Hydrophobic Polyacids in Denaturant Solutions 142'7

4

n 0 - -1

Y 8 2 -I "

0 0 2

(Log KL Icalc Figure 6. Experimental values, (log K L ) ~ ~ ~ ~ for alkali metal ions with triphenylphosphine oxide in tetrahydrofuran at 25' vs. values, (log KL)calcd, calculated using eq 20 based on an iori- dipole model corrected for surrounding polarizable solvent.

scale of our plot. The value of B found corresponded to a value of the radius of the sphere b = 6.9 X 10-8 cm. The plot of (log &Jexpt us. values of (log KL)calcd calculated from eq 19 using the B value (= 0.695 X 1016 cm-2) found above appears in Figure 6.

The radius of the sphere containing cation and ligand (or solvent molecule) required to fit our data to eq 20 (or

19) is very reasonable. The diameter of a molecule of THF may be estimated from models to be -5.5 A. This would lead to an estimate of b of 6.5 to 7 A . For Ph3P0, the ef- fective b value would be expected to be somewhat larger.

We conclude that ion-dipole interaction, modified to include the effects of surrounding polarizable solvent, can account for the values of cation-ligand association con- stants of the alkali metal cations with triphenylphosphine oxide in tetrahydrofuran solvent. Of course, we cannot say that we have proven that these are the only factors that affect the process. As we have pointed out above, we have neglected several factors in our equation which are known to be involved: ion-induced dipole interaction, dipole- dipole repulsion in the complex, and van der Waals dic- persion interaction. We have advanced what we consider to be good reasons for neglecting the contributions due to these factors. In order to assess the relative importance of these terms we have neglected, it would be advantageous to have cation-ligand association data for several other di - pole ligands with the alkali metal ions. We hope to accu- mulate such information in the future.

We further conclude that the effect of surrounding PO- larizable solvent on the energy of interaction between ions and solvent molecules in the first solvation layer28 should be taken into account in efforts to calculate energies of solvation.21929

Conformational Transitions of Hydrophobic Polyacids in Denaturant Solutions. The Effect of Urea

P. Dubin' and U. P. Strauss"

School of Chemistry, Rutgers University, The State University of New Jersey, New Brunswick, New Jersey 08903 (Received November 27, 1972)

Publication costs assisted by the United States Public Health Service

The effect of urea on the charge-induced transition from hypercoiled to extended conformation of alkyl vinyl ether-maleic acid copolymers has been investigated in 0.04 M sodium chloride solutions in the presence and absence of 5 M urea. Potentiometric titrations indicate that the free energy of stabilization of the uncharged compact state relative to the hypothetical uncharged extended state is reduced upon addition of urea from 310 to 120 cal/mol of residue for the polyacid with butyl side chains, and from 1070 to 830 cal/mol of residue for the hexyl copolymer. Intrinsic viscosity results obtained with the ethyl and butyl copolymers a t various degrees of dissociation show that urea expands the molecular dimensions of the copolymers in both hypercoiled and extended conformations. Urea appears to destabilize the hyper. coiled state by enhancing the solvent affinities of the hydrophobic side chains as well as the polar back. bone of the polyacids.

Introduction Considerable interest has arisen concerning the made of

action of denaturants on biological marcromolecules, inas- much as such processes are thought to elucidate the forces stabilizing the native state. The denaturant activity of such widely varied substances as alcohols, detergents, and

urea and guanidine compounds is currently rationalized in terms Of the Several types Of interactions which maintain secondary and tertiary structure in biopolymers. Studies

(1) Present address: Dynapol Corporation, Palo Alto, Calif. 94304.

The Journal of Physical Chemistry, Voi. 77, No. 11, 1972

1428 P. Dubin and U. P. Strauss

with small molecule model compounds, chemically similar to protein moieties, have been of great value in identifying plausible sites of interaction and estimating corresponding thermodynamic contributions to the denaturation pro- cess? It is nonetheless quite difficult to specify exactly the mechanism of action of a denaturant, such as urea, which is primarily effective a t very high concentrations. For example, considerable evidence exists from model compound studies to the effect that urea (1) interacts strongly with peptide backbone groups, thus weakening interpeptide hydrogen and (2) enhances the sol- ubility of nonpolar groups, thus reducing hydrophobic in- teractions, either by altering the bulk properties of the solvent or through more localized effect~,~,5,6 It is difficult to assess the relative significance of such effects in studies with natural biopolymers.7,s

An alternative approach to this problem may be avail- able through the use of synthetic macromolecular model compounds. In our laboratory, for example, we have stud- ied hydrolyzed copolymers of maleic anhydride and alkyl vinyl ethers which undergo conformational transitions re- sembling those occurring in the denaturation of biopoly- mers. When their alkyl side chains contain four or more methylene groups, these copolymers exist a t low pH in highly compact conformations with intrinsic viscosities similar to those of globular proteins.SJ0 These "hyper- coiled" states, stabilized by cooperative hydrophobic in- teractions among side chains, undergo reversible confor- mational transitions to expanded polyelectrolyte configu- rations upon progressive neutralization of the carboxylic acid group~.~JO

We present here an exploratory study of the effects of concentrated urea on such copolymers and their confor- mational transitjons. Our aims are to investigate whether there are, indeed, effects similar to those observed with proteins, and if so, whether and to what extent any such effects can be ascribed to action on specific chemical groupings, such as the hydrophobic groups belonging to the side chains and/or the more hydrophilic backbone. Attempts a t such a differentiation appear promising with our synthetic copolymers, because the length of their hy- drophobic side chains may be varied, a procedure not gen- erally feasible with natural biopolymers. Specifically, we shall consider the effect of 5 M urea on the difference in free energy between the uncharged compact and (hypo- thetical) expanded states of the butyl and hexyl vinyl ether copolymers in dilute sodium chloride solution. This free energy difference is obtained from potentiometric ti- tration data for these polyacids, in conjunction with those of the nonhypercoiling homolog, the ethyl vinyl ether co- polymer.lO These data are supplemented by results ob- tained under the same conditions giving the effect of 5 M urea on the dependence of the intrinsic viscosity on the degree of dissociation for the ethyl and butyl vinyl ether copolymers.

Experimental Section Materials. The preparation and characterization of the

copolymers of maleic anhydride and ethyl vinyl ether (A- VI), butyl vinyl ether (B-11), and hexyl vinyl ether (C) have been described previously.lO Mann Ultrapure urea was recrystallized from hot ethanol. Fischer Certified 0.200 N sodium hydroxide and hydrochloric acid solutions were used as titrants. All water was deionized and doubly distilled.

Preparation of Polymer Solutions. All studies were per- formed in 0.04 M NaC1, either with or without 5 M urea. In order to avoid degradation, polymer stock solutions were prepared in the dark at 4" by stirring a weighed amount of the dry anhydride form of the polymer with an amount of NaOH solution equivalent to the total carbox- ylic acid content of the polymer, neutralizing with HC1, and, finally, adding sufficient water and NaCl to make the solution 0.02 N in carboxylic acid and 0.08 M in NaC1. Prior to titration or viscometry, an aliquot of the refriger- ated stock solution was mixed with an equal volume of either water or freshly prepared 10 M urea. Polymer con- centrations were determined from titration equivalence points.

Potentiometric Titrations. Titrations of polymer solu- tions and appropriate solvent blanks were performed as previously describedlO under nitrogen and at 30°, using a Radiometer pHM26 pH meter equipped with Radiometer G202c glass and K4016 calomel electrodes. Titrant, 0.200 N NaOH, was added with a 2.0-ml Gilmont micrometer buret. The value of the degree of dissociation CY (defined as unity a t the equivalence point of the first acid group), corresponding to a desired pH, was determined from the volumes of added titrant a t that pH and a t the first equi- valence point, corrected for the volume of 0.200 N HC1 re- quired to bring a blank to the desired pH from the pH of the equivalence point.1° It was assumed that the activity coefficient of hydrogen ion is the same in polymer and blank solutions.

Viscometry. Viscosities were measured in a Cannon- Ubbelhode dilution viscometer a t 30" by successively add- ing appropriate volumes of solvent adjusted to the pH of the initial polymer solution. The use of solvent as diluant, as opposed to external dialysis solution, affects the slope of the reduced viscosity-concentration plots, but not the values of the intrinsic viscosity obtained by extrapolation of those plots.11 Because of acid-catalyzed hydrolysis, so- lutions containing urea at low pH exhibit a slow pH drift (ca. +0.1 pH unit/hr) resulting in a gradual increase in the measured viscosity of the polymer solution. When the pH drift was' small, either because the pH was not very low and/or the measurements were made rapidly, the pH of the final, most dilute solution was used to obtain the value of CY corresponding to the extrapolated value of the intrinsic viscosity. In order to obtain reliable measure- ments a t the lowest pH values, and to verify the other data, the following procedure was employed to maintain constant pH throughout the measurement. A sample of polymer solution identical with that introduced into the viscometer was placed in the (thermostated) titration ves- sel and diluted with solvent simultaneously along with the viscometer sample. Constant pH was maintained in the duplicate sample by intermittent addition of 0.200 N HC1, and identical volumes of HC1 were added to the dilution bulb of the viscometer. With this procedure, the pH of the

C. Tanford, Advan. Protein Chem., 24, 1 (1970). Y. Nozaki and C. Tanford, J. B i d . Chem., 238, 4074 (1963). D. R. Robinson and W. P. Jencks, J. Amer. Chem. SOC., 87, 2462 (1965). D. B. Wetlaufer, S. K. Malik, L. Stoller, and R. L. Coffin, J. Amer. Chem. SOC., 86,508 (1964). W. Bruning and A. Holtzer, J. Amer. Chem. SOC., 83,4865 (1961). J. R. Warren and J. A. Gordon, J. B id . Chem., 245, 4097 (1970). T. T. Herskovits, H. Jaillet, and 8. Gadegbeku, J. B id . Chem., 245, 4544 (1 970). P. Dubin and U. P. Strauss, J. Phys. Chem., 71, 2757 (1967). P. Dubin and U. P. Strauss, J. Phys. Chem., 74, 2842 (1970). U. P. Strauss, J. Polymer. Sci., 33, 291 (1958).

The Journal of Physical Chemistry, Vol. 77, No. 11. 1973

Hydrophobic Polyacids in Denaturant Solutions

I I

I m m

40

1429

- - = u u - n 0.80 0.45

- I

r I

5.0,

4.0

I I 0 0.5 1.0

a Figure 1. Potentiometric titration data at 30", plotted according to eq 2: (0) ethyl, ( A ) butyl, and (U) hexyl copolymers in 0.04 M NaCl (ordinate on right): (0 ) ethyl, (A) butyl, and (H) hexyl copolymers in 5 M urea 4- 0.04 M NaCl (ordinate on left).

final solution in the viscometer usually differed from the initial value by less than 0.02 pH units. The total amount of HC1 added was always too small to have a measurable effect on polymer concentration or ionic strength. For urea solutions above pH 5 , viscosity measurements could be made in the usual way, as no pH drift was observed.

Results

acid) is given by the well-known equation12J3 Potentiometric Titrations. The pH of a poly(monoprotic

where bGion io the change in the excess free energy of the polyacid accompanying an incremental change in the de- gree of dissociation, and KO is the intrinsic ionization con- stant, obtained by extrapolating the left-hand side of eq 1 to a = 0, a t which point the second term of the right-hand side of eq 1 vanishes. The analogous equation for a poly- (diprotic acid) is1O

pH 4- log y = pKlo + (RT In lOF1(aGiOn/da) (2) where

112

and where a is defined as unity a t the equivalence point of the first acid group, and Kl0 and Kzo are the first and second intrinsic ionization constants of a diprotic acid res- idue.

Titration data for the ethyl, butyl, and hexyl copoly- mers, in 0.04 M NaCl with and without 5 M urea, are pre- sented in Figure 1 in the form of pK10 + (RT In 10)-1. PGion /ba ) , obtained from eq 2, against a. By means of a procedure described previously,1° pK10 and pKz0 are found to be 3.8 and 6.6, respectively, for all three copoly- mers in the absence of urea. The corresponding values in the presence of urea are found to be 4.4 and 7 . 2 . A similar decrease in acid strength in the presence of urea has been

A * - 0.175 P

fl.10 0.165) t0.14y * Y I

0.050 0.010 0 0

c,eq. I;'

Figure 2. The dependence of the reduced viscosity on polymer concentration at varying degrees of dissociation: (0) ethyl co- polymer in 0.04 M NaCI; (A ) butyl copolymer in 0.04 M NaCI; (Q, 0 , 0 ) ethyl copolymer in 0.04 M NaCl 4- 5 M urea (proce- dures 1, 2, and 3, respectively, see text); (A, V I A) butyl co- polymer in 0.04 M NaCl 4- 5 M urea (procedures 1, 2, and 3 , respectively).

noted for a number of simple dicarboxylic acids.14 The procedure for extrapolating the curve for the butyl copoly- mer in the presence of urea is based on the assumption that the values of pKIO and pK20, respectively, are identi- cal for all three copolymers, and that the titration curves' for the butyl and hexyl copolymers coincide at low values of U. These assumptions are supported by previous titra- tion data in pure water,1° as well as by our present find- ings in the absence of urea.

The regions of negative slope in the titration curves of the butyl and hexyl copolymers in both solvents corre- spond to transitions from hypercoiled to expanded confor- mations.lO The standard state free energy change per mole of residue for the (hypothetical) conformational transition a t zero charge, AGtO, is proportional to the area between the curve of the copolymer undergoing the transi- tion and that of the ethyl copolymer in the same solvent. The details of and justification for this method of deter- mining AGtO have been given previously.lO The value of AGtO for the butyl copolymer, expressed in cal per mole residue, is 310 f 10 in the absence of urea, and 120 f 20 in the presence of urea. The corresponding values for the hexyl copolymer are 1070 f 30 and 830 f 30. These error limits encompass the range of reasonable extrapolations of the data to a = 0. The areas between the curves are found to be rather insensitive to the extrapolated value of pK10. I t is noteworthy that these AGtO values are considerably more precise than those obtained by a similar procedure for the helix-coil transition of ionizable polypeptides, in which case the titration curve for the random coil confor- mation is not experimentally accessible.15

Viscosity. Several representative extrapolations to infi- nite dilution of viscosity data for the ethyl and butyl co- polymers in the two solvents are shown in Figure 2 . In order to facilitate comparisons between the two copoly-

(12) A. Katchalsky and J. Gillis, Red. Trav. Chim. Pays-Bas, 66, 879 (1949).

(13) A. Arnold and J. Th. G. Overbeek, Red. Trav. Chim. Pays-Bas, 68, 192 (1950).

(14) M. Levy and J. P. Magoulas, J. Amer. Chem. SOC., 64, 1345

(15) See, for example, T. V. Barskaya and 0. B. Ptitsyn, Biopolymers, (1 962).

IO, 2181 (1971).

The Journal of Physical Chemistry, Vol. 77, No. 17, 1973

1430 P. Dubin and U. P. Strauss

' 6 2 6.4 d 6 0:8 Ib 1'2 I

a

Figure 3. Intrinsic viscosity as a function of a for ethyl and butyl copolymers in 0.04 Iw NaCI, with and without 5 M urea. Symbols identified in Figure 2. Inset shows A a vs. a for ethyl (0) and butyl (A) copolymers (see text).

mers of different monomer weights, concentration units are equivalents liter-I, each monomer residue contribut- ing 2 equiv of carboxylic acid.

The slopes of the reduced viscosity plots for both poly- mers in 0.04 M NaCl are negligibly small a t low charge, and increase with a, presumably as a result of enhanced intermolecular interactions. Since the diluting solvent was not in dialysis equilibrium with the polymer solution, the observed slopes may be lower than those corresponding to the true Huggins' constants, if one component is preferen- tially absorbed by the polymer.ll Such interactions may explain the negative slope of the butyl copolymer in urea a t low a.

Semilogarithmic plots of intrinsic viscosity against a are presented in Figure 3. This type of plot was chosen in order to facilitate the empirical extrapolation to a = 0. It is of interest that comparable data for polyacrylic and polymethacrylic acids16 are also found to conform to lin- ear semilogarithmic plots in the region of low a.

Discussion Inspection of the appropriate areas of Figure 1 reveals

immediately that the free energies of stabilization of the uncharged butyl and hexyl hypercoils are considerably re- duced in the presence of 5 M urea. In the case of the butyl copolymer, AGtO is reduced by 190 f 25 cal/mol (about 60%), while the corresponding decrease in AGtO of the hexyl copolymer is 240 f 35 cal/mol (about 22%).

The effect of denaturants on the conformational stabili- ty of the native state of proteins has frequently been ana- lyzed in terms of the behavior of small molecule model compounds.2 In such treatments, it is customary to as- sume that hydrophobic peptide side chains are withheld from the solvent in the native state and exposed upon de- naturation. The presumably additive contribution of one such group to the overall denaturant effect on the free energy of the conformation change then corresponds to its free energy of transfer from pure solvent to denaturant so- lution, which may be estimated from appropriate model compound partition or solubility measurements.

Let us examine the consequences of applying an analo- gous procedure for the case of the butyl and hexyl copoly- mers of this study. The values of AGtO result from several

contributions, which include the hydrophobic interactions among alkyl side groups, other interactions of the various groups with one another and with the solvent components, and conformational changes of the polymer chain. If we assume additivity of the interactions involving the two extra methylene groups of the hexyl copolymer, one-half the difference in the AG,O values of the butyl and hexyl copolymers, both obtained in the same solvent, should represent lGCHzO, the free energy of transfer of a methy- lene group from the hypercoiled form to the extended form, in the chosen solvent. We find AGCHz' to be 380 f 15 cal/mol in the absence of urea, and 355 f 15 cal/mol in the presence of urea. If we assume further, as is fre- quently done in the analogous treatment of denaturant effects on proteins, that in the compact conformation the environment of the methylene groups of the hydrophobic side chains is independent of the solvent, we obtain from the difference in the l G C H 2 0 values a value of -25 f 20 cal/mol for the transfer of methylene groups from 0.04 M NaCl to the same solvent containing 5 M urea. For com- parison, the data of Wetlaufer, et C Z ~ . , ~ for the solubility of various hydrocarbons in water and in concentrated urea may be interpolated to give a value of -60 f 10 cal/mol for the same process, and those of Nozaki and Tanford,3 for solubilities of leucine and glycine in water and aqueous urea, lead to a value of -50 f 10 cal/mol. Our result is of the right magnitude. However, this may be somewhat for- tuitous in view of the assumptions implicit in the course of the calculations. For instance, the effect of urea on the intrinsic viscosity of the butyl copolymer a t low values of a, to be discussed below, throws some doubt on the as- sumption of nonaccessibility to solvent of the hydrophobic side chains in the hypercoiled conformation. The implicit assumption that the hydrophobic side chains are com- pletely exposed to the solvent in the random coil confor- mation may also be questioned. These considerations suggest that the agreement may in part be due to a can- cellation of neglected effects, and that therefore in com- parisons of this type applied to biological or other syn- thetic macromolecules caution should be exercised in evaluating the significance of apparent agreement of this sort.

Further information about the effects of urea on the conformations of these copolymers may be obtained from the intrinsic viscosity results shown in Figure 3. The in- trinsic viscosity is a measure of the molecular dimensions, which, in turn, depend on the interactions of the macro- molecules with the solvent. Whereas the potentiometric titration method as applied here yields only the effect of urea on the difference in free energies of the hypercoiled and hypothetical extended states at zero charge, the in- trinsic viscosity should allow an estimate of the effect of urea on the copolymers in their actual states a t both zero and finite charge. The data show that urea raises the in- trinsic viscosity of both ethyl and butyl copolymers over the whole range of a investigated. Since the butyl copoly- mer is in its hypercoiled form a t low values of a and in its random coil form a t high values of a, and the ethyl copol- ymer is in its extended form under all conditions, we see that urea significantly enhances the solvation of both compact and extended forms. However, a meaningful quantitative interpretation of the effects of urea on the two forms from a numerical comparison of the intrinsic

(16) I. Noda, T. Tsuge, and M. Nagasawa, J. Phys. Chem., 74, 710 (1970).

The Journal of Physical Chemistry, Vol. 77, No. 11, 1973

Hydrophobic Polyacids in Denaturant Solutions 1431

viscosity changes appears impossible, primarily because of the lack of adequate theoretical expressions relating the viscosity of the hypercoiled conformation to the free ener- gy of interaction with the solvent.

Nevertheless, a somewhat more limited quantitative in- terpretation of the results may be obtained by means of an alternative procedure for which no explicit mathemati- cal relationship involving the intrinsic viscosity is needed. In this procedure the intrinsic viscosity is employed essen- tially as an indicator for comparing the effect of urea on the macromolecular dimensions with that brought about by a change in the degree of dissociation without urea. At any value of a, the change from 0 to 5 M urea leads to a certain increase in the intrinsic viscosity, as shown by the vertical displacement between the “no urea” and “urea” curves of Figure 3. At not too high values of a (see below) an identical intrinsic viscosity increase may be brought about by the addition of an appropriate amount of sodium hydroxide (in the absence of urea). The resulting increase in the degree of dissociation, denoted by Aa, is given by the horizontal distance between the appropriate urea curve (at a ) and no urea curve (at a + 101) of Figure 3, and is plotted as a function of a for the two copolymers in the inset of that figure.

The increase in [v] accompanying an increase in a is due to an enhancement of the solvent affinity of the car- boxylate groups attached to the polymer backbone. If urea, like NaOH, affected the solvent affinity of the back- bone region alone, 1 a would be independent of alkyl side chain length. In fact, Aa is observed to be greater for the butyl than for the ethyl copolymer, from a = 0 to cy = 0.4, beyond which the method is not applicable because of the maxima in the [v] us. a C U I V ~ S . ~ ~ The difference in Aa for the two copolymers represents clear-cut evidence that

urea enhances the solvent affinity of the hydrophobic side chains. On the other hand, the relatively large value of for the ethyl copolymer indicates that urea also enhances the solvent affinity of groups close to or part of the poly- mer backbone. The slight decrease of l a with increasing a may reflect a relative change in the solvent affinities of carboxylic acid groups and carboxylate groups, in favor of the former, brought about by the addition of urea. Such a preference is indicated by the observation that the disso- ciation of carboxylic acids in aqueous solution is generally diminished by the addition of ~ r e a . 1 ~

Concentrated urea thus reduces the stability of the hy- percoiled relative to the random coil conformation by a combination of effects which involve enhancements of the solvent affinities of both kinds of conformations. Intra- molecular interactions of the hydrophobic side chains as well as of the groups located in the immediate domain of the polymer backbone are affected. In view of the complex interdependence of these interactions with one another and with the macromolecular dimensions, a quantitative resolution of the effects of urea into additive contributions of the various chemical groups appears to be unwarranted with the data thus far available.

Acknowledgment. The support of this research by grants from the United States Public Health Service (Grant KO. GM 12307) and from S. C. Johnson and Son, Inc., is gratefully acknowledged. We also wish to thank Sylvia Taylor for expert technical assistance.

(17) Such maxima, which are observed for these poly(diprotic acids) with sodium but not with tetramethylammonium ion as the counter.. ion, may be attributed to the specific binding of sodium ion to the! polyacids above (Y = l.i8,19

(18) A. W. Schultz and U. P. Strauss, J. Phys. Chem., 76, 1767 (1972). (19) A. J. Begala and U. P. Strauss, J. Phys. Chem., 76,354 (1972).

The Journal of Physical Chemistry, Voi. 77, NO. 11, 1973