Chitinase 3-like 1 is neurotoxic in multiple sclerosis patient-derived cortical neurons

Dear Editor,

We are pleased to present our latest findings regarding the neurotoxic role of Chitinase 3-like 1 (CHI3L1) in multiple sclerosis (MS). CHI3L1, a 40 kD glycoprotein, is primarily produced by activated astrocytes and microglia in the central nervous system (CNS), and it has garnered considerable attention due to its implications in inflammation and tissue remodelling.¹ It is notably increased in several conditions, including MS, and accumulating evidence supports CHI3L1 as a biomarker in early MS, with elevated cerebrospinal fluid (CSF) levels associated with increased disability risk.^{2, 3} This association led us to investigate whether CHI3L1 simply reflects glial activation or if it exerts direct neurotoxicity. Our prior work in murine neurons demonstrated CHI3L1's neurotoxic effects,⁴ prompting us to explore its impact on MS patient-derived human induced pluripotent stem cells (hiPSC). Here, we aim to characterize these effects at both molecular and functional levels, further exploring CHI3L1's potential as a biomarker and therapeutic target for MS.

The first step in our investigation involved refining a human neuronal model using two MS-derived hiPSC lines (Table S1), MS-10 and MS-6, matured for 28 and 40 days (Figure S1A). To ensure the model's suitability, we meticulously characterized the neuronal cultures through immunofluorescence and calcium imaging, evaluating neuronal and astrocytic proportions (Figure S1B–E,J,L), cortical fate (Figure S1F–I,K,M), dendrite growth (Figure S2A), synaptic (Figure S2B) and neurotransmitter markers (Figure S3) and the onset of sporadic and synchronous neuronal activity (Figure S4). The neuronal cultures exhibited varying percentages of astrocytes, which increased over time for both cell lines (Figure S1J,L). Cortical fate was delineated by robust Tbr1 immunoreactivity alongside limited CTIP2 expression (Figure S1K,M). Notably, dendritic growth persisted until day 28 (Figure S2A), coinciding with the expression of synaptic markers like synapsin and PSD-95 (Figure S2B). By day 28, most neurons from each line co-expressed glutamatergic and GABAergic markers, but by day 40, the loss of vGAT immunoreactivity suggested a predominant population of glutamatergic cells (Figure S3). Additionally, fluorescence calcium imaging revealed a progressive increase in spontaneous and synchronous neuronal activity over time, culminating in a sporadic and synchronous pattern by day 40 (Figure S4).

Our investigation progressed to examine the impact of CHI3L1 on neuronal morphology and synaptic plasticity by treating MS-10 and MS-6 neuronal cultures at day 28 with CHI3L1 (300 ng/mL) or vehicle (PBS) for 24 and 72 h (Figure 1A). The analysis revealed a 17.5% reduction in dendritic arborization by 24 h and a 19% reduction by 72 h, along with a 16.5% decrease in dendrite length at 72 h (Figure 1B–D). Additionally, synaptic plasticity assessment unveiled CHI3L1-induced decreases in synapsin area (23.3%) and active synapses (47.9%) at 72 h, accompanied by a trend towards decreased PSD-95 levels (Figure 1E–H), indicating compromised structural integrity and synaptic function. The CHI3L1-induced reductions in dendritic arborization and synaptic connectivity are consistent with hallmark features of neurodegenerative diseases, including MS.

We then investigated the impact of CHI3L1 on neuronal and population activity. MS-10 and MS-6 neuronal cultures at day 40 were treated with CHI3L1 (300 ng/mL) or vehicle (PBS) for 4, 24 and 72 h (Figure 2A). Using fluorescent calcium imaging, we monitored neuronal activity and applied advanced computational techniques⁵ for analysis. Our results revealed a notable increase in fluorescence amplitude within CHI3L1-treated cultures, which was only significant at the 4-h mark (Figure 2B), indicative of heightened excitability. While no significant differences were observed in the percentage of active neurons (Figure 2C), mean neuronal activity (Figure 2D) or the inter-burst interval (IBI; Figure 2E), our network behaviour analysis unveiled dynamic shifts over time. Initially, there was an increase in the strength of collective events (SCE), followed by a gradual decline at 24 and 72 h (Figure 2F). Although our assessment of effective connectivity did not reveal significant differences, the observed trends suggest altered network integration dynamics over time (Figure 2G–I), mirroring the shifts noted in collective behaviour.

Following the characterization of CHI3L1-induced neurotoxic effects, we investigated the underlying molecular mechanisms. Neuronal cultures derived from MS-10 cells at day 28 were treated with CHI3L1 (300 ng/mL) or vehicle (PBS) for 12 and 24 h, followed by gene expression analysis using microarrays. This analysis identified numerous differentially expressed genes (DEGs) linked to neurodegenerative disorders and synaptic activity (Tables S2–5). Seven DEGs (RIOK2, DENND2C, CFAP61, RASA2, LRRC66, UHMK1 and GNMT) were validated through quantitative real-time polymerase chain reaction (Figure 3A). Notably, some of the validated DEGs, such as RIOK2,⁶ have been reported to be involved in neurodegenerative disorders, while others like UHMK1⁷ and GNMT⁸ play roles in neurite growth and neurogenesis. Functional analysis revealed enrichment in categories related to receptor-ligand activity, signalling receptor activator activity and proinflammatory processes at 12 h (Figure 3B), transitioning to synaptic activity-related processes at 24 h (Figure 3C), as confirmed by Gene Set Enrichment Analysis (GSEA) (Figure 3D). These findings underscore CHI3L1's impact on genes and pathways critical to neuronal function and neurodegeneration, particularly at the 24-h mark.

In parallel, we conducted a protein phosphorylation array analysis to explore the signalling pathways underlying CHI3L1's neurotoxicity. MS-10-derived neuronal cultures exposed to CHI3L1 (300 ng/mL) or vehicle (PBS) at day 28 were analyzed for key protein phosphorylation levels at early time points. We observed consistent increases in the phosphorylation levels of STAT1, particularly at Y701 (Figure 4A). Immunoblot analyses confirmed STAT1-Y701 phosphorylation at 2 h post-exposure (Figure 4B). TRRUST enrichment analysis identified IRF1 and STAT1 as potential transcription factors governing the gene expression response to CHI3L1 treatment at 24 h, implicating the interferon response pathway (Figure 4C). Comparison with the Transcription Factor Target Gene Database supported the involvement of IRF1 and STAT1 in mediating CHI3L1's effects on hiPSC-derived neurons (Figure 4D).

The elucidation of neurodegenerative processes in MS is vital for targeted therapeutic strategies. In this study, we observed neurotoxic effects on dendritic morphology, synaptic function and neuronal excitability, indicative of its potential as an MS prognostic biomarker. Transcriptomic analyses unveiled a complex signature involving pathways and genes related to inflammation and synaptic function, alongside activation of STAT1 post-CHI3L1 treatment. Understanding its intricate molecular mechanisms may unveil new therapeutic targets for inhibiting CHI3L1-mediated neuronal signalling, offering promising avenues for targeted interventions in MS.