Uptake and Presentation of Myelin Basic Protein by Normal Human B Cells

B cells may play both pathogenic and protective roles in T-cell mediated autoimmune diseases such as multiple sclerosis (MS). These functions relate to the ability of B cells to bind and present antigens. Under serum-free conditions we observed that 3–4% of circulating B cells from healthy donors were capable of binding the MS-associated self-antigen myelin basic protein (MBP) and of presenting the immunodominant peptide MBP85-99, as determined by staining with the mAb MK16 recognising the peptide presented by HLA-DR15-positive cells. In the presence of serum, however, the majority of B cells bound MBP in a complement-dependent manner, and almost half of the B cells became engaged in presentation of MBP85-99. Even though complement receptor 1 (CR1, CD35) and CR2 (CD21) both contributed to binding of MBP to B cells, only CR2 was important for the subsequent presentation of MBP85-99. A high proportion of MBP85-99 presenting B cells expressed CD27, and showed increased expression of CD86 compared to non-presenting B cells. MBP-pulsed B cells induced a low frequency of IL-10-producing CD4+ T cells in 3 out of 6 donors, indicating an immunoregulatory role of B cells presenting MBP-derived peptides. The mechanisms described here refute the general assumption that B-cell presentation of self-antigens requires uptake via specific B-cell receptors, and may be important for maintenance of tolerance as well as for driving T-cell responses in autoimmune diseases.


Introduction
In addition to producing antibodies, B cells are highly efficient antigen-presenting cells (APCs) and produce a variety of cytokines [1]. B cells are capable of taking up small amounts of their cognate antigen and presenting it to T cells [2]. Complement receptors (CRs) may contribute to antigen uptake by B cells, either by cross linking CR2 and the B-cell receptor (BCR), or as a BCRindependent internalisation receptor [3,4]. In contrast to antigenspecific BCRs, CRs recognise antigens coated with fragments of complement component 3 (C3) or in the context of complementcoated immune complexes [4][5][6][7][8][9][10][11]. CR2-mediated antigen uptake by B cells bypasses the need for antigen specificity, and increases the proportion of B cells engaging in antigen-presentation [12]. We have previously shown that CR2 contributes to B-cell binding of the self-antigen thyroglobulin, which is capable of forming immune complexes with naturally occurring or disease-associated autoantibodies [12,13]. It is not known, however, whether CR2dependent uptake is sufficient for presentation of self-antigens to occur. Depending on the circumstances, this could either potentiate immune responses or mediate T-cell tolerance.
Recently, much research has focused on a subset of B cells with immunoregulatory potential, known as regulatory B cells (Bregs) [14][15][16][17]. These B cells assist in maintaining peripheral tolerance by secreting immunoregulatory cytokines [15,17]. The phenotypic definition of Bregs is still controversial because production of the immunomodulating cytokine interleukin-10 (IL-10) is their only hallmark [14]. Moreover, several studies have demonstrated crosstalk between Bregs and regulatory T cells (Tregs) [18][19][20] and, apart from IL-10 production [20], especially the expression of CD80 and CD86 seems important in this interaction [18,20]. Activated B cells derived from MS patients show decreased IL-10 production [21]. Usually, polyclonal stimuli such as toll-like receptor ligands are used to stimulate human B cells to produce IL-10 (for review see [22]), but the self-antigen thyroglobulin also induces IL-10 production by approximately 1% of normal B cells [23].
Propathogenic B cells are involved in the maintenance of autoimmune diseases, as demonstrated by the beneficial effect of the B cell-depleting antibody rituximab in a number of autoimmune diseases [24]. These include relapsing-remitting multiple sclerosis (MS) [25,26], an inflammatory, demyelinating disease of the central nervous system (CNS) characterised by an immunological attack on the myelin sheath in the CNS orchestrated by autoreactive CD4+ T cells [27]. MS is associated with the human histocompatibility leukocyte antigen (HLA)-DR15 haplotype [28], indicating that major histocompatibility complex class II-restricted presentation of CNS-derived antigens is important in the disease process. Reduced relapse rates in the first 24 weeks of B-cell depletion without a significant influence on total antibody level [25] suggest that the pathogenic role of B cells is associated with antigen-presentation [29] and secretion of pro-inflammatory cytokines [30], rather than with antibody production. B-cell numbers are elevated in the CNS in the majority of MS patients [31].
B-cell antigen presentation is usually studied by indirect measurement of the resulting T-cell response [5,6,10,12,32]. Using CD4+ T-cell activation as read-out, we and others have previously examined the ability of B cells to present self-antigens such as thyroglobulin [23] and aggrecan [33]. However, antigen presentation that leads to downregulation of T-cell responses is difficult to assess in this manner, and information about the proportion and phenotype of the B cells presenting the antigen is usually missing. Here we examine directly the B-cell uptake and presentation of the self-antigen myelin basic protein (MBP), a selfantigen considered to be involved in the pathogenesis of MS [27], exploiting the recognition of the immunodominant peptide MBP85-99 presented on HLA-DR15 by mAb MK16 [28]. We also aimed to determine the role of complement receptors in the process, and to characterise the phenotypic profile of the B cells that most efficiently present MBP85-99.

Cells and serum
Peripheral blood mononuclear cells (PBMCs) and serum were isolated from healthy blood donors attending the Blood Bank at Copenhagen University Hospital Rigshospitalet in tubes containing heparin or no anti-coagulant (BD Bioscience, San Jose, CA). Buffy coats or heparin blood derived from 6 HLA-DR15-positive donors were used for experiments analysing the presentation of MBP peptide. Another 25 healthy blood donors with unknown HLA-tissue type were used for i) assessing the role of complement inactivation and blockade of CR1/CR2/FcRs on surface binding of MBP, ii) verifying classical complement activation in the presence of serum and iii) in co-culture studies of B cells and T cells. The donors were anonymous to the investigators, and thus no local Ethical Committee approval was required according to Danish legislation. Gradient centrifugation over LymphoPrep (Axis-Shield, Oslo, Norway) was used to isolate PBMCs. The cells were washed twice in phosphate buffered saline (GIBCO, Invitrogen, Carlsbad, CA, USA) and were resuspended in Roswell Park Memorial Institute (RPMI) 1640 medium containing HEPES (Biological Industries Israel Beit-Haemek Ltd, Kibbutz Beit-Haemek, Israel), L-glutamine (GIBCO) and gentamicin (GIBCO). Cells were either used directly in MBP surface-binding experiments or stored in liquid nitrogen before use in MBP peptidepresentation experiments. PBMCs were labelled with 5-carboxy-29,79-dichlorofluorescein diacetate succinimidyl ester (CFSE) at 0.25 mM in RPMI 1640 for 10 min at 37uC. CD19+ B cells and CD3+ T cells were purified from freshly purified PBMCs using the Human B cell Enrichment Kit or CFSE-labelled PBMCs using the Human CD3 Selection Kit (StemCell Technologies Inc, Vancouver, Canada).
Unless otherwise stated, serum from blood group AB donors (Lonza, Basel, Switzerland) was used as the source of normal human serum (NHS).

Antigens
Whole human MBP was purchased from HyTest Ltd. (Turku, Finland) and was used either unconjugated or conjugated with biotin using the LYNX rapid conjugation kit (AbD serotec, Kidlington, UK), according to the manufacturer's instructions. Tetanus toxoid (Statens Serum Institut, Copenhagen, Denmark) and thyroglobulin (Biogenesis Ltd., Poole, England) were used as control antigens.

Antibodies
The monoclonal antibody MK16 that recognizes MBP85-99 in the context of HLA-DRB1*1501 [28] was used as probe for antigen presentation. MK16 was originally obtained by phage display technology [28] in the Fab format, and was subsequently modified into a murine IgG1 antibody expressed in Chinese hamster ovary (CHO) cells [34]. The MK16 IgG1 antibody (referred to in the following as MK16) was affinity-purified by protein A from the supernatant of the MK16-expressing CHO cells grown in HAMS F-12 media (GIBCO) supplemented with 10% fetal calf serum (FCS; Biological Industries) and 0. Proliferation assay in co-cultures of B cells and T cells 250,000 CFSE-labelled CD3+ T cells were mixed with 100,000 CD19+ B cells plus 30 mg/ml MBP. Cells were cultured in RPMI 1640 in round-bottomed 96-well plates containing 30% (v/v) NHS for 7 days at 37uC and 5% CO 2 . As positive controls, 250,000 CFSE-labelled CD3+ T cells were added to wells coated with anti-CD3 (OKT3) at a concentration of 0.5 mg/ml (eBioscience, San Diego, CA, USA). At day 7, cells were stained for expression of CD4, and proliferation was measured by flow cytometry on a FACS Calibur cytometer (BD Biosciences). Background proliferation was assessed in cultures without added MBP.
IL-10 and TNF-alpha secretion assay 500,000 CD3+ T cells were mixed with 100,000 CD19+ B cells 630 mg/ml MBP. Cells were cultured in RPMI 1640 in flatbottomed 96-well plates containing 30% (v/v) autologous serum for 18 h at 37uC under 5% CO 2 . As positive control, cells were cultured with staphylococcal enterotoxin B (SEB) at a concentration of 1 mg/ml. Cultures containing only T cells were used as negative controls. Production of IL-10 and TNF-alpha were measured using MACS cytokine secretion assay for IL-10 and TNF-alpha according to the manufacturer's instructions (Miltenyi Biotec, Bergisch Gladbach, Germany). Samples containing only detection antibodies were included as controls and these values were subtracted from all other samples. B-cell preparations contained 97.760.47% B cells, 1.560.43% T cells, and 0.6860.15% monocytes, while T-cell preparations contained 98.960.40% T cells, 0.92%60.38 B cells, and 0.1560.05% monocytes. For each sample, between 100,000 and 150,000 CD4+ T cells were recorded using a FACS Canto flow cytometer (BD Biosciences).

Assessment of MBP deposition on B cells
0.3610 6 PBMCs from healthy donors were incubated for 30 min at 37uC under 5% CO 2 in LGM-3 media (Lonza, Walkersville, MD, USA) with 30 mg/ml biotinylated MBP either in the absence of serum, or in the presence of 30% (v/v) autologous serum, 30% (v/v) heat-treated (30 min at 56uC) autologous serum, or 30% (v/v) autologous serum containing 50 mM EDTA or sodium polyanethole sulfonate (SPS) at a concentration of 0.2 mg/ml or 2 mg/ml respectively (Sigma, St Louis, MO). Cells were washed twice in PBS/2% FCS and then incubated with 0.6 mg/ml streptavidin-PE and APC-anti-human CD19 for 30 min at 4uC. Binding of MBP to B cells was measured as mean fluorescence intensity (MFI) values on the total CD19+ population. Cells were analyzed by flow cytometry using a FACS Calibur cytometer (BD Biosciences). To stain for dead cells, 7actinomycin D (7-AAD) (BD Biosciences) was added to samples before acquisition.

Assessment of MBP presentation
Aliquots of MK16 were conjugated with biotin using the LYNX rapid conjugation kit (AbD serotec, Kidlington, UK), according to the manufacturer's instructions, or with fluorescein isothiocyanate (FITC; Sigma-Aldrich GmbH, USA) to an FITC:protein ratio of 6:1.
0.5610 6 HLA-DR15+ PBMCs were incubated for 18 h at 37uC under 5% CO 2 in media containing 30% v/v AB serum plus 30 mg/ml of whole MBP. In some experiments thyroglobulin and tetanus toxoid were included as controls at a final concentration of 30 mg/ml. Next, the cells were incubated with IgG for intravenous use (IVIg; CSL Behring, Bern, Switzerland) at a concentration of 6 mg/ml and 2% mouse serum (Statens Serum Institut, Copenhagen, Denmark) to block unspecific binding. Subsequently, MK16 was incubated at a concentration of 50 ng/ml for 30 min at 4uC in 2% FCS; antibodies against cell-surface markers were included in the same step. Following two washes, streptavidin-PE (BD Biosciences) was incubated with the samples for 30 min at 4uC in experiments using biotinylated MK16. Finally, cells were analysed on a FACS Canto flow cytometer (BD Biosciences). To exclude dead cells, 7-AAD was added to samples before acquisition.
Assessment of complement deposition on B cells 0.3610 6 PBMCs were incubated for 5 or 15 min at 37uC in LGM-3 media 630 mg/ml of MBP and 630% v/v autologous serum. Afterwards the tubes were kept on ice, and cells were stained with FITC-anti-human C3, biotin-anti-human C1q, and PE-Cy7-anti-human CD19 followed by a second stain by streptavidin-PE. To exclude dead cells, 7-AAD was added to samples before acquisition. Cells were analysed on a FACS Canto flow cytometer (BD Biosciences).

Inhibition of B-cell binding and presentation of MBP by CR1 and CR2 blockade
To inhibit uptake and presentation of MBP by B cells, 0.3610 6 PBMCs and 1610 6 PBMCs respectively were incubated for 30 min at room temperature with monoclonal anti-CR1 (clone 3D9), polyclonal anti-CR2 antibodies, or a combination of the two, each at a final concentration of 10 mg/ml. As negative control, an irrelevant antibody, anti-human glycophorin (GP)-A (CD235a) was used at a concentration of 10 mg/ml. In experiments analysing presentation of MBP, cells were incubated for 1.5 h or 4 h at 37uC with 30 mg/ml MBP in media containing 30% v/v NHS. Excess MBP was washed away, and cells were incubated for 18 h at 37uC in media containing 30% v/v NHS. Otherwise, the experiments were carried out as described above.
Background MFI values from samples incubated without MBP were subtracted from all values.
Student's paired t-test was used. Kolmogorov-Smirnovs test was used to test for normality. P-values,0.05 were considered statistically significant.

Binding of MBP to normal B cells
We assessed the binding of MBP to B cells in cultures of normal PBMCs. In the absence of serum, little binding occurred: only 2.561.5% (mean6SEM) of the cells stained positive for MBP binding (Figs. 1A and B). By contrast, addition of autologous serum to the medium resulted in a shift of the entire B-cell population towards higher MBP binding. Under these conditions, 65.468.2% of the cells had MFI values above the negative control (no addition of MBP).
In view of previous findings that complement promotes the uptake of antigens by B cells [6,9,10,12], we examined the effect of heating serum to 56uC, which is known to inactivate heat-labile factors of the complement system [35]. Moreover, we also added EDTA or sodium polyanethole sulphonate (SPS) as a different means of preventing complement activation [36,37]. As shown in Fig. 1C, heat treatment of serum lowered the binding of MBP to B cells by 61.3% on average, while EDTA reduced the binding by 86.1%. SPS reduced the binding of MBP to B cells by 71.6% at a concentration of 0.2 mg/ml and 81.9% at 2.0 mg/ml. Taken together, these data strongly imply that complement enhances the binding of MBP to B cells. Accordingly, we observed that C3fragments and C1q co-deposited with MBP on the B-cell surface ( Figs. 2A and B).
Antibody-mediated blockade of either CR1 or CR2 markedly lowered the binding of MBP to the B cells, while simultaneous blockade of both receptors virtually abrogated MBP binding (Figs. 2C and D). On the contrary, blockade of FccRIIa,b,c (CD32) known to be expressed by B cells, did not affect the binding of MBP to the B cells (Fig. S1).
Presentation of the MBP85-99 peptide by normal HLA-DR15+ B cells after culture with whole MBP protein To study antigen presentation by B cells, isolated PBMCs from HLA-DR15-positive donors were incubated with whole MBP. Subsequently, the mAb MK16 was used as probe for presentation of the immunodominant MBP peptide MBP85-99 [28] (Figs. 3A  and B). As shown in Fig. 3B, only 3.762.4% CD19+ B cells presented MBP peptides in the absence of serum. In the presence of serum, however, 42.269.4% of the B-cell population presented MBP85-99. Binding of the MK16 antibody to B cells from DR15negative donors was also examined to validate the antibody's specificity (Fig. 3C). As expected, MK16 did not bind to MBPstimulated B cells from DR15-negative subjects, nor to B cells incubated with a different self-antigen, human thyroglobulin, or tetanus toxoid, a foreign recall antigen.

Influence of serum concentration and complement activity on B-cell presentation of MBP85-99
Little B-cell presentation of MBP85-99 was observed after incubation of PBMCs with MBP in medium containing only 0.1% of serum (Fig. 4A). At serum concentrations above 3%, however, the peptide was efficiently presented. To examine if complement was the serum factor responsible for enhancing the presentation of MBP85-99, in analogy to its role in binding of MBP by B cells, SPS was used as complement inhibitor and, indeed, dosedependently reduced the presentation of MBP85-99 (Fig. 4B).
The presence of polyclonal anti-CR2 antibodies during incubation of PBMCs with MBP for 1.5 or 4 h markedly lowered the presentation of MBP85-99 by B cells (Fig. 4C). By contrast, coincubation with anti-CR1 mAb had no effect (Fig. 4C).
The most outstanding finding was that a high proportion (around 50%) of B cells presenting MBP85-99 (MK16 positive cells) expressed CD27, which is considered a memory B-cell marker [38], compared to only 20% of the MK16 negative B cells, as shown in Fig. 5B.
Notably, MBP85-99-presenting B cells were not enriched with any of the markers CD5, CD1d, or CD24, which have all been associated with Bregs [20,23,39], nor with IgM (Fig. S2). CD86 and CD80 were found to be constitutively expressed by B cells, and their expression was independent on addition of MBP (Figs. 5C and D). Interestingly, however, B cells presenting MBP85-99 showed increased expression of CD86 compared to MBP85-99 negative B cells (Fig. 5C). We did not observe a corresponding increase in the expression of CD80 (Fig. 5D).
In agreement with their lack of phenotypic Breg markers, B cells presenting MBP85-99 with high efficiency did not produce IL-10 or IL-6 when stimulated with MBP alone (Fig. S3).

Low frequency of IL-10 secreting MBP specific CD4+ T cells in co-cultures of T cells and B cells
In co-cultures of purified CD3+ T cells and purified CD19+ B cells pulsed with MBP, no T-cell proliferation was induced, whereas anti-CD3 stimulated T cells proliferated as expected (data not shown). We did, however, observe a low frequency of IL-10 producing CD4+ T

Discussion
While the ability of B cells to take up and present foreign antigens has been investigated intensively (for review see [40]), little is known about the capacity of human B cells to take up and present self-antigens and thereby modulate CD4+ T-cell activation. In this study we dissect the uptake and presentation of the self-antigen MBP by B cells.
We observed that 2.561.5% of normal B cells were capable of binding MBP when suspended in serum-free medium. We previously showed that a similar proportion of B cells bound a different self-antigen, human thyroglobulin, under similar conditions [12]. It is likely that the B-cell subsets in question bear polyreactive B-cell receptors reactive with a variety of self-and non-self-antigens [41].
In the presence of 30% (v/v) serum, the majority of B cells bound MBP. We show here that B-cell uptake of MBP is dependent on active complement and functional complement receptors. Thus, the binding of MBP was markedly reduced by: i) heat inactivation of serum complement; ii) EDTA-mediated chelation of calcium and magnesium, essential for complement activation; iii) inactivation of complement by SPS [37]; and iv) blockade of CR1 or CR2. Accordingly, we observed that C1q and C3 co-deposited with MBP on B cells. Taken together with the finding that NHS contains antibodies capable of binding MBP [42,43], this suggests that MBP is incorporated into immune complexes that activate complement via the classic pathway of activation, facilitating the uptake of MBP by B cells as previously shown for exogenous antigens [4][5][6][9][10][11][12]32] and thyroglobulin [12]. We did not detect any lowered binding of MBP when blocking CD32a, b, c indicating that FccRs do not participate in this process.
When examining the presentation of MBP85-99 peptides after culturing of B cells with whole MBP we found that 3.762.4% of B cells presented MBP85-99 in the absence of serum, and that approximately half of the B cells presented the peptide when the serum concentration exceeded 3% (v/v). Thus, a surprisingly high proportion of B cells became engaged in antigen-presentation when serum was present. Inhibition of complement by SPS and antibody-mediated blockade of CR2 abrogated presentation, while anti-CR1 antibody did not inhibit presentation, even though it significantly reduced binding of MBP to B cells. CR2 thus seems to be the important receptor for antigen internalisation by B cells in our system, supporting prior observations [4,44]. Using a monoclonal antibody recognising the 46-61 determinant from hen egg lysozymes presented on mouse MHC class II molecule I-A k one study has previously demonstrated presentation of peptides by bulk non-specific B cells after administration of soluble protein [45], supporting our findings. The fact that bulk B cells can engage in antigen presentation may have important implications for maintenance of tolerance by Bregs [17], as well as for B-cell-driven pathogenic processes in MS, where pathogenic T cells are believed to be activated outside the blood-brain barrier [46]. Our data suggest that B cells, irrespective of specificity, may take up and present MBP via CR2 in lymph nodes and spleen, or take up MBP outside lymphoid tissue and migrate to secondary lymphoid organs to present MBP-derived peptides to T cells. Transitional B cells may take up MBP outside the blood-brain barrier and, as described by Lee-Chang et al. [31,47], cross the barrier by virtue of a4 and b1 integrin expression. The majority of MS patients show elevated numbers of B cells in the CNS [31]; B cells have been demonstrated in a subset of cortical lesions in patients with early-stage MS [48], and some patients have ectopic B-cell follicles containing T cells, B cells and plasma cells in the cerebral meninges [49,50]. Complement-activating immune complexes  may also form intrathecally, when anti-MBP antibodies are present in cerebrospinal fluid, as is the case in children at least [42], and immunopathological evidence of complement activation has been demonstrated in plaques [51]. It is possible that higher quantities and affinities of anti-MBP antibodies in patients, as compared to healthy controls, may lead to formation of immunecomplexes with stronger pro-inflammatory potential than those involved in this study.
We were not able to identify subsets with particular pro-or antiinflammatory potential among MBP85-99-presenting B cells, neither in terms of production of IL-10 or IL-6, nor in terms of expression of CD24, CD1d, CD5 or IgM, which have been associated with Bregs [20,23,39]. Approximately half of the B cells presenting MBP85-99 expressed CD27, which has been associated with a Breg phenotype [52,53]. However, it is usually considered a memory B-cell marker [38], hence the significance of this observation is not clear. We did not investigate if the expression of CR1 and CR2 on MBP85-99-presenting B cells was similar to that of B cells not engaged in presentation of the peptide. Others have shown that CD27+ memory B cells have been shown to express higher levels of CR1, but not CR2, than CD27-naïve B cells [54]. Moreover, we observed significantly upregulated expression of CD86 on MBP85-99-presenting B cells. Increased expression of co-stimulatory molecules has usually been associated with an immune activating phenotype of APCs [55], but recently expression of B7 (CD80/CD86) on murine B cells was shown to be central to regulation of CD4+CD25 high Tregs in experimental autoimmune encephalomyelitis [18]. Accumulating data also support a role for B cells in the generation of human Tregs [56]. Concordantly, B cells pulsed with MBP induced low frequencies of IL-10-producing CD4+ T cells in half of the cocultures tested in this study, while MBP induced no TNF-alphaproducing CD4+ T cells. This is to be expected if B cells are involved in the silencing of potentially self-reactive T cells in healthy humans, as indicated by animal experiments [15,17].
In conclusion, our study demonstrates that B cells, irrespective of specificity, can become engaged in the presentation of the MSrelevant auto-antigen MBP in a complement-dependent manner. While CR1 and CR2 cooperate in the binding of MBP, engagement of CR2 is crucial for subsequent presentation of the