Vol. 16, 1505-1512 (1977)
BIOPOLY MERS
Calorimetric Investigation of Some ElastinSolvent Systems GIUSEPPINA CECCORULLI, MARIASTELLA SCANDOLA, and GIOVANNI PEZZIN, Centro d i Studio per la Fisica delle Macromolecole del C.N.R. Via Selmi 2, Bologna, Italy Synopsis The interaction of ox ligamenturn nuchae elastin (native, purified, and soluble) with different solvents (water, ethylene glycol, methanol, and 2,2,2-trifluoroethanol) was investigated by means of differential scanning calorimetry. The unfreezable solvent content and the corresponding solvent to elastin residue molar ratio were determined from the melting endotherms of the freezable solvent present in these systems. The molar ratio obtained in the case of water and ethylene glycol agrees with the hypothesis of a direct solvation of the main chain peptide group and is interpreted according to a model previously proposed for polyamides. Quite different molar ratios are obtained in the case of the two monofunctional solvents and no model can be proposed a t present to explain their interaction with the protein.
INTRODUCTION In previous papers1-3 it has been shown that both native and purified elastin, in the dry state, are brittle glassy solids that do not undergo phase transitions from -80 to 300 "C; only a glass-to-rubber transition is revealed a t about 200 "C by calorimetric and dynamic-mechanical measurements. As expected for any polymer-solvent system, the glass transition temperature is strongly dependent on the amount of water or of other solvents present in the protein; in fact, it is lowered to room temperature a t a water content of about 30%and a t a dimethylsulfoxide or methanol content of about 2Ph.l As stated about ten years ago, this lowering of tg is "doubtless the result of the removal of barriers to the rotational and translational motions of the macromolecular chain segments due to their interaction with the solvent" 2; in other words, no rubberlike properties a t room temperature would be exhibited by the swollen elastin unless a direct solvation of the macromolecular chain had taken place. It has been ~ h o w nthat ~ . ~in synthetic hydrogels, as well as in cellulose, only part of the total absorbed water is involved in a direct interaction with the polymer, so that both free and bound water are present in these systems. Similarly, both freezable and unfreezable water have been found in highly hydrated collagen.6 1505 63 1977 by John Wiley & Sons, Inc.
1506
CECCORULLI, SCANDOLA, AND PEZZIN
285
280
275
270
T('K)
Fig. 1. DSC melting endotherms of freezable water in water-purified elastin systems at 0.36; - 0.53;- A - A 0.84; - - - 0.97. different water regains. R:
-. -.
The present paper shows that analogous results are obtained for elastin, i.e., water and other solvents interact with the protein at the molecular level up to a critical amount, above which solvent cluster formation takes place. The amount of unfreezable solvent present in different elastin samples (both native and purified, as well as soluble elastin) has been obtained by the calorimetric determination of the melting heat of the freezable solvent in samples of known total solvent content. The results support the hypothesis of a direct solvation of the protein main chain.
EXPERIMENTAL Samples of native ox ligamentum nuchae elastin and of purified collagen-free elastin were obtained as reported previously.3 Samples of soluble elastin were isolated, according to the procedure suggested by Partridge e t a1.: by treating elastin with oxalic acid, dializing the soluble protein so obtained against water and then recovering it by freeze drying. A 1.2% water solution of the protein was coacervated a t 50 "C for 10 min, and the high-molecular-weightfraction (aelastin) was isolated from the coacervate by dissolution into cold water and freeze drying. The lower-molecularweight fraction p elastin was isolated from the supernatant. The calorimetric measurements were carried out by means of PerkinElmer differential scanning calorimeters, models 1B and 2. Dry samples of about 5-7 mg were placed in the DSC aluminum pans and carefully weighed on an automatic microbalance. Suitable amounts of solvent were
ELASTIN-SOLVENT SYSTEMS
1507
60
-
m
\
4
2
40
3 \
U 20
0
0.3
0.5
0.7
0.9
1.1
R (g/g)
Fig. 2. Plots of the freezable water melting heat, over dry elastin weight (Q/w), against the and purified elastin ( 0 ) . water regain R for native elastin (0)
allowed to be absorbed and the pans were sealed and reweighed to measure the total solvent regain R (solvent to dry protein weight ratio). The samples were stored at room temperature for several days; in order to determine the quantities of freezable and unfreezable solvent, the sample temperature was lowered in the DSC to a t least 50 "C below the melting point of the pure solvent, and then the solvent melting endotherm was recorded a t a heating rate of 8 "C/min. The solvents used were distilled water, ethylene glycol (C.Erba RP) distilled under vacuum in a nitrogen stream, 2,2,2-trifluoroethanol (Schuchardt, Munchen) rectified over sulphuric acid, and methanol (C.Erba RP) rectified over sodium.
RESULTS AND DISCUSSION Typical melting endotherms of water-swollenpurified elastin, at different total regains R , are shown in Figure 1. It is seen that the onset of melting starts a t about -3 "C, and that the narrow peak centered a t about 1"C is followed by a broad shoulder which extends up to 10 "C. A double peak was invariably found in the differential scanning calorimeter (DSC) runs relative to water/elastin systems. Whereas the shape of the melting endotherm was somewhat dependent on the crystallization conditions, the area was constant, so that the heat of melting Q (in calories) of the freezable water could be confidently derived from the area of the melting curve, after
CECCORULLI, SCANDOLA, AND PEZZIN
1508
300
n
200
m
-m \
0
v
P 100
0
Fig. 3. Plots of the freezable water melting heat, over dry elastin weight (Q/w), against the water regain R for soluble (Y elastin ( 0 )and soluble 0 elastin (0).
calibration with a high purity standard. The Q values were plotted as Q/w (were w is the dry protein weight) versus the total regain R. The results obtained for water-swollen purified and native elastin are shown in Figure 2. From a least-square analysis the intercepts with the R axis are found to be Ro = 0.335 for purified and Ro = 0.377 for native elastin, and the slopes are found to correspond to 77.6 and 77.5 cal/g, respectively, in good agreement with the AH measured, using the same procedure, for pure water (AH = 77.2 cal/g). The unfreezable water contents for the soluble a and P elastin were determined, as described above, from the melting peak area of samples whose total water regains ranged from 1.0 to 4.0 (Fig. 3). The results, collected in Table I, show that very close values are found for native, purified, and even for the high-molecular-weight soluble a elastin, indicating that hydration is similar for the three different samples. The slightly higher Ro value shown by the low-molecular-weightsoluble p elastin is probably due to hydration of the additional hydrophilic groups present as chain ends. As far as the mechanism of watedelastin interaction is concerned, it has to be pointed out that elastin has in common with the other proteins and with synthetic polymers such as poly-a-amino acids and polyamides the presence of the hydrophilic peptide group CO-NH in the main chain. The sorption of water by polyamides is explained in terms of a direct solvation of the only hydrophilic site in the macromolecule, the peptide group,
ELASTIN-SOLVENT SYSTEMS
\NH/
0 It C
1509
6 I1 \
/C\NH/
Fig. 4. Model for the water-polyamide interaction (Ref. 8).
whereas the sorption of water by proteins and poly-a-amino acids is a rather complex phenomenon, involving also the solvation of polar side groups. Since elastin has a very peculiar aminoacidic composition with very few polar side groups (less than five residues percent), the suggested site for solvation appears to be the main chain peptide group and a comparison with the solvation of polyamides can be made. In this connection, a specific model for the water/polyamide interaction has been proposed by Puffr and Sebenda8 on the basis of several experimental results including sorption isotherms, sorption heats, ir, and nmr spectra. The model assumes that three water molecules are bound to two neighboring amide groups: the first molecule (firmly bound water) forms a double H bond between the two CO groups, whereas the other two water molecules (loosely bound water) are inserted in the preexistent H bonds between the CO groups and the amide hydrogen atoms, the total bound water in this model being therefore 1.5 moles of water per CO-NH group, as shown in Figure 4. Since elastin has a chain conformation of low order with neither a helix nor other ordered structures sufficiently developed to be recognizable by X rays or ir spectroscopy, it can be practically assumed that all regions of the protein are accessible to the solvent. On the basis of this consideration, the water/elastin molar ratios for the different elastin samples were directly obtained from the respective Ro values (the average molecular weight of the elastin residue being taken as 84).9 It is evident that the unfreezable water in purified and soluble a elastin corresponds to a molar ratio rather close to 1.5 moles of water per mole of elastin residue, i.e., to the molar ratio of Puffr and Sebenda's models for polyamides. It is therefore likely that the hydration of elastin follows a similar mechanism. Other estimates of
1510
CECCORULLI, SCANDOLA, AND PEZZIN
-
T('K)
I
I
I
2 70
260
250
Fig. 5. DSC melting endotherms of freezable ethylene glycol in glycol-purified elastin systems at different glycol regains. R : - - - 1.36;- 1.63;- * - 1.88.
the amount of "bound" and "free" water in elastin are found in the papers of Hoffman and co-workerslOJ1and of Ellis and Packer.12 According to Hoffman's diffusion data, a t 20 "C only 0.1 glg of water are bound (out of a total content of 1.2 g/g). On the other hand, from Ellis and Packer's nmr data the total water content is about 2.2 glg and the bound water about 0.6 g/g at 25 "C. Our unfreezable water results lay in between these two values, and it is difficult to attribute the discrepancies only to the different techniques employed to evaluate the amount of bound water or to the different procedure followed to purify the samples. As far as the other solvents are concerned, Figure 5 shows some of the melting peaks for the purified elastinlethylene-glycol system; the melting range is as broad as about 13 "C, starting a t about -16 "C and showing a shoulder on the low-T side a t the highest solvent regain. The results are summarized in Figure 6 where a Ro of 1.04 and a slope corresponding to an ' enthalpy of fusion of 35.7 callg are obtained from a least-square analysis of the data. This AH value is lower than that reported in the literature13 but it agrees with that measured by us, under the same experimental conditions, for pure ethylene glycol (AH = 35.4 callg). The glycol/elastin molar ratio (see Table I) is again rather close to 1.5; according to this result, the bifunctional glycol molecule seems to interact with elastin in a way similar to that discussed above for water.
ELASTIN-SOLVENT SYSTEMS
1.0
1.2
1.8
1.6
1.4
1511
2.0
R (g/gl
Fig. 6. Plot of the freezable ethylene glycol melting heat, over dry purified elastin weight
(Q/w),against the glycol regain R .
Some preliminary measurements on purified elastin swollen with monofunctional solvents, such as methanol and 2,2,2-trifluoroethanol, have also been carried out. The amount of unfreezable solvent was calculated in these cases from the area of the melting endotherm and from the heat of fusion AH (cal/g) determined directly on the pure solvent. The Ro values so obtained are 0.65 and 3.58 for methanol and trifluoroethanol, respectively, and the corresponding solvent to protein ratios are 1.7 and 3.0. TABLE I Solvation Data of Elastin-Solvent Systems
Protein
Solvent
R oa
Slope (cal/g)
Molar Ratiob
Native Purified Soluble (Y Soluble p
H2O H z0 HzO H20 CHzOH
0.377 0.335 0.342 0.416
77.5 77.6 76.0 75.1
1.56 1.60 1.94
1.04
35.7
1.41
Purified
I
CHzOH a
Unfreezable solvent to dry protein weight ratio. Unfreezable solvent to elastin residue molar ratio.
1512
CECCORULLI, SCANDOLA, AND PEZZIN
Although it can be reasonably assumed that methanol and trifluoroethanol also interact with elastin predominantly at the protein peptide group, at present it is difficult to propose a model of interaction in line with that suggested above for water and able to account for the solvent-elastin molar ratios obtained. In order to get a clear insight into this problem further experimental work is in progress.
References 1. Gotte, L., Pezzin, G. & Stella, G. D. (1966) in Biochimie et Physiologie du Tissu Conjonctif, Comte, P., Ed., Lyon, pp. 145-154. 2. Gotte, L., Mammi, M. & Pezzin, G. (1968) in Symposium on Fibrous Proteins, Crewther, W. G., Ed., Butterworths, Australia, pp. 236-245. 3. Pezzin, G., Scandola, M. & Gotte, L. (1976) Biopolymers 15,283-292. 4. Jhon, M. S. & Andrade, J. D. (1973) J. Biomed. Muter. Res. 7,509-522. 5. Froix, M. F. & Nelson, R. (1975) Macromolecules 8,726-730. 6. Dehl, R. E. (1970) Science 170,73&739. 7. Partridge, S. M., Davis, H. F. & Adair, G. S. (1955) Biochem. J. 61,ll-21. 8. Puffr, R. & Sebenda, J. (1967) J. Polym. Sci. 16,79-93. 9. Mammi, M., Gotte, L. & Pezzin, G. (1968) Nature 220,371-373. 10. Hoffman, A. S. (1971) in Biomaterials, Bement, A. L., Ed., University of Washington, Seattle, pp. 285-312. 11. Mukherjee, D. P., Hoffman, A. S. & Frauzblau, C. (1974) Biopolymers 13, 24472459. 12. Ellis, G. E. & Packer, K. J. (1976) Biopolymers 15,813-832. 13. Handbook of Chemistry and Physics (1971-72) Weast, R. C., Ed., Chemical Rubber, Cleveland.
Received August 26,1976 Revised October 22,1976
BIOPOLY MERS
Calorimetric Investigation of Some ElastinSolvent Systems GIUSEPPINA CECCORULLI, MARIASTELLA SCANDOLA, and GIOVANNI PEZZIN, Centro d i Studio per la Fisica delle Macromolecole del C.N.R. Via Selmi 2, Bologna, Italy Synopsis The interaction of ox ligamenturn nuchae elastin (native, purified, and soluble) with different solvents (water, ethylene glycol, methanol, and 2,2,2-trifluoroethanol) was investigated by means of differential scanning calorimetry. The unfreezable solvent content and the corresponding solvent to elastin residue molar ratio were determined from the melting endotherms of the freezable solvent present in these systems. The molar ratio obtained in the case of water and ethylene glycol agrees with the hypothesis of a direct solvation of the main chain peptide group and is interpreted according to a model previously proposed for polyamides. Quite different molar ratios are obtained in the case of the two monofunctional solvents and no model can be proposed a t present to explain their interaction with the protein.
INTRODUCTION In previous papers1-3 it has been shown that both native and purified elastin, in the dry state, are brittle glassy solids that do not undergo phase transitions from -80 to 300 "C; only a glass-to-rubber transition is revealed a t about 200 "C by calorimetric and dynamic-mechanical measurements. As expected for any polymer-solvent system, the glass transition temperature is strongly dependent on the amount of water or of other solvents present in the protein; in fact, it is lowered to room temperature a t a water content of about 30%and a t a dimethylsulfoxide or methanol content of about 2Ph.l As stated about ten years ago, this lowering of tg is "doubtless the result of the removal of barriers to the rotational and translational motions of the macromolecular chain segments due to their interaction with the solvent" 2; in other words, no rubberlike properties a t room temperature would be exhibited by the swollen elastin unless a direct solvation of the macromolecular chain had taken place. It has been ~ h o w nthat ~ . ~in synthetic hydrogels, as well as in cellulose, only part of the total absorbed water is involved in a direct interaction with the polymer, so that both free and bound water are present in these systems. Similarly, both freezable and unfreezable water have been found in highly hydrated collagen.6 1505 63 1977 by John Wiley & Sons, Inc.
1506
CECCORULLI, SCANDOLA, AND PEZZIN
285
280
275
270
T('K)
Fig. 1. DSC melting endotherms of freezable water in water-purified elastin systems at 0.36; - 0.53;- A - A 0.84; - - - 0.97. different water regains. R:
-. -.
The present paper shows that analogous results are obtained for elastin, i.e., water and other solvents interact with the protein at the molecular level up to a critical amount, above which solvent cluster formation takes place. The amount of unfreezable solvent present in different elastin samples (both native and purified, as well as soluble elastin) has been obtained by the calorimetric determination of the melting heat of the freezable solvent in samples of known total solvent content. The results support the hypothesis of a direct solvation of the protein main chain.
EXPERIMENTAL Samples of native ox ligamentum nuchae elastin and of purified collagen-free elastin were obtained as reported previously.3 Samples of soluble elastin were isolated, according to the procedure suggested by Partridge e t a1.: by treating elastin with oxalic acid, dializing the soluble protein so obtained against water and then recovering it by freeze drying. A 1.2% water solution of the protein was coacervated a t 50 "C for 10 min, and the high-molecular-weightfraction (aelastin) was isolated from the coacervate by dissolution into cold water and freeze drying. The lower-molecularweight fraction p elastin was isolated from the supernatant. The calorimetric measurements were carried out by means of PerkinElmer differential scanning calorimeters, models 1B and 2. Dry samples of about 5-7 mg were placed in the DSC aluminum pans and carefully weighed on an automatic microbalance. Suitable amounts of solvent were
ELASTIN-SOLVENT SYSTEMS
1507
60
-
m
\
4
2
40
3 \
U 20
0
0.3
0.5
0.7
0.9
1.1
R (g/g)
Fig. 2. Plots of the freezable water melting heat, over dry elastin weight (Q/w), against the and purified elastin ( 0 ) . water regain R for native elastin (0)
allowed to be absorbed and the pans were sealed and reweighed to measure the total solvent regain R (solvent to dry protein weight ratio). The samples were stored at room temperature for several days; in order to determine the quantities of freezable and unfreezable solvent, the sample temperature was lowered in the DSC to a t least 50 "C below the melting point of the pure solvent, and then the solvent melting endotherm was recorded a t a heating rate of 8 "C/min. The solvents used were distilled water, ethylene glycol (C.Erba RP) distilled under vacuum in a nitrogen stream, 2,2,2-trifluoroethanol (Schuchardt, Munchen) rectified over sulphuric acid, and methanol (C.Erba RP) rectified over sodium.
RESULTS AND DISCUSSION Typical melting endotherms of water-swollenpurified elastin, at different total regains R , are shown in Figure 1. It is seen that the onset of melting starts a t about -3 "C, and that the narrow peak centered a t about 1"C is followed by a broad shoulder which extends up to 10 "C. A double peak was invariably found in the differential scanning calorimeter (DSC) runs relative to water/elastin systems. Whereas the shape of the melting endotherm was somewhat dependent on the crystallization conditions, the area was constant, so that the heat of melting Q (in calories) of the freezable water could be confidently derived from the area of the melting curve, after
CECCORULLI, SCANDOLA, AND PEZZIN
1508
300
n
200
m
-m \
0
v
P 100
0
Fig. 3. Plots of the freezable water melting heat, over dry elastin weight (Q/w), against the water regain R for soluble (Y elastin ( 0 )and soluble 0 elastin (0).
calibration with a high purity standard. The Q values were plotted as Q/w (were w is the dry protein weight) versus the total regain R. The results obtained for water-swollen purified and native elastin are shown in Figure 2. From a least-square analysis the intercepts with the R axis are found to be Ro = 0.335 for purified and Ro = 0.377 for native elastin, and the slopes are found to correspond to 77.6 and 77.5 cal/g, respectively, in good agreement with the AH measured, using the same procedure, for pure water (AH = 77.2 cal/g). The unfreezable water contents for the soluble a and P elastin were determined, as described above, from the melting peak area of samples whose total water regains ranged from 1.0 to 4.0 (Fig. 3). The results, collected in Table I, show that very close values are found for native, purified, and even for the high-molecular-weight soluble a elastin, indicating that hydration is similar for the three different samples. The slightly higher Ro value shown by the low-molecular-weightsoluble p elastin is probably due to hydration of the additional hydrophilic groups present as chain ends. As far as the mechanism of watedelastin interaction is concerned, it has to be pointed out that elastin has in common with the other proteins and with synthetic polymers such as poly-a-amino acids and polyamides the presence of the hydrophilic peptide group CO-NH in the main chain. The sorption of water by polyamides is explained in terms of a direct solvation of the only hydrophilic site in the macromolecule, the peptide group,
ELASTIN-SOLVENT SYSTEMS
\NH/
0 It C
1509
6 I1 \
/C\NH/
Fig. 4. Model for the water-polyamide interaction (Ref. 8).
whereas the sorption of water by proteins and poly-a-amino acids is a rather complex phenomenon, involving also the solvation of polar side groups. Since elastin has a very peculiar aminoacidic composition with very few polar side groups (less than five residues percent), the suggested site for solvation appears to be the main chain peptide group and a comparison with the solvation of polyamides can be made. In this connection, a specific model for the water/polyamide interaction has been proposed by Puffr and Sebenda8 on the basis of several experimental results including sorption isotherms, sorption heats, ir, and nmr spectra. The model assumes that three water molecules are bound to two neighboring amide groups: the first molecule (firmly bound water) forms a double H bond between the two CO groups, whereas the other two water molecules (loosely bound water) are inserted in the preexistent H bonds between the CO groups and the amide hydrogen atoms, the total bound water in this model being therefore 1.5 moles of water per CO-NH group, as shown in Figure 4. Since elastin has a chain conformation of low order with neither a helix nor other ordered structures sufficiently developed to be recognizable by X rays or ir spectroscopy, it can be practically assumed that all regions of the protein are accessible to the solvent. On the basis of this consideration, the water/elastin molar ratios for the different elastin samples were directly obtained from the respective Ro values (the average molecular weight of the elastin residue being taken as 84).9 It is evident that the unfreezable water in purified and soluble a elastin corresponds to a molar ratio rather close to 1.5 moles of water per mole of elastin residue, i.e., to the molar ratio of Puffr and Sebenda's models for polyamides. It is therefore likely that the hydration of elastin follows a similar mechanism. Other estimates of
1510
CECCORULLI, SCANDOLA, AND PEZZIN
-
T('K)
I
I
I
2 70
260
250
Fig. 5. DSC melting endotherms of freezable ethylene glycol in glycol-purified elastin systems at different glycol regains. R : - - - 1.36;- 1.63;- * - 1.88.
the amount of "bound" and "free" water in elastin are found in the papers of Hoffman and co-workerslOJ1and of Ellis and Packer.12 According to Hoffman's diffusion data, a t 20 "C only 0.1 glg of water are bound (out of a total content of 1.2 g/g). On the other hand, from Ellis and Packer's nmr data the total water content is about 2.2 glg and the bound water about 0.6 g/g at 25 "C. Our unfreezable water results lay in between these two values, and it is difficult to attribute the discrepancies only to the different techniques employed to evaluate the amount of bound water or to the different procedure followed to purify the samples. As far as the other solvents are concerned, Figure 5 shows some of the melting peaks for the purified elastinlethylene-glycol system; the melting range is as broad as about 13 "C, starting a t about -16 "C and showing a shoulder on the low-T side a t the highest solvent regain. The results are summarized in Figure 6 where a Ro of 1.04 and a slope corresponding to an ' enthalpy of fusion of 35.7 callg are obtained from a least-square analysis of the data. This AH value is lower than that reported in the literature13 but it agrees with that measured by us, under the same experimental conditions, for pure ethylene glycol (AH = 35.4 callg). The glycol/elastin molar ratio (see Table I) is again rather close to 1.5; according to this result, the bifunctional glycol molecule seems to interact with elastin in a way similar to that discussed above for water.
ELASTIN-SOLVENT SYSTEMS
1.0
1.2
1.8
1.6
1.4
1511
2.0
R (g/gl
Fig. 6. Plot of the freezable ethylene glycol melting heat, over dry purified elastin weight
(Q/w),against the glycol regain R .
Some preliminary measurements on purified elastin swollen with monofunctional solvents, such as methanol and 2,2,2-trifluoroethanol, have also been carried out. The amount of unfreezable solvent was calculated in these cases from the area of the melting endotherm and from the heat of fusion AH (cal/g) determined directly on the pure solvent. The Ro values so obtained are 0.65 and 3.58 for methanol and trifluoroethanol, respectively, and the corresponding solvent to protein ratios are 1.7 and 3.0. TABLE I Solvation Data of Elastin-Solvent Systems
Protein
Solvent
R oa
Slope (cal/g)
Molar Ratiob
Native Purified Soluble (Y Soluble p
H2O H z0 HzO H20 CHzOH
0.377 0.335 0.342 0.416
77.5 77.6 76.0 75.1
1.56 1.60 1.94
1.04
35.7
1.41
Purified
I
CHzOH a
Unfreezable solvent to dry protein weight ratio. Unfreezable solvent to elastin residue molar ratio.
1512
CECCORULLI, SCANDOLA, AND PEZZIN
Although it can be reasonably assumed that methanol and trifluoroethanol also interact with elastin predominantly at the protein peptide group, at present it is difficult to propose a model of interaction in line with that suggested above for water and able to account for the solvent-elastin molar ratios obtained. In order to get a clear insight into this problem further experimental work is in progress.
References 1. Gotte, L., Pezzin, G. & Stella, G. D. (1966) in Biochimie et Physiologie du Tissu Conjonctif, Comte, P., Ed., Lyon, pp. 145-154. 2. Gotte, L., Mammi, M. & Pezzin, G. (1968) in Symposium on Fibrous Proteins, Crewther, W. G., Ed., Butterworths, Australia, pp. 236-245. 3. Pezzin, G., Scandola, M. & Gotte, L. (1976) Biopolymers 15,283-292. 4. Jhon, M. S. & Andrade, J. D. (1973) J. Biomed. Muter. Res. 7,509-522. 5. Froix, M. F. & Nelson, R. (1975) Macromolecules 8,726-730. 6. Dehl, R. E. (1970) Science 170,73&739. 7. Partridge, S. M., Davis, H. F. & Adair, G. S. (1955) Biochem. J. 61,ll-21. 8. Puffr, R. & Sebenda, J. (1967) J. Polym. Sci. 16,79-93. 9. Mammi, M., Gotte, L. & Pezzin, G. (1968) Nature 220,371-373. 10. Hoffman, A. S. (1971) in Biomaterials, Bement, A. L., Ed., University of Washington, Seattle, pp. 285-312. 11. Mukherjee, D. P., Hoffman, A. S. & Frauzblau, C. (1974) Biopolymers 13, 24472459. 12. Ellis, G. E. & Packer, K. J. (1976) Biopolymers 15,813-832. 13. Handbook of Chemistry and Physics (1971-72) Weast, R. C., Ed., Chemical Rubber, Cleveland.
Received August 26,1976 Revised October 22,1976
