Lymphotoxin-dependent elevated meningeal CXCL13:BAFF ratios drive grey matter injury

Abstract

In multiple sclerosis (MS), B cell-rich tertiary lymphoid tissues (TLT) in the brain leptomeninges associate with cortical grey matter injury. Using a model of Th17 cell-driven Experimental Autoimmune Encephalomyelitis (EAE) in mice, we found that inhibitors of Bruton’s tyrosine kinase (BTKi) prevented TLT formation and cortical pathology in a BAFF-dependent manner. BTKi reduced expression of Lymphotoxin (LT) ligands, and co-treatment with an LTβ receptor agonist abrogated the benefits of BTKi. TLT and cortical pathology tracked with a high ratio of CXCL13:BAFF in the leptomeninges which was reduced by BTKi. Moreover, we observed high CXCL13:BAFF ratios in post-mortem cerebral spinal fluid (CSF) from MS patients with pathologically confirmed leptomeningeal inflammation, as well as in living MS patients with radiologically confirmed paramagnetic rim lesions. In summary, using EAE we revealed a molecular circuit that leads to TLT formation and cortical injury with translational relevance for detection of this pathology in MS patients.

INTRODUCTION

The frequent observation of lymph node–like organized aggregates of immune cells in chronically inflamed tissues ¹, which we refer to as tertiary lymphoid tissues (TLTs) ², suggests that lymphoid tissue neogenesis may have a role in supporting immune responses at sites of chronic inflammation. Pathological studies of brains from people with multiple sclerosis (MS) have demonstrated the formation of TLT of variable size and organization in the brain leptomeninges of the central nervous system (CNS) ^3–9. These TLTs are rich in B cells ³ and are associated with underlying grey matter injury including (sub-pial pattern) cortical demyelination, neuronal loss, microglial activation and a more severe disease course ^5,6,8,10.

The mechanistic steps that lead to leptomeningeal TLT are unknown. Moreover, fluid biomarkers or radiological proxies of TLT and grey matter injury are lacking^11,12. Current MS therapies are effective at reducing white matter lesions¹³ which can be easily tracked radiologically. On the other hand, (cortical) grey matter lesions have a different and ill-understood etiology ¹⁴, with no effective treatment. Grey matter lesions ultimately lead to neurodegeneration and brain volume loss in MS¹⁵.

Rodent models of MS (experimental autoimmune encephalomyelitis [EAE]) that recapitulate brain leptomeningeal TLTs ^2,16 represent our best tool for studying the molecular requirements for TLT formation. We previously described a passive model of EAE whereby myelin-primed Th17 cells adoptively transferred into SJL/J recipient mice induce paralysis and robust leptomeningeal TLTs². Compared to young SJL/J recipient mice, old SJL/J recipients exhibit severe and unremitting axon and synapse loss accompanied by brain atrophy ¹⁷. In both aged and young recipients, TLTs are associated with degradation of the glial limitans and an inward gradient of subpial cortical demyelination and microglia/macrophage accumulation ^2,18, as seen in MS ^5,6,8.

Remibrutinib (LOU064) is a brain-penetrant selective covalent inhibitor of Bruton’s tyrosine kinase (BTKi) ¹⁹ and has shown pharmacologic efficacy in rodent models of MS but has not been tested for its impact on cortical pathology ^20–23. Herein we used remibrutinib to define the circuitry that drives leptomeningeal TLT and used these findings to identify a potential biofluid-based means for ascertaining brain-compartmentalized inflammation in MS.

RESULTS

BTK inhibition prevents subpial injury in young EAE.

Bruton’s tyrosine kinase (BTK) is an intracellular kinase that mediates signals initiated by surface receptor engagement emanating from the B cell receptor on B cells and the FcγR on myeloid cells ²⁰. Given the therapeutic efficacy of B cell depletion in MS¹³, BTK has been tested as a therapeutic target²⁴. To determine the impact of BTK inhibitors (BTKi) on subpial cortical grey matter pathology independent of its role in T cell priming, we adoptively transferred myelin-specific Th17 cells primed in untreated SJL/J donor mice into recipient mice to induce EAE. EAE mice were then prophylactically treated with BTKi before the onset of clinical symptoms until endpoint (Extended Data Fig. 1a). This treatment regimen resulted in a significant accumulation of the drug in the spleen (Extended Data Fig. 1b) and in the brain (Extended Data Fig. 1c). Pre-primed myelin-specific Th17 cells drive spinal cord white matter injury and paralysis in EAE ^25,26, and enter the leptomeninges prior to B cell accumulation at this site. Thus, similar to what we have previously observed in mice treated with a B cell depleting agent (anti-CD20) ²⁷, BTK inhibition did not significantly impact spinal cord-related readouts including paralysis (Extended Data Fig. 1d-g), the accumulation of neurofilament light (NfL) chain in the serum (Extended Data Fig. 1h-i), or spinal cord pathology (Extended Data Fig. 2).

We previously showed that the cortex, including the glial limitans, is injured in areas adjacent to the TLT but not in areas distal to the TLTs ¹⁸. Therefore, we next examined the impact of BTKi on cortical demyelination and microgliosis, both of which are associated with disease progression in MS ^4,5,8. Immunohistochemistry (IHC) of subpial cortical gray matter showed that, compared to vehicle-treated controls, treatment with BTKi reduced disruption of the glial limitans, cortical demyelination, myeloid cell accumulation and axonal loss (Fig. 1a-d). Importantly, in addition to prophylactic administration, the protective effects of BTKi in the cortex were also observed in a therapeutic drug delivery modality (Fig. 1 and Extended Data Fig. 3). In summary, BTKi abrogates cortical pathology in a passive model of EAE.

Fig. 1. BTK inhibition spares the glial limitans and prevents cortical grey matter demyelination, myeloid cell accumulation and axonal loss in young SJL/J adoptive transfer EAE mice.

Immunohistochemical staining and quantification of somatosensory cortex tissue from naïve, vehicle–treated and BTK inhibitor (BTKi)–treated SJL/J adoptive transfer EAE mice at peak disease (day 12 after adoptive transfer) for glial fibrillary acidic protein (GFAP) visualizing the glial limitans (a) (naïve=7, p-vehicle-treated: n=6, p-BTKi-treated: n=7, t-vehicle treated: n=8, t-BTKi: n=7),proteolipid protein (PLP) visualizing myelin (b) (naïve=7, p-vehicle-treated: n=7, p-BTKi-treated: n=7, t-vehicle treated: n=8, t-BTKi: n=7), ionized calcium binding adaptor molecule 1 (IBA1) visualizing microglia and macrophages (c) (naïve=7, p-vehicle-treated: n=7, p-BTKi-treated: n=7, t-vehicle treated: n=8, t-BTKi: n=8) and neurofilament light chain (NfL) visualizing axons (d) (naïve=7, p-vehicle-treated: n=7, p-BTKi-treated: n=7, t-vehicle treated: n=7, t-BTKi: n=6). White arrowheads in A-C indicate the glial limitans; in A-K indicate microglia/macrophages. Zoomed subpial areas are taken at equivalent depth from the pia mater across groups. Data are shown as means ± SD. Statistical analysis was conducted using Kruskal-Wallis tests. p, prophylactic; t, therapeutic.

BTK inhibition prevents subpial injury in old EAE.

Brain atrophy is a key feature of aged MS^28,29 and in the aged version of our EAE model¹⁷. Therefore, we assessed whether BTKi protect the brain from atrophy and subpial cortical grey matter pathology in old mice that received pre-primed Th17 cells (Extended Data Fig. 4). As in young mice, BTK inhibition did not impact paralysis as expected (Extended Data Fig. 4). Using 7-tesla magnetic resonance imaging (7T MRI) we assessed the impact of BTKi on brain volume. Excitingly, therapeutic administration of BTKi ameliorated whole brain atrophy, with beneficial effects particularly observed in the somatosensory cortex (SSC) (Fig. 2a). Moreover, compared to vehicle-treated controls, mice treated with BTKi exhibited reduced disruption of the glial limitans, minimal cortical demyelination and myeloid cell accumulation, as well as reduced axonal injury (Fig. 2b-e). In summary, therapeutic administration of BTKi abrogates brain atrophy and cortical pathology in aged EAE mice.

Fig. 2. BTK inhibition prevents brain atrophy, spares the glial limitans and prevents cortical grey matter demyelination, myeloid cell accumulation and axonal loss in old SJL/J adoptive transfer EAE mice.

(a) Post-mortem 7T MRI of whole brain and somatosensory cortex (SSC) from naïve (n=4), EAE vehicle-treated (n=5) and EAE BTKi-treated (n=8) old SJL/J mice, expressed as a ratio to axial length. Immunohistochemical staining and quantification of somatosensory cortex tissue from naïve, vehicle–treated and BTK inhibitor (BTKi)–treated SJL/J adoptive transfer EAE mice at peak disease (day 12 after adoptive transfer) for glial fibrillary acidic protein (GFAP) visualizing the glial limitans (b), proteolipid protein (PLP) visualizing myelin (c), ionized calcium binding adaptor molecule 1 (IBA1) visualizing microglia and macrophages (d) and neurofilament light chain (NfL) visualizing axons (e). Zoomed subpial areas are taken at equivalent depth from the pia mater across groups. n = 5 (naïve), 4 (vehicle-treated), 4 (BTKi–treated). Data are shown as means ± SD. Statistical analysis was conducted using One-Way ANOVA with multiple comparison.

Enforced LTβR signaling overrides BTKi neuroprotection.

We next asked whether the protection conferred by BTKi on brain pathology is connected with leptomeningeal TLT formation. We found that, compared to vehicle-treated controls, BTKi reduced both the number and size of TLTs adjacent to the cortex (Fig. 3a,b) and generally throughout the brain meninges (Extended Data Fig. 5). In vehicle-treated controls TLTs were largely populated by B220⁺ B cells and CD3⁺ T cells - both cell populations were significantly reduced in BTKi-treated mice with remaining B cells appearing dispersed through the leptomeninges (Fig. 3c and Extended Data Fig. 6). While we did not phenotype BTKi-sensitive B cells in the leptomeninges, we previously showed that not only mature but anti-CD20 resistant immature B cells accumulate in the leptomeninges during EAE³⁰. Future work should address which of these B cell subsets respond to BTKi.

Fig. 3. BTK inhibition reduces the size and number of B cell-rich leptomeningeal aggregates in SJL/J adoptive transfer EAE mice.

(a) Representative hematoxylin & Eosin (H&E)–stained sections of the leptomeninges overlaying the cortices of vehicle–treated or BTK inhibitor (BTKi)–treated SJL/J adoptive transfer EAE mice at peak disease (day 12 after adoptive transfer) with treatment administered in the therapeutic modality. (b) Quantification of the number and size of leptomeningeal TLTs shown in a (p-vehicle treated: n=7, p-BTKi-treated: n=7, t-vehicle treated: n=4, t-BTKi-treated: n=4). (c) Representative IF images of leptomeninges stained for B220⁺ B cells and CD3⁺ T cells at peak disease. (d) Histological score of B and T cells shown in D-I (p-vehicle treated: n=16, p-BTKi-treated: n=17, t-vehicle treated: n=6, t-BTKi-treated: n=7). Data are shown as means ± SEM. Statistical analysis was conducted using a one-way Anova with multiple comparison. Experiments were performed 3 times with similar results. ns, not significant. p: prophylactic; t: therapeutic

We previously showed that accumulation and organization of B cells in the inflamed leptomeninges is mediated by lymphotoxin β receptor (LTβR)-derived signals in radioresistant stromal cells². LTα and LTβ form a membrane-bound heterotrimer (LTα1β2) that stimulates LTβR which is primarily expressed by stromal cells and myeloid cells. LTβR signaling in fibroblasts is required for lymph node development in utero and can promote TLT in various tissues³¹. Given that TLT were blunted in BTKi treated EAE mice, we hypothesized that BTKi may be indirectly impacting LTβR signaling in stromal cells by interfering with the expression of LTβR ligands in a BTK-sensitive target cell. Strengthening this hypothesis, using qPCR we observed a reduction in Lta and LTb expression in the leptomeninges of BTKi-treated EAE (Fig. 4a).

Fig. 4. BTKi affects LTβR signaling in B cells.

(a) qPCR of Lta and Ltb in the leptomeninges of vehicle–treated or BTKi–treated SJL/J adoptive transfer EAE mice at peak disease. (b-d) Single-cell RNA sequencing of leptomeninges from young (n=2), and old (n=2) SJL/J mice at acute A/T EAE and appropriate age-matched, naïve controls young and old EAE mice. Data shown represent 2 biological and experimental repeats, each with an n of 1 for each group analyzed. (b) Violin plot showing relative gene expression of Btk and lymphotoxin ligand genes (Lta, Ltb) in identified cell clusters stratified by age. (c) Uniform manifold projection (UMAP) of 16,568 leptomeningeal cells after unsupervised clustering. (d) Gene co-expression of Btk with Ltb, Lta projected on the global UMAP. (e) Flow cytometry of the LTβ positive population in CD19+ B220+ B cells from naive SJL/J (n=7) or Ltb^−/− mice (n=7) splenocytes stimulated with mouse anti-CD40 (5ug/ml) + LPS (1ug/ml) (Stim) ex vivo or pre-treated 1 hour with BTKi (10nM). Data in (a) and (e) are shown as means ± SD. Statistical analysis was conducted using two-sided unpaired t test for (a) and two-sided Mann-Whitney for (e). ns, not significant. n = 7 in each group.

We therefore assessed the expression of Btk, Lta and Ltb in the inflamed leptomeninges by mining our previously published scRNAseq database of leptomeninges from young vs old EAE mice at the acute time point¹⁷. We found that the expression of Btk, Lta and Ltb in the leptomeningers was comparable between young and old EAE mice, and that Btk expression was primarily observed in leptomeningeal B cells and to a lesser extent in myeloid cells. Ltb was expressed in B and T cells as well as neutrophils and Ly6c⁺ myeloid cells. In general, Lta expression was less pronounced than Ltb (Fig. 4b) in line with previous data reporting that Lta expression is regulated differently than Ltb, and potentially less stable ³². Importantly, LTa, Ltb and Btk were found to be co-expressed only in B cells (Fig. 4c,d).

Given the unique expression of LT ligands and Btk in B cells in the inflamed leptomeninges, we next performed an in vitro experiment where we pre-treated naive SJL/J splenocytes with BTKi, then stimulated splenocytes with an anti-CD40 + LPS. This in vitro protocol significantly induced upregulation of LTαβ surface protein detected by LTβR-Ig by flow cytometry compared to Ltb^−/− splenocytes (Extended Data Fig. 7). Compared to vehicle treated splenocytes, we found a lower percentage of LTαβ⁺ B cells among stimulated splenocytes treated with BTKi (Fig. 4e). In summary, B cells uniquely express Lta, Ltb and Btk in the inflamed leptomeninges, and upregulation of LTαβ on activated B cells is BTK-dependent.

To determine if LTβR signaling was causally associated with BTK-driven subpial cortical gray matter damage and microgliosis in EAE mice, we administered an LTβR agonist antibody (αLTβR) ³³ at days 9 and 11 post-adoptive transfer of Th17 cells concomitant with BTKi treatment. Combined BTK inhibition and αLTβR treatment reversed the beneficial effects conferred by BTKi alone on disruption of the glial limitans, demyelination, microgliosis and axonal loss in the subpial gray matter (Fig. 5a-d). In summary, these data link BTK with LTβR-mediated signals in mediating subpial grey matter damage.

Fig. 5. BTK inhibition prevents glial limitans disruption, gray matter demyelination, myeloid cell accumulation and axonal loss in the brain in a Lymphotoxin-dependent manner.

Immunohistochemical staining and quantification of somatosensory cortex tissue from vehicle–treated, BTK inhibitor (BTKi) and αHEL–treated, or BTKi and αLTβR-treated SJL/J adoptive transfer EAE mice at peak disease (day 12 after adoptive transfer) for glial fibrillary acidic protein (GFAP) visualizing the glial limitans (a), proteolipid protein (PLP) visualizing myelin (b), ionized calcium binding adaptor molecule 1 (IBA1) visualizing microglia and macrophages (c) and neurofilament light chain (NfL) visualizing axons (d). Zoomed subpial areas are taken at equivalent depth from the pia mater across groups. Data are shown as means ± SD. Statistical analysis was conducted using one-way ANOVA. n = 7 (vehicle and αHEL-treated), 12 (BTKi and αHEL-treated), 7 (vehicle and αLTβR-treated), 8 (BTKi and αLTβR-treated).

BTKi alters leptomeningeal CXCL13 and BAFF.

Passive transfer of encephalitogenic Th17 cells into SJL/J mice results in an approximately 800-fold increase in Cxcl13 that is exquisitely LTβR-dependent². Given that BTKi reduced the expression of LTβR ligands in the leptomeninges, we next wished to determine if BTKi inhibition affects CXCL13 accumulation in this compartment. Staining for CXCL13 protein within TLT revealed significantly reduced leptomeningeal CXCL13 staining in BTKi-treated mice compared to control-treated mice. In contrast, the combination of BTKi with LTβR agonism reversed the CXCL13 reduction induced by BTKi treatment alone (Fig. 6a).

Fig. 6. BTK inhibition reduces accumulation of CXCL13 and boosts the levels of BAFF in the leptomeninges in a Lymphotoxin-dependent manner.

(a) Representative immunohistochemical staining and quantification of CXCL13 in the leptomeninges of vehicle–treated or BTK inhibitor (BTKi) and αHEL–treated, or BTKi and αLTβR-treated SJL/J adoptive transfer EAE mice at peak disease (12 days after adoptive transfer). (b) Quantification of B220⁺ B cells in the leptomeninges of vehicle–treated or BTKi and αHEL–treated, or BTKi and αLTβR-treated SJL/J adoptive transfer EAE mice at peak disease. (c) Correlation between the percentage of B220+ B cell staining and the percentage of CXCL13⁺ staining in the leptomeninges of vehicle–treated or BTKi and αHEL–treated, or BTKi and αLTβR-treated SJL/J adoptive transfer EAE mice at peak disease. (d) Representative immunofluorescent staining of B220⁺ B cells and BAFF in the leptomeninges of vehicle–treated or BTKi and αHEL–treated, or BTKi and αLTβR-treated SJL/J adoptive transfer EAE mice at peak disease. Nuclei are stained with DAPI. (e) Quantification of BAFF in the leptomeninges of vehicle–treated or BTKi and αHEL–treated, or BTKi and αLTβR-treated SJL/J adoptive transfer EAE mice at peak disease. (f) Correlation between the percentage of B220+ B cell staining and the percentage of BAFF⁺ staining in the leptomeninges of vehicle–treated or BTKi and αHEL–treated, or BTKi and αLTβR-treated SJL/J adoptive transfer EAE mice at peak disease. (g) CXCL13:BAFF ratio in the leptomeninges of vehicle–treated or BTKi and αHEL–treated, or BTKi and αLTβR-treated SJL/J adoptive transfer EAE mice at peak disease. Data are shown as means ± SD. Statistical analysis in (a, b, e, g) was conducted using one-way ANOVA. Correlation analysis in (d, f) was conducted using the two-sided non-parametric Spearman correlation coefficient.

Given our previous observation that LTβR signaling is required not only for expression of Cxcl13 but also for the organization of B cells within TLT², we next asked whether CXCL13 levels correlated with accumulation of B cells in the leptomeninges in the different treatment groups (control, BTKi, BTKi + αLTβR). Indeed, BTKi treatment resulted in a reduction in the amount of B220 signal within the TLT that was reversed by αLTβR agonist antibody administration (Fig. 6b), and a significant positive correlation was observed between B220 and CXCL13 staining within TLT (Fig. 6c).

Depletion of B cells with anti-CD20 therapy has been previously shown to augment levels of the B cell survival BAFF in the CSF of EAE mice and MS patients, potentially due to the removal of the main BAFF-consuming cell type (B cells) ²⁷. Staining for BAFF by immunofluorescence revealed that it is expressed at steady state in the LM of naïve mice. However, during EAE, leptomeningeal BAFF expression is significantly reduced and only detected in areas that are relatively poor in B220⁺ B cells. This paucity in BAFF was restored in the leptomeninges by BTKi treatment, but not in mice that received both BTKi and αLTβR agonist antibody treatment (Fig. 6d,e). Overall, a strong anti-correlation between B220 and BAFF staining was observed (Fig. 6f).

The disparate directionality of BAFF vs CXCL13 expression in the leptomeninges in the various treatment groups led us to hypothesize that, on a per-mouse basis, these two molecules were potentially acting in opposition. To assess this, we examined the CXCL13:BAFF ratio for each animal. Vehicle-treated mice with prominent TLT exhibited a high CXCL13:BAFF ratio, whereas BTKi-treated mice with ablated TLT exhibited a low CXCL13:BAFF ratio which was reversed with αLTβR treatment (Fig. 6g). Therefore, BTK inhibition significantly decreases the CXCL13:BAFF ratio in the leptomeninges of EAE mice in an LTβR-dependent manner concomitant with abrogation of TLT formation.

Lastly, we asked if the accumulation of BAFF in the leptomeninges of BTKi treated mice was causally associated with protection against subpial cortical gray matter pathology in EAE mice. To test this, we administered an anti-BAFF mAb (αBAFF) at days 9 and 11 post-adoptive transfer of Th17 cells concomitant with BTKi treatment. Combined BTK inhibition and αBAFF treatment reversed the beneficial effects conferred by BTKi resulting in a restoration of glial limitans disruption, demyelination, microgliosis and axonal loss in the subpial gray matter (Fig. 7a-d). Therefore, BTKi-mediated protection against EAE-induced sub-pial grey matter injury is causally associated with a restoration of BAFF back to homeostatic levels in the leptomeninges.

Fig. 7. BTK inhibition spares the glial limitans and prevents cortical grey matter demyelination, myeloid cell accumulation and axonal loss in young SJL/J adoptive transfer EAE mice in a BAFF-dependent manner.

Immunohistochemical staining and quantification of somatosensory cortex tissue from vehicle–treated, BTKi-treated, BTKi + Isotype–treated and BTKi + anti-BAFF-treated SJL/J adoptive transfer EAE mice at peak disease (day 12 after adoptive transfer) for glial fibrillary acidic protein (GFAP) visualizing the glial limitans (a), proteolipid protein (PLP) visualizing myelin (b), ionized calcium binding adaptor molecule 1 (IBA1) visualizing microglia and macrophages (c) and neurofilament light chain (NfL) visualizing axons (d). Zoomed subpial areas are taken at equivalent depth from the pia mater across groups. Data are shown as means ± SD. Statistical analysis was conducted using one-way ANOVA. n = 4 (vehicle-treated), 4 (BTKi–treated), 5 (BTKi + Isotype-treated), 5 (BTKi + anti-BAFF-treated).

In summary, BTKi treatment reduces leptomeningeal CXCL13 levels, B cell accumulation and TLT formation in an LTβR-dependent manner, corresponding with an increased level in leptomeningeal BAFF which has the capacity to protect against cortical grey matter injury.

High CSF CXCL13:BAFF reflects compartmentalized inflammation in MS.

To test the relevance of our EAE findings in MS, we sourced a cohort of post-mortem brain tissues from MS patients from which matched CSF had also been collected at rapid autopsy (Supplementary Table 1). Importantly, we previously stratified these brain tissues into those with high vs low leptomeningeal B cell counts based on the median level of B cell counts across the cohort, discovering that patients with high numbers of leptomeningeal B cells had larger subpial grey matter lesions as well as chronic active-inactive lesions³⁴. We accessed the matched CSF from this cohort and used a sensitive ELISA-based method to measure the ratio of CXCL13:BAFF in this compartment. We also measured CSF levels of neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP), fluid biomarkers which are used to monitor MS relapsing disease activity^35,36. While CSF NfL and GFAP levels were not significantly different between MS patients who showed evidence of high vs low levels of leptomeningeal B cells (Fig. 8a,b), the ratio of CXCL13:BAFF in the CSF was significantly elevated in MS patients with high leptomeningeal inflammation compared to MS patients with low leptomeningeal inflammation (Fig. 8c).

Fig. 8. An elevated ratio of CXCL13:BAFF in the CSF is associated with compartmentalized inflammation in MS.

Quantification of neurofilament light chain (NfL), glial fibrillary acidic protein (GFAP) and CXCL13 and BAFF by ELISA-based method in CSF collected post-mortem from MS donors with high (n = 25) versus low (n = 22) meningeal CD20⁺ B cell count (a-c) or in CSF collected from a living cohort of MS patients with (n = 13) or without (n = 8) paramagnetic rim lesions (PRLs) (d-f). Data are shown as means ± SD. Statistical analysis was conducted using two-sided Mann-Whitney (a-c) or ANCOVA (d-f).

Encouraged by these data, we next examined levels of NfL, GFAP and the CXCL13:BAFF ratio in the CSF of a cohort of living patients which exhibited MRI-confirmed paramagnetic rim lesion load (PRLs). PRLs are another measure of compartmentalized inflammation which represent the in vivo imaging correlate of histologically-defined chronic active white matter lesions ^37–40 (Supplementary Table 1). We found that while CSF NfL and GFAP levels were not different between MS patients with or without PRLs (Fig. 8d,e), the ratio of CXCL13:BAFF in the CSF was significantly elevated in MS patients with PRLs versus those MS patients without evidence of PRLs (Fig. 8f). In summary, these data indicate that a high ratio of CXCL13:BAFF in the CSF is associated with elevated leptomeningeal B cells and PRLs in MS.

DISCUSSION

The molecular mechanisms that underlie the formation and maintenance of TLTs are remarkably similar to that of secondary lymphoid organs, with both processes involving the lymphotoxin (LT) pathway ^41,42 and the production of B cell homing chemokines such as CXCL13 ^2,43. Here, using remibrutinib, a selective covalent BTKi in Phase 3 for MS ²⁴, our study revealed a leptomeningeal circuitry that drives cortical injury in EAE. We found that the presence of leptomeningeal B cell rich aggregates requires both BTK and Lymphotoxin signaling, and that these aggregates are associated with low levels of BAFF and high levels of CXCL13 within the leptomeninges. In addition, underlying cortical pathology in Th17-induced EAE mice was prevented by BTKi even upon therapeutic treatment of aged mice which normally exhibit non-remitting grey matter injury and brain volume loss¹⁷. In MS patients, a high CXCL13:BAFF ratio in the CSF was associated with a greater degree of compartmentalized inflammation.

BTKi had no significant effect on spinal cord demyelination in our model. This is not surprising since the method of EAE induction is fundamentally different in the passive SJL/J EAE model compared to the active C57Bl/6 model used in other studies^20,22,23, where antigen persists in the form of an adjuvant depot. We hypothesize that since adoptively transferred Th17 cells enter the leptomeninges prior to B cells², the therapeutic impact of B cell depletion²⁷ and BTKi is too late to impact this initial inflammatory acute phase of disease. Indeed, we found that the T cells that infiltrate the leptomeninges in BTKi-treated mice show no appreciable changes in intracellular interferon-γ (IFN-γ), interleukin-17A (IL-17A) or granulocyte-macrophage colony-stimulating factor (GM-CSF) expression compared to vehicle treated mice (Extended Data Fig. 8). This also implies that, once primed, T cells play a minimal role in driving cortical injury. In agreement, scRNAseq of the leptomeninges of aged SJL/J mice, which, compared to young (remitting) SJL/J EAE mice, exhibit severe EAE and cortical shrinkage, revealed very limited differences in gene expression in T cells compared to B cells ¹⁷. In support, we recently showed that B cells are a key determinant of grey matter pathology in this model since anti-CD20 B cell depletion therapy prevents pathology ²⁷.

We previously found that the B cell chemokine Cxcl13 accumulates in the leptomeninges in response to Th17 cells². In line with the high local expression of CXCL13 in the inflamed EAE leptomeninges, CXCL13 has also been detected in the CSF of postmortem MS cases with high levels of meningeal inflammation and grey matter demyelination ⁴⁴. Several proinflammatory cytokines (IFNγ, TNF, IL2, and IL22) and other molecules (CXCL10, LTα, IL6, and IL10) have been detected together with CXCL13 in the CSF ⁴⁴, suggesting an environment of sustained B-cell activity and lymphoid-neogenesis in MS cases with high levels of leptomeningeal inflammation. Here we show that enforced Lymphotoxin signaling can restore high levels of CXCL13 in the leptomeninges in BTKi-treated mice, suggesting that this signal lies downstream of BTK. A positive feedback loop involving CXCR5-mediated induction of LTα₁β₂ expression by B cells is required for optimal B cell organization in the spleen ⁴⁵. We propose that a similar positive feedback loop mediates accumulation of B cells in leptomeninges which indirectly results in an augmentation of CXCL13 production by stromal cells via LTβR signaling.

Interestingly, we noted that the accumulation of CXCL13 and B cells negatively correlated with the amount of BAFF in the EAE leptomeninges. We previously showed that high BAFF levels induced by anti-CD20 treatment are neuro-protective in MS and EAE ²⁷. Moreover, MS disease activity is exacerbated in patients treated with atacicept, a soluble TACI decoy receptor that binds and neutralizes BAFF ⁴⁶. Although treatment with the BTKi does not deplete B cells ⁴⁷ we did nevertheless notice a substantial reduction in the amount and organization of B cells within vestigial TLTs. A potential mechanism of action of BTKi in preventing cortical pathology may therefore be via increased local levels of BAFF. As we showed previously, BAFF may act by promoting survival of stressed neurons²⁷, in agreement with its neuroprotective effects previously described^48,49, although this is likely not the only mechanism of action of BAFF in the leptomeninges. We have previously shown that Tnfsf13b is primarily expressed by myeloid cells in the leptomeninges of SJL/J EAE ¹⁷. Given that B cells are the major consumers of BAFF, a reduction in B cell numbers should increase availability of BAFF, as demonstrated in Wang et al. ²⁷, although whether myeloid cells (or other cells) further contribute to the pool of available BAFF remains possible. Therefore, according to our model, T cells initiate TLT formation whereas B cells maintain TLT via the LT pathway whilst causing pathology due to their consumption of BAFF, which under homeostatic conditions is important for neuroprotection in the underlying grey matter.

Consistent with our model, we noted that in EAE the CXCL13:BAFF ratio was high in mice with leptomeningeal inflammation and cortical injury (vehicle treated, or BTKi treatment with Lymphotoxin beta Receptor stimulation) and low in mice which were spared these outcomes (BTKi-treated without Lymphotoxin beta Receptor stimulation). We therefore hypothesized that the CXCL13:BAFF ratio may be used as a proxy for B cell-rich TLTs in MS. Indeed, we found that the CXCL13:BAFF ratio was high in the CSF of MS patients with high levels of leptomeningeal B cells compared to those with low levels of leptomeningeal B cells. Identifying and tracking meningeal inflammation in living patients - which could be used to monitor the effect of interventions targeting TLTs - is challenging. While contrast-enhanced FLAIR imaging ¹² and more recently 7 Tesla MRI ^50,51 have the capacity to visualize leptomeningeal enhancement, whether this represents TLTs or scar tissue resulting from stromal cell remodelling underpinning these structures ^52,53 remains unclear ⁵⁴.

Using the same post-mortem brain tissue, we had previously shown that the proportion of chronic active white matter lesions positively correlates with the extent of leptomeningeal inflammation ³⁴. The mechanistic link between chronic active white matter lesions and leptomeningeal inflammation could be potentially due to augmented myelin antigen priming of T cells in the periphery that subsequently migrate to the leptomeninges ⁵⁵ or due to retrograde degeneration propagating backward toward cortical neurons, resulting in the formation of grey matter lesions as supported by the link of damage in these two compartments ^56–59. Regardless of the mechanism, B cell-rich leptomeningeal aggregates accumulate in MS patients with a higher proportion of chronic active white matter lesions³⁴. Using this knowledge, we elected to examine the relationship between CXCL13:BAFF ratios in the CSF and PRLs, the radiological equivalent of chronic active white matter lesions ⁴⁰. Remarkably, high CSF ratios of CXCL13:BAFF, but not CSF levels of NfL or GFAP, were associated with the presence of PRLs in MS patients. Thus, we propose that measuring CXCL13:BAFF ratios in the CSF can help identify patients with compartmentalized inflammation which includes both leptomeningeal TLT and PRLs, enabling. better patient stratification and treatment monitoring. By extension, we predict that MS patients who have a high CXCL13:BAFF ratio in the CSF may particularly benefit from treatment with BTKi.

There are some limitations to our study. In addition to B cells, myeloid cells express BTK and could be the target of BTKi^20,60,61. Since our choice of the SJL/J model was predicated on our quest to understand how BTK inhibitors impact cortical grey matter pathology, we are consequently limited by the lack of genetic tools in the SJL/J background that would isolate the impact of BTKi on specific immune cell types in the LM. Ideally an experiment that silences BTK in myeloid vs B cells in vivo would complement our findings. Nevertheless, in the absence of such tools, we can draw some interpretations from our scRNAseq findings: Unlike B cells, myeloid cells do not co-express both Ltb and Lta and thus cannot provide a signal to LTβR expressing cells. Thus, despite our finding that Btk is expressed in leptomeningeal myeloid cells, BTK signaling in myeloid cells is not necessary for leptomeningeal inflammation and cortical pathology since this can proceed in the presence of BTK inhibition when the Lymphotoxin pathway is agonized, although a role for myeloid cell-intrinsic BTK in grey matter pathology independent of LTβR signaling remains possible. We also recognize that our characterization of myeloid cells in tissue do not necessarily distinguish microglial cells from macrophages, which is challenging to do in situ⁶². Lastly, we have not unraveled the signaling pathway that leads to BTK-dependent upregulation of LTαβ in the LM. We previously showed that ICOS-dependent CD40 ligand expression on CD4⁺ T cells triggers the upregulation of LTαβ on B cells in the spleen⁶³. Since we show here that BTKi reduces LTαβ upregulation in response to CD40 stimulation, this suggests that BTKi, which has been shown to operate downstream of CD40^64,65, may promote B cell-intrinsic LTαβ expression by influencing CD40 signaling, although this remains to be tested. Proving this in vivo would be tricky as it would require blockade of CD40 ligand which will likely have other impacts beyond LTαβ expression.

In summary, our study reveals a novel BTK-dependent circuit that establishes B cell rich TLT in the leptomeninges during EAE, demonstrates that dissolving these structures ameliorates grey matter injury in a BAFF-dependent manner, and links high ratios of CXCL13:BAFF in the CSF with compartmentalized inflammation in MS patients. Future studies should characterize the impact of BTKi on CXCL13:BAFF ratios in MS patients and on leptomeningeal/CSF B cell phenotype.

METHODS

Post-mortem human tissue retrieval.

Tissue blocks for this study were obtained from the Netherlands Brain Bank (Amsterdam, the Netherlands). For the characterization of leptomeningeal immune cells, tissue blocks from 22 donors with progressive (primary progressive or secondary progressive) MS were selected based on the presence of leptomeninges adjacent to the cortex in the tissue blocks. Tissue blocks were dissected based on the identification of lesions as guided by macroscopical examination and/or by postmortem MRI (since 2001) of 1 cm thick coronal brain slices ⁶⁵. The tissue blocks used for the analysis of leptomeningeal inflammation and subpial demyelination performed in this study were dissected from the supratentorial cortex at locations that included the occipital or the parietal or the temporal or the frontal lobes. Matched cerebrospinal fluid was obtained from the Netherlands Brain Bank. Detailed clinical-pathological and demographic data of all donors are provided in Supplementary Table 1. The age at the time of death of MS patients ranged from 41 to 89 years (median: 61.5 years), with a mean postmortem delay of 8 h and 14 min (SD, ± 2 h 3 min). The clinical diagnosis of MS and its clinical course were determined by a certified neurologist and confirmed by a certified neuropathologist based on the neuropathological analysis of the patient’s brain autopsy. Tissue samples were obtained from MS patients and controls with approval of the Vrije Universiteit Medical Center Review Board who has reviewed and agreed with the procedures of the Netherlands Brain Bank concerning “Donation of brain material for scientific research”. All patients signed informed consent. Where it concerns persons who, for reasons of their health, are unable to give informed consent (incompetent persons), informed authorization is obtained from the legal representative as defined in the Netherlands Civil Code (Burgerlijk Wetboek). The use of post-mortem study was also approved by the University of Toronto (protocol #30813).

Immunohistochemistry of post-mortem human tissue.

For the classification of cortical gray matter lesions, sections were stained by immunohistochemistry for the PLP marker of myelin. Leptomeningeal immune cells were identified by immunohistochemistry for CD3 to detect T cells and CD20 to detect B cells (see Supplementary Table 2). IHC was performed as previously described ^66,67. Sections of 7 μm thickness were cut from FFPE tissue blocks, collected on Superfrost Plus glass slides (VWR international), and dried overnight at 37°C. Sections were deparaffinized in xylene (2 × 15 min) and rehydrated through a series (100%, 70%, 50%) of ethanol. Endogenous peroxidase activity was blocked by incubation in methanol (Merck KGaA) with 0.3% H₂O₂ (Merck KGaA) for 20 min at 21ºC. Sections were then rinsed in PBS and pretreated with microwave antigen retrieval (3 min at 900 W followed by 10 min at 90 W) in either 0.05 M Tris-buffered saline (pH 7.6) or 10 mM Tris/1 mM EDTA buffer pH 9.0 (see Supplementary Table 2). Sections were incubated overnight at 4°C in the appropriate primary antibody (see Supplementary Table 2) diluted in Normal Antibody Diluent (Immunologic) and the next day with the BrightVision poly-HRP-Anti Ms/Rb/Rt IgG biotin-free (diluted 1:1 in PBS, Immunologic) for 30 min at 21ºC. The immunostaining was visualized with DAB (Vector Laboratories) for 4 min at 21ºC, and sections were counterstained with hematoxylin (Sigma Chemie GmbH), dehydrated in ethanol, and mounted with Pertex (Histolab).

Quantification of subpial demyelination and leptomeningeal inflammation of post-mortem human tissue.

For the quantification of subpial demyelination, brain slides stained for PLP identifying myelin were used. Subpial (type III) lesions — which span from the pial surface typically to cortical layers 2–4 — were imaged at original magnification ×5 using a light microscope (Axioscope), connected to a digital camera (AxioCam MRc), and Zen pro 2.0 imaging software. The RGB images were separated into single-color channels using the color deconvolution plugin in ImageJ Pro Plus 7.0 imaging software (MediaCybernetics). The single-color channel for PLP was subjected to thresholding to create a mask that captured the specific staining. The area fraction measurement was applied to each image to quantify the percentage of thresholded staining. The scale represents the percentage of PLP⁺ subpial cortical area in each brain section examined. For the quantification of leptomeningeal inflammation, leptomeningeal segments were randomly selected for imaging at 20× original magnification with a light microscope (Olympus BX41TF) connected to the Cell D software (Olympus). Immune cells were quantified in leptomeningeal areas that were adjacent to type III (subpial) gray matter lesions (GMLs) and in areas that were adjacent to normal appearing gray matter (NAGM). A total of 71% ± 20% (mean ± SD) of intact leptomeninges were available for scoring in the MS cohort. CD20⁺ B cell counts were done over a total leptomeningeal area of 47.53 mm² from MS patients, of which 37.02 mm² was adjacent to GMLs and 10.51 mm² was adjacent to NAGM; and 5.652 mm² of leptomeningeal area was adjacent to non-neurological control cortex. The leptomeningeal area (in mm²) was measured using the “measurement” function of the Image Pro Plus 7.0 imaging software. Cell numbers were expressed as mean number per mm of intact leptomeninges. Considering the large dynamic range of the meningeal count for CD20+ B cells (range: 0.89–16.27 cells/ mm meninges) in MS cases, the median was chosen as a cutoff to stratify the MS cases in those with high (above the median) or low (below the median) meningeal lymphocyte count (7.30 cells/mm for B cells) ⁶⁸. Since B cells are a proxy of leptomeningeal inflammation ⁶⁹, the B cell cutoff was used in this study to stratify donors with high vs low leptomeningeal inflammation.

NIH cohort and metadata

The Ella-based proteomic data for the NIH cohort were collected following the approval of the institutional review board, and after written consent was obtained as part of the National Institute of Neurological Disorders and Stroke’s “Evaluation of Progression in Multiple Sclerosis by Magnetic Resonance Imaging (MRI)” protocol (NCT00001248). All participants provided informed consent prior to inclusion. Per this protocol, participants were seen in the clinic, undergo imaging and biospecimen collection. Experienced MS clinicians documented clinical history and determined the EDSS. We performed Ella assays on the cerebrospinal fluid (CSF) of these participants. These included 21 cases (Supplementary Table 1), all of whom were radiologically and clinically inactive at the time of sample collection. Radiologically and clinically inactive was defined as the absence of any new and/or gadolinium-enhancing lesion(s), and absence of clinical relapse respectively within 6 months of sample acquisition.

In vivo MRI and PRL determination

For the NIH cohort, in vivo MRI acquisition for participants involved using a Siemens Magnetom 7T scanner equipped with a bird-cage-type transmit coil and a 32-channel receive coil. The imaging protocol included a pre-contrast, high resolution 3-dimensional (3D) gradient dual-echo sequence for acquiring T2*-weighted (T2*) and phase contrast images (repetition time [TR]: 74 milliseconds; echo times [TE]: 18, 29.5, 50.0 and 52.4 milliseconds; flip angle [FA]: 10°; acquisition time [AT]: 11 min 50 seconds per 30 mm slab; 0.5 mm isotropic resolution). Axial T2*-weighted images were motion- and B0-correcged. Magnitude and unwrapped, filtered phase images were obtained as well. In addition, a T2-weighted FLAIR (fluid-attenuated inversion recovery) sequence was obtained for visualization of white matter lesions.

The paramagnetic rim lesions (PRLs) were determined by an experienced neuroradiologist based on the NAIMS consensus criteria ⁷⁰. To qualify as a PRL, the lesion must be identifiable on T2* or FLAIR and exhibit a discrete paramagnetic rim signal on susceptibility-sensitive sequences including phase. The paramagnetic rim signal should be continuous through at least two-thirds of the outer edge of the white matter portion of the lesion, and is observed as a hypo-intensity. DSR rated all the PRL in the NIH cohort.

Ella simple-plex immunoassay of human CSF

The amount of NfL, CXCL13 and BAFF in the CSF collected post-mortem from people with MS (See Supplementary Table 1) was quantified using Ella simple-plex immunoassay (Bio-Techne), which has been previously validated for the measurement of analytes in MS biofluids ⁷¹. NfL, CXCL13 and BAFF levels were measured using a custom Simple Plex cartridge (Bio-Techne catalogue numbers SPCKC-PS- 008065; SPCKB-CS- 010515; ST01B-PS- 002448) on Ella™ instrument, according to the manufacturers’ instructions. Ella™ was calibrated using the in‐cartridge factory standard curve. All samples were measured in triplicate, with a 1:2 dilution.

Mice.

Female 6- to 10-week-old (young) SJL/J mice were obtained from Envigo and housed at the University of Toronto animal facilities under specific pathogen–free conditions, in a closed caging system with a 12-hour light/12-hour dark cycle at 21ºC with 50% relative humidity. They were provided with a standard irradiated chow diet (Teklad; Envigo, 2918) and acidified water (reverse osmosis and ultraviolet sterilized). For experiments involving old mice, 6 to 10-week-old SJL/J mice were purchased and aged at our animal facility until they were 8 to 12 months-old. LTβ^−/− mice were originally from B&K Universal and bred in our animal facility. All animal experiments were conducted with ethical approval (protocol #20011363) from the University of Toronto, Faculty of Medicine animal care committee.

Induction and clinical assessment of experimental autoimmune encephalomyelitis (EAE).

Adoptive transfer (A/T) experimental autoimmune encephalomyelitis (EAE) in young (6 to 10 weeks) or old (8 to 12 months) SJL/J mice (SJL/J A/T EAE) was induced as previously described ⁷². Briefly, SJL/J donor mice were immunized with 100 μg PLP_139–151 (HSLGKWLGHPDKF; Canpeptide) in an emulsion of incomplete Freund’s adjuvant (BD Difco), supplemented with 200 μg Mycobacterium tuberculosis H37 Ra (BD Difco, 231141) in a total volume of 300 μL administered as three 100 μL subcutaneous injections on the back and flanks. On day 9 after immunization, cells from the spleen and draining lymph nodes (inguinal, cervical, axillary, and brachial) were harvested. Lymphocytes were separated using Percoll gradient and restimulated ex vivo with PLP_139–151 (10 μg/mL) in the presence of anti–IFN-γ (20 μg/mL, Bioceros), anti–IL-4 (20 μg/mL, Bioceros), and IL-23 (10 ng/mL, R&D Systems) for 72 h at 37°C. In total, 1 10⁷ cells were injected intraperitoneally into SJL/J recipient mice.

Recipient mice were weighed and scored daily according to a composite scale that we ^72,73 and others ⁷⁴ have previously published. Briefly, the composite scale measures mobility impairments in each limb and the tail. Each limb is graded from 0 (asymptomatic) to 3 (complete paralysis), and the tail is graded from 0 (asymptomatic) to 2 (limp tail). Assessment of the righting reflex is scored from 0 to 2, with 0 being assigned for a normal righting reflex, 1 for slow righting reflex, and 2 for a delay of more than 5 s in the righting reflex. Each criterion was measured in 0.5 increments. Thus, the composite score ranges from 0 (nonsymptomatic) to 16 (fully quadriplegic mouse with limp tail and significantly delayed righting reflex) ^74,75.

Bruton’s tyrosine kinase (BTK) inhibitor and ɑLTβR or anti-BAFF treatment.

Remibrutinib (LOU064) was synthesized by Novartis. The drug was administered in SJL/J recipient mice twice a day by oral gavage at a dose of 30 mg/kg, diluted in sterilized Methylcellulose/Tween80 vehicle (Fujifilm Wako Chemicals and Bioshop respectively) in an application volume of 10 ml/kg. For young mice, the drug was administrated from day 3 and up to day 11 post-adoptive transfer in the prophylactic modality or from day 9 and up to day 13 post-adoptive transfer in the therapeutic modality. Control mice received Methylcellulose/Tween 80 vehicle solution twice a day by oral gavage at the same time points after adoptive transfer. For experiments including ɑLTβR treatment, clone AFH6 (Kind gift of Dr. Ray Jupp, Mestag Therapeutics) and isotype control (ɑHEL; IgG1 antibody) were administered intraperitoneally on D8 and D12 post-A/T (100μg/mouse). On days where the ɑLTβR/ɑHEL were to be administered along with the BTKi (on D12 post-A/T), ɑLTβR/ɑHEL were administered first prior to vehicle or BTKi treatment. For experiments including anti-BAFF treatment, anti-BAFF antibody (Sandy-2, Adipogen) or mouse IgG1 isotype controls (SouthernBiotech) was delivered intraperitoneally at a dose of 2 mg/kg on days −2 and 5 post adoptive transfer. For old mice, the drug was administrated from day 9 to day 21 for the histology readout or from day 9 to day 39 for the MRI readout. Mice were randomized into either the treatment or control groups.

Tissue harvest and processing.

For histology/immunohistochemistry and serum neurofilament light (NfL) measurements at acute and post-acute disease phases, mice were euthanized at the peak phase of disease corresponding to 12–14 days post-adoptive transfer or at the post-acute phase corresponding to 22 or 40 days post-adoptive transfer, depending on the readout. At the time of sacrifice, blood was collected via cardiac puncture and serum was separated and stored at −80°C until used. Mice were intracardially perfused with 30 ml of PBS. Brains and spinal cords were dissected from the skull and spinal columns. The 2 brain hemispheres were sagitally divided. One hemisphere was postfixed together with the spinal cord in 10% buffered formalin (MilliporeSigma) for 1 week at 21ºC prior to being processed into paraffin for pathological analysis. The other hemisphere was postfixed in 10% buffered formalin for 1 hour at 21ºC followed by immersion in 30% sucrose overnight at 4°C, then placed in plastic molds with optimal cutting temperature mounting medium (Thermo Fisher Scientific), frozen in a bath of 2-O-methylbutane on dry ice (Fisher Chemical, Thermo Fisher Scientific), and stored at –80°C until they were sectioned for staining.

To test for the BTK occupancy, a subset of BTKi-treated mice and vehicle-treated mice were euthanized at peak disease and transcardially perfused. Spleen and brains were collected and frozen for later measurements.

BTK occupancy.

Meso Scale Discovery (MSD)-based sandwich immunoassays for free and total BTK were performed as described in ⁷⁶. Briefly, for free BTK measurements, samples were incubated with a biotinylated covalent BTK probe, then samples were added to streptavidin ELISA plates to allow binding of probe-bound BTK. The binding of the probe to BTK is mutually exclusive with compound binding to BTK. Plate-bound BTK was detected with an anti-BTK antibody (D3H5, Cell Signaling Technology). For total BTK measurements, an ELISA plate was coated with D3H5 anti-BTK to capture total BTK. An anti-BTK antibody (no. 53, BD Biosciences) directed to a different epitope was then used to detect captured BTK. The respective free BTK levels for each sample were normalized to the total BTK level in the same sample, and these ratios were expressed as percentage of the vehicle control samples.

Preparation of samples for MRI.

Mice destined for MRI were sacrificed by CO₂ asphyxiation and transcardially perfused using a peristaltic pump at a rate of 1 mL/min. Mice were first perfused with 40 mL of PBS containing 2 mM ProHance (Bracco Diagnostics) and 400 USP heparin (Fresnius Kabi), followed by 30 mL of PBS containing 2 mM ProHance and 4% paraformaldehyde (EMS). Skulls were decapitated and placed into PBS containing 2 mM ProHance and 4% paraformaldehyde (EMS). After an overnight incubation at 4°C, skulls were transferred to PBS containing 2 mM ProHance with 0.02% sodium azide (Fisher Scientific). Following 30-day incubation, skulls were scanned for MRI at the Mouse Imaging Centre in The Centre for Phenogenomics in Toronto, Ontario, Canada.

Anatomical image acquisition.

A 7-Tesla 306 mm horizontal-bore magnet (BioSpec 70/30 USR, Bruker) with a ParaVision 6.0.1 console ⁷⁷ was used to image brains in skulls. Four to eight samples were imaged in parallel using a custom-built 8-coil solenoid array. Anatomical image acquisition was modeled from Spencer Noakes et al. ⁷⁸ with the following scan parameters: T2W 3D FSE cylindrical k-space acquisition sequence, TR/TE/ETL = 350 ms/12 ms/6, TEeff = 30 ms, 4 effective averages, FOV/matrix size = 20.2 × 20.2 × 25.2 mm/504 × 504 × 630, total imaging time = 13.2 h. The resulting anatomical images had an isotropic resolution of 40 μm voxels.

MRI registration and analysis.

To assess any changes to the mouse brains due to age and treatment, all anatomical brain images were registered together using the mni_autoreg ⁷⁹ and Advanced Normalizations Tools ⁸⁰ tool kits. The resulting consensus average and Jacobian determinants were used to quantifying volumetric differences between each MRI image and the average. The MAGeT pipeline ⁸¹ was used to segment images using a published classified MRI.

Single molecular array (SIMOA) assay for serum neurofilament light chain (NfL).

The amount of NfL in mouse serum was quantified with a single-molecule array (Simoa) NF-light assay (Quanterix). In brief, magnetic beads were conjugated with monoclonal capture antibodies (mAB47:3, UmanDiagnostics), then incubated with diluted mouse serum (1:8 or 1:16 dilution) and biotinylated detection antibodies (mAB2:1, UmanDiagnostics). Upon adding streptavidin-conjugated β-galactosidase (Quanterix), Resorufin β-D-galactopyranoside (Quanterix) was added for detection. The experiment was performed on a Simoa HD-X Analyzer (Quanterix). The assay was performed in duplicates, and the mean of the 2 measured serum NfL values per sample is reported.

Single-cell RNA sequencing.

Data from our previously published leptomeninges single cell dataset (GSE201568 ⁸²) was loaded into RStudio (v2023.03.0) and analyzed using the Seurat package (v5.1.0 ⁸³). After quality control, data scaling and normalization, unsupervised clustering of 16,568 cells was performed, resulting in identification of 11 unique cell clusters. Using canonical lineage markers, we were able to identify B cells (Ms4a1, Cd19, Ighm, Ighd), Cd8+ T cells (Cd3, Cd8a), Cd4+ Th1 cells (Cd3e, Cd4, Tbx21), Cd4+ Th17 cells (Cd3e, Cd4, Rorc, Il17a), neutrophils (Cxcr2, Ly6g, Itgam), Ly6c+ myeloid cells (Itgam, Ly6c2, Csf1r, Aif1), Microglia/macrophages (Itgam, Aif1, Hexb, Tmem119, P2ry12), DCs (Itgax, Flt3, Ccr7), stromal cells (Col1a1, Pdgfra, Fn1, Pdpn), and endothelial cells (Pecam1, Cd34). Visualization of gene expression and co-expression on individual clusters was performed using the VlnPlot and FeaturePlot functions.

Dissection of the leptomeninges for T cell stimulation and flow cytometry.

Under a dissection microscope, leptomeninges were removed from the brainstem, cerebellum, ventricles, hypothalamus, and cortex into digestion buffer (RPMI 1640 with 10 mM HEPES, 1% Pen-Strep, 1% GlutaMAX, 2% FBS pH 7.4) containing DNAse I (60 μg/mL) and collagenase P (1 mg/mL). Collected cells from leptomeninges were washed twice in ice-cold PBS and resuspended in complete RPMI (10% FBS from Gibco, Thermo Fisher Scientific; l-glutamine from MilliporeSigma, sodium pyruvate from MilliporeSigma, penicillin from MilliporeSigma, streptomycin from MilliporeSigma, HEPES pH 7.0 from Gibco, Thermo Fisher Scientific; and β-mercaptoethanol from Gibco, Thermo Fisher Scientific, in RPMI-1640 medium from MilliporeSigma). Cells were counted with a hemocytometer and incubated at 37°C for 5 h with an intracellular cytokine restimulation buffer (PMA [MilliporeSigma, stock 500 μg/mL] used at 1:100,000; ionomycin [MilliporeSigma, stock 0.5 mg/mL] used at 1:1000; and Brefeldin A [eBioscience, Thermo Fisher Scientific, stock 100x] used at 1:1000 in complete RPMI). At the end of the incubation, cells were collected, washed twice in ice-cold PBS, and resuspended in PBS with 2% FBS and stained with LIVE/DEAD Fixable Aqua (Life Technologies, Thermo Fisher Scientific) at 1:1000 in PBS for 30 min at 4°C, followed by staining with anti–CD4- (mAb, RM4–5), and anti–CD3- (mAb, 17A2) in PBS with 2% FBS and 1 mg/mL Fc block. Cells were then permeabilized with CytoFix/CytoPerm (BD Biosciences) for 20 min at 4°C, then stained with anti–GM-CSF-FITC (mAb, MP1– 22E9), anti–IL-17A-( mAb, 17B7), and anti–IFN-γ- (mAb, XMG1.2) in 1× Perm/Wash buffer with 1 mg/mL Fc block (see Supplementary Table 2). Cells were acquired by FACS using the BD LSR X-20 II instrument and FACSDiva software (BD Biosciences). Data analysis was done with FlowJo V10.5.3 software.

In vitro culture of splenocytes and LTβ staining by flow cytometry.

SJL/J and LTβ^−/− mice were euthanized by CO₂ asphyxiation and spleens were collected and mashed through 70-μm filters into culture medium [RPMI 1640 (Sigma-Aldrich) supplemented with 10% fetal bovine serum (Gibco), 1× penicillin-streptomycin (Sigma-Aldrich), 1× L-Glutamin (Sigma-Aldrich)]. Splenocytes were resuspended in 5mL of RBC lysis buffer (eBioscience 00–4333-57) for 5 min to remove red blood cells. Cells were washed and resuspended in culture media at 2×10⁶/mL. 200ul of cell suspension /well were plated in 96-well U bottom culture plates. Cells were placed in the incubator at 37°C 5%CO2 for 1 hour to rest, then BTK inhibitor (or culture medium) was added at a concentration of 10nM for 1 hour. InVivoMAb anti-mouse CD40 (BioxCell BE0016–2) (5ug/ml) + LPS (Sigma-Aldrich L2880) (1ug/ml) were added to the culture overnight (18–20 h). Cells were harvested, washed with PBS and stained with Live/Dead Fixable Aqua (Invitrogen) at a dilution of 1:1000 for 20 min on ice. Surface markers were stained in 2% fetal bovine serum in PBS for 30 min on ice (Supplementary Table 2). Samples were acquired on a BD FACSymphony™ A3 using the FACSDiva software. Flow cytometry data were analyzed using FlowJo v10.0.

Quantitative polymerase chain reaction (qPCR) on Leptomeninges.

Leptomeninges were collected into digestion buffer (RPMI 1640 with 10 mM HEPES, 1% Pen-Strep, 1% GlutaMAX, 2% FBS pH 7.4) containing DNAse I (60 μg/mL) and collagenase P (1 mg/mL). RNA was extracted with Qiagen RNeasy Microkit (Qiagen # 74004). RNA concentration and quality were assessed by NanoDrop. cDNA synthesis was performed using Invitrogen SuperScript IV First-Strand Synthesis System, and cDNA was subjected to qPCR reactions with Rpl19 as internal control and using the following primers for LT-α: Forward primer: AGCCCATCCACTCCCTCAGAAG, Reverse Primer: TGCTCTCCAGAGCAGTGAGTTC; LT-β: Forward primer: CCTGTTGTTGGCAGTGCCTATC, Reverse Primer: GACGGTTTGCTGTCATCCAGTC (Integrated DNA Technologies)

Statistics

All statistical tests were run using GraphPad Prism (version 10.2.1). All quantification data were subjected to Shapiro-Wilk normality test for equality of variances. Only P values less than 0.05 were considered significant. The statistical tests used for each experimental dataset are indicated in the Fig. legends. All murine experiments were performed at least three times with reproducible results. In all plots, each dot represents a measurement from a distinct sample. No statistical methods were used to pre-determine sample sizes but our sample sizes are similar to those reported in our previous publications^68,73,82,84.

Extended Data

Extended Data Figure 1. Prophylactic BTK inhibition does not affect clinical outcome or serum neurofilament light (NfL) levels in SJL/J adoptive transfer EAE mice.

(a) Female 6- to 10-week-old SJL/J mice received an adoptive transfer of 10 million encephalitogenic T_H17 cells to induce EAE. Mice were treated via oral gavage with 30 mg/kg Bruton’s tyrosine kinase inhibitor (BTKi, Remibrutinib, (LOU064) or vehicle control twice a day prophylactically from day 3 to day 11 after adoptive transfer and were harvested at peak EAE (day 12 after adoptive transfer). (b, c) BTK occupancy in the spleen and brain of vehicle-treated (n = 5 for each) or BTKi-treated (n = 7 for each) SJL/J adoptive transfer EAE mice at peak disease. (d-g) Clinical scores and area under the curves (AUC) of the clinical scores in 2 experiments where SJL/J adoptive transfer EAE mice were treated with the BTK inhibitor (n = 10 and n=7) or vehicle control (n = 10 and n=6) prophylactically. (h) Serum neurofilament light chain (NfL) in naïve (n = 4), vehicle-treated (n = 4) and BTKi-treated (n = 7) mice at peak disease. (i) Correlation between the AUC and the amount of serum NfL in vehicle–treated or BTKi–treated (n=8) SJL/J adoptive transfer EAE mice at peak disease. Data are shown as means ± SD. Statistical analysis in b, c, e, g was conducted using a two-sided Mann-Whitney test. Statistical analysis in H was conducted using a two-sided Kruskal-Wallis test with post hoc Dunn’s multiple comparisons test. Statistical analysis in d, f was conducted using a One-Way ANOVA test with post hoc Bonferroni multiple comparisons test. Correlation analysis in i was conducted using two-sided Pearson correlation coefficient. *P ≤ 0.05 and **P ≤ 0.01. (b):p=0.002, (c):p=0.0051, (h): naïve vs vehicle-treated p= 0.0216 and naïve vs BTKi-treated p= 0.0403.

Extended Data Figure 2. Prophylactic BTK inhibition do not affect spinal cord pathology in SJL/J adoptive transfer EAE mice.

Histology and immunohistochemical staining of spinal cord tissue from naïve (a), prophylactically vehicle–treated (b) and BTK inhibitor (BTKi)–treated (c) SJL/J adoptive transfer EAE mice at peak disease (day 12 after adoptive transfer) for H&E visualizing inflammation, luxol fast blue (LFB) visualizing myelin, ionized calcium binding adaptor molecule 1 (IBA1) visualizing microglia and macrophages and neurofilament light chain (NfL) visualizing axons. Black arrowheads indicate the inflamed and damaged spinal cord white matter area. (d) Quantification of a-c. Data are shown as means ± SD. Statistical analysis was conducted using Kruskal-Wallis tests with post hoc Dunn’s multiple comparisons test. P > 0.05. (naïve: n= 5, 5, 7 and 7 respectively), (vehicle-treated: n=10 for each graph), (BTKi–treated: n=10 for each graph).

Extended Data Figure 3. Therapeutic BTK inhibition spares the glial limitans, reduces cortical grey matter demyelination, myeloid cell accumulation and axonal loss in SJL/J adoptive transfer EAE.

Immunohistochemical staining of somatosensory cortex tissue from naïve, therapeutically vehicle–treated and BTK inhibitor (BTKi)–treated SJL/J adoptive transfer EAE mice at peak disease (day 12 after adoptive transfer) for glial fibrillary acidic protein (GFAP) visualizing the glial limitans (a), proteolipid protein (PLP) visualizing myelin (b), ionized calcium binding adaptor molecule 1 (IBA1) visualizing microglia and macrophages (c) and neurofilament light chain (NfL) visualizing axons (d). White arrowheads in a indicate the glial limitans; in c indicate microglia/macrophages.

Extended Data Figure 4. Therapeutic BTK inhibition does not affect clinical outcome in old SJL/J adoptive transfer EAE mice.

a) Female 8 to 12 moths-old SJL/J mice received an adoptive transfer of 10 million encephalitogenic T_H17 cells to induce EAE. Mice were treated via oral gavage with 30 mg/kg Bruton’s tyrosine kinase inhibitor (BTKi, Remibrutinib, LOU064) or vehicle control twice a day therapeutically from day 9 to day 21 post-adoptive transfer (b, c) when the tissue was harvested for pathology assessment or day 39 post adoptive transfer (d, e) when the tissue was harvested for MRI assessment. Timeline and area under the curves (AUC) of the clinical scores from the pathology (b, c; n = 5 for BTKi treated and n=5 for vehicle treated mice) and the MRI (d, e; n=7 for BTKi treated and n=7 for vehicle treated mice) experiments. Data are shown as means ± SD. Statistical analysis in B and D was conducted using a two-sided One-Way ANOVA test with post hoc Bonferroni multiple comparisons test. Statistical analysis in C and E was conducted using a two-sided Mann-Whitney test.

Extended Data Figure 5. Effects of prophylactic or therapeutic BTK inhibition on leptomeningeal TLTs in SJL/J adoptive transfer EAE mice.

Representative hematoxylin & Eosin (H&E)–stained sections of the leptomeninges overlaying the hippocampi (a, b), hypothalami (c, d), cerebelli (e, f) and brainstems (g, h) of vehicle–treated or BTK inhibitor (BTKi)–treated SJL/J adoptive transfer EAE mice at peak disease (day 12 after adoptive transfer). Quantification of the number and size of leptomeningeal TLTs shown in A-H for the prophylactic (i) or therapeutic (j) treatment modality. Data are shown as means ± SD. Statistical analysis was conducted using two-sided multiple Mann-Whitney tests. ns, not significant; *P ≤ 0.05. (vehicle-treated: n=7 for each treatment modality), (BTKi–treated: n=7 for each treatment modality).

Extended Data Figure 6. Effects of prophylactic or therapeutic BTK inhibition on leptomeningeal B and T cells in SJL/J adoptive transfer EAE mice.

Histological score of B220⁺ B cells and CD3⁺ T cells in the leptomeninges of prophylactically (a) or therapeutically (b) vehicle–treated or BTK inhibitor (BTKi)–treated SJL/J adoptive transfer EAE mice at peak disease (day 12 after adoptive transfer). Data are shown as means ± SD Statistical analysis was conducted using multiple Mann-Whitney tests. ns, not significant; *P ≤ 0.05. (a) B220, p=0.011994; CD3, p=0.003767. (b) B220, hippocampus p=0.020979; hypothalamus p=0.028571; cerebellum p=0.011072. CD3, cerebellum p=0.049534. n =7 (vehicle-treated), 7 (BTKi–treated).

Extended Data Figure 7. Gating strategy for B cells treated with BTKi.

(a) Gating strategy for B cells. (b) Representative plots of LTβ⁺CD19⁺B220⁺ B cells from naive SJL/J splenocytes either unstimulated or stimulated with mouse anti-CD40 (5ug/ml) + LPS (1ug/ml) ex vivo after pre-treatment with BTKi (10nM) or equivalent Vol of culture medium for 1 hour.

Extended Data Figure 8. Prophylactic BTK inhibition does not impact the pro-inflammatory profile of T cells in the leptomeninges of SJL/J adoptive transfer EAE mice.

Flow cytometry quantification of interferon (IFN)-γ (a), interleukin (IL)-17A (b) and granulocyte-macrophage colony-stimulating factor (GM-CSF) (c) production by CD3⁺CD4⁺ T cells from single cell suspensions isolated from the leptomeninges and stimulated for 4 hours in 1X Cell Stimulation Cocktail with 1X Brefeldin A. (d) Gating strategy of a-c. Data are shown as means ± SD .Statistical analysis was conducted using a two-sided Mann-Whitney test. ns, not significant. n = 5 (vehicle-treated), 7 (BTKi–treated).

DATA AVAILABILITY STATEMENT

All data supporting the findings of this study are available within the paper and its Extended Data Information.