SIV-specific CD8⁺ T cells are clonotypically distinct across lymphoid and mucosal tissues

CD8⁺ T cells are required to control human immunodeficiency virus (HIV) and simian immunodeficiency virus (SIV) but largely fail to protect infected humans and monkeys from eventual progression to AIDS (1). These lentiviruses replicate primarily in lymphoid tissues and preferentially infect CD4⁺ follicular helper T (T_FH) cells (2). Elite control of HIV/SIV replication has been attributed to various qualitative properties of virus-specific CD8⁺ T cells (3–8), including the expression of CXCR5, which allows follicular access and direct proximity to infected CD4⁺ T_FH cells (2, 9, 10). However, most studies to date have been limited to evaluations of peripheral blood samples, which may not fully recapitulate the whole-body response to HIV/SIV.

Antigen-specific CD8⁺ T cell populations generally incorporate several clonotypes, defined by the expression of distinct T cell receptors (TCRs) (11, 12). The composite properties of these constituent clonotypes in turn dictate the overall functionality of a given antigen-specific CD8⁺ T cell population (13). In the context of HIV/SIV, morphologically constrained repertoires directed against biologically variable epitopes have been associated with reproducible patterns of viral escape (14), whereas certain aspects of morphologically diverse repertoires directed against biologically constrained epitopes have been associated with protection, including the early mobilization of public clonotypes (15), which are commonly generated via convergent recombination and therefore occur in multiple individuals (16, 17), and the subsequent mobilization of cross-reactive private clonotypes (18). It nonetheless remains unclear how SIV-specific CD8⁺ T cells traffick in vivo to mediate these effects at sites of viral replication.

Recent work has identified discrete subsets of tissue-resident memory T cells, which occupy various epithelial, lymphoid, and mucosal sites (19–22) and differ from circulating memory T cells (23), constitutively expressing the C-type lectin CD69 and lacking the lymph node (LN) homing markers CCR7 and CD62L (24). In line with this concept of localized adaptive immunity, previous studies have revealed a degree of clonotypic disparity between circulating and mucosal SIV-specific CD8⁺ T cells, both in the context of acute infection (25) and in response to vaccination (26). Similarly, tissue-resident CD8⁺ T cells isolated from the LNs of elite controllers in another study exhibited skewed clonotypic profiles relative to specificity-matched CD8⁺ T cells isolated from venous blood samples, whereas the corresponding CD69^– HIV-specific CD8⁺ T cells were clonotypically similar across the same anatomical compartments (27). These data suggest that distinct clonotypes may occupy distinct tissue niches, but to date, a comprehensive survey has not been performed across multiple anatomical sites.

In this study, we compared the clonotypic, functional, and phenotypic characteristics of SIV-specific CD8⁺ T cells across matched samples of peripheral blood and various lymphoid and mucosal tissues isolated from chronically infected rhesus macaques with high (>10,000 RNA copies/mL plasma) or low viral loads (VLs) (<10,000 RNA copies/mL plasma). We found that public clonotypes were commonly shared across anatomical sites, whereas private clonotypes were more often tissue specific. In addition, we identified correlates of immune control, including the presence of polyfunctional CXCR5⁺ SIV-specific CD8⁺ T cells in lymphoid tissues, such as the spleen and various LNs.

SIV-specific CD8⁺ T cells occur at similar frequencies in lymphoid and mucosal tissues. In preliminary experiments, we used fluorochrome-labeled pentameric complexes of Mamu-A*01-CTPYDINQM Gag_181–189 (CM9) and/or Mamu-A*02-YTSGPGIRY Nef_159–167 (Nef) and Mamu-A*02-GSENLKSLY Gag_71–79 (Gag) to evaluate the distribution of SIV-specific CD8⁺ T cells across matched samples of peripheral blood and various tissue homogenates obtained from chronically infected rhesus macaques with high (>10,000 RNA copies/mL plasma; n = 16) or low VLs (<10,000 RNA copies/mL plasma; n = 6) (Table 1 and Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/JCI129161DS1). No significant anatomical differences in response magnitude were detected between or within these outcome-defined groups (Figure 1A). Accordingly, the frequency of SIV-specific CD8⁺ T cells in the spleen correlated with the frequencies of SIV-specific CD8⁺ T cells in the GI tract (Figure 1B), the peripheral blood, and the axillary/inguinal lymph nodes (ALNs/ILNs) (Table 2). In contrast, the frequency of SIV-specific CD8⁺ T cells in the peripheral blood correlated only with the frequency of SIV-specific CD8⁺ T cells in the spleen, and the frequency of SIV-specific CD8⁺ T cells in the mesenteric lymph nodes (MLNs) correlated only with the frequency of SIV-specific CD8⁺ T cells in the GI tract (Table 2).

Figure 1

SIV-specific CD8⁺ T cells occur at similar frequencies in lymphoid and mucosal tissues. (A) Frequency of SIV-specific (CM9/Nef/Gag) CD8⁺ T cells across various anatomical sites. Horizontal bars indicate median values. GI, gastrointestinal. (B) Correlation between the frequency of SIV-specific CD8⁺ T cells in the GI tract and the frequency of SIV-specific CD8⁺ T cells in the spleen. (C–E) Frequency correlations for CD27⁺ (C), CD28⁺ (D), and CD69⁺ SIV-specific CD8⁺ T cells (E) in the GI tract versus the spleen. Data were acquired from Mamu-A*01⁺ and Mamu-A*02⁺ rhesus macaques (n = 22). Significance was determined using the Wilcoxon rank sum test (A) or Spearman’s rank correlation with linear regression (B–E).

Table 1

Characteristics of rhesus macaques used in this study

Table 2

Frequency correlations for SIV-specific CD8⁺ T cells across various anatomical sites

In further analyses, we compared the phenotypes of lymphoid and mucosal SIV-specific CD8⁺ T cells, focusing on expression of the costimulatory molecules CD27 and CD28 and the tissue residency marker CD69. Strong correlations were detected between SIV-specific CD8⁺ T cells in the spleen and SIV-specific CD8⁺ T cells in the GI tract with respect to the expression frequencies of CD27 (Figure 1C) and CD28 (Figure 1D). No such association was observed for CD69 (Figure 1E). It was also noted that CD27, CD28, and CD69 were not differentially expressed on the surface of SIV-specific CD8⁺ T cells as a function of VL (data not shown).

Expression of CXCR5 on SIV-specific CD8⁺ T cells correlates inversely with VL. As expected, higher frequencies of CD4⁺ T cells in the GI tract and higher numbers of CD4⁺ T cells in the peripheral blood were detected in rhesus macaques with low VLs relative to rhesus macaques with high VLs (Figure 2, A and B). Moreover, the frequency of SIV-specific CD8⁺ T cells in the GI tract correlated with the frequency of CD4⁺ T cells in the GI tract and the number of CD4⁺ T cells in the peripheral blood (Figure 2, C and D).

Figure 2

Expression of CXCR5 on SIV-specific CD8⁺ T cells correlates inversely with VL. (A) Frequency of CD4⁺ T cells in the GI tract. (B) Number of CD4⁺ T cells in the peripheral blood. (C) Correlation between the frequency of SIV-specific CD8⁺ T cells in the GI tract and the frequency of CD4⁺ T cells in the GI tract. (D) Correlation between the frequency of SIV-specific CD8⁺ T cells in the GI tract and the number of CD4⁺ T cells in the peripheral blood. (E) Frequency of CXCR5⁺ SIV-specific CD8⁺ T cells in the spleen. (F) Correlation between the frequency of CXCR5⁺ SIV-specific CD8⁺ T cells in the spleen and the amount of viral DNA in CD4⁺ T_FH cells. (G) Correlation between the frequency of CXCR5⁺ SIV-specific CD8⁺ T cells in the spleen and VL. Data were acquired from Mamu-A*01⁺ and Mamu-A*02⁺ rhesus macaques (n = 22). Horizontal bars indicate median values (A, B, and E). Significance was determined using the Mann-Whitney U test (A, B, and E) or Spearman’s rank correlation with linear regression (C, D, F, and G).

Earlier work suggested that SIV-specific CD8⁺ T cells can enter lymphoid follicles in Asian macaques with low VLs (9, 10). In line with this observation, we found higher frequencies of CXCR5⁺ SIV-specific CD8⁺ T cells in the spleens of rhesus macaques with low VLs relative to rhesus macaques with high VLs (Figure 2E). To assess the biological relevance of this association, we measured SIV gag DNA in flow-sorted CD4⁺ T_FH cells from the same tissues (Supplemental Figure 2). The frequency of CXCR5⁺ SIV-specific CD8⁺ T cells in the spleen correlated inversely with the amount of CD4⁺ T_FH cell–associated viral DNA (Figure 2F) and the plasma burden of viral RNA (Figure 2G).

SIV-specific CD8⁺ T cells are polyfunctional in rhesus macaques with low VLs. To extend these findings, we assessed the functionality of SIV-specific CD8⁺ T cells in response to cognate peptide stimulation, measuring surface mobilization of CD107a and the simultaneous induction of IFN-γ, IL-2, IL-21, and TNF-α. The inclusion of IL-21 in the readout was designed to identify T_FH-like activity (28). Overall, SIV-specific CD8⁺ T cells were more polyfunctional in the ALNs/ILNs of rhesus macaques with low VLs relative to rhesus macaques with high VLs, but very few of these cells in either group produced IL-21 (Figure 3A).

Figure 3

SIV-specific CD8⁺ T cells are polyfunctional in rhesus macaques with low VLs. (A) Pie charts depict relative expression of CD107a, IFN-γ, IL-2, IL-21, and/or TNF-α among memory CD8⁺ T cells in response to stimulation of ALNs/ILNs with overlapping peptides spanning SIV Gag (n = 6 Mamu-A*01⁺ rhesus macaques per group). (B) Pie charts depict relative expression of CD107a, granzyme B, IFN-γ, MIP-1β, RANTES, and/or TNF-α among CXCR5⁺ or CXCR5^– memory CD8⁺ T cells in response to stimulation of splenocytes with overlapping peptides spanning SIV Gag (n = 4 Mamu-A*01⁺ rhesus macaques per group). Significance was determined using the permutation test in Simplified Presentation of Incredibly Complex Evaluations (SPICE) (detailed in Table 3).

In similar experiments, we compared the functionality of CXCR5⁺ and CXCR5^– SIV-specific CD8⁺ T cells, measuring surface mobilization of CD107a and intracellular granzyme B, IFN-γ, macrophage inflammatory protein–1β (MIP-1β), RANTES, and TNF-α. CXCR5⁺ SIV-specific CD8⁺ T cells were more polyfunctional in the spleens of rhesus macaques with low VLs relative to rhesus macaques with high VLs and produced a wider range of effector molecules relative to CXCR5^– SIV-specific CD8⁺ T cells (Figure 3B and Table 3).

Table 3

Functionality comparisons for CXCR5⁺ versus CXCR5^– memory CD8⁺ T cells in rhesus macaques with high versus low VLs

SIV-specific CD8⁺ T cell clonotypes targeting CM9 are structurally diverse. The early mobilization of public clonotypes and the subsequent mobilization of cross-reactive private clonotypes directed against biologically constrained epitopes have been associated with enhanced control of HIV/SIV (15, 18). We therefore used an unbiased molecular approach to characterize all expressed TCRβ locus (TRB) gene rearrangements in flow-sorted Mamu-A*01–restricted CM9-specific CD8⁺ T cell populations (Supplemental Figure 2) isolated at necropsy from each anatomical site in rhesus macaques with high (Supplemental Figure 3) or low VLs (Supplemental Figure 4). Data were normalized before analysis to correct for differences in sample size (29, 30). The number of unique clonotypes per anatomical site was similar in rhesus macaques with high (median 6, range 1–16) and low VLs (median 7, range 1–13). Moreover, no consistent anatomical differences in repertoire diversity were apparent between or within these outcome-defined groups, and accordingly, no correlations were detected between repertoire diversity and either the amount of CD4⁺ T_FH cell–associated viral DNA or the plasma burden of viral RNA (data not shown and Figure 4A). There were some conspicuous anatomical differences in repertoire similarity, however, most notably between lymphoid and mucosal sites in certain rhesus macaques with high or low VLs (Figure 4B).

Figure 4

Clonotypic architecture of CM9-specific CD8⁺ T cell populations. (A) Repertoire diversity in rhesus macaques with high (n = 6) or low VLs (n = 6). (B) Repertoire similarity in rhesus macaques with high (n = 6) or low VLs (n = 6). (C) TRBV gene use in rhesus macaques with high VLs (n = 6). (D) TRBV gene use in rhesus macaques with low VLs (n = 6). (E) CDR3β amino acid use in rhesus macaques with high VLs (n = 6). (F) CDR3β amino acid use in rhesus macaques with low VLs (n = 6). Significance was determined using the permutation test in SPICE (C and D).

A vast majority of CM9-specific clonotypes in both groups incorporated CDR3β loops spanning 14 amino acids, with equivalent representation at each anatomical site (Supplemental Figure 5). Irrespective of anatomical location, preferential use of TRBV6-1 and TRBV9 was observed in rhesus macaques with high VLs (Figure 4C), and preferential use of TRBV6-1, TRBV10-2, and TRBV13 was observed in rhesus macaques with low VLs (Figure 4D). Similarly, preferential use of TRBJ1-5 and TRBJ2-1 was observed in rhesus macaques with high VLs (Supplemental Figure 6A), and preferential use of TRBJ1-5 was observed in rhesus macaques with low VLs (Supplemental Figure 6B). No clear motifs were observed among non–germline-encoded CDR3β residues, irrespective of VL (Figure 4, E and F).

SIV-specific CD8⁺ T cell clonotypes targeting CM9 are anatomically discrete. To substantiate the observed anatomical differences in repertoire similarity, we assessed the distribution of CM9-specific clonotypes across peripheral blood and tissue sites, again comparing rhesus macaques with high versus low VLs. A degree of clonotype sharing was apparent in each individual rhesus macaque, most notably between the spleen and the peripheral blood and between the MLNs and the GI tract (Figure 5, A and B). However, many clonotypes were either unique to a particular anatomical site or preferentially located in a particular anatomical site, most commonly the GI tract (Figure 5, A and B). No obvious anatomical differences in clonotype sharing were detected between groups, although unique clonotypes were significantly more common in the MLNs of rhesus macaques with low VLs relative to rhesus macaques with high VLs (Figure 5C). Equivalent levels of sharing between groups were also observed for clonotypes stratified by the number of occupied sites (Figure 5D). In general, the number of anatomical sites occupied by a given clonotype increased as the frequency of a given clonotype increased at any given anatomical site, and this was true across all sampled tissues (Figure 5E and Table 4). However, there were exceptions to this pattern, especially in the GI tract, where some clonotypes occurred at high frequencies with no sharing (e.g., rhesus macaque 591) or sharing restricted to the MLNs (e.g., rhesus macaque 863), and other clonotypes occurred at low frequencies with sharing across all anatomical sites (e.g., rhesus macaques 760 and F98) (Figure 5, A and B). Moreover, approximately 40% of clonotypes were unique to each site, rising to more than 55% in the peripheral blood and the GI tract, and most of the shared clonotypes in the peripheral blood were found in the spleen but not in the GI tract (Figure 5F). These clonotypically distinct SIV-specific CD8⁺ T cell populations were nonetheless functionally equivalent across various tissue sites, at least after stimulation with nonlimiting concentrations of CM9 (Figure 5G).

Figure 5

CM9-specific CD8⁺ T cell clonotypes are anatomically discrete. (A) Frequency of individual clonotypes across various anatomical sites in rhesus macaques with high VLs. (B) Frequency of individual clonotypes across various anatomical sites in rhesus macaques with low VLs. (C) Proportion of clonotypes detected in 1 (unique) or more than 1 (shared) anatomical site in rhesus macaques with high (n = 6) or low VLs (n = 6). Significance was determined using the χ² test. **P < 0.01. (D) Proportion of clonotypes shared across the indicated number of anatomical sites in rhesus macaques with high (n = 6) or low VLs (n = 6). Significance was determined using the Mann-Whitney U test. (E) Correlation between the number of occupied anatomical sites and clonotype frequency in the GI tract. Significance was determined using Spearman’s rank correlation with linear regression. (F) Schematic representation of clonotype sharing across distinct anatomical sites (n = 12 Mamu-A*01⁺ rhesus macaques). Values in parentheses refer to unique clonotypes at each anatomical site (%). (G) Pie charts depict relative expression of CD107a, IFN-γ, MIP-1β, RANTES, and/or TNF-α among clonotypically distinct (TCR Vβ1⁺ or TCR Vβ23⁺) CM9-specific memory CD8⁺ T cells in response to stimulation of various tissue homogenates with overlapping peptides spanning SIV Gag (n = 4 Mamu-A*01⁺ rhesus macaques per group). Clonotypes found in multiple tissues (circulating) were distinguished from clonotypes found in 1 tissue (resident). Significance was determined using the permutation test in SPICE.

Table 4

Correlations between the number of occupied anatomical sites and clonotype frequency at each anatomical site

Public clonotypes were identified on the basis of sharing at the amino acid level across more than 1 rhesus macaque in our cohort and/or across previously reported data sets from other rhesus macaques (refs. 14, 15, 31 and Figure 6, A and B). No significant differences in the anatomical distribution of public clonotypes were detected between or within groups (Figure 6C). However, public clonotypes with identical nucleotide sequences were shared across anatomical sites more commonly than private clonotypes in individual rhesus macaques, irrespective of VL (Figure 6D).

Figure 6

Public CM9-specific CD8⁺ T cell clonotypes are commonly shared across multiple anatomical sites. (A) Frequency of public (colored bars) or private clonotypes (white bars) across various anatomical sites in rhesus macaques with high VLs. (B) Frequency of public (colored bars) or private clonotypes (white bars) across various anatomical sites in rhesus macaques with low VLs. (C) Proportion of public or private clonotypes detected at each anatomical site in rhesus macaques with high (n = 6) or low VLs (n = 6). Significance was determined using the χ² test. (D) Proportion of public or private clonotypes shared across the indicated number of anatomical sites in Mamu-A*01⁺ rhesus macaques (n = 12). Significance was determined using the Mann-Whitney U test.

Virus-specific CD8⁺ T cells are necessary for the control of HIV and SIV (32–34). However, the precise determinants of immune efficacy remain unclear, potentially because current paradigms are based primarily on studies of circulating CD8⁺ T cells. We report here that SIV-specific CD8⁺ T cells are clonotypically distinct across multiple anatomical compartments in rhesus macaques with chronic infection or AIDS. Public clonotypes were commonly shared across anatomical sites in individual rhesus macaques, as were the more dominant clonotypes at any particular anatomical site, but overall, approximately 40% of clonotypes were site specific, and unique clonotypes were observed most often in the peripheral blood and the GI tract. Moreover, rhesus macaques with low VLs harbored higher frequencies of polyfunctional CXCR5⁺ SIV-specific CD8⁺ T cells in various lymphoid tissues, such as the spleen and the ALNs/ILNs.

Expression of CXCR5 is required for entry into B cell follicles (35, 36). We found that the frequency of CXCR5⁺ SIV-specific CD8⁺ T cells in the spleen correlated inversely with viral replication, measured in terms of CD4⁺ T_FH cell–associated viral DNA and plasma viral RNA. Although these findings are consistent with the hypothesis that follicular access licenses the optimal delivery of antiviral effector functions (37–39), it is important to note that high levels of antigen expression can exhaust follicular CXCR5⁺ virus-specific CD8⁺ T cells as a consequence of sustained activation via the TCR (40). This phenomenon may explain, at least in part, the preferential occurrence of CXCR5⁺ SIV-specific CD8⁺ T cells in rhesus macaques with low VLs. Follicular CXCR5⁺ SIV-specific CD8⁺ T cells are nonetheless optimally positioned to suppress viral replication in vivo, potentially via noncytolytic mechanisms (41).

Our current understanding of HIV/SIV-specific CD8⁺ T cell responses is derived mainly from studies of peripheral blood, which is easily accessible in humans and other primates (1, 3, 4, 14, 15, 18, 32–34, 42). Other windows may provide better insights, however, because these viruses replicate in lymphoid tissues, and very few lymphocytes in the total body pool circulate at any given time (43). In line with previous studies, we found a degree of clonotype sharing across anatomical compartments (25, 26, 44), which has likewise been reported in the context of influenza virus infection (45). However, we also identified anatomically unique clonotypes, most frequently in the peripheral blood and the GI tract. The clonotypic disparity between peripheral blood and lymphoid tissues is consistent with the recent description of tissue-resident HIV-specific CD8⁺ T cells (27), which provide durable antiviral immunity at the site(s) of viral challenge/replication (27, 46, 47). Our data therefore suggest 2 mutually inclusive possibilities. The first is that individual CD8⁺ T cells are primed in situ and do not traffick extensively between anatomical compartments during chronic infection, implying that systemic vaccination may be required to elicit long-term immunity against HIV/SIV (26). The second is that public clonotypes are more readily primed during the early stages of infection as a function of prevalence within the naive pool, which can be attributed to convergent recombination (48, 49), and subsequently form highly stable tissue-resident populations, which are then more likely to be shared across sites of viral replication (27, 50). Longitudinal studies will be necessary to address these possibilities and determine which route(s) of vaccine delivery are likely to favor the generation of effective immunity against HIV/SIV.

In summary, we have demonstrated that SIV-specific CD8⁺ T cells are clonotypically, functionally, and phenotypically distinct across multiple anatomical sites in outcome-defined groups of rhesus macaques with chronic infection or AIDS. A greater understanding of these anatomical complexities may therefore be required to identify the key attributes of effective virus-specific CD8⁺ T cell responses, which in turn may inform the rational design of next-generation vaccines against HIV/SIV.

Study design. The goal of this study was to assess the functional and phenotypic characteristics of SIV-specific CD8⁺ T cell clonotypes across distinct anatomical compartments in rhesus macaques (Macaca mulatta) with high (>10,000 RNA copies/mL plasma; n = 16) or low VLs (<10,000 RNA copies/mL plasma; n = 6) during the chronic phase of infection with SIVmac239, SIVsmE543, or SIVsmE660 (Table 1 and Supplemental Figure 1). Selection criteria included the availability of cryopreserved tissue homogenates. PBMCs, spleen, ALNs, ILNs, MLNs, colon, and jejunum were obtained at necropsy from Mamu-A*01⁺ (n = 12) and Mamu-A*02⁺ rhesus macaques (n = 11) with established SIV infection or AIDS (Table 1). Samples were homogenized as described previously (51). Clinical endpoints were defined as (a) weight loss more than 25%; (b) major organ failure or medical conditions unresponsive to treatment; (c) complete anorexia for 4 days; (d) inability to consume sufficient nutrients to maintain body weight without assistance for 7 days; (e) distress vocalization unresponsive to intervention for 7 days; and/or (f) tumors arising from other than experimental means that impaired movement, ulcerated, or grew to more than 10% of body weight. Euthanasia was performed according to guidelines established by the American Veterinary Medical Association.

Flow cytometric sorting of SIV-specific CD8⁺ T cells. Tissue homogenates were washed twice in RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 mM l-glutamine, and 1% penicillin/streptomycin (all from HyClone, GE Healthcare Life Sciences). SIV-specific CD8⁺ T cells were identified using CM9 and/or Nef and Gag Pro5 MHC Class I Pentamers (ProImmune). Surface markers were identified using pretitrated concentrations of the following antibodies: (a) anti-CD8a (RPA-T8) and anti-CXCR5 (MU5UBEE) from eBioscience; (b) anti-CD3 (SP34-2), anti-CD8a (SK1), anti-CD27 (O323), anti-CD69 (FN50), and anti-CCR7 (3D12) from BD Biosciences; (c) anti-CD4 (OKT4), anti-CD95 (DX2), and anti–PD-1 (EH12.2H7) from BioLegend; and (d) anti-CD28 (CD28.2) from Beckman Coulter. Bulk CD4⁺ T_FH cells and SIV-specific CD8⁺ T cells were flow-sorted using a FACSAria (BD Biosciences). Bulk CD4⁺ T_FH cells were defined as live, CD3⁺, CD4⁺, CD28⁺, CD95⁺, CXCR5⁺, PD-1⁺ lymphocytes, and SIV-specific CD8⁺ T cells were defined as live, CD3⁺, CD8⁺, CM9/Nef/Gag⁺ lymphocytes (Supplemental Figure 2).

Functional analysis of SIV-specific CD8⁺ T cells. Cell suspensions were stimulated with a pool of peptides spanning SIV Gag (15-mers overlapping by 11 amino acids) at a final concentration of 2 μg/peptide/10⁶ cells in the presence of anti-CD28 (CD28.2; Beckman Coulter), anti-CD107a (H4A3; BioLegend), brefeldin A (1 μg/mL; MilliporeSigma), and monensin (1 μg/mL; BD Biosciences) for 16 hours at 37°C. After stimulation, cells were washed in PBS, stained for surface markers, fixed/permeabilized, and stained for intracellular markers. The following antibodies were used in these experiments: (a) anti-CD4 (OKT4), anti-CXCR5 (MU5UBEE), and anti–IL-21 (3A3-N2) from eBioscience; (b) anti-CD3 (SP34-2), anti-CD8a (SK1), anti-CD95 (DX2), anti–granzyme B (GB11), anti–IL-2 (MQ1-17H12), anti–MIP-1β (D21-1351), and anti-RANTES (2D5) from BD Biosciences; (c) anti–IFN-γ (4S.B3) and anti–TNF-α (MAb11) from BioLegend; and (d) anti–TCR Vβ1 (BL37.2) and anti–TCR Vβ23 (AF23) from Beckman Coulter. Data were acquired using an LSR Fortessa (BD Biosciences). SIV-specific CD8⁺ T cells were defined as live, CD3⁺, CD8⁺, function⁺/pentamer⁺ lymphocytes (Supplemental Figure 2). Functional profiles were analyzed using SPICE v5.35 (National Institute of Allergy and Infectious Diseases).

Quantification of cell-associated SIV DNA. SIV DNA was quantified in CD4⁺ T_FH cells as described previously (52, 53). Briefly, flow-sorted cells were lyzed using 25 μL of a 1:100 dilution of proteinase K (Roche) in 10 mM Tris-HCl (pH 8.4). Quantitative PCRs were performed in duplicate using 5 μL of cell lysate per reaction with TaqMan Master Mix (Thermo Fisher Scientific) over 50 cycles of 95°C for 15 seconds and 60°C for 1 minute after a single holding step of 95°C for 5 minutes. SIV_mac239 gag was amplified using the forward primer GTCTGCGTCATYTGGTGCATTC and the reverse primer CACTAGYTGTCTCTGCACTATRTGTTTTG with the probe CTTCRTCAGTYTGTTTCACTTTCTCTTCTGCG. SIV_{smE543/smE660} gag was amplified using the forward primer GGCAGGAAAATCCCTAGCAG and the reverse primer GCCCTTACTGCCTTCACTCA with the probe AGTCCCTGTTCRGGCGCCAA. Cell numbers were quantified relative to monkey albumin (54). Reactions were processed using a StepOnePlus Real-Time PCR System (Applied Biosystems, Thermo Fisher Scientific), and template copies were calculated using StepOne Software (Applied Biosystems, Thermo Fisher Scientific).

Clonotype analysis. Pentamer-labeled CD8⁺ T cells were flow-sorted viably into 100 μL of RNAlater (MilliporeSigma). All expressed TRB gene rearrangements were amplified without bias using a template-switch anchored reverse transcription PCR as described previously (55). Sequences were aligned using Sequencher v5.4.6 (Gene Codes Corporation). TRBV and TRBJ genes were identified according to the ImMunoGeneTics nomenclature (56). Data were normalized using a randomization procedure to account for differences in sample size as described previously (29, 30). Repertoire diversity was assessed using the Shannon diversity index (29), and repertoire similarity was assessed using the Morisita-Horn similarity index (30). Anatomical overlap was visualized using Circos (http://circos.ca/) and InteractiVenn (http://www.interactivenn.net/), and amino acid use was visualized using WebLogo (https://weblogo.berkeley.edu/).

Statistics. Matched groups were compared using the Wilcoxon rank sum test. Unmatched groups were compared using the Mann-Whitney U test. Correlations were evaluated using Spearman’s rank correlation with linear regression. Proportions were compared using the χ² test. All statistical analyses were performed using Prism v7.0 (GraphPad Software Inc.). P values less than 0.05 were considered significant. Functional profiles were compared using the permutation test with relative expression values in SPICE v5.35 (National Institute of Allergy and Infectious Diseases).

Study approval. Experimental procedures were approved by the National Institute of Allergy and Infectious Diseases Division of Intramural Research Animal Care and Use Program as part of the NIH Intramural Research Program (protocols LMM6 and LVD26). Rhesus macaques were housed and sustained in accordance with standards established by the Association for Assessment and Accreditation of Laboratory Animal Care.