Integrating Human Brain Proteomic Data with Genome-Wide Association Study Findings Identifies Novel Brain Proteins in Substance Use Traits

Background: Despite the growing number of genetic risk loci identified for substance use traits (SUTs), the impact of these loci on protein abundance and their potential as therapeutic targets are unknown. Methods: To address this, we performed a proteome-wide association study (PWAS) by integrating human brain proteomes from discovery (Banner; N = 152) and validation (ROSMAP; N = 376) datasets with genome-wide association study (GWAS) summary statistics for 4 SUTs. The sample comprised 4 GWAS of European-ancestry individuals for smoking initiation [Smk] (N = 1,232,091), alcohol use disorder [AUD] (N = 313,959), cannabis use disorder [CUD] (N = 384,032), and opioid use disorder [OUD] (N = 302,585). We conducted transcriptome-wide association studies (TWAS) with human brain transcriptomic data to examine the overlap of genetic effects at the proteomic and transcriptomic levels and tested significant genes for causality through Colocalization analysis. Results: Twenty-seven genes (Smk=21, AUD=3, CUD=2, OUD=1) were significantly associated with cis-regulated brain protein abundance. There was evidence for causality in 6 genes (Smk: NT5C2, GMPPB, NQO1, SRR, and ACTR1B; AUD: CTNND1), which act by regulating brain protein abundance. Cis-regulated transcript levels for 8 genes (Smk=6, CUD=1, OUD=1) were associated with SUTs, indicating that genetic loci could confer risk for these SUTs by modulating both gene expression and proteomic abundance. Conclusions: Functional studies of the high-confidence risk proteins (SRR for Smk and CTNND1 for AUD) identified here are needed to determine whether they are modifiable targets and useful in developing medications and biomarkers for these SUTs.