Proc. Nat. Acad. Sci. USA Vol. 72, No. 5, pp. 1729-1733, May 1975

Motion of Subfragment-1 in Myosin and Its Supramolecular Complexes: Saturation Transfer Electron Paramagnetic Resonance (F-actin/myosin filaments/spin label/nonlinear spin response)

DAVID D. THOMAS*, JOHN C. SEIDELt, JAMES S. HYDEt, AND JOHN GERGELYt * Department of Biophysics, Stanford University, Stanford, California 94305; t Department of Muscle Research, Boston Biomedical

Research Inst., 20 Staniford St., Boston, Massachusetts 02114; and t Varian Associates, Instrument Division, Palo Alto, California 94303

Communicated by Harden M. McConnell, February, 3, 1975 Molecular dynamics in spin-labeled musABSTRACT cle proteins was studied with a recently developed electron paramagnetic resonance (EPR) technique, saturation transfer spectroscopy, which is uniquely sensitive to rotational motion in the range of 10-7-10- sec. Rotational correlation times (r2) were determined for a spin label analog of iodoacetamide bound to the subfragment-I (S-i) region of myosin under a variety of conditions likely to shed light on the molecular mechanism of muscle contraction. Results show that (a) the spin labels are rigidly bound to the isolated S-i (i-2 = 2 X 10-7 sec) and can be used to estimate values of r2 for the S-i region of the myosin molecule; (b) in solutions of intact myosin, S-i has considerable mobility relative to the rest of the myosin molecule, the value of r2 for the S-i segment of myosin being less than twice that for isolated S-i, while the molecular weights differ by a factor of 4 to 5; (c) in myosin filaments, T2 increases by a factor of only about 10, suggesting motion of the S-i regions independent of the backbone of the myosin filament, but slower than that in a single molecule; (d) addition of F-actin to solutions of myosin or S-i increases T2 by a factor of nearly 103, indicating strong immobilization of S-i upon binding to actin. Saturation transfer spectroscopy promises to provide an extremely useful tool for the study of the motions of the crossbridges and thin filaments in reconstituted systems and in glycerinated muscle fibers.

Current models of muscle contraction involve motion of the crossbridges between thick and thin filaments (1, 2). These crossbridges correspond to the subfragment-1 (S-1) and subfragment-2 (S-2) regions of myosin (3). Evidence for segmental flexibility within the myosin molecule, characterized by rotational correlation times (T2) on the order of 10-i sec, has been obtained by observing the rate of depolarization of fluorescent probes attached to the S-1 region of myosin in solution (4). Studies on the fluorescence polarization of tryptophyl residues in glycerinated muscle fibers suggest that the orientation of crossbridges may depend on the functional state, namely, contraction, relaxation, or rigor (5). However, there are no reliable measurements on the rates of S-1 motion in thick filaments or in actomyosin complexes. Some of the physiologically important motions of S-1, and of the other protein components of the contractile and regulatory machinery, are likely to have correlation times much longer than those observed in solution. Previous electron paramagnetic resonance (EPR) studies on spin-labeled muscle proteins (6-9) have been useful in detecting changes in mobility that could be attributed to conformational changes

near the label. However, the conventional EPR methods used in these earlier studies, like fluorescence depolarization methods, were limited to correlation times on the order of 10-7 sec and shorter. Quasielastic light scattering has been used to investigate correlation times longer than 10-s sec in F-actin (10), but few physicochemical methods are suitable for measurements in the range 10-7 < T2 < 10-s sec. New EPR techniques have been developed whose maximum sensitivity to rotational motion is in this time range (11-15). These techniques are referred to as saturation transfer spectroscopy because they depend on the diffusion of saturation through the spectrum of a nitroxide spin label. This spectral diffusion arises from rotational diffusion that modulates the anisotropic magnetic interactions. These techniques employ partially saturating microwave radiation and hence probe the nonlinear spin response. They are optimally sensitive to the rotational motion of the spin label if T2 is comparable to the spin-lattice relaxation time, (T1), which is about 10-5 sec for slowly tumbling nitroxides (16). One of these techniques has proven applicable to the study of solutions of spin-labeled proteins: absorption EPR, detected 900 out-of-phase with respect to the unsaturated second harmonic EPR response to the modulation field, Hm cos Wmt (12, 13, 15). If T2 1wm A ' T71 -- 10-5 sec, and the spin system is partially saturated, the competition between spectral diffusion and field modulation, in governing the passage of spins through the resonance condition, gives rise to a significant signal component that lags behind the modulation by 900 and is sensitive in shape to rotational diffusion. In the present study we have used the saturation transfer spectra of a spin label rigidly attached to the S-1 region of myosin to study the motion of this segment in intact myosin, in enzymatically cleaved fragments of myosin, in synthetic thick filaments, and in complexes with actin. MATERIALS AND METHODS

Preparation of Spin-Labeled Proteins. The preparation of myosin and spin-labeling of sulfhydryl-1 (-SHI) groups with N - (1 -oxyl - 2,2,6,6-tetramethyl -4- piperidinyl)iodoacetamide (IASL) was carried out as described previously (17, 18). To minimize the amount of weakly immobilized label (r2 < 10-9 sec) bound to myosin, 1 mole of the label was used per mole of myosin. In labeling sulfhydryl-2 (-SH2) groups, the more reactive groups were blocked with N-ethylmaleimide and the -SH2 groups were then labeled with N-(1-oxyl-2,2,6,6-tetramethvyl4-piperidinyl)-maleimide (MSL) (19). Heavy meromyosin (HMM) and S-1 labeled at the -SH, groups were prepared by digesting IASL myosin with trypsin (17). The labeled myosin was dialyzed exhaustively against a solution

Abbreviations: EPR, electron paramagnetic resonance; S-1, subfragment-1; -SHI, sulfhydryl-1; -SH2, sulfhydryl-2; HAIM, heavy meromyosin; Hb, hemoglobin; MISL, maleimide spin label; IASL, iodoacetamide spin label. 1729

1730

Biophysics: Thomas et al.

Proc. Nat. Acad. Sci. USA 72

(1975)

TABLE 1. T2 determination for IASL myosin and fragments in solution

V2' experimentst

vi experiments*

Myosin HMM S-1

AS(G) 0.68 it 0.10 0.795 i 0.10 1.01 4± 0.10

107 (see) T2/T2(MyO) 2.53 i 0.35 1.0 2.15 ±L 0.30 0.85 ± 0.20 1.63 i 0.20 0.64 i 0.15

T2 X

C'/C

Tg X 107 (see) r2/T2(MyO) -0.67 i= 0.05 3.7 i 0.6 1.0

-0.71 i1 0.04 3.2 d 0.4 0.86 4 0.25 -0.81 ±4 0.04 2.2 i 0.4 0.59 i 0.25

Averagest Tg X

107

(see) 3.0 2.6 1.85

T2/'Tr

(Myo) 1.0

0.85 0.62

r2/r2(Myo) is the ratio of T2 to that obtained for myosin by the same method. Uncertainties in AS and C'/C are standard deviations. Data from four preparations were combined and extrapolated to infinite dilution. * AS is the decrease in the separation of the outer hyperfine extrema, as described in text. T2 was determined from a plot of AS versus Tg (24, see text). t C is the height of the central peak of v2' spectra above the baseline, and -C' is the depth of this peak below the baseline. T2 was determined from a plot of C'/C versus r2, obtained from MSL Hb reference spectra. -r2 values are averages of columns 2 and 5, weighted according to uncertainties. Ratio values (last column) are averages of columns 3 and 6.

containing 0.5 M KCl and 0.05 M Tris HCl (pH 7.5). HMM and S-1 were dialyzed against solutions containing 0.1 M KCl and 0.05 M TriseHCl (pH 7.5). Myosin filaments were prepared by dialysis of IASL myosin in 0.5 M KCl and 1 mM Tes (N-tris[hydroxymethyl]methyl-2-aminomethane sulfonic acid) (pH 7.0) against 100 volumes of a solution containing 0.137 M KCI and 0.02 M Triss HCl (pH 8.3). Actin was prepared from an acetone powder by extraction at 40, followed by two polymerization-depolymerization cycles (20). Polymerization was induced by adding 0.1 M KCl and 0.6 mM MgCl2. F-actin was spin-labeled with MSL essentially as described previously (21). Human oxyhemoglobin was labeled with MSL (22) in 0.10 M potassium phosphate buffer, pH 7.0, and concentrated to 250 mg/ml by pressure dialysis. Glycerol (Eastman, "spectro" grade), buffer, and the concentrated hemoglobin solution were then combined to produce samples containing 25 mg/ml of protein in glycerol concentrations from 0% to 90% (v/v). These samples are referred to as maleimide-spinlabeled hemoglobin (MSL Hb). Similarly, solutions of IASL S-1 (20 mg/ml) in various glycerol concentrations were prepared. Viscosities were measured for the glycerol-water solvents at 200, 50, and - 12°, using Cannon-Fenske calibrated capillary viscometers. EPR Experiments. In this work, a Varian E-9 X-band spectrometer was employed in the absorption mode. For conventional EPR, 100 kHz field modulation was used. The resulting spectra are designated vi. A straightforward modification (12, 13, 15) permitted modulation of the magnetic field at 50 kHz with phase-sensitive detection at 100 kHz. In this configuration, normal 2nd derivative EPR spectra, which we designate v2, are obtained when the reference phase is set "in-phase." Saturation transfer spectra were recorded "900 out-of-phase" and are designated v2'. The out-of-phase setting was obtained by adjusting the reference phase to obtain a minimum signal (typically 1% of the maximum 2nd derivative signal) with the incident microwave power well below saturation (typically 1 mW). Saturation transfer spectra were recorded at 63 mW incident power, corresponding to a microwave field amplitude of 0.25 G, as determined by the method of perturbing metal spheres (23). The modulation amplitude was 5.0 G. Samples were contained in a standard Varian quartz flat cell, and the temperature of the entire E-231 cavity structure was controlled to within 1° by circulation of cooled nitrogen gas through it.

The conventional EPR method of McCalley et al. (24) was used to estimate correlation times on the order of 10-7 sec. Our vl experiments were performed with low modulation amplitude (2.0 G) and subsaturating microwave power (typically 10 mW). For each sample, at least 10 determinations were made of S, the separation in gauss between the outer extrema of the spectrum. Since the spin label was very strongly immobilized by addition of F-actin to IASL myosin (see below) the value of S for acto-IASL-myosin was used as the rigid limit value So. A plot of AS versus r2 for isotropic Brownian rotation (24) was then used to determine x2, where AS = So -S and 2 is the correlation time for Brownian reorientation of the nitroxide principal axis. Reference Spectra. Effective correlation times were determined by comparison of v2' spectra with collections of reference spectra. One such collection was a set of computer-simulated spectra calculated with modified Bloch equations (13). The only parameter varied was x2, the correlation time for reorientation of the molecule-fixed principal axis of the hyperfine and g tensors of a nitroxide spin label undergoing isotropic Brownian rotational diffusion. To simplify calculation, these tensors were assumed to have axial symmetry. When possible (i.e., for 72 > 10-6 see), 72 has been determined by matching the wings of a v2' spectrum with those of reference spectra, resulting in a 72 for the reorientation of the principal axis of the spin label, essentially independent of any deviation from axial symmetry or from isotropic rotation (25). In addition to the calculated reference spectra, empirical reference spectra were obtained from spin-labeled Hb and S-1 in aqueous glycerol solutions. Dilution of Hb below 25 mg/ ml and of S-1 below 20 mg/ml resulted in no significant change in vl or v2' spectra. Therefore, the solutions are dilute enough that T2 is determined by the solvent viscosity, and we have estimated T2values for Hb from the expression (26)

[1]

3kT Hb is assumed to be a sphere of effective radius R = 29 i (12) undergoing isotropic Brownian rotational diffusion. v is the measured solvent viscosity. Since S-1 is believed to have approximately the shape and hydrodynamic properties of a prolate ellipsoid with an axial ratio of 3 to 5 (1, 4), Eq. 1 cannot be used directly to estimate 72 for the protein. The long axis of a rigid ellipsoid with an axial ratio of 4 should change its orientation with a 72 that is

Proc. Nat. Acad. Sci. USA 72

IASL S-

MSL Hb

90 -12

90 5

S-1 Motion in Myosin: Saturation Transfer EPR

(1975)

2.3

XA 10

3,-12 80

3s16

6.0 x 5 80 lo-5

_ 0.9

90 20

T2 T % 60

r2 =2.3 x10-4 ACTOMYOSIN sec

x 10-4 r2=2 sec

1.9 x 20 80 l0-5

10 5

3.1

L

! V

80 20

I0,6

LASL

MSL Hb 90% GLYC -12;0

1731

.,

2.7 x 20 60 0-6

I 4.4 60 20

X

1.3 40 20

-

V

io-7

2.7 0

20

x8

I0-

x37-

4

v

.-/\.

N.

4.8

x 20 30

lo-7

1.6 x 20 0 lo-7

I

-

FIG. 1. Representative V2' reference spectra obtained from 25 mg/ml of MSL Hb (left) and 20 mg/ml of IASL S-1 (right), in appropriate buffers (see text). Spectra were recorded and solvent viscosities were measured at glycerol concentration (%) and temperatures (T, in 0C) shown. Values of r2 were estimated for the rotational diffusion of rigid bodies shaped like hemoglobin and S-1 as described in the text, and spectra are arranged so that these r2 values for adjacent Hb and S-1 samples are similar. We have compiled a collection of 40 MSL Hb spectra, available on request.

3.4 times longer than that of a sphere of the same volume (27). We, therefore, obtained estimates of the 72 for reorientation of the long axis of isolated S-1 by using Eq. 1 to calculate the 72 for a spherical protein of the same molecular weight (115,000, ref. 3) and multiplying this 72 by 3.4. The spin label concentration, determined by double integration of v1 spectra, was 0.49 mM for the Hb solutions and 0.10 mM for the S-1 solutions. High quality v2' spectra were obtained in 2-4 min for Hb and in 30 min for S-1. We consider the Hb spectra more reliable and have weighted them more than the other reference spectra in the determination of 72. The conclusions of this study are essentially the same regardless of which set of reference spectra is emphasized. RESULTS Myosin and fragments in solution Conventional EPR Experiments. The left side of Table 1 shows the 72 values obtained in v1 experiments. The values given were obtained from a series of protein concentrations by extrapolation to infinite dilution. Within the 10-20% statistical uncertainties, these values were unchanged by dilution beyond 20 mg/ml. Saturation Transfer Experiments. S-1 and Hb reference spectra in Fig. 1 were recorded under conditions such that

FIG. 2. (Right) V2' spectra from IASL myosin (20 mg/ml), IASL myosin filaments (20 mg/ml), and acto-IASL myosin (1 ml of the 20 mg/ml IASL myosin solution plus 1 ml of a 5 mg/ml solution of F-actin), in appropriate buffers (see text), at 200, all from the same preparation of IASL myosin. 72 values are from comparison with reference spectra, averaged for four myosin and actin preparations. The spin label concentrations were (top to bottom) 20 1M, 40 sM, and 40 /AM, and each spectrum was obtained in a 30 min scan. (Left) MSL Hb reference spectra which are the closest lineshape matches to the spectra on the right. 72 values were calculated with Eq. 1. Spectra were recorded and viscosities were measured at glycerol concentrations and temperatures shown. Ticks are at 20 G intervals.

adjacent S-1 and Hb spectra correspond to similar T2 values for the protein, estimated from molecular size and shape, as described above. Compared on this basis, the Hb and S-1 lineshapes agree fairly well with each other and with calculated lineshapes (13), particularly in the wings of the spectra, with small discrepancies between calculated and experimental spectra for T2 > 10-4 sec. This agreement indicates that each label is undergoing rotational diffusion at a rate predicted for the host protein, as indicated for MSL Hb in other studies (12, 15, 22,24). The center of Table 1 shows the v2' results, extrapolated to infinite dilution, for solutions of IASL myosin and the two fragments. The averages of the results obtained by the two methods are shown at the right of Table 1. For one preparation, S-1 and HMM were also studied in the presence of 0.5 M KCl, resulting in no significant differences from the results in Table 1. These values of 72 reflect primarily the rate of reorientation of the principal axis of the spin label; knowledge of the orientation and motion of this axis with respect to S-1 is important for extracting information about S-1 rotation from the spin label spectrum. The 72 found for the label on the isolated S-1, 1.85 X 10-7 sec, is approximately the value predicted for the reorientation of the long axis of a prolate ellipsoid with an axial ratio of 4 to 5 and the same volume as S-1 (27), and is four times that predicted for a sphere of the same volume (Eq. 1). The rotation of such an ellipsoid about its long axis would be about as fast as the rotation of the sphere (27). The long 72 observed indicates that the principal axis of the label is rigidly aligned parallel to the long axis of S-1 (within 300), and that S-1 has little internal flexibility. Assuming that these conditions do not change, we can use the spin label in the following experiments to probe the motion of S-1 as a

1732

Biophysics: Thomas et al. IASL ACTOMYOSIN

Proc. Nat. Acad. Sci. USA 72

(1975)

Preliminary Experiments Ussing MSL. Saturation transfer experiments on myosin labeled at the -SH2 groups with MSL show that this label is not rigidly attached to S-1 and is immobilized by F-actin to a lesser extent than is the iodoacetamide label at -SH1. The spectrum of maleimide-spinlabeled F-actin (Fig. 3, bottom) reveals slow rotation of the label (r2 = 5 X 10-4 sec), suggesting that this label is rigidly bound and that its motion, therefore, reflects the motion of the F-actin filament. DISCUSSION

MSL F ACTIN

FIG. 3. V2' spectra for acto-IASL-myosin (1 ml of a 20 mg/ml IASL myosin solution plus 1 ml of a 5 mg/ml F-actin solution), acto-IASL-S-1 (0.5 ml of a 20 mg/ml IASL S-1 solution plus 1 ml of a 5 mg/ml F-actin solution), and MSL F-actin (5 mg/ml), in appropriate buffers (see text) at 200, all from the same preparations of myosin and F-actin. Estimated correlation times are (top to bottom) 2 X 10-4 sec, 1 X 10-4 sec, and 5 X 10-5 sec, by comparison with reference spectra. Arrows in the MSL F-actin spectrum indicate where a small amount of loosely bound label affects the lineshape. 72 is therefore estimated mainly from the high-field portion of the spectrum. Ticks are at 20 G intervals.

whole, using the v2' spectra to estimate 72 for the reorientation of the long axis of S-1. The estimated values are at least lower bounds for the actual times. Supramolectlar complexes

IASL Myosin. The saturation transfer method is most powerful in the study of large complexes in which the mobility of myosin is reduced: Fig. 2 shows v2' spectra and estimated correlation times for monomeric myosin, myosin filaments, and actomyosin. Aggregation of myosin in low salt reduces the mobility of the label on the S-1 segment of myosin by a factor of about 10, and F-actin immobilizes this label by a factor of nearly 103. The value of 2 X 10-4 sec is a reliable lower bound for the T2 for reorientation of the long axis of S-1 in actomyosin, demonstrating the extremely rigid binding of the S-1 region to actin. The resulting immobilization is nearly 100 times that due to filament formation, which apparently has a less direct effect on S-i. We have also found that addition of F-actin to these filaments has a large immobilizing effect on S-1, based on the v2' spectrum (not shown here). Fig. 3 demonstrates that F-actin immobilizes S-1 isolated after cleavage, as well as the S-1 segment of intact myosin. The 72 for acto-IASL-S-1 is about half that for acto-IASLmyosin. These results provide further evidence that the spin label is rigidly bound to the protein and indicate that S-1 undergoes little or no independent motion when bound to F-actin in the absence of nucleotides. The actomyosin and acto-S-1 samples contained about one actin monomer per S-1. When G-actin was added to an IASL S-1 solution in a 1:1 molar ratio, both the vl and V2' spectra indicated only a slight

immobilization.

Calculated v2' spectra are moderately sensitive to changes in electron spin relaxation times (13). We performed steady-state saturation studies to determine H112, the microwave field strength required for half saturation and an indicator of relaxation times (28). The results suggest that high glycerol concentrations and low temperatures increase relaxation times, independently of motion. Based on the effects of increased relaxation times on calculated v2' spectra, the observed Hl/2 variations in Hb reference samples might increase the uncertainties of IASL correlation times to as much as a factor of 2 for T2 > 10-6 sec. No corrections were attempted for this possible error. In general the 72 values reported here should be considered accurate to within a factor of 2, although the precision is several times better. Since the motions of the muscle proteins are likely to involve restricted angular ranges, anisotropic rotation, and/or several different correlation times, our estimated effective correlation times may not describe the motions in detail. However, our evidence that the label is rigidly bound, with its principal axis parallel to the major axis of the rigid S-1 segment, argues that these effective correlation times are meaningful characteristic times for reorientation of the major axis of S-1. Better relaxation time measurements and extensions of the theory will be required to improve the accuracy and to investigate the details of complex motions (13, 15). Although the v1 and v2' experiments (Table 1) yield slightly different results, both methods show that (a) 72 decreases by only about 40% when S-1 is cleaved from the rest of myosin, even though the molecular weight decreases by more than a factor of 4, and (b) 72decreases by only about 15% when more than half of the "tail" is removed, producing HMM. As Mendelson et al. (4) pointed out when they obtained similar results from the depolarization of the fluorescence of a dyebound to S-1, these results indicate flexibility within the HMM region of myosin, possibly at the postulated "hinge" (1) between S-1 and S-2 (subfragment-2). Our data are also consistent with, but do not require, flexibility at the junction between HMM and the rest of myosin. Since the S-1 motion in HMM and myosin is less than twice as slow as the motion of the isolated subfragment, it appears that the motions of the two S-1 "heads" are somewhat independent. There seems to be little flexibility within the S-1 region itself, since the rotation of isolated S-1 is about four times slower than expected for a sphere of the same volume. A rigid S-1 region would be a requirement for models of contraction such as those proposed by H. E. Huxley (1) and by A. F. Huxley and Simmons (2). The rotational mobility of S-1 in myosin filaments is more relevant to the contraction mechanism than is S-1 mobility in solution. The increase in T2 by a factor of 10 shows that the mobility of S-1 markedly decreases upon formation of synthetic myosin filaments. However, the fastest component of

Proc. Nat. Acad. Sci. USA 72

(1975)

rotational motion of an ellipsoid shaped like one of these filaments (molecular weight about 5 X 107, ref. 29) would be about 10 times slower than observed for the spin label (27). The additional motion is probably due to rotation of S-i, independent of the filament as a whole. The v2' lineshape is more difficult to match with a reference spectrum than are the lineshapes from other samples, suggesting that S-1 motion in filaments may be more anisotropic or restricted to a narrower angular range. Because the S-1 motion in synthetic thick filaments is in a time range of high sensitivity for v2' experiments, this method promises to be useful in more detailed studies of thick filaments. We conclude that F-actin strongly immobilizes S-1, whether S-1 is present in intact myosin monomers, in synthetic myosin filaments, or as the proteolytic fragment. Explanation of the results in terms of immobilization of only that portion of S-1 that contains the spin label, leaving the remainder of S-1 free to move, is less plausible. This would require flexibility within S-1 itself, which is inconsistent with the long T2 of isolated S-1. A direct interaction between actin and the spin label is also unlikely, since (a) the spin label can react with -SHI groups whether F-actin is present or not, (b) the spin labels bound to -SH1 groups are equally accessible to dithiothreitol in the presence and absence of F-actin, and (c) the interactions among myosin, ATP, and actin are not greatly altered by labeling the -SHI groups (7). The extremely long T2 in actomyosin (even longer than in acto-S-1) and the evidence for independent rotation of the two S-1 segments indicate that both "heads" are bound to actin and thereby immobilized, in a preparation containing about two actin monomers per myosin molecule. In previous studies, the immobilization of IASL by F-actin was indicated by small changes in the unsaturated vl spectrum (7, 9). It was also noted that at saturating microwave power levels the vl spectrum became more sensitive to the effect of actin. Similar effects were observed in the present study, but the v2' method is a much more sensitive monitor of rotational effects on the nonlinear spin response. Motions of the label in actomyosin and in synthetic thick filaments occur in a time range of optimal sensitivity for the saturation transfer technique; this makes it possible to detect the large difference between the effects of filament formation and F-actin binding. Fluorescence depolarization measurements, having questionable applicability in this time range, yielded no significant difference between the two effects, and the estimated times were shorter (4). The motion of the maleimide spin label bound to F-actin (T2 = 5 X 10-i sec, Fig. 3) is much faster than the time scale accessible to the light scattering methods of Oosawa et al. (10), who found evidence for F-actin flexing motions with correlation times longer than 10-i sec. The spectra in Fig. 3 suggest that MSL, attached directly to actin, and IASL, attached to S-1 and myosin, may both be reporting the motion of the F-actin structure, which is slowed by the binding of S-1 and myosin. Although there is evidence that the principal axis of the maleimide label is approximately perpendicular to the filament axis (21), we cannot yet assign the observed T2 to any particular mode of actin filament motion. The technique of saturation transfer spectroscopy promises

S-1 Motion in Myosin: Saturation Transfer EPR

1733

to be a useful tool in studying rotation in biological systems in the time range of 10-7-10-' sec, and may be particularly useful in studies of molecular dynamics in highly organized systems such as the thick and thin filaments of muscle. We thank Prof. H. M. McConnell for many helpful discussions, particularly regarding the theoretical aspects of this work, which was supported in part by the Office of Naval Research under Grant no. 00014-67-A-0122-0045. D.D.T. has received fellowship support from the National Science Foundation, the Danforth Foundation, and the Josephine de Karman Trust. In addition, this work was supported in part by grants from the National Institutes of Health (HI,5949 and HL 15391), the National Science Foundation (GB43484), and the Muscular Dystrophy Associations of America, Inc., and by General Research Support Grant SO1 RR05711. 1. Huxley, H. E. (1969) Science 164, 1356-1366. 2. Huxley, A. F. & Simmons, R. M. (1972) Cold Spring Harbor Symp. Quant. Biol. 37, 669-680. 3. Lowey, S., Slayter, H. S., Weeds, A. G. & Baker, H. (1969) J. Mol. Biol. 42, 1-29. 4. Mendelson, R. A., Morales, M. F. & Botts, J. (1973) Biochemistry 12, 2250-2255. 5. Botts, J., Cooke, R., dos Remedios, C., Duke, J., Mendelson, R., Morales, M. F., Tokiwa, T., Viniegra, G. & Yount, R.

(1972) Cold Spring Harbor Symp. Quant. Biol. 37, 195-200. 6. Seidel, J. C. & Gergely, J. (1973) Ann. N.Y. Acad. Sci. 222, 574-587, and references cited there. 7. Seidel, J. C. (1973) Arch. Biochem. Biophys. 157, 588-596. 8. Seidel, J. C. & Gergely, J. (1973) Arch. Biochem. Biophys. 158, 853-863. 9. Tokiwa, T. (1971) Biochem. Biophys. Res. Commun. 44, 471. 10. Oosawa, F., Fujime, S., Ishiwata, S. & Mihashi, K. (1972)

Cold Spring Harbor Symp. Quant. Biol. 37, 277-285.

11. Hyde, J. S. & Dalton, L. R. (1972) Chem. Phys. Lett. 16,

568-572. 12. Hyde, J. S. & Thomas, D. D. (1973) Ann. N.Y. Acad. Sci. 222, 680-692. 13. Thomas, D. D. & McConnell, H. M. (1974) Chem. Phys. Lett. 25, 470-475. 14. Smigel, M. D., Dalton, L. R., Hyde, J. S. & Dalton, L. A. (1974) Proc. Nat. Acad. Sci. USA 71, 1925-1929. 15. Thomas, D. D., Seidel, J. C., Hyde, J. S. & Gergely, J. (1975) J. Supramol. Struct. 3, in press. 16. Huisjen, M. & Hyde, J. S. (1974) Rev. Sci. Instrum. 45,

669-675. 17. Nauss, K. M., Kitagawa, S. & Gergely, J. (1969) J. Biol. Chem. 244, 755-765. 18. Seidel, J. C., Chopek, M. & Gergely, J. (1970) Biochemistry 9, 3265-3272. 19. Seidel, J. C. (1972) Arch. Biochem. Biophys. 152, 839-848. 20. Drabikowski, W. & Gergely, J. (1962) J. Biol. Chem. 237, 3412-3417. 21. Burley, R. W., Seidel, J. C. & Gergely, J. (1972) Arch. Biochem. Biophys. 146, 597-602. 22. Ohnishi, S., Boeyens, J. C. A. & McConnell, H. M. (1966) Proc. Nat. Acad. Sci. USA 56, 809-813. 23. Freed, J. H., Leniart, D. S. & Hyde, J. S. (1967) J. Chem. Phys. 47, 2672-2773. 24. McCalley, R. C., Shimshick, E. J. & McConnell, H. M. (1972) Chem. Phys. Lett. 13, 115-119. 25. McConnell, H. M. & Gaffney McFarland, B. (1970) Quart. Rev. Biophys. 3, 91-136. 26. Carrington, A. & McLachlan, A. D. (1967) in Introduction to Magnetic Resonance (Harper and Row, New York), p. 189. 27. Tao, T. (1969) Biopolymers 8, 609-632. 28. Castner, T. G., Jr. (1959) Phys. Rev. 115, 1506-1515. 29. Josephs, R. & Harrington, W. F. (1966) Biochemistry 5,

3474-3487.