Inside the MS Risk Blueprint: Pinpointing the Immune Cells and Brain Microglia Where Genetics Hits

Genome-wide association studies (GWAS) have been remarkably successful in multiple sclerosis (MS): the field has cataloged 201 independent genome-wide significant associations outside the MHC and 32 within the MHC, yielding 551 candidate risk genes, plus 416 additional “suggestive” effects that together explain ~50% of MS heritability. The catch is that most associated variants sit in noncoding DNA, which makes it difficult to tell which genes are actually affected—and in which cell types those effects matter most for disease biology and treatment.

A three-layer map: genetics + epigenetics + the 3D genome
This study tackled that “context problem” by integrating MS GWAS summary statistics with single-cell and bulk chromatin accessibility (open chromatin maps), histone modification profiles (epigenetic marks that label enhancers/promoters), and 3D chromatin interaction data (which links far-away regulatory DNA to the genes it physically contacts). The authors used GARFIELD to test whether MS-associated variants concentrate in regulatory DNA across many immune and brain cell types (while controlling for confounders like allele frequency and LD), then used H-MAGMA to convert variant signals into cell-specific gene lists using promoter capture Hi-C/HiChIP. For downstream clinical relevance, they built cell-specific polygenic risk scores (CPRS) and tested them in large cohorts, including an MS GWAS dataset with 14,802 MS cases and 26,703 controls, plus UK Biobank and UCSF-EPIC validation cohorts.

Where risk variants land: open chromatin highlights B cells, monocytes, and microglia
When the team asked, “In which cells are MS risk variants sitting inside open regulatory DNA?”, the strongest answer was immune cells—with a notable brain-resident immune signature. Using scATAC-seq-derived open chromatin regions, MS GWAS signals were significantly enriched across all tested peripheral immune cell types at stringent GWAS thresholds (T < 10⁻⁵ to 10⁻⁸), with naïve B cells showing slightly higher enrichment among immune subsets. In the brain, the signal was strikingly selective: enrichment was significant in microglia, but not in astrocytes, oligodendrocytes, OPCs, or neurons. In broader regulatory datasets (ENCODE/Blueprint), the pattern held—highest enrichment in B cells and monocytes—and there was also modest enrichment in some CNS accessibility datasets (including brain microvascular endothelial cells) with odds ratios ~1.34 to 2.52 at T < 10⁻⁵.

Risk variants behave like switches: enhancers and promoters are the hotspots
The next question was functional: are these variants sitting in “quiet” DNA or in regulatory elements that actually control gene activity? By overlapping MS GWAS loci with histone-mark ChIP-seq peaks, the authors found significant enrichment particularly in H3K27ac and H3K4me1 (classic active/primed enhancer marks) and H3K4me3 (an active promoter mark). Importantly, B cells consistently showed the strongest enrichment across these regulatory annotations. When mapped onto ENCODE candidate cis-regulatory elements (cCREs), MS hits were especially enriched in proximal and distal enhancer-like signatures (pELS and dELS)—again, most prominently in B cells and monocytes—supporting a model where MS risk is often mediated through enhancer/promoter “control panels,” not protein-coding changes.

From SNPs to genes: H-MAGMA expands the target list and spotlights microglial epigenetic control
Using H-MAGMA with immune and microglial 3D interaction maps, the study identified 1,247 MS-associated genes in B cells, 1,148 in monocytes, and 1,183 in microglia (FDR < 0.05). There was substantial overlap: 717 genes were shared across all three cell types, while 234 (B cell), 136 (monocyte), and 281 (microglia) genes were unique to each. These predicted gene sets also aligned well with prior MS prioritizations: they overlapped 283/551 (51.4%) for B cells, 265/551 (48.1%) for monocytes, and 294/551 (53.4%) for microglia. Pathway analysis of the shared core most strongly highlighted immune biology (with “cytokine signaling in immune system” as the top term), while unique sets pointed to cell-specific themes—most notably a microglial enrichment in “PRC2 methylates histones and DNA,” driven in part by the microglia-unique gene DNMT3A, a DNA methyltransferase linked to epigenetic regulation.

Turning biology into prediction: cell-specific polygenic risk scores (CPRS) quantify risk in real cohorts
To translate cell-specific biology into measurable human risk, the authors computed CPRS in UK Biobank phase 1 (601 MS cases / 109,990 controls) for model selection and validated in UK Biobank phase 2 (1,354 cases / 252,065 controls) and UCSF-EPIC (494 cases / 449 controls). Using CPRS built from unique SNPs per cell type, predictive performance in UKBB2 was modest but highly significant: B cell R² = 2.2% (P = 1×10⁻⁶⁹), AUC = 64%; monocyte R² = 2.4% (P = 1.68×10⁻⁸⁴), AUC = 64%; microglia R² = 2.0% (P = 7.2×10⁻⁷²), AUC = 63%; and combined R² = 2.9% (P = 2.9×10⁻¹⁰²), AUC = 65%. In UCSF-EPIC, performance improved (consistent with cleaner case definition and case/control balance): combined R² = 3.4% (P = 2×10⁻¹⁷), AUC = 69%, and monocyte CPRS reached R² = 3.1% (P = 4×10⁻¹⁶), AUC = 68%. Clinically intuitive stratification also emerged: individuals in the top 5% of CPRS had roughly a 3- to 5-fold higher MS risk than those near the median, with the monocyte-specific tail showing particularly elevated risk.

Linking DNA risk to the clinic: white matter volume and relapse activity connect genetics to disease expression
The most “patient-facing” part of the paper is the attempt to connect cell-specific genetic burden to MS phenotypes. In UCSF-EPIC MS subjects with MRI outcomes (n = 461), CPRS associations survived multiple-testing correction most clearly for white matter volume (WMV). The strongest reported relationship was for monocyte-linked genetic burden: β = −0.13 with R² = 2.1% for WMV, and when excluding the MHC region the association strengthened to β = −0.14 with R² = 2.2% (P < 0.05); an association with total brain volume also appeared (β = −0.10, R² = 1.66%, P < 0.05). Using unique SNP CPRS (Table 2), WMV associations remained notable, especially for microglia (β = −0.13, R² = 2.03%, P < 0.05) and monocytes (β = −0.11, R² = 1.54%, P < 0.05). Beyond imaging, relapse activity over a 5-year interval also tracked with cell-specific risk: B cell β = 0.29 (P < 0.005), microglia β = 0.29 (P < 0.005), and monocyte β = 0.25 (P < 0.05). The authors note key limitations—especially the lack of scATAC-seq from MS patients—but the overall message is sharp: MS genetic risk is not “everywhere,” it is concentrated in regulatory DNA active in B cells, monocytes, and microglia, and that cell-specific burden can be linked (in small but measurable effect sizes) to clinically meaningful outcomes like white matter loss.

Disclaimer: This blog post is based on the provided research article and is intended for informational purposes only. It is not intended to provide medical advice. Please consult with a healthcare professional for any health concerns.

References:
Ma, Q., Shams, H., Didonna, A. et al. Integration of epigenetic and genetic profiles identifies multiple sclerosis disease-critical cell types and genes. Commun Biol 6, 342 (2023). https://doi.org/10.1038/s42003-023-04713-5