THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1986 by The American Society of Biological Chemists, Inc.

Vol. 261, No. 27, Issue of September 25. pp. 12477-12485.1486 Printed in U.S.A.

The Systematic Characterization by Aqueous Column Chromatography of Solutes Which Affect Protein Stability*

(Received for publication, May 13, 1986)

Michael W. WashabaughS and Kim D. Collins8 From t k Department of Biochemistry, T k Johns Hopkins University School of Hygiene and Public Health, Baltimore, Maryland 21205

We have systematically characterized, by aqueous column chromatography on a size exclusion cross- linked dextran gel (Sephadex@ G-lo), 12 solutes, 11 of which are known to affect protein stability. Six are chaotropes (water structure breakers) and destabilize proteins, while five are polar kosmotropes (polar water structure makers) and stabilize proteins. Analysis of the chromatographic behavior of these neutral (ethyl- ene glycol, urea), positively charged (Tris, guanidine, as the hydrochloride salts) and negatively charged (Sol-, HPOl-, F-, Cl-, Br-, Cl,CCO;, I-, SCN-, as the sodium salts, in order of elution) solutes at pH 7 as a function of sample concentration (up to 0.6 M), sup- porting electrolyte, and temperature yields four con- clusions, based largely on the behavior of the anions. 1) Chaotropes adsorb to the gel according to their position in the Hofmeister series, with the most chao- tropic species adsorbing most strongly. 2) Chaotropes adsorb to the gel less strongly in the presence of chao- tropes (a salting in effect) and more strongly in the presence of polar kosmotropes (a salting out effect). 3) Polar kosmotropes do not adsorb to the gel, and are sieved through the gel according to their position in the Hofmeister series, with the most kosmotropic spe- cies having the largest relative hydrodynamic radii. 4) The hydrodynamic radii of polar kosmotropes is in- creased by chaotropes and decreased by polar kosmo- tropes.

These results suggest that a chaotrope interacts with the first layer of immediately adjacent water molecules somewhat less strongly than would bulk water in its place; a polar kosmotrope, more strongly.

Escherichia coli dihydroorotase (~-5,6-dihydroorotate ami- dohydrolase (EC 3.5.2.3)) is a dimer with a subunit molecular weight of 38,300, and is subject to dilution inactivation (Wash- abaugh and Collins, 1986). We attribute this dilution-associ- ated inactivation to dissociation of the dimer to monomers and subsequent unfolding of the monomers. It is known that cold-sensitive enzymes, which inactivate at low temperature by oligomer disaggregation or monomer unfolding or both, can be stabilized by ethylene glycol (Penefsky and Warner, 1965), glycerol (Jarabak et al., 1966; Penefsky and Warner, 1965), propylene glycol (Graves et al., 1965), or inorganic phosphate (Irias et aL, 1969; Kono and Uyeda, 1973; Shukuya and Schwert, 1960). We found that ethylene glycol and inor-

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Present address: Brandeis University, Graduate Dept. of Bio- chemistry, Waltham, MA 02254.

§ To whom correspondence should be addressed.

ganic phosphate had additive stabilizing effects on crude extracts of E. coli dihydroorotase (Table l), similar to the additive stabilizing effects of glycerol and inorganic sulfate on yeast cy-isopropylmalate isomerase (Bigelis and Umbarger, 1976). The effectiveness of various neutral salts (neutral referring to pH 7) in stabilizing dilute E. coli dihydroorotase (Fig. 1) correlates with their position in the Hofmeister series (Hofmeister, 1888)’ (Collins and Washabaugh, 1986); polar kosmotropes‘ (polar water structure makers) stabilize the enzyme and chaotropes (Hamaguchi and Geiduschek, 1962) (water structure breakers) destabilize the enzyme. In order to determine the molecular mechanism relating the strength with which these various salts and two neutral compounds bind water Hofmeister’s (Hofmeister, 1888) suggested expla- nation for the origin of salt-specific effects on protein solu- bility) to their effect on protein stability, we have studied the interfacial behavior of these solutes by chromatographing them on a size exclusion cross-linked dextran gel (Sephadex@ G-10).

EXPERIMENTAL PROCEDURES3

RESULTS AND DISCUSSION

Sephadex G-10 is epichlorohydrin cross-linked dextran in beaded form, which separates solutes below a molecular weight of about 700 by a “size-exclusion’’ mechanism: small molecules penetrate the beads and have a longer path length through a packed column (slow elution), while larger mole- cules are excluded from the interior of the beads and have a shorter path length through a packed column (fast elution) (Fischer, 1980; Kremmer and Boross, 1979; Pharmacia, 1970). This separation mechanism as described involves no binding of the solute to the Sephadex gel. In Fig. 2, we have used a macromolecular, uncross-linked dextran of average molecular weight 40,000 to measure Vo, the excluded volume (corre- sponding to the shortest gel sieving path length through the

‘The anions examined by Hofmeister (Hofmeister, 1888) were sulfate, phosphate (HPOi-), acetate, citrate, tartrate, bicarbonate, chromate, chloride, nitrate, and chlorate (C10;). We have deposited an English translation of this paper in The National Translations Center, John Crerar Library, University of Chicago, 5730 S. Ellis Avenue, Chicago, IL 60637, from which a copy may be obtained for $5.00 (nonsubscribing individuals) or for no charge (subscribing uni- versities) by specifying paper 85-20000.

* From the Greek noun kosmos, meaning “order.” Portions of this paper (including “Experimental Procedures” and

Table 2) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No, 86M-1594, cite the authors, and include a check or money order for $1.60 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press.

12477

12478 Aqueous Solutes

column), and tritiated water (THO4) to measure Vi, the in- cluded volume (corresponding to the longest gel sieving path length through the column). The elution position of a solute can be described by Kd, where Kd = (V, - Vo)/( Vi - Vo) and V, is the elution volume of the solute. Thus Kd is the relutioe elution position of a solute, and can vary between Kd = 0 (for the dextran polymer used to measure the excluded volume) and K d = 1 (for the tritiated water used to measure the included volume). A K d of greater than 1 cannot be explained

TABLE 1 E. coli dihydroorotase stability

Crude enzyme extracts were prepared in potassium phosphate buffer (pH 7.0) with or without 25% (v/v) aqueous ethylene glycol (EG) and dihydroorotase activity was assayed with N-carbamyl-DL- aspartate as substrate as described under “Experimental Procedures.” (While this batch (Fisher Scientific Co., lot 726434) of reagent grade ethylene glycol stabilized E. coli dihydroorotase for 48 h, most com- mercial preparations required extensive purification to prevent sub- stantial loss of activity over this time period (Washabaugh and Collins, 1983)). Reagent grade chemicals were used without purifica- tion in preparation of the extract. 2-ml aliquots were incubated at 4 “C and assayed after 48 h. Activity is expressed as a per cent of that present before incubation.

Conditions Activity

%

0.10 M KHZPO, 29 0.10 M KH,PO, + 25% EG 98 1.0 M KHZPO, 106 1.0 M KHZPO, + 25% EG 95

I I 1

so -

c1-

GUAN’

. 2 . 4 . 6

SOLUTE CONC. (MOLAR)

FIG. 1. Effect of Hofmeister solutes on the stability of dilute E. coli dihydroorotase as measured by enzyme activity. 0.2-ml samples containing pure dihydroorotase (10 pg/ml) in 0-0.60 M solute plus 0.10 M NaCl (the latter for charge screening), 0.002 M Tris phosphate buffer (pH 7.0) were incubated under nitrogen at 30 “C in 0.5-ml plastic centrifuge tubes. The solute was either Na2S04, NaCI, or guanidine (GUAN+) HCl. Thus the concentration of NaCl in the middle curve varied from 0.1 to 0.7 M. Dihydroorotase activity was measured over a 24-h period using dihydro-DL-orotate as substrate as described under “Experimental Procedures.” Dihydroorotase activity loss was first order in all cases with no abrupt changes occurring because of solute addition. The Kt for competitive inhibition of dihydroorotase by 0.10 M NazS04 was determined in a separate

pH 8.0 and greater than 0.6 M at pH 5.80; we thus conclude that experiment and, if any inhibition occurred, was greater than 1 M at

sulfate does not stabilize the enzyme by binding at the active site.

The abbreviation used is: THO, [3H]H~0.

by the simple gel sieving mechanism described above; the most plausible and general explanation for a Kd > 1 is adsorp- tion of the solute to the gel.

In all experiments reported in this paper, the column of Sephadex G-10 was equilibrated with aqueous 0.1 M NaCl (pH 7) throughout the experiment to suppress any ion ex- change effects resulting from the low concentration of nega- tive charges on the gel (Eaker and Porath, 1967; Neddermeyer and Rogers, 1968; Ortner and Pacher, 1972). Experiments with 0.1 M NaCl as the sole supporting electrolyte (Figs. 2-5) are referred to as “bolus” experiments since the test solute (the solute whose elution position is being determined) is added to the top of the column in a small, localized bolus; experiments in which an additional electrolyte is added to the 0.1 M NaCl elution solution (Figs. 6-9) are referred to as “uniform” experiments, because the effect of the additional electrolyte spread uniformly throughout the column on the test solute is being measured.

Fig. 2 illustrates the chromatographic behavior of halide anions when 1.0 ml of 0.1 M sodium halides (separately or together) in aqueous 0.1 M NaCl are passed through a Seph- adex G-10 column (1.5-cm diameter X 85.5 cm tall) at 30 “C equilibrated with aqueous 0.1 M NaC1. The amount of sepa- ration obtained decreases greatly as the extent of epichlorohy- drin cross-linking decreases, that is, separation decreases in the series Sephadex G-10, G-15, G-25j5 (Marsden and Hag- lund, 1984). Surprisingly, the halide anions are separated according to their order in the Hofmeister series, with the smallest anion (as determined by x-ray crystallography in the solid; see Collins and Washabaugh, 1986) eluting first and the largest anion eluting last; a gel sieving mechanism would predict the reverse order. Fluoride, which is a kosmotrope (see below and Collins and Washabaugh, 1986), diffuses through the gel with an apparent hydrodynamic radius greater than that of chloride; chloride is approximately neutral on the Hofmeister scale, that is, chloride appears to have only a small effect on water “structure.” Bromide and iodide are chaotropes (see below and Collins and Washabaugh, 1986) and elute after THO, establishing that some mechanism other than gel sieving must be largely determining their chromato- graphic behavior. The late eluting iodide peak has a sharp leading edge and a long trailing edge.

Fig. 3 illustrates that the solutes within each of the cate- gories neutral, cationic, and anionic, elute from Sephadex G- 10 in their order in the Hofmeister series.6 Those solutes which are chaotropic and tend to denature proteins elute later than tritiated water, while those which are kosmotropic and tend to stabilize proteins elute sooner than tritiated water. While there are positive, neutral, and anionic species which elute both before and after tritiated water, the range of chromatographic positions is greatest for the anions. As the bolus concentration of chaotropes is increased, adsorption to the gel becomes less (a salting in effect). This also explains the trailing edge of the iodide peak in Fig. 2: as the concen- tration of iodide decreases on the trailing edge of the iodide sample bolus, the salting in effect is lost and the iodide adsorbs more strongly to the gel.

The double line across Fig. 4 is our best estimate of the region of “normal” behavior for solutes on Sephadex G-10. The line connecting the halide anions demonstrates that their elution position is determined largely by factors other than gel sieving on the basis of molecular weight. Clearly, those

M. W. Washabaugh and K. D. Collins, unpublished data. We have not included data for NaClO, in Fig. 3 because we have

shown that it reacts with Sephadex G-10, generating soluble alde- hydes (data not shown); however, a 3 M bolus of NaC104 in 0.1 M NaCl elutes with an apparent K d of 2.12

12479

FIG. 2. Chromatography of halide anions on Sephadex G-10. 1.0-ml samples containing 0.10 M sodium hal- ides (separately or together) in 0.10 M NaCl containing THO and 0.5% dextran were chromatographed on a Sephadex G- 10 column (1.5 X 85.5 cm) at 30 “C and a flow rate of 0.5 ml/min. The eluent was 0.10 M NaC1. 0.65-ml fractions were collected. Elution profiles were deter- mined as described under “Experimental Procedures.” The eluate peak positions were related as the distribution coeffi- cient, &, defined by & = (v, - vo)/(Vi - Vo) where Vo is the excluded volume (labeled “dextran”), V, is the included volume (labeled “THO), and V, is the elution volume for a given solute.

Aqueous Solutes K,= 0 . 4 1 . 65 1 1 . 2 3 2

0 100

?- UREA 0

L;-;-;-.-.”=* P, v b; 0

I C - - m - = . = = = $ GUAN?^

L * - ruxs+*

Bt”

~ - 0 c1- 0 EG

,- - -

0 ,c

I I I so: . 0 . 2 . 4 .6

BUUS CONC. IWLAR)

FIG. 3. Chromatography of neutral, cationic, and anionic solutes on Sephader G-10. 1.0-ml samples containing 0.001-0.60 M solute in 0.10 M NaCl containing THO and 0.5% dextran were chromatographed on a Sephadex G-10 column (1.5 X 85.5 cm) at 30 “C and a flow rate of 0.5 ml/min. The eluent was 0.10 M NaCl. 0.65- or 1.30-ml fractions were collected. Elution profiles were deter- mined as described under “Experimental Procedures.” The eluate peak positions are presented as the distribution coefficient, Kd, cal- culated as described in the legend to Fig. 2. EG, ethylene glycol; TCA-, trichloroacetate; GUAN+, guanidinium; TRZS’, protonated Tris. So- dium salts of indicated anions were used.

solutes which have a large effect on water structure chromat- ograph anomalously, with kosmotropes eluting earlier than expected and chaotropes eluting later than e~pected.~ Fig. 4

’ A plausible intramolecularly hydrogen bonded structure involving

200 300 400 500 FRACTION

2 t

SCN‘ 0

t I 1 ,+

EG TRIS+

0 I 1 1.5 2 2.5 3

LOG,, MOLECULAR WEIGH1

FIG. 4. Anomalous chromatographic behavior on Sephadex G-10. & values for eluate peak positions of samples containing 0.10 M solute from Fig. 3 are plotted as K d versus loglo molecular weight of the solute at pH 7 as the sodium or hydrochloride salt. The points labeled 1-6 represent glycine and its homopolymers through hexagly- cine. The double line is our best estimate of the region of normal behavior for solutes on Sephadex G-10. The line connecting the halide anions demonstrates that their elution position is largely determined by factors other than gel sieving on the basis of molecular weight. Symbols are defined in the legend to Fig. 3.

again demonstrates that while neutral, cationic, and anionic solutes are well capable of anomalous chromatographic be- havior, it is the anions which show the largest variation.

Fig. 5 indicates that the elution positions of the chaotropes bromide and iodide (which have K d > 1) are affected by

one water molecule can be drawn for glycine; this larger structure is consistent with the chaotrope glycine eluting slightly faster than expected on the basis of its anhydrous molecular weight.

12480 Aqueous Solutes 2

l

u- r( E

0

l r

-1

1 F- e e . - e -

1 I 1

3.2 3 . 4 3.6 I / T (K+xIOl

FIG. 5. Temperature-dependence of elution of the sodium halides from Sephadex G-10.1.0-ml samples simultaneously containing 0.10 M NaF, NaC1, NaBr, and NaI in 0.10 M NaCl plus THO and 0.5% dextran were chromatographed on a Sephadex G-10 column (1.5 X 85.5 cm) at temperatures between 4.5 and 50 “C at a flow rate of 0.35-0.69 ml/min. The eluent was 0.10 M NaC1. The elution positions were determined as described under “Experimental Procedures.” The distribution coefficient, K d , is defined in the legend to Fig. 2. The temperature dependence data on the right are plotted as In Kd versus 1/T. The slope of the line in this experiment yields adsorption associated enthalpies of -1.27 f 0.06 kcal/mol for Br- and -3.02 f 0.06 kcal/mol for I-. The coefficient of determination of the linear least squares fit is 0.992 for Br- and 0.999 for I-.

THIOCYANATE AS TEST SOLUTE CKORIDE AS TEST S a m

SUPPORTING ELECTROLYTE CONC. [MOLAR) FIG. 6. Uniform experiment with thiocyanate as test solute

on Sephadex G-10. 0.5-ml samples containing 9 X M K[’‘C] SCN in 0-0.60 M uniform solute plus 0.10 M NaCl, THO, and 0.5% dextran were chromatographed on a Sephadex G-10 column (1.5 X 34.5 cm) at 30 ‘C and a flow rate of 1.5 ml/min. The eluent was 0- 0.60 M uniform solute as indicated on the horizontal axis plus 0.10 M NaC1. Thus the concentration of NaCl as the uniform solute varied from 0.1 to 0.7 M. 0.65-ml fractions were collected. The uniform solute was Na2S04, NaC1, or NaSCN. The test solute elution profile was determined as described under “Experimental Procedures.” The dis- tribution coefficient, Kd, is defined in the legend to Fig. 2; the tritiated water elution position is indicated by a broken line.

0 .2 .4 .6 SUPPORTING ELECTROLYTE CONC . (MOLAR)

FIG. 7. Uniform experiment with chloride as test solute on Sephadex G-10. 0.5-ml samples containing 5 X M Na[36Cl]Cl in 0-0.60 M uniform solute plus 0.10 M NaC1, THO, and 0.5% dextran were chromatographed on a Sephadex G-10 column (1.5 X 34.5 cm) at 30 “C and a flow rate of 1.5 ml/min. The eluent was 0-0.60 M uniform solute as indicated on the horizontul axis plus 0.10 M NaCl. Thus the concentration of NaCl as the uniform solute varied from 0.1 to 0.7 M. 0.65-ml fractions were collected. The uniform solute was Na2S0,, NaCl, or NaSCN. The test solute elution profile was deter- mined as described under “Experimental Procedures.” The distribu- tion coefficient, Kd, is defined in the legend to Fig. 2; the tritiated water elution position is indicated by a broken l ine .

Aqueous Solutes 12481

SULFATE AS TEST SOLUTE

t t

1 1

0 u 0 .2 . 4 . 6 SUPPORTING ELECTROLYTE CONC. (MOLAR1

FIG. 8. Uniform experiment with sulfate as test solute on Sephadex G-10.0.5-ml samples containing 9 X M Na2[35S]S04 in 0-0.60 M uniform solute plus 0.10 M NaCI, THO, and 0.5% dextran were chromatographed on a Sephadex G-10 column (1.5 X 34.5 cm) at 30 ‘C and a flow rate of 1.5 ml/min. The eluent was 0-0.60 M uniform solute as indicated on the horizontal ark plus 0.10 M NaCl. Thus the concentration of NaCl as the uniform solute varied from 0.1 to 0.7 M. 0.65-ml fractions were collected. The uniform solute was Na,SO,, NaCl, or NaSCN. With NaSCN as the uniform solute, the Kd of Na2[3SS]S04 was determined twice at each of the five NaSCN concentrations. The test solute elution profile was determined as described under “Experimental Procedures.” The distribution coeffi- cient, K d , is defined in the legend to Fig. 2.

temperature with iodide showing the stronger dependence, while the elution positions of fluoride and chloride (which have Kd < 1) are not detectably affected by temperature. This suggests that gel sieving (associated with K d < 1) is a temper- ature-independent process while adsorption to the gel (asso- ciated with Kd > 1) tends to be a temperature-dependent process. The Gibbs-Helmholtz equation for the temperature dependence of the elution position is In K d = -AGO/RT = -aHO/RT + AsO/R, yielding physical adsorption associated enthalpies (Ruthven, 1984) of -1.27 f 0.06 kcal/mol for Br- and -3.02 -+ 0.06 kcal/mol for I-. This calculation assumes identical, noninteracting binding sites. Any adsorption asso- ciated enthalpy for C1- and F- must be less than -0.3 kcall mol, which is less than RT (0.6 kcal/mol). The organic chao- tropes urea (Haglund and Marsden, 1980; Janson, 1967), guanidinium, and thiocyanate, which run more slowly than their molecular weights would predict (Fig. 4) also appear to be “sticky” in this system, while the organic polar kosmotrope ethylene glycol elutes at approximately the position predicted by its molecular weight (Fig. 4; Haglund and Marsden, 1980; Bywater and Marsden, 1983). Nonpolar solutes adsorb to the relatively nonpolar surface (Holmberg, 1983) of the highly cross-linked Sephadex gels (Determann and Walter, 1968; Haglund and Marsden, 1980, 1984a, 1984b; Janson, 1967; Ujimoto and Kurihara, 1981; Yano and Janado, 1980) with a temperature dependence (Haglund and Marsden, 1984b; Uji- mot0 and Kurihara, 1981; Yano and Janado, 1980) opposite to that of the chaotropes in Fig. 5 . (Chaotropes adsorb to Bio- Gel P-2 polyacrylamide (Pecsok and Saunders, 1968) and Spheron 300 hydroxyethyl methacrylate (Borak, 1978) with a temperature dependence similar to that shown in Fig. 5.)

More work is necessary to characterize trichloroacetate, with its highly asymmetrical charge distribution, as either a chao- trope or a detergent. Those polar kosmotropes which are charged typically have high local charge densities, while those chaotropes which are charged typically have low local charge densities.

Fig. 6 presents a uniform experiment where the supporting electrolyte at the concentration indicated has been added to the 0.1 M aqueous NaCl used throughout the experiment. The test solute ( i e . the sample chromatographed on the column of Sephadex G-10 (1.5 x 34.5 cm)) is [’4C]thiocyanate (SCN-), 0.5 ml of 9 X M in the appropriate concentration of supporting electrolyte. Since adsorption to the column ap- pears to dominate the chromatographic behavior of thiocya- nate, we conclude that adding a kosmotrope (SO:-) to the aqueous 0.1 M NaCl supporting electrolyte causes a large increased adsorption of [14C]thiocyanate to the gel (a salting out effect), adding the approximately neutral (in the Hof- meister sense) solute C1- causes a slightly decreased adsorp- tion (a salting in effect), and adding a chaotrope (SCN-) causes a large decreased adsorption of [‘4C]thiocyanate (a salting in effect).

Fig. 7 presents another uniform experiment, this time with 36Cl- as the test solute (the sample chromatographed on the column), 0.5 ml of 5 X M in the appropriate concentration of supporting electrolyte. A kosmotrope (SO:-) added to the aqueous 0.1 M NaCl supporting electrolyte causes slower elution of 36Cl- from the column, the temperature dependence of which (Fig. 9, upper curves) indicates adsorption to the column (a salting out effect), as also suggested by its elution position after tritiated water (dotted line in Fig. 7). Addition of the approximately neutral (in the Hofmeister sense) anion C1- to the supporting electrolyte causes a slightly faster elu- tion of 36Cl-; addition of thiocyanate to the supporting elec- trolyte causes a substantially faster elution. These latter two effects will be interpreted from Figs. 8 and 9.

Fig. 8 presents a third uniform experiment, this time with [35S]SO:- as the test solute (the sample chromatographed on the column), 0.5 ml of 9 X M in the appropriate concen- tration of supporting electrolyte. A kosmotrope (SO:-) added to the aqueous 0.1 M NaCl supporting electrolyte causes slower elution of [35S]SO:- from the column, the temperature inde- pendence of which (Fig. 9, lower curve) indicates that adsorp- tion is not involved, and suggests a decrease in the effective hydrodynamic radius of the test solute. Any adsorption asso- ciated enthalpy for [35S]SO:- in Fig. 9 must be less than -0.13 kcal/mol, which is less than RT (0.6 kcal/mol). Addition of the approximately neutral (in the Hofmeister sense) anion C1- to the supporting electrolyte causes no change in the elution of [35S]SOz- from the column (Fig. 8, middle curue). Since there appear to be no adsorption phenomena associated with sulfate on Sephadex G-10, this suggests that the slightly faster elution of 36Cl- from the column with C1- added to the supporting electrolyte (Fig. 7, middle curue) is due to a less- ened adsorption of the test solute, as in the middle curve of Fig. 6 (a salting in effect). This reasoning implies (a) that C1- is slightly chaotropic (see Samoilov, 1972), and (b ) that 36Cl- adsorbs weakly to Sephadex G-10 (in spite of its apparent temperature-independent elution (Fig. 5, third curue from the top). The adsorption associated enthalpy for 36Cl- is -0.96 0.07 kcal/mol in Fig. 9. Addition of the chaotrope SCN- to the aqueous 0.1 M NaCl supporting electrolyte causes faster elution of [35S]SO:- (Fig. 8, bottom curve), suggesting an increase in the effective hydrodynamic radius of the test solute.

Can thiocyanate’s actions in Fig. 8 be explained solely by direct effects (i.e. adsorption to the Sephadex G-10) Or must

12482 Aqueous Solutes

SUPPORTING ELECTROLYTE INCLUOES 0 .3 H SO4-

1.5 1 - , , , I , I , I , , , 1 1 I

1 -

c1- -\- 7

P

” - Y

Y

c - . 5 - - -1 -

so,= e a I

so,= . I - I I 1 1 1

0 ’ l ‘ l ~ l ’ l ’ l ‘ I I 1

0 10 20 30 40 50 -e

3 -

3.2 3 . 4 I / T (K”x 10’1

3.6 T PC1

FIG. 9. Temperature-dependence of elution of sodium chloride and sodium sulfate under conditions favoring interaction with Sephadex G-10. 0.5-ml samples simultaneously containing 5 X M Na[36C1]C1 and 9 X M N~Z[~‘SS]SO~ in 0.30 M Na2S0, plus 0.10 M NaCl, THO and 0.5% dextran, were chromatographed on a Sephadex G-10 column (1.5 X 34.5 cm) at temperatures between 2.5 and 50 “C at a flow rate of 0.69-1.3 ml/ min. The eluent was 0.30 M Na2SO4 plus 0.10 M NaCl. The elution positions were determined as described under “Experimental Procedures.” The distribution coefficient, Kd, is defined in the legend to Fig. 2. The temperature dependence data on the right are plotted as In K d versus 1/T. The slope of the line in this experiment yields an adsorption associated enthalpy of -0.96 +: 0.07 kcal/mol for C1-. The coefficient of determination of the linear least squares fit is 0.997 for C1-.

TABLE 3 Chromatography of THO and dextran on Sephadex G-10

0.5-ml samples containing THO and 0.5% dextran (average molec- ular weight, 40,000) in the appropriate eluent were chromatographed on a column of Sephadex G-10 (1.5 X 34.5 cm) at 30 “C and a flow rate of 1.5 ml/min. 1.30-mi fractions were collected. All eluents contained 100 mM NaCl. Vi and VO refer to the fraction number containing the peak elution of THO and dextran, respectively.

K v o v; - vo added to 100 m M NaCl Supporting electrolyte

No added electrolyte

100 mM NaSCN

100 mM NaZSO4

100 mM NaCl

300 mM NaSCN

300 mM Na2S0,

300 mM NaCl

600 mM NaSCN

600 mM NazS04

600 mM NaCl

80 79 80 I9 79 80 79 80 I9 79 80 81.5 78 I9 83 I9 81 82 79 18

41 39 41 38 40 40 40 39 40 39 40 40 40 39 41 39 38 41 39.5 39.5 40 40 42 39.5 39 39 40 39 43 40 40 39 41 40 41 41 40 39 39 39 ”

Mean value 80 f 0.6 40-t 0.5 40 +: 0.4

indirect (i.e. water structure mediated) effects be involved? Because sulfate does not appear to adsorb to Sephadex G-10, and since it seems unreasonable to believe that anions in aqueous solution interact directly with other anions (i.e. with no intervening water molecules), the sulfate-mediated effects (Fig. 8, upper curue) must be indirect, mediated by water molecules, and, we would like to argue, through changes in water structure. The thiocyanate effects (Fig. 8, lower curue), however, are less clear cut. We have interpreted the decreased adsorption of the chaotropes to the Sephadex G-10 as their concentration is raised (the bolus experiments of Fig. 3) to a salting in effect. But in adsorptive thin layer chromatography, an overloaded sample (an excess over binding sites) can result in faster elution by a self-displacement mechanism, since each sample molecule in the overloaded region spends a smaller fraction of its time adsorbed to the chromatographic support. Can the effects on [35S]SOf- elution caused by adding thio- cyanate to the supporting electrolyte also be explained by phenomena resulting from the adsorption of thiocyanate to Sephadex G-lo? The pores of the Sephadex G-10 appear not to be occluded, because tritiated water always elutes at the same fraction (Table 3) irrespective of additions to the sup- porting electrolyte, indicating no change in pore accessibility. Also unlikely is charge repulsion between a thiocyanate- coated column and the [“S]lSOq- being sieved through the pores because of the charge screening effects of the 0.1 M NaCl present in all experiments. Two additional arguments support the interpretation of Fig. 8 as resulting from changes

Aqueous Solutes 12483

in hydrodynamic radius of the test solute as opposed to adsorption phenomena: (a) the saturation at 0.3 M of the SO:- effect on [”S]SOf, upper curve, whereas the upper curves of Figs. 6 and 7 (where adsorption is involved) do not saturate in this concentration range, and (b) the small changes in Kd of [35S]SOi-, always remaining well below 1.

We therefore conclude that at least in some instances thiocyanate must be acting through intervening water mole- cules. And, because the effect of adding thiocyanate to the supporting electrolyte on [35S]SO:- sieving is opposite in sign to that of adding sulfate to the supporting electrolyte, the effect cannot be due simply to thiocyanate’s negative charge but must be due to changes in structure of the intervening water molecules.

The surprising saturation of the SCN- mediated enhance- ment of the hydrodynamic radius of [“S]SO:- at 50 mM SCN- (Fig. 8, lower curve) is consistent with a non-uniform distri- bution of anions in solution, and this general issue is discussed further elsewhere (Collins and Washabaugh, 1986).

GENERAL DISCUSSION

A representative Hofmeister series (Hofmeister, 1888)’ for anions is given by the order in which the following species elute from a Sephadex G-10 column (Fig. 3).

SO:- HPOZ- > F- > GI- > Br- > I- (= C10;)6 > SCN-

Those species in this series which elute before C1- are polar kosmotropes and stabilize proteins; those which elute after C1- are chaotropes and destabilize proteins. C1- has little effect on water structure or, in the range of 0.1-0.7 M, on protein stability (Fig. 1). While the chromatographic behavior of Hofmeister anions on size exclusion (sieving) gels has been examined before (Borak, 1978; Bywater and Marsden, 1983; Deguchi, 1975; Deguchi et al., 1977; Egan, 1968; Kremmer and Boross, 1979; Kura et al., 1977; Lindqvist, 1962; Marsden, 1973; Neddermeyer and Rogers, 1968; Neddermeyer and Rog- ers, 1969; Pecsok and Saunders, 1968; Saunders and Pecsok, 1968; Sinibaldi and Lederer, 1975; Ueno et al., 1970; Wilson and Greenhouse, 1976; Yoza et al., 1971; Zeitler and Stadler, 1972), we have examined their behavior more systematically here; our explicit treatment of test solute relative hydrody- namic radius (based on molecular weight), the measurement of clear-cut supporting electrolyte-mediated changes in its value (using temperature independence to rule out adsorp- tion), and the consideration of its relationship to its other behaviors in a system such as this are all new; and finally, for the first time, we have thoroughly integrated these results with the other information available on the interaction of these species with water. We present a review of water struc- ture effects, the Hofmeister series, and their relationship to protein stability elsewhere (Collins and Washabaugh, 1986).

Our results indicate that those solutes which destabilize native (folded) protein structure adsorb to the gel; in addition, they increase the solubility and hydrodynamic radii of test solutes chromatographed in their presence. In contrast, those solutes which stabilize native protein structure do not adsorb to the gel, and, they decrease the solubility and hydrodynamic radii of test solutes chromatographed in their presence. This pattern of behavior is readily explained if we assume that a chaotrope interacts with the first layer of immediately adja- cent water molecules somewhat less strongly than would bulk water in its place, and that a polar kosmotrope interacts with the first layer of immediately adjacent water molecules more strongly than would bulk water in its place. This assertion that the water molecules in the first hydration shell surround-

ing chaotropes are loosely held is consistent with the views of Bernal and Fowler (1933), Gurney (1953), Samoilov (Samo- ilov, 1965, 1957b, 1972; Buslaeva and Samoilov, 1985), Kres- tov (1965), Luck (Luck, 1985; Kleeberg and Luck, 19831, Swain and Bader (1960), and Hertz (Endom et aZ., 1967; Engel and Hertz, 1968), but not with those of Frank and Wen (Frank and Wen, 1957; Wen, 1982), who believe that the first hydra- tion shell of ions is always tightly held. Those polar solutes which adsorb to Sephadex G-10 are sticky because they are easily dehydrated. The driving force for adsorption is the formation of strong water-water interactions in bulk solution from the weakly held water molecules in the first hydration shell around chaotropes, as well as London dispersion inter- actions between the chaotrope and the many di-substituted 1,4-dioxane and related structures known to be attached to the dextran backbone of the highly cross-linked Sephadex gels (Holmberg, 1983). Polar kosmotropes do not adsorb to Sephadex G-10 because their interaction with the water mol- ecules in their first hydration shell is stronger than bulk water-water interactions, making dehydration difficult. Being less polarizable, polar kosmotropes also have less favorable London dispersion interactions with the relatively nonpolar surface of the highly cross-linked dextran gels.

Polar kosmotropes compete more effectively for nearby water molecules than does bulk water; chaotropes compete less effectively for nearby water molecules than does bulk water. This is the basis for the salting out and the salting in phenomena observed on Sephadex G-10, and also explains why polar kosmotropes decrease the hydrodynamic radii of nearby solutes while chaotropes increase the hydrodynamic radii of nearby solutes. These explanations are presented in greater detail and integrated into the existing literature else- where (Collins and Washabaugh, 1986).

Acknowledgments-We thank R. Daniel Camerini-Otero (National Institutes of Health), B. E. Conway (Ottawa), Richard L. Schowen (Kansas), and John I. Brauman (Stanford) for helpful scientific discussions; Diskin Clay (Johns Hopkins) for generating the word “kosmotrope”; and Lawrence Grossman (Johns Hopkins) for advice and encouragement.

REFERENCES Ashworth, M. R. F. (1979) Analytical Methods for Glycerol, Academic

Bernal, J. D. & Fowler, R. H. (1933) J. Chem. Phys. 1,515-548 Bigelis, R. & Umbarger, H. E. (1976) J. Bwl. Chern. 251,3545-3552 Borak, J. (1978) J. Chromatogr. 155,69-82 Bowler, R. G. (1944) Biochern. J. 38,385-388 Buslaeva, M. N. & Samoilov, 0. Ya. (1985) in The Chemical Physics

of Sobation, Part A. Theory of Soluation (Dogonadze, R. R., Kal- man, E., Kornyshev, A. A., and Ulstrup, J., eds) pp. 391-414, Elsevier/North Holland, New York

Bywater, R. P. & Marsden, N. V. B. (1983) in The Journal of Chromatography Library 22A, Chromatography, Fundamentals and Applications of Chromatographic and Electrophoretic Methods. Part A: Fundamentals and Techniques (Heftmann, E., ed) pp. A257- A330, Elsevier/North Holland, New York

Chen, P. S., Jr., Toribara, T. Y. & Warner, H. (1956) Anal. Chern.

Collins, K. D. & Washabaugh, M. W. (1986) Q. Reu. Biophys. 19, in

Deguchi, T. (1975) J. Chromatogr. 108,409-414 Deguchi, T., Hisanaga, A. & Nagai, H. (1977) J. Chromatogr. 133,

Determann, H. & Walter, I. (1968) Nature 219 , 604-605 Eaker, D. & Porath, J. (1967) Sep. Sci. Technol. 2,507-550 Egan, B. Z. (1968) J. Chromatogr. 34 , 382-388 Endom, L., Hertz, H. G., Thul, B. & Zeidler, M. D. (1967) Ber.

Engel, G. & Hertz, H. G. (1968) Ber. der Bunsenges. 72,808-834 Fields, R. (1972) Methods Enzymol. 25,464-468 Fischer, L. (1980) Gel Filtration Chromatography. Laboratory Tech-

Press, Orlando, FL

28,1756-1758

press

173-179

Bunsenges. Phys. Chem. 71 , 1008-1031

12484 Aqueous Solutes nigues in Biochemistry and Molecular Biology (Work, T. S. & Burdon, R. H., eds) Elsevier/North-Holland, New York

Frank, H. S. & Wen, W.-Y. (1957) Discuss. Faraday SOC. 24, 133- 140

Graves, D. J., Sealock, R. W. & Wang, J. H. (1965) Biochemistry 4 ,

Gurney, R. W. (1953) Ionic Processes in Solution, McGraw-Hill, New

Haglund, A. C. & Marsden, N. V. B. (1980) J. Polym. Sci. Polym.

Haglund, A. C. & Marsden, N. V. B. (1984a) J. Chromatogr. 301,

Haglund, A. C. & Marsden, N. V. B. (1984b) J. Chromatogr. 301 ,

Hamaguchi, K. & Geiduschek, E. P. (1962) J. Am. Chem. SOC. 8 4 ,

Hofmeister, F. (1888) Naunyn-Schmiedebergs Arch. Pathol. 24,247- 260

Holmberg, L. (1983) Ph.D. Thesis, Swedish University of Agricultural Sciences, Uppsala

Horowitz, W. (ed) (1975) Official Methods of Analysis of the Associa- tion of Official Analytical Chemists, 12th Ed, p. 223, Association of Official Analytical Chemists, Washington D. C.

Irias, J. J., Olmsted, M. R. & Utter, M. F. (1969) Biochemistry 8 ,

Janson, J.-C. (1967) J. Chromatogr. 28,lZ-20 Jarabak, J., Seeds A. E., Jr. & Talalay, P. (1966) Biochemistry 5 ,

Kleeberg, H. & Luck, W. A. P. (1983) J. Solution Chem. 12,369-381 Komatsu, S. & Nomura, T. (1967) Nippon Kagaku Kaishi 88,63-66 Kono, N. & Uyeda, K. (1973) J. Biol. Chem. 248,8603-8609 Kremmer, T. & Boross, L. (1979) Gel Chromatography, John Wiley

& Sons, New York Krestov, G. A. (1966) Chem. Abstr. 6 4 , 10478a ((1965) Bull. Inst.

Higher Educ. Chem. Chem. Technol. 8, 734-740, Russian original) Kura, G., Koyama, A. & Tarutani, T. (1977) J . Chromatogr. 144,

Lindqvist, B. (1962) Acta Chem. Scand. 16 , 1794-1798 Luck, W. A. P. (1985) in Water and Ions in Biological Systems

(Pullman, A., Vasilescu, V. and Packer, L., eds) pp. 95-126, Plenum Press, New York

Maniatis, T., Fritsch, E, F. & Sambrook, J. (1982) in Molecular Cloning, A Laboratory Manuul, p. 437, Cold Spring Harbor Labo- ratory, Cold Spring Harbor, NY

290-296

York

Lett. Ed. 18,271-279

47-55

365-376

1329-1338

5136-5148

1269-1279

245-252

Mantel, M. & Nothmann, R. (1977) Analyst 102,672-677 Marczenko, Z. (1976) in Spectrophotometric Determination of Ele-

ments (Ramsay, C. G., Ed) pp. 260-262, John Wiley & Sons, New York

Marsden, N. V. B. (1973) Naturwissenschuften 6 0 , 257 Marsden, N. V. B. & Haglund, A. C. (1984) J. Inclusion Phenom. 2 ,

21-34

Moore. S. & Stein, W. H. (1954) J. Biol. Chem. 211,907-913 Neddermeyer, P. A. & Rogers, L. B. (1968) Anal. Chem. 40 , 755-762 Neddermeyer, P. A. & Rogers, L. B. (1969) Anal. Chem. 41 , 94-102 Nozaki, Y. (1972) Methods Enzymol. 26,43-50 Nyc, J. F. & Mitchell, H. K. (1947) J. Am. Chem. SOC. 6 9 , 1382-1384 Oosting, M. & Reijnders, H. F. R. (1980) Fresenius’ 2. Anal. Chem.

Ortner, H. M. & Pacher, 0. (1972) J. Chromatogr. 71,55-65 Pecsok, R. L. & Saunders, D. (1968) Sep. Sci. Technol. 3, 325-355 Penefsky, H. S. & Warner, R. C. (1965) J. Biol. Chem. 240, 4694-

4702 Pharmacia (1970) Sephudex, Gel Filtration in Theory and Practice,

Pharmacia Fine Chemicals AB, Uppsala Rajagopal, G. & Ramakrishnan, S. (1975) Anal. Biochem. 6 5 , 132-

136 Ruthven, D. M. (1984) Principles of Adsorption and Adsorption Proc-

esses, Wiley Interscience, New York Samoilov, 0. Ya. (1965) Structure of Aqueous Electrolyte Solutions

and the Hydration of Ions, Consultants Bureau, New York (English translation of 1957 Russian original)

301,28-29

Samoilov, 0. Ya. (1957b) Discuss. Faraday SOC. 24,141-146 Samoilov, 0. Ya. (1972) in Water and Aqueous Solutions; Structure,

Thermodynamics, and Transport Processes (Horne, R. A., ed) pp. 597-612, Wiley Interscience, New York

Saunders, D. & Pecsok, R. L. (1968) Anal. Chem. 40,44-48 Schales, 0. & Schales, S. S. (1941) J. Biol. Chem. 140,879-884 Scott, T. A., Jr. & Melvin, E. H. (1953) Anal. Chem. 26 , 1656-1661 Shindler, D. B. & Prescott, L. M. (1979) Anal. Biochern. 97,421-422 Shiryaeva, T. M. (1973) Zauod. Lab. 39,794-795 Shukuya, R. & Schwert, G. W. (1960) J. Biol. Chem. 235,1658-1661 Sinibaldi, M. & Lederer, M. (1975) J. Chromutogr. 107,210-212 Swain, C. G. & Bader, R. F. W. (1960) Tetrahedron 10, 182-199 Ueno, Y., Yoza, N. & Ohashi, S. (1970) J. Chromatogr. 52,321-327 Ujimoto, K. & Kurihara, H. (1981) J. Chromutogr. 208 , 183-200 Washabaugh, M. W. & Collins, K. D. (1983) Anal. Biochem. 134,

Washabaugh, M. W. & Collins, K. D. (1984) J. Biol. Chem. 2 5 9 ,

Washabaugh, M. W. & Collins, K. D. (1986) J. Biol. Chem. 2 6 1 , 5920-5929

Wen, W.-Y. (1982) in Ions and Molecules in Solution; Studies in Physical and Theoretical Chemistry 27 (Tanaka, N., Ohtaki, H., and Tamamushi, R., eds), pp. 45-59, Elsevier Scientific Publishing Co., Amsterdam

Williams, W. J. (1979) Handbook of Anion Determination, Butter- worths, London

Wilson, N. & Greenhouse, V. Y. (1976) J. Chromatogr. 118, 75-82 Yano, Y. & Janada, M. (1980) J. Chromatogr. 200,125-136 Yoza, N., Ogata, T., Ueno, Y. & Ohashi, S. (1971) J. Chromatogr. 6 1 ,

Zeitler, H.-J. & Stadler, E. (1972) J. Chromatogr. 74,59-71

144-152

3293-3298

295-305

Aqueous Solutes 12485