Therapeutic Targets for Sepsis: Multicenter Proteome-Wide Analyses and Experimental Validation

Abstract

Sepsis is a life-threatening condition with high mortality, underscoring the urgent need for effective therapeutic targets. We conducted a proteome-wide Mendelian randomization (MR) analysis using plasma protein data from the FinnGen, UKB-PPP, and deCODE cohorts to identify proteins causally associated with sepsis. The analysis included 16,074 cases and 363,227 controls in FinnGen and 11,643 cases and 474,841 controls in the UK Biobank, spanning four exposure-outcome combinations. Proteins were prioritized based on a false discovery rate <0.05 in one combination and p < 0.05 in another. Colocalization and phenome-wide association studies (PheWAS) were performed to evaluate causality and potential off-target effects. Three proteins─dual specificity phosphatase 13 (DUSP13), inhibin beta C chain (INHBC), and toll-like receptor 1 (TLR1)─showed strong evidence of colocalization with sepsis risk. PheWAS confirmed broader disease associations for DUSP13 and TLR1, while INHBC showed no significant adverse associations and is considered druggable. TLR1 is currently under clinical investigation. ELISA-based experimental validation in 20 sepsis patients and 20 controls demonstrated elevated serum levels of DUSP13 and INHBC and reduced levels of TLR1 in sepsis. These findings identify DUSP13, INHBC, and TLR1 as promising therapeutic targets for sepsis, supported by genetic, phenotypic, and experimental evidence.

Keywords: PheWAS; colocalization analysis; druggability; proteome-wide Mendelian randomization; sepsis; therapeutic targets.

1

Study design. FinnGen: FinnGen study, UKB-PPP: UK Biobank Plasma Proteomics Project, deCODE: deCODE Health Study, UKBB: UK Biobank, MHC: Major histocompatibility complex, SNP: Single nucleotide polymorphism, FDR: False discovery rate, MR: Mendelian randomization, PPH4: Posterior probability of colocalization, MAF: Minor allele frequency, ELISA: Enzyme-linked immunosorbent assay.

2

Multicenter proteomic Mendelian randomization analysis of plasma proteins and sepsis. Manhattan plots show associations between (a) plasma proteins from FinnGen and sepsis from the UK Biobank, (b) plasma proteins from UKB-PPP and sepsis from FinnGen, (c) plasma proteins from deCODE and sepsis from the UK Biobank, and (d) plasma proteins from deCODE and sepsis from FinnGen. Each point represents an SNP linked to a plasma protein, with the x-axis showing genomic positions and the y-axis displaying -log10 (p value). The dotted lines indicate p = 0.05, and significant associations post-FDR correction are indicated. FinnGen: FinnGen study, UKB-PPP: UK Biobank Plasma Proteomics Project, deCODE: deCODE Health Study, SNP: Single nucleotide polymorphism.

3

UpSet plot showing the selection process for the proteins included in the study. The left panel displays the set sizes for each data set, whereas the right bars represent the intersection sizes, indicating the number of shared proteins across the various data set combinations. The green bars highlight proteins that passed the FDR correction in one group and had a P-value <0.05 in at least one other group; these proteins were selected for further colocalization and functional analyses. The orange bars represent the intersection of proteins with P-values <0.05 across data setsthese proteins did not pass the FDR correction and were excluded from subsequent analyses. In total, 42 unique proteins were selected for further analyses.

4

Mendelian randomization analyses and Bayesian colocalization results for three candidate causal proteins associated with sepsis across four analysis cohorts. The y-axis lists the protein names and data sources, and the x-axis presents the OR with 95% CI and PP.H4 for each protein. Error bars represent the OR for sepsis per 1 SD increase in protein expression, as estimated using the inverse variance weighted method. FinnGen: FinnGen study, UKB-PPP: UK Biobank Plasma Proteomics Project, deCODE: deCODE Health Study, UKBB: UK Biobank, nsnp: number of SNPs, OR: odds ratio, CI: confidence interval, SD: standard deviation, FDR: false discovery rate, PP.H4: posterior probability of Hypothesis 4 using the approximate Bayes factor.

5

Serum concentrations of DUSP13, INHBC, and TLR1 in patients with sepsis and nonsepsis controls. ELISAs were conducted to measure the serum expression levels of DUSP13, INHBC, and TLR1 in both patients with sepsis (n = 20) and nonsepsis (n = 20) groups. Boxplots show the distribution of protein concentrations (pg/mL) in each group. Statistical significance was assessed using the Mann–Whitney U test, with p < 0.05 for DUSP13 and p < 0.001 for INHBC and TLR1.