Functional Characterization of Cardiac Actin Mutants Causing Hypertrophic (p.A295S) and Dilative Cardiomyopathy (p.R312H and p.E361G)

The human mutant cardiac α-actins p.A295S or p.R312H (plus p.R312K) and p.E361G correlated with hypertrophic or dilative cardiomyopathy, respectively, were expressed by using the baculovirus/Sf21 insect cell system. After purication their biochemical and cell biological properties were analysed and compared to wild type (wt) cardiac actin identically obtained or conventionally isolated from bovine hearts. DNase I inhibition and their polymerization behaviour indicated that all c-α-actins had maintained their native state. Cardiomyopathy type specic differences were observed except for the p.R312K mutant, which behaved like wt c-α-actin. The extent of myosin-S1 ATPase stimulation by the c-actin variants and its Ca 2+ -sensitivity after decoration with tropomyosin (cTm) and troponin complex (cTn) varied being highest for the HCM p.A295S and lower for both DCM mutants. Similar Ca 2+ -sensitivity differences were observed by recording the uorescence increase of pyrene-cTm in the absence or presence of myosin-S1 and/or the actin-binding N-terminal fragment of cardiac myosin binding protein C (N-cMyBP-C). Transfection experiments showed the incorporation of the c-actin variants into existing cytoskeletal elements of non-muscle cells. Wt and p.A295S c-α-actin preferably incorporated into the microlament system and p.R312H and p.E361G into the submembranous actin network of MDCK cells. Transduction of neonatal rat cardiomyocytes with adenoviral constructs coding for HA-tagged c-α-actins showed their incorporation into thin laments of nascent sarcomeric structures at their plus ends (Z-lines) except the p.E361G mutant, which preferably incorporated at the minus ends. Our data indicate functional differences of the c-α-actins that may be causative for the different cardiomyopathy phenotypes.


Introduction
Cardiomyopathies (CM) are generally characterized by electric dysfunction, hypertrophy or dilation.
Hypertrophic cardiomyopathy (HCM) is characterized by an eccentric left ventricular wall thickness accompanied by an increased stiffness of the cardiac muscle resulting in reduced ventricular lling during diastole.Dilated cardiomyopathy (DCM) is characterized by brosis and enlarged left and/or right ventricle with a reduced ejection volume during systole.Whereas HCM has a genetic background in about 90% of all cases with a prevalence of 1:500 in the general population, DCM is in only 20% genetically based with a prevalence of 1: 2,500.In the remaining 80% of all cases DCM develops as a complication of bacterial or viral infections, adverse life style, alcoholism, drug abuse or chemotherapy [53,54,66].
The genetically caused (familial) cardiomyopathies (CM) are caused in the vast majority by mutations in genes encoding sarcomeric [3,53] or cytoskeletal proteins [15,36], i.e. actin binding proteins.About 1,400 different mutations causative for HCM or DCM have so far been described.In HCM the most frequently affected sarcomeric proteins are the ß-myosin heavy chain and the myosin binding protein C (MyBP-C) [9,53,54].It has been observed that mutations in genes of cytoskeletal proteins occur more frequently in familial DCM, in particular of components of the intercalated disc [15].In about 20% of the acquired DCM cases, titin gene truncations are observed as de novo mutations [23].But mutations leading to HCM or DCM have been identi ed also within the same gene like for instance in the ACTC gene coding for cardiac α-actin (c-α-actin).Typically, the CM mutations act in a dominant negative manner and are therefore in most cases single-allelic.Consequently, it is supposed that the mutated protein overrides in a poisonous manner the counterpart of the healthy allele [3,39].The expression dosage of the mutant and healthy allele are usually equal, but deviations will contribute to the resulting phenotype [36,56].
Mutations of the c-α-actin gene are with an incidence of 4-6% an infrequent nding in patients with familial HCM, the incidence may be even lower in patients with familial DCM [11,17,18,39,44].This low incidence is in contrast to myopathies of skeletal muscle, which are caused to about 30% by mutations of the skeletal muscle actin gene (ACTA1) [36,46].
The rst mutations of cardiac α-actin, which were found in DCM patients, were the p.R312H and p.E361G mutations located in subdomain 4 and 1 of actin, respectively [44].The rst actin mutants causing HCM were p.H90Y and p.R97C both located in subdomain 1 [45].Subsequently, a number of further mutations in c-α-actin were described that lead to HCM [46].Presently 40 mutations have been identi ed [36], of which 14 different HCM and two DCM causing mutations are known [16], most of which are located within subdomains 1 and 4. Figure 1 shows the 3D structural model of G-actin [27] indicating the positions of the c-α-actin point mutations correlated to HCM and DCM.
Dysfunctions connected to the different mutations largely depend on the type of amino acid exchange, their localisation within the actin molecule, and the effect on the interaction with speci c binding partners [45,46,63].Thus for example, mutations in subdomain 4 affect the stability of the actin lament, though subdomain localization appears not decisive for HCM or DCM development [6,9,11,18,39].The development of a certain cardiomyopathy phenotype does not follow a single pathway.It is complicated by the fact that missense mutations in many different sarcomeric proteins can lead to the same disease phenotype and conversely point mutations at different locations in the same protein can induce diverse outcomes [66,56].Therefore for the development of a speci c disease phenotype it may be decisive how a mutation affects the overall architecture and/or functionality of the sarcomere or the whole cardiomyocyte by for instance its modi ed interactions with different binding partners.On the way to such an understanding it is however necessary to analyse the altered properties of every mutation in a single protein.
In a previous communication we described the expression of two mutants of c-actin causing HCM (p.Y166C and p.M305L) by the Sf21/baculovirus system and their puri cation in native state that subsequently allowed their biochemical and cell biological characterization [39].In the current study we investigated the functional properties of further mutants of cardiac α-actin puri ed after recombinant expression as native and tag-free proteins with the aim to analyse the mechanisms, by which these mutants might induce the development of the HCM or DCM phenotype.Detailed knowledge of these processes is still lacking.The mutant p.A295S located in subdomain 3 has been reported to cause HCM with high penetrance [17] and was shown to perturb the interaction of actin laments with tropomyosin [64].The c-α-actin mutations p.R312H located in subdomain 4 and p.E361G in subdomain 1 (Fig. 1) are reported to cause DCM [2,11,55] with a benign outcome [11].
Here we compared wild-type cardiac actin with the listed mutants in a number of different assays.In summary, we observed subtle differences in the biochemical and cell biological properties of the isolated c-α-actin mutants In comparison to wt c-α-actin.Our data indicate that after puri cation the c-α-actin variants were in native state and able to polymerize albeit with different kinetics and form normally appearing laments of varying lengths.Further detected altered properties concern the Ca 2+ -sensitivity after decoration with tropomyosin (Tm) and troponin (Tn) complex or their interactions with other isolated sarcomeric and cytoskeletal proteins.These alterations might however have a larger impact when interacting with other regulatory proteins during cardiomyocyte development or within the intact sarcomere.

Clones:
The pcDNA3.1/NT-GFP-TOPO®-WT-α-cardiacactin and the mutants p.A295S, p.R312K, and p.E361G were donated from Dr. Cora-Ann Schoenenberger (University Basel, Switzerland).The p.R312H mutant was generated by site-directed mutagenesis from the p.R312K variant.Since the biochemical properties of the p.R312K variant were found to be very similar to wt c-actin, the results of its analysis are included in the Supplementary Information.The c-α-actin containing plasmids served as templates for cloning the c-αactin variants into p3xHA-C1 plasmid.The p3xHA-C1 plasmid was a kind gift from Dr. T. Engel (Leibniz-Institut für Arteriosklerosis, Münster University, Germany), who deleted cDNA of EGFP from pEGFP-C1 plasmid (Clontech) and instead cloned into this plasmid the cDNA of a three times repeated hemagglutinin-tag (HA).The primers used for amplifying the actin cDNAs were as follows: 5'-GTTATGTGTGACGACGAGGAGACC-3' and 5'-ATTGCCCTTTTAGAAGCATTTGCG-3'.PCR inserts were cloned into p3xHA-C1 using XbaI and XhoI sites.
The deletion construct of human gelsolin G4-6 was kindly supplied by Dr. A.G. Weeds (MRC-LMB, Cambidge, UK) and subcloned from shuttle vector pKN172 into the cold-shock expression plasmid pCOLD II (Takara Bio Inc., Kusatsu, Japan) using the restriction sites for BamHI and HindIII enzymes obtained from Fermentas (Vilnius, Lithuania).The pCOLD II plasmid provides a His-Tag sequence for a nity chromatography, which was fused to the N-terminus of G4-6and subsequently used to a nitypurify the c-α-actins [39,42].Arp2/3 complex isolated from Acanthameba castellani was kindly supplied by Prof. M. Barber (San Francisco, California, USA) and mDia3-FH2 by Prof. Alfred Wittinghofer (MPI, Dortmund, Germany).

Reagents
Pyrene-maleinimide was obtained from Sigma Aldrich (Munich, Germany).All reagents were of analytical grade.

Protein expression and puri cation
Rabbit skeletal muscle and bovine c-α-actin were prepared from dried acetone powder and human cardiac muscle wt α-actin and its mutants were expressed in the baculovirus/Sf21-system [2,39] as detailed in the Supplementary Information.Preprations of myosin subfragment 1 (myosin-S1) from skeletal muscle and bovine cardiac muscle [58], of cardiac tropomyosin (cTm) [4,21] and troponin complex (cTn) [1,10,12], the N-terminal C0C2 fragment of human cardiac MyBP-C [51], and the gelsolin deletion mutant G4-6 [42] were performed with modi cations of published procedures as detailed in the Supplementary Information.

ATPase assay
Stimulation of the ATPase activity of ß-cardiac or skeletal muscle myosin-S1 activated by human wt cardiac α-actin or the cardiomyopathy inducing mutants was performed at 25 °C using a modi ed version of NADH-coupled assay according to [35] in a buffer containing 40 mM HEPES, pH 7.4, 25 mM KCl, 2 mM MgCl 2 , 0.5 mM DTT, 0.2 mM NADH and an ATP regeneration system consisting of 0.05 mg/ml pyruvate kinase, 0.5 mM PEP, and 0.02 mg/ml LDH.The reaction was started by addition of myosin-S1 to a nal concentration of 1 µM.NADH oxidation was followed by measuring the decrease in absorption at 340 nm (ε = 6220 M 1 cm − 1 ) [48,61] using a spectrophotometer (DU 800, Beckman Coulter, Krefeld, Germany).The ATPase rates were determined by linear curve tting and repeated at least three times for each condition with at least two different c-actin variant puri cations.

DNase I inhibition assay
The DNase I inhibition assay was performed as described [34].The DNase test solution contained 50 µg/ml salmon sperm DNA (Sigma-Aldrich D1626) in 10 mM Tris-HCl, pH 8.0, 1 mM MgCl 2 and 0.1 mM CaCl 2 .To determine the endonuclease activity of DNase I, a 10 µl pre-incubation reaction containing 3.2 µM DNase I from bovine pancreas (Sigma-Aldrich DN25) and zero to 6.4 µM of G-actin was prepared in G-buffer and incubated at room temperature for 20 min.Aliquots of the samples were added to 0.8 ml of 50 µg/ml DNA-solution and the absorbance was immediately monitored at 260 nm for 10 min using the Beckman DU 800 spectrophotometer.From the ratio of the initial linear rates of DNase I activity were determined and expressed as Kunitz units (KU/min; 1 KU = ΔOD 260 nm of 0.001).

Gel electrophoresis
Polyacrylamide gel electrophoresis in the presence of SDS (SDS-Page) was performed as given [30].
Native gel electrophoresis was performed on 10% polyacrylamide gels without SDS and run as described previously [47].

Immunoblotting:
Cells were lysed 10 mM Tris-HCl, pH 7.4, 100 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM NaF, 20 mM Na 4 P 2 O 7 , 2 mM Na 3 VO 4 , 1% Triton X-100, 10% glycerol, 0.1% SDS, 0.5% deoxycholate) and vortexed for 30 s and frozen until use.After thawing, the extracts were vortexed again and centrifuged at 20,817x g at 4 °C for 5 min.The protein concentration was estimated according to Bradford [7].30 µg of protein extracts were separated on 12.5 % SDS-Page gels.Proteins were transferred to nitrocellulose membrane using a wet blotter [24,59].Subsequently the membranes were blocked for 1 h in Tris-buffered saline with 1% Tween-20 (TBS-T) containing 5 % non-fat milk powder (blocking solution) and then incubated overnight at 4 °C with primary antibody diluted in blocking solution (goat anti-HA 1:500, mouse anti-αcardiac actin 1:200, mouse anti-α-actinin 1:2,000, mouse anti-GFP at 1:2,500 and rabbit anti-actin C11 at 1:1,000 dilution).After three washing steps with TBS-T for 15 min at room temperature, the nitrocellulose sheets were incubated with secondary antibodies linked to horse radish peroxidase (HRP) diluted in blocking solution (1:2,000) directed against either mouse or rabbit or goat for 1 h at RT.The nitrocellulose membranes were developed with the help of enhanced chemiluminescence (ECL) system (GE Amersham, Stepstone, UK).Occasionally membranes were subsequently stripped, re-blotted and immunostained for total actin.

Actin polymerization assays:
Actin polymerization rates were determined by the increase in uorescence caused by incorporation of pyrene-labelled actin into actin laments [28,39].Pyrene-labelled actin was pre-cleared by dialysis against G-buffer (10 mM Tris-HCl, pH 8.0, 0.2 mM CaCl 2 , 7 mM β-mercaptoethanol, 1 mM ATP) and centrifugation at 100,000xg for 30 min.In these tests we used pyrene-labelled skeletal muscle actin that was added to the c-actins at a ratio of 20:1 (0.25 to 5 c-α-actin).Since pyrene-labelled skeletal-actin on its own at 0.25 µM did not show signi cant polymerization, i.e. increase in uorescence.Therefore, we assume that the increase in uorescence observed after mixing it with globular c-α-actin in G-buffer was solely due to the polymerization of the c-α-actins.Polymerization was induced by addition of 2 mM MgCl 2 and 0.1 M KCl.Though The increase of pyrene uorescence with excitation wavelength of 365 nm was monitored at 385 nm using a Schimadzu RF5001PC spectro uorometer.

Critical concentration
To determine the critical concentration of c-α-actin polymerization the varying concentrations of the c-αactins supplemented with 5% pyrene-actin were polymerized in the presence of 2 mM MgCl 2 and 0.1 M KCl overnight.The actin concentrations varied from 0.1 to 10 µM.The steady-state uorescence of polymerized actin was plotted versus monomeric actin concentration and the critical concentration was calculated form the intersections with the abscissa.
Determination of the Ca 2+ -dependence of Tm movement on cardiac F-actins: The Ca 2+ -dependence of pyrene-labelled cTm movement on polymerized c-actin variants was determined by the increase of the eximer pyrene-uorescence at excitation and emission wavelengths of 340 nm and 480 nm, respectively, using an In nite 200 microplate reader (Tecan, Männedorf, Switzerland).Thin laments were reconstituted from each c-actin variant with pyrene-labelled cTm and reconstituted cTn complex, myosin-S1, and N-cMyBP-C each added at a 1:6 molar ratio to actin subunits.Distinct free Ca 2+concentrations in the presence of 1 mM ATP were generated in black 96-well plates [10].Fluorescence intensities were corrected for background uorescence and normalized to F max = 1 and F min = 0. Nine experiments were performed for each c-α-actin variant and condition.The data were tted using a normalized Hill equation (Sigma Plot, Systat Software, Erkrath, Germany).

Generation of recombinant adenoviruses
For the generation of recombinant adenoviruses (Ad) the AdEasy™ kit (Qbiogene) was applied [37,39].DNA sequences encoding wt-and the mutants A295S-, R312H-, and E361G-c-α-actins fused at the Nterminus to a HA-tag were ampli ed by PCR with the primers: 5'ATCATGGATTACCCATACGATGTTC-3' and 5'-ATCGCCCTTTTAGAAGCATTTGCG-3'.As templates served p3xHA-C1 plasmids encoding wt and the mutant cardiac α-actins.The EcoRV site was used to clone PCR inserts into pAdTrack-CMV shuttle plasmid.Electro-competent bacteria E.coli BJ5183 were simultaneously transformed with the shuttle plasmid linearized with the help of PmeI and adenoviral AdEasy-1 DNA backbone.Following homologous recombination in bacteria, clones were screened by restriction with the PacI enzyme that in the case of positive clones resulted in two 33 kb and 4.5 kb fragments.Lipofectamine™ (Invitrogen) reagent was used to transfect HEK293 cells with linearized pAdEasy-1 construct encoding wild type and the mutant cardiac α-actins.Since pAdEasy-1 lacks of E1 and E3 genes critical for successful packaging of adenoviruses, it was crucial to generate adenoviral particles in HEK293 cells, which contain these two genes.The adenoviral DNA encoded additionally EGFP enabling tracking the generation of viral particles.
After two to three weeks, the cells were lysed liberating viral particles.HEK293 cells were twice re-infected with recombinant adenoviruses in order to obtain higher amounts of viral particles.For more details concerning the structure of recombinant adenoviral DNA and the steps of recombinant adenoviruses generation see [22].The correctness of DNA constructs was veri ed by sequencing.
Cell culture and immunohistological procedures Cardiomyocytes from 1-5 days old rats were isolated following the modi ed protocol described in [16,37] as detailed in the Supplementary Information.

Cell transfection
The cells were seeded on glass coverslips in 6-well plates (3 x 10 5 cells/well) and transfected with the help of MATra-A reagent (Iba, Munich, Germany) with 3 µg of DNA encoding either for GFP-actins or HAactins.For Western blot analysis the cells were seeded in 6 cm plastic Petri dishes and transfected with 5 µg DNA.24 h or 48 h after transfection the cells on coverslips were xed with 4% formaldehyde (FA) or harvested in Petri dishes in lysis buffer using a rubber policeman.
NRCs were infected with 20 µl of adenoviruses added to 2 ml of medium following the procedure described [39].72 h after infection the cells were either xed with warm (37 °C) 4% formaldehyde (FA) for immunocytochemistry or harvested with the help of rubber policeman for Western blotting.For controls, cells were infected with only EGFP-encoding viruses.

Confocal microscopy
Control cells, transfected cells and those infected with adenoviruses were xed with warm (37 °C) 4% FA for 20 min at RT and permeabilized with 0.1% Triton X-100 in PBS for 6 min.For staining with anti-cardiac α-actin-antibody, we additionally xed the cells with ice-cold methanol for 6 min at 4 °C.After xation the coverslips or plastic dishes were blocked for 30 min with 3% BSA in PBS.All antibodies were diluted in PBS containing 3% BSA.The cells were immunostained either with goat anti-HA IgGs, or with monoclonal antibodies directed against anti-β-actin, anti-c-α-actin, anti-sarcomeric α-actinin, and anti-myomesin.The secondary IgGs were conjugated either with Alexa Fluor® 488 or Alexa Fluor® 568.In the case of double immunostaining, when the goat anti-HA antibody was applied, donkey anti-mouse IgGs were used in order to avoid cross-reactivity.F-actin was visualized by staining with TRITC-conjugated phalloidin (Sigma-Aldrich).The nuclei were visualized with the help of Hoechst 33342 (Riedel-de-Haen).The coverslips or plastic dishes were washed several times with PBS for 5 min.After all incubations and washing steps the cells were mounted in DAKO cytomatic uorescent mounting medium.Immuno uorescence microscopy was performed using a Zeiss LSM 800 laser-scanning microscope (Jena, Germany).For documentation at least 5 cells were photographed from three independent experiments and a representative image is presented.Co-localization analysis was performed by using ZEN 2007 software (Carl Zeiss Vision GmbH, Goettingen, Germany) and con rmed when the Pearson's correlation coe cient was > 0.3.
Control cells, transfected cells and those infected with adenoviruses were xed with warm (37 °C) 4% FA for 20 min at RT and permeabilized with 0.1% Triton X-100 in PBS for 6 min.For staining with anti-cardiac α-actin-antibody, we additionally xed the cells with ice-cold methanol for 6 min at 4 °C.After xation the coverslips or plastic dishes were blocked for 30 min with 3% BSA in PBS.All antibodies were diluted in PBS containing 3% BSA.The cells were immunostained either with goat anti-HA IgGs, or with monoclonal antibodies directed against anti-β-actin, anti-c-α-actin, anti-sarcomeric α-actinin, and anti-myomesin.The secondary IgGs were conjugated either with Alexa Fluor® 488 or Alexa Fluor® 568.In the case of double immunostaining, when the goat anti-HA antibody was applied, donkey anti-mouse IgGs were used in order to avoid cross-reactivity.F-actin was visualized by staining with TRITC-conjugated phalloidin (Sigma-Aldrich).The nuclei were visualized with the help of Hoechst 33342 (Riedel-de-Haen).The coverslips or plastic dishes were washed several times with PBS for 5 min.After all incubations and washing steps the cells were mounted in DAKO cytomatic uorescent mounting medium.Immuno uorescence microscopy was performed using a Zeiss LSM 800 laser-scanning microscope (Jena, Germany).For documentation at least 5 cells were photographed from three independent experiments and a representative image is presented.Co-localization analysis was performed by using ZEN 2007 software (Carl Zeiss Vision GmbH, Goettingen, Germany) and con rmed when the Pearson's correlation coe cient was > 0.3.
Control cells, transfected cells and those infected with adenoviruses were xed with warm (37 °C) 4% FA for 20 min at RT and permeabilized with 0.1% Triton X-100 in PBS for 6 min.For staining with anti-cardiac α-actin-antibody, we additionally xed the cells with ice-cold methanol for 6 min at 4 °C.After xation the coverslips or plastic dishes were blocked for 30 min with 3% BSA in PBS.All antibodies were diluted in PBS containing 3% BSA.The cells were immunostained either with goat anti-HA IgGs, or with monoclonal antibodies directed against anti-β-actin, anti-c-α-actin, anti-sarcomeric α-actinin, and anti-myomesin.The secondary IgGs were conjugated either with Alexa Fluor® 488 or Alexa Fluor® 568.In the case of double immunostaining, when the goat anti-HA antibody was applied, donkey anti-mouse IgGs were used in order to avoid cross-reactivity.F-actin was visualized by staining with TRITC-conjugated phalloidin (Sigma-Aldrich).The nuclei were visualized with the help of Hoechst 33342 (Riedel-de-Haen).The coverslips or plastic dishes were washed several times with PBS for 5 min.After all incubations and washing steps the cells were mounted in DAKO cytomatic uorescent mounting medium.
Immuno uorescence microscopy was performed using a Zeiss LSM 800 laser-scanning microscope (Jena, Germany).For documentation at least 5 cells were photographed from three independent experiments and a representative image is presented.Co-localization analysis was performed by using ZEN 2007 software (Carl Zeiss Vision GmbH, Goettingen, Germany) and con rmed when the Pearson's correlation coe cient was > 0.3.

Electron microscopy
For negative staining F-actin samples were diluted to 0.1 mg/ml and adsorbed to freshly glow-discharged carbon-coated copper grids (200 mesh) for 45 sec.Negative staining with 1.0% uranyl acetate was performed as described [26,48].The samples were examined in a Zeiss electron microscope EM923 (SESAM) run at 150 kV tted with a TemCamF416 camera (Tietz Video and Image Processing Systems, Gauting, Germany).

Data evaluation
DNA sequences were analysed in DNAstar Lasergene software (DNASTAR Inc., Madison, Wisconsin, USA).Densitometric analysis of bands was performed with the help of the Ultra Quant 6.0 software (Thermo Fisher Scienti c, Schwerte, Germany).Graphs were plotted in Excel 2007 (Microsoft®) or in Origin 8.5 (OriginLab).In both the myosin-S1 ATPase activity and Ca 2+ -dependence of cTm movement assays the data are given as mean values (± SEM, standard error of the mean).In addition, the signi cance of the Ca 2+ -dependency data were analysed by the Student´s t-test and the R-square (R 2 ) analysis using Sigma Plot Software (Systat, Erkrath, Germany).

Expression and puri cation of the cardiac α-actin variants
Expression of the c-α-actins (wt plus the mutants) was achieved in Sf21 insect cells by using the baculovirus system and puri ed after a nity binding to gelsolin G4-6 as reported previously [39] (shown for wt and the p.E361G mutant in the Online Supplementary Information; Fig. S1A,B).The isolated proteins showed only one main band after SDS-Page (Fig. 2A) and were shown to be cardiac α-actin by immunoblotting using an anti-cardiac α-actin antibody (dot blots Fig. 2B and Western blots in Fig. 2C).
Since it was suspected that this puri cation procedure does not discriminate between the expressed cardiac and endogenous insect cell actin, we performed also immunoblots with an anti-ß-actin antibody [37,39] assuming that this antibody recognized the endogenous actin of the insect Sf21-cells.Indeed, after expression in Sf21-cells the puri ed cardiac actins possessed ß-actin immune-reactivity (Fig. 2C).Cardiac α-actin was also conventionally puri ed from acetone powder obtained from bovine hearts.After SDS-Page its migration behaviour and reactivity in Western blots against anti-cardiac α-actin was found identical to recombinant wt c-α-actin.In contrast, prominent anti-β-actin reactivity was observed for puri ed bovine c-α-actin (Fig. 2C) probably due to the presence of β-actin in either broblastic cells or submembranously within the cardiomyocytes themselves.
Initially we had aimed to selectively extract the expressed cardiac α-actins using a N-terminal His-tag.This procedure, however, led to a considerable lower expression and furthermore the puri ed cardiac αactins were suspected not to adopt a fully native con guration as they failed to polymerize.We generated also c-α-actin constructs containing a C-terminal thymosin β4-His-tag [6], however, in our hands also this approach led to only reduced expression and only small amounts could be puri ed that showed also reduced polymerizability.It was because of these failures that we switched to the previously used procedure using His-tagged gelsolin G4-6 as means to effectively a nity purify the Sf21-cell expressed cardiac actins.

Tests for native con guration of the cardiac actin variants
After their puri cation the native state of the α-actins was veri ed by their ability to inhibit DNase I.All the puri ed c-α-actin variants inhibited deoxyribonuclease I (DNase I) activity.Wt c-α-actin inhibited with a lower e ciency than skeletal muscle actin (Fig. 3A).The p.R312H mutant inhibited with a similar activity as wt c-α-actin, whereas the p.E361G and the p.A295S mutants showed a higher or lower inhibitory capacity, respectively (Fig. 3A).These results suggested that the c-α-actin mutants might have attained slightly different conformations though none of the mutations is located within the DNase I-binding loop located in subdomain 2 [27].Native gel electrophoresis used as additional test for their native conformation con rmed that all c-α-actin variants were able to bind thymosin β4 and gelsolin-segment 1 (G1) as described previously for other c-α-actin mutants (not shown; [39]).

Polymerization behaviour of recombinant c-actins
The ability of the isolated cardiac actins to polymerize was analysed by measuring the uorescence increase of added pyrenyl-actin [28].Conventionally puri ed bovine cardiac actin showed almost identical polymerization behaviour as expressed wt c-α-actin (Fig. 3B) giving additional con rmation that the expression of c-α-actins by the baculovirus/Sf21 system did not cause an impairment of its functionality.Clear differences, however, were observed between the mutant c-α-actin variants.The p.E361G DCM mutant showed the fastest polymerization kinetics (Figure 3F).Wt c-α-actin and the p.A295S (HCM) mutant polymerized with almost identical speed, whereas the DCM p.R312H mutant showed reduced kinetics and extent of polymerization (Fig. 3E) as also veri ed by the calculation of the half times of their polymerization (Table 1A).The critical concentrations of their polymerization (Cc) were determined separately (see Table 1B).The data obtained indicated that the Ccs did not differ signi cantly between wt c-α-actin and the mutants except the p.R312H mutant, which exhibited a higher Cc (Table 1) in agreement with previous data showing a higher thermal instability [67].[52].
Therefore, we analysed the polymerization behaviour of the c-α-actin variants in the presence of the nucleators Arp2/3 complex and the active FH2-domain of the formin mDia3, which induce either actin lament networks or straight laments, respectively [29].In these experiments the nucleators were added to actin at a 100:1 ratio before initiating polymerization.The data showed that both nucleators stimulated the polymerization rate of wt c-α-actin (Fig. 3D), whereas the polymerization of the mutants was affected differently: The HCM p.A295S mutant was clearly stimulated by mDia3-FH2, but only minimally by Arp2/3 complex (Fig. 3E).In contrast, the DCM mutant p.R312H was stimulated by both nucleators albeit more strongly by mDia3-FH2 (Fig. 3G).In contrast, Arp2/3 complex had only a small stimulating effect on the rate of polymerization of the p.E316G mutant, whereas mDia3-FH2 appeared not to affect its polymerization behaviour (Fig. 3H).The data (compiled in Table 2) show that the polymerization of most c-α-actin variants was stimulated by mDia3-FH2, except the p.E316G mutant.In summary, these data show signi cant differences in the polymerization behaviour of the analysed c-α-actin variants and its modulation by these nucleating actin-binding proteins that might have functional consequences during cardiomyocyte development.

Table 2
Effect of polymerisation nucleators on cardiac actin variants.Polymerisation parameters of 5 µM cardiac actin variants in the absence or presence of 50 nM Arp2/3 complex or 50 nM mDia3-FH2 (data taken from Fig. 3D-G).The experiments were repeated three times with two different protein preparations.
Half times of polymerization (t 1/2 ) in minutes in the presence of nucleators (data taken from Fig. 3D-G Electron microscopy of laments formed by the c-actin variants Filament formation was further veri ed by electron microscopy after negative staining (EM) before and after decoration with cardiac tropomyosin (cTm) and troponin complex (cTn) at molar ratio of 7:1:1 (Fig. 4).The data indicated formation of normally appearing actin laments at this resolution for bovine and wt recombinant c-α-actin (Fig. 4C).Decoration of with cTm and cTn led to straighter and apparently longer laments (Fig. 4).EM showed that the mutants p.A295S and p.R312H formed more curved laments and frequently exhibited strand breaks (Fig. 4D,E).The p.E361G mutant formed curved and often only short laments (Fig. 4F).After decoration with cTm/cTn, all mutants formed more regular and longer laments though the p.R312H mutant appeared to retain the wavy appearance and fragmentations were still occasionally detectable (Fig. 4E´).

Ca 2+ -dependence of the myosin-S1 ATPase stimulation by c-α-actin variants
The main function of lamentous actin is the stimulation of the myosin ATPase activity during its cyclic interaction with myosin-heads leading to muscle contraction.Therefore we determined the stimulation of the myosin-S1 ATPase by the c-actin variants under two different conditions.The c-actin concentration dependency is given in the Supplementary Information and showed that decoration of the c-actin variants with cTm and cTn increased their stimulatory activity except for the p.R312H mutant (Fig. S3 and Table S1).
Next we determined the Ca 2+ -concentration dependence of the myosin-S1 ATPase stimulation after decoration of the lamentous c-actins with cTm/cTn (see Materials and methods).The absolute ATPase rates (shown in Figure 5A,B and D-F) indicate a rather similar extent of stimulation albeit the maximal ATPase stimulation was highest for bovine c-α-actin and p.A295S (Table 3).The superposition of the normalized rates indicate that bovine and recombinant wt c-α-actin stimulate the myosin-S1 ATPase with a similar Ca 2+ -dependence (Fig. 5C and Table 3).The superposition of the Ca 2+ -dependences of the stimulation of the myosin-S1 ATPase by wt and the three c-α-actin mutants shown in Figure 5G indicated that in comparison to wt c-α-actin the p.A295S mutant showed a slightly higher and the two DCM mutants (p.R312H and p.E361G) a rather similar Ca 2+ -sensitivity (Fig. 6G).The pCa 50 values and differences in Hill coe cients are compiled in Table 3.The data additionally show that the HCM p.A295S mutant possesses in comparison to cTm/cTn decorated wt c-α-actin and the two DCM mutants a higher Ca 2+ -sensitivity (p.R312H and p.E361G), but a lower Hill coe cient (Table 3), i.e. cooperative propagation ability of the myosin-S1 binding (Table 3).The reduced Hill coe cient of the DCM mutants might be due to the fact that the p.R312H and p.E361G mutations are located within the cTm and the myosin head binding regions, respectively (see also Discussion).
Student t-distribution tests and R square (R 2 ) analysis indicated that the measurements were signi cant for each special condition albeit not of high signi cance when comparing different mutants with each other or different conditions (varying compositions of actin-binding proteins, see also Table 3).

Ca 2+ -dependence of tropomyosin movement on c-α-actin variants
Binding of one Ca 2+ -ion to the cardiac troponin subunit TnC induces a conformational alteration of the Tn complex leading to an azimuthal movement of Tm along the actin lament and the exposure of binding sites for myosin heads [13].Functionally three different Tm positions have been identi ed: (i) in relaxed muscle the blocked (B-) state, that inhibits myosin head attachment, (ii) upon cytosolic Ca 2+ -ion increase and binding to TnC the Tm moves to the closed (C-) state, that allows binding of myosin heads with low a nity, (iii) and nally induced by myosin head binding the open (M-) state allowing their binding with high a nity [5,19,20,32,62].The extent of Tm movement is largest (azimuthal angle of about 20°) when going from the B-to the C-state, whereas the transition to the M-state requires only a smaller azimuthal movement (angle change about 5°).Due to a lower energy barrier the transition from C-to M-state occurs much more readily in cardiac than skeletal muscle [32].
Cardiac Tm is a homodimer composed of two α-chains.Labelling with pyrenyl-maleinimide occurs at Cys190 of both chains producing a uorescent eximer [21] that allows to monitor the Ca 2+ -dependency of cTm movement on F-actin from the closed (C-) to the open (M-) state [25].Using this assay we analysed whether the observed Ca 2+ -sensitivities of the ATPase stimulation were paralleled by a similar behaviour of the Tm movement, i.e., a uorescence increase of pyrene-cTm.Therefore, polymerized c-α-actins were decorated with pyrene-labelled cTm and cTn at a 6:1:1 molar ratio (see Materials and methods).Due to the low energy barrier of the C-to M-state transition of cardiac muscle, we expected to observe only a single step in uorescence increase of pyrene-cTm.
The measurements were performed for four different conditions: in the absence and presence of cardiac myosin-S1 plus 1 mM ATP and for both cases with the additional presence of the F-actin binding Nterminal fragment of cardiac myosin binding protein C comprising the C0, C1, M, and C2 domains (subsequently named N-cMyBP-C; for details see Discussion).The C1-domain has been shown to bind to subdomain 1 of actin and cTm, thereby increasing the Ca 2+ -sensitivity of native cardiac thin laments [43,51,57].In native muscle the M-domain is phosphorylated in response to external signals, however, in our experiments we used recombinantly expressed and therefore non-phosphorylated N-cMyPB-C.Like cTm and cTn cardiac N-MyBP-C and cardiac myosin-S1 (where indicated) were added at a 6fold lower molar concentration to actin subunits.
In a rst series of experiments we compared recombinant wt with conventionally puri ed bovine c-α-actin (Fig. 6).Surprisingly, the Ca 2+ -sensitivity of the cTm movement in the presence of only cTm/cTn was lower than that of the stimulation of the myosin-S1 ATPase (Fig. 6; compare Table 3 and 4).The data, however, showed that the addition of myosin-S1 did not increase the Ca 2+ -sensitivity of bovine but of the c-α-actins (Fig. 6; Table 4).However, in the presence of myosin-S1 both bovine and wt c-α-actin exhibited almost identical Ca 2+ -dependency of cTm movement (Fig. 6).Addition of N-cMyBP-C, however, resulted in an increase in Ca 2+ -sensitivity of both bovine and recombinant wt c-α-actin in the absence and presence myosin-S1 (Fig. 6; Table 4).These data indicate that also under conditions more closely resembling the natural situation both bovine and recombinant c-α-actin behaved identically.
Next we tested the Ca 2+ -sensitivity of the mutant c-α-actins after decoration with cTm and cTn under identical conditions.The p.A295S mutant (correlated with HCM) showed an increased Ca 2+ -sensitivity when compared to the p.R312H and p.E361G mutants (correlated to DCM).All mutants showed a decreased Ca 2+ -sensitivity when compared to wt c-α-actin.Differences were observed between the mutant c-α-actins in their Ca 2+ -sensitivity and the effects of addition of myosin-S1 and N-cMyBP-C (Fig. 6; Table 4).In the presence of myosin-S1 the p.A295S mutant showed the highest Ca 2+ -sensitivity in comparison to wt and both DCM mutants (Fig. 6E, Table 4).Addition of N-cMyBP-C had a slightly " atting" effect on the Ca 2+ -sensitivity, i.e. a decrease of the Hill coe cient from 1.7 to 0.98 (Table 4).
Table 4 Parameters of the Ca2+-ion dependence of uorescence increase of pyrene-cTm; i.e. cTm movement The c-actin variants were decorated with pyrene-labelled cTm at a ratio of 6:1 and the dependence of the increase of pyrene-uorescence on Ca2+-concentration was determined (see Materials and methods) in the absence or additional presence of myosin-S1 alone and/or plus N-cMyBP-C.Hill coe cients, pCa50-values and the R2-coe cient were calculated by the Sigma Plot Systat Software (Erkrath, Germany  The p.R312H mutant showed a different behaviour.In the absence of myosin-S1 and N-cMyBP-C this mutant showed the steepest increase Ca 2+ -sensitivity, i.e. the highest Hill coe cient (Table 4), which was reduced by about 50% after addition of myosin-S1 and/or N-cMyBP-C.Its Ca 2+ -sensitivity was not affected by N-cMyBP-C, but further decreased by myosin-S1 (Fig. 6E).
In contrast, for the p.E361G mutant we observed in the absence of myosin-S1 and N-cMyBP-C the highest Ca 2+ -sensitivity that was, however, reduced after addition of myosin-S1 and/or N-cMyBP-C (Fig. 6; Table 4).
Student-t test distribution and R-square (R 2 ) analysis indicated that the measurements were signi cant for each special condition albeit not of high signi cance when comparing different mutants with each other (see Table 3 and 4, which gives the R 2 -values for each condition).Nevertheless, the data suggested that in the presence of only myosin-S1 the HCM p.A295S mutant showed a higher and the DCM R312H and E361G mutants a lower Ca 2+ -sensitivity than wt c-actin corresponding to a higher and lower contractility supposedly typical for HCM and DCM, respectively.

Transfection of established cell lines with cardiac actin variants
The physiological or cell biological effects of these cardiac actin variants were tested also by transfection experiments into established cell lines like HeLa, C2C12 (not shown) and MDCK cells (Fig. 7).For the immunohistochemical localisation the c-α-actins were N-terminally tagged with a HA-(hemagglutinin) extension by standard procedures (see Materials and methods).After transfection (see Materials and methods), the cells were xed, immunostained with anti-HA, TRITC-phalloidin and Hoechst 33342, and examined by confocal microscopy.The results obtained indicated that all c-α-actin variants incorporated into the cellular cytoskeleton of the three cell lines.The data collected for the MDCK cells are shown in Figure 7A-D and demonstrate that wt and p.A295S (HCM) mutant c-α-actins are preferentially incorporated into stress bres forming either lamentous structures composed of apparently only the transfected c-α-actins as indicated by anti-HA staining (green; arrow in Fig. 7A,C) or copolymerizing with the endogenous actin (yellow arrow heads) in the merged images (Fig. 7A´´).In contrast, the p.R312H and p.E361G mutants did not signi cantly form stress bres, but appeared to incorporate into apical structures of the polarized MDCK cells like the sub-membranous actin belt underneath presumed junctional contacts (arrows; Fig. 7B,D) and microvillar extensions (arrows in Fig. 7C,D).

Infection of rat neonatal cardiomyocytes with the cardiac actin variants
In order to observe speci c effects of the cardiac α-actin variants in their natural cell type, adenoviral constructs of the c-α-actin variants were generated as described [37,39] (see also Materials and methods) and used to transduce neonatal rat cardiomyocytes (NRCs) isolated from one to three days old newborn rats.For these experiments the c-α-actins were N-terminally HA-tagged for selective immunostaining.
Subsequent TRITC-phalloidin staining allowed to observe their incorporation into the intracellular lament system, in particular into nascent sarcomeres (Figure 7E-I).The data indicate that the exogenous c-αactins preferentially incorporated into or formed centrally located (around the nuclei) lamentous structures, whereas the TRITC-phalloidin staining appeared to be concentrated in peripheral laments (Fig. 7E-I).Higher magni cation of NRCs transfected with wt HA-c-α-actin showed a gradual increase in the concentration of the exogenous c-α-actins towards the cell centre (Fig. 7E), whereas the TRITCphalloidin staining became weaker towards the cell centre (Fig. 7E) as indicated by a shift in the staining colour from yellow (overlap region) to green (Fig. 7E´).Sarcomeric structures were apparent in the periphery and within the yellow overlap region, whereas they appeared to vanish in the perinuclear region at the cell centre (Fig. 7E) suggesting that the transfected c-α-actin were generated within the perinuclear region and initially formed micro laments before being transformed into sarcomeric structures.
Only the p.R312H mutant showed an additional peripheral localisation when HA-tagged (Fig. 7G).The fact that the centrally located lamentous c-α-actins were only weakly stained by TRITC-phalloidin may suggest that the HA-tag of the exogenous c-α-actins impaired its binding.
Since the N-terminal HA-tag blocks also the binding of the monoclonal anti-cardiac α-actin antibody [39], it became possible to differentiate by immunostaining the localisation of endogenous and exogenous cα-actins.Indeed the obtained data supported the assumption of a distinct distribution of endogenous and exogenous c-α-actins in infected NRCs. Figure 8 shows a clear concentration of the anti-c-α-actin mAB immunostaining for endogenous actin (red; Fig. 8A and merge A´´) at the cell periphery or in peripheral extensions of NRCs expressing wt HA-c-α-actin (anti-HA immunostained in green; Fig. 8A).Similar results were obtained for NRCs infected with the p.A295S mutant (Fig. 8B; only merged image), the p.R312H mutant (Fig. 8C), or the p.E361G mutant (Fig. 8D).In NRCs transduced with vectors coding for the p.A295S or p.E361G mutant, the endogenous c-α-actin was concentrated in peripheral extensions but also present underneath the plasma membrane along the whole cell periphery where it appeared to partially co-localise with the exogenous mutant α-actins (Fig. 8B,D).Nevertheless, the exogenous c-α-actins were clearly concentrated in the middle of the infected NRCs.
Finally, transduced NRCs were counterstained with anti-α-actinin or -myomesin to identify the site of incorporation of the exogenous c-α-actins.Three days after isolation the NRCs had formed intracellular sarcomeric structures, which in some instances spanned the whole cytoplasm.Immunostaining with antiα-actinin (Fig. 8E-I) or -myomesin (Fig. 8J-M) identi ed the Z-or M-lines, to which the thin laments are attached by their plus ends or mark the minus ends, respectively.The data indicated that wt, p.A295S and p.R312H incorporated preferentially at the plus ends, i.e. at the region of the Z-lines (co-localisation with anti-α-actinin staining in Fig. 8E´,F,G), whereas the p.E361G c-actin did not colocalize with the anti-αactinin staining (Fig. 8I).Only occasionally was a colocalisation of the p.E361G staining with anti-αactinin detected (see below).The anti-myomesin staining revealed that wt, p.A295S and p.R312H did not colocalize with myomesin (Fig. 8I,K,L), but a signi cant colocalisation of p-E361G with the anti-myomesin staining (Fig. 8M). Figure 8M also indicated the additional presence of a faint anti-HA-band between the stronger anti-myomesin band colocalising with the anti-HA-p.E361G staining (arrow heads in Fig. 8M) indicating that a small fraction of HA-p.E61G had incorporated at the region of the Z-line.The preferential incorporation of the p.E361G actin at the minus end might be due to the fact that the mutated site is within the binding region of α-actinin, a Z-line component.These data suggest that in case of a high expression level of the p.E361G mutant the thin lament attachment to Z-line might be weakened leading to a reduced stability of the sarcomere.

Discussion
Hypertrophic and dilative cardiomyopathies (CM) are the most frequent genetic diseases of the heart and caused to a high percentage by point mutations of genes encoding sarcomeric proteins.Here we analysed the biochemical and physiological properties of three mutations of cardiac α-actin correlated to cardiomyopathies.For this study wild type cardiac α-actin, the p.A295S mutant causing HCM, and the p.R312H and p.E361G mutants correlated to DCM were expressed using the Sf21/baculovirus system, isolated in native state and analysed.It was hoped that analysing actin mutants causative for either of these two CM forms might reveal distinctly different properties that could be speci cally attributed to a particular CM disease phenotype.
After puri cation the cardiac α-actins were pure as judged by SDS-Page, but appeared to contain about 10% cytoplasmic ß-actin as revealed by Western blotting with an anti-ß-actin speci c antibody.We assumed that the source of ß-actin was the Sf21 insect cells.Bovine cardiac α-actin puri ed from acetone powder contained about the same amount of ß-actin presumably present underneath the cardiomyocyte sarcolemma or in vessel walls.Unfortunately it was not possible to purify the c-α-actins after N-terminal tagging.Nevertheless, we are con dent that the properties and activities of the puri ed αactins are attributable to the cardiac α-actins, since the observed deviations from the wild type behaviour can only be due to alterations of the properties of the mutant forms.
The puri ed c-α-actins appeared to be in native state as veri ed by their ability to inhibit DNase I activity.Bovine, wt and the p.E361G mutant c-α-actin showed identical polymerization behaviour, whereas both p.A295S and p.R312H mutants polymerized much slower though the critical concentrations of polymerization (Cc) were similar and showed only a small difference from the Cc of bovine c-α-actin.
Electron microscopy indicated that all α-actin variants formed typical laments.The mutants, however, appeared to show lament breaks that became less frequent after decoration with cTm and cTn, a treatment that also led to more and longer laments.
In muscle the function of F-actin is to stimulate the myosin ATPase activity.Myosin-S1 of skeletal or cardiac muscle represents the globular N-terminal domain of myosin that contains both the actin binding and the separate ATPase site.Our data show that every c-α-actin isoform was able to stimulate the myosin-S1 ATPase in a concentration dependent manner (see Fig. S3).Decoration with cTm/cTn led to an increase of the myosin-S1 ATPase stimulatory activity of wt, p.A295S and p.E361G c-α-actin probably due to stabilisation of the lament structure as indicated by EM, except for the p.R312H mutant (see Fig. S3) that also after cTm/cTn decoration retained a wavy structure and whose myosin-S1 stimulatory activity did not increase after cTm/Tn decoration (Fig. 4).Previously, a similar behaviour of the p.R312H mutant has been reported that was releated to a reduced stability [41,67].These effects are re ected by the kinetic parameters (see Table S1), which in particular indicate a higher myosin-S1 ATPase stimulatory capacity (V max ) of p.A295S after decoration with cTm/cTn than wt and the DCM c-α-actin variants (Table S1).
Decoration of all c-α-actin variants with cTm/cTn conferred Ca 2+ -sensitivity of their myosin-S1 ATPase stimulatory activity underlining their full functionality.Slight differences in the pCa 50 were observed.The p.A295S HCM mutant showed an at least equal or slightly higher Ca 2+ -sensitivity than wt c-α-actin, which was clearly higher than that of the two DCM variants (see Tables 3 and 4).The data are in line with the assumption that HCM is correlated with a high and DCM with decreased Ca 2+ -sensitivity leading to a higher myosin-ATPase activity already at low Ca 2+ -concentration [62].
The pCa 50 differences between the c-α-actins were principally maintained when determining the uorescence increase of pyrene-cTm presumably paralleling the cTm movement from the C-to M-state though it occurred at lower Ca 2+ -ion concentrations (Fig. 6 and Table 4).The reasons for this difference are presently not clear.It is possible that an internal conformational change of pyrene-cTm leads to an earlier response of the uorescent eximer [25].Using this procedure the p.A295S mutant showed the highest Ca 2+ -sensitivity in the presence of only myosin-S1 (Fig. 6E), whereas that of the p.R312H and p.E361G mutants was lower than of recombinant wt c-α-actin (Fig. 6E and Table 4).Thus these data are in agreement with the notion that HCM or DCM are correlated with and increased or decreased myosin activity, respectively [56, ].
In these measurements we also included the N-terminal fragment C0C2 of cMyBP-C.MyBP-C was rst discovered in skeletal muscle bound to the thick laments, where it is restricted to the overlap region of thick and thin laments [43].MyBP-C is able to attach also to actin and titin thereby connecting thin and thick laments.Meanwhile three isoforms of MyBP-C were identi ed: skeletal, smooth and cardiac muscle speci c isoforms [51,57].Cardiac MyBP-C is composed of eight immunoglobin-like and three bronectin-like domains, which are numbered from C0 (N-terminus) to C10 [51].C0 and C1 are linked by a exible proline-alanine rich linker and between C1 and C2 is placed the M-domain, which is phosphorylated by a number of signalling protein kinases [51,57].In our experiments we used the Nterminal and actin-binding fragment C0-C2 of cardiac MyBP-C, which has been shown to broaden the Ca 2+ -response (decrease of the Hill coe cient) and thereby to increase the Ca 2+ -sensitivity of native thin laments at low Ca 2+ -concentrations [51].Due to its recombinant expression the C0C2-fragment was nonphosphorylated and probably unable to bind myosin-S1.
Our data show N-cMyBP-C was able to increase the Ca 2+ -sensitivity of bovine and wt c-α-actin both in the absence and presence of myosin-S1, whereas its effect on the mutant c-α-actins varied from none to a slight increase in Ca 2+ -sensitivity of p.R312H and p.E361G in the absence of myosin-S1 (Table 4).These different responses might be due to alterations of the interaction of N-cMyBP-C with the mutated actins.
Indeed, previous data have shown that the a nity of the C0C2 fragment is reduced to mutated c-actins especially to the p.E361G mutant [9], because this mutation in subdomain 1 might be located close to the presumed binding region of N-cMyBP-C [51].Nevertheless, it appeared that in the presence of myosin-S1 N-cMyBP-C shifted the Ca 2+ -sensitivity of p.A295S below that of wt (compare Fig. 6E with 6F) suggesting that it might reduce deleterious effects of c-actin mutants.
Finally, we tested the behaviour of the cardiac actins when transfected into established cell lines or rat neonatal cardiomyocytes.The transfection of non-muscle cell lines indicated incorporation of all c-αactins into the cellular cytoskeleton though we noted a preference of wt and p.A295S c-α-actins for the micro lament system, whereas the p.R312H and E361G mutants incorporated into the submembranous cortical F-actin network, a behaviour more typical of cytoplasmic actins.This might appear surprising, since the six mammalian actins contain identical residues at positions 295, 312, and 361 (A, R and E, respectively).Therefore the mutations at the positions 312 and 361 might have reduced the strength of their interactions with lament forming actin binding proteins, especially of the p.E361G mutant, whose rate of polymerization was not stimulated by mDia3-FH2 (Fig. 3).
In contrast, after transduction of neonatal rat cardiomyocytes (NRCs) the c-α-actin variants were found to form or incorporate into sarcomeric structures (Fig. 7,8).A clear separation, however, of the intracellular localisation of exogenous and endogenous c-actins was found in one to two days old NRCs.These cells showed a more central localisation of the exogenous actins, irrespective of the variant, in sarcomeric and micro lamentous organisations, and a peripheral of the endogenous c-α-actin in sarcomeric form (Fig. 7A).Sarcomerogenesis has been shown to start at the cell periphery at sites of attachment to the extracellular matrix and to subsequently extend centripetally [8].These sites (also termed protocostameres) might have been initiated and developed before translation of the adenovirally introduced exogenous c-α-actin variants.It is also possible that the mRNA of endogenous c-α-actin contained additional localisation signals that directed its translation to the cell periphery and therefore attracted preferentially newly formed endogenous c-α-actins [31].Such signals were absent in the adenoviral constructs.
Finally, we determined the site of incorporation of the exogenous c-α-actins into the sarcomeric thin laments by co-staining with anti-α-actinin and -myomesin, markers of the Z-and M-line, respectively.All c-α-actins appeared to incorporate at the Z-line, i.e. the thin lament plus end, except the p.E361G mutant, which preferentially incorporated in the region of the M-line.The p.E361G residue is localised in subdomain 1 of actin and has been implicated in binding to α-actinin [65].This mutation has been shown to reduce its a nity α-actinin [65] and thereby impairs lament plus end incorporation.However, the immunostaining showed a weak HA-p.E361G immunostaining between the myomesin bands indicating albeit reduced plus-end incorporation.

Conclusions
In summary, our data show that recombinant wt behaved very similar to conventionally puri ed bovine c-α-actin.This concordance strengthens the assumption that the isolated mutant c-α-actins were also in native state and exhibited their intrinsic functionality.Our results describe a number of differences of the c-actin mutants in their biochemical and cell biological behaviour.It can be stated that each variant showed individual differences: The HCM p.A295S mutant showed an EM appearance and polymerizability similar to wt c-actin, but was less e ciently stimulated by Arp2/3 complex but stronger by mDia3-FH2 than wt c-actin, a behaviour that might lead to its preferable incorporation into stress bre systems.In agreement with previous reports we also observed that p.A295S possessed a slightly higher Ca 2+ -sensitivity and stimulatory activity of the myosin-S1 ATPase than the other c-actins [64].It has been proposed that its higher Ca 2+ -sensitivity was due to a charge disturbance of the A-triad, since residue 295 is located on a helix opposite the A-triad containing loop [27,32,36].The A-triad is a sequence of positvely charged residues (K326, K328, and R147) that is implicated in stabilising tropomysin binding in the closed-blocked (C,B) position.Since residue A295 is in close vincinity to the A-triad, the alanine to serine mutation may distort the positive charge patch of the A-triad and lead to the observed higher Ca 2+ -sensitivity of cTm movement in the presence of myosin-S1.This behaviour agrees with the current hypothesis that HCM is caused by a hypercontractility.Our data however show that in the absence of myosin-S1 or the presence of N-cMyBP-C this effect less pronounced suggesting that this effect might be restricted to the MyBP-C free overlap region, i.e. to both sides of the Z-line, the region of p.A295S incorporation.
One pecularity of the DCM p.R312H mutant was its steep increase in the Ca 2+ -sensitivity and Hill coe cient (of 3.9), that was however considerably reduced by myosin-S1 and not observed in the ATPase measurements (Fig. 5).The p.R312H residue is located in subdomain 3 of actin (Fig. 1) and close to the A-triad that is involved in electrostatic binding of tropomyosin [13,36].Furthermore, the neighbouring residue D311 forms the strongest electrostatic contact to Tm [13].Therefore a mutation leading to a reduction of positive charges of this region might lead to reduced a nity of the p.R312H mutant to cTm and be the reason for the high Hill coe cient (of 3.9) observed in the pyrene-cTm measurement indicative for a low energy barrier of the transition from B-to M-state (Table 4) [11].In agreement with previous reports [36,64] the DCM p.R312H mutant showed a reduced polymerizability (Fig. 3C), though positively affected by the nucleators.Interestingly another recently detected amino acid exchange at this position, the p.R312K variant, did not affect actin polymerization, but was not in uenced by the two nucleators (see Supplementary Information).EM indicated that p.R312H-actin laments were wavy and showed strand breaks that were still observed after decoration with cTm/Tn.Both results suggested reduced stability of p.R312H laments, which might explain the observation that decoration with cTm/Tn did not further increase its myosin-S1 ATPase stimulatory capacity.In the presence of both myosin-S1 and N-cMyBP-C the Ca 2+ -sensitivity decreased slightly most probably due to their simultaneous binding to cTm and c-actin (Table 4).It appears however probable that the reduced lamentous stability of p.R312H cactin leading to reduced myosin-S1 ATPase activation might be causative for DCM development.
The DCM p.E361G mutation is located in subdomain 1 the binding region of myosin heads and α-actinin [5,67].The p.E361G mutation is reported to cause a three fold lower a nity to α-actinin [67] that might be responsible for its preferred incorporation into into cytoskeletal F-actin networks of culture cells and the thin lament minus-ends, although a small fraction was observed to also incorporate into the plus-ends (Fig. 8M).This might lead to a reduction of the stability of the attachment of thin laments to the Z-lines.
Indeed, it was observed that the hearts of transgenic mice expressing the E361G mutant developed contractile dysfunction under stress and signs of DCM [55], although this mutant showed a high rate of polymerization that was not further stimulated by Arp2/3 complex and mDia3-FH2.
Furthermore, previous investigations have shown that the p.E361G mutation abolishes the Ca 2+ -ion regulatory effect of cTnI phosphorylation [55].Phosphorylation of two N-terminal serines of cTnI by protein kinase A (PKA) after ß1-adrenergic stimulation increases the Ca 2+ -sensitivity and by interaction with TnC the rate Ca 2+ -dissociation and subsequently the heart rate [55,66].The p.E361G mutant was shown to blunt the increase in Ca 2+ -sensitivity affected cTnI phosphorylation, thereby the adrenergic responsiveness of cardiac muscle is lost leading to a decrease of its Ca 2+ -sensitivity further supporting the development of DCM [55].
In summary, our data indicate that each c-actin mutation possesses its own speci c alterations in the mode of interactions with particular sarcomeric proteins.Apart from the general accepted notion that in the Ca 2+ -sensitivity affecting the contractlity can lead to either HCM or DCM, our data demonstrate that modi es interactions of mutant c-actins with other sarcomeric proteins during development and in adult cardiac muscle may also play an essential role for CM establishment.

(
3D structural model of G-actin derived from the skeletal muscle actin:DNase I complex[27] indicating the positions of the mutated residues of the c-actin mutants investigated.In addition subdomains (SD) as well as the position of ATP and Ca2+ are indicated.

Figure 3 Properties
Figure 3

Table 1
B) independently determined critical concentrations of polymerization (Cc) of the cardiac actin variants (determined as detailed in Materials and methods).The experiments were repeated three times with two different protein preparations.Variations are given as SEM.
contains also the data obtained for the p.R312K variant, which has been supposed to cause also DCM (personal communication (Dr.Schoenenberger, Basel, Switzerland).Since no clinical reports exist demonstrating a correlation of p.R312K to CM and its polymerization properties closely resembled wt c-α-actin (see Supplementary Information), we did include this variant in further studies.(During cardiomyocyte differentiation and sarcomerogenesis a number of different actin-binding proteins are involved in the processes of micro lament transformation into sarcomeres [8].Indeed recent data have shown the presence of actin polymerization nucleating factors (nucleators) in cardiac muscle

Table 3
Parameters from Ca2+ -ion dependence of myosin-S1 ATPase stimulation.Parameters are derived from experiments depicted in Figure5.All measurements were performedfor each c-actin variant in lamentous form after decortion with cTm and cTn at a ratio of 6:1:1.Hill coe cients, pCa50-values and the R-squared (R2) values were calculated by the Sigma Plot Systat Software (Erkrath, Germany).N gives the number of experiments.Variations are given as SEM.The fold stimulation was calculated from the ratio of the maximal (max.)/minimal(min.)ATPase activity.
). N gives the number of experiments.Variations are given as SEM.