Baseline CD4+ T cell counts correlates with HIV-1 synonymous rate in HLA-B*5701 subjects with different risk of disease progression

Abstract

HLA-B*5701 is the host factor most strongly associated with slow HIV-1 disease progression, although risk of progression may vary among patients carrying this allele. The interplay between HIV-1 evolutionary rate variation and risk of progression to AIDS in HLA-B*5701 subjects was studied using longitudinal viral sequences from high-risk progressors (HRPs) and low-risk progressors (LRPs). Posterior distributions of HIV-1 genealogies assuming a Bayesian relaxed molecular clock were used to estimate the absolute rates of nonsynonymous and synonymous substitutions for different set of branches. Rates of viral evolution, as well as in vitro viral replication capacity assessed using a novel phenotypic assay, were correlated with various clinical parameters. HIV-1 synonymous substitution rates were significantly lower in LRPs than HRPs, especially for sets of internal branches. The viral population infecting LRPs was also characterized by a slower increase in synonymous divergence over time. This pattern did not correlate to differences in viral fitness, as measured by in vitro replication capacity, nor could be explained by differences among subjects in T cell activation or selection pressure. Interestingly, a significant inverse correlation was found between baseline CD4+ T cell counts and mean HIV-1 synonymous rate (which is proportional to the viral replication rate) along branches representing viral lineages successfully propagating through time up to the last sampled time point. The observed lower replication rate in HLA-B*5701 subjects with higher baseline CD4+ T cell counts provides a potential model to explain differences in risk of disease progression among individuals carrying this allele.

Figure 1. HIV-1 gag p24 genealogies for all six HLA-B*5701 subjects displaying different branch sets.

Internal and external branches are shown for each subject (P1-P6), colored in purple and black, respectively. Orange branches indicate one of the possible backbone paths in the genealogy (see Methods). Specific estimates were based on 200 randomly sampled trees from the posterior distribution obtained with a Bayesian coalescent framework enforcing a relaxed molecular clock. Branch lengths are drawn proportional to the time scale (in days) at the bottom of each tree.

Figure 2. HIV-1 mean nucleotide substitution rate for different branches for all six HLA-B*5701 subjects.

Estimates along internal branches (purple), backbone paths (orange) and external branches (black) are shown for the HRPs (P1-P3) and the LRPs (P4-P6).

Figure 3. HIV-1 mean nonsynonymous and synonymous substitution rates for internal branches in HLA-B*5701 subjects.

Subject- specific estimates were based on 200 randomly sampled trees from the posterior distribution obtained with a Bayesian coalescent framework enforcing a relaxed molecular clock. HRPs (P1-P3) and LRPs (P4-P6) are shown in orange and purple, respectively. The p-value is marked in bold when there is a significant difference between the two groups of patients. Along the y-axis, internal refers to all internal branches; backbone refers to rates averaged along all the possible backbone paths.

Figure 4. HIV-1 mean nonsynonymous and synonymous divergence for internal branches over time in HLA-B*5701 subjects.

Subject- specific estimates were based on 200 randomly sampled trees from the posterior distribution obtained with a Bayesian coalescent framework enforcing a relaxed molecular clock. HRPs (P1-P3) and LRPs (P4-P6) are shown in orange and purple, respectively. Along the y-axis, internal refers to all internal branches; backbone refers to rates averaged along all the possible backbone paths.

Figure 5. HIV-1 mean synonymous substitution rates along backbone paths vs. CD4⁺ T cell count at baseline for each subject.

HIV-1 synonymous rates (dS) in gag p24 within each HLA-B*5701 subject (y-axis) were plotted against baseline CD4⁺ T cell counts (10–11 wpi). The squared correlation coefficient (r²) is given in the graph and resulted highly significant (p = 0.002).