Biochem. J. (1973) 131, 335-342Printed in Great Britain

Effect of Alcohols and Neutral Salt on the Thermal Stabilityof Soluble and Precipitated Acid-Soluble Collagen

By ALLAN E. RUSSELLLeather Industries Research Institute, Rhodes University, Grahamstown, South Africa

(Received 5 July 1972)

The effects of mono- and poly-hydric alcohols in the presence of KCl on the intrinsicstability of collagen molecules in dilute acid solution were compared with correspondingsolvent and salt effects on the increased stability of the aggregated molecules in salt-precipitated fibrils. Salt addition decreased solubility and increased the thermal stabilityof fibrils, but progressively decreased the stability of collagen molecules in solution. Incontrast, the alcohols enhanced solubility and decreased fibril stability, the effects in-creasing with solvent hydrocarbon chain length and with decreasing hydroxyl/methylene-group ratio. Molar destabilization of dissolved collagen by alcohols was lower than forfibrils, and at low salt concentration, both ethylene glycol and glycerol were structuralstabilizers. Electron-micrograph studies indicated that salt-precipitated fibrils tended toadopt the native aggregation mode, and qualitatively similar solvent effects wereobserved in insoluble collagens. Implications of the experimental findings are discussed interms of a model in which electrostatic and apolar interactions mainly govern the excessof stability in collagen fibrils whereas intrinsic stability of single molecules is a functionof polar interactions and polypeptide-chain rigidity.

Previous studies from these laboratories havereported the use of related organic perturbants ascollagen structural probes in attempts to elucidatelyotropic mechanisms and gain an insight into factorsgoverning the intrinsic stability of collagen moleculesin the dispersed state. Perturbant structural andfunctional features such as linear hydrocarbon chainlength, chain isomerism and position and numberof potential hydrogen-bonding groups were foundto constitute the main factors determining solventeffects on thermal stability and renaturation kineticsof collagen in solution (Russell & Cooper, 1969a,b,1970, 1972; Hart et al., 1971; Cooper et al., 1971).Similar solvent effects have been reported by otherworkers for soluble collagen (Herbage et al., 1968;Schnell, 1968; Harrap, 1969; Bianchi et al., 1970) andglobular proteins (Schrier & Scheraga, 1962;Herskovits et al., 1970a,b,c; Kaminsky et al., 1972).

In extending these solvent studies to the molecularaggregates present in fibrous collagen, it was foundthat certain perturbants differed markedly in theiractivities in the soluble and insoluble collagensystems. Thus, ethylene glycol, glycerol and propane-1,3-diol progressively stabilized collagen in solution(Harrap, 1969; Hart et al., 1971), but destabilizedmature fibrous collagen at the low-solvent extreme(Russell & Cooper, 1971). Although relative activitiesand the influence of solvent structure on perturbantactivity in soluble and insoluble collagens werequalitatively similar for the other solvents examined,

Vol. 131

molar activities at infinite dilution were consistentlygreater in insoluble collagen.The different response of collagen in the dispersed

and aggregated states toward the glycols and glycerolsuggests that different factors predominate in thestabilization of the two forms. Accordingly, in thepresent study, the phenomenon and its structuralimplications have been further explored. In particular,the effects of the lower mono- and poly-hydricalcohols in the presence of neutral salt have beencompared with respect to their effects on the thermalstability of soluble, salt-precipitated and insolublecollagens as a function of solvent composition.

Experimental

Materials

Acid-soluble calf skin collagen was extracted andpurified by the method previously described afterremoval of the neutral-salt-soluble fraction (Cooper& Davidson, 1965). Both the acid-soluble collagenand the insoluble collagen fragments recovered afterextraction were neutralized, dialysed free of saltsand freeze-dried. An area of intact insoluble collagenfor shrinkage-temperature studies was prepared fromlimed and acetone-dehydrated sheep skin as previous-ly described (Russell & Cooper, 1971) to give acollagen matrix conforming to standard commercialhide powder [Official Methods of Analysis (1965)

335

A. E. RUSSELL

Society of Leather Trades' Chemists, Redbourn].Reagents used were A.R. grade or laboratory-gradematerials.

Measurement of thermal-transition temperatures'Melting' temperature of acid-soluble collagen in

solution. 'Melting curves' were obtained by polari-metry by using the '30min method' (von Hippel &Wong, 1965) on collagen solutions containing approx.1 mg of dry protein/ml in potassium acetate buffercontaining various concentrations ofKCl and organicsolvent within the range of protein and salt solubilityand 0.15M with respect to total acetate concentrationin the final mixture (pH4.8 in pure aqueous solution).The temperature at the midpoint of the transitionwas taken as the 'melting' point.

Dispersion temperature of salt-precipitated acid-soluble collagen. Fibres of acid-soluble collagenwere precipitated by mixing equal volumes of a solu-tion of acid-soluble collagen (2mg/ml) in 0.15m-potassium acetate buffer (pH4.8) and a solution oflM-KCl in 0.15M-potassium acetate buffer. Thecollagen was thus allowed to precipitate in 0.15M-potassium acetate buffer at ambient temperature at afinal composition of lmg of protein/ml and 0.5M-KC1, when precipitation was substantially completein a few minutes. Precipitations were also carried outat lower KCI concentrations (0.25, 0.125M), requiringcorrespondingly longer precipitation times. Thermal-stability measurements were carried out as previouslydescribed (Russell & Cooper, 1971) by suspending asmall portion of the precipitated fibres in aqueousorganic solvent media of various compositions in-corporating 0.15 M-potassium acetate buffer and vari-ous concentrations of KCI. Addition of KCI to theorganic solvent-buffer medium was necessary tosuppress protein solubility, since in the absence ofsalt,the precipitate tended to redisperse. The suspensionwas heated (0.5-1°C/min) in a test tube supportedin a beaker of water on a magnetic stirrer/hotplateand the temperature of incipient dissolution of thefibres as gelatin was noted.

Shrinkage temperature of mature collagen fibres.The temperature of incipient length contraction (or'shrinkage temperature') ofmature insoluble collagenstrips exposed to various aqueous organic solventcompositions incorporating 0.15M-potassium acetatebuffer and various concentrations of KCl, wasmeasured as previously described [Russell & Cooper(1971); Official Methods of Analysis (1965) Societyof Leather Trades' Chemists, Redbourn]. With ex-tracted insoluble calf skin fragments recovered afterthe preparation of acid-soluble collagen, shrinkagetemperatures were measured by observing dimen-sional changes in individual fibres exposed to solventmedia by using a Reichert 'Thermopan' microscopefitted with a Kofler hot stage. A portion of dry

collagen was shaken with excess of solvent medium;the wetted fibres were placed in the centre of a ringof light oil on a microscope slide and sealed with acover slip. The temperature was then raised at acontrolled rate (0.5-1°C/min) and the temperatureof incipient dimensional change noted.

Electron microscopy

The collagen fibrils precipitated at various KCIconcentrations were positively stained in 1% (w/v)phosphotungstic acid adjusted to pH6.0-6.3 byaddition of 1% (w/v) NaOH. A drop of fibril sus-pension was placed on a formvar-coated grid onfilter paper, the grid was flooded twice with waterto remove salts and drained. The grid was then heldin staining solution for 30-60s and subsequentlywashed by immersion in water four times anddrained on filter paper. The positively stained fibrilswere examined in the Hitachi transmission electronmicroscope.

Results

Effect of ethylene glycol and glycerol at various KClconcentrations on thermal stability ofdissolvedandsalt-precipitated acid-soluble collagen

Variations in transition temperatures of the dis-solved and salt-precipitated forms of acid-solublecollagen, respectively, with solvent concentration atvarious concentrations of added KCI are shown inFig. 1. At a given salt concentration, the dispersiontemperature of the precipitated collagen decreasedas the ethylene glycol or glycerol concentration in-creased up to the limiting solvent concentration atwhich the precipitate redissolved at ambient tem-perature. Similar destabilization by ethylene glycoland glycerol at low solvent proportions in theabsence of salt was noted previously with maturesheep skin collagen (Russell & Cooper, 1971). At agiven solvent concentration, however, thermal stabil-ity of the precipitate increased as the concentrationof added KCl increased, and the limiting solventconcentration for dissolution of the precipitate atambient temperature shifted to higher solvent pro-portions. Molar destabilization effects at the varioussalt concentrations examined were similar for bothethylene glycol and glycerol. No significant differ-ences were apparent when effects of ethylene glycolon fibrils precipitated in 0.5M- and 0.25M-KCI werecompared, although precipitation rates at the highersalt concentration were more rapid (Fig. la).

Specific-rotation measurements in the solutionsobtained at temperatures fractionally above the fibril-dispersion temperatures yielded values correspondingto those of gelatin ([OC] -460o), indicating thatfibril dissolution under these conditions resulted from

1973

336

STABILITY OF SOLUBLE AND PRECIPITATED ACID-SOLUBLE COLLAGEN

60 F

50

so 40Ii-

0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

Concn. of ethylene glycol (M)

(b)

I I ' W30

0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

Concn. of glycerol (M)

Fig. 1. Effect of(a) ethylene glycol and (b) glycerol at various concentrations of added KCl on thermal stability ofprecipitated and dissolved acid-soluble calfskin collagen

The solvent medium was 0.15M-potassium acetate buffer (pH4.8) containing various amounts of KCI. Opensymbols correspond to experimental values obtained for collagen fibrils precipitated initially in 0.5M-KCl(see the Experimental section); identical trends were obtained for fibrils formed in 0.25M-KCI. Closed symbolscorrespond to values for dissolved collagen. Discontinuities marked by vertical broken lines denote approxi-mate limiting solvent concentrations at which precipitated collagen redissolved. Back-extrapolated broken lines

represent hypothetical trends for soluble collagen. Concn. of KCI: *, 0, 1.OM; A, A, 0.75M; *, 0, 0.5M; v,

v, 0.25M; *, OM.

a helix-to-coil transition in the material. On the otherhand, optical-rotation measurements at ambienttemperature on the material in solution at solventconcentrations above the fibril-dissolution limit gavespecific-rotation values corresponding to those ofnative collagen (x]`5-1330o), indicating that normalprotein dissolution had occurred under these con-ditions without accompanying internal structuralchange.Both solvent and salt effects on the dissolved

collagen differed markedly from correspondingeffects on the precipitated form. Thus, addition ofKCI progressively lowered the stability of the dis-persed collagen at all organic solvent concentrations,whereas at low salt concentration, ethylene glycol andglycerol acted as structural stabilizers, as noted pre-viously in the absence of neutral salt (Harrap, 1969;Hart et al., 1971). Solvent stabilization effects de-creased with salt addition, however, and at higherconcentrations of KCI, ethylene glycol and the iso-meric propanediols (1.OM-KCI; Figs. la and 4) werestructural destabilizers. Comparison of plot gradientson either side of the fibril-dissolution limit indicatedthat molar destabilization of soluble collagen by thesesolvents was considerably less than for the pre-cipitated form.A tentative back-extrapolation of the approxi-

mately linear trends observed for soluble collagen tozero ethylene glycol concentration is shown in Fig.l(a), and a similar extrapolation of the curved plots

Vol. 131

obtained with glycerol was attempted in Fig. 1(b).The divergent trends apparent in both cases reflectan increase in molar destabilization by KCI as theconcentration of solvent increased.

Comparison of effects of ethylene glycol and KCI on

thermal stability ofinsoluble andprecipitated collagens

Qualitatively similar effects were apparent on com-paring ethylene glycol activity at various concentra-tions of KCI (Fig. 2) on precipitated acid-solublecollagen, extracted insoluble calfskin collagen (micro-scopic methods) and mature sheep skin collagen(macroscopic shrinkage method). As found abovefor precipitated collagen, thermal stability of theinsoluble collagens decreased with solvent addition,but increased with salt addition. With mature sheepskin collagen a converging trend with increasingsolvent concentration was apparent, resulting in inter-section of the plots at higher solvent proportions.Similar but less pronounced convergence was alsoevident with the precipitated and insoluble calf skincollagens.

Quantitatively, transition temperatures for the pre-cipitated collagen were approx. 5°C higher than thoseof the insoluble collagen from which it was extracted.Shrinkage temperatures of the mature sheep skincollagen, however, were approx. 12°C higher thanthose of the precipitated calf skin collagen at thesame salt concentration.

M

I.-

044a)0U,-

337

14

"I

A. E. RUSSELL

60

C600

0

4-a

I I } f I I_0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0

Concn. of ethylene glycol (M)

Fig. 2. Comparison of ethylene glycol activity atvarious concentrations ofaddedKCI on thermal stabil-ity ofprecipitated acid-soluble collagen and insoluble

fibrous collagens

0

950 _ ''40)

Ce~~~~~~----- ----

30 ,.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

Concn. of solvent (M)

Fig. 3. Effect of solvent hydrocarbon chain length atconstant hydroxyl/methylene-group ratio on thermalstability ofprecipitated and dissolved acid-soluble calf

skin collagen

The solvent medium was 0.15M-potassium acetatebuffer (pH4.8) containing l.OM-KCI. For other de-tails see Fig. 1. Solvents: o, methanol; A, A, ethyleneglycol; *, o, glycerol.

The solvent medium was 0.15M-potassium acetatebuffer (pH4.8) containing various amounts of KCI.Values for insoluble sheep skin collagen (o, A) weredetermined by the macroscopic shrinkage method;values for acid-soluble calf skin collagen precipitatedin 0.5M-KCI (O, A) and extracted insoluble calf skincollagen (e, A) were determined by the microscopicmethod (see the Experimental section). o, 0, *,1.OM-KCI; A, A, A, 0.5M-KCI.

Electron-microscope studies

Collagen fibrils precipitated at various concentra-tions of KCl were examined in the electron micro-scope to determine the effect of salt on aggregationmode. At initial collagen concentrations of 1 mg/mlin 0.15M-potassium acetate buffer, precipitation in0.5M-KCI was virtually instantaneous, whereas at0.25M- and 0.125M-KCI, turbidity development waspreceded by induction periods of approx. 1 h and 4hrespectively.

In 0.125M-KCI, extended fibrils were obtained witha higher degree of native 64nm (640A) striationapparent. Microfibrillar strands comprising the largerfibrils and showing a small degree of twisting werealso discernible.At 0.25M-KCI a mass of entangled, twisted fibrils

was obtained in which striation appeared to beobscured. Native striations were apparent in thethinner less-dense fibrils, showing various degreesof twisting.

At 0.5M-KCI a high yield of tightly twisted fibrilsof extended length was obtained without visiblestriation.

Structuralfactors in solvent activity

The effects of methanol, ethylene glycol andglycerol on thermal stability of precipitated anddissolved acid-soluble collagen are compared at 1.OM-KCl in Fig. 3. Molar destabilization of precipitatedcollagen by the three solvents was comparable not-withstanding the change in molecular size. As notedabove, however, solvent activity varied significantlywith collagen in solution, where glycerol was astructural stabilizer and ethylene glycol a destabilizerunder these conditions. The precipitated collagenremained insoluble over the methanol concentrationrange examined.The effect of decreasing ratio of solvent hydroxyl/

methylene groups is shown in Fig. 4, in which theactivities of glycerol, propane-1,2-diol, propane-1,3-diol and propan-1-ol are compared. A progressiveincrease in molar destabilization of the precipitatedcollagen was apparent as the hydroxyl/methylene-group ratio decreased. Positional isomerism had littleinfluence on propanediol activities, which were com-parable. With collagen in solution, however, propane-1,2-diol was the more active isomer, whereas aprogressive increase in destabilization with decreasinghydroxyl-group content was suggested by comparison

1973

338

STABILITY OF SOLUBLE AND PRECIPITATED ACID-SOLUBLE COLLAGEN

o.~ ~ ~ Cnn sovnoM

00

'0

30 .-.I-

0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

Concn. of solvent (m)

Fig. 4. Effect ofdecreasing solvent hydroxyl/methylene-group ratio at constant hydrocarbon chain length on

thermal stability of precipitated and dissolved acid-soluble calfskin collagen

The solvent medium was 0.15M-potassium acetatebuffer (pH4.8) containing 1.OM-KC1. For other de-tails see Fig. 1. Solvents: *, o, glycerol; A, A,

propane-1,3-diol; *, o, propane-1,2-diol; v, propan-1-ol.

60

50

'0

V

040

.

H

30L1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

Concn. of solvent (M)

Fig. 5. Effect of increasing hydrocarbon chain lengthin monohydric alcohols on thermal stability of pre-

cipitated acid-soluble calf-skin collagen

The solvent medium was 0.15M-potassium acetatebuffer (pH4.8) containing 1.OM-KCI. No fibril dis-solution occurred over the experimentally accessiblerange. Solvents: o, methanol; A, ethanol; o, propan-l-ol.

of glycerol and the propanediols. Solvent immiscibil-ity prevented extension of the results for propan-1-ol.The effect of hydrocarbon chain length on activity

Vol. 131

of monohydric alcohols is shown in Fig. 5. A pro-gressive increase in molar destabilization of theprecipitated collagen with increasing chain lengthwas apparent, as noted in mature sheep skin collagen(Russell & Cooper, 1971) and acid-soluble calf skincollagen in solution (Russell & Cooper, 1969a).Molar activity decreased progressively as the solventconcentration increased. No protein-solubility limitwas apparent over the extended range of solventconcentrations examined, so that solvent effects onsoluble collagen under these conditions were notmeasurable.

Discussion

The results of the present study indicate thatcollagen aggregation to form fibrils in vitro causes asignificant gain in thermal stability relative to thatfor single molecules under the same conditions. Aninsight into the origin of the stability increase andthe underlying factors involved may be gained fromconsideration of the different responses of collagenin the dispersed and aggregated states to solvent andneutral-salt perturbants.

Salt effects on soluble andprecipitated collagensThe effects of KCI in accelerating collagen pre-

cipitation and stabilizing the resultant fibrils aresuggestive of an increase in protein-protein inter-action in acid medium in the presence of salt. Similarcorrelation between formation rate and fibril stabilityas influenced by various environmental factors suchas pH value, temperature and ionic strength has beennoted previously (Cooper, 1970). At physiologicalpH values, however, neutral-salt effects on collageninteraction appear to differ considerably from thosein acid medium. Thus Bensusan & Hoyt (1958) andWood & Keech (1960) found a progressive decreasein precipitation rate with increasing ionic strength asa result of salt addition in the pH range 7.0-8.5.Similarly in the present study, addition of KCI toprecipitated and insoluble collagens in alkalinemedia produced slight destabilization, but markedlystabilized both forms at acid pH values (Table 1).These pH-dependent salt effects indicate that

charge interactions are implicated in collagen-fibrilaggregation and stabilization. In acid media at lowsalt concentration, mutual repulsion between collagenmolecules carrying an excess of positive chargewould be expected to minimize aggregation. As theKCl concentration is increased, however, preferentialanion binding may decrease the cationic characterof the protein, resulting in a progressive precipitationof isoelectric collagen at rates governed by Cl- ionconcentration. Electrophoretic evidence for such pre-ferential binding for Cl- ions on collagen has been

339

A. E. RUSSELL

Table 1. Effect ofKCI on thermal stability ofprecipitated and insoluble collagens in acidic and neutralpH ranges

Values reported are the dissolution and shrinkage temperatures for precipitated acid-soluble calf skin collagenand insoluble sheep skin collagen respectively. For details see the text.

Temperature (°C)

0.15M-Potassium acetatebuffer, pH4.8

0.15M-Potassium acetatebuffer, pH8.5

PrecipitatedInsoluble acid-solublecollagen collagen60.062.363.066.067.0

52.052.051.651.251.0

presented by Bensusan & Hoyt (1958). Conversely,dissolution of precipitated fibrils will also dependon concentration of Cl- ions available to maintainelectrical neutrality. It follows that additionalstructural stabilization in fibrils can be expectedarising from an increase in polypeptide-chain rigidityin the associated molecules, the effect increasing withsalt addition. At pH values above the isoelectricpoint, however, when the protein carries an excessof negative charge, no comparable effects due to C1-ion binding can be expected.A similar mechanism of charge interaction may be

invoked to account for the effects of KCl on thethermal stability of insoluble collagens, in which dis-solution is prevented by the presence of covalentcrosslinks (reviewed by Veis, 1967). Variation inthermal stability with salt concentration was qualita-tively similar to that noted for fibrils precipitated in0.5M- and 0.25M-KCI, the latter showing evidenceof the striated, native aggregation mode. The largernumber of covalent crosslinks conferring rigidity inmature sheep skin would appear to account for thegenerally higher stability of this material relative tothe precipitated and insoluble calf skin collagens.Reasons for the slightly lower stability of insolublecalf skin collagen compared with the precipitatedmaterial are not apparent.

Destabilization of acid-soluble collagen by KClin alcohol-buffer media is consistent with previousreports oflyotropic effects ofneutral salts on collagen(reviewed by von Hippel, 1967) and other proteins(reviewed by von Hippel & Schleich, 1969). Althoughlyotropic mechanisms remain controversial, neutral-salt effects are currently considered to involve ion-dipole association orhydrogen-bonding ofa hydratedspecies at the peptide bond leading to a decrease indouble-bond character and consequent loss in poly-

peptide-chain rigidity. Evidence for such interactionstems from neutral-salt effects on conformationalchanges in poly-L-proline, which lacks internal hydro-gen bonding (Schleich & von Hippel, 1969). Alterna-tively, for proteins, alteration of water structure byions has also been suggested to favour the exposureof internal hydrophobic groups to the solvent, result-ing in unfolding of the native structure.For individual collagen molecules in dilute solu-

tion, a lyotropic mechanism involving changes in side-chain solvation is less plausible, since all side chainsare located on the molecular surface (Ramachandran,1967) and hence are relatively accessible to the solventin both the native and denatured states. Alterna-tively, a mechanism involving binding of ions atpeptide bonds in soluble collagen with charge inter-actions constituting an additional factor in precipi-tated and insoluble collagen, would account for thedistinctive effects of KCl on protein stability in thetwo states.

Alcohol effects on soluble andprecipitated collagens

Destabilization of precipitated acid-soluble col-lagen and the influence of hydrocarbon structure andhydroxyl-group substitution on solvent activity in thepresent study were consistent with effects of aqueousalcohols at low proportion on insoluble collagenpreviously reported (Schnell & Zahn, 1965; Russell& Cooper, 1971). Analogous activity-structurerelationships in retardation of collagen precipitationby monohydric alcohols (Bensusan, 1960) suggeststhat these solvent effects on stability and fibril forma-tion reflect decreases in protein-protein interactionas a result of competitive solvent binding in pre-

dominantly aqueous media. Evidence for directsolvent-protein interaction moderated by hydro-

1973

Concn. ofKC1(M)00.250.500.751.00

Precipitatedacid-solublecollagen43.646.048.551.553.5

Insolublecollagen

63.562.361.059.760.5

340

STABILITY OF SOLUBLE AND PRECIPITATED ACID-SOLUBLE COLLAGEN 341

carbon chain length, has been reported in studies ofthe binding of alcohol homologues to collagen mem-branes (Bianchi et al., 1970).The influence of hydrocarbon chain length on

lyotropic activity has led previous workers to con-clude that alcohols disrupt internal hydrophobicbonding in proteins generally (Kauzman, 1959;von Hippel & Wong, 1965; Herskovits et al., 1970a)and, by analogy, in soluble and insoluble collagens(Schnell & Zahn, 1965; Herbage et al., 1968; Schnell,1968; Harrap, 1969). For collagen fibrils, the positivetemperature, dependence of formation (Gross, 1958;Cooper, 1970; Cassel, 1971) is indicative of an endo-thermic, and hence entropy-driven, process, implicat-ing solvent-labile hydrophobic interactions betweenapolar side chains in adjacent molecules in aggrega-tion and stabilization.A qualitative difference between the stabilizing

factors predominating in individual molecules andcollagen fibrils is suggested by the observation thatactivities for ethylene glycol and glycerol and for theisomeric propanediols differed substantially forcollagen in solution, but were comparable for fibrils.Moreover, molar effects of the solvents for precipit-ated fibrils were consistently greater than for isolatedmolecules with reversal of effects occurring in thelatter in some cases.For single collagen molecules in dilute solution, the

external location of polar and apolar side chainssuggests that structural stability is primarily a func-tion of interpeptide hydrogen bonding and chainrigidity conferred by rotational restrictions at pyrrol-idine residues and peptide bonds (reviewed byRamachandran, 1967). Recent calorimetric studiesof denaturation, however, indicate that internalhydrogen bonding does not contribute greatly totropocollagen stability (Privalov & Tikopulo, 1970;Cooper, 1971). Alternatively, the observation thatsolvent binding to peptide groups induces conforma-tional changes in poly-L-proline (Strassmair et al.,1969) suggests that as for neutral salts, polar inter-actions of solvents can induce similar changes inbond rotation in collagen. The role of hydrocarbonstructure in the activity of aqueous binary solventsin non-protein systems lacking hydrophobic bonds(Pittz & Bello, 1969; Gerlsma, 1970; Russell &Cooper, 1971) indicates that polar interactions of thesolvent may be influenced through the orderingeffects of the hydrocarbon chain on local waterstructure (Russell & Cooper, 1969a, 1970).Comparison of propanediol activities on collagen

in solution with those of ethylene glycol and glycerolin the present study confirmed that molar activitydecreased with increasing hydroxyl/methylene-groupratio as noted previously (Harrap, 1969; Hart et al.,1971). Thus, for weakly active solvents of higherpolar/methylene-group ratio such as glycerol andethylene glycol, stabilization of dissolved collagen

with increasing solvent concentration appeared toreflect a further effect, namely, a progressive strength-ening of internal hydrogen bonding as a result of adecrease in medium dielectric constant. Significantly,when the dielectric constant was increased by saltaddition, stabilization effects decreased and ethyleneglycol became an active destabilizer.

In general, the solvent and salt effects observed areconsistent with a model in which long-range electro-static forces and hydrophobic interactions governaggregation and the resultant excess of stability incollagen fibrils, whereas the intrinsic stability ofindividual molecules is a function of polypeptide-bond rotational constraints and, under conditions oflow dielectric constant particularly, of intramolecularpolar interactions.

The assistance of Mr. R. Cross of the Rhodes UniversityElectron-Microscopy Unit is gratefully acknowledged.This work was supported by the annual grants of theSouth African Livestock and Meat Industries ControlBoard and the Council for Scientific and IndustrialResearch.

References

Bensusan, H. B. (1960) J. Amer. Chem. Soc. 82, 4995-4998

Bensusan, H. B. & Hoyt, B. L. (1958) J. Amer. Chem. Soc.80, 719-724

Bianchi, E., Rampone, R., Tealdi, A. & Ciferri, A. (1970)J. Biol. Chem. 245, 3341-3345

Cassel, J. M. (1971) Biophys. Prop. Skin 1, 63-100Cooper, A. (1970) Biochem. J. 118, 355-365Cooper, A. (1971) J. Mol. Biol. 55, 123-127Cooper, D. R. & Davidson, R. J. (1965) Biochem. J. 97,

139-147Cooper, D. R., Russell, A. E. & Hart, G. J. (1971)

Biochem. J. 125, 1069-1074Gerlsma, S. Y. (1970) Starke-Starch 22, 3-9Gross, J. (1958) J. Exp. Med. 108, 215-226Harrap, B. S. (1969) Int. J. Protein Res. 1, 245-252Hart, G. J., Russell, A. E. & Cooper, D. R. (1971)

Biochem. J. 125, 599-604Herbage, D., Hanus, A. & Vallet, G. (1968) Biochim.

Biophys. Acta 158, 544-554Herskovits, T. T., Gadegbeku, B., & Jaillet, H. (1970a)

J. Biol. Chem. 245, 2588-2598Herskovits, T. T., Jaillet, H., & Gadegbeku, B. (1970b)

J. Biol. Chem. 245, 4544 4550Herskovits, T. T., Jaillet, H. & De Sena, A. T. (1970c)

J. Biol. Chem. 245, 6511-6517Kaminsky, L. S., Yong, F. C. & King, T. E. (1972)

J. Biol. Chem. 247, 1354-1359Kauzman, W. (1959) Advan. Protein Chem. 14, 1-60Pittz, E. & Bello, J. (1969) Trans. Faraday Soc. 66, 537-545Privalov, P. L. & Tikopulo, E. L. (1970) Biopolymers

9, 127-139Ramachandran, G. N. (1967) in Treatise on Collagen(Ramachandran, G. N., ed.), vol. 1, pp. 103-183,Academic Press, New York

Vol. 131

342 A. E. RUSSELL

Russell, A. E. & Cooper, D. R. (1969a) Biochemistry 8,3980-3990

Russell, A. E. & Cooper, D. R. (1969b) Biochem. J. 113,221-223

Russell, A. E. & Cooper, D. R. (1970) Biochemistry 9,2802-2806

Russell, A. E. & Cooper, D. R. (1971) Biochemistry 10,3890-3896

Russell, A. E. & Cooper, D. R. (1972) Biochem. J. 127,855-863

Schleich, T. & von Hippel, P. H. (1969) Biopolymers 7,861-877

Schnell, J. (1968) Arch. Biochem. Biophys. 127, 496-502Schnell, J. & Zahn, H. (1965) Makromol. Chem. 84,

192-202

Schrier, E. E. & Scheraga, H. A. (1962) Biochim. Biophys.Acta 64, 406-408

Strassmair, H., Knof, S. & Engel, J. (1969) Hoppe-Seyler'sZ. Physiol. Chem. 350, 1153-1154

Veis, A. (1967) in Treatise on Collagen (Ramachandran,G. N., ed.), vol. 1, pp. 367439, Academic Press, NewYork

von Hippel, P. H. (1967) in Treatise on Collagen(Ramachandran, G. N., ed.), vol. 1, pp. 326-338,Academic Press, New York

von Hippel, P. H. & Schleich, T. (1969) Struct. Stabil.Biol. Macromol. 2, 560-574

von Hippel, P. H. & Wong, K.-Y. (1965) J. Biol. Chem.240, 3909-3923

Wood, G. C. & Keech, M. K. (1960) Biochem. J. 75, 588-598

1973