2D2xTh EAE mice – which develop spontaneous chronic EAE and meningeal inflammatory lesions in the spinal cord - were generated, held and clinically assessed as previously described (16–18). All animal procedures were approved by the standing committee for experimentation with laboratory animals of the administration of Upper Bavaria (ROB-55.2-2532.Vet_02-16-100).

Six mice were initially used in the current experiment. Mice were randomly assigned to either the treatment or control group. The treatment was initiated when a mouse reached an EAE score ≥ 3 and was conducted as previously described (16). In brief, B cell depletion was achieved by weekly injection of 100 μg murine aCD20 mAbs (clone 18B12, provided by Roche). In vivo grade mouse IgG2a kappa isotype control (Crown bio, Cat: C0006) was used in the control group. One mouse from the treated group was excluded from further analysis due to insufficient effect of the aCD20 treatment. Figure 1A shows the experimental design. Table 1 lists the clinical characteristics of the 5 mice included in the final analyses and Figure 1B their clinical disease course.

Mice from both groups were sacrificed 30-35 days after the first injection. Animals were first anesthetized, after which CSF was obtained by puncture of the cisterna magna and stored on ice. Subsequently mice were sacrificed. After thorough perfusion with ice cold PBS, the spinal cord was dissected. A small test piece was cut from the thoracolumbar part of the spinal cord, which was used to histologically confirm the presence of mELT. Meninges were gently separated from the spinal cord in a petri dish with PBS and, using fine forceps, cut into small pieces under a dissection microscope. Subsequently meninges were digested in 600 µl digestion mix consisting of 4µl DNase, 60µl Collagenase IV and 536µl RPMI Medium for 30 minutes at 37°C on a shaker. Cells were isolated from the digested meninges by passing through 30 µl cell strainers, and washed. Cells from meninges and CSF were finally centrifuged (450xg, 5 min, 4 degrees) and resuspended in 100 µl FACS Buffer.

A brief incubation (10 min, 4°C) with Fc Block reagent was performed to block nonspecific antibody binding. Next, samples were labelled with DNA barcoded TotalSeq antibodies against mouse CD4, CD8a, CD19, Ly6G, and NK1.1, as previously described (11). At the same time, samples were incubated with 2 different TotalSeq Hashtag antibodies, in order to allow pooling samples per two: CSF und meningeal cells were incubated (20 minutes at 4° in the dark) with TotalSeq-C 0301 anti-mouse #1 (clone M1/42; 30-F11, Biolegend, dilution 1:400 in FACS buffer) and TotalSeq anti-mouse 0302 #2 (clone M1/42;30-F11, Biolegend, dilution 1:400 in FACS buffer), respectively. The incubation was followed by 3 wash steps with FACS Buffer and cell counting (Cellometer Auto 2000, Nexcelom Bioscience). Ultimately, the cell concentration was adjusted to 1x10³ cells/µl in RPMI/10% FCS, and CSF and meningeal cells were pooled together in a 1:4 ratio per mouse.

Cells were loaded onto 10x Genomics Chromium Next GEM Chip G and the Chromium Controller was run according to the manufacturer’s instructions (10x Genomics, USA) with a targeted cell recovery of 10,000 cells. Library preparation of Cell Surface Protein Library (CSP Library) and Gene Expression Library (GEX Library) was performed using Chromium Next GEM Single Cell V(D)J Reagent Kits v1.1, following the 10x Genomics protocol CG000208. The quality of library preparation was assessed using the Agilent 2100 Bioanalyzer.

Real-Time PCR was performed to quantify the libraries. All single cell gene expression, BCR and cell surface protein (CSP) libraries were multiplexed and sequenced using Illumina NovaSeq 6000 (S2 Flow cell), to obtain 26 x 91 bp paired-end reads for gene expression, cell surface protein and BCR libraries, taking into account the differences in read depth requirements. Novaseq sequencing was performed at the Helmholtz Zentrum München (HMGU) by the Genomics Core Facility.

We aimed to examine and compare the immune cell composition of CSF and mELT. Furthermore, we aimed to analyze the cellular differences in these tissues between the aCD20 treated (B cell depleted) and the control treated (B cell competent) mice. After filtering out low quality cells and corrections using “harmony” library, we obtained the transcriptome from a total of 9,585 single cells: 5,445 cells from mELT and 4,140 cells from CSF. We performed clustering and, in this way, classified the cells into 11 final cell clusters (Figure 2). Based on a combination of automatic cluster annotation (CIPR version 0.1.0, “immgen” database) and marker gene expression (see Methods section), we identified the following 11 cell types: T cells (Cd3e), more specifically naïve CD4+ T cells (Cd4, Ccr7, Sell), activated CD4+ T cells (Cd4, Rora, Pdcd1, Il2), CD8+ T cells (Cd8a), and regulatory T cells (Foxp3), B cells (Cd79a, Cd19) and plasma cells (Sdc1, Tnfrsf17), a cluster containing natural killer (NK) cells and NKT cells (Klrb1c, Klrk1), myeloid lineage cells, separated into dendritic cells (DC) (Flt3), macrophages (Cd68, Adgre1), monocytes (Cd68), and granulocytes (S100a8 and S100a9). Erythroid cells (Hbb-bt) and oligodendrocytes (Olig1), probably coming from blood or CNS contamination respectively, were not considered in further analyses.

Figure 3 and Table 2 illustrate the cell composition of mELT and CSF from control treated (B cell competent) mice. Both CSF and mELT consist mainly of B and CD4+ T lymphocytes. Their relative numbers (percentage) were comparable in both tissues (Table 2). Other cell types like monocytes, plasma cells, NK cells, macrophages and granulocytes are also equally represented in both CSF and mELT from control treated mice. Overall, regarding the immune cell composition, CSF and mELT were closely related in control treated 2D2xTh EAE mice.

As previously shown (16), aCD20 treatment did not alter the EAE course (Figure 1B). The cellular composition of B cell depleted, aCD20 treated CSF seems to be considerably different compared to the B cell competent control group. Due to insufficient data for the mELT aCD20 condition (Table 2), we compared CSF from aCD20 treated versus control treated mice as a surrogate (Figure 4, Table 2). Like previously shown for mELT (16), aCD20 treatment depleted virtually all B cells from CSF. In addition, aCD20 treated CSF contained less naïve CD4+ T-cells. Similarly, we observed a reduction of monocytes, granulocytes, and plasma cells, but their number was relatively low overall. Interestingly, we observed a substantial increase of macrophages in CSF samples from the aCD20 treated group when compared to the control treated group (1,277 vs. 227 cells; 62,5% vs. 9,6%). Numbers of activated CD4+ T cells, NK cells, Tregs, CD8+ T cells were similar in both conditions.

In order to identify differences in gene expression in CSF between the aCD20 and isotype conditions, a differential gene expression analysis was conducted for each cell cluster within the respective tissue samples. Only those cell types were included in the analysis, where >50 cells were available in both conditions. Figure 5 illustrates the 10 genes with highest fold change among the significant upregulated and downregulated genes in CSF from aCD20 treated mice compared to the control treated group. A full list of significant genes can be found in Supplementary Table 1.

Differential gene expression analysis showed that the majority of differentially regulated genes was upregulated in the aCD20 treated group in comparison to the control group (Figure 6). One of the most differentially expressed genes, upregulated in the aCD20 treated group in each of the examined cell types, was Ferritin Heavy Chain 1 (Fth1), which has a major function in iron storage. Studies have demonstrated an upregulation of the Fth1 gene in brain tissue of patients with MS lesions (22). However, an association with upregulated Fth1 gene in immune cells of CSF is currently unknown. Another highly upregulated gene in all examined cell types was Mbp. The Mbp gene encodes for the myelin basic protein, which is a structural component of myelin sheath. Thus, Mbp expression in our data may represent contamination by myelin. However, according to Marty et al, Mbp expression is detected in T cells isolated from lymph nodes and spleen (23). Several variations in the Mbp gene have been associated with an increased susceptibility to developing MS and its clinical course (24). Furthermore, we observed a significant upregulation of the Pfn1 gene in all immune cell populations. Even though no association with MS has been observed yet, studies have shown that excessive Pfn1 activity can result in the immune system targeting the body’s own cells resulting in other autoimmune diseases (25). Apoe was among the top 10 upregulated genes in activated and naïve CD4+ T cells. Apoe exhibits notable immunomodulatory properties by attenuating immune activation through downregulation of immune stimulatory proteins on antigen-presenting cells. Multiple scientific investigations have established a potential association between Apoe and autoimmune conditions, including MS (26–28). Interestingly, both Apoe and Mbp also showed higher expression in mELT than in spleen and lymph nodes in many immune celltypes in our previous study, comparing mELT with secondary lymphoid organs (11). Macrophages in CSF of B cell depleted mice exhibited upregulated expression of Atf3. Atf3 correlates with macrophages infiltration and plays a role in inflammation, cell division, and apoptosis (29). Additionally, we observed an upregulation of the Junb gene in immune cells from aCD20 treated mice. Junb has been identified as a significant contributor to the activation of T cells and the differentiation and proliferation of Tregs (30). In addition, several subunits of cytochrome c oxidase (Cox5a, Cox5b, Cox8a, Cox6a1), which play an important role in regulation of ATP synthesis via oxidative phosphorylation (OXPHOS) (31), were found to be upregulated after CD20 treatment in all studied cell types. Typically, resting lymphocytes generate energy through oxidative phosphorylation and fatty acid oxidation, whereas activated lymphocytes rapidly shift to glycolysis (32). Finally, also translation initiation factors including Eif1, and Eif3f, were upregulated by CD20 treatment in all studied cell types. Interestingly, another translation initiation factor has been found to play a role in MS development (33). aCD20 mAb treatment leads to an increased expression of proinflammatory chemokines CCL3 and CCL4 in macrophages, which can lead to the attraction of more macrophages and monocytes. In addition, also CXCL14 and CCRL2 were increased in aCD20 mAb-treated CSF samples.

While genes upregulated after aCD20 treatment show more consistency across the different cell types, there is more variability observed in downregulated genes between different cell types. Among the genes found to be downregulated in the CSF after aCD20 treatment was Txnip, which was downregulated in activated CD4+ T cells, naive CD4+ T cells, Tregs, and NK cells. The Txnip gene has been investigated for its role in inflammatory activation and its implication in neurodegenerative disorders (34). In NK cells, the Gzma gene is downregulated in CSF from aCD20 treated mice. The Gzma gene, also known as Granzyme A, encodes a serine protease enzyme that is primarily found in the granules of cytotoxic T lymphocytes (CTLs) and NK cells. Gzma has been associated with immune activation and promotion of inflammation (35). A downregulated gene in macrophages is the Ccl8 (C-C Motif Chemokine Ligand 8) gene. Ccl8 has been linked to a severe disease course in MS (36). Furthermore, Ccl8 is also known for its role in the induction of inflammation and is highly expressed in tumor-associated macrophages (37).

Although some of the top downregulated genes seemed to be related to inflammation and some of the top upregulated genes are related to regulation of inflammation, overall, gene expression analyses comparing CSF immune cells of B cell depleted and non-depleted mice did not show a uniform pro- or anti-inflammatory pattern.

Secondly, we compared the gene expression profiles of immune cells in mELT with CSF in control treated, B cell competent EAE mice. In mELT compared to CSF, overall, more genes were downregulated than upregulated (Figures 7, 8). A full list of significant genes is in Supplementary Table 2.

However, there was an overlap between the top 10 upregulated genes when comparing mELT versus CSF in the control treatment condition (Figure 7) with CSF in aCD20 treatment versus CSF in control treatment (Figure 5). Among the top 10 upregulated genes were once again Mbp, Mobp, Fth1, and Tpt1. These genes have been consistently associated with crucial biological processes, including myelin formation (Mbp, Mobp), iron homeostasis (Fth1), and cellular stress response (Tpt1). The expression of Mbp, Mobp and Plp1, which is also among the top 10 upregulated genes in the naïve CD4+ T cell dataset, could possibly represent myelin contamination in mELT. In B cells, the expression of immediate early genes (IEGs) such as Egr1, Jund, and Fos, which are rapidly induced in response to antigen receptor or cytokine stimulation (38), exhibited higher levels of expression in mELT compared to CSF. Interestingly, we found IEG also to be higher expressed in mELT compared to lymph nodes and spleen in our previous study (11). It is noteworthy that one of the most upregulated genes in B cells was Fos. Previous studies have associated Fos with MS (39).

The majority of genes was downregulated in mELT compared to CSF, including the Klf2 gene, also referred to as Krüppel-like factor 2, downregulated in activated CD4+ T cells and NK cells, naïve CD4+ T cells and B cells. Klf2 is recognized as a negative regulator of EAE-induced neuroinflammation through the inhibition of pro-inflammatory factors. It has been demonstrated that the absence of Klf2 exacerbates neurological dysfunction and neuroinflammation (40). Furthermore, in mELT of control-treated mice, there were additional downregulated genes that seem noteworthy - Hspa8 (all cell types), and Gzma (NK cells). Gzma has been introduced above. Hspa8 belongs to the Hsp70 family of chaperones and exhibits a protective potential in neurodegenerative diseases (41). Additionally, Hsp70 serves as a marker for inflammatory processes in MS. It has been demonstrated that patients with MS have an elevated level of Hsp70 in their serum. Specifically, clinically isolated syndrome and relapsing-remitting MS are associated with higher serum concentrations of Hsp70 compared to progressive MS (42).

In summary, gene downregulation outweighed upregulation in control treated mELT compared to CSF – a finding that remains difficult to interpret.

Pathway enrichment analysis was used to identify differentially regulated pathways among differentially expressed genes (both up and downregulated) from 1) the CSF in the aCD20 vs. control treated conditions and 2) B cell competent immune cells between mELT and CSF.

For the first analysis, the top 5 pathways per cell type are shown in Figure 9. The most relevant seem to be T cell activation (in naïve T cells), antigen processing and presentation of exogenous peptide antigen and cytokine signaling in immune system (both macrophages). Moreover, oxidative phosphorylation, which is known to be important for many immune cell functions (32), was among the top pathways for all cell types. Yet, the majority of regulated pathways seem not to be primarily linked to inflammation. Overall, this analysis suggests that the absence of B cells in the CSF does not significantly change the inflammatory state of the remaining immune cells, whereas the phenotype (cell type composition) is clearly different.

Comparing B cell competent immune cells between mELT and CSF (Figure 10), we found an enrichment of relevant immune-related pathways, for example response to cytokine stimulus/signaling in activated CD4+ T cells, adaptive immune system response in B cells and naïve CD4+ T cells, and activation, cell-cell recognition and TCR signaling in naïve T cells.