Pressure Effects on Folded Proteins in Solution HYDROGEN EXCHANGE AT ELEVATED PRESSURES*

The observed rate constants for base-catalyzed hydrogen exchange reactions between solvent water and peptide nitrogen in lysozyme, ribonuclease A, oxidized ribonuclease A, and poly(oL-lysine) are all enhanced by an increase in pres- sure. Activation volumes have been calculated from the pressure effect on these rate constants. For the folded proteins lysozyme and ribonuclease A, AV$ for base-catalyzed exchange changes from about mllmol at to -3 at 2500 kg/cm”. The same quantity, determined for the random coil polypeptides oxidized ribonuclease A and poly(oL-lysine), does not show this dependence upon pressure. These effects can be understood either in terms of penetration the proteins the of a small degree of mechanisms which such penetration could occur are

The observed rate constants for base-catalyzed hydrogen exchange reactions between solvent water and peptide nitrogen in lysozyme, ribonuclease A, oxidized ribonuclease A, and poly (oL-lysine) are all enhanced by an increase in pressure. Activation volumes have been calculated from the pressure effect on these rate constants. For the folded proteins lysozyme and ribonuclease A, AV$ for base-catalyzed exchange changes from about +9 mllmol at atmospheric pressure to -3 mllmol at 2500 kg/cm". The same quantity, determined for the random coil polypeptides oxidized ribonuclease A and poly(oL-lysine), does not show this dependence upon pressure. These effects can be understood either in terms of solvent penetration of the folded proteins or the onset of a small degree of pressure induced unfolding. Possible mechanisms by which such penetration could occur are discussed. Conformational changes in enzymes dissolved in water can be caused by several variables, such as temperature, pH, solvent composition, and pressure, as well as by addition of substrates, inhibitors, and denaturants. Studies of these changes continue to advance the still incomplete understanding of how enzymes function. The kinetics of proton exchange between solvent and protein is a powerful probe of the conformational properties of folded proteins since the protons on all but a few core peptide nitrogens will exchange with water under conditions such that the native state is much more stable than the denatured state (l-4).
That water enters and leaves folded proteins is an experimental fact; how it does so seems more and more clearly to be tied to the conformational dynamics of protein structure and the role such movements play in enzyme catalysis.  (6)(7)(8). We have adopted the method favored by Whalley (8) We have applied this technique to base-catalyzed hydrogen exchange between water and the folded proteins lysozyme and RNase A as well as to both the acid-and base-catalyzed hydrogen exchange between water and the random coil polypeptides polylysine and oxidized RNase.
Our analysis of these results emphasizes the importance of peptide group solvation prior to exchange of hydrogen between peptide nitrogen and solvent. It is by a consideration of the equilibrium between solvated and unsolvated peptide bonds within a folded protein that we are able to reconcile the two major effects reported herein: that the activation volume of base-catalyzed hydrogen exchange is slightly more positive for folded proteins than for random coil polypeptides at 1 atm and that it is pressure-dependent for the former but not for the latter.
Bovine pancreatic ribonuclease A was Sigma protease-free type XII-A.
The oxidized RNase used was a preparation by the method of Hirs (9) from bovine pancreatic RNase A (Sigma Type III-A), which had been frozen since its production.
Poly ( Temperature control was ?O.Ol" at 25", ?0.03" at 2". Prior to a kinetic run, the pressure vessel was assembled with the sample container (a loml Hamilton gas-tight syringe with the plunger shortened to -1 cm) flushed with buffer and then emptied completely.
After eluting a small volume (0.5 to 1.0 ml) of the in-exchange through a column packed with Sephadex G-25 medium grade, the protein fraction, hereafter called the out-exchange, was injected by syringe into the sample container through the sample,delivery tube. The air bubble thus produced in the hydraulic fluid was removed by pressurizing the system to 250 atm and depressurizing until no bubbles were visible in the oil which was released on depressurizing.
The pressure was then increased at about 1000 kg/cm'/min until final pressure was achieved.
Heating due to compression can be estimated from Equation 3, which gives the temperature rise in the limit of an adiabatic compression.
(%), = sp(x)p 3 Although this can be significant (3-5" for aqueous solutions up to 3 kbar), the time scale of our experiments and the excellent heat transfer characteristics of the apparatus plus the use of chilled outexchange solutions justifies our reporting as the experimental temperature, the temperature of the bath in which the apparatus was placed. Assembled properly, the pressure vessel repeatedly held 3000 kg/cm' for 24 h with no detectable leakage. A small amount (0.2 ml) of sample was discarded when aliquots were removed from the pressure vessel so that each kinetic point contained no sample which had been within the sampling apparatus and thus at an undetermined pressure and temperature. Aliquots of out-exchange were placed on thermostatted medium grade Sephadex G-25 columns and the time recorded.
In the lysozyme and RNase A work, done at 25", the columns were at 6". For the experiments on polylysine and oxidized RNase, at Z", the columns were cooled to -0.5".
Experiments with lysozyme and RNase A were carried out at pH 7.00 using Tris buffer to take advantage of the pressure independence of the, pH of this buffer system (13). In the polylysine and oxidized RNase experiments, HCl/NaCl was used to adjust the pH. All activation volumes are corrected for the change in (OH-) caused by solvent compression at elevated pressure using volume data for water obtained by Vedam and Holton (14).

RESULTS
The hydrogen exchange reaction involves proton removal as the slow step when base catalysis is predominant (151, as shown in Equation 4. When acid catalysis predominates, protonation is considered the rate-determining step. This has traditionally been written as N-protonation (151, although more and more evidence indicates that peptide bonds are oxygen bases (Ref. 16  holds, the ionization volume of water, AVKw , must be taken into account to obtain AVS,,, from the pressure effect on IzobS since (OH-) as well as ko, is pressure-dependent.
Hamann (17) found AVK, to be -20.4 ml/ mol. One can either use AV,_ to adjust the acidity of each reaction so that (OH-) is the same at each pressure or, as we chose to do, take advantage of the relationship expressed in Equation 9 which is shown under "Appendix" to obtain at acidities appreciably more basic than the PH,,,~,. In Equation 9, AVL,, is the apparent activation volume which results when kobS is used in Equation 2. AV&, z nv: on + Lqw 9 The same complication does not arise on the acid side of the pH,[, where Equation 8 applies since (H+) is so high that any pressure-induced change in K,v produces an undetectable change in (H+ From the rate constants recorded in Table I, AV$,,,,, for polylysine was calculated as -14 * 1 ml/m01 at pH 3.00 and 3.50. From Equation 9, AV&,,, = +6 ? 1 ml/mol. From the work at pH 0.51 and 1.00, AVS,, can be estimated as 0 ? 1 ml/ mol. Our study of oxidized RNase was carried out at 2.0", pH 2.5, conditions at which the exchange reaction was slow enough to be observable. This is the PH,,,~, for oxidized RNase (5) Fig. 2 shows the pressure invariance of AI'&+,,, at these conditions for oxidized RNase at three different values of H,,,. From this plot, a value of -6 ? 1 ml/m01 is obtained for A%,,, <I atm~ for oxidized RNase. Using the value for AV&,, obtained from the polylysine experiment and the known value for AL',,, we estimate from Equation 11 a value of +8 2 2 ml/m01 for AL'&,,, from the oxidized RNase data. This is in fair agreement with either the polylysine value or the value obtained from the lysozyme and RNase experiments. There is more uncertainty in the oxidized RNase AV&,,, than in the we can find no other significant difference between the two studies.
RNase results, however, is the demonstration that the pressure invariance of AVS,,, found for polylysine is not a special feature of this homopolypeptide.
Oxidized RNase can be thought of as a protein with no tertiary structure; it has the same diversity in amino acid sequence as RNase A, from which it was derived, yet AVf,,, determined using it is independent of pressure while the analogous experiment on native RNase A yields fundamentally different results (see below).
Pressure Dependence of Exchange from Lysozyme and RNase A -The exchange profiles of folded proteins are curved when displayed on a semilogarithmic plot because of the distribution of rates which extends over almost 6 orders of magnitude (3). Fig. 3 shows our RNase A results plotted in this way; the lysozyme data, also obtained at pH 7.0, in 0.05 M Tris buffer, 25", 1.5 mg/ml, look very similar. Apparent activation volumes at different values of H,, can be obtained from Fig. 3 and an analogous figure for lysozyme by determining the time required to reach a given H,,, at two different pressures and substituting ln(t,/t,) for ln (K,lk,) in Equation 2. An alternative treatment which is completely equivalent, but easier to apply for the case in which the hydrogen exchange follows a simple power law in t, involves plotting log H,,, versus log t. Fig. 4 shows our lysozyme data displayed in this way.
Extrapolation of the plots in Fig. 5 3. Pressure dependence (P) of out-exchange from RNase A, at 25.0", 0.05 M Tris, pH 7.00, 1.5 mg/ml. 7.0 since the pH,,, of the proteins is about 3. Consequently, Equation 9 applies and AVS,,, for both proteins is +9 ml/mol. The difference between this value and that obtained using polylysine is significant. Another difference between the results for the random coil polypeptides and the folded proteins in this study is seen by comparison of Figs. 2 and 5. At pressures above 1000 kg/cm', AVS,,, for the folded proteins becomes significantly more negative, while for polylysine and oxidized RNase, it remains constant as pressure is increased.

DISCUSSION
Before considering the implications our results have for protein conformation, a comparison of the activation volumes we have found for polylysine with those reported for similar reactions is in order. Neuman and co-workers (18,19) have determined activation volumes for base-catalyzed hydrolysis of several esters. le Noble (20) estimates activation volumes of about -10 ml/m01 for the bond formation part of such biomolecular reactions. The values reported by Neuman and coworkers (18,19) are more positive than this by at least 6 ml/ mol for all but one of their substrates, which they interpret as a decrease in solvation of hydroxide ion as the reaction proceeds to the transition state. This is quite reasonable since, as le Noble points out in his review (201, the volume contraction accompanying water dissociation (AV,, = -20.4 ml/ mol) is not shared equally between the 2 ions produced; hydroxide ion is responsible for 12 to 15 ml/mol of the total, indicating that it is quite strongly solvated.
The biomolecular displacement of chloride ion from NH&l, CH,NHCl, and (CH,),NCl by hydroxide ion had AV$ of -2 to 0 ml/mol (21), more positive by 7 to 10 ml/m01 than for similar reactions in which carbon is the central atom which is thought due to a lesser electrostriction of solvent around the transition state when the central atom is nitrogen rather than carbon.
Thus, our own value of i-6 ml/m01 for AVt,,, from polylysine is not unreasonable; the catalyst is hydroxide ion, strongly solvated in the ground state, while the transition state has the negative charge shared by both the nitrogen and the carbonyl oxygen of the peptide group in addition to the oxygen of the hydroxide ion so the electrostriction of solvent in the transition state should be decidedly less than in the ground state. The value of 0 ml/m01 for AVS,, for polylysine, more negative by 6 ml/m01 than AVS,,, , also fits this picture; since hydronium ion does not have such a tightly constricted solvation shell as hydroxide, the solvation contribution to the activation volume will not be as negative in the acid-catalyzed as in the base-catalyzed reaction.
The oxidized RNase study serves to validate that the pressure invariance of AVS,,, found in the case of the homopolymer polylysine also holds for heteropolymers and to substantiate the value of AVS,,, found at 1 atm for polylysine. The data for oxidized RNase show somewhat larger scatter and the form of the plot of activation volume uersus pressure is somewhat less well defined. We did not pursue this study further since oxidized RNase, although a good model for unfolded protein at 1 atm, has not been studied at higher pressures so we could not rule out the possibility of partial refolding. There could also be some refolding at higher basicities. The oxidized RNase data support our use of polylysine as a model for the peptide group in the unfolded solvated state. The difference in AVS,,, of polylysine and oxidized RNase at 1 atm, 6 2 1 and 8 * 2 ml/mol, is not statistically significant, and, as stated earlier, since the value obtained for polylysine is more precise we take"+6 ml/mol as the actual value of the activation volume of the base-catalyzed exchange reaction in the absence of conformation effects.  the plots in Fig. 5 gives -11 rt 2 ml/mol for AV$&, o atmj for both lysozyme and RNase A. Using Equation 9 gives AVS,,,, = +9 ki 2 ml/mol for both proteins, leaving a difference between the value for the proteins and that for polylysine of +3 ml/mol. A more significant result, however, is the effect of elevated pressure on AVS,,, values calculated using Equation 9 for the two proteins. As shown in Table II, AVS,,, changes from +9 ml/mol at 1 atm to -3 ml/m01 at 2500 kg/cm2 for the proteins while it remains constant at +6 ml/m01 for polylysine and between +7 and + 10 ml/m01 for oxidized RNase. Since its pressure variation is so small in the case of the random coil polypeptide, it seems unlikely that AV&,, actually changes significantly with pressure in the case of the proteins; the observed change has to be due to some other pressure-sensitive process occurring in the proteins but not in polylysine or oxidized RNase.
The difference in behavior between random coil polypeptides and folded proteins could be due to pressure-enhanced solvent penetration of the proteins, or to the onset of pressureinduced reversible denaturation.
Insofar as the results do not allow us to distinguish between them, both explanations are briefly discussed in the following sections.

Pressure-enhanced
Solvent Penetration of Folded Proteins -Hydrogen exchange from buried peptide groups at pH 7.0 is base-catalyzed and involves the rate-determining production of an ion, -CON--.
Stabilization of this ion may be accomplished by other polar groups of the protein itself in a few of the exchange events, but structural constraints limit this means of stabilization.
It seems apparent, then, that the charge development in most of the reactions is stabilized by water. The solvation of OH-is probably not larger than a single water molecule at the time that OH-abstracts a proton from a peptide group, since recent fluorescence quenching studies3 (22) indicate that the size of molecules which penetrate to quench fluorescence of buried tryptophans in folded proteins such as RNase T, and lactic acid dehydrogenase is strictly limited. Since in our experiments with lysozyme and RNase A the ratio of H,O to OH-concentrations is 5.5 x lo8 and since H,O is smaller than OH-(H,O), water must penetrate to a buried peptide group many, many times for every time that an OH-(H,O) does. Thus, the exchange reaction can be written as a solvation step The rate constant for base-catalyzed exchange can then be written as k,, = KK, and AV$,,, = AV, + AV&, Since the activation volume for base-catalyzed hydrogen exchange from the random coil polypeptides is pressure-invariant, it is not unwarranted to assume that AVS,, is also independent of Effects on Hydrogen Exchange of Proteins pressure and equal to the value obtained for polylysine of +6 ml/mol. The difference between AVVS,,, for the proteins and polylysine is seen to be AV,, the volume change associated with solvating an exchangeable peptide group. At 1 atm, our results indicate that AV, is small and positive, having a value of +3 2 2 ml/mol, but as the pressure is increased AV, becomes more negative, reaching -11 ? 2 ml/m01 at 2500 kg/ cmZ.
an increasing fraction of the exchange will proceed via the unfolding pathway as the pressure is increased, and the apparent AVS,,,, will decrease.
For AV, to decrease with pressure, the compressibility of the product of the solvation reaction must be greater than the sum of the compressibility of the reactants. That is, for a given increment of pressure, VcoNTo,oL must decrease more than does the sum of the volumes VcONToC, + V,L so. This conclusion is not at odds with existing compressibility data. Liquid water has a low compressibility as compared to other liquids, losing only about 10% of its volume on going from 1 to 2500 kg/cm2, and both lysozyme (23) and RNase A (24) are reported to have low compressibilities.
This finding implies that, since a solvated peptide group can occupy a smaller volume at elevated pressure than can a nonsolvated peptide group and the water necessary to solvate it, elevating the pressure on an aqueous protein solution will tend to force water into the protein. However, our 1 atm value of AV, indicates that transfer of water into the protein is disfavored from a volume standpoint at atmospheric pressure. Examination of Fig. 5 and use of Equation 9 shows that for RNase A the movement of water into the protein has a favorable AV at pressures above about 1500 kg/cm'. For lysozyme the turning point is about 1000 kg/cm2. The pressure denaturation of lysozyme has been studied by Li et al. (25) and is a rather complicated phenomenon involving multiple domains. Thus, quantification via Equation 17 would be difficult. Brandts et al. (24) have studied Rnase A at lower pH than the present experiments; extrapolation of their results to pH 7 would indicate that the AV,, for denaturation would be small and perhaps even positive. However, their experiments were carried out only up to 3200 kg/cm*. Thus, although the data, as such, would exclude unfolding reaction as the basis for observed pressure effects and support the alternate based on increased solvent accessibility of the native state, the validity of our extrapolation and the absence of further pressure-dependent transitions has to be verified by additional experiments. The present study reveals unequivocally that the solvation process preceding the exchange of hydrogens at atmospheric pressure has a very small positive or no volume change associated with it. If we combine this with the observation that practically all hydrogens in the folded state of a protein exchange with such a low energy mechanism (2, 31, we see that the conformational substates responsible for structural motility of a protein appear to have very little volume work associated with it.

APPENDIX
The way we obtained AV&, and AV&,