Biochimica et Biophysica Acta 1830 (2013) 4564–4572 Contents lists available at SciVerse ScienceDirect Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbagen Hofmeister ions control protein dynamics Balázs Szalontai a, Gergely Nagy b,c,1, Sashka Krumova d, Elfrieda Fodor e, Tibor Páli a, Stefka G Taneva d,2, Győző Garab f, Judith Peters c,3, András Dér a,⁎ a Institute of Biophysics, Biological Research Centre of the Hungarian Academy of Sciences, H-6726 Szeged, Temesvári krt 62, Hungary Wigner Research Centre for Physics, Institute for Solid State Physics and Optics, Hungarian Academy of Sciences, P.O Box 49, H-1215 Budapest, Hungary Institut Laue–Langevin, BP 156, 38042 Grenoble Cédex 9, France d Institute of Biophysics and Biomedical Engineering, Bulgarian Academy of Sciences, Acad G Bonchev Str., Bl.21, 1113 Sofia, Bulgaria e Institute of Biochemistry, Biological Research Centre of the Hungarian Academy of Sciences, H-6726 Szeged, Temesvári krt 62, Hungary f Institute of Plant Biology, Biological Research Centre of the Hungarian Academy of Sciences, H-6726 Szeged, Temesvári krt 62, Hungary b c a r t i c l e i n f o Article history: Received 16 April 2013 Received in revised form 24 May 2013 Accepted 28 May 2013 Available online June 2013 Keywords: Hofmeister effect Differential scanning calorimetry Neutron scattering Fourier transform infrared spectroscopy Protein structural fluctuation Bacteriorhodopsin a b s t r a c t Background: Recently, we have elaborated a thermodynamic theory that could coherently interpret the diverse effects of Hofmeister ions on proteins, based on a single physical parameter, the protein–water interfacial tension (Dér et al., Journal of Physical Chemistry B 2007, 111, 5344–5350) This theory, implying a “liquid drop model”, predicts changes in protein conformational fluctuations upon addition of Hofmeister salts (containing either kosmotropic or chaotropic anions) to the medium Methods: Here, we report experimental tests of this prediction using a complex approach by applying methods especially suited for the detection of protein fluctuation changes (neutron scattering, micro-calorimetry, and Fourier-transform infrared spectroscopy) Results: It is demonstrated that Hofmeister salts, via setting the hydrophobic/hydrophilic properties of the protein–water interface, control conformational fluctuations even in the interior of the typical membrane transport protein bacteriorhodopsin, around its temperature-induced, unusual α(II) → α(I) conformational transition between 60 and 90 °C We found that below this transition kosmotropic (COOCH− ), while above it chaotropic (ClO− ) anions increase structural fluctuations of bR It was also shown that, in each case, an onset of enhanced equilibrium fluctuations presages this phase transition in the course of the thermotropic response of bR Conclusions: These results are in full agreement with the theory, and demonstrate that predictions based on protein–water interfacial tension changes can describe Hofmeister effects and interpret protein dynamics phenomena even in unusual cases General significance: This approach is expected to provide a useful guide to understand the principles governing the interplay between protein interfacial properties and conformational dynamics, in general © 2013 Elsevier B.V All rights reserved Introduction Hofmeister effects (HEs) cover a wide range of salt-induced phenomena on proteins, and on colloidal particles in general, including changes of solubility, denaturation, or enzyme kinetics In his original ⁎ Corresponding author at: Institute of Biophysics, Biological Research Centre, Hungarian Academy of Sciences, H-6726 Szeged, Temesvári krt 62., Hungary Tel.: +36 62 599 606; fax: +36 62 433 133 E-mail addresses: szalontai.balazs@brc.mta.hu (B Szalontai), nagy@ill.fr, jpeters@ill.fr (G Nagy), sakrumo@gmail.com (S Krumova), fffrida7@gmail.com (E Fodor), stefka.germanova@ehu.es (S.G Taneva), der.andras@brc.mta.hu (A Dér) Present address: Paul Scherrer Institute, Laboratory for Neutron Scattering, 5232 Villigen PSI, Switzerland Present address: Unidad de Biofísica (CSIC-UPV/EHU) and Departamento de Bioquímica y Biología Molecular, Universidad del País Vasco, 48080 Bilbao, Spain Present address: University Joseph Fourier, UFR PhITEM, BP 53, 38041 Grenoble Cédex 9, France; Institut de Biologie Structurale, 41 rue Jules Horowitz, 38027 Grenoble Cédex 1, France 0304-4165/$ – see front matter © 2013 Elsevier B.V All rights reserved http://dx.doi.org/10.1016/j.bbagen.2013.05.036 work, Hofmeister [1] reported that certain salts (primarily their anions) either decrease or increase protein solubility when present in the solution at moderate or high concentrations (usually above 100 mM), and arranged them into the so-called Hofmeister series (HS) according to the sign and the magnitude of the effect they exerted The HS of the most important anions, in descending order of their precipitat> HPO2− > CH3COO− > Cl− > ing ability, is as follows: F− ≈ SO2− 4 − − − − − − NO3 > Br > ClO3 > I > ClO4 > SCN Members in the series left of the “Hofmeister-neutral” Cl− are precipitants, while the ones right of it are solubilizers An impressive number of follow-up studies show that the same HS emerges in other phenomena, like denaturation of proteins, inhibition or activation of enzymes, as well (for reviews, see [2–4]) HEs live their renaissance in the past few years Besides their significance in colloid chemistry, preparative biochemistry and biotechnology, they have been successfully applied in studying the function and dynamics of macromolecular structures [5,6] In addition, a recent review calls the attention to their role in basic pathophysiologic issues [7] B Szalontai et al / Biochimica et Biophysica Acta 1830 (2013) 4564–4572 Despite the widespread occurrence of HEs, and the extensive research efforts focused on them, their interpretation has remained a matter of debate [8], because of the lack of a unifying formalism covering the entire spectrum of salts from “salting out” to “salting in” effects, i.e from precipitants to solubilizers One approach [9] correlated these attributes with effects on water structure, in particular on the fraction of hydrogen-bonded water molecules: precipitants promote H-bond formation between the water molecules in their vicinity, and are therefore called kosmotropes, while solubilizers break H-bonds between water molecules, and are therefore called chaotropes In another branch of interpretations, dispersion forces were suspected to be the main factor responsible for HEs [10] Although these are certainly among the main microphysical factors responsible for the Hofmeister effects, all attempts aiming to assign a single physical parameter as an approximate measure for the whole diversity of HEs, however, have failed for more than a century In a recent work, we have demonstrated that a unified, phenomenological formalism is able to qualitatively account for the entire range of HE-related phenomena [11] The crucial difference between our approach and the previous attempts of similar intention was that, instead of building on the conventionally used air–water interface [9], we took protein–water interfacial tension (γpw) as the principal physical parameter to describe Hofmeister effects The most important conclusion of our theory is that HEs are manifested via the surface-dependent term of the free energy (Gs = γpw · ASA, where ASA is the solvent accessible surface area) of the proteins (or, in general, of colloid disperse systems) [12] In other words, they modify the hydrophobic effect at protein–water interfaces [13] The addition of kosmotropic or chaotropic salts to the solvent increases or decreases the solute–water interfacial tension, respectively, and thus, Gs should increase or decrease accordingly The alteration of Gs is the driving force behind the observed HEs affecting either the aggregation or the conformation of the proteins Note that γpw should, necessarily, depend also on the quality of the solvent-exposed protein surface, and may even take negative values, too (e.g., in case of proteins of naturally open conformation) [11,14] An implication of our theory is that HEs are manifested by changing transition rates and equilibrium constants in reactions accompanying major conformational changes that involve changes in the water-exposed surface area of macromolecules and supramolecular assemblies (Such effects have recently been exemplified by a detailed study of Hofmeister effects on the function of photoactive yellow protein [5].) It was also established that interfacial tension and protein stability are interconnected by protein conformational fluctuations, thus providing the keystone for the microscopic interpretation of HEs [11,15] In consequence of the above view, the same HS should appear for salt-induced protein conformational fluctuations as for precipitation or conformational changes In spite of the success of this theory in describing the diverse Hofmeister phenomena, so far direct experimental evidences for salt-induced changes of protein conformational fluctuations have been missing Hence, here we used a complex experimental approach to monitor temperature-induced changes of protein structure and dynamics, with methods specially adapted for observing changes in conformational fluctuations (neutron scattering, differential scanning calorimetry (DSC), and Fourier transform infrared spectroscopy (FTIR)) Our model object was the prototypical retinal protein, bacteriorhodopsin (bR) from Halobacterium salinarum [16] Also known as the simplest ion pump in biological systems, bR is one of the best-characterized and most robust membrane proteins, subjected to non-specific anion binding effects Despite, both its structure and function have been shown to be influenced by interactions with Hofmeister anions [11,17] In the present study, special attention was paid to the reversible conformational change of bR in the course of its heat denaturation, the α(II) → α(I) conformational transition between 60 and 90 °C, where enhanced changes of structural 4565 fluctuations were expected to arise [18,19], due to the associated alterations of the solvent-exposed surface So far, there have been only a few reports about proteins having an “open” conformation as their naturally most stable form [3], and bR is considered to be such one Namely, previous FTIR measurements on bR detected α(II) helices below ca 50 °C [18], and between 60 and 90 °C, bR undergoes an α(II) → α(I) conformational transition [19] In α(I), however, the helix is known to be more tightly packed than in α(II) [18,20] It implies that, in contrast to most proteins, here the accessible surface area (ASA) of the protein is higher at lower temperatures (α(II)), and ASA decreases upon the temperatureinduced transition of the secondary structure (to α(I)) For such a case, our theory [11] predicts an unusual phenomenon to occur: instead of kosmotropic, here chaotropic salts should have a stabilizing effect; consequently, they are expected to shift the α(II) → α(I) transition to higher temperatures, while kosmotropes are expected to destabilize the open α(II) protein conformation, therefore shifting the transition toward lower temperatures, as it has been indicated by our former DSC experiments [11] We took advantage of this unusual feature of bR to test the predictive power of our HE-interpreting theory Materials and methods 2.1 Bacteriorhodopsin-containing purple membrane (bR) isolation Purple membranes were isolated according to the standard procedure [21] Purple membranes were always re-suspended in D2O-based buffers, due to the needs of neutron scattering and FTIR experiments (10 mM HEPES in D2O at pD 6.6, containing 500 mM kosmotropic, Hofmeister-neutral, or chaotropic salts, NaCOOCH3, NaCl, or NaClO4, respectively, as required) The bR samples, treated with these salts will be denoted as bR-NaCOOCH3, bR-NaCl, and bR-NaClO4 throughout the paper 2.2 Neutron scattering experiments For neutron scattering experiments the suspensions of bR-containing purple membranes were centrifuged for 20 at 40,000 g The pellets were placed in flat rectangular aluminum sample holder with an area (30 × 40 mm2) adapted to the dimensions of the incident neutron beam The sample holders were hermetically closed and used for the experiment The experiments were carried out on the IN13 backscattering spectrometer at the Institut Laue Langevin (ILL, France) Two samples were measured by elastic incoherent neutron scattering to determine the atomic mean-square displacements bu2> as a function of temperature One of the samples was bR-NaClO4, measured in the 40–91 °C temperature range, the other one was bR-NaCOOCH3 measured between 40 and 87 °C Temperature scans were done upon increasing the temperature in steps of 7–10 °C below the transition region, and of 3–4 °C around the transition region Additional vanadium, buffer and empty cell measurements were recorded for correction and normalization purposes [22] In the examined time range (8 μeV energy resolution, corresponding to a time window of 100 ps through Heisenberg's uncertainty principle) on backscattering spectrometers, H-atom motions reflect the motions of the chemical groups to which they are bound [23] The scattered elastic incoherent intensity can be described within the Gaussian approximation [24] by D 2E u Q Iel Q ; ẳ EịI0 exp ð1Þ where I0 is the intensity hitting the sample, bu2> is the average atomic mean square displacement, ω corresponds to the energy transfer, and Q to the momentum transfer between neutron and target The average 4566 B Szalontai et al / Biochimica et Biophysica Acta 1830 (2013) 4564–4572 value of the atomic mean square displacements (MSD) for the protein can thus be obtained for each temperature from the slope of the semi-logarithmic plot of the incoherent scattering function through D u E ¼ −6 d lnIel ðQ ; ϖ ¼ ΔEÞ : dQ ð2Þ The data treatment, including subtraction of the scattering from the empty sample holder, normalization to vanadium and absorption correction based on the correction formula of Paalman–Pings coefficients [25] was performed using the ILL program LAMP (Large Array Manipulation Program [26]) 2.3 Differential scanning calorimetry experiments Differential scanning calorimetry (DSC) is a useful tool to characterize the thermodynamics of protein conformational changes [27], and has been applied to monitor the steps of thermal denaturation of bR, too [11,28,29] Direct information about the specific heat of the protein (Cp) can be derived from the experiments, and Cp is instantaneously related to enthalpy fluctuations, according to the fluctuation-dissipation theorem [27]: Cp ¼ b δH > : kT stretching vibration with smaller contributions from CC, NC stretching vibrations, and CO in-plane bend [30] To keep the Amide I region free from the disturbing strong contribution of the H2O-bending band at 1643 cm−1, and being in agreement with the neutron scattering and DSC experiments, all infrared spectra were recorded in D2O environment The presence of D2O induces the exchange of the peptide \NH groups to \ND ones, which affects the Amide I and II regions differently The H/D isotope effect is weak in the Amide I region, since N\H vibrations have little contribution to these modes, as discussed above (the shape of the Amide I region does not change considerably, only the frequencies of the component bands shift down by a few wavenumbers; the deuterated Amide I band will be indicated throughout the paper as Amide I′) Temperature-dependent changes in the Amide I′ remain characteristic to the changes of the secondary structure elements of the protein To monitor protein dynamics, we used the Amide II region, where H/D exchange exerts a major effect, since the Amide II region downshifts to about 1450 cm−1 (Amide II′) Therefore, the Amide II band (around 1550 cm−1) disappears upon H/D exchange The rate of this disappearance is informative about protein structural fluctuations [31,32], since the access of the external D atoms to the peptide \NH groups depends on the protein dynamics under the given environmental factors (temperature, molecular interactions) ð3Þ DSC measurements were performed on bR membrane suspensions of mg·mL−1 protein concentrations, by using a VP-DSC high-sensitivity differential scanning calorimeter (MicroCal Inc., Northampton, MA) equipped with built-in twin reference and sample cells of 0.5 mL volume Temperature scans were performed in the 10–110 °C range, at a heating rate of 0.5 °C·min−1, and the differential heat capacities were recorded At the end of the heating scans samples were cooled back to 10 °C where they were allowed to equilibrate for 30 before being subjected to a second heating scan with the same measuring protocol These re-heating scans served later as baselines for the first recordings since they did not show any transition peak present in the above mentioned temperature range Positions of the pre- and main transition peaks were determined by using the built-in integration routine of the DSC analysis software (MicroCal-Origin, DSC application) 2.4 Infrared measurements FTIR spectra were recorded on a Philips PU9800 Fourier transform infrared spectrometer, averaging 128 scans at cm−1 spectral resolutions For the infrared experiments, 1.5 mL aliquots of bR-containing purple membranes dispersed in the required D2O-based buffer + Hofmeister salt, were concentrated in a Hettich (Germany) EBA 12R tabletop centrifuge (4 °C, 24,000 ×g, min) The pellet (about 20 μL) was placed between CaF2 windows separated by an aluminum spacer (15 μm) Temperature dependence of the infrared absorption spectra was measured by recording 35 × FTIR spectra between and 95 °C in about °C steps, leaving for reaching thermal equilibrium At each temperature, two absorption spectra were measured by recording for each single beam background (empty spectrometer) and sample spectrum (CaF2 window plus purple membrane dispersion) by using a sample shuttle A water-thermostated cell holder controlled the temperature of the sample The accuracy of the temperature setting was about 0.1 °C In water suspensions of most biological membranes, the Amide I band is centered at around 1650 cm−1, arising mainly from the C_O stretching vibration (with minor contributions from the out-of-phase C\N stretching vibration, the C\C\N deformation and the N\H in-plane bend) The Amide II mode near 1550 cm−1 is the out-of-phase combination of the NH in-plane bend and the CN 2.4.1 Singular value decomposition (SVD) analysis Temperature-dependent changes in the infrared spectra were studied by SVD analysis [33] Briefly, the D (n ∗ m) data matrix was decomposed as D = S ∗ W ∗ VT, where n is the number of wavelengths, and m is the number of temperatures at which the measurements were made S (n ∗ m) contains the orthonormal abstract spectrum vectors; the W (m ∗ m) diagonal matrix consists of the singular values in descending order, while V (m ∗ m) contains the orthonormal vectors describing the temperature dependence The s1 base spectrum (the first column of the S matrix) shows the average spectrum over the temperature range involved; s2 shows the largest changes accompanying the temperature increase, which have to be combined with the s1 spectrum to get back the actual spectrum at a given temperature The v1 and v2 amplitude vectors give the temperature dependence of the average, and the largest change, respectively, manifested in the s1, s2 spectra, etc In the present SVD analysis, we considered only the average (s1, v1) and the largest change (s2, v2) of the spectra recorded upon increasing temperature (5–95 °C) The higher order SVD components (s3, v3, s4, v4…, etc.) were neglected because their weights were low This approach is evidently an approximation, which, however, can be justified since the weights (wi) of the si base spectra were decreasing very rapidly (to about 2–3% of w1 by w3, data not shown) All calculations were carried out with the SPSERV© software of Cs Bagyinka (Biological Research Centre, Szeged, Hungary) For more details on the use of SVD analysis on biological membrane proteins, see our earlier work on dynamics and temperature-induced denaturation of proteins in thylakoid membranes [34] Results 3.1 Neutron scattering experiments Among the three applied methods, neutron scattering gives the most direct information about conformational fluctuations It probes mainly the mean square displacements (MSD) of hydrogen atoms in biological samples, since the incoherent cross-section of H is much larger than that of any other atom in the samples Since H atoms are usually homogeneously distributed in biological samples, the MSD are assumed to reflect the global dynamics of the system as well B Szalontai et al / Biochimica et Biophysica Acta 1830 (2013) 4564–4572 The bR-containing purple membranes consist of two types of molecules, bR and the lipid components of the membranes, therefore, in principle, we should see in the scattering experiments contributions from both lipids and bR Nevertheless, by choosing the temperature range where bR, but not the lipid, is expected to have a major structural transition, and taking into account the 75:25 weight ratio of bR to lipids in purple membranes, one can assume that the changes seen in H-dynamics are mostly due to H-atoms of bR, and that lipid-related H atoms provide, at most, a smoothly changing background A further factor to be taken into account is the H/D exchange at the \NH groups of the solvent accessible peptide bonds This means that especially at higher temperatures, the decreasing population of non-D2O-accessible peptide groups, which are either in the interior or at the lipid interface of bR, will have the major contribution to the neutron scattering signals, i.e., this method monitors the least accessible, internal parts of the protein Bacteriorhodopsin dynamics has been studied by incoherent neutron scattering over the past decades [35–39] In these experiments, two populations of motions have been found to arise when investigating the thermotropic behavior of fully hydrated bR with neutron scattering They were assigned to dynamical transitions around −123 °C and −23 °C, where onsets of additional conformational motions occur [37] In the present case, however, we measured at much higher temperatures (in the 40–91 °C range), in order to monitor Hofmeister effects on protein fluctuations associated with the α(II) → α(I) transition of bR Neutron scattering experiments were carried out on bR-containing purple membranes suspended in D2O-based kosmotropic acetate and chaotropic perchlorate salt solutions (500 mM NaCOOCH3 and NaClO4, respectively) In Fig 1, MSD values of hydrogen atoms are plotted as a function of temperature They are associated with the sample flexibility via the fluctuation-dissipation theorem [40] At higher temperatures, where the solvent of the protein is fluid, the elastically scattered intensity decreases drastically, as the motions are leaving the instrumental time window This fact leads to high errors (compare our errors bars with those in Ref [41]) of the calculated MSD values Nevertheless, a break in Fig MSDs as the function of temperature for bR-containing purple membranes suspended in D2O-based 0.5 M kosmotropic (NaCOOCH3) and chaotropic (NaClO4) Hofmeister salt solutions Straight lines represent linear fits used to extract the force constants bk> A change in the slope of the MSD values can be observed at around 75 °C for bR-NaCOOCH3, and at around 80 °C for bR-NaClO4 The lower conformational flexibility below the transition and the higher transition temperature in bR-NaClO4 compared to bR-NaCOOCH3 suggest a higher dynamical stability for the chaotrope-treated sample as compared to the kosmotrope-treated one 4567 the slope of the MSD(T) function for kosmotrope-treated bR-NaCOOCH3 around 75 °C, and for the chaotrope-treated bR-NaClO4 around 80 °C is clearly visible The mean force constants, which are associated with the stability of the whole sample, can be separately calculated for the two formula [36,42] The temperature regions according to the hki ¼ d0:00276 hu2 i=dT fitted straight lines and their slopes of the obtained MSD values and the force constants derived from their slopes are shown in Fig Note, that the chaotropic sample below the transition has a higher mean force constant (a lower slope of the fitted curve) associated with a lower flexibility as compared to that of the kosmotropic sample Above the transition, however, the calculated 〈k〉 is slightly lower for the chaotropic sample than for the kosmotropic one, but the difference is within the error of the measurements In other words, below the transition the resistance against structural fluctuations is higher (i.e., the conformational flexibility is lower) in bR-NaClO4 than in bR-NaCOOCH3, while above the transition, they are about the same in the two cases (or might even be oppositely related) This implies a higher “dynamical stability” of bR-NaClO4 than that of bR-NaCOOCH3, being also reflected by the up-shift of the transition temperature of bR by the chaotropic salt 3.2 Calorimetric measurements In order to correlate fluctuation changes with thermodynamic transitions, complementary differential scanning calorimetry (DSC) experiments were performed under similar ambient conditions (i.e., in D2O-based media) For the DSC measurements, however, the set of salt solutions was extended by the Hofmeister-neutral chloride, in order to allow a full comparison with results obtained earlier with H2O-based salts [11], too Fig shows that the main unfolding transition peak of the Cp(T) curve is located at around 100 °C, slightly depending on the particular Hofmeister salt, while a minor peak (the so-called “pre-transition”) Fig DSC heating endotherms of bR-containing purple membranes suspended in D2O-based solutions of 0.5 M of various Hofmeister salts Temperature profiles of excess molar heat capacities from top to bottom correspond to samples in the presence of chaotropic NaClO4 (red), neutral NaCl (green), and kosmotropic NaCOOCH3 (blue), respectively The two vertical lines indicate the temperature range, in which the α(II) → α(I) transition of bR takes place in the various Hofmeister salts The numbers indicate the critical temperatures (Tc) associated to the reversible α(II) → α(I) transition (“pre-transition”) and the irreversible (denaturing) maxima The scanning rate was 0.5 °C·min−1 The data were subjected to baseline subtraction by using the second heating scans as baselines, and were normalized to the corresponding protein concentrations For better visibility, the NaClO4 and NaCOOCH3 curves were up- and downshifted, respectively 4568 B Szalontai et al / Biochimica et Biophysica Acta 1830 (2013) 4564–4572 appears in the 75-90 °C range In all cases, transitions could not be detected in the second heating scans, indicating that during the first scan a complete and irreversible heat-denaturation of the protein occurred This indicates that the main transition peak of the Cp(T) curve most probably reflects the final, irreversible denaturation step of the bR accompanied also by a decomposition of the purple membrane Therefore, in the following we not consider the analysis of this peak From our point of view, much more interesting is the minor, reversible pre-transition in the 75–90 °C range, which exhibits strong Hofmeister salt dependence This transition has been assigned to the α(II) → α(I) conformational change [19,28] The chaotropic ClO− ions shifted the critical temperature (Tc) of this transition up to ca 88 °C, as compared to the Hofmeister-neutral Cl− around 82 °C, while the kosmotropic CH3COO− ions downshifted the same transition to about 77 °C It is worth mentioning that these values are by a few degrees higher than the ones measured on non-deuterated bR [11], and that the peaks related to the α(II) → α(I) transitions are getting sharper when moving from the kosmotropic acetate to the chaotropic perchlorate (Fig 2) From the Cp(T) curves, the molar enthalpy and heat capacity changes (ΔH and ΔCp values, respectively) associated with the α(II) → α(I) transition were also determined according to Jelesarov and Bosshard [43] for bR suspended in the three different salt solutions (Table 1) The molar enthalpy changes were determined from the integral of the Cp(T) functions, and found to be the same for the three different salts within the error, implying that the main differences between the initial and product secondary structures of the protein (α(II) and α(I) states, respectively) are presumably the same in all the three cases The corresponding ΔCp values, determined from lines fitted to the Cp(T) functions before and after the transition, are close to zero within the temperature range of interest 3.3 Fourier transform infrared (FTIR) experiments To reveal the sequence and the ion dependence of the secondary structure changes in bR, FTIR experiments were carried out on different Hofmeister salt-treated bR samples The salt content of the FTIR samples was the same as those of used in the neutron scattering and DSC measurements The Amide I′ and Amide II regions of the infrared spectra, characteristic of the secondary structure and dynamics of proteins, were recorded and analyzed The sequence of the secondary structure changes (from α(II) through α(I) to β), could be examined by the temperature dependence of the evolution of the related infrared absorption bands in the difference spectra, obtained by subtracting the absorption spectrum recorded at the lowest temperature from the later ones Since all the three samples behaved qualitatively similar (see Fig S1 in Supplementary data) we present, as an example, the evolution of changes in the Amide I′ region upon temperature increase only for the kosmotrope-treated bR-NaCOOCH3 (Fig 3) (The same results for the other two salts are given in Fig S2.) In the difference spectra depicted here (Fig 3A), first, a shallow valley around 1667 cm−1 is the only feature, which slowly deepens toward higher temperatures In agreement with the literature, this frequency was assigned to α(II) helices [19,28] Around 80–85 °C, the changes become more characteristic, partly due to the deepening of the 1667 cm−1 band, Table Molar enthalpy and heat capacity changes during the temperature-induced α(II) → α(I) transition of bacteriorhodopsin, and the corresponding critical absolute temperatures ΔH (kcal·mol−1) ΔCp(kcal·mol−1·K−1) Tc (°C) bR-NaOOCH3 bR-NaCl bR-NaClO4 11.4 ± 0.5 0.4 ± 0.5 77 ± 0.5 9.7 ± 3.1 −0.5 ± 0.5 82 ± 0.5 10.9 ± 0.4 −0.7 ± 0.5 88 ± 0.5 Fig Secondary structure-related changes in the Amide I′ region of bR in the presence of 0.5 M NaCOOCH3 (kosmotropic) Hofmeister salt upon increasing temperatures A — Difference absorption spectra obtained by subtracting the first infrared absorption recorded at the lowest temperature from the later ones B — The temperature dependence of the maxima of the three bands extracted from the spectra depicted in panel A Note that first α(I) starts to grow at the expense of α(II), then both α-structures diminish into the emerging β-band, which is indicative of irreversible protein denaturation and partly due to an emerging band at around 1652 cm−1, assigned to α(I) helices [44] Above 90 °C, the strong, β-structure-related band at around 1622 cm−1 dominates the difference spectra (one may observe a weaker pair-band around 1685 cm−1 as well in Fig 3A as it should be [44,45]) This pair of bands was assigned earlier to anti-parallel β-sheets [44] The appearance of this type of β-structure always accompanies the heat-induced denaturation of proteins [45] For better visualizing the sequence of events, we have plotted just the maxima of the three bands as a function of temperature in Fig 3B From Fig one may conclude that on the way to heat-induced denaturation of bR, first the α(II) structure is transformed to α(I), then α(I) (and the remaining α(II)) to β-sheets The same sequence of events could be established for bR-NaCl, and bR-NaClO4, as well, except that the corresponding spectral changes started at higher temperatures (shown in Fig S2) B Szalontai et al / Biochimica et Biophysica Acta 1830 (2013) 4564–4572 In order to give a more rigorous analysis of the similarities in the secondary structures of bR in the presence of different Hofmeister salts, and of the differences between the dynamics and stability of these structures upon heat-induced changes, we analyzed the temperature series of the recorded infrared spectra with SVD (Fig 4) (The advantage of this pure mathematical procedure is that it does not require any model for the decomposition of the measured spectra, and uses the whole data set on the full spectral and temperature range investigated [33].) The s1 and s2 base spectra provide information about the spectral changes of the samples in the investigated temperature range Fig 4A shows the s1 base spectra (corresponding to the average structure of the temperature series) derived from the Amide I′ regions of bR in the presence of different Hofmeister salts recorded between and 93 °C It can be seen that s1 spectra are very similar, i.e the majority of the secondary structure elements contributing to Fig Singular value decomposition (SVD) of the Amide I′ region of bR infrared spectra recorded between and 93 °C in different Hofmeister salts: NaClO4 — chaotropic, NaCl — neutral, and NaCOOCH3 — kosmotropic A — The s1 base spectra of the SVD analysis, i.e., the average spectra over the temperature range studied B — The s2 base spectra, i.e the largest changes associated with the temperature-induced conformational changes Note the negative band at around 1667 cm−1 assigned to α(II) helices, and the positive bands at around 1652, and 1622 cm−1, assigned to α(I) and β structures, respectively, into which the α(II) structures are transformed upon heat treatment 4569 the Amide I′ region are not affected by the Hofmeister salts (in other words, the basic structure of bR is the same in all cases) The s2 spectra (the largest changes detected by the SVD analysis, whose weights were about 7% (bR-NaClO4), 6% (bR-NaCl), and 7% (bR-NaCOOCH3) of their corresponding s1 base spectrum) are also similar (Fig 4B), meaning that the temperature-induced spectral changes can be described by the same phenomenology in all the three cases, namely α(II) (around 1667 cm−1) is the source structure of the temperature-induced changes, and it is transformed into α(I) (1652 cm−1) [19,28] and β (1622/1685 cm−1) [44,45] structures by increasing temperature (It should be noted that the s base spectra, in themselves, provide no information about the sequence of these transitions.) By the v2 vectors, the SVD analysis can reveal the temperature dependence of the spectral changes This is shown in Fig 5, for the Amide I′ and II regions of Hofmeister salt-treated bR samples The increased values of the v2 vectors between 80 and 90 °C (Fig 5) reflect the increased contribution of the corresponding s2 spectra to the spectral changes, representing the transition from the original α(II) structure to the product states, α(I) and β In the Amide I′ band, the critical temperatures, where these large-scale structural rearrangements start are 78 °C for bR-NaCOOCH3 (kosmotrope), 82 °C for bR-NaCl (HE-neutral), and 87 °C for bR-NaClO4 (chaotrope) The critical temperatures (where the slow regime ends and the fast one starts) were defined by the maxima of the second derivatives of the v2(T) curves (Note that these temperatures coincide well with the declination temperatures in the α(II) curves in Fig 3B.) For the temperature dependence of the Amide II H/D exchange (protein dynamics), the same sequence of Hofmeister salt effects was found as for the transition temperatures of the Amide I′ band In the Amide II v2 vectors, up to about 70–80 °C, there is a slow, monotonously growing H/D exchange due to the increasing thermal fluctuations in bR, and above 70–80 °C, a faster H/D exchange starts As discussed above, in the Amide II region only the disappearance of the original peak takes place; therefore, the corresponding Amide II s2 base spectra were just negative bands centered at around 1550 cm−1, and are not shown (For details, see Figs S3 and S4 in Supplementary data.) The temperature dependence of the v2 vectors of the Amide II region exhibits similar, albeit less dramatic, slope-changes as those of Fig Temperature dependence of heat-induced structural changes as reflected by the v2 vectors of the SVD analysis of the Amide I′ (Am-I) and II (Am-II) regions of the infrared absorption spectra of bR recorded in the presence of different Hofmeister salts Numbers in corresponding colors indicate the critical temperatures of the unfolding of the original α(II) forms in the presence of different Hofmeister salts 4570 B Szalontai et al / Biochimica et Biophysica Acta 1830 (2013) 4564–4572 Amide I′, but they start at 6–8 °C lower temperatures It is remarkable that the slope of the bR-NaCOOCH3 (kosmotrope) curve in Fig 5, below about 80 °C, is higher than that of the bR-NaClO4 (chaotrope) Above 80 °C, however, this relation reverses: the temperature dependence of the Amide II band disappearance (H/D exchange) is steeper in bR-NaClO4 than in bR-NaCOOCH3 (The HE-neutral bR-NaCl curve is in between the two extremes in both cases.) This means that under the critical temperature the chaotropic perchlorate, above it the kosmotropic acetate decreases the flexibility of bR Discussion Comparison of the complementary information provided by the three different experimental methods is expected to give important clues for the interpretation of Hofmeister effects by conformational fluctuations, and, thereby, for understanding the interplay between surface properties and protein dynamics, too The sequence and the character of the reactions the bR undergoes during temperature increase were revealed by FTIR and DSC measurements FTIR experiments prove that the amount of the native α(II) elements in bR decreases dramatically in the 80–90 °C temperature range as evidenced by the dramatic decrease of 1667 cm−1 band in Fig The SVD analysis also shows here the increase of the contribution of the s2 base spectra (see the steep increase of the v2 vectors in Fig 5) In agreement with the prediction [11], the chaotropic Na-perchlorate stabilizes, while the kosmotropic Na-acetate destabilizes the native conformation as compared to the Hofmeisterneutral NaCl (see the critical temperatures in Fig 5) The critical temperatures assigned to the slope-brakes of the SVD v2 vectors of the Amide I′ band (Fig 5) seem to correlate with the pre-transition peaks of the Cp(T) curves in the DSC measurements (Fig 2) This correlation becomes understandable if we consider the following definition of heat capacity [27]: C p Tị ẳ T ! G : ∂T2 ð4Þ This means that the peaks of the Cp(T) curves are proportional to the curvature of the free energy function within a narrow temperature range This curvature is expected to be the highest at those temperatures where the most significant acceleration of free energy changes takes place, due, e.g., to the onset of large-scale conformational changes From the FTIR measurements, on the other hand, the slope breaks of the v2(T) intensity functions of the second SVD base spectra (s2) mark just such points (Fig 5), hence they should, in fact, well approximate the critical temperatures established from the Cp(T) curves (Fig 2) (An exact coincidence cannot be expected, because v2(T) is not strictly proportional to G(T).) About conformational fluctuations, the neutron-scattering and the H–D exchange FTIR measurements provide the most direct information They demonstrate that the intensity of fluctuations is influenced by Hofmeister salts in the entire investigated temperature range When approaching the critical temperature of the α(II) → α(I) phase transition, the salt-induced effects are getting enhanced (Figs and 5) The disappearance of the Amide II band of bR due to H/D exchange upon increasing temperatures shows the same order of critical temperatures for the three Hofmeister salts (CH3COO− b Cl− b ClO− ) as observed for conformational changes via the Amide I' band (Fig 5) Here, however, due to the rather smooth and broad transition from the low-rate H/D exchange to the high-rate one, it would be difficult to assign precise values to the critical temperatures Nevertheless, it can be safely stated that the high-rate disappearance of the Amide II band of bR starts always at a few degrees lower temperature than that characteristic for the corresponding conformational changes as determined from the thermotropic responses of the Amide I′ region and of Cp (Figs and 2, respectively) Since intra-molecular fluctuations are needed to make the peptide \NH groups accessible for the D+ ions, the increase of the H/D exchange rate already below the critical temperatures determined from Cp(T) or v2(T) is indicative of an increased rate of equilibrium fluctuation changes in the protein preceding its conformational transition It is very interesting, how the Hofmeister salts affect these equilibrium fluctuations below and above the critical temperature From the FTIR experiments it turns out that below the Amide II critical temperatures, the kosmotropic NaCOOCH3 allows the steepest H/D exchange in bR, i.e it induces larger conformational fluctuations than the HE-neutral NaCl or the chaotropic NaClO4 does (see the Am-II curves in the low-temperature region of Fig 5) Around the critical temperatures, large-scale structural rearrangements start, leading to protein denaturing at the end In the high-temperature region, the temperature dependence of the H/D exchange becomes steeper, too Here, evidently, larger segments of the bR molecules become at least temporally accessible to the external D2O-rich medium on their way to complete denaturing Interestingly, for bR above the critical temperature, the fluctuation–Hofmeister salt relationship seems to reverse: here, the slope of the H/D exchange curve becomes the steepest for bR in the chaotropic perchlorate and the flattest for bR in the kosmotropic acetate salt The above findings are in good agreement with the results of the neutron scattering experiments, where the MSD(T) functions directly measure fluctuations of the non-exchanged H atoms of amino acids, that are supposedly buried in the innermost part of the protein From ca 40 to 75 °C, the MSD values (characteristic of the fluctuations), as well as the slopes of the MSD(T) curves (characteristic of the “flexibility”), are higher for the kosmotrope-treated bR-NaCOOCH3 sample than for the chaotrope-treated bR-NaClO4 one On the other hand, the higher flexibility of bR-NaClO4 as compared to that of bR-NaCOOCH3 above the critical temperature is apparently opposite to what was observed below this temperature, similarly to the v2 vectors of the Amide II H/D exchange data The temperatures corresponding to the change of the force constant also seem to be in accordance with the slope-brakes in the Amide II H/D exchange curves (Figs and 5) Since the Amide II H/D exchange signal monitors more the outer residues (where the H/D exchange has already taken place at a given temperature), this correlation implies that the dynamical properties of the protein surface determine the internal dynamics, too The onset of the fluctuation changes appears close to, but by a few degrees lower than the critical temperatures found for Amide I′ and for the Cp peak, and alludes to a casual relationship between increased intramolecular dynamics and the corresponding structural changes A similar phenomenon has been observed by neutron-scattering measurements on bR at low temperatures [41], and has been interpreted by the help of a double-well potential model [42], showing that the fluctuations start to increase at temperatures where the system “probes” not only one, but both of the potential wells This occurs already well below the critical temperature of the transition [41,42] In order to interpret the effects revealed by the present study, here we also present a phenomenological free energy landscape model based on the assumptions of Dér at al [11], illustrating the energetics of the temperature-induced conformational changes of bR in the α(II) → α(I) transition regime The free energy of the bacteriorhodopsin-solvent system is depicted as a function of the solvent accessible surface area (ASA), in Fig (Note that possible changes of the lipid–protein interface are not considered in this model.) The two potential wells correspond to the α(I) and α(II) conformations of bR, having smaller and larger ASAs, respectively The middle curve represents the situation for bR-NaCl at its critical temperature (T = Tc(NaCl)) The equally leveled bottoms of the potential minima have a positive or negative slope corresponding to the more closed α(I) or to the more open B Szalontai et al / Biochimica et Biophysica Acta 1830 (2013) 4564–4572 4571 Conclusions Using a complex experimental approach, we demonstrated that Hofmeister-active anions affect the magnitude of protein conformational fluctuations in the entire temperature range studied, even in the interior of the bacteriorhodopsin molecule Chaotropic perchlorate is found to decrease, and kosmotropic acetate to increase fluctuations of the more open α(II) conformation, while they act vice versa on the more closed α(I) form We also found that, in each salt, an onset of enhanced equilibrium fluctuations of the protein precedes the temperature-induced α(II) → α(I) phase transition The obtained results are in accordance with the predictions of theories outlined earlier [11,42], and provide general implications to a better understanding of the governing principles of conformational fluctuations, as essential attributes of protein dynamics and function [17] Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.bbagen.2013.05.036 Fig Double-well model of the free energy landscape of bacteriorhodopsin at the α(II) → α(I) transition temperature of bR-NaCl (T = Tc(NaCl)) Gibbs solute-water free energy (Gs) of bR is plotted as a function of the solvent accessible surface area (ASA) in various salt solutions The more closed secondary structure element (with smaller ASAs) is α(I), while the more open one (with bigger ASAs) is α(II) The middle curve (green) represents undisturbed landscape of the Hofmeister-neutral bR-NaCl The upper curve (blue) is the landscape of the kosmotropic bR-acetate, distorted by adding an increasing linear baseline reflecting the increased protein–water interfacial tension caused by the presence of the kosmotropic acetate (NACOOCH3) salt [11] The lower curve (red) was obtained by adding a negative background to the original landscape to show the effect of the decreased protein–water interfacial tension, due to the effect of the chaotropic perchlorate [11] The salt-dependence of the magnitude of the ASA fluctuations (W) and the free energy differences between the α(I) and α(II) states (∆G) are shown by the arrow-headed lines kT is the depth of the Boltzmann “energy fluid” Indices kosmo, neut, chao, refer to kosmotropic, neutral, and chaotropic Hofmeister salts, respectively α(II) conformations, respectively (In this context, namely, an “open” conformation also implies that ASA tends to reach a local maximum, while in a “closed” conformation it “seeks after” a local minimum, within the borders of the free energy well.) Kosmotropic salts (like acetate) contribute with an additional linear background of a positive, while chaotropic salts (e.g., perchlorate) with an additional line of a negative slope to the landscape at the same temperature, due to the respective salt-induced increase or decrease of the protein–water interfacial tension (based on Gs = γpw · ASA) The model predicts stabilization of the α(II) conformation versus α(I) by chaotropic, and its destabilization by kosmotropic salts (ΔGchao > 0, while ΔGkosmo b 0) This relationship can also be quantitatively checked by the help of the experimental results Namely, based on the ΔH and ΔCp values for the α(II) → α(I) transition determined from the calorimetric experiments, the corresponding ΔG values can be calculated at any given temperature according to the Gibbs–Helmholtz equation: ΔG ¼ ΔHc ð1−T=T c ịC p ẵT c T ị ỵ T lnT=T c Þ ð5Þ where Tc is the critical temperature and ΔHc is the enthalpy change of the transition up to Tc, while ΔCp is the heat capacity difference between the α(I) and α(II) states Eq (5) assumes a temperature-independent approximation of ΔCp Since for bR in all the three salts ΔCp is zero within the error, this condition is fulfilled in our case, and at the same time it allows a simple approximation of ΔHc with ΔH/2 From the average experimental values of Table 1, at T = Tc(NaCl) = 82 °C (5) yields positive (0.126 kcal·mol−1), zero and negative (−0.096 kcal·mol−1) ΔG values for bR-NaClO4, bR-NaCl and bR-NaCOOCH3, respectively, in agreement with the implications of the model On the other hand, the model also predicts a lower level of conformational fluctuations for chaotropic salts (Wchao) as compared to kosmotropic ones (Wkosmo) below the phase transition, and an opposite relationship above it, giving a straightforward interpretation of the main experimental results of this study Acknowledgements This research was sponsored by the Hungarian research grants: KTIA-OTKA 78367 (A.D.), OTKA K-101821 (A.D.), K-75818 (B.Sz.), K-101633 (T.P.) and CNK-80345, as well as by a bilateral grant between the Hungarian and Bulgarian Academies of Sciences We would like to thank the Institut Laue–Langevin (Grenoble, France) for providing beam time for the experiments We also wish to thank Dr Jérôme Combet for the discussions concerning the neutron scattering data treatment and Karine Huard for her help in sample preparation References [1] F Hofmeister, Zur Lehre von der Wirkung der Salze II, Arch Exp Pathol Pharmacol 24 (1888) 247–260 [2] P Lo Nostro, B.W Ninham, Hofmeister phenomena: an update on ion specificity in biology, Chem Rev 112 (2012) 2286–2322 [3] M.G Cacace, E.M Landau, J.J Ramsden, The Hofmeister series: salt and solvent effects on interfacial phenomena, Q Rev Biophys 30 (1997) 241–277 [4] K.D Collins, M.W Washabaugh, The Hofmeister effect and the behaviour of water at interfaces, Q Rev Biophys 18 (1985) 323–422 [5] P Khoroshyy, A Dér, L Zimányi, Effect of Hofmeister cosolutes on the photocycle of photoactive yellow protein at moderately alkaline pH, J Photochem Photobiol B 120 (2013) 111–119 [6] C Ganea, A Babes, C Lupfert, E Grell, K Fendler, R.J Clarke, Hofmeister effects of anions on the kinetics of partial reactions of the Na+, K+-ATPase, Biophys J 77 (1999) 267–281 [7] R.M Davidson, S Seneff, The initial common pathway of inflammation, disease, and sudden death, Entropy 14 (2012) 1399–1442 [8] W Kunz, P Lo Nostro, B.W Ninham, The present state of affairs with Hofmeister effects, Curr Opin Colloid Interface Sci (2004) 1–18 [9] W Melander, C Horvath, Salt effects on hydrophobic interactions in precipitation and chromatography of proteins — interpretation of lyotropic series, Arch Biochem Biophys 183 (1977) 200–215 [10] M.C Gurau, S.M Lim, E.T Castellana, F Albertorio, S Kataoka, P.S Cremer, On the mechanism of the Hofmeister effect, J Am Chem Soc 126 (2004) 10522–10523 [11] A Dér, L Kelemen, L Fábián, S.G Taneva, E Fodor, T Páli, A Cupane, M.G Cacace, J.J Ramsden, Interfacial water structure controls protein conformation, J Phys Chem B 111 (2007) 5344–5350 [12] T.J Richmond, Solvent accessible surface area and excluded volume in proteins: analytical equations for overlapping spheres and implications for the hydrophobic effect, J Mol Biol 178 (1984) 63–89 [13] A Dér, Salts, interfacial water and protein conformation, Biotechnol Biotechnol Equip 22 (2008) 629–633 [14] Y Zhang, P.S Cremer, The inverse and direct Hofmeister series for lysozyme, Proc Natl Acad Sci U S A 106 (2010) 15249–15253 [15] A Neagu, M Neagu, A Dér, Fluctuations and the Hofmeister effect, Biophys J 81 (2001) 1285–1294 [16] W Stoeckenius, R.H Lozier, R.A Bogomolni, Bacteriorhodopsin and the purple membrane of halobacteria, Biochim Biophys Acta Rev Bioenerg 505 (1979) 215–278 [17] A Dér, J.J Ramsden, Evidence for loosening of a protein mechanism, Naturwissenschaften 85 (1998) 353–355 [18] S Krimm, A.M Dwivedi, Infrared spectrum of the purple membrane: clue to a proton conduction mechanism? Science 216 (1982) 407–708 [19] J Wang, M.A El-Sayed, The effect of protein conformation change from α(II) to α(I) on the bacteriorhodopsin photocycle, Biophys J 78 (2000) 2031–2036 [20] S.M Barnett, C.M Edwards, I.S Butler, I.W Levin, Pressure-induced transmembrane αII- to αI-helical conversion in bacteriorhodopsin: an infrared spectroscopic study, J Phys Chem B 101 (1997) 9421–9424 4572 B Szalontai et al / Biochimica et Biophysica Acta 1830 (2013) 4564–4572 [21] D Oesterhelt, W Stoeckenius, Isolation of the cell membrane of Halobacterium halobium and its fractionation into red and purple membrane, Methods Enzymol 31 (1974) 667–678 [22] M Bée, Quasielastic Neutron Scattering: Principles and Applications in Solid State Chemistry, Biology, and Materials Science, Adam Hilger, Bristol, England and Philadelphia, 1988 [23] F Gabel, D Bicout, U Lehnert, M Tehei, M Weik, G Zaccai, Protein dynamics studied by neutron scattering, Q Rev Biophys 35 (2002) 327–367 [24] J.C Smith, Protein dynamics: comparison of simulations with inelastic neutron scattering experiments, Q Rev Biophys 24 (1991) 227–291 [25] H.H Paalman, C.J Pings, Numerical evaluation of X-ray absorption factors for cylindrical samples and annular sample cells, J Appl Phys 33 (1962) 2635–2639 [26] D Richard, M Ferrand, G.J Kearley, Analysis and visualisation of neutronscattering data, J Neutron Res (1996) 33–39 [27] N.V Prabhu, K.A Sharp, Heat capacity in proteins, Annual Review of Physical Chemistry, 2005, pp 521–548 [28] S.G Taneva, J.M.M Caaveiro, A Muga, F.M Goni, A pathway for the thermal destabilization of bacteriorhodopsin, FEBS Lett 367 (1995) 297–300 [29] Z.A Tokaji, E.A.B Fodor, A.A Szabó-Nagy, T.A Páli, Hydroxylamine as a thermal destabiliser of bacteriorhodopsin, Eur Biophys J 39 (2010) 1605–1611 [30] C Zscherp, A Barth, Reaction-induced infrared difference spectroscopy for the study of protein reaction mechanisms, Biochemistry 40 (2001) 1875–1883 [31] P Zavodszky, J.T Johansen, A Hvidt, Hydrogen-exchange study of conformational stability of human carbonic-anhydrase-B and its metallocomplexes, Eur J Biochem 56 (1975) 67–72 [32] E Goormaghtigh, V Cabiaux, J.M Ruysschaert, Determination of soluble and membrane protein structure by Fourier transform infrared spectroscopy II Experimental aspects, side chain structure, and H/D exchange, Subcell Biochem 23 (1994) 363–403 [33] E.R Henry, J Hofrichter, Singular value decomposition: application to analysis of experimental data, Methods Enzymol 210 (1992) 129–192 [34] H.A Laczkó-Dobos, B.b Szalontai, Lipids, proteins, and their interplay in the dynamics of temperature-stressed membranes of a cyanobacterium, Synechocystis PCC 6803, Biochemistry 48 (2009) 10120–10128 [35] K Wood, U Lehnert, B Kessler, G Zaccai, D Oesterhelt, Hydration dependence of active core fluctuations in bacteriorhodopsin, Biophys J 95 (2008) 194–202 [36] G Zaccai, Moist and soft, dry and stiff: a review of neutron experiments on hydration-dynamics–activity relations in the purple membrane of Halobacterium salinarum, Biophys Chem 86 (2000) 249–257 [37] U Lehnert, V Réat, M Weik, G Zaccaï, C Pfister, Thermal motions in bacteriorhodopsin at different hydration levels studied by neutron scattering: correlation with kinetics and light-induced conformational changes, Biophys J 75 (1998) 1945–1952 [38] 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 Natl Acad Sci U S A 90 (1993) 9668–9672 [39] J Fitter, R.E Lechner, G Buldt, N.A Dencher, Internal molecular motions of bacteriorhodopsin: hydration-induced flexibility studied by quasielastic incoherent neutron scattering using oriented purple membranes, Proc Natl Acad Sci U S A 93 (1996) 7600–7605 [40] H.B Callen, T.A Welton, Irreversibility and generalized noise, Phys Rev 83 (1951) 34–40 [41] D.J Bicout, G Zaccai, Protein flexibility from the dynamical transition: a force constant analysis, Biophys J 80 (2001) 1115–1123 [42] G.A.B Zaccai, How soft is a protein? A protein dynamics force constant measured by neutron scattering, Science 288 (2000) 1604–1607 [43] I Jelesarov, H.R Bosshard, Isothermal titration calorimetry and differential scanning calorimetry as complementary tools to investigate the energetics of biomolecular recognition, J Mol Recognit 12 (1999) 3–18 [44] J.L Arrondo, A Muga, J Castresana, F.M Goni, Quantitative studies of the structure of proteins in solution by Fourier-transform infrared spectroscopy, Prog Biophys Mol Biol 59 (1993) 23–56 [45] I.H van Stokkum, H Linsdell, J.M Hadden, P.I Haris, D Chapman, M Bloemendal, Temperature-induced changes in protein structures studied by Fourier transform infrared spectroscopy and global analysis, Biochemistry 34 (1995) 10508–10518