Atlas of Proteomic signatures of brain structure and its links to brain disorders

Abstract

Individual variation in brain structure influences deterioration due to disease and comprehensive profiling of the associated proteomic signature advances mechanistic understanding. Here, using data from 4997 UK Biobank participants, we analyzed the associations between 2920 plasma proteins and 272 neuroimaging-derived brain structure measures. We identified 5358 associations between 1143 proteins and 256 brain structure measures, with NCAN and LEP proteins showing the most associations. Functional enrichment implicated these proteins in neurogenesis, immune/apoptotic processes and neurons. Furthermore, bidirectional Mendelian randomization revealed 33 associations between 32 proteins and 23 brain structure measures, and 21 associations between nine brain structure associated proteins and ten brain disorders. Moreover, the significant associations between the identified proteins and mental health were mediated by brain volume and surface area. In summary, this study generates a comprehensive atlas mapping the patterns of association between proteome and brain structure, highlighting their potential value for studying brain disorders.

Fig. 1. Schematic overview of study design.

A The association between 2920 plasma proteins and measures of brain structure from five categories. B The annotation of preferred chromosome positions and functional categories for proteins associated with brain structure. Functional categories were annotated based on Gene Ontology (https://geneontology.org). C The protein-brain structure associations satisfied FDR-corrected P < 0.05 were further used for MR analysis that enhanced the relationships between plasma proteins and brain structures. Since the MR requires the cohort of exposure and outcome not to overlap, the GWAS of brain structure measures were conducted based on UKB participants for whom neuroimaging data was available but proteomic data was not. The proteins exhibited significant MR associations with brain structure measures (FDR-corrected P < 0.05) and were eligible for the following three sub-analyses, i.e., SMR, protein-disease MR and mediation analyses. D To explain the possible mechanism of why plasma proteins and brain structure were connected, SMR analysis was performed to examine whether the pQTL of brain structure-related plasma proteins were associated with eQTL of brain cortex because of pleiotropy. E The MR analysis investigating whether proteins exhibiting significant association with brain structure also affect the risk of brain disorders. F The mediation analysis investigates whether the brain structure could mediate the association between plasma proteins associated with brain structure and mental health. The mediator and outcome were constructed as latent variables. FA fractional anisotropy, MD mean diffusivity, SNP single- nucleotide polymorphism, pQTL protein quantitative trait loci, GWAS genome-wide association study, eQTL expression quantitative trait loci, MR mendelian randomisation. Created in BioRender. Li, Z. (2025) https://BioRender.com/ghla3ez.

Fig. 2. Association between plasma proteins and brain structures.

A For each modality, each dot represents the most significant association between the specific protein and all corresponding metrics (i.e., regional or tract-wise) in Fig. 2B. Positive associations are shown in red and negative associations are shown in blue. Standardised coefficients are shown. B The distribution of the count of significantly associated proteins across different brain regions and tracts. Positive and negative associations are shown together. C The top 30 proteins with the highest total number of significantly positive (top panel) and negative (bottom panel) associations across modalities. D The top five proteins with the highest number of significant associations for each modality. E The UpSet plot showing the relationship of significantly associated proteins between different modalities. The counts of associations are shown in rows, and the categories of proteins are shown in columns. Source data are provided as a Source Data file. FA fractional anisotropy, MD mean diffusivity.

Fig. 3. Physical and functional annotation of brain structure associated proteins.

A The overview of the annotation categories, including protein-coding gene position, tissue enrichment, cellular enrichment and GO biological process enrichment. B The distribution of coding genes for brain structure association proteins across different chromosomes. The density of the associated proteins is shown, which accounts for the length of the chromosomes. C The distribution of brain structure association proteins across different UKB-defined protein categories. D The enrichment of coding genes for brain structure-associated proteins across GO biological processes. The positively and negatively associated proteins are shown separately, only volume and MD measures showed significant enrichment. E The enrichment of coding genes for brain structure-associated proteins across tissues. The GTEx v8 gene expression data and FUMA GENE2FUN were used. The enrichment in brain tissue is highlighted in the grey bar. The significant enrichments after the FDR correction are shown with a black border. F The enrichment of brain structure-associated proteins across brain cell types. The cell type gene sets are obtained elsewhere, the significant enrichments are shown with coloured circles. For all the enrichment analyses, the statistical significance was determined with hypergeometric tests and the significance threshold was set to FDR-corrected P < 0.05. The UKB 2,920 proteins are used as a background. All statistical tests were two-sided. Source data are provided as a Source Data file. GO go ontology, OPC oligodendrocyte precursor cell, FA fractional anisotropy, MD mean diffusivity, DEG differential expression gene. Created in BioRender. Li, Z. (2025) https://BioRender.com/ghla3ez.

Fig. 4. The association between protein and brain structure in the forward MR.

The forest plot shows the significant MR relationships of the IVW method at a clumping P threshold of $5$×${10}^{-6}$. The significance was determined with a Wald test, and all MR results at a nominal P < 0.001 are shown. Raw P-values are shown in the rightmost column. The MR relationships that meet the significance threshold of FDR-corrected P < 0.05 are marked with an asterisk. The centre of the error bar means the estimated effect of the MR relationship using the IVW method. All statistical tests were two-sided. Source data are provided as a Source Data file. FA fractional anisotropy, MD mean diffusivity, IV independent variant.

Fig. 5. Association between brain structure-associated proteins and gene expression of brain.

The expression of coding genes across tissues for proteins that exhibited significant MR associations with brain structure. Source data are provided as a Source Data file. TPM transcripts per million.

Fig. 6. The association between protein and disease in the forward MR.

The forest plot shows the significant MR relationships of the IVW method at a clumping P threshold of $5$ × ${10}^{-6}$. Raw P-values are shown in the rightmost column. The significance was determined with a Wald test, and the MR relationships meet the significance threshold of FDR-corrected P < 0.05 are shown. The centre of the error bar means the estimated effect of the MR relationship using the IVW method. All statistical tests were two-sided. Source data are provided as a Source Data file. ADHD attention deficit hyperactivity disorder, ASD autism spectrum disorder, BP bipolar disorder, MDD major depressive disorder, SCZ schizophrenia, AD Alzheimer’s disease, PD Parkinson’s disease, MS multiple sclerosis, ALS amyotrophic lateral sclerosis, IV independent variant, OR odd ratio, CI confidence interval.

Fig. 7. The mediation effect of brain structure on association between protein and mental health.

The SEM model was established for assessing the mediation effect, and the significance of the paths were examined with bootstrapping and t test. The observable variables are shown with rectangles, and the latent variables are shown with ellipses in the top row. Standardised coefficients of path a and path b, as well as the corresponding significance values, are shown. The proportion mediated by brain structure is shown on the rightmost column. The potential high collinearity of observable variables in constructing the latent variables is addressed with an iterative selection strategy. As an exploratory analysis, all proteins that exhibited MR associations with brain structure at nominal P < 0.001 were included. The proteins exhibiting significant MR association with brain structure at FDR-corrected P < 0.05 were highlighted in bold. All statistical tests were two-sided. Source data are provided as a Source Data file. Prop proportion.