ABSTRACT
A thorough understanding of the role of human immunodeficiency virus (HIV) intrahost evolution in AIDS pathogenesis has been limited by the need for longitudinally sampled viral sequences from the vast target space within the host, which are often difficult to obtain from human subjects. CD8+ lymphocyte-depleted macaques infected with simian immunodeficiency virus (SIV) provide an increasingly utilized model of pathogenesis due to clinical manifestations similar to those for HIV-1 infection and AIDS progression, as well as a characteristic rapid disease onset. Comparison of this model with SIV-infected non-CD8+ lymphocyte-depleted macaques also provides a unique opportunity to investigate the role of CD8+ cells in viral evolution and population dynamics throughout the duration of infection. Using several different phylogenetic methods, we analyzed viral gp120 sequences obtained from extensive longitudinal sampling of multiple tissues and enriched leukocyte populations from SIVmac251-infected macaques with or without CD8+ lymphocyte depletion. SIV evolutionary and selection patterns in non-CD8+ lymphocyte-depleted animals were characterized by sequential population turnover and continual viral adaptation, a scenario readily comparable to intrahost evolutionary patterns during human HIV infection in the absence of antiretroviral therapy. Alternatively, animals that were depleted of CD8+ lymphocytes exhibited greater variation in population dynamics among tissues and cell populations over the course of infection. Our findings highlight the major role for CD8+ lymphocytes in prolonging disease progression through continual control of SIV subpopulations from various anatomical compartments and the potential for greater independent viral evolutionary behavior among these compartments in response to immune modulation.
IMPORTANCE Although developments in combined antiretroviral therapy (cART) strategies have successfully prolonged the time to AIDS onset in HIV-1-infected individuals, a functional cure has yet to be found. Improvement of drug interventions for a virus that is able to infect a wide range of tissues and cell types requires a thorough understanding of viral adaptation and infection dynamics within this target milieu. Although it is difficult to accomplish in the human host, longitudinal sampling of multiple anatomical locations is readily accessible in the SIV-infected macaque models of neuro-AIDS. The significance of our research is in identifying the impact of immune modulation, through differing immune selective pressures, on viral evolutionary behavior in a multitude of anatomical compartments. The results provide evidence encouraging the development of a more sophisticated model that considers a network of individual viral subpopulations within the host, with differing infection and transmission dynamics, which is necessary for more effective treatment strategies.
INTRODUCTION
Viral evolutionary changes often have a profound impact on human immunodeficiency virus type 1 (HIV-1) disease progression (reviewed in reference 1), with the most well-characterized changes involving shifts in coreceptor usage and adaptation to specific tissues and cell types (2). Cellular and receptor tropism shifts have been linked to faster AIDS progression (3) as well as to the onset and progression of AIDS-related neuropathology (4). Therefore, understanding the evolutionary processes and dynamics of the viral subpopulations (i.e., within tissues), or phylodynamics (5), within an infected host is crucial for clarifying the complex interplay between the virus and various aspects of the host immune system and antiretroviral therapies. However, access to the multitude of infected tissues and distinct cell populations is often difficult in the human host, especially at regular intervals over the entire disease course.
Infection of rhesus macaques with pathogenic simian immunodeficiency virus (SIV) is a widely used model of HIV-1 infection and AIDS progression (6). SIV-infected macaques exhibit clinical manifestations similar to those of untreated HIV-1-infected humans, albeit on a shorter time scale of approximately 1 to 3 years (6). Unfortunately, the low incidence, maintenance costs, and disease timeline associated with the macaque model limit its usefulness in terms of producing rapid and statistically sound results. Such limitations have led to the development of rapid disease models, such as the use of antibody-mediated depletion of the CD8+ lymphocyte arm of the antiviral response (7–9), which results in a compressed timeline of disease progression of 6 months or less (7, 10, 11).
Rapid models of AIDS progression have provided invaluable insight into the immunopathological (12–14) and neuropathological (15–17) mechanisms of HIV infection. Moreover, the role of the cellular immune response, specifically CD8+ lymphocytes, in controlling viral replication has become increasingly evident based on results obtained from these models and HIV-infected patients (12, 18). CD8+ lymphocytes, which include cytotoxic T cells and natural killer cells, are able to suppress viral binding and transcription and to eliminate the infected cell population (19), with the exception of macrophages, which are resistant to such lysis (20). CD8+ T-lymphocyte immune responses to HIV, however, are unable to eradicate infection because of the high mutation and replication rates of the virus resulting in emergence of adapted populations, as well as progressive impairment of the immune system. Given their impact on viral infectivity, particularly during the early stages of infection (21, 22), it is not surprising that the level of HIV-specific recognition of virus by CD8+ lymphocytes has been associated with slower disease progression (23) and has been proposed as an addition to current combined antiviral therapeutic strategies (24, 25). Despite knowledge of the critical roles of both viral evolutionary changes and the CD8+ response in HIV disease progression, few studies have focused on the impact of the cellular immune response on viral intrahost population dynamics, or phylodynamics (5), much less on viral subpopulations within tissues, which vary in immune cell composition. For this study, we utilized a phylodynamics approach to garner more detailed insight into the relationship of CD8+ lymphocytes and viral evolution at the levels of individual infected tissues and peripheral cell populations sampled longitudinally from SIV-infected macaques with or without CD8+ lymphocyte depletion. The results showed that while patterns of SIV evolution and selection pressure in non-CD8+ lymphocyte-depleted animals were characteristic of sequential population turnover and continual viral adaptation, these patterns were less clear in the CD8+ lymphocyte-depleted model, which was characterized by more distinct evolutionary behaviors in individual tissues and circulating leukocyte populations. The overall findings suggest a role for this arm of the immune system not only in driving viral evolution, in agreement with other studies, but also in doing so in a manner that affects viral subpopulations within a multitude of infected tissues and cell types.
RESULTS
CD8+ lymphocyte depletion affects the time to simian AIDS (SAIDS) but not SIV phylogenetic resolution.Four CD8+ lymphocyte-depleted (Mac251-DEP) and 6 non-CD8+ lymphocyte-depleted, or naturally progressing (Mac251-NP), macaques were inoculated with the SIVmac251 viral swarm and analyzed for this study. Longitudinal full-length gp120 sequences were successfully obtained from various tissue and cell populations—consisting of bone marrow (BM), bronchoalveolar lavage fluid (BAL fluid) macrophages, sorted peripheral CD3+ T lymphocytes and CD14+ monocytes, and plasma—from over the course of infection (Fig. 1). Additionally, sequences from meninges and three distinct lobes within the brain (frontal, temporal, and parietal cortices) were obtained at necropsy. Results of likelihood mapping and substitution saturation analyses for all sequence data sets indicated sufficient resolution for further phylogenetic analysis (data available upon request).
Timeline of infection and sampling for the CD8+ lymphocyte-depleted and naturally progressing macaque cohorts. Six CD8+ lymphocyte-depleted and 12 naturally progressing macaques were inoculated (orange) with an SIVmac251 viral swarm at 0 dpi. Depletion of CD8+ cells was obtained using an anti-CD8 (α-CD8) monoclonal antibody at 3 time points between 6 and 12 dpi (blue), as described previously (41). Lymphoid and nonlymphoid tissue samples, excluding the brain and meninges, were serially sampled at the depicted times (pink), whereas the brain and meninges were sampled postmortem (green). Two macaques from each cohort (“+2”) were sacrificed early, at 21 dpi (pink/green), simulating early brain sampling.
The temporal range of AIDS onset in Mac251-NP macaques (204 to 373 days postinfection [dpi]) indicated a significantly slower progression (P < 0.01) than that in Mac251-DEP animals (75 to 118 dpi) (Fig. 2). As reported previously (2), patterns for viral loads and CD4+ T-cell counts were also investigated in each animal in order to rule out possible confounding factors (such as set point viral load and CD4+ T-cell nadir) responsible for differences in disease progression between the Mac251-DEP and Mac251-NP cohorts. These findings were critical for reliable inferences of the impact of CD8+ lymphocyte depletion on viral evolution in the context of differential disease progression.
Differences in time to disease progression for CD8+ lymphocyte-depleted and naturally progressing macaques. Significant differences in the mean times to SAIDS onset (reported with standard deviations) were determined using Welch's t test. **, P ≤ 0.01.
Measurable evolution and temporal clustering (TC) contribute to differences in macaque cohort tree topologies despite similar evolutionary rates.Differences in viral population dynamics between the two cohorts were measured in terms of overall evolutionary rate, rate variation among lineages, and temporal patterns in the underlying gp120 tree topology. Regression analysis of the distance within the maximum likelihood (ML) tree (Fig. 3 and 4) of each tip (collected taxon) from the inferred root (root-to-tip distance) against sample collection time was used to determine if the SIV population in each macaque was measurably evolving (i.e., there was sufficient temporal resolution), as well as if the evolutionary rate was “clocklike” over time (Fig. 5).
ML phylogenetic tree reconstruction for each naturally progressing macaque. The following macaques were sampled longitudinally: N02, N04, N09, N05, N10, and N12. Animals N06 and N07 were sampled at a single time point (21 days postinfection). (Trees for N02, N09, and N10 were adapted from reference 2.)
ML phylogenetic tree reconstruction for each CD8+ lymphocyte-depleted macaque. The following macaques were sampled longitudinally: D03, D04, D05, and D06. Animals D01 and D02 (boxed) were sampled at a single time point (21 days postinfection). For all ML trees, the GTR + G model of nucleotide substitution was used in RAxML, with bootstrapping (1,000 replicates) analysis. Trees for this cohort were inferred by maximum likelihood using sequence data described in Strickland et al. (41).
Linear regression analysis of temporal signals in maximum likelihood phylogenetic trees for longitudinally sampled naturally progressing and CD8+ lymphocyte-depleted macaques. Linear regression was performed between the root-to-tip genetic divergence (y axis) in the RAxML-derived maximum likelihood trees and the sampling time point (x axis) of each sequence (yellow dots). The R2 value for each linear regression is given in Table 1. Longitudinal data are presented for six naturally progressing macaques (A) and four CD8+ lymphocyte-depleted macaques (B).
Although it was less pronounced for the Mac251-DEP animals, sufficient temporal resolution, indicated by a positive slope of the regression line, was present for all of the macaques. Positivity indicated continual divergence (measurable evolution) of the SIV population from the viral inoculum. However, the stronger linear relationship between divergence and sampling time (greater R2 value), indicating stronger adherence to a strict molecular clock, was significantly greater for the Mac251-NP macaques than for the Mac251-DEP macaques (Fig. 5 and Table 1). Despite the more clocklike nature of evolution in the Mac251-NP animals, the uncorrelated relaxed clock evolutionary model, wherein evolutionary rates are distributed lognormally across tree branches, was a better fit to describe SIV intrahost evolution in all animals in both cohorts according to Bayesian model testing (Bayes factor [BF] values of >20) (26). Estimates of mean evolutionary rates according to the relaxed clock model did not differ significantly between the two cohorts. Yet the level of rate variation among intrahost viral lineages, as measured by the coefficient of variation, was significantly greater for the Mac251-DEP macaques (Table 1). Although additional analyses are required to determine if evolutionary rate variation is tissue and/or cell population dependent in Mac251-DEP animals, the results indicated an impact of CD8+ lymphocytes on viral population dynamics within the host.
Analysis of temporal signals for time-stamped phylogeniesd
In order to determine if stricter clocklike evolution in the Mac251-NP animals was the result of time-progressive population bottlenecks and subsequent turnover, which are characteristic of HIV infection (27–29), we analyzed the extent of phylogenetic clustering of taxa according to time of sampling for both cohorts. This phenomenon, referred to as temporal clustering (TC) or temporal structure (30), was present to a significantly greater (P ≤ 0.001) degree in the Mac251-NP macaques, as measured quantitatively using the TC statistic (Fig. 6). These results suggest that although CD8+ cells do not significantly alter the evolutionary rate of the virus, they play a major role in continual viral population turnover and evolutionary adaptation to the host immune response over the course of disease progression, even considering viral subpopulations from multiple anatomical locations.
Temporal clustering within maximum likelihood phylogenies for longitudinally sampled naturally progressing and CD8+ lymphocyte-depleted macaques. The temporal clustering statistic was determined using MacClade v4.08, based on comparison of the temporal clustering tree score for the given maximum likelihood tree to that for a null distribution of 1,000 replicates. Error bars represent standard deviations among all macaques. ***, P ≤ 0.001 using the Student t test.
Reduced temporal structure in CD8+ lymphocyte-depleted macaques is not attributed to increased reservoir formation.A lack of or low-level temporal structure, as observed for the Mac251-DEP animals, can often be explained by the reemergence (or reactivation) of archival genomes from long-lived cell populations, or reservoirs (27, 31, 32). One reasonable indication of a potential tissue/cellular reservoir for virus is the presence of metapopulation structure, or compartmentalization, within a phylogeny, which is characterized by the clustering of sequences based on their respective tissue/cellular origins (31). A few of these viral lineages may escape their residual compartment, giving rise to a peripherally sampled virus (derived from plasma or peripheral blood mononuclear cells [PBMCs]) that is more closely related to older sequences associated with the former compartment than the more recently sampled peripheral virus, thereby reducing the level of temporal clustering. However, differences in tissue/cell-specific clustering patterns were not observed between the ML phylogenetic trees belonging to the two cohorts (Fig. 3 and 4), as confirmed using both distance- and tree-based quantitative tests of compartmentalization (Table 2). Aside from previously reported brain sequence clustering (2), sequences from individual tissues and/or leukocyte populations appeared to cluster monophyletically for 2 or 3 macaques within each cohort (Fig. 3 and 4). Quantitative compartmentalization analysis was therefore performed on the basis of two hypotheses: (i) each tissue/cell type harbors a distinct SIV subpopulation (complete compartmentalization hypothesis) and (ii) only one tissue/cell type (not necessarily the same in each animal) harbors an SIV subpopulation distinct from the panmictic population of viral strains infecting the remaining tissues/cell types (single-compartment hypothesis). The analysis to test the first hypothesis revealed several Mac251-DEP and Mac251-NP animals with significant tissue-specific viral populations by use of the tree correlation coefficient (TCC), which measures both the number of branches (rb) and the distances along the branches (r) separating sequences from defined tissue/cell type “compartments” (33). None of the macaques in either cohort exhibited significant tissue compartmentalization as assessed by the Simmonds association index (SAI) (Table 2), which was used to account for uncertainty in the tree topology by use of bootstrapping (34). These results were not surprising given the appearance of clustering for only certain tissues/cell populations within the ML trees. The second quantitative analysis, to assess whether at least one tissue in each animal harbored a distinct SIV subpopulation (single-compartment hypothesis) by use of the more sensitive SAI value alone, revealed that, aside from brain sequence compartmentalization (discussed in detail in reference 2), BAL fluid macrophage and/or meningeal sequences were significantly compartmentalized in five different macaques across both cohorts (Table 2). Interestingly, the meningeal sequences were significantly compartmentalized among the Mac251-NP cohort only for macaque N10, which had SIV-associated encephalitis (SIVE). Viruses from the meninges of macaque D03 (also SIVE+) were also significantly compartmentalized (Table 2), suggesting that meningeal sequences can be genetically distinct from both peripheral and parenchymal brain sequences, consistent with an earlier study by Matsuda et al. (35). Additionally, this compartmentalization may be associated with SIVE, although the unavailability of meningeal sequences for the remaining SIVE+ Mac251-DEP animals prohibited a robust statistical approach to this hypothesis.
Analysis of tissue-dependent compartmentalization of viral RNA sequences
The occurrence of distinct viral subpopulations in meninges and lungs may be explained by macrophage tropism, as productive HIV and SIV infections of meninges (36) and lungs and/or epithelial lining fluid have predominantly been associated with cells of the monocyte/macrophage lineage (37–39) rather than with T lymphocytes. This finding is consistent with previous reports of monocyte/macrophage viral compartmentalization and the role of these cells as a potential reservoir (31, 40). Because this compartmentalization was independent of CD8+ lymphocyte depletion, macrophage tropism or macrophage reservoir formation is not likely to have been the primary contributor to reduced TC in the Mac251-DEP animals.
Time-dependent patterns in viral divergence accumulation distinguish CD8+ lymphocyte-depleted from naturally progressing animals.In order to determine if differences in temporal patterns in tree topology between the two cohorts were the result of differences in time-dependent viral subpopulation evolution, within-tissue diversity and tissue-dependent divergence from the viral swarm were investigated as a function of sampling time (Fig. 7A). The Mac251-NP cohort displayed similar increasing rates (across macaques and tissues) of both viral diversity and divergence until approximately 90 dpi. Following 90 dpi, the rate of viral diversity accumulation for most tissue and cell types decreased significantly (P ≤ 0.05) until 182 to 189 dpi and then remained relatively low (∼1.12E−05 substitution/site/day) until the last time point (average of ∼270 dpi) (Fig. 7B). Alternatively, following 90 dpi, divergence accumulation increased at a higher rate (P = 0.07) until 180 dpi, when it declined significantly (P ≤ 0.05), to ∼3.36E−05 substitution/site/day. It is important that despite the significant decline averaged across all tissues and cell populations, divergence increased significantly (P ≤ 0.05) at the last time point for virus from CD14+ monocytes and cell-free virus in plasma. An increase was also observed for BAL fluid macrophage sequences, although it was statistically insignificant (P = 0.07).
Genetic distance calculations over time for viral sequences derived from longitudinally sampled peripheral cell populations and meninges of naturally progressing and CD8+ lymphocyte-depleted macaques. (A) Viral diversity (solid lines) and divergence (dashed lines) in macaques were estimated as numbers of nucleotide substitutions/site for gp120 sequences derived from various cell populations over time for four different time points (dpi) and for meninges at necropsy. Divergence was estimated as the genetic distance from the inoculating SIVmac251 viral swarm sequences. The average viral divergence and diversity within the meninges and the brain compartments over time for the CD8+ lymphocyte-depleted macaques are highlighted in the inset. (B) Divergence and diversity accumulation rates. The depicted last time point (LTP) is an average of sampling times among all macaques prior to euthanization. Error bars represent standard deviations among all macaques. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; NA, not available (Welch's unpaired t test).
Compared to that for the Mac251-NP cohort, greater variation in diversity and divergence patterns across anatomical locations as well as macaques at the last sampling time point was observed for Mac251-DEP macaques (Fig. 7 and 8). No consistent pattern in divergence and diversity estimates was apparent across individual tissues or cell populations within the Mac251-DEP macaques over time (Fig. 7A); however, a significantly increased (P ≤ 0.05) rate of sequence divergence accumulation was observed during the last sampling time interval relative to that for the previous time interval (Fig. 7B). Although this result was in stark contrast to the reduced accumulation rate observed for the Mac251-NP cohort (Fig. 7B), the elevated BAL fluid macrophage (P ≤ 0.001), CD14+ monocyte (P = 0.06), and plasma (P = 0.07) sequence divergences characterizing the Mac251-DEP animals during the last time interval closely resembled the evolutionary events in the Mac251-NP cohort (Fig. 7A).
Genetic distance calculations for viral sequences derived from early-infection (21 dpi) and necropsy samples from naturally progressing and CD8+ lymphocyte-depleted macaques. Distances were estimated as numbers of nucleotide substitutions/site for gp120 sequences collected at 21 dpi and at necropsy, or the last time point (LTP), and were averaged across naturally progressing macaques (gray) and CD8+ lymphocyte-depleted macaques (black). Black asterisks (*) indicate a statistically significant elevation in viral diversity and/or divergence for naturally progressing macaques relative to that for CD8+ lymphocyte-depleted macaques, whereas gray asterisks indicate statistical significance for the reverse. Error bars represent standard deviations among all macaques. *, P < 0.05; **, P < 0.01; ***, P < 0.001; NA, not available (Welch's unpaired t test). (A) Within-tissue viral divergence from the inoculating viral swarm. (B) Within-tissue viral diversity.
When the tissues and cell populations were compared directly between the two cohorts, notable differences in divergence, but not diversity, were observed. Divergence estimates at 21 dpi did not differ significantly between the cohorts, but significantly reduced divergence (P ≤ 0.05) was observed by necropsy for six of the nine sampling locations in the Mac251-DEP animals (Fig. 8A). Within the Mac251-NP cohort, greater divergence estimates were observed for the longer-living macaques (data not shown), providing evidence of the association of viral divergence with the timeline to AIDS progression, though the extent of this relationship requires further investigation. BAL fluid macrophage and meningeal sequences were exceptions to this pattern, as their divergence rates were greater in the Mac251-DEP animals. We believed it important that these animals developed SIVE and/or meningitis within 4 months of infection (41) (both associated with macrophage and meningeal infections), whereas only one macaque within the Mac251-NP cohort developed either pathology, suggesting a relationship to development of AIDS comorbidity rather than a more rapid AIDS progression. Alternatively, viral diversity estimates did not differ significantly between the two cohorts for all sampling locations at 21 dpi and for seven of the nine locations at necropsy (Fig. 8B). This finding indicated that differences in viral divergence patterns between the two cohorts were not imposed by cloning-induced errors (after PCR error correction) that have been reported to influence measured viral diversity (42). The overall findings related to diversity and divergence indicate that CD8+ lymphocyte depletion limits viral divergence from the founder strain but not viral diversity. Moreover, differences in the temporal pattern of divergence accumulation suggested that reduced viral divergence in the Mac251-DEP animals at necropsy was not simply a function of observed time, although these animals progressed significantly faster to AIDS, but was related to the temporal patterns of CD8 immune selective pressure.
CD8+ lymphocyte-depleted animals exhibit a compressed timeline of selective pressure, with differing contributions from anatomical compartments.As selective pressure is an important driving force behind changes in viral diversity and divergence, site-specific selection analysis within gp120 was utilized to determine the proportion of viral amino acid sites experiencing either purifying or diversifying selection at each sampling time point for macaques within both cohorts (Fig. 9A) and for individual tissues and/or cell populations (Fig. 9B). Although purifying selection appeared to dominate for both cohorts, the Mac251-DEP macaques harbored a larger proportion of sites under diversifying selection than that for the Mac251-NP macaques at all time points (Fig. 9A). Significant differences (P < 0.05) between the two cohorts for both diversifying and purifying selection were observed by 92 dpi, with the difference in diversifying selection potentially explained by viral evolution in the CD14+ monocyte population (P < 0.08), which may have contributed to the similar pattern observed for cell-free virus in plasma (P < 0.05) (Fig. 9B). Despite the lack of significance until 92 dpi, the proportion of sites experiencing either diversifying or purifying selection was significantly larger (P ≤ 0.05) in the CD3+ T-lymphocyte population at 21 dpi for the Mac251-DEP animals than for the Mac251-NP animals. The reason for this finding is unclear, and further investigation is required to determine the specific relationship between an absence of CD8+ immune pressure during acute infection and evolutionary dynamics in peripheral T cells. Importantly, however, when necropsy samples were compared directly between the two cohorts, the difference in selective pressure was no longer significant, suggesting a compressed timeline of selection, likely the result of the eventual partial rebound of CD8+ cells in the depleted animals (43).
Site-specific selection over time for SIV sequences within individual naturally progressing and CD8+ lymphocyte-depleted macaque tissues/cell populations. (A) The fast, unconstrained Bayesian approximation for inferring selection (FUBAR) model implemented in HyPhy was used to measure selection at individual sites within viral gp120 sequences for individual macaques at each time point. (B) The proportions of sites with posterior probabilities of an increased (diversifying selection) or decreased (purifying selection) rate of nonsynonymous relative to synonymous substitutions of >0.9 are reported as averages across naturally progressing (open triangles) and CD8+ lymphocyte-depleted (filled triangles) macaques for individual tissues/cell populations (colored accordingly) at each sampled time point. Error bars represent standard deviations among all macaques. *, P ≤ 0.05; **, P ≤ 0.01; NA, not available (Welch's unpaired t test).
DISCUSSION
Early modulation of the cell-mediated arm of the host immune system in the CD8+ lymphocyte-depleted, or Mac251-DEP, macaque model of HIV/AIDS is thought to play a major role in evolutionary processes of the virus at both the population and tissue/cell subpopulation levels. The complex interplay between the host's primary immune defenses and viral population dynamics, referred to as phylodynamics (5), can be observed as an evolutionary fingerprint in the phylogenetic reconstruction of serially sampled viral lineages (5, 27, 44). Analysis of phylogenetic characteristics, such as tree topology, was therefore used to provide information about the extent of this relationship.
A distinctive pattern of virus-host interaction is evident in the staircase topology of HIV-1 intrahost genealogies, representing sequential population turnover and continual viral adaptation to the multitude of host immune responses over time (27–29). One of the central findings of this study was that this characteristic tree topology was evident only for naturally progressing (Mac251-NP), non-CD8+ lymphocyte-depleted macaques. The temporal structure of the Mac251-NP macaque phylogenies did not fully resemble the clear staircase pattern typical of HIV-1-infected patient plasma samples (28); however, the deviation may be due in part to the incorporation of sequences from anatomical locations other than the blood. Distinct infected anatomical locations likely experience immune-driven selective pressure differing in both type and degree, resulting in the occurrence of bottlenecks and subsequent expansion events at different time points over the course of infection for the individual viral subpopulations. This variation would essentially obscure the presence of well-defined bottlenecks if sequences from all locations were treated as a single intrahost viral population. The dissipation of temporal structure in the Mac251-DEP phylogenies relative to that for not only previously reported HIV-1 plasma sequences but also the Mac251-NP animals could not be explained similarly by sampling strategy, as the same locations were sampled in both cohorts at similar time points up to necropsy. Nor was the difference between these two groups supported by the explanation of differing levels of archival genome activation from viral reservoirs (31, 32). The reduced temporal structure in these animals may instead be explained by the absence of significant viral population turnover and adaptation as a result of a virtually nonexistent CD8+ immune response.
Similar to tree topology, analyses of viral sequence diversity and divergence in HIV-infected patients have been used in an attempt to understand disease progression, often producing conflicting results (e.g., see references 45 and 46). The most well-known study of evolutionary patterns associated with moderate disease progression was reported by Shankarappa et al. (28), who showed that viruses from PBMCs of HIV-1-infected patients exhibited a consistent three-phase pattern characterized by the rates of change in both viral diversity and divergence, as follows: an early phase, characterized by a linear increase in both diversity and divergence; an intermediate phase, characterized by a continued increase in divergence but stabilization or decline in diversity; and a late phase, characterized by the stabilization of divergence and continued stability or decline in diversity, which has been explained statistically by reduced selective pressure due to deterioration of the immune response. Despite the fact that the utilized sampling strategy was based on estimated times of viral population turnover rather than changes in viral load and CD4+ T-cell count, as in the study of Shankarappa et al., we have shown a strikingly similar three-phase pattern for naturally progressing SIV-infected macaques. Furthermore, we were able to dissect these patterns based on individual tissue and cellular locations, revealing a deviation from this trend in CD14+ monocytes, BAL fluid macrophages, and plasma, for which genetic divergence from the viral swarm continued to increase during the last sampling time interval for both macaque cohorts. This finding supports the “cellular exhaustion” hypothesis for both animal models, wherein depletion of the primary CD4+ T-cell target leads to infection of alternative cell types, such as monocytes and macrophages (36–39, 47, 48). Additionally, it highlights the differing evolutionary processes occurring in the various sampled anatomical compartments.
Immune selective pressure exerted on the virus by CD8+ lymphocytes is known to be an important environmental factor in driving viral diversity and divergence (49–52). With this in mind, the deviation of the Mac251-DEP viral populations from the three-phase model of Shankarappa et al. and the increased variation in evolutionary patterns among the macaques as well as among individual tissues/cell types could be explained by an altered viral fitness landscape. The depletion of significant immune selection pressure would promote a flatter landscape, as opposed to immune selection-driven uphill movement toward particular phenotypes that are more successful with respect to immune evasion. This flattened fitness landscape would also explain the overall reduced sequence divergence and compressed timeline of site-specific selection patterns in the rapidly progressing animals. In other words, our findings indicate that, in depleted animals, dramatically reduced selective pressure in the absence of CD8+ lymphocytes allows the virus to move more easily through the evolutionary landscape representation of possible phenotypes and to adapt more quickly to CD4+ target cell loss as well as the eventual CD8+ lymphocyte rebound, resulting in faster disease onset.
In summary, phylodynamics analysis in the absence of antiretroviral therapy (ART) is an important first step in understanding the evolutionary capabilities of individual viral subpopulations within the host and their relationship with specific immune selective pressures. The results described in this study indicate a critical role for the cell-mediated immune response in shaping viral evolution within and among individual tissues and cell populations that is distinct from the influence of ART. Although further investigation in the presence of ART is needed to define the links between disease progression, evolutionary variation, and viral fitness among HIV-targeted anatomical compartments, the current study implies the need for consideration of combined ART (cART) strategies designed to treat these compartments as individual but connected populations with differing infection and transmission dynamics.
MATERIALS AND METHODS
Study population.Two macaque cohorts were used in this study, which included 6 CD8+ lymphocyte-depleted (D01 to D06) and 12 naturally progressing, or non-CD8+ lymphocyte-depleted (N01 to N12), Indian rhesus macaques (Macaca mulatta), referred to here as Mac251-DEP and Mac251-NP macaques, respectively. Both cohorts were infected intravenously with the same SIVmac251 viral swarm (1 ng SIV p27) 2006 stock, which was described previously (53, 54). CD8+ lymphocyte depletion was achieved by subcutaneous administration of the anti-CD8 antibody cM-T807 (at 6, 8, and 12 dpi) (43). Two animals from each cohort were euthanized at 21 dpi in order to evaluate early evolutionary events within and related to the brain. The remaining animals were euthanized at the onset of SAIDS (75 to 118 dpi) (refer to Fig. 1 for a timeline). Criteria for development of AIDS included (i) weight loss of >15% of body weight in 2 weeks or >30% of body weight in 2 months; (ii) documented opportunistic infection; (iii) persistent anorexia for >3 days, without an explicable cause; and (iv) severe intractable diarrhea, progressive neurological signs, or significant cardiac and/or pulmonary signs, as previously described (55). A pathological diagnosis of SIV encephalitis (SIVE) was determined postmortem by a veterinary pathologist, included the presence of microglial nodules and multinucleated giant cells, and was confirmed by immunohistochemistry staining for SIV p27, as previously described (56–59).
Ethical guidelines.Procedures involving the Mac251-DEP animals were performed with the approval of Tulane University's Institutional Animal Care and Use Committee (IACUC). The treatment and handling of macaques in this cohort have been described previously (43). Procedures involving the Mac251-NP animals, which were housed at the New England Primate Research Center, were conducted according to the standards of the American Association for Accreditation of Laboratory Animal Care and IACUC protocol 04802, and treatment of all animals was in accordance with the Guide for the Care and Use of Laboratory Animals (60). Further detailed information on the handling and supervisory guidelines for the Mac251-NP cohort was published previously (61). All possible measures were taken to minimize discomfort of the animals, and the guidelines for humane euthanasia of rhesus macaques were followed.
Sample collection and sequencing.Plasma viral loads were monitored by quantitative PCR (qPCR) methods targeting a conserved sequence in the group antigen gene (gag) as previously described (62, 63). Plasma, fluorescence-activated cell sorter (FACS)-sorted peripheral CD3+ T lymphocytes and CD14+ monocytes, unelicited BAL fluid macrophages, and bone marrow (BM) aspirates were collected at two or three time points—21 dpi, 60 dpi, and necropsy—for 4 Mac251-DEP macaques and at four time points—21 dpi, 90 dpi, 180 dpi, and necropsy—for 6 Mac251-NP macaques (Fig. 1). Viral genomic RNA was extracted from these samples and the SIVmac251 inoculum as previously described (2, 41, 43, 61). Viral genomic material was also extracted from meninges and brain tissue sections from the parietal, frontal, and temporal lobes at necropsy, when available. Full-length envelope glycoprotein gp120 RNA sequences derived from the Mac251-DEP cohort were obtained using bulk PCR and cloning methods (41, 43). Due to the potential limitations of clonal analysis, a modified single-genome sequencing protocol based on previously published methods (42) was used for all samples obtained from the Mac251-NP cohort as well as for frontal lobe samples from the Mac251-DEP cohort. Sequences were aligned as previously described (41), and approximately 20 gp120 sequences per tissue per time point were obtained after removal of potential recombinants. Detailed information regarding sample collection, sequencing protocols, and the sequence alignment procedure has been reported previously (2, 43). All sequences used in this study are accessible in GenBank under accession numbers JF765272 to JF766081 (Mac251-DEP cohort) and accession numbers KR998525 to KR999900 , KX081185 to KX081229 , and KX081254 to KX082629 (Mac251-NP cohort and inoculating viral swarm).
Viral diversity and divergence estimates.Estimates of mean pairwise viral diversity within individual tissues/cell populations, as well as the mean divergence of longitudinally sampled sequences from the inoculating viral swarm, were calculated in MEGA v5.2.2 (64; http://www.megasoftware.net ), using the maximum composite likelihood model of nucleotide substitution (65) and 1,000 bootstrap replicates. Due to the influence of the bulk PCR/cloning method on sequence heterogeneity (42), mutated sites representing <1% of observed point mutations (estimated PCR error rate) were removed from the Mac251-DEP alignments prior to phylogenetic analysis, as described previously (54). After removal of these sites, overall viral diversity and divergence within tissue/cell populations during early and late infections were compared for the two cohorts to determine if differences between the two animal models could be explained by the sequencing methodology.
Phylogenetic and molecular clock analyses.Evaluation of phylogenetic resolution satisfying resolved phylogenetic relationships among SIV sequences from each macaque was performed using likelihood mapping (66) implemented in IQ-TREE (67; http://www.iqtree.org/ ), with searching for all possible quartets by use of the best nucleotide substitution model selected by a hierarchical likelihood ratio test. The absence of substitution saturation, which decreases the phylogenetic information contained in the sequences, was also assessed using DAMBE6 (68; http://dambe.bio.uottawa.ca/DAMBE/dambe.aspx ). Maximum likelihood (ML) reconstruction of phylogenetic trees for both macaque cohorts was performed in RAxML v8.0.25 (69; http://sco.h-its.org/exelixis/web/software/raxml/index.html ), using the general time-reversible (GTR) model of nucleotide substitution (70) with gamma-distributed rate variation across sites and 1,000 bootstrap replicates. Trees were viewed and modified in FigTree v1.4.0 (http://tree.bio.ed.ac.uk/software/figtree/ ) and can be found in previously published studies (2, 41).
Temporal resolution for each macaque phylogeny was assessed using a linear regression of root-to-tip genetic distances inferred from the ML trees against sampling time in the program Path-O-Gen v1.3 (71), now called TempEst (72; http://tree.bio.ed.ac.uk/software/tempest/ ). Trees were rooted either by using the known infecting viral swarm sequences or by selecting the root resulting in the best root-to-tip correlation (both methods produced identical results). Deviation from a strict molecular clock model was assessed based on the coefficient of determination (R2) values from the regression analysis and the coefficient of variation for evolutionary rates, assuming a relaxed clock model, in BEAST v1.8.0 (73; http://beast.bio.ed.ac.uk ). The HKY model of nucleotide substitution (74) with gamma-distributed rate variation across sites (4 categories) and the Bayesian skyride and/or Bayesian skyline plot (BSP) demographic model (75, 76) were used for parameter estimation. For further information on prior distributions, XML files are available upon request. Model testing was performed based on Bayes factor (BF) comparison (26), using the harmonic mean estimator in Tracer v1.5 (available from the BEAST software package). A ln(BF) value of >6 was considered significant evidence in favor of the more complex model. Statistical differences in R2 values and mean evolutionary rates reported in this study for the Mac251-NP animals and previously for the Mac251-DEP animals (41) were determined using two-tailed Student's t test (unpaired).
The degree of topological temporal structure within the ML phylogenies was assessed using a temporal clustering analysis performed in MacClade v4.08 (77; http://macclade.org/download.html ), as previously described (30). Briefly, sequences were assigned states according to sampling time, and the number of state changes within the maximum likelihood phylogeny (inferred without assuming a molecular clock) was compared to that for a null distribution of 1,000 replicates, for which the character states were randomized (while the phylogeny itself was held constant). The final result is referred to as the temporal clustering (TC) statistic, with values ranging from 0 to 1, with 0 indicating an absence of temporal structure (i.e., an intermix of sequences sampled at different time points) and 1 indicating perfect temporal structure (i.e., all sequences from the same sampling time cluster together and are the direct ancestors of sequences from the following sampling time).
Metapopulation structure analysis.Results of metapopulation structure (i.e., the existence of distinct subpopulations), or compartmentalization, analyses were obtained from two separate compartmentalization tests implemented in HyPhy (78; http://hyphy.org/ ) in order to evaluate the extent of distinct viral subpopulations within individual tissues. Analyses included both tree- and distance-based methods. Tree correlation coefficients were calculated based on the number of branches (rb) or branch length (r) separating sequences within separate defined compartments (33) in the RAxML-derived ML trees. Statistical significance was determined using a null distribution of permutated sequences (1,000 permutations), with P values of ≤0.05 considered significant. The Simmonds association index (SAI or AI) was determined for sequence alignments, using SIVmac239 as a reference sequence. The SAI represents the mean ratio for 100 bootstrap replicates of the association value calculated from the test sequences to that for 10 sample-reassigned controls. The association value (d) is defined as follows: d = (1 − f)/2n − 1, where n is the number of sequences below the node and f is the frequency of the most common sample type (79). Bootstrapping (1,000 replicates) was used as a test of significant compartmentalization, for which support of >80% was considered significant. A more thorough description of such tests has been published previously (34).
Selection analysis.Because statistical measures of metapopulation structure can be affected by selection as well as migration dynamics, an unrestricted branch-site random effects model, referred to as BUSTED (branch-site unrestricted statistical test for episodic diversification) (https://datamonkey.org ), was used to test for gene-wide episodic diversifying selection (80). The analysis was restricted to internal branches, which are assumed to capture at least one round of virus replication, to mitigate the biasing effects of transient deleterious mutations on the ratio of nonsynonymous to synonymous substitution rate estimates along terminal branches, where selection has not had time to fully filter such population-level variation (81, 82).
In addition, nucleotide sequence alignments for all tissues/cell populations at individual time points for each macaque were used to determine site-specific selection over the course of infection for both macaque cohorts. The fast, unconstrained Bayesian approximation for inferring selection (FUBAR) model (83; http://datamonkey.org ) was used to identify potential individual amino acid sites under selection within viral gp120 sequences for individual macaques as well as for individual tissues/cell types at each time point. Sites with posterior probabilities of an increased (diversifying) or decreased (purifying) rate of nonsynonymous relative to synonymous substitutions of >0.9 were considered to have experienced a significant level of selective pressure. Macaques were then classified according to SIVE diagnosis or early sacrifice in order to determine similarities and differences among classifications across macaque cohorts.
Statistical analysis.The statistical significance of differences in evolutionary parameters between cohorts was determined using either Welch's unpaired t test (unequal variance) or the Mann-Whitney U test, based on the Shapiro-Wilk test of normality (α = 0.05).
ACKNOWLEDGMENTS
This work, including the efforts of B.R.M., D.J.N., P.A., T.H.B., K.C.W., and M.S., was funded by the National Institutes of Health (NIH) (grants R01 NS063897, R01 NS040237, and F31 MH109398).
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
FOOTNOTES
- Received 7 July 2017.
- Accepted 24 August 2017.
- Accepted manuscript posted online 20 September 2017.
- Copyright © 2017 American Society for Microbiology.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
