Convergent molecular, cellular, and cortical neuroimaging signatures of major depressive disorder

Abstract

Major depressive disorder emerges from the complex interactions of biological systems that span genes and molecules through cells, networks, and behavior. Establishing how neurobiological processes coalesce to contribute to depression requires a multiscale approach, encompassing measures of brain structure and function as well as genetic and cell-specific transcriptional data. Here, we examine anatomical (cortical thickness) and functional (functional variability, global brain connectivity) correlates of depression and negative affect across three population-imaging datasets: UK Biobank, Brain Genomics Superstruct Project, and Enhancing NeuroImaging through Meta Analysis (ENIGMA; combined n ≥ 23,723). Integrative analyses incorporate measures of cortical gene expression, postmortem patient transcriptional data, depression genome-wide association study (GWAS), and single-cell gene transcription. Neuroimaging correlates of depression and negative affect were consistent across three independent datasets. Linking ex vivo gene down-regulation with in vivo neuroimaging, we find that transcriptional correlates of depression imaging phenotypes track gene down-regulation in postmortem cortical samples of patients with depression. Integrated analysis of single-cell and Allen Human Brain Atlas expression data reveal somatostatin interneurons and astrocytes to be consistent cell associates of depression, through both in vivo imaging and ex vivo cortical gene dysregulation. Providing converging evidence for these observations, GWAS-derived polygenic risk for depression was enriched for genes expressed in interneurons, but not glia. Underscoring the translational potential of multiscale approaches, the transcriptional correlates of depression-linked brain function and structure were enriched for disorder-relevant molecular pathways. These findings bridge levels to connect specific genes, cell classes, and biological pathways to in vivo imaging correlates of depression.

Keywords: astrocytes; gene expression; major depressive disorder; neuroimaging; somatostatin interneurons.

Fig. 1.

Consistent imaging correlates of depression and negative affect across datasets. (A) Differences in cortical thickness between moderate/severe recurrent MDD and controls in the UKB and ENIGMA, and their spatial correlation (r_s = 0.29). Association of (B) RSFA and (C) GBC to moderate/severe MDD in the UKB, and negative affect in the GSP, and their spatial correlation (RSFA: r_s = 0.40; GBC: r_s = 0.41). Each dot represents one of 200 parcels from the functional atlas from Schaefer et al. (34). Significance was established using permuted spin tests to retain the structure of spatial autocorrelation of effects. UKB effects reflect comparison of n = 2,136 individuals with lifetime history of recurrent MDD (controls, n = 12,223). ENIGMA data are metaanalytic estimates of MDD (ns = 1,206 to 1,302) relative to controls (ns = 7,350 to 7,449) (8). GSP data reflect relationships to a continuous measure of trait negative affect in n = 947 healthy young adults. Dashed line is the regression line of best fit.

Fig. 2.

SST marker genes are spatially associated to in vivo imaging correlates of depression. (A) Normalized AHBA cortical expression of three gene markers for SST interneurons: SST (SST), cortistatin (CORT), and neuropeptide Y (NPY). Each dot on the cortical surface represents expression in a single AHBA tissue sample, which is averaged across 200 bihemispheric cortical parcels. (B) SST marker expression is spatially correlated with depression-related shifts in cortical thickness (r_avg = −0.25), RSFA (r_avg = 0.37), and GBC (r_avg = −0.38) in UKB data. Circles in the dot plots are cortical parcels, colored by relative SST expression. (C) Permutation analyses revealed that the strength of the spatial association was greater than what is expected by random selection of 10,000 triplets of brain-expressed genes, as well alternative permutation strategies (SI Appendix, Fig. S4). (D and E) SST marker spatial associations are consistent in replication data for cortical thickness (r_avg = −0.51), RSFA (r_avg = 0.21), and GBC (r_avg = −0.26). SST_mark denotes average of SST, NPY, and CORT spatial correlations. (F) Principal components analysis of AHBA transcriptional data revealed a rostrocaudal gradient of expression, which was spatially correlated to SST gene markers and cortical correlates of depression. *P < 0.05.

Fig. 3.

Association between in vivo depression-linked imaging phenotypes and ex vivo gene dysregulation in depression. (A) AHBA spatial gene expression was correlated to each the six depression-linked anatomical and functional neuroimaging maps, then averaged. (B) Standardized case−control expression differences were calculated using postmortem metaanalytic data from Gandal et al. (4). (C) Average AHBA spatial correlation to depression maps were selectively correlated to postmortem depression down-regulation (r = 0.047, P = 3.4e-8), but not that of other disorders. (D) Binned analysis revealed a parallel relationship between gene down-regulation in depression and AHBA correlates of in vivo depression effects (r_s = 0.72, P = 5.3e-7), which was also present for BD (r_s = 0.49, P = 1.7e-3). *P < 0.05. Error bars = 95% CI.

Fig. 4.

Integrative single-cell analyses implicate excitatory neurons, SST interneurons, and astrocytes. (A) AHBA genes were rank-ordered by spatial correlation to each depression MRI phenotype (e.g., UKB thickness, GSP RSFA). FGSEA identified astrocytes, OPC, and Ex8 (CBLN2+POSTN+) neurons as enriched across all modalities. RSFA gene correlates were multiplied by −1 to match the direction of thickness and GBC effects. Warm colors indicate positive enrichment, and numbers in each cell are corrected P values. (B) FGSEA enrichment plot showing that astrocyte marker genes tend to be spatially correlated to in vivo depression maps. Each black line on the x axis is the position of an astrocyte specific gene. (C) Average AHBA expression of astrocyte marker genes, which was significantly spatially correlated to each depression imaging map (r_avg = −0.20). (D) FGSEA analysis of genes down-regulated in ex vivo tissue samples from the cortex of patients revealed broad enrichment across cell classes, that was most pronounced in astrocytes and SST interneurons.

Fig. 5.

Genome-wide risk for depression is primarily enriched for inhibitory interneurons, but not glia. Polygenic cell enrichment analyses were conducted across the eight superordinate cell categories across two methods, (Left) LDSC and (Right) MAGMA. Cell-specific genes are defined using data from the MTG, DFC, and V1C. For LDSC, inhibitory interneuron markers show increased polygenic risk for depression (3). For MAGMA, inhibitory and excitatory genes show enrichment for polygenic depression risk, but the effect is limited to differentially expressed genes defined from the MTG. Dashed line reflects false-discovery rate corrected P < 0.05. Inh, inhibitory; Exc, excitatory; Ast, astrocytes; Mic, microglia; Oli, oligodendrocytes; Per, pericytes; End, endothelial.

Fig. 6.

Transcriptional correlates of in vivo depression-linked imaging phenotypes are enriched for depression-relevant pathways. (A) Genes were rank-ordered by average spatial correlation to depression imaging maps, then split into deciles. The top gene deciles had the greatest number of enrichment terms across ontological categories. (B) The top gene decile was enriched for depression and other psychiatric disorders. (C) Subset of significant enrichment terms for the top decile of MDD imaging correlated genes. Hierarchical clustering is based on overlap of genes in each category. Blue indicates that the gene is included in a given enrichment term. Full enrichment terms are available in Dataset S1.