Citation:TranslPsychiatry(2013)3,e223;doi:10.1038/tp.2012.142 &2013MacmillanPublishersLimited Allrightsreserved 2158-3188/13 www.nature.com/tp Pharmacometabolomic mapping of early biochemical changes induced by sertraline and placebo RKaddurah-Daouk1,MBBogdanov2,WRWikoff3,HZhu1,SHBoyle1,EChurchill1,ZWang4,AJRush5,RRKrishnan1,5, EPickering6,MDelnomdedieu7andOFiehn3 Inthisstudy,wecharacterizedearlybiochemicalchangesassociatedwithsertralineandplaceboadministrationandchanges associatedwithareductionindepressivesymptomsinpatientswithmajordepressivedisorder(MDD).MDDpatientsreceived sertraline or placebo in a double-blind 4-week trial; baseline, 1 week, and 4 weeks serum samples were profiled using a gas chromatographytimeofflightmassspectrometrymetabolomicsplatform.IntermediatesofTCAandureacycles,fattyacidsand intermediates of lipid biosynthesis, amino acids, sugars and gut-derived metabolites were changed after 1 and 4 weeks of treatment.Someofthechangeswerecommontothesertraline-andplacebo-treatedgroups.Changesafter4weeksoftreatment inbothgroupsweremoreextensive.PathwayanalysisinthesertralinegroupsuggestedaneffectofdrugonABCandsolute transporters, fatty acid receptors and transporters, G signaling molecules and regulation of lipid metabolism. Correlation betweenbiochemicalchangesandtreatmentoutcomesinthesertralinegroupsuggestedastrongassociationwithchangesin levelsofbranchedchainaminoacids(BCAAs),lowerBCAAslevelscorrelatedwithbettertreatmentoutcomes;pathwayanalysis inthisgrouprevealedthatmethionineandtyrosinecorrelatedwithBCAAs.Lowerlevelsoflacticacid,higherlevelsofTCA/urea cycle intermediates, and 3-hydroxybutanoic acid correlated with better treatment outcomes inplacebo group. Results of this studyindicatethatbiochemicalchangesinducedbydrugcontinuetoevolveover4weeksoftreatmentandthatmightexplain partiallydelayedresponse.Responsetodrugandresponsetoplacebosharecommonpathwaysbutsomepathwaysaremore affectedbydrugtreatment.BCAAsseemtobeimplicatedinmechanismsofrecoveryfromadepressedstatefollowingsertraline treatment. TranslationalPsychiatry(2013)3,e223;doi:10.1038/tp.2012.142;publishedonline22January2013 Introduction arenotwelldefined.Responsetocurrenttherapiesintreating Majordepressivedisorder(MDD)isacommonandpotentially MDDvariesconsiderably,with40%ofpatientsnotresponding life-threateningpsychiatricillnessthatisasignificantcauseof and60%notremittingafteraninitialtrialoftherapy.19Boththe disability worldwide.1 Selective serotonin (5-HT) reuptake delayed onset of antidepressant therapeutic action and the inhibitors(SSRIs)arethemajorclassofantidepressantsfor variability in the response prevent clinicians from predicting thetreatmentofMDD.AntidepressanteffectsofSSRIsresult whether an antidepressant is going to work for a particular from inhibition of presynaptic 5-HT transporter, followed patient. The detailed molecular mechanisms underlying by changes in 5-HT neurotransmission and secondary variationintreatmentresponseindepressionandthegenetic changesinotherneurotransmittersystems,suchascatecho- and biochemical basis remain unknown although numerous lamines,2–5neurotransmitteraminoacids6–10andacetylcho- studiesimplicatethenorepinephrineand5-HTsystems.20–23 line.11ItwasshownthatSSRIsinducedelayedtranscriptional The placebo effect adds complexity, as 30–40% of MDD andtranslationalchangesthatmediatemolecularandcellular patientsrespondtoplacebothroughmechanismsthatarenot neuroplasticity.12–14 Different antidepressants, including yetunderstood.24,25Furtherinsightsintothemechanismsof SSRIs,followingtheirchronicadministrationhavebeenfound responsetoplacebovsresponsetomedicationsareneeded toincreasebiomarkersforenergymetabolism.15–18 andanabilitytoidentifyearlyonwhomightbearesponderto The onset of antidepressant therapeutic action typically placebo can stream line clinical trials. Development of does not occur until at least 2–4 weeks of treatment; biomarkers that can predict early on no response should mechanismsunderlyingthisdelayedeffectofantidepressants havegreatclinicalutility. remain largely unknown and the biochemical changes that Metabolomicsenablestheidentificationandquantification happenoverthefirst4weeksthatleadtotherapeuticbenefit of hundreds to thousands of compounds that can report on 1DepartmentofPsychiatryandBehavioralSciences,DukeUniversity,Durham,NC,USA;2DepartmentofNeurologyandNeuroscienceWeillCornellMedicalCollege, NewYork,NY,USA;3UCDavisGenomeCenter,UniversityofCalifornia,Davis,NC,USA;4DepartmentofStatisticsandBioinformaticsResearchCenter,NorthCarolina StateUniversity,Raleigh,NC,USA;5Duke-NUSGraduateMedicalSchool,Singapore;6PfizerGlobalR&D,ClinicalResearchStatistics,Groton,CT,USAand7Pfizer GlobalR&D,NeuroscienceClinicalResearch,Groton,CT,USA Correspondence:DrRKaddurah-Daouk,DukeUniversityMedicalCenter,Box3903,Durham,NC27710,USA. E-mail: rima.kaddurahdaouk.duke.edu or O Fiehn, Metabolomics Research and Core Laboratories UC Davis Genome Center, Room 1314þ1315, First Floor, 451HealthSciDriveDavis,CA95616,USA Keywords:depression;metabotype;metabolomics;pharmacometabolomics;personalizedmedicine;subclassificationofdisease Received3May2012;revised9October2012;accepted15November2012 Earlybiochemicalchangesinducedbysertralineandplacebo RKaddurah-Daouketal 2 changes in biochemical pathways and should provide novel GC-TOF mass spectrometry. Data were collected as insights into the complex metabolic mechanisms of disease described in Trupp et al.33 Briefly, the study design was and drug effects.26,27 Metabolomics provides powerful tools entered into the SetupX database.34 Samples were ali- for characterization of complex phenotypes involving endo- quotted and maintained at (cid:2)801C until use, at which point genous and exogenous mechanisms and for studies of 30ml of samples were extracted and derivatized with complex diseases, such as depression. We have started to methoxyamine and MSTFA to trimethylsilation. Fatty acid applymetabolomicstoolstomapglobalplasmabiochemical methylestersinternalstandardswereincludedinthesample. changesindepressivesubjectsinremissionandinmechan- One microliter of sample was injected using a Gerstel liner isms involved in response to SSRIs. In prior studies we exchange system onto an Agilent 6890 gas chromatograph identifiedperturbationsinlipidmetabolism,dicarboxylicfatty (SantaClara,CA,USA)interfacedtoaLecoPegasusIVtime acids, branched chain and other amino acids, including of flight mass (TOF) spectrometer. Data were processed GABA,aswellasinketonebodiesinplasmaofsubjectswith usingBinBase35andcompoundspectraandretentiontimes MDD.28,29InastudybyJietal.,30weusedapharmacome- were matched against the Fiehn laboratory library or tabolomics-informed pharmacogenomic research strategy NIST05.36 Intensities from all identified metabolites were andidentifiedthatglycineandasingle-nucleotidepolymorph- summed and this value was used for data normalization. ism (SNP) in the glycine dehydrogenase gene were asso- Thesedatawereusedforstatisticalanalysis. ciated with response to escitalopram treatment. Merging pharmacometabolomicswithpharmacogenomicsusing‘1000 Statisticalmethods Genomes’ SNP imputation, we provided additional insights Metabolic signatures of sertraline and placebo (univariate about selective serotonin reuptake inhibitor response.31 We analysis). Paired t-tests were used to determine which also demonstrated that metabolic profiles prior to treatment metabolites changed significantly from baseline to week 1 could help distinguish responders from non-responders to and to week 4. These analyses were conducted separately sertralineandplacebo.32 for the sertraline and placebo groups. A significant t-test In this current study, a hypothesis-generating metabolo- indicates that a metabolite is changing differently in the two micsapproachwasusedtomapglobalbiochemicalchanges groups.Todeterminewhich metaboliteschanged differently inducedby1weekand4weekstreatmentwithsertralineand betweenthesertralineandplacebogroups,wecompared1- placebo. We also correlated biochemical changes and weekchangescoresand4-weekchangescoresbetweenthe treatmentoutcomesat1and4weeksasafirststeptowards two groups using t-tests for independent groups. Metabolite determiningwhichbiochemicalchangescouldbeassociated values were log transformed before data analysis. withthedelayedresponsetosertralinetreatment. Q-values37werecalculatedtocontrolforfalsediscoveryrate. Methods Correlation with response analysis. Pearson correlation analysis was used to examine associations of changes in Study design and patients. Study participants were out- metabolite levels with concurrent changes in depressive patients, 18–65 years of age, from a randomized, double- symptoms. Analyses were conducted separately for 1- and blind, flexible dosing, placebo-controlled study performed at 4-week change. Correlations of the sertraline group and 12 clinical sites. Patients had a primary diagnosis of non- placebo group were compared using the two-sample Z-test psychotic MDD by DSM-IV criteria, with symptoms of after hyperbolic tangent transformation. For these tests, depression present for at least 1 month prior to screening, Q-values37werecalculatedforcontrollingthefalsediscovery and a total baseline score 422 on the 17-item Hamilton rate. Rating Scale for Depression (HAMD )30 at screening. The 17 full exclusion criteria for patients enrolled in the study was Response-pathway oriented analysis. After testing the reported before.32 Patients were randomized to treatment marginaleffectsofmetabolites,wethenaimedtoidentifysets (oraladministration)witheithersertralineorplaceboataratio ofmetaboliteswithgroupeffectsonthedrugresponsethrough of 1:1. Sertraline dosing was started at 50mg per day at the following steps. First,anindividually significant metabolite baseline(week0),withdoseincreasedupto100mgperday waschosenbasedontheaforementionedmarginalcorrelation at week 1 and up to 150mg per day at week 2, as seen analysis; second, a small subset of metabolites of strong needed by the treating clinician. The primary outcome marginal correlations with the preselected metabolite was measure was the HAMD score. HAMD scores were identified; third, a principal component regression (PCR) 17 17 determined and serum samples were collected at baseline analysis was used to regress the drug response on several (before treatment started), 1 week and at 4 weeks after principal components. These principal components are linear treatment;thesampleswerecollectedfromfastingsubjects. combinations of the subset of metabolites and therefore Acompletelistoftheinclusionandexclusioncriteriausedin representthegroupeffectofthesubset. thisstudy,subjectsmedications,havebeendescribedinour previousreport.32Briefly,nopsychotropicmedicationswere Pathway enrichment and network modeling. Pathways and used by the participants. The study protocol was developed networks were analyzed using multiple approaches. Meta- in accordance with the principles of the Declaration of Mapp38 was used to calculate metabolic networks, which Helsinki. All patients provided written informed consent. were displayed using Cytoscape. MetaMapp uses KEGG The study was sponsored and monitored by Pfizer (Groton, reaction pairs to build a metabolic network and then adds CT,USA).Eachsite’sIRBapprovedandoversawthestudy. non-mappedcompoundsviaTanimotosimilarity.39 TranslationalPsychiatry Earlybiochemicalchangesinducedbysertralineandplacebo RKaddurah-Daouketal 3 Results Four-week treatment with sertraline. Biochemical changes following 4-week treatment with sertraline were far more Sample characteristics. The sertraline and placebo extensive than changes seen after 1 week of treatment groups did not differ significantly in age (44±11.4 years (Table 1). Prominent effects were found on levels of fatty vs 40±12.7 years, P¼0.12), gender (29/43 (69%) acids (linoleic, palmitic, arachidonic, oleic, palmitoleic and female vs 32/46 (70%) female, P¼0.83) or race (34/43 heptadecanoicacidsallshoweddecreases)andglycerol,as (79%)whitevs34/46(74%)white,P¼0.57).Theresponse well as changes in metabolites within the TCA and urea rate was slightly higher for sertraline than for placebo, but cycles (decreases in ornithine, citrulline and increase in this difference was not statistically significant (26/43 (60%) aconitic acid). Other changes included metabolites from the vs 23/46 (50%)) responders for sertraline and placebo purine pathway (xanthine decreased), one carbon metabo- groups, respectively. lism (cysteine increased), tryptophan (5-methoxytryptamine decreasedandindole-3-aceticacidincreased)andaketone body (3-hydroxybutanoic acid decreased). With Q-values Metabolicsignaturesofsertralineafter1and4weeksof cutoff controlled at 0.1, significant changes were decreased treatment. A GC-TOF metabolomics platform was used to levels of linoleic, arachidonic and palmitic acids, and of semiquantitatively measure 348 compounds in patient ornithine and glycerol. A metabolic network reconstruction serum. Of these, 143 were known compounds, 17 com- that highlights the metabolites changed after 4-week treat- pounds were identified by spectral matching to the NIST mentwith sertralineisshowninFigure1b. libraryand188wereunknown. Pathway enrichment analysis for 4-week treatment with sertralineisshowninTable2.Severalpathwaysinvolvedin One-week treatment with sertraline. Several metabolites signalingwereenrichedinthisanalysisincludingtransmem- were changed following 1 week of treatment with sertraline; brane transporters of small molecules, signaling by GPCR, they included intermediates of TCA cycle, sugars and fattyacidreceptorsandtransporters. metabolites involved in fatty acid biosynthesis (Supplementary Table 1). All metabolites except glycerol Metabolic signatures of placebo after 1 and 4 weeks of showed an increase in concentration over this period. treatment However, with Q-value cutoff set at 0.1 none of these One-weektreatment. Metabolitesthatchangedafter1-week changes was found to be significant. A metabolic network treatmentwithplaceboareshowninSupplementaryTable1. reconstructionhighlightingthemetabolitesthatchangedafter Changes included amino acids homoserine and aspartate, 1weekofsertraline isshowninFigure1a. one metabolite, xanthine, from the purine pathway, and the Figure1 Compoundschangingsignificantly(Po0.05)inthefirst(a,lefttop)andfourth(b,righttop)weekofsertralinetreatmentup(green)ordown(red)areshownas enlargednodes.Placebotreatmentinthefirstweek(c,leftbottom)andfourthweek(d,rightbottom)isalsoshown.Allcompoundsthatweremeasuredintheexperimentare shown.GraylinesindicatenetworkconnectionsbaseduponstructuralsimilarityandboldbluelinesindicateKEGGreactionpairs. TranslationalPsychiatry Earlybiochemicalchangesinducedbysertralineandplacebo RKaddurah-Daouketal 4 Table1 Metabolicchanges(Po0.05)after4-weektreatmentwithsertralineandplacebo Metabolite Sertraline P Q Metabolite Placebo P Q t-value Percent t-value Percent change change Linoleicacid (cid:2)3.791 (cid:2)20.471 0.001 0.036 Elaidicacid (cid:2)5.259 (cid:2)30.801 5.500E(cid:2)06 0.003 Ornithine (cid:2)3.788 (cid:2)18.435 0.001 0.036 Fructose 4.213 487.726 1.443E(cid:2)04 0.014 Glycerol (cid:2)3.597 (cid:2)15.845 0.001 0.052 Hippuricacid 4.389 331.668 8.420E(cid:2)05 0.014 Arachidonicacidisomer (cid:2)3.306 (cid:2)16.947 0.002 0.088 Linoleicacid (cid:2)4.175 (cid:2)22.794 1.617E(cid:2)04 0.014 Palmiticacid (cid:2)3.444 (cid:2)14.788 0.002 0.072 Oleicacid (cid:2)4.206 (cid:2)21.281 1.472E(cid:2)04 0.014 Arachidonicacid (cid:2)3.123 (cid:2)15.713 0.004 0.113 Palmitoleicacid (cid:2)4.520 (cid:2)24.734 5.620E(cid:2)05 0.014 Hippuricacid 3.031 268.553 0.005 0.115 Methioninesulfoxide 3.082 77.910 0.004 0.113 Oleicacid (cid:2)3.024 (cid:2)0.299 0.005 0.115 Propane-1,3-diolNIST (cid:2)3.099 (cid:2)11.241 0.004 0.113 Aconiticacid 2.912 26.214 0.006 0.122 a-Ketoglutaricacid (cid:2)2.997 (cid:2)11.514 0.005 0.115 Cystine 2.667 327.080 0.012 0.161 Palmiticacid (cid:2)2.978 (cid:2)10.948 0.005 0.115 Palmitoleicacid (cid:2)2.597 131.426 0.014 0.172 Homoserine (cid:2)2.879 (cid:2)4.833 0.006 0.122 5-Methoxytryptamine (cid:2)2.530 425.326 0.016 0.184 Inulobiose 2.835 88.022 0.007 0.127 Xanthine (cid:2)2.538 (cid:2)8.746 0.016 0.184 Myristicacid (cid:2)2.843 (cid:2)9.520 0.007 0.127 Myristicacid (cid:2)2.480 (cid:2)8.007 0.018 0.190 Cytidine-50-diphosphate (cid:2)2.590 1.726 0.013 0.172 Indole-3-acetate 2.462 21.503 0.019 0.191 Glycerol (cid:2)2.535 (cid:2)7.201 0.015 0.183 Heptadecanoicacid (cid:2)2.358 (cid:2)7.089 0.024 0.201 Arabitol 2.479 15.601 0.018 0.190 3-Hydroxybutanoicacid (cid:2)2.345 33.418 0.025 0.201 Creatinine 2.473 148.690 0.018 0.190 Xylitol 2.197 163.060 0.035 0.228 Hydroxylamine (cid:2)2.481 (cid:2)8.192 0.018 0.190 HydroxycarbamateNIST 2.152 298.069 0.039 0.228 2-Deoxytetronicacid 2.434 31.055 0.020 0.193 Asparagines 2.132 53.085 0.040 0.228 Histidine 2.393 59.236 0.022 0.200 Uricacid (cid:2)2.111 (cid:2)6.143 0.042 0.230 Proline 2.375 30.962 0.023 0.201 Succinicacid (cid:2)2.067 (cid:2)5.740 0.046 0.235 Glutamine 2.330 82.385 0.025 0.201 Maltose 2.302 171.497 0.027 0.204 3-Hydroxybutanoicacid (cid:2)2.245 9.976 0.031 0.213 Pseudouridine (cid:2)2.158 (cid:2)5.284 0.037 0.228 2-DeoxyerythritolNIST 2.131 133.213 0.039 0.228 GABA 2.133 139.734 0.039 0.228 Glutamicacid 2.098 1264.390 0.042 0.230 Sucrose 2.107 597.169 0.042 0.230 Hydrocinnamicacid (cid:2)2.089 (cid:2)6.049 0.043 0.230 Xanthine (cid:2)2.092 (cid:2)5.011 0.043 0.230 Quinicacid 2.070 210.924 0.045 0.233 Oxoproline (cid:2)2.050 (cid:2)5.221 0.047 0.235 Benzylalcohol (cid:2)2.043 17.474 0.048 0.235 SignificantP-valuesareinbold. Table2 Pathwayenrichmentanalysisoftheeffectofsertralineexposurefrombaselinetoweek4 Pathwayname PT OV Metabolites P Q Transmembranetransportofsmallmolecules R 9 170(192) 1.21E(cid:2)06 1.51E(cid:2)03 SignalingbyGPCR R 7 84(93) 1.30E(cid:2)06 1.51E(cid:2)03 Signaltransduction R 7 89(98) 1.93E(cid:2)06 1.51E(cid:2)03 RegulationoflipidmetabolismbyPPAR-a R 3 5(5) 3.60E(cid:2)06 2.12E(cid:2)03 SLC-mediatedtransmembranetransport R 8 147(165) 4.85E(cid:2)06 2.28E(cid:2)03 Freefattyacidreceptors R 4 19(23) 8.34E(cid:2)06 3.27E(cid:2)03 Transportoffattyacids R 3 9(9) 2.97E(cid:2)05 9.96E(cid:2)03 GPCRligandbinding R 5 66(75) 9.26E(cid:2)05 0.027 Ga(q)signalingevents R 4 38(43) 1.45E(cid:2)04 0.038 Nucleotidemetabolism W 3 16(17) 1.91E(cid:2)04 0.045 Fattyacid,triacylglycerolandketonebodymetabolism R 5 83(89) 2.78E(cid:2)04 0.059 Glucosehomeostasis W 3 21(21) 4.44E(cid:2)04 0.087 ArachidonateproductionfromDAG R 2 5(5) 5.32E(cid:2)04 0.096 ClassA/1(Rhodopsin-likereceptors) R 4 56(64) 6.62E(cid:2)04 0.111 Abbreviations:PCR,principalcomponentregression;PT,sourceofpathway;R,Reactome;W,Wikipathways;OV,numberofoverlappingmetabolites;P,P-value calculatedformetabolites;Q,Q-valuecalculatedformetabolitesaftercorrectionforFDR. fatty acids linoleic, oleic and palmitoleic acids. Each treatment with placebo, compared with 1 week (Table 1). metabolite showed a concentration decrease. However, Reductionswerefoundinlevelsoffattyacids(linoleic,oleic, similarly to 1-week treatment with sertraline, none of these myristic and palmitoleic acids) and glycerol. Other changes changeswassignificantwithQ-valuecutoff of0.1. included TCA cycle metabolites (decreased a-ketoglutarate, and increased glutamine and glutamic acid), amino acids Four-week treatment with placebo. Significantly more pro- (decreased homoserine, and increased proline, GABA and found biochemical changes were observed after 4 weeks of glutamic acid), metabolites linked to one carbon cycle TranslationalPsychiatry Earlybiochemicalchangesinducedbysertralineandplacebo RKaddurah-Daouketal 5 (increased creatinine and decreased homoserine) Table3 Correlationswithtreatmentoutcomes:correlationsbetweenbiochem- (Figure 1d). Both xanthine from the purine pathway, and icalchangesandchangesinHam-D the ketone body 3-hydroxybutanoic acid were decreased. Metabolite Cor P-value With Q-values cutoff controlled at 0.1, significant changes werefoundforelaidic,hippuric,linoleic,oleicandpalmitoleic (a)One-weeksertraline 5-Methoxytryptamine 0.405 0.016 acidsandfructose. Arachidonicacidisom (cid:2)0.398 0.018 Ribose 0.389 0.021 Associations of changes in metabolites with treatment Cystine 0.370 0.029 outcomes Trehalose 0.358 0.035 a-Ketoglutaricacid (cid:2)0.339 0.047 One-week treatment with sertraline. The associations Xylose (cid:2)0.335 0.049 between changes in individual metabolites and treatment outcomewerealsoexamined.Correlationsbetweenchanges (b)One-weekplacebo Phosphoricacid 0.510 0.001 inindividualmetabolitesandtreatmentoutcomeasassessed 4-Hydroxyproline (cid:2)0.439 0.005 using HAMD17 scores are presented in Table 3a. Decrease Cholesterol 0.397 0.011 in 5-methoxytryptamine (5-MTPM) levels was the most Malate (cid:2)0.323 0.042 significant in association with symptoms reduction following (c)Four-weeksertraline 1weekoftreatmentwithsertraline.Symptomsreductionwas Valine 0.507 0.002 also associated with decrease in levels of ribose, trehalose Leucine 0.478 0.004 and cystine. Increases in the levels of arachidonic and a- Lacticacid (cid:2)0.435 0.009 Pseudouridine (cid:2)0.409 0.015 ketoglutaricacidswereassociatedwithsymptomsreduction Conduritol-b-epoxi (cid:2)0.406 0.015 after1-weeksertraline treatment. Isoleucine 0.382 0.024 MannonicacidNIST 0.380 0.024 Cystine 0.362 0.032 Four-week treatment with sertraline. After 4 weeks of Inulobiose (cid:2)0.361 0.033 sertralinetreatment,themostsignificantchangesassociated with symptoms reduction were decreases in the levels of (d)Four-weekplacebo branched chain amino acids (BCAAs), valine, leucine and Lacticacid 0.383 0.015 Ribose6 (cid:2)0.371 0.018 isoleucine (Table 3c). Similarly to 1 week, at 4 weeks of 3-Hydroxybutanoicacid (cid:2)0.355 0.025 sertraline treatment symptoms reduction were associated Oxalicacid 0.346 0.029 with increase in levels of cysteine. Improvement of the HydroxycarbamateNIST 0.345 0.029 Citrulline (cid:2)0.336 0.034 symptoms after 4 weeks on sertraline was also associated Indole-3-acetate 0.323 0.042 withincreasesinthelevelsoflacticacidandpseudouridine. 1-Monostearin 0.318 0.046 One-weektreatmentwithplacebo. After1weekoftreatment Negative correlations indicate better response with increase in metabolite; with placebo, decrease in phosphoric acid and increases in positiveindicatesbetterresponsewithdecreaseinmetabolite. 4-hydroxyprolineandmalate(Table3b)wereassociatedwith symptoms reduction. Changes in several metabolites within TCA/urea cycle also correlated with changes in depression acids is warranted in studying biological mechanisms of scores. sertraline. Four-week treatment with placebo. Changes in lactic acid, Discussion sugars, oxalic acid, ketone body, indole-3-acetic acid, urea cycleintermediatecirtrulineandalaninewereassociatedwith Recently, weconductedametabolomicsstudytodetermine symptoms reduction at 4 weeks of placebo treatment whetherthebaselinemetabolicprofile(thatis,metabotype)of (Table 3d). In contrast to the sertraline findings, no a patient with MDD would allow a prediction of patient’s associationbetweenchangesinlevelsofBCAAswasnoted. response to treatment with sertraline or placebo.32 In that study,utilizinganelectrochemistry-basedanalyticalplatform Response-pathway oriented analysis. Since valine was that enables quantification of redox active compounds, we foundtobethemostsignificant metaboliteshowingcorrela- foundthatmetabolitesfromphenylalanine,tryptophan,purine tionwithresponseinthesertralinegroup,weselectedalistof and tocopherol pathways enabled separation of responders 26 metabolites (Supplementary Table 2) marginally corre- and non-responders to treatment with the SSRI or placebo. lated with valine in sertraline group including leucine, Thecurrentstudyextendsonthesefindingsby(1)analyzing isoleucine and tyrosine. A PCR analysis with variable differentclassesofmetabolitesusingGC-TOFmassspectro- selection identified four principal components jointly and metry analytical platform; (2) identifying the biochemical significantly associated with drug response. In the first changes that occur in MDD patients 1 week and 4 weeks principal component (Supplementary Table 3), isoleucine, aftertreatmentwithsertralineorplacebo,and(3)identifying leucine, methionine, valine and tyrosine possessed the whichchangesareassociatedwithareductionindepressive largest weights. Thus, this metabolite set characterized by symptomsafter1and4weeksoftreatmentwithsertralineor major amino acids explored the possible relationship placebo. between amino acids, their transport and function, and Severalmetabolitesfrommultiplepathwayswerechanged depression, suggesting that further investigation of amino insertraline-andplacebo-treatedsubjects.Inbothsertraline- TranslationalPsychiatry Earlybiochemicalchangesinducedbysertralineandplacebo RKaddurah-Daouketal 6 and placebo-treated patients, the larger effects were glycolysisinmice,52and5-HThadstimulatoryandinhibitory observedafter4weeksoftreatmentcomparedwith1week. effects on glycogen synthesis in hepatocytes mediated by Multiple mechanisms could account for the biochemical 5-HT1/2Aand5-HT2Breceptors,respectively.53 changes observed in this study. Progression of the disease We also analyzed associations between changes in itself,aswellasbehavioralchanges(forexample,changesin individual metabolites and treatment outcome. After 1 week diet) associated with improvement of the symptoms after ofsertralinetreatment,thedecreasein5-methoxytryptamine treatment,couldexplainsomeoftheeffectssuchaschanges andincreasesinarachidonicacidanda-ketoglutaratelevels infattyacidsandindicesofenergymetabolism.Thus,itwas were the most significant in association with symptoms suggestedthatplasmalevelsoflinoleicacidmayserveasa reduction.Associationofdecreasein5-MTPMwithsymptoms possible biomarker for residual depression and anxiety in reductionsuggestspotentialroleofmethoxyindolebranchof older people with previous depression;40 it was also shown the tryptophan pathway in recovery form depressed state; earlier that increased dietary intake of oleic acid was currently we are investigating different branches of the associated with reduced risk of severe depressed mood pathway in the effects of SSRIs. Changes in the levels of amongwomenwhileincreaseddietaryintakeoflinoleicacid arachidoniclevelsmostlikelyreflectincreaseintheactivityin wasassociatedwithincreasedriskofseveredepressedmood phospholipase A2, resulting from increased 5-HT levels at amongmen.41Itwasrecentlyfoundthatomega-3augmenta- earlier time point.44,45 Association of increased levels of a- tion of SSRI citalopram treatment produced a significantly ketoglutaratewithsymptomsreductionisnotclear;however, greaterreductioninHAM-Dscorescomparedwithcitalopram it may be related to the activation of energy metabolism treatment alone.42 In a yeast model, sertraline has been followingSSRItreatment.Insupportofthismechanismisthe shown to inhibit phospholipases A1 and D and activates observation that improvement of symptoms at 4 weeks of phospholipase A2.43 It is thus possible, if the effect on treatment with sertraline was associated with decreased phospholipases is similar in mammalian systems, that the levelsoflacticacid.Themostsignificantchangesassociated changes in fatty acids observed in this study could be the with symptoms reduction after 4 weeks of treatment with result of sertraline inhibition and/or activation of specific sertraline were decreases in the levels of BCAAs, valine, phospholipases. It was also shown in primary cultures of leucine and isoleucine. In our previous study, we have mouseastrocytes thatchronic SSRIstreatment upregulates demonstrated that higher levels of leucine and isoleucine calcium-dependentphospholipaseA2(PLA2),whichreleases were associated with depression.29 BCAAs have well- arachidonic acid from the membrane-bound phospholi- established anabolic effects on protein metabolism that pids.44,45 Additional mechanism, involving PLA2 activation involves activation of mTOR pathway. In this respect, it is and increased release of arachidonic acid following SSRIs particularly interesting that antidepressant effect of NMDA could involve membrane phospholipids remodeling and receptorantagonistketamineresultsfromactivationofmTOR changes in signal transduction pathways.46,47 The changes pathway involving BCAAs.54 In a recent study, Webhofer noted in fatty acids in the placebo group are of interest and etal.18foundthatinmicevaline,leucineandisoleucinelevels couldreflectchangesinfattyacidbiosynthesisordegradation wereincreasedafterchronictreatmentwithSSRIparoxetine. asdisease progresses. It can alsobe common towhat was Detailed mapping of effects of SSRIs, as well as of other observedinthesertralinegroup. antidepressants, on metabolism in animals and in humans Several studies implicated impairment of energy metabo- should be compared at a more detailed level to enable lismpathwaysinbothpathogenesisofdepressionandeffects translationalresearch. ofantidepressants.Inananimalmodelofdepression,levels Of interest, there were several metabolites that closely of aconitate hydratase and pyruvate dehydrogenase were correlated with changes in BCAAs and these included found to be regulated in opposite directions by stress and methionine and tyrosine. Methionine and the methionine antidepressant drug treatment, thus representing molecular cycle are closely linked to the folate cycle, to one carbon correlatesofbothvulnerabilitytodepressionandresponseto metabolism and to methylation processes that have been drug.48 Chronic administration of different antidepressants, implicatedindepression. including SSRIs, was associated with increased levels of While there is strong evidence that MDD is influenced by biomarkers for energy metabolism, such as activity of brain inheritance,55 GWA studies for possible genetic factors that creatinekinase,ofmitochondrialrespiratorychaincomplexes contribute to MDD risk, as well as GWA studies for genes IIandIV,andofcitratesynthaseandsuccinatedehydrogen- associated with SSRI response in MDD patients, have ase.17,49,50Recentlyinananimalmodelsignificantchangesin provided somewhat disappointing results.56,57 However, it amino-acidmetabolismrelatedtoenergymetabolismviaTCA wasdemonstratedthattheSNPintheABCB1geneencoding cycle,andincreasedlevelsofglucose,glucose-6-phosphate ATP-binding cassette, subfamily B (MDR/TAP), member 1 andfructose-6-phosphatefollowingchronicparoxetinetreat- (ABCB1)transporterproteinexpressedintheendothelialcells ment were found.18 Additional mechanisms related to the oftheblood–brainbarrier,isassociatedwithseverityofMDD observedchangesinenergymetabolismfollowingsertraline and its response to citalopram and escitaloppram.58,59 treatmentcouldberelatedtotheeffectsof5-HTonglucose Recently,inGWAstudyKohlietal.60identifiedtheSNPina metabolism and glycolysis. It was demonstrated that 5-HT gene encoding a neuron-specific neutral amino-acid trans- stimulates glucose uptake in skeletal muscle and activates porter, SLC6A15, as an MDD risk gene. While the exact skeletalmuscle6-phosphofructo-1-kinaseactivityinadose- function of the SLC6A15 SNP is not well understood, from responsethrough5-HT2Areceptors.51Recently,itwasfound both human and animal studies, it is known that SLC6A15 that 5-HT through 5-HT2A receptors controls hepatic belongstothesolutecarrier6(SLC6)genefamily,whichalso TranslationalPsychiatry Earlybiochemicalchangesinducedbysertralineandplacebo RKaddurah-Daouketal 7 includesthetransportersformonoamines(includingserotonin 2. PriscoS,EspositoE.Differentialeffectsofacuteandchronicfluoxetineadministrationon transporter,theprimarytargetforSSRIs)andneurotransmit- thespontaneousactivityofdopaminergicneuronesintheventraltegmentalarea.BrJ Pharmacol1995;116:1923–1931. ter amino acids, as well as a sodium-dependent transporter 3. BymasterFP,ZhangW,CarterPA,ShawJ,ChernetE,PhebusLetal.Fluoxetine,butnot for neutral amino acids, including BCAAs.61 Earlier it was other selective serotonin uptake inhibitors, increases norepinephrine and dopamine shown that in the SLC6A15 knockout mice, leucine uptake extracellularlevelsinprefrontalcortex.Psychopharmacology(Berl)2002;160:353–361. into cortical synaptosomes is significantly reduced.62 Zhou 4. DremencovE,ElMansariM,BlierP.Effectsofsustainedserotoninreuptakeinhibitionon et al.63 demonstrated that SSRIs sertraline and fluoxetine thefiringofdopamineneuronsintheratventraltegmentalarea.JPsychiatryNeurosci 2009;34:223–229. have high affinity for the prokaryotic leucine transporter 5. KitaichiY,InoueT,NakagawaS,BokuS,KakutaA,IzumiTetal.Sertralineincreases homolog (LeuT)of SLC6A15. Interestingly, SSRIs also bind extracellularlevelsnotonlyofserotonin,butalsoofdopamineinthenucleusaccumbens andstriatumofrats.EurJPharmacol2010;647:90–96. to the homologous norepinephrine and dopamine transpor- 6. GolembiowskaK,DziubinaA.Effectofacuteandchronicadministrationofcitalopramon ters, although with much lower affinity thanto theirprincipal glutamateandaspartatereleaseintheratprefrontalcortex.PolishJPharmacol2000;52: targetserotonintransporter,64,65andonlyoneortwodifferent 441–448. functional group substitutions are sufficient to convert an 7. Wang SJ, Su CF, Kuo YH. Fluoxetine depresses glutamate exocytosis in the rat cerebrocorticalnerveterminals(synaptosomes)viainhibitionofP/Q-typeCa2þchannels. SSRI into a norepinephrine reuptake inhibitor with higher Synapse2003;48:170–177. affinitytonorepinephrinetransporter.66,67Therefore,itisnot 8. BhagwagarZ,WylezinskaM,TaylorM,JezzardP,MatthewsPM,CowenPJ.Increased unlikely that observed effect of sertraline on BCAAs could brainGABAconcentrationsfollowingacuteadministrationofaselectiveserotoninreuptake inhibitor.AmJPsychiatry2004;161:368–370. resultfromtheSLC6A15SNPindepressedpatients.Allthese 9. MarstellerDA,Barbarich-MarstellerNC,PatelVD,DeweySL.Brainmetabolicchanges findings and the association between the improvement of following4-weekcitalopraminfusion:increased18FDGuptakeandgamma-aminobutyric depression and changes in BCAAs levels observed in our acidlevels.Synapse2007;61:877–881. studystrongly suggest potential role of BCAA in depression 10. SchipkeCG,HeuserI,PetersO.Antidepressantsactonglialcells:SSRIsandserotonin elicitastrocytecalciumsignalinginthemouseprefrontalcortex.JPsychiatrRes2011;45: and in the effects of SSRI; further studies are needed to 242–248. examinepotentialroleofBCAAsindepressionandtheeffects 11. MineurYS,PicciottoMR.Nicotinereceptorsanddepression:revisitingandrevisingthe of different antidepressants, in both patients and animal cholinergichypothesis.TrendsPharmacolSci2010;31:580–586. 12. MartinowichK,LuB.InteractionbetweenBDNFandserotonin:roleinmooddisorders. models. Neuropsychopharmacology2008;33:73–83. One of the limitations of this study is that despite 13. RantamakiT,HendolinP,KankaanpaaA,MijatovicJ,PiepponenP,DomeniciEetal. comprehensive biochemical analysis it included relatively Pharmacologically diverse antidepressants rapidly activate brain-derived neurotrophic small number of subjects. However, this study should be factorreceptorTrkBandinducephospholipase-Cgammasignalingpathwaysinmouse brain.Neuropsychopharmacology2007;32:2152–2162. viewedashypotheses-generatingtotestwhethermetabolo- 14. VidalR,Pilar-CuellarF,dosAnjosS,LingeR,TrecenoB,VargasVIetal.Newstrategiesin mics approach could provide insights into pathways impli- thedevelopmentofantidepressants:towardsthemodulationofneuroplasticitypathways. cated in SSRI and placebo response and mechanism of CurrPharmDes2011;17:521–533. 15. AgostinhoFR,ReusGZ,StringariRB,RibeiroKF,FerraroAK,BenedetJetal.Treatment delayedresponse.Replicationandvalidationofthesefindings witholanzapine,fluoxetineandolanzapine/fluoxetinealterscitratesynthaseactivityinrat inotherclinicaltrialsisimportantforfuturestudies,however, brain.NeurosciLett2011;487:278–281. this is outside the scope of this first report. Whether and to 16. AgostinhoFR,ScainiG,FerreiraGK,JeremiasIC,ReusGZ,RezinGTetal.Effectsof what extent peripheral biochemical changes reflect CNS olanzapine,fluoxetineandolanzapine/fluoxetineoncreatinekinaseactivityinratbrain. BrainResBull2009;80:337–340. (centralnervoussystem)changesrelatedtomechanismsof 17. SantosPM,ScainiG,RezinGT,BenedetJ,RochiN,JeremiasGCetal.Braincreatine depression and drug effects also should be addressed in kinaseactivityisincreasedbychronicadministrationofparoxetine.BrainResBull2009; futurestudies.Depressionisadiseasethataffectsbothbrain 80:327–330. 18. WebhoferC,GormannsP,TolstikovV,ZieglgansbergerW,SillaberI,HolsboerFetal. andalsotherestofthebody,asiswellknownfromthecortisol Metabolite profiling of antidepressant drug action reveals novel drug targets beyond data;68,69therefore,itispossiblethattheperipheralchanges monoamineelevation.TranslPsychiatry2011;1:e58. reflectbothdirectCNSandadditionalCNS-linkedperipheral 19. Trivedi MH. Major depressive disorder: remission of associated symptoms. J Clin effectsandmaybedirectperipheraleffectsalso.Irrespective Psychiatry2006;67(Suppl6):27–32. 20. MalhiGS,ParkerGB,GreenwoodJ.Structuralandfunctionalmodelsofdepression:from of the exact mechanism that leads to the changes and to sub-typestosubstrates.ActaPsychiatrScand2005;111:94–105. response, the presence of consistent changes can be an 21. Ressler KJ, Nemeroff CB. Role of serotonergic and noradrenergic systems in the importantsignaturethatcouldbethebasisoffurtheriterative pathophysiologyofdepressionandanxietydisorders.DepressAnxiety2000;12(Suppl1): 2–19. understandingfrombothanimalandhumanstudies. 22. InselTR,WangPS.TheSTAR*Dtrial:revealingtheneedforbettertreatments.Psychiatr In conclusion, our data for the first time illustrate that Serv2009;60:1466–1467. biochemical changes following SSRI treatment are progres- 23. KatzMM,BowdenCL,FrazerA.Rethinkingdepressionandtheactionsofantidepressants: sive and some of the pathways that are modified in the first uncoveringthelinksbetweentheneuralandbehavioralelements.JAffectDisord2010; 120:16–23. weekmightleadtosubsequentchangesnotedinthefollowing 24. KirschI.Challengingreceivedwisdom:antidepressantsandtheplaceboeffect.McGillJ weeks.Thepresenceofconsistentchangescouldprovidean Med2008;11:219–222. opportunity for developing biomarkers that can predict 25. Walsh BT, Seidman SN, Sysko R, Gould M. Placebo response in studies of major depression:variable,substantial,andgrowing.JAMA2002;287:1840–1847. responseornoresponsetoantidepressanttreatment. 26. Kaddurah-Daouk R, Krishnan KR. Metabolomics: a global biochemical approach tothestudyofcentralnervoussystemdiseases.Neuropsychopharmacology2009;34: 173–186. ConflictofInterest 27. Kaddurah-DaoukR,KristalBS,WeinshilboumRM.Metabolomics:aglobalbiochemical approachtodrugresponseanddisease.AnnuRevPharmacolToxicol2008;48:653–683. Theauthorsdeclarenoconflictofinterest. 28. PaigeLA,MitchellMW,KrishnanKR,Kaddurah-DaoukR,SteffensDC.Apreliminary metabolomicanalysisofolderadultswithandwithoutdepression.IntJGeriatrPsychiatry 2007;22:418–423. 1. Kessler RC, Berglund P, Demler O, Jin R, Koretz D, Merikangas KR et al. The 29. Steffens DC, Wei J, Krishnan KR, Karoly ED, Mitchell MW, O’Connor CM et al. epidemiologyofmajordepressivedisorder:resultsfromtheNationalComorbiditySurvey Metabolomic differences in heart failure patients with and without major depression. Replication(NCS-R).JAMA2003;289:3095–3105. JGeriatrPsychiatryNeurol2010;23:138–146. TranslationalPsychiatry Earlybiochemicalchangesinducedbysertralineandplacebo RKaddurah-Daouketal 8 30. JiY,HebbringS,ZhuH,JenkinsGD,BiernackaJ,SnyderKetal.Glycineandaglycine 51. Coelho WS, Costa KC, Sola-Penna M. Serotonin stimulates mouse skeletal muscle dehydrogenase (GLDC) SNP as citalopram/escitalopram response biomarkers in 6-phosphofructo-1-kinase through tyrosine-phosphorylation of the enzyme altering its depression:pharmacometabolomics-informedpharmacogenomics.ClinPharmacolTher intracellularlocalization.MolGenetMetab2007;92:364–370. 2011;89:97–104. 52. CoelhoWS,DaSilvaD,Marinho-CarvalhoMM,Sola-PennaM.Serotoninmodulates 31. AboR,HebbringS,JiY,ZhuH,ZengZB,BatzlerAetal.Mergingpharmacometabolomics hepatic6-phosphofructo-1-kinaseinaninsulinsynergisticmanner.IntJBiochemCellBiol withpharmacogenomicsusing‘1000Genomes’single-nucleotidepolymorphismimputa- 2012;44:150–157. tion:selectiveserotoninreuptakeinhibitorresponsepharmacogenomics.Pharmacogenet 53. TudhopeSJ,WangCC,PetrieJL,PottsL,MalcomsonF,KieswichJetal.Anovel Genomics2012;22:247–253. mechanism for regulating hepatic glycogen synthesis involving serotonin and cyclin- 32. Kaddurah-DaoukR,BoyleS,MatsonW,SharmaS,MatsonS,ZhuHetal.Pretreatment dependentkinase-5.Diabetes2012;61:49–60. metabotypeasapredictorofresponsetosertralineorplaceboindepressedoutpatients:a 54. LiN,LeeB,LiuRJ,BanasrM,DwyerJM,IwataMetal.mTOR-dependentsynapse proofofconcept.TranslPsychiatry2011;32:1–7. formationunderliestherapidantidepressanteffectsofNMDAantagonists.Science2010; 33. TruppM,ZhuH,WikoffWR,BaillieRA,ZengZB,KarpPDetal.Metabolomicsreveals 329:959–964. aminoacidscontributetovariationinresponsetosimvastatintreatment.PLoSOne2012; 55. Baker M, Dorzab J, Winokur G, Cadoret R. Depressive disease. Evidence favoring 7:9. polygenicinheritancebasedonananalysisofancestralcases.ArchGenPsychiatry1972; 34. ScholzM,FiehnO.SetupX–apublicstudydesigndatabaseformetabolomicprojects.Pac 27:320–327. SympBiocomput2007169–180. 56. Psychiatric GCCC, Cichon S, Craddock N, Daly M, Faraone SV, Gejman PV et al. 35. SkogersonK,WohlgemuthG,BarupalDK,FiehnO.ThevolatilecompoundBinBasemass Genomewide association studies: history, rationale, and prospects for psychiatric spectraldatabase.BMCBioinformatics2011;12:321. disorders.AmJPsychiatry2009;166:540–556. 36. KindT,WohlgemuthG,LeedoY,LuY,PalazogluM,ShahbazSetal.FiehnLib:mass 57. WrayNR, PergadiaML, Blackwood DH, Penninx BW,Gordon SD, NyholtDR etal. spectralandretentionindexlibrariesformetabolomicsbasedonquadrupoleandtime-of- Genome-wideassociationstudyofmajordepressivedisorder:newresults,meta-analysis, flightgaschromatography/massspectrometry.AnalChem2009;81:10038–10048. andlessonslearned.MolPsychiatry2012;17:36–48. 37. StoreyJD,TibshiraniR.Statisticalmethodsforidentifyingdifferentiallyexpressedgenesin 58. LeeSH,LeeMS,LeeJH,KimSW,KangRH,ChoiMJetal.MRP1polymorphisms DNAmicroarrays.MethodsMolBiol2003;224:149–157. associated with citalopram response in patients with major depression. J Clin 38. HartmanAL,LoughDM,BarupalDK,FiehnO,FishbeinT,ZasloffMetal.Humangut Psychopharmacol2010;30:116–125. microbiomeadoptsanalternativestatefollowingsmallboweltransplantation.ProcNatl 59. LinKM,ChiuYF,TsaiIJ,ChenCH,ShenWW,LiuSCetal.ABCB1genepolymorphisms AcadSciUSA2009;106:17187–17192. are associated with the severity of major depressive disorder and its response to 39. KarnovskyA,WeymouthT,HullT,TarceaVG,ScardoniG,LaudannaCetal.Metscape2 escitalopramtreatment.PharmacogenetGenomics2011;21:163–170. bioinformaticstoolfortheanalysisandvisualizationofmetabolomicsandgeneexpression 60. KohliMA,LucaeS,SaemannPG,SchmidtMV,DemirkanA,HekKetal.Theneuronal data.Bioinformatics2012;28:373–380. transportergeneSLC6A15confersrisktomajordepression.Neuron2011;70:252–265. 40. Jadoon A, Chiu CC, McDermott L, Cunningham P, Frangou S, Chang CJ et al. 61. BroerS.TheSLC6orphansareformingafamilyofaminoacidtransporters.NeurochemInt Associationsofpolyunsaturatedfattyacidswithresidualdepressionoranxietyinolder 2006;48:559–567. peoplewithmajordepression.JAffectDisord2012;136:918–925. 62. DrgonovaJ,LiuQR,HallFS,KriegerRM,UhlGR.Deletionofv7-3(SLC6A15)transporter 41. WolfeAR,OgbonnaEM,LimS,LiY,ZhangJ.Dietarylinoleicandoleicfattyacidsin allows assessment of its roles in synaptosomal proline uptake, leucine uptake and relationtoseveredepressedmood:10yearsfollow-upofanationalcohort.ProgNeuro- behaviors.BrainRes2007;1183:10–20. PsychBiolPsych2009;33:972–977. 63. ZhouZ,ZhenJ,KarpowichNK,LawCJ,ReithME,WangDN.Antidepressantspecificityof 42. GertsikL,PolandRE,BreseeC,RapaportMH.Omega-3fattyacidaugmentationof serotonintransportersuggestedbythreeLeuT-SSRIstructures.NatStructMolBiol2009; citalopramtreatmentforpatientswithmajordepressivedisorder.JClinPsychopharmacol 16:652–657. 2012;32:61–64. 64. EshlemanAJ,CarmolliM,CumbayM,MartensCR,NeveKA,JanowskyA.Characteristics 43. RaineyMM,KorostyshevskyD,LeeS,PerlsteinEO.Theantidepressantsertralinetargets ofdruginteractionswithrecombinantbiogenicaminetransportersexpressedinthesame intracellularvesiculogenicmembranesinyeast.Genetics2010;185:1221–1233. celltype.JPharmacolExpTher1999;289:877–885. 44. LiB,ZhangS,LiM,HertzL,PengL.Chronictreatmentofastrocyteswiththerapeutically 65. Tatsumi M, Groshan K, Blakely RD, Richelson E. Pharmacological profile of relevantfluoxetineconcentrationsenhancescPLA2expressionsecondaryto5-HT2B- antidepressants and related compounds at human monoamine transporters. Eur J induced,transactivation-mediatedERK1/2phosphorylation.Psychopharmacology2009; Pharmacol1997;340:249–258. 207:1–12. 66. EildalJN,AndersenJ,KristensenAS,JorgensenAM,Bang-AndersenB,JorgensenMetal. 45. ZhangS,LiB,LovattD,XuJ,SongD,GoldmanSAetal.5-HT2Breceptorsareexpressed Fromtheselectiveserotonintransporterinhibitorcitalopramtotheselectivenorepinephrine onastrocytesfrombrainandincultureandareachronictargetforallfiveconventional transporterinhibitortalopram:synthesisandstructure-activityrelationshipstudies.JMedChem ’serotonin-specificreuptakeinhibitors’.NeuronGliaBiol2010;6:113–125. 2008;51:3045–3048. 46. LeeLH,ShuiG,FarooquiAA,WenkMR,TanCH,OngWY.Lipidomicanalysesofthe 67. WongDT,BymasterFP.Developmentofantidepressantdrugs.Fluoxetine(Prozac)and mousebrainafterantidepressanttreatment:evidenceforendogenousreleaseoflong- otherselectiveserotoninuptakeinhibitors.AdvExpMedBiol1995;363:77–95. chainfattyacids?IntJNeuropsychopharmacol2009;12:953–964. 68. HerbertJ.Cortisolanddepression:threequestionsforpsychiatry.PsycholMed2012; 47. RapoportSI.Invivofattyacidincorporationintobrainphosholipidsinrelationtoplasma 8:1–21. availability,signaltransductionandmembraneremodeling.JMolNeurosci 2001;16: 69. VreeburgSA,HoogendijkWJ,vanPeltJ,DerijkRH,VerhagenJC,vanDyckRetal.Major 243–261. depressivedisorderandhypothalamic-pituitary-adrenalaxisactivity:resultsfromalarge 48. MalleiA,GiambelliR,GassP,RacagniG,MatheAA,VollmayrBetal.Synaptoproteomics cohortstudy.ArchGenPsychiatry2009;66:617–626. oflearnedhelplessratsinvolveenergymetabolismandcellularremodelingpathwaysin depressive-like behavior and antidepressant response. Neuropharmacology 2011; 60: 1243–1253. Translational Psychiatry is an open-access journal 49. ReusGZ,StringariRB,RezinGT,FragaDB,DaufenbachJF,ScainiGetal.Administration ofmemantineandimipraminealtersmitochondrialrespiratorychainandcreatinekinase published by Nature Publishing Group. This work is activitiesinratbrain.JNeuralTransm2012;119:481–491. licensedundertheCreativeCommonsAttribution-NonCommercial-No 50. ScainiG,SantosPM,BenedetJ,RochiN,GomesLM,BorgesLSetal.EvaluationofKrebs DerivativeWorks3.0UnportedLicense.Toviewacopyofthislicense, cycleenzymesinthebrainofratsafterchronicadministrationofantidepressants.Brain ResBull2010;82:224–227. visithttp://creativecommons.org/licenses/by-nc-nd/3.0/ SupplementaryInformationaccompaniesthepaperontheTranslationalPsychiatrywebsite(http://www.nature.com/tp) TranslationalPsychiatry