Oligodendrocytes regulate the adhesion molecule ICAM-1 in neuroinflammation

Recently, oligodendrocytes (Ol) have been attributed potential immunomodulatory effects. Yet, the exact mode of interaction with pathogenic CNS infiltrating lymphocytes remains unclear. Here, we attempt to dissect mechanisms of Ol modulation during neuroinflammation and characterize the interaction of Ol with pathogenic T cells. RNA expression analysis revealed an upregulation of immune-modulatory genes and adhesion molecules (AMs), ICAM-1 and VCAM-1, in Ol when isolated from mice undergoing experimental autoimmune encephalomyelitis (EAE). To explore whether AMs are involved in the interaction of Ol with infiltrating T cells, we performed co-culture studies on mature Ol and Th1 cells. Live cell imaging analysis showed direct interaction between both cell types. Eighty percentage of Th1 cells created contacts with Ol that lasted longer than 15 min, which may be regarded as physiologically relevant. Exposure of Ol to Th1 cells or their supernatant resulted in a significant extension of Ol processes, and upregulation of AMs as well as other immunomodulatory genes. Our observations indi-cate that blocking of oligodendroglial ICAM-1 can reduce the number of Th1 cells initially contacting the Ol. These results suggest that AMs may play a role in the interaction between Ol and Th1 cells. We identified Ol interacting with CD4 + cells in vivo in spinal cord tissue of EAE diseased mice indicating that our in vitro findings are of interest to further scientific research in this field. Further characterization and understanding of Ol interaction with infiltrating cells may lead to new therapeutic strategies enhancing Ol protection and remyelination potential. Oligodendrocytes regulate immune modulatory genes and adhesion molecules during autoimmune neuroinflammation Oligodendrocytes interact with Th1 cells in vitro in a physiologically relevant manner Adhesion molecules may be involved in Ol-Th1 cell interaction.


| INTRODUCTION
Neuroinflammation is a complex response of the central nervous system (CNS) against injury, pathogens, misfolded proteins and agents that disturb CNS homeostasis. This response is mainly orchestrated by resident glial cells, although upon brain blood barrier (BBB) disruption, it may also involve the peripheral immune system. Neuroinflammation is a hallmark for many neurological diseases, but major infiltration of autoreactive lymphocytes is a pivotal feature of multiple sclerosis (MS). The autoimmune response in MS targets the myelin layer and results in characteristic demyelinating lesions, oligodendrocyte (Ol) death and eventually axonal damage as well as neuronal loss.
The mechanisms that lead to antigen presentation and autoimmunity against myelin antigens are not completely understood. Yet, mounting evidence indicates that auto reactive CD4 + T helper cells play an important role in the development of the disease (Kaskow & Baecher-Allan, 2018;Severson & Hafler, 2010;Traugott et al., 1982).
Currently, the exact mechanisms behind CD4 + T cell induced Ol damage remain largely unclear (Antel et al., 1998;Zaguia et al., 2013). Evidence indicates that T cells may injure Ol via cellcell contact, soluble factors or by modulating the inflammatory milieu (Kirby et al., 2019;Moore et al., 2015;Popko & Baerwald, 1999;Zaguia et al., 2013). Due to the high metabolic demands of myelin production, Ol are very sensitive to inflammation, thus they have been disregarded as active players in neuroinflammation (McTigue & Tripathi, 2008). However, recent studies are challenging this established concept and suggest that Ol may act as immunomodulators of their environment by expressing a variety of inflammation-related genes and perhaps even being able to interact with the peripheral immune system (Falcão et al., 2018;Jäkel et al., 2019;Kirby et al., 2019;Madsen et al., 2020;. Immune cells mainly communicate via chemokines and cytokines, polypeptides with the capacity to influence cell behavior. However, CD4 + T cells also employ direct cell-cell contact for communication (Becher et al., 2000). The adhesion molecules (AMs) intracellular adhesion molecule 1 (ICAM-1) and vascular cell adhesion molecule 1 (VCAM-1) play a fundamental role in lymphocyte-endothelium interaction and T cell activation (Haghayegh et al., 2019;Steiner et al., 2010) and are upregulated in Ol in inflammatory conditions . Thus, they represent suitable candidates for mediating Ol interaction with infiltrating lymphocytes.
Here we explore the possibility that AMs may be involved in T cell-Ol interplay during autoimmune neuroinflammation. Our results support the novel concept of Ol developing an immune phenotype under certain inflammatory settings. Additionally, we describe a Th1 cell-Ol interaction in vitro and observe the formation of stable contacts that may be considered physiologically relevant. A Th1 cell related inflammatory environment exerts morphological changes on Ol and induces the expression of immune related genes as well as AMs, ICAM-1, and VCAM-1. Our data suggest that ICAM-1 may play a role in the interaction between Ol and Th1 cells.

| Mice strains and maintenance
C57BL/6J mice were initially purchased from Charles River breeding laboratories (Sulzfeld, Germany). All mice were housed at the central animal laboratory (ZTL), the animal care facility of the University Hospital in Regensburg (Germany) under a 12 h day/night cycle and standardized environmental conditions receiving normal or powder chow and tap water ad libitum. All mice strains bred in-house were backcrossed on a C57BL/6J background for at least 10 generations. All experiments were in accordance with the German laws for animal protection and were approved by the local ethic committees for animal welfare (55.22532-2-450/55.2-2532.2-395 tetraparesis, respiratory distress, moribund; 10: death (Linker et al., 2002). Mice that exceeded a clinical score of 7 were killed in accordance with animal welfare. Animals were perfused via the left cardiac ventricle with 4% PFA solution. The CNS was removed and embedded either in paraffin or OCT cryo embedding matrix. For gene expression analysis, animals were perfused via the left cardiac ventricle with PBS and O4 + cells were isolated from whole brain or spinal cord before RNA isolation and qPCR analysis. Extraction of tissue for the analysis of Ol-Th1 cell interaction in vivo was performed 4 days after the mice reached peak of disease.

| Cuprizone model
To induce full toxic demyelination of the corpus callosum (CC), 8-10 weeks old C57BL/6J mice were exposed to 0.2% cuprizone (Bis [cyclohexanone] oxaldihydrazone, C9012, Sigma, St. Louis, MO, USA) mixed in powder standard rodent chow for 5 weeks. During this time, mice were monitored for weight loss and trained in a rotor rod to assess motor coordination. For remyelination analysis, animals were returned to normal diet in fresh cages for 3 days. Finally, mice were sacrificed at point of maximum demyelination (week 5) and at early remyelination point (week 5.5) and the brain was extracted for Ol isolation. Age-matched mice receiving normal powder chow were used as naïve cuprizone controls.

| Neonatal oligodendrocyte isolation and culture
The brains from wild type (WT) mice younger than postnatal day 7 (P7) were extracted under sterile conditions and kept in cold hank's balanced salt solution (HBSS) (without Ca 2+ or Mg +2 ) until dissociation according to the neural tissue disassociation kit protocol Miltenyi,Bergisch Gladbach,Germany). CD140 + cells were isolated using MACS magnetic columns following the anti-CD140 microbeads protocol Miltenyi,Bergisch Gladbach,Germany). Briefly, magnetic labeled CD140 antibody binds to cells expressing the corresponding antigen. These cells are applied to a MACS column and retained in a strong magnetic field. The cells can be collected by removing the column from the magnetic field and pushing them out of the column. Labeled cells were plated on slides or plates pre-coated with poly-L-lysine (PLL) overnight at 37 C and kept in MACS neuro media Cells were used after 5 DIV for co-culture.

| T cell isolation and Th1 cell differentiation
The spleens of adult 2D2 mice were extracted under sterile conditions and conserved in cold HBSS (without Ca +2 or Mg +2 ) until tissue disassociation. The spleen was triturated with a syringe plunge over a 70 μm filter, the filter was washed with RPMI 1640 media (Gibco, Darmstadt, Germany) and the solution was collected in a 50 ml falcon tube and centrifuged for 10 min, 300g at 4 C. After decanting the supernatant, the pellet was resuspended in 5 ml of NH 4 Cl (0.14 M) and incubated for 10 min at room temperature (RT). The erythrocyte digestion was stopped with 20 ml of serum containing media, filtered with a 70 μm filter and centrifuged for 10 min, 300g at 4 C. The pellet was resuspended in 10 ml of PBS and the cell number was determined. CD4 + naïve T cells were isolated using the naïve CD4 T cell isolation kit according to the manufacturer's instructions Miltenyi,Bergisch Gladbach,Germany). After isolation, 100,000 cells were plated in a 96 well flat bottom plate precoated overnight with 2 μg/ml anti-CD3 (100,331, Biolegend, San Diego, CA, USA) at 4 C.
To induce a Th1 phenotype, in addition to pre-absorbed CD3, we added soluble 2 μg/ml anti-CD28 ( 2.6 | Co-culture of oligodendrocytes and Th1 cells T cells were harvested, resuspended and counted. Next, they were centrifuged for 5 min at 300 g and resuspended to a density of 1 Â 10 6 cells/ml in co-culture media, 50% RPMI with 10% heat inactivated horse serum and 50% of proliferation media for OPC co-culture and differentiation media for Ol co-culture. The T cells were seeded with Ol at a ratio of 5:1, respectively, and cultured for 24 h at 37 C under constant 5% CO 2 . After incubation, both cell types were collected separately for either FACS analysis, RNA analysis or immunofluorescence staining. The supernatant of Th1 cells was collected and added to Ol in a 1:1 proportion with proliferation (OPC) or differentiation (Ol) media for 24 h.

| Live cell imaging of co-culture experiments
For live cell imaging of the co-culture, Th1 or Th0 cells were added right before transport to the microscope, in addition of 20 mM Hepes.
Cells were imaged for 1.5 h at 37 C, 5% humidity with an inverted Leica AF6000LX microscope, equipped with a Leica DFC350 FX digital camera (Leica, Wetzlar, Germany). Bright field images were taken every 5 min under a 10x objective, using the Leica LASX software.
The images were exported as videos and analyzed with the ImageJ software using the cell counter plugin to determine the number of T cells in contact with Ol and Ol area. The number of Th1 cells in contact was normalized to the OPC/Ol area. The manual tracking plugin was used to follow the crawling behavior of the Th1 cells.

| Adult oligodendrocyte isolation
Adult Ol were isolated from whole brain or spinal cord. The tissue of adult mice was extracted and kept in cold HBSS (without Ca 2+ or Mg 2+ ) until tissue disassociation. In brief, brains were dissociated in a gentle MACS Octo dissociator following the instructions provided by the adult brain dissociation kit (130-107-677, Miltenyi, Bergisch Gladbach, Germany).
After dissociation, O4 + cells were isolated using MACS magnetics columns following the anti-O4 microbeads protocol (130-094-543, Miltenyi, Bergisch Gladbach, Germany). The cells were preserved in BL buffer (Z6012, Promega, WI, USA) at À80 C until RNA isolation was performed or directly used in flow cytometry analysis.

| Immunohistochemistry
Mice were euthanized in a CO 2 chamber and perfused via the left car- To quantify ICAM-1 expression in Ol and Ol-CD4 + T cell interaction in spinal cord, whole spinal cord sections were imaged with FV3000 confocal microscope, with the 10 or 20Â objective. The white matter area was defined in the ImageJ software and the cell counter plugin was used to identify double positive cells.

| Flow cytometry (FC)
Ol and Th1 cells were analyzed by staining of extra and intra-cellular markers. When intracellular cytokine staining was required, lymphocytes were stimulated with ionomycin (1 μM) and PMA (50 ng/ml) in the presence of monensin (2 μM (Martin, 2011) and reads with a length of less than 60 bp after trimming were discarded. Read quality was assessed before and after trimming with FastQC (v0.11.8). Trimmed reads were mapped to the mouse reference genome GRCm38 with the GENCODE annotation 23 using the spliceaware aligner STAR (v2.6.1c) (Dobin et al., 2013). Reads mapping to nonoverlapping exons are counted and summarized as reads per gene using Subread FeatureCounts (v1.6.1) (Liao et al., 2014). Differential expression analysis between the two groups was performed on the count matrices

| Statistical analysis
Statistical analysis was performed using the GraphPad Prism software (GraphPad Software Inc., La Jolla, CA). All data were analyzed by oneway ANOVA followed by Tukey's post-test or Mann-Whitney U test.

| RESULTS
3.1 | RNA sequencing analysis from brain derived Ol of EAE mice revealed regulation of immune related genes RNA sequencing was performed in O4 + cells extracted from the brain of MOG-EAE mice at the peak of disease (mean clinical score = 5.5 downregulating 126 genes (Tables S1 and S2) in comparison to the naïve group ( Figure S1c). A gene ontology analysis revealed that most upregulated genes were involved in immune regulation processes and response to external stimuli (Figure 1b) whereas downregulated genes corresponded to a wider range of categories related to membrane transport and cell signaling (Table S1). Heatmap visualization highlighted differential expression of the upregulated genes, compared to the naïve group, inside the gene ontology category of "regulation of immune response" and involved in processes such as regulation of cell-cell adhesion, peptide binding and cytokine and chemokine activity (Figure 1c). All upregulated genes are also depicted in a heat map ( Figure S1d). Additionally, we validated these results via qPCR analysis on Ol from the EAE model during disease progression (Figure 1d) (the clinical course and scores at time of euthanasia are shown in Figure S3f,g) and confirmed their significant upregulation at the peak of disease in comparison to Ol from aged matched naïve controls (mice without immunization, referred to as naïve in the figures). The genes depicted in the heat map were subjected to a pathway analysis. As expected, they are involved in many neuroinflammation-related signaling pathways (Figure 1e).

| Live cell imaging of MOGTh1 cell-Ol coculture demonstrated the formation of stable contacts
Next, we implemented a co-culture with differentiated primary Ol and pro-inflammatory Th1 cells harboring a MOG specific T cell receptor (TCR; referred to as "Th1 cells") to explore how an inflammatory environment directly affects Ol. At the time of co-culture, the Ol expressed MOG both at the RNA and protein level ( Figure S2a,b) and $80% of the T cells were IFNγ positive ( Figure S2c). In addition, we co-cultured Th1 cells with OPC. The OPC and Ol culture were characterized via ICC, showing a clear distinction between the cell types in each culture. OPC cultures were mainly composed of CD140 + cells whereas the Ol culture mainly contained CNPase + cells and greatly reduced CD140 expression ( Figure S2d,e). We performed live cell imaging of the co-culture to investigate the interaction between both cell types and observed that Th1 cells were able to form stable contacts with OPC and Ol, lasting longer than 15 min (Figure 2a,c).
Mostly, Th1 cells attached to the Ol processes ( Figure 2b) and a small percentage presented a crawling behavior along the Ol process ( Figure 2d). This crawling behavior was unique to Th1 cells and was not observed with Th0 cells (Figure 2d), and was rarely observed in the OPC co-culture (data not shown). Interestingly, Ol experienced morphological changes during co-culture with Th1 cells. They signifi-  Quantification of the live cell imaging revealed that less Th1 cells contacted Ol processes when the ICAM-1 blocking antibody was present in culture (Figure 5a). A similar result was obtained when the corresponding integrin LFA-1 was blocked on Th1 cells (Figure 5a).
As compared to Ol, significantly less Th1 cells contacted OPC, and we observed no effect of blocking ICAM-1 on the number of Th1 cells contacting OPC (Figure 5a). ICAM-1 block did not interfere with the formation of stable contacts in OPC or Ol ( Figure S5b). Th1 cell density did not vary between co-culture and crawling behavior was not altered by blocking antibodies (Figure S5a,c). Since the crawling behavior was rarely observed in the OPC co-culture, this was not fur-

| DISCUSSION
The metabolic demands of myelin production render Ol susceptible targets of neuroinflammation. Thus, Ol have been considered mere victims of inflammatory processes. Yet, OPC and Ol may be actively involved in immune modulatory processes, giving origin to the concept of inflammatory Ol (iOl) (Harrington et al., 2020). Consistently, our RNA sequencing analysis demonstrates that Ol from EAE diseased mice upregulated genes involved in antigen presentation, regulation of cytokine and chemokine production, response to IFNγ, cell-cell adhesion, among others. Falcão et al. (2018) described a similar regulation of immune related genes in Ol and OPC isolated from the spinal cord at the peak of EAE (Falcão et al., 2018). A similar immune phenotype has also been described in MS derived tissue samples (Jäkel et al., 2019). The potential of Ol to modulate the inflammatory milieu opens a new venue of research for therapeutic targets that may skew their immune modulatory capabilities to potentiate remyelination and survival during neuroinflammatory events.
We employed a co-culture system to analyze the impact of proinflammatory Th1 cells on OPC and mature Ol and to investigate possible interactions between these cell types. Here, we describe the formation of stable contacts between Th1 cells and Ol in vitro. Lasting significant interactions of T cells have been described to be longer than 10 min, implying that this contact has the potential to be physiologically relevant (Bartholomäus et al., 2009;Rezai-Zadeh et al., 2009). Previous studies have shown that OPC respond to injury rapidly and efficiently (Hampton et al., 2004;Levine & Reynolds, 1999;Rhodes et al., 2006 (Knapp, 1997;Pan et al., 2020). These findings emphasize the need to understand the effects of activated T cells in Ol subpopulations, as well as to reanalyze whether targeting specific inflammatory factors could prove more beneficial than complete blocking of T cell activation, a therapeutic approach currently used in certain neuroinflammatory conditions. Interestingly, we observed that mature Ol may have a slight modulatory effect on Th1 cells by inducing LFA-1 downregulation. After 24 h of co-culture, we did not find any particular effect of OPC in Th1 cells. Yet, Falcão et al. (2018) reported that co-culture of Th1 cells with OPC for 72 h resulted in a higher proliferation and number of cells producing IFNγ and TNFα (Falcão et al., 2018). Since mature Ol react differently than OPC to inflammatory conditions (Madsen et al., 2020), it could be expected that they also have distinct effects on the surrounding cells as suggested by our results.
ICAM-1 and VCAM-1 have a pivotal role in T cell activation and extravasation. Hence, they were considered as suitable candidates to facilitate the interaction with Ol. Ol exposed to Th1 cells and their supernatant or isolated from diseased EAE mice showed upregulation of both ICAM-1 and VCAM-1. In contrast, Ol from mice undergoing a cuprizone diet, hence exposed to a local glial response, did not regulate AMs, suggesting that the observed upregulation is  (Steiner et al., 2010). In the absence of ICAM-1, VCAM-1 served as an alternative ligand for T cells keeping a certain level of cell arrest and only blocking both molecules culminated in absolute T cell detachment (Bartholomäus et al., 2009;Steiner et al., 2010). Double blocking was not analyzed in the present work, opening the possibility that Th1 cells contacted Ol via alternative ligands such as ICAM-2 or 3, VCAM-1, and MHCII (Falcão et al., 2018;Miyamoto et al., 2016). Blocking ICAM-1 has resulted in ambiguous, possibly stage specific consequences, reporting both beneficial and detrimental outcomes in EAE (Archelos et al., 1993;Kawai et al., 1996;Willenborg et al., 1993). A few studies have characterized ICAM-1 expression in glial cells (Frohman et al., 1989;Sobel et al., 1990). However, Ol are rarely mentioned, emphasizing the need for further investigations on the role of ICAM-1 in oligodendroglial cells. Future studies should focus on oligodendroglia specific knockout of ICAM-1, rather than constitutive blockage, to study the specific effects that this molecule may have in myelinating glia and its interactions with infiltrating lymphocytes.
We also reported the presence of Ol-T cell interaction in vivo, most cells in contact were found close to lesions but we also detected some in the gray matter. Recently, Ol-T cell interaction has been described in MS derived brain tissue (Larochelle et al., 2021). This study also described Ol-Th17 cell interactions in mice that are MHCII independent, thus further strengthening a possible role of AMs in this interaction.
In conclusion, we observed a clear transcriptional remodeling in Ol in the context of the EAE model or when exposed to a Th1 cell  (Pan et al., 2020).
Additional studies on chronic EAE may offer some insights whether this is the case in autoimmune diseases as well. Finally, we report the formation of stable contacts between Th1 cells and Ol in vitro, which may be regulated to some extent by ICAM-1. AMs have long been therapeutic targets in autoimmunity. However, the approved drugs efalizumab and natalizumab, were either restricted from the market or bear some limitations due to a higher risk of progressive multifocal leukoencephalopathy (Seminara & Gelfand, 2010