The detection of plasma proteins may indicate broader systemic processes associated with AD pathophysiology, including neuroinflammation, oxidative stress, and metabolic dysregulation. As such, plasma proteins can serve as valuable biomarkers of disease states, offering accessible targets for therapeutic intervention. Although the primary pathological manifestations of AD occur within the brain, systemic alterations—such as those measurable in plasma—can influence disease progression. Consequently, plasma proteins present a potentially modifiable therapeutic pathway, particularly for strategies targeting the peripheral regulation of neuro degenerative processes.

In this study, we employed MR to identify plasma circulating proteins associated with AD, with five (GSTP1, BIN1, Siglec-3, SERPINF2, GRN) showing statistically significant associations with AD and strong evidence of CA, suggesting their potential causal roles in AD pathogenesis. Protein–protein interaction (PPI) analysis revealed interactions between identified targets and current AD drug targets, highlighting GSTP1 and BIN1 as promising new targets for drug development. We preliminarily conducted virtual screening and identified several potential small-molecule inhibitors.

Previous research results have shown that smoking is associated with AD neuropathology in preclinical models and humans. Smoking related oxidative stress in the brain is a potential mechanism that promotes AD pathology and increases AD risk [29]. And our results found that quitting smoking may prevent AD by affecting BIN1 levels, which is consistent with previous findings. Previous studies have yielded contradictory results regarding the impact of dried fruit intake on AD. Deng's study [30] found that an increase in dried fruit intake was associated with a decrease in AD risk, while Liao et al.'s study found a positive correlation between an increase in dried fruit intake and AD risk [29]. And our reduction in dried fruit intake may lower the risk of AD by affecting SERPINF2. We believe that the reason for this difference may be due to differences in sample size, or the types of dried fruits studied.

A “causality” identified by MR might be horizontal pleiotropy, genetic confounding due to linkage disequilibrium (LD), or reverse causality. To limit the bias from horizontal pleiotropy, we only used cis-pQTLs as the instruments, given their direct role in the transcription and/or translation of related genes [31]. In addition, Bayesian CA was also used to exclude the bias introduced by LD. With PPH4 exceeding 0.75 in different windows or when PPH3 + PPH4 exceeding 0.8 as the critical threshold for posterior probability, the five proteins identified were likely to share the same variant of AD [32]. Bidirectional MR was conducted in the study and no proteins showed reverse causality.

Growing evidence suggests that variations and expression levels of multiple genes play a critical role in neuroinflammation and neuro degenerative processes within the pathophysiology of AD. GSTP1 is the most widely studied member of the GST family [33]. It is supposed to involve in metabolism, detoxification and elimination of potentially genotoxic foreign complexes, metabolizes a variety of carcinogenic compounds and protects cells against DNA damage and canceration [34]. However, it may play the villain in AD. In 2016, GSTP1 Ile105Val polymorphism was found to be associated with increased risk of AD [35]. Then, in 2017, a study reported that GSTP1 was involved in anxiety and depression behaviors in 10-month-old triple transgenic mice of AD, and melatonin could serve as a potential candidate drug to improve the neuropsychiatric behaviors in AD via modulating the expression of the GSTP1 [36]. Our study further proved that GSTP1 is positively associated with AD in humans, with one standard deviation (SD) increase raising the risk by 62%. BIN1, its down-regulation was founded to be related to cancer progression and also correlates with ventricular cardiomyopathy and arrhythmia preceding heart failure while increased BIN1 expression maybe linked with increased susceptibility for AD [37]. A study investigated the role of BIN1 in regulating neuroinflammatory responses in microglia by employing various molecular techniques, including immunostaining, RT-PCR, siRNA knockdown, and Cre-lox conditional deletion in both mouse and human models, with findings analyzed through gene expression panels, flow cytometry, and pathway analysis. The result shows that BIN1 is primarily involved in clathrin-mediated endocytosis and has been implicated in the regulation of neuroinflammation through its role in microglial activation [38]. Our study quantitatively underscores the enormous risk of BIN1 protein level, showing that each SD increase raises the likelihood of AD by 126%. Furthermore, Siglec-3 -related research has become a new hot topic. Siglec-3 is an immunomodulatory receptor expressed on microglia and is implicated in the inhibition of microglial-mediated clearance of Aβ. [39] Our study demonstrated the positive association between increased Siglec-3 protein level and AD risk, this could be explained as a stronger inhibitory ability on Aβ clearance mediated by microglia, however, the effect size is relatively small. For every one SD increase of Siglec-3 in plasma concentration, the risk of AD merely increases by 3%. In contrast to these risk factors, GRN is a multifunctional protein involved in the regulation of neuroinflammation and the promotion of neuronal survival. Reduced GRN levels have been linked to increased neuroinflammation and neuronal death, contributing to the pathogenesis of AD [40]. GRN was identified as a protective factor as well in our research, with each standard deviation increase in plasma concentration reducing AD risk by 34%. In addition, a study assessed the methylation state of the brain's DNA in relation to AD using 708 prospectively collected autopsied brains [41]. By comparison, although SERPINF2 has been mentioned in previous literature, its specific causal relationship with AD has not been clearly validated. Therefore, our study provides causal evidence for SERPINF2 protein as a potential regulatory factor for Alzheimer's disease for the first time.

Despite advancements in therapies in recent years, current treatment options for AD remain challenging. The pathophysiology of AD involves multiple complex processes, with glutamatergic excitotoxicity recognized as a significant contributor to neuronal damage and death. The NMDA receptor, a subtype of glutamate receptors, is central to synaptic plasticity and cognitive functions, including learning and memory. In pathological conditions, particularly in AD, excessive glutamate release leads to sustained activation of NMDA receptors. This chronic activation results in an influx of calcium ions into neurons, triggering a cascade of intracellular events that culminate in neuronal injury and apoptosis—a phenomenon known as excitotoxicity [42]. Memantine Hydrochloride exerts its therapeutic effects by acting as a non-competitive antagonist at the NMDA receptor, specifically binding to the NMDA receptor subunit 2B (encoded by the GRIN2B gene). This action selectively blocks the prolonged calcium ion influx associated with excitotoxicity while preserving normal physiological synaptic transmission, which is essential for cognitive function [43]. Protein–protein interaction analysis based on the STRING database identified GLRA1 as another target of Memantine Hydrochloride. Moreover, GLRA1 and BIN1 appear to be connected within the network, suggesting a potential interaction or co-expression relationship. BIN1 is involved in membrane dynamics, particularly in endocytosis and membrane curvature, whereas GLRA1 is a component of the glycine receptor, a ligand-gated ion channel involved in inhibitory neurotransmission [38]. The association between GLRA1 and BIN1 may indicate a functional relationship in synaptic activity or neuronal signaling. BIN1's role in membrane remodeling could influence the localization or trafficking of GLRA1, thereby affecting glycinergic neurotransmission. Additionally, Donepezil, a medication primarily used to treat cognitive symptoms in AD, mainly targets and inhibits acetylcholinesterase, an enzyme responsible for the breakdown of acetylcholine in the synaptic cleft [44]. Acetylcholine plays a critical role in memory formation and cognitive function, leading to improvements in symptoms such as memory, attention, and other cognitive abilities. Interestingly, in our study, Donepezil was also determined to act on the IL1B target, and the correlations between IL1B and Siglec-3, SERPINF2 were then identified. IL1B is a critical pro-inflammatory cytokine that plays a central role in the inflammatory response and is often involved in the pathogenesis of various diseases, including autoimmune disorders [45]. The interaction between IL1B and Siglec-3, a sialic acid-binding immunoglobulin-like lectin primarily expressed on myeloid cells, suggests an underlying regulatory mechanism in which Siglec-3 modulates IL1B-driven immune responses. This interaction potentially influences the activation state of myeloid cells during inflammation, highlighting the importance of Siglec-3 in immune regulation [46]. SERPINF2, also known as alpha-2 antiplasmin, is a serine protease inhibitor that primarily inhibits plasmin, the major enzyme responsible for fibrinolysis [47]. The interaction between IL1B and SERPINF2 implies a role in modulating inflammation-driven proteolytic activity, which could be particularly relevant in pathological conditions where inflammation and fibrinolysis are closely linked, such as in tissue repair, or certain inflammatory diseases.

Furthermore, to the best of our knowledge, this is the first study to discover the positive causal association between GSTP1 and AD and no interactions between selected AD drug targets and GSTP1 were observed. GSTP1 plays a critical role in cellular detoxification by catalyzing the conjugation of glutathione to various electrophilic compounds. Structurally, GSTP1 is a homodimer, with each subunit comprising approximately 210 amino acids. The enzyme's structure features a highly conserved N-terminal domain that binds glutathione and a more variable C-terminal domain responsible for interacting with substrates [48]. The active site of GSTP1 is formed by a combination of residues from both domains, creating a catalytic pocket essential for its enzymatic activity. In fact, this detoxifying enzyme is over-expressed in erythrocytes when unusual amounts of toxins are present in the body [49]. Therefore, the reverse causality between GSTP1 and AD cannot be completely ruled out although no significant results were indicated in the reverse MR. More research is needed to elucidate the specific mechanism between GSTP1 and AD. Nonetheless, we still did some virtual screening work to search for GSTP1 inhibitors as it is particularly promising in the context of cancer therapy, not just for AD treatment. The overexpression of GSTP1 has been associated with resistance to chemotherapy, as it can detoxify chemotherapeutic agents, reducing their efficacy [50]. Therefore, inhibitors targeting GSTP1 could potentially enhance the effectiveness of chemotherapy by preventing the enzyme from neutralizing these drugs. Six potential lead compounds for GSTP1 inhibitors were identified in our study, and more experiments including Enzyme-Linked Immunosorbent Assay (ELISA) and Surface Plasmon Resonance (SPR) are needed to further evaluate these intermolecular interactions.

Currently approved AD treatment drugs only provide symptomatic benefits. Emerging epidemiological and clinical studies suggest that lifestyle changes offer an alternative therapeutic route for slowing cognitive decline and AD development [51]. For example, adherence to a Mediterranean diet, including core foods such as fish, olive oil, fruit, and green leafy vegetables, may reduce the risk of AD [52]. Crisp foods generally refer to vegetables and fruits, which have a crisp and refreshing taste and are rich in fibre, vitamins, and minerals. In our study, more crisp food intake was significantly associated with lower BIN1 levels and thus could be a protective lifestyle factor of AD. This may be related to the involvement of vitamins in crisp foods in alleviating oxidative stress processes. Conversely, dried fruit intake was found to be negatively associated with SERPINF2, thus a risk lifestyle factor of AD. This result is consistent with a recently published MR study [53], and our findings may contribute to explaining the mechanism of more dried fruit intake increasing AD risk. For addictive behaviours, drinking and smoking are the focus of attention. Literature indicates that former/active smoking is related to a significantly increased risk for AD [29], while the relationships between alcohol drinking has dose-related associations with AD [22]. In our study, alcohol intake was not found to be related to AD, while never smoking is identified as a protective lifestyle factor for AD, with never-smokers having relatively lower BIN1 protein levels.

This study possesses several strengths. First, it employs a comprehensive methodology that combines MR, CA, PPI network construction, and virtual screening, allowing for a thorough investigation of potential therapeutic targets for AD. Second, five plasma proteins were identified as targets significantly associated with AD, particularly GSTP1 and BIN1, which have not been extensively studied in the context of AD. Third, the successful virtual screening identified six promising compounds as potential GSTP1 inhibitors, highlighting the utility of computer-aided drug discovery in identifying therapeutic candidates. Furthermore, the study offers valuable insights into the influence of lifestyle factors on AD risk, suggesting that behaviors such as increased intake of crisp foods and smoking cessation may mitigate AD risk through their impact on specific proteins. Finally, the study’s rigorous validation techniques, including Bayesian CA and reverse MR, enhance the credibility of these findings, demonstrating the interdisciplinary impact of integrating genetic, proteomic, and virtual screening approaches to advance the understanding of AD pathogenesis and identify new therapeutic strategies.

However, this study is not without limitations. First, all investigated plasma circulating proteins had only one or two cis-acting SNPs and lacked trans-pQTLs, which limited the application of various analyses, including alternative MR algorithms, as well as tests for heterogeneity and pleiotropy. Second, this study primarily utilized genetic data from European populations, and it was difficult to generalize the results to other ancestries. Variations in genetic architecture across populations could result in different associations between cis-pQTLs and AD, thereby affecting the broader applicability of the study's results. Third, despite the identification of highly promising small-molecule compounds through virtual screening, the lack of experimental validation due to inherent limitations in conditions remains a drawback. Finally, the datasets selected in this article are all European datasets, and the final research results may not be applicable to non-European populations. In future research, we will continuously expand the data sources to improve the generalizability and applicability of the research results.