Increased activity of IRE1 improves the clinical presentation of EAE

Activation of the endoplasmic reticulum (ER) stress sensor inositol‐requiring enzyme‐1α (IRE1α) contributes to neuronal development and is known to induce neuronal remodeling in vitro and in vivo. On the contrary, excessive IRE1 activity is often detrimental and may contribute to neurodegeneration. To determine the consequences of increased activation of IRE1α, we used a mouse model expressing a C148S variant of IRE1α with increased and sustained activation. Surprisingly, the mutation did not affect the differentiation of highly secretory antibody‐producing cells but exhibited a beneficial effect in a mouse model of experimental autoimmune encephalomyelitis (EAE). Although mechanical allodynia was unaffected, significant improvement in motor function was found in IRE1C148S mice with EAE relative to wild type (WT) mice. Coincident with this improvement, there was reduced microgliosis in the spinal cord of IRE1C148S mice, with reduced expression of proinflammatory cytokine genes. This was accompanied by reduced axonal degeneration and enhanced 2',3'‐cyclic nucleotide 3'‐phosphodiesterase (CNPase) levels, suggesting improved myelin integrity. Interestingly, while the IRE1C148S mutation is expressed in all cells, the reduction in proinflammatory cytokines and in the microglial activation marker ionized calcium‐binding adapter molecule (IBA1), along with preservation of phagocytic gene expression, all point to microglia as the cell type contributing to the clinical improvement in IRE1C148S animals. Our data suggest that sustained increase in IRE1α activity can be beneficial in vivo, and that this protection is cell type and context dependent. Considering the overwhelming but conflicting evidence for the role of ER stress in neurological diseases, a better understanding of the function of ER stress sensors in physiological contexts is clearly needed.


| INTRODUCTION
Inositol requiring enzyme 1 alpha (IRE1α) is the ancient and most conserved stress sensor that initiates the unfolded protein response (UPR).When activated, IRE1α cleaves the transcript XBP1 and removes a short intron. 1 The two remaining RNA ends require non-conventional ligation, 2 mediated by the RNA 2′,3′-cyclic phosphate and 5′-OH ligase RtcB, 3 creating spliced XBP1 (XBP1s).XBP1s encodes an active basic-leucine zipper transcription factor, 1 which regulates the expression of hundreds of stress responsive genes, such as chaperones (e.g., Grp78/BiP) and lipid biosynthesis genes for ER expansion. 4,5Inability of the UPR to re-establish protein homeostasis following stress is linked to several neurodegenerative diseases such as Alzheimer's or Parkinson's diseases (reviewed in Ref. [6]).The UPR is also an important regulator of neuroinflammation. 7It is upregulated in experimental autoimmune encephalomyelitis (EAE), a demyelinating disease that affects nerve conduction and leads to paralysis and serves as a common mouse model for multiple sclerosis (MS), (reviewed in Ref. [5]) which currently has no cure.The disease is mediated by T cells that autoreact to myelin components and cause inflammation in the central nervous system. 8Because several UPR proteins are upregulated in MS and EAE lesions, 9,10 it stands to reason that modulating the UPR response may be a useful therapeutic intervention.Yet, it is unclear whether upregulation of the UPR seen in neurodegeneration, including in MS or EAE, represents an adaptive response that is insufficient, or an overactivated response with a known potential to be maladaptive.The maladaptive potential of overactivated IRE1 pathway is linked to pro-apoptotic signaling. 11Despite that, selective activation of the IRE1α pathway can be protective in stress conditions. 12Indeed, ectopic overexpression of spliced XBP1 was neuroprotective in a model of EAE/optic neuritis when coupled with inhibition of PERK pathway. 13The goal of our study was to determine whether increasing activation of IRE1 in a physiological context has a protective effect in the mouse EAE model.We had previously shown that a conserved cysteine residue (human C148/murine C150) in the luminal domain of IRE1α serves as a critical binding site for the negative regulator of IRE1, the protein disulfide isomerase family A member 6 (PDIA6), and thus determines the magnitude and duration of IRE1 activation. 14IRE1C148S mutant protein exhibits enhanced and prolonged activity in mouse embryonic fibroblasts or 293T cells under stress conditions, without inappropriate activation in unstressed cells. 14Thus, we generated a mouse carrying the IRE1C148S mutant form of IRE1α with the prediction that enhancing IRE1α activity would lead to higher capacity to cope with stress and therefore benefit cells with high secretory loads, such as oligodendrocytes and neurons in the CNS and antibody-secreting B cells in the periphery, by increasing their secretory capacities.To our surprise, plasma cells from IRE1C148S had the same antibody secretion capabilities as their WT counterparts.However, enhanced IRE1α activity was beneficial for reducing the symptoms of EAE in mice, showing potential for IRE1α activation in the treatment of MS.

| Key results
• Generation and characterization of a novel IRE1C148S variant mouse with activity-enhancing mutation.
The purified PCR product (Qiaquick PCR purification kit, Qiagen) was sequenced using the forward primer (Genewiz, South Plainfield, New Jersey).

| RNA extraction and quantitative real-time PCR
Total RNA from lumbar spinal cords from WT and IRE1C148S 31d post-EAE (and controls) were extracted using TRIzol™ (ThermoFisher Scientific) and further purified on RNAeasy mini columns using on column DNA digestion according to manufacturer's protocol (Qiagen).
For each sample, an aliquot of RNA (500 ng) was reverse transcribed using the Ominiscript RT kit (Qiagen).Realtime PCR quantification was performed using Sybr green and primers specific for the gene of interest, using the standard curve method, where a standard curve is run with known amounts of the target PCR product, and the experimental sample concentrations are determined based on the standard curve.Every RT reaction is also subjected to amplification with β-actin primers as a control, and all data are normalized to β-actin.

| Western blot analysis
Proteins were extracted from different brain regions (cortex, hippocampus, cerebellum) and spinal cord of WT and IRE1C148S mutant mice as previously described. 15Equal amounts of each sample (20 g) were loaded on an 8% or 15% SDS-PAGE depending on the size of the protein to be detected.After transfer to nitrocellulose membrane using the TurboBlot system (Biorad), primary antibodies were applied: rabbit anti-CNPase (1:1000, Cell signaling), rat anti-MBP (1:40 000, Millipore), mouse anti-NF200 (1:1000, Sigma), and rabbit anti-IBA1 (1:500, Cell signaling).Species-specific HRP-conjugated secondary antibodies from ImmunoJax were used.Following incubation with West Pico ECL substrate (Thermo Fisher Scientific), blots were imaged using the Gel Doc XR+ (BioRad) and analyzed with image J software (NIH).Protein quantification was normalized to the total protein for each sample determined after transfer to the membrane using Ponceau S solution (Sigma).

| EAE induction
Induction of EAE was achieved by subcutaneous immunization of adult mice with an emulsion of MOG 35-55 (Bio-Synthesis Inc.) peptide in complete Freund's adjuvant containing heat-inactivated Mycobacterium tuberculosis (MT) H37Ra as previously described by Ref. [16].Each mouse received 2 s.c.injections of 50 μL each into the posterior right and left flanks (total of 100 μg of MOG and 200 μg of MT per mouse).A second injection of MOG/ CFA was administered 7 days later.Mice were monitored daily for clinical symptoms and body weight loss.Clinical scores were as follows: 0-no motor deficits, 1-tip of the tail is flaccid, 2-Tail completely flaccid, 3-Hindlimb paralysis or ataxia, impaired balance/ambulation, 4-Forelimb paralysis, and 5-Moribund.

| Pain testing
Mechanical allodynia measurement was performed using the Von Frey test as previously described. 17In brief, filaments of different sizes (0.02 g-2 g) are applied to the hind paw of the mouse until a withdrawal response is elicited.The response is consodered painful only when the withdrawal response is paired with cognitive awareness. 2.9 | Electron microscopy, G-ratio, and degenerated axons WT and IRE1C148S naive or EAE d31 mice were deeply anesthetized with ketamine and xylazine prior to perfusion with PBS followed by 4% PFA in PBS.Spinal cords were dissected out then transferred to 2% glutaraldehyde, 100 mM sucrose in 0.05 M Sorensen's phosphate buffer and sent to the Transmission Electron Microscopy Core facility of the Miami Project to Cure Paralysis (University of Miami Miller School of Medicine) for processing.The G-ratio, a measure of myelin thickness and compaction, calculated as fiber (axon + surrounding myelin) diameter/ axon diameter, was determined as previously described. 18he total number of axons out of 13 pictures from each animal was obtained as well as the number of degenerated axons identified as axons with darkened axoplasm, and percentages of degenerated axon were calculated.

| B-cell isolation and in vitro differentiation
Spleens of WT and IREC148S mice (about 4-month-old) were removed and the tissues were disrupted between two frosted slides and filtered through a Nitex mesh, followed by red blood cell lysis with ACK lysis buffer. 19Splenocytes from 10-week-old mice were plated at 1 × 10 6 cells/mL in 96-well culture plates in complete growth media (RPMI, 10% FBS, HEPES, Pen/Strep, NEAA, Na-Pyruvate, 2-mercaptoethanol).Cultures were treated with LPS 0.1 mM and IL5 (100 ng/mL) where indicated and incubated at 37°C/5% CO 2 for the indicated times.

| Flow cytometry
Staining of blood cells and flow cytometry was performed as described in Ref. [19,20].Following red blood cell lysis, splenocytes were stained with the fixable viability Zombie-Aqua dye (Biolegend) according to manufacturer's protocol.Flow cytometry was performed using either BD LSRII or LSR Fortessa analyzers, and data analysis was performed using Flowjo 8.8.7 software.Cells were also sorted and analyzed on a BD FACSymphony™ at the Flow Cytometry Facility of The Wistar Institute, and the data analyzed with FlowJo v10.8 software (BD Life Sciences).

| Elispot
Mice (WT and IRE1C148S) were immunized i.p. with NP-LPS (50 μg) or received saline as control, and spleens were collected 3 days later.Antibody secretion was assayed with Elispot plates (Millipore MSIPN4W) coated with NP 26 -BSA as described in Ref. [19,20].In Brief, after blocking the plates with complete culture medium (RPMI 1640 + 10% FBS) cells were applied in a serial dilution and incubated overnight at 37°C.Following washes (PBS/0.1% Tween 20), biotinylated anti-IgM antibodies were applied and detected using ExtrAvidin-alkaline phosphatase (Sigma) followed by BCIP/NBT reagent (Sigma) for visualization of the spots.Quantification of the spots (number and size) was done using ImmunoSpot software (Cellular Technology).

| Statistical analysis
Statistical analyses were performed with GraphPad Prism 6 software.For the comparison of two groups, the twotailed t-test was used.For more than 2 groups, one-way ANOVA followed by Fischer's post hoc test for multiple comparison was used.For the behavior, a two-way ANOVA was performed with the result considered significant when p < .05.

| Generation and characterization of a novel IRE1C148S variant mouse with activity-enhancing mutation
The studies described here were designed to determine whether augmenting the physiological ER stress response affects an autoimmune demyelinating disease.To this end, IRE1C148S mutant mouse was generated using CRISPR technology to introduce a point mutation in the IRE1α gene, replacing the highly conserved cysteine 148 (murine position-C150) with a serine residue (Figure 1A).The cysteine 148 is involved in the activation of IRE1α and is also the target of protein disulfide isomerase family A member 6 (PDIA6), an oxidoreductase that attenuates IRE1α activity. 14Because of the mutation, IRE1α activity is prolonged in vitro, in 293T cells, leading to more robust splicing of XBP1, an immediate substrate of IRE1. 21In vivo as well, the level of XBP1s is increased significantly in the spinal cord of mutant mice, as shown by quantitative RT-PCR (Figure 1B), with no change in the levels of IRE1α protein (Figure 1C).Increased XBP1s was also detected in several brain regions (Figure S1).Therefore, these data confirm that in vivo and in cell culture the Cys148-to-Ser mutation increases IRE1α activity.
Phenotypically, the mutant mice were indistinguishable from WT mice.Since IRE1α activity increases during development both in the early and the active myelination phases 22 and EAE targets myelin, we assessed the spinal cord of IRE1 mutant mice for possible differences in myelin compared to WT mice that could confound our results.Gross morphology of the spinal cord appeared normal, and we found no significant changes in white and gray matter between mutant and WT mice, either histologically or by quantifying the major cytosolic component of the myelin sheath, myelin basic protein (MBP) (Figure 1D-H).We further quantified the number of oligodendrocyte precursor cells (OPCs) (Figure 1I) and oligodendrocytes (Figure 1J) in the spinal cord and found no significant differences.

| The IRE1C148S mutation improves clinical outcome in mice following EAE
Since no significant difference was observed between WT and IRE1C148S mice under normal physiological conditions, we sought to determine whether a difference becomes evident under conditions that trigger IRE1α activation, such as in the course of the autoimmune demyelinating disease model EAE.A clear phenotype was found when EAE was induced by injection of MOG 35-55 peptide in complete Freund's adjuvant (but omitting the pertussis toxin as previously described 15 ).As shown in Figure 2A-F, the day of onset, defined as the first day when mice have a clinical score of 2 for two consecutive days, 23 occurred significantly later for the IRE1C148S mutant mice than for control mice.Although differences between the groups in either the day of peak disease (Figure 2B) or in the maximum clinical score at peak disease (Figure 2C) did not reach statistical significance, suggesting that IRE1C148S mice do develop EAE similarly to control mice, the number of days IRE1C148S animals stayed with a clinical score of 2 or above was significantly reduced compared to WT animals (8.3 days vs. 12.4 days, respectively, Figure 2D).Indeed, the extent of the disease was significantly reduced in IRE1C148S, as shown by the area under the curve (Figure 2E), and the cumulative disease index (35 vs. 25, p < .01, Figure 2F).Furthermore, unlike WT mice, IRE1C148S mutants showed a significant improvement in the clinical score by the end of the experiment (Day 31) in comparison with the peak disease (Figure 2A).The IRE1C148S mice tended to lose less weight than control mice during the acute phase of the disease between Days 18 and 20 (Figure 2G).Thus, according to several clinical criteria, the mutant IRE1C148S had a beneficial effect against EAE.Since pain is recognized as an important symptom of the disease, 24,25 we also assessed whether enhanced IRE1 activity would influence EAE-induced pain.However, mechanical allodynia assessment using Von Frey testing showed that both WT and IRE1C148S mutant mice developed pain which persisted for the duration of the study, with no statistical significance between groups (Figure 2H).

| Enhanced IRE1 activity reduces myelin pathology and neuronal damage
To begin investigating changes that would help explain improvements in functional recovery, we analyzed myelin pathology and neuronal damage.To this end, we utilized EM to analyze naive and diseased spinal cord tissue.
G-ratio analysis was used to investigate myelin thickness of axons between 0.5 and 2 μm in diameter in diseased WT and C148S mice relative to naive controls (Figure 3A,B).As expected, inducing EAE in WT mice resulted in some decrease in G-ratio suggesting myelin damage (Figure 3A).In contrast, such decrease was not observed in IRE1C148S mice (Figure 3B).Similar results could also be observed by staining spinal cord lumbar sections from WT and IRE1C148S with the myelin marker MBP, with noticeable areas of demyelination in WT but not mutant mice (Figure 3C).Further, we observed reduced axonal degeneration in diseased IRE1C148S mice relative to diseased WT mice (Figure 3D).Western blot analysis demonstrated a significant increase in 2′, 3′-cyclic nucleotide 3′-phosphodiesterase (CNPase), an abundant protein in oligodendrocytes (Figure 3E).Although we did not detect large differences in the neuronal marker NF200, the decrease in diseased WT animals was prevented in the IRE1 mutant animals (Figure 3F).

| Increased myelin production but no difference in proliferation or differentiation of IRE1C148S mutant oligodendrocyte precursor cells
Since EAE is a demyelinating disease, and we found increased CNPase expression in IRE1C148S EAE compared to WT mice (Figure 3E), we asked whether oligodendrocytes from IRE1C148S mutant mice were more efficient at making myelin proteins, in vitro.First, we compared the proliferative capabilities of both WT and IRE1C148S oligodendrocyte precursor cells (OPC) using EdU incorporation assay, and found no significant difference in the percentage of EdU + cells, as shown in Figure 4A,B.Next, we asked whether mutant affected OPC maturation toward differentiated oligodendrocytes.We cultured OPCs from WT and IRE1C148S, differentiated them for 5 days with thyroid hormone (T3), and then evaluated the number of mature oligodendrocytes double labeled for CC1 and MBP.
Neither the percentage of mature CC1 + oligodendrocytes (Figure 4C) or CC1 + MBP + myelinating oligodendrocytes (Figure 4D) exhibited a significant difference in IRE1C148S compared to WT cells, suggesting that the more active IRE1 does not affect the proliferation or differentiation of oligodendrocytes.Finally, we tified the production of the myelin sheath by integrating the intensity of MBP staining over each sheath area and found an increase in myelin production in the IRE1C148S mutant oligodendrocytes (Figure 4E).These results suggest that mutant oligodendrocytes produce more myelin than WT cells.Using two different inhibitors of IRE1α, A106 and 4μ8C, we confirmed that while OPCs still differentiate into mature oligodendrocytes (CC1 + -cells), the production of myelin in culture is inhibited (Figure 4F), suggesting a direct role for IRE1α in MBP expression and myelin production.

| Enhanced IRE1α activity reduces microgliosis and inflammatory gene expression.
Neuroinflammation is a significant mediator of EAE disease pathology.Therefore, we investigated changes in spinal cord neuroinflammation in diseased WT and IRE1C148S mice.Astrocytes and microglia are two cell types that are activated under CNS pathologic conditions, with GFAP and IBA1 as their markers of reactivity, respectively.GFAP expression was upregulated in both WT and IRE1C148S EAE mice compared to their respective control group (Figure S2).In contrast, reduced IBA1 upregulation was observed in diseased IRE1C148S mice (Figure 5A), suggesting reduced activation of microglial cells in response to the disease.We next determined the levels of the proinflammatory molecules CCL2, TNF, and IL1β in the spinal cords of 31d EAE mice.Gene expression of all three cytokines was reduced in diseased IRE1C148 mice in comparison with diseased WT mice (Figure 5C-E).We also determined the levels of triggering receptor expressed on myeloid cells (TREM) 2, a receptor involved in phagocytosis, which were elevated in both diseased WT and IRE1C148S mice (Figure 5F).Based on these data, we speculate that sustained IRE1 activity promotes reparative microglia resulting in attenuated neuroinflammation and enhanced phagocytosis.

| Enhanced IRE1 activity in plasma cells does not increase antibody production
Apart from myelination, another example of dedicated high-volume secretion is production of antibodies by activation B cells.Therefore, we examined the antibody responses of IRE1C148S mice in vitro and in vivo.First, the mutation did not affect the development of a normal immune system, resulting in the expected subsets of splenic leukocytes (Figure S3).Second, we asked whether differentiation of plasma cells, or their ability to secrete antibodies, were affected by the IRE1C148S mutation.Mice were immunized with NP-LPS and the appearance of plasma cells (B220 + CD138 + ) in spleens 3 days post-injection was assessed by flow cytometric analysis.The absolute numbers of live plasma cells per organ and the percentages of plasma cells were slightly reduced in the IRE1C148S compared to WT (Figure 6A-C).Purified splenic B cells were then stimulated in vitro with LPS plus IL5 for up to 72 h.This treatment resulted in differentiation of B cells into plasma cells, as monitored by the percentage of CD138 + cells, but with no difference in the time course of differentiation whether the B cells carried WT or C148S IRE1 (Figure 6D). 19nti-NP antibody production by plasma cells was compared by ELISPOT assays, and both the frequency of antigen-specific cells (Figure 6E) and the appearance of their plaques (Figure 6F,G) were similar for both genotypes.Therefore, contrary to our expectation, hyperactive IRE1α does not enhance the differentiation of antibody-producing cells nor does it increase antibody secretion in vitro or in vivo.

| DISCUSSION
There is a considerable body of literature on the effects of decreasing or ablating IRE1α expression on the central and peripheral nervous system. 26On the contrary, there is very little information about the effects of upregulating IRE1α expression or activity, and this work describes a genetic approach to filling this information gap.We took advantage of our previous work in cell lines and primary cells in culture, 14 and characterized a transgenic mouse model where an overactive IRE1α gene replaces the normal gene.The data presented in this manuscript show that in two secretory cell types-myelin producing and antibody-producing cells, increased IRE1α activity has different outcomes.
Our data show a beneficial effect of enhancing the activity of IRE1α on the progression of EAE, a mouse model of MS.Both the severity of motor defects and their duration are reduced.The most pronounced improvement is at the end of the time course (31d post-immunization), when disease progression seems to reverse consistently in IRE1C148S animals.
Previous reports have demonstrated that IRE1α signaling plays an important role in development of the immune system and in immunological response to infection and disease. 27In particular, targeting the IRE1α gene 28 and its direct substrate XBP1 29 have demonstrated that this signaling pathway is crucial for the development function of antibody-producing cells. 30Moreover, in to the XBP1 splicing activity of IRE1α, its RIDD activity is also involved in the differentiation of antibody-secreting cells, 31 suggesting that the quality of IRE1α output changes at various stages of differentiation. 32Because the IRE1α response is not constant, we asked if increased IRE1 activation further enhances either antibody production or plasma cell cellularity, using a genetic approach. 14,31Surprisingly, our results demonstrate that there is no significant enhancement, indicating that the output from activation of wild-type IRE1α in the B-cell compartment is already optimal.dynamics of IRE1α activation have also been determined in models of inflammatory arthritis and EAE, though much less extensively.It was determined that deleting IRE1α in myeloid cells ameliorated chronic inflammation and disease pathology. 33Conversely, ectopic overexpression of spliced XBP1 with a viral vector was beneficial in a model of EAE/optic neuritis, when coupled with inhibition of PERK pathway, 13 suggesting that activation of the IRE1α branch of UPR is protective but insufficient in this disease model.Therefore, we used the mutant IRE1C148S to determine how increased amplitude and duration of IRE1α activity affect EAE.In contrast to the effects on antibody secretion, we found that prolonging IRE1α activation improves functional outcome and reduces neuroinflammation and myelin damage in the EAE model.These results suggest a new interpretation to reports that luteolin or other flavonoids are therapeutic in models of EAE, spinal and brain injury [34][35][36][37] : The positive therapeutic effects may be due to the ability of flavonoids to augment IRE1α activity 38 which then impact many other proteins.9][40] The increased splicing of XBP1 in IRE1C148S and in luteolin-treated cells may lead to transcriptional upregulation of target genes that are only moderately upregulated with the normal level of XBP1s in wild-type cells.
IRE1α is unusual in that it possesses several enzymatic activities, each with outputs to different downstream pathways and distinct consequences.While we have only tested the XBP1 splicing activity of IRE1α, it is quite likely that other activities of IRE1α-higher RIDD activity and/ or recruitment of JNK-are also affected by the C148S mutation.2][43] The marked differences in the effects on B lineage cells and CNS tissue could thus be attributed to the complexity of IRE1α signaling and to its differential regulation in various cell types, including immune cells. 32,44Both deletion and robust pharmacological activation of IRE1α evidently impacts several critical signaling pathways including targets of XBP1s, of RIDD, 45 JNK activation 11 and the mTOR pathway, 31,46 but the activation of these signaling pathways is likely different in distinct cell types.
The preservation of CNPase levels, along with decreased axonal degeneration and improved G-ratios in EAE animals carrying IRE1C148S mutation, indicates improved myelin integrity.Interestingly, as suggested by the findings of Unlu et al., CNPase itself could modulate the outcome of IRE1 activation. 47They demonstrated that CNPase is a component of the machinery that completes splicing of XBP1.Together with the RtcA gene product, CNPase acts to resolve the RNA end with 2′3′ cyclic phosphate, so that it can be ligated to 5′-OH RNA end by the ligase RtcB.In this manner, CNPase regulates the availability of spliced XBP1.Furthermore, the fine-tuning of XBP1 splicing by CNP and RtcA significantly alters the expression levels of direct targets of XBP1. 47This may explain the involvement of CNPase in disease progression, as described in this work.
Microglial cells, also known as the resident macrophages of the brain, constantly monitor the CNS environment and respond to its perturbations.When they become inflammatory, they change their gene expression profile, including the upregulation of IBA1 48 and of proinflammatory cytokines.We observed that a significant increase in IBA1 and in levels of several cytokines in WT EAE mice was markedly reversed in IREC148S EAE mice.
While IL1β gene expression was significantly reduced in IREC148S EAE mice compared to WT, the levels were still significantly higher than IREC148S naive control mice and may be the reason why although there is improvement in motor function, mice still experience pain.It is also possible that increased IRE1α activity in peripheral leukocytes could lead to higher levels of prostaglandins and related pain, through XBP1s regulation of the prostaglandin's synthase Ptgs2. 49The data also suggest that the pain and locomotor aspects of the disease are independent.Interestingly, in a EAE model where microglia are made more inflammatory at peak EAE by ablation of TNFR2, there is downregulation of many UPR genes, including XBP1 (Figure S4, analysis of GEO set: GSE78082 from Ref. [50]), along with an upregulation of proinflammatory cytokines and a worse outcome for the mice.Our data show that the EAE IRE1C148S mutant mice with increased IRE1α activity and XBP1s levels have lower levels of inflammation, showing that IRE1C148S microglia are less inflammatory than WT microglia while still retaining their phagocytic activity.Our mouse model has altered IRE1α activity in all cells and tissues, raising the question of which cell(s) is mostly accountable for the improved phenotype.Short of the definitive experiment of altering IRE1α in cell-specific fashion, we currently must rely on the indirect evidence to suggest that the most relevant cell type for IRE1 impact on EAE is microglia.While we see changes in neurons, oligodendrocytes, and microglia, the decrease in proinflammatory cytokines and IBA1 levels in IREC148S EAE mice suggests a direct, cell-autonomous effect in the latter, whereas the improvements in neurons and oligodendrocytes may be either direct or secondary to decreased inflammation.Furthermore, the difference between cells that are affected by the hyperactive IRE1 and cells that are not, like plasma cells, is unexpected and fascinating.Microglia are not "professional" secretory cells like oligodendrocytes or plasma cells, which are programmed for high-rate production of myelin or immunoglobulin, respectively.Perhaps, they are not programmed to increase the secretory capacity by orders of magnitude, and therefore, the impact of the overactive IRE1α is more pronounced in them.
At the molecular level, a likely explanation for cell type different effects maybe cell-specific interactome of IRE1α and unique XBP1s-regulated gene set expression. 51This is becoming more and more plausible, as there is an expanding number of IRE1α interactors that are being defined, such as the cytoskeleton-associated filamin A, 52 TXNIP, 53 and the TRAF6-mediated ubiquitination machinery. 54he involvement of IRE1α in the inflammatory process in MS is not entirely unexpected.Prior studies in macrophages already showed that proinflammatory cytokine production is regulated 11,55 by IRE1α activation as a consequence of TLR activation. 33,56In that system, the activation of IRE1α is accompanied by activation of the TRAF2-ASK axis. 57aken together, the data suggest that activation of IRE1α signaling above and beyond the usual level is beneficial in this mouse model of MS, and that specific activators of IRE1α activity may be useful tools in the treatment of MS.

F I G U R E 1
Spinal cord from naive IRE1C148S variant with activity-enhancing mutation is not significantly different from WT mice.(A) Schematic diagram of IRE1α protein and its main domains.The N-terminal domain resides in the lumen of the ER and serves as the stress sensing device.The four cysteine residues in this domain are indicated, and the sequence encoding amino acids around Cys148 is listed below the diagram.TM, the transmembrane region that spans the ER membrane and is responsive to lipid stress.58(B) Quantitation of XBP1s transcripts in spinal cords of WT and IRE1C148S mice, using q-PCR.(C) Quantification of IRE1α protein levels in the spinal cords of WT and IRE1C148S mice by Western blot.n = 5/group.(D) Representative thoracic spinal cord sections from WT and IRE1C148S mice immunostained for myelin basic protein (MBP).(E) Spinal cord area quantification in WT and IRE1C148S mice.n = 3/group.(F, G) Quantification of spinal cord areas: White matter (WM) area (E) and Gray matter (GM) area (F), n = 3/group.(H) Western blot analysis of spinal cord extracts from WT and IRE1C148S for MBP.n = 5/group.(I) Quantification of the number of oligodendrocyte precursor cells (PDGFRα+NG2+-cells) per mm 2 in the spinal cord of WT and IRE1C148S.n = 3/group.(J) Quantification of the number of oligodendrocytes per mm 2 (CC1 + -cells).n = 3/group.

F
I G U R E 2 IRE1C148S rescues clinical phenotypes in the EAE model.(A) Mice were MOG-immunized on Day 0 and Day 7 and monitored daily for motor clinical score until Day 31 post-immunization.n = 20 WT, n = 18 IRE1C148S.(B) WT and IRE1C148S mutant mice day of peak disease.The graph represents the mean ± stdev of n = 20 WT, n = 18 IRE1C148S.(C) The highest clinical score reached by each WT and IRE1C148S mutant mice.The graph represents the mean ± stdev of n = 20 WT, n = 18 IRE1C148S.(D) The number of days that each WT and IRE1C148S mutant mice stayed with a clinical score ≥ 2. The graph represents the mean ± stdev of n = 20 WT, n = 18 IRE1C148S.(E) IRE1C148S mutant mice show a significant reduction in disease severity as measured by the area under the curve (AUC) compared to WT mice.n = 20 WT, n = 18 IRE1C148S.(F) Table comparing different parameters of EAE between WT and IRE1α mutant mice.The day of onset is defined as the first day when a mouse has a clinical score of 2 for two consecutive days.The cumulative disease index (CDI) was calculated as the sum of the clinical scores between d7 and d31.(G) Body weight (BW) was recorded daily after the second immunization and plotted as a percentage of initial BW.The graph shows the average ± SEM of n = 20 WT and 18 IRE1C148S with motor deficits.(H) Mechanical allodynia in WT and IRE1C148S mice following EAE induction were determined using the Von Frey method.n = 15 WT and n = 10 IRE1C148S.

F I G U R E 3
Reduction in myelin damage and reduced neuropathology in hyperactive IRE1C148S mice.(A) G-ratio calculated in the lumbar spinal cord from WT mice naive control (n = 3) and 31d post-EAE (n = 5).Bar graphs represent the mean ± SEM.Multiple t-tests analysis, *p < .05,**p < .01. (B) G-ratio calculated in the lumbar spinal cord from IRE1C148S mice naive control (n = 3) and 31d post-EAE (n = 4).Bar graphs represent the mean ± SEM. (C) Micrographs of spinal cord from WT and IRE1C148S EAE mice (31d postimmunization) immunostained for myelin basic protein.Hoechst is used as a nuclei counterstain.Scale bar: 50 μm.White arrows indicate areas of demyelination.(D) Graph shows the percentage of degenerated axons in WT (n = 5) and IRE1C148S (n = 4) mice 31d post-EAE.(E, F) Western blot analysis of lumbar spinal cord from WT and IRE1C148S naive and 31 days post-EAE for myelin protein CNPase (E) and neuronal marker NF200 (F).Signal for each protein was normalized to the total protein amount quantified by Ponceau red staining of the membrane.Bar graphs represent the mean ± SEM of n = 3/group.One-way ANOVA and Fisher's test, *p < .05.

F I G U R E 4
No difference in proliferation or differentiation rate of IRE1C148S mutant oligodendrocyte precursor cells but increase in myelin basic protein (MBP).(A) Proliferation of oligodendrocytes precursor cell (OPC) from WT and IRE1C148S was determined by EdU incorporation.Representative images are shown.(B) Quantification of the percentage of EdU + -cells in OPC cultures derived from WT and IRE1C148S.(C, D) Quantification of the percentage of mature oligodendrocytes (CC1 + -cells) between WT and IRE1C148S mutant (C), and the percentage of mature myelinating oligodendrocytes (CC1 + MBP + , D). Bar graph represents the means ± SEM of n = 3/group.(E) Quantification of the myelin sheath via MBP immunostaining.MBP integrated density was calculated for MBP + CC1 + oligodendrocytes from WT and IRE1C148S in triplicate.A paired T-test was used.*p < .05. (F) Effects of A106 and 4-8°C, two inhibitors of IRE1α on OPC differentiation into myelinating (CC1 + MBP + ) oligodendrocytes.White arrows indicate CC1 + oligodendrocytes.

F I G U R E 5
Reduction in proinflammatory cytokine levels but preservation of phagocytosis gene expression.(A) Western blot analysis of lumbar spinal cord from WT and IRE1C148S naive and 31 days post-EAE for IBA1, a marker of reactive microglia.Bar graph represents the mean ± SEM of n = 3/group.*p < .05,***p < .001.(B) Correlation analysis between IBA1 protein levels in the spinal cord and the clinical score at d31 post-EAE.(C-F) mRNA levels for proinflammatory cytokines CCL2 (C), TNF (D), IL1β (E), and TREM2, a marker of microglial phagocytosis (F), were determined in the lumbar spinal cord of WT and IRE1C148S naive and 31d post EAE mice by quantitative real-time PCR.Bar graphs represent the mean ± SEM of n = 3-5/group.*p < .05,**p < .01,***p < .001,****p < .0001.

F I G U R E 6
Hyperactive IRE1α does not increase production of antibodies by B cell.(A) Flow cytometric analysis of the differentiation of B cells into plasma cells.Representative contour plots of splenocytes from WT and IRE1C148S mice, 72 h after I.P. injection of NP-LPS (or Saline as a control).Plasma cells were gated as Dump [CD4, CD8a, Ter-119, F4/80]-IgD-, B220+ /− , CD138+ cells.The top plots show steady-state controls, and the bottom plots are the immunized mice.(B, C) Quantification of antigen-specific plasma cells (PC) in the spleens of WT and IRE1C148S mice (n = 3 mice per group) (B) Histogram of total number of NP-LPS-induced plasma cells (PC) in each spleen, (C) The percentage of PC in each NP-LPS immunized spleen.(D) Time course of in vitro differentiation of WT and IRE1C148S B cells into plasma cells following LPS and IL5 stimulation over 72 h in culture.(E) Anti-NP-LPS antibodies production assessed using ELISpot.The number of antibody-secreting cells (ASC) from each spleen is shown as spots per million cells (n = 3 mice per group).(F) The amount of antibodies secreted from PC of each genotype is compared based of the calculated average spot size.(G) Representative ELISpots for WT and IRE1C148S.
Our data suggest that sustained increase in IRE1α activity can be beneficial in vivo, and that this protection is cell type and context dependent.Considering the overwhelming but conflicting evidence for the role of ER stress in neurological diseases, a better understanding of the function of ER stress sensors in physiological contexts is clearly needed.
2 | MATERIALS AND METHODS2.1 | MiceIRE1α C148S mice were generated using the CRISPR technology in the Transgenic and Chimeric Mouse the clinical improvement in IRE1C148S animals.