Myosin ATP turnover rate is a mechanism involved in thermogenesis in resting skeletal muscle fibers

Abstract

Thermogenesis by resting muscle varies with conditions and plays an active role in homeostasis of body weight. The low metabolic rate of living resting muscles requires that ATP turnover by myosin be inhibited relative to the purified protein in vitro. This inhibition has not been previously seen in in vitro systems. We used quantitative epifluorescence microscopy of fluorescent nucleotides to measure single nucleotide turnovers in relaxed, permeable skeletal muscle fibers. We observed two lifetimes for nucleotide release by myosin: a fast component with a lifetime of ≈20 s, similar to that of purified myosin, and a slower component with a lifetime of 230 ± 24 s. We define the latter component to be the “super relaxed state.” The fraction of myosins in the super relaxed state was decreased at lower temperatures, by substituting GTP for ATP or by increased levels of myosin phosphorylation. All of these conditions have also been shown to cause increased disorder in the structure of the thick filament. We propose a model in which the structure of the thick filament modulates the nucleotide turnover rates of myosin in relaxed fibers. Modulation of the relative populations of the super relaxed and conventional relaxed states could have a profound effect on muscle thermogenesis, with the capacity to also significantly alter whole-body metabolic rate.

The metabolic activity of resting skeletal muscle is of interest because it plays a significant role in the whole-body resting-energy expenditure (1). Muscle metabolism is involved in cold-induced thermogenesis, in consumption of calories from excess food intake, and is a major regulator of blood-sugar levels (1–5). The mechanism of muscle thermogenesis and its regulation remain an active area of investigation. Here we identify another mechanism that plays a role in thermogenesis of resting skeletal muscle, the modulation of myosin ATPase activity by the structure of the thick filament.

Our study also addresses a long-standing discrepancy concerning myosin kinetics. Ferenczi et al. (6) noted an inconsistency between the ATP turnover rate of purified frog myosin and the rate of oxygen consumption of living, resting frog muscle. The low metabolic activity of living, resting muscle (6, 7) sets an upper limit on the myosin ATPase activity, and this limit was less than one-fifth of the rate observed for myosin in vitro. A similar difference is seen in mammalian fibers (8, 9). The ATP turnover rate of rabbit myosin [0.16 s⁻¹ at in vivo temperatures (6, 9) would also have to be inhibited by more than a factor of 5 to be compatible with the resting-energy consumption of rabbit muscle [0.7 J × L⁻¹ s ×⁻¹ (8, 9)]. These observations show unambiguously that myosin in living, resting vertebrate muscle is inhibited by a large factor relative to purified myosin measured in vitro.

Although more than 30 years have passed since the original observation, the mechanism of myosin inhibition, unambiguously required to explain the energy output of living fibers, remains unknown. For all in vitro systems studied, including purified proteins, myofibrils, and skinned fibers, myosin has an ATPase activity similar to, or only a little slower than, purified myosin (9). The higher activities in skinned fibers or myofibrils arise in part because these preparations contain ATPases in addition to that of the relaxed myosin, membrane-associated pumps, damaged regions of the filaments that are not regulated, and so forth. The true myosin ATPase activity of relaxed permeable fibers must be measured using a method that can detect a slow turnover of nucleotides by myosin in the presence of a faster component on other proteins. The turnover lifetime of myosin measured here is the average time it takes a myosin head to hydrolyze and release a nucleotide after binding it. The turnover rate is 1/lifetime.

Results

Measuring Single Nucleotide Turnovers in Permeable Muscle Fibers.

The turnover rate of myosin was followed by incubating permeable rabbit psoas fibers in fluorescent nucleotides followed by a chase with ATP. In the permeable fiber preparation, the membrane has been removed chemically, allowing access to a relatively intact myofilament array. The fluorescent ATP analog, 2′(3′)-O-(N-methylanthraniloyl-ATP (mantATP), binds readily to myosin and activates ATPase activity to a similar extent as ATP (10, 11). Quantitative epifluorescence microscopy was employed to measure biochemical kinetics. The fluorescent images provided high signal-to-noise measurements of fluorescence intensity within small, defined areas [supporting information (SI) Fig. S1]. Single fibers were mounted on the scope in a cell that allowed solutions to be rapidly exchanged. In our standard experiment, fibers were first incubated in a relaxing solution containing 250-μM mantATP, and this solution was then exchanged for a relaxing solution containing 4-mM ATP. During this chase phase, the fiber remained relaxed and the fluorescence intensity decreased as the bound mant nucleotides were released and replaced by ATP.

The fluorescence intensity as a function of time following an initial incubation in mantATP and chase with ATP is shown in Fig. 1. There is a rapid decay in fluorescence intensity, with a lifetime of 19 s, and a slow decay in fluorescence intensity with a lifetime of 258 s. The fluorescence intensity as a function of time was fit by a double exponential decay, and the magnitudes and lifetimes of the two exponential components are given in Table 1. Quantitative interpretation of the rapid phase is complex, but its time course is compatible with: (i) the release of nucleotides by a fraction of normally relaxed myosin heads, which have a turnover time of 16 s at 24°C (9); (ii) release of nonspecifically bound nucleotides, presumably fast; and (iii) diffusion of released nucleotides out of the fiber occurring in ∼10 s (12). The lifetime for the second phase is much slower, 230 ± 24 s, and as argued below, it arises from the slow release of nucleotides by a second fraction of myosin heads with a very slow ATP turnover rate (see Table 1). Also shown in Fig. 1 is the rise in fluorescence intensity occurring in the inverse experiment in which the fibers were initially incubated in 4-mM ATP, and chased with 250-μM mantATP. The rise in fluorescence intensity mirrors the fall in intensity that occurred in the experiment described above. This observation shows that the mantATP turnover rate is similar to the ATP turnover rate and thus interactions, if any, of the protein with the mant moiety of the analog do not significantly alter the results. The experiments described above were done with rabbit psoas fibers, which are mainly fast twitch fibers. A similar decay in fluorescence was also obtained with rabbit soleus fibers, which are slow twitch fibers (see Table 1 and Fig. S2).

Fig. 1.

The changes in fluorescence intensity for single muscle fibers during the chase phase of the single nucleotide turnover experiments are shown as a function of time. The fluorescence intensity occurring during a chase with ATP following an incubation with mantATP is shown (○). The rise in fluorescence intensity occurring during the inverse experiment in which the fiber was first incubated with ATP followed by a chase with mantATP is shown (□). As can be seen, the fluorescence changes in two phases: a rapid phase which has a lifetime of 15 to 25 s and a slow phase with a lifetime of ∼230 s. The slower phase is attributed to the slow release of nucleotides from a fraction of myosin heads, which are in a super relaxed state. The fluorescence intensity is also shown after incubation with mantATP followed by a chase with 4 mM ADP (△). In ADP the fiber is not relaxed, all nucleotides are released in the first rapid exponential phase, and the exponential component with a slow lifetime is almost eliminated. The data were fit with a double exponential function. The fits are defined as follows: ATP chase P1 = 0.58 ± 0.01, T1 = 19 ± 1 s, P2 = 0.34 ± 0.01, T2 = 258 ± 12 s; mantATP chase, P1 = 0.58 ± 0.01, T1 = 16 ± 1 s, P2 = 0.33 ± 0.01, T2 = 216 ± 36 s; ADP chase, P1 = 0.94 ± 0.01, T1 = 12 ± 1 s, P2 = 0.05 ± 0.01, T2 = 175 ± 30 s. The ADP chase could also be fit with a single exponential, P1 = 0.97 ± 0.01, T1 = 13 ± 1 with approximately the same χ². Note: P1 plus P2 do not sum to 1, which shows that a small residue of nucleotides, usually <0.05, remain bound for even longer than the times measured; see discussion of Fig. S2. Myosin in the fibers was unphosphorylated; the temperature was 24°C.

Table 1.

Parameters of the two exponential fit to the fluorescence intensity during the chase phase

Incubationa	Chase	P1c	T1 (sec)	P2	T2 (sec)	nd
mantATP	ATP	0.65 ± 0.04	20 ± 2	0.31 ± 0.01	230 ± 24	24
ATP	mantATP	0.58 ± 0.05	19 ± 3	0.33 ± 0.02	246 ± 25	8
mantATPb	ATP	0.63 ± 0.02	17 ± 3	0.32 ± 0.02	156 ± 20	5
mantATP	ADP	0.96 ± 0.02	17 ± 5	0.02 ± 0.01	60 ± 240e	7
mantADP	ATP	0.91 ± 0.01	14 ± 6	0.06 ± 0.01	276 ± 180	5
mantATP +ATP	ATP	0.92 ± 0.01	11 ± 3	0.04 ± 0.01	222 ± 80	5

^aThe first two columns give the nucleotides in the initial incubation and the chase solutions.

^bThis row gives the averages for soleus fibers, all other rows were done with psoas fibers.

^cP1 and P2 are the magnitudes and T1 and T2 are the lifetimes for the first and second phases of the two exponential function that was fit to the data.

^dErrors are standard errors of the mean for n experiments.

^eFor some measurements, the second component was too small to be well defined, and the errors in the lifetime were large.

Slow Component of Fluorescence Decay Arises from Nucleotides Released by Myosin.

A number of experiments showed that the slow exponential phases seen in Fig. 1 arose from the slow turnover of nucleotides by a fraction of myosin heads. Competition with ATP in the initial incubation was used to determine what fraction of mant nucleotides bound in the fiber was specific to ATPase sites. Both the initial binding of mant nucleotides and the magnitude of the slow decay of fluorescence in the chase phase were reduced by addition of ATP to the initial solution, along with the 250-μM mantATP (Fig. 2). The data show that 41 ± 2% of the binding in the initial incubation is nonspecific. Inclusion of 5-mM ATP in the initial incubation also eliminated the slow decay of fluorescence in the chase phase, as expected if this phase arises from specific binding (see Table 1 and Fig. S3).

Fig. 2.

Determination of the fraction of mant nucleotides bound specifically to ATP binding sites in the fiber. A fiber was incubated in 250-μM mantATP and the fluorescence measured. ATP was then added to the concentrations shown and the fluorescence obtained was normalized to the initial fluorescence in the absence of added ATP. The data of this figure were fit to a simple competition model (solid line), which defined the fraction of mant nucleotides bound nonspecifically to the fiber as 0.41 ± 0.02, and the ratio of the apparent affinities of ATP to mantATP, as 0.57 ± 0.07. This ratio is consistent with previous observations showing tighter binding of mant nucleotides relative to their unlabeled counterparts (11). The temperature was 24°C; the myosin in the fibers was 7% phosphorylated.

The slow phase is only seen if the fiber is relaxed in both incubation and chase portions of the protocol. When ADP binds to the myosin head, it attaches strongly to actin and displaces the regulatory proteins, activating the fiber. This action results in further attachment of myosin heads to actin and rapid exchange of all nucleotides. In three different control experiments, in which mantADP replaced mantATP or ADP replaced ATP, in the incubation or chase solutions, virtually all nucleotides (>90%) were released in the rapid phase and the slow decay was absent or greatly reduced (see Fig. 1, Table 1, and Figs. S2 and S3). These decay traces also showed that diffusion of free nucleotides out of the fiber after they are released from binding sites is rapid. This result is expected as diffusion from the cell center to the periphery can be calculated to take place in ≈10 seconds (12). Thus, the slow phase required a relaxed fiber in all solutions, and myosin is the only nucleotide binding protein that is altered by relaxation.

Does the fluorescence intensity of specifically bound mantATP in the fiber correspond to that expected from binding to myosin? The fluorescence intensity of the fiber was quantified by comparison with 250-μM mantATP in a thin-walled quartz capillary (ID = 50 μm). The fluorescence intensity expected from mantATP bound to myosin was calculated using the known concentration of myosin heads (220 μM) and the increase in quantum yield upon binding to myosin (2.5) (10). The fluorescence intensity expected from full occupancy of myosin heads with mantATP is 0.96 ± 0.06 of the specific fraction of the fiber fluorescence. This equivalence indicates that most of the specific fluorescence arises from nucleotides bound to myosin.

Confocal microscopy was used to image the mant nucleotides bound to the fiber during the chase phase without interference from out of focus light. Fig. 3 shows that after 180 s in the chase, the mant-nucleotides were found predominantly in bright bands. At this point in the chase, the nonspecific fluorescence has been largely eliminated, as shown by the rapid decay of fluorescence in a chase after an incubation in which specific binding of mantATP is eliminated by competition with ATP (see Fig. S3). Thus, the remaining intensity seen here is because of the component of mant nucleotides that are being released slowly. The bright bands were identified as the A-bands by comparison with a bright field image and by measurement of their dimensions at different sarcomere lengths. The width of the bright bands remained constant as the sarcomere length was decreased from 2.9 μm, (bright band = 1.67 ± 0.05 μm) to 2.2 μm (bright band = 1.57 ± 0.04 μm). The width of the A-bands, which contain the myosin filaments, is known to be constant as a function of sarcomere length and equal to 1.6 μm. Thus, this image provides further and more direct evidence that the slow phase arises from mant nucleotides bound to myosin.

Fig. 3.

Confocal image of a muscle fiber that was incubated in mantATP followed by a 3-min chase with ATP, showing that mant nucleotides are bound to the A-bands, which contain the myosin filaments. This result provides direct evidence that during the chase phase the mant nucleotides are primarily bound to myosin. The fiber width is 35 μm; the distance between the bands (sarcomere length) is 2.2 μm.

Thus, we conclude that the slow decay of fluorescence is caused by the slow release of nucleotides from myosin heads in the relaxed fibers. Myosin provides the only known nucleotide binding site in the fibers that would be expected to react to the changes in conditions described above; it is the major protein of the A-band and the only single enzyme that has sufficient concentration to explain the magnitudes of the specific initial binding and of the slowly released component. Values of P2 reported here are given as a fraction of total fluorescence bound in the initial incubation, and 40% of this fluorescence is bound nonspecifically. Thus, assuming that all of the specifically bound mant nucleotides are bound to myosin, a value of P2 of 0.33 represents a larger fraction, 0.56, of the fluorescence arising from binding to myosin.

Factors that Affect the Population of Myosin Heads that Release Nucleotides Slowly.

In the Discussion we propose that our observed inhibition of myosin ATPase activity can be explained by the structure of the thick filament. To test this hypothesis, we explored conditions that affect the thick filament structure, including relaxation in different nucleotides, changing temperature, and phosphorylation of myosin. The results of these experiments are described below.

The experiment was repeated with different nucleotides replacing ATP. When the fiber was relaxed in 250-μM mantATP and chased with 4-mM GTP, the fluorescence intensity decayed rapidly. The data could be fit with a rapid phase with a lifetime similar to that seen in ATP and a slower phase, with a reduced amplitude (P2 = 0.12 ± 0.03) and shortened lifetime (T2 = 72 ± 18 s) (Fig. S4). If the fiber was relaxed in 4-mM GTP and chased with 250-μM mantATP, the fluorescence returned to its full value rapidly with a reduced slow exponential rise, which mirrored the decay in fluorescence described above. These observations show that although the fiber is mechanically relaxed in GTP [i.e., force and stiffness are low (13)], the slow release of nucleotides seen with ATP is greatly reduced in magnitude and lifetime. In contrast, when CTP replaced ATP in the chase phase, the slow exponential decay was similar to that seen in ATP in amplitude (P2 = 0.33 ± 0.01) and lifetime (T2 = 174 ± 12 s) (see Fig. S4). This observation shows that the slow exponential component is present when the fiber is relaxed in CTP.

The fluorescent decay was measured as a function of the temperature. The fraction of slowly exchanging nucleotides decreased as the temperature decreased, and the life time for release became shorter (Fig. 4). Measurement of the fluorescent intensity in the presence and absence of 5-mM ATP showed that the specificity of mantATP binding did not change with temperature; thus, changes in P2 arise from changes in the nucleotides bound to myosin. The value of T2 decreased with temperature from 258 ± 36 s at 30°C to 108 ± 14 s 12°C. The van ’t Hoff plot of P2 is linear, with an enthalpy of 57 ± 8 kJ/mole (see Fig. 4). Linear extrapolation of the plot to 39°C, the in vivo temperature of the rabbit, indicates that 75% of the myosin heads would be in the component that releases nucleotides slowly.

Fig. 4.

Decay of fluorescence in the chase phase for the highest and lowest temperatures measured (A) and the van 't Hoff plot for the equilibrium between super relaxed and relaxed states (B). The fluorescence decay during chase phases are shown for two different temperatures, 30°C (○) and 12° (□). The fibers were incubated in 250-μM mantATP followed by a chase with 4 mMATP. The fits to the data in (A) are defined as follows: 12°C, P1 = 0.80 ± 0.01, T1 = 26 ± 1 s, P2 = 0.16 ± 0.01, T2 = 138 ± 10 s; 30°C, P1 = 0.57 ± 0.02, T1 = 28 ± 2 s, P2 = 0.33 ± 0.02, T2 = 246 ± 35 s. (B) The van 't Hoff plot for the equilibrium between myosin heads that release mantATP slowly (super relaxed state) and those that release mantATP rapidly (relaxed state). The population of the super relaxed state is the fraction of specifically bound nucleotides that are released slowly (P2/0.59). The population of the relaxed state is 1-P2/0.59. The slope of the plot defines an enthalpy for the transition from the relaxed to the super relaxed states of 57 ± 8 kJ/mole. Temperature was controlled to within 1°C by a Peltier device. Myosin in the fibers was 5 to 7% phosphorylated.

The amplitude and lifetime of the slow exponential component were also altered by changing phosphorylation of the myosin regulatory light chain (LC2) (Fig. 5 and Fig. S5). Phosphorylation of the myosin regulatory light chain causes a decrease in both the amplitude and lifetime of the slow exponential decay. Experiments similar to that shown in Fig. 2 demonstrated that the changes in P2, discussed above, were not because of changes in specificity.

Fig. 5.

Effect of myosin phosphorylation on the fluorescence during the chase phase and on the amplitude of the slow exponential decay. (A) The effect of myosin phosphorylation on the fluorescence intensity in the chase phase: 7% phosphorylation (□), 85% phosphorylation (○). (B) The amplitude of the slow exponential decay during the chase phase is shown as a function of the level of phosphorylation of the myosin regulatory light chain. The temperature was 24°C. Error bars show the SEM for 10 to 20 different runs. The solid line in (B) is an interpolation between the points. Fits to the data in (A) are defined as follows: 7% phosphorylation, P1 = 0.64 ± 0.01, T1 = 14 ± 1 s, P2 = 0.28 ± 0.01, T2 = 230 ± 40 s; 85% phosphorylation, P1 = 0.73 ± 0.01, T1 = 11 ± 1 s, P2 = 0.22 ± 0.01, T2 = 125 ± 6 s.

Above, we showed that the slow release of nucleotides was eliminated when the fiber was not relaxed. Conditions that produce rigor alter two aspects of the fiber, a strong interaction with actin is allowed, and the structure of the thick filament is disrupted (14, 15). In the following section, we discuss the hypothesis that the structure of the thick filament plays a role in the slow release of nucleotides. To determine whether rigor conditions altered the slow release of nucleotides because of an interaction of myosin heads with actin, or by an altered structure of adjacent heads in the thick filament, we measured nucleotide release in an ADP chase at long sarcomere lengths. In an ADP chase at long sarcomere lengths (at which interaction with actin is not possible) there are two phases of fluorescent decay, a rapid phase (T2 = 20 s) and a slow decay (P2 = 0.35 ± 0.1), with a lifetime of 45 ± 12 s (Fig. S6). The magnitude of the second phase is much greater than that observed at full overlap, but the lifetime is shorter than that seen in the chase with ATP. Chase with ATP at long sarcomere lengths did not alter the slow release of nucleotides. This result shows that binding of ADP to some myosin heads increases the rate of release of mant ATP from adjacent heads. At full filament overlap the binding of myosin heads to actin accelerates this process so that virtually all myosin heads release nucleotides rapidly in the fast phase.

Discussion

Resolving a Discrepancy.

We propose a model of myosin heads in muscle fibers in which they occupy one of three states. In the active state they are interacting with actin and turning over nucleotides in less than 0.1 s. In resting fibers they can be in a state with a turnover time similar to that of purified myosin (≈15 s), or they can be in a second state with a much longer turnover time (230 s at 24°C). We term these two resting fiber states the “relaxed state” and the “super relaxed state,” respectively.

This model resolves the discrepancy described above between the ATPase activity of myosin and the metabolic rate of living fibers (6). When myosin is in the super relaxed state its ATP turnover time is 230 s, which is sufficiently slow to be compatible with the low metabolic activity of living fibers. If all myosin heads were in the super relaxed state, which is likely in living muscle under basal conditions, their ATPase activity would account for only 12% of the metabolic rate of resting living rabbit muscle = 0.7 J × L⁻¹ × s⁻¹ (8). It is also probable that in vivo the turnover time of myosin in the super relaxed state is even slower than observed here in an in vitro preparation.

A Structural Model to Explain how Myosin ATP Turnover Is Inhibited.

Recent studies of the structure of the thick filament in resting fibers provide a reasonable hypothesis to explain the slow release of nucleotide from myosin observed here in skeletal muscle fibers. The myosin heads in resting muscle fibers are found in an ordered helical array bound to the core of the thick filament. This structure has been determined at highest resolution for tarantula thick filaments (16). The two myosin heads each interact closely with one another in a structure first seen in relaxed smooth-muscle myosin (17). Additional interactions occur with the heads of adjacent myosin molecules along the helical track. The ordered helical array has been found in a variety of myosin filaments, including those from scallop and vertebrate cardiac muscle, suggesting that this structure is conserved across different species and muscle types (18, 19).

Of particular importance for our hypothesis is the observation that in scallop muscle the myosin heads are very inhibited when in the ordered helical array, taking 30 min for turnover and release of ATP (20). A similar slow release of nucleotides, lifetime >50 min, is seen in a folded conformation of monomeric smooth-muscle myosin, in which the two myosin heads have the same structure as that seen in the thick-filament array (21). Scallop muscle is regulated via calcium binding to the light chain; tarantula and smooth muscle fibers are regulated by phosphorylation of the myosin LC2. Although mammalian skeletal muscles are not regulated via the LC2, the observation of a very low ATP turnover rate associated with an ordered helical array in one type of muscle lends support to the possibility that the same thing occurs in mammalian skeletal muscle.

Several experiments suggested that the myosin heads with slow nucleotide release times are in the ordered helical array discussed above. Three factors have been shown to decrease the fraction of myosin heads in the ordered helical array: lower temperatures (14, 22) substitution of GTP or ADP for ATP (14)*, and myosin phosphorylation (23). When myosin is >80% phosphorylated, the value of P2 drops by 35% (see Fig. 5). This level of phosphorylation creates a high degree of disorder in EM images of myosin filaments of relaxed-rabbit thick filaments (23). The ordered array is undetectable by x-ray diffraction in the presence of GTP or ADP (14). When mantATP is chased with GTP, both P2 and T2 are considerably reduced (see Fig. S4). When mantATP is chased with ADP at long sarcomere lengths, where actin binding is removed from the picture, the amplitude of P2 remains high, but T2 is reduced to 45 s. Both GTP and ADP produce an open state of myosin that is thought to be incompatible with the structure of myosin in the array (14). In contrast, CTP supports formation of the ordered array (14) and also supports the formation of the super relaxed state (see Fig. S4). Thus, in each of the conditions discussed above, factors that reduce helical order in diffraction experiments also greatly reduce P2, T2, or both.

The amplitude of the fraction of slowly released nucleotides, P2, decreases and the lifetime is shorter for lower temperatures (see Fig. 4). The temperature dependence of P2 is qualitatively but not quantitatively similar to that seen by x-ray diffraction. In particular, the x-ray diffraction indicates a fraction of ordered myosin heads that is greater than the population of super relaxed heads found here (14, 15, 24). The enthalpy for the transition to the ordered state measured by diffraction, 96 kJ/mole (14), is considerably greater than that measured here for the transition to the ordered state, 57 kJ/mole. The quantitative differences between the two results could be interpreted as an indication that the heads in the super relaxed state are a subset of the heads that are structurally ordered in the diffraction experiments. Levine et al. observed two different populations of ordered heads with different degrees of order in cardiac thick filaments, supporting this hypothesis (25). On the other hand, there are a number of experimental uncertainties in both techniques that could also account for the difference.

Together, the above correlations, along with the observation of a slow turnover rate in other muscles, support a parsimonious model in which the ATP turnover rate of myosin in resting skeletal muscle is inhibited when myosin heads are bound to the thick filament core. It is possible that multiple ordered thick-filament structures exist, with only a subset of these in a super relaxed state; however, more work is required to address this possibility.

In the three-dimensional images of the thick filament, the two myosin heads adopt very different structures (16, 18, 26). Are both heads in the super relaxed state or is only one? In our experiments, the highest fraction of myosin heads in the super relaxed state is only ∼55%, leaving open the possibility that one of the two heads is not in the super relaxed state. However, if 50% of the myosin heads are in the normal relaxed state, the rate of thermogenesis of living resting muscle would be much higher than observed. Thus, as discussed below, we believe that a very high fraction of myosin heads are in the super relaxed state in living muscle.

How do myosin heads in the super relaxed states become active with the rapidity seen during skeletal muscle activation? Our observation that at long sarcomere lengths the binding of ADP to some myosin heads greatly accelerates the release of nucleotides from adjacent myosin heads in the super relaxed state shows that interactions occur between myosin heads in the filament, as is also suggested by structural studies (16, 18, 26). This cooperativity could lead to rapid activation of super-relaxed myosin heads. In an active fiber the interaction of some fraction of myosin heads with actin could be communicated to adjacent super relaxed heads, inducing them to leave the super relaxed state and also interact with actin. It is reasonable to assume that at all times there is a population of heads in the relaxed state, and that these heads can sense the activation of the thin filament.

Role of Muscle Thermogenesis in Whole-Body Metabolism.

The resting-energy expenditure can be roughly divided into two components, an obligatory thermogenesis that involves basic cell functions, and an adaptive thermogenesis that changes with conditions (27, 28). Although muscle has a low metabolic rate per unit volume, it contributes about 25% of the obligatory thermogenesis due its large mass (1). Thermogenesis by muscle also plays a large role in adaptive thermogenesis. Thermogenesis is increased by cold exposure and by consumption of excess calories, mediated at least in part by epinephrine (2, 3, 5, 28). Muscle thermogenesis is also activated by infusion of leptin, a signal released by fat cells to burn off excess calories (4). Muscle may be an optimal tissue for adaptive thermogenesis because of its low basal rate and its unique reserve metabolic capacity.

Our results suggest that changes in the relative populations of the relaxed and super relaxed states have the potential to produce significant changes in muscle thermogenesis. The low baseline rate of thermogenesis, (0.7 J × L⁻¹ × s⁻¹ in the rabbit, 0.5 J × L⁻¹ × s⁻¹ in humans) demands that virtually all myosin heads are in the super relaxed state. The transition of only 20% of these myosin heads into the relaxed state, with a turnover time of 6 s, would increase the rate of muscle thermogenesis by a factor of 2. This is approximately equal to the typical change in muscle thermogenesis seen to occur upon overfeeding or cold exposure in humans and rodents (28). If all of the heads were in the relaxed state, muscle thermogenesis would increase by a factor of 4 to 5, increasing whole-body metabolism by ≈80%. Thus, modulation of the population of the super-relaxed myosin heads would produce sufficient thermogenesis to significantly influence whole-body metabolic rates. Although our results do not prove that this mechanism occurs, they show that the potential for thermogenesis is great, and we identify one pathway that would modulate it.

Connection Between Thermogenesis and Muscle Mechanics.

It is reasonable to assume that the transition from the relaxed state to the active state would be more rapid than that from the super relaxed state. This appears to be the case for fast twitch muscle, where myosin phosphorylation, epinephrine, and cold exposure all lead to potentiation of twitch tensions relative to tetanic tensions (29–31). The correlation between thermogenesis and mechanics provides further evidence that the contractile proteins are involved in thermogenesis. This connection is a natural aspect of the model proposed here. The relative populations of the relaxed state and super relaxed state influence twitch potentiation at low levels of calcium, and thermogenesis at even lower levels of calcium. Thus, the use of myosin to burn off excess calories has an advantage over other proposed mechanisms of thermogenesis; it also makes the fiber more easily activated. However the same interventions, mentioned above, do not cause potentiation of twitches in slow twitch fibers (29–31). The super relaxed state was also found in slow twitch fibers (see Table 1 and Fig. S2). A possible explanation is that the slower time course of activation of twitch tension in slow twitch fibers may not be as sensitive to the speed of the transitions between myosin states.

Summary

We show that skeletal muscle myosin has an alter ego: it works as a motor protein in active muscle, and in relaxed muscle it plays a role in determining muscle thermogenesis. Our results resolve a 30-year-old discrepancy between the activity of myosin observed in vitro and in vivo. We identify one pathway that can modulate thermogenesis by myosin, and we feel that there are multiple pathways to be discovered. Modulation over the full possible range would produce a profound effect on muscle thermogenesis and, thus, on resting energy expenditure. We have provided experimental evidence supporting a very plausible structural model to explain this modulation. Our results bring together two previously unrelated fields, the study of myosin structure and function and the field of whole-body metabolism. This synthesis has the potential to provide theraputic targets to develop previously untried approaches to treat important problems in human health (28). An increase in muscle thermogenesis would aid in the regulation of body weight and in controlling high blood-sugar levels. On the other hand, decreasing muscle thermogenesis might also help treat patients suffering from insufficient metabolic capacity, such as occurs in heart failure.

Materials and Methods

Rabbit psoas and soleus fibers were harvested and chemically skinned as described previously (32). Animal care and killing protocols were followed according to University of California at San Francisco institutional guidelines. Phosphorylated fibers (i.e., fibers with > 80% phosphorylated LC2s) were produced by addition of phosphatase inhibitors, 20 mM NaF and 20 mM phosphate, to the basic skinning and storage solutions along with 5-mM ATP, as described previously (see Fig. S5) (32). The control fibers prepared without the phosphatase inhibitors and ATP in the skinning solution were typically 5 to 10% phosphorylated. Incubation of these fibers in the relaxing solution with 4-mM ATP for 60 min at 24°C reduced phosphorylation levels to zero (see Fig. S5). The level of phosphorylation was assayed using isoelectric focusing gels stained with Pro-Q (Molecular Probes) and visualized using a Typhoon 9400 Variable Mode Imager (Amersham Biosciences).

Single muscle fibers were mounted on a coverslip and the ends were secured with small amounts of grease. A simple flow cell was constructed by mounting the coverslip with the fiber on a glass slide between two other coverslips forming a small cell, which had a thickness of ∼270 μm. In the standard experiment, the fibers were first incubated for 2 min in a solution containing 250-μM mantATP. The results did not depend on the duration of the incubation, from 2 to 10 min. The solution was then exchanged for one containing 4-mM ATP. Fluorescence images were acquired on a Nikon TE2000 microscope with a Coolsnap HQ2 camera, using a DAPI filter set (Chroma #89000) and Sutter DG4 arc lamp for excitation. Images were obtained with a 200-ms exposure time. The average fluorescence intensities within a rectangular region, 10 to 20 μm × ≈100 μm, were determined using the Nikon software NIS-Elements. Four rectangular regions were defined: two placed over the center of the fiber and one adjacent to the fiber on either side, used to measure the background (see Fig. S1). Fiber fluorescence was determined by subtracting the background intensity from the fiber intensity. Two basic temporal sequences were used. In one, data were obtained every 15 s. In the second, data were obtained every 5 s for the first 2 min to better define the early phase, and for every 20 s thereafter. Fluorescence intensities were normalized by the fiber intensity before the chase in the ATP chase experiments. In the mantATP chase experiments, the fiber was washed for 1 min in rigor following the chase to remove all nucleotides and was then reincubated in mantATP. The intensities subsequently obtained were used to normalize the data obtained during the chase. The data were fit to a two exponential decay (or rise) using a nonlinear least-squares algorithm in KaleidaGraph v3.6 (Synergy Software). The fit also defined the 95% confidence limits provided in the figure legends. Photo bleaching during the experiment was determined to be ≈0.5%, and was thus negligible. Most experiments were performed at room temperature, which was stabilized at 24 ± 1°C. In some experiments, temperature was varied using a Peltier device. To determine the specificity of nucleotide binding in the fibers, the fiber and background intensities were obtained as a function of temperature in the presence of 250-μM mantATP, and the presence and absence of 4-mM ATP. The fraction of fluorescence intensity that was specific for ATP binding sites did not change with temperature.

A confocal microscope was used to visualize mant nucleotides bound to muscle fibers. A confocal microscope was used because the striations resolved in the images obtained with the Nikon TE2000 were obscured by out of focus light because of the high numerical aperture of the objective. Confocal images were acquired with a Nikon C1si confocal mounted on an FN-1 microscope. A 405-nm laser was used for illumination and a 450/35-nm emission filter was used for detection. Images were acquired with a 60/1.4 NA oil immersion objective.

The basic buffer contained 120-mM KAc (potassium acetate), 5-mM K-phosphate, 5-mM magnesium acetate, 4-mM EGTA, and 50-mM 3-(N-morpholino)propanesulfonic acid (Mops), pH 6.8 and either 250 μM mantATP, or 4 mM of another nucleotide.