Librational fluctuations in protein glasses

Biochimica et Biophysica Acta 1834 (2013) 1591–1595

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbapap

Librational fluctuations in protein glasses Derek Marsh a,⁎, Rosa Bartucci b, Rita Guzzi b, Luigi Sportelli b, Mikael Esmann c a b c

Max-Planck-Institut für biophysikalische Chemie, 37070 Göttingen, Germany Dipartimento di Fisica, Laboratorio di Biofisica Molecolare, Università della Calabria, 87036 Arcavacata di Rende, CS, Italy Department of Physiology & Biophysics, Aarhus University, 8000 Aarhus, Denmark

a r t i c l e

i n f o

Article history: Received 3 January 2013 Received in revised form 25 April 2013 Accepted 3 May 2013 Available online 10 May 2013 Keywords: Glass transition Arrhenius behaviour Na,K-ATPase Spin label EPR

a b s t r a c t Librational motions in the region of the protein “glass” (or dynamic) transition are analysed for spin-labelled haemoglobin, serum albumin and β-lactoglobulin by EPR spectroscopy. A discontinuity in the temperature dependence of the mean-square librational amplitude, bα2>, occurs in the region of 200 K as found for the mean-square atomic displacement, br 2>, at the protein dynamic transition by Mössbauer spectroscopy and neutron scattering. The discontinuity in bα2> vs. T can be described by the Vogel–Tammann–Fulcher equation, implying a finite glass transition temperature. Above the dynamic transition, bα2> vs. 1/T can be approximated by the Arrhenius law with activation energies similar to those usually found for br 2>, and relaxation processes in glass-forming media and the hydration shells of proteins. Similar results are found for librational fluctuations of membranous Na,K-ATPase spin-labelled either on superficial \SH groups or on those essential to activity. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Several aspects of the conformational behaviour of solvated proteins strongly resemble the properties of glass-forming liquids [1–4]. At a characteristic temperature, in the region of 200 K, the hydrated protein becomes frozen into a non-equilibrium distribution of almost isoenergetic conformational substates [5,6]. This recalls the heterogeneous metastable distribution of local environments that is found in the glassy state [7]. Below 200 K, the protein motions are restricted to harmonic vibrations and librations, as in a solid and characteristic also of the vitreous state. In this regime, the mean-square atomic displacements, br 2>, of the protein increase linearly with temperature. Above 200 K, a dynamic transition [8] takes place beyond which thermal fluctuations establish equilibrium between the different conformational substates. The atomic displacements of the hydrated protein then increase much more rapidly with increasing temperature [3,8], which characterises the onset of stochastic (i.e., diffusive) motions. This situation corresponds to the onset of rapid translational diffusion that takes place on melting of vitreous systems at the glass transition. The position of the dynamic protein transition depends on the rate of heating and on the timescale of motions to which the experimental method for detecting the transition is sensitive [3,9], just as for the glass transition. The dynamic transition of the glass-like protein state was characterised originally by the discontinuity in temperature dependence of translational displacements, from measurements of br 2> ⁎ Corresponding author. Tel.: +49 551 201 1285. E-mail address: [email protected] (D. Marsh). 1570-9639/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bbapap.2013.05.001

either by Mössbauer spectroscopy [10,11] or quasi-elastic neutron scattering [8,12–14]. The purpose of this communication is to show that the dynamic protein transition is accompanied also by the onset of Brownian torsional fluctuations, as recorded by the electron paramagnetic resonance (EPR) spectra of spin-labelled proteins. In particular, it is found that the temperature dependence of the librational motion above the transition is very similar to that of the atomic displacements, which are recorded by the br 2> parameter. Previously, we have demonstrated the multi-state nature of the energy landscape of spin-labelled proteins in the glassy state, both by analysis of conventional EPR line shapes and by time-resolved EPR [6,15]. 2. Method of analysis The mean-square amplitude of librational motion, bα 2>, can be derived from the motionally averaged hyperfine splittings, 2bAzz>, in the EPR spectra of the spin-labelled proteins, according to Ref. [16]: D

α

2

E

¼

Azz −hAzz i Azz −Axx

ð1Þ

where the angular brackets indicate a motionally averaged hyperfine tensor (of principal elements Axx, Ayy and Azz) — see also Ref. [17]. Eq. (1) is limited to small librational amplitudes. In the stochastic regime, the mean-square diffusive displacement at short times t′ is given by an Einstein-type relation: 2

b α >¼ 2Dα t′

ð2Þ

1592

D. Marsh et al. / Biochimica et Biophysica Acta 1834 (2013) 1591–1595

where Dα is the effective rotational diffusion coefficient for libration within the torsional potential well. The temperature dependence of the effective diffusion coefficient is determined by an activation barrier that is associated with the rotational friction coefficient, fα, where Dα = kBT / fα [18]. For a spin-label EPR experiment of this type, the characteristic timescale of observation, t′, is determined by the inverse intrinsic line width ≈T2⁎ and the effective diffusion coefficient (or correlation time, τc). Resolution of a measurable shift in hyperfine splitting requires that the numerator in Eq. (1) should be at least ~ 10 −2 of the line width. In the temperature range above 200 K, where the current measurements are made, this corresponds to an experimental uncertainty in bα 2> that is ~ 5 × 10 −3 rad 2 [6]. From Eqs. (1) and (2), the corresponding minimum value of t′ is:  =T 2

1 10 ≈0:005τ c 2Dα γe ðAzz −Axx Þ

0.08

0.04 0.02 0.00

2

ð3Þ

where γe is the gyromagnetic ratio of the electron. Of course, the effective diffusion coefficient and τc are temperature dependent, but for −11 τc ≈ 2 × 10 −9 s (see Ref. [6]) Eq. (3) predicts that t′min ≈10 s. This corresponds to the temperature below which bα 2> becomes too small to be measured by conventional CW EPR. Note that this limit on resolution is not instrumental but determined by the spectroscopic properties of the spin-labelled system, because EPR spectra are recorded with field-modulation amplitudes that are much less than the intrinsic line widths. This therefore differs from the situation that obtains with neutron scattering (see, e.g., Ref. [19]). In the temperature range over which the librational amplitudes are measurable by EPR, inhomogeneous line broadening is small and therefore T2⁎ ≈ T2, the true transverse relaxation time that determines the Lorentzian line width [6].

A) Hb/6-MSL

0.06

libration amplitude, <α2> (rad )

t′min ¼

−2

0.10

160 0.03

180

200

220

240

260

280

B) HSA/5-MSL

0.02 0.01

water glycerol

0.00 160 0.04 0.03

180

200

220

240

260

280

220

240

260

280

C) β-LG/5-MSL

0.02 0.01 0.00 160

180

200

temperature (K) Fig. 1. Temperature dependence of the mean-square amplitude of librational motion, bα2>, for: A. 6-MSL-labelled haemoglobin in water (solid circles), and in 60% v/v glycerol–water (open circles); B. 5-MSL-labelled human serum albumin in water (solid circles); C. 5-MSL-labelled β-lactoglobulin in water (solid circles), and in 60% v/v glycerol–water (open circles). Solid lines are fits of Eq. (4) that were performed on a log scale vs. 1/T.

3. Soluble proteins Fig. 1 shows the temperature dependence of the mean-square librational amplitude, bα2>, for spin-labelled human haemoglobin (Hb), human serum albumin (HSA) and bovine β-lactoglobulin (β-LG). Hb is spin-labelled on Cys-β93 of the β-subunits with 4-maleimido2,2,6,6-tetramethylpiperidine-1-oxyl (6-MSL); HSA is spin-labelled on Cys-34 with 3-maleimido-tetramethylpyrrolidine-1-oxyl (5-MSL); and β-LG is spin labelled on Cys-121 with 5-MSL. Both Hb and HSA are α-helical proteins, whereas β-LG consists of a β-barrel. Data are given in Fig. 1 for the proteins in water (i.e., buffer) and in a glassforming 60% v/v glycerol–water mixture. Librational motions of appreciable amplitude set in at temperatures of 180−200 K, for the proteins in water and slightly above this for the protein in glycerol–water mixtures. The diffusive librations that occur above 200 K are a property of the hydrated protein, because they are virtually absent for the lyophilised spin-labelled proteins. Time-resolved (spin echo-detected) EPR spectra show that they take place on a sub-nanosecond time scale [6,20]. Considerable variation is found between the librational amplitudes, bα2>, of the different systems in Fig. 1. This arises from heterogeneity between the different single sites of labelling, which would become averaged if many residues were labelled on a particular protein. Mean square displacements, br2>, averaged over all non-exchangeable protons by neutron scattering, are found to be rather similar for different α-helical proteins, for example, myoglobin and lysozyme [19]. Recently, however, it was found that corresponding values of br 2> for a β-barrel protein are of much smaller amplitude [21]. This correlates with the smaller values of bα2> obtained for β-LG than for Hb in Fig. 1. The temperature dependence of the librational motion that is recorded by EPR strongly resembles that of the diffusive displacements measured by Mössbauer spectroscopy and quasi-elastic neutron scattering [8,11]. In particular, it can be described by the Vogel–

Tammann–Fulcher equation that is appropriate to glass-forming solvents [3]: D

α

2

E

¼ A expð−BT o =ðT−T o ÞÞ

ð4Þ

where To the temperature at which 〈α2〉 → 0, and A and B are fitting constants. An equation of this form is also the basis for the freevolume model of translational diffusion [22,23]. Qualitative fits of Eq. (4) to the spin-label EPR data are shown by the solid lines in Fig. 1. Note that the fitting parameters can be determined only with limited precision, because the fractional errors in the data become rather large as 〈α 2〉 → 0 (see the discussion surrounding Eq. (3) above). Nonetheless, the fits are adequate to establish that To increases when 60% glycerol is added to the medium. Although the form of the temperature dependence of the librational amplitudes is rather similar in 60% aqueous glycerol to that in water, reflecting only the difference in effective glass transition temperatures, the absolute values of 〈α2〉 are lower in 60% glycerol. This effect of solvent viscosity is expected for the diffusion coefficient (i.e., Dα = kBT / fα), or more generally from Kramers theory for the coupling of dynamic processes to the environment [24]. Most significantly, the temperature dependence of the spin-labelled proteins reported here strongly resembles the librational motions found for small spin labels in glass-forming solvents by Paschenko et al. [25] and in our own measurements with 5-MSL alone and the TEMPONE (2,2,6,6-tetramethylpiperidine-4-oxo-1-oxyl) spin label in 60% aqueous glycerol (data not shown). It is also interesting to note that qualitatively similar data are found for the librational motions of spin-labelled lipid chains in membranes [26]. The latter results are

D. Marsh et al. / Biochimica et Biophysica Acta 1834 (2013) 1591–1595

directly relevant to influences of the lipid environment on integral membrane proteins, such as the Na,K-ATPase which is considered below. When sufficiently above the transition, the temperature dependence of the atomic displacements and other dynamic processes in the various proteins conforms with the Arrhenius law [3]. It is readily seen that Eq. (4) becomes simply: 〈α 2〉 ≈ A exp(− BTo/T), for T ≫ To. Fig. 2 shows Arrhenius plots of the temperature dependence of the mean-square librational amplitude, bα 2>, of the solvated spinlabelled proteins and of the proteins in glycerol–water. In this higher temperature regime, they are reasonably well described with a single activation energy, Ea (cf. Eq. (2)). Such behaviour is expected if the librational motion is characterised by activated random jumps of sufficiently small amplitude, αo, as to be otherwise unrestricted (cf. Ref. [3]). Table 1 lists the activation energies that are deduced for the spin-labelled proteins in Fig. 2. The values of Ea are in the range estimated from the mean-square displacement, br 2>, in neutron scattering or Mössbauer studies on solvated proteins and correspond to that for relaxation processes in glass-forming media and the hydration shells of proteins [3,27–29]. From measurements of br 2> [3] and dielectric relaxation [29], activation energies of Ea = 35 and ≈ 30 kJ/mol, respectively, are deduced for myoglobin. For comparison, the activation energies associated with the librational motion of TEMPONE and 5-MSL alone in glass-forming 60% aqueous glycerol are 34 ± 4 and 49 ± 3 kJ/mol, respectively (data not shown). A comparable value of 30 ± 1 kJ/mol has also been found for a glycerol–water mixture with a glass transition temperature of Tg = 177 K [25]. Although the nature of a molecular process cannot be derived from the activation energy alone, it is striking that measurements from proteins and glass systems, and from the different diffusive processes registered by the various physical methods, correlate so well in their magnitudes.

4. Comparison with previous EPR results on soluble proteins Earlier EPR studies on carbonmonoxy haemoglobin [30], myoglobin and lysozyme [31], and on methaemoglobin [32], each spin-labelled with 6-MSL, have found similar low-temperature behaviour to that reported here for oxyHb, β-LG and HSA. In each case, a discontinuity in temperature dependence of the outer hyperfine splitting is found at around 200−210 K for the hydrated proteins, but not for the

<α2> (rad2)

0.01

1E-3 3.3

4 4.0

4. 4.5

5.0

Table 1 Activation energies, Ea, characterising the temperature dependence of the mean-square librational amplitude, bα2>, in the diffusive regime for different proteins (see Figs. 2 and 3). Protein

Spin label

Medium

Ea (kJ mol−1)

Hb Hb β-LG β-LG HSA Na,K-ATPase, shark Class I Na,K-ATPase, shark Class II Na,K-ATPase, kidney Class I Na,K-ATPase, kidney Class II

6-MSL 6-MSL 5-MSL 5-MSL 5-MSL 5-MSL 5-MSL 5-MSL 5-MSL

Water 60% v/v glycerol Water 60% v/v glycerol Water Water Water Water Water

19.2 29.5 22.1 41.6 24.7 29.2 31.1 31.1 31.7

a

± ± ± ± ± ± ± ± ±

0.7 3.0a 1.7 6.8 1.8 1.5 2.5 2.1 1.6

Omitting T ≥ 270 K.

lyophilized proteins. Parallels with the results from Mössbauer spectroscopy were drawn in Ref. [31]; spin-label data for myoglobin from this reference, analysed by the methods of Fig. 2, yield an activation energy of 20 ± 2 kJ mol−1 over a similar temperature range. It was suggested in Ref. [32] that, over the range 200−260 K, the enhanced temperature dependence of spin-labelled metHb could be explained by a temperature-dependent relaxation of the librational potential well; analysis by the present model yields an activation energy in the region of 18 ±3 kJ mol−1 in this range. An alternative interpretation in terms of slow 60°-jumps with correlation times 4−20 ns, over the range 260−220 K, was also suggested for the metHb data [32]. Our own data with time-resolved EPR, however, indicate correlation times extending into the sub-nanosecond range for a variety of spin-labelled proteins, including Hb, in this temperature range [6,20]. In Ref. [30], the earliest of these studies, the temperature dependence for HbCO was interpreted in terms of hydrogen bonding, rather than angular libration. However, it was found subsequently that the spectral changes in the range 200−260 K cannot be attributed to this type of polarity effect and must be dynamic in origin [33]. Application of the present analysis yields an activation energy of 16−22 kJ mol−1 also from this data for HbCO. Thus each of the hydrated proteins studied yields temperature dependences and effective activation energies in the region above 200 K and below the freezing point of bulk water that are consistent with the present interpretation. It is interesting to note that these values of Ea are comparable to the enthalpy of H-bond formation with nitroxides, possibly implicating hydrogen bonding with the protein hydration shell.

5. Na,K-ATPase

Hb/H2O Hb/60% glycerol HSA/H2O β-LG/H2O β-LG/60% glycerol

0.1

1593

5. 5.5

103/T(K) Fig. 2. Arrhenius plots characterising the temperature dependence of the librational amplitude, bα2>, in the diffusive regime for 6-MSL-labelled Hb in water (solid circles) and 60% v/v glycerol–water (open circles); 5-MSL-labelled β-LG in water (solid squares) and 60% v/v glycerol–water (open squares); and 5-MSL-labelled HSA in water (triangles). Solid lines are linear regressions.

In contrast to the small water-soluble proteins discussed above, the Na,K-ATPase is a large integral membrane protein that is composed of 12 transmembrane α-helices, in addition to a sizeable globular cytoplasmic domain that bears the nucleotide binding site and catalytic centre. The protein contains several cysteine residues that can be spin-labelled with maleimide derivatives [34,35]. One superficial population of \SH groups (Class I) can be labelled in the presence of glycerol and ATP to produce a fully active enzyme, whereas a second population (the Class II groups) is essential for activity and labelled only in the absence of glycerol and protecting ligands [36]. Fig. 3 shows the temperature dependence of the mean-square librational amplitude for membranous Na,K-ATPase, spin-labelled on either Class I or Class II \SH groups and suspended in aqueous buffer, in the form of an Arrhenius plot. Data are shown both for the enzyme from the salt gland of shark, which has a low physiological temperature, and for the mammalian enzyme from pig kidney. In each case, the librational amplitude is reduced to a very small level below a characteristic temperature in the region of 200 K [6]. Additionally, the temperature dependence in the diffusive regime above

1594

D. Marsh et al. / Biochimica et Biophysica Acta 1834 (2013) 1591–1595

0.1

References

<α2> (rad2)

Shark I Shark II Kidney I Kidney II 0.01

1E-3

3.6

3.8

4.0

4.2

4.4

4.6

4.8

5.0

103 /T(K) Fig. 3. Temperature dependence of the mean-square amplitude of librational motion, bα2>, for Na,K-ATPase from shark salt gland (squares) or porcine kidney (circles), spin-labelled with 5-MSL on either Class I (solid symbols) or Class II (open symbols) \SH groups. Solid lines are linear regressions for the Arrhenius plots.

200 K can be described by the Arrhenius law (Fig. 3). The activation energies deduced from Fig. 3 are listed in Table 1. The values of the activation energy are similar for the different Na, K-ATPase systems, and are comparable to those obtained for the water-soluble proteins in 60% aqueous glycerol and the small spin labels in the same glass-forming solvent. It appears that the librational motions in this large and complex membrane transport enzyme resemble more closely those of the smaller proteins coupled to the glass-forming solvents (in a Kramers fashion) than to the soluble proteins in water alone. Interesting is the similarity between Class I groups and Class II groups of the Na,K-ATPase. Presumably, the less accessible Class II groups are coupled to the aqueous environment via the cytoplasmic domain which contains the superficial Class I groups. A similar coupling to the aqueous environment must be mediated by the surface polar groups of bacteriorhodopsin, the dynamic transition of which has been investigated by neutron scattering techniques [12]. In this connection, it is interesting to note that the different enzymatic reactions of the Na,K-ATPase are found to depend strongly on viscosity of the suspending medium [37].

6. Conclusion To conclude: spin-label EPR is a useful technique for investigating the glass-like properties of a wide variety of enzymes and proteins, including membrane proteins. Whereas discussion of the exact nature of the dynamic transition and glass-like behaviour of protein systems may still be in a state of flux and development (see, e.g., Refs. [9,38]), it is worthwhile to note that that the treatment given above is based solely on correlations between experimental temperature dependences, and does not rely for its validity on a specific theoretical model. Particularly with regard to the latter point, a reviewer of this paper remarks that the glass transition refers only to the solvent and its effect on the protein, and the glass temperature Tg is always lower than Td, the dynamical transition temperature, because different time scales are involved. The time scale relevant to the EPR experiment is discussed above. Recent data, on the other hand, suggest that the protein dynamical transition is a real physical effect and not solely limited by instrumental resolution [38]. The same reviewer further remarks that the dynamical transition was originally introduced as a two-step process: a fast, amplitude-controlled motion emerging above Tg, which then acts as a precursor to the more collective α-relaxation and diffusion occurring on a longer time scale.

[1] I.E.T. Iben, D. Braunstein, W. Doster, H. Frauenfelder, M.K. Hong, J.B. Johnson, S. Luck, P. Ormos, A. Schulte, P.J. Steinbach, A.H. Xie, R.D. Young, Glassy behavior of a protein, Phys. Rev. Lett. 62 (1989) 1916–1919. [2] D. Ringe, G.A. Petsko, The ‘glass transition’ in protein dynamics: what it is, why it occurs, and how to exploit it, Biophys. Chem. 105 (2003) 667–680. [3] P.W. Fenimore, H. Frauenfelder, B.H. McMahon, R.D. Young, Bulk-solvent and hydration-shell fluctuations, similar to α- and β-fluctuations in glasses, control protein motions and functions, Proc. Natl. Acad. Sci. U.S.A. 101 (2004) 14408–14413. [4] H. Frauenfelder, G. Chen, J. Berendzen, P.W. Fenimore, H. Jansson, B.H. McMahon, I.R. Stroe, J. Swenson, R.D. Young, A unified model of protein dynamics, Proc. Natl. Acad. Sci. U.S.A. 106 (2009) 5129–5134. [5] H. Frauenfelder, F. Parak, R.D. Young, Conformational substates in proteins, Ann. Rev. Biophys. Biophys. Chem. 17 (1988) 451–479. [6] R. Guzzi, R. Bartucci, L. Sportelli, M. Esmann, D. Marsh, Conformational heterogeneity and spin-labelled \SH groups: pulsed EPR of Na,K-ATPase, Biochemistry 48 (2009) 8343–8354. [7] G. Biroli, J.P. Garrahan, Perspective: the glass transition, J. Chem. Phys. 138 (2013), (12A301). [8] W. Doster, S. Cusack, W. Petry, Dynamical transition of myoglobin revealed by inelastic neutron-scattering, Nature 337 (1989) 754–756. [9] W. Doster, The protein-solvent glass transition, Biochim. Biophys. Acta 1804 (2010) 3–14. [10] F. Parak, E.W. Knapp, D. Kucheida, Protein dynamics — Mossbauer spectroscopy on deoxymyoglobin crystals, J. Mol. Biol. 161 (1982) 177–194. [11] F. Parak, H. Hartmann, K.D. Aumann, H. Reuscher, G. Rennekamp, H. Bartunik, W. Steigemann, Low-temperature X-ray investigation of structural distributions in myoglobin, Eur. Biophys. J. Biophys. Lett. 15 (1987) 237–249. [12] M. Ferrand, A.J. Dianoux, W. Petry, G. Zaccai, Thermal motions and function of bacteriorhodopsin in purple membranes — effects of temperature and hydration studied by neutron-scattering, Proc. Nat. Acad. Sci. U.S.A. 90 (1993) 9668–9672. [13] A.M. Tsai, D.A. Neumann, L.N. Bell, Molecular dynamics of solid-state lysozyme as affected by glycerol and water: a neutron scattering study, Biophys. J. 79 (2000) 2728–2732. [14] J.H. Roh, V.N. Novikov, R.B. Gregory, J.E. Curtis, Z. Chowdhuri, A.P. Sokolov, Onsets of anharmonicity in protein dynamics, Phys. Rev. Lett. 95 (2005) 038101. [15] R. Guzzi, M. Babavali, R. Bartucci, L. Sportelli, M. Esmann, D. Marsh, Spin-echo EPR of Na, K-ATPase unfolding by urea, Biochim. Biophys. Acta 1808 (2011) 1618–1628. [16] S.P. Van, G.B. Birrell, O.H. Griffith, Rapid anisotropic motion of spin labels. Models for motion averaging of the ESR parameters, J. Magn. Reson. 15 (1974) 444–459. [17] R. Bartucci, R. Guzzi, M. De Zotti, C. Toniolo, L. Sportelli, D. Marsh, Backbone dynamics of alamethicin bound to lipid membranes: spin-echo EPR of TOAC spin labels, Biophys. J. 94 (2008) 2698–2705. [18] A. Einstein, The motion of elements suspended in static liquids as claimed in the molecular kinetic theory of heat, Ann. Phys. 17 (1905) 549–560. [19] W. Doster, The dynamical transition of proteins, concepts and misconceptions, Eur. Biophys. J. 37 (2008) 591–602. [20] F. Scarpelli, R. Bartucci, L. Sportelli, R. Guzzi, Solvent effect on librational dynamics of spin-labelled haemoglobin by ED- and CW-EPR, Eur. Biophys. J. 40 (2011) 273–279. [21] J.D. Nickels, H. O'Neill, L. Hong, M. Tyagi, G. Ehlers, K.L. Weiss, Q. Zhang, Z. Yi, E. Mamontov, J.C. Smith, A.P. Sokolov, Dynamics of protein and its hydration water: neutron scattering studies on fully deuterated GFP, Biophys. J. 103 (2012) 1566–1575. [22] M.H. Cohen, D. Turnbull, Molecular transport in liquids and glasses, J. Chem. Phys. 31 (1959) 1164–1169. [23] M.D. King, D. Marsh, Free volume model for lipid lateral diffusion coefficients. Assessment of the temperature dependence in phosphatidylcholine and phosphatidylethanolamine bilayers, Biochim. Biophys. Acta 862 (1986) 231–234. [24] H.A. Kramers, Brownian motion in a field of force and the diffusion model of chemical reactions, Physica 7 (1940) 284–304. [25] S.V. Paschenko, Y.V. Toropov, S.A. Dzuba, Y.D. Tsvetkov, A.K. Vorobiev, Temperature dependence of amplitudes of libration motion of guest spin-probe molecules in organic glasses, J. Chem. Phys. 110 (1999) 8150–8154. [26] D.A. Erilov, R. Bartucci, R. Guzzi, D. Marsh, S.A. Dzuba, L. Sportelli, Librational motion of spin-labeled lipids in high-cholesterol containing membranes from echo-detected EPR spectra, Biophys. J. 87 (2004) 3873–3881. [27] P.W. Fenimore, H. Frauenfelder, B.H. McMahon, R.D. Young, Proteins are paradigms of stochastic complexity, Physica A 351 (2005) 1–13. [28] G.U. Nienhaus, H. Frauenfelder, Structural fluctuations in glass-forming liquids: Mössbauer spectroscopy on iron in glycerol, Phys. Rev. B 43 (1991) 3345–3350. [29] G.P. Singh, F. Parak, S. Hunklinger, K. Dransfeld, Role of adsorbed water in the dynamics of metmyoglobin, Phys. Rev. Lett. 47 (1981) 685–688. [30] M.E. Johnson, Apparent hydrogen bonding by strongly immobilized spin labels, Biochemistry 20 (1981) 3319–3328. [31] J. Ruggiero, R. Sanches, M. Tabak, O.R. Nascimento, Motional properties of spin labels in proteins: effects of hydration, Can. J. Chem. 64 (1986) 366–372. [32] H.J. Steinhoff, K. Lieutenant, J. Schlitter, Residual motion of hemoglobin-bound spin labels as a probe for protein dynamics, Z. Naturforsch. 44c (1989) 280–288. [33] H.J. Steinhoff, A simple method for determination of rotational correlation times and separation of rotational and polarity effects from EPR spectra of spin-labeled

D. Marsh et al. / Biochimica et Biophysica Acta 1834 (2013) 1591–1595 biomolecules in a wide correlation time range, J. Biochem. Biophys. Methods 17 (1988) 237–248. [34] M. Esmann, K. Hideg, D. Marsh, Conventional and saturation transfer EPR spectroscopy of Na+/K(+)-ATPase modified with different maleimide-nitroxide derivatives, Biochim. Biophys. Acta 1159 (1992) 51–59. [35] M. Esmann, K. Hideg, D. Marsh, Analysis of thiol-topography in Na, K-ATPase using labelling with different maleimide nitroxide derivatives, Biochim. Biophys. Acta 1112 (1992) 215–225.

1595

[36] M. Esmann, Sulfhydryl groups of (Na+ + K+)-ATPase from rectal glands of Squalus acanthias — titration and classification, Biochim. Biophys. Acta 688 (1982) 251–259. [37] M. Esmann, N.U. Fedosova, D. Marsh, Osmotic stress and viscous retardation of the Na, K-ATPase ion pump, Biophys. J. 94 (2008) 2698–2705. [38] G. Schiró, F. Natali, A. Cupane, Physical origin of anharmonic dynamics in proteins: new insights from resolution-dependent neutron scattering on homomeric polypeptides, Phys. Rev. Lett. 109 (2012) 128102-1–128102-5.