Differential Dynamics of Regulatory T-Cell and Th17 Cell Balance in Mesenteric Lymph Nodes and Blood following Early Antiretroviral Initiation during Acute Simian Immunodeficiency Virus Infection

Tregs contribute to SIV/HIV disease progression by inhibition of antiviral specific responses and effector T-cell proliferation. Tregs also cause tissue fibrosis via transforming growth factor β1 production and collagen deposition, which are associated with microbial translocation and generalized immune activation. Early ARV initiation upon viral exposure is recommended globally and results in improved immune function recovery and reduced viral persistence. Here, using an acute SIV infection model of rhesus macaques, we demonstrated for the first time that despite clear improvements in mucosal CD4 T cells, in contrast to blood, Treg frequencies in MLNs remained elevated following early ARV initiation. The particular Th17/Treg balance observed in MLNs can contribute, in part, to the maintenance of mucosal fibrosis during suppressive ARV treatment. Our results provide a better understanding of gut mucosal immune dynamics following early ARV initiation. These findings suggest that Treg-based treatments could serve as a novel immunotherapeutic approach to decrease gut mucosal damage during SIV/HIV infections.

maintaining gut mucosal immune responses, remains understudied. We therefore assessed here the dynamic of Tregs in these tissues of SIV-infected rhesus macaques (RMs) after early ARV initiation.

Early ARV initiation results in total CD4 T-cell recovery and decreased levels of T-cell immune activation in both blood and MLNs.
We recently demonstrated an early depletion of memory CD4 T cells in MLNs (29) and, as expected, our results showed that early ARV treatment preserves CD4 T cells in both blood and MLNs (Fig.  1). Since SIV/HIV infections promote T-cell immune activation (8,30), we then evaluated the effect of early ARV initiation on T-cell immune activation by assessing the expression of HLA-DR and CD39, an ectonucleotidase that converts ATP into immunosuppressive adenosine in concert with CD73, both known as markers of T-cell activation (13,(31)(32)(33). Consistent with the drastic reduction of plasma viral load in ARV-treated RMs (Table 1), we observed a significant decrease in CD8 T cells expressing HLA-DR ( Fig.  2A and B) and CD39 ( Fig. 2C and D), both in peripheral blood and MLNs, compared to untreated animals in acute and chronic phases. Interestingly, despite a rebound in viremia after ARV interruption, the level of CD8 T-cell immune activation remains low both in the blood and MLNs. Similarly, HLA-DR expression by CD4 T cells in MLNs followed the same tendency as CD8 T cells described above, whereas no change was observed in frequencies of CD4 ϩ CD39 ϩ in the MLNs of different study groups (Fig. 2E to H). Altogether, these results demonstrate that early ARV initiation decreases T-cell immune activation in both blood and MLNs.
Early ARV initiation restores levels of memory CCR6 ؉ Th17 cells in both blood and MLNs. It is well documented that a subset of Th17 cells, known as memory CCR6 ϩ T cells, are depleted rapidly following SIV/HIV infections (5,9,11). We therefore assessed the effect of early ARV treatment on the dynamics of memory CCR6 ϩ Th17 cells. As expected, a rapid decline has been observed in peripheral memory CCR6 ϩ Th17 cells in untreated RMs starting from the acute phase of SIV infection ( Fig. 3A and B). Accordingly, these cells are also low in MLNs of untreated RMs in both acute and chronic phases. Importantly, early ARV treatment resulted in significant memory CCR6 ϩ Th17 cell recovery compared to untreated SIV-infected RMs in acute and chronic phases in both compartments. Interestingly, the levels of this population in the blood were even higher in ARV-treated RMs compared to healthy RMs and remain elevated in RMs in which ARV has been interrupted (Fig. 3B). To functionally confirm these data, we measured the in vitro production of IL-17a in cells isolated from the MLNs. We observed, here again, a decline in IL-17a-expressing CD4 T cells following acute SIV infection which was recovered by early ARV (Fig. 3C). We observed higher levels of IL-17aexpressing cells in ARV-treated RMs compared to untreated SIV-infected RMs and, of importance, our results demonstrated a drastic decrease in IL-17a-expressing CD4 T cells after ARV interruption (Fig. 3C). Since the contribution of IL-17a-expressing CD4 ϩ FoxP3 ϩ in inflammatory intestinal diseases has been previously reported (34), we also assessed the expression of IL-17 in these cells, as shown in Fig. 3D. A very small proportion of CD4 ϩ FoxP3 ϩ cells produced IL-17a, a level which remains similar in the different study groups. Altogether, these results demonstrate that early ARV initiation preserves memory CCR6 ϩ Th17 cells.
Treg frequencies in MLNs remain unchanged despite early ARV initiation. Both acute and chronic SIV/HIV infections are associated with increases in total Treg (CD4 ϩ Foxp3 ϩ ) and CD39 ϩ Treg (CD4 ϩ Foxp3 ϩ CD39 ϩ ) frequencies in peripheral blood and GALT of untreated individuals which contribute to a poor effector T-cell-specific immune response, GALT fibrosis, and dysfunction and disease progression (4, 13, 14, 23, 33, 35, 36). Here, we observed that the frequencies of Tregs  and of CD39 ϩ Tregs in the blood of SIV-infected RMs increased during both acute and chronic phases of infection (Fig. 4). Early ARV treatment decreased the frequencies of blood Tregs and CD39 ϩ Tregs compared to acutely SIV-infected RMs.
Paradoxically, compared to blood, our results indicated that early ARV treatment did not decrease the frequencies of Tregs and CD39 ϩ Tregs within MLNs (Fig. 4B to D). The Th17/Treg ratio is commonly used as an indicator of gut mucosal immunity. Here again, our results demonstrate that early ARV initiation was able to recover the memory CCR6 ϩ Th17/Treg ratio only in blood and not in MLNs (Fig. 5A). To gain further insight into the dynamics of the events, we compared available matched blood and MLN specimens of early ARV-treated animals to measure Treg and CCR6 ϩ Th17 cell frequencies. We observed systematically higher Treg frequencies and lower memory CCR6 ϩ Th17 frequencies and CCR6 ϩ Th17/Treg ratios in the MLNs compared to blood (Fig. 5B). We and others have previously reported that the tryptophan-catalyzing enzyme IDO-1 is involved in the degradation of ROR␥t, the Th17 cell transcription factor, resulting in changes in the Th17/Treg balance in favor of Tregs, in line with SIV/HIV disease progression (24,25,37). In line with the lack of significant improvement in the memory CCR6 ϩ Th17/Treg ratio, no effect of early ARV initiation was observed on IDO-1 expression in MLNs, and a negative correlation was observed between total  Treg (CD4 ϩ FoxP3 ϩ ) frequencies and ROR␥t mRNA expression ( Fig. 5C and D). Overall, these results indicated that early ARV initiation reduced Treg frequencies and significantly improved the memory CCR6 ϩ Th17/Treg ratio only in the blood and not in the MLNs of SIV-infected RMs.
Early ARV initiation decreases Treg thymic output but does not affect Treg accumulation within the MLNs. Tregs may originate either directly from the thymus (thymic Tregs) or from naive CD4 T cells in the context of inflammation (14). To evaluate the origin of Treg generation and to explain the differences in Treg dynamics between blood and MLNs, we assessed the expression of CD31 and Helios on Tregs. CD31 is a marker for recently migrated CD4 T cells from the thymus (37), whereas Helios is a specific marker of thymic Tregs (38,39). Only in blood did SIV infection induce a rapid increase in the frequency of recently thymic migrated CD31 ϩ Tregs during the acute phase compared to uninfected RMs; this effect was normalized following early ARV treatment. In contrast to blood, a decrease in the frequencies of CD31 ϩ Tregs was observed in the MLNs (Fig. 6A to C), which follows the total CD4 dynamics within MLNs, as shown in the Fig. 1. We then showed a significant decrease in Helios expression by Tregs in the blood of treated SIV-infected RMs versus untreated RMs in the acute phase of infection. Of interest, we observed that the frequencies of Helios ϩ Tregs remained unchanged in MLNs regardless of ARV initiation (Fig. 6A, D, and E). In treated animals, we observed higher frequencies of CD31 ϩ Tregs in the MLNs versus blood but no difference in Helios ϩ Tregs between the two compartments ( Fig. 6F), suggesting that in MLNs, Tregs are differentiated locally from newly generated thymic CD4 T cells, which are homing to this tissue. In agreement with this observation, we also documented higher Treg activation (HLA-DR ϩ ) in the MLNs of SIV-infected animals, which were not affected by early ARV treatment (Fig. 7A). In addition, a very small portion of Tregs coexpress Helios and HLA-DR (Fig. 7B).
Early ARV treatment decreases the frequencies of CTLA-4 ؉ PD-1memory CD4 T cells only in blood. Recently, a new subset of memory CD4 ϩ T cells expressing CTLA-4 ϩ PD-1that shares characteristics with Tregs contributing to viral persistence in lymph nodes has been described (21). Here, we observed that SIV infection induced both peripheral blood CTLA-4 ϩ PD-1 -CD45RA -CD4 ϩ and FoxP3 ϩ CTLA-4 ϩ PD-1 -CD45RA -CD4 ϩ cells during the acute and chronic phases of SIV infection (Fig. 8). Early ARV initiation significantly decreased the level of these subsets, reaching normal frequencies compared to healthy RMs. In contrast to peripheral blood, the frequencies of both populations remained unchanged in the MLNs within all study groups. Therefore, early ARV initiation decreased the levels of CTLA-4 ϩ PD-1 -Tregs only in peripheral blood and not in the MLNs of SIV-infected RMs.
TGF-␤1 and collagen-1 mRNA expression persists in MLNs despite early ARV treatment. It is well known that TGF-␤1 production by Tregs in mucosal tissues starts very rapidly after SIV/HIV infection, resulting in collagen deposition and tissue fibrosis; this is one of the main obstacles in the recovery of gut mucosal immunity following ARV (23,28,40). Our results showed that early ARV initiation did not improve the expression of the GALT fibrosis markers TGF-␤1 and collagen-1 mRNA within the MLNs (Fig. 9A and  B). Accordingly, we observed a positive trend between TGF-␤1 and collagen-1 expression (Fig. 9C).

DISCUSSION
It is well known that GALT represents a major site of SIV/HIV pathogenesis and immune dysfunction during suppressive ARV treatment (2,41). Increased frequencies and the immunosuppressive functions of Tregs are hallmarks of SIV/HIV infections, and these cells are involved in GALT fibrosis via TGF-␤1 production, as well as the inhibition of antivirus-specific immune responses (13,14,42,43). The initiation of ARV treatment upon viral exposure or diagonosis is recommended in clinical practice, since it results in a better CD4 recovery and nadir, a smaller reservoir size, and a reduction in immune activation (2,27). Here, for the first time, we assessed the impact of early ARV initiation at day 4 postinfection on Treg dynamics in the MLNs versus the blood of RMs. The early time point of ARV initiation at day 4 has been previously shown in RMs to suppress peripheral VL, while SIV DNA has been already seeded within their MLNs (44). Our results demonstrated that, despite improvements in the levels of immune activation and recovery of memory CCR6 ϩ Th17 cells in both blood and MLNs after early ARV initiation, the dynamics of the Tregs were distinct. We observed the accumulation of total Tregs and various subsets of Tregs, including CD39 ϩ Tregs, thymic Tregs, and CTLA-4 ϩ PD-1 -Tregs, remained unchanged within the MLNs compared to blood. Accordingly, early ARV initiation failed to decrease the expression of TGF-␤1 and collagen-1 as markers of GALT fibrosis.
SIV/HIV infections drive generalized immune activation, even under suppressive ARV treatment (8,30,41,45). In addition to HLA-DR as a marker of T-cell activation, the ectonucleotidase CD39 is a functional marker of T-cell activation and, in concert with CD73, hydrolyzes inflammatory ATP into ADP and finally into immunosuppressive adenosine (32,46). CD39/adenosine pathways have been reported by us and others to be involved in disease progression during SIV/HIV infections (33,35,47). As expected, we observed higher levels of T-cell immune activation in untreated SIV-infected RMs during both acute and chronic phases of infection. Early ARV initiation decreased the levels of immune activation considerably, along with a decline in the peripheral viral load. Our results are in line with previous reports about the effect of early treatment during the acute phase on the control of immune activation levels in the periphery and in the GALT of HIV-infected individuals (48)(49)(50). In HIV infection, ARV are able to partially to completely normalize the frequencies of Tregs (19,51,52), but the effect of very early ARV initiation on Treg dynamics, specifically within MLNs, has been understudied. Consistent with previous reports, we observed an increase in both total Tregs and CD39 ϩ Tregs in the blood of SIV-infected animals in acute infection and an even greater increase in chronic stages of infection. Importantly, in contrast to peripheral blood, early ARV initiation failed to decrease the levels of total Tregs and CD39 ϩ Tregs, and  even a slight increase in the percentages of these populations was observed in MLNs. This could be due to the migration of CCR6 -Tregs into the GALT following ARV initiation, as previously reported in chronic HIV infection (53).
The accumulation of Tregs and higher levels of TGF-␤1 within the GALT results in tissue fibrosis via collagen-1 deposition, which begins in the acute phases of SIV/HIV infections (23,40). Importantly, GALT fibrosis occurs even in HIV elite controllers, despite their natural suppression of the viremia (26). We showed that early ARV treatment did not impact the levels of both TGF-␤1 and collagen-1 mRNA expression, and a strong positive correlation was observed between TGF-␤1 and collagen-1 expression. Importantly, we have previously reported that in the GALT, TGF-␤1 impairs the cytotoxic functions of CD8 T cells and induces their apoptosis, resulting in viral dissemination within the MLNs (4). Therefore, while this remains to be confirmed by immunohistochemistry, our results suggest a link between persistent Treg homing in the MLNs and consequently TGF-␤1 production and collagen deposition within the MLNs, despite a decrease in T-cell immune activation after early ARV initiation.
Several sudies have shown that memory CCR6 ϩ Th17 cells are highly permissive to SIV/HIV infections, and a rapid depletion of these cells happens rapidely following infection (5-7, 11, 54). CCR6, the CCL20 and ␤-defensin receptor, regulates the migration of T cells to the GALT (55) and is widely used as a marker of Th17 cells (5,11,41). The Th17/Treg ratio is frequently used as an indicator of GALT immunological functions. Our results showed that early ARV initiation restored the levels of memory CCR6 ϩ Th17 cells in both the periphery and the MLNs, even at higher levels than in noninfected animals. These dynamics have been also confirmed by the assessment of IL-17aproducing cells from the MLNs. This was expected since it has been already demonstrated that ARV treatment restores CD4 ϩ T memory stem cell homeostasis in SIVinfected RMs (56). Accordingly, following early ARV initiation, the memory CCR6 ϩ Th17/Treg ratio was only restored in blood but remained unchanged in the MLNs in line with the accumulation of Tregs in this tissue. Of particular importance, in matched specimens from early ART-treated animals, systematic higher Treg frequencies and lower memory CCR6 ϩ Th17 frequencies were observed in the MLNs compared to blood, which demonstrates clearly their differential dynamics in these tissues. A recent study showed that the later initiation of ARV treatment at 6 weeks postinfection was not able to recover the Th17/Treg ratio in the GALT during the acute SIV infection of RMs (57). The consistent increased frequencies of Tregs in MLNs could be linked to IDO-1 enzyme expression (4). IDO-1 catabolizes tryptophan into immunosuppressive kynurenine and promotes a Th17/Treg imbalance in favor of Tregs via the degradation of ROR␥t during HIV infection, as reported by us and others (18,58,59). In addition, IDO-1 is highly expressed by macrophages, and accumulation of these cells in the GALT and MLNs has been reported during SIV infection (4,60). Our results showed that, along with the subsequent increase in Tregs in MLNs, IDO-1 mRNA expression was not decreased following early ARV treatment. Altogether, these findings demonstrate that despite decreased immune activation and recovery of memory CCR6 ϩ Th17 cells, the accumulation of Tregs within the MLNs seems to be not reversible.
To better evaluate the origin of Tregs in MLNs, we assessed the expression of both CD31 and Helios as markers of recently thymic migrated CD4 T cells and thymic Tregs, respectively (37)(38)(39). We observed an increase in CD31 ϩ Tregs during acute SIV infection in the absence of ARV in line with viral load and immune activation, whereas ARV normalized the frequencies of recently migrated CD31 ϩ Tregs in contrast to their dynamics in the MLNs. It has been reported that Helios ϩ Tregs proliferate more in vivo and produce higher levels of TGF-␤1 compared to Helios -Tregs (61), and the expression of Helios ϩ Tregs in blood was correlated with the control of HIV replication (62). Accordingly, we demonstrated that the frequencies of thymic Tregs were decreased only in the blood after early ARV initiation, whereas the frequencies of Helios ϩ Tregs remained unchanged following SIV infection. This observation, together with the higher frequencies of CD31 ϩ Tregs but not of Helios ϩ Tregs in the MLNs, suggests that Tregs within the MLNs could be differentiated locally form the newly generated thymic CD4 T cells, in agreement with the recovery of total CD4 T cells following ART. Our study also demonstrated that SIV infection increased the frequencies of both CD4 ϩ memory CTLA-4 ϩ PD-1and FoxP3 ϩ CTLA-4 ϩ PD-1 -T cells in SIV-infected RMs; these frequencies were normalized by early ARV initiation only in the blood. These CTLA-4 ϩ PD-1subsets share Treg characteristics and were reported to contribute to viral persistence within the peripheral lymph nodes of ARV-treated SIV-infected RMs (21). Here, we extended this initial observation by demonstrating the persistence of this population in the MLNs despite early ARV treatment. Although our findings remain to be confirmed in further studies, these findings suggest that at least a subset of Tregs might contribute to the viral persistence in the GALT despite very early ARV initiation.
In summary, here, we present evidence indicating differential dynamics of Tregs in the blood and MLNs of SIV-infected RMs following early ARV initiation. Indeed, despite the beneficial impact of early ARV initiation within the first few days of SIV infection on immune activation and memory CCR6 ϩ Th17 cells in both blood and MLNs, the frequencies and homing of various subsets of Tregs within the MLNs, as well as markers of tissue fibrosis, were not improved. These findings highlight the role of Tregs in SIV/HIV pathogenesis in GALT and represent Tregs as potential targets for further immunotherapeutic strategies.

MATERIALS AND METHODS
Ethics statement. All RMs were housed at the nonhuman primate facilities of the Laval University, Quebec City, Quebec, Canada, in accordance with the rules and regulations of the Canadian Council on Animal Care (http://www.ccac.ca). This protocol was approved by the Laval University Animal Protection Committee (no. 106004). Animals were fed by standard monkey chow diet supplemented daily with fruit and vegetables and water ad libitum. Social enrichment was delivered and overseen by a veterinary staff, and overall animal health was monitored daily. Animals showing significant signs of distress, disease, and weight loss were evaluated clinically and were humanely euthanized, using an overdose of barbiturates, according to the guidelines of the Veterinary Medical Association.
Animals, infection protocol, and ARV cocktail. A total of 32 female RMs, seronegative for SIVmac, STLV-1 (simian T leukemia virus type 1), SRV-1 (type D retrovirus), and herpes B viruses were enrolled in this study (Table 1). A total of 25 animals were infected intravenously with 20 50% animal infectious doses of SIVmac251 virus, and the specimens were collected in both acute and chronic phases of infection. To evaluate the effect of early ARV initiation, seven monkeys were treated 4 days after the infection in a daily manner with an ARV cocktail that included tenofovir (20 mg/kg), emtricitabine (20 mg/kg), indinavir (2 mg/kg), ritonavir (20 mg/kg), and raltegravir (20 mg/kg). After 8 weeks of ARV treatment, the therapy was interrupted in three RMs. Two groups of nontreated animals in both acute (n ϭ 13) and chronic (n ϭ 4) phases were also assessed. To assess the uninfected baseline status, blood specimens were obtained from 10 animals 3 days before the SIVmac251 infection. MLN specimens from two uninfected animals were also assessed to establish a baseline status in the MLNs.
Cell isolation. At the indicated time points (Table 1), peripheral blood and MLNs were collected upon euthanasia. Blood specimens were collected in EDTA tubes, and staining for flow cytometry analysis was performed using whole blood. MLNs were mechanically disrupted and grinded over a 70-m cell strainer to avoid any collateral effect of enzymatic digestion by collagenase or other proteases on the expression of surface markers. The whole blood specimens obtained without euthanasia were frozen, and flow cytometric analyses were performed in batch on frozen samples. Ex vivo flow cytometric analysis. Multicolor flow cytometry was performed using whole blood and MLN cells. Predetermined optimal concentrations of fluorochrome-conjugated antibodies were used for the staining, as detailed in Table 2. Dead cells were excluded from the analysis using a Live/Dead Fixable Aqua dead cell stain kit (Invitrogen). After the extracellular staining, FoxP3, Helios, and CTLA-4 staining was performed following permeabilization with a transcription factor buffer set (BD Bioscience). Flow cytometry acquisition was performed on a three-laser BD Fortessa-X20 cytometer, and results were analyzed by using FlowJo v10.2.
IL-17A production following in vitro stimulation. Isolated cells from MLNs were stimulated with phorbol myristate acetate (PMA) at 50 ng/ml and ionomycin (1 g/ml) for 4 h, followed by incubation in the presence of brefeldin A and monensin (both from BD Bioscience) for 4 h. The intracellular production of IL-17a and gamma interferon (IFN-␥) was assessed by flow cytometry. Viral RNA quantification. Viral loads in the sera of SIV-infected RMs were quantified by quantitative real-time PCR (RT-PCR) using a PureLink viral RNA/DNA kit (Invitrogen). A plasmid encoding the gag gene of SIVmac251 was used as a standard. Amplifications were carried out with a 7500 real-time PCR system (Applied Biosystems) using the following parameters: 50°C/5 min, 95°C/20 s, and 40 cycles of 95°C/15 s and 60°C/1 min. Samples were run in duplicates, and results are expressed as SIV RNA copies/ml, with a limit of detection of 40 copies/ml (63).
Quantification of mRNA expression by RT-PCR. Total RNA was extracted from snap-frozen MLN tissues obtained upon euthanasia using an AllPrep DNA/RNA/miRNA Universal kit (Qiagen, Germany). cDNA was then synthetized using M-MLV reverse transcriptase (Invitrogen) on total RNA (140 ng) according to the manufacturer's protocol. The specific primers used for the detection of TGF-␤1, collagen-1, IDO-1, ROR␥t, and housekeeping gene ACTB are described in Table 3. Reactions were done in duplicate using a LightCycler 480 Instrument II (Roche, Switzerland) with a LightCycler 480 SYBR green I Master (Roche, Germany). The cycling conditions were as follows: 5 min at 95°C; 40 amplification cycles of 15 s at 95°C, 20 s at 59°C, and 20 s at 72°C; 1 cycle of 5 s at 95°C and 30 s at 65°C; and finally 10 s at 40°C. We generated a negative control without cDNA to confirm that no contamination occurred. The negative control yielded no amplification or a crossing point (Cp) value up to or equal to 35. The expression of each gene was quantified relative to the housekeeping gene ACTB.
Statistical analysis. Prism v6.01 (GraphPad, San Diego, CA) was used for statistical analyses. The results are presented as medians with interquartile ranges. Differences among the different groups of animals were determined by using a Mann-Whitney rank test for unpaired variables. A Pearson rank correlation test was used to identify the associations among study variables. A Wilcoxon matched-pair signed-rank test was used to compare paired variables. Statistical significances (P Յ 0.05) are indicated in the figures.