Research Article 351 The constant beat: cardiomyocytes adapt their forces by equal contraction upon environmental stiffening Nils Hersch, Benjamin Wolters, Georg Dreissen, Ronald Springer, Norbert Kirchgeßner*, Rudolf Merkel and Bernd Hoffmann` InstituteofComplexSystems,ICS-7:Biomechanics,ForschungszentrumJu¨lichGmbH,52425Ju¨lich,Germany *Presentaddress:InstituteofAgriculturalSciences,ETHZu¨rich,Universita¨tsstraße2,8092Zu¨rich,Switzerland `Authorforcorrespondence([email protected]) BiologyOpen2,351–361 doi:10.1242/bio.20133830 Received13thDecember2012 Accepted23rdDecember2012 Summary Cardiomyocytesareresponsibleforthepermanentbloodflow contraction amplitude remains stable, which in turn leads to by coordinated heart contractions. This vital function is increased force levels allowing cells to adapt almost accomplished over a long period of time with almost the instantaneously to changing environmental stiffness. sameperformance,althoughheartproperties,asitselasticity, Furthermore, our data strongly indicate specific adhesion to change drastically upon aging or as a result of diseases like flat substrates via both costameric and focal adhesions. The myocardialinfarction.Inthispaperwehaveanalyzedlaterat generalappearanceofthecontractileandadhesionapparatus embryonic heart muscle cells’ morphology, sarcomere/ remainsalmostunaffectedbysubstratestiffness. costamere formation and force generation patterns on substratesofvariouselasticitiesrangingfrom,1to500kPa, (cid:2)2013.PublishedbyTheCompanyofBiologistsLtd.Thisis n which covers physiological and pathological heart stiffnesses. an Open Access article distributed under the terms of the e p Furthermore,adhesionbehaviour,aswellassinglemyofibril/ CreativeCommonsAttributionNon-CommercialShareAlike O sarcomere contraction patterns, was characterized with high License (http://creativecommons.org/licenses/by-nc-sa/3.0). y spatial resolution in the range of physiological stiffnesses g o (15 kPato90 kPa).Here,sarcomereunitsgenerateanalmost Keywords: Cardiomyocyte, Tractionforce microscopy, Cell ol stable contraction of ,4%. On stiffened substrates the adhesion, Mechanoresponse, Myofibril,Sarcomere Bi Introduction processes of those cells go along with extensive functional Mechanical properties of tissues strongly affect embedded changes. In contrast, myocardial embedded cardiomyocytes, mammalian cells. The consequences extend from simple, small responsible for the regular heartbeat, do not seem to adapt adaptations of rather isolated cellular structures all the way to morphologicallyandfunctionallytosuchenvironmentalchanges. initiation of very complex differentiation pathways. One of the Cardiomyocytes of healthy myocardial tissue interact most potent mechanical signals is the elasticity of the mechanically with their environment via costameric adhesions environment. Various cell types respond with simple to surrounding extracellular matrix molecules (ECM) and via reinforcements of adhesion and cytoskeletal structures upon cell–cell contacts at intercalated disks to other myocytes. In substratestiffening(Giannoneetal.,2003;Krishnanetal.,2009). disease affected tissues adhesion structures of cardiomyocytes More extensive adaptations have been described for elasticity become remodeled. As a consequence, adhesion sites of dependent stem cell differentiation. Here, cell fate could be cardiomyocytes increase at sites of scar formation to the ECM directed by substrate stiffness in direction of neuronal, bone or aswellastonon-spontaneouslycontractingcellslikeendothelial muscle progenitors(Discher et al., 2009; Engleret al., 2006). In cells or myofibroblasts (van den Borne et al., 2010). Resulting addition, various other cell types undergo further differentiation changes in adhesion and heart tissue elasticity can affect the processes after environmental stiffening without changing their volume of blood filling the heart and therefore also the stroke tissue localization. Prominent examples can be found in volume according to the Frank–Starling mechanism and the mammalian heart myocardial tissue, especially during aging force–velocityrelation(Fukudaetal.,2001;MartynandGordon, andafterinfarctionorothercoronaryheartdiseases(Herrmannet 2001;RassierandPavlov,2010).Elevatedheartloadgoesalong al., 2003; Hinz, 2009; Koltai et al., 1984; Mazhari et al., 2000). with enhanced forces during cardiomyocyte contraction due to Whileelasticitiesofhealthymyocardialtissuesrangefrom10to cellstretchingandthereforeincreasedsarcomerelengthsasforce 30 kPa, stiffnesses in heart stroke affected areas increase to generating subunit of myofibrils from originally ,1.9 mm to values of up to 150 kPa (Berry et al., 2006; Gupta et al., 1994; extended lengths of up to 2.4 mm (Rassier and Pavlov, 2010). Jacot et al., 2010). Triggered by this stiffening myocardial Even isolated neonatal cardiomyocytes recognize substrate fibroblasts differentiate together with bone marrow stem cells elasticity as very potent signal inducing maturation and intomyofibroblasts,leadingtoscartissueformationwithdrastic functional development. This was shown, for example, changes of myocardial physical properties. Differentiation examining cells grown on substrates mimicking the natural Elasticity-driven muscle forces 352 elasticity of ,10 kPa and on stiffer ones (50 kPa). In the latter cardiomyocytes adhere during maturation or how and where case sarcomere alignment was reduced in comparison to the remodeled adhesion sites transmit traction forces. Adaptations in former. Additionally, cellular forces were reported to be highest morphology, structure and composition of adhesion structures to on 10 kPa substrates and reduced on softer and stiffer materials substrate elasticity are also not fully elucidated. In this paper we (Jacot et al., 2008; Jacot et al., 2010). For neonatal haveaddressedtheseimportantquestionsofadhesionbehaviorand cardiomyocytes grown on elastomeric micropillars, twitch forcegeneration/transmissionofratlateembryoniccardiomyocytes. forces increased on stiffened posts while twitch velocities Cellsweregrownonwell-characterizedsiliconerubbersubstratesof leveled down. Elevated forces were accompanied by increased various elasticities in the natural range. Traction force sarcomere lengths and z-band widths (Rodriguez et al., 2011). measurements were performed and analyzed in depth using two Other studies indicate that beating frequency and fraction of independent algorithms. Special care was taken to achieve high isolated cardiomyoctes corresponded best to natural spatialresolutiontodeterminetheforcestransmittedateverysingle characteristics on substrates resembling native myocardial putative adhesion structure. Moreover, the local contractile strain elasticities (Bajaj et al., 2010; Engler et al., 2008). Remarkably, alongsinglemyofibrilswasdetermined. several aspects of how cardiomyocytes respond to substrate In summary our data prove that cultured late embryonic rat elasticity are still under debate since other studies indicate a cardiomyocytes exhibit very stable contraction amplitudes of completelossofmyofibrilsonsoft1 kPasubstrates(Bajajetal., sarcomeric units along separate myofibrils. This behavior 2010),anelevated number of beating myocytesafter maturation generates increased force values upon substrate stiffening. At all on very soft substrates (below 1 kPa) (Shapira-Schweitzer and substrate elasticities analyzed adhesion and force transmission Seliktar, 2007) or increasing force values upon substrate occuralongthewholelengthofmyofibrilsatsitesofcostameres stiffening to up to 144 kPa (Bhana et al., 2010). Furthermore, and at remodeled myofibril ends. Interestingly, in the atomic force microscopy measurements indicate a completely physiological elasticity range substrate stiffness did not affect unaffected cell elasticity of cardiomyocytes on substrate thelengthsofrelaxedorcontractedsarcomericunits.Thisisnotin elasticities ranging from 8 to 50 kPa (Shi et al., 2011). agreementwiththeFrank–Starlingmechanismandindicatesthat, Afterisolationfromtissueandgrowthasisolatedcellsonflat despite it being well established in tissue, various regulatory substratesmyocytesdisplayanalteredadhesionpattern(Hilenski mechanisms connected with this mechanism might be et al., 1992; Simpson et al., 1993). This adaptation includes incompletelyactiveinisolatedcardiomyocytes. n e reorganization of costameric structures, resulting in adhesive p sites that are structurally similar to focal adhesions (FA), which Results O are typically found in non-muscle cells (Decker et al., 1990). The sarcomericforce generationsystem isformed y g Simultaneously, intercalated disks at the end of myofibrils independentlyof environmental elasticity o remodel to adherens junctions as well as to FA like structures. In order to analyze contractile and adhesive behavior of ol Bi Allthesestructuresprovidestrongadhesionsitesresponsiblefor cardiomyocytes in an elasticity range of ,1 to 500kPa silicone mechanicalcommunicationbetweentheECMandthecontractile rubber substrates were fabricated. In a first step, late embryonic machinery of the cultured cells (Goncharova et al., 1992; myocytes were grown for three days on substrate elasticities Simpson et al., 1993). The disassembly of the contractile spanning the physiological range of 15 to 150kPa and, apparatus upon isolation and a reassembly from pre-myofibrils additionally, on artificially very soft (,1kPa) or very stiff afterattachmenttocellculturesubstratesseemtobedecisivefor substrates (500kPa) and analyzed for differences in cell shape. this environmentally controlled adaptation (Du et al., 2003; When grown on substrate elasticities of 15kPa and stiffer a large Sanger et al., 2005). percentageofcellswerecharacterizedbyashapeindexsibelow0.8, Previous work has shown that both environmental stretch and correspondingtoamoreangularshape(Fig.1A,B).Just30%ofall cell contractility increase size and frequency of adhesion cellsexhibitedaroundish,mostlyconvexshape(siabove0.8).On structures (Miller et al., 2000; Sharp et al., 1997; Simpson et verystiffsubstratesalmostallcellswereangularinshape(fractionof al.,1993).Furthermore,inearlyworkDanowskiandco-workers roundish cells below 10%). Only on very soft substrates (,1kPa) cultivated adult cardiomyocytes on a thin and flexible silicone theratiobetweenroundandangularshapedmyocyteswasbasically membrane and observed pleat-like wrinkles with a spacing of invertedwith,80%ofcellsexhibitingaconvexoutline(Fig.1A,B). ,1.9 mm (Danowski et al., 1992). Pleats coincided with the Sinceedgecurvaturedependsonthemechanicalstressformed distribution and distance of z-disks. These observations clearly withinthenetworkofinneractinfibers(Bischofsetal.,2009),we indicated the transmission of mechanical force at sites of analyzed whether sarcomeres, the force generating units of costameric adhesions. Later traction force microscopy myocytes, were differently formed or organized depending on measurements were evaluated with the assumption that forces substrate stiffness. To this end we used a-actinin as marker were exclusively applied at FAs at the ends of myofibrils and protein (Fig. 2). Interestingly, myocytes always displayed stress fibers. These analyses resulted in force values of ,15 nN regularly striated myofibrils irrespective of substrate elasticity per FA and force dipole values in the 6 pNm range (Balaban et (,1 kPato500 kPa).Thisobservationwasconfirmedwithactin al., 2001; Cesa et al., 2007). The force dipole, also called as marker (Fig. 2) indicating that the whole force generating generalizedfirstmoment,isjustthesumofthescalarproductsof system is functional and largely unaffected by changes in the vectors force and its position. For a cell of the size of a environmental elasticity. More elaborate analyses of myofibril cultivated cardiomyocyte a force dipole of 6 pNm roughly formation within the most relevant physiological range of corresponds to overall contractile forces in the 200 nN range. substrate elasticities between 15 and 90 kPa revealed that Although cardiomyocytes have been intensively analyzed for neither the number of costameres per cell, nor the mean area more than 20 years, several important questions are still not yet pera-actininclusterorthefractionofthecellarealabeledbya- answered. For example, it is not entirely clear how neonatal actinin was significantly affected by substrate elasticity (Fig. 3). Elasticity-driven muscle forces 353 Fig.1. Substrateelasticitydependentcellshapeadaptationofmyocytes. Fig.3. Substrateelasticityindependentformationofcostameres.GFP-a- Freshlyisolatedratcardiomyocyteswereincubatedfor3daysonsilicone actinintransfectedcells(A)weregrownfor3dayson15kPa,30kPaand rubbersubstratesrangingintheirelasticityfrom,1to500kPaasgiven. 90kPasubstrates.Livecellimagesweresegmentedusingimageprocessing n Subsequently,cellswereanalyzedforeitherround(shapeindex,si.0.8)or (B).Scalebar:10mm.Meana-actininclustersize(C),relativecoverageof e elongated,angular(si#0.8)cellshape(A)usingimageprocessing.Left cellswitha-actininclusters(D)andthetotalnumberofa-actininclustersper Op micrographsi50.96,rightmicrographsi50.45.Scalebars:50mm.Distribution cell(E)weredetermined.Allresultsareindicatedwiththeirrespectives.d.; betweenbothshapesisgiveninB.Roundcellshape5darkgreyandelongated n516cellsfor15kPa,20cellsfor30kPaand13cellsfor90kPa. y shapedcells5lightgrey.n5numberofanalyzedcells.Meanvalues(black) g includings.d.(grey)ofshapeindicesforroundandelongatedcellsaregiven o ol nexttothebars. With respect to the cytoskeleton, the only effect observed on Bi artificially soft ,1 kPa substrates was a slightly reduced order andalignmentofindividualmyofibrilswithrespecttoeachother. Since reduced order of myofibrils might indicate an altered coupling of costameres to intracellular structures and also to the outer environment, we additionally analyzed the localization of plectin. This protein is vital for the mechanical coupling of sarcomericborders(z-bands)totheintermediatefilamentprotein desminand,viathislink,forthecouplingbetweenmyofibrilsand the whole dystrophin/glycoprotein complex embedded in the plasma membrane (Konieczny et al., 2008). These analyses revealedalocalizationofplectinatsitesofsarcomereseparating z-bands on all substrate elasticities tested. Interestingly, while two distinct small plectin spots were found on both sides of z- bands on ,1 kPa, these spots visibly enlarged in size in an elasticity dependent manner (Fig. 4A). This structural reinforcement resulted in plectin signals co-localizing with actinalongitsfullsarcomericlengthonstiff500 kPasubstrates. The structural studies performed above suggest a functional force generation system as well as a proper connection of myofibrils to the plasma membrane. However, whether plasma membrane connected costameres also interacted with the extracellular matrix was still unclear. We therefore cultivated myocytesonsubstrateswithelasticitiesfrom,1 kPato500 kPa and stained them for vinculin as marker for FA as well as for integrin-dependentcostamerelinkedadhesionsites.Asshownin Fig.2. Stableformationoftheforcegeneratingsystemofmyocytes. Fig. 4B,thesemicrographsdisplaydistinctsignalsatbothendsof Cardiomyocytesweregrownon,1kPa,15kPaand500kPastiffelastomeric every single myofibril. Furthermore, vinculin co-localized with substratesfor3daysandanalyzedafterfixationformyofibrilformationusing costameres. These signals were variable in intensity between a-actinin(left)andactin(middle)asmarkerproteins.Overlayimagesare given(right).Scalebars:50mm. individual cells and slightly blurred on very soft substrates. Elasticity-driven muscle forces 354 Fig.4. Myofibrilalignmentandplasmamembraneattachment. Cardiomyocytesweregrownfor3daysonsubstratesofindicatedelasticities. Subsequently,cellswerefixedandstainedforplectinandactin(A)orvinculin andactin(B).InsetsinAshowenlargementsoftheareaswithinthewhite squares.Notetheelasticitydependentreinforcementofplectinandvinculin aroundsitesofz-bands.Scalebars:20mm. Nevertheless, these vinculin stainings might argue for an interaction with integrin, and therefore with the environment, at n sites of costameres on each substrate elasticity tested. e p O Forceistransmittedatsitesofcostamericandfocaladhesions y Tofurthertesttheadhesivebehaviorofcardiomyocytes,traction og force microscopy was performed on spontaneously contracting Fig.5. Costamericandfocaladhesiondependenttractionforcesof ol GFP-a-actinin transfected cells grown on 15 kPa, 30 kPa, and myocytes.GFP-a-actinintransfectedcardiomyocytes(A–E)andGFP-VASP transfectedmyofibroblasts(F–J)weregrownonfluorescentbead Bi 90 kPa stiff substrates (see also supplementary material Movie micropatterned15kPa(exemplarilyshownhere),30kPaand90kPa 1).Here,forcecalculationwasbasedontheassumptionofpoint elastomericsubstrates.Spontaneouslycontractingmyocyteswereanalyzedby forces acting at interactively chosen adhesion sites. From the livecellmicroscopytovisualizethesubstratedeformationfield(A,green arrows)uponcontraction.Usinginteractivelychosenforceinductionpointsand forces determined in this way substrate deformation fields were elasticitytheorycellforces(notshown)andtheresultingbackcalculated back calculated (Fig. 5, top) and compared with the actual substratedeformations(A,yellowarrows)weredeterminedforeachtimepoint. measured displacements. From those calculations residual Asresultwereceivedaresidualdisplacementvectorforeachmarkerbead displacements for each tracked marker could be constructed. (B–D,right,bluearrows).Choosingasforceinductionpoints(yellowsquares) eitherFAsonly(B),costameresonly(C)orbothtogether(D)backcalculations WheneverexclusivelyFAswerechosenasforceinductionpoints reproducedtheactualdeformationfieldwithverydifferentaccuracy.Frame major residual displacements remained present below the center rate517Hz.Ascontrol,measuredsubstratedeformationsandbackcalculated of cells (Fig. 5B,E). Vice versa, choosing only costameres as substratedeformationsofstablyadheredmyofibroblasts(F)weredeterminedas force induction points resulted in back calculated displacements describedformyocytesusingFAs(G),alatticemimickingthelengthoftwo sarcomeres(,4mm)alongstressfibers(H)andboth(I)asforceinduction that fitted the measured ones well below the interior of the cell points.Here,fordisplacementcalculationanimageofthesamesubstratearea while major residuals remained at the periphery. (Fig. 5C,E). wastakenaftercellremoval.Foreachbackcalculationthenormalizedsquared Only combining FAs and costameres resulted in low residual deviation(x2)betweenfitresultandmeasureddatawasdeterminedas displacements over the whole field (Fig. 5D,E). exemplarilygiveninE,J.Notethatformyocytesresidualdeformationsarelow onlyifbothFAsandcostameresaretakenasforceinductionpoints,whilefor As additional control we identically analyzed similar traction myofibroblastsFAsalonearefullysufficient.Scalebarsfordisplacements force experiments on myofibroblasts, a cell type well known to (yellowandgreen)inA,F:0.5mm;forthecellsize(white):10mm;andfor adhereexclusivelyviaFAswithstrongcontractilestressfibersin residualdeformation(lightblue):0.2mm. between.WheneverweusedasforceinductionpointseitherFAs alone (Fig. 5G,J), or together with them a fictitious pattern, adhesion and force transmission of cardiomyocytes at sites of similartotheonesusedinmyocytesforcostameres,alongstress FAs and costameres. fibers (Fig. 5I,J), we found very low residual displacements. In contrast, high residual displacements resulted if exclusively Forcegeneration increaseswith substrate stiffening pointsalongstressfiberswereused(Fig. 5H,J).Interestingly,for Based on the results given above we evaluated the sum of all this cell type combining all types of force induction points was contractile forces of GFP-a-actinin transfected cells using either not further reducing the residual displacements. Thus our both FAs and costameres as force induction points or, implementation of traction force microscopy was well able to alternatively, a hexagonal lattice of fictitious points with a discriminate between the different modes of force application. lattice constant of 5 mm, covering the whole area of the cell Taken together, these data clearly prove the simultaneous (Fig. 6A,B). The latter choice was necessary to increase the Elasticity-driven muscle forces 355 In cell force retrieval algorithms based on force fields the completeinterfacebetweencellandsubstrateisusedasputative forceinductionarea.Here,thesameelasticitydependentincrease of generated cell forces was found albeit with slightly different absolute values (Fig. 7D; supplementary material Fig. S1B). Furthermore, substrate deformation and force fields of GFP-a- actinintransfectedcells(Fig. 7A)werenotjustlocalizedmainly at FA positions but also along most parts of myofibrils independent on substrate elasticity (Fig. 7B,C). These data confirm once more that adhesion of myocytes occurs simulta- neously at FAs and costameres. Contractionsof separatemyofibril sarcomeric unitsarestable Elevating force values on stiffening substrates with a virtually unaffectedcellforcegeneratingsystemopenedthepossibilityof elasticity adapted contractile amplitudes of myofibrils. If this increaseinforcewasbasedonaFrank–Starlinglikemechanism, it should be accompanied by elasticity dependent sarcomere Fig.6. Stiffnessdependentforcegenerationofmyocytes.Aftergrowthof lengths and contractions (de Tombe et al., 2010; Hanft et al., myocytesfor3daysonbeadmicropatternedsubstrates(A)withelasticities 2008). For this reason we analyzed the contraction patterns of rangingfrom,1to500kPa(D),tractionforcemicroscopywasperformedon discretemyofibrilsoverseveralcontractions(exemplarilyshown spontaneouslycontractingcells.Applyingelasticitytheorybasedonsubstrate inFig. 8A).Thiswasdoneforcellsgrownonsubstratesspanning deformations(B,yellowarrows)withahexagonalgridusedasforceinduction the most relevant natural elasticity range of 15 to 90 kPa. points(A),forcespergridpoint(B,redarrows)weredeterminedforevery imageofthetimestack.Scalebars:10mm.Thetimecourseofthesumofall Foreverymyofibrilwecouldidentifyacentralspotthatbarely contractileforcesfortheexemplarilygivencellisindicated(C).Maximum movedduringcontraction(Fig. 8B,C,blackarrowhead).Onboth valuesofthesumofallcontractileforceswereaveragedforallcellsper sides GFP-a-actinin labeled z-bands displaced upon contraction n substrateelasticityandareindicatedwiththerespectives.d.inDusinga e logarithmicsubstratestiffnessscale.Thedottedcurvedisplaysafitted towardsthemyofibrilcenter.Thisz-banddisplacementalongthe Op proportionalforceincrease.Thedashedcurveindicatesafittedlinearforce filament(Fig. 8C)revealedastablecontractilestrain,i.e.relative increase.n>15cellsperelasticity. shortening, for about the central half of each myofibril. As y g contractionssumupalongthefibrilthiscorrespondstoalinearly o increasing displacement of the more outside located sarcomeres ol number of analyzable cells since transfection rates of vital Bi myocytesarealwaysverylow.Onallsubstrateelasticitiestested within the inner part of myofibrils (Fig. 8B,C). Interestingly, displacementsofoutmostsarcomericunitsoftenleveleddownto bothmethodsrevealedhighlycomparablecontractileforceswith a plateau like displacement caused most likely by simple surprisingly small differences (Table 1). We therefore analyzed transmission of inner myofibril displacement without additional cell forces of single myocytes on substrate elasticities in the contractility (Fig. 8C). range of ,1 to 500 kPa using point forces on such a hexagonal In order to characterize the exact contractile behavior of grid. Independent of substrate elasticity, data prove stable discrete myofibrils, we automatically detected the linear region spontaneous myocyte contractions over time (Fig. 6C). In good of each displacement-position curve. Within this region a line agreement with former analyses (Jacot et al., 2008), these data was fitted to the displacements. The slope of this line is the revealed low force generation of myocytes grown on soft contractile strain (relative shortening) of the myofibril (see substrates (Fig. 6D; supplementary material Fig. S1A). With Materials and Methods). Averaging slope values of 2 to 4 increasing stiffness, contractile forces rose continuously from independent myofibrils per cell characterized itsmean myofibril valuesofonly10 nN(s.d.36 nN,n522)on,1 kPasubstratesto contractile behavior (Fig. 8D). Mean values for various cells more than 800 nN (s.d. 311 nN, n514) on 500 kPa stiff grown on 15 kPa, 30 kPa or 90 kPa revealed a large spread in substrates. Within the physiological range from 15 to 150 kPa cellaveragedmyofibrilcontractilestrainbetween2%and8%for thisincreasewasalmostlinear.Increasingcontractileforcesupon 15 kPa substrates (n57 cells with 15 myofibrils), 2% and 12% substratestiffeningwent alongwithunchangedaverage contrac- for30 kPasubstrates(n513cellswith29myofibrils)and1%and tion frequencies, with 38 (s.d. 18) spontaneous contractions per 6% for 90 kPa substrates (n58 cells with 20 myofibrils) minute on soft (,1 to 15 kPa, n530 cells) and 36 (s.d. 18) on (Fig. 8D). However, with contraction values of ,4% (4.7% s.d. stiff (150 to 500 kPa, n530 cells) substrates. 2.1% for 15 kPa, 4% s.d. 2.9% for 30 kPa, 3% s.d. 1.7% for Table 1. Comparison of the averaged sum of all contractile cell forces using either FA and costameres or alternatively a hexagonal lattice as force induction points for cells grown on 15, 30 and 90 kPa stiff substrates. s.d. is given in parentheses. n indicates number of cells analyzed. Elasticity[kPa] FA+costamer[nN] Hexagonallattice[nN] Difference[%] 15n516 74(24) 74(26) 3.9(8.4) 30n520 174(94) 175(93) 2.3(1.9) 90n513 262(82) 261(80) 2.3(3.3) Elasticity-driven muscle forces 356 markedlydifferentappearanceofthecontractileapparatus.Such elasticity independent behavior is in contrast to results reported formanyothercelltypes:fibroblastsaswellasendothelialcells, for example, respond to increasing substrate elasticities with reinforcement processes of cytoskeletal stress fibers as well as adhesion structures (Goffin et al., 2006). Furthermore, several cell types as stem cells or cardiac fibroblasts differentiate in a substrateelasticitydependentmannerwithprominentchangesin almostalltypesofcytoskeletalandadhesionstructures(Engleret al., 2004; Engler et al., 2006; Gilbert et al., 2010). Atthismomentwecanonlyspeculateaboutthereasonforthis constant behavior of myocytes. Since we have used late embryonic rat myocytes (d19) most isolated cells are already partially differentiated and able to contract spontaneously as single cells. Beating itself is a mechanical signal that has been shown to stimulate further muscle cell differentiation and myofibril formation (Damon et al., 2009; Miller et al., 2000). It might therefore be possible that myocytes beyond a certain Fig.7. Forcefieldsalongthecontractileapparatus.GFP-a-actinin differentiation state generate their own mechanical late transfectedmyocytes(A)weregrownfor3daysonbeadmicropatterned substrateswithelasticitiesasindicatedin(D).Substratedeformationfields differentiation signal causing the observed elasticity weredetermined(B)andentireforcefieldswereretrieved(C).Scalebar: independent myofibril formation. Such a behavior would be 10mm.Thesumofallmaximumcontractileforceswereaveragedoverallcells physiologically advantageous. In the myocard cardiomyocytes analyzedpersubstrateelasticityandareindicatedwiththerespectives.d.inD arepermanentlyformed,althoughatlowrates,bydifferentiation usingalogarithmicsubstratestiffnessscale.Thedottedcurvedisplaysafitted proportionalforceincrease.Thedashedcurveindicatesafittedlinearforce of tissueembeddedprecursor cells. Thishappensthroughoutthe increase.n>20cellsperelasticity. whole lifetime of the organism. Because on one hand cardiomyocyte function critically depends on the structure of n e 90 kPa, Fig. 8D) overall averaged myofibril contractile strains theirforce producingapparatusand,ontheother hand,myocard p stiffness varies enormously during aging or disease progression, O showed no significantly different values for all substrate myofibril formation should not depend on environmental y elasticities analyzed. Notably, substrate elasticity did also not g influence sarcomeric length for the inner myofibril region. No stiffness. However, it needs to be mentioned that published o results on substrate elasticity dependent functional and ol matter which elasticity was analyzed, contraction started from morphological adaptations of cardiomyocytes are inconsistent Bi basically identical sarcomeric lengths of 1.9 mm in relaxed state to some extent (Bajaj et al., 2010; Bhana et al., 2010; Engler et (s.d. 0.13 mm for 15 kPa, s.d. 0.17 mm for 30 kPa, s.d. 0.11 mm al.,2008;Jacotetal.,2008;Shietal.,2011).Itmightwellbethat for 90 kPa). species of origin, exact time point of isolation and even the These data clearly point to a morphological and functional chemical nature of the cell culture substrate have unexpectedly highly stable myofibril contractile mechanism that is basically strong impact. unaffectedandrobustagainstchangesinenvironmentalstiffness. If plated on soft substrates, contractile forces of our cardiomyocytes were very low. This result is in very good Discussion agreementwithvariousotherstudies(Engleretal.,2008;Jacotet Cardiomyocytes are the mechanically most active cell type in al., 2008). With increasing substrate elasticities contractile cell mammals. Over decades these cells pump almost constant forces rose. Force values did not peak at a distinct elasticity but volumes of blood although environmental conditions within the insteadgraduallyincreased.Thisbehaviorresultedin,80times myocard can change drastically. For example, the molecular highercontractileforceson500 kPasubstratescomparedtovery compositionof the extracellular matrix and the relative amounts soft ,1 kPa ones. Even for the more natural elasticity range of cell types populating it change substantially during between 15 and 90 kPa this increase was still ,4-fold. This embryogenesis, aging or diseases like heart strokes. Most of almostlinearbehaviorisinmarkedcontrastwithnon-monotonic these changes go along with alterations in tissue elasticity behavior reported by Engler and co-workers and Jacot and co- causingstiffeningfromelasticitiesinthe10 kParangefoundfor workers, who found a maximum of forces at close to nature healthy young myocard to values of up to 100 kPa and substrate elasticity (Engler et al., 2008; Jacot et al., 2008). A sometimes even higher (Berry et al., 2006; Gupta et al., 1994; probablereasonforthiscontrastisthattheircellswereisolatedat Jacot et al., 2008). muchearlierstagesofembryonicdevelopmentcomparedtoours. In light of this background it is even more surprising that Cardiomyocytes isolated at identical developmental state cardiomyocytes are morphologically and functionally barely generated highly comparable contractile forces when tested on affected by substrate elasticity. In the present study we could 12 kPa substrates (Balaban et al., 2001). Possibly due to the clearly show that myocytes from late rat embryos form electricalstimulationperformedinexperimentsbyJacotandco- contractile myofibrils by a very stable mechanism. These fibrils workers, forces generated by single myocytes reported by them arenotvisiblyaffectedinnumber,lengthorregularityofstriation (Jacot et al., 2008) were higher on very soft substrates while when cells are grown on elastomeric substrates with stiffnesses comparable to our results on stiffer ones. covering the complete natural range. Even extreme elasticity Interestingly, with values in the range of 3 to 10 mN overall valuesofverysoft,1 kPaorverystiff500 kPadidnotresultin forces reported for isolated skinned myocytes are much higher Elasticity-driven muscle forces 357 (Diffee and Chung, 2003; Korte and McDonald, 2007) than the myocardial tissues. Such cardiomyocytes are characterized by highest contractile force of 1 mN observed here on stiff well-aligned myofibrils spanning the whole cell in a defined substrates. This discrepancy is most likely due to several direction and are therefore different from the prenatal aspects. Most importantly, effective Ca2+ concentrations cardiomyocytes used here. Finally, traction force microscopy is available to induce myosin activity are clearly different and not well suited to determine the highest possible contraction optimalforskinnedcellscausingmaximumcontractionsinthose forces of cardiomyocytes. These forces are reached under cells. Furthermore, cardiomyocytes used for skinned cell isometric conditions where the contraction amplitude vanishes. preparations are usually isolated from adult, fully differentiated While this situation might be approached using very stiff substrates, the resulting tiny substrate deformations cannot be detected anymore. Inadditiontoforces,wealsocalculatedtheworkperformedby myocytes during one contraction according to Butler and co- workers(Butleretal.,2002).However,averyhighscatterofthe resultingvalues(about1to5 fJ)maskedallpossibledependence on substrate stiffness. Our results point at the elasticity independent presence and functionofmyofibrilsasreasonforthecontinuousforceincrease with substrate stiffening. These fibrils are already present in rat cardiomyocytes isolated at day 19 of embryonic development. Subsequent incubation on any substrate elasticity used here had no significant effect on morphology or function of this force generating structure. Although primary cardiomyocytes are by nature a highly heterogeneous cell pool and, additionally, our functional analyses were performed on the mechanically most activesubfraction,wenotednodependenceoftheoverallnumber ofmyocytesorthenumberofspontaneouslycontractingcellson n e substrateelasticity.Substrateelasticitydependentbeatfrequency p analyses as done elsewhere (Bajaj et al., 2010; Engler et al., O 2008) were not performed here due to the naturally high y g differences of contraction rates between cardiomyocyte o subtypes (Joyner et al., 2007; Watanabe et al., 1983). ol Bi The only morphological difference that occurred only on the very softest substrates was a more roundish shape of myocytes with possibly slightly less aligned myofibrils. This phenotype is mostlikelycauseddirectlybylowcellforcegenerationonthese substrates. Bischofs and co-workers developed a theoretical model that describes the relation between cell shape and cell force distribution based on outer adhesion sites and connecting inwardcurvedactinarcs(Bischofsetal.,2009).Softersubstrates resultinsmallerforcesandtherefore,accordingtothismodel,in less curvature. Consequently, reduction of outer adhesion sites withconnectingactinbundlesasgivenforourcardiomyocyteson very soft substrates (Fig. 4B) should result in rounded cell shapes. The slightly reduced order of myofibrils on very soft substrates fits very well with current knowledge. In independent papers Englerandco-workersand Bajajandco-workersshowed similar results for cardiomyocytes grown on very soft polyacrylamide gels (Bajaj et al., 2010; Engler et al., 2008). Fig.8. Constantsarcomericcontractilestrainuponsubstratestiffening. ContractionofindividualmyofibrilsofGFP-a-actinintransfectedcellswas analyzedovertimealongalinespanningtheirfulllength(A,left)andplotted astime–spaceplot(A,right;blackarrowindicatesdirectionoftime).Scalebar: 10mm.Grayvaluesalongthemyofibrilatrelaxed(t1)andcontractedstate(t2) showdisplacementstowardsthecenter(blackarrowheadinB,C)ofthe myofibrilandenablecalculationofdisplacementsofz-bandsalongthe myofibril(C).Notethealmostlineardependenceofdisplacementonpositionin theinnerhalfofthemyofibril.Fittedlinescharacterizingthelinear displacementsectionwereaveragedover2–4independentmyofibrilspercell. Theresultingslopes(D)describetheaveragemyofibrilcontractilityofeach cellon15kPasubstrates(n57),30kPa(n513)and90kPa(n58).Mean contractilemyofibrilstrainaveragedoverallcellsperelasticityisgivenas straightlineandseparatelyin(D),bottomright,withincludeds.d. Elasticity-driven muscle forces 358 Although generated force values increase drastically with The model above includes cell adhesion all along the substrate stiffening, not only myofibril structure, but also their contractile apparatus, which is well supported by our traction contractility,seemtobebasicallyunaffectedbyelasticitywithin force experiments.Here,costameric andfocaladhesionsequally the natural range of 15 to 90 kPa. Our quantitative analyses of participate in force transmission, especially on stiff substrates. myofibrilcontraction(cf.Fig. 8)revealedrelativelystablestrain Only the very central sarcomere units show a clear contractility. valuesofabout4%forcentrallylocalizedsarcomeres.Apossible Unfortunately, finding a clear correlation between contractile tendency of reduced contractions on stiffer substrates was not length of myofibrils and substrate stiffness over all substrate significant and even if it were, this behavior would indicate that elasticitiesanalyzedherewasnotpossibleduetolimitednumbers cardiomyocytes generate elevating force values on stiffer of analyzable myofibrils of sufficient length. Such experiments substrates with even less contraction. were out of the scope of this work due to the very low Measurements of sarcomere lengths during single myofibril transfection rates achieved for cardiomyocytes but could finally contractions (e.g. Fig. 8) very clearly showed different behavior prove or disprove our model in the future. in the middle and at the ends of myofibrils. While centrally Most importantly our data argue for a force generating located sarcomeres contracted, the outer ones were moved at apparatus in cardiomyocytes achieving stable contraction constant length. From basic mechanics this is a very surprising amplitudes on varying substrate elasticities. Uncoupling findingbecausealongthemyofibrilforcesmustaddupresulting cardiomyocytes from surrounding tissue tension leads to stable inthehighesttensilestressatthecenterofthefibril.Atthesame sarcomerelengthsonphysiologicalsubstrateelasticitiesbetween time traction forces transmitted by costameres from the elastic 15 and 90 kPa. Substrate stiffening sarcomeres have to generate substrate to the myofibril were indeed low at the center and higher forces in order to keep their contraction value, which increased towards fiber ends (e.g. Fig. 7). Exactly the same again argues for a force generating machinery never working at behavior is found in technical mechanics for fibers of finite full load on softsubstrates.Such a regulatorymechanism would length embedded in a strained elastic matrix (Hsueh, 1989). As therefore be an ideal system to respond instantaneously to sarcomeric lengths were in the range of 1.9 mm where in tissue environmental changes or just stress factors causing temporal sarcomere length raises with applied stress a passive material tissue stiffening connected with the necessity of enhanced force should display contraction mostly at the ends and less in the values. Sarcomere length regulation as described by the Frank– n middle.Takentogetherbasicmechanicsandourobservationscan Starlingmechanisminstiffermyocard,willcertainlyadditionally e only be reconciled by assuming that in our experiments elevateforcesgeneratedbymyocytesbutisnotthecrucialfactor p O individual sarcomeres can sustain the same force at a whole for the force adaptation observed here. y range of lengths. In turn, such a behavior would also imply that g sarcomeresatstablelengthareabletogeneratearangeofforces o ol as we could clearly identify in our experiments (Fig. 8). Materials and Methods Bi At first such a material behavior appears highly implausible. Isolationofratmyocytesandmyofibroblasts Cardiomyocytesandcardiacfibroblastsweresimultaneouslyisolatedfrom19-day- However, it should be seen in the context of the well-studied oldWistarratembryos.PregnantratswereCO anesthetizedanddecapitated.The 2 spontaneous oscillations of isolated myofibrils (Fabiato and pupswereremovedandalsodecapitatedundersterileconditions.Theheartofeach Fabiato, 1978; Linke et al., 1993). In this system spontaneous embryo was quickly isolated, washed in Hank’s balanced salt solution (HBSS, Sigma, St. Louis, USA), cut into small pieces and digested in 8ml of a 0.5% oscillations of sarcomeric lengths and even traveling density Trypsin/0.2%EDTAsolutioninHBSStodisintegratethetissue.Thesupernatant waves occur at intermediate levels of excitation, which lead to was discarded and resulting cell aggregates were incubated with 100ml DNase the same conclusion as presented above (Anazawa et al., 1992). solution(10,000U/ml,Sigma,St.Louis,USA)for3minutesandthenfilledwith As we observed spontaneous contractions in isolated cells 8ml of TE solution for further 5minutes. While cell aggregates precipitated withoutcentrifugation,disintegratedsinglecellsremainedinthesupernatantand excitation of the contractile system by cellular calcium were transferred into TE-blocking solution (F10 Ham’s medium supplemented oscillations was also merely at an intermediate level. with33%fetalbovineserum).ThedisintegrationstepandDNasetreatmentwas In the abovementioned experiments on isolated myofibrils all repeatedonetotwomoretimesandsupernatantswerecombined.Cellswerethen collected by centrifugation at 200g, maintained at 37˚C and 5% CO in a sarcomeres within the fibril experience the same force and 2 humidifiedincubatorandculturedinF10Ham’smediumsupplementedwith10% identical conditions. This is contrary to our case where traction fetal bovine serum, a 1:100 dilution of an antibiotic solution (10,000 units forces transferred from the substrate to the fibril grow towards penicillinand10mg/mlstreptomycinin0.9%NaCl,Sigma)anda1:200dilution the ends while tension in the fibril is maximal at center. This of solution containing insulin (1mg/ml), transferrin (0.55mg/ml), and sodium selenite(0.5mg/ml)inEBSS(Sigma,St.Louis,USA).Myocyteswereseededon could explain why we observed neither travelling waves nor fibronectin (2.5mg/cm2) (BD Biosciences, San Jose, CA) coated cross-linked oscillations. In addition, since tensile stress within the myofibril siliconerubbersubstratesofdiverseandcontrolledelasticities. and traction forces transmitted by costameres counteract, Forsomeexperimentscardiacfibroblastswereculturedforadditional3dayson standardpolystyrenecellculturedishesinsupplementedF10Ham’smediumto relatively small alterations in composition and structure might inducetheirdifferentiationtomyofibroblasts.Differentiatedcellsweretrypsinized be sufficient to modify the contractile behavior of myofibrils of andusedforcontrolexperimentsasdescribedbelow. isolated cardiomyocytes. Evidence for such small differences between centrally and distally located costameres was found in Celltransfection vivo (Bennett et al., 2006; Bennett, 2012) as well as in vitro Cardiomyocytes and myofibroblasts were transfected using Nucleofector technology (Lonza–Amaxa Biosystems, Cologne, Germany). 26106 cells were (Goncharova et al., 1992; Sanger et al., 2005; Simpson et al., resuspendedin100mlNucleofectorsolution.2mgpurifiedplasmid-DNA(GFP-a- 1993). Taken together, suboptimal levels of excitation and actininformyocytesandGFP-VASPformyofibroblasts)wasaddedtothesample gradual stress transfer from the substrate, perhaps together with andcellsweretransfectedaccordingtothemanufacturer’sinstructions(program slightvariationsofmyofibrilcompositiontowardstheirends,are G09).Transfectionratesrangedbetween0.1and1%forcardiomyocytesandwere ,5% for myofibroblasts. After transfection, resuspended cells were seeded on sufficient to reconcile our observations with present knowledge fibronectincoatedcross-linkedsiliconerubbersubstrates atacelldensityof2– on myofibril contractility. 46104perPetridish(ø3.5cm). Elasticity-driven muscle forces 359 Fabricationofsiliconerubbersubstrates segmentation routine was developed. It started with bandpass filtering of the For cell force analysis of cardiomyocytes, cells were seeded on cross-linked imagesbysubtractingtwobinomiallyfilteredimages(969binomialfilteredminus elastomeric silicone rubber substrates (Sylgard 184, Dow Corning GmbH). 57657binomialfiltered).Pixelsizewas0.2mm.Subsequently,foreachpixelof Substrate preparation and characterization of elastomer material properties thispreprocessedimagelocalaverageandlocalstandarddeviationwerecalculated (Young’s modulusand Poisson’s ratio)were performed as described previously withina91691environment.Thesevalueswereusedtoreplaceeachpixelvalue (Cesaetal.,2007).Inbrief,variousratios(seebelow)ofbase(vinylterminated bythez-score,i.e.thelocalaveragewassubtractedandtheresultnormalizedby polydimethylsiloxane) and cross-linker (methylhydrosiloxane–dimethylsiloxane the local standard deviation. All connected regions exceeding a threshold size copolymer)wereprepared.Theprepolymermixturesweredegassedandonepart (usually10pixels)andathresholdz-score(usually0.7)wereclassifiedasclusters. of the mixture was mixed with red fluorescent beads (0.2mm diameter, In some micrographs highly elongated and obviously artificial clusters were FluoSpheres Crimson carboxylate-modified beads, Invitrogen) and subsequently identifiedbythisroutine.Theseartifactswereremovedbythefollowingcriterion. appliedonapreviouslysilanized(1H,1H9,2H,2H9-perfluorooctyl-trichlorosilane, Foreachclustertheellipsewithequalcentralsecondmomentswascalculated.The Sigma)micro-structuredsilicondioxidemold.Themicrostructurewascomposed distributionofthesemi-majoraxesoftheseellipseswastransformedintothez- of asquare lattice consistingof500nm highmicrodots withan edgelength of score(definitionseeabove).Allareasexceedinganinteractivelychosenthreshold 2.5mm and a lattice constant of 3.5mm. The layer thickness ofthe fluorescent value(typically15)wereconsideredartifactsandremoved.Fromtheremaining bead/prepolymermixturewasreducedtolessthan500nmbywipingthesurface datawecalculatedaverageclustersize,thenumberofclusterswithinagivencell with lint-free tissue. The remaining non-cross-linked prepolymer lacking andthefractionofthecellareacoveredbyclusters.Tothisendcellareaswere fluorescent beads was added on top of the thin fluorescent bead/prepolymer interactivelymarkedbypolygonsasabove. layer.Glassslides(80mmthick,MenzelGmbH,Germany)servedasspacersand wereplacedonthesiliconemixturebesidethemicrostructure.Subsequently,they Cellforceretrieval werecoveredwithanadditional80mmcoverslipontoptoobtainaconstantlayer Traction forces applied by the cell to the elastic substrate caused substrate thickness.Cross-linkingwasperformedat60˚Covernight.Elastomericsubstrates deformations that were visualized by tracking fluorescent marker beads in the weresubsequentlygluedtothebottomof3.5cmPetridishestocoverpredrilled uppermost layer of the substrate (see above). From the displacements of these 1.5cmholes.Aftercuring,thesesiliconerubbersdisplayedaPoisson’sratioof0.5 particlescellforceswerecalculated.Thesecalculationsarebasedontheverywell andaYoung’smodulusof500kPa(20:1ratio;basetocross-linkerw/w),150kPa understoodbehavioroflayeredelasticmaterialsinresponsetoexternalforces(cf. (30:1), 90kPa (35:1), 30kPa (45:1), 15kPa (50:1) and ,1kPa (70:1). The Merkeletal.,2007;andreferencestherein).Thespecificsiliconeelastomersystem elasticityfortheverysoft,1kPasubstratescouldnotbeaccuratelyanalyzeddue usedherehasbeenshowntofulfilltheprerequisitesoftheabovementionedtheory, totheirsoftness andstickiness. Therefore,theelasticity ofthosesubstrates was i.e.theyarehomogeneousandisotropicmediaexhibitinglinearelasticityuptovery estimatedbyextrapolatingexistingcalibrations. largeextensions(Balabanetal.,2001;Cesaetal.,2007;GutierrezandGroisman, 2011). However, for rapidly contracting cells like cardiomyocytes viscoelastic Microscopy material behavior is a concern, especially for the softest rubbers. We therefore Livecellanalyseswereperformedat37˚Cand5%CO (cellincubatorXL2,Zeiss, measured the force relaxation after a rapid indentation step (0.2second) with a 2 Germany)usinganinvertedmicroscope(Axiovert200,Zeiss,Germany),equipped cylindricalpunch.Theratioofforceafterthedurationofanaveragecontraction withaCCDcamera(Orcablack-and-white, Hamamatsu,Japan)utilizing a636 (,0.3second)andatverylongtimeswas2.1fortheverysoftestrubber(,1kPa) n e Antiflex EC Plan-Neofluar oil immersion objective (PH3, NA51.25, Zeiss, and1.3for15kParubber.Asstifferelastomersexhibitfasterforcerelaxationand p Germany).ImagesweretakenusingtheimagingsoftwareOpenBox(version1.77, theoverallaccuracyofthetechniqueisabout30%,weconcludedthatviscoelastic O Informationssysteme Schilling, Germany). Excitation of GFP-a-actinin and of materialbehaviorplaysasignificantroleforthesoftestelastomeronly. fluorescentbeadswasdonewithamercuryarclamp(HBO100,Osram,Germany). Moreover,forisolatedcellsofthesizeofmyocytesalayerof80mmthickness y g FluorescentimagesofGFP-a-actininwereobtainedusinga505–530nmbandpass behavesalmostexactlylikeanelastichalfspace(Merkeletal.,2007).Thisgreatly o filter,forthefluorescentbeadsa590nmlong-passfilterwasused. simplifies the underlying mathematics and enabled us to use the analytical ol Immunofluorescentlylabeledcellswereanalyzedwithaninverseconfocallaser expressions derived first by Boussinesq (Boussinesq, 1885). These expressions Bi scanningmicroscope(LSM710,Zeiss,Germany)usinga406Plan-Neofluaroil correctlydescribethedisplacementscausedbyapointforceactingonthesurface immersionobjective(Ph3,NA1.3,Zeiss,Germany)anda636Plan-Apochromat of the elastomer. Our algorithms (see below) assume a test force distribution, oil immersion objective (Ph3, NA 1.4, Zeiss, Germany). Confocal micrographs calculate the cumulative displacement caused by all these forces and iterally weretakenusinganargonionlaser(488nm)witha490–530nmbandpassfilter improve the force distribution until it reproduces the measured displacements. andahelium–neonlaser(543nm)witha550–812nmbandpassfilter. Unfortunately, in traction force microscopy this conceptually straightforward approachinvolvessomelessthanintuitivemethodsofnumericalmathematics.For detailsseee.g.Merkelandco-workersandSchwarzandco-workers(Merkeletal., Immunofluorescencetechniques 2007;Schwarzetal.,2002).Ourimplementationworksasdescribedbelow. Myocytes (36104 cells per substrate) were seeded on silicone rubber substrates BeaddisplacementswerefirstdeterminedasdescribedbyMerkelandco-workers with different stiffness ranging from ,1 to 500kPa and stained after 3 days. (Merkeletal.,2007).Inshort,thefirstimageofthesequencewasusedasreference Myocyteswerefixedusing3.7%formaldehyde(Merck,Germany)incytoskeleton image.Asmallregionaroundarandomlychosenmicrobeadwasusedastemplate. buffer(CB:150mMNaCl,5mMMgCl2,5mMEGTA,5mMglucose,1mg/ml Allotherbeadswerelocalizedbycross-correlatingthistemplatewiththeimageand streptomycin,10mMMES,pH6.1)for20minutesat37˚C.Subsequently,cells finding local maxima of the normalized correlation coefficient with sub-pixel werewashedoncewith30mMglycineinCBfor10minutes,permeabilizedwith accuracy.Onlyplaceswherethenormalizedcross-correlationcoefficientexceededa 5% Triton X-100 (Sigma, St. Louis, USA) in CB for 3minutes at room thresholdvalue(typically0.8)wereacceptedasbeadpositions.Insubsequentframes temperature, rinsedtwice withCB and then treated with blocking solution (5% microbeadswerelocalizedusingbead-specifictemplatesandtheirdisplacements skim milk powder (Sigma) in CB) for 45minutes at 37˚C. Primary antibodies withrespecttothereferenceimagewerecalculatedwithanaccuracyof25nm. (anti-vinculin,clonehVin-1(Sigma,St.Louis,USA);anti-a-actinin,cloneEA53 Forceretrievalfromthesedisplacementsismorecomplicatedasmathematically (Sigma, St. Louis, USA); anti-plectin, clone 10F6 (Santa Cruz Biotechnology, thedisplacementvectorfieldoriginatesfromtheforcefieldbyaconvolutionwith SantaCruz,CA)andsecondaryantibodiesconjugatedtoCy3(Dianova,Hamburg, astiffnessdependentGreens’tensorthatexhibitsasingularityattheoriginand Germany)werebothdiluted1:100in1%skimmilkpowderinCBandapplied decaysasoneoverdistance.Invertingthisconvolutionisanumericallyill-posed eachfor45minutesat37˚CwiththreeadditionalwashingstepswithCBbetween problemrequiringregularizedleastsquaresfitorrelatedalgorithms(Demboand theprimaryandsecondaryantibody.ForactinstainingAlexaFluor488phalloidin Wang,1999;Merkeletal.,2007;Schwarzetal.,2002). (Invitrogen,Carlsbad,USA)wasusedina1:200dilutioninCBtogetherwiththe Inthisworkweusedtwodifferentalgorithms.Oneisbasedonasuperposition secondaryantibodyfor45minutesat37˚C. ofoptimallychosenpointforceslocalizedatmanuallypredefinedlocations(cell adhesionsites),theothercalculatesentireforcefields.Thepointforcealgorithm Cellshapedetermination wasdevelopedbySchwarzandco-workersandusedasdescribedbyCesaandco- Forquantifyingcellroundnessweusedadimensionlessshapeindexsidefinedas workers(Schwarzetal.,2002;Cesaetal.,2007).Theforcefieldalgorithmusedis si54pAP22whereAdenotestheareaofthecellandPitsperimeter.Foranyshape described in detail by Houben and co-workers (Houben et al., 2010). Both siisinbetween1(circle)and0(line).Forcalculatingsitheshapeofacellwas algorithms yield force distributions. To assess the quality of these solutionswe approximatedbyapolygonwithinteractivelychosencornerpoints(dependingon calculated the displacements resulting from the retrieved force distribution and cell shape typically between 20 and 40). Cells exhibiting si above 0.8 were compareditwiththemeasuredbeaddisplacements.Besidesavisualcomparison we calculated the normalized squared deviation, x2, between fit result and classifiedasround,allothersasangular. measured data according to x2~½2ðNB{NFÞs2(cid:2){1PNB ð~uuBi{~uuFiÞ2 where NB DetectionofGFP-a-actininclustersinmicrographs signifies the total number of tracked beads with dispi~la1cements~uuBi (measured) FluorescencemicrographsofcellstransfectedwithGFP-a-actininexhibitedbright and~uuFi (calculated).NFisthetotalnumberofforceinductionpointsandsthe clusters indicating FAs and z-bands. For the localization of these clusters a standarddeviationofdisplacementmeasurement(here25nm). Elasticity-driven muscle forces 360 For the characterization of cell contractions we calculated the sum of all Bajaj, P., Tang, X., Saif, T. A. and Bashir, R. (2010). Stiffness of the substrate contractile forces. For the point force algorithm we proceeded as described by influencesthephenotypeofembryonicchickencardiacmyocytes.J.Biomed.Mater. Merkelandco-workers(Merkeletal.,2007).Here,forceapplicationpointswere Res.A95A,1261-1269. chosenonaregularhexagonalgrid(latticeconstant5mm)coveringthewholecell Balaban,N.Q.,Schwarz,U.S.,Riveline,D.,Goichberg,P.,Tzur,G.,Sabanay,I., area.Subsequently,thecellcenterwascalculatedasthecenterofgravityofthese Mahalu,D.,Safran,S.,Bershadsky,A.,Addadi,L.etal.(2001).Forceandfocal points. The contractile parts of all forces, i.e. the force components pointing adhesion assembly: a close relationship studied using elastic micropatterned substrates.Nat.CellBiol.3,466-472. towardsthecellcenter,weresummeduptoyieldthetotalcontractileforce.This Bennett,P.M.(2012).Frommyofibriltomembrane;thetransitionaljunctionatthe procedureenablesthecalculationofoverallcellcontractionforcevaluesalthough intercalateddisc.Front.Biosci.17,1035-1050. accordingtoNewton’sprinciplesallforcevectorshereandforanyotherstably Bennett,P.M.,Maggs,A.M.,Baines,A.J.andPinder,J.C.(2006).Thetransitional adheredcellsumuptozero. junction:anewfunctionalsubcellulardomainattheintercalateddisc.Mol.Biol.Cell Intheothercase,theforcefieldalgorithm,thewholemicrographiscoveredbya 17,2091-2100. mesh comprising almost the same number of vertices as beads were tracked Berry,M.F.,Engler,A.J.,Woo,Y.J.,Pirolli,T.J.,Bish,L.T.,Jayasankar,V., (typically2,000–4,000).Forcesarecalculatedateachvertex.Thereforewecould Morine, K. J., Gardner, T. J., Discher, D. E. and Sweeney, H. L. (2006). calculate the contractile force in almost the same way as before. In effect the Mesenchymal stem cell injection after myocardial infarction improves myocardial geometrical center of the cell was determined and within the cell area the compliance.Am.J.Physiol.HeartCirc.Physiol.290,H2196-H2203. contractilepartsofallforceswereadded. Bhana,B.,Iyer,R.K.,Chen,W.L.K.,Zhao,R.,Sider,K.L.,Likhitpanichkul,M., For some control analyseswe performed traction force experiments on GFP- Simmons, C. A. and Radisic, M. (2010). Influence of substrate stiffness on the VASP transfected myofibroblasts. Bead microstructured elastomer substrate phenotypeofheartcells.Biotechnol.Bioeng.105,1148-1160. preparation, bead displacement determination and traction force analyses (point Bischofs,I.B.,Schmidt,S.S.andSchwarz,U.S.(2009).Effectofadhesiongeometry forcealgorithm)wereperformedasdescribedaboveforcardiomyocytes.Inorder andrigidityoncellularforcedistributions.Phys.Rev.Lett.103,048101. toobtainareferenceimageofthenon-deformedsubstratebeneaththeanalyzed Boussinesq, J. (1885). Application Des Potentiels A` L’e´tude De L’e´quilibre Et Du cells, myofibroblasts were removed manually from the substrate using a glass MouvementDesSolideE´lastiques.Paris:Gauthier-Villars. micropipettemountedonamicroinjector(InjectManNI2,Eppendorf,Germany). Butler,J.P.,Tolic´-Nørrelykke,I.M.,Fabry,B.andFredberg,J.J.(2002).Traction fields, moments, and strain energy that cells exert on their surroundings. Am. J. As force induction points we either used FAs labeled by GFP-VASP or an Physiol.CellPhysiol.282,C595-C605. interactively chosen lattice of ,4mm spacing beneath the central cell body to Cesa, C. M., Kirchgessner, N., Mayer, D., Schwarz, U. S., Hoffmann, B. and mimicadhesionsitesalongmyofibrilsofcardiomyocytes. Merkel,R.(2007).Micropatternedsiliconeelastomersubstratesforhighresolution analysisofcellularforcepatterns.Rev.Sci.Instrum.78,034301. High-resolutionmyofibrilcontractionanalysis Damon,B.J.,Re´mond,M.C.,Bigelow,M.R.,Trusk,T.C.,Xie,W.,Perucchio,R., FluorescentmicrographsofcellstransfectedwithGFP-a-actininclearlyshoweda Sedmera,D.,Denslow,S.andThompson,R.P.(2009).Patternsofmuscularstrain sarcomerestructure.Duetothecorrespondingbrightnessvariationalongthefibril intheembryonicheartwall.Dev.Dyn.238,1535-1546. contractions could be accurately quantified by image correlation. In detail we Danowski,B.A.,Imanaka-Yoshida,K.,Sanger,J.M.andSanger,J.W.(1992). Costameres are sites of force transmission to the substratum in adult rat proceededasfollows.Theimagewasdenoisedwitha363binomialfilter(pixel cardiomyocytes.J.CellBiol.118,1411-1420. n size0.2mm).Subsequently,amyofibrilwastracedwithconnectingstraightlines. deTombe,P.P.,Mateja,R.D.,Tachampa,K.,AitMou,Y.,Farman,G.P.and e At distances closely corresponding to the original pixel size gray values were p determinedbyinterpolationfromtheneighboringpixels.Inthefollowingwewill Irving,T.C.(2010).Myofilamentlengthdependentactivation.J.Mol.Cell.Cardiol. O 48,851-858. refertotheresultingvectoras‘‘profile’’.Grayvalueextractionwasrepeatedfor Decker,M.L.,Simpson,D.G.,Behnke,M.,Cook,M.G.andDecker,R.S.(1990). y eachimageofthesequencealongthesameline.Intheresultssectiontheseprofiles Morphologicalanalysisofcontractingandquiescentadultrabbitcardiacmyocytesin og are shown in the space–time plane. Such plots are sometimes also called long-termculture.Anat.Rec.227,285-299. ol kymographs. Dembo,M.andWang,Y.L.(1999).Stressesatthecell-to-substrateinterfaceduring Bi Beforetheactualcalculationofsarcomeredisplacementseachprofilewasagain locomotionoffibroblasts.Biophys.J.76,2307-2316. smoothedbyabinomialfilterofwidth3.Thefirstprofileofasequencewasused Diffee, G. M. and Chung, E. (2003). Altered single cell force-velocity and power as reference. It was subdivided into overlapping templates. Spacing between propertiesinexercise-trainedratmyocardium.J.Appl.Physiol.94,1941-1948. template centers was 5pixels, template width was chosentypically between 15 Discher,D.E.,Mooney,D.J.andZandstra,P.W.(2009).Growthfactors,matrices, and 25pixels. At each later time point displacements along the myofibril were andforcescombineandcontrolstemcells.Science324,1673-1677. found by cross-correlating each template with the respective profile. Sub-pixel Du,A.,Sanger,J.M.,Linask,K.K.andSanger,J.W.(2003).Myofibrillogenesisin resolution was achieved by fitting a parabola to the maxima of the cross- thefirstcardiomyocytesformedfromisolatedquailprecardiacmesoderm.Dev.Biol. correlationcoefficientanddeterminingitssymmetrypoint.Thisprocedureyielded 257,382-394. displacement profiles with one displacement value every five pixels. For each Engler, A. J., Griffin, M. A., Sen, S., Bo¨nnemann, C. G., Sweeney, H. L. and contraction of the myofibril the time point of the maximum displacement was Discher,D.E.(2004).Myotubesdifferentiateoptimallyonsubstrateswithtissue-like stiffness:pathologicalimplicationsforsoftorstiffmicroenvironments.J.CellBiol. found. Displacement profiles at maximum contractions (typically five) were 166,877-887. averagedandfurtheranalyzed. Engler,A.J.,Sen,S.,Sweeney,H.L.andDischer,D.E.(2006).Matrixelasticity Todeterminethecentralslopeoftheseaverageddisplacementprofileswefirst directsstemcelllineagespecification.Cell126,677-689. determined the useful fit range. This was taken to be the range between the Engler,A.J.,Carag-Krieger,C.,Johnson,C.P.,Raab,M.,Tang,H.Y.,Speicher, numericallydeterminedinflectionpointsofthedisplacementprofile.Intheactual D. W., Sanger, J. W., Sanger, J. M. and Discher, D. E. (2008). Embryonic experimentsthisregionwas5to10mmwide.Withinthisrangeastraightlinewas cardiomyocytes beat best on a matrix with heart-like elasticity: scar-like rigidity fittedtothedata.Itsslopecorrespondstothecontractilestrain(relativeshortening) inhibitsbeating.J.CellSci.121,3794-3802. ofthemyofibrilinitscenterregion.Subsequently,thecontractilestrainvaluesof Fabiato,A.andFabiato,F.(1978).Myofilament-generatedtensionoscillationsduring allevaluatedmyofibrilswithinacell(mostlytwotofour)wereaveraged. partialcalciumactivationandactivationdependenceofthesarcomerelength-tension relationofskinnedcardiaccells.J.Gen.Physiol.72,667-699. Fukuda, N., O-Uchi, J., Sasaki, D., Kajiwara, H., Ishiwata, S. and Kurihara, S. Acknowledgements (2001).Acidosisorinorganicphosphateenhancesthelengthdependenceoftensionin We thank W. Rubner for helpful discussions and expert technical ratskinnedcardiacmuscle.J.Physiol.536,153-160. support. We also thank E. Noetzel for critical reading of the Giannone,G.,Jiang,G.,Sutton,D.H.,Critchley,D.R.andSheetz,M.P.(2003). manuscript and T. Jonas for cell isolation and helpful discussions. Talin1iscriticalforforce-dependentreinforcementofinitialintegrin–cytoskeleton bondsbutnottyrosinekinaseactivation.J.CellBiol.163,409-419. This work was supported by the German Bundesministerium fu¨r Gilbert,P.M.,Havenstrite,K.L.,Magnusson,K.E.,Sacco,A.,Leonardi,N.A., BildungundForschung(BMBF,projecttitle:MechanoSys,BMBF- Kraft, P., Nguyen, N. K., Thrun, S., Lutolf, M. P. and Blau, H. M. (2010). Nr.0315501A). Substrateelasticityregulatesskeletalmusclestemcellself-renewalinculture.Science 329,1078-1081. Goffin,J.M.,Pittet,P.,Csucs,G.,Lussi,J.W.,Meister,J.J.andHinz,B.(2006). Competing Interests Focaladhesionsizecontrolstension-dependentrecruitmentofa-smoothmuscleactin Theauthors have nocompeting intereststodeclare. tostressfibers.J.CellBiol.172,259-268. Goncharova, E. J., Kam, Z. and Geiger, B. (1992). The involvement of adherens junction components in myofibrillogenesis in cultured cardiac myocytes. References Development114,173-183. Anazawa,T.,Yasuda,K.andIshiwata,S.(1992).Spontaneousoscillationoftension Gupta,K.B.,Ratcliffe,M.B.,Fallert,M.A.,Edmunds,L.H.andJrandBogen, andsarcomerelengthinskeletalmyofibrils.Microscopicmeasurementandanalysis. D. K. (1994). Changes in passive mechanical stiffness of myocardial tissue with Biophys.J.61,1099-1108. aneurysmformation.Circulation89,2315-2326.