Lymph node migratory dendritic cells modulate HIV-1 transcription through PD-1 engagement

T-follicular helper (Tfh) cells, co-expressing PD-1 and TIGIT, serve as a major cell reservoir for HIV-1 and are responsible for active and persistent HIV-1 transcription after prolonged antiretroviral therapy (ART). However, the precise mechanisms regulating HIV-1 transcription in lymph nodes (LNs) remain unclear. In the present study, we investigated the potential role of immune checkpoint (IC)/IC-Ligand (IC-L) interactions on HIV-1 transcription in LN-microenvironment. We show that PD-L1 (PD-1-ligand) and CD155 (TIGIT-ligand) are predominantly co-expressed on LN migratory (CD1chighCCR7+CD127+) dendritic cells (DCs), that locate predominantly in extra-follicular areas in ART treated individuals. We demonstrate that TCR-mediated HIV production is suppressed in vitro in the presence of recombinant PD-L1 or CD155 and, more importantly, when LN migratory DCs are co-cultured with PD-1+/Tfh cells. These results indicate that LN migratory DCs expressing IC-Ls may more efficiently restrict HIV-1 transcription in the extra-follicular areas and explain the persistence of HIV transcription in PD-1+/Tfh cells after prolonged ART within germinal centers.


Introduction
One of the major obstacles to HIV-1 eradication resides in the capacity of HIV-1 to rapidly establish a latent reservoir, transcriptionally silent, which is not susceptible to both the host immune response and cART [1][2][3][4][5][6].Different cell lineages including CD4 T cells and monocytes/macrophages [7][8][9] may contribute to the HIV-1 reservoir.Central memory and transitional memory CD4 T cells serve as major cellular compartments of the latent HIV-1 reservoir in blood [6].More recently, blood memory CD4 T cells with stem-cell like properties [10] or expressing CXCR3 and/or CCR6 were also shown to contain latently HIV-infected cells [11][12][13].However, blood contains only 2% of the total lymphocytes that reside predominantly within lymphoid organs [14], and lymphocyte populations within the lymphoid tissues may be phenotypically and functionally distinct from those in blood [15] with differential cell composition and extensive cell heterogeneity regarding T follicular helper (Tfh) cells [16].Notably, previous studies have demonstrated that lymphoid organs are the major anatomic site for HIV infection, production and spreading and that high concentration of virus particles and CD4 T cells with active virus replication were predominantly restricted to germinal centers (GCs) in viremic individuals [17][18][19].
Interestingly, recent studies have underscored that while HIV or SIV DNA containing CD4 T cells were consistently detected in cells located in lymph node (LN) follicular and extra-follicular areas in long-term cART treated individuals [11,20,21], HIV or SIV RNA detection were mainly restricted to CD4 T cells in LN GC areas of HIV viremic controllers [22], SIVinfected elite controller macaques [23] or cART-treated aviremic HIV-infected individuals [24].Of note, Tfh cells represent the major cellular compartment for HIV production and replication in viremic individuals [25,26] and the major CD4 T cell population for persistent HIV-1 transcription in long-term treated individuals [24] as compared to any other blood or LN memory CD4 T cell populations.The limited access of cytotoxic CD8 T cells to GCs [23], the sub-optimal antiretroviral drug penetration in lymphoid tissues [27] and the higher level of activation of Tfh cells [26] are possible explanations for lymphoid organs serving as primary anatomic sites for HIV infection and persistence.
Multiple cellular mechanisms are involved in the establishment and the maintenance of HIV-1 latency including 1) epigenetic silencing induced by histone deacetylation and DNA methylation [28,29], 2) limiting cellular levels of the essential Tat cofactor P-TEFb and the transcription initiation factors NF-kB and NFAT [30] and 3) condensed chromatin at the viral long terminal repeat [31].However, under certain circumstances, HIV-1 transcription/ production might be reactivated.The parameters associated with HIV-1 reactivation include T-cell receptor (TCR)-mediated signaling via NF-κB [29,32], cytokine and chemokine stimulations [33] or epigenetic DNA modifications such as acetylation and methylation [34].Interestingly, Fromentin et al. showed that the expression of immune checkpoint molecules (ICs) such as PD-1, LAG-3 and TIGIT on memory CD4 T cells was associated with HIV-infected cells in distinct blood memory CD4 T cell subsets during ART [35], suggesting that IC signaling may contribute to maintain HIV-1 latency in HIV-1 infected memory CD4 T cells [36,37].
Given that previous studies have underscored the involvement of IC/IC-L interactions in the functional impairment of Tfh cells in viremic HIV-infected individuals [38][39][40], we hypothesized that IC/IC-ligand (IC-L) interactions may contribute to modulate HIV latency/ virus reactivation in the LN microenvironment.In the present study, we therefore investigated 1) the expression and distribution of ICs and IC-Ls in blood and LN mononuclear cells isolated from cART treated aviremic, viremic and HIV-uninfected subjects using mass cytometry and in situ in lymph node compartments (germinal centers and extra-follicular zones) by immunohistochemistry, and 2) the impact of IC/IC-L interactions on TCR-mediated T-cell proliferation and HIV-1 transcription/production.
We demonstrate that PD-1 and TIGIT, the two major ICs expressed on Tfh cells ex vivo, are functionally active and regulate TCR-mediated HIV-1 transcription and production in vitro.However, PD-L1 (PD-1-ligand) and CD155 (TIGIT-ligand) were predominantly coexpressed on LN migratory (CD1c high CCR7 + CD127 + ) dendritic cells (DCs) which located mainly in the extra-follicular areas of cART treated subjects.These findings suggest that the strength of the inhibitory signal resulting from the IC/IC-L interactions might be selectively reduced in GCs of ART treated subjects, thus creating a microenvironment less constrained for cell activation and HIV transcription.In support of this hypothesis, we demonstrate that LN migratory DCs via IC/IC-L interactions modulate TCR-mediated HIV-1 reactivation and production from LN PD-1 + /Tfh cells of cART treated HIV-infected individuals and that the levels of HIV-1 transcription in LN memory CD4 T cells correlated with the reduced frequency of LN migratory DCs.
These findings indicate that the IC-L-mediated modulation of HIV-1 transcription in treated subjects is more efficient in extra-follicular areas and underscore that an imbalance in IC/IC-L interactions is a novel mechanism contributing to HIV-1 persistence in LNs.

PD-1 and TIGIT are the two major ICs expressed on blood and LN memory CD4 T-cell populations including T follicular helper cells
We simultaneously collected blood and LN from 10 viremic and 10 cART treated aviremic HIV-1 infected individuals and 7 HIV-uninfected subjects.Mononuclear cells isolated from blood and LNs were then stained with a mass cytometry panel encompassing 38 markers including cell lineage markers, ICs and IC-ligands, i.e.PD-1, CTLA-4, LAG-3, TIM-3, TIGIT, PD-L1, PD-L2 and CD155.The extracellular expression of CTLA-4, LAG-3, TIM-3 and TIGIT was assessed in blood and LN memory CD4 T-cell populations identified on the basis of the expression of PD-1 and/or CXCR5, i.e.CXCR5 -PD-1 -, CXCR5 + PD-1 -, CXCR5 -PD-1 + , CXCR5 int PD-1 int and CXCR5 high PD-1 high CD4 T cells (Fig 1A and 1B).Of note, CXCR5 high PD-1 high CD4 T cells correspond to LN Tfh cells (Fig 1B).Consistent with a previous study [24,26], LN Tfh cells of viremic untreated HIV-1 infected individuals were increased as compared to HIV-uninfected subjects and their percentage dropped after prolonged cART to levels observed in HIV-uninfected subjects (P<0.05)(Fig 1C).
The representative examples and cumulative data indicated that in blood, PD-L1, PD-L2 and CD155 were predominantly expressed on monocytes and to a lower extent on type 2 conventional DCs (cDC2; HLA-DR + CD11c + CD1c + ) and plasmacytoïd DCs (pDC; HLA-DR + CD11c -CD123 + ), while IC-Ls were poorly expressed on blood B-cell populations of HIVuninfected, viremic and aviremic ART treated HIV-1 infected individuals (P<0.05)(S3 Fig and Fig 2A).Of note, no correlation was observed between the frequencies of IC-L expressing blood monocytes with HIV viral load in viremic HIV-infected individuals (S4 Fig).However, although not statistically significant, but a trend towards a negative correlation was observed between the frequencies of PD-L-expressing monocytes and the duration of ART in treated individuals (r = -0.58;P = 0.08) (S4 Fig) , suggesting that the expression of PD-Ls on monocytes may require longer treatment period to normalize.
In LN, IC-Ls were significantly more expressed on CD1c high DCs than on pDCs and LN Bcell populations, including GC B cells of HIV-uninfected, viremic and aviremic ART treated HIV-1 infected individuals (P<0.[41,42].Lymph node CD1c high CCR7 + CD127 + DCs were therefore referred to as LN migratory DCs.
We next investigated the influence of HIV-1 infection and treatment initiation on the frequency of LN migratory DCs of untreated viremic and treated aviremic HIV-1 infected individuals.Of note, non-reactive LNs obtained from HIV-uninfected subjects were used as control.The cumulative data indicated that the frequencies of LN migratory DCs of viremic

HIV-1 infection and cART treatment initiation influences PD-L1 expression and tissue distribution
We then determined whether active and persistent virus transcription detected in LN PD-1 + / Tfh cells [24] may result from reduced IC/IC-L interactions in the GC areas of cART treated HIV-1 infected individuals.For this purpose, the percentage of cells expressing PD-1 and the proportion of PD-L1 positive tissue surface were determined in GC and extra-follicular areas of LN sections collected from untreated viremic and treated aviremic HIV-1 infected individuals using immunohistochemistry staining as previously described [24].Of note, to determine whether the spatial distribution of PD-1 and PD-L1-expressing cells observed in viremic HIV-infected individuals was associated with non-specific immune activation/inflammation or was specifically associated with HIV infection/replication, LN sections collected from HIV-uninfected individuals suffering from lymphadenopathy ("reactive LNs") were used as control.
The representative examples and the cumulative data indicate that PD-1 and PD-L1 expressing cells were detected in both GC and extra-follicular areas of HIV-uninfected subjects, viremic and cART treated HIV-1 infected individuals (Fig 3A -3F).Comparison of serial immuno-labeled sections showed that cells expressing PD-1 were predominantly lymphocytes, while PD-L1 expressing cells were predominantly mononucleated histiocytes or dendritic cells.Cells expressing high levels of PD-1 (PD-1 high cells) were essentially detected in the GC areas.Consistent with previous study [24], the size of GCs/mm 2 and the number of PD-1 high cells/mm 2 of treated aviremic HIV-1 infected individuals were significantly reduced as compared to viremic untreated HIV-1 infected individuals (P<0.05)(Fig 3G and 3H).
Taken together, these results suggest that due to the different distribution in PD-L1 expression between extra-follicular and GC areas, the strength of the inhibitory signal resulting from PD-1/PD-L1 interactions may be weaker within GCs.The initiation of cART further influences the expression of both PD-1 and PD-L1 being associated with significant reduction in both extra-follicular and GC areas.

PD-1/PD-L1 interaction modulates HIV-1 production in LN of treated aviremic HIV-infected individuals
We then investigated whether PD-1 was functionally active on LN PD-1 + /Tfh cells.To address this issue, PD-1 high and PD-1-negative LN memory (CD45RA -) CD4 T cells were sorted from 5 treated aviremic HIV-infected individuals (Fig 4A).Consistent with a previous study, Tfh cells represented about 69% of the sorted PD-1 high memory CD4 T cells (Fig 4B) [24].Sorted cell-populations were labelled with carboxyfluorescein (CFSE) and stimulated with coated anti-CD3/anti-CD28 monoclonal antibodies (MAbs) in the presence or in the absence of PD-L1 recombinant protein.The impact of PD-L1/PD-1 interaction on T-cell proliferation

LN migratory (CD1c high CCR7 + CD127 + ) dendritic cells expressing PD-L1 or PD-L2 modulate HIV-1 transcription/production in LN of treated aviremic HIV-infected individuals
We next explored the potential role of LN migratory DCs expressing PD-L1 and/or PD-L2 to modulate TCR-mediated reactivation of HIV-1 production from LN CD4 T cells expressing or not PD-1.To address this issue, PD-1-positive and PD-1-negative LN memory (CD45RA -) CD4 T cells were sorted from 4 treated aviremic cART treated individuals.Sorted cell-populations were stimulated with coated anti-CD3/anti-CD28 MAbs in the presence or in the absence of autologous LN migratory DCs cultured with or without blocking anti-PD-L1/2 MAbs.Reactivation of HIV-1 production was assessed using HIV-1 RNA detection in day 6 culture supernatants in the presence of emtricitabine.Of note, sorted LN migratory DCs expressed more than 80% of PD-L1 and/or PD-L2, respectively (Fig 5A ).
Interestingly, the frequencies of LN migratory DCs of cART treated HIV-infected individuals inversely correlated with the levels of cell-associated HIV-1 gag RNA found in sorted LN memory CD4 T cells (r = -0.828;P = 0.05) (Fig 5C ).
Taken together, these data demonstrate that LN migratory DCs could modulate HIV-1 transcription in LN of treated aviremic HIV-infected individuals through a mechanism involving PD-L/PD-1 interactions.

Anti-PD-1 monoclonal antibody Pembrolizumab reactivates HIV-1 replication from latently infected CD4 T cells in vitro
Additional evidence in support of the role of PD-1/PD-L1/2 interactions in regulating TCRmediated HIV-1 reactivation was obtained evaluating the efficiency of anti-PD-1 MAb (pembrolizumab) to reactivate HIV-1 from latently infected resting memory CD4 T cells isolated from blood using a "modified" quantitative viral outgrowth assay (QVOA) [44].For this purpose, resting memory CD4 T cells isolated from aviremic long-term treated HIV-1 infected individuals were cultured under different experimental conditions including 1) unstimulated (negative control), 2) stimulated with anti-CD3/anti-CD28 MAbs for 3 days in the presence of IL-2 (positive control), 3) exposed to isotype control, and 4) exposed to clinically relevant concentration of pembrolizumab for 14 days in the presence of autologous irradiated CD8-depleted PBMCs that include IC-L expressing cells.Of note, the use of autologous CD8-depleted PBMCs prevented the increase of transcriptional noise induced by mixed leukocyte reaction generated by the use of heterologous PBMCs and, the irradiation of CD8-depleted PBMCs prevented the reactivation of HIV-1 transcription from the feeder cells [44].The QVOA was performed in the absence of ART to allow amplification of the reactivated virus and in a multiple replicate/limiting dilution format to allow the estimation of frequencies.
The reactivation of HIV-1 replication induced in the various conditions was then assessed by HIV-1 RNA detection in day 14 culture supernatants using Roche Taqman assay as previously described and by the estimation of the frequencies of infected cells [44].The levels of HIV-1 RNA induced under the various conditions were generated using the 5 replicates of the highest cell concentration (5.10 5 cells/condition) of the "modified QVOA".The cumulative data generated from 10 treated aviremic HIV-1 infected individuals indicated that pembrolizumab induced significantly higher levels of HIV-1 RNA in the culture supernatants as compared to untreated cultures or cells exposed to isotype control (P<0.05)(Fig 6A).The frequencies of cells containing replication competent virus were then evaluated in all conditions by the detection of HIV-1 RNA in QVOA supernatants and expressed as RNA-unit per million (RUPM) [45].The results indicate that the average RUPM frequency following pembrolizumab exposure was significantly higher as compared to the unstimulated cells or cells exposed to the isotype control and represented about 14 cells containing replication competent virus per million (P<0.05) (Fig 6B ).
Taken together, these data demonstrate that anti-PD-1 MAbs can efficiently reverse HIV-1 latency in vitro.

Discussion
Increasing number of evidences indicate that B-cell follicles might be anatomical sanctuaries for active transcription in both HIV/SIV viremic controllers [22,23,26] and in ART treated aviremic HIV-infected individuals [24].While multiple mechanisms may be involved in the regulation of HIV transcription, recent studies suggested that IC molecule expression may contribute to control HIV-1 transcription and therefore the maintenance of HIV latency in HIV-infected memory CD4 T cells [35][36][37].Interestingly, we recently showed that Tfh cells expressing high levels of PD-1 serve as a major site of active and persistent virus transcription  (A-B).Statistical significance (P values) was obtained using One-way ANOVA (Kruskal-Wallis test) followed by Wilcoxon Matched-pairs two-tailed Signed Rank test (A) or using Extreme Limiting Dilution analysis (ELDA) (http://bioinf.wehi.edu.au/software/elda/)(B)."US" corresponds to unstimulated, "Iso" corresponds to isotype control, "3/28" corresponds to anti-CD3/anti-CD28 MAbs, "anti-PD-1" corresponds to Pembrolizumab.https://doi.org/10.1371/journal.ppat.1007918.g006 in ART treated aviremic individuals [24].These observations prompted us to investigate the involvement of IC/IC-L interactions in the regulation of HIV-1 transcription in lymph node tissues.
We showed that Tfh cells, predominantly co-expressed PD-1 and TIGIT directly ex vivo in HIV-uninfected, viremic and aviremic ART treated HIV-infected individuals.We subsequently showed that PD-1 and TIGIT expressed on LN CD4 T-cell populations, including PD-1 + /Tfh cells, were functionally active.Of note, in the present manuscript, we cannot exclude that LN PD-1 + /Tfh cells may also encompass T follicular regulatory (TFR) cells.
Analysis of immunostained LN tissue sections indicated that PD-L1 expressing cells were detected in both extra-follicular and follicular regions of untreated viremic HIV-1 infected individuals and ART initiation was associated with a significant reduction of the proportion of PD-L1 positive tissue surface, even more pronounced in GC areas.An in-depth mass cytometry analysis revealed that PD-L1, PD-L2 and CD155 were predominantly co-expressed on a specific LN CD1c high DC subpopulation expressing markers of migratory DCs i.e.CCR7 and CD127 [41,42].Consistent with previous studies, IC-Ls were also detected at low levels on blood monocytes and on blood and LN B-cell populations including GC B cells [38,46,47].Interestingly, the frequency of migratory DCs directly correlated with HIV-1 viral load and was significantly reduced in aviremic ART treated HIV-infected individuals, thus indicating that HIV-1 replication influences the frequency of migratory DCs.
Migratory DCs differ from tissue to tissue but share the capacity to transport antigens to the draining LNs during both homeostatic conditions and infections [42].In lymph nodes, migratory DCs preferentially locate at the T cell-B cell border where they contribute to induce Tfh-dependent antibody responses [48,49].
Multiple cellular parameters such as TCR-mediated signaling, cytokine and chemokine stimulations or epigenetic DNA modifications can trigger reactivation of HIV-transcription [50], which might be in turn modulated by various mechanisms.In the present study, we investigated whether IC-Ls in general and migratory DCs expressing IC-Ls, in particular, could modulate HIV transcription in LN CD4 T cells including Tfh.Consistent with previous study, LN PD-1 + /Tfh cells produced significantly higher levels of HIV-1 RNA than PD-1-negative CD4 T cells [24].In particular, TCR-mediated HIV production of PD-1 + /Tfh cells was strongly reduced in vitro in presence of recombinant PD-L1 or CD155.
More importantly, we demonstrated that LN migratory DCs could modulate TCR-induced HIV-1 transcription in LN by a mechanism involving PD-L/PD-1 interactions.Indeed, migratory DCs specifically regulated TCR-mediated HIV production of PD-1-positive CD4 T cells and anti-PD-L1/2 blocking MAb treatment partially restored the levels of HIV-1 RNA detected in PD-1-positive CD4 T-cell supernatants only.These data indicate that LN migratory DCs expressing IC-Ls may more efficiently restrict HIV-1 transcription in the extra-follicular areas versus GCs.In addition, the frequency of migratory DCs inversely correlated with HIV-1 transcription from LN memory CD4 T cells, suggesting that IC-L expressing migratory DCs might contribute to control HIV-1 transcription and maintain HIV-1 latency in extra-follicular areas.Notably, recent studies indicated that LN migratory DCs may harbor tolerogenic properties by a mechanism involving the production of immunosuppressive cytokines such as TGF-β and IL-10 and the induction of regulatory T cells [51][52][53].
Finally, we postulated that inhibition of PD-1/PD-Ls interactions might reverse HIV-1 latency and evaluated the efficiency of anti-PD-1 MAbs (pembrolizumab) to reactivate HIV-1 from latency using a "modified" QVOA.We demonstrated using primary CD4 T cells isolated from aviremic ART treated HIV-infected individuals and co-cultured with autologous irradiated CD8-depleted PBMCs, that clinically relevant concentration of pembrolizumab could efficiently reverse HIV-1 latency in vitro.While the precise mechanism regulating HIV latency in blood remains to be established, one may postulate that IC-L expressing monocytes may contribute to effects observed upon PD-1 blockade in vitro.These data support the recent observations suggesting the involvement of PD-1 in the establishment of HIV latency [36].In addition, while one study found no effect [54], two other studies underscored the potential of PD-1 blockade in HIV cure strategies in combination or not with additional latency reversing agents [36,55].Therefore, immune checkpoint blocking antibodies (ICBs) represent a novel form of latency reversing agent that are currently being explored in the context of HIV functional cure [56].Indeed, ICBs may potentially on one hand reverse HIV latency, thereby allowing for the expression of HIV proteins on the cell surface, and on the other hand, rescue the function of exhausted HIV-specific CD8 T cells to facilitate the elimination of reactivated cells [57].In this regard, a phase II dose-escalation study of anti-PD-L1 antibody therapy in HIVinfected individuals was recently terminated due to safety concerns [58,59].Interestingly however, the data collected indicated an increase in Gag-specific CD4 and CD8 T-cell responses in a fraction of individuals with no changes in levels of cell associated HIV RNA or DNA in blood [58].Unfortunately, no data were collected regarding the impact of ICB treatment on the tissue reservoirs.Taken together, these data indicate that future clinical trials based on ICBs should probably carefully estimate the risk-benefit ratio for HIV-infected individuals on stable suppressive ART and consider a thorough evaluation of the impact of ICB treatment on both blood and tissue reservoirs.
Recent studies suggested that IC signaling may contribute to maintain HIV-1 latency in HIV-1 infected memory CD4 T cells [36,37,55], however, these studies were performed exclusively on cells isolated from blood, while the present study provides four lines of evidence on the role of IC/IC-L interactions in regulating HIV transcription in LN tissues that include: 1) PD-1/PD-L1 interactions strongly impact the reactivation of TCR-mediated HIV-1 production from LN memory CD4 T cells 2) the modulation of HIV-1 transcription by LN migratory DCs through a mechanism involving PD-L/PD-1 interactions 3) the relationship between the levels of HIV-1 transcription and the frequency of PD-L1/2 expressing migratory DCs and 4) PD-1 blockade with anti-PD-1 monoclonal antibody treatment can reactivate of HIV-1 replication from latently infected CD4 T cells.
These findings represent a step forward in our understanding of the mechanism regulating the persistence of HIV transcription in lymphoid tissues.

Ethics statement
Fifty-seven HIV-1 infected adult volunteers and twelve HIV-uninfected subjects were enrolled in the present study.No statistical method was used to predetermine sample size.The present study was approved by the Institutional Review Board of the Centre Hospitalier Universitaire Vaudois, and all subjects which were adults gave written informed consent.The 57 HIV-1-infected individuals studied had a documented diagnosis of HIV-1 infection between 0.4 and 28.4 years.Treated HIV-1 infected individuals received ART treatment for 0.3 to 23.1 years.No exclusion criteria was implemented except with regards to the reactivation of HIV-1 latency from blood resting memory CD4 T-cell experiments, which were exclusively performed on cells isolated from treated aviremic HIV-1 infected individuals with undetectable viremia (HIV-1 RNA levels <50 copies per ml of plasma) for at least 12 months.Inguinal lymph node biopsies and blood samples from HIV-infected individuals were collected the same day.In addition, LN sections collected from HIV-uninfected individuals suffering from lymphadenopathy were also collected and were referred to as "reactive" LNs.

Cell isolation
Blood mononuclear cells were isolated as previously described [24] and lymph node mononuclear cells were isolated by mechanical disruption as previously described [60].Blood mononuclear cells and lymph node mononuclear cells were cryopreserved in liquid nitrogen for longterm storage.

Mass cytometry
Freshly isolated matched blood and lymph node mononuclear cells were resuspended (10 6 cells/ml) in complete RPMI medium and incubated (30 min; 4 ˚C) directly ex vivo (no permeabilization) with metal-conjugated antibodies directed against a panel of 37 parameters (Fluidigm/DVS Science) including lineage markers for T-cell, B-cell and antigen presenting-cell populations and ICs such as PD-1, CTLA-4, TIM-3, LAG-3, TIGIT as well as IC-ligands such as PD-L1, PD-L2 and CD155.Cells were washed and fixed (10 min; room temperature) with 2.4% PFA.Total cells were identified by DNA intercalation (1μM Cell-ID Intercalator, Fluidigm/DVS Science) in 2% PFA at 4 ˚C overnight.Labeled samples were acquired on a CyTOF1 instrument that was upgraded to CyTOF2 (Fluidigm) using a flow rate of 0.045 ml/ min.Data were analyzed using Fluidigm Cytobank software package (Cytobank, Mountain View, CA).At least 100,000 events were acquired for each sample.

CD4 T-cell proliferation
Sorted LN PD-1 high and PD-1 -memory (CD45RA -) CD4 T-cell populations (2 × 10 5 cells) were washed twice, labeled with CFSE at 37˚C for 7 min (Life Technologies) and cultured for 6 days at 37˚C and 5% CO 2 in 96-well U-bottom plates coated with 10 μg/ml anti-CD3 (BD) and 10 μg/ml anti-CD28 (BD) in presence or not of recombinant IC ligands at 100 μg/mL in complete RPMI.The proliferation of CD4 T-cell populations was assessed by quantifying the percentage of CFSE low cells on a LSR SORP cell analyzer (BD).Of note, all experiments were performed in presence of 75 nM emtricitabine to prevent re-infection of stimulated CD4 T cells.

Histopathology
Lymph nodes were cut into slices and fixed in B-plus or formalin before routine processing and embedding in paraffin blocks.Serial tissue sections (4-μm) were stained according to standard routine protocols by using a Ventana benchmark platform (Roche) with antibodies against PD1 (192106 R&D Systems) and PD-L1 (sp263, Ventana).PD1 and PD-L1-imunostained slides were digitalized using a Hamamatsu Nanozoomer 1.0 scanner (model C9600-01) at 40× with the NDPScan software (v.2.5.89).Scanning area and focus points were set manually.Image analysis was performed with the Tissue IA-specific module of the Slidepath Software Digital Image Hub (DIH) (version 4.0.7).The surfaces of the entire tissue section and of individual germinal centers (manually circumscribed) were measured.Cell density was estimated using the "measure stained cells algorithm" at 20× to quantify PD-1-positive cells, while PD-L1-positive areas were quantified using "measure stained area algorithm" at 20×.

HIV-1 production
Sorted LN PD-1 high and PD-1 -memory (CD45RA -) CD4 T-cell populations (2 × 10 5 cells) were cultured for 6 days at 37˚C and 5% CO 2 in 96-well U-bottom plates coated with anti-CD3 (BD) and anti-CD28 (BD) (10 μg/ml) in presence or in absence of recombinant IC ligands at 100 μg/mL in complete RPMI.In some experiments, the combination of two ICligands was also tested.Of note, all experiments were performed in presence of 75 nM emtricitabine to prevent re-infection of stimulated CD4 T cells.Supernatants were collected at days 6 and quantification of HIV-1 production was performed by assessing HIV-1 RNA levels by COBAS AmpliPrep/TaqMan HIV-1 Test (Roche; Switzerland) as previously described [44].

Fig 5 .
Fig 5. LN migratory dendritic cells expressing PD-L1 and/or PD-L2 modulate HIV-1 transcription/production in LN of treated aviremic HIV-infected individuals.(A) Level of expression of PD-Ls on LN (CD1c high CCR7 + CD127 + ) migratory DCs from one representative aviremic ART treated HIV-infected individual (#058).(B) Cumulative data of the levels of HIV-1 RNA (copies/ml) in the culture supernatants at day 6 of LN PD-1 -or PD-1 + CD4 T cells following anti-CD3/anti-CD28 stimulation and co-culture with autologous LN migratory DCs in the presence or absence of blocking anti-PD-L1/L2 MAbs (N = 4).(C) Correlation between the levels of cell-associated unspliced HIV-1 RNA detected in LN memory (CD45RA -) CD4 T cells and the frequencies of LN migratory DCs in aviremic ART treated HIV-infected individuals (N = 6).HIV-infected individuals were color-coded (B).Red stars indicate statistical significance (P<0.05).Statistical significance (P values) was obtained using one-way ANOVA (Kruskal-Wallis test) followed by ratio Paired-t test (B) or using Spearman rank test for correlations (C).https://doi.org/10.1371/journal.ppat.1007918.g005