Metabotropic Glutamate Receptor-4 Modulates Adaptive Immunity and Restrains Neuroinflammation

a r t i c l e s

Metabotropic glutamate receptor-4 modulates adaptive immunity and restrains neuroinflammation

Francesca Fallarino1, Claudia Volpi1, Francesco Fazio2, Serena Notartomaso2,3, Carmine Vacca1, Carla Busceti2, Silvio Bicciato4, Giuseppe Battaglia2, Valeria Bruno2,5, Paolo Puccetti1, Maria C Fioretti1, Ferdinando Nicoletti2,5,6, Ursula Grohmann1,6 & Roberto Di Marco2,3,6

High amounts of glutamate are found in the brains of people with multiple sclerosis, an inflammatory disease marked by progressive demyelination. Glutamate might affect neuroinflammation via effects on immune cells. Knockout mice lacking metabotropic glutamate receptor-4 (mGluR4) were markedly vulnerable to experimental autoimmune encephalomyelitis (EAE, a mouse model of multiple sclerosis) and developed responses dominated by interleukin-17–producing T helper (TH17) cells. In dendritic cells (DCs) from those mice, defective mGluR4 signaling—which would normally decrease intracellular cAMP formation—biased TH cell commitment to the TH17 phenotype. In wild-type mice, mGluR4 was constitutively expressed in all peripheral DCs, and this expression increased after cell activation. Treatment of wild-type mice with a selective mGluR4 enhancer increased EAE resistance via regulatory T (Treg) cells. The high amounts of glutamate in neuroinflammation might reflect a counterregulatory mechanism that is protective in nature and might be harnessed therapeutically for restricting immunopathology in multiple sclerosis.

Classical neurotransmitters such as monoamines and acetylcholine mGluR8, which are also coupled to Gi and Go proteins in heterologous regulate the magnitude and quality of immune responses1–4. The expression systems (further detailed in Supplementary Note). recent discovery of interleukin-17 (IL-17)-producing TH17 cells Although inflammatory cytokines and glutamate are thought to con- as a distinct subset of effector cells has provided new insight into tribute to neurodegeneration in multiple sclerosis as well as in EAE, 5,6 the intricacies of immunopathology . TH17 cells have key roles early alterations of the neuronal compartment occurring in this disor- in inflammation and autoimmunity5,7 and develop under strict, der are partially independent of demyelination. How immune cells and 7,8 bidirectional influence by Treg cells . Although TH17 cells act soluble cytokines are connected temporally and causally with synaptic as mediators of autoimmunity in EAE and multiple sclerosis5,9, transmission and neurodegeneration remains elusive12. In EAE, adminis- © 2010 Nature America, Inc. All rights reserved. All rights Inc. America, Nature © 2010 several reports raise the question of whether TH1 or TH17 cells are tration of an iGluR antagonist increases oligodendrocyte survival but does the predominant pathogenic T cells. There is growing evidence that not reduce neuroinflammation13. Conversely, l-2-amino-4-phosphono- clinically similar forms of autoimmune demyelinating disease can butanoate (l-AP4), an orthosteric agonist specific for group III be driven by myelin-specific T cells of distinct lineages with vary- mGluRs, increases recovery rate in Lewis rats with EAE14. ing degrees of dependence on IL-17A production to achieve their pathological effects10. A recent report suggests that interferon-β is RESULTS effective in reducing EAE sustained by TH1 cells but exacerbates EAE is exacerbated by mGluR4 deficiency 11 −/− disease by TH17 cells . mGluR4-knockout (Grm4 ) mice and their wild-type (WT) Glutamate, the major excitatory neurotransmitter in the central counterparts were immunized with myelin oligodendrocyte glyco- nervous system (CNS), activates ligand-gated ion channels (iono- protein (MOG35–55), and EAE clinical scores were recorded daily over tropic glutamate receptors; iGluRs) as well as G-protein coupled a period of 40 d. A lack of mGluR4 was associated with earlier onset receptors (metabotropic glutamate receptors or mGluRs). mGluRs (P < 0.005), as well as more severe (P < 0.05) and ultimately fatal form a family of eight subtypes, subdivided into three groups on the disease in >40% of the hosts (Fig. 1a and Supplementary Table 1). basis of their amino acid sequence and G-protein coupling. Group I Morphologically, white matter demyelination and inflammatory includes mGluR1 and mGluR5, which are coupled to Gq protein; infiltrates were more prevalent in the spinal cord of MOG-vaccinated group II includes mGluR2 and mGluR3, which are coupled to Grm4−/− mice than in WT mice (Fig. 1b). EAE was associated with + + + Gi and Go proteins; group III includes mGluR4, mGluR6, mGluR7 and infiltration of CD4 , CD8 and B220 cells from peripheral lymphoid

1Department of Experimental Medicine and Biochemical Sciences, University of Perugia, Perugia, Italy. 2Istituto Neurologico Mediterraneo, Neuromed, Pozzilli, Italy. 3Department of Health Sciences, University of Molise, Campobasso, Italy. 4Department of Biomedical Sciences, University of Modena and Reggio Emilia, Modena, Italy. 5Department of Human Physiology and Pharmacology, University of Rome ‘La Sapienza’, Rome, Italy. 6These authors contributed equally to this work. Correspondence should be addressed to U.G. ([email protected]). Received 27 January; accepted 21 June; published online 25 July 2010; doi:10.1038/nm.2183

nature medicine VOLUME 16 | NUMBER 8 | AUGUST 2010 897 A r t i c l e s

WT Grm4–/– Figure 1 mGluR4-deficient mice are highly susceptible to EAE. Active EAE was a b −/− –/– induced in WT and mice by immunization with the MOG peptide. 5 Grm4 Grm4 (a) Clinical EAE scores (means ± s.d.) over time and linear regression analysis. 4 WT P P < 0.001 Data are from one experiment representative of three, which is also referred 3 MB to as ‘Exp. No. 2’ in Supplementary Table 1. (b) Histopathological analysis 2 100 µm of spinal cord sections from representative WT and Grm4−/− mice at 30 d 1 after immunization. Spinal cord sections were stained with H&E to assess Mean clinical score 0 inflammation and immunostained for myelin-binding protein (MBP) to evaluate H&E 0 5 10 15 20 25 30 35 40 45 myelin depletion. Arrows indicate demyelinated areas and inflammatory cellular 50 µm Time (d) after immunization infiltrates in MBP and H&E immunostaining, respectively.

organs to the CNS in both Grm4−/− and WT mice, but the percentages we did not observe any Grm6 transcripts in any of the examined cell of CD4+ and CD8+ T cells—as well as CD11b+ and CD11c+ cells— types, including brain tissue (data not shown and Supplementary were markedly higher in the CNS of Grm4−/− mice at the peak of Note), peculiar expression patterns characterized the other genes. In disease (Supplementary Table 2 and Supplementary Figs. 1–3). particular, Grm4 expression was highest in CD4+ and CD11c+ cells, When we extended our analysis to littermates from heterozygote establishing those cells as potential targets for mGluR4-mediated + − breeding—with cohorts of mice being matched for gender and age— effects. Activation of CD4 CD25 cells under TH0-, TH1- and induc- −/− +/− the disease course was also more severe in Grm4 and Grm4 mice ible Treg (iTreg)-favoring conditions led to substantial Grm4 upregu- +/+ than in Grm4 mice (Supplementary Fig. 4). lation, whereas we observed no effect in a TH2-biased environment. Grm4 expression was also upregulated in activated natural Treg EAE cytokine profile is affected by mGluR4 signaling (nTreg) cells (Fig. 3b) as well as in iTreg cells (data not shown). Gene −/− We evaluated the cytokine profile in Grm4 relative to WT mice expression was virtually absent in a TH17-biased setting (Fig. 3b). after MOG vaccination. In sorted CD4+ T cells from brain-infiltrating leukocytes (BILs), we found a significant increase in Rorc transcripts a 10 d 30 d 30 d 10 (encoding the TH17 specification factor), a reduction in Foxp3 (Treg) ** ** WT transcripts, and no change in Tbx21 (coding for Tbet; a T 1 marker) e –/– H 8 Grm4–/– Grm4 in Grm4−/− mice at the peak of neurologic signs (30 d after immuniza- 6 ** WT tion; Fig. 2a). A similar pattern was evident earlier (at 10 d) in CD4+ * 4 T cells sorted from pooled lymph nodes of mGluR4-deficient mice Foxp3 2 (Fig. 2a). No change occurred in Gata3 (a TH2 marker) in any groups mRNA fold chang (Fig. 2a). It seems that the absence of mGluR4 tipped the balance of 0 transcriptional activation in favor of inflammatory genes in response Tbx21 Gata3 Rorc Foxp3 Tbx21 Gata3 Rorc Foxp3Tbx21 Gata3 Rorc Foxp3 to MOG vaccination. LN BIL The cytokine profile in mice with EAE (high IL-17A and low Control MOG b LN

) ** ) ) transforming growth factor-β (TGF-β) from T cells and high IL-6 ) –1 ** –1 0.8 –1 –1 4 4 4 ** ** and low IL-27 from DCs) suggested that the lack of mGluR4 3 ** 3 3 ** ** 0.6 ** ** (ng ml 0.4 favored the emergence of T 17 over T cells, which would sustain 2 2 ** (ng ml 2 β

H reg γ inflammation and exacerbate clinical signs of EAE (Fig. 2b,c and 1 1 1 0.2 IL-10 (ng ml IFN- © 2010 Nature America, Inc. All rights reserved. All rights Inc. America, Nature © 2010 TGF- 0 IL-17A (ng ml 0 0 0 Supplementary Results). WT Grm4 –/– WT Grm4 –/– WT Grm4 –/– WT Grm4 –/–

+ Grm4 is expressed in accessory and adaptive immune cells CD11c cells Control MOG-vaccinated ) ) ) –1 0.4 –1 0.4 0.4 Expression of Grm4, Grm6, Grm7 and Grm8 was examined by –1 0.3 0.3 0.3 real-time RT-PCR (rRT-PCR) in WT brain cells, lymph node cells ** ** ** ** ** 0.2 * 0.2 0.2 and total splenocytes in addition to specifically sorted cell types 0.1 0.1 0.1 IL-6 (ng ml IL-12 (ng ml + + + + + IL-10 (ng ml (CD4 , CD8 , γδ, B220 , CD11b and CD11c ; Fig. 3a). Although 0 0 0 ) ) ) –1 –1 0.8 0.8 –1 ** ** Figure 2 mGluR4 deficiency alters T cell differentiation and cytokine ** 30 H 0.6 0.6 ** production. Active EAE was induced in WT and Grm4−/− mice as in Figure 1. ** 20 (ng ml ** **

0.4 β 0.4 (a) Tbx21, Gata3, Rorc and Foxp3 transcript levels in CD4+ T cells sorted 0.2 0.2 10 IL-23 (ng ml IL-27 (pg ml

from BILs at 30 d post-MOG vaccination and from lymph node (LN) cells at TGF- 0 0 0 10 and 30 d by rRT-PCR, using Gapdh (encoding glyceraldehyde 3-phosphate WT Grm4 –/– WT Grm4 –/– WT Grm4 –/– dehydrogenase) for normalization. Data (means s.d. from three experiments) ± –/– are presented as fold change in normalized transcript expression in EAE mice c BIL WT Grm4 ) ) ) ) –1 –1 relative to vehicle-injected counterparts (in which fold change = 1; dotted 2.0 –1 * –1 0.4 0.4 1.2 + * line). Inset, cytofluorometric analysis of Foxp3 in CD4 cells gated from BILs 1.5 0.3 0.3 0.9 harvested at 30 d after vaccination. (b) Cytokine production by CD4+ cells 1.0 0.2 0.2 0.6 purified from lymph nodes at 10 d after vaccination and restimulated with 0.5 0.1 0.1 0.3 IL-6 (ng ml IL-10 (ng ml IL-12 (ng ml MOG in vitro for 24 h (top) and by CD11c+ cells purified from spleens at 10 d 0 0 0 IL-17A (ng ml 0 ) ) ) ) –1

and incubated for 24 h (bottom) with medium alone (MOG-vaccinated), as –1 –1 0.8 –1 1.2 0.8 40 * compared to nonvaccinated mice (control). (c) Cytokine production by BILs 0.6 0.9 0.6 30 −/− (ng ml

purified from MOG-vaccinated WT or Grm4 mice at 30 d and incubated 0.4 (ng ml 0.6 0.4 20 β * γ for 24 h with medium alone. In b and c, data are means ± s.d. from three 0.2 0.3 0.2 10 IL-23 (ng ml IL-27 (pg ml IFN- experiments, each performed in triplicate. *P < 0.01 and **P < 0.001. 0 0 TGF- 0 0

898 VOLUME 16 | NUMBER 8 | AUGUST 2010 nature medicine a r t i c l e s

Control a Grm4 Grm7 Grm8 b Control Anti-CD3/CD28 cLPS d CpG 3 6 ** ** 5 ** Grm4 Grm4 LPS – + –– * 4 Anti-CD3/CD28 – – – + 2 4 * * 3 mGluR4

2 β-actin 1 2

1 + + Relative expression cDC CD4 CD25 0 0 0 Relative expression of + + + + + 0 1 g Relative expression of Brain LN Spleen CD4 CD8 γδ B220 CD11b CD11c 2 re cDC pDC T H T H T H 17 iT Splenic subsets T H CD4+CD25– CD4+CD25+

Figure 3 CD4+ T cells and DCs express the highest levels of mGluR4. (a) rRT-PCR analyses for Grm4, Grm7 and Grm8 in brain neural cells (brain), pooled lymph node (LN) cells, total splenocytes (spleen) and splenic CD4+, CD8+, γδ+, B220+, CD11b and CD11c+ purified cells. (b) rRT-PCR analyses for Grm4 + + + − in lymph node CD4 CD25 (nTreg) and CD4 CD25 T cells under basal conditions (control) or after 48 h incubation with antibodies to CD3 and CD28 + − (Anti-CD3/CD28). Activation of CD4 CD25 cells was performed in neutral conditions (TH0) or in the presence of the appropriate recombinant cytokine and neutralizing antibody mixtures (see Online Methods) for differentiation into TH1, TH2 or TH17 cells. (c) rRT-PCR analyses for Grm4 in cDCs or pDCs isolated from the spleen and cultured overnight in medium alone (control) or in the presence of LPS or CpG-ODN, respectively. In a–c, data are means ± s.d. of three experiments and are presented as fold change in transcript expression relative to Gapdh. *P < 0.01 and **P < 0.001 (activated versus respective control). (d) mGluR4 protein expression assayed by immunoblot analysis of whole-cell lysates from nTreg cells activated by antibodies to CD3 and CD28 and cDCs stimulated with LPS. Blots were stripped and reprobed with β-actin–specific antibody. Data are from one of three experiments.

Thus, at the T cell level, mGluR4 expression correlated positively increased Grm4 expression (Fig. 3c). Modulation of mGluR4 expres- with indicators of a potentially regulatory environment. sion in activated nTreg cells and LPS-conditioned cDCs was confirmed Searching for mGluR4+ DC subsets, we purified conventional DCs at the protein level (Fig. 3d), further supporting the hypothesis that (cDCs; CD11b+CD11chigh) and plasmacytoid DCs (pDCs; mPDCA- mGluR4 activation within an immunologic synapse contributes to 1+CD11clow) from the spleen. We found that both cDCs and pDCs cross-talk and reciprocal influence between T and accessory cells. expressed Grm4 and that the Toll-like receptor ligands lipopolysaccha- ride (LPS) for cDCs and CpG-oligonucleotide (CpG-ODN) for pDCs mGluR4 is required for DCs to orchestrate TH cell differentiation We examined any possible regulatory function of mGluR4 in the interaction between CD4+ T cells and DCs in cells cultured in vitro a cDC 2.0 WT * either singly or in combination. We observed no significant differ- –/– * 1.5 Grm4 * ences in cytokine production (Supplementary Fig. 5a) or T cell pro- * 1.0 * liferation (data not shown) between mGluR4+ and mGluR4− CD4+ ) 0.5 –1 0 T cells cultured alone. In addition, induction of Foxp3 was similar IL-6 IL-10 IL-12 IL-23 TNF-α TGF-β IL-6 IL-10 IL-12 IL-23 TNF-α TGF-β −/− + − C LPS in activated WT and Grm4 CD4 CD25 T cells in iTreg-favoring conditions (Supplementary Fig. 5b). The suppressive capacity of iT

) reg pDC cDC pDC –1 0.3 * 15 and nTreg cells was likewise comparable (data not shown). In contrast, © 2010 Nature America, Inc. All rights reserved. All rights Inc. America, Nature © 2010 IL-27 IFN-α Cytokine (ng ml * −/− * both cDCs and pDCs from Grm4 mice produced higher amounts of 0.2 * 10 * 0.1 5 IL-6 and IL-23, but less IL-12 and IL-27, than their WT counterparts 0 0 in response to LPS or CpG-ODN, respectively (Fig. 4a). We observed Cytokine (pg ml β β CCLPS CpG C CpG IL-6 IL-6 no significant phenotypic differences in immature and mature DCs IL-10 IL-23 IL-10 IL-23 C TGF- CpG TGF- between the two genotypes (data not shown). WT CD4 + WT cDC + –/– We obtained notable results in CD4 T cell–DC cocultures. Under b WT CD4 + WT CD4 + c WT CD4 + Grm4 cDC −/− WT cDC Grm4–/– cDC –/– unskewed conditions (T 0), the use of Grm4 cDCs yielded a con- ) Grm4 CD4 + WT cDC H –/– –/– 40.9% 11.1% 12.8% 4.7% Grm4 CD4 + Grm4 cDC 60 siderable increase in total T cell recovery (Supplementary Fig. 6). The ** + + 50 ** percentage of IL-17A CD4 T cells increased (P < 0.001), whereas the 13.9% 23.5% 40 * –/– –/– Grm4 CD4 + Grm4 CD4 + 30 * –/– WT cDC Grm4 cDC 20 * Figure 4 The absence of mGluR4 in DCs favors emergence of T 17 cells. 37.4% 8.5% 9.7% 4.2% * H 10 (a) Splenic cDCs and pDCs from WT or Grm4−/− mice were incubated γ 0 with medium alone (C) or in the presence of LPS (cDCs) or CpG-ODN 9.1% 21.4% Cytokine-producing cells (% IL-17A IFN-γ IL-17A + IFN-γ IFN- (pDCs). After 24 h, cytokine concentrations were measured in culture IL-17A supernatants. (b) IL-17A and IFN-γ expression in CD4+ T cells, purified WT CD4 + WT cDC from WT or Grm4−/− lymph nodes and activated with CD3- and CD28- d WT CD4 + Grm4–/– cDC –/– * specific antibodies in the presence of cDCs purified from spleens of 5 * Grm4 CD4 + WT cDC 50 ) * ) * –/– –/– –1 –1 Grm4 CD4 + Grm4 cDC either genotype. Numbers in dot plots indicate the percentage (at 4 d) 4 * 40 * * of IL-17A+, IFN-γ+ and IL-17A+IFN-γ+ cells in gated CD4+ cells. One * 3 30 experiment representative of four. (c) Mean percentages (± s.d.) 2 * * 20 of the four experiments, one of which is depicted individually in b. * * 1 10 *P < 0.001 and **P < 0.0001. (d) Cytokine concentration was measured Cytokine (pg ml Cytokine (ng ml in supernatants from cocultures set up as in b. In a and d, data are means 0 0 IFN-γ IL-17A IL-2 IL-6 IL-10 IL-12 IL-23 TGF-β IL-27 ± s.d. of three experiments, each performed in triplicate. *P < 0.01.

nature medicine VOLUME 16 | NUMBER 8 | AUGUST 2010 899 A r t i c l e s

** ** Figure 5 mGluR4-dependent signaling controls cAMP formation and a ** ** * ** cytokine production. (a) Intracellular cAMP formation, measured in cDCs ** * 1.6 Control ** stimulated with forskolin (FSK) or LPS for 10 min in the presence or FSK 1.2 absence of l-AP4 or PHCCC. (b) Cytokine concentrations in supernatants LPS 0.8 from WT or Grm4−/− cDCs stimulated with LPS for 24 h in the presence 0.4 of PHCCC (WT) or KT 5720 (Grm4−/−). All data are means ± s.d. of three (nmol per well) cAMP formation 0 experiments, each performed in triplicate. *P < 0.01 and **P < 0.001. – 5 –– ––5 5 –– ––5 5 –– – 5 L-AP4 (µM) –– 2.5 10 40 40 –– 2.5 10 40 40 –– 2.5 10 40 40 PHCCC (µM) −/− 3.0 25 cocultures involving Grm4 cDCs (Fig. 4d). Because IL-27 is known b WT cDCs Control 2.5 ** LPS 20 15 2.0 * to counter the effect of IL-6 in directing TH17 cell development , an * 15 16 1.5 ** effect that can limit EAE , the decrease in IL-27 during activation of 1.0 ** ** 10 ) ) + 0.5 5 naive CD4 T cells might favor the emergence of T 17 cells. –1 –1 H 0 0 PHCCC –– + + –– + + – – + + – – + + – – + + –– + + –/– mGluR4 signaling modulates cAMP in dendritic cells 3.0 ** Grm4 cDCs 25 2.5 20 Pathways leading to increased cAMP formation in DCs (as occurs 2.0 15 1.5 * 17,18 Cytokine (ng ml ** Cytokine (pg ml in response to LPS ) decrease cytokines associated with T 1 1.0 ** ** ** 10 H 0.5 5 responses (that is, IL-12 and IL-27) and increase TH17-associated 0 0 KT 5720 –– + + –– + + – – + + – – + + – – + + –– + + cytokines (IL-6 and IL-23). In presynaptic nerve terminals and micro- IL-6 IL-10 IL-12 IL-23 TGF-β IL-27 glia, mGluR4 activation lowers intracellular cAMP formation in a Gi protein-dependent fashion19. We challenged WT DCs with forskolin, percentage of IFN-γ+CD4+ T cells—a portion of which also expressed which stimulates adenylyl cyclase, or with LPS in the presence of IL-17A—was significantly reduced (P < 0.0001) following culture with l-AP4 or N-phenyl-7-(hydroxyimino)cyclopropa[b]chromen-1a- mGluR4-deficient cDCs but not T cells (Fig. 4b,c). In line with this carboxamide (PHCCC). l-AP4 is an orthosteric agonist of all group data, we observed no additive effects when cocultures consisted of III mGluRs, whereas PHCCC acts as a selective enhancer (that is, a DCs and T cells both deficient in mGluR4 expression (Fig. 4b,c). positive allosteric modulator) of mGluR4 (Supplementary Note)20. It seems that the effect of mGluR4 deficiency was wholly depend- Although l-AP4 and PHCCC had no effect on basal cAMP levels, ent on DCs. Measurements of cytokine concentrations in culture both drugs reduced the increase in intracellular cAMP induced in supernatants showed decreased amounts of TH1-associated IL-2 in cDCs by either forskolin or LPS (Fig. 5a and Supplementary Results).

–/– Vehicle PHCCC a WT + Vehicle b WT + PHCCC Grm4 + PHCCC d WT + PHCCC 4 4 4 4 * 4 LN CD4 –/– 3 3 * 3 3 3 ** Grm4 + vehicle 2 ** 2 2 5 2 2 Grm4–/– + PHCCC 1 1 1 1 1 0 0 0 0 0

4 MBP IL-4 IL-10 IL-17A IFN-γ TGF-β 3 100 µm 0.3 0.3 * 0.6 ** * ** 1.5

) 60 2 0.2 * 0.2 0.4 DC –1 40 1.0 1 0.1 0.1 0.2 20 0.5 0 0 0 0 0 Mean clinical score 0 IL-6 IL-12 IL-23 IL-27 TGF-β

H&E ** 0 5 10 15 20 25 30 35 40 45 1.5 © 2010 Nature America, Inc. All rights reserved. All rights Inc. America, Nature © 2010 ** 0.3 0.3 * 0.6 50 µm 1.0 * Time (d) after immunization Cytokine (ng ml 0.2 0.2 0.4 0.5 0.1 0.1 0.2

0 0 0 0 BIL c IL-6 IL-10 IL-12 IL-17A 1.2 * 1.5 ** Vehicle ** 0.15 ** 30 * 10 Vehicle 0.8 PHCCC ** 0.10 20 1.0 PHCCC 0.05 10 0.4 0.5 8 * 0 0 0 0 ** IFN- TGF- 6 ** ** IL-23 IL-27 γ β Foxp3 4 e f Vehicle → CD4+CD25– 5 + – e mRNA fold change 2 PHCCC → CD4 CD25 2.5 Vehicle

+ e 4 Vehicle → CD4 PHCCC + 2.0 0 PHCCC → CD4 Tbx21 Gata3 Rorc Foxp3 Tbx21 Gata3 Rorc Foxp3 3 LN BIL 1.5 2 1.0 Figure 6 PHCCC attenuates EAE. (a) Clinical EAE scores (means ± s.d.) 1 over time of WT and Grm4−/− mice vaccinated with MOG on day 0 0.5 Mean clinical scor 0 and treated subcutaneously daily with PHCCC from day 1. Data are Mean clinical scor 0 ******** *** representative of two experiments with similar results and correspond to 0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20 25 30 35 40 45 50 the experiment shown in Supplementary Table 3. (b) Histopathological Time (d) after cell transfer Time (d) after immunization analysis of spinal cord sections taken 30 d after vaccination from representative WT mice treated as in a. (c) Transcript levels of the indicated genes evaluated in CD4+ T cells sorted from lymph nodes at 10 d and BILs at 30 d after vaccination as in Figure 2a. Inset, Foxp3 expression in gated CD4+ BILs (30 d). (d) Cytokine production by lymph node CD4+ cells (10 d), splenic DCs (10 d) and BILs (30 d) from WT mice treated with PHCCC or vehicle as in a. Measurements (ng ml−1 with the exception of IL-27, pg ml−1) were performed at 24 h after restimulation with MOG in vitro (lymph node CD4+ cells) or after incubation with medium alone (DCs and BILs). Data are means ± s.d. of three experiments, each performed in triplicate. *P < 0.01 and **P < 0.001. (e) EAE induced in WT mice by adoptive transfer of CD4+ cells purified at 18 d after vaccination from lymph nodes and splenic cells of WT or Grm4−/− mice treated as in a and restimulated with MOG in vitro. In two groups (CD4+CD25−), CD25+ cells were depleted in vitro before transfer (n = 9). (f) Clinical RR-EAE (means ± s.d.) scores in SJL/J mice treated daily with PHCCC at the onset of the first clinical RR-EAE attack (see also Supplementary Table 5). *P < 0.05. In a and f, control mice were treated with vehicle alone. In a, b, e and f, data are representative of two experiments with similar results.

900 VOLUME 16 | NUMBER 8 | AUGUST 2010 nature medicine a r t i c l e s

PHCCC also increased production of IL-12, IL-27 and TGF-β in LPS- recognition by myelin-reactive T cells21. Moreover, DCs modulate the treated but not untreated DCs, whereas the production of IL-6, IL-10 recruitment of encephalitogenic and Treg cells into CNS tissue. This and IL-23 was reduced (Fig. 5a). capacity is influenced by DC surface expression of co-stimulatory A similar pattern was reproduced by using mGluR4-deficient DCs or inhibitory molecules. The fact that DCs accumulate in the CNS in combination with KT 5720, a specific inhibitor of protein kinase A before T cells and can direct T cell responses suggests that they are (PKA) (Fig. 5b). In contrast, PHCCC had no effect on cytokine release key determinants of CNS autoimmune outcomes21. in LPS-treated Grm4−/− DCs or in WT DCs transfected with Grm4- Although synaptic communication and plasticity require glutamate, specific siRNA (Supplementary Figs. 7 and 8). Likewise, we observed excessive glutamate release and signaling through iGluRs results in impaired PHCCC effects in WT DCs treated with 8-bromo-cAMP, a excitotoxicity, neural death and loss of myelin-producing oligodendro- standard PKA activator (Supplementary Fig. 8). cytes22,23. Synaptic alterations are a feature of neuroinflammatory diseases12. These alterations are largely mediated by inflammatory mGluR4 signaling contributes to EAE resistance in vivo cytokines released by infiltrating T cells and activated microglia and Repeated in vivo administration of PHCCC starting 1 d after MOG are responsible, at least in part, for irreversible dendritic pathology12. immunization (Fig. 6a, Supplementary Fig. 9 and Supplementary High amounts of glutamate are found in the brains of individuals Table 3), but not after the onset of neurological signs (Supplementary with multiple sclerosis24–26, and gluamate in addition to inflammatory Fig. 10), significantly attenuated disease in WT but not Grm4−/− mice cytokines, is a major contributor to neurodegeneration in multiple (P < 0.005; Fig. 6b and Supplementary Fig. 11). The effect was dose sclerosis as well as in EAE. dependent and observable within a range of doses far below those Like other neurotransmitters1–4, owing to the diverse nature and resulting in toxicity (Supplementary Fig. 12). Although no signifi- functions of its receptors, glutamate might affect neuroinflamma- cant change was observed in the relative proportion of immune cells tion via specific effects on cells of the immune system27,28. In this populations in BILs after PHCCC treatment, a substantial increase study, we found that EAE was exacerbated in mice lacking mGluR4, did occur in the proportion of CD4+CD25+ and CD4+Foxp3+ cells and the immune response was dominated by IL-17–producing in lymph nodes (Supplementary Table 4). In both BILs and lymph TH17 cells. Treatment of WT mice with a selective mGluR4 enhancer + nodes, the protective effect of PHCCC in WT mice was associated induced a protective response dominated by CD4 Treg cells. with a transition to a response dominated by Foxp3+CD4+ T cells and Constitutive expression of mGluR4 occurred in all peripheral DCs + TH1 cells, and a reduction in TH17-mediated immunity (Fig. 6c,d and in specific CD4 T cell subtypes, including Treg cells but not and Supplementary Table 4). A protective effect of Treg cells was TH17 cells. Notably, an absence of mGluR4 in DCs—rather than + + also evident in EAE adoptive transfer experiments with CD4 cells CD4 T cells—tipped the balance of TH cell differentiation in favor from PHCCC-treated WT mice, which would transfer disease of the TH17 phenotype. only when CD25+ cells were depleted (Fig. 6e and Supplementary Studies in both multiple sclerosis and EAE have demonstrated Fig. 13). Although the effects of PHCCC were attenuated but still an association between the development of demyelinating plaques 29–31 significant (P = 0.002, drug versus vehicle) on discontinuing drug and the accumulation of TH17 cells in the CNS and periphery . treatment (Supplementary Fig. 14), our data suggest that the use of Our data show that glutamate exerts immunoregulatory effects by an mGluR4 enhancer supports the development of Treg cells, whose acting on the DC and affecting TH cell development and adaptive 7 regulatory activity may affect TH17 cells . The use of bone marrow immunity. Effector functions of T lymphocytes are under the con- chimeras further demonstrated that the protective effects of mGluR4 trol of DCs, which are capable of integrating a variety of incoming signaling occur mostly in the periphery, by recruitment of myeloid cells signals and directing the downstream response32,33. Recent evidence © 2010 Nature America, Inc. All rights reserved. All rights Inc. America, Nature © 2010 expressing Grm4 (Supplementary Fig. 15). Although PHCCC treat- indicates an inhibitory role for GABA in autoimmune inflamma- ment had no effect on EAE once neurological symptoms had devel- tion, whereby GABAergic agents act directly on DCs, decreasing oped in the setting of WT mice treated with MOG (Supplementary signaling events and diminishing subsequent adaptive inflammatory Fig. 10), parallel data in a model of relapsing-remitting responses to myelin proteins34. In our model system, activation of EAE (RR-EAE) revealed that PHCCC reduced the number and sever- mGluR4 was found to modulate cAMP concentrations and cytokine ity of relapses when administered during the first clinical attack responses to inflammatory signals and inhibit TH17 cell develop- (Fig. 6f and Supplementary Table 5). ment in response to myelin antigens. Although unable to restore a Whereas these data provide the first experimental evidence that gluta- durable tolerant state, the effects of mGluR4 signaling contributed mate acts at the interface between the nervous and immune systems, to redirecting the TH response in mice treated with a positive allo the cytokine-modulating activity by the selective mGluR4 enhancer steric modulator. could also be demonstrated in human DCs (Supplementary Fig. 16), in During inflammation plasma glutamate concentrations rise to which expression of Grm4 and other class III members was assessed by levels that are expected to saturate mGluR4 (>40 μM)24,35. Under Affymetrix profiling of public data sets (Supplementary Fig. 17). those conditions, activation of mGluR4 by endogenous glutamate might limit an otherwise uncontrolled TH17 cell response, DISCUSSION and PHCCC might reinforce this negative control by amplifying Multiple sclerosis is a chronic, immune-mediated demyelinating dis- stimulation of mGluR4 by glutamate. The high levels of glutamate ease of the CNS. Clinical and histopathological features suggest an in neuroinflammation might exert an effect that is protective in inflammatory etiology involving resident CNS innate cells as well as nature and could be exploited therapeutically in multiple scle- invading adaptive immune cells. Encephalitogenic myelin-reactive rosis. Second-generation mGluR4 enhancers with improved T cells have been implicated in the initiation of an inflammatory cas- pharmacodynamics are being investigated for the treatment cade, eventually resulting in demyelination and axonal damage. DCs of Parkinson’s disease36. The use of these molecules might pave the may represent key modulators of this immunopathological cascade. In way to innovative therapies aimed at rescuing a proper TH cell EAE, CNS microvessel-associated DCs are essential for local antigen balance in neuroinflammation.

nature medicine VOLUME 16 | NUMBER 8 | AUGUST 2010 901 A r t i c l e s

Methods 14. Besong, G. et al. Activation of group III metabotropic glutamate receptors inhibits the production of RANTES in glial cell cultures. J. Neurosci. 22, 5403–5411 Methods and any associated references are available in the online (2002). version of the paper at http://www.nature.com/naturemedicine/. 15. Neufert, C. et al. IL-27 controls the development of inducible regulatory T cells and Th17 cells via differential effects on STAT1. Eur. J. Immunol. 37, 1809–1816 Note: Supplementary information is available on the Nature Medicine website. (2007). 16. Batten, M. et al. Interleukin 27 limits autoimmune encephalomyelitis by suppressing Acknowledgments the development of interleukin 17–producing T cells. Nat. Immunol. 7, 929–936 This work was supported by Fondazione Italiana Sclerosi Multipla Project (2006). 17. Schnurr, M. et al. Extracellular nucleotide signaling by P2 receptors inhibits IL-12 No. 2008/R/2 (to G.B., U.G. and R.D.M.). We thank G. Andrielli for digital art and and enhances IL-23 expression in human dendritic cells: a novel role for the cAMP image editing, P. Scarselli for technical support and S. Iacobelli for statistical advice. pathway. Blood 105, 1582–1589 (2005). 18. Li, K. et al. Cyclic AMP plays a critical role in C3a-receptor–mediated regulation AUTHOR CONTRIBUTIONS of dendritic cells in antigen uptake and T-cell stimulation. Blood 112, 5084–5094 F. Fallarino designed and performed experiments. C. Volpi, F. Fazio, S.N., C. Vacca (2008). and C.B. performed experiments. S.B. analyzed microarray data. G.B. and V.B. 19. Conn, P.J. & Pin, J.P. Pharmacology and functions of metabotropic glutamate contributed to experiment design. P.P. and M.C.F. supervised research. F.N. receptors. Annu. Rev. Pharmacol. Toxicol. 37, 205–237 (1997). contributed to experiment design and supervised research. R.D.M. designed 20. Maj, M. et al. (–)-PHCCC, a positive allosteric modulator of mGluR4: characterization, mechanism of action and neuroprotection. 45, 895–906 experiments, supervised research and analyzed data. U.G. designed experiments, Neuropharmacology (2003). supervised research and wrote the manuscript. 21. Zozulya, A.L., Clarkson, B.D., Ortler, S., Fabry, Z. & Wiendl, H. The role of dendritic cells in CNS autoimmunity. J. Mol. Med. 88, 535–544 (2010). COMPETING FINANCIAL INTERESTS 22. Matute, C., Domercq, M. & Sanchez-Gomez, M.V. Glutamate-mediated glial injury: The authors declare no competing financial interests. mechanisms and clinical importance. Glia 53, 212–224 (2006). 23. McDonald, J.W., Althomsons, S.P., Hyrc, K.L., Choi, D.W. & Goldberg, M.P. Published online at http://www.nature.com/naturemedicine/. Oligodendrocytes from forebrain are highly vulnerable to AMPA/kainate receptor– Reprints and permissions information is available online at http://npg.nature.com/ mediated excitotoxicity. Nat. Med. 4, 291–297 (1998). reprintsandpermissions/. 24. Stover, J.F. et al. Neurotransmitters in cerebrospinal fluid reflect pathological activity. Eur. J. Clin. Invest. 27, 1038–1043 (1997). 25. Srinivasan, R., Sailasuta, N., Hurd, R., Nelson, S. & Pelletier, D. Evidence of 1. Woiciechowsky, C. et al. Sympathetic activation triggers systemic interleukin-10 release elevated glutamate in multiple sclerosis using magnetic resonance spectroscopy at in immunodepression induced by brain injury. Nat. Med. 4, 808–813 (1998). 3 T. Brain 128, 1016–1025 (2005). 2. Steinman, L. Elaborate interactions between the immune and nervous systems. Nat. 26. Newcombe, J. et al. Glutamate receptor expression in multiple sclerosis lesions. Immunol. 5, 575–581 (2004). Brain Pathol. 18, 52–61 (2008). 3. Jutel, M. et al. Histamine regulates T-cell and antibody responses by differential 27. Pacheco, R. et al. Glutamate released by dendritic cells as a novel modulator of expression of H1 and H2 receptors. Nature 413, 420–425 (2001). T cell activation. J. Immunol. 177, 6695–6704 (2006). 4. Tracey, K.J. Reflex control of immunity. Nat. Rev. Immunol. 9, 418–428 (2009). 28. Storto, M. et al. Expression of metabotropic glutamate receptors in murine 5. Langrish, C.L. et al. IL-23 drives a pathogenic T cell population that induces thymocytes and thymic stromal cells. J. Neuroimmunol. 109, 112–120 (2000). autoimmune inflammation. J. Exp. Med. 201, 233–240 (2005). 29. Lock, C. et al. Gene-microarray analysis of multiple sclerosis lesions yields new 6. Ivanov, I.I. et al. The orphan nuclear receptor RORγt directs the differentiation targets validated in autoimmune encephalomyelitis. Nat. Med. 8, 500–508 program of proinflammatory IL-17+ T helper cells. Cell 126, 1121–1133 (2006). (2002). 7. Bettelli, E. et al. Reciprocal developmental pathways for the generation of pathogenic 30. Haak, S. et al. IL-17A and IL-17F do not contribute vitally to autoimmune neuro- effector TH17 and regulatory T cells. Nature 441, 235–238 (2006). inflammation in mice. J. Clin. Invest. 119, 61–69 (2009). 8. Zhou, L., Chong, M.M. & Littman, D.R. Plasticity of CD4+ T cell lineage 31. Segal, B.M. Th17 cells in autoimmune demyelinating disease. Semin. Immunopathol. differentiation. Immunity 30, 646–655 (2009). 32, 71–77 (2010). 9. Cua, D.J. et al. Interleukin-23 rather than interleukin-12 is the critical cytokine for 32. Banchereau, J. & Steinman, R.M. Dendritic cells and the control of immunity. autoimmune inflammation of the brain. Nature 421, 744–748 (2003). Nature 392, 245–252 (1998). 10. Goverman, J. Autoimmune T cell responses in the central nervous system. Nat. 33. Grohmann, U. et al. Functional plasticity of dendritic cell subsets as mediated by Rev. Immunol. 9, 393–407 (2009). CD40 versus B7 activation. J. Immunol. 171, 2581–2587 (2003). 11. Axtell, R.C. et al. T helper type 1 and 17 cells determine efficacy of interferon-β 34. Bhat, R. et al. Inhibitory role for GABA in autoimmune inflammation. Proc. Natl. in multiple sclerosis and experimental encephalomyelitis. Nat. Med. 16, 406–412 Acad. Sci. USA 107, 2580–2585 (2010).

© 2010 Nature America, Inc. All rights reserved. All rights Inc. America, Nature © 2010 (2010). 35. Pampliega, O. et al. Association of an EAAT2 polymorphism with higher glutamate 12. Centonze, D. et al. The link between inflammation, synaptic transmission and concentration in relapsing multiple sclerosis. J. Neuroimmunol. 195, 194–198 neurodegeneration in multiple sclerosis. Cell Death Differ. 17, 1083–1091 (2008). (2010). 36. Conn, P.J., Christopoulos, A. & Lindsley, C.W. Allosteric modulators of GPCRs: a 13. Pitt, D., Werner, P. & Raine, C.S. Glutamate excitotoxicity in a model of multiple novel approach for the treatment of CNS disorders. Nat. Rev. Drug Discov. 8, sclerosis. Nat. Med. 6, 67–70 (2000). 41–54 (2009).

902 VOLUME 16 | NUMBER 8 | AUGUST 2010 nature medicine ONLINE METHODS inhibitor of PKA44) was used at the final concentration of 300 nM; 8-bromo- Mice. Grm4+/− (B6.129-Grm4tm1Hpn/J) mice, on a C57/BL6 background, were cAMP (8-Br-cAMP, Sigma-Aldrich; a membrane-permeable analog of cAMP purchased from The Jackson Laboratory. The Grm4−/− offspring of hetero- and activator of PKA45) was used at 100 μM. zygotes were used to establish colonies of Grm4−/− mice. Although characterized by altered spatial learning and memory, Grm4−/− mice do not show Flow cytometry and real-time reverse transcription PCR, small interfering any gross motor abnormalities or alterations of fine motor coordination37. RNA synthesis and transfection. We performed flow cytometry, rRT-PCR, C57BL/6 mice 4- to 10-weeks-old, referred to as WT controls, were purchased siRNA synthesis and transfection as previously described41,42,46. For details, see from Charles River Breeding laboratories. Littermates (Grm4+/+, Grm4+/− and the Supplementary Methods and Supplementary Table 6. Grm4−/−) from the heterozygote (Grm4+/−) breeding were included in selected experiments. All mice used in these studies were genotyped by PCR of DNA Western blot analysis and determination of cytokines, intracellular cAMP isolated from tail clippings. Ten-week-old female SJL/J mice (Charles River and extracellular glutamate concentrations. We performed all of these analyses Breeding laboratories) were used in RR-EAE experiments. All in vivo studies as previously described41,42,47–49. For details, see the Supplementary Methods were in compliance with national (Italian Approved Animal Welfare Assurance and Supplementary Table 7. A-3143-01), Perugia University and Neuromed Institute Animal Care and Use Committee guidelines. Meta-analysis of dendritic cell gene expression data. We performed this analysis as previously described50 and outlined in the Supplementary Methods. Experimental autoimmune encephalomyelitis induction and treatment with PHCCC in vivo. Induction and evaluation of MOG-induced EAE was done Statistical analyses. EAE data were analyzed by the nonparametric Mann– as outlined in the Supplementary Methods and as previously described38,39. Whitney U test. To compare mean day of onset and maximal score, Mann- PHCCC (3 mg per kg body weight; Tocris Cookson) was administered daily Whitney’s rank-sum test was used51. Linear regression analysis of the individual subcutaneously starting 1 d after MOG immunization. For adoptive transfer disease curves in MOG-induced EAE was carried out using the mean clinical of EAE, mice were immunized with MOG, then CD4+ T lymphocytes were score as the dependent variable and time (d) as the independent variable. Paired purified from spleens and lymph nodes, restimulated and transferred into WT data, including mean daily clinical scores in RR-EAE, were evaluated by Student’s recipients as described in detail in the Supplementary Methods. RR-EAE was t test. All in vitro determinations are means ± s.d. from at least three independent induced in SJL/J mice by vaccination with proteolipid protein peptide 139–151 experiments, unless otherwise indicated. All n values were computed by power (Genemed Synthesis). PHCCC (3 mg per kg body weight) was administered analysis to yield a power of at least 80% with an α-level of 0.05. daily subcutaneously at the onset of the first clinical RR-EAE attack (see also Supplementary Methods). Mice were monitored daily, and neurological effects Additional methods. Detailed methodology is described in the Supplementary were scored as follows: 0, no signs of disease; 1, flaccid tail; 2, inability to right; Methods. 3, paralysis of one hind limb; 4, paralysis of both hind limbs; and 5, moribund or death. The induction of EAE in bone marrow chimeras was performed as 37. Pekhletski, R. et al. Impaired cerebellar synaptic plasticity and motor performance in mice lacking the mGluR4 subtype of metabotropic glutamate receptor. outlined in Supplementary Methods. J. Neurosci. 16, 6364–6373 (1996). 38. Fazio, F. et al. Switch in the expression of mGlu1 and mGlu5 metabotropic glutamate Histology and immunohistochemistry. At 30 d after MOG immunization, receptors in the cerebellum of mice developing experimental autoimmune spinal cords were removed, embedded in paraffin, cut at 5 μM and stained by encephalomyelitis and in autoptic cerebellar samples from patients with multiple sclerosis. 55, 491–499 (2008). μ Neuropharmacology H&E to reveal CNS inflammatory infiltrates. For immunohistochemistry, 5- M 39. Orabona, C. et al. Enhanced tryptophan catabolism in the absence of the molecular deparaffinized sections were stained with rabbit polyclonal MBP-specific anti- adapter DAP12. Eur. J. Immunol. 35, 3111–3118 (2005). body, mouse monoclonal antibody to class II major histocompatibility complex, 40. Fallarino, F. et al. The combined effects of tryptophan starvation and tryptophan rat monoclonal antibody to CD4 or rabbit polyclonal antibody to CD8. For catabolites down-regulate T cell receptor ζ-chain and induce a regulatory phenotype in naive T cells. J. Immunol. 176, 6752–6761 (2006). details, see Supplementary Methods. 41. Grohmann, U. et al. Reverse signaling through GITR ligand enables dexamethasone to activate IDO in allergy. Nat. Med. 13, 579–586 (2007). © 2010 Nature America, Inc. All rights reserved. All rights Inc. America, Nature © 2010 Leukocyte isolation and stimulation. Purification of CD4+, CD4+CD25− and 42. Romani, L. et al. Defective tryptophan catabolism underlies inflammation in mouse CD4+CD25+ T cells from pooled lymph nodes and DCs from spleens, either chronic granulomatous disease. Nature 451, 211–215 (2008). 43. Grohmann, U. et al. CTLA-4–Ig regulates tryptophan catabolism in vivo. Nat. unfractionated or fractionated into cDCs and pDCs, was carried out as previ- Immunol. 3, 1097–1101 (2002). 40–43 + 6 ously described . For cytokine induction, CD4 cells (1 × 10 per well) were 44. Huang, Y.Y., Martin, K.C. & Kandel, E.R. Both protein kinase A and mitogen- cultured in the presence of 2 × 105 irradiated, T cell–depleted splenocytes (as activated protein kinase are required in the amygdala for the macromolecular antigen-presenting cells) and 10 μM MOG peptide for 2 d. For in vitro differen- synthesis–dependent late phase of long-term potentiation. J. Neurosci. 20, + − −1 6317–6325 (2000). tiation, purified naive CD4 CD25 T cells were activated for 5 d with 1 μg ml 45. Chen, T.C., Hinton, D.R., Zidovetzki, R. & Hofman, F.M. Up-regulation of the cAMP/ of plate-bound CD3- and CD28-specific antibodies in the presence of various PKA pathway inhibits proliferation, induces differentiation, and leads to apoptosis combinations of recombinant cytokines and blocking antibodies as detailed in malignant gliomas. Lab. Invest. 78, 165–174 (1998). in the Supplementary Methods. Cytokine production was measured in DC 46. Fallarino, F. et al. Therapy of experimental type 1 diabetes by isolated Sertoli cell xenografts alone. J. Exp. Med. 206, 2511–2526 (2009). culture supernatants harvested after 24-h cell incubation with medium 47. Ngomba, R.T. et al. Positive allosteric modulation of metabotropic glutamate 4 −1 −1 alone, 1 μg ml LPS (for cDCs) or 10 μg ml CpG-ODN (for pDCs). The (mGlu4) receptors enhances spontaneous and evoked absence seizures. Supplementary Methods contain details. Purification of mouse splenic CD11b+ Neuropharmacology 54, 344–354 (2008). and B220+ cells, lymph node γδ T cells, BILs and human CD11c+ cells was 48. Corti, C., Aldegheri, L., Somogyi, P. & Ferraguti, F. Distribution and synaptic localisation of the metabotropic glutamate receptor 4 (mGluR4) in the rodent CNS. performed as outlined in Supplementary Methods. Neuroscience 110, 403–420 (2002). 5 + − In T cell–DC cocultures, 5 × 10 CD4 CD25 T cells were incubated with 49. Matrisciano, F. et al. Defective group-II metaboropic glutamate receptors in the 1.6 × 105 unfractionated DCs (ratio 3:1) for 48–72 h in the presence of 2.5 μg ml−1 hippocampus of spontaneously depressed rats. Neuropharmacology 55, 525–531 soluble antibody to CD3 (145-2C11, BD Pharmingen), as previously described40. (2008). 50. Bisognin, A. et al. A-MADMAN: annotation-based microarray data meta-analysis For in vitro cell treatments, PHCCC was dissolved in DMSO at an initial con- tool. BMC Bioinformatics 10, 201 (2009). centration of 10 mM and diluted in medium to a final concentration of 40 μM, 51. Fleming, K.K. et al. Statistical analysis of data from studies on experimental unless specified otherwise; KT 5720 (Sigma-Aldrich; a specific, cell-permeable autoimmune encephalomyelitis. J. Neuroimmunol. 170, 71–84 (2005).

doi:10.1038/nm.2183 nature medicine