Alterations in Thin Filament Regulation Induced by a Human Cardiac Troponin T Mutant That Causes Dilated Cardiomyopathy Are Distinct from Those Induced by Troponin T Mutants That Cause Hypertrophic Cardiomyopathy*

We have compared the in vitroregulatory properties of recombinant human cardiac troponin reconstituted using wild type troponin T with troponin containing the ΔLys-210 troponin T mutant that causes dilated cardiomyopathy (DCM) and the R92Q troponin T known to cause hypertrophic cardiomyopathy (HCM). Troponin containing ΔLys-210 troponin T inhibited actin-tropomyosin-activated myosin subfragment-1 ATPase activity to the same extent as wild type at pCa8.5 (>80%) but produced substantially less enhancement of ATPase atpCa4.5. The Ca2+ sensitivity of ATPase activation was increased (ΔpCa50 = +0.2pCa units) and cooperativity of Ca2+ activation was virtually abolished. Equimolar mixtures of wild type and ΔLys-210 troponin T gave a lower Ca2+ sensitivity than with wild type, while maintaining the diminished ATPase activation atpCa4.5 observed with 100% mutant. In contrast, R92Q troponin gave reduced inhibition at pCa8.5 but greater activation than wild type at pCa4.5; Ca2+sensitivity was increased but there was no change in cooperativity.In vitro motility assay of reconstituted thin filaments confirmed the ATPase results and moreover indicated that the predominant effect of the ΔLys-210 mutation was a reduced sliding speed. The functional consequences of this DCM mutation are qualitatively different from the R92Q or any other studied HCM troponin T mutation, suggesting that DCM and HCM may be triggered by distinct primary stimuli.

Dilated cardiomyopathy (DCM) 1 is defined clinically by cardiac chamber dilatation with reduced contractile performance in the absence of underlying coronary artery disease. The heart appears thin walled and distended and, at the microscopic level, there is moderate myocyte hypertrophy and death, along with replacement fibrosis. Echocardiographic screening of relatives of affected individuals suggests that ϳ25-35% of cases are familial (1,2). The disease is frequently inherited with an associated phenotype such as conduction disease, skeletal myopathy, or sensorineural hearing loss. To date, as many as 18 loci that cause DCM as the predominant phenotype have been identified, and in all but two of these, the disease is inherited in an autosomal dominant manner (3,4). For 10 of the loci, the disease genes have been identified. These encode a diverse range of proteins, including components of the sarcomere: actin (ACTC) (5); ␤-myosin heavy chain (MYH7) (6); titin (TTN) (7); ␣-tropomyosin (TPM1) (8); and cardiac troponin T (TNNT2) (6).
In a recent report, Kamisago et al. (6) identified two mutations in ␤-myosin heavy chain and one in cardiac troponin T (the deletion of lysine 210) in kindreds having autosomal dominant dilated cardiomyopathy without conduction disease, skeletal muscle dysfunction, or other accompanying phenotypes. It was noteworthy that affected subjects did not have ventricular hypertrophy, and histology from one subject showed mildly increased interstitial fibrosis without the myocyte and myofibrillar disarray characteristic of hypertrophic cardiomyopathy (HCM). These mutations therefore appear to cause dilated cardiomyopathy directly and induce a phenotype that is distinctly different from HCM.
HCM is known to be caused by mutations in at least 10 genes, all but one of which encodes a sarcomeric protein (3,9). In contrast to the contractile protein gene mutations that cause DCM, the functional consequences of the HCM mutations have been extensively characterized (reviewed in Refs. 3, 10, and 11). Most mutations in sarcomeric proteins have been found to increase maximum shortening speed and/or Ca 2ϩ sensitivity in vitro, which may result in energetic compromise through increased cost of force production in vivo (9,(12)(13)(14)(15). In HCM families, some individuals go on to develop a dilated cardiomyopathy phenotype, presumably through induction of apoptosis (3). Thus, one plausible hypothesis to explain how different mutations in the same gene can cause different cardiomyopathies is that DCM mutations produce similar, but more severe, perturbations of contractile protein function, sufficient to result in cell death. Alternatively, the DCM mutations in sarcomeric protein genes could initiate disease through qualitatively different perturbations of contractility (6,16).
Functional analysis of different mutations within a single gene that produce the divergent phenotypes of HCM and DCM provides a valuable opportunity to investigate the triggers that discriminate between these two disease pathways. In this report, we have focused on the ⌬Lys-210 troponin T mutant that causes DCM. In common with the HCM troponin T mutations, it is highly likely that this apparently subtle mutation acts in a dominant-negative manner and is incorporated into the thin filament, where it affects normal thin filament function. The deleted amino acid forms one of a stretch of four lysine residues in human cardiac troponin T (amino acids 207-210). These lie within the globular C-terminal T2 domain (residues 188 -288) FIG which binds to troponins I and C as well as to tropomyosin (17) and may therefore affect thin filament function by a variety of mechanisms.
We have compared the changes in thin filament function caused by the ⌬Lys-210 mutation with those caused by a mutation in troponin T that causes HCM. For the latter we have used the R92Q troponin T mutant, which has been extensively characterized in transgenic mouse models (13,18,19) and myofibrils (20, 21) 2 but has not yet been examined in reconstituted thin filaments in vitro. We found that the ⌬Lys-210 mutation had distinctive effects upon thin filament function: the maximally activated ATPase activity and filament sliding speed were decreased, and Ca 2ϩ activation became non-cooperative. This pattern of changes was quite unlike the effect of the R92Q HCM mutation (increased ATPase activity and sliding speed and higher Ca 2ϩ sensitivity with unaltered cooperativity) but closely resembled the properties of troponin ex-tracted from end stage failing human hearts studied by the same techniques (22,23).

MATERIALS AND METHODS
Purification and Preparation of Proteins-Rabbit skeletal muscle actin, rabbit and human cardiac muscle ␣-tropomyosin, and subfragment-1 (S-1) derived by chymotryptic digestion of whole rabbit skeletal muscle myosin were prepared as previously described (22,24). Recombinant wild type human troponin subunits were overexpressed in BL21(DE3)pLysS Escherichia coli and subsequently purified (24). pMW172 expression constructs encoding R92Q and ⌬Lys-210 troponin T were made, respectively, by subcloning from an existing plasmid cytomegalovirus construct (12) and using a two-step PCR protocol for site-directed mutagenesis.
Whole troponin complexes were formed using a development of our established protocol (25). The subunits were mixed in a ratio of 1.5 troponin C:1 troponin I:1 troponin T in 6 M urea, 1 M KCl, 10 mM imidazole, 50 M CaCl 2 , 1 mM dithiothreitol, 0.01% sodium azide, pH 7.0, and the concentrations of first urea and then KCl were reduced using a stepwise dialysis protocol to 0 and 200 mM, respectively. The mixtures were centrifuged (12,300 ϫ g, 5 min) to remove insoluble material and intact troponin purified by gel filtration using a Sepharose 2 R. Willott and C. Ashley, unpublished data.  Fig. 1. For the binding experiments (C), thin filaments were reconstituted using biotin-phalloidin-labeled actin, and filaments were selected from the motility assay mixture (100 nM actin) with streptavidin-coated Dynabeads, which were recovered by a magnetic separator. SDS-PAGE (8 -18% gradient) analysis of wild type and mutant filaments (C). Troponin T, I, and C and tropomyosin are clearly incorporated into both wild type and mutant thin filament under these conditions. 200 column. Complexes were then dialyzed into the appropriate assay buffer. The final proportions of individual subunits were measured by scanning densitometry and found to be 1.00:0.95:1.12 (troponin T:troponin I:troponin C; n ϭ 3) for wild type and both mutants.
Actin-Tropomyosin-activated Myosin ATPase Assay-Assays were carried out as previously described using 0.5 M myosin S-1 and thin filaments reconstituted using either 3.5 M actin, 1 M tropomyosin, and 1 M troponin or 3.5 M actin, 0.5 M tropomyosin, and 0.5 M troponin in 5 mM PIPES, 3.87 mM MgCl 2 , 1 mM dithiothreitol, pH 7.0, at 37°C (25). The free Ca 2ϩ concentration was set using 1 mM EGTA and the appropriate concentration of CaCl 2 as previously described (25).
In Vitro Motility Assay-The movement of thin filaments over a bed of immobilized skeletal muscle heavy meromyosin (HMM) was investigated using the in vitro motility assay as we have described (24,26). F-actin was labeled with rhodamine-phalloidin () as described by Kron et al. (27). HMM at 100 g/ml was infused in buffer A (50 mM KCl, 25 mM imidazole-HCl, pH 7.4, 4 mM MgCl 2 , 1 mM EDTA, 5 mM dithiothreitol) to provide a coating of immobilized HMM on the coverslip. The surface was blocked by infusing 0.5 mg/ml bovine serum albumin in buffer A, and then reconstituted thin filaments were infused. All experiments were carried out at 28°C with the following final concentrations of reconstituted thin filament proteins: 10 nM rabbit skeletal actin, 30 nM human cardiac tropomyosin, 0 -20 nM reconstituted human cardiac troponin in buffer A plus 0.1 mg/ml glucose oxidase, 0.02 mg/ml catalase, 3 mg/ml glucose, 0.5% methylcellulose, 1 mM MgATP, Ϯ troponin at assay concentration. Both tropomyosin and troponin were titrated to determine saturating concentrations. The movement of actin () tropomyosin filaments over the immobilized skeletal muscle HMM was observed under a Zeiss epifluorescence microscope (ϫ63/1.4 objective). 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 fraction of filaments moving and velocity of motile filaments using the automatic tracking program described by Marston et al. (28).

RESULTS
Troponin complexes were reconstituted from recombinant human cardiac troponins I, C, and T using wild type troponin T, ⌬Lys-210 mutant troponin T, or R92Q mutant troponin T. Thin filaments were assembled using rabbit skeletal actin, rabbit or human ␣-tropomyosin, and wild type or mutant troponin. The functional properties of wild type and mutant troponin T were compared by assay of thin filament activation of skeletal muscle myosin S-1 ATPase activity and by the in vitro motility assay using skeletal muscle heavy meromyosin. Actin cosedimentation assays carried out as previously described (25) showed that the binding of the wild type and both mutant troponins to actin-tropomyosin were indistinguishable (data not shown).
In ATPase experiments carried out under activating conditions, the addition of wild type troponin increased ATPase activity, reaching a plateau at 207 Ϯ 3% of the ATPase rate obtained using actin-tropomyosin alone, whereas when troponin containing ⌬Lys-210 troponin T was added the maximum ATPase activation was only 167 Ϯ 6% of the actin-tropomyosin rate. This difference was consistently seen in four different troponin preparations and was highly significant (p Ͻ0.001) (Fig. 1A).
At pCa 5.4, troponin increased the speed of actin-tropomyosin filament sliding in the in vitro motility assay, reaching a plateau at 10 -15 nM troponin ( Fig. 2A). As previously observed (24,27), wild type troponin increased sliding velocity to 125 Ϯ 2.6% that of actin-tropomyosin filaments. However, filaments reconstituted with troponin containing ⌬Lys-210 troponin T gave a significantly reduced sliding speed (111 Ϯ 2.7% of actintropomyosin), and this difference was highly reproducible (p Ͻ0.0001, 11 separate experiments, Fig. 1B). Troponin containing ⌬Lys-210 mutant troponin T also induced a small but significant decrease (p Ͻ0.05) in the fraction of filaments motile under activating conditions compared with wild type troponin (Figs. 1C and 2B). Gel electrophoresis of the reconstituted thin filaments showed that both wild type and mutant thin filaments contained a full complement of bound troponin subunits and tropomyosin (Fig. 2C).
Under relaxing conditions (pCa9), ATPase activation and filament motility of thin filaments containing wild type and ⌬Lys-210 mutant troponin T were indistinguishable. ATPase activity was inhibited by ϳ90% (Fig. 1A), and in vitro motility was switched off by troponin such that Ͻ10% of the filaments were moving, and the speed of the remaining moving filaments was about 40% slower than actin-tropomyosin filaments, in agreement with previous measurements (24, 29) using this system (Fig. 1, B and C). The maximum switch-off of motility required 10 -15 nM troponin with both wild type and ⌬Lys-210 mutant troponin T. Comparing data from paired preparations of thin filaments, we found no significant difference in the fraction of filaments moving at saturating troponin concentrations (Fig. 1C) nor at partially saturating concentrations (2.5-5 nM troponin). These results demonstrate that neither the EC 50 of inhibition caused by troponin containing ⌬Lys-210 mutant troponin T, nor the maximum inhibition of motility, was different from wild type.
Thin filaments reconstituted with troponin containing the HCM troponin T mutation R92Q reproducibly gave enhanced activation of ATPase activity (224 Ϯ 5% of actin-tropomyosin activity compared with 207 Ϯ 3% using wild type) and thin filament sliding velocity (Fig. 1, A and B). In addition, at pCa9 thin filaments containing R92Q troponin T reproducibly gave less inhibition of ATPase (79% compared with 90% using wild type troponin) and less inhibition of filament motility than wild type (p ϭ Ͻ0.05; see Fig. 1, A and C).
We determined the Ca 2ϩ concentration dependence of thin filament activation of S-1 ATPase activity and motility using concentrations of troponin that gave maximal inhibition of filament activity at pCa9 (1 M for ATPase, 15 nM for motility assay; Fig. 3 and Table I). A consistent pattern of results was obtained with both techniques: the Ca 2ϩ -dependent curves for  Table I. ATPase activation, fraction of filaments motile, and filament velocity of thin filaments containing ⌬Lys-210 mutant troponin T were much less steep (i.e. less cooperative) than those obtained using wild type troponin, and 50% activation was shifted to lower Ca 2ϩ concentrations. The data were fitted to the Hill equation: rate ϭ a ϩ b/ (1 ϩ 10 nH(pCa-pCa50) ). The ⌬Lys-210 mutation increased calculated pCa 50 by 0.19 Ϯ 0.03, 0.28 Ϯ 0.04, and 0.31 Ϯ 0.03 pCa units for ATPase, fraction of filaments motile, and sliding speed, respectively. The apparent Hill coefficients, n H , derived from fits of the mutant troponin data were close to unity (n H ϭ 1.24 Ϯ 0.06, 0.88 Ϯ 0.19, and 0.90 Ϯ .12, respectively) and the ATPase data were well fitted by a simple saturation equation (rate ϭ V max [Ca 2ϩ ]/ (EC 50 ϩ[Ca 2ϩ ]) with EC 50 ϭ 0.25 Ϯ 0.03 M), indicating that these thin filaments were not cooperatively activated by Ca 2ϩ . In contrast, the R92Q troponin T mutation had a different effect; the Ca 2ϩ sensitivity of ATPase regulation was increased with no change in cooperativity (Table I).
The disease caused by both the DCM and HCM troponin T mutations is autosomal dominant, and it is likely that the cardiac thin filaments of affected individuals contain similar proportions of wild type and mutant troponin. Our previous work has shown that the effect on thin filament function of mixtures of wild type and mutant troponin is not directly predictable from the functional properties of the mutant troponin alone (24,25). Thin filaments were reconstituted using stoichiometric amounts (7:1:1, respectively) of actin, tropomyosin, and total troponin (either wild type, mutant, or 50:50 wild type mutant). Filaments containing an eqimolar mix of wild type and mutant gave activation relative to actin-tropomyosin at pCa4.5 that was similar to the level using 100% mutant complex and significantly less (p Ͻ0.001) than that obtained with wild type troponin (Fig. 4A). Similarly, in the in vitro motility assay the speed of sliding of a 50:50 mixture was close to that of 100% mutant (2.31 Ϯ 0.04, 2.34 Ϯ 0.06, 2.53 Ϯ 0.01 m/sec for 100% mutant, 50:50 mixture, and 100% wild type, respectively). In contrast, thin filaments reconstituted with an equimolar mix of wild type troponin T and R92Q mutant troponin T gave similar levels of activation and inhibition as 100% wild type (Fig. 4B).
Interestingly, whereas 100% ⌬Lys-210 troponin T mutant gave an increase in Ca 2ϩ sensitivity of ATPase activation of ϩ0.23 pCa units, the ⌬Lys-210/wild type mixture resulted in a significant decrease in Ca 2ϩ sensitivity compared with wild type; ⌬pCa 50 ϭ Ϫ0.27 (p Ͻ0.001; n ϭ 4) (Fig. 4A). The apparent Hill coefficient for the 50:50 mixture was intermediate between 100% wild type and 100% mutant and was significantly different from that obtained with wild type troponin alone (p Ͻ0.001). The same pattern of results was observed using in vitro motility assay: in thin filaments with a 50:50 mixture of wild type and ⌬Lys-210 troponin, the Ca 2ϩ sensitivity of the fraction motile parameter was less than wild type by 0.07 pCa units (data not shown).

DISCUSSION
The deletion of lysine 210 in cardiac troponin T has been reported to be a cause of inherited dilated cardiomyopathy (6). When human cardiac troponin T with this mutation was incorporated into reconstituted thin filaments, we found a pattern of functional changes in vitro that was distinctly different from the changes previously observed with hypertrophic cardiomyopathy mutations. The Ca 2ϩ -activated rate of actomyosin ATP hydrolysis and the thin filament sliding speed were reduced compared with wild type troponin, the Ca 2ϩ activation curve became non-cooperative, and pCa 50 was increased. It is particularly noteworthy that a 50:50 mixture of wild type and mutant troponin T, which is likely to reflect the situation in vivo, still reduced the maximally activated actomyosin ATPase and filament sliding speed to the same level as 100% mutant troponin T but gave Ca 2ϩ sensitivity significantly lower than wild type.
The deleted amino acid forms one of a stretch of four lysine residues in human cardiac troponin T (amino acids 207-210). These amino acids lie within the C-terminal chymotryptic T2 fragment known to bind tropomyosin, troponin C, and troponin I (17). Studies of peptides and deletions within this region have indicated that troponin I binds to heptad repeat sequences C-terminal to these four lysines (approximately residues 229 -268) (30), whereas both troponin C and tropomyosin may interact close to or directly with these residues (31). The recently reported (32) crystal structure of the T2 fragment in complex with troponins I and C shows that residue 210 forms part of a short ␣-helix N-terminal to the separate helix involved in a coiled coil interaction with troponin I that has no direct association with any other troponin subunit. The change in maximum sliding speed and ATPase is compatible with experiments that have shown that one function of troponin T is to determine the cross-bridge turnover rate (12,26,33); the remarkable uncoupling of cooperative activation could be related to the role of troponin T in determining the size of the thin filament cooperative unit through its interaction with tropomyosin (34,35). The recent report of a DCM mutation within the N-terminal T1 domain of troponin T, which only binds to tropomyosin (36), and two mutations in tropomyosin itself (8) also suggest that alterations in tropomyosin-troponin T interactions may be responsible for the appearance of the DCM phenotype (reviewed in Ref. 16).
The decreased Ca 2ϩ sensitivity observed with 50:50 wild type/⌬Lys-210 mixtures was surprising given the significant increase in Ca 2ϩ sensitivity obtained in experiments using 100% DCM mutant troponin T. This emphasizes our previous findings that the regulatory properties of a 50:50 wild type/ mutant troponin mixture are often quite different from both wild type and mutant troponin in a way that could not be predicted from the functional properties of the mutant troponin alone (24,25,29). The laboratory of Morimoto has recently reported a similar decreased Ca 2ϩ sensitivity in rabbit heart trabeculae in which endogenous troponin was displaced upon incubation with human troponin T (either wild type or mutant) and human complex reconstituted in situ by the addition of human troponins I and C (37). It appears that these treated trabeculae contained about 50% human mutant troponin T and 50% endogenous rabbit troponin T. Decreased Ca 2ϩ sensitivity together with the decreased cross-bridge turnover rate and reduced cooperativity suggest that the expression and incorporation of this DCM troponin T mutant in vivo may result in myofilaments that are markedly less responsive to Ca 2ϩ , contract more slowly, and thus are unable to produce sufficient force during activation.
Interestingly, the changes in contractility predicted to be caused by the DCM troponin T mutant closely resemble the alterations in myofilament regulation identified in end stage failing human hearts. Investigations using tissue from hearts affected by a number of aetiologies (including non-familial DCM) have shown a consistent pattern of reduced unloaded shortening speed and reduced myofibrillar ATPase activity (38 -40). Reports are less clear over possible changes in Ca 2ϩ sensitivity that has been suggested to be increased or unaltered; however, recent work suggests that additional factors such as sarcomere length and phosphate concentration are involved in determining which way the Ca 2ϩ sensitivity will change (41)(42)(43). The reduced unloaded shortening speed correlates with the reduced ATPase and sliding speed caused by the ⌬Lys-210 mutation in troponin T. The effects of the ⌬Lys-210 mutation also correlate with the altered properties of troponin extracted from failing hearts. Failing heart troponin was found to confer a slower maximum sliding speed and a higher Ca 2ϩ sensitivity with reduced cooperativity (22,23). Thus it is possible that a reduced maximum rate of cross-bridge turnover is a common causative feature of end stage heart failure, because a defect of this kind could not be fully compensated for by increasing Ca 2ϩ sensitivity or modifying Ca 2ϩ handling (44). It remains unclear whether these contractility changes in end stage failing hearts mediate the progress of heart failure or whether they are compensatory epiphenomena. However, the data presented here that show a troponin T mutant that is responsible for inherited DCM causing similar changes in contractility supports the notion that alterations to troponin may contribute directly to the progression of acquired forms of DCM.
The functional alterations caused by the ⌬Lys-210 troponin T mutation are in many respects opposite from the data presented here for the R92Q HCM mutant and from our analyses of other HCM troponin T mutants (Ref. 24). 3 Furthermore, this combination of properties is distinct from any other reported in vitro biochemical study on HCM troponin T mutants (21,45,46), suggesting that the reduced maximal activation, depressed cooperativity, and, at an equimolar ratio with wild type troponin, diminished Ca 2ϩ sensitivity may provide a specific stimulus for the production of the dilated, rather than hypertrophic, phenotype.