Influence of Solvent Composition on the Mechanical Properties of Arterial Elastin C. P.
WINLOVE and K. H. PARKER
Physiological Flow Studies Unit, Imperial College, London SW7 2AZ, England
SY NOPSlS
We have investigated the effects of changes in solution composition on the mechanical properties of rings of arterial elastin. The time course of force equilibration a t constant strain following a change in the composition of the bathing solution was measured. Both the force developed during slow extension and force relaxation following rapid straining were also measured in each of the test solutions. The results are difficult to summarize because all of the primitive quantities measured-sample dimension, slope of the forceextension curve, force overshoot and time of relaxation-as well as the derived quantities such as elastic modulus changed in different and apparently uncorrelated ways. Changes in pH and ionic composition of the bathing solution had small effects consistent with the low fixed charge density of elastin. Solutions of glucose, sucrose, and ethylene glycol had larger effects consistent with changes in hydrophobic interactions. The viscosity of the solution that penetrated the intrafibrillar space of the elastin appeared to be a major determinant of the dynamic response.
I NTRODU CTI 0N Elastin makes a major contribution to the mechanical properties of ligaments and blood vessels, and there have been numerous studies on the mechanical properties of the isolated protein.‘ Attention has largely been directed towards inferring the structure and organization of the elastin molecule from thermomechanical measurements, and in some of these studies fairly exotic solvent mixtures have been employed. There has, however, been relatively little consideration of the effects of solvent composition per se on mechanical properties. Studies on the effects of solvents are important not only to the clarification of the fundamental elastic mechanisms but also to the understanding of the interactions of elastin with its physiological environment. In addition, it has been suggested that solventmediated changes in the mechanical properties of vascular elastin could be important to its pathology. ‘9
1990 John Wiley & Sons, Inc. CCC 0006-3525/90/040729-07 $04.00 Biopolymers, Vol. 29, 729-735 (1990)
Mukherjee et a1.4 reported that straight chain alkyl sulphates (SDS) and lauric acid reduced the strength and stiffness of ligament elastin while increasing extensibility. These effects were attributed to a reduction in “noncovalent cross link density through disruption of hydrophobic bonds.” However, Bush et a1.’ showed that SDS had the reverse effect in arterial elastin and dismissed the earlier results as artefactual. They also showed that dimethyl sulfoxide (DMSO) had no effect after making a simple correction for changes in tissue volume. Since SDS and DMSO swell elastin to a similar extent, these results eliminated the possibility that changes in elastin hydration were responsible. No definitive alternative explanation was offered, but it was suggested that either the negative charge on SDS or its effect on elastin conformation might be responsible. Following a suggestion’’ that an effect of increased plasma sodium levels on the mechanical properties of elastin may be responsible for some of the changes associated with hypertension, Bush et al., also studied the effect of 0.15 and 0.31M NaCl on the stress strain curves of arterial elastin. The results were negative, although an earlier study5 showed that 6 h exposure to 4M CaCl, 729
730
WINLOVE AND PARKER
doubled stiffness but left breaking strain unaffected. Differences in experimental conditions and the adoption of different definitions of mechanical parameters make it difficult to make comparisons between the two studies, and they underline the need for further work in this area. In the present paper we investigate the effects of both Na' and Ca2+ on the mechanical properties of arterial elastin. We also report measurements over a range of pH in order to investigate the influence of the degree of ionization of the elastin itself. It is widely believed that the existence of large hydrophobic domains in the elastin molecule is more important to its conformation and mechanical properties than its charge structure,6*' and we have therefore also studied its mechanical properties in solutions of glucose and sucrose, which are known to alter hydrophobic interactions in proteins.8 In order to characterize these interactions, we have measured the enthalpies of interaction of these solutes with elastin calorimetrically. The dynamic properties have been less extensively studied than the static ones,g but these too might be sensitive to changes in solution composition. For example, elastin fibers have an internal structure consisting of small hydrated pores" and changes in solution viscosity might influence the rate at which this intrafibrillar water can redistribute. In the present work we characterize the dynamic properties by the stress relaxation following a step increase in strain. By using a range of small neutral solutes with different viscosities but similar enthalpies of interaction with elastin, we are indeed able to demonstrate an effect of solute viscosity. By comparing the effects of the small solutes with those of larger solutes in solutions of comparable bulk viscosity, we are further able to show that intrafibrillar rather than extrafibrillar solution viscosity is important. MATERIALS A N D METHODS
Materials Rings of elastin were prepared from unbranched portions of the descending thoracic aortas of young pigs by the alkali digestion method." This procedure removes tenacious glycoprotein contaminants, but in order to check that mechanical properties were not markedly sensitive to the small amount of end group modification it produces,12a small number of rings were prepared by the milder process of CNBr digestion.13 Some specimens were delipidated by immersion in a 1 : 1: 1 mixture of ethanol,
ether, and acetone overnight and decalcified by soaking in 12.5 m M EDTA a t 37°C for 1 h. All solutions were prepared in deionized water using Analar grade reagents. For the experiments on the effects of changing pH, deionized water was brought to the required pH by small additions of HCl or NaOH, and used immediately. Where appropriate, solution viscosities were measured in a Ostwald viscometer and densities were determined picnometrically. Methods
Mechanical testing was performed on an Instron Model 1122. Rings of elastin approximately 10 mm in width were looped over L-shaped stainless steel supports 2 mm in diameter mounted in the jaws of the Instron and kept immersed in a bath of solution a t room temperature. Two experimental protocols were adopted. In the first, the sample was allowed to equilibrate overnight in the starting solution (usually deionized water) and then mounted in the test cell. The crosshead was then raised until a load of 1 N was recorded. This position was taken as the reference state of the specimen. Force-extension curves were then recorded for an extension of 5 mm ( - 20% strain) at a strain rate of 10 mm/min (preliminary experiments showed similar results for crosshead speeds between 1 and 100 mm/min). Repeat measurements were in general identical, although there was some hysteresis between increasing and decreasing load. The sample was then returned to its reference length and reextended a t a rate of 1000 mm/min. The stress relaxation transient was recorded and when equilibrium had been reached the bathing medium was changed with the sample still extended. The time course of stress reequilibration was followed, and when equilibrium had been attained the sample was returned to its reference length and the sequence of measurements was repeated. The series of solution changes was always completed by a return to the starting solution. A t the end of the experiment the relaxed dimensions of the specimen were measured with calipers and micrometer, and its wet weight and its weight after drying for 48 h a t 55°C were recorded. This protocol had the advantage of eliminating effects of variation between specimens but it was not practical to follow the equilibration of more than a few samples in situ. Groups of specimens were therefore studied under a simpler protocol. Rings of elastin were weighed and measured after 48 h equilibration in deionized water, pH 7.4. Their mechanical properties were measured in the In-
INFLUENCE OF SOI'VENT COMPOSITION
stron with deionized water as the bathing solution. Batches of samples were then immersed in the test solutions for 48 h and retested on the Instron using the same solutions in the bath. Their weights were measured and they were immersed in deionized water overnight in order to remove solutes prior to oven drying for 48 h a t 55°C to determine their dry weights. Because of circumferential variations in the specimens, a small dot of Evans blue dye was applied t o each specimen to ensure th at they were oriented in the same way in successive tests. The mechanical test consisted of extending the specitnen to a force of 1 N and recording the distance hetween the supports. This was taken as the reference length. Th e force developed during a further extension of 5 mm at a rate of 10 mm/min was recorded. T he specimen was allowed to relax a t the same rate and then reextended 5 mm at a rate of 1000 mm/min while recording the force as a function of time until equilibrium was reached. Rings of elastin were employed in the present study to avoid the problems of clamping and slippage that are claimed to have introduced artefacts i t i t o earlier studies.? The possibility existed, however, t ha t errors could arise if a change in solution composition affected the interaction between the elastin and the supports. In preliminary experiments. supports made of different materials with different radii of curvature were employed, and the displac,ement of ink marks on the free and curved surfaces of the elastin was observed during straining. All of these experiments confirmed that the interaction with the supports was constant and negligible. The stress distribution around the supports is complex and makes the interpretation of the elastic properties in absolute terms difficult, but the conclusions of the present work rest on comparative rather than absolute properties. Another source of difficulty in quantitative analysis is the dimensional changes that accompany changes in solution composition. Some authors have corretteti :heir data on the basis of estimated changes in sample cross-sectional area, but this procedure appears to us to make assumptions that are not justified by our present level of understanding of the niechanics of elastin matrices. We therefore make no such correction in the present work. Calorimetry
A flow microcalorimeter (Microscal) was used to measure the enthalpies of interaction of elastin with the various solutions used in the mechanical tests. Approximately 50 mg of dry, ground elastin
731
was packed into the calorimeter chamber and allowed to equilibrate under a constant flow of deionized water. The enthalphy change following a step change of solution from deionized water to the test solution was recorded. The procedure has been described in more detail elsewhere."
RESULTS AND DISCUSSION
The results reported herein are for the alkaliextracted elastin preparation a t the extreme conditions of concentration and pH th a t were tested. Tests were made a t a number of intermediate conditions, but the results were qualitatively similar but smaller. Also, the results obtained for CNBrdigested elastin were indistinguishible from those obtained for the alkali-extracted elastin. Tests before and after delipidating and decalcifying alkaliextracted elastin showed no significant length change ( L / L , , = 0.987 f .013, n = ,5) but a significant increase in slope of the force-extension curve ( S / S , 1.088 .005, n = 5). Illustrative data obtained for a single specimen in different nonionic solutions under the first protocol are shown in Figure 1. The first panel shows that the force-extension curves, which were similar in shape for all of the different solutions, were very linear during extension with a relatively small amount of hysteresis during unloading. The response of the sample to rapid extension (1000 mm/min) is shown in the second panel. Except for sucrose, the sample responded ver?' quickly and with little overshoot in the measured force. In sucrose, however, the overshoot of the force is quite large and the time constant of stress relaxation is on the order of 0.5 s. T h e time course of the measured force at cons ta n t extension after a change of solution is shown in Figure 2 for the same sample. Changing from deionized water, pH 7.4, to glucose caused an increased force. Changing to sucrose caused a further increase in force while a further change to polyethylene glycol (PEG) caused a sharp decrease in force. A further change to deionized water, the original test solution, caused the force to return to its original value. The slight differences in final and initial forces between parts a, b, and c of Figure 2 arise from the various tests performed a t each equilibrium condition. As discussed in the methods section. it was not practical to make a significant number of measurements in different solutions for the same sample
732
WINLOVE AND PARKER
Figure 1. The results of mechanical tests performed on one sample of elastin in different solutions. The left-hand panel shows the force as a function of extension at a crosshead speed of 10 mm/min during loading (upper curve) and unloading (lower curve). S is the slope of the loading curve. The right-hand panel shows the force as a function of time during and after a rapid (1000 mm/min) extension of 5 mm. The solutions are (a) deionized water, pH 7.4; (b) 2M glucose; (c) 2 M sucrose; and (d) 76 mg/mL PEG 2oooO. A 5 mrn extension corresponds to - 20% strain.
2 [
*
1
j
I
I I I
I I I
I
0
1
1
0
1 0 t
1
(h)
Figure 2. The measured force a t a constant extension as a function of time after changing the solution in the test chamber. (a) Deionized water, pH 7.4 to 2 M glucose; (b) 2 M glucose to 2M sucrose; and (c) 2 M sucrose to 76 mg/mL PEG 20000.
INFLUENCE OF SOLVENT COMPOSITION
733
Table I The Effect of Solvent Composition on the Force Per Unit Extension S, Reference Length L, and Elastic Modulus E Solvent
( S - S,),/S, ( W )
PH 2 pH 11
-4.8 f 4.8" 5.9 f 3.6
- 1.7
1M NaCl 1M CaCl
-3.2 f 1.2 4.8 5 3.8
-7.3 5 1.P 0.5 f 3 . ~ 5 ~
2 M glucose 2 M sucrose 2M EG 76 mg/mL PEG
1.4 f 3.7d 7.2 f 5.8 -3.1 ? 1.4 0.3 f 3.gd
( L - Lo)/Lo @)
k 5.6d 14.3 f 5.6
- 11.5 2 2.Eih -7.5 f 6.9 -4.2 f 1.3'
0.6 k 3.4"
( E - &)/Eo ('0 3.3 k 0.8 7.9 k 2.3 -4.7 f 1.1 -4.1 k 0.7 - 10.1
* 0.9"
- 13.7 f 4.0" -1.1 f 0.7 0.3 f 0.9"
"All data are expressed as percentage change from the value measured in deionized water, pH ;.-1 (Mean 5 I SD).The values in deionized water were S,, = 0.471 .OX2 N/mm and L,, = 24.9 i 1.9 mm. See text for a discussion of the calculation of E,). EG, ethylene glycol; PEG, polyethylene glycol 20000. " p < .0001.
'i,
.Os, not significant.
*
=
0.030 t .011 N.
734
WINLOVE AND PARKER
around the circumference, and strictly, the cross section of the solid elastin rather than the whole sample is the appropriate dimension. We have, therefore, used a reference area based upon the wet weight of the sample prior to the measurements, the measured circumference a t a force of 1 N and a mean density, 1.06 g/mL (determined for relaxed elastin samples measured in a separate study by successi;re weighings in air and in water)". The elastic modulus based on this area in deionized water was 0.25 k .03 MPa ( n = 53). As an example of the sensitivity of the elastic modulus to the choice of reference state, the modulus based upon the net weight of the samples after the mechanical tests was 0.36 k .06 MPa ( n = 53). These results compare well with the values of 0.33 MPa reported by Bush et al. for porcine arterial elastin prepared by GuHCl extraction followed by autoclaving,2 and 0.41 MPa obtained by Aaron and Gosline for single autoclaved ligamentum nuchae elastin fibersL6 The final column in Table I shows the percentage change in the elastic modulus, ( E - E o ) /Eo. Both increasing and decreasing pH caused an increase in elastic modulus, while both salt solutions caused a slight decrease. This indicates that ionic interactions do make a contribution to the mechanical properties of the elastin either through direct inter- or intrafibrillar interactions, or through exclusion of solutes from the intrafibrillar medium giving rise to osmotic effects, although as expected from the low fixed charge density on elastin, the effect is small. The decrease in elastic modulus in 1M NaCl is consistent with the observations of Bush et al.,, who reported a small decrease in 0.31M NaC1. The largest changes were seen with the sugar solutions: both 2M glucose and 2M sucrose caused large and statistically highly significant decreases in the elastic modulus. Ethylene glycol had only a small effect and the effect of PEG, albeit a t a much lower monomer concentration, was insignificant. For most of the solutions tested the change in elastic modulus was due primarily to changes in the reference length. Titration curves of elastin,17 which shorn7 an increase in the net charge on elastin for both pH 2 and 11, and the increase in length at high pH, may reflect changes in the balance of electrostatic forces. There is, however, no information about the distribution of charge sites on the elastin fibers that would allow us to analyze the effects in detail. Changing to 1M NaCl caused a significant decrease in circumference, while 1M CaC1, had a small and variable effect.. The insignificant effect of Ca2 was unexpected in view of suggestions that Ca2+ binds specifically a t neutral &
sites and reports that Ca' induces conformational changes in e l a ~ t i n . ' ~ - ~ " The large effect of the sugars on the conformation of the elastin samples could be interpreted in two ways. Nonenzymic glycosyiation, or glycation, of proteins involves a reaction between free amino groups and the carbonyl group on the open chain form of the sugar. This reaction is reversible and has a time constant of hours. Glycation is reported to stiffen isolated ~ollagen.~' Solutes such as s u crose are also known to affect the conformation of many proteins through, it is thought, effects on hydrogen bonding.* The enthalpies of interaction with ground elastin of both sugars were similar (AH = - 16.9 mJ/g dry weight of elastin per mole of glucose and AH = -20.6 mJ/g dry weight of elastin per mole of sucrose) and comparable in magnitude to that expected for hydrophobic interaction.22Both were linear with concentration up to 2 molar and, as far as could be determined, the interaction was reversible as expected of a hydrophobic interaction. Our results could perhaps be taken as a further demonstration of the importance of hydrophobic interactions in elastin mechanic~.~ Another factor that may affect the behavior of the elastin is the accessibility of the protein structure to solutes of different size. Chromatographic"' and experiments have shown that ethylene glycol, glucose, sucrose, and PEG 20000 are excluded to different extents from the intrafibrillar space. This exclusion from parts of the protein matrix would influence the solute-fiber interaction. Another effect of sohte exclusion from the intramolecular space could be the generation of local osmotic gradients. Increasing the osmotic pressure of the bathing solution has been shown to reduce the spacing between collagen f i b e r ~ ~but ~,,~ similar effects would be difficult to demonstrate in the more amorphous structure of elastin. It was difficult to quantify the results of the rapid (1000 mm/min) extension experiments because of transients produced by the mounting assembly. One quantity that could be measured was the "force overshoot," A F, defined as the difference between the maximum force attained and the asymptotic steady state force. The percentage change of the peak force, ( A F - A
WINLOVE and K. H. PARKER
Physiological Flow Studies Unit, Imperial College, London SW7 2AZ, England
SY NOPSlS
We have investigated the effects of changes in solution composition on the mechanical properties of rings of arterial elastin. The time course of force equilibration a t constant strain following a change in the composition of the bathing solution was measured. Both the force developed during slow extension and force relaxation following rapid straining were also measured in each of the test solutions. The results are difficult to summarize because all of the primitive quantities measured-sample dimension, slope of the forceextension curve, force overshoot and time of relaxation-as well as the derived quantities such as elastic modulus changed in different and apparently uncorrelated ways. Changes in pH and ionic composition of the bathing solution had small effects consistent with the low fixed charge density of elastin. Solutions of glucose, sucrose, and ethylene glycol had larger effects consistent with changes in hydrophobic interactions. The viscosity of the solution that penetrated the intrafibrillar space of the elastin appeared to be a major determinant of the dynamic response.
I NTRODU CTI 0N Elastin makes a major contribution to the mechanical properties of ligaments and blood vessels, and there have been numerous studies on the mechanical properties of the isolated protein.‘ Attention has largely been directed towards inferring the structure and organization of the elastin molecule from thermomechanical measurements, and in some of these studies fairly exotic solvent mixtures have been employed. There has, however, been relatively little consideration of the effects of solvent composition per se on mechanical properties. Studies on the effects of solvents are important not only to the clarification of the fundamental elastic mechanisms but also to the understanding of the interactions of elastin with its physiological environment. In addition, it has been suggested that solventmediated changes in the mechanical properties of vascular elastin could be important to its pathology. ‘9
1990 John Wiley & Sons, Inc. CCC 0006-3525/90/040729-07 $04.00 Biopolymers, Vol. 29, 729-735 (1990)
Mukherjee et a1.4 reported that straight chain alkyl sulphates (SDS) and lauric acid reduced the strength and stiffness of ligament elastin while increasing extensibility. These effects were attributed to a reduction in “noncovalent cross link density through disruption of hydrophobic bonds.” However, Bush et a1.’ showed that SDS had the reverse effect in arterial elastin and dismissed the earlier results as artefactual. They also showed that dimethyl sulfoxide (DMSO) had no effect after making a simple correction for changes in tissue volume. Since SDS and DMSO swell elastin to a similar extent, these results eliminated the possibility that changes in elastin hydration were responsible. No definitive alternative explanation was offered, but it was suggested that either the negative charge on SDS or its effect on elastin conformation might be responsible. Following a suggestion’’ that an effect of increased plasma sodium levels on the mechanical properties of elastin may be responsible for some of the changes associated with hypertension, Bush et al., also studied the effect of 0.15 and 0.31M NaCl on the stress strain curves of arterial elastin. The results were negative, although an earlier study5 showed that 6 h exposure to 4M CaCl, 729
730
WINLOVE AND PARKER
doubled stiffness but left breaking strain unaffected. Differences in experimental conditions and the adoption of different definitions of mechanical parameters make it difficult to make comparisons between the two studies, and they underline the need for further work in this area. In the present paper we investigate the effects of both Na' and Ca2+ on the mechanical properties of arterial elastin. We also report measurements over a range of pH in order to investigate the influence of the degree of ionization of the elastin itself. It is widely believed that the existence of large hydrophobic domains in the elastin molecule is more important to its conformation and mechanical properties than its charge structure,6*' and we have therefore also studied its mechanical properties in solutions of glucose and sucrose, which are known to alter hydrophobic interactions in proteins.8 In order to characterize these interactions, we have measured the enthalpies of interaction of these solutes with elastin calorimetrically. The dynamic properties have been less extensively studied than the static ones,g but these too might be sensitive to changes in solution composition. For example, elastin fibers have an internal structure consisting of small hydrated pores" and changes in solution viscosity might influence the rate at which this intrafibrillar water can redistribute. In the present work we characterize the dynamic properties by the stress relaxation following a step increase in strain. By using a range of small neutral solutes with different viscosities but similar enthalpies of interaction with elastin, we are indeed able to demonstrate an effect of solute viscosity. By comparing the effects of the small solutes with those of larger solutes in solutions of comparable bulk viscosity, we are further able to show that intrafibrillar rather than extrafibrillar solution viscosity is important. MATERIALS A N D METHODS
Materials Rings of elastin were prepared from unbranched portions of the descending thoracic aortas of young pigs by the alkali digestion method." This procedure removes tenacious glycoprotein contaminants, but in order to check that mechanical properties were not markedly sensitive to the small amount of end group modification it produces,12a small number of rings were prepared by the milder process of CNBr digestion.13 Some specimens were delipidated by immersion in a 1 : 1: 1 mixture of ethanol,
ether, and acetone overnight and decalcified by soaking in 12.5 m M EDTA a t 37°C for 1 h. All solutions were prepared in deionized water using Analar grade reagents. For the experiments on the effects of changing pH, deionized water was brought to the required pH by small additions of HCl or NaOH, and used immediately. Where appropriate, solution viscosities were measured in a Ostwald viscometer and densities were determined picnometrically. Methods
Mechanical testing was performed on an Instron Model 1122. Rings of elastin approximately 10 mm in width were looped over L-shaped stainless steel supports 2 mm in diameter mounted in the jaws of the Instron and kept immersed in a bath of solution a t room temperature. Two experimental protocols were adopted. In the first, the sample was allowed to equilibrate overnight in the starting solution (usually deionized water) and then mounted in the test cell. The crosshead was then raised until a load of 1 N was recorded. This position was taken as the reference state of the specimen. Force-extension curves were then recorded for an extension of 5 mm ( - 20% strain) at a strain rate of 10 mm/min (preliminary experiments showed similar results for crosshead speeds between 1 and 100 mm/min). Repeat measurements were in general identical, although there was some hysteresis between increasing and decreasing load. The sample was then returned to its reference length and reextended a t a rate of 1000 mm/min. The stress relaxation transient was recorded and when equilibrium had been reached the bathing medium was changed with the sample still extended. The time course of stress reequilibration was followed, and when equilibrium had been attained the sample was returned to its reference length and the sequence of measurements was repeated. The series of solution changes was always completed by a return to the starting solution. A t the end of the experiment the relaxed dimensions of the specimen were measured with calipers and micrometer, and its wet weight and its weight after drying for 48 h a t 55°C were recorded. This protocol had the advantage of eliminating effects of variation between specimens but it was not practical to follow the equilibration of more than a few samples in situ. Groups of specimens were therefore studied under a simpler protocol. Rings of elastin were weighed and measured after 48 h equilibration in deionized water, pH 7.4. Their mechanical properties were measured in the In-
INFLUENCE OF SOI'VENT COMPOSITION
stron with deionized water as the bathing solution. Batches of samples were then immersed in the test solutions for 48 h and retested on the Instron using the same solutions in the bath. Their weights were measured and they were immersed in deionized water overnight in order to remove solutes prior to oven drying for 48 h a t 55°C to determine their dry weights. Because of circumferential variations in the specimens, a small dot of Evans blue dye was applied t o each specimen to ensure th at they were oriented in the same way in successive tests. The mechanical test consisted of extending the specitnen to a force of 1 N and recording the distance hetween the supports. This was taken as the reference length. Th e force developed during a further extension of 5 mm at a rate of 10 mm/min was recorded. T he specimen was allowed to relax a t the same rate and then reextended 5 mm at a rate of 1000 mm/min while recording the force as a function of time until equilibrium was reached. Rings of elastin were employed in the present study to avoid the problems of clamping and slippage that are claimed to have introduced artefacts i t i t o earlier studies.? The possibility existed, however, t ha t errors could arise if a change in solution composition affected the interaction between the elastin and the supports. In preliminary experiments. supports made of different materials with different radii of curvature were employed, and the displac,ement of ink marks on the free and curved surfaces of the elastin was observed during straining. All of these experiments confirmed that the interaction with the supports was constant and negligible. The stress distribution around the supports is complex and makes the interpretation of the elastic properties in absolute terms difficult, but the conclusions of the present work rest on comparative rather than absolute properties. Another source of difficulty in quantitative analysis is the dimensional changes that accompany changes in solution composition. Some authors have corretteti :heir data on the basis of estimated changes in sample cross-sectional area, but this procedure appears to us to make assumptions that are not justified by our present level of understanding of the niechanics of elastin matrices. We therefore make no such correction in the present work. Calorimetry
A flow microcalorimeter (Microscal) was used to measure the enthalpies of interaction of elastin with the various solutions used in the mechanical tests. Approximately 50 mg of dry, ground elastin
731
was packed into the calorimeter chamber and allowed to equilibrate under a constant flow of deionized water. The enthalphy change following a step change of solution from deionized water to the test solution was recorded. The procedure has been described in more detail elsewhere."
RESULTS AND DISCUSSION
The results reported herein are for the alkaliextracted elastin preparation a t the extreme conditions of concentration and pH th a t were tested. Tests were made a t a number of intermediate conditions, but the results were qualitatively similar but smaller. Also, the results obtained for CNBrdigested elastin were indistinguishible from those obtained for the alkali-extracted elastin. Tests before and after delipidating and decalcifying alkaliextracted elastin showed no significant length change ( L / L , , = 0.987 f .013, n = ,5) but a significant increase in slope of the force-extension curve ( S / S , 1.088 .005, n = 5). Illustrative data obtained for a single specimen in different nonionic solutions under the first protocol are shown in Figure 1. The first panel shows that the force-extension curves, which were similar in shape for all of the different solutions, were very linear during extension with a relatively small amount of hysteresis during unloading. The response of the sample to rapid extension (1000 mm/min) is shown in the second panel. Except for sucrose, the sample responded ver?' quickly and with little overshoot in the measured force. In sucrose, however, the overshoot of the force is quite large and the time constant of stress relaxation is on the order of 0.5 s. T h e time course of the measured force at cons ta n t extension after a change of solution is shown in Figure 2 for the same sample. Changing from deionized water, pH 7.4, to glucose caused an increased force. Changing to sucrose caused a further increase in force while a further change to polyethylene glycol (PEG) caused a sharp decrease in force. A further change to deionized water, the original test solution, caused the force to return to its original value. The slight differences in final and initial forces between parts a, b, and c of Figure 2 arise from the various tests performed a t each equilibrium condition. As discussed in the methods section. it was not practical to make a significant number of measurements in different solutions for the same sample
732
WINLOVE AND PARKER
Figure 1. The results of mechanical tests performed on one sample of elastin in different solutions. The left-hand panel shows the force as a function of extension at a crosshead speed of 10 mm/min during loading (upper curve) and unloading (lower curve). S is the slope of the loading curve. The right-hand panel shows the force as a function of time during and after a rapid (1000 mm/min) extension of 5 mm. The solutions are (a) deionized water, pH 7.4; (b) 2M glucose; (c) 2 M sucrose; and (d) 76 mg/mL PEG 2oooO. A 5 mrn extension corresponds to - 20% strain.
2 [
*
1
j
I
I I I
I I I
I
0
1
1
0
1 0 t
1
(h)
Figure 2. The measured force a t a constant extension as a function of time after changing the solution in the test chamber. (a) Deionized water, pH 7.4 to 2 M glucose; (b) 2 M glucose to 2M sucrose; and (c) 2 M sucrose to 76 mg/mL PEG 20000.
INFLUENCE OF SOLVENT COMPOSITION
733
Table I The Effect of Solvent Composition on the Force Per Unit Extension S, Reference Length L, and Elastic Modulus E Solvent
( S - S,),/S, ( W )
PH 2 pH 11
-4.8 f 4.8" 5.9 f 3.6
- 1.7
1M NaCl 1M CaCl
-3.2 f 1.2 4.8 5 3.8
-7.3 5 1.P 0.5 f 3 . ~ 5 ~
2 M glucose 2 M sucrose 2M EG 76 mg/mL PEG
1.4 f 3.7d 7.2 f 5.8 -3.1 ? 1.4 0.3 f 3.gd
( L - Lo)/Lo @)
k 5.6d 14.3 f 5.6
- 11.5 2 2.Eih -7.5 f 6.9 -4.2 f 1.3'
0.6 k 3.4"
( E - &)/Eo ('0 3.3 k 0.8 7.9 k 2.3 -4.7 f 1.1 -4.1 k 0.7 - 10.1
* 0.9"
- 13.7 f 4.0" -1.1 f 0.7 0.3 f 0.9"
"All data are expressed as percentage change from the value measured in deionized water, pH ;.-1 (Mean 5 I SD).The values in deionized water were S,, = 0.471 .OX2 N/mm and L,, = 24.9 i 1.9 mm. See text for a discussion of the calculation of E,). EG, ethylene glycol; PEG, polyethylene glycol 20000. " p < .0001.
'i,
.Os, not significant.
*
=
0.030 t .011 N.
734
WINLOVE AND PARKER
around the circumference, and strictly, the cross section of the solid elastin rather than the whole sample is the appropriate dimension. We have, therefore, used a reference area based upon the wet weight of the sample prior to the measurements, the measured circumference a t a force of 1 N and a mean density, 1.06 g/mL (determined for relaxed elastin samples measured in a separate study by successi;re weighings in air and in water)". The elastic modulus based on this area in deionized water was 0.25 k .03 MPa ( n = 53). As an example of the sensitivity of the elastic modulus to the choice of reference state, the modulus based upon the net weight of the samples after the mechanical tests was 0.36 k .06 MPa ( n = 53). These results compare well with the values of 0.33 MPa reported by Bush et al. for porcine arterial elastin prepared by GuHCl extraction followed by autoclaving,2 and 0.41 MPa obtained by Aaron and Gosline for single autoclaved ligamentum nuchae elastin fibersL6 The final column in Table I shows the percentage change in the elastic modulus, ( E - E o ) /Eo. Both increasing and decreasing pH caused an increase in elastic modulus, while both salt solutions caused a slight decrease. This indicates that ionic interactions do make a contribution to the mechanical properties of the elastin either through direct inter- or intrafibrillar interactions, or through exclusion of solutes from the intrafibrillar medium giving rise to osmotic effects, although as expected from the low fixed charge density on elastin, the effect is small. The decrease in elastic modulus in 1M NaCl is consistent with the observations of Bush et al.,, who reported a small decrease in 0.31M NaC1. The largest changes were seen with the sugar solutions: both 2M glucose and 2M sucrose caused large and statistically highly significant decreases in the elastic modulus. Ethylene glycol had only a small effect and the effect of PEG, albeit a t a much lower monomer concentration, was insignificant. For most of the solutions tested the change in elastic modulus was due primarily to changes in the reference length. Titration curves of elastin,17 which shorn7 an increase in the net charge on elastin for both pH 2 and 11, and the increase in length at high pH, may reflect changes in the balance of electrostatic forces. There is, however, no information about the distribution of charge sites on the elastin fibers that would allow us to analyze the effects in detail. Changing to 1M NaCl caused a significant decrease in circumference, while 1M CaC1, had a small and variable effect.. The insignificant effect of Ca2 was unexpected in view of suggestions that Ca2+ binds specifically a t neutral &
sites and reports that Ca' induces conformational changes in e l a ~ t i n . ' ~ - ~ " The large effect of the sugars on the conformation of the elastin samples could be interpreted in two ways. Nonenzymic glycosyiation, or glycation, of proteins involves a reaction between free amino groups and the carbonyl group on the open chain form of the sugar. This reaction is reversible and has a time constant of hours. Glycation is reported to stiffen isolated ~ollagen.~' Solutes such as s u crose are also known to affect the conformation of many proteins through, it is thought, effects on hydrogen bonding.* The enthalpies of interaction with ground elastin of both sugars were similar (AH = - 16.9 mJ/g dry weight of elastin per mole of glucose and AH = -20.6 mJ/g dry weight of elastin per mole of sucrose) and comparable in magnitude to that expected for hydrophobic interaction.22Both were linear with concentration up to 2 molar and, as far as could be determined, the interaction was reversible as expected of a hydrophobic interaction. Our results could perhaps be taken as a further demonstration of the importance of hydrophobic interactions in elastin mechanic~.~ Another factor that may affect the behavior of the elastin is the accessibility of the protein structure to solutes of different size. Chromatographic"' and experiments have shown that ethylene glycol, glucose, sucrose, and PEG 20000 are excluded to different extents from the intrafibrillar space. This exclusion from parts of the protein matrix would influence the solute-fiber interaction. Another effect of sohte exclusion from the intramolecular space could be the generation of local osmotic gradients. Increasing the osmotic pressure of the bathing solution has been shown to reduce the spacing between collagen f i b e r ~ ~but ~,,~ similar effects would be difficult to demonstrate in the more amorphous structure of elastin. It was difficult to quantify the results of the rapid (1000 mm/min) extension experiments because of transients produced by the mounting assembly. One quantity that could be measured was the "force overshoot," A F, defined as the difference between the maximum force attained and the asymptotic steady state force. The percentage change of the peak force, ( A F - A
