PD patients were screened for eligibility at the Neurosurgery Department of the First Affiliated Hospital of Soochow University between January 2022 and October 2022. These idiopathic PD patients without dementia [17] were enrolled according to the Movement Disorders Society criteria [18]. Participants all underwent subthalamic nucleus deep brain stimulation (STN-DBS) electrode implantation and micro recording was performed intraoperatively. After surgery, these patients were followed up for 1 year. During follow-up, we performed DBS programming for all participants based on the patient’s condition, at least once every 2 weeks. Unified Parkinson’s Disease Rating Scale (UPDRS) scores, Hoehn and Yahr (H&Y) scores, levodopa equivalent daily dose (LEDD), and clinical phenotypes [19] were evaluated at the medication-off state at baseline and after follow-up. All evaluations were conducted by two professional technicians blinded to the grouping.

According to the criteria of DBS response that defined the assessment of significant improvement as a 50% or greater decrease in UPDRS score at the end of follow-up, participants were assigned to the non-responder group or the responder group by two researchers. To analyze global protein expression in the non-responders and responders, we collected serum from these patients and used LC-MS/MS to profile DEPs. The upregulated and downregulated DEPs (the candidate DEPs) were then verified by enzyme-linked immunosorbent assay (ELISA) in a new set of samples, including 21 PD patients (10 Non-responders and 11 Responders) having undergone DBS.

This study was authorized by the Ethics Committee of the First Affiliated Hospital of Soochow University (Ethics Committee approval code: 2022-422). Informed consent was obtained from all patients.

Eleven serum samples were used for proteomic analysis. First, we employed the Pierce Top 14 Abundant Protein Depletion Spin Column kit (Thermo Fisher Scientific, Waltham, MA, USA) to remove high-abundance proteins from each sample. After collecting the supernatant, we determined protein concentrations using the Pierce BCA Protein Assay kit (Thermo Fisher Scientific, USA). Protein solutions were digested using 100 mM TEAB buffer, followed by addition of trypsin at 1:100 trypsin-to-protein mass. The peptides were recovered by centrifugation at 12,000 g for 10 min at room temperature, and the peptides were recovered with ultrapure water once, and then the two peptide solutions were combined [20].

The peptides were added to solvent A, and directly onto a reversed-phase analysis column. On the EASY-nLC 1200 UPLC system (Thermo Fisher Scientific, USA), the gradient with solvent B was increased from 4% to 20% for 68 minutes, from 20% to 32% for 14 minutes, up to 80% for 4 minutes, and then remained at 80% for the last 4 minutes. The peptides were placed in a nanoliter electrospray ionization (NSI) source and then using the Orbitrap Exploris 480 Mass Spectrometer (Thermo Fisher Scientific, USA). With the m/z scan range from 400 to 1200, intact peptides were found in the Orbitrap at 60,000 resolution. Subsequently, we selected peptides for analysis and detected the fragments in the Orbitrap at 30,000 resolution [21].

The LC-MS/MS data were processed using the Proteome Discoverer search engine (v2.4.1.15) (Thermo Fisher, USA) [21]. Mass spectrometry data were collected from the FASTA database (homo_sapiens 9606_PR_20201214; 75,777 entries; (https://www.uniprot.org/proteomes/UP000005640) concatenated with the reverse decoy database. Trypsin (full) was used as the cleavage enzyme that allows up to two missing cleavages. The mass value of the precursor ion was 10 ppm and that of the fragment ion was 0.02 Da. The fixed modification was carbamidomethyl (Cys), while the variable modifications included protein N-terminal acetylation, oxidation (Met), Met-loss (Met), and Met-loss + acetyl (Met). The false discovery rate for proteins and peptides was set at 1% and the minimum peptide length was set at 6. Default values were used for all other parameters.

The characteristics of the participants who underwent proteomic analysis are shown in Table 1. Before DBS surgery, no obvious differences were found in sex, age, disease duration, H&Y scale score, UPDRS score, clinical phenotype, or LEDD in the medication-off state. At the end of the follow-up, the UPDRS and LEDD scores were significantly reduced in the responder group (p 0.05).

The UniProt-GOA database (http://www.ebi.ac.uk/GOA/) was used for gene ontology (GO) annotation, and eggnog-mapper software (http://eggnog5.embl.de/) was used to obtain the GO functions of proteins. We divided proteins into three categories. Statistical significance was set at p 0.05 for GO annotation and enriched pathway analysis. Interacting proteins belonging to the search dataset only were selected and non-interacting candidate proteins were removed. Hierarchical clustering was then performed according to the functional classification of DEPs analyzed using GO annotation. Finally, we analyzed category enrichment and selected those categories enriched in more than one cluster with a p-value 0.05.

For the validation study, three proteins were selected as being highly differentially expressed (absolute fold change $\ge$1.5) from the top five upregulated and downregulated DEPs, including nidogen-1 (NID1), creatine kinase M-type (CKM), and bleomycin hydrolase (BLMH). ELISA was used to detect NID1 (EHNIDI, Thermo Fisher Scientific, USA), CKM (ab264617; Abcam, Cambridge, UK), and BLMH (EH46RB; Thermo Fisher Scientific, USA) in serum samples from another 21 PD patients (10 non-responders and 11 responders). Procedures were performed according to the manufacturer’s instructions. In brief, diluted serum samples were placed into 96-well plates, in duplicate wells, with Biotinylated Detection Ab Working Solution. Blank, negative, and positive control wells were also used. After incubation for 2.5 hours at 37 °C in the microplate shaker, the plates were washed. Tetramethylbenzidine substrate solution was then added to mediate color development and the plates were incubated at 37 °C for 10–20 minutes. Subsequently, we used a DR-200Bn spectrophotometer (Wuxi Huawei Delang Instrument Co., Ltd., Wuxi, Jiangsu, China) to detect the optical absorbance of each sample 450 nm. Finally, the concentration of each analyte was analyzed.

IBM SPSS software (version 24.0; IBM Corp., Armonk, NY, USA) was used for statistical analysis. The comparisons of variables between non-responders and responders were analyzed using two-sample t-tests or Mann–Whitney U tests for continuous data, and chi-squared tests for categorical data. Furthermore, we used logistic regression and receiver operating characteristic (ROC) curve analyses to determine the diagnostic performance of candidate proteins. Statistical significance was set at p 0.05.

Eighteen PD patients were screened for eligibility. In accordance with the inclusion and exclusion criteria, three patients with dementia were excluded and two patients declined to participate in this study. Finally, 13 participants were enrolled. These patients underwent STN-DBS electrode implantation and were then followed up for 1 year after DBS surgery. Three patients were subsequently lost during follow-up, and 10 PD patients (four non-responders and six responders) were enrolled for proteomics analysis. The study workflow is shown in Fig. 1.

Fig. 1.

Study workflow. LC‒MS/MS, liquid chromatography-tandem mass spectrometry.

First, we evaluated mass spectrometry-based proteomics data based on multifaceted quality control indexes. Proteins above 10 kDa were evenly distributed without significant molecular weight drift (Fig. 2A). The proportion of proteins with more than two peptides was 72% (Fig. 2B). Most of the peptides were between 10 and 30 amino acids in length (Fig. 2C). Most proteins had a coverage of less than 20% (Fig. 2D). In total, we detected 601,818 spectra, 7875 unique peptides, 8741 peptides, 130,117 matched spectrums, 1290 proteins, and we quantified 1157 proteins. Compared with the non-responder group, there were 10 downregulated and four upregulated proteins in the responder group (Fig. 2E). Significant differences were defined as a $\ge$1.5 fold change combined with a p-value 0.05.

Fig. 2.

Mass spectrometry data for quality control and proteomic mapping of DEPs. (A) Identified protein mass distribution. (B) Number of peptides per protein distribution. (C) Distribution of identified peptide lengths. (D) Protein sequence coverage distribution. (E) Number of total spectrums, matched spectrums, identified peptides, unique peptides, identified proteins, and quantifiable proteins. DEPs, differentially expressed proteins.

Fourteen DEPs between responders and non-responders are illustrated in the volcano map in Fig. 3A, and DEP information is shown in Table 2. The DEPs were mainly located in the extracellular space (57%), cytoplasm (22%), and plasma membrane (7%) (Fig. 3B). GO analyses of the DEPs revealed that the top five biological processes were response to stimulus process (eight DEPs), cellular process (12 DEPs), metabolic process (seven DEPs) multicellular organismal process (eight DEPs), and biological regulation (eight DEPs) (Fig. 3C green). The top three cellular components were cell (12 DEPs), intracellular (11 DEPs), and protein-containing region (three DEPs) (Fig. 3C orange). The top three molecular functions were catalytic activity (four DEPs), binding (eight DEPs) and molecular function regulator (two DEPs) (Fig. 3C purple). DEPs were also categorized according to functionality and homology (Fig. 3D). The top category was cellular processes and signaling, which included six DEPs (Fig. 3D). The second category was metabolism, which included three DEPs (Fig. 3D). Furthermore, the data revealed that cellular processes and metabolism of biological functions might be associated the response and clinical outcomes of DBS. Three DEPs (CKM, BLMH, and NID1) were closely associated with cellular processes and metabolism.

Fig. 3.

Protein function annotation and category analysis of differentially expressed proteins. (A) The horizontal axis and vertical axis of the volcano plot shows the logarithmically converted value of the relative quantitative value of the protein and the logarithmically converted p-value after the log-log conversion, respectively. In the volcano plot, the red dot indicates significantly upregulated proteins and the blue dot indicates significantly downregulated proteins. (B) Sub cellular localization and classification of differentially expressed proteins. (C) The secondary classification of differentially expressed proteins under each Gene Ontology category. (D) Clusters of Orthologous Groups of protein/EuKaryotic Orthologous Groups (COG/KOG) category analysis of the up and downregulated differentially expressed proteins.

The results of the enrichment analysis of protein GOs and hierarchical analysis of enrichment-based clustering of GOs are illustrated in Fig. 4A. Enriched DEPs were classified into four subgroups, labeled Q1–Q4. Next, the four subgroups were hierarchically clustered using a heatmap, as illustrated in Fig. 4B. Based on the 14 DEPs between the non-responder group and responder group, ontological functions were explored using UniProt-GOA, and were mainly clustered into 22 GO functional categories. The most significantly enriched biological processes of the DEPs were cellular amino acid metabolic process (CKM and BLMH), cellular protein modification process (BLMH, CKM, TF, MAN2B1, PSMA7, MINPP1, and FLG), protein modification process (BLMH, TF, MAN2B1, PSMA7, MINPP1, and FLG), and ion homeostasis (SPINK1, TF, and MYOC). For biological processes, the major functional categories of downregulated proteins were the carboxylic acid metabolic processes (CKM, BLMH), positive regulation of cell adhesion (MYOC and HLA-DPA1), alpha-amino acid metabolic process (CKM, BLMH), and cellular amino acid metabolic process (CKM, BLMH) in Q1 (Fig. 4B). Thus, the biological functions of the DEPs were related to metabolic and protein modification processes.

Fig. 4.

Enrichment analysis and enrichment-based clustering analysis of GO. (A) Enrichment analysis of protein GOs. (B) Hierarchical analysis of enrichment-based clustering of GOs. GO, gene ontology.

Based on the results of bioinformatics analysis, three DEPs were closely related to the therapeutic responses to DBS in PD patients, including CKM, BLMH, and NID1. In the validation study, ELISA was used to verify candidate DEPs in additional patient samples. The general characteristics of patients and serum levels of the candidate DEPs are shown in Table 3. As illustrated in Fig. 5, compared with the non-responders, CKM and BLMH levels were decreased in the responder group (Fig. 5A,B). However, no differences were found in the serum levels of NID1 (Fig. 5C). Additionally, binary logistic regression and ROC analyses indicated that CKM and BLMH could be predictive biomarkers for therapeutic responses to DBS, with p-values of 0.014 and 0.017 and areas under the curve of 0.818 (95% confidence interval (CI) 0.634–1.000) (Fig. 5D) and 0.809 (95% CI 0.605–1.000) (Fig. 5E). When CKM and BLMH were included in the modelling analysis, CKM and BLMH contributed prominently, with an area under the curve of 0.873 (95% CI 0.721–1.000) (Fig. 5F).

Fig. 5.

Changes in the serum levels of candidate DEPs in each group. (A) Levels of CKM in non-responders and responders. (B) Levels of BLMH in non-responders and responders. (C) Levels of NID1 in non-responders and responders. The performance of baseline degree centrality of CKM (D), BLMH (E), and the combined value of CKM and BLMH (F) in distinguishing responders from non-responders, using receiver operating characteristic (ROC) curve analysis. *p 0.05; **p 0.01; NS (Not Significant), p $>$ 0.05. AUC, area under the curve.