MMP7 cleaves remyelination‐impairing fibronectin aggregates and its expression is reduced in chronic multiple sclerosis lesions

Abstract Upon demyelination, transient expression of fibronectin precedes successful remyelination. However, in chronic demyelination observed in multiple sclerosis (MS), aggregates of fibronectin persist and contribute to remyelination failure. Accordingly, removing fibronectin (aggregates) would constitute an effective strategy for promoting remyelination. Matrix metalloproteinases (MMPs) are enzymes known to remodel extracellular matrix components, including fibronectin. Here, we examined the ability of MMPs to degrade fibronectin aggregates. Our findings reveal that MMP7 cleaved fibronectin aggregates resulting into a prominent 13 kDa EIIIA (16 kDa EDA)‐containing fragment. MMP7 was upregulated during lysolecithin‐induced demyelination, indicating its potential for endogenous fibronectin clearance. In contrast, the expression of proMMP7 was substantially decreased in chronic active and inactive MS lesions compared with control white matter and remyelinated MS lesions. Microglia and macrophages were major cellular sources of proMMP7 and IL‐4‐activated, but not IFNγ+LPS‐activated, microglia and macrophages secreted significant levels of proMMP7. Also, conditioned medium of IL‐4‐activated macrophages most efficiently cleaved fibronectin aggregates upon MMP‐activating conditions. Yet, coatings of MMP7‐cleaved fibronectin aggregate fragments inhibited oligodendrocyte maturation, indicating that further degradation and/or clearance by phagocytosis is essential. These findings suggest that MMP7 cleaves fibronectin aggregates, while reduced (pro)MMP7 levels in MS lesions contribute to their persistent presence. Therefore, upregulating MMP7 levels may be key to remove remyelination‐impairing fibronectin aggregates in MS lesions.

component of the signaling microenvironment and participates in the regulation of OPC behavior, its remodeling serving as an effective mechanism to regulate repair. Specifically, following central nervous system (CNS) demyelination, extensive ECM remodeling leads to alterations in ECM expression profiles (Back et al., 2005;Zhao, Fancy, Franklin, & ffrench-Constant, 2009;Lau et al., 2012;Hibbits, Yoshino, Le, & Armstrong, 2012;Stoffels et al., 2013). For example, following lysolecithininduced demyelination and in chronic-relapsing (cr-)EAE, remyelinationimpairing fibronectin and chrondroitin sulfate proteoglycans (CSPGs), and remyelination-promoting laminin are readily expressed (Milner et al., 2007;Lau et al., 2012;Stoffels et al., 2013). Evidently, the regulation and transient expression of these distinct ECM molecules, is essential in maintaining the proper physiological environment for timely development of OPCs into mature, myelinating oligodendrocytes. Indeed, a dysbalance in expression of these ECM entities contributes to an impaired differentiation of OPCs, as observed in 70% of MS lesions (Lucchinetti et al., 1999;Kuhlmann et al., 2008;Chang et al., 2012). More specifically, in chronic MS lesions, but not in toxin-induced lesions, fibronectin is persistently present as aggregates, which frustrates OPC differentiation, and thereby impedes remyelination (Stoffels et al., 2013). Moreover, fibronectin precludes the ability of laminin, also present in MS lesions (Van Horssen, B€ o, Vos, Virtanen, & de Vries, 2005), to facilitate myelin membrane formation (Baron et al., 2014), emphasizing the necessity to remove aggregated fibronectin to allow remyelination.
Interestingly, fibronectin mRNA is hardly present in chronic MS lesions (Stoffels et al., 2013) and aggregates are formed extracellularly (Mao & Schwarzbauer, 2005, our unpublished observations), indicating that a perturbed clearance, rather than an altered expression of fibronectin, is responsible for its accumulation. Selective removal of astrocytederived fibronectin from the lesion site reveals that although dimeric fibronectin promotes OPC proliferation following demyelination, reduced numbers of OPCs suffice for successful remyelination (Stoffels, Hoekstra, Franklin, Baron, & Zhao, 2015). Therefore, removal of dimeric or aggregated fibronectin may represent a therapeutic strategy to promote remyelination in MS. Important players in controlled ECM degradation are matrix metalloproteinases (MMPs). Commonly, MMPs are synthesized and secreted as pro-enzymes that are subsequently activated by various proteinases, including other MMPs and plasmin (Lu, Takai, Weaver, & Werb, 2011).
Under healthy conditions, MMPs are transiently involved in the regulation of ECM dynamics upon injury, and are under strict (local) control at various levels, including gene transcription, synthesis, secretion, propeptide activation, and inhibition by physiological tissue inhibitors of MMPs (TIMPs). MMPs are implicated in the pathogenesis of MS, and some MMPs, including MMP3, MMP7, and MMP9, are upregulated in MS lesions (Cuzner et al., 1996;Maeda & Sobel, 1996;Cossins et al., 1997;Anthony et al., 1997;Lindberg et al., 2001). While fibronectin is a substrate for at least 14 distinct MMPs (Lu et al., 2011), it is unknown whether they dissociate, cleave and/or degrade fibronectin aggregates.
Given the role of MMPs in ECM remodeling in injured CNS, we examined here whether a perturbed expression and/or malfunctioning of MMP3, MMP7 or MMP9 contributes to the inability to clear dimeric fibronectin and/or fibronectin aggregates in MS lesions. We demonstrate that MMP7 cleaved fibronectin aggregates and that proMMP7 is weakly expressed in chronic MS lesions compared with remyelinated lesions. IL-4-activated microglia and macrophages were major cellular sources of proMMP7. Hence, local targeting of MMP7 levels in chronic MS lesions may represent a first step to remove remyelination-inhibiting fibronectin aggregates.

| MS lesions
Autopsy samples of human brain material were obtained from the Netherlands Brain Bank and with the approval of the VU University Medical Ethical Committee (Amsterdam, The Netherlands). Patients and controls, or their next of kin, had given informed consent for the use of their brain tissue and clinical details for research purposes.
For immunohistochemical analysis paraffin-embedded (n 5 12) or snap-frozen (n 5 5) tissue from MS patients and non-neurological controls (n 5 3) were used. Western blot studies were performed on nine control white matter (CWM), eight (chronic) active MS lesion [(c)aMS)], nine chronic inactive MS lesion (ciMS), and two remyelinated MS lesion (rMS) homogenates. Brain tissue was homogenized as previously described (Maier et al., 2007). CWM did not show any histological signs of inflammation and demyelination and was obtained from subjects without clinical signs of neurological disease. MS lesions were classified as previously described (van der Valk & De Groot, 2000).

| Toxin-induced demyelination 2.2.1 | Lysolecithin
To induce local demyelination 8-10-week-old female C57BL/6 mice (for RT-qPCR) or Sprague Dawley rats (for Western blot) were injected with 1 ml of 1% lysolecithin (Sigma) in spinal cord white matter (Zhao, Li, & Franklin, 2006). At indicated time points, animals were sacrificed and tissue processed for Western blot and qPCR analysis as previously described (Zhao et al., 2006;Stoffels et al., 2013). For immunohistochemistry, animals were perfused with 4% paraformaldehyde via the left ventricle, after which the dissected spinal cord containing lesions was treated with 20% sucrose in phosphate-buffered saline (PBS) overnight. The brains were cryosectioned at 12 lm thickness and stored at 2808C until further processing. Control spinal cord tissues were taken from non-lesioned thoracic segments of spinal cord, distant from the lesion site. Experiments were performed in compliance with UK Home Office regulations.

| Cuprizone
To induce robust and reproducible demyelination of the corpus callosum, 9-week old male C57BL/6 mice (Harlan, Horst, the Netherlands) were individually housed and subjected to a standard powder chow diet containing 0.2% cuprizone (bis(cyclohexanone)-oxaldihydrazone, Sigma, St. Louis, MO). After 5 weeks, animals returned to standard chow. Tissue was processed as described for lysolecithin-induced lesions. Demyelination of the corpus callosum was confirmed by Sudan black staining (0.1% in 70% ethanol for 5 min). All experimental procedures were approved by the Animal Ethical Committee of the University Medical Center Groningen (the Netherlands).
The mixed glia flasks were shaken overnight on the orbital shaker at 240 rpm, and the floating OPCs obtained by this procedure were further purified via differential adhesion (Bsibsi et al., 2012). Isolated OPCs (>97% Olig2-positive) were plated at a density of 1.0 3 10 6 per 10-cm dish (in 6 ml) for Western blotting or on 8-well Permanox chamber slides (Nunc) at a density of 30,000 cells per well for the maturation assays. OPCs were cultured for 2 days in SATO medium (Maier, Baron, & Hoekstra, 2005) supplemented with growth factors FGF-2 (10 ng/ml, Peprotech) and PDGF-AA (10 ng/ml, Peprotech). To obtain mature oligodendrocytes, OPC maturation was initiated by growth factor withdrawal and culturing in SATO supplemented with 0.5% FBS for 7 days.
For experimental analysis, macrophages were plated and treated as described for microglia.

| Generation of fibronectin aggregates
Deoxycholate (DOC)-insoluble aggregated fibronectin was prepared from primary rat astrocytes or MS lesions homogenates. Astrocytes were plated at a density of 1 3 10 6 cells per 10-cm dish, and after 1 hr treated with Toll-like receptor 3 agonist poly(I:C) (50 lg/ml, GE Healthcare, Munich, Germany). After 48 hr, astrocytes were removed by water-lysis for 2 hr at 378C. aggregate-containing pellet was washed three times in PBS, followed by resuspension in PBS with a syringe and 25-gauge needle. The quality of the fibronectin aggregates, that is, the lack of dimeric fibronectin and/or smaller products, and the extent of aggregation, was routinely checked by Western blot.

| Fibronectin-free serum
Fibronectin was depleted from serum with a gelatin sepharose 4B column (GE healthcare) according to manufacturer's instructions. The generated fibronectin-free serum was filtered (0.2 lm) and stored in aliquots at 2208C. The absence of fibronectin in serum was confirmed by Western blot.

| Lactate dehydrogenase and MTT assay
OPCs were cultured in 24-wells plates at a density of 50,000 cells per well on the indicated substrates. After 48 hr the medium [lactate dehydrogenase (LDH) assay] and cells [3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; MTT assay] were analyzed as described (Stoffels et al., 2013). Briefly, the release of LDH into the medium was measured using a commercial LDH assay kit (Roche) according to manufacturer's instructions. The effect on cell viability was determined with an MTT assay, for which cells were incubated with MTT diluted in culture medium (0.5 mg/ml, Sigma) for 4 hr. MTT-formazan crystals were collected in dimethyl sulphoxide and absorption measured at 560 nm.

| Toxin-induced lesions
Sections were blocked and permeabilized with PBS containing 5% normal donkey serum and 0.1% Triton-X-100. Sections were incubated with a mixture of primary antibodies (Table 1) overnight at 48C, followed by a 2-hr incubation with appropriate Alexa-conjugated secondary antibodies. Nuclei were visualized by Hoechst 33342 or DAPI (1 mg/ml, Sigma). Sections were analyzed with confocal laser scanning microscopy (Leica SP8 AOBS).

| MS lesions
Luxol-fast blue staining was used to identify shadow plaques. To visualize myelin loss, microgliosis and (pro)MMP3 and (pro)MMP7 expression, 5-mm serial sections of active (n 5 3), chronic active (n 5 3) and inactive lesions (n 5 3), shadow plaques (n 5 3) and healthy controls (n 5 3) were single-stained or double-stained. Slides were deparaffinized in xylene and descending ethanol concentrations. After endogenous peroxidase inhibition with H 2 O 2 , antigen retrieval and/or blocking (Table 1), sections were incubated with primary antibodies for 1 hr or overnight ( Table 1) After blocking with H 2 0 2 , slides were incubated overnight with both primary antibodies (Table 1), which were visualized with DAB and liquid permanent red (LPR) as described above. Sections were analyzed and representative pictures were taken with an Olympus BX41 microscope, equipped with a Leica MC170 HD camera.

| Immunocytochemistry
Cells were fixed with 4% paraformaldehyde in PBS for 20 min, permeabilized for 5 min in ice-cold methanol and blocked with 4% BSA for 30 min.
The cells were incubated for 1-2 hr with anti-MBP, followed by incubation with an appropriate TRITC-conjugated secondary antibody (1:50, Jackson ImmunoResearch, Westgrove, UK) and DAPI for 25 min. Cells were covered with mounting medium (Dako) to prevent image fading and analyzed with conventional immunofluorescence microscopy (Olympus AX70 or Leica DMI 6000 B). Oligodendrocytes were characterized by morphology, that is, cells with typical astrocytic morphology were excluded (<3%), and in each experiment at least 250 cells were manually scored as either MBPnegative or MBP-positive, while in addition MBP-positive cells were classified as myelin membrane-forming or non-myelin membrane-forming.

| Western blot analysis
Cells were collected by scraping with PBS and centrifuged at 9,200g for 5 min. Cell pellets were sonicated in TNE buffer (50 mM Tris-HCl, 150 mM NaCl, and 5 mM EDTA, pH 7.5) for 10 s on ice. Total protein concentration was measured by a Bio-Rad DC Protein Assay (Bio-Rad Laboratories, Herculas, CA) using BSA as standard. Equal amounts of protein (50 lg for cell lysates and brain homogenates) or equal volumes of medium (40 ml) were mixed with SDS-reducing loading buffer, denatured at 958C for 5 min and subjected to Western blotting as previously described (Bsibsi et al., 2012). Primary antibodies used are indicated in Table 1. The signals were detected using the Odyssey Infrared Imaging System (LI-COR Biosciences, Lincoln, NE) and analyzed using Scion image software.

| Real-time quantitative polymerase chain reaction reaction (RT-qPCR)
Total  (Table 2) and absolute SYBR Green Rox Mix (ThermoScientific, Breda, the Netherlands) were mixed. Gene expression was calculated by the 2 -DDct method.
GADPH and HMBS were used as housekeeping genes.

| Statistics
Data are expressed as mean 6 SEM for at least three independent experiments. Statistical analysis was performed with a one sample

| R E SU LTS
3.1 | proMMP3 and MMP7 are upregulated upon toxin-induced demyelination Immunohistochemical analysis and in situ hybridization studies revealed that expression of fibronectin-degrading proteinases MMP3 and MMP7 is enhanced in active demyelinating MS lesions (Anthony et al., 1997;Cossins et al., 1997;Lindberg et al., 2001). To examine whether the enhanced expression of these MMPs in MS lesions is a natural response to a demyelinating insult, we first examined their expression levels in two different animal models of toxin-induced demyelination.
Notably, we and others have previously shown that upon toxininduced demyelination, dimeric fibronectin is transiently increased during demyelination, and its clearance commensurate with remyelination (Zhao et al., 2009;Hibbits et al., 2012;Stoffels et al., 2013;Espitia Pinzon et al., 2017), making it an ideal model to examine the regulation of "natural" fibronectin-degrading proteolytic enzymes involved in ECM remodeling upon CNS demyelination. Focal demyelination in spinal cord white matter is induced by a single injection of the detergent lysolecithin, being maximal at 5 days post lesion (DPL), which is followed by robust remyelination at 14-21 DPL (Zhao et al., 2006). As shown in  (Ulrich et al., 2006;Skuljec et al., 2011). Also, (pro)MMP7 and MMP7 were present during demyelination and remyelination, while in contrast to lysolecithininduced demyelination, no difference in their ratio was observed (Figure 2b,d). Immunohistochemical analysis confirmed the similar (pro) MMP7 expression levels (Supporting Information Figure S1). In contrast to lysolecithin-induced lesions, (pro)MMP7 was scarcely localized to Iba1-positive microglia/macrophages and appears to localize extracellularly (Supporting Information Figure S1, insets), while a prominent increase in Iba1-positive microglia/macrophages was evident at 3 and 5 weeks demyelination. This discrepancy may be explained by taking into account that in the lysolecithin-induced lesions (pro)MMP7 levels were examined during early remyelination, whereas in the cuprizone model (pro)MMP7 levels were analyzed at conditions at which remyelination nears completion. This suggests that microglia/macrophages transiently express MMP7, and therefore, the presence of (pro)MMP7 during remyelination upon cuprizone-induced demyelination may be underestimated. Hence, these findings show that (pro)MMP3 and (pro) MMP7 are present in demyelinated areas, and that their expression levels are enhanced during the remyelination process, indicating that they may be involved in ECM remodeling upon injury, including fibronectin clearance, which we examined next.

| MMP7-cleaved fibronectin aggregate coatings inhibit OPC maturation
To assess whether the MMP7-mediated proteolytic digestion of fibronectin aggregates is sufficient to overcome the aggregated fibronectin-mediated inhibition of myelin membrane formation, OPCs were plated on intact fibronectin aggregates or aggregates that were cleaved with recombinant MMP7 (see Figure 3). As shown in Figure  In each experiment, the data of cells cultured on PLL was set at 100% (horizontal line). The percentages of MBP-positive cells or myelin membranes in cells cultured on PLL were respectively 33.6.8% 6 9.5% and 24.0% 6 10.2% for the pFn-related experiments, 34.9.8% 6 8.6% and 26.1% 6 8.2% for the cFnrelated experiments, and 43.8% 6 11.9% and 29.1% 6 8.7% for the aFn-related experiments. Statistical differences with cells cultured on PLL (one sample t test, ***p < .001, n 5 5) and intact fibronectin (Student's t test, ##p < .01) are indicated. Note, that MMP7-cleaved plasma, cellular and aggregated fibronectin inhibited OPC differentiation compared with their intact counterparts. (e and f) Cytotoxicity (e, LDH release in culture medium) and cell viability (f, MTT reduction) assays of OPCs cultured for 2 days at the indicated substrates. Bars depict mean 1 SEM of four independent experiments. In each experiment, the data of respective intact substrates was set at 100% (horizontal line). Statistical analyses were performed using the one-sample t test when compared with the intact substrate (not significant) [Color figure can be viewed at wileyonlinelibrary.com] tissue. Of note, the anti-MMP7 antibody does not distinguish proMMP7 from MMP7 (Figures 1b and 2b). In control tissue, (pro)MMP7 is diffusely expressed throughout the white matter (Figure 5a). In active MS lesions, characterized by HLA-DR-expressing microglia/macrophages in the demyelinated area (Figure 5a, van der Valk & De Groot, 2000), (pro) MMP7 localized to HLA-DR-positive microglia/macrophages (Figure 5a, arrow), corroborating previous studies (Anthony et al., 1997;Cossins et al., 1997). Co-stainings of MMP7 with human phenotype-specific differentially-activated microglia/macrophages markers (Peferoen et al., 2015) in active MS lesions showed that ( Therefore, we next examined in vitro which type(s) of resident cells, present in MS lesions, is (are) the major source(s) of these MMPs.

| IL-4-activated microglia and macrophages are a major cellular source of secreted proMMP7
Several in vitro and in vivo studies reveal that resident microglia and astrocytes, as well as infiltrating macrophages produce MMPs (Anthony et al., 1997;Cossins et al., 1997;Skuljec et al., 2011). To identify cellular sources of (pro)MMP7 expression and release, and to explain its impaired expression in chronic MS lesions, we examined the presence MMP7 in lysates and in conditioned medium of cultured microglia and astrocytes, as well as bone marrow-derived macrophages (referred to as "macrophages") under resting (control) and/or MS-relevant conditions that do not affect cell viability. In addition, given their role in MS pathology and/ or their ability to activate MMP7, we also examined cellular sources of (pro)MMP3 and (pro)MMP9. Models of microglia and macrophage activation are often simplified to classical activation, evoked by exposure to IFNg and LPS, or alternative activation, generated by exposure to IL-4.
These phenotypes are regarded as the two extremes of a continuum of microglia/macrophage activation states (Wolf, Boddeke, & Kettenmann, 2017;Murray, 2017). As shown in Figure 6a MMP7 and MMP9 were hardly present in microglia and macrophages lysates and conditioned medium. Importantly, consistent with the corresponding phenotype, iNOS expression was increased upon exposure to IFNg1LPS, and arginase-1, a marker for alternative activated microglia and macrophages, was expressed upon IL-4 treatment (Figure 6a,d). Proinflammatory-activated astrocytes, a condition that is relevant to MS lesions (Nair, Frederick, & Miller, 2008;Sofroniew & Vinters, 2010), showed increased levels of MMP7, but not proMMP3 and proMMP7 compared with control astrocytes (Supporting Information Figure S3).
ProMMP7 and proMMP9, but not MMP7, were detected in proinflammatory-activated astrocyte conditioned medium, while the MMPs were virtually absent in conditioned medium of control astrocytes (Supporting Information Figure S3). While OPCs are present in most MS lesions, their differentiation seems to be inhibited (Lucchinetti et al., 1999, Kuhlmann et al., 2008, Chang et al., 2012, which may account for the observed alterations in MMP expression levels. Upon differentiation of OPCs to myelinating oligodendrocytes, proMMP3 expression significantly increased, while proMMP7 and proMMP9 were hardly detected at either differentiation stage (Supporting Information Figure S3). ProMMP3 was virtually absent in conditioned medium of primary oligodendrocyte cultures (data not shown). Hence, pro-inflammatory cytokine-activated astrocytes, and IL-4-activated microglia and macrophages secrete proMMP7 levels that may potentially degrade fibronectin aggregates.
3.6 | MMP-activated conditioned medium of IL-4activated microglia and macrophages cleaves fibronectin aggregates As IL-4-activated microglia, macrophages, and cytokine-activated astrocytes secreted significant proMMP7 levels, we next determined WANG ET AL.

| 1635
whether MMPs present in the conditioned medium of either cell type were sufficient to cleave aggregated fibronectin. Incubation with aggregated fibronectin for 72 hr at 378C showed that conditioned medium of either cell type was insufficient in cleaving aggregated fibronectin ( Figure 7). As Western blot analysis revealed that MMP7 was present as proenzyme (Figure 6), we treated the conditioned medium with APMA, a general activator of MMPs. Interestingly, APMA-treated, that is, MMP-activated, conditioned medium from IL-4-activated F IGURE 5.
macrophages, and to a lesser extent microglia cleaved aggregated fibronectin (Figure 7a,b). In IL-4-activated macrophages and microglia, similar to recombinant MMP7-mediated cleavage of aggregated fibronectin, a major degradation product visualized with the anti-EIIIA fibronectin antibody was a 13 kDa fragment (Figure 7b cf Figure 3b).
Also, the 9 kDa product that appears upon recombinant MMP7cleavage of aggregated fibronectin using the polyclonal fibronectin antibody was visible (Figure 7a cf Figure  Upon MMP-activating conditions, conditioned medium of cytokine-activated astrocytes, known to contain proMMP7 and proMMP9 (Supporting Information Figure S3), was not able to fragment aggregated fibronectin (Figure 7a,b). This indicates that other factors present in astrocyte-conditioned medium may prevent MMP7mediated cleavage, or that expression levels of MMPs are too low.
Expression profiling showed that upon pro-inflammatory cytokine exposure, next to transcripts of MMP3, MMP7, MMP9, MMP12, MMP13, and ADAMST, also the natural inhibitors of MMPs, TIMP1, and TIMP3, were increased compared with "resting" astrocytes (Table   3). Hence, macrophages and microglia are potential cellular sources to degrade fibronectin aggregates, and once secreted by IL-4-activated microglia/macrophages, MMPs, most likely MMP7, require an appropriate, extracellular activation, for example by other cellular sources, to cleave aggregated fibronectin.

| D ISC USSION
Fibronectin aggregates present in chronic MS lesions impair OPC maturation, and contribute to remyelination failure (Stoffels et al., 2013).
These MMPs are present at the rim of active lesions, and less so or confined to perivascular cuffs, in chronic MS lesions (Maeda & Sobel, 1996;Lindberg et al., 2001). The upregulated MMPs are likely part of the neuroinflammatory response that also encompasses disruption of the blood-brain barrier, allowing entry of blood-derived cells (Rosenberg, Estrada, Dencoff, & Stetler-Stevenson, 1995;Buhler et al., 2009).
Our present findings show that MMP7 cleaved fibronectin aggregates into small, EIIIA/EDA containing fragments. Previous studies revealed that MMP7 is present in macrophages in active demyelinating lesions (Anthony et al., 1997;Cossins et al., 1997  Actin serves as a loading control. Bars depict mean 1 SEM relative to control. Statistical differences as assessed with a one-sample t test are indicated (*p < .05, n 4) Lindberg et al., 2001). MMP7 is implicated in blood-brain barrier disruption, axonal injury and is involved in shedding of signaling molecules, including TNF, that contribute to the pathology of MS (Chandler et al., 1995;Gearing et al., 1995;Kieseier et al., 1998;Newman et al., 2001;Buhler et al., 2009). In contrast, in the present work, we demonstrated that upon lysolecithin-induced demyelination, MMP7 was upregulated during remyelination, and also present in remyelinated MS lesions. Furthermore, others have shown that MMP7 levels also remain elevated during the remission phase of EAE (Kieseier et al., 1998). As MMP7 efficiently degrades CSPGs and fibronectin, both known to inhibit OPC maturation and transiently expressed upon demyelination while being persistent in MS lesions (Lau et al., 2012;Stoffels et al., 2013), MMP7 may aid to their timely degradation. In addition, since MMP7 is able to degrade MBP (Chandler et al., 1995) (Imai et al., 1995), which is highly expressed by astrocytes upon demyelination ( Skuljec et al.,  (Lu et al., 2011), and may be a rate limiting step in proper MMP7 activation. Furthermore, while proMMP7 was secreted by pro-inflammatory cytokine-activated astrocytes, fibronectin aggregates remained virtually intact upon incubation with astrocyte conditioned medium at MMP-activating conditions. Our RT-qPCR analysis showed that pro-inflammatory cytokines increased TIMP1 and TIMP3 mRNA expression in astrocytes suggesting that such inhibitory molecules may attenuate MMP activity, adding complexity to regulation of local MMP7 activity.
The decreased MMP7 levels in chronic active and inactive MS lesions may be due to dysfunctional activation of microglia and macrophages. Indeed, microglia and macrophages in inflammatory MS lesions have an intermediate activation status, expressing classical and alternative activated phenotypes (Vogel et al., 2013;Peferoen et al., 2015). The presence of alternative activated microglia/macrophages is essential for remyelination (Miron et al., 2013), and this may rely on appropriate ECM remodeling (Agrawal et al., 2013).
MMP7-mediated cleavage of fibronectin aggregates was not sufficient to overcome aggregate-induced inhibition of OPC maturation but rather differentiation was decreased when OPCs were plated on MMP7-treated fibronectin aggregate coatings, compared with intact fibronectin aggregate coatings. This may be due to exposure of OPCs to EIIIA/EDA-containing fragments, released upon MMP7-induced cleavage of the aggregates. The EIIIA/EDA domain, particularly when it is present as a fragment, acts as an endogenous ligand for TLR4 (Okamura et al., 2001;Bhattacharyya et al., 2014). However, OPCs do not express TLR4 (Lehnardt et al., 2003;Sloane et al., 2010;Bsibsi et al., 2012), suggesting that upon release of EIIIA/EDA-containing fragment the remaining aggregate may be remodeled, which altered the binding and signalling activity to OPCs in a different fashion. Also, MMP7treated EIIIA-lacking plasma fibronectin coatings inhibited OPC differentiation compared with intact plasma fibronectin, corroborating that other fragments than an EIIIA-containing fragment inhibit OPC differentiation. Therefore, the remaining aggregates may require further processing and clearance by other proteases, e.g., MMP12 ( Skuljec et al., 2011), and/or phagocytosis. It is of particular interest in this regard that fibronectin fragments, including EIIIA/EDA itself, can activate MMP expression (Saito et al., 1999;Yasuda et al., 2003). All data relative to (untreated) control are shown, which was set to 1 at each independent experiment; Data are expressed as mean 6 SEM (n 4); Statistical differences with control as assessed with a one sample t test are indicated (in bold, *p < .05. ***p < .001).
Thus, local addition, and specific and timely activation of MMP7 may be a therapeutic option to clear remyelination-inhibiting fibronectin aggregates from MS lesions. However, due to potential side effects such as MMP7-induced cleavage of (beneficial) laminin (Rosenberg, 2002a;Lu et al., 2011), MMP7-mediated activation of harmful cytokines Yamamoto et al., 2014), or its toxicity to axons (Newman et al., 2001), such an approach will require careful targeting and timing of MMP7 activation. Accordingly, further knowledge as to why the cleavage pattern of aggregated fibronectin differs from that of plasma fibronectin, as well as identification of MMP7 cleavage sites in the aggregates will be key to induce MMP7 to specifically cleave aggregated fibronectin and enhance remyelination.

CONFLICTS OF INTEREST
The authors declare no conflict of interest.