Dilated cardiomyopathy mutations in three thin filament regulatory proteins result in a common functional phenotype.

Dilated cardiomyopathy (DCM), characterized by cardiac dilatation and contractile dysfunction, is a major cause of heart failure. Inherited DCM can result from mutations in the genes encoding cardiac troponin T, troponin C, and alpha-tropomyosin; different mutations in the same genes cause hypertrophic cardiomyopathy. To understand how certain mutations lead specifically to DCM, we have investigated their effect on contractile function by comparing wild-type and mutant recombinant proteins. Because initial studies on two troponin T mutations have generated conflicting findings, we analyzed all eight published DCM mutations in troponin T, troponin C, and alpha-tropomyosin in a range of in vitro assays. Thin filaments, reconstituted with a 1:1 ratio of mutant/wild-type proteins (the likely in vivo ratio), all showed reduced Ca(2+) sensitivity of activation in ATPase and motility assays, and except for one alpha-tropomyosin mutant showed lower maximum Ca(2+) activation. Incorporation of either of two troponin T mutants in skinned cardiac trabeculae also decreased Ca(2+) sensitivity of force generation. Structure/function considerations imply that the diverse thin filament DCM mutations affect different aspects of regulatory function yet change contractility in a consistent manner. The DCM mutations depress myofibrillar function, an effect fundamentally opposite to that of hypertrophic cardiomyopathy-causing thin filament mutations, suggesting that decreased contractility may trigger pathways that ultimately lead to the clinical phenotype.

ious environmental insults, such as viral myocarditis or alcohol toxicity, are known to cause DCM; however, echocardiographic screening of relatives of affected individuals has suggested that about one-third of cases are familial (1,2). Around 20 loci have been identified for Mendelian traits in which DCM is the predominant phenotype, and in all except 2 of these, the disease is inherited in an autosomal dominant manner (3). In some instances DCM is inherited with an associated phenotype such as conduction disease, skeletal myopathy, or sensorineural hearing loss.
The underlying disease genes have been identified at some of the mapped loci, and others have been implicated from candidate gene studies; these disease genes encode a wide variety of proteins, suggesting that diverse abnormalities can trigger the observed cardiac remodeling. Many of the genes encode components of the cytoskeleton: these include proteins that colocalize at the sites of attachment of the cytoskeleton to the membrane (costameres), for example, desmin (4), and Z-disk components, such as muscle LIM protein (5). This has led to the proposals that the primary deficit in DCM is either in force transmission, due to destabilization of the membrane attachment at the costameres (6), or due to defective stretch sensing at the Z-disk (5). However, mutations in at least six genes encoding components of the contractile apparatus (␤-myosin heavy chain (7), titin (8), cardiac actin (9), ␣-tropomyosin (10), cardiac troponin T (TnT) (7), and cardiac troponin C (TnC) (11)) can also cause autosomal dominant DCM, and some, if not all of these, are likely to cause changes in force generation.
It is known that different mutations in the same contractile protein genes cause hypertrophic cardiomyopathy (HCM) (12), a more common familial disease with a contrasting phenotype of a hypertrophied ventricle with a small cavity and preserved systolic function. The finding that both HCM and DCM can be caused by different mutations in the same sarcomeric protein raises the question of whether there are two separate programs that remodel the heart, or whether these different morphologies reflect gradations of a single pathway (12). For example, transgenic mice bearing a truncation allele of myosin-binding protein C (which produces HCM in man) show hypertrophy in heterozygotes but dilated cardiomyopathy in homozygotes (13); this might suggest that severity of the underlying defect (in this case, determined by dosage), rather than the qualitative nature of the defect, could determine the resultant phenotype. However, the underlying histology is quite different because the hallmark feature of HCM, myocyte disarray, is absent in hearts of individuals with DCM due to contractile protein gene mutations (7).
Analysis of the function of mutated protein in reconstituted contractile systems has been used to address these questions. * This work was supported by the British Heart Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Many studies of different HCM thin filament mutations have been performed; these indicate a variety of functional abnormalities that generally include increased Ca 2ϩ sensitivity and cross-bridge turnover rate (14). In contrast, experiments have only been reported for two of the DCM mutations to date, both being mutations in troponin T. Incorporation of the ⌬Lys-210 mutant into skinned cardiac fibers was found to result in a decrease in the Ca 2ϩ sensitivity of tension generation (15,16). Our in vitro actin-activated ATPase and motility analyses of the ⌬Lys-210 mutant using thin filaments reconstituted with an equal amount of wild-type and mutant troponin (the expected in vivo ratio) showed that this mutation caused reduced Ca 2ϩ sensitivity and decreased thin filament activation (17). A second TnT mutant (R141W) was shown to cause reduced Ca 2ϩ sensitivity in one study (18), but a further study found no significant shift, leading to the suggestion that a rightward shift in Ca 2ϩ sensitivity was not a consistent feature of these DCM mutants (16).
To resolve uncertainty regarding the functional phenotype of the sarcomeric DCM mutations and hence the relationship between DCM and HCM pathways, we extended our work to examine all reported autosomal dominant DCM mutations in thin filament regulatory proteins. We have studied five TnT mutants (R131W (11), R141W (19), R205L (11), ⌬Lys-210 (7), and D270N (11)), two ␣-tropomyosin mutants (E40K (10) and E54K (10)), and the single novel mutation in cardiac TnC (G159D (11)) in in vitro actin-activated ATPase and motility assays and measured the effect of two TnT mutants on isometric tension in skinned fibers. We show that reduced Ca 2ϩ sensitivity is indeed a consistent property of all eight of these mutants and find that seven of these mutants also reduce thin filament activation. This suggests that this specific defect in myofilament regulation provides the primary stimulus for a pathway leading to myocyte death and ensuing ventricular dilatation and dysfunction.

MATERIALS AND METHODS
Actin and the myosin proteolytic fragments heavy meromyosin (HMM) and subfragment-1 (S-1) were prepared from rabbit skeletal muscle by standard methods. Human cardiac tropomyosin and troponin were purified as previously described (20). Recombinant wild-type human troponin subunits were overexpressed in BL21(DE3)pLysS Escherichia coli and subsequently purified. pMW172 expression constructs encoding DCM mutations in TnT and TnC were made from the wildtype construct using a two-step PCR protocol. Wild-type and DCM mutant human Ala-Ser ␣-tropomyosin constructs were made, and the protein was expressed and purified as described previously (21). Whole troponin complexes were reconstituted from subunits using our previously described techniques (17,22).
The movement of thin filaments over a bed of immobilized HMM was investigated by in vitro motility assay as described previously (17,(23)(24)(25). We used F-actin labeled with tetramethylrhodamine isothiocyanate phalloidin () and skeletal muscle HMM on cover glasses coated with silicone by soaking in 0.2% dichloroedimethylsilane in chloroform. Two 50-l aliquots of HMM at 100 g/ml were infused in buffer A (50 mM KCl, 25 mM imidazole-HCl, 4 mM MgCl 2 , 1 mM EDTA, 5 mM dithiothreitol, pH 7.4) to provide a coating of immobilized HMM on the coverslip. This was followed by 2 ϫ 50 l of buffer B (buffer A ϩ 0.5 mg/ml bovine serum albumin) and then by 2 ϫ 50 l of 10 nM labeled actin Ϯ associated tropomyosin-troponin in buffer A. 50 l of buffer C (buffer B ϩ 0.1 mg/ml glucose oxidase, 0.02 mg/ml catalase, 3 mg/ml glucose, 0.5% methylcellulose, Ϯ troponin at assay concentration) and 50 l of buffer D (buffer C ϩ 1 mM ATP) were then infused. All experiments were carried out at 28°C with final concentrations of reconstituted thin filament proteins as follows: rabbit skeletal actin, 10 nM; human cardiac tropomyosin, 30 nM; and reconstituted human cardiac troponin, 0 -20 nM. Both tropomyosin and troponin were titrated to determine saturating concentrations. Four 15-s videos were recorded in each cell before any significant photobleaching had occurred. Videos were digitized, and the movement was analyzed to determine the fraction of filaments moving and the velocity of motile filaments using the automatic tracking program described by Marston et al. (24).
The Ca 2ϩ -dependent regulation of isometric contraction was measured in chemically skinned guinea pig heart trabeculae by replacing endogenous troponin with whole wild-type or mutant human cardiac recombinant troponin using the method introduced by She et al. (26). Albino Dunkin Hartley guinea pigs weighing ϳ250 g were killed humanely by CO 2 asphyxiation by authorized United Kingdom Home Office procedures. Trabeculae were removed from the left ventricle and immersed in 2ϫ relaxing solution (relaxing solution: 60 mM BES (pH 7.2), 6.36 mM MgCl 2 , 10 mM EGTA, 6.26 mM Na 2 ATP, and 54.02 mM K 2 propionate) containing 50% (v/v) glycerol and 1% (v/v) Triton X-100 for 12 h at 2°C. The trabeculae were then transferred to 2ϫ relaxing solution with 50% (v/v) glycerol and stored at Ϫ20°C for up to 7 days.
The trabeculae of ϳ2.0 mm in length and 0.2 mm in width were suspended between a static hook, and another was attached to the semi-conductor element of a force transducer. The trabeculae were immersed in relaxing solution at 12°C until a stable baseline was achieved, and the length and width of the fiber were measured. The fiber was then immersed in a zero EGTA solution (60 mM BES (pH 7.2), 6.09 mM MgCl 2 , 6.32 mM Na 2 ATP, and 74.70 mM K 2 propionate). Maximum tension was measured using a pCa 4.5 solution (60 mM BES (pH 7.2), 6.11 mM MgCl 2 , 10 mM EGTA, 6.31 mM Na 2 ATP, 43.66 mM K 2 propionate, and 10.07 mM CaCl 2 ), and each fiber was maximally activated three times. The maximum force generated by these fibers prior to exchange was 30.27 Ϯ 6.88 mN/mm Ϫ2 , and the pCa 50 was 5.73 Ϯ 0.13.
For the exchange of troponin, the trabeculae were activated in a pCa 4.5 solution and then immediately placed into a quick rinse solution (10 mM imidazole (pH 7.0), 5 mM EGTA, and 15 mM EDTA) for 5 min. During this time, the fiber developed a rigor force that was approximately one-third of the total Ca 2ϩ -sensitive force. Following this, the trabecula was placed in a rigor solution (170 mM NaCl, 10 mM imidazole (pH 7.0), 2.5 mM EGTA, and 2.5 mM EDTA) for 10 min, followed by 20 min in exchange solution (170 mM NaCl, 10 mM imidazole (pH 6.8), 5 mM MgCl 2 , 5 mM EGTA, and 5 mM dithiothreitol) without any troponin. During this time, the temperature was increased to a nominal temperature of 22°C. The fiber was incubated in 0.5 mg/ml human recombinant cardiac troponin for 1 h. Subsequent to the troponin exchange, the fiber was washed three times (10 min each) in rigor solution and then washed twice (20 min each) in relaxing solution. During the latter incubation, the temperature of the cooling bath was returned to 12°C. The fiber was maximally activated in a pCa 4.5 solution, and after that, the fiber was immersed sequentially in solutions containing increasing quantities of Ca 2ϩ , and the level of contraction was determined. The pCa-force curves were fitted to a four-parameter Hill equation using SigmaPlot 2002, and from this, the pCa 50 and Hill coefficient were calculated.
Trabeculae were separated by SDS-PAGE and silver-stained. Human recombinant cardiac TnI migrates slightly faster than the guinea pig cardiac TnI subunit, and densitometric analysis was used to quantitate both proteins. A standard curve of cardiac TnI was used for calibration.
All data are presented as mean Ϯ S.E. unless otherwise indicated. Means of mutant data were compared with wild-type using the paired Student's t test.

RESULTS
Human cardiac troponin was reconstituted using recombinant troponin T, I, and C subunits. Wild-type troponin and troponin containing either a TnT or TnC DCM mutant were compared in thin filaments reconstituted with recombinant human Ala-Ser ␣-tropomyosin (for ATPase assays) or native human cardiac tropomyosin (for in vitro motility studies). To study mutations in tropomyosin, thin filaments were reconstituted using recombinant wild-type or mutant Ala-Ser ␣-tropomyosin with either recombinant troponin (ATPase assays) or native human troponin (in vitro motility). The Ala-Ser N-terminal addition substitutes for the absence of N-terminal acetylation of the bacterially produced peptide (27,28). Direct binding measurements, made under the conditions of the ATPase assay, showed normal stoichiometry for binding of all troponin subunits and tropomyosin to actin (data not shown).

Effect of DCM Mutations on Maximum Activation and Inhibi-
tion-Addition of wild-type or mutant troponin to actin-tropomyosin filaments resulted in Ca 2ϩ -sensitive regulation of S-1 ATPase activity and filament motility. Titration experiments showed that the concentration of troponin needed to obtain a maximal effect was the same for wild-type and all mutant proteins; a typical set of titrations for wild-type troponin and troponin containing the TnT R131W mutation is shown in Fig. 1. At pCa 4.5, wild-type troponin activated acto-S-1 ATPase activity to 202 Ϯ 1% of the actin-tropomyosin level, whereas at pCa 8.5, ATPase was reduced to 21 Ϯ 1% of actin-tropomyosin activity as previously noted (17,22) (Fig. 2A). In equivalent experiments in which wild-type TnT, TnC, or tropomyosin was substituted with a DCM mutant component, the maximal activation was significantly reduced (p Ͻ 0.001) for seven of the eight mutants tested (Figs. 1 and 2A). The most marked reduction was obtained with the TnT R131W mutant (to 109 Ϯ 4% of the actin-tropomyosin level). The ␣-tropomyosin E54K DCM mutant was unique because it had no effect upon maximal ATPase activation. In con- trast to the changes observed in activating Ca 2ϩ , at pCa 8.5, no significant differences were detected between wild-type and six of the eight mutants; small, statistically significant decreases in the inhibited rate were detected using the TnT R205L and ␣-tropomyosin E54K mutants.
In the in vitro motility assay, addition of troponin to actintropomyosin increased the sliding speed at pCa 5.4 by up to 28% but did not alter the fraction of filaments motile (Figs. 1 and 2, C and D). At pCa 9, sliding speed of motile filaments was 50 -70% of actin-tropomyosin filaments, and the fraction of filaments motile was reduced to 10 -15% (20). The presence of DCM mutations in the thin filaments caused significant decreases compared with wild-type in the troponin-dependent speed enhancement at pCa 5.4 (Figs. 1 and 2C; p Ͻ 0.001 for TnT R131W; p Ͻ 0.01 for TnT R141W, TnT R205L, TnT ⌬Lys-210, and TnC G159D; and p Ͻ 0.05 for ␣-tropomyosin E40K). In contrast, neither the fully activated nor the fully inhibited fraction of filaments motile was significantly different from wild-type for any of the mutations (Fig. 2D). Again, the ␣-tropomyosin E54K mutation was unique because it had no effect on sliding speed at pCa 5.
The Effect of DCM Mutations on Ca 2ϩ Sensitivity-We determined the effect of DCM mutations on the Ca 2ϩ concentration dependence of troponin regulation of actomyosin ATPase and in vitro motility parameters, using concentrations of troponin that gave maximal inhibition under relaxing conditions.  Table  I). For wild-type thin filaments, the mean pCa 50 for ATPase regulation was 6.44 Ϯ 0.02 (n ϭ 14); in the motility studies, the mean pCa 50 was 6.69 Ϯ 0.05 (n ϭ 18) for fraction of filaments motile and 6.35 Ϯ 0.05 (n ϭ 18) for filament velocity. With the exception of the TnT ⌬Lys-210 mutation previously described (17), all the DCM mutations caused a rightward shift of the curve for all three measured parameters in direct comparison with wild-type in paired experiments. The decrease in pCa 50 was in the range 0.09 (TnT R141W) to 0.48 (␣-tropomyosin E54K) pCa units in the ATPase assays and between 0.18 (TnT R270N) and 0.40 (TnT R205L) in motility studies. All the differences in pCa 50 were significant in the ATPase assay, and the differences in pCa 50 were also significant in in vitro motility assay when there were sufficient independent measurements for statistical analysis. The Hill coefficient, n H , a measure of cooperativity of Ca 2ϩ activation, was significantly reduced for every TnT and TnC mutant, but no significant difference in n H was detected with the two ␣-tropomyosin mutants ( Fig. 4D and Table I obtained from in vitro motility assay (data not shown).
The Effect of Equimolar Mixtures of Wild-type and Mutant Troponin or Tropomyosin-We examined the regulatory behavior of thin filaments composed of 1:1 mixtures of wild-type and mutant troponin or tropomyosin. Because these DCM mutations are dominant, affected individuals are heterozygous, and therefore it is likely that wild-type and mutant protein are present at approximately a 1:1 ratio in heart muscle. In 1:1 mixtures with wild-type protein, there was a consistent pattern of results; all the DCM mutations caused decreases in pCa 50 , which were statistically significant (Figs. 3 and 5 and Table I). For most TnT mutations and the TnC mutation, the changes in pCa 50 and n H were less than that observed with 100% mutant protein. As previously reported, the TnT ⌬Lys-210 mutation, which anomalously increased Ca 2ϩ sensitivity on its own, decreased Ca 2ϩ sensitivity in the physiological 1:1 mixture (17). TnT R141W gave a greater reduction in Ca 2ϩ sensitivity of ATPase (but not motility) regulation when present in a 1:1 mixture with wild-type than when on its own, whereas ␣-tropomyosin E54K showed a greater decrease in pCa 50 using 1:1 mixtures than with 100% mutant in motility assay but not in ATPase studies.
The Effect of DCM Mutations on Force Production in Troponin-substituted Permeabilized Muscle-We examined the effect of two DCM mutations in TnT on tension generation in skinned guinea pig trabeculae, selecting one mutation lying within the extended N-terminal T1 region (R141W) and a second mutation in the globular T2 domain (R205L). Whole recombinant human cardiac troponin was used to displace the endogenous complex using the method of She et al. (26). The amount of exchange was quantified by densitometric analysis of cardiac TnI in trabeculae analyzed by SDS-PAGE and silver staining; human and guinea pig forms were distinguished by different mobilities. No significant differences in the level of incorporation were found between trabeculae treated with wild-type troponin (human troponin constituted 58 Ϯ 5% of total after exchange), trabeculae treated with TnT R141W-containing complex (52 Ϯ 6%), and trabeculae treated with TnT R205L-containing complex (55 Ϯ 5%).
The maximal Ca 2ϩ -sensitive tension measured in trabeculae treated with wild-type troponin was 21.44 Ϯ 1.62 mN/mm 2 (n ϭ 6 fibers). The trabeculae treated with TnT R141W-containing complex showed a nonsignificant increase in maximal Ca 2ϩsensitive tension with a value of 27.08 Ϯ 3.32 mN/mm 2 (n ϭ 5). In contrast, fibers containing the R205L mutant generated a significantly diminished amount of force compared with the wild-type control (12.63 Ϯ 1.86 mN/mm 2 , n ϭ 5; p Ͻ 0.001).
Steady-state pCa-force curves were achieved by sequentially  Table I. B, ⌬pCa 50 for thin filament sliding speed. Significance tests gave p Ͻ 0.001 for TnC G159D; p Ͻ 0.01 for TnT R141W, TnT⌬Lys-210, and ␣-tropomyosin E54K; and p Ͻ 0.05 for TnT D270N and ␣-tropomyosin E40K. C, ⌬pCa 50 for fraction of filament motile. Significance tests gave p Ͻ 0.001 for TnT R141W and TnT ⌬Lys-210 and p Ͻ 0.05 for TnC G159D and ␣-tropomyosin E40K. D, n H for activation of S-1 ATPase activity. raising the level of free Ca 2ϩ . Fibers containing either TnT R141W or TnT R205L gave a significantly reduced Ca 2ϩ sensitivity compared with wild-type (pCa 50 wild-type ϭ 5.58 Ϯ 0.01; TnT R141W ϭ 5.48 Ϯ 0.04, ⌬pCa 50 ϭ Ϫ0.10, p Ͻ 0.001; TnT R205L ϭ 5.37 Ϯ 0.06, ⌬pCa 50 ϭ Ϫ0.21, p Ͻ 0.001) with no significant alterations in the Hill coefficient (Fig. 6). The magnitude of the ⌬pCa 50 was similar to the 1:1 mixtures of mutant and wild-type troponin measured in vitro (Fig. 5). DISCUSSION The finding of mutations in genes encoding thin filament proteins that cause contrasting disease phenotypes provides a valuable opportunity to dissect disease mechanisms and the relationship of structure to function. In this work, we made a comprehensive investigation of all eight known DCM mutations in TnT, TnC, and ␣-tropomyosin and used a range of techniques that measure different aspects of contractility.
To model the effect of these mutations as closely as possible, we have incorporated them into the human cardiac troponin and tropomyosin sequences and used native human cardiac tropomyosin or troponin as partner protein wherever possible. This is important because our previous work on HCM mutations has shown that the origin of the thin filament partner proteins may influence the functional properties (22). In addition to comparing the effect of wild-type and mutant protein, we investigated the functional properties of 1:1 wild-type/mutant mixtures, such as might be expected in heterozygous individuals with autosomal dominant DCM. The ratio of wildtype/mutant protein present in tissue of individuals with autosomal dominant cardiomyopathies has proved difficult to measure, and accurate reports are limited. Existing data for the ␣-tropomyosin D175A and ␤-myosin heavy chain R403G HCM mutations (21,29) suggest wild-type and mutant protein are indeed present in equimolar ratios. Our previous experiments investigating mixtures of mutant and wild-type proteins have shown non-linear relationships, such that the effect of 1:1 mixtures cannot be predicted by interpolation from the findings with 100% mutant and 100% wild-type preparations (17,22). Using these conditions, we found a strikingly consistent functional phenotype caused by all eight mutations associated with DCM, which is independent of both the location of the mutation in the protein and the protein bearing the mutation. This consistency provides a clear starting point for exploring the downstream consequences of these mutations and will aid confirmation of the pathogenic role of novel putative DCM-causing mutations, as illustrated by the G159D allele, which is the first validated disease-causing mutation in TnC in man.
Common Functional Effects of DCM Mutations-The feature common to all the DCM mutations we investigated is a decrease in the Ca 2ϩ sensitivity by 0.1-0.3 pCa unit (1.3-2.0-fold increase in EC 50 ) in 1:1 wild-type/mutant mixtures (Fig. 5). This decreased Ca 2ϩ sensitivity was observed in ATPase and in vitro motility assays (Figs. 3 and 5) and in isometric force measurements on TnT R141W and R205L made with approximately 50% troponin-substituted muscle fibers (Fig. 6). This pattern of results is in agreement with the reduction of Ca 2ϩ sensitivity caused by the ⌬Lys-210 and R141W TnT mutations in partially substituted rabbit skeletal muscle fibers reported by Morimoto and co-workers (15,18) and also with the effect of ⌬Lys-210 troponin T in fully substituted pig heart fibers reported by Venkatraman et al. (16). The latter group found that the TnT R141W mutant produced a nonsignificant 0.04-pCa unit decrease in Ca 2ϩ sensitivity, much less than that we observed in comparable experiments (Figs. 4 and 6). However, a significant decrease in Ca 2ϩ sensitivity was reported when the TnT R141W mutation was incorporated into the TnT T1 isoform present in fetal heart (30).
With the exception of tropomyosin E54K, all the mutants also showed a diminished maximum ATPase rate and slower sliding speed than wild-type both in 100% and 50:50 mixtures (Fig. 2). A decrease in Hill coefficient, reflecting loss of cooperativity, was also common to all the TnT and TnC mutations but was not found with the two ␣-tropomyosin mutations (Fig. 4). The lack of effect of the tropomyosin mutations on cooperativity may just reflect limitations of the assay used because Ala-Ser-␣-tropomyosin differs subtly from native acetylated tropomyosin in some aspects, including cooperativity (31).
The functional properties of these DCM mutations contrast sharply with mutations in the same proteins associated with hypertrophic cardiomyopathy. Most investigators have found increased rate of cross-bridge turnover at pCa 5, impaired relaxation at pCa 9, and/or increase in Ca 2ϩ sensitivity using a range of investigation techniques (12, 14, 25, 32). An apparent exception to this pattern is the work of Burton et al. (33), who examined two HCM mutations (R145G and G203S) in TnI TABLE I Effect of mutations on Ca 2ϩ sensitivity of actin-tropomyosin-troponin activation of S-1 ATPase Ca 2ϩ dependency of thin filament-activated S-1 ATPase was determined for wild-type and mutant troponin/tropomyosin. pCa 50 and n H were derived by fitting the data to the Hill equation. The table shows ⌬pCa 50 (calculated as the difference between wild-type and mutant pCa 50 values in paired experiments) and n H for all DCM mutants averaged from at least three independent experiments. Comparative data for 100% mutant and 50:50 wild-type/mutant mixtures are shown. The significance of the differences between wild-type and mutant results was determined by paired Student's t test. Significance is indicated as follows: NS, p Ͼ 0.05 (nonsignificant); *, p Ͻ 0.05; **, p Ͻ 0.01; and ***, p Ͻ 0.001.

Mutation
ATPase ⌬pCa  using chemically skinned guinea pig trabeculae. The vanadate technique was used to remove endogenous troponins I and C, and exogenous human cardiac troponins I and C were then used to reconstitute troponin in situ. In these experiments, the R145G mutant gave a small increase in Ca 2ϩ sensitivity (much smaller than that reported elsewhere (34)) and the G203S TnI gave a decrease in Ca 2ϩ sensitivity of tension development. However, our subsequent experimentation, using the troponin replacement method described here, has established that the G203S mutant does not decrease Ca 2ϩ sensitivity, and we suggest that the earlier results were due to an artifact of the vanadate technique. 2 Molecular Basis of Contractile Abnormalities and Structure/ Function Implications-Four of the mutations are located in the "regulatory core" of troponin, for which two x-ray diffraction structures have been published (35,36). The TnT mutations at Arg-205 and Lys-210 are located close to Thr-203, a major site of protein kinase C phosphorylation (37), in the N terminus of the T2 domain (Fig. 7A). The two DCM mutations involve a loss of positive charge, and the phosphorylation involves a gain of two negative charges at physiological pH. Both the mutations and phosphorylation at Thr-203 produce a decrease in maximal ATPase and Ca 2ϩ sensitivity. In the troponin crystal structures, these amino acids are in "helix 1" (residues 191-220 in the cardiac sequence), which is remote from TnC and TnI; therefore, it is likely that this helix primarily interacts with tropomyosin and actin. This is in accord with the 2 L. Preston, C. Redwood, and C. Ashley, unpublished data.  (38) based on deletion mutation studies, in which helix 1 is proposed to bind to tropomyosin and/or actin in the absence of Ca 2ϩ and be released from the thin filament surface when Ca 2ϩ activates the filament. The orientation of both mutated amino acids is away from the regulatory core, and therefore the mutations could be in a TnT-tropomyosin interface, where they may be expected to alter Ca 2ϩ activation of the thin filament.
In both published Ca 2ϩ -troponin crystal structures, the TnC Gly-159 residue does not appear to interact with TnI or TnT in the presence of Ca 2ϩ . However, the structure of skeletal troponin in the absence of Ca 2ϩ shows that this residue is close to the N-terminal helix of TnI helix, notably Gln-15, so that the mutation at TnC Gly-159 could reposition the C terminus of TnC relative to the N terminus of TnI.
Intriguingly, an earlier NMR study of Blumenschein et al. (39) suggested that residue Gly-160 of skeletal muscle TnC.4Ca 2ϩ (equivalent to cardiac TnC Gly-159) interacts with the fast skeletal muscle TnT peptide 160 -193 (equivalent to human cardiac TnT 188 -224, corresponding to helix 1 and containing the 203, 205, and 210 residues). However, there is no evidence for this in either the cardiac or skeletal troponin crystal structures.
The cardiac troponin crystal structure suggests a structural role for the TnT residue Asp-270. The ␥-carboxyl group of Asp-270 is hydrogen-bonded to two amino acids of TnC, Tyr-111 and Arg-147, which form part of the high affinity Ca 2ϩbinding loops at sites III and IV of the C-terminal domain, respectively, and are themselves coordinated to the Ca 2ϩ ions (Fig. 7B). The substitution of aspartate with asparagine would eliminate one of the two hydrogen bonds.
The DCM mutations R131W and R141W lie in the N-terminal T1 region of TnT. This domain is predicted to be a largely ␣-helical peptide that has been shown to bind along tropomyosin in an antiparallel orientation that extends over the junction between adjacent tropomyosin molecules (40). The two mutations could affect tropomyosin-TnT binding, which in turn could affect tropomyosin interactions with actin and TnT interactions with TnI and TnC. Troponin T1 binding to the tropomyosin overlap zone is thought to be necessary for the in-crease in cooperative unit size induced by troponin binding to actin-tropomyosin (41). The precise part of TnT involved in this interaction is unknown: Hinkle and Tobacman suggest 112-136, which includes Arg-131 (42), whereas Palm et al. suggest 92-110 (43).
The two DCM mutations in tropomyosin are located near the N terminus, and both mutated amino acids lie in the "e" position of the heptad repeat, which usually forms an interchain salt bridge with an amino acid in the "g" position that stabilizes the coiled-coil structure of tropomyosin. The 2 Å crystal structure of tropomyosin 1-81 (Fig. 7C) shows that Glu-40 is linked to Arg-35 and that Glu-54 is linked to Lys-49. The mutation of glutamic acid to lysine would abolish the interaction in either position and thus destabilize tropomyosin structure, at least locally. A recent study of two HCM mutations in this region of tropomyosin, A63V and K70T, has indeed demonstrated destabilization of the N-terminal half of tropomyosin (44), and it has been shown that Ca 2ϩ and troponin-dependent conformational changes are propagated to this region of tropomyosin in thin filaments (45). It is noted that Glu-40 and Glu-54 are in different positions in the 19-amino acid actin-binding pseudorepeat, which may account for their different effects upon maximally activated ATPase and sliding (46,47).
Implications for Disease Pathways in DCM-Our data address the issue of whether sarcomeric HCM and DCM result from two separate disease pathways, or whether these different outcomes reflect gradations of a single pathway (12). In the single pathway model, the extent rather than the type of contractile defect might determine the outcome: moderate contractile dysfunction might be compensated and thereby lead to hypertrophy, whereas analogous abnormalities that are more severe may might bring about myocyte death. However, because the contractility changes observed with the DCM mutants are qualitatively opposite to those caused by HCM mutations in the same gene (12,14,25), our data strongly argue that a separate pathway is initiated by the DCM thin filament mutations.
All aspects of the aberrant thin filament regulation observed in vitro (decreased Ca 2ϩ sensitivity, maximum activation, and cooperativity) strongly suggest that these DCM mutations will cause decreased force across the range of Ca 2ϩ concentrations FIG. 7. Location of mutations in troponin and tropomyosin structure. A, partial troponin structure determined by x-ray diffraction (Protein Data Bank code 1J1D) (35). TnC is blue, TnI is green, and TnT is cyan. Ribbon representation was drawn in RASMOL to show the helices. The sites of mutations TnT R205L, ⌬Lys-210, D270N, and TnC G159D are shown in space filling format. B, detail of the hydrogen bonding between TnT Asp-270 and TnC Tyr-111 and Arg-147 visualized using Swiss-PDB viewer. C, location of Glu-40 (red) and Glu-54 (yellow) in the crystal structure of ␣-tropomyosin 1-81 (Protein Data Bank code 1IC2) (54) showing the interchain interactions with Arg-35 (green) and Lys-48 (cyan), respectively. that operate in vivo. These changes are opposite to the increased activation that characterizes thin filament mutations that produce HCM, which result in an increased cost of force production that may lead to energetic compromise (48). Depressed contractility is a consistent feature of end-stage heart failure, and this has been correlated with defects at the level of the contractile apparatus (49,50), leading to slower crossbridge cycling and altered Ca 2ϩ sensitivity, as well as abnormal calcium regulation and signaling. Alterations in troponin function have been associated with end-stage heart failure (51,52), but it has been unclear how frequently a primary reduction in myocyte contractility initiates human heart failure (53). However, for the class of inherited DCM mutations studied in this work, reduced contractility is likely to be the initial stimulus. It is not yet clear how these fundamental changes in myofilament function trigger pathways leading to apoptosis and subsequent remodeling. The other genetic abnormalities that lead to DCM are diverse, but most could involve either defects in force transmission or force production. Thus, most DCM mutations share in common the potential for both diminished contractility and a failure of normal signaling, both of which in turn may impair myocyte viability. In the case of the thin filament DCM mutations, one possibility is that it is the combination of the deficits in myofilament activation that is ultimately unsustainable.