Physical and Chemical Properties of Rubber-Like Elastic Fibres From the Octopus Aorta

Compliance and elastic recoil in the walls of major arteries provide an important pulse smoothing mechanism for the high pressure closed circulatory systems of vertebrates. The elasticity of vertebrate arteries at physiological pressures is based primarily on elastin, a fibrous connective tissue protein with rubber-like properties (McDonald, 1974). Although elastin is the only arterial elastomer which has been described in detail, histological reports of ‘elastic’ fibres in the blood vessels of some molluscs, crustaceans and annelid worms (Elder & Owen, 1967; Elder, 1973) suggest that at least some of the more advanced invertebrate circulatory systems contain an elastic component. However, these invertebrate ‘elastic’ fibres cannot be elastin because elastin occurs exclusively among the vertebrates (Sage & Gray, 1979, 1980). Thus, elasticity of invertebrate blood vessels may be based on rubber-like proteins which have not yet been described.

To determine if any elastin analogues are present in invertebrate blood vessels, we began to investigate the connective tissue architecture and mechanical properties of the arteries of cephalopod molluscs. In our initial study we found that the octopus aorta is a compliant and elastic tube, and that the vessel wall contains connective tissue fibres that are highly extensible (Shadwick & Gosline, 1981). These fibres stain with aldehyde fuchsin (an ‘elastic’ fibre stain) and are present throughout the artery wall and also as a layer that lines the vessel lumen (Shadwick & Gosline, 1983). Preliminary experiments with fibres isolated from the luminal layer of the octopus aorta provided evidence that this material is an insoluble protein with rubber-like properties (Shadwick & Gosline, 1981). The current study extends these observations, and presents a quantitative analysis of the physical and chemical properties of the octopus elastic fibre protein, based on the theory of rubber elasticity.

The kinetic theory of rubber elasticity is a useful mathematical description of the mechanical properties of highly extensible elastic polymers (see Treloar, 1975, for a complete description). Briefly, this theory assumes that: (1) the material is composed of flexible, linear polymer molecules, (2) these molecules are randomly coiled, so that the end-to-end separation of each molecular chain is much smaller than its fully extended length, (3) the molecular chains are kinetically free and in constant motion due to thermal agitation and (4) the chains are joined together by permanent crosslinks to form an isotropic three-dimensional network.

The kinetic theory treats the individual chains as ‘ideal’ random walks. Each polymer chain is considered to be a series of random links, where each link has complete rotational freedom with respect to its neighbours. This allows the whole chain to take on the statistically most probable conformations. When an ideal network is mechanically deformed the chains are forced to adopt new conformations of lower probability. In other words, the conformational entropy of the network is decreased with increasing deformation, and an elastic force arises from the tendency of these random-coil chains to return spontaneously to the state of highest entropy (i.e. the undeformed state).

If the polymer chains in a real rubber network are able to change their conformation with little or no distortion of the chemical bonds in the polymer backbone, then elastic energy will be stored in the network as a decrease in conformational entropy, with little or no change in internal energy. However, there are some important differences between real polymer molecules and ideal random chains. Because there are physical restrictions to rotation around the covalent bonds in the backbone of a polymer, it is not possible for an individual bond to exhibit the complete rotational freedom assumed to occur between the links in an ideal random chain. Thus, in the application of the kinetic theory to real molecules it is necessary to replace the concept of an ideal random link with a ‘functional’ random link. This functional link may be defined as a length of molecule which contains enough bonds with partial rotation so that the segment has essentially the same kinetic freedom as an ideal random link.

The Gaussian model holds only for relatively small extensions. As a real network is deformed the chains are pulled out from their coiled conformations, and eventually the network becomes stiffer than predicted by the theory. Treloar (1975) developed a non-Gaussian model which accounts for the finite extensibility of chains in real networks. The shape of the non-Gaussian force-extension curve is highly dependent on the number of random links (n) in each network chain. Decreasing n causes a greater deviation from the Gaussian curve, for which n is infinite. By comparing the force-extension curve for a real network with the non-Gaussian model (see Treloar, 1975, equation 6.22) we can estimate the number of functional random links in the network chains.

These relationships provide a simple method (known as the thermoelastic experiment) for determining the entropy and enthalpy components of the elastic force for any material, under conditions of constant pressure. However, the kinetic theory of rubber elasticity predicts specifically that the change in internal energy will be zero when the network is deformed at constant volume. Thus, only under conditions of constant volume will the enthalpy change be equal to the internal energy change, and the entropy component will be due solely to changes in the conformational entropy of the polymer chains. For water-swollen protein rubbers the distinction between testing at constant pressure and constant volume may be important if the volume of the protein network varies with temperature. If so, then (∂H/∂L)_T,P will not be just the internal energy change associated with the conformation of the molecule, but it will also have an energy component associated with the mixing of water and the protein; similarly (∂S/∂L)_T,P will also have a component due to this mixing process (Gosline, 1980).

The dorsal aorta was dissected from specimens of Octopus dofleini that had been cold-anaesthetized and killed by decerebration. Tissues were minced, blotted and weighed before and after drying at 60°C to determine the water content. Extraction and purification of elastic connective tissue from the artery wall was based on methods for elastin (Steven, Minns & Thomas, 1974; Rasmussen, Bruenger & Sandberg, 1975; Sage & Gray, 1977) which assume that the material is covalently crosslinked into an insoluble network. Dry aortae were ground in a glass tissue homogenizer and then extracted with continuous agitation in 98% formic acid at room temperature for periods of either 6, 18, 24, 48 or 72h. After centrifugation at 12 000 g for 20 min the pellets were resuspended and washed with distilled water several times. The residues were dried at 60°C to a constant weight. As an alternative method aortae were extracted in 6 mol l⁻¹ guanidine HCl +1% mercaptoethanol for 48 h with continuous agitation, followed by centrifugation, washing in water, and autoclaving for 4h at 1 atmosphere pressure and 120°C.

Other samples of aorta were minced, weighed and then washed for 24 h in several volumes of distilled water. This was to remove soluble salts, proteins and carbohydrates. The insoluble material was separated by centrifugation (15 min at 5000 g) and autoclaved in distilled water for 6h to solubilize collagen and precipitate other proteins. After centrifugation for 20 min at 12000 the supernatant was passed through a l-ptm Millipore filter, and made up to a known volume with water. This autoclave-soluble fraction was assumed to be a crude collagen extract. The total dry weight of this fraction was determined by drying aliquots of the solution. The amount of collagen in the artery wall was then estimated by comparing the amino acid composition of the collagen preparation with published data for purified octopus collagen.

Fresh aortae were cut open longitudinally and pinned out flat on wax under filtered sea water. With the use of a stereo microscope and fine forceps native elastic fibres were stripped from the internal elastic layer which lines the vessel lumen (Shadwick & Gosline, 1983). Some fibres were used for mechanical tests immediately, while others were stored either frozen or at 4°C in a 0-01% solution of penicillin + streptomycin. Some fibres were exposed to 98% formic acid at room temperature for up to 72 h and also used for mechanical tests and amino acid analyses.

Water content of the elastic fibres was determined at 20 °C in the following way. Small (1ml) polyethylene capsules with snap caps were weighed empty and dry. Samples of native elastic fibres (about 300 mg) were put into open capsules and equilibrated over a solution of 0·08moll⁻¹ NaCl in distilled water (relative humidity = 99·7%) in a vacuum chamber. This allowed the fibres to become fully hydrated without surface condensation. The chamber was opened and the capsules were immediately capped and weighed. Then the capsules were opened and the contents were dessicated in a vacuum chamber with phosphorus pentoxide for 48 h. The chamber was opened and the capsules were covered immediately and weighed. An empty capsule served as control and showed no weight change throughout the hydration and drying procedure.

To determine protein content, samples of native elastic fibres were dried, weighed, dissolved in hot alkali (0·1 moll⁻¹ NaOH, 100°C for 5 min) and assayed for protein by the method of Dorsey, McDonald & Roels (1977), using bovine serum albumin and gelatin as standards. This is a heated biuret-folin technique which is equally chromogenic for different proteins.

Carbohydrate content of native elastic fibres was determined by the phenolsulphuric method of Lo, Russel & Taylor (1970). In this case the samples and standards were heated to 100 °C for 5 min in the acid reagent to dissolve the elastic fibres. The assay is sensitive to any carbohydrate with free or potential reducing groups, and includes hexoses, pentoses and uronic acid but not amino sugars. The results are expressed as glucose equivalents.

Samples for amino acid analyses were hydrolysed in 6 mol l⁻¹ HC1 at 105 °C for 24 h in evacuated glass tubes. The tubes were opened and the acid was diluted and evaporated under vacuum. The samples were chromatographed with a Beckman 118C automatic amino acid analyser and peak areas were integrated manually.

Force-extension tests were conducted by stretching elastic fibres on a micro-force transducer stage fitted to a Wild M21 compound microscope, as previously described (Aaron & Gosline, 1981; Shadwick & Gosline, 1981). A bundle of around 10 to 15 μm diameter, consisting of 2 to 5 fibres, was mounted onto the transducer by attaching one end to a slender glass cantilever and the other end to a moveable glass plate with rubber cement. The two ends of the fibre bundle were allowed to dry in order to be glued to the glass, while the middle section of the sample remained in a drop of water. After gluing, the transducer stage was flooded with water. The sample was stretched by sliding the glass plate over a thin film of high vacuum grease. The grease kept the plate stationary at each extension. Tensile force in the sample was calculated from the deflection of the glass cantilever using simple beam theory. Extension of the sample was determined by measuring changes in the spacing of surface features on the fibres. For each sample, the length at which the minimum tensile force was detectable was defined as the unstrained length. All measurements were made optically under the microscope at 300× with a Wild travelling crosshair micrometer eyepiece. The error in these measurements was estimated to be not greater than 5% (Aaron & Gosline, 1981). These experiments were done at 20 °C.

Force-temperature measurements were also made by using the micro-force transducer. A fibre sample was stretched and held at a constant length at 35 °C for 2h. The sample was cooled, in steps of a few degrees, down to 5°C. After 2 min at each temperature the new force was measured. Force measurements were also made as the sample was heated back to 35 °C.

The effect of temperature on swelling of native elastic fibres in water was investigated. A fibre sample was mounted on the transducer stage by attaching one end while leaving the other end free to move. A 1-mm diameter probe from an electronic thermometer was placed immediately adjacent to the sample to monitor temperature. The preparation was covered with water and viewed under the microscope. A temperature controlled microscope stage was used to heat and cool the sample. Length measurements were made optically, with the filar micrometer, over the temperature range 3 to 60 °C.

The water content of the artery wall was 70% by weight. The autoclave-soluble fraction was 30% of the tissue dry weight. Hydroxyproline was present in this fraction as 42 residues per 1000. In purified octopus collagen, hydroxyproline accounts for 81 residues per 1000 (Kimura, Takema & Kubota, 1981). Therefore collagen must make up about 15% of the dry weight of the artery wall.

The results of formic acid extractions on pieces of whole artery wall are given in Fig. 1. This showed that extraction for at least 24 h was required to remove soluble contaminants, and that longer exposure to formic acid does not reduce the amount of insoluble residue. Assuming that the formic acid insoluble fraction is indeed the elastic fibre component then these fibres comprise about 5% of the dry weight of the artery wall. This assumption is consistent with the amino acid data presented below. A 48-h extraction on an aorta by the guanidine HCl/mercaptoethanol and autoclave procedure yielded an insoluble residue that was about 6% of the dry weight (Fig. 1 ).

The biuret-folin assay showed that the elastic fibres are 98–100% protein. The carbohydrate content of the fibres was 0·4–0·5%. Each gram of native fibres contained 0-53 g of water at 20°C. This gives a volume fraction v −0’40, assuming a value of 1·33×10³kgm⁻³ for the density of dry protein (v is the ratio of the unswollen protein volume to the water-swollen protein volume).

The elastic fibre protein could not be dissolved and retained its elasticity after 72 h in 98% formic acid. The protein was also resistant to prolonged autoclaving. Elastin and other protein rubbers are similarly insoluble. However, unlike elastin, the octopus elastic protein was dissolved within 10min in 0·1 moll⁻¹ NaOH at 100°C.

The amino acid composition of purified elastic fibres from the internal elastic layer of the octopus aorta is distinctly different from that of the protein rubbers elastin, resilin and abductin; the octopus protein has a relatively low glycine content and a high proportion of large amino acids (Table 1). The average amino acid weight is 110.

Analyses of the insoluble residues from whole aortae extracted for 6, 18, 24 and 72 h in formic acid and for 48 h in guanidine HCl with mercaptoethanol followed by autoclaving are given in Table 1. Residues after 6 and 18 h in formic acid contain some collagen, as indicated by the high hydroxyproline and glycine content. Extraction of whole aortae for 24 h or longer gave a constant composition as well as constant weight recovery (Fig. 1), and this composition is virtually identical to the elastic protein of the isolated fibres. These results suggest that the residues after prolonged extraction in formic acid or guanidine represent only the elastic protein, and that this protein comprises about 5% of the dry weight of the artery wall. If the elastic fibres make up 5% of the dry weight of the artery wall, and these fibres contain 53% water while the whole wall is hydrated to an average of 70%, then the hydrated elastic fibres will occupy about 3% of the hydrated volume of the artery wall.

Fibres stripped from the internal elastic layer of fresh aortae were highly extensible and elastic. However, when allowed to dry out, the fibres lost their elasticity and became stiff and brittle. Hydrated fibres mounted on the test apparatus could be extended by over 100% without breaking. At each extension the elastic fibres reached equilibrium quickly; there was no appreciable hysteresis or change in properties with consecutive cycles of extension. We found no mechanical differences between elastic fibres tested fresh and those tested after cold or frozen storage. Fig. 2 shows force-extension data for native fibres and also for fibres which were exposed to formic acid for 72 h, rehydrated in water, and then tested. According to theory, a graph of f vs (λ–λ⁻²), for a Gaussian network, yields a straight line with a slope equal to Gv^⅓. Up to λ = 1·5 (i.e. λ–λ⁻²= 1·05) the data fit the Gaussian relationship, while at higher extension ratios the points deviate upwards from the straight line. The elastic modulus, G, was calculated from the slope of the regression lines, using v = 0·40. For the native fibres G = 4·65×10⁵ Nm⁻² and for the formic acid fibres G = 4·62×10⁵ N m⁻². This is almost identical to the modulus of elastin (Aaron & Gosline, 1981) and means that the elastic fibres have about the same stiffness as a common rubber band. These results indicate that long exposure to formic acid had no effect on the elastic properties of the fibres, at least up to extensions of λ =1·5. Comparison of the two curves in Fig. 3 confirms that the formic acid treated fibres were mechanically identical to native fibres, even in the non-Gaussian range of extensions (λ>l·5).

To estimate the number of functional links per random chain between crosslinks in the elastic protein, the experimental curve for native fibres was compared to theoretical plots for random networks with different values of n, where n is the number of ideal links per random chain (Fig. 4). Although the curve for the elastic fibres does not have the exact shape of any one theoretical curve it does fall between the plots for n = 4 and n = 5 at extension ratios greater than 1·6. On the basis of this analysis it is concluded that there are, on average, 4 or 5 functional links in the random chains of the octopus elastic protein.

The effect of temperature on the swollen length of unstrained elastic fibres is shown in Fig. 5. The data were normalized to the specimen length at 20°C, the temperature at which the volume fraction was determined. The length of the fibre sample varied linearly with temperature between 0°C and 35 °C, with a coefficient of linear expansion of −10⁻³/°C. The thermoelastic measurements were made over this temperature range. Above 35 °C there appeared to be a greater rate of decrease in sample length. The reason for this shift in the swelling curve above 35 °C is not yet understood.

Fig. 6 shows that the force required to hold a fibre sample at constant length increased linearly with temperature, and that the slopes of the force-length plots (curves A, B, C) increased with length. In addition, Fig. 6 gives regression lines for the data when corrected from constant length to constant extension ratio (curves A′, B′, C′). The difference between these two sets of curves is explained below.

Fig. 5 shows that with increasing temperature the unstrained fibre length decreases. Therefore, when a fibre sample is held at a constant extended length while the temperature is increased, λ will increase due to decreased swelling. To obtain force-temperature curves at constant λ the following procedure was used. Force-length isotherms in Fig. 7 were plotted from the uncorrected regression lines in Fig. 6; the abscissa is normalized to the unstrained length at 20 °C. The zero-force lengths at other temperatures were determined from the swelling data (Fig. 5). Based on these starting dimensions, lines of constant λ were plotted across the isotherms in Fig. 7. Now the force-temperature curves for constant λ can be drawn, and these are shown in Fig. 6. In principle, this correction method reduces the slope of the force-temperature curve by eliminating the force increase with temperature that arises due to the decrease in swollen dimensions of the protein.

The linearity of the force-temperature curves allows us to proceed with the thermodynamic analysis of the constant length data according to equation 5, and of the constant A data according to equation 8. The results are given in Table 2 and plotted in Fig. 8 for T=20°C. When uncorrected for swelling changes, the extension of the elastic fibres appears to be associated with a large negative enthalpy change, (∂H/∂L)T.P, and at the same time a large decrease in entropy, −T(∂S/∂L)T.P. However, when the force-temperature data are analysed in terms of constant A (swelling correction), the internal energy change, (∂H/∂L)T.V, is much less than the enthalpy change. In fact (∂H/∂L)T.V is close to zero at low extensions, as predicted by the kinetic theory for an ideal rubber. The slope of the constant λ force-temperature curve represents the decrease in conformational entropy that results from deformation of the molecular network at constant volume. This change in entropy accounts for almost all of the elastic force at extensions up to λ = 1·42, and for 77% of the force at λ = 162 (Table 2).

This study presents evidence that the elastic fibres in the wall of the octopus aorta are composed of a rubber-like protein. This protein, which we have called the octopus arterial elastomer (OAE), appears to have physical and chemical properties that can be described by the kinetic theory of rubber elasticity. Like other protein rubbers and lightly crosslinked synthetic rubbers the OAE protein has a relatively low elastic modulus and high extensibility. These properties, along with the high degree of insolubility, are consistent with the model of a crosslinked network of thermally agitated, flexible, random-coil molecules. The network is hydrated under physiological conditions, with the volume fraction of protein in a swollen sample, v = 0·40 at 20°C. For elastin, which is more hydrophobic, v = 0·55 at 20°C (Gosline, 1978b). Resilin and abductin are also water-swollen in vivo, with v equal to about 0·45 and 0·50 respectively (Weis-Fogh, 1961b; Alexander, 1966). The OAE fibres become brittle and glassy when dehydrated. To have rubber-like properties the network chains must be kinetically free and have random configurations. Without water, or some other polar solvent, protein chains will fold into a fixed conformation, held by intra-molecular hydrogen bonds. Presumably, water disrupts these hydrogen bonds and imparts kinetic freedom to the network (Gosline, 1976, 1978a).

Random conformation in the OAE molecular network is suggested by the observation that the fibres are non-birefringent when unstressed (Shadwick & Gosline, 1981), in contrast to the large intrinsic birefringence of highly ordered protein molecules such as collagen fibres (Aaron & Gosline, 1980). When stretched to large extensions the OAE fibres become highly birefringent, as do other protein rubbers, presumably because deformation imposes a degree of order upon the network chains, and this gives rise to optical anisotropy (Gosline, 1980). We observed that the birefringence increased with extension of the OAE fibres, but a more detailed study is needed to determine the exact stress-birefringence relation in this protein. That the molecules of the OAE protein are held together by stable covalent crosslinks is shown by the long-range reversible elastic properties (i.e. there was virtually no creep in extended samples) and by the insolubility of the fibres in protein solvents such as formic acid and guanidine HC1.

The thermodynamic analysis of the OAE protein, tested in water as an open system, shows that the elastic force has a large negative enthalpy component (Fig. 8). In fact, at low extensions the enthalpy change is more than twice as great as the total force (Table 2). Consequently, the decrease in entropy is also much larger than the force, in order to satisfy equation 3. A similar result was obtained for hydrated elastin tested as an open system (Meyer & Ferri, 1936). This result was incorrectly attributed to strain-induced crystallization, a phenomenon which occurs in natural rubber at very high extensions (Treloar, 1975). It is now known that the very large enthalpy change is a direct result of swelling changes that occur in the elastin molecule. Specifically, this involves changes in enthalpy associated with mixing of water and non-polar side groups in the protein (Gosline, 1976). The large negative enthalpy term observed for the OAE fibres in this study may be attributed, for the most part, to changes in temperature-dependent network swelling.

When the thermoelastic data are corrected to constant volume, the internal energy change accompanying extension is reduced to nearly zero and the elastic force is predominantly due to changes in the conformational entropy of the system, as predicted by the kinetic theory (Fig. 8 and Table 2). However, (∂H/∂L)T.V at λ =1·62 is not zero, but actually represents 23% of the total force. At high extensions there may actually be a small positive internal energy contribution to the elastic force due to interactions between chains. On the other hand, (∂H/∂L)_T.V may be essentially zero, and the result for λ =1·62 may represent an over-correction of the constant length data. The temperature swelling relationship for the elastic fibres has been studied only for samples under no tensile load (Fig. 5). The degree of swelling of a polymeric molecule is inversely proportional to the network stiffness (Flory, 1953). Since the OAE protein is in the non-Gaussian region at λ=1·62 the network stiffness has effectively increased, and therefore one might expect the magnitude of the swelling change with temperature to be less. The swelling correction that was applied is probably quite valid for extensions in the Gaussian region but may be too large at extensions where the network becomes stiffer and non-Gaussian; thus (∂H/∂L)_T.V may have been overestimated at the high extension.

The isolation and characterization of the OAE in this study is a significant addition to our fairly limited knowledge of an unusual group of molecules, the protein rubbers. The tremendous variation in amino acid composition among the four known protein rubbers (Table 1) suggests to us that each protein arose independently during evolution in response to selection for similar mechanical design. The intriguing question of what type of sequence patterns allow these special proteins to exist as random, kinetically-free chains is still unanswered. In contrast, most proteins have stable conformations which arise uniquely from each amino acid sequence. Based on the compositions of elastin, resilin and abductin it was suggested that a large proportion of the small amino acids (glycine, alanine and serine) may be necessary in order to minimize steric hindrances and permit a kinetically-free network to exist under normal physiological conditions (Gosline, 1980). Table 1 shows that this is not the case for the OAE protein, where glycine, alanine and serine together make up only 23% of the composition, compared with 57–74% in the other protein rubbers.

An alternative strategy in ‘designing’ a protein rubber might be to have a high proportion of charged amino acids (aspartic and glutamic acids, arginine and lysine) which prevent the formation of stable secondary structures. The charged amino acids account for 33% of the OAE protein. The large side groups on the charged amino acids might also hinder the flexibility of the molecular chains, as mentioned above. However, a relatively inflexible network which becomes non-Gaussian at low extensions may be a useful characteristic for a protein rubber that provides elastic stability to the wall of a blood vessel. The reasons for this are discussed in the following paper, a detailed analysis of the mechanical properties of the octopus aorta.

We thank Arthur Martin and Brian Hartwick for help in obtaining the animals used in this study. This research was funded by an operating grant to JMG from NSERC of Canada.