Insufficient Production and Tissue Delivery of CD4⁺Memory T Cells in Rapidly Progressive Simian Immunodeficiency Virus Infection

Animals and Veterinary Procedures.

34 purpose-bred male RMs (Macaca mulatta) of Indian genetic background and free of Cercopithicine herpesvirus 1, D-type simian retrovirus, simian T lymphotrophic virus type 1, and SIV infection were used in this study. 14 were infected with WT SIVmac239, 4 with SIVmac239 (Δnef), 4 with SIVmac155T3, 6 with rhesus CMV (RhCMV) strain 68.1, and 6 were studied as uninfected controls. SIV infections were initiated with intravenous injection of 5-ng equivalents of SIV p27 (2.7–3.0 × 10⁴ infectious centers). RhCMV infections were initiated with subcutaneous injection of 10⁶ PFU. Bronchoalveolar lavage was performed by infusion and aspiration of 10-ml aliquots of sterile, pyrogen-free saline using a 3-mm pediatric fiberoptic endoscope (Olympus) gently wedged into a subsidiary bronchus. For in vivo 5-bromo-2′-deoxyuridine (BrdU) pulse labeling, 10 mg/ml of filter-sterilized BrdU, pH 7.2 (Sigma-Aldrich), was prepared in HBSS (Ca²⁺/Mg⁺⁺-free; Fisher Scientific) and administered intravenously at 30 mg/kg body weight at 24-h intervals for 4 d. All RMs were housed at the Oregon National Primate Research Center in accordance with the standards of the Center's Animal Care and Use Committee and the “NIH Guide for the Care and Use of Laboratory Animals” (17). RMs that developed disease states that were not reasonably clinically manageable were killed in accordance with the recommendations of the Panel on Euthanasia of the American Veterinary Medical Association (18).

Viruses and Viral Quantification.

The WT SIVmac239 (nef open), SIVmac239(Δnef), and SIVmac155T3 clonal virus stocks were prepared from pBR-SIVmac239, pBR-SIVmac239 Δnef, and pBR-SIVmac155T3 plasmids, respectively. The full-length pBR-SIVmac239 plasmid was provided by S. Wong (Oregon Health & Science University, Beaverton, OR; reference 19). The sequence of pBR-SIVmac239 Δnef (20) was modified to contain an Mlu1 site at position 9503 that spans the 181-bp nef deletion and results in a frame shift of downstream nef-coding elements (insuring stable nef inactivation). The pBR-SIVmac155T3 env recombinant was derived by replacing the env sequence in pBR-SIVmac239 from the Sph1 (nucleotide 6450) to Cla1 (nucleotide 8073) sites with the PCR-derived SIVmac155T3 sequence (21). The SIVmac155T3 recombinant clone has 22 amino acid differences compared with SIVmac239 env, and is a novel T cell–tropic SIVmac variant that uses CXCR4 as the primary coreceptor for virus entry (CCR5 use is reduced by 10-fold; unpublished data). Plasmid DNAs were transfected into COS1 cells using TransIT-LT1 (Mirus Corp.). The culture medium was collected from the cells 72 h after transfection, sterile filtered, and frozen in aliquots in liquid nitrogen. The virus stocks were assayed for p27-gag concentration by SIV core antigen assay (Beckman Coulter) and for infectious particles by sMAGI assay (22). Plasma SIV RNA was assessed using a real-time RT-PCR assay, essentially as described previously (threshold sensitivity = 100 SIV gag RNA copy equivalents/ml of plasma; interassay CV ≤25%; reference 23). RhCMV strain 68.1 was obtained from the American Type Culture Collection and propagated and titered in RM fibroblasts (24).

Immunofluorescent Staining and Flow Cytometric Analysis.

PBMCs, peripheral lymph node cells, and bronchoalveolar lymphocytes (BALs) were obtained and stained for flow cytometric analysis as described previously (25). Six-parameter flow cytometric analysis was performed on a two-laser FACSCalibur instrument using FITC, PE, peridinin chlorophyll protein-Cy5.5 (True Red), and allophycocyanin (APC) as the four fluorescent parameters. List mode multiparameter data files were analyzed using the PAINT-A-GATE Plus software program (BD Biosciences). Each analysis included a lineage-defining marker, CD4 or CD8β (with or without CD3ɛ), and two to three phenotyping markers. Memory and naive T cell subsets in peripheral blood and lymph node were delineated based on CD28 and CD95 expression patterns using previously described criteria (25). The specificity of staining and criteria for setting positive versus negative markers for CCR5, Ki-67, and BrdU expression were determined using isotype-matched negative control mAbs, and in the case of BrdU, BrdU analysis of study animals before BrdU pulsing.

mAbs.

mAbs L200 (CD4; True Red and APC conjugated), SP34 (CD3; FITC, PE, True Red, and APC), CD28.2 (CD28; FITC, PE, True Red, and APC), B56 (Ki-67; FITC and PE), B44 (anti-BrdU; FITC), DX2 (CD95; PE and APC), 12G5 (CXCR4; PE), 3A9 (CCR5; PE), and IgG1 and IgG2 isotype-matched controls were obtained from BD Biosciences. mAb 2ST8.5h7 (CD8β; PE, unconjugated) was obtained from Beckman Coulter. Purified 2ST8.5h7 was custom conjugated to True Red by BD Biosciences.

Statistical Analysis.

Between-group differences in quantitative dependent variables (except BrdU) were tested at all time points (days 0–205 after infection) and during acute (days 0–42) and plateau (days 42–205) phases of infection. BrdU decay rates were compared for 70 d after the last BrdU pulse. Given the small sample size and reported simulation results (26), we elected not to use mixed model ANOVA procedures. Instead, we took the mean of all time points for each animal within each interval, insuring that for each animal, a single value was modeled and, therefore, that data were independently distributed. Given the small and unequal sample sizes across groups in some tests, we conducted further sensitivity analyses to evaluate the robustness of our findings by redoing all tests using the nonparametric Wilcoxin rank sum test. In all cases but one, the conclusion with respect to statistical significance for the t test was confirmed by nonparametric testing, lending substantial confidence to the results. In the one case in which they were different (see Fig. 4), the difference was modest.

Clinical and Virologic Characteristics of WT and Variant SIVmac239 Infection.

22 RMs were monitored for 205 d after IV infection with (a) WT (CCR5-tropic) SIVmac239 (n = 14), (b) the (CXCR4-tropic) SIVmac239 env recombinant SIVmac155T3 (n = 4), or (c) the attenuated SIVmac239(Δnef; n = 4). 10 of the 14 WT SIVmac239-infected RMs remained free of SIV-related disease during the first 200 d of infection, and all 10 of these RMs survived at least 1 yr after infection. In contrast, the remaining four WT SIVmac239-inoculated RMs developed progressive disease requiring euthanasia at days 107–175 after infection. Necropsy revealed one or more AIDS-defining opportunistic infections in all four of these animals (Table I). Viral replication in the 14 RMs inoculated with WT SIVmac239 was quite high, with peak viral loads (almost invariably at day 10 after infection) ranging from 2 × 10⁷ to 2 × 10⁸ copies/ml, and plateau phase viral loads from 10⁵ to >10⁸ copies/ml (Fig. 1 A). One RM with the protective Mamu A*O1/B*17 MHC alleles (27) manifested spontaneous, temporary control of viral replication to <100 copy equivalents/ml at day 56 after infection, but reverted to a high replicative state (viral loads > 4 × 10⁵) ∼40 d later. In keeping with previous studies (16, 28, 29), the plasma viral loads in the four rapid progressors, particularly the post-acute phase (>day 42 after infection) viral levels, were significantly higher than in the normal progressor group (Fig. 1, legend). However, it is noteworthy that two of that latter group manifested viral replication kinetics overlapping that of the rapid progressors, yet maintained immune competence for >1 yr after infection.

The peak viremias observed in the SIVmac239(Δnef)-infected RMs, as is characteristic of these infections (30), was blunted compared with WT SIVmac239, and post-acute viral loads stabilized in the 2–3 log range (Fig. 1 A). SIVmac155T3 is a novel variant of SIVmac239 with 22 amino acid substitutions in gp120 that confer dominant CXCR4 tropism. Peak viral loads with this clone were similar to those of WT SIVmac239, but plateau phase viremia was considerably reduced, ranging from 10³ to 10⁵ copies/ml (Fig. 1 A). All of the SIVmac239(Δnef)- and SIVmac155T3-infected RMs studied here remained healthy for >1 yr after infection.

SIV Coreceptor Tropism, Patterns of CD4⁺ T Cell Depletion, and Early Disease Progression.

Although total peripheral blood CD4 counts have long been the benchmark of immunologic assessment of disease progression in both HIV and SHIV infection (1, 2, 31, 32), this parameter was not predictive of early pathogenesis in the SIV-infected RMs studied here (Fig. 1 B). Although the healthy SIVmac239(Δnef)-infected RMs in this work showed the expected stability of these counts (indeed, on average, CD4⁺ T cell numbers rose 39% in these RMs), the clinically indistinguishable SIVmac155T3-infected cohort showed rapid and profound initial depletion (>90%), followed by a delayed, slow rebound. Although the postinfection course of total CD4 counts in the clinically stable WT SIVmac239-infected RMs showed considerable variability, overall, these animals manifested a slow decline in total CD4⁺ T cell counts, averaging 44% over the 200 d of observation. Paradoxically, the WT SIVmac239-infected RMs destined for early pathogenicity (rapid progressors) uniformly demonstrated a striking (approximately threefold) CD4⁺ T lymphocytosis, starting at about day 42 after infection, and extending until the onset of overt disease (usually 2–3 wk before they were killed).

CD4⁺ T cells in peripheral blood are a composite of two phenotypically and functionally distinct subpopulations: the naive and memory/effector subsets (25), each with a characteristic pattern of coreceptor expression. As illustrated in Fig. 2, A and B (top), the primary coreceptor for WT SIVmac239, CCR5, is almost exclusively expressed by a subset of memory CD4⁺ T cells (mean CCR5⁺ ± SD for 21 uninfected RMs = 21.8 ± 7.7% of CD4⁺ memory T cells in peripheral blood). The alternative coreceptor CXCR4 is robustly expressed by both subsets, with the intensity of expression highest on naive cells (Fig. 2 B, top). Infection with CCR5- versus CXCR4-tropic SIV resulted in dramatically different changes in CD4⁺ T cell subset composition, in keeping with coreceptor-directed targeting. As shown in Fig. 2 A, a typical CCR5-tropic WT SIVmac239 infection induced a 75% loss of CD4⁺ memory T cells by day 49 after infection, and within the memory population, a 75% depletion of the CCR5-expressing population. Concomitant with this striking depletion of CCR5-expressing, CD4⁺ memory T cells, we observed a prominent change in the proliferation status of this subset, as measured by the cell cycle marker Ki-67 (25). Whereas before infection, CCR5⁺ CD4⁺ memory T cells were predominantly Ki-67⁻, after infection-associated depletion, this subset was predominantly Ki-67^bright. Thus, circulating CCR5⁺ CD4⁺ memory T cells are markedly depleted in SIVmac239 infection with the small, residual CCR5⁺ subset appearing to be maintained by a proliferative process. In contrast, a typical CXCR4-tropic SIVmac155T3 infection profoundly and preferentially depleted the naive CD4⁺ subset in the same timeframe (Fig. 2 B). CXCR4-expressing CD4⁺ memory T cells were not resistant to this virus, but this subset appeared to down-regulate CXCR4 expression so as to escape wholesale depletion. The frequencies of CCR5⁺ cells within the CD4⁺ memory T cell population actually increased in SIVmac155T3-infected RMs due to CCR5 expression by the proliferating subset (see below).

Analysis of absolute peripheral blood naive and memory T cell numbers in the different cohorts confirms and extends these findings (Fig. 3). In SIVmac239(Δnef)-infected RMs, absolute numbers of naive and memory CD4⁺ T cells in blood fluctuated over time, but overall, preinfection levels of both subsets were maintained for >200 d after infection. In SIVmac155T3 infection, the aforementioned rapid loss of circulating CD4⁺ T cells can be seen to be primarily the result of naive T cell depletion (essentially complete depletion was also observed in lymph nodes). Circulating CD4⁺ memory T cell numbers declined modestly during acute infection with this X4-tropic virus, but rebounded, and remained stable thereafter. In contrast, among WT SIVmac239-infected normal progressors, naive CD4⁺ T cell numbers were stable, whereas memory CD4⁺ T cell numbers demonstrated an abrupt decline in acute infection (after a brief CD4⁺ memory T lymphocytosis at day 7 after infection), followed by a slow decline thereafter. 4 of the 10 normal progressors showed some fluctuating or transient rebound in these CD4⁺ memory T cell numbers, but even for these RMs, the overall trend was downward. Among the SIVmac239-infected rapid progressors, the above-described CD4⁺ T lymphocytosis is clearly attributable to the naive subset, which given the low turnover of this population (25), likely reflects a redistribution of these cells from tissue to blood. Circulating CD4⁺ memory T cells uniformly decline in these RMs, but the kinetics and extent of this loss are largely overlapping with that of the normal progressors.

Representation of the CCR5-expressing component of the CD4⁺ memory population briefly increased in blood at day 7 after infection in all WT SIVmac239-infected RMs, but then rapidly declined (Fig. 4 A). In the clinically stable WT SIVmac239-infected RMs, the decline during acute infection was accompanied, as indicated above, by the appearance of Ki-67⁺ CCR5⁺ CD4⁺ T cells, which either maintained this fraction at a low level (at least 5% of the memory fraction; 5 of 10 RMs), or increased it back toward normal levels (the other 5 normal progressors). The initial decline of this population was similar in rapid progressors, but in contrast to the normal progressors, there was little evidence of rebound, and after day 42 after infection, the frequency of CCR5-expressing cells among CD4⁺ memory T cells was <3%, significantly lower than the normal progressor group (Fig. 4). The CCR5-expressing memory fraction increased in RMs infected with CXCR4-tropic SIV, and did not change in attenuated (Δnef) SIV infection.

Previous work has established that CCR5 is expressed by the vast majority of CD4⁺ memory T cells in mucosal tissues such as the intestinal and vaginal mucosae and the tissue–air interface of the lung, and that these populations are subject to extensive, acute depletion during CCR5-tropic SIV infection (4, 13, 33–35). Using the latter (BALs) as a repeatedly accessible “window” into the mucosal tissue compartment, we determined the relationship between the rate and extent of “effector” site CD4⁺ T cell depletion and clinical course in our study cohorts. As shown in Fig. 4 B, CD4⁺ T cell numbers in BALs were stable during early infection with either attenuated SIVmac239(Δnef) or CXCR4-tropic SIVmac155T3 infections, but rapidly declined with CCR5-tropic WT SIVmac239 infection. Normal progressors lost, on average, 82% of their lung CD4⁺ T cells by day 21 after infection (range: 63–96%), and 86% by day 28 after infection (range: 71–97%). Thereafter, about half of these RMs were able to transiently and partially regenerate CD4⁺ T cell percentages in the lung, whereas in the others, the CD4⁺ T cell fraction declined to ≤1%. WT SIVmac239-infected rapid progressors demonstrated a uniform 98–99% decline in CD4⁺ T cell frequencies by day 21 after infection, and showed no significant regeneration. Statistical assessments of the differences in the extent and kinetics of CD4⁺ (memory) T cell depletion in BALs in WT SIVmac239-infected normal versus rapid progressors were equivocal, with t tests suggesting significance after, but not before, day 42 after infection and nonparametric testing indicating the reverse (Fig. 4 B).

CD4⁺ Memory T Cell Regenerative Failure Precedes and Predicts Rapid Progression.

Next, we assessed whether differences in CD4⁺ memory T cell proliferation and turnover correlated with clinical outcome, particularly in animals with substantial acute-phase CD4⁺ memory T cell depletion in tissue. Two approaches were used to investigate this issue. First, we serially determined the fraction of CD4⁺ (and for comparison, CD8⁺) memory T cells expressing the proliferation marker Ki-67 (Fig. 5). Ki-67 expression provides a running average of the fraction of cells that have recently gone through S phase of the cell cycle with moderate/bright and dim staining reflecting S phase in the 3–4 and 5–8 d before sampling, respectively (25). Second, we labeled proliferating cells in vivo at strategic time points after infection with the thymidine analogue BrdU, followed by serial measurement of the fraction of CD4⁺ memory cells retaining this DNA label. Peak BrdU incorporation immediately after the last BrdU dose provides an independent measurement of the proliferating fraction, and the relative kinetics of BrdU decay during the wash-out period provides insight into the relative fate of the proliferating component over that time period. To facilitate interpretation of BrdU decay kinetics in SIV-infected RMs, we included two control groups in the study: healthy uninfected RMs (n = 6) and healthy RMs with primary RhCMV infection (n = 6).

These analyses revealed that SIV infection, in general, results in profound changes in CD4⁺ memory T cell turnover. In the first 7–10 d after infection with any of the SIV variants, coincident with peak viral replication, most animals showed a transient drop in the fraction of Ki-67⁺ cells within the circulating CD4⁺ memory T cell population. However, starting on days 14–21 after infection, the frequency of Ki-67⁺ cells within this population dramatically increased (Fig. 5). For WT SIVmac239-infected normal progressors, the fraction of CD4⁺ memory T cells expressing Ki-67 on day 42 after infection increased from preinfection values by an average (±SEM) of 2.6 (±0.3)-fold (an average increase of 27 ± 2.1 percentage points). A similar increase in Ki-67 expression was observed in CD4⁺ memory T cells in lymph nodes (not depicted) and in CD8⁺ memory T cells (Fig. 5). Most noteworthy, however, was the observation that the infection-associated increase in CD4⁺ memory T cell proliferation was sustained in all SIV-infected animals except those destined for rapid progression. In all four rapid progressors, the CD4⁺ memory proliferative response initiated with normal kinetics (peak percent Ki-67 within the CD4⁺ memory subset in blood during the acute phase, day 14–30 after infection, was 4.1- ± 0.9-fold higher than preinfection levels), but failed by day 42 after infection, with CD4⁺ memory T cell proliferative frequencies returning back to or toward baseline levels (Fig. 5). After day 42 after infection, the difference in CD4⁺ memory T cell proliferative frequencies between RMs destined for normal versus rapid progression was highly significant (P < 0.0001 by t test). Significantly, proliferative failure was CD4 memory T cell specific. There was no difference in the postinfection proliferative kinetics of CD8⁺ memory T cells between rapid and normal progressing WT SIVmac239-infected RMs.

Peak BrdU labeling in the acute phase of infection (days 10–13 or 24–27 after infection) confirmed the CD4⁺ memory T cell proliferative “burst,” with mean ± SEM of maximum percent BrdU⁺ of blood CD4⁺ memory T cells as 5.4 ± 0.5% for uninfected RMs, 20.6 ± 1.7% for WT SIVmac239-infected normal progressors, 16.2 ± 5.1% for WT SIVmac239-infected rapid progressors, 21.9 ± 5.3% for SIVmac155T3-infected RMs, and 14.7 ± 1.2% for SIVmac239(Δnef)-infected RMs (all SIV cohorts were significantly different from uninfected controls with t test p-values from 0.02 to <0.001). During the post-BrdU pulse “washout” period, BrdU⁺ memory T cells decline in frequency in a multiphasic process (25). In the initial 4 d of washout (phase 1), most BrdU⁺ cells are Ki-67^{moderate–high}, indicating many are still cycling, and labeled cells decline in frequency because of proliferative dilution (>5 postlabel divisions), as well as loss from the circulating pool (Fig. 6, A and B). From washout days 4–14 (phase 2), labeled cells have largely left the cell cycle (manifested by loss of Ki-67 expression and stabilization of BrdU staining intensity), and display a relatively rapid removal from the recirculating pool, as compared with the considerably slower, steady loss at later time points (phase 3). In SIV-infected RMs, loss of labeled cells in phases 1 and 2 was dramatically increased versus uninfected RMs (80 to >90% vs. ∼50%), consistent with a great predominance of high turnover/short-lived cells within the CD4⁺ memory proliferative burst. This CD4⁺ memory BrdU decay pattern was observed in acute SIV infection of all types, as well as primary RhCMV infection (Fig. 6 C), suggesting it is a common response to viral infection. Importantly, there was no significant difference between the CD4⁺ memory BrdU decay kinetics of the WT SIVmac239-infected RMs destined for normal versus rapid progression, or between either of these groups and the SIVmac155T3-infected RMs. Thus, the rate and extent to which proliferating CD4⁺ memory T cells are lost from the recirculating pool in acute SIV infection does not correlate with rapid progression.

As mentioned above, Ki-67 frequencies in the CD4⁺ memory population of WT SIVmac239-infected normal progressors remain high throughout the period of observation (Fig. 5), and plateau phase (after day 100 after infection) BrdU analysis of this population (Fig. 6 C, far right) demonstrates continued rapid decay, suggesting that the high turnover state is similarly maintained. Significantly, in the one WT SIVmac239-infected rapid progressor that survived long enough for a plateau phase BrdU decay analysis (i.e., after the onset of proliferative failure), CD4⁺ memory T cell BrdU decay kinetics was similar to that of uninfected RMs, suggesting that collapse of the infection-associated proliferative burst abrogates the high turnover/short-lived component of CD4⁺ memory proliferation, leaving a low turnover/long-lived component behind.

Delivery of CD4⁺ Memory T Cells to the Pulmonary Mucosal Surface in Stable versus Rapidly Progressive Infection.

The potential impact of CD4⁺ memory T cell proliferative failure on CD4⁺ memory populations in effector sites, and its possible link with rapid disease progression, was first revealed by comparison of BrdU versus Ki-67 staining patterns and BrdU kinetics in the blood, lymph node, and lung of (plateau phase) WT SIVmac239-infected normal progressors. As shown in Fig. 7 A, 24 h after the last dose of a 4-d BrdU pulse, the vast majority of BrdU-labeled CD4⁺ memory T cells in the blood and lymph node are also Ki-67⁺, indicating that they are relatively close to their last S phase. In contrast, at the exact same time, BrdU-labeled CD4⁺ memory T cells in the lung are predominantly Ki-67⁻ or dim, strongly suggesting that these cells are more remote from their most recent S phase, and did not go through S phase in the lung itself. This impression is strengthened by the observation that although maximum BrdU labeling in the blood and lymph node occurs 24 h after the last BrdU dose, maximum BrdU labeling in the lung occurs 7 d later (Fig. 7 C). These data lead to the conclusion that the pulmonary mucosal surface in these RMs is continually supplied by an influx of new CD4⁺ memory T cells that are the products of prior proliferative events occurring elsewhere.

Thus, the postinfection CD4⁺ memory proliferative response appears to provide a steady influx of new CD4⁺ memory T cells in mucosal effector sites, potentially countering the severe CD4⁺ T cell depletion in these sites and preventing overt immunodeficiency. Although accurate direct measurement of BrdU-labeled CD4⁺ T cell influx into the lungs was not possible in severely depleted SIV-infected animals, the phenotypic signatures of lung CD4⁺ T cells in normal versus rapid progressors strongly support this hypothesis. As shown in Fig. 7 B, newly produced, BrdU-labeled memory CD4⁺ T cells in the blood and lymph node might be either CCR5⁺ or CCR5⁻, but those arriving in the lung are almost all CCR5⁺. Moreover, although BrdU⁺ cells arriving in lung have undergone S phase elsewhere, many of them retain low level Ki-67 reactivity (Fig. 7 A). Taken together, these observations suggest that most new emigrants to lung will express Ki-67, CCR5, or both markers. As illustrated in Fig. 8 A, BAL CD4⁺ T cells of all WT SIVmac239-infected normal progressors did indeed predominantly express either or both CCR5 and Ki-67 at all plateau phase time points (mean ± SEM for BAL CD4⁺ T cells lacking both Ki-67 and CCR5 = 12.1 ± 4.5%), notably including those RMs with profound CD4⁺ T cell depletion in the lung (e.g., <1% CD4⁺ T cells). In striking contrast (Fig. 8 B), the residual BAL CD4⁺ T cells of the rapid progressors were almost entirely negative for both CCR5 and Ki-67 (mean ± SEM for BAL CD4⁺ T cells lacking both Ki-67 and CCR5 = 84.6 ± 2.4%). These differences were highly significant (P < 0.0001 by t test). Thus, even though normal and rapid progressors may have had identical overall frequencies of lung CD4⁺ T cells, the phenotypes of these T cells strongly suggest that the lungs of the clinically stable RMs were receiving a continuous influx of newly produced CD4⁺ effector T cells, whereas the lungs of animals destined for rapid progression were not.