Diminished Viral Control during Simian Immunodeficiency Virus Infection Is Associated with Aberrant PD-1^hi CD4 T Cell Enrichment in the Lymphoid Follicles of the Rectal Mucosa

This article is featured in In This Issue, p.4275

The humoral and cellular immune responses are critical for the control of HIV and SIV infections. The CD4 T cells play a key role in regulating the magnitude and function of humoral and cellular immunity (1–5). HIV preferentially infects virus-specific CD4 T cells, with memory CD4 T cells being the primary target of HIV infection (1, 2). During acute HIV/SIV infection, massive depletion of memory CD4 T cells occurs predominantly at mucosal sites, with over one-half of all memory CD4 T cells in SIV-infected rhesus macaque (RM) being destroyed directly by viral infection. Virus-specific CD8 T cells are induced during acute infection and are important in the containment of viral replication (4, 5). CD4 T cell help has also been shown to play a vital role in the control of HIV infection because individuals capable of controlling virus to low or undetectable levels maintain a high frequency of HIV-specific CD4 T cells with high functional avidity (6–8). In addition, depletion of CD4 T cells during acute SIV infection leads to abrogation of initial postpeak viral decline (9).

In the setting of chronic infection, T cells have been shown to upregulate the inhibitory receptor programmed death-1 (PD-1), as well as other inhibitory receptors such as CTLA4, LAG-3, Tim-3, and 2B4 (10–15). Sustained expression of these inhibitory receptors has been associated with immune dysfunction in murine (19, 20), nonhuman primate (16–20), and human model systems (11, 12, 20, 21). In the context of chronic HIV and SIV infections, it has been well established that there is an appreciable increase in both the frequency and expression of PD-1 on antiviral CD8 T cells and a preferential depletion of PD-1⁺ B cells. PD-1⁺ Ag-specific CD8 T cells exhibit impaired proliferation, decreased Ag specific cytokine production, and compromised survival (16, 17, 22, 23). Alternatively, in vivo blockade of PD-1 enhances antiviral CD8 T cell function and viral control (19, 22, 24). Despite the comprehensive characterization of PD-1 on CD8 T cells during chronic SIV/HIV infection, the role of PD-1 on CD4 T cells has received far less attention in the context of viral infection, specifically in sites of preferential viral replication.

Preliminary studies of PD-1 on CD4 T cells during chronic HIV infection have shown that the frequency of PD-1⁺ CD4 T cells in the blood correlates with plasma viral load and decreased CD4 T cell counts and that subsequent in vitro PD-1 blockade of PBMCs can augment proliferative capacity of virus-specific CD4 T cells (13, 25). It is known that follicular helper CD4 T (Tfh) cells in the lymphoid tissue express high levels of PD-1 (26–28). Recent studies have demonstrated that the frequency of PD-1^hi Tfh cells increase significantly in lymph nodes (LNs) of HIV-infected humans and SIV-infected nonhuman primates (NHPs) during the chronic stage (29–32). The reasons for this increase are not yet fully understood. Although human studies suggested a direct relationship between the frequency of PD-1⁺ or Tfh cells and plasma viremia, this association was not observed in NHP studies. Petrovas et al. (29) demonstrated a direct relationship between higher soluble CD14 levels in plasma and the frequency of Tfh cells, suggesting a role for microbial translocation in the gut in regulating Tfh cells in the lymphoid tissue. However, there is no information available on the status of PD-1^hi CD4 T cells in the gut, a preferential site of virus replication in HIV-infected humans or SIV-infected NHPs, and a site that is constantly exposed to high levels of pathogenic and nonpathogenic bacteria. In addition, it is not clear whether vaccine-elicited CD8 T cells have any effect on PD-1^hi or Tfh cells in the LN and rectum following SIV infection.

In this study, to understand the influence of chronic SIV infection on PD-1⁺ CD4 T cells in the gut of RM, we studied the PD-1 expression on CD4 T cells in the rectal mucosal tissue (rectum) and compared it with LNs in the context of SIV-naive, chronic uncontrolled SIV infection, and vaccine-mediated controlled SIV infection. Our results showed a preferential increase in the frequency of PD-1^hi CD4 T cells in the rectum and LN of uncontrolled SIV infection and revealed important differences between rectal mucosa and LNs.

Young adult RMs from the Yerkes breeding colony were cared for under the guidelines established by the Animal Welfare Act and the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals using protocols approved by the Emory University Institutional Animal Care and Use Committee. Noncontrollers were either unvaccinated or received a DNA/MVA SIV vaccine, and all vaccine controllers received the DNA/MVA SIV vaccine (33, 34). All animals were infected with SIVmac251 intrarectally.

Indian-origin RMs (Macaca mulatta) were unvaccinated or vaccinated with DNA/MVA SIV (DM) vaccine. Vaccination consisted of two DNA primes on weeks 0 and 8 and two MVA boosts on weeks 16 and 24. Both DNA and MVA immunogens expressed SIV239 Gag, Pol, and Env as described previously (35) Vaccinated animals either received DM vaccine alone (n = 3), the CD40L adjuvant during DNA prime and MVA boosts (DM40L) (n = 7), or rapamycin for 28 d during each of the MVA boosts (DMRapa) (n = 19). All animals were challenged weekly with SIVmac251 starting 21–24 wk after the final MVA immunization with a dose of 647 TCID₅₀ (1.25 × 10⁷ copies of viral RNA) until they were productively infected. All animals were infected by seven challenges under these conditions. Please see Supplemental Table 1 for additional details. Dr. N. Miller (NIH, Bethesda, MD) provided the challenge stock. The criteria for defining controllers and noncontrollers were based on a plasma viral load cutoff of 10⁴ RNA copies/ml plasma at 24 wk postinfection (36). SIV RNA levels were determined using a quantitative PCR (37).

Mononuclear cells were isolated from the blood, axillary LN, and rectal tissue, and flow cytometry analysis was performed as described previously (16). Briefly, PBMCs were isolated from blood collected in sodium citrate cell preparation tubes (BD Biosciences). Mononuclear cells from the LN were isolated from axillary and mesenteric LNs processed in complete medium and treated with ammonium–chloride–potassium lysing buffer to remove residual RBCs. Mononuclear cells from the rectum were isolated after tissue was digested for 2 h in complete medium with 10% FBS, 1% penicillin/streptomycin, 0.05% gentamicin, 1% HEPES, 200 U/ml collagenase IV (Worthington, Lake Wood, NJ), and DNase I (Roche, Indianapolis, IN). Digested tissue was then passed through decreasing size needles (16-, 18-, and 20-gauge, five to six times).

The following Abs were used: FITC-conjugated Bcl-2 (clone Bcl-2/100; BD Biosciences), PE-conjugated CXCR5 (clone MU5UBEE; eBioscience), PerCP-conjugated CD3 (clone SP-34-2; BD Biosciences), PeCy7-conjugated CD28 (clone CD28.2; eBioscience), PE-Texas Red–conjugated CD95 (clone DX2; BD Biosciences), Brilliant Violet 421–conjugated CD279 (PD-1; clone EH12.1; BioLegend), V500-conjugated CD8 (clone SK1; BD Biosciences), allophycocyanin-conjugated CCR5 (clone 3A9; BD Biosciences), Live Dead-IR stain (Invitrogen), Alexa 700–conjugated Ki-67 (clone B56; BD Biosciences), FITC-conjugated Bcl-6 (clone K112-91; BD Biosciences), Brilliant Violet 650–conjugated CD4 (clone OKT4; BioLegend), FITC-conjugated IL-17A (clone eBio64Dec15; eBioscience), PE-conjugated IL-21 (clone 3A3-N2; BD Biosciences), PerCP-conjugated CD4 (clone L200; BD Biosciences), PeCy7-conjugated anti-CD279 (PD-1; clone EH12.1; BioLegend), PacBlue-conjugated CD3 (clone SP-34-2; BD Biosciences), allophycocyanin-conjugated IL-2 (clone MQ1-17H12; BD Biosciences), and Alexa 700–conjugated IFN-γ (clone B27; BD Biosciences).

Fresh blood, LN, and rectal samples were suspended in RPMI 1640 medium (Life Technologies) with 10% FBS (HyClone, Thermo Fisher Scientific), 100 IU/ml penicillin, and 100 μg/ml streptomycin (Lonza). Stimulations were conducted in the presence of anti-CD28 Ab and anti-CD49d Ab (1 μg/ml; BD Pharmingen). One million cells were stimulated with either 200 ng/ml PMA and 1 μg/ml ionomycin, CD3/CD28 beads (at a 1:2 ratio beads to cells; Miltenyi Biotec) or pooled peptides spanning the entire SIV Gag protein (single pool of 125 peptides with each peptide at a concentration of 1.0 μg/ml; NIH AIDS Research and Reference Reagent Program catalog number 6204) in the presence of brefeldin A (5 μg/ml; Sigma-Aldrich, St. Louis, MO) and GolgiStop (0.5 μl/ml; BD Pharmingen) after 2 h of stimulation for 4 h at 37°C in the presence of 5% CO₂. At the end of stimulation, cells were washed once with FACS wash (PBS containing 2% FBS and 0.25 g sodium azide) and surface stained with anti-CD3, anti-CD8, anti-CD95, and anti-CD279 (PD-1) at room temperature for 20 min. Cells were then fixed with cytofix/cytoperm (BD Pharmingen) for 20 min at 4°C and washed with Perm wash (BD Pharmingen). Cells were then incubated for 30 min at 4°C with Abs specific to IL-2, IL-17A, IFN-γ, IL-21, and CD4, washed once with Perm wash, once with FACS wash, and resuspended in PBS containing 1% formalin. Cells were acquired on LSR-Fortessa with four lasers (205, 288, 532, and 633 nm) and analyzed using the FlowJo software (Tree Star). At least 50,000 events were acquired for each sample.

Mononuclear cells isolated from the blood, LN, and rectum were stained with Live/Dead Near-IR Dead Cell stain (Life Technologies) at room temperature for 15 min in PBS to stain for dead cells. Cells were then washed with FACS wash and stained on the surface using Abs specific to CD3, CD4, CD8, CD28, PD-1, CD95, CXCR5, and CCR5 and then treated with 1× BD FACS Lysing solution for 10 min at room temperature, permeabilized with 1× BD Permeablizing solution for 10 min at room temperature, washed with FACS wash, stained with anti-Ki67 and anti–Bcl-2 Abs, washed twice with FACS wash, and assessed by flow cytometry.

Rectal tissues were fixed in SafeFix (Fisher Scientific) and embedded in paraffin. Embedded tissue blocks were cut into five microsections, deparaffinized, and rehydrated for immunohistochemical analysis. Some tissues also embedded unfixed in OCT medium (TissueTek, Sakura, Finetek, Torrance, CA) for CD8 staining. Immunofluorescence staining was performed for CD20, CD4, CD8, and PD-1 to examine the distribution of PD-1^hi cells in tissues, as described previously (38). In brief, heat-induced epitope retrieval was performed with DIVA Decloaker and then blocked with the SNIPER reagent (Biocare, Walnut Creek, CA) for 15 min and in PBS/0.1% Triton X-100/4% donkey serum for 30 min at room temperature. Subsequently, the sections were incubated with rabbit anti-human CD20 (Thermo Scientific, Rockford, IL), mouse anti-human CD4 (clone BC/1F6; Abcam, Cambridge, MA), mouse anti-human CD8 (clone LT8; Abcam, Cambridge, MA), and goat anti-human PD-1 (R&D Systems, Minneapolis, MN) Abs diluted 1:20 to 1:100 in blocking buffer for 1 h at room temperature. Thereafter, the sections were incubated with secondary Abs (Alexa Fluor 488/Cy3/Cy5–conjugated appropriate donkey anti-mouse/rabbit/rat/goat Abs; Jackson ImmunoResearch, West Grove, PA) diluted 1:1000 in blocking buffer for 30 min at room temperature. For the frozen tissues, 5-μm-thick sections were fixed with 4% paraformaldehyde for 10 min, followed by washing in 1× TBS buffer (Biocare Medical, Concord, CA). The same process was then performed without heat-induced epitope retrieval. Finally, the sections were mounted in warm glycerol gelatin (Sigma-Aldrich) containing 4 mg/ml n-propyl gallate (Fluka, Switzerland). Between each step, the sections were washed three times. All images were acquired and analyzed with an Axio Imager Z1 microscope (Zeiss) using various objectives.

Mononuclear cells isolated from the LN and rectal tissue were processed and stained with anti-CD3, anti-CD279 (PD-1), anti-CD95, and anti-CD8 for 25 min at 4°C and the CD95⁺PD-1^neg, CD95⁺PD-1^int, and CD95⁺PD-1^hi, and CD95⁻ (naive) CD4 T cell populations were sorted using a FACSAriaII (BD Biosciences). In all sorting experiments, the grade of purity on the sorted cells was >93%. SIV RNA levels were determined using a quantitative PCR (37).

Mononuclear cells isolated from the LNs of SIV-infected Mamu A*01⁺ SIV controller RMs were processed, stained with Live/Dead IR, anti-CD3, anti-CD4, anti-CD8, anti-CD95, and anti-CXCR5 Abs, and sorted for CD95⁺ CD8 T cells and CD95⁺CXCR5^hi CD4 T cells (Tfh cells) using a FACSAriaII (BD Biosciences). Tfh cells were then pulsed with P11c peptide for 1 h at 37°C at a concentration of 0.1 μg/ml and washed. CD8 T cells were cocultured with unpulsed or pulsed Tfh cells at a 2:1 ratio of CD8 T cells to Tfh cells with no stimulation or anti-CD3/CD28 stimulation at one bead to two cells (Miltenyi Biotec) for 5 d. Cells were then harvested and analyzed using flow cytometery.

Statistical analyses were performed using Prism (version 5.0d; GraphPad Software). Statistical significance (p values) was obtained using non-parametric Mann–Whitney U test (for comparisons between groups/subsets) or Spearman rank test (for correlations). Statistical analyses of global cytokine profiles were performed by partial permutation tests using SPICE software (National Institute of Allergy and Infectious Diseases, NIH) as described previously (39).

To understand the role of PD-1 on CD4 T cells during chronic SIV infection, we characterized PD-1 expression in the rectum compared with the LN and peripheral blood of SIV-naive RM. We observed three subsets of PD-1 expressing memory (CD95⁺) CD4 T cells namely PD-1^neg, PD-1^int, and PD-1^hi cells (Fig. 1A). Interestingly, the PD-1^hi CD4 T cells were present predominantly in the rectum and LN, with 2–4% in the LN and 8–12% in the rectum and <1% in the blood (Fig. 1A). Thus, PD-1^hi memory CD4 T cells are enriched at sites of preferential SIV replication.

Because Tfh cells in the LN are known to express high levels of PD-1, we phenotyped the PD-1^hi CD4 T cells in the rectum for CXCR5 and Bcl-6, markers used to define Tfh cells in the LN (Fig. 1B). Interestingly, the majority of PD-1^hi but not PD-1^int and PD-1^neg memory CD4 T cells in the rectum and LN expressed CXCR5 and Bcl-6, suggesting that PD-1^hi cells in the rectum phenotypically may predominately be Tfh cells. We also accessed PD-1 subsets for the level of CCR5 coreceptor expression. In contrast to PD-1^int cells that expressed high levels of CCR5, the majority of PD-1^hi cells in the rectum and LN did not express the viral coreceptor CCR5, suggesting that these may not be ideal targets for the virus (Fig. 1C).

We next investigated the influence of SIV infection on PD-1^hi CD4 T cells in the rectum and LN of unvaccinated and vaccinated animals. For this purpose, we used samples from a cohort of SIV-infected RMs that were either vaccine controllers (<10⁴ RNA copies/ml plasma at week 24 postinfection) or noncontrollers (>10⁴ RNA copies/ml plasma at week 24 postinfection, which include both unvaccinated and vaccinated animals) (Supplemental Fig. 1A). Please note that all conclusions made below remained true even if we defined controllers based on a set point viral load of less than 10³ RNA copies/ml. As expected, the frequency of total CD4 T cells was significantly lower in the rectum of noncontrollers compared with vaccine controllers (Supplemental Fig. 1B). However, the frequency of PD-1^hi cells within the memory CD4 T cell compartment was dramatically higher both in the rectum and LN of noncontrollers compared with SIV-naive and vaccine controllers (Fig. 2A). The frequency of PD-1^hi cells in the vaccine controllers was comparable to the uninfected RMs (Fig. 2A). To get an estimate of the cell number, we expressed the frequency of PD-1^hi cells as a percentage of total lymphocytes and found a significant increase in PD-1^hi CD4 T cells in the LN as a percentage of lymphocytes (Supplemental Fig. 1C), but this was not observed in the rectum (Supplemental Fig. 1B). This finding suggests that despite a decrease in total memory CD4 T cells in the GALT, PD-1^hi CD4 T cells remain enriched at this site. Consistent with the increase in the frequency of PD-1^hi cells as a percentage of memory, the mean fluorescence intensity (MFI) of PD-1 was higher on memory CD4 T cells in the noncontrollers than in uninfected and vaccine controllers (data not shown). Furthermore, the frequency of PD-1^hi cells in the LN and rectum correlated directly with plasma viremia (Fig. 2C).

To understand the kinetics of expansion of PD-1^hi cells following infection, we followed the frequency of PD-1^hi cells longitudinally in the rectum of SIV noncontrollers in a separate study and found a similar increase in PD-1^hi cells during the course of SIV infection, with the increase being observed as early as 2 wk postinfection (Fig. 2C). These data demonstrated that despite the loss of total memory CD4 T cells, the PD-1^hi memory CD4 T cells are enriched at preferential sites of virus replication in uncontrolled chronic SIV infection very early postinfection, whereas interestingly such enrichment is not seen in vaccine controllers. We also characterized PD-1 expression in the jejunum of a small group of chronically SIV-infected animals, but we did not find PD-1^hi CD4 T cells (data not shown) at this site. There are two possible explanations for this observation: PD-1^hi CD4 T cells are depleted from the jejunum during chronic SIV infection or PD-1^hi CD4 T cells are not present in the jejunum because of a limited number of GALT structures in this region (40, 41).

Next, we investigated the expression of CXCR5 on the PD-1^hi cells following SIV infection in the rectum and LN to understand their Tfh phenotype and localization. In contrast to SIV-naive animals, a significant fraction of PD-1^hi cells in the rectum did not express the Tfh marker CXCR5 following SIV infection (Fig. 3A). This was true for both noncontrollers and controllers, except that it was more pronounced in controllers. However, similar to SIV-naive animals, the majority of PD-1^hi cells in the LN expressed CXCR5, although there was a small decrease in the controllers (Fig. 3A). Although we observed a decrease for CXCR5 expression on PD-1^hi cells in the rectum of noncontrollers, because the majority of memory CD4 T cells were PD-1^hi (Fig. 2A), the overall frequencies of CXCR5⁺ and CXCR5⁻ PD-1^hi cells within the memory CD4 T cell compartment was also higher in the noncontrollers compared with uninfected RMs (Fig. 3B). Consistent with this increase in CXCR5⁺ PD-1^hi cells in the noncontrollers by flow cytometry, immunofluorescence analysis of rectal tissue revealed B cell follicles with significantly higher density of PD-1^hi CD4 T cells in noncontrollers compared with controllers (Fig. 3C, 3D) (38). We phenotyped CXCR5⁻ and CXCR5⁺ cells for the expression of Bcl-6 in a limited number of SIV-infected RM (Fig. 3E). These analyses revealed that in the LN CXCR5⁻ cells express lower levels of Bcl-6 compared with CXCR5⁺ cells; however, interestingly in the rectum, both the subsets seem to express Bcl-6 at similar levels. These results argue that the phenotype of Tfh cells in the rectum could be different from that of LN during chronic SIV infection and a thorough characterization of their localization and function is critical before they can be classified as Tfh on the basis of CXCR5 and Bcl-6 expression.

The enrichment of PD-1^hi cells in noncontrollers could be because of increased proliferation of these cells while maintaining their survival potential, so we studied the expression of Bcl-2 (antiapoptotic protein) and Ki-67 (marker for proliferating cells) ex vivo. The PD-1^hi memory CD4 T cells in the rectum and LN of noncontrollers showed either comparable or higher levels of Bcl-2 expression (Fig. 4A, Supplemental Fig. 2) and markedly enhanced Ki-67 expression compared with uninfected animals (Fig. 4B). However, this was also true for vaccine controllers, suggesting that the observed higher proliferation or Bcl-2 expression of PD-1^hi CD4 T cells alone did not markedly contribute to their enrichment in noncontrollers. These results demonstrated that the uncontrolled chronic SIV infection is associated with an enrichment of PD-1^hi cells at preferential sites of virus replication with preserved survival potential and high proliferation status.

To assess the cytokine production capability of these PD-1^hi memory CD4 T cells, we stimulated cells isolated from the rectum and LN with either PMA/ionomycin (non–TCR-driven cytokine production) or anti-CD3/CD28 (TCR-driven cytokine production) (Fig. 4C). In general, a significant fraction of PD-1^hi cells failed to produce cytokines following anti-CD3/CD28 stimulation (data not shown). However, following stimulation with PMA/ionomcyin, a significant fraction of PD-1^hi cells in the uninfected animals produced cytokines IFN-γ, IL-2, and IL-21 (Fig. 4C) and a small fraction produced IL-17 (data not shown). In the LN, they produced predominantly IL-2, followed by IL-21, IFN-γ, and IL-17. However, in the rectum, they produced predominantly IL-21 and IL-2, followed by IFN-γ. PD-1^hi cells in the noncontrollers and vaccine controllers largely maintained IL-21 production but noncontrollers showed decreased production of IL-2 and IFN-γ. However, this defect was not observed in the rectum (Fig. 4C). These results demonstrated that the PD-1^hi cells that accumulate at the preferential sites of virus replication during uncontrolled SIV infection maintain the potential to produce IL-21, which may contribute to hypergammaglobulinemia by aiding in the maintenance and proliferation of memory B cells. PD-1^hi cells show decreased production of IL-2 and IFN-γ during chronic uncontrolled SIV infection, possibly limiting the potential for the generation of functional Ag-specific humoral responses at these sites. The failure of these cells to express cytokines following TCR-driven stimulation could be because of inhibition by PD-1 signaling and needs further investigation.

We then assessed the expression of the viral coreceptor CCR5 on these cells to see whether these cells can be preferentially infected and killed by the virus (Fig. 5A). In general, only a small fraction (2–5%) of PD-1^hi cells expressed CCR5 in the uninfected animals, and these levels were even lower in noncontrollers. Interestingly, the fraction of PD-1^hi cells expressing CCR5 increased dramatically in the rectum of vaccine controller compared with SIV naive and noncontroller RMs. As expected, the overall frequency of CCR5⁺ total memory CD4 T cells declined in the rectum and LN of SIV-infected noncontrolling animals; however, this was not evident in controllers (Fig. 5B).

In an attempt to better understand the contribution of these PD-1^hi cells to viral production and persistence during chronic SIV infection, we further studied the infection status of these cells in the rectum and LN of noncontroller animals (Fig. 5C). The levels of viral RNA were significantly higher in PD-1⁺ cells than in PD-1–negative cells. Within PD-1⁺ CD4 T cells, viral RNA was present in both PD-1^int and PD-1^hi cells. To approximate the production of virus on a per cell basis, we determined the ratio of viral RNA to viral DNA and observed a significantly higher ratio in PD-1^hi cells in the rectum compared with PD-1^int and PD-1^neg cells (Fig. 5C). Cell-associated viral RNA and DNA were also found predominately in the PD-1⁺ CD4 T cell subsets in vaccine controllers compared with noncontrollers, albeit at 50- to 100-fold lower levels likely because of lower plasma viremia (data not shown). These data demonstrate that a significant fraction of PD-1^hi cells in the rectum are productively infected during uncontrolled SIV infection and actively support viral production in lymphoid sites known to highly contribute to viral persistence.

To understand the relationship between antiviral CD8 T cells and the frequency of PD-1^hi cells, we determined the frequency of Gag CM9 tetramer–specific CD8 T cells in Mamu A01⁺ animals of our controllers and noncontrollers (Fig. 6). Previous studies estimated that a significant fraction of the total SIV-specific CD8 T cell response in Mamu A01 animals is directed to this single epitope during the chronic phase (42), and the use of tetramer allows us to determine frequency of CD8 T cells independent of function. These properties make it easier to interpret the data. The frequency of tetramer⁺ CD8 T cells was significantly higher in the LNs of controllers than in noncontrollers (Fig. 6A), and interestingly, this difference was not observed in the blood (Fig. 6A), suggesting that LN-resident CD8 T cells could have contributed to enhanced control in these animals. Furthermore, the frequency of granzyme B⁺tetramer⁺ CD8 T cells was also higher in controllers than in noncontrollers (Fig. 6A). In addition, the frequency of tetramer⁺ or granzyme B⁺tetramer⁺ CD8 T cells correlated inversely with the frequency of total PD-1^hi and CXCR5+ PD-1^hi CD4 T cells (Fig. 6B). These results strongly suggested that LN-resident highly functional antiviral CD8 T cells might limit the aberrant expansion of PD-1^hi as well as CXCR5⁺PD-1^hi CD4 T cells during chronic SIV infection. The antiviral CD8 T cells could mediate this effect either directly by killing the infected PD-1^hi CD4 T cells or indirectly by controlling the virus replication that in turn may limit their expansion. To gain more insight into the mechanism, we cocultured antiviral CD8 T cells and Gag CM9 peptide–pulsed CXCR5⁺ PD-1^hi CD4 T cells obtained from three controller RMs for 5 d and found that in two of three animals antiviral CD8 T cells limit the expansion of the CXCR5⁺PD-1^hi CD4 T cells (Fig. 6C). Interestingly, the ability of CD8 T cells to limit the expansion of CD4 T cells was associated with higher frequency of Gag CM9–specific CD8 T cells at baseline in vitro (Fig. 6C). In conclusion, these results strongly suggest that LN-resident antiviral CD8 T cells have the potential to kill the CXCR5⁺PD-1^hi CD4 T cells and highlight a role for these cells in limiting the aberrant expansion of PD-1^hi CD4 T cells in the vaccinated controllers in vivo.

Depletion of CD4 T cells in the GALT is an important hallmark of SIV/HIV infection, and recent studies have sought to investigate the immune responses and cellular populations affected in the mucosal sites during pathogenic infection. In the early 1980s, germinal centers were considered as potential sites for long-lasting viral reservoirs and sanctuaries for viral recrudescence (43), but many questions were left unanswered. Recently, germinal centers of lymphoid sites and, in particular, CD4+ Tfh cells in the LN have emerged as an important population of interest during chronic HIV/SIV infection. In particular, several groups have described an increase in the frequency of CXCR5⁺PD-1^hiBcl-6⁺ Tfh cells in the LN during chronic SIV/HIV infection, and these cells have been speculated to significantly contribute to B cell dysfunction and hypergammaglobulinemia observed during chronic infection (29–31, 44). Despite the existing knowledge of Tfh cells in the LN during chronic SIV/HIV infection, the existence of these cells and their contribution to SIV pathogenesis have not been studied in the mucosal tissue, one of the most important sites of preferential viral replication and persistence.

Our data demonstrate that lymphoid follicles in the rectum of RMs with uncontrolled chronic SIV infection are highly enriched in actively proliferating PD-1^hi CD4 T cells that retain survival potential and harbor a significant fraction of virus-infected cells. This enrichment is impressive considering the widespread depletion of memory CD4 T cells in the gastrointestinal tract following SIV infection. Furthermore, a significant fraction of these PD-1^hi cells reside in B cell follicles of lymphoid aggregates in the rectum. Strategically, this seems to be an important mechanism by which these virus-infected PD-1^hi CD4 T cells can contribute to ongoing viral replication and persistence while avoiding antiviral CD8 T cell responses, as it has been shown that germinal centers of LNs may act as viral sanctuaries during SIV infection. In addition, these cells maintained their ability to produce IL-21, which may support the uncontrolled or constant proliferation of memory B cells leading to B cell dysfunction and hypergammaglobulinemia commonly seen in chronic SIV and HIV infections (44).

With little known about PD-1^hi CD4 T cells residing in the mucosal tissue of SIV naive and SIV-infected RMs, our study revealed some similarities and differences in immune responses in the rectal mucosa compared with the LN. In the absence of SIV infection, PD-1^hi memory CD4 T cells are present at higher frequencies in the rectum compared with the LN. This higher frequency in the rectum is most likely due to constitutive germinal centers present in the GALT as a result of continued sampling of microbial Ags, which stimulate differentiation and maintenance of CD4 Tfh cells (45–47). Moreover, during chronic SIV infection, PD-1^hi cells increased as a frequency of total memory CD4 T cells in both the LN and rectum. This may be because of either preferential depletion of PD-1^int and PD-1^neg cells or preferential differentiation to PD-1^hi cells. Nevertheless, this happens in both compartments (LN and rectum). Although we only directly observed an enrichment of PD-1^hi CD4 T cells in the rectum during chronic uncontrolled SIV infection, we speculate that we would find a similar enrichment in other mucosal sites known to have a high density of GALT structures, such as the terminal ileum (41). In contrast to the LN, there was not an increase in the frequency of PD-1^hi cells as a percentage of total lymphocytes in the rectum. This may likely be because of higher depletion of memory CD4 T cells in the GALT compared with LN during chronic uncontrolled HIV/SIV infection.

The increased frequency of PD-1^hi memory CD4 T cells in the LN and rectum both consistently associated with viral control, but interestingly we observed an increase in the frequency of CCR5⁺PD-1^hi CD4 T cells and decrease in CXCR5⁺PD-1^hi CD4 T cells in the rectum of vaccine controllers, an observation much less apparent in the LN. We can speculate that altered chemokine receptor expression may allow for differential homing of these PD-1^hi CD4 T cells to regions close to the periphery of the germinal center or outside the B cell follicle. This change in localization may allow for increased immune pressure directed at PD-1^hi CD4 T cells. In addition, because of the enrichment of these PD-1^hi CD4 T cells in the rectum of SIV-infected noncontrollers, as a result of memory CD4 T cell depletion, these PD-1^hi CD4 T cells seemed to express higher levels of Bcl-2, which was not observed in the LN. We can hypothesize that as a result of CD4 depletion, immune activation, and widespread epithelial damage in the GALT, these PD-1^hi cells that localized to germinal centers of rectal aggregates may increase Bcl-2 expression to promote survival and persistence in the face of extensive immune dysregulation. In addition, in contrast to LN PD-1^hi CD4 T cells in SIV noncontrollers that experienced decreased cytokine polyfunctionality, PD-1^hi CD4 T cells in the rectum seemed to retain their ability to produce IFN-γ and IL-21. PD-1^hi cells in the rectum, compared with PD-1^int, and PD-1^neg memory CD4 T cells, also seem to represent the most active cellular subset for ongoing viral production as they contained the highest RNA to DNA ratio per cell. These observations suggest that despite an observed aberrant enrichment of PD-1^hi CD4 T cells in both the LN and rectum of SIV noncontrollers, site-specific immune response during chronic SIV infection may contribute to localized differences in PD-1^hi CD4 T cell subsets.

A critical finding of our study is that the vaccine controllers do not show an enrichment of PD-1^hi CD4 T cells in the B cell follicles of rectum or LN. We speculate that multiple mechanisms could have contributed to this outcome. We observed an increase in CCR5 expression and a decrease in CXCR5 expression on PD-1^hi CD4 T cells in the controllers. This shift in chemokine receptor expression could promote T cell migration away from the germinal center area of B cell follicles toward T cell zones leading to enhanced killing by cytotoxic CD8 T cells. A recent study demonstrated that the Tfh cells move from one germinal center to another germinal center within a LN (48). Therefore, another plausible mechanism that prevents the aberrant enrichment of these PD-1^hi cells is that before reaching the germinal center of lymphoid follicles, these PD-1^hi cells are targeted and killed by antiviral CD8 T cells. In addition, the controllers maintained higher frequency of granzyme B⁺ antiviral CD8 T cells that correlated inversely with the frequency of PD-1^hi CD4 T cells. It is possible that these CD8 T cells restrict virus replication in PD-1^hi CD4 T cells through cytolytic as well as noncytolytic mechanisms. Furthermore, our ongoing work suggests that some of the antiviral CD8 T cells express CXCR5 and thus could migrate to B cell zone (data not shown).

Thus, our results highlight the importance of SIV-specific CD8 T cells at sites of ongoing viral replication and persistence. Our results also suggest the role of functional antiviral CD8 T cells in limiting the aberrant enrichment of SIV⁺PD-1^hi CD4 T cells at lymphoid sites during chronic uncontrolled SIV infection. Finally, it is conceivable that manipulating the localization of Tfh and antiviral CD8 T cells to and from the germinal center may enhance immune-mediated control of HIV/SIV-infected target cells. These data highlight the importance of generating strong and potent antiviral CD8 T cells at sites of active viral replication and persistence and support the rationale for using targeted therapies that promote Tfh migration out of the GC to allow for increased clearance by antiviral CD8 T cells and enhanced viral control.

We thank the veterinary staff at Yerkes for animal care, Center for AIDS Research virology core for viral RNA and quantitative PCR analysis, and Center for AIDS Research immunology core for help with flow cytometry. Also, we thank the NIH AIDS Research and Reference Reagent Program for the provision of peptides.

R.R.A. and V.V. are co-inventors of PD-1 technology that has been licensed to Genentech by Emory University. All other authors have no financial conflicts of interest.