Monocytes, Not T Cells, Show Strongest Gene Disruption in Multiple Sclerosis

Multiple sclerosis has long been framed as a disease of wayward lymphocytes, but a new systems-level analysis of immune cells from patients suggests that the loudest molecular alarm is ringing in a different compartment altogether. By reanalysing bulk RNA-sequencing data from CD14+ monocytes, CD4+ T cells and CD8+ T cells drawn from the same large public cohort, researchers report that the myeloid cells—rather than the T-cell populations that dominate most therapeutic thinking—display the most extensive transcriptional, pathway and regulatory disruption in the disease. The study, published in Discover Immunity, was released as an open-access research article on 18 October 2026.

The team drew on dataset GSE137143 from the Gene Expression Omnibus, comprising 427 patient samples across the three immune cell types. After quality control, the analysed material included 120 multiple sclerosis samples and 21 healthy controls for CD4+ T cells, 121 cases and 21 controls for CD8+ T cells, and 122 cases and 22 controls for CD14+ monocytes. Principal component analysis flagged 16 noisy samples, which were removed iteratively by two independent reviewers until each cell type formed a compact, homogeneous cluster; the removed samples included both patients and controls, which the authors say rules out disease-related selection bias. Differential expression was then computed with the DESeq2 package in R.

The resulting gene lists differed sharply in character across cell types. In CD14+ monocytes, 428 differentially expressed genes emerged, 296 of them upregulated and 132 downregulated. CD8+ T cells yielded 267 differentially expressed genes, with 116 upregulated and 151 downregulated, while CD4+ T cells produced 346, of which 98 were upregulated and 248 downregulated. Notably, the T-cell changes skewed toward gene loss, whereas the monocyte signature was dominated by activation—consistent with the authors’ hypothesis that the myeloid compartment, an early innate driver of central nervous system inflammation, would prove the most extensively rewired of the three.

Functional enrichment painted a coherent picture of myeloid stress. Upregulated genes in CD14+ cells mapped to 17 Kyoto Encyclopedia of Genes and Genomes pathways, including cytokine–cytokine receptor interaction, NF-kappa B signalling and cell adhesion molecules, all pointing toward heightened inflammatory signalling. Downregulated genes yielded no significant KEGG pathways but produced nine biological-process terms centred on lipoprotein metabolism, including negative regulation of lipoprotein particle clearance and negative regulation of low-density lipoprotein receptor activity. The authors note that lipid accumulation in macrophages can increase pro-inflammatory cytokine production, and that lipoprotein handling in the central nervous system intersects with immune cell function, though these links remain interpretive rather than demonstrated in this dataset.

To move beyond simple gene lists, the researchers built protein–protein interaction networks for each cell type using the STRING database. The CD14+ network contained 86 nodes and 166 edges; CD8+ T cells produced 104 nodes and 40 edges; CD4+ T cells yielded 50 nodes and 110 edges. Hub genes were selected by an objective betweenness-centrality drop-off criterion, yielding 15 hubs for monocytes, 10 for CD8+ T cells and 10 for CD4+ T cells. In monocytes, CTLA4 topped the ranking with a betweenness score of 2399 and was overexpressed in the disease state—an intriguing result, since CTLA4 is classically a T-cell checkpoint molecule, although prior work cited by the authors shows monocytes can express it constitutively and secrete it in soluble form. Other monocyte hubs included KIF5A, THBS1, EGR2, KCNA2, OCLN, FLT4, LEF1 and NEFL, with IL13 and FGF8 downregulated.

Clustering analysis with the MCODE algorithm found three significant subnetworks only within the monocyte network; no comparable clusters passed the threshold in either T-cell type. Functional annotation of these clusters revealed distinct programs: cluster 1 enriched for NF-kappa B and chemokine signalling alongside regulation of apoptotic process and cytokine production; cluster 2 for Wnt signalling; and cluster 3 for calcium signalling, including calcium ion transmembrane import into the cytosol. Within the shared biology of clusters 2 and 3, the calcium channel gene CACNA1H and the Wnt effector LEF1 were overexpressed, while SOST, a Wnt inhibitor, was downregulated—a combination the authors interpret as potential disinhibition of Wnt activity in monocytes, though they emphasise that the functional consequences remain untested.

Upstream regulatory analysis, drawing on eight transcription factor, kinase and long non-coding RNA databases, again singled out the myeloid compartment. Five transcription factors—LEF1, ID1, EGR3, FOXJ1 and EGR2—were identified in CD14+ monocytes, whereas only LEF1 and ID1 appeared among differentially expressed regulators there, and CD4+ T cells yielded no regulatory elements beyond the network analysis. In CD8+ T cells, the transcription factor ZBTB7C, the kinase ALPK2 and the lncRNA SMIM2-AS1 emerged as candidate regulators. Within the monocyte clusters, LEF1 and the lncRNA TRBV11-2 were shared between clusters 1 and 2, while cluster 3 showed downregulation of the transcription factor ZNF483 alongside upregulation of the potassium channel genes KCNA2 and KCNMA1. The lncRNA LINC00200, overexpressed in monocytes, showed an inverse expression pattern with the downregulated anti-inflammatory cytokine IL13, leading the authors to propose—explicitly as a hypothesis—a possible suppressive regulatory relationship.

Because the primary analysis rested on a single cohort, the team performed a limited external assessment rather than a full independent validation. Queries of CellMarker 2.0 and MSigDB provided contextual support for cell-type expression patterns of selected candidates, and two additional datasets, GSE26927 and GSE1910 along with a macrophage-versus-monocyte RNA-seq comparison, were used to probe specific genes. For example, EGR2 and LEF1 showed differential expression between CD14+ monocytes and other immune cells including microglia in normal reference data, and EGR1, flagged as a downregulated hub in CD4+ T cells, appeared as a significant gene with a 0.5 log2 fold change in an independent multiple sclerosis versus control dataset. The authors are explicit that no experimental validation of the hub genes or predicted regulatory relationships was performed in the study itself.

Two findings receive extended discussion as hypothesis-generating leads. RNF150, an E3 ubiquitin ligase, was markedly overexpressed—roughly threefold in log2 terms—specifically in CD14+ monocytes from patients. Prior mouse and human work cited in the paper links RNF150 to pro-inflammatory macrophage polarisation through degradation of the RNA-binding protein ELAVL1, but the authors stress that whether the same mechanism operates in multiple sclerosis monocytes is untested and requires direct functional validation. Separately, the hominoid-specific gene TBC1D3 showed a striking dichotomy: approximately 22-fold upregulation in both T-cell subsets, concurrent with downregulation of the EGR transcription factor family, versus a more moderate six-fold increase in monocytes accompanied by upregulation of EGR2 and EGR3. The authors frame TBC1D3 as a possible context-dependent rheostat—potentially amplifying pathogenic T-cell states while supporting compensatory myeloid responses—but again present this as interpretation awaiting experimental testing.

The study’s limitations are clearly stated. The primary differential-expression and regulatory-network analyses derive from a single large publicly available cohort; hub-gene prioritisation and inferred regulatory networks were not independently reconstructed or experimentally validated; and the external checks were selective rather than comprehensive. Within those bounds, the authors position their work as a systems-level framework: a demonstration that CD14+ monocytes carry the most pronounced transcriptional, functional and regulatory alterations among the three examined populations, and an argument that myeloid-centred programs—inflammatory, metabolic, Wnt-related and calcium-related—deserve a larger share of mechanistic attention in multiple sclerosis, a disease whose current licensed therapies predominantly target lymphocyte trafficking or depletion. Whether the identified hub genes and regulators hold up as functional drivers or biomarkers will depend on the independent cohorts and experimental models the authors themselves call for.

Subject of Research: Systems biology analysis of gene expression and regulatory networks in CD14+ monocytes, CD4+ T cells and CD8+ T cells in multiple sclerosis

Article Title: Immune cell heterogeneity in multiple sclerosis reveals myeloid-specific dysfunction and distinct regulatory programs across CD14+, CD4+, and CD8+ immune cells

Article References: Salari, A., Yazdani, M., Sharifipasandi, M., Akbari, M., Saadati, P., Sahabjo, A., Shahsavari, M. H., Mousavi, S. M., Tabatabayi, M. A., Estahri, B., Babayi, S., Jalali, A., & Shiran, M. G. (2026). Immune cell heterogeneity in multiple sclerosis reveals myeloid-specific dysfunction and distinct regulatory programs across CD14+, CD4+, and CD8+ immune cells. Discover Immunity, 3(1), Article 19. https://doi.org/10.1007/s44368-026-00037-8

Image Credits: AI Generated

DOI: 10.1007/s44368-026-00037-8

Keywords: multiple sclerosis, CD14+ monocytes, CD4+ T cells, CD8+ T cells, bulk RNA-seq, differential expression, protein-protein interaction networks, hub genes, Wnt signalling, calcium signalling, transcription factors, neuroinflammation