Abstract
Introduction: Disease-induced neurotoxicity represents a significant clinical challenge across various neurological conditions, causing long-term or even fatal damage to the central nervous system (CNS). Cerebrospinal fluid (CSF), a critical component of CNS homeostasis, functions by mediating immune surveillance and regulating neuroimmune responses. However, the specific characteristics of disease-associated neurotoxicity and the cellular heterogeneity within CSF remain largely uncharacterized.
Methods and Results: We established the multi-disease single-cell transcriptome atlas of CSF immunity, profiling 154,111 cells from 105 samples across 8 neurotoxic conditions, including aging, Alzheimer's disease and mild cognitive impairment (AD/MCI), CAR-T associated neurotoxicity (neuro-CAR-T), COVID with neurological sequelae (neuro-COVID19), postherpetic neuralgia (PHN), viral encephalitis (VE), multiple sclerosis (MS) and idiopathic intracranial hypertension (IIH). Leveraging disease-specific cellular heterogeneity, we identified three principal neurotoxicity drivers: i) Lipid metabolism-correlated inflammation: Characterized by the dominance of PPAR pathway enriching SPP1+ macrophages, and lipid-metabolizing COX4I1+ macrophages as key drivers of neuroinflammation (e.g., neuro-CAR-T and VE). (ii) Virally-triggered neuroinflammation: Featured by interferon signaling-mediated antiviral responses and characterized by the enrichment of anti-virus NK cells and CD14+CD16+ intermediate monocytes, particularly prominent in VE and neuro-COVID-19. iii) B-lineage dichotomy: Defined by plasma cell expansion in inflammatory contexts (e.g., MS, VE) contrasting with B cell prevalence in senescent states (e.g., aging and AD/MCI). Furthermore, integrative analysis of transcriptomic profiles and disease-related neurotoxic factors delineated four distinct molecular patterns of neurotoxic pathogenesis: (i) an interferon-dominant, virally-triggered pattern, (ii) a lipid-metabolism-dysregulated, CAR-T-induced pattern, (iii) a translation-perturbed, ribosome-related pattern, and (iv) an aging-associated pattern with prominent mitochondrial impairment. Notably, despite transcriptional differences across patterns, comparative analyses deciphered the critical convergent role of myeloid compartment in driving neuroinflammation.
Significance: Our study presented one of the largest human CSF single-cell RNA immunology atlases to date, comprising 154,111 cells across 8 neurotoxic conditions. We delineated conserved and disease-specific neurotoxic mechanisms, identifying myeloid cells as pivotal regulators of neuroinflammation. The discovered molecular patterns offer diagnostic biomarkers for early detection, reveal both universal and specific therapeutic targets, and shed light on potential precision medicine for neurotoxic complications.
