Rank-Sensitive and Cell-Type-Localized Coupling between Vitamin D-Related Transcriptional Programs and Brain Molecular Alterations in Major Depressive Disorder

Published: 26 May 2026| Version 1 | DOI: 10.17632/4jdp3752bp.1
Contributor:
Gao Biao

Description

Background The relationship between vitamin D (VD) and major depressive disorder (MDD) remains biologically plausible but clinically unresolved. We asked whether VD-related transcriptional programs show rank-level coupling with MDD-associated molecular alterations in human brain tissue and whether candidate signals can be localized to specific cellular compartments. Methods Public bulk and single-nucleus RNA-seq datasets were integrated across predefined evidence layers. VD-related signatures were tested in GSE102556 as the multi-region bulk discovery layer, GSE101521 as the independent DLPFC bulk validation layer, and GSE144136 as the male DLPFC snRNA-seq localization layer. Bulk analyses used limma-based modeling and preranked enrichment. snRNA-seq localization used donor-level module scoring at the donor × major cell-type level, and restricted pseudobulk sensitivity used edgeR/limma-voom followed by focused fgsea. Nuclei were not treated as independent biological replicates. Results GSE102556 nominated VD-related rank-level signals, whereas GSE101521 showed lower-in-MDD enrichment for Tier 3 neuroimmune/neuroplasticity and refined main VD signatures, opposite to the GSE102556 pan-region direction. Thus, the bulk evidence indicated rank sensitivity with directional discordance rather than simple replication. In GSE144136, donor-level module scoring identified two within-signature FDR-controlled lower-in-MDD_suicide localization signals: the VD core metabolism/receptor axis in oligodendrocytes and a shared discordant main VD leading-edge program in excitatory neurons. Restricted pseudobulk sensitivity supported the excitatory-neuron leading-edge signal (NES = −2.154; global FDR = 0.000281), whereas the oligodendrocyte VD core-axis signal was not evaluable after pseudobulk filtering because of insufficient ranked-universe detectability. The original GSE144136 study provides relevant neuronal and oligodendrocyte-lineage context, but not direct replication of the mature oligodendrocyte VD core-axis signal. Conclusions VD-related transcriptional programs appear rank-sensitive to MDD-associated brain molecular alterations, but the direction of coupling is heterogeneous across datasets. snRNA-level analyses localized candidate signals to oligodendrocytes and excitatory neurons. These findings are hypothesis-generating transcriptomic evidence, not proof of vitamin D deficiency effects, supplementation response, or cell-type-specific causality.

Files

Steps to reproduce

Download all R scripts and the supplementary table from the Mendeley Data repository. Set the working directory to the project folder, install the required R packages listed in the scripts, and run the scripts sequentially according to their step numbers. Public GEO input data are retrieved or processed as specified in the scripts. The final outputs correspond to the submitted supplementary workbook Supplementary_Tables_S1_to_S9_SUBMISSION_READY_LEAN_submission_clean_REPAIRED.xlsx.

Institutions

Categories

Depression, RNA Sequencing, Cell Type (Life Sciences), Vitamin D, Single-Cell Transcriptomics

Licence