Exploring Potential Causal Links: How Parkinson’s Disease Brain Damage Impacts Liver Health–A Mendelian Randomization and Transcriptomic Study

Introduction

Parkinson’s disease (PD) is clinically defined by movement-related manifestations such as slowed movements (bradykinesia), muscle stiffness, resting tremors, and impaired balance coordination, which together constitute parkinsonism.¹ Ranking as the second most common neurodegenerative condition after Alzheimer’s disease, PD demonstrates the fastest-growing epidemiological burden among neurological disorders, currently affecting more than six million individuals globally.² Although advancing age remains the primary risk determinant, modifiable environmental exposures and behavioral patterns significantly contribute to the development of age-associated neurodegenerative conditions.³ The neuropathological hallmarks of PD include progressive loss of dopamine-producing neurons in the substantia nigra pars compacta (SNpc) coupled with abnormal aggregation of α-synuclein (α-syn) proteins forming characteristic Lewy bodies (LBs) inclusions.⁴ Disease progression involves multiple pathogenic mechanisms including oxidative stress from excessive free radicals, disrupted cellular energy metabolism, and compromised mitochondrial functionality, all contributing to the selective degeneration of dopaminergic neurons.⁵

Recent clinical evidence indicates a significant connection between the central nervous system (CNS) and liver diseases, particularly in patients with PD.^6,7 Understanding the interrelationship between the CNS and liver diseases is essential for assessing the overall condition and disease progression in patients with PD. Clinical data suggest that as neurodegenerative disorders progress, the liver, a critical organ for maintaining metabolic homeostasis, is the first to show functional impairment, followed by the kidneys and heart.⁸ The liver plays a vital role in various physiological activities, including cholesterol homeostasis, immune support, protein synthesis, and whole-body metabolism.⁹ This organ is composed of highly diverse groups of cells that work together to carry out the organ’s functions in both healthy and diseased individuals.¹⁰ Studies have demonstrated that several metabolic disorders, including PD, are associated with dysregulation of liver-related physiological processes.¹¹ Liver P450 subsystem damage has been reported in patients with delayed-onset PD.^12,13 The liver is pivotal in maintaining the homeostasis of whole - body glutathione (GSH), and among all organs, it holds the highest GSH concentration.¹⁴ GSH is of great significance in the cellular antioxidant defense system and the detoxification process.¹⁵ The availability of cysteine is a key factor influencing GSH synthesis. Conversely, in specific cell types, an abnormally high level of cysteine can give rise to oxidative stress and cytotoxicity.¹⁶ Additionally, Squitti et al¹⁷ reported that patients with PD residing near iron alloy factories had elevated levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT), along with subclinical liver dysfunction.

Emerging studies indicate that hepatic dysfunction in PD patients may intensify cerebral damage, mirroring the pathophysiological mechanisms observed in acquired hepatocerebral degeneration (AHD). This neurological disorder, typically associated with end-stage liver cirrhosis (LC) or portosystemic shunting, manifests through progressive motor abnormalities and cognitive deficits similar to those seen in advanced PD cases. Movement disorders are the most common neurological manifestations of AHD, with approximately 20% of patients with AHD exhibiting parkinsonism. The exact pathophysiology of AHD remains unclear. None of the proposed hypotheses encompass all the plausible mechanisms believed to contribute to its development. Emerging evidence points to compromised hepatic detoxification pathways as a critical driver in acquired hepatocerebral degeneration (AHD), where impaired toxin elimination—particularly manganese and ammonia—creates a neurotoxic milieu. While hyperammonemia was historically emphasized in AHD pathogenesis, contemporary research prioritizes manganese dysregulation as a central mechanistic framework. The neurotoxic effects of ammonia appear mediated through astrocytic dysfunction, specifically disrupting the glutamate-glutamine cycle. This process critically depends on glutamine synthetase activity, an astrocyte-specific enzyme essential for converting glutamate to glutamine, thereby maintaining cerebral excitatory neurotransmission homeostasis.¹⁸ The increase in ammonia concentration in chronic liver disease inhibits the clearance of glutamine by astrocytes through glutamine synthetase, leading to the accumulation of glutamine and mild cerebral edema.¹⁹ Overall, liver damage exacerbates brain injury in PD. However, the impact of brain injury on liver function in patients with PD, considering the intricate interplay between the liver and the brain, remains insufficiently studied.

This study presents groundbreaking evidence for a bidirectional causal relationship between PD-related brain injury and liver dysfunction through an innovative combination of Mendelian randomization (MR) and transcriptomic analysis. Leveraging genome-wide association study (GWAS) data, we employed bidirectional MR to systematically evaluate the PD-liver interaction, revealing serum cysteine as a crucial metabolic mediator that exhibits concentration-dependent effects - neuroprotective at low levels but hepatotoxic at elevated concentrations. Importantly, we identified specific immune-inflammatory hub genes (neutrophil cytosolic factor 1 (NCF1), spleen-associated tyrosine kinase (SYK)) through which the anti-PD drug tolcapone exerts its hepatotoxic effects, providing the first molecular explanation for its known liver injury risks.

The clinical implications of these findings are particularly significant given that PD patients typically require long-term pharmacotherapy with drugs like levodopa. Our multi-omics approach, integrating MR with transcriptomic profiling of PD and liver injury models, demonstrates that PD-associated brain injury can disrupt hepatic metabolic function through both circulating metabolites (cysteine) and shared molecular pathways. This systemic perspective advances our understanding of PD beyond its classical neurological manifestations, revealing it as a disorder of inter-organ communication. Methodologically, we employed a rigorous analytical pipeline: (1) bidirectional MR to establish causal directions in the PD-liver axis; (2) serum metabolite analysis to identify cysteine as a key mediator; (3) transcriptomic comparison to uncover shared gene networks in PD and hepatitis B virus (HBV); and (4) experimental validation of hub genes in peripheral blood mononuclear cells (PBMCs) and liver cells. This comprehensive strategy not only confirms the brain-liver connection in PD but also provides clinically actionable biomarkers for monitoring hepatic health in PD patients undergoing long-term treatment, potentially enabling personalized medication regimens to minimize side effects while maintaining therapeutic efficacy. The flowchart of this study is presented in Figure 1.

Figure 1 The flow chart of this study. Abbreviations: GSEA, gene set enrichment analysis; WGCNA, Weighted correlation network analysis; PPI, protein-protein interaction networks.

Materials and Methods

MR Analysis

Establishment of the Program

The study design followed the basic principles of MR.^20,21 PD was initially treated as the exposure variable, while relevant liver injury diseases were considered outcome variables. Concurrently, serum metabolites were examined as outcome variables to identify the metabolites that PD may affect. Subsequently, these serum metabolites were further screened as exposure variables to determine their impact on the liver. The overall findings revealed the specific metabolic pathways through which PD affects the liver. During the analysis, stringent inclusion and exclusion criteria were applied, selecting single nucleotide polymorphisms (SNPs) significantly associated with the exposure target as instrumental variables (IVs) for two-sample MR. A series of sensitivity analyses were conducted on the statistically significant results. Notably, publicly available summary statistics from GWAS were used in this analysis, and the original GWAS was approved by the relevant institutional review boards.

Serum Metabolite Data

The serum metabolite GWAS involved comprehensive genetic association scans of 486 metabolites in 7,824 adult individuals from two European populations, resulting in complete GWAS statistical data for human blood metabolites. Utilizing structural and biochemical profiling, the metabolites were stratified into nine distinct biosignature clusters—eight encompassing principal biochemical superfamilies and one comprising unclassified entities. The defined categories spanned essential biomolecular domains: nitrogenous bases (nucleotides), nitrogen metabolism derivatives (amino acids/peptides), saccharide derivatives (carbohydrates), redox mediators (cofactors/vitamins), bioenergetic intermediates, lipid species, along with xenobiotic compounds.²²

PD and Hepatitis Sample Data

The GWAS data for diseases were obtained from the Open GWAS database (https://gwas.mrcieu.ac.uk/), which comprises at least 12.6 billion genetic associations from over 14,582 complete GWAS datasets, representing a diverse range of human phenotypes and disease outcomes across different populations. In this experiment, two types of diseases were included as the analysis targets: PD and liver injury-related diseases. Liver injury-related diseases include viral hepatitis (such as hepatitis B and hepatitis C), autoimmune hepatitis, nonalcoholic fatty liver disease, liver fibrosis and cirrhosis. In addition, it also includes GWAS data about liver function abnormalities. Specifically excluded were liver cancer and alcoholic fatty liver disease types, as their primary causes were not related to brain injury-related diseases. Detailed information on specific diseases is presented in Supplementary Table S1.

Selection of IVs

To obtain effective and significant SNPs as IVs, different treatments were applied when different targets were considered as exposures. When serum metabolites and PD were considered as exposures separately, the threshold for screening SNPs significantly associated with exposure was set at P < 1×10⁻⁵ and P < 5×10⁻⁸, respectively. Additionally, SNPs in linkage disequilibrium were excluded using the European population reference group from the 1000 Genomes Project, with parameters set as r² < 0.01 (500 kb) and r² < 0.001 (1000 kb). Furthermore, the selected SNPs were harmonized using summary data from GWAS outcomes, and any palindromic SNPs or SNPs not available in the GWAS results were discarded. The filtered SNPs were then utilized as final IVs MR analysis. To validate the instrumental strength assumption for the analysis, the proportion of exposure variance explained by genetic variation (R²) was estimated using the following formula where β_i represents the effect size of genetic instrument variable i, N is the effective sample size, se(β_i) is the standard error of the effect size for genetic variation i, and K is the number of independent genetic variations.²³ Additionally, the F-statistic was computed as and compared to an empirical threshold of 10 to assess the strength of these genetic instruments.²⁴

Two-Sample MR

In bidirectional Mendelian randomization (MR) investigations, analytical approaches were stratified by genetic instrument complexity. For mono-SNP exposures, causal estimates were derived from the Wald ratio estimator, computed as the quotient of outcome-associated to exposure-associated SNP effects. Poly-SNP instruments employed inverse variance-weighted (IVW) meta-analyses under fixed/random effects paradigms, supplemented by pleiotropy-robust methodologies including weighted median estimator, MR-Egger regression, and simplified/weighted modeling frameworks to validate result stability.²⁵ Instrumental variable (IV) heterogeneity was quantified through Cochrane’s Q statistics, with random-effects IVW implementation when exceeding statistical significance threshold (α=0.05). The weighted median approach maintained estimation consistency when valid IVs contributed >50% analytical weight,²⁶ whereas MR-Egger accommodated directional pleiotropy even under majority invalid IV scenarios.²⁷ Outlier detection via MR-PRESSO enabled IVW estimate correction through iterative exclusion of heterogeneous variants, complemented by leave-one-out sensitivity analyses to identify SNP-driven result distortions.

All computations were implemented through R v4.0.3, utilizing TwoSampleMR for core MR methodologies (IVW, weighted median, MR-Egger) and MRPRESSO for outlier detection protocols.

Analysis of Differentially Expressed Genes (DEGs) in PD and HBV Based on GEO Database

DEGs Acquisition

Parkinson’s disease transcriptomic profiles were sourced from GEO repositories (GSE20141/GSE7621; 26 PD cases vs 17 controls), while hepatitis B virus (HBV) data derived from GSE83148 (122 hBV-infected individuals vs 6 controls). All microarray datasets were profiled on the Affymetrix HG-U133 Plus 2.0 array (GPL570). Differentially expressed genes were identified using stringent thresholds (|logFC| > 0.58 and P.adj < 0.05). Validation cohorts encompassed PD series (GSE20163/GSE20164/GSE99039) and HBV datasets (GSE38941/GSE58208), with comprehensive cohort characteristics detailed in Table 1.

Table 1 The Detailed Information of the Dataset

Gene Set Enrichment Analysis (GSEA) and Immune Infiltration Analysis

Gene Set Enrichment Analysis (GSEA v4.1.0)²⁸ was implemented to interrogate molecular pathways underlying PD and HBV pathogenesis. Immune microenvironment profiling leveraged single-sample GSEA (ssGSEA) via the GSVA package,²⁹ quantifying 24 reference-curated immune cell signatures within the transcriptomic datasets.³⁰ Differential immune cell abundance across PD/HBV cohorts versus controls was visualized through hierarchical clustering (heatmaps) and comparative distribution plots (boxplots). Significance thresholds for DEGs and pathway enrichment were established at Benjamini-Hochberg P-value < 0.05.

Biofunctional Enrichment Analysis and Weighted Correlation Network Analysis (WGCNA)

KEGG pathway and GO enrichment analyses for DEGs were conducted using the “clusterProfiler” and “GOplot” packages within the R environment. WGCNA was carried out to detect co-expressed gene modules, investigate relationships between gene networks and phenotypes of interest, and identify hub genes within these networks. A weighted adjacency matrix was constructed, and a correlation power (soft thresholding parameter) was determined to emphasize strong gene correlations while downweighting weak ones. This adjacency matrix was subsequently transformed into a topological overlap matrix (TOM) to assess gene network connectivity. The TOM aggregated adjacent genes based on the network’s gene ratio and computed corresponding dissimilarity values. Average linkage hierarchical clustering, based on TOM dissimilarity measurements, was applied to group genes with similar expression patterns into modules, which were depicted by branches and distinct colors in the cluster tree, establish module relationships, and filter genes.³¹

Protein-Protein Interaction (PPI) Analysis and Screening of Hub Genes

In R software (version 4.2.0), the Venn diagram visualization of overlapping genes between PD and HBV datasets derived from WGCNA analysis was generated using the “venneuler” package. The protein-protein interaction network for 30 DEGs was constructed through the STRING database (https://cn.string-db.org/). Top-ranked hub genes were selected through maximal clique centrality (MCC) scoring implemented in the Cytohubba module of Cytoscape platform (v3.9.0). Comparative expression profiles of these key genes in PD/HBV patients versus healthy individuals were visualized using boxplot graphical representations. Subsequent validation identified six hub genes through cross-referencing PBMC data from PD cohort GSE99039 with the initial 10 candidate hub genes.

Identification of Diagnostic Genes

Diagnostic biomarker identification was conducted through receiver operating characteristic (ROC) curve evaluation employing the pROC package within the R software. Quantitative assessment of gene discriminatory capacity involved computing area under ROC curve (AUC) metrics, with a selection criterion established at AUC thresholds exceeding 0.70 for potential diagnostic biomarkers. The predictive performance of key hub genes was systematically assessed through these analytical procedures.

Anti-PD Drugs and Molecular Docking

Pharmacological agents targeting PD were acquired from the DrugBank repository (https://go.drugbank.com/). Target prediction for these compounds was executed through the Comparative Toxicogenomics Database (CTD, https://ctdbase.org/). Network visualization of drug-target relationships was achieved using Cytoscape 3.9.0’s graph theoretical framework. Toxicogenomic associations with hepatic injury were investigated by mining clinical evidence from PubMed (https://pubmed.ncbi.nlm.nih.gov/). Computational docking simulations between tolcapone and prioritized hub genes were conducted using AutoDock Vina 1.2.2,³² with ligand structures sourced from PubChem (https://pubchem.ncbi.nlm.nih.gov/)³³ and protein conformations obtained from PDB (the Protein Data Bank, http://www.rcsb.org/). Structural preparation involved: (1) Conversion to PDBQT format with removal of crystallographic water molecules; (2) Optimization of charge states through polar hydrogen addition; (3) Establishment of 30ų cubic docking grids centered on protein functional domains, featuring 0.05nm grid resolution for precise spatial sampling. Comprehensive validation encompassed both molecular interaction patterns of tolcapone with hub gene products and systematic evaluation of hepatotoxic potential through integrated database analyses.

In vivo and in vitro Experimental Verification

Animal

Experimental male C57BL/6J mice (6–8 weeks, 25–30g) were procured from Cyagen Biosciences (USA) and maintained in controlled vivarium conditions (12h photoperiod, 22±2°C ambient temperature, 55±5% humidity) with ad libitum access to food and water. All experimental protocols complied with NIH guidelines for laboratory animal welfare and received ethical approval from Ethics Committee of Henan University (no. HUSOM2021-161). Following two-week environmental acclimatization, subjects underwent masked randomization into two cohorts (n=8/group): Saline (0.9% NaCl) and MPTP-induced PD model (20 mg/kg/d i.p. 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine), both administered for 15 consecutive days.³⁴ Terminal procedures at day 16 included: Dual-modality sampling (n=4/group): Intracardiac saline perfusion followed by 4% paraformaldehyde fixation for brain/liver histology; Hematological analysis (n=8): Retro-orbital blood collection; Molecular preservation (n=4/group): Rapid dissection of substantia nigra (SN) and hepatic tissue under cryogenic conditions, with subsequent snap-freezing in liquid nitrogen and −80°C archival storage.

Cell Culture Assay

The MN9D and AML12 cells line were purchased from Servicebio (China, Wuhan). Dopaminergic neuronal cell line MN9D, mouse liver cell line AML12, and co-cultured MN9D and AML12 cells at a 1:1 ratio were cultured in high-glucose Dulbecco’s modified Eagle’s medium (DMEM) (Gibco Laboratories, Grand Island, NY, USA) containing 10% fetal bovine serum (FBS), penicillin (100 units/mL), and streptomycin and incubated at 37°C in a humidified atmosphere with 5% CO₂. The culture medium was replaced every 2–3 days.

Cell Viability Testing

The MN9D and AML12 cells were treated with different concentrations of 1-methyl-4-phenylpyridinium (MPP⁺) and Cys (0, 5, 10, 20, 40, 80, 160, 320, 640, and 1280 μM) for 24 h. The treated cells were then seeded in a 96-well plate with cell suspension at a density of 5×10³/200 μL/well and incubated at 37°C in a humidified atmosphere with 5% CO₂. 10 μL of CCK-8 (catalog number: G4103, Servicebio) solution was added to each well, and the plates were further incubated for 4 h. The absorbance at 450 nm was measured using a microplate reader.

Cys, ROS, ALT, and AST Level Determination

The SN, liver, and blood samples from the saline and MPTP groups, the MPP⁺-induced PD cell model, and the MPP⁺-induced PD cell model were co-cultured with AML12 cells and cysteine content was detected using a cysteine colorimetric assay kit.(catalog number: E-BC-K352-M, Elabscience). reactive oxygen species (ROS) level determination in AML12 cells was performed using ROS detection kit (catalog number: G1706, Servicebio), examining different concentrations of Cys (0, 5, 10, 20, 50, 100, 200, and 300 μM) and control, MPP⁺ (60 μM), Cys (30 μM), and the Cys+MPP⁺ group. Blood and AML12 cell levels of ALT and AST were measured using ALT and AST assay kits (catalog numbers: SYP-M0414QX and SYP-M0413QX, UpingBio).

Total Protein Extraction and Western Blot Analysis

Cells across various treatment groups were lysed using RIPA buffer containing PMSF (product codes: G2002 and G2008; Servicebio). The lysis process involved sonication followed by centrifugation at 12,000 ×g for 10 minutes at 4°C. The protein concentration in the resulting supernatant was quantified via a BCA assay. Subsequently, proteins were separated by SDS-PAGE and transferred onto a nitrocellulose membrane (Millipore, IPFL00010, Germany). Membranes were blocked with 5% nonfat milk in TBST for 1 hour at room temperature, then incubated overnight at 4°C with primary antibodies diluted in TBST containing 1% nonfat milk. The primary antibody panel included: α-syn, nuclear factor-κB (NF-κB p65), tumour necrosis factor α (TNFα), interleukin-1β (IL-1β), IL-6, nuclear factor-erythroid 2-related factor 2 (Nrf2), heme oxygenase 1 (HO-1), NAD(P)H quinone dehydrogenase 1 (NQO1), Caspase-3 (CASP3), NCF1, p-NCF1, selectin P ligand (SELPLG), SYK, p-SYK, GTPase-activating protein 30 (ARHGAP30) (1:1000 dilution, Affinity), tyrosine hydroxylase (TH) (1:500 dilution, Affinity), SAM and SH3 domain-containing 3 (SASH3) (Huabio, ER61213), and GAPDH (1:3000 dilution, Affinity). After primary antibody incubation, membranes were probed with HRP-conjugated secondary antibodies in 1% nonfat milk-TBST for 2 hours at room temperature. Following three 10-minute TBST washes, protein bands were visualized using chemiluminescence (Bio-Rad). Densitometric analysis of Western blot images was performed using ImageJ software.

Analysis of Immunofluorescence and Immunohistochemistry

MN9D and AML12 cells were seeded on slides, washed with PBS, and fixed with 4% paraformaldehyde at room temperature for 30 minutes. For SH-SY5Y cells cultured on chamber slides, permeabilization was achieved using 0.1% Triton X-100 for 10 minutes, followed by three PBS washes. The cells were then incubated overnight at 4°C with primary antibodies (α-syn, TH, NF-κB p65, Nrf2, and CASP3; 1:200 dilution, Affinity) diluted in 10% goat serum. After primary antibody incubation, slides were washed three times with PBS and incubated with fluorescent secondary antibodies (1:2000 dilution, Invitrogen) for 2 hours at room temperature. Nuclei were stained with DAPI for 10 minutes to visualize cell nuclei.

For immunohistochemistry, 20-μm-thick serial sections of brain and liver tissues were used. Antigen retrieval was performed by heating in 1× citrate antigen retrieval solution at 98°C for 10 minutes, followed by permeabilization with 0.1% Triton X-100 for 10 minutes. Sections were treated with hydrogen peroxide for 20 minutes in the dark to quench endogenous peroxidase activity, then blocked with 10% goat serum in PBS. Primary antibodies (TH, CASP3, and LC3A/B; all at 1:200 dilution, Affinity) were applied, and immunoreactions were visualized using a DAB kit (G1212, Servicebio). Images were captured using a Leica DMI4000B microscope (Wetzlar, Germany). Densitometric analysis of scanned immunofluorescence and immunohistochemistry images was conducted using ImageJ software to quantify signal intensity.

HE and Masson Staining

Hepatic tissue sections were stained using the HE high-definition histochemical staining kit (G1076, Servicebio) and Masson’s trichrome staining kit (G1006, Servicebio) to evaluate morphological changes. The stained liver tissues were observed under a light microscope.

Statistical Analysis

Data are presented as the mean ± SEM. Comparisons between groups were performed using an unpaired Student’s t-test and one-way ANOVA analysis of variance. ROCs were used to evaluate AUCs and the predictive abilities of diagnostic genes. All statistical tests (t-tests, ANOVA) used P < 0.05, with adjustments for multiple comparisons (Bonferroni) where applicable.

Results

MR Analysis

The Impact of PD on Liver Function

To investigate the impact of PD on liver function, IVs were selected for MR analysis. After a series of quality control steps, different numbers of SNPs were selected as IVs for various liver disease outcomes. The mean F-statistics for the IVs for different outcomes were all >30, indicating no evidence of weak instrument bias (Supplementary Table S2). When MR-PRESSO identified outlier SNPs during the overall test, MR analyses were re-performed after their removal, and no evidence of pleiotropic effects was found after exclusion (P > 0.05) (Supplementary Table S3).

Using a threshold of P_IVW < 0.05 for initial screening, the effects of PD on liver function were preliminarily assessed. The results indicated that when 33 liver function abnormalities or indicators were used as outcomes for MR analysis, PD had no significant causal effect on liver function, except for chronic hepatitis (finn-b-K11_CHRONHEP, finn-b-K11_DISLIVOTH), which showed a significant negative association. This association was further supported by two other auxiliary MR methods, all showing consistent direction and magnitude (Supplementary Table S3). Additionally, the P-value of Cochran’s Q calculated by MR-Egger and IVW methods indicated no heterogeneity, and the intercept of the MR-Egger test suggested no horizontal pleiotropy. These results suggest that PD may reduce the incidence of hepatitis to some extent. The unusual finding that PD may reduce the incidence of chronic hepatitis piqued our interest, as previous studies have suggested that PD’s regulation of distant target organs may occur through serum metabolites. Next, we examined the effects of PD on serum metabolites.

Effect of PD on Serum Metabolites

To analyze the impact of PD on serum metabolites, IVs were screened for MR analysis. The mean F-statistic values of the screened IVs were all >30, indicating no evidence of weak instrument bias. When MR-PRESSO identified outlier SNPs during the overall test, they were removed, and MR analyses were rerun. After removal, no evidence of pleiotropy was observed (P > 0.05) (Supplementary Table S4).

Using a threshold of P_IVW < 0.05, 14 metabolites influenced by PD were preliminarily screened (Supplementary Table S5). Among these, 7 compounds, including lysine, were found to be reduced, while the remaining 7 compounds, including eicosapentaenoate (EPA; 20:5n3), were increased (Supplementary Figure S1). These findings were corroborated by other auxiliary MR methods, which provided robust support with all algorithms showing consistent direction and magnitude (Supplementary Table S5). The P-value of Cochran’s Q calculated by MR-Egger and IVW methods indicated no heterogeneity, and the intercept of the MR-Egger test suggested no horizontal pleiotropy. These pathways were considered potential pathways through which PD affects liver function.

Effect of Serum Metabolites on Liver Function

Having identified the types of serum metabolites influenced by PD and chronic hepatitis, the next step was to explore how PD-influenced serum metabolites affect chronic hepatitis. Effective IVs for screening target serum metabolites were selected (Supplementary Table S6) based on statistical power parameters.

Using a threshold of P_IVW < 0.05, the preliminary screening of serum metabolites affecting hepatitis revealed no significant results when finn-b-CHRONHEP_NAS was the outcome (Supplementary Table S7). However, when finn-b-K11_CHRONHEP was used as the outcome, the effect of M31453 was found to be positively associated with chronic hepatitis (Supplementary Table S7), suggesting that cysteine mediates the effect of PD on chronic hepatitis (Figure 2). However, cysteine undergoes dynamic changes in the body, and how PD specifically influences liver function through cysteine as a mediator is crucial.

Figure 2 Cysteine mediates the effect of PD on chronic hepatitis. (A) Scatter plot of the MR analysis results between PD outcome and cysteine. (B) Effect estimates of the MR analysis results between PD outcome and cysteine. (C) Funnel plot of the MR analysis results between PD outcome and cysteine. (D) Leave-one-out sensitivity analysis results of the MR analysis results between PD outcome and cysteine. (E) Scatter plot of the MR analysis results between cysteine and hepatitis outcome. (F) Effect estimates of the MR analysis results between cysteine and hepatitis outcome. (G) Funnel plot of the MR analysis results between cysteine and hepatitis outcome. (H) Leave-one-out sensitivity analysis results of the MR analysis results between cysteine and hepatitis outcome.

Dual Role of Cysteine

Determination of Cysteine Content

The concentrations of cysteine were separately determined in the SN, liver, and serum. In the MPTP group (PD model), cysteine levels were increased compared to the saline group (Figure 3A–C). Moreover, the IC50 of MPP⁺ in MN9D cells was 62.7 μM, while in AML12 cells, it was 34.38 μM. The IC50 of cysteine in MN9D cells was 52.54 μM, and in AML12 cells, it was 132.6 μM (Supplementary Figure S2A–S2D). Additionally, in the MPP⁺-induced PD cell model, the MPP⁺ group exhibited an increase in cysteine concentration compared to the control group (Figure 3D). When the PD cell model from MN9D was co-cultured with AML12 cells, there was an increase in cysteine concentration, although without statistical significance (P > 0.05) (Figure 3E).

Figure 3 Determination of cysteine levels and the protective effect of cysteine in the PD cell model. (A) Determination of cysteine levels in the SN of the MPTP-induced PD mice model. (B) Determination of cysteine levels in the liver of the MPTP-induced PD mice model. (C) Determination of cysteine levels in the blood of the MPTP-induced PD mice model. (D) Determination of cysteine levels in the MPP+-induced PD cell model. (E) Co-culture of MPP+ induced MN9D PD cell model with AML12 liver cells for cysteine level determination. (F) Determination of TH and α-syn levels in the MPP+-induced PD cell model following treatment with different concentrations of cysteine. (G) Statistical graph of TH protein. (H) Statistical graph of α-syn protein. Compared with the Saline or control groups, *P < 0.05, **P < 0.01, ***P < 0.001. Compared with the MPP+ group, #P < 0.05, ##P < 0.01. n = 3.

Protective Effect of Cysteine in the PD Cell Model

Tyrosine hydroxylase (TH) and α-syn are important indicators of PD. Compared to the control group, the MPP⁺ group exhibited decreased TH expression and increased α-syn expression. Different concentrations of cysteine increased TH expression and decreased α-syn protein expression compared to the MPP⁺ group (Figure 3F–H). Co-localization of TH and α-syn was also investigated, and the results revealed that the MPP⁺ group had decreased TH fluorescence intensity and increased α-syn fluorescence intensity. Compared to the MPP⁺ group, different concentrations of cysteine increased TH fluorescence intensity but decreased α-syn fluorescence intensity in a concentration-dependent manner (Supplementary Figure S2E–S2F).

Reduced Cysteine Concentration Protects Liver Cells

Cysteine, a natural component of glutathione, contains a reactive thiol group (-SH), which provides protective effects on thiol proteases and intoxicated hepatic parenchymal cells. Low concentrations of cysteine can protect liver cells. Compared to the control group, the MPP⁺ group (60 μM) exhibited increased levels of AST, ALT, proinflammatory factors (NF-κB p65, TNFα, IL-1β, IL-6), oxidative stress markers (Nrf2, HO-1, NQO1), CASP3, and ROS. After treatment with low concentrations of cysteine (30 μM), the levels of AST, ALT, inflammatory factors (NF-κB p65, TNFα, IL-1β, IL-6), CASP3, and ROS decreased compared to the MPP⁺ group, while the oxidative stress markers (Nrf2, HO-1, NQO1) increased (Supplementary Figure S3). Nuclear entry of NF-κB p65 and Nrf2, as well as CASP3 fluorescence intensity, were also measured. Compared to the control group, the MPP⁺ group exhibited enhanced fluorescence intensity of NF-κB p65, Nrf2, and CASP3, while after cysteine treatment, compared to the MPP⁺ group, the fluorescence intensities of NF-κB p65 and CASP3 decreased while that of Nrf2 increased (Supplementary Figure S4). These findings indicate that cysteine can activate the antioxidant pathway, protecting AML12 liver cells from oxidative damage caused by ROS produced during MPP⁺ metabolism.

High Levels of Cysteine Lead to Hepatocyte Damage

The extent of damage to AML12 cells following treatment with different concentrations of cysteine was subsequently assessed. Notably, as the concentration of cysteine increased, the damage to AML12 cells also increased. Compared to the control group, the ALT levels increased significantly when the cysteine concentration reached 50 μM, whereas the AST levels increased significantly at 100 μM (Figure 4A and B). Additionally, the expression of inflammatory markers (NF-κB p65, TNFα, IL-1β, IL-6) also increased with higher cysteine concentrations (Figure 4C–G). Nuclear entry of NF-κB p65 was observed, with fluorescence intensity increasing after treatment with 50 μM cysteine (Figure 4H and I). As cysteine concentration increased, the production of inflammatory factors and ROS was promoted, leading to cellular damage (Figure 4J and K). Increased levels of cysteine may promote the generation of nitrite in the body, thereby forming reactive nitrogen oxide species such as nitric oxide (NO). These reactive nitrogen oxide species can induce oxidative stress, leading to damage to cell membranes, exacerbation of inflammatory responses, and oxidative damage to intracellular proteins, nucleic acids, and other molecules, thereby affecting the normal function of cells.

Figure 4 High concentrations of cysteine lead to hepatocyte damage and inflammation. (A and B) Determination of ALT and AST levels in AML12 cells following treatment with different concentrations of cysteine. (C) Western blot images showing the protein expression of NF-κB p65, TNFα, IL-1β, and IL-6. (D–G) Statistical graphs showing the protein expression of NF-κB p65, TNFα, IL-1β, and IL-6. (H) Determination of NF-κB p65 protein immunofluorescence intensity. (I) Statistical graph showing the average immunofluorescence intensity of NF-κB p65 (Scale bar, 50 μm). (J) Determination of ROS levels using a green fluorescence probe (Scale bar, 50 μm). (K) Statistical graph showing the average fluorescence intensity of ROS. Compared with the control group, *P < 0.05, **P < 0.01, ***P < 0.001. n = 3.

Analysis of DEGs in Patients with PD or HBV

Differential Genes Obtained from the GEO Database

The expression matrices of the GSE20141 and GSE7621 datasets for PD were normalized using samples from the GPL570 platform (Supplementary Figure S5A). Principal component analysis (PCA) confirmed the repeatability of the data (Supplementary Figure S5B). Volcano plots were used to visualize DEGs, and we used an adjusted threshold of |logFC| > 0.58 and P.adj < 0.05, identifying a total of 2925 DEGs (Supplementary Figure S5C). Similarly, the expression matrix of the GSE83148 dataset for HBV was normalized using samples from the GPL570 platform (Supplementary Figure S6A). PCA and mean−variance trend verified the repeatability and lack of trend in the data (Supplementary Figure S6B and S6C). Volcano plots were used to visualize DEGs, and we used an adjusted threshold of |logFC| > 0.58 and P.adj < 0.05, identifying a total of 5204 DEGs (Supplementary Figure S6D).

GSEA and Immune Infiltration Analysis

GSEA of the PD datasets revealed that the DEGs were mainly involved in membrane trafficking, dopaminergic neurogenesis, cell cycle mitosis, mitochondrial protein import, protein folding, and autophagy (Figure 5A). In the HBV datasets, the DEGs were mainly involved in the regulation of proinflammatory and profibrotic mediators, chemokine receptors bind chemokines, metabolism of amino acids and derivatives, chemokine signaling pathway, toll-like receptor signaling pathway, and the IL8-CXCR2 pathway (Figure 5B). Heatmaps illustrating immune infiltration in the PD and HBV datasets are presented in Figure 5C and D. Compared to the normal controls, cytotoxic cells and Tcm cells were highly expressed in the PD model, whereas iDC, mast cells, and Th2 cells were lowly expressed (Figure 5E). In the HBV dataset, aDC, B cells, cytotoxic cells, iDC, Th1 cells, and Th2 cells were highly expressed, whereas eosinophils, neutrophils, Tcm, and TFH were lowly expressed (Figure 5F). There were no significant differences in immune cell expression between the PD or HBV datasets and normal controls (Supplementary Figure S7).

Figure 5 Immune infiltration analysis and GSEA of DEGs. (A) GSEA of the PD dataset. (B) GSEA of the HBV dataset. (C) Heatmap of immune infiltration analysis in the PD dataset. (D) Heatmap of immune infiltration analysis in the HBV dataset. (E) Boxplot of differentially expressed immune cells in the PD dataset. (F) Boxplot of differentially expressed immune cells in the HBV dataset. Compared with the normal group, *P < 0.05, ** P < 0.01, ***P < 0.001, ****P < 0.0001.

Biofunctional Enrichment Analysis

We identified 642 DEGs from the intersection of the PD and HBV datasets (Figure 6A). Pathway enrichment analysis of these 642 DEGs revealed involvement in several key pathways, including transcriptional misregulation in cancer, osteoclast differentiation, Chagas disease, leishmaniasis, Fc gamma R-mediated phagocytosis, legionellosis, pertussis, inflammatory bowel disease, hepatitis C, B cell receptor signaling pathway, amoebiasis, axon guidance, T cell receptor signaling pathway, and Th17 cell differentiation (Figure 6B). The enriched biological processes (BPs) mainly included axonogenesis, regulation of hemopoiesis, positive regulation of CD4-positive alpha-beta T cell activation, regulation of CD4-positive alpha-beta T cell differentiation, positive regulation of lymphocyte differentiation, dopamine biosynthetic process, and regulation of alpha-beta T cell differentiation (Figure 6C). The enriched cellular components (CC) included growth cone, site of polarized growth, actin-based cell projection, dendritic spine, neuron spine, distal axon, filopodium, transcription regulator complex, nuclear heterochromatin, and dendritic spine head (Figure 6D). The enriched molecular function (MF) included voltage-gated potassium channel activity involved in cardiac muscle cell action potential repolarization, microtubule binding, Wnt-protein binding, phosphatidylinositol-3,4-bisphosphate binding, p53 binding, tau-protein kinase activity, dynactin binding, voltage-gated potassium channel activity involved in ventricular cardiac muscle cell action potential repolarization, ATPase binding, and SH3 domain binding (Figure 6E).

Figure 6 Signaling pathway and biofunctional enrichment analysis. (A) Identification of 642 common genes from the overlapping of PD and HBV datasets. (B) KEGG enrichment analysis. (C–E) Enrichment of BP, CC, and MF.

WGCNA Identification of Hub Genes

To comply with the scale-free network distribution principle, determining a suitable value for the adjacency matrix weight parameter (termed “power”) was critical. By setting power values ranging from 1 to 30, corresponding correlation coefficients and average network connectivities were calculated. A higher correlation coefficient (up to 1) indicates closer adherence to scale-free distribution, though maintaining adequate gene connectivity was also essential—thus, the power value needed to balance sufficiently high correlation coefficients with substantial network connectivity. In this analysis, a power value of 8 was selected (Figure 7A). Using this parameter, a weighted co-expression network model was constructed, partitioning 642 genes into 4 modules. The grey module represented gene sets unassigned to any module and lacked analytical relevance (Figure 7B). Clustering heatmaps for all genes are shown in Figure 7C. Modules associated with various traits were identified using an absolute correlation coefficient threshold of ≥ 0.3 and P < 0.05 (Figure 7D). For the HBV dataset’s WGCNA, a power value of 30 was applied, dividing 233 genes into 3 modules. Again, the grey module denoted unclassified, non-relevant gene sets (Supplementary Figure S8). Supplementary Figure S9 lists the top 50 genes ranked within each module for both PD and HBV datasets.

Figure 7 WGCNA of the PD dataset GSE20141 and GSE7621. (A) Parameters for constructing the WGCNA network. (B) The upper part of the figure shows the gene clustering tree constructed based on the dissTOM matrix generated using weighted correlation, whereas the lower part shows the distribution of genes in each module. Genes in the same color belong to the same module. The colors obtained using the Dynamic Tree Cut method represent modules identified by this method. Some modules were merged due to correlations, indicated as “Merged Dynamic” at the bottom, and these merged modules were used for subsequent analysis. (C) Heatmap of all gene clusters. (D) Heatmap showing the correlation between trait modules.

PPI Analysis and Hub Gene Screening

We identified 30 DEGs through the intersection of WGCNA results from PD and HBV datasets (Supplementary Figure S10A). PPI enrichment analysis revealed 29 nodes and 31 edges, with an average node degree of 2.14 and an average local clustering coefficient of 0.456. The PPI enrichment P-value was <1.0×10⁻¹⁶ (Supplementary Figure S10B). Using the MCC algorithm in cytohubba, 10 hub genes were identified (Supplementary Figure S10C). These hub genes were found to be highly expressed in the SN and PBMC of patients with PD, as well as in the liver and PBMC of patients with HBV (Supplementary Figure S11), suggesting that these genes play important roles in the progression of PD and HBV and are involved in multifunctional systemic metabolic regulation that is closely related to disease progression. Their biological functions are listed in Table 2. Furthermore, the hub genes were validated using the PD PBMC GSE99039 dataset (P.adj < 0.05), resulting in 6 hub genes: NCF1 NCF4, SELPLG, SASH3, SYK, and ARHGAP30 (Supplementary Figure S10D).

Table 2 Biological Functions of Hub Genes

Single-Cell Type Analysis

Clustering analysis of gene cell types in the brain revealed that the genes were distributed across various clusters corresponding to excitatory neurons (c-0—c-11, c-13, c-16, c-18—c-20, c-22, c-24, c-27, c-29, c-31—c-34, c-36, c-39, c-40, c-44), oligodendrocytes (c-12, c-37), inhibitory neurons (c-14, c-15, c-17, c-21, c-23, c-25, c-26, c-28, c-30, c-38, c-41), astrocytes (c-35), oligodendrocyte precursor cells (c-42), and microglial cells (c-43). These genes are primarily expressed in microglial cells (Figure 8A). Studies have suggested that damage and overactivation of microglial cells near the risk areas within the basal ganglia and SN, trigger oxidative stress, mitochondrial autophagy and dysfunction, α-syn accumulation, and the release of pro-inflammatory cytokines. The neuronal damage mediated by these injurious factors further exacerbates microglial cell activation, further contributing to the pathological progression of PD.^35,36

Figure 8 Single-cell type analysis of hub genes in (A) brain and (B) liver.

In the liver, gene cell type clustering analysis revealed distribution across various clusters corresponding to hepatocytes (c-0, c-4, c-9, c-12, c-18), T cells (c-1, c-8, c-13, c-14), macrophages (c-2, c-15), plasma cells (c-3), endothelial cells (c-5), Kupffer cells (c-6, c-10), smooth muscle cells (c-7), fibroblasts (c-11), B cells (c-16), cholangiocytes (c-17), and erythroid cells (c-19) (Figure 8B). Notably, certain genes exhibit predominant expression in specific cell types: ARHGAP30, SELPLG, and SASH3 in T cells; CASP1 and NCF4 in macrophages; EVI2B and SYK in Kupffer cells; IRF8 and SELL in B cells; and NCF1 in plasma cells. These cell types play crucial roles in immune responses and antibody production within the liver, thereby maintaining its immune defense function.

Prognostic Analysis

ROC curves were used to evaluate the diagnostic value of 6 hub genes in PD and HBV. In the SN of patients with PD, the AUC values were determined as follows: ARHGAP30 (AUC = 0.719), CASP1 (AUC = 0.735 and 0.648), EVI2B (AUC = 0.692 and 0.852), IRF8 (AUC = 0.781 and 0.597), NCF1 (AUC = 0.735 and 0.673), NCF4 (AUC = 0.765 and 0.735), SASH3 (AUC = 0.749 and 0.796), SELL (AUC = 0.667 and 0.607), SELPLG (AUC = 0.749 and 0.745), and SYK (AUC = 0.713 and 0.602) (Figure 9A and B). In the liver of patients with HBV, the AUCs were ARHGAP30 (AUC = 0.895 and 1.000), CASP1 (AUC = 0.891 and 1.000), EVI2B (AUC = 0.836 and 1.000), IRF8 (AUC = 0.939 and 0.982), NCF1 (AUC = 0.939 and 1.000), NCF4 (AUC = 0.970 and 1.000), SASH3 (AUC = 0.992 and 1.000), SELL (AUC =0.906 and 0.971), SELPLG (AUC =0.964 and 1.000), and SYK (AUC =0.945 and 1.000) (Figure 9C and D). Furthermore, validation using PBMC datasets from patients with PD or HBV revealed that these genes had high diagnostic value (AUC > 0.483) (Figure 9E and F), indicating that these genes may serve as potential targets for the diagnosis of patients with PD or HBV.

Figure 9 Diagnostic value of hub genes. (A) PD brain SN datasets GSE20141 and GSE7621, from the GPL570 platform, including 26 patients with PD and 17 healthy controls. (B) PD brain SN datasets GSE20163 and GSE20164, from the GPL96 platform, including 14 patients with PD and 14 healthy controls. (C) HBV liver tissue dataset GSE83148, from the GPL570 platform, including 122 patients with HBV and 6 healthy controls. (D) HBV liver tissue dataset GSE38941, from the GPL570 platform, including 17 patients with HBV and 10 healthy controls. (E) PD PBMC dataset GSE99039, from the GPL570 platform, including 233 healthy controls and 205 patients with PD. (F) HBV PBMC dataset GSE58208, from the GPL570 platform, including 12 patients with HBV and 5 healthy controls.

Anti-PD Drugs

Through a search of the DrugBank database, we identified 34 drugs with potential therapeutic effects in PD. Their pharmacological actions and mechanisms are detailed in Supplementary Table S8. The therapeutic targets of these 34 drugs primarily include inflammatory factors (IL1α, IL1β, IL24, TNF, RELA, Fos, and secretary leukocyte protease inhibitor (SLPI)), chemokines (C-X-C Motif Chemokine Ligand 8 (CXCL8)), apoptosis-related factors (BAX, CASP3, CASP9, and BCL2), hepatic drug-metabolizing enzymes (cytochrome P450 family 1 subfamily A member 1 (CYP1A1) and CYP1A2), oxidative stress regulators (Nrf2 and catalase (CAT)), monoamine oxidases (monoamine oxidase A (MAOA) and MAOB), neurotrophic factors, dopamine receptors D1 (DRD1) and DRD2), transporter protein families (solute carrier family 22 member 2 (SLC22A2) and SLC7A11), prolactin (PRL), potassium voltage-gated channel subfamily H member 2 (KCNH2), catechol-O-methyltransferase (COMT), prostaglandin-endoperoxide synthase 2 (PTGS2), α-syn, mitogen-activated protein kinase (MAPK), and bone morphogenetic protein 6 (BMP6) (Figure 10). Additionally, it was found that tolcapone, one of the 34 drugs, could induce liver injury (Supplementary Table S9), prompting further experimental validation regarding whether tolcapone indeed causes liver damage.

Figure 10 The drug-target network. Red represents anti-PD drugs, and blue represents drug-related targets.

Molecular Docking

The binding poses and interactions of tolcapone with hub genes were obtained using Autodock Vina v.1.2.2, which generated binding energies for each interaction. The results showed that tolcapone binds to hub genes via hydrogen bonds and strong electrostatic interactions. The binding energies of tolcapone with NCF1, NCF4, SELPLG, SASH3, SYK, and ARHGAP30 are −8.0 kcal/mol, −7.6 kcal/mol, −5.9 kcal/mol, −6.8 kcal/mol, −9.3 kcal/mol, and −6.0 kcal/mol, respectively (Figure 11). The binding energies and corresponding amino acid sites are detailed in Table 3.

Table 3 Binding Energies and Corresponding Amino Acid Sites

Figure 11 Molecular docking of tolcapone with various hub genes (A) NCF1 (AF-P14598-F1), (B) NCF4 (AF-Q15080-F1), (C) SELPLG (AF-Q14242-F1), (D) SASH3 (AF-O75995-F1), (E) SYK (AF-Q15046-F1), and (F) ARHGAP30 (AF-A0A0A0MRJ8-F1).

Effect of Tolcapone on Hub Genes and Its Impact on Liver Damage

First, the effects of tolcapone on the hub genes were assessed. Compared to the control group, in the MPP⁺-induced PD cell model, the expression levels of NCF1, SELPLG, SASH3, SYK, and ARHGAP30 were elevated. Treatment with Tolcapone resulted in a concentration-dependent decrease in the expression of these proteins (Figure 12A–F). Additionally, tolcapone led to increased levels of ALT and AST in AML12 liver cells in a concentration-dependent manner (Figure 12G and H), indicating tolcapone-induced liver injury. Compared to the control group, MPP⁺ treatment resulted in elevated concentrations of AST and ALT. Upon administration of tolcapone at a concentration of 100 μM, significant increases in AST and ALT concentrations were observed compared to the MPP⁺ group (Figure 12I and J). These findings suggest that both the PD modeling agent and tolcapone can induce liver injury, warranting further validation in a PD mouse model.

Figure 12 Effect of tolcapone on hub genes and hepatocellular damage. (A) Determination of NCF1, SELPLG, SASH3, SYK, and ARHGAP30 protein expression levels in MPP+-induced PD cell model treated with different concentrations of Tolcapone. (B–F) Statistical graphs showing the expression levels of NCF1, SELPLG, SASH3, SYK, and ARHGAP30 proteins. (G and H) Determination of ALT and AST concentrations in AML12 liver cells treated with different concentrations of tolcapone. (I and J) Determination of AST and ALT concentrations in MPP+-induced AML12 liver cells treated with different concentrations of tolcapone. Compared with the control group, *P < 0.05, **P < 0.01, ***P < 0.001. Compared with the MPP+ group, #P < 0.05, ##P < 0.01, ###P < 0.001. n = 3.

Compared to the saline group, MPTP led to a significant decrease in TH levels (Figure 13A and B). MPTP treatment also resulted in a significant decrease in liver weight (Figure 13C and D). The liver histology of mice in the saline group revealed intact lobular structure, clear outlines of hepatic parenchymal cells, evenly distributed nuclei, and an orderly arrangement of liver cells. In the MPTP group, extensive diffuse degeneration and necrosis were observed near the hepatic portal vein, with disordered arrangement of liver cells at the edges, most of which were in a swollen state, along with inflammatory cell infiltration (Figure 13E). Masson staining revealed collagen fiber deposition in the hepatic portal area in the saline group. Compared to the saline group, pseudo-lobules were observed in the hepatic portal and central vein areas of mice in the MPTP group, with the formation of fibrous septa and an increase in collagen fiber deposition (Figure 13F). Immunohistochemistry revealed decreased autophagy and increased apoptosis in the MPTP group compared to the saline group, with decreased levels of LC3 high positive and low positive and increased levels of Negative; increased levels of CASP3 high Positive and Low Positive and decreased level of Negative (Figure 13G–J, Table 4). Compared to the saline group, the MPTP group exhibited an increase in ALT and AST concentrations (Figure 13K and L). After co-culturing MN9D and AML12 cells, the ALT and AST concentrations increased in the MPP⁺-induced PD cell model of AML12 liver cells compared to the control group (Figure 13M and N).

Table 4 Proportion of LC3A/B and CASP3 Positive Degree

Figure 13 Liver damage in the MPTP-induced PD mice model. (A) Determination of TH content in the MPTP-induced PD mice model. (B) Statistical graph of TH cell count. (C) Anatomy of the liver in the MPTP-induced PD mice model. (D) Statistical graph of liver weight. (E) Analysis of liver HE staining in the MPTP-induced PD mice model. (F) Analysis of liver Masson staining in the MPTP-induced PD mice model. (G) Immunohistochemical staining of LC3A/B in liver. (H) Statistical graph of LC3A/B-positive ratio. (I) Immunohistochemical staining of CASP3 in liver. (J) Statistical graph of CASP3-positive ratio. (K and L) Determination of ALT and AST levels in the blood of MPTP-induced PD mice model. (M and N) Co-culture of MPP+-induced MN9D PD cell model with AML12 liver cells to determine ALT and AST levels. Compared with the Saline or control group, *P < 0.05, **P < 0.01, ***P < 0.001. n=3.

Discussion

Through MR analysis, we found that PD does not have a significant causal impact on most liver diseases. However, a negative association was found with chronic hepatitis (finn-b-K11_CHRONHEP, finn-b-K11_DISLIVOTH). We speculate that the influence of PD on the liver is dynamic and complex, and the current data is still incomplete. Nevertheless, experimental validation has confirmed some effects of PD on the liver. Further analysis of serum metabolites revealed a positive causal effect between PD and cysteine, a serum metabolite associated with chronic hepatitis (finn-b-K11_CHRONHEP). This finding suggests that PD may influence the development of chronic hepatitis by modulating cysteine levels. The experimental study further investigates the role of cysteine in PD cell models. Specifically, moderate levels of cysteine have protective antioxidant effects on liver cells, while high concentrations of cysteine may lead to increased oxidative stress and damage to liver cells. Therefore, controlling the concentration of cysteine is crucial for maintaining the health and function of liver cells. Additionally, comprehensive bioinformatics analysis revealed the biological functions and single-cell localization information of common genes in PD and HBV diseases. Furthermore, the study confirmed that the use of tolcapone in PD treatment can exacerbate liver damage.

The liver is the largest digestive organ in the human body and has a strong ability to metabolize foreign substances. It is the main organ responsible for maintaining systemic metabolic homeostasis.³⁷ Liver injury is often accompanied by the release of certain metabolites related to the occurrence of neurodegenerative disorders.³⁸ Currently, there is increasing evidence to suggest that PD is not only a neurodegenerative disorder but also a systemic disease, implying that the pathological changes in PD also affect peripheral organs outside the brain. Therefore, studying other peripheral tissues may provide new insights into the pathogenesis of PD. There is growing clinical evidence suggesting a connection between the CNS and liver metabolism, especially in patients with PD.^39,40 Controlling further liver damage in patients with PD is of great significance for maintaining their systemic homeostasis and avoiding potential risks associated with medication. This study demonstrates that PD can affect liver function through serum metabolites. Specifically, disruptions in cysteine metabolism were detected in the SN, liver, and blood of the PD animal model, indicating the importance of cysteine in the pathogenesis of PD. In the MPTP-induced PD mice model, increased levels of cysteine were observed in the SN, liver, and blood of the mice. Decreased cysteine concentration in the MPP⁺-induced PD cell model led to reduced α-syn aggregation and increased TH expression, indicating decreased cellular damage. Thus, at low concentrations, cysteine’s anti-inflammatory, anti-apoptotic, and antioxidant effects protect liver cells from oxidative damage. However, high cysteine concentrations promoted inflammation and oxidative stress, resulting in excessive ROS production and liver cell damage. Therefore, cysteine levels could serve as biomarkers for detecting liver damage in patients with PD and as targets for future therapeutic interventions.

Furthermore, shared hub genes between PD and HBV were identified, and several studies have reported the functions of these genes. For example, the NCF1 and NCF4 genes play an important role in immune responses. They are involved in the respiratory burst generation during neutrophil activation, providing cells with redox capabilities to kill bacteria and regulate immune activity during inflammation.⁴¹ Variations in these genes may influence neutrophil function and inflammatory responses in both PD and HBV infections. SELPLG plays an important role in inflammatory responses and the adhesion of immune cells. SELPLG binds to P-selectin on the surface of endothelial cells, facilitating the migration and infiltration of leukocytes and immune cells.⁴² In PD and HBV, variations in the SELPLG gene may regulate leukocyte adhesion and the inflammatory response. The protein encoded by the SASH3 gene contains sterile alpha motif (SAM) and Src homology 3 (SH3) domains, which play important roles in signal transduction and protein interactions. The SAM domain may be involved in protein interactions and signal transduction, whereas the SH3 domain may mediate the binding of the protein to other domains.⁴³ In PD and HBV, SASH3 may regulate inflammation and immune responses by participating in signal transduction pathways. The protein encoded by the SYK gene is a tyrosine kinase that is involved in signal transduction and immune response in B cells and macrophages. SYK plays a critical role in the signaling process, regulating the activation of immune cells and inflammatory responses.⁴⁴ In PD and HBV, variations in the SYK gene may affect the activation of immune cells and inflammatory response. The ARHGAP30 gene is involved in the regulation of the cytoskeleton and cell migration.⁴⁵ ARHGAP30 may affect the pathogenesis of PD and HBV by regulating cell migration and inflammatory response. In summary, the shared genes between PD and HBV may impact the function of neutrophils and inflammatory responses, regulate leukocyte adhesion and inflammatory responses, as well as participate in signal transduction and immune responses.^46–48 The innovation of our research lies in identifying potential genes that could be used as therapeutic targets.

This study demonstrated that while drugs effective for treating PD show promise, they may also target pathways leading to liver damage. Previous studies have demonstrated that tolcapone can cause mitochondrial uncoupling via oxidative phosphorylation, resulting in damage to liver cell energy production. In Phase III clinical trials for PD treatment, 1% of patients taking 100 mg of tolcapone three times daily had serum ALT levels increase above three times the upper limit of normal. This percentage increased to 3% among those taking 200 mg three times daily.^49,50 These observations confirm the hepatotoxic potential of the drug. In our study, treating PD cell models with tolcapone resulted in decreased expression levels of NCF1, NCF4, SELPLG, SASH3, SYK, and ARHGAP30, along with other classical treatment indicators, compared to MPP⁺ treatment. However, as the concentration of tolcapone increased, the levels of AST and ALT also increased. In conclusion, our results indicate that tolcapone treatment potentially regulates the expression of PD-related genes. However, it is important to note that tolcapone treatment can also cause liver damage. Therefore, when using tolcapone or similar drugs for PD treatment, close monitoring of patients’ liver function indicators should be considered, and appropriate treatment and management measures should be taken to balance the therapeutic effects with the adverse effects.^51,52 In neurotoxin-induced models of PD, neurotoxins such as 6-hydroxydopamine (6-OHDA), MPTP, MPP⁺, paraquat, and rotenone may directly damage the liver through mechanisms such as oxidative stress, inflammation, and mitochondrial dysfunction.^53–55 We also observed an increase in AST and ALT levels in the MPP⁺-induced liver cells. While lower concentrations of tolcapone were found to decrease AST and ALT levels, higher concentrations led to increased levels of these enzymes. Similarly, in MPTP-induced PD animal models, an increase in AST and ALT levels was observed, along with reduced liver autophagy and increased apoptosis.

While our study demonstrates a causal link between PD and liver dysfunction mediated by cysteine and shared immune pathways, alternative explanations warrant consideration. First, pharmacological interventions for PD (eg, levodopa) may independently alter liver metabolism, though our experimental models controlled for this via MPTP/MPP⁺ toxicity. Second, lifestyle factors (eg, diet, alcohol) and comorbidities (eg, metabolic syndrome) could confound clinical associations; future MR analyses should incorporate these covariates where feasible. Lastly, gene-environment interactions (eg, toxin exposure) may modulate the PD-liver axis, emphasizing the need for integrated omics and environmental data in subsequent work. Prospective studies should track liver function in PD patients stratified by medication use, dietary habits, and genetic risk profiles to disentangle confounding effects. Additionally, organoid models co-culturing neuronal and hepatic cells could elucidate bidirectional brain-liver communication pathways.

The identification of cysteine as a biomarker and hub genes (eg, NCF1) as therapeutic targets opens avenues for personalized medicine in PD management. The integration of MR, transcriptomics, and experimental validation mitigates biases inherent to any single approach. Although this study proposes several new suggestions, it still has some limitations. The research data is mainly based on cell models and animal experiments and further validation is needed in humans. More in-depth exploration is needed to investigate the mechanisms and interrelationships between PD and liver metabolism. More in-depth research and clinical observations are needed to understand the mechanisms of liver damage and the clinical impact of tolcapone treatment. Future treatment strategies need to focus on maximizing the therapeutic effectiveness of PD medications while reducing their hepatotoxic effects. More biological and bioinformatics analyses should be conducted to reveal the deeper connections between PD and liver metabolic homeostasis. Prospective clinical cohorts are needed to validate the temporal dynamics of PD-associated liver damage and assess causality beyond genetic proxies. These limitations and future research plans will contribute to further improving the understanding of the relationship between PD and liver diseases and provide important guidance and evidence for future treatments and clinical practices.

Conclusion

Research shows that PD-related brain damage does indeed have an impact on liver metabolic function. Specifically, PD-related brain damage may disrupt the concentration of homocysteine, and high levels of homocysteine may lead to liver cell damage and oxidative stress. By comparing the gene expression differences in the transcriptomes of PD and liver injury tissues, some common genes or functions that may interact across tissues were identified. For example, specific genes involved in immune and inflammatory responses (such as NCF1, NCF, and SELPLG) may lead to liver injury. The study suggests that homocysteine can serve as a biomarker for PD-related liver injury and a therapeutic target, providing important insights for future treatment strategies for PD. The study also evaluated the impact of the anti-PD drug tolcapone on liver function. It was found that besides having significant therapeutic.