Inflammation in multiple sclerosis induces a specific reactive astrocyte state driving non‐cell‐autonomous neuronal damage

An in-depth understanding of the neurodegenerative component of multiple sclerosis (MS) is crucial for the design of therapeutic approaches that may stop disease progression. Astrocytes have emerged as key contributors to the pathogenesis of MS. 1 However, the mechanisms underlying the regulation of maladaptive astrocytic responses remain unknown. In this report, we show that a high inflammatory activity in MS patients at disease onset induces a specific reactive astrocyte state that triggers synaptopathy and contributes to neuronal damage in vitro and ex vivo suggesting potential mechanisms that may ulti-mately lead to neurodegeneration. To


Inflammation in multiple sclerosis induces a specific reactive astrocyte state driving non-cell-autonomous neuronal damage
Dear Editor An in-depth understanding of the neurodegenerative component of multiple sclerosis (MS) is crucial for the design of therapeutic approaches that may stop disease progression. Astrocytes have emerged as key contributors to the pathogenesis of MS. 1 However, the mechanisms underlying the regulation of maladaptive astrocytic responses remain unknown. In this report, we show that a high inflammatory activity in MS patients at disease onset induces a specific reactive astrocyte state that triggers synaptopathy and contributes to neuronal damage in vitro and ex vivo suggesting potential mechanisms that may ultimately lead to neurodegeneration.
To investigate whether astrocytes are essential contributors to neuronal damage in MS, we cultured purified astrocytes with cerebrospinal fluid (CSF) samples from MS patients with high inflammatory activity at disease onset (MS-High, Table S1). Then, we examined the effect of astrocytic secretomes on neurons ( Figure 1A). Astrocytes became reactive upon high inflammatory CSF exposure ( Figure 1B) and induced morphological alterations typically observed in neurodegenerative disorders, such as a less complex dendritic tree due to decreased arborisation ( Figure 1C, D). Moreover, these abnormalities were accompanied with synaptic plasticity impairment ( Figure 1E, F). Considering that a high lesion load at disease onset has been associated with an increased risk of neurological disability development, 2 we assessed whether the non-cellautonomous effect on neuronal plasticity could be influenced by the degree of inflammatory activity of MS patients ( Figure 2A and Table S1). Interestingly, we observed a direct correlation between the degree of inflammatory exposure and the extent of both astrocyte-mediated synaptopathy ( Figure 2B, C) and dendrite arborisation impairment ( Figure 2D, E).
We next characterised the secretomes from astrocytes exposed to high inflammatory MS microenvironment and found an altered pro-inflammatory profile comprised of 23 upregulated factors ( Figure 3A). Functional enrichment and interactome analysis revealed a set of proinflammatory pathways enriched following the MS-High CSF exposure ( Figure S1). Moreover, nuclear factor NFkappa-B p105 subunit (Nfkb1) and cellular tumour antigen p53 (Trp53) were identified as the transcription factors regulating the MS-High-associated astrocyte secretome ( Figure 3B), both involved in NF-ĸB signalling.
SerpinE1, also known as plasminogen activator inhibitor 1 (PAI-1), which has been shown to exacerbate axonal damage and demyelination in MS animal models 3 and be regulated by NF-ĸB in neuroinflammation, 4 was significantly increased in the MS-High secretomes ( Figure 3A and Table  S2). Considering its potential role as a mediator of neurodegeneration, we validated by ELISA SerpinE1 increased levels in secretomes from astrocytes exposed to MS-High condition ( Figure 3C).
By using omics technologies, we studied whether secretomes that alter neuronal plasticity are associated with a specific reactive astrocyte state in MS patients with high inflammatory activity. Astrocytes stimulated with MS-High CSF exhibited a specific reactive gene ( Figure 3D) and protein ( Figure 3E) expression profile. We identified a MS-High-associated reactive gene signature comprised of 7 differentially expressed genes ( Figure 3F) that were validated by qPCR ( Figure 3G). This reactive gene expression fingerprint was mostly comprised of downregulated immediate early response genes (Nr4a1, Klf6, Egr2 and Fosb). Interestingly, Nr4a1 and Klf6 have been reported to promote anti-inflammatory responses by specifically repressing NF-ĸB activity. 5,6 To further decipher the MS-Highspecific reactive astrocyte state, we performed a functional enrichment analysis integrating all datasets obtained from CSF exposed astrocytes: secretomes and gene/protein expression. Overall, this revealed a prominent inflammatory signature in MS-High astrocytes ( Figure S2). 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).  , Nfkb1, Trp53 and transcription factor p65 (Rela), were identified as the transcription factors regulating the MS-High astrocyte-specific fingerprint ( Figure 3H). The inhibition of NF-кB activation in astrocytes ameliorated immune infiltrate, axonal damage and demyelination, by preventing the establishment of the astrocytemediated pro-inflammatory microenvironment that leads to disease progression in EAE. 7,8 Moreover, a common MS risk variant (rs7665090) has been found associated to increased NF-кB signalling in astrocytes, driving increased lymphocyte infiltrate and lesion size. 9 These findings provide evidence that a high inflammatory microenvironment in MS patients may mediate disease progression by enhancing NF-кB signalling in astrocytes, which modifies their secretome content resulting in both immunemediated neurodegeneration and potential direct neurotoxic effects.
Next, we investigated whether this reactive astrocyte state is associated with a specific CSF proteome in MS patients with high inflammatory activity. LC/MS analysis showed that CSF from MS-High patients have a specific proteome profile ( Figure 4A, B). To elucidate the mechanisms underlying astrocyte reactivity, we performed an integrative omics data analysis at CSF, reactive astrocytes, and secretomes levels. Mitogen-activated protein kinase (ERK)-1/2 cascade, NF-ĸB-inducing kinase (NIK)/NF-ĸB signalling, Nfkb1 and Trp53 were found as the most significantly enriched pathways and transcription factors by the exposure to a highly inflammatory MS microenvironment ( Figure 4C and Table S3). These data reinforced the role of an enhanced NF-ĸB signalling in the MS-High reactive astrocytes.
Remarkably, we identified the prognostic biomarker chitinase 3-like 1 (CHI3L1) 10 upregulated in the MS-High CSF ( Figure 4B) and represented in the aforementioned pathways (Table S3). To investigate whether CHI3L1 could be a mediator of the MS-High reactive astrocyte state, we stimulated astrocytes with CHI3L1 at concentrations above the cut-off value that demonstrated prognostic implica-tions in MS patients 10 ( Figure 4D). CHI3L1 stimulation (600 ng/ml) downregulated Nr4a1 and Klf6 expression, both involved in the inhibition of NF-ĸB signalling 5,6 ( Figure 4E and Figure S3). Protein interactome computation revealed an interaction between the NF-ĸB transcription module (Nr4a1 and Klf6) potentially controlled by CHI3L1 ( Figure 4F). Noteworthy, we also found interactions between Mapk3 (ERK1), Mapk1 (ERK2) and Nr4a1 in the NF-ĸB transcription module, which might be regulated by CHI3L1 and SerpinE1 ( Figure 4G). Finally, to address whether CHI3L1 could be a potential driver of astrocytemediated neuronal damage we used P7 murine myelinating organotypic brain slice cultures that generate compact myelin ex vivo and mimic in vivo microenvironment. After 48 h, CHI3L1 (600 ng/ml) induced axonal damage reducing total neurofilament area ( Figure 4H).
Our findings provide evidence that the degree of inflammatory activity in MS patients at disease onset has the potential to induce a specific reactive state in astrocytes that trigger neuronal damage ( Figure S4). This reactive state, mainly associated with the NF-ĸB signalling, could be exploited as a prognostic biomarker that reflects a potential detrimental effect of MS astrocytes on neuronal plasticity.

A C K N O W L E D G E M E N T S
The authors thank the Advanced Optic Microscopy Unit (Campus Casanova) of the Centre Científic i Tecnològic (CCIT, Universitat de Barcelona) for their advice with microscopy techniques.

C O N F L I C T O F I N T E R E S T
The authors report no competing interests. G.G.D.B. is now an employee of Evotec.

F U N D I N G I N F O R M AT I O N
The study was funded by the 'Red Española de Esclerosis Múltiple (REEM)' sponsored by the 'Fondo de Investigación 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 (μm 2 ) 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 Ministry of Science and Innovation, Spain; the 'Ajuts