Probing the basis of disease heterogeneity in multiple sclerosis using genetically diverse mice

1 Multiple sclerosis (MS) is a complex disease with significant heterogeneity in disease course and 2 progression. Genetic studies have identified numerous loci associated with MS risk, but the genetic basis of 3 disease progression remains elusive. To address this, we leveraged the Collaborative Cross (CC), a genetically 4 diverse mouse strain panel, and experimental autoimmune encephalomyelitis (EAE). The thirty-two CC strains 5 studied captured a wide spectrum of EAE severity, trajectory, and presentation, including severe-progressive, 6 monophasic, relapsing remitting, and axial rotary (AR) -EAE, accompanied by distinct immunopathology . Sex 7 differences in EAE severity were observed in six strains. Quantitative trait locus analysis revealed distinct genetic 8 linkage patterns for different EAE phenotypes, including EAE severity and incidence of AR-EAE. Machine 9 learning-based approaches prioritized candidate genes for loci underlying EAE severity (Abcc4 and Gpc6) and 10 AR-EAE (Yap1 and Dync2h1). This work expand s the EAE phenotypic repertoire and identifies novel loci 11 controlling unique EAE phenotypes, supporting the hypothesis that heterogeneity in MS disease course is driven 12 by genetic variation. 13 14 Summary 15 The genetic basis of disease heterogeneity in multiple sclerosis (MS) remains elusive. We leveraged the 16 Collaborative Cross to expand the phenotypic repertoire of the experimental autoimmune encephalomyelitis 17 (EAE) model of MS and identify loci controlling EAE severity, trajectory, and presentation. 18 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint

19 Multiple sclerosis (MS) is an autoimmune disease of the central nervous system (CNS) characterized by 20 demyelination, gliosis, axonal loss, and progressive neurological dysfunction, representing the leading cause of 21 non-traumatic neurological disability in young adults1. The pathogenesis of MS is not fully understood, but current 22 evidence suggests that activation of myelin-reactive CD4 T cells triggers an inflammatory cascade in the CNS, 23 recruiting other immune cells, which mediate subsequent tissue destruction and pathology 2,3. Significant 24 heterogeneity in disease presentation and severity exists, with disease courses characterized as: (i) clinically 25 isolated syndrome (CIS), denoted as the first inflammatory and demyelinating event preceding MS diagnosis4; 26 (ii) relapsing remitting MS (RR-MS), defined by distinct flar es of disease progression with minimal disease 27 progression between relapses and an overall less severe disease course; (ii i) secondary progressive MS (SP-28 MS), which typically transitions into RR-MS and presents as a progressive deterioration over time; and (i iii) 29 primary progressive MS (PP-MS), which presents as a severely debilitating disease course with early disease 30 progression and poor prognosis5. In addition, significant sex differences have also been noted, with MS incidence 31 being approximately three times higher in women, while disease course has been shown to be more severe in 32 men6,7. The biological underpinnings behind this heterogeneity in disease presentation remain largely unknown, 33 and their characterization should yield prognostic indicators to better inform personalized therapeutic 34 intervention8. 35 Previous studies have shown that up to 30% of MS risk can be attributed to genetic factors9. Early studies in 36 MS families demonstrated a significant genetic component, which was mapped to the major histocompatibility 37 (MHC) locus, with HLA-DRB1*15:01 being the strongest risk allele10,11. Nonetheless, the MHC locus accounted 38 for only a portion of the total heritable risk for this polygenic disease. Subsequent genome-wide association 39 studies (GWAS) have identified 200 non-MHC loci, as well as 32 independent loci within the MHC, associated 40 with MS incidence9. However, despite the strong genetic association with MS disease risk, the genetic basis for 41 MS disease course heterogeneity remains poorly understood . An additional limitation of these association 42 studies lies in that the causality of the identified genes cannot be demonstrated in humans. In this regard, mouse 43 models have been proposed as the next crucial step in the post-GWAS era12. 44 Several mouse models of MS exist, with experimental autoimmune encephalomyelitis (EAE) being the 45 principal immune -mediated model of this disease. This model has been instrumental in improving our 46 understanding of MS pathogenesis at the cellular and molecular levels, as well as developing new disease 47 modifying therapies 13,14. However, animal models of MS, in particular EAE, have been criticized for failing to 48 capture many relevant aspects of the human disease15,16. We propose that this shortcoming is in part due to the 49 failure to consider the importance of genetic heterogeneity that is so pronounced in human populations; a gap 50 that we attempted to address in this study. With regard to genetics approaches in animal models , the most 51 common approach has been to use reverse genetics to delete a candidate gene of interest on a fixed genetic 52 background, typically C57BL/6 (B6). While effective, such genetically modified mice are rarely representative of 53 natural genetic variation present in the human population 17, where variation in many native genomic elements 54 regulates the expression and activity of genes in a time- and tissue-specific manner. Additionally, the candidate 55 gene is pre-selected based on prior knowledge. In contrast, unbiased forward genetics approaches utilize natural 56 genetic variants that control a phenotype of interest, analogous to allelic variants naturally segregating in human 57 populations18. Genetic crosses between two or more populations of interest and subsequent mapping can reveal 58 loci containing genes linked to the phenotype , akin to GWAS in humans . Importantly, the causal effects of 59 candidate genes can be confirmed using a number of approaches, including congenic mapping/positional 60 cloning, targeted genome editing, transgenic complementation, and pharmacological targeting12. 61 While conventional laboratory inbred strains of mice are an important tool in forward genetics, they represent 62 artificially selected organisms originating from a small founder population 19. As such, they lack the range of 63 genetic diversity and selective evolutionary pressure present in human populations. These limitations can be 64 overcome by incorporating into the experimental design so -called wild-derived inbred strains 20, thereby more 65 accurately modeling human genetic diversity. This approach has been used previously in our lab by leveraging 66 wild-derived PWD/PhJ (PWD) mice21. Our initial studies showed that PWD mice have a decreased susceptibility 67 to EAE, associated with a downregulation of proinflammatory and MS -associated genes in immune cells22. To 68 begin to map loci driving these phenotypes, we employed the B6.Chr PWD chromosome substitution (consomic) 69 strain panel, which carry PWD chromosomes on the B6 background 23. Using B6.Chr PWD consomic mice, we 70 demonstrated that PWD-derived alleles profoundly regulated EAE severity, often in a sex -specific manner24,25. 71 While B6.ChrPWD consomic mice could be used as a starting point for mapping of specific gene variants driving 72 EAE phenotypes (e.g. using congenic mapping), this is a laborious process. Further, the genetic diversity, 73 although improved, is still limited to 2 allelic variants per gene (PWD and B6) and captured only classic EAE 74 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint symptomatology (ascending paralysis). To capture a broader spectrum of disease heterogeneity more 75 representative of MS, we turned to the Collaborative Cross (CC) mouse genetic resource. The CC is a panel of 76 multi-parental recombinant inbred lines designed specifically for the analysis of complex and polygenic 77 phenotypes. They represent a unique resource to discover and dissect the exact contributions of genetic, 78 environmental, and developmental components to the etiology of common complex human diseases 26. These 79 mice were generated using 8 founder strains. Five of these are conventional laboratory inbred strains: C57BL/6J 80 (B6), A/J, 129S1/SvImJ (129S1), NOD/ShiLtJ (NOD), NZO/HlLtJ (NZO), and three are wild -derived strains: 81 CAST/EiJ (CAST), PWK/PhJ (PWK), and WSB/EiJ (WSB). Of note, the PWK strain is a close relative of the 82 PWD strain used in our previous studies, as it was developed in parallel from the same wild-derived population21. 83 Collectively, these founder genomes cover ~90% of the known genetic variation in mice. Using a combinatorial 84 breeding design, through a long series of intercrosses between these different strains and their offspring, the CC 85 strains were generated and fixed as homozygotes , allowing for unlimited numbers of repeat phenotypic 86 measurements in genetically identical individuals26,27. 87 Since becoming commercially available, the CC mice have been applied to several research fields, including 88 examining the genetic susceptibility to infectious diseases and control of immunologic phenotypes 28-36, as well 89 as allergic responses 37-40, but to our knowledge this model has not yet been applied to study organ-specific 90 autoimmunity. Here, we developed an EAE induction protocol for use in CC mice, using strains carrying H2b and 91 H2g7 MHC haplotypes. Using this approach, we characterized EAE phenotypes in 32 different CC strains, 92 including: EAE incidence/susceptibility, EAE subtype (classic or atypical), and overall disease course 93 presentation. This revealed a wide variation in EAE phenotypes and identified several strains with clinically 94 relevant phenotypes, including but not limited to: EAE resistance, severe progressive disease, relapsing remitting 95 disease, monophasic disease, and atypical axial rotary (AR)-EAE, as well as the presence of genotype-specific 96 sex differences. Subsequent follow up analysis revealed distinct immunopathology associated with EAE 97 phenotypes of interest. Furthermore, utilization of this approach allowed for quantitative trait loci (QTL) mapping 98 and subsequent candidate gene prioritization, revealing distinct linkage patterns and identifying novel loci and 99 genes controlling unique EAE phenotypes. Taken together, this characterization greatly expands the phenotypic 100 repertoire of the EAE model, bringing the model one step closer to human disease relevance by addressing the 101 role of genetic diversity in disease presentation. Importantly, together with emerging genome -wide studies in 102 humans41,42, these findings strongly support the hypothesis that heterogeneity in MS disease course is driven by 103 natural genetic variation. 104 105

106 MOG35-55 induced EAE in H2 b and H2 g7 Collaborative Cross (CC) strains captures a broad range of clinically 107 relevant disease phenotypes 108 EAE, like MS, is initiated by autoreactive CD4 T cells recognizing myelin antigens presented on major 109 histocompatibility complex (MHC) class II molecules. These autoreactive CD4 T cells are typically elicited by 110 immunization with myelin antigens together with adjuvants such as complete Freund’s adjuvant (CFA) and 111 pertussis toxin (PTX), with the latter likely serving to disrupt the blood -brain barrier 43. S everal different 112 immunogens are used for EAE induction. These include crude mouse spinal cord homogenate ( mSCH), 113 recombinant myelin proteins, or, most commonly, peptides derived from myelin proteins, including myelin 114 oligodendrocyte glycoprotein 35-55 peptide (MOG35-55) classically used in C57BL/6J (B6) mice 8-10. It has been 115 well documented that different strains of mice have varying susceptibility to the EAE peptide immunogens based 116 on the ability of their MHC allelic variants to bind to and present these immunogens44-46. This is potentially 117 problematic to the design of a universal EAE induction protocol for CC strains, as the original 8 CC founder 118 strains have contributed 7 different MHC haplotypes to the resulting CC strains (Table S1). Given this, we initially 119 attempted to induce EAE utilizing mSCH, to capture all potential neuroantigens and therefore provide potential 120 antigen binding across various MHC allelic variants. However, initial experiments in a subset of CC strains 121 demonstrated that mSCH-based induction, although fairly effective in B6 mice, had variable and low penetrance 122 of EAE in CC mice (Fig. S1). 123 As an alternative, we chose a peptide-based EAE induction approach utilizing the classic MOG35-55 peptide, 124 based on the identification of compatible allelic variants at the MHC locus (called H2 in the mouse). Previous 125 studies have shown successful using MOG35-55-based EAE induction in three classic laboratory strains, which 126 were also used as founders for the CC: B6, 129S1/SvImJ (129S1), and NOD/ShiLtJ (NOD)44-46. These strains 127 carry either an H2b (B6 and 129S1) or a H2g7 (NOD) MHC haplotype (Table S1), both which would be expected 128 to be found in the CC strains. Using the available CC genotype data (see Materials and Methods), we determined 129 that a total of 32 out of the 69 commercially available CC strains carry an H2b or H2g7 haplotype at MHC class II 130 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint loci ( Table 1). These 32 CC strains are thus potentially compatible with the MOG35-55 based EAE induction 131 approach and were selected for inclusion in our study (Fig. 1A). 132 Male and female mice (n=5 of each sex), 9-14 weeks old, of each of the selected 32 CC strains, as well as 133 B6 reference controls, were immunized subcutaneously with 200 µg MOG35-55 emulsified in CFA and received a 134 single intraperitoneal injection of 200 ng pertussis toxin (PTX) as an ancillary adjuvant, both administered on day 135 0. Due to availability, CC mice were obtained and studied in 4 separate cohorts (Table 1), each of which included 136

B6 controls. Mice w ere observed daily for a total of 50 days for the presence of clinical disease 137 symptoms using the both the classical EAE24 and a modified axial rotary-EAE (AR-EAE)43 scoring scale as 138 previously described (see Materials and Methods for a detailed description). Daily disease scores (Fig. S2 – S7 139 and Table S2 – S4) were utilized to calculate EAE quantitative trait variables (QTVs) and derive EAE phenotypes 140 (Table 1 and Table S5) (see Materials and Methods). 141 In control B6 mice, this protocol resulted in a high penetrance (~92%) of typical symptomatology manifesting 142 as ascending paralysis (here termed “classic EAE”), with a moderately severe chronic disease phenotype (Fig. 143 1B – D and Fig. 2A and B). In contrast, the CC strains demonstrated a broad range of EAE phenotypes, with 144 the overall incidence of EAE (of any type) ranging from 10 -100% (Fig. 1B; Table 1, and Table S5). Analysis of 145 disease course and EAE phenotypic traits (see Materials and Methods) classified CC strains into unique disease 146 profiles/subtypes (Table 1). Besides classic EAE, a number of strains exhibited a high incidence of atypical AR-147 EAE, manifesting as severe ataxia and axial rotational body movements ( Fig. 1B; Table 1, and Table S5). 148 Additionally, a number of strains exhibited relapsing-remitting (RR)-EAE, or monophasic disease (Fig. 1C; Table 149 1, and Table S5). We used cumulative disease score (CDS; the total sum of daily scores accounting for both 150 classic and AR-EAE subtypes) (see Materials and Methods) as a single quantitative variable capturing overall 151 EAE severity/duration, which ranged greatly across the CC panel, with several strains demonstrating significantly 152 lower CDS compared with B6, and two strains demonstrating significantly higher CDS ( Fig. 1D; Table 1, and 153 Table S5). 154 Because MHC class II alleles are the major genetic determinant of susceptibility to MS10,11,47, we next asked 155 whether the limited H2 haplotypes captured in our subset CC strains influenced EAE incidence and/or severity. 156 Stratifying the CC strains by major H2 haplotype (H2b or H2g7) demonstrated no significant difference in EAE 157 cumulative disease score or incidence of any of the major EAE subtypes (Fig. 1E – K). Further subsetting the 158 H2 haplotype by founder strain of origin (B6, 129S1, or NOD) also did not reveal any significant differences (Fig. 159 S8). 160 While many CC strains manifested clinical EAE presentation similar to B6, several strains captured extreme 161 ends of the different phenotypes studied ( Fig. 2A-D; Table 1 and Table S5). These included diversity in both 162 susceptibility and severity, from nearly completely resistant (CC011), to highly susceptible (CC028); with the 163 latter presenting with rapidly progressing severe disease, with 70% (5/5 males and 2/5 females) reaching 164 quadriplegia/humane endpoint by Day 39 (Fig. 2A). CC004 mice presented with the highest incidence (79%; 165 Fig. 1B) of atypical AR-EAE with a severe chronic disease course ( Fig. 2B), with a significantly higher overall 166 CDS compared with B6 mice ( Fig. 1D). Additionally, several strains exhibited diversity in disease course. Four 167 strains exhibited a ≥ 50% incidence of RR -EAE, with CC002 being the most robust among them ( Fig. 1C and 168 2C). CC043 mice demonstrated a secondary progressive disease course ( Fig. 2D). Several other strains, in 169 particular CC068, presented with a monophasic disease course, which in the case of CC068 was staggered in 170 day of onset by sex (Fig. 1C and 2E). Taken together, these results demonstrate that the genetic diversity in CC 171 mice captures a wide spectrum of clinically relevant EAE phenotypes, with several strains capturing extreme 172 EAE phenotypes of interest for follow-up analysis. 173 MS exhibits well -documented sex differences in both disease incidence (higher in females) and disease 174 progression (more severe in males) 6. Sex differences in EAE have also been reported, predominantly in SJL/J 175 mice and less so in B6 mice 48. With our sample size (n=5 per sex per CC strain), while we were likely 176 underpowered to detect subtle sex differences in individual strains, we could potentially capture larger effects if 177 any were present. A two-way analysis of variance of the effect of strain and sex on CDS (analyzed separately 178 for classic EAE, AR-EAE, or combined disease) revealed a highly significant effect of strain (as expected), no 179 overall effect of sex, and a significant strain by sex interaction in the case of classic EAE CDS ( Fig. 3A-C). A 180 post-hoc analysis of the effect of sex on disease course within each CC strain identified significant and bi -181 directional effects of sex on EAE CDS in a total of 5 different strains across the different disease types (Fig. 3A-182 C, Table 1, and Table S6). Examination of disease course revealed distinct differences for classic EAE in CC046 183 (greater severity in males) (Fig. 3D) and CC042 (greater severity in females) (Fig. 3E), and for AR-EAE course 184 in CC038 (greater severity in males) (Fig. 3F) and CC072 (greater severity in females) (Fig. 3G). Taken together, 185 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint these results demonstrate that the effect of sex on EAE is highly genotype -dependent, indicating the presence 186 of gene-by-sex interactions. 187 Distinct immunopathology in the spinal cord and brain is associated with classic and AR-EAE clinical phenotypes 188 in CC028 and CC004 mice 189 To determine the immunopathological basis of the distinct EAE phenotypes identified in CC mice, we 190 specifically focused on strains exhibiting RR -EAE (CC002), AR -EAE (CC004), and severe progressive EAE 191 (CC028), together with reference B6 controls. In the experiments described above, brain and spinal cord were 192 collected at D50 post EAE induction (or at time of humane endpoint euthanasia), fixed in formalin, and processed 193 for sectioning and staining with hematoxylin and eosin (H&E) or Luxol fast blue (LFB) to assess immune cell 194 infiltration and demyelination, respectively, using a semi-quantitative assessment by a blinded observer, similar 195 to what we have previously published 49,50 (see Materials and Methods). Analysis of spinal cord inflammation in 196 B6 mice revealed expected focal inflammatory infiltrates, which were comparable in CC002 mice, and 197 significantly reduced in CC004 mice (Fig. 4A and B). In contrast, spinal cords from CC028 mice demonstrated 198 significantly greater levels of inflammation compared with B6 reference controls, characterized by extensive and 199 dense infiltration of inflammatory cells (Fig. 4A and B). Surprisingly, the overall extent of demyelination in the 200 spinal cord showed no significant difference in between the three CC strains and B6 reference control (Fig. 4C 201 and D). Assessment of brain pathology revealed a CC004 -specific increase in inflammation compared with B6 202

controls, characterized by prominent perivascular infiltrates and moderate to severe parenchymal 203 infiltration in the cerebellum (Fig. 4E and F). Likewise, LFB-stained CC004 brain sections revealed a significant 204 increase in the level of demyelination, defined by w idespread white matter pallor affecting nearly all ( >75%) of 205 the sample (Fig. 4G and H). Taken together, these data demonstrate that AR-EAE clinical presentation in CC004 206 mice is associated with lesions in the cerebellum rather than the spinal cord, consistent with previous findings 207 on AR-EAE43,51, while severe and rapidly progressing classical EAE in CC028 mice is associated with augmented 208 spinal cord inflammation. 209 Severe EAE in CC028 and CC004 is associated with a greater abundance of myeloid rather than lymphoid cells 210 in the CNS 211 To further characterize the extent of immune infiltration and immunopathology in severe-progressive and AR- 212 EAE phenotypes in CC028 and CC004 mice, EAE was induced and followed for 14 days (D14) in order prevent 213 CC004 and CC028 mice from succumbing to humane endpoints, while also capturing peak disease activity, with 214 disease course recapitulating our previous results above ( Fig. 5A and B). At D14 brain and spinal cord were 215 collected and processed independently for leukocyte isolation, staining, and flow cytometric analysis (see 216

and Methods). Resulting cells were gated as shown in Figure 5C to assess key immune cell 217 populations, including microglial cells (CD45 intCD11b+CX3CR1+), myeloid cells (CD45+CD11b+Cx3CR1low/-), 218 neutrophils (CD45+CD11b+CX3CR1-Ly6G+), B cells (CD45 +CD11b-CD19+), and T cells (CD45 +CD11b-CD19-219 TCRβ+), as well CD4 + T cells producing the two signature cytokines, IFN γ and IL-17, for Th1 and Th17 cells, 220 respectively. Assessment of these populations revealed an increase in total CD11b+ cell frequency in both the 221 brain and spinal cord of CC004 and CC028 compared with B6 mice (Fig. 5D and J). Further examination of this 222 population in the spinal cord revealed no significant differences in the population of microglia, myeloid cells, or 223 neutrophils between the three strains/phenotypes (Fig. 5E-G). However, analysis of these same populations in 224 the brain revealed a significantly greater population of microglia cells in CC028 mice compared with B6 ( Fig. 225 5K). Additionally, we found a significant increase in myeloid cells in the brain of CC004 mice compared with B6 226 (Fig. 5L), which was mostly accounted for by an increase in neutrophils (Fig. 5M). The increased frequencies of 227 myeloid cells in the brain and spinal cord of CC004 and CC028 mice compared with B6 mice were 228 counterbalanced by a lower frequency of B and T cells, reported as frequency of CD45+ cells (Fig. 5H-I, N-O). A 229 similar pattern was seen for IFNγ production by CD4 T cells, which was reduced in CC004 and CC028 compared 230 with B6 mice, in both tissues (Fig. 5P and R). This decrease in Th1 cells was not accompanied by a reciprocal 231 change in Th17 cells, as evidenced by lack of significant differences in IL-17 expression in CD4+ T cells (Fig 5Q 232 and S). Taken together, these findings suggest that the severe classic EAE phenotype of CC028 mice is 233 surprisingly associated with a decrease in lymphocyte abundance in the spinal cord, including greatly reduced 234 Th1 cells. While a similar decrease in lymphocytes is seen in the brain and spinal cord of AR-EAE CC004 mice, 235 there is an additional brain-specific increase in neutrophil infiltration, which is in alignment with the brain-specific 236 increase in inflammation and demyelination observed during histopathological analysis (Fig. 4). 237 Relapsing remitting EAE in CC002 mice is driven by both peripheral immune responses and non-hematopoietic 238 derived factors 239 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint While peripheral immune cells initiate disease in EAE and MS, CNS -intrinsic factors play an important role 240 in regulating disease progression. To determine which of these two distinct mechanisms serve as the basis for 241 the genetically regulated relapsing remitting EAE phenotype in CC 002 mice, reciprocal bone marrow chimera 242 EAE experiments, between MHC-matched control B6 (H2b) and CC002 (H2b) mice w ere conducted (see 243

and Methods). In order to assess chimerism, we used congenic B6 mice carrying the CD45.1 allele 244 (B6.SJL-Ptprca Pepcb/BoyJ), since we determined that CC002 mice carry a A/J-derived haplotype at the Ptprc 245 locus, and therefore the CD45.2 allele. The donor/recipient combinations and respective CD45 alleles are 246 detailed in Figure 5A, and we note that for the CC002→CC002 chimeras, congenic markers could not be used. 247 At 8 weeks post bone marrow ablation and reconstitution, EAE was induced as above and observed for a total 248 of 30 days, at which point spleen and CNS tissues were collected for immunophenotyping by flow cytometry (see 249

and Methods). 250 Flow cytometric analysis using CD45.2 and CD45.1 markers demonstrated successful bone marrow 251 chimerism across all groups assessed, with some variation in chimerism across cell types as expected. A n 252 average of 95.4% (range 86.6 – 99.4%) of total splenic leukocytes were donor derived. For CD11b+ and CD19+ 253 cells, the average chimerism was greater than 98% (Fig. 6B and C) cells. CD4+ and CD8+ T cells displayed an 254 average of 81.2% and 78.2% chimerism, respectively, with increased to persistence of B6 host T cells in the 255 CC002→B6 group ( Fig. 6D and E). Subsequent assessment of donor cell frequency in the spinal cord 256 demonstrated similar trends to those observed in the spleen, most notably the persistence of a significant fraction 257 of host CD4+ and CD8+ T cells in the CC002→B6 group (Fig. 6F and G). 258 Analysis of EAE course demonstrated that phenotypes for the control groups matched our original findings 259 (Fig. 2C), with chronic EAE observed in B6→B6 chimeras and RR-EAE in CC002→CC002 chimeras (Fig. 6H), 260 confirming that the bone marrow ablation and transplantation procedure did not alter these phenotypes. Analysis 261 of the experimental groups demonstrated that bone marrow from B6 donors was sufficient to alter the disease 262 course in CC002 hosts, resulting in a “B6-like” chronic EAE phenotype (Fig. 6H). Meanwhile, transfer of CC002 263 bone marrow into B6 hosts resulted in remitting EAE, although without relapse ( Fig. 6H). Assessment of CNS 264 infiltrating cells by flow cytometry revealed a significant difference between CC002 and B6 mice with a greater 265 frequency of CD11b+ cells (Fig. 6I) and a reduced population of TCRβ+ cells (Fig. 6J) in CC002 mice compared 266 with B6. This reduction in TCRβ+ cells was driven primarily by a significant reduction in frequency of CD8+ cells 267 in the strains with remitting EAE phenotypes (CC002 →B6 and CC002 →CC002) compared to chronic EAE 268 phenotypes (B6→B6 and B6→CC002) (Fig. 6K), an effect that was not observed in the CD4 + cell population 269 (Fig. 6L ). Taken together, these results suggest that while remission of EAE in CC002 mice is intrinsic to 270 peripheral immune system, CNS-intrinsic genetic factors may drive relapse. Alternatively, the lack of relapse in 271 CC002→B6 chimeras could be driven by the significant percentage of remaining host (B6) T cells. 272 QTL mapping reveals distinct loci associated with EAE subtype, course, and severity 273 Besides the identification of novel phenotypes, an advantage of the CC model is the ability to genetically 274 map loci controlling specific phenotypic traits of interest, with mapping power and resolution increasing with the 275 number of unique strains studied. We performed genome -wide association mapping for two major EAE 276 quantitative traits: 1) EAE incidence (Fig. 1B), as a measure of disease susceptibility, and 2) cumulative disease 277 score (CDS; Fig. 1D), as a cumulative measure of disease severity, duration, and incidence. For incidence, we 278 either measured total EAE incidence, or subsetted incidence by: 1) disease subtype (classic EAE and AR-EAE), 279 and 2) disease course (chronic, RR -EAE, and monophasic EAE). Association mapping was performed using 280 R/qtl2 software52, with genome-wide significance logarithm of odds (LOD) thresholds determined by permutation 281 analysis, using a relaxed threshold (given the limited number of strains studied and the complexity of the traits) 282 of 20% to identify top QTL (see Materials and Methods). QTL were designated Eaecc (EAE QTL identified in CC 283 mice) and numbered in the order of their description in the manuscript (Eaecc1-6). 284 The first trait mapped was total EAE incidence (any disease subtype). While no single QTL passed the 285 significance threshold, the lead QTL on proximal Chr4 (Eaecc1; LOD score = 6.79) fell just short of significance 286 (Fig. S9A and B; Table 2). Subsetting EAE incidence by disease subtype and disease course revealed distinct 287 linkage patterns for each classification and identified several QTL, some of which passed the significance 288 threshold. While analysis of classic EAE (i.e. non -AR) incidence yielded a top QTL on Chr5 that fell short of 289 significance (Eaecc2; LOD score = 6.64; Fig. S9C and D; Table 2), AR-EAE incidence mapping revealed a lead 290 QTL on Chr9 (LOD score = 7.96) that passed 20% genome-wide significance (Eaecc3; Fig. 7A; Table 2). Closer 291 examination of the founder effects at this QTL revealed that WSB, and to a lesser extent NZO , alleles were 292 associated with higher incidence of AR-EAE, while NOD alleles were associated with lower incidence (Fig. 7B). 293 Consistent with this, analysis of genotype by phenotype distribution revealed that of the 32 studied CC strains, 294 the top five CC strains with the highest AR-EAE incidence (CC004, CC083, CC084, CC072, and CC038) carried 295 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint either WSB or NZO alleles at the QTL region, aligning with the observed QTL founder effects ( Fig. 7C ). 296 Subsetting EAE incidence by disease course revealed no major associations for chronic EAE ( Fig. S9E), but 297 RR-EAE and monophasic EAE revealed one QTL each on Chr18 (Eaecc4; LOD score = 6.40) and Chr6 (Eaecc5; 298 LOD score = 7.31), respectively, that fell just short of the significance threshold ( Fig. S9F - I; Table 2). Taken 299 together, these results reveal distinct linkage patterns for the incidence of different EAE subtypes or disease 300 courses, suggesting that these phenotypes are controlled by several distinct major loci, further supporting the 301 idea that disease course in MS is genetically controlled. 302 Mapping of total EAE severity (encompassing both classic and AR -EAE presentation), using CDS as the 303 quantitative trait, revealed a distinct linkage pattern, with the top associated QTL on distal Chr14 passing 20% 304 genome-wide significance (Eaecc6; LOD score = 4.65; 20% threshold LOD score = 4.65) ( Fig. 7D; Table 2). 305 Closer examination of the founder effects at this QTL revealed that WSB alleles were associated with more 306 severe EAE, while PWK (and to a lesser extent 129S1 and B6) alleles were protective (Fig. 7E and F). Taken 307 together, these results reveal a novel QTL controlling EAE severity in CC mice, supporting the notion that disease 308 severity in MS is a genetically regulated trait. 309 Candidate gene prioritization nominates candidate genes for QTL controlling EAE traits 310 The ultimate goal of genetic mapping is to identify the causative genes underlying the associated phenotypes 311 of interest, which, in recombinant inbred strain populations like the CC is often impeded by statistical resolution 312 due to large haplotype blocks. This limitation can be overcome by gene prioritization approaches 53-57. Our 313 prioritization analysis focused on the two QTL reaching 20% genome wide significance: AR-EAE incidence (Chr9; 314 95% confidence interval ~9.1Mb) and EAE severity (Chr14; 95% confidence interval ~14.7Mb), as well as the 315 suggestive narrow QTL associated with incidence of monophasic EAE (Chr6; 95% confidence interval ~2.5Mb) 316 (Table 2). While these QTL are moderately high resolution, these loci still contain numerous genes: Chr9 (81 317 genes, 36 protein-coding), Chr14 (130 genes, 19 protein -coding), and Chr6 (68 genes, 32 protein -coding). To 318 prioritize lead candidate genes in an unbiased manner, we utilized a machine learning-based approach that we 319 developed and described previously 55-57, depicted schematically in Figure 8A. Briefly, this approach trains 320 support vector machine (SVM) classifiers to distinguish trait-associated genes from randomly drawn genes using 321 feature vectors derived from tissue-specific connectivity networks ( functional networks of tissues in mouse 322 (FNTM) database)58. The trained models are then asked to classify each positional candidate gene as trait -323 related or not trait-related. Here, we used the top 500 genes (as determined by -log10(p value)) associated with 324 MS from the National Human Genome Research Institute GWAS catalog59 as the training set. We derived feature 325 vectors for training from two tissue-specific mouse gene interaction networks from FNTM: 1) the immune system, 326 the initiator and driver of pathology in EAE and MS) and 2) the CNS, as the target organ in EAE and MS (see 327

and Methods; Table S7). 328 For the QTL associated with AR-EAE incidence on Chr9, this approach prioritized a total of four genes (two 329 immune system-specific and two CNS-specific) passing the false positive rate (FPR) cutoff at 0.05, with the two 330 top prioritized genes being: Yap1 (immune specific) and Dync2h1 (CNS-specific) (Fig. 8B and C; Table 2). For 331 the EAE severity QTL on Chr14, this approach prioritized two genes passing the FPR cutoff of 0.05: Gpc6 (CNS-332 specific) and Abcc4 (both tissues) ( Fig. 8D and E; Table 2). Additionally, Gpc5, which is located adjacent to 333 Gpc6 within the Crh14 QTL, is itself an MS -GWAS candidate, although it was not ranked utilizing the above -334 described approach, due to insufficient connectivity in the networks . Assessment of the narrow QTL on Chr6 335 associated with monophasic EAE incidence prioritized a total of three genes (one CNS-specific and two in both 336 tissues) passing the threshold of 0.05 FPR, with the top prioritized genes in the immune system and CNS 337 identified as Il17ra and Wnk1, respectively, and both genes passing the FPR cutoff of 0.05 in both tissues (Fig. 338 8F and G; Table 2). Gene prioritization for the remaining suggestive QTL, including Chr4 total EAE incidence, 339 Chr5 classic EAE incidence, and Chr18 RR-EAE incidence, identified additional lead candidate genes, including 340 Tox, Klf3, and Fbn2, respectively (Fig. S10 and Table 2). Taken together, this analysis prioritizes several genes 341 as plausible candidates driving the EAE phenotypes of interest via effects on the immune system or the CNS. 342 Because missense variants have a high potential impact on gene function, we identified nonsynonymous 343 single nucleotide variants (nsSNPs) distinguishing the founder alleles exerting the strongest opposing effects at 344 each of the three lead QTL (Fig. 7B and D; Fig. S9G), focusing only on the top prioritized candidate genes (Fig. 345 7). Assessment of nsSNPs segregating between WSB (high AR -EAE incidence) and NOD (low AR -EAE 346 incidence) at the top two prioritized genes (Yap1 and Dync2h1) found no divergent nsSNPs. However, analysis 347 of the remaining genes passing the 0.05 FPR cutoff revealed one nsSNP matching the strain distribution in Birc3 348 (Table 3). For the Chr14 QTL controlling EAE severity, comparison of nsSNPs differentiating WSB (high EAE 349 CDS) from PWK (low EAE CDS) at the top two prioritized genes revealed one nsSNP in Gpc6, and zero in Abcc4 350 (Table 3). For the Chr6 QTL controlling monophasic EAE incidence, comparison of PWK (high monophasic EAE 351 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint incidence) vs. 129S1 (low monophasic EAE incidence) alleles in Il17ra revealed no segregating nsSNPs, while 352 two nsSNPs were identified in Wnk1 (Table 3). Taken together, these results, combined with our prioritization 353 analysis above, highlight potential coding variants driving the EAE phenotypes of interest, to be functionally 354 validated in future studies, although we also do not discount the importance of non-coding regulatory variants. 355 356

357 The genetic basis of disease course and/or pathology in MS remains obscure. To bridge the gap between 358 genetic association studies and the cellular underpinnings of MS, animal models are needed , but most current 359 mouse models fail to address the importance of natural genetic heterogeneity that is so pronounced in human 360 populations. To expand the genetic diversity of the standard EAE model , we leveraged the CC - a highly 361 genetically diverse mouse strain panel. This approach revealed a wide spectrum of distinct EAE phenotypes in 362 CC mice. Immunological profiling of CC strains with clinically relevant phenotypes revealed distinct pathology 363 associated with different EAE profiles . Additionally, QTL mapping and m achine learning -based functional 364 candidate gene prioritization revealed distinct genetic linkage patterns and identified several top candidate genes 365 associated with the different EAE phenotypes captured by the CC mice. Taken together, these results 366 demonstrate that introducing natural genetic diversity into the standard EAE model captures a wide spectrum of 367 clinically relevant EAE phenotypes, providing a novel resource for the research community. Importantly, our 368

also provide support for the hypothesis that heterogeneity in disease course in MS has a strong genetic 369 component. 370 The major risk genes for MS incidence reside in the MHC locus 9,11. While we originally intended to use the 371 full set of six MHC haplotypes represented across the CC population, this was precluded by the low efficacy of 372 the “MHC-agnostic” EAE induction using mSCH ( Fig. S1). Our MOG peptide -based EAE induction approach, 373 while more efficient, cost-effective, and reproducible, allowed the inclusion of only two different compatible MHC 374 haplotypes, H2b from 129S1 and B6 founder strains, and H2g7 from NOD. While we found no significant effect of 375 these two specific haplotypes on EAE phenotypes of interest ( Fig. 1E-K and Fig. S8), this does not rule out a 376 major role of additional MHC alleles in disease progression, which has been suggested in a several small 377 association studies in humans47,60. This can be addressed in future studies using immunization with recombinant 378 encephalitogenic proteins, such as MOG. Another approach could be to generate intercrosses between the H2-379 compatible CC strains in this study and additional CC strains carrying other H2 haplotypes, to create 380 heterozygosity at the MHC locus, a situation more representative of human genetics. Nonetheless, our approach, 381 by limiting the MHC diversity, may have in fact increased our power to detect non -MHC genetic effects across 382 the rest of the genome. 383 Relevant to the expansion of the EAE repertoire, this study represents a novel application of the widely used 384 MOG35-55 EAE induction protocol to achieve a variety of disease course profiles across a panel of genetically 385 distinct mice. Historically, the most common approach to reproducibly modeling a variety of distinct EAE disease 386 profiles has been to utilize varying myelin derived peptides for EAE induction in corresponding susceptible strains 387 of mice 61-63. For example, chronic EAE disease profiles have been most commonly modeled using MOG 35-55 388 induced EAE in B6 mice, while proteolipid protein (PLP) peptide 135-151-induced EAE in SJL/J (H2s) mice has 389 been commonly used to model RR-EAE61. However, with the methods presented here, we were able to model a 390 variety of distinct EAE disease profiles, including both chronic disease as well as RR disease, such as seen in 391 CC002, using a single peptide induction approach, and in many cases on the same H2 background. As such, 392 the CC mice studied represent novel models for various MS disease courses for which adequate model 393 representation has been somewhat limited. These include but are not limited to: CC068, with monophasic EAE 394 as a potential model for clinically isolated syndrome; CC043, as a model for secondary progressive (SP) -MS, 395 given the presence of a relapsing remitting phase that transitions into a progressive increase in clinical severity; 396 CC028, as a potential model for primary progressive (PP) -MS, given the rapid onset and progressive increase 397 in severity of disease; and CC004 with AR -EAE, as a model capturing ataxic symptomology of MS and brain 398 lesions. 399 The occurrence of AR -EAE has been previously documented in several models, although the incidence 400 varies and depends on the induction method. Greer and colleagues found a high penetrance of AR-EAE in 401 PLP109-209 immunized C3H/HeJ mice (H2k), where it is was accompanied by distinct neuropathology, involving 402 predominantly the cerebellum or brain stem 51, consistent with our own findings. Follow -up studies from the 403 Goverman group demonstrated that recombinant MOG-induced EAE in the closely related C3H/HeB (H2k) strain 404

predominantly in AR disease, whereas C3H.SW MHC congenic mice (carrying H2b) lose this phenotype 405 and develop classic EAE, suggesting that AR -EAE incidence in this model is MHC -dependent64. Interestingly, 406 they also found that AR-EAE was associated with higher Th17:Th1 ratios and could be induced more readily by 407 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint adoptive transfer of Th17 cells. Consistent with this, several knockout strains on the B6 background, such as 408 IFN-γ or Socs3-deficient mice, also show increased incidence of AR -EAE and brain pathology, which is often 409 associated with increased neutrophil infiltration 43,65-68. In our own studies, we found a similar increase in brain 410 pathology in CC004 mice, which exhibited predominantly AR-EAE (Fig. 3). However, while we also found a brain-411 specific increase in neutrophil infiltration and a pronounced reduction in T helper IFN production in CC004 mice 412 compared with B6, we did not find a corresponding increase in IL -17 production, which was in fact reduced by 413 about two fold, albeit non-significantly (Fig. 4), suggesting that an increased Th17:Th1 ratio is not likely to be a 414 major driver of AR-EAE in CC004 mice. Importantly, our QTL mapping results identified a non-MHC linked locus 415 driving AR -EAE ( Eaacc3) on Chr9, implicating novel genes in this phenotype (discussed below). F uture 416 mechanistic studies can further dissect whether the AR -EAE phenotype in CC004 mice is immunologically 417 driven, and if so, whether neutrophils and/or Th17 cells play a major role. Importantly, cerebellar pathology and 418 ataxia are a common feature of MS, whereas longitudinally extensive spinal cord pathology, while a major feature 419 of “classic” EAE models, is less common in MS69,70. 420 Other distinct models of EAE disease progression in laboratory inbred strains of mice have been previously 421 described. In particular, secondary progressive disease has been described in MOG 35-55 induced NOD mice 71 422 and mSCH -induced EAE in Biozzi ABH mice 72. Much debate continues as to whether these models truly 423 represent progressive disease representative of MS44, which has more neurodegenerative and less inflammatory 424 characteristics5. While we identified two CC strains with either rapidly progressive severe disease (CC028) or 425 secondary progression following a remission (CC043), we acknowledge that it is as yet unclear whether these 426 represent valid models of progressive MS. In fact, compared with B6 mice, we observed greater histologic 427 evidence of spinal cord inflammation in CC028 at early time points in disease (Fig. 4), suggesting a more severe 428 inflammatory disease, although flow cytometric analysis also revealed reduced lymphocyte frequencies and 429 increased frequencies of CD11b+ cells. Future studies focused on later time points can help to clarify the 430 immunopathologic basis of progressive EAE in CC028 and CC043 mice. 431 In MS, sex differences in incidence (greater in females) and severity (greater in males) have been well 432 documented6,7, but there is limited evidence for gene -by-sex interactions73, and GWAS studies typically use 433 pooled male and female data to maximize power9. In contrast, the EAE model allows one to directly assess the 434 effect of sex on the same fixed genotype, i.e. strain. While no uniform effect of sex was seen across all CC 435 strains when combined, bi-directional sex differences were noted in several individual CC strains. Sex differences 436 in EAE incidence were noted in CC038 (60% incidence in males and 0% incidence in females) and CC072 (0% 437 incidence in males and 60% incidence in females). Similarly, sex differences in EAE severity (CDS) were noted 438 in CC046 (greater severity in males) and CC042 (greater severity in females ). These findings suggest the 439 existence of gene-by-sex interactions, if only in a subset of the studied CC strains, the mechanisms of which can 440 be elucidated in future studies. These findings are fully consistent with our published data in the B6.Chr PWD 441 consomic model24 and studies in classic inbred strains 74, where sex differences are typically present only on a 442 specific genetic background, e.g., in SJL and not B6 mice. As such, our studies provide new genetic models to 443 study sex differences using a uniform EAE induction method. Given that strain -by-sex interactions were only 444 seen in a subset of strains, our QTL mapping was performed using sex as an additive, rather than interactive 445 covariate. Additionally, given the limited number of mice of each sex per strain, we opted not to perform sex 446 segregated QTL analysis. However, future mapping studies could focus on subpopulations (e.g. F2 or CC-RIX) 447 derived from CC strains exhibiting the most robust sex differences and use sex as an interactive covariate, or 448 use sex-segregated QTL analysis, to identify specific QTL and candidate genes interacting with sex. Interestingly, 449 a recent GWAS analysis of MS severity (rather than incidence; discussed further below), report ed two sex-450 specific associations41. 451 One prominent advantage of the CC model is the ability to map loci controlling specific traits of interest, which 452 we applied to different EAE phenotypes. Given the availability of extensive GWAS assessing MS incidence using 453 case-control studies, we could leverage these results for comparison with our own EAE QTL. Candidate gene 454 prioritization of the classic EAE incidence QTL, Eaecc2, identified two genes passing an FPR of 0.05, Klf3 and 455 Rhoh (Fig. S10 C and D), both of which have been associated with MS risk in the 2019 GWAS analysis 9. 456 Furthermore, comparison of all identified genes within EAE QTL (without prioritization) to the MS incidence 457 GWAS, revealed additional shared genes between these analyses, including: Txk/TXK and N4bp2/N4BP2 458 associated with Eaecc2, Chd7/CHD7 and Ints8/INTS8 associated with Eaecc1 (EAE incidence), all identified in 459 the 2019 MS GWAS9, as well as Gpc5/GPC5 associated with EAE severity (Eaecc6), which was identified in an 460 early MS GWAS in 2009 75. Taken together, these results identify overlapping genes driving predominantly 461 disease susceptibility, rather than disease course, in both EAE and MS, supporting a shared genetic architecture 462 between the human disease and its model, in concordance with our previous EAE QTL mapping studies in SJL 463 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint and B10.S mice76,77. With regard to our approach using machine learning-based candidate gene prioritization to 464 identify the top positional candidate genes for each EAE QTL, we note as a caveat that the candidates may be 465 somewhat biased towards those associated with disease risk/susceptibility rather than disease progression. The 466 reason for this is that our approach requires a large training set of “true positive” genes (200-500 genes) known 467 to be associated with the phenotype of interest, typically by GWAS (Table S7). The vast majority of these genes 468 (available from the GWAS catalog) were associated with MS risk, although a small number also represented 469 emerging GWAS hits for MS progression (discussed below). Therefore, our top prioritized candidate genes are 470 likely biologically more associated with MS genes, which may be genetically distinct from progression. Hence, 471 we consider all genes within our QTL as plausible candidates, and we can continue to improve our prioritization 472 using larger training datasets focused specifically on progression, as those data become available in the future. 473 While there is a wealth of genetic associations with MS risk/incidence from large case -control studies, 474 genome-wide association studies of disease severity and/or progression have been more challenging to perform. 475 However, new studies and associations are beginning to emerge41,42,78,79. We used these studies to generate a 476 list of genes associated with MS severity/progression (See Materials and Methods; Table S8) and asked whether 477 any overlapped with our EAE QTL. Of particular interest, this analysis revealed an overlapping gene from the 478 EAE severity QTL Eaecc6, Spry2/SPRY2, which is the closest gene associated with a suggestive hit on Chr13 479 (rs2876767) from the most recent International MS genetics Consortium GWAS analysis of MS severity and 480 progression41. Additionally, our comparison also revealed two overlapping genes associated with Eaecc1 and 481 Eaecc2, Chd17/CHD17 (associated with MS severity 42) and Ppargc1a/PPARGC1A (associated with 482 accumulation of disability in MS78), respectively. Taken together, these studies further highlight the value of using 483 the CC model to map genes associated with MS progression as an orthogonal approach to resource -intensive 484 GWAS in humans. In particular, mouse QTL studies could prioritize/refine the mapping of the candidate genes 485 from human GWAS, as in the example of Spry2/SPRY2, above. 486 Beyond overlapping MS GWAS genes, candidate gene prioritization for promising QTL, those passing 20% 487 genome wide significance (AR -EAE incidence - Eaecc3, and EAE severity - Eaecc6) or having a particularly 488 narrow 95% confidence interval (2.52Mb, monophasic EAE incidence - Eaecc5) revealed genes with relevance 489 to MS pathology. Of note, assessment of Eaecc6 identified Abcc4/ABCC4, encoding the ATP-binding cassette 490 transporter ABCC4, as the top prioritized gene in both the immune system and CNS tissue networks. Relevant 491 to MS pathology, ABCC4 has been implicated in blood brain barrier function 80 and effector immune cell efflux 492 activities, the latter of which has been shown to be involved in the pathogenesis of rheumatoid arthritis (RA) 81, 493 an autoimmune disease that shares a genetic burden with MS82,83. While the role of ABCC4 in MS is still unclear, 494 other members of the ABCC subfamily, specifically ABCC1 and ABCC2, have been shown to involved in 495 neuroinflammation80, with previous studies revealing increases in ABCC1 in the brain tissue of MS patients 496 compared to healthy controls84. Taken together, these findings represent a plausible mechanism in which variants 497 in Abcc4/ABCC4 could contribute to increased inflammation, acting either within the CNS or peripheral immune 498 system to drive disease severity. 499 In terms of AR -EAE incidence ( Eaecc3), Yap1/YAP1, which is a key component of the Hippo signaling 500 pathway85, and Dync2h1/DYNC2H1, which encodes a dynein motor protein that has been shown to be involved 501 in intraciliary transport, were the top prioritized genes for Eaecc3 for the immune system and CNS networks, 502 respectively. Recent studies have suggested the involvement of the Hippo signaling pathway in autoimmune 503 pathogenesis, with emphasis on the balance between T regulatory cell (Treg) and pro -inflammatory Th17 cell 504 differentiation through Yap-Taz expression85, as well as specific evidence for involvement in RA, inflammatory 505 bowel disease, and psoriasis86, all of which have been shown to have genetic overlap with MS 82,83. In contrast, 506 less direct evidence for involvement in EAE/MS pathogenesis exists for Dync2h1/DYNC2H1. However, 507 Dync2h1/DYNC2H1 has been shown to have a role in neurodevelopment, maintenance, neuronal transport87-89, 508 and has been shown to be associated with retinopathies 90, suggesting a potential involvement in 509 neurodegeneration. In further support of this, Dync2h1 expression was downregulated in the spinal cord of mice 510 infected with Theiler’s murine encephalomyelitis virus (TMEV)91, which has been used to model certain aspects 511 of MS, including axonopathy. Taken together, these candidate genes could represent a potential mechanism 512 behind AR-EAE incidence through immune dysregulation via various impacts of altered Yap1 expression setting 513 the stage for increased inflammation, and/or axonopathies via impaired transport associated with Dync2h1 514 resulting in impaired neurological function. 515 Analysis of Eaecc5 revealed Il17ra/IL17RA, encoding the receptor for the pro-inflammatory cytokines IL-17A 516 and IL-17F, and Wnk1/WNK1, encoding a serine/threonine protein kinase that is involved in CNS signaling, as 517 the top prioritized genes in the immune system and CNS networks, respectively. The association of Il17ra/IL17RA 518 with monophasic EAE incidence is particularly plausible, given the known increased Il17ra/IL17RA expression 519 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint in the CNS 92 and involvement of IL -17 and Th17 cells in MS and EAE pathogenesis 93,94. Specifically, studies 520 have shown that the frequency of Th17 cells is increased during MS relapses 95 and that assessment of serum 521 IL-17A levels in RR -MS patients who were non -responsive to interferon-β therapy revealed that those with 522 elevated IL-17A had shorter disease duration then those with lower IL -17A serum levels96. Furthermore, IL-17 523 has been implicated as having a significant role in the pathogenesis of both clinically isolated syndrome and 524 early MS97,98. Altogether, our findings of the association of Il17ra with monophasic EAE incidence is in support 525 of these previous findings, and the monophasic EAE strain CC068 represents another potential model to further 526 elucidate the role of IL -17 in MS onset and duration. In terms of CNS expression of Wnk1/WNK1, it has been 527 well accepted that Wnk1/WNK1 is involved in pathogenic signaling in the CNS 99 and has been implicated as a 528 crucial component in neuronal axon development and maintenance 100. While direct implication of Wnk1/WNK1 529 in MS pathogenesis has been limited to suggestive involvement in phosphorylation differences in B cells as 530 associated with MS susceptibility alleles 101, the involvement of Wnk1 in CNS dysregulation could represent a 531 potential mechanism for association with monophasic EAE incidence. Interestingly, we identified two 532 nonsynonymous SNPs distinguishing the two major phenotype-associated Wnk1 alleles, both of which change 533 the position of threonine residues ( Table 3), providing a potential mechanism of differential post -translational 534 regulation by upstream serine/threonine kinases 102. Additionally, given the regulation of WNT signaling by 535 WNK1103, another intriguing link between our candidate gene prioritization of Wnk1/WNK1 and MS comes from 536 a recent GWAS study of MS relapse hazard, which identified a minor allele in WNT9B as the lead candidate 537 gene driving relapse risk104. 538 Our study is far from the first to apply forward genetics in the mouse to EAE. A number of groups have applied 539 classic forward genetics approaches, such as F2 crosses or congenic mapping, using intercrosses between two 540 susceptible and resistant standard inbred strains (B10.S and SJL in the majority of the studies), as reviewed in 541 Olsson et al., 2006105. All together, these studies identified a large number of QTL, but, with a few exceptions106, 542 the causative genes have not identified. In part, this is due to the large size of the identified EAE QTL, and thus 543 limited resolution. The other possibility is that the polygenic nature of EAE phenotypes made it difficult to 544 physically map candidate gene variants in cases where the genetic effects were driven by combined modest 545 effects of several allelic variants or involved epistatic interactions76. Another limitation of these approaches is the 546 use of classic inbred strains of mice, which provided limited allelic diversity, compared with CC model, which 547 provides up to 8 different alleles, 3 of them wild -derived and divergent from classic strains. In this regard, we 548 note that the majority of the allelic effects in our CC EAE QTL are driven by one or more wild-derived haplotypes 549 (Fig. 7 and Fig. S9), which in previous studies have been observed to exert large phenotypic effect sizes, 550 facilitating mapping efforts 107. The CC offers additional advantages over traditional mapping populations, 551 including the ability to generate CC intercross (CC-RIX), or F1, mapping populations, and using in parallel the 552 Diversity Outbred (DO) population, which carries the same founder allelic variants but with a very high degree of 553 recombination, allowing for higher resolution mapping comparable to human GWAS 27. This, combined with the 554 growing availability of bioinformatic tools and shared resources, and extensive and still growing human GWAS 555 databases, makes the CC and the DO promising systems genetics tools to continue to dissect the genetic 556 architecture of MS and other complex autoimmune diseases , as evidenced by a recent study of systemic 557 autoimmunity induced by silica exposure in the DO108. 558 Our study is the first (to our knowledge) to apply the CC resource in an autoimmune model of MS. However, 559 the Threadgill and Brinkmeyer-Langford groups have utilized the CC to study neurological disease induced by 560 TMEV CNS infection in a number of studies109-113. These investigators included strains of multiple H2 haplotypes, 561 and thus the overlap with strains in our own study is limited . However, they did identify CC002 and CC011 as 562 strains with minimal neurological progression 109, which is somewhat in line with the EAE phenotypes of these 563 strains (RR-EAE and resistant, respectively). Additionally, sex differences in TMEV induced disease also varied 564 by strain 109, in line with our findings in EAE. While these studies were likely underpowered for QTL mapping 565 (ranging from 5-18 CC and/or CC-RIX strains per study), and complicated by the fact that host genetic variation 566 impacts viral clearance in this model, they nonetheless provide strong evidence that neurological sequelae 567 following adaptive immune inflammation and demyelination in the CNS are genetically regulated, in full 568 agreement with our own EAE studies. 569 Our identification and initial characterization of EAE phenotypes in CC mice represents what we hope will be 570 the first of many utilizing the CC resource in the pursuit of the genetic underpinnings behind MS disease 571 heterogeneity. An obvious extension of this will include the validation and functional characterization of the 572 identified candidate genes, as well as further mapping efforts to increase the resolution and statistical 573 significance of EAE QTL. Because expression QTL (eQTL) often underlie phenotypic QTL, and as such are often 574 used to help fine-mapping in human GWAS, future EAE QTL mapping could be facilitated by the generation of 575 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint immune and CNS cell specific eQTL databases for CC and Diversity Outbred populations. Similarly, PheWAS 576 approaches and/or strain correlations could be used to find colocalized traits and intermediate phenotypes, such 577 as immune or neurological phenotypes, with the latter enhanced by the ever growing immune phenotyping 578 studies in CC strains as a key resource 36. Other efforts will focus on further mechanistic characterization of 579 specific EAE disease course attributes unique to subsets of CC strains as described here. Importantly, the work 580 presented here provides to the research community an easily accessible improved animal model of MS that 581 accounts for host genetic diversity and captures novel phenotypes lacking in traditional models. 582 583

584 Mice 585 For peptide-based EAE studies, male and female mice of each of the 32 Collaborative Cross (CC) strains 586 (Table 1) were purchased between 2021 and 2022 from the Mutant Mouse Resource and Research Center 587 (MMRRC) at University of North Carolina at Chapel Hill, an NIH -funded strain repository, in collaboration with 588 the UNC Systems Genetics Core Facility (UNC; Chapel Hill, North Carolina, USA), with the exception of CC020, 589 which was purchased from The Jackson Laboratory (Bar Harbor, Maine, USA). While CC strains were originally 590 generated and bred at Oak Ridge National Laboratory (USA), the International Livestock Research Institute 591 (Kenya)/Tel Aviv University (Israel), or Geniad Ltd (Australia),all CC strains had been maintained at UNC (or 592 transferred from UNC to The Jackson Laboratory and then back to UNC for clean rederivation) for 10 years from 593 these original diverse locations 114-117. All CC strains are referred to in the text by their abbreviated name 594 (“CC###”); full strain names are provided in Table 1. Due to availability and the number of mice studied, mice 595 were obtained and studied across four cohorts – denoted as C1 – 4 (Table 1). To reduce confounding by batch 596 effects, male and female B6 mice were purchased from Jackson Laboratory to serve as reference controls in 597 each cohort. Once at the vivarium at the University of Vermont, mice were rested for a range of 14 – 29 days, 598 depending on quarantine requirements, prior to experimentation. For follow up studies utilizing CC002, CC004 599 and CC028, including CNS flow cytometry and reciprocal bone marrow chimeras, three females and two males 600 of each strain were purchased from the MMRRC in collaboration with the UNC Systems Genetics Core Facility 601 in 2022 and bred at the vivarium at the University of Vermont until sufficient numbers of offspring were obtained 602 for experimentation. The experimental procedures used in this study were approved by the Institutional Animal 603 Care and Use Committee of the University of Vermont. 604 Selection of CC Strains for Peptide-induced EAE 605 To identify CC strains that are compatible with myelin oligodendrocyte glycoprotein 35 -55 peptide (MOG35-606 55) induced experimental autoimmune encephalomyelitis (EAE), the H2 haplotype of each strain was 607 characterized. Founder strain contributions at the H2 locus, specifically in the region encompassing MHC class 608 I (H2K only) and all class II genes, as defined by Chr17: 33,918,830–34,347,345 bp, were identified using the 609 UNC Systems Genetics Collaborative Cross Viewer tool (Chapel Hill, North Carolina, USA), using sequenced 610 and most recent common ancestor ( MRCA) genotypes. Given that previous studies have suggested MOG35-55 611 binding compatibility with H2b (C567BL/6J ( B6) and 129S1 /SvlmJ (129S1) ) and H2g7 (NOD/ShiLtJ (NOD) ) 612 haplotypes44-46, CC strains with founder contributions, either homozygous or heterozygous, from the three 613 abovementioned strains were selected for study, for a total of 32 investigated CC strains (see Table 1 for strain 614 list and corresponding H2 haplotype). 615 Mouse Spinal Cord Homogenate Induced EAE Studies 616 To investigate the use of mouse spinal cord homogenate (mSCH) as a potential EAE immunogen in CC mice, 617 three CC strains (CC003/Unc, CC008/GeniUnc, and CC010/GeniUnc) with different H2 haplotypes, and four CC 618 founder strains (129S1, A/J, B6, and NOD) were selected for analysis. CC mice were purchased in 2020 from 619 the UNC Systems Genetics Core Facility 114,116, then bred and housed within the vivarium at the University of 620 Vermont until sufficient numbers were obtained for experimentation. CC founder strains were purchased from 621 The Jackson Laboratory and then rested for two weeks prior to experimentation. 622 mSCH was prepared from Swiss Webster mice (Charles River Laboratories) and EAE was induced using a 623 modified approach from previously described22,25. Briefly, mice were injected subcutaneously with 0.15 ml of an 624 emulsion containing 5 mg mSCH in PBS and 50% complete Freund’s adjuvant (CFA; Sigma, USA), 625 supplemented with additional 4 mg/ml heat-killed Mycobacterium tuberculosis (Difco Laboratories, USA). On D0 626 or D0 and D2, mice were administered an intraperitoneal injection of 200 ng pertussis toxin (PTX; List Biological 627 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Laboratories, USA) as an ancillary adjuvant . At 7 days post induction , mice were scored daily for presence of 628 clinical disease symptoms as previously described22,24, and expanded below. 629 Peptide-based EAE Induction and Scoring 630 Approximately five male and five female mice, 8-14 weeks old, of each of the 32 CC strains were immunized 631 subcutaneously with 200 µg MOG35-55 (New England Peptide) emulsified in CFA (Sigma), supplemented with 632 additional 4 mg/ml heat -killed Mycobacterium tuberculosis (Difco Laboratories ), and received a single 633 intraperitoneal injection of 200 ng PTX (List Biological Laboratories) on D0 as an ancillary adjuvant. Mice were 634 studied across a total of four cohorts. Five male and five female B6 mice per cohort, immunized as above, served 635 as reference controls for cohort effects and EAE induction. 636 Mice were observed daily for a total of 50 days starting at 5 days post induction for the presence of clinical 637 disease symptoms, using both classic EAE22,24 or a modified axial rotary (AR)-EAE scoring scale43. Briefly, classic 638 EAE clinical scores were assigned as follows: 0, asymptomatic; 1, tail paralysis; 2, t ail paralysis and hind limb 639 weakness; 3, h ind limb paralysis ; 4, hind limb paralysis with incontinence ; and 5, moribund/quadriplegic. AR-640 EAE clinical scores were assigned as follows: 0, asymptomatic; 1, slight head tilt; 2, pronounced head tilt; 3, 641 inability to walk in a straight line; 4 mouse is moving/lying on its side, will continuously fall to its side after being 642 made to stand; 5, mouse rolls or spins continuously - axial rotation. Mice reached humane endpoints after 643 presenting with a score of 5 (classic or AR -EAE) for 72 hours, at which point mice were euthanized and their 644 daily disease score was recorded as a 5 (classic or AR-EAE depending on presentation) for the remainder of the 645 experiment. Similarly, mice that died after having presented with EAE for more than two continuous days were 646 given a daily disease score of 5 for the remainder of the experiment (designated as classic or AR-EAE depending 647 on previous presentation). Mice that died without any EAE clinical signs or having only presented with EAE for 648 two or fewer days were excluded from the study. 649 EAE Disease Phenotype/Quantitative Trait Variable (QTV) Classification 650 Raw daily disease scores, reported as both classic and AR-EAE as described above, for the full 50-day time 651 course were pooled by strain and sex upon completion of all four cohorts. Raw daily disease scores were utilized 652 to derive the following daily score classifications: classic disease score (reports disease scores that only 653 correspond to the classic-EAE scoring system and negates/assigns a score of 0 to any AR score if present in 654 the raw score; Table S3), AR disease score (reports disease scores that only correspond to the AR-EAE scoring 655 system and negates/assigns a score of 0 to any classic score if present in the raw score; Table S4), and disease 656 score (reports disease scores regardless of EAE subtype and derives an average score for any occurrence of 657 simultaneous classic and AR-EAE scoring, i.e. a mouse that was reported to have a classic score of 3 and an 658 AR score of 2 on the same day was assigned a score of 2.5; Table S2). Subsequently, these three disease score 659 classifications (disease score, classic disease score, and AR disease score) were utilized to determine 660 cumulative disease score (CDS; the total sum of all daily disease scores), classic cumulative disease score 661 (classic-CDS; the total sum of all daily classic disease scores), and AR cumulative disease score (AR-CDS; the 662 total sum of all daily AR disease scores), respectively. CDS, classic -CDS, and AR-CDS were calculated per 663 mouse and averaged by sex and strain to allow for analysis of both strain and sex effects. 664 To assess incidence of EAE and EAE subtypes, the following quantitative trait variables (QTVs) were 665 calculated per individual mouse: EAE incidence, classic EAE incidence, AR-EAE incidence. EAE incidence was 666 determined utilizing daily disease scores and classified as a minimum of two consecutive days of a score above 667 0. Classic EAE incidence and AR-EAE incidence were assessed in a binary and mutually exclusive manner, as 668 determined by raw daily disease score. First, mice were categorized into three groups: classic EAE only (mice 669 that only presented with clinical presentations aligning with the classic EAE scoring system), AR-EAE only (mice 670 that only presented with clinical presentations aligning with the AR -EAE scoring system), and “combined type” 671 (mice that presented with clinical presentations of both the classic and AR -EAE scoring system, either on the 672 same or different days during the experiment). Mice that were classified as being in the classic EAE only or AR-673 EAE only group were assigned as either classic EAE incidence or AR EAE incidence, respectively. For mice 674 classified as being in the “combined type group”, EAE subtype incidence was assigned as classic EAE incidence 675 if the number of days in which a mouse presented with only a classic EAE score (excluding days of simultaneous 676 classic and AR-EAE scores) was greater than the number of days in which a mouse presented with only an AR-677 EAE score (excluding days of simultaneous classic and AR-EAE scores). The inverse was utilized to assign AR-678 EAE incidence. For all incidence calculations, individual mice were assigned a value of 0 or 1 given the absence 679 or presence of the assigned trait, respectively. To assess incidence by strain and sex, values (assigned 0 and 1) 680 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint per QTV were tallied by either strain or sex and then divided by the corresponding number of mice per strain or 681 sex and multiplied by 100 to calculate percent incidence. 682 To determine EAE disease course phenotypes, individual mice were classified as having either a relapsing 683 remitting (RR), monophasic, or chronic EAE disease course as assessed using daily disease scores (Table S2). 684 RR-EAE was defined as a scoring pattern in which a mouse presents with an initial disease period (defined by 685 at least two consecutive days of a disease score >0), followed by remission period ( characterized by disease 686 score of 0 for three or more consecutive days ), and proceeded by a relapse period of at least two consecutive 687 days of a disease score >0. Mice characterized as having RR -EAE could have one o r more remission and 688 relapse cycles. Classification of monophasic EAE was defined as a scoring pattern in which a mouse presented 689 with a consistent disease score of 0 after having presented with a disease score >0 for at least two consecutive 690 days. Chronic EAE was defined as a scoring pattern in which a mouse presented with persistent disease scores 691 >0 from time of disease onset until experiment termination (or humane endpoint) that could not be classified as 692 either RR-EAE or monophasic EAE. Percent incidence of RR, monophasic, and chronic EAE by strain and sex 693 was determined as described above. 694 CNS Histopathology 695 On day 50 post EAE induction, or upon reaching humane endpoint criteria, mice were euthanized, and brain 696 and spinal cord tissues were collected for histopathological assessment as previously described 49. In brief, the 697 skull and vertebral column were removed, and diffusion fixed in 10% neutral buffered formalin (Fisher Scientific, 698 USA). Post fixation, brain and spinal cord were extracted from calvaria and vertebral columns, respectively, and 699 dissected into thirds based on overarching anatomical region (brain: front brain, mid brain, and hind brain – 700 including cerebellum; spinal cord: cervical, thoracic, and lumbar). Tissues were subsequently embedded in 701 paraffin, sectioned (coronal and longitudinal for brain and spinal cord, respectively), and stained with luxol fast 702 blue (LFB) and/or hematoxylin and eosin (H&E) by the pathology department at the University of Vermont 703 Medical Center. 704 Histopathological assessment was conducted in a blinded fashion and evaluated representative areas of the 705 brain and spinal cord corresponding to the above -mentioned regions. Brain and spinal cord regions were 706 assessed for degree of inflammation and demyelination using a semi -quantitative scale adapted from previous 707 studies49,50. Inflammation was evaluated using H&E stained tissues and scored on a scale of 0 – 4 based on 708 severity using the following criteria: 0, no inflammation; 1, f ew inflammatory cells scattered or in small clusters; 709 2, o rganized clusters of inflammatory cells without significant extension beyond small lesions ; 3, s ignificant 710 organized clusters of inflammatory cells with patchy infiltration of surrounding tissue, central involvement of larger 711 lesions; and 4, e xtensive and dense infiltration of inflammatory cells with affecting over half of the sample, +/ - 712 diffuse gliosis. Extent of demyelination was evaluated from tissues stained with LFB (co-stained with H&E) and 713 scored on a scale of 0 – 4 based on severity using the following criteria: 0, no demyelination – deep blue staining; 714 1, small, patchy area(s) of white matter pallor on LFB, no well -defined lesions; 2, defined area of white matter 715 pallor forming isolated lesion(s); 3, confluent foci of white matter pallor on LFB with some spared areas in the 716 sample; and 4, w idespread white matter pallor on LFB affecting nearly all of the sample (over ~75%) . 717 Inflammation and demyelination scores were assigned to each of the three dissected regions per tissue (brain 718 and spinal cord)/mouse. Overall inflammation and demyelination scores were reported per mouse for both brain 719 and spinal cord and were determined by the highest scored region of that tissue. 720 Flow Cytometry 721 To assess CNS infiltrating cells, mice were anesthetized under isoflurane, euthanized by exsanguination via 722 transcardial perfusion with PBS, and brain and spinal cord tissues were collected and processed independently 723 for flow cytometric staining as previously described118,119. In brief, brain and spinal cord tissues were mechanically 724 dissociated by Dounce homogenization until a single cell suspension was created . Resulting cell suspensions 725 were filtered with a 70-μm strainer followed by Percoll gradient (37%/70%) centrifugation and collection of the 726 interphase for subsequent processing dependent on the objective of flow cytometry analysis. 727 For intracellular cytokine analysis, cells were stimulated with 5 ng/mL PMA, 250 ng/mL ionomycin, and 728 brefeldin A (Golgi Plug reagent; BD Bioscience, USA) for 4 hours prior to staining. For flow cytometry analysis of 729 intracellular cytokines and surface staining, cells were stained with UV-Blue LIVE/DEAD fixable stain (Invitrogen, 730 USA) and then surface stained with antibodies against the following: CD45, CD19, CD11b, TCRβ, CD4, CD8, 731 Ly6G, and CX3CR1 (BioLegend, USA). Cells were then fixed and permeabilized with 1% paraformaldehyde 732 (Sigma Aldrich, USA) and 0.05% saponin permeabilization buffer and labeled with antibodies against IFN γ, IL-733 17A, and GM-CSF (BioLegend). 734 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint For flow cytometry staining to assess chimerism in addition to CNS infiltrating cells in CC002 and B6 mice, 735 spleens were collected at time of euthanasia in addition to spinal cord. Spinal cord tissues were processed as 736 described above and spleens were processed as previously described 118. Briefly, spleens were mechanically 737 dissociated by a syringe plunger between two pieces of mesh to create a single cell suspension followed by red 738 blood cell lysis by incubation in 0.8% ammonium chloride solution (STEMCELL Technologies) and subsequent 739 staining. Cells were stained with UV -Blue LIVE/DEAD fixable stain (Invitrogen) and then surface stained with 740 antibodies against the following: CD45 .1, CD45.2, CD19, CD11b, TCRβ, CD4, CD8, Ly6G, and CX3CR1 741 (BioLegend, USA) prior to fixation with 1% paraformaldehyde (Sigma Aldrich). 742 All stained cells were analyzed utilizing a Cytek Aurora and SpectroFlo software versions 2.2 – 3.3 (Cytek 743 Biosciences, USA) and spectral unmixing was performed with appropriate single -color controls and 744 autofluorescence correction from an unstained control group. Flow cytometry data analysis was performed using 745 FlowJo software versions 10.8.1 – 10.10 (BD Biosciences). 746 Reciprocal Bone Marrow Chimeras 747 To identify CD45 allele status in CC002 mice, founder strain contributions for the gene Ptprc 748 (Chr1:137,990,599-138,103,446 bp) were identified using the UNC Systems Genetics Collaborative Cross 749 Viewer tool (Chapel Hill, North Carolina, USA), using both sequenced and MRCA genotypes. Recipricoal bone 750 marrow chimeras were generated between B6.SJL-Ptprca Pepcb/BoyJ (B6.CD45.1) and CC002 mice as 751 previously described118. In brief, 9-15 week old recipient mice were irradiated twice with 550 rads 4-6 hours apart. 752 Post irradiation, mice were injected via retro-orbital vein with 4x106 whole bone marrow cells from respective 753 unmanipulated age- and sex-matched donors. Resulting chimeric mice were rested for 8 weeks to allow for 754 maximal immune reconstitution, at which point EAE was induced as described above and disease course was 755 observed for a total of 30 days. At 30 days post EAE induction, mice were euthanized, and spleen and spinal 756 cord tissues were collected to assess rate of chimerism and CNS immune cell profiles via flow cytometry as 757 described above. 758 Quantitative Trait Loci (QTL) Mapping 759 To map genetic variants that are associated with EAE QTV (including: CDS, total EAE incidence, AR -EAE 760 incidence, classic EAE incidence, RR-EAE incidence, monophasic EAE incidence, and chronic EAE incidence), 761 the R package, R/qtl252, was utilized to preform quantitative trait loci (QTL) mapping. QTVs were calculated as 762 described above. For QTL mapping with CDS, individual mouse data was utilized and underwent covariate batch 763 correction, using experimental batch as a covariate, and rank Z normalization to address potential batch effects. 764 For QTL mapping addressing incidence QTVs, strain incidence percentages were utilized, and rank Z 765 normalization was performed. CC genotype probabilities and kinship matrices were derived utilizing the CC 766 genome “sequenced” data set available from the UNC Systems Genetics Core Facility (Chapel Hill, North 767 Carolina, USA ( http://csbio.unc.edu/CCstatus/CCGenomes/#genotypes). Resulting Manhattan plots 768 demonstrating logarithmic of odds (LOD) traces and significance thresholds of 20% genome wide significance 769 were generated given analysis utilizing 1000 permutations. Allele effect plots were generated for lead QTL for 770 each QTV. For analysis of large QTL effects, QTV specific allele effect plots were generated in addition to classic 771 allele effect plots, allowing for subsequent CC strain-specific genotype by phenotype analysis. 772 Candidate Gene Prioritization 773 To identify and prioritize potential causative genes associated with the Eaecc QTL (Table 2) in an unbiased 774 manner, we leveraged a machine learning-based approach that we developed and described previously55-57. We 775 trained support vector machines (SVM) to classify randomly selected genes from previously identified MS GWAS 776 genes as reported by the National Human Genome Research Institute GWAS catalog 59. We used the top 500 777 as determined by -log10(p value) for training ( Table S7). Mouse orthologs were identified for all genes in the 778 training set and positional candidates were removed so as not to be used for SVM training. Feature vectors for 779 training the SVMs were based on connection weights in one of two functional networks of tissues in mouse 780 (FNTM) network – hemolymphoid system, as a proxy for the immune system, or CNS. The feature vector for a 781 single true positive gene consisted of its connection weights to all other true positive genes in the training set. 782 Any genes without connections to other genes in training set were trimmed off, resulting in a total of 271 and 273 783 positive-labeled genes for training the immune system and CNS networks, respectively ( Table S7). For each 784 tissue-specific network we trained 100 independent SVMs. Each SVM was trained to classify the roughly 271 785 positively labeled MS genes from a matched set of randomly drawn genes from outside the set of GWAS 786 candidates. All trained SVMs were then used to classify positional candidate genes in each QTL. The final score 787 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint for each gene was the -log10 of it false positive rate (FPR) averaged across all 100 SVMs. FPR for gene x was 788 defined as the following: FPRx = FP/(FP + TN), where FP is the number of false positive genes and TN is the 789 number of true negative genes using a cutoff score equal to that of gene x. Significance for prioritization analysis 790 was determined using an FPR of 0.05. 791 To determine direct overlap between the identified candidate genes associated with EAE QTL and mouse 792 orthologs of genes associated with MS risk by GWAS, we compared all genes within a given EAE QTL to the 793 complete list of reported genes from the 2019 GWAS analysis 9, as well as the genes associated with MS 794 susceptibility from the original list of the top 500 genes used in the SVM training pipeline ( Table S7). Similarly, 795 we generated a list of mouse orthologs of genes associated with MS severity/progression based on recent 796 studies41,42,78,79 (Table S8), to determine overlap between the identified candidate genes associated with EAE 797 QTL and genes associated with MS severity. 798 Statistical Analysis 799 Statistical analysis not pertaining to QTL mapping and candidate gene prioritization were carried out using 800 GraphPad Prism software, versions 9.1.2 – 10.2.1. Details of the analyses are provided in the figure legends; 801 including the specific tests used to assess the significance of the observed differences and indication of 802 adjustments for multiple comparisons when appropriate. Comparisons were assessed for effects between B6 803 and CC strains or within strain sex effects as indicated. All center values represent the mean, and error bars 804 represent the standard error of the mean. A P value <0.05 was considered significant. Comparisons are indicated 805 by the brackets and P values are reported using asterisks where significant (* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, 806 **** p ≤ 0.0001). A lack of an indicated asterisk p value therefore signifies a lack of a significant difference. 807 808 Data Availability 809 The data underlying the figures and tables presented are available in this article and its supplementary 810 material. 811 812

813 The authors would like to acknowledge the University of Vermont Medical Center Histology Lab, the Flow 814 Cytometry and Cell Sorting facility, the Department of Pathology and Laboratory Medicine at the University of 815 Vermont Larner College of Medicine for the use of their facilities and resources. They also acknowledge the 816 University of North Carolina Systems Genetics Core Facility, and in particular the core director Darla Miller for 817 her advice and help coordinat ing mouse experiments. Additionally, the authors would also like to specifically 818 acknowledge Dr. DeWitt in the Department of Pathology and Laboratory Medicine for consulting on the guidelines 819 used for histopathology scoring , as well as the members of the Krementsov Lab over the course of this study 820 (Theresa Montgomery, Ph.D , Bristy Sabikunnahar, Ph.D , Dan Peipert, and Sydney Caldwell) for their assistance 821 in tissue harvests. 822 This work was supported by RG-1901-33309 from the National Multiple Sclerosis Society as well as 823 R21AI145306 and R01AI172166 from the NIH | National Institute of Allergy and Infectious Diseases to DNK . 824 With additional support by U19AI100625 from the NIH | National Institute of Allergy and Infectious Diseases to 825 MTF. Additional support for EAH was provided by 5T32AI055402-08 from the NIH | National Institute of Allergy 826 and Infectious Diseases. The distribution of the CC mice used in this study was supported by U42OD010924 827 from the NIH to the Mutant Mouse Resource and Research Centers (MMRRC) at UNC. 828 829 Author Contributions 830 DNK, EAH, JMM, MTF, RML, and CT designed the research; EAH , AT, TL-W, KGL, and KH performed the 831 research; EAH, AT, MTF, JMM, and DNK analyzed the data and interpreted results; EAH and DNK wrote the 832 manuscript; EAH, DNK, AT, TL-W, MTF, RML, and CT edited the final manuscript. 833 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint

834 835 1 Thompson, A. J., Baranzini, S. E., Geurts, J., Hemmer, B. & Ciccarelli, O. Multiple sclerosis. The Lancet 836 391, 1622-1636 (2018). https://doi.org:https://doi.org/10.1016/S0140-6736(18)30481-1 837 2 Frohman, E. M., Racke, M. K. & Raine, C. S. Multiple sclerosis--the plaque and its pathogenesis. N Engl 838 J Med 354, 942-955 (2006). https://doi.org:10.1056/NEJMra052130 839 3 Greenstein, J. I. Current concepts of the cellular and molecular pathophysiology of multiple sclerosis. 840 Developmental Neurobiology 67, 1248-1265 (2007). https://doi.org:https://doi.org/10.1002/dneu.20387 841 4 Efendi, H. Clinically Isolated Syndromes: Clinical Characteristics, Differential Diagnosis, and 842 Management. Noro Psikiyatr Ars 52, S1-s11 (2015). https://doi.org:10.5152/npa.2015.12608 843 5 Antel, J., Antel, S., Caramanos, Z., Arnold, D. L. & Kuhlmann, T. Primary progressive multiple sclerosis: 844 part of the MS disease spectrum or separate disease entity? Acta Neuropathologica 123, 627-638 (2012). 845 https://doi.org:10.1007/s00401-012-0953-0 846 6 Voskuhl, R. R. The effect of sex on multiple sclerosis risk and disease progression. Mult Scler 26, 554-847 560 (2020). https://doi.org:10.1177/1352458519892491 848 7 Voskuhl, R. R. & Gold, S. M. Sex-related factors in multiple sclerosis susceptibility and progression. Nat 849 Rev Neurol 8, 255-263 (2012). https://doi.org:10.1038/nrneurol.2012.43 850 8 Lee, C. Y. & Chan, K. H. Personalized Use of Disease -Modifying Therapies in Multiple Sclerosis. 851 Pharmaceutics 16 (2024). https://doi.org:10.3390/pharmaceutics16010120 852 9 Consortium, I. M. S. G. Multiple sclerosis genomic map implicates peripheral immune cells and microglia 853 in susceptibility. Science 365 (2019). https://doi.org:10.1126/science.aav7188 854 10 Ebers, G. C., Sadovnick, A. D. & Risch, N. J. A genetic basis for familial aggregation in multiple sclerosis. 855 Nature 377, 150-151 (1995). https://doi.org:10.1038/377150a0 856 11 Hollenbach, J. A. & Oksenberg, J. R. The immunogenetics of multiple sclerosis: A comprehensive review. 857 Journal of Autoimmunity 64, 13-25 (2015). https://doi.org:https://doi.org/10.1016/j.jaut.2015.06.010 858 12 Ermann, J. & Glimcher, L. H. After GWAS: mice to the rescue? Curr Opin Immunol 24, 564-570 (2012). 859 https://doi.org:10.1016/j.coi.2012.09.005 860 13 Steinman, L. & Zamvil, S. S. How to successfully apply animal studies in experimental allergic 861 encephalomyelitis to research on multiple sclerosis. Annals of Neurology 60, 12 -21 (2006). 862 https://doi.org:10.1002/ana.20913 863 14 Ben-Nun, A. et al. From classic to spontaneous and humanized models of multiple sclerosis: impact on 864 understanding pathogenesis and drug development. Journal of Autoimmunity 54, 33 -50 (2014). 865 https://doi.org:10.1016/j.jaut.2014.06.004 866 15 Behan, P . O. & Chaudhuri, A. EAE is not a useful model for demyelinating disease. Multiple Sclerosis 867 and Related Disorders 3, 565-574 (2014). https://doi.org:https://doi.org/10.1016/j.msard.2014.06.003 868 16 Sriram, S. & Steiner, I. Experimental allergic encephalomyelitis: a misleading model of multiple sclerosis. 869 Ann Neurol 58, 939-945 (2005). https://doi.org:10.1002/ana.20743 870 17 Sittig, L. J. et al. Genetic Background Limits Generalizability of Genotype -Phenotype Relationships. 871 Neuron 91, 1253-1259 (2016). https://doi.org:10.1016/j.neuron.2016.08.013 872 18 Bramwell, K. K., Teuscher, C. & Weis, J. J. Forward genetic approaches for elucidation of novel regulators 873 of Lyme arthritis severity. Front Cell Infect Microbiol 4, 76 (2014). 874 https://doi.org:10.3389/fcimb.2014.00076 875 19 Yang, H., Bell, T. A., Churchill, G. A. & Pardo -Manuel de Villena, F. On the subspecific origin of the 876 laboratory mouse. Nat Genet 39, 1100-1107 (2007). https://doi.org:10.1038/ng2087 877 20 Poltorak, A., Apalko, S. & Sherbak, S. Wild -derived mice: from genetic diversity to variation in immune 878 responses. Mamm Genome 29, 577-584 (2018). https://doi.org:10.1007/s00335-018-9766-3 879 21 Gregorová, S. & Forejt, J. PWD/Ph and PWK/Ph inbred mouse strains of Mus m. musculus subspecies-880 -a valuable resource of phenotypic variations and genomic polymorphisms. Folia Biol (Praha) 46, 31-41 881 (2000). 882 22 Bearoff, F. et al. Natural genetic variation profoundly regulates gene expression in immune cells and 883 dictates susceptibility to CNS autoimmunity. Genes Immun 17, 386 -395 (2016). 884 https://doi.org:10.1038/gene.2016.37 885 23 Gregorová, S. et al. Mouse consomic strains: exploiting genetic divergence between Mus m. musculus 886 and Mus m. domesticus subspecies. Genome Res 18, 509 -515 (2008). 887 https://doi.org:10.1101/gr.7160508 888 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint 24 Bearoff, F. et al. Identification of Genetic Determinants of the Sexual Dimorphism in CNS Autoimmunity. 889 PLoS ONE 10 (2015). https://doi.org:10.1371/journal.pone.0117993 890 25 Montgomery, T. L. et al. Interactions between host genetics and gut microbiota determine susceptibility 891 to CNS autoimmunity. Proceedings of the National Academy of Sciences of the United States of America 892 117, 27516-27527 (2020). https://doi.org:10.1073/pnas.2002817117 893 26 Threadgill, D. W., Miller, D. R., Churchill, G. A. & de Villena, F. P. -M. The collaborative cross: a 894 recombinant inbred mouse population for the systems genetic era. ILAR journal 52, 24 -31 (2011). 895 https://doi.org:10.1093/ilar.52.1.24 896 27 Saul, M. C., Philip, V. M., Reinholdt, L. G. & Chesler, E. J. High-Diversity Mouse Populations for Complex 897 Traits. Trends in Genetics 35, 501-514 (2019). https://doi.org:10.1016/j.tig.2019.04.003 898 28 Lorè, N. I., Iraqi, F. A. & Bragonzi, A. Host genetic diversity influences the severity of Pseudomonas 899 aeruginosa pneumonia in the Collaborative Cross mice. BMC Genetics 16 (2015). 900 https://doi.org:10.1186/s12863-015-0260-6 901 29 Vered, K., Durrant, C., Mott, R. & Iraqi, F. A. Susceptibility to klebsiella pneumonaie infection in 902 collaborative cross mice is a complex trait controlled by at least three loci acting at different time points. 903 BMC Genomics 15, 865 (2014). https://doi.org:10.1186/1471-2164-15-865 904 30 Graham, J. B. et al. Genetic diversity in the collaborative cross model recapitulates human West Nile 905 virus disease outcomes. mBio 6, e00493-00415 (2015). https://doi.org:10.1128/mBio.00493-15 906 31 Gralinski, L. E. et al. Genome Wide Identification of SARS-CoV Susceptibility Loci Using the Collaborative 907 Cross. PLOS Genetics 11, e1005504 (2015). https://doi.org:10.1371/journal.pgen.1005504 908 32 Green, R. et al. Oas1b-dependent Immune Transcriptional Profiles of West Nile Virus Infection in the 909 Collaborative Cross. G3 (Bethesda) 7, 1665-1682 (2017). https://doi.org:10.1534/g3.117.041624 910 33 Green, R. et al. Identifying protective host gene expression signatures within the spleen during West Nile 911 virus infection in the collaborative cross model. Genom Data 10, 114 -117 (2016). 912 https://doi.org:10.1016/j.gdata.2016.10.006 913 34 Graham, J. B. et al. Baseline T cell immune phenotypes predict virologic and disease control upon SARS-914 CoV infection in Collaborative Cross mice. PLOS Pathogens 17, e1009287 (2021). 915 https://doi.org:10.1371/journal.ppat.1009287 916 35 Graham, J. B. et al. Extensive Homeostatic T Cell Phenotypic Variation within the Collaborative Cross. 917 Cell Rep 21, 2313-2325 (2017). https://doi.org:10.1016/j.celrep.2017.10.093 918 36 Graham, J. B. et al. Unique immune profiles in collaborative cross mice linked to survival and viral 919 clearance upon infection. iScience 27, 109103 (2024). 920 https://doi.org:https://doi.org/10.1016/j.isci.2024.109103 921 37 Orgel, K. et al. Genetic diversity between mouse strains allows identification of the CC027/GeniUnc strain 922 as an orally reactive model of peanut allergy. J Allergy Clin Immunol 143, 1027 -1037.e1027 (2019). 923 https://doi.org:10.1016/j.jaci.2018.10.009 924 38 Risemberg, E. L. et al. A mutation in Themis contributes to anaphylaxis severity following oral peanut 925 challenge in CC027 mice. J Allergy Clin Immunol (2024). https://doi.org:10.1016/j.jaci.2024.03.027 926 39 Donoghue, L. J. et al. Collaborative cross strain CC011/UncJ as a novel mouse model of T2-high, severe 927 asthma. Respir Res 24, 153 (2023). https://doi.org:10.1186/s12931-023-02453-y 928 40 Kelada, S. N. et al. Integrative genetic analysis of allergic inflammation in the murine lung. Am J Respir 929 Cell Mol Biol 51, 436-445 (2014). https://doi.org:10.1165/rcmb.2013-0501OC 930 41 Consortium, I. M. S. G. Locus for severity implicates CNS resilience in progression of multiple sclerosis. 931 Nature 619, 323-331 (2023). https://doi.org:10.1038/s41586-023-06250-x 932 42 Jokubaitis, V. G. et al. Not all roads lead to the immune system: the genetic basis of multiple sclerosis 933 severity. Brain 146, 2316-2331 (2023). https://doi.org:10.1093/brain/awac449 934 43 Abromson-Leeman, S. et al. T-cell properties determine disease site, clinical presentation, and cellular 935 pathology of experimental autoimmune encephalomyelitis. The American Journal of Pathology 165, 936 1519-1533 (2004). https://doi.org:10.1016/S0002-9440(10)63410-4 937 44 Baker, D. et al. Autoimmune encephalomyelitis in NOD mice is not initially a progressive multiple sclerosis 938 model. Annals of Clinical and Translational Neurology 6, 1362 -1372 (2019). 939 https://doi.org:10.1002/acn3.792 940 45 Mendel, I., Kerlero de Rosbo, N. & Ben -Nun, A. A myelin oligodendrocyte glycoprotein peptide induces 941 typical chronic experimental autoimmune encephalomyelitis in H -2b mice: fine specificity and T cell 942 receptor V beta expression of encephalitogenic T cells. Eur J Immunol 25, 1951 -1959 (1995). 943 https://doi.org:10.1002/eji.1830250723 944 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint 46 Croxford, A. L., Kurschus, F. C. & Waisman, A. Mouse models for multiple sclerosis: Historical facts and 945 future implications. Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease 1812, 177-183 946 (2011). https://doi.org:10.1016/j.bbadis.2010.06.010 947 47 Isobe, N. et al. Association of HLA Genetic Risk Burden With Disease Phenotypes in Multiple Sclerosis. 948 JAMA Neurol 73, 795-802 (2016). https://doi.org:10.1001/jamaneurol.2016.0980 949 48 Alvarez-Sanchez, N. & Dunn, S. E. Potential biological contributers to the sex difference in multiple 950 sclerosis progression. Front Immunol 14, 1175874 (2023). https://doi.org:10.3389/fimmu.2023.1175874 951 49 Krementsov, D. N., Case, L. K., Hickey, W. F. & Teuscher, C. Exacerbation of autoimmune 952 neuroinflammation by dietary sodium is genetically controlled and sex specific. The FASEB Journal 29, 953 3446-3457 (2015). https://doi.org:https://doi.org/10.1096/fj.15-272542 954 50 Butterfield, R. J. et al. Identification of genetic loci controlling the characteristics and severity of brain and 955 spinal cord lesions in experimental allergic encephalomyelitis. American Journal of Pathology 157, 637-956 645 (2000). https://doi.org:Doi 10.1016/S0002-9440(10)64574-9 957 51 Muller, D. M., Pender, M. P. & Greer, J. M. A neuropathological analysis of experimental autoimmune 958 encephalomyelitis with predominant brain stem and cerebellar involvement and differences between 959 active and passive induction. Acta Neuropathol 100, 174-182 (2000). 960 52 Broman, K. W. et al. R/qtl2: Software for Mapping Quantitative Trait Loci with High-Dimensional Data and 961 Multiparent Populations. Genetics 211, 495-502 (2019). https://doi.org:10.1534/genetics.118.301595 962 53 Guan, Y ., Ackert-Bicknell, C. L., Kell, B., Troyanskaya, O. G. & Hibbs, M. A. Functional genomics 963 complements quantitative genetics in identifying disease -gene associations. PLoS Comput Biol 6, 964 e1000991 (2010). https://doi.org:10.1371/journal.pcbi.1000991 965 54 Guan, Y . et al. Tissue-specific functional networks for prioritizing phenotype and disease genes. PLoS 966 Comput Biol 8, e1002694 (2012). https://doi.org:10.1371/journal.pcbi.1002694 967 55 Brabec, J. L., Lara, M. K., Tyler, A. L. & Mahoney, J. M. System -Level Analysis of Alzheimer's Disease 968 Prioritizes Candidate Genes for Neurodegeneration. Front Genet 12, 625246 (2021). 969 https://doi.org:10.3389/fgene.2021.625246 970 56 Tyler, A. L. et al. Network-Based Functional Prediction Augments Genetic Association To Predict 971 Candidate Genes for Histamine Hypersensitivity in Mice. G3 Genes|Genomes|Genetics 9, 4223-4233 972 (2019). https://doi.org:10.1534/g3.119.400740 973 57 Lahue, K. G. et al. Identification of novel loci controlling inflammatory bowel disease susceptibility utilizing 974 the genetic diversity of wild -derived mice. Genes & Immunity 21, 311 -325 (2020). 975 https://doi.org:10.1038/s41435-020-00110-8 976 58 Goya, J. et al. FNTM: a server for predicting functional networks of tissues in mouse. Nucleic Acids 977 Research 43, W182-W187 (2015). https://doi.org:10.1093/nar/gkv443 978 59 Sollis, E. et al. The NHGRI-EBI GWAS Catalog: knowledgebase and deposition resource. Nucleic Acids 979 Res 51, D977-d985 (2023). https://doi.org:10.1093/nar/gkac1010 980 60 Cree, B. A. Genetics of primary progressive multiple sclerosis. Handb Clin Neurol 122, 211-230 (2014). 981 https://doi.org:10.1016/B978-0-444-52001-2.00042-X 982 61 Gold, R., Linington, C. & Lassmann, H. Understanding pathogenesis and therapy of multiple sclerosis via 983 animal models: 70 years of merits and culprits in experimental autoimmune encephalomyelitis research. 984 Brain 129, 1953-1971 (2006). https://doi.org:10.1093/brain/awl075 985 62 Constantinescu, C. S., Farooqi, N., O'Brien, K. & Gran, B. Experimental autoimmune encephalomyelitis 986 (EAE) as a model for multiple sclerosis (MS). Br J Pharmacol 164, 1079 -1106 (2011). 987 https://doi.org:10.1111/j.1476-5381.2011.01302.x 988 63 Ransohoff, R. M. Animal models of multiple sclerosis: the good, the bad and the bottom line. Nature 989 Neuroscience 15, 1074-1077 (2012). https://doi.org:10.1038/nn.3168 990 64 Stromnes, I. M., Cerretti, L. M., Liggitt, D., Harris, R. A. & Goverman, J. M. Differential regulation of central 991 nervous system autoimmunity by T(H)1 and T(H)17 cells. Nat Med 14, 337 -342 (2008). 992 https://doi.org:10.1038/nm1715 993 65 Liu, Y. et al. Preferential Recruitment of Neutrophils into the Cerebellum and Brainstem Contributes to 994 the Atypical Experimental Autoimmune Encephalomyelitis Phenotype. J Immunol 195, 841-852 (2015). 995 https://doi.org:10.4049/jimmunol.1403063 996 66 Yan, Z. et al. Deficiency of Socs3 leads to brain -targeted EAE via enhanced neutrophil activation and 997 ROS production. JCI Insight 5 (2019). https://doi.org:10.1172/jci.insight.126520 998 67 Spanier, J. A., Nashold, F. E., Olson, J. K. & Hayes, C. E. The Ifng Gene Is Essential for Vdr Gene 999 Expression and Vitamin D3-Mediated Reduction of the Pathogenic T Cell Burden in the Central Nervous 1000 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint System in Experimental Autoimmune Encephalomyelitis, a Multiple Sclerosis Model. The Journal of 1001 Immunology 189, 3188-3197 (2012). https://doi.org:10.4049/jimmunol.1102925 1002 68 Miller, N. M., Wang, J., Tan, Y . & Dittel, B. N. Anti -inflammatory mechanisms of IFN -γ studied in 1003 experimental autoimmune encephalomyelitis reveal neutrophils as a potential target in multiple sclerosis. 1004 Frontiers in Neuroscience 9 (2015). https://doi.org:10.3389/fnins.2015.00287 1005 69 Kutzelnigg, A. et al. Widespread demyelination in the cerebellar cortex in multiple sclerosis. Brain Pathol 1006 17, 38-44 (2007). https://doi.org:10.1111/j.1750-3639.2006.00041.x 1007 70 Wilkins, A. Cerebellar Dysfunction in Multiple Sclerosis. Front Neurol 8, 312 (2017). 1008 https://doi.org:10.3389/fneur.2017.00312 1009 71 Basso, A. S. et al. Reversal of axonal loss and disability in a mouse model of progressive multiple 1010 sclerosis. J Clin Invest 118, 1532-1543 (2008). https://doi.org:10.1172/JCI33464 1011 72 Al-Izki, S. et al. Practical guide to the induction of relapsing progressive experimental autoimmune 1012 encephalomyelitis in the Biozzi ABH mouse. Mult Scler Relat Disord 1, 29 -38 (2012). 1013 https://doi.org:10.1016/j.msard.2011.09.001 1014 73 Schmidt, H., Williamson, D. & Ashley -Koch, A. HLA -DR15 haplotype and multiple sclerosis: a HuGE 1015 review. Am J Epidemiol 165, 1097-1109 (2007). https://doi.org:10.1093/aje/kwk118 1016 74 Spach, K. M. et al. Cutting edge: the Y chromosome controls the age -dependent experimental allergic 1017 encephalomyelitis sexual dimorphism in SJL/J mice. J Immunol 182, 1789 -1793 (2009). 1018 https://doi.org:10.4049/jimmunol.0803200 1019 75 Baranzini, S. E. et al. Genome-wide association analysis of susceptibility and clinical phenotype in 1020 multiple sclerosis. Hum Mol Genet 18, 767-778 (2009). https://doi.org:10.1093/hmg/ddn388 1021 76 Blankenhorn, E. P . et al. Genetics of experimental allergic encephalomyelitis supports the role of T helper 1022 cells in multiple sclerosis pathogenesis. Annals of Neurology 70, 887 -896 (2011). 1023 https://doi.org:10.1002/ana.22642 1024 77 Oksenberg, J. R. & Hauser, S. L. Decoding multiple sclerosis. Ann Neurol 70, A5 -7 (2011). 1025 https://doi.org:10.1002/ana.22680 1026 78 Jackson, K. C. et al. Genetic model of MS severity predicts future accumulation of disability. Ann Hum 1027 Genet 84, 1-10 (2020). https://doi.org:10.1111/ahg.12342 1028 79 Fransen, N. L. et al. Post-mortem multiple sclerosis lesion pathology is influenced by single nucleotide 1029 polymorphisms. Brain Pathol 30, 106-119 (2020). https://doi.org:10.1111/bpa.12760 1030 80 Al Rihani, S. B. et al. Disease-Induced Modulation of Drug Transporters at the Blood–Brain Barrier Level. 1031 International Journal of Molecular Sciences 22, 3742 (2021). 1032 81 Oerlemans, R. et al. Expression profiling of ABC transporters in peripheral blood lymphocytes and 1033 monocyte-derived macrophages of rheumatoid arthritis patients. JMCM 3, 47 -60 (2020). 1034 https://doi.org:10.31083/j.jmcm.2020.02.007 1035 82 Richard-Miceli, C. & Criswell, L. A. Emerging patterns of genetic overlap across autoimmune disorders. 1036 Genome Medicine 4, 6 (2012). https://doi.org:10.1186/gm305 1037 83 Lincoln, M. R. et al. Genetic mapping across autoimmune diseases reveals shared associations and 1038 mechanisms. Nature Genetics 56, 838-845 (2024). https://doi.org:10.1038/s41588-024-01732-8 1039 84 Kooij, G. et al. Adenosine triphosphate-binding cassette transporters mediate chemokine (C -C motif) 1040 ligand 2 secretion from reactive astrocytes: relevance to multiple sclerosis pathogenesis. Brain 134, 555-1041 570 (2010). https://doi.org:10.1093/brain/awq330 1042 85 Yamauchi, T. & Moroishi, T. Hippo Pathway in Mammalian Adaptive Immune System. Cells 8 (2019). 1043 https://doi.org:10.3390/cells8050398 1044 86 Kong, H., Han, J.-J., Gorbachev, D. & Zhang, X.-A. Role of the Hippo pathway in autoimmune diseases. 1045 Experimental Gerontology 185, 112336 (2024). 1046 https://doi.org:https://doi.org/10.1016/j.exger.2023.112336 1047 87 Mikami, A. et al. Molecular structure of cytoplasmic dynein 2 and its distribution in neuronal and ciliated 1048 cells. Journal of Cell Science 115, 4801-4808 (2002). https://doi.org:10.1242/jcs.00168 1049 88 Hirokawa, N., Niwa, S. & Tanaka, Y. Molecular Motors in Neurons: Transport Mechanisms and Roles in 1050 Brain Function, Development, and Disease. Neuron 68, 610 -638 (2010). 1051 https://doi.org:https://doi.org/10.1016/j.neuron.2010.09.039 1052 89 Dahl, T. M. & Baehr, W. Review: Cytoplasmic dynein motors in photoreceptors. Mol Vis 27, 506-517 1053 (2021). 1054 90 Vig, A. et al. DYNC2H1 hypomorphic or retina -predominant variants cause nonsyndromic retinal 1055 degeneration. Genet Med 22, 2041-2051 (2020). https://doi.org:10.1038/s41436-020-0915-1 1056 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint 91 Kreutzer, M. et al. Axonopathy is associated with complex axonal transport defects in a model of multiple 1057 sclerosis. Brain Pathol 22, 454-471 (2012). https://doi.org:10.1111/j.1750-3639.2011.00541.x 1058 92 Sarma, J. D. et al. Functional interleukin-17 receptor A is expressed in central nervous system glia and 1059 upregulated in experimental autoimmune encephalomyelitis. Journal of Neuroinflammation 6, 14 (2009). 1060 https://doi.org:10.1186/1742-2094-6-14 1061 93 Khan, A. W., Farooq, M., Hwang, M. J., Haseeb, M. & Choi, S. Autoimmune Neuroinflammatory Diseases: 1062 Role of Interleukins. Int J Mol Sci 24 (2023). https://doi.org:10.3390/ijms24097960 1063 94 Milovanovic, J. et al. Interleukin-17 in Chronic Inflammatory Neurological Diseases. Frontiers in 1064 Immunology 11 (2020). https://doi.org:10.3389/fimmu.2020.00947 1065 95 Brucklacher-Waldert, V., Stuerner, K., Kolster, M., Wolthausen, J. & Tolosa, E. Phenotypical and 1066 functional characterization of T helper 17 cells in multiple sclerosis. Brain 132, 3329 -3341 (2009). 1067 https://doi.org:10.1093/brain/awp289 1068 96 Bălaşa, R., Bajko, Z. & Huţanu, A. Serum levels of IL -17A in patients with relapsing –remitting multiple 1069 sclerosis treated with interferon -β. Multiple Sclerosis Journal 19, 885 -890 (2013). 1070 https://doi.org:10.1177/1352458512468497 1071 97 Frisullo, G. et al. IL17 and IFNgamma production by peripheral blood mononuclear cells from clinically 1072 isolated syndrome to secondary progressive multiple sclerosis. Cytokine 44, 22 -25 (2008). 1073 https://doi.org:10.1016/j.cyto.2008.08.007 1074 98 Graber, J. J. et al. Interleukin-17 in transverse myelitis and multiple sclerosis. J Neuroimmunol 196, 124-1075 132 (2008). https://doi.org:10.1016/j.jneuroim.2008.02.008 1076 99 Krueger, E. M., Miranpuri, G. S. & Resnick, D. K. Emerging role of WNK1 in pathologic central nervous 1077 system signaling. Ann Neurosci 18, 70-75 (2011). https://doi.org:10.5214/ans.0972.7531.1118212 1078 100 Izadifar, A. et al. Axon morphogenesis and maintenance require an evolutionary conserved safeguard 1079 function of Wnk kinases antagonizing Sarm and Axed. Neuron 109, 2864 -2883.e2868 (2021). 1080 https://doi.org:https://doi.org/10.1016/j.neuron.2021.07.006 1081 101 Kotelnikova, E. et al. MAPK pathway and B cells overactivation in multiple sclerosis revealed by 1082 phosphoproteomics and genomic analysis. Proc Natl Acad Sci U S A 116, 9671 -9676 (2019). 1083 https://doi.org:10.1073/pnas.1818347116 1084 102 Xiu, M., Li, L., Li, Y. & Gao, Y. An update regarding the role of WNK kinases in cancer. Cell Death Dis 13, 1085 795 (2022). https://doi.org:10.1038/s41419-022-05249-y 1086 103 Serysheva, E. et al. Wnk kinases are positive regulators of canonical Wnt/beta-catenin signalling. EMBO 1087 Rep 14, 718-725 (2013). https://doi.org:10.1038/embor.2013.88 1088 104 Vandebergh, M. et al. Genetic Variation in WNT9B Increases Relapse Hazard in Multiple Sclerosis. Ann 1089 Neurol 89, 884-894 (2021). https://doi.org:10.1002/ana.26061 1090 105 Olsson, T., Jagodic, M., Piehl, F. & Wallstrom, E. Genetics of autoimmune neuroinflammation. Curr Opin 1091 Immunol 18, 643-649 (2006). https://doi.org:10.1016/j.coi.2006.08.001 1092 106 Ma, R. Z. et al. Identification of Bphs, an autoimmune disease locus, as histamine receptor H1. Science 1093 297, 620-623 (2002). https://doi.org:10.1126/science.1072810 1094 107 Di Francesco, A. et al. Regulators of health and lifespan extension in genetically diverse mice on dietary 1095 restriction. bioRxiv, 2023.2011.2028.568901 (2023). https://doi.org:10.1101/2023.11.28.568901 1096 108 Mayeux, J. M. et al. Silicosis and Silica -Induced Autoimmunity in the Diversity Outbred Mouse. Front 1097 Immunol 9, 874 (2018). https://doi.org:10.3389/fimmu.2018.00874 1098 109 Eldridge, R. et al. Antecedent presentation of neurological phenotypes in the Collaborative Cross reveals 1099 four classes with complex sex -dependencies. Scientific Reports 10, 7918 (2020). 1100 https://doi.org:10.1038/s41598-020-64862-z 1101 110 Perez Gomez, A. A. et al. Genetic and immunological contributors to virus -induced paralysis. Brain, 1102 Behavior, & Immunity - Health 18, 100395 (2021). 1103 https://doi.org:https://doi.org/10.1016/j.bbih.2021.100395 1104 111 Brinkmeyer-Langford, C. L. et al. Host genetic background influences diverse neurological responses to 1105 viral infection in mice. Scientific Reports 7, 12194 (2017). https://doi.org:10.1038/s41598-017-12477-2 1106 112 Lawley, K. S. et al. Host genetic diversity drives variable central nervous system lesion distribution in 1107 chronic phase of Theiler's Murine Encephalomyelitis Virus (TMEV) infection. PLoS One 16, e0256370 1108 (2021). https://doi.org:10.1371/journal.pone.0256370 1109 113 Lawley, K. S. et al. Viral Clearance and Neuroinflammation in Acute TMEV Infection Vary by Host Genetic 1110 Background. International Journal of Molecular Sciences 23, 10482 (2022). 1111 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint 114 Welsh, C. E. et al. Status and access to the Collaborative Cross population. Mamm Genome 23, 706-1112 712 (2012). https://doi.org:10.1007/s00335-012-9410-6 1113 115 Iraqi, F. A., Churchill, G. & Mott, R. The Collaborative Cross, developing a resource for mammalian 1114 systems genetics: a status report of the Wellcome Trust cohort. Mamm Genome 19, 379-381 (2008). 1115 https://doi.org:10.1007/s00335-008-9113-1 1116 116 Morahan, G., Balmer, L. & Monley, D. Establishment of “The Gene Mine”: a resource for rapid 1117 identification of complex trait genes. Mammalian Genome 19, 390 -393 (2008). 1118 https://doi.org:10.1007/s00335-008-9134-9 1119 117 Chesler, E. J. et al. The Collaborative Cross at Oak Ridge National Laboratory: developing a powerful 1120 resource for systems genetics. Mamm Genome 19, 382-389 (2008). https://doi.org:10.1007/s00335-008-1121 9135-8 1122 118 McGill, M. M. et al. p38 MAP Kinase Signaling in Microglia Plays a Sex-Specific Protective Role in CNS 1123 Autoimmunity and Regulates Microglial Transcriptional States. Front Immunol 12, 715311 (2021). 1124 https://doi.org:10.3389/fimmu.2021.715311 1125 119 Montgomery, T. L. et al. Lactobacillus reuteri tryptophan metabolism promotes host susceptibility to CNS 1126 autoimmunity. Microbiome 10, 198 (2022). https://doi.org:10.1186/s40168-022-01408-7 1127 1128 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint B6 CC011CC038CC072CC040CC020CC044CC018CC023CC042CC036CC075CC001CC031CC068CC010CC051CC074CC003CC006CC037CC046CC059CC028CC032CC002CC004CC030CC043CC041CC061CC084CC083 0 20 40 60 80 100 EAE Incidence by Subtype Incidence (%) Classic EAE Axial Rotary EAE 91.7% Relapsing Remitting EAE B6 CC040CC030CC041CC011CC020CC044CC038CC072CC042CC028CC032CC031CC068CC004CC023CC075CC083CC036CC001CC018CC061CC074CC003CC010CC043CC037CC084CC046CC006CC059CC051CC002 0 20 40 60 80 100 EAE Incidence by Disease Course Incidence (%) Chronic EAEMonophasic EAE B6 CC011CC020CC044CC068CC040CC036CC072CC006CC023CC074CC038CC031CC032CC051CC003CC018CC059CC061CC002CC010CC046CC075CC001CC042CC083CC037CC084CC030CC041CC043CC004CC028 0 50 100 150 200 EAE Disease Severity Cumulative Disease Score************ *** ** ** ** ** ** * * * * **** *** H2b H2g7 0 50 100 150 200 EAE Severity CDS H2b H2g7 0 25 50 75 100 EAE Incidence (%) H2b H2g7 0 25 50 75 100 Classic EAE Incidence (%) H2b H2g7 0 25 50 75 100 AR-EAE Incidence (%) H2b H2g7 0 25 50 75 100 Chronic EAE Incidence (%) H2b H2g7 0 25 50 75 100 Monophasic EAE Incidence (%) H2b H2g7 0 25 50 75 100 RR-EAE Incidence (%) A B C D E F G H I J K Figure 1 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Figure 1. MOG35-55 induced EAE in CC strains results in heterogeneous disease profiles. EAE was induced in 8-13 week old male and female B6 and H2b or H2g7 CC mice (32 strains, ~5 males and ~5 females per strain) by s.c. immunization with 200 µg of MOG35-55 emulsified in CFA, and a single i.p. injection of 200 ng of PTX on D0 (see Materials and Methods). Mice were observed daily for a total of 50 days starting at 5 days post induction for the presence of clinical disease symptoms which were quantified to assess the overall EAE disease profile, as described in Materials and Methods. ( A) Workflow illustrating the study design and selection of CC strains compatible with MOG35-55 EAE induction for inclusion based on founder MHC haplotype (H2b or H2g7). (B) Percent EAE incidence per strain, in CC strains, with B6 mice shown as a reference control. Color of bars illustrates percent incidence of major EAE subtype (classic : grey, or axial rotary (AR): orange). (C) Percent incidence of relapsing remitting (RR; green bars) and monophasic (light green bars) EAE in CC strains, with B6 mice shown as a reference control. ( D) Comparison of EAE disease severity in CC strains, as calculated using cumulative disease score (see Materials and Methods), versus B6 reference controls. Significance of differences of each CC strain from B6 reference controls was determined via one-way ANOVA with Dunnett’s multiple comparisons test and indicated by asterisks where significant ( * p <0.05, ** p<0.01, *** p<0.001, **** p< 0.0001), with corresponding bar colors indicating directionality as compared to B 6, ie. blue = less severe and red = more severe. (E-K) Distribution of strain (E) cumulative disease score (CDS), (F) total EAE incidence, and incidence of (G) classic EAE, (H) AR-EAE, (I) chronic EAE, (J ) RR-EAE, and (K) monophasic EAE, grouped by H2b and H2g7 homozygous haplotypes. Each data point in panels ( E - K) represents the average for a single strain. Significance of differences between haplotype groups was determined by unpaired T-test and indicated by asterisks where significant. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint 0 10 20 30 40 50 0 1 2 3 4 5 Resistant and Severe-Progressive Days Post Induction Classic Disease Score B6 CC011 CC028 0 10 20 30 40 50 0 1 2 3 4 5 Axial Rotary Days Post Induction AR Disease Score B6 CC004 0 10 20 30 40 50 0 1 2 3 4 5 Relapsing Remitting Days Post Induction Classic Disease Score CC002 0 10 20 30 40 50 0 1 2 3 4 5 Monophasic Days Post Induction Classic Disease Score CC068 Females CC068 Males 0 10 20 30 40 50 0 1 2 3 4 5 Secondary Progressive Days Post Induction Classic Disease Score CC043 A B C D E Figure 2 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Figure 2. EAE in CC strains captures clinically relevant disease courses. EAE was induced and observed for 50 days in CC and B6 reference control mice as described in Figure 1. Post observation, disease course profiles for each strain were derived from daily disease scores (see Materials and Methods) , revealing several strains that captured phenotypes of interest. These include d: (A) severe progressive disease in CC028 (5M + 5F) mice (shown in red) and EAE resistance in CC011 (5M + 5F) mice (shown in blue), compared to B6 (18M + 18F) reference controls (shown in grey) (sexes pooled), (B) AR-EAE in CC004 (7M + 7F) mice (shown in orange) (sexes pooled), (C ) relapsing remitting ( RR)-EAE in CC002 (6M + 5F) mice (sexes pooled), (D) secondary progressive EAE in CC043 (5M + 5F) mice (sexes pooled), and (E) monophasic EAE in CC068 (4M + 5F) mice (sexes shown separately due to timing of disease onset). All panels show classic EAE scores, except (B), which shows AR-EAE scores, as indicated on the Y axis in each panel. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint 0 10 20 30 40 50 0 1 2 3 4 5 CC046 Days Post Induction Classic Disease Score Females Males CC001CC002CC003CC004CC006CC010CC011CC018CC020CC023CC028CC030CC031CC032CC036CC037CC038CC040CC041CC042CC043CC044CC046CC051CC059CC061CC068CC072CC074CC075CC083CC084 0 50 100 150 200 Axial Rotary EAE Severity by Sex AR CDS FemaleMale ✱✱ ✱ ✱ Strain:✱✱✱✱ Sex: ns Interaction: ns 0 10 20 30 40 50 0 1 2 3 4 5 CC042 Days Post Induction Classic Disease Score Females Males 0 10 20 30 40 50 0 1 2 3 4 5 CC072 Days Post Induction AR Disease Score Females Males CC001CC002CC003CC004CC006CC010CC011CC018CC020CC023CC028CC030CC031CC032CC036CC037CC038CC040CC041CC042CC043CC044CC046CC051CC059CC061CC068CC072CC074CC075CC083CC084 0 50 100 150 200 EAE Severity by Sex Cumulative Disease Score FemaleMale ✱ ✱✱ ✱✱✱ Strain:✱✱✱✱ Sex: ns Interaction: ns CC001CC002CC003CC004CC006CC010CC011CC018CC020CC023CC028CC030CC031CC032CC036CC037CC038CC040CC041CC042CC043CC044CC046CC051CC059CC061CC068CC072CC074CC075CC083CC084 0 50 100 150 200 Classic EAE Severity by Sex Classic CDS FemaleMale ✱✱ ✱✱✱ ✱✱✱ Strain:✱✱✱✱ Sex: ns Interaction: ✱ 0 10 20 30 40 50 0 1 2 3 4 5 C0038 Days Post Induction AR Disease Score Females Males M > F M < F A B C D E F G Figure 3 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Figure 3. EAE in CC strains demonstrates bi-directional effects of sex on disease course. EAE was induced and observed for 50 days in CC mice as described in Figure 1. Disease severity was assessed for effects of sex within strain using (A) cumulative disease score (CDS; the total sum of disease scores), (B) classic cumulative disease score (classic-CDS; the total sum of classic disease scores), and (C) AR cumulative disease score (AR-CDS; the total sum of AR disease scores) (see Materials and Methods). Significance of differences was determined by two-way ANOVA and displayed with Fishers LSD multiple comparisons (see Table S6) . Comparisons are indicated by asterisks where significant (* p <0.05, ** p<0.01, *** p<0.001, **** p< 0.0001). Disease course profiles of sex differences in classic EAE presentation in ( D) CC046, and (E) CC042 mice, and AR-EAE presentation in (F) CC038, and (G) CC072 mice. Panels (D) and (E) display classic EAE scores while panels (F) and (G) display AR-EAE scores, as indicated on the Y axis in each panel. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint B6 CC002CC004CC028 0 1 2 3 4 Brain Demyelination Score ✱✱ B6 CC002CC004CC028 0 1 2 3 4 Spinal Cord Demyelination Score B6 CC002CC004CC028 0 1 2 3 4 Brain Inflammation Score ✱ B6 CC002CC004CC028 0 1 2 3 4 Spinal Cord Inflammation Score ✱ ✱ B6 CC002 CC004 CC028 A BLFB C D E F G H H&E CC002 CC004 CC028 CC002 CC004 CC028 B6 CC002 CC004 CC028 H&E B6 LFB B6 Figure 4 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Figure 4. Severe progressive EAE in CC028 mice and AR-EAE in CC004 mice is associated with distinct pathology in the spinal cord and brain, respectively. EAE was induced and observed for a total of 50 days as described in Figure 1. At day 50, or at time of humane endpoint euthanasia, spinal cord and brains were collected and processed for sectioning and staining with H&E +/- LFB. Histopathologic evaluation of samples from B6 reference control (9M + 9F), CC002 (6M + 5M), CC004 (7M + 7F) and CC028 (5M + 5F) mice, sexes pooled, was performed as described in Materials and Methods . (A) Spinal cord inflammation scores, displayed as strain averages, and (B) corresponding representative spinal cord images for B6, CC002, CC004 and CC028 mice. Images were captured with a 10x objective, scale bar represents 100 μm, and arrows mark focal inflammatory infiltrates. (C ) Spinal cord demyelination scores displayed as strain averages, and ( D) corresponding representative spinal cord images for B6, CC002, CC004 and CC028 mice. Images captured at 10x objective, scale bar represents 100 μm, and arrows denote areas of demyelination. ( E) Brain inflammation scores displayed as strain averages, and (F) corresponding representative brain images for B6, CC002, CC004 and CC028 mice. Images captured at 5x objective, scale bar represents 200 μm, and arrows mark regions of inflammatory infiltrates . (G) Brain demyelination scores displayed as strain averages, and ( H) corresponding representative brain images for B6, CC002, CC004 and CC028 mice. Images captured at 5x objective, scale bar represents 200 μm, and arrows mark areas of demyelination. Significance of difference determined by or dinary one-way ANOVA, with Fishers LSD multiple comparisons to B6 reference controls, or by Brown-Forsythe and Welch ANOVA, with unpaired T test with Welch’s correction for multiple comparison testing when appropriate. Comparisons are indicated by asterisks where significant (* p<0.05, ** p<0.01, *** p<0.001, **** p< 0.0001). was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint B6 CC004CC028 0 20 40 60 80 100 CD11b+ Cells Frequency of CD45+ ✱✱ ✱✱ B6 CC004CC028 0 20 40 60 80 100 Microglia Frequency of CD45+ B6 CC004CC028 0 20 40 60 80 100 Myeloid Cells Frequency of CD45+ B6 CC004CC028 0 20 40 60 80 100 Neutrophils Frequency of CD45+ B6 CC004CC028 0 5 10 15 20 B Cells Frequency of CD45+ ✱ ✱✱ B6 CC004CC028 0 10 20 30 40 TCRβ+ Cells Frequency of CD45+ ✱ ✱ B6 CC004CC028 0 20 40 60 80 100 CD11b+ Cells Frequency of CD45+ ✱ ✱✱✱ B6 CC004CC028 0 20 40 60 80 100 Microglia Frequency of CD45+ ✱ B6 CC004CC028 0 20 40 60 80 100 Myeloid Cells Frequency of CD45+ ✱✱ B6 CC004CC028 0 20 40 60 80 100 Neutrophils Frequency of CD45+ ✱✱ B6 CC004CC028 0 5 10 15 20 B Cells Frequency of CD45+ ✱✱ ✱✱✱ B6 CC004CC028 0 10 20 30 40 TCRβ+ Cells Frequency of CD45+ ✱✱ B6 CC004CC028 0 20 40 60 80 CD4+ IFNγ+ Frequency of CD4+ ✱ ✱ B6 CC004CC028 0 5 10 15 20 CD4+ IL-17+ Frequency of CD4+ B6 CC004CC028 0 20 40 60 80 CD4+ IFNγ+ Frequency of CD4+ ✱✱✱✱ ✱✱✱✱ B6 CC004CC028 0 5 10 15 20 CD4+ IL-17+ Frequency of CD4+ 0 5 10 15 0 1 2 3 4 5 Classic EAE Days Post Induction Classic Disease Score B6 CC004 CC028 0 5 10 15 0 1 2 3 4 5 Axial Rotary EAE Days Post Induction AR Disease Score B6 CC004 CC028 A B C Spinal Cord Brain Spinal Cord Brain D E F G H I J K L M N O P Q R S Figure 5 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Figure 5. Severe EAE in CC028 and CC004 mice is associated with unique CNS immune profiles. EAE was induced in 8-13 week old male B6, CC004, and CC028 mice (5 per strain) by s.c. immunization with 200 µg of MOG35-55 emulsified in CFA, and a single i.p. injection of 200 ng of PTX on D0 (see Materials and Methods). Mice were observed daily for a total of 14 days to capture peak disease activity. On day 14 spinal cord and brains were collected and processed for flow cytometric staining (see Materials and Methods). Disease course profiles for B6, CC004, and CC028 mice displayed as (A) classic EAE or (B) AR-EAE. (C) Representative gating scheme for flow cytometric analysis. Scatterplots demonstrating key immune cell subsets in the spinal cord of B6, CC004 and CC028 mice, including (D) CD11b+ cells (CD45+CD11b+), (E) microglial cells (CD45intCD11b+CX3CR1+), (F) myeloid cells (CD45+CD11b+Cx3CR1low/-), (G ) neutrophils (CD45+CD11b+CX3CR1-Ly6G+), (H) B cells (CD45+CD11b-CD19+), and (I) T cells (CD45+CD11b-CD19-TCRβ+). Scatterplots demonstrating key immune cell subsets in the brain of B6, CC004 and CC028 mice, including ( J) CD11b+ cells, (K) microglial cells, (L) myeloid cells, (M) neutrophils, (N) B cells, and (O) T cells. Scatterplots demonstrating CD4+ T cells (CD45+CD11b-CD19- TCRβ+CD4+) producing (P) IFNγ and (Q) IL-17 in the spinal cord of B6, CC004 and CC028 mice. Scatterplots demonstrating CD4 + T cells producing ( R) IFN γ and ( S) IL -17 in the brain of B6, CC004 and CC028 mice. Significance of differences of each CC strain from B6 reference controls was determined via one-way ANOVA with Dunnett’s multiple comparisons test and indicated by asterisks where significant (* p <0.05, ** p<0.01, *** p<0.001, **** p< 0.0001). was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint B6 →B6 B6 →02 02→B6 02→02 0 2 4 6 8 10 Spinal Cord CD8+ Cells Frequency of CD45+ ✱ ✱ ✱ B6 →B6 B6 →02 02→B6 02→02 0 5 10 15 20 Spinal Cord CD4+ Cells Frequency of CD45+ B6 →B6 B6 →02 02→B6 0 25 50 75 100 Spleen CD19+ Cells Percent B6 →B6 B6 →02 02→B6 0 25 50 75 100 Spleen CD11b+ Cells Percent B6 →B6 B6 →02 02→B6 0 25 50 75 100 Spleen CD8+ Cells Percent B6 →B6 B6 →02 02→B6 0 25 50 75 100 Spleen CD4+ Cells Percent B6 →B6 B6 →02 02→B6 02→02 0 20 40 60 80 100 Spinal Cord CD11b+ Cells Frequency of CD45+ ✱ B6 →B6 B6 →02 02→B6 0 25 50 75 100 Spinal Cord CD8+ Cells Percent A C D F E G H I J K B L Donor Host Donor Host B6 →B6 B6 →02 02→B6 0 25 50 75 100 Spinal Cord CD4+ Cells Percent B6 →B6 B6 →02 02→B6 02→02 0 5 10 15 20 Spinal Cord TCRβ+ Cells Frequency of CD45+ ✱ 0 5 10 15 20 25 30 35 0 1 2 3 4 5 Days Post Induction Classic Disease Score B6→B6 02→02 B6→02 02→B6 Figure 6 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Figure 6. Peripheral immune and CNS intrinsic factors drive relapsing remitting EAE in CC002 mice. B6 and CC002 mice were subjected to bone marrow ablation and reconstitution to create reciprocal bone marrow chimeric mice (see Materials and Methods), designated as B6→B6 (7M + 4F), B6→02 (7M + 1F), 02→B6 (6M +4F), 02→02 (6M + 1F) and illustrated in the schematic in panel ( A). Mice were rested for a total of 8 prior to EAE induction by s.c. immunization with 200 µg of MOG35-55 emulsified in CFA, and a single i.p. injection of 200 ng of PTX on D0 (see Materials and Methods). Mice were observed daily for a total of 34 days. On day 34 spleen, spinal cord and brain tissues were collected and processed for flow cytometric staining (see Materials and Methods). Percent chimerism was assessed in B6→B6, B6→02, and 02→B6 mice for splenic (B) CD11b + cells (CD45+CD11b+CD19-), (C) CD19+ cells (CD45+CD11b-CD19+), (D) CD4+(CD45+CD11b-CD19-TCRβ+CD4+) and (E) CD8+ T cells (CD45+CD11b-CD19-TCRβ+CD8+), as well as infiltrating ( F) CD4+ and (G) CD8+ T cells in the spinal cord. (H) Disease course profiles for B6→B6, B6→02, 02→B6, and 02→02 mice displayed as classic EAE. Comparison of spinal cord infiltrating immune cell populations in B6→B6, B6→02, 02→B6, and 02→02 mice for (I) CD11b + cells, (J) CD19+ cells, (K) CD8+, and (L) CD4+ T cells. Significance of differences between groups was determined via one-way ANOVA with Tukey's multiple comparisons test and indicated by brackets and asterisks where significant (* p<0.05, ** p<0.01, *** p<0.001, **** p< 0.0001). was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint A B C D Cumulative Disease Score E AR EAE Incidence F 8 6 4 2 0 LOD Score 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 X Chromosome 0.150.2 5 4 3 2 0 LOD Score 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 X Chromosome 0.150.2 1 2 1 0 -2 -1 8 6 2 4 0 QTL EffectsLOD Score 0 5 10 15 20 25 30 Chr9 Position (Mb) Eaecc3 A/J B6 129S1 NOD NZO CAST PWK WSB Eaecc6 2 1 0 -2 -1 4 3 1 2 0 QTL EffectsLOD Score 95 100 105 110 115 120 125 Chr14 Position (Mb) A/J B6 129S1 NOD NZO CAST PWK WSB Eaecc6 Founder Genotype Distribution Normalized CDS (A) A/J (B) B6 (C) 129S1 (D) NOD (E) NZO (F) CAST (G) PWK (H) WSB Founder Genotype Eaecc3 Genotype x Phenotype CC Strains Low High AR Incidence Founder Effect HighLow Founder Genotype (A) A/J (B) B6 (C) 129S1 (D) NOD (E) NZO (F) CAST (G) PWK (H) WSB CC Founder Allele Probability 1 0 0.5 Figure 7 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Figure 7. QTL analysis reveals distinct genetic linkage patterns for AR-EAE incidence and EAE severity. EAE was induced and evaluated in 32 CC strains , as described in Figure 1. EAE disease phenotypes and quantitative trait variables were calculated, and quantitative trait loci (QTL) mapping was performed (see

and Methods). (A ) Manhattan plot demonstrating logarithm of odds (LOD) traces for QTL mapping assessing genome association with AR-EAE incidence. Genome wide significance thresholds of 15% (solid line) and 20% (dashed line) were determined by permutations (n=1000). (B) Corresponding CC founder allele effects plot for lead QTL identified on chromosome 9 - Eaecc3. (C) Heatmap demonstrating CC strain distribution based on genotype by phenotype analysis for the chromosome 9 AR-EAE incidence QTL – Eaecc3. (D) Manhattan plot demonstrating LOD traces for QTL mapping assessing genome association with EAE severity as determined by cumulative disease score. Permutation was used to determine Genome wide significance thresholds of 15% (solid line) and 20% (dashed line) were determined by permutations (n=1000) . (E) Corresponding CC founder allele effects plot for lead QTL identified on chromosome 14 – Eaecc6. (F) Box and whisker plot demonstrating distribution of CC founder alleles within studied CC strains. Panels (B ), (C), (E), and (F ) use the conventional letter and color designations for CC founder strains as follows: A/yellow: A/J, B/grey: C57BL/6J, C/pink: 129S1/SvlmJ, D/dark blue: NOD/ShiLtJ, E/light blue: NZO/HILtJ, F/green: CAST/EiJ, G/red: PWK/PhJ, and H/purple: WSB/EiJ. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint

Analysis Data MS GWAS Genes Trained Immune SVM Models Positional Candidate Genes Immune Network CNS Network Train SVM Trained CNS SVM Models EAE QTL mapping Ranked Genes for Immune Network Ranked Genes for CNS Network Score Positional Candidates Position on Chr9 (Mb) AImmune SystemCNS D E F Eaecc3 Eaecc6 Eaecc5 G Position on Chr9 (Mb) C B Figure 8 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Figure 8. Machine learning-based functional candidate gene prioritization nominates distinct genes associated with QTL for AR-EAE incidence, EAE severity, and monophasic EAE incidence. Support vector machine (SVM) classifiers were trained using MS GWAS genes and integrated with extensive tissue specific connectivity networks for the CNS and immune system to rank gene candidates associated EAE QTL in the context of either the CNS or immune system (see Materials and Methods), as illustrated by the schematic in panel (A). Ranked candidate genes for the AR -EAE incidence QTL on chromosome 9 – Eaecc3 for the (B) immune system network and (C) CNS network, genes are plotted by genomic position on the x axis and - log(FPR) on the y axis with the dotted lines demonstrating QTL boundaries. Ranked candidate genes graphed independent of genomic position for the cumulative disease score QTL on chromosome 14 – Eaecc6 for the (D) immune system network and (E) CNS network, and the monophasic EAE incidence QTL on chromosome 6 – Eaecc5 for the (F) immune system network and (G) CNS network. The solid line in panels B – G corresponds to the FPR threshold of 0.05. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Table 1. Characteristics of CC strains used in EAE studies and their phenotypes Haplotype Founder Strain CC001/Unc CC001 b B6 9 (4M, 5F) C1 Chronic (Classic) None CC002/Unc CC002 b 129S1 11 (6M, 5F) C1,C2, and C3 Relapsing Remitting (Classic) None CC003/Unc CC003 b/g7 129S1/NOD 10 (5M, 5F) C1 and C2 Monophasic (Classic) None CC004/TauUnc CC004 b 129S1 14 (7M, 7F) C1 and C2 Severe Chronic (AR) None CC006/TauUnc CC006 b/g7 B6/NOD 10 (5M, 5F) C1 and C2 Mild Relapsing Remitting (Classic) None CC010/GeniUnc CC010 b B6 10 (5M, 5F) C1 and C2 Chronic/Relapsing Remitting (Classic) None CC011/Unc CC011 g7 NOD 10 (5M, 5F) C1 and C3 Resistant (Classic) None CC018/Unc CC019 b 129S1 9 (5M, 4F) C1 and C3 Chronic (Classic) None CC020/GeniUncJ CC020 b B6 10 (5M, 5F) C4 Mild Monophasic (Classic) None CC023/GeniUnc CC023 b 129S1 10 (5M, 5F) C1 and C2 Mild Chronic (Classic) None CC028/GeniUnc CC028 b 129S1 10 (5M, 5F) C1 and C3 Severe Progressive (Classic) M > F CC030/GeniUnc CC030 g7/PWK NOD/PWK 10 (5M, 5F) C1 and C2 Chronic (Classic) None CC031/GeniUnc CC031 b B6 9 (4M, 5F) C3 Mild Monophasic (Classic) None CC032/GeniUnc CC032 b B6 10 (5M, 5F) C1 and C2 Mild Monophasic (Classic) None CC036/Unc CC036 g7 NOD 10 (4M,6F) C1 and C2 Mild Monophasic (Classic) None CC037/TauUnc CC037 b B6 10 (5M, 5F) C1 Chronic/Relapsing Remitting (Classic) None CC038/GeniUnc CC038 b 129S1 10 (5M, 5F) C3 Mild Chronic (AR) M > F * CC040/TauUnc CC040 b B6 9 (4M, 5F) C1 Mild Chronic (Classic) None CC041/TauUnc CC041 b 129S1 10 (5M, 5F) C1 Chronic (Classic) None CC042/GeniUnc CC042 b 129S1 10 (5M, 5F) C1 and C2 Chronic (Classic) M < F * CC043/GeniUnc CC043 g7 NOD 10 (5M, 5F) C1 and C2 Secondary Progressive (Classic) None CC044/Unc CC044 b/g7 B6/NOD 10 (5M, 5F) C1 and C2 Mild Monophasic (Classic) None CC046/Unc CC046 g7 NOD 10 (5M, 5F) C3 Chronic (Classic) M > F CC051/TauUnc CC051 b 129S1 10 (5M, 5F) C1 Relapsing Remitting (Classic) None CC059/TauUnc CC059 b/a 129S1/A/J 10 (5M, 5F) C1 Relapsing Remitting (Classic) None CC061/GeniUnc CC061 b 129S1 11 (6M, 5F) C1,C2, and C3 Monophasic (AR & Classic) None CC068/GeniUnc CC068 b 129S1 9 (4M, 5F) C1 and C2 Mild Monophasic (Classic) None CC072/GeniUnc CC072 b B6 10 (5M, 5F) C1,C2, and C3 Mild Chronic (AR) M < F * CC074/Unc CC074 g7 NOD 10 (5M, 5F) C1 and C2 Mild Monophasic (Classic) None CC075/Unc CC075 b 129S1 10 (5M, 5F) C1 and C2 Chronic (Classic) None CC083/Unc CC083 g7 NOD 10 (5M, 5F) C1 Chronic (AR & Classic) None CC084/TauUnc CC084 b/z B6/NZO 10 (5M, 5F) C3 Chronic (AR & Classic) M < F * H2 *Denotes statistical significance (p<0.05) utilizing uncorrected Fisher's LSD ††Sex difference reported as most significant P-value in any of the following: cumulative disease score, classic cumulative disease score, and axial rotary cumulative disease score † EAE phenotype denotes severity (Mild, average- no title, and severe ) as determined by cumulative disease score analysis compa red to B6 reference controls (Figure 1A), disease course (chronic, relapsing remitting, or monophasic), as determined by greatest percent incidence except in the case of equal incidences which are shown with a /, and subtype (classic or AR), where AR EAE is classified by >25% AR-EAE incidence CC Strain Abbreviated Strain Name n Cohort (C#) EAE Phenotype† Course (Subtype) Sex Difference†† was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Table 2. Major EAE phenotype QTL detected in the CC Lead Candidate Genes (Immune) Lead Candidate Genes (CNS) Total EAE Incidence Eaecc1 4: 4.33 (4.04-11.82) 7.31 6.79 Trp53inp1, Plekhf2, Tox, Gem, Chd7 Trp53inp1, Tox, Sdcbp, Plekhf2 Classic EAE Incidence Eaecc2 5: 74.06 (51.50-75.92) 7.50 6.64 Rbpj, Klf3, Stim2, Fam114a1, Pcdh7 Klf3, Rasl11b, Apbb2, Atp8a1, Stim2 AR-EAE Incidence Eaecc3 9: 4.75 (3.37-12.50) 7.91 7.96 Yap1, Kbtbd3 Dync2h1, Birc3 RR-EAE Incidence Eaecc4 18: 57.43 (56.12-60.12) 7.20 6.40 - Fbn2, Isoc1 Monophasic EAE Incidence Eaecc5 6: 120.17 (118.97-121.49) 7.65 7.31 Il17ra, Wnk1 Wnk1, Il17ra, Cecr2 Cumulative Disease Score Eaecc6 14: 113.22 (103.97-118.63) 4.65 4.65 Abbc4 Abbc4, Gpc6 Top 5 Prioritized Genes FPR≥0.05 QTV QTL Symbol Chr: Position (95% Confidence Interval) (Mb) 20% Genome Wide Significance LOD Score LOD Score was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint Table 3. Coding non-synonymous variants in top prioritized QTL candidate genes QTV QTL Symbol QTL QTL Founder Effects Genes ≤ 0.05 FPR SNPs Differentiating Founder Effects Functional Annotation (RS Number) Yap1 None Dync2h1 None Birc3 Cn:Birc3:WR:400 (rs36889432 G/T) Kbtbd3 None Abbc4 None Gpc6 Cn:Gpc6:CF:298 (rs32608490 G/T) Il17ra None Wnk1 Cn:Wnk1:PT:2075 (rs239559624 G/T) & Cn:Wnk1:TN:1043 (rs6257337 G/T) Cecr2 None WSB (high) Vs. NOD (low) WSB (high) Vs. PWK (low) PWK (high) Vs. 129S1 (low) AR-EAE Incidence 9: 3.37-12.50 14: 103.97-118.63Cumulative Disease Score Monophasic EAE Incidence 6: 118.97-121.49 Eaecc3 Eaecc6 Eaecc5 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 4, 2024. ; https://doi.org/10.1101/2024.06.03.597205doi: bioRxiv preprint