The macrophage migration inhibitory factor pathway in human B cells is tightly controlled and dysregulated in multiple sclerosis

Abstract In MS, B cells survive peripheral tolerance checkpoints to mediate local inflammation, but the underlying molecular mechanisms are relatively underexplored. In mice, the MIF pathway controls B‐cell development and the induction of EAE. Here, we found that MIF and MIF receptor CD74 are downregulated, while MIF receptor CXCR4 is upregulated in B cells from early onset MS patients. B cells were identified as the main immune subset in blood expressing MIF. Blocking of MIF and CD74 signaling in B cells triggered CXCR4 expression, and vice versa, with separate effects on their proinflammatory activity, proliferation, and sensitivity to Fas‐mediated apoptosis. This study reveals a new reciprocal negative regulation loop between CD74 and CXCR4 in human B cells. The disturbance of this loop during MS onset provides further insights into how pathogenic B cells survive peripheral tolerance checkpoints to mediate disease activity in MS.


Introduction
MS is a chronic autoimmune disease of the CNS, in which infiltrating proinflammatory immune cells mediate local pathology [1]. The strong effects of anti-CD20 monoclonal antibody therapy in the homing of neuroantigen-specific T cells to the CNS [6]. Which and how immune subsets are regulated by MIF to promote disease activity in MS patients remains to be determined.
In murine B cells, triggering of the cognate receptor of MIF, CD74 (invariant chain), results in enhanced proliferation and proinflammatory cytokine production via NF-κB [7,8]. Besides functioning as an MHC class II chaperone protein, CD74 also has an MHC class II-independent role in B-cell maturation [9]. Interestingly, MIF is also a noncognate ligand for chemokine receptor CXCR4 [10], which probably cooperate with CD74 to regulate B-cell development and function through MIF [11][12][13].
This study explores whether MIF, CD74, and CXCR4 expression in B cells is associated with early MS disease activity, and how the regulation and downstream effects of MIF receptors CXCR4 and CD74 affect human B-cell function. We show that MIF and CD74 are downregulated and CXCR4 is upregulated in blood B cells from early MS patients. This is dependent on a B cell-intrinsic negative regulation loop between MIF, CXCR4, and CD74, which mediates their proinflammatory activity, proliferation, and sensitivity to Fasmediated apoptosis.

The expression ratio of MIF receptors CXCR4 and CD74 on B cells is increased during rapid MS onset
To determine whether the B cell-intrinsic MIF pathway is differentially regulated in early MS, we assessed the expression levels of MIF receptors CXCR4 and CD74 on blood B cells of relapsingremitting MS (RRMS) patients and healthy controls (HC). CXCR4 was 1.4-fold increased on B cells from 15 RRMS patients compared to 15 age-and gender-matched HC (p = 0.002, Fig. 1A, B, and D), which was reproduced and validated in additional cohorts (Supporting Information Fig. 1A and B). In contrast, CD74 expression was 1.4-fold reduced on B cells in RRMS versus HC (p = 0.038, Fig. 1A, C, and E). The ratio of CXCR4 and CD74 expression levels on B cells was even further enhanced in RRMS (2.1-fold, p = 0.004; Fig. 1F; Supporting Information Fig. 1C and D), suggesting that both MIF receptors are dysregulated on a per-patient basis.
In RRMS and high-risk CIS blood, transitional (IgM + CD27 − CD38 hi CD24 hi ) as well as naive mature (IgM + CD27 − CD38 −/dim ) B-cell subsets displayed the highest CXCR4/CD74 expression ratios as compared to classswitched (CD27 + /CD27 − IgG + and IgA + ) and nonclass-switched (IgM + CD27 + ) memory subsets ( Fig. 3; Supporting Information Fig. 2), implying that the CXCR4 hi CD74 lo phenotype of B cells in early MS reflects a more immature state. These data demonstrate that MIF receptors CXCR4 and CD74 are inversely expressed on B cells, which is dysregulated during early disease onset in MS.

MIF is predominantly expressed by B cells in healthy blood and downregulated in early MS patients
MIF levels in both serum and plasma were not different between CIS patients and healthy controls ( Fig. 4A and Supporting Information Fig. 3A) or high-risk and low-risk CIS subgroups ( Fig. 4B and Supporting Information Fig. 3B), and did not correlate with CXCR4 and CD74 expression levels on B cells from the same individuals (Supporting Information Fig. 3C and D).
However, among blood immune subsets, MIF mRNA was predominantly expressed by B cells compared to paired T cells, monocytes and dendritic cells (Fig. 4C), but downregulated in B cells from RRMS patients versus HC (p < 0.01) and even further in B cells of CIS patients (CIS vs. HC, p < 0.0001; CIS vs. RRMS, p < 0.01; Fig. 4D). We found no differences in MIF mRNA levels between B cells from low-risk and high-risk CIS subgroups (Fig.  4E). Low MIF mRNA levels corresponded to high CXCR4/CD74 surface expression ratios in B cells of patients and controls (p = 0.005; Fig. 4F). This point to the existence of a disturbed regulation loop between MIF and MIF receptors in B cells of early MS patients.

MIF, CD74, and CXCR4 are part of a reciprocal negative regulation loop in human B cells
In vitro, B cells showed enhanced MIF expression and secretion after activation with anti-IgM, which was comparable between patients and controls ( Fig. 5A and B). To determine potential crosstalk between MIF, CXCR4, and CD74, we used the specific MIF inhibitor ISO-1 [16] to block MIF-mediated signaling in in vitro-activated B cells. ISO-1 treatment of these B cells resulted in a CXCR4 upregulation and CD74 downregulation (Fig. 5C), reflecting the inverse correlation between MIF and CXCR4/CD74 expression levels in B cells ex vivo (Fig. 4F). In parallel to this, in the human B-cell line Raji, which abundantly expresses CD74 [17], the percentage of CD74 hi cells decreased after MIF knockdown using three distinct shRNA constructs ( Fig. 5D and E). This indicates that CXCR4 and CD74 surface expression is inversely regulated by endogenous MIF in B cells. After treatment with anti-CD74 antibody (LN2), CXCR4 surface expression was increased, whereas MIF expression was reduced in in vitro-activated B cells ( Fig. 5F and H). Vice versa, activated B cells treated with CXCR4 antagonist  AMD3100 [18] showed increased CD74 and MIF levels ( Fig. 5G and H). These data demonstrate that MIF, CD74, and CXCR4 expression in human B cells is tightly and mutually controlled.

CXCR4/CD74 controls the inflammatory, proliferative, and Fas-mediated apoptotic potential of B cells
To determine how the reciprocal negative regulation of CD74 and CXCR4 is associated with the function of B cells, we compared the proinflammatory, proliferative, and survival capacity of in vitro-activated B cells before and after blocking of these MIF receptors. Treatment of these B cells with anti-CD74 antibody LN2 (24 h) suppressed the induction of proinflammatory genes NFKB1, IL6, and TNF (Fig. 6A). This was not found in in vitro-activated B cells treated with CXCR4 antagonist AMD3100 (Fig. 6A). These differences were verified on protein level (Supporting Information Fig. 4). Also B-cell proliferation was inhibited after treatment for 3 days with LN2 antibody and not with AMD3100, as determined by CFSE labeling (Fig. 6B and C). Finally, surface expression of the death receptor Fas (CD95) was triggered in AMD3100-and not in LN2-treated B cells after 3 days of activation (Fig. 6D).  These results imply that in humans, CD74 primarily boosts the proinflammatory and proliferative capacity of B cells, while CXCR4 makes B cells less sensitive for Fas-mediated apoptosis [19].

Discussion
The aim of this study was to elucidate the impact of the B-cell intrinsic MIF pathway on early disease onset in MS patients. We demonstrate that decreased CD74 and increased CXCR4 expression on B cells in blood are associated with early MS diagnosis. This was shown for CIS patients who will rapidly develop MS as well as for clinically definite MS patients. In vitro experiments supported the inverse regulation of MIF/CD74 and CXCR4 expression in B cells ex vivo, which differentially controlled their proinflammatory capacity, proliferation, and sensitivity to Fasmediated apoptosis (Supporting Information Fig. 5). The observed CXCR4 hi CD74 lo B-cell phenotype in early MS blood points to the presence of more immature B-cell populations with senescent features that survived peripheral tolerance checkpoints in MS [3].
There are several lines of evidence implying that CD74 downregulation and CXCR4 upregulation disrupt the selection of imma-ture B cells. During B-cell development, "new emigrant" transitional subsets in blood are negatively selected in secondary lymphoid organs for autoreactive clones before developing into naive mature subsets. B cells in CD74-deficient mice revealed an arrest at the transitional stage [20] and a reduced lifespan [21]. Consistently, decreased CD74 expression impaired B-cell maturation in patients with X-linked lymphoproliferative disease [22]. Both cell autonomous and nonautonomous roles of CD74 could explain these defects in B-cell development. As a cell surface receptor, CD74 triggers B-cell proliferation in mice [8] and humans (current study), supporting the lowered proliferative capacity of naive (CD74 lo ) versus memory (CD74 hi ) populations [23]. The reduced expression of CD74 on B cells in early MS blood thus might reflect a functional state of anergy, contributing to the persistence of pathogenic immature B cells in the periphery [24]. This is underlined by the downmodulation of proinflammatory cytokines TNF-α and IL-6, as well as MIF after blockade of CD74 on B cells. Alternatively, defective processing of CD74 results in the accumulation of an N-terminal fragment, which interferes with B-cell receptor signaling to suppress B-cell maturation in an MHC class IIindependent manner [9,25,26]. In addition to these cell-intrinsic effects, a loss of CD74 can also influence T helper cell-mediated  (F) Correlation between MIF mRNA levels and CXCR4/CD74 surface expression ratios in B cells from patients and healthy controls (n = 30). B cells were sorted, mRNA was measured in duplicates, and surface expression was analyzed in six independent experiments with 2 HC and 2-4 patients per experiment. r = Spearman's correlation; *p < 0.05, **p < 0.01, ****p < 0.0001. selection of naive mature B cells via altered MHC class II antigen presentation [27].
In contrast to CD74, CXCR4 was found to be the most abundant on naive B cells, which controls their development in germinal centers. CXCR4 hi B cells are localized in the dark zone to undergo somatic hypermutation, whereas antigen-and T helper cell-based selection of CXCR4 lo B cells occurs in the light zone [28,29]. Our data show that blocking of CXCR4 signaling in B cells increases CD95 (Fas) expression, which is essential for the elimination of autoreactive clones by T helper cells [30]. Hence, it may be speculated that overexpression of CXCR4 on B cells as observed in early MS results in the escape of naive populations from T helper cellbased selection in the light zone, via reduced Fas expression and enhanced migration to the dark zone. Central tolerance checkpoints were not defective in MS patients [3], making it unlikely that the abundance of CXCR4 affects precursor B-cell selection in the bone marrow [31]. In vivo studies need to be performed in the future to confirm these roles of CD74 and CXCR4 in peripheral B-cell tolerance in MS.
To our knowledge, the coregulation of MIF, CD74, and CXCR4 in human B cells and MS patients has never been studied before. Although abundantly expressed by B cells compared to other immune populations, MIF is downregulated in blood B cells from early MS patients. This downregulation links to the survival of autoreactive naive B cells, as seen in atherosclerotic mice [13] and MS [3], and coincides with decreased CD74 and increased CXCR4 surface expression, as part of a tightly controlled regulation loop in B cells. The reciprocal expression of CD74 and CXCR4 was supported by the increased migration capacity of B cells toward CXCL12 after inhibition of CD74 [12,32]. Since MIF has a higher affinity for CD74 than for CXCR4 [7,10], a possible underlying mechanism of this reciprocal expression is that  MIF-mediated endocytosis of CD74 results in interaction with the adaptor molecule β-arrestin, thereby preventing binding to and internalization of CXCR4 [33][34][35] (Supporting Information Fig. 5).
Extracellular MIF levels were not different in the blood of early MS patients. However, previous studies showed increased levels of MIF in MS CSF [36] as well as in MS lesions and not normal-appearing white matter [37]. This suggests that local MIF production predominantly attracts CXCR4 hi B cells from the blood [10,12] to mediate early MS disease activity. Another CXCR4 ligand, CXCL12, was also found to be abundant in MS CSF, but did not correlate to local B-cell infiltration and activation [38]. After recruitment to the CNS, CXCR4 hi CD74 lo B cells will probably be activated, resulting in increased CD74 expression and MIF production. CD74 triggering by autocrine MIF may then enhance their Figure 6. CD74 and CXCR4 on human B cells differentially control proinflammatory gene expression, proliferation, and sensitivity to Fas-mediated apoptosis. B cells from healthy blood were in vitro-activated with a-IgM and subsequently treated with anti-CD74 antibody (LN2) or AMD3100. Data were compared to their respective controls for relative NF-κB, IL-6, and TNF-α mRNA expression after 24 h using qPCR (A, n = 4-6, measured in duplicates), as well as CFSE-based proliferation (B and C, n = 5) and Fas (CD95) surface levels (D, n = 5) after 3 days using FACS. Stimulations have been done in three individual experiments with two controls per group per experiment. All used controls were set at 1 (dotted line). Data are shown as mean ± SEM. Paired t-tests were performed to compare groups. *p < 0.05, **p < 0.01. ability to proliferate, as shown for tumor cells [39], and produce proinflammatory cytokines to mediate CNS pathology in MS.
This study shows that MIF and MIF receptors CD74 and CXCR4 are coordinately expressed in B cells to control their inflammatory, proliferative, and apoptotic potential in humans. The dysregulation of this B cell-intrinsic loop in early MS pleads for future studies on the processing and cooperation of CD74 and CXCR4 during autoreactive B-cell development. Also more insights into the effects of autocrine and T helper cell-derived MIF on their development will lead to better understanding of the role of B cells as central players in MS and other autoimmune diseases.

Patients
Patient characteristics are summarized in Supporting Information Table 1. All clinically isolated syndrome (CIS) and relapsingremitting MS (RRMS) patients as well as healthy controls (HC) were included in MS Center ErasMS at Erasmus MC (Rotterdam, The Netherlands). CIS was defined as a first clinical attack of demyelination in the CNS [14]. Clinically definite MS (CDMS) was diagnosed when a patient experienced two attacks with clinical evidence of two separate lesions according to the Poser criteria [40]. CIS patients were sampled within 4 months after their first attack. From our prospective cohort, we selected CIS patients who did not develop CDMS for at least 5 years of follow-up (lowrisk CIS) and CIS patients who were diagnosed with CDMS within 1 year after CIS diagnosis (high-risk CIS). Fatigue severity was assessed using Krupp's fatigue severity scale [41]. RRMS patients were diagnosed according to the McDonalds criteria [42] and were age-and gender-matched with healthy control subjects. For functional studies, buffy coats (Sanquin, Amsterdam, The Netherlands) were obtained from healthy volunteers. All patients and controls gave written informed consent and study protocols were approved by the medical ethics committee of the Erasmus MC.

Peripheral blood sampling
PBMCs and plasma were isolated from whole blood with the use of CPT TM heparin tubes, while serum was isolated using coagulation tubes (both BD Biosciences, San Jose, CA). Samples were processed according to the manufacturer's instructions. PBMCs were C 2018 The Authors. European Journal of Immunology published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.
www.eji-journal.eu stored in liquid nitrogen; plasma and serum were stored in -80°C until analysis.

Human MIF ELISA
ELISA was used to measure human MIF levels in serum and plasma using the four-span approach, as previously described [43]. In brief, anti-human MIF polyclonal antibodies raised in chicken and rabbit were used as capture and trapping antibodies. Microtiter plates were coated with a duck anti-chicken antibody. A HRPlabeled goat anti-rabbit antibody was used for the detection. Distinct concentrations of rhMIF (R&D Systems, Minneapolis, MN) were used to generate a standard curve. The analytic sensitivity of the human MIF ELISA was 39 pg/mL and the coefficients of variation were 6% for intrarun and 12% for interrun.

Flow cytometry and cell sorting
In-depth flow cytometric analysis of B cells were performed using anti-human monoclonal antibodies against CD3, CD19, CD24, CD27, CD38, CD69, CD74, CD95, CXCR4, HLA-DR, IgA, IgD, IgG, and IgM. Details of these antibodies are indicated in Supporting Information Table 2. All measurements were performed on an LSRFortessa TM flow cytometer and data were analyzed using FACSDiva 8.1 software (both BD Biosciences). Guidelines for the use of flow cytometry in immunological studies have been followed [44].

Intracellular cytokine staining
Cells were stimulated for 5 h using phorbol 12-myristate 13acetate (PMA; 20 ng/mL) and ionomycin (500 ng/mL, both Sigma-Aldrich), in the presence of BD GolgiStop TM . Stimulated cells were stained with BD Horizon TM Fixable Viability Stain 700, fixed and permeabilized using BD Cytofix/Cytoperm TM according to the provided protocol and stained for IL-6 and TNF-α (Supporting Information Table 2, BD Biosciences).

RNA isolation and quantitative PCR
CD19 + B cells, CD3 + T cells, CD14 + monocytes, and CD56 − HLA-DR + dendritic cells were sorted using a high-speed cell sorter (FACSAria III TM ; BD Biosciences), resulting in a purity of more than 95%. Subsequently, mRNA was isolated using GenElute TM Mammalian RNA Kit (Sigma-Aldrich, St. Louis, MO) and reversely transcribed into cDNA following a standard laboratory protocol with the use of SuperScript II R Reverse Transcriptase (Invitrogen, Paisley, UK). Primers and probes were selected by using the Universal Probe Library Assay Design Centre (Roche Applied Science, Penzberg, Germany). To determine target gene mRNA expression levels, qPCR was performed using an Applied Biosystems 7900 Sequence Detector, which was programmed for the initial step of 2 min at 50°C and 10 min 95°C, followed by 40 thermal cycles of 15s at 95°C and 1 min at 60°C. For the calculation of relative mRNA levels, CT values per gene were related to standard curves, which were generated for each gene of interest. The 18S levels were measured as a control to normalize for RNA input. Primer sequences are listed in Supporting Information Table 3.

In vitro activation and modulation of B cells
CD19 + B cells of healthy donors were isolated untouched by depleting all other cell types from PBMCs using the B-cell isolation kit II and MACS (Miltenyi Biotec, Bergisch Gladbach, Germany). CD19 + B cells of patients and healthy controls were sorted using FACS and were cultured in RPMI supplemented with 10% fetal calf serum and 1% penicillin/streptavidin for 24 hours or 3 days with various stimuli; anti-IgM (F(ab') 2, 10 μg/mL, Jackson ImmunoResearch Inc., West Grove, PA) with or without MIF inhibitor ISO-1 (100 μM, R&D Systems), a neutralizing anti-human CD74 antibody (LN2; 10 μg/mL, BD Biosciences) or AMD3100 (10 μg/mL, Sigma-Aldrich). Proliferation rates were addressed by labeling B cells with CFSE (eBioscience, San Diego, CA) before in vitro activation. For MIF knockdown, the human B-cell line Raji was transfected with MIF shRNA-containing pLKO.1 constructs (MISSION R shRNA Library, Sigma-Aldrich) using Nucleofector Kit V from Lonza (Basel, Switzerland). Three different MIF shRNA constructs were used: #1 GACAGGGTCTACATCAACTAT, #2 CTACATCAACTATTACGACAT, and #3 CCTGCACAGCATCG-GCAAGAT. MIF mRNA and CD74 surface expression was analyzed and compared to scrambled shRNA controls at day 3.

Statistical analyses
Statistical analyses were performed using Graphpad Prism 7 (GraphPad Software Inc., San Diego, CA mean ± SEM. Datasets were tested for normal distribution. Twotailed t tests or ANOVA were used to compare groups. Correlations between two parameters were tested by using Pearson's or Spearman correlation coefficients. p-values < 0.05 were considered as statistically different.