Analysis of myosin heavy chain functionality in the heart.

Comparison of mammalian cardiac alpha- and beta-myosin heavy chain isoforms reveals 93% identity. To date, genetic methodologies have effected only minor switches in the mammalian cardiac myosin isoforms. Using cardiac-specific transgenesis, we have now obtained major myosin isoform shifts and/or replacements. Clusters of non-identical amino acids are found in functionally important regions, i.e. the surface loops 1 and 2, suggesting that these structures may regulate isoform-specific characteristics. Loop 1 alters filament sliding velocity, whereas Loop 2 modulates actin-activated ATPase rate in Dictyostelium myosin, but this remains untested in mammalian cardiac myosins. Alpha --> beta isoform switches were engineered into mouse hearts via transgenesis. To assess the structural basis of isoform diversity, chimeric myosins in which the sequences of either Loop 1+Loop 2 or Loop 2 of alpha-myosin were exchanged for those of beta-myosin were expressed in vivo. 2-fold differences in filament sliding velocity and ATPase activity were found between the two isoforms. Filament sliding velocity of the Loop 1+Loop 2 chimera and the ATPase activities of both loop chimeras were not significantly different compared with alpha-myosin. In mouse cardiac isoforms, myosin functionality does not depend on Loop 1 or Loop 2 sequences and must lie partially in other non-homologous residues.

Myosin, the molecular motor of the heart, generates force and motion by coupling its ATPase activity to its cyclic interaction with actin. Myosin is a hexameric protein and is composed of two heavy chains (MHC) 1 and two essential and two regulatory myosin light chains. Structurally, MHC is composed of a number of discrete domains: a helical rod necessary for thick filament formation, and a globular head that contains the actin-binding site, catalytic, and motor domains (1).
In the mammalian heart, two functionally distinct MHC isoforms, termed V 1 and V 3 , are present. V 1 is a homodimer of two ␣-MHC molecules, whereas V 3 is a ␤␤-homodimer. Expression of V 1 and V 3 is controlled both developmentally and hormonally. In the mouse, ␤-MHC expression in the ventricles predominates prenatally. However, via thyroid hormone regulation, ␤-MHC expression is silenced at birth, and ␣-MHC is transcribed (2). The functional differences between V 1 and V 3 myosin in terms of shortening velocity, force generation, and ATPase activity are profound. For example, rabbit V 1 myosin has a 2-3-fold faster actin filament sliding velocity than V 3 , but generates only half the average isometric force (3,4). Likewise, both the Ca 2ϩ -stimulated and actin-activated ATPase activities of rabbit V 1 myosin are ϳ2-3 times greater than for V 3 myosin (3,5). Similar differences in actin velocity and myofibrillar ATPase activity have been observed between mouse V 1 and V 3 myosin, but there is no difference in their average force generation (6).
Although the proteins are functionally distinct, the primary amino acid sequences of mouse ␣and ␤-MHC are 93% identical. Thus cardiac isoform diversity must lie in the non-identical residues (127 of 1938 amino acids in mice). The differences in the enzyme kinetics and mechanics of the myosin interactions that are observed between the two cardiac isoforms are believed to reside in two, hypervariable "loops," so called because their structures cannot be defined via x-ray crystallography due to their relative disorder. Loop 1 (L 1 ), which is located between residues 213 and 223, is at the mouth of the nucleotide pocket while Loop 2 (L 2 ), at positions 624 -646, cradles the long cleft running from the actin binding site to the nucleotide binding pocket (1,7). Comparison of MHC sequences within the human sarcomeric MHC family shows that these domains of sequence variability are conserved (8). Spudich and co-workers (9 -11) have proposed that Loop 1 modulates velocity through ADP release, whereas Loop 2 helps regulate the actin-activated ATPase rate. Data obtained from studies with chimeric Dictyostelium and smooth muscle myosins corroborated the model. For example, chimeras were constructed in which 9 amino acids in the L 2 region of Dictyostelium myosin II were substituted with the corresponding residues from other myosins such as rabbit skeletal muscle myosin, chicken smooth muscle myosin, or rat cardiac myosin (11). The actin-activated ATPase activities of the chimeras correlated well with the activities of the myosin from which the Loop 2 sequence was derived. Thus, myosin's ATPase activity could be specifically modulated de-pending on the sequence of Loop 2. However, a number of studies indicate that the loops may not influence myosin kinetics and mechanics as proposed and thus have varying roles depending on the structure of the myosin backbone. Rat and pig ␤-MHC, which have identical Loop 1 sequences apart from a single conservative substitution, have 3-4-fold differences in ATPase activity and ADP dissociation (12). Sweeney et al. (13) showed that the properties of Loop 1 chimeras with a smooth muscle backbone are a function of loop size/flexibility rather than related to the properties of the myosin from which Loop 1 was derived. Furthermore, chimeric myosins that consisted of a Dictyostelium MHC backbone with carp loop sequences did not exhibit changes in sliding velocity if Loop 1 was substituted, although Loop 2 substitution did lead to the expected modulation of actin-activated ATPase activity (14). Taken together, these studies indicate that the role of the surface loops for MHC functionality depends on the interplay of the surface loops with other regions important for myosin mechanics and kinetics.
In contrast to the abundance of detailed studies on in vitro function of various MHC isoforms, our current knowledge of how differences found on the single molecule level are reflected in in vivo cardiac function is limited. Cardiac isoform shifts can be achieved by endocrine intervention but hypothyroidism not only results in a nearly complete V 1 3 V 3 shift in rodent hearts, it also induces a number of structural changes including mitochondrial swelling, as well as rupture and loss of continuity of the myofilaments (15). Thus one cannot dissect MHC isoform shift induced functional changes from contractile impairment due to structural damage. Transgenesis avoids these issues. To date, only a single transgenic study has dealt directly with cardiac myosin isoform substitution, showing that contractile function of TG mouse hearts with low-level expression of Myctagged rat ␤-MHC was reduced by 15% (16). This disproportionate impairment of contractile function might be due to the presence of a heterologous species (rat) cDNA being placed into the mouse context, resulting in a dominant negative effect of the TG protein.
The present study is the first to investigate directly the functional significance of cardiac isoform diversity by using TG mice in which ventricular V 1 is largely replaced by V 3 . This approach has the advantage of effecting isoform replacement in the heart without the pleiotropic stimuli that are normally used to induce MHC isoform transitions, such as pressureoverload induced hypertrophy or changes in hormonal status (17,18). A significant V 1 3 V 3 shift resulted in the expected changes in the heart at the single motor and biochemical levels as well as in fiber mechanics and kinetics. However, in contrast to a hypothyroidism-effected replacement, cellular structure appeared normal and whole organ function was preserved with relatively minor effects on systolic and diastolic hemodynamics in the intact animal. Furthermore, we generated transgenic (TG) mice in which we substituted the sequences of Loop 1 and/or Loop 2 of mouse ␣-MHC with the respective sequences of ␤-MHC and assessed the mechanical and enzymatic characteristics of the chimeric MHCs. These experiments were designed to test whether a sequence substitution in the Loop 1 and/or Loop 2 region is sufficient to confer ␤-like activity to the ␣-MHC molecule. Complete replacement of the endogenous ␣-MHC protein with the chimeric myosin resulted in surprisingly minor differences in enzyme kinetics indicating that, for these cardiac isoforms, other variable regions or residues must play a predominant role in determining overall ATPase activity and velocity of shortening.

EXPERIMENTAL PROCEDURES
Generation of TG Animals-TG mice expressing full-length mouse ␤-MHC were generated. ␤-MHC cDNA was produced using a combina-tion of a cDNA containing the 3Ј one-third of the RNA (19), and RT-PCR with ␤-MHC-specific primers (GenBank TM accession number AY056464). The product contained the entire ␤-MHC cDNA and was linked to the mouse ␣-MHC promoter (Fig. 1A). Multiple TG founders (FVB/N) were generated. Line 102 with the highest degree of MHC replacement (ϳ40%) was bred to homozygosity resulting in 73% replacement of ␣with ␤-MHC (Fig. 1D). Neither the heterozygous or homozygous TG animals showed an overt phenotype, and all animals had a normal life span. To shift isoform expression in non-transgenic (NTG) hearts from ␣to ␤-MHC, adult mice received an iodine-deficient diet containing 0.15% propylthiouracil (PTU) for 8 weeks.
Chimeric myosins in which either only the sequence of L 2 or of both Loop 1 and Loop 2 (L 1 ϩL 2 ) of ␣-MHC was exchanged for the respective sequence of ␤-MHC were subsequently constructed. A diagram of the TG constructs for the chimeric MHCs is shown in Fig. 5A. Constructs for L 2 , and L 1 ϩL 2 chimeric MHC were made using standard PCR methodology. Inner sets of oligonucleotides were designed so that the overlap encoded the amino acids that form Loop 1 or Loop 2 of the ␤-isoform (Fig. 5B). Fragments generated by PCR amplification were cloned back into full-length ␣-MHC and placed into the mouse ␣-MHC promoter cassette (20). Finally, the DNA was excised free of plasmid sequence and used to generate multiple TG founders (FVB/N).
ATPase Assays-MHC was purified from individual mouse hearts (21). F-actin was prepared from acetone powder of chicken pectoralis muscle according to a protocol modified from Pardee and Spudich (22). Actin-activated ATPase activity was measured at actin concentrations ranging from 10 to 80 M (21). Ca 2ϩ -stimulated Mg 2ϩ -ATPase activity of myofibrillar preparations from ventricular tissue (23) was determined as previously described (24).
Miscellaneous Methods-Transcript analysis was performed with RNA blots with transcript-specific probes as described previously (19). The two cardiac myosins were separated on denaturing gels in the presence of glycerol, essentially as described by Reiser and co-workers (25,26). Fiber isolation and their analyses have been described in detail (27). The in vitro motility assays (28) and determination of cardiac hemodynamics in the isolated working heart model and the closed chest intact mouse model were carried out as described (29,30).
Statistical Analysis-All data are expressed as mean Ϯ S.E. Comparisons between NTG and TG littermates were evaluated using Student's t test, and a p value of Ͻ0.05 was considered statistically significant.

Mediating an ␣ 3 ␤ MHC Switch via Transgenesis-To
understand the functional consequences of MHC isoform shifts in the heart we used transgenesis to effect an ␣ 3 ␤ shift. A cDNA encoding the mouse ␤-MHC cDNA was produced using a combination of a cDNA that encoded the 3Ј-terminal one-third of the RNA (19), and RT-PCR with ␤-MHC-specific primers. In our hands, for unknown reasons, the ␤-MHC cDNA full-length clone was difficult to isolate, and, upon bi-directional sequencing, numerous errors were apparent, particularly in the PCRderived portion of the molecule even though a "proof-reading" enzyme was used. Each error was repaired using site-directed mutagenesis. Multiple clones of the final construct were sequenced in order to exclude any remaining PCR-induced errors and, in comparison to both the rat and human clones, a consensus sequence for the murine ␤-MHC was derived (Gen-Bank TM accession number AY056464). ␣-MHC is expressed in both the mouse atria and ventricles in post-birth animals, and so the cDNA was placed into the ␣-MHC promoter cassette, which contains the full-length ␣-MHC promoter upstream of the single cloning site, and a human growth hormone polyadenylation signal (hGH) downstream (Fig. 1A).
In order to confirm that the transcript was the correct size, a Northern blot was made using RNAs derived from each of the lines ventricles, and hybridized against a myosin cDNA probe. All TG lines showed only a single RNA species, which comi-grated along with the endogenous MHC RNA (Fig. 1B). Subsequently, the levels of expression were quantitated using oligonucleotide probes hybridized to RNA dot blots. Probes specific to either the ␣-MHC 3Ј-UTR (which is contained within the transgene, Fig. 1A) or an oligonucleotide specific for hGH were used. Both probes gave consistent results with respect to the lines' relative expression to one another, while the ␣-MHC probe enabled us to determine the degree of overexpression. For each of the 5 lines tested, TG transcript levels were quite modest, ranging from 1.7 to 3.0-fold with respect to the endogenous message (Fig. 1C). These levels were relatively low compared with those that are sometimes observed when other contractile proteins were overexpressed using transgenesis (27,31,32). However, we have noted in other related studies that high copy numbers and high levels of MHC expression (Ͼ12fold overexpression) can be lethal, presumably because of the relatively insoluble nature of the intact protein.
Cardiac Myosin Heavy Chain Replacement-TG mice were analyzed for ventricular MHC protein content at 10 -12 weeks. The ␣and ␤-MHC proteins can be separated on 5% glycerol gels (Fig. 1D). Hypothyroidism resulting from PTU treatment resulted in nearly complete replacement (90%) of ␣with ␤-MHC. Our previous contractile protein-based TG studies showed that the cardiomyocyte rigidly controls the overall stoichiometry of the contractile protein pool, such that TG overexpression at the mRNA level does not lead to increases of overall protein content (30,32,33). That is, there is no "overexpression." The steady state levels of endogenous protein are down-regulated and replaced proportionally by the TG protein. Therefore, we could achieve partial or even complete replacement of the endogenous MHC with TG proteins. When the highest expressing line (Line 102) was bred to homozygosity, 73% of the total MHC was ␤-MHC (Fig. 1D).
We reasoned that if a phenotype were likely to be present, it would be most easily detected in line 102, and this line became the focus of our analyses. Immunohistochemical staining using a V 3 -specific antibody derived from the hypervariable Loop 2 region of ␤-MHC (Fig. 5B) showed only traces of this isoform in NTG ventricles ( Fig. 2A). Confocal analysis confirmed that the PTU-treated animals showed significant accumulation of ␤-MHC (Fig. 2B), but the characteristic striated pattern was somewhat blunted, consistent with the major effects that hypothyroidism has on cardiomyocyte morphology. In contrast with the PTU-treated mice, striated morphology was well conserved in the ␤-MHC TG cardiomyocytes, with the pattern of staining confirming the correct incorporation of TG protein into the sarcomere (Fig. 2C). Cardiac histology was examined using young adult animals (8 -12 weeks) (Fig. 2, D and E) and aged animals (1 and 2 years). 2 No significant differences in the gross morphology of either heterozygous or homozygous TG hearts were observed, and no differences in heart rate or chamber showing that the TG mRNA is the same size as the endogenous message (NTG). Line identities are shown above their respective lanes. The probe corresponded to the mRNA 3Ј-UTR. C, relative and absolute levels of overexpression at the transcript level were quantitated from a series of dot blots. Probes included oligonucleotides corresponding to either the hGH or 3Ј-UTR sequences. The gray boxes denote the amount of TG MHC expression relative to endogenous MHC mRNA levels. D, to induce a V 1 3 V 3 shift in NTG hearts, mice were treated with PTU. Ventricular protein was loaded onto a 5% glycerol gel and electrophoresed to separate the ␣and ␤-MHC-encoded isoforms. The PTU-treated hearts contained Ͼ90% V 3 . Line 102 heterozygotes, which have 40% replacement, were bred to homozygosity, and protein derived from those hearts (lane 3, ␤-TG) contained 73-75% V 3 .

FIG. 2. Incorporation of TG protein into the sarcomere and cardiomyocyte structure.
A-C, representative images of longitudinal cryosections from 4-month-old ventricles double-labeled with an antibody against mouse ␤-MHC Loop 2 (green) and fluorescence-tagged phalloidin (red). A, NTG; B, NTG treated with PTU; C, line 102 ␤-TG homozygotes. Note the lack of any ␤-MHC (green) in the NTG cardiomyocyte and the relative lack of striations in the PTU-treated animals when compared with either panels A or C. D and E, NTG and homozygous line 102 TG ventricles, respectively. Longitudinal sections stained with hematoxylin-eosin display regularly aligned myofibrils with distinct Z bands and M lines. The insets demonstrate well-defined striations in both sets of cardiomyocytes. F and G, thin sections from NTG and homozygous line 102 TG ventricles, respectively, prepared for transmission electron microscopy. No abnormalities could be detected by observers who were blinded to sample identity. weight could be detected ( Table I). Quantitation of the molecular markers of hypertrophy, which we have found to be a very sensitive marker of any response at the cellular level, was carried out at the transcript level as described previously (34), and no differences could be detected. 2 Similarly, we could not detect any obvious differences in the cardiomyocytes from homozygous TG mouse hearts using either light or electron microscopy (Fig. 2, D-G). No early deaths or overt ill health was noted in any of the TG animals during the first year and a half of life as compared with the NTG experimental cohorts. We conclude that the ␣ 3 ␤ transition is benign in terms of the animals overall cardiac morphology of the animals and that no early mortality or morbidity presents under normal animal husbandry conditions.
Consequences of Isoform Replacement in Intact Fibers-In light of the unremarkable phenotype at the whole animal level, we wished to confirm that changes in isoform content had affected the mechanical and kinetic properties of the skinned myofibers. Ventricular papillary muscles were isolated from line 102, PTU-treated and NTG mice. Line 102 heterozygotes have ϳ40% ␤-MHC while the homozygotes show ϳ73% replacement. The skinned fiber is a complex system in which the contractile machinery operates against the internal cytoskeletal structures in both the cardiomyocytes and connective tissue. Therefore, V max in a fiber is never truly unloaded, as is assumed to be the case in the in vitro actin motility assay (see below).
We first wished to compare the effects of ϳ40% replacement versus the fibers derived from PTU-treated animals, in which ϳ95% of the cardiac myosin consisted of ␤-MHC. Fibers were isolated from 9-week-old animals in order to minimize the effects of any secondary pathology that might develop later in life, and the unloaded shortening and maximum shortening velocities, as well as the relative power that the fibers developed, were measured (Fig. 3). As expected on the basis of the degree of ␣-MHC replaced by ␤-MHC, the values derived from line 102 heterozygotes were intermediate between the NTG (100% ␣-MHC) and PTU (90% ␤-MHC) data. Significant, graded decreases in the unloaded shortening velocity were noted (Fig. 3A) from NTG (3.80 Ϯ 0.14 m. l./s, n ϭ 7) to line 102 (2.72 Ϯ 0.26 m. l./s, n ϭ 4) to the PTU-derived fibers (1.51 Ϯ 0.24 m. l./s, n ϭ 3). The same gradual decreases were also observed in the force-velocity data used to derive the maximum shortening velocities (Fig. 3B). The power-force relationships, and maximum power produced followed the same trend (Fig.  3C), and the data show that the shift in MHC isoform content leads directly to changes in cross-bridge cycling rates.
We confirmed both the trend and stability of these changes by developing a cohort of heterozygotes and homozygotes over the course of a year and subsequently carrying out fiber measurements comparing these two populations to NTG fibers (Fig.   4). Similar graded decreases in the unloaded shortening velocity (Fig. 4A), maximum shortening velocity (Fig. 4B), and maximum power produced (Fig. 4C) were noted when the NTG, heterozygotes, and homozygotes were compared. No changes in the calcium-force relationship could be observed in any of the TG fibers. 2 Replacement of Cardiac Myosin with an ␣/␤ Chimera-The  above data clearly showed that the cardiomyocyte is tolerant of significant MHC isoform shifts that are transgenically imposed. We next explored the structural basis of the different cardiac myosins' unique functionalities by replacing the endogenous MHC with ␣/␤ chimeras, the working hypothesis being that the functional differences between the isoforms presumably are caused by the different loop sequences. The structure of Loop 1 is thought to modulate the rate of Mg 2ϩ -ATP binding and Mg 2ϩ -ADP release while the structure of Loop 2 affects the rate of myosin attachment to actin (9 -11). Two constructs, in which the sequences of either Loop 1 and Loop 2 (L 1 ϩL 2 ) or only Loop 2 (L 2 ) of mouse ␣-MHC were substituted by the corresponding ␤-MHC sequences were made and used to generate TG mice (Fig. 5A). In order to detect the TG protein, an antibody to the ␤-MHC Loop 2 sequence was generated (Fig.  5B). Quantification of protein replacement in hearts from L 1 ϩL 2 -and L 2 -TG mice by Western blotting (Fig. 5C) showed nearly complete replacement (L 1 ϩL 2 , 100%; L 2 , 84%). This was confirmed by mass spectroscopy, in which the tryptic peptide of the endogenous protein (LMATLFSTYASADTGDSGK, mono-isotopic mass 1935.90) was replaced by the respective fragment containing ␤-MHC-sequence (LLSNLFANYAGADAPADK, mono-isotopic mass 1850.93).
In Vitro Motility Assays-To determine the effects on motor function, in vitro actin motility assays were performed using MHC that had been isolated from heterozygous line 102 ␤-TG mice (40% replacement), from high-replacement (100%) L 1 ϩL 2 TG hearts, and from NTG as well as from PTU-treated hearts (Fig. 6). V 1 3 V 3 replacement had a clear effect on molecular motor velocity, with the isoform switches in the PTU and ␤-MHC TG preparations significantly decreasing the sliding velocity of myosin. For the 40% replacement of V 1 with V 3 in line 102 heterozygotes, the observed values were intermediate between the NTG and PTU-derived samples, as expected considering the results of our previous studies in which we compared filament sliding velocities of mouse V 1 /V 3 mixtures in varying proportions, and observed a linear relationship between relative isoform content and filament sliding velocity (6). In the present study, actin filament sliding velocity of the L 1 ϩL 2 chimeric MHC was not significantly different from that of ␣-MHC, indicating that the loops did not confer "␤-like" activity on the molecule.
ATPase Assays-The different myosins are characterized by their unique enzymatic activities. We determined both the myofibrillar Ca 2ϩ -stimulated Mg 2ϩ -ATPase and actin-activated ATPase activities of myosins purified from NTG, L 1 ϩL 2 -, L 2 -, and ␤-TG hearts. As expected, a 2-fold difference in myofibrillar Ca 2ϩ -stimulated Mg 2ϩ -ATPase activity was observed between NTG and PTU hearts (Fig. 7A). Myosin isolated from the ␤-MHC-expressing TG line 102 homozygote hearts exhibited depressed activities that were consistent with a 75% replacement of V 1 with V 3 (Fig. 7B). The chimeric myosins, while displaying somewhat diminished activities at the two highest calcium concentrations tested, had myofibrillar ATPase values

FIG. 4. Contractile properties of isolated ventricular fibers II.
A, slack test comparison between NTG, TG line 102 heterozygotes, and TG line 102 homozygotes. The change in length (⌬length) was plotted versus the time lag between the onset of a release and the onset of tension recovery (⌬time). Straight lines were then fitted by the leastsquared method. The maximum shortening velocities were also determined using the slack test. Units are in muscle lengths per s (m.l./s). B, force-velocity relationships and maximum shortening velocities determined by isotonic quick releases under constant load at pCa ϭ 5. C, relative power was extrapolated from the force-velocity relationships. Multiple fiber preparations were derived from 3 to 4 animals per group. Power output is defined as the relative power (P/P 0 ) multiplied by the velocity (m.l./s). Values are expressed as means Ϯ S.E. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001 versus NTG. Absolute values differ slightly from the data set in Fig. 3 because of minor alterations in the fiber apparatus. However, relative differences are conserved. that were closer to V 1 than V 3 (Fig. 7C). For the L 1 ϩL 2 -TG hearts in which a 100% replacement had been effected, enzymatic activity of this myosin was more like V 1 than V 3 , indicating that the loops did not confer full "␤-like" activity on the molecule. There were no significant differences in the pCa 50 values between any of the groups. The slightly different ATPase values between L 1 ϩL 2 and L 2 myofibrils is most likely due to the higher degree of protein replacement in the L 1 ϩL 2 -TG mouse hearts.
To explore this phenomenon further, data were obtained for the more physiologically relevant actin-activated Mg 2ϩ -ATPase activity (Fig. 8). While MHC isolated from line 102 homozygotes exhibited the expected decrease in enzymatic activity (Fig. 8A) neither L 1 ϩL 2 nor L 2 ␣/␤ chimeric MHCs showed any significant differences from the V 1 enzyme at any actin concentration (Fig. 8, B and C).
Cardiac Hemodynamics in Isolated Working Hearts and in Vivo-Although the V 1 3 V 3 TG animals appeared overtly healthy in the unstressed state and showed no signs of morbidity or increased mortality, we reasoned that whole organ function must be affected because of the slower motor velocity. To that end, we determined cardiac hemodynamics for the line 102 homozygotes using the isolated working heart preparation and in the intact animal (Table II). The unpaced isolated working hearts, with ϳ75% V 3 , displayed a significantly reduced heart rate with concomitant reductions in both systolic and diastolic parameters (Table II). We repeated the measurements under paced conditions (397 bpm) so that heart rate would not directly affect systolic or diastolic function and again noted significant decreases in the values of both dP/dt min and dP/dt max .
In vivo data from intact mice were also consistent with a decrease in systolic and diastolic function in ␤-TG mice. Mean arterial pressure, LV systolic pressure, and LV dP/dt max tended to be lower in TG animals compared with wild type (Table II). Furthermore, dP/dt 40 (dP/dt at 40 mm Hg developed pressure) was also significantly lower in TG mice, suggesting that the decreased rate of contraction could not be accounted for by the observed differences in afterload and is more likely due to actual differences in myocardial contractility. Likewise, both dP/dt min and the time constant of relaxation (tau, ) also dem- and C show the NTG and PTU data as dotted lines to facilitate comparison. A 2-fold difference in myofibrillar Ca 2ϩ -stimulated Mg 2ϩ -ATPase activity was observed between the NTG and PTU hearts. ATPase activity of L 1 ϩL 2 -TG myofibrils was reduced, but not to the level of the PTU group. ATPase activity of L 2 -TG myofibrils was also different from the PTU group, but not from the NTG group. In contrast, ␤-TG myofibrils showed significantly reduced ATPase activity close to that seen in the PTU group. *, p Ͻ 0.05 versus NTG; ‡, p Ͻ 0.05 versus PTU, n ϭ 3-5, myofibrils from 3-to 6-month-old hearts.

FIG. 8. In vitro actin-activated ATPase activities.
A-C, actinactivated ATPase activities of myosin preparations from NTG and ␤-TG (A), L 1 ϩL 2 (B), and L 2 (C) myosin. For comparison, data from panel A are shown again as dotted lines in panels B and C. ␤-MHC, but not L 1 ϩL 2 and L 2 chimeric myosin, showed significantly reduced actinactivated ATPase activity. Myosin was purified from 3-to 6-month-old hearts. *, p Ͻ 0.05 versus NTG. onstrate significant impairment of relaxation in the TG animals, consistent with data from the isolated hearts. Interestingly, in contrast to the isolated unpaced working heart, there was no difference in heart rate between the NTG and ␤-TG mice, presumably because of compensation via neurohumoral mechanisms in the intact animal. Supporting this hypothesis, we found that administration of the ␤-adrenergic blocker propranolol revealed a difference in heart rate between TG and wild type mice (434 Ϯ 29 versus 474 Ϯ 11). Finally it is important to note that out of the 6 mice in the ␤-TG group, 2 were hemodynamically unstable under anesthesia and died before completion of the protocol, further underscoring their cardiovascular deficit.

DISCUSSION
The aim of this study was to investigate the functional consequences of myosin isoform diversity and how the different motor abilities influence cardiac contractile function. In mammalian adult hearts, alterations in MHC isoform expression occur in response to various pleiotropic stimuli such as hypertrophy, failure, hypo-or hyperthyroidism. While these partial/ complete isoform switches correlate with changes in cardiac performance, there are a myriad of other structural and functional changes that accompany these processes (17,18), such that it is impossible to ascribe the changes in heart function to modifications in MHC isoform content alone. To test the hypothesis that changes in isoform content in the absence of such global processes could alter heart function, we used transgenesis to generate mice that overexpressed specifically in the heart the ␤-MHC-encoded isoform.
Cardiac-specific transgenesis also was effective in replacing the endogenous myosin with ␣/␤ chimeric myosins, which could then be used to investigate whether myosin functionality could be critically altered by exchanging loop sequences between the cardiac MHC isoforms. Based on chimeric studies in Dictyostelium myosin II, Spudich et al. (9 -11) postulated that the structures of the myosin surface Loops 1 and 2 could serve as modulators of the enzymatic and mechanical properties of the MHC. However, a number of other investigators reported evidence that the structure of the myosin backbone could profoundly influence surface loop function (12,13). Rat and pig ␤-MHC, which have identical Loop 1 sequences apart from a single conservative substitution, have 3-4-fold differences in ATPase activity and ADP dissociation (12). The present study addresses this controversy in full-length mammalian cardiac MHC. Our data demonstrate that the sequences of the surface loops of mouse cardiac MHC, in isolation, are of minor importance in determining isoform-specific characteristics. In light of the previous chimeric studies, the sequences of the ␣and ␤-loops in mouse cardiac MHC may be dependent upon other variable amino acids for their ability to influence both the kinetics and mechanics of cardiac muscle contraction.
Changes in charge and length of Loop 2 in Dictyostelium myosin can modulate MHC function (35). Neither mouse ␣and ␤-Loop 1 nor mouse ␣and ␤-Loop 2 differ in net charge (Fig.  5B), but charge distribution and spacing varies considerably between the two isoforms. These differences in the three-dimensional arrangement of charges may influence myosin-nucleotide and actin-myosin interactions (8). This concept is supported by data demonstrating that the structural differences between the rat ␣and ␤-loops are sufficient to alter Dictyostelium MHC function (11). Loop 2 of Dictyostelium myosin II was exchanged for Loop 2 of rat ␣or ␤-MHC, resulting in actin-activated ATPase activities of the two chimeras that closely reflected ATPase activities of rat ␣and ␤-MHC. Mouse and rat ␣-Loop 2 are identical, and mouse and rat ␤-Loop 2 differ only in one amino acid (mouse A631V rat), a conservative substitution. Our data indicate that Loop 2 exchanges between mouse ␣and ␤-MHC do not affect the physiologically relevant actin-activated Mg 2ϩ -ATPase activity: neither the L 1 ϩL 2 nor L 2 chimeras were different from that of ␣-MHC. Similarly, ATPase activity of myofibrils from L 2 -TG hearts was not altered. Only at non-physiologically high calcium concentrations was the ATPase activity of L 1 ϩL 2 myofibrils reduced, indicating that Loop 2 plays only a very minor role in regulating this aspect of isoform functionality. Moreover, the present chimeric study also demonstrates that switches in Loop 1 sequence do not alter ADP release rate and thereby actin filament sliding velocity in mouse cardiac MHC, in contrast to the previous Dictyostelium data (10). The data, together with other chimeric-based studies indicate that the structure of the MHC backbone influences whether or not loop exchanges between isoforms can affect MHC function.
These seemingly contradictory data can be reconciled by hypothesizing that the neighboring structures of the myosin backbone direct the flexible loops into certain conformations and thereby promote or ameliorate their influence on actinmyosin interactions or nucleotide binding and release. Alignment of the amino acid sequences of the backbones of Dictyostelium myosin II and mouse cardiac ␣-MHC reveals only 33% identity (64% homology). Consequently, it is not surprising that the Dictyostelium myosin II backbone could provide different atomic interactions with a chimeric loop than would the mouse cardiac MHC backbone. Furthermore, it is well known that within the myosin head even widely separated regions can critically influence one another, and this may underlie a partial explanation for the importance of the backbone TABLE II Hemodynamic measurements in the isolated working heart model and in vivo All results are given as mean Ϯ S.E. 4-month-old mice were used for working heart experiments; 3-month-old mice were used for in vivo studies. Line 102 homozygotes were used (␤-TG). The unpaced and paced models were carried out on slightly different apparati and absolute values between these two groups cannot be directly compared; however, NTG and TG cohorts within the unpaced or paced groups can be compared directly. ND, not determined; MAP, mean arterial pressure; LVP sys , left ventricular systolic peak pressure; dP/dt, change in pressure in relation to time; dP/dt 40  sequence for loop function. Therefore, alignment of more closely related MHC isoforms in order to find regions that could influence isoform-specific characteristics appears to be the most feasible approach for determining the critical amino acids that underlie the differing functionalities of the unique isoforms. By comparing MHC isoform sequences across various mammalian species in the context of all available functional data, we identified only 8 non-conservative amino acid substitutions in ␣-MHC (residues 2, 210, 442, 452, 801, 1092, 1637, and 1681) and only 4 residues in the ␤-MHC (residues 424, 573, 1201, and 1368) that may be responsible for species-specific MHC-isoform functionality rather than surface Loops 1 and 2 (6). Functional analysis of these residues may tell us which regions of the backbone are of importance for MHC function and would provide the basis for further structure-function studies comparing ␣-MHC and ␤-MHC. Functional differences between the cardiac MHC isoform were manifested at the single motor, biochemical, and fiber levels, in a manner that reflected the altered V 1 /V 3 ratios. Thus, it is clear that changes in myosin enzymatic activity are reflected by concomitant changes in motor velocity, which, in turn, lead to changes in fiber contractility. These changes are all consistent with the correlations that have been previously noted (36 -38). Comparing the data with those studies, the unloaded shortening velocity decreased 30% in TG mice exhibiting 40% replacement with ␤-MHC, by 56% in TG mice with 73% replacement, and by 60% in the PTU-treated animals (Figs. 3 and 4), a value in the same range as measured by Fitzsimons et al. (39), who found an 80% decrease in rat single cardiac myocytes that express essentially pure ␤-MHC. In our hands, substituting the majority of ␣-MHC with ␤-MHC either via transgenesis or by inducing hypothyroidism resulted in a similar impairment in fiber mechanics. This indicates that, in this setting, the MHC isoform switch has a predominant effect compared with other pleiotropic effects of PTU.
Our data also demonstrate how the in vitro mechanical and kinetic differences of the MHC isoforms are reflected at the whole organ and intact animal levels. Although the unstressed animal is overtly healthy, the 2-fold differences in actin filament sliding velocity and ATPase activity between ␣and ␤-MHC observed on the molecular level resulted in altered systolic and diastolic function in isolated hearts and in vivo. In unpaced isolated hearts, a 17% decrease in left ventricular pressure and a 31% decrease in dP/dt max together with a 45% increase in dP/dt min was seen, and when differences in heart rate were removed by atrial pacing, these differences persisted. Left ventricular pressure measurements from intact animals revealed similar contractile deficits in ␤-TG mice.
Recently, contractile function of TG mouse hearts with lowlevel expression of Myc-tagged rat ␤-MHC was measured in Langendorff preparations (16). In those hearts, dP/dt max was reduced by 15% although replacement of the endogenous MHC with the tagged ␤-MHC was only 12%. In light of our data, the relationship between contractility and relative isoform content may be non-linear and supporting this hypothesis, in hypothyroid hearts predominantly expressing ␤-MHC, a small amount of ␣-MHC expression can significantly augment myocyte power output (41).
The spontaneous heart rate of the isolated hearts, as well as heart rate in propranolol-treated mice, was reduced in ␤-TG mice as compared with NTG animals. A trend toward lower heart rates in ␤-TG isolated hearts was also reported by Tardiff et al. (16). These data support the concept that there might be an intrinsic feedback mechanism adapting heart rate to the kinetic properties of the MHC (42). In vivo, it is the autonomic nervous system that regulates sinus node firing, even in anes-thetized animals, and the system is regulated through both the sympathetic and parasympathetic neurons (43). Based on the normal histology, unremarkable sarcomeric structure of the cardiomyocytes and the lack of hypertrophy in the ␤-TG mice, we conclude that as long as the heart is able to adequately respond to autonomic regulation, and contractile function is sufficiently compensated to maintain cardiac output, no overt or subclinical phenotype will present.
While it has been generally accepted that isoform shifts can result in different functional endpoints, until very recently the potential of a V 1 3 V 3 transition to affect human heart disease remained problematic, as it was thought that the human ventricle contained only the V 3 isoform in either the normal or diseased state. However, evidence now exists that an isoform shift does occur in the failing human ventricle, with ␣-MHC mRNA accounting for as much as 34% of the total MHC transcript in the normal heart (44). Down-regulation of ␣-MHC in the failing human ventricle at both the RNA and protein levels occurs (45,46). Our data show that a V 1 3 V 3 shift in mouse hearts reduces contractile function. Taken together with the potential energy-conserving effect of a V 1 3 V 3 shift and the fact that even small amounts of V 1 might impact favorably on cardiac function, it is now critical to understand what role an isoform shift might play in disease onset and progression. Crossing the V 3 TG mice into different mouse models of hypertrophy and failure should provide insight into the potential role(s) the different cardiac MHCs may play in the pathogenesis of cardiac disease.