Specific Anions Effects of on the Stability of Azurin in Ice - The Journal

Jul 30, 2008 - This investigation represents a first attempt to gain a quantitative estimate of the effects of the anions sulfate, citrate, acetate, c...
0 downloads 0 Views 142KB Size
J. Phys. Chem. B 2008, 112, 10255–10263

10255

Specific Anions Effects of on the Stability of Azurin in Ice Giovanni B. Strambini* and Margherita Gonnelli Istituto di Biofisica, Consiglio Nazionale delle Ricerche, Via Moruzzi 1, 56124 Pisa, Italy ReceiVed: April 07, 2008; ReVised Manuscript ReceiVed: June 04, 2008

This investigation represents a first attempt to gain a quantitative estimate of the effects of the anions sulfate, citrate, acetate, chloride and thiocyanate on the thermodynamic stability (∆G °) of a model globular protein in ice at -15 °C. The method, based on guanidinium chloride denaturation of the azurin mutant C112S from Pseudomonas aeruginosa, distinguishes between the effects of cooling to subfreezing temperatures from those induced specifically by the formation of a solid ice phase. The results confirm that, both in liquid and frozen states, kosmotropes (sulfate, citrate and acetate) increase significantly protein stability, relative to chloride, whereas the chaotrope thiocyanate decreases it. Throughout, their stabilizing efficacy was found to rank according to the Hofmeister series, sulfate > citrate > acetate > chloride > thiocyanate, although the magnitude of ∆(∆G°) exhibited a distinct sensitivity among the anions to low temperature and to ice formation. In the liquid state, lowering the temperature from +20 to -15 °C weakens considerably the stabilizing efficacy of the organic anions citrate and acetate. Among the anions sulfate stands out as the only strong stabilizer at subfreezing temperatures while SCN- becomes an even stronger denaturant. Freezing of the solution in the presence the “neutral” salt NaCl destabilizes the protein, ∆G° progressively decreasing up to 3-4 kcal/mol as the fraction of liquid water in equilibrium with ice (VL) is reduced to less than 1%. Kosmotropes do attenuate the decrease in protein stability in ice although in the case of citrate and acetate, their efficacy diminishes sharply as the liquid fraction shrinks to below 2.7%. On the contrary, sulfate is remarkable for it maintains constantly high the stability of azurin in liquid and frozen solutions, down to the smallest VL (0.5%) examined. Throughout, the reduction in ∆G° caused by the solidification of water correlates with the decrease in the denaturant m value, an indirect indication that protein-ice interactions generally lead to partial unfolding of the native state. It is proposed that binding of the kosmotropes to the ice interface may inhibit protein adsorption to the solid phase and thereby counter the ice perturbation. Introduction In vitro, freeze-thawing of protein solutions may result in irreversible protein aggregation and severe loss of catalytic activity of enzymes, for which reasons many proteins cannot be stored in ice or lyophilized from it without partial inhibition of their function. The cold lability of protein molecules is also believed to be one of the causes of freeze injury and death for many living organisms.1 Both prokaryotic and eukaryotic cells adopt a common strategy in protecting their proteins by producing or importing low molecular weight organic substances (such as sugars, polyols and methylamines) called osmolytes.2–4 The adverse effects of freezing on protein stability are presumably a combined action of multiple factors, prominent among which are: low temperature (eventually leading to cold denaturation), the large over 100 fold freeze-concentration of solutes in the liquid phase and the solidification of water with possible adsorption of the macromolecule to the ice-liquidus interface. The relative contributions of these individual stresses, however, as well as the extent to which cosolutes effects on cryoprotection are thermodynamic or kinetic in nature are still open issues.5 Because the underlying cryoprotective mechanisms are still not fully understood accurate prediction of cosolutes effects on protein stability in ice is premature. As a result formulations and protocols for the stabilization of therapeutic proteins during long-term storage in ice and industrial freeze-drying processes are still largely worked out empirically, case by case, drawing * Corresponding author. Telephone: +39 050 315 3046. Fax: +39 050 315 2760. E-mail: [email protected].

mostly on long time experience and guided by thermodynamic data pertaining to stabilizing additives in warm temperature liquid solutions6 Until recently, pertinent low temperature thermodynamic data was scarce in both liquid and frozen solutions. The major obstacle to the determination of the thermodynamic stability of proteins in ice and, consequently, of the influence of cosolutes in preserving the native fold, is represented by the intrinsic complexity of frozen solutions, challenging quantitative evaluation of chemical eqiulibria by ordinary spectroscopic techniques. Lately, an experimental approach was developed in this laboratory, based on laser-excited CCD-detected Trp fluorescence, capable of determining protein unfolding eqiulibria in ice.7 A suitable model protein was found, the mutant C112S of azurin from Pseudomonas aeruginosa (Caz), that in the presence of guanidinium chloride (Gdn) denatures rapidly and reversibly both in liquid and frozen solutions, thus providing an assessment of its thermodynamic stability (∆G°) in ice. The method is based on the linear relationship between the equilibrium free energy change and the denaturant concentration, ∆G ) ∆G° - m[Gdn], where the m value represents the slope of the free energy plot and ∆G° the protein stability at zero denaturant concentration.8 For this protein it was found that it is the solidification of water, rather than cold temperature or freeze-concentration of solutes that poses a significant stress on the native fold, as represented by an up to 30-40% decrease of ∆G° as the pool of liquid water in equilibrium with ice shrinks to below 1%. Successive investigations demonstrated that among the natural osmolytes, members of the methylamine class offer wide ranging degrees

10.1021/jp8030122 CCC: $40.75  2008 American Chemical Society Published on Web 07/30/2008

10256 J. Phys. Chem. B, Vol. 112, No. 33, 2008 of protection against the ice perturbation, not necessarily correlated to their stabilizing influence in ordinary solutions.9 Unexpectedly, popular sugars (sucrose and trehalose) and polyols (sorbitol and glycerol), commonly employed to preserve the activity of pharmaco proteins during freeze-drying, did not attenuated the ice perturbation suggesting that cryoprotection by polyhydric compounds is probably of kinetic rather than of thermodynamic origin.10 Dissolved salts are often found in bioprocessing media either as buffer components or precipitating agents but not all salts are equivalent in precipitating or stabilizing proteins. Specific ion effects exhibit a recurring trend called the Hofmeister series,11–14 which is more pronounced for anions than for cations. Depending on the degree of ordering of the water molecules in the first hydration shell electrolytes are divided into kosmotropes (strongly hydrated) and chaotropes (weakly hydrated), with the pair Na+ and Cl- usually taken as the dividing line of the respective series.15,16 In solution, kosmotropes have stabilizing and salting out effects on globular proteins whereas chaotropes are known to destabilize folded proteins and give rise to salting in behavior. However, exceptions are found and the basis for ion specificity leading to the Hofmeister series is still the subject of active investigation.14,17 Despite a long standing and extensive examination of the effects of small electrolytes on protein stability in ordinary liquid solutions there is little if any systematic evaluation of their influence at subfreezing temperatures in either liquid or frozen solutions. The present investigation examines for the first time the effect of a series of anions, ranging from kosmotropes to chaotropes, on the freeze-induced destabilization of azurin, attempting to dissect the specific contributions of low temperature and of the solidification of water. The choice of anions and the maximum concentration employed was severely restricted by the low solubility of many salts at subfreezing temperatures, or by their small common ion product with some other ion present in the denaturing salt mixture (Na+, Cl-, and Gdn+). In other cases the restriction is owed to a relatively high eutectic temperature of the salt, above - 15 °C. The examined series includes sulfate, citrate, acetate, chloride and SCN-. In all cases but sulfate (due to the low solubility of sodium sulfate) the common cation was Na+. For sulfate the common cation was NH4+ and the comparison between sulfate and the reference anion chloride was obtained from the difference between (NH4)2SO4 and NH4Cl. The results obtained at +20 and -15 °C emphasize that although the effect on protein stability ranks according to the Hofmeister series under all conditions examined, the response to low temperature and to ice formation is significantly different among the anions. Materials and Methods All chemicals were of the highest purity grade available from commercial sources and were used without further purification. Tris(hydroxymethyl)aminomethane (Tris), NaCl Suprapur, was from Merck (Darmstadt, Germany). NH4CL, (NH4)2SO4, NaSCN, sodium citrate, sodium acetate, and Guanidinium chloride (Gdn) were from Sigma-Aldrich (Deisenhofen, Germany). The concentration of Gdn was determined by refractive index18 and confirmed by density measurements.19 Water, doubly distilled over quartz, was purified by a Milli-Q Plus system (Millipore Corporation, Bedford, MA). Protein Expression and Purification. The preparation of C112S azurin was done following a procedure analogous to that described by Karlsson et al. 20 for the wild type protein except for the omission of any CuSO4 addition in both growth and

Strambini and Gonnelli purification medium.21 The plasmid carrying the wild type sequence was a generous gift of Prof. A. Desideri (Universita` di Roma, ”Tor Vergata”). The C112A mutant was constructed using the QuikChange kit (Stratagene, La Jolla, CA) and confirmed by sequencing. Sample Preparation. It must be stressed that with frozen samples fine control of the concentration of each solute in the denaturation mixture is crucial for the reproducibility of unfolding eqiulibria and the precision of the derived thermodynamic parameters. To this end we adopted rigorous gravimetric control of sample composition and prepared all samples for denaturation in ice by diluting with water the original solute mixtures used for denaturation in liquid solutions, a procedure that assures exactly the same composition of solutes in the liquidus at all VL examined. For each salt, equilibrium denaturation profiles in liquid solutions (at +20 and - 15 °C) were obtained by examining a series of about 25 azurin samples, appositely prepared to contain a constant amount (from 0.3 to 1.0 M) of a particular salt (additive), a variable concentration of denaturant (Gdn) and a balancing amount of reference salt (NaCl or NH4Cl) as required to maintain the freezing temperature (Tf) of the solution close to -15 °C, the temperature of denaturation experiments in ice. The samples (6 µM in protein, 150 µL volume) were prepared by adding 20 µL of protein stock (45 µM, in 0.1 mM Tris pH 7.5) to 130 µL of appositely prepared salt mixtures, here referred to as mother mixtures (mxs), containing a Gdn-NaCl or a GdnNH4Cl salt mixture of variable Gdn content sufficient to fully denature the protein, with or without sub molar quantities of one of the four salts (additives) examined here. The mxs were prepared gravimetrically from two parent stocks (stocks B), equimolar in additive, one containing Gdn the other NaCl or NH4Cl. The concentration of the reference salt in the mxs was adjusted to maintain a Tf close to -15 °C, once diluted with the protein aliquot (130 + 20 µL). Stocks B, in turn, were prepared gravimetrically by adding to 50 mL volumetric flasks a constant amount of additive salt and either Gdn (8.1 M) or the reference salt, NaCl (5.0M) or NH4Cl (4.3 M). Their pH was adjusted to between 7.2 and 7.5 with 30 mM Tris. Each series is labeled according to the additive: NaCl reference and NH4Cl reference (no additive), sulfate, citrate, acetate and thiocyanate. Frozen samples are characterized by a solid phase of pure water (ice) in equilibrium with a liquid phase (liquidus) comprising liquid water and practically all the solutes (protein, salts and additives) at relatively high concentration, [solutes]liquidus. The samples for denaturation experiments in ice were prepared in the same fashion as above, but subsequent to diluting the mxs from about 2- to 200-fold with pure water. Because the starting liquid samples from which they are prepared have a Tf ≈ -15 °C, at this temperature, any water added to these solutions will freeze out leaving a liquidus of exactly the same composition of the original stock. For any given dilution, the magnitude of VL of the corresponding frozen sample is simply given by VL ) [solutes]liquid/ [solutes]liquidus (from the equality 1[solutes]liquid ) VL[solutes]liquidus and VL ) 1 in the liquid state). In these experiments VL ranged from 0.54 to 54%, which corresponds to total solutes concentrations in the starting liquid state, at 20 °C, ranging from about 0.020 to about 2.0 M. Frozen samples for fluorescence measurements were prepared by placing liquid samples (150 µL) in capped spectrosil quartz tubing, 4 mm ID, and freezing them in a cold bath (-20 °C). Subsequently, ice was melted leaving a small seed from which new ice was formed at the bath temperature, set at -15 °C, the

Anions and Protein Stability in Ice

J. Phys. Chem. B, Vol. 112, No. 33, 2008 10257

TABLE 1: Sample Composition, Isosbestic Wavelength (λiso), and Liquid to Ice Correction Factor (fc)a sample

[Ref salt] (M)

NH4Cl (NH4)2SO4 NaCl Na-citrate Na-acetate NaSCN

3.62 2.35 3.50 2.51 2.18 3.25

[additive] (M)

λiso(20 °C) (nm)

λiso(-15 °C) (nm)

fc(∆G0)

fc(m)

346.9 343.5 345.0 341.7 341.8 344.0

338.8 335.6 338.5 335.6 335.3 336.6

1.00 1.02 0.92 1.00 0.985 0.99

1.00 0.98 0.96 0.98 0.98 0.99

0.73 0.50 1.01 0.32

a The reference salt is NaCl except when the additive is (NH4)2SO4 in which case the reference salt is NH4Cl. Indicated is the concentration of reference salt that when added to the Additive salt yields a solution with a freezing temperature just above -15 °C.

temperature of the experiment. This procedure improved signal reproducibility, when compared to freezing by sudden cooling the solution to very low temperatures, presumably because of a more even spatial distribution of solutes in frozen samples. No phase separation due solute crystallization in ice was observed in any of the samples examined in this study. For each salt and salt reference at least two independent series, each of 25 protein samples, were prepared for every value of VL (100, 54, 2.7, and 0.54%,) and the fluorescence spectrum of frozen samples was measured again after melting and refreezing the samples. In all, over 1200 protein samples have been examined and over 2400 fluorescence spectra have been measured. The results reported here represent the average values from these measurements. Determination of Freezing Temperatures. In the preparation of stocks B (having a Tf close to -15 °C) the concentration of NaCl and Gdn needed to be adjusted for each methylamine by means of freezing point determinations. For the determination of freezing temperatures 1 mL samples were placed in long round quartz cell (4 mm ID), sealed off from the atmosphere, and slowly frozen in a cold ethanol-water bath (-20 °C) to avoid cracking of the cell. After 0.5 h equilibration the bath temperature was raised by consecutive, 0.05 °C steps intercalated by 10 min equilibration time after each temperature change. Before ice was completely melted the solution was gently stirred (by means of a temperature equilibrated thin plastic rod) to eliminate any temperature and concentration gradient through the sample. The freezing temperature, measured with a precision (0.001 °C) thermometer (Hart Scientific, model 1502, Utah), was the temperature of the bath at which no further ice could be detected. The reproducibility of Tf measurements was generally better than 0.1 °C. Analysis of Denaturation Profiles in Frozen Samples. The analysis of denaturation profiles in ice requires knowledge of the denaturant concentration in the liquidus, [Gdn]liquidus. This is given by the [Gdn]liquid times the freeze concentration factor. Usually the latter is derived for each solute by determining the liquidus curve for every change of sample composition (Gdn concentration). However, because it requires a large number of freezing point determinations, this procedure is lengthy and timeconsuming, even with modern freezing point determination methods based on differential scanning calorimetry.22 In this study we adopted a less rigorous but simpler approach. It takes advantage of the fact that when the amount of ice is small, i.e., in predominantly liquid samples (VL > 96%), the ∆G0 and m value of frozen solutions are expected to be the same as those of liquid solutions. We then determined the effects of this approximation by comparing the thermodynamic parameters obtained from isothermal denaturation curves with predominantly liquid (VL > 96%) and liquid samples, both plotted against the [Gdn] in the liquid state. In general, the procedure yields a few percent larger estimates of ∆G 0 and m value from the denaturation profiles in ice. Consequently, all ∆G0 and m

values obtained with frozen samples have been multiplied by the correcting factors fc reported in Table 1and determined as follows. Liquid samples were prepared to have a freezing temperature a little higher (typically 0.1-0.2°) than -15 °C and were first analyzed in the liquid state, at -15 °C. Subsequently, they were frozen in liquid nitrogen and then let to equilibrate at -15 °C for 2 h before a new denaturation curve was determined after adequate stirring. Because the freezing temperature is slightly higher than -15 °C, a small amount of water remained frozen and a tiny ice ring floated on top of each sample. The factors fc represents the ratios ∆G0(liquid)/ ∆G0(liquidus) and m(liquid)/m(liquidus) obtained from the analysis of these denaturation profiles using the [Gdn] in liquid solutions at 20 °C. This procedure is rapid and the assumption that the protein stability is not affected by a small amount of ice is quite reasonable as for VL > 50% no difference in ∆G0 and m value could be found between liquid and frozen samples.7 Fluorescence Measurements. Prior to fluorescence measurements liquid and frozen samples were equilibrated for at least 2 h at the bath temperature and then rapidly transferred to the sample holder of the fluorometer, set at the same temperature. Kinetic runs following temperature jumps showed that 10 to 15 min equilibration time was sufficient to give a constant degree of denaturation both in ice and in liquid solution, which remained invariant over a period of 1 week. Fluorescence spectra were measured on a homemade apparatus7 that implements pulsed UV-laser excitation and emission detection by means of a CCD camera. The main advantage of laser over traditional lamp-monochromator sources is a considerable reduction of the spurious background signal, while the merit of CCD detection over the traditional scanning monochromator-photomultiplier assembly is simultaneous acquisition of the entire spectrum. The latter feature is crucial for avoiding spectral distortions caused by instability of excitation intensity and changes of optical properties of frozen samples, such as excitation/emission collection efficiency and transmittance of the medium, during spectral acquisition. Pulsed excitation (λex ) 288 nm), horizontally polarized with the respect to the optical plane, was provided by a frequency-doubled Nd/ Yag-pumped dye laser (Quanta Systems, Milan, Italy) with pulse duration of 5 ns, pulse frequency up to 10 Hz and energy per pulse varying from 10 to 1000 µJ. The emission collected at 90° from the excitation direction was passed through a 290 nm long-pass filter (WG290, Lot-Oriel, Milan Italy) and dispersed by a 0.3 m focal length triple grating imaging spectrograph (SpectraPro-2300i, Acton Research Corporation, Acton, MA) set to a band-pass of 2.0 nm. The emission intensity in the spectral range 295-435 nm was monitored by a back-illuminated 1340 × 400 pixels CCD camera (Princeton Instruments Spec-10:400B (XTE), Roper Scientific Inc., Trenton, NJ) cooled to -60 °C. Fluorescence intensity variations among frozen samples were reduced by rotating the sample (3 Hz) during spectral acquisition and by averaging 8-10 pulses during

10258 J. Phys. Chem. B, Vol. 112, No. 33, 2008

Strambini and Gonnelli

1.5 s acquisition time. To avoid vapor condensation on optical components the whole apparatus was maintained under nitrogen atmosphere. Data Analysis. The fraction of native azurin, fN, in the denaturation equilibrium was determined by assuming a twostate equilibrium. The method employed is based on the shift of the center of spectral mass of the fluorescence spectrum, υg ) υiFi/νFi, Fi the fluorescence intensity at υi ) 1/λi, according to the relationship23

fN ) [1 + (ν - υD)/(υN - υ)φ]-1

(1)

where υN and υD are the center of spectral mass of native and denatured azurin, respectively, and φ is the ratio of their fluorescence quantum yields, φD/φN. The parameter φD/φN is given by the ratio νFiD/νFiN as obtained from the fluorescence spectrum of D and N each normalized at λiso. It should be noted that the above expression for fN takes into account all variations of molar absorptivity and fluorescence yield between N and D states of the protein and, moreover, it is totally independent of excitation intensity, extent of light scattering in frozen samples and protein concentration within the exciting beam. Note also that because the fluorescence yield varies substantially with temperature and the additive salt employed, in particular for the D state, the parameters of equation 1and the isosbestic point were determined for each additive both at +20 and -15 °C, the two temperatures examined here. The isosbestic points are collected in Table 1. The thermodynamic parameters ∆G°, m value, and D1/2 (the mid point denaturant concentration) describing the unfolding transition were derived from nonlinear least-squares fitting of fN data to the expression fN ) 1/{1+ exp[-(∆G0 - m[Gdn])/ RT] }.8,24,25 The data were fitted using the program Origin 7 (Originlab Corporation). We point out that the above estimate of ∆G° is subject to the validity of the linear extrapolation method, LEM, when Gdn is the denaturant. Concern has been raised that with some proteins, particularly if sensitive to the ionic strength of the medium, ∆G° estimated by LEM using Gdn may differ from that obtained using urea as denaturant, and that in general the latter is in better agreement with calorimetric studies. The reason may lie on a different experimental conditions, dilute buffer for urea vs molar salt concentration for Gdn, respectively, but also on a non linear free energy vs [Gdn] plot in the in the low concentration range, presumably induced by the screening of electrostatic interactions by Gdn.26 To address this concern we have repeated the denaturation of azurin in dilute buffer employing urea as denaturant. The estimate of azurin stability was, within experimental error, the same between urea and Gdn denaturation. In practice, for the determination of protein stability in ice the use of Gdn/NaCl salt mixture is an obligatory choice, for it permits one to maintain constant conditions of freezing temperature, pH, and roughly constant ionic strength (ions concentrations, apart from the exchange of Na+ with Gdn+) across the transition. Also, the protein stability in a medium of relatively high ionic strength is more pertinent to the frozen state as due to the large freezeconcentration of buffering salts, proteins in ice invariably find themselves in a medium of relatively high, 2-5 M, electrolyte concentration. The alternative use of urea as denaturing agent in ice is not practicable mainly because the much larger concentration of urea required to unfold the protein, compared to Gdn, would not permit the solution to freeze (at -15 °C). At even lower freezing temperatures the denaturant would probably crystallize out as the eutectic temperature of urea is relatively high (-13 °C).

Figure 1. Changes in fluorescence spectrum and intensity that characterize the denaturation of C112S azurin (6 µM) in liquid samples (VL ) 100%), at -15 °C. The samples contain 0.73 M (NH4)2SO4 and 2.35-2.92 M Gdn/NH4Cl salt mixture, ranging in Gdn content from 0 to 100% (all concentrations refer to the starting liquid solution at 20 °C). N and D refer to the spectrum of native and fully denatured protein. The excitation wavelength is 288 nm. All spectra are corrected for instrumental response.

Results Figure 1 shows a typical example of the changes in fluorescence spectrum and intensity (λex ) 288 nm) that characterize Caz denaturation. The spectra refer to liquid samples (VL ) 100%), at -15 °C, obtained from solutions containing 0.73 M (NH4)2SO4 and 2.35-2.92 M Gdn/NH4Cl salt mixture, ranging from 0 to 100% in Gdn content (all concentrations are referred to the liquid solutions at 20 °C). The Figure emphasizes the large spectral difference between N and D states as well as a clearly identified isosbestic point characteristic of two-state reactions. The salt concentration and isosbestic points are reported for each additive in Table 1.Throughout, the N and D spectra are practically identical between liquid and frozen state, the only difference being a somewhat larger background contribution with frozen samples. As pointed out before7 background subtraction was not found to be necessary as it had negligible effects on the estimate of fN . As in previous experiments with other cosolutes, Caz denaturation in liquid and frozen solutions was found to be completely reversible, irrespective of the salt employed. In ice, reversibility was indicated by complete recovery of the native fluorescence upon thawing fully denatured samples as well as by promptly reversible shifts on the equilibrium following up or down temperature jumps. The lack of hysteresys in the unfoldingequilibrium,evenafterseveralconsecutivecooling-warming cycles, is consistent with a highly reversible process that excludes the formation of protein aggregates (a condition not met by the wild type protein). Lastly, ice is a spatially nonuniform medium in which, depending on freezing conditions, the distribution of solutes, and therefore the degree of unfolding, could vary from one site to another. Tests devised to check the spatial uniformity of the degree of azurin denaturation consisted in measuring the fluorescence spectrum in different positions of the sample after reducing the excitation beam diameter from 2.5 (normal) to 0.5 mm, to enhance spatial resolution. The results obtained with frozen samples with fN values close to 0.5, for maximum sensitivity, indicated right through that despite an up to 1.5 fold variation of the fluorescence intensity between different sites across the sample, presumably reflecting distinct excitation/ fluorescence collection efficiencies or spatially non uniform protein concentration, no significant differences could be

Anions and Protein Stability in Ice

Figure 2. Effect of various salts on the equilibrium denaturation curves of C112S azurin (6 µM), at 20 °C. Key: (9) Ammonium sulfate (0.73 M (NH4)2SO4 plus 2.35 M NH4Cl); (1) Na-citrate (0.5 M Na-citrate plus 2.51 M NaCl); (2) Na-acetate (1.01 M Na-acetate plus 2.18 M NaCl); (b) NaSCN (0.32 M NaSCN plus 3.25 M NaCl); (full line) NaCl reference (3.5 M salt mixture); (dashed line) NH4Cl reference (3.62 M salt mixture). The standard error in fN is typically less than 3%. The lines drawn through the points represent the best fits to the data.

detected in the degree of denaturation. Thus, it appears that under the freezing conditions adopted here and at the spatial resolution of the monitoring beam both the liquidus composition and the stability of azurin in ice are uniform throughout. Effect of Anions on the Stability of Caz in Liquid Solutions at +20 and at -15 °C. The effect of each additive salt (at the concentration indicated in Table 1) on the equilibrium unfolding curve at 20 °C is displayed in Figure 2. Simply from the shift of the profile to higher or lower denaturant concentration, with respect to the reference anion Cl-, it is evident that sulfate, citrate, and acetate stabilize the protein whereas SCN- destabilizes it. The parameters ∆G° and m value obtained by leastsquares fitting a two-state model to the data are collected in Table 2. D1/2 (mol/liter) represents the Gdn concentration at the transition midpoint and is given by the ratio ∆G°/m. The parameters for the NaCl salt reference are in good agreement with the values reported by Sandberg et al. 27(∆G° ) 9.0 kcal/ mol, m ) 5.74 kcal/mol · M, and D1/2 ) 1.57 M) and by our previous study.7 Note that although across the transition the concentration of the reference salt, NaCl or NH4Cl, decreases by roughly the same amount by which the Gdn concentration increases (a compensation needed to keep the sample Tf constantly close to -15 °C) as long as the linear extrapolation method is valid to obtain ∆G° the variable NaCl or NH4Cl level does not affect its estimate. The stabilizing effectiveness of the individual anions, relative to Cl-, is derived from the difference in ∆G° with respect to their reference salt, NH4Cl for ammonium sulfate and NaCl for sodium citrate, sodium acetate and NaSCN. The change in ∆G° estimated on a molar basis, ∆(∆G°)/[anion], is +3.4 kcal/mol for sulfate, +2.8 kcal/mol for citrate, +1.5 kcal/mol for acetate, and -5.5 kcal/mol for SCN-. The results attest to significant anions effects on the stability of Caz and rank their magnitude in the order, sulfate > citrate > acetate > Cl- > SCN-, which is largely in accord with the Hofmeister series.28 Lowering the temperature from 20 to -15 °C in most cases decreases the stabilizing effectiveness of the additive salt. For citrate and acetate the value of ∆(∆G°) obtained at ambient temperature is practically halved at -15 °C whereas for SCN- it becomes even more negative, i.e. the anion is an even more efficient denaturant. The change is marginal, within the range of the

J. Phys. Chem. B, Vol. 112, No. 33, 2008 10259 experimental error, for both NaCl and ammonium sulfate (Table 2). Thus, while the ranking order of the anions effectiveness is unchanged, sulfate stands out as the only strong stabilizer at subfreezing temperatures. With regards to the large decrease of ∆G° derived for NaSCN we point out that in this case the unfolding equilibrium at -15 °C showed evidence of departure from a two-state reaction and, consequently, the estimated ∆G° based on the simple two-state model is only approximate its uncertainty presumably greater than that indicated by the experimental error. Indeed, the fluorescence spectrum at intermediate degrees of denaturation shows significant deviations from the isosbestic point indicating that additional molecular species are formed having a fluorescence spectrum distinct from that of native or denatured species. A detailed characterization of the intermediates and a more refined multistate analysis of the unfolding equilibrium are beyond the scope of the present investigation and were not undertaken. The m value is a measure of the denaturant effectiveness to unfold the protein under specific experimental conditions. Cosolutes will generally alter the m value only if they interfere with the interaction between denaturant and protein. For Caz in NaCl, m ) 5.83 kcal/mol · M, at 20 °C. It decreases somewhat (≈ 0.5 kcal/mol · M) when the reference salt NaCl is replaced by NH4Cl but also in the presence of sub molar concentrations of Na-acetate or NaSCN (Table 2). The decrease is more pronounced (≈1.9 kcal/mol · M) with ammonium sulfate and Na-citrate. Because of the smaller m value in the presence of sulfate and citrate their denaturation transition is shifted to larger Gdn concentrations (D1/2 ) ∆G°/m) than would be predicted by the increase in protein stability (Figure 2). At lower temperature, -15 °C, m values tend to increase, the only exception to the trend being SCN-, for which it decreases further. Part of the increase is artificial because in the analysis of low temperature (-15 °C) data we have used Gdn concentrations that refer to 20 °C, which do not take into account the increase in solutes concentration caused by cooling. While uncorrected concentrations have no influence on the estimate of ∆G°, they will yield larger apparent m values. The significant variability in the temperature-induced change in m value among the salts, however, and especially the opposite trend found with NaSCN, indicates that temperature affects differently the denaturing effectiveness of Gdn in the presence of the various anions. Anions Effect on the on the Stability of Azurin in Ice. The only difference between frozen samples prepared from increased dilution of the above liquid samples is that a proportionally greater fraction of water will form ice (VL decreases). As the liquidus composition is constant with VL freezing will affect the unfolding equilibrium and ∆G° only if the protein free energy in the liquidus, of either N or D state, is perturbed by the presence of ice. Denaturation in frozen samples was monitored at three selected VL (54, 2.7 and 0.54%). Sample denaturation profiles in ice upon a 100 fold decrease in VL, from 54 and 0.54%, are shown in Figure 3 for ammonium sulfate, Na-citrate and their respective reference salts, NH4Cl and NaCl. A first observation to make about the frozen state is that with very fine control of temperature and sample composition the reproducibility of fN measurements in ice and the precision of the thermodynamic parameters derived from these transitions are comparable to those of liquid samples (see standard deviation). A common trend, particularly evident in the case of neutral NaCl, is that decreasing the liquid fraction leads to a progressively less steep unfolding transition, with only minor shifts in the transition midpoint, D1/2. The outcome is a

10260 J. Phys. Chem. B, Vol. 112, No. 33, 2008

Strambini and Gonnelli

TABLE 2: Thermodynamic Parameters Relative to the Effects of Specific Salts on the Denaturation of Caz in Liquid and Frozen Solutionsa T (°C) 20

-15

-15

-15

-15

a

VL (%) 100

100

54

2.7

0.54

additive

∆G0 (kcal/mol M)

NH4Cl-ref (NH4)2SO4 NaCl-ref Na-citrate Na-acetate NaSCN NH4Cl-ref (NH4)2SO4 NaCl-ref Na-citrate Na-acetate NaSCN NH4Cl-ref (NH4)2SO4 NaCl-ref Na-citrate Na-acetate NaSCN NH4Cl-ref (NH4)2SO4 NaCl-ref Na-citrate Na-acetate NaSCN NH4Cl-ref (NH4)2SO4 NaCl-ref Na-citrate Na-acetate NaSCN

7.99 ( 0.06 10.51 ( 0.18 9.30 ( 0.15 10.73 ( 0.31 10.8 ( 0.18 7.70 ( 0.11 7.22 ( 0.10 10.21 ( 0.14 9.05 ( 0.21 9.68 ( 0.15 9.80 ( 0.14 6.62 ( 0.10 7.49 ( 0.12 10.10 ( 0.08 9.40 ( 0.20 9.68 ( 0.10 9.59 ( 0.12 6.66 ( 0.06 6.94 ( 0.12 10.20 ( 0.13 7.01 ( 0.27 10.01 ( 0.24 9.49 ( 0.17 5.28 ( 0.12 6.27 ( 0.28 10.35 ( 0.19 5.90 ( 0.25 8.31 ( 0.27 8.51 ( 0.32 4.84 ( 0.19

∆(∆G0)/[additive] (kcal/mol M) +3.43 +2.86 +1.50 -5.28 +4.07 +1.26 +0.75 -8.00 +3.55 +0.56 +0.19 -8.20 +4.44 +6.00 +2.48 -5.71 +5.56 +4.82 +2.61 -3.50

m value (kcal/mol M)

D1/2 (M)

5.25 ( 0.05 3.95 ( 0.07 5.83 ( 0.19 3.90 ( 0.13 5.17 ( 0.09 5.01 ( 0.07 6.50 ( 0.08 4.60 ( 0.06 6.69 ( 0.16 4.10 ( 0.06 5.50 ( 0.08 4.11 ( 0.08 6.74 ( 0.12 4.60 ( 0.03 6.81 ( 0.22 4.09 ( 0.07 5.40 ( 0.07 4.30 ( 0.05 6.31 ( 0.11 4.65 ( 0.06 5.02 ( 0.11 4.32 ( 0.10 5.27 ( 0.09 4.36 ( 0.10 5.68 ( 0.26 4.70 ( 0.09 4.10 ( 0.17 3.72 ( 0.12 4.75 ( 0.18 4.04 ( 0.15

1.52 2.66 1.59 2.75 2.09 1.49 1.11 2.22 1.35 2.36 1.78 1.18 1.11 2.19 1.38 2.37 1.77 1.19 1.10 2.19 1.40 2.32 1.80 1.17 1.10 2.20 1.44 2.23 1.79 1.20

∆(∆G0) is calculated relative to the common cation reference salt assuming a linear dependence on the salt concentration. D1/2 ) ∆G0/m.

decrease of both m value and ∆G° as the liquid water pool shrinks. An exception to this trend is ammonium sulfate, which stands out for a remarkably constant denaturation curve at all VL examined. The thermodynamic parameters as function of VL, reported in Table 2, are conveniently displayed in Figure 4. The presence of a solid ice phase per se does not necessarily affect the stability of the protein, at least not before the fraction of liquid water falls below 50%. That is, the denaturation profile and ∆G° in wet ice, at VL ) 54%, are not detectably different from those in the liquid solution at the same temperature. For NaCl the data confirm the progressive decrease of ∆G° at small VL, about 3.5 kcal/mol between VL ) 54 and 0.54% (Figure 4, top panel), demonstrating that reduction of the liquid phase can lower significantly the stability of azurin, a phenomenon we have referred to as the “ice perturbation”. The perturbation is attenuated in the presence of Na-citrate and Na-acetate as in both cases ∆G° is maintained large up to the solidification of over 97% of water. Protection by these organic acid anions drops significantly, about 40%, at the smallest VL (0.54%). With ammonium sulfate the ice perturbation is totally inhibited, the protein stability being, within experimental error, practically identical all the way from completely liquid solution up to 99.5% of frozen water. Lastly, we observe that the ice perturbation is less pronounced with NH4Cl, relative to NaCl while in the presence of NaSCN ∆G° remains constantly low at all VL. Discussion The decrease of protein stability in ice is presumed to arise from a combined effect of low temperature, freeze-concentration (over 100-fold) of cosolutes and the presence of the solid ice

phase.5 In general, the efficacy of a cosolute to offset the freezeinduced instability of a protein will be a cumulative effect of its influence in the liquidus, at low temperature, and of its capacity to neutralize the ice perturbation. By maintaining solutes composition constant in liquid and frozen samples, this investigation has attempted to dissect the specific contribution of some anions with respect to low temperature and to ice formation. For azurin in the “neutral” salt NaCl, the stability is little affected by cooling from ambient temperature to -15 °C but it is significantly reduced with the progressive solidification of water, a clear indication that for this protein it is primarily the ice interface that destabilizes the globular fold. In the choice of cosolutes, among them buffering salts, to help preserve the integrity of pharmaco proteins during low temperature storage or industrial freeze-drying one of the guiding criteria is the ability to stabilize proteins in ordinary liquid solutions. As a rule of thumb, for small electrolytes this is provided by the ranking in the Hofmeister series. Such a criterion, however, presupposes constancy in the stabilizing effectiveness of salts from ambient to subfreezing temperatures as well as between liquid and frozen states that as yet has not been established for any salt. The results obtained with familiar anions, ranging from kosmotropes to chaotropes, confirmed that their efficacy on the stability of azurin in the frozen state ranks in the same order as with room-temperature solutions. In conformity to the Hofmeister series, kosmotropes increase the stability in the order sulfate > citrate > acetate > Cl-, while the chaotrope SCN- decreases it. Notwithstanding conformity to the Hofmeister series the individual response of the anions to low temperature and to ice formation are quite distinct,

Anions and Protein Stability in Ice

J. Phys. Chem. B, Vol. 112, No. 33, 2008 10261

Figure 3. Effect of decreasing VL, between 54% (empty symbol) and 0.27% (full symbol) on equilibrium denaturation curves of C112S azurin (6 µM), in ice at -15 °C: (9, 0) ammonium sulfate; (2, ∆) ammonium chloride; (1, 3) Na-citrate and (b, O) NaCl. The lines drawn through the points represent the best fits to the data.

presumably reflecting differences in the underlying molecular mechanisms. Effect of Cooling on the Preferential Interactions. Lowering the temperature from 20 to -15 °C weakens considerably the stabilizing efficacy of the organic anions citrate and acetate but not that of sulfate. On a molar basis, as shown in Figure 5, at -15 °C acetate and citrate have become almost equivalent to Cl-, whereas sulfate maintains a stabilizing effect as high as at ambient temperature. Our current understanding of specific ion effects on protein stability leading to the Hofmeister series emphasizes the important role of the protein-water interface and its modification by selective enrichment-exclusion of the ions.12,14,15,17 At the molecular level the mechanism underlying stabilization of the folded state by electrolytes refers to the solvation properties of the ions, stabilizers making the aqueous solution a worse solvent for polypeptides (preferential hydration or solvophobic effect) and vice versa.15 Both sulfate and carboxylate anions form strong H-bonds with water whose immobilization is believed to make polypeptide hydration more costly in free energy terms. For citrate and acetate an additional contribution to the solvophobic effect may come from the repulsive interaction between the nonpolar alkyl moiety and the peptide backbone unit, in a manner similar to that reported for neutral osmolytes.29 Thus, we may speculate that the loss of stabilizing efficacy of the organic anions at subfreezing temperatures reflect the general decrease of the hydrophobic interaction, a phenomenon believed to be mainly responsible also for the cold denaturation of proteins.30 Should this behavior be common to organic acid salts then this class of compounds would not be expected to play a significant role in countering protein cold denaturation and preserve the functionality of enzymes in organisms living at subfreezing temperatures.

Figure 4. Thermodynamic parameters (∆G0, m value) relative to Gdn denaturation of C112S azurin at -15 °C, in the liquid state (VL ) 100%) and in ice of decreasing VL: (9) (NH4)2SO4; Na-citrate (1); Na-acetate (2); NaSCN (b); and the reference salts (∆) NH4Cl and (0) NaCl. The data are taken from Table 2. The vertical dotted line separates liquid from frozen samples.

Decreasing temperatures make SCN- a stronger denaturant, the denaturing strength becoming comparable to that of the guanidinium cation. Contrary to kosmotropes, chaotropes like SCN- destabilizes the native fold by interacting directly with the macromolecule.15 The greater efficacy of the destabilizer at subfreezing temperatures might then reflect an increased binding affinity and, or, a greater accessibility to SCN- binding sites in the protein. Attenuation of the Ice Perturbation by Kosmotropes. Even though it is clear that protein may unfold at the surface of ice the driving force of ice-induced protein degradation is not fully understood. The drop in ∆G0 at small VL, when the specific surface of ice is very large, has the characteristics of an adsorption process in which the macromolecule is partitioned between the ice interface and the surrounding liquidus and when bound to ice the compact native state becomes destabilized. The lower panel of Figure 4 points out that the decrease in protein stability at small VL is in each case directly linked to the reduction in m value. This is confirmed by the observation that the ratio ∆G°/m ) D1/2 is largely independent of VL (Table 2). The direct correlation established between the magnitude of the m value and the change in solvent-accessible surface area of the macromolecule on unfolding (∆ASA)31,32 suggests that the decreased stability in ice is associated to a marked reduction of

10262 J. Phys. Chem. B, Vol. 112, No. 33, 2008

Figure 5. Effect of kosmotropic anions, relative to Cl-, on the stability of azurin in liquid and frozen solutions, at -15 °C. ∆(∆G0) is calculated relative to the common cation reference salt assuming a linear dependence of (∆G0) on the salt concentration. The corresponding quantities at 20 °C are included as open symbols

∆ASA between N and D states. In other words, protein-ice interactions change the structure of either native or denatured state, or both, reducing the surface area of the macromolecule that becomes exposed on unfolding. Protein-ice interactions were first proposed to explain the alterations of the protein tertiary structure in ice 33as well as the irreversible denaturation of some proteins on freeze-thawing34 and have been attributed to partial loosening and expansion of the native fold structure upon adsorption to the interface.7 All three kosmotropic anions examined here were found to contrast effectively the ice perturbation. The increase in ∆G0, relative to Cl-, is up to 5-6 kcal/mol for sulfate and citrate and a little less than halfsfor acetate (Figure 5). Protein adsorption to ice can, in principle, be countered either by changing the nature of the ice-liquidus interface, from attractive to repulsive with respect to polypeptides, or by altering some feature of the macromolecule, like the overall net charge, that might negatively affect its adsorption to the ice interface. The chemical-electrical features of the interface can be modified by electrolyte binding or associating (double layer) directly to ice whereas the protein net charge can be made more negative by anion binding to the protein. Although we have no evidence of direct electrolyte binding to ice we speculate that such an hypothesis, first advanced to account for the remarkable stabilizing influence of the osmolyte trimethyl-N-oxide (TMAO)9 in frozen solutions, may explain the relative effectiveness of sulfate, citrate and acetate to counter the ice perturbation. All three anions make strong interactions with water and could therefore bind directly to ice giving rise to an ordered interface layer expected to be repulsive with respect to polypeptides. For a sulfate-rich ice interface the driving force making protein adsorption to the solid phase very costly in free energy terms would be the strong preferential hydration of the protein induced by the anion, the same interaction that is responsible for the large increase in protein stability in the liquid state, even at -15 °C. For the organic anions the nature of the repulsive interaction may be quite different. By binding to ice through the carboxylic end these ions would expose the alkyl moiety to the liquidus and thereby build up a nonpolar interface from which polypeptides tend to be excluded. The greater ef-

Strambini and Gonnelli fectiveness of citrate to counter the ice perturbation, twice as large relative to acetate (Figure 5), would be accounted for by a larger alkyl chain combined to stronger binding to ice by the doubly ionized tricarboxylic citrate (at pH near neutrality). Although hydrophobic interfaces can also destabilize the folded state of proteins here we are simply suggesting that by preventing the direct interaction of proteins to ice the overall net effect of a nonpolar coating may be stabilizing. In summary, the efficacy of inorganic sulfate to preserve the native fold of azurin in ice stands above that of the other anions examined here. Should its action extend to proteins in general, the salt could be profitably employed in the long term storage of proteins in frozen media. The addition of relatively small quantities of (NH4)2SO4 (about 25% of the solutes fraction, i.e., >5 mM in a 15-20 mM buffer) could be sufficient to effectively neutralize degradation processes leading to enzyme inactivation and protein aggregation, a finding which provides a rationalization for the common practice of adding mM quantities of the salt to solutions of enzymes to be stored at subfreezing temperatures. Although carboxylic acids seem to have little impact on countering protein cold denaturation they can effectively neutralize the perturbation induced by the solidification of water, and should therefore be preferred as buffering salts in the development of protein freeze-drying formulations. By the same token, organic acids and acidic amino acids could play a role in the cryoprotection of organisms surviving in frozen water. Lastly, these preliminary results caution that the solvophobic property of an ion at ambient temperature is not sufficient guarantee of effective cryoprotection in ice, as the impact of low temperature and of specific interactions that may occur between protein and the ice interface are unpredictable. References and Notes (1) Franks, F. Biophysics and biochemistry at low temperature; Cambridge University Press: London, 1985. (2) Hochachka, P. W.; Somero, G. N. Biochemical Adaptation. Mechanism and Process in Physiological EVolution; Oxford University Press: Oxford, U.K., 2002. (3) Yancey, P. H. J. Exp. Biol. 2005, 208, 2819. (4) Yancey, P. H.; Clark, M. E.; Hand, S. C.; Bowlus, R. D.; Somero, G. N. Science 1982, 217, 1214. (5) Bhatnagar, B. S.; Bogner, R. H.; Pikal, M. J. Pharm. DeV. Technol. 2007, 12, 505. (6) Carpenter, J. F.; Prestrelski, S. J.; Arakawa, T. Arch. Biochem. Biophys. 1993, 303, 456. (7) Strambini, G. B.; Gonnelli, M. Biophys. J. 2007, 92, 2131. (8) Pace, C. N. Methods Enzymol. 1986, 131, 266. (9) Strambini, G. B.; Gonnelli, M. Biochemistry 2008, 47, 3322. (10) Strambini, G. B.; Balestreri, E.; Galli, A.; Gonnelli, M. J. Phys. Chem. B 2008, 112, 4372. (11) Collins, K. D.; Washabaugh, M. W. Q. ReV. Biophys. 1985, 18, 323. (12) Zhang, Y.; Cremer, P. S. Curr Opin Chem Biol 2006, 10, 658. (13) Kunz, W.; Lo Nostro, P.; Ninham, B. W. Curr. Opin. Colloid Interface Sci. 2004, 9, 1. (14) Cacace, M. G.; Landau, E. M.; Ramsden, J. J. Q. ReV. Biophys. 1997, 30, 241. (15) Collins, K. D. Methods 2004, 34, 300. (16) Vrbka, L.; Jungwirth, P.; Bauduin, P.; Touraud, D.; Kunz, W. J. Phys. Chem. B 2006, 110, 7036. (17) Der, A.; Kelemen, L.; Fabian, L.; Taneva, S. G.; Fodor, E.; Pali, T.; Cupane, A.; Cacace, M. G.; Ramsden, J. J. J. Phys. Chem. B 2007, 111, 5344. (18) Nozaki, Y Methods Enzymol. 1972, 26 PtC, 43. (19) Protein structure a pratical approch; Creighton, T. E., Ed.; Oxford University Press: Oxford, U.K., New York, and Tokyo, 1989. (20) Karlsson, B. G.; Pascher, T.; Nordling, M.; Arvidsson, R. H.; Lundberg, L. G. FEBS Lett. 1989, 246, 211. (21) Sandberg, A.; Leckner, J.; Shi, Y.; Schwarz, F. P.; Karlsson, B. G. Biochemistry 2002, 41, 1060. (22) Bhatnagar, B. S.; Cardon, S.; Pikal, M. J.; Bogner, R. H. Thermochim. Acta 2005, 425, 149.

Anions and Protein Stability in Ice (23) Paladini, A. A., Jr.; Weber, G. Biochemistry 1981, 20, 2587. (24) Pace, C. N.; Shaw, K. L. Proteins 2000, 4, 1. (25) Santoro, M. M.; Bolen, D. W. Biochemistry 1988, 27, 8063. (26) Perez-Jimenez, R.; Godoy-Ruiz, R.; Ibarra-Molero, B.; SanchezRuiz, J. M. Biophys. J. 2004, 86, 2414. (27) Sandberg, A.; Leckner, J.; Karlsson, B. G. Protein Sci. 2004, 13, 2628. (28) Hofmeister, F. Arch. Exp. Pathol. Pharmakol. 1888, 24, 247. (29) Street, T. O.; Bolen, D. W.; Rose, G. D. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 13997.

J. Phys. Chem. B, Vol. 112, No. 33, 2008 10263 (30) Lopez, C. F.; Darst, R. K.; Rossky, P. J. J. Phys. Chem. B 2008, 112, 5961. (31) Courtenay, E. S.; Capp, M. W.; Saecker, R. M.; Record, M. T., Jr Proteins 2000, 4, 72. (32) Myers, J. K.; Pace, C. N.; Scholtz, J. M. Protein Sci. 1995, 4, 2138. (33) Strambini, G. B.; Gabellieri, E. Biophys. J. 1996, 70, 971. (34) Chang, B. S.; Kendrick, B. S.; Carpenter, J. F. J. Pharm. Sci. 1996, 85, 1325.

JP8030122