Myelin Oligodendrocyte Glycoprotein 35-55

miR-223 promotes regenerative myeloid cell phenotype and function in the demyelinated central nervous system

Abstract
In the injured central nervous system, myeloid cells, including macrophages and microglia, are key contributors to both myelin injury and repair. This immense plasticity emphasizes the need to further understand the precise molecular mechanisms that contribute to the dynamic regulation of myeloid cell polarization and function. Herein, we demonstrate that miR-223 is upregulated in multiple sclerosis (MS) patient monocytes and the alternatively- activated and tissue-regenerating M2-polarized human macrophages and microglia. Using miR-223 knock-out mice, we observed that miR-223 is dispensable for maximal pro- inflammatory responses, but is required for efficient M2-associated phenotype and func- tion, including phagocytosis. Using the lysolecithin animal model, we further demonstrate that miR-223 is required to efficiently clear myelin debris and promote remyelination. These results suggest miR-223 constrains neuroinflammation while also promoting repair, a finding of important pathophysiological relevance to MS as well as other neurodegenera- tive diseases.

1 | INTRODUCTION
Multiple sclerosis (MS) is a chronic neuroinflammatory disease characterized by immune-mediated destruction of myelin within the central nervous system (CNS) (Reich, Lucchinetti, & Calabresi, 2018). In both MS and its relevant animal models, including experi- mental autoimmune encephalomyelitis (EAE) and lysolecithin- induced demyelination, brain-resident microglia and peripheral blood-derived macrophages are critical mediators of demyelin- ation, remyelination, and brain repair. In the inflamed CNS, these specialized myeloid cells demonstrate considerable phenotypic and functional plasticity, which can either potentiate inflammation or promote repair depending on their activation status and cross-talk with other neural and immune cells (Giles et al., 2017; Lloyd & Miron, 2016). In particular, polarization of macrophages and micro- glia to an alternative and repair-promoting (M2) phenotype is associated with CNS repair and remyelination (Miron et al., 2013). Thus, factors responsible for fostering myeloid cell phenotype and function are of critical relevance to MS microRNAs (miRs) are small-noncoding RNA molecules that post-transcriptionally regulate gene expression by binding mRNA 30 untranslated regions. microRNAs play important roles in many cel- lular processes, including inflammation and myeloid cell phenotype and function. Currently, the functional roles of microRNAs in MS and neuroinflammation are not fully understood. In particular, one microRNA that has been implicated in MS pathology is microRNA- 223 (miR-223). miR-223 has diverse roles in modulating inflamma- tory processes (Haneklaus, Gerlic, O’Neill, & Masters, 2013) and its expression is upregulated in both MS patient peripheral blood mononuclear cells (PBMCs) (Keller et al., 2009) and early active, demyelinating MS lesions (Junker et al., 2009). Importantly, these reports have not identified precise cell types in which miR-223 expression is altered. miR-223 is predominantly expressed in cells of hematopoietic lineage (Chen, Li, Lodish, & Bartel , 2004), and is further enriched in myeloid cells (Johnnidis et al., 2008).

miR-223 has been reported to regulate both pro-inflammatory (M1) and anti-inflammatory (M2) macrophage polarization (Ying et al., 2015; Zhuang et al., 2012), and granulocyte function. Initial reports characterizing mice deficient for miR-223 (miR-223 KO) uncovered exaggerated inflammatory phenotypes, particularly innate immune hyper-activation during aging or following endotoxin challenge (Johnnidis et al., 2008). Subsequent reports utilizing miR-223 KO mice have demonstrated that expression of miR-223 is protective in several inflammatory animal models including colitis (Neudecker et al., 2017b; Zhou et al., 2015), lung injury (Neudecker et al., 2017a), liver injury (Li et al., 2017) and obesity (Zhuang et al., 2012). Furthermore, within the CNS, miR-223 overexpression is protective in models of ischemic stroke (Harraz, Eacker, Wang, Dawson, & Dawson, 2012) and intrace- rebral hemorrhage (Yang, Zhong, Xian, & Yuan ,2015). In contrast, recent reports have also demonstrated that miR-223 is pathogenic in EAE (Cantoni et al., 2017; Ifergan, Chen, Zhang, & Miller, 2016) with mice lacking miR-223 displaying impaired pathogenic T cell activation. Although this may suggest that miR-223 expression is detrimental in MS, the EAE model does not fully reflect the pathological features of MS and is an inefficient model to study repair and remyelination (Franklin & Ffrench-Constant, 2017).Taken together, these reports suggest a Janus role for miR-223 in controlling neuroinflammation that is currently unresolved. Further- more, it is currently unknown how miR-223 contributes to repair and remyelination in the CNS following demyelination.In the current study, we have sought to identify the role of miR- 223 in regulating MS-relevant neuroinflammation. We demonstrate that miR-223 expression is elevated in MS patient monocytes and M2 myeloid cells. miR-223 deficiency delays EAE onset, but is dispensable for severity and M1 polarization. Furthermore, miR-223 is required for efficient M2 polarization and phagocytosis in myeloid cells, and mice lacking miR-223 display impaired CNS remyelination. These results suggest a complex, but overall beneficial role for miR-223 in regulating CNS inflammation and remyelination.

2 | MATERIALS AND METHODS
2.1 | Animals
All animal experiments were approved by the Memorial University Animal Care Committee in accordance with Canadian Counsel on Animal Care guidelines. All animals were kept in 12 hr light dark- cycles with access to food and water ad libitum and housed at the Memorial University Faculty of Medicine animal care facility. miR-223−/− and miR-223−/y knockout mice (B6.Cg-PtprcaMir223tm1Fcam/J) were maintained on a C57BL/6 background and purchased from Jackson Laboratories. Wild-type control CD45.1+ mice (B6.SJL-Ptprca Pepcb/BoyJ) were purchased from Jackson Laboratories, as the miR-223 knockout mice were initially backcrossed and maintained on this strain (Johnnidis et al., 2008). Genotyping was performed using the common 50-TTCTGCTATTCTGGCTGCAA-30; wild-type 50-CAGTGTCACGCTCCGTGTAT-30; and knockout 50-CTTCCTCG-
TGCTTTACGGTATCG-30 primers purchased from Integrated DNA Technologies. miR-223 knockout mice lacked expression of mature miR-223 as demonstrated by qPCR (Supplementary Figure 1A).

2.2 | EAE Induction and Scoring
Experimental autoimmune encephalomyelitis was induced in 6- to 8– week-old littermate or age- and sex-matched animals with similar clin- ical scores obtained. Animals were immunized with 200 μg of subcu- taneous MOG35–55 (Genscript) in a 200-μL emulsion of Complete
Freund’s Adjuvant (CFA; Difco) bilaterally at the base of the tail. On days 0 and 2, pertussis toxin (400 ng, Enzo) was injected intraperito- neally (i.p.). The clinical scoring system used was as follows: 0, normal; 1, limp tail; 2, walking deficits; 3, unilateral hindlimb paralysis; 4, bilat- eral hindlimb paralysis; 5, moribund (required sacrifice). When disease severity was between the defined scores, intermediate scores of 0.5 were given. All mice were scored by a blinded investigator.

2.3 | Lysolecithin-Induced Focal Demyelination
Demyelinating lesions were induced in the corpus callosum of 12- to 24-week-old littermate male mice. Under isoflourane anesthesia, mice
were injected with 2 μL of 1% lysolecithin (Sigma, Oakville, Canada) dissolved in PBS at stereotaxic coordinates 1.2 mm posterior, 0.5 mm lateral, and 1.4 mm deep to bregma using a Hamilton syringe and a nanopump (Harvard Apparatus). Following surgery, mice received 5 mg/kg Metacam® and were returned to their home cage for 7 or 14 days. To quantify demyelination, coronal sections were stained with FluoroMyelin™ (outlined below). Lesion size was measured as demyelinated area at the level of the anterior hippocampus in all ani- mals by a blinded investigator. Demyelinated area was quantified using ImageJ software.

2.4 | In Vivo LPS Challenge
Littermate 6- to 8-week-old male miR-223 KO mice and WT control were injected I.P. with 5 mg/kg of LPS (Serotype O55:B5). Blood was collected 2 hr post-injection via cardiac puncture using heparinized syringes. Plasma was than isolated for ELISAs.

2.5 | Patient Samples
All experiments involving human participation were approved by the Newfoundland Health Research Ethics Board. MS patients were recruited through the Health Research Innovation Team in Multiple Sclerosis (HITMS) at Memorial University of Newfound- land, St. John’s, NL, Canada. Venous blood was drawn from relapsing–remitting MS (RRMS) patients and healthy controls with informed consent. PBMCs were isolated following ficoll-density gradient centrifugation, and CD14+ monocytes were subsequently isolated to ~95% to 98% purity using anti-CD14 magnetic beads (Miltenyi). For microRNA and mRNA expression assays, cells were
immediately lysed and stored in QiaZOL® reagent at −80 ◦C.

2.6 | Cell Culture
Human monocyte-derived macrophages (MDMs) were cultured from iso- lated CD14+ monocytes. Monocytes were cultured at 5 × 105 cells/ml in RPMI media containing 10% fetal bovine serum (FBS), penicillin/paraformaldehyde (PFA). Brains and spinal cords were dissected and post-fixed in 4% PFA for 24 hr at 4 ◦C. Tissues were paraffin-embedded and sectioned at 6 μm for H&E or eriochrome cyanine staining. Lysoleci-
thin brains were cryoprotected in 30% sucrose in PBS at 4 ◦C. Tissue streptomycin (Gibco ; 1×), and GlutaMAX™ (Gibco ; 1×) supplemen- ted with macrophage colony stimulating factor (M-CSF; 25 ng/ml; Peprotech). Macrophages were used in experiments after 5 to 6 days of differentiation. Human adult microglia were isolated and cultured from surgically resected CNS tissue as previously described (Durafourt, Moore, Blain, & Antel, 2013). All human MDMs and microglia were derived from normal populations and not patients with MS. Mouse bone marrow-derived macrophages (BMDMs) were obtained from bone mar- row of adult mice and cultured in DMEM containing 10% FBS, penicillin/ streptomycin, and glutamine supplemented with M-CSF (10 ng/ml; Peprotech) as previously described (Ying, Cheruku, Bazer, Safe, & Zhou, 2013). Mouse microglia were prepared from mixed glial cultures of P0-2 cortices cultured in DMEM media containing 10% FBS, penicillin/strep- tomycin (Gibco®; 1×), and GlutaMAX™ (Gibco®; 1×). Microglia (99% purity) were isolated by mild-trypsinization (Saura, Tusell, & Serratosa, 2003) and maintained in astrocyte-conditioned media for 24 to 48 hr prior to experimentation. Cell activation and polarization was performed as previously described (Durafourt et al., 2012; Moore et al., 2013). For LPS activation assays, cells were stimulated with LPS (100 ng/ml; sero- type 0127:B8) for 6 hr. For polarization assays, M1 cells were activated with IFNγ (20 ng/ml; Peprotech) for 1 hr, followed by LPS (100 ng/ml) for 48 hr. M2 cells were activated with IL-4 (20 ng/ml; Peprotech) and IL-13 (20 ng/ml; Peprotech) for 48 hr. Resting, untreated cells were used as “M0” controls. Cells and supernatants were collected as described below. Murine neurospheres were derived from P0 to P2 mouse pups as previously described (Moore et al., 2011). Briefly, cortical neurospheres were grown in DMEM-F12 containing penicillin/streptomycin, B27 supplement, EGF (20 ng/ml), FGF2 (20 ng/ml), and heparin (2 μg/ml). After 1 week in culture, neurospheres were plated on 12 mm laminin-coated coverslips in wells of a 24-well culture dish and differentiated for 7 days in DMEM containing 1% FBS and N1 supplement.

2.7 | microRNA Transfection
microRNA transfection of cells was performed using Lipofectamine RNAiMAX® (Invitrogen, Burlington, Canada). Lipofectamine reagent was diluted 1:100 in serum-free media, and then incubated with micro- RNA mimics (mirVana™) for 20 to 30 min. microRNA–lipofectamine complexes were then added to cell cultures yielding final concentra- tions of 1, 10, and 30 nM for 48 to 72 hr prior to experimentation. For all microRNA mimic experiments, a non-coding control microRNA (mirVana™) was transfected at a concentration of 30 nM. Transfection efficiency (>90%) was determined by transfection of a Cy3 fluores- cently labeled microRNA, and a 30 nM dose of miR-223 mimics corre- sponded to an approximately 200-fold increase in mature miR-223 levels in mouse BMDMs (Supplementary Figure 1B and C).

2.8 | Immunohistochemistry
Mice were anesthetized by intraperitonal injection of sodium pento- barbital and perfused intracardially with PBS followed by 4% nal sections were cut at 14 to 16 μm. All sections were blocked and permeabilized in blocking solution (PBS containing 10% normal goat serum, 2% horse serum, 0.1% Triton-X) for 1 hr at room temperature (RT). Primary antibodies were diluted in blocking solution and incu- bated with tissue sections overnight at 4 ◦C. Sections were washed 3x in PBS-Tween20 (0.05%) followed by incubation with secondary antibodies for 1 hr at RT. Sections were then washed 3x in PBS- Tween20 and mounted in Fluoromount-G™ (Southern Biotech). For myelin staining, slides were incubated with FluoroMyelin™ Green (1:300 in PBS; Thermo, Mississauga, Canada) for 20 minutes at RT following staining with secondary antibodies. DAPI (1:1000) was used to stain nuclei. Primary antibodies included: Iba1 (1:500; Wako), GFAP (1:1000; Thermo, Mississauga, Canada), MBP (1:200; BioLegend), NG2 (1:500; Millipore), and Arg1 (1:200; SantaCruz). All secondary antibodies were purchased from ThermoFisher and used at 1:500. All images were acquired on a Zeiss AxioObserver.Z1. Cell counts and quantification were performed by a blinded observer.

2.9 | RNA Isolation and qPCR
Cells were lysed in QiaZOL® reagent and stored at −80 ◦C. Total RNA was isolated by RNeasy® column extraction with a DNase treatment step (Qiagen, Germantown, MD). RNA was quantified using a Nanodrop™ One.For gene expression assays, RNA (200 ng) was reverse transcribed using M-MLV reverse transcriptase (Invitrogen, Burlington, Canada). Individual gene expression assays were performed using specific TaqMan® probes and normalized to the endogenous control gene 18S. For microRNA expression assays, microRNA-specific reverse transcription primers were multiplexed and RNA (10 ng) was reverse transcribed following the TaqMan® MicroRNA Reverse Transcription kit protocol. MicroRNA expres- sion was normalized to RNU48 or SNO202 for human or mouse samples respectively. Fold changes were calculated using the ΔΔCt method.

2.10 | Western Blotting
Cell cultures were lysed in 1× Lamelli sample buffer and boiled for 5 min at 95 ◦C. Samples were separated on NuPAGE™ 4–12% Bis-Tris Gels (Invitrogen, Burlington, Canada) and transferred to 0.45 pore Immobilon- P PVDF membranes (Millipore) for 1 hr at 100 V. Membranes were probed with antibodies specific to RASA1 (Clone B4F8; 1:1000) or β-Actin (Clone C-4; 1:1000) followed by HRP-linked anti-mouse IgG
(1:2000; sc-2005). Quantification of band intensity was measured using ImageJ software (NIH). Protein loading was normalized relative to β-actin.

2.11 | Flow Cytometry
For EAE experiments, spinal cords were homogenized from PBS perfused mice and myelin was removed using a 30% to 70% Percoll gradient. Iso- lated mononuclear cells were resuspended in FACS buffer (1% FBS in PBS). For in vitro assays, BMDMs were collected in FACS buffer by
gentle scraping using a cell lifter. The following antibodies were used for analysis CD45 (Clone: A20; Pacific Blue), CD11b (Clone: M1/70, FITC) CD80 (Clone: 16-10A, FITC), CD86 (Clone: GL1, PE), CD206 (Clone:C068C2, PE-Cy7). Cell viability was determined by staining with LIVE/ DEAD™ Fixable Aqua (Thermo, Mississauga, Canada) according to man- ufacturer’s protocol. Following a 30-min incubation, cells were washed and fixed in 1% PFA and acquired using a MoFlo® Astrios™ flow cyt- ometer (Beckman Coulter, Mississauga, Canada) or FACSCalibur™ (BD). Data were analyzed using FlowJo® software.

2.12 | ELISAs
Cell culture supernatants were collected and stored at −80 ◦C. Supernatants were assayed for TNF or IL-6 levels using ELISA kits (BD Biosciences) according to manufacturer’s instructions.

2.13| Phagocytosis Assays
Myelin was isolated as previously described (Norton & Poduslo, 1973). Briefly, adult mouse brains were homogenized, subjected to sucrose density gradient centrifugation and osmotic shocks to sepa- rate myelin from other cellular components. The protein component of myelin was measured using the BCA assay (Sigma, Oakville, Canada). To assess myelin phagocytosis by IHC, cells were treated with 25 μg/ml of myelin for 24 hr and stained for Iba1 and MBP. For live-cell phagocytosis assays, pHrodo® Green-labeled zymosan A bio- particles (Thermo, Mississauga, Canada), or pHrodo® Red-labeled myelin were added to BMDM cultures and incubated in a live-imaging plate reader (Cytation5) at 37 ◦C with 5% CO2. Fluorescence intensity was measured at 5-min intervals over a period of 6 hr.

2.14 | ROS Assays
Bone marrow-derived macrophages were loaded with 20 μM carboxy- H2DCFDA (Thermo, Mississauga, Canada) for 15 min in serum free DMEM. BMDMs were washed and media was replaced with phenol red free medium containing treatment conditions. Relative fluorescence intensity (Ex480/Em520) was determined using a fluorescence plate reader after 3 hr stimulation with LPS (100 ng/ml).

2.15 | Data Analysis
Data analysis was performed using Prism 6 software (GraphPad), and results are presented as the mean SEM. All statistical tests used are described in the figure legends. p < .05 were considered statistically significant (*p < .05, **p < .01, ***p < .001). 3 | RESULTS 3.1 | miR-223 expression is increased in treatment naïve MS patient monocytes and M2-polarized primary human myeloid cells We first sought to measure the expression of miR-223 in MS patient circulating immune cells and performed qPCR for miR-223 in CD14+ monocytes and whole PBMCs isolated from untreated RRMS patients and age- and sex-matched healthy controls (Supplemental Table 1). miR-223 expression was significantly upre- gulated in CD14+ monocytes (Figure 1a), however, no differences were observed in whole PBMCs (Figure 1b) and suggested a monocyte-specific dysregulation of miR-223 in RRMS patients. As miR-223 has previously been reported to influence macrophage polarization in mice, we further assessed miR-223 expression in M1- and M2-polarized primary human macrophages and microglia. Compared with M1-polarized cells, we observed increased miR- 223 expression in M2-polarized human monocyte-derived macro- phages and human adult microglia (Figure 1c,d). 3.2 | miR-223 knockout mice display delayed EAE onset, but similar disease severity Based on our observation that miR-223 expression was increased in MS patient monocytes and M2-polarized myeloid cells, we hypothesized that miR-223 may influence disease course and pathology in EAE, an animal model of MS. In a series of three-independent EAE experiments, our pooled in vivo data demonstrated that miR-223 knockout mice had only a modest delay in EAE onset; no differences in disease severity were observed as demonstrated by both clinical score and percent weight FIGURE 1 Relative expression of miR-223 in MS patient samples and polarized myeloid cells. (a) A significant increase in miR-223 was observed in isolated CD14+ monocytes from untreated RRMS patient (n = 10) in comparison to controls (n = 18). (b) No change was observed in whole PBMCs from MS patients. n = 9/group;(c, d) miR-223 expression was significantly increased in M2 human macrophages and human microglia. For all panels, statistical significance was determined by an unpaired (a, b) or paired (c, d) Student's t-test. Data are expressed as mean SEM change (Figure 2a,b). Histology of EAE spinal cords demonstrated similar immune cell infiltration and demyelination in WT and miR-223 KO mice at 21 days (Figure 2c). Furthermore, flow cytometry of EAE spinal cords at 21 days demonstrated similarities in both peripheral-derived macro- phages and microglia numbers (Figure 2d,e). 3.3 | Myeloid cells derived from miR-223 knockout mice display a similar pro-inflammatory phenotype and function compared with WT-littermates.Given the role of pro-inflammatory and activated myeloid cells in anti- gen presentation and demyelination, we measured the ability of miR-223 KO-derived myeloid cells to express co-stimulatory molecules and pro-inflammatory cytokines in response to IFNγ and LPS stimulation (M1). With the exception of a ~20% reduction in IL-6 release by miR-223 KO BMDMs, the release of TNF and IL-6 from macrophages or microglia did not differ between cells derived from either miR-223 KO mice or WT littermates (Figure 3a). miR-223 KO macrophages demonstrated slightly reduced co-stimulatory molecule (CD80 and CD86) expression at 48 hr following stimulation (Figure 3b). To assess function, no changes in intracellular ROS production were measured either at baseline or following LPS activation (Figure 3c). Furthermore, pro-inflammatory cytokine responses following peripheral LPS injections were similar between both genotypes (Figure 3d). 3.4 | Overexpression of miR-223 reduces human and murine myeloid cell activation To further elucidate a potential role of miR-223 and negate any compen- satory mechanisms in the miR-223 KO mice during pro-inflammatory myeloid cell activation, we overexpressed miR-223 using mature miR-223 mimics in both human and mouse macrophages and microglia prior to activation with LPS. Compared with transfection with a non- coding microRNA control mimic, cells transfected with a miR-223 mimic showed decreased TNF secretion upon LPS challenge (Figure 4a,b). Fur- thermore, transfection of miR-223 mimics significantly reduced protein FIGURE 2 Knockout of miR-223 results in delayed EAE onset but does not influence overall disease severity. (a, b) EAE was induced in miR-223 KO (n = 21) and WT (n = 17) mice, and clinical score and body weight were recorded for 21 days. Statistical significance was determined by repeated-measures two-way ANOVA with Sidak’s multiple comparisons test. (c) Histological sections show similar immune cell infiltration and demyelination at 21 days post-induction. (d, e) flow cytometric analysis of myeloid cell infiltration demonstrates no significant differences in macrophage or microglial cell numbers at 21 days post-induction. n = 4/group; statistical significance determined by unpaired Student’s t-test. Data are expressed as mean SEM. Scale bars are 200 μm [Color figure can be viewed at wileyonlinelibrary.com] FIGURE 3 M1 polarization is unaffected in miR-223 KO cells. (a) Secretion of TNF and IL-6 following LPS (100 ng/ml) stimulation of WT and KO myeloid cells. n = 3 to 5/group; statistical significance was determined by unpaired Student's t-tests. (b) BMDMs were polarized to an M1 phenotype for 48 h with IFNγ (20 ng/ml) and LPS (100 ng/ml) and surface expression of the costimulatory molecules CD80 and CD86 were assessed (histograms, dotted line: Untreated, solid line: M1). n = 3 to 5/group; statistical significance determined by two-way ANOVA with Tukey's multiple comparisons test. (c) ROS production was determined in BMDMs loaded with carboxy-H2DCFDA and stimulated with LPS (100 ng/ml) for 3 h. n = 3/group; statistical significance determined by two-way ANOVA with Tukey’s multiple comparisons test. (d) Plasma cytokine responses were determined in mice 2 hr post-injection of LPS (5 mg/kg; i.p.). n = 7 to 8/group; statistical significance determined by unpaired Student’s t-tests. Data are expressed as mean SEM levels of the miR-223 target RASA1 (Sun et al., 2015) in macrophages (Figure 4c,d and Supplementary Figure 2), particularly at the highest con- centration of miR-223 mimics (30 nM). 3.5 | miR-223 is essential for efficient M2 polarization and phagocytosis We next assessed the ability of miR-223 to influence the alternatively activated, tissue-repairing ‘M2’ phenotype in myeloid cells. Upon stim- ulation with IL-4 and IL-13 for 48 hr, miR-223 KO macrophages dem- onstrated significantly impaired M2 polarization as measured by expression of the mannose receptor, CD206 (Figure 5a,b). Compared with WT-derived cells, further evidence of an impaired M2 phenotype was noted in miR-223 KO cells using qPCR for the common M2 markers Arg1 and Ym1. While levels of Arg1 remained unchanged, Ym1 expression was significantly reduced in both miR-223 KO- derived macrophages and microglia (Figure 5c), confirming reduced M2 polarization. As phagocytosis is a key function of M2 myeloid cells (Durafourt et al., 2012), we further investigated whether miR-223 KO macrophages displayed a deficit in phagocytosis ability. Following exposure to highly purified myelin for 24 hr, fewer miR-223 KO macrophages were MBP+ compared with WT controls (Figure 6a,b). To further investigate phagocytosis, WT and miR-223 KO macro- phages were exposed to pHrodo®-labeled Zymosan A particles or purified myelin. In real-time, phagocytosis was measured by fluores- cence intensity. As pHrodo® is a pH-sensitive dye, increases in rela- tive fluorescent unites (RFUs) are indicative of particle uptake into phagosomes and subsequent phagosomal maturation. Within these phagocytosis assays, miR-223 KO macrophages demonstrated signif- icantly impaired phagocytosis in all phenotypes compared with WT controls (Figure 6c,d). 3.6 | miR-223 knockout mice display impaired CNS Remyelination and myelin debris clearance Given the observed deficit in M2 polarization and phagocytosis, we reasoned that miR-223 KO mice would display impaired remyelination in the context of lysolecithin-induced focal demyelination. We induced demyelinating lesions miR-223 KO mice and WT littermate controls and collected tissue at 7 and 14 DPI. At 7 DPI, KO mice dem- onstrated a trend toward increased lesion size (p = .08) (Figure 7a). At 14 DPI, miR-223 KO mice had a significantly increased lesion sizes FIGURE 4 Overexpression of miR-223 in mouse and human myeloid cells decreases pro-inflammatory responses upon LPS activation. Myeloid cells were transfected with miR-223 mimics (1 to 30 nM) prior to activation with LPS (100 ng/ml). Transfection of miR-223 into human (a) and mouse (b) macrophages and microglia result in reduced TNF secretion following LPS activation (100 ng/ml, 6 h). n = 2 to 3/group (human microglia) or n = 4/group; statistical significance from non-coding (NC) miRNA was determined by repeated-measures one-way ANOVA followed by Dunnett’s multiple comparisons. (c, d) overexpression of miR-223 reduces expression of the miR-223 target RASA1. Data are expressed as mean SEM FIGURE 5 miR-223 is required for efficient M2 myeloid cell polarization. (a) BMDMs were polarized to an M2 phenotype with IL-4 (20 ng/ml) and IL-13 (20 ng/ml), and polarization was assessed by expression of the mannose receptor CD206. n = 3 to 4/group; statistical significance determined by two-way ANOVA followed by Tukey’s multiple comparisons test. (c, d) BMDMs and microglia were polarized to an M2 phenotype, and M2-associated gene expression was determined by qPCR. Myeloid cells lacking miR-223 show reduced Ym1 expression following M2 polarization. n = 3 to 4/group; statistical significance determined by two-way ANOVA followed by Tukey’s multiple comparisons test. Data are expressed as mean SEM [Color figure can be viewed at wileyonlinelibrary.com] FIGURE 6 Myeloid cells lacking miR-223 display reduced phagocytic ability. (a, b) BMDMs from WT and KO mice were treated with purified mouse myelin (25 μg/ml; 24 hr) and the number of myelin-laden macrophages was determined. n = 6/group; statistical significance determined by unpaired Student’s t-test. (c, d) Polarized BMDMs were incubated with pHrodo-labeled Zymosan A (c) or myelin (d) and relative fluorescence (corresponding to phagocytosis) was recorded. Phagocytic rates were calculated as the slope of the linear portion of the curves. n = 3 to 4/group; statistical significance determined by unpaired repeated-measure two-way ANOVAs followed by Tukey’s multiple comparisons test (RFU) or unpaired Student’s t-tests (phagocytic rate). Data are expressed as mean SEM. Scale bars are 50 μm [Color figure can be viewed at wileyonlinelibrary.com](Figure 7b), indicative of impaired remyelination. Analysis of CC1+ cells within lesions demonstrated miR-223 KO mice had a modest, but not statistically significant, decrease in CC1+ cells (Figure 7c). Analysis of OPC differentiation from WT and miR-223 KO neurospheres dem- onstrated no significant differences (Supplemental Figure 3), suggest- ing the remyelination deficit observed in vivo is not oligodendrocyte intrinsic.To further investigate the mechanisms leading to impaired remye- lination in vivo, we assessed Iba1, FluoroMyelin™ and Arg1 immuno- reactivity in WT and miR-223 KO lysolecithin lesions. It was apparent that in the miR-223 KO animals, demyelinated lesions contained numerous myelin debris puncta, often deposited within macrophages and microglia (Figure 8a). Notably this staining pattern was absent in WT mice, suggesting a debris clearance failure in the miR-223 KO mice. Further analysis of the lesions demonstrated that while there were no significant differences in myeloid cell infiltration (Figure 8b), there were increased numbers of myelin-laden cells in the miR-223 KO animals (Figure 8c). Quantification of Arg1+ myeloid cells revealed no significant differences between WT and miR-223 KO animals (Supplemental Figure 4). 4 | DISCUSSION Recently, it has become clear that diverse myeloid cell phenotypes are key contributors to inflammation and repair in MS and its relevant ani- mal models (Rawji, Mishra, & Yong, 2016). However, the precise molecular mechanisms that underlie these phenotypes, particularly in the context of MS, have yet to be fully elucidated. Our data has demon- strated that an immunoregulatory microRNA, miR-223, is dysregulated in MS patient myeloid cells, and contributes to reparative myeloid cell activation and CNS remyelination. Importantly, we have demonstrated that miR-223 is required for efficient anti-inflammatory M2 myeloid cell activation, debris clearance via phagocytosis, and CNS remyelination in vivo. Using samples derived from untreated RRMS patients, we have observed that miR-223 is upregulated in circulating monocytes, but not in whole PBMCs (Figure 1). While a previous report has demon- strated an increase in miR-223 expression in the whole blood of MS patients (including granulocytes, red blood cells, and platelets) (Keller et al., 2009), our results suggest that the change does indeed occur in the monocytes and was not observed in PBMCs due to a dilution effect of RNA from other cell populations. It has previously been reported miR-223 expression is also upregulated in active MS lesions (Junker et al., 2009), although this upregulation is not sustained in chronic lesions. Previous reports have also demonstrated that miR- 223 expression is decreased in the plasma of MS patients (Fenoglio et al., 2013), potentially in the exosomal compartment where miR-223 is highly abundant (Hunter et al., 2008). The upregulation of miR-223 in MS is consistent with observed upregulation of miR-223 in a multi- tude of diseases associated with chronic inflammation (Haneklaus, Gerlic, O'Neill, & Masters, 2013). FIGURE 7 Deletion of miR-223 impairs CNS remyelination. miR-223 KO or WT littermates were injected with lysolecithin into the corpus callosum. (a) At 7 days post-injection, miR-223 KO mice had larger lesions compared with WT controls. n = 4/group; statistical significance determined by unpaired Student’s t-test. (b) Lesion sizes at 14 days post-injection were significantly larger in miR-223 KO mice than WT controls. n = 3 to 5/group; statistical significance determined by unpaired Student’s t-test. (c) CC1+ cell density was modestly reduced in miR-223 KO mice in comparison to WT controls. n = 3 to 5/group; statistical significance determined by unpaired Student's t-test. Data are expressed as mean SEM. Scale bars are 200 or 50 μm [Color figure can be viewed at wileyonlinelibrary.com] limit tissue damage and promote the transition to a regenerative phe- notype. Why this mechanism fails in MS (and other inflammatory dis- eases) is currently unknown. In response to an acute demyelinating insult, peripheral blood- derived macrophages and CNS-resident microglia are recruited to lesion sites and engage in an inflammatory program that is initially pro-inflammatory. However, this is often followed by a conversion to an anti-inflammatory and regenerative phenotype that promotes tissue repair and remyelination (Eming, Wynn, & Martin, 2017). Dys- regulation of this process results in remyelination failure (Miron et al., 2013), a hallmark pathological feature of chronic MS lesions that can ultimately lead to axonal degeneration and progressive clini- cal impairment (Chang, Tourtellotte, Rudick, & Trapp, 2002; Trapp and Nave 2008). Within chronic, but not active or remyelinating MS lesions, myeloid cells lack expression of classic M2-associated cell surface markers, transcription factors, and cytokines (Peferoen et al., 2015; Vogel et al., 2013), and may be indicative of an inefficient transition to the regenerative phenotype, thus resulting in remyelina- tion failure in progressive MS. Using primary bone marrow-derived macrophages lacking miR-223 expression, our data demonstrated that these cells have an impaired ability to polarize toward the M2 phenotype (Figure 5). These results are consistent with previous reports whereby macrophages derived from miR-223 KO mice have impaired M2 polarization following exposure to IL-4 and/or PPARγ ligands and investigated in the context of obesity (Ying et al., 2015; Zhuang et al., 2012). In the context of brain injury, we have further elaborated on these findings by demonstrating that M2 polarization is also impaired in brain-resident microglia (Figure 5). This result is of great clinical relevance given that transcriptional profiles between blood-derived macrophages and microglia significantly differ between these myeloid cell subtypes (Butovsky et al., 2014). Com- pared with wild-type cells, we observed that both CD206 and Ym1 expression were reduced in M2-polarized miR-223 KO myeloid cells. in vitro expression of Arg1 in M2 polarized myeloid cells was similar between miR-223 KO and WT cells (Figure 5). While Arg1 is a marker of M2 macrophage polarization, its expression is not an abso- lute requirement (Murray & Wynn, 2011). It is also worth noting that the M1/M2 model of myeloid cell polarization does not fully capture FIGURE 8 Myelin debris clearance is impaired in miR-223 KO mice following lysolecithin-induced demyelination. (a) Representative immunofluorescent images of WT and miR-223 KO lesions at 14 days post-injection. (b, c) Quantification of Iba1+ macrophages/microglia (b), myelin-laden macrophages (c) in lysolecithin lesions. n = 3 to 5/group; statistical significance determined by unpaired Student’s t-test. Data are expressed as mean SEM. Scale bars are 20 μm [Color figure can be viewed at wileyonlinelibrary.com] the diversity of myeloid cells, and in particular microglia (Ransohoff, 2016), suggesting a complex effect of miR-223 on myeloid cell pheno- type that may extend beyond the binary M1/M2 distinction. Further- more, macrophages lacking miR-223 demonstrated a functional impairment in phagocytosis in vitro (Figure 6), which also translated in vivo, as mice lacking miR-223 displayed impaired myelin debris clear- ance following lysolecithin-induced demyelination (Figure 8). This impairment is likely a significant factor contributing to the decreased remyelination capacity in the miR-223 KO mice since myelin debris has been shown to negatively influence OPC differentiation and remyelina- tion (Kotter, Li, Zhao, & Franklin , 2006; Neumann, Kotter, & Franklin , 2009). Arg1 immunoreactivity was similar in vivo in WT and miR-223 KO mice, suggesting that the functionality, but not recruitment, of M2 polarized myeloid cells is impaired in vivo. In mouse macrophages, the anti-inflammatory role of miR-223 has been suggested given its ability to target the pro-inflammatory transcription factor NFAT5, as well as multiple STAT transcription fac- tors, including STAT1, STAT3, and STAT5 (Chen et al., 2012; Moles, Bellon, & Nicot, 2015; Pinatel et al., 2014; Ying et al., 2015). miR-223 has also been shown to target pro-inflammatory RASA1 (also known as p120-RasGAP), a result that we have confirmed in both mouse and human macrophages (Figure 4). RASA1 is a GTPase activating protein that terminates Ras signaling following activation of receptor tyrosine kinases (RTKs) (King, Lubeck, & Lapinski, 2013). Although the precise mechanism by which RASA1 influences the phenotype of myeloid cells, particularly alternative activation, is currently unknown, Ras regulates numerous signaling pathways including ERK, PI3K, and MAPK. Additionally, RASA1 is inhibited by SOCS3 (Cacalano, Sanden, & Johnston, 2001), further suggesting inhibition of RASA1 is critical for anti-inflammatory functions. Herein, we demonstrated that the M1-polarization phenotype and its associated functions in vitro (as measured by ROS and cytokine levels) were relatively similar in both macrophages and microglia derived from miR-223 KO and WT mice (Figure 3). To further investigate any potential contribution of miR-223 toward the M1 pro-inflammatory phenotype in vivo, we performed an acute LPS challenge and observed equivalent peripheral cytokine responses between miR-223 KO mice and WT-littermates (Figure 3). We also performed EAE in miR-223 KO mice and observed similar peak clinical scores compared with WT mice (Figure 2). Previous EAE studies using miR-223 KO mice had demonstrated reduced EAE severity due to impaired pathogenic T cell differentiation (Cantoni et al., 2017; Ifergan et al., 2016). The authors described this impaired T cell differen- tiation as myeloid cell dependent because antigen-presenting cells derived from miR-223 KO mice failed to sufficiently drive IL-17/GM-CSF production by MOG-specific T cells (Ifergan et al., 2016). The C57Bl/6-MOG35–55 model of EAE is T cell dependent (Rangachari & Kuchroo, 2013). Consistent with our data, previous adoptive transfer EAE experiments using in vitro polarized T cells demonstrated equivalent clinical score in WT and miR-223 KO mice (Cantoni et al., 2017), suggest- ing a similar pro-inflammatory myeloid cell polarization despite an impaired T-cell response. In addition to assessing the potential involvement of miR-223 in pro-inflammatory myeloid cells, we also performed both in vitro and in vivo experiments to investigate how miR-223 could regulate the M2 anti-inflammatory and tissue-regenerating myeloid cell pheno- type. Unlike EAE, the distinct effector cells in the lysolecithin- induced demyelinating model include the macrophages and microglia and is a more ideal model to study remyelination (Franklin & Ffrench-Constant, 2017; Ousman & David, 2000). The failure of miR-223 KO mice to efficiently remyelinate in this model coupled by their inherent inability to adopt a robust M2-like phenotype and phagocytic function is consistent with previous studies implicating a critical role of the alternatively activated macrophage/microglia dur- ing remyelination (Miron et al., 2013).An important component within this study involved assessing a role for miR-223 in macrophages and microglia derived from both mouse and human species. Based on in vitro overexpression data, the anti-inflammatory role for miR-223 is conserved and is suggestive that miR-223 may not only be relevant in MS, but also in other neurologi- cal diseases associated with inflammatory components. For example, levels of miR-223 in leukocytes are increased in patients with acute ischemic stroke (Wang et al., 2014b), while exogenous miR-223 limits injury in models of ischemic and hemorrhagic stroke (Harraz et al., 2012; Yang et al., 2015). Based on our data, miR-223 may exert its beneficial effects in these models through its ability to promote M2 myeloid cell polarization and subsequent repair. In addition, myeloid- derived miR-223 can be transferred to neurons during neuroinflam- mation (Prada et al., 2018), suggesting that miR-223 may also exert neuroprotective and pro-regenerative effects directly within neu- rons. Studies investigating miR-223 in other neurological diseases associated with activation of innate immunity, such as Alzheimer's or ALS, will further elucidate the role of miR-223 in neuroinflammation, neurodegeneration, and brain repair. A protective role of miR-223 has also been described in non-CNS tissues. Knockout of miR-223 leads to exaggerated inflamma- tion in mouse models of lung injury, liver injury, obesity, sepsis, and colitis (Li et al., 2017; Neudecker et al., 2017a, 2017b; Wang et al., 2014a; Zhou et al., 2015; Zhuang et al., 2012). In these studies, the effects of miR-223 have been attributed to the suppression of pro- inflammatory genes, and the hyper-activation of innate immunity and enhanced tissue destruction/impairments in tissue repair. Thus, our data demonstrating impaired CNS remyelination following lysolecithin-induced demyelination are consistent with the predom- inant view that miR-223's role is largely anti-inflammatory (Yuan et al., 2018). We have also demonstrated that forced overexpres- sion of miR-223 can suppress both macrophage and microglial inflammatory responses in vitro (in human and mouse), supporting the possibility for delivery of miR-223 as a potential therapeutic agent. This is important as both microRNAs (Rupaimoole & Slack,2017) and remyelination-targeting therapeutics (Plemel, Liu, & Yong, 2017) are translated to clinical populations. Additionally, the ability for current disease-modifying therapies to modulate levels of miR-223 should be assessed in future studies. In summary, this study demonstrates that miR-223 is upregu- lated in MS patient myeloid cells and is essential for efficient anti- inflammatory myeloid cell polarization. Moreover, we have shown that miR-223 is essential for promoting efficient debris clearance and remyelination in the CNS following focal demyelinating injury. Thus, miR-223 Myelin Oligodendrocyte Glycoprotein 35-55 dysregulation may represent an important molecular mechanism underlying impaired remyelination in progressive MS.