key: cord-0754681-dotjy9ur
authors: Edwards, Julie A; Denis, François; Talbot, Pierre J
title: Activation of glial cells by human coronavirus OC43 infection
date: 2000-08-01
journal: J Neuroimmunol
DOI: 10.1016/s0165-5728(00)00266-6
sha: c46612b3da6c1968adba5d16892f1024b0d1b1fc
doc_id: 754681
cord_uid: dotjy9ur

Multiple sclerosis (MS) is an immune-mediated demyelinating disease that could be triggered by a viral infection. Coronaviruses induce an MS-like disease in rodents, are neuroinvasive in humans and can infect primary cultures of human astrocytes and microglia. Infection of the human astrocytic cell line U-373MG by the OC43 strain of human coronavirus caused an upregulation of IL-6, TNF-α, and MCP-1 mRNA expression. This virus also modulated the activity of matrix metalloproteinases-2 and -9 and augmented nitric oxide production in both U-373MG cells and the human microglial cell line CHME-5. Thus, a coronaviral infection of glial cells could lead to the production of inflammatory molecules that have been associated with central nervous system pathologies such as MS.

1. Introduction al., 2000) . Interestingly, primary cultures of human astrocytes and microglia are susceptible to HCoV infection Multiple sclerosis (MS) is an inflammatory demyelinat- (Bonavia et al., 1997) . Thus, this virus may be involved in ing disease of the central nervous system (CNS) that is demyelinating diseases by a mechanism of virus-induced characterized by mononuclear cell infiltration into the myelin damage following infection of astrocytes and / or CNS, myelin degradation and oligodendrocyte loss. It is microglia. the most common neurological disease affecting young It has been suggested that upon infection of glial cells, adults (Ewing and Bernard, 1998) . The etiology and the upregulation of pro-inflammatory molecules might pathogenesis of MS have yet to be elucidated but are participate in the pathogenesis of several inflammatory probably multifactorial, involving both genetic and en-CNS diseases (Chao et al., 1996; Sriram and Rodriguez, vironmental factors (Ewing and Bernard, 1998) . Several 1997) . Given our in vitro observations of glial cell viruses have been associated with demyelinating processes infection by HCoV, we hypothesize that HCoV may partici- (Sarchielli et al., 1993) and the human coronaviruses pate in MS pathogenesis through an indirect mechanism (HCoV) represent candidate environmental triggers of MS whereby glial cell infection in genetically susceptible pathogenesis. The murine counterpart of HCoV, murine individuals may lead to the upregulation of pro-inflammahepatitis virus (MHV), can induce an MS-like disease in tory molecules that could participate in the destruction of genetically susceptible animals (Lane and Buchmeier, the oligodendrocyte-myelin unit observed in MS. 1997) and HCoV are both neurotropic and neuroinvasive Several inflammatory molecules including cytokines, (Arbour et al., 1999a (Arbour et al., , 1999b (Arbour et al., , 2000 . We have previously chemokines, matrix metalloproteinases (MMPs), and nitric characterized acute and persistent infections of human oxide (NO), have been associated with CNS pathologies. neural cell lines by both known viral serogroups, OC43

Moreover, a wide range of cytokines and chemokines have and 229E (Lachance et al., 1998; Arbour et al., 1999a,b) .

been detected in MS lesions. More specifically, IL-1a and Importantly, HCoV-OC43 RNA was detected in human b, IL-6, TNF-a, IFN-g, MIP-1a, and MCP-1, have probrains (Murray et al., 1992; Stewart et al., 1992; inflammatory properties that could participate in CNS demyelination (Benveniste, 1997; Munoz-Fernandez and Fresno, 1998 (MBP) within the encephalitogenic portion of the protein; previously described (Mounir and Talbot, 1992) . The fifth this would be expected to contribute to immune responses passage of HCoV-OC43 stocks kept at 2908C, having a 6 against MBP epitopes and demyelination (Chandler et al., titer of 3.5310 TCID / ml, was used in all experiments. 50 1995 Cossins et al., 1997) . Astrocytes and microglia The CHME-5 cell line consists of human fetal microglican produce NO, a non-specific inflammatory mediator of al cells immortalized by transfection with the SV40 large T oligodendrocyte death. Furthermore, active NO synthase antigen (Janabi et al., 1995) . The U-373MG cell line was was detected in astrocytes present at the level of MS obtained from ATCC and was originally derived from a plaques, at the lesion edge where demyelination occurs grade III human astrocytoma. Cells were grown in Dulbec- (Mitrovic et al., 1996) .

co's modified Eagle's medium (DMEM), supplemented In the present study, the capacity of human coronavirus with 10% (v / v) heat-inactivated fetal bovine serum (FBS) infection to modulate the expression of pro-inflammatory and 50 mg / ml gentamicin (Gibco Laboratories, Grand molecules in astrocytes and microglia was assessed. The Island, NY). targeted molecules were those associated with MS pathology (Bilzer and Stitz, 1996; Benveniste, 1997; Giovannoni 2.2. Expression of cytokines and chemokines in glial cell et al., 1997; Sarchielli et al., 1997; Link, 1998; Munoz-lines Fernandez and Fresno, 1998; Lee et al., 1999) , those detected in animal models of MS such as experimental For evaluation of mRNA expression, CHME-5 and Uautoimmune encephalomyelitis (EAE; Cross et al., 1994;  373MG cell monolayers were grown to 80% confluence on Renno et al., 1995; Benveniste, 1997; Clements et al., plastic 140320-mm Petri dishes. Positive controls for 1997; Kennedy et al., 1998) , those modulated following cytokine or chemokine expression were obtained by treatinfection by neurotropic viruses such as Theiler's virus, ing cells with 5 ng / ml phorbol myristate acetate (PMA; MHV-JHM, human immunodeficiency virus (HIV), human Sigma, Oakville, Ontario, Canada) or, for IL-12 induction, T-lymphotropic virus-1 (Bilzer and Stitz, 1996) , and with 10 ng / ml of lipopolysaccharide (LPS; Sigma). To finally those known to be expressed in glial cells (Reiling verify the influence of an exogenous source of IFN-g, cells et al., 1994; Bilzer and Stitz, 1996) .

were treated with 100 U / ml of IFN-g (Roche Diagnostics, We present in vitro data that is consistent with the Laval, Quebec, Canada). Cells were infected with HCoVhypothesis that the OC43 strain of HCoV, which we have OC43 at a multiplicity of infection (MOI) of 0.25 and shown to be neuroinvasive in humans, could be involved treatments and infections were carried out for up to 3 days in CNS pathologies by an indirect mechanism. We show at 338C in a humidified atmosphere containing 5% (v / v) that HCoV-OC43 infection alters IL-6, TNF-a, and MCP-1 CO .

2 mRNA expression in infected astrocytes and upregulates Total cellular RNA was extracted using TRIzolE NO production and MMP-2 and MMP-9 secretion in (Gibco). Five mg of RNA were reverse-transcribed accordinfected astrocytes and microglia. These results provide ing to manufacturer's instructions (Roche Diagnostics) in new insights into how infection by a neurotropic and the presence of 50 U of ExpandE Moloney murine neuroinvasive human virus might contribute to inflamma-leukemia virus reverse transcriptase (Mo-MuLV-RT), 60 U tory immunopathologies of the CNS, such as MS.

of RNAGuardE RNase inhibitor (Roche Diagnostics) and 100 pmol of oligo(dT) (Roche Diagnostics) for 90 min at 428C using 0.4 mM of deoxynucleotide triphosphates 2. Materials and methods (dNTPs)-Na salt, 13 RT buffer (50 mM Tris-HCl, pH 8.3, 40 mM KCl, 5 mM MgCl , 0.5%, v / v, Tween-20) and 10 2

The HCoV-OC43 serogroup was chosen for these studies mM dithiothreitol (DTT; Roche Diagnostics). Primers used because we have previously shown that the glial cell lines for amplification of HCoV-OC43, glyceraldehyde 3-phostested are susceptible to infection by this coronavirus phate dehydrogenase (GAPDH), cytokines and strain, which belongs to the same antigenic group as MHV, chemokines are described in Table 1 . PCR conditions were a murine coronavirus that can induce a pathology similar optimized for salt concentration, amount of primer, number to MS in mice (Lai and Cavanagh, 1997) . Since infiltrating of cycles and annealing temperatures and are described in T cells also contribute to MS pathology, and these cells are Table 2 . All primers were intron-spanning to allow poa major source of IFN-g, we also evaluated the effect of an tential DNA contamination to be assessed. A preliminary exogenous source of IFN-g on the induction of inflamma-step involved amplification of GAPDH in all samples in tory molecules in glial cells (Bilzer and Stitz, 1996) .

order to compare amplification efficiencies. The concentration of cDNA was normalized using the integrated 2.1. Virus and cell lines densitometric value (IS-1000 Digital Imaging System, Alpha Innotech Corporation, CA) obtained from GAPDH HCoV-OC43 was originally obtained from the American PCR amplification. The same concentrations were then Type Culture Collection (ATCC, Manassas, VA), plaque used in all other amplification steps. PCR amplifications purified twice, and grown on HRT-18 cells (ATCC) as were performed using 2.5 U of ExpandE high-fidelity 

PCR system DNA polymerase (Roche Diagnostics) using tion). Volume integration of signals (AlphaImagerE softthe buffer supplied by the manufacturer (10 mM Tris-HCl, ware, Canberra Packard Canada) was normalized against pH 8.3, 50 mM KCl) and 0.4 mM dNTPs-Li salt (Roche the signals obtained from untreated cells, to yield the Diagnostics). A hot start was carried out at 948C for 5 min stimulation indices. followed by 5 min at 50-608C (see Table 2 ). After addition of ExpandE polymerase, conditions used were: 1 2.3. Secretion of matrix metalloproteinases by glial cells min denaturation at 948C, 2 min annealing, 2 min elongation at 728C (annealing temperature and number of cycles

The secretion and gelatinase activity of MMP-2 and are indicated in Table 2 ), followed by a final elongation MMP-9 were detected by gelatin-based zymography, using step of 10 min at 728C. a modified version of the method of Heussen and Dowdle DNA amplicons were separated by electrophoresis on (1980) . Supernatants of cells treated with PMA, infected 1.5% (w / v) agarose gel and signals analyzed using the with HCoV-OC43, or treated with conditioned media IS-1000 Digital Imaging System (Alpha Innotech Corpora-derived from infected cells (Giraudon et al., 1996) were concentration was determined against a sodium nitrite standard, and background levels were subtracted using acellular medium. 

HCoV-OC43-infected and IFN-g-treated U-373MG astrocytic cells. chemokines MIP-1a and MCP-1 was determined by semiquantitative RT-PCR in the astrocytic and microglial cell lines. Upon HCoV-OC43 infection, we observed an upregulation in the expression of mRNA for IL-6 ( Fig. 1A) , TNF-a (Fig. 1B) and MCP-1 (Fig. 1C) in the U-373MG IFN-g or MIP-1a was detected in these cells (data not astrocytic cell line. In contrast, the CHME-5 microglial shown). cells expressed none of the molecules tested. The levels of

The kinetics of IL-6 induction after viral infection or IL-1a, IL-1b, and IL-12 were not altered following IFN-g treatment were similar, with a 3-and 5-fold infection of U-373MG cells with HCoV-OC43 or upon maximal upregulation observed at 24 h, respectively (Fig. treatment with IFN-g and no expression of IL-10, TGF-b, 1A). In both cases, IL-6 expression diminished after the 24-h time point, but remained above basal level (Fig. 1A) . led to a 2-fold induction that peaked at 12 h and remained TNF-a mRNA expression was also upregulated in U-at this level (Fig. 1C ). 373MG cells after infection or IFN-g treatment (Fig. 1B) . IFN-g treatment led to a maximal induction of 2.8-fold, 3.2. Induction of matrix metalloproteinase secretion in reached at 24 h, which then decreased to a stable level of glial cells 1.8 for as long as 72 h. Viral infection caused a steady increase in TNF-a mRNA expression which peaked at Since IL-6 and TNF-a have been reported to increase 2.4-fold at 72 h, and further time points would be required MMP activity (Giraudon et al., 1997), we evaluated the to verify whether this upward trend is maintained (Fig. secretion of MMP-2 and MMP-9 (type IV collagenases) 1B). The expression of mRNA for the MCP-1 chemokine upon viral infection. Two major bands of gelatinolysis at was also increased following either IFN-g treatment or 66 and 92 kDa were detected (Fig. 2) that were, respecviral infection. While IFN-g caused a steady increase in tively, identified as MMP-2 and pro-MMP-9 gelatinases, MCP-1 levels that reached 3.5-fold at 72 h, viral infection on the basis of their molecular mass and inhibition of activity by EDTA and 1,10-phenanthroline. In some 3.3. Induction of nitric oxide production in glial cells instances, a third band of 82 kDa was observed which corresponds to active MMP-9 cleaved from its latent form, Since nitric oxide (NO) is a pro-inflammatory molecule pro-MMP-9 / 92 kDa. While MMP-2 was constitutively that has been implicated in MS pathology (Bagasra et al., expressed in both glial cell lines, its activity was increased 1995) and astrocytes are the main producers of NO within at 24 h post-infection when compared to untreated cells.

the CNS (Lee and Brosnan, 1996) , we evaluated NO Both cell lines also expressed MMP-9. Untreated U-production following HCoV infection. Untreated U-373MG astrocytic cells secreted very low levels of pro-373MG astrocytic cells produced an average of 2.4 mM of MMP-9 which were significantly increased at 24 h post-NO at all time points tested (Fig. 3A) , presumably through infection, and a faint digestion site, corresponding to active the constitutive isoform of nitric oxide synthase (cNOS). MMP-9, was detected at 48 h post-infection ( Fig. 2A) .

Viral infection led to a 2-fold increase in NO production at Similarly, the CHME-5 microglial cells showed MMP-9 48 h, which was sustained up to 72 h (Fig. 3A) . The secretion and an active form of the enzyme at 48 h (Fig. CHME-5 microglial cells also produced NO but in lower 2B). In order to determine whether increased MMP quantities compared to that produced by U-373MG assecretion was a direct consequence of viral infection or trocytic cells (Fig. 3B) . was due to soluble factors released upon viral infection, culture supernatants from infected cells were used to treat non-infected cells. While the supernatants were non-infec-4. Discussion tious, the kinetics of gelatinase activity was similar to that observed upon HCoV-OC43 infection (data not shown).

We have shown by semi-quantitative RT-PCR the These results strongly suggest that MMP secretion is upregulation of mRNA expression for IL-6, TNF-a and mediated by soluble factors produced by infected cells.

MCP-1 following HCoV-OC43 infection or IFN-g treatment of astrocytes. Moreover, detection of gelatinase activity by zymography suggested a modulation of MMP-2 and MMP-9 activity in infected astrocytes and microglia that was mediated by soluble molecules. Finally, nitrite quantification suggested that viral infection of both cell types led to the release of NO, presumably through iNOS activation. Thus, our data clearly show that the expression of pro-inflammatory molecules (cytokines, chemokines, MMPs, and NO) that have been associated with CNS pathologies is increased in glial cells following infection by the OC43 strain of human coronavirus.

Astrocytes are known to be the main producers of IL-6 within the CNS, and its expression can be induced by cytokines such as TNF-a, IL-1b, IFN-g, and TGF-b (Bilzer and Stitz, 1996) . The upregulation of IL-6 mRNA expression observed following HCoV infection was described after Theiler's virus infection of astrocytes (Benveniste, 1998 ). The precise role of IL-6 in CNS pathologies remains unclear since it appears to have both pro-and anti-inflammatory properties through the promotion of leukocyte recruitment by induction of chemokine and adhesion molecules or the downregulation of TNF-a expression and the upregulation of soluble TNF-a receptor and IL-1R antagonist expression (Benveniste, 1998) . Given that IL-6 induction following HCoV infection did not abrogate TNF-a induction in our study, this argues that IL-6 does not repress TNF-a induction in astrocytes. In addition, IL-6 was shown to modulate the expression of MMP and TIMP (tissue inhibitor of matrix metalloproteinase; Giraudon et al., 1997) .

Infected U-373MG astrocytic cells showed a gradual relevant to CNS pathologies, given that transgenic mice overexpressing murine TNF-a in the CNS spontaneously enhanced MMP-2 and MMP-9 secretion in infected cells. developed chronic CNS inflammation and treatment with Since MMP-2 can cleave and activate MMP-9, the inanti-murine TNF-a antibody prevented demyelination creased MMP-9 activity might be attributed in part to (Probert et al., 1995) . Moreover, it has been shown that increased MMP-2 expression (Fridman et al., 1995) . TNF-a can cause oligodendrocyte death and myelin dam-Increased activity of MMP-2 was reported in other viral age in vitro (Robbins et al., 1987) . Finally, TNF-a can infections, such as upon exposure to HIV-1 gp41 peptides induce MMP and iNOS expression, which can be harmful (Chong et al., 1998) . Interestingly, we also observed an to CNS cells (Lane et al., 1996; Giraudon et al., 1997) .

upregulation of MMP-2 and MMP-9 expression in cells The contribution of TNF-a to demyelination remains treated with supernatant derived from infected cells. controversial since it has been reported not to be required MMP synthesis and activity are regulated by gene for the development of EAE (Frei et al., 1997) .

transcription and translation, secretion, activation, or Expression of the MCP-1 chemokine can be upregulated through inhibitors (Kleiner and Stetler-Stevenson, 1993) . It in CNS pathologies such as MHV-induced neurological would thus appear that HCoV-OC43 infection can influence disease in mice, EAE and MS (Glabinski and Ransohoff, the protease cascade indirectly through cellular factors 1999). Given that astrocytes are the main producers of induced following viral infection, such as IL-6 and TNF-a, MCP-1 within the CNS (Glabinski and Ransohoff, 1999) , as previously reported (Gottschall and Yu, 1995) . It would the increased expression of MCP-1 observed following therefore be of interest to evaluate the effect of these HCoV-OC43 infection might contribute to MS pathology cytokines on MMP secretion and activity in glial cells. It is by promoting leukocyte infiltration into the CNS (Calvo et clear that cytokines and possibly viral proteins can regulate al. , 1996) . Lymphocyte infiltration in the CNS is well the synthesis, secretion and activity of MMPs. The comknown to contribute to the pathology of MS (Bilzer and bined effects of pro-inflammatory cytokines and MMP Stitz, 1996) and IFN-g is an antiviral cytokine produced by secretion on the health of the oligodendrocyte-myelin unit activated Th1 and NK cells that has been detected within and their precise role in MS pathogenesis remain to be MS lesions (Munoz-Fernandez and Fresno, 1998) .

investigated. It is known, however, that MMPs can partici-Interestingly, systemic IFN-g administration worsened pate in autoimmunity by cleavage of myelin components the clinical state of MS patients (Munoz-Fernandez and and degradation of the extracellular matrix, thereby alter-Fresno, 1998). Since glial cells are not the main producers ing cell-cell connectivity. Since MMPs can sustain or of IFN-g (data not shown; Bilzer and Stitz, 1996) we direct the action of TNF-a by converting pro-TNF-a to its found it pertinent to evaluate the effect of an exogenous active form (Giraudon et al., 1996) , it would appear that a source of IFN-g, hypothetically produced by activated T self-amplification loop could exist in vivo. cells, on expression of cytokines and chemokines. In this

We have shown increased NO production by infected study, IFN-g led to increased IL-6, TNF-a, and MCP-1 astrocytes and microglia. Observations made in other mRNA expression similar to that observed following laboratories suggest that astrocytes represent the main HCoV infection. It is tempting to speculate that the source of NO in the CNS (Lee and Brosnan, 1996) , which combined effects of viral infection and lymphocyte-derived is consistent with our in vitro results. It has been shown IFN-g would have synergistic or additive effects that could that human glial cells can be induced to express iNOS contribute to inflammatory responses and MS pathology.

mRNA after exposure to pro-inflammatory cytokines such The CHME-5 microglial cells expressed none of the as IFN-g and TNF-a (Ogura and Esumi, 1996) . Thus, the cytokines and chemokines tested. This could be a result of increased NO production we observed might be attributed the transfection technique used to immortalize these cells, to TNF-a induction following infection. The observed and serves to underline the need to pursue these studies pattern of NO production suggests that it is mediated by with primary cultures. To our knowledge, this microglial the transcription of the iNOS gene, since it is known to be cell line is the only one presently available worldwide. One tightly regulated and to generate greater amounts of NO current aim of our laboratory is to develop a new microgli- (Liu et al., 1994) . Given that in situ analysis of brain al cell line, to be used as an experimental tool for further autopsies have shown iNOS immunoreactivity in the CNS studies of human coronavirus neurotropism and neuroinva-of MS patients (Bagasra et al., 1995) , viral infection might siveness.

contribute to the pathology by promoting inflammatory There is growing evidence suggesting that MMP activi-cytokine release and iNOS induction. ty, in particular MMP-9, contributes to myelin degradation

In conclusion, astrocytes and microglia respond to and promotes leukocyte trafficking across the blood-brain infection by the OC43 strain of human coronavirus by barrier (Goetzl et al., 1996; Chandler et al., 1997) . It has producing various inflammatory molecules. This argues for been suggested that IFN-b used therapeutically in MS an indirect inflammatory mechanism by which HCoVtargets MMP-9 (Stuve et al., 1996) . Also, MMPs can OC43 could be implicated in an inflammatory pathology of process pro-inflammatory cytokines, such as TNF-a and the CNS, such as MS. We have shown that HCoV-OC43 can degrade myelin basic protein to release immunogenic can influence the expression of cytokines, chemokine and peptides (Cuzner and Opdenakker, 1999) . We observed MMP and lead to an increased production of NO. These Cossins, J.A., Clements, J.M., Ford, J., Miller, K.M., Pigott, R., Vos, W., inflammatory mediators could act in concert to orchestrate Van Der Valk, P., De Groot, C.J.A., 1997. Enhanced expression of an inflammatory pathology in the CNS.

MMP-7 and MMP-9 in demyelinating multiple sclerosis lesions. Acta Neuropathol. 94, 590-598. Cross, A.H., Misko, T.P., Lin, R.F., Hickey, W.F., Trotter, J.L., Tilton,

Raised serum nitrate and nitrite levels in patients with multiple sclerosis

Acute and persistent infection of human neural cell lines Extracellular matrix-remodeling metalloproteinases and infection of bu human coronavirus OC43

Neuroinvasion by of matrix metalloproteinases (MMPs) and their natural inhibitor human respiratory coronaviruses

Activation of the Psy

USA 92, 12041-12045. receptors in CNS pathology

Matrix metalloproteinases in sclerosis and experimental allergic encephalomyelitis

Cytokines regulate gelatinase A and B

Cytokine actions in the central nervous system. (matrix metalloproteinase 2 and 9) activity in cultured rat astrocytes

Immunopathogenesis of virus diseases affecting Green

Analysis of nitrate, nitrite, and

Electrophoretic analysis of plasand OC43

Production of sulfate and copolymerized substrates

monocyte chemotactic protein-1 by rat brain macrophages

primary cultures of embryonic microglia cells with the SV40 large T Matrix metalloproteinases degrade myelin basic protein

Matrix metalloproteincephalomyelitis are regulated by differential expression of the CC ases, tumor necrosis factor and multiple sclerosis: an overview. J. chemokines macrophage inflammatory protein-1a and monocyte Neuroimmunol. 72, 155-161. chemotactic protein-1

Increased activity of matrix 897. metalloproteinase-2 in human glial and neuronal cell lines treated with Lachance

The molecular biology of coronaviruses

Murine coronavirus infection: a bined matrix metalloproteinase and tumor necrosis factor-a inhibitor. paradigm for virus-induced demyelinating disease

Expression of G., 1995. Spontaneous inflammatory demyelinating disease in transinflammatory cytokines and inducible nitric oxide synthase in brains genic mice showing central nervous system-specific expression of of SIV-infected rhesus monkeys: applications to HIV-induced central tumor necrosis factor alpha

Nitric oxide synthase: mRNA expression of different

Serum gelatinase B, TIMP-1 and TIMP-2 levels in multiple sclerosis, 1941-1944. a longitudinal clinical and MRI study

The cytokine storm in multiple sclerosis. Mult. Sclerosis. TNF-alpha expression by resident microglia and infiltrating 4, 12-15. leukocytes in the central nervous system of mice with experimental

Cytoskeletal alterations in human fetal 154, 944-953. astrocytes induced by interleukin-1 beta

An in vitro model of by stimulated astrocytes

Role of viruses in the etiopathogenesis of multiple sclerosis

The role of tumor necrosis production by mononuclear cells of patients with multiple sclerosis

Indictment of the microglia as the villain Neurobiol

Human coronavirus gene coronavirus RNA and antigen in multiple sclerosis brain. Ann. expression in the brains of multiple sclerosis patients

Nitric oxide synthase expression in human Stuve

 R.G., 1994. Aminoguanidine, an inhibitor of inducible nitric oxide synthase, ameliorates experimental autoimmune encephalomyelitis in SJL mice. J. Clin. Invest. 93, 2684-2690 This work was supported by grant MT-9203 from the