Intra-Host Evolutionary Rates in HIV-1C env and gag during Primary Infection

Abstract

Background

HIV-1 nucleotide substitution rates are central for understanding the evolution of HIV-1. Their accurate estimation is critical for analysis of viral dynamics, identification of divergence time of HIV variants, inference of HIV transmission clusters, and modeling of viral evolution.

Methods

Intra-patient nucleotide substitution rates in HIV-1C gag and env gp120 V1C5 were analyzed in a longitudinal cohort of 32 individuals infected with a single viral variant. Viral quasispecies were derived by single genome amplification/sequencing from serially sampled blood specimens collected at median (IQR) of 5 (4–6) times per subject from enrollment (during Fiebig stages II to V) over a median (IQR) of 417 (351–471) days post-seroconversion (p/s). HIV-1C evolutionary rates were estimated by BEAST v.1.6.1 using a relaxed lognormal molecular clock model. The effect of antiretroviral therapy (ART) on substitution rates in gag and env was assessed in a subset of six individuals who started ARV therapy during the follow-up period.

Results

During primary HIV-1C infection, the intra-patient substitution rates were estimated at a median (IQR) of 5.22E-03 (3.28E-03–7.55E-03) substitutions per site per year of infection within gag, and 1.58E-02 (9.99E-03–2.04E-02) substitutions per site per year within env gp120 V1C5. The substitution rates in env gp120 V1C5 were higher than in gag (p<0.001, Wilcoxon signed rank test). The median (IQR) relative rates of evolution at codon positions 1, 2, and 3 were 0.73 (0.48–0.84), 0.67 (0.52–0.86), and 1.54 (1.21–1.71) in gag, and 1.01 (0.86–1.15), 1.05 (0.99–1.21), and 0.86 (0.67–0.94) in env gp120 V1C5, respectively. A first to the third position codon rate ratio > 1.0 within env was found in 25 (78.1%) cases, but only in 4 (12.5%) cases in gag, while a second to the third position codon rate ratio > 1.0 in env was observed in 26 (81.3%) cases, but in gag only in 2 (6.3%) cases (p<0.001 for both comparisons, Fisher’s exact test). No ART effect on substitution rates in gag and env was found, at least within the first 3–4 months after ART initiation. Individuals with early viral set point ≥ 4.0 log₁₀ copies/ml had higher substitution rates in env gp120 V1C5 (median (IQR) 1.88E-02 (1.54E-02–2.46E-02) vs. 1.04E (7.24E-03–1.55E-02) substitutions per site per year; p=0.017, Mann-Whitney sum rank test), while individuals with early viral set point ≥ 3.0 log₁₀ copies/ml had higher substitution rates in gag (median (IQR) 5.66E-03 (3.45E-03–7.94E-03) vs. 1.78E-03 (4.57E-04–5.15E-03); p=0.028; Mann-Whitney sum rank test).

Conclusions

The results suggest that in primary HIV-1C infection, (1) intra-host evolutionary rates in env gp120 V1C5 are about 3-fold higher than in gag; (2) selection pressure in env is more frequent than in gag; (3) initiation of ART does not change substitution rates in HIV-1C env or gag, at least within the first 3–4 months after starting ART; and (4) intra-host evolutionary rates in gag and env gp120 V1C5 are higher in individuals with elevated levels of early viral set point.

1. INTRODUCTION

HIV-1 is one of the fastest evolving RNA viruses. Multiple factors, such as the error-prone nature of viral reverse transcriptase, high replication rate, high population size within host, compartmentalization, frequent recombination, and superinfection, play important roles in HIV-1 evolutionary dynamics (Coffin et al., 1997; Perelson et al., 1996; Preston et al., 1988; Rambaut et al., 2004; Suzuki et al., 2000). Evolutionary rates, or rates of nucleotide substitution, represent the cumulative effects of virus mutation rate, generation time, effective population size and fitness, and are central for understanding the evolution of HIV-1 (Duffy et al., 2008; Jenkins et al., 2002). An accurate estimation of HIV-1 evolutionary rates is critical for analysis of viral dynamics, identification of divergence time of HIV variants, inference of HIV transmission clusters, and modeling of viral evolution. The HIV-1 evolutionary rates within the host are associated with disease progression (Edo-Matas et al., 2011; Lemey et al., 2007; Williamson, 2003; Wolinsky et al., 1996), which highlights the importance of a systematic assessment of intra-host evolutionary rates in HIV-1 infection.

The nucleotide substitution rate is defined as the number of nucleotide substitutions per site per year. Previous studies estimated the rate of nucleotide substitution in HIV-1 (inter-patient level) at about 1.0×10⁻³ per site per year (Duffy et al., 2008; Gojobori et al., 1990; Goudsmit and Lukashov, 1999; Korber et al., 2000; Korber et al., 1998; Leitner and Albert, 1999; Li et al., 1988; Salemi et al., 2001; Suzuki et al., 2000; Yusim et al., 2001), and pointed to different substitution rates among HIV-1 genes (Korber et al., 2000; Leitner and Albert, 1999; Li et al., 1988; Salemi et al., 2001). Using the maximum likelihood method, substitution rates in partial gag and env of HIV-1 were estimated at 2.5×10⁻³ per site per year (Jenkins et al., 2002). Applying a Bayesian framework and hierarchical models of phylogenetic analysis, intra-host substitution rates in HIV-1 env were estimated at 9.2 ×10⁻³ per site per year among disease progressors and 7.0×10⁻³ per site per year among long-term non-progressors (Edo-Matas et al., 2011). Analysis of synonymous and nonsynonymous rates using well-characterized datasets of prospectively followed individuals infected with HIV-1B (Shankarappa et al., 1999; Shriner et al., 2004) revealed intra-host evolutionary rates in env at 6.3×10⁻³ to 1.0×10⁻² per site per year (Lemey et al., 2007; Lemey et al., 2006; Pybus and Rambaut, 2009). The intra-host evolutionary rates depend on the stage of infection and are lower as disease progresses (Lee et al., 2008; Pybus and Rambaut, 2009).

Little is known about intra-host evolutionary rates in HIV-1 non-B subtypes, particularly in subtype C. Evolutionary rates for HIV-1C were reported at 9.7×10⁻³ per site per year (Maljkovic Berry et al., 2007). Inter-patient evolutionary rates in HIV-1C were estimated at 0.05–2.95×10⁻³ per site per year for gag and at 3.1–4.8×10⁻³ per site per year for env (Walker et al., 2005). Abecasis et al. obtained similar estimates of substitution rates for HIV-1C env, and about 3 times lower rates for HIV-1C pol (Abecasis et al., 2009).

In this study we assessed the in vivo intra-host substitution rates in HIV-1 subtype C gag and the V1C5 region of env gp120 during primary infection. Our sample set of prospectively collected HIV-1C quasispecies from 32 subjects can be compared to a comprehensive set of HIV-1B env sequences described by Shankarappa et al. (Shankarappa et al., 1999). While the follow-up period in our study was shorter (~400 days p/s vs. about 8–10 years), the sample size was larger (n=32 vs. n=8). We assessed levels and distribution of intra-host substitution rates in primary HIV-1C infection, compared rates between HIV-1C gag and env, evaluated the relative rates of HIV-1C evolution across codon positions, estimated the effect of antiretroviral therapy (ART) on viral substitution rates, and addressed potential associations between viral evolutionary rates and HIV-1 RNA load in the early phase of HIV-1C infection.

2. MATERIAL AND METHODS

2.1

Study subjects were enrolled in the primary HIV-1 subtype C infection cohort in Botswana (Novitsky et al., 2009b; Novitsky et al., 2009c; Novitsky et al., 2008) from April 2004 to April 2008. The study was approved by Institutional Review Boards in Botswana and the US. Written informed consent was obtained from each participant. The entire cohort was comprised of 42 individuals with estimated time of seroconversion (Novitsky et al., 2009a; Novitsky et al., 2011b; Novitsky et al., 2009b; Novitsky et al., 2010b; Novitsky et al., 2009c). However, 10 of those individuals were infected with multiple viral variants (Novitsky et al., 2011c). Therefore, only a subset of 32 individuals who were infected with unrelated viral variants with limited diversity of viral quasispecies at the time point of sampling closest to the estimated time of seroconversion (HIV infection with a single viral variant) were analyzed in this study, including 6 subjects identified with acute HIV-1 infection (Fiebig stage II) and 26 subjects identified soon after seroconversion within Fiebig stage IV or V (Table 1). Time “zero” corresponded to seroconversion, as a more accurate, measurable, and less subjective time point than estimation of the time of HIV infection (Novitsky et al., 2010b). There were 5 male (1 acute) and 27 female (5 acute) participants. The median age at enrollment was 28 years (IQR 25–34, range 20–56). All subjects were Botswana nationals, and all infections were HIV-1 subtype C (Novitsky et al., 2009a; Novitsky et al., 2009c). Serial measurements of HIV-1 RNA and CD4 T cell counts were performed over time (Novitsky et al., 2011a; Novitsky et al., 2009c). The early viral set point was estimated as a mean value of HIV-1 RNA measurements in plasma during the period 100 to 300 days post-seroconversion (p/s) (Novitsky et al., 2011a; Novitsky et al., 2010a). The median (IQR) follow-up period was 417 (351–471) days post-seroconversion in env, and 418 (350–471) days p/s in gag. Six of 32 (19%) subjects initiated ART within the observed period of time due to a drop in CD4+ T cells.

Table 1.

Study subjects. Age is shown at enrollment. HIV-1 RNA load is shown as mean for measurements over the period from 100 to 300 days post-seroconversion (p/s). Time of ART initiation is presented only for subjects who started ART within the follow-up period (subjects who started ART at later time points are not shown).

Patient code	Gender	Age	HIV-1 RNA load, mean 100–300 days p/s	gag quasispecies				env quasispecies				Time of ART initiation (if any) during follow-up, days p/s
				Earliest time point, days p/s	Latest time point, days p/s	N of time points	N of gag sequences	Earliest time point, days p/s	Latest time point, days p/s	N of time points	N of env sequences
OC-2381	M	25	5.11	27	1,446	10	159	86	1,496	12	149
OE-2530	M	26	5.40	52	236	5	68	52	236	4	66
B-2865	F	20	5.12	10	223	6	72	10	223	6	78	199
OM-2869	F	23	4.11	86	178	2	38	86	178	2	25
OQ-2990	F	20	5.61	52	415	5	44	52	415	5	75	270
OS-3079	M	42	3.74	62	1,213	10	150	62	583	7	82
OU-3091	F	27	3.71	13	471	7	116	13	471	8	78
OX-3251	F	46	1.78	58	555	7	57	58	464	6	66
OY-3244	F	37	3.36	51	514	8	48	51	514	7	33
OZA-3285	F	31	3.56	78	540	7	73	78	540	7	60
C-3312	F	32	5.44	4	445	7	143	4	445	6	95	362
OZL-3354	F	30	3.47	6	1,259	8	97	6	1,259	6	62
PA-3390	F	24	4.13	65	227	4	67	65	227	4	44
E-3430	F	35	3.51	30	404	8	99	30	404	7	89
PC-3481	F	32	5.32	54	418	6	80	54	418	6	51	388
F-3505	F	53	4.28	7	447	7	45	7	447	7	58
PD-3508	F	25	4.22	59	428	5	44	59	428	5	44
G-3603	M	34	3.93	4	438	6	85	4	345	5	69
QA-5099	M	28	5.09	67	389	4	64	6	389	4	42	150
QC-5340	F	22	4.01	80	415	4	47	80	415	4	56
QD-5355	F	27	4.30	79	420	4	59	79	420	4	48
QG-5511	F	37	1.96	48	415	5	11	48	415	4	14
H-5582	F	26	3.90	16	347	6	88	16	378	7	99
QI-5715	F	27	5.34	20	476	5	44	20	476	4	48	112
QJ-5768	F	29	3.59	8	471	7	77	8	471	6	68
QM-5849	F	37	2.49	48	415	4	21	48	357	3	27
QR-5943	F	26	3.37	7	485	5	58	7	485	4	51
QS-6020	F	26	2.22	44	351	5	31	44	351	4	34
QT-6024	F	24	3.59	44	349	5	64	44	349	5	77
QU-6029	F	25	5.06	84	266	4	43	84	266	4	55
RA-6343	F	56	4.25	54	422	6	84	54	422	4	54
RB-6380	F	28	5.25	52	332	4	40	52	332	4	69
Median		28		50	419	6	64	50	417	5	59
IQR1		25		15	351	5	44	12	351	4	47
IQR3		34		60	472	7	84	60	471	6	76

2.2. HIV-1C sequences

The methodology of generating gag and env quasispecies was presented in detail elsewhere (Novitsky et al., 2009a; Novitsky et al., 2011b; Novitsky et al., 2009b; Novitsky et al., 2010b; Novitsky et al., 2011c). Briefly, for both cDNA and DNA templates, a strategy of single-genome amplification by limiting dilutions (Liu et al., 1996; Palmer et al., 2005) was employed with some modifications. The analyzed region of gag corresponded to nucleotide positions 841 to 2,217 of HXB2, and the analyzed region in env corresponded to nucleotide positions 6,615 to 7,757 of HXB2. The FastStart High Fidelity Enzyme Blend (Roche Diagnostics, Indianapolis, IN) was used for all PCR amplifications. Amplicons were purified by Exo-SAP (Dugan et al., 2002), and were sequenced directly at both strands on the ABI 3730 DNA Analyzer using BigDye technology. Sequence contigs were assembled by SeqScape v.2.6. Multiple sequence alignments were performed by using MUSCLE v.3.5 (Edgar, 2004) followed by minor manual adjustments in BioEdit (Hall, 1999). Analysis was performed using separate alignments for each subject. Only intra-host insertions and deletions were present in alignments. In contrast to the alignment of viral sequences that originate from different hosts, the number of indels in the intra-host alignment was relatively small, and therefore, the individual alignments were not gap-stripped. The obtained sequences were tested by HYPERMUT v.2.0 (Rose and Korber, 2000) and hypermutated sequences were excluded from analysis.

2.3. Estimating viral substitution rates

To estimate evolutionary rates in HIV-1C gag and env gp120 we used a comprehensive BEAST package v.1.7 (Drummond and Rambaut, 2007) that incorporates phylogenetic uncertainty into estimates of evolutionary parameters (Gray et al., 2011) as described previously (Stadler et al.). Briefly, the optimal evolutionary model estimated by jModelTest (Posada, 2008) was used for inferring the maximum likelihood (ML) tree by PhyML (Guindon et al., 2009; Guindon et al., 2010). The ML tree was used as a starting tree in Bayesian analysis. The independent HKY+G model was used for each codon, which allowed us to estimate relative rates of evolution at each codon position and codon rate ratio values of the 1^st to 3^rd codon position, and 2^nd to 3^rd codon position. We employed a relaxed molecular clock approach that explicitly models and estimates the level of rate variation among lineages (Drummond et al., 2006). The posterior distribution for the coefficient of variation that does not include zero indicates that the relaxed clock model provides a better fit to the data than the strict clock model (Drummond et al., 2006; Gray et al., 2011). The prior on the HKY transition/transversion parameter (κ) was Gamma distribution with a shape of 0.05 and a scale of 20. The prior on the alpha shape parameter (α) was an exponential distribution with a mean of 1. The ucld.stdev (S) parameter of the lognormal relaxed clock had an exponential prior with mean of 0.333. Viral evolutionary rates before and after the ART were estimated in a subset of six subjects who started ART during the follow-up period (Table 1) by assigning temporal partitions corresponding to the sampling intervals before and after initiation of ART.

2.4. Statistical analysis

Data are summarized with medians (inter quartile range for 25% and 75%). Comparisons of continuous outcomes between two groups were based on the Mann-Whitney Rank Sum test/Wilcoxon test. Analysis of relative evolutionary rates at different codon positions was performed using One Way Analysis of Variance. Associations between substitution rates and HIV-1 RNA in plasma were assessed using linear regression analysis. All reported p-values are 2-sided and not adjusted for multiple analyses.

2.5. Accession numbers

A total of 1,963 HIV-1 subtype C sequences used in this study were deposited to GenBank previously with the following accession numbers: GQ275453–GQ275667, GQ275750–GQ276051, GQ276137–GQ276247, GQ870874–GQ870904, GQ276248–GQ276284, GQ276418–GQ276698, GQ276766–GQ276795, GQ276797–GQ277080, GQ277179–GQ277222, GQ277224–GQ277293, GQ277295–GQ277374, GQ277404–GQ277443, GQ277446–GQ277569, GQ375107–GQ375128, GQ870874–GQ870939, GQ870957–GQ871036, and GQ871050–GQ871183. Additionally, 2,219 HIV-1C sequences were deposited to GenBank within this study: accession numbers KC628761–KC630979. The supplementary Table S1 includes information for all viral sequences used in this study including sequence ID, former sequence ID (if any), GenBank accession numbers, patient code, source of amplification, date of sampling, and the number of days from estimated seroconversion.

3. RESULTS

3.1. Intra-host evolutionary rates in HIV-1C gag and env gp120 V1C5

The gag- and env-based maximum likelihood trees with all subjects, 5 sequences per subject randomly selected from different time points of sampling, are shown in Supplementary Figures S1A and S1B for gag and env, respectively. The individual phylogenetic trees for each study subject for HIV-1C gag are presented in Supplementary Figures S02 to S33, and for the HIV-1C V1C5 region of env gp120 are shown in Supplementary Figures S34 to S65.

The individual intra-patient nucleotide substitution rates within HIV-1C gag and env gp120 V1C5 among study subjects are presented in Figure 1A, while a summary is shown in Figure 1B. The median (IQR) of the nucleotide substitution rate within HIV-1C gag was 5.22×10⁻³ (3.28×10⁻³ to 7.55×10⁻³) substitutions per site per year ranging from 3.39×10⁻⁴ to 2.25×10⁻². The median (IQR) of the substitution rate in env gp120 V1C5 was 1.58×10⁻² (9.99×10⁻³ – 2.04×10⁻²) substitutions per site per year of infection, ranging from 1.88×10⁻⁴ to 5.85×10⁻². The observed evolutionary rates were significantly higher in env gp120 V1C5 (~ 3-fold) than in gag (p<0.001, Wilcoxon test). The distribution of nucleotide substitution rates in env gp120 V1C5 and gag was skewed toward smaller values, suggesting that the majority of HIV-1C-infected individuals exhibit relatively low evolutionary rates, while a small proportion of subjects maintain elevated levels of viral evolutionary rates during primary HIV-1C infection. The linear regression analysis demonstrated a statistically significant (p=0.015) but weak (adjusted R² = 0.155) direct correlation between substitution rates in env gp120 V1C5 and gag (Fig. 1C).

Figure 1.

Nucleotide substitution rates within HIV-1C gag and env gp120 V1C5 among 32 study subjects infected with a single viral variant. A: Individual intra-host rates of nucleotide substitution, shown as substitutions per site per year with lower HPD within bars and upper HPD outside bars. Subjects identified within acute HIV-1C infection are highlighted in red. Asterisks highlight subjects who initiated ART during the follow-up period. B: Summary of nucleotide substitution rates within HIV-1C gag and env gp120 V1C5. The boxplots highlight the median and the 1^st and the 3^rd quartiles of nucleotide distribution rates in gag and env gp120 V1C5. Whiskers correspond to the “1.5 rule”: less than Q1 – 1.5*IQR and greater than Q3 + 1.5*IQR. Open circles indicate outliers. C: Analysis using linear regression model to assess potential association between substitution rates in env gp120 V1C5 and gag.

3.2. Relative evolutionary rates at codon positions

Within HIV-1C gag, the median (IQR) relative rates of evolution were 0.73 (0.48 to 0.84), 0.67 (0.52 to 0.86), and 1.54 (1.21 to 1.71) at codon positions 1, 2, and 3, respectively. Thus, in gag, the relative evolutionary rates at the 1^st and 2^nd codon positions were significantly lower than at the 3^rd codon position (p<0.001, One Way Analysis of Variance), which is expected for a continuous protein-coding region. In contrast, within the analyzed region of HIV-1C env gp120 V1C5, the median (IQR) relative rates of evolution at codon positions 1, 2, and 3 were 1.01 (0.86 to 1.15), 1.05 (0.99 to 1.21), and 0.86 (0.67 to 0.94), respectively. Therefore in env gp120 V1C5, the relative evolutionary rates at the 1^st and 2^nd codon positions were significantly higher than at the 3^rd codon position (p<0.001, One Way Analysis of Variance), suggesting a substantial selection pressure in the analyzed region of HIV-1C env.

The codon rate ratio values can be used to estimate the ratio of the evolution rates at different codon positions. The codon rate ratio values of the 1^st to 3^rd codon positions, and the 2^nd to 3^rd codon positions, were found to be lower in HIV-1C gag than in env gp120 V1C5 (p<0.001 for both comparisons, Wilcoxon test; Fig. 2). The first to the third position codon rate ratio > 1.0 was found in only 4 (12.5%) gag cases but in 25 (78.1%) env cases, while the second to the third position codon rate ratio > 1.0 was observed in only 2 (6.3%) gag cases but in 26 (81.3%) env cases (p<0.001 for both comparisons, Fisher’s exact test).

Figure 2.

The codon rate ratio values of the 1^st to 3^rd codon position, and 2^nd to 3^rd codon position in HIV-1C gag and env gp120 V1C5. The gray shadow at the bottom highlights the area below 1.0 and denotes the proportion of subjects below and above 1.0 for each codon rate ratio.

3.3. ART and viral evolutionary rates

To address whether initiation of ART alters the levels of substitution rates in gag and/or in env we compared viral evolutionary rates before and after the ART in a subset of six subjects who dropped CD4 levels and started ART during the follow up period. The median (IQR) time of ART initiation was 234 (162 to 339) days p/s. The median (IQR) time of follow-up after ART initiation was 114 (43 to 216) days. Viral substitution rates in gag decreased in 3 of 6 subjects, and increased in the other 3 subjects (Fig. 3A left panel). In env gp120 V1C5, substitution rates increased in one subject, and decreased in the other 5 subjects (Fig. 3A, right panel). Overall, the substitution rates in gag and env gp120 V1C5 did not differ before and after initiation of ART (p>0.1 for both comparisons), at least within the first 3–4 months after starting ART. Analysis of relative evolutionary rates at codon positions demonstrated a similar non-significant difference between viral substitution rates before and after subjects were placed on ART. The changes in codon rate ratio values of the 1^st to 3^rd codon position (Fig. 3B), and 2^nd to 3^rd codon position (Fig. 3C), did not reach statistical significance in either gag or env gp120 V1C5 (p>0.1 for all comparisons).

Figure 3.

Nucleotide substitution rates before and after initiation of ART in six subjects who initiated ART during the follow-up period. A: Nucleotide substitutions per site per year in HIV-1C gag and env gp120 V1C5. B: Codon position ratio 1^st to 3^rd. C: Codon position ratio 2^nd to 3^rd.

3.4. Viral substitution rates and HIV-1 RNA load in plasma

We assessed potential associations between the levels of HIV-1 RNA in plasma and intra-host substitution rates in HIV-1C gag and env gp120 V1C5 in the linear regression model, and using two thresholds of HIV-1 RNA in plasma, 3.0 log₁₀ and 4.0 log₁₀ copies/ml. There was weak (adjusted R²=0.165) but statistically significant (p=0.012) association between substitution rates in gag and HIV-1 RNA load. No statistically significant association was found between substitution rates in env and HIV-1 RNA load. When analyzed within groups with the defined threshold of HIV-1 RNA load, the substitution rates in HIV-1C gag were higher in subjects with early viral set point above 3.0 log₁₀ copies/ml than in subjects with set point below 3.0 log₁₀ (median (IQR) 5.66×10⁻³ (3.45×10⁻³ to 7.94×10⁻³) vs. 1.78×10⁻³ (4.57×10⁻⁴ to 5.15×10⁻³); p=0.028; Mann-Whitney sum rank test). However, no difference in the gag substitution rates was observed between subjects with early viral set point above and below 4.0 log₁₀ copies/ml (p=0.33). In contrast, substitution rates in env gp120 V1C5 did not differ between subjects with set point below and above 3.0 log₁₀ copies/ml (p=0.20), but reached statistical significance between subjects with early viral set point greater than and lower than 4.0 log₁₀ copies/ml (median (IQR) 1.88×10⁻² (1.54×10⁻² to 2.46×10⁻²) vs. 1.04×10⁻² (7.24×10⁻³ to 1.55×10⁻²) substitutions per site per year; p=0.017; Mann-Whitney rank sum test).

4. DISCUSSION

Using a Bayesian phylogenetic framework we estimated the intra-host evolutionary rates in two structural HIV-1C genes, gag and env, during primary infection. We demonstrated substantial heterogeneity in the substitution rates between HIV-1C-infected subjects during the early stage of infection, provided evidence of higher rates in env than in gag, demonstrated differential patterns in the relative evolutionary rates at different codon positions in gag and env, assessed rates before and after initiation of ART, and evaluated associations between nucleotide substitution rates and levels of HIV-1 RNA load.

The HIV-1C nucleotide substitution rates were estimated at 5.22×10⁻³ (3.28×10⁻³ to 7.55×10⁻³) substitutions per site per year in gag, and at 1.58×10⁻² (9.99×10⁻³ to 2.04×10⁻²) substitutions per site per year in env gp120. As expected, our estimates of intra-host evolutionary rates in HIV-1C infection are higher than estimates of the inter-host rates, and are slightly higher than previous estimates of intra-host rates in HIV-1C by Walker et al. (Walker et al., 2005) and Abecasis et al. (Abecasis et al., 2009). One of the potential reasons (which is also a study limitation) is related to the nature of frequent serial sampling and including transient viral polymorphisms that might not be fixed yet. As a result, our estimates of the HIV-1C evolutionary rates could be higher than estimates made over a longer period of time with less frequent sampling. It is known that frequent sampling over a short time period may result in estimates that approach the viral mutation rate (Duffy et al., 2008).

Reasons underlying evolutionary rate variation between subjects are likely to include viral replication and population size, virus–host interactions and viral adaptation. While natural selection is the dominant force determining intra-host evolutionary dynamics of HIV-1 (Pybus and Rambaut, 2009), it is not surprising that the extent of selection pressure during the course of HIV infection differs between subjects and can cause variation in nucleotide substitution rates between individuals. In addition, upon virus transmission to a new host, reverse mutations toward subtype consensus (Friedrich et al., 2004; Leslie et al., 2004; Li et al., 2007; Novitsky et al., 2009b) are likely to affect evolutionary rates due to differences in the host genetic makeup between the previous and new hosts. Viral transient polymorphisms including deleterious mutations can also contribute to variability in HIV substitution rate between subjects (Edwards et al., 2006; Lemey et al., 2006).

As expected, the intra-host evolutionary rates in HIV-1C primary infection were higher in env than in gag, which is in line with previous studies (Abecasis et al., 2009; Walker et al., 2005). Three codon positions in the protein-coding genes have very different substitution rates and evolutionary dynamics (Yang and Yoder, 2003). The analysis of relative evolutionary rates across codon positions suggested a substantial difference between env and gag in primary HIV-1C infection in the context of selection pressure. The relative nucleotide substitution rates in the V1C5 region of gp120 were slightly above 1.0 at codon positions 1 and 2, while only 0.86 at codon position 3. This observation provides evidence for a strong selective pressure on the V1C5 region of gp120 during primary HIV-1C infection. In contrast, the relative evolutionary rates in gag were higher at the third codon position, highlighting differences between the two structural proteins during primary HIV-1C infection. The proportions of individuals with the codon rate ratio above and below 1.0 in gag and env provided additional evidence for difference between these genes.

Initiation of ART dramatically reduces the level of viral replication. Thus, it was important to address whether intra-host evolutionary rates differ before and after ART administration. In the limited subset of subjects who started ART during the follow-up period, and the reduced number of analyzed time points due to “before” and “after” subdivision, we found no evidence for statistically significant changes in the viral substitution rates, at least over the first 3–4 months after initiation of ART. In subjects with the highest evolutionary rates in env, OC and OE, the mean HIV-1 RNA load over the early stage of infection was also high, 5.11 and 5.40 log₁₀ copies/ml, respectively. However in the next three subjects with the highest evolutionary rates in env, OX, OZA, and PA, the early stage HIV-1 RNA load was relatively low or modest, 1.78, 3.56, and 4.13 log₁₀ copies/ml, respectively. Interestingly, none of the subjects with high evolutionary rates dropped their CD4 count and started ART during the follow-up period. We cannot exclude the possibility that nucleotide substitution rates might change at a later stage of ART.

In the linear regression analysis we demonstrated that higher nucleotide substitution rates in gag but not in env are associated with higher HIV-1 RNA levels in plasma during primary HIV-1C infection. In addition rates in both gag and env were higher among individuals with higher levels of HIV-1 RNA, although the threshold for early viral set point differed for gag and env. The observed associations suggest that higher levels of HIV replication result in higher intra-host nucleotide substitution rates, at least during primary infection.

In summary, the study utilized a unique prospective set of HIV-1C gag and env quasispecies to estimate intra-host evolutionary rates during primary HIV-1C infection, to assess the effect of ART initiation on rates, and to evaluate associations between evolutionary rates and early HIV-1 RNA. Better understanding of the ranges and dynamics of nucleotide substitution rates could help to improve accuracy in estimating divergence time of circulating HIV variants and reconstruction of HIV transmission histories.

Supplementary Material

01

S1. Phylogenetic relationships among representative viral sequences in 32 subjects 5 sequences per subject. Optimal evolutionary model identified by jModelTest was used for phylogenetic inference by maximum likelihood method in PhyML. The support of splits was estimated by bootstrapping (n=100). Individual branches are collapsed to highlight independent clustering of each subject. A: HIV-1C gag. Optimal evolutionary model is TVM (substitution code 012314) + G, α = 0.363. B: HIV-1C env gp120 V1C5 region. Optimal evolutionary model is TPM1uf (substitution code 012210) + G, α = 0.384.

02

S2–S33. Individual phylogenetic trees for 32 subjects based on HIV-1C gag. Trees were generated in BEAST v. 1.7 analysis. Maximum clade credibility trees were produced by TreeAnnotator v1.7.4 after burning in about 25% of 10,000 trees generated in Bayesian analysis. Node bars indicate 95% HPD of the tree height. The scale at the bottom is shown in days. In cases with low phylogenetic signal and disproportionally large 95% HPD (subjects QG, QI, and QM) node bars are not shown for clarity of the phylogenetic structure.

34

S34–S65. Individual phylogenetic trees for 32 subjects based on HIV-1C env gp120 V1C5 region. Trees were generated in BEAST v. 1.7 analysis. Maximum clade credibility trees were produced by TreeAnnotator v1.7.4 after burning in about 25% of 10,000 trees generated in Bayesian analysis. Node bars indicate 95% HPD of the tree height. The scale at the bottom is shown in days. In cases with low phylogenetic signal and disproportionally large 95% HPD (subjects OM, QG, QI, and QM) node bars are not shown for clarity of the phylogenetic structure.

Highlights.

• HIV-1C quasispecies from 32 subjects analyzed prospectively

• We assessed levels and distribution of intra-host substitution rates in primary HIV-1C infection

• We compared rates between HIV-1C gag and env

• We evaluated the relative rates of HIV-1C evolution across codon positions

• We estimated the effect of ART on viral substitution rates

• We addressed associations between viral evolutionary rates and HIV-1 RNA load