International Journal of Molecular Sciences Article Deciphering the Genetic Crosstalk between Microglia and Oligodendrocyte Precursor Cells during Demyelination and Remyelination Using Transcriptomic Data Jennifer Enrich-Bengoa 1,2,† , Gemma Manich 2,3,† , Irene R. Dégano 4,5,6, * and Alex Perálvarez-Marín 1,2, * 1 Biophysics Unit, Department of Biochemistry and Molecular Biology, School of Medicine, Universitat Autònoma de Barcelona, 08193 Cerdanyola del Vallès, Catalonia, Spain 2 Institut de Neurociències, Universitat Autònoma de Barcelona, 08193 Cerdanyola del Vallès, Catalonia, Spain 3 Medical Histology Unit, Department of Cell Biology, Physiology and Immunology, School of Medicine, Universitat Autònoma de Barcelona, 08193 Cerdanyola del Vallès, Catalonia, Spain 4 Centro de Investigación Biomédica en Red of Cardiovascular Diseases (CIBERCV), Instituto de Salud Carlos III, 28029 Madrid, Spain 5 Registre Gironí del Cor (REGICOR) Study Group, Hospital del Mar Medical Research Institute (IMIM), 08003 Barcelona, Catalonia, Spain 6 Faculty of Medicine, University of Vic-Central University of Catalonia, 08500 Vic, Catalonia, Spain * Correspondence: irene.roman@umedicina.cat (I.R.D.); alex.peralvarez@uab.cat (A.P.-M.); Tel.: +34-93-881-55-37 (I.R.D.); +34-93-581-4504 (A.P.-M.) † These authors contributed equally to this work. Abstract: Demyelinating disorders show impaired remyelination due to failure in the differentiation of oligodendrocyte progenitor cells (OPCs) into mature myelin-forming oligodendrocytes, a process Citation: Enrich-Bengoa, J.; Manich, driven by microglia–OPC crosstalk. Through conducting a transcriptomic analysis of microarray G.; Dégano, I.R.; Perálvarez-Marín, A. studies on the demyelination–remyelination cuprizone model and using human samples of multiple Deciphering the Genetic Crosstalk sclerosis (MS), we identified molecules involved in this crosstalk. Differentially expressed genes between Microglia and (DEGs) of specific regions/cell types were detected in GEO transcriptomic raw data after cuprizone Oligodendrocyte Precursor Cells treatment and in MS samples, followed by functional analysis with GO terms and WikiPathways. during Demyelination and Additionally, microglia–OPC crosstalk between microglia ligands, OPC receptors and target genes Remyelination Using Transcriptomic Data. Int. J. Mol. Sci. 2022, 23, 14868. was examined with the NicheNet model. We identified 108 and 166 DEGs in the demyelinated https://doi.org/10.3390/ corpus callosum (CC) at 2 and 4 weeks of cuprizone treatment; 427 and 355 DEGs in the remyelinated ijms232314868 (4 weeks of cuprizone treatment + 14 days of normal diet) compared to 2- and 4-week demyelinated CC; 252 DEGs in MS samples and 2730 and 12 DEGs in OPC and microglia of 4-week demyelinated Academic Editors: Antonio Rescifina CC. At this time point, we found 95 common DEGs in the CC and OPCs, and one common DEG and Giuseppe Floresta in microglia and OPCs, mostly associated with myelin and lipid metabolism. Crosstalk analysis Received: 22 September 2022 identified 47 microglia ligands, 43 OPC receptors and 115 OPC target genes, all differentially expressed Accepted: 23 November 2022 in cuprizone-treated samples and associated with myelination. Our differential expression pipeline Published: 28 November 2022 identified demyelination/remyelination transcriptomic biomarkers in studies using diverse platforms Publisher’s Note: MDPI stays neutral and cell types/tissues. Cellular crosstalk analysis yielded novel markers of microglia ligands, OPC with regard to jurisdictional claims in receptors and target genes. published maps and institutional affil- iations. Keywords: corpus callosum; cuprizone; demyelination; oligodendrocytes; microglia Copyright: © 2022 by the authors. 1. Introduction Licensee MDPI, Basel, Switzerland. This article is an open access article Multiple sclerosis (MS) is the most common demyelinating disorder of the central distributed under the terms and nervous system (CNS), with 3.0 million patients worldwide. MS is more common in conditions of the Creative Commons women than in men, and has a great impact on patients’ quality of life [1]. MS is considered Attribution (CC BY) license (https:// as a multifactorial disease that is triggered by genetic predisposition and environmental creativecommons.org/licenses/by/ factors, leading to neurological impairment and disabling symptoms [2,3]. Within MS, 4.0/). current treatments are disease-modifying therapies mainly focused on treating attacks, Int. J. Mol. Sci. 2022, 23, 14868. https://doi.org/10.3390/ijms232314868 https://www.mdpi.com/journal/ijms women than in men, and has a great impact on patients’ quality of life [1]. MS is consid- ered as a multifactorial disease that is triggered by genetic predisposition and environ- mental factors, leading to neurological impairment and disabling symptoms [2,3]. Within Int. J. Mol. Sci. 2022, 23, 14868 2 of 17 MS, current treatments are disease-modifying therapies mainly focused on treating at- tacks, ameliorating the symptoms, and suppressing or modulating inflammation [4,5]. Therefore, a better understanding of the molecules that are involved in this pathology ameliorating would be key the tosymptoms, discoveringand suppressing novel therapies.or modulating inflammation [4,5]. Therefore, a better understanding of the molecules that are involved MS and most other demyelinating disorders in this pathology are characterized would be key by neuroinflammation to discovering novel therapies. that leads to demyelination, oligodendrocyte death and axonal injury. In the CNS, myelin MS and by is produced most other demyelinating myelinating disorders oligodendrocytes are characterized that originate by neuroinflamma- from oligodendrocyte precur- tion that leads to demyelination, oligodendrocyte death and axonal sor cells (OPCs), which can proliferate, migrate, and differentiate into injury. theseInmature the CNS, oli- myelin is produced by myelinating oligodendrocytes that originate from oligodendrocyte godendrocytes that are able to generate myelin. Different studies suggest that these dif- precursor cells (OPCs), which can proliferate, migrate, and differentiate into these mature ferent steps towards mature oligodendrocytes are orchestrated by microglia, astrocytes oligodendrocytes that are able to generate myelin. Different studies suggest that these and OPCs [6–11]. In this crosstalk, both astrocytes and microglia can play a dual role, different steps towards mature oligodendrocytes are orchestrated by microglia, astrocytes being either beneficial or detrimental for (re)myelination through interaction with oli- and OPCs [6–11]. In this crosstalk, both astrocytes and microglia can play a dual role, being godendrocytes. Beneficial roles of microglia promote OPC survival and maturation, foster either beneficial or detrimental for (re)myelination through interaction with oligodendro- OPC differentiation (i.e., via galectin-3), and contribute to myelin formation during devel- cytes. Beneficial roles of microglia promote OPC survival and maturation, foster OPC opment and remyelination [10–14]. However, the complete role of microglia in (re)mye- differentiation (i.e., via galectin-3), and contribute to myelin formation during development lination is poorly understood. and remyelination [10–14]. However, the complete role of microglia in (re)myelination is The objectives of this study were (i) to identify transcriptomic markers of demye- poorly understood. lination and remyelination in CC, microglia, OPC and human MS samples using a com- The objectives of this study were (i) to identify transcriptomic markers of demyelina- mon pipeline, and (ii) in tion and remyelination to CC, study the microglia microglia, OPC andligands as well human as the OPC MS samples receptors using a common and target genes involved in the microglia–OPC crosstalk in demyelination using pipeline, and (ii) to study the microglia ligands as well as the OPC receptors and target available transcriptomic genes involved indata thefrom cuprizone crosstalk microglia–OPC demyelination–remyelination in demyelination using mouse model available experi- transcrip- ments. tomic data from cuprizone demyelination–remyelination mouse model experiments. 2.2.Results Results 2.1. 2.1.Identification IdentificationofofDemyelination Demyelinationand andRemyelination RemyelinationMarker MarkerGenes Genesininthe theCuprizone CuprizoneModel Model 2.1.1. 2.1.1.Differentially DifferentiallyExpressed ExpressedGenes Genes(DEGs) (DEGs)ininthe theCC CC After After22weeks weeksofofdemyelination demyelinationtreatment treatment there therewere were108 108DEGs DEGs in in the the CC, CC, 45 45 down- down- regulated regulated and 63 upregulated (Figure 1A, Table S2). The 10 most significant DEGs be- and 63 upregulated (Figure 1A, Table S2). The 10 most significant DEGs be- tween tween2-week 2-weekcuprizone-treated cuprizone-treatedmice miceand andthe thecontrol controlwere wereSerpinb1a, Serpinb1a,Gdf15, Gdf15,Pigz, Pigz, Tgm1, Tgm1, B230206H07Rik, B230206H07Rik,Ninj2, Ninj2,Eif4ebp1, Eif4ebp1,Slc34a3, Slc34a3,Trib3 Trib3and andCdkn1a. Cdkn1a.Among Amongthese thesegenes, genes, Gdf15, Gdf15, Tgm1, Eif4ebp1, Trib3 and Cdkn1a were downregulated in the CC of cuprizone-treated Tgm1, Eif4ebp1, Trib3 and Cdkn1a were downregulated in the CC of cuprizone-treated mice, mice,while whileSerpinb1a, Serpinb1a,Pigz, Pigz,B230206H07Rik, B230206H07Rik,Ninj2 Ninj2and andSlc34a3 Slc34a3were wereupregulated. upregulated.There There were three Wikipathways associated with the DEGs (p53 signaling, white fat cell differenti- were three Wikipathways associated with the DEGs (p53 signaling, white fat cell differ- ation and hypertrophy model), but no GO molecular functions. entiation and hypertrophy model), but no GO molecular functions. Figure 1. Differentially expressed genes (DEGs) in the corpus callosum (CC), oligodendrocyte progenitor cells (OPCs) and microglia. Volcano plots of −log10-adjusted p-values and log fold changes. Significant DEGs are shown in red and the 10 most significant DEGs are shown in blue. CC analysis at early (2-week cuprizone treatment) (A) and late (4-week cuprizone treatment) (B) demyelination. (C) Microglia analysis. (D) OPC analysis. (E,F) Analysis of early (E) and late (F) demyelination when compared to remyelinating the CC. Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 3 of 18 Int. J. Mol. Sci. 2022, 23, 14868 Figure 1. Differentially expressed genes (DEGs) in the corpus callosum (CC), oligodendrocyte pro- 3 of 17 genitor cells (OPCs) and microglia. Volcano plots of –log10-adjusted p-values and log fold changes. Significant DEGs are shown in red and the 10 most significant DEGs are shown in blue. CC analysis at early (2-week cuprizone treatment) (A) and late (4-week cuprizone treatment) (B) demyelination. After 4 weeks of demyelination treatment we identified 166 DEGs in the CC: 80 down- (C) Microglia analysis. (D) OPC analysis. (E,F) Analysis of early (E) and late (F) demyelination when regulated compared to and 86 upregulated remyelinating (Figure 1B, Table S3 and Figure S1A,B). The 10 most signif- the CC. icant DEGs between 4-week cuprizone-treated mice and the control were Moxd1, Gdf15, Ddit3,After Ccng1, Atf15, Trib3, 4 weeks Slc34a3, Pigz,treatment of demyelination Xrcc3 andwe Ninj2. 166 Moxd1, Of these, identified DEGs in Gdf15, Ddit3, the CC: 80 Ccng1, Atf15 and Trib3 were significantly downregulated in the CC of cuprizone-treated downregulated and 86 upregulated (Figure 1B, Table S3 and Figure S1A,B). The 10 most mice, significant Slc34a3,between while DEGs Pigz, Xrcc3 and Ninj2 4-week were significantly cuprizone-treated mice upregulated (Figure and the control were 2).Moxd1, There were sixDdit3, Gdf15, significant Ccng1,GO molecular Atf15, functionsPigz, Trib3, Slc34a3, associated withNinj2. Xrcc3 and the DEGs identified, Of these, Moxd1,such as Gdf15, being a structural constituent of the myelin sheath (Figure 3A). Four Wikipathways Ddit3, Ccng1, Atf15 and Trib3 were significantly downregulated in the CC of cuprizone- were associated with treated mice, the DEGs, while including Slc34a3, cholesterol Pigz, Xrcc3 and Ninj2biosynthesis and metabolism were significantly (Figure upregulated 3D). (Figure There were 75 common DEGs at 2 and 4 weeks of demyelination, all downregu- 2). There were six significant GO molecular functions associated with the DEGs identified, lated/upregulated at both time points. Among the 10 most significant DEGs, 5 genes were such as being a structural constituent of the myelin sheath (Figure 3A). Four Wikipath- present at 2 and 4 weeks: Gdf15, Atf5 and Trib3 were downregulated compared to control, ways were associated with the DEGs, including cholesterol biosynthesis and metabolism while Slc34a3 and Ninj2 were upregulated. (Figure 3D). Figure2.2.Demyelinating Figure Demyelinatingmarker markergenes genesand andmicroglia–OPC microglia–OPCinteractions interactionsin inthe thecuprizone cuprizonemodel. model.In In this thisstudy, study,different differentdatasets datasetscomparing comparingthe thecontrol controlto to4-week 4-weekcuprizone-treated cuprizone-treatedmice micewere wereanalyzed analyzed in interms termsof ofthe theidentified identifiedDEGs; DEGs;upregulated upregulatedgenes genesare areshown shownin ingreen greenand anddownregulated downregulatedgenesgenes are are shown shown in in red. red. In total, 95 DEGs DEGs were were found found in in common common between betweenboth bothCC CCandandininOPCs, OPCs,andand1 1DEG DEGwas wasfound foundexpressed expressed inin both both microglia microglia and OPCs. In the microglia–OPC microglia–OPC crosstalk, crosstalk, we we found found 47 ligands 47 ligands in in microglia, microglia, 4343OPC OPCreceptors receptorsand and115 115target genes target genesin in OPCs. Microglia OPCs. ligands Microglia are pre- ligands are sented in orange and OPC receptors are presented in purple. Created with Biorender.com. presented in orange and OPC receptors are presented in purple. Created with Biorender.com. Int. J. Mol. Sci. 2022, 23, 14868 4 of 17 Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 4 of 18 Figure 3. GO annotation and WikiPathways analysis of DEGs. (A–C) Dot plots including GO mo- Figure 3. GO annotation and WikiPathways analysis of DEGs. (A–C) Dot plots including GO lecular functions associated with the 20 most significant DEGs in the CC during demyelination: (A) molecular functions Comparison associated of controls withcuprizone-treated to 4-week the 20 most significant mice, and DEGs fromin the demyelination early CC during demyelination: or late (A) demyelination Comparison of to controls to 4-week remyelination; cuprizone-treated (B) 2-week-treated mice, andtofrom mice compared early demyelination remyelination; (C) 4-week- or late treated mice compared demyelination to remyelination. to remyelination; In these graphs, (B) 2-week-treated micedotcompared size showstothe number of DEGs remyelination; (C)for 4-week- each GO function, the -log10 of the p-value for over-represented significant GO terms is illustrated, treated mice compared to remyelination. In these graphs, dot size shows the number of DEGs for each and the color gradation expresses the amount of DEGs in the set divided by the total number of GOgenes function, thethe within -log10 of the GO term. p-value (D–H) for over-represented Dot plots of the significantlysignificant GO termsofisWikiPathways expressed functions illustrated, and the color gradation associated withexpresses the DEGs the amountinofthe identified DEGs in the CC (D) andset divided OPCs after 4byweeks the total number treatment of cuprizone of genes within the (E), GOand term.with the common (D–H) DEGs Dot plots identified of the in the CC significantly and OPC functions expressed analysis (F).ofWikipathways WikiPathways associ- associated ated with DEGs detected in the CC at early and late demyelination compared to remyelinating CC with the DEGs identified in the CC (D) and OPCs after 4 weeks of cuprizone treatment (E), and with were analyzed for (G) 2-week-treated mice compared to remyelination and (H) 4-week-treated mice the compared common to DEGs identifiedIninthese remyelination. the CC andthe graphs, OPC analysis x-axis (F). Wikipathways represents the number of DEGs associated with DEGs in the differ- detected in the CCset ential expression at divided early and latenumber by the demyelination of genes incompared the pathway. to Color remyelinating scale showsCCthewere analyzed adjusted for p-value. Dot size illustrates (G) 2-week-treated mice the number of compared to genes for each pathway. remyelination and (H) 4-week-treated mice compared to remyelination. In these graphs, the x-axis represents the number of DEGs in the differential expression There were 75 common DEGs at 2 and 4 weeks of demyelination, all downregulated/ set divided by the number of genes in the pathway. Color scale shows the adjusted p-value. Dot size upregulated at both time points. Among the 10 most significant DEGs, 5 genes were pre- illustrates the number of genes for each pathway. sent at 2 and 4 weeks: Gdf15, Atf5 and Trib3 were downregulated compared to control, while 2.1.2. DEGsSlc34a3 and Ninj2 were upregulated. in Microglia There 2.1.2. were DEGs 12 DEGs when comparing microglia from 4-week cuprizone-treated and in Microglia control mice. All DEGs were upregulated: Olr1, Cd69, Itgax, Clic4, Lpl, Clec7a, Gas2l3, There were 12 DEGs when comparing microglia from 4-week cuprizone-treated and Fam20c, controlMmp12, Lgals3, mice. All DEGsRab7 wereand Cxcr4 (Figure upregulated: Olr1, 1C, Table Cd69, S4Clic4, Itgax, and Figure S1C). Lpl, Clec7a, There was Gas2l3, oneFam20c, GO molecular Mmp12, Lgals3, Rab7 and Cxcr4 (Figure 1C, Table S4 and Figure S1C). There wasbut no function associated with these DEGs (carbohydrate binding), Wikipathways. one GO molecular function associated with these DEGs (carbohydrate binding), but no Wikipathways. 2.1.3. DEGs in OPCs 2.1.3. DEGs in Analysis of OPCs the DEGs of OPCs between late-demyelinated cuprizone-treated and control mice revealed Analysis of the 2730 DEGsDEGs (those of OPCs between a p-value < 0.001 cuprizone-treated with late-demyelinated are presented in and Table S5). Of these DEGs, 1372 were downregulated and 1358 were upregulated in control mice revealed 2730 DEGs (those with a p-value < 0.001 are presented Table S5). (Figure 1D). The 10 most significant DEGs were all upregulated: Slc7a1, Tagln2, Atf5, Trib3, Cdkn1a, Col5a3, Gadd45b, Egr1, Ppp1r15a and Ccnd1 (Figure 2). There were 226 GO biological processes associated with the DEGs, such as regulation of axon extension, neurogenesis and neuron projection development. Associated WikiPathways included cholesterol and sphingolipid metabolism (Figure 3E). Of these DEGs, 1372 were downregulated and 1358 were upregulated (Figure 1D). The 10 most significant DEGs were all upregulated: Slc7a1, Tagln2, Atf5, Trib3, Cdkn1a, Col5a3, Gadd45b, Egr1, Ppp1r15a and Ccnd1 (Figure 2). There were 226 GO biological processes associated with the DEGs, such as regulation of axon extension, neurogenesis and neuron Int. J. Mol. Sci. 2022, 23, 14868 5 of 17 projection development. Associated WikiPathways included cholesterol and sphingolipid metabolism (Figure 3E). 2.1.4. Common DEGs 2.1.4. Common DEGs inin the the CC, CC, Microglia Microglia and and OPCs OPCs There There were were no no DEGs DEGs in in the the three tissue/cell types three tissue/cell types(CC, (CC,microglia microglia and and OPCs) OPCs) between between 4-week cuprizone-treated and control mice. However, there was 4-week cuprizone-treated and control mice. However, there was one common one common DEGDEGin mi- in croglia andand microglia OPCs (Lgals3), OPCs and and (Lgals3), 95 common DEGsDEGs 95 common in thein CC and the CCOPCs and (Table OPCs S6 and S6 (Table Figure and 4). The 4). Figure 95 The common DEGs were 95 common DEGsassociated with 86 were associated withGO86biological processes, GO biological including processes, gli- including ogenesis, gliogenesis,oligodendrocyte oligodendrocyte development development and nervous and nervoussystem systemdevelopment, development,among amongothers. others. These common common 95 95 DEGs DEGs were also associated associated with with 5 Wikipathways, Wikipathways, such such as spinal spinal cord injury (Figure 2F). Common differentially Figure 4. Common Figure differentially expressed expressed genes genes (DEGs) (DEGs) in in the the corpus callosum callosum (CC), microglia and and oligodendrocyte oligodendrocyte progenitor progenitor cells cells (OPCs) (OPCs) during demyelination. demyelination. Venn Venn diagram diagram showing showing the the number number of of DEGs DEGsshared sharedbetween between murine murine CC,CC, OPCOPC and and microglia microglia samples samples after aafter a 4/5-week 4/5-week cu- cuprizone prizone treatment. treatment. No common No common DEGs wereDEGs were found found between between microglia, microglia, OPCs and OPCs CC, and CC,common but one but one com- DEG mon DEG was found between microglia and OPCs, and 95 shared DEGs were found between CC was found between microglia and OPCs, and 95 shared DEGs were found between CC and OPCs. and OPCs. 2.2. Identification of Common Demyelination Marker Genes in Multiple Sclerosis Samples and in Samples 2.2. from the Cuprizone Identification of CommonMice Model Demyelination Marker Genes in Multiple Sclerosis Samples and in Samples Therefrom werethe 252Cuprizone DEGs when Micecomparing Model MS and control samples (Table S7), and 74 of theseThere DEGswerewere252 also identified in the cuprizone DEGs when comparing MS and model analyses control samples (Figure S2). (Table Ofand S7), the 74 com- of mon 74 these DEGs, DEGs 72 were were differentially also identified expressed in the in OPCs cuprizone modeland 13 in the analyses CC (Tables (Figure S2). OfS8the and S9). com- Moreover, mon 11 out 74 DEGs, 72ofwere 74 ofdifferentially the DEGs were identified expressed ininOPCs both and OPC13 and in CC thesamples (Table CC (Tables S8 S10). and One of the 11 common DEGs in MS samples and in CC and OPC S9). Moreover, 11 out of 74 of the DEGs were identified in both OPC and CC samplessamples was myelin basic protein, (Table S10). Onewhich is 11 of the a factor commonin the pathogenesis DEGs of MSand in MS samples [15]. in In CCaddition, and OPCanother samplesofwasthe 72 common DEGs in MS samples and in OPCs was myelin-associated oligodendrocytic myelin basic protein, which is a factor in the pathogenesis of MS [15]. In addition, another basic of the protein. 72 common DEGs in MS samples and in OPCs was myelin-associated oligodendro- cytic basic protein. 2.3. Identification of Remyelination Marker Genes in the CC in the Cuprizone Model After 2 weeks 2.3. Identification of demyelination of Remyelination treatment, Marker Genes inthe theCC CC had 415 in the DEGs compared Cuprizone Model to the re- myelinating CC, including 15 downregulated and 400 upregulated. The 10 most significant After 2 weeks of demyelination treatment, the CC had 415 DEGs compared to the DEGs were all upregulated: D16Ertd472e, Aoc1, Krt15, Serpinb1c, Slain1, Cxcl10, Sntn, Lrig3, remyelinating CC, including 15 downregulated and 400 upregulated. The 10 most signif- Ppfibp2 and Dusp15 (Figure 1F and Table S11). These DEGs were associated with seven icant DEGs were all upregulated: D16Ertd472e, Aoc1, Krt15, Serpinb1c, Slain1, Cxcl10, Sntn, GO molecular functions, related to myelin sheath structure (Figure 3B), and with three Lrig3, Ppfibp2 and Dusp15 (Figure 1F and Table S11). These DEGs were associated with WikiPathways, including cholesterol metabolism (Figure 3G). seven GO molecular functions, related to myelin sheath structure (Figure 3B), and with When comparing the CC having undergone 4 weeks of demyelination and the re- three WikiPathways, myelinating CC, thereincluding were 355 cholesterol metabolism DEGs, 27 were (Figure and downregulated 3G). 328 were upregulated (Figure 1F and Table S12). The 10 most significant DEGs in this case were Ninj2, Mog, Cyp3a13, Trib3, Gm5067, Shgl3, Msmo1, E330037M01Rik, Plekhh1 and Insig1. From these, the only downregulated DEG was Trib3, while the rest were upregulated. There were 23 GO molecular functions associated with the DEGs, including being a structural con- stituent of the myelin sheath and amyloid-beta binding (Figure 3C and Table S13). Three WikiPathways were associated with the DEGs: cholesterol biosynthesis and metabolism, and omega-9 fatty acid synthesis (Figure 3H). Int. J. Mol. Sci. 2022, 23, 14868 6 of 17 There were 159 common DEGs in the remyelinating CC compared to 2 or 4 weeks of demyelination. All were either up- or downregulated in both analyses. 2.4. Cell Type Composition in CC Samples Estimated cell fractions of CC samples are presented in Table S14. The cell types showing the largest proportion in CC samples were neurons, astrocytes and oligodendro- cytes, while the ones showing the lowest proportions were pericytes, endothelial cells and OPCs. Cell fraction estimation was similar between samples treated, or not treated, with cuprizone. However, in control samples, the proportion of astrocytes was higher and the proportion of oligodendrocytes was lower than in cuprizone-treated samples. There was high accuracy in the predictions, as shown by the Pearson correlation coefficient and the root mean square error (RMSE) between CC data and cell type predicted data (Table S15). 2.5. Microglia–Oligodendrocyte Interaction in Cuprizone-Induced Demyelination–Remyelination: Potential Receptor–Ligand Pairs to Promote Remyelination In total, 47 potential ligands were identified in microglia that showed a similar po- tential to regulate the OPC gene expression set of interest following cuprizone treatment (Figures 2 and 5A, Table S14). All of them showed active target links in OPC genes differ- entially expressed after demyelination. Ligands showed similar expression in cuprizone- treated and control samples, except for the increased expression of two genes—Plau (plas- minogen activator urokinase) and Spp1 (secreted phosphoprotein 1)—which showed bor- derline significant differences in expression (Table S15, adjusted p-value = 0.052). A total of 43 potential receptors for ligands identified in OPC after cuprizone treatment were detected in a ligand–receptor analysis (Figures 2 and 5B). The maximal interaction potential was discovered for the following microglia ligand–OPC receptor pairs: CSF1– CSF1R, GAS6–AXL/TYRO3, IGF1–IGF1R, JAM2–JAM3, PTPRC–CD22, APOE–VLDLR, LRPAP1–VLDLR, SEMAD4D–PLXNB1. From these, the cytokine pair CSF1–CSF1R, GAS6– AXL/TYRO3, the TGF-β signaling transductor SEMAD4D–PLXNB1, the tyrosine kinase signal receptor PTPRC (also known as CD45) and ligand CD22, and the growth factor pair IGF1–IGF1R are involved in cell migration, survival and/or phagocytosis [16–21]; integrins JAM2–JAM3 function as cell adhesion molecules [22], and in APOE–VLDLR, LRPAP1–VLDLR, the receptors are involved in lipid transport. Out of 43 receptors, 15 were differentially expressed in OPCs from cuprizone-treated compared to control mice (Table S18). From these, Jam3 was the most differentially expressed in OPCs, being down- regulated (Figure 5C). Overall, 115 DEGs that could be predicted targets of the identified ligands were identified in OPCs (Figures 2 and 6A). Of those, 107 were differentially expressed in OPCs from cuprizone-treated compared to control mice (Figure 6B and Table S19). The most differentially expressed were Bbc3, Gadd45b and Hist1h3d, all of which were upregulated in OPCs after cuprizone treatment. Int.J.J.Mol. Int. Mol.Sci. Sci.2022, 23,x14868 2022,23, FOR PEER REVIEW 7 7ofof18 17 Figure Figure 5. 5. Identification Identificationof ofinteracting interactingmicroglia microglialigands ligandsandandoligodendrocyte oligodendrocyteprogenitor progenitorcell cell(OPC) (OPC) receptors using the NicheNet model. (A,B) Expression of microglia ligands in microglia receptors using the NicheNet model. (A,B) Expression of microglia ligands in microglia samples samples (A), (A), and and OPC OPC receptor receptor expression expression in in OPC OPC samples samples (B) (B) from from cuprizone-treated cuprizone-treated (Cupri) (Cupri)and and control control mice (Ctr). (C) Predicted interactions between identified microglia ligands and OPC receptors. In all mice (Ctr). (C) Predicted interactions between identified microglia ligands and OPC receptors. In all figures, the color scale indicates the levels of expression, the darker the red or purple colors are, the figures, the color scale indicates the levels of expression, the darker the red or purple colors are, the higher the expression (red) or the regulatory potential of microglia ligands for OPC receptors (pur- higher the expression (red) or the regulatory potential of microglia ligands for OPC receptors (purple). ple). Int. J.J.Mol. Int. Mol.Sci. Sci.2022, 23,x14868 2022,23, FOR PEER REVIEW 8 8ofof18 17 Figure 6. Identification of interacting microglia ligands and oligodendrocyte progenitor cell (OPC) Figure 6. Identification of interacting microglia ligands and oligodendrocyte progenitor cell (OPC) target genes using the NicheNet model. (A) Expression of OPC target genes detected in samples target genes using the NicheNet model. (A) Expression of OPC target genes detected in samples of of OPC OPC fromfrom cuprizone-treated cuprizone-treated (Cupri) (Cupri) or control or control (Ctr)(Ctr) mice. mice. (B) Predicted (B) Predicted interactions interactions between between mi- microglia ligands and OPC target genes. In all figures, the color scale indicates the levels of expression, croglia ligands and OPC target genes. In all figures, the color scale indicates the levels of expression, thedarker the darkerthethered redor orpurple purplecolors colorsare, are,the thehigher higherisisthe theexpression expression(red) (red)or orthe theregulatory regulatorypotential potential of of microglia microglia ligands ligands for for OPC OPC target target genes genes (purple). (purple). Int. J. Mol. Sci. 2022, 23, 14868 9 of 17 3. Discussion In this in silico study, we analyzed demyelination/remyelination transcriptomic mark- ers in mouse CC, microglia and OPCs, as well as human MS samples, and we examined the crosstalk between microglia and OPCs during demyelination. In demyelination, we identified 166 DEGs in the CC, 12 in microglia and 2730 in OPCs. Ninety-five genes were differentially expressed in both CC and in OPCs, and one gene was differentially expressed in both microglia and OPCs. In remyelinating compared to demyelinating CC, there were 611 DEGs. These demyelination/remyelination markers were associated with molecular functions and pathways related to myelin, central nervous system cell populations and lipid metabolism. Regarding the microglia–OPC crosstalk during demyelination, we iden- tified 47 ligands in microglia, 43 OPC receptors and 115 target genes in OPCs (Figure 3). In human MS samples, we identified 252 DEGs, and from those, 13 were shared with demyelinated mouse CC (11 of them were also common with OPCs). This limited over- lapping between species may be explained by the restricted transcriptomic MS sample availability, the area heterogeneity, the different nature of the lesions (chronic versus acute demyelination), and the limitations of the cuprizone model. However, the overlapping of some relevant genes detected in human MS samples involved in myelination, such as Mobp or Mbp, highlights the validity of this transcriptomic analysis, and urges the undertaking of further studies to investigate the role of other common DEGs between human MS and CC, such as Jam3 or CD22. Common genes among the 10 most significant DEGs in early and late demyelinating CC, which included Gdf15, Ninj2 and Trib3, have been previously associated with myelination. Gdf15 is a member of the transforming superfamily of proteins, such as TGF-β [23], and was downregulated in demyelinating CC in our study. Gdf15 is found in oligodendrocytes of the CC, and it is overexpressed in patients with stabilized MS [24], and also in functional recovery after traumatic spinal cord injury (SCI) [25]. It could be possible that the observed downregulation is linked to the severe decrease in oligoden- drocytes in the CC. Ninj2 was found upregulated during early and late demyelination and when comparing late demyelination–remyelination. Ninj2 is associated with neurite outgrowth and myelination networks [26,27]. Therefore, the observed Ninj2 upregulation during demyelination may indicate a feedback loop to restore myelin in the CC. Trib3 was downregulated in early and late demyelination and during remyelination when compared to late demyelination. Trib3 has been associated with myelin destruction [28,29]. The 12 DEGs identified in microglia included Lpl [30] and Olr1 [31], which have been previously associated with myelination. In addition, the common DEG in demyelinating microglia and OPCs, Lgals3, has been consistently identified in microglia and oligoden- drocytes in the cuprizone demyelination–remyelination model [32,33], and its role in oligodendrocyte maturation has been dissected [34]. Three of the 10 most significant DEGs in demyelinating OPCs have been associated with demyelination, and were also found among the most significant DEGs in the CC: Atf5, Trib3 and Cdkn1a. Atf5 was downregulated in the CC but upregulated in OPCs. Atf5 downregulation is necessary for the differentiation of neurons, astrocytes and OPCs, while Atf5 is not expressed in mature oligodendrocytes [35]. One explanation for being downregulated in the CC and upregulated OPCs could be that during demyelination, OPCs are not capable of differentiating due to the overexpression of Atf5. However, other cell types may be trying to counterbalance this effect by downregulation of Atf5. Trib3 was also downregulated in the CC under both demyelination and remyelination conditions, but upregulated in OPCs. Trib3 is expressed in the endoplasmic reticulum of oligodendrocytes, and its overexpression has been linked to stress responses, such as myelin breakdown or cuprizone [29,30]. Our results for OPCs agree with the available literature. The observed downregulation in the CC could be related to the different cell types present in the CC, in addition to oligodendrocytes and OPCs. On the other hand, Cdkn1a (p21) is a ubiquitously expressed cell-cycle-dependent kinase inhibitor that has been found expressed in most oligodendrocytes of demyelinated lesions [36]. Other cell-cycle-dependent kinases of the same family, such as Cdkn1c, were identified as DEGs in human MS samples. Int. J. Mol. Sci. 2022, 23, 14868 10 of 17 GO term analysis showed that DEGs were associated with different biological and molecular processes related to the CNS, such as to constituents of myelin, axon exten- sion and neurogenesis, while associated pathways included cholesterol and sphingolipid metabolism. Cholesterol and sphingolipids are major components of myelin, and their al- tered metabolism is linked to different neurological and neurodegenerative disorders [37,38]. The association of the most significant DEGs with demyelination/remyelination, as well as the whole DEGs with myelin-related functions and pathways, argues for the robustness of the method for DEG analysis. Among the 47 potential microglia ligands, Spp1 and Plau expression were higher in cuprizone-treated compared to control mice. Spp1 upregulation has been observed in response to injury in different animal models of CNS trauma [39,40], in experimental autoimmune encephalomyelitis (EAE) [41], and in microglia/macrophages and astrocytes during cuprizone-induced demyelination [42]. Spp1 has also been related to enhanced myelin formation and prevents apoptosis of human OPCs [43,44]. Our results are in line with previous results published based on the cuprizone model [42], showing that that Spp1 is upregulated during cuprizone-induced demyelination. In addition, Plau has been associated with the promotion of axon outgrowth in myelin presence after injury [35]. Regarding the identified OPC target genes, the most differentially expressed were Bbc3, Gadd45b and Hist1h3d. Bbc3 is a pro-apoptotic gene that promotes premyelinating oligodendrocyte cell death [45]. Our results indicate that Bbc3 is upregulated in cuprizone- treated compared to control mice. These results indicate that Bbc3 overexpression in cuprizone-treated mice could be linked to fostering programmed cell death in OPCs. Moreover, Gadd45b, which was upregulated in cuprizone-treated OPCs, downregulates JNK signaling [46], a signaling event that is necessary for OPC proliferation [47]. Hist1h3d is a gene belonging to the histone cluster family, and is involved in transcriptional control; its expression has been described in other cell types, such as microglia [48]. The performed analysis also revealed that among the 43 identified receptors in OPCs, 15 were differentially expressed. Some of them were upregulated (PLXNB2, TNFRSF1A, ACVR1, LRP1, ITGB5 and AXL) or downregulated (JAM3, HHIP, LDLR, FGFR2, ITGB4, EPHB1, LPAR1, TYRO3 and SORL1) after cuprizone treatment. Most of these OPC recep- tors have been associated with CNS myelin and oligodendrocytes, and related pathologies. Regarding upregulated receptors, mutated Tnfrsf1a is a risk factor for MS due to it interfer- ing in the TNF-α signaling pathway, leading to an increase in proinflammatory signals [49]. In addition, TNF has been associated with an accelerated onset of EAE [50]. Therefore, the increase in Tnfrsf1a after cuprizone treatment could modulate the TNF signaling pathway and contribute to the observed demyelinating phenotype in the cuprizone model. Muta- tions in the Acvr1 gene arrest oligodendrocyte differentiation [51], and Acvr1 can regulate OPC differentiation and myelin production [52]. Our results reinforce the role of Acvr1 during demyelination by inhibiting oligodendrocyte differentiation. Lrp1 is a negative regulator of OPC differentiation, and its deletion is associated with OPC proliferation and the generation of new mature oligodendrocytes [53]. Therefore, Lrp1 increase in the cuprizone model would reduce the number of newborn mature oligodendrocytes, and OPCs would not differentiate into oligodendrocytes. Itgb5 is a member of the integrin family, which is necessary for oligodendrocyte proliferation, survival and maturation [54]. Upregulation of Itgb5 after cuprizone treatment could promote OPC functioning. Finally, Axl is necessary for the maintenance of axonal integrity and remyelination, since its loss induces higher oligodendrocyte apoptosis, reduced remyelination and extensive axonal damage after cuprizone treatment [16,17]. Therefore, Axl increase after cuprizone treatment could be an attempt to protect cells from injury. Regarding the identified downregulated OPC receptors, Jam3 has been reported as a Jam2 counter receptor [55,56]. Jam2 inhibits oligodendrocyte wrapping but does not affect OPC differentiation, proliferation or migration. Therefore, the downregulation of Jam3 could contribute to the loss of myelin integrity that occurs in cuprizone-induced demyelina- tion. Interestingly, Jam3 was also identified as a common DEG between human MS samples Int. J. Mol. Sci. 2022, 23, 14868 11 of 17 and CC analysis. Hhip interacts with Sonic Hedgehog (shh), and it has been reported that Shh signaling protects against demyelination and favors remyelination by OPC prolifer- ation [57]. Ldlr has been related to correct myelination, both directly [58] and through cholesterol metabolism. In addition, low levels of Ldlr have been associated with failure in OPC differentiation [59], consistent to the observed behavior at 4 weeks of cuprizone treatment. Fgfr2 is necessary for upregulating myelin gene expression, regulating myelin thickness and OPC differentiation, and fostering myelin growth during developmental myelination [50]. Additionally, Fgfr2 promotes remyelination in chronic demyelinated lesions but not in acute lesions [60]. Itgb4 has been related to OPC proliferation [61], and its deletion leads to delayed peripheral nerve regeneration in the peripheral nervous system (PNS) [62]. Ephb1 has been described as implicated in T cell differentiation and migration to inflammatory sites in both EAE and MS [63], and stimulates myelin sheath formation [64]. Lpar1 expression has been reported as necessary for oligodendrocyte differentiation and myelination, and its deletion has been correlated with myelin alterations and oligodendro- cyte death [65]. The observed downregulation of Lpar1 in cuprizone-treated mice is in line with published data, contributing to oligodendrocyte death during demyelination. Finally, Tyro3 expression occurs at the onset of remyelination and enhances myelination [66]. Our results from the microglia–OPC crosstalk analysis show that most of the identified microglia ligands, OPC receptors and target genes play important roles in myelination. These results point also to ligands, receptors and target genes that have not been previously associated with myelination, and that could also be involved in this process. In addition, our results highlight the potential of the NicheNet model to study cellular crosstalk not only in co-cultured cell types, but also with transcriptomic data of independent studies where cell types have received the same treatment. This observation may open the possibility of analyzing the crosstalk between cell types that cannot be co-cultured or for which transcriptomic data are available. We combined all the available studies on the CC, microglia and OPCs in mice treated with cuprizone analyzed with microarrays, and used the same pipeline to obtain DEGs; we validated our results by comparing them with human MS samples. DEGs were associated with molecular functions and pathways related to myelination, and some of them were identified in patients with MS. We conducted a crosstalk analysis based on two microglia studies and one OPC study with independent transcriptomic data, and the results from the crosstalk analysis are in accordance with the demyelination/remyelination processes. There are, however, some limitations that should be considered. Some experimental condi- tions of the studies used were different, such as slight time point variation for cuprizone demyelination (either 4 or 5 weeks), or the use of only males in the OPC transcriptome but both sexes in the rest of the studies [67–70]. These changes may be factors that could introduce variability in DEGs detection, despite other factors affecting cuprizone treatment, such as the animal strain or age, being the same for all cases [71]. Additionally, because the number of samples in the analyzed studies was small, the number of DEGs in cell types and tissues after cuprizone treatment could be affected, which could impact in the crosstalk analysis. In addition, we did not know the cell type composition of the corpus callosum samples. However, we performed an in silico composition analysis that yielded the presence of six cell types, with neurons, astrocytes and oligodendrocytes being the most abundant (Tables S12 and S13). Since most of the DEGs, ligands, targets and receptors identified in the analysis were associated with myelination, and some of them were shared with human MS samples, we consider our results to be robust. 4. Materials and Methods 4.1. Data Two searches were performed in GEO/ArrayExpress. One to identify studies that analyzed gene expression by microarray in mice treated, or not treated, with cuprizone, and another one to identify studies that analyzed gene expression by microarray in MS patient white matter samples (Table S1). Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 12 of 18 Two searches were performed in GEO/ArrayExpress. One to identify studies that an- Int. J. Mol. Sci. 2022, 23, 14868 12 of 17 alyzed gene expression by microarray in mice treated, or not treated, with cuprizone, and another one to identify studies that analyzed gene expression by microarray in MS patient white matter samples (Table S1). Raw Raw data data for for the the six six studies studies identified identified were were obtained obtained from from GEO. GEO. There There were were two two studies on mice microglia, one on the mouse CC, one on mice OPCs, and two on studies on mice microglia, one on the mouse CC, one on mice OPCs, and two on humanhuman white white matter matter [67–70,72]. [67–70,72]. 4.2. Identification of Demyelination/Remyelination Biomarkers 4.2. Identification of Demyelination/Remyelination Biomarkers The pipeline to identify demyelination/remyelination DEGs in samples from mice, The pipeline to identify demyelination/remyelination DEGs in samples from mice, and to identify DEGs in human white matter samples from MS patients and controls, was and to identify DEGs in human white matter samples from MS patients and controls, was as follows. We performed data preprocessing, annotation, gene filtering and DEG analysis as follows. We performed data preprocessing, annotation, gene filtering and DEG analysis in all the analyses. In addition, we performed a functional analysis for the DEGs identified in all the analyses. In addition, we performed a functional analysis for the DEGs identified in the studies with mouse samples (Figure 7). Demyelination DEGs were examined in the in the studies with mouse samples (Figure 7). Demyelination DEGs were examined in the CC, microglia and OPC, regarding the studies, as well as in the MS samples. Remyelination CC, microglia and OPC, regarding the studies, as well as in the MS samples. Remye- DEGs were only examined in the CC. lination DEGs were only examined in the CC. Figure 7. Methodology workflow and summary of results. * In OPCs from cuprizone-treated mice compared to OPCs from control mice. CC, corpus callosum; DEGs, differentially expressed genes; GO, gene ontology; MS, multiple sclerosis; OPC, oligodendrocyte progenitor cell. Int. J. Mol. Sci. 2022, 23, 14868 13 of 17 Quality control of raw data was performed with boxplots and density plots of log intensities and with MA plots. In Affymetrix studies, we also inspected images and plots of relative log expression values and of normalized unscaled standard errors. Preprocessing was performed as follows. Background correction was performed with convolution of normal and exponential distributions, then, quantile normalization was used. Finally, summarization was performed with the median-polish method for Affymetrix studies, and by replacing array replicates with their average for Agilent studies. For the microglia analysis and for the white matter analysis, normalized data of the two available studies were combined using quantile normalization. Normalized data were examined with boxplots. After preprocessing, the data were annotated and filtered. We excluded probes with no symbol or low expression, considering, in that case, the 30th percentile of lower intensity in several arrays as the smallest experimental group. Gene filtering was included in Agilent studies for control and negative flagged probes. Sample clustering by treatment and scan date was analyzed with hierarchical clustering and principal component analysis. There was no scan date clustering. Treatment clustering was clear in microglia and OPC data, and moderate in CC and white matter data. We excluded the most different half of the samples in the CC analyses based on sample clustering. For the microglia and the white matter analyses, clustering by study was identified and corrected with the ComBat method. Differential expression was analyzed with moderated t-tests including an intensity trend for prior variance and robustifying for outlier sample variances. p-values were adjusted for multiple comparisons using the Benjamini–Hochberg (BH) method. Duplicated assigned genes were removed by selecting the one with the lowest p-value. In the analyses of mice samples, DEGs were visualized with volcano plots. Sample clustering by gene expression of all DEGs and by gene expression of the most significant DEGs was visualized with heatmaps (Figure S1). Functional analysis of DEGs was per- formed on the three cell types/regions separately, and on the DEGs identified in more than one cell type/region. Considering the large number of DEGs in OPCs, functional analysis was undertaken with an adjusted p-value < 0.001. Then, we performed over-representation functional analyses with gene ontology (GO) molecular-level functions and WikiPathways. p-values were obtained with the hypergeometric test and adjusted with the BH method. Significant GO molecular functions and pathways were visualized with dot plots. 4.3. Cell Type Analysis in CC Samples CIBERSORTx [73] was used to perform the cell type composition analysis. We used, as the signature matrix, the gene expression signature of the mouse brain obtained from single-cell RNA sequencing by Donovan et al. [74] Mixture data were the normalized and filtered expression data of the five CC samples included in the 4-week demyelination analysis. Quantile normalization and 100 permutations were used to obtain the results. Pearson’s correlation coefficient and RMSE of signature genes between original mixture data and the estimated mixture were computed. Sensitivity analyses were performed using the raw and normalized data, by enabling batch correction, by not including quantile normalization and by increasing the number of permutations. 4.4. Microglia and OPC Crosstalk We applied the NicheNet model [75] to examine the intercellular communication between microglia (sender cells) and OPCs (receiver cells). OPC and microglia genes used for the analysis had been previously normalized, annotated and filtered. Gene filtering for low expression included genes below the 60th (microglia) and the 75th (OPCs) percentile of intensity. The gene set of interest was defined as the OPC DEGs with a p-value < 0.01. OPC background genes were all those genes expressed by OPCs except for genes contained in the set of interest. Potential microglia ligands were selected from those expressed by microglia that were able to bind putative OPC receptors. Potential OPC receptors were those expressed by these cells that could bind a putative microglia ligand. Putative ligand– receptor associations were restricted to the ones described in curated databases. Int. J. Mol. Sci. 2022, 23, 14868 14 of 17 Ligand activity was determined by how well the ligands predicted changes in the expression in the gene set of interest when compared to background genes, and was categorized using the Pearson correlation coefficient. OPC target genes were analyzed according to those expressed by OPCs that belonged to the 100 most strongly predicted target genes of the top-ranked microglia ligands. Both ligands and receptors were organized by hierarchical clustering. Microglia ligand and OPC receptor/target gene association was visualized with heatmaps. Expression of the identified microglia ligand and OPC receptor/target gene association was compared between cuprizone samples and controls using t-tests, with adjusted p-values for multiple comparisons with the BH method. Analyses were conducted in R-4.0.4. 5. Conclusions In this study, we established a strategy to select genes associated with different cell types or tissues in de-/remyelination processes in cuprizone-treated mice. Genes identified in this study were related to molecular functions and pathways associated with myelination. These genes are potential candidates for determining demyelination and remyelination in cuprizone-treated mice, and the whole strategy could be used to analyze different cell types in other demyelination and remyelination models and/or treatments. We also performed an in silico analysis to identify ligands, receptors and target genes involved in the microglia–OPC crosstalk during demyelination. Some of them were associated with demyelination, as they were differentially expressed in samples from cuprizone-treated compared to control mice. This result suggests that crosstalk analysis can be studied using data from different experiments, at least in cuprizone-induced demyelination experimental models. We discovered novel potential demyelination biomarkers as well as novel ligands, receptors and target genes involved in microglia–OPC crosstalk, which demand additional validation studies. Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijms232314868/s1. Author Contributions: Conceptualization, J.E.-B., G.M., I.R.D. and A.P.-M.; methodology, software, validation, formal analysis, investigation, resources, I.R.D. and J.E.-B.; writing—original draft prepa- ration, J.E.-B., G.M. and I.R.D.; writing—review and editing, supervision, project administration and funding acquisition, I.R.D. and A.P.-M. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by Spanish Government MCIN/AEI/10.13039/501100011033 (Project BFU2017-87843-R- and Project PID2020-120222GB-I00 to A.P.-M.) and by the Carlos III Health Institute and the European Regional Development Fund (CIBERCV CB16/11/00229 and FIS PI21/00163 to I.R.D.), and by the Marató TV3 Foundation (202119-33 to I.R.D.). Data Availability Statement: Datasets used in this study (GSE100663, GSE84113, GSE66926, GSE48872, GSE38010, GSE52139) can be found in the GEO dataset browser of the NIH (https://www.ncbi.nlm. nih.gov/gds/, accessed on 31 October 2022). Conflicts of Interest: The authors declare no conflict of interest. References 1. Walton, C.; King, R.; Rechtman, L.; Kaye, W.; Leray, E.; Marrie, R.A.; Robertson, N.; La Rocca, N.; Uitdehaag, B.; van der Mei, I.; et al. Rising prevalence of multiple sclerosis worldwide: Insights from the Atlas of MS, third edition. Mult. Scler. J. 2020, 26, 1816–1821. [CrossRef] [PubMed] 2. Ghasemi, N.; Razavi, S.; Nikzad, E. Multiple Sclerosis: Pathogenesis, Symptoms, Diagnoses and Cell-Based Therapy. Cell J. 2017, 19, 1–10. 3. Gelfand, J.M. Multiple sclerosis: Diagnosis, differential diagnosis, and clinical presentation. Handb. Clin. Neurol. 2014, 122, 269–290. [CrossRef] 4. Hauser, S.L.; Cree, B.A. Treatment of Multiple Sclerosis: A Review. Am. J. Med. 2020, 133, 1380–1390.e2. [CrossRef] [PubMed] 5. Dobson, R.; Giovannoni, G. Multiple sclerosis—A review. Eur. J. Neurol. 2019, 26, 27–40. [CrossRef] Int. J. Mol. Sci. 2022, 23, 14868 15 of 17 6. Pang, Y.; Cai, Z.; Rhodes, P.G. Effects of lipopolysaccharide on oligodendrocyte progenitor cells are mediated by astrocytes and microglia. J. Neurosci. Res. 2000, 62, 510–520. [CrossRef] [PubMed] 7. Liu, Z.; Xu, D.; Wang, S.; Chen, Y.; Li, Z.; Gao, X.; Jiang, L.; Tang, Y.; Peng, Y. Astrocytes induce proliferation of oligodendrocyte progenitor cells via connexin 47-mediated activation of the ERK/Id4 pathway. Cell Cycle 2017, 16, 714–722. [CrossRef] [PubMed] 8. Willis, C.M.; Nicaise, A.M.; Bongarzone, E.R.; Givogri, M.; Reiter, C.R.; Heintz, O.; Jellison, E.R.; Sutter, P.A.; TeHennepe, G.; Ananda, G.; et al. Astrocyte Support for Oligodendrocyte Differentiation can be Conveyed via Extracellular Vesicles but Diminishes with Age. Sci. Rep. 2020, 10, 828. [CrossRef] [PubMed] 9. Schnädelbach, O.; Fawcett, J. Astrocyte influences on oligodendrocyte progenitor migration. Prog. Brain Res. 2001, 132, 97–102. [CrossRef] 10. Kalafatakis, I.; Karagogeos, D. Oligodendrocytes and Microglia: Key Players in Myelin Development, Damage and Repair. Biomolecules 2021, 11, 1058. [CrossRef] 11. Santos, E.N.; Fields, R.D. Regulation of myelination by microglia. Sci. Adv. 2021, 7, eabk1131. [CrossRef] [PubMed] 12. Miron, V.E.; Boyd, A.; Zhao, J.-W.; Yuen, T.J.; Ruckh, J.M.; Shadrach, J.L.; van Wijngaarden, P.; Wagers, A.J.; Williams, A.; Franklin, R.J.M.; et al. M2 microglia and macrophages drive oligodendrocyte differentiation during CNS remyelination. Nat. Neurosci. 2013, 16, 1211–1218. [CrossRef] [PubMed] 13. Traiffort, E.; Kassoussi, A.; Zahaf, A.; Laouarem, Y. Astrocytes and Microglia as Major Players of Myelin Production in Normal and Pathological Conditions. Front. Cell. Neurosci. 2020, 14, 79. [CrossRef] [PubMed] 14. HDomingues, H.S.; Portugal, C.C.; Socodato, R.; Relvas, J.B. Oligodendrocyte, Astrocyte, and Microglia Crosstalk in Myelin Development, Damage, and Repair. Front. Cell Dev. Biol. 2016, 4, 71. [CrossRef] 15. Martinsen, V.; Kursula, P. Multiple sclerosis and myelin basic protein: Insights into protein disorder and disease. Amino Acids 2021, 54, 99–109. [CrossRef] 16. Hoehn, H.J.; Kress, Y.; Sohn, A.; Brosnan, C.F.; Bourdon, S.; Shafit-Zagardo, B. Axl−/− mice have delayed recovery and prolonged axonal damage following cuprizone toxicity. Brain Res. 2008, 1240, 1–11. [CrossRef] 17. Ray, A.K.; Dubois, J.C.; Gruber, R.C.; Guzik, H.M.; Gulinello, M.E.; Perumal, G.; Raine, C.; Kozakiewicz, L.; Williamson, J.; Shafit- Zagardo, B. Loss of Gas6 and Axl signaling results in extensive axonal damage, motor deficits, prolonged neuroinflammation, and less remyelination following cuprizone exposure. Glia 2017, 65, 2051–2069. [CrossRef] 18. Saher, G.; Brügger, B.; Lappe-Siefke, C.; Möbius, W.; Tozawa, R.-I.; Wehr, M.C.; Wieland, F.; Ishibashi, S.; Nave, K.-A. High cholesterol level is essential for myelin membrane growth. Nat. Neurosci. 2005, 8, 468–475. [CrossRef] 19. Shafit-Zagardo, B.; Gruber, R.C.; DuBois, J.C. The role of TAM family receptors and ligands in the nervous system: From development to pathobiology. Pharmacol. Ther. 2018, 188, 97–117. [CrossRef] 20. Coughlin, S.; Noviski, M.; Mueller, J.L.; Chuwonpad, A.; Raschke, W.C.; Weiss, A.; Zikherman, J. An extracatalytic function of CD45 in B cells is mediated by CD22. Proc. Natl. Acad. Sci. USA 2015, 112, E6515–E6524. [CrossRef] 21. Vodrazka, P.; Korostylev, A.; Hirschberg, A.; Swiercz, J.M.; Worzfeld, T.; Deng, S.; Fazzari, P.; Tamagnone, L.; Offermanns, S.; Kuner, R. The semaphorin 4D-plexin-B signaling complex regulates dendritic and axonal complexity in developing neurons via diverse pathways. Eur. J. Neurosci. 2009, 30, 1193–1208. [CrossRef] [PubMed] 22. Ebnet, K. Junctional Adhesion Molecules (JAMs): Cell Adhesion Receptors With Pleiotropic Functions in Cell Physiology and Development. Physiol. Rev. 2017, 97, 1529–1554. [CrossRef] [PubMed] 23. Scheld, M.; Fragoulis, A.; Nyamoya, S.; Zendedel, A.; Denecke, B.; Krauspe, B.; Teske, N.; Kipp, M.; Beyer, C.; Clarner, T. Mitochondrial Impairment in Oligodendroglial Cells Induces Cytokine Expression and Signaling. J. Mol. Neurosci. 2018, 67, 265–275. [CrossRef] [PubMed] 24. Amstad, A.; Coray, M.; Frick, C.; Barro, C.; Oechtering, J.; Amann, M.; Wischhusen, J.; Kappos, L.; Naegelin, Y.; Kuhle, J.; et al. Growth differentiation factor 15 is increased in stable MS. Neurol.-Neuroimmunol. Neuroinflamm. 2020, 7, e675. [CrossRef] 25. Golakani, M.H.; Mohammad, M.G.; Li, H.; Gamble, J.; Breit, S.N.; Ruitenberg, M.J.; Brown, D.A. MIC-1/GDF15 Overexpression Is Associated with Increased Functional Recovery in Traumatic Spinal Cord Injury. J. Neurotrauma 2019, 36, 3410–3421. [CrossRef] 26. Araki, T.; Milbrandt, J. Ninjurin2, a Novel Homophilic Adhesion Molecule, Is Expressed in Mature Sensory and Enteric Neurons and Promotes Neurite Outgrowth. J. Neurosci. 2000, 20, 187–195. [CrossRef] 27. Allen, M.; Wang, X.; Burgess, J.D.; Watzlawik, J.; Serie, D.J.; Younkin, C.S.; Nguyen, T.; Malphrus, K.G.; Lincoln, S.; Carrasquillo, M.M.; et al. Conserved brain myelination networks are altered in Alzheimer’s and other neurodegenerative diseases. Alzheimer’s Dement. 2017, 14, 352–366. [CrossRef] 28. YShimotsuma, Y.; Tanaka, M.; Izawa, T.; Yamate, J.; Kuwamura, M. Enhanced Expression of Trib3 during the Development of Myelin Breakdown in dmy Myelin Mutant Rats. PLoS ONE 2016, 11, e0168250. [CrossRef] 29. Abe, H.; Tanaka, T.; Kimura, M.; Mizukami, S.; Saito, F.; Imatanaka, N.; Akahori, Y.; Yoshida, T.; Shibutani, M. Cuprizone decreases intermediate and late-stage progenitor cells in hippocampal neurogenesis of rats in a framework of 28-day oral dose toxicity study. Toxicol. Appl. Pharmacol. 2015, 287, 210–221. [CrossRef] 30. Bruce, K.D.; Gorkhali, S.; Given, K.; Coates, A.M.; Boyle, K.E.; Macklin, W.B.; Eckel, R.H. Lipoprotein Lipase Is a Feature of Alternatively-Activated Microglia and May Facilitate Lipid Uptake in the CNS During Demyelination. Front. Mol. Neurosci. 2018, 11, 57. [CrossRef] Int. J. Mol. Sci. 2022, 23, 14868 16 of 17 31. Gaultier, A.; Wu, X.; Le Moan, N.; Takimoto, S.; Mukandala, G.; Akassoglou, K.; Campana, W.M.; Gonias, S.L. Low-density lipoprotein receptor-related protein 1 is an essential receptor for myelin phagocytosis. J. Cell Sci. 2009, 122, 1155–1162. [CrossRef] [PubMed] 32. Hoyos, H.; Rinaldi, M.; Mendez-Huergo, S.; Marder, M.; Rabinovich, G.; Pasquini, J.; Pasquini, L. Galectin-3 controls the response of microglial cells to limit cuprizone-induced demyelination. Neurobiol. Dis. 2014, 62, 441–455. [CrossRef] [PubMed] 33. Deczkowska, A.; Keren-Shaul, H.; Weiner, A.; Colonna, M.; Schwartz, M.; Amit, I. Disease-Associated Microglia: A Universal Immune Sensor of Neurodegeneration. Cell 2018, 173, 1073–1081. [CrossRef] [PubMed] 34. Thomas, L.; Pasquini, L.A. Galectin-3 Exerts a Pro-differentiating and Pro-myelinating Effect Within a Temporal Window Spanning Precursors and Pre-oligodendrocytes: Insights into the Mechanisms of Action. Mol. Neurobiol. 2020, 57, 976–987. [CrossRef] 35. Mason, J.L.; Angelastro, J.M.; Ignatova, T.N.; Kukekov, V.G.; Lin, G.; Greene, L.A.; Goldman, J.E. ATF5 regulates the proliferation and differentiation of oligodendrocytes. Mol. Cell. Neurosci. 2005, 29, 372–380. [CrossRef] 36. Shen, K.; Reichelt, M.; Kyauk, R.V.; Ngu, H.; Shen, Y.-A.A.; Foreman, O.; Modrusan, Z.; Friedman, B.A.; Sheng, M.; Yuen, T.J. Multiple sclerosis risk gene Mertk is required for microglial activation and subsequent remyelination. Cell Rep. 2021, 34, 108835. [CrossRef] 37. Raddatz, B.B.; Sun, W.; Brogden, G.; Sun, Y.; Kammeyer, P.; Kalkuhl, A.; Colbatzky, F.; Deschl, U.; Naim, H.Y.; Baumgärtner, W.; et al. Central Nervous System Demyelination and Remyelination is Independent from Systemic Cholesterol Level in Theiler’s Murine Encephalomyelitis. Brain Pathol. 2015, 26, 102–119. [CrossRef] 38. Li, Y.; Xu, X.; Zhang, D.; Cheng, W.; Zhang, Y.; Yu, B.; Zhang, Y. Genetic variation in the leukotriene pathway is associated with myocardial infarction in the Chinese population. Lipids Health Dis. 2019, 18, 25. [CrossRef] 39. Ellison, J.A.; Barone, F.C.; Feuerstein, G.Z. Matrix Remodeling after Stroke: De Novo Expression of Matrix Proteins and Integrin Receptors. Ann. N. Y. Acad. Sci. 1999, 890, 204–222. [CrossRef] 40. Wang, X.; Louden, C.; Yue, T.-L.; Ellison, J.A.; Barone, F.C.; Solleveld, H.A.; Feuerstein, G.Z. Delayed expression of osteopontin after focal stroke in the rat. J. Neurosci. 1998, 18, 2075–2083. [CrossRef] 41. Chabas, D.; Baranzini, S.E.; Mitchell, D.; Bernard, C.C.A.; Rittling, S.R.; Denhardt, D.T.; Sobel, R.A.; Lock, C.; Karpuj, M.; Pedotti, R.; et al. The influence of the proinflammatory cytokine, osteopontin, on autoimmune demyelinating disease. Science 2001, 294, 1731–1735. [CrossRef] [PubMed] 42. Selvaraju, R.; Bernasconi, L.; Losberger, C.; Graber, P.; Kadi, L.; Avellana-Adalid, V.; Picard-Riera, N.; Van Evercooren, A.B.; Cirillo, R.; Kosco-Vilbois, M.; et al. Osteopontin is upregulated during in vivo demyelination and remyelination and enhances myelin formation in vitro. Mol. Cell. Neurosci. 2004, 25, 707–721. [CrossRef] 43. Minor, K.; Phillips, J.; Seeds, N.W. Tissue plasminogen activator promotes axonal outgrowth on CNS myelin after conditioned injury. J. Neurochem. 2009, 109, 706–715. [CrossRef] [PubMed] 44. Mazaheri, N.; Peymani, M.; Galehdari, H.; Ghaedi, K.; Ghoochani, A.; Kiani-Esfahani, A.; Nasr-Esfahani, M.H. Ameliorating Effect of Osteopontin on H2O2-Induced Apoptosis of Human Oligodendrocyte Progenitor Cells. Cell. Mol. Neurobiol. 2018, 38, 891–899. [CrossRef] [PubMed] 45. Sun, L.O.; Mulinyawe, S.B.; Collins, H.Y.; Ibrahim, A.; Li, Q.; Simon, D.J.; Tessier-Lavigne, M.; Barres, B.A. Spatiotemporal Control of CNS Myelination by Oligodendrocyte Programmed Cell Death through the TFEB-PUMA Axis. Cell 2018, 175, 1811–1826. [CrossRef] [PubMed] 46. de Smaele, E.; Zazzeroni, F.; Papa, S.; Nguyen, D.U.; Jin, R.; Jones, J.; Cong, R.; Franzoso, G. Induction of gadd45beta by NF-kappaB downregulates pro-apoptotic JNK signalling. Nature 2001, 414, 308–313. [CrossRef] 47. Zhang, J.-X.; Feng, Y.-F.; Qi, Q.; Shen, L.; Wang, R.; Zhou, J.-S.; Lü, H.-Z.; Hu, J.-G. JNK is necessary for oligodendrocyte precursor cell proliferation induced by the conditioned medium from B104 neuroblastoma cells. J. Mol. Neurosci. 2014, 52, 269–276. [CrossRef] 48. Solga, A.C.; Pong, W.W.; Walker, J.; Wylie, T.; Magrini, V.; Apicelli, A.J.; Griffith, M.; Griffith, O.L.; Kohsaka, S.; Wu, G.F.; et al. RNA-sequencing reveals oligodendrocyte and neuronal transcripts in microglia relevant to central nervous system disease. Glia 2015, 63, 531–548. [CrossRef] 49. Caminero, A.; Comabella, M.; Montalban, X. Role of tumour necrosis factor (TNF)-α and TNFRSF1A R92Q mutation in the pathogenesis of TNF receptor-associated periodic syndrome and multiple sclerosis. Clin. Exp. Immunol. 2011, 166, 338–345. [CrossRef] 50. Kassiotis, G.; Pasparakis, M.; Kollias, G.; Probert, L. TNF accelerates the onset but does not alter the incidence and severity of myelin basic protein-induced experimental autoimmune encephalomyelitis. Eur. J. Immunol. 1999, 29, 774–780. [CrossRef] 51. Fortin, J.; Tian, R.; Zarrabi, I.; Hill, G.; Williams, E.; Sanchez-Duffhues, G.; Thorikay, M.; Ramachandran, P.; Siddaway, R.; Wong, J.F.; et al. Mutant ACVR1 Arrests Glial Cell Differentiation to Drive Tumorigenesis in Pediatric Gliomas. Cancer Cell 2020, 37, 308–323.e12. [CrossRef] [PubMed] 52. Quan, L.; Uyeda, A.; Muramatsu, R. Central nervous system regeneration: The roles of glial cells in the potential molecular mechanism underlying remyelination. Inflamm. Regen. 2022, 42, 7. [CrossRef] [PubMed] 53. Auderset, L.; Pitman, K.A.; Cullen, C.L.; Pepper, R.E.; Taylor, B.V.; Foa, L.; Young, K.M. Low-Density Lipoprotein Receptor-Related Protein 1 (LRP1) Is a Negative Regulator of Oligodendrocyte Progenitor Cell Differentiation in the Adult Mouse Brain. Front. Cell Dev. Biol. 2020, 8, 1067. [CrossRef] [PubMed] Int. J. Mol. Sci. 2022, 23, 14868 17 of 17 54. O’Meara, R.W.; Michalski, J.-P.; Kothary, R. Integrin Signaling in Oligodendrocytes and Its Importance in CNS Myelination. J. Signal Transduct. 2011, 2011, 354091. [CrossRef] [PubMed] 55. Arrate, M.P.; Rodriguez, J.M.; Tran, T.M.; Brock, T.A.; Cunningham, S.A. Cloning of human junctional adhesion molecule 3 (JAM3) and its identification as the JAM2 counter-receptor. J. Biol. Chem. 2001, 276, 45826–45832. [CrossRef] 56. Liang, T.W.; Chiu, H.H.; Gurney, A.; Sidle, A.; Tumas, D.B.; Schow, P.; Foster, J.; Klassen, T.; Dennis, K.; DeMarco, R.A.; et al. Vascular endothelial-junctional adhesion molecule (VE-JAM)/JAM 2 interacts with T, NK, and dendritic cells through JAM 3. J. Immunol. 2002, 168, 1618–1626. [CrossRef] 57. Ferent, J.; Zimmer, C.; Durbec, P.; Ruat, M.; Traiffort, E. Sonic Hedgehog Signaling Is a Positive Oligodendrocyte Regulator during Demyelination. J. Neurosci. 2013, 33, 1759–1772. [CrossRef] 58. Xie, Y.; Zhang, X.; Xu, P.; Zhao, N.; Zhao, Y.; Li, Y.; Hong, Y.; Peng, M.; Yuan, K.; Wan, T.; et al. Aberrant oligodendroglial LDL receptor orchestrates demyelination in chronic cerebral ischemia. J. Clin. Investig. 2021, 131, e128114. [CrossRef] 59. Pietiäinen, V.; Vassilev, B.; Blom, T.; Wang, W.; Nelson, J.; Bittman, R.; Bäck, N.; Zelcer, N.; Ikonen, E. NDRG1 functions in LDL receptor trafficking by regulating endosomal recycling and degradation. J. Cell Sci. 2013, 126, 3961–3971. [CrossRef] 60. Furusho, M.; Dupree, J.L.; Nave, K.-A.; Bansal, R. Fibroblast Growth Factor Receptor Signaling in Oligodendrocytes Regulates Myelin Sheath Thickness. J. Neurosci. 2012, 32, 6631–6641. [CrossRef] 61. Zhang, W.; Zhang, X.; Zhang, L.; Xu, D.; Cheng, N.; Tang, Y.; Peng, Y. Astrocytes increase exosomal secretion of oligodendrocyte precursor cells to promote their proliferation via integrin β4-mediated cell adhesion. Biochem. Biophys. Res. Commun. 2020, 526, 341–348. [CrossRef] [PubMed] 62. van der Zee, C.E.E.M.; Kreft, M.; Beckers, G.; Kuipers, A.; Sonnenberg, A. Conditional Deletion of the Itgb4 Integrin Gene in Schwann Cells Leads to Delayed Peripheral Nerve Regeneration. J. Neurosci. 2008, 28, 11292–11303. [CrossRef] [PubMed] 63. Darling, T.K.; Lamb, T.J. Emerging Roles for Eph Receptors and Ephrin Ligands in Immunity. Front. Immunol. 2019, 10, 1473. [CrossRef] [PubMed] 64. Linneberg, C.; Harboe, M.; Laursen, L.S. Axo-Glia Interaction Preceding CNS Myelination Is Regulated by Bidirectional Eph- Ephrin Signaling. ASN Neuro 2015, 7, 1759091415602859. [CrossRef] [PubMed] 65. Diaz, B.G.; Riquelme, R.; Varela-Nieto, I.; Jiménez, A.J.; De Diego, I.; Gómez-Conde, A.I.; Matas-Rico, E.; Aguirre, J.; Chun, J.; Pedraza, C.; et al. Loss of lysophosphatidic acid receptor LPA1 alters oligodendrocyte differentiation and myelination in the mouse cerebral cortex. Brain Struct. Funct. 2015, 220, 3701–3720. [CrossRef] 66. Akkermann, R.; Aprico, A.; Perera, A.A.; Bujalka, H.; Cole, A.E.; Xiao, J.; Field, J.; Kilpatrick, T.J.; Binder, M.D. The TAM receptor Tyro3 regulates myelination in the central nervous system. Glia 2017, 65, 581–591. [CrossRef] 67. Martin, N.A.; Nawrocki, A.; Molnar, V.; Elkjaer, M.L.; Thygesen, E.K.; Palkovits, M.; Acs, P.; Sejbaek, T.; Nielsen, H.H.; Hegedus, Z.; et al. Orthologous proteins of experimental de- and remyelination are differentially regulated in the CSF proteome of multiple sclerosis subtypes. PLoS ONE 2018, 13, e0202530. [CrossRef] 68. Poliani, P.L.; Wang, Y.; Fontana, E.; Robinette, M.L.; Yamanishi, Y.; Gilfillan, S.; Colonna, M. TREM2 sustains microglial expansion during aging and response to demyelination. J. Clin. Investig. 2015, 125, 2161–2170. [CrossRef] 69. Moyon, S.; Dubessy, A.L.; Aigrot, M.S.; Trotter, M.; Huang, J.K.; Dauphinot, L.; Potier, M.C.; Kerninon, C.; Parsadaniantz, S.M.; Franklin, R.J.; et al. Demyelination causes adult CNS progenitors to revert to an immature state and express immune cues that support their migration. J. Neurosci. 2015, 35, 4–20. [CrossRef] 70. Han, M.H.; Lundgren, D.H.; Jaiswal, S.; Chao, M.; Graham, K.L.; Garris, C.S.; Axtell, R.C.; Ho, P.P.; Lock, C.B.; Woodard, J.I. Janus-like opposing roles of CD47 in autoimmune brain inflammation in humans and mice. J. Exp. Med. 2012, 209, 1325–1334. [CrossRef] 71. Kipp, M.; Nyamoya, S.; Hochstrasser, T.; Amor, S. Multiple sclerosis animal models: A clinical and histopathological perspective. Brain Pathol. 2017, 27, 123–137. [CrossRef] [PubMed] 72. Lieury, A.; Chanal, M.; Androdias, G.; Reynolds, R.; Cavagna, S.; Giraudon, P.; Confavreux, C.; Nataf, S. Tissue remodeling in periplaque regions of multiple sclerosis spinal cord lesions. Glia 2014, 62, 1645–1658. [CrossRef] [PubMed] 73. Steen, C.B.; Liu, C.L.; Alizadeh, A.A.; Newman, A.M. Profiling Cell Type Abundance and Expression in Bulk Tissues with CIBERSORTx. In Stem Cell Transcriptional Networks, Methods and Protocols; Kidder, B.L., Ed.; Humana: New York, NY, USA, 2020; Volume 2117, pp. 135–157. 74. Donovan, M.K.R.; D’Antonio-Chronowska, A.; D’Antonio, M.; Frazer, K.A. Cellular deconvolution of GTEz tissues powers Discovery of disease and cell-type associated regulatory variants. Nat. Commun. 2020, 11, 955. [CrossRef] 75. Browaeys, R.; Saelens, W.; Saeys, Y. NicheNet: Modeling intercellular communication by linking ligands to target genes. Nat. Methods 2020, 17, 159–162. [CrossRef] [PubMed]