doi:10.1016/j.jmb.2004.12.001 J. Mol. Biol. (2005) 346, 551–556

The Increase in Denaturation Temperature FollowingCross-linking of Collagen is Caused by Dehydration ofthe Fibres

Christopher A. Miles1*, Nicholas C. Avery1, Victor V. Rodin2 andAllen J. Bailey1

1Collagen Research GroupUniversity of Bristol, LangfordBristol BS40 5DU, UK

2MATI Russian StateTechnological University3, Orshanskaya str, Moscow121552, Russian Federation

0022-2836/$ - see front matter q 2004 E

Abbreviations used: MDA, malonhexamethylene diisocyanate; DSC,calorimetry.E-mail address of the correspond

chris.miles@bristol.ac.uk

Differential scanning calorimetry (DSC) was used to study the thermalstability of native and synthetically cross-linked rat-tail tendon at differentlevels of hydration, and the results compared with native rat-tail tendon.Three cross-linking agents of different length between functional groupswere used: malondialdehyde (MDA), glutaraldehyde and hexamethylenediisocyanate (HMDC). Each yielded the same linear relation between thereciprocal of the denaturation temperature in Kelvin, Tmax, and the watervolume fraction, 3 (1/TmaxZ0.0007313C0.002451) up to a critical hydrationlevel, the volume fraction of water in the fully hydrated fibre. Thereafter,water was in excess, Tmax was constant and the fibre remained unchanged,no matter how much excess water was added. This Tmax value and thecorresponding intrafibrillar volume fraction of water were as follows:84.1 8C and 0.48 for glutaraldehyde treated fibres, 74.1 8C and 0.59 forHMDC treated fibres, 69.3 8C and 0.64 for MDA treated fibres, and 65.1 8Cand 0.69 for untreated native fibres. Borohydride reduction of the nativeenzymic aldimines did not increase the denaturation temperature of thefibres. As all samples yielded the same temperature at the same hydration,the temperature could not be affected by the nature of the cross-link otherthan through its effect on hydration. Cross-linking therefore causeddehydration of the fibres by drawing the collagen molecules closertogether and it was the reduced hydration that caused the increasedtemperature stability. The cross-linking studied here only reduced thequantity of water between the molecules and did not affect the waterin intimate contact with, or bound to, the molecule itself. The enthalpyof denaturation was therefore unaffected by cross-linking. Thus, the“polymer-in-a-box” mechanism of stabilization, previously proposed toexplain the effect of dehydration on the thermal properties of nativetendon, explained the new data also. In this mechanism, the configura-tional entropy of the unfolding molecule is reduced by its confinement inthe fibre lattice, which shrinks on cross-linking.

q 2004 Elsevier Ltd. All rights reserved.

Keywords: collagen; DSC; cross-link; stability; polymer-in-a-box

*Corresponding author

Introduction

The covalent cross-linking of collagen moleculesin connective tissues, such as tendon, bone and skinis essential for proper tissue function. In the absence

lsevier Ltd. All rights reserve

dialdehyde; HMDC,differential scanning

ing author:

of these intermolecular cross-links, the molecules inthe fibre are capable of sliding relative to each other;consequently, the fibre is weak and extensible. Suchan effect is well known to be induced by lathyriticagents that inhibit the oxidative-deamination ofspecific lysine residues in the first step of cross-linkformation by one or more isoforms of the enzymelysyl oxidase.1,2 Impaired cross-link formation isalso seen in the connective tissues of oim mice, amodel for a variant of Osteogenesis imperfecta inwhich the a2 chain is absent resulting in thehomotrimer [a1(I)]3 forming the functional fibre.

d.

Figure 1. The denaturation temperature of collagen inrat-tail tendon measured at different hydrations andcross-linked with different agents: (a) glutaraldehyde, (b)HMDC, (c) MDA. Squares represent the denaturationtemperatures of the native tendon before cross-linking.Circles and triangles denote cross-linked tendons fromdifferent rats. The continuous lines represent theequations quoted previously for native tendon.15 Thickline (intermediate hydration): 1/TmaxZ0.0007313C0.002451; thin line (high hydration): 1/TmaxZ0.002956.The dotted lines are the means of fully hydrated cross-linked fibres. All scans at 10 8C/minute.

552 Thermal Stability of Cross-linked Collagen Fibres

Oim collagen fibres are abnormally hydrated, withlarger than normal interaxial separations of thecollagen molecules, and this is a possible cause ofthe reduced cross-linking.3 Increased cross-linkconcentrations and a change in the quality andtype of the cross-linking and tissue hydration areobserved with maturation, and a further increaseoccurs with ageing of mature tissue by non-enzymicglycation cross-linking (for a review see Bailey et al.4).

The lysine-aldehyde derived cross-links formedin the early stages of life simply connect onemolecule head to tail to another. The concentrationsof these immature cross-links increase initially, butare then replaced by “mature” cross-links formedby making a further cross-link between an imma-ture cross-link and a third residue.4 These maturecross-links may therefore connect two or threemolecules depending on whether the third residueis on a different molecule or on a different alphachain within the same molecule. Cross-linking to athird molecule must occur transversely with themolecules in register, which would account forfurther increase in mechanical strength.

The accumulation of increased concentrations ofmature cross-links with age is associated with amarked increased thermal stability5 that may bemonitored by calorimetry6,7 or hydrothermal iso-metric tension measurement.8 Analysis of skin fromsubjects with diabetes showed increased tempera-ture stability over controls in humans,9 reflecting anaccelerated ageing. A similar effect has beenobserved in experimental diabetes in rats10,11 andthis increase in denaturation temperature can befollowed by cross-linking with glucose in vitro,which also increases the thermal denaturationtemperature.12 Historically in vitro cross-linkingwith a variety of agents has been used to stabilizecollagen, for example hides for leather13 andtendons for high tensile biomaterials,14 but themechanism underlying the increased denaturationtemperatures at the molecular level is unknown.

At least two factors may be considered aspotentially important. The first mechanism is aloss of chain entropy brought about by the reducednumber of molecular configurations available dueto molecules being increasingly attached covalently.The second is dehydration. Cross-links mightdehydrate the fibre by closer binding of themolecules. Indeed, dehydration caused by subject-ing the fibre to a lowwater activity environment hasbeen shown to result in very substantial increases inthe temperature stability of collagen fibres.15,16

Here we have attempted to differentiate betweenthe mechanisms experimentally, based on thesepreviously studied effects of hydration (at a fixedlevel of cross-linking) on the thermal stability ofrat-tail tendon.15 We have measured the thermalstability at different levels of hydration in rat-tailtendon cross-linked to different degrees and com-pared the results with those obtained inthe previous work. We thereby deduce that dehy-dration is the dominant mechanism increasing thethermal stability of cross-linked collagen fibres.

Results

Chemical cross-linking

When samples of rat-tail tendon, cross-linked

Thermal Stability of Cross-linked Collagen Fibres 553

with bifunctional agents of different length, werescanned by differential scanning calorimetry (DSC),the temperature of the characteristic collagendenaturation endotherm changed with hydrationas shown in Figure 1. In all cases there was a linearrelation between 1/Tmax and the water volumefraction up to a critical hydration level (the volumefraction of water in the fully hydrated fibre).Thereafter, water was in excess and Tmax wasconstant and the fibre remained unchanged nomatter how much excess water was added. Thislevel differed between the different treatments, andis represented by the intersection of the dashed linewith the solid line in each case. Thus, theglutaraldehyde treated fibre possessed the highesttemperature stability with TmaxZ84.1 8C and thelowest volume fraction of water of 0.48, and wasconsequently the most closely packed. The volumefraction of water in the HMDC and MDA cross-linked fibres in excess water were 0.59 and 0.64,respectively, and the corresponding temperatures74.1 8C and 69.3 8C. It is emphasized that thecontinuous lines in these Figures are the least-squares lines reported previously15 for native rat-tail tendon and clearly fit the present data, showingthat the temperature stability of cross-linked fibresis determined only by the hydration of the fibresand not by any other property of the cross-linkitself.

The enthalpy of denaturation of the fibres is alsoshown (Figure 2), revealing that the enthalpy of thefully swollen cross-linked fibres and of fibreshydrated beyond a critical level is constant andthis value is the same as reported for native rat-tailtendon. Below a given level of hydration, theenthalpy falls in both cross-linked fibres and nativefibres.

Figure 2. The denaturation enthalpy of collagen in rattail tendon fibres measured at different hydrations andcross-linked with different agents: (a) glutaraldehyde,

Stabilization of native enzymic cross-links

Native rat-tail tendon with and without boro-hydride reduction of the enzymic aldimine cross-link yielded the same denaturation temperature(63.5 8C and 64.7 8C) and enthalpy (55.3 J/g and59.8 J/g), respectively, when measured in water.

(b) HMDC, (c) MDA. Squares denote the native tendonbefore cross-linking. Circles, triangles and crosses arecross-linked tendons from different rats. The continuousline is the enthalpy of native rat-tail tendon at highhydration recorded previously,15 58.6 J gK1, now found toalso equate with those of artificially cross-linked collagenat high hydrations.

Discussion

The experiment has clearly demonstrated thatthe increased temperature stability of collagen incross-linked fibres is determined mainly by theintrafibrillar water content. Cross-linking causesdehydration of the fibres and it is the reducedhydration that causes the increased temperaturestability. Equally the results show that fibres cross-linked by different agents yield exactly the sametemperature and enthalpy of denaturation pro-vided they are measured at the same intrafibrillarwater content. Cross-linking affects the denatura-tion temperature of fibres fully saturated with waterbecause it reduces the extent to which fibres canswell in a given environment. More generally, in a

particular environment, fibres with different cross-linking will equilibrate to different intrafibrillarfluid contents and therefore have different tem-perature stabilities (see Figure 1).A similar effect is seen during the ageing of the

collagen fibre as the cross-links mature. Thetrivalent structures of the pyridinoline and pyrrolemature cross-links (e.g. Figures 6 and 7 of Bailey

554 Thermal Stability of Cross-linked Collagen Fibres

et al.4) would clearly draw the collagen moleculescloser together than the bivalent immature cross-links. This would explain the correlation betweenthe temperature stability of calf tendon and pyri-dinoline or pyrrole concentrations.5 Further, ageingof mature fibres is believed to occur through non-enzymic glycation4 when further increases indenaturation temperature are observed.12

Increasing the stability of the mature aldiminebond by borohydride reduction does not affect thelength of the bond between the twomolecules in thenative fibre. Thus, the fact that reduction had noappreciable effect on the temperature of denatura-tion of rat-tail tendon compared to non-reducedtendon provides evidence that the transition is notinitiated by a thermally induced breakage of across-link. The data also show that any change inlength/tension relationship of the cross-link onreplacing a carbon–carbon double bond by a singlebond, did not have an appreciable effect on thetemperature stability and therefore we infer noappreciable change in the intrafibrillar hydration inbuffered saline.

We have found previously that the reciprocal ofthe denaturation temperature (in Kelvin) is linearlydependent upon the volume fraction of intrafibrillarfluid.15,16 This has been confirmed and reinforcedby the current work. The present data showing thatthe previously determined relation, defined byleast-squares analysis of measurement of naturallycross-linked tail tendons, also holds when thetendons have been cross-linked synthetically usingexternal agents.

Thus the “polymer-in-a-box” mechanism ofstabilization,15 previously proposed to explain thenative tendon data, explains the new data also.Briefly in this mechanism, the rate of unfolding isdepressed by the proximity of the surroundingcollagen molecules in the fibre. These molecules actlike the walls of a box to reduce the number ofpossible molecular configurations and therefore theentropy of the unfolding molecule, thus reducingthe rate of unfolding. For example, we havepreviously suggested that collagen unfolding pro-ceeds via an “activated state” in which the inter-chain bonds in a thermally labile region have to bebroken before the whole molecule can unzip.17 Theentropy of this activated state is reduced by theconfinement of the molecules in the box.18 Conse-quently the Gibbs free energy of the activated stateis increased and therefore the probability ofactivation and the rate of collagen denaturationare reduced (as analysed).15 Cross-linking thereforestabilizes the collagen molecules in a fibre byreducing the separation of the molecules, i.e. byreducing the lateral dimensions of the box.

The temperature stability of the collagenmolecule in a fibre is composed of two terms.

(a)

The intrinsic temperature stability of themolecule itself, without the stabilizing inter-actions of surrounding collagen molecules. Thisis the intercept with the vertical line, water

volume fractionZ1. It is also the stability of themolecule in dilute solution.

(b)

The second term is the temperature stabilitygained by the collagen–collagen interactions,i.e. the stability gained from being confined inthe box. This is given by the slope ! (1Kthevolume fraction of water). The volume fractionof water determines the dimensions of the boxand the slope is determined by specific statis-tical characteristics of the collagen chains.15

The stability of the triple helix is governed bymany factors in addition to the cross-linkingstudied here, notably its primary structure19 par-ticularly its hydroxyproline content,20 pH,21 and thepresence of salts. Ions affect the denaturationtemperature of collagen in solution22 and in fibres23

and the effect depends on ion concentration and onthe identity of the ions involved.24 Thus, a particu-lar salt solution at a particular pH value may have astabilizing or a destabilizing effect on the moleculeitself, and it may swell or shrink the fibre. Theresults of these studies indicate that analysis ofthe temperature stability of fibres would be clarifiedby separating the effects of confinement of themolecule in the fibre lattice from the intrinsic effectson the molecule itself. Here we have thereforeanalysed the collagen in the presence of pure water.

Within the limits of experimental error, cross-linking affected neither the intrinsic temperaturestability nor the slope of the relation between thereciprocal of the denaturation temperature andwater content. Hence the cross-linking must havehad a negligible effect on the intrinsic stability of themolecule and a negligible effect on the relevantchain statistics.

The major effect was caused by the cross-linksreducing the axial separation of the molecules. Atan equilibrium separation, the net force on themolecule is zero, i.e. the sum of the attractive forcesmust equal the sum of those of repulsion.25 Otherthings being equal, we therefore deduce that the netresult of adding a cross-link is to increase theattractive force or reduce the repulsive force. This isequivalent to the cross-links sustaining a tension,which in principle might be calculated by measur-ing the pressure required to dehydrate the fibre bythe same amount as the cross-linking. In such amanner, we speculate that it should be possible toinvestigate the apparent extension/tension charac-teristics of specific cross-links. The actual forcesinvolved are potentially complex and such experi-ments might allow the effect of individual com-ponents to be investigated. We envisage forexample, that adding a cross-link could changethe charge and disrupt the hydration networkaround the molecule, as well as allowing a tensionto develop within the cross-link itself by distortionof the unstressed molecular conformation.Although natural cross-links, and the syntheticcross-links we have studied appear to dehydratethe collagen fibre we do not envisage that all cross-links will necessarily sustain a tension. We also

Thermal Stability of Cross-linked Collagen Fibres 555

foresee the possibility that a synthetic cross-linkcould theoretically be in compression. This wouldtend to rehydrate the fibre, and have novelmechanical consequences.

The primary function of native cross-linking is toimpart desired mechanical characteristics on thecollagen fibres in connective tissue. Covalent linksbetween the molecules will undoubtedly reduce theextensibility of the fibre by preventing longitudinalslippage of one molecule against another, andincrease stiffness by developing a lateral networkof linkages. However, the association betweencross-linking and hydration, which the presentpaper has brought into focus, and betweenhydration and mechanical properties (obvious toanyone who has watched a collagen fibre dry andchange from a wet flexible thread to a dry stiffbristle) implies that there is a relation betweencross-linking and mechanical properties over andabove that which is normally discussed. The resultsof this paper have therefore generated a question.How far are themechanical properties of connectivetissue caused by the collagen cross-linking per seand howmuch by the dehydration the cross-linkinginduces? In theory this question might be answeredby experiments analogous to those here, by com-paring the mechanical properties of syntheticallycross-linked and native fibres over a range ofhydrations.

To summarize, when the collagen in rat-tailtendon is cross-linked with glutaraldehyde, hexa-methylene diisocyanate, or malondialdehyde itstemperature stability is increased solely by dehy-dration of the fibre. The dehydration is broughtabout by the cross-links drawing the collagenmolecules closer together. The stabilization isconsistent with the polymer-in-a-box mechanism,previously suggested to explain the increase intemperature stability brought about by dehy-dration. The cross-linking studied here onlyreduced the quantity of water between themolecules and did not affect the water in intimatecontact with or bound to the molecule itself. Theenthalpy of denaturation was therefore unaffectedby the cross-linking treatments.

Materials and Methods

Rat-tail tendon

Tails were excised from the carcasses of five to six weekold rats and frozen at K20 8C until required. On removalfrom the freezer, tails were thawed and the tendonsremoved, cleaned of all visible contaminants and washedin distilled water.

Cross-linking procedure

The cross-linking was carried out with three differentcompounds in which the reactive groups were separatedto different extents: malondialdehyde (1!CH2), glutar-aldehyde (3!CH2) and hexamethylene diisocyanate (6!CH2). Borohydride reduction was used to stabilize the

pre-existing aldimine cross-link without altering itslength.

Cross-linking with glutaraldehyde

Tendons were placed in 0.5% aqueous solutions ofglutaraldehyde at 4 8C for 24 hours. After removal andwashing in copious amounts of distilled water, the cross-linked tendons were kept in water at 4 8C.

Cross-linking with hexamethylene diisocyanate(HMDC)

The method used by Rault et al.26 was adopted.Tendons were placed in 1% (w/v) HMDC (Sigma) inwater buffered to pH 5.5 and frozen at K19 8C for 22hours. Samples were then thawed and washed in copiousquantities of water and kept in water at 4 8C.

Cross-linking with malondialdehyde (MDA)

Tendons were placed in 150 mM MDA (a gift from DrD. A. Slatter, Biochemistry Department, CambridgeUniversity) at pH 7.5 and left for different lengths oftime at 20 8C. Following treatment, the tendon wasthoroughly washed and stored in water at 4 8C.27

Pre-existing cross-link reduction with borohydride

Equal masses of blot dried rat-tail tendon were placedin two separate containers with 5 ml of fresh phosphate-buffered saline (PBS). Sodium borohydride equivalent to1% of the blot-dried tendon mass was dissolved in 100 mlof 0.01 M NaOH and added to one container. 100 ml of0.01 M NaOH alone was added to the second, controlcontainer. Both open containers were left for one hour atroom temperature in the fume hood. The tendons werethen removed with forceps from each incubation mixture,separately rinsed twice with 5 ml of distilled water andplaced in 5 ml of 0.01 M HCl for approximately 30seconds to stop any residual reaction. Each sample wasthen rinsed twice more with excess distilled water andfinally rinsed with a single change of PBS before beingplaced in 5 ml of fresh PBS.

Sample preparation at different hydrations andcalorimetry of fibres

Control and cross-linked tendons in water were dabbeddry, weighed in open aluminum DSC pans and left todehydrate on the balance for different lengths of time inorder to obtain a range of hydrations. After the chosentime pans were hermetically sealed, weighed (G0.01 mg)and left to equilibrate at 4 8C. After at least one dayof equilibration in the sealed pans, samples werere-weighed to ensure that there had been no leakageand scanned in a Perkin–Elmer DSC-7, fitted withIntracooler, at 10 8C/minute from 5 8C to an appropriatespecified temperature using an empty pan as reference.Following scanning the pans were pierced and placed inan oven at 105 8C overnight and re-weighed for dry mass(md) and water mass (mw) determinations. The volumefraction of water, 3, was calculated using the relation:

3Zmw

rwmw

rwCmd

rd

where rw and rd, the densities of water and dry tendon,

556 Thermal Stability of Cross-linked Collagen Fibres

were taken as 1.00!103 kg mK3 and 1.38!103 kg mK3,respectively.

Calorimetry of solutions

Collagen solutions were stirred and degassed for eightminutes using a Thermovac apparatus (Microcal Inc.,Northampton, MA) and scanned in a VP-DSC (Microcal)from 10 8C to 60 8C at 1 8C/minute. Numerical analysis ofthe data was undertaken with Microcal software using acubic interpolation for the baseline.

References

1. Kagan, H. M. & Trackman, P. C. (1991). Properties andfunction of lysyl oxidase. Am. J. Respr. Cell Mol. Biol. 5,206–210.

2. Maki, J. M., Tikkanen, H. & Kivirikko, K. I. (2001).Cloning and characterization of the fifth human lysyloxidase isoenzyme: the third member of the lysyloxidase-related subfamily with four scavenger recep-tor cysteine-rich domains. Matrix Biol. 20, 493–496.

3. Miles, C. A., Sims, T. J., Camacho, N. P. & Bailey, A. J.(2002). The role of the alpha 2 chain in the stabilizationof the collagen type I heterotrimer: a study of the typeI homotrimer in oim mouse tissue. J. Mol. Biol. 321,797–805.

4. Bailey, A. J., Paul, R. G. & Knott, L. (1998).Mechanisms of maturation and ageing of collagen.Mech. Ageing Dev. 106, 1–56.

5. Horgan, D. J., King, N. L., Kurth, L. B. & Kuypers, R.(1990). Collagen crosslinks and their relationship tothe thermal properties of calf tendons. Arch. Biochem.Biophys. 281, 21–26.

6. Le Lous, M., Flandin, F., Herbage, D. & Allain, J. C.(1982). Influence of collagen denaturation on thechemorheological properties of skin assessed bydifferential scanning calorimetry and hydrothermaltension measurement. Biochim. Biophys. Acta, 717,295–300.

7. Flandin, F., Buffevant, C. & Herbage, D. (1984). Adifferential scanning calorimetric analysis of the age-related changes in the thermal stability of rat skincollagen. Biochim. Biophys. Acta, 791, 205–211.

8. Allain, J. C., Le Lous, M., Bazin, S., Bailey, A. J. &Delaunay, A. (1978). Isometric tension developedduring the heating of collagenous tissues—relation-ships with collagen cross-linking. Biochim. Biophys.Acta, 533, 147–155.

9. Melling, M., Pfeiler, W., Karimian-Teherani, D.,Schnallinger, M., Sobal, G., Zangerle, C. & Menzel,E. J. (2000). Differential scanning calorimetry, bio-chemical, and biomechanical analysis of human skinfrom individuals with diabetes mellitus. Anat. Rec.259, 327–333.

10. Andreassen, T. T., Seyer-Hansen, K. & Bailey, A. J.(1981). Thermal stability, mechanical properties andreducible cross-links of rat-tail tendon in experimen-tal diabetes. Biochim. Biophys. Acta, 677, 313–317.

11. Yue, D. K., McLennan, S., Delbridge, L., Handelsman,

D. J., Reeve, T. & Turtle, J. R. (1983). The thermalstability of collagen in diabetic rats: correlation withseverity of diabetes and nonenzymatic glycosylation.Diabetologia, 24, 282–285.

12. Bailey, A. J., Sims, T. J., Avery, N. C. & Miles, C. A.(1993). Chemistry of collagen cross-links—glucose-mediated covalent cross-linking of type IV collagen inlens capsules. Biochem. J. 296, 489–496.

13. Covington, A. D. (1997). Modern tanning chemistry.Chem. Soc. Rev. 26, 111–126.

14. Paul, R. G. & Bailey, A. J. (2003). Chemical stabili-zation of collagen as a biomimetic. Sci. World J. 3,138–155.

15. Miles, C. A. & Ghelashvili, M. (1999). Polymer-in-a-box mechanism for the thermal stabilization ofcollagen molecules in fibres. Biophys. J. 76, 3243–3252.

16. Miles, C. A. & Burjanadze, T. V. (2001). Thermalstability of collagen fibres in ethylene glycol. Biophys.J. 80, 1480–1486.

17. Miles, C. A., Burjanadze, T. V. & Bailey, A. J. (1995).The kinetics of the thermal denaturation of unrest-rained rat tail tendon determined by differentialscanning calorimetry. J. Mol. Biol. 245, 437–446.

18. Doi, M. & Edwards, S. F. (1986). The Theory of PolymerDynamics, Oxford University Press, Oxford, UK.

19. Persikov, A. V., Ramshaw, J. A. M., Kirkpatrick, A. &Brodsky, B. (2002). Peptide investigations of pairwiseinteractions in the collagen triple-helix. J. Mol. Biol.316, 385–394.

20. Burjanadze, T. V. & Kisiriya, E. L. (1982). Dependenceof thermal stability on the number of hydrogen-bondsin water-bridged collagen structure. Biopolymers, 21,1695–1701.

21. Dick, Y. P. & Nordwig, A. (1966). Effect of pH on thestability of the collagen fold. Arch. Biochem. Biophys.117, 466–468.

22. Komsa-Penkova, R., Koynova, R., Kostov, G. &Tenchov, B. G. (1996). Thermal stability of calf skincollagen type I in salt solutions. Biochim. Biophys. Acta,1297, 171–181.

23. Luescher, M., Ruegg, M. & Schindler, P. (1974). Effectof hydration upon the thermal stability of tropocolla-gen and its dependence on the presence of neutralsalts. Biopolymers, 13, 2489–2503.

24. Lim, J. J. (1976). Transition temperature and enthalpychange dependence on stabilizing and destabilizingions in the helix-coil transition in native tendoncollagen. Biopolymers, 15, 2371–2383.

25. Leikin, S., Rau, D. C. & Parsegian, V. A. (1994). Directmeasurement of forces between self-assembled pro-teins: temperature-dependent exponential forcesbetween collagen triple helices. Proc. Natl Acad. Sci.USA, 91, 276–280.

26. Rault, I., Frei, V., Herbage, D., AbdulMalak, N. &Huc,A. (1996). Evaluation of different chemical methodsfor cross-linking collagen gel, films and sponges.J. Mater. Sci.-Mater. M7, 215–221.

27. Slatter, D. A., Avery, N. C. & Bailey, A. J. (2004).Identification of a new cross-link and unique histidineadduct from bovine serum albumin incubated withmalondialdehyde. J. Biol. Chem. 279, 61–69.

Edited by M. Moody

(Received 8 September 2004; received in revised form 27 November 2004; accepted 1 December 2004)