IL-10 Producing B Cells Ability to Induce Regulatory T Cells Is Maintained in Rheumatoid Arthritis

Despite growing evidence highlighting the relevance of increasing IL-10-producing B cells (B10+cells) in autoimmune diseases, their functions in patients are still unknown. The aim of this study was to evaluate the functions of CpG-induced B10+ cells isolated from healthy controls (HC) and rheumatoid arthritis (RA) patients, on naïve T cell differentiation. We demonstrated that CpG-induced B10+ cells from HC drove naïve T cell differentiation toward regulatory T cells (Treg cells) and IL-10-producing T cells (Tr1) through IL-10 secretion and cellular contacts. B10+ cells from HC did not decrease T helper 1 (Th1) nor and tumor necrosis factor α producing T cell (TNFα+ T cell) differentiation. We showed that in RA, B10+ cells could also induce Treg cells and Tr1 from naïve T cells. Contrary to HC, B10+ cells from RA patients increased naïve T cell conversion into Th1. Interestingly, PD-L2, a programmed death-1 (PD-1) ligand that inhibits PD-L1 and promotes Th1 differentiation, was overexpressed on RA B10+ cells compared to HC B10+ cells. Together, our findings showed that CpG-induced B10+ cells may be used to increase Treg cells in patients with RA. However, CpG may not be the most adequate stimuli as CpG-induced B10+ cells also increased inflammatory T cells in those patients.

Different subsets of Breg cells can decrease inflammatory responses (4)(5)(6). In humans, immature transitional CD24 hi CD38 hi B cells (7,8,11) and mature follicular CD24 hi CD27 + B cells (12)(13)(14) were shown to decrease Th1, Th17, TNFα + T cells and also to increase Treg cells and Tr1 through IL-10 production. However, the presence of CD24 hi CD38 hi and CD24 hi CD27 + B cells does not necessarily reflect their functionality. In fact, in patients with autoimmune diseases, while the abundance of CD24 hi CD38 hi and CD24 hi CD27 + B cells is comparable to those in healthy patients, they have lost the ability to induce Treg cells or to decrease Th1 and TNFα + T cells (7,8,12). Thus, a marker for Breg cells which closely correlates with their functions is needed, both in healthy individuals and in patients. As both CD24 hi CD38 hi and CD24 hi CD27 + B cells are able to produce IL-10 after a stimulation with CpG, IL-10 production has been extensively used to define Breg cells, also known as B10 + cells (12,15,16). However, it is unknown whether any type of B cell secreting IL-10 has regulatory functions, in healthy subjects and in patients. Indeed, while the functions of CD24 hi CD38 hi and CD24 hi CD27 + B cells have been extensively described, CpG-induced B10 + cell regulatory functions remain fully elusive.
The objective of this study was to determine whether CpG-induced IL-10-producing B cells is a relevant functional definition for Breg cells in healthy subjects and in patients with RA.

MaTerials anD MeThODs subjects
Healthy subjects were either blood donors or patients seen in the department of Rheumatology (Teaching hospital, Montpellier) for mild osteoarthritis or mechanical pain with no general pathology or infection and receiving no immunomodulatory drugs. To be included, patients with RA had to fulfill ACR/EULAR 2010 criteria, be free of biological disease-modifying anti-rheumatic drugs and have no glucocorticoid or less than 10 mg/day. All subjects signed a written informed consent for the study in accordance with the 2013 Declaration of Helsinki and as approved by the Medical Ethics Committee of Nimes hospital, France (CPP_2012-A00592-41). Characteristics of the controls and patients are detailed in Table 1.

cytokines assessment
PBMCs from healthy individuals were stimulated by CpG (10 µg/mL) for 24 h. PMA and ionomycin were then added for the last 4 h of culture and BFA (10 µg/mL, Sigma Aldrich) was added for the last 2 h. PBMCs were then stained using FITC or V450-conjugated anti CD19 (clone HIB19, BD Pharmingen) and Fixable Viability Dye eFluor 506 (Thermofisher) before fixation (cytofix/cytoperm buffer BD biosciences) and permeabilization (Perm wash buffer, BD Pharmingen). Intracellular staining for IL-10, IFNγ, and TNFα was then performed using APCconjugated anti IL-10 (clone JES3-19F1, BD Pharmingen) and PeCy7-conjugated anti-IFNγ PeCy7 (4SB3, BD Pharmingen) or anti-TNFα (clone MAb11, BD Pharmingen). Percentages of IFNγ and TNFα positive cells among B10 + and B10 neg cells were measured by flow cytometry, using FACS CANTO II (BD biosciences). Gating strategy and representative flow cytometry dot plot are shown in Figure S2 in Supplementary Material. We also measured cytokines concentrations in supernatants from isolated B10 + and B10 neg cells. To do this, B10 + and B10 neg cells were sorted as described above, and kept in culture for 24 h. Then, cytokine dosages in supernatants were performed using Human IFN-γ ELISA MAX Deluxe and Human TNF-α ELISA MAX Deluxe (BioLegend, San Diego, CA, USA) according to the manufacturer's instructions.
Gating strategies for flow cytometry analysis were performed using FlowJo software V10 (Treestar, Ashland, OR, USA). First, PBMCs were gated based on forward scatter and side scatter, followed by exclusion of doublets and dead cells. Gating strategies are shown in Figure S1A  Blocking antibodies B cells were first incubated with Blocking antibody against PDL-1 (CD274, clone MIH1, BD Pharmingen) or PDL-2 (CD273, R&D Systems, Lille, France) before the co-culture with T cells.

statistical analysis
Comparison between paired data were performed using Wilcoxon matched pairs test. Significance was ascribed to p values <0.05. To compare variations between healthy controls (HC) and patients, we expressed data as median ± interquartile range (IQR) 25-75 and significance was assessed using Mann-Whitney test. All analyses were performed in Graph Pad Prism 5 (San Diego, CA, USA). resUlTs cpg-induced B10 + cells Produced More Pro-inflammatory cytokines Than B10 neg cells in hc TLR9 ligation by CpG is the most potent and the most commonly used inducer of B10 + cells. However, it also promotes release of pro-inflammatory cytokines by B cells (17). As the effect of CpG on the release of pro-inflammatory cytokines by Breg cells is unknown, we first studied the secretion profile for TNFα and IFNγ of B10 + , induced by CpG, isolated from HC. Despite their secretion of the anti-inflammatory cytokine IL-10, B10 + are also significantly more TNFα + and IFNγ + than B10 neg ( (Figures 1C,D).
These results were confirmed in a co-culture using CD4 + CD45RA + CD62L + T cells, considered as a more accurate population of naïve T cells. Indeed, B10 + cells significantly increased the differentiation of naïve CD4 + CD45RA + CD62L + T cells into Treg cells, compared to B10 neg cells ( (Figures 2E,F). il-10 is necessary, but not sufficient, for cpg-induced B10 + cell regulatory Functions It has been shown that CD24 hi CD38 hi and CD24 hi CD27 + B cell regulatory functions were mainly mediated by IL-10 and cell contacts. To assess the respective effect of IL-10 and cellular contacts in our system, we first neutralized IL-10 with a blocking antibody. IL-10 blocking significantly reduced CD4 + CD25 − conversion into Treg cells (7.09% [2.86; 16.55] with anti-IL-10 antibody vs 13.90% [7.09; 24.65] without it; p = 0.008; n = 8) ( Figure 3A). However, Treg cell conversion was not solely attributable to IL-10 since addition of recombinant IL-10 (10 and 100 ng/mL) to B10 neg cells could not lead to the same increase in Treg cells (Figure 3A). To test whether cellular contacts were needed to generate Treg cells, we separated B10 + and T cells with a membrane insert, so that cytokine transfers were still effective but direct contacts were prevented. We found that cellular contacts were essential since adding insert to the co-cultures lead to a drastic decrease in Treg cells (Treg decrease of 90% [80.56; 93.33]; p = 0.02; n = 7) (Figures 3B,C).
To further characterize the cellular interactions between B10 + cells and T cells, we assessed the surface expression on B10 + cells and on B10 neg cells of the co-stimulatory molecules CD80, CD86, and the program death molecule ligands 1 and 2 (PD-L1 and PD-L2), known to trigger Treg cell functions. We found that all of them, with the exception of CD86, were more often expressed and with a higher level of expression on B10 + cells compared to B10 neg cells ( Figure 3D; Figure S3 (Figure 3E) suggesting that co-stimulatory molecules rather than programmed death-1 (PD-1) ligands are critical for CpG-induced B10 + cell induction of Treg cells.
B10 + cells From ra Patients expressed a higher ratio of PD-l2 to PD-l1 Than in hc PD-L1 was shown to inhibit Th1 differentiation whereas PD-L2 could promote it (18,19). As RA B10 + cells induced Th1 differentiation while HC B10 + cells did not, we compared the expression of PD-L1 and PD-L2 on B10 + cells from RA and HC. Interestingly, RA B10 + cells were less PD-L1 + (10.10%  (Figures 6A,B) leading to a higher ratio of PD-L2 + to PD-L1 + on RA B10 + cells ( Figure 6C). This variation in PD-L1 and PD-L2 expression was also true when analyzing their expression levels by median of fluorescence intensity ( Figure 6D). This might explain why RA B10 + cells induced Th1 differentiation but not HC B10 + cells.

DiscUssiOn
The aim of the present study was to determine whether CpGinduced IL-10-producing B cells were a relevant functional definition for regulatory B cells in healthy subjects and in patients with RA. Functions of human B10 + cells have poorly been assessed in controls and in RA patients. We found that in HC, CpG-induced B10 + cells could increase the differentiation of naïve T cells into Treg cells and Tr1, through IL-10 secretion and cellular contacts. Using blocking antibodies, we showed that CD80 and CD86, but not the PD-1 ligands were required for B10 + cells to induce Treg cells. Interestingly, B10 + cells from RA patients could also increase the conversion of naïve T cells into Treg cells and Tr1. Thus, we showed that, unlike the CD24 hi CD27 + or CD24 hi CD38 hi B cells (8,12), B10 + cells seems to be a consistent functional definition for B cells inducing Treg subsets.
Another important finding of this paper is that CpG-induced B10 + cells failed to limit Th1 and TNFα + T cells differentiation, in contrast to previously described CD24 hi CD27 + or CD24 hi CD38 hi B cell functions (7,8,11,12,14). Most of the functional studies about CD24 hi CD27 + or CD24 hi CD38 hi B cells did not use any stimulation. CpG stimulation enhanced the production of IFNγ and TNFα by B cells which may counterbalance the effect of IL-10 on pro-inflammatory T cell differentiation. Therefore, we believe that CpG stimulation influenced the effect of B10 + cells on proinflammatory T cell differentiation. Along with this, Iwata et al. did not find any specific effect of CpG stimulated CD24 hi CD27 + B cells on TNFα producing T cells differentiation (20). While type of stimulation is highly important, duration of stimulation may also modify B10 + function. Besides, Iwata described two types of B10 + cells; B cells that express IL-10 after only 5 h of stimulation, called B10 cells, and B cells that require 48 h of stimulation, called progenitor B10 cells. It is likely that, depending on the duration of the stimulation, different subtypes of B cells are expressing IL-10, and that these may have different functions. In our system, we did a 24 h stimulation, to minimize mortality (12). Thus, determining the best stimulation capable of inducing regulatory but not pro-inflammatory subsets should be address in further studies to optimize therapeutic potential of B10 + cells.
While B10 + cells of patients have a similar effect on Treg cell differentiation as in HC, RA B10 + cells increased Th1 cell frequency. A higher PD-L2 to PD-L1 ratio on RA B10 + cells compared to HC B10 + cells might explain this difference. In fact, Riccomi and Palma (19) showed that in co-culture of unexperienced B cells and CD4 + T cells, blocking PD-L2 inhibited IFNγ production whereas neutralizing PD-L1 enhanced proliferation and IFNγ production. In the same way, Karunarathne et al. (18) found that PD-L2 expressed on dendritic cells, by inhibiting PD-1 binding to PD-L1, increased T cell activation and improved Th1 response. In our study, because of the low number of B10 + cells in RA patients, we could not perform functional experiment to confirm PDLs involvement in Th1 differentiation.
Notably, PD-L1 and PD-1 blocking antibody have been used in cancer therapies (21). As PDLs seem to be implicated in Breg cell functions, further studies could be useful to support the choice between PD-L1 or PD-1 blocking antibodies in treating cancer.
Together, our findings show that CpG-induced B10 + cells may be used to increase Tregs in patients with RA but not to decrease pro-inflammatory T cells.

eThics sTaTeMenT
This study was carried out in accordance with the recommendations of the local Ethic Committee. The protocol was approved by the Medical Ethics Committee of Nimes hospital, France (CPP_2012-A00592-41). All subjects gave written informed consent in accordance with the Declaration of Helsinki.

aUThOr cOnTriBUTiOns
JMi designed and performed the experiments; collected, analyzed, and interpreted the data; and wrote the manuscript. RA designed and performed the experiments, interpreted the data. MH and LM critically read the manuscript and gave scientific advice. BC and JMo contributed to patient enrollment into the study and critically read the manuscript. CD obtained financial support, designed the experiments, conducted patient enrollment into the study, interpreted the data, and wrote the manuscript.

acKnOWleDgMenTs
We thank Naomi Taylor, Valérie Dardalon, Gaspard Cretenet, and all their team for their advice; Sumaiya Hoque for English correction of the manuscript. We acknowledge the imaging facility MRI, member of the national infrastructure France-BioImaging supported by the French National Research Agency (ANR-10-INBS-04, «Investments for the future»), and more especially Myriam Boyer, Sarah Gailhac for the cell sorting.