Primate lentiviruses are typically apathogenic in their evolutionarily coadapted host species but can be lethal when transferred to new host species. Why such infections are pathogenic in humans and rhesus macaques (RMs) but not in sooty mangabeys (SMs), a natural host, remains unclear. Studies of chronically infected animals point to the importance of diminished immune activation in response to the infection in SMs. In this study, we sought the causes and timing of the differences in immune activation in a comparative study of acute SIV infection in RMs and SMs. Surprisingly, we show that in acute infection immune activation is comparable in SMs and RMs but thereafter, SMs quickly resolve immune activation, whereas RMs did not. Early resolution of immune activation in SMs correlated with increased expression of PD-1 and with preservation of CD4+ T cell counts and lymphatic tissue architecture. These findings point to early control of immune activation by host immunoregulatory mechanisms as a major determinant of the different disease outcomes in SIV infection of natural vs non-natural hosts.
African primate species naturally infected with SIV are generally spared from disease (1, 2). In contrast, viruses from naturally infected species, when transferred to a new nonadapted host species, can be highly pathogenic, even lethal, as with SIV infection in rhesus macaques (RMs)5 (Macaca mulatta) or HIV type 1 infection in humans. Indeed, SIV-infected natural hosts, e.g., sooty mangabeys (SMs) (Cercocebus atys), typically maintain normal peripheral blood CD4+ T cell counts and do not develop simian AIDS despite acute depletion of CD4+ T cells from gut-associated lymphoid tissue (3, 4) or despite levels of virus replication as high as, or even higher than, levels found in HIV-infected individuals (5, 6, 7). In contrast, pathogenic infection occurs with SM-derived viruses in RMs, which is the best-characterized and most widely used animal model for studies of AIDS pathogenesis (8, 9).
These different responses to infection have motivated studies aimed at identifying mechanisms of disease by comparing experimental SIV infection in pathogenic and nonpathogenic animal models. Although the reasons for the nonpathogenic outcome of SIV infection in SMs remain incompletely understood, one consistent and striking difference is that SIV-infected SMs do not manifest the chronic generalized immune activation that characterizes pathogenic SIV and HIV infections (5), a process that is thought to play a central role in driving CD4+ T cell depletion through bystander activation and loss of uninfected T cells. Of particular predictive importance, increased levels of CD8+ T cell activation have been shown to directly correlate with accelerated progression to AIDS and death in HIV-infected humans and SIV-infected RMs (10, 11, 12, 13, 14, 15, 16). Collectively, these findings have led to the widely held view that the lesser degree of chronic immune activation observed in SIV-infected SMs is a major determinant that favors the preservation of the CD4+ T cell population and underlies the nonpathogenic nature of SIV infection in natural hosts (5).
So how do SIV-infected SMs attenuate immune activation despite continued viral replication? We sought answers in studies reported of acute SIV infection, to identify the kinetics and potential mechanisms of the disease-sparing differences in immune activation between nonpathogenic SIV infection of SMs and pathogenic infection of RMs. Surprisingly, we found that early immune activation in SMs was as high or higher than activation seen in RMs, but SMs rapidly resolved acute immune activation in contrast to RMs. Resolution of high levels of acute immune activation was strongly associated with a rapid induction of the negative regulator, programmed cell death-1 (PD-1), on T cells within lymphatic tissue (LT), and with preservation of CD4+ T cell populations. These findings suggest that early control of immune activation in SIV-infected SMs is mediated by PD-1, with sustained effects on immune activation that distinguish nonpathogenic from pathogenic infections.
Materials and Methods
Animals and SIV infection
Five SMs and RMs were initially infected with uncloned SIVsmm infection as previously described (17, 18). Briefly, each animal was inoculated i.v. with 1 ml of plasma from an experimentally SIVsmm-infected SM sampled at day 11 postinfection, with a viral load at 1 × 107 copies/ml. Although all RMs became infected with the primary SIVsmm virus, the viral kinetics were significantly delayed in all RMs and attenuated in three of five RMs (data not shown). For this reason, comparative studies were undertaken between SMs infected with the primary SIVsmm virus listed in this experiment and RMs infected with the well-characterized SIVmac239 viral isolate. Thus, five new RMs were inoculated i.v. with 10,000 TCID50 of SIVmac239. The rationale for choosing this high dose of SIVmac239 was to compare SIVsmm infection in SMs with a fully pathogenic SIV infection in RMs. We further extended our analysis to chronic infection with analyses of LT samples obtained from both longitudinally derived animals from this experiment and animals from cross-sectional studies (five SMs and six RMs) previously described (19). In addition to LT, 34 naturally SIV-infected and 19 uninfected SMs were included in the cross-sectional peripheral blood analysis. Blood collection was performed by venipuncture. For lymph node biopsy, animals were anesthetized with ketamine or Telazol; the skin over the axillary or inguinal region was clipped and surgically prepped. An incision was made over the lymph node, which was exposed by blunt dissection and excised over clamps. A portion of each lymph node biopsy was placed immediately in fixative (4% neutral-buffered paraformaldehyde) and paraffin embedded, while the remaining portion was homogenized and passed through a 70-μm cell strainer to mechanically isolate lymphocytes for flow cytometric analysis. All animals were housed and cared for at the Yerkes National Primate Research Center (Atlanta, GA) in accordance with the regulations of the American Association of Accreditation of Laboratory Animal Care standards. These studies were approved by the Emory University and University of Pennsylvania Institutional Animal Care and Use Committees.
Viral load measurements
SIV plasma viral load was determined as previously described (5).
Immunohistochemistry and in situ hybridization
Immunohistochemical staining, in situ hybridization, and quantitative image analysis (QIA) were performed as previously described (19, 20, 21). In brief, immunohistochemistry was performed using a biotin-free polymer approach (MACH-3; Biocare Medical) on 5-μm tissue sections mounted on glass slides, which were dewaxed and rehydrated with double-distilled H2O. Ag retrieval was performed by heating sections in 1× DIVA Decloacker reagent (Biocare Medical) in a 95°C water bath for 30 min followed by cooling to room temperature. Nonspecific Ig-binding sites were blocked with Blocking Reagent (Biocare Medical) for 1 h at room temperature. Endogenous peroxidase was blocked with 3% (v/v) H2O2 in PBS (pH 7.4). Primary Abs were diluted in 10% Blocking Reagent in TNB (0.1 M Tris-HCL (pH 7.5), 0.15 M NaCl, and 0.5% Blocking Reagent (NEN)) and incubated overnight at 4°C. Tissue sections were washed, and mouse, goat, or rabbit MACH-3 secondary polymer systems (Biocare Medical) were applied according to the manufacturer’s instructions. Sections were developed with 3,3′-diaminobenzidine (Vector Laboratories) or Vulcan Fast Red (Biocare Medical) and counterstained with Harris Hematoxylin (Surgipath). These were dehydrated, mounted in Permount (Fisher Scientific), and examined by light microscopy using an Olympus BX60 upright microscope with the following objectives: ×10 (0.3 NA), ×20 (0.5 NA), and ×40 (0.75 NA). Light micrographs were taken using a Spot color mosaic camera (model 11.2) using Spot acquisition software (version 4.5.9; Diagnostic Instruments). Immunofluorescent confocal microscopy was performed as previously described (19) using an Olympus Fluoview confocal microscope. Primary Abs used for immunohistochemical and immunofluorescent experiments were the following: mouse anti-human CD4 (clone 1F6; Vector Laboratories), mouse anti-human granzyme B (clone GZB01; Neomarkers), rabbit monoclonal anti-human Ki67 (clone SP6; Neomarkers), rabbit monoclonal anti-human CD3 (clone SP7; Neomarkers), mouse anti-human CD20 (clone L26; DakoCytomation), mouse anti-human CD25 (clone 4C9; Vector Laboratories), rabbit monoclonal anti-human active caspase 3 (clone 8G10; Cell Signaling Technology), mouse anti-human collagen I (clone COL-1; Sigma-Aldrich), mouse anti-human FoxP3 (clone 236A/E7; Abcam), goat polyclonal anti-human PD-1 (R&D Systems), donkey anti-mouse IgG Alexa Fluor 488 and 555, donkey anti-goat IgG Alexa Fluor 488 and 568, and donkey anti-rabbit IgG Alexa Fluor 488 and 555 (all from Molecular Probes). Isotype-matched negative control Abs were used in every experiment and uniformly showed no staining (data not shown). QIA on 10–20 randomly acquired high-powered brightfield images (×200 or ×400 magnification) from the paracortical T cell zone (TZ) was performed by either manually counting the cells per image or by determining the percentage of area of the image that was occupied by chromagen using an automated action program in Adobe Photoshop CS with tools from Reindeer Graphics, as previously described (19, 20). Collagen type I QIA was performed essentially as previously described (19) with the exception that images were collected from a fluorescent microscope on Vulcan Fast Red-stained samples.
Immunophenotyping by flow cytometry
Seven- to eleven-color flow cytometric analysis was performed on either whole blood or cryopreserved samples according to standard procedures. Immunophenotyping of lymphocytes was performed using a panel of mAbs that were originally designed to detect human molecules, but that we and others have shown to be cross-reactive with SMs and RMs (5). The Abs used were the following: anti-CD4 PerCP or PerCP-Cy5.5 (clone L200), anti-CD8 Pacific Blue (clone RPA-T8), anti-Ki67 FITC (clone B56), anti-CD3 Alexa Fluor 700 (clone SP34-2), anti-CD69 PerCP (clone L78), anti-HLA-DR PerCP (clone G46-6), anti-CD95 PE-Cy5 (clone DX2) all from BD Pharmingen. Additionally, anti-CD28 PE-Cy7 (clone CD28.2; eBioscience), anti-CD14 PE-TR (clone RMO5; Beckman Coulter), anti-CD20 PE-TR (clone B9E9; Beckman Coulter), anti-CD8 Pacific Orange or QD655 (clone RPA-T8), which was courtesy of Dr. M. Betts (University of Pennsylvania, Philadelphia, PA), anti-PD-1 PE (clone EH12), courtesy of Dr. G. Freeman (Harvard University, Cambridge, MA), and LIVE/DEAD Fixable Violet Dead Cell Stain kit (Invitrogen) were used. Samples assessed for Ki67 were surface stained first with the appropriate Abs, then permeabilized using Perm 2 (BD Pharmingen), and stained intracellularly with anti-Ki67. Flow cytometric acquisition and analysis of samples was performed on at least 100,000 events on a LSRII flow cytometer driven by the DiVa software package (BD Biosciences). Analysis of the acquired data was performed using FlowJo software (Tree Star).
Linear regression models were used to assess the effect of PD-1 on the logarithm of the density of Ki67+ cells. The logarithms were used to better conform to the assumptions of homoskedasticity and normally distributed errors. The effect of time was incorporated by dichotomizing as <14 days postinfection or 14 days or longer. This dichotomization was motivated by our previous studies that showed peak acute infection is approximately at this point. The models considered included time (dichotomized), species and the log of the PD-1 level. All interactions were considered, including subject specific effects, but were found to be generally unnecessary (i.e., the autocorrelation in the longitudinal data for each animal was negligible after allowing for covariates of interest). The model best supported by the data (in terms of including only significant predictors and explaining most of the variation in the response variable) included a main effect for time (p = 0.0003), the interaction between PD-1 and species (p = 0.0006), and the interaction between species and time (p = 0.0007). The values for p for the F test for the entire model was 5.206 × 10−5 with a value for R2 of 0.40. Standard diagnostics indicated no serious violations of the usual linear model assumptions.
Primary SIV infection of SMs is associated with transient immune activation
Although several published studies have described the immunological and virological differences and similarities between nonpathogenic SIV infection of natural hosts and pathogenic SIV infection of RMs, most of these studies have focused on the peripheral blood or on the chronic stage of infection (1, 2, 5, 6, 7, 8, 9, 22, 23, 24, 25). In this study, we focused on acute infection, concentrating particularly on analyses of LT that we could obtain from experimental SIV infection in SMs. This was the focus because so many of the major pathogenic determinants in response to SIV infection that set the stage for disease outcome in RMs such as viral replication, CD4+ T cell depletion, and pathological damage to the LT niche, transpire early in the acute phase of infection within tissues (19, 20, 21, 26, 27, 28, 29).
We undertook longitudinal tissue-based analyses of serial lymph node biopsy before infection and days 7, 14, and 30 and 1.3 years (for SMs) after i.v. inoculation of five SMs with uncloned SIVsmm strain and five RMs infected with SIVmac239 strain. We further extended our analysis to chronic infection with analyses of LT samples obtained in cross-sectional studies, and also analyzed peripheral blood samples collected at frequent intervals in both acute and chronic infections to determine viral load and immunological parameters to directly compare LT and blood.
Given the extensive studies in chronic SM infection and our previous preliminary studies of acute SIV infection showing attenuated immune activation throughout all stages of SIVsmm infection (5, 6, 7, 17), we were somewhat surprised to find remarkably similar levels of generalized immune activation (as measured by total CD25 staining) and cell proliferation (total Ki67 staining) in LT within the first 2 wk of infection in all five SIVsmm-infected SMs compared with the five SIVmac239-infected RMs (Fig. 1, A–D). This apparent discrepancy with previous studies (17) can be explained by the fact that in this earlier study of experimental SIV infection of SMs, we were only able to examine a single lymph node sample, and thus did not appreciate the transient early phase of immune activation associated with acute infection in SMs.
However, whereas immune activation and proliferation in the TZ of LT continued to increase in RMs, there was a remarkable reduction in both activated CD25+ and proliferating Ki67+ cells by day 30 postinfection in SMs, with continued decreases in chronic infection, to levels still statistically significantly higher than baseline for Ki67 but much lower than those observed in RMs (Fig. 1, A and B). Flow cytometric analysis of specific T cell subsets from lymph node (Fig. 1,E, top) and peripheral blood (Fig. 1 E, bottom) were consistent with our immunohistochemical and QIA of LT, although the kinetics and peak of proliferating Ki67+ T cells based on proportions of cells was earlier and slightly more elevated in RMs compared with SMs for reasons we later discuss. These data immediately suggested and are consistent with the conclusion that it is the acute resolution of immune activation and cell proliferation that distinguishes SIV infection in SMs from SIV infection in RMs.
Resolution of immune activation in SMs results in preserved CD4+ T cell homeostasis and lower levels of apoptosis and fibrosis
We next investigated the benefits of acute resolution of immune activation to the SM host. First, the CD4+ T cell population in LT was severely depleted in chronically infected RMs but largely preserved in SMs (Fig. 2 A). The tissue analysis of the CD4+ T cell loss in RMs, but relative preservation in SMs, was also reflected in both the peripheral blood CD4+ T cell count and the frequency of CD4+ T cells within LT, in which there was a precipitous acute decline in CD4+ T cells in RMs that remained at low levels into chronic infection, whereas SMs showed only modest changes from baseline values (data not shown).
The link between resolution of early immune activation and maintenance of CD4+ T cells may reflect, at least in part, the different levels of apoptosis in LT (measured by active caspase 3 expression) in SMs and RMs, which mirrored the magnitude and kinetics of immune activation and proliferation (Fig. 2, B and C). At day 14, when RMs and SMs had equivalent levels of both CD25+ activated cells and Ki67+ proliferating cells in LT (Fig. 1), the levels of active caspase 3-positive cells were also similar. However, apoptosis in LT declined precipitously back to baseline levels in SMs by day 30, whereas in RMs levels continued to increase concomitant with immune activation (Figs. 1 and 2, B and C).
Second, because we have previously shown that collagen deposition and damage to the secondary LT architecture begins during the acute stage of infection, continues throughout the chronic stage of disease, and negatively impacts the CD4+ T cell population (19, 27, 28, 29), we asked whether early resolution of immune activation in SMs would prevent or moderate LT injury by this mechanism. We confirmed our previous findings of early and progressive deposition of collagen I in the TZ of secondary LT in all RMs, but found no evidence of increased collagen I deposition at any time point in the TZ of SMs (Fig. 3, A and B). Importantly, it is likely that the sparing effects on collagen deposition are directly related to resolution of immune activation and cellular proliferation, because of the strong positive correlation (Fig. 3 C) between the level of Ki67+ and CD25+ cells within LT and the amount of collagen deposition in the late acute and chronic stages of infected animals (R = 0.89; p = 0.0006).
Activation of immunosuppressive regulatory T cell pathways in SMs is not correlated with resolution
We then investigated potential immune regulatory pathways that might mediate resolution of immune activation in SMs in acute infection, focusing particularly on known highly immunosuppressive pathways (i.e., natural FoxP3+ regulatory T cells, IL-10+ Tr1, TGF-β+ Th3, and IDO expression) with kinetics that would be strongly associated with early resolution of immune activation in SMs but not RMs. Using Abs that stained TGF-β+, IL-10+, and IDO-positive cells in both SMs and RMs, we documented increases in all three cell types in the TZ early in RM LT (by day 14 in most animals), consistent with our previous findings (20), but we did not see either consistent increases in cells with these immunosuppressive molecules or temporal changes in their numbers consistent with a role in immune activation resolution in SMs (data not shown). Interestingly, at baseline, SMs had ∼5-fold more FoxP3+ cells compared with RMs (Fig. 4), suggesting that species differences may exist in homeostatic FoxP3+ regulatory T cell maintenance between RMs and SMs, but we found no consistent increase or association between the kinetics of FoxP3+ regulatory T cells that could explain early resolution of immune activation in SMs. In fact, in SMs, FoxP3+ cells actually decreased in LT during the period of immune activation resolution (Fig. 4). Collectively, these data demonstrated no clear relationship between the expression of highly immunosuppressive regulatory T cell populations and early resolution of immune activation in SMs.
Rapid, massive but transient up-regulation of PD-1 in the lymph nodes of SIV-infected SMs
We next sought to determine whether up-regulation of inhibitory receptors that attenuate TCR signaling in T cells might mediate resolution of immune activation in SMs during acute infection. We found a remarkably rapid and substantial increase in cells within the TZ expressing PD-1, a regulator of peripheral tolerance (30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40) increasingly implicated in CD8+ T cell exhaustion or dysfunction in persistent viral infections (41, 42, 43, 44, 45, 46, 47). PD-1+ cells increased several-fold by 14 days after infection, and were maintained at high levels throughout the acute phase in SMs (Fig. 5, A and B). The PD-1 response in SMs differed dramatically from RMs, in which there was no significant change in the frequency of PD-1+ cells in the TZ during the acute phase of infection. However, we found as others have previously documented in the chronic phase of SIV and HIV infections (41, 42, 43, 44, 45, 46, 47) substantial levels of PD-1+ cells in all chronically infected RMs (Fig. 5, A and B).
The focus on changes in LT and on changes in the absolute size of relevant populations of cells proved opportune, as FACS analysis of PD-1 expression in peripheral blood provided a far less clear picture of the potential mechanism responsible for immune activation resolution during acute infection in SMs. Using flow cytometry to ascertain the proportions of PD-1+ T cell populations, we did not detect increased levels of PD-1 expression in “bulk” (i.e., not necessarily SIV-specific) CD4+ or CD8+ T cells derived from the peripheral blood of SMs during acute infection or RMs in chronic disease (Fig. 5,C). In fact, peripheral blood analysis of T cells in RMs showed and initial rapid decrease in both CD4+PD-1+ and CD8+PD-1+ T cells that did not return to baseline levels until ∼day 300 postinfection (Fig. 5,C). Also inconsistent with our LT analysis in acute infection, we saw a decline in peripheral blood CD4+PD-1+ and no change in CD8+PD-1+ T cells in SIV-infected SMs. However, our LT studies were consistent with the peripheral blood results when analyzing PD-1 expression on T cells comparing SIV uninfected to chronically infected SMs, in which we found no significant difference in the proportions of CD4+PD-1+ or CD8+PD-1+ T cell populations between chronically infected SMs (n = 32) and SIV-negative SMs (n = 19) (Fig. 5, D and E). This observation highlights the differences that can exist in two important immunological compartments (peripheral blood and LT).
PD-1 up-regulation in lymph nodes of SMs likely involves predominantly CD8+ T cells
We next characterized the types of cells expressing PD-1 in the TZ by double label immunofluorescence confocal microscopy, and found that at peak (day 14) most of the PD-1+ cells in SMs cells were CD3+ T cells (Fig. 5,F, bottom left) and not B cells or macrophages (Fig. 5,F, top panels). Although PD-1+ cells in SMs consisted of both CD3+CD4+ and CD3+CD4− T cells, most of the PD-1+ cells were in fact CD4− T cells during acute infection in SMs and chronic infection in RMs (Fig. 5 F, bottom right, and data not shown). We were unable to directly stain the fixed tissue with Abs to CD8, but we think that the CD3+ PD-1+ CD4− T cells are most likely CD8+ T cells.
PD-1 up-regulation in SMs precedes the down-modulation of immune activation
Given the dramatic increase in PD-1 expression in the LT T cells of SMs during acute infection, we next sought to demonstrate that PD-1 expression is associated with reduced T cell immune responses and resolution of immune activation. Although at this time we do not have the reagents to ascertain SIV-specific T cell responses in tissues in SMs, the possibility that PD-1 expression on T cells in the acute phase of infection in SMs is involved in resolution of immune activation was supported by the observation of a concomitant reduction in the frequency of effector T cells (as measured by granzyme B expression) in SMs but not RMs. At day 14, a time point in which SMs had dramatically increased PD-1 expression and RMs still had low baseline levels of PD-1, the level of granzyme B-positive cells was >2-fold higher in RMs than in SMs (Fig. 6, A and B). Whereas activated effector granzyme B-positive T cells in SMs declined to baseline levels by day 30 postinfection and remained at these low levels during the chronic stage of disease, RMs from day 14 throughout the chronic stage of disease maintained high levels of effector granzyme B-positive T cells (Fig. 6, A and B). Importantly, the kinetics of granzyme B-positive effector cell induction within LT mirrored the kinetics of Ki67+ immune activation in both SMs and RMs (Fig. 6,C). Moreover, as further evidence that PD-1 expression on T cells attenuates effector function, we found that in SMs at peak levels (day 14) nearly all the PD-1+ cells within LT were granzyme B-negative (Fig. 6 D). Collectively, these data support the notion that early PD-1 expression on T cells attenuates effector function and is involved in acute resolution of immune activation in SMs.
Correlations between viral replication, PD-1 up-regulation and the resolution of immune activation in SMs but not in RMs
RMs and SMs had equivalently high levels of SIV RNA-positive productively infected cells in LT (Fig. 7). In both species, CD25+, Ki67+, and granzyme B-positive effector cells increased in parallel with viral replication, but in SMs levels decreased in parallel with increased expression of PD-1 that followed shortly after the peak of viral replication. In RMs, however, there was no increase in PD-1 and levels of effector cells remained high from acute through the chronic stages of infection. These high levels of effector cells were sustained in RMs even though levels of viral replication had substantially declined at this time and PD-1 levels had increased (Fig. 8). Thus, the timing and relationships between viral replication, immune activation, and PD-1 expression were profoundly different in the two species, with a clear relationship with viral replication, early increased expression of PD-1, and the resolution of immune activation in SMs but not RMs.
We used linear regression models to assess the level of statistical significance of the apparent relationship between PD-1 and immune activation, and found a main effect for time (p = 0.0003), the interaction between PD-1 and species (p = 0.0006), and the interaction between species and time (p = 0.0007). The value for p the F test for the entire model was 5.206 × 10−5 with an R2 value of 0.40. These data strongly support the notion that early PD-1 expression in SMs, but not in RMs, impacts the level of immune activation, proliferation, and effector response.
The objective of the experiments reported in this study, like previous studies we and others have undertaken (19, 48, 49, 50), was to try to understand through comparative analysis why SIV infection in a natural host generally does not cause generalized CD4+ T cell depletion and AIDS despite levels of virus replication that approximate levels in pathogenic SIV infections in non-natural hosts (i.e., Asian macaques) and HIV-1 infection in humans. Because previous studies have clearly established that lower levels of immune activation and T cell proliferation are a major distinguishing feature of nonpathogenic infections (1, 5, 17, 22, 23, 24, 25), we investigated the potential underlying mechanisms of this difference by conducting a comparative analysis of the early stages of nonpathogenic infection of SMs and pathogenic SIV infection of RMs. By concentrating on the initial stages of infection and studying LT at multiple time points within the first 30 days of infection, we were able to uncover a significant, albeit transient, phase of immune activation and proliferation in SMs that was not detected in previous studies in which a more limited sampling was performed (17). In this study, we found that the initial immune activation and T cell proliferation were actually comparable between SMs and RMs in terms of CD25 and Ki67 expression, although the fraction of effector (i.e., granzyme B-positive) T cells was higher in RMs during both acute and chronic infection. Interestingly, the most striking difference between the two species was the rapid and substantial resolution of this state of immune activation in SMs but not in RMs, an event that appears to be sufficient to protect SMs from subsequent CD4+ T cell depletion, chronic T cell apoptosis, and LT niche-damaging fibrosis, i.e., all phenomena that contribute to disease progression during pathogenic HIV and SIV infections (19, 27, 28, 29).
The quantitative comparative analysis of changes in LT in response to SIV infection in its earliest stages in natural and non-natural hosts turned out to be critical in recognizing early immune activation and resolution in SMs. Although some of the changes in blood and LT documented in our FACS analysis in this experiment were similar to the tissue analysis, a clear picture of the rapid resolution of acute immune activation and in particular the relationship to PD-1 expression were not apparent in peripheral blood. These data highlight the differences in peripheral blood and LT, possibly related to altered lymphocyte trafficking in infection, which point to the need for LT analysis during this dynamic window of infection.
Because PD-1 is known to play an important role in resolving immune activation and T cell proliferation (30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40), we investigated PD-1 expression and discovered that a critical difference between nonpathogenic and pathogenic SIV infections is the rapid increase in PD-1 expression in SMs, particularly in CD4− (CD8+) T cells, which parallels the decreased T cell proliferation and effector functions associated with immune activation. Notably, there was a strong positive association between the level of immune activation over time (CD25+ activated, Ki67+ proliferating, and granzyme B-positive effector cells) and the level of PD-1 expression in LT in SMs, whereas the increase in the level of immune activation over time was not affected by the level of PD-1 expression in RMs (Fig. 6). Collectively, these data strongly support a role for early induction of PD-1 expression on CD8+ T cells in the resolution of acute immune activation in SIV infection.
Although the mechanisms are presently unknown, our findings also point to critical differences in PD-1 function in tolerance vis-à-vis CD8+ T cell dysfunction related to the timing, magnitude, and duration of PD-1 expression. We show in this study that early transient elevation of PD-1 expression in SMs is associated with the resolution of immune activation, whereas the delayed induction and then sustained expression of PD-1 at high levels in RMs did not “shut down” T cell activation after the acute window of infection, with untoward consequences for immune function persistent in SIV and HIV infections and other chronic viral infections of CD8+ T cell “exhaustion” and dysfunction (41, 42, 43, 44, 45, 46, 47).
The mechanisms underlying these differences in the timing of PD-1 expression and relationship to viral replication in nonpathogenic vs pathogenic SIV infections also need further investigation. In SMs, the kinetics and levels of PD-1 expression initially paralleled the frequency of SIV RNA-positive cells in LT, peaking around day 14 postinfection, and then declined by the chronic stage of infection after the decline in viral replication. By contrast, in RMs, increased PD-1 expression was not linked to peak viral replication, occurred at a time when viral loads were at lower set point levels and when infected cells in LT were dramatically reduced, and was associated with the sustained immune activation in RMs. Thus, although the mechanism of induction of PD-1 expression in SMs is directly related to virus replication, the mechanism of induction in RMs appears to be indirectly mediated by mechanisms associated with chronic immune activation.
It is tempting to speculate that decreased immune activation mediated by early PD-1 expression in SMs may also represent an important cause of the lower viral set point levels compared with RMs in this study by reducing the total number of activated cellular substrates of SIV. Thus, lowering immune activation early after infection not only attenuates the immunopathogenic effects driven by persistent chronic immune activation (i.e., CD4+ T cell loss, collagen deposition, and others) but also reduces the availability of the major SIV/HIV target cell in chronic disease.
Although the timing of the immunoregulatory responses appears to be crucial for protection from disease in nonpathogenic SIV infections in natural hosts, the mechanisms responsible for this phenotype may differ among natural host species. In the case of SIV infection of African green monkeys, an early regulatory T cell response may mediate a disease-sparing attenuation of immune activation (48), whereas we have shown that a later regulatory T cell response before clearing infection suppresses the SIV-specific immune response but does not dampen generalized immune activation in RMs (20). In this study, we show that an early PD-1 response may limit immune activation in SMs and that the later PD-1 response in RMs may further contribute to a dysfunctional immune response ultimately leading to disease and death. Further study on immunosuppressive mechanisms that attenuate immune activation in SMs but not in RMs is actively being pursued by our laboratories.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Grants R01 AI48484 and AI056997 (to A.T.H.), Grants R01 AI66998 and AI52755 (to G.S.), Grant R01 HL75766 (to S.S.), Grant T32 AI07421 (to J.D.E.) from the National Institutes of Health, and by Grant RR0165 from the Yerkes National Primate Research Center. This project has been funded in part with federal funds from the National Cancer Institute, National Institutes of Health under contract N01-CO-12400.
The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. government.
Abbreviations used in this paper: RM, rhesus macaque; SM, sooty mangabey; TZ, T cell zone; QIA, quantitative image analysis; LT, lymphatic tissue; PD-1, programmed cell death-1.