Lipid-antigen presentation by CD1d(+) B cells is essential for the maintenance of invariant natural killer T cells.

B cells perform many immunological functions, including presenting lipid antigen to CD1d-restricted invariant natural killer T (iNKT) cells, known to contribute to maintaining tolerance in autoimmunity. Patients with systemic lupus erythematous (SLE) display dysregulated B cell responses and reduced peripheral iNKT cell frequencies. The significance of these defects and how they relate to SLE pathogenesis remain elusive. We report that B cells are essential for iNKT cell expansion and activation in healthy donors but fail to exert a similar effect in SLE patients. Defective B cell-mediated stimulation of iNKT cells in SLE patients was associated with altered CD1d recycling, a defect recapitulated in B cells from healthy donors after stimulation with interferon-α (IFN-α) and anti-immunoglobulin (Ig). iNKT cell number and function were restored in SLE patients responding to anti-CD20 treatment upon normalization of CD1d expression exclusively in repopulated immature B cells. We propose that healthy B cells are pivotal for iNKT cell homeostasis. Copyright Â

Introduction

Systemic lupus erythematosus (SLE) is a complex autoimmune disease with an unclear etiology (Rahman and Isenberg, 2008). Aberrant B cell responses and the production of autoantibodies are considered hallmarks of the disease (Lipsky, 2001). The important role of B cells in SLE pathogenesis is further proven by the clinical success of B cell depletion therapy (CD20 mAb; rituximab) (Leandro et al., 2005). As well as producing antibodies, B cells release cytokines and chemokines and present both peptide and lipid antigen (Batista and Harwood, 2009; Lund and Randall, 2010). Although the majority of studies have focused on the effect that peptide-antigen presentation has on CD4+ T cell differentiation, there is little information regarding the effect that B cells, presenting lipid antigen via CD1d, have on invariant natural killer T (iNKT) cell activation and differentiation.

iNKT cells perform critical functions in a broad range of immune responses including protection from specific pathogens and tumors, promotion of airway hyperreactivity, and the maintenance of tolerance in autoimmunity (Berzins et al., 2011; Wilson and Delovitch, 2003). Changes in iNKT cell frequency have been reported in patients with autoimmune disease. However, the cause of this reduction remains to be ascertained (Kukreja et al., 2002; Tudhope et al., 2010). Activation of iNKT cells occurs via presentation of exogenous or endogenous lipid antigen by CD1d expressed on a variety of antigen-presenting cells (APCs). Although the nature of the natural activating ligand(s) remains controversial, a marine-sponge-derived glycolipid, α-galactosylceramide (αGalCer), potently activates iNKT cells (Kawano et al., 1997). Engagement of the invariant T cell receptor (iTCR) by CD1d-lipid complexes leads to rapid iNKT cell activation, the prompt production of T helper 1 (Th1) cell and Th2-like cytokines, and the upregulation of several costimulatory molecules (Cerundolo et al., 2009). These events contribute to the reciprocal activation of APCs, for example, the release of interleukin-12 (IL-12) by dendritic cells (DCs) and the promotion of B cell maturation into plasma cells (Barral et al., 2008; Lang et al., 2008). Conversely, marginal zone (MZ) B cells activate iNKT cells via DCs (Bialecki et al., 2009), supporting an indirect function for B cells in iNKT cell homeostasis. Overall, the effect that B cell lipid-antigen presentation has on CD1d-restricted iNKT cell function in humans remains unclear.

We examined whether B cells are required for the in vitro and in vivo maintenance of iNKT cells from healthy donors and SLE patients. We demonstrated that B cells sustained iNKT cell homeostasis and activation in healthy donors but not in SLE patients. Patients were characterized by a decrease in CD1d cell surface expression exclusively on B cells and not on other lipid-antigen-presenting cells, a phenomenon that could be mimicked in vitro by simultaneous stimulation with interferon-α (IFN-α) and B cell receptor (BCR) engagement, factors associated with SLE pathogenesis (Bennett et al., 2003; Lipsky, 2001). We have shown that SLE patients responding to B cell depletion therapy present normalized CD1d expression prevalently on repopulated CD19+CD24hiCD38hi immature B cells and this positively correlated with the restoration of iNKT cell number and function.

Results

B Cells Promote iNKT Cell Proliferation and Activation in Healthy Donors

Previous work shows that peripheral blood mononuclear cell (PBMC) stimulation with αGalCer and IL-2 leads to an exponential expansion of iNKT cells after 7–14 days (Watarai et al., 2008). To determine the role of B cells presenting lipid antigen in this process, we depleted B cells from PBMCs before stimulation with αGalCer and IL-2. Removal of B cells from PBMCs impeded the capacity of iNKT cells to expand in response to αGalCer and IL-2 compared to undepleted PBMCs. Depletion of CD14+ monocytes, CD19−CD1d+, CD19−CD1c+ DCs, or CD19−CD1c+ DCs together with CD14+ monocytes did not significantly reduce the capacity of iNKT cells to expand in response to αGalCer and IL-2 (Figure 1A, Figure S1A available online, and data not shown for CD19−CD1c+ DCs or CD19−CD1c+ DCs with CD14+ monocytes). Removal of B cells led to a reduction in expression of CD25 and Ki67, markers associated with activation and proliferation, thus supporting the importance of B cells in the maintenance of iNKT cell homeostasis (Figure 1B). No difference was observed in the expression of CD69 (data not shown).

Figure 1

B Cells Support iNKT Cell Expansion and Activation in Healthy Donors

B cells or monocytes were depleted from PBMCs obtained from healthy donors (n = 12). Depleted or undepleted PBMCs were cultured with αGalCer and IL-2 for 4 and 7 days. Supernatants were collected for cytokine analysis and cells were stained with CD3 and iTCR mAbs.

(A) Representative flow cytometry plots showing frequency of iNKT cells at day 0, 4, and 7 in undepleted PBMCs from healthy donors and in B cell- or monocyte-depleted PBMCs; bar chart shows cumulative data as mean ± SE. ∗p < 0.05; one-way ANOVA.

(B) Representative histograms showing CD25 and Ki67 expression on iNKT cells from undepleted or B cell-depleted PBMCs. Bar chart showing mean ± SE. ∗p < 0.05, ∗∗p < 0.01; one-way ANOVA.

(C) Cytokines in supernatants collected from experiment (A) were measured by cytometric bead array (CBA). Bar charts show mean ± SE from 15 healthy donors. Data representative of four separate experiments. ∗p < 0.05; paired t test.

(D) Negatively purified B cells pulsed with αGalCer for 1 hr and cocultured 1:1 with positively purified iNKT cells (105/well) with IL-2 ± isotype control or blocking CD1d mAb for 48 hr. Proliferation was determined by 3[H]TdR incorporation. Bar charts show fold change ± SE in proliferation normalized to unpulsed B cells in coculture with iNKT cells with isotype control alone. Cumulative data from three independent experiments with three healthy donors each. Comparisons were between αGalCer (αGC) pulsed and unpulsed cultures; ∗p < 0.01; one-way ANOVA.

We then proceeded to investigate the effect that removal of B cells from PBMCs had on cytokine production driven by iNKT cells upon stimulation with αGalCer and IL-2. In vitro depletion of B cells led to a reduction of IFN-γ, tumor necrosis factor-α (TNF-α), IL-2, and IL-10 production at both day 4 and 7 after stimulation (Figure 1C and data not shown for IL-2). No difference in IL-4 and IL-6 production were detected between B cell-depleted and undepleted cultures (data not shown). Intracellular detection of cytokines specifically produced by iNKT cells further supported these data (Figure S1B).

To determine whether B cells directly, or indirectly via other APCs, support iNKT cell proliferation, we cultured purified iNKT cells with B cells preloaded with αGalCer in the presence of IL-2. The results in Figure 1D display the fold increase in proliferation, compared to unpulsed B cells and demonstrate that B cells support iNKT cell expansion directly, as shown by the fact that blockade of the CD1d-iTCR interaction inhibited iNKT cell proliferation. Thus, B cells appear to be important for the maintenance of iNKT cell number and function in healthy donors.

iNKT Cells in SLE Patients Are Numerically Decreased and Functionally Impaired

Numerical defects in iNKT cell number have been previously reported in patients with autoimmune disease including SLE (Tudhope et al., 2010; Wither et al., 2008). We demonstrated that patients with SLE and other autoimmune diseases (OAD) including Sjögren's syndrome and rheumatoid arthritis (RA) have significantly reduced iNKT cell numbers in peripheral blood (PB) (Figure 2A). The ratio of CD4+ and CD8+ iNKT cell subsets was comparable in both healthy donors and patients (data not shown). To confirm that reduced iNKT cell number in SLE patients was not due to iTCR internalization, we stained for both surface and intracellular iTCR expression. The results in Figure 2B showed a lower iNKT cell frequency in SLE patients compared to healthy donors.

Figure 2

iNKT Cells in SLE Patients Are Numerically Decreased and Functionally Impaired

PBMCs from healthy donors or SLE patients were stained ex vivo for surface expression of CD3 and iTCR.

(A) Representative flow cytometry plots and cumulative data showing mean iNKT cell percentages and absolute numbers from 69 healthy donors, 254 SLE patients, and 20 patients with other autoimmune disease (OAD). ∗∗p < 0.01, ∗∗∗∗p < 0.0001; one-way ANOVA.

(B) PBMCs from healthy donors or SLE patients were stained for CD3 and iTCR extracellularly only (EC) or extra- and intracellularly (EC+IC). Representative flow cytometry plots and cumulative data showing iNKT cell percentages from five healthy donors and four SLE patients.

(C) Representative flow cytometry plots displaying frequencies of IFN-γ+, TNF-α+, and IL-10+ iNKT cells, mean ± SE. ∗p < 0.01; two-tailed t test.

Analysis of the ex vivo cytokine profile revealed a decrease in IFN-γ and an increase in IL-10-producing iNKT cell frequencies in SLE patients compared to healthy donors. No differences were observed in the number of iNKT cells producing TNF-α (Figure 2C). Thus, iNKT cells from SLE patients display an altered phenotype and cytokine milieu compared to healthy donors.

Abnormal B Cells from SLE Patients Induce Impaired iNKT Cell Number and Function

Unlike those from healthy donors, iNKT cells from SLE patients did not expand in vitro in response to αGalCer and IL-2 stimulation (Figure 3A). Although iNKT cells failed to expand, analysis of iNKT cells from SLE patients stained with 4′6 diamidino-2-phenylindole (DAPI) did not show increased mortality (data not shown). The analysis of cytokines in the supernatants at day 7, measured with a cytometric bead array (CBA), revealed that iNKT cells from SLE patients were incapable of driving the release of IL-2, IFN-γ, and TNF-α (Figure 3B). In contrast, production of IL-10 driven by iNKT cells was increased in SLE patients compared to healthy donors (Figure 3B), although production of IL-4 and IL-6 remained unchanged (data not shown).

Figure 3

iNKT Cells from SLE Patients Are Refractory to Further In Vitro Stimulation with αGalCer and IL-2

PBMCs were cultured with αGalCer and IL-2 for 7 days. Supernatants were collected for cytokine analysis and cells were stained with CD3 and iTCR mAbs.

(A) Representative flow cytometry plots and cumulative data from 21 healthy donors and 22 SLE patients showing frequencies of iNKT cells before and after αGalCer and IL-2 stimulation. ∗∗p < 0.01; paired t test.

(B) Cytokines from supernatants were measured by CBA. Representative CBA plot and bar charts with cumulative data are shown. Mean ± SE. ∗p < 0.05; two-tailed t test.

(C) Experimental design. B cells from healthy donors were negatively isolated, cocultured with B cell-depleted PBMCs from SLE patients, and stimulated for 7 days with αGalCer and IL-2. Alternatively, negatively isolated B cells from SLE patients were cultured with B cell-depleted PBMCs from healthy donors. As controls, B cells from healthy donors were cultured with allogeneic B cell-depleted PBMCs from healthy donors or B cells from SLE patients were cultured with allogeneic B cell-depleted PBMCs derived from SLE patients.

(D) Bar chart shows mean ± SE fold increase in percentage of iNKT cells at day 7 compared to day 0. ∗p < 0.05, ∗∗p < 0.01; one-way ANOVA.

(E) IFN-γ and TNF-α from supernatants collected in experiment (D) were measured by CBA. Representative CBA plot and cumulative data from three independent experiments each with five healthy donors and five SLE patients, mean ± SE. ∗p < 0.05, ∗∗p < 0.01; one-way ANOVA.

(F) Monocytes from healthy donors or SLE patients were isolated with anti-CD14 beads, stimulated with 1,000 IU/ml GM-CSF and 500 IU/ml IL-4, and allowed to differentiate into DCs. At day 4, immature DCs were stimulated with 100 ng/ml lipopolysacharide (LPS) and αGalCer for 24 hr. αGalCer-pulsed mature DCs were cultured for 7 days with autologous CD14+ monocyte-depleted PBMCs from healthy donors or SLE patients with 200 IU/ml IL-2. Purity of monocyte-derived DCs is shown in Figure S5D. Representative flow cytometry plots showing frequency of iNKT cells at day 0 and 7 from cultures with monocyte-derived DCs together with autologous monocyte-depleted PBMCs from healthy donors and SLE patients; bar chart shows cumulative data, mean ± SE from five healthy donors and five SLE patients. ∗p < 0.05; paired t test.

(G) Cytokines from supernatants collected in experiment (F) were measured by CBA. Results from five healthy donors and five SLE patients shown, representative of two separate experiments.

Because SLE patients are characterized by severe B cell abnormalities (Lipsky, 2001), we assessed their potential contribution to driving iNKT cell dysfunction. B cells were depleted from PBMCs from healthy donors and replaced with negatively purified B cells from SLE patients. Two controls were used: purified B cells from healthy donors cultured with allogeneic PBMCs from healthy donors depleted of B cells and purified B cells from SLE patients cultured with allogeneic PBMCs from SLE patients depleted of B cells (schematic representation of this experiment is depicted in Figure 3C). In PBMC from healthy donors, replacement of B cells with B cells from SLE patients reduced iNKT cell expansion and their capacity to produce IFN-γ and TNF-α in response to αGalCer and IL-2 stimulation compared to the response observed in healthy allogeneic control (Figures 3D and 3E). Production of IL-2 and IL-10 under these experimental conditions was below the limit of detection (data not shown). Replacement of B cells from SLE patients with B cells from healthy donors did not rescue the capacity of iNKT cells from SLE patients to proliferate in response to lipid-antigen stimulation (Figure 3D) but did restore their ability to produce IFN-γ and TNF-α (Figure 3E). No restoration effect was observed when allogeneic B cells from SLE patients were cultured with PBMCs from SLE patients depleted of B cells. Additionally, no detectable iNKT cell expansion or cytokine production was seen when B cell-depleted PBMCs from SLE patients were cultured alone (data not shown).

The results presented so far could be interpreted in two different ways. One explanation is that B cells from SLE patients are defective and fail to support iNKT cell proliferation and differentiation. Alternatively, iNKT cells from SLE patients could be intrinsically defective independent from the action of B cells. However, the results displayed in Figures 3F and 3G show that monocyte-derived dendritic cells (mDCs) isolated from SLE patients were able to support both expansion and cytokine release of iNKT cells from SLE patients, thus refuting the hypothesis that iNKT cells from SLE patients are intrinsically defective.

B Cells from SLE Patients Present Defects in CD1d Surface Expression and Intracellular Recycling Compared to Healthy Donors

Our results indicated that B cell-mediated iNKT cell proliferation was CD1d dependent (Figure 1D). This prompted us to examine the expression of CD1d on B cells from healthy donors and SLE patients with active disease. The expression of surface CD1d on B cells from SLE patients was decreased compared to B cells from healthy donors, whereas the expression of CD1d on monocytes was equivalent in SLE patients and healthy donors (Figure 4A). Given the profound reduction in expression of CD1d on B cells from SLE patients, we tested whether increasing CD1d cell surface expression on these cells could restore their function. We used an established method shown previously to force cytotoxic T lymphocyte antigen 4 to the surface of regulatory T (Treg) cells (Flores-Borja et al., 2008). Negatively isolated B cells from healthy donors and SLE patients were briefly stimulated with phorbol myristate acetate (PMA), fixed, and cultured with unmanipulated, autologous B cell-depleted PBMCs in the presence of αGalCer and IL-2 (Figure 4B). The increment in CD1d expression on the surface of B cells from SLE patients restored their capacity to promote iNKT cell proliferation and cytokine production (Figures 4C and 4D), equivalent to the rate observed after forcing CD1d to the surface of B cells from healthy donors. Blocking CD1d expression with CD1d mAb, specifically on B cells, prevented this increase. Similar results were obtained when the internalization of CD1d was inhibited in B cells with 0.4 M hypertonic sucrose, known to prevent clathrin-mediated endocytosis (Idkowiak-Baldys et al., 2006), hence retaining CD1d on the surface (data not shown).

Figure 4

Defects in CD1d Surface Expression and Intracellular Recycling on B Cells from SLE Patients

PBMCs from healthy donors and SLE patients with active disease were surface stained with CD19, CD14, and CD1d mAbs.

(A) Representative histograms showing surface expression of CD1d on CD19+ B cells or on CD14+ cells from SLE patients compared to healthy donors. Scatter plots showing mean fluorescent intensity (MFI) of CD1d on B cells from 42 healthy donors and 33 SLE patients. ∗p < 0.05, ∗∗p < 0.01; one-way ANOVA.

(B–D) B cells were negatively isolated from eight SLE patients and six healthy donors and stimulated with PMA for 1 hr to force CD1d to the surface. CD1d-bearing B cells were then fixed and cultured with autologous B cell-depleted PBMCs with αGalCer and IL-2 ± blocking CD1d mAb for iNKT cell expansion for an additional 7 days.

(B) Representative histograms showing increased CD1d surface expression after PMA treatment of B cells from SLE patients.

(C) Bar chart shows mean ± SE fold increase in percentage of iNKT cells at day 7 in CD1d “forced” cells compared to CD1d unforced cells ± blocking CD1d mAb.

(D) Cytokines in supernatants collected from experiment (C) were measured by CBA. Representative CBA plot and cumulative results from two independent experiments each with five healthy donors and five SLE patients, mean ± SE.

(E and F) B cells were negatively isolated from three healthy donors and three SLE patients and stimulated with PMA for 1 hr. B cells were then stained with CD1d mAb on ice. B cells were incubated at 37°C for several time points to allow CD1d endocytosis.

(E) Internalization was determined by flow cytometry after quenching of surface CD1d expression and expressed as mean ± SE percentage CD1d internalization. One of three experiments is shown.

(F) Representative confocal microscopy analysis showing CD1d localization on B cells from five healthy donors and five SLE patients at 0 and 5 min after CD1d endocytosis was allowed to proceed. Cells stained for CD1d (green) and DAPI (blue). Scale bars represent 1 μm. Arrows indicate CD1d internalization.

(G–J) B cells from healthy donors or SLE patients were stained with unconjugated CD1d mAb, incubated at 37°C for 20 min to allow recycling of intracellular CD1d, followed by detection of newly formed surface CD1d with CD1d-PE mAb.

(G) Cumulative data showing MFI of CD1d recycling to the surface of B cells from healthy donors (n = 6) compared to SLE patients (n = 6). Differences in CD1d MFI in B cells incubated at 37°C compared to B cells left on ice are shown, mean percentage ± SE from two experiments. ∗∗p < 0.01; two-tailed t test; similar results were obtained when intracellular protein trafficking was blocked with Brefeldin A, instead of leaving cells on ice (data not shown).

(H) Protein immunoblot analysis of total cellular CD1d protein in B cells from either healthy donors or SLE patients. Cumulative data show quantification relative to actin control. Mean ± SE of three healthy donors and three SLE patients.

(I) PBMCs were cultured in the presence or absence of proteosome inhibitor MG123 for 4 hr. CD1d expression was assessed by flow cytometry. Cumulative data showing fold change in CD1d expression on B cells cultured with MG123 compared to B cells cultured with DMSO as control. Mean ± SE of five healthy donors and five SLE patients.

(J) B cells from healthy donors were negatively isolated and stained on ice for CD1d. B cells were incubated at 37°C for 15 min in the presence or absence of IFN-α and/or anti-Ig F(ab')2 to allow CD1d endocytosis. Internalization was determined as in (E). Bar charts show mean ± SE. CD1d internalization in five healthy donors.

(K and L) iNKT cells were cultured 1:1 with autologous CD19+ B cells for 3 days with αGalCer and IL-2 ± IFN-α and anti-Ig. Both iNKT and B cells were purified by flow cytometry.

(K) Representative flow cytometry plots and cumulative data (on the right) showing frequency of Ki67+ iNKT cells.

(L) IFN-γ in supernatants collected from experiment (K) measured by ELISA. Bar charts with cumulative data from three healthy donors, mean ± SE.

CD1d molecules are synthesized in the endoplasmic reticulum and expressed on the cell surface before being reinternalized and processed through the endosomal and lysosomal compartments (De Silva et al., 2002; Jayawardena-Wolf et al., 2001). Next, we examined whether abnormal CD1d recycling contributed to its reduced surface expression on B cells from SLE patients. We have used PMA stimulation as a tool to study the kinetics of CD1d internalization (Flores-Borja et al., 2008). The results in Figure 4E revealed that B cells from SLE patients had an increased rate of CD1d internalization, which peaked at 5 min, compared to B cells from healthy donors. In contrast, CD1d internalization remained stable over time in B cells from healthy donors. Further analysis by confocal microscopy revealed that CD1d is localized at or near the plasma membrane, immediately after PMA stimulation (0 min), in B cells from both SLE patients and healthy donors (Figure 4F). CD1d on B cells from SLE patients was aggregated compared to a more diffuse distribution on B cells from healthy donors (Figure 4F). After allowing endocytosis to proceed at 37°C, CD1d on B cells from healthy donors remained localized at or near the plasma membrane after 5 min; however, in B cells from SLE patients, CD1d appeared to migrate into the cell, confirming the results obtained by flow cytometry (Figure 4F). Together, these results suggest that increased CD1d internalization in B cells from SLE patients could contribute to its reduced surface expression.

We next compared CD1d recycling dynamics in B cells from SLE patients and healthy donors by using a method previously established (Yuan et al., 2006). B cells were incubated at 4°C with saturating amounts of unconjugated CD1d mAb to block cell surface CD1d molecules. We then incubated B cells at 37°C for 20 min to allow intracellular CD1d molecules to recycle to the cell surface. Because recycled intracellular CD1d molecules were not protected by a prebound CD1d mAb, we could assess their presence by staining with a CD1d-PE mAb and flow cytometry. The results in Figure 4G showed that in B cells from SLE patients, more CD1d was trafficked to the cell surface compared to B cells from healthy donors. To ensure that only preformed CD1d was measured during the time course of the experiment, we inhibited synthesis of new CD1d with cyclohexamide (Chen et al., 2007).

We assessed whether decreased CD1d surface expression reflected a reduction in the total cellular pool of CD1d. No difference in the total cellular expression of CD1d was observed in B cells from SLE patients compared to B cells from healthy donors by protein immunoblotting (Figure 4H). Furthermore, reduced surface CD1d expression was not associated with increased CD1d degradation as indicated by the fact that blocking proteosomal (Figure 4I) and lysosomal function (with MG132 and chloroquine, respectively) (data not shown) did not reveal any differences in total cellular CD1d expression between B cells from healthy donors or from SLE patients. It has been shown previously that surface CD1d molecules belong to a long-lived pool, which continuously recycle between intracellular compartments and the plasma membrane (Jayawardena-Wolf et al., 2001). Therefore, our results suggest that reduced surface expression of CD1d is not associated with an increased degradation of CD1d or with a reduction in total cellular CD1d but rather with a potential defect in trafficking of CD1d in B cells from SLE patients.

So far, our results suggest that inflammatory cues present in SLE patients may be responsible for the reduced expression of CD1d on B cells, which may affect iNKT cell homeostasis. Increased BCR-related signaling and IFN-α production are hallmarks of SLE pathogenesis (Bennett et al., 2003; Lipsky, 2001). Following this line of reasoning, we stimulated B cells from healthy donors with IFN-α alone or in combination with anti-immunoglobulin (Ig) (to cross-link BCRs) and assessed the effect on CD1d internalization. No effect on CD1d recycling was observed when B cells from healthy donors were stimulated with either IFN-α or anti-Ig alone. In contrast, simultaneous stimulation with IFN-α and anti-Ig increased CD1d recycling on B cells from healthy donors, to the rate observed in B cells from SLE patients (Figure 4J). Additionally, IFN-α- and anti-Ig-stimulated B cells had a reduced capacity to directly induce iNKT cell proliferation and IFN-γ production (Figures 4K and 4L).

iNKT Cell Number and Function Is Recovered in SLE Patients Responding to Rituximab Treatment after B Cell Repopulation

Our in vitro results showed that B cells and their expression of CD1d are pivotal for the maintenance of iNKT cell number and function. Therefore, we exploited the opportunity to investigate the effect that in vivo B cell depletion has on iNKT cell homeostasis in SLE patients undergoing B cell-depletion therapy. After rituximab treatment, patients can either relapse or remain in full or partial remission, concomitant with B cell repopulation (Leandro et al., 2005; Ng et al., 2007). Patients were divided into four groups: prior to rituximab treatment (SLE with active disease assessed by British Isles Lupus Assessment Group [BILAG] global score > 6); B cell depleted (BCD); responding to rituximab therapy once B cells had repopulated (rituximab B cell repopulated responding; RBRr; BILAG global score < 6); and not responding to treatment after B cell repopulation (rituximab B cell repopulated nonresponding; RBRnr; global score BILAG > 6). A detailed definition of the BILAG index clinical assessment, clinical information, and treatment regimes of SLE patients included in the study are provided in Table S1.

We compared iNKT and B cell frequencies in the four patient groups (Figures 5A–5C). The number of iNKT cells remained low in PB of patients during the B cell depletion phase (BCD) independent of clinical response and in patients not responding to treatment (RBRnr). However, iNKT cell frequency and absolute number were dramatically restored in the responding group of patients (RBRr) (Figure 5C). Longitudinal analysis of 37 individual patients either responding (RBRr) or not responding (RBRnr) to B cell depletion therapy confirmed that iNKT cell numbers were mostly recovered in the responding patients (76.9%) but generally remained low in the RBRnr (87.5%) or in patients with active SLE treated with nonbiological therapies (Figures S2A and Figures S2B–S2D). Although there is a general variability in iNKT cell numbers among healthy donors (Lee et al., 2002), the frequencies of iNKT cells remained constant in the same healthy donor over time (Figure S2A).

Figure 5

iNKT Cell Number Is Recovered in SLE Patients Responding to Rituximab after B Cell Repopulation

Ex vivo PBMCs from SLE patients before, during, and after rituximab treatment stained for expression of CD19, CD3, and iTCR.

(A) Representative flow cytometry plots showing frequency of iNKT and B cells from SLE patients, B cell-depleted SLE patients (BCD), repopulated nonresponding SLE patients (RBRnr), and repopulated responding patients (RBRr).

(B and C) Scatter plots show mean percentage of B cells (B) and percentage and absolute number of iNKT cells (C) in PB of 254 SLE, 92 BCD, 32 RBRnr, and 113 RBRr patients. ∗p < 0.05, ∗∗p < 0.01; one-way ANOVA.

(D and E) PBMCs isolated from SLE patients before, during, and after rituximab treatment were cultured with αGalCer and IL-2 for 7 days. Supernatants were collected and cells were stained with CD3 and iTCR mAbs. Representative flow cytometry plots at day 7 (D) and bar charts showing the cumulative data of iNKT cells (E) before and after the 7 day culture from 10 SLE, 10 BCD, 13 RBRnr, and 13 RBRr patients. ∗∗p < 0.01; paired t test.

(F) Cytokines from supernatants collected in experiment (D) were measured by CBA. Bar charts of cumulative results from three independent experiments with 10 SLE, 10 BCD, 13 RBRnr, and 13 RBRr patients, mean ± SE. ∗p < 0.05, ∗∗p < 0.01; one-way ANOVA.

(G) Ex vivo PBMCs from SLE patients before, during, and after rituximab treatment stained for expression of CD19 and CD1d. Scatter plots showing CD1d MFI on B cells in PB of 42 healthy donors, 33 SLE patients, 21 RBRnr, and 66 RBRr patients. ∗p < 0.05; one-way ANOVA.

(H) Bar chart showing CD1d internalization on B cells from three healthy donors, three SLE patients, three RBRnr, and three RBRr patients after 5 min. One of three experiments is shown. Mean percentage ± SE. ∗∗∗p < 0.001; one-way ANOVA.

(I) Confocal microscopy analysis depicting CD1d localization on B cells from one RBRnr and one RBRr patient before (0) and 5 min after CD1d endocytosis. Cells were stained for CD1d (green) and DAPI (blue). Scale bars represent 1 μm. Arrows indicate areas of CD1d internalization.

Similar to iNKT cells from SLE patients prior to rituximab treatment (Figure 3A), iNKT cells from B cell-depleted patients (BCD) and from those not responding to treatment (RBRnr) failed to expand (Figures 5D and 5E) or to produce a healthy cytokine profile in response to αGalCer and IL-2 stimulation in 7 day cultures (Figure 5F). In contrast, the capacity of iNKT cells to expand and to drive production of IL-2, TNF-α, IFN-γ, and IL-10 was normalized in B cell-repopulated patients responding to treatment (RBRr) (Figures 5D–5F). Of note, CD1d surface expression and internalization rate was restored in RBRr but not in RBRnr patients (Figures 5G–5I). These results suggest that in this set of patients, newly repopulated B cells may be reset to a healthy status, as recently suggested (Anolik et al., 2007), and no longer display the abnormality present in patients with active SLE.

Immature B Cells from SLE Patients Present Reduced CD1d Expression

We have shown previously, and confirmed here, that immature B cells (CD19+CD24hiCD38hi) from healthy donors express increased amounts of CD1d compared to mature B cells (Figures 6A and 6B; Blair et al., 2010). No significant differences were observed in the frequency of immature B cells between SLE, RBRnr, and RBRr patients (Figure 6A); however, analysis of mature (CD19+CD24intCD38int), memory (CD19+CD24hiCD38−), and immature B cells in healthy donors and in our cohort of SLE patients revealed that CD1d expression was reduced exclusively in immature B cells from SLE patients (Figure 6B). CD1d expression remained low on immature B cells from RBRnr patients but was restored to healthy amounts in RBRr patients (Figure 6B). A statistically significant positive correlation between immature B cell and iNKT cell frequencies (but not mature or memory B cells) was seen only in healthy donors and in RBRr patients (Figure 6C). The significance of these results was increased further when those immature B cells expressing the highest CD1d were correlated with iNKT cell frequency (Figure 6D).

Figure 6

CD1dhi Immature B Cells Positively Correlate with the Frequency of iNKT Cells in Healthy Donors and Patients Responding to Rituximab

PBMCs from either healthy donors or SLE patients were stained for expression of CD19, CD24, CD38, and CD1d to identify B cell subsets.

(A) Representative flow cytometry plot showing CD19+CD24hiCD38hi (immature), CD19+CD24+CD38− (memory), and CD19+CD24intCD38int (mature) B cells (Blair et al., 2010). Representative histograms displaying CD1d expression on mature, memory, and immature B cells from healthy donors, SLE patients, RBRnr, and RBRr patients.

(B) Scatter plots showing frequency of immature B cells in 25 healthy donors, 134 SLE patients, 21 RBRnr, and 66 RBRr patients.

(C) Volcano plot showing individual Spearman's correlations for frequency of iNKT cells compared to frequency of mature, memory, and immature B cells from healthy donors, SLE patients, RBRnr, and RBRr patients versus the statistical significance.

(D) Representative dot plot showing the gating strategy for CD1dhi expression on immature B cells. Correlation plots show the relationship between frequency of CD1dhi B cells and iNKT cells in RBRr patients. Each dot, representative of a patient, is colored based on its disease activity calculated with the BILAG scoring system. Spearman's correlation.

Immature B Cells from Healthy Donors Are More Proficient in Promoting iNKT Cell Proliferation and Cytokine Production than Other B Cell Subsets

Mature or immature B cells from healthy donors were isolated from PBMCs and cocultured 1:1 with isolated autologous iNKT cells for 3 days in the presence of αGalCer and IL-2. iNKT cell proliferation and IFN-γ production was consistently higher in immature B cell cocultures, compared to mature B cells (Figures 7A and 7B). The experiments described above are technically difficult to perform because of the scarcity of both cell populations in PB. To dissect further the role of CD1dhi immature B cells in iNKT cell homeostasis, we depleted mature or CD1dhi immature B cells from PBMCs prior to stimulation with αGalCer and IL-2. Depletion of immature B cells reduced iNKT cell expansion and cytokine production compared to whole PBMCs and depletion of mature B cells (Figures 7C and 7D). These results indicate a preferential role for immature B cells compared to mature B cells in iNKT cell homeostasis in healthy donors.

Figure 7

iNKT Cell Proliferation and Cytokine Production Is Preferentially Induced by Immature Compared to Mature B Cells

Flow cytometry-sorted iNKT cells from healthy donors were cultured 1:1 with autologous flow cytometry-sorted immature or mature B cells for 3 days with or without αGalCer and IL-2.

(A) Representative flow cytometry plots of three healthy donors showing frequencies of Ki67+ iNKT cells.

(B) IFN-γ in supernatants collected from experiment (A) measured by ELISA. A representative experiment of two is shown.

PBMCs from healthy donors were stained for CD19, CD24, and CD38 and gated as shown in Figure 6A. Immature or mature B cells were depleted, by flow cytometry sorting, from PBMCs and the remaining PBMCs were collected, cultured for 7 days with αGalCer and IL-2, and surface stained with CD3 and iTCR mAbs.

(C) Representative flow cytometry plots depicting iNKT cells from undepleted, mature, or immature B cell-depleted PBMCs of three healthy before and after 7-day culture.

(D) Cytokines in supernatants collected from experiment (C) measured by ELISA. Bar charts show cumulative data from six healthy donors, mean ± SE. ∗p < 0.05; two-tailed t test.

We propose that among different B cell subsets, immature B cells with high expression of CD1d are pivotal for iNKT cell homeostasis in healthy donors. Furthermore, defects in CD1d expression in immature B cells, possibly induced by factors contributing to SLE pathogenesis, may play a role in iNKT dysfunction in these patients.

Discussion

Several reports have demonstrated that B cells interact either directly or indirectly with iNKT cells (Bialecki et al., 2009; Tonti et al., 2009) and that B cell presentation of αGalCer to iNKT cells generates a specific activation signal that drives memory B cell or plasma cell differentiation (Barral et al., 2008; Galli et al., 2007; Lang et al., 2008; Leadbetter et al., 2008). However, the effect that lipid-antigen presentation by B cells has on iNKT cell function remains largely uninvestigated.

We report a previously unappreciated role for B cells in the maintenance of iNKT cell homeostasis. B cells from SLE patients showed altered CD1d recycling and impaired iNKT cell activation a phenomenon that could be observed in B cells from healthy donors after simultaneous stimulation with IFN-α and anti-Ig. IFN-α is implicated in SLE etiology and pathogenesis (Pascual et al., 2003). SLE-associated autoantibodies can drive IFN-α release and its increased production can inhibit immune regulation (Pace et al., 2010). Excessive signaling via the BCR and elevated circulating IFN-α may account for the defective CD1d-mediated lipid presentation by B cells leading to reduced iNKT cell frequency. Our results suggest a sustained B cell dysfunction as indicated by the fact that patients with reduced disease activity undergoing nonbiological treatment do not recover CD1d expression on B cells (data not shown). Although CD1d expression on B cells is required for iNKT cell proliferation and activation, further investigations are needed to understand whether factors other than CD1d (which could have been induced in our experimental system) are important for the maintenance of iNKT cells.

A remaining conundrum is why in vivo APCs expressing a normal amount of CD1d do not compensate for the reduction of CD1d on B cells, especially because DCs are shown to be the most effective iNKT cell stimulators (Bezbradica et al., 2005; Im et al., 2009). It is possible that in addition to CD1d, B cells provide other signals essential for iNKT cell expansion, absent on DCs and monocytes. Alternatively, a determining factor may be the physiological location of the interaction. More experiments are needed to fully clarify the mechanisms underlying reduced surface expression of CD1d exclusively on B cells from SLE patients including its intracellular location during recycling and its impact on iNKT cell homeostasis.

It has been suggested that the reduction of iNKT cells in autoimmunity is genetically linked, a finding supported by the concomitant iNKT cell reduction identified in first-degree relatives of SLE patients (Wither et al., 2008). Of note, many of these relatives suffered from autoimmune diseases, which could account for their reduced iNKT cell frequencies. The longitudinal analysis of rituximab-treated patients shows that reduced iNKT cell frequency was a reversible phenomenon, restored in responding patients upon reconstitution of immature B cells with normal CD1d expression. Furthermore, iNKT cells from SLE patients were able to respond normally when primed by autologous mDCs, suggesting an acquired rather than intrinsic defect.

It is possible that reduced iNKT cell number in PB could be the result of iNKT cells migrating to sites of inflammation. Inflammation per se did not affect iNKT cell number in PB, as shown by the fact that disease activity did not reflect a variation in iNKT cell frequency. Restoration of iNKT cell numbers was a primary feature of patients responding to rituximab treatment after B cell repopulation, as shown by the fact that SLE patients with inactive disease treated with nonbiological therapies or during the B cell-depletion phase (no detectable B cells in PB) had a lower number of iNKT cells compared to healthy donors. Of interest, the peripheral numerical deficiency seems to be confined to iNKT cells as other immune-regulatory cells, such as regulatory B cells (Breg) are increased in the periphery of patients with active SLE (Blair et al., 2010).

iNKT cells from SLE patients failed to proliferate and release IFN-γ but produced equivalent amounts of IL-4. Previous work in mice has shown that a single administration of αGalCer induces iNKT cell anergy, defined by their inability to proliferate and produce IFN-γ while producing IL-4. Anergy was reversed after addition of exogenous IL-2 (Parekh et al., 2005). Expression of programmed death-1 (PD-1) was increased in iNKT cells from SLE patients (data not shown), characteristic of an anergic or exhausted phenotype in iNKT and conventional T cells (Keir et al., 2008; Parekh et al., 2009). In vivo, multiple factors contribute to the profound hyporesponsive phenotype of iNKT cells from SLE patients, including exposure to elevated quantities of immunosuppressive cytokines such as IL-10 or transforming growth factor-β (TGF-β), abundantly present in these patients (Isenberg and Rahman, 2006; and E.C.J., data not shown).

The biological significance of iNKT cell deficiency in autoimmune patients remains controversial. This is primarily a result of the lack of uniformity in the methods used to identify iNKT cells and to the limited functional analysis that has been performed (Berzins et al., 2011). Most information supporting an immune-regulatory role for iNKT cells in autoimmunity comes from studies performed in mice (Cerundolo et al., 2009). Experimental strategies focused on increasing the number or stimulation of iNKT cells in a variety of experimental models show an enhanced degree of protection against the development of autoimmunity (Mars et al., 2004). Protective mechanisms exerted by iNKT cells include inhibition of autoreactive B cells, induction of tolerogenic DCs, and the indirect induction of Treg cells (Ly et al., 2006; Naumov et al., 2001; Wermeling et al., 2010; Yang et al., 2011). The finding showing that iNKT cell numbers were restored in patients responding to rituximab only after the repopulation of B cells with normalized CD1d expression suggests that iNKT cells may contribute directly or indirectly to the amelioration of SLE.

The benefit of rituximab therapy in SLE patients has been attributed to different mechanisms of action including depletion of effector B cells from target organs followed by a reduction of autoantibody production, restoration of normal Th1 and Th2 cytokine balance, and an increase of Treg cells, although this remains controversial (Edwards and Cambridge, 2006; Sfikakis et al., 2007). We provide an additional mechanism showing that B cell-depletion therapy may rebalance altered immune responses in SLE patients by restoring defective iNKT cells to healthy function in responding patients. Immature B cells from healthy donors, expressing the highest amount of CD1d among different B cell subsets, possess a regulatory function and suppress CD4+ T cell activation via the provision of IL-10 (Blair et al., 2010). In healthy donors, this population promotes higher iNKT cell proliferation and IFN-γ production compared to mature B cells. We have shown that CD1d expression was reduced exclusively on immature B cells from SLE patients and was recovered in responding patients. Statistical analysis of our data revealed a significant positive correlation between the iNKT cell percentages, immature B cells, and a better clinical outcome. Expansion of immature B cells has been shown previously to correlate with a long-term favorable response in SLE and in RA patients treated with rituximab (Anolik et al., 2007; Leandro et al., 2006). It is tempting to envisage a scenario where Breg cells, as opposed to other CD1d-expressing B cells, impart the differentiation of iNKT cells with immune-regulatory function via presentation of endogenous lipid or protein antigen.

In summary, we report that B cell-iNKT cell interaction is important for the maintenance of iNKT cell homeostasis and that it requires the presence of functionally normal B cells. Additionally, we propose that B cell-iNKT cell cross-talk is disrupted in SLE patients and that rituximab treatment may work, in part, by resetting the outcome of B cell-iNKT cell interactions.

Experimental Procedures

Patients and Controls

Table S1 provides detailed information of patient characteristics and treatment. Ethical approval was obtained from University College Hospital ethics committee.

PBMC, B Cell, and iNKT Cell Isolation

PBMCs were isolated by Ficoll-Paque Plus (GE Healthcare) gradient centrifugation, resuspended to 107 cells/ml in FCS (Biosera) and 5% DMSO (Sigma-Aldrich), and frozen until subsequent use. Unmanipulated B cells were negatively isolated and monocytes or B cells were depleted from PBMCs by positive selection with magnetic beads (Miltenyi, Biotec). B cell subsets or iNKT cells were isolated with the FACSAria (Beckton, Dickinson) based on expression of CD19, CD24, CD38 or CD3, and iTCR, respectively. Alternatively, iNKT cells were isolated with anti-iTCR magnetic beads (Miltenyi, Biotec). Figures S5A–S5C shows purity of B cells and iNKT cells. Absolute numbers from healthy donors were calculated by multiplying 10× the total number of cells isolated from 100 ml of blood by the fraction of PBMCs contained within each subset determined by flow cytometry.

Cell Cultures

For iNKT cell expansion, PBMCs were cultured in RPMI 1640, L-glutamine and NaHCO3 (Sigma-Aldrich), 10% human serum type AB (Lonza), 100 U/μg/ml penicillin/streptomycin (Sigma-Aldrich), 200 IU/ml recombinant human (rh)IL-2 (R&D Systems), and 100 ng/ml αGalCer (Alexis Biochemicals). For iNKT and B cell cocultures, negatively isolated B cells were pulsed with 100 ng/ml αGalCer for 1 hr at 37°C. B and iNKT cells were cocultured with isotype control or 10 μg/ml purified CD1d blocking mAb (51.1). Alternatively, PBMCs from healthy donors or SLE patients were cultured in the presence or absence of a protease inhibitor (MG123; Sigma-Aldrich) for 4 hr to determine the rate of CD1d proteosomal degradation. The efficacy of MG123 was confirmed by assessing the accumulation of ubiquitinated protein species by protein immunoblotting (data not shown). Experimental procedures related to CD1d quantitative immunoblotting, CD1d internalization, and recycling are described in Supplemental Experimental Procedures.

Forcing CD1d to the B Cell Surface

Negatively isolated B cells from SLE patients were stimulated with 10 ng/ml PMA (Sigma-Aldrich) for 1 hr to force CD1d to the surface, fixed with 2% PFA (Sigma-Aldrich), and cultured with untouched, autologous B cell-depleted PBMCs for 7 days in the presence of 200 IU/ml rhIL-2 and 100 ng/ml αGalCer for iNKT cell expansion.

Thymidine Incorporation

B and iNKT cell cocultures were incubated with 0.5 μCi (0.0185 MBq)/well [3H]thymidine (TdR; Hartmann Analytik) for the last 16 hr of culture. Cells were harvested (Tomtec) prior to quantification of [3H]TdR incorporation on a scintillation counter (PerkinElmer).

Flow Cytometry

For iNKT cell staining, CD3-FITC, iTCR-PE, CD69-PECy7, and CD25-APC mAbs were used. For staining CD1d on monocytes and B cells, CD14-FITC, CD19-PECy7, CD38-PECy5, CD24-FITC, and CD1d-PE mAbs were used. Purity of monocyte-derived DCs was determined with DC-SIGN-APC mAb. Ki67 expression was detected according to manufacturer's instructions (BD, Bioscience). For analysis of human intracellular cytokine production, PBMCs were stimulated for 4 hr with 50 ng/ml PMA, 250 ng/ml ionomycin (Sigma-Aldrich), and Golgi-Plug (BD, Biosciences) in complete medium. For intracellular staining, cells were stained with combinations of CD3-FITC, iNKT-PE, IFN-γ-PECy7, IL-10-APC, or TNF-α-APC mAbs or appropriate isotype controls. Data were acquired on an LSRII (Beckton, Dickinson) and analyzed with FlowJo (TreeStar). Supernatants were collected from PBMC cultures for iNKT cell expansion. IL-2, IL-4, TNF-α, IL-6, IL-10, and IFN-γ were measured in supernatants by CBA (BD Biosciences) as described by the manufacturer's instructions and analyzed on a FACSArray (Beckton, Dickinson). Alternatively, TNF-α, IL-10, and IFN-γ were measured in supernatants by ELISA (R&D Systems).

Statistical Analysis

Values are expressed as absolute mean ± SE. Data were tested for normal distribution by the Kolmogorov-Smirnov test and analyzed for significance in GraphPad Prism (La Jolla, CA) by two-tailed t test, paired t test, or one-way ANOVA as specified. Correlation coefficients and their significance were calculated by two-tailed Spearman's rank correlation.