Massive Restrain of Cytotoxic B cells in the Peripheral Blood During Fingolimod and Natalizumab Treatments in Multiple Sclerosis Patients

Recently, the success of anti-CD20 monoclonal antibody therapy brought a new light over the role of B cells in multiple sclerosis (MS) pathogenesis. Due to the expression pattern of CD20 during B cells ontogeny, this role seems to be extended beyond the antibodies' production and secretion. Therefore, here we investigated whether not only classical cytotoxic CD8 + T lymphocytes but also non-classical cytotoxic B cells may occur in the peripheral blood from relapsing-remitting MS (RRMS) patients. Methods 104 RRMS patients during different treatment and 58 healthy donors were studied. CD19, GzmB, Runx3 and CD49d expression was assessed by ow cytometry analyses.


Introduction
Multiple sclerosis (MS) is an immune-mediated and demyelinating disease of the central nervous system (CNS). The in ammatory response is characterized by the perivascular leukocyte in ltration, mainly in the white matter. However, recent magnetic resonance imaging (MRI) studies revealed a much broader CNS damage, including gray matter atrophy and normal-appearing white matter (NAWM) abnormalities [1][2][3].
Moreover, leptomeningeal contrast enhancement is resent in around 19% of patients during the initial phase of the disease [4]. Noteworthy, brain atrophy is not correlated with the number of relapses, activity lesions, or with in ammatory in ltration in the gray matter [2,3]. Cortical brain lesions seem to be regionally associated with meningeal in ammation [3]. Meningeal in ammation leads to local leukocytes aggregates, which are formed mainly by B cells, T cells, macrophages, follicular dendritic cells, and plasmacytoid dendritic cells [4,5].
Although the pathophysiology of MS is very complex, it is well accepted that the disease is mediated mainly by T lymphocytes [6]. However, recently, it has been shown a successful outcome by targeting B cell in patients with MS using anti-CD20 monoclonal antibodies (rituximab, ocrelizumab, or ofatumumab) [7]. Nevertheless, the CD20 molecule is not expressed on pro-B cells or differentiated plasma cells. Thus, the bene cial effect of anti-CD20 treatment appears to be extended beyond the autoantibody production and release. Besides the central role in the humoral response, B cells can exert antigen presentation, helper, and regulatory functions as well as direct effector activity [8]. Indeed, secretory products of B cells are cytotoxic to oligodendrocytes and neurons in vitro [9,10].
B cells may differentiate in granzyme B (GzmB)-producing cells upon insu cient T cell help [11,12]. Recently, we have been demonstrated an enhancement of the cytotoxic activity of B cells during refractory neuromyelitis optica spectrum disorder [13]. Of note, GzmB inhibition induces neuroprotection in MS experimental model, although it did not alter the numbers of CD4 + or CD8 + T lymphocytes in ltrated in the CNS [14].   to the membrane and for permeabilizing the cell surface. To investigate the cytotoxic pro le held intracellular staining with anti-human GzmB antibody -AlexaFluor® 700 (BD Biosciences®) and antihuman Runx3 -eFluor660 (eBioscience™) in all samples. The acquisition was performed in FACSVerse® ow cytometer (BD Biosciences®), and the analysis used the FlowJo® software.

B cells isolation and in vitro stimulation
After PBMCs obtaining, B cells were isolated using the EasySep® Human B Cell Enrichment Kit with EasySep® magnet. After the isolation, 2 × 10 4 B cells were stimulated during 16 h in culture, with CPG-ODN (2.5µL/mL) and human recombinant IL-21 (50 ng/mL), as previously described [16]. After the culture period, the supernatants were stored at -80 °C for the investigation of GzmB production using a Cytometric Bead Array (CBA).

Cytometric Bead Array (CBA)
According to the manufacturer's protocol, 50 microliters of B cells stimulated supernatants, and solutions for calibration curve construction were incubated with beads containing monoclonal antibodies to GzmB. After incubation for 2 hours, it was added revealing antibody conjugated to the uorochrome PE. The acquisition was made in ow cytometer FACSCanto (BD Bioscience®), and the analysis used the FCAP Array software (BD Bioscience®).

Statistical Analyses
The statistical signi cance of the results was determined using a nonparametric analysis of variance

GzmB expression in CD8 + T and CD19 + B lymphocytes by untreated RRMS patients
Flow cytometry analysis of PBMCs ( Fig. 1a-b) showed that CD8 + GzmB + T lymphocytes from untreated RRMS patients are not different in terms of percentage when compared to healthy donors (p = 0.0567; CI 95%: -0.4-20.9) (Fig. 1c). On the other hand, after we excluded the outlier values (open circles), the RRMS group presented a signi cantly higher percentage of CD19 + GzmB + cells in comparison with the healthy individual (p = 0.0198; CI 95%: -0.007-2.28) (Fig. 1d). In order to evaluate the cytotoxic activity of those cells, we sorted out CD19 + cells from RRMS patients and healthy controls and analyzed the releasing of GzmB after ODN-CPG and IL-21 stimulation. Puri ed CD19 + cells from RRMS patients presented a signi cantly higher release of GzmB in comparison with CD19 + B cells from healthy individuals (p = 0.0346), even though some of those patients were under treatment (Fig. 1e).
GzmB expression in CD8 + T and CD19 + B lymphocytes by treated RRMS patients Next, we explored the effect of treatments (GA, IFN, FTY, and NTZ) over those cytotoxic lymphocytes ( Fig. 2a and 2d). Flow cytometry analysis did not reveal differences in the percentage of total circulating CD8 + or B lymphocytes between all groups ( Fig. 2b and 2e). However, the expression of GzmB was signi cantly higher in CD8 + T lymphocytes from patients treated with NTZ concerning the other treatments (NTZ vs GA: p < 0.01, NTZ vs IFN: p < 0.001, NTZ vs FTY: p < 0.01), untreated MS patients (p < 0.001), or healthy donors (p < 0.001) ( Fig. 2a and 2c). In parallel, the expression of GzmB was signi cantly higher in B cells from patients treated with NTZ concerning the rst line immunomodulatory therapies (NTZ vs GA: p < 0.001, NTZ vs IFN: p < 0.001), untreated patients (p < 0.001), or healthy donors (p < 0.001) (Fig. 2f). Surprisingly, the expression of GzmB was also signi cantly higher in B cells from patients treated with FTY in relation to GA (p < 0.05), untreated patients (p < 0.05), or healthy donors (p < 0.001) (Fig. 2f).

Expression of Runx3 and CD49d in cytotoxic B lymphocytes
Classically, the expression of Runx3 is a pivotal mechanism of the cytotoxic program in CD8 + T lymphocytes [15]. It has been shown that the ectopic expression of Runx3 is directly related to the cytotoxic activity of non-classic cytotoxic cells [16]. Consistently, there is a strong positive correlation (R 2 = 0.8430; p < 0.0001) between the expression of Runx3 and GzmB, which is independent of the treatment ( Fig. 3a and 3b). Strikingly, considering only Runx3 + B lymphocytes, NTZ-treated RRMS patients present a signi cantly higher expression of GzmB in comparison with other treatments (Fig. 3c). Therefore, it seems that NTZ treatment tackles speci cally the cytotoxic B cells. In order to evaluate this hypothesis, we analyzed the expression of GzmB in CD19 + CD49d − or CD19 + CD49d + lymphocytes (Fig. 3d). As we hypothesize, the expression of GzmB is signi cantly higher in CD49d-expressing B lymphocytes ( Fig. 3e  and 3f).

Discussion
Here, we demonstrated that B cells from patients with RRMS present cytotoxic behavior. Moreover, NTZtreatment seems to restrain these cells in the peripheral blood effectively. These results may present a novel pathophysiological mechanism for the disease as well as for its treatment.
Until now, a vast collection of studies has consistently demonstrated the enhancement of cytotoxic activity during MS. Despite a possible role of GzmB in regulatory pathways [17], CD8 + T cells are a classical cytotoxic subset that can induce direct axonal and/or neuron injury through the MHC class Imediated manner and GzmB releasing [18]. Relapses during RRMS are associated with increased CD8 + T lymphocytes in CSF [19]. Noteworthy, cytotoxic behavior seems not to be restricted to these classical CD8 + T lymphocytes. Recently, Larochelle and colleagues described the migratory capacity of CD8 + GzmB + and CD4 + GzmB + T lymphocytes into the CNS-parenchyma from fatal MS relapse after NTZ discontinuation [20]. Moreover, our group also has previously identi ed the cytotoxic activity of B cells in neuroin ammatory diseases [13].
Our results demonstrated a slight but signi cant increase percentage of GzmB-producing B cells in untreated patients with RRMS in comparison with healthy individuals. Those cytotoxic B cells are capable of producing GzmB as well as to release the protease after IL-21 and CpG stimulation. Of note, untreated RRMS patients are signi cantly younger and with shorter time of disease diagnosis than treated RRMS patients. Many authors have proposed that B cells would take a later role in MS pathophysiology, especially in the formation of ectopically leukocytes aggregates [21]. Therefore, it is plausible that during the early stage of the disease, the cytotoxic activity of B cells is less prominent. Nevertheless, it is di cult to test this possibility since the patients are treated at a very early stage upon diagnosis.
Regarding treatment, our data showed that FTY-treated RRMS patients present a signi cantly higher percentage of CD19 + GzmB + in comparison with healthy, untreated or GA-treated RRMS patients. The effect of FTY over the B cell population seems to be a lesser extent than T cells. On the one hand, the treatment with FTY inhibits the pro-in ammatory B cell function favoring the regulatory phenotype [22,23]. On the other hand, FTY-treatment does affect the T-independent response [24]. In this context, human cytotoxic B cells differentiate upon incomplete T-cell help [11,25]. Thus, it is possible that FTY-treatment does not act over those cells, which would explain the increased proportion of granzyme B-expressing cells in the peripheral blood.
Besides, NTZ-treated RRMS patients present a consistent and signi cant retrains of CD8 + T lymphocytes and cytotoxic B cells in the peripheral blood. The restrain of activated T cells is the rationale of the NTZ treatment [26]. In fact, during fatal rebound following NTZ withdrawal CNS-in ltrating lymphocytes, mainly CD8 + T cells, express high amounts of GzmB. Curiously, monocytes/macrophages, and B cells were enriched in the CNS parenchyma compared to the CSF [20]. Here, we demonstrated that GzmBexpressing B cells are preferentially CD49d positive, which would explain the consistent restrain of cytotoxic B cells by NTZ treatment.
In contrast, NTZ promotes activation and pro-in ammatory enhancement of B cells in vitro [22]. Moreover, those B cells with an in ammatory pro le increase in the PB of NTZ-treated RRMS patients progressively [22]. Nevertheless, the restrain of those pro-in ammatory B cells in the PB seem to be crucial for the effectiveness of the treatment.
Runx3 is a master regulator of the cytotoxic program in CD8 + T lymphocytes and natural killer cells [27,28]. The ectopic expression of Runx3 has been extensively associated with non-classical cytotoxic cells [13,16,29]. Our data demonstrated a correlation between the expression of GzmB and Runx3 by B cell, which was independent of the treatment. Mechanistically, the ectopic expression of Runx3 by RRMS patient's B cells might explain the enhancement of cytotoxic activity. Of note, the expression of Runx3 is required for the proliferation of immortalized B cell during the Epstein-Barr virus (EBV) infection, which has been extensively associated with MS [30]. Noteworthy, the ectopic expression of Runx3 by dendritic cells regulates the activity of the CD49d gene promoter [31].
Collectively, our ndings support that beyond antigen presentation to T cells, releasing of lytic factors, secretion of anti and pro-in ammatory cytokines, cytotoxic behavior might emerge in the context of antibody-independent functions developed by B cells during MS. In this context, anti-CD20 monoclonal antibodies have been reported as a viable option for escalation, aiming to prevent relapses after NTZ therapy in MS patients [32]. Anti-CD20 monoclonal antibodies main deplete naïve and memory B cells, preserving antibody-secreting B cells, which is con rmed by the sustained levels of immunoglobulins in the CSF from MS treated patients [33]. The future investigation might reveal whether memory B cells responsible for antigen presentation, secretion of pro-in ammatory cytokines, and even, GzmB-expressing subsets are targeted by anti-CD20 depleting therapies during MS.

Limitations
We are aware that the size of the cohort and the transversal nature of our study did not allow us to understand the clinical relevance of our nds better. Moreover, although we were able to establish a strong correlation between Runx3 and GzmB expression by B cells, there are not enough tools to imply causality to the phenomenon. In vitro generation of cytotoxic B cells will be necessary to clarify the role of Runx3 expression in the ectopic cytotoxic activity of these cells.

Conclusions
Further comprehension of the cytotoxic activity in MS physiopathology might provide an accessible and powerful tool for monitoring progression and therapy effectiveness during disease. Moreover, the cytotoxic mechanism might be a valuable target to develop new therapeutic strategies.