversity in this region. We found that genetic diversity varied at different time points, from patient to patient (Table 2). For example, five patients had higher genetic variation at baseline (0.78–2.15, mean 1.59 ± 0.53) compared to post-treatment values (0.34–0.82, mean 0.55 ± 0.22). This indicates that genetic diversity can dramatically change over time within the same patient. It also suggests that these patients had significant greater quasispecies variation at the baseline than the virus population at the endpoint after therapy. The other two patients, P6 and P7, harboring no drug resistance mutations, showed very homogeneous viral populations, with genetic diversity values from 2.19 to 1.93 and from 1.05 to 1.11, respectively.

Table 2.  Genetic diversity and non-synonymous/synonymous mutation ratios among plasma HIV-1 populations in different time points of seven patients
PatientWeeks after treatmentGenetic diversitydN/dS
P1 02.150.032
960.820.102
P2 01.390.084
960.440.153
P3 00.780.152
720.390.229
 01.880.064
P4321.110.091
960.340.140
P5 01.770.063
361.170.032
880.750.077
P6 02.190.071
282.210.055
361.930.092
P7 01.050.179
441.160.122
801.110.081

The dN/dS ratio compares non-synonymous to synonymous substitutions in order to study the kinetics of evolution. A dN/dS value equal to 1 indicates neutrality; purifying selection occurs when a dN/dS ratio is substantially less than 1. Of seven patients, all sequences had a dN/dS ratio of less than 1, indicating purifying selection in this region (Table 2). However, as the patients became exposed to antiretroviral drugs, the dN/dS ratio increased significantly in six subjects (mean value from 0.076 ± 0.037 to 0.124 ± 0.055). The increment showed positive selection for amino acid changes under drug pressure except in patient P7, which again did not exhibit any drug-resistant mutation.

On average, the genetic diversity was decreased from 1.59 to 0.55 in treatment-failure viral populations, whereas non-synonymous/synonymous mutation ratios increased from 0.067 to 0.118, respectively.

Genetic diversity has been considered to help virus quasispecies escape selective pressure (15, 16). Some studies have reported that new drug-resistant viruses were associated with a strong but transient reduction in genetic diversity that later became more divergent (8). Other research, however, indicated that viral populations were compartmentalized and reconstituted by turnover from new viral founders within alternative cellular sites (17, 18). Increased diversity resulted from recombination, either through restricting evolutionary bottleneck effects or through the generation of novel genotypes (7). We observed that two patients, P4 and P5, had exhibited divergent viral populations during the middle therapy time point. These dynamic shifts were consistent with baseline and endpoint populations. The scenario promotes the concept of a persistent viral evolutionary mechanism producing distinct viral lineages instead of breaking down baseline viral populations.

It had been shown that viral populations from different time points could not easily be separated through phylogenetic tree analysis under conditions of natural infection (19). However, within well-defined selection pressure, we found certain HIV subclusters were supported with high bootstrap values and separated from others by long branches. The divergence of virus population decreased in accordance. Genetic diversity changes in a viral population at different time points can be said to be an indication of the selection pressure on the viral population. As MDR viruses may occur from pre-existing minor species or directly correlated with distinct viral lineages that persistently evolve and accumulate over time, patients should be monitored by resistance testing and their treatment adjusted as soon as viruses begin to diversify genetically.

Ancillary