This study was approved by the Ethics Committee of Shanghai University of Sport (No. 102772019RT044) and conducted in accordance with the Helsinki Declaration of 1975. All participant information remained confidential and all patients provided written informed consent. Twelve individuals with METH dependence were recruited from compulsory detoxification programs in Zhejiang province, China. Twelve control subjects were recruited from the community within Changhai Street, Yangpu District, Shanghai. These participants were selected according to the following criteria: (1) men aged 18–45 years; (2) with long-term use of METH (2 or more years); (3) confirmed to meet the criteria for methamphetamine dependence as outlined in the DSM-V (Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition) based on the Structured Diagnostic Interview [18]; (4) without a serious infectious disease, autoimmune disease, severe mental illness, neurological disease, Human Immunodeficiency Virus (HIV) infection, obvious disease that would affect organ quality, hypertension, hyperlipidemia, diabetes, tumor, or other disease; (5) without use of nonamphetamine-type drugs; and (6) without consumption of pharmaceutical drugs, cocaine-like substances, alcohol, or tobacco within 2 weeks of serum sampling. The METH-dependent group had a mean age of 29.83 years (standard deviation [SD] = 5.80 years), an average duration of drug use spanning 4.18 years (SD = 1.40 years), a typical weekly drug usage frequency of 2.21 times (SD = 1.94 times), and a mean quantity of 0.265 g per drug consumption event (SD = 0.11 g). Participants in the control group were 12 healthy men (mean age 32.58 years, SD = 6.32 years) who met the same criteria as the METH-dependent group, except without history of drug misuse. The 12 control participants were randomly assigned to one of three subgroups (A, B, or C), with four participants in each subgroup. Similarly, the 12 participants with METH dependence were randomly assigned to one of three subgroups (D, E, or F), with four participants in each subgroup.

Fasting venous blood samples (5 mL) were drawn from the 24 participants and clotted at 4 °C for 30 min. The samples were then centrifuged at 1000 $\times$g for 10 min, then the supernatants were collected. The samples were subgrouped as described above to create a total of six serum samples (A, B, C, D, E, and F). Albumin and Immunoglobulin G (IgG) Depletion SpinTrap columns (No. 28-9480-20, GE Healthcare, Buckinghamshire, UK) were used, and the manufacturer’s instructions were followed to remove the high-abundance proteins from the serum. The total protein concentration was then determined using a Pierce™ Coomassie (Bradford) Protein Assay Kit (Cat. No. 23200, Thermo Fisher Scientific, Rockford, IL, USA).

Sodium dodecyl sulfate–polyacrylamide gel electrophoresis was used to separate the proteins in the serum sample (40 µg/sample). The gels were cut and the pieces were washed twice with double-distilled water. Silver dye decolorizer (equal volumes of 100 mmol/L sodium thiosulfate (Thermo Fisher Scientific, Fair Lawn, NJ, USA) and 30 mmol/L potassium ferricyanide (Thermo Fisher Scientific, Fair Lawn, NJ, USA)) was used to destain the gel, which was then washed twice with water. Acetonitrile was used to dehydrate the gel fractions. The reducing reagent (dithiothreitol, I1149-25G, 10 mmol/L, SIGMA, Saint Louis, MO, USA) was allowed to react for 1 h at 37 °C with the gel fractions, which were then dehydrated with acetonitrile. A cysteine blocking reagent (iodoacetamide, 25 mmol/L, Thermo Fisher Scientific, Fair Lawn, NJ, USA) was added and reacted with the gel fractions for 30 min at room temperature in the dark. The fractions were washed twice and dehydrated. The gel pieces were then digested using trypsin (V5113, Promega, Madison, WI, USA) overnight at 37 °C. An acetonitrile (50%) and trifluoroacetic acid (0.1%) solution (400 µL) was added and incubated at 37 °C for 30 min. This step was repeated twice. Finally, the eluates from each respective sample were combined, and the peptides were desalted and freeze dried.

The peptides were analyzed using an Orbitrap Fusion Lumos Tribrid (Thermo Fisher Scientific, San Jose, CA, USA) mass spectrometer using a chromatographic column (ChromXP, C18, 3 µm, 120 Å; catalog No. 805-00120, SCIEX, Framingham, MA, USA). A binary solvent system composed of water containing buffer A (0.1% formic acid and 2% acetonitrile in water) and buffer B (98% acetonitrile and 0.1% formic acid) was used for liquid chromatography. The peptides were separated by a gradient of buffer B (5% B, 0–5 min; linearly increasing concentrations of 5%–40% B within 100 min; increased to 80% B in 1 min and washed for 5 min; decreased to 5% B in 1 min and isocratic at 5% B for 13 min). The solvent system flow rate was 300 nL/min. The mass spectrometer was operated in the scan range of 300–1400 m/z, with a spray voltage at 2500 V and a dynamic exclusion time of 12 s. All data were acquired using “high-low” mode, in which during a maximum 3-s cycle time, the most abundant multiply-charged parent ions were selected with high resolution (120,000 at m/z 200) from the full scan for higher-energy collisional dissociation fragmentation. Precursor ions with singly charged and charge states over 7 were excluded. The target value for the second mass spectrometry (MS2) spectra was set at 5000 (the automated gain control was enabled with a maximum injection time of 35 ms), the isolation window was set to 1.6 m/z, and 35% was used for the normalized collision energy. Data were acquired using Xcalibur software 4.1 (Thermo Fisher Scientific, Waltham, MA, USA) [19].

All mass-spectrometric data were analyzed using MaxQuant 1.5.2.8 (University of Halle-Wittenberg, Halle, Germany; https://www.maxquant.org) [20] against the human UniProt FASTA database (https://www.uniprot.org/proteomes/UP000005640). Carbamidomethyl cysteine was searched as a fixed modification, and oxidized methionine and protein N-terminal acetylation as variable modifications. Enzyme specificity was set to trypsin/P. Two missing cleavage sites were allowed. For mass spectrometry and tandem mass spectrometry, the tolerance of the main search for peptides was set at 7 ppm and 20 ppm, respectively. The peptide, protein, and site false discovery rates were fixed at a significance level not greater than 0.01. Label-free protein quantitation was performed with a minimum ratio count of 2.

Quantitative data for the detected proteins were analyzed using MetaboAnalyst software 5.0 (McGill University, Montreal, Canada; https://www.metaboanalyst.ca) [21]. Proteins were removed from further analysis if the detection values were missing in more than 50% of the samples from either the METH-dependent or control group or if the quantitative values were consistent across all samples.

For proteins with missing quantitative values, missing values were imputed as one-half the minimum positive value in the original data using the minimum fill method. To obtain normalized protein abundance data for subsequent analyses, the summed intensity values for all proteins measured in each sample were calculated, and individual protein intensities within a sample were divided by the sample’s total intensity value. Hierarchical clustering analysis and visualization of the acquired protein profiles for all samples and genes were performed using Multiple Array Viewer software 4.9.0 (Agilent Technologies, Santa Clara, CA, USA; https://webmev.tm4.org) [22]. Principal component analysis of the proteins was carried out using MATLAB software R2019b (MathWorks, Natick, MA, USA; https://www.mathworks.com) [23]. The orthogonal partial least squares discriminant analysis was performed using SIMCA software 14.1 (Umetrics, Umeå, Sweden; https://www.sartorius.com/en/products/process-analytical-technology/data-analytics-software/mvda-software/simca) [24].

To identify proteins with significantly different expression between METH-dependent and control groups, Statistical Package for the Social Sciences (SPSS, 24.0) (International Business Machines Corporation, Armonk, NY, USA; https://www.ibm.com/products/spss-statistics) software [25] was utilized. A threshold filter requiring at least a two-fold change in the protein ratio between groups and a p-value 0.05 was applied. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were conducted using the Database for Annotation, Visualization and Integrated Discovery. A general pathway and network analysis was performed with Ingenuity Pathway Analysis (IPA) software (QIAGEN, Hilden, Germany; https://digitalinsights.qiagen.com/IPA) [26].

A total of 459 proteins were identified by proteomic workflow, and 308 proteins with quantitative information were finally obtained using MetaboAnalyst software analysis (Supplementary Table 1). We used the Multiple Array Viewer software to perform hierarchical clustering analysis at both the sample- and gene-levels of those 308 proteins (Fig. 1). The clustering results indicated that the protein expression profiles of the METH-dependent group samples (D, E, and F) exhibited a high degree of resemblance, suggesting that these samples share a closely related pattern of protein expression. Similarly, the protein expression profiles within the control group samples (A, B, and C) displayed a substantial similarity, reflecting a consistent pattern of expression distinct from the METH-dependent group. This similarity, as evidenced by the proximity of the samples to one another on the dendrogram and the homogeneity in expression levels represented by the color coding on the heat map, underscores the comparative analysis of gene expression levels within each group. Significant differences in the protein expression profiles were observed between the two sets of groups. Correlation analysis of the samples showed high intragroup correlation but low intergroup correlation between the METH-dependent and control groups (Fig. 2).

Fig. 1.

Hierarchical clustering analysis of genes. Each column represents a sample and each row represents a gene. The tree structure above the colored area represents the clustering of similarity between samples, and the tree structure on the left represents the clustering of similarity between genes. Red indicates a high level of expression, and green indicates a low level of expression. Letters along the x-axis indicate the participant group. Along the y-axis, positive numbers indicate upregulation and negative numbers indicate downregulation.

Fig. 2.

Correlation analysis of the samples. The ordinate and abscissa represent the six samples: A, B, C, D, E, and F. The color depth represents the degree of correlation between the samples as indicated in the color bar.

We used MATLAB software to perform two- and three-dimensional principal component analyses of the 308 proteins (Fig. 3) and completed the orthogonal partial least squares discriminant (OPLS) analysis using SIMCA software (Fig. 4). Both the principal component analysis and OPLS results indicated close clustering of samples within the same group and a clear separation between METH-dependent and control groups based on protein expression levels.

Fig. 3.

Principal Component Analysis. Letters indicate the participant group, with A, B, and C representing samples from healthy individuals and D, E, and F representing samples from individuals with methamphetamine dependence. METH, methamphetamine.

Fig. 4.

Orthogonal partial least squares discriminant analysis. Letters indicate the participant group, with A, B, and C representing samples from healthy individuals and D, E, and F representing samples from individuals with methamphetamine dependence.

The misuse of METH has been associated with cognitive impairment, including AD [28], a neurodegenerative disorder characterized by amyloid beta (A$\beta$) plaques, neuronal death, and memory decline [29]. In the present study, we found DEPs in long-term METH users related to A$\beta$ accumulation and AD pathology, including APP, SAA1, CETP, and adipocyte serum membrane-associated protein (APMAP).

APP is a single transmembrane neuronal protein widely distributed in tissues and concentrated at neuronal synapses [30]. APP can be cleaved into different fragments by $\alpha$-, $\beta$-, or $\gamma$-protease, including cleavage by $\beta$- and $\gamma$-protease that generates A$\beta$ [31]. As the precursor of A$\beta$, APP is implicated in early neuronal damage, but may also have positive protective roles in cell adhesion, maintaining synaptic membrane stability, inhibiting serine protease activity, and participating in the immune response of the CNS [32]. Our results showed an elevated expression of APP in individuals with METH dependence, suggesting increased risk of A$\beta$ plaque formation and neurodegeneration. These results are consistent with previous studies, which suggest METH abuse is related to types of cognitive decline such as deficits in cognitive control, decision-making, and social skills [33], all of which are clinical symptoms of AD [34].

CETP promotes cholesterol transfer from high-density lipoprotein to low-density lipoprotein or very low-density lipoprotein, which may be associated with atherosclerosis. Intracellular cholesterol accumulation also stimulates A$\beta$ production [35]. The high expression of CETP in chronic METH users in the present study suggests heightened risks of cardiovascular disease and A$\beta$-associated neurodegenerative diseases.

APMAP is a membrane protein ubiquitously expressed in various tissues and organs, where it promotes pre-adipocyte differentiation into mature adipocytes to maintain normal physiological metabolism of fat cells. Studies have shown that inhibition of APMAP expression destabilizes APP and increases A$\beta$ production [36]. The present study found decreased serum APMAP protein in METH users compared with healthy controls, which may lead to an increase in the level of A$\beta$ to thereby increase the risk of neurodegeneration.

Collectively, our findings indicate that METH misuse modulates the expression of proteins such as CETP and APMAP, implicating cardiovascular function and neurodegenerative pathways. The upregulation of CETP is suggestive of an augmented risk for cardiovascular pathology and A$\beta$-related neurodegenerative processes, whereas the downregulation of APMAP may signal an enhanced progression towards neurodegeneration through increased A$\beta$ accumulation. These proteomic alterations shed light on the systemic ramifications of METH abuse and underscore the imperative for integrated treatment modalities that concurrently tackle the multifaceted repercussions of METH dependency and its extensive health sequelae.

Prolonged stimulation of the nervous system by METH impairs dopaminergic nerve endings and induces neuronal necrosis [37], potentially contributing to the development of PD [10]. Microglia are important neuroimmune cells in the CNS. Following nervous system damage, microglia become activated, releasing neurotrophic factors to repair tissue, engulfing damaged neurons, and promoting repair, thus acting in a neuroprotective capacity [38]. However, in neurodegenerative diseases, overactivated microglia trigger inflammation and release neuroinflammatory factors, such as nitric oxide, interleukins 6 and 1$\beta$, and tumor necrosis factor $\alpha$, inducing neurotoxicity [12]. Microglia overactivation has been associated with the NF-$\kappa$B signaling pathway [39].

The present study also found significantly elevated expression of DPP4 and CSF1R in the METH group compared with the healthy control group. CSF1R regulates microglial proliferation and differentiation [40]. DPP4, also known as CD26, is a cell surface serine protease that activates the extracellular signal-regulated kinase 1/2 (ERK1/2)–NF-$\kappa$B signaling pathway [41]. Because NF-$\kappa$B can activate microglia, DPP4 may have dual roles: protecting and repairing damaged neurons, while also promoting inflammatory reactions and aggravating damage to the nervous system.

We also observed increased serum LRRTM2 and EFEMP1 protein expression in the METH group. LRRTM2, a neural cell adhesion molecule containing leucine-rich repeats, is involved in neurodevelopmental processes, including axon growth, synapse formation, and axon bundling [42, 43]. EFEMP1, an extracellular matrix glycoprotein containing epidermal growth factor-like domains, regulates glial cell development, migration, differentiation, and synaptic proliferation, suggesting roles in nervous system development, learning, and memory [44, 45]. While we did not examine the functional effects of elevated LRRTM2 and EFEMP1 in chronic METH users, it is still reasonable to suspect that these two proteins may contribute to METH-induced neurological changes. These results provide new potential molecular targets for exploring cognitive deficiency in METH users.

Prolonged METH use compromises both the innate and adaptive arms of the immune system, increasing susceptibility to viral and bacterial infections, indicating adverse effects on immune function [46]. Multiple mechanisms likely contribute to METH-induced immunosuppression. First, METH elevates synaptic dopamine, which is metabolized by monoamine oxidase into reactive oxygen species (ROS). This oxidative stress damages DNA, lipids, and proteins, directly impairing immune cell viability [47]. Second, METH inhibits the function of innate immune cells, such as natural killer cells, dendritic cells, monocytes, macrophages, and granulocytes, thereby suppressing frontline defenses against pathogens [48, 49]. We found several differentially expressed immune-related proteins in METH users that may represent compensatory responses to counteract immunosuppression. Immunoglobulin lambda variable 3 (IGLV3)-16, IGLV3-25, and immunoglobulin heavy variable 3-64 showed decreased expression, suggesting impaired antibody production. Meanwhile, immunoglobulin kappa variable 1-6, IGLV6-57, PIGR, C1qA, and ficolin-3 were upregulated, suggesting attempts to boost humoral immunity.

Immunoglobulins are antigen-binding antibody proteins composed of two light chains and two heavy chains. As immune complexes, immunoglobulins can activate the complement cascade. The complement system is an enzymatic cascade that amplifies immunity through classic, alternative, and lectin pathways [50]. Complement activation on pathogen surfaces mediates cell lysis, while split products interact with adaptive immunity [51]. PIGR is a crucial transmembrane glycoprotein receptor that transports immunoglobulins across mucosal epithelial cells into secretions [52]. Each immunoglobulin molecule requires one PIGR molecule for secretion; therefore, PIGR deficiency reduces antibody levels on the mucosal surface. C1q, part of the complement C1 complex, binds immunoglobulin (Immunoglobulin G [IgG] or Immunoglobulin M [IgM]) complexes to trigger the classic pathway. Elevated C1qA in the brain of individuals with AD suggests a pathogenic role [53, 54]. Ficolin-3 is a pattern recognition receptor that activates the lectin complement pathway through its collagen-like and fibrinogen-like domains, contributing to innate pathogen sensing [53, 55]. The differential expression of these immune-related proteins indicates that METH-induced dysfunction provokes compensatory responses to bolster both innate and adaptive immunity against infections. However, more research is needed to fully elucidate the mechanisms of METH-induced immunosuppression and compensatory immune protein expression. Additionally, the discovery of differentially expressed immune-related proteins in METH users may inform the creation of targeted immunomodulatory treatments. Such therapies would aim to recalibrate the immune equilibrium, mitigating the increased vulnerability to infectious diseases observed among individuals with METH dependence.

METH directly enters neuronal mitochondria through its lipophilic properties, impairing the respiratory chain and disrupting cellular energy metabolism. Brain function maintenance requires substantial energy input, and therefore METH-induced energy deficits can promote neuronal death and neurodegenerative diseases [56, 57].

Lactate dehydrogenase (LDH) interconverts pyruvate and lactate, participating in cellular anaerobic respiration [58]. In AD models, early stage neuronal lactate dehydrogenase (LDHA) upregulation reduces mitochondrial activity and ROS production, conferring A$\beta$ toxicity resistance and improving survival [59]. However, late stage LDHA downregulation shunts pyruvate into mitochondria, elevating ROS and apoptosis, impairing cognition. In our study, LDHB was upregulated in the METH group, possibly an adaptive response similar to early AD, countering A$\beta$/ROS by supplementing anaerobic Adenosine Triphosphate (ATP) generation upon METH-induced respiratory damage. However, LDHB upregulation may also reflect a shift toward aerobic glycolysis and reduced mitochondrial pyruvate metabolism. Further research is required to determine whether LDHB upregulation is beneficial or maladaptive, and to elucidate effects across cell types. Nonetheless, our data indicate perturbed energy homeostasis in METH-induced neurodegeneration.

Calcium is a key intracellular messenger regulating neuronal functions such as long-term potentiation, neuronal perception, synaptic plasticity, and cell proliferation [60]. Calcium dysregulation can promote neurodegeneration [61, 62]. The calcium-binding protein S100A8 mediates inflammatory/oxidative stress responses. Its expression in immune cells is stimulated by nerve injury, eliciting neuroinflammation [63]. Cadherin-5 (CDH5) is a calcium-dependent cell adhesion glycoprotein involved in cell recognition, migration, and tissue morphogenesis. Cadherin expression is developmentally regulated, including N-cadherin enrichment in developing neurons [64]. We found that S100A8 and CDH5 were upregulated in the METH-dependent group. This may represent adaptive responses, with S100A8 counteracting drug-induced calcium perturbations and neuroinflammation, while CDH5 promotes structural plasticity repairing neuronal damage.

It is worth mentioning that, unlike in rodent models, serum represents a minimally invasive human sample. The association between changes in serum proteins and cognitive impairments suggests that the systemic effects of METH extend beyond the blood-brain barrier, manifesting as peripheral biomarkers that may reflect concurrent changes in the CNS [65]. This implies that METH’s systemic impact transcends the blood-brain barrier. Serum proteins can be utilized for the early detection and monitoring of the progression of neurodegenerative diseases associated with METH abuse. For instance, the observed increase in serum levels of APP in this study potentially signals an elevated risk of amyloid plaque formation. Similarly, changes in serum levels of proteins such as CETP and APMAP may indicate underlying neuropathological processes. The differential expression of these genes in serum likely indicates a systemic response to METH’s neurotoxicity.