Regulatory T Cells Increase After rh-MOG Stimulation in Non-Relapsing but Decrease in Relapsing MOG Antibody-Associated Disease at Onset in Children

Background Myelin oligodendrocytes glycoprotein (MOG) antibody-associated disease (MOGAD) represent 25% of pediatric acquired demyelinating syndrome (ADS); 40% of them may relapse, mimicking multiple sclerosis (MS), a recurrent and neurodegenerative ADS, which is MOG-Abs negative. Aims To identify MOG antigenic immunological response differences between MOGAD, MS and control patients, and between relapsing versus non-relapsing subgroups of MOGAD. Methods Three groups of patients were selected: MOGAD (n=12 among which 5 relapsing (MOGR) and 7 non-relapsing (MOGNR)), MS (n=10) and control patients (n=7). Peripheral blood mononuclear cells (PBMC) collected at the time of the first demyelinating event were cultured for 48 h with recombinant human (rh)-MOG protein (10 μg/ml) for a specific stimulation or without stimulation as a negative control. The T cells immunophenotypes were analyzed by flow cytometry. CD4+ T cells, T helper (Th) cells including Th1, Th2, and Th17 were analyzed by intracellular staining of cytokines. Regulatory T cells (Tregs, Foxp3+), CD45RA-Foxp3+ Tregs and subpopulation naive Tregs (CD45RA+Foxp3int), effector Tregs (CD45RA-Foxp3high) and non-suppressive Tregs (CD45RA-Foxp3int) proportions were determined. Results The mean onset age of each group, ranging from 9.9 to 13.8, and sex ratio, were similar between MOGR, MOGNR, MS and control patients as analyzed by one-way ANOVA and Chi-square test. When comparing unstimulated to rh-MOG stimulated T cells, a significant increase in the proportion of Th2 and Th17 cells was observed in MOGAD. Increase of Th17 cells was significant in MOGNR (means: 0.63 ± 0.15 vs. 1.36 ± 0.43; Wilcoxon-test p = 0.03) but not in MOGR. CD4+ Tregs were significantly increased in MOGNR (means: 3.51 ± 0.7 vs. 4.59 ± 1.33; Wilcoxon-test p = 0.046) while they decreased in MOGR. CD45RA-Foxp3+ Tregs were significantly decreased in MOGR (means: 2.37 ± 0.23 vs. 1.99 ± 0.17; paired t-test p = 0.021), but not in MOGNR. MOGR showed the highest ratio of effector Tregs/non suppressive-Tregs, which was significantly higher than in MOGNR. Conclusions Our findings suggest that CD4+ Th2 and Th17 cells are involved in the pathophysiology of MOGAD in children. The opposite response of Tregs to rh-MOG in MOGNR, where CD4+ Tregs increased, and in MOGR, where CD45RA-Foxp3+ Tregs decreased, suggests a probable loss of tolerance toward MOG autoantigen in MOGR which may explain relapses in this recurrent pediatric autoimmune disease.


INTRODUCTION
Pediatric acquired demyelinating syndromes (ADS) are rare immune-mediated acute demyelinating disorders of the central nervous system (CNS) with an incidence of 0.6 to 1.6 for 100,000 children per year in western countries (1)(2)(3). Myelin oligodendrocytes glycoprotein (MOG) antibodies (Abs) are found in about 25% of pediatric ADS (4) now referred to as MOG antibody-associated disease (MOGAD). MOG protein represents only 0.05% of myelin proteins and is expressed exclusively on the outer surface of the myelin sheath and the plasma membrane of oligodendrocytes. Its cell surface location makes it accessible to immune reactions (5) becoming a target of autoimmune responses that cause inflammation and CNS demyelination (6,7). The course of MOGAD can be either non-relapsing (MOGNR) or relapsing (MOGR). Unlike MOGAD, multiple sclerosis (MS) patients do not have MOG-Abs (8)(9)(10). MS is an ADS characterized by the recurrence of demyelination episodes resulting in subsequent neurological damage. Both MOGAD and MS have T and B lymphocytes infiltration in the brain, but in MS, CD8 + T cell and B cell infiltration is higher (11) than in MOGAD where CD4 + T cell infiltration is predominant (12). There are evidence that autoreactive CD4 + T-cells are involved in both MOGAD and MS pathogenesis, but further research is required to understand their role in the disease onset and evolution (13,14).
In experimental autoimmune encephalomyelitis (EAE) in rodents, both cell-mediated and humoral immune responses are involved in brain inflammation and demyelination (15,16), sustained by a T/B cell cooperation as studied using transgenic mice with MOG-specific T and B cell receptors (17)(18)(19)(20)(21). In vitro experiments in humans have shown that patient sera containing MOG-Abs, activate the complement pathway (22)(23)(24), induce natural killer cells and antibody-dependent cell-mediated cytotoxicity (6), and can result in the disruption of the oligodendrocyte cytoskeleton (25,26).
A non-human primate model of EAE in cynomolgus macaques was developed by sensitization with recombinant human (rh)-MOG, emulsified in incomplete Freund's adjuvant (IFA), which induced a disease similar to MOGAD in children (27). Interestingly, in this model, it was recently shown that the intradermal routing of MOG into resident dendritic cell asialoglycoprotein receptor (DC-ASGPR) + cells using recombinant antibody DC-ASGPR fused to MOG prevented the breach of immune tolerance against MOG after rh-MOG/ IFA sensitization. Phenotyping of blood lymphocytes indicated that only the control animals had an increase activation of CD4 + T cells in the days preceding the onset of EAE. In contrast, animals treated with anti-DC-ASGPR-MOG had an increase in MOG-specific T regs upon rh-MOG/IFA re-administration (28).
These results prompted us to evaluate the response of CD4 + T cells to rh-MOG stimulation in vitro from children with different forms of pediatric ADS. We compared the CD4 + T cells immunological phenotypes of MOGAD patients to MS and control patients with non-inflammatory neurological diseases, and assessed the cells functional responses after stimulation with rh-MOG protein in vitro. We focused our analysis on the three main T helper (Th) cells corresponding to three lineages of CD4 + lymphocytes triggered upon antigenic activation. These CD4 + T cells are referred to as Th type-1 (Th1), Th type-2 (Th2) or Th type-17 (Th17) cells, according to their phenotype. Th1 cells produce interferon-g (IFN-g) and tumor necrosis factor-alpha (TNF-a) and are effective against intracellular bacteria and viruses, but are also involved in autoimmune diseases. Th2 cells secrete interleukin-4 (IL-4), -5, -10 and -13, which upregulate antibody production through B cells activation. Th17 cells secrete IL-17 and TNF-a and are involved in tissue inflammation, activation of neutrophils and in autoimmunity. We then studied regulatory T (T reg ) cells that secrete IL-10 and transforming growth factor-beta (TGF-b), which modulate Th cell activity and suppress some of their functions, inducing tolerance to antigens. We then subdivided the MOGAD group into non-relapsing MOGAD (MOGNR) and relapsing MOGAD (MOGR) to further evaluate relations between relapse and immunological response to rh-MOG.

Patients and Controls
Twenty-two children ≤ 18 years old, from the French cohort KIDBIOSEP, followed for a first demyelinating episode in the national reference center for rare inflammatory brain and spinal diseases at Bicetre Hospital, from January 2011 to May 2018, were included. ADS was defined as an acute neurological deficit lasting more than 24 h in the CNS, affecting the optic nerve, brain, cerebellum, brainstem and/or spinal cord associated with T2 lesions on magnetic resonance imaging (MRI). Relapse was defined as a new episode of CNS demyelination at least 1 month after the first episode or 3 months after the first episode if the first attack is an acute demyelinating encephalomyelitis (ADEM), and lasting for at least 24 h in the absence of fever or infection. The MS diagnosis was made according to the 2013 IPMSSG criteria and the revised 2010 MacDonald criteria. We classified our patients into 3 groups: MOGAD patients, including nonrelapsing (MOGNR) and relapsing (MOGR) subgroups based on the progression of the disease after the blood sampling, ADS without MOG-Ab corresponding to MS patients and control patients (CTRL). Seven control patients were included from pediatric neurology department of Bicetre Hospital for noninflammatory neurological diseases, such as intracranial hypertension (n=1), psychosomatic syndromes (n=2), genetical peripheral neuropathies (n=2), psychomotor developmental delays (n=1) or stroke (n=1), for which blood samples were performed for diagnosis. Demographic data of included children are presented in Table 1.

Ethics
This study complied fully with French national and local ethics committee guidelines. The national cohort of first demyelinating episode "Kidbiosep 2004" (No. 910506) was authorized by the Commission Nationale de l'Informatique et des Liberteś and the Comitéde Protection des Personnes of Paris-Saclay University. An informed consent form was signed by parents of each included child.

Blood Samples
Blood samples were collected on heparinized tubes from all patients at their first demyelinating event, within the first 3 months and before starting immunosuppressive or immunoregulatory therapy. For all patients, PBMCs were isolated by Ficoll density gradient centrifugation. Briefly, heparinized blood was centrifuged at 700 × g for 15 min and the top layer containing plasma was removed, transferred in cryovials and stored at -20°C. The remaining blood was diluted with an equal volume of isosmotic 0.9% wt/vol NaCl solution layered over 15 ml of the Ficoll-Paque PLUS (GE Healthcare). Gradients were centrifuged at 700 × g for 30 min at room temperature. The PBMC interface was removed by pipetting and washed with 0.9% wt/vol NaCl solution by centrifugation at 700 × g for 15 min. Non-viable cells were identified by staining with trypan blue and cell viability was calculated using the total cell count and the count of non-viable cells. Approximately 1 million PBMCs were transferred in cryovials in 1 ml 90% heatinactivated fetal calf serum (FCS) and 10% dimethyl sulfoxide (DMSO; Sigma) and immediately placed into a freezing box, placed overnight into an −80°C freezer. On the following day, the vials were placed at −150°C for storage.
Cell-Based Assay for Detection of Antibodies to Cell Surface MOG in Plasma HEK293A cells transfected with full-length human MOG were used as antigenic substrate in combination with control cells as previously described (29). Briefly, these stable MOG cells were used to detect patient plasma Ig by flow cytometry. As a control,

Statistical Analysis
We performed statistical analyses of cellular immunophenotyping using GraphPad Prism version 8.0.1 (GraphPad Software, San Diego, CA, USA). Data are presented as mean ± standard error of the mean (SEM). For all data sets which could be accurately modeled by a Gaussian normal distribution, an unpaired t-test was used for analysis of differences between groups; otherwise, the Mann-Whitney U-test was used. Within each group, paired comparison of non-stimulated versus rh-MOG stimulated was performed for normal distribution using the paired t-test; otherwise, the non-parametric Wilcoxon signed rank test was used. Statistical significance was assigned to values of p < 0.05, and the symbols used were p <0.05 (*), p <0.01 (**) and p <0.001 (***). For demographic data, one-way ANOVA test, Chi-square test and Fisher's exact test were used.

Demographic Characteristics of Patients
Demographic data are summarized in Table 1. Twenty-two children were included in this study. The mean onset age of each group were similar as analyzed by one-way ANOVA, but that of MS was significantly higher than that of MOGR (unpaired t-test p = 0.009). The gender proportion of each group was similar as analyzed by Chi-square test, but MOGR occurred only in female in our cohort (5/5, 100%), a proportion significantly higher than that in MOGNR (Fisher's exact test p = 0.027) and in MS (Fisher's exact test p = 0.044).  Figure 1A). The ratio of rh-MOG-stimulated-Th2 cells to unstimulated-ones was significantly higher in the MOGAD group than in the control group (means: 1.83 ± 0.64 vs. 0.91 ± 0.8; Mann-Whitney U-test p = 0.036) ( Figure 1D). When comparing MOGNR and MOGR, the percentage of Th2 cell was increased upon rh-MOG stimulation in MOGNR (means: 1.28 ± 0.67 vs.

CD45RA -Foxp3 + T regs Decreased Upon rh-MOG Stimulation in MOGR
We next evaluated the percentage of Foxp3 + T regs among CD45RAcells to evaluate the balance between CD4 + Foxp3 + T regs and effector/memory CD45RA -Foxp3 + T reg cells. This latter cell percentage did not differ after rh-MOG stimulation in MOGAD (means: 1.91 ± 0.2 vs. 1.83 ± 0.26; Wilcoxon-test p = 0.294), MS (means: 2.52 ± 0.33 vs. 2.51 ± 0.28; Wilcoxon-test p = 0.865) and control groups ( Figure 2E). The same result was observed when expressed as a ratio (ratio of Foxp3 + T reg cells among CD45RAcells in rh-MOG-stimulated versus unstimulated conditions) ( Figure 2F). When subdividing MOGAD into MOGNR and MOGR, it appeared that the percentage of Foxp3 + cells in CD45RAfraction was significantly decreased in the MOGR group after rh-MOG stimulation (means: 2.37 ± 0.23 vs. 1.99 ± 0.17; paired t-test p = 0.021) but not in the MOGNR group ( Figure 2G). Interestingly, the ratio of rh-MOG-stimulated-CD45RA -Foxp3 + T reg cells to unstimulated-ones was significantly lower in MOGR as compared to MS (means: 0.84 ± 0.03 vs. 1.04 ± 0.05; unpaired t-test p = 0.035), and as compared to MOGNR (mean: 1.1 ± 0.08; unpaired t-test p = 0.037). These results indicate a decrease in Foxp3 + cells among CD45RAcells in MOGR after stimulation by rh-MOG ( Figure 2H). subsets: naive T reg cells (nT regs : CD4 + CD45RA + FoxP3 int ), effector T reg cells (eT regs : CD4 + CD45RA -Foxp3 high ) which are both suppressive, and non-suppressive Foxp3+ cells (non-T regs : CD4 + CD45RA -Foxp3 int ). The majority of nT regs are thought to recently originate from the thymus, that may subsequently convert into eT regs (31). This strategy of analysis allows better identification of the Foxp3 expressing cells exhibiting suppressive properties (31). To investigate whether differences in these T reg subsets existed among our groups of patients, we quantified these populations (nT regs , eT regs and non-T regs among Foxp3 + cells) as well as T effector/memory cells among Foxp3-cells (T eff/mem : CD4 + CD45RA -Foxp3 -). The gating strategy used is presented in Figure 3A.

T regs Subpopulation Differences Between MOGR and MOGNR
We observed that nT regs tended to be lower in MOGR than in any other group of patients, and significantly lower than in control patients ( Figure 3B). No change in the percentage of eT regs was observed between MOG-stimulated and unstimulated conditions in MOGNR, MOGR, MS and control groups. However, when comparing the proportions of eT regs upon rh-MOG stimulation, MS exhibited a higher proportion of eT regs than MOGNR and control ( Figure 3C). This was specific to the eT regs fraction since no difference in the percentages of non-T regs Foxp3 + cells among MOGR, MOGNR MS and control, in MOGstimulated or unstimulated conditions was observed ( Figure 3D). As a control, we also determined the proportion of T eff/mem Foxp3cells (CD4 + CD45RA -Foxp3 -). This fraction was not affected by rh-MOG stimulation as no difference was observed between stimulated and unstimulated conditions. We observed that MS exhibited higher proportion of T eff/mem Foxp3cells compared to the control group ( Figure 3E). Lastly, we evaluated the ratio of eT regs Foxp3 + cells/non-T regs Foxp3 + cells to evaluate the balance between suppressive and effector cells within the activated fraction. The impact of rh-MOG stimulation on this ratio was not significant in all group considered. However, the MOGR showed a significantly higher eT regs /non-T regs ratio than MOGNR and control, in both non-stimulated and rh-MOG stimulated conditions ( Figure 3F).

DISCUSSION
One main finding of this study is that there are differences in the CD4 + T cells immunological phenotypes of ADS clinical subsets, with a significant increase in CD4 + Th2 and Th17 cells following stimulation with rh-MOG in MOGAD children at onset of demyelinating events but not in MS and control patients. Within MOGAD patients, a significant increase of Th17 induced by rh-MOG stimulation was observed in patients without relapse (MOGNR). Importantly, CD4 + Foxp3 + T regs were significantly increased in response to rh-MOG in MOGNR, while CD45RA -Foxp3 + T regs decreased upon rh-MOG stimulation in MOGR patients.
Our results point at a major role of MOG-specific CD4 + T cells in MOGAD pathogenesis. The absence of significant changes in MS cells upon rh-MOG stimulation suggests that CD4 + A B D E F C FIGURE 3 | Ratio eT regs to non-T regs is higher in MOGR than in MOGNR. (A) Representative dot plot showing gating strategy for Foxp3-expressing subsets. The different Foxp3 + subsets were analyzed following the differential expression of CD45RA and Foxp3 staining, gated on CD3 + CD4 + . Naive T regs (nT regs : CD4 + CD45RA + Foxp3 int ) (B) and effector T regs (eT regs : CD4 + CD45RA -Foxp3 high ) are both known to be suppressive in vitro whereas non-suppressive T regs (non-T regs CD4 + CD45RA -Foxp3 int ) lack suppressive activity and are pro-inflammatory (31). lymphocytes do not respond to MOG antigen in MS, reminiscent of the absence of MOG-Abs in this pediatric ADS (9,10). Several groups also showed no difference in number of circulating MOGautoreactive T-cells in MS compared to healthy controls (32). However, in a recent study, a novel and highly sensitive method for detection of antigen-specific T-cells using bead-bound MOG as stimulant allowed to detect circulating autoreactive CD4 + T-cells producing IFN-g, IL-22 or IL-17A in 46-59% of adult MS patients.
The patients included in this study were adults with MS under natalizumab, which blocks the very late antigen 4 (VLA-4) dependent cell migration across the blood-brain barrier into CNS, and could result in an accumulation of MOG specific Tcells in the circulation of treated patients, that may have increase their numbers in this assay (33) whereas our patients, at onset of the disease, had not been treated yet. The rates of MOG-specific Th1 lymphocytes are low in our ADS patients as compared to that found in other studies (34). This may be explained by the fact that in this latter study PMAionomycin is used to stimulate the cells, whereas we compared rh-MOG stimulated cells to unstimulated ones.
Involvement of Th2 inflammation in autoimmunity including ADS is increasingly found as a component of these diseases (35). In some EAE models, a harmful Th2 response upon exposure to MOG autoantigen was observed beside the classical Th1/Th17 responses in mice (36) and in marmosets (37). These studies suggest that a Th2-type immune response plays a role in the development of EAE, giving further importance to the increase in Th2 cells observed in our MOGAD patients. Recently, in neuromyelitis optica spectrum disorder (NMOSD), an autoimmune demyelinating disorder characterized by auto-Abs targeting the astrocytic aquaporin-4 (AQP4) water channel in the serum of patients, it was found that a therapeutic strategy promoting a shift from Th1/Th17 to Th2 responses is potentially deleterious in NMOSD (35). In other autoimmune diseases with autoantibodies such as systemic lupus erythematosus (SLE), beside the well-established role of Th1/Th17, a Th2 environment and increased basophils are associated with lupus nephritis in human (38).
In rodent EAE, it has also been reported that IL-17-deficient mice have a less severe disease than wild-type mice (39), and that IL-17 worsened EAE in mice (40). In EAE in marmosets, it was found that treatment with an anti-IL-17A antibody induced a moderate delay of clinical scores, without abrogating EAE development (41). Th17 have often been associated with demyelinating diseases in children and adults, such as MS (42,43) and MOGAD (44). Our results support this finding and suggest that Th17 may be more particularly involved in MOGNR patients where it is significantly increased as compared to MOGR patients.
In MOGNR patients, beside the increase of MOG-specific Th2 and Th17 responses, we observed a significant increase in CD4 + Foxp3 + T reg cells. On the contrary, in MOGR patients, effector/memory CD45RA -Foxp3 + T regs significantly decreased upon rh-MOG stimulation. Th17 is known to induce autoimmune tissue injury, whereas T regs inhibit autoimmunity and tissue injury. Disruption of the Th17/T reg balance is thought to be involved in the development of various autoimmune disorders and chronic inflammation (45,46). When we assessed specifically T regs , the ratio of CD4 + Foxp3 + T cell percentage upon stimulation versus unstimulated conditions was higher in MOGNR compared to MOGR, MS and control. Focusing on the percentage of Foxp3 + cells among CD45RAcells showed the relative contribution of Foxp3 + cells among effector/memory cells. Interestingly, we observed a reduced percentage of Foxp3 + cells among CD45RAcells upon rh-MOG-stimulation in MOGR but not MOGNR. Because Foxp3 expressing cells may include cells with different suppressive activity and notably a non-suppressive fraction, we further studied the Foxp3 + fraction. In MOGR, nT regs tended to be lower than in any other group of patients. On the contrary, eT regs were higher in MOGR than in MOGNR and control groups. Although the eT regs fraction decreased upon stimulation by rh-MOG in MOGR, the percentages in stimulated conditions in MOGR remained higher than in MOGNR and control group. Foxp3 is essential for differentiation and suppressive function of T regs . In human, conventional non T reg Tcells have been shown to transiently express Foxp3 upon activation (47). Additionally, emerging evidence suggests that Foxp3 expression is not always stable in T regs and can be lost. Several studies on T reg in other autoimmune diseases, such as type 1 diabetes and rheumatoid arthritis, suggest a loss of Foxp3 expression and the generation of pathogenic Th17 cells (48), in association with Foxp3 instability in T regs in these diseases. Different mechanisms would be involved in controlling the stability and expression of Foxp3. Demethylation of an evolutionarily conserved element within the Foxp3 genomic locus, conserved noncoding sequences 2 (CNS2), was described as an important feature regarding its expression stability (49,50). Recently, downregulation of Foxp3 mRNA expression was described in PBMC of NMOSD patients (51). The decrease, significant for CD45RA -Foxp3 + T regs and trend for eT regs , we found in MOGR upon rh-MOG stimulation could also be explained by apoptosis of T regs . In healthy condition, T regs are resistant to apoptosis induced by T-cell receptor (TCR) (52), including T regs expressing selfreactive TCR (53). However, in autoimmune thyroiditis, it was recently suggested that apoptosis of T regs through self-reactive TCR activation can drive autoimmunity (54). In a study based on PBMC stimulation in vitro, examination of AQP4-specific T-cells revealed a significantly reduced frequency of T regs in NMOSD patients in response to rhAQP4, in comparison to healthy controls (55). Decrease of Foxp3 + T reg cells could influence the multiphasic evolution of MOGAD and be linked to a poor control of inflammation. In our study, it is tempting to speculate that MOGNR patients, which have higher Th17 in response to rh-MOG than MOGR at onset, can control autoimmunity since their T regs are also increased upon rh-MOG stimulation. On the contrary, MOGR patients, in which T regs and particularly CD45RA -Foxp3 + T regs are decreased in response to rh-MOG, may have intermittent loss of tolerance toward this autoantigen and therefore enter into a relapsing form of MOGAD.
In primates treated with rhMOG/IFA, an EAE is induced having immune-inflammatory characteristics closer to MOGAD than to MS (27). Subsequent treatment of these primates with anti-DC-ASGPR-MOG increased their number of MOG-specific CD4 + CD25 + FOXP3 + CD39 + T regs as compared to controls. This increase likely precluded EAE seen in the controls (28) and reminded the increase in T regs upon rh-MOG stimulation found in MOGNR patients.
One major limitation of our study is the small sample size of patients including MOGAD, but this is inherent to the fact that ADS are rare diseases. The timing of the samples was not always similar in all patients and there might be temporal modifications of immune cells, which could also have biased our work. The absence of healthy control is also a limitation.

CONCLUSIONS
In conclusion, an increase of Th2 and Th17 after stimulation by rh-MOG was observed in ADS children with MOGAD, particularly in MOGNR. T regs have differences in their subset pattern and behave differentially in MOGR and MOGNR. CD4 + Foxp3 + T cell are increased upon rh-MOG stimulation specifically in MOGNR while in MOGR CD45RA -FoxP3 + T regs decreased upon rh-MOG stimulation. This may reflect instability or apoptosis of T regs induced by rh-MOG that may subsequently contribute to a probable loss of tolerance toward MOG autoantigen in MOGR which may explain relapses in this recurrent pediatric autoimmune disease. These results suggest that T regs are targets to develop new therapeutic strategies of MOGAD.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Commission Nationale de l'Informatique et des Liberteś. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.