Abstract Objects To explore the long-term influence of methamphetamine abuse on metabolomics character, with gas chromatography-mass spectrometry (GS-MS) technology, and the potential regulatory network using the bioinformatics method. Methods Forty withdrawal methamphetamine abusers (WMA) were recruited from Shanghai Gaojing Forced Isolation Detoxification Institute. Forty healthy controls (HC) were recruited from society. GS-MS technology was used to detect metabolic products in serum. A bioinformatics method was used to build a regulatory network. Q-PCR was used to detect the candidate gene expressions, and ELISA was used to detect the regulatory enzyme expressions. Results Four pathways were significantly changed in the MA compared to the HC: (1) the arginine synthesis pathway, (2) alanine, aspartic acid and glutamate metabolic pathway, (3) cysteine and methionine metabolic pathway, and (4) the ascorbate and aldarate pathway (enrichment analysis p < 0.05, Impactor factor > 0.2). When focusing on the ‘Alanine, aspartate, and glutamate metabolism’ pathway, a regulatory network was established, and the expression of candidate regulatory genes and enzymes was verified. It was found that the expression of DLG2 (Discs large MAGUK scaffold protein 2), PLA2G4 (Phospholipase A2 group IVE), PDE4D (Phosphodiesterase 4D), PDE4B (Phosphodiesterase 4B), and EPHB2 (Ephrin type-B receptor 2) were significantly different between the two groups (p < 0.05), However, after adjusting for age and BMI, only DLG2, PLA2G4, and EPHB2 remained significant (p < 0.05). The expression of enzymes was not significantly different (p > 0.05). Conclusion Methamphetamine abuse influences the metabolic process in the long term, and DLG2, PLA2G4, and EPHB2 may regulate the glutamate metabolism pathway. Keywords: methamphetamine abuse, metabolomics, bioinformatics, regulatory network, glutamate metabolic pathway Introduction Methamphetamine (METH) is a widely abused addictive psychostimulant, which accounts for a considerable share of global disease burden. METH abuse is increasing rapidly ([35]Yang X. et al., 2018), especially in East and South-East Asia, and Oceania. Long-term abuse of METH causes serious physical and mental damage. Compared to other psychostimulants, METH is better at penetrating the center nervous system (CNS) and has a longer duration of action ([36]Yang X. et al., 2018), making METH more addictive while the damage to the CNS is more serious and persistent. METH addicts often suffer from permanent psychotic symptoms ([37]Mellsop et al., 2019) and cognitive decline, even when deprived from drugs, bringing great difficulties for patients who try to return to society. These symptoms are mostly attributed to METH-induced long-term neurotoxicity and excitotoxicity ([38]Ashok et al., 2017). The neurochemical mechanisms underlying neurotoxicity and excitotoxicity are complex and not well understood. Glutamate, as an important excitatory neurotransmitter is thought to activate various downstream signaling pathways, for example the apoptotic pathway, and eventually produce neurotoxicity ([39]Courtney and Ray, 2014; [40]Yang X. et al., 2018). A previous study found that methamphetamine may result in glutamine system disturbance ([41]Tasic et al., 2017), and glutamate + glutamine in the right inferior frontal cortex is decreased in methamphetamine users ([42]Zheng et al., 2016). Metabolomics is a potentially powerful tool for understanding the global biological consequences of disease and drug administration. In recent years, with the improvement of metabolomics technology, metabolomics research in psychotic diseases has made great progress ([43]Bortolato et al., 2009; [44]Zhang et al., 2016; [45]Proitsi et al., 2017; [46]Rodríguez Cerdeira et al., 2017). However, in the field of substance addiction, metabolomics research is still in its initial stages. A few studies focus on methamphetamine addiction, using animals as study subjects. Energy metabolic disturbance is believed to be the biochemical base of toxicity and addictive property of MTHE ([47]Zhang et al., 2016). Neurotransmitters in serum or plasma are also disrupted by METH, especially in the glutamate-glutamine-GABA system ([48]Rist et al., 2017). Nevertheless, metabolic character differences between animals and humans still exist. This study uses withdrawal methamphetamine abusers (WMA) as objects and explore the metabolomics characters and possible regulatory mechanisms that may be the basis of the long-term neurotoxicity of methamphetamine. Materials and Methods Ethics Statement Informed consent of the participants was obtained after the nature of the procedures used in the study had been fully explained. The study was carried out in accordance with the last version of the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of Shanghai Mental Health Center (SMHC IRB 2016-11R). All participants were identified using numbers, instead of names. Research Participants Forty male WMA were recruited from Shanghai Gaojing Forced Isolation Detoxification Institute in July 2017, and 40 male healthy controls were recruited from general society. The entire sample consisted of a Chinses Han population. The WMA accepted assessment from two senior psychiatric doctors and met the diagnosis criterion for methamphetamine abuse disorder according to the Diagnostic and Statistical Manual of mental disorders, fifth edition (DSM-5). Participants did not have the opportunity to use any illicit substance during the last 2 months. They completed a self-report form that included basic demographic information and accepted an interview of Addiction Severity Index (ASI). Exclusion criteria for both groups were: (1) history of craniocerebral trauma or epileptic seizure; (2) metal implants in the brain or an implanted pacemaker; (3) history of major physical diseases, organic mental disorders, and severe mental disorders. People in the control group did not have any history of illegal substance abuse. Venous blood was collected at 7:30–8:00 am before breakfast. Gas Chromatography – Mass Spectroscopy (GC-MS) After standing at room temperature for 30 min, venous blood was centrifuged at 4°C, 3000g, for 15 min. The upper layer of the serum was collected. Sample preparation, GC-MS experimental process, and GC-MS data processing were carried out using previously published methods ([49]Morris et al., 2012; [50]Li et al., 2014). Compound identification was performed by comparing the retention indices and mass spectral data with those previously generated from reference standards of known structures present in the JiaLib metabolite database using the proprietary software XploreMET. The current JiaLib comprises over 1,200 mammalian metabolites over a 15-year accumulation. The reference chemicals present in JiaLib were commercially purchased from Sigma-Aldrich (St. Louis, MO, United States), Santa Cruz (Dallas, TX, United States), Nu-Chek Prep (Elysian, MN, United States), and synthesized in the laboratory. XploreMET adopted a specific algorithm hypergeometric, which can enrich the information of metabolites, weight the contribution of metabolites, and focus on the key metabolic network. The P and impact value was used to measure the importance of metabolic pathway. Bioinformatics Analysis Based on the results of untargeted metabolomics, a glutamate metabolism pathway was noticed. Glutamate is a very important excitatory neurotransmitter, and a previous study has confirmed that glutamate + glutamine in the right inferior frontal cortex was decreased in methamphetamine users ([51]Zheng et al., 2016). In serum or plasma, the glutamate-glutamine-GABA system was also disrupted by METH ([52]Rist et al., 2017). So, we chose the ‘alanine, aspartic acid and glutamate metabolic pathway’ to build the regulatory network. The four differential expressed metabolites were treated as candidate metabolites. We collected 133 risk genes from previous genome-wide association studies ([53]Heikkinen et al., 2018; [54]Irwin et al., 2018). The molecular network was constructed by connecting candidate metabolites, enzymes, intermediate proteins, and risk genes. First, we investigated the relative enzymes linked to these candidate metabolic compounds according to the KEGG database and the Human Metabolome Database (HMDB). To further explore correlation networks linked to the relative enzymes and risk genes, we searched the functional proteins network database (STRING v10.5) setting the requirement that the score must be more than 700 and that no more than one intermediate protein must be present between the relative enzyme and risk gene. Finally, we built a network including four enzymes: ASNS (Asparagine synthetase), AGT (Angiotensinogen), GAD1 (Glutamate decarboxylase 1), and GAD2 (Glutamate decarboxylase 2); 13 genes: DLG2 (Discs large MAGUK scaffold protein 2), NRG1 (Neuregulin 1), PLA2G4E (Phospholipase A2 group IVE), STX8 (Syntaxin 8), NEDD4L (NEDD4 like E3 ubiquitin protein ligase), PRKG1 (Protein kinase cGMP-dependent 1), PDE4D (Phosphodiesterase 4D), PDE4B (Phosphodiesterase 4B), SERPINA5 (Serpin family A member 5), PDE6C (Phosphodiesterase 6C), CD55 (CD55 molecule), EPHB2 (Ephrin type-B receptor 2), and RAPGEF5 (Rap guanine nucleotide exchange factor 5), and several intermediate proteins. RNA Extraction and Gene Expression Analysis RNA was extracted from the serum samples with Trizol reagent (Invitrogen, Carlsbad, CA, United States) according to the previous protocol ([55]Gan et al., 2018), and stored at −80°C for subsequent cDNA synthesis. Following RNA quantitation, 1ug of total RNA was used for cDNA synthesis (5 × All-In-One RT Master Mix, ABM, Canada), and the quantitative gene expression analysis was performed using TB Green Fast qPCR Mix (TaKaRa, Shiga, Japan), and Lightcycle^® 96 real-time PCR machine (Roche Diagnostics GmbH, Germany). Primers are listed in [56]Table 1. GAPDH was used as a housekeeper gene. The relative gene expression was calculated using the 2^−△CT method ([57]Kohno et al., 2016). The CT values of duplicates were averaged for the final analysis. TABLE 1. Primers used in Q-PCR. gene Primers (F: Forward; R: Reverse) GAPDH F AGGGCTGCTTTTAACTCTGGT R CCCCACTTGATTTTGGAGGGA DLG2 F CCCAATGGGATGGCAGACTT R AGTATTCCCACCTCCCTCCC NRG1 F GCAACTTTGTTTCCCGGGTC R TTGGGGGCAAAAGTCACACT PLA2G4E F AATCAGTGCTCCCTTGAGCC R GCATTACCTGGATGGGGACC STX8 F CCCCTGGTTCTCCACATACG R TCTGTCTTCGGTCCCCTTCA NEDD4L F AGGGTTAGCTTCCTGTTGGC R TGGTACGGGGTTGAGAATGC PRKG1 F ACTTGGTGCACCTCTCAACA R AGTCATTGCCAAGGTCCCAG PDE4D F TCTTACAGCCCACGGGGATA R AGGCAGAATCAACCCATGCT PDE4B F GTGTCGTTCACCGTGAGAGT R CGGGGTGAAGAGAGGAGGTA CD55 F CATCTTTCCTTCGGGTTGGC R CACCATCAACACCCCTGGTT EPHB2 F AGTCTGGGGAGGGACTCATC R CTGTGTGGCCATGGAAGCTA RAPGEF5 F GGGCTTTCACTGCTACCCAT R GATTGGCTGCGCCATTTAGG SERPINA5 F GCTATGGCCCATCTGTATGCT R TTCCCCAAGGCACCTGTATG PDE6C F TTTTGGAGAGGCACCACCTG R AACTGTTTCAAACTGCCGCT [58]Open in a new tab Elisa The serum levels of AGT, GAD1, GAD2, and ASNS were assessed using an ELISA kit (catalog number SEC904Hu, Cloud-Clone Corp, Shanghai, China for GAD1, and catalog numbers MBS726474, MBS2501551, MBS090144; Mybiosource, Southern California, San Diego, United States, for AGT, GAD2, and ASNS, respectively). All assays were performed according to the manufacturer’s directions and were performed in duplicate. The average value of the duplicates was used as the final value. Statistical Analysis General data was presented as mean and standard error of the mean (Mean ± SD), and analyzed using SSPSS 22.0. Data was analyzed using the student t test, a was 0.05 and Cohen’s d was used to calculate the effect size of the t test. 2^–△CT was used to calculate the relative expression of genes, and the expression ratio of the WMA group relative to the control group was calculated as previously described ([59]Kohno et al., 2016). Metabolomics data was managed with the XploreMET system (v3.0, Metabo-Profile, Shanghai, China), and R was used to analyze the data. The NIST 2017 database was used for metabolite certification and Metaboanalyst 3.0 was used for the metabolic pathway analysis. Partial least squares discriminant analysis (PSL-DA) was performed, and a t test was used after log transformation. Differential expressed metabolites were screened according to fold change (FC) and p value. FC > 1. 5 and p < 0. 05 showed that the metabolite was significantly differentially expressed. Based on the selected differential metabolites, a pathway analysis was performed using a hypergeometric test and KEGG pathways ([60]Yang W. et al., 2018) and the p values were calculated as follows: [MATH: Pp⁢ t⁢w(PtwH< /mi>⁢i⁢t⁢s |TotalA⁢l⁢l ,PtwA⁢l⁢l,TotalH⁢i⁢t⁢s(P⁢t⁢ wA⁢l⁢lP⁢t⁢ wH⁢i⁢t⁢ s)⁢(T⁢o⁢ t⁢a⁢lA⁢l⁢l-P⁢t⁢wA⁢l⁢lT⁢o⁢ t⁢a⁢lH⁢i⁢t⁢s-P⁢t⁢wH⁢i⁢t⁢s) (T⁢o⁢t< /mi>⁢a⁢lA⁢ l⁢lT⁢o⁢t< /mi>⁢a⁢lH⁢ i⁢t⁢s) :MATH] Ptw[Hits] is the hit number of differential metabolites in a certain metabolic pathway. Total[All] is the number of metabolites in all pathways of a certain species, such as Homo sapiens. Ptw[All] is the total number of metabolites in a certain pathway and Total[Hits] is the hit number of differential metabolites in all pathways. A significant p value means significantly changed metabolic pathway activities that need to be focused on. For multiple tests, a false discovery rate (FDR) was applied to correct the raw p values. Results Demography Characteristic The demographic characteristic are shown in [61]Table 2. Age, BMI, and education level between the two groups were not significantly different. However, WMA individuals have lower marriage rates and higher unemployment (p < 0.05). TABLE 2. Demographic characteristics of subjects. WMA (Mean ± SD) Control (Mean ± SD) t value p value Age (years) 34.02 ± 6.82 32.02 ± 8.57 1.13 0.260 BMI (kg/m2) 24.44 ± 3.64 23.54 ± 3.23 1.14 0.260 Education (years) 5.95 ± 3.55 6.97 ± 3.14 1.14 0.256 Marital status 16.537 0.000 Married 9 (23.1%) 20 (52.6%) Divorced/Separated 15 (38.4%) 1 (2.6%) Unmarried 15 (38.4%) 17 (44.7%) Employment 8.271 0.004 Unemployed 18 (46.1%) 6 (15.8%) The age began to abuse (years) 25.20 ± 7.52 Withdrawal time (month) 3.20 ± 1.32 Duration of abuse (years) 6.51 ± 3.26 Average dosage (g) 0.64 ± 0.35 Frequency Everyday 24 (60.0%) 3–5 times/week 12 (30.0%) 1 time/week 2 (5.0%) 1–3 times/month 2 (5.0%) [62]Open in a new tab Metabolomics Profile and Pathways Thirty-five differentially expressed metabolites were screened by t test (p < 0.05), in which 21 metabolites could be found by searching metabolic pathways in the KEGG information database (as shown in [63]Table 3). Since age and BMI may influence the metabolic function, we adjusted age and BMI. After that, there were 19 metabolites differentially expressed in two groups, including two kinds of alcohols, seven kinds of amino acids, one kind of carbohydrate, one kind of fatty acid, three kinds of nucleotides, and five kinds of organic acids. TABLE 3. Differentially expressed metabolites in pathways. Metabolites species Metabolites Metabolic pathways FC value Before adjusting __________________________________________________________________ After adjusting __________________________________________________________________ Cohen’s d p value p value Alcohols Glycerin Galactose metabolism, glyceride metabolism 0.6 0.000 0.000 0.898 Alcohols Inositol Galactose metabolism, inositol metabolism, inositol phosphate metabolism, phosphatidylinositol phosphate metabolism 0.7 0.001 0.001 0.813 Amino acids 1-Methylhistidine Histidine metabolism 1.4 0.001 0.001 –0.802 Amino acids L- Aspartic acid Ammonia recovery, arginine and proline metabolism, aspartic acid metabolism, B-alanine metabolism, malate-aspartic acid shuttle, transcription/translation, urea cycle 1.5 0.001 0.002 –0.761 Amino acids Homocysteine Betaine metabolism, catecholamine biosynthesis, glycine and serine metabolism, homocysteine degradation, methionine metabolism 1.4 0.004 0.004 –0.692 Amino acids L- Methionine Betaine metabolism, glycine and serine metabolism, methionine metabolism, spermidine and spermine biosynthesis, transcription/translation 0.9 0.005 0.009 0.658 Amino acids L- Glutamic acid Amino sugar metabolism, ammonia recovery, glutamate metabolism, phenyl acetate metabolism, purine metabolism, pyrimidine metabolism, transcription/translation, urea cycle 1.1 0.011 0.015 –0.597 Amino acids Ornithine Metabolism of arginine and proline, metabolism of glycine and serine, biosynthesis of spermine and spermine, urea cycle 1.3 0.013 0.020 –0.582 Amino acids L- Cysteine Cysteine Metabolism, Glutathione Metabolism, Glycine and Serine Metabolism, Methionine Metabolism, Pantothenate and CoA Biosynthesis, Tauurine and Tauurine Metabolism, Transcription/Translation 0.9 0.028 0.016 0.517 Amino acids L- Isoleucine Degradation, transcription/translation of valine, leucine and isoleucine 1.1 0.029 0.088 –0.512 Amino acids L- Proline Arginine and proline metabolism, transcription/translation 1.2 0.032 0.068 –0.502 Carbohydrate Sorbitol Fructose and mannose degradation, galactose metabolism 0.9 0.025 0.027 0.525 Fatty acids Linoleic acid α-linolenic acid and linoleic acid metabolism 0.8 0.007 0.008 0.643 Nucleotides Hypoxanthine Purine metabolism 1.4 0.000 0.000 –1.046 Nucleotides Xanthine Purine metabolism 1.2 0.000 0.000 –0.900 Nucleotides Guanosine hydrate Purine metabolism 0.7 0.005 0.006 0.658 Organic acids Glyceric acid Glyceride metabolism, glycine and serine metabolism 0.7 0.000 0.000 1.023 Organic acids Pyruvic acid Alanine metabolism, amino glucose metabolism, ammonia recovery, tricarboxylic acid cycle, cysteine metabolism, gluconeogenesis, glucose-alanine cycle, glycine and serine metabolism, glycolysis, pyruvic aldehyde degradation, pyruvate metabolism, acetyl mitochondrial transfer, urea cycle 0.6 0.000 0.000 0.872 Organic acids Vanillylmandelic_acid Tyrosine metabolism 0.8 0.002 0.002 0.744 Organic acids Hydroxypropionic_acid Propionic acid metabolism 0.9 0.033 0.027 0.497 Organic acids Fumaric acid Arginine and proline metabolism, aspartic acid metabolism, citric acid cycle, mitochondrial electron transport chain, phenylalanine and tyrosine metabolism, tyrosine metabolism, urea cycle 0.7 0.037 0.047 0.488 [64]Open in a new tab According to the results of the Metabolic Pathway Enrichment Analysis (MPEA), four pathways were found to be significantly changed; (1) arginine synthesis pathway, (2) alanine, aspartic acid and glutamate metabolic pathway, (3) cysteine and methionine metabolic pathways, and (4) ascorbate and aldarate pathway (enrichment analysis p < 0.05,Impactor factor > 0.2) ([65]Figure 1). FIGURE 1. [66]FIGURE 1 [67]Open in a new tab Significant pathways (the pathways which p < 0.05, and Impactor factor > 0.02 were marked out). Glutamine Metabolites Regulatory Network Previous research found that methamphetamine may result in glutamine system disturbance ([68]Tasic et al., 2017), either in the brain and serum or in plasma ([69]Zheng et al., 2016; [70]Rist et al., 2017). This study also found a change in the glutamate pathway. We therefore focused on the ‘Alanine, aspartate, and glutamate metabolism’ pathway, and based on the bioinformatics method, the regulatory network was built. As shown in [71]Figure 2, we found three metabolites (pyruvate, L-aspartate, and L-glutamine), which can be regulated by four enzymes, and which can be related to 13 candidate genes. Each candidate gene influences the regulatory enzyme through one intermediate protein ([72]Figure 2). FIGURE 2. [73]FIGURE 2 [74]Open in a new tab The regulatory network model for alanine, aspartic acid and glutamate metabolism pathway(enzyme, posttranslational modification, activate, inhibition, catalysis, reaction, and combine. Yellow oval, metabolites of glutamate pathway; Green parallelogram, regulatory enzymes; Purple diamond, intermediate proteins; Red square, candidate genes). Candidate Gene Expression According to the regulatory network, we used Q-PCR to verify the 13 candidate genes expressed. 2^–△ ^CT was used to calculate the relative expression of genes ([75]Kohno et al., 2016) ([76]Table 4). DLG2, PLA2G4, PDE4D, PDE4B, and EPHB2 were significantly differentially expressed between the two groups (p < 0.05). However, after adjusting for age and BMI, only DLG2, PLA2G4, and EPHB2 remained significant (p < 0.05). STX8, RAPGEF5, and PDE6C were not detected because of their low expression. TABLE 4. Expression of candidate genes. WMA Control Before adjusting __________________________________________________________________ After adjusting __________________________________________________________________ Cohe^n′sd t value p value t value p value DLG2 0.250 ± 0.104 0.190 ± 0.083 2.804 0.006 –2.348 0.022 0.638 NRG1 0.220 ± 0.071 0.210 ± 0.066 1.207 0.231 –1.002 0.320 0.146 PLA2G4 0.050 ± 0.028 0.020 ± 0.024 4.260 0.000 –3.672 0.000 1.150 STX8 – – – – – – – NEDD4L 0.140 ± 0.041 0.140 ± 0.053 –0.532 0.596 0.528 0.599 –0.101 PDE4D 0.140 ± 0.038 0.120 ± 0.050 2.007 0.048 –1.665 0.100 0.450 PDE4B 0.170 ± 0.039 0.150 ± 0.040 2.191 0.032 –1.722 0.090 0.506 CD55 0.110 ± 0.032 0.100 ± 0.036 0.912 0.365 –0.609 0.545 0.294 EPHB2 0.280 ± 0.137 0.190 ± 0.065 3.731 0.000 –3.324 0.001 0.839 RAPGEF5 – – – – – – – SERPINA5 0.220 ± 0.076 0.190 ± 0.052 1.904 0.061 –1.639 0.106 0.461 PDE6C – – – – – – – [77]Open in a new tab For the DLG2, PLA2G4, PDE4D, PDE4B, and EPHB2 genes, the expression in the WMA group was 1.473, 4.226, 1.334, 1.153, and 1.595 times higher than that in control group, respectively ([78]Figure 3). FIGURE 3. [79]FIGURE 3 [80]Open in a new tab Relative expression fold of candidate genes. *Means that these genes are significantly different expressed in two groups. Regulatory Enzyme Expression ELISA was used to detect the regulatory enzyme expression. As [81]Table 5 shows, no significant difference was found between the two groups (p > 0.05). TABLE 5. Expression of candidate enzymes. WMA Control t value p value Cohen’s d ASNS 2.28 ± 1.65 2.21 ± 1.44 0.18 0.86 0.045 AGT 3.31 ± 1.23 3.52 ± 1.64 –0.63 0.53 –0.145 GAD1 0.49 ± 0.13 0.50 ± 0.12 –0.34 0.74 –0.080 GAD2 3.88 ± 0.92 3.71 ± 1.56 0.59 0.54 0.133 [82]Open in a new tab Discussion This study found that peripheral metabolites of WMA patients were mainly changed in amino acid metabolism. The most obvious pathways were; (1) the arginine synthesis pathway, (2) the alanine, aspartic acid and glutamate metabolic pathways, (3) the cysteine and methionine metabolic pathways, and (4) the purine metabolic pathway. Previous studies on methamphetamine addicted animals found that abnormal energy metabolism was the biochemical basis of toxicity and addiction ([83]Cservenka and Ray, 2017; [84]Moallem et al., 2018). An increase in energy metabolism was found in the serum of amphetamine-administered rats. After 5 consecutive days of administration, the difference between the administration and control group decreased ([85]Moallem et al., 2018). After high dose, high frequency, or long time administration, the energy reserve may be depleted ([86]Moallem et al., 2018). The patients in this study were basically deprived from METH for about 2 months. This may explain why abnormal energy metabolism was not observed—the energy reserves may have been consumed. In addition, amphetamines cause serum/plasma neurotransmitter disturbances mainly in the glutamate-glutamine-GABA system ([87]Moallem et al., 2018). Glutamate-Glutamine disturbance was observed both in withdrawal animals and in the WMA patients ([88]Shima et al., 2011; [89]Chen et al., 2015; [90]Su et al., 2018), and this disturbance may be related to a continuous search for drugs and the high relapse risk ([91]Zheng et al., 2014). Glutamate, aspartic acid, and methionine are elevated in rat neuroblastoma cells 48 h after methamphetamine treatment ([92]Bu et al., 2013). These studies showed that the effect of methamphetamine on the glutamate system could last for a long time. In this study, the “alanine, aspartic acid, and glutamate pathway” was found to be abnormal, and the expression of related metabolites was significantly changed, which is consistent with previous studies. We chose the glutamate metabolic pathway and used bioinformatics methods to establish a regulatory network model, which predicts that GAD1, GAD2, ASNS, and AGT may directly regulate the glutamate pathway metabolites- pyruvate, L-aspartate, L-glutamine. The regulatory enzyme is regulated by DLG2, NRG1, PLA2G4E, STX8, NEDD4L, PRKG1, PDE4D, PDE4B, SERPINA5, PDE6C, CD55, EPHB2, and RAPGEF5 genes only through one intermediate protein. We verified the expression of the candidate genes and enzymes, and finally found five regulatory genes that were differentially expressed in the two groups, namely DLG2, PLA2G4, PDE4D, PDE4B, and EPHB2 genes. After adjusting for age and BMI, only DLG2, PLA2G4, and EPHB2 remained significant. DLG2 plays a very important role in synaptic development and stability of plasticity ([93]McClay et al., 2013; [94]Su et al., 2020). The function of the DLG2 gene is associated with complex cognitive and learning tasks ([95]Parsegian and See, 2014), and is also associated with developmental and intellectual disorders ([96]Christopher et al., 2007) such as Parkinson’s disease ([97]Darna et al., 2015). The DLG2 gene polymorphism was relevant for methamphetamine addiction ([98]Irwin et al., 2018). This study also found that the expression of the DLG2 gene was up-regulated in WMA subjects. There was a long-term cognitive impairment in methamphetamine addicted populations ([99]Crocker et al., 2014), as well as dysfunction of the glutamate system ([100]Maker et al., 2018). The DLG2 gene may play an important role in the development, plasticity, and stability of glutamatergic synapses and may further affect the cognitive function of WMA patients. The protein coded by the PLA2G4E gene is involved in the regulation of membrane-mediated transport and affects phospholipase activity ([101]Zhu et al., 2016). PLA2G4E is strongly expressed in neurons and regulates endogenous cannabinoids, by mobilizing intracellular calcium ([102]Zhu et al., 2016). The endogenous cannabinoid system also regulates the function of the glutamate system. Cannabinoid 1 receptors affect glutamatergic and GABAergic regulation of anxiety and fear responses ([103]Zheng et al., 2011; [104]Nithianantharajah et al., 2013). The PLA2G4E gene expression changed the most in this study, reaching 4.226 times that of the control group. That may influence cannabinoid, glutamate, and GABA system functions. The PDE4D and PDE4B genes belong to the phosphodiesterase (PDE family) superfamily, and play an important role in the degradation of cAMP ([105]Reggiani et al., 2017). Since cAMP is an important second messenger, PDE4D/4B may play a role in signal transduction by regulating cAMP concentration. PDE4D/4B is widely expressed in the human brain. Previous studies have shown that PDE4D/4B is associated with neuropsychiatric disorders such as depression, schizophrenia ([106]Wu et al., 2018), methamphetamine addiction ([107]Irwin et al., 2018), alcohol-induced neuroinflammation ([108]Ikeda et al., 2013), and stroke ([109]Zhong et al., 2016). Since signal transduction of neurotransmitters in the brain, such as dopamine, glutamate, and GABA, etc., requires the participation of second messenger cAMP, some researchers believe that the phosphodiesterase family (PDE) may be used as an important drug target for schizophrenia and affective disorder ([110]Wu et al., 2018). This study found that the expression of PDE4D/4B in the peripheral blood of WMA patients was significantly higher than that of the control group. However, after adjusting for age and BMI, this difference was no longer significant. The expression ratios of the WMA group compared to the control group were also close to 1, showing that these two genes may not be important regulatory genes in the glutamine system of WMA. The EPHB2 gene encodes Ephrin type-B receptor 2, which is a tyrosine kinase receptor involved in triggering and maturation. In mature brains, EPHB2 is highly abundant in large dendritic axes and in the frontal cortex and dendritic spines in the hippocampus, and is also involved in the regulation of glutamate receptor distribution and excitatory synapse formation ([111]Su et al., 2017). EPHB2 also regulates synaptic plasticity and leads to anxiety behavior ([112]Capestrano et al., 2014) and reversal of cognitive deficits ([113]Ruehle et al., 2012). EPHB2 reverses cognitive dysfunction in the Alzheimer’s disease model, possibly related with glutamate receptor-NMDA’s function (N-methyl-D-aspartic acid receptor), and EPHB2 is a key regulator of synaptic localization of NMDA receptors ([114]Hu et al., 2017). Knockout EPHB2 induces depression-like behavior and memory impairment in mice, which is mediated by NMDA receptor subtypes ([115]Zhen et al., 2018). This study found a significant increase of the EPHB2 gene in WMA patients, which elicits that an abnormal expression of the EPHB2 gene may be a potential mechanism of glutamate system dysfunction and emotional or cognitive symptoms of methamphetamine addicts. The expression of candidate regulatory enzymes was not differentially expressed between two groups. However, the activity of enzymes was not detected, which may influence the metabolites. Another complex regulatory mechanism could not be excluded. Conclusion MA use has a long-term and extensive impact on the human body, which can be reflected by the peripheral metabolites. The glutamate system metabolic disturbance may be regulated by the genes that are associated with multiple pathophysiological processes such as synaptic plasticity, signal transduction, cAMP degradation, phosphorylation, and synaptic localization, which may be related to clinical symptoms of withdrawal MA, such as long-term cognitive impairment, anxiety and fear responses, and so on. There were some limitations in this study. First, only one time point was selected for the exploration of serum metabolomics in the addiction state of methamphetamine. If long-term follow-up can be performed during the acute phase and withdrawal period, the changes in metabolomics characteristics over time will be observed and the results will be more systematic. Second, the exploration of the mechanism is limited to biochemical and gene regulation mechanisms, and if it can be combined with the analysis of central neurotransmitter changes, it will be more in-depth. Third, the enzyme activity was not detected. Fourth, the sample size was limited, and all the enrolled participants were from the Shanghai area only. The conclusion can therefore only be extended to Shanghai Han nationality. A multi-center collaboration may therefore facilitate further discoveries. Data Availability Statement The raw data supporting the conclusions of this article will be made available by the authors. Ethics Statement The studies involving human participants were reviewed and approved by Ethics Committee of Shanghai Mental Health Center. The patients/participants provided their written informed consent to participate in this study. Author Contributions SP: conceptualization, data curation, formal analysis, methodology, writing – original draft, review, and editing. HS: investigation, project administration, and supervision. TC: investigation and project administration. XL: investigation and project administration. JD: conceptualization, methodology, project administration, resources, and supervision. HJ: conceptualization, methodology, project administration, resources, and supervision. MZ: conceptualization, funding acquisition, methodology, resources, supervision, and writing – review and editing. All authors contributed to the article and approved the submitted version. Conflict of Interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Acknowledgments