Neurotherapeutics (2021) 18:920–937 https://doi.org/10.1007/s13311-021-01016-7 ORIGINAL ARTICLE Selected Clostridia Strains from The Human Microbiota and their Metabolite, Butyrate, Improve Experimental Autoimmune Encephalomyelitis Laura Calvo‑Barreiro1,2,3 · Herena Eixarch1,2,3 · Thais Cornejo2,4 · Carme Costa1,2,3 · Mireia Castillo1,2,3 · Leyre Mestre3,5 · Carmen Guaza3,5 · María del Carmen Martínez‑Cuesta6 · Takeshi Tanoue7 · Kenya Honda7 · Juan José González‑López2,4 · Xavier Montalban1,2,3 · Carmen Espejo1,2,3 Accepted: 26 January 2021 / Published online: 7 April 2021 © The Author(s) 2021 Summary Gut microbiome studies in multiple sclerosis (MS) patients are unravelling some consistent but modest patterns of gut dysbiosis. Among these, a significant decrease of Clostridia cluster IV and XIVa has been reported. In the present study, we investigated the therapeutic effect of a previously selected mixture of human gut-derived 17 Clostridia strains, which belong to Clostridia clusters IV, XIVa, and XVIII, on the clinical outcome of experimental autoimmune encephalomyelitis (EAE). The observed clinical improvement was related to lower demyelination and astrocyte reactivity as well as a ten- dency to lower microglia reactivity/infiltrating macrophages and axonal damage in the central nervous system (CNS), and to an enhanced immunoregulatory response of regulatory T cells in the periphery. Transcriptome studies also highlighted increased antiinflammatory responses related to interferon beta in the periphery and lower immune responses in the CNS. Since Clostridia-treated mice were found to present higher levels of the immunomodulatory short-chain fatty acid (SCFA) butyrate in the serum, we studied if this clinical effect could be reproduced by butyrate administration alone. Further EAE experiments proved its preventive but slight therapeutic impact on CNS autoimmunity. Thus, this smaller therapeutic effect highlighted that the Clostridia-induced clinical effect was not exclusively related to the SCFA and could not be reproduced by butyrate administration alone. Although it is still unknown if these Clostridia strains will have the same effect on MS patients, gut dysbiosis in MS patients could be partially rebalanced by these commensal bacteria and their immunoregulatory proper- ties could have a beneficial effect on MS clinical course. Key Words Gut Microbiota · Clostridia Strains · Short-chain Fatty Acid · Immune Regulation · Experimental Autoimmune Encephalomyelitis · Multiple sclerosis 5 * Carmen Espejo Grupo de Neuroinmunología, Departamento de carmen.espejo@vhir.org Neurobiología Funcional y de Sistemas, Instituto Cajal, CSIC, Avda. Doctor Arce 37, 28002 Madrid, Spain 1 Servei de Neurologia‑Neuroimmunologia, Vall d’Hebron 6 Department of Food Biotechnology and Microbiology, Institut de Recerca, Hospital Universitari Vall d’Hebron, Instituto de Investigación en Ciencias de la Alimentación, Centre d’Esclerosi Múltiple de Catalunya, Pg. Vall d’Hebron CIAL (CSIC-UAM), C/ Nicolás Cabrera 9, 28049 Madrid, 119‑129, 08035 Barcelona, Spain Spain 2 Universitat Autònoma de Barcelona, 7 RIKEN Center for Integrative Medical Sciences, 1‑7‑22 08193 Bellaterra Cerdanyola del Vallès, Spain Suehiro‑cho, Tsurumi‑ku, Yokohama, Kanagawa 230‑0045, 3 Red Española de Esclerosis Múltiple (REEM), Instituto de Japan Salud Carlos III, Ministerio de Economía y Competitividad, Fondo de Investigación Sanitaria, Madrid, Spain 4 Servei de Microbiologia, Hospital Universitari Vall d’Hebron, Pg. Vall d’Hebron 119‑129, 08035 Barcelona, Spain Selected Clostridia Strains from The Human Microbiota and their Metabolite, Butyrate, Improve… 921 Introduction were found in the serum of Clostridia-treated mice. Further in vivo assays with the SCFA butyrate proved its preventive Multiple sclerosis (MS) is an autoimmune, degenerative, effect on CNS autoimmunity but a slight therapeutic clini- chronic, and demyelinating disease that affects the cen- cal impact on EAE clinical course. Our results suggest that tral nervous system (CNS) [1]. As a complex disease, MS the therapeutic effect performed by this commensal bacteria depends on both genetic and environmental factors and it population was not exclusively related to the SCFA and could not is thought to be developed in genetically predisposed indi- be reproduced by butyrate administration alone. Although viduals after exposure to different environmental factors. it is still unknown whether these 17 Clostridia strains will Ground breaking data from research in experimental auto- have the same effect on MS patients, previously-defined gut immune encephalomyelitis (EAE) models and subsequent dysbiosis regarding Clostridia cluster IV and XIVa could studies in MS patients have connected gut microbiota to be compensated by bacteria administration and their immu- innate and adaptive immune responses characteristic of noregulatory properties could have a beneficial effect on MS MS pathogenesis [2]. clinical course. Gut microbiome studies in MS patients are unravel- ling some consistent but modest patterns of gut dysbio- sis compared to healthy controls [3–9]. Among these gut microorganisms associated to MS disease course, a signifi- Methods cant decrease of Clostridia cluster IV and XIVa has been reported [5]. On the other hand, previous in vivo studies Mice have shown that a cocktail of 17 Clostridia strains isolated from a human faecal sample, which partially belong to C57BL/6JOlaHsd 8-week-old female mice purchased previously-mentioned clusters, promoted both the number from Envigo (Venray, Holland) were used. Mice were and function of regulatory T (Treg) cells in the colon and housed under standard light- and climate-controlled successfully suppressed experimental colitis in mice [10]. conditions, and standard chow and water were provided Subsequent analysis of caecal contents from Clostridia- ad libitum. All experiments were performed in strict treated mice showed a higher concentration of short-chain accordance with EU (Directive 2010/63/EU) and Spanish fatty acids (SCFAs): acetate, propionate, butyrate, and regulations (Real Decreto 53/2013; Generalitat de Cata- isobutyrate [10]. Simultaneously, it was demonstrated lunya Decret 214/97). The Ethics Committee on Animal that both propionate and butyrate promoted the genera- Experimentation of the Vall d’Hebron Research Institute tion of splenic Treg cells in vivo [11]. Similarly, naïve approved all procedures described in this study (protocol ­CD4+ T cells treated with propionate and butyrate in vitro number: 35/15 CEEA). promoted Treg cell populations and, more important, a propionate-rich diet ameliorated the clinical course of Induction and Assessment of EAE EAE mice due to an increase in Treg cell population and interleukin (IL)-10 production [12]. Keeping this in mind, Anaesthetised mice were immunised by subcutaneous it is reasonable that the reduction of these Clostridia clus- injection of 100 μl of phosphate-buffered saline contain- ters may be associated with the pathogenesis of MS due to ing 200 μg of peptide 35–55 from myelin oligodendrocyte their intrinsic immunoregulatory properties depending, to glycoprotein ­(MOG35–55, Proteomics Section, Universitat a greater or lesser extent, on SCFA production. Pompeu Fabra, Barcelona, Spain) emulsified in 100 μl of In the present study, we investigated the therapeutic effect complete Freund’s adjuvant [incomplete Freund’s adjuvant of a previously selected mixture of 17 Clostridia strains from (IFA, F5506, Sigma-Aldrich, St. Louis, MO, USA) contain- the human microbiota on the clinical outcome of estab- ing 4 mg/ml Mycobacterium tuberculosis H37RA (231141, lished EAE. We observed a clinical improvement that was BD Biosciences, BD, Franklin Lakes, NJ, USA)]. At 0 and related to lower demyelination and astrocyte reactivity as 2 days postimmunisation (dpi), mice were intravenously well as a trend to lower microglia reactivity/infiltrating mac- injected with 250 ng of pertussis toxin (P7208, Sigma- rophages and axonal damage in the CNS, and to an enhanced Aldrich). Mice were weighed and examined daily for neu- immunoregulatory response of Treg cells in the periphery. rological signs in a blinded manner and corrective meas- Likewise, microarrays studies highlighted increased anti- ures and endpoint criteria to ensure EAE-incident animals inflammatory responses related to interferon beta (IFN-β) welfare were included as previously detailed [13]. All data production in the periphery and lower activation, differentia- presented are in accordance with the ARRIVE guidelines tion, and proliferation of immune cells in the CNS. Finally, for animal research and with the guidelines suggested for higher levels of the immunoregulatory SCFA butyrate EAE publications [14]. 922 L. Calvo-Barreiro et al. Motor Function Assessment of a 200 µl volume containing 100 mg of sodium butyrate or water (vehicle) via oral gavage was performed once daily The day before the end of the experiment (27 or 34 dpi), until the end of the experiment (28 dpi). motor performance was evaluated using a Rotarod apparatus (Ugo Basile, Gemonio, Italy) that was set to accelerate from Ex vivo Splenocyte Proliferative Capacity a speed of 4 to 40 rotations per minute in a 300-s time trial [13]. Briefly, once mice were placed on the rotating cylinder, Splenocyte suspensions were prepared by grinding spleens the amount of time (sec) that the animals walked on the cyl- through a 70-μm nylon cell strainer at the end of the experi- inder without falling was recorded. Each mouse was given ment (28 or 35 dpi). Splenocytes were seeded at 2 × ­105 cells four trials, and the value for each mouse depicts the mean of per well in supplemented X-VIVO™ 15 medium (BE02- these four measurements. 060F, Lonza, Basel, Switzerland) within 96-well plates as previously detailed [13]. Splenocyte cultures were stimu- lated with 5 μg/ml M ­ OG35-55 (antigen-specific stimulus) Bacterial Strains and Treatments or 5 μg/ml phytohaemagglutinin-L (PHA-L, L2769, Sigma- Aldrich) (polyclonal stimulus) and compared to non-stimulated A consortium of 17 Clostridia strains was originally iso- (control) condition. After 54 h, 75 μl of supernatant were lated from a stool sample of a Japanese healthy volunteer harvested and stored at − 80 °C to further assess cytokine as described previously [10]. The bacterial isolates were secretion. At the same time, 75 μl of supplemented medium individually cultured in BD Schaedler Agar with Vitamin containing 1 μCi of ­[3H]-thymidine (NET027Z, PerkinElmer, K1 and 5% Sheep Blood (254042, BD) in an anaerobic jar Waltham, MA, USA) were added to each well. Splenocyte (Biomérieux, Marcy l’Etoile, France) under strict anaerobic cultures were maintained under the same conditions for an conditions at 37ºC during three to four days. To prepare the additional 18 h and incorporated radioactivity was measured bacterial mixture, Clostridia strains were again individu- in a beta-scintillation counter (Wallac, Turku, Finland). Five ally grown in BD BBL Thioglycollate Medium (221742, replicates per experimental condition ­(MOG35-55, PHA-L, BD) in an anaerobic jar under strict anaerobic conditions at and control) were assessed per mouse. Stimulation indices 37ºC until they reached confluence. Later, they were mixed were calculated by dividing the mean counts per minute at equal amounts of media volume in an anaerobic cham- (cpm) of ­MOG35-55 or PHA-L condition by the mean cpm ber and stored in ready-to-use individual cryotubes supple- of the control condition. The results are shown as the stimulation mented with 5% glycerol at -80ºC until use. index. Preventive approach: Since oral gavage administration diminished EAE incidence in a pilot study as a preven- Cytokine Detection by Luminex tive approach (data not shown), the butyrate treatment was administered via water feeding bottle. Seven days before Cytokine secretion pattern [granulocyte–macrophage col- mice immunisation, a 50 ml water volume containing 1.1 g ony-stimulating factor (GM-CSF), IFN-γ, IL-2, IL-4, IL-6, of sodium butyrate (200 mM, 303410, Sigma-Aldrich) or IL-10, IL-12p70, IL-17A, IL-21, IL-22, IL-23 and tumour not supplemented (vehicle) was prepared daily and added necrosis factor alpha (TNF-α)] was assessed in the superna- in the water feeding bottle of the corresponding cage. This tants of ­MOG35-55-stimulated splenocytes by using a Pro- concentration of butyrate was selected taking into account cartaPlex Multiplex Immunoassay (Invitrogen, Carlsbad, previous research on SCFA propionate in the EAE model CA, USA), according to the manufacturer’s instructions. [12]. The following day, the remaining volume was meas- Samples were acquired in a Magpix instrument (Luminex ured and the intake of sodium butyrate per mouse was esti- Corporation, Austin, TX, USA) and data were analysed with mated daily. Daily treatment was performed until the end of a ProcartaPlex Analyst software (Thermo Fisher Scientific, the experiment (35 dpi). Waltham, MA, USA). The results are shown as the concen- Therapeutic approach: After attaining a clinical score tration of each cytokine (pg/ml). equal to or greater than 1 (between 12 and 15 dpi), mice were randomised into clinically equivalent experimental groups. Flow Cytometry Analysis Administration of a 250 µl volume containing Clostridia strains or Thioglycollate Medium, supplemented with 5% Spleen cell suspensions were prepared as described pre- glycerol (vehicle) via oral gavage was performed once daily viously. Cell subsets were analysed using fluorochrome- until the end of the experiment (28 dpi). Regarding butyrate conjugated monoclonal antibodies (mAbs) after discrimina- treatment, after attaining a clinical score equal to or greater tion of dead cells by Fixable Viability Stain (BD Pharmin- than 1 (between 13 and 15 dpi), mice were randomised into gen). For analysis of the Treg cell population, CD3ε (BD clinically equivalent experimental groups. Administration Biosciences Cat# 553,061, RRID: AB_394594), CD4 Selected Clostridia Strains from The Human Microbiota and their Metabolite, Butyrate, Improve… 923 (BD Biosciences Cat# 561,090, RRID:AB_10562560), Biosciences Cat# 560,689, RRID:AB_1727506) to assess and CD25 (Thermo Fisher Scientific Cat# 12–0251-81, proper ex vivo stimulation. Samples were acquired in a RRID:AB_465606) were used. FoxP3 intracellular stain- CytoFLEX flow cytometer (Beckman Coulter, Brea, CA, ing was performed using fluorochrome-labelled anti- USA) and data were analysed with CytExpert 2.3 software FoxP3 mAb (Thermo Fisher Scientific Cat# 17–5773-80, (Beckman Coulter). RRID:AB_469456). CD39 (Thermo Fisher Scientific Cat# 25–0391-80, RRID:AB_1210767), CD62L (BD Biosciences Histological Analysis Cat# 563,252, RRID:AB_2738098), HELIOS (BD Bio- sciences Cat# 563,801, RRID:AB_2738428), and ICOS Preventive approach: At 35 dpi, spinal cords of euthanised (BD Biosciences Cat# 564,592, RRID:AB_2738858) were mice were collected, fixed in a 4% paraformaldehyde solu- also selected and evaluated in the Treg cell population. For tion, embedded in paraffin, and cut into 4-μm thick coronal dendritic cell subpopulations, mAbs specific for B220 (BD sections. Demyelination was assessed by Klüver-Barrera Biosciences Cat# 553,088, RRID:AB_394618), CD11b staining and inflammatory infiltration by haematoxylin and (BD Biosciences Cat# 562,605, RRID:AB_11152949), eosin staining, as previously described [15]. Images were CD11c (BD Biosciences Cat# 553,802, RRID:AB_395061), acquired using a Leica DFC550 fluorescence microscope CD8a (Thermo Fisher Scientific Cat# 47–0081-80, (Leica Microsystems, Wetzlar, Germany) and LAS V4.5 RRID:AB_1272221), and CD317 (Thermo Fisher Scien- visualisation software (Leica Microsystems), obtained at a tific Cat# 17–3172-82, RRID:AB_10596356) were used. magnification of 10X, and analyzed using ImageJ software. For analysis of macrophage, neutrophils, and MDSC For every staining, one mosaic image from the thoracic spi- populations, CD11b (BD Biosciences Cat# 562,605, nal cord was selected per mouse and evaluated in a blinded RRID:AB_11152949), CD206 (BioLegend Cat# 141,712, manner. The results are shown as the individual score per RRID:AB_10900420), F4/80 (Thermo Fisher Scientific mouse. Cat# 12–4801-82, RRID:AB_465923), Ly6C (BD Bio- Therapeutic approach: At 28 dpi, spinal cords of euthan- sciences Cat# 560,593, RRID:AB_1727557), and Ly6G ised mice were collected, fixed in a 4% paraformaldehyde (BD Biosciences Cat# 551,460, RRID:AB_394207) were solution, embedded in paraffin, and cut into 4-μm thick used; for B cell subsets, B220 (BD Biosciences Cat# coronal sections. Demyelination was assessed by rabbit 552,772, RRID:AB_394458), CD1d (BD Biosciences polyclonal anti-myelin basic protein (MBP, Millipore Cat# Cat# 553,846, RRID:AB_2073521), CD5 (BD Biosciences AB980, RRID:AB_92396); T cells, by anti-CD3 (Agilent Cat# 550,035, RRID:AB_398457), CD19 (BD Biosciences Cat# A0452, RRID:AB_2335677); axonal damage, by Cat# 560,143, RRID:AB_1645234), CD138 (BD Bio- mouse monoclonal purified anti-neurofilament H, nonphos- sciences Cat# 562,610, RRID:AB_11153126), and MHCII phorylated antibody (SMI32, BioLegend Cat# 801,701, (BD Biosciences Cat# 562,363, RRID:AB_11153297) RRID:AB_2564642); and reactive microglia/infiltrat- were used. For analysis of T cell activation status, CD3ε ing macrophages and astroglia reactivity, by lectin from (BD Biosciences Cat# 553,062, RRID:AB_394595), Lycopersicon esculentum (LEA, L0401, Sigma-Aldrich) CD4 (Thermo Fisher Scientific Cat# 46–0042-80, and mouse monoclonal anti-glial fibrillary acidic protein RRID:AB_1834432), CD8a (BD Biosciences Cat# (GFAP, Sigma-Aldrich Cat# C9205, RRID:AB_476889), 563,152, RRID:AB_2738030), CTLA-4 (BD Biosciences respectively. Immunofluorescences were carried out as pre- Cat# 564,331, RRID:AB_2738751), PD-1 (BioLegend viously described [13]. Images were acquired using a Leica Cat# 135,217, RRID:AB_10900085), and TIM-3 (Thermo AF6000 fluorescence microscope (Leica Microsystems) and Fisher Scientific Cat# 25–5870-82, RRID:AB_2573483) LAS AF visualisation software (Leica Microsystems), and were used. For intracellular cytokine determination, ex vivo mosaic images were obtained at a magnification of 20X and stimulation of splenocytes was performed as previously analysed using ImageJ software as previously detailed [13]. detailed [13]. Then, CD3ε (BD Biosciences Cat# 553,061, RRID:AB_394594), CD4 (Thermo Fisher Scientific Cat# In vivo Intestinal Permeability 46–0042-80, RRID:AB_1834432), and CD8a (BD Bio- sciences Cat# 563,152, RRID:AB_2738030) staining was At 28 dpi, EAE mice were weighed and orally gavaged with performed. Cytokine intracellular staining was performed an isotonic solution of 0·9% NaCl with fluorescein sodium using fluorochrome-labelled anti-IFN-γ (BD Biosciences salt (NaF, F6377, Sigma-Aldrich) at 10 μg/g mouse body Cat# 563,376, RRID:AB_2738165), anti-IL-4 (BD Bio- weight or without NaF (negative control mice). After 1 h, sciences Cat# 554,436, RRID:AB_398556), anti-IL-10 mice were euthanised, blood samples were collected in hep- (BD Biosciences Cat# 563,276, RRID:AB_2738111), arinised tubes, and NaF concentration in plasma was meas- and anti-IL-17A (BD Biosciences Cat# 561,020, ured by spectrophotofluorimetry in a Thermo Scientific App- RRID:AB_10584331) mAbs, and anti-CD69 mAb (BD liskan (Thermo Fisher Scientific) as previously described 924 L. Calvo-Barreiro et al. [13, 16]. NaF standard concentrations were used as reference 3,800 rpm for 5 min and the organic layer was separated and both samples and calibration curve were performed in and dried under nitrogen stream. The extracts were reconsti- duplicate. tuted in 1 ml water:methanol (50:50) and 3 µl were injected into the liquid chromatography tandem mass spectrometry Transcriptome Studies (LC–MS/MS) system. The LC–MS/MS system consisted in an Acquity ultra-performance liquid chromatography At 28 dpi, spinal cords of euthanised mice were collected, system (Waters, Milford, MA, USA) coupled to a triple immersed in liquid nitrogen, and stored at -80ºC until use. quadrupole mass spectrometer provided with an electro- Total RNA was isolated from spinal cords using QIAzol lysis spray interface (Xevo TQ-S micro, Waters). The chroma- reagent (79306, Qiagen, Hilden, Germany) and RNeasy tographic separation was performed in an Acquity UPLC Mini Kit (74104, Qiagen) with on-column DNase digestion BEH C18 column (Waters). Concentrations of SCFA were (79254, Qiagen) in order to remove any genomic DNA trace obtained by external calibration. Four quality control sam- according to the manufacturer’s instructions. Regarding sple- ples, which consisted of water spiked with known amounts nocytes, splenocyte suspensions were prepared as described of the analytes, were injected in the same batch as the sam- previously at the end of the experiment (28 dpi), aliquoted in ples. Additionally, four blank samples were included in fetal bovine serum (FBS, Biowest, Nuaillé, France) supple- the batch to evaluate the ubiquitous contribution of each mented with 10% dimethyl sulfoxide (DMSO), and stored in metabolite. Samples were processed using the TargetLynx liquid nitrogen until use. RNeasy Mini Kit together with on- software (Waters). column DNase digestion was also used to isolate splenocyte total RNA according to the manufacturer’s instructions. Prior to cDNA generation with WT Pico Reagent Kit-HT (Applied Statistics Biosystems, Thermo Fisher Scientific), RNA quality was ana- lysed by capillary electrophoresis (Bioanalyzer 2100, Agilent, When two experimental groups were compared in an only Santa Clara, CA, USA). Subsequently, Clariom S Pico Assay independent experiment, all study variables were analysed HT, mouse arrays were processed according to the manufac- using the Student t-test. When variances were equal or turer’s instructions in a GeneTitan Multi-Channel instrument different between experimental groups, Satterthwaite or (Applied Biosystems, Thermo Fisher Scientific). Pooled method was used, respectively. All analyses were carried out with the Proc TTEST procedure. All tests were Determination of SCFA Levels in Serum two-tailed, and statistical significance was set at a p value of < 0·05. When two or more experimental groups were Mice were euthanised and the total blood volume was col- compared in several independent experiments, all study var- lected by cardiac puncture in non-heparinised tubes at the iables were analysed using the differences of least-squares end of the experiment (28 dpi). Blood samples were kept means. Normal or lognormal distributions were assumed for on ice until further processing. Next, blood samples were all studied variables depending on data distribution. When centrifuged at 3,500 revolutions per minute (rpm) and 4ºC repeated measures within mice were performed, compound for 15 min to separate the serum. Finally, serum samples symmetry was used as the covariance structure. When only were obtained and stored at -80ºC until use. Approximately a potential clustering effect of the experiments was present, 200 µl of accurately measured serum samples were individ- a variance components structure showed an acceptable fit. ually mixed with 30 µl of the internal standard (consisting in All analyses were carried out with the Proc MIXED pro- 100 µg/mL of 4-methylvaleric acid in methanol) and 300 µl gram, except for analysis of lognormal distribution vari- of acetonitrile. After protein precipitation, the mixtures were ables, for which Proc GLIMMIX was used. All tests were centrifuged at 5,000 rpm and the supernatant transferred to two-tailed, and statistical significance was set at a p value a clean microtube and derivatised with o-benzyl hydroxy- of < 0·05. lamine as previously reported [17]. The derivatised analytes For the transcriptome studies, the statistical analysis of were detected by a selected reaction monitoring (SRM) the data was performed using the statistical language R (R method. Briefly, 50 µl of N-(3-dimethylaminopropyl)-N’- version 3.6.1, Copyright© 2018 The R Project for Statisti- ethylcarbodiimide hydrochloride 1 M were mixed with 50 µl cal Computing) and the libraries developed for the micro- of o-benzyl hydroxylamine 1 M and added into the mixture. array analysis in the Bioconductor Project (https://​www.​ Extracts were left at room temperature and shaken every bioco​nduct​or.​org). The analysis of biological significance 15 min in a vortex. After one hour, 1 ml of ultrapure water has been based on GSEA on different annotation databases was added. Derivatised extracts were transferred to a clean [18]. The analysis has been performed over two annotation 10 ml tube and mixed with 4 ml of ethyl acetate. After vig- databases: the "Gene Ontology" (GO) and the Reactome orous shaking for one minute, samples were centrifuged at Pathway Knowledge base [19]. Selected Clostridia Strains from The Human Microbiota and their Metabolite, Butyrate, Improve… 925 Data Sharing test was performed to assess motor coordination skills at 27 dpi. No statistically significant differences were observed Completed metadata worksheet, raw data, and processed regarding weight loss (p = 0·190) (Fig. 1c) or the Rotarod data from the microarray studies were deposited in GEO test (Vehicle: 14·37 ± 10·26 s, n = 15, vs. Clostridia strains: database: GSE155438. 23·03 ± 19·16 s, n = 15, p = 0·361) (Fig. 1d). Finally, no differences were observed regarding intestinal permeabil- ity in the Clostridia strains group (1281·66 ± 525·50 ng of Results NaF/ml, n = 9, p = 0·280) compared to the vehicle group (1516·88 ± 349·60 ng NaF/ml, n = 9) (Fig. 1e). Clostridia Administration Improves the EAE Clinical Outcome as a Therapeutic Approach Pathological Signs are Reduced in the CNS of Clostridia‑Treated EAE Mice Mice were treated once daily with Clostridia strains or vehicle after attaining a clinical score equal to or greater Within the spinal cord white matter, demyelination was than 1 and being randomised into clinically equivalent lower in mice treated with Clostridia strains (6·19 ± 2·44%, experimental groups, from 12–15 dpi to the end of the n = 9, p = 0·017) than in vehicle-treated mice (11·10 ± 6·66%, experiment (28 dpi). Treatment with Clostridia strains n = 9) (Fig. 2a-c). CNS inflammation was evaluated measur- improved the clinical outcome [area under the curve (AUC): ing T cell inflammatory infiltrate density and reactive micro- 48·00 ± 15·41, n = 15, p = 0·017] compared to vehicle treat- glia and infiltrating macrophages presence. No differences ment (AUC: 59·38 ± 8·09, n = 15) (Fig. 1a,b). Mice were were observed in T cell inflammatory infiltrate (p = 0·364) weighed daily to monitor their well-being and the Rotarod (Fig. 2d-f) but a tendency to lower reactive microglia and Fig. 1  Clostridia Strains Improve the Clinical Outcome of EAE were weighed and orally gavaged with an isotonic solution of NaF Mice as a Therapeutic Approach. C57BL/6JOlaHsd mice were at 10 μg/g mouse body weight. After 1 h, mice were euthanised, and immunised by subcutaneous injection of ­MOG35–55 and weighed and plasma samples were collected. Intestinal permeability was evalu- examined daily for neurological signs. After attaining a clinical score ated by measuring the NaF concentration in mouse plasma and no equal to or greater than 1 and being randomised (12–15 dpi), mice were differences were observed between the experimental groups e. The treated once daily with Clostridia strains (250 μl confluent Clostridia graph presents the results of a representative experiment (n = 9 mice culture) or vehicle (250 μl thioglycollate medium), weighed and per group). Moreover, higher levels of the SCFA butyrate are pre- examined daily for neurological signs in a blinded manner until the sented in the serum of Clostridia-treated mice f. The graph presents end of the experiment (28 dpi). Daily oral therapeutic treatment with the results of a representative experiment (n = 6 mice per group). The Clostridia strains improves the clinical score a and the AUC b and data are presented as the means ± standard errors of the mean a, c or does not affect weight loss c or motor coordination skills d compared the means ± standard deviations b, d, e, f. Black circle, Vehicle; white to vehicle group. The charts present the combined results of two inde- circle, Clostridia strains. Abbreviations: NaF: fluorescein sodium salt pendent experiments (n = 15 mice per group). At 28 dpi, EAE animals 926 L. Calvo-Barreiro et al. infiltrating macrophages was observed in the Clostridia- significant differences were observed regarding PD-1 treated group (1·52 ± 0·43%, n = 9, p = 0·068) compared expression level (data not shown). Further study of the T to vehicle-treated mice (2·03 ± 0·81%, n = 9) (Fig. 2g-i). cell activation status (CTLA-4, LAG-3, and TIM-3) did Regarding astrocyte reactivity, a reduction was observed in not show differences between experimental groups (data the Clostridia-treated mice (1·06 ± 0·39%, n = 9, p = 0·041) not shown). Moreover, Clostridia-treated mice presented compared to the vehicle group (1·50 ± 0·42%, n = 9) (Fig. 2j- increased levels of CD25 activation marker in ­CD4+ T lym- l). Finally, the level of axonal damage was also lower in mice phocytes (1646·43 ± 59·14 MFI, n = 9, p = 0·002) compared treated with the Clostridia strains (0·73 ± 0·26%, p = 0·052) with vehicle-treated animals (1541·57 ± 64·04 MFI, n = 9) than in the vehicle group (0·99 ± 0·25%, n = 9), although no in the periphery (Fig. 5b). No differences were observed statistical significance was reached (Fig. 2m-o). between experimental groups regarding B cell and myeloid cell [dendritic cells (DCs), macrophages, neutrophils, and Clostridia Strains Increase Antigen‑Specific myeloid-derived suppressor cells (MDSCs)] populations Proliferation but do not Alter Disease‑Related (data not shown). Cytokines Clostridia Strains Reduce Immunological Responses Whereas an increase in antigen-specific proliferation was in the CNS and Promote an IFN‑β Response observed under Clostridia strains treatment (stimulation in the Periphery index: 6·66 ± 3·41, n = 15, p = 0·030) compared to the vehicle group (stimulation index: 4·78 ± 2·21, n = 13) (Fig. 3a), no Prior to studying the differentially expressed genes in our differences were detected in the polyclonal immune response target tissues: spinal cord and splenocytes, different types of (Fig. 3b). In addition, the cytokine secretion pattern in the quality controls were performed to confirm that every array supernatants of the antigen-specific stimulated splenocytes was suitable for normalisation process and, once normal- did not differ between these two experimental groups and ised, that it was adequate for differential expression analysis. some of them (IL-12p70, IL-21 and IL-23) were even not Later, a batch factor was included in the linear model used detected (Supp. Figure 1). Likewise, when we specifically for differential expression analysis when required. In order measured the intracellular cytokine (IFN-γ, IL-4, IL-10, and to increase statistical power and reduce unnecessary noise, IL-17A) production patterns in T helper (Th) and cytotoxic those genes whose standard deviation was below the 65 per- T cells from the periphery, no differences were found (data centile of all standard deviations were removed. A list of 7, not shown). 104 genes was included in the differential expression analy- sis. After correcting for multiple testing by the Benjamini Clostridia Strains Increase the Immunoregulatory and Hochberg method [20], no genes were found differ- Capacity of Treg Cells entially expressed in Clostridia-treated mice compared to vehicle mice in either of our target tissues with an adjusted Immunoregulatory responses are implicated in the resolution p value of 0·25 or below. Thus, no individual gene met the of inflammation and can potentially alleviate clinical signs threshold for statistical significance possibly because the in the EAE model. In this sense, the increase in the ­CD62L+ relevant biological differences are modest in relation to the Treg cell population in mice treated with Clostridia strains noise of the microarray technology. In order to overcome (29·13 ± 5·73%, n = 9, p = 0·055) compared to vehicle- this setback, we performed a gene set enrichment analysis treated mice (23·06 ± 6·66%, n = 9) could be related to the (GSEA) to determine whether members of a given Gene Set observed clinical improvement (Fig. 4a), although no sta- correlate with the phenotypic class distinction between our tistical significance was reached. Moreover, the expression experimental groups (Clostridia- and vehicle-treated mice) of the inducible molecules CD62L and CD25 was aug- within the list of selected genes (7,104 genes) [18]. mented in Treg cells of mice treated with Clostridia strains Clostridia strains reduced the activation of the immune (CD62L: 3741·50 ± 353·03 MFI, n = 9, p = 0·001; and CD25: response in general and T cells in particular, the lymphocyte 1688·18 ± 75·23 MFI, n = 9, p = 0·001) compared to that in differentiation, and the mononuclear cell proliferation (T cells the vehicle group (CD62L: 3169·08 ± 268·67 MFI, n = 9; and lymphocytes, specially) in the CNS of EAE mice [adjusted and CD25: 1551·92 ± 68·57 MFI, n = 9) (Fig. 4b,c). No dif- p < 0·05 and normalised enrichment score (NES) < -2·3] ferences were observed between the experimental groups in (Fig. 6a). Besides these biological processes (GO database), the rest of the studied Treg cell subsets (data not shown). Clostridia treatment was also related to the reduction of sev- On the other hand, Clostridia-treated mice presented a eral pathways (Reactome Pathway Knowledge base) related decrease in the PD-1+ Th cell population (17·75 ± 2·37%, to the immune response: immunoregulatory interactions n = 9, p = 0·044) compared to vehicle-treated mice between lymphoid and non-lymphoid cells, cell surface inter- (21·31 ± 4·27%, n = 9) (Fig. 5a), although no statistically actions at the vascular wall, innate immune system, and IL-2 Selected Clostridia Strains from The Human Microbiota and their Metabolite, Butyrate, Improve… 927 928 L. Calvo-Barreiro et al. ◂Fig. 2  Clostridia Strains Ameliorate Histopathological signs in compared to vehicle treatment (AUC: 83·95 ± 31·28, n = 5) the CNS of EAE Mice. At 28 dpi, spinal cords of euthanised EAE as a preventive approach (Fig. 7a,b). Butyrate intake not mice were collected, fixed in a 4% paraformaldehyde solution, embed- ded in paraffin, and cut into 4-μm thick coronal sections. Demyelina- only affected clinical score but also modified the maximum tion was assessed by anti-myelin basic protein (MBP) antibody; T score (2·20 ± 2·05, n = 5, p = 0·049) compared to vehicle cell infiltrates, by anti-CD3 antibody; reactive microglia/infiltrating treatment (4·50 ± 0·87, n = 5). As expected, butyrate-treated macrophages and astroglia reactivity, by lectin from Lycopersicon mice also reflected their clinical improvement in other esculentum and anti-glial fibrillary acidic protein (GFAP) antibody, respectively; and axonal damage, by anti-neurofilament H, nonphos- clinical variables such as the accumulated weight loss (AUC: phorylated (SMI32). Daily therapeutic administration of Clostridia 318·30 ± 187·50, n = 5, p = 0·033) and motor coordination strains reduces demyelination a-c and reactive astroglia j-l, does not skills (40·02 ± 15·00 s, n = 5, p = 0·069) compared to vehicle modify T cell infiltrates d-f and tends to diminish reactive microglia/ group (AUC: -152·90 ± 364·50, n = 5; and motor coordina- infiltrating macrophages g-i and axonal damage m–o. The graphs pre- sent the results of a representative experiment (n = 9 mice per group). tion skills: 18·65 ± 14·60 s, n = 4) (Fig. 7c,d). Arrows indicate ­CD3+ cells. Scale bar indicate 100 µm. The data are Within the spinal cord white matter, histopathologi- presented as the means ± standard deviations. Abbreviations: DAPI: cal signs such as demyelination (demyelination score: 4′,6-diamidino-2-phenylindole, GFAP: glial fibrillary acidic protein, 0·80 ± 0·45, n = 5, p = 0·002) (Fig. 7e) and CNS inflamma- LEA: lectin from Lycopersicon esculentum, MBP: myelin basic pro- tein, SMI32: neurofilament H, nonphosphorylated tion (cell inflammation score: 2·20 ± 0·84, n = 5, p = 0·034) (Fig. 7f) were diminished in the butyrate-treated mice com- pared to the vehicle group (demyelination score: 2·50 ± 0·58, family signaling (adjusted p < 0·05 and NES < -2). Regarding n = 4; and cell infiltration score: 3·50 ± 0·58, n = 4). the biological effect of Clostridia treatment in the periphery, On the other hand, antigen-specific ­(MOG 35–55) and Clostridia-treated mice showed an increased defence response polyclonal (PHA-L) immune responses were not altered by to other organisms that was partially related to the innate the preventive administration of the SCFA butyrate immune response, the IL-1 production, and the response to (stimulation index: 4·62 ± 2·32, n = 5, p = 0·394 and IFN-β (adjusted p < 0·1 and NES > 2) (Fig. 6b). This defence 31·43 ± 8·05, n = 5, p = 0·894, respectively) compared response was probably related to the high number of micro- with those in the vehicle group (stimulation index: organisms administered to mice and it was in line with the 3·61 ± 0·56, n = 4 and 32·30 ± 10·86, n = 4, respectively) immune pathways observed: neutrophil degranulation and (data not shown). Finally, no between-group differences innate immune system (adjusted p < 0·05 and NES > 1·5). in the cytokine secretion pattern (GM-CSF, IFN-γ, IL-2, IL-4, IL-6, IL-10, IL-12p70, IL-17A, IL-21, IL-22, IL-23, Clostridia Strains Increase the Levels of the SCFA and TNF-α) were observed for antigen-specific stimulatory Butyrate in Mouse Serum condition (data not shown). Next, we tested the therapeutic effect of butyrate administra- Since the administered Clostridia strains are highly known tion as a potential translational approach. Since a mean daily for their capacity of SCFA production, we studied whether butyrate intake (90·67 ± 19·38 mg/mouse) was estimated in the the levels of several SCFA such as acetate, propionate, preventive approach and it was identified as an effective dose and butyrate, were altered as a result of the oral adminis- in the EAE model, 100 mg of this SCFA was selected as the tration of these bacteria. The concentration of the SCFA suitable butyrate quantity for the therapeutic approaches. Mice butyrate was increased in the serum of Clostridia-treated were treated once daily with butyrate or vehicle after attaining mice (1·74 ± 0·40 µM, n = 6, p = 0·047) compared to that in a clinical score equal to or greater than 1 and being randomised vehicle-treated mice (1·23 ± 0·33 µM, n = 5) (Fig. 1f) but no into clinically equivalent experimental groups, from 13–15 dpi differences were observed in the rest of the studied SCFAs to the end of the experiment (28 dpi). A slight clinical effect (data not shown). was observed in the butyrate group (AUC: 50·24 ± 13·64, n = 19, p = 0·069) compared to that in vehicle-treated mice (AUC: Butyrate Ameliorates the Clinical Signs and Reduces 56·89 ± 13·93, n = 18) (Fig. 8a,b), although these differences CNS Autoimmunity in EAE Mice were not statistically significant. No differences were observed regarding accumulated weight loss (Fig. 8c) but, compared to Since Clostridia strains increased the production of the the vehicle group (16·57 ± 11·23 s, n = 17), butyrate treatment immunoregulatory SCFA butyrate, we studied whether the improved motor coordination skills (25·61 ± 12·05 s, n = 19, previously observed clinical effect could be reproduced by p = 0·005) (Fig. 8d). butyrate administration alone. Within the spinal cord white matter, demyelination Daily intake of butyrate (90·67 ± 19·38 mg/mouse), (6·59 ± 1·71%, n = 8, p = 0·063), T cell inflammatory infiltrate from seven days before mice immunisation to the end of density (58·83 ± 28·18 cells/mm2, n = 8, p = 0·073), and astro- the experiment (35 dpi), significantly reduced the clinical glia reactivity (1·00 ± 0·46%, n = 8, p = 0·074) were slightly signs of EAE mice (AUC: 27·20 ± 27·12, n = 5, p = 0·016) Selected Clostridia Strains from The Human Microbiota and their Metabolite, Butyrate, Improve… 929 Fig. 3  Clostridia Strains Increase Autoreactive Splenocytes ex the proliferative capacity of antigen-specific cells in response to vivo. Splenocyte cultures were prepared at the end of the experi- MOG stimulation a but do not alter the polyclonal response b. The ment (28 dpi) and stimulated with 5 μg/ml M ­ OG35-55 or 5 μg/ml graphs present the combined results of two independent experiments PHA-L and compared to non-stimulated (control) condition. After (n = 15 mice per group). The data are presented as the means ± stand- 54 h in vitro, supernatants were harvested to further assess cytokine ard deviations. Abbreviations: M ­ OG35-55: peptide 35–55 from myelin secretion and cell cultures were supplemented with [­ 3H]-thymidine to oligodendrocyte glycoprotein, PHA-L: phytohaemagglutinin-L quantify cell proliferation at 72 h in vitro. Clostridia strains increase reduced in butyrate-treated mice compared to vehicle treat- XIVa [5], we thought that not only this bacteria cluster could ment (demyelination: 9·50 ± 3·91%, n = 7; T cell inflamma- increase immunoregulatory responses but also it could have tory infiltrate density: 124·42 ± 120·38 cells/mm2, n = 7; and the potential to partially rebalance gut dysbiosis in MS astroglia reactivity: 1·71 ± 0·62%, n = 7) (Fig. 9a-c, 9d-f, and patients. Our study is one of the few preclinical assays that 9j-l), although differences did not reach statistical significance. uses human isolated bacteria in an EAE model [21, 22]. Regarding microglia reactivity and infiltrating macrophages, no Likewise, it is the first one that provides commensal bacteria statistically significant differences were observed between the with the potential to partially rebalance gut dysbiosis in MS butyrate (3·54 ± 1·46%, n = 8, p = 0·127) and the vehicle group patients. (6·10 ± 2·89%, n = 7) (Fig. 9g-i). However, the level of axonal Previous studies in EAE and MS patients have damage was lower in mice treated with butyrate (0·96 ± 0·43%, observed alterations in the intestinal permeability [16, n = 8, p = 0·011) than in the control group (2·08 ± 1·05%, n = 7) 23, 24]. It has been proposed that this intestinal dys- (Fig. 9m-o). As seen in the preventive approach, butyrate function could be contributing to disease progression treatment did not modify antigen-specific (stimulation index: or the pathophysiology of the disease. In fact, intestinal 6·52 ± 3·33, n = 11 vs. 7·75 ± 2·74, n = 10, p = 0·129, respec- permeability increases as early as in the induction and tively) or polyclonal immune responses (stimulation index: the inflammatory phases of EAE and CNS autoreac- 28·41 ± 24·27, n = 11 vs 26·28 ± 10·83, n = 10, p = 0·577, respec- tive T cells have been related to this damaged intestinal tively) compared to the vehicle group as a therapeutic approach integrity [16, 24]. However, no differences in gut perme- (data not shown). ability were observed between our experimental groups when intestinal permeability was evaluated in the chronic phase of the experimental disease. Conversely, other Discussion studies have revealed a reduced intestinal permeability in probiotic-treated mice in both the induction and the Our results show that oral treatment with human gut- inflammatory phases of EAE as a preventive approach derived Clostridia strains improves the clinical outcome [24]. Keeping this in mind, it is possible that once intes- of established EAE as a therapeutic approach. Previous tinal features are established, the ability to revert this studies showed that these 17 Clostridia strains performed intestinal dysfunction may be limited. both clinical and histological benefits in an experimental We observed a decrease of CNS demyelination in accord- model of colitis [10]. These bacteria were originally iso- ance to clinical features in Clostridia-treated group. Moreo- lated from a healthy human faecal sample and their selection ver, we wanted to deepen into other important pathological was grounded in their capacity of inducing Treg cells [10]. hallmarks of the experimental disease, such as T cell infiltra- As the selected bacteria belong to Clostridia clusters IV, tion, reactive microglia/infiltrating macrophages and astro- XIVa, and XVIII and it has been reported that MS patients cyte reactivity, and axonal damage. Although every patho- present a significant decrease of Clostridia clusters IV and logical feature tended to decrease in line with the attenuation 930 L. Calvo-Barreiro et al. Fig. 4  Clostridia Strains Increase Immunoregulatory Proper- CD62L b and CD25 c inducible surface markers in Treg cells in the ties of Treg Cells in the Periphery. Splenocyte suspensions were periphery. Representative dot plots a or histograms b, c of every prepared by grinding spleens of EAE mice through a 70-μm nylon inducible surface marker in Treg cells are shown. The graphs present cell strainer at 28 dpi. Cell subsets were analysed after exclusion of the results of a representative experiment (n = 9 mice per group). The doublets and dead cells from the gating scheme. Clostridia treatment data are presented as the means ± standard deviations. Abbreviations: increases the ­CD62L+ Treg cell population a and the expression of MFI: median fluorescence intensity of clinical signs, only astrogliosis was significantly affected have a negative effect on the CNS pathology, the cytokine by Clostridia strains administration. secretion pattern from MOG-stimulated immune cells, or Regarding the immunological effects, Clostridia strains any other studied immune response in the periphery. On only performed subtle immune regulation in the periphery. the contrary, this result could be in line with the observed Whereas a higher antigen-specific immune proliferation increase in the ­CD62Lhigh Treg cell subset in the periphery. was detected under Clostridia treatment, which was ini- However, since immunophenotyping was not performed tially thought to be adverse, this immune response did not within the antigen-specific cells, we cannot explain the Selected Clostridia Strains from The Human Microbiota and their Metabolite, Butyrate, Improve… 931 Fig. 5  Clostridia Strains Modify Th Population in the Periphery. Splenocyte suspen- sions were prepared by grinding spleens of EAE mice through a 70-μm nylon cell strainer at 28 dpi. Cell subsets were analysed after exclusion of doublets and dead cells from the gating scheme. Clostridia-treated mice presented a decrease in the inducible PD-1+ Th cell population a and increased lev- els of CD25 activation marker in ­CD4+ T lymphocytes b. Representative dot plots a and histogram b are shown. The graphs present the results of a representative experiment (n = 9 mice per group). The data are presented as the means ± stand- ard deviations. Abbreviations: MFI: median fluorescence intensity increased immune cell proliferation by means of this Treg Although Clostridia strains treatment did not modify the cell population alone. overall Treg cell population in the periphery, it increased In the transcriptome studies, an activation of the innate both the percentage of Treg cells that express CD62L and immune system was observed under Clostridia treat- the expression level of this surface molecule within Treg cell ment. We hypothesise that, rather than a disease-related population. It has been previously described that ­CD62Lhigh immune response, the high number of administered bac- Treg cells displayed higher immunosuppressive properties teria could be promoting this innate immune response in due to a higher ability to suppress proliferation of effector the periphery. On the other hand, transcriptome studies T cells and to induce apoptosis in Th cells in vitro [31]. also highlighted an upregulation of genes related to the Clostridia treatment also promoted higher CD25 (IL-2 recep- response to IFN-β in the periphery of Clostridia-treated tor α-chain) levels in the cell surface of Treg and Th cells mice. Whereas this cytokine could also be fostered as a compared to vehicle treatment, these latter depending directly response to bacterial infection, the biological or chemi- on the former. High levels of IL-2 receptor in Treg cells cal stimuli that triggered this cytokine by the immune have been proposed as an immunoregulatory mechanism system have not been studied. It is relevant to mention that deprives proliferating effector cells from this growth that IFN-β is a first-line drug to treat MS and it has been factor [32, 33]. Consequently, a lower availability of IL-2 described to ameliorate EAE clinical course. Moreover, will diminish effector immune cell proliferation. Besides pre- the deletion of the IFN-β gene has been shown to worsen venting proliferative processes, IL-2 deprivation also leaves EAE clinical course [25]. Regarding its mechanisms of effector cells without a key cytokine for metabolic processes action, some studies have demonstrated that its antiin- and survival. Finally, Clostridia treatment presented a lower flammatory effect is mediated by the innate immune population of Th cells that express the immunoinhibitory cells that inhibit proinflammatory Th17 cells [26, 27], receptor PD-1, which has been defined as an inducible mol- whereas other groups have described that IFN-β would ecule involved in the termination of the immune response be inhibiting T cell activation, reducing immune cell traf- after T cell activation [34]. This later effect suggests that ficking across the BBB, or inducing antiinflammatory a higher number of T cells from vehicle-treated mice have cytokines among others [28]. Together with the enhanced been previously activated and, therefore, a higher amount response to IFN-β, an increase in the genes related to the of Th cells expressed PD-1 molecule to negatively regulate production of IL-1β in the periphery was also observed excessive immune activation. in Clostridia-treated mice. This result may be in line with Within the CNS of Clostridia-treated mice, transcriptome previous studies in EAE and MS, which pointed to a role studies highlighted the reduction in several biological processes of the NLRP3 inflammasome and its related cytokine, related to the activation, differentiation, and proliferation of IL-1β, in the effective clinical response to IFN-β in EAE immune cells. Particularly, the reduction of the mononuclear mice and RRMS patients [29, 30]. cell proliferation was strongly connected to lymphocyte and T 932 L. Calvo-Barreiro et al. Fig. 6  Clostridia Strains Reduce the Immune Response in the vation of the immune response and T cells, in particular; 2) the dif- CNS and Increase the Innate Immune Response in the Periphery. ferentiation of lymphocytes; and 3) the proliferation of mononuclear At 28 dpi, spinal cords and splenocytes of euthanised mice were col- cells and lymphocytes and T cells, specially, in the CNS. On the other lected and total RNA was isolated. Subsequently, cDNA was gener- hand, Clostridia strains promote gene sets related to: 1) the defense ated and Clariom S Pico Assay HT, mouse arrays were processed. response to other organisms; 2) the innate immune response; and 3) The analysis of biological significance, based on the GSEA method, IL-1, IL-1β, and IFN-β production. In these plots, the size of the GO highlights that Clostridia strains alter different gene sets from the term nodes is related to the number of genes found in that category, "Gene Ontology" (GO) database in both the CNS a and the peripheral the color of the gene nodes refers to their logFC, and the color of the immune system b. Specifically, Clostridia strains reduced: 1) the acti- edges links each gene with its GO term cell proliferation. Whereas the histopathological studies did not downregulation of genes related to the immune cell prolifera- show a decrease in T cell density within the spinal cord white tion in the CNS since IL-2 is the prototypical T cell growth fac- matter, the trend towards a lower LEA-positive cells (infiltrat- tor. Finally, the decrease in the innate immune system pathway ing macrophages and reactive microglia) is consistent with could partially be pointing to the reduction of macrophages in these gene expression results. A few immune pathways were the CNS in line with the histopathological studies. also downregulated in the CNS of Clostridia-treated mice. The The SCFA butyrate and, to a lesser extent, propionate, reduction in the cell surface interactions at the vascular level have been demonstrated to foster in vitro activation of DCs may be related to the downregulation of genes related to the and subsequent promotion of Treg cells as well as in vivo immune cell proliferation in the CNS as well as to the upregula- generation of Treg cells in peripheral compartments [11]. tion of genes related to the response to IFN-β in the periphery. Moreover, SCFAs have been stood out as the real sources Since the IFN-β limits immune cell trafficking to the CNS by of the antiinflammatory immune responses exerted by downregulating T cell adhesion molecules that interacts to BBB some probiotic bacteria, including Clostridia strains [10]. adhesion molecules [35], the lower cell surface interaction at Therefore, the increased levels of butyrate in the serum of the vascular level may be partially related to the diminished Clostridia-treated mice indicated that this SCFA might be immune cell responses in the CNS. On the other hand, the acting as an antiinflammatory molecule in the periphery. In reduction in the IL-2 signalling could also be associated to the fact, the presence of this bacterial product could partially Selected Clostridia Strains from The Human Microbiota and their Metabolite, Butyrate, Improve… 933 934 L. Calvo-Barreiro et al. ◂Fig. 7  Butyrate Treatment Reduces the Pathological Clinical Signs Since our results, together with previous studies, connected the and the CNS Inflammation and Demyelination of EAE Mice as a SCFA butyrate with immunoregulatory responses, we decided Preventive Approach. Daily oral preventive treatment with sodium butyrate (daily intake: 90·67 ± 19·38 mg) from 7 days before mice to test the oral administration of butyrate in the chronic EAE immunisation until the end of the experiment (35 dpi) improves clini- model. In our experimental model, preventive administration cal score a, AUC b, weight loss c, and tends to improve motor func- of butyrate performed a beneficial effect on CNS autoimmunity tion skills d compared to vehicle group. Moreover, sodium butyrate by halting both demyelination and CNS inflammation. These reduces CNS demyelination e and inflammation f. The graphs pre- sent the results of a single experiment (n = 5 mice per group). Arrows results are in accordance with previous studies that reported a indicate demyelinated and inflamed areas within the spinal cord white significant amelioration of CNS demyelination in a cuprizone matter. Scale bars indicate 100 µm or 500 µm. The data are presented model as a preventive approach [36]. Considering a poten- as the means ± standard errors of the mean a, c or the means ± stand- tial translational approach, we tested the therapeutic effect of ard deviations b, d-f. Black triangle, Vehicle; white triangle, Butyrate butyrate administration in the EAE model. A slight but not statistically significant clinical effect on established EAE was observed. A tendency to lower demyelination, CNS inflamma- explain the increase in Treg cell subsets as well as in their tion, and astrogliosis was detected in line with clinical outcome. higher immunosuppressive properties. It is worth mentioning On the contrary, a noticeable improve concerning axonal dam- that the detection of butyrate in the mouse serum illustrates age was observed. In this way, the remarkable difference on how the oral administration of these Clostridia strains pro- EAE clinical course between the preventive and the therapeutic moted a systemic effect that links the gut environment with approaches suggest that SCFA butyrate would be acting only on the peripheral immune compartment. the development of the immune response and has little effect on Fig. 8  Therapeutic Administration of Butyrate Causes Slight administration of butyrate slightly improves the clinical score a and Clinical Effects in EAE Mice. C57BL/6JOlaHsd mice were immu- the AUC b and does not affect weight loss c. Furthermore, butyrate nised by subcutaneous injection of ­MOG35–55 and weighed and exam- improves motor function skills compared to vehicle group d. The ined daily for neurological signs. After attaining a clinical score equal charts present the combined results of three independent experiments to or greater than 1 and being randomised (13–15 dpi), mice were (Vehicle, n = 18 and Butyrate, n = 19). The data are presented as the treated once daily with sodium butyrate (100 mg/dose) or water (vehi- means ± standard errors of the mean a, c or the means ± standard devi- cle), weighed and examined daily for neurological signs in a blinded ations b, d. Black triangle, Vehicle; white triangle, Butyrate manner until the end of the experiment (28 dpi). Daily therapeutic Selected Clostridia Strains from The Human Microbiota and their Metabolite, Butyrate, Improve… 935 936 L. Calvo-Barreiro et al. ◂Fig. 9  Some Pathological Signs are Improved by the Therapeutic Our data show that the human gut-derived Clostridia strains Administration of Butyrate in EAE Mice. At 28 dpi, spinal cords improved the clinical outcome of established EAE as a therapeu- of euthanised EAE mice were collected, fixed in a 4% paraformalde- hyde solution, embedded in paraffin, and cut into 4-μm thick coronal tic approach. The observed clinical improvement was related to sections. Demyelination was assessed by anti-myelin basic protein lower histopathological features in the CNS and was associated (MBP) antibody; T cell infiltrates, by anti-CD3 antibody; microglia to an enhanced immunoregulatory response of Treg cells in the reactivity/infiltrating macrophages and reactive astroglia, by lectin periphery. Likewise, transcriptome studies highlighted increased from Lycopersicon esculentum and anti-glial fibrillary acidic protein (GFAP) antibody, respectively; and axonal damage, by anti-neurofil- antiinflammatory responses related to IFN-β in the periph- ament H, nonphosphorylated (SMI32). Therapeutic administration of ery and lower activation, differentiation, and proliferation of butyrate decreases axonal damage m–o, tends to reduce CNS demy- immune cells in the CNS. Further in vivo assays with the SCFA elination a-c, T cell infiltration d-f and reactive astroglia j-l, and does butyrate proved its preventive effect on CNS autoimmunity but not alter microglia reactivity/infiltrating macrophages g-i in the spi- nal cord white matter. The graphs present the results of a representa- a slight therapeutic clinical impact on EAE clinical course. Our tive experiment (Vehicle, n = 7; and Butyrate, n = 8). Arrows indicate results suggest that the therapeutic effect performed by this com- ­CD3+ cells. Scale bars indicate 100 µm. The data are presented as the mensal bacteria population was not exclusively related to the SCFA means ± standard deviations. Abbreviations: DAPI: 4′,6-diamidino- and could not be reproduced by butyrate administration alone. 2-phenylindole, GFAP: glial fibrillary acidic protein, LEA: lectin from Lycopersicon esculentum, MBP: myelin basic protein, SMI32: neuro- Although it is still unknown whether these 17 Clostridia strains filament H, nonphosphorylated will have the same effect on MS patients, previously-defined gut dysbiosis regarding Clostridia cluster IV and XIVa in MS patients could be rebalanced by Clostridia strains administration the neuroinflammatory processes. However, the limited immune and its immunoregulatory properties could have a beneficial studies performed in this experimental model do not allow for effect on MS clinical course. understanding the mechanism of action of butyrate treatment on CNS autoimmunity. Our studies only proved that butyrate Supplementary Information The online version contains supplementary material available at https://​doi.​org/​10.​1007/​s13311-​021-​01016-7. treatment does not modify proliferation or cytokine secretion pattern of antigen-specific cells ex vivo. Additionally, the SCFA Acknowledgments This work was supported by the Fondo de Inves- butyrate seems to behave similar to propionate treatment since tigación Sanitaria, Instituto de Salud Carlos III, co-funded by the only preventive but not therapeutic oral administration after the European Union (European Regional Development Fund/European EAE induction phase had a significant beneficial effect on EAE Social Fund) “A way to build Europe” under Grant PI15/00840 and RD16/0015/004; and “Agència de Gestió d’Ajuts Universitaris clinical course [12]. Nevertheless, the translational approach to i de Recerca” (AGAUR; Generalitat de Catalunya) under Grant evaluate propionate efficacy in MS patients did perform a ben- 2017SGR527. eficial clinical effect on MS disease course [37]. Oral propion- ate supplementation increased the number and immunoregula- Required Author Forms Disclosure forms provided by the authors are tory function of Treg cells and decreased the number of Th17 available with the online version of this article. cells, reversing the Treg/Th17 cell imbalance in MS patients. Open Access This article is licensed under a Creative Commons Attri- Regarding the CNS compartment, oral propionate supplementa- bution 4.0 International License, which permits use, sharing, adapta- tion increased propionate concentration in the CSF and this, in tion, distribution and reproduction in any medium or format, as long turn, was associated to the increase in subcortical grey matter in as you give appropriate credit to the original author(s) and the source, treated MS patients. More importantly, MS patients who were provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are continuously supplemented with SCFA propionate for at least included in the article’s Creative Commons licence, unless indicated one year reduced their annual relapse rate and disease progres- otherwise in a credit line to the material. If material is not included in sion [37]. Thus, this translational study made us aware of the the article’s Creative Commons licence and your intended use is not therapeutic potential of butyrate administration even though permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a its therapeutic clinical impact in the EAE model was not sta- copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. tistically significant. Further studies should be performed to investigate its immunoregulatory effect on EAE pathogenesis and to value its translational potential. On the other hand, the References smaller therapeutic effect in the butyrate-treated animals than in Clostridia-treated mice, even though higher levels of the SCFA 1. Dendrou CA, Fugger L, Friese MA. Immunopathology of multiple butyrate were present in the former group, highlighted that the sclerosis. Nat Rev Immunol. 2015;15:545-558. beneficial outcome exerted by the oral administration of the bac- 2. Calvo-Barreiro L, Eixarch H, Montalban X, Espejo C. Combined therapies to treat complex diseases: The role of the gut microbiota teria was not only related to the production of the SCFA. Thus, in multiple sclerosis. Autoimmun Rev. 2018;17:165-174. the Clostridia-induced clinical effect could not be reproduced by 3. Cantarel BL, Waubant E, Chehoud C, et al. Gut microbiota in mul- butyrate administration alone. tiple sclerosis: possible influence of immunomodulators. Journal Selected Clostridia Strains from The Human Microbiota and their Metabolite, Butyrate, Improve… 937 of investigative medicine : the official publication of the American 22. Salehipour Z, Haghmorad D, Sankian M, et al. Bifidobacterium Federation for Clinical Research. 2015;63:729-734. animalis in combination with human origin of Lactobacillus 4. Jangi S, Gandhi R, Cox LM, et al. Alterations of the human gut plantarum ameliorate neuroinflammation in experimental model microbiome in multiple sclerosis. Nat Commun. 2016;7:12015. of multiple sclerosis by altering CD4+ T cell subset balance. 5. Miyake S, Kim S, Suda W, et al. Dysbiosis in the gut microbiota Biomed Pharmacother = Biomedecine & pharmacotherapie. of patients with multiple sclerosis, with a striking depletion of 2017;95:1535–1548. species belonging to clostridia XIVa and IV clusters. PloS one. 23. Buscarinu MC, Cerasoli B, Annibali V, et al. Altered intestinal 2015;10:e0137429. permeability in patients with relapsing-remitting multiple sclero- 6. Forbes JD, Chen CY, Knox NC, et al. A comparative study of the sis: a pilot study. Mult Scler. 2017;23:442-446. gut microbiota in immune-mediated inflammatory diseases-does 24. Secher T, Kassem S, Benamar M, et al. Oral administration of the a common dysbiosis exist? Microbiome. 2018;6:221. probiotic strain Escherichia coli Nissle 1917 reduces susceptibil- 7. Tremlett H, Fadrosh DW, Faruqi AA, et al. Gut microbiota in early ity to neuroinflammation and repairs experimental autoimmune pediatric multiple sclerosis: a case-control study. Eur J Neurol. encephalomyelitis-induced intestinal barrier dysfunction. Front 2016;23:1308-1321. Immunol. 2017;8:1096. 8. Chen J, Chia N, Kalari KR, et al. Multiple sclerosis patients have 25. Teige I, Treschow A, Teige A, et al. IFN-beta gene deletion leads a distinct gut microbiota compared to healthy controls. Sci Rep. to augmented and chronic demyelinating experimental autoim- 2016;6:28484. mune encephalomyelitis. J Immunol. 2003;170:4776-4784. 9. Zeng Q, Junli G, Liu X, et al. Gut dysbiosis and lack of short chain 26. Shinohara ML, Kim JH, Garcia VA, Cantor H. Engagement of fatty acids in a Chinese cohort of patients with multiple sclerosis. the type I interferon receptor on dendritic cells inhibits T helper Neurochem Int. 2019;129:104468. 17 cell development: role of intracellular osteopontin. Immunity. 10. Atarashi K, Tanoue T, Oshima K, et al. Treg induction by a ration- 2008;29:68-78. ally selected mixture of Clostridia strains from the human micro- 27. Guo B, Chang EY, Cheng G. The type I IFN induction pathway biota. Nature. 2013;500:232-236. constrains Th17-mediated autoimmune inflammation in mice. J 11. Arpaia N, Campbell C, Fan X, et al. Metabolites produced by Clin Investig. 2008;118:1680-1690. commensal bacteria promote peripheral regulatory T-cell genera- 28. Dhib-Jalbut S, Marks S. Interferon-beta mechanisms of action in tion. Nature. 2013;504:451-455. multiple sclerosis. Neurology. 2010;74 Suppl 1:S17-S24. 12. Haghikia A, Jorg S, Duscha A, et al. Dietary Fatty Acids Directly 29. Inoue M, Williams KL, Oliver T, et al. Interferon-beta therapy Impact Central Nervous System Autoimmunity via the Small against EAE is effective only when development of the disease Intestine. Immunity. 2016;44:951-953. depends on the NLRP3 inflammasome. Sci Signal. 2012;5:ra38. 13. Calvo-Barreiro L, Eixarch H, Ponce-Alonso M, et al. A com- 30. Malhotra S, Rio J, Urcelay E, et al. NLRP3 inflammasome is asso- mercial probiotic induces tolerogenic and reduces pathogenic ciated with the response to IFN-beta in patients with multiple responses in experimental autoimmune encephalomyelitis. Cells. sclerosis. Brain. 2015;138:644-652. 2020;9:906. 31. Lange C, Scholl M, Melms A, Bischof F. CD62L(high) Treg 14. Baker D, Amor S. Publication guidelines for refereeing and report- cells with superior immunosuppressive properties accumulate ing on animal use in experimental autoimmune encephalomyelitis. within the CNS during remissions of EAE. Brain Behav Immun. J Neuroimmunol. 2012;242:78-83. 2011;25:120-126. 15. Gutiérrez-Franco A, Eixarch H, Costa C, et al. Semaphorin 7A 32. Caridade M, Graca L, Ribeiro RM. Mechanisms underlying CD4+ as a potential therapeutic target for multiple sclerosis. Mol Neu- Treg immune regulation in the adult: from experiments to models. robiol. 2017;54:4820-4831. Front Immunol. 2013;4:378. 16. Nouri M, Bredberg A, Westrom B, Lavasani S. Intestinal barrier 33. Arce-Sillas A, Alvarez-Luquin DD, Tamaya-Dominguez B, et al. dysfunction develops at the onset of experimental autoimmune Regulatory T cells: molecular actions on effector cells in immune encephalomyelitis, and can be induced by adoptive transfer of regulation. J Immunol Res. 2016;2016:1720827. auto-reactive T cells. PloS one. 2014;9:e106335. 34. Agata Y, Kawasaki A, Nishimura H, et al. Expression of the PD-1 17. Gomez-Gomez A, Soldevila A, Pizarro N, Andreu-Fernandez V, antigen on the surface of stimulated mouse T and B lymphocytes. Pozo OJ. Improving liquid chromatography-tandem mass spec- Int Immunol. 1996;8:765-772. trometry determination of polycarboxylic acids in human urine by 35. Muraro PA, Leist T, Bielekova B, McFarland HF. VLA-4/CD49d chemical derivatization. Comparison of o-benzyl hydroxylamine downregulated on primed T lymphocytes during interferon-beta and 2-picolyl amine. J Pharm Biomed Anal. 2019;164:382–394. therapy in multiple sclerosis. J Neuroimmunol. 2000;111:186-194. 18. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment 36. Chen T, Noto D, Hoshino Y, Mizuno M, Miyake S. Butyrate sup- analysis: a knowledge-based approach for interpreting genome- presses demyelination and enhances remyelination. J Neuroin- wide expression profiles. Proceedings of the National Academy of flamm. 2019;16:165. Sciences of the United States of America. 2005;102:15545-15550. 37. Duscha A, Gisevius B, Hirschberg S, et al. Propionic acid shapes 19. Fabregat A, Jupe S, Matthews L, et al. The Reactome Pathway the multiple sclerosis disease course by an immunomodulatory Knowledgebase. Nucleic Acids Res. 2018;46:D649-D655. mechanism. J Neuroinflamm. Cell. 2020;180:1067-1080. 20. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Publisher’s Note Springer Nature remains neutral with regard to Series B Stat Methodol. 1995;57:289–300. jurisdictional claims in published maps and institutional affiliations. 21. Mangalam A, Shahi SK, Luckey D, et al. Human gut-derived com- mensal bacteria suppress CNS inflammatory and demyelinating disease. Cell Rep. 2017;20:1269-1277.