Fewer than 50% of all patients with major depressive disorder (MDD) treated with currently available antidepressants (ADs) show full remission. Moreover, about one third of the patients suffering from MDD does not respond to conventional ADs and develop treatment-resistant depression (TRD). Ketamine, a non-competitive, voltage-dependent N-Methyl-D-aspartate receptor (NMDAR) antagonist, has been shown to have a rapid antidepressant effect, especially in patients suffering from TRD. Hippocampi of ketamine-treated mice were analysed by metabolome and proteome profiling to delineate ketamine treatment-affected molecular pathways and biosignatures. Our data implicate mitochondrial energy metabolism and the antioxidant defense system as downstream effectors of the ketamine response. Specifically, ketamine tended to downregulate the adenosine triphosphate (ATP)/adenosine diphosphate (ADP) metabolite ratio which strongly correlated with forced swim test (FST) floating time. Furthermore, we found increased levels of enzymes that are part of the ‘oxidative phosphorylation’ (OXPHOS) pathway. Our study also suggests that ketamine causes less protein damage by rapidly decreasing reactive oxygen species (ROS) production and lend further support to the hypothesis that mitochondria have a critical role for mediating antidepressant action including the rapid ketamine response.
Mood disorders including major depressive disorder (MDD) are a leading cause of disability worldwide1,2. The prevalence for mood disorders has been estimated at 21 million people across 28 European countries with an annual cost of 106 billion Euro3,4. Globally, approximately 350 million people suffer from MDD, with women more likely to be affected than men5,6. The currently available antidepressants (ADs) require chronic long-term treatment of several weeks before a therapeutic effect is achieved7,8. Less than 50% of all patients treated with current ADs show full remission9. Moreover, about one third of the patients suffering from MDD do not respond to conventional ADs and suffer from treatment-resistant depression (TRD)10. Fifteen percent of severely depressed individuals commit suicide, the second leading cause of death for ages 15- to 29-years6.
In 2000 Berman et al. demonstrated for the first time a rapid antidepressant effect of a low dose of ketamine within 2 h and 4 h of treatment that lasts for up to 10 days11. Ketamine is a non-competitive, voltage-dependent N-Methyl-D-aspartate receptor (NMDAR) antagonist12,13,14,15,16,17. The antidepressant effect of ketamine is also observed in patients suffering from TRD18,19. Moreover, ketamine diminishes suicidal ideation20. However, psychomimetic side effects of ketamine have prevented its use as a first-line drug21,22,23,24,25.
A low dose of ketamine was initially reported to act through the activation of the mammalian target of rapamycin complex 1 (mTORC1) causing an increase of synaptic protein levels and an elevated number of spines and neuronal activity in the medial prefrontal cortex of rats26. Furthermore, increased levels of postsynaptic proteins required for formation, maturation, and function of new spines were observed 2 h after ketamine treatment and lasted up to 72 h26. More recent studies of ketamine’s antidepressant effects implicate a ketamine metabolite as the actual active compound. Results from studies in mice treated with the ketamine metabolite (2 R, 6 R)-hydroxynorketamine (HNK) suggest that the antidepressant-like effect of the drug is independent of NMDAR binding and implicate the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) and its signaling pathway to be critical for the activity27.
In previous studies we have shown an important role of mitochondrial energy metabolism for ketamine’s antidepressant mode of action. A time-dependent metabolomics profiling study implicated ‘glycolysis’ and the ‘citrate cycle’ pathways. In the present paper we report on an extension of our analyses on ketamine’s fast acting antidepressant activity that further support a critical role of mitochondrial pathways impacting cellular oxidative stress defense mechanisms28.
Our previous study on ketamine’s mode of action suggested an involvement of mitochondrial energy metabolism, a finding that is also supported by its link to glutamate neurotransmission. Glutamate is metabolized from α-ketoglutarate, a ‘citrate cycle’ metabolite, and can be recycled through this pathway. Interestingly, glutamate neurotransmission also stimulates glucose utilization and ‘glycolysis’, the upstream pathway of the OXPHOS pathway. Therefore, we investigated whether ketamine’s fast antidepressant-like effects are reflected by metabolite ratios that are part of the ‘citrate cycle’ and the connected OXPHOS pathway, including ATP/ADP, NADH/NAD and GTP/GDP ratios. In addition, we correlated these metabolites with antidepressant-like behavior using the FST floating time as a readout. Interestingly, the ATP/ADP metabolite ratio tended to be decreased 2 h and 24 h after a single injection of a low dose of ketamine. In addition, ATP and ADP levels significantly and strongly correlated with the FST floating time at 24 h (Fig. 1A). ATP is produced from ADP by OXPHOS complex V and NADH serves as an electron donor for OXPHOS complex I28,29. The NADH/NAD ratio neither showed a significant difference nor a significant correlation with the FST floating time (Fig. 1B). GTP is synthesized from GDP as a by-product of succinate-CoA to succinate conversion in the ‘citrate cycle’. Succinate is an electron donor for the OXPHOS complex II subunit A that we have previously shown to be elevated 2 h after ketamine treatment28. We found that the GTP/GDP metabolite ratio is indeed significantly elevated 2 h upon ketamine treatment and tended to be downregulated at the 24 h time point. In addition, GTP levels significantly correlated with FST floating time at both time points (Fig. 1C).
Based on the above results we next wanted to investigate whether AMP-activated protein Kinase (AMPK) is affected by ketamine treatment. AMPK is the major energy status sensor that promotes catabolic pathways that generate ATP and inhibits anabolic pathways that consume ATP. A low ATP/ADP ratio activates AMPK resulting in its phosphorylation, leads to decreased activities of pathways that consume energy (anabolic pathways) and increased activities of pathways that produce energy (catabolic pathways)30,31,32,33,34,35,36. Previous data from a study in rats suggest that ketamine’s antidepressant-like effects are mediated by an increased anabolic rate, thereby inducing cell growth and differentiation26. The result of an elevated anabolism is a higher energy demand. This is supported by our results of an increased pAMPK/AMPK ratio and decreased ATP/ADP ratio and decreased ATP levels at the 24 h time point28. While total AMPK protein levels are not significantly changed, the phosphorylated form of the protein, pAMPK, shows a statistically significant increase at the 24 h time point (Fig. 2, Supplementary Figure 1A and B).
We next investigated whether the observed energy status alterations upon ketamine treatment are associated with changes in the expression of proteins related to cellular energy metabolism. Time-dependent quantitative proteomics profiling analyses were performed 2 h and 24 h after a single injection of a low dose of ketamine. Five mice per time point and treatment group were chosen and MF and CF from ketamine- and vehicle-treated mice were isolated from the hippocampus and quantified using 15N metabolically labelled reference proteins by LC-MS/MS analyses.
After proteomics data processing and missing value imputation by k-nearest neighbor (KNN), a total of 889 hippocampal MF and 1173 CF proteins were processed by statistical analyses for the 2 h time point. For the 24 h time point 929 MF and 1091 CF proteins were used for further statistical analyses (Supplemental Table S1). The protein profiles allowed a separation of ketamine- and vehicle-treated mice using multivariate PLS-DA for both time points (Fig. 3). The quality criteria of PLS-DA models were assessed by accuracy, R2, and Q2 values. They indicate good models for the 2 h time point for MF and CF proteins of ketamine- compared to vehicle-treated animals. The PLS-DA models were only weak for the 24 h time point, which might result in false-positives (Supplemental Table S2). To increase the robustness of the statistical analyses we combined the VIP-score of the PLS-DA with SAM analyses in order to obtain a reliable list of MF and CF proteins that significantly contribute to ketamine’s antidepressant effect. This resulted in 61 MF and 95 CF proteins that were statistically significant altered at the 2 h time point when comparing ketamine- to vehicle-treated mice (Supplemental Table S3). Consistent with previous analyses by us we found proteins that are part of major energy metabolism pathways with altered levels after ketamine treatment28. The most significant alterations were detected already 2 h after a single injection of ketamine, in line with the rapid antidepressant effect of ketamine. Subsequent pathway enrichment analyses highlighted altered OXPHOS, ‘purine metabolism’, ‘fructose and mannose metabolism’, ‘proteasome’, ‘protein processing in endoplasmic reticulum’ and ‘ribosome’ (Table 1 ). We found that already at the 2 h time point a low dose of ketamine resulted in increased levels of several OXPHOS proteins including complex I NADH dehydrogenase [ubiquinone] flavoprotein 2 (Ndufv2), cytochrome c1, heme protein (Cyc1), and complex I assembly factor (Tmem126B) and decreased levels of NADH dehydrogenase [ubiquinone] complex I and assembly factor 7 (Ndufaf7) (Supplemental Table S3 and Fig. 4A). Furthermore, Ndufv2, Cyc1, Ndufaf7 and Tmem126B protein levels significantly correlated with floating time in the FST (Fig. 4B). These results are consistent with our previous metabolomics profiling study that had also indicated mitochondrial energy metabolism, specifically OXPHOS, to be affected by ketamine treatment28.
Increased OXPHOS activity can result in higher ROS levels. The phenomenon referred to as oxidative stress causes an imbalance of ROS and the cellular defense system made up of antioxidant molecules and enzymes like peroxiredoxins (Prdxs) that contribute to the antioxidant capacity of the cell37,38,39,40,41. We therefore next analysed ketamine’s effects on oxidative stress. Cytoplasmic Prdx1 and mitochondrial Prdx3 protein levels were significantly lower at the 2 h and 72 h time points, respectively (Fig. 5A, Supplementary Figure 1C). In addition, we found decreased total antioxidant capacity (TAC) 72 h after ketamine injection (Fig. 5B, supplementary Figure 1D). ROS radicals promoting oxidative stress result in cellular damage by reacting with deoxyribonucleic acid (DNA), ribonucleic acid (RNA), lipids, and proteins. Protein modifications by ROS become apparent through the introduction of carbonyl groups into protein side chains. Since ROS are difficult to measure due to their highly reactive nature we assessed their presence indirectly by the analysis of carbonylated (damaged) proteins37. Our results indicate that ketamine results in statistically significant reduction of protein carbonylation 72 h after ketamine treatment (Fig. 5C).
Ketamine treatment reduced the ATP/ADP metabolite ratio at 2 h and 24 h and both ATP and ADP levels significantly correlated with FST floating time at 24 h suggesting an energy deficit. This is consistent with our previous results where we found a significant reduction of ATP levels at 24 h28. Also in line with this is our observation of an increased AMPK activation following ketamine treatment. In the current study we observed several metabolite level and metabolite ratio changes that are part of the ‘citrate cycle’ and ‘glycolysis’ as early as 2 h after ketamine administration, reflecting the fast antidepressant effect of the drug. The ‘citrate cycle’ is connected to the OXPHOS pathway which produces energy in the form of ATP. ‘Glycolysis’ is a major catabolic cytoplasmic pathway where glucose is degraded to pyruvate, which is further metabolized by pyruvate dehydrogenase complex to acetyl-CoA that enters the ‘citrate cycle’. The latter is a mitochondrial matrix pathway that is coupled to the OXPHOS pathway29. We observed metabolite ratio changes that are either due to protein level or protein activity alterations of OXPHOS complex II enzymatic subunit A that processes these metabolites28,42. Our previous time-dependent hippocampal metabolomics profiling analyses indicated statistically significant biomolecular changes that are part of several pathways including the ‘citrate cycle’ and ‘glycolysis’28. Mitochondrial abnormalities including alterations in energy metabolism have been implicated in the pathobiology of affective disorders43,44,45. Taken together, these findings provide further evidence for an important role of mitochondrial energy metabolism in ketamine’s mode of action to generate ATP which compensates for the increased energy demand at activated synapses. We propose that ketamine treatment causes a cellular energy deficit by first promoting anabolic processes (that consume ATP) through the activation of mTORC1. This is counterbalanced in a second phase by the activation of AMPK that promotes catabolic pathways (to generate ATP).
Ketamine affects glutamate neurotransmission and is tightly linked to mitochondrial energy metabolism. Upon release, glutamate is taken up by glial cells and subsequently metabolized to glutamine. In addition, glutamate is metabolized from α-ketoglutarate, a ‘citrate cycle’ metabolite, and can also be recycled through this pathway. Interestingly, glutamate neurotransmission also stimulates glucose utilization and ‘glycolysis’, the upstream pathway of the OXPHOS pathway46. Magnetic resonance spectroscopy analysis of rat medial prefrontal cortex with 13C-labeled glucose demonstrated 13C-enrichment of glutamate, glutamine and GABA upon ketamine treatment, reflecting neurotransmitter release coupled to glycolytic and ‘citrate cycle’ activity. In addition, other studies have shown elevated medial prefrontal cortex metabolism and an involvement of mitochondria in MDD pathobiology43,47,48,49. Mitochondrial respiration in peripheral blood mononuclear cells from depressed patients is lower compared to cells from control subjects and correlates with depressive symptom severity50. Furthermore, mitochondrial disorders are often comorbid with psychiatric disorders including MDD and patients frequently suffer from TRD. These patients have a primary impairment of mitochondrial functioning through either nuclear or mitochondrial DNA mutations. Mitochondrial impairment mostly affects high energy consuming organs like muscles and brain. The mitochondrial disorder is often recognized only later in life of MDD patients51,52,53. Previous studies have connected the antidepressant response with drugs elevating ATP levels. Spectroscopy data have shown decreased NPT levels (mainly ATP) in the brain of depressed patients. Another magnetic resonance spectroscopy study could demonstrate lower NPT levels (mainly ATP) in fluoxetine responders compared with non-responders. Previous data from a study in rats have suggested that the antidepressant-like effect observed after a single injection of ketamine is mediated by an increased anabolic rate-mediating cell growth and differentiation, processes that are highly energy consuming54,55. The high ketamine treatment efficacy for TRD patients might be due to the drug’s beneficial effects on mitochondrial energy metabolism. We submit that mitochondria play a critical role in the development of MDD as well as AD treatment response, at least in a subset of patients with a defined symptomatology.
In order to better understand the biomolecular pathways and cellular mechanisms underlying ketamine’s mode of action, a time-dependent proteomics profiling analysis was also performed. We identified several pathways that were affected and associated with mitochondrial energy metabolism. The PLS-DA showed a time-dependent separation of ketamine- and vehicle-treated mice that was evident by significant protein level alterations and overrepresented pathways. Of the pathways that we found to be affected by ketamine treatment we focused our attention on OXPHOS, in line with results previously obtained by us that had implicated mitochondrial energy metabolism to play a major role in the drug’s mode of action. Ketamine affects the OXPHOS pathway as is evident by several complex I protein levels (Ndufv2, Cyc1, Ndufaf7 and Tmem126B) that correlate with FST floating time in a treatment time-dependent manner.
Taken together, we propose a mechanism by which ketamine stimulates glutamatergic neurotransmission and neuronal activity, ultimately resulting in a long-term potentiation (LTP)-like process. This activation results in the activation of mTORC1. The mTORC1 mediated anabolic processes include protein synthesis and mitochondrial biogenesis within 30 minutes and 1 h upon ketamine treatment. Pretreatment with rapamycin, an mTOR inhibitor, prevents ketamine’s antidepressant-like effect26. Furthermore, other studies indicated that mTOR inhibition decreases mitochondrial respiration, whereas mTOR hyperactivation elevates the expression of oxidative metabolism genes and mitochondrial DNA copy number56,57,58. Hence, ketamine may not only increase protein translation through mTORC1, but also mitochondrial biogenesis. LTP-like processes require energy in the form of ATP followed by AMPK activation, which we found in the present study, in order to promote catabolic processes to generate ATP. Ketamine’s antidepressant-like effects are known to last for one to two weeks. This long-term effect can be explained by the observed LTP-like process. Late phase LTP with spine formation and strengthening of synaptic connections are known to last from days to weeks. Increased OXPHOS activity can result in elevated ROS levels, a phenomenon referred to as oxidative stress, which causes protein, DNA, RNA and lipid modifications and damage. We found that protein carbonylation decreases over time until it reaches significance at the 72 h time point. This can be explained by either decreased ROS levels or alternatively by an activation of a cellular defense system that captures ROS. The cellular defense system includes antioxidant molecules and enzymes such as Prdxs that combat ROS. In addition, a protein quality control system that degrades damaged proteins counteracts oxidative stress38,39,40,41,59,60. We found that total antioxidant capacity (TAC) and cytoplasmic Prdx1 levels were lower at the 2 h and 72 h time points and mitochondrial Prdx3 levels lower at the 72 h time point in ketamine- vs. vehicle-treated mice. The reduced protein carbonylation we observed in ketamine- vs. vehicle-treated mice could be indicative of increased protein quality control processes in the cell. This is supported by our proteomic data that indicate an upregulation of the ‘proteasome’ pathway at the 2 h time point. Taken together, our combined proteomic results suggest that ketamine administration in mice leads to increased degradation of damaged proteins and homeostatic redox adjustment.
The close connection between the glutamatergic system and mitochondrial energy metabolism make a compelling case for the here observed facilitative effects of energy metabolism and the antioxidant defense system following ketamine administration. Further experiments including blocking pathway enzymes and receptors including OXPHOS complex I and AMPK that we found to be involved as well as different ketamine doses are required to confirm these findings. We only assessed ketamine’s behavioral effect using FST floating time and did not carry out any additional behavioral analyses like the learned helplessness, chronic mild stress and novelty suppressed feeding tests that have been previously reported by others. However, using several consecutive behavioral stress assays will in all likelihood affect the animals’ metabolome and proteome and therefore skew the data.
Animals and ketamine treatment
The animals and ketamine treatment were housed and carried out as previously described28. The experiments were performed in accordance with European Communities Council Directive 86/609/EEC. The protocols were approved by the committee for the Care and Use of Laboratory Animals of the Government of Upper Bavaria, Germany.
Forced swim test
The forced swim test and floating time analysis were carried out as previously described28.
Isolation of cytoplasmic protein fraction (CF) and membrane-associated protein fraction (MF)
CF and MF were prepared by repeated tissue homogenization and extraction of non-MF proteins and solubilization of MF proteins with sodium dodecyl sulfate (SDS). Tissues were homogenized for 30 s in 2 M NaCl, 10 mM Hepes/NaOH, pH 7.4, containing 1 mM ethylenediaminetetraacetic acid (EDTA), phosphatase inhibitor cocktail 2 and 3 (Sigma-Aldrich, Munich, Germany), protease inhibitor cocktail tablets ‘cOmplete’ (Roche Diagnostics, Mannheim, Germany), incubated for 10 min and homogenized for 30 s and further with a ultrasonicator for 3 × 10 s on ice. Homogenates were centrifuged at 16100 g at 4 °C for 20 min with the supernatant containing CF proteins. Pellets were rehomogenized in 0.1 M Na2CO3 and 1 mM EDTA containing 1 mM EDTA, phosphatase inhibitor cocktail 2 and 3 (Sigma-Aldrich, Munich, Germany), protease inhibitor cocktail tablets ‘cOmplete’ (Roche Diagnostics, Mannheim, Germany), pH 11.3, mixed at 4 °C for 30 min and collected by centrifugation (16100 g at 4 °C for 20 min). Subsequently, pellets were extracted with 5 M urea, 100 mM NaCl, 10 mM Hepes, pH 7.4 and 1 mM EDTA containing 1 mM EDTA, phosphatase inhibitor cocktail 2 and 3 (Sigma-Aldrich, Munich, Germany), protease inhibitor cocktail tablets ‘cOmplete’ (Roche Diagnostics, Mannheim, Germany) and then washed twice with 0.1 M Tris-HCl, containing 1 mM EDTA, phosphatase inhibitor cocktail 2 and 3 (Sigma-Aldrich, Munich, Germany), protease inhibitor cocktail tablets ‘cOmplete’ (Roche Diagnostics, Mannheim, Germany) pH 7.6. Pellets were solubilized in 2% SDS, 50 mM dithiothreitol (DTT) and 0.1 M Tris-HCl, containing 1 mM EDTA, phosphatase inhibitor cocktail 2 and 3 (Sigma-Aldrich, Munich, Germany), protease inhibitor cocktail tablets ‘cOmplete’ (Roche Diagnostics, Mannheim, Germany) pH 7.6, at 90 °C for 1 min and stored at −20 °C until further analysis.
Hippocampal MF and CF proteins from 8-week-old male C57BL/6 mice treated with ketamine for 2 h, 14 h, 24 h and 72 h were fractionated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot was performed based on standard protocols. After electrophoresis, proteins were transferred to PVDF membranes (Immobilon-P, Millipore, Billerica, USA). Primary antibodies were against adenosinmonophosphate-activated protein kinase (AMPK, Abcam, Cambridge, UK), phosphorylated AMPK (pAMPK, Cell Signaling, Merck, Darmstadt, Germany), peroxiredoxin 1 (Prdx1, Abcam, Cambridge, UK), peroxiredoxin 3 (Prdx3, Abcam, Cambridge, UK). Anti-rabbit, anti-mouse and anti-goat ECL horseradish peroxidase-linked secondary antibodies (GE Healthcare Life Sciences, Little Chalfont, Buckinghamshire, UK) were used. The densitometric analyses were performed with the Image Lab software (Bio-Rad, Munich, Germany).
Protein oxidation analysis
The OxyBlotTM Kit (Merck Millipore, Darmstadt, Germany) was used according to the manufacturer’s instructions.
Total antioxidant capacity (TAC)
The TAC Assay Kit (BioVision, Milpitas, USA) was used according to the manufacturer’s instructions. Absorbance was measured at 570 nm using a plate reader (iMarkTm Microplate Absorbance Reader, Bio-Rad, Munich, Germany).
Protein sample preparation for quantitative proteomics analyses
Hippocampal CF and MF proteins were mixed 1:1 with CF and MF 15N-labeled protein standards, respectively. The latter were derived from C57BL/6 mice that had been fed a 15N-labelled diet (Silantes, Munich, Germany) for 12 weeks. 50 μg of the 14N/15N protein mixture were separated by SDS-PAGE. Separated proteins were stained with Coomassie Brilliant Blue for 20 min and destained overnight. Each gel lane was cut into 16 2.5 mm slices per biological replicate and each slice further cut into smaller pieces.
The gel pieces were then covered with 100 µl of 25 mM NH4HCO3/50% acetonitrile (ACN) for complete destaining and mixed for 10 min at room temperature. The supernatant was discarded and this step repeated twice. Proteins were reduced with 75mL 1x DTT/25 mM NH4HCO3 and incubated at 56 °C for 30 min in the dark. The supernatant was discarded and for alkylation 100 µl iodoacetamide (IAM) was added to the gel pieces and mixed for 30 min at room temperature. The supernatant was discarded and the gel pieces washed twice with 100 µl 25 mM Na4HCO3/50% ACN and incubated for 10 min at room temperature. The supernatant was discarded and gel pieces dried for 20 min at room temperature. Proteins were digested with 50 µl trypsin solution (5ng/µl trypsin/25 mM NH4HCO3) overnight at 37 °C. Peptides were extracted from the gel pieces by incubation in 50 µl 2% formic acid (FA)/50% ACN for 20 min at 37 °C followed by 5 min sonication. This step was repeated twice with 50 µL of 1% FA/50% ACN. The supernatants were then combined and dried (SpeedVac Plus, SC 210 A, Savant, Hyannis, MA, USA). The pellet was stored at −20 °C.
Liquid chromatography tandem mass spectrometry (LC-MS/MS)
Hippocampal MF and CF proteins were identified and quantified with a Dionex Ultimate 3000 RSLC nanoUPLC (Thermo Fisher Scientific, Waltham, USA) online coupled to a QExactiveTM OrbitrapTM mass spectrometer (Thermo Fisher Scientific, Waltham, USA). Separation of peptides was performed by reversed-phase chromatography at a flow rate of 300 nl/min (Thermo Scientific PepMap C18, 2 µm particle size, 100 A pore size, 75 µm × 50 cm). Peptides were loaded onto a pre-column (Thermo Scientific PepMap 100 C18. 5 µm particle size, 100 A pore size, 300 µm × 5 mm) in 0.1% FA for 3 min at a flow rate of 10 µl/min. Chromatography was performed using the following solvents: solvent A water, 0.1% FA, solvent B 80% ACN, 20% water, 0.1% FA. A linear gradient of 2–40% B for 30 min was used. The LC eluant was sprayed into the mass spectrometer by means of an easy-spray source (Thermo Fisher Scientific, Waltham, USA). All m/z values of eluting ions were measured in the Orbitrap mass analyzer, set at a resolution of 70000. Data dependent scans (Top 20) were employed to automatically isolate and generate fragment ions by high energy collisional dissociation in the quadrupole mass analyser and measurement of the resulting fragment ions was performed in the Orbitrap analyzer, set at a resolution of 17500. Peptide ions with charge states of 2+ to 4+ and above were selected for fragmentation.
Orbitrap raw files were converted to mzXML files using MSConvert software. The in-house software package iSPY, which was adapted from an earlier version of a peptide quantitation program known as iTracker, was used to identify and quantify peptides61,62. The software was used to convert mzXML to mgf files that were then imported into Mascot and searched against the SwissProt Mouse database (November 2013) and a decoy database. The search was then performed using TAXONOMY = Mus. The databases were searched using the following settings: variable modifications of carbamidomethyl, oxidation), 20 ppm peptide tolerance, 0.1 Da MS/MS tolerance, 2 missed cleavages and peptide charge states of +2, +3, or +4. In iSPY, Mascot dat output files were run through Percolator for improved identification. Non-unique peptides were discarded. Only peptides with a protein type 1 error of less than 0.01 were kept in the final dataset. The heavy and light peak intensities for each peptide were calculated in iSPY using retention time and sequence information from the MS1 spectra and Mascot search, respectively. Briefly, the intensities for a pre-specified number of isotopomeric peaks were calculated by scanning through a retention time window spanning a set distance on either side of the maximum intensity value. The 14N and 15N peptide isotopic peaks from the MS1 dataset were used to compare the theoretical mass difference between the heavy and light peptides, and the typical isotopic distribution patterns. Only quantifiable peptides for which both heavy and a light peak intensities were identified in five replicates were included in the dataset.
Isolation of polar metabolites from mouse tissue and metabolomics analyses
Statistics and data analyses
Identification of significant metabolite and protein level alterations
Metabolite intensities as well as protein ratios were median-normalised and auto-scaled for statistical analysis. Significant protein level changes 2 h and 24 h after ketamine treatment were identified by multivariate partial least squares-discriminant analyses (PLS-DA) and high-dimensional feature selection significance analysis of microarrays and other omic-datasets (SAM) using MetaboAnalyst64,65. The quality of the PLS-DA models were assessed for R2, Q2 and accuracy values with variable influence of projection (VIP)-score ≥ 1.0 and for SAM with q ≤ 0.10 and false discovery rate (FDR) ≤ 0.1066. We improved robustness of our data analyses and increased confidence in significantly altered metabolites and proteins as well as in the subsequent analyses of overrepresented KEGG pathways by only considering the overlap between the two different statistical methods.
Identification of significantly enriched pathways
Pathway analyses were performed using MetaboAnalyst64 applying a hypergeometric algorithm for overrepresentation analysis and relative-betweeness centrality for pathway topology analysis. Pathways were considered affected if they were significantly enriched for all significantly altered metabolites (PHolm-corrected ≤ 0.05). Proteomic pathway enrichment was assessed by String with an FDR ≤ 0.1067,68.
Calculation of metabolite pair ratios
The metabolite pair ratios were calculated as described28.
All data generated or analysed during this study are included in this published article (and its Supplementary Information files).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
KW was supported by the International Max Planck Research School for Molecular and Cellular Life Sciences. The technical support by Maria Holzapfel is gratefully acknowledged.