Similarities between Human Immunodeficiency Virus Type 1 and Hepatitis C Virus Genetic and Phenotypic Protease Quasispecies Diversity

ABSTRACT

Human immunodeficiency virus type 1 (HIV-1) and hepatitis C virus (HCV) are two highly variable RNA viruses that cause chronic infections in humans. Although HCV likely preceded the AIDS epidemic by some decades, the global spread of both viruses is a relatively recent event. Nevertheless, HCV global diversity is higher than that of HIV-1. To identify differences in mutant diversity, we compared the HIV-1 protease and HCV NS3 protease quasispecies. Three protease gene quasispecies samples per virus, isolated from a total of six infected patients, were genetically and phenotypically analyzed at high resolution (HIV-1, 308 individual clones; HCV, 299 clones). Single-nucleotide variant frequency did not differ between quasispecies from the two viruses (HIV-1, 2.4 × 10⁻³ ± 0.4 × 10⁻³; HCV, 2.1 × 10⁻³ ± 0.5 × 10⁻³) (P = 0.1680). The proportion of synonymous substitutions to potential synonymous sites was similar (3.667 ± 0.6667 and 2.183 ± 0.9048, respectively) (P = 0.2573), and Shannon's entropy values did not differ between HIV-1 and HCV (0.84 ± 0.02 and 0.83 ± 0.12, respectively) (P = 0.9408). Of note, 65% (HIV-1) and 67% (HCV) of the analyzed enzymes displayed detectable protease activity, suggesting that both proteases have a similar mutational robustness. In both viruses, there was a rugged protease enzymatic activity landscape characterized by a sharp peak, representing the master sequence, surrounded by a collection of diverse variants present at lower frequencies. These results indicate that nucleotide quasispecies diversification during chronic infection is not responsible for the higher worldwide genetic diversity observed in HCV.

IMPORTANCE HCV global diversity is higher than that of HIV-1. We asked whether HCV genetic diversification during infection is responsible for the higher worldwide genetic diversity observed in HCV. To this end, we analyzed and compared the genotype and enzymatic activities of HIV-1 and HCV protease quasispecies existing in infected individuals. Our results indicate that HIV-1 and HCV protease quasispecies have very similar genetic diversity and comparable rugged enzymatic activity landscapes. Therapy for HCV has expanded, with new therapeutic agents such as the direct-acting antivirals (DAAs). DAAs, which target HCV NS3 protease and other virus proteins, have improved cure rates. However, major questions remain to be elucidated regarding the virologic correlates of HCV eradication. The findings shown here may help our understanding of the different therapeutic responses observed during chronic HCV infection.

INTRODUCTION

The activities of human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) and hepatitis C virus (HCV) RNA polymerase are not subject to error-correcting mechanisms, and as with other RNA viruses, replication of these viruses is highly error prone (1). The in vitro error rate of HCV polymerase has been estimated at ∼10⁻³ mutations per site for transitions and about 100-fold lower rates for transversions (2). Slightly lower in vitro error rates of purified HIV-1 reverse transcriptase have been reported, ranging from 1.9 × 10⁻⁴ to 5.9 × 10⁻⁴ mutations per site (3, 4). Even lower HIV-1 mutation rates have been found using an ex vivo assay (3.4 × 10⁻⁵ mutations per site per cycle) (5). Of interest, a similar in vivo mutation rate has been reported recently for HCV (2.5 × 10⁻⁵ mutations per site per cycle) (6); however, HCV is highly variable, with multiple subtypes and a global diversity higher than that of HIV-1 (7).

A possible explanation for the higher diversity of HCV might be its earlier global spread, which HIV-1 has followed; however, in contrast to the origins of HIV-1, which are well established, those of HCV are much less clear (8). The extensive genetic heterogeneity of HCV in sub-Saharan Africa strongly suggests that this virus was endemic to this geographical area long before its global spread in the last 100 to 200 years. Recently, HCV diversity has been hypothesized to depend on other parameters of the HCV life cycle, such as the long-lived nature of infected cells compared to that of cells infected with HIV-1, the existence of multiple replication complexes within an infected cell that may allow for the accumulation of diversity, and the turnover rate of both of these replication complexes and infected cells (6).

The error-prone nature of HIV-1 and HCV replication results in circulation of these viruses in infected individuals as a complex mutant spectrum of genetically related variants known as quasispecies (9, 10). Quasispecies diversity challenges host immune virus surveillance, antiviral therapies, and the development of effective vaccines (1). Of note, the genetic structure of viral quasispecies determines virus adaptability (11) and pathogenesis (12–14).

We hypothesized that an error-prone polymerase or virus life cycle would affect quasispecies diversity. To address this hypothesis, we analyzed and compared, at high resolution, the genotype and enzymatic activities of three HIV-1 proteases and three HCV NS3 proteases previously isolated from six infected individuals (15, 16). Our results indicate that HIV-1 and HCV protease quasispecies have very similar genetic diversity and comparably rugged enzymatic activity landscapes.

MATERIALS AND METHODS

Three subtype B HIV-1-infected individuals and three genotype 1b HCV-infected individuals were chosen for this study (15, 16). Two HIV-1 individual samples (Fig. 1A and B), with HIV-1 viral loads of 17,094 and 150,662 RNA copies/ml blood, respectively, and CD4⁺ T cell counts of 471 and 607 cells/μl blood, respectively, were asymptomatic, while one sample (designated C in Fig. 1) was isolated from an individual with an advanced stage of the disease, a CD4 T cell count below 200 cells/μl (10 cells/μl), and a high viral load (297,000 RNA copies/ml). The three HIV-1-infected individuals were naive for antiretroviral therapy. Two HCV individuals (designated A and B in Fig. 1) were coinfected with HIV-1, whereas the third subject (C in Fig. 1) was HCV monoinfected. HCV viral loads for these three samples were 2,364, 109,142, and 1,300,000 IU/ml, respectively. Alanine aminotransferase levels were 13, 37, and 862 U/liter, respectively. HCV-coinfected individuals A and B were under antiretroviral treatment and had undetectable HIV-1 RNA. CD4⁺ T cell counts for these two coinfected patients were 571 and 992 cells/μl, respectively. After the samples used in this study were taken, individuals A and B were treated for 48 weeks with pegylated alpha 2b-interferon in combination with ribavirin. Individual A showed a sustained response to therapy, whereas individual B did not respond to therapy. Individual C's sample was taken during acute infection; afterward, this individual's HCV infection spontaneously resolved. For the six study patients, time from primary infection was unknown; nevertheless, no patient was acutely infected or showed symptoms of disease progression (i.e., AIDS or liver fibrosis).

FIG 1.

Neighbor-joining phylogram of HIV-1 protease and HCV NS3 protease sequences from subtype B HIV-1 samples A, B, and C and genotype 1b HCV samples A, B, and C. Phylogenetic reconstruction was generated using a p-distance model as implemented in the MEGA6 software package. The three HIV-1 quasispecies were genetically more related to each other than were the three HCV samples.

The HIV-1 protease gene was amplified by PCR from proviral peripheral blood mononuclear cell DNA as previously described (15). The PCR products were digested with EcoRI and XhoI, ligated into a Uni-ZAP XR vector kit (Stratagene), and packaged with a Uni-ZAP XR Gigapack cloning kit (Stratagene) as previously described (15). To ensure that multiple proviral HIV-1 protease templates were present in each analyzed quasispecies, for each sample, 40 different PCR amplifications were performed and pooled before cloning. At least 10 HIV-1 proviral DNA copies were used as a starting template in the first PCR amplification. Twenty-five picograms of an HIV-1 HXB2 plasmid control was amplified and cloned as described above, and 14 subclones were sequenced to estimate diversity due to PCR Taq errors. The different HIV-1 proteases were sequenced with the flanking oligonucleotides T3 (5′-AATTAACCCTCACTAAAGGG-3′) and T7 (5′-TCGAGGTCGACGGTATC-3′) using the ABI PRISM dRhodamine terminator cycle sequencing kit (Applied Biosystems). HCV RNA extraction and amplification were performed as previously described (16). The PCR products were digested with EcoRI and XhoI and ligated to pBSK⁻ (Stratagene) to generate a ß-gal-HCV NS3 protease fusion protein as previously described (16). To ensure that multiple HCV NS3/4 protease templates were present in each analyzed quasispecies, for each sample, 40 different PCR amplifications were performed and pooled before cloning. Endpoint RNA limiting dilution was performed before RT-PCR to ensure that an excess of template HCV RNA was present in the PCR amplification mixtures. At least 25 RNA molecules were added in each of the PCR amplifications. The different proteases were sequenced with the flanking oligonucleotides T3 and T7. For phylogenetic analysis and genetic distance calculations, the MEGA6 software package was used (17). To determine possible selective pressures, the proportion of synonymous substitutions per potential synonymous sites (dS) and the proportion of nonsynonymous substitutions per potential nonsynonymous sites (dN) were calculated with the MEGA6 software package. The dS/dN ratio, used to infer selection pressure in protein coding genes, also was calculated.

The catalytic efficiencies of the different HIV-1 and HCV proteases were determined using a bacteriophage lambda-based genetic screening, as we have previously described (15, 16, 18–20). This genetic screen relies on the phage lambda regulatory circuit in which the viral repressor cI is specifically cleaved to initiate the lysogenic-to-lytic switch. The introduction of an HIV-1 protease or an HCV NS3 protease in a wild-type phage will cleave a mutant cI repressor containing a specific HIV-1 protease cleavage site (MA/CA) or HCV NS3 protease cleavage site (NS5A/NS5B), allowing the phage to go into the lytic replication cycle. As we have previously demonstrated, cI repressor cleavage is directly proportional to the protease catalytic efficiency (15, 16, 18–20).

Statistical analyses with χ² and unpaired t tests were performed with GraphPad Prism version 4.00 for Windows.

Nucleotide sequence accession numbers.

The sequences reported in this study have been deposited in the GenBank database (accession no. DQ193605 to DQ193912 for HIV-1 and EF013788 to EF014086 for HCV).

RESULTS

We included in this study individual clones carrying the HIV-1 protease or the HCV NS3 protease coding region generated from a single-time-point-plasma sample from three HIV-1- and three HCV-infected individuals, as we have previously described (15, 16). We analyzed 105, 103, and 100 individual clones from the three HIV-1 samples and 99, 97, and 103 clones from the three HCV samples. Neighbor-joining phylogenetic reconstructions of all HIV-1 protease and HCV NS3 protease nucleotide sequences were performed to determine the evolutionary relationships of the different variants (Fig. 1). As expected, sequences from each individual produced a monophyletic group. Visual inspection of the former trees showed that the three HIV-1 quasispecies were genetically more related to one another than were the three HCV samples, suggesting a more recent common ancestor for HIV-1 quasispecies than for the HCV quasispecies. Indeed, mean pairwise HIV-1 intersample p distance was 0.0477 ± 0.0000 (means ± standard errors), a value significantly lower than that observed among the HCV samples of 0.1019 ± 0.0000 (P < 0.0001 by unpaired t test).

Single-nucleotide variant (SNV) frequencies were 2.4 × 10⁻³ ± 0.4 × 10⁻³ and 2.1 × 10⁻³ ± 0.5 × 10⁻³ mutations per nucleotide (means ± standard errors) for HIV-1 and HCV, respectively (Table 1), and the they did not differ significantly (χ² = 1.900, df = 1, P = 0.1680). A higher number of nonsynonymous SNVs was observed in the HIV-1 proteases, but this difference was not significant (frequency of 1.1 × 10⁻³ ± 0.5 × 10⁻³ and 0.9 × 10⁻⁴ ± 0.3 × 10⁻⁴ mutations per nucleotide, respectively; χ² = 2.407, df = 1, P = 0.1208) (Table 1). No differences were found in intrasample mean pairwise nucleotide diversity (0.0157 ± 0.0014 and 0.0087 ± 0.0028, respectively; P = 0.0938 by unpaired t test), nucleotide quasispecies for Shannon's entropy (0.84 ± 0.02 and 0.83 ± 0.12, respectively; P = 0.9408 by unpaired t test), or between the proportion of synonymous substitutions to potential synonymous sites (dS) (3.667 ± 0.6667 and 2.183 ± 0.9048 for HIV-1 and HCV, respectively; P = 0.2573 by unpaired t test). These high Shannon's entropy values are indicative of an absence of PCR resampling in the study data set. To determine the contribution of Taq polymerase errors to the level of genetic diversity, a nested PCR was initiated with an HIV-1 protease plasmid, and 14 subcloned lambda phage clones were sequenced. Only two clones were mutated, one with a synonymous U-to-C substitution and another one with a nonsynonymous A-to-G substitution (15). This result suggested that few PCR errors were generated under our experimental conditions. Nevertheless, it cannot be dismissed that some of the substitutions analyzed here were generated during PCR amplification. A significantly lower proportion of nonsynonymous substitutions to potential nonsynonymous sites (dN) was observed in HCV (0.9800 ± 0.0152 and 0.3467 ± 0.0982; P = 0.0031), but no differences were found when the dS/dN ratios were compared (5.717 ± 1.559 and 7.433 ± 4.333, respectively; P = 0.7282).

TABLE 1.

Summary of population metrics

Protease	No. of bases sequenced	No. of mutations detected	No. of mutations per gene	No. of SNVsa	No. of nonsynonymous SNVs	Snb
HIV-1	9.2 × 104	782	2.5	218	104	0.84
HCV	1.5 × 105	700	2.3	311	137	0.83

^aSingle-nucleotide variants.

^bShannon's entropy.

The six study patients were chronically infected and in the asymptomatic phase of the disease. However, the time since acute infection was unknown. During HIV-1 or HCV infection, the level of virus diversity may change over time, specifically during acute infection or late stages of the disease (21–23). In order to control for this variable, the study sample diversity was compared to that of other sample data sets that we previously analyzed using the same methodology. Thus, the HIV-1 protease study samples were compared to 10 samples from naive chronically infected HIV-1 patients (24). No differences were found when the mean pairwise nucleotide diversity (0.0157 ± 0.0014 and 0.0143 ± 0.0022, respectively; P = 0.7544 by unpaired t test), Shannon's entropy (0.84 ± 0.02 and 0.86 ± 0.05, respectively; P = 0.7851 by unpaired t test), or dS were compared (3.667 ± 0.6667 and 3.840 ± 0.6327, respectively; P = 0.8909 by unpaired t test). Similarly, when the HCV NS3 protease study samples were compared with 56 samples from naive chronically infected HCV genotype 1 patients that we previously analyzed (18), again, no differences were found when the mean pairwise nucleotide diversity (0.0087 ± 0.0028 and 0.0121 ± 0.0008, respectively; P = 0.3362 by unpaired t test), Shannon's entropy (0.83 ± 0.12 and 0.87 ± 0.027, respectively; P = 0.4308 by unpaired t test), or dS (2.183 ± 0.9048 and 3.200 ± 0.5600, respectively, P = 0.1926 by unpaired t test) were compared. Importantly, when HCV NS3 samples were compared with a data set of 18 samples from acutely HCV genotype 1-infected patients (25), significant differences were found in the mean pairwise nucleotide diversity (0.0087 ± 0.0028 and 0.0024 ± 0.0007, respectively; P = 0.0060 by unpaired t test) or Shannon's entropy (0.83 ± 0.12 and 0.35 ± 0.049, respectively; P = 0.0014 by unpaired t test). These results strongly suggested that no biases, in terms of virus diversity, were introduced in the study sample data set.

The average number of mutations per gene was very similar for HIV-1 and HCV at 2.5 ± 0.2 and 2.3 ± 0.8, respectively (P = 0.8094, unpaired t test), as was the distribution of mutations per gene (Fig. 2). Remarkably, some hyper-edited sequences with 10 and 11 mutations also were found in both viruses (Fig. 2). Again, comparable frequencies of transitions and transversions were detected in HIV-1 and HCV quasispecies, with transitions averaging 2.1 × 10⁻³ and 1.8 × 10⁻³ substitutions per site and transversions averaging 3.2 × 10⁻⁴ and 3.3 × 10⁻⁴ substitutions per site, respectively (χ² = 2.308, df = 1, P = 0.1287 for transitions; χ² = 0.003388, df = 1, P = 0.9536 for transversions). Nevertheless, some differences were found in nucleotide-specific changes (Fig. 3). A significantly higher number of A-to-G and G-to-A transitions was found in HIV-1 compared with the number of U-to-C and C-to-U transitions (χ² = 16.66, df = 1, P < 0.0001), whereas a significantly higher number of U-to-C and C-to-U transitions was found in HCV (χ² = 67.27, df = 1, P = 0.0095). Differences in the frequencies of the different transversions also were detected between HIV-1 and HCV (Fig. 3). These nucleotide-specific differences in substitution frequency most likely reflect the different molecular mechanisms of HIV-1 and HCV polymerase fidelity. The results of these analyses of mutation frequencies are consistent with biochemical observations (2, 4) and in vivo HIV-1 and HCV mutational patterns (18, 24).

FIG 2.

HIV-1 protease and HCV NS3 protease population structure. The histograms display the proportion of genomes of each virus protease containing the given number of nucleotide mutations (Hamming distance from the respective quasispecies master sequence). The distribution of mutations per gene was very similar for both proteases.

FIG 3.

Frequency of the different types of nucleotide substitutions found in the HIV-1 protease and HCV NS3 protease sequences. Mutations were relative to the corresponding quasispecies master sequence. Differences in the frequencies of the different nucleotide substitutions were observed within HIV-1 and HCV proteases.

We next compared the relative enzymatic activity of the different identified single mutant proteases (21, 42, and 42 out of 105, 103, and 100 sequences for the three HIV-1 quasispecies, respectively, and 33, 47, and 32 out of 99, 97, and 103 sequences for the three HCV quasispecies, respectively). The enzymatic activities of the different variant proteases of the HIV-1 and HCV quasispecies analyzed in this study were related to the activity of the HIV-1 HXB2 protease and HCV I389 NS3 protease (100%), respectively. The enzymatic activities of the three master (most abundant) sequences were 75% ± 29% and 97% ± 59% for the HIV-1 and HCV quasispecies, respectively (means ± standard deviations), and the amino acid master sequences represented 41% ± 15% and 56% ± 11% of the total number of the HIV-1 and HCV sequences, respectively. The second most represented sequence form of the HIV-1 and HCV quasispecies had catalytic efficiencies of 43% ± 24% and 63% ± 46%, respectively, and corresponded to 14% ± 11% and 4% ± 2% of the sequences.

When enzymatic activities of individual protease clones were compared to the activity of the corresponding master sequence, HIV-1 and HCV displayed a very comparable protease enzymatic activity landscape (Fig. 4). Within both virus quasispecies there were proteases with undetectable enzymatic activity (lethal). The percentage of lethal (less than 1% of the activity) HIV-1 or HCV protease clones was 35% ± 20% and 33% ± 5%, respectively (12% ± 6% and 13% ± 2%, respectively, if we considered the all-study population size instead of the different identified single mutant proteases). Therefore, 65% and 67% of all identified HIV-1 and HCV single mutant proteases analyzed, respectively, displayed detectable enzymatic activity. Of note, 11% ± 12% and 29% ± 18% of the protease variants from the HIV-1 and HCV quasispecies, respectively, showed an equal or higher catalytic efficiency than the master protease. Most of the HIV-1 or HCV high-activity minority variants had only one substitution compared to the master sequence. This significant number of beneficial mutations observed either in the HIV-1 or HCV quasispecies indicates the presence of several adaptive peaks and a highly dynamic population structure in which selection for minor genetic components may drive the population to new regions of sequence space.

FIG 4.

Enzymatic activity distribution of quasispecies individual amino acid clones. Enzymatic activity values were relative to the corresponding quasispecies master sequence. HIV-1 and HCV displayed a very comparable protease activity landscape.

DISCUSSION

Prior work has extensively documented that HIV-1 and HCV, like other RNA viruses, generate genetically complex mutant populations known as viral quasispecies (26). The diversity and genetic structure of HIV-1 and HCV quasispecies determine the rapid adaptation and spread of these two viruses (8, 27). RNA virus quasispecies diversity also has been associated with virus pathogenesis (12). Indeed, HCV quasispecies diversification during the acute phase of hepatitis C predicts whether the infection will resolve or become chronic (28). In HIV-1 infection, a high rate of CD4 T cell loss has been associated with relative evolutionary stasis of the quasispecies virus population (29). However, no direct comparison of the HIV-1 and HCV intrapatient diversification has been reported.

HIV-1 and HCV share some characteristics. Both are RNA viruses, and both produce persistent infections in humans. The higher global diversity observed in HCV has given rise to the question of whether this virus has a higher rate of genetic diversification during chronic infection. In this study, we compared the genotype and enzymatic activities of HIV-1 and HCV protease quasispecies isolated from infected individuals (15, 16), using the same experimental approach to obtain and analyze both HIV-1 and HCV quasispecies.

We found that HIV-1 and HCV proteases have similar nucleotide diversity and heterogeneity and detected the same frequency of total and nonsynonymous SNVs in HIV-1 and HCV quasispecies. Similar Shannon's entropy values also were found. Of note, no differences were observed when the proportion of synonymous substitutions to potential synonymous sites (dS) was compared. We also found a significantly higher proportion of nonsynonymous substitutions to potential nonsynonymous sites (dN) in HIV-1. This result is not surprising, because the accumulation of nonsynonymous diversity depends not only on mutation rate but also on the particular selective forces acting on a particular protein, virus, or host (e.g., immune response, cell-to-cell transmission, and host cell restriction factors). The most widely used method to infer selection pressure in protein-coding genes calculates the evolutionary rate dS/dN ratio. This metric indicates how quickly a protein's constituent amino acids change relative to synonymous changes, and a dS/dN ratio greater than 1 implies purifying (stabilizing) selection. The dS/dN values obtained with HIV-1 and HCV clearly demonstrate that in both viruses, the protease-coding region is under purifying selection, indicating that the observed mutations are not the result of positive Darwinian selection (dS/dN < 1) but rather of virus life history factors (e.g., polymerase mutation rate and replication rate). Our results strongly suggest that the higher worldwide genetic diversity observed in HCV and corroborated here by the significantly higher HCV intrasample p distances is not attributable to quasispecies diversification during chronic infection.

Using a comparable methodology, a previous study found a proviral DNA HIV-1 protease diversity similar to that found in this report (24) (0.0157 ± 0.0014 and 0.0143 ± 0.0022, respectively). In contrast, deep sequencing, using a primer identifier (ID) to minimize PCR errors, of two HIV-1 protease naive samples from the same patient displayed lower diversity (intrasample mean pairwise nucleotide diversities of 0.0080 and 0.0079, respectively) (30). In this study, viral RNA, instead of proviral DNA, was analyzed. Previous studies performed with the HCV NS3 protease of genotype 1 also showed an equivalent intrasample mean pairwise nucleotide diversity, ranging from 0.0133 to 0.0174 in HCV monoinfected patients (31) (32) and between 0.0075 and 0.0120 in HCV/HIV-1-coinfected patients (18, 31, 33, 34). A more recent study analyzed, by deep sequencing and primer ID, the HCV NS3 protease of genotype 1 from 20 HCV monoinfected patients and 20 HCV/HIV-1 coinfected patients (35). A lower intrasample mean pairwise nucleotide diversity was observed in HCV/HIV-1-coinfected patients (0.0095 ± 0051) than in HCV monoinfected patients (0.0143 ± 0.0069). Remarkably, the diversity found in the HCV monoinfected patient data set (0.0143 ± 0.0069) is very close to the one found in our HIV-1 protease data set (0.0157 ± 0.0014) and to that from a previous HIV-1 protease diversity study (0.0143 ± 0.0022) (24). This large study supports the nucleotide diversity similarity between HIV-1 and HCV protease quasispecies.

Different rules govern global virus diversity versus intrahost virus diversity. Although a high mutation rate provides viruses with the potential for rapid intrahost diversification, the virus must adapt to the local environment within a host during the course of infection, particularly for chronically infecting viruses, without losing both its replication capability and its transmissibility (36). Maintenance of a functional requirement for transmission may generate a selective constraint on interhost sequence changes (23, 37, 38). Due to the proviral stage in vivo, persistence of a retrovirus such as HIV-1 differs in important aspects from persistence of a nonintegrating virus such as HCV. These differences may affect the different global diversifications of these two viruses. The possibility of latency, during which the integrated genome displays the stasis typical of the host DNA, cannot be discarded. In contrast, persistence of HCV implies continuous genome replication with the introduction of mutations. Recent work with hepatitis B virus (HVB) has found that the evolutionary rate of HBV between hosts was considerably lower than that within hosts (39). HIV-1 mutations that are adaptive in one individual may be maladaptive in another owing to different genetic backgrounds, such as human leukocyte antigen (HLA), and would revert after transmission (40, 41). As a result, intrahost substitution rates would be faster than that of interhost substitution, because not all of the mutations accumulating within hosts were maintained after transmission (36).

Most notably, this study is the first, to our knowledge, to investigate and compare the structure of the HIV-1 and HCV quasispecies protease enzymatic activity landscapes. Several studies have analyzed the HIV-1 and HCV genetic structure and the evolution of different viral genes in detail; nevertheless, data regarding the phenotypic structure of HIV-1 and HCV quasispecies are scarce. Our results show compelling evidence for the similarity between the enzymatic activity landscapes of HIV-1 and HCV proteases, although different selective pressures may be operating in HIV-1 or HCV infection. Distinct selective forces may determine the spectrum and relative activity of the observed mutations, but we identified some common traits between these two proteases. The proportion of lethal protease genes is of the order of 10% to 15% in both HIV-1 and HCV. Although there are no equivalent data for HCV, a very similar rate of defective genomes has been found in the HIV-1 tat gene (9) or gag or env quasispecies (42). PCR recombination may affect the functionality of the studied proteases (43, 44). However, lethal or highly deleterious substitutions should have a similar impact on most of the different sequence backgrounds and affect similarly the different study samples. Every quasispecies displayed a large number of enzymatic activity optima or peaks, suggesting that in both protease quasispecies, complexity does not exclusively depend on the enzyme catalytic efficiency. Because the amino acid master sequence represented 41% and 56% (HIV-1 and HCV, respectively) of the total number of sequences analyzed, the structure of the HIV-1 and HCV protease quasispecies enzymatic activity landscapes is characterized by a sharp peak, representing the master sequence, surrounded by a collection of diverse variants present at lower frequencies. Both landscapes were rugged. Several single substitutions were lethal and led the master sequence to drop down the activity peak. Of note, at other positions, single substitutions sent the master sequence to a new local optimum (neutral or beneficial) peak. Interestingly, this finding was predicted in the first theoretical description of a quasispecies. The quasispecies concept had its origin in a theoretical formulation of molecular evolution that emphasized error-prone replication of simple RNA replicons as an essential feature of self-organization and adaptability of primitive life forms (45, 46).

Our study has some limitations that are worth noting. Although previous work showed that the patterns of increase in viral diversity (the breadth of population at a given time point) were similar in the HIV-1 DNA and RNA populations (22, 47), it cannot be discarded that some of the analyzed HIV-1 DNA sequences were archival, in contrast to HCV populations, which are replicating at all times. Other possible study limitations are the assumption that intrapopulation heterogeneity will be similar with HCV of different genotypes, the sample size, and the presence of HCV patients coinfected with HIV-1. Therefore, further work should include a larger sample size with more HCV monoinfected patients.

Our findings may have implications for understanding therapeutic responses in chronic HCV infection. Therapy for HCV has expanded with new therapeutic agents, such as the DAAs, which target HCV NS3 protease and other virus proteins and have improved cure rates. However, the virologic correlates of HCV eradication are incompletely understood. The similarities between HIV-1 and HCV proteases found here show that it is likely that similar to HIV-1, drug-resistant mutants will preexist therapy. Indeed, we and others have shown that this is the case for NS3 protease inhibitors (18, 48; reviewed in reference 49). Therefore, further work should include the study of other virus enzymes and proteins to evaluate whether the findings shown here can be extended to other virus genomic regions.