摘要:

The Critically Branched State in a Covalent Synthetic System and in the Reversible Gelation of Gelatin BY C. A. L. PENICHE-COVAS, S. B. DEV, M. GORDON," M. JUDD AND K. KAJIWARA Dept. of Chemistry and Institute of Polymer Science, University of Essex, Colchester Received 28th December, 1973 The critically branched state of matter (close to a gel point) is further explored theoretically and experimentally. Fluctuations and effects due to cutting the high molecular weight tail of the distribu- tion are examined in terms of asymptotic formulae derived from the Flory-Stockmayer distribution for random fifunctional condensates. " Tailcutting " can arise from preparative and instrumental limitations, or from statistics alone when systems of relatively small size are involved. A law of quasi-invariance is deduced, according to which random critically branched materials will appear to have constant weight average molecular weight M,, irrespective of their real (higher) M,, if the molecular weight distribution is always cut at a fixed size-limit.Model polycondensates of decamethylene glycol/benzene 1 : 3 : 5 triacetic acid (DMG/BTA) are further studied near the gel point. The relation 7 cc Mw for the bulk viscosity 7 has been con- firmed and serves to locate the gel point. The relative conversion a/% can be determined to a few parts in ten thousand. Light scattering and intrinsic viscosities have been measured on sol fractions extracted from three post-gel samples containing about 1 to 5 % gel. The statistical inference that the newly-formed gel must be practically ring-free (" tree-like ") is confirmed, since highly branched sol molecules up to size lo8 are extractable with high efficiency.The evolution in time of Young's modulus E (300-2000 dyn cm-2) of jellies prepared by isothermal gelation of aqueous gelatin has been followed in a micro-sphere rheometer. A classical gelation process by random cross-linking is postulated, each cross-link consisting of a rapidly formed helical section comprising n primary chains, locally twisted together. With a " front-factor " g around Q, the classical rubber elasticity equation E = 3gNeRT/ Vis reasonably fitted to the measurements without adjustment of parameters, for n = 3 but not n = 2. 1. INTRODUCTION The classical theory of gelation l* allows not merely the calculation of gel points, but also provides the statistical framework for treating the critically branched state on either side of, and close to, such a point. Experimental investigations of this state are important both for a programme of extending our understanding towards the real state of gels and rubbers, and because life processes occur in critically branched materials.The characterization 3-9 of model polycondensates of decamethylene glycol (DMG) and benzene 1 : 3 : 5 triacetic acid (BTA) is here carried forward, using critically branched samples whose COOH end-groups were stabilized by benzhydrylation with diphenyl diazomethane (DDM), and whose OH end-groups were-for the first time- additionally stabilized by acetylation with ketene (K). The applicability to DMG/ BTA of the classical theory based on tree-like molecules (ring-free, apart from the trivial benzene rings introduced with BTA) is well documented.The kinetic substitu- tion effect of reacted functionalities on neighbouring unreacted ones is not measurable ; the cyclization effect is small though definitely measurable in DMG/BTA. The gel point was found to be 0.722+0.006 in five kinetic runs by Love lo (cf. ref. (6), (7)) and more recently 0.71 8 & 0.005 in five such runs by Ross-Murphy.' The classical 165166 REVERSIBLE GELATION OF GELATIN theory predicts 0.707 = 2-3. From these results, from the effect of dilution on the gel point,' from chemical kinetics," and from statistical calculations based on bond lengths and angles (cf. ref.(12)) together with rotational barriers,13 a revised estimate of the cyclization effect assigns to bulk condensates of DMG/BTA at the gel point about one intramolecularly formed ester link for each hundred (rather than ' fifty-five) or so intermolecular ones. This paper shows theoretically that it is meaningful to search for sol molecules of very high molecular weight ; evidence is then presented that critically branched sol molecules of molecular weight M = 108 can in fact readily be extracted from a gel. 2. FLUCTUATIONS AND ASYMPTOTIC FORMULAE FOR CRITICALLY BRANCHED SAMPLES For irreversible systems, the familiar macroscopic fluctuations near a critical point, take the form of compositional differences from sample to sample. For reversible systems, Whittle l4 showed how the kinetic equations for f-functional random condensation could be generalized to a stochastic form, from which, in accord with the classical Gibbs distribution, the number fraction nx of x-mer emerges as a Poisson variable at fixed conversion.For an infinite system this converges to deter- ministic values. In the close vicinity of the gel point such statistical fluctuations are of potential importance for the experimentalist, and for cell chemistry. Some statistical formulae are quoted below (see section 4.1) for random poly- condensates of DMG and BTA, i.e., a difunctional and trifunctional unit. Many properties of such condensates can, however, be adequately approximated by setting f = 2.5 in the model of anf-functional random polycondensate, whose weight fraction distribution is (at conversion a) : w, = [cfx-x)!f/(x- 1)!(fx-2x+2)!]aX-1(1 --a)fX-2X+2.(1) To calculate physical quantities near a critical point requires high-order approxima- tions. Using Stirling's approximation in the form n ! - (2nn)3nn exp ( - n) (2) (n - t)" N n"/e' (not merely nn), (3) and taking exponential limits as the weight fraction w, of eqn (1) reduces to forf > 2 to W, = BA~/X*+U(X) (f> 2, x 3 03). Here and At the gel point,' and therefore so that for any f > 2. A = (I - a)f-2a(f- 1)f-l /(f- 2)f-2 B = (1 - a)Zf(f- 1)+/(2n)+.(f- 2)s. = l/Cf- 1) A , = 1 ; B, = f/[2nCf- 1)cf-2)]3 w, - const x-3 (gel point) (4) (5) (7) (9)PENICHE-COVAS, DEV, GORDON, JUDD AND KAJIWARA 167 Direct computation shows that the simple form w, - BA"/x* (eqn (4)) forms an excellent approximation to eqn (l), at or near the gel point (see table 1).Of course, the factorials on the right of eqn (1) could not be computed individually for large x without computer " overflow ", but In w, can be evaluated readily (using the recur- rence relation of the I' function), and hence wx(< 1) by exponentiation. TABLE ~.-COMPAFUSON OF ASYMPTOTIC FORM wx = B & / d (EQN (4)) WITH EXACT NUMERICAL CALCULATION FOR CRITICALLY BRANCHED RANDOM POLYCONDENSATE (f = 2.5) number x of r---- 7 r.---------------7 10 3.648 x lo-' 2 . 1 3 0 ~ 3.642~ 2.126~ lo00 3.647 x 3.625 x 3.642~ 3.619~ 5000 3 . 2 6 0 ~ 3.256~ 3.257 x 3.253 x DPw = 5x lo3 DP, = 106 repeat units wx (eqn (4)) H ' ~ (exact) wx (eqn (4)) wx (exact) 100 1.154~ 1.085 x 1.152~ 1.084~ 10-3 Note that columns 3 and 5 are almost identical.The practically constant distribution at the low molecular-weight end extends to the sol fraction of critically branched samples (section 6a). 3. MATHEMATICAL THEORY OF GELATION I N FINITE SYSTEMS (i) EMERGENCE OF THE GEL MOLECULE The w,-distribution (eqn (1)) applicable to the thermodynamic limit of a system comprising an infinity of repeat units, becomes statistically heterogeneous l4 at the gel point. From the lack of normalization : Q) c w, = l-w,(a) 1 one infers the emergence of a delta-function contribution wg(a)6(co - x ) to the distribution, placed at x = 00 and of intensity w,(a). Thus w,(a) becomes the weight fraction of the infinitely large gel molecule. The largest molecule present becomes separated in size by an infinite gap from the rest (sol).The private a-value of any very large tree-like molecule is necessarily extremely close to 2/f, i.e., larger by a factor 2(f- 1)lfthan the average a, over the whole system. As a grows > a,, the need to sustain the growth of w, at its inflated a-value (= a,,,) causes a of the rest of the system (E &), i.e., of the sol, to decrease. For finite systems of N repeat units, the mean Zmax of the size xma, of the largest molecule (averaged over many such systems) is calculated as a function of a, to lie within less than a factor of three from the solution of N J [w,(a)/x] dx = 3. %ax This estimate arises because the probabilityp(x,,, > ZmaJ - 3, and the error limits are easily established, using (4).Fig. 1 shows a plot of such solutions, based on eqn (4)-(6), for a system of 3 mg of 2.5 functional system with repeat units of molecular weight M = 200. The catastrophic growth rate of the largest molecule very close to the gel point is evident even in this finite system, though the largest molecule now grows by a few powers of ten rather than to infinity. The point A marks the smallest size of a gel molecule, viz a calculated gel fraction of 2 % of the 3 mg sample, on which the classical theory of rubber elasticity could be approximately ~erified.~ This cor- responds to a molecular weight of 4 x168 REVERSIBLE GELATION OF GELATIN At fixed a, large fluctuations about Zm,, of the largest molecule from system to system occur near the gel point. Eqn (4), (8) and (11) also show that, at the gel point, Xmax increases (for any f > 2) by a factor lO2I3 = 4.64 when the size N of the system is increased 10-fold. 0.999 1.000 1.001 1.002 4% FIG.1.4rowth of gel molecule in finite system. Logarithm of the molecular weight of the largest molecule in a random condensation system of l O I 9 repeat units of molecular weight 200 (i.e., about 3 m g total weight) and mean functionalityf = 2.5 as function of conversion near the gel point (using eqn (4)-(6)y and transformation (13), to solve eqn (11) by computer). A: approximate size of gel molecule found to obey rubber elasticity theory.3 B : data point calculated at gel point proper (i.e., using eqn (4), (8) and (1 1) in slide-rule calculation). C : indication of error limits due to asymptotic approximation, see text.(ii) EFFECT OF CUTTING THE TAIL OF THE w,-DISTRIBUTION : THE QUASI-INVARIANCE PRINCIPLE Cutting the high molecular weight tail of the distribution, from a chosen limiting number of repeat units L, is calculated with negligible error to reduce the weight average degree of polymerization to DPw,,,t = .N[Ds,,.,.,- 1; w,x dx] = N[DPW,,...,-B AXf3 dx . (12) s 1 Here the normalizer may be taken as unity for L > 1000. DPw,cut is evaluated from (12) using the transformation B f A 5 -* dx = B( n/ k)* erfc (kL)* with k = -In A.PENICHE-COVAS, DEV, GORDON, JUDD AND KAJIWARA 169 a 9 FIG. 2.-Effect of cutting molecular weight tail of random distribution (eqn (l), f = 2.5, and repeat unit of M = 200) on M,, comparing values before and after cutting.(a) Ratio Mw,cut/Mw,uncut as function of molecular weight L beyond which species are removed from distribution. The four curves refer to different Mw,uncut, viz A : los, B : lo6, C : lo’, D : lo8. (bj Principle of quasi-invariance revealed by plotting analogous results for fixed MW,CUt, viz A : 5 x lo5, B : 5 x lo6, C : 5 x 10’. Nearly vertical course of plots implies that for fixed cutting limit (e.g., instrumental cut-off) and widely varying Mw practically constant will be observed.170 REVERSIBLE GELATION OF GELATIN The complementary error function was computed from a slightly modified version of the series given by Matta and Reichel,16 arid found extreniely accurate when checked against published values obtained by polynomial approximati~ns.~~ Fig.2 assesses the effect of tail-cutting in terms of the ratio DPw,cut/DPw,uncut, by plotting this quantity against L, either at constant DPwBuncut (fig. 2a) or at constant DPw,,Ut (fig. 2b). The almost vertical course of the latter curves reveals a remarkable principple of quasfinvariance of the random distribution, which follows mathematically (forf > 2) from eqn (1) : if we measure DP, for samples approaching the gel point in an instrument with a certain built-in cut-off level L at the high molecular weight end, these samples will all show an almost exactly constant result (DPW,,,J, depending only on the fixed value of L, while DPw,uncut could vary 100-fold or more. This law of quasi-invariance might also be a useful feature for control of cell processes involving precipitation.4. CALIBRATION OF CRITICALLY BRANCHED SAMPLES ON THE BASIS OF CHEMICAL KINETICS The basic measurement techniques have been described in previous publications,6* and in detail in three theses.ll* 18* l9 The structure of critically branched DMG/BTA polycondensates is determined as a function of the relative fractional conversion a/a,. A i B C D E 1 I 1 2 3 log107 FIG. 3.-Typical plots of log (melt viscosity) against logM, (calculated from eqn (16) and (17)). The gel time tc was assumed to be due to occur A 0.25 min ; B 2 min ; C 3 min ; D 4 min ; E 7 min after the last viscosity measurement was completed at 9 - lo00 poise (after which the sample was quenched). The linearity of C is accepted as the criterion for choosing the correct tc.The effect of accepting B or D would be to change the value of ar/occ calculated for the quenched sample from 0.999 596 to 0,999 730 or 0.999 461 respectively,PENICHE-COVAS, DEV, GORDON, JUDD AND KAJIWARA 171 This is one of the most accurately measurable parameters of polymer science,* for example the great accuracy of measurements of time and the enormous rate of change of, e.g., the bulk viscosity with time near gelation (fig. 3) can be exploited to deter- mine a/a,. 4.1. VISCOSITY-MOLECULAR WEIGHT RELATIONSHIP The bulk viscosity q at various conversions a of the stoichiometric polycondensate DMG/BTA was determined l9 by the Stokes-Fax& formula from the viscosity of fall under gravity of a steel sphere of 0.5 mm diameter through the reaction mass at 90°C.The mass was intermittently stirred until close to the gel point ( - 50 poise). The typical plot of log Mw (see below) against log q in fig. 3 illustrates how the crude estimated of the gel time tc was refined, through the linearisation of the plot. In this way, t, is found to an accuracy of f l min, which corresponds to a change Acc - 0.0002. This gel time of ca. 1200 min, after an initial reaction period of 4 min at 170°C to achieve rapid homogenization, was reproducible to a few minutes. For sixteen runs, the slope tc of the linearized plot of In ~7 = k’ In M,+k” (1 5 ) had an average value of 1.03 and a standard deviation of 0.03 in good agreement with earlier results.3* The slope of unity is generally observed also for melts of linear polymers in the range of M , up to 30 000-50 000, when a sudden transition increases the slope to about 3.4.The transition is often attributed to the onset of chain entanglements and, as fig. 3 shows, it is not reached in a critically branched system even at M, = 5 x lo5. No doubt this is due to the high segment density and low weight-average niean square radius in such systems. 4.2. DETERMINATION OF M, AND RELATED PARAMETERS l9 With the aid of tc as a scaling factor, determined as just shown to about 1 part in 1200, and accepting a, = 0.721 as the experimental value (see above), 1 -(a/a,) is determined near the gel point as follows. Taking a, = 0.720 as the best experimental value (see section l), a is computed from the equation : (16) This has the form dictated by the third-order mass law equation.The temperature- independent constant (= 2(da/dt),t,) in (1 6) has been adjusted from 0.257 (from the mass law) to 0.272, the mean experimental value, found by Ross-Murphy l1 with an accuracy of +6 % from five experimental runs at 170°C. This induces the same +6 % error in 1 -(a/.,) or in M, near the gel point, while an error in cc, itself of + 1 % also induces an error of f 1 % in 1 -(ct/ac) or in M,. The difference between the experimental value, 0.272, and the mass-law value, 0.257, is about equal to the probable error in measurements, but it happens to correspond to the expected devi- ation from 3rd order kinetics by virtue of the small cyclization component.ll The calibration in the earlier work by Love,lo using less refined equipment, had yielded the value 0.329 for the calibration constant in eqn 16.The effect of the revision of the calibration on the front factor in the measurements of Young’s modulus of DMG/ BTA gels is discussed in section 7.3. The calculation of M, from a/a, was carried out in this work using the following formulae, which generalize that given for DP, earlier, to allow for the nature of different monofunctional groups attached to BTA units and to DMG units. For (( 1 -a,)/( 1 - 0 1 ) ) ~ = 1 - (0.272[1- (t/t,)]/( 1 - u,)] (a, = 0.720).1 72 REVERSIBLE GELATION OF GELATIN fig. 3 both these end units are hydroxyls ; in stabilized samples they represent benz- hydryl and acetyl units respectively. From the routine cascade formalism, we obtain : [I -2a2]Mw = [M,, Mrt, M2, M3] x 3-2a(l +a), 241 -a), 4a(l -a), 3(1 -a) 241 -a), 2-41 +2a), 2(1 -a), 3a(l -a) 1, 1+2a2, 3a a, 2% 1 +a2 (17) where the Mi are molecular weights and the rnj mass fractions of the units.The weight-average degree of polymerization DP, is given by the same equation with Mi = 1 (all i). The indices represent 1 = end unit on BTA; 1’ = end unit on DMG ; 2 = DMG ; 3 = BTA. The mass fractions are given thus : m1 = 2(1-a)Ml/N (18) ml’ = 2(1-a)Ml/N (1 9) m2 = M2/N (20) m3 = 2M3/3x (21) (22) and the normalizer N 2( 1 - a)(M1. + Mi) + Mz + +M3. The structure parameter b was calculated from the measured mean square radius (S2), by the following formula, which is derived using the multitype cascade formalism on the assumption of Gaussian subchains* : 2b2 2DP, (1 -2a2)2<S2), = -[4a2((3-2a2)rn1 +2(1 +a-a2)m1’+ 2(1 +2a)m2 + 3(1 -a)ni3> +(1 -2a2)((5 +6a- 6a2)m1 + 3( 1 + 2a - 2a2)m ‘ + 2( 1 + 4a)m2 + 3( 1 + 2a)m3)] (23) which is well approximated for critically branched samples by ( S 2 ) , N b2/(1 -2a2) (as a -+ a, = 2-9.(24) The parameter b represents the root-mean-square distance between the centres of neighbour units in solution; more precisely a weighted average over such root- mean-square distances for different types of neighbour pairs. Only the pair DMG/ BTA contributes appreciably to this average when a - a,. by the cascade method. For the range here treated, an amply accurate treatment, which gives M, to 0.3 %, replaces a in eqn (17)-(24) by For the sol fraction, the Mi, M, and ( S 2 ) , are exactly calculable 19* 61 = (1-a2)+ [I < a/a, < 1.0041 (25) for DMG/BTA, irrespective of end-group stabilization.s = m“(1 -a2))”/a4]+m”[(1 -a)(1 +a--2)/a3]+ for the case in hand, with the mi given directly by eqn (18)-(22). The usual calculation of the sol fraction itself leads to the formula : m2[(l -a)(l +a-a2)/a3l2+m3[(l -a2)3/a6] f 1 -G (26)PENICHE-COVAS, DEV, GORDON, JUDD AND KAJIWARA 173 9 - 8 - \ c G 2 7 - 6 - 5. PROPERTIES OF SOL FRACTIONS FROM CRITICALLY BRANCHED DMG/BTA STABILIZED AFTER EXTRACTION Light-scattering from dilute solutions of pre-gel samples of DMG/BTA/DDM gives Zimm plots from which reliable values of M,, (S2>>,, and b are found, and the constancy of b was confirmed over a 15-fold range of M,. The value of b will be compared with a priori calculations from the rotational isomeric state theory in a further paper.We here report on three experiments to characterize the structure of extracted sol fractions in terms of b, M, and (S2), from light scattering, in terms of intrinsic viscosity and the calibration procedure based on chemical kinetics. Three condensation runs were allowed to over-run their gel times by 6, 7, and 27min respectively ; these times (ca. 1200 min) were determined (subsequent to the run) to about 1 min by the linearisation procedure of their viscosity plots (cJ: fig. 3). I L I 5.1. TECHNIQUE OF SOL EXTRACTION l9 For our weak jelly, an adaptation of the method of Purdon and Mate 22 was found to be the only suitable one to extract the sol. After rapid cooling of the jelly, methyl ethyl ketone is added and the gel broken by vigorous manual stirring.A fine suspension of highly gelatinous particles, much less than 1 mm in diameter, is then produced in less than 5 min. The sample is transferred to two 200-cm3 beakers and diluted to a total volume of 300 cm3. The small gel fraction, originally in sus- pension, then sedimented. The top half of the solution in each beaker was removed for centrifuging at 1000 r.p.m. for 1 h, while the remaining bottom half is diluted I 1 twofold with further MEK. This process is repeated six times. The gel is then transferred to the centrifuge tubes and alternately washed and centrifuged another twice. All the extracts after centrifugation are bulked, and stabilized with DDM and ketene.lg The delay in the stabilization till after sol extraction is estimated to lead to an increase in conversion, due to further polycondensation in the cold and diluted sample, of order which is negligible. There was visibly more gel in the sample 111 which had been allowed to over-run the gel point by 27 min, compared with samples I and I1 (6 and 7 min over-run respectively).The extraction technique does have the disadvantage that a sufficiently quantitative estimate of a small gel174 REVERSIBLE GELATION OF GELATIN content is not possible (Purdon and Mate 23). However, the sol fraction was avail- able in solution for sensitive physical measurements and produced excellent Zimm plots. An example is shown as fig. 4. 6. RESULTS AND DISCUSSIONS The properties of sol fractions (table 2), in comparison with those of pre-gel samples, confirm (a) the classical theory due to Flory,l* l5 that the sol fraction exhibits essentially the same random distribution (e.g.eqn (1)) as pre-gel samples, but with progressively diminishing conversion parameter & (e.g., eqn (25)), and Mw, (S2), ; and (b) the conclusion that DMG/BTA bulk condensates produce almost ring-free (" tree-like ") gels up to values of a/ac of at least 1.001, i.e., to the point at which the modulus of such gels becomes measurable and is found in satisfactory agreement with the classical theory of ela~ticity.~ The evidence for these two conclusions is now summarized. TABLE 2.-cOMPARISON OF PROPERTIES OF 3 SOL FRACTION WITH THEORY (BASED ON NEW CALCULATIONS, USING DATA BY PENICHE-cOVAS9) 2 3 4 5 6 7 8 9 10 l.111 <W,/ run t - f c 102mIgm-~~fwx1O-~ Mw*xlO-5: a/& (a/&)* G 10-3nm2 b/nm cf.fig. 3 exp. exp. eqn (16), (17) eqn (17) eqn (17) eqn (26) exp. eqn (23) and col. 4 and col. 5 I 5 0.206 8.19 5.49 1.0005 1.0007 0.0077 7.10 2.66 I1 6 0.200 4.39 4.58 1.0009 1.0008 0.0092 4.11 2.72 I11 27 0.130 1.75 1 .03 1.0035 1.0036 0.0413 1.77 2 x mean 2.74f0.1 * The starred parameters were calculated from the kinetic calibration. using B (eqn (25)) at the moment of stopping the reaction ; the other parameters were measured directly. (a) RANDOM STATISTICS OF SOL FRACTION That pre-gel samples of DMG/BTA obey random co-condensation statistics with remarkable accuracy is attested most directly by results based (i) on the low and (ii) on the high molecular weight end of the distribution : (i) the two monomers and the dimer were shown to follow closely the correct functions of conversion a, and (ii) the gel point to lie within 2 % of the theoretical value.When the reaction is allowed to over-run the gel time by from 5 to 27 min, a increases by only at most 0.5 % beyond a,, so notable changes can occur at most in the high molecular-weight part of the distribution, leaving the part which becomes the sol fraction practically un- changed (cJ: footnote to table 1). That such notable changes do occur is shown theo- retically in fig. I, and practically by the arrest of the motions of the steel ball, and by the presence of visible gel particles after the system is dispersed in MEK. The conformity of the extracted sol fraction with random statistics is further attested as follows.Though the low intrinsic viscosity measurements (table 2) cannot yet be interpreted in terms of quantitative theory, they agree qualitatively with a highly branched molecular structure ; moreover, they agree approximately with the intrinsic viscosity values of pre-gel samples l9 of comparable M,. The rapid drop in intrinsic viscosity with increasing relative conversion a/a, reflects the statistical prediction, which is more quantitatively revealed in the measurements by light scattering of M," and {S'),. The value of a/@, calculated from the measured M, (using e.g., (17) and a, = 2-9, is regarded as the most accurate available. (We recall that the real sample contains about 1 % (section 1) of cyclic links at the gel point, but that the relative overall conversion a/a, can be calculated in critically branchedPENICHE-COVAS, DEV, GORDON, JUDD AND KAJIWARA 175 samples with negligible error from the tree-like model of eqn (17)).The comparison in table 2, columns 6 and 7, with the values (a/a,)* calculated from the quenching time via the chemical calibration (section 4.2) is surprisingly good. The agreement was noticeably worse, though still quite acceptable, in seven pre-gel samples l9 quenched in the range 0.999 < a/a, < 1 .OOO to be reported elsewhere. The predicted rapid fall of M, and (S2), of the sol fraction with conversion is rather quantitatively confirmed by the results in columns 4, 5 and 9. Whittle recently put forward a theory 25 that in equilibrium samples the sol fraction distribution is independent of conversion.One would then expect the same to hold in random irreversible poly- condensations, but this is not borne out by the present results in table 2. A typical l9 Zimm plot (fig. 4) of a sol fraction, stabilized with DDM and K, exhibits the rectilinearity predicted by theory of the scattering envelopes for random polycondensates. The structure factor b of 2.74f0.1 nm of the three sol fractions is in excellent agreement with the mean b of 2.70k0.2 for seven pre-gel ~amp1es.l~ This agreement is the most telling argument for the essentially identical form of the distribution of pre-gel and sol-fraction samples. The small extent of the downturn at low angles in the Zimm plot confirms (fig. 4) the virtual absence after centrifugation of gel particles in suspension, and the conclusion that the sol fractions represented statistically homogeneous polycondensates of essentially random distribution and M , around 5 x lo6 is inescapable.The extremely low second virial coefficient (slope of line A in fig. 4) is, moreover, typical for critically branched samples. (b) CONFIRMATION OF THE ALMOST TREE-LIKE STRUCTURE The quasi-invariance law illustrated in fig. 2b, shows that if a random 2.5 functional polycondensate (closely resembling the case of DMG/BTA) returns a value of M, - 5 x lo5 in some measurement, a sharp cut-off of the molecular weight distribu- tion of the sample can be operative at a limiting molecular weight only well above lo8. In practice, any preparative or instrumental cut-off would not occur sharply at a fixed limit, but the conclusion that the sol-fraction samples contain most of the species predicted by the random statistics up to molecular weight around lo8 can be firmly drawn.In particular, extraction of sol molecules of molecular weight up to lo8 from the gel must be very efficient. This could not be the case if the gel contained a substantial proportion of intramolecular (cyclic) ester links. The best current estimate of this proportion is around 1 % for critically branched DMG/BTA prepared by bulk esterification (see introduction). The essentially tree-like structure of gel and sol is represented schematically in fig. 5. We estimate that in run I or I1 (table 2) the gel molecule of M - (or - 20 mg) was mechmically cut into fewer than lo6 particles of M 4 Each particle contained 95 % or more of sol fraction and became swollen with several times its weight of solvent.Before swelling, a particle would weigh rather more than 1 pg. Eqn (9) shows that there would only be about 1 molecule of each x-mer fraction of fixed x (- 5 x lo5) near M - lo8 to be extracted from a 1 pg particle near the gel point, so that experi- ments on sol fractions can perhaps be performed under conditions where the statistical truncation of the molecular weight distribution becomes significant. Flory wrote in 1953 “the gel must surely possess an abundance of intramolecular con- nections. These are an essential part of its network ”. Though he was clearly not thinking of weak gels in the critically branched region just after gelation, such gels are of special significance in biochemistry.It is obvious that the proportion of cyclic links will increase in DMGJBTA gels, as elsewhere, at higher conversions, but the when dispersing it in MEK. Fig. 5 runs counter to the current notion of the actual structure of a gel.176 REVERSIBLE GELATION OF GELATIN theory of rubber elasticity (and presumably swelling equilibrium) can be confirmed on samples in which cyclic links are rare accidents. This is fortunate, because the theoretical calculations of the number of active network chains are always based, implicitly or with explicit apologies, on tree-like models of gels. Gels which are almost though never absolutely, ring-free are accessible to synthesis, and their properties promise to be interesting.FIG. 5.-Schematic representation of extraction of large sol molecule (thin line) from surface of gef particle (thick line) in the essentially tree-like poIycondensate DMG /BTA. Critically branched DMG/BTA bulk samples have only about 1 intramolecular (cyclic) ester link, e.g., at A, B, C, in 100 such links. 7. CLASSICAL GELATION OF AQUEOUS GELATIN The reversible formation of weak gels, typical of living systems, is exemplified in vitro by aqueous gelatin solutions. The applicability of the classical theory of gelation to such systems is frequently doubted. By means of a moving-sphere micro-rheometer,26 the passage through the gel point induced by covalent linking has previously been charted for BMG/BTA and the results fitted to molecular models.3* ' Isothermal measurements on aqueous gelatin solutions in the same apparatus have now revealed a qualitatively very similar pattern : the viscosity rapidly diverges at the gel point, and Young's modulus E and dE/dt both rise from zero starting at the gel point.We now show that the appropriate molecular network model fits the evolution of E with the progress of gel formation roughly but satisfactorily. 7.1. THE TREE-LIKE STRUCTURE OF THE POSTULATED MODEL Ferry 27 summarized the evidence for a cross-linked network in gelatin gels. However, there is uncertainty whether two 28 or three 29 chains are involved in the helical segments which serve as cross-links, though the triple helix of collagen, the precursor of gelatin, is generally a~cepted.~O-~~ In the case of carrageenan gels, two chain cross-links have been described.33 Precisely such problems are usefully studiedPENICHE-COVAS, DEV, GORDON, JUDD AND KAJIWARA 177 in the state of critical branching, which occurs close to and on either side of the gel point.Here Young's modulus is proportional to the statistical network parameter N,, the number of elastically active network chains 3 4 9 35 per (single) primary chain in the system : where g is the " front factor ", R the gas constant, T the absolute temperature, and Vmol the volume per mole of primary chains. The classical statistical model is based on a tree-like structure for the molecular graph ; it is readily generalised for any number, n, of chains involved in a cross link formed at random between homodisperse primary chains, thus where there are x possible reaction sites on each primary molecule, a fraction a of which have reacted at any given time, and u is the extinction p r ~ b a b i l i t y .~ ~ For the 2 oon I a/% FIG. 6.-Plots of measured Young's modulus against relative conversion a/a,, from optical rotation measurements (cf. fig. 7), for three isothermal gelation runs at 26.9"C. The gel point (a/a, = 1) is determined to within A(a/ctc) -= 0.05 by the sharp divergence of the viscosity (not shown). Gel times tc : A-76 min, M-80 min, 0-77 min. Four theoretical curves plotted (eqn (27)-(30)) : A, x = 6 sites per primary chain, and n = 2 primary chains involved in each helix; B, x = 20, n = 2 ; C , x = 8 , n = 3 ; D , x = 8 , n = 4 .case of two chains forming a cross link (n = 2) eqn (1) reduces to eqn (34) of Dobson and Fig. 6 shows theoretical plots (calculated as shown below of N, for assumed x and n. As can be seen from curves A (n = 2), C (n = 3) and D (n = 4),178 REVERSIBLE GELATION OF GELATIN N, is very sensitive to n ; curves A (x = 6) and €3 (x = 20) show Ne to be relatively insensitive to the number of sites on each primary molecule. By computer, u was eliminated between (29) and (30), and q between (28) and (29), thus giving N, for various assumed values of n, x, and a ; or, using The theoretical curves A-D in fig. 6 were calculated as follows. a, = [(x - l)(n - I)]-l, the following is known Ne = Ne(n7 x, a l 4 . (32) Next Ne was inserted in eqn (27), together with Vmol and g, to produce curves A-D.Here Vmol = 1 . 8 0 ~ lo6 ml was calculated for 5.7 % aqueous solution of gelatin of molecular weight 105 OOO. The front factor g was taken as 3, the value proposed by James and Guth, and recently supported by diverse theoretical t r e a t r n e n t ~ . ~ ~ ' ~ ~ Having explained the construction of theoretical curves in fig. 6, we turn to the measured points in this figure. 7.2. CALIBRATION OF a/a, AS A FUNCTION OF t / t c The progress of helix formation, leading to gelation, after quenching an 5.7 % aqueous solution of Ilford SC200 gelatin, of essentially pure alpha chains (M = 105 OW), to a temperature between 26 and 29°C has been followed isothermally in 21 2201 0 0 5 0 I00 150 2 0 0 time/min FIG. 7.--Rate curve, serving for calibration in fig.6, of specific optical rotation as function of time for 5.7 % aqueous gelatin at 26.9"C. 0, gel time 72 min, 0, gel time 73 min. The smooth curve re- presents the optimal polynomial fitted : [a]~=165.56+0.395 22t-0.000 154 4t2+3.036 X 10-6t3-2.90~ 10-'t4. terms of optical rotation [a],,. At 26.9"C7 [a], increases by about 32+ 3" up to the gel point. The small volume change, about 1 in lo4 up to the gel point, is found roughly proportional to the change in [a],, but somewhat less accurately measurable (in a dilatometer). We assume that the change in [a], is proportional to the amount of helix formed, and to the number of helical cross-links, i.e., that each helical cross-link forms in a time short compared to the experimental time scale. The gel point in the polarimeter was measured by the bubble rise technique. The typical polarimetric rate curves shown in fig.7 are found to be reproducible and similar to those of Smith.40 Our rate curves were smoothed by computer as shown (fig. 7), and served for calibra-PENICHE-COVAS, DEV, GORDON, J U D D AND KAJLWARA 179 tion in fitting the experimental points in fig. 6 as follows. The gel point was deter- mined from the divergence of the viscosity in each micro-rheometer run. For each subsequent measurement of E during a run, the time t was noted, and t/tc was con- verted to a/ac using the polynomial equation of the smoothed plot in fig. 7. The error in tc was equivalent to < 0.05 in a/@,. 7.3. COMPARISON OF THEORY WITH MEASUREMENTS Of the three runs shown in fig.6, two fit quite well and one fits moderately well to the preferred model (curve C, n = 3). Although such a fit represents, in principle, a verification of classical rubber elasticity theory without adjustment of floating parameters, there could be fortuitous cancellation of fairly substantial errors in the calibration procedure. Besides, several assumptions are inherent in the attempted fit of the model, especially the relatively rapid formation of each helical section, and that the gelatin sample was sufficiently homodisperse for application of eqn (28). A detailed evaluation of assumptions had better await a more extensive investigation. The present results do, however, lend strong confirmation to the view that gelation and the initial formation of a “ network ” occur by the classical random mechanism as treated in rubber elasticity theory for covalent structures.Two points of detail deserve mention. Near the gel point (1 < a/a, < 1.2) the experimental points lie above the theoretical curve C. It is likely that relatively slow relaxation accounts for these high E values, an effect observed in the past with a variety of gels in this region, which generally become perfectly elastic bodies once sufficient crosslinking has occurred. As regards the front factor g, the value 0.5 assumed is confirmed by the measurements to only about 100 %, because of the possible error in calibration and in the gel time. Assuming the value of g = 1, curve D could be fitted, but a quadruple helix is unlikely. Curves for a double helix (A, B) would require an improbably large reduction in g.The measurements on DMG/BTA gels were originally fitted to the old value of g = 1.0. The revision (section 4) of the underlying chemical calibration of the (a/a,)-scale by about 20 % (viz 100 x 0.272/ 0.329) raises the fitted g from 1.0 to about 1.6, which is clearly too high. This dis- crepancy remains to be investigated further. C. A. L. P. C. thanks the Essex University Latin American Centre and K. K. and S. B. D. thank the Science Research Council for fellowships M. J. thanks Ilford Limited for his studentship. P. J. Flory, J. Amer. Chem. SOC., 1941, 63, 3083, 3091, 3096. W. H. Stockmayer, J. Chem. Phys., 1943, 11,45. M. Gordon, T. C. Ward and R. S. Whitney, Polymer Networks, ed., A. J. ChompK and S.Newman (Plenum Press, New York and London, 1970), p. 1. N. S. Clarke, C. J. Devoy and M. Gordon, Brit. Polymer J., 1971, 3, 194. M. Gordon, J. A. Love, T. G. Parker and W. B. Temple, J. Prakt. Chem., 1971, 213,411. J. Khlal, M. Gordon and C. Devoy, Makromol. Chem., 1972, 152,233. ’ M. Gordon and K. Kajiwara, Plaste Kautschuk, 1972, 4, 245. * W. Burchard, K. Kajiwara, M. Gordon, J. Kdal and J. W. Kennedy, Macromolecules, 1973, 6, 642. H. Suzuki, C. G . Leonis and M. Gordon, Makromol. Chem., 1973, 172,227. S . B. Ross-Murphy, Ph.D. Thesis (Essex University, 1974). lo A. J. Love, Ph.D. Thesis (Strathclyde University, 1968). l 2 M. Gordon and W. B. Temple, Makronzol. Chem., 1972, 160, 263. l3 M. Gordon, K. Kajiwara, C. A. L. Peniche-Covas and S. B. Ross-Murphy, to be published. l 4 P. Whittle, Proc. Cambridge Phil. SOC., 1965, 61, 475 ; Proc. Roy. SOC. A , 1965, 285, 501. l 5 P. J. Flory, The Principles ofPolymer Chemistry (Cornell University Press, Ithaca, 1953), p. 375.180 REVERSIBLE GELATION OF GELATIN l6 F. Matta and A. Reichel, Math. Comp., 1971,25, 342. '* R. S. Whitney, PhB. Thesis (Essex University, 1972). l9 C. A. L. Peniche-Covas, PhB. Thesis (Essex University, 1973). 2o M. Gordon and G. R. Scantlebury, J. Chem. SOC., 1967, 1. 21 G. R. Dobson and M. Gordon, J. Chem. Phys., 1964,41,2389. 22 J . R. Purdon and R. 0. Mate, Pohmer Letters, 1963, 1, 451. 23 J. R. Purdon and R. 0. Mate, J, Polymer Sci., 1-A, 1972, 10, 3111. 24P. J. Flory, ref. (15), p. 377. 25 P. Whittle, Suppl. Adu. AppZ. Prob., 1972, 199. 26 M. Gordon, S. C. Hunter, J. A. Love and T. C. Ward, Nature, 1968,217, 735. 27 J. D. Ferry, Advances in Protein Chemistry, 1948, 4, 1. 28 I. H. Coopes, J. Polymer Sci., A-I, 1970, 8, 1793. 29 W. F. Harrington and N. V. Rao, Biochemistry, 1970, 9, 3714. 30 G. N. Ramachandran, Treatise on CoZlagen, ed., G. N. Ramachandran (Academic Press, 31 A. Rich and F. H. C. Crick, Mol. Biol., 1961, 3,483. 32 W. Tranb, A. Yonath and D. M. Segal, Nature, 1969, 221, 914. 33 D. A. Rees, I. W. Steele amd F. B. Williamson, J. Polymer Sci., 1969, 28, 261. 34 L. C. Case, J. Polymer Sci., 1960,45, 397. 35 J. Scanlan, J. PoZymer Sci., 1960, 43, 501. 36 G. R. Dobson and M . Gordon, J. Chem. Phys., 1965,43,705. 37 S . Imai and M. Gordon, J. Chem. Phys., 1969,50,3889. 38 S . F. Edwards, in Polymer Networks, ed., A. J. Chompff and S. Newman (Plenum Press, New 39 €3. E. Eichinger, Macromolecules, 1972, 5,496. 40 C. R. Smith, J. Amer. Chem. SOC., 1919, 41, 135. C. W. Clenshaw, Math. Tab. Nat. Phys. Lab. (H.M.S.O., 1965), 5. London, 1967), vol. 1, p. 103. York and London, 1971), p. 83.

ISSN:0301-7249    

DOI:10.1039/DC9745700165

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Studies of the Collagen Fold Formation and Gelation in Solutions of a Monodisperse a Gelatin BY D. EAGLAND," G. PILLING AND R. G. WHEELER School of Chemistry, University of Bradford, Received 28th December, 1973 The reversion of a monodisperse 01 gelatin to the collagen fold over a concentration range up to 5 % w/v has been studied by n.m.r., dielectric relaxation, differential scanning calorimetry, optical rotation, viscometry and light scattering techniques. The folding process is shown to be intra- molecular and the minimum concentration for gelling is correctly predicted from the behaviour of less concentrated solutions. Collagen is the principal protein in mammals, accounting for approximately 60 % of their total protein content. It is unique among proteins because of its amino acid composition; it is the only mammalian protein which contains large amounts of hydroxyproline, and is also extremely rich in both proline and glycine.In contrast with other proteins the sulphur containing protein cysteine is completely absent, and methionine is only present in trace amounts. The high pyrrolidine ring content of the protein (-25 %) precludes the formation of extensively intramolecularly hydrogen bonded structures, and the lack of sulphide groups prohibits the formation of disulphide linkages which are a major stabilising factor in the native helix conformation of many proteins. Several models for the structure of collagen have been proposed which differ in detail, but all have a common feature that the presence of glycine at every third residue position is required along the poly- peptide chain ; this is in agreement with the known glycine content (- 33 %) of the total amino acid residues of the protein.The basic structure of the collagen molecule is visualised as a triple helix of coiled coils consisting of three left handed helices of a polyproline 11 type conformation, winding in a right handed sense about a common axis; the presence of glycine at every third residue position allows the possibility of forming a limited number of intramolecular hydrogen bonds between the N H groups on the backbone of one chain and the -C=O groups on the backbone of another chain. At temperatures above 40°C, collagen loses its ordered triple helical structure to adopt a random coil conformation which consists of single, double and triple strands, a, p and y gelatins.Cooling of the gelatins initiates a reversion to the helical form, the so-called reversion to the collagen fold, but this is not a simple transition to an intact collagen unit, at least not in the case of the single strand a-gelatin ; the situation is more uncertain with double and triple strand gelatins and unfortunately almost completely confused in many systems reported, containing a mixture of a, p and y chains. If the concentration of the gelatin(s) in solution is sufficiently high ( 2 0 . 5 % wlv), then on cooling, in addition to regeneration of a helix conformation, gelation occurs and the material sets to a semi-solid mass. Several mechanisms for the gelling process have been proposed ; these mechanisms have usually been advanced on the basis of 181182 COLLAGEN FOLD A N D GELATION OF a GELATIN studies of relatively high concentration gels which contained a mixture of gelatins or the double strand B gelatin.5 A further serious complicating factor in both the process of regeneration of the collagen fold and gelation is the electrolyte content of the gelatin solution.It has been suggested that low concentrations of the order of 0.1-0.15 mol dm-3 are without effect upon the reversion process but this has recently been shown to be in error ; it is, therefore, particularly difficult to correlate data and mechanisms put forward for different gelatins at different concentrations containing different quantities of different electrolytes. This report is concerned with the examination of a pure sample of a gelatin obtained by the use of very mild techniques with the exclusion of salts.The reversion of the a-gelatin to the collagen fold and gelation have been observed by a variety of techniques at concentrations up to and beyond that at which gelling occurs. Consistent data were obtained from all the techniques used in the investigation which must inspire greater confidence in any deductions which may be formulated since, as pointed out by Brandts,8 it is especially dangerous in studies of biopolymers to deduce on the basis of limited experimental data. EXFERIMENTAL (i) Preparation of a gelatin : the starting material was obtained from the tail tendons of young specimens (<3 months) of the common rat R. ruttus; after dissecting from the tail the tendons were washed with 0.5 mol dm-3 sodium phosphate solution to remove blood, serum, etc.Studies using radio tagged phosphate solutions confirmed that no detectable quantities of salt were carried through into subsequent steps of the purification. The tendons were subsequently washed with several aliquots of distilled water prior to being allowed to swell and dissolve at 45" in distilled water. The a gelatin was isolated from the resulting mixture of 8, a and y gelatins by a coacervation procedure using an ethanol+ water system. 8 and y gelatins were removed as a coacervate phase by adjusting the volume composition of the total solvent system to 2.5 : 1, ethanol: water ; a gelatin was removed as a coacervate from the supernatant liquid by adjustment of the ethanol : water ratio to 4 : 1.The coacervate was dried in a stream of dust-free air until the ethanol present had been removed by evaporation, the solid a gelatin remaining was dissolved in distilled water to give a clear transparent solution. The concentrations of all gelatin solutions were obtained by air drying at 120°C to a constant weight. Infra-red studies confirmed that, after drying, water was absent from the spectrum of a gelatin. The final material was shown to have a molecular weight of 110 OOO+ 10 OOO by light scattering and viscometric studies.1° The good agreement on the value of the molecular weight between a method based on number average molecular weights and one based upon weight average molecular weights confirms that the material under investigation is a single strand monodisperse a gelatin.(ii) Distilled water was obtained from a continuous distillation apparatus based upon the design of Franks l1 ; all water used had a specific resistance better than 5 x lo7 ohm-l cm-'. (iii) Ethanol : absolute alcohoI, reagent grade. OBSERVATION OF THE COLLAGEN FOLD AND GELATION OPTICAL ROTATION The instrument chosen as most suitable was the Bendix NPL Automatic Polarimeter Type 143, an electronically controlled, self-balancing photoelectric polarimeter which utilises the Fmaday electro-optic effect to measure the rotation of the plane of polarisation of light passing through the cell. There are no moving parts and the sensitivity (0.OOOl" of arc) is such that, with no sacrifice of relative precision, the path length of the sample can be about one tenth of that required in other instruments.This means, in practice, that a sampleD. EAGLAND, G . PILLING AND R. G. WHEELER 183 volume of about 1 ml may be used without loss of precision; such a small volume, with its short thermal equilibration time is essential for the accurate observation of the initial, rapid changes in optical rotatory power exhibited by gelatin solutions during the reformation of the collagen fold. Prior to measurements being taken on a gelatin solutions the sample was held at 45" for four hours to eradicate the thermal history of the material. VISCOSITY Ubbelohde suspended level capillary flow viscometers type BS 1P SL71 of 25 ml capacity were chosen. The gelatin solutions used gave efflux times of the order of 200s, so that shear rates were of the order of 200s-'; under these operating conditions the Reynolds number was in the range 10-100, so that flow was always non-turbulent. All viscometers were stored in chromic acid and, before use, were thoroughly rinsed with distilled water followed by AR acetone; they were finally blown dry in a stream of filtered air.As in the case of optical rotation studies the thermal history of the gelatin solution was eradicated by heating to 45" for several hours ; this was followed by transferring the gelatin solution at this temperature to the viscometer, already placed in a thermostat bath at the required temperature. Prior to any efflux times being recorded care was taken to wet the sides of the capillary of the viscometer with gelatin solution. All efflux times were measured with a normal stop watch graduated in units of 0.2 s.The quench time relevant to each reading was taken as the time at which half the solution had drained back into the lower bulb of the viscometer. Blank determinations with water in which the viscometer was quenched from 45" to 10" revealed that thermal equilibrium was not attained until ten minutes had elapsed ; readings were not attempted therefore for quench times less than ten minutes. LIGHT SCATTERING All scattering measurements were undertaken using a Sofica model 42000 photogonio- diffusometer. The wavelength of the incident light was 546nm and the intensity of the scattered light was measured as a function of the scattering angle.An advantage of this instrument is that certain angles can be present on the cell mounting stage and the stage will subsequently move progressively through these angles with a considerable saving in time. NUCLEAR MAGNETIC RESONANCE N.m.r. spectra were obtained using a Varian XL-100 high resolution spectrometer operating at 100.1 MHz and equipped with a Fourier Transform accessory. Proton spectra were obtained in 6 min, accumulating 200 transients in 20 blocks of 10 each from samples in 12 mm tubes. Fourier transformed spectra were displayed in the " absolute intensity " mode in order to facilitate monitoring changes in intensity from one spectrum to the next. The integrated area was measured for various sections across the spectrum. The gelatin sample was again warmed to 45" prior to cooling to 15", the temperature at which the n.m.r.spectra were obtained ; the quench time after cooling the sample was taken to be the midpoint of the six-minute accumulation period. Where necessary data were stored on magnetic tape and processed after an experimental run was complete. Samples for n.m.r. were obtained by dissolving the a gelatin in distilled water, freeze- drying overnight and redissolving in 99.8 % D20 just prior to their examination ; samples of 0.202 and 0.506 % w/v were used, the higher concentration solution becoming a gel during the period of examination of the proton resonance spectrum. DIELECTRIC RELAXATION Dielectric measurements at spot frequencies in the range 150 kHz to 5 MHz were made using a Wayne Kerr B 201 RF Bridge and a sample cell of the type described by Jordan and Grant.12 Relaxation measurements were carried out using three variants of the techniques of Time Domain Spectroscopy (TDS).139 l4 In both cases a Hewlett-Packard 180 Time184 COLLAGEN FOLD AND GELATION OF a GELATIN Domain Reflectometry system (12.4 GHz) was used in conjunction with an Intertechnique Physioscope 800 as a signal averager.The shortest relaxation processes were observed by a combination of the " Direct Reflection " and " Transmission Coefficient Ratio " methods as discussed fully e1se~here.l~ Using these methods a dielectric relaxation spectrum is computed in terms of the frequency dependence of the complex permittivity (frequency range 100 MHz to 10 GHz).For the longer time relaxation processes the " Thin Sample " approach of Fellner-Feldegg l6 was used. Using the Physioscope 800 to subtract out the effects of some of the unwanted reflections in the TDS waveform in the manner described elsewhere l7 the dielectric spectrum in this case appears as a time-dependent rather than a frequency dependent form. DIFFERENTIAL SCANNING CALORIMETRY A Perkin-Elmer DSC 1B differential scanning calorimeter was used according to the technique described in detail elsewhere. l8 Essentially the method involved weighing the gelatin into an aluminium sample pan, water was added by micropipette, the pan was sealed and weighed. The sealed pan was heated at 70°C for 30 min and then cooled to - 70°C at a cooling rate of 8°C min-'. After being held at - 70°C for 5 min the temperature of the sample was raised to 70°C at a heating rate of 8 O C min-l.After completion of the experiment the weight of the sealed pan was checked to guard against inadvertent leakage of material. RESULTS OPTICAL ROTATION The variation of optical rotation of a gelatin solutions of concentrations ranging from 0.034 % w/v to 1.23 % w/v were observed at 5, 10 and 15°C as a function of time following quenching from 45" ; behaviour representative of all the solutions is shown in fig. 1, which depicts the variation of optical rotation with time of a series of a gelatin solutions at 10°C. The rate at which the optical rotation changed as a function of quench time, expressed as a plot of -loglo (daldt) against t, for the /2' 0.3- 8 a 8 - I I 1 50 I00 150 t/min FIG.1.-The variation of optical rotation as a function of quench time for several a gelatin solutions at 10T.D. EAGLAND, G. PILLING AND R . G . WHEELER 185 gelatin solutions at 15°C is shown in fig. 2 ; plots of a similar nature were also obtained for the 5 and 10" data. Linearity of such a plot is an indication that the reversion process under observation is occurring by a first order kinetic mechanism ; fig. 2 shows that two linear regions are apparent, the first over a time interval from zero quench time to approximately 20min and a subsequent region from 20 min onward. A similar behaviour pattern was observed by von Hippel and Harrington I9 for the rate of change of specific optical rotation during collagenase digestion of Fro.2.-- Log, (rate of change of optical rotation) as a function of time for a gelatin solutions at 15°C. The rate constant of the first order process was determined from the slope of the (-log (daldt), t ) plot ; values of this constant determined from the initial slope are not of a high order of accuracy, particularly for the lower gelatin concentrations, due to the instrument operating at its limit of sensitivity. The trend however, taking the data at 15" as an example, is a decrease of the rate constant from approximately 1.6 min-I for a gelatin concentration of 0.034 % w/v to approximately 0.3 min-1 for a gelatin concentration of 0.538 % w/v. This trend shows that, although a first order kinetic plot is obtained in each case, the value of the rate constant decreases with increasing gelatin concentration. Behaviour of this kind has been called pseudo-first order,20 indicating that although the initial reversion process is first order with respect to gelatin and is, therefore, an intramolecular process, another species is involved in this reversion step.Determinations of the rate constant for the slow process (table 1) reveal that this is a genuine first order process since, within the limits of experimental error, a constant value for the rate constant is obtained at each of the three temperatures (1.7f0.5 x min-I). Two exceptions, however, can be noted, the lowest gelatin concentra- tions 0.034 and 0.052 % w/v at 15°C where the rate constant for the slow process is approximately zero and 0.8 x min-' respectively.This suggests that at con- centrations below 0.1 % w/v of gelatin an additional factor may be involved in this folding process. No clearly defined difference in the magnitude of the rate constant186 COLLAGEN FOLD AND GELATION OF 01 GELATIN 1.3 1.4 1.4 TABLE DETERMINATION OF THE RATE CONSTANT FOR THE " SLOW " FOLDING PROCESS OF 01 GELATIN DETERMINED FROM OPTICAL ROTATION DATA gelatin concn. rate constant klmin-1 x 102 % wlv 5" 10" IS" 0.034 zero 0.052 0.8 0.085 1.5 0.104 2.8 0.160 2.1 0.21 4 I .8 0.230 0.260 1.3 0.270 2.0 0.290 0.316 1.8 0.320 0.332 1.8 0.390 0.424 2.3 0.450 0.480 0.530 2.5 0.538 1.5 0.580 0.630 1.4 0.760 2.1 1.23 2.5 2.8 1.7 2.5 2.5 as a function of temperature could be discerned, suggesting that the energy of activa- tion for this slow process is quite small, certainly <4 kJ mol-l.It should be noted that even at the higher gelatin concentrations, which are well in excess of the minimum concentration required for gel formation, the kinetic interpretation of the optical rotation data does not show any marked difference from that obtained for lower gelatin concentrations. This suggests that the process of reversion to the collagen fold is unaffected by the process of gelation. FIG. 3.-The variation of the viscosity of 01 gelatin solutions at 5°C as a function of quench time.D. EAGLAND, G . PILLING AND R . G . WHEELER 187 VISCOSITY Changes in the viscosity of the gelatin solutions were observed as a function of quench time for a variety of concentrations; fig. 3 illustrates the data obtained at a temperature of 5°C.An obvious difference in behaviour for the highest concentra- tion of gelatin (0.63 % w/v) can be seen. Kinetic analysis of the specific viscosity changes, for solutions other than the 0.63 % w/v solution, in a similar manner to that used in analysis of the optical rotation studies, resulted in linear plots of -log,, (d(qsp)/dt) against t, Some of which are shown in fig. 4. The collected d u e s of the conc "lo 3 -*--*-*. 4 i 0.278 0.186 --., 0.138 -,o.o9c 0.045 -t-+-i-/- 100 209 300 -00 506 609 ti me / m i n FIG. 4.-- Log,, (rate of change of specific viscosity) as a function of time for o! gelatin solutions at 15°C. rate constant of the flow process at 5, 10 and 15°C are listed in table 2 and show that, with the exception of gelatin solutions of concentration <0.1 % w/v, the process is truly first order and no detectable difference in the rate constant at the three tempera- tures can be seen.The 0.63 % w/v solution at 5" and the 1.25 % w/v solution at 15" do not follow a first order flow process; it seems clear therefore that zt these higher gelatin concentrations which are well in excess of the minimum required for gelation at the particular temperature, the viscous flow being observed is not a first order process, in contrast with the situation for lower concentration solutions. At low gelatin concentrations, <O. 1 % w/v, the flow process appears to be pseudo first order in character, as with the change in optical rotation. This appears to confirm that at these lower concentrations a species other than gelatin is involved in the reversion process.For all other gelatin concentrations at 5, 10 and 15" the value of the rate constant determined for the flow process agrees, within experimental error, with the value determined from optical rotation studies. It appears therefore that the rate of change of flow is directly determined by the rate of folding of the gelatin molecule. No opportunity was available to observe the viscosity changes associated with the Initial changes in optical rotation because of the longer time interval required for the iarger volume of solution to reach the equilibrium temperature.188 COLLAGEN FOLD AND GELATION OF Ct GELATIN TABLE 2.-DETERMINATION OF THE RATE CONSTANT FOR THE " SLOW '' FLOWING PROCESS OF a GELATIN DETERMINED FROM VISCOSITY DATA gelatin concn.rate constant klmin-lx 102 % wlv 5 O 10" 15" 0.045 0.065 0.090 0.133 0.138 0.166 0.1 86 0.190 0.202 0.278 0.290 0.300 0.320 0.328 0.390 0.400 0.41 8 0.450 0.460 0.462 0.500 0.530 0.630 0.650 0.760 1.6 1.4 1.7 1.8 1.6 1.3 5.0 3.1 3.4 1.1 1.3 0.9 1.2 0.9 1 .o 1 .o 1.2 0.8 1.2 1.7 1.3 0.6 1.1 1.3 0.8 LIGHT SCATTERING The reduced intensities of scattered light Kc/RO obtained from gelatin solutions w/v at 5, 10 and 15" were expressed as a function of up to a concentration of 0.23 sin2 8!2 where 8 is the angle of-the scattered light. According to Zimm * 16n2ri -(1+cos2 8) = R0 1 + 2 A , c + ~ sin2 - 3A Kc where K = -(-) 2n2ng dn NA4 dc and no : refractive index of the medium, n : refractive index of the mixture, A : wave- length of incident light in the medium, A2 : second virial coefficient, r, : radius of gyration of the molecule.Fig. 5 illustrates the dependence of Kc/Ro upon sin2 S/2 at 15". It is important to note that the intensities of scattering observed were found to be independent of quench time in all cases; this means that the changes in optical rotation and viscosity which occurred with time were not reflected in the scattering of light produced by the solutions. lo4 was obtained for the a gelatin by extrapolation of the reduced intensity to zero concentration and zero angle ; in addition, since no change of the intercept of the Kc/& axis occurred with time, there is no A weight average molecular weight of lo5D. EAGLAND, G . PILLING A N D R. G . WHEELER 189 evidence for aggregation of a gelatin molecules during formation of the collagen fold which was observed by optical rotation studies.Several authors have suggested 22* 23 that the folding process observed in dilute gelatin solutions such as those considered here corresponds to the reformation of the collagen triple helix in the form of a rigid rod. Privalov 24 has derived the following equation 2 4 +, sin 812 where ll is the osmotic pressure and ,u is the mass per unit length of the molec~ile, for the scattering of a rigid rod-like molecule. A plot of ( K C / R , J ~ = ~ versus sin 8/2 FIG. 5.-The sin2 8/2+ 1Oc ngular dependence of the reduced scattering intensity KclRe fc cc gelatin solutions at 15°C. should therefore give an intercept of 2/lJ2M; in this case however the intercept is negative, a situation with no physical reality.Privalov reported essentially similar results for solutions of tropocollagen and concluded that the molecule could not approximate to the shape of a rigid rod. The radius of gyration of the gelatin molecule may be determined from the Zimm equation but the assumption is made that the solutions are ideal; for a non-ideal solution the scattering is diminished, giving an increased apparent size for the particles. Deviations from ideality have been discussed by Zernicke et aLZ5 who showed that, as a result of interparticle interference of the scattered light, for nearest neighbours of spheres the following equation results where Ie/Icl 80 -0) is the disymmetry, v is the number of particles per unit volume and L is the effective diameter of the particle.A plot of 1 -[~e/rcl8o-e,lV/[Ie/r~(18~-~,Iv=~ against v resulted in a value for the effective radius of the gelatin molecule of 22 nm at all three temperatures. In the absence of highly charged species it is expected 26190 COLLAGEN FOLD AND GELATION OF a GELATIN that this effective radius will be the same as the real radius of the molecule. A sphere of real radius 22 nm has a radius of gyration equal to 17.2 nm which is in agreement with the value quoted by Boedtker and D ~ t y . ~ NUCLEAR MAGNETIC RESONANCE SPECTRA The n.m.r. spectra of both the 0.2 and 0.5 % w/v solutions of a gelatin when freshly prepared were typical of a random coil protein, in agreement with earlier data.27 The only change observed in each spectrum as a function of time after 1 I I I 510 Id0 160 2b0 tlmin FIG.6.-Plots of integrated intensity against quench time of a 0.5 ”/, w/v solution for various regions of the a gelatin spectrum at 15°C. El : 61.02-61.47 ppm. 0 : 60.40-61.02 ppm. x : 62.50- 63.30 ppm. quenching was a decrease in the total intensity of the high resolution component, until, after several days, up to 90 % of the spectrum was lost. There were no chemical shift changes or changes in the line widths of peaks in the residual spectrum. If the disappearance of the spectrum intensity, I, obeys a first order rate law d In I/dr = - k a plot of In It against t should be linear with a slope of -k. Fig. 6 shows such a plot for various regions of the spectrum of the 0.5 % solution; a similar linear TABLE THE SLOPES OF LINES OBTAINED BY PLOTTING In It AGAINST t/min x lo4 FROM THE N.M.R.SPECTRUM OF a GELATIN AT 15°C 0.40-1.02 ppm 124+ 16 l05k 7 1.02-1.47 ppm 130+ 9 104f 8 2.50-3.30 ppm 123f 17 l05f 6 mean 126f 10 l05k 5 region of spectrum 0.2 % solution 0.5 % solution relationship was obtained for the 0.2 % solution. The values of the rate constant for the disappearance of spectrum are given in table 3 ; the mean value is very similar to the rate constant obtained from studies of the optical rotation and viscosity of the gelatin solutions.D . EAGLAND, G . PILLING AND R . G . WHEELER 191 DIELECTRIC STUDIES The primary aini of the dielectric studies was to examine the changes, if any, in the bulk solvent during the renaturing and gelling of the gelatin solutions.A secondary aim was to monitor any changes in the relaxation properties of the a gelatin occurring during the renaturing and gelling processes. FIG. 7.-The microwave dielectric spectrum of a 0.5 %gelatin solution at 5°C. Data for pure water at the same temperature are given by the dotted and dashed lines. The Time Domain Spectrum (TDS) of a 0.5 % wlv solution held at 5°C over a period of several hours showed no significant changes. The spectrum shown in fig. 7 is very close to that obtained 28 for pure liquid water and indicates a relaxation time, z, of 19 3 ps (compared to the liquid water value of 15 ps). This relaxation time demonstrates that no large scale rigid ordering of the bulk solvent occurs in the 0.5 % solution of a gelatin.Not all of the water in the system is appearing in the microwave spectrum ; some of the water is presumably hydrating the gelatin and undergoing a much more restricted motion. A quantitative estimate of the amount is difficult because of the low solute concentration but it is certainly less than 10 molecules of water per amino acid residue. When the TDS time scale was focused on the motion of the gelatin molecules, changes in the dielectric spectrum were apparent in both 0.5 % and 4.0 % gelatin concentrations as a function of quench time. The final state, long after gelation had occurred, was characterised by a dielectric spectrum showing a single relaxation process with a relaxation time, z, of approximately 350 ns ; the purity of the exponential decay of the spectrum indicates that the species is an essentially rigid spherical species.The amplitudes of the relaxation process at the two concentrations, shown in table 4, are not proportional to the gelatin concentration; this suggests that the gelatin species responsible for the relaxation is less favoured at the high gelatin concentration of 4.0 %.192 COLLAGEN FOLD AND GELATION OF GELATIN Before the final state was reached the spectrum also contained contributions from a longer time process (z > 2 ps) ; this component decreased in amplitude, eventually to insignificance, with increasing quench time. Although gelation of both the 0.5 and 4.0 % solutions occurred during the examination no discontinuities in the dielectric relaxation parameters which could be attributed to gelatin were observed.This was also found to be true for “spot ” frequency measurements of permittivity and conductivity in the range 100 kHz-1 MHz. TABLE 4.-THE DIELECTRIC RELAXATION PARAMETERS FOR a GELATIN GELS AT 5°C concn. A& 0.5 % 8.2 4.0 % 36.5 Tins 350 320 The rate of disappearance of the longer time scale component in a 5 % solution of the a gelatin was monitored by observing the change in permittivity at a frequency of 1 MHz and a temperature of 25°C. The process can apparently be divided into two consecutive first order processes ; the first, and much more rapid, process being substantially complete within the first thirty minutes or so and followed by a second much slower process. The similarity with the behaviour of optical rotation at a lower temperature and for lower gelatin concentrations is notable.DIFFERENTIAL SCANNING CALORIMETRY The data showed that on heating from -70°C to +7O”C only one large endo- thermic peak was observed at 0°C. There was no evidence of heat absorption due to melting of the gel; thus this result, obtained from a 25 % gel, confirms the con- clusion drawn from the use of the other techniques, that the gelling process does not give rise to detectable changes in the folding or flow of the gelatin molecule. Measurement of the heat required to produce the ice melting peak revealed that for the 25 % w/v a gelatin gel 14 % of the water present was not frozen, even at - 70” ; this gave a value expressed on a weight for weight basis of 0.426 x kg H,O/kg a gelatin as the hydration of the a gelatin.DISCUSSION The optical rotation studies suggest that, for all the concentrations of a gelatin studied, the process of reversion is first order in character and therefore an intra- molecular process, i.e., generation of the collagen fold occurs within one gelatin molecule and not by an intermolecular process involving links between gelatin chains. The data also show that for low a gelatin concentrations (<O. 1 % w/v) another species other than gelatin is involved in the folding process which, nevertheless, is still first order with respect to the gelatin. At higher a gelatin concentrations, the initial folding process is also dependent upon a species other than gelatin. The low concentration behaviour of a gelatin, at least for the slow process, can possibly be explained by considering the natural ionisation of the bulk solvent. The isoionic pH of the a gelatin solution is 4.5, which is independent of gelatin concentration ; this gives a hydrogen ion concentration of 3.2 x g ions dm-3.At a gelatin concentration of 0.1 % there are moles of a gelatin per cubic decimetre hence there are 3 solvated hydrogen ions per molecule of gelatin. However, at 0.01 % gelatin, there are some 30 solvated hydrogen ions per molecule of gelatin. There will also be an equivalent number of hydroxyl ions to affect the gelatin moleculeD. EAGLAND, G . PILLING AND R. G . WHEELER 193 and it may be this relative increase in ratio of ions to gelatin molecule with decreasing concentration below 0.1 % that is responsible for the decrease in the rate constant by suppressing the folding process.Viscosity studies of the low concentration gelatin solutions appear to confirm this interpretation; whereas the rate of the optical rotation data decreases with decreasing concentration, indicating greater difficulty in folding due to a more extended structure of the gelatin, the viscosity studies show that the rate constant for flow increases with decreasing concentration, a situation which could arise from the greater ease of flow of an extended flexible molecule. The initial rapid folding process observed at gelatin concentrations higher than 0.1 % involves only a small proportion of each gelatin molecule ; this is confirmed by the n.m.r. and dielectric relaxation spectra since the slow process is clearly associated with the disappearance of a random coil conformation of the pr~tein.~' The data of von Hippel et a1.,l9 on the effect of collagenase digestion upon the optical rotation of gelatin solutions suggests that the initial folding process involves the -gly-pro-hydroxypro- regions of the macromolecule.Collagenase digestion is specific for the sequence X-Pro, -R, -Gly, -Pro-Y and the required sequence where R is hydroxyproline is less than 10 % of the total number of residues in the protein. Thus the initial folding process of the gelatin chain occurs at the -gly, -pro, -hydroxypro, -pro- sequence along the chain and the remainder of the chain remains in a random coil configuration. Segal 29 has shown from optical rotation studies of the hexapolypeptides (gly-ala- pro-gly-pro-pro),, (gly-pro-ala-gly-pro-pro),, (gly-ala-pro-gly-pro-ala), and (gly-ala- ala-gly-pro-pro), that all four show a collagen type reversible dependence of the transition temperature between native and denatured conformations.Brown 30 has shown that the polytripeptide (pro-ala-gly), is also capable of producing a collagen type structure and Isemura 31 concludes from studies of the oligomers of tert-amyloxycarbonyl-L-proline that the tetramer is the smallest oligomer capable of producing a left hand helix, probably of one turn. Such evidence suggests that the initial folding process in a gelatin is the production of an approximately single turn of a left hand helix at the X-pro-hydroxypro-gly-pro-Y sequences along what is otherwise a random coil conformation of the gelatin molecule.The calorimetric studies of Privalov and Tiktopu10,~~ together with the O.R.D. data of McClain and Wiley 33 and the theoretical study of Cooper 34 all point to the structural stabilisation of the gly-pro-hydroxypro regions of collagen involving some type of water bridging between the pyrrolidine rings. A mechanism such as this would certainly be in agreement with the pseudo first order kinetic path observed here for the initial folding process of a gelatin. The much slower folding process subsequently occurring in the gelatin molecule is a true first order process; shown by the agreement upon the rate of the process, being approximately 1.5 x min-l, from the optical rotation, viscosity and n.m.r. data.The folding is thus intra-molecular and involves no other species than gelatin. The experimental rate constant was utilised to obtain values for the half life of the reversion process at each temperature and subsequently infinite-time values of the optical rotation. Plots of infinite-time values of the specific opticaI rotation as a function of concentration, extrapolated to infinite dilution at each temperature gave values for [alC=o of 300,250 and 225 deg g-1 cm3 dm-l at 5,lO and 15". Since these values are proportional to the final equilibrium concentrations of the folded form of the gelatin a van't Hoff plot of log [a]c, against 1 /T gave a value for AH" of folding 57--G1 94 COLLAGEN FOLD AND GELATION OF a GELATIN of - 18.8 kJ mol-I. Assuming a value of 4 kJ mol-l for the activation energy of the folding process then that of the unfolding process is of the order of 23 kJ mol-' ; this observation of an activation energy of unfolding being much greater than that of folding is similar to that found by Pohl 35 for globular proteins.By utilising the rate process approach of Eyring et aZ.,36 the rate constant k for the folding process can be used to obtain AG* the free energy of activation for folding RT Nh k = - exp (- AG*/RT) where Nis the Avogadro number and ir is Planck's constant. AG* was found to be 47.5 kJ mol-1 leading to a value of the entropy of activation for folding, AS*, of approximately - 160 J mo1-l K-l. The data strongly suggests that folding is dominated by the entropy changes involved in the process and gives little evidence for the formation of hydrogen bonds, in agreement with Privalov 32 and Cooper.34 The evidence obtained from O.R.studies is then that the folding is a very slow process but that the activation energy barrier to folding is low (-4 kJ mol-I). The n.m.r. studies of the same folding process show that the disappearance of the high resolution component of the a gelatin spectrum involves a line width transition- the lines become too broad to be observed (probably 2250 Hz in width). Since all the resonances undergo the transition at exactly the same rate it seems likely that all parts of the molecule are equally involved. The broadening is due to reduced freedom of motion. The transition is two-state * in character, since the resonances do not show gradual broadening; thus the conclusion must be that the transition involves the whole molecule becoming rigid-a similar conclusion arises from the dielectric studies of the gelatin relaxation.Since the low activation energy of folding does not form a major barrier to folding, the small rate constant must reflect the time which the molecule takes to produce a conformation suitable for it to undergo a rapid transition to a rigid structure. We can now visualise that in a process somewhat analogous to that proposed by Harrington and Rao 37 for very dilute gelatin solutions this rate determining step random coil int ram0 lecu tar triple helix region n uc lea t ed random coil nu cleated triple helix .-- : ;-• fast 8 FIG. 8.-A diagrammatic picture of the nucleation and folding sequences in a portion of the o( gelatin chain.D.EAGLAND, G . PILLING A N D R. G . WHEELER 195 requires the formation of a small region of triple helix by the gelatin chain folding back upon itself twice, bringing the single turns of left-handed helix into conjunction with one another. After this step has been achieved the remainder of the molecule contained within the loops can be expected to “ zip up ” with great rapidity, apparently by a mechanism which does not require the formation of appreciable numbers of hydrogen bond^.^^-^^* 38 A diagrammatic picture of the sequence is given in fig. 8 ; it should be borne in mind that because of the total number of amino acid residues in the molecule and the percentage of appropriately folded left handed turns (- 10 %) several such “zipped up” regions can be expected to occur along each gelatin molecule.This would inevitably leave a proportion of the molecule unavailable for triple helix formation which would explain why the initial optical rotation of the collagen is never totally recovered on renaturation, and the small residual spectrum observed in the N.M.R. studies even after several days. 7.0 6G 5.0 x 3 4.0 - M 0 -.I 1 3.c 2 .c I +l---t--t---+--i 0.1 0.2 0.3 0.4 0.5 conc % FIG. 9.-A plot of c/log vrel against concentration for a gelatin solutions at 15°C. Leaving aside for the moment discussion of the viscosity data concerning the highest gelatin concentrations, that pertaining to concentrations greater than 0.1 % indicates that the flow process reflects the influence of the formation of the nucleated triple helix and the subsequent change in the size of the flowing unit.The size of the flowing unit at infinite dilution of the gelatin solution and its variation with time can be determined by use of the Valid equation 39 2.54 In qrel = -~ derived froiii siuciie; of high volume fraction dispersions of spherical particles, where 1 - k 4196 COLLAGEN FOLD AND GELATION OF a GELATIN 9 is the determined volume fraction, 2.5 is the Einstein coefficient and k is known as the "self-crowding" factor. Making the transposition 4 = cV;, where c is the concentration and V," is the effective flowing volume of the gelatin molecule at infinite dilution, a plot of c/log qrel versus concentration is linear, fig. 9 is one such typical graph.The values of V," obtained from this treatment are given in table 5 as a function of time ; a plot of log (dV:/dt) against time results in essentially the same rate constant as that determined from O.R. and n.m.r. studies. TABLE 5.-THE EFFECTIVE FLOWING VOLUME OF THE Cf, GELATIN MOLECULE AS A FUNCTION OF QUENCH TLME AND TEMPERATURE quench time Ve/l. mol-1 /min 5" 10" 1 5" 0 145 1 1189 1316 10 1474 1474 1345 20 1630 1675 1365 50 2258 1842 1396 100 2559 2047 1417 150 2852 2142 1498 200 3040 1675 1548 The variation of the effective flowing volume of the gelatin molecule as a function of concentration can be obtained from the expression originally used by Breslau and Miller 40 for studies of solutions of inorganic salts and extended by Eagland and Pilling 41 to solutions of the symmetrical tetraalkyl amnionium halides -2.5~ + C(2.5~)~ -4(10.05c2)(1 -qteJ]' ve = 2( 10.05)2 Fig.10 shows the variation of the equilibrium value of the effective flowing volume as a function of concentration and temperature; fig. 10 also shows the excluded m 53 1 2 3 4 5 6 7 8 9101112 conc/mol I.-' x lo5 FIG. 10.-The equilibrium value of the effective flowing volume of the ct gelatin molecule as a function of concentration and temperature.D. EAGLAND, G. PILLING AND R. G. WHEELER 197 volume as a function of concentration in the solution arising from the point of first contact of spheres. Several points of interest arise from this figure, first a maximum in the flowing volume appears at all three temperatures in the region of 0.1 % gelatin concentration which seems to confirm the model proposed for these lower concentra- tion solutions. Secondly, the point ofintersection of the flowing volume curve with that of excluded volume should be the minimum concentration at which gelling will occur ; the experimentally determined gelling points at each temperature give confirmation of this proposal.The highest gelatin concentrations studied are in excess of that which corresponds to contact between spheres, and a different flowing process might be expected to occur ; this is confirmed by the experimental data, flow is no longer first order in character. It must be pointed out however that optical rotation in these solutions is still a first order process; helix formation is thus still intramolecular although contact occurs between the gelatin molecules in the flowing process.TABLE 6.-THE VISCOSITY AVERAGE MOLECULAR WEIGHT OF a GELATIN AS A FUNCTION OF QUENCH TIME AND TEMPERATURE quench time /min 5" mol. wt x 10-5 1 oo 1S0 0 0.97 1.02 0.98 10 1.10 1.15 1 .oo 20 1.28 1.28 1.04 50 1.41 1.41 1.10 100 1.68 1.52 1.13 150 1.82 1.63 1.15 200 2.09 1.74 1.17 The viscosity data can alternatively be examined using the model of Pouradier and Venet lo to determine the apparent change in the molecular weight of the gelatin with time. where [q] is the intrinsic viscosity of the solution, N is the molecular weight of the species ; a and x are constants characteristic of the solute and solute/solvent inter- actions. Taking a = 1.66 x and x = 0.885,37 the viscosity number average molecular weight was determined at 5, 10 and 15" as a function of quench time (table 6).All the data show, within experimental error, an initial particle weight of lo5 which increased with time; in addition the rate of increase of molecular weight expressed as a plot of log (dM,,/dt) versus time gave a rate constant of 1.7 x min-l, in agreement with the value obtained by other methods. Since aggregation of gelatin molecules is ruled out, it is a reasonable assumption that the change of particle weight is due to association of water with the gelatin. The number of water molecules associated with each gelatin chain afeer 200min, if this correlation is accepted, are 5.25 x lo3 at 5", 4.1 x lo3 at 10" and 1.5 x lo3 at 15". Expressing these results as kg H,O/kg gelatin they become 0.95 x 0.74 x and 0.27 x respectively which are in reasonable agreement with the value of 0.426 and obtained by us from d.s.c.measurements, and with that determined by Privalov 42 and Kuntz 39 ; this is particularly so when one remembers that techniques such as d.s.c. and n.m.r. only " see " the most firmly bound water molecules and flow is much more sensitive to less firmly bound water. If the number of moles of " bound '' water per mole of gelatin are combined with the mrresponding increase in flowing volume an " apparent molar volume " for the associated water is obtained. Fig. 11 shows the change in the apparent molar volume198 COLLAGEN FOLD AND GELATION OF a GELATIN after a quench time of 200 min plotted as a function of temperature ; too much must not be read into the extrapolation but it is interesting to note that the molar volume of the associated water reduces to its expected value at the denaturation temperature of the gelatin.It seems unlikely that the water itself can produce so open a structure I0 20 310 40 i o 60 temp/"C FIG. 11 .-The " apparent molar volume " of water associated with the geIatin molecule as a function of temperature. that such large values of molar volume can be sustained, but it may be that the combination of gelatin and its water of hydration produces a large exclusion volume of these dimensions which is lost on denaturation. It is clear that the minimum concentration of a gelatin required for gelling can be predicted from the flow behaviour of the more dilute solutions.The flowing species must be reasonably spherical ; this is confirmed by dielectric relaxation, light scattering and viscosity studies. Gelling occurs at the point of first contact between the gelatin spheres. The size of the sphere is, apparently, largely a measure of the hydration associated with the gelatin molecule ; the d.s.c. results suggest that, on average, each amino acid residue of the gelatin chain is associated with 2.4 mol. of water, in good agreement with the data of K ~ n t z . ~ ~ A picture, very probably oversimplified, of this situation is that of a sphere of approximately 45 nm diameter surrounded by a shell of water molecules, between 2 and 3 layers in depth, of -1-2nm thickness; the dielectric relaxation studies show that a more extensive rigid hydration layer cannot be sustained.Vold44 has shown that for small colloidal particles of this diameter a water shell of this thickness would satisfy the van der Waals-London dispersion attractive force of each particle with the result that the particles exert no attractive force on each other. However, as the gelling concentration is approached, first contact between the spheres will be through the hydration layers and Vincent 45 has shown that, when the structured hydration regions around colloidal particles are physically forced into contact, the structured medium between the particles becomes a means of transmitting attractive forces between the particles. It is possible, therefore, that gelling at these low concentrations is due to the predominance of van der Waals-London attractive forces over electrostatic repulsive forces, perhaps by the creation of a secondary minimum in the total potential energy interaction curve between the particles.This is not to say that such a mechanism can be expected to dominate at concentrations much in excess of the minimum required for gelling,D . EAGLAND, G . PILLING AND R . G . WHEELER 199 i.e. up to 5 %. Our viscosity studies show that in such cases flow behaviour is completely different to that used by us to predict gelling of the lower concentration solutions and gelling may involve tangling of the gelatin chains ; this however appears to be simply physical entanglement for O.R. and dielectric relaxation studies suggest that helix formation is still first order in these solutions.How far increasing the a gelatin concentration may be taken before the entanglement interferes with the first order folding process remains an open question at the present time. Grateful acknowledgments are due to Dr. E. G. Finer, for n.m.r. data, Dr. M. C. Phillips for d.s.c. measurements and Dr. A. Suggett for dielectric relaxation measure- ments. Particular thanks are also due to Dr. A. Suggett and Dr. F. Franks for stimulating discussions on the mechanism of the collagen fold. G. N. Ramachandran and G. Kartha, Nature, 1954,174,269. A. Rich and F. C. Crick, J. Mol. Biol., 1961,3,483. A. Veis, Macromolecular Chemistry of Gelatin (Academic Press, New York, 1964) p. 349. J. E. Eldridge and J. D. Ferry, J. Phys. Chem., 1954,58,992.H. Boedtker and P. Doty, J. Phys. Chem., 1954,58,968. P. H. von Hippel and T. Schleich, Structure and Stability of Biological Macromolecules, Eds., S . N. Timasheff and G. D. Fasman (M. Dekker, New York, 1969), p. 417. ’ D. Eagland, Comprehensive Treatise on Water, Ed. F. Franks (Plenum Press, New York), Vol. 4 to be published in 1974. J. F. Brandts, Structure and Stability of Biological Macromolecules, Eds., S . N. Timasheff and G. D. Fasman (M. Dekker, New York, 1969), p. 213. J. Pouradier and A. M. Venet, J. Chim. Phys., 1950,47, 391. lo J. Pouradier and A. M. Venet, J. Chim. Phys., 1950,47,11. l1 F. Franks, Chem. Ind., 1961, 204. l2 B. P. Jordan and E. H. Grant, J. Phys. E, 1970,3,764. l3 A. Suggett, Dielectric and Related Molecular Processes, Ed. M. Davies (Chem. SOC. London, l4 A. Suggett, High Frequency Dielectric Measurement, Ed. J. Chamberlain and G. W. Chantry l5 H. W. Loeb, G. M. Young, P. A. Quickenden, A. Suggett, Ber. Bunsenges.phys. Chem., 1971, l6 H. Fellner-Feldegg, J. Phys. Chem., 1972, 76, 2116. l7 A. Suggett and P. A. Quickenden, to be published. l8 B. D. Labrooke and D. Chapman, Chern. Phys. Lipids, 1969,3,304. l9 P. H. von Hippel and W. F. Harrington, Biochim. Biophys. Acta, 1959, 36,427. 2o S. Glasstone, Text Book of Physical Chemistry (Macmillan, London 2nd Edn, 1955), p. 1051. 21 B. H. Zimm, J. Chem. Phys., 1948,16,1093. 22 P. J. Flory and R. R. Garrett, J. Amer. Chem. SOC., 1958, 80,4836. 23 P. J. Flory and E. S. Weaver, J. Amer. Chem. SOC., 1960,82,4518. 24 P. L. Privalov, I. N. Serbyuk and E. I. Tiktopulo, Biopolymers, 1971,10, 1777. 2 5 F. Zernicke and J. Prins, 2. Phys., 1927,41,184. 26 A. Weissberger and B. Rossiter, Physical Methods of Chemistry (Wiley, New York, 1972), 27 P. I. Rose, Science 1971,171, 573. 28 A. Suggett and P. A. Quickenden, unpublished results. 29 D. M. Segal, J. Mol. Biol., 1969,43,497. 30 F. R. Brown, III, A. J. Hopfinger and E. R. Blout, J. Mol. Biol., 1972,63,85. 31 T . Isemura, H. Okabayashi and S. Sakakibara, Biopolymers, 1968, 6, 307. 32 P. L. Privalov arid E. I. Tiktopulo, Biopolymers, 1970,9, 127. 33 P. E. McClain and E. R. Wiley, J. Biol. Chem., 1972,247,692. 34 A. Cooper, J. Mol. Biol., 1971, 55, 123. 35 F. M. Pohl, Europ. J. Biochem., 1968,4, 373. 36 S. Glasstone, K. J. Laidler and H. Eyring, The Theory of Rate Processes (McGraw-Hill, 37 W. F. Harrington and N. V. Rao, Biochemistry, 1970,9, 3714. 1972), p. 100. (I.P.C. Press London, 1973), p. 96. 75, 1155. Chap. 2. New York, 1941).200 COLLAGEN FOLD AND GELATION OF a GELATIN 38 K. A. Piez and M. R. Sherman, Biochemistry, 1970,9, 4134. jg J. Vand, J. Phys. Chem., 1948,52,277. 40 B. R. Breslau and I. F. Miller, J. Phys. Chzm., 1970,74, 1056. 41 D. EagIand and G. Pilling, J. Phys. Chem., 1972,76,1902. 43 P. L. Pnvalov and G. M. Mrevlishvili, Biojizika, 1967, 12, 22. 43 I. D. Kuntz, Jr., J. Amer. Chem. SOC., 1971, 93, 514. 4s B. Vincent, J. Colloid Inter$ Sci., 1973, 42, 270. M. J. Vold, J. Colloid Sci., 1961, 16, 1.

ISSN:0301-7249    

DOI:10.1039/DC9745700181

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

GENERAL DISCUSSION Prof. M. Gordon (University of Essex) (partially communicated) : Tombs’s model proposal adds one sphere at a time to a growing aggregate, and focuses attention on the asymmetry characteristics, packing properties, etc., of the random particles produced. Thus the model seems to be aimed at the particulate type of gel, which Flory called type IV in his Introduction. I suggest that since the spheres of Tombs’s model are polyfunctional, e.g., tetrafunctional in fig. 2, it is more realistic to allow aggregation of aggregates to take place, cf. the linking between polymer molecules in polyfunctional condensation reactions. This automatically leads to a critical (gel) point, and would re-classify globulin gels with type I1 or 111. The Tombs model can hardly be justified in terms of precise information conveyed to a site on an aggregate, restricting the sticking mechanism to the acceptance of a single sphere at a time.On the other hand, sites situated in a concavity of a large aggregate are admittedly screened from accepting a link to another large aggregate, so the mechanism would not be a completely random aggregation. But these screen- ing effects in a neighbourhood of a site are covered by the extension of random poly- condensation mechanisms to first, second, . . . nth shell substitution effects. Dr. M. P. Tombs (Unilever) said: Inspection of electron micrographs of protein gels suggests that the pieces of the gel strand between junction points consists of many peptide chains aggregated together. My primary concern was to see how aggregates of suitable shape to form such pieces of strands could arise.Certainly, gel formation will require the aggregation of the aggregates, and there is no reason to suppose that this is not partly contemporaneous with the primary aggregation to form the strands. If, however, the initial stages of aggregation do not provide suitable shapes, then aggregation of the aggregates will lead not to gelation but to coagulation and phase separation. Aggregation alone is not enough, it must be the right kind of aggregation. The collision frequency efficiency factor for protein aggregations suggests that about one collision in lo8 leads to adhesion (cf. ref. (1) of my paper) and for the initial stages of aggregation addition of single spheres seems to be reasonable. Our model, which is not, and is not claimed to be, particularly realistic is concerned with early stages of the aggregation process to produce the pieces of strand which electron micrographs reveal.Gordon does, however, raise a more general point, which is whether the theory of polyfunctional condensation as applied to simple polymers could be applied to this case. In principle it should be possible, though I find it difficult to see what the significance of a critical gel point would be in this kind of system. More significant would be the critical conditions which determine gelation as opposed to coagulation. This must involve such factors as the degree of orientation of the interactions, which for the most part we do not know, and the detailed kinetics of, probably, different kinds of aggregations proceeding at different rates, and with different temperature dependences.It seems to me that any satisfactory account, which is likely to be no more than qualitative at the moment, must explain not only gelation but also coagula- tion. Further refinement of our simple model would produce something like the existing theory of poly functional condensation but this hardly seems worthwhile until we can specify the nature of the aggregation more clearly. 201202 GENERAL DlSCUSSION Dr. A. J. Hyde (University of Strathclyde) (partly communicated) : It is of interest, in connection with Tombs' paper and especially his last slide, to note that gels of rather similar structure may be obtained from a completely different and non-globular protein-silk fibroin, the archetype of fibrous proteins.The fibroin is dissolved at about 2 % concentration in one of a range of unpleasant solvents in which it almost certainly has a random coil structure (probably expanded by protonation or other charging of the coil). The system is then dialysed against water and one finishes with an approximately 1 % unstable aqueous solution of fibroin. As the salts or acid dialyse out of the system, the solvent quality changes from being poor to being almost a non solvent. After some weeks this forms a weak thixotropic gel. The constituent units of the gel are rodlike and have a most frequent length between 150 and 200 nm and a diameter of ca. 12 nm.' More rapid production of instability by adding alcohol or ammonium sulphate leads to shorter, malformed rods.The diameter of the rods is slightly greater than the value for the diameter of a fibroin molecule (MW 3 x lo5) in a compact globular form. The length of the rods bears no obvious relation to the extended length of the fibroin molecule (ca. 1300 nm) but is comparable with the length available for a linear aggregation process before the growing rod meets another rod in the system. It would thus seem that the rods are formed by linear aggregation of almost collapsed fibroin coils rather than side to side aggregation of extended chains. The length of time take to form the gel would also indicate a slow aggregation process and it is notable that Tombs' fibrillar structures were formed by heating his protein solution at the lowish temperature of 70°C, whereas the rougher treatment of 100°C gave more globular aggregates. It is also of interest to note that similar effects were discussed by Joly in the 1953 Faraday Society Protein Discussion.Dr. M. P. Tombs (Unilever) said: The papers by Barbu and Joly and Joly were important in showing clearly that aggregation as a consequence of denaturation was responsible for many of the phenomena, such as an increase in viscosity, exhibited by solutions of denaturing proteins. They relied almost entirely on flow birefringence to establish the presence of aggregates and to some extent their shape. The slide referred to by Hyde which illustrates the aggregation of albumin has been more fully reported in ref. (1) of my paper and in this case the interchain links were probably disulphides and therefore oriented.It is possible to explain the different behaviour at different temperatures in terms of the different accessibility of the intra-chain disulphides, though the mechanism suggested by Hyde may also play a part. The case of silk fibroin mentioned by Hyde is an interesting one because our somewhat limited experience suggests that for globular proteins at least, highly oriented aggregations of this kind are uncommon. It may well be that further investigations will reveal the conditions required to bring about and control the extent of oriented interactions, (or to use an equivalent way of describing it, the degree of randomness). Our analysis suggests that gel formation can occur as a result of nearly random aggregation, but the efficiency of gel formation measured by some such parameter as the minimum concentration requirement will depend on how random the orientation is, and this is not easily accessible to quantitative analysis at present.There are a number of attempts to predict the effect of random aggregation of spherical particles but while we may be reasonably sure that protein aggregations are not entirely random we do not know how non-random they are. It appears that F. Happy, A. J. Hyde and B. Manogue, Biopolymers, 1967,5,749. Barbu and Joly, Disc. Faraday Soc., 1953, 13, 77. Joly, Disc. Furday Soc., 1958, 25, 150. H. A. van Eekal, J. Catalysis, 1973,29,75.GENERAL DISCUSSION 203 6500- 6000 5500 f E 7 5000- 7 4500 400d: they do not have to be very non-random to produce gels, but to go any further we shall have to know much more about the mutual interactions of partly disrupted protein molecules.- - - 0 I I I I 0 20 40 60 80 11 Prof. C. A. Smolders (Enschede) said: Chain formation during aggregation of particles is not unusual in the literature. Sutherland,' using various statistical models, gave probabilities for chain formation ; Turkevich gave experimental evidence for the formation of chains of colloidal gold particles on slow coagulation, and Thomas and MacCorkle proposed a theory of oriented flocculation using the DLVO theory for colloid stability. Our own experimental work shows that con- centrated colloidal solutions of amorphous iron hydroxide particles (0.6 M in iron, particle diameter 40 A) give gels in a reproducible way if coagulation proceeds slowly (e.g., by slow titration or dialysis of the acid-stabilized sol). Electron microscopy shows chains of primary particles to be present to a large extent.204 GENERAL DISCUSSION mixtures above 77 % (v/v), a reverse change in [m’]234 takes place, the kinetics of which are initially second order, becoming more complex.The change is accompan- ied by a rise in turbidity at 320 nm whose kinetics parallel those of the change in optical rotation. The latter has been shown not to be a consequence of the turbidity. At higher protein concentrations still, c 2 8 g l.--I, a firm gel is formed on standing overnight. This gel is soluble in urea, melts on raising the temperature and exhibits considerable syneresis over a period of time.Infra-red spectra of films cast from the more dilute turbid solution and from the gel both show identical spectra indicating a large increase in /%structure in the aggre- gated protein. The change in [m’]234 is in keeping with a loss in cr-helix and an in- crease in #?-structure and the kinetics are taken to indicate that this conformational change is coincident with intermolecular aggregation. It is suggested therefore that the creation of the initial fibre aggregates shown by Tombs to lead to a gel network is, in the case of lysozyme, stabilised by the formation of intermolecular hydrogen bonded, #?-pleated structures. Dr. D. Eagland (University of Bradford) said : Our work with the cx-gelatin derived from rat tail tendons has shown that, at gelatin concentrations well in excess of the minimum required to produce a gel, helix formation is still an intramolecular cess; there is no evidence for helix formation between chains.This suggests helical segments do not serve as crosslinks in the gel as suggested in this paper. 0 I..... 0 0.05 0.40 0.15 0.20 0.2s molarity FIG. 1.-The difference in density between an a-gelatin gel (1.3 % w/v gelatin) containing butylammonium bromide and the solution containing only salt, plotted as a function of salt con- centration. pro- that tetra- Density determinations upon solutions of a gelatin and tetrabutyl ammonium bromide, using a quartz tube when filled with sample and then with water gave the data shown in fig. 1. The density of the gdatin gel shows marked gelatin gel shows marked changes as a function of the electrolyte concentration of the gel, implying aGENERAL DISCUSSION 205 considerable effect upon the structure, and number of cross links, within the gel.This problem appears not to have been considered in the particular case presented in the paper of Gordon and his co-workers. Prof. M. Gordon (Essex University) said: In reply to Eagland, there is evidence for the existence of intermolecular helix regions in both collagen and gelatin gels. Single a-chains have been renaturated to form triple helix collagen molecules indisting- uishable under the electron microscope from native collagen molecules.’ Even single-stranded fragments of collagen, 36 amino acids long, have been renatured to a molecule whose molecular weight is thrice that of the original fragments, which has the circular dichroism of the original helical It has been pointed out that the protein concentration inside an average random- coil gelatin molecule would be expected to be about 0.2 % and that intra and inter- molecular cross links are equally probable at this concentration.As dilution favours the intra molecular cross-links, in the 5 % solution the interchain bonds will predom- inate. The model of isolated spheres causing gelation by contact forces is not real- istic at high protein concentrations. We are not excluding intramolecular chain folding as a side reaction. This is expected as reflecting the normal competition between intra and intermolecular “ bonds ”. In other types of system it has been treated with high precision, and ab~olutely.~ In the present case, the effect is compensated by the rescaling of the abscissa (to ala,).The situation is analogous to that, say, in the theory of corresponding states. There the accurate superposition of the physical properties of liquid/gas systems with force fields as diverse as those of argon and CH4, is achieved by plotting quantities as ratios relative to their values at a critical point. Within the rescaling theories of the critically branched state, one similarly starts without special parameters for pH or electrolyte content. The theory suggests it, and experiments on a surprising variety of systems so far supports the theory. For the triple helix model the relevant curve C gives practically a zero parameter fit (though more accurate experiments would be desirable).The same rescaling answers Eagland’s second point. Dr. W. M. Marrs (BFMIRA, Leatherhead) said: The absence of an endotherm on melting 25 % a-gelatin gels is an interesting result in view of the fact that relatively dilute (-5 %) commercial gelatin gels produce an endotherm on melting in the temperature region 30-45°C. The commercial gelatin gels, however, were matured for at least 16 h whereas the a-gelatin gels were only permitted 3-4 min in a temperature region in which the formation of gel structure proceeds at an appreciable rate. Does not Eagland think it possible, therefore, that under these conditions of setting, the amount of structured phase produced may be insufficient to make detection of the endotherm on melting a practical possibility? Dr.M. Pyrlik and Prof. G. Rehage (Clausthal) (partly communicated): We are interested in Eagland’s interpretation of the results of his DSC-measurements that the thermograms do not show any endothermic effect due to the melting of the gelatin gels. In most papers dealing with structures of gelatin gels crystallinity of the junc- tion points of the three-dimensional-network is assumed and often confirmed by K. Altgelt, A. J. Hodge and F. 0. Schmitt, Proc. Nat. Acad. Sci. U.S.A., 1961, 47, 1914. K. A. Piez and M. R. Sherman, Biochemistry, 1970, 9,4129. M. P. Drake and A. Veis, Biochemistry, 1964, 3, 135. see e.g., W. B. Temple, Makromol. Chem., 1972, 160, 277.206 GENERAL DISCUSSION X-ray techniques (e.g., ref.(I), (2)). Accordingly, the appearance of a melting heat is to be expected. Also the helix-coil-transformation should contribute to this heat, because melting and helix-coil-transitions are assumed to occur in the same tempera- ture region. A possible explanation for the lack of an endothermic peak in your thermograms could be that at a cooling rate of 8"/min the time allowed for forming small crystallites is too short. For it can be assumed that below the freezing temp- erature of the water no considerable diffusion and therefore crystallization of the macromolecules can occur, the time allowed for the helix-coil transition and crystal- lization of gelatin molecules is only about 4-5 min. Probably only the formation of poorly ordered crystallites with a very broad size distribution is possible during such a short period.On subsequent heating in the D.S.C., the melting of these poor crystal- lites gives a very flat and broad endothermic maximum. 1 i a w b s 10 30 50 T/"C rate 16"/min for both thermograms. FIG. 1.-Thermograms of a 36 wt.-% aqueous gelatin gel with different thermal history. Heating Our own results give evidence for this behavio~r.~ Melting of a gel aged at 0°C for 24h gives thermogram a. After melting, this gel was cooled down to -20°C with 2"/min, and then heated again to give thermogram b, which shows only a very flat and broad maximum compared with the peak of curve a. This maximum can vanish almost completely due to another slope adjustment of the D.S.C. Appar- ently, even with the comparatively slow cooling rate of 2"/min a proper formation of the crystallites is not possible.This observation agrees with X-ray measurements, which show that good crystal reflections are only observable after ageing times of at least 24 h.4 J. R. Katz, S. C. Derlson and W. T. Bon, Rec. Truv. Chim., 1932, 51, 513. J. E. Jolley, Photogr. Sci. Eng., 1970, 14, 169. K. Bergmann, Diploinarbeit (TU Clausthal, 1974). ' K. Herrmann, 0. Gzrngross and W. Abits, Kolloicl-Z., 1932, 60, 276.GENERAL DISCUSSION 207 Pro€. W. Prim (Syracuse University) said: I wish to make two comments on the paper by Eagland et al., (1) intramolecular unfolding or refolding rates in proteins are normally of the order of microseconds or faster. In view of this it seems peculiar that your proposed intramolecular folding process is so slow.(2) The angular dependence of the light scattering of hot gelatin solutions, as well as that of cold gelatin gels indicates that there must be at least some aggregation. Would this necessitate changes in your treatment? Dr. D. Eagland (University of Bradford) said: 1. The rates to which Prins refers are applicable to the refolding of globular proteins. There is considerable experi- mental evidence that the growth stage of the collagen fold or triple helix observed by us is a much slower process. We would, however, agree withhis comment in regard to the nucleation step ; this appears to be exceedingly fast, we were only able to obtain approximate data for this step even with our most rapid technique of optical rotation.2. Aggregation in gelatin gels containing B or y gelatins is quite possible, we can only say that for this particular single strand a-gelatin at low concentrations, observa- tions by several techniques other than light scattering gave no indication of aggrega- tion. Dr. G. Stainsby (University of Leeds) said: If gelling, in solutions at the lower concentrations, is essentially due to the predominance of van der Waals-London attractive forces over electrostatic repulsive forces, then the ionic composition of the environment surrounding the gelatin particles is a most important factor. Isoionic gelatin, in good quality distilled water, is the system most likely to exhibit this mode of gelation. It has for long been known that, for commercial gelatins, dilute isoionic gels differ quite markedly from gels made at the other pH values, and from gels made in the presence of salts.The main differences are : isoionic gels are very weak, they are opaque rather than transparent (and hence contain large aggregates) and they readily exhibit syneresis. These differences would be expected to persist in gels made using pure a-gelatin. It seems unlikely that the DLVO-type mechanism of gelation, suggested in the paper by Eagland et al., would apply to all type of gelatin gel, but only to dilute isoionic gels. Do the authors agree with this view? Secondly, have the authors evidence to show that the a-gelatin was isoionic at pH 4.5? This is a very low pH value indeed for isoionic a-gelatin. Pure collagen from most mammalian tissues is generally thought to be isoionic when the pH value is about 9, as swelling is then at a minimum.The extracted gelatin, having the same amino acid composition as the collagen, would also be expected to be isoionic at pH 9, when the use of acids, alkalis and buffer salts is avoided in the extraction procedure, as in the present experiments. It seems to me that the a-gelatin used in this investigation is most probably not isoionic but is at a pH value not far removed from the " natural " pH of the collagen in the tail tendon. This is normally not quite so acidic, but the pH usually falls just before and also after death. Since a-gelatin is a highly flexible polyelectrolyte a knowledge of the net charge per molecule is of importance when considering, for example, the relationship of dilute solution viscosity to molecular weight.The constants for this relationship, used in the paper, were deduced by Pouradier and Venet for isoionic gelatin which had been extracted from collagen after a prolonged alkaline treatment which released additional acidic groups from asparagine and glutamine. When these additional groups are not released, and the gelatin has the same charge profile as a-gelatin, both constants are very different, k for example increasing from 1.66 x to 1.10 x for isoionic208 GENERAL DISCUSSION solutions. It seems unlikely, however, that either set of constants is appropriate for use with a-gelatin at pH 4.5. Dr. D. Eagland (University of Bradfordj(part1y communicated) : In reply to Stainsby, the conditions for gelling in which van der WaaIs and electrostatic forces can be expected to predominate should be a salt free, isoionic solution of the gelatin; this is precisely why these conditions were chosen for the investigation.The authors therefore are in full agreement with the view that this is the system most likely to exhibit this mode of gelation. Several differences are apparent between the dilute isoionic gels formed by the x-gelatin derived from soluble rat tail tendon collagen and those derived from commercial gelatins as exemplified by Stainsby ; for example, the rat tail a-gelatin gel is a clear water white gel, exhibiting no syneresis. In addition a 0.3 % w/v solution of rat tail tendon a-gelatin produces a quite rigid gel when cooled to 5"C, whereas a 1.0 % w/v solution of a-gelatin derived from an acid precursor commercial gelatin (pIq) remains a viscous solution even after severaI days of standing at this temperature.We would agree that although a DLVO mechanism appears predominant in the system investigated it would be unwise to attempt the extrapolation to every type of gelatin gel. This, however, has not been our intention, although clearly a DLVO mechanism must have some role to play in the gelling process; we have shown that gelling by a dilute isoionic solution of a rat tail tendon a-gelatin can be predicted from the behaviour of more dilute solutions of the protein which are not, in themselves, capable of gelation. The problem of defining the isoionic point of a gelatin or collagen often gives rise to difficulties ; the isoionic point of this material was determined, after exhaustive dialysis in accord with the proposals of Alberty,l by a titration technique.The value obtained is distinctly different to the value obtained for an acid precursor gelatin (usually in the range 7.0 to 9.0), but close to the value (5.5) obtained by Stainsby, and reported in a previous Faraday Discussion,2 for a gelatin obtained by alcohol coacervation of collagen extracted from bone by steam treatment. that a reliable location of the isoionic point is difficult to determine in a heterogeneous system consisting of insoluble collagen fibres and water ; it is, therefore, particularly difficult to equate a value determined for such a system with that obtained by quite vigorous treatment of the parent insoluble collagen with acid in order to produce a soluble denatured material.The a-gelatin used in this investigation was derived from a soluble collagen obtained from the tail tendons of very young immature rats; such tendons on swelling in cold salt-free distilled water yield a soluble collagen without further treatment. It is perhaps significant that the gelatin of similar isoionic point obtained by Stainsby was derived, albeit by a much more vigorous, but nevertheless neutral and salt free procedure. It appears that an isoionic u-gelatin derived by a neutral procedure is very different from a nominally similar fraction obtained from isoionic acid precursor material. In addition, Mazurov and Orekhovich and Veis and Cohen have shown that it is not possible to regenerate collagen type fibrils from acid extracted gelatins whereas Jackson and Fessler have reported that extraction with cold salt solutions at physio- R. A. Alberty in The Proteins, ed. H. Neurath and K. Bailey, (Academic Press, New York, 1953), Vol. 1, Part A, p. 478. G. Stainsby, Disc. Faraday SOC., 1954, 18, 288. A. Veis, in Macromokcular Chemistry of Gelatin (Academic Press, New York, 1964), p. 7. V. 1. Mazurov and V. N. Orekhovich, Biokhymia, 1959,24,28. A. Veis and J. Cohen, J, Phys. Chem., 1958,62,459. D. S . Jackson and J. F. Fessler, Nature, 1955, 176, 169. It has been pointed out by VeisGENERAL DISCUSSION 209 logical ionic strengths produces a material capable of regenerating a collagen type fibril. It would appear, therefore, unsafe to attempt the explanation of all gelatin behaviour on the basis of acid precursor material, as it would be equally unsafe to extrapolate the behaviour found in this system as a general explanation for other gelatin gels. Finally we would point out that the molecular weight of the material has been determined by vapour pressure osmometry, in addition to light scattering and visco- metry techniques, the results agreeing within experimental error ; it appears that any variations in the Pouradier and Venet constants for this system, which might occur, are not severe enough to seriously affect the accuracy of the result.

ISSN:0301-7249    

DOI:10.1039/DC9745700201

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Formation of Mucous Gel from the High Molecular Weight Mucoprotein of Gastric Mucus BY ADRIAN ALLEN, ROGER H. PAIN" AND DAVID SNARY Dept. of Biochemistry, Ridley Building, University of Newcastle upon Tyne, Newcastle upon Tyne, NE1 7RU Received 2nd January, 1974 The highly viscous and gelatinous nature of gastric mucus is shown to depend upon the properties of a mucoprotein of protein content 13.6 % and molecular weight 2x lo6. At 25"C, the viscosity of this mucoprotein is strongly concentration dependent and at concentrations about 4 g I.-' rises asymptotically, indicating the onset of a reversible sol-gel transition. Reducing agents (mercapto- ethanol) and trypsin split the mucoprotein into four and two sub-units of molecular weight 5.2 x lo5 and 9.8 x lo5 respectively. These reductions in molecular size are accompanied by loss of the viscous and gelatinous properties.The viscosity of the mucoprotein undergoes two sharp changes in KCI, one between 0 and 0.05 M KCI which is attributed to a polyelectrolyte charge shielding phenomenon, The second transition near 1 .O M KCI is interpreted as a marked increase in asymmetry accompanied by contraction and reduced solvation. Increasing concentrations of CsCl reduce the viscosity of the mucoprotein, bringing about a transition which is essentially complete by 0.1 M CsCl and is as great in extent as both transitions in KCI. In CsCl the highly expanded mucoprotein is contracted to a molecule having the same symmetry but an even smaller volume and lower solvation than in high KCI concentrations. Both salt-induced transitions are non-reversible.A mechanism is dis- cussed in terms of the effect of the cs+, K+ and C1- ions on water structure and water-mucoprotein interactions. Guanidinium chloride causes partial expansion of the CsCl treated material with a limited aggregation of four mucoprotein molecules on its removal. The viscosity at finite con- centration increases markedly and reversibly with temperature and at 45°C the threshold concentra- tion for reversible gel formation is reduced to 2 g I.-'. There is no evidence for intramolecular changes. The temperature dependence is characteristic of intermolecular entropy driven interactions and the results are interpreted to show that the mechanism of gel formation in gastric mucus involves micelle junctions. Mucus is a slimy, viscous secretion of epithelial tissues which in mammals lines the surface of the gastro-intestinal, respiratory and urinary tracts.In the stomach, a layer of mucous gel protects the underIying mucosa from ulceration by the acid and proteolytic enzyme contents as well as lubricating the passage of food.2* The essential components of the mucous secretions are macromolecular mucoproteins consisting of a large proportion of polysaccharide chains covalently bound to a rela- tively small proportion of p r ~ t e i n . ~ These mucoproteins are produced in vivo by the mucosal cells as large molecular weight molecules which then interact to form the mucous gel. This gel possesses considerable stability under the conditions prevailing in the stomach.In order to understand the mechanism of formation of this gel and its stability it is necessary first to investigate the conformation of the mucoprotein molecule and then to study its interactions in a well characterized system before finally studying the gel itself. This paper reports some results obtained on the first two of these stages of investigation. We have isolated, by water extraction of pig gastric mucus, a mucoprotein which accounts for the high viscosity and gel forming properties of the parent mucus.5 This mucoprotein is isolated without enzymic digestion and has a chemical analysis 210A . ALLEN, R. H. PAIN AND D. SNARY 21 1 shown in table 1. Sedimentation analysis shows the mucoprotein as a single poly- disperse peak of s&w = 18.7s and Ms,D = 2 x lo6.The carbohydrate side chains consist on average of about 15 monosaccharide units arranged in a specific sequence and attached by seryl and threonyl ether links to the protein core?' The portion of the protein core carrying the sugar chains is high in serine, threonine and proline (36 % of total amino acids) and the size of the carbohydrate chains is such that most of the protein must be completely buried by a sheath of polysaccharide. There is also a section of the protein core susceptible to attack by proteolytic enzymes which has an amino acid analysis more typical of an average globular proteing The characteristic polydispersity of these molecules is thought to be due to differences in the number and completeness of the polysaccharide chains between individual molecules. The single NH,-terminal aspartic acid found for this mucoprotein argues against the polydispersity being due to the protein component.M,,D 4.9 x 105 (1 sub-unit) mucopro tein 7 M,,D 5.2 x 105 (1 sub-unit) protein 13.6 % (4 sub-units) +- trypsin M,,D 9.6 x 105 protein 10.5 % (2 sub-units) c 0.2 M mercaptoethanol -+ I ' I 1 FIG. 1.-The effect of mercaptoethanol and trypsin on the purified mucoprotein. The water extractable mucoprotein (Ms,D = 2 x lo6) of gastric mucus is composed of four mucoprotein subunits each one quarter of the molecular weight of the parent molecule (Ms,D = 5 . 2 ~ lo5). These four subunits have the same N-terminal amino acid (aspartic acid) and are joined together covalently by disulphide bridges, as shown by the action of mercaptoethanol.O Trypsin splits the mucoprotein into two subunits (MSsD = 9.8 x lo5) which when treated with mercaptoethanol give subunits of similar molecular size to those obtained from the undigested molecule (fig. l).' EXPERIMENTAL PREPARATION OF MATERIAL Stomachs were collected in ice immediately after slaughter of the animals. The stomachs were opened from the cardio-oesophageal entrance along the lesser curvature to the pyloric sphincter. The contents of the stomach were removed and the stomach was washed212 GASTRIC MUCOUS GEL thoroughly with cold water. The mucosal surface of the cardiac section was scraped with a scalpel and the scrapings were placed in ice-cold 0.02 % (w/v) azide solution against which they were dialyzed (4°C for 3 days, 3 changeslday).The noo-diffusible, water-extractable mucoprotein was separated from the water-insoluble gel and mucosal tissue by centrifugation TABLE 1 .-CHEMICAL COMPOSITION OF THE WATER EXTRACTABLE MUCOPROTEIN (MOLECULAR WEIGHT 2 X lo6) FROM PIG GASTRIC MUCUS % by weight of freeze-dried material galactose fucose glucosamine galactosamine sialic acid protein phosphate sulphate 26.0 11.3 19.5 8.3 0.2 13.6 0.2 3.1 (5000 g x 10 min). The water-extractable mucoprotein was concentrated by vacuum dialysis. This preparation was used for preliminary experiments on the effect of mercapto- ethanol, trypsin, KCl and CsCl on the viscosity of the mucoprotein. The preparation is a crude aqueous extract of mucus, which contains smalIer mucoprotein fragments, but whose viscosity is characteristic of, and dependent almost entirely upon, the high molecular weight mucoprotein which is the subject of this paper.For characterization of its properties the high molecular weight mucoprotein was first purified from the remaining water extractable mucoprotein by gel filtration on Sepharose 4B.ll PHYSICAL MEASUREMENTS The viscosity of solutions in which shear dependence was absent, or uncorrected for, was determined in a capillary viscometer of the type described by Schachman l2 having a capacity of 2 ml and a flow time for water at 25°C of 220 s. Viscosities at different shear stress were measured in a zero shear capillary viscometer. The rate of approach of the menisci in the two arms of a capillary U-tube was measured, the difference in height of these menisci being proportional to the shear stress applied to the solution.Shear dependence of viscosity was indicated by non-linearity of a plot of logAh against time where Ah is the difference in height between menisci. Sedimentation coefficients were obtained and extrapolated to zero concentration.13 (Height, area) diffusion coefficients were measured as described.13 Zero time corrections, At, were always measured and the values of DoAt never exceeded 7x cm2. Values of 1/D were plotted against c0/2 to give D&w. Molecular weights were calculated from values of s&w and D& using the Svedberg equation. The value for the partial specific volume used in this equation was 0.64 ml g-l. Since the low solubility and scarcity of material made the experimental determination of the partial specific volume impracticable, a graph of protein content against partial specific volume for a number of mucoproteins was used to obtain a value of 3 for the pig-gastric fractions. l3 RESULTS Mucus, as obtained from the pig stomach, is a viscous, slimy material.The aqueous extract, when concentrated, has the visual properties characteristic of a gel although it is never homogeneous in consistency. The most reproducible, measurable para- meter by which this material is defined as a gel is the viscosity of the aqueous solution. If, as concentration is increased, the viscosity at zero shear rises asymptotically towards infinite viscosity, the formation of structure due to intermoIecular interaction isA . ALLEN, R.H. PAIN AND D . SNARY 213 indicated. The concentration at which such a rise in viscosity occurs is here termed the " gel threshold ". THE MUCOPROTEIN Aqueous extracts of gastric mucus were purified by gel filtration to give a muco- protein which, in 0.2 M KCl, possesses the sedimentation, diffusion and molecular weight properties shown in table 2. The viscous properties in this solvent are shown TABLE 2.-PHYSICAL PROPERTIES OF THE WATER EXTRACTABLE MUCOPROTEIN FROM PIG GASTRIC MUCUS solvent physical property 0.2 M KCl 2.5 M KCl 3.5 M CsCl -+ 0.2 M KCl s; 5 ,w Is 18.7 16.1 33 Ksld g-l 260 46 120 Ms,D 1.9x lo6 1.9x 106 2.2x lo6 o;5,wlm* s-1 x 107 0.69 0.56 0.94 [VlIml g-' 320 460 160 W V ) 0.81 0.1 0.75 flfo 4.45 4.66 3.27 All solvents contained : the appropriate concentration of salt, 0.02 % (w/v) Na azide ; 0.02 M K acetate brought to pH 5.5 with acetic acid.in fig. 2. At a mucoprotein concentration of about 4gl.-I, the reduced specific viscosity extrapolated to zero shear rises sharply and gel formation sets in. This behaviour is reversible. Fig. 3 illustrates the non-Newtonian shear dependence of viscosity, as shown by the deviation from linearity. Shear dependence is present down to the lowest measurable concentrations of mucoprotein. At concentrations n 1 M G 0 II 3 - 2- I - ? concentration of mucoprotein (1. gel) FIG. 2.-Variation of viscosity with concentration for water-extractable mucoprotein 0.2 M KCl(0) : 2.5 M KCl + 0.2 M KCl (0) : 3.5 M CsCl-+ 0.2 M KCI (A). All solutions measured in 0.2 hl KCl buffer, pH 5.5 ; all points refer to zero shear.214 GASTRIC MUCOUS GEL well below the gel threshold, the viscosity is still markedly concentration dependent, with a Huggins constant of 3 as the concentration approaches zero.The high value for the intrinsic viscosity and the large frictional ratio (see table 2) means that the molecule must be asymmetric or expanded or both. The value of the ratio K,l[q], where K, = l/s"(ds/dc), has been shown l4 to be an indicator of molecular asymmetry, independent of molecular size. Thus solvation and asymmetry can be separated by measurement of flow properties. Approximately spherical molecules take values in the range 1.5 to 1.7 and the magnitude of the intrinsic viscosity indicates whether the structure is compact [q] x 2.5, or expanded, [q] > 2.5.Values of K,/[q] less than 1.5 are indicative of an asymmetric conformation with a stiff and relatively free-draining structure. The experimental value for the mucoprotein is subject to 0.8 qr 0.6 Oa2 t 2 0 4 0 6 0 8 0 time/& FIG. 3.-The effect of CsCl on the shear dependence of viscosity for the water extractable muco- protein 0.2 M KCI (0). 3.5 M CsCl --+ 0.2 M KCI (A). Ah is the hydrostatic pressure difference in the variable shear viscometer and is proportional to the shear stress. Deviation from linearity of this plot represents non-Newtonian behaviour. considerable error owing to the shear dependence of viscosity and to the relatively narrow concentration range over which sedimentation velocity measurements can be made. The mean value for Ks/[q] is 1.2 k0.3, which indicates that the mucoprotein is only moderately asymmetric. The high intrinsic viscosity (320 ml g-l) shows, therefore, that the molecule is highly expanded and solvated.The " effective hydro- dynamic volume '' can be calculated from the relation 4nN Ve = {&} 3M assuming the molecule is spherical. f i s the frictional coefficient calculated from the diffusion coefficient, q0 is the viscosity of the solvent, Nis Avogadro's number and M the molecular weight. The mucoprotein has a value Ye = 56mlg-' compared with the partial specific volume of 0.64. Allowing for the assumptions involved, an idea is gained of the degree of expansion of the mucoprotein. It is concluded that the mucoprotein in 0.2 M KCl is a fairly symmetrical, highly expanded molecule. with the carbohydrate chains presumably extended into the solvent and hydrated.A .ALLEN, R . H . PAIN A N D D. SNARY 215 DEPENDENCE OF INTERMOLECULAR INTERACTION O N THE NATIVE CONFORMATION (a) INTEGRITY OF THE MOLECULAR STRUCTURE The mucoprotein has been shown to be held together by disulphide bonds.1° When these are broken by mild reduction with mercaptoethanol, in the absence of denaturing agents, four sub-units of similar molecular weight are produced. The viscosity is reduced considerably (by 75 % 16), and the mucoprotein no longer forms a gel. This explains at the molecular level the reduction in viscosity produced in bronchial mucus when reducing reagents such as N-acetylcysteine are used as drugs for the relief of bronchial congestion.15 Proteolytic enzymes such as trypsin bring about limited digestion of the mucoprotein with the production of large " sub-units ".g The viscosity is again reduced and the ability to form strong intermolecular inter- actions is lost.The viscosity of the water extractable mucoprotein was reduced by 72 % after incubation with trypsin for 24 h at 37°C. Thus the structural integrity of the native mucoprotein is essential for its biological function of forming a gel at low mucoprotein concentrations. It is not clear whether the loss of this ability is a result of conformational change resultant on chemical modification or of reduced size and, perhaps, " valency " of the sub-units. (b) INTEGRITY OF THE MUCOPROTEIN CONFORMATION During attempts to probe the nature of the intermolecular interactions responsible for gel formation, the effect of increasing the ionic strength was studied.13 When it was discovered that the primary effect of increasing KCl concentration was on the conformation of the mucoprotein, the influence of CsCl was also studied as part of a programme to investigate the effect of different ions.The dependence of viscosity on salt concentration is shown in fig. 4 for both KC1 and CsCl. At concentrations below 0.05 M KCl, the viscosity of the mucoprotein rises to a much higher value in 0.2 n I M r( 0. I 0.1 0.2" 1.0 2.0 3.0 4.0 molarity of salt, KCl or CsCl FIG. 4.-Viscosity of unfractionated water extractable mucoproteinas a functionof salt concentration. KCI (0) ; CsCl(0). Mucoprotein concentration, 3.1 g I.-'.216 GASTRIC MUCOUS GEL the absence of salt.In this region the viscosity is highly sensitive to minor traces of ions, in a manner reminiscent of the classical polyelectrolyte effect. The charged polymer is highly expanded in the absence of counterions due to charge repulsion and contracted when the repulsion is removed by charge shielding. The charged groups on the mucoprotein are aspartate, glutamate, (18 % of protein), lysine and arginine (5.9 % of protein) on the protein component and mainly sulphate on the carbohydrate, approximately 1 sulphate for every 8 carbohydrate residues. Since the viscosity was measured only at finite concentration, it is not possible to state whether the change in viscosity is due primarily to changes in molecular interaction or to a conformational change in the mucoprotein. The second transition occurs around 1 M KCl, where the viscosity drops from one plateau region to another (fig.4). The properties of the molecule at zero con- centration are listed for solutions on either side of this transition (table 2). The shear dependence of viscosity shown in 0.2 M KCl is considerably reduced in 2.5 M KCI but the intrinsic viscosity in the latter is slightly higher. Combination of the various flow property parameters shows that the molecular weight is unchanged but that the molecule has become much more asymmetric (low K,/[q]). The fact that the frictional ratio is only slightly higher in 2.5 M KC1 implies that the molecule must be contracted in order to account for the increased asymmetry contribution to the frictional ratio.This evidence suggests, therefore, that at high concentrations of KCl the mucoprotein contracts from an approximately spherical, expanded molecule to an ellipsoid which is less expanded and solvated. The lack of change in viscosity between 0.1 M and 0.5 M KCl indicates that the high salt transition is qualitatively different from the low salt transition. The greater extent of the transition induced by low concentrations of CsCl (fig. 4) shows, there- fore, that, while part of this transition may have the same basis as that occurring in low concentrations of KCl, the remainder must be qualitatively different and most probably similar in basis to that occurring at higher concentrations of KCl. The flow properties of the mucoprotein exposed to 3.5 M CsCl and then measured in 0.2 M KCl show that the inolecule is now little changed in symmetry from the native mucoprotein in 0.2 M KCl since the value of KJq] is not greatly reduced (table 2).The large increase in s;5,w and decrease in intrinsic viscosity show that the molecule must be contracted. The volume, V,, calculated from eqn (1) is reduced from 56 to 22 ml g-' after exposure to CsCI. The Huggins constant is reduced to 0.27 and there is no shear dependence of viscosity. The fact that the above flow properties were measured in 0.2 M KC1 after removing the CsCl underlines the fact that this transition is not reversible. The transition induced by high concentrations of KCl is also only partially reversible and the reasons in both cases for non-reversibility must be intra- rather than intermolecular. The ability to form a gel is eliminated after treatment with both salts and it is an attractive hypothesis to suggest that the groups normally associated with intermolecular inter- action are now involved in intramolecular interactions in the contracted molecule.The explanation for these effects of salts on the conformation must be sought in terms other than polyelectrolyte charge-shielding effects. Two possible hypotheses emerge, one based on changes in solvent environment and the other on specific ion- mucoprotein interactions. Certain of the alkali metal and halide ions are thought to have a structure-breaking effect on water and experiment shows that the effect is greater for Cs+ than for K+.17* l8 Frank and Evans l 9 have calculated the " struc- ture breaking entropy " contributed by the effect of ions on the water beyond the close, irrotationally bound water immediately adjacent to the ion.These vatues amount to +90 and + 120 J K-l mol-l for KCl and CsCl respectively. It has been proposedA . ALLEN, R . H . P A I N A N D D . SNARY 21 7 that the degree of solvation of a sugar residue depends upon the compatibility of its hydroxyls with the water structure and it has been shown that this solvation is reduced when the structure is disrupted thermally.20* 21 Thus, solvation of the sugar moiety of the mucoprotein may be reduced by the addition of ions with a consequent shrink- age of the molecule as intra-molecular inter-saccharide interactions become ener- getically favoured.The phenomenon is paralleled on an intermolecular basis by the action of sucrose, ethylene glycol or ammonium sulphate in reducing the ability of the water to interact with pectin, thereby favouring association between polysaccharide chains.22 The second type of mechanism, involving specific ion interactions, is suggested by the relatively low concentrations at which CsCl brings about the dramatic change in mucoprotein conformation. Alkali metal ions are known to be chelated by ether oxygen atoms such as exist in the glycosidic bond of the polysaccharide moiety. 30 This would require a critical steric assembly of chelating groups in what is currently thought to be a fairly flexible molecule. In both these mechanisms, emphasis has been placed on the polysaccharide com- ponent, which makes up 86 % of the mucoprotein. It must be left an open question as to whether the protein moiety can play a part in these conformational changes.The flow properties measured in guanidinium chloride (GuCl) are changed (see table 3) in a direction which suggests increased expansion. This is analogous to the known unfolding effect of GuCl on proteins and suggests that GuCl solubilizes groups involved in stabilizing the contracted structure. On removal of GuCI, however, the molecule aggregates in a fairly specific way, still sedimenting within a single boundary with a high degree of optical recovery. TABLE 3.-THE EFFECT OF GUANIDIMUM CHLORIDE ON THE PHYSICAL PROPERTIES OF THE WATER-EXTRACTABLE MUCOPROTEIN PREVIOUSLY TREATED WITH 3 .5 M CSCl mucoprotein (solvent) SOZS.WlS &/ml g-1 D;,,,/crnz s-1 x lo7 mol. wt. KCP 33 120 0.94 2 . 3 ~ lo6 GuCl 26 290 - 2 . 7 ~ lo6 after GuCl a 52 260 0.44 8 . 2 ~ lo6 aexamined in 0.18 M KCI, 0.02 M acetate pH 5.5; bexamined in 4 M GuCl, 0.018 M KCI, 0.002 M acetate, pH 5.5 ; C molecular weight from s;s,w and D&w ; dmolecular weight by sedimenta- tion equilibrium. (C) NATURE OF FORCES INVOLVED I N INTERMOLECULAR INTERACTION In looking for a factor which might influence association of the mucoprotein without affecting the conformation, the effect of temperature was investigated. Raising the temperature is seen (fig. 5) to bring about a marked increase in the viscosity measured at zero shear and at finite concentration of mucoprotein.This increase is fully reversible. The extrapolations to zero concentration at the different tempera- tures are compatible with there being no variation in intrinsic viscosity with tempera- ture and this is supported by the sedimentation behaviour (fig. 6). Thus, the changes in viscosity are associated largely, if not completely, with increased intermolecular interaction and, in keeping with this, the gel threshold is reduced from 4 g 1.-' at 25°C to 2 g 1.-l at 45°C. Further, the shear dependence of viscosity is substantially increased as the temperature is raised (fig. 7). It is concluded, therefore, that the interaction of mucoprotein molecules-the initial step in formation of mucous gel- has a negative temperature coefficient. These results suggest that entropy driven interactions are involved in .the association of mucoprotein molecules.Many218 GASTRIC MUCOUS GEL protein-protein interactions have an unfavourable enthalpy of interaction z 4 and this is usually associated with the classical entropy driven, hydrophobic interacti~n.~~ For a similar temperature coefficient to pertain to an association reaction dependent upon hydrogen bond formation or ionic interaction, it would be necessary to invoke temperature dependent conformational changes which would steadily expose more and more hydrogen bonding or ionic groups as the temperature was raised. The various known mechanisms for gel formation have been reviewedzz and include micelle junctions, entanglement, microcrystallites and double-helix junctions.In the present context it is interesting to note that the reversible formation of gels by 0-methyl cellulose has a similar temperature coefficient, is accompanied by an increase in volume and is inhibited by cations of the classical chaotropic series.26 These phenomena point towards the involvement of hydrophobic interactions 27 at micelle junctions.2z The behaviour of the mucoprotein suggests that such micelle junctions may be involved in mucous gel formation. 'i I f J I 1 I I 2 3 concentration/mg ml-' FIG. 5.-Temperature dependence of the reduced specific viscosity, corrected to zero shear, of the water extractable mucoprotein 20°C (m) ; 25°C (+) ; 37°C (A) ; 45°C ( 0 ) ; 20°C from 45°C (0). All measurements were in 0.2 M KCI : 0.02 % (w/v) azide ; 0.02 K acetate brought to pH 5.5 with acetic acid.i 0.1 0 concentration/mg ml-' FIG. 6.adimentation of the waterextractable mucoprotein. Plot of 1 js25,w as a function of muco- protein concentration 25°C (0); 40°C (corrected to 25°C) (A). All measurements were in 0.2 M KCI : 0.02 % (wlv) azide ; 0.02 K acetate brought to pH 5.5 with acetic acid.A . ALLEN, R. H . P A I N A N D D . SNARY 21 9 The protein component contains 39 % of non-polar amino acid residues, namely proline, alanine, valine, isoleucine, leucine, tyrosine and phenylalanine, which may take part in hydrophobic bonding. The sugar moiety may also be able to take part in these interactions, by virtue of the methyl groups on fucose and N-acetyl hexosamine which comprise 11.3 % and 27.8 % respectively of the total carbohydrate.This \ \ \ \ \ \ \ I I I 2 4 6 h/cm FIG. 7.-Reduced specific viscosity of the water extractable mucoprotein as a function of concentra- tion and of temperature. Plot of against shear force expressed by hydrostatic head, h.9 1.66 mg ml-' at 45°C (A) ; 2.79 mg ml-' at 25°C (@) ; 1.64 mg ml-' at 25°C (0). All measure- ments were in 0.2 M : 0.02 % (w/v) azide; 0.02 K acetate brought to pH 5.5 with acetic acid. content is similar to the low degree of methylation required to cause 0-methyl cellulose to The possession of A and H blood group activity means that at least some of the fucose and N-acetyl hexosamine residues are terminal and on the outside of the molecule.* From the accepted structure of these mucoprotein molecules, much of the surface must be carbohydrate in character and these residues are, therefore, highly likely to be involved in the interactions leading to gel formation. DISCUSSION The above results show that both the high molecular weight chemical structure and the highly expanded conformation of the mucoprotein must be preserved for reversible intermolecular interaction and gel-formation to take place, presumably through groups on the outer surface of the molecule.This soluble mucoprotein system comprises approximately 20 % of the stomach mucus the remainder being water-insoluble gel. There is, however, chemical, physical and immunological evidence that the mucoprotein is representative of the insoluble portion of the mucus 29 and that the latter is formed originally by association of monomeric mucoprotein molecules by non-covalent interaction.It is possible to further solubilize the water- insoluble gel using reagents which break non-covalent interactions, such as deoxy- cholate, urea and guanidinium chloride. It seems reasonable to suggest that this portion of the gel is formed in vivo over a period of time by an ageing process, under conditions when high local concentrations of mucoprotein are being generated.220 GASTRIC MUCOUS We thank the Medical Research Council and their financial support. H. A. Platt, Ann. N.Y. Acad. Sci., 1966, 130, 925. H. W. Florey, Proc. Roy. SOC. By 1955, 143, 147. N. G. Heatlev, Gastroenterology, 1959, 37, 304. GEL the Science Research Council for A. Gottschalk; GZycoproteins (Amsterdam, Elsevier, 1972). A. Allen and D. Snary, Gut, 1972, 14,666. K. 0. Lloyd and E. A. Kabat, Proc. Nat. Acad. Sci., 1968, 61, 1470. ' B. L. Slomiany and K. Meyer, J. Biol. Chem., 1972,247, 5062. W. M. Watkins, Science, 1966, 152, 172. A. Allen and B. Starkey, Trans. Biochem. SOC., 1974, 2, 630. lo B. Starkey, D. Snary and A. Allen, Biochem. J., 1974, 141, 633. l 1 D. Snary and A. Allen, Biochem. J., 1971, 123, 845. l2 H. K. Schachman, Methods Enzymology, 1957, 4, 32. l3 D. Snary, A. Allen and R. H. Pain, European J. Biochem., 1971, 24, 184. l4 J. M. Creeth and C. G. Knight, Biochem. J., 1967,105,1135. l6 D. Snary, A. Allen and R. H. Pain, Biochem. Biophys. Res. Comm., 1970,40,844. A. L. Sheffner, Ann. N. Y. Acad. Sci., 1963, 106,298. W. P. Jencks, CataZysis in Chemistry and Enzymology (McGraw-Hill, London, 1969), chap. 7. R. A. Robinson and R. H. Stokes, EZectroZyte SoZutions (Butterworth, London, 1970), chap. 1. l9 H. S . Frank and M. W. Evans, J. Chem. Phys., 1945,13, 507. 2o F. Franks, D. S. Reid and A. Suggett, J. Solution Chem., 1973, 2, 99. 21 M. J. Tait, A. Suggett, F. Franks, S. Ablett and P. A. Quickenden, J. Solution Chem., 1972,1, 22 D. A. Rees, Adv. Carbohyd. Chem. Biochem., 1969, 24,267. 23 D. Snary, A. Allen and R. H. Pain, European J. Biochem., 1973, 36, 72. 24 W. Kauzmann, Adv. Protein Chem., 1959, 14, 1. 2 5 C. Tanford, Adv. Protein Chem., 1970,24, 1. 26 E. Heyman, Trans. Faraday SOC., 1935,31,846. 27 W. B. Dandiiker and V. A. de Saussure, in Chemistry of Biosurfaces, Vol. I , ed. M. L. Hair z8 W. B. Neely, J. Polymer Sci., 1966, 1, 311. 29 D. Snary and A. Allen, Biochem. J., 1972,127, 577. 30 R. C. Helgeson, J. M. Timko and D. J. Cram, J. Arner. Chem. SOC., 1973,95,3023. 131. (Dekker, New York, 1971).

ISSN:0301-7249    

DOI:10.1039/DC9745700210

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Chain Conformations in the Sol-Gel Transitions for Poly- saccharide Systems, and their Characterisation by Spectroscopic Methods BY T. A. BRYCE, A. A. MCKINNON, E. R. MORRIS, D. A. REES” AND D. THOM Unilever Research Colworth/Welwyn Laboratory, Colworth House, Sharnbrook, Bedford MK44 1LQ Received 2nd January, 1974 The sol-gel transition has been studied for two polysaccharide systems, which form gels by the noncovalent association of chain segments in ordered conformations, but which differ in the geometry of these conformations. For t-carrageenan they are double helices, the formation and melting of which is monitored by [I3C]-n.m.r. and by optical rotation. For alginate, the associations involve the site-binding of Ca2+ ions in an egg-box or sandwich-like geometry, the formation of which is monitored and related to similar structures in solid films using circular dichroism spectroscopy.Despite the low concentration of dispersed polysaccharide (of the order of 1 %), each gel network is a very highly associated structure in which almost all sequences of sugar residues having the appro- priate covalent structure to permit the ordered binding, are actually involved in binding to other chains. In recent years a picture of polysaccharide gelation has emerged in which solvent is trapped in the interstices of a 3-dimensional network held together by “ tie points ” which involve the co-operative association of long regions of polymer chains.’. The associated regions are known as “junction zones ” and the polysaccharide seg- ments in them have a regular covalent structure and exist in ordered conformations.Since the chain typically exists in the sol as a random coil, the interconversion of sol and gel may involve a conformation transition which can be detected and characterized by spectroscopic methods. Such behaviour is already familiar for the gelation of a protein, gelatin, which shows a sharp shift in optical activity that can be attributed to the formation of the triple helix structure and can be used to derive information about the mechanism and kinetics of the formation of this str~cture.~. A polysaccharide system of this type is rc-carrageenan for which the changes in optical rotation which occur with the sol + gel + sol sequence have been explained in terms of a double helix model for the junction zones, corresponding to the double helix derived previously by diffraction analysis of oriented fibred More recent evidence suggests that these double helices are actually combined into larger aggregates.The gelation of certain polysaccharide mixtures such as agarose with some galactomannans may be accompanied by complex changes in optical rotation which give evidence for double transitions that involve co-aggregation or ordered binding between parts of the unlike chains.8 Yet other examples of polysaccharide systems are alginates and pectins which form gels by the ordered association of their polycarboxylic chains, brought about by calcium ions. The perturbation of the n-n* circular dichroism band of the carboxylate chromo- phore in this process can be used to derive information about the nature of the micro- crystalline junction zones and the forces which hold them together.g* l o In this paper, we give more examples of the scope of spectroscopic methods in general and of chiroptical methods in particular, in investigating the geometry of the 221222 POLYSACCHARIDE GELATION ordered associations and in advancing our understanding of the structures of poly- saccharide gels.We have examined &-carrageenan and alginate as examples of two different types of polysaccharide gel. Both covalent structures (I and 11) can be regarded as block copolymers in which certain sequences are able to adopt the ordered conformations to enter that remain " soluble " F-- 0 junction zones, and are interspersed with other sequences and play a complementary role in the gel framework.In +n 0 s 0,- (b) Formula I -1 7 . . I &-Carrageenan Note that this formula describes a particular sample which contains no 2,6-disulphate residues. The integcrs rn and n do not necessarily have unique values, even within a given polysaccharide chain. r- -1 L, - -J9 (b) Formula I1 I1 Alginate This is a schematic formula because the sequence of (a), (b) ad (c) in the chain is unknown. The integers m, n and p do not necessarily have unique values, even within a given poly- saccharide chain. &-carrageenan the junction zone is a double helix (fig. 1) and the helix-forming sequences are those shown as I(a); the helix-breaking " soluble sequences " are shown as I@). The latter can be removed by chemical degradation to give shorter chains, known as " carrageenan segments ", which retain the ability to convert from coil to double helix but have lost the ability to gel.In alginate (11) the sequences which have the most important role injunction zones are shown as II(b) ; II(a) can share this function but, depending on the conditions of the experiment, may also function with Il(c) as j unction-breaking " soluble sequences ". This gelation involves the co-operative binding of Ca2+ by polyguluronate sequences, i.e., by II(6), in inicrocrystallineBRYCE, MCKINNON, MORRIS, REES A N D THOM 223 junction zones which, in the light of the evidence from c.d.,'O have been described in terms of an " egg-box " model in which there is regular packing and co-ordination of the cations in the interstices of associated, extended segments of the polymer chains (fig.2). Although the interactions of alginate with Ca2+ ions and the strength and texture of the resulting gels are dominated by this association of poly-a-L-guluronate sequences, the c.d. evidence suggests that the poly-P-D-mannuronate sequences niay also play a part.g* lo FIG. 1.-Schematic representation of the sol-gel transition for &-carrageenan, involving double helix junction zones. /- FIG. 2.-Schematic representation of the sol-gel transition for alginate, involving egg-box junction zones. EXPERIMENTAL &-Carrageenan was obtained from a commercial source (Auby Gel, France; X-52) and was segmented by Smith degradation then treated with alkaline borohydride.' ' Further details of the product are given in another paper at this Discussion.12 Sodium alginate was obtained from Alginate Industries Ltd., and purified by extensive dialysis against deionized water followed by filtration through glass paper, then isolated by freeze-drying.Analysis showed that the ratio of mannuronate to guluronate residues was 57 : 43, and the ratio of polyguluronate : polymannuronate : poly(mannuronosylgu1uronate) sequences i.e., II(6) : II(a) : II(c), was 21 : 38 : 41.13 SPECTROSCOPIC METHODS Optical rotation measurements were made at 365 nm and 436 nm, using the Perkin Elmer 141 polarimeter with jacketed cells of pathlength 1 cm, using methods described previously.8 [l3C]-N.rn.r. spectra were measured at 25.2 MHz with the Varian XL-100 high resolution spectrometer equipped with the Fourier transform accessory.Solutions were 6 % (w/v) in &-carrageenan and all spectra were proton decoupled, were summed over about 65 O00 transients, and were recorded using a sweep width of 5000 Hz (198.4 p.p.m.). Chemical shifts are in terms of SC which describes the downfield shift of the resonance in p.p.m. relative to DSS. C.d. spectra were recorded with the Cary 61 spectropolarimeter using a 10 s integration period. All measurements were at 25°C after Equilibration of the224 POLYSACCHARIDE GELATION sample. Cell blanks were recorded with deionized water and 6 mM calcium chloride for solutions and gels respectively. Spectra of films were recorded using specially constructed holders and, of course, required no baseline calibration.Difference spectra were calculated by subtraction of ellipticities at intervals of 2.5 nm from 260 nm to 192.5 m. SAMPLE PREPARATION Solutions of &-carrageenan were prepared by dispersion in water or in 0.1 M sodium chloride, in a sealed tube, followed by heating in an autoclave at 15 lb/in2 for about 20 min. To minimize any interference from dust, solutions were passed through a Millipore filter (1.2 pm) before making any measurements. Calcium alginate gels were made by passing a solution of sodium alginate (0.1 % at pH 7.0) through a Millipore filter (3 pm) then dialyzing in a c.d. cell of pathlength 10 mm, against a large excess of well stirred aqueous calcium chloride (6mM). The cell was removed from the solution at various times and, after drying and cleaning the end plates, the spectrum was recorded and the cell was returned to continue the dialysis.The experiment was continued until no further change was observed in the c.d. spectrum. Alginate films were cast on PTFE to give a film of known effective thickness (equivalent to a solution pathlength of 0.188 mm), as follows. An ac- curately measured volume (1 ml) of sodium alginate solution (1 % w/v) was confined to a known area by rubber rings (26 mm diam.) sealed onto the PTFE plate by silicone grease. To obtain films of the sodium salt, the solutions were dried over silica gel for 1-2 days. To obtain films of the calcium salt, the surface of the solution was lightly sprayed with calcium chloride solution (6 mM) to form a skin of calcium alginate then the plate was immersed in the same calcium chloride solution for 16 h before drying over silica gel as for the sodium salt.RESULTS AND DISCUSSION CARRAGEENAN GELS It has already been shown that the segments of &-carrageenan which correspond to the regions I(a), give a temperature-induced shift in optical rotation which agrees closely in sign and magnitude with calculations based on the double helix model from diffraction ana1~sis.l~ To make this correlation it was necessary at the time to assume that the conversion from coil to helix was complete ; this assumption was later verified by monitoring the change in particle weight during the tran~iti0n.l~ Using [13C]n.m.r. at natural abundance, we now give further evidence that the segments go to a state that is essentially fully helical.We will also use these spectra to charac- terize certain other details of the transitions of segmented and gelling L-carrageenans. Earlier attempts l 6 to study the helix-coil transition of carrageenan by C'H1n.m.r. were because the spectra were complex and poorly resolved. This was not surprising because peaks in the high resolution proton spectra of polysaccharides are very often broad and overlapping owing to restricted segmental motion, and only the anomeric proton resonance is generally useful in structure determinations. However, well re- solved [' 3C]spectra can often be obtained for polysaccharides, partly because of larger chemical shift differences and partly because the smaller magnetic moment of the carbon nucleus results in diminished dipolar broadening.At 80°C, when the carrageenan segments were judged from optical rotation to exist entirely as the random coil, the spectrum showed well defined peaks which should be capable of complete assignment when appropriate model spectra are available (fig. 3). Provisional assignments are that the signals at 104 and 94p.p.m. are the resonances of the anomeric carbon nuclei and that the signal at 63p.p.m. is the resonance of the hydroxymethyl carbon. Other signals are in the region between 70 and 80 p.p.m. All these peaks disappeared from the spectrum (fig. 3) when the polysaccharide was near the limit o f conversionBRYCE, MCKINNON, MORRIS, REES AND THOM 225 to helix as judged by optical rotation (i.e. at 1SOC). Evidently the mobility of residues in the helix is so severely restricted that the relaxation times for the carbon nuclei are decreased such as to cause broadening and hence the loss of peaks from the high resolution spectrum.Evidently also, the limit of the optical rotation change does indeed correspond to complete helix formation as we have always assumed, at least within the sensitivity of n.rn.r. detection, because no signals at all remain for the sample in this state. 110 100 90 80 70 60 50 p.p.m. from DSS FIG. 3.-Comparison of [‘%]-n.m.r. spectra for &-carrageenan segments at 80°C (upper spectrum) and 15°C (lower spectrum). The concentration was 6 % (w/v) in 0.1 M sodium chloride. At intermediate temperatures, which correspond to partial conversion of coil to helix, the [13C]n.m.r. resonances are diminished in amplitude to the extent that would be expected from estimates made of helix content from optical rotation.However, the signals are not broadened or shifted and therefore show no evidence of any time-averaging of resonances between helix and coil; this is consistent with the “ two state all-or-none model ” for the carrageenan transition.12 A similar progres- sive disappearance of intensity without other changes in the high resolution spectrum has been seen with [lH]-n.m.r. spectra for a-gelatin, and has been interpreted in essentially the same way.17 The molecular weight data that are available for segments l 2 . l 5 and for intact carrageenans l8 would indicate that each chain (I) may contain 8 or more helix- forming regions [I(a)].The conversion to helix of so many independent segments within each chain might be expected to be hindered, especially as the network develops to tie down chain ends and prevent the further coaxial twisting that is necessary to propagate helix formation. Now that we can be sure of the extent of helix formation by the segments, it is possible to investigate this situation by comparison of the optical rotation transitions in the polysaccharide and the segments. Fig. 4 shows one com- parison that is typical of a series at different concentrations. Helix formation begins 57-H226 POLYSACCHARIDE GELATION at about the same temperature in both but seems to be relatively hindered in the system in which the network develops, in that the cooling curves diverge progressively.Nevertheless, the final value for the optical rotation of the gel corresponds to the conversion to helix of 70-90 % of the regions having the appropriate covalent struc- ture. The remarkable conclusion is reached that at least 6 independent double helices are formed within each chain. Evidently pathways do exist by which the system can surmount topological obstacles to achieve a very high conversion to helix when this is favoured thermodynamically. I 10 2 0 30 4C 5 0 5 0 70 temperature/"C FIG. 4.-Comparison of the optical rotation changes with temperature for native c-carrageenan (0) and &-carrageenan segments (a), at 436 nm. The measurements are for 6 % (w/v) solutions in dis- tilled water, in a 1 cm cell. Readings are independent of thermal history.ALGINATE GELS Our new c.d. studies show that the environment of carboxylate chromophores in alginate gels in which chain-association occurs spontaneously under the influence of Ca2+, is very similar to that in dried films of various salts which are prepared by forcing the chains to associate by evaporation of solvent, and very different from the environment in solutions in which no association exists. Because alginate salts are characteristically crystalline in the solid state (although crystallinity is enhanced by annea1ing),lg it is likely that the associations in the film and the gel are in crystal- lites having the same lattice but perhaps having different overall dimensions. The evidence for this conclusion is that the c.d. spectra of solid films of calcium, sodium and other alginates, differ from the spectrum of the solution (fig.5) in preciselyBRYCE, MCKINNON, MORRIS, REES A N D THOM 227 the way that we have already observed ’* lo for the gel and the sol (fig. 6). These comparisons are conveniently shown by difference spectra for films and gels having the same amount of calcium alginate in the light path, referred to a solution of sodium alginate (fig. 6). These curves have exactly the same shape, width and position, and differ only in that the magnitude is slightly larger for the film than for the gel. wavelength A/nm FIG. 5.-Comparison of c.d. spectra for a solution (0.1 %, w/v) of sodium alginate (-), a film of sodium alginate (- - -), and a film of calcium alginate (. - -).Also shown are difference spectra for the sodium alginate film (--) and the calcium alginate film (0 O), referred to sodium alginate solution and sodium alginate film respectively. Taking this magnitude as a measure of chain-association, it may be concluded that the extent of association in the gel is over 90 % of that in the calcium alginate film. A sodium alginate film shows a difference spectrum that is very similar in form but which is much diminished in magnitude (fig. 5), suggesting that sodium alginate associ- ates in essentially the same way as the calcium salt but that the spectroscopic perturba- tion of the carboxylate chromophores by Na+ is much less than by Ca2+. X-Ray diffraction studies of alginates show that polyguluronate sequences [II(b)] usually crystallize most readily, and that all salt forms of these sequences retain a chain conformation having a two fold screw axis and an axial rise of 8.7 A per disac- charide residue which implies the lc ring conformation.19~ 2 o This agrees with the structures for the junction zones that have been suggested lo* 21 to be formed from such sequences when the sol -+ gel transition is induced with Ca2+ ions.Crystallization228 POLYSACCHARIDE GELATION of polymannuronate sequences can also occur and in the calcium salt form these crystallize with each chain in a three-fold helix with an axial rise of about 15 A which implies the cl ring conf~rmation.~~ The qualitative and quantitative agree- ment between the c.d. spectra of calcium alginate gels and films therefore supports our suggestion lo that the association of polymannuronate sequences occurs in gel formation, as well as the association of polyguluronate sequences. 0- 9, 1.. \ / * * * - a wavelength A/nm FIG. 6.-Change in the c.d. spectrum with diffusion of Caz+ to a final concentration of 6 mM into a solution of sodium alginate (0.1 % w/v). Spectra correspond to the solution (-) and the final gel (- - -). Also shown are difference spectra for the gel (a * -) and the calcium alginate film (--), both referred to sodium alginate solution. CONCLUSIONS Spectroscopic studies, and especially chiroptical studies, have led us to propose quite detailed models for the structure of carrageenan and alginate gels and for the extent of conformation change which is involved in each sol + gel transition.The structure of each gel is viewed as being very similar to that of a dry, condensed film and having almost the same degree of ordered chain-association as expected in a solid state structure. The important difference is that the gel is expanded in the third dimension, with minimum disruption of ordered conformations, to form large interstices which are filled with solvent. For alginate and carrageenan gels, the orderedBRYCE, MCKINNON, MORRIS, REES AND THOM 229 associations have different types of geometry but each associates to almost the maximum level possible. We thank Mr. R. N. Robertson for recording the [13C]-n.m.r. spectra, and Mr. G. A. Young for expert technical assistance. D. A. Rees, Adv. Carbohydrate Chem. Biochem., 1969,24,267. D. A. Rees, Biochem. J., 1972,126,257. W. Traub and K. A. Piez, Adu. Protein Cliem., 1971, 25,243. P. J. Flory and E. S. Weaver, J. Amer. Chem. SOC., 1960,82,4518. D. A. Rees, I. W. SteeIe and F. B. Williamson, J. Polymer Sci. C, 1969, 29, 261. Biol., 1969, 45, 85, I. C. M. Dea, A. A. McKinnon and D. A. Rees, J. Mol. Biol., 1972,68,153. E. R. Morris, D. A. Rees and D. Thorn, Chem. Comm., 1973,245. lo G. T. Grant, E. R. Morris, D. A. Rees, P. J. C. Smith and D. Thom, F.E.B.S. Letters, 1973, 32, 1. A. A. McKinnon, F. B. Williamson and D. A. Rees, Chem. Comm., 1969, 701 ; for further details, see ref. (8). ' N. S. Anderson, J. W. Campbell, M. M. Harding, D. A. Rees and J. M. B. Samuel, J. Mol. ' A. A. McKinnon, Ph.D. Thesis (University of Edinburgh, 1973). l 2 T. A. Bryce, A. H. Clark, D. A. Rees and D. S. Reid, paper read at this Discussion. l 3 A. Penman and G. R. Sanderson, Carbohydrate Res., 1972,25,273. l4 D. A. Rees, W. E. Scott and F. B. Williamson, Nature, 1970,227, 390. l5 R. A. Jones, E. J. Staples and A. Penman, J.C.S. Perkin ZI, 1973, 1608. l6 F. B. Williamson, Ph.D. Thesis (University of Edinburgh, 1970). l7 D. Eagland, et al., paper read at this Discussion. 10 D. A. I. Goring and E. G. Young, Canad. J. Chem., 1955, 33,480. 19W. Mackie, Biochem. J., l971,125,89P. 2o E. D. T. Atkins, I. A. Nieduszynski, W. Mackie, K. D. Parker and E. E. Smolko, Biopolymers, 21 D. Thorn, Ph.D. Thesis (University of Edinburgh, 1973). 1973,12,1865, 1879.

ISSN:0301-7249    

DOI:10.1039/DC9745700221

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Helix-Coil Transition in Gelling Polysaccharides BY D. S . REID, T. A. BRYCE, A. H. CLARK AND D. A. REES Unilever Research ColworthlWelwyn Laboratory, Colworth House, Sharnbrook, Beds., England Received 10th December, 1973 The coil/helix transition in solutions of a degraded i-carrageenan in 0.1 M NaCl has been studied using optical rotation methods and differential scanning calorimetry. Provided the molecular weight distribution profile is included in the model, a simple two-state all-or-none mechanism is an adequate. description of the transition process. The enthalpy and entropy changes which accompany the transfer of a disaccharide residue from the double helical conformation to the coil conformation are, based on the two-state all-or-none model, AH; = 1.2 kcal (mol of residue)-' and AS; = 3.55 cal (mol of residue)-' K-'.Certain polysaccharides, such as the carrageenans and agarose (fig. l), can, in dilute aqueous solution, form thermally reversible gels. During the gelation process a characteristic change in the optical rotation (0.R) of the system occurs. From a consideration of this phenomenon, and a study of the X-ray diffraction patterns obtained from certain polysaccharide fibres, it has been inferred 1-4 that these poly- saccharides are capable of existing in a double helical conformation, and that the gel results from the association of polysaccharide chains to form a three-dimensional network, the junction zones containing double helical regions. It has also been shown that the transition from sol to gel can be detected by differential scanning calorimetry.L AGAROSE n L CARRAGEE NAE) FIG. 1 .-Structure of agarose and carrageenan. R = H x-carrageenan ; R = SO; i-carrageenan. In protein and nucleic acid systems which undergo a thermodynamically reversible coil/helix transition, much useful information has been obtained from a comparison of the directly-measured calorimetric behaviour of the system with the thermo- dynamic properties calculated indirectly from the temperature dependence of the 230D. S . REID, T. A . BRYCE, A . H . CLARK AND D. A . REES 23 1 equilibrium between helical and coil conformations. A similar study involving polysaccharide coillhelix transitions might therefore increase our understanding of the mechanisms of this transition. For such a study, we require a material which undergoes a thermodynamically reversible conformational change, and which does not gel, since the existence of a gel network may produce undesirable side effects.We have available a material which meets these criteria-a degraded i-carrageenan. This material exhibits a reversible transition (as observed by O.R.) and does not gel. EXPERIMENTAL A sample of i-carrageenan (Auby Gel X-52) was degraded by the procedure described by McKinnon et aL4p The degraded X-52 is ideally an alternating copolymer of p-D galactose-4-sulphate and 3,6-anhydro-a-~-galactose-2-sulphate. A range of concentrations of the potassium salt of degraded X-52 in 0.1 M NaCl was prepared. The optical rotation of these solutions was followed as a function of temperature, using a Perkin-Elmer 141 polarimeter with a 1.00 cm cell.A Perkin-Elmer differential scanning calorimeter, DSC-2, was used in order to measure the enthalpy change associated with the coil/helix transition, both cooling and heating scans being employed. The methods of sample preparation have been de~cribed.~ From the area of the DSC peak associated with the transition, the enthalpy change could be calculated. THEORY If we assume that each chain segment has a characteristic optical rotation when in the helical state, and a characteristic optical rotation when in the coil form, then the proportion of residues in the helical state at any temperature Tis given by a, the degree by which the O.R. curve has approached its " all-helix " value from the " all-coil " value.The equilibrium can be described as A +nB where A represents the helix and B the coil. brium constant for this two-state all-or-none process is We assume that n = 2. The equili- K = [BI2/[A]. The concentration of A is ccc, and that of B is 2c(l -a), where c is the molar con- centration of helices in the " all-helix " system. Assuming the system to be mono- disperse, the following relationships would hold : d In K AHo d(l/T) - -- R ' -- assuming AN" is constant. -AH" l n c = +constant (at constant a), ( n - l)RT (3) In (ca) = n ln(c[l - or]) + constant (at constant T). (4) Hence AH" and n could be derived. However, since the system is unlikely to be mono- disperse, these relationships are not necessarily valid. In order to test the applica- bility of the simple two-state all-or-none mechanism, we must adopt the following procedure, which requires knowledge of the molecular weight distribution of the material.We assume that for each distinct molecular species of chain length i, the mechanism of association is two-state all-or-none, represented by Ki Ai +2Bi232 HELIX-COIL TRANSITION IN GELLING POLYSACCHARIDES where A, represents a double helix formed by the association of two coiled chains Bi, of length i residues, (i.e., disaccharide units). We assume that the enthalpy change AH: associated with this process is given by AH: = 2iAH,", ( 5 ) where AH," is the enthalpy of dissociation per mole of residue, (i.e., disaccharide unit) determined calorimetrically. Similarly, the entropy of dissociation, AS': is given by ASP = 2iAS,".(6) K,, the dissociation constant for the double helix, is given by AH; AS: In&= --+-, RT R and can be calculated at any temperature. From Ki, a,, the fraction of chains of length i which are in the helical form, can be calculated, since Ki = 4( 1 - ui)2ci/at, (8) where ci is the molar concentration of helices Ai in the "all-helix" state. a, the overall fraction of residues which are in helical form, is given by In order to test the model, the following stepwise procedure is adopted. (1) A preliminary estimate of the characteristic '' all-helix " and " all-coil " rotations (as functions of temperature) is made (for a chosen concentration)-see fig. 5- and the a(T) curve is calculated from the experimental data. (2) Values of AS," are chosen, and a(T) curves calculated using eqn (5)-(9).(3) The value of AS," which optimizes the fit of the theoretical a(T) curve to the experimental a(T) curve at small values of a is found. This value is adopted in all subsequent calculation. (4) Using the theo- retical value of a at some low temperature, where a > 0.9, a corrected " all-helix " optical rotation is calculated (using the estimated " all-coil " rotation) such that the revised a (experimental) is equal to a (theoretical). (5) The temperature dependence of the revised " all-helix " rotation is assumed to be that of the '' a l l a i l " rotation. (6) The experimental a(T) curve is recalculated. (7) The " all-coil " rotation base line is allowed to vary within its established error limits, and steps (4)-(6) repeated.(8) The values of '' all-helix " and " all-coil " rotations which give the best fit of ex- perimental and theoretical a(T) curves are identified, and the specific rotations for both coil and helix are expressed as functions of temperature. (9) Using the values of AS:, coil specific rotation and helix specific rotation established by the above procedure the theoretical a(T) curves, and hence the theoretical optical rotation curves for other concentrations are calculated, and compared with the experimental results. RESULTS AND DISCUSSION The molecular weight distribution profile of the degraded i-carrageenan was ob- tained using a technique of gel electrophoresis, that had been calibrated using i- carrageenan standards previously characterized by equilibrium sedimentation.This profile is shown in fig. 2. The average molecular weights were found to be an = 51 000, M, = 71 OOO for dimeric (i.e., double helical) species. From the molecular weight distribution profile, the concentration ci of any species in a solution can be calculated, knowing the weight of polysaccharide contained in a given volume of the solutj on.D. S. REID, T . A . BRYCE, A . H . CLARK AND D . A . REES 233 1.1- 0.s- 5 8 e 0.7 a, .r( c1 Y - 0 a .-.I c. 0 0.5 0.3 i FIG. 2.-Molecular weight distribution profile of degraded i-carrageenan sample ; fraction of chains of length i plotted against i. - - - T/"C FIG. 3.-Change of optical rotation of solutions of degraded X-52 as a function of temperature. A, 1.04 %; B,3.0 %; C,4.04 %; D,5.66 %.234 HELIX-COIL TRANSITION I N GELLING POLYSACCHARIDES The optical rotation curves, at a wave length of 365 nm, for several concentrations are shown in fig.3, and the results obtained from the differential scanning calorimetry are summarized in table 1. In order to demonstrate that the simple approach, TABLE 1 .-SCANNING CALORIMETRIC DATA FOR THE COlL/HELIX TRANSITION IN DEGRADED i-c ARR AGEENAN heating scans cooling scans cal g- 1 cal g-1 % concentration AH(he1ix --f coil) AH(coil+ helix) 3.00 2.13 3.03 4.93 - 2.15 2.22 2.07 4.98 2.24 Since residue weight for i-carrageenan (K+ salt) is 542, AH; = 1.2 kcal (mol of residue)-’, based on heating scans. ignoring polydispersity, is misleading AHtH was calculated from a van’t Hoff plot (fig. 4) for a 3 % solution. AH& was found to be 34.7 kcal/mol (1 cal = 4.184 J), which corresponds to an average chain length of 14 residues.This compares to M, = 51 000, an average chain length of 47 residues. Clearly, when a sample is polydisperse, the simple van’t Hoff approach is unsatisfactory. There are, however, methods of data analysis based on eqn (4) which may allow n to be determined without requiring prior knowledge of the molecular weight distribution. These methods, which will be described in a future publication, do not give detaiIed thermodynamic information relating to the energetics of the transition. K x 1 0 3 / ~ FIG. 4.van’t Hoff plot for a 3 % solution of degraded X-52. In order to obtain a thermodynamic description of the transition, we have used the procedure outlined in the theoretical section, in conjunction with the known molecular weight distribution profile to fit the optical rotation curve for a 3 % solution of the degraded i-carrageenan (fig.5). The fitting parameters so established are listed in table 2. The expected optical rotation curves for other concentrations were calculated. These are compared with the experimental optical rotation curves in fig. 6. From the results of these comparisons, it is clear that the model of the association process described in the theoretical section is capable of adequately accounting for theD. S . REID, T. A . BRYCE, A . H . CLARK A N D D. A . REES 235 experimental observations. Hence the association process which produces the- junctions in the gelation of i-carrageenan can be considered as a simple two-state all- or-none process, in which the two chain segments are either completely independent of one another in the coil conformation, or fully associated in a double helix.The lack of sharpness of the transition, compared to that which one would expect for a species of average chain length 47 residues is a consequence of the polydisperse nature of the sample. When this polydispersity is taken into account, one can adequately account for the observed behaviour of the system on the basis of a two- state all-or-none model of chain association. *I- \ 0.3 T/"C FIG. %-The fitting of the experimental optical rotation curve (solid line) for a 3 % solution of degraded X-52 to the theory. Line A is the initial estimate of the characteristic " all-helix " rotation, line B is the final estimate of the characteristic " all-helix " rotation and line C the final estimate of the characteristic " all-coil " rotation. The points are calculated from the theory, using the para- meters listed in table 2.TABLE 2.-FImG PARAMETERS FOR THE TWO-STATE ALL-OR-NONE MECHANISM AH: = 1.2 kcal (mol of residue)-l, AS: = 3.55 cal (mol of residue)-' K-l. Specific rotation (helix) = 211-0.192t; specific rotation (coil) = 134-0.1922, where t is the temperature (in "C). The question arises whether the values of AH," and AS: used to fit the experimental data are in accord with what one might expect., In the solid state, it has been sug- gested that there is one hydrogen bond per disaccharide residue, which contributes, together with van der Waals forces etc., to the stabilization of the helix.This hydro- gen bond is also expected to contribute to the stabilization of the helix in solution. The relative contributions of this hydrogen bond, van der Waals forces, differences in the solvation of the helix and the coil to the helix stabilization are uncertain. This236 HELIX-COIL TRANSITION IN GELLING POLYSACCHARIDES makes it difficult to predict AH," or AS,". However, the results are of similar order to those determined for some other coil/helix processes, as widely different as coil-single helix coil-double helix and coil-triple helix and so would appear to be reasonable. 1.1- 0.0 a G 0.1- 0 (d - .r( Y c. 2 - .- 8 c) a O 0.s- 0.3 - - -EXPERIMENTAL 0 THEORETICAL n C A T/"C points calculated using the parameters listed in table 2 ; concentrations as in fig.3. FIG. 6.-A comparison of the experimental and theoretical optical rotation curves. Theoretical CONCLUSION The coil/helix transition of degraded i-carrageenan can be adequately described by a simple two-state all-or-none mechanism, provided that the molecular weight distribution of the carrageenan sample is included in the calculation. This is the first time that such a mechanism has been shown to apply to a coil/helix transition in a polysaccharide. The model employed might be made more realistic by the inclusion of end-effects (i.e., a helix initiation parameter), but such an improvement would only be justified if one had a more nearly monodisperse material available for study. We thank Dr. N. F. Stanley, of Marine Colloids Inc., for establishing the mole- cular weight distribution profile. D. A. Rees, Biochem. J., 1972, 126,257. Biol., 1969, 45, 85. * N. S. Anderson, J. W. Campbell, M. M. Harding, D. A. Rees and J. W. B. SamueI, J. Mol.B. S . REID, T. A. BRYCE, A . H. CLARK AND D . A. REES 237 D. A. Rees, I. W. Steele and F. B. Williamson, J. Polymer Sci. C, 1969, 28, 261. A. A. McKinnon, D. A. Rees and F. B. Williamson, Chem. Comm., 1969,701. D. S. Reid and D. J. Tibbs, Thermal Analysis, Proc. 3rd I.C.T.A. (Davos, 1971), ed. H. G. Wiedemann (Birkhauser Verlag, Basel), vol. 3, p. 423. 1. C . M. Dea, A. A. McKinnon and D. A. Rees, J. Mol. Riol., 1972,68, 153. J. Applequist and V. Damle, J. Amer. Chem. Suc., 1965, 87, 1450. N. Gb and Y. Suezaki, Biopolymers, 1973,12,1927. ’ J. Hermans, J. Phys. Chem., 1966, 70, 510.

ISSN:0301-7249    

DOI:10.1039/DC9745700230

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

GENERAL DISCUSSION Prof. M. Gordon (University of Essex) said: Pain mentioned that about 20 % of the mucous gel was soluble, but was found to have the same composition as the remaining insoluble part. He attributed the partial insolubilisation to some kind of ageing process. However, the existence of a sol and a gel fractions in the critically branched structure just after gelation is a statistical consequence of the random linking process, as first explained by Flory long ago (cf. the paper by Covas et al., this Discussion, in which references are given). Thus no special explanation like an ageing process is needed. Dr. C. Marriott (University of Nottingham) said : In the light of the erratic behav- iour exhibited by sulphydryl compounds in breaking down bronchial mucus in vivo, I would be interested to know how long the mercaptoethanol was left in contact with the mucus before the gel structure was lost.Also was any fall in viscosity after treatment with the proteolytic enzyme trypsin detectable after periods as short as one hour? Dr. A. Allen (University of Newcastle upon Tyne) said: In all experiments samples were dialysed for 24 h against buffer containing 0.2 M mercaptoethanol. This treatment always gave complete breakdown of the mucoprotein into the 4 subunits with loss of gelling ability. Incubation with crystalline trypsin (I mg/ml) for one hour reduced the specific viscosity of the water soluble mucus by 45 %. Six hours incuba- tion with trypsin reduced the specific viscosity of the mucus by 72 % after which fur- ther incubation, even with fresh additions of trypsin, caused no more decrease in the specific viscosity.Prof. D. Rees (Unilever) said: What is the evidence that part of the protein core of the mucoprotein is free from carbohydrate side chains ? Dr. A. Allen (University of Newcastle upon Tyne) said: Enzymatic digestion of the mucoprotein with trypsin removes 22 % of the protein without any loss of the carbo- hydrate side chains. Amino acid analysis shows the protein remaining in the digested mucoprotein to be rich in serine, threonine and proline residues. Threonyl- and seryl- ether linkages are the mode of attachment for the carbohydrate side chains to the protein. The amino acid composition of the part of the protein hydrolysed by trypsin is more characteristic of a globular protein.Cleavage by trypsin of this protein splits the mucoprotein into two subunits of equal molecular weight. Prof. F. Franks (Unilever) said : The experiments described by Pain show the high degree of specificity which is one of the chief characteristics of biopolymer gels. On the question of the hydrodynamic volume, the figure of 120 OOO 1. mol-1 appears at first sight even more unbelievable than the 2000 1. mol-1 quoted by Eagland for a-gelatin. Even allowing for an expanded molecule, the peptide backbone must have some degree of tertiary structure which would permit the carbohydrate residues to take up their specific conformations at the periphery of the molecule. Can the dimensions suggested by V, be reconciIed with any loosely coiled " bottlebrush " structure having the composition of the mucoprotein? At the beginning of the Discussion section the authors imply that gelling takes 238GENERAL DISCUSSION 239 place via carbohydrate residues.Is this compatible with the suggestion made earlier that micelle/hydrophobic type contacts are involved ? Dr. R. H. Pain (University of Newcastle upon Tyne) said: The figure of 56 1. kg-1 for the effective hydrodynamic volume is not excessive when compared with values for other polysaccharide-rich molecules. The radius of gyration of hyaluronic acid leads to a volume of 6 x lo3 ml g-1 and the volume of exclusion for albumin for for the same polymer is 25 ml g-l. Calculation of the volume swept out by our mucoprotein molecule, based on a " bottle brush '' structure with polysaccharide chain lengths of 15 residues, leads to a value of 58 ml g-l, which suggests that the value of Ye calculated from flow properties is not unrealistic.It also indicates that, as the molecule has been shown to be spherical, it must be folded, probably at the sections of protein not involved in sugar linkage. Certainly, we believe the flow pro- perties to be compatible with a model structure based on the chemical evidence outlined in the introduction. While the mucoprotein contains units of " bottle brush " structure, the chemical evidence that it is not a simple loosely coiled bottle brush structure but, as with the laboratory bottle brush, contains regions of protein with very low serine and threonine content and no carbohydrate. Further, the molecule is composed of units joined together covalently by disulphide bonds.One realises the assumptions involved in talking about an effective hydrodynamic volume, but the values of V, quoted in the paper may be taken to substantiate the concept of this mucoprotein as a highly swollen molecule and also the contraction which occurs on exposure to CsCl. With regard to the second part of Franks' question, we said that gelling takes place presumably through groups on the outer surface of the molecule. Immunological and chemical evidence demands that a high proportion of these outer groups must be carbohydrate in nature. There is also evidence, however, that protine moieties are on the surface as shown by proteolytic digestion studies (cf. fig. 1). It is therefore an open question as to whether protein or carbohydrate groupings-or both-are involved in the intermolecular interactions. Although the classical " hydrophobic " interaction has been discussed in terms of proteins, we have put forward grounds in Section (c) of our paper for considering the occurrence of entropy-driven interactions between polysaccharide moieties in this mucoprotein.Mr. H. Beltman and Prof. J. Lyklema ( Wugeningen) said) said: In view of the conclusion by Pain et ul. that hydrophobic interactions govern the association of mucoproteins, it could perhaps be interesting to study also the effect of hydrophobic bond breakers like urea, methanol, etc. We have observed a marked methanol effect with PVA gels. In a PVA-resorcinol gel, hydrophobic bonding between the macro- molecular backbones probably causes gelation.Addition of methanol to the gel leads to dissolution. However, in PVA-congo red gels factors other than hydrophobic bonding are responsible for the crosslinking and in this case methanol addition has no measurable effect. Dr. R. H. Pain (Unioersity of Newcastle upon Tyne) said: Urea, guanidinium chloride and deoxychlorate will break down the mucoprotein intermolecular inter- actions (see Discussion). The fact that guanidinium chloride also brings about an expansion of the isolated mucoprotein molecule (see table 3 of our paper) means that B. N. Preston, M. Davies and A. G. Ogston, Biochem. J., 1965, 96, 449. * T. C. Laurent and A. G. Ogston, Biochem. J., 1963, 89,249.240 GENERAL DISCUSSION it is impossible at the moment to say whether non-covalent bond breakers are dissolv- ing the gel by acting primarily on the intra- or intermolecular interactions.Prof, M, Gordon (University qf Essex) said : Rees spoke of lateral aggregation of helices in the gel. For the gelation of gelatin, we showed (Covas et al., this Discussion) that the modulus as function of helix content (cross-link density) behaves as expected if the early gel stage involves helical sections formed in random positions. In gelatin also, bundles of aggregated helical sections are sometimes inferred to exist from electron micrographs taken on matured gels. It is difficult to see how aggregation c’ould occur by translational movement of helical sections through the initial random network once it is developed beyond the essentially tree-like structure. Does Rees consider the lateral aggregates to consist of crystals with a single space- lattice, or are the aggregates due to a kind of repeated epitaxial nucleation near the surface of a double-stranded helix? Prof.D. A. Rees (Unilever) said : In agarose and K-carrageenan, it appears (for example by monitoring the helix formation by optical rotation and the aggregate formation by light scattering) that many, perhaps most, of the aggregates do in fact form at the same time as the helices and might indeed drive helix formations so I don’t think this is inconsistent with your observations. I do believe that some aggregation occurs in the gel network when it is almost fully developed but I accept that this must involve some distortion of the network, unless there is a contraction of network volume.Such changes in volume are in fact well known for polysaccharide gels, in the phenomenon of syneresis or “ weeping ” and are very troublesome in practica1 applications. Some movement of helices between one aggregate and another is also indicated by Prins’ work (this Discussion). The aggregates in agarose are not crystalline-at least, in films prepared from agarose gels we have never been able to maintain or restore the crystallinity if it ever existed, in that the diffraction diagram corresponds to the cylindrically averaged transform of the agarose helix with no Bragg reflections. Rather, I believe that the driving force for aggregation is in the rigidity of the agarose helix-a phenomenon that I believe was first described by Flory for certain other examples.2 These comments do not apply to i-carrageenan because, as far as we know, the individual helices are always separate and “ solubIe ” in gels of this polysaccharide.Prof. F. Franks (Unilever) said : The postulate of 8 independent helical segments per polymer chain obviously raises certain topological problems and one might speculate on the nature of a transition state for this process. To ease this situation, I would ask Rees if one might assume that the junction zones are not formed independently but sequentially ? Prof. D. A. Rees (Unilever) said: Yes, for biological reasons the helices must form sequentially and what is more they must form in sequence from the centre-most segments in each chain towards the ends.I can’t conceive of a mechanism by which this can be achieved by kinetic control because each helical segment is nucleated independently. Rather, I propose a thermodynamic reason for the high helix content of the gel. This is simply that the state of high helix content corresponds (at the appropriate temperature) to the minimum free energy and by the making and breaking of helices the system eventually finds this minimum or a state somewhere near it. Good evidence that helices can re-shuffle in this way is provided by the observation P. J. Flory, Proc. Roy. SOC. A , 1956,234,60.GENERAL DISCUSSION 241 that the heterogeneity of chain length of the short carrageenan segments which we describe in our paper, does not lead them to form trimefs or higher aggregates in which, say, one long segment combines with two short segments.(This would be detected in either M, or M, or both). Equally, conformational mdeting is complete (n.m.r. optical rotation) and therefore there must be some mutual “ selection ” process which leads to helix formation between chains of the same chain length. I suggest that his can only happen by re-shuffling of partners and that similar re- shuffling could explain how the topological problems are overcome in formation of the gel structure. Dr. D. S . Reid ( Unilever) said : I have already commented on the dangers associated with the employment of the van’t Hoff procedure for estimating molar enthalpies, in connection with the paper presented by Park. The example contained in my paper illustrates the dangers very well.The use of the van’t Hoff procedure in this case leads 1 0.06 i FIG. 1.-Probability density as a function of chain length i. to an estimate of the molar enthalpy of the coil-double helix reaction which is much smaller than one would expect on the basis of M , and the calorimetric enthalpy of chain association. If a process is considered to be two-state all-or-none, we can define a “ van’t Hoff average ” molecular weight as where Ah is the calorimetric enthalpy of the process, per gram of material. Using the model described in the paper, we have calculated the variation of helical content of a TABLE - - - - - distribution M~H/MH M n l M H ii?wlM~ MnlMw MvHIGn monodisperse 50 50 50 1 1 1 45 50 51 0.98 0.90 2 45 60 64 0.94 0.75 3 44 70 77 0.91 0.63 4 43 77 85 0.91 0.56 5 41 93 1 09 0.85 0.44 MH is the molecular weight of the double helix repeat unit, 1082242 GENERAL DISCUSSION number of hypothetical systems, using the established values for AH," and AS,". These systems had the molecular weight distributions illustrated in fig. 1. The van't Hoff heats for these hypothetical transitions were estimated in the usual way, and hence MVH calculated. In the table, the different molecular weight averages for these distributions are listed. It can be seen that as the polydispersity of the material increases, MVH decreases. Thus, the van't Hoff approach, which ignores polydispersity, sees the broadening due to polydispersity as being due to a reduction in the molecular weight of the species undergoing the all-or-none transition. Clearly, unless a material is known to be monodisperse, estimates of chain lengths, molar heats, etc. based solely on van't Hoff heats should be viewed with suspicion.

ISSN:0301-7249    

DOI:10.1039/DC9745700238

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

N.m.r. of Agarose Gels BY W. DERBYSHIRE * AND I. D. DUFF Department of Physics, University of Nottingham, Nottingham NG7 2RD Received 28th December, 1973 'H spin lattice and spin-spin relaxation together with diffusion rates have been studied in agarose gels as functions of gel concentration and temperature in the ranges 0.9 to 18 % by weight, and from 273 to 323 K. Data have been interpreted in terms of a two phase model where rapid exchange occurs between bulk water molecules and those " bound " to macromolecules. The amount of bound water has been estimated from the amplitude of the signal due to unfrozen water at tempera- tures below the freezing point. The implications of this method of estimating hydration are dis- cussed, and comparisons made with observations on frozen gels,fand on water adsorbed on to agarose.This estimate, 0.59g of water per g of agarose, is satisfactory for an explanation of spin lattice relaxation, but certain anomalies occur when applied to spin-spin relaxation. Interpretation of diffusion in terms of the Wang theory leads to a hydration value of 0.35. Previous estimates of hydration have been 0.1 and 3.5. There is currently considerable interest in the role of water in biological systems. This necessitates an understanding of the interactions between water and various macromo1ecules,1-6 where the macromolecules may be in solution or may have a structural role. Agarose gels provide an interesting, relatively well characterised system where conditions are more than usually under the experimentalist's control. Agarose gels are therefore interesting systems of study in themselves, but they can also be used as a model system for investigating the role of water in foodstuffs where gel systems are frequently encountered, and as a model for muscle systems.A number of n.ni.r. investigations have already been ~ndertaken.~-l The general features are that the n.m.r. relaxation times of the water protons are considerably shorter than those in bulk water. Two explanations have been offered for the occurrence of similar relaxation behaviour in cellular systems. In the first of these, cellular water is envisaged as being more structured than bulk water, having rates of molecular motion intermediate between those of ice and water. 9-20 The second explanation assumes that the structure of water adjacent to the surface is dominated by the presence of that surface, and that the relaxation rates of this surface water are modified.13*18 If a rapid molecular exchange occurs between this surface layer and bulk water, population-weighted average relaxation rates are obtained.Hence the presence of a small proportion of water having an increased relaxation rate can produce a marked reduction in the observed relaxation times. It would be expected that the value of any observable parameter would vary in some reasonably regular manner with distance from the surface, attaining a value characteristic of the bulk liquid at large distances. The simple two phase theories ignore this regular variation and assume that the system can be discussed in terms of two types of water, " bound " and " free ".Within the framework of this theory there is interest in assigning a value to the amount of bound water and in determining its physical properties. It would not be surprising if estimates of hydration based on different techniques were not identical, but it might be hoped that discrepancies would not be too great. 243244 N.M.R. OF AGAROSE GELS Zimmerman and Brittin have developed expressions relating the observable relaxation rates and populations to the real ones for all rates of exchange between two phases. Woessner and Zimmerman 22 have shown that in the situation where a small fraction of molecules exist in a bound state having a much greater relaxation rate than bulk water where Tis the observable relaxation time (TI or T2), and Ta and Tb are the relaxation times of the bulk and bound phases.zb is the lifetime of a water molecule in the bound phase, c is the macromolecular concentration, mass of solute per mass of solvent, and h is the amount of water hydrated by unit mass of macromolecule. A concentration study of relaxation rate is insufficient for an unambiguous determination of h, T b and q,. In an attempt to overcome this problem, Woessner and Snowden 2 s investigated the concentration and temperature dependences of relaxation rates in agarose gels, but, in addition, studied an ensemble of gels of equivalent gel concentration, but where the H,O to D20 ratio in the aqueous com- ponent was varied in a contralled manner. After a somewhat involved discussion the authors extracted a value of h of 0.1.In contrast, Langdon and Thomas,23 using radiotracer techniques to investigate the effects of agarose concentration on ionic transport in agarose gels, obtained a value of 3.6 for the hydration coefficient based QL~ the obstruction effect. The work reported here is concerned with an alternative approach. It is assumed that the physical properties of water adjacent to the macromolecular surface are modified and that in particular this water does not freeze. Signal attenuation upon freezing is used to provide an estimate of the hydration coefficient. This paper is concerned with an investigation of the implications of this choice. EXPERIMENTAL Gels in the range 1 to 18 % by weight were prepared by adding a measured weight of agarose, obtained from BDH Ltd., batch number 1190360 to a known amount of deionised water.The relaxation rates were sufficiently short that degassing was not required. Sealed tubes were heated in boiling water for several hours and allowed to cool slowly. Samples were used not less than twelve hours and within one week of boiling. It was found that reproducible results could be obtained, irrespective of past thermal treatment, provided that samples were reboiled before use. Care was taken to ensure that measurements were made only on samples in the gel and not the sol state. The amount of water present initially in the agarose was estimated by heating samples at 380 K until a constant weight was obtained. The choice of 380 K was a compromise between a temperature sufficiently high for the removal of water of hydration, but not so high as to cause serious dehydroxylation.Measurements of relaxation times were made on a fully phase-coherent spectrometer constructed in this laboratory and operating at a frequency of 10.7 MHz. Spin lattice relaxation was observed by recording the initial height of the signal decay after the 90" pulse in a 180" z 90" pulse sequence as the pulse spacing z was varied. Spin-spin relaxation was observed using a Gill-Meiboom 24 modification of the Carr-Purcell 25 pulse sequence. For T2 measurements the puke spacing was adjusted such that 200 echoes were observed over a time of 10 T2. In practice, z spacings between 400 ps and 4 ms were used, dependent upon the gel concentration. The dependence of the measured T2 upon pulse spacing was investigated over two orders of magnitude of z from 100 to 10 OOO 180" pulses per second, for several agarose concentrations and several temperatures.The purpose of this was to test for the occurrence of exchange between chemically shifted sites,26* 27 or of diffusion through field gradients.2s The measured T2 values showed no z dependence, indicating thatW. DERBYSHIRE AND I . D. DUFF 245 these effects were negligible. Signal strength was enhanced using a DL 102 signal averager manufactured by Data Laboratories and the data transmitted onto a teletype ASR33. Spin-lattice and spin-spin relaxations were observed to be single exponential over two and two and a half orders of magnitude of signal attenuation respectively.Measurements of signal attenuation upon freezing were made by recording the initial height of the free induction decay after a single 90" r.f. pulse. 60 MHz C.W. spectra at temperatures below 0°C showed a narrow line superposed on a broad one. The dead time of the pulse spectro- meter, 2 5 p was inadequate for the observation of the broad, ice component or of the protons in the agarose molecules. The dependence of spectrometer sensitivity upon tempera- ture was calibrated by using a glycerol-water sample that was known not to freeze. The temperature of the sample was controlled by a simple gas flow system providing a tempera- ture stability and homogeneity better than 0.5 K. Diffusion measurements were made using a Bruker pulsed field gradient unit type B-KR 300 Z 18.The amplitude of an echo signal observed after a 9O"z 180" r.f. puke sequence was measured as a function of the area under two equivalent field gradient pulses, one applied between the two r.f. pulses, and the second between the 180" r.f. pulse and the echo.28 By investigating the dependence upon the field gradient pulse separation it is possible to detect the occurrence, within the appropriate time scale, of restricted or barrier limited diffusion. To avoid problems of calibrating the field gradient unit, measurements were made relative to bulk water using similar spectrometer settings. RESULTS AND DISCUSSION SIGNAL ATTENUATION UPON FREEZING The temperature dependence of the normalised signal intensity of a 10 % agarose gel is shown in fig. 1. After freezing there is some evidence of a small reduction in signal intensity only slightly in excess of experimental error.In practice, we chose to neglect this small temperature dependence, and arbitrarily used the signal at 263 K in determining signal attenuation. A graph of relative signal amplitude at 263 K against agarose concentration is linear with a zero intercept, as shown in fig. 2, an observation consistent with expectation. The gradient corresponds to a hydration value of 0.59, almost the geometric mean of the previous estimates, 0.1 l3 and 3.6.23 I 0.0 $03 213 223 233 2L3 253 263 5 3 temp./K FIG. 1 .-The temperature dependence of the normalised signal intensity of the non-freezing component. To check the reasonableness of this procedure a sample was prepared containing 0.59 g of water per g of dry agarose.A small reduction in signal amplitude, = 10 %, did occur on freezing. This was interpreted as a 10 % reduction in hydration value resulting from a change in the secondary (or tertiary) structure of agarose, with the246 N . M . R . OF AGAROSE GELS blocking of some hydration sites, as the agarose concentration increased from 10 to 170 % that of water. Conversely, the linearity of the plot of bound water against concentration shown in fig. 2 indicates that no changes in hydration occur up to agarose concentrations of 18 %. agarose concentration % wt/wt FIG. 2.-The gel concentration dependence of signal attenuation upon freezing. SPIN LATTICE RELAXATION At all temperatures and agarose concentrations studied, graphs of relaxation rate against the parameter hc/(l -hc) were linear, fig.3, again indicating that no changes in hydration behaviour occurred within this concentration range. Fig. 4 shows the temperature dependences of Tl for six of the fourteen concentrations studied. For all samples, with the exception of the lowest concentration, spin lattice relaxation was dominated by the second term of eqn (1). As the residence time q, would be expected to decrease with increasing temperature, the observed increase in TI with temperature is explained by assuming that T 1 b is to the high temperature side of the Tl mini- mum,29* 30 exceeds Tb, and hence rapid exchange occurs. Values of TI, estimated from the intercepts of the lines in fig. 3 are plotted against reciprocal temperature in fig.5. The values obtained are comparable to those measured directly on a sample of water used in preparing the gels. Considering that the values of TI, extracted from the intercepts are subject to relatively large errors, the agreement is considered satisfactory. Values of T l b extracted from the gradients of the concentration plots shown in fig. 3 are shown as curve (iii) and compared with TI values observed on the sample containing 0.59 g of adsorbed water per g of agarose, (curve iv). Considering the possible conformational change in the two different concentration ranges investigated, the agreement must be considered satisfactory and taken as evidence that the estimate of relative hydration as 0.59 is not grossly in error. If the amount of water adsorption is reduced, the relaxation time of the adsorbed water is shortened and the agreement improved, with T 1 b calculated on the basis of a hydration value obtained from signal attenuation upon freezing.As an alternative to estimating h from a freezing experiment, an adsorption study could be performed and the bound layer selected as the maximum amount of water that could be adsorbed that does not exhibit a freezing phenomenon. The objections to this are twofold, firstly that it would be inconvenient in practice, and secondly that the agreement would not be improved, a 10 % reduction in h would result in a corresponding 10 % reduction in the value of T,,,.W. DERBYSHIRE AND I . D . DUFF 247 FIG. 3.-Graph of spin lattice relaxation rate against hc/(l - hc).Whilst the amplitudes of the observable signals of frozen samples were pro- portional to the agarose concentration, the relaxation rates were concentration independent. Both spin lattice and spin spin relaxation were observed to be single 10.5% 16% -- 3.1 3.2 3.3 3.4 3.5 3.6 3.7 103 KIT FIG. 4.-Temperature dependence of spin lattice relaxation for several concentrations.248 N . M . R . OF AGAROSE GELS exponential and a broad Tl minimum was observed, but at the temperature of the Tl minimum the Tl to T2 ratio was 50 and not 1.6,29* 30 and so it must be concluded that whatever motion is responsible for relaxation, that motion is not isotropic and describable by a single correlation frequency. At 263 K the value of Tl was 48 f 2 ms. 3231 3pJ 303 293 a 0 3 273K 3.1 3.2 3.3 U 2 5 3.6 3.7 3.9 1 0 3 ~ 1 ~ FIG.5.-Temperature dependence of the spin lattice relaxation times of the two components. (i) Tla estimated from intercepts of fig. 3. (ii) Tl of deionised water used to prepare gels. (iii) Tlb estimated from gradients of lines in fig. 3. (iv) Tl of adsorbed water. Extrapolation of the T I , to this temperature yields 32&2 ms. Although not con- sistent within experimental error the agreement is comparable to that with the adsorbed water and must again be considered as good. Relaxation times of adsorbed water and of the non-freezable component are consistent within experimental error. In summary, the Tl data of agarose gels may be fitted surprisingly well by a simple two-phase model where the hydration value of 0.59 is obtained from the signal attenuation upon freezing.The model has been tested by a comparison of the spin lattice relaxation behaviour of gels with that of frozen gels and with agarose containing 0.59 g of water per g of agarose. Whilst the agreement is not ideal it is very much better than that which would be obtained if previously published hydration values 0.1 or 3.6 were employed. SPIN-SPIN RELAXATION Fig. 6 shows the temperature dependence of T2 for six of the fourteen gel con- centrations studied. Like spin-lattice relaxation, spin-spin relaxation is dominated by the presence of a bound phase. A significant feature is the observation of a minimum in T2 value at 303 K. Predictions of the two phase model are confirmed, that a T2 minimum will occur when TZb E q,, and that the temperature and value of the T2 minimum will be concentration independent and dependent respectively. This interpretation is similar to that adopted by Woessner and Snowden l3 and by Child et aZ.9-11 The observation by Woessner and Snowden that the temperature of the minimum was dependent upon the H,0--D20 isotopic composition, and thew.DERBYSHIRE AND r. D. DUFF 249 failure of Lillford and Ablett 31 to observe a T2 minimum in the same temperature range using 1 7 0 relaxation, provides support for this model and strong evidence against any explanation based on conformational changes. As in spin lattice ---I- ' 4 3.1 3.2 3.3 3.L 3.5 3.6 3.7 103K/T FIG. 6.-Temperature dependence of spin-spin relaxation for several concentrations.relaxation, concentration plots of spin-spin relaxation were linear within the con- centration and temperature ranges studied (fig. 7). However, the T2 data is anoma- lous in a number of respects. T2, values estimated from intercepts (and therefore - . - I . . . 1 I . J 0 .01 .02 .03 .a! .05 .06 .07 .08 .09 .I0 hc(1- hc)-' FIG. 7.4raphs of spin-spin relaxation rate against hc/( 1 - hc). subject to large errors) are in the range 150 to 300 ms (fig. 8), an order of magnitude shorter than values observed for bulk water. Secondly T2, has a negative, instead of the expected positive, temperature coefficient. The presence of paramagnetic250 N . M . R . OF AGAROSE GELS 300. v) a G 200- impurities have been sought without success by e.s.r. and by chemical methods, but it would be difficult to rationalise an explanation based on this effect with the observation of large Sr, values in agarose S O ~ S .~ ~ " ~ I3 5001 4 00 323K 313 303 203 283 273K 31 3.1 3.2 33 3.4 3.5 3.6 3.7 3.8 100 103K/T FIG. 8.-Calculated temperature dependence of Tza. 9 The phenomenon might possibly be related to that reported by Glasel 32 who reported that nuclei in confined spaces have anomalously low relaxation times. However, we have not investigated this feature in any further detail. The gradients of the concentration plots of fig. 7 yield values of T2b +rb. By performing an iterative calculation it is possible to obtain the separate temperature dependences of T 2 b and q,. Over the temperature ranges in common the spin-spin relaxation times of the adsorbed water are similar to those of the non-freezing component. Neither are consistent with the values estimated for T 2 b from fig.7. At 263 K the T2 values of the adsorbed water and of the non freezing component are 1.2 ms whereas the extrapolated value At temperatures below the T2 minimum, q, > T2b and exchange is therefore slow. Double exponential relaxation behaviour is predicted on the basis of the Zimmerman- Brittin equations, and yet the observed relaxation was single exponential within experimental error. The most favourable conditions for the observation of a doubIe exponential decay occur in the most concentrated gel at the lowest temperature, 273 K. For an 18 % gel the true value of Pb( =hc) is 0.1. At 273 K values of T,, T b and q, are estimated from the various graphs to be 300 ms, 100 p s and 1.7 ms.On in- serting these values into the Zimmerman-Brittin equations the calculated relaxation becomes Of T 2 b is 55 /AS. M ( t ) = Mo { 0.91 exp ( -- ;:)f0.09exP(-;)} where T,' and TL are the apparent relaxation times of 14.7 ms and 94 ps. Within experimental error the relaxation is dominated by the first term, thus explaining the experimental observations. The value for Ti obtained experimentally was 15.2+ 0.2 ms. Another puzzling feature is that, in the unfrozen gels, exchange perpendicular toW. DERBYSHIRE AND I . D. DUFF 25 1 the surface between the bound and bulk phases is slow at temperatures below 303 K. In contrast, in the frozen state where the only unfrozen water resides in a layer adjacent to the macromolecules, and where it is to be expected that a number of different binding sites exist, the observation of a single component T2 relaxation implies that exchange along the surface is rapid, even at a temperature of 190 K.One proposed explanation which was both considered and rejected was the mis- match theory. At room temperature the structure of the agarose repeat unit is compatible with that of an ice-lattice where the lattice spacing is extrapolated from the value at 273 K. The compatibility of structures does not apply at much lower temperatures. It was therefore envisaged that in the frozen gels there would be a region of disorder. Within this region, reorientation and, therefore, diffusion of water molecules could occur readily. One purpose of performing the experiments on adsorbed samples was to test this theory.In adsorbed samples below 273 K there was in one case, only a small amount of ice present and in the second no ice present. Spin-spin relaxation, however, was still single component. If the argument were reversed and the value of h estimated from a T2b value equated with the relaxation times of the non-freezable component, or of the adsorbed water, h would be considerably increased from the value 0.59. SELF DIFFUSION Fig. 9 shows a typical result with the echo attenuation having an exponential dependence upon (gS)2, where g is the amplitude and S the duration of the field gradient pulses. In practice, 6 was maintained constant and the value of the self diffusion coefficient D extracted from the gradient using the expression A = A .exp ( -y2g2d2D(A-+d)). (3) A the echo amplitude is A . in the absence of the field gradient pulses, and A is the separation of the field gradient pulses. The concentration dependence of D is plotted in fig. 10. Such measurements were made at three temperatures and three A 0 10 20 30 (gS)2 arbitrary units FIG. 9.-Dependence of echo amplitude upon the area of the field gradient pulse,252 N . M . R . OF AGAROSE GELS spacings. time scale have reported similar concentration dependences. 22* 33-35 Other workers, mainly using radio tracer techniques which have a different If the two phase model used earlier is applied D = Da(l - hc)+ hcDb. (4) In an alternative model, Jason 36 has suggested that Dby the diffusion in the bound state is controlled by the hopping of water molecules between active sites which have a concentration dependent separation D = Da(l -hC)+PDb,C* (5) where /? is a temperature dependent constant.Unfortunately, errors of measure- ment were such that fig. 10 is inadequate for distinguishing between these models. Certainly on either basis fig. 10 is not compatible with a value of h less than unity. In the limit of Db, = 0, fig. 10 corresponds to a value of h of 1.9, a lower limit. U d 5 10 15 20 0.60 gel concentration/( %) FIG. 10.-Concentration dependence of diffusion coefficient. However, Wang 37 has developed another model, where a diffusing molecule, not necessarily water, encounters barriers in the form of macromolecules with a hydration layer surrounding them.D is predicted to have a linear concentration dependence. On applying the Wang equation, values of the hydration coefficients of 0.35,0.28 and 0.39 were obtained at 273, 293 and 313 K with errors of 0.05. These values are subject to a possible systematic error in that numerical factors, equivalent to a molecular shape factor are incorporated. The order of magnitude agreement between 0.34 and 0.59 is considered to be satisfactory in the circumstances. Support for the use of the Wang model was also obtained from studies of the temperature dependence of the diffusion coefficient in the range 273 to 323 K . The temperature dependence was describable by an Arrhenius equation with an activation energy of 15+2 kJ mol-' similar to that observed in a water sample used in preparation of the gels, and similar to the value obtained by Simpson and Carr 38 using a continuous field gradient technique, and only slightly smaller than the value of 20 kJ mol-1 obtained by Wang using an l 8 0 radiotracer,W.DBRBYSHIRE AND I . D. DUFF 253 In an effort to detect the occurrence of restricted diffusion D/Do was determined for a range of A values between 4 and 46ms. No A dependence was observed. In addition, graphs of echo attenuation had a single exponential dependence upon (gQ2 (A-@) over two orders of magnitude of signal attenuation, indicating that a single diffusion coefficient was appropriate, and confirming that diffusion was not restricted. CONCLUSIONS Some of the experiments reported here have previously been reported by other workers, and where so our measurements are consistent with the earlier work.The two phase model with an estimate of hydration 0.59 based on signal attenuation upon freezing is almost certainly an oversimplification but even so is adequate to explain the spin-spin relaxation data in qualitative terms, and almost adequate to explain the spin-lattice relaxation results quantitatively. However, aspects of the spin-spin relaxation behaviour are confusing and could be an example of a more fundamental phenomenon causing a reduction in the T2 values of nuclei in confined spaces. Diffusion measurements are not explicable in terms of a two phase model with a hydration value of this order, in fact the Wang obstruction theory seems to be more appropriate, yielding a hydration value of 0.36, a reasonable fit on considering the different natures of the techniques and the uncertainties of the Wang treatment.The agreement in the two values is particularly encouraging after the previous spread in reported values of h ; 0.1 to 3.6. We thank S.R.C. for a CAPS award to I. D. D. and acknowledge Unilever Research Laboratories for support, particularly J. Clifford, N. Pryce, P. Lillford and F. Franks for helpful discussions. M. J. Tait and F. Franks, Nature, 1971, 230,91. G. Zundel, Hydration and Intermolecular Interactions (Academic Press, New York, 1969.) L. P. Kayushin, Water in Biological Systems (Consultants Bureau, Plenum, New York, 1969). H. H. G. Jellinek, Water Structure at the Water-Polymer Interface (Plenum, New York, 1972).B. E. Conway, Rev. Macromof. Chem., 1972,7,113. M. Aizawa, J. Mizuguchi, S. Suzuki, S. Hayashi, T. Suzuki, N. Mitomo and H. Toyarno, Bull. Chem. SOC. Japan, 1972,45, 3031. M. Aizawa and S . Suzuki, Buff. Chem. SOC. Japan, 1970,44,2967. J. Clifford and T. F. Child, Proc 1st. European Biophys. Cong., 1971,461. ' J. A. Walter and A. B. Hope, Prog. Biophys. Mol. Biol., 1971, 23, 3. l o T. F. Child and N. G. Pryce, Biopolymers, 1972, 11,409. ' I T. F. Child, N. G. Pryce, M. J. Tait and S . Ablett, Chem. Comm., 1970, 1214. l2 D. E. Woessner, B. S. Snowden Jr. and Y . C. Chiu, J. Colloid Interf. Sci., 1970, 34, 283. D. E. Woessner and B. S. Snowden Jr., J. Colloid Interf. Sci., 1970,34,290. l4 0. Hechter, T. Wittstruck, N. McNiven and C. Lester, Proc. Nat. Acad. Sci., 1960,46, 783. G. Sterling and M. Masuzawa, Makromol. Chem., 1968,116, 140. l6 F. W. Cope, Biophys. J., 1969,9, 303. l 7 C. F. Hazlewood, B. L. Nichols and N. F. Chamberlain, Nature, 1969,222,747. l9 F. W. Cope, Biophys. J., 1969, 9, 303. 2o F. W. Cope, Nature (New BioZogy), 1972, 237, 215. 2 1 J. R. Zimmerman and W. E. Brittin, J. Phys. Chem., 1957, 67, 1328. 2 2 D. E. Woessner and J. R. Zimmerman, J. Phys. Chem., 1963, 67, 1590. 23 A. G. Langdon and H. C. Thomas, J. Phys. Chem., 1971,75, 1821. 24 S. Meiboom and D. Gill, Rev. Sci. Instr., 1958, 29, 688. *' H. Y. Carr and E. M. Purcell, Phys. Rev., 1954, 94, 630. 26 A. Allerhand and H. S. Gutowsky, J. Chem. Phys., 1964,41,2115. 27 Z . Luz and S . Meiboom, J. Chem. Phys., 1963,39, 366. 28 E. 0. Stejskal and J. E. Tanner, J. Chem. Phys., 1965,42,288. R. K. Outhred and E. P. George, Biophys. J., 1973,13,83.254 N.M.R. OF AGAROSE GELS 29 N. Bloembergen, E. M. Purcell and R. V. Pound, Phys. Rev., 1948,73, 679. 30 R. Kubo and K. Tomita, J. Phys. Soc. Japan, 1954,9, 888. 31 P, Lillford and S. Ablett, private communication. 32 J. A. Glasel, Nature, 1970,227, 704, 33 A. L. Slade, A. E. Cremers and H. C. Thomas, J. Phys. Chem., 1966,70,2840. 34 F. S. Nakayama and R. D. Jackson, J. Phys. Chem., 1963,67,932. 35 G. E. Spalding, J . Phys. Chem., 1968, 73, 3380. 36 A. C. Jason, J. Sci. Food. Agric., 1965, 16, 281. 37 J. H. Wang, J. Amer. Chem. SOC., 1954,76,4755. 38 J. H. Simpson and H. Y. Carr, Phys. Reu., 1958, 3, 1201.

ISSN:0301-7249    

DOI:10.1039/DC9745700243

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Rheological and Swelling Properties of Alginate Gels BY A. J. M. SEGEREN, J. V. BOSKAMP AND M. VAN DEN TEMPEL" Unilever Research, Olivier van Noortlaan 120 Vlaardingen, The Netherlands Received 30th November, 1973 The mechanical and swelling properties of alginate gels are studied. Discussion of the results of the dynamic and swelling measurements in terms of the theories of ideal rubberlike elasticity leads to cross-link densities, which are in satisfactory agreement. Stress-strain curves, dynamic measurements on swollen gels and measurements of solution viscosity are used to obtain information about the number of statistical chain elements in a sodium alginate chain. The results show that the number of statistical chain elements in an average strand between cross-links is much smaller than would be required in order to make the theory of ideal rubberlike elasticity applicable.An explanation for the observed " Gaussian " behaviour is suggested, on the basis of a distribution of the '' strength " of the cross-links. A dilute solution of sodium alginate in water can be transformed into a rigid gel by cross-linking with calcium ions. The alginate chain consists of blocks of p-D-mannuronic and blocks of a-L-guluronic acid residues, separated by blocks in which both acids occur in a predominantly alternating fashi0n.l. that a cross-link is formed when a block of twenty or more contiguous guluronic or mannuronic acid residues in a chain is linked to a similar block in another chain by the proper number of calcium ions. Although the chemistry of the cross-linking process in these systems appears to be well established, the situation is less satisfactory with regard to our knowledge of the concentration of cross-links in a given gel, and of the mechanical or thermal stability of the cross-links during loading or heating of the gel.Information of this type can be obtained from rheological and swelling experiments. The results of such experiments on polymer gels are usually interpreted on the basis of the theory of ideal rubberlike elasticity. This theory is more or less valid for gels containing non- ionized polymer in a non-polar solvent. Application of the theory to an aqueous polyelectrolyte gel is of doubtful validity, but has not yet been investigated in detail. It is the purpose of the present investigation to determine how far theories based on the concept of ideal rubberlike elasticity can be used in describing the behaviour of an aqueous polyelectrolyte gel during deformation, swelling and heating.It is believed EXPERIMENTAL Four samples of sodium alginates were used ; their properties are given in table 1. Gels were prepared by mixing a solution of 2.8 g of sample A in 180 g water, containing 2.5 g sodium citrate 2 aq., 3.9 g sucrose and 0.8 g dicalcium orthophosphate, with a solution of 2.5 g citric acid 1 aq. in 20 g water. In the acid mixture, calcium ions are slowly released from the orthophosphate and react with the carboxylic groups of the alginates. The standard gel contains 0.19 mol sodium and 3 x mol/l calcium. Sucrose was added because it is present in many applications.Gel properties were measured in the Weissenberg rheogoniometer, model R18, using either the concentric cylinder or the parallel-plate configuration. The metal surfaces con- tacting the samples were made rough to prevent slip. Swelling was measured by weighing 255256 RHEOLOGICAL AND SWELLING PROPERTIES OF ALGINATE GELS the superficially dried samples after immersion for the required period in the swelling solution. Solution viscosities were measured with an Ubbelohde viscometer. TABLE 1 .-CHARACTERIZATION OF SODIUM ALGINATE SAMPLES sample MIG * M n t A 0.40 1.3 x 105 B 1.1 1.1 x 105 C 1.14 9.3 x 104 D 0.40 4 . o ~ 104 * ratio mannuronic/guluronic acid in chain. t number average molecular weight determined by means of the Mark-Houwink equation, using the osmometric value for sample D as a standard.RESULTS AND DISCUSSION The moduli of the standard gel are almost independent of the frequency over about 2 decades of the time scale (fig. 1). The storage modulus was independent of the amplitude of the deformation up to an amount of shear of 0.1, which was the highest shear that could be obtained in the instrument used, and the ratio G / T was inde- pendent of temperature between 22 and 49°C. These observations, together with the low loss factor G"/G', suggest that we are dealing with a gel that behaves according to the classical theory of rubber elasticity. Under these conditions, the number average molecular weight of the chains between cross-links M, can be obtained from the well-known relation, valid for a network of Gaussian chains : G' = cRT/M..(1) Here, c is the concentration of the polymer in the network (9.4 kg/m3), which was determined by weighing the dried network after swelling for three weeks in pure water. , d l I d 2 10' I d 0 10' frequency 1I-b FIG. I.-Moduli of standard alghate gel at 23°C and amplitude of deformation of 3 x lW3.A . 3. M. SEGEREN, J . V. BOSKAMP AND M. V A N DEN TEMPEL 257 This is supposed to extract suluble compounds and chains that do not form part of the network from the gel. The resulting value of M, is 1.22 x lo4, corresponding with 11 cross-links per molecule of sample A. Further evidence in support of the assumption of Gaussian coils between cross-links is obtained from the stress-strain curve measured at a constant, low rate of shear (fig.2). The initial slope of this curve corresponds with a modulus of 2200 N m-2, which is in satisfactory agreement with the dynamic modulus of fig. 1. The increasing slope at shear values (7) exceeding 0.13 would be due to a finite extensibility of strands between cross-links. Using Treloar’s t h e ~ r y , ~ one finds that the number of statistical chain elements in a chain between cross-links must be less than 2 in order to explain the observed deviation from ideal rubber theory at a shear as small as 0.13. This number, however, is too small even to allow application of ideal rubber theory to the undeformed network. This is the first contradiction in the results of our measurements ; the second will appear from a consideration of the mechanical properties of swollen gels.4 3 m 2 X n I E 0 0 0.2 0.4 0.6 0.8 amount of shear (-) FIG. 2.-Stress-strain curve in torsion of standard alginate gel, measured at a shear rate of 0.0144 s-l at 23°C. The cross-link density in an ideal rubberlike network can also be determined from equilibrium swelling data, provided that the cross-links are not affected by the swelling process. Equilibrium is established if the chemical potential of the solvent in the swelling medium equals that in the swollen gel. The latter is affected by three contri- butions. One contribution to the solvent chemical potential in the gel is due to the mixing free energy of polymer and solvent ; its determination requires knowledge of the Flory-Huggins interaction parameter x.The second contribution is due to the extension of the polymer network, and the third to mixing of solvent and mobile ions. In a first, crude analysis of the results of equilibrium swelling experiments (fig. 3), it was assumed that ionic effects are fully accounted for in the value of the interaction parameter x, determined from intrinsic viscosity data (see below). Tbjs essentially 57-1258 RHEOLOGICAL AND SWELLING PROPERTIES OF ALGINATE GELS means that the effects of ionized constituents on the polymer conformation are assumed to be the same in the swollen gel and in the solution used for viscosity determination. Any further effects of an unequal distribution of charged constituents between the swollen gel and the equilibrium liquid are ignored.Under these conditions, the equilibrium degree of swelling qe must satisfy 2- 0 .- Y E .I3 1 - 8 3 - Here l/q, and I/qo are the volume fractions of polymer in the swollen gel and in the I I I I 0 2 0 40 6 0 80 01 temperature/"C FIG. 3.-Equilibrium swelling ratio in various solvents as a function of temperature ( = distilled water, 0 = 0.005 mol NaCI, 0 = 0.01 mol NaCl, V = 0.02 mol NaCI, A = 0.2 mol NaCI). X I 0 ." 3 3 0.5 0 I I 1 O T ~~ 0 5 I0 15 0 5 10 I5 salt concentration+ FIG. 4.-Intrinsic viscosity of the four alginate samples as a function of the NaCl concentration at various temperatures (0 = 23S°C, 0 = 35°C A = 50°C).A . J . M . SEGEREN, J . v. BOSKAMP AND M . VAN DEN TEMPEL 259 original gel respectively, and u1 is the molar volume of the solvent.This equation can be used, together with the results shown in fig. 3, to obtain the average molecular weight of strands between cross-links after the interaction parameter x has been deter- mined from the intrinsic viscosity data shown in fig. 4. In this graph, the intrinsic viscosity of each of the alginate samples, at a given temperature, has been plotted as a function of the (salt concentration)-*. This is known to result in straight lines. TABLE 2.-cROSS-LINKS PER MOLECULE IN STANDARD GEL (n) DETERMINED FROM EQUILIBRIUM SWELLING (mol/l) x n X n x n - - - 0 . 1 3 20 lo-z 0 . 1 8 1 0 0.20 1 3 0.22 20 0.2 0 . 3 0 1 3 dissolution - - NaCl conc. 23.5/"C 35pc 50l"C 5~ 10-3 - 2x - - 0 . 2 5 1 2 dissolution The value of x to be used in a given swelling experiment was found by selecting the intrinsic viscosities of the four polymers, at the temperature and salt concentration used in that particular swelling experiment, to determine the parameter B of the Stock- mayer-Fixman relation : Here, K is a constant that need not be specified for our purpose, 4o is a universal constant having the value of 2.87 x where p is the density of the polymer.Straight lines were usually obtained in the plots of [q]/M* against M* ; the values of x derived from the slopes of these lines are given in table 2. The table also shows the number of cross-links per molecule of sample A in the swollen gel, calculated by means of eqn (2) from the equilibrium swelling data. The cross-link density shows satisfactory agreement with that ob- tained from modulus measurements.This is surprising, since the Gaussian coil approximation was again used in the second member of eqn (2), and this is expected to be even less valid in a swollen gel than in the original gel. So long as the Gaussian coil approximation for the strands between cross-links is valid, the rigidity of the swollen gel should not differ from that of the original gel. The maximum degree of swelling where an unchanged modulus may be expected can be estimated from the results shown in fig. 2. Here it was found that the shear modulus remained unchanged up to an amount of shear y m = 0.13. The correspond- ing principal extension ratio Anz is 1.067, determined from the relation y = 1- l/A On this basis, one would expect unchanged rigidity up to a swelling ratio qm/qO = 1.: = 1.21.Fig. 5 and 6 show that the modulus starts to increase already at lower swelling ratios. The rise in modulus upon swelling is largely irreversible (cf. fig. 5), suggesting that the phenomenon might be due not only to finite chain extensibility but also to forma- tion of new cross-links. In order to discuss the extensibility of the chains between cross-links, we return to the viscosity measurements of fig. 4. The increasing viscosity at lower salt concentrations is due to an increased rigidity of the chain, and this can be expressed in terms of an increased length L of a statistical chain element. The mean-square end-to-end distance ( r 2 ) of a single polymer molecule in solution is determined by [q]/Mf = K+O.516,BM*. (3) and B = [2/[6 x ~023"o,211[3 -XI, ( r 2 ) = Pb,L, (5)260 RHEOLOGICAL AND SWELLING PROPERTIES OF ALGINATE GELS where P is the degree of polymerization and 6, the length of a monomer unit. The value of (r2> can be estimated from intrinsic viscosity data, using an equation pro- posed by Flory and FOX,’ Values of the length L obtained in this way are between 11.5 and 15 nm at 1 mol and mol salt per litre, respectively. These are of the same order of magnitude as those found by Smid~rrd.~ In the cases where swelling resulted in an increased modulus, the salt concentration in the external solution was lower than in the gel.[ql = 2.1 x 1 0 2 3 ~ 2 ) q ~ . (6) h N k A I I 1 I 1.s 2 1021 swelling ratio FIG. L-Rigidity of swollen gels measured at 23°C and a frequency of 2 Hz (0 = distilled water 0 = 0.01 mol NaCI, A = 0.1 mol NaCI, A = 0.2 mol NaCl ; - - - gel volume reduced to q = qo by drying in air).The resulting reduction of the salt concentration in the swollen gel is indeed seen to cause chain stiffening, and this is partly responsible for the increased modulus after swelling. A statistical chain element of 15 nm length contains about 30 monomer units, and has a molecular weight of about 6000. Taking into account that the average chain between cross-links consists of about 2 statistical chain elements, we find M, = 1.2 x lo4, in satisfactory agreement with results from modulus and equilibrium swelling measurements. The above interpretation of the results of swelling experiments is based on the assumption that the number of cross-links remains unchanged.This is probably not correct. In the initial stages of swelling in a dilute salt solution, the Ca/Na ratio in the gel may be expected to increase due to the ion-exchange capacity of the gel. This makes more calcium available at the surface of the polymer chain, whichA . J . M. SEGEREN, J . V . BOSKAMP AND M. VAN DEN TEMPEL 261 can be used for cross-linking. In later stages, calcium ions will leave the gel towards the surrounding solution ; this ultimately results in liquefaction of the gel, the more rapid if sufficient sodium ions are available for replacing calcium at the polymer surface. Immersion of the gel in the more concentrated salt solution$ caused lique- faction in about 10 days, whereas 3 weeks were required to attain the maximum modulus in the more dilute solutions. The equilibrium swelling data reported in fig.3 relate to a situation where the modulus ofthe swollen gel is beyond the maximum. 3 10 h I E - 2 10 I i -5 swelling ratio Fic. 6.-Rigidity of swollen gels measured at 40°C and a frequency of 2 Hz (0 = distilled water, 0 = 0.005 mol NaCl, 0 = 0.01 mol NaCl, V = 0.02 mol NaCl). The results of this study indicate that the number of statistical chain elements in an average strand between cross-links is much smaller than would be required in order to make the statistical theory of rubber elasticity applicable. Nevertheless, the use of this theory gives reasonable results in the interpretation of modulus and equilibrium swelling measurements. Doubts about the applicability of the theory arise when one considers the rise in modulus at shear values exceeding 0.13, and the large effect of swelling on the rigidity.A possible explanation for this peculiar behaviour will now be suggested. A cross-link consisting of a single calcium ion between two carboxylic acid groups in different chains is insufficient to keep the chains together during deformation. A permanent cross-link requires two or more chains to run parallel over an appreciable distance in an orientation suitable for the formation of a number of calcium bonds. The required number for a permanent cross-link has been estimated to be at least 20. It appears obvious that the actual gel will obtain a distribution of bond lengths, i.e., the number of calcium ions participating in the formation of cross-links will vary262 RHEOLOGICAL AND SWELLING PROPERTIES OF ALGINATE GELS between 1 and, say, 30.The weak bonds, containing only few calcium ions, act like entanglements in the sense that they contribute to the modulus only in small deformations and in measurements of sufficiently short duration (high frequency). On the other hand, breaking of some of the weaker bonds enhances chain mobility, and this may result in the formation of more and stronger bonds by better alignment of neighbouring molecules. This effect appears to predominate at shear values exceed- ing 0.13, and in the initial stages of swelling in dilute salt solutions. The apparent applicability of the Gaussian coil concept to some of our results would then be due to an accidental cancelling of these two effects: breaking of weak bonds and formation of stronger ones. In the same way, the increasing firmness of the geI with tempera- ture can be described. Due to Brownian movement at higher temperatures the weaker bonds will break, giving the chain the possibility of better alignment and resulting in the formation of more and stronger bonds. A more quantitative analysis of a net- work containing mobile bonds would be required in a further analysis of this hypo- thesis. D. A. Rees, Biochem. J., 1972,126,257. 0. Smidsrrd, Curbohyd. Res., 1970, 13, 359. L. R. G. Treloar, The Physics of Rubber Elasticity (Oxford University Press, Oxford, 2nd ed., 1958), chap. VI. A. J. Staverman, Thermodynamics of Polymers, Encyclopedia of Physics, ed. S . Fliigge (Springer, Berlin, 1962), vol. 13, p. 489. D. T. F. Pals and J. J. Hermans, Rec. Truv. Chem., 1952, 71,433. W. H. Stockmayer and M. Fixman, J. Polymer Sci. C, 1963, 1, 137. ' P. J. Flory and T. G. Fox, J. Amer. Chem. SOC., 1951, 73, 1904.

ISSN:0301-7249    

DOI:10.1039/DC9745700255

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Molecular Basis for some Physical Properties of Alginates in the Gel State BY OLAV sMIDSR0D Institute of Marine Biochemistry, 7034 Trondheim-NTH, Norway (Formerly Norwegian Institute of Seaweed Research) Received 3rd December, 1973 Alginate is a binary linear heteropolymer containing 1.4-linked /3-D-mannuronic and or-L-guluronic acid residues in varying proportions. The monomers are arranged in a blockwise fashion along the chain ; it contains the two homopolymeric blocks (MM and GG) together with blocks of the alter- nating sequence (MG) in the same molecule. Data from light scattering and viscometric experiments show that their relative unperturbed dimensions increase in the order : Ion exchange data show that GG-blocks are characterized by a selective binding of calcium ions in solution, a strong autocooperative binding of calcium between the chains in the gel state, and hysterisis due to a slow dissociation of the cooperatively-formed functions between the chains.The two other blocks have much lower selectivity for calcium ions, no autocooperative binding mechanism and no detectable hysteresis. The mechanical strength of calcium alginate gels is mainly due to junctions formed by the GG-blocks. The modulus of rigidity of gels formed by different cations is directly dependent on their ability to bind to the polyuronides by a cooperative inter-chain binding mechanism. Alginates, which comprise the intercellular substance in brown algae occur in the plants as a mixed sodium-magnesium-calcium-strontium salt, the relative proportions of these ions most probably being determined by an ion-exchange equilibrium- reaction with sea-water.l Their function as a structural element in the plant is prob- ably due to their ability to accumulate divalent metal ions and form gels of the re- quired mechanical strength with these ions.Much work at this Institute has been based upon the general idea that the mechani- cal properties of alginate gels are dependent mainly upon (a) the mechanical pro- perties (stiffness of the individual polymer chains and (b) the nature and the strength of the forces whereby divalent metal ions bind different chains together. For under- standing the gel-forming properties of alginates, it was therefore necessary to study both the conformational properties of the isolated molecular chains, by carrying out experiments with dilute solutions, and also their ability to bind divalent metal ions.Results in these two fields are, therefore, presented first, and then the properties of the alginate gels themselves are discussed. However, any understanding of the physical properties of alginates must be based upon a detailed knowledge of their chemistry, and it is therefore necessary to begin by reviewing the relevant facts concerning the fine structure of alginates. MG-blocks < MM-blocks < GG-blocks. CHEMICAL STRUCTURE OF ALGINATES Alginates are composed of 1 +linked residues of #l-D-mannuronic and a-L-guluronio acid in varying Haug et aL9 established that a part of alginate was resistant to hydrolysis in mineral acid, and that this part could be divided into two 263264 ALGINATE GELS fractions, each enriched to more than 90 ”/o in one of the two acids, and both having degrees of polymerization above 20.A non-resistant material of intermediate com- position could be prepared with about the same degree of polymerization lo and was found l2 to contain a large fraction of the two monomers arranged in a strictly alternating sequence. It was thus established that alginate is a block-copolymer con- taining two kinds of homopolymeric blocks (MM-blocks and GG-blocks) together with blocks with an alternating sequence (MG-blocks). Attempts have also been made 3-15 to give a statistical description of the monomer sequence in alginates by applying models similar to tbose used for synthetic addition The stochastic growth of the polymer was simulated by using a set of initial and conditional probabilities, the latter being determined by the identities of one or more preceding units in the growing chain.Both random and non-random degradation of chains built up according to these principles could be simulated, and, by comparison with experimental results for degraded alginate, it was found I4 that its structure resembled a chain in which the conditional probabilities were dependent upon the identities of at least the two preceding units in a growing chain (i.e., according to the “penultimate-unit ” theory of addition copolymerization). However, no experimental results are available for testing whether this type of statistics (a second- order Markov chain) can account for all the structural details in alginate.The distribution of the lengths of the different types of blocks and even their average length is for the same reason not known with any accuracy. In spite of this, the possibility of isolating and studying alginate fragments representative of the three kinds of block, coupled with semi-quantitative m2thods I** l9 for the determination of their relative amounts in alginate samples, suggest that a qualitative understanding of their function in forming gels may be obtained. l7 CONFORMATIONAL PROPERTIES OF ALGINATE I N SOLUTION Since most divalent metal ions except Mg2+ form gels or precipitates with alginate,* the study of its solution properties bas mainly been confined to monovalent cations as the counterion.The main results 20* 21 obtained in aqueous sodium chloride solutions from viscosity and light scattering studies with a sample prepared from Laminaria digitata containing 38.5 % L-guluronic acid residues are given in TABLE I.-VALUES OF THE INDICES K AND a IN THE MARK-HOUWINK EQUATION [ V ] ( d / g ) = KMa AND FOR THE KUHN STATISTICAL SEGMENT LENGTH, A,, AT DIFFERENT IONIC STRENGTHS I = 0.01 4.8 x 1.15 825 I = 0.1 2 . 0 ~ 10-5 1 .o 390 I = 1.0 0.91 x 10-4 0.87 230 I = m 1.2 x 10-4 0.84 155 K a AmIA * *The values of A,,, were calculated from viscosity data for a sample with Mw = 2.7 x lo6 using values of @ in the FIory equation [.I] = @(p)s/M taken from the Bloomfield-Zimm theory.22 table I . It is seen that the extension of the alginate molecule is very dependent upon ionic strength, but even at infinite ionic strength (an extrapolated condition) the molecule is highly extended.By using a modification of the Burchard-Stockmayer- Fixman extrapolation 24 an indication was obtained that the un- perturbed dimensions were only marginally lower than those found at infinite ionic strength, suggesting a high degree of mechanical inflexibility in the polymer chain. By letting the mannuronic acid and the guluronic acid residues adopt the 4C,0. SMIDSROD 265 and 'C, ring conformation, respectively,8 Whittington 2 5 v 26 found by conformational analysis that the hindrance to rotation around the glycosidic linkages in alginate was severely restricted, thus supporting this conclusion. Light scattering and viscosity studies of alginate samples containing different amounts of the three types of blocks have indicated that their relative extension in- creases in the order 27 : MG-blocks < MM-blocks < GG-blocks, both in 0.1 M aqueous sodium chloride and in the unperturbed state.By letting the uronic acids adopt the same ring conformation as above, the same relative order was obtained by conformational analysis.26* 27 Th? incomplete knowledge of the monomer sequence in the different alginates has prevented an experimental deter- mination of the absolute magnitude of the stiffness parameters for the three t y p s of blocks, and a comparison of theoretical and experimental results on an absolute scale has therefore not been possible. Very little is known about the influence of divalent metal ions on the flexibility of the three types of monomer sequence. It has been found that magnesium and sodium alginate of the same molecular weight have the same intrinsic viscosity at infinite ionic strength.21 As will be shown later, the binding of magnesium, calcium, strontium and barium ions by MM-blocks and MG-blocks is similar in strength.Considering the severe hindrance to rotation around the glycosidic linkages in alginate, it is therefore reasonable to believe that the order of flexibility among these two blocks is maintained in the presence of divalent metal ions. It has been indicated 28 that single chain segments of GG-blocks may bind divalent metal ions selectively by a binding site composed of COO-, O(5) and O(4) in one unit and O(2) and O(3) in the preceding ~ n i t .~ ' - ~ O The GG-blocks should therefore be even more stiff in the presence of alkaline earth salts than in sodium chloride solutions. ION BINDING PROPERTIES OF ALGINATES The ion-binding properties of alginates have received considerable attention and have been investigated by a number of experimental techniques. Counterion activities have been measured with permselective rne~nbranes,~~ osmotic and Donnan eq~ilibria,~ cation-sensitive electrode^,^ and by using tetramethylmurexide as an auxiliary ligand.34* 35 The activity coefficient of calcium ions in solutions containing calcium uronates of a certain, low, concentration, was found to decrease with in- creasing degree of polymerization below DP x 20, and to be lower for calcium poly- guluronate than for calcium polymannuronate.Selectivities in exchange-reactions between monovalent and divalent metal ions have been investigated by potentiometric titration 2 * 36* 37 and by ion-exchange equilibrium experiment^.^^'^^ One general conclusion from these studies is that the affinity of alginates for divalent metal ions increases with increasing content of L- guluronic acid residues in the alginates. Most of the recent studies in this laboratory have dealt with exchange reactions between pairs of divalent metal ions 43-45 or pairs of trivalent metal ions 46 using equilibrium dialysis as the experimental technique. The results may be expressed as selectivity coefficients defined by the equation where A and B represents the two cations, Xrepresents the equivalent fraction of the counterions in the polymer phase and C represents the concsntration of the ions in266 ALGINATE GELS solution.Since magnesium ions do not cause gelation in alginate solutions, selec- tivity coefficients in exchange between this ion-type and the other algaline-earth metal ions are of particular interest. Results from calcium-magnesium exchange reactions with fragments approaching the three types of sequence in composition are given in fig. 1. For comparative pur- poses we have found it convenient to plot k g . against Xca instead of using more traditional plotting methods. Increasing values of kf: with X, indicate that the binding of the ion B is associated with a cooperative process 47 ; while decreasing values of k i indicate that the system is heterogeneous with respect to binding 100 so 3 0 igs 10 5 3 I 0 0 .5 1 .o xca FIG. 1 .-The selectivity coefficient k& for aIghate fragments against the equivalent fraction XC, of calcium bound to the polyelectrolyte. Curve 1 : fragment with 90 % G.A., DPn = 50 ; curve 2 : alternating fragment, 38 % G.A., DPn = 20; curve 3 : fragment with 90 % M.A., DPn = 26. 0, Dialysis of the fragments in their sodium form ; x , dialysis first against 0.2 M CaCI2, then against the different mixtures of CaCI2 and MgC12. Total concentration of CaCI, + MgClz = 0.2 M in all dialyzates. Change of dialyzate once a day for 10 days. Concentration of polyuronide 10 mg/ml. It is seen that the GG-blocks are characterized by a significant selectivity for cal- cium ions in solution (k& - 7 for Xca 4 0), a strong autocooperative binding of calcium ions, signs of marked heterogeneity with respect to binding sites, and hys- terisis.The two other fragments have low selectivity and little or no signs of auto- cooperativity and hysterisis. In all three fragments, phase-separation occurs when the concentration of calcium ions in the dialyzates is increased. In a series of experiments the two phases were separated by centrifugation and analysed separately for calcium and magnesium ions. By comparing the results (table 2) with those in fig. 1 it is seen that the increasing values0. SMIDSRBD 267 TABLE 2.-FRACTIONAL PRECIPITATION OF ALGINATE FRAGMENTS AFTER EQUILIBRIUM DIALY SlS AGAINST AQUEOUS SOLUTIONS CONTAINING DIFFERENT MOLAR RATIOS R BETWEEN MAGNESIUM AND CALCIUM IONS Total alginate concentration 1 % (w/v).Total salt concentration 0.2 M (MgCI2+ CaCl,). R2i 400 200 100 66 50 20 100 40 20 3 1 0.1 100 40 20 3 1 0.1 (a) Guluronic-acid rich fragment (90 % G.A., DP, = 50) insoluble fraction soluble fraction total % of total Xca kii! Xca k"M", xcs 4.6 0.017 7.0 0.019 4.6 0.033 6.9 0.036 41.5 0.34 51.2 0.067 7.2 0.18 82.7 0.41 46.6 0.065 4.7 0.35 90.7 0.45 41.4 0.42 94.5 0.65 37.1 0.62 (b) Alternating fragment (38 % G.A., DP, = 20) 5.2 0.03 6.6 0.063 12.2 0.13 73.0 0.60 4.5 0.367 1.8 0.527 75.0 0.79 3.8 0.58 1.4 0.738 86.0 0.977 (c) Mannuronic-acid rich fragment (90 % M.A., DP, = 26) 0.7 0.0 10.5 14.3 0.483 2.8 0.33 1.5 48.0 0.78 3.5 0.58 1.4 82.0 0.95 1.9 0.92 1.2 I 0 I 1 1 1 1 l 1 1 1 l 0.5 XB 0.017 0.038 0.079 0.352 0.677 0.935 I .o k% 7.6 7.4 22.6 36.0 35.5 32.0 3.1 2.7 3.0 3.3 2.8 4.3 1.8 1.6 1.7 1.6 2.1 1.5 FIG.2.-The selectivity coefficient k:, against XB for different alkaline earth ions B for an alginatc. fragment with 90 % M.A., DPn = 26. Curve 1 : B = Ca; curve 2 : B = Sr; curve 3: B = Ba.263 ALGJNATE GELS of kgg with increasing X,, for the GG-blocks is due to an increasing amount of precipitate with a much higher selectivity than the soluble phase. In the two other fragments the selectivities in the two phases are not very different. These results and other evidence 47-49 indicate that the precipitation of the GG-blocks is associated with some regular packing of chain segments enabling a selective binding of ions in sequences between the chains.The hysterisis shown in fig. I indicates that the junc- tions between the chains formed in this way are kinetically very stable towards dissocation. The marked signs of heterogeneity with respect to binding sites may be explained by assuming that the chains are unable to pack into perfectly aligned dimers or other higher-order structures thus giving sites for both inter-chain and intra-chain binding. I0 5 I 0 f i l l I l l I I I 0.5 XB I .o FIG. 3.-The selectivity coefficient kEg against XB for different alkaline earth ions B for an alternating fragment with 38 % G.A., DPn = 20. Curve 1 : B = Ca ; curve 2 : B = Sr ; curve 3 : B = Ba. It was of interest to examine more closely the effect of ion-size in these reactions, and fig. 2 , 3 , 4 and 5 show the behaviour of the three fragments and of pectic acid in the exchange reactions Ca-Mg, Sr-Mg and Ba-Mg.Hysterisis is generally connected with autocooperative inter-chain-binding of metal ions,45 but for the sake of clarity, only the results from experiments associated with the introduction and not with the removal of gel-forming metal ions are included in the figures. A full analysis of the curves in terms of selectivity coefficients for both inter-chain and intra-chain binding, the fractions of these two binding sites, and the hetereo- geneity due to impure fragments, has not yet been carried out. The following order of binding strength for the autocooperative inter-chain binding may, however, be inferred from the results in the figures : GG-blocks MM-blocks MG-blocks Pectate Ba > Sr > Ca 9 Mg, Ba > Sr - Ca - Mg, Ba - Sr - Ca - ME, Ba < Sr - Ca 9 Mg.The exact chemistry and geometry of the inter-chain binding site in different0. SMJDSRPID 269 polyuronides is not known. It has been suggested by several authors 28* 4 5 * 30* 49 that the carboxyl groups, hydroxyl groups and the ring oxygen atoms are all involved in the binding sites. Rees et aZ.49 have recently introduced the term" egg-box-model " iO3OC 300C IOOC 30C mg IOC 3 c IC I I 03 1.0 XB FIG. 4.-The selectivity coefficient k k , against XB for different alkaline earth ions B for a fragment * Due to experimental difficulties because of the high selectivities, no reliable values of k& at XB, < 0.25 have yet been obtained. The stipled curve is drawn based on the facts that incipient gel- formation occur in alginate 42 at XBa < 0.05, and that k z approach 100 at low X ~ ~ - v a l u e ~ .~ ' with 90 % G.A., DPn = 50. Curve 1 : B = Ca; curve 2: B = Sr; curve 3*: B = Ba. to describe the kind of chain-packing in polyuronides that might accommodate ions of different size. In this terminology, the results above suggest that the size of the holes is larger in GG-blocks and MM-blocks than in pectate. Experiments with n.m.r. using the lanthanides as probe ions are currently being carried out at this Institute 2 9 9 5 0 * 51 for a further evaluation of the different binding sites in poly- uronides. The tendency for polyuronide chains to exhibit an autocooperative incorporation of divalent metal ions between chains should also be dependent of the entropic loss associated with the formation of the required regular structure.It is thus interesting270 ALGINATE GELS that the MG-blocks show no signs of autocooperative binding whereas the GG-blocks and pectate (which also is believed to be a very stiff molecule 5 2 * 53) have the strongest tendency for autocooperative binding. FIG. 5.-The selectivity coefficient kzg against XB for different alkaline earth ions B for pectate. Pectic acid bought from Fluka in the spring 1971. According to analysis45 it contains 1.3 % Rhamnose and a total of 10.5 % of neutral sugars, [.I] = 1.5(100 ml/g). Curve 1 : B = Ca ; curve 2 : B = Sr; curve3: B = Ba. ALGINATES AS GELS Experimental data on the mechanical properties of well-defined polyuronide gels have been very scarce in the literature, probably because of the lack of suitable ercperi- mental methods for measuring them.We have found it convenient to determine the stress-strain diagram on small cylindrical pellets prepared by dialysis of sodium polyuronates against solutions of the different metal salts.54 By this method it has been possible to determine the modulus of rigidity as a function of such parameters as concentration, molecular weight and chemical structure of the polyuronides. In addition, the gels could be characterized by their turbidity and the reduction in volume associated with the sol-gel transition. Results from a series of 3 % (w/v) calcium alginate gels are given in fig. 6. It is seen that very high turbidity and very low modulus is typical for polymannuronate, whereas polyguluronate is characterized by very high stiffness and transparency.The calcium polymannuronate may be better described as a volumineous slurry of aggregates rather than a gel. Taking into account the ion-exchange data presented above, the results indicate a requirement for an auto-cooperative incorporation of0. SMIDSROD 27 1 calcium ions between regularly packed chains, for a sol-gel transition to occur. When an increasing amount of blocks containing L-guluronic acid is contained in the alginate, stable junctions seem to be introduced which hinder the formation of aggregates of MM-blocks. The MM-blocks presumably function as single chain segments between junctions in the gels. 100 8 0 2 0 0 2c 40 6 0 8C ! 00 % guluronic acid FIG.6.-Modulus of rigidity, G p 3 , light transmittance, and gel volume relative to sodium alginate, for 3 % (w/v) gels of different chemical composition. Dialysis 2 days against 0.34 M CaC12. Alginates : Fucus vesiculosis receptacles (10 % G.A.), Ascophyllum nodosum (37 % G.A.), Laminaria digitata (30-46 % G.A.), Laminaria hyperburea stipe (64 and 72 % G.A.). Curve 1 : modulus; curve 2 : transmittance for a cell with 2 mm light pass ; curve 3 : volume of gel. In table 3 are given results for two alginate samples having different contents of the alternating sequence. The sample rich in the alternating sequence is characterized by low modulus and high volume. The function of the MG-blocks seems, therefore, to be more to bind water than to form junctions. This may be due to the higher flexibility of this type of sequence.For the same reason one would expect that gels rich in the alternating sequence should be less brittle than other alginate gels, and it is well known by the producers of alginate 5 6 that when jellies which can be deformed to a great extent without breaking are wanted, alginate from Ascophyllum nodosum which contains much of the alternating sequence,l* should be used. The clear correlation between gel-formation and ability for a strong inter-chain binding of calcium ions that is so far indicated seems to be a general phenomenon, as indicated by the results with different alkaline-earth ions given in fig. 7. The following order for the moduli is indicated from the experiments : GG-blocks MM-blocks Pectic acid Ba > Sr > Ca 9 Mg, Ba > Sr - Ca - Mg, Ba < Sr - Ca 9 Mg.These series are identical with those given above for the strength of the inter-chain type of binding of metal ions. This simple, but somewhat surprising, result suggestsT TABLE 3.-cOMPARISON OF THE MODULUS OF RIGIDITY, Gc=3$ FOR AN ALGINATE SAMPLE RICH IN THE ALTERNATiNG SEQUENCE AND A SAMPLE PREPARED FROM Laminaria digitata Dialysis against 0.34 M CaClz ; 3 % (w/v) alginate in the gels. Fractionation results taken from table 2 of ref. (27). gel % G.A. % % % kplcrnz % MG-blocks MM-blocks GG-blocks Gc= 3 x 10 volume Laminaria digitata, Tarva 29/8, [q] = 10, DP, = 2500 38.5 30 41 29 8 .O 55 acid soluble alginate fraction, [q] = 2.0, DP, > 500 35.5 65 25 10 2.5 100 that the single-chain segments in between junctions are very restricted in their move- ment so that the energy applied for compressing the gels is partly used to rupture junctions. To judge the soundness of this hypothesis, some information regarding the chain-structure of the gels is needed.I I I I 0 0 / N 0 3 0‘ I I I I I /[3 \ \ \ \ \ \ \ \ Ca Sr BCl FIG. 7.-Modulus of rigidity, Gc= 3 for different alkaline earth polyuronate gels. Dialysis for 2 days against 0.34 M alkaline earth chlorides. Curve 1 : alginate from Fucus vesiculosusus receptacles (10 % G.A.) ; curve 2 : alginate from Laminaria digitata, 38.5 % G.A. ; curve 3 : pectate from Fluka. In fig. 8 are given results from measurements of the modulus of rigidity as a func- tion of the degree of polymerization of the alginate. The figure shows that a degree of polymerization as low as 65 does give a gel, suggesting that the distance between the junctions must be small, and of the same order of magnitude as the length of theFIG.9.-A 4OO-A section of 4 % alginate gel embedded in Epon. Alginate : Laminaria Iiyperborea, stipe, 64 % G.A., DPww 2 0 0 . To face page 27310. SMIDSR0D 273 Kuhn statistical segment. It is understandable, therefore, that the forces can be transferred through this stiff network structure causing rupture of junctions. I I I I I I I I I I 1 1 '"t 9 x X X X X 1 X I 1 I I I I 1 I I 1 5 0 0 1000 2 0 0 0 3 0 0 0 4000 5 0 0 0 D P W FIG. S.-Modulus of rigidity Gc=3 of alginate gels against weight average degree of polymerization.Dialysis for 2 days against 0.34 M CaCI,. Alginate : Laminaria digitata, 38.5 % G.A. Finally, some indication is needed of the size and number of the junctions them- selves in different types of gels. Thiele 5 7 has shown by microscopy that it is possible to confer diffetent structural features upon an alginate gel by varying the type of metal ions and the conditions whereby they are allowed to diffuse into sodium alginate solutions. Belavtseva et al. have shown by freeze-etching and electron-microscopy that fibrillar structural elements with a diameter of about 20-500A are present in calcium alginate gels. The high turbidity of many of the gels in fig. 6 suggests that some aggregates or other higher-order structures are present in the gels, but the lov turbidity and the small shrinkage in guluronic acid-rich calcium-alginate gels suggest that the junctions contributing to the stiffness of the gels are small in size.Moreover, the linear relationship between the modulus and the square of the alginate concentra- tion observed earlier 5 5 may most easily be explained if the junctions were a result of a bimolecular (bisegmental) crosslinking reaction. We have, therefore, started a re- investigation of the chain structure in alginate gels.59 It has been found that the appearance on the photographs of the structure is critically dependent on the choice of the technique for preparing the sections. A photograph of a guluronic acid-rich gel, prepared first as the calcium salts and then transferred to the lead form by dialysis before embedding in Epon and sectioning, is given in fig.9. In this and many other photographs 6o we have seen no signs of fibrils of diameter 200-500A. The main impression from the photographs is that of a highly crosslinked network of chain molecules, the junctions consisting of only two or a few chain segments. More experimental results are, however, needed to get a clearer understanding of the formation of different chain structures in alginate. A. Haugh and 0. Smidsrprd, Nature, 1967, 215, 1167. A. Haug, Report no. 30 (Norwegian Institute of Seaweed Research, Trondheim, 1%4). E. Percival and R. H. McDowell, Chemistry and Enzymology of Marine Algal Pulysaccharides (Academic Press, London and New York, 1967).274 ALGINATE GELS S. K. Chanda, E. L. Hirst, E.G. V. Percival and A. G. Ross, J. Chem. Soc., 1952, 1833. ' E. D. T. Atkins, W. Mackie and E. E. Smolko, Nature, 1970,225, 626. ti E. D. T. Atkins, W. Mackie, K. D. Parker and E. E. Smolko, J. Polymer Sci. B, 1971,9, 311. ' A. Haug and B. Larsen, Biochim. Biophys. Acta, 1969,192,557. A. Penman and G. R. Sanderson, Carbohyd. Res., 1972,25,273. A. Haug, B. Larsen and 0. Smidsrsd, Acta Chem. Scand., 1966,20, 183. lo A. Haug, B. Larsen and 0, Smidsrsd, Acta Chem. Scand., 1967, 23,691. l1 B. Larsen, 0. Smidsrsd, A. Haug and T. Painter, Acta Chem. Scand., 1969, 23,2375. l2 B. Larsen, 0. Smidsrsd, T. Painter and A. Haug, Acta Chem. Scad., 1970,24, 726. l3 T. Painter, 0. Smidsrsd, B. Larsen and A. Haug, Actu Chem. Scand., 1968,22,1637. l4 B. Larsen, T. Painter, A.Haug and 0. Smidsrsd, Acta Chem. Scand., 1969,23, 355. l5 0. Smidsrsd and S . G. Whittington, Macromolecules, 1969, 2,42. T. Alfrey, J. J. Bohrer and H. Mark, Copolymerization (Interscience Publishers, New York, 1952). G. E. Ham, Copolymerization (Interscience Publishers, New York, 1964). A. Haug, S. Myklestad, B. Larsen and 0. Smidsrsd, Acta Chem. Scand., 1967, 21, 1637. l9 A. Haug, B. Larsen and E. Baardseth, Pruc. 6th Int. Seaweed Symp., Santiago de Compostela 1968 (Pergamon Press, 1969), pp. 443. 2o 0. Smidsrsd and A. Haug, Acta Chem. Scand., 1968,22,797. 0. Smidsrrd, Carbohyd. Res., 1970, 13, 359. 22 V. A. Bloomfield and B. H. Zimm, J. Chem. Phys., 1966,44,315. 23 W. Burchard, Makromol. Chem., 1961, 50,20. 24 W. H. Stockmayer and M. Fixman, J. PoZymer Sci. C, 1963, 1, 137.25 S. G. Whittington, Biopolymers, 1971, 10, 1481. S. G. Whittington, Biopolymers, 1971, 10, 1617. '' 0. Smidsrsd, R. M. Glover and S. G. Whittington, Carbohyd. Res., 1973, 27, 107. 28 0. Smidsrsd, A. Haug and S. G. Whittington, Acta Chem. Scand., 1972,26,2563. 29 T. Anthonsen, B. Larsen and 0. Smidsrsd, Actu Chem. Scad., 1972,26,2988. 30 S. J. Angyal, Pure Appl. Chem., 1973, 35, 131. 31 D. R. Hardistry and S. M. Neale, J. Polymer Sci., 1960, 46, 195. 32 A. Katchalsky, R. E. Cooper, J. Upadhay and A. Wasserman, J. Chem. Soc., 1961, 5198. 33 0. Smidsrsd and A. Haug, J. Polymer Sci. C, 1967, 16,1687. 34 R. Kohn, 1. Furda, A. Haug and 0. Smidsrad, Acta Chem. Scand., 1968,22,3098. 35 R. Kohn and B. Larsen, Acta Chem. Scand., 1972,26,2455. 36 A. Haug, Acta Chem. Scad., 1961,15,1794. 37 A. Haug and 0. Smidsrsd, Acta Chem. Scand., 1970,24, 843. 38 J. L. Mongar and A. Wassermann, Disc. Farday Soc., 1949, 7, 118. 39 J. L. Mongar and A. Wassermann, J. Chem. SOC., 1952, 492. 40 J. L. Mongar and A. Wassermann, J. Chem. Soc., 1952, 500. " A. Haug, Acta Chem. Scand., 1959, 13,1250. 42 A. Haug and 0. Smidsrsd, Acta Chem. Scand., 1965, 19,341. 43 A. Haug and 0. Smidsrsd, Nature, 1967,215,757. 44 0. Smidsrsd and A. Haug, Acta Chem. Scand., 1968,22,1989. 45 0. Smidsrsd, Report No. 34 (Norwegian Institute of Seaweed Research, Trondheim, 1973). 46 G. Lunde, 0. Smidsrsd and A. Haug, Acta Chem. Scad., 1972,26,3421. 47 0. Smidsrsd and A. Haug, Acta Chem. Scand., 1972, 26,2063. 48 E. R. Morris, D. A. Rees and D. Thom, Chem. Comm., 1973,245. 49 G. T. Grant, E. R. Morris, D. A. Rees, P. J. C. Smith and D. Thom, FEBS Letters, 1973, 32, 195. T. Anthonsen, B. Larsen and 0. Smidsrsd, Acta Chem. Scad., 1973, 27, 2671. H. Grasdalen, T. Anthonsen, B. Larsen and 0. Smidsrad, Acta Chem. Scand., in press. 5 2 0. Smidsrsd and A. Haug, Biopolymers, 1971,10, 1213. 53 G. M. Hallman and S. G. Whittington, Macromolecules, 1973, 6, 386. 54 0. Smidsrsd, A. Haug and B. Lian, Acta Chem. Scand., 1972,26,71. 5 5 0. Smidsrsd and A. Haug, Acta Chem. Scad., 1972,26,79. '' R. H. McDowell, Properties of Alginates (Alginate Industries Limited, London, 1961). 57 H. Thiele, Histolyse und Histogenese. Gewebe und ionotrope Gele. Prinzit, einer Strukturbildung 58 Y. M. Belavtseva, V. B. Tolstognzov, D. B. Izymov and M. M. Genia, Biofizika, 1972,17,744. 5 9 0. Smidsrsd and 0. Skipnes in Report no. 34 (Norwegian Institute of Seaweed Research, 6o 0. Skipnes and 0. Smidsrsd, unpublished results. (Munchen, 1967). Trondheim, 1973).

ISSN:0301-7249    

DOI:10.1039/DC9745700263

出版商:RSC    

年代:1974

数据来源: RSC