Integrating genome-wide association studies with quantitative trait loci prioritised PSMA4 as a risk gene for pneumonia

Extract

Pneumonia ranks as the eighth leading cause of death worldwide and leads to significant healthcare costs [1]. A deeper understanding of pneumonia susceptibility and pathogenesis is essential for developing effective preventive and therapeutic strategies. Previous genome-wide association studies (GWAS) from large biobanks have identified genetic loci to be related to pneumonia susceptibility, including loci on chromosome 15 (near genes CHRNA3/5), the major histocompatibility complex region on chromosome 6, and those near genes IL127, PBX3 and APOBR [2], and MUC5AC, TNFRSF1A and CYP1A1 [3].

Shareable abstract

By integrating the largest pneumonia GWAS meta-analysis with multiple QTL data, PSMA4 emerged as a likely risk gene for pneumonia as well as a potential therapeutic target https://bit.ly/435kxK0

To the Editor:

Pneumonia ranks as the eighth leading cause of death worldwide and leads to significant healthcare costs [1]. A deeper understanding of pneumonia susceptibility and pathogenesis is essential for developing effective preventive and therapeutic strategies. Previous genome-wide association studies (GWAS) from large biobanks have identified genetic loci to be related to pneumonia susceptibility, including loci on chromosome 15 (near genes CHRNA3/5), the major histocompatibility complex region on chromosome 6, and those near genes IL127, PBX3 and APOBR [2], and MUC5AC, TNFRSF1A and CYP1A1 [3]. However, most of the GWAS associations are confined to non-coding regions, suggesting gene regulatory effects rather than direct protein sequence alteration.

To bridge this gap, in this study we integrated GWAS with quantitative trait loci (QTL), including expression or splicing QTLs (e/sQTLs), to prioritise genes associated with pneumonia risk. e/sQTLs are genetic variants that influence gene expression level or splicing, linking non-coding genetic variants to their functional effects on gene regulation. We performed a multi-layer integrative analysis combining GWAS, QTL enrichment analysis, Mendelian randomisation and colocalisation, followed by transcriptomic validation.

We used the publicly available pneumonia GWAS meta-analysis between FinnGen release 12 [4], Million Veteran Program [5] and UK Biobank [6], as released by FinnGen consortium. This GWAS meta-analysis has the largest sample size to date, with a total of 151 723 pneumonia cases and 1 360 172 controls. 21 independent loci reached genome-wide significance level (p<5×10⁻⁸) in meta-analysis, corresponding to 2.2–7.5% changes in risk of pneumonia. The most significant association was observed on chr15q25.1 (rs55853698, p=1.05×10⁻²⁹), which was also the only locus that reached genome-wide suggestive significance (1×10⁻⁵) in all three studies. A transcriptome wide association study (TWAS) was performed, using the FUSION models pre-trained on the Genotype Tissue Expression (GTEx) v8 dataset. TWAS integrates GWAS with eQTL data to test whether genetically predicted gene expression levels are associated with a disease or trait. We tested three tissues which are biologically relevant to pneumonia pathophysiology: lung, whole blood, and spleen. Top associations include PSMA4 and ARFRP1 in whole blood, IREB2 and CHRNA3 in lung, and ADAMTS7 and LIME1 in spleen.

QTLEnrich [7] was applied to assess enrichment of pneumonia GWAS among GTEx v8 tissue-specific eQTLs [7]. The analysis revealed significant enrichment of pneumonia genetic associations in eQTLs of most tested tissues, with whole blood showing the highest enrichment (adjusted fold enrichment=1.46, adjusted p<0.001, figure 1a). Compared to whole blood, lung showed a moderate level of enrichment, with an adjusted fold enrichment of 1.31.

FIGURE 1.

a) Enrichment of pneumonia genome-wide association studies (GWAS) in Genotype Tissue Expression dataset v8 tissue-specific expression quantitative trait loci (eQTLs), showing the top 20 tissues with highest adjusted fold enrichment estimated by QTLEnrich. b) Volcano plots showing the result of primary transcriptome-wide Mendelian randomisation (TWMR) analysis. The lower and upper horizontal dashed line represent the raw (0.05) and Bonferroni adjusted (0.05/ngenes) p-value cutoff respectively. The vertical dashed lines represent the beta=0.05 cutoff. c) Forest plots showing the association of PSMA4 with pneumonia across different GWAS (left) and quantitative trait loci (QTL) (right) datasets in additional TWMR analysis. d) Locus zoom plot showing the colocalisation between eQTLGen PSMA4 whole blood eQTL and pneumonia. MVP: Million Veteran Program; sQTL: splicing QTL.

Given the highest enrichment of pneumonia GWAS in whole blood, we next performed a transcriptome-wide Mendelian randomisation (TWMR) to screen for genes of which the expression level in blood contributed to pneumonia risk. The eQTLGen whole blood eQTL data (n=31 684) [8] was used as an exposure and pneumonia GWAS meta-analysis as the outcome. The initial screen identified 33 genes showing significant associations with pneumonia (figure 1b). To assess the cross-GWAS consistency of these findings, we repeated TWMR analysis using the eQTLGen data as exposure but applied to each of the three individual GWAS cohorts. PSMA4 and HYKK were significant in all three GWAS cohorts and TRBV3-1 was significant in FinnGen and Million Veteran Program (figure 1c), while other genes were only significant in one cohort. To further validate the findings across different QTL datasets, we conducted TWMR using the following datasets as exposure: GTEx whole blood e/sQTL, GTEx lung e/sQTL, and peripheral blood mononuclear cells single-cell eQTL data from Aquino et al. [9]. We found PSMA4 was significant in GTEx whole blood e/sQTL, GTEx lung sQTL and monocyte cell-type-specific eQTLs (figure 1c), while other genes cannot be replicated in any of these additional QTL datasets. To summarise, PSMA4 was significant in TWMR analysis across different GWAS and blood/lung QTL datasets, suggesting robust and possibly cell-type-specific regulatory effects.

We further tested the colocalisation between eQTLGen cis-eQTLs of PSMA4 and pneumonia, a method to test whether the same genetic variant underlies both a disease association and a gene regulatory effect. Two Bayesian colocalisation tools were used, and the result showed the posterior probability that PSMA4 shared a single causal variant with pneumonia was 99.7% estimated by the software Coloc and 99.3% estimated by the software HyPrColoc (figure 1d). The putative shared variant was rs55853698, the lead single-nucleotide polymorphism of pneumonia GWAS meta-analysis.

Transcriptomic analysis was further performed to verify the association between PSMA4 and pneumonia. In a whole-blood RNA sequencing dataset of severe community-acquired pneumonia (SCAP) [10], we found PSMA4 expression was higher in SCAP patients (n=56) compared to healthy people (n=21), with a log2 fold change of 0.77 (FDR=6.25×10⁻⁴, estimated with edgeR trimmed mean of m-values (TMM) method). In this dataset, high PSMA4 expression level was also associated with increased mortality after adjusting for age and sex (OR 2.45, 95% CI 1.21–5.75). We further performed a gene set variation analysis, and we found proteasome-related pathways were ubiquitously upregulated in SCAP patients. In a large-scale peripheral blood mononuclear cells single-cell RNA sequencing dataset of COVID-19 by Stephenson et al. [11], we found PSMA4 was highly expressed in dendritic cells, monocytes and plasmablasts, but less in lymphocytes. We further found upregulation of PSMA4 in COVID-19 primarily occurred in CD14 monocytes, followed by CD16 monocytes, dendritic cells (DCs) and plasmablasts.

In this study, we proposed PSMA4 as a potential risk gene for pneumonia by integrating GWAS with QTL. We thought such association was relatively robust, due to consistent evidence including: a) the top GWAS hits were near PSMA4 locus and were consistent across three GWAS studies, b) PSMA4 being significant in TWAS, c) PSMA4 being significant in TWMR analysis across different GWAS and blood/lung QTL datasets, d) increased expression in transcriptomic datasets, e) consistent cell-type-specificity between single-cell QTL and transcriptomic data.

Our findings position PSMA4 as a potential target for therapeutic intervention for pneumonia. PSMA4 is as a subunit of the 20S proteasome which plays an essential role in selective protein degradation, a process critical to the regulation of cell cycle, gene transcription and signal transduction. Upregulated PSMA4 expression has been observed in a series of inflammatory conditions, including sepsis-associated acute respiratory distress syndrome (ARDS) [12], ankylosing spondylitis [13], and primary biliary cholangitis [14], implicating a potential common mechanism involving proteasome-related inflammatory dysregulation. Proteasome inhibitors have already demonstrated the capacity to attenuate excessive inflammation in murine models of sepsis [15], and recent evidence has identified PSMA4 as a promising candidate for drug targeting in sepsis-related ARDS [12]. Future research should focus on validating the functional role of PSMA4 in immune cells and determining the applicability of proteasome as a therapeutic target for pneumonia.

Strengths of this study include the use of a large population comprising 151 723 pneumonia cases and 1 360 172 controls, as well as the application of multiple complementary approaches to reinforce the conclusion. This study has several limitations. Most GWAS and QTL datasets were derived from participants of European ancestry, potentially limiting the generalisability of the findings to non-European populations. Moreover, biobank-based datasets are susceptible to pleiotropic effects, where genetic variants influence multiple traits, complicating causal inference and the interpretation of trait-specific associations.

In conclusion, by integrating the largest pneumonia GWAS meta-analysis with multiple QTL data, we identified PSMA4 as a robust candidate gene for pneumonia risk. Consistent evidence from GWAS, TWAS, TWMR, colocalisation, and transcriptomic analysis supports a potential regulatory role of blood PSMA4 in pneumonia pathogenesis.

Data availability

All GWAS and QTL datasets used in this study are publicly available and can be accessed from the official websites. Any processed data in this study can be accessed from the corresponding author upon reasonable request.