Splenocytes used in this study were procured, recovered, and processed by the Network for Pancreatic Organ Donors with Diabetes (nPOD) and delivered in cryovials in liquid nitrogen and then kept at −80°C in our laboratory until used. Samples were from three cohorts: 14 T1D, 8 diabetes-free subjects with T1D-associated autoantibodies (Ab⁺), and 10 non-diabetic (ND) subjects. Clinical characteristics of donors are shown in Table 1.

To minimize any effect of inter-experimental variation and maintain consistency of gating used to identify positive subpopulations in different diabetic, Ab⁺, and ND subjects, we performed our analysis by matching one T1D sample with at least one Ab⁺ and/or ND sample in each experiment, and the latter were used to draw gates to reduce any bias that could have resulted from analyzing T1D separately from Ab⁺ and ND samples. In total, we collected and performed statistical analysis in data pooled from 10 to 14 independent experiments. In each experiment, frozen splenocytes were thawed by using a standard operating procedure provided by nPOD. Briefly, cryovials were removed from liquid nitrogen, rapidly thawed with vigorous shaking for 2–3 min in a 37°C water bath, and added to pre-warmed medium (RPMI+ 10% FBS+ 50 U/mL Benzonase). Cells were centrifuged, washed, and counted by using trypan blue, and 1 × 10⁶ cells were used per application, unless stated otherwise. The absence of information regarding the absolute numbers of lymphocytes per spleen (we also received small pieces) made determination of absolute numbers of each subset per subject just a mere reflection of their frequency among acquired events, and for this reason we presented the data only as frequency of their indicated subsets.

Fluorochrome-conjugated monoclonal antibodies (mAbs) specific for human antigens were obtained, unless stated otherwise, from BD Pharmingen (San Jose, CA, USA), BioLegend (San Diego, CA, USA), or eBioscience (San Diego, CA, USA). The following anti-human mAbs (clone) were used: FasL (NOK-1), Fas (DX2), CD19 (H1B19), CD5 (L17F12), CD40 (5C3), CD86 (IT2.2), CD20 (2H7), CD27 (LG.7F9), CD38 (HIT2), CD22 450 (S-HCL-1), CD10 (HI10a), CD1d (51.1), IL-10 (JES3-9D7), active caspase 3 (C92-605), IL-17 (eBio64DEC17), and IFN-γ (B27). Surface expression of human FasL was analyzed by using biotin-conjugated NOK-1 mAb. Previously described (25, 26) humanized anti-human FasL (RNOK203) was provided by Dr. Hideo Yagita of Juntendo University and Dr. Toshihiro Maeda of The Chemo-Sero Therapeutic Research Institute (Kaketsuken) in Japan.

Single cell suspensions were stained with predetermined optimal concentrations of indicated mAbs. Samples were then washed and acquired using FACS LSRII flow cytometer and analyzed by Flow-JOV10 software. Three types of specificity controls were used. First, compensations were properly set using single-stained samples. Second, Fluorescence-Minus-One unstained samples and those stained with isotype controls were used to specifically gate positive cells and set up quadrants. Third, comparison of positive and negative cells in samples from donors (T1D, Ab⁺, and ND), and when appropriate, stimulated vs unstimulated samples provided internal controls. For intracellular staining, single cell suspensions were stimulated for 6 h with phorbol myristate acetate (PMA) (50 ng/mL) and ionomycin (500 ng/mL) in the presence of Golgi-Plug (2 mmol/L). Surface staining was followed by cell fixation, permeabilization, and intracellular staining with anti-IL-10, IL-17, TNFα, or IFN-γ mAb as indicated.

Splenocytes were stimulated with 10 µg/mL of affinity purified polyclonal goat F(ab’)2 anti-human IgM (Jackson ImmunoResearch) for 24 h at 37°C in the presence or absence of humanized anti-human FasL-neutralizing mAb (RNOK203) or the general caspase inhibitor Z-VAD-FMK (BD Biosciences). During the last 6 h of incubation, cultures were stimulated with PMA and ionomycin in the presence of Golgi-Plug, followed by surface staining and intracellular detection of IL-10 and active caspase 3 using specific mAbs. It is noteworthy that number of cells obtained per sample was limiting and did not allow a more direct and extensive analysis such as multi-color FACS analysis for FasL, Fas, CD5 on B cells without culture with anti-FasL mAb. Such analysis would have directly addressed whether IL-10-producing CD5⁺ B cells are specifically targeted for apoptosis.

Unless otherwise indicated, all data are shown as mean ± SEM. Statistical analysis was performed using the unpaired Mann–Whitney U test. A two-tailed value of p ≤ 0.05 was considered statistically significant. Statistical analysis and graph preparation were performed by using GraphPad Prism software (GraphPad Software, Inc., La Jolla, CA, USA).

It has long been known that FasL regulates insulitis and autoimmune diabetes development in NOD mice. In addition, our recent data showed that genetic or antibody inactivation of FasL leads to significant increases in IL-10-producing CD5⁺ B cells in the pancreata (23). To investigate relevance of these findings to the human disease, we analyzed IL-10- and FasL-expressing B and T cells in splenocytes from T1D, Ab⁺, and ND subjects. Overview of the gating strategy used to identify lymphocytes, exclude doublets and dead cells is outlined (Figure S1 in Supplementary Material). B cells are often analyzed as one population, yet they are heterogeneous and contain CD5⁺ B cells that are viewed as autoreactive and expanded in T1D patients [Figure S2 in Supplementary Material; Ref (27–30)]. CD5⁺ B cells also harbor the IL-10-producing regulatory subset. We therefore analyzed them separately from the CD5⁻ B cell subpopulation. CD5 is otherwise a pan marker of T cells. We stimulated samples with PMA/ionomycin and used the depicted gating strategy (Figure 1A) to distinguish between CD5⁺ CD19⁺ and CD5⁻ CD19⁺ B cells and CD5⁺ CD19⁻ T cells. Each subset was analyzed for IL-10 production in various subjects. Our results show that the frequency of IL-10^pos CD5⁺ B cells was significantly higher in Ab⁺ as compared to T1D subjects (Figures 1A,B). The difference was also significant when examined at the level of the absolute numbers of IL-10^pos CD5⁺ B cells (Figure S3 in Supplementary Material). On the other hand, the frequency, but not the absolute numbers, of IL-10^pos CD5⁺ B cells was higher in the Ab⁺ subjects than ND subjects (Figure S3 in Supplementary Material). There were no significant differences, however, in the frequency of IL-10-expressing CD5⁻ B cells (IL-10^pos CD5⁻ B cells) among the subjects in the three groups, excluding generalized upregulation of IL-10 in B cells of Ab⁺ subjects. Consequently and in line with the fact that CD5⁻ B cells comprised the bulk of B cells, the overall frequency of IL-10-expressing cells among total B cells (IL-10^pos B cells) was not statistically different among subjects in the three groups (Figure S4 in Supplementary Material). The frequency of IL-10^pos T cells, however, was significantly less in T1D as compared to ND, but not Ab⁺ subjects (Figure 1B). We concluded that there is selective expansion of IL-10^pos CD5⁺ B cells in Ab⁺ as compared to T1D subjects. The results extend the idea that the regulatory mechanisms in at-risk individuals, unlike in T1D progressors, are augmented and actively working to control islet autoreactivity (31, 32).

Next, we determined whether FasL expression is dysregulated in B or T cells in any of the three subject groups. We detected significant differences that, as above, were limited to the CD5⁺ B cell subpopulation, albeit in the opposite direction, as the frequency of FasL-expressing CD5⁺ B cells (FasL^hi CD5⁺) was significantly higher in T1D as compared to Ab⁺ and ND subjects (Figure 2A). The frequency of FasL-expressing CD5⁻ B cells (FasL^hi CD5⁻) was generally low and slightly higher in T1D, but the difference among the groups did not reach statistical significance (Figure 2B). Likewise, the frequency of FasL-expressing T cells was comparable in the three groups, thereby excluding a generally dysregulated expression of FasL in T1D subjects (Figure 2B). In addition, CD5⁺ B cells in T1D, Ab⁺, and ND subjects had comparable surface expression of CD86 and CD40, confirming that the detected changes in IL-10 and FasL expression by these cells was specific and not due to generalized modulation of their surface profile (Figure S5 in Supplementary Material). Taken together, our results show IL-10^pos CD5⁺ B cells are selectively expanded in Ab⁺ subjects as opposed to selective expansion of FasL^hi CD5⁺ B cells T1D subjects.

The differential expansion of FasL^hi and IL-10^pos CD5⁺ cells in Ab⁺ and T1D indicated that they might represent two separate subpopulations. To directly test this notion, we simultaneously analyzed FasL and IL-10 expression at the single cell level in the CD5⁺ B cells from T1D, Ab⁺, and ND subjects by FACS after rapid PMA/ionomycin stimulation to upregulate IL-10. As predicted, we detected two distinct subpopulations that express either IL-10 or FasL in three types of donors (Figure 3A). The distinction, however, was not absolute as some cells were dual positive for FasL and IL-10. Nonetheless, the inability of FasL-expressing cells to produce cytokines was not limited to IL-10 or CD5⁺ B cells as it was the case with IFN-γ, TNFα, and IL-17 and for CD5⁻ B cells and T cells (Figure 3A; Figure S6 in Supplementary Material). Whether FasL^hi CD5⁺ B cells secrete cytokines other than those examined here cannot be formally ruled out and await to be determined. Differences between IL-10^pos and FasL^hi CD5⁺ cells extend to activation markers. IL-10^pos CD5⁺ B cells expressed significantly higher levels of CD1d and CD27 and lower levels of CD10 and CD22 than FasL^hi CD5⁺ B cells; but both expressed comparable levels of CD38 (Figure 3B; Figure S7 in Supplementary Material).

FasL^hi CD5⁺ B cells expressed less Fas and were less susceptible to AICD than FasL^neg CD5⁺ B cells, including IL-10^pos CD5⁺ B cells. This was examined by the analysis of intracellular levels of active caspase 3 in anti-IgM-stimulated cultures. In this experiment, we stimulated splenic cultures with anti-IgM for 24 h and analyzed gated FasL^hi and FasL^neg subpopulations of CD5⁺ B cells for AICD. Most of apoptotic cells belonged to the FasL^neg subpopulation (Figure 4A) in which IL-10^pos cells resided. Therefore, in a second set of experiments, we examined whether AICD of IL-10^pos CD5⁺ cells was mediated by the Fas pathway. Consistent with this result, there was high Fas expression on IL-10^pos CD5⁺ B cells (Figure 4B, plot), which often denote sensitivity to AICD (33). The frequency of IL-10^pos CD5⁺ cells was significantly higher in T1D than in Ab⁺ and ND subjects (Figure 4B, graphs). On the other hand, mean level of Fas expression (MFI) was higher in T1D but was not statistically different than its levels in Ab⁺ and ND subjects. Moreover, apoptosis of IL-10^pos cells was inhibited by FasL-neutralizing mAb to an extent that was comparable to that induced by the pan caspase inhibitor, ZVAD (Figure 4C). It is noteworthy that cultures treated with anti-FasL mAb often reduced the overall frequency of IL-10-producing B cells, suggesting perhaps a role for FasL in regulating IL-10 production that is worthy of future investigation. Taken together, these results show that most FasL^hi CD5⁺ B cells did not produce any of examined cytokines and that they differed from IL-10^pos CD5⁺ B cells in their activation state and susceptibility to AICD.

In summary, our results extend the long-known role for FasL in regulating pathogenesis of autoimmune diabetes in NOD mice to the human disease by showing dysregulated homeostasis of FasL^hi CD5⁺ B cells in T1D subjects. Future studies using conditional knock mice that lack FasL only in B cells will determine whether FasL expression in B cells plays an essential role in driving the diabetic process.