Integrative plasma proteomics and genomics identify causal drivers and drug targets for major psychiatric conditions

Introduction

Despite decades of psychopharmacological development, most therapies still target pathways identified in early drug discoveries, and many patients do not respond adequately to treatment, suggesting that additional mechanisms are involved. Understanding the biological pathways underlying psychiatric conditions is therefore an important research priority for identifying effective treatment targets. Advances in large-scale population cohorts, genomics, and proteomics, together with developments in causal inference methods, have enabled the identification of numerous plausible targets, but prioritising those with the strongest evidence remains a key challenge. Using large-scale integrative proteomic and genetic analyses of the most comprehensive collection of plasma proteins to date, we aimed to identify proteins associated with major psychiatric conditions and prioritise them as potential causal drivers and novel drug targets.

Methods

We analysed 2,920 plasma proteins assayed using Olink in 52,615 UK Biobank participants with available proteomic data. We examined cross-sectional associations with prevalent lifetime anxiety, depression, bipolar disorder, and psychotic disorders at baseline (blood collection) using multivariable logistic regression, and longitudinal associations with incident cases during follow-up using Cox proportional hazards models. For proteins associated with psychiatric conditions, we conducted pathway enrichment analyses to characterise shared biological processes and applied two-sample Mendelian randomisation using common cis- and trans- protein coding variants to assess potential causal effects, complemented by genetic colocalisation and sensitivity analyses to evaluate horizontal pleiotropy and reverse causation. Additional genetic analyses using rare and ultra-rare protein-coding variants and gene expression in blood and brain were conducted to strengthen causal inference. Multiple testing correction was applied across all analyses. We integrated findings across proteomic and genetic approaches to prioritise proteins based on consistency of evidence and evaluated therapeutic tractability for those with supporting evidence.

Results

We identified widespread cross-sectional and longitudinal associations between circulating plasma proteins and major psychiatric conditions, with 3,349 protein-condition associations involving 1,545 proteins passing multiple testing correction, many of which remained consistent after covariate adjustment and sensitivity analyses. Pathway enrichment analyses of associated proteins highlighted immune, inflammatory, and extracellular matrix processes across psychiatric conditions. Using Mendelian randomisation and genetic colocalisation analyses, we identified 24 proteins meeting stringent criteria of causality in at least one psychiatric condition. Of these, five proteins, DDR1, LTB, PDIA3, DSG3, and PBXIP, showed concordant directional effects across proteomic and genetic analyses. Among these, DDR1 showed the most robust and consistent signal for depression, with directionally concordant evidence across cross-sectional and longitudinal associations, MR using common and rare variants, and genetic colocalisation. DDR1 (discoidin domain receptor 1) is a collagen-activated receptor tyrosine kinase involved in cell-matrix interactions and inflammatory signalling, providing a plausible biological link to psychiatric disease processes, and representing a potentially druggable candidate [1,2].

Conclusions

In this large-scale integrative study combining proteomic and genetic analyses, we identified and prioritised circulating plasma proteins potentially involved in the aetiology of major psychiatric disorders. Our findings highlight biologically plausible and potentially druggable targets, particularly within immune, inflammatory, and extracellular matrix pathways, and suggest opportunities for drug repurposing.

References

1. Wu, D., Ding, Z., Lu, T., Chen, Y., Zhang, F., & Lu, S. (2024). DDR1-targeted therapies: current limitations and future potential. Drug discovery today, 29(5), 103975. https://doi.org/10.1016/j.drudis.2024.103975
2. Fowler, A. J., Hebron, M., Balaraman, K., Shi, W., Missner, A. A., Greenzaid, J. D., Chiu, T. L., Ullman, C., Weatherdon, E., Duka, V., Torres-Yaghi, Y., Pagan, F. L., Liu, X., Ressom, H., Ahn, J., Wolf, C., & Moussa, C. (2020). Discoidin Domain Receptor 1 is a therapeutic target for neurodegenerative diseases. Human molecular genetics, 29(17), 2882–2898. https://doi.org/10.1093/hmg/ddaa177