Insights into Kidney Protein Handling through a Genome-Wide Association Study (GWAS) of the Urine Proteome

Background: Urine contains small amounts of protein under physiological conditions. Levels of urine proteins are determined by filtration from plasma, exosomes and membrane patches of kidney and urothelial cells, and active tubular secretion and reabsorption. Genetic screens of the urine proteome and comparisons to similar plasma studies may offer insights into renal protein handling, particularly in kidney tubules. Methods: We performed genome-wide association studies with 2886 proteins quantified by the antibody-based Olink Explore 3072 platform from urine samples of 1246 German Chronic Kidney Disease study participants with eGFR>45 and ≤60 mL/min/1.73m2 and UACR<300 mg/g. In a discovery-replication design, we studied relations of protein levels and 6 million common genetic variants, adjusted for urine dilution and potential confounders. Significantly associated (p<1.7x10-11) variants were assigned to protein quantitative trait loci (pQTLs). To examine significant urine pQTLs in detail, we performed finemapping and identified independent credible sets (CS) of variants most likely to cause association signals in each locus, comparisons to pQTLs from corresponding plasma proteome screens (UK Biobank), and genetic colocalization analyses. Results: We identified 173 pQTLs, arising from 127 distinct genetic regions and 170 unique proteins. The 173 pQTLs contained 153 cis associations, with the index variant in close proximity to the protein’s cognate gene, supporting the precision and validity of urine proteomics. Median protein variance explained by pQTLs was 10.3%, with a maximum of 67.2% for variant chr8:142681841:T:TG and prostate stem cell antigen levels (PSCA locus, GWAS p-value 1.2x10-451). Finemapping revealed 254 independent CS (1-8 CS per region). Colocalization with plasma pQTLs detected shared signals, reflecting genetic fingerprints on the systemic proteome recovered from urine. Urine-specific associations were found for at least 15 proteins including mucin-13, where a regulatory variant altering transcription factor binding in an enhancer region determined urine levels of this epithelial barrier protein. Conclusion: Genetic screens of the urine proteome can reveal mechanisms involved in human renal protein handling. Our ongoing investigations include connections of pQTLs to the transcriptome, the urine and plasma metabolome, and clinical traits and diseases. Funding: Government Support – Non-U.S.