A multi-omic approach implicates novel protein dysregulation in post-traumatic stress disorder

Abstract

Background: Post-traumatic stress disorder (PTSD) is a common and disabling psychiatric disorder. PTSD involves multiple brain regions and is often comorbid with other psychiatric disorders, such as major depressive disorder (MDD). Recent genome-wide association studies (GWASs) have identified many PTSD risk loci and transcriptomics studies of postmortem brain have found differentially expressed genes associated with PTSD cases. In this study, we integrated genome-wide measures across modalities to identify convergent molecular effects in the PTSD brain.

Methods: We performed tandem mass spectrometry (MS/MS) on a large cohort of donors (N = 66) in two prefrontal cortical areas, dorsolateral prefrontal cortex (DLPFC), and subgenual prefrontal cortex (sgPFC). We also coupled the proteomics data with transcriptomics and microRNA (miRNA) profiling from RNA-seq and small-RNA sequencing, respectively for the same cohort. Additionally, we utilized published GWAS results of multiple psychiatric disorders for integrative analysis.

Results: We found differentially expressed proteins and co-expression protein modules disrupted by PTSD. Integrative analysis with transcriptomics and miRNA data from the same cohort pointed to hsa-mir-589 as a regulatory miRNA responsible for dysregulation of neuronal protein networks for PTSD, including the gamma-aminobutyric acid (GABA) vesicular transporter, SLC32A1. In addition, we identified significant enrichment of risk genes for other psychiatric disorders, such as autism spectrum disorder (ASD) and major depressive disorder (MDD) within PTSD protein co-expression modules, suggesting shared molecular pathology.

Conclusions: We integrated genome-wide measures of mRNA and miRNA expression and proteomics profiling from PTSD, MDD, and control (CON) brains to identify convergent and divergent molecular processes across genomic modalities. We substantially expand the number of differentially expressed genes and proteins in PTSD and identify downregulation of GABAergic processes in the PTSD proteome. This provides a novel framework for future studies integrating proteomic profiling with transcriptomics and non-coding RNAs in the human brain studies.

Keywords: Major depressive disorder; MicroRNAs; Multi-omics; PTSD; Prefrontal cortex.

Fig. 1

Multiomic data overview and differential expression analysis of proteins in PTSD. A A schematic overview of the study design and analysis with multiple -omics datasets employed in this study. Dashed lines indicate integration of data-inferred associations in disease. B Distribution of gene and protein expression (all transcripts; gray), protein-coding transcripts (black) and peptides (red). C Volcano plots showing DEPs in DLPFC (top) and sgPFC (bottom) of PTSD brains. Red (P < 0.05 and log fold change > 0) indicates upregulated proteins, and blue (P < 0.05 and log fold change < − 0.18) indicates downregulated proteins. D Top significantly enriched biological pathways in DLPFC and sgPFC for PTSD and MDD differentially expressed proteins

Fig. 2

Network Co-expression analysis of proteomics data. A Module-trait correlation between protein expression correlations in the DLPFC of PTSD brains and demographic features (PrimaryDx, age of death, PMI, sex, ancestry). Color in each cell reflects correlation between module eigenprotein and PrimaryDx and the number represents significance (− log₁₀(P-value)) of that correlation. Module names are abbreviated as color codes only. B Boxplots of eigenproteins and volcano plots of differential expression patterns of the proteins in PTSD-associated modules PTSD-PM-skyblue (left) and red (right) (vertical dash lines indicate log₁₀(fold change) = ± 0.18. Module PTSD-PM-skyblue, P = 0.03; module PTSD-PM-red, P = 0.04. P indicates eigenprotein change significance between PTSD and CON. C Cell type enrichment analysis showing the enrichment of cell type markers in protein modules. Brown dashed line indicates significantly enriched markers, P < 0.05

Fig. 3

Gene expression and transcriptome network analysis of PTSD cortical regions. A Volcano plots showing DEGs in DLPFC (left) and sgPFC (right) of PTSD cortical regions. Red (P < 0.05) indicate upregulated genes and blue (P < 0.05) indicate downregulated ones. Red lines indicate P-value = 0.05 and black lines indicate |log fold change| > 0.18. B Pathway enrichment analysis of differentially expressed genes in PTSD and MDD DLPFC and their overlap with sgPFC. C Enrichment of DEGs and cell types in gene modules constructed by WGCNA for DLPFC (top) and sgPFC (bottom). Color indicates significance of enrichment (−log₁₀(P-value)). Module names are abbreviated as indices only. Comparison of transcriptional and protein differential expression are shown in PTSD DLPFC (D) and sgPFC (E))

Fig. 4

miRNA dysregulation in PTSD and its effect on the proteome. A Comparison of miRNA abundance captured using bulk mRNA-seq and smRNA-seq results. B Volcano plots showing differential expression patterns of miRNAs in PTSD DLPFC and sgPFC regions. Dashed lines: log₂(fold change) = ± 0.5 and P-value = 0.05. C Examples of correlations between miRNA MIR589 and disease-associated protein LY6H. D miRNA enrichment analysis of DEPs to identify regulatory miRNAs and their targets. Bar heights (signed − logP) indicate significances of enrichment

Fig. 5

miRNA-regulated disease-associated protein modules. A Correlations between miRNAs with PTSD-associated protein module PTSD-PM-skyblue. Colors in heatmap indicate the correlation levels between miRNAs and proteins. miRNA-protein pairs with P-value < 0.05 are plotted. B Disease-specific protein abundances (log₁₀ Intensity) of hsa-mir-589 targets (CACNA2D1, NEGR1, OPCML, CNTN1), and hsa-mir-6786 targets (ATP6V0A1, LY6H) in module PTSD-PM-skyblue. C Comparison of transcriptional and protein differential expression in PTSD module PTSD-PM-skyblue and grouped linear regression by proteins association with miRNAs. Red, proteins without miRNA association; blue, proteins with miRNA association. D Key driver analysis (KDA) plot of PTSD protein module PTSD-PM-skyblue (top) and MDD protein module MDD-PM-grey60 (bottom). Pink circles indicate key drivers. Colors of individual slices refer to putative miRNA regulation of the protein. E Top: enrichment of GWAS risk genes from TWAS analysis in the proteomic modules of PTSD (top) and MDD (bottom), including Alzheimer’s disease (ALZ), autism spectrum disorder (ASD), bipolar disorder (BIP), major depressive disorder (MDD), post-traumatic stress disorder (PTSD) and schizophrenia (SCZ). Bolded proteins are DEPs (P-value < 0.05) in DLPFC. Colors of bars correspond to their module names. miRNA name abbreviations: MIR103A1, hsa-mir-103a1; MIR218 -1, hsa-mir-218–1; MIR379, hsa-mir- 379; MIR589, has-mir-589; MIR6786, hsa-mir-6786. PM- indicates protein module