10
All Gelatin Networks: 1. Biodiversity and Physical Chemistry ² Christine Joly-Duhamel, Dominique Hellio, and Madeleine Djabourov* Laboratoire de Physique et Me ´ canique des Milieux He ´ te ´ roge ` nes, Ecole de Physique et Chimie, UMR ESPCI-CNRS 7636, 10 Rue Vauquelin, 75231 Paris Cedex 5, France Received February 15, 2002. In Final Form: May 16, 2002 Gelatin gels are well-known for their ability to form nonpermanent, physical gels at room temperature and for their numerous applications in photographic and food industries. The difficulties in understanding the processes of gelation and the properties of the physical gels is partly due to the nature of the molecules that exhibit this state: these are often biopolymers. Concerning gelatin, a regain of interest appeared recently, in relation with the development of new types of moleculessmammalian gelatins were almost exclusively considered so far in the literature. In this series of two papers we investigated gelatin samples from various origins, mammalian and fish gelatins, with the aim of comparing their properties to the reference samples generally used in photographic and food applications. The samples were characterized by their imino acid composition, which indeed varies according to the biodiversity of the species from which they are extracted. The physical and chemical aspects are mainly reported in this paper, while the companion paper deals with the rheological properties. The influence of the thermal treatments, of the gelatin concentration, of the molecular weight, and of the solvent (aqueous and mixed solvents) was put in evidence and finally blending of samples was achieved. Optical rotation measurements were mainly performed; they allow us to fully characterize the development of the triple helices, which depends on the parameters mentioned before. A systematic comparison of the temperatures of helix formation and melting was undertaken. The results are discussed in the context of the models of helix-coil transitions often used for proteins and polypeptides and sometimes for polysaccharides. Formation of physical networks in protein and polysac- charide solutions has been intensively investigated over the last years 1-3 because of their many important ap- plications or simply because they are difficult to under- stand. Some of these biopolymers undergo a specific conformational transition from coil to helix during which the physical network is formed. The relation between the molecular characteristics and the rheology of the networks is one of the important goals of these studies, compared to the more conventional rubberlike, chemical networks. The formation of extended helical structures characterizes the networks. The coexistence of ordered structures, which are often described as rigid rods, and of flexible coils is the starting point of theoretical models for the elasticity of these networks. The adequacy of these models to describe the real networks is difficult to assess. The physical networks obtained by cooling the solutions are disordered and their properties vary considerably from batch to batch, which makes comparisons between measurements ob- tained by different techniques or different authors very delicate. The variability of chemical composition of the biopolymer, the molecular weight, the polydispersity, the modifications of the environment such as pH, ions, additives such as sugars, polyols, etc., the thermal history, or the passage of time itself have been shown to sub- stantially influence the mechanical and thermal properties of these physical gels. It is thus difficult to identify the relevant parameters, which should be kept in mind in order to model their elasticity on the basis of a stress- strain relationship, at a molecular level. The question that still remains is how to find the right parameters to enter in such models? On one hand, many parameters have been identified experimentally; on the other hand, the models make use of very few of them. Also, one does not necessarily know how to measure some particular pa- rameters that are introduced in the models. This is the paradoxical situation of physical gels! In this series of two papers we consider gelatin gels. When a gelatin solution is cooled below room temperature, the protein coils start to form triple helices and progres- sively a 3D network is formed. The triple helices are reminiscent of the native structure of collagen. The residues can only partially recover their native conforma- tion; a certain proportion still remains in the random coil conformation, even when the samples are annealed for hours or days. The gel is definitely not an equilibrium state. When temperature is raised back above approxi- mately 30 °C, the reverse transition, helix to coil, takes place and the gel becomes liquid. Many parameters, as those mentioned before, have a dramatic influence on the thermal and mechanical properties of the gels. The aim of this investigation is to take into account a large number of these parameters, which govern gel formation, and derive the common features that can serve as a basis for modeling the physical gels. We addressed these questions in the context of the emergence of new types of gelatins such as those extracted from fish skins of various species aimed to supplant the traditional ones. Fish gelatins are certainly different from mammalian gelatins. Fishes being, in general, nonhomoeothermic animals (they do not have a constant body temperature, like mammals), their body temperature depends on the temperature of the water where they live (rivers, seas, etc.). Therefore, fish collagens exhibit a large biodiversity in relation with their environ- ment that the mammalians do not have, especially in their ² This article is part of the special issue of Langmuir devoted to the emerging field of self-assembled fibrillar networks. * To whom correspondence should be addressed: e-mail [email protected]. (1) Faraday Discussion 101, Gels, 1995. (2) te Nijenhuis, K. Adv. Polym. Sci. 1997, 130. (3) Wiley Polymer Networks Group Review Series; Stepto, R. T. F., Ed.; John Wiley & Sons Ltd.: Chichester, U.K., 1998 and 1999; Vols. 1 and 2. 7208 Langmuir 2002, 18, 7208-7217 10.1021/la020189n CCC: $22.00 © 2002 American Chemical Society Published on Web 07/10/2002

All Gelatin Networks:  1. Biodiversity and Physical Chemistry †

Embed Size (px)

Citation preview

Page 1: All Gelatin Networks:  1. Biodiversity and Physical Chemistry               †

All Gelatin Networks: 1. Biodiversity and PhysicalChemistry†

Christine Joly-Duhamel, Dominique Hellio, and Madeleine Djabourov*

Laboratoire de Physique et Mecanique des Milieux Heterogenes, Ecole de Physique et Chimie,UMR ESPCI-CNRS 7636, 10 Rue Vauquelin, 75231 Paris Cedex 5, France

Received February 15, 2002. In Final Form: May 16, 2002

Gelatin gels are well-known for their ability to form nonpermanent, physical gels at room temperatureand for their numerous applications in photographic and food industries. The difficulties in understandingthe processes of gelation and the properties of the physical gels is partly due to the nature of the moleculesthat exhibit this state: these are often biopolymers. Concerning gelatin, a regain of interest appearedrecently, in relation with the development of new types of moleculessmammalian gelatins were almostexclusively considered so far in the literature. In this series of two papers we investigated gelatin samplesfrom various origins, mammalian and fish gelatins, with the aim of comparing their properties to thereference samples generally used in photographic and food applications. The samples were characterizedby their imino acid composition, which indeed varies according to the biodiversity of the species from whichthey are extracted. The physical and chemical aspects are mainly reported in this paper, while the companionpaper deals with the rheological properties. The influence of the thermal treatments, of the gelatinconcentration, of the molecular weight, and of the solvent (aqueous and mixed solvents) was put in evidenceand finally blending of samples was achieved. Optical rotation measurements were mainly performed;they allow us to fully characterize the development of the triple helices, which depends on the parametersmentioned before. A systematic comparison of the temperatures of helix formation and melting wasundertaken. The results are discussed in the context of the models of helix-coil transitions often used forproteins and polypeptides and sometimes for polysaccharides.

Formation of physical networks in protein and polysac-charide solutions has been intensively investigated overthe last years1-3 because of their many important ap-plications or simply because they are difficult to under-stand. Some of these biopolymers undergo a specificconformational transition from coil to helix during whichthe physical network is formed. The relation between themolecular characteristics and the rheology of the networksis one of the important goals of these studies, comparedto the more conventional rubberlike, chemical networks.The formation of extended helical structures characterizesthe networks. The coexistence of ordered structures, whichare often described as rigid rods, and of flexible coils is thestarting point of theoretical models for the elasticity ofthese networks. The adequacy of these models to describethe real networks is difficult to assess. The physicalnetworks obtained by cooling the solutions are disorderedand their properties vary considerably from batch to batch,which makes comparisons between measurements ob-tained by different techniques or different authors verydelicate. The variability of chemical composition of thebiopolymer, the molecular weight, the polydispersity, themodifications of the environment such as pH, ions,additives such as sugars, polyols, etc., the thermal history,or the passage of time itself have been shown to sub-stantially influence the mechanical and thermal propertiesof these physical gels. It is thus difficult to identify therelevant parameters, which should be kept in mind in

order to model their elasticity on the basis of a stress-strain relationship, at a molecular level. The question thatstill remains is how to find the right parameters to enterin such models? On one hand, many parameters havebeen identified experimentally; on the other hand, themodels make use of very few of them. Also, one does notnecessarily know how to measure some particular pa-rameters that are introduced in the models. This is theparadoxical situation of physical gels!

In this series of two papers we consider gelatin gels.When a gelatin solution is cooled below room temperature,the protein coils start to form triple helices and progres-sively a 3D network is formed. The triple helices arereminiscent of the native structure of collagen. Theresidues can only partially recover their native conforma-tion; a certain proportion still remains in the random coilconformation, even when the samples are annealed forhours or days. The gel is definitely not an equilibriumstate. When temperature is raised back above approxi-mately 30 °C, the reverse transition, helix to coil, takesplace and the gel becomes liquid. Many parameters, asthose mentioned before, have a dramatic influence on thethermal and mechanical properties of the gels. The aimof this investigation is to take into account a large numberof these parameters, which govern gel formation, andderive the common features that can serve as a basis formodeling the physical gels. We addressed these questionsin the context of the emergence of new types of gelatinssuch as those extracted from fish skins of various speciesaimed to supplant the traditional ones. Fish gelatins arecertainlydifferent frommammaliangelatins.Fishesbeing,in general, nonhomoeothermic animals (they do not havea constant body temperature, like mammals), their bodytemperature depends on the temperature of the waterwhere they live (rivers, seas, etc.). Therefore, fish collagensexhibit a large biodiversity in relation with their environ-ment that the mammalians do not have, especially in their

† This article is part of the special issue of Langmuir devoted tothe emerging field of self-assembled fibrillar networks.

* To whom correspondence should be addressed: [email protected].

(1) Faraday Discussion 101, Gels, 1995.(2) te Nijenhuis, K. Adv. Polym. Sci. 1997, 130.(3) Wiley Polymer Networks Group Review Series; Stepto, R. T. F.,

Ed.; John Wiley & Sons Ltd.: Chichester, U.K., 1998 and 1999; Vols.1 and 2.

7208 Langmuir 2002, 18, 7208-7217

10.1021/la020189n CCC: $22.00 © 2002 American Chemical SocietyPublished on Web 07/10/2002

Page 2: All Gelatin Networks:  1. Biodiversity and Physical Chemistry               †

imino acid composition. Accordingly, there is a wide rangeof gelation temperatures for these gelatins, different fromthe traditional ones.4-7

In the first paper of the series we compare the thermalproperties of gelatin gels of various sources under varioustreatments: either cooling, annealing, or heating condi-tions. All gelatin samples were fully characterized at amolecular scale. Structural properties, basically the helicalcontent, were determined for gelatins from differentsources (mammalian and fish), various molecular weights,and isoelectric points. The fish gelatins came from fishskins either from cold seas (cod) or warm seas (tuna,megrim). They were compared to lime-processed bovineossein, considered as a reference type for photography,and to pig skin gelatin, which is mostly used in foodapplications. We investigated these samples mainly inaqueous solutions, at their natural pHs. Their propertiesin the presence of additives (mixed solvents) were alsobriefly examined. We also investigated blends of gelatins.The structural data are discussed in the context of thewell-known models for helix-coil transitions. In the secondpaper we focus on the rheological properties (shear moduli)in the linear regime, for gels formed in exactly the sameconditions as for the structural investigation. The cor-relation between the two types of properties was system-atically sought and allowed to define a unique mastercurve for all gelatins, for all conditions of gel formation,without any adjustable parameter.

This paper contains three sections: materials andmethods, results of optical rotation measurements, anda theoretical discussion based on the Zimm-Bragg modelof helix formation.

I. Materials and MethodsThe reference to collagen being necessary for understanding

the difference between the various gelatin samples, we first recallsome properties of these collagens. In particular, the meltingtemperatures of the collagens are important for the comparisonbetween species.

Collagens. Several solutions of soluble collagens were ex-amined: calf skin (provided by SKW), sole from Coletica(Neptigene 2), and tuna and cod that we prepared ourselves withthe following protocol: the fish skins were washed twice indemineralized water during 30 min, then in ammonium acetate(2 L, 0.5 M), and then again in demineralized water. Theextraction of collagen was made in acetic acid in a cold storeduring 2 days. The skins first swelled and after 2 days in aceticacid were dislocated and collagen was dissolved. The solutionswere filtered and centrifuged during 1 h at 12000g in order toeliminate impurities from skins. For further purification, thecollagen was precipitated by adding salt to a final concentrationof 0.8 M NaCl. The suspension was again centrifuged. The solidphase (collagen) was collected, washed, and once more dissolvedin acetic acid in a cold store and centrifuged. Finally, the solutionwas cleaned by dialysis in order to eliminate the remaining saltin acetic acid. The concentration of collagen in solution wasdetermined by solvent elimination (amount of dry material). Theimino acid composition of mammalian and fish gelatins and thecorresponding melting temperatures of the native collagens arepresented in Table 1.

Gelatin Samples. A large variety of gelatin samples wasinvestigated. Two samples were from mammalians, provided bySKW (France): lime-processed demineralized bovine osseinextractions with two different molecular weights A1 and A2,acidic extractions from pig skins, also with two molecular weights,

B1 and B2. Various gelatin acidic extractions from fish skinswere also studied: cod gelatin provided by Icetec (Reykjavik,Iceland), megrim from Instituto del Frio (Madrid, Spain), andtuna from SKW (Boulogne, France). The molecular characteristicsand the isoelectric points of these gelatins are summarized inTable 2 and were provided by the laboratories mentioned above.

The A1, B1, tuna, and cod samples may be considered as high-quality extractions, containing high proportions of single, linearcoils (called R-chains, Mw ≈ 125 000 g/mol): 48% in A1, 25% inB1, 31.5% in tuna, and 35% in cod. The megrim sample contained35% â-chains (double the molecular weight of the R-chains) andonly 22% R-chains, meaning that the extraction did not fullyseparate the single chains; they were still partially cross-linked.For each sample, the molecular weight distribution also includessome smaller molecular weights (limited to 25 000 g/mol) andoccasionally larger ones (cross-linked coils). We also had at ourdisposal a very degraded gelatin (hydrolyzed sample) of type Afrom SKW, with a very low molecular weight, which is anongelling sample (Mw ) 11 200 g/mol and an index of polydis-persity I ) 1.6). Its molecular weight distribution is entirelylocated below the previous ones. This sample was interesting toinvestigate for illustrating the strong molecular weight effectsin helix formation. The isoelectric points pI of gelatins dependon the extraction process: they are around 5 for basic extractionsand around 8 or 9 for acidic extractions. Basic extractions increasethe number of COO- groups by hydrolysis of lateral groups ofAsp and Glu.

The solvent for gelatin solutions was mainly demineralizedwater. We also used mixed solvents of water and glycerol(Rectapur 98%, Prolabo) or sorbitol [D-(-)-sorbitol, M ) 182.17g/mol, Prolabo) from Merck Eurolab. The pH of the solutions wasclose to 5.7 for basic extractions and 5.1 for acidic extractions.

Protocol of Gelatin Dissolution. The required quantitiesof gelatins were mixed with water and allowed to swell overnightat 4 °C. The A and B samples (the mammalian gelatins) werethen dissolved at 45 °C during 30 min and the fish gelatins at35 °C, by use of a magnetic stirrer. The tuna and cod solutionswere filtered (0.45 µm disposable filters, Millipore). The crucialaspects in theprotocol ofdissolutionare thecontrol of temperatureand time during which the solution is kept at high temperature,because of the risk of hydrolysis, which modifies the molecularweight distributions. It was also noticed that the solutions whichgelled do not recover completely after heating their initial stageof dispersion, because some local associations of the chains(especially by hydrophobic groups or entanglements) still persist.Higher temperatures or longer periods of heating inducesubstantial degradation of the chains and must be avoided.Because of these restrictions, each solution was used only once.

Methods. Optical rotation was measured on a Perkin-Elmer341 polarimeter, equipped with a PC computer, with a softwarespecially developed to collect simultaneously the optical rotationangle and the temperature of the sample versus time. The glasscell has an optical path of 1 cm and is jacketed to allowtemperature to be controlled from an external circulating bath.The wavelength could be varied as the emission lines of Hg andNa lamps. Temperature was regulated by a Julabo FS18 baththat was programmed to execute various steps of cooling,annealing, or heating. Cooling ramps of 0.5 ˚C/min and heatingramps of 0.05 ˚C/min were mainly applied. The conformationalchange from coil to helix is accompanied by an important changein the optical rotation of the solution. In the case of collagentriple helices, it is possible to derive the amount of helices or the

(4) Leuenberger, B. H. Food Hydrocolloids 1991, 5, 353-361.(5) Gilsenan, P. M.; Ross-Murphy, S.-B. In Wiley Polymer Networks

Group Review Series; Stepto, R. T. F., Ed.; John Wiley & Sons Ltd.:Chichester, U.K., 1999; Vol. 2, Chapt. 28.

(6) Gilsenan, P. M.; Ross-Murphy, S. B. Food Hydrocolloids 2000,14, 191-195.

(7) Gilsenan, P. M.; Ross-Murphy, S. B. J. Rheol. 2000, 44, 871-883.

Table 1. Imino Acid Composition of Gelatins andMelting Temperatures of Corresponding Collagens

imino acid (g/100 g of protein)

Pro Hypa total Tm(collagen), °C

bovine gelatin 13.03 12.23 25.26 36b

pig skin gelatin 13.99 11.15 25.14 36b

tuna 11.61 10.49 22.1 29sole or megrim 12.86 9.06 21.92 28cod 10.29 6.72 17.01 15

a Hydroxyproline. b Calf skin collagen.

Gelatin Networks: Biodiversity and Physical Chemistry Langmuir, Vol. 18, No. 19, 2002 7209

Page 3: All Gelatin Networks:  1. Biodiversity and Physical Chemistry               †

fraction of amino acids in the helical conformation from thespecific optical rotation of the solution. The procedure is explainedin ref 8.

The helix amount ø

is derived from

where λ is the wavelength (most of the experiments wereperformed at λ ) 436 nm), [R]λ

exp ) R/cl is the specific opticalrotation of the protein in solution, c is the concentration (gramsper cubic centimeter), l is the optical path (decimeters), R is theoptical rotation angle (degrees) measured experimentally, [R]λ

col-

lagen is the specific optical rotation of native soluble collagen (ø) 1), which contains only triple helices, and [R]λ

coil is the specificoptical rotation of the coils (ø ) 0). An average value of 100 g/molper amino acid was taken.

At high temperature, the chains have a random coil confor-mation. The specific optical rotations for the different gelatinsin coil conformation were derived directly from the measurementin solutions at high temperature. The exact concentration ofgelatin was derived by weighing after drying the solutions (thegranules of gelatin contain usually around 10% humidity). Thespecific optical rotations of the different gelatins in the coil stateare given Table 3.

The specific optical rotation in the coil conformation isindependent of concentration below c ) 10% g/cm3. However, itvaries slightly with temperature and solvent (for mixed solvents).We determined experimentally a drift of [R] (with the units givenbefore) by ∆[R]/∆T ) 0.8/°C for sample A, 0.7/°C for tuna, or0.5/°C for cod. This small variation may have at least twoorigins: changes of the index of refraction of the solvent or changesof local conformation of the coils. Thus, we kept the values of[R]coil

436nm at the ultimate temperature before it starts to varystrongly at the beginning of the coil to helix transition.

We were also able to measure the specific optical rotation ofnative collagens, calf skin, and sole skin at 436 nm. We foundthe following values:

[R]helix436nm ) -800 deg cm3 g-1 dm-1 for calf skin collagen

and[R]helix

436nm ) -810 deg cm3 g-1 dm-1 for sole collagen.The value for calf skin is in agreement with the literature.8-9

We could not estimate precisely the values for all collagens,because all samples were not sufficiently pure; thus we decidedto use a unique value for the specific optical rotation for allcollagens, which is the value measured for the sole collagen andclose to that of calf skin collagen (100% of residues in helixconformation).

Viscosity measurements were performed in dilute and sem-idilute gelatin solutions at 40 °C. An Ubbelohde viscometer was

used for the determination of the intrinsic viscosity (c < 0.02g/cm3) and a stress-controlled rheometer, AR 1000 from TAInstruments, with a concentric cylinder with double gap cup forlarger concentrations (from 0.02 to 0.08 g/cm3).

II. Results: Optical Rotation MeasurementsMelting Curves for Collagens. The first important

step for understanding the gelation of the different gelatinsamples is the melting or the helix-coil transition ofsoluble collagens. The temperature of this transition iswell characterized when the extraction of collagen pre-served its native structure (helices). Optical rotation canbe used to derive the melting temperature Tm and toestablish the dependence of the helical content withtemperature. In the experiments that we performed onfish collagens, the temperature was raised at a fixed rateof 0.05 °C/min. The optical rotation angle, R recordedduring the temperature ramp, is a sigmoid going from Rof the 100% helix to R of the 100% coil (ø varies between1 and 0). The derivative of this curve, dR/dT or dø/dT,exhibits a peak giving the average melting temperature.The width of the curve indicates the spread of temper-atures. The results for our collagens are shown in Figure1. The calf skin preparation was certainly a good one:there is a large peak and the width is only a few degrees.It exhibits the highest melting temperature. All the fishcollagens melted at lower temperatures compared to calfskin, the cod having the largest deviation. The cod collagenextract was not as good as the other preparations, as itappears from the shape of the melting curve, where thetotal amount of soluble collagen was lower than expectedand the width of the curve larger than in the othercollagens.

(8) Djabourov, M.; Leblond, J.; Papon, P. J. Phys. France 1988, 49,319-332; 49, 333-343.

(9) von Hippel, P. H.; Wong, K.-Y. Biochemistry 1963, 2, 1399-1413.

Table 2. Molecular Characteristics of the Gelatin Samples

gelatin

A1 A2 B1 B2 megrim tuna cod

Mw (g/mol) 145 700 102 200 168 500 74 600 226 300 122 600 120 500Mn 86 300 52 400 88 370 39 540 136 325 74 780 62 435I ) Mw/Mn 1.68 1.95 1.91 1.89 1.66 1.64 1.93pI 5 5 8.7 7.6 ∼8.9 ∼8.9 ∼8.9

Table 3. Specific Optical Rotation of Gelatin Solutions in the Coil Conformation

gelatin

A1 A2 B1 B2 megrim tuna cod

temperature (°C) 35 35 35 35 30 30 15[R]436nm

coil -290 ( 2 -278 ( 1 -285 ( 1 -264 ( 1 -263 ( 2 -261 ( 1 -245 ( 1

ø ) number of residues in helical conformationtotal number of residues

ø )[R]λ

exp - [R]λcoil

[R]λcollagen - [R]λ

coil(1)

Figure 1. Melting curves of soluble collagens of various originsin acidic solutions, derived from optical rotation measurements(dR/dT) at a heating rate of 0.05 °C/min.

7210 Langmuir, Vol. 18, No. 19, 2002 Joly-Duhamel et al.

Page 4: All Gelatin Networks:  1. Biodiversity and Physical Chemistry               †

The correlation between the imino acid content of thecollagens, given Table 1 and their melting temperatures,is clear: the larger this content, the higher the meltingtemperature in acidic solutions. Gustavson10 first noticedthe correlation between the denaturation temperaturesof collagens and the environmental temperatures of thefishes. Later on, the relationship between the denaturationtemperatures and the imino acid content (Pro + Hyp,hydroxyproline) has been established for vertebrate andinvertebrate species11 in relation with the mechanism ofstabilization of the triple helices by the pyrrolidineresidues.Morerecentstudies,12 however, stress thespecificrole of Hyp for collagens of vertebrates, because, in thiscase, the location of Hyp in the sequence of the protein isdifferent from that in invertebrates.

Renaturation and Melting of Triple Helices. Thefollowing gelatin solutions have been investigated:

(a) aqueous solutions, single-component gelatins,(b) mixed solvents, single-component gelatins, and(c) aqueous solutions, blends of gelatins.We report first on the single component gelatins.(a) Aqueous Solutions, Single-Component Gela-

tins.Helix formation was investigated for gelatin solutionsat a fixed concentration of 4.5% g/cm3; temperature wasdecreased linearly in time. Gelatins of various sourcesare compared in Figure 2, where helix amount is plottedversus temperature. The coil to helix transition starts atdifferent temperatures; one can notice that it varies inthe same order as the melting temperatures of collagen.However, this type of curve depends on the rate of coolingand concentration. Thus, from these experiments onecannot define precisely a temperature at which helicesstart to form but a range of temperatures obviously belowthe melting temperatures of the native collagens inaqueous solutions. The rate of helix formation follows thesame trend for all gelatins: the lower the temperature,the larger the helix amount formed during the coolingramp. A long annealing time provides additional amountsof helices, which never reach 100% within the “normaltime scale” of observation. We could notice an increase,up to ø ) 0.55 for cod gelatin or 0.65 for A1 (see, for instance,Figure 4)

The renaturation appears to be very sensitive to themolecular weight. This is illustrated in Figure 3 for A1,A2, B1, and B2 for the same concentration (4.5% g/cm3).The thermal history is also displayed, the final temper-ature being 20 °C. The kinetics of helix formation is shown.A1 and B1 reach a helix amount of about 0.5, while A2and B2 reach only 0.15 or 0.25 after 20 000 s (5.5 h for thecooling ramp and annealing). The hydrolyzed sample(nongelling gelatin) is also shown on the same graph. Thehelical renaturation is much more restricted in the lattercase (ø ≈ 0.03).

We summarize in Figure 4 the results for the helixamounts, after annealing for 15 h at various finaltemperatures indicated in the plot. All gelling samplesare represented with the exception of the very lowmolecular weight sample. The progressive shift of thecharacteristic temperatures of helix formation, in relationwith the molecular composition, is clearly put in evidence.The helical content is lower for the low molecular weightgelatins, for the same temperature and annealing timethan for the high molecular weight.

Finally we examine the effect of gelatin concentrationin Figure 5: helix amounts versus time are shown at fourdifferent concentrations (from 2% to 8% g/cm3), with thesame thermal history. The curves corresponding to thehighest concentrations were fairly noisy. This problemappeared several times at low temperatures for reasons

(10) Gustavson, K. H. In The chemistry and reactivity of collagen;Academic Press: New York, 1956.

(11) Harrington, W. F.; Rao, N. V. In Conformation of biopolymers;Ramachandran, G. N., Ed.; Academic Press: New York, 1967; Vol. 2.,pp 513-531.

(12) Privalov, P. L. Adv. Protein Chem. 1982, 35, 1-104.

Figure 2. Helix formation for gelatins of various origins versustemperature at a cooling rate of 0.5 °C/min. (concentration 4.5%g/cm3).

Figure 3. Helix formation versus time for A1, A2 and B1, B2samples and for the hydrolyzed gelatin. The thermal history(cooling and annealing) is shown on the same graph.

Figure 4. Helix amounts at different temperatures afterannealing for 15 h for the gelatins of various sources andmolecular weights. The concentration is 4.5% g/cm3. The linesare guides for the eye.

Gelatin Networks: Biodiversity and Physical Chemistry Langmuir, Vol. 18, No. 19, 2002 7211

Page 5: All Gelatin Networks:  1. Biodiversity and Physical Chemistry               †

that can either be the scattering of light, by someinhomogeneities of the gel, or a stress birefringence thataccompanies the slight contraction of the gel, when helicesare formed. This is the limitation of this technique, whenused for higher concentrations and low temperatures. InFigure 5 one sees that the rate of helix formation duringthe cooling ramp was sensitive to the concentration, inthe first stages: the larger the concentration, the largerthe helix amounts. However, when the three samples werekept at constant temperature (10 °C) and allowed tomature, the amount of helices was very close, the moreconcentrated solution exhibiting even a slightly smalleramount.

We proceed now with the melting or “unwinding” of thehelices. The melting can be followed by optical rotationwhen temperature is slowly increased, after the solutionswere kept for annealing. As for collagens, the derivativedø/dT exhibits a peak that indicates the melting temper-ature range, and from the width of the peak one can deducethe sharpness of the transition. It is observed that thethermal stability of the helices is strongly related to therenaturation conditions: the position of the peaks and thewidth of the melting curves are related to the annealingtemperature. This is clearly illustrated in Figure 6 for A1and A2 (top panel) and for fish gelatins, cod and tuna(middle panel). The concentrations for these solutions was4.5% g/cm3, and the annealing temperatures were dif-ferent, as indicated by arrows. This is the temperature atwhich the heating ramp started. The hysteresis betweenthe formation and maturation temperature (initial tem-perature) is thus observed and an average meltingtemperature appears for each sample. After maturation,the lowest temperatures provide the largest amounts ofhelices and the largest distributions of melting temper-atures. Under identical conditions, the low molecularweight samples form less stable helices. Finally, themelting curves for three concentrations of A1 are shownin Figure 6 (bottom panel): there is no systematic shift ofthe peak position, but the shape of the melting curve andthus the distribution of the melting temperatures appearsslightly different. The analysis of the melting curves ispresented in the Discussion section.

(b) Mixed Solvents, Single-Component Gelatins.The solvent composition also influences the formation andmelting of helices. We used, as an example, mixtures ofwater and glycerol. Glycerol is miscible with water in allproportions and is used for making capsules for phar-maceutical applications. The densities of the mixedsolvents were measured in order to determine theconcentrations of gelatin solutions in grams per cubic

centimeter. Densities, measured at 22 °C with increasingweight fractions of glycerol, vary between 1 g/cm3 for purewater and 1.25 g/cm3 for pure glycerol. To derive the helicalcontent in these mixed solvents, we checked the validityof the Drude equation, which is known for gelatins inaqueous solutions.8 This equation was valid for the mixedsolvents containing 30 and 50 wt % glycerol + water. Also,we noticed that the specific optical rotation of the coilconformation changes with the composition of the mixed

Figure 5. Helix renaturation for A1 at various concentrationsand same cooling ramp.

Figure 6. (a, top panel) Melting curves dø/dT for A1 and A2after cooling at different temperatures indicated by the arrowsand annealing for 15 h. The melting temperatures varyaccording to the initial temperature and depend on themolecular weight. (b, middle panel) Melting curves dø/dT forfish gelatins (cod and tuna) cooled and annealed at differenttemperatures indicated by the arrows. (c, bottom panel) Effectof concentration on the melting curves dø/dT for A1.

7212 Langmuir, Vol. 18, No. 19, 2002 Joly-Duhamel et al.

Page 6: All Gelatin Networks:  1. Biodiversity and Physical Chemistry               †

solvent. We found for A1 in water [R]coil436nm ) -290 ( 2

deg cm3 g-1 dm-1, whereas with 30% glycerol [R]436nmcoil )

- 275 ( 1 deg cm3 g-1 dm-1 or 50% glycerol [R]436nmcoil )

-252 ( 3 deg cm3 g-1 dm-1 measured at T ) 35 °C for thesame wavelength. With these modifications, the helixamounts were derived, as in pure water. The helix amountsversus temperature, during cooling, are shown in Figure7 for A1 at a concentration of 4.5% g/cm3. There is clearlya shift of the beginning of helix formation, toward highertemperatures, which is progressively enhanced by theaddition of glycerol (up to 50 wt %, for larger amounts ofglycerol; gelatin A1 was no more soluble). The sametendency was observed with sorbitol and glucose syrup at30 wt % (data not shown). The melting of helices in thesemixed solvents also exhibits a marked shift toward highertemperatures. According to the literature,13 the equivalentshift is observed in melting of soluble collagen with themixed solvents (polyols).

(c) Aqueous Solutions, Blends of Gelatins. Weconsider now the blending in aqueous solutions of twosamples of gelatins, from different sources, with the aimof determining if two gelatins are compatible and can formhelices containing chains of both species. To this end wedesigned the following experiments: we mixed A1 andcod gelatins, which lie at the two extremities of the meltingtemperatures of collagens. We considered single solutionsof each species at 4% concentration and a blend at thetotal concentration of 8% g/cm3 containing equal propor-tions of the two gelatins. Solutions were cooled to 0.8 °C,annealed for 5 h, and finally heated again. In Figure 8(top and middle panels), we see both steps, formation andmelting of the helices. The formation step (top panel)clearly shows that the building of the helices was achievedin two waves: A1 starts first, during the cooling ramp,when the helix amount is superposed on the pure A1 data,followed by the cod sample. However, compared to theaqueous solutions of the cod alone, the gelation of codstarted earliersat a higher temperaturesaround 15 °Cinstead of 4 °C. Eventually, after 5 h the total concentrationof helices (cø) in grams of helices per cubic centimeter wasthe same as calculated by the addition of the twopopulations (Figure 8, top panel). Melting is also achieved

in two steps, each species separately: the derivative ofdR/dT is seen in Figure 8 (middle panel) and also the curvecalculated as the sum of the melting of A1 and cod insingle solutions, which shows that the melting peak of A1in the blend is almost identical to the single component,a small fraction of mixed helices melt in the intermediaterange of temperatures, between the melting peaks of codand A1 as single components. The melting peak of the codcomponent is broadened and apparently shifted towardhigher temperatures. This peak may represent helicesthat are mixtures of cod and A1 and/or pure cod helices.It is therefore not possible to define “one average melting(13) Gekko, K.; Koga, S. J. Biochem. (Tokyo) 1983, 94, 199-205.

Figure 7. Helix formation in mixed solvents water + glycerolfor A1 during cooling. The curves obtained with the mixedsolvents are noisy and were not smoothed. There is no systematiccorrelation between noisy measurements and the use of mixedsolvents as the problem appears even in pure water and it isnot systematic.

Figure 8. (a, top panel) Helix formation in a blend of cod andA1 gelatins at a total concentration of 8% g/cm3 and equalproportions of each species. (b, middle panel) Helix melting inthe same blend after annealing of 5 h. (c, bottom panel) Helixmelting of a blend of tuna and A1 gelatins with a totalconcentration of 9% g/cm3 and equal proportions of each species.

Gelatin Networks: Biodiversity and Physical Chemistry Langmuir, Vol. 18, No. 19, 2002 7213

Page 7: All Gelatin Networks:  1. Biodiversity and Physical Chemistry               †

temperature”, the blend clearly containing the twopopulations and a small fraction of mixed helices. Thetotal spread of the melting temperatures is equivalent tothe addition of the two individual melting curves, withthe differences given before. Blends of tuna and A1 werealso investigated: they represent the closest meltingtemperatures of the collagens. The melting curve of abinary mixture of c ) 9% g/cm3 of equal proportions of thetwo constituents is shown in Figure 8 (bottom panel). Thereis a broad melting curve, but compared to the curvecalculated as the sum of the two individual components,one can define in this case an average melting temperatureat half-height of the ø versus T curves. For variousproportions of the constituents the melting temperaturesvary as shown in Figure 9. The total width of the meltingpeak varies between 6 and 3 °C and is much smaller thanin Figure 8 (middle), for instance. A similar investigationbased on rheological measurements was recently pub-lished by Gilsenan and Ross-Murphy.6 Their meltingtemperatures are different from ours because they arenot derived in the same way. We also briefly investigatedthe effect of pH: as these gelatins (tuna and A1) havedifferent isoelectric points, we suspected coacervationeffects to play a role in this synergy. By changing the pHfrom the natural one (5.7 in demineralized water) to 3 orto 9, we noticed an effect on the kinetics of helix formationbut not on the overall behavior under melting conditions.Thus, we conclude that the formation of mixed triplehelices is enhanced between the species that have closeimino acid compositions, independently of the pH.

In conclusion of the experiments on helix formation andmelting, the following features were established:

(i) The temperature range for helix renaturationdepends on the imino acid composition of the gelatin andis directly related to the collagen melting temperatures.

(ii) The amount of helices for a given imino acidcomposition depends on temperature, but no equilibriumexists for the renaturation of helices. The helix amountversus temperature shows hysteresis between formationand melting and a distinct and marked dependence onmolecular weight. The dependence of the helix amount onconcentration (2 < c < 8 g/cm3) is not important afterannealing, when the kinetic effects are damped.

(iii) Mixed solvents (water + glycerol, or sorbitol, orsucrose,14 etc.) enhance the formation of helices by shifting

the process toward higher temperatures. An equivalentshift appears on the native collagen when dissolved insuch solvents.

(iv) Blends of gelatins with extreme amino acid com-positions show little synergy. Melting of the blends clearlyreflects the presence of the two initial sets of populations.Formation of mixed triple helices is, however, favored inblends of gelatins with close compositions.

III. DiscussionThe discussion is focused on the thermal properties of

the helices. The well-known theory for coil-helix transi-tions proposed by Zimm and Bragg15 is based on astatistical mechanical model in which the conformationsof the chains depend on two parameters, s and σ. Theparameter s is an equilibrium constant for the additionof one residue to a sequence of residues in helicalconformation, this residue being initially in the coilconformation adjacent to the helical portion. It character-izes the propagation step. The parameter σ is a factorthat characterizes the difficulties of starting a newsequence (σ , 1) the nucleation step. σs is the equilibriumconstant for the initiation of a new sequence by makingthe first hydrogen bond. The nucleation step is much lessfavorable than the propagation step. For very long chains,the Zimm-Bragg model assumes that helical and randomcoil sequences coexist in the same molecule in thetemperature interval of the melting transition. s has acritical value at 1, in the neighborhood of which thetransition from random coil to helical conformation occurs.The sharpness of the transition depends on both s and thechain length. However, above a certain value of the chainlength the transition becomes independent of chain length.It is also assumed in this model that s depends ontemperature and σ does not. The model describes anequilibrium transition and thus fully reversible.

The relation of s to the temperature is in the form of avan’t Hoff relation:

where T is the absolute temperature, R is the gas constant,and ∆H is the enthalpy change for adding one helical unitto the preexisting helical section. Thus s is related to theenergy of hydrogen bonding through the Boltzmann factorand σ to the entropy of formation of the first turn of thehelix, which is relatively independent of the temperature.Integration of eq 2 between the limits Tm and T allows usto relate the theory and the data:

Tm is the temperature of the midpoint of the meltingtransition for the very long chains. One can derive thefraction of residues in the helical form as a function of s,σ, and N, the number of residues (monomers) of the chain.From the Zimm-Bragg theory, Flory16 showed that thebreadth ∆T for the transition from helix to random coilfor very long chains is given by

(14) Nishinari,K.;Watase,M.;Kohyama,K.;Nishinari,N.; Oakenfull,D.; Koide, S.; Ogino, K. Polym. J. 1992, 24, 871-877.

(15) Zimm, B. H.; Bragg, J. K. J. Chem. Phys. 1959, 31, 526-535.(16) Flory, P. J. Polym. Sci. 1961, 49, 105-128.

Figure 9. Melting temperatures for blends of tuna and A1gelatins (total concentration of 9% g/cm3) and various propor-tions of tuna.

d(ln s)dT

) ∆HRT

(2)

ln s ) ∆HR (T - Tm

TTm) (3)

∆T )2RTm

2σ1/2

∆H(4)

7214 Langmuir, Vol. 18, No. 19, 2002 Joly-Duhamel et al.

Page 8: All Gelatin Networks:  1. Biodiversity and Physical Chemistry               †

The melting temperature Tm is evaluated at half-heightof the transition and the breadth ∆T between 3/4 and 1/4of its amplitude. At the midpoint of the transition, theaverage number of residues in one sequence of triple helixis given by σ -1/2. In other words, σ -1/2 is a measure of thesize of the cooperative unit.

Melting curves of collagens have been analyzed in theframework of these statistical mechanical theories.11 Fromthe breadth of the transition, the parameter σ can bederived. For different species of collagens, the enthalpychange ∆H of the transformation of one helical residue torandom coil was taken from experimental data ∆H ≈ 1.2kcal/mol ) 5 kJ/mol in acidic solutions. In the case of calfskin collagen with our data Tm ) 273 + 36 ) 309 K, ∆T) 2 °C, and thus σ ) 0.4 × 10-4 or σ -1/2 ≈ 158 residuesin the cooperative melting unit. The number of residuesin the cooperative unit per chain is one-third of the total.The significance of the size of the cooperative unit incollagens was discussed by Privalov12 in the context ofbiological aspects. The author raises the question whethernative collagen consists of a discrete or a continuousstructure. The distribution of the imino acids in thesequence of various collagens suggests the existence of arepeat unit, as a reminiscence of a primordial gene.

The pyrrolidine content of collagens modifies theirmelting temperatures. The imino acid content shouldmainly affect the entropy of the melting transition: thepresence of pyrrolidine rings in the protein chain decreasesthe configurational entropy of the random coil, which wasconfirmed by the calculations by Josse and Harrington.17

On the basis of the assumption that the pyrrolidine ringshave a conformational change that is zero at the coil-helix transition, the melting temperature is then relatedto the pyrrolidine content by

where p is the percentage of imino acids. Thus increasingthe percentage of imino acids increases the meltingtemperature.18 In this model, the enthalpic term isconstant, while in a more refined analysis,11-12 it was alsoassumed that the enthalpic term per residue may varywith the pyrrolidine content, decreasing with this content.One finds qualitatively a good agreement with the firstassumption, but a more quantitative estimation of Tm isobtained with the more refined analysis. It is also known19

that the melting temperatures and the enthalpies ofdenaturation of collagen vary with solvent compositionand pH and therefore there are no absolute values forthese parameters. In agreement with Privalov,12 a moreelaborate theory that takes into account the interactionof the macromolecules with the solvent should be neces-sary.

The influence of the imino acid content on the coopera-tive unit is also difficult to corroborate: Harrington andRao11 suppose that the cooperative unit diminishes whenthe total pyrrolidine content increases. The free energyof stabilization of the helical sequence will be increasedby insertion of pyrrolidine residues and thus a smallernumber of hydrogen-bonded residues can stabilize thesequence. Their experimental results do not show a clearcorrelation. Our results on tuna and sole do not allow ustoeither confirmordispute this statement.Thecooperativeunit σ -1/2 varies as Tm

2/∆T. For a lower imino acid content

Tm is decreased, the difference being 21 °C between theextremes, cod and calf skin (relative shift of 7%). Thechange of the breadth of the melting ∆T from one collagento another is difficult to evaluate with great accuracy dueto the significant effects of both the extraction conditionsand the rate of heating known to modify the shape of themelting curves. Only fluctuations of the cooperative unitsare observed, which cannot be connected unambiguouslyto the imino acid content.

While the Zimm-Bragg treatment could by applied fordenaturation of native collagens (with the assumption ofa continuous helical structure), the model used in thecontext of the gelatin gels gives inconsistent results. Twoaspects were considered in this investigation: the influ-ence of the gel formation temperature and of the molecularweight. We examine thus the helix-coil transition of therefolded triple helices.

Influence of the Gelation Temperature. The melt-ing temperatures of refolded triple helices were derivedexperimentally, as for collagen, as the temperaturecorresponding to the middle of the curve of R(T) or ø(T).Figure 10 summarizes these measurements: the meltingtemperatures of the gels are plotted versus annealingtemperatures (gelation temperatures) in Figure 10 (toppanel) for A1 and A2 and in Figure 10 (bottom panel) forthe three fish gelatins, cod, megrim, and tuna. The meltingtemperatures were derived at a fixed concentration of 4.5%

(17) Josse, J.; Harrington, W. F. J. Mol. Biol. 1964, 9, 269-287.(18) Harrington, W. F. J. Mol. Biol. 1964, 9, 613-617.(19) Harrington, W. F.; Rao, N. V. Biochemistry 1970, 9, 3714-

3724.

Tm ) ∆H(1 - p)∆S

(5)Figure 10. (a, top panel) Melting temperatures versusannealing temperatures for A1 and A2. The extrapolationtoward the melting temperature of mammalian collagens isshown. (b, bottom panel) Melting temperatures versus anneal-ing temperatures for megrim, tuna, and cod.

Gelatin Networks: Biodiversity and Physical Chemistry Langmuir, Vol. 18, No. 19, 2002 7215

Page 9: All Gelatin Networks:  1. Biodiversity and Physical Chemistry               †

g/cm3. At different concentrations of the A1 sample (Figure6, bottom panel), we observe a variation of meltingtemperature of the gels of 0.4 °C (between 26 and 26.4 °C)when concentration varied between 2 and 8% g/cm3 forgels cooled at 10 °C. Thus, we assume that the meltingtemperature is mainly related to the gelation temperature.The extrapolated melting temperatures, at the highestgelation temperatures, tend toward the native collagen ofeach species, i.e., the perfect sequence of triple helices.The comparison between the native collagen and therefolded triple helices, however, exhibits important limi-tations: renaturation is essentially a nonequilibrium,nonreversible process. The derivatives dR/dt or dø/dt arenonsymmetrical peaks as shown in Figure 6, in particularfor gels containing a large amount of helices (low gelationtemperature). Using this model, Harrington and Rao11

have calculated the sizes of the cooperative units σ -1/2

from the melting of refolded helices, by normalizing to 1the helix amounts observed at each cooling temperatureand measuring the melting temperatures and the breadthfor each thermal history (∆H is taken as a constant). Thesecalculations, when applied to our own gels, show unex-pected results in which the apparent cooperative lengthsσ -1/2 strongly depend on the gelation temperature (forinstance, σ -1/2 decreases with gelation temperaturebetween 50 and 37 residues for A1). Both the meltingtemperatures and the cooperative units are much smallerfor refolded helices than for native collagen and thus arenot only a function of the imino acid and solvent composi-tion, as stated in equilibrium transitions theories (Tm andσ should be independent of temperature). The apparentstrong decrease in the cooperative unit size is due bothto variations of the midpoint of the transition and of thebreadth of the melting curve. A small proportion of verystable triple helices that have the stability of the collagenitself appears occasionally [see Figure 6 (top)].

An interesting feature appears also in Figure 10 relatedto the extrapolation of the melting temperatures versusgelation temperature. An interpretation of this effect isbased on the nucleation theories in supercooled states.These well-known models of crystallization of liquids orpolymers in solution rely on the presence of an interfacialtension between the crystal and the liquid. The nucleationtheories state that the minimum number of moleculesrequired for a stable nucleus to form in the supercooledstate strongly depends on the supercooling, defined as∆Tsuper ) |T - Tm|. The critical nucleus size is inverselyproportional to ∆Tsuper. If this concept is applied to theincipient formation of the triple helices, one may expectthat the lower the temperature of gelation, the shorterthe nucleated sequences. At T ) Tm the nucleus is infinitelylarge and its melting temperature is that of the infinitecrystal. The particular feature in gelatin gels is that themelting of the helices, even after long periods of annealing,at fixed temperatures, still keeps the memory of the coolingtemperature: one can suppose that helices, while growing,“freeze in” the defects such as loops or mismatching ofamino acid sequences (memory effect). A parallel betweenthe critical size of the nucleus and the cooperative unitsize is made in ref 19.

Influence of the Molecular Weight. The effect of themolecular weight on helix renaturation and melting iseven more pronounced than the effects mentioned aboveand has important practical consequences. The meltingcurves ø(T) for the various molecular weights for the Atype samples are shown in Figure 11. The temperaturerange of this plot is chosen between 10 and 40 °C. Thenumber of residues per chain N are calculated from Mwvalues. In the Zimm-Bragg model the influence of the

length of the coils has been theoretically determined: twolimits are generally considered, the infinitely long chainsand the short chains. The difference in helix-coil transi-tions of long and short chains is that the latter tend tofollow a “one sequence approximation” (called the “zippermodel” for nucleic acids), whereas long chains have atendency to alternate random and helical sequences, asstated before. The cutoff between the two regimes is thelimit of N of the order of the cooperative unit length, σ -1/2;actually, the limit should be Νlimit ) 2σ -1/2. Applying thiscriterion to collagen, one finds Nlimit ≈ 300 (≈100 unitsper chain). This means that the hydrolyzed sample, whichis the “nongelling sample”, N ) 120, should be consideredas the limit of short chains. It is expected from these modelsthat the helix-coil transition becomes substantiallyindependent of the chain length above ∼Nlimit, as foundexperimentally by Zimm et al.20 for polypeptides withvarious molecular weights. The transition is particularlybroad for the short chains and much steeper for the longestones. The overall tendency observed for gelatin andcollagen melting, Figure 11, is in agreement with thisframework. However, one notices a large influence of themolecular weight on the melting curves even for gelatinsof high molecular weights, which is not justified in themodel. We also observed that the shorter chains exhibita more reversible transition during cooling and slowmelting (weak hysteresis). The shorter chains may indeedapproach the “one sequence model” suggested by theZimm-Bragg model. This is also in agreement with thenongelling property of these gelatins. Thus one mayconclude that nonreversibility should be associated withbranching or network formation.

In conclusion, we stress the limitations of the equilib-rium theories to describe the helix-coil transitions ingelatins and collagens. The fact that the renaturation ofthe helices creates junctions between the individual chainsleads to a completely different situation from the classicalmodels of the reversible transition. When the formationof a network is taken into account, one can explain, atleast qualitatively, some of these differences that createa nonequilibrium situation for the helix-coil transition,8with important hysteresis effects. It is interesting tocompare these results to the model calculations proposedby Viebke et al.21 on the double helix formation of

(20) Zimm, B. H.; Doty, P.; Iso, K. Proc. Natl. Acad. Sci. U.S.A. 1959,45, 1601-1607.

(21) Viebke, C.; Picullel, L.; Nilsson, S. Macromolecules 1994, 27,4160-4166.

Figure 11. Effect of the molecular weight on the melting ofhelices in the gels: Comparison with collagen.

7216 Langmuir, Vol. 18, No. 19, 2002 Joly-Duhamel et al.

Page 10: All Gelatin Networks:  1. Biodiversity and Physical Chemistry               †

κ-carrageenans. Experiments were performed with samplescovering a large range of molecular weights. The authorsarrived at the conclusion that branching on the helicallevel is not required for the formation of the gels (3Dnetworks) of κ-carrageenans, which then are essentiallybuilt by the side-by-side aggregation of helices, at thesuperhelical level. The equilibrium models for coil to helixtransitions that the authors propose, based on the Zimm-Bragg model, seem to satisfactorily reproduce theirexperimental data and allow them to derive the cooper-ativity parameter of the transition for this polysaccharide.Our conclusions and theirs thus complement each other.The comparison is, however, limited at this stage becausegelation of carrageenans is related to a different mech-anism than gelatin: side-by-side aggregation of the triplehelices is not observed in gelatins. The triple helicalsequences are stable in aqueous solutions. The fibers oftriple helices, similar to native collagen ones, do not re-form in gels.

Mixed solvents have a marked shift on the meltingtemperature of the helices for single-component gelatinsolutions. Our experiments concerned tuna, cod, andbovine gelatins. The increased thermal stability of collagenwas known already,13 by thermodynamic measurements(calf skin collagen). The interpretation of this effect ismainly based on the influence of sugars and polyols onthe structure of water. These compounds have a waterstructure-making character.13,22,23 This effect is also knownfor globular proteins. In the presence of these compoundsa preferential hydration of the protein takes place suchthat the hydrophobic groups exclude the polyhydriccompounds from their immediate vicinity. The hydro-phobic interactions are strengthened and the protein self-association is enhanced. To get a better insight into theseinteractions, we performed intrinsic viscosity measure-ments of the gelatins in the presence of the mixed solventssuch as water + glycerol. We found the following effects:for tuna gelatin the intrinsic viscosity is [η]) 53.8 cm3/gin pure water and 50.9 cm3/g with 30% glycerol, thusshowing that the hydrodynamic volume is reduced in thepresence of glycerol. The Huggins interaction coefficientkH is increased from 0.21 to 0.28, indicating a strongerattraction between the protein coils in the presence ofglycerol. For A1, the effect was less pronounced and cannotbe quantified. At higher gelatin concentrations, above 0.02g/cm3, the viscosity η of semidilute solutions is stronglyincreased as shown in Figure 12 for tuna, and a similareffect is seen for A1 for the same concentration range. Theenhancement of viscosity is noticed when either ηsolution,or the ratio ηsolution/ηsolvent, or the difference (ηsolution - ηsolvent)is plotted versus concentration. Such an effect can beascribed to the formation of efficient entanglementsbetween the protein chains in the presence of glycerol.These interactions may involve the hydrophobic groups,as observed in “associating polymers”, which are hydro-phobically modified water-soluble polymers. However,gelatin solutions remain Newtonian, in contrast with theaqueous solutions of associating polymers. Recently,Wulansari et al.24 discussed the Newtonian behavior of

gelatin solutions (in a phosphate buffer, at pH 7) incomparison with polysaccharide solutions, which ingeneral are non-Newtonian. The authors raised thequestion of the entanglement mechanism of the proteincoils. Our results also stress the role of the entanglementsin the semidilute solutions and also the specific role playedby the solvent.

ConclusionIn this paper we have extensively examined the helix

formation of gelatin solutions of many sources, molecularweights, and isoelectric points and we have modified thethermodynamic conditions for gelation and melting byvarying the temperature, changing the solvent (purewater, mixed solvents), and blending different gelatins.This investigation provides a clear comparison betweenthe molecules themselves (imino acid compositions, mo-lecular weights) and puts into evidence the influence ofthe environment on the helix formation. In view of suchdiversity of behavior, the search for the common featuresbetween all gelatin gels becomes imperative, and this isthe topic of the next paper.

Analysis of our experimental results in the context ofthe well-known Zimm-Bragg theory shows that a moreelaborate theory that takes into account the building ofthe network and interaction of the macromolecules withthe solvent should be necessary.

Acknowledgment. This work was performed in thecontext of European Contract FAIR CT 97-3055. We thankall our partners for numerous and fruitful exchangesduring the project. We are indebted in particular to Dr.M. Gudmundsson, Dr. P. Montero, and Dr. G. Takerkartfor their expertise in gelatin preparation and character-ization. We also thank Dr. M.-M. Giraud-Guille and Dr.L. Besseau for their precious help in the preparation ofsoluble collagen solutions and Dr. A. Ajdari for manystimulating discussions. We thank the reviewer for thethorough examination of our manuscript and for veryinteresting comments and suggestions that we took intoaccount in the final version.

LA020189N

(22) Arakawa, T.; Timasheff, S. N. Biochemistry 1982, 21, 6536-6544.

(23) Gekko, K.; Timasheff, S. N. Biochemistry 1981, 20, 4667-4676.(24) Wulansari, R.; Mitchell, J. R.; Blanshard, J. M. V.; Paterson, J.

L. Food Hydrocolloids 1998, 12, 245-249.

Figure 12. Viscosity versus concentration for A1 in pure waterand in a mixed solvent of water and glycerol.

Gelatin Networks: Biodiversity and Physical Chemistry Langmuir, Vol. 18, No. 19, 2002 7217