Crosstalk between HSF1 and HSF2 during the heat shock response in mouse testes

(1)

ContentslistsavailableatScienceDirect

The International Journal of Biochemistry

& Cell Biology

j ourna l h o m e pa ge :w w w . e l s e v i e r . c o m / l o c a t e / b i o c e l

Crosstalk between HSF1 and HSF2 during the heat shock response in mouse testes

Joanna Korfanty

^a

, Tomasz Stokowy

^a,d

, Piotr Widlak

^a

, Agnieszka Gogler-Piglowska

^a

, Luiza Handschuh

^b,c

, Jan Podkowi ´nski

^b

, Natalia Vydra

^a

, Anna Naumowicz

^a,e

,

Agnieszka Toma-Jonik

^a

, Wieslawa Widlak

^a,∗

aMariaSkłodowska-CurieMemorialCancerCenterandInstituteofOncology,GliwiceBranch,Wybrze ˙zeArmiiKrajowej15,44-101Gliwice,Poland

bMicroarrayandDeepSequencingLaboratory,InstituteofBioorganicChemistryPolishAcademyofSciences,Noskowskiego12/14,61-704Pozna´n,Poland

cDepartmentofHematologyandBoneMarrowTransplantation,PoznanUniversityofMedicalSciences,Szamarzewskiego84,60-569Pozna´n,Poland

dDepartmentofClinicalScience,UniversityofBergen,Postboks7800,5020Bergen,Norway

eInstituteofAutomaticControl,TheSilesianUniversityofTechnology,44-100Gliwice,Poland

a r t i c l e i n f o

Articlehistory:

Received14May2014 Receivedinrevisedform 24September2014 Accepted6October2014 Availableonline19October2014

Keywords:

Heatshockresponse Spermatogenesis ChIP-Seq

a b s t r a c t

HeatShockFactor1(HSF1)istheprimarytranscriptionfactorresponsiblefortheresponsetocellular stress,whileHSF2becomesactivatedduringdevelopmentanddifferentiation,includingspermatogen- esis.Althoughbothfactorsareindispensableforproperspermatogenesis,activationofHSF1byheat shockinitiatesapoptosisofspermatogeniccellsleadingtoinfertilityofmales.Tocharacterizemecha- nismsassistingsuchheatinducedapoptosiswestudiedhowHSF1andHSF2cooperateduringtheheat shockresponse.Forthispurposeweusedchromatinimmunoprecipitationandtheproximityligation approaches.Welookedforco-occupationofbindingsitesbyHSF1andHSF2inuntreated(32^◦C)orheat shocked(at38^◦Cor43^◦C)spermatocytes,whicharecellsthemostsensitivetohyperthermia.Atthe physiologicaltemperatureoraftermildhyperthermiaat38^◦C,thesharingofbindingsitesforbothHSFs wasobservedmainlyinpromotersofHspgenesandotherstress-relatedgenes.Stronghyperthermia at43^◦CresultedinanincreasedbindingofHSF1andreleasingofHSF2,henceco-occupationofpro- moterregionswasnotdetectedanymore.ThecloseproximityofHSF1andHSF2(and/orexistenceof HSF1/HSF2complexes)wasfrequentatthephysiologicaltemperature.Temperatureelevationresulted inadecreasednumberofsuchcomplexesandtheywerebarelydetectedafterstronghyperthermiaat 43^◦C.WehaveconcludedthatatthephysiologicaltemperatureHSF1andHSF2cooperateinspermato- geniccells.However,temperatureelevationcausesremodelingofchromatinbindingandinteractions betweenHSFsaredisrupted.Thispotentiallyaffectstheregulationofstressresponseandcontributesto theheatsensitivityofthesecells.

1. Introduction

Proteotoxicstress,e.g.induced byhyperthermia, provokes a rapid response to maintain homeostasis, so called heat shock

Abbreviations:ChIP-Seq,chromatinimmunoprecipitationcombinedwithhigh- throughputsequencing;HSF,heatshockfactor;HSP,heatshockprotein;HSE,heat shockelement;HSR,heatshockresponse;PLA,proximityligationassay.

∗Correspondingauthor.Tel.:+48322789669;fax:+483227899840.

E-mailaddresses:[email protected](J.Korfanty),[email protected] (T.Stokowy),[email protected](P.Widlak),[email protected]

(A.Gogler-Piglowska),[email protected](L.Handschuh), [email protected](J.Podkowi ´nski),[email protected](N.Vydra), [email protected](A.Naumowicz),[email protected](A.Toma-Jonik), [email protected](W.Widlak).

response(HSR).Thisstress-inducedresponseisexecutedbyheat shockproteins(HSPs),whicharemajormolecularchaperonescon- tributingtoproteinrepairand degradationinstressconditions, butalsoassistingproteinfoldingduringbiosynthesis.Mammalian HSPsareclassiﬁedaccordingtomolecularweightintoseveralfam- ilies:HSPH(HSP110),HSPC(HSP90),HSPA(HSP70),DNAJ(HSP40), HSPB(smallHSPs,sHSPs),andtwochaperoninfamilies:HSPD/E (HSP60/HSP10)andCCT(TRiC)(Kampingaetal.,2009).EachHSP familyincludesmembersthatareeitherinduciblebystress(e.g.

HSPA1), constitutively expressed, or both (e.g. HSPH1, HSPA8, HSP90AA1).ExpressionofsomeHSPsisdevelopmentallyregulated orrestrictedtospeciﬁccells(Rupiketal.,2011;Jietal.,2012).The HSRisregulatedbyHeatShockFactors(HSFs),which aremajor transcriptionalactivatorsofHSPgenes.SeveralmembersoftheHSF familyhavebeenfoundinvertebrates(Vydraetal.,2014).Once http://dx.doi.org/10.1016/j.biocel.2014.10.006

(2)

activatedtheyformtrimersand bindspeciﬁcallytoHeatShock Elements(HSEs) throughoutthe genome.In mammals,HSF1 is theprimarytranscriptionfactor responsiblefortheresponseto differentformsofcellularstress,whileHSF2becomes activated duringdevelopmentanddifferentiation(e.g.duringspermatogenesis).Nevertheless,bothHSF1andHSF2canformheterotrimers andcooperateeitherduringstressorunderphysiologicalcondi- tions(Mathewetal.,2001;Heetal.,2003;Ostlingetal.,2007;

Shinkawaetal.,2011).

DespitethehighdegreeofconservationoftheHSR,different cellsvaryintheirabilitytoinduceHSPssynthesis,andconsequently insensitivitytoharmfulconditions.Interestingly,sometypesof cells,e.g.spermatocytes,lackthetypicalHSRandarehypersensi- tivetoelevatedtemperatures(Yinetal.,1997).Inthemajorityof mammals,malegonadsarelocatedoutsidethemainbodycavityto providethelowertesticulartemperaturerequiredforcorrectsper- matogenesisandfertility.Increasingthetemperatureoftestisupto thebodytemperature(oraboveit)leadstotheactivationofHSF1.

However,inducibleHspa1(Hsp70i)genesexpressionisblockedin heatshockedmurinespermatocytes(Izuetal.,2004;Vydraetal., 2006), although HSF1 binds totheirpromoters (Kus-Li´skiewicz etal.,2013).Moreover,theconstitutivelyexpressedtestis-speciﬁc variantofHSP70(HSPA2)isdepletedafterHSF1activation(Widlak et al.,2007).Hence,an over-expression of constitutivelyactive HSF1inmiceleadstotheapoptoticdeathofspermatocytesand maleinfertility(Nakaietal.,2000;Widłaketal.,2003;Vydraetal., 2006).Ontheotherhand,endogenousHSF1appearstobeimpor- tantforspermatogenesisbecauseHsf1nullmales,althoughfertile, producelessspermthanwildtypemice(Salmandetal.,2008).Ithas beenshownthatHSF1isrequiredforthetranscriptionalregulation ofsexchromosomalmulticopygenesduringpostmeioticrepres- sion(Akerfeltetal.,2010).Reductionoffertilitywasalsoobserved inHsf2nullmales(Wangetal.,2003).Moreover,doubleHsf1and Hsf2knockoutcausesmalesterilityandacompletelackofmature sperminmice(Wangetal.,2004).IthasbeenshownthatbothHSF1 andHSF2arerequiredforcorrectchromatinorganizationduring normalspermatogenesis(Akerfeltetal.,2008,2010),andthatboth factorscanformheterotrimersatthechromatin(Sandqvistetal., 2009).TheseﬁndingsindicatefunctionalcrosstalkbetweenHSF1 andHSF2duringspermatogenesisinnormalconditions.However, theinterplaybetweenbothfactorsduringtheheatshockresponse intestishasneverbeenstudied.Aimingtoelucidatethemecha- nismsofsuchinteraction,herewestudiedthechromatinbindingof HSF1andHSF2inmousespermatocytessubjectedtohyperthermia.

2. Materialsandmethods 2.1. Isolationofspermatocytes

Adult(10–16-week-old), inbredFVB/Nmalemicewereused forspermatocytesisolation(20malesperoneisolation)byunit gravitysedimentationinlinearBSAgradientasdescribedearlier (Kus-Li´skiewiczetal.,2013).Isolatedfractioncontainedupto80%

ofspermatocytesandwascontaminated mainlybyroundsper- matids.Aftereachisolationcells wereequallydividedforthree groups:control(culturedat32^◦C),andheatshockedat38^◦Corat 43^◦C.

2.2. Hyperthermiatreatment

ForChIPexperimentsheatshockwasperformedasdescribed indetailselsewhere(Kus-Li´skiewiczetal.,2013):anequalvolume (10ml)ofCO₂saturated,pre-heatedmedia(to53^◦Cor60^◦C)were addedtothecellsuspension,whichimmediatelyraisedthetem- peratureofthemediafrom32^◦Cto38^◦Cor43^◦C,respectively.The

tubesweresubmergedinawaterbathattheappropriatetemper- atureforanadditional5,10,or20min(thesesampleswerepooled andtreatedasheatshockedsample).Immediatelyafterheatshock cellswereﬁxedfor10minbyaddingformaldehydetoﬁnalcon- centration1%,whilethecellmediumwasquicklycooledtoroom temperature.Thewhole-bodyhyperthermiawasperformedinvivo inawaterbathat38^◦Cor43^◦Casdescribedearlier(Widlaketal., 2007).TheanimalexperimentswerecarriedoutaccordingtoPolish legislation,andwereapprovedbytheLocalCommitteeofEthics andAnimalExperimentationattheMedicalUniversityofSilesiain Katowice,Poland(DecisionNo82/2009)andbytheinstitutional animalcarepolicyoftheCancerCenterandInstituteofOncology (Gliwice,Poland).

2.3. Chromatinimmunoprecipitation(ChIP)

For analysesof HSF1 and HSF2 binding,the ChIPassay was carried out according to the protocol of ChIP kit of Upstate Biotechnology(LakePlacid,NY)usingproteinA-sepharosebeads (Amersham).For30␮gofchromatinsonicatedto100–500bpfrag- ments,3␮gof rabbitanti-HSF1(cat.noADI-SPA-901, EnzoLife Sciences),or5␮gofgoatanti-HSF2(cat.noAF5227,R&DSystems, USA)polyclonalantibodieswereused.Fornegativecontrolschro- matinsampleswereproceededwithoutantibody,withanti-TetR rabbitpolyclonalantibody(Abcam),withIgGfromrabbitserum, or withnormal goatserum;allsuchcontrolsgenerated similar results.ImmunoprecipitatedDNAwasanalyzedbyPCR(ChIP-PCR) toassessqualityofpreparationbeforesequencingandtovalidate ChIP-Seqresults.Primerscharacteristicsusedinanalysesarepre- sentedinSuppl.Table1.

2.4. High-throughputsequencing,dataanalysisandfunctional annotation

In each experimentalpoint two PCR-veriﬁed ChIP replicates werecollectedandcombinedinonesamplebeforeDNAsequenc- ing.Sequencinglibrariesweregenerated usingChIP-SeqSample Prep Kit (Illumina). Template ampliﬁcation and cluster gener- ation were performed using the cBot and TruSeq SR Cluster Kit v2 cBot-GA, and 80 nucleotides were sequenced with Illu- mina Genome Analyzer IIx using TruSeq SBS Kit v5 reagents.

Afterqualityfiltering(averagephred>30)andremovalofdupli- cates, readswere mappedto themouse genome (mm10)with Bowtie2 (Langmead et al., 2009). A minimum fold enrichment of five times over negative control was set as a cutoff crite- rion for target sites. The peaks were called with Model-based Analysis of ChIP-Seq (MACS) 1.4.2 (Feng et al., 2012). HSF1 and HSF2targetsiteswereannotatedtogenomicregionsusing HOMER software (Heinz et al., 2010). Fifty percent of peak length was centered on the summit point, and peaks that fell on exon-intron boundaries are indicated as exons. The den- sity signals of HSF1 and HSF2 on the mouse genome were visualized with the Integrative Genomics Viewer version 2.2.1 (Thorvaldsdóttir et al., 2013). The consensus DNA sequences for HSF1 and HSF2 were identified in silica by motif analysis of large DNA datasets (MEME-ChIP Version 4.9.1) (Bailey, 2011; Machanick and Bailey, 2011)using a120-bp regioncen- teredonthesummit point.Biologicalprocesses associatedwith HSF1 or HSF2 bound genes were analyzed with NucleoAnnot application created within the confines of the GENEPI Low-RT project(FP6-036452;availableonSilesianBioinformaticPlatform:

http://cellab.polsl.pl/index.php/software/standalone-app, August, 2014).Thehypergeometrictestwasappliedforcalculationofthe Pvalueforenrichedgeneontologyterms.

(3)

2.5. Proximityligationassay

TodetecttheHSF1/HSF2interactionstheDuoLinkinsituProx- imityLigationAssay(PLA)(OlinkBioscience,Uppsala,Sweden)was usedaccordingtothemanufacturer’sprotocol.Reactionswereper- formedonsections(8␮m)offormalin-fixed(4%inPBS,overnight at 4^◦C) and paraffin-embedded mouse testes. For each experimentalpoint (control,heatshocked for15min,for30min,and for30minwith2hrecovery)threemaleswereused.Anantigen retrievalstepin0.01McitratebufferpH6.0wasperformedbefore theprocedure.SectionswerewashedinPBS(3×5min),incubated inBlockingSolution(OlinkBioscience),andimmunolabeledwith primaryantibodies.WeusedthesameantibodiesasforChIP(1:90 dilution,1%BSAinPBS,overnight,4^◦C);negativecontrolswere proceededwithoutoneprimary antibodyor bothgivingsimilar results.ThenthesecondaryantibodieswithattachedPLAprobes (PLAProbeanti-RabbitPLUSandPLAProbeanti-GoatMINUS;sup- pliedintheDuolinkkit)wereused.Signalsofanalyzedcomplexes wereobservedbyconfocalmicroscopyat×600magnificationorby fluorescentmicroscopyat×1008magnification;redfluorescence signalindicatedcloseproximity(<40nm)ofproteinsrecognized bybothantibodies(Fredrikssonetal.,2002).Imagesfromconfo- calmicroscopyweretakenin25focalplanes(every340nm)and combinedforoneimage.Imagesfromfluorescentmicroscopywere takenintwo–threefocalplanes.Thesamesetting(acquisitiontime) wasusedforalltheimages. Atleastsevenimageswerechosen withtubulesindifferentdevelopmentalstagesandallspotsinside tubuleswerecounted.

3. Results

3.1. Genome-wideidentiﬁcationofHSF1andHSF2chromatin bindingsitesinmousespermatocytes

Weused theChIP-Seq approach tocharacterizeactualbind- ingsitesforHSF1andHSF2inisolatedmousespermatocytesthat wereeitheruntreatedor heat-shockedfor 5–20minat 38^◦C or at43^◦C.Durationoftheheatshockwaspreviouslyestablishedas optimalforHSF1activation(Kus-Li´skiewiczetal.,2013).Weused aspermatocyte-enrichedfractionoftesticularcellsbecausethese cellsarethemostsensitivetodamageatelevatedtemperatures (Yinetal.,1997);thisapproachalsoallowstheavoidingofpossible interferenceofsomatictesticularcells(Kus-Li´skiewiczetal.,2013).

TheChIP-Seqprovidedhigh-resolutionmapsofHSF1andHSF2tar- getsitesinthemousegenome(Suppl.Dataset1;GeneExpression Omnibusaccessionno.GSE56735).Underphysiologicaltempera- ture(thatis32–33^◦Cformousetestes),1,562bindingsiteswere identiﬁedforHSF1and1,284forHSF2.Anelevationofthetem- peraturecausedchangesintheHSF1andHSF2chromatinbinding.

ThenumberofbindingsitesofHSF1,afteraninitialdecreaseat 38^◦C,reacheditsmaximumat43^◦C,whilethebindingofHSF2 graduallydecreasedwithtemperatureelevation.Asimilarproﬁle wasobservedincaseofbothglobalgenomebindingandbinding atpromoterregions(Fig.1).Inbothcontrolandheat-shockedcells eithertranscriptionfactoroccupiedchromatinprimarilyininter- genicsequences(44–61%)andintrons(32–40%),while5-12% of allHSF1-bindingsitesand1.5–4%ofallHSF2-bindingsiteswere locatedinpromoter regions(Fig.2).EitherHSF1 orHSF2 bind- ingsites weredetected in promotersof 60 genes encoding for HSPsandotherstress-relatedproteins(Suppl.Table2).TheGene toGOBP(GeneOntologyBiologicalProcess)analysesrevealedthat amonggenesoccupiedinpromotersbybothHSF1andHSF2inall conditions(exceptHSF2bindingat43^◦C,whichisminimal)these associatedwithclassicalHSF-relatedfunctions(e.g.,genesinvolved inresponsetostressandinproteinfolding)areover-represented

Fig.1.ThenumberofHSF1andHSF2bindingsitesinmousespermatocytes:control andheatshocked(HS)for5–20minat38^◦Cand43^◦C.Promotersweredeﬁnedas theregion−1000bp,+100bparoundthetranscriptionstartsiteofRefseqgenes.

(Suppl.Dataset2).GOanalysesofHSFstargetsinnon-promoter regions(includingexonsandintrons)showedthatinallconditions genesinvolvedinDNA-dependent(regulationof)transcription,in transport,andin(protein)phosphorylationareover-represented.

3.2. SharedbindingofHSF1andHSF2atpromotersof stress-relatedgeneswasremodeledatanelevatedtemperature

Weobserved thatseveralbindingsitesweresharedbyboth transcription factors,which wascharacteristic for promotersof Hspsandotherstress-relatedgenes.TheChIP-Seqanalysisrevealed thatHSF1andHSF2couldsimultaneouslyoccupygenepromoters atthephysiologicaltemperatureand/orfollowing“mild”hyper- thermiaat38^◦C,butnotat43^◦C.Ingeneral,for159geneswhich promoterregionsboundeitherHSFatthephysiologicaltempera- tureorat38^◦C,promotersof12genes(∼8%)weresimultaneously occupiedbybothfactors.Importantly,for12chaperoneandco- chaperonegenesthatboundeitherHSFintheseconditionsthere were9genes,which promoterswereco-occupiedbyHSF1 and HSF2at33^◦Cor38^◦C(Table1;Suppl.Tables3AandB).Inmarked contrast,forabout3,300bindingsitesforeitherfactor detected outsidepromoterregionsatthephysiological temperatureorat 38^◦C,only∼0.4%(12sites)weresharedbyHSF1andHSF2.After temperatureelevationupto43^◦C,remodelingofHSFsbindingto genepromoterswasobserved:bindingofHSF1andreleasingof HSF2.Such temperature-related changeswere observedinpro- motersofchaperoneandco-chaperonegenes(Hspa8,Hsp90aa1, Hsp90ab1,Hspd1,Hspe1,Dnaja1,Hsph1,Cct6a,Stip1,St13)(Fig.3), genes coding for proteins involved in ubiquitination (Ube2g2, Ubqln1, Uspl1), and someother genes(Acot7, Aldh1a2,Ccdc117, Rsrp1,Gm10069,Hnrnpa2b1,Ptges3,Setx,Slc35e2,Spo11)(Fig.4;

Table1;Suppl.Tables2and3).Asaresult,targetsitessharedby both HSF1and HSF2 werenot detectedinpromoter regions in cellssubjectedtoheatshockat43^◦C(onlythreesuchsiteswere detectedoutsidepromoterregions;Suppl.Table3C).Consequently, lookingforDNAmotifsenrichedintheChIP-Seqdatasetswefound

Fig.2.DistributionofHSF1andHSF2bindingsiteswithindifferentgenomicregions incontrolandheatshockedspermatocytes.ExonsandintronsarefromRefSeq;

promoterregion:−1000bpto+100bpfromthetranscriptionstartsite.

(4)

Table1

SharedbindingofHSF1andHSF2topromotersofselectedgenesinmousespermatocytes,controlandheatshockedfor5–20minat38^◦Corat43^◦C(formoredetailssee Suppl.Table2).

GeneName (EntrezID)

Full name

PeakScore

HSF1 HSF2

Control HS38^◦C HS43^◦C Control HS38^◦C HS43^◦C Stress-relatedgenes

Hspa8(15481) Heatshockprotein8 118.46 98.22 705.02 115.80 116.27 –

Hsp90aa1(15519) Heatshockprotein90,alpha(cytosolic),classA member1

402.16 597.48 358.41 232.67 129.05 –

Hsp90ab1(15516) Heatshockprotein90,alpha(cytosolic),classB member1

– 300.93 1747.56 105.83 69.18 –

Dnaja1(15502) DnaJ(Hsp40)homolog,subfamilyA,member1 – 99.29 388.24 83.21 – –

Hspd1/Hspe1^a(15510/15528) Heatshockprotein1(chaperonin)/heatshock protein1(chaperonin10)

100.35 227.90 1176.39 74.60 216.35 –

Hsph1(15505) Heatshock105kDa/110kDaprotein1 – 59.20 272.90 – 129.88 –

Cct6a(12466) ChaperonincontainingTcp1,subunit6a(zeta) – – 55.11 83.35 – –

Ube2g2(22213) Ubiquitin-conjugatingenzymeE2G2 138.33 70.81 177.11 69.51 – –

Ubqln1(56085) Ubiquilin1 – – 276.22 51.95 – –

Uspl1(231915) Ubiquitinspeciﬁcpeptidaselike1 104.37 311.85 792.74 157.96 143.70 –

St13(70356) Suppressionoftumorigenicity13 104.37 280.31 254.59 211.47 92.66 –

Stip1(20867) Stress-inducedphosphoprotein1 65.91 184.28 708.82 105.83 – –

Othergenes

Acot7(70025) Acyl-CoAthioesterase7 276.39 183.61 222.30 82.43 – –

Aldh1a2(19378) Aldehydedehydrogenasefamily1,subfamilyA2 148.23 – 118.88 – 129.88 –

Ccdc117(104479) Coiled-coildomaincontaining117 – 55.64 80.19 56.78 – –

Rsrp1(27981) Arginine/serinerichprotein1 – – 109.51 199.39 – –

Gm10069(791299) Predictedgene10069 – 57.95 214.29 – 109.02 –

Hnrnpa2b1(53379) HeterogeneousnuclearRibonucleoproteinA2/B1 – – 364.14 – 112.15 –

Ptges3(56351) ProstaglandinEsynthase3(cytosolic) – – 135.92 – 66.52 –

Setx(269254) Senataxin – 53.35 156.13 71.16 – –

Slc35e2(320541) Solutecarrierfamily35,memberE2 89.68 151.70 112.08 – 107.82 –

Spo11(26972) SPO11meioticproteincovalentlyboundtoDSB homolog(S.cerevisiae)

– 157.82 192.04 85.43 – –

aGenesoriented“head-to-head”.

classicalHSEmotifsinpromoterstargetedbyHSF1andHSF2atthe physiologicaltemperatureandat38^◦C,whileat43^◦CHSEmotifs werefoundonlyintheHSF1-IPsample(Suppl.Table4).Usingthe gene-speciﬁcChIP-PCRapproachwevalidatedsuchtemperature- inducedremodelingof HSF1 andHSF2 bindingin promotersof selectedgenes(Hspa8,Hspe1,Hsph1,Spo11,Stip1,St13,Uspl1).The obtainedresultsconﬁrmedthatanelevationofthetemperature resultedinagradualincreaseinthebindingofHSF1andadecrease inthebindingofHSF2totargetHSEmotifs(Suppl.Fig.1).

3.3. HyperthermiacauseddisruptionofHSF1/HSF2interactions inmousetestes

Using the proximity ligation assay (PLA) we studied direct interactionsbetweenHSF1andHSF2(whichpotentiallyincluded heterotrimers)inmousetestesatthephysiologicaltemperature (32–33^◦C), and following heat shock at 38^◦C or 43^◦C (Fig.5).

TheHSF1/HSF2complexeswereclearlydetectedinspermatogonia, spermatocytesandspermatidsofcontroluntreatedanimals.Many complexes werelocatedon theboundarybetween thenucleus and cytoplasm, which suggests HSF1-HSF2 interactions in the dense(sex)body(Table2).Complexeslocalizedinthecytoplasm of elongatingspermatids near theluminal center of the cross- sectionswerealsoabundant(Fig.5C).Importantly,thenumberof HSF1/HSF2complexesdramaticallydecreasedfollowingtheheat

Table2

IntracellulardistributionofHSF1/HSF2complexesdetectedinsituinmousetestes atphysiologicalconditions.

Nucleus Boundary Cytoplasm

Spermatogonia 10% 35% 55%

Spermatocytes 38% 36% 26%

Roundspermatids 25% 45% 30%

shock,whichwasthemoststrikingafter15minutesofhyperther- miaat43^◦C(Fig.5AandG).After“mild”hyperthermia,performed at38^◦C,thechangesweresmallerandappearedgradually.After twohoursofrecoveryatthephysiologicaltemperaturethenumber ofdetectedHSF1/HSF2complexesstartedtorisefromthemini- mum.Thisindicatesareconstitutionofdirectinteractionbetween bothtranscriptionfactorsdisruptedafterexposuretoelevatedtem- peratures.

4. Discussion

InthepresentworkgenomicbindingsitesforHSF1andHSF2 transcriptionfactorswerecharacterizedinmousespermatocytes usingchromatinimmunoprecipitationcombinedwithnextgen- eration sequencing (ChIP-Seq). Several hundred actual binding sitesweredetectedatthephysiologicaltemperature,whichisin agreementwith previousreports describing theimportant role ofHSF1orHSF2inmousetestes(Akerfeltetal.,2008,2010).In fact,manylessHSF1bindingsiteswerefoundatthephysiological temperature using ChIP combined with promoter tiling arrays (Kus-Li´skiewicz et al., 2013).We suppose that such variability betweenbothtechnologiescanbeobservedwhenthebindingis weaker,which isthecase forHSF1bindingatthephysiological temperature. Whenthebindingis stronger (e.g.following heat shock),itcouldbeeasilydetectedusingeithermethod.ChIP-Seq allowedustoidentifythemajority(∼90%)ofHSFsbindingsitesin intronsandintergenicregions.Wehaveevidencethatthebinding ofHSF1tosomeintronscouldhavefunctionalimportance,forboth suppressionoractivationofgeneexpression(unpublished),yetthe roleofHSFsbindingoutsidepromotersismostlyspeculativeatthe moment.IthasbeensuggestedthatthelocationoftheHSF1binding (promoterversusdistalregions)couldbeconnectedwiththemode of regulation (positive versus negative, respectively) (Mendillo

(5)

Fig.3.RemodelingofHSF1andHSF2bindinginpromoterregionsofselectedgenes codingforHSPsorproteinsinvolvedinubiquitination,estimatedbytheChIP-Seq approach.HSFsbindingatthephysiologicaltemperature(C)andafterheatshockat 38^◦Cand43^◦CisvisualizedbypeaksbuiltwiththeIntegrativeGenomicsViewer abovetheschemeofthegeneorganization(lines–introns,boxes–exons).Approx- imately6kbisshown.Thescaleforeachsampleissetto0–50(exceptahighly enrichedHSF1bindingat43^◦C,whichisshownindividuallyintheﬁgureonthe right)tobettervisualizedifferencesinsampleswithlowerbinding.Ctr(−),negative controlwithoutspeciﬁcantibody.

etal.,2012).Furthermore,sinceHSFsareknowntoinitiatechro- matinremodeling(Sullivanetal.,2001;Jollyetal.,2004;Xingetal., 2005), itispossiblethattheirbindingtoDNAinintragenicand intergenicregionscouldhaveaninﬂuenceonthetranscriptionof noncodingRNAs,whichhasbeenshownforsomaticcells(i.e.atSat III)(Jollyetal.,2004).Itisassumedthat70–90%ofthemammalian genomeistranscribedinsomecontextsaslongnon-codingRNAs (lncRNAs),andthattranscriptionof thegenomeis substantially morewidespreadinthetestis(whereextensivechromatinremod- elingoccurs)comparedtosomatictissues(Soumillonetal.,2013).

Therefore, it is possible that in spermatocytes, the binding of

Fig.4.RemodelingofHSF1andHSF2bindinginpromotersofsomenon-Hspgenes.

HSFsbindingatthephysiologicaltemperature(C)andafterheatshockat38^◦Cand 43^◦CisvisualizedbypeaksbuiltwiththeIntegrativeGenomicsViewerabovethe schemeofthegeneorganization(lines–introns,boxes–exons).Approximately 7kbisshown.ThescaleforAcot7issetto0–50,forRsrp1andSpo11,0–30.Ctr(−), negativecontrolwithoutspeciﬁcantibody.

HSFsdetectedoutside theclassicalpromoter regionreflectsthe regulation of non-coding RNAs. Furthermore, some identified bindingsitescouldresultfromatransientsequence-independent chromatin binding corresponding tothe HSFs search for more specific targets, which mechanism was recently suggested for HSF1byHerbomeletal.(Herbomeletal.,2013).

Wefoundthatinspermatocytesatthephysiologicaltempera- ture,thepromotersofseveralgenesareco-occupiedbybothHSF1 andHSF2,whichindicatedtheimportanceofHSF1-HSF2crosstalk duringspermatogenesis.Inagreementwiththisobservation,the short-distanceproximity(<40nm)ofbothtranscriptionfactorswas detectedinspermatogeniccellsinsituatthephysiologicaltem- peratureusingtheproximityligationassay.Inthisassaythesignal observedinnucleimightpossiblycorrespondalsotoHSF1andHSF2

(6)

Fig.5. HSF1/HSF2complexesinmousespermatogeniccellsassessedbyProximityLigationAssay.(A)ThenumberofallHSF1/HSF2complexesperseminiferoustubule cross-section.Meanvalues±SDfromatleastsevensectionsfromtwo-threetestesarepresented.(B–I)Tubulecross-sections;HSF1/HSF2complexesarevisibleasredspots (asterisksindicateunspeciﬁcﬂuorescence),nucleiarestainedinblue.(B)Negativecontrolwithoutprimaryantibodies.(C–I)Testesofnottreatedorheat-shockedanimals.

SC–spermatocytes,RS–roundspermatids,ES–cytoplasmofelongatingspermatids;insetsshowblow-upsofselectedregions.

homotrimersboundtoDNAseparatelyatadistancesmallerthan

∼110–120bp(orkeptincloseproximitybychromatinlooping).

However,thesignalsobservedincytoplasmindicateanexistence ofdirectHSF1/HSF2interactions,mostpossiblyintheformofhet- erotrimers(orheterodimers).PutativeHSF1/HSF2complexeswere observedinthenucleus(presumably boundtoDNA)andinthe cytoplasm.Theywerealsofoundontheboundarybetweennucleus andcytoplasm,whichsuggeststheirlocalizationinthedense(sex) bodies,structuresassociatedwithsynapsisandtheformationof theXYbodyduringmeiosis.Therefore,ourobservationisinagree- mentwiththepreviousﬁndingthatbothHSF1andHSF2occupysex chromatinduringmeioticrepression(Akerfeltetal.,2008,2010).

Interestingly,manyHSF1/HSF2complexeswerelocalizednearthe luminalcenterofthecross-sectionsofthetesticulartubules,inthe cytoplasmofelongatingspermatids.Suchcytoplasmiccomplexes couldformbeforebothfactorsgainthecompetencetobindDNA and/orrepresentcomplexesreleasedfromDNA.Theexistenceof HSF1/HSF2complexesduringmousespermatogenesisatthephys- iologicaltemperaturehasbeenalreadyshownbySandqvistetal.

(Sandqvistetal.,2009).Additionally,co-localizationofHSF1and HSF2,mostprobablyintheformofheterotrimers,wasshownin

thenuclearstressgranules/bodies(nSBs),whichwereformedin response toheat shockinhumancells (Alastaloetal., 2003).It wasalsodemonstratedthatHSF1-dependenttranscriptioncould bemodulatedbytheHSF1/HSF2ratio,bothatthephysiological temperatureandduringstress(Heetal.,2003;Loisonetal.,2006;

Ostlingetal., 2007;Sandqvistetal.,2009).Thissuggestscoop- erationofHSF1andHSF2notonlyduring“normal”processesat thephysiologicaltemperature,butalsoduringresponsetostress.

AlthoughnSBs are notformed inrodent cells, one shouldcon- siderthecooperationofHSF1andHSF2intheregulationofstress responseinmousespermatogeniccells.

Geneswhosepromotersareco-occupiedbybothHSF1andHSF2 in mousespermatocytesencodemainlyfor chaperonesandco- chaperonesthatfacilitateproteinfolding.Thus,bothfactorscould participateintheregulationofthebasalleveloftranscriptionof thesegenesatthephysiologicaltemperature.Someofthesegenes couldbestillco-regulatedbybothfactorsduringmildhyperther- miaat38^◦C,yetstrongerhyperthermiaat43^◦Ccausedcomplete remodelingofHSFsbinding.Moststrikingly,temperatureelevation to43^◦CresultedinanincreasedbindingofHSF1topromotersof chaperonesandotherstress-relatedgenes,whileHSF2wasalmost

(7)

completelyreleasedfromthepromotersofsuchgenes;tonote, suchincreasedHSF1 bindingwasnotassociatedwithactivation oftranscription(Kus-Li´skiewiczetal.,2013).Inagreementwith thisobservation, thenumberof HSF1/HSF2complexesdetected in spermatogenic cells in situ by the proximity ligation assay wasmarkedly (∼10-fold)reduced at43^◦C.Temperature-related changesintheDNA-bindingabilityofHSF1andHSF2havebeen previouslyobserved usinginvitro experimentalmodel.Recom- binantHSF1acquiredHSE-bindingability(examinedbygel-shift assay)atatemperatureabove39^◦C(withmaximumat42–43^◦C), whileHSF2lostHSE-bindingstartingfrom39^◦C,uptoacomplete lossat43^◦C(Sargeetal.,1991).However,laterinvivostudiesusing differentsomaticcellsrevealedthatfollowingheatshock,atleast insomeHSPspromoters,bothHSFscouldbindatthesametime (Trinkleinetal.,2004;Ostlingetal.,2007;Ahlskogetal.,2010;

Shinkawaetal.,2011).Morerecently,aChIP-Seqstudyperformed inhumanK562erythroleukemiacellsalsoshowedtheinvolvement ofbothHSF1andHSF2intheregulationofgenescodingforchaper- onesandco-chaperones,wherethetotalnumberofHSF1andHSF2 targetlociwasincreasedaftertemperatureelevation(Vihervaara etal.,2013).HereweshoweddifferentinvolvementofHSF1and HSF2intheresponsetothermalstress,apparentlyspeciﬁcforsper- matogeniccells.Hence,bothfactorsplayadifferentroleinsomatic andspermatogeniccellsatnormal,physiologicalconditions,but alsobehavedifferently at anelevatedtemperature.It hasbeen shownthatoverexpressionofHSF1issufﬁcienttotriggerapopto- sisinspermatogeniccellsintheabsenceofactivationofHSPgenes (Vydraetal.,2006;Widlaketal.,2007).Thus,oneshouldassume thatdisturbancesinHSF1/HSF2interactionsandtheirchromatin bindingobservedinspermatogeniccellssubjectedtostronghyper- thermiahaveanapparentimpactontheviabilityofthesecellsin stressconditions.

5. Conclusion

DuringheatshockinmousetestesinteractionsbetweenHSF1 andHSF2aredisruptedandtheirbindingtochromatinisremod- eled.Thiscouldcontributetotheheatsensitivityofspermatogenic cells,sincethecooperationofbothfactorsisrequiredforcorrect spermatogenesis.

Acknowledgments

TheauthorsthankDrRyszardSmolarczyk,DrEwaMałusecka, andMrsUrszulaBojkoforexperttechnicalassistance.Thiswork wassupportedbythePolishMinistryofScienceandHigherEdu- cation(grantnumberNN301002439),andPolishNationalScience Centre(grantnumber2011/03/N/NZ3/03926).Allthecalculations werecarriedoutusingGeCONiIinfrastructurefundedbyproject numberPOIG.02.03.01-24-099/13.

AppendixA. Supplementarydata

Supplementary data associated with this article can be found,intheonlineversion,athttp://dx.doi.org/10.1016/j.biocel.

2014.10.006.

References

AhlskogJK,BjörkJK,ElsingAN,AspelinC,KallioM,Roos-MattjusP,SistonenL.

Anaphase-promotingcomplex/cyclosomeparticipatesintheacuteresponseto protein-damagingstress.MolCellBiol2010;30:5608–20.

AkerfeltM,HenrikssonE,LaihoA,VihervaaraA,RautomaK,KotajaN,SistonenL.

PromoterChIP-chipanalysisinmousetestisrevealsYchromosomeoccupancy byHSF2.ProcNatlAcadSciUSA2008;105:11224–9.

AkerfeltM,VihervaaraA,LaihoA,ConterA,ChristiansES,SistonenL,Henriksson E.Heatshocktranscriptionfactor1localizestosexchromatinduringmeiotic repression.JBiolChem2010;285:34469–76.

AlastaloT-P,HellesuoM,SandqvistA,HietakangasV,KallioM,SistonenL.Forma- tionofnuclearstressgranulesinvolvesHSF2andcoincideswiththenucleolar localizationofHsp70.JCellSci2003;116:3557–70.

BaileyTL.DREME:motifdiscoveryintranscriptionfactorChIP-seqdata.Bioinforma OxfEngl2011;27:1653–9.

FengJ,LiuT,QinB,ZhangY,LiuXS.IdentifyingChIP-seqenrichmentusingMACS.

NatProtoc2012;7:1728–40.

FredrikssonS,GullbergM,JarviusJ,OlssonC,PietrasK,GústafsdóttirSM,OstmanA, LandegrenU.Proteindetectionusingproximity-dependentDNAligationassays.

NatBiotechnol2002;20:473–7.

HeH,SoncinF,GrammatikakisN,LiY,SiganouA,GongJ,BrownSA,Kingston RE, Calderwood SK. Elevated expression of heat shock factor (HSF) 2A stimulatesHSF1-inducedtranscriptionduringstress.JBiolChem2003;278:

35465–75.

HeinzS,BennerC,SpannN,BertolinoE,LinYC,LasloP,ChengJX,MurreC,Singh H,GlassCK.Simplecombinationsoflineage-determiningtranscriptionfactors primecis-regulatoryelementsrequiredformacrophageandBcellidentities.

MolCell2010;38:576–89.

HerbomelG,Kloster-LandsbergM,FolcoEG,ColE,UssonY,Vourc’hC,DelonA, SouchierC.Dynamicsofthefulllengthandmutatedheatshockfactor1inhuman cells.PloSOne2013;8:e67566.

IzuH,InouyeS,FujimotoM,ShiraishiK,NaitoK,NakaiA.Heatshocktranscrip- tionfactor1isinvolvedinquality-controlmechanismsinmalegermcells.Biol Reprod2004;70:18–24.

JiZ-L,DuanY-G,MouL-S,AllamJ-P,HaidlG,CaiZ-M.Associationofheatshockpro- teins,heatshockfactorsandmaleinfertility.AsianPacJReprod2012;1:76–84.

JollyC,MetzA,GovinJ,VigneronM,TurnerBM,KhochbinS,Vourc’hC.Stress- inducedtranscriptionofsatelliteIIIrepeats.JCellBiol2004;164:25–33.

KampingaHH,HagemanJ,VosMJ,KubotaH,TanguayRM,BrufordEA,Cheetham ME,ChenB,HightowerLE.Guidelinesforthenomenclatureofthehumanheat shockproteins.CellStressChaperones2009;14:105–11.

Kus-Li´skiewiczM,Pola ´nskaJ,KorfantyJ,OlbrytM,VydraN,TomaA,WidłakW.

Impactofheatshocktranscriptionfactor1onglobalgeneexpressionproﬁles incellswhichinduceeithercytoprotectiveorpro-apoptoticresponsefollowing hyperthermia.BMCGenomics2013;14:456.

LangmeadB,TrapnellC,PopM,SalzbergSL.Ultrafastandmemory-efﬁcientalign- mentofshortDNAsequencestothehumangenome.GenomeBiol2009;10:R25.

LoisonF,DebureL,NizardP,leGoffP,MichelD,leDréanY.Up-regulationoftheclus- teringeneafterproteotoxicstress:implicationofHSF1-HSF2heterocomplexes.

BiochemJ2006;395:223–31.

MachanickP,BaileyTL.MEME-ChIP:motifanalysisoflargeDNAdatasets.Bioin- formaOxfEngl2011;27:1696–7.

MathewA,MathurSK,JollyC,FoxSG,KimS,MorimotoRI.Stress-speciﬁcactivation andrepressionofheatshockfactors1and2.MolCellBiol2001;21:7163–71.

MendilloML,SantagataS,KoevaM,BellGW,HuR,TamimiRM,FraenkelE,Ince TA,WhitesellL,LindquistS.HSF1drivesatranscriptionalprogramdistinctfrom heatshocktosupporthighlymalignanthumancancers.Cell2012;150:549–62.

NakaiA,SuzukiM,TanabeM.Arrestofspermatogenesisinmiceexpressinganactive heatshocktranscriptionfactor1.EMBOJ2000;19:1545–54.

OstlingP,BjörkJK,Roos-MattjusP,MezgerV,SistonenL.Heatshockfactor2(HSF2) contributestoinducibleexpressionofhspgenesthroughinterplaywithHSF1.J BiolChem2007;282:7077–86.

RupikW,JasikK,BembenekJ,WidłakW.Theexpressionpatternsofheatshockgenes andproteinsandtheirroleduringvertebrate’sdevelopment.CompBiochem PhysiolAMolIntegrPhysiol2011;159:349–66.

SalmandPA,JungasT,FernandezM,ConterA,ChristiansES.Mouseheat-shockfac- tor1(HSF1)isinvolvedintesticularresponsetogenotoxicstressinducedby doxorubicin.BiolReprod2008;79:1092–101.

SandqvistA,BjörkJK,AkerfeltM,ChitikovaZ,GrichineA,Vourc’hC,JollyC,Salminen TA,NymalmY,SistonenL.Heterotrimerizationofheat-shockfactors1and2 providesatranscriptionalswitchinresponsetodistinctstimuli.MolBiolCell 2009;20:1340–7.

SargeKD,ZimarinoV,HolmK,WuC,MorimotoRI.Cloningandcharacterization oftwomouseheatshockfactorswithdistinctinducibleandconstitutiveDNA- bindingability.GenesDev1991;5:1902–11.

ShinkawaT,TanK,FujimotoM,HayashidaN,YamamotoK,TakakiE,TakiiR, PrakasamR,InouyeS,MezgerV,etal.Heatshockfactor2isrequiredformain- tainingproteostasisagainstfebrile-rangethermalstressandpolyglutamine aggregation.MolBiolCell2011;22:3571–83.

SoumillonM,NecsuleaA,WeierM,BrawandD,ZhangX,GuH,BarthèsP,Kokkinaki M,NefS,GnirkeA,etal.Cellularsourceandmechanismsofhightranscriptome complexityinthemammaliantestis.CellRep2013;3:2179–90.

SullivanEK,WeirichCS,GuyonJR,SifS,KingstonRE.Transcriptionalactivation domainsofhumanheatshockfactor1recruithumanSWI/SNF.MolCellBiol 2001;21:5826–37.

ThorvaldsdóttirH,RobinsonJT,MesirovJP.Integrativegenomicsviewer(IGV):

high-performancegenomicsdatavisualizationandexploration.BriefBioinform 2013;14:178–92.

TrinkleinND,ChenWC,KingstonRE,MyersRM.Transcriptionalregulationandbind- ingofheatshockfactor1andheatshockfactor2to32humanheatshockgenes duringthermalstressanddifferentiation.CellStressChaperones2004;9:21–8.

VihervaaraA,SergeliusC,VasaraJ,BlomMAH,ElsingAN,Roos-MattjusP,Sistonen L.Transcriptionalresponsetostressinthedynamicchromatinenvironmentof cyclingandmitoticcells.ProcNatlAcadSciUSA2013;110:E3388–97.

VydraN,MaluseckaE,JarzabM,LisowskaK,Glowala-KosinskaM,BenedykK,Widlak P,KrawczykZ,WidlakW.Spermatocyte-speciﬁcexpressionofconstitutively

(8)

activeheatshockfactor1inducesHSP70i-resistantapoptosisinmalegermcells.

CellDeathDiffer2006;13:212–22.

VydraN,TomaA,WidlakW.PleiotropicroleofHSF1inneoplastictransformation.

CurrCancerDrugTargets2014;14:144–55.

WangG,ZhangJ,MoskophidisD,MivechiNF.Targeteddisruptionoftheheatshock transcriptionfactor(hsf)-2generesultsinincreasedembryoniclethality,neu- ronaldefects,andreducedspermatogenesis.GenesNYN2003;2000(36):48–61.

WangG,YingZ,JinX,TuN,ZhangY,PhillipsM,MoskophidisD,MivechiNF.Essential requirementforbothhsf1andhsf2transcriptionalactivityinspermatogenesis andmalefertility.GenesNYN2004;2000(38):66–80.

WidlakW,VydraN,MaluseckaE,DudaladavaV,WiniarskiB,Sciegli ´nskaD,Wid- lakP.Heatshocktranscriptionfactor1down-regulatesspermatocyte-speciﬁc

70kDaheatshockproteinexpressionpriortotheinductionofapoptosisin mousetestes.GenesCellsDevotedMolCellMech2007;12:487–99.

WidłakW,BenedykK,VydraN,GłowalaM,Sciegli ´nskaD,MałuseckaE,Nakai A,KrawczykZ.Expressionofaconstitutivelyactivemutantofheatshock factor1underthecontroloftestis-speciﬁchst70genepromoterintrans- genicmiceinducesdegenerationofseminiferousepithelium.ActaBiochimPol 2003;50:535–41.

XingH,WilkersonDC,MayhewCN,LubertEJ,SkaggsHS,GoodsonML,HongY, Park-SargeO-K,SargeKD.Mechanismofhsp70igenebookmarking.Science 2005;307:421–3.

YinY,HawkinsKL,DeWolfWC,MorgentalerA.Heatstresscausestesticulargerm cellapoptosisinadultmice.JAndrol1997;18:159–65.