Introduction

Psoriatic arthritis (PsA) is a chronic immune-mediated inflammatory arthropathy that is intricately linked with psoriasis1. It primarily manifests through joint pain, stiffness, and swelling. Following a diagnosis of psoriasis, approximately one-third of patients will ultimately develop PsA2. The global prevalence of PsA ranges from 0.1 to 1% in the general population, with an estimated 520,000 cases in the United States alone3,4. PsA not only increases the risk of disability but also raises the likelihood of mortality among affected individuals, as demonstrated by a standardized mortality rate of 1.622,5. Therefore, there is an urgent need to identify and develop drug targets and biomarkers for PsA to enhance therapeutic interventions.

Currently, numerous plasma proteins have been identified as disease biomarkers, signifying their differential expression (or exclusive expression) in patients with the disease. Beyond their role as disease indicators, these proteins may also serve as potential drug targets and provide new insights into the disease pathogenesis. Previous studies have revealed several circulating proteins associated with the risk of PsA6,7,8,9,10. However, most of these studies were limited in their ability to fully understand the causal role of protein markers in PsA risk due to the narrow selection of proteins, reliance on observational designs, and small sample sizes.

Genome-wide association studies (GWAS) have successfully identified genetic variants associated with plasma protein levels, which are referred to as protein quantitative trait loci (pQTLs). Multiple large-scale proteomic investigations have revealed a vast array of over 18,000 pQTLs, encompassing a broad spectrum of more than 4800 proteins11,12,13,14,15,16,17,18. These pQTLs offer an opportunity to utilize Mendelian randomization (MR) to evaluate the causal impact of potential drug targets on the human disease phenome. MR is a powerful methodological approach for inferring causal relationships between an exposure and an outcome by utilizing genetic variants as instrumental variables19,20,21. This method significantly alleviates concerns associated with reverse causation and confounding biases. Proteome-wide MR has recently provided crucial insights into comprehending the etiology and prioritizing druggable targets for stroke, inflammatory bowel disease, acute pancreatitis, and Alzheimer’s disease22,23,24,25.

Here, we conducted a multi-omics approach to identify protein biomarkers and potential drug targets for psoriatic arthritis. This involved integrating proteome-wide MR using multiple large-scale plasma protein measurements, validation analysis with independent datasets, protein-protein Interaction (PPI) network analysis, druggability assessment, single-cell RNA-sequencing (scRNA-seq) analysis, and phenome-wide association studies (PheWAS). Our findings showed that seven plasma proteins were associated with PsA, including IL23R, ERAP2 and IFNLR1. This study relies on bioinformatics analyses and underlines the importance of further mechanistic validation, particularly through animal model experiments.

Methods

Data sources

The primary outcome data for this study were sourced from the GWAS conducted by Mehreen Soomro et al., which included 5065 PsA cases and 21,286 controls26. Additional GWAS summary data for PsA were obtained from the FinnGen study, encompassing 213,879 participants (1637 cases and 212,242 controls)27. Both sets of GWAS data were derived from two independent, non-overlapping samples of European ancestry. Seven large-scale proteomic studies provided the summary statistics of genetic associations with plasma proteins11,12,13,14,15,16,17. Ethical approval was secured from the relevant authorities. The foundational information regarding these datasets is detailed in Supplementary Data 1–2.

Mendelian randomization analysis

To explore the causal relationships between plasma proteins and PsA, we employed the “TwoSampleMR” package (https://github.com/MRCIEU/TwoSampleMR) to perform MR analysis. The Wald ratio method was employed to generate effect estimates when examining a plasma protein instrumented by a single Single Nucleotide Polymorphism (SNP). For proteins instrumented by two or more SNPs, we predominantly utilized the Inverse Variance Weighted (IVW) method. We mapped SNPs to the human genome Build 37 (NCBI GRCh37) to standardize genomic coordinates. The selection of pQTLs was based on the following criteria: (i) a genome-wide significant threshold of P < 5 × 10–8 to identify highly correlated SNPs with plasma proteins; (ii) SNPs and proteins within the Major Histocompatibility Complex (MHC) region (chr6:25.5–34.0 Mb) were excluded due to their intricate linkage disequilibrium (LD) structure; (iii) LD clumping was performed using the 1000 Genomes Project European reference panel to identify independent pQTLs for each protein (r2 < 0.001); (iv) The R2 and F-statistic (R2 = 2 × EAF × (1-EAF) × beta2 ; F = R2 × (N − 2) / (1 − R2)) were utilized to assess the strength of genetic instruments, where F-statistics less than 10 were defined as weak instruments and were further excluded. R2 represented the proportion of variability in protein levels explained by each genetic instrument (Supplementary Data 3)18,28. An essential aspect of MR is ensuring that SNP effects on exposure align with the effects on the outcome allele. Palindromic SNPs, such as those with A/T or G/C alleles, were excluded to prevent potential strand orientation or allele coding biases. Ambiguous or duplicated SNPs were eliminated during the harmonization process. Both data harmonization and MR analyses were conducted using the TwoSampleMR package. Specifically, data harmonization utilized the ‘harmonise_data()’ function with default settings. MR analysis employed the ‘mr()’ function. False Discovery Rate correction was applied for multiple testing correction, with a significance level set at P < 0.05. Independent GWAS datasets were used for replication study. Replication MR analysis was additionally conducted for the identified proteins using PsA GWAS summary data sourced from FinnGen. A significance level of P value < 0.05 was defined for replication.

Protein-protein interaction network construction

Examining potential interactions among identified proteins, protein–protein interaction (PPI) networks offer enhanced insights into intracellular protein interactions. In this study, we constructed the PPI network utilizing the STRING database (https://string-db.org/). We further searched for PsA-related drug targets in the DrugBank databases, assessed whether the identified proteins can interact with PsA-related drug targets, which prioritized the potential druggable targets29. Additionally, GeneMANIA (https://genemania.org/) was employed for PPI analysis and data visualization30.

Evaluation of druggability for candidate proteins

Evaluating protein-drug interactions is crucial for determining the viability of target genes as potential drug targets. In this study, we will utilize the Drug Signatures Database (DSigDB, http://dsigdb.tanlab.org/DSigDBv1.0/) for this purpose31. DSigDB, boasting 22,527 gene sets and 17,389 distinct compounds across 19,531 genes, serves as a substantial database linking medications and chemicals to their target genes. The identified target genes are submitted to DSigDB, enabling the prediction of drug candidates and the assessment of the medicinal activity associated with the target genes.

Single-cell RNA-sequencing and data analysis

Evaluating the specific cellular groups in which candidate proteins express in PsA aids in assessing the potential of these proteins as drug targets. We further conducted single-cell transcriptome sequencing on peripheral blood samples from three PsA patients32. Mononuclear cells were isolated through density gradient separation from freshly obtained paired peripheral blood samples. Subsequently, the samples were enriched for CD4 and CD8 T cells using fluorescence-activated cell sorting (FACS). The cells were then processed using a 10x Chromium Controller and sequenced with Illumina HiSeq 4000. Quality control, filtering, and clustering analyses were conducted using the R package Seurat. Cells with fewer than 500 genes were excluded to remove low-quality cells. Probable doublets were filtered out by excluding cells with more than 2500 genes or 10,000 UMIs. Additionally, cells with a mitochondrial fraction exceeding 10% were excluded. In total, 29,072 cells successfully passed the filtering process.

Phenome-wide association studies

We further investigated the side effects associated with candidate proteins related to PsA through Phenome-wide association studies (PheWAS) across diverse diseases. Summary statistics were employed to assess the impact of Single Nucleotide Polymorphisms (SNPs) on outcomes, with a sample size of up to 408,961 individuals from the UK Biobank cohort33. Researchers conducted Genome-Wide Association Studies (GWAS) using SAIGE v.0.29, a generalized mixed model method, to address imbalanced distributions of cases and controls. Phenotypic outcomes for diseases or conditions were defined using “PheCodes,” a system that organizes codes from the International Classification of Diseases and Related Health Problems (ICD-9/-10) corresponding to specific phenotypic outcomes34.

Cell isolation and culture

The healthy donor was recruited from the Second Affiliated Hospital of Wenzhou Medical University. The informed consent was obtained. This study was approved by the ethics committee of the Second Affiliated Hospital of Wenzhou Medical University. All ethical regulations relevant to human research participants were followed. Peripheral blood mononuclear cells (PBMCs) were isolated from their blood using lymphocyte separation medium (07851, STEMCELL, China) and density gradient centrifugation. PBMCs (5 × 106 cells) were seeded into each well of a 24-well plate and cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (11875085, Thermo Fisher Scientific, China). Cells were then stimulated with 20 ng/mL IL-23 (CW54, NOVOProtein, China) and 50 ng/mL IL-17A (C774, NOVOProtein, China) for 48 h.

qRT-PCR

Total RNA was isolated using the RNAeasy™ Animal RNA Isolation Kit with Spin Column (R0027, Beyotime, China) and reverse-transcribed with reverse transcriptase (RR092A, TaKaRa, China). RNA concentration was measured on a NANODROP 2000 (Thermo Fisher Scientific, USA). qPCR was performed using SYBR qPCR Master Mix (BL705A, Biosharp, China) on a LightCycler 480 real-time PCR system (Roche, USA) according to the manufacturer’s instructions. Relative mRNA expression levels were calculated with the 2 − ΔΔCT method, using β-actin as the internal control. All experiments were performed in triplicate and repeated three times. Primer sequences are provided in Supplementary Table 1. Results were analyzed using ANOVA in GraphPad Prism 8.0 software, with a p value < 0.05 considered statistically significant.

Statistics and reproducibility

The data analysis in this study was conducted using R software (version 4.1.3) along with open-source packages. Specifically, the “TwoSampleMR” package (https://github.com/MRCIEU/TwoSampleMR) was employed to perform MR analyses. The construction of PPI network was carried out using the STRING database (https://string-db.org/), and additional PPI analysis and data visualization were conducted using GeneMANIA (https://genemania.org/). To correct for multiple testing in MR analyses, the FDR method was applied. Detailed information on the R packages and online resources used is provided in the Methods section. All tools and data sources used are publicly available and ensure the reproducibility of the results.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Results

Identification of candidate proteins associated with PsA

The study design flowchart is detailed in Fig. 1. Utilizing either the Wald ratio or IVW method, 557 proteins were found to have significant associations with PsA risk after False Discovery Rate (FDR < 0.05) adjustment in the Discovery outcome dataset (Supplementary Data 4). During the replication stage, seven proteins (IL23R, ERAP2, IFNLR1, KIR2DL3, NEO1, CLSTN3 and POLR2F) were successfully validated in the FinnGen dataset (P < 0.05) (Fig. 2, Supplementary Data 5). Interestingly, the MR estimates showed that genetically predicted lower NEO1 levels were associated with an increased risk of PsA. The other six proteins (IL23R, ERAP2, IFNLR1, KIR2DL3, CLSTN3, and POLR2F) showed a positive association with PsA risk, suggesting that elevated levels of these six proteins are correlated with an increased risk of PsA.

Fig. 1: Study overview.
figure 1

This figure provides a comprehensive overview of the study design, highlighting the key stages of the research process, including data collection and analysis.

Full size image
Fig. 2: MR results showing the associations between plasma proteins and PsA risk.
figure 2

A The volcano plot illustrates the impact of seven candidate plasma proteins associated with PsA, derived from MR analyses using either IVW or Wald ratio. The forest plot displays the estimated effects of seven candidate proteins from the MR analysis. B displays the results from the discovery dataset. C displays the results from the replication dataset.

Full size image

PPI networks verified causal proteins

To further validate the observed findings, we conducted analysis for 140 PsA-related drug targets in the DrugBank databases (Supplementary Data 6) and constructed PPI networks for seven proteins with these PsA-related drug targets (Supplementary Fig. 1). Notably, five out of the seven proteins (IL23R, ERAP2, IFNLR1, KIR2DL3, and NEO1) were found to interact with established PsA-related drug targets. Drawing from these findings, we categorized these proteins into two tiers. The five proteins that passed the Replication and Discovery MR and demonstrated interactions with reported PsA-related drug targets were classified into tier 1. The remaining two proteins (CLSTN3 and POLR2F) were assigned to tier 2.

Network functional analysis of candidate proteins

The GeneMANIA-constructed protein-protein interaction (PPI) network (https://genemania.org/) not only encompasses the seven candidate proteins but also includes an additional 20 genes with potential interactions (Fig. 3). These connections are primarily based on shared protein domains (89.90%) and genetic interactions (10.10%). The functional analysis of the network highlights the involvement of candidate drug targets and associated genes related to aminopeptidase activity and metalloexopeptidase activity, indicating a robust correlation with immune processes. This finding aligns with the characteristics of autoimmune diseases, consistent with previous research on autoimmune diseases35.

Fig. 3: Functional interaction network analysis of PsA-related proteins.
figure 3

The PPI network was constructed using GeneMANIA. Each node is color-coded to signify the functional pathway associated with the respective gene.

Full size image

Expression quantitative trait loci MR analysis of causal proteins

To explore whether proteins with causal implications exhibit parallel effects at both the transcriptional and proteomic levels in PsA, we conducted Expression Quantitative Trait Loci (eQTL) MR analysis for these seven potential biomarkers. Only POLR2F was not included in this step due to the absence of suitable SNPs as instrumental variables, thereby other six proteins (IL23R, ERAP2, IFNLR1, KIR2DL3, NEO1, and CLSTN3) were analyzed with eQTL datasets (Supplementary Data 7). Ultimately, five candidate proteins (ERAP2, IFNLR1, KIR2DL3, NEO1, and CLSTN3) did not exhibit a discernible causal association with PsA at the transcriptional level. Despite the IL23R protein demonstrating a causal connection with PsA at the transcriptional level, the observed causal effect contradicted that at the proteomic level, suggesting a divergence between the proteomic and transcriptional manifestations of IL23R.

Druggability and clinical development stages of candidate protein targets

We utilized the DrugBank database to systematically evaluate the druggability and current status of drug development for the seven candidate proteins. Among the seven proteins, one protein (IL23R) has three approved drugs: ustekinumab, tildrakizumab, and risankizumab. These three drugs are used to treat moderate-to-severe plaque psoriasis. The remaining six proteins lack information regarding drug-related details (Supplementary Data 8). In addition, the DSigDB database was utilized in this study to predict potentially effective intervention drugs. The potential chemical compounds were highlighted based on adjusted p values (P < 0.05) (Supplementary Data 9). Notably, the findings revealed that scopoletin and esculetin were the two most significant drugs associated with ERAP2. Moreover, theophylline exhibited connections to POLR2F and KIR2DL3.

Cell-specific expression of seven candidate proteins in PsA

The immune cells play a pivotal role in the pathogenesis of PsA, exerting substantial influence on the progression of the disease. Specifically, T cells emerge as central role, serving as integral contributors to the immunological milieu in PsA. Therefore, to investigate the specific expression of seven candidate proteins in different T cell subsets, we conducted single-cell transcriptome sequencing analysis on T cells derived from peripheral blood samples of three PsA patients. T cells were grouped into 12 clusters and (Supplementary Fig. 2). Based on known marker genes32,36, T cells were categorized into 11 cell types (Fig. 4A and Supplementary Fig. 3): regulatory T cells (Treg), CD4+ naive T cells (CD4_naive_T_cell), CD8+ naive T cells (CD8_naive_T_cell), CD4+ effector T cells (CD4_effector_T_cell), mucosal-associated invariant T cells (MAIT), CD8+ exhausted T cells (CD8_exhausted_T_cell), CD8+ tissue-resident memory T cells (CD8_tissue_resident_memory_T_cell), CD8+ effector memory T cells (CD8_effector_memory_T_cell), HLA-DRA + CD8 + T cells (HLADR_CD8_T_cell), MX1 + T cells (MX1_T_cell), and ITGB1 + CD4+ central memory T cells (ITGB1_CD4_central_memory_T_cell). Interestingly, we observed specific expression of IL23R in MAIT cells, while ERAP2, CLSTN3, and POLR2F were expressed in various T cell subsets (Fig. 4B). Conversely, IFNLR1, KIR2DL3, and NEO1 showed no detectable expression in specific T cell subsets (Supplementary Fig. 4). Consistent with our findings, prior studies emphasize the crucial role of MAIT cells producing IL-17A in the pathogenesis of PsA37. These findings imply a potential significant role for IL23R in MAIT cells in the context of PsA.

Fig. 4: Single-cell transcriptomics analysis reveals differential expression patterns of candidate proteins in T cell subtypes of PsA patients.
figure 4

A 2D visualization of single-cell types by UMAP, with a total of 11 cell types. Different colors represent distinct cell types. B The UMAP plot illustrates the expression patterns of four candidate proteins across distinct T cell subtypes.

Full size image

Analysis of phenome-wide association studies of proteins associated with PsA

We then conducted a phenome-wide association study (PheWAS) on 1403 diseases in the UK Biobank to determine whether the five PsA-associated proteins (tier 1) have beneficial or adverse effects on other diseases. The results revealed that increased protein expression of IL23R was linked to an increased risk of PsA, while reduced risks were observed for other autoimmune diseases, such as inflammatory bowel disease and ulcerative colitis (Fig. 5). Increased protein expression of ERAP2 was associated with an increased risk of PsA and was also correlated with an elevated risk of abnormal heart sounds (Fig. 5). Moreover, increased protein expression of IFNLR1 and NEO1 was associated with an elevated risk of hypercholesterolemia (Fig. 5 and Supplementary Fig. 5). Conversely, elevated protein expression of KIR2DL3 was associated with a decreased risk of hypothyroidism (Supplementary Fig. 6). These findings suggest that tier 1 candidate proteins, which are the most promising candidate drug targets for PsA, need to be carefully considered for their potential side effects as drug targets.

Fig. 5: PheWAS analysis of associations between candidate proteins and various disease outcomes.
figure 5

Each point in the plot represents a disease. Points above the dashed line indicate statistically significant disease associations. Different colors represent distinct disease system categories.

Full size image

IFNLR1 expression significantly upregulated in simulated inflammation

Previous studies indicate that IL-17 and IL-23 play significant roles in mediating PsA38,39. To further investigate this pathway, PBMCs from a healthy donor were stimulated with IL-17A and IL-23 in vitro for 48 h to simulate an inflammatory environment. As shown in Fig. 6, qPCR analysis revealed a significant upregulation of IFNLR1 mRNA expression relative to the normal control group. These results further support that IFNLR1 may play a key regulatory role in PsA pathogenesis.

Fig. 6: mRNA expression changes of candidate genes following IL-17A and IL-23 stimulation.
figure 6

Analysis of ERAP2, IFNLR1, and KIR2DL gene expression in PBMCs from a healthy donor showed significant upregulation of IFNLR1 expression following 48 h of co-stimulation with IL-17A and IL-23, relative to the control group. ERAP2 expression also demonstrated an upward trend. Experiments were repeated three times. Error bars represent standard deviation (SD). Statistical significance was designated as follows: *p < 0.05; **p < 0.01; ***p < 0.001.

Full size image

Discussion

In this study, we conducted a comprehensive investigation into the causal associations between 4853 plasma proteins and the risk of PsA. The discovery and replication proteome-wide MR identified seven protein markers. Genetically determined higher levels of six proteins (IL23R, ERAP2, IFNLR1, KIR2DL3, CLSTN3, and POLR2F), and lower levels of one protein (NEO1), were associated with an increased risk of PsA. Additionally, we constructed PPI networks for the seven proteins associated with PsA-related drug targets to validate their potential as candidate drug targets for PsA. In summary, we categorized the seven candidate proteins with causal associations with PsA into two tiers: five proteins (IL23R, ERAP2, IFNLR1, KIR2DL3, NEO1) with the most compelling evidence (Tier 1), and the second tier includes two proteins (CLSTN3 and POLR2F) with strong evidence. Additionally, we conducted a thorough assessment of the cell-specificity of the seven candidate proteins in different T-cell subsets present in PsA. Subsequently, we utilized the DrugBank and DSigDB databases to evaluate their potential as therapeutic targets for treating PsA. Notably, IL23R has already been developed for treating moderate to severe plaque psoriasis. Given this, other six proteins are likely to be repurposed as therapeutic targets for PsA, especially the Tier 1 candidate proteins. We then conducted a PheWAS analysis to assess the side effects and safety profile of the five tier 1 candidate proteins as potential drug targets. Moreover, experimental evidence showed that IFNLR1 expression is significantly upregulated under simulated inflammatory conditions.

The interleukin 23 receptor (IL23R), encoded by this gene, serves as a subunit of the receptor complex for IL23A/IL2340. This protein forms a pairing with the receptor molecule IL12RB1/IL12Rbeta1, and both are essential for IL23A signaling. Studies have reported an elevated expression of the IL23R gene associated with an increased risk of PsA, consistent with the findings of this study41,42,43. Furthermore, existing studies have reported that the expression of IL23R is higher in lesional skin compared to non-lesional skin and synovium in PsA41. IL23R has a close relationship with IL23, and drugs targeting IL23 (ustekinumab, tildrakizumab, risankizumab) are currently used for treating moderate to severe plaque psoriasis. The potential for future development of IL23R as a therapeutic target for PsA is significant.

More importantly, this study identified six candidate proteins associated with PsA, including ERAP2, IFNLR1, KIR2DL3, NEO1, POLR2F, and CLSTN3.

The endoplasmic reticulum aminopeptidase 2 (ERAP2) gene encodes a zinc metalloaminopeptidase belonging to the M1 protease family. It is localized in the endoplasmic reticulum and plays a role in N-terminal trimming of antigenic epitopes for presentation by major histocompatibility complex (MHC) class I molecules. Specific mutations in this gene have been linked to the inflammatory arthritis syndrome ankylosing spondylitis and pre-eclampsia44,45. This gene is located near a closely related aminopeptidase gene on chromosome 5. Currently, there is limited research on the association between ERAP2 and PsA. Although a study has reported that ERAP2 influences the risk of PsA in the Romanian population, the mechanisms underlying how ERAP2 affects the risk of PsA are poorly understood. Future research is required to unveil the relationship between ERAP2 and the risk of PsA. Further epidemiological studies and experimental research are necessary to determine ERAP2 as a potential therapeutic target for treating PsA.

IFNLR1 was prioritized based on robust supporting evidence (tier 1). Interferon lambda receptor 1 (IFNLR1), belonging to the class II cytokine receptor family, forms a receptor complex with interleukin 10 receptor beta (IL10RB). The receptor complex interacts with three closely related cytokines-interleukin 28 A (IL28A), interleukin 28B (IL28B), and interleukin 29 (IL29). Studies have demonstrated elevated IFNL protein levels in the sera of patients with rheumatoid arthritis compared to those of healthy controls46,47. As the receptor for IFNL1, IFNLR1 binds to IFNL1 and mediates corresponding biological effects. This interaction is a pivotal component of the interferon signaling pathway, crucial for the normal functioning of the immune system. Although no direct evidence is reported regarding the association between IFNLR1 protein and the risk of PsA, the upregulation of IFNLR1 gene expression may be associated with the pathogenesis of Systemic Lupus Erythematosus (SLE)48. Therefore, IFNLR1 holds the potential to serve as a highly promising therapeutic target for treating PsA. Our findings reveal that IFNLR1 expression is significantly upregulated under simulated inflammatory conditions, reinforcing its potential as a key regulator in PsA pathogenesis. By stimulating PBMCs with IL-17A and IL-23, two cytokines well-recognized for their pivotal roles in PsA, we successfully mimicked an inflammatory environment and observed a marked increase in IFNLR1 mRNA expression. This aligns with prior studies highlighting the IL-17/IL-23 axis as a central driver of PsA inflammation. The upregulation of IFNLR1 under these conditions suggests that it may serve as a downstream effector or modulator of this pathway. These data, when combined with GWAS analysis and single-cell findings, strongly implicate IFNLR1 as a potential therapeutic target for PsA.

Treatment for PsA includes traditional disease-modifying antirheumatic drugs (DMARDs), biologic therapies such as TNF inhibitors (TNFi), IL-17 inhibitors (IL-17i), IL-12/23 inhibitors (IL-12/23i), and new targeted oral agents, including a phosphodiesterase-4 inhibitor and a Janus kinase (JAK)/signal transducer and activator of transcription (STAT) inhibitor49,50. In recent years, although new therapies targeting specific cytokines and signaling pathways have emerged and the number of therapeutic options has increased, overall improvements in treatment outcomes have been disappointing. At least 40% of PsA patients exhibit no response or only a partial response, and clinical trials have failed to consistently achieve a 20% improvement according to American College of Rheumatology criteria in more than 60% of patients1,51. Personalized treatment for this heterogeneous patient population is crucial, underscoring the urgent need to identify and develop new drug targets and biomarkers for PsA to enhance therapeutic interventions. This study employed proteome-wide Mendelian randomization to identify seven plasma proteins associated with PsA (e.g., IL23R, ERAP2 and IFNLR1), offering potential therapeutic targets for its treatment.

A notable strength of this study lies in its systematic examination of the associations between plasma protein biomarkers and PsA risk, employing a two-stage proteome-wide MR design. This approach benefits from large sample sizes, extensive proteome coverage, and minimal risk of confounding bias. The consistent findings across multiple rigorous analyses confirm the robustness of the study. Supplementary evidence from scRNA-seq analysis, PPI networks, and druggability evaluation sheds light on the potential pathogenic effects of candidate proteins on PsA and further prioritizes druggable targets.

There are several limitations in this study. First, some candidate proteins lack drug information. Second, the data for the seven plasma proteins were derived from different experimental platforms, potentially introducing batch effects. Third, further experimental research is needed to explore these seven proteins as potential PsA drug targets, thus investigating their mechanisms and functions.

This study employed proteome-wide mendelian randomization to identify seven plasma proteins associated with PsA, offering novel insights into the disease’s etiology and presenting promising targets for the selection of biomarkers and therapeutic drugs. Our results primarily rely on predictive approaches, emphasizing the importance of conducting additional mechanistic investigations. Validation through animal experiments will be crucial to support and strengthen the reliability of these findings.