Received: 1 February 2022 Revised: 3 April 2022 Accepted: 8 April 2022 DOI: 10.1002/ctm2.837 LETTER TO EDITOR Inflammation in multiple sclerosis induces a specific reactive astrocyte state driving non-cell-autonomous neuronal damage Dear Editor found an altered pro-inflammatory profile comprised of An in-depth understanding of the neurodegenerative 23 upregulated factors (Figure 3A). Functional enrich- component of multiple sclerosis (MS) is crucial for the ment and interactome analysis revealed a set of pro- design of therapeutic approaches that may stop disease inflammatory pathways enriched following the MS-High progression. Astrocytes have emerged as key contribu- CSF exposure (Figure S1). Moreover, nuclear factor NF- tors to the pathogenesis of MS.1 However, the mecha- kappa-B p105 subunit (Nfkb1) and cellular tumour anti- nisms underlying the regulation of maladaptive astrocytic gen p53 (Trp53) were identified as the transcription fac- responses remain unknown. In this report, we show that tors regulating the MS-High-associated astrocyte secre- a high inflammatory activity in MS patients at disease tome (Figure 3B), both involved in NF-ĸB signalling. onset induces a specific reactive astrocyte state that triggers SerpinE1, also known as plasminogen activator inhibitor synaptopathy and contributes to neuronal damage in vitro 1 (PAI-1), which has been shown to exacerbate axonal dam- and ex vivo suggesting potential mechanisms that may ulti- age and demyelination in MS animal models3 and be reg- mately lead to neurodegeneration. ulated by NF-ĸB in neuroinflammation,4 was significantly To investigate whether astrocytes are essential contrib- increased in the MS-High secretomes (Figure 3A and Table utors to neuronal damage in MS, we cultured purified S2). Considering its potential role as a mediator of neurode- astrocytes with cerebrospinal fluid (CSF) samples from generation, we validated by ELISA SerpinE1 increased lev- MS patients with high inflammatory activity at disease els in secretomes from astrocytes exposed to MS-High con- onset (MS-High, Table S1). Then, we examined the effect of dition (Figure 3C). astrocytic secretomes on neurons (Figure 1A). Astrocytes By using omics technologies, we studied whether secre- became reactive upon high inflammatory CSF exposure tomes that alter neuronal plasticity are associated with (Figure 1B) and induced morphological alterations typi- a specific reactive astrocyte state in MS patients with cally observed in neurodegenerative disorders, such as a high inflammatory activity. Astrocytes stimulated with less complex dendritic tree due to decreased arborisation MS-High CSF exhibited a specific reactive gene (Figure 3D) (Figure 1C, D). Moreover, these abnormalities were accom- and protein (Figure 3E) expression profile. We identified a panied with synaptic plasticity impairment (Figure 1E, F). MS-High-associated reactive gene signature comprised of Considering that a high lesion load at disease onset has 7 differentially expressed genes (Figure 3F) that were vali- been associated with an increased risk of neurological dis- dated by qPCR (Figure 3G). This reactive gene expression ability development,2 we assessed whether the non-cell- fingerprint was mostly comprised of downregulated imme- autonomous effect on neuronal plasticity could be influ- diate early response genes (Nr4a1, Klf6, Egr2 and Fosb). enced by the degree of inflammatory activity of MS patients Interestingly, Nr4a1 and Klf6 have been reported to pro- (Figure 2A and Table S1). Interestingly, we observed a mote anti-inflammatory responses by specifically repress- direct correlation between the degree of inflammatory ing NF-ĸB activity.5,6 To further decipher the MS-High- exposure and the extent of both astrocyte-mediated synap- specific reactive astrocyte state, we performed a functional topathy (Figure 2B, C) and dendrite arborisation impair- enrichment analysis integrating all datasets obtained from ment (Figure 2D, E). CSF exposed astrocytes: secretomes and gene/protein We next characterised the secretomes from astrocytes expression. Overall, this revealed a prominent inflam- exposed to high inflammatory MS microenvironment and matory signature in MS-High astrocytes (Figure S2). This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2022 The Authors. Clinical and Translational Medicine published by John Wiley & Sons Australia, Ltd on behalf of Shanghai Institute of Clinical Bioinformatics. Clin. Transl. Med. 2022;12:e837. wileyonlinelibrary.com/journal/ctm2 1 of 8 https://doi.org/10.1002/ctm2.837 2 of 8 LETTER TO EDITOR F I G U R E 1 Astrocytes exposed to CSF from MS patients with high inflammatory activity induce synaptic plasticity impairment. (A) Flowchart depicting CSF collection from a cohort of MS patients with high inflammatory activity (MS-High; N = 4), followed by exposure of astrocytes to the CSF in vitro, secretome collection and co-culture with neurons. (B) qPCR for the astrocyte reactive marker lipocalin-2 (Lcn2) of primary astrocytes exposed for 6 and 12 h to medium (Basal) or CSF. One-way ANOVA analysis; n = 2 independent experiments. (C) Illustrative confocal images of primary cortical neurons treated with media (Basal) or MS-High-exposed astrocyte secretomes. Neurons were immunostained with MAP2 (green) and DAPI (blue). (D) Graphs represent individual data of neuronal area, total dendrite tree length and number of dendrite segments per neuron. Least Squares Means Estimates test, n = 2 replicates per condition, n = 2 independent experiments. (E) Illustrative confocal images of primary cortical neurons treated with Control or MS-High-exposed astrocyte secretomes. Neurons were immunostained with MAP2 (white), the pre-synaptic marker Synaptophysin (green) and the post-synaptic marker PSD-95 (red). Arrows in high-magnification insets point to co-localization of Synaptophysin and PSD-95 (synapses). (F) Graphs represent individual data of the density of PSD-95, Synaptophysin and PSD-95/Synaptophysin double-positive puncta. Least Squares Means Estimates test; n = 40 neurons per condition, n = 2–3 dendrites per neuron, n = 2 independent experiments. Data are shown as mean (standard error of the mean, SEM). *p < .05, **p < .01, ***p < .001. Scale bars: 40 μm (B) and 10 μm (D). DIV: days in vitro LETTER TO EDITOR 3 of 8 F I G U R E 2 Astrocyte-mediated synaptic plasticity impairment correlates with the degree of inflammatory activity in MS patients. (A) Flowchart depicting CSF collection from a cohort of MS patients with high inflammatory activity (MS-High; N = 9), MS patients with low inflammatory activity (MS-Low; N = 9) and non-inflammatory neurological controls (NINC; N = 9) followed by exposure of astrocytes to the CSF in vitro, secretome collection and co-culture with neurons. (B) Illustrative confocal images of primary cortical neurons treated with secretomes from astrocytes exposed to CSF. Neurons were immunostained with MAP2 (white), the pre-synaptic marker Synaptophysin (green) and the post-synaptic marker PSD-95 (red). Arrows in high-magnification insets point to co-localisation of Synaptophysin and PSD-95 (synapses). (C) Graphs represent individual relative numbers of PSD-95, Synaptophysin and PSD-95/Synaptophysin double-positive puncta density normalised to untreated neurons (NT). Least Squares Means Estimates test, Tukey–Kramer multiple comparisons test; n = 3 independent secretome samples per group, n = 180 neurons per condition (n = 60 for non-treated neurons), n = 2–3 dendrites per neuron, n = 2 replicates per condition, n = 3 independent experiments. (D) Illustrative confocal images of primary cortical neurons treated with MS-High and MS-Low-exposed astrocyte secretomes. Neurons were immunostained with MAP2 (green) and DAPI (blue). (E) Graphs represent individual data of neuronal area, total dendrite tree length and the number of dendrite segments per neuron. Least Squares Means Estimates test; n = 3 independent secretome samples per group, n = 2 replicates per condition, n = 2 independent experiments. Data are shown as mean (SEM). *p < .05, **p < .01, ***p < .001. Scale bars: 10 μm (B) and 40 μm (D). DIV: days in vitro 4 of 8 LETTER TO EDITOR F I G U R E 3 Astrocytes exposed to CSF from MS patients with high inflammatory activity have a specific reactive astrocyte state in vitro. (A) Proteome profiler array of secretomes from CSF-exposed primary astrocytes showing 23 inflammation-related molecules upregulated in the MS-High condition. FDR analysis. n = 3 independent secretome samples per group, n = 3 independent experiments. (B) TRRUST enrichment analysis identifying Nfkb1 as a candidate key transcription factor modulator of upregulated astrocyte-secreted factors in the MS-High condition (FDR = 1.9 × 10–8 ). (C) Dot plot showing SerpinE1 levels determined by ELISA in astrocyte-derived secretomes. Least Squares Means Estimates test and Tukey–Kramer multiple comparisons test; n = 3 independent secretome samples per group. (D–F) Gene expression microarrays and liquid chromatography/mass spectrometry analysis of reactive astrocytes exposed to CSF from MS patients with high inflammatory activity (MS-High), low inflammatory activity (MS-Low) and non-inflammatory neurological controls (NINC). (D, E) Heatmaps and (F) volcano plots showing normalised Log2 gene (D, F) and protein (E) expression satisfying p value < .01 and absolute FC > 0.5; n = 3 independent biological samples per group. FDR analysis; n = 3 independent biological samples per group. (G) mRNA expression levels measured by qPCR of the specific gene expression signature associated with astrocyte exposure to CSF from MS-High patients. Individual values represent average FC = 2–(average ∆∆Ct) in mRNAs of CSF-exposed astrocytes relative to non-stimulated astrocytes (Basal). Least Squares Means Estimates test and Dunnett–Hsu test for multiple comparisons; n = 3 independent biological samples per group. (H) TRRUST enrichment analysis identifying Nfkb1, Trp53 and Rela as candidates key transcription factor modulator of MS-High-reactive astrocyte signature. Data are shown as mean (SEM). *p < .05, **p < .01, ***p < .001 LETTER TO EDITOR 5 of 8 F I G U R E 4 CHI3L1 is a potential mediator of the reactive astrocyte state by enhancing the pro-inflammatory NF-ĸB signalling pathway. (A, B) Liquid chromatography/mass spectrometry analysis of CSF from MS patients with inflammatory activity (MS-High; N = 9), low inflammatory activity (MS-Low; N = 9) and non-inflammatory neurological controls (NINC; N = 9). (A) Heatmap and (B) volcano plots showing normalised Log2 protein expression satisfying p-value < .01 and absolute FC > 0.5 identifying upregulation of CHI3L1 in MS-High CSF (inset). (C) Plots showing bioinformatic annotation analysis integrating CSF, reactive astrocytes and astrocytic secretome data sets of the 6 of 8 LETTER TO EDITOR Furthermore, Nfkb1, Trp53 and transcription factor p65 tions in MS patients10 (Figure 4D). CHI3L1 stimulation (Rela), were identified as the transcription factors regulat- (600 ng/ml) downregulated Nr4a1 and Klf6 expression, ing the MS-High astrocyte-specific fingerprint (Figure 3H). both involved in the inhibition of NF-ĸB signalling5,6 The inhibition of NF-кB activation in astrocytes amelio- (Figure 4E and Figure S3). Protein interactome compu- rated immune infiltrate, axonal damage and demyelina- tation revealed an interaction between the NF-ĸB tran- tion, by preventing the establishment of the astrocyte- scription module (Nr4a1 and Klf6) potentially controlled mediated pro-inflammatory microenvironment that leads by CHI3L1 (Figure 4F). Noteworthy, we also found interac- to disease progression in EAE.7,8 Moreover, a common tions between Mapk3 (ERK1), Mapk1 (ERK2) and Nr4a1 in MS risk variant (rs7665090) has been found associated to the NF-ĸB transcription module, which might be regulated increased NF-кB signalling in astrocytes, driving increased by CHI3L1 and SerpinE1 (Figure 4G). Finally, to address lymphocyte infiltrate and lesion size.9 These findings pro- whether CHI3L1 could be a potential driver of astrocyte- vide evidence that a high inflammatory microenviron- mediated neuronal damage we used P7 murine myelinat- ment in MS patients may mediate disease progression by ing organotypic brain slice cultures that generate compact enhancing NF-кB signalling in astrocytes, which modi- myelin ex vivo and mimic in vivo microenvironment. After fies their secretome content resulting in both immune- 48 h, CHI3L1 (600 ng/ml) induced axonal damage reduc- mediated neurodegeneration and potential direct neuro- ing total neurofilament area (Figure 4H). toxic effects. Our findings provide evidence that the degree of inflam- Next, we investigated whether this reactive astrocyte matory activity in MS patients at disease onset has the state is associated with a specific CSF proteome in MS potential to induce a specific reactive state in astrocytes patients with high inflammatory activity. LC/MS analysis that trigger neuronal damage (Figure S4). This reactive showed that CSF from MS-High patients have a specific state, mainly associated with the NF-ĸB signalling, could proteome profile (Figure 4A, B). To elucidate the mech- be exploited as a prognostic biomarker that reflects a anisms underlying astrocyte reactivity, we performed an potential detrimental effect of MS astrocytes on neuronal integrative omics data analysis at CSF, reactive astrocytes, plasticity. and secretomes levels. Mitogen-activated protein kinase (ERK)-1/2 cascade, NF-ĸB-inducing kinase (NIK)/NF-ĸB AC K N OW L E D G E M E N T S signalling, Nfkb1 and Trp53 were found as the most signif- The authors thank the Advanced Optic Microscopy Unit icantly enriched pathways and transcription factors by the (Campus Casanova) of the Centre Científic i Tecnològic exposure to a highly inflammatory MS microenvironment (CCIT, Universitat de Barcelona) for their advice with (Figure 4C and Table S3). These data reinforced the role microscopy techniques. of an enhanced NF-ĸB signalling in the MS-High reactive astrocytes. CONFLICT OF INTEREST Remarkably, we identified the prognostic biomarker The authors report no competing interests. G.G.D.B. is now chitinase 3-like 1 (CHI3L1)10 upregulated in the MS-High an employee of Evotec. CSF (Figure 4B) and represented in the aforementioned pathways (Table S3). To investigate whether CHI3L1 could F U N D I N G I N F O R M AT I O N be a mediator of the MS-High reactive astrocyte state, we The study was funded by the ‘Red Española de Esclero- stimulated astrocytes with CHI3L1 at concentrations above sis Múltiple (REEM)’ sponsored by the ‘Fondo de Inves- the cut-off value that demonstrated prognostic implica- tigación Sanitaria’ (FIS; project reference: PI15/01111), MS-High condition. ERK1/2 cascade (p = 4.7 × 10–11 ), NIK/NF-ĸB signalling (p = .004) and Nfkb1 (FDR = 5.3 × 10–8 ) exhibit substantial changes, pointing to enhanced NF-ĸB signalling. (D) Flowchart illustrating primary purified astrocyte cultures stimulated with either PBS (Vehicle) or CHI3L1 at 300 and 600 ng/ml, using the same exposure conditions as previously in the CSF stimulation experiments. (E) qPCR assessment of CHI3L1 stimulation. The 600 ng/ml CHI3L1 concentration induces a reduced expression of Nr4a1 (p = .03) and Klf6 (p = .02). Individual values represent average FC = 2–(average ∆∆Ct) in mRNAs in CHI3L1-stimulated astrocytes relative to vehicle. Least Squares Means Estimates test and Tukey–Kramer multiple comparisons test; n = 8 independent biological samples per group. (F) Network diagram of differentially regulated contributors of enhanced NF-ĸB signalling in MS-High-exposed reactive astrocytes predict CHI3L1 as an upstream regulator of astrocyte reactivity (p = 3.5 × 10–5 ). (G) A node of interaction between Mapk3 (ERK1), Mapk1 (ERK2), Trp53 and Nr4a1 in the NF-ĸB transcription module may potentially be regulated by CHI3L1 and would also affect Serpine1 expression (p = .0002). (H) Illustrative confocal images of P7 murine organotypic brain slices treated at 7 DIV with either PBS (Vehicle) or CHI3L1 (600 ng/ml) for 48 h. Neurons were immunostained with neurofilament heavy chain (NFh, green) and DAPI (blue). Graphs represent individual data of averaged NFh area (μm2 ) per stack in each field of view (FOV). Least Squares Means Estimates test; n = 7 mice (Vehicle, n = 6). Data in E and H are shown as mean (SEM). *p < .05. Scale bar: 30 μm (H). DIV: days in vitro LETTER TO EDITOR 7 of 8 Ministry of Science and Innovation, Spain; the ‘Ajuts Institut de Recerca Vall d’Hebron (VHIR), Hospital per donar Suport als Grups de Recerca de Catalunya’, Universitari Vall d’Hebron, Pg. Vall d’Hebron 119–129 sponsored by the ‘Agència de Gestió d’Ajuts Universi- 08035 Barcelona, Spain. taris i de Recerca’ (AGAUR), Generalitat de Catalunya, Email: manuel.comabella@vhir.org; Spain; and Wellcome Trust (110138/Z/15/Z, to D.C.F.), clara.matute@vhir.org United Kingdom. ORCID Clara Matute-Blanch1 Clara Matute-Blanch https://orcid.org/0000-0002-1779- Verónica Brito2,3,4 0008 Luciana Midaglia1 Verónica Brito https://orcid.org/0000-0002-4137-0708 Luisa M Villar5 Luciana Midaglia https://orcid.org/0000-0001-7981- Gerardo Garcia-Diaz Barriga2 2228 Alerie Guzman de la Fuente6 Luisa M Villar https://orcid.org/0000-0002-9067-3668 Eva Borrás7 Gerardo Garcia-Diaz Barriga https://orcid.org/0000- Sara Fernández-García2,3,4 0002-9713-3795 Laura Calvo-Barreiro1 Alerie Guzman de la Fuente https://orcid.org/0000- Andrés Miguez1 0001-8366-4969 Lucienne Costa-Frossard5 Sara Fernández-García https://orcid.org/0000-0003- Rucsanda Pinteac1 2281-4526 Eduard Sabidó7,8 Laura Calvo-Barreiro https://orcid.org/0000-0002- Jordi Alberch2,3,4 8524-3305 Denise C. Fitzgerald6 Andrés Miguez https://orcid.org/0000-0001-6014-3685 Xavier Montalban1 Lucienne Costa-Frossard https://orcid.org/0000-0002- Manuel Comabella1 6512-4413 Rucsanda Pinteac https://orcid.org/0000-0001-5215- 1 Servei de Neurologia-Neuroimmunologia, Centre 3912 d’Esclerosi Múltiple de Catalunya (Cemcat). Institut de Eduard Sabidó https://orcid.org/0000-0001-6506-7714 Recerca Vall d’Hebron (VHIR), Hospital Universitari Vall Jordi Alberch https://orcid.org/0000-0002-8684-2721 d’Hebron. Universitat Autònoma de Barcelona, Barcelona, Denise C. Fitzgerald https://orcid.org/0000-0002-8608- Spain 430X 2 Departament de Biomedicina, Facultat de Medicina, Xavier Montalban https://orcid.org/0000-0002-0098- Institut de Neurociències, Universitat de Barcelona, 9918 Barcelona, Spain Manuel Comabella https://orcid.org/0000-0002-2373- 3 Institut d’Investigacions Biomèdiques August Pi i Sunyer 6657 (IDIBAPS), Hospital Clínic, Universitat de Barcelona, Barcelona, Spain 4 Centro de Investigación Biomédica en Red sobre REFERENCES Enfermedades Neurodegenerativas (CIBERNED), Madrid, 1. Wang D, Ayers MM, Catmull DV, Hazelwood LJ, Bernard CC, Orian JM. Astrocyte-associated axonal damage in pre- Spain 5 Departments of Neurology and Immunology, Hospital onset stages of experimental autoimmune encephalomyeli- tis. Glia. 2005;51(3):235-240. https://doi.org/10.1002/glia.20199. Universitario Ramón y Cajal, Instituto Ramón y Cajal de Aug 15. Investigacion Sanitaria, Madrid, Spain 2. Tintore M, Rovira A, Rio J, et al. Defining high, medium and 6 Wellcome-Wolfson Institute for Experimental Medicine, low impact prognostic factors for developing multiple sclero- Queen’s University Belfast, Belfast, UK sis. Brain. 2015;138(7):1863-1874. https://doi.org/10.1093/brain/ 7 Proteomics Unit, Universitat Pompeu Fabra (UPF), awv105. Jul. Pt. Barcelona, Spain 3. Pelisch N, Dan T, Ichimura A, et al. Plasminogen activator 8 Proteomics Unit, Centre de Regulació Genòmica (CRG), inhibitor-1 antagonist TM5484 attenuates demyelination and axonal degeneration in a mice model of multiple sclerosis. PLoS Barcelona Institute of Science and Technology (BIST), One. 2015;10(4):e0124510. https://doi.org/10.1371/journal.pone. Barcelona, Spain 0124510. 4. Kasza A, Kiss DL, Gopalan S, et al. Mechanism of plasminogen Correspondence activator inhibitor-1 regulation by oncostatin M and interleukin- Manuel Comabella and Clara Matute-Blanch, Centre 1 in human astrocytes. J Neurochem. Nov 2002;83(3):696-703. d’Esclerosi Múltiple de Catalunya (Cemcat), https://doi.org/10.1046/j.1471-4159.2002.01163.x. 8 of 8 LETTER TO EDITOR 5. Popichak KA, Hammond SL, Moreno JA, et al. Compen- ology by orchestrating in the CNS the inflammatory response satory expression of Nur77 and Nurr1 regulates NF-kappaB- of resident and peripheral immune cells and by suppressing dependent inflammatory signaling in astrocytes. Mol Pharma- remyelination. Glia. 2014;62(3):452-467. https://doi.org/10.1002/ col. 2018;94(4):1174-1186. https://doi.org/10.1124/mol.118.112631. glia.22616. Mar. Oct. 9. Ponath G, Lincoln MR, Levine-Ritterman M, et al. Enhanced 6. Masilamani AP, Ferrarese R, Kling E, et al. KLF6 deple- astrocyte responses are driven by a genetic risk allele asso- tion promotes NF-kappaB signaling in glioblastoma. Oncogene. ciated with multiple sclerosis. Nat Commun. 2018;9(1):5337. 2017;36(25):3562-3575. https://doi.org/10.1038/onc.2016.507. Jun https://doi.org/10.1038/s41467-018-07785-8. Dec 17. 22. 10. Canto E, Tintore M, Villar LM, et al. Chitinase 3-like 1: 7. Brambilla R, Persaud T, Hu X, et al. Transgenic inhibition prognostic biomarker in clinically isolated syndromes. Brain. of astroglial NF-kappa B improves functional outcome in 2015;138(4):918-931. https://doi.org/10.1093/brain/awv017. Apr. experimental autoimmune encephalomyelitis by suppressing Pt. chronic central nervous system inflammation. J Immunol. 2009;182(5):2628-2640. https://doi.org/10.4049/jimmunol. 0802954. Mar 1. S U P P O RT I N G I N F O R M AT I O N 8. Brambilla R, Morton PD, Ashbaugh JJ, Karmally S, Lambertsen Additional supporting information may be found in the KL, Bethea JR. Astrocytes play a key role in EAE pathophysi- online version of the article at the publisher’s website.