Breast cancer (BC) is the most common malignant tumor of women in clinical practice and seriously endangers women’s physical and mental health. In the year 2020, there were a staggering 2.26 million new cases of BC worldwide, with its incidence having surpassed that of lung cancer to become the global leader (1). With a deepening understanding of BC treatment, a new era of comprehensive treatment has emerged (2). Pharmacological therapy is an indispensable component of the comprehensive treatment approach for BC patients, but it brings about a series of challenges (3). Firstly, the existing medications come with significant side effects (i.e., gastrointestinal reactions, bone marrow suppression, and myocardial structural damage). Secondly, many patients show no response to pharmacological treatment (i.e., treatment-resistant). As such, novel and effective drug therapy targets for BC are still needed.

Proteins play a pivotal role in the pathogenesis and progression of diseases, with circulating proteins often serving as viable targets for pharmacological interventions due to their amenability to direct manipulation. In preceding research, several circulating proteins linked to BC have been documented, including cellular communication network factor 1 (CCN1) (4), serum secreted clusterin (sCLU) (5), and insulin-like growth factor 1 (IGF1) (6). However, most of these studies are observational and yield results that are susceptible to the possibility of confounding or reverse causation bias.

Randomized controlled trials (RCTs) serve as the gold standard for establishing the relationship between drugs (protein targets) and BC. However, the implementation of RCTs is challenged by significant financial and time costs. Mendelian Randomization (MR) is an epidemiological approach that employs genetic variation as a proxy for exposure to predict causal relationships with outcomes (7, 8). Owing to the random allocation of genetic variations during conception, MR is largely impervious to the influence of confounders and reverse causality. The efficacious application of drug-targeted MR analyses has extended to various disorders, such as multiple sclerosis (9), type 1 diabetes (10), and COVID-19 (11). However, there have been limited studies utilizing the drug target MR approach to identify potential targets for BC up to the present.

In this study, we conducted MR by integrating large-scale BC genome-wide association studies (GWAS) results with cis-protein quantitative trait loci (cis-pQTL) of 2,004 plasma proteins, aiming at identifying plasma proteins that can serve as potential therapeutic targets for BC. To ensure the stability of causal relationships, we conducted a series of sensitivity analyses, including reverse causality detection, pleiotropy test, and Bayesian co-localization analysis. For significant MR results, we utilized cis-pQTL data from the deCODE cohort for replication to validate the preliminary results. In subsequent analyses, we performed protein-protein interaction (PPI) and Gene Ontology (GO) enrichment analyses on suggestive causal proteins (p < 0.05), and visualized a PPI network diagram between prioritized proteins and established BC drug targets. Then, we conducted a phenome-wide MR (Phe-MR) against 525 disease phenotypes to explore side effects. Finally, we further affirmed the drugability of potential drug targets through mRNA expression analysis and molecular docking. The study design is shown in Figure 1 .

All data used in this study were derived from publicly available GWAS statistics and did not require new ethical approval.

Despite the steady progress in breast cancer treatment approaches, breast cancer therapy still faces numerous challenges. Therefore, the development of new drugs for breast cancer is urgently needed. In this study, based on a large-scale pQTL dataset, we identified 15 prioritized proteins that may influence BC outcomes [11 for BC_overall, 3 for BC_ER(+), and 2 for BC_ER(-)], 12 of which were also found in a repeated study. To mitigate reverse causality, horizontal pleiotropy, or genetic confounding due to linkage disequilibrium (LD), we conducted a series of sensitivity analyses to strengthen our conclusions. Reverse causality detection was performed using Steiger filtering and Bidirectional MR analysis. Encouragingly, we did not identify any proteins with reverse causal effects on BC. Next, phenoscanner was used to detect horizontal pleiotropy. The genetic instruments of MST1 and GPX1 appear to be associated with a variety of proteins, suggesting that MST1 and GPX1 might function as hub proteins regulating multiple pathways. Therefore, we temporarily excluded MST1 and GPX1 from the list of candidate drug targets due to their complex biological functions. Phenoscanner also revealed associations between SNP and several allergic diseases and digestive system disorders. Finally, Bayesian colocalization was employed to assess whether the MR findings were impacted by linkage disequilibrium. Out of the 16 potential targets, 7 (43.75%) potential targets passed the colocalization test with 0.7 as the critical threshold for posterior probability. All in all, through pleiotropy scanning, reverse causality detection, colocalization analysis, and replication analysis, we further identified five potential drug targets: TLR1, A4GALT, SNUPN, and CTSF for BC_overall and TLR1 for BC_ER(+). Utilizing a multifaceted approach, including GO enrichment analysis, PPI analysis, molecular docking simulations, and mRNA expression analysis, we have extended the confirmation of the pharmacological viability of the predicted target proteins. Last but not least, the Phe-MR indicated a few potential safety concerns.

TLR1 is a member of the Toll-like receptors (TLRs) family, expressed on the surface of immune cells. TLRs are integral components of the innate immune system, crucial for protecting the host against bacterial and viral infections. Emerging evidence indicates that the TLRs/NF-κB signaling cascade assumes a substantial role in the etiology and recidivism of BC (28). Upon engagement with either endogenous or exogenous ligands, TLRs stimulate intracellular signaling pathways, culminating in the release of diverse cytokines, including tumor necrosis factor-alpha and interleukin-1, which orchestrate chronic inflammatory responses. This activation cascade in turn triggers NF-κB signaling, potentiating tumor cell proliferation. Moreover, González-Reyes et al., through immunohistochemistry, protein blotting, and real-time PCR analysis, identified a significant increase in the mRNA levels of TLR3, TLR4, and TLR9 in recurrent breast cancer samples (29). These findings align with our discoveries, underscoring the potential promise of TLR1 antagonists in BC treatment. However, the potential latent off-target effects, such as inadvertent immunosuppression, cast uncertainty on the feasibility of small molecule TLR1 antagonists. As evident in our phe-MR study, adverse effects like arthrosis and chalazion could be mediated by immune suppression. Furthermore, our PPI analysis has unveiled an intricate interplay between TLR1 and CXCR4, as well as AKT1. Notably, CXCR4 has been identified as the therapeutic target of Balixafortide, whereas AKT1 as the therapeutic target of Capivasertib and Ipatasertib, all of which have been subjected to rigorous scrutiny in phase III clinical trials for breast cancer. This compelling convergence of evidence amplifies the hypothesis that TLR1 stands as a promising novel therapeutic avenue for BC intervention.

Snurportin 1, also known as SNUPN, is a protein involved in the transport of small nuclear ribonucleoproteins between the cytoplasm and the nucleus of a cell. Historically, research on SNUPN has been limited, although there is a more extensive study of a protein that collaborates with SNUPN, known as exportin 1 (XPO1). SNUPN and XPO1 mutually recognize each other, forming a nuclear pore complex or a cargo together, leading to abnormal cellular localization of oncogenes, tumor suppressor genes, and signaling pathway mediators. This disruption of cellular homeostasis contributes to the initiation and progression of tumorigenesis, particularly in the context of leukemia (30). A recent proteome association study has further confirmed the association between SNUPN and breast cancer, which was in consistent with our finding (31).

Cathepsin F is a member of the Cathepsin family, playing a pivotal role in protein degradation and metabolic processes within cells. Research indicates that Cathepsin F may be associated with the development and progression of certain tumors, such as lung cancer (32), thyroid cancer (33), and cervical cancer (34), although there is a lack of studies investigating its relationship with breast cancer. Other members of the cathepsin protease family have been reported to have connections with breast cancer. For example, Vashum Y et al. indicated that CTSK is specifically associated with breast cancer bone metastasis by promoting adipocyte differentiation (35). Additionally, Cathepsin D has been found to be upregulated and secreted by breast cancer cells, promoting tumor invasion and metastasis by degrading the extracellular matrix and basement membrane in an acidic environment (36). Based on the aforementioned studies, there is reason to believe that the Cathepsin family plays a significant role in the occurrence and progression of breast cancer. Therefore, CTSF as a potential therapeutic target for breast cancer is credible.

Lactosylceramide 4-alpha-galactosyltransferase, encoded by the A4GALT gene in humans, participates in the pathway of sphingolipid metabolism, specifically responsible for attaching a galactose moiety to lactosylceramide molecules, thereby forming an α-1,4-galactose linkage (37). Despite the limited extent of previous investigation into 4AGALT, we postulate that the sphingolipid metabolism mediated by 4AGALT could intricately modulate the composition and structure of cell membranes, consequently exerting influence over the tumor microenvironment. Both our MR and co-localization analyses robustly affirm the association between A4GALT and breast cancer. However, further research is requisite to elucidate the underlying mechanistic basis of this relationship.

This study possesses numerous merits. Primarily, it is widely recognized that the drug development process inherently demands substantial time, exorbitant costs, and encounters a notable risk of failure. Employing drug targeting MR, through an analysis of vast-scale biological data encompassing the genome and proteome, pertinent plasma proteins associated with BC were identified, thereby pinpointing potential drug targets and amplifying the efficiency of drug development. Additionally, the MR method obviates confounding and reverse causality inherent in observational studies. Thirdly, complementary analyses such as co-localization analysis, phe-MR, drug targets PPI analysis, and replication analysis contribute to the comprehensiveness and reliability of our research findings.

Still, this study is not without certain limitations. Firstly, our analysis was exclusively focused on individuals of European descent, posing challenges in extrapolating the findings to other ancestral groups. Furthermore, while interventions targeting circulating plasma proteins may exert systemic impacts, the attainment of precise modulation within specific tissues remains uncertain. Of paramount importance, it is noteworthy that MR analysis does not entirely recapitulate clinical trials, as patient responses to pharmaceutical interventions inherently manifest diversity within clinical practice. Consequently, clinical trials are warranted to meticulously evaluate the preliminary efficacy and safety profile of these latent drug targets for BC intervention.

In conclusion, this study embraced a thorough genetic methodology to evaluate the intricate interrelationship of plasma proteins with BC and its subcategories. It is noteworthy to emphasize that our study findings underscore the feasibility of TLR1, A4GALT, SNUPN, and CTSF as viable therapeutic targets for BC_overall or BC_ER(+) subtypes. Subsequent investigations hold the promise of corroborating our observations and delving into the underlying mechanisms that warrant exploration.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

TY: Writing – original draft, Conceptualization. YL: Writing – original draft, Conceptualization. YW: Writing – original draft, Conceptualization. XQ: Writing – review & editing, Data curation. ZW: Writing – review & editing, Data curation. SQ: Writing – review & editing, Data curation. TJ: Software, Writing – review & editing. JL: Writing – review & editing, Software. LF: Writing – review & editing, Software. CZ: Writing – review & editing, Software. CW: Writing – original draft, Conceptualization.