Elucidate biomarkers and the molecular pathways associated with genetic variants that contribute to the etiology of Parkinson’s disease

Parkinson’s disease (PD) is a prevalent neurological disorder associated with aging. Its hallmark characteristics are the degeneration of dopaminergic neurons and the presence of Lewy bodies in the substantia nigra [18]. PD is characterized by symptoms such as resting tremors, postural instability, bradykinesia, and rigidity. The clinical presentation encompasses non-motor symptoms such as cognitive decline, hyposmia, sleep difficulty, depression, and autonomic dysfunction [8]. These symptoms worsen over time, impairing the patient’s ability to move and overall well-being. The number of PD cases tends to increase in the next few decades, resulting in a significant burden on health systems [18]. Nevertheless, the precise etiology of PD remains uncertain, but genetic factors, the aging process, and environmental factors are pivotal in the development of PD [40].

Over the last decade, around 20 genes have been linked to PD, or parkinsonism, in familial cases [7]. Although detrimental mutations in genes are crucial in the development of PD, only a small number of patients actually possess these mutations [52]. With the development of technology, genome-wide association studies (GWASs) can identify common variations with a small effect size, which have been found to significantly influence the genetic predisposition to PD. Over 300 prevalent variations (e.g., SNCA, LRRK2, BST1, GCH1, VPS13C, TMEM175, and MAPT) have been linked to age at onset, progression, and the risk of sporadic PD by GWASs [6, 21, 37, 58]. Although there have been extensive collaborative efforts involving multiple cohorts, the currently identified genetic variations can only account for 22–36% of the heritability of PD in the population of European ancestry [58]. Genetic variations have increased the challenge of finding approaches that can accurately identify, treat, and prevent this condition. Although numerous molecular pathways underlying the pathogenesis of PD associated with these variations have been identified, such as endocytosis, lysosomal function, and mitochondrial function, these molecular pathways seem insufficient to describe the precise effects of genetic variations on PD etiology [45, 58]. Thus, it is imperative to elucidate the precise molecular mechanisms underlying PD pathogenesis associated with genetic variations. Our hypothesis proposes that genetic variations in key genes may impact the biological mechanisms associated with PD. These genetic variations have the capacity to impact signaling pathways that play significant roles in the pathogenesis of PD. Understanding the link between them plays the most important role in the development of diagnostic tools and preventive therapies for PD. Therefore, the objective of this study is to determine the key genetic variations and their corresponding biological pathways that contribute to the etiology of PD using GWAS data from a total of 68 studies.

To evaluate the impact of excluding 312 records with missing beta values, a sensitivity analysis was conducted. Beta values for a subset of excluded records were estimated using odds ratios (when available) or mean beta values from similar SNPs in the dataset. Re-analysis of hub genes and pathways with these imputed values showed no significant changes in the prioritization of SNCA, LRRK2, SH3GL2, TMEM175, BST1, RIT2, or MCCC1, nor in the enrichment of key pathways (e.g., synaptic vesicle endocytosis, dopamine secretion), confirming the robustness of our findings.

After identifying the potential genetic variants, we proceeded to carry out the enrichment analysis. The pathway analysis was performed using the CytoscapeClueGO plug-in (version 2.5.8). The KEGG, Reactome, and WikiPathways databases were selected in order to obtain a complete compilation of signaling pathways. The enrichment analysis was conducted using the two-sided hypergeometric test, using a Bonferroni step-down adjustment. The terms demonstrated a correlation with each other, as evidenced by a κ score of 0.4. The ClueGO plug-in enables the incorporation of Gene Ontology (GO) terminology and KEGG/BioCarta pathways [4]. This plug-in enables the viewing of biochemical pathways and gene ontologies related to the genetic variants under investigation. Furthermore, it allows for the assessment of functional annotations by comparing them between two clusters. The present study employed a plug-in tool to clarify the molecular pathways that are associated with the genetic changes under research and are linked to PD. The protein-protein interaction (PPI) networks were constructed using the STRING v12.0 database, and further modifications were made using Cytoscape v3.9.1. The assessment was performed by examining physical interactions that displayed a significant level of certainty, indicated by a physical score over 0.4 in the STRING database. The Cytoscape plug-in, CytoHubba, was employed to extract a hub network. The creation of this network involved combining subnetworks formed by the use of degree, closeness, and betweenness techniques, employing the technique of intersectional merging. The Venn diagram tool was used to determine the proteins that are shared throughout the subnetworks, including the top 10 nodes, which were ranked using degree, closeness, and betweenness methods [12]. To visualize the candidate genes at GWAS loci, we submitted the putative genes into the PD GWAS Locus Browser (https://pdgenetics.shinyapps.io/GWASBrowser/).

The identification of proteoforms resulting from post-translational modifications was performed by leveraging the UniProt database after conducting PPI analysis [79]. The biomarkers were categorized using the Panther classification system [80]. The expression of hub proteins in brain tissues was assessed using the Human Protein Atlas, accessible at the online address: https://www.proteinatlas.org/. The ChIP-X Enrichment Analysis Version 3 (CHEA3) was employed to identify the transcription factors that may be accountable for the identified genomic alterations [42]. The present algorithm employs data acquired from the ENCODE and ReMap GTEx datasets to detect transcription factors [43]. The creation of the integrated regulatory network was carried out using the software application Cytoscape. The approach entailed the selection of the top 10 transcription factors, which were chosen based on their average rank score. Later on, the MicroRNA Enrichment Turned Network (MIENTURNET) was employed to generate and analyze networks that depict the relationships between microRNA and its target biomarkers [50]. We performed microRNA-target interaction analysis using the MIENTURNET tool, with 310 curated variant-associated genes as input. Enrichment analysis was conducted using KEGG, Reactome, and WikiPathways databases, applying a significance threshold of p < 0.05 after Benjamini-Hochberg correction. miRNAs were identified based on its significant enrichment in PD-relevant pathways, a minimum interaction score defined by MIENTURNET defaults, and further validated through brain-specific expression profiles from the Human Protein Atlas and supportive literature evidence. The functional enrichment analysis employed the KEGG, Reactome, and WikiPathway databases in combination with the disease ontology database. A significance level of 0.05 was established in order to identify the functional annotations that exhibited significant enrichment throughout the whole gene group in the input set. The P-values underwent correction using the Benjamini-Hochberg approach [51]. The study utilized the R programming language version 4.0.2 for performing general data transformations and groupings. The activities were completed using the tidyverse package collection (version 1.3.2). The charts were created using the R packages ggplot2 v3.4.0, ggrepel v0.9.3, and ggpubr v0.5.0. The figure panels were created using the online platform BioRender.com.

This study used data from 68 studies (after cleaning data) about genetic variants implicated in the pathogenesis of PD. A total of 542 variant and risk alleles were located across all chromosomes, especially chromosomes 4 and 17. As shown in Fig. 1A–B, SNCA and TMEM175 located in chromosome number 4 showed high frequencies and traits (e.g., age at onset, cognitive progression, motor progression, composite progression, and tremor dominant and postural instability gait difficulty, etc.). The majority of these genes included various variant and risk alleles, which exert doubled nature (increase (\(\uparrow\)) or decrease (\(\downarrow\)) risk of PD). Most of these genes exert binding (21.6%) and catalytic activities (18.0%) (Fig. 1C). After the cleaning process, there were 310 genetic variations, including 167 with a higher risk of PD and 143 with a lower risk of PD (Supplementary Table S2). Furthermore, there were 14, 12, and 16 genetic variations associated with the cognitive, motor, and composite progression of PD (Fig. 1D). There was significant evidence of impaired synaptic function, vesicle-mediated transport, and neuron projection associated with variant and risk alleles of PD (Fig. 1E).

Three hub genes were identified after network topological analysis (Fig. 2A–D), including SNCA (\(\uparrow\)rs5019538 and \(\uparrow\)rs356182), (LRRK2 (\(\uparrow\)rs34637584 and \(\uparrow\)rs76904798), and SH3GL2 (\(\uparrow\)rs10756907 and \(\downarrow\)rs13294100). It has been well known that SNCA and LRRK2 are the prevalent genes underlying PD etiology. So, this study only analyzed the expression of SH3GL2. As shown in Fig. 3A–B, the expression of SH3GL2 was mostly in the brain, especially in the cerebral cortex. SH3GL2 is also found in the clathrin-coated endocytic vesicle membrane, glutamatergic synapse, and presynapse (Fig. 3C). SH3GL2 was found to be a part of cluster 2 neurons – nucleosome; SH3GL2 showed high correlations with H2B clustered histone 15 (r = 0.874, cluster 34) and mitogen-activated protein kinase 9 (r = 0.856, cluster 37) (Fig. 3D). We also found SH3GL2 interacted with other enzymes (SYNJ1, CBL, PTPN23, and ITCH), transporters (LRRK2, EGRR, and DNM2), and other proteins in the brain (Fig. 3E–F). There was substantial evidence of impaired dopamine secretion, receptor recycling, and oxidoreductase activity and increased amyloid-beta formation associated with genetic variations with a higher risk of PD (Fig. 3G). Significant evidence indicated improved synaptic vesicle pathway, neuron projection development, and regulated histone methylation and excitatory postsynaptic potential related to genetic variants that carry a lower risk of PD (Fig. 4A–C).

Next, we further analyzed the molecular mechanisms underlying the pathogenesis of the PD subgroup, including age at onset, cognitive progression, motor progression, composite progression, and tremor dominant and postural instability gait difficulty. Genetic variations (SNCA (\(\uparrow\)rs356203), AAK1 (\(\downarrow\)rs7577851), OCA2 (\(\downarrow\)rs17565841), GABRG3 (\(\downarrow\)rs17565841), and ATF6 (\(\downarrow\)rs10918270)) associated with age at onset of PD were found to be related to SNARE (soluble N-ethylmaleimide sensitive factor attachment protein receptor) complex assembly, ly-tyrosine transmembrane transporter activity, recruitment of AP-2 complex and clathrin, GABAA receptors, and ATF6 (activating transcription factor 6)-mediated unfolded protein response (Fig. 5A). In terms of cognitive progression, genetic variations (APOE (\(\downarrow\)rs429358), (NTRK2 (\(\downarrow\)rs148603475), SLCO1B3 (SLC28A3 (\(\downarrow\)rs148603475)) were related to negative regulation of amyloid beta formation and transport of vitamins, nucleosides, and related molecules (Fig. 5B). As shown in Fig. 5C, genetic variations (AQP10 (\(\downarrow\)rs35950207), SNCAIP (\(\uparrow\)rs5870994), ANO2 (\(\downarrow\)rs74709761), CADM1 (\(\uparrow\)rs4436579), and PTPRD (\(\uparrow\)rs7870456)) were underlying the motor progression of PD via altering aquaporins that passively transport urea out of cells, binding to alpha-synuclein and Ca²⁺, Necl-1:Necl-2 trans heterodimer interaction, and synaptic adhesion (Fig. 5D). Furthermore, we found genetic variations (APOE (\(\uparrow\)rs429358), GPR32 (\(\uparrow\)rs4802739), GPR321 (\(\uparrow\)rs4802739), SNCAIP (\(\uparrow\)rs17367669), SQOR (\(\uparrow\)rsrs17554587), and SULT1C2 (\(\uparrow\)rs13424530)) underlying the composite progression of PD by altering complement receptor activity, lipoprotein particle clearance, binding alpha synuclein, SQR (sulfide quinone oxidoreductase), oxidizing sulfide to bound persulfide, and cytosolic sulfonation (Fig. 5E). On the other hand, we also found genetic variations (GABRG2 (\(\uparrow\)rs11949046), CYP4Z1 (\(\uparrow\)rs116504637), CDH13 (\(\downarrow\)rs13330839), FANCF (\(\downarrow\)rs55971529)) were involved in the pathogenesis of tremor dominant and postural instability gait difficulty by altering low-density lipoprotein particle-mediated signaling, cellular response to histamine, lauric acid metabolic process, and FANCD2 deubiquitination (Fig. 5F).

Most miRNAs are found in the central nervous system, especially during brain development [46]. Dysregulation of miRNA transcript expression has been observed in various neurodegenerative diseases, including PD [75]. We performed microRNA–target interaction analysis using the MIENTURNET tool, with 310 curated variant-associated genes as input. Enrichment was assessed against KEGG, Reactome, and WikiPathways databases, applying a significance threshold of p < 0.05 after Benjamini–Hochberg correction. We identified three candidate miRNAs (hsa-miR-16-5p, hsa-miR-17-5p, and hsa-miR-20a-5p); however, only hsa-miR-20a-5p showed significant enrichment (false discovery rate (FDR) = 0.0394) in PD-relevant pathways. Specifically, hsa-miR-20a-5p was associated with inositol phosphate metabolism, adhesion junction, phosphatidylinositol signaling, and Th17 cell differentiation (Fig. 6A, Supplementary Table S3). We also identified the transcription factors underlying the pathogenesis of PD associated with genetic variations. As shown in Fig. 6B, MYT1L (myelin transcription factor 1 like) was the predominant transcription factor (the first ranking based on mean-rank) among the top 10 transcription factors. The expression of MYT1L was found in the SH-SY5Y cell line and across the brain, especially the cerebral cortex.

To address the contribution of effect sizes to biomarker identification, we analyzed the beta values of the 232 genetic variants associated with PD and related traits from 68 GWAS studies (Supplementary Table S1). Beta values, representing the change in PD risk or trait progression per risk allele, were categorized by effect direction: 125 variants were risk-increasing (mean beta = 0.245, SD = 0.346, range = 0.0529–2.4289), and 107 were risk-decreasing (mean beta = − 0.209, SD = 0.231, range = − 2.84 to − 0.0529). These statistics highlight the variability in effect sizes, with risk-increasing variants showing larger magnitudes on average, driven by outliers such as LRRK2 (rs34637584, beta = 2.4289, p = 4e-82) and GBA1 (rs421016, beta = 1.979, p = 1e-14). To illustrate the distribution of effect sizes, box plots were generated for risk-increasing and risk-decreasing variants across traits (Supplementary Fig. S1). These plots highlight the variability in beta values and emphasize variants with |beta| >0.3, particularly for LRRK2, GBA1, and SNCA, which are critical for PD pathogenesis.

High-effect-size variants were mapped to hub genes identified in our network analysis. LRRK2 (rs34637584, beta = 2.4289) and GBA1 (rs421016-G, beta = 1.979) showed the largest effects, reinforcing their roles in autophagy and lysosomal dysfunction. SNCA variants (e.g., rs356203, beta = 1.4337) underscored its centrality in Lewy body pathology. TMEM175 (rs34311866-C, beta = −0.613) and SH3GL2 (rs10756907, beta = −0.0926) aligned with lysosomal and synaptic vesicle pathways, respectively. BST1 (rs4698412, beta = 0.1035) and RIT2 (rs12456492, beta = −0.0983) showed moderate effects, consistent with their roles in neuroinflammation and dopamine signaling. MCCC1 (rs10513789, beta = 0.173) was linked to oxidative stress pathways. These findings confirm that variants with large effect sizes correspond to biologically relevant pathways, enhancing the reliability of our biomarker prioritization.

This study further analyzed the molecular mechanisms underlying the pathogenesis of PD associated with genetic variations. We observed three hub genes (SNCA, LRRK2, and SH3GL2) that carry different variants underlying the pathogenesis of PD. Other biomarkers (APOE, NTRK2, SLCO1B3, SLC28A3, AQP10, SNCAIP, ANO2, CADM1, PTPRD, GPR32, GPR321, SQOR, SULT1C2, GABRG2, CYP4Z1, CDH13, and FANCF) were also related to the clinical characteristics of PD (e.g., cognition, motor, age at onset, etc.). There was significant evidence of altered synaptic function and neuron projection development associated with the studied genetic variations.

The present work entailed a comprehensive examination of data generated from the GWAS database, with the aim of identifying noteworthy genetic differences and elucidating the underlying molecular pathways connected with the etiology of PD arising from genetic variants. Six common biomarkers (SNCA, TMEM175, BST1, RIT2, LRRK2, and MCCC1) associated with PD were detected across all 68 studies. SNCA (\(\uparrow\)rs5019538 and (\(\uparrow\)rs356182), LRRK2 (\(\uparrow\)rs34637584 and (\(\uparrow\)rs76904798), and SH3GL2 (\(\uparrow\)rs10756907 and (\(\downarrow\)rs13294100) were the predominant biomarkers underlying PD. Other biomarkers (APOE, NTRK2, SLCO1B3, SLC28A3, AQP10, SNCAIP, ANO2, CADM1, PTPRD, GPR32, GPR321, SQOR, SULT1C2, GABRG2, CYP4Z1, CDH13, and FANCF) have been found underlying the clinical traits of PD, including age at onset, cognitive progression, motor progression, composite progression, and tremor dominant and postural instability gait difficulty. There was substantial evidence of impaired dopamine secretion, receptor recycling, and oxidoreductase activity and increased amyloid-beta formation associated with genetic variations with a higher risk of PD. Significant evidence indicated improved synaptic vesicle pathway, neuron projection development, and regulated histone methylation and excitatory postsynaptic potential related to genetic variants that carry a lower risk of PD. We also identified hsa-miR-20a-5p and MYT1L that play a crucial role in elucidating the genetic variants associated with PD. The aforementioned discoveries serve as a fundamental basis for potential therapeutic interventions targeting PD, with a particular emphasis on the genetic variations and mechanisms associated with the condition.