Sergio Enrique Baranzini

Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the CNS whose risk and course are shaped by both genetic and environmental factors, among which the intestinal microbiota ...

Background: Accessing large-scale clinical and biomedical databases remains a significant barrier for clinicians and researchers, requiring substantial computational expertise. Agentic artificial intelligence frameworks, in which large language models (LLMs) orchestrate multi-step reasoning and query execution under interactive human supervision, offer the potential to democratize data access and accelerate evidence generation. Methods: We applied the Model Context Protocol (MCP) to integrate, within a single agentic workflow, an Observational Medical Outcomes Partnership (OMOP)-standardized electronic health record (EHR) database from an academic health system (>7 million subjects), with the Scalable Precision Medicine Open Knowledge Engine (SPOKE), a curated knowledge graph integrating relationships biomedical concepts from expert-maintained resources. Using this implementation (MedCP), we evaluated the approach across 100 benchmarking clinical research tasks, 617 biomedical factual accuracy questions (BiomixQA), an integrative case study linking clinical co-occurrence with molecular similarity, and a standardized protocol for generating and replicating real-world studies. Results: Across the 100-task benchmark, knowledge-graph access improved mean scores overall for GPT-5.5 and Claude Opus 4.8, though the pattern differed between models; on BiomixQA, SPOKE grounding raised multiple-choice accuracy for both models, without changing true/false accuracy. In the case study, disease co-occurrence patterns extracted from the EHR correlated with molecular network similarity, surfacing mechanistic hypotheses from real-world data. Applied to the replication of published observational studies, the research protocol compressed timelines from months to hours, lowering the technical barrier to query generation and execution while study design and interpretation remained under expert supervision. Conclusions: An agentic AI infrastructure that combines institutional EHR data with curated biomedical knowledge via MCP can serve as a transparent, domain-grounded, supervised research assistant for real-world evidence, supporting both hypothesis generation and testing.

OBJECTIVES: Short-chain fatty acids (SCFAs) are immunometabolites produced by the gut microbiome. In animal models, SCFAs affect traumatic brain injury (TBI) severity by modulating the immune response and serving as an energy source. The goal of this study was to assess whether SCFAs are associated with functional outcome in adult patients with moderate-to-severe TBI (msTBI). DESIGN: Prospective cohort study. SETTING: Urban Trauma Center. PATIENTS: Adults (age ≥ 15 yr) who had TBI with Glasgow Coma Scale 3–12, intracranial hemorrhage on head CT scan, and at least one reactive pupil. Blood samples had to be collected within 3 hours of trauma. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Univariate and multivariate analyses demonstrated that plasma SCFAs were associated with better functional outcomes at discharge and 6 months, an association driven primarily by differences in plasma acetate and propionate. K-means clustering of acetate and propionate levels identified two patient clusters with distinct discharge and 6-month functional outcomes but similar clinical, biomarker, and radiographic injury severity. Cluster 1 ( n = 47) had higher SCFA levels compared with cluster 2 ( n = 76) and cluster 1 had more favorable outcomes at discharge (Glasgow Outcome Scale 4–5: 83% vs. 55%; p = 0.003) and 6 months (Extended Glasgow Outcome Scale 4–8: 78% vs. 45%; p = 0.005). Multivariable logistic regression adjusting for the International Mission for Prognosis and Analysis of Clinical Trials in TBI (IMPACT) _lab model identified an independent association between the SCFA cluster and functional outcome at discharge ( p = 0.001) and 6 months ( p = 0.03). Adding the SCFA cluster to the IMPACT _lab model improved the area under the receiver operating characteristic curve for the prediction model for a favorable outcome. CONCLUSIONS: Our study suggests that SCFA levels are associated with functional outcome after msTBI. Future studies will focus on identifying mechanisms through which SCFAs may improve msTBI outcomes and what drives interpatient variation in their levels, which could position SCFAs as prognostic biomarkers and therapeutic targets in TBI.

Background: Mucosal-associated invariant T (MAIT) cells recognize microbial vitamin B2 and B9 metabolites via MR1 and have been implicated in multiple sclerosis (MS). How patient-specific gut microbiota shape human MAIT-cell pathogenicity at the clonal level remains unknown. Material and Methods: We generated 62 MAIT-cell clones from relapsing–remitting multiple sclerosis (RRMS) patients and 50 from healthy controls (HCs). Clones were stimulated with riboflavin- or folate-pathway metabolites, paraformaldehyde-fixed patient-matched gut bacterial isolates, or interleukin (IL)-12/IL-18. Activation markers, cytokine secretion, cytotoxicity, and competitive MR1–ligand inhibition were assessed. Intestinal permeability was evaluated using I-FABP, LBP, GLP-2, and fecal α-1-antitrypsin. Results: MS-derived MAIT clones showed markedly enhanced activation, increased interferon-gamma (IFN-γ), IL-17, and granulocyte-macrophage colony-stimulating factor (GM-CSF) secretion, and greater cytotoxicity compared with HCs when stimulated with riboflavin-producing taxa isolated from the same patients. Importantly, several responses diverged from predictions based on murine models and genomic inference, including mixed cytokine profiles and graded competitive inhibition by folate-derived ligands. These findings highlight species-specific differences in MR1 ligand handling and MAIT-cell activation. Activation required uptake of intact bacteria and acid-dependent MR1 loading. MS patients exhibited significant intestinal barrier dysfunction, linking dysbiosis to systemic MAIT-cell hyperactivation. Conclusion: At clonal resolution, this study demonstrates that patient-specific microbial metabolism, MR1–ligand competition, and epithelial barrier disruption cooperate to amplify MAIT-cell pathogenicity in MS, revealing human-specific mechanisms not predicted by animal models.

Purpose of review: This review highlights the latest advances in the genetics of multiple sclerosis (MS). While earlier research defined the polygenic architecture of disease susceptibility, recent efforts have focused on unraveling the genetics of disease progression and dissecting the interplay between host genetics and environmental triggers. Recent findings: The discovery of the first locus associated with MS severity ( DYSF-ZNF638 ) revealed central nervous system (CNS)-intrinsic mechanisms of progression distinct from immune-mediated susceptibility. Recently, a multi-ancestry genome-wide association studies confirmed most of the previously identified susceptibility loci (most of which are involved in immune biology) and identified 4 additional associations. Mounting evidence on EBV-host genetic convergence highlights critical gene-environment interactions in disease etiopathogenesis. Summary: Advances in genetics are broadening the conceptual framework of MS pathogenesis, separating the immunological triggers of disease susceptibility from distinct, CNS-specific drivers of disease progression. Future efforts leveraging longitudinal, deeply phenotyped cohorts and functional validations will empower translational applications for prevention, prognostication, and treatment.

SUMMARY: Women are at increased risk of autoimmune diseases, including multiple sclerosis (MS); however, the degree to which sex differences in the gut microbiota impact autoimmunity remains largely unexplored. Our 27-cohort meta-analysis revealed 60 sex-associated gut bacterial species. Leveraging an independent clinical cohort, we demonstrate that female-enriched species significantly associate with MS status and clinical disability (EDSS). Top female-enriched species Eggerthella lenta drove disease in the experimental autoimmune encephalomyelitis (EAE) MS model, consistent with brain and gut lamina propria T cell infiltration and MS-associated T helper (Th) signatures. E. lenta induced intestinal Th1 and Th17 in healthy mice, independent of bacterial viability. Mechanistically, we demonstrate that TLR2 directly drives E. lenta -induced IFN-γ production in Th cells and is necessary for exacerbation of EAE. Together, we identify a causal host-microbe axis contributing to sex differences in autoimmunity and provide a framework for evaluating sex as a biological variable in human microbiome research. HIGHLIGHTS: 27-cohort meta-analysis identifies a robust sex-signature in human gut microbiota. Female-enriched species are associated with MS risk and severity. Female-enriched Eggerthella lenta exacerbates the EAE model. E. lenta impacts neuroinflammation via toll-like receptor 2.

Time-restricted feeding (TRF) improves metabolic health, yet the pathways linking feeding rhythms to host–microbe metabolic integration remain unclear. Using clinical profiling, humanized mouse models, and network analysis, we characterize a liver-driven metabolic axis mediating TRF’s benefits in type 2 diabetes (T2DM). We show that hepatic HNF4α–CYP8B1 pathway dysregulation contributes to T2DM by increasing 12α-hydroxylated bile acids and altering the gut microbial environment, specifically depleting Parabacteroides distasonis . TRF suppresses this hepatic pathway, shifting the bile acid pool toward non-12α-hydroxylated species like ursodeoxycholic acid (UDCA), thereby creating a permissive niche for the restoration of beneficial gut microbes, including P. distasonis . Although supplementation with P. distasonis or UDCA provides partial metabolic improvements, only TRF corrects the upstream hepatic driver, HNF4α. These findings suggest TRF improves T2DM by realigning the liver–bile acid–microbiota axis, highlighting feeding timing as a regulator of host and microbial metabolism.