Regulatory B Cells Inhibit Cytotoxic T Lymphocyte (CTL) Activity and Elimination of Infected CD4 T Cells after In Vitro Reactivation of HIV Latent Reservoirs

During HIV infection, IL-10/IL-10 receptor and programmed death-1 (PD-1)/programmed death-1-ligand (PD-L1) interactions have been implicated in the impairment of cytotoxic T lymphocyte (CTL) activity. Despite antiretroviral therapy (ART), attenuated anti-HIV CTL functions present a major hurdle towards curative measures requiring viral eradication. Therefore, deeper understanding of the mechanisms underlying impaired CTL is crucial before HIV viral eradication is viable. The generation of robust CTL activity necessitates interactions between antigen-presenting cells (APC), CD4+ and CD8+ T cells. We have shown that in vitro, IL-10hiPD-L1hi regulatory B cells (Bregs) directly attenuate HIV-specific CD8+-mediated CTL activity. Bregs also modulate APC and CD4+ T cell function; herein we characterize the Breg compartment in uninfected (HIVNEG), HIV-infected “elite controllers” (HIVEC), ART-treated (HIVART), and viremic (HIVvir), subjects, and in vitro, assess the impact of Bregs on anti-HIV CTL generation and activity after reactivation of HIV latent reservoirs using suberoylanilide hydroxamic acid (SAHA). We find that Bregs from HIVEC and HIVART subjects exhibit comparable IL-10 expression levels significantly higher than HIVNEG subjects, but significantly lower than HIVVIR subjects. Bregs from HIVEC and HIVART subjects exhibit comparable PD-L1 expression, significantly higher than in HIVVIR and HIVNEG subjects. SAHA-treated Breg-depleted PBMC from HIVEC and HIVART subjects, displayed enhanced CD4+ T-cell proliferation, significant upregulation of antigen-presentation molecules, increased frequency of CD107a+ and HIV-specific CD8+ T cells, associated with efficient elimination of infected CD4+ T cells, and reduction in integrated viral DNA. Finally, IL-10-R and PD-1 antibody blockade partially reversed Breg-mediated inhibition of CD4+ T-cell proliferation. Our data suggest that, possibly, via an IL-10 and PD-L1 synergistic mechanism; Bregs likely inhibit APC function and CD4+ T-cell proliferation, leading to anti-HIV CTL attenuation, hindering viral eradication.


Introduction
CD8 cytotoxic T lymphocyte (CTL) activity is critical in controlling viral replication during HIV infection (reviewed in [1]). Individuals who naturally control HIV replication in the absence of therapy (''elite controllers'', HIV EC ) often exhibit robust CTL activity [2,3]. In contrast, CTL function is severely attenuated in individuals who do not control HIV, and this impaired CTL activity is not restored even with successful ART [2].
In ART-treated HIV-infected subjects (HIV ART ), viral replication is suppressed but the virus persists because early in infection HIV establishes latent reservoirs and upon ART interruption, HIV replication is detected [4,5]. The establishment of stable latent reservoirs [6] dictates lifelong ART treatment associated with financial cost and potential toxicity, thus, therapies leading to HIV eradication are urgently warranted. Recent studies have focused on using small molecules that, unlike antibodies, reactivate the latent reservoirs without inducing unrestrained T cell activation [7]. However, data from a seminal study by Shan et al [8] indicate that reactivating the viral reservoir using the FDAapproved histone deacetylase inhibitor (HDACi), suberoylanilide hydroxamic acid (SAHA) was not associated with the death of infected CD4 + T cells as was previously hypothesized. In contrast, Shan et al determined that post reactivation of latent reservoirs an efficient CTL response was indispensable to clear infected cells. Since CTL responses are impaired in HIV ART subjects, Margolis and Hazuda [9] suggest that HIV eradication would require a dual approach: reactivation of the latent reservoir without inducing global activation, concomitant with strategies to boost the immune response, specifically anti-HIV CTL responses. This indicates that understanding and delineating the mechanisms underlying CTL impairment in ART-treated HIV-infected subjects is critical before HIV eradication becomes viable. We have shown that in HIV ART subjects, IL-10 expressing regulatory B cells (Bregs, CD19 + CD24 hi CD38 hi ) attenuate anti-HIV CTL activities in vitro by directly inhibiting the proliferation of antigen-specific cytotoxic CD8 + T cells in a partially IL-10 dependent manner [10]. Similarly, Das et al report that CD19 + CD24 hi CD38 hi Bregs impair CTL activity during chronic Hepatitis B virus infection [11]. However, interactions between proliferating CD4 + T cells and antigen presenting cells (APC) are also critical in generating effective CTL responses [12]. Interestingly, studies show that activated B cells negatively regulate CD4 + T cell proliferation and APC function [13][14][15], indirectly attenuating the generation of effective CTL, however this has not been investigated in human viral infections.
In this study, the goal was twofold: a comprehensive characterization of the Breg compartment in HIV-infected subjects including ''elite controllers'' and assessing the anti-HIV CTL inhibitory role for Bregs in the clinically relevant context of latent reservoir reactivation. We determine phenotypic and functional similarities between Bregs from HIV EC and HIV ART subjects. Further, in vitro after SAHA treatment, Bregs directly and indirectly attenuate anti-HIV CTL activity. The mechanism likely involves modulation of mediators of CTL generation via IL-10 and/or PD-L1. To our knowledge, these data represent the first report demonstrating possible mechanisms by which Bregs directly attenuate HIV-specific CTL generation and function in a human viral infection with potential therapeutic importance in eradication of HIV.

HIV-infected ''Elite Controllers'' (HIV EC ) and Uninfected (HIV NEG ) Individuals have Comparable Breg-frequency
We have previously shown in vitro that after stimulation with HIV peptides, Bregs from HIV ART subjects directly attenuate the proliferation of HIV-specific CD8 + T cells and anti-CTL activities in a partially IL-10-dependent manner [10]. We characterized Bregs (gating strategy shown in Figure 1a Breg-Depleted SAHA-Treated PBMC from HIV EC and HIV ART Individuals Exhibit Heightened Frequency of HIVspecific Cytotoxic CD8 + T cells We next investigated the impact of Bregs on CTL activity in a clinically relevant setting. Results from in vitro and in vivo studies have demonstrated that treatment with the histone deacetylase inhibitor (HDACi), suberoylanilide hydroxamic acid (SAHA) leads to reactivation of HIV latent reservoirs [16][17][18]. In SAHA-treated PBMC from HIV EC , Breg depletion led to enhanced CD107a expression (averagely, 451% increased expression) on CD8 + T cells of all subjects investigated (n = 4, Figure 2a, left panel). There was an increase in CD107 + CD8 + T cells (averagely, 251% increased expression) in 80% of the HIV ART subjects (5 of 6, Figure 2a, right panel). This observed increase in the frequency of cytotoxic CD8 + T cells was also associated with an increased frequency of CD8 + T cells expressing HIV gag CTL-associated SL9 epitope [19][20][21] (averagely 282% increased expression) in the HIV ART (n = 4) subjects studied (Figure 2b).
In Breg-depleted SAHA-Treated PBMC from HIV EC and HIV ART Subjects, Heightened Frequency of CTL-Competent CD8 + T cells is Associated with Efficient Elimination of HIV Infected CD4 + T Cells and Reduction of Viral DNA After determining that Breg depletion resulted in elevated frequency of CTL-competent CD8 + T cells (identified by CD107a expression), we investigated if this leads to enhanced clearance of infected cells. By intracellular cytokine staining (Figure 3a), we determined that Breg depletion led to a significantly reduced frequency of infected CD4 + T cells in PBMC from HIV EC (p = 0.0021) and HIV ART , (p = 0.0236) (HIV-1 core protein positive cells, Figure S1). Finally, by quantitative RT-PCR with primers hybridizing in the HIV LTR we determined that Breg depletion was associated with a significant decrease (p = 0.0292) in viral DNA in SAHA-treated PBMC from HIV ART (n = 4) ( Figure 3b).

After SAHA Treatment, Breg Depletion leads to Enhanced Expression of Antigen Presentation Molecules and Heightened Proliferation of CD4 + T cells
In these SAHA treated samples, we next investigated the underlying mechanisms by which Bregs mediate inhibition of CTL activity. The generation of robust CTL responses requires interactions between antigen presenting cells (APC) and proliferating CD4 + T cells [12]. We determined that in the HIV EC subjects, Breg depletion led to enhanced DC expression of MHC-I

Bregs from HIV EC and HIV ART Individuals Express Elevated PD-L1 Levels and Breg Inhibition of CD4 + T-cell Proliferation is Partially PD-L1 and IL-10 Dependent
We previously demonstrated that TLR activated Bregs upregulate PD-L1 expression [10]. Further, during HIV infection, exhausted PD-1 hi T cells significantly contribute to viral persistence [22][23][24][25] and in vivo PD-1 blockade has been shown to lead to reduction in viral load [26,27]. We therefore investigated PD-L1 expression levels on Bregs from HIV NEG , HIV VIR , HIV AVIR , and HIV EC (Figure 5a). We determined that Bregs from HIV ART , and HIV EC have comparable levels of PD-L1 expression. In contrast, Bregs from HIV EC had higher levels of PD-L1 expression compared to Bregs from HIV NEG (p = 0.0929) and HIV VIR (p = 0.0421). Bregs from HIV ART expressed 34% and 38% more PD-L1 than Bregs from HIV NEG and HIV VIR respectively.
Interestingly, in HIV EC , HIV ART and HIV VIR , compared to other B cell subsets, Bregs expressed significantly higher levels of PD-L1 ( Figure 5b).
To assess the contribution of IL-10 and PD-L1 to Bregimmunoregulation, we employed CD4 + T cell proliferation coculture assays. Purified VPD450-labeled CD4 + T cells were stimulated with MACSiBeads (Miltenyi, T cell Activation/ Expansion kit) and IL-2, either cultured with medium alone, cocultured with non-Breg B cells, with Bregs, or co-cultured with Bregs supplemented either with anti-IL-10 receptor (IL-10R) blocking antibody, a PD-1 blocking antibody or both anti-IL-10R and anti-PD-1 blocking antibodies combined, as described elsewhere [22]. We determined that CD4 + T cells co-cultured with Bregs proliferated significantly less that those co-cultured with non-Bregs (p = 0.0316, Figure 6a). Further, in the Bregs cocultures, addition of IL-10R, PD-1 antibody and both IL-10R and PD-1 blocking antibodies led to 14%, 19% and 24% respective increased CD4 + T-cell proliferation compared to the control ( Figure 6b); comparable results have been reported in similar antibody blocking assays [28].

Discussion
In vitro data from a seminal study by Shan et al, [8] indicate that despite efficient reactivation of the reservoirs using SAHA, effective clearance of infected cells requires robust anti-HIV CD8 + CTL responses, which remain attenuated in ART-treated HIV-infected subjects. Thus understanding the mechanisms underlying the attenuated CTL activity in ART-treated HIV-infected subjects is crucial before viral eradication is feasible. We have shown that in vitro, Bregs inhibit the generation of anti-HIV CTL activity after antigenic stimulation [29], here we provide compelling data indicating that in vitro, after SAHA-mediated reactivation of HIV reservoirs, Bregs exert a similar anti-HIV CTL inhibitory role.
Regulatory B cells (Bregs) have been attributed divergent phenotypic (reviewed in [30]) markers, thus IL-10 production remains the de facto Breg marker. However, our group as well as others have shown that during human viral infections, CD19 + CD24 hi CD38 hi B cells are highly IL-10 competent and exert a regulatory function [11,29]. Here we furnish further evidence supporting this notion. We find that amongst B cell subsets, Bregs express the highest levels of immunosuppressive mediators IL-10 and PD-L1. Further, unlike Bregs, a non-Breg subset (CD19 + CD24 lo CD38 lo ) did not robustly inhibit CD4 + T cell proliferation.
To perform an exhaustive assessment of the Breg compartment during HIV pathogenesis, we used PBMC from HIV-infected ''elite controllers'', ART-treated and viremic subjects as well as uninfected subjects as controls. Robust and polyfunctional CTL activities in HIV-infected ''elite controllers'' have been associated with their ability to control HIV replication without ART [3,31]. We find that HIV EC and HIV NEG subjects exhibit comparable Breg-frequency that was slightly reduced in HIV VIR subjects and lowest in the HIV ART subjects. Some reports indicate that IL-10competent B cells are susceptible to apoptosis [32], and we observed that compared to other B cell subsets, Bregs from HIVinfected subjects express significantly higher levels of Annexin V and lower anti-apoptotic Bcl-2 expression ( Figure S2), suggesting that Bregs may be more prone to undergo cell death in HIV infected individuals. These data collectively suggest that HIV causes a loss of these cells, and that this loss is not readily reversed with ART.
Regarding the expression of immunosuppressive mediators IL-10 and PD-L1 by Bregs, we determined that Bregs from HIV EC and HIV ART subjects constitutively expressed comparable levels of IL-10 and PD-L1, higher than in HIV NEG . However, Bregs from HIV VIR expressed the most IL-10 though lower PD _ L1 levels that were comparable to Bregs from HIV NEG .
We have previously shown in vitro that, Breg inhibition of CD8 + T cell proliferation is partially IL-10 dependent [10]. However, our finding that Bregs also express high levels of PD-L1 prompted us to investigate if Bregs also exert their immunoregulatory function via PD-L1-PD-1 interactions. Data from our antibody blockade study, suggest that Bregs-immunoregulatory function likely involves a synergistic effect of IL-10 and PD-L1, comparable to results from a study by Brooks et al [33]. In autoimmune settings, murine Bregs exert immunoregulatory functions via cognate T-cell ligands including CD40, MHCII and IL-21 [34]. However, we did not find any difference in the expression of CD40, MHCII and IL-21R on the Bregs in our subjects (data not shown), indicating key differences between Breg immunoregulatory pathways in autoimmune diseases and viral infections. There is growing interest in blocking the PD-1 pathway as part of an HIV cure strategy [26,35]. Our data suggest that in the context of long-term effective ART, blocking either PD-1L or PD-1 may result in the generation of effective generation of anti-HIV CTL, and ultimately enhanced clearance of infected cells.
Finally, we assessed the impact of Bregs on anti-HIV CTL activity after SAHA-treatment. We used degranulation (CD107a expression) as the measure of CTL-competency and determined that following reactivation of latent reservoirs using SAHA, Bregdepletion modulated critical mediators of robust CTL generation including antigen presentation and CD4 + T cell proliferation [12,36]. Murine Bregs have been shown to modulate APC function and inhibit CD4 + T cell proliferation [13,14,37] but to our knowledge this is the first time this has been demonstrated for IL-10-competent Bregs during a human viral infection. In these Breg-depleted samples, the robustness of the CTL response was reflected by the enhanced expression of CTL-competent CD107a + and HIV-specific CD8 + T cells. Finally, we determined significantly enhanced clearance of infected CD4 + T cells, a significant reservoir of latent HIV [4,5,38].
Taken together our results suggest that during HIV infection CD19 + CD24 hi CD38 hi Bregs represent the predominant IL-10 producing B-cell subset, consistent with previous data suggesting a similar role for Bregs during hepatitis B viral infection [39]. Further, we provide novel evidence indicating that Bregs represent the B-cell subset with highest levels of PD-L1 expression in HIV infection, although more studies are warranted to delineate the patterns of Breg IL-10 and PD-L1 expression. Finally, our results suggest a role for Bregs in attenuating CTL responses after reactivation of HIV latent reservoirs. Ineffective CTL responses present a critical hurdle in the quest for HIV eradication. Thus further elucidation of Breg phenotype and regulation could potentially lead to therapies boosting the anti-HIV responses and HIV eradication during curative measures.

Ethics Statement and Study Participants
All studies were performed after signed informed written research consent by each study subject. The study was reviewed and approved by the Institutional Review Board of the Rush University Medical Center, University of California-San Francisco (UCSF), and the University of Iowa City VAMC and University of Iowa. HIV-uninfected (HIV NEG ) subjects had a median CD4 count of 777 cells/ml (range: 380-1487). HIVinfected viremic (HIV VIR ) subjects had a median CD4 count of 495 cells/ml (range: 240-1136) and median viral load of 40909 copies/ml (range: 11023-3140000). Antiretroviral-treated HIVinfected (HIV ART ) subjects had a median CD4 count of 435 cells/ ml (range: 212-1076) and median viral load of 40 copies/ml (range: 0-800). HIV-infected ''Elite'' controllers (HIV EC ) had a median CD4 count of 763 cells/ml (range: 454-1595) and median viral load of 48 copies/ml (range: 20-1226). HIV EC were characterized as HIV-infected subjects capable of maintaining their viral load at ,2000 copies/ml without ART as previously described [3].
Staining of HIV-specific CD8 + T cells with HLA-A * 0201-Restricted Peptide Complex The frequency of antigen-specific CD8 + T cells was determined by binding to APC-labeled HLA-A2-restricted SL9 (SLYNT-VATL) HIV-Gag epitope MHC-I-Dextramer (Immudex, Copenhagen, Denmark). Cells of HLA-A2 typed HIV + individuals were washed twice with PBS, and incubated with 10 ml Dextramer for 10 minutes at room temperature, stained with antibodies and analyzed by flow-cytometry.

LTR Real-Time PCR for HIV-1 DNA Quantification
RNA was isolated using Qiagen RNeasy Kit, according to the manufacturer's recommendations. Subsequently, cDNA was synthesized using the Quantitect Reverse Transcription kit (Qiagen, Valencia, CA). Real-time RT-PCR was performed using a Quantitect SYBR Green PCR kit (Qiagen) in a 7900HT Fast Real-Time PCR system (Applied Biosystems, Foster City, CA). Melting curve analysis was performed to ensure that the primers amplified the desired amplicon and that primer-dimers were absent. Primers used were: Long Terminal Repeats (LTR) mRNA forward 59-TCAAGTAGTGTGTGCCCGTCTGTT-39 and reverse 59-AGCTCCTCTGGTTTCTCTTTCGCT-39; and GAPDH mRNA forward 59-CTTCAACGACCACTTTGT-39 and reverse 59-TGGTCCAGGGGTCTTACT-39. Fold change in RNA expression was calculated by relative quantification using the comparative cycle threshold method. GAPDH expression was used as an endogenous control.

Statistical Analysis
Results are expressed as mean 6 standard error of the mean (SEM) or as indicated. GraphPad Prism software, version 5.03 was used for all statistical analysis. The statistical significance p value between group parameters was determined using either unpaired or paired tests (as indicated, with a confidence level of 95%). The statistical dependence between variables was calculated using the Spearman rank correlation analysis. p values of ,0.05 were considered statistically significant. Figure S1 Detection of infected CD4+ T cells using the KC-57 antibody. To determine the specificity of the KC-57 antibody, total or Breg-depleted PBMC from an HIV ART subject, were supplemented with HIV-peptides (2 ug/ml of gag, pol, env and nef) and as a control PBMC were left unstimulated. After 4 days in culture, intracellular staining for KC-57 was performed and the frequency of stained infected cells determined by flow cytometry. Representative dot plots from 3 independent experiments are shown.