Myelin-reactive antibodies initiate T cell-mediated CNS autoimmune disease by opsonization of endogenous antigen

MOG p35-55 TCR transgenic 2D2 mice were kindly provided by Dr. Kuchroo (Boston, USA). MOG Ig heavy chain knockin (Th) mice were kindly provided by Dr. Wekerle (Munich, Germany). Fcγ receptor knockout (FcγR^−/−) mice were kindly provided by Dr. Nimmerjahn (Erlangen, Germany). Wild-type (WT) C57BL/6 mice were purchased from Charles River (Sulzfeld, Germany). All murine experiments were carried out as approved by the government of Upper Bavaria (protocol number 55.2-1-54-2531-67-09) and the government of lower Saxony (protocol number 33.9-42502).

B cell depletion was achieved by weekly intraperitoneal (i.p.) injections of 200 µg of murine anti-CD20 or anti-ragweed isotype control Ab (both provided by Genentech, South San Francisco, USA) in 200 µl PBS.

Mouse MOG peptide (p) 35-55 (MEVGWYRSPFSRVVHLYRNGK) was synthesized by Auspep (Parkville, Australia). Murine MOG protein 1-117 (mMOG) and human MOG protein (hMOG) were kindly provided by C.C.A. Bernard and synthesized, purified and refolded as previously reported [8]. 8–10 week-old-female Th or WT mice were immunized subcutaneously with either a suboptimal dose of 35 µg mMOG or an optimal dose of 50 µg mMOG in Complete Freund’s Adjuvant (CFA) followed by 200 ng of pertussis toxin (Sigma-Aldrich, St. Louis, USA) i.p. at the day of immunization and 2 days thereafter. Alternatively, WT mice were immunized with 100 µg MOG p35-55 in CFA followed by two injections of 300 ng pertussis toxin. For spontaneous EAE experiments, Th mice were bred with 2D2 mice (Thx2D2). To induce EAE by transfer of MOG-specific Ab, a serum preparation containing anti-MOG Ab or monoclonal Ab clone 8.18C5 were injected into the tail vein of WT or 2D2 recipients. 150 µl Th serum or control serum was injected twice a week up to a total of five injections; 150 µg 8.18C5 Ab or isotype control Ab (clone: MOPC-21, Bio X cell, West Lebanon, USA) was injected twice a week up to a total of ten injections. In recipients, no immunization or adjuvant treatment was added.

Mice were assessed for clinical signs of EAE as follows: 0 = no clinical disease, 1 = tail weakness, 2 = hind limb weakness, 3 = one paralyzed hind limb, 4 = two paralyzed hind limbs, 5 = moribund or dead. To evaluate balance and general motor function the elevated beam test was used; mice were placed on a raised beam with a maximal height of 40–50 cm and a length of 100 cm and the time needed to traverse was measured. Mice were evaluated daily starting 2 weeks prior to the respective treatment.

Mice were injected with 40 µg mMOG in 10 µl PBS or 10 µl of 10 % Evans blue percutaneously into the cisterna magna with a 30-gauge needle in 45° anteflexion of the head.

Serum containing high titers of pathogenic anti-MOG Ab or control serum was obtained from Th mice immunized with 100 µg mMOG (Th serum) or from WT mice immunized with 100 µg MOG p35-55 (control serum), respectively. Donor mice were immunized to ensure the maximum capacity of obtained myelin-specific Ab. 14 days after immunization, blood was obtained by puncture of the left ventricle; serum was separated by centrifugation, pooled for further use and stored at −20 °C.

Anti-MOG monoclonal Ab clone 8.18C5 was generated by hybridoma cells kindly provided by Dr. Linington (Glasgow, UK). Hybridoma cells were cultured in complete medium (RPMI, 5–10 % fetal calf serum, 50 U/ml penicillin, 50 µg/ml streptomycin, 2 mM l-glutamine, 1 mM sodium pyruvate, 0.05 mM β-mercaptoethanol) in large-scale flasks (Greiner bio-one, Kremsmuenster, Austria) and Ab was purified using rProtein A/Protein G Sepharose columns (rProtein A/Protein G GraviTrap, GE Healthcare, Little Chalfont, UK), according to manufacturer's recommendations. 8.18C5 IgG was digested by ficin (Pierce Mouse IgG1 Fab and F(ab′)₂ Preparation Kit, Thermo Scientific, Waltham, USA), according to manufacturer’s recommendations; integrity and binding capacity of 8.18C5 Ab and resulting 8.18C5 F(ab′)₂ fragments were verified by 6 % sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) under non-reducing conditions, followed by protein staining with Coomassie brilliant blue G250 (Biorad, Munich, Germany) and competitive anti-MOG enzyme linked immunosorbent assay (ELISA), respectively.

IgG Ab preparations were isolated by Protein G Sepharose 4 Fast Flow (GE Healthcare) from plasma exchange fluid obtained from two patients with neuromyelitis optica spectrum disorders (NMOSD), and plasma from one healthy blood donor. One patient was a 62-year-old Caucasian woman with past history of relapsing bilateral optic neuritis. The patient was admitted because of a severe brainstem relapse that was treated with corticosteroids and plasma exchange followed by rituximab. The other patient was a 55-year-old Caucasian woman with a relapsing disease since the age of 48, who presented with a severe optic neuritis relapse in January 2016 that was treated with corticosteroids and plasma exchange followed by rituximab. In both cases, cerebrospinal fluid showed no oligoclonal bands, and serum samples were negative for anti-AQP-4 Ab but positive for anti-MOG Ab (titer 1:1280, patient #1; 1:5120, patient #2) measured by a recombinant live cell-based immunofluorescence assay with HEK293 cells as previously described [30]. Prior to our in vitro assay, the general ability to bind soluble recombinant hMOG was tested by ELISA using an anti-MOG Ab positive IgG preparation. The healthy blood donor was tested anti-MOG Ab negative, and IgG preparation served as negative control. Samples were kindly provided by Dr. Reindl (Innsbruck, Austria) and Dr. Saiz (Barcelona, Spain). The use of the patient’s material was approved by the Ethic Committee of the Hospital Clinic of Barcelona, written consent was obtained.

To block Fcγ receptors in vivo, mice were injected i.p. daily with 100 µg anti-CD16/CD32 Ab (Clone: 2.4G2, TONBO bioscience, San Diego, USA) in 100 µl 1xPBS starting 2 days prior to further treatment.

In vivo proliferation of T cells was evaluated by i.p injection of 200 µl Bromdesoxyuridin (BrdU; 10 mg/ml) 24 h before flow cytometry evaluation using a BrdU Flow kit (BD Pharmingen, San Diego, USA), according to manufacturer´s recommendations.

To generate bone marrow-derived macrophages (BMDM), bone marrow isolated from hind limbs of WT or FcγR^−/− mice was cultured at 37 °C for 7 days in medium containing 30 % conditioned L929 cell supernatant (DMEM, 30 % L929 supernatant, 10 % fetal calf serum, 5 % horse serum, 50 U/ml penicillin, 50 µg/ml streptomycin, 0.05 mM β-mercaptoethanol). For bone marrow-derived dendritic cell (BMDC) generation, bone marrow was isolated from hind limbs of WT mice and cultured for 7 days in complete medium (RPMI, 10 % fetal calf serum, 50 U/ml penicillin, 50 µg/ml streptomycin, 2 mM l-glutamine, 1 mM sodium pyruvate, 0.05 mM β-mercaptoethanol) containing 25 ng/ml GM-CSF (Sigma-Aldrich).

Adherent BMDM or BMDC were harvested using cell scraper, 0.5 × 10⁵ cells/well were plated into 96-well flat-bottom plates (Sarstedt, Nuembrecht, Germany) and pre-stimulated with 500 ng/ml lipopolysaccharide (LPS, Sigma-Aldrich). 24 h thereafter, cells were incubated for 2 h with DyLight-405 labeled conformational mMOG (mMOG-DyLight-405) or ovalbumin (OVA)-FITC (Life Technologies, Thermo Fisher Scientific) in the presence of 50 µg/ml 8.18C5 Ab, 8.18C5 F(ab′)₂ fragment, anti-OVA Ab (clone: TOSG1C6, BioLegend, San Diego, USA) or isotype control Ab (clone: MOPC-21). To investigate phagocytosis of conformational hMOG, murine BMDM which are known to recognize human IgG via their Fc receptor [37] were incubated with hMOG-DyLight-405 for 2 h in the presence of 15 µg/ml of the respective IgG preparation. Where indicated, Fcγ receptors were blocked using anti-mouse CD16/CD32 Ab (Clone: 2.4G2, 1:100 dilution).

Adherent BMDM were harvested using cell scraper, 0.5 × 10⁵ cells/well were plated into 96-well flat-bottom plates and pre-stimulated with 50 ng/ml LPS and mMOG in the presence of 50 µg/ml 8.18C5 Ab, 8.18C5 F(ab′)₂ fragment or isotype control Ab. Where indicated, Fcγ receptors were blocked using anti-mouse CD16/CD32 Ab (Clone: 2.4G2, 1:100 dilution). 24 h thereafter, 1 × 10⁵ MACS-purified (Pan T cell Isolation Kit, Miltenyi, Bergisch Gladbach, Germany) carboxyfluorescein succinimidyl ester (CFSE)-stained (CFSE Cell Division Tracker Kit, BioLegend) or unstained T cells from 2D2 mice were added per well. 72 h thereafter, T cell proliferation and differentiation were evaluated by flow cytometry or ELISA, respectively.

To obtain single cell suspensions of lymphoid tissues, respective spleens and lymph nodes were carefully dissected and passed through 70 µm strainer. Freshly obtained blood was mixed 1:2 with 1 mM EDTA and erythrocytes were lysed using BD Pharm Lyse Buffer, according to manufacturer’s recommendation. B cells were stained for B220 (BioLegend) CD19 (BioLegend) and/or CD20 (BioLegend), T cells for CD3 and/or CD4 (all BD Bioscience). T cell activation was determined by expression of CD69 (BD Bioscience). T cell differentiation was evaluated by intracellular cytokine staining for IFN-γ (eBioscience, San Diego, USA) and IL-17A (BD Bioscience) after 4-h incubation with phorbol 12-myristate 13-acetate (PMA, 50 ng/ml) and ionomycin (0.5 μg/ml) in the presence of monensin (1 µl Golgi-Stop per ml cell suspension, BD Bioscience). Dead cells were excluded using an Aqua Dead Cell Stain Kit (Invitrogen, Thermo Fisher Scientific, 405 nm excitation). CNS-infiltrating cells were isolated by discontinuous density gradient (Percoll) [24] and stained in a similar manner for CD3, CD4, IFN-γ and IL-17A. Myeloid APC were stained for CD11b and CD11c (both BioLegend). Expression of Fcγ receptor (FcγR) I and III of BMDM and BMDC was determined using anti-CD64 Ab (FcγRI, BioLegend) and anti-CD16 Ab (FcγRIII, BioLegend). APC secretion of IL-1α and IL-23 (both BioLegend) was determined after 5-h incubation with 500 ng/ml LPS in the presence of monensin (1 µl Golgi-Stop per ml cell suspension. BD Bioscience).

ELISA for IFN-γ, IL-17 or GM-CSF was performed using paired monoclonal Ab per manufacturer’s recommendations (DuoSet, R&D Systems, Wiesbaden‐Nordstadt, Germany). The results are expressed as an average of duplicates/triplicates ± SEM. iMark microplate reader and software was used for data analysis (Bio-Rad Laboratories Inc., Hercules, CA, USA).

96-well plates were coated with 10 µg/ml conformational mMOG or conformational hMOG in 1xPBS overnight. Diluted samples were added for 2 h. After washing, plate-bound Ab of murine samples and 8.18C5 Ab were detected with horseradish peroxidase (HRP)-conjugated anti-mouse IgG, directed against the Fc part of the bound Ab (1:6000; Sigma-Aldrich) or against the whole molecule (1:5000; Sigma-Aldrich). Anti-MOG Ab in human samples were detected with HRP-conjugated anti-human IgG (1:1000; Sigma-Aldrich). Plates were read at 450 nm wavelength by a Tecan Genios plate reader and analyzed using Magellan6 software or iMark microplate reader and software.

Brain and spinal cord tissue was PBS-perfused and cryofixed or perfused and fixed with 4 % paraformaldehyde and paraffin embedded. To determine myelin loss and inflammatory infiltration, vertically or horizontally oriented sections were stained with Luxol Fast Blue and periodic acid Schiff (LFB/PAS) or hematoxylin and eosin (H&E), respectively. Infiltration of immune cells was determined by immunohistochemistry using Ab against CD3 (1:150 dilution; DCS, Hamburg, Germany), B220 (1:200 dilution; BD Pharmingen) or MAC3 (1:200 dilution, BD Bioscience). Evaluation of histologic samples was performed in a blinded manner.

EAE experiments were evaluated for significance using the Mann–Whitney test. The Student t test was used for all other statistical comparison. A value of p < 0.05 was considered significant. Data are presented as mean ± SEM.

To delineate the relative pathogenic contribution of myelin-specific Ab from myelin-specific B cells, we first investigated the effect of preventive B cell depletion on actively induced EAE in Th mice constitutively containing a high frequency of MOG-recognizing B cells as well as plasma cells producing pathogenic anti-MOG Ab (Suppl. Figure 1a). Results were compared to WT mice, in which only a comparably small fraction of B cells can respond to mouse MOG protein immunization, and developing Ab titers are non-pathogenic [32]. As shown in suppl. Figure 1b, three injections of anti-mouse CD20 Ab efficiently depleted virtually all CD20 positive (⁺) B cells from blood, bone marrow, lymph node and spleen of WT mice, sparing only CD20 negative B cells in the bone marrow. Corresponding results were obtained in secondary lymphoid organs of Th mice following anti-CD20 Ab treatment (Suppl. Figure 1c). Upon immunization with a suboptimal dose of mMOG protein, WT mice developed mild but consistent EAE with a slight benefit in the B cell-depleted group [52]; using the identical induction regimen, Th mice developed fulminant EAE irrespective of anti-CD20 Ab treatment prior to immunization (Fig. 1a) or after EAE onset (Suppl. Figure 1d). These findings were closely reflected by a massive CNS demyelination and inflammation in both B cell-depleted and isotype-treated Th mice (Fig. 1b), along with an unhindered Th1/Th17 differentiation in the absence of B cells (Suppl. Figure 1e). In line with the fact that anti-CD20 Ab spare CD20 negative plasma cells, constitutively produced anti-MOG Ab remained high in preventatively anti-CD20 Ab treated Th mice during the priming- and CNS-infiltration period (Fig. 1c).

Next, we applied anti-CD20 Ab treatment to Thx2D2 mice (Suppl. Figure 1f) containing MOG-recognizing B cells, MOG-specific T cells and high titers of constitutively produced pathogenic anti-MOG Ab. To our surprise, preventative B cell depletion at the age of 3–4 weeks neither affected incidence of spontaneous EAE (18 anti-CD20-treated vs. 22 isotype-treated in equally sized groups) nor its severity (Fig. 1d). As in immunized Th mice, depletion of CD20⁺ B cells did not affect generation of encephalitogenic Th1 and Th17 cells (Fig. 1e) nor lower endogenous anti-MOG Ab production at the time of disease induction (Fig. 1f); analogous clinical and immunological results were obtained when anti-CD20 Ab treatment was initiated in fully established spontaneous EAE (Suppl. Figure 1g–h).

These findings endorsed a prominent pathogenic role of myelin-reactive Ab independent of B cells, which we next investigated directly and independent of anti-CD20 Ab treatment by adoptive transfer. At first, serum from immunized Th mice (Th serum) was transferred into MOG p35-55-immunized WT mice. As shown in Fig. 1g, repeated intravenous (i.v.) injections of Th serum indeed substantially accelerated EAE severity in WT recipients and, of note, onset and severity closely resembled EAE in immunized Th mice. Histologic analysis revealed accentuated CNS inflammation and demyelination in Th serum recipients (Fig. 1h, left and middle), which is in line with the reported demyelinating nature of anti-MOG Ab in ongoing EAE [26, 46]. Exceeding this anticipated result, Th serum recipients furthermore displayed a massive CNS infiltration of CD3⁺ T cells (Fig. 1h, right).

Together with the observation that early B cell depletion failed to halt spontaneous EAE development in Thx2D2 mice, these findings fueled our hypothesis that in lieu of antigen-specific B cells, anti-MOG Ab may be capable of triggering activation and expansion of myelin-reactive T cells in a manner sufficient to cause EAE. To test this hypothesis, we transferred serum from Th mice or purified 8.18C5 Ab into otherwise naïve 2D2 recipients, which contain MOG-specific T cells but no transgenic B cells (Fig. 2a). Strikingly, five out of fourteen 2D2 mice receiving Th serum developed classical EAE with ascending paralysis, one mouse developed atypical EAE with severe ataxia (Table 1). Histologic analysis confirmed that all EAE-diseased Th serum recipients showed extensive CNS demyelination, inflammation and infiltration by CD3⁺ T cells (Fig. 2b), while none of the twelve 2D2 mice receiving control serum showed signs of clinical or histological EAE. To confirm that myelin-reactive Ab and not any other serum component had triggered EAE, we next transferred 150 µg of purified 8.18C5 Ab, the amount equivalent to its content in 150 µl Th serum. As indicated in Fig. 2c, 2D2 mice receiving a total of 10 injections showed a rapid increase in anti-MOG Ab, comparable to the level observed in Thx2D2 mice with spontaneous EAE (gray area). Again, thirteen out of twenty-three 8.18C5 Ab recipients, but none of the seventeen 2D2 mice receiving control Ab, developed severe clinical or histologic EAE (Table 1; Fig. 2d). When evaluated by the more sensitive elevated beam test, 8.18C5 Ab recipients without paralysis yet showed a significant decline in their agility (Table 1; Fig. 2e). CNS damage was quantified in all 8.18C5 Ab-transferred 2D2 mice, which revealed a pronounced inflammation and CNS infiltration by both B and T cells in the majority of these mice, while extensive demyelination was restricted to the paralyzed subgroup (Fig. 2f + g). Here, all histologic parameters were in their extent comparable to Thx2D2 mice with spontaneous fulminant EAE, while in some cases, the degree of infiltration by highly activated MAC3⁺ macrophages even exceeded the level in diseased Thx2D2 mice. To investigate in which sequence peripherally applied anti-MOG Ab, MOG-reactive T cells and activated myeloid APC may have triggered EAE, we transferred anti-MOG Ab into naïve WT mice (Suppl. Figure 2a). Recipients showed no signs of paralysis, impaired agility or, importantly, CNS demyelination (Suppl. Table 1; Suppl. Figure 2b + c), suggesting that in the absence of ongoing CNS inflammation, myelin within the CNS is not recognized by peripherally applied myelin-reactive Ab. Alternatively, myelin-reactive Ab could have triggered activation of peripheral myelin-reactive T cells prior to CNS entry. To investigate this hypothesis, we evaluated proliferation of T cells in lymph nodes and spleen of healthy, naïve 2D2 mice after peripheral injection of 8.18C5 Ab and co-administration of mMOG protein intrathecally (Fig. 3a; Suppl. Figure 3a + c). As shown in Fig. 3b–d, 8.18C5 Ab triggered in vivo expansion of CD4⁺ T cells in spleen, inguinal and cervical lymph nodes 4 days after mMOG injection and 6 days after the initial administration of 8.18C5 Ab. While at this early time point only a trend toward an enhanced Th17 differentiation could be observed, administration of 8.18C5 Ab lead to a distinct upregulation of the early activation marker CD69 on CD4⁺ T cells (Fig. 3e). Further, co-administration of 8.18C5 Ab and mMOG was associated with a concomitant upregulation of MHC class II and an enhanced release of IL-1α by CD11b⁺ myeloid APC (Fig. 3f). To a lesser extent, T cell proliferation and activation of myeloid APC were also observed in 2D2 mice receiving control Ab, which may be reflective of endogenous anti-MOG Ab development in these mice when MOG is administered intrathecally. Irrespectively, 8.18C5 Ab-mediated in vivo activation and proliferation of T cells as well as activation of myeloid APC could be completely abolished by addition of an in vivo Fc-blocking Ab (Suppl. Figure 3b; Fig. 3b–f). In conjunction, these findings indicate that myelin-specific Ab trigger activation of peripheral myelin-specific T cells and highlight Fc-mediated opsonization of antigen as most plausible mechanism.

Activation of CD4⁺ T cells requires antigen presentation in the context of MHC class II. Prior to its presentation, antigen must be taken up and presented by the immune system, which, particularly at low concentrations, can be facilitated by specific binding to B cells [9, 17, 38] or Ab [22, 31]. In the latter case, Ab-decoration triggers Fc receptor-mediated internalization and processing of the antigen by APC [49], followed by its presentation to adaptive immune cells. This mechanism of Ab-mediated antigen opsonization is believed to be central for instance in development of systemic lupus erythematosus (SLE) [15]. To investigate whether in our setting, MOG-specific Ab may have triggered in vivo activation of myelin-specific T cells [35] by opsonization of rare CNS auto-antigen, we developed an in vitro setting capable of dissecting the respective components; we generated BMDM or BMDC from WT mice expressing high levels of Fc receptors or from knockout mice lacking Fcγ receptors (FcγR^−/−) (Fig. 4a + b). Further, we digested the 8.18C5 Ab to obtain an antigen-recognizing, but Fc-cleaved F(ab′)₂ fragment (Fig. 4c; Suppl. Figure 4a + b). Using these tools, we demonstrated that intact 8.18C5 Ab strongly promoted recognition and internalization of fluorescence-labeled mMOG by either BMDC or BMDM (Fig. 4d + e, left). This effect strictly depended on cellular expression of Fcγ receptors (Fig. 4e, right); along the same lines, the Fc-cleaved 8.18C5 F(ab′)₂ fragment failed to enhance internalization and rather neutralized antigen recognition (Fig. 4f). Of interest, Ab-mediated opsonization similarly occurred when the random control antigen OVA was used in combination with anti-OVA Ab (Fig. 4g), corroborating that Ab opsonize any fitting antigen.

To understand the downstream immunological consequences of Ab-mediated antigen opsonization we co-cultured WT BMDM with naïve 2D2 T cells, again in the presence of mMOG. In line with our in vivo findings, addition of 8.18C5 Ab specifically accelerated T cell proliferation and differentiation into encephalitogenic T cells, while blockade of Fcγ receptors on APC reversed this effect (Fig. 5a–c; Suppl. Figure 4c). Along the same lines, the 8.18C5 F(ab′)₂ fragment again failed to enhance internalization and subsequent presentation of mMOG to myelin-specific T cells (Fig. 5d). In summary, these results indicate that in development of CNS autoimmune disease, systemic auto-Ab facilitate APC recognition of otherwise undetected CNS antigen, resulting in activation, pro-inflammatory differentiation and CNS infiltration of auto-reactive T cells.

To exemplify that human IgG is capable of opsonizing human myelin protein, we next investigated IgG preparations obtained from two NMOSD patients with high titers of anti-MOG Ab determined in an established cell-based assay [30]. As a prerequisite to be tested in our in vitro assay, we first ensured that patient-derived anti-MOG Ab recognized soluble hMOG in vitro by a conventional ELISA (Fig. 6a). Applying the IgG preparations to our phagocytosis assay, we indeed observed that both anti-MOG Ab containing IgG preparations significantly enhanced the APC uptake of hMOG, when compared to an anti-MOG Ab negative IgG preparation obtained from a healthy individual (Fig. 6b). Ab-mediated opsonization could be largely reversed by addition of an Fc receptor blocking Ab (Fig. 6c), mechanistically corroborating that under these conditions human anti-MOG Ab facilitate APC uptake of human CNS antigen in an Fc-dependent manner.