Decreased T Follicular Regulatory Cell/T Follicular Helper Cell (T_FH) in Simian Immunodeficiency Virus–Infected Rhesus Macaques May Contribute to Accumulation of T_FH in Chronic Infection

Several key findings over the past few years have energized efforts toward the development of an effective HIV vaccine, including the discovery and characterization of a number of broadly neutralizing Abs (bnAbs) that develop in a subset of HIV-infected individuals. However, the mechanisms involved in shaping Ab responses to immunization with HIV Ags or natural HIV infection remain incompletely understood (1). Importantly, HIV-Env–specific bnAbs develop at relatively late stages of HIV infection and show peculiar genetic and molecular features, including a high level of divergence from germline predecessors, which indicates that they are the products of extensive somatic hypermutation within germinal centers (GCs), as well as the presence of unusually long CDR3 regions (2). Perplexingly, there appear to be no direct or predictable routes to the development of these bnAbs from the germline predecessors, and it remains unclear whether this process is driven by antigenic mutations and/or escape as opposed to specific intrinsic aspects of the B cell or Th cell response (3). A better understanding of the mechanisms responsible for the development of bnAbs is crucial to harness this type of immunity for HIV prevention and therapy in humans.

T follicular helper cells (T_FH) are critical for the development and maintenance of GCs, and competition for survival signals from T_FH via molecules such as CD40L and IL-21 is thought to be a key mechanism of selection of high-affinity B cells (4). The regulation of T_FH frequency, and by extension the regulation of their impact on GC B cell development and function, is vital to the quality of the humoral immune response (5). Although the presence of too few T_FH may lead to abortive GC formation and defective B cell responses, an overexpansion is associated with the prevalence of autoantibodies (6, 7). It is possible that an expansion of T_FH also lowers the selection pressure on GC B cells and leads to the emergence of low-affinity B cells (8).

Several studies have shown that T_FH accumulate during the chronic stages of HIV/SIV infection. This accumulation occurs even though these cells support high levels of viral replication and represent an important component of the persistent virus reservoir under antiretroviral therapy (9). The chronic expansion of T_FH in the case of HIV/SIV infection with persistent virus replication may be a direct result of antigenic persistence. As expected, HIV/SIV-associated expansion of T_FH is associated with dysregulation of B cell responses with ineffective memory cell formation and hypergammaglobulinemia (10, 11). Whether and to what extent this T_FH expansion is also related to a deficit in the physiological regulation of specific T_FH immune response within the lymph nodes remains unknown. However, this possibility would be consistent with the well-known observation that the chronic phase of pathogenic HIV/SIV infections is associated with a state of generalized immune activation that is resistant to the normal mechanisms of immune regulation.

Under normal circumstances, regulation of T_FH function is mediated, at least in part, by a recently described subset of CD4⁺ T cells termed T follicular regulatory cells (T_FR). T_FR are thought to develop from thymic-derived regulatory T cells (T_REG) that express lineage-associated markers such as FOXP3, CD25, and low levels of CD127. These T_FR migrate into the follicles of lymph nodes by virtue of their expression of CXCR5 (and downmodulation of CCR7) and, similarly to T_FH, express high levels of Bcl-6 and PD1 (12, 13). Notably, the role of T_FR in the immunopathogenesis of HIV/SIV infections is currently unknown, both in terms of ability to negatively regulate HIV-specific B cell responses (including, potentially, the production of bnAbs) and to suppress local virus-induced immune activation. Indeed, none of the previous reports on T_FH dynamics in the context of SIV or HIV infection have distinguished between cells that either do or do not coexpress T_REG-associated markers. Thus, the T_FR subset within the broader CXCR5⁺Bcl-6⁺PD1⁺ is not fully characterized in the setting of HIV/SIV infection.

In this study, we described and characterized phenotypically, histologically, and genomically the T_FR population that is found within the GCs of rhesus macaques (RM) and express markers associated with both T_FH and T_REG. The hypothesis that T_FR play a suppressive role in vivo is supported by the observation that their frequency is inversely correlated with both the levels of T_FH and GC B cells and the percentage of proliferating CD4⁺ T cells. In the setting of SIV infection, we found that T_FR show a slow in vivo proliferative response after the initial infection and exhibit only a small increase in their frequency within the total CD4⁺ T cell pool during the chronic phase. In conjunction with the large expansion of T_FH observed post SIV infection, this phenomenon leads to a significantly decreased T_FR/T_FH ratio in the lymph nodes of chronically SIV-infected RM. These data suggest that, during SIV infection, a lack of T_FR expansion may allow for a progressive accumulation of T_FH cells in the lymph nodes of chronically infected RM, thus indirectly contributing to the aberrant immune activation that characterizes this pathogenic infection.

This study involved a total of 40 Indian origin female RM divided as follows: 1) 10 healthy, unvaccinated, and SIV-uninfected RM; 2) 10 healthy, SIV-immunized, but SIV-uninfected RM; 3) 11 unvaccinated SIV-infected animals; and 4) 9 vaccinated and SIV-infected RM. Animals were vaccinated with a SIVmac239 Gag-, Pol-, and Env-expressing DNA vaccine with inactivating mutations in proteases, half of which also coexpressed GM-CSF. These were followed by two boosts of a SIVmac239 Gag-, Pol-, and Env-expressing modified vaccinia Ankara (MVA) vaccine as described previously (14). All infections were a result of SIVsmE660 intravaginal challenge at 2.06 × 10⁴ 50% tissue culture infective dose grown in RM PBMCs. Lymph node biopsies were collected for measurement of a number of immunological parameters at day −35 preinfection and days 14 and 168 postinfection (i.e., acute and chronic phase, respectively). Spleen and lymph nodes were collected at necropsies performed at 6 mo postinfection. All animals were housed at Yerkes National Primate Center at Emory University and were cared for in accordance with National Institutes of Health guidelines and following protocols approved by the Institutional Animal Care and Use Committee.

Lymphocytes were isolated from freshly obtained lymph node and spleens by passing homogenized tissue through a 70-μm cell strainer and lysing blood cells with ACK lysis buffer. Tissue collection and processing was performed as previously described (15). Cells to be later used for sorting were cryopreserved at −80°C in FBS media containing 10% DMSO.

Multicolor flow-cytometric analysis was performed on mononuclear cells isolated from blood and lymph nodes according to standard procedures using mAbs directed against RM markers and human markers that also cross-react with the same markers in RM. For optimum staining of intracellular markers, permeabilization of cells using the eBioscience FOXP3-permeabalization buffer was performed as recommended by the manufacturers. Predetermined optimal concentrations of the following Abs and reagents were used: CD3–Alexa 700 (clone SP34-2), CD4-allophycocyanin-Cy7 (clone OKT-4), Bcl-6–PE–Texas Red (clone K112-91), Ki67-FITC (clone B56), CCR5-PE (clone 3A9), and CTLA4-BV421 (clone BNI3) from BD; CXCR5-PerCP eFlour 710 (clone MU5BEE), PD1-PeCy7 (clone J105), and CD127-PeCy5 (clone eBio-RDR5) from eBioscience; and CD20-BV650 or PE-CF594 (clone 2H7), CD25-BV711 (clone BC96), Helios-FITC (clone 22F6), and FOXP3-allophycocyanin (clone 150D) from Biolegend; and Live/Dead Fixable Aqua from Invitrogen. Flow-cytometric data were acquired using LSRII flow cytometer using BD’s FACSDiva software. Acquired data were analyzed using Flow Jo version 9.3.2 following the gating strategy described in Fig. 1. Further analyses were performed using PRISM (GraphPad) and Excel (Microsoft Office 2011) software.

Cryopreserved cells were thawed in a 37°C water bath and rested overnight for 8–10 h and then stained for sorting. Splenocytes from five SIV-uninfected and unvaccinated RM, as well as five unvaccinated SIV-infected animals were used for sorting of T_FH, T_FR, and T_REG. Cell populations were sorted using FACSAria II flow cytometer. Cells were first gated based on light scatter followed by positive gating on cells negative for Live/Dead Fixable Aqua and positive for CD3 and CD4. After collecting bulk CD4⁺ cells the following three populations were collected: T_FR (CXCR5⁺PD1^hiCD127⁻CD25⁺), T_FH (CXCR5⁺PD1^hiCD127^+/−CD25⁻), and T_REG (CXCR5^+/−PD1^lo/intCD127⁻CD25⁺).

Immunohistochemistry was performed on 5-μm tissue sections mounted on glass slides, which were deparaffinized and rehydrated with double-dH₂O. Ag retrieval was performed in 1× Dako Target Retrieval Solution (pH 6.0) in a pressure cooker heating slides to 122°C for 30 s. Slides were then rinsed in ddH₂O and incubated for 10 min using Dako Protein block. Slides were then incubated with rabbit anti-CD4 (1:200), mouse anti-FOXP3 (1:100), and goat anti-PD1 (1:500) for 1 h at room temperature. Next, slides were washed in TBS with 0.05% Tween 20. Slides were then incubated for an hour in the dark with secondary Ab mixture containing donkey anti-rabbit Alexa 488 (1:500), donkey anti-mouse Alexa 594 (1:500), and donkey anti-goat Alexa 647 (1:500). After washing in TBS with 0.05% Tween 20, Prolong Gold with DAPI was applied to all the slides. Confocal microscope images were obtained using Olympus FV10i Confocal Microscope with CellSens 1.9 Digital Imaging software.

Cell-associated viral DNA was measured in sorted cell populations from RM lymph nodes by RT-PCR as previously described (16–18).

Total RNA was prepared using the QIAGEN RNeasy Micro Kit. Libraries were generated using the Clontech SMARTer HV kit, barcoding, and sequencing primers were added using Nextera XT DNA kit. Libraries were validated by microelectrophoresis, quantified, pooled, and clustered on Illumina TruSeq v3 flow cell. Clustered flow cell was sequenced on an Illumina HiSequation 1000 in 100-base single-read reactions.

RNA sequencing (RNA-Seq) data were submitted to the Gene Expression Omnibus repository at the National Center for Biotechnology Information. RNA-Seq data were aligned to a provisional assembly of Indian Macaca mulatta (MaSuRCA rhesus assembly v.7_20130927) using STAR version 2.3.0e (19, 20). Transcripts were annotated using the provisional UNMC annotation v7.6. Transcript assembly, abundance estimates, and differential expression analysis were performed using Cufflinks v2.1.1 and Cuffdiff (21). All samples had read counts >12,000,000 and unique mapping percentages in the range of 63–76%; no samples were excluded from the analysis for technical issues. Differentially expressed genes were defined by pairwise comparison of each phenotype. Differential gene lists were uploaded to Ingenuity Pathway Analysis software (v1.0 Ingenuity Systems, http://www.ingenuity.com) and pathways with significant enrichment by Fisher’s exact test and the Benjamini-Hochberg multiple testing correction. Heat maps and other visualization were generated using Partek Genomics Suite v6.6. RNA-Seq data are publically available at the Gene Expression Omnibus repositories (accession number: GSE69756, http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE69756).

Total RNA was prepared using the QIAGEN RNeasy Micro Kit from sorted T_FR, T_FH, and T_REG. RNA quantity was measured using Nanodrop analysis and reverse transcribed as previously described for RNA sequencing. Finally, 0.1 μl cDNA was used for real-time SYBR green PCR analysis using an ABI 7900 HT real-time PCR instrument (Applied Biosystems). Primer sequences for PCR were GAPDH: forward (Fwd) 5′-GCACCACCAACTGCTTAGCAC-3′, reverse (Rev) 5′-TCTTCTGGGTGGCAGTGATG-3′; IL-2RA: Fwd 5′-GGCTTCATTTTCCCACGGT-3′, Rev 5′-GCAGCTGGCGGACCAA-3′; IL-6R: Fwd 5′-TTCGGCCGGACTGTTCTG-3′, Rev 5′-GCACCCCATCTCCGACG-3′; SLAMF6: Fwd 5′-TGG AAC ATC TCT TGC CTT CAT AG-3′, Rev 5′-GTT GCT GAG TTT CAG GGA GTA G-3′; SAP/SH2D1A: Fwd 5′-CTC TGC AGT ATC CAG TTG AGA AG-3′, Rev 5′-GGC TTT CAG GCA GAC ATC A-3′; XIAP: Fwd 5′-GAG GAA CCC TGC CAT GTA TAG-3′, Rev 5′-GTG TAG TAG AGT CCA GCA CTT G-3′; PRDM1: Fwd 5′-TGT GGT ATT GTC GGG ACT TTG-3′, Rev 5′-GCT TGA GAT TGC TCT GTG TTT G-3′; CCL20: Fwd 5′-GCA ACT TTG ACT GCT GTC TTC-3′, Rev 5′-CAG CAT TGA TGT CAC AGG TTT C-3′; PD1: Fwd 5′-TCCTTGGCCACTGGTGTTC-3′, Rev 5′-CTTCTCCTGAGGGAAGGAGC-3′; IL-10: Fwd 5′-AAGACCCTCAGGCTGAGGCT-3′, Rev 5′-TCCACGGCCTTGCTCTTG-3′; IL-21: Fwd 5′-TGTGAATGACTTGGACCCTGAA-3′, Rev 5′AAACAGGAAATAGCTGACCACTCA-3′. Relative RNA transcript levels were calculated normalized to primer efficiency and housekeeping gene RNA (GAPDH).

Except for RNA-Seq data, all statistical analyses were conducted using GraphPad Prism 5.0. Comparisons of mean fluorescence intensity (MFI) between cell populations in uninfected RM were made using Wilcoxon signed rank tests (Fig. 2). Mann–Whitney U tests were used to compare frequencies of populations in uninfected, acutely infected, and chronically infected RM (Figs. 4, 5). Spearman rank correlation tests were used to analyze all correlations (Fig. 6). All p values <0.05 were defined as significant.

Recent studies of GC T_FH have defined these cells based on their surface expression of the chemokine receptor CXCR5 and very high levels of the coinhibitory receptor PD1 (10). However, a fraction of these canonically defined T_FH also express the lineage-specific T_REG marker FOXP3 and have been therefore defined as T_FR as proposed previously (22–24). In this article, we identified CD4⁺ T_FR by flow cytometry by their coexpression of CXCR5, PD1, FOXP3, and CD25 within lymph nodes of uninfected RM. The gating strategy used to define T_FH, T_FR, and T_REG throughout this study is shown in Fig. 1A. Notably, the gating strategy for T_REG includes both CXCR5⁺ and CXCR5⁻ cells. To confirm the presence of T_FR within GCs, we conducted an immunohistochemistry analysis. As shown in Fig. 1B, single cells with nuclear expression for FOXP3 and surface expression of PD1 were identified within GCs of uninfected RM. These T_FR can also be readily identified within GCs of SIV-infected RM (Fig. 1C). Interestingly, several bona fide T_REG, identified by their expression of FOXP3, but not PD1, are visible in the T cell zone just outside the GC (Fig. 1C). Presumably, some of these T_REG migrate into the GC and upregulate T_FH-like markers along their differentiation pathway to T_FR. Fig. 1B and 1C also show that, as expected, non-FOXP3–expressing “true” T_FH are also seen within GCs of the same animals.

We next performed a comprehensive examination of the T_FR phenotype in healthy, SIV-uninfected RM. As shown in Fig. 2, our analysis of relative MFIs for T_FR markers confirmed that T_FR express FOXP3 and CD25 at comparable levels with T_REG (Fig. 2A) and both CXCR5 and PD1 at comparable levels with T_FH (Fig. 2B). We next examined in T_FR the expression patterns of a series of markers (i.e., CD127, CTLA4, Bcl-6, and Helios) that have been linked to either T_FH or T_REG phenotype and function (1, 25). CD127, the IL-7R α-chain, is expressed at low levels on T_REG in humans (26–28). As expected, we found that T_FR express CD127 at lower levels than the bulk of CD4⁺ T cells, and similar or even lower levels than those observed in T_REG and T_FH (Fig. 2C). CTLA4 is a key negative T cell regulator that is constitutively expressed on T_REG and, upon ligation, induces downmodulation of cytokine production and inhibition of cell cycle progression (25). Consistent with previous reports in murine models (12), we observed that T_FR express CTLA4 at a higher frequency and MFI than both T_REG and T_FH cell populations (Fig. 2D). This is consistent with a putative role of T_FR as negative regulators of GC responses. Helios is a transcription factor expressed in thymus-derived natural T_REG (29). As shown in Fig. 2F, T_FR express Helios at levels that are even higher than those observed in T_REG in terms of both frequency of positive cells and MFI, thus suggesting that T_FR originate from natural T_REG in RM, as well as in mice.

To further define the functional features of T_FR in RM, we next examined the transcriptional profiles of T_FH, T_FR, and T_REG using RNA-Seq by Illumina technology. Splenocytes from five healthy, SIV-uninfected, and unvaccinated RM were sorted into “bulk” CD3⁺CD4⁺ T cells, T_REG, T_FH, and T_FR based on the following phenotypic markers: T_FR (CXCR5⁺PD1^hiCD127⁻CD25⁺), T_FH (CXCR5⁺PD1^hiCD127^+/−CD25⁻), and T_REG (CXCR5^+/−PD1^lo/intCD127⁻CD25⁺). In mice, T_FR are derived from thymic T_REG precursors and acquire homing markers that allow them to traffic to GCs, while maintaining a transcriptome and suppressive function that most closely resembles T_REG (12, 13). To examine the transcriptional profile of T_FR relative to T_REG and T_FH in healthy, SIV-uninfected RM, we first performed principal component analysis on a subset of the most highly expressed transcripts detected in T_FH, T_REG, and T_FR (Fig. 3A). The transcriptomes of each subset were clearly distinct and grouped by subset, with T_REG displaying the highest degree of intrasubset variability, and T_FH and T_FR subsets being more tightly distributed. We next compared the expression of several canonical T_FH and T_REG transcripts between the three subsets. As shown in Fig. 3B–D, T_FH- and T_REG-related genes showed expression patterns that behaved as predicted with genes such as IL-10 expressed in T_FR and T_REG but absent in T_FH. Importantly, RNA-Seq data confirmed that T_FR share expression of T_REG signature transcripts such as FOXP3, GZMB, PRDM1, and IL-2RA (Fig. 3). However, we found that several other T_REG-specific transcripts were expressed at lower levels in T_FR than in T_REG, including TRAF6, CD74, CCL20, and IL-1R1. Similar to previous studies in mice, T_FR also showed elevated expression of the prototypical T_FH genes CXCR5, PD1/PDCD1, BCL-6, CXCL13, and ICOS. In fact, T_FR demonstrated a peculiarly high expression of the T_FH- and T_REG-specific genes Bcl-6 and FOXP3, respectively. Interestingly, for several genes (SH2D1A/SAP, IL-21, CXCR5, IL-10), gene expression was higher in T_FR than either T_REG or T_FH, thus suggesting that the CD25⁺CXR5⁺PD1^hi phenotype may represent a more transcriptionally active population than classical T_REG or T_FH. This latter set of RNA-Seq data provides strong evidence that T_FR are indeed a distinct cell subset and that the somewhat hybrid transcriptional profile of T_FR is not simply due to the sample being a mixture of T_REG and T_FH. Notably, elevated expression of IL-10 in T_FR compared with T_REG has been previously reported in murine studies (13).

To then compare the profile of gene expression between T_FR with T_REG and T_FH subsets without using any a priori information, we defined T_FH and T_REG expression signatures by statistically contrasting RNA-Seq data from T_FH and T_REG with bulk CD4⁺ T cells. After exclusion of transcripts that had zero expression in any of the populations, a total of 88 genes made up the combined T_FH and T_REG signature of which 12 genes were T_REG related. Many, but not all, canonical T_REG and T_FH genes were also identified as significantly upregulated compared with bulk CD4⁺ T cells. The lack of statistical significance for some prototypical transcripts is likely due to the presence T_REG and T_FH subsets within the bulk CD4⁺ population used as a comparator sample. Nevertheless, we found that T_FR show similar levels of expression of T_FH signature genes such as Bcl-6, TIGIT, CD200, LATm, and BATF (Fig. 3C). T_FR cells also express mRNA for key T_FH-related genes that are important for B cell help, including IL-21, SH2D1A, CD40L, and CD84. One notable difference between our data in RM and previously published mouse studies is that we observed high expression of IL-21 in T_FR, suggesting that these cells have acquired some specific genomic and functional features in primates. As shown in Supplemental Fig. 2A–C, the expression patterns of IL-21, SH2D1A, SLAMF6, PD1, IL-6R, CCL20, IL-2RA, IL-10, PRDM1, and XIAP were confirmed by RT-PCR quantification. In addition, levels of protein expression of IL-21 by T_FR, T_FH, and T_REG were also measured by flow cytometry and further confirmed the pattern observed by RNA sequencing and RT-PCR (Supplemental Fig. 2D).

The dynamics of T_FR in the setting of HIV or SIV infection have not been previously investigated; in fact, all published studies of T_FH dynamics during HIV/SIV infection used a definition of these cells that included T_FR as well. To study the kinetics of T_FR, T_FH, and T_REG post SIV infection of RM, we measured the frequency of these cells within the lymph nodes preinfection, 2 wk postinfection, and 6 mo postinfection with SIVsmE660. The RM included in these kinetics analyses included both unvaccinated and animals that were challenged after immunization with a SIVmac239 Gag-, Pol-, and Env-expressing DNA vaccine (with or without GM-CSF) followed by two boosts of a SIV239 Gag-, Pol-, and Env-expressing MVA vaccine. As previously reported, we found a significant increase (p < 0.0001) in frequency of T_FH at 24 wk postinfection (Fig. 4A). Interestingly, the frequency of T_FR measured as percent of total CD4⁺ T cells also showed a significant (p = 0.0001) increase during chronic SIV infection (Fig. 4A). However, when the frequency of T_FR is measured as percentage of total T_FH, we found that the T_FR decrease significantly at both the acute (p = 0.0385) and the chronic (p = 0.0016) stages of SIV infection (Fig. 4B). Accordingly, the overall ratio of T_FR to T_FH also decreased significantly (p = 0.0018) at the week 24 postinfection time point as compared with baseline (Fig. 4C). The increase of both T_FH and T_FR as a percent of CD4⁺ T cells postinfection is likely the result of proliferation driven by Ag persistence and virus-mediated depletion of other CD4⁺ T cell subsets. However, the relative decrease in the frequency of T_FR when measured as percentage of T_FH suggests that the low frequency of these regulatory cells might contribute to the expansion and accumulation of T_FH in chronically SIV-infected RM. Notably, we found no significant changes in T_REG frequencies after SIV infection within the lymph nodes. To better define the kinetics of T_FH and T_FR during SIV infection, we next measured the level of cell proliferation using the well-established marker Ki67. We observed that T_FH show a significant increase in proliferating cells during the acute (p < 0.0001) and chronic (p = 0.0001) phases of infection (Fig. 4D). T_FR have a similar pattern of proliferation, with a significant increase in proliferating cells during the acute (p < 0.0001) and chronic (p = 0.0376) phases of infection (Fig. 4D). In contrast, the level of Ki67 expression in T_REG remains relatively low throughout our analysis with a small, significant increase (p = 0.0141) during the chronic phase of infection (Fig. 4D).

Several studies have shown that, during HIV and SIV infection, T_FH are highly infected with the virus despite their relative increase within the total CD4⁺ T cell pool (9). Although the actual in vivo life span of T_FH, either infected or uninfected, remains unknown in the setting of HIV/SIV infection, the presence of a notable fraction of these cells expressing the proliferation marker Ki67 suggests that their number could be maintained through continual replenishing from precursors located outside the GC. To measure the level of direct SIV infection of T_FR, T_FH, and T_REG, we sorted these subpopulations from the lymph nodes of a subset of our studied animals and quantified the levels of total cell-associated SIV-DNA by RT-PCR. This analysis revealed that T_FH, T_FR, and T_REG derived from chronically SIV-infected RM all harbor comparably high levels of cell-associated viral DNA (Fig. 5A). Interestingly, the levels of SIV infection were similarly high between T_FR and T_FH even though the surface expression levels of the main SIV coreceptor CCR5 were significantly higher in T_FR as compared with T_FH (Fig. 5B).

To further examine the relationship between T_FR and T_FH and GC B cells, we next performed a set of correlation analyses in the RM included in this study. We observed that, in healthy uninfected RM, the frequency of T_FR (as fraction of the total T_FH pool) is negatively correlated with the percentages of T_FH (as fraction of total CD4⁺ T cells) and GC B cells (as fraction of total B cells) (Fig. 6A). In addition, we found that, in the same animals, the frequency of T_FR (as fraction of T_FH) is negatively correlated with the level of CD4⁺ T cell proliferation as measured by Ki67 expression (Fig. 6A).

We next performed the same correlation analyses in our cohort of SIV-infected RM. The SIV-infected RM included in these regression analyses included both unvaccinated and animals that were challenged after immunization with an SIVmac239 Gag-, Pol-, and Env-expressing DNA vaccine (with or without GM-CSF) followed by two boosts of an SIV239 Gag-, Pol-, and Env-expressing MVA vaccine. In the SIV-infected RM, similar to what was observed in uninfected animals, the frequency of T_FR (as fraction of T_FH) is negatively correlated with the percentages of both T_FH and GC B cells (Fig. 6B). However, the negative correlation between frequency of T_FR (as fraction of T_FH) and the level of CD4⁺ T cell proliferation as measured by Ki67 expression is not seen in SIV-infected RM (Fig. 6B). The negative correlation between T_FR cells (as a frequency of T_FH cells) and both T_FH and GC B cell frequencies is consistent with the hypothesis that T_FR cells play a role in regulating T_FH and GC responses under normal circumstances and in the setting of chronic SIV infection.

To further define the effect of SIV infection on T_FR, we next compared the transcription profiles of T_FR that were isolated from unvaccinated, chronically SIV-infected and uninfected RM (Fig. 7). We performed RNA-Seq analysis, and transcripts that were significantly differentially expressed in T_FR sorted from SIV-infected versus uninfected RM were analyzed by Ingenuity Pathway Analysis. Unsurprisingly, a large proportion of genes induced during SIV infection in T_FR (CD3G, FOS, CD4, ZAP70, PIK3CD, STAT3) were components of T cell proliferation, activation of T cell effector function, and costimulatory activation (data not shown). The enhanced T cell activation was consistent with our observation that T_FR cells express higher levels of the proliferation marker Ki67 compared with T_REG (Fig. 4D). We also observed that several genes implicated in pathways regulating apoptosis or cell-cycle control were perturbed in SIV-infected RM. Of particular interest was the observation that the proapoptotic gene FASLG was >100-fold induced, whereas the antiapoptotic regulator XIAP was significantly downregulated. This latter finding was again validated by RT-PCR (Supplemental Fig. 2B). Thus, although we observed a significant increase of the proliferation marker Ki67 in T_FR after SIV infection (Fig. 4D), a proapoptotic shift of gene expression may partly explain why only a modest increase in T_FR frequency was observed (Fig. 4A).

T_FH require IL-6 signaling and STAT3 expression for differentiation and, once mature, produce several factors that support B cell activation. Conversely, IL-2R signaling drives STAT5 to activate Blimp1/PRDM1, which ultimately blocks T_FH differentiation (30). However, T_FR express both Blimp/PRDM1 and Bcl-6. In this study, we find that both IL-6 and IL-2 signaling genes are enriched in T_FR after SIV infection. However, several of these genes, such as MAPK1, are common to different cytokine signaling pathways, thus making it difficult to establish whether SIV infection causes a shift in the T_FR/T_FH differentiation pressure. Interestingly, downstream signaling for IL-10, a regulatory cytokine produced by both T_FR and T_REG, is also enriched in T_FR postinfection. ICOS-ICOSL signaling was also enriched in T_FR postinfection. These data suggest that T_FR may be engaged in similar T_FH-like cell-surface receptor–ligand interactions with B cells. In addition to genes that were identified with differential expression without any a priori knowledge, we also examined genes with known function in T_FR and T_REG. Postinfection, T_FR show a significant increase in the expression of PD1, IL-6R, SLAMF6, and CD84, that is, all markers associated with T_FH differentiation and function (Fig. 7B). We also found a significant decrease in STAT3 and IL-2RA in T_FR after SIV infection and a nonsignificant decrease in Bcl-6 expression. Finally, as expected, we also observed several other changes in expression patterns of the T_FH and T_FR signature gene sets as we had previously determined (Supplemental Fig. 1). Overall, these data indicate a complex remodeling of gene expression in T_FR post SIV infection of RM.

T_FH are critical to the development of the humoral response to infections, and their role in the setting of HIV and SIV infection (and vaccination) is the subject of intense investigation. However, some aspects of the complex T_FH response to HIV/SIV infection remain poorly understood, including: 1) their role in promoting the development of broadly neutralizing HIV/SIV-specific Abs, and 2) their role in the immunopathogenesis of the infection. In particular, the mechanisms by which T_FH accumulate during the chronic stage of infection despite high levels of direct virus infection are unclear. Importantly, a series of recent studies have shown that T_FH include a subset of cells that are derived from thymic T_REG precursors, express the classical T_REG markers (i.e., FOXP3 and CD25, as well as low levels of CD127), and acquire T_FH markers (i.e., PD1, CXCR5, and Bcl-6) while migrating into the GC of lymph nodes, where they are thought to act as regulators of the host humoral immune response. To the best of our knowledge, this study, together with the independently generated set of data that are included in the accompanying article by the group of Franchini and Vaccari (31), represents the first description of the main features of T_FR in a nonhuman primate species. In this work, we also investigated the dynamics of this cell subset during SIV infection of RM.

The main findings of this study are the following: 1) T_FR show distinct yet overlapping phenotype as compared with T_FH and T_REG based on a combination of flow cytometric, histological, and transcriptional analyses; 2) in healthy, SIV-uninfected RM, the frequencies of T_FR are negatively correlated with the levels of both T_FH and GC B cells; 3) post SIV infection, the T_FR/T_FH ratio is reduced; and 4) T_FR sorted from SIV-infected RM harbor comparable levels of cell-associated viral DNA as compared with T_FH and T_REG. Collectively, these data indicate that although T_FR closely resemble T_FH in several aspects, they are also clearly distinguishable from this cell subset in terms of both their immunophenotype and transcriptional profile. It is therefore important that, in future studies of T_FH, a distinction be made between T_FR and true “non-T_FR” T_FH to fully take into account the complexity of the different CD4⁺ T cell subsets that are present in the GC.

The observation that T_FR express proteins typically expressed by T_FH, such as CD40L, as well as proteins typically expressed by T_REG, like IL-10 and CTLA4, is consistent with studies in mice showing that T_FR are thymic-derived T_REG that migrate into the follicle and, in a manner similar to T_FH, upregulate CXCR5, Bcl-6, and PD1 in a B cell–dependent manner (24). The production of IL-21 by T_FR is an intriguing new finding and suggests that T_FR play a more complex role in RMs than has been described in mice. In addition, principal component analysis suggests that the transcriptional profile of T_FR tends to be more similar to T_FH than T_REG. Further studies are required to fully investigate the functional role played by T_FR in RMs and, specifically, in the context of SIV infection. The finding that >90% of T_FR express Helios as measured by flow cytometry is also consistent with the natural T_REG origin of these cells. Importantly, these immunophenotypic and RNA-Seq data were complemented by histological analyses showing that T_FR are found within GCs of both uninfected and infected RM. Although CD4⁺FOXP3⁺ T_REG were found in abundance outside the GC, CD4⁺PD1⁺ T_FH and CD4⁺PD1⁺FOXP3⁺ T_FR were both only seen within the GC. The hypothesis that T_FR play an immune-regulatory role in vivo is supported by the observation that their frequency is inversely correlated with the levels of both T_FH and GC B cells. These data are consistent with mouse studies indicating that: 1) T_FR suppress T_FH in vitro and prevent the outgrowth of non–Ag-specific B cells (13); and 2) T_FR may inhibit Ab production without an effect on T cell activation, thus indicating an ability to directly regulate B cells (32).

In the setting of in vivo SIV infection, we found that T_FR exhibited only a small increase in their frequency within the total CD4⁺ T cell pool. In conjunction with the large expansion of T_FH that is consistently observed post SIV infection, the minor expansion of T_FR leads to a significantly decreased T_FR/T_FH ratio in the lymph nodes of chronically SIV-infected RM. We confirmed this trend in the ratio of T_FR/T_FH cells after SIV infection by quantifying the number of T_FH and T_FR cells by immunohistochemistry (Supplemental Table I). Although the present set of results does not allow us to determine whether and to what extent the kinetics of T_FH and T_FR during SIV infection are mechanistically linked, it is conceivable that the limited T_FR expansion facilitates progressive accumulation of T_FH in chronically SIV-infected RM, thus indirectly contributing to the aberrant immune activation that characterizes this pathogenic infection. In contrast, it is also possible that the strong proliferation of T_FH and associated increase in PD1 expression post SIV infection hampers the development or differentiation of T_FR as suggested previously (32).

Comparison of the transcriptional profiles of T_FR cells before and post SIV infection showed a significant upregulation of transcripts typically expressed by activated T_REG including FOSB, FABP5, USP2, and USP13 (data not shown) (33), thus suggesting that T_FR may be involved in the generalized immune activation associated with pathogenic SIV infection. Interestingly, we also observed a downregulation of several T_FH and T_REG signature genes as established by our own algorithm. This somewhat unexpected observation indicates that SIV infection has a complex effect on the in vivo phenotype and function of T_FR. A better understanding of the overall contribution of T_FR to the immunopathogenesis of AIDS, in terms of causing either the virus-associated B cell dysfunction or the changes in the lymph node architecture and function, will require further in vitro and in vivo investigation of the suppressive effect by these T_FR on the function of either T_FH or GC B cells.

Although CD4⁺ T cells are the main target for HIV and SIV infection, substantial differences exist between various CD4⁺ T cell subsets in terms of their relative levels of direct virus infection in vivo (34–36). In this study, we tested the possibility that the decrease in the T_FR/T_FH ratio observed during SIV infection of RM was associated with higher level of virus infection in T_FR as compared with T_FH. However, our comparative analysis of the cell-associated viral burdens in sorted T_FR, T_FH, and T_REG revealed similar levels of SIV-DNA in the three CD4⁺ T cell subsets, even though T_FR exhibited higher levels of the SIV coreceptor CCR5 as compared with the other two subsets.

In summary, to our knowledge, the presented data provide the first comprehensive description of T_FR in healthy, uninfected RM, as well as the first examination of the kinetics of these cells in the setting of pathogenic SIV infection. These results support the hypothesis that these cells play an important immune-regulatory role in vivo, and that a relative decline of the T_FR/T_FH ratio may be involved in establishing a state of chronic immune activation in the B cell areas of lymph nodes during pathogenic HIV and SIV infection.

We thank Drs. Barbara Cervasi and Kiran Gill at the Flow Cytometry Core (Emory Vaccine Center) and Drs. Prachi Sharma and Deepa Kodandera at the Molecular Pathology Core Lab (Yerkes National Primate Research Center). We also thank the animal care and veterinary staff at the Yerkes National Primate Research Center.

The authors have no financial conflicts of interest.