Introduction

Protein dynamics is crucial for biological function, especially for ATPases and regulatory proteins. Protein structural change is usually a requirement for ATPase activity during the interaction with a nucleotide. Conformational sampling is a requirement for successful function of a regulatory protein. The combination of site-directed spin labeling (SDSL) and electron paramagnetic resonance (EPR) is a well-established method to determine protein dynamics and protein conformations (Berliner 1976). Following protein site-directed cysteine mutagenesis, a nitroxide spin probe covalently binds to the mutated Cys residue and reports on local dynamics of the probe itself, conformational dynamics of protein domains, and global protein tumbling. Anisotropy of the nitroxide spin probe magnetic tensors provides exceptional sensitivity and resolution of EPR to probe orientation in the external magnetic field, producing high-resolution sensitivity to the spin probe’s rotational motion.

Several motional models have been developed to interpret EPR spectra of a spin probe in terms of its dynamics. Simple models, based on partial averaging of nitroxide magnetic tensors, are applicable to the approximation of fast (subnanosecond) and restricted motion of a spin probe, and are useful to describe protein conformation at the labeling site (Timofeev and Nikolsky 2003; Van et al. 1974). In this model, the EPR lineshape depends only on parameters describing the restriction of spin probe motion, e.g., the cone angle in a wobbling-in-cone model (Van et al. 1974) or the angle of oscillation (Timofeev and Nikolsky 2003). Advanced models, based on quantum mechanical description of evolution of spin probe ensemble magnetization, were developed to model slow restricted motion of a spin probe within a protein, and protein overall tumbling in solution (Freed 1976; Hustedt et al. 1997; Robinson et al. 1992). In this case, more parameters are required to describe the spin probe EPR lineshape: diffusion tensors that reflect the rotational rates and anisotropies of both spin probe and protein, and potentials that reflect restrictions of these motions. These variable parameters, along with unknown spin–spin relaxation time of a probe (responsible for EPR spectra broadening), make fits of single-frequency EPR spectra ambiguous. For example, the effects of increased rotational rate can be hard to distinguish from those of increased rotational amplitude. This ambiguity is greatly decreased in a simultaneous fit of EPR spectra that are obtained for the same labeled protein at different frequencies. Ambiguity is reduced because of the sensitivities of EPR spectra to rotational rates and amplitudes are quite frequency dependent.

EPR and nitroxide: winning combination

Resolution to spin probe orientation in the external magnetic field

In the nitroxide radical, an unpaired electron is localized in the π orbital of the N–O bond, with the z-axis oriented perpendicular to the N–O bond, and the x-axis oriented parallel to the N–O bond (Fig. 1). This asymmetry of electron density leads to orientation dependence of the nitroxide EPR lineshape in the external magnetic field. The distribution of the unpaired electron density in the nitroxide is reflected in the g-factor tensor (Equation 1).

$$ H = \left( {hv} \right)/\left( {g{\beta_e}} \right) + {m_1}A, $$
(1)

where H is the resonant magnetic field, ν the resonant frequency, h Planck’s constant, β e Bohr’s magneton, m I = (−1, 0, +1) the nitrogen nuclear spin quantum number, and A the hyperfine splitting tensor. Equation 1 determines the magnetic field of resonance; the higher the EPR frequency, the better resolution for g-factor components (first term in Equation 1). The interaction of the unpaired electron with the nitrogen nucleus results in splitting of the EPR line into a triplet (due to the second term in Equation 1). The splitting is orientation-dependent and determined by the electron–nucleus hyperfine interaction tensor A. Higher electron density in the z direction produces larger electron–nucleus interaction and consequently a larger EPR line splitting when the z axis of the spin probe is oriented parallel to the external magnetic field. y- and x-components of electron density are smaller, producing smaller line splitting when y- or x-axes of the nitroxide are oriented along the magnetic field H. The value of the hyperfine splitting does not depend on the frequency ν (Equation 1), so the center of each triplet spectrum is determined entirely by the g-tensor and depends on frequency ν (first term in Equation 1). Increasing ν increases the resolution of the x-, y-, and z-components in the EPR spectrum. Thus, the resolution to g-factor spectral components becomes comparable to the hyperfine splitting A z at 1.036 T, and 29.094 GHz for a typical nitroxide (g x = 2.0087, g y = 2.0058, g z = 2.0018, A x = 7.1 G, A y = 7.2 G, A z = 36.7 G). At lower magnetic field and frequency (for example, 0.34 T, 9.4 GHz, X-band), the lineshape of the nitroxide EPR spectrum is determined mainly by the anisotropy of the hyperfine tensor, while in higher fields (3.4 T, 94 GHz, W-band), both g-tensor and hyperfine tensor contribute to EPR spectral lineshape (Figs. 1 and 2).

Fig. 1
figure 1

The position and splitting of the nitroxide spin probe EPR spectrum depends on probe orientation in the external magnetic field H (Equation 1). In the “powder” spectrum all spin probes’ orientations are present. Resonance line position and splitting depends on tensors g and A, respectively. H = 0.34 T, ν = 9.4 GHz (X-band), H = 3.4 T, ν = 94 GHz (W-band)

Full size image
Fig. 2
figure 2

X-band (0.34 T, 9.4 GHz) and W-band (3.4 T, 94 GHz) “powder” EPR spectra of iodoacetamido-TEMPO-labeled myosin head S1. x, y, and z are the spectral components of g and A tensors, not resolved in X-band and clearly resolved in W-band due to better g-factor resolution

Full size image

Resolution to spin probe dynamics

Fast isotropic motion averages g- and A-tensors of the nitroxide spin probe, reducing them to scalar values g 0 = 1/3(g x + g y + g z), and A 0 = 1/3(A x + A y + A z) and producing a three-line EPR specrtrum, similar to the EPR spectrum of an oriented nitroxide in the absence of motion (Fig. 3, top left). The frequency of averaging motion for A- and g- tensors is, accordingly, 2π(A z – A x)β e g 0/h and 2π(g x – g z)β e H/h. Rotational motion of a spin probe much faster than these averaging frequencies does not change the EPR spectrum lineshape, while slower significantly changes the EPR lineshape, until the powder limit is reached (Fig. 3, bottom row). The restrictions of a spin probe motion also affect the EPR lineshape (Fig. 3, left to right). Another factor affecting the EPR lineshape is the spin–spin relaxation time, T2, of the nitroxide, which contributes to the linewidth.

Fig. 3
figure 3

Lineshapes of nitroxide spin probe EPR spectra (X-band, left panel, and W-band, right panel, X- and W-band spectra are not to scale) at different rotational correlation times and order parameters (reflecting restriction of motion modeled as a Maier-Saupe potential; Freed 1976)

Full size image

EPR spectra interpretation and fit

There are several models to describe spin probe dynamics. The simplest model assumes fast motion of a spin probe much faster than the averaging rate. Restrictions of spin probe motion lead to partial averaging of the magnetic tensors and a more complex EPR spectrum lineshape. Several models of motion are developed to describe different character of a fast motion of a spin probe: wobbling in a cone, uniaxial rotation, or limited angle oscillations (Van et al. 1974; Nordio 1976; Timofeev and Nikolsky 2003). The spectral lineshape depends on the angle of oscillation (or the angle of a cone); the angle can be immediately found from the apparent hyperfine splitting, A z′, due to the relationship of A z (the value of z-component of hyperfine splitting tensor, found for immobilized spin probe) and A z′ (apparent value of z-component of hyperfine splitting tensor, determined from the spectrum) in Euler transformations (Van et al. 1974). High frequency and low frequency EPR spectra are equivalent in these models, the high frequency EPR experiment provides better insight into the character of a spin probe motion due to better resolution of x, y, and z spectral components.

A more complex model, assuming slow and restricted motion of a spin probe, considers the evolution of the total magnetization of the sample. Two approaches are developed currently, the microscopic order, macroscopic disorder (MOMD) approach, assuming slow and restricted motion of a spin probe within randomly distributed immobile protein molecules (Freed 1976). The slow relaxing local structure (SRLS) approach (Polimeno and Freed 1995) assumes slow restricted motion of a spin probe within a protein, which itself tumbles in solution. Slow motion of a spin probe is described with the diffusion coefficient DR; usually, the diffusion coefficient of a spin probe has tensor value. To account for isotropic tumbling of labeled protein, additional diffusion coefficient is introduced. The restrictions of spin probe motion within the protein are modeled with irreducible spherical harmonic functions, using their coefficients as variable parameters in the EPR spectral fits (Budil et al. 1996). Figure 3 shows simulated X-band EPR spectra for a spin probe moving isotropically in a Maier-Saupe potential (Freed 1976), the first term of the spherical harmonics function series. According to Fig. 3, EPR spectra lineshape is extremely sensitive to spin probe motion on the nanosecond time scale, reflecting the spin probe’s rotational correlation time and restrictions of motion, as well as resolving multiple populations of spin probes with different motions. Figure 3 also illustrates how hard it is to distinguish between spectra of spin probe with slow unrestricted motion and fast restricted motion; the outer extrema splittings are similar for these types of spin probe motion. The lines of EPR spectrum are narrower for a faster moving spin probe, but spectral broadening, usually unknown due to the unknown spin probe’s relaxation time T2, masks sharp spectral features, making it hard to distinguish the spectral response of a spin probe’s fast restricted motion or slow unrestricted motion. This problem is solved in multifrequency EPR, which can distinguish between fast restricted and slow unrestricted spin probe motion.

Analysis of EPR spectra, obtained at different frequencies

Two tensors, g-factor and hyperfine splitting tensor A, determine the magnetic properties of a spin probe and the lineshape of its EPR spectrum (Equation 1). The g-tensor determines the value of resonant magnetic field; the resolution of x-, y-, and z-components of the g-tensor depends on the frequency of the resonance: the higher the microwave frequency, the better the resolution to g-tensor components (Fig. 2). The value of the hyperfine splitting does not depend on resonant frequency and remains constant. Molecular motion of a spin probe modulates both tensors, and the modulation of g- and A-tensors depends on the microwave frequency, making multifrequency EPR a valuable tool to determine the molecular dynamics of a spin probe, and the labeled protein.

To illustrate the difference in the effect of molecular motion on g- and A-tensors at different microwave frequencies, one can consider the motion that averages the tensors. The frequency of spin probe motion that averages the g-tensor can be calculated as 2π(g x – g z)β e H/h, which depends linearly on the resonant magnetic field strength H. The greater H, the higher is the averaging frequency of spin probe motion (Fig. 4, “g-factor averaged”). In contrast, the frequency of averaging for the A-tensor is 2π(A z – A x)β e g 0/h, independent of H (Fig. 4, “hyperfine splitting averaged”). The area under the lines “g-factor” and “hyperfine splitting” represents the spin probe motional regime with non-averaged tensors g and A; the area above these lines represents motion, averaging g- and A-tensor values, respectively. Thus, at X-band (9 GHz, 0.34 T), the g-tensor gets averaged at slower spin probe motion than the A-tensor. At W-band (94 GHz, 3.4 T), this is reversed (Fig. 4), so the W-band EPR spectrum is sensitive to faster motion. At a certain rate, both g- and A-tensors at X-band are completely averaged, but at W-band, the g-tensor is still anisotropic. This simple consideration of the rate of molecular motion, averaging tensors g and A, explains the different sensitivity of the EPR experiment at different frequency to a molecular motion.

Fig. 4
figure 4

The effect of molecular motion on g- and A-tensors averaging. Field position of g-components is frequency dependent and at higher resonant frequency g-factor averages at faster motion rate. The value of hyperfine splitting does not depend on resonant frequency and averaged A-tensor value is a constant number. a Both g- and A-tensors are averaged, three line EPR spectrum; b g-tensor is averaged, A-tensor is not; c g-tensor is not averaged, A-tensor is averaged; d both g- and A-tensors are not averaged

Full size image

Initially, the multifrequency approach for EPR spectral analysis was realized separately for higher and lower frequency EPR. First, the high-frequency EPR spectrum was fitted in the MOMD approach, assuming the absence of protein tumbling, to determine spin probe rotational dynamics (correlation times, order parameters). Then, the determined parameters of spin probe motion were fixed in the fit of low frequency EPR spectra to determine the diffusion coefficient of labeled protein global tumbling in solution (Barnes et al. 1999; Borbat et al. 2001). It was shown later that even at high frequency, global tumbling affects the EPR spectrum lineshape (Liang et al. 2004), and simultaneous fit of spectra at all frequencies was adopted with SRLS (Liang et al. 2004; Sezer et al. 2009).

The choice of “good” initial parameters for a spectral fit is important. The multifrequency approach decreases the ambiguity in this case. The analysis of EPR spectra splitting at different frequencies helps to narrow the region of the initial parameters, leading to convergence in multifrequency spectra fit (Nesmelov et al. 2007).

T4L-MTSSL dynamics in solution

Bacteriophage T4 lysozyme (T4L) is a well-characterized enzyme in terms of structure and stability, and was chosen as a model system to study protein backbone flexibility and side chain dynamics (Barnes et al. 1999; Borbat et al. 2001; Columbus and Hubbell 2002; Columbus et al. 2001; Liang et al. 2004; Sezer et al. 2009). Several residues of Cys-less T4L were mutated to cysteine and labeled with MTSSL: S44 and Q69 (Barnes et al. 1999; Borbat et al. 2001), D72 and V131 (Liang et al. 2004; Sezer et al. 2009). All mutated residues are solvent exposed and located on alpha helices. For S44C and Q69C, EPR spectra at 9 and 250 GHz were acquired (T = 10C) (Barnes et al. 1999; Borbat et al. 2001); for D72C and V131C, EPR spectra at 9, 94, 170, and 250 GHz were acquired at different temperatures (T = 2°C–37.5°C) (Liang et al. 2004; Sezer et al. 2009). Multifrequency EPR spectra were fitted sequentially (high frequency first, to determine local motion of a spin probe in the MOMD approach, and then low frequency to determine global tumbling of labeled protein, in the SRLS approach) (Barnes et al. 1999; Borbat et al. 2001; Liang et al. 2004), or simultaneously in the SRLS approach (Liang et al. 2004). All studies were agreed on the diffusion coefficient for lysozyme global tumbling at T = 10°C, τ c = 11–15 ns (Barnes et al. 1999; Liang et al. 2004). All four labeling sites showed a two-component EPR lineshape, especially at a lower temperature (Liang et al. 2004). The motion of a spin probe at all labeling sites was determined to be anisotropic for both populations; with sub-nanosecond correlation time about x and y axes and tens of nanoseconds correlation time in the x–y plane about the z axis (\( {\tau_\bot } = 0.08ns - 0.8ns \), except for the minor population of V131C, \( {\tau_\bot } = 6.4ns \), and for the minor population of Q69C, τ y = 2.1–3.3 ns; \( {\tau_\parallel } = 3.7ns - 33ns \), except for the major population of Q69C, \( {\tau_\parallel } = 0.4ns - 0.8ns \)). The restoring potential for a spin probe is characterized by two coefficients, c20 and c22; the motion of the major component is less restricted than the motion of the minor component (major component, c20 = 0.8–2.8 kT, c22 = −1.7 to 0.6 kT; minor component, c20 = 2.0–2.5 kT, c22 = −1.5 to −2.1 kT; except for S44C and Q69C, minor component, c20 = 4.1 kT–5.2 kT). The mean distribution of spin label orientation for both components is shown in Fig. 5 (major component, c20 = 1.8 kT, c22 = −0.5 kT; minor component c20 = 3.8 kT, c22 = −1.8 kT).

Fig. 5
figure 5

MTSSL at T4 lysozyme: two populations of the spin probe within protein, with less (a) and more (b) restricted motion. Probability distributions of the probe’s z-axis orientation, obtained from 9 and 250 GHz EPR spectra

Full size image

The recently developed approach of EPR spectra simulation from MD trajectories gives a detailed look at the spin probe orientation and interaction with surrounding residues. Analysis of D72C and V131C labeling sites (Sezer et al. 2009) shows that spin probe indeed accepts two conformations, due to a large energy barrier for motion around the disulfide bond. It was shown previously that alanine mutation of adjacent residues on the helix (i ± 3, i ± 4) does not affect X-band EPR spectra of labeled T4L (Columbus and Hubbell 2002; Mchaourab et al. 1996). MD simulations show that the spin probe participates in diverse hydrophobic and electrostatic interactions with several neighbor residues simultaneously, which restricts the spin probe motion and explains the absence of dramatic effect of i ± 3, i ± 4 alanine mutations.

TOAC-PLB dynamics in a membrane

Phospholamban (PLB) is a 52-amino-acid amphipathic integral membrane protein that regulates the active transport of calcium in the heart through the interaction with the cardiac sarcoplasmic reticulum Ca-ATPase (SERCA) (MacLennan and Kranias 2003; Reddy et al. 1999). The NMR structure of monomeric PLB in detergent micelles (PDB 1N7L) (Zamoon et al. 2003) reveals a predominant L-shaped conformation with two approximately perpendicular helices, cytoplasmic (residues 1–16) and transmembrane (residues 21–52), connected by a flexible hinge. According to solid-state NMR (Mascioni et al. 2002; Traaseth et al. 2006) and EPR (Karim et al. 2004; Kirby et al. 2004), PLB’s predominant conformation in the membrane has the transmembrane domain approximately perpendicular to the membrane. NMR data show the cytoplasmic domain oriented almost perpendicular to the transmembrane domain, lying along the membrane surface and interacting with lipid headgroups. EPR quenching studies confirm interaction of cytoplasmic domain with the surface of membrane (Karim et al. 2004). However, cross-linking (James et al. 1989) and functional mutagenesis (Toyofuku et al. 1994) data suggest that the cytoplasmic domain of PLB extends well above the membrane surface when it interacts with SERCA, so most models of the PLB-SERCA complex show PLB straightened almost linearly on the surface of SERCA, with the entire molecule, including the cytoplasmic domain, nearly perpendicular to the membrane surface (Hutter et al. 2002; Karim et al. 2004, 2006; Toyoshima et al. 2003; Zamoon et al. 2005).

To understand how PLB undergoes this dramatic structural transformation in a membrane, PLB labeled with TOAC (paramagnetic amino acid that reports directly the dynamics of the peptide backbone; Karim et al. 2004) was synthesized. TOAC-PLB was reconstituted into DOPC/DOPE lipid vesicles (1/200 protein/lipid ratio), EPR spectra were acquired at 9.4 and 94 GHz at T = 4C. The spin probe was strongly immobilized, when TOAC is inserted into the transmembrane domain of PLB, consistent with a well-ordered α-helix. It was determined that the transmembrane domain rotates slowly (\( {\tau_\parallel } = 105ns \)) around the helix axis and performs a rapid (\( {\tau_\bot } = 2ns \)), but very restricted (c20 = 14 kT) wobbling motion perpendicular to this axis in the membrane (Fig. 6). TOAC-PLB with the labeled site in the cytoplasmic domain shows clear evidence for two resolved conformations, one a well-ordered helix (the T conformation, τ iso = 3.5 ns, c20 = 4.2 kT, 84% mole fraction; Fig. 6b), and the other dynamically disordered on the nanosecond time scale (the R conformation, τ iso = 0.68 ns, c20 = 0.52 kT, 16% mole fraction; Fig. 6c). The disordered R conformation binds preferentially to SERCA (Karim et al. 2006; Zamoon et al. 2005) and is enhanced by phosphorylation of PLB (Karim et al. 2006), resulting in SERCA activation. The T conformation interacts with the surface of the membrane, slowing its motion and stabilizing its helical conformation. It is likely that the large rate and amplitude of the cytoplasmic helix rotational motion is due, at least in part, to the partial unfolding of this helix, as suggested previously by both NMR (Metcalfe et al. 2004) and EPR (Karim et al. 2004) data.

Fig. 6
figure 6

TOAC-labeled PLB in lipid bilayer: probability distribution of PLB domains orientation in a membrane, revealed from 9.4 and 94 GHz EPR spectra. TOAC label is rigidly attached to PLB backbone and distribution of spin probe’s z-axis orientation, measured by EPR reflects distribution of PLB domain orientations. a Transmembrane domain, very narrow distribution; b cytoplasmic domain, T conformation; c cytoplasmic domain, R conformation, practically unrestricted motion

Full size image

Myosin S1-IASL dynamics in solution

Myosin is a motor enzyme involved in muscle contraction and cell motility through a cycle of actin-activated ATP hydrolysis. Myosin’s head domain, subfragment 1 (S1), is responsible for the coupling of ATP hydrolysis and force generation. During the ATPase cycle, myosin shows two dramatic conformational changes, the powerstroke, when the force gets generated, and the recovery stroke, when myosin molecule primes to perform a powerstroke. Three structural states of myosin were originally identified by intrinsic fluorescence (Bagshaw and Trentham 1974; Johnson and Taylor 1978): M, the apo state, M*, the post-powerstroke state, trapped with ADP nucleotide, and M**, the pre-powerstroke state, trapped with ADP.Vi. These structural states are assumed to be tightly coupled to biochemical states, which are defined by the nucleotide bound to the active site. These conformational states of myosin S1 have also been observed and characterized by EPR through the changes in mobility of a nitroxide spin probe (IASL) at the SH1 labeling site (C707) (Agafonov et al. 2008; Barnett and Thomas 1987; Nesmelov et al. 2008; Ostap et al. 1993; Seidel et al. 1970) in the force-generating region of myosin.

Magnetic tensor values were determined from fitting of W-band EPR spectra of IASL-S1 at T = 80 K, and were kept constant. The rotational correlation time of global S1 tumbling were determined as D Riso = 1.3 × 106 rad2 s−1, τ c = 128 ns (Nesmelov et al. 2008) and was kept constant. The rotational correlation time and restrictions of spin label motion within myosin, Lorentzian broadening, and the angle between the magnetic frame and the diffusion frame of the spin label (β d) were variable parameters. Figure 7 shows the probability of the spin label orientation distribution for each biochemical state of IASL-S1. In the apo S1 biochemical state, multifrequency EPR unambiguously detects a single structural state of myosin in the force-generating domain, M. The motion of the spin label is complex; the molecular frame of the spin label is tilted in the diffusion frame (β d = 32o), the spin probe moves slowly on the nanosecond time scale (\( {\tau_\bot } = 9.9ns \), \( {\tau_\parallel } = 64ns \)), and the motion is very restricted (c20 = 4.36 kT), Fig. 7a. These slow restricted motions probably reflect those of the secondary structural elements in close contact with the labeled site. Indeed, our modeling of the structure of D. discoideum myosin in the apo S1 state (1FMV; Bauer et al. 2000) shows that the spin probe is located between the converter and N-terminal domains of myosin (Fig. 8). The spin label is squeezed between myosin domains, but not buried under the surface, since g- and A-tensor values do not reflect changes in polarity for different myosin states. In the S1.ADP state, a single structural state (M*) is also determined. In this conformation of the force-generating domain, the spin probe moves slightly faster, its motion is less restricted relative to the apo S1, and the tilt of the molecular frame of the spin label in the diffusion frame is less (\( {\tau_\bot } = 6.8ns \), \( {\tau_\parallel } = 25.2ns \), c20 = 1.88 kT, c22 = −0.36 kT, β d = 5o), Fig. 7b. The structure of D. discoideum myosin with bound ADP (1MMA; Gulick et al. 1997) shows a slightly different position of the converter relative to N-terminal domain, compared to that of apo S1, allowing more space for the spin label to move, in qualitatively agreement with EPR data. A much more complex structural picture emerges in the S1.ADP.Vi complex, which is presumed to correspond to the pre-powerstroke structural state M**. The EPR spectrum in this complex shows at least two, and probably three, structural states, while only one structural state is revealed by the crystal structure, 1VOM (Smith and Rayment 1996). One spectral component (component 1, 16%) is similar to that of IASL-S1.ADP, suggesting the presence of M* structural state. The two other spectral components (corresponding to higher mobility, with mole fractions 78% and 6%) have different dynamics of the spin probe (component 2, 78%, τ x = 1.0 ns, τ y = 1.8 ns, τ z > 300 ns, c20 = 4.96 kT, c22 = −4.91 kT, c40 = −1.93 kT; Fig. 7c; component 3, 6%, τ x = 0.2 ns, τ y = 0.1 ns, τ z > 300 ns, c20 = 6.42 kT, c22 = −6.24 kT; Fig. 7d), there is no tilt of the molecular frame relative to the diffusion frame. These spectral components could be assigned as the M** structural state of myosin (Barnett and Thomas 1987). According to the structure of the S1.ADP.Vi state, the spin probe is located between the converter and N-terminal domains, while the relay helix is bent in this structural state and is not likely to affect spin label motion. The two observed distributions of spin label orientation in the M** structural state (78% and 6%) could reflect a slight difference in relative position of the converter and N-terminal domains: rotation of the spin label about the x axis is allowed in one state (M1**, 6%; Fig. 7d) and stopped in another state (M2**,78% Fig. 7c). This slight difference in relative position of domains within one myosin population (M**) could be interpreted in terms of flexibility between myosin domains in the M** state on the microsecond and slower time scale (Houdusse and Sweeney 2001), and could be treated as a slow exchange between myosin conformations, producing two spectral components.

Fig. 7
figure 7

IASL labeled myosin S1; the spin probe is located between the converter and N-terminal domains of myosin (Fig. 8). Orientational distribution of the spin probe’s z-axis within myosin in three structural states, M, apo myosin (a), M*, myosin with ADP bound (b), M**, myosin with the nucleotide analog ADP.Vi bound (distributions (c) and (d)). Both distributions (c) and (d) reflect spin probe motion about its x axis, restricted (c) and unrestricted rotation (d). Domain flexibility on the microsecond time scale is proposed to explain two distributions of the spin probe orientation in one structural state of myosin (Agafonov et al. 2008; Nesmelov et al. 2008)

Full size image
Fig. 8
figure 8

IASL probe within myosin S1, squeezed between the converter domain (blue) and the N-terminal domain (gray). Pre-powerstroke M** structural state (1vom.pdb). According to Fig. 7c, d, the spin probe oscillates (c) or freely rotates (d) about its x-axis (N–O bond, blue and red atoms of the spin probe sticks image). Two spin probe distributions within one structural state M** might reflect domains flexibility within myosin on microsecond time scale

Full size image

Conclusion

Multifrequency EPR is a unique spectroscopic tool, allowing detailed characterization of protein dynamics. With multifrequency EPR, the spin probe’s local dynamics can be unambiguously separated from protein overall tumbling. Analysis of spin probe motion within a protein reveals dynamics of protein domains and characterizes protein conformation. Multiple conformations of a spin labeled protein can be distinguished with multifrequency EPR. A model system, spin labeled T4 lysozyme, demonstrates these important features of the multifrequency EPR approach. Subsequent studies of TOAC-labeled phospholamban in lipid bilayer and IASL-labeled myosin in solution have used this approach for detailed characterization of functional dynamics of regulatory peptide in a membrane and structural changes of myosin during its conformational changes in the ATPase cycle. These studies confirm the unique properties of multifrequency EPR in determination of protein functional dynamics.