key: cord-0320194-38f70q38
authors: Kim, Dongkyun; Rai, Nagendra Kumar; Burrows, Amy; Kim, Sohee; Tripathi, Ajai; Weinberg, Samuel E.; Dutta, Ranjan; Sen, Ganes C.; Min, Booki
title: IFN-induced protein with tetratricopeptide repeats 2 (Ifit2) limits autoimmune inflammation by regulating myeloid cell activation and metabolic activity
date: 2022-03-11
journal: bioRxiv
DOI: 10.1101/2022.03.11.483954
sha: d71f2bbed39b5b60837f6942ec45c5f17a1231f8
doc_id: 320194
cord_uid: 38f70q38

Besides anti-viral functions, Type I IFN expresses potent anti-inflammatory properties and is being widely used to treat certain autoimmune conditions, such as multiple sclerosis (MS). In murine model of MS, experimental autoimmune encephalomyelitis (EAE), administration of IFNβ effectively attenuates the disease development. However, the precise mechanisms underlying the treatment remain elusive. In this study, we report that IFN-induced protein with tetratricopeptide repeats 2 (Ifit2), a type I and type III IFN-stimulated gene, plays a previously unrecognized immune regulatory role during autoimmune neuroinflammation. Mice deficient in Ifit2 display greater susceptibility to EAE and escalated immune cell infiltration in the central nervous system. Ifit2 deficiency is also associated with microglial activation and increased myeloid cell infiltration. Unexpectedly, myelin debris clearance and the subsequent remyelination is impaired in Ifit2-/- CNS tissues. Clearing myelin debris is an important property of reparative M2 type myeloid cells to promote remyelination. Indeed, we observed that bone marrow derived macrophages, CNS infiltrating myeloid cells, and microglia from Ifit2-/- mice express cytokine and metabolic genes associated with proinflammatory M1 type subsets. Taken together, our findings uncover a novel regulatory function of Ifit2 in autoimmune inflammation in part by modulating myeloid cell function and metabolic activity.

Type I IFN (IFN-I) is widely used to treat acute and chronic inflammation including 47 autoimmunity in the central nervous system (CNS), especially multiple sclerosis (MS) (1-4). 48

Despite its broad applications, the precise mechanisms by which IFN-I elicits its treatment 49 effects within the CNS remain largely unclear. In mouse model of MS, experimental 50 autoimmune encephalomyelitis (EAE), IFNb administration effectively ameliorates the severity 51 of the disease (5, 6). IFN-I is also induced within the CNS during EAE, and it was reported that 52 endogenously produced IFN-I plays a key role in regulating EAE pathogenesis, since mice 53 deficient in IFN-I receptor (IFNAR) or in TIR domain-containing adaptor-inducing interferon-b 54 (TRIF), a cytosolic adaptor molecule needed to induce IFN-I production, develop exacerbated 55 EAE (7, 8) . The use of cell type specific Ifnar -/mice further uncovers that IFN-I signaling in T 56 cells or neurons is dispensable and that the signaling in myeloid cells was essential for the anti-57 inflammatory properties of endogenously produced IFN-I (7). Nevertheless, the cellular and 58 molecular mechanisms by which endogenously produced IFN-I mediates its anti-inflammatory 59 functions remain unknown. 60 61 IFN-I orchestrates both innate and adaptive immunity by inducing IFN stimulated genes 62 (ISGs) in multiple cell types (9, 10). IFN-induced tetratricopeptide repeats (Ifit) is a group of 63 ISG protein family consisted of 4 families in human (3 families in mice, Ifit1/ISG56, 64 Ifit2/ISG54, and Ifit3/ISG60). Ifit proteins play important anti-viral roles through their ability to 65 bind viral RNA and to interfere with translation (11-13). The anti-viral functions of Ifit overlap 66 amongst the Ifit family proteins, although the mechanisms utilized by each Ifit member could be 67 unique and distinct (13) (14) (15) . We recently reported that Ifit2 protects mice from lethal neurotropic 68 MHV infection (16) . Pronounced mortality of Ifit2 -/mice during the infection is accompanied by 69 uncontrolled virus replication and spread throughout the brain (16). Interestingly, there is 70 significantly impaired microglial activation and reduced recruitment of inflammatory cells to the 71 brain, suggesting an Ifit2's ability to control microglial activation. In addition to broad anti-viral 72 functions, Ifit is known to play a diverse regulatory role in immunity. In LPS activated 73 macrophages, Ifit1 not only supports IFN-I expression but also limits the inflammatory gene 74 expression (17). Ifit3 negatively regulates TLR3 induced IFNb and Cxcl10 expression in human 75 CNS mononuclear cells were isolated by 70%/30% Percoll density gradient. Surface staining of 120 mononuclear cells was performed with anti-CD4 (RM4-5), anti-CD44 (IM7). anti-CD25 121 For immunohistochemistry, the WT and Ifit2 -/mice were perfused intracardially with 4% 134 paraformaldehyde (PFA) under anesthesia. The spinal cord was isolated and post-fixed with 4% 135 PFA overnight and then placed in 30% sucrose at 4°C. After the samples had sunk, the spinal 136 cord is dissected into cervical, thoracic and lumbar tissue blocks, embedded in OCT compound 137 (Scigen-Tissue Plus) and frozen on dry ice. Cryostat sections (10μm) were cut on a Leica 138 cryotome and affixed to Superfrost™ Plus Microscope Slides (FisherScientific). Sections were 139 thawed, air-dried, rehydrated with PBS and further non-specific binding was blocked using 5% 140 goat serum. The following antibodies were incubated in 0.1% Triton with 2% goat serum in PBS 141 at 4°C overnight: rabbit anti-GFAP (1: 1000, DAKO, Z0334), rabbit anti-IBA-1 (1: 500, Abcam, 142 AB178846), rabbit anti-TMEM119 (1:250, Abcam, ab209064), rat anti-MBP (1: 500, Millipore-143 Sigma, MAB386) and mouse anti-NF200 (1:500, Sigma, N5389) antibodies. Double staining 144 was performed for MBP and GFAP, or IBA-1 and TMEM119). The sections were then washed 145 in PBS and incubated with a secondary antibody conjugated to Alexa Fluor 488 or Alexa Fluor 146 555 (1:1000, Invitrogen) for 2 hrs. After washing, slides were mounted with prolong Gold 147 antifade reagent with DAPI (Invitrogen). Images were taken with a Zeiss AX10 (Imager Z2) 148 confocal microscope. 149 150 Organotypic Slice culture 151

Organotypic cerebellar slice culture were established from 5 to 8-day-old pups from WT and 152

Ifit2 -/mice according to Tripathi et al. (21) . Briefly, brains of P5 to P8 pups were removed and 153 placed in ice-cold artificial cerebral spinal fluid (2 mM calcium chloride, 10 mM glucose, 3 mM 154 potassium chloride, 4 mM magnesium sulphate, 26 mM sodium bicarbonate, 2 mM Ascorbic 155 acid). Excess brain regions were removed and cerebellum was embedded in 2% low-melting 156 agarose with artificial cerebrospinal fluid. Using Vibratome, 300 μm sagittal slices of the 157 cerebellum were sectioned. Two to three slices were placed onto cell-culture inserts (Millicell 158 0.4m, Millipore) in media containing 50% MEM, 25% Horse Serum, 25% Hank's Buffer, 1% 159 GlutaMax, 10 mg/ml Glucose and 1x antibiotic-antimycotic. To study demyelination, slices, 160 after five days in culture, were incubated with 0.5mg/ml lysolecithin for 20h. For remyelination, 161 slices were washed with media (3x) and cultured an additional 6-7 days for remyelination. For 162 immunofluorescence staining, slices were fixed with 4% PFA for 30 min, permeabilized with 2% 163 Triton X-100 for 30 min, and blocked with 10% donkey serum and 0.1% Triton X-100 in PBS 164 for 1 h at room temperature before being incubated in the primary antibodies against 165 neurofilament (mouse anti-NF200, 1:750) and myelin (rat anti-MBP, 1:250) overnight at 4 °C. 166

Primary antibodies were visualized by incubating sections with the appropriate Alexa 167 fluorophore-conjugated secondary antibodies for 1h at room temperature. After mounting the 168 slices, images were taken with a Zeiss AX10 (Imager Z2) confocal microscope. 169 170

Statistical significance was determined by the Mann-Whitney test using a Prism software 172 (GraphPad, San Diego, CA). p<0.05 was considered statistically significant. 173 174

We first examined Ifit mRNA expression in the CNS tissues during EAE. C57BL/6 mice were 177 immunized with MOG peptide in CFA followed by pertussis toxin administration to induce 178 EAE. CNS tissues were harvested at disease onset (7 day), peak (day 14), and recovery (day 21) 179 phases. Expression of all Ifit mRNAs (Ifit1, Ifit2, and Ifit3) in the CNS was determined by qPCR 180 analysis. The expression pattern closely matched with the disease activity (Fig 1) . It was 181 previously reported that endogenous type I IFN production is localized in the target tissues 182 during EAE and that the resulting enhanced IFNAR signaling activity is related to the disease 183 severity (7). Since type I IFN is the primary stimulus inducing IFIT expression (22), increased 184

Ifit mRNA expression in EAE is likely induced by IFN signaling elicited from the endogenously 185 produced type I IFN. It is interesting to note that low but detectable basal level expression of Ifit2 186 mRNA is observed in naïve CNS tissue (Fig 1) . 187

188

We previously reported using Ifit whole locus (IFIT-WL) knockout mice that viral replication is 190 drastically increased in the absence of the Ifit whole locus (19) . To investigate if the lack of Ifit 191 family proteins affects autoimmune inflammation, wild type and IFIT-WL KO mice were 192 induced for EAE. The onset and initial progress appeared to be similar between the groups. 193

However, IFIT-WL KO animals exhibited more severe disease based on the clinical score (Fig 194 2A) . Consistent with the disease severity, the total numbers of CD4 T cells infiltrating the CNS 195 tissue were significantly increased in KO mice, although the proportion was comparable (Fig  196   2B ). Foxp3 + regulatory T (Treg) cells accumulated in the inflamed sites are important in 197 resolving inflammation (23). The proportion of CNS infiltrating Treg cells was similar between 198 the groups ( Fig 2B) ; however, the absolute number of Treg cells was greater in KO mice. 199 Surface expression of Treg cell associated markers, such as ICOS, GITR, and CD25, remained 200 comparable ( Fig 2C) . These results suggest that the susceptibility seen in KO mice is not 201 attributed to defects in Treg cell recruitment or activation. We then found that CNS accumulation 202 of effector CD4 T cells expressing encephalitogenic cytokines, IL-17, TNFa, GM-CSF, and 203

IFNg, were markedly increased in both proportions and total numbers ( Fig 2D) . This was further 204 supported by gene expression level shown in Supplementary Fig 1. None of the Ifit mRNA was 205 found in IFIT-WL KO mice, while all three Ifit mRNAs were drastically increased in wild type 206 CNS tissues (Supp Fig 1A) . Encephalitogenic cytokine mRNA expression was consistent with 207 flow cytometry data. Il17a, Il1b, Tnfa, and Rorc mRNA expression was markedly elevated in 208 IFIT-WL KO mice compared to those in WT control mice (Supp Fig 1B) . These results 209 demonstrate that Ifit proteins play an important immune regulatory role in controlling 210 encephalitogenic inflammatory responses. 211 212

While the canonical functions of the Ifit proteins could be overlapping, there is evidence that 214 each Ifit could still play a distinct role. For example, Ifit2 protects mice from lethal VSV induced 215 neuropathogenesis, while Ifit1 protein is dispensable in anti-VSV immunity (24). Ifit2 also plays 216 a unique role in protecting mice from coronavirus induced encephalitis by regulating microglial 217 activation and leukocyte migration (16). Thus, we investigated if Ifit2 plays a role in EAE 218 independent of other Ifit proteins. Ifit2 -/mice were induced for EAE. Analogous to what we 219 observed from IFIT-WL KO mice, we found that the disease onset and initial progress were 220 comparable in WT and Ifit2 -/mice. However, Ifit2 -/mice failed to recover from the acute 221 paralysis and continued to display severe disease as in IFIT-WL KO mice ( Fig 3A) . CD4 T cells 222 infiltrating the CNS tissues were markedly increased in Ifit2 -/mice, although the proportions of 223 CD4 T cells seem comparable between the groups as seen in IFIT-WL KO mice ( Fig 3B) . 224

Proportion of CNS infiltrating Treg cells were similar (data not shown); however, the absolute 225 numbers were elevated in Ifit2 -/mice ( Fig 3B) . Likewise, Treg cell marker expression such as 226 ICOS, GITR, and CD25 remained unchanged (data not shown). We then examined 227 proinflammatory cytokine expression profiles in CNS infiltrating CD4 T cells. As shown in Fig  228   3C , CD4 T cell expression of inflammatory cytokines was drastically increased in Ifit2 -/mice. 229

Consistent with escalated proinflammatory CD4 T cell profiles in the CNS, the expression of key 230 transcription factors for Th1 and Th17 immunity, Tbx21 and Rorc, was similarly increased in 231

Ifit2 -/mice ( Fig 3D and data not shown) . Therefore, Ifit2 alone may play an immune regulatory 232 function to control the development of encephalitogenic immune responses. 233 234 Ifit2 deficiency did not affect IFN-I expression in EAE (data not shown). In support, 235 other Ifit, Ifit1 and Ifit3, mRNA expression in the CNS tissues of Ifit2 -/mice remained 236 comparable to that of wild type mice, strongly suggesting that endogenous IFN-I production is 237 not affected by the absence of Ifit2 (Fig 3E) . This result also suggests that the susceptibility of 238

Ifit2 -/mice to EAE cannot be attributed to impaired IFN-I or other Ifit expression and that Ifit2 239 plays a unique role in modulating inflammatory responses. 240 241 Severe EAE and increased T cell infiltration in the CNS were further corroborated by 242 enhanced inflammatory chemokine expression. CCL2 and CCL3 production in the CNS tissue 243 critically regulates EAE pathogenesis through recruiting macrophages and dendritic cells (25, 244 26) . Consistent with severe pathogenesis in Ifit2 -/mice, we found that Ccl2 and Ccl3 mRNA 245 expression particularly in the spinal cord was significantly elevated (Fig 3F) . CCL7 is an 246 alternative CCR2 ligand potentially involved in EAE pathogenesis (27, 28), and indeed the Ccl7 247 expression was also found increased in the Ifit2 -/spinal cord ( Fig 3F) . Overexpression of 248 CXCL1 in astrocytes enhances EAE development by supporting inflammatory cell recruitment 249 (29). Likewise, CXCR3 ligands, CXCL9 and CXCL10 have been implicated in EAE, where the 250 expression is significantly increased in the CNS (30). Consistent with these reports, we found 251 that Ifit2 deficiency markedly increased all those chemokines Cxcl1, Cxcl9, and Cxcl10 ( Fig 3F) . 252 253 APC-derived IL-12 family cytokines are especially important in tuning the activation of 254 pathogenic T cells such as Th17 type cells. We found that IL-23p19 subunit expression was 255 dramatically increased in Ifit2 -/mice, although IL-12p35 and IL-12p40 subunit expression 256 remained unchanged (Supp Fig 2) . IL-27, composed of IL-27p28 and EBI3 subunits and induced 257 by IFN-I signaling, plays an important role in encephalitogenic inflammation (8, 31) . While IL-258 27 could mediate protective roles in EAE pathogenesis, its expression also represents the extent 259 of inflammatory activity. In support, we found elevated expression of both subunits of IL-27 in 260

Ifit2 -/mice (Supp Fig 2 and 

An imbalance between demyelination and remyelination could trigger disease process in 266 multiple sclerosis, and clearing myelin debris derived from demyelination is an important 267 process promoting remyelination (32). Myeloid cells including monocyte-derived macrophages 268 or microglia are the central cell types mediating myelin debris clearance (32). Macrophages and 269 microglia undergo differentiation into two distinct subsets, proinflammatory M1 and regulatory 270 M2 phenotype cells, and M2 phenotype cells are specifically known to induce reparative 271 processes (33). In case of demyelinating inflammation, it was reported that M2 phenotype 272 microglia remove myelin debris from the lesion, further promoting remyelination process (34, 273 35). We thus compared infiltrating myeloid cells and CNS resident microglial cells. Infiltrating 274 myeloid cells defined as CD45 high CD11b high cells were highly abundant in Ifit2 -/mice (Fig 4A) . 275

Their MHCII expression was significantly higher in Ifit2 -/mice. Importantly, expression of 276 iNOS, a marker for M1 type pro-inflammatory subset, was also higher in Ifit2 -/cells. On the 277 other hand, expression of the mannose receptor, CD206, a marker for M2 type anti-inflammatory 278 subset, was substantially lower in Ifit2 -/cells ( Fig 4A) . Microglia, tissue resident macrophages in 279 the CNS, have also been implicated in neuroinflammation (36), and are important sources of 280 cytokines and chemokines. They can similarly polarize into M1 or M2 phenotype cells. In EAE, 281 microglia undergo expansion and M1 type differentiation (37, 38) . In support, we observed 282 greater expansion of microglia population (CD45 int CD11b high ) in Ifit2 -/mice ( Fig 4B) . Likewise, 283 microglia expression of MHCII molecule was drastically elevated in Ifit2 -/mice. Unlike 284 infiltrating myeloid cells, upregulation of M1-like marker, iNOS, was not observed. However, 285 downregulation of CD206 was pronounced ( Fig 4B) . Therefore, Ifit2 deficiency appears to have 286 significant impact on myeloid cells and microglia activation and polarization, possibly 287 contributing the pathogenesis. 288 289

We next investigated if Ifit2 deficiency alters demyelination/remyelination during autoimmune 291 inflammation of EAE. Immunohistochemistry of coronal spinal cord sections from WT and Ifit2 -292 /mice with EAE revealed significant changes in myelination and axonal density in the cervical, 293 thoracic, and lumber spinal cords. As seen by lower MBP staining in spinal cord sections, EAE 294 generated a notable myelin damage in Ifit2 -/relative to WT at the peak of disease (Supp Fig 3A) . 295

The ventral white matter of the thoracic area of the Ifit2 -/mice spinal cord showed widespread 296 and notable myelin degradation. In the WT and Ifit2 -/spinal cords, we performed double 297 immunostaining with antibodies to MBP and NF200 to determine the condition of demyelinated 298 axons and found a rather intact myelin ring (green) around the axons (red). Interestingly, we 299 found several MBP + myelin sheaths in Ifit2 -/mice that were devoid of NF200 labeled axons 300 (Supp Fig 3A, arrows) , implying that axons have been destroyed or damaged and are 301 consequently negative for NF200 antibody. 302 303

Microglia, macrophages, and astrocytic activation are all required for immune-mediated 306 demyelination. We looked at whether the spinal cords of MOG 35-55 immunized WT and Ifit2 -/-307 mice revealed any differences in peripheral macrophage infiltration and/or microglial activation 308 to see if greater microgliosis/ macrophage activity was associated with severe EAE-related 309 pathology in Ifit2 -/animals. In the cervical, thoracic, and lumbar areas of the WT and Ifit2 -/mice 310 spinal cords, immunohistochemistry for ionized calcium binding adaptor molecule-1 (IBA-1) 311 was used to mark macrophage/microglia. The density of IBA-1 + cells in the ventral white matter 312 of all regions of the spinal cord was significantly greater in Ifit2 -/mice than in WT mice, 313 indicating more immune cell infiltration in Ifit2 -/mice ( Fig 5A) . MBP + myelin debris are 314 engulfed by IBA + macrophages, and immune cell infiltration and phagocytosis are more evident 315 in the thoracic area (Fig 5B) , according to our findings. Of note, infection with the neurotrophic 316 coronavirus MHV-RSA59 resulted in decreased microglial activation and infiltration of 317 peripheral immune cells (NK, CD4 T cells) into the brain in Ifit2 -/mice (39), suggesting a 318 unique mode of pathogenesis of EAE in the absence of Ifit2. Microglia and peripheral 319 macrophages have different origins, and microglia are activated prior to peripheral immune 320 infiltration and actively assist in their recruitment, which may also be true in EAE. We stained 321 the spinal cord sections for transmembrane protein 119 (TMEM119), which has emerged as a 322 microglia-specific marker, to assess the alteration in microglial numbers. At the peak of disease, 323 TMEM119 staining revealed that Ifit2 -/mice had a higher number of microglia than WT mice, 324 supporting the flow cytometry data shown in Fig 4B. Ifit2 -/mice's TMEM119 + microglia have 325 an active phenotype with amoeboid morphology and retracted microglial processes (Fig 5C) . 326

Several reports have showed that astrocyte regulate CNS inflammation by affecting infiltration 327 of peripheral immune cells and activation of resident microglia (40, 41). We next determined 328 astrocyte activation by immunohistochemistry using antibody against GFAP. We observed that 329

Ifit2 -/mice show higher GFAP + astrocytes compared to WT at the peak of EAE. The degree of 330 astrocyte reactivity is positively correlated with GFAP expression (42, 43), hence indicates more 331 astrocyte activation in Ifit2 -/mice. 332 333

Ifit2 -/-CNS tissues 335 M2 type microglia and macrophages support CNS remyelination (34). Since both infiltrating 336 monocytes and microglia expressing M2 phenotype (CD206) were diminished in Ifit2 -/mice, we 337 sought to test if such defects in M2 type subset differentiation are associated with impaired 338 remyelination and with severe pathogenesis seen in Ifit2 -/mice. We used the model for 339 organotypic slice cultures and lysolecithin (LPC, 0.5mg/ml) induced demyelination and 340 remyelination in P10 control and Ifit2 -/brains using our recent protocols (21, 44) . Slices stained 341 for myelin protein MBP and axonal marker NF200 show demyelination by LPC (Fig 6C) 342 compared to control (Fig 6A) . The extent of demyelination was, however, strikingly pronounced 343 in slices from Ifit2 -/brains (Fig 6G) . Of particular interest was the lack of myelin debris 344 clearance in the Ifit2 -/brains (Fig 6G, shown by the yellow outline). We then examined the 345 extent of remyelination as the lack of myelin clearance by the microglia and/or macrophages 346 could have negative impact on remyelination (32). Indeed, spontaneous remyelination was 347 observed in control slices (Fig 6E) , while remyelination was severely impaired in Ifit2 -/brains 348 ( Fig 6I) . Of note, Ifit2 deficiency did not induce axonal loss (Fig 6D, 6F , 6H, and 6J), suggesting 349 defects in remyelination in Ifit2 -/brains. As the overall number of OL lineage cells were 350 comparable in control and Ifit2 -/mouse brains (data not shown), defective remyelination appears 351 to be attributed to impaired myelin debris clearance. 352 353

Altered M1/M2 phenotypes in infiltrating myeloid cells and in microglia (Fig 4) , exacerbated 355 demyelination (Fig 5) , and defects in myelin debris clearance and remyelination observed in 356

Ifit2 -/brain tissues prompted us to test if Ifit2 deficiency dysregulates metabolic programming of 357 those myeloid cells. Notably, inflammatory M1 phenotype cells predominantly utilize glycolytic 358 pathways as an energy source, whereas reparative M2 phenotype cells preferentially utilize 359 oxidative phosphorylation (45, 46) . We first measured proinflammatory and metabolic gene 360 expression in WT and Ifit2 -/bone marrow derived macrophages (BMDM) following stimulation 361 with various stimuli, including LPS, IFNg, and IL-4 to promote M1 and M2 differentiation, 362 respectively. We found that Ifit2 -/-BMDM expressed greater proinflammatory cytokines, Tnfa 363 and Il1b, following LPS stimulation. In addition, glycolytic gene, Hif1a and Hk2 expression was 364 also elevated in Ifit2 -/-BMDM cells (Fig 7A) . Interestingly, IFNg stimulation dramatically 365 enhanced Il1b mRNA expression in Ifit2 -/-BMDM cells, although other tested genes remained 366 unchanged. IL-4 stimulation, which promotes M2 macrophage differentiation (47), also 367 strikingly enhanced Il1b and Hif1a expression in Ifit2 -/-BMDM cells (Fig 7A) . Therefore, Ifit2 368 deficiency is associated with dysregulated inflammatory cytokine and metabolic gene expression 369 in myeloid cells. 370

To further investigate metabolic program in vivo, we isolated infiltrating myeloid cells 372 (CD45 high CD11b high ), microglia (CD45 int CD11b high ), and CD45 low cells (including 373 oligodendrocytes and astrocytes) from wild type and Ifit2 -/mice induced for EAE. Genes 374 involved in glycolytic pathway were measured. As shown in Fig 7B, expression of Hk2, Pdk1,  375 and Hif1a was strikingly enhanced in Ifit2 -/mice compared to WT mice. Of note, the enhanced 376 gene expression was most pronounced in infiltrating myeloid cells and microglia. Therefore, 377 these results strongly suggest that Ifit2 may control metabolic programming of myeloid cells via 378 M1/M2 type differentiation, which may then have direct impact on disease progression, 379 particularly remyelination process through myelin debris clearance. 380

Discussion 383

In the present study, we report an unexpected role of Ifit2, one of the type I IFN-induced gene 384 product known for the anti-viral functions, in modulating autoimmune inflammation. Mice 385 deficient in the Ifit2 gene were highly susceptible to myelin antigen-reactive autoimmunity, 386 EAE. The susceptibility was closely associated with heightened encephalitogenic T cell 387 responses that were accompanied by elevated proinflammatory cytokine and chemokine 388 production. Importantly, we noticed that the lack of Ifit2 resulted in severe demyelination and 389 impaired remyelination process, which appear to contribute to the exacerbated disease. Myelin 390 debris clearance predominantly mediated by reparative macrophage and microglia subset is 391 known to play a key role in supporting remyelination. One notable feature of proinflammatory 392 and tissue repair subsets of myeloid cells is their metabolic program, i.e., glycolytic pathways in 393 proinflammatory M1 type cells vs. oxidative phosphorylation in reparative M2 type cells. 394

Indeed, we found that, in Ifit2 -/infiltrating myeloid and CNS resident microglia, expression of 395 genes involved in glycolytic pathways is markedly increased. Therefore, our results demonstrate 396 that Ifit2 plays a novel regulatory function limiting inflammatory responses, in part by balancing 397 metabolic profiles of myeloid lineage cells. 398 399 To our knowledge, the present study reports for the first time that Ifit2 expresses a novel 400 immune regulatory function in autoimmunity. In EAE, all three Ifit family proteins (Ifit1, Ifit2, 401 and Ifit3) are strongly induced in the spinal cord, and the kinetics of the expression closely 402 mirror the disease activity. Given that endogenous IFN-I production in EAE is predominantly 403 found in the target CNS tissues (7), Ifits induced by endogenous IFN-I are also likely localized in 404 the CNS. The entire Ifit locus knockout animals display exacerbated disease, and Ifit2 deficiency 405 alone results in similar susceptibility to EAE, suggesting that Ifit2's ability to regulate 406 autoimmune inflammation may be unique to Ifit2 protein. In support, IFN-I, Ifit1, and Ifit3 407 expression in Ifit2 -/mice remain unchanged, strongly suggesting that high susceptibility of Ifit2 -/-408 mice to EAE is not due to defective IFN-I signaling or other Ifit protein production. 409 410 Mice deficient in Ifnar are highly susceptible to EAE induction, suggesting that IFN-I 411 stimulation and subsequent induction of Ifit2 are crucial in limiting inflammatory responses (7). 412

Likewise, the lack of Trif, an adaptor molecule crucial to induce IFN-I production, results in 413 exacerbated EAE (8) . The precise upstream signal that activates TRIF to induce IFN-I 414 production within the CNS remains to be determined. Protective functions of TRIF suggest that 415 ligands activating TLR3 or TLR4 may be involved in inducing IFN-I production in EAE. In 416 support, TLR3 stimulation was previously shown to suppress EAE (48) . Moreover, neonatal 417 exposure to LPS can suppress EAE by promoting tolerogenic dendritic cells (49). However, 418 pretreatment of LPS only prior to EAE induction is known to delay onset of the disease, while 419 the severity remains unchanged (50). Paradoxically, mice deficient in Tlr4 surprisingly develop 420 more severe EAE, suggesting the pathogenic role (51). Alarmin S100A9 is an alternative ligand 421 for TLR4 (52), and S100A9-deficient mice were previously shown to develop exacerbated EAE, 422 suggesting protective roles (53). It will be important to precisely determine endogenous signals 423 capable of activating the TRIF-IFN-I-IFIT signaling pathways in EAE pathogenesis. 424 425 Cell type specific Ifnar-deficient mouse models were previously used to determine the 426 target cells responsible for the severe EAE seen in Ifnar -/mice (7). T cell-or brain-expression of 427

Ifnar is found dispensable, while targeting the Ifnar in myeloid lineage cells phenocopies global 428 IFNAR deficiency, suggesting that resident and inflammatory monocytes as well as CNS 429 resident microglia are likely target cells of IFN-I (7). Based on these findings, we speculate that 430 myeloid lineage cells are likely the source of Ifit2-expressing cells. Our recent report similarly 431 identified a unique function of Ifit2 that protects the host from lethal neurotropic MHV infection 432 by modulating microglia activation and recruitment (16). The exact mechanisms by which IFN-I 433 is endogenously produced (i.e., via TRIF-dependent mechanism) and acts on the target cells (i.e., 434 via IFNAR-IFIT2-dependent mechanism) to mediate protective roles will require further 435 investigation. 436 437 It is interesting to note that Ifit2 deficiency is closely associated with severe 438 demyelination and defective remyelination. Following demyelination, macrophages and 439 microglia play a crucial role in clearing myelin debris, which is known to be instrumental to 440 support for the subsequent remyelination to occur (32). From the LPC-induced 441 demyelination/remyelination organotypic slice culture and immunofluorescence microscopic 442 examination experiments, we observed that Ifit2 deficiency is associated with defective myelin 443 debris clearance and remyelination processes. Importantly, myelin debris clearance is one key 444 property of M2-type macrophage/microglia (32). Therefore, it is possible that Ifit2 deficiency 445 may impair transition from proinflammatory M1 phenotype cells to reparative M2 phenotype cell 446 differentiation, which occurs during EAE (34, 54). M1 and M2 type macrophages (and 447 microglia) utilize different metabolic programs, glycolysis and oxidative phosphorylation, 448 respectively. Indeed, we noticed elevated glycolytic gene expression in Ifit2 -/-BMDM. In 449 addition, Ifit2 -/-BMDM expression of proinflammatory cytokines, especially IL-1b, was 450 markedly increased compared to wild type cells, even after IL-4 stimulation, which is known to 451 promote M2 differentiation (47). Consistent with this, IFN-I promotes fatty acid oxidation and 452 oxidative phosphorylation (55). Therefore, we would argue that Ifit2 may be an important switch 453 that balances metabolic programming of macrophages (and microglia) to trigger remyelinating 454 tissue repair processes by supporting M2 differentiation. While IFN-I is a widely used 455 therapeutic option in MS patients, there is a cohort of patient refractory to the treatment (56). 456

Better defining the precise cellular and molecular mechanisms will be a subject of great 457 importance to identify novel therapeutic approaches for both IFN-I-responsive and -refractory 458 Taqman primers Assay ID

Il17a Mm00439618_m1

Hif1a Mm00468869_m1

Hk2 Mm00443385_m1

Mm00554300_m1

Type I interferons as anti-inflammatory mediators

Regulation of type I interferon signaling in immunity 465 and inflammation: A comprehensive review

Defining and scoring response to IFN-beta in 467 multiple sclerosis

Interferons in relapsing remitting multiple sclerosis: a systematic review

Interferon-beta therapy against EAE is effective only when 473 development of the disease depends on the NLRP3 inflammasome

Engagement of the type 476 I interferon receptor on dendritic cells inhibits T helper 17 cell development: role of 477 intracellular osteopontin

Distinct and nonredundant in vivo functions of IFNAR on 481 myeloid cells limit autoimmunity in the central nervous system

The type I IFN induction pathway constrains 483 Th17-mediated autoimmune inflammation in mice

The Dual Nature of Type I and Type II Interferons. 486 Front Immunol

Type I 488 interferons in infectious disease

Emerging 490 Functions of Human IFIT Proteins in Cancer

The broad-spectrum antiviral functions of IFIT 492 and IFITM proteins

Human IFIT3 Modulates IFIT1 RNA Binding Specificity and Protein Stability

IFIT1 is an antiviral protein that 500 recognizes 5'-triphosphate RNA

Crystal structure of ISG54 reveals a novel RNA binding structure and 503 potential functional mechanisms

Ifit2 deficiency restricts microglial activation and

leukocyte migration following murine coronavirus (m-CoV) CNS infection

IFIT1 Exerts Opposing Regulatory Effects on the Inflammatory and 510

Interferon-stimulated gene 60 (ISG60) constitutes a 514 negative feedback loop in the downstream of TLR3 signaling in hCMEC/D3 cells

Biphasic and 517 cardiomyocyte-specific IFIT activity protects cardiomyocytes from enteroviral infection

Ifit2 deficiency results in uncontrolled neurotropic coronavirus 521 replication and enhanced encephalitis via impaired alpha/beta interferon induction in 522 macrophages

Interferon induced IFIT family genes in host antiviral defense

New insights into the resolution of inflammation

Interferon-induced Ifit2/ISG54 533 protects mice from lethal VSV neuropathogenesis

Production of CCL2 by central nervous 536 system cells regulates development of murine experimental autoimmune 537 encephalomyelitis through the recruitment of TNF-and iNOS-expressing macrophages 538 and myeloid dendritic cells

Crucial role of 540 CCL3/MIP-1alpha in the recurrence of autoimmune anterior uveitis induced with myelin 541 basic protein in Lewis rats

CCR2+Ly-6Chi monocytes are crucial for the effector phase 544 of autoimmunity in the central nervous system

miR-23b Suppresses Leukocyte Migration and Pathogenesis of 547 Experimental Autoimmune Encephalomyelitis by Targeting CCL7

Induced CNS expression of CXCL1 augments

Molecular dissection of reactive astrogliosis and glial scar 597 formation

Heterogeneity of reactive 599 astrocytes

Hippocampal demyelination and memory dysfunction are associated 603 with increased levels of the neuronal microRNA miR-124 and reduced AMPA receptors

The 606 Metabolic Signature of Macrophage Responses

Glycolytic Stimulation Is Not a Requirement for M2 Macrophage 609 Differentiation

Microglial 611 and macrophage polarization-new prospects for brain repair

Cutting Edge: 614 TLR3 stimulation suppresses experimental autoimmune encephalomyelitis by inducing 615 endogenous IFN-beta

Early life exposure to lipopolysaccharide suppresses experimental 618 autoimmune encephalomyelitis by promoting tolerogenic dendritic cells and regulatory T 619 cells

Lipopolysaccharide pretreatment modulates 621 the disease course in experimental autoimmune encephalomyelitis

Unexpected 624 regulatory roles of TLR4 and TLR9 in experimental autoimmune encephalomyelitis

627 CD14 Is a Co-Receptor for TLR4 in the S100A9-Induced Pro-Inflammatory Response in 628 Monocytes

Identification of human S100A9 as a novel target for treatment of 631 autoimmune disease via binding to quinoline-3-carboxamides

Targeting the shift from M1 to M2 macrophages in experimental 636 autoimmune encephalomyelitis mice treated with fasudil

Type 1 Interferons Induce Changes in Core Metabolism that Are Critical for Immune 641 Function

Pharmacogenomic analysis of 644 interferon receptor polymorphisms in multiple sclerosis

Fig 1. Expression of Ifit mRNA in the CNS during EAE. C57BL/6 mice were injected with 659

CFA/MOG35-55 followed by pertussis toxin on the day of immunization and 48 h later to induce 660

day), peak (day 14), and 661 recovery (day 21) phases. Data were normalized to naive mice. n = 3-4 per group

**p 662 < 0.01; as determined by Mann-Whitney nonparametric test

Increased susceptibility to EAE in IFIT-WL KO mice. WT (n = 9

= 9) mice induce for EAE. (A) Evaluated for clinical signs for 20 d. (B) Absolute number of 669

KO 670 mice at the peak of disease (day 17 post immunization). (C) The Mean Fluorescence Intensity 671 (MFI) of CD44, ICOS, GITR and CD25 on CD4 + Foxp3 + Treg cells in the CNS. (D) Total 672 numbers of GM-CSF, IFNγ, IL-17, and TNFa expressing CD4 + T cells infiltrating the CNS at 673 the peak of disease

0.001; as determined by Mann-Whitney nonparametric test

Ifit2 -/-mice were more susceptible to EAE than WT mice. WT (n = 10) and Ifit2 -/-(n = 679

mice induce for EAE. (A) Time course of the development of EAE. (B) Absolute number of 680

CNS-infiltrating CD4 + T cells and CD4 + Foxp3 + Treg cells in the CNS of WT and Ifit2 -/-mice at 681 the peak of disease (day 17 post immunization). (C) Numbers of GM-CSF, IFNγ

TNFa expressing CD4 + T cells infiltrating the CNS at the peak of disease (day 17 post 683 immunization). n = 8-9 per group. (D-F) qPCR analysis of the indicated mRNAs in the brain and 684 spinal cords from naïve, WT, and Ifit2 -/-mice at the peak of disease

Representative flow 691 cytometry plots of infiltrating myeloid cells (A) and of microglia (B) from CNS of WT and Ifit2 692 -/-mice. Absolute numbers of myeloid cells (CD45 hi CD11b hi ) and microglia

The Median Fluorescence Intensity (MFI) of MHCII, iNOS and CD206 expression measured by 694 flow cytometry. n = 4 per group

Fig 5. Ifit2 deficiency augments number of macrophage/microglia following EAE induction 700 at the peak of disease. Immunohistochemical staining of macrophage/microglia cell types in 701

ventral white matter of cervical, thoracic, and lumbar spinal cord of WT and Ifit2 -/-mice

Inset images showing 703 engulfment of myelin debris by IBA1 + macrophage/microglia in cervical, thoracic, and lumbar 704 regions (arrow, Scale bar: 5 µm) (B) TMEM119 staining, a marker of CNS local microglia in the 705

ventral white matter of cervical, thoracic, and lumbar spinal cord of WT and Ifit2 -/-(Scale bar: 706 20µm). Insets display examples of microglial cells in WT (ramified

and Ifit2 -/-(amoeboid, round) mice (Scale bar: 5 µm)

Mouse brain organotypic sections from control mice 714 stained for myelin basic protein (MBP) and axonal marker NF200. Organotypic slices from 715 control and Ifit2 -/-P10 mice were treated with lysolecithin (LPC) (0.5mg/ml) for 18 hrs to induce 716 demyelination

There was comparable number of axons between the two 719 conditions (D,H). Following demyelination, the slices were allowed to recover and remyelinate 720 for 7 days. Similar staining was carried out

Ifit2 -/-brain slices had drastically lower extent of remyelination (I-J) despite 722 almost similar levels of axons between conditions (F,J). There was myelin debris found during 723 remyelination (yellow border; I)

Cytokines and glycolytic gene expression in Ifit2 -/-mice. (A) mRNA expression of Tnfa

Hif1a, and Hk2 in bone marrow-derived macrophages from WT and Ifit2 -/-mice were 730 incubated with lipopolysaccharide (LPS; 1 µg/mL) or IFNg (20ng/ml) or IL-4 (5000U/mL

= 3 per group. (B) CNS tissues harvested at the peak of disease were measured for 732

glycolytic gene (Hk2, Pdk1, and Hif1a) expression by qPCR. (C) Sorted CD45 high CD11b high 733 (infiltrating monocyte), CD45 int CD11b high (microglia) and CD45 low

oligodendrocyte) cells from CNS of EAE induced mice (day 17 post immunization) and the 735 indicated gene expression were measured by qRT-PCR. n = 3-4 per group

< 736 0.01; as determined by Mann-Whitney nonparametric test