Role for the kinase SGK1 in stress, depression, and glucocorticoid effects hippocampal neurogenesis

SGK1, depression, hippocampal neurogenesis

RoleforthekinaseSGK1instress,depression,andglucocorticoideffectsonhippocampalneurogenesis

ChristophAnackera,b,c,1,AnnamariaCattaneoa,d,KseniaMusaelyana,PatriciaA.Zunszaina,MarkHorowitza,

RaffaellaMoltenie,AlessiaLuonie,FrancescaCalabresee,KatherineTanseyf,MassimoGennarellid,g,SandrineThuretc,JackPricec,RudolfUherf,h,MarcoA.Rivae,andCarmineM.Pariantea,b

Stress,PsychiatryandImmunologyLaboratory,DepartmentofPsychologicalMedicine,InstituteofPsychiatry,King’sCollegeLondon,LondonSE59NU,UnitedKingdom;bBiomedicalResearchCentreforMentalHealth,SouthLondonandMaudsleyNationalHealthServiceFoundationTrust,NationalInstituteforHealthResearch,LondonBR33BX,UnitedKingdom;cCentrefortheCellularBasisofBehaviour,InstituteofPsychiatry,King’sCollegeLondon,LondonSE59NU,UnitedKingdom;dGeneticandBiologySection,DepartmentofBiomedicalSciencesandBiotechnology,UniversityofBrescia,25123Brescia,Italy;eDepartmentofPharmacologicalandBiomolecularSciences,UniversityofMilan,20133Milan,Italy;fMedicalResearchCouncilSocial,Genetic&

DevelopmentalPsychiatryCentre,InstituteofPsychiatry,King’sCollegeLondon,LondonSE58AF,UnitedKingdom;gGeneticUnit,IstitutodiRicoveroeCuraaCarattereScienti coSanGiovannidiDio,FatebenefratelliCentre,25100Brescia,Italy;andhDepartmentofPsychiatry,DalhousieUniversity,Halifax,NovaScotia,CanadaB3H2E2

EditedbyBruceS.McEwen,TheRockefellerUniversity,NewYork,NY,andapprovedApril8,2013(receivedforreviewJanuary17,2013)

a

Stressandglucocorticoidhormonesregulatehippocampalneuro-genesis,butthemolecularmechanismsmediatingtheseeffectsarepoorlyunderstood.Hereweidentifytheglucocorticoidreceptor(GR)targetgene,serum-andglucocorticoid-induciblekinase1(SGK1),http://wendang.chazidian.comingahumanhippocampalprogenitorcellline,wefoundthatasmallmoleculeinhibitorforSGK1,GSK650394,counteractedthecortisol-inducedreductioninneurogenesis.More-over,geneexpressionandpathwayanalysisshowedthatinhibitionoftheneurogenicHedgehogpathwaybycortisolwasSGK1-dependent.SGK1alsopotentiatedandmaintainedGRactivationinthepresenceofcortisol,andevenaftercortisolwithdrawal,byincreasingGRphos-phorylationandGRnucleartranslocation.ExperimentscombiningtheinhibitorforSGK1,GSK650394,withtheGRantagonist,RU486,demonstratedthatSGK1wasinvolvedinthecortisol-inducedre-ductioninprogenitorproliferationbothdownstreamofGR,byregulatingrelevanttargetgenes,andupstreamofGR,byincreasingGRfunction.Corroboratingtherelevanceofthese ndingsinclinicalandrodentsettings,wealsoobservedasigni cantincreaseofSGK1mRNAinperipheralbloodofdrug-freedepressedpatients,aswellasinthehippocampusofratssubjectedtoeitherunpredictablechronicmildstressorprenatalstress.Our ndingsidentifySGK1asamediatorfortheeffectsofcortisolonneurogenesisandGRfunction,withpar-ticularrelevancetostressanddepression.

antidepressantsneuroplasticity

nonmutuallyexclusivepossibilityisthatSGK1alsodirectlyen-hancesGRfunctionandpotentiatesglucocorticoideffects.Hereweexaminedboththesepossibilities.

Wehavepreviouslyusedahumanhippocampalprogenitorcelllinetoexaminetheeffectsofglucocorticoidsandantidepressantsonneurogenesis(3,6,12),andwehaveshownthatinhibitionofHedgehogsignalingcriticallycontributestothereductioninneuronaldifferentiationupontreatmentwiththehumanglucocorticoidhor-mone,cortisol(6).HereweinvestigatetheroleofSGK1intheseeffectsofcortisolinvitro.Moreover,weanalyzeSGK1geneex-pressionintheperipheralbloodofdrug-freedepressedpatients,andinthehippocampusofratsexposedtounpredictablechronicmildstress(UCMS)orprenatalstress(PNS),twoinvivostressmodelsknowntoprecipitatedepressivebehaviorandtodecreasehippo-campalneurogenesis(13–15).Our ndingsidentifySGK1asakeyenzymeinvolvedinthedownstreammechanismsbywhichgluco-corticoidsreduceneurogenesisandintheupstreampotentiationandmaintenanceofGRfunction,evenafterglucocorticoidwithdrawal.Results

havepreviouslyreportedthatGRactivationincreasesSGK1ex-pression,andthatGR-mediatedconcentrationsofcortisol(CORT)decreaseproliferationandneuronaldifferentiationofhumanhip-pocampalprogenitorcells(3,6).Inparticular,wehaveshownthattheconcentrationof100μMCORTexertsconsistentGR-medi-atedeffectsoncellproliferationandneuronaldifferentiation,andinducesGRtransactivationwithoutsaturatingthereceptorre-sponse(6).Thisconcentrationwasthususedforthepresentstudy.HereweusedthesmallmoleculeinhibitorforSGK1,GSK650394(16),totestwhetherSGK1ismechanisticallyinvolvedintheeffectsofCORTonneurogenesis.Ashypothesized,GSK650394(at10nM,50nM,and100nM)dose-dependentlycounteractedtheCORT-inducedreductioninBrdU-positive,proliferatingprogenitorcells(one-wayANOVA,P=0.01,F1,4=4.17;Fig.1AandB).

Authorcontributions:C.A.,S.T.,J.P.,R.U.,M.A.R.,andC.M.P.designedresearch;C.A.,A.C.,K.M.,P.A.Z.,M.H.,R.M.,A.L.,andF.C.performedresearch;M.G.contributednewre-agents/analytictools;C.A.,A.C.,P.A.Z.,andK.T.analyzeddata;andC.A.andC.M.P.wrotethepaper.

Con ictofintereststatement:J.P.actedasaconsultantandreceivedpaymentfromReNeuronGroupwithinthelast2y.C.M.P.hasreceivedfeesasaspeakerorasamem-berofadvisoryboard,aswellasresearchfunding,frompharmaceuticalcompaniesthatcommercializeoraredevelopingantidepressantsinthelast3y,suchasLilly,Servier,andJanssen.M.A.R.hasreceivedcompensationasspeaker/consultantforAs-traZeneca,Bristol-MyersSquibb,EliLilly,Servier,andTakeda.P.A.Z.hasreceivedspeakerfeesfromServier.

ThisarticleisaPNASDirectSubmission.

FreelyavailableonlinethroughthePNASopenaccessoption.

1

SGK1MediatestheCortisol-InducedDecreaseinNeurogenesis.We

|hypothalamus–pituitary–adrenalaxis|stemcells|

lucocorticoidhormonesareconsistentlyincreasedinseverelydepressedpatientsandinrodentstressmodels(1).Alargebodyofevidencehasdemonstratedthatchronicallyhighcon-centrationsofglucocorticoidsimpairhippocampalneurogenesis,thatis,thedevelopmentofnewneuronsfromstemcellsinthedentategyrus,byactivatingtheglucocorticoidreceptor(GR)(1–6).However,themechanismsbywhichGRactivationreducesneurogenesisarenotfullyunderstood.HereweexploretheroleoftheGRtargetgene,serum-andglucocorticoid-regulatedki-nase1(SGK1),intheseeffects.

SGK1isaserine/threoninekinaseimplicatedinthecellularstressresponseandneuronalfunction(7).WeandothershavepreviouslyshownthatglucocorticoidsincreaseSGK1expressioninhumanneuralstemcellsandinrodentneurons(3,8).Impor-tantly,SGK1mediatessomeglucocorticoideffectsonbrainfunc-tion;forexample,acutestressenhancesneuronalexcitabilityandworkingmemoryintherodentprefrontalcortexviaglucocorticoid-inducedactivationofSGK1(9),andchronicstresscausesmor-phologicaloligodendrocyteabnormalitiesbyup-regulatingSGK1inthemurinecorpuscallosum(10).ThesedatasuggestthatSGK1maybeadownstreammediatorofglucocorticoideffectsonthebrain.However,recentevidencehasalsosuggestedthatenhancedSGK1expressionisassociatedwithgreaterGR-mediatedgeneexpressioninresponsetoglucocorticoids(11).Therefore,another,

G

Towhomcorrespondenceshouldbeaddressed.E-mail:Christoph.Anacker@kcl.ac.uk.

http://wendang.chazidian.com/lookup/suppl/doi:10.1073/pnas.1300886110/-/DCSupplemental.

SGK1, depression, hippocampal neurogenesis

GSK650394alonedidnotexertanyeffectsonproliferationat

theseconcentrations(one-wayANOVA,P=0.84,FFig.1C),indicatingthatSGK1indeedmediatesthe1,3effect=0.178;ofCORT,butdoesnotelicitinganyeffectbyitself.

WealsoexamineddifferentiationintoDcx-positiveneuroblastsandintomicrotubule-associatedprotein2(MAP2)-positiveneurons

ACORT (100 µM)+

veh

CORT (100 µM)

GSK650394 (100nM)

BrdU Hoechst BrdU Hoechst BrdU Hoechst

B

C

)

)

egegnnahahcc %(% (sllesllce +c U+dUrBdrBCORT

**

100100100100_

_

_

GSK650394_

10nM 50nM 100nM

10nM

50nM

100nM

DCORT (100 µM)+

veh

CORT (100 µM)

GSK650394 (100nM)

Dcx Hoechst DcxHoechst DcxHoechst

MAP2 Hoechst MAP2 Hoechst MAP2 Hoechst

E

F

)

)

eeggnnaahhcc %%(( sslllleecc ++X2CPDAMCORT (100++++_

+

_

+

Fig.1.CORT(100μM)decreasesneurogenesisviaanSGK1-dependenteffect.(A)BrdUimmunocytochemistry(ICC)wasusedtoassessproliferation.(B)TheSGK1inhibitor,GSK650394(10nM,50nM,100nM)dose-dependentlycoun-teractedtheCORT-inducedreductioninproliferation(n=4).(C)GSK650394alonedidnotchangeproliferation(n=4).(D)ICCfordoublecortin(Dcx)andmicrotubule-associatedprotein2(MAP2)wasusedtoassessneuronaldiffer-entiation.(EandF)GSK650394(100nM)counteractedtheCORT-induceddecreaseinDcx-positiveneuroblasts(n=3;E)andinMAP2-positiveneurons(n=3;F).Dataaremean±SEM*P<0.05,**P<0.01;***P<0.001.

(Fig.1D).Inlinewithour ndingsonproliferation,thedoseofGSK650394thatmosteffectivelycounteractedtheeffectsofCORTonproliferation(100nM,asshowninFig.1B)alsocounteractedtheCORT-inducedreductioninDcx-positive(P=0.003;Fig.1E)andinMAP2-positivecells(P=0.03;Fig.1F).Takentogether,thesedatademonstratethatCORTdecreaseshippocampalprogenitorcellproliferationandneuronaldifferentiationviaanSGK1-dependentmechanism.

CortisolIncreasesSGK1Expression.Wethenfurthercharacterized

thetimecourseofCORT-inducedSGK1expression.SGK1mRNAwasmarginallyelevatedafter1h(1.3±0.2fold,P=0.11),butsigni cantlyincreasedafter3h(1.7±0.08fold,P=0.01),12h(1.7±0.03,P=0.002)and72h(1.7±0.09fold,P=0.02)oftreatment(Fig.2A).Moreover,theGR-antagonist,RU486(50nM),abolishedtheCORT-inducedincreaseinSGK1mRNA(Fig.S1A),con rmingthatthiseffectisGR-dependent.Inlinewithour ndingsonthemRNAlevel,nochangesinSGK1proteinwereobservedafter1hofCORTtreatment(1.4±0.5fold,P=0.5),whereastreatmentfor12hsigni cantlyincreasedSGK1proteinlevels(by4.4±1.1fold,P=0.02)(Fig.S1B).Finally,after12hoftreatment,SGK1expressionwassigni cantlyin-creasedonlybyhighconcentrationsofCORT(100μM),butnotbylowerconcentrations(1–100nM)(Fig.2B).Thisresultisinlinewithourpreviousdata,showingthat100μMCORTpre-dominantlyaffectstheGR,whereas1–100nMCORTpre-dominantlyaffectstheMRinthesamecells(6).

SGK1IsInvolvedinCortisol-InducedChangesinMolecularSignalingPathways.Wehavepreviouslyshownthatcortisoldecreasesneuro-naldifferentiationbyinhibitingHedgehogsignaling(6).HereweconductedAffymetrixgeneexpressionmicroarrayandpathwayanalysisonmRNAsamplesofcellstreatedwithCORT(100μM)andGSK650394(http://wendang.chazidian.comingthisapproach,wecon rmandextendourprevious ndings,byshowingthatCORTsigni cantlydecreasesHedgehogsignaling(vehiclevs.100μMCORT;P=0.04,n=5)andthatthiseffectiscounteractedbyGSK650394(vehiclevs.100μMCORT+100nMGSK650394;P=0.22,n=5).Ac-cordingly, Hedgehogrmedthatquantitativereal-timePCR(qRT-PCR)analysiscon-genes,theCORT-inducedglioma-associateddecreaseoncogenein(Gli)expressionandSmoothenedofthekey(Smo)wascounteractedFig.2D).

=6.63,byFig.GSK6503942C;forSmo(one-way:P=0.006,ANOVAFforGli:P=0.005,F1,31,3=6.9,SGK1MediatestheCortisol-InducedIncreaseinGRPhosphorylation.

Next,wewantedtotestwhetherSGK1mayalsoactasan“up-stream”regulatoroftheGR.Therefore,weinvestigatedacriti-calcomponentofGRfunction:phosphorylationattheserineresiduesS203,S211,andS226.CORTtreatmentfor12hinducedphosphorylationatthephospho-sitesS203(by1.5±0.1fold,P=0.007),S211(by2.1±0.4fold,P=0.01),andS226(by1.5±0.2fold,P=0.03).Interestingly,theSGK1-inhibitor,GSK650394(100nM),blockedtheCORT-inducedphosphorylationatS203(P=0.01)andS211(P=0.04),butnotatS226(P=0.4)(Fig.3A–C).Importantly,SGK1wasinvolvedinGRphosphorylationonlywhenSGK1expressionwasincreased(thatis,after12hofcortisoltreatment),butnotatlow,basallevels.Infact,after1hofCORTtreatment,whenSGK1levelsarenotyetelevated(asshowninFig.2AandFig.S1B),GRphosphorylationwasalreadyincreasedatS203(by2.5±0.5-fold,P=0.04),S211(by6.9±2.4-fold,P=0.04),andS226(by2.8±0.9-fold,P=0.07),buttheseeffectswerenotcounteractedbyGSK650394(Fig.S2A–C).Kinasesinvolvedinthisearly,SGK1-independentGRphosphorylationarepresentedinSIResultsandFig.S3.SGK1MediatesCortisol-InducedGRNuclearTranslocation.Having

demonstratedthatSGK1mediatesthecortisol-inducedincreaseinGRphosphorylationattheS203andS211phospho-sites,whichenhanceGRnucleartranslocation,butnotattheS226site,whichinhibitsnucleartranslocation,wethendirectlyassessedSGK1-dependentchangesinGRnucleartranslocationuponcortisol

E

CNEICSORUEN

SGK1, depression, hippocampal neurogenesis

SGK1 mRNA (fold change)

A

1h

SGK1 mRNA (fold change)

B

3h12h72h

CORT0nM1nM10nM100nM

1M10M100M

C

Gli mRNA (fold change)

D

SMO mRNA (fold change)

CORT(100M)GSK650394 (100nM)

Fig.2.CORTincreasesSGK1expressionandinhibitsHedge-hogsignaling.(A)CORT(100μM)increasesSGK1mRNAex-pressionafter3,12,and72h(

n=3).(B)CORT(1nMto100μMfor12h)dose-dependentlyincreasesSGK1mRNAexpression(n=6).(CandD)GSK650394

(100nM)counteractstheCORT(100μM)-induceddecreaseoftheHedgehogpathwaygenesGli(n=6;C)andSMO(n=6;D).Dataaremean±SEM*P<0.05,**P<0.01;***P<0.001.

treatment.Asexpected,after12hoftreatment,CORTinducedGRnucleartranslocationasindicatedbya3.5-foldincreaseinGRnuclearprotein(P=0.008;Fig.4AandB).CotreatmentwithGSK650394counteractedthiseffect(P=0.02;Fig.4B),demonstratingthatCORTinducesGRtranslocationviaanSGK1-dependentmechanism.Incontrast,cytoplasmiclevelsofGRweredecreaseduponCORTtreatment(by44%,P=0.02),butthisdecreasewasnotcounteractedbyGSK650394(Fig.4C),suggestingthatthiseffectmayratherre ectanSGK1-independentdecreaseinoverallGRexpression.This ndingwascon rmedbyaseparatesetofexperimentsinwhichwefoundthatCORTdecreasesGRproteinexpressioninwholecelllysates(by20%,P=0.04),andthatthiseffectwasalsonotcounteractedbyGSK650394(Fig.S4).Takentogether,thesedatademonstratethatSGK1isalsoinvolvedinregulatingGRactivationandtranslocationduringglucocorticoidexposure.

SGK1MaintainsGRActivationintheAbsenceofCortisol.HavingshownthatCORTincreasesSGK1levels,andthattheseincreasedSGK1levelsmediatetheCORT-inducedGRphosphorylationandnucleartranslocation,wethentestedthehypothesisthatSGK1maymaintainGRactivationevenafterCORThasbeenremoved.Therefore,wetreatedcellswithCORTfor3h,atimelongenoughtosigni cantlyinduceSGK1mRNAexpression(asshowninFig.2A).WethenextensivelywashedcellsinCORT-freemediaandinvestigatedGRnucleartranslocationafteratotalincubationtimeof12h,thatis,afterfurther9hofincubationintheabsenceofCORT(Fig.4D).At12h,eventhoughCORThadnotbeenpresentforthelast9h,SGK1proteinexpression

wasstillincreased(by～2-fold,P=0.04)andtheGRwasstilltranslocatedtothenucleus(by～4.5-fold,P=0.04;Fig.4E).Indeed,thispersistentGRtranslocationintheabsenceofCORTwasmaintainedbySGK1,becauseitwascounteractedwhenSGK1wasblockedduringthe9-hperiodaftertheendoftheinitial3-hCORTtreatment(P=0.02;Fig.4E,thirdcolumn).Again,andinlinewithourdataoncontinuousCORTtreatmentfor12h,cytoplasmicGRwasdecreasedevenafterCORTremoval(by17%,P=0.02),andthiseffectwasnotcounteractedbytheSGK1inhibitor(Fig.4F),likelyre ectinganSGK1-independentreductioninoverallGRexpression,asdiscussedabove.

SGK1MediatestheDecreaseInProgenitorProliferationAfterCortisolRemoval.Inlightofour ndingthatSGK1maintainsGRtrans-

locationevenafterCORTremoval,wenexttestedwhetherprogenitorcellproliferationwasalsodecreasedundertheseconditions.WethustreatedcellswithCORTfor3h,washedthemextensivelyinCORT-freemedia,andtheninvestigatedproliferation69hlater(thatis,afteratotalincubationtimeof72h).EventhoughCORTwasonlypresentforthe rst3hofincubation,itwassuf cienttoreduceproliferationtolevelscom-parabletothoseaftercontinuous72-htreatment(～11%,P=0.0009;Fig.5A, rstcolumn).WethenincubatedcellswiththeSGK1inhibitor,GSK650394,duringthe69-hperiodaftertheinitial3-hCORTstimulus,thatis,intheabsenceofanyCORT.Ashypothesized,thistreatmentcompletelyabolishedtheCORT-inducedreductioninproliferation(P=0.0012;Fig.5A,secondcolumn).ThesedatathussuggestthattheCORT-inducedSGK1expressionmaintainsGRactivationevenafterCORTremoval,

A

S203GRtotal

fold induction (GR-S203/GR)GR-S203 / GRtotal

B

S211GRtotal

fold induction (GR-S211/GR)

GR-S211 / GRtotal

C

S226GRtotal

GR-S226 / GRtotal

fold induction (GR-S226/GR)

Fig.3.SGK1mediatesCORTeffectsonGRphos-phorylation.After12h,CORT(100μM)increasesGRphosphorylationatS203(n=6;A),S211(n=6;B),andS226(n=6;C).GSK650394(100nM)counter-actedtheincreaseinS203andS211phosphoryla-tion.Dataaremean±SEM*P<0.05,**P<0.01.

CORT (100M)GSK650394 (100nM)

SGK1, depression, hippocampal neurogenesis

AB

CORT (100M)DE

F

GRtotalGRtotallaminB1

ACTB

n=5

n=5

)

1Bn)

BTimCalA//RRGG(( nonoiitcuctdudnnii d dllooffCORT (100M)GSK650394 (100nM)

Fig.4.SGK1mediatesandmaintainsCORTeffectsonGRnucleartrans-location.(A)CotreatmentwithCORT(100μM)andGSK650394(100nM)for12h.(B)GSK650394counteractedtheCORT-inducedincreaseinnuclearGR(n=7).(C)TheCORT-induceddecreaseincytoplasmicGRwasnotcounteractedbyGSK650394(n=7).(D)CORTtreatmentforthe rst3handGSK650394treatmentduringthesubsequent9hintheabsenceofanyCORT.(E)CORT(foronly3h)wassuf cienttoinduceGRnucleartranslocationafteratotalincubationof12h.GSK650394(duringthe9-hperiodaftercortisol)abolishedtheCORT-inducedincreaseinnuclearGR(n=5).(F)CORT(foronly3h)de-creasedcytoplasmicGR.ThiseffectwasnotcounteractedbysubsequentGSK650394treatment(n=5).Dataaremean±SEM*P<0.05,**P<0.01.

andthatthisSGK1-dependentGRactivation(intheabsenceofCORT)isfunctionallyrelevantforprolongingtheeffectsofCORTonhippocampalprogenitorproliferation.

SGK1EffectsonProliferationOccurbothUpstreamandDownstreamoftheGR.Intheabove-describedexperiments,wehavedemon-

stratedaroleforSGK1inmediatingcortisoleffectsonhumanhippocampalprogenitorcells.WenowwantedtoclarifywhethertheseeffectsofSGK1werepredominantlydownstreamorup-streamofGRactivation.Therefore,wetreatedcellsasabove(CORTforonly3h,theninvestigatedproliferation69hlater),butinthepresenceoftheGRantagonist,RU486(50nM),duringthe69-hperiodafterCORTtreatment.Wehadpre-viouslydemonstratedthatcoincubationwithRU486blockstheCORT-inducedreductioninproliferationafter72h(6).Inthesenovelexperimentalconditions,iftheeffectsofthe3-hCORTtreatmentonproliferationwereblockedbyRU486(duringthe

69haftertheinitialCORTstimulus),itwouldindicatethatthedecreaseinproliferationwasdependentontheabilityofSGK1tomaintainGRactivation;incontrast,iftheeffectsofthe3-hCORTtreatmentwerenotblockedbyRU486(duringthe69hafterCORT),itwouldindicatethatincreasedlevelsofSGK1weresuf cienttoexertalldownstreameffectsleadingtodecreasedproliferation.Interestingly,treatmentwithRU486onlypartiallycounteractedthereductioninproliferation(by～60%,P=0.04;Fig.5A,thirdcolumn).ThesedatathussuggestthatSGK1medi-atestheCORT-inducedreductioninhippocampalprogenitorproliferationbyactingbothupstreamofGR,byenhancingGRfunction,aswellasdownstreamofGR,byregulatingneuro-genesis-relevantmoleculartargets(Fig.5B).

IncreasedSGK1mRNAExpressioninDepressedPatients.Tocorrob-

oratetheclinicalrelevanceofourcellularandmolecular ndings,wemeasuredSGK1mRNAexpressioninperipheralbloodofpatientswithmajordepression,aconditioninwhichbothhyper-cortisolemiaandreducedneurogenesisareconsideredrelevantpathophysiologicalmechanisms.Wehavepreviouslydescribedcandidategeneexpressionpro lesassociatedwithGRfunctionandneuroplasticityintheGenome-basedTherapeuticDrugsforDepression(GENDEP)study(17).HereweanalyzedSGK1expressioninasubpopulationofthiscohort.Allpatients(n=25)weredrug-freeforatleast2wkbeforeassessment,and18ofthese25patientsweredrugnaïve.Controls(n=14)hadnopresentorpastpsychiatricdisorderanddidnottakeanyhormonaltreatment.Therewerenosigni cantdifferencesinageandsexbetweenpatientsandcontrols(P=0.97forage,andP=0.74forsex).Thefullpatients’clinicalcharacteristicsandinclusioncriteriaaredescribedintheSIMaterialsandMethods.

Inlinewithourdataonhumanhippocampalprogenitorcells,wefoundthatdepressedpatientshadsigni cantlyhigherSGK1mRNAlevels(by～2.5-fold;controls:1.26±0.16,patients:3.11±0.24,P<0.0001,Fig.6A).Eventhoughoursampleincludedmoremalethanfemalepatients(n=15males,n=10females),therewasnoeffectofsexonSGK1mRNAexpressionincontrols

A

)

egnahc %( sllec +UdrBCortisol (100M)

*

GSK650394(100nM)RU486(50nM)

Fig.5.SGK1effectsonproliferationoccurupstreamanddownstreamoftheGR.(A)CORT(100μM)foronlythe rst3hofa72-hincubationperiodwassuf cienttoreducethenumberofBrdU-positivecells( rstcolumn).TreatmentwithGSK650394(100nM)duringthe69-hperiodaftertheinitial3-hCORTstimulusabolishedtheCORT-inducedreductioninproliferation(secondcolumn),whereastheGRantagonistRU486(50nM)partiallycounteractedtheeffectofCORT(thirdcolumn).n=10.Dataaremean±SEM*P<0.05,**P<0.01;***P<0.001.(B)DiagramdepictingtheproposedmodelofSGK1-dependentregulationofneurogenesisandGRfunction.

E

CNEICSORUEN

SGK1, depression, hippocampal neurogenesis

(femalecontrols:1.27±0.3,n=7,malecontrols:1.24±0.13,n=7;P=0.9)ordepressedpatients(femaledepressed:3.14±0.3,n=10,maledepressed:3.09±0.3,n=15;P=0.9).Moreover,wecorrelatedSGK1expressionwiththeexpressionofGR-relevantgenesthatwehadpreviouslyanalyzedinthesameindividuals(17).ConsistentwithourinvitrodatashowingdecreasedGRinthepresenceofincreasedSGK1expression,SGK1mRNAlevelscor-relatednegativelywithGRlevelsinourclinicalsample(r= 0.32,P=0.046).Moreover,SGK1correlatedpositivelywithmRNAlevelsoftheGR-targetgene,FK506bindingprotein5(FKBP5)(r=0.45,P=0.004),andnegativelywithgenesinvolvedinneuro-plasticity[brain-derivedneurotrophicfactor(BDNF):r= 0.32,P=0.05;VGF:r= 0.33,P=0.037;p11:r= 0.43,P=0.007].

IncreasedHippocampalSGK1ExpressioninRodentStressModels.

Finally,weanalyzedSGK1mRNAexpressioninthehippocam-pusofadultratsexposedtotwowellestablishedrodentmodelsofdepression,characterizedbybothincreasedcorticosteronelevelsandreducedneurogenesis(13–15):unpredictablechronicmildstress(UCMS)andprenatalstress(PNS).UCMShasbeenshowntoreducehippocampalneurogenesisintheventralandinthedorsalhippocampus(18).Accordingly,inourexperiments,UCMSsigni cantlyincreasedSGK1expressionbothintheventralhippocampus(by～1.4-fold,P<0.001,n=8)andinthedorsalhippocampus(by～1.2-fold,P<0.01,n=8)(Fig.6B).Inaddition,wehadpreviouslyshownthatPNScausesincreasedbloodcor-ticosteronelevelsandmolecularsignalingchangesintherathippocampus,includingareductioninHedgehogsignaling(6).Forthecurrentstudy,weanalyzedSGK1mRNAexpressioninthewholehippocampusofthesameratsandfoundanincreaseof～1.4-fold(P=0.02,n=5,Fig.S5).ThesedatainrodentstressmodelsthusfurthersupportanimportantroleforSGK1instressandglucocorticoideffectsonthehippocampus.Discussion

HerewedemonstratethattheGRtargetgene,SGK1,mediatesthecortisol-induceddecreaseinproliferationandneuronaldif-ferentiationofhumanhippocampalprogenitorcells,byactingbothdownstreamofGRactivation(viaSGK1-dependentinhi-bitionoftheHedgehogpathway)andupstreamofGRactivation(viaSGK1-dependentGRphosphorylationandnucleartrans-location).Additionally,we ndthatSGK1mRNAisincreasedintheperipheralbloodofdrug-freedepressedpatientsandinthehippocampusofratsafterunpredictablechronicmildstressandprenatalstress.Our ndingsidentifySGK1asamechanismformediatingglucocorticoideffectsonneurogenesisandforenhancingGRactivation,whichmaybeofparticularrelevanceforstress-inducedmentaldisorders,suchasdepression.

AlthoughSGK1wasinitiallydescribedforitsroleinregulatingsodiumtransportinrenalcollectingductcells(19),recentstudieshaveprovidedevidenceforanovelroleofthiskinaseinstressandglucocorticoidactionsonthebrain(9,10).Ourstudyextendstheseprevious ndingsbyshowingthatSGK1alsomediatestheinhibitoryeffectsofcortisolonproliferationandneuronaldifferentiationof

A

)

oi

B

)

tear ng AnaohNiRsc sdmel ro1pf(Kx GeA NSevitmRa l1erK(GSventraldorsal

Fig.6.SGK1mRNAexpressionisincreasedindrug-freedepressedpatients(ctrls,n=14;patients,n=25)(A)andtheventralanddorsalhippocampusofratsuponunpredictablechronicmildstress(UCMS,n=8;B).Dataaremean±SEM;**P<0.01;***P<0.001.

humanhippocampalprogenitorcells.Moreover,we ndthatin-hibitionofHedgehogsignalingmaybeoneofthemolecularmech-anismsresultingfromtheincreasedSGK1expressionbycortisol.Importantly,wedemonstratethatSGK1isnotonlyamecha-nism“downstream”ofGRactivation,butisinfactalsoinvolvedinregulatingGRfunction.Speci cally,theGR-mediatedincreaseinSGK1expressionuponcortisoltreatmentfurtherenhancesGRphosphorylationattheserineresiduesS203andS211,whicharephosphorylationsitesknowntofacilitatenucleartranslocationofthereceptor(20,21).Accordingly,SGK1enhancesandmaintainsGRnucleartranslocationevenaftercortisolwithdrawal.ThissustainedSGK1-mediatedGRactivationintheabsenceofcor-tisolhasfunctionalconsequencesandresultsinareductioninprogenitorproliferationthatisabolishedbyinhibitingSGK1evenaftercortisolhasbeenremoved.Thisinterpretationisconsistentwithstudiesshowingthatasingle-nucleotidepolymorphismintheSGK1promoterregion,withaputativeeffectonincreasingSGK1expressionuponglucocorticoidtreatment,enhancesglucocorticoid-mediatedgenetranscription(11).Therefore,weproposethatGR-dependentSGK1up-regulationmaybeamechanismformaintainingandprolongingGRactivationafterexposuretoincreasedglucocorticoidlevels,suchasduringstress,andevenafterglucocorticoidlevelshavenormalized.

Ourhumaninvitrodataappeartoberelevantalsoinvivobothinrodentsandinhumans,aswehaveidenti edincreasedSGK1mRNAexpressioninthehippocampusofratsexposedtotwodif-ferenttypesofstress,andinthebloodofdrug-freedepressedpatients.Speci cally,UCMSincreasedSGK1expressionbothintheventralandinthedorsalhippocampus,whichisinlinewithstudiesshowingthatUCMSalsodecreasesneurogenesisinbothoftheseregions(18).Consideringthattheventralhippocampushasbeenimplicatedinmood,whereasthedorsalhippocampushasbeenimplicatedpredominantlyincognition(22),theroleofSGK1intheeffectsofglucocorticoidsonneurogenesisandbehaviorinrodentstressmodelswillbeaninterestinglineoffutureinvestigation.ThenegativecorrelationbetweenSGK1andlevelsofthreegenesinvolvedinneuroplasticityinourclinicalsample(BDNF,VGF,andp11),mayofferanindicationthatthismechanismmayalsopotentiallybeinvolvedinregulatingneurogenesisinde-pression.However,althoughadulthippocampalneurogenesishasrepeatedlybeenimplicatedinstress-induceddepressivesymptomsandantidepressanttreatmentresponseinrodents(23,24),thefunctionalroleofneurogenesisinhumansremainselusive.Indeed,comparedwithrodents,relativelyfewneuronsarebeingaddedtothehumanbrain(25),andhumanpostmortembrainstudieshavedeliveredcon ictingresultswithregardstochangesinhippocam-palcellproliferationindepressionanduponantidepressanttreatment(26–28).Itisalsoimportanttomentionthatonepre-viousreporthasexaminedSGK1mRNAexpressionindepressedpatients,andfoundnodifferencecomparedwithcontrols(29);thismaylikelybeduetothefactthatintheirstudy,29ofthe40patientswereonantidepressants,andnotdrug-freeasinourstudy.Interestingly,undercertainconditionsstressmayalsoleadtoanincreaseinneurogenesisinthepresenceofelevatedglucocorticoids,particularlywhenthestresshashedonicvalue,forexample,duringmating(30),environmentalenrichment(31)orphysicalexercise(32).Indeed,somestudieshavesuggestedthatconcomitantregulationofprotectivefactors,suchasoxytocin,maycontributetotheneuro-genesisincreaseinthesesituations(33).WhetherSGK1alsocon-tributestosuch“protective”effectsofglucocorticoidsremainstobeelucidated.

Somelimitationsofourstudyneedtobeconsidered:First,al-thoughwehavedeterminedthatthecortisoleffectsinourinvitroexperimentsarein uencedbyalbumininthecellculturemedia(SIResultsandFigS6),itisdif culttoestimatehowthecortisolcon-centrationsinourcellcultureexperimentscomparewithphysio-logicalcortisolconcentrationsinthehumanhippocampusinvivo.Second,thesamplesizeofourclinicalpopulationissmallandthechangesinSGK1mRNAexpressionshouldthusbereplicatedinasecondcohortofdrug-freedepressedpatients.Althoughallde-pressedpatientsincludedinthisstudyweredrug-freeforatleast2wkatthetimeofassessment,and～70%wereinfactdrug-naïve,