Mechanical Properties Of The Octopus Aorta

Investigations on the mechanical properties of blood vessels have been numerous since Roy (1880) first gave a detailed quantitative description of the long-range elastic behaviour of the mammalian aorta. He showed that this vessel was very distensible at physiological pressure, but became much less distensible as the pressure was increased. Roy demonstrated two other important properties of arteries: that inflation was accompanied by viscous energy losses, and that the artery wall was thermodynamically similar to natural rubber.

More recent studies have added considerable detail to these fundamental observations (for reviews see Dobrin, 1978; Cox, 1979). We now know that the mammalian artery wall has non-linear elastic properties which arise from a complex arrangement of two fibrous connective tissue proteins: elastin, which is highly extensible and rubber-like, and collagen, which is stiff and relatively inextensible. At low physiological pressures the artery wall is highly compliant, as the pressure is resisted largely by the elastin. With increasing pressures in the physiological range and above, the artery wall becomes much stiffer due to mechanical recruitment of more elastin fibres and finally, to the transfer of load to the collagen network.

The role of the arteries as a passive elastic component in the high pressure closed circulatory system of mammals is well established (McDonald, 1974). In addition to being blood conduits, large arteries are important in reducing the pulsatility of blood flow from the heart to the body tissues. The arteries are distended with blood during systole, while passive elastic recoil provides continued flow during diastole. The elasticity of the large arteries is an important determinant of the dynamics of blood flow in the cardiovascular systems of mammals (McDonald, 1974) as well as other vertebrates (Speckman & Ringer, 1966; Satchell, 1971; Langille & Jones, 1975, 1977; Burggren, 1977).

Until recently there were no reports in the literature on the mechanical properties of blood vessels from any invertebrate animal. This was not surprising since most invertebrates have relatively simple, open circulatory systems. However, in cephalopods the vascular system has attained a high level of complexity; the system is closed, and has an extensive network of blood vessels and capillaries (Wells, 1978; Browning, 1980). Observations on blood pressures in the octopus (Johansen & Martin, 1962; Wells, 1979), nautilus (Bourne, Redmond & Johansen, 1978) and squid (Bourne, 1982) suggested that the large arteries function as elastic reservoirs in these animals. We measured the mechanical properties of the aorta of several species of cephalopod, and found that these vessels were indeed highly extensible elastic structures (Shadwick & Gosline, 1981; Gosline & Shadwick, 1982). We also showed previously that.the octopus aorta contains a connective tissue network made up of a protein with rubber-like properties. We proposed that this protein provides the long-range elasticity in the octopus vessel wall, and thus is a functional analogue of elastin (Shadwick & Gosline, 1985).

In this paper we present a quantitative analysis of the non-linear and viscoelastic properties of the octopus aorta in vitro. Our results indicate that this vessel is mechanically similar to the major arteries of vertebrates, and is suitably designed to function as an efficient elastic energy storage element in the circulatory system. We also demonstrate that the elastic properties of the artery wall at physiological pressure can be attributed to the presence of the rubber-like protein which we have called the octopus arterial elastomer (OAE). .

We have considered the artery wall as an elastic structure for which the mechanical response is independent of the rate of deformation. In reality, biological, materials exhibit time-dependent elasticity because they have properties both of elastic solids and viscous fluids. A perfectly elastic solid deforms instantly under a given stress to a constant strain. Removal of the stress is followed by instantaneous recovery of the original dimensions because all the work done in deformation is stored as elastic energy and can be recovered without loss. On the other hand, a stress applied to a purely viscous fluid will cause deformation at a constant strain rate, determined by the viscosity. Energy used to drive viscous flow is lost as heat. The response of a viscoelastic material to stress or strain will be between these two extremes, depending on the relative contribution of the elastic and viscous components (Wainwright, Biggs, Currey & Gosline, 1976). There are several ways to quantify viscoelastic properties of arteries in vitro. One method is by dynamic mechanical testing over a range of frequencies. This technique provides a direct measure of viscoelasticity in a form that can be related to haemodynamic parameters such as pulse wave velocity and vascular impedance. Arteries in vivo are subjected to a static stress arising from the mean blood pressure, and a dynamic stress applied by the pressure pulse from each heartbeat. To simulate these conditions in vitro a small sinusoidal strain oscillation is imposed upon a static load, and the resulting stress oscillation is observed at each frequency. If the material exhibits linear viscoelasticity the stress will also be sinusoidal and will be phase shifted ahead of the strain by an angle δ. The magnitude of δ indicates the relative importance of viscosity to the overall mechanical response of the material. In theory, δ may vary from 0° for pure elasticity to 90° for pure viscosity.

Live specimens of Octopus dofleini, 10-15 kg in weight, were obtained from the waters of Puget Sound, Washington, and Barkley Sound, British Columbia. The animals were maintained in good health in a recirculating sea water system at 10 °C, and fed on fish, crabs and clams.

Experiments were performed on segments of the dorsal aorta, the major vessel which leaves the systemic heart. Aortae were dissected from animals which had been anaesthetized by chilling and killed by decerebration. The samples were washed and kept at 4 °C in saline (0·4 μ.m filtered sea water) containing 1% of a solution of penicillin, streptomycin and fungizone (Gibco Labs), and used for mechanical tests within 12 h of death. These tests were at 10°C, or in some cases at room temperature (20°C). No differences in the passive mechanical response of the aortae could be detected between these temperatures.

Artery segments about 5 cm long were cannulated at one end with a tubular metal connector of appropriate diameter and held in the saline-filled chamber. This connector was linked to a saline pressure reservoir and either a linear infusion pump for quasi-static tests or a sinusoidal pump for dynamic tests. After saline perfusion to clear air bubbles from the system, the distal end of the specimen was ligated. This arrangement left the artery free to lengthen as it was inflated. Most samples of aorta were taken between the heart and the digestive gland where no branch vessels were present (see Wells, 1983). In some cases a more distal portion of the aorta at about the level of the crop was used, and branch vessels were ligated. Leak-free preparations were normally obtained without difficulty.

Quasi-static pressure-volume curves were generated by very slow, continuous cycles of inflation and deflation of the vessel segment by the linear pump. The pump consisted of a calibrated glass syringe with a spring-loaded piston which could be advanced or retracted at a predetermined rate by a reversible, variable-speed motor (Cole-Parmer Master Servo-Dyne). A cycle consisted of taking the artery from zero pressure to about 10 kPa (=100cmH₂O) and back to zero. The cycles were repeated until stable P-V curves were obtained. Infusion rates were varied from 10 to 15 μl s⁻¹, depending on the specimen size, so that each cycle took 2 to 3 min. For a typical unpressurized specimen r = 0·12cm, R = 0·23 cm and L = 3·5cm. A preconditioning period consisting of several inflation cycles was required in order to obtain stable P-V curves.

Transmural pressure was measured through an 18 gauge needle which was glued into the artery connector and coupled to a BioTec BT70 pressure transducer by a short length of polyethylene tubing (PE 190). Changes in the external diameter or length of the artery were determined with a video measuring system. This consisted of a video dimension analyser (Instruments for Medicine and Physiology VDA 303) that provides an analogue voltage which is proportional to the distance between two reference points in the video image. The dimension measurement is made on the horizontal axis of the television and the resolution is about 0·1% of the full width of the screen (Fung, 1981).

Experiments to demonstrate the effect of vascular muscle activation on the mechanical behaviour of the artery wall were carried out using 5-hydroxytryptamine (5-HT-creatine sulphate complex, Sigma Co.) to stimulate and acetylcholine (ACh-chloride, Sigma Co.) to relax the muscles. The drugs were dissolved in saline and introduced into the artery lumen by infusion after opening the distal ligature. As the drug solution flowed through the vessel it was allowed to mix into the surrounding bath. The ligature was then closed and the artery was inflated, using the same drug solution in the pump.

Dynamic inflation experiments involved measuring small pressure and diameter oscillations over a frequency range of 0·05 to 1·0 Hz. These experiments were carried out on artery segments at three different mean pressures: 3, 4 or 6kPa (corresponding to circumferential strains of approximately 0·6, 1·0, 1·2 respectively). Small sinusoidal volume changes were imposed on the inflated artery by a diaphragm pump driven by an electromagnetic vibrator (Ling Systems 200). The amplitude and frequency of the vibrator were controlled by a signal generator which also provided the frequency reference for a phase analyser. A needle valve inserted between the pump and the static pressure head ensured that flow from the pump back into the reservoir would be negligible. The volume amplitude was adjusted to give diameter changes of less than ±5% of the mean value. Under these conditions the pressure (P) and diameter (D) oscillations were sinusoidal with little distortion, and P-D hysteresis loops were essentially linear. The amplitudes and the phase shift (δ ) between P and D signals at each frequency were measured by the phase analyser (SE Labs SM272DP Transfer Function Analyser) as described by Denny & Gosline (1980). The P and D signals were also recorded continuously on a two-channel strip chart recorder.

The frequency response characteristics of the pressure transducer were tested by a free-vibration method using a step function of pressure (‘pop’ test, Gabe, 1972). The transducer and tubing together had a damped resonant frequency of 45 Hz, and a damping factor of 0·26. Over the frequency range used the pressure signal was delayed by 0·7° per Hz, and had an amplitude error of less than 4% at 10 Hz. The VDA output is processed through an internal single RC 15 Hz filter, which at 10 Hz gives a 25% attenuation and a 135° phase lag. The frequency response of the VDA over the relevant frequencies was measured with the phase analyser. Appropriate corrections for phase and amplitude losses were made to the pressure and diameter data.

After each inflation experiment the unstressed length and external radius were recorded. The thickness of the vessel wall was measured on frozen-cut transverse sections, using a microscope micrometer. The vessel wall volume was calculated on the basis of uniform cylindrical geometry and assuming isovolumetric deformation. This allowed us to determine vessel length L, and inside radius r, from V and R at any pressure, or R and r from V and L. Equations 2–5 were used to obtain biaxial stress-strain curves. Values of tangential elastic moduli were computed from equations 9 and 10 by the method of Dobrin & Doyle (1970). Dynamic pressure and radius data were used with equation 13 to obtain values of the dynamic incremental elastic modulus in the circumferential direction.

Quasi-static uniaxial tests of the aorta were carried out on a tensile testing machine (Instron model 1120). Short transverse sections of aorta (3-4mm long) were mounted over two rigid L-shaped stainless steel bars, one of which was secured to the base of the test frame, while the other one was attached to a force transducer on the moveable cross-head of the Instron. This transducer gave reliable measurements of forces from about 0·1 to 500 g. Force-extension tests were conducted by slow cyclic stretching at constant rates from 0·5 to 5 mm min⁻¹ (i.e. strain rates of 0·002 to 0·02 s⁻¹). No differences in the response of the artery specimens were detected over the range of strain rates used. Extension of the sample was determined from the cross-head displacement. Data were recorded on an X-Y plotter. Since the ring samples became flattened under a force of less than 1 g, they were considered to be two parallel sheets of tissue, each having a length equal to one-half of the circumference of the ring. Thus uniaxial extension of ring samples resulted in stresses which were essentially in the circumferential direction (Attinger, 1968; Goedhard & Knoop, 1973). Preconditioning of the samples was achieved by repeated strain cycling until relatively stable force-extension curves were obtained. During the experiments the samples were immersed in a saline bath at 10°C.

In another series of experiments the longitudinal elastic properties of the aorta were investigated. Vessel segments about 5 cm long were mounted in the saline bath by ligating the ends over tubular connectors which were attached to the Instron test frame. Longitudinal force-extension tests were done according to the procedure described for the ring samples. A hole through one connector allowed fluid movement in and out of the vessel lumen, thus maintaining the transmural pressure at zero throughout the test.

Sample dimensions were measured at the end of each test. Strain calculations were based on an unstressed sample length which was measured at the crosshead position where a positive force was just detectable. Stress was calculated as the force divided by the cross-sectional area of the specimen, which for rings was taken as 2hL (see Fig. 1A), while for the longitudinal samples it was π(R²—r²). With the assumption of isovolumetric deformation of the artery wall, the cross-sectional area, and thus the stress, at each strain could be calculated from the initial dimensions.

Dynamic mechanical tests were conducted on artery ring samples by using a forced strain oscillation technique which was analogous to the dynamic inflation. Ring samples were loaded to predetermined strains and then subjected to small sinusoidal strain increments at 0·1-10 Hz, by an electromagnetic vibration apparatus (Gosline & French, 1979). Deformation of the sample was measured by a strain gauge displacement transducer attached to the vibrator shaft. Force was measured by a semiconductor strain gauge transducer which was linked, by the sample, to the vibrator. Experiments were performed at mean strains of 0·3 and 0·6. The displacement amplitude was kept small enough (±2% of the mean strain) to give relatively linear force-extension hysteresis loops. Force and displacement signals were conditioned with matched carrier amplifiers (SE Labs type 4300), and the amplitudes and phase relationship measured by the Transfer Function Analyser. The resonant frequency of the force transducer was two orders of magnitude greater than the highest test frequency and therefore no signal corrections were necessary. The force and extension data were used to calculate the dynamic incremental modulus, according to equation 12.

The first four consecutive inflation cycles for an aortic segment are shown in Fig. 2. The first inflation curve is highly sigmoid and lies well above the deflation curves.

Since the product of pressure (P) and volume (V) is work, then the area under the inflation curve is the total work done (= the energy used) to distend the artery, while the area under the deflation curve is the energy recovered by elastic recoil. The area enclosed by the P-V loop is the energy lost as heat through viscous processes and, when expressed as a percentage of the total energy, is called the mechanical hysteresis. For the first inflation cycle shown in Fig. 2 the hysteresis was 43%. With subsequent inflations the curves shifted to the right and the hysteresis loops became smaller. Three to four cycles were normally enough to obtain a stable response with about 25% to 30% hysteresis. The instability of hysteresis curves during the first few strain cycles is a phenomenon which is typical of arteries tested in vitro, and probably represents a process of internal structural readjustment to stresses after being excised and kept in an unpressurized state (Bergel, 1961a; Fung, 1972). The most important feature of the P-V curves is that the volume distensibility of the aorta, (dV/VdP), decreased markedly with increasing distension over the range of physiological blood pressures, approximately 2-5kPa (Johansen & Martin, 1962).

Inflation of the octopus aorta always resulted in an increase in length as well as in circumference. The changes in these dimensions which occurred in a typical experiment are shown in Fig. 3. At a pressure of 5 kPa, the increase in length was 30% (ϵ_L = 0·30), while the circumference had increased by 120% (ϵL=1’20). Since luminal volume increases with the square of the radius, but only linearly with vessel length, it can be seen that circumferential strain contributes the major portion of the volume distensibility of the aorta. Above physiological pressures only relatively small changes in longitudinal and circumferential strain were observed. The aorta has a relatively thick wall at zero pressure, the ratio h/R being about 0·45 (±0·05). However, due to the large circumferential strains, at 5 kPa pressure h/R is reduced to about 0·10 (±0·01). In mammalian arteries h/R is about 0·20 at zero pressure and decreases to about half of this value at physiological pressures (Cox, 1979).

Fig. 4 presents data, taken from the inflation limb of P-V loops, in terms of stress and strain in the circumferential and longitudinal directions. The observed stress-strain relations for the octopus aorta were always non-linear. The stress-strain data for distal segments of the aorta indicate that in this region the aorta is somewhat less extensible, circumferentially and longitudinally, than the main part of the vessel which lies more proximal to the heart. While these differences in extensibility may be due to an ‘elastic taper’ in the aorta, as occurs in the vertebrate aorta (McDonald, 1974), the possibility was not investigated further in this study. The experiments to follow deal only with the major proximal portion of the aorta which is characterized by curves B and D in Fig. 4.

Immediately after excision, many aortae exhibited a brief period of spontaneous muscle activity which was severe enough to cause visible undulatory movements along the length of the vessel. However this phenomenon did not persist for more than about 30 min in cold saline. Spontaneous muscle activity was also indicated by brief, large pressure transients or by small pressure pulsations which occasionally occurred during the slow inflation of a vessel. This activity appeared to be abolished by several consecutive inflation cycles during the normal preconditioning period, or by exposure of the artery to a high pressure (10 kPa) for several minutes.

Since the main objective of this study was to measure the passive mechanical properties of the artery wall, it was necessary to be able to distinguish between the active and relaxed states of the vascular muscle. To do this, we used drugs which had previously shown excitatory or inhibitory effects on the octopus circulation in vivo (Johansen & Huston, 1962) or on the aorta in vitro (Wells & Mangold, 1980). Prolonged periods of muscle activation (i.e. several minutes) could be produced by treatment with 5-HT. Fig. 5 shows the results of an experiment in which a preconditioned aortic segment was subjected to three sets of inflations, the first with normal saline (A), the second after infusion with 5-HT (B), and the third following infusion with ACh (C). P-V curves for the control and ACh-treated artery are virtually identical. In the presence of 5-HT, however, the volume distensibility of the aorta was greatly reduced, particularly over the physiological range of pressures. Muscle activation increased hysteresis slightly above the control level (28% and 22% respectively for the example in Fig. 5). The results suggest that the muscle in preconditioned, control arteries was in the relaxed state. Thus, we can regard the results of tests on preconditioned artery samples where drugs were not used as representing the passive elastic properties of the vessel wall.

Stress-strain curves, calculated from the loading portion of P-V curves, are shown for the longitudinal and circumferential directions in Fig. 6. Here, strain is expressed with reference to the initial dimensions of the relaxed vessel. Note that 5-HT did not produce a measurable change in the external diameter at zero pressure, although the length appeared to increase by about 6%. For all strains in the octopus aorta, the circumferential and longitudinal stresses were elevated by activation of the muscle, with the exception of the initial part of the longitudinal extension. The difference between the stress curves for the stimulated and relaxed states is the active stress component, and this is shown in Fig. 6. Active stress in the circumferential direction increased continuously with strain, reaching an apparent peak at ϵ_c = 1·0. In the longitudinal direction, the active stress increased to a plateau level at ϵ_L>0·20. These results suggest that the magnitude of the active stress response is maximal at some optimum circumference or length. This optimum dimension appears to occur at circumferential and longitudinal strains that correspond to a pressure of about 3·5 kPa, i.e. in the middle of the physiological range.

Fig. 7 shows examples of force-extension curves obtained from quasi-static uniaxial tests of the aorta in the longitudinal and circumferential directions. We found that hysteresis was greater longitudinally than circumferentially (39% vs 26% respectively, in this case). Data from the loading portion of force-extension cycles were used to construct stress-strain curves, examples of which are plotted in Fig. 8. These curves resemble the J-shaped biaxial stress-strain curves in Fig. 4, except that the artery wall appears to be more extensible, for a given stress, when loaded uniaxially than biaxially. This difference was expected according to equations 9 and 10, and was most noticeable in the circumferential direction.

Values of the tangential elastic moduli, EL and Ec, were calculated from uniaxial stress-strain data using equation 1 (Fig. 9). Over the range of extensions used the artery wall increased in stiffness by approximately two orders of magnitude: EL varied from 3X10³ to l·8x10⁵Nm⁻², while Ec varied from about 5X10³ to 5X10^sNm⁻². These values of modulus are characteristic of an elastic material which has from 1/200 to 1/2 the stiffness of an ordinary rubber band. The large increase in elastic modulus which accompanies increasing strain explains the decreased volume distensibility observed at high pressures (Fig. 2).

Uniaxial tests on isolated artery rings are simple and convenient, but the results from these tests may be unreliable at very high strains where fibre orientations within the vessel wall must differ considerably between uniaxial and biaxial specimens. Therefore we wanted to calculate the circumferential elastic modulus from biaxial inflation data. This requires the knowledge of either Poisson’s ratio or the longitudinal modulus. Since Poisson’s ratios were not measured directly, we decided to use values of E_L from uniaxial tests in order to determine VCL and Ec from the biaxial data. To do this, we assumed that the uniaxial EL values are close to the true biaxial values. The assumption is reasonable since the longitudinal uniaxial tests involved only small strains (up to 30%) which presumably did not cause too extensive an alteration in fibre orientation.

Values of VCL calculated from equation 9, varied from 0·20 to 0·54 with a mean of 0·30. We used these values to determine E_c by solving equation 10. E_c is plotted as a function of pressure in Fig. 10, and as a function of strain in Fig. 11. Ec increased from 9x 10³ to 2·3x 10⁵ N m⁻², over the approximate physiological pressure range, which corresponds to circumferential strains of 0·5 to 1·18 respectively. The ratio of the moduli, E_C/EL, varied between 1·05 and 3·30 (Fig. 10). It appears that the octopus aorta has greater stiffness circumferentially than longitudinally at all pressures, and has a Poisson’s ratio which is generally less than 0·5. Under these conditions equation 9 yields positive values of Δϵ_L,: that is, the artery should lengthen as it is inflated.

The curve for E_c obtained from uniaxial tests (Fig. 9) is redrawn in Fig. 11 for comparison with E_c values determined from inflation tests. For strains below 1·10, tests on artery rings yielded values of Ec which are in close agreement with those obtained from inflation of vessel segments. At strains above 1·10 the two curves diverge. The ring samples may be expected to show mechanical behaviour which is different from the. intact vessel, particularly at high extensions, because the artery wall is not a homogeneous material. At large strains it is likely that fibre networks in the artery wall will be considerably reorientated, in ways which may differ in the two types of samples. Since these analytical methods are not exact, the results must be regarded as only approximations of the true values. It is quite satisfying, therefore, that uniaxial and biaxial tests on the aorta gave similar values of E_c for strains which include the physiological range. This indicates that uniaxial tests on ring samples are legitimate for extensions up to about 100%.

The presence of a significant viscous component in the mechanical response of octopus artery wall was indicated by hysteresis in the quasi-static inflations. Other viscoelastic phenomena which we observed during step-wise inflation of the aorta were stress-relaxation and creep (data not shown). Since arterial viscoelasticity is important in determining haemodynamic relationships, we decided to characterize this property by dynamic testing.

A quantitative analysis of the frequency-dependent mechanical properties of the octopus aorta is based on the combined results of dynamic mechanical tests on intact vessel segments and ring samples. In these tests small sinusoidal strain oscillations were superimposed on different levels of pre-strain (Fig. 12A). Examples of a pressure-radius loop from a dynamic inflation experiment (Fig. 12B) and a force-extension loop from an artery ring vibration test are given (Fig. 12C). Both show that the strain increments were sufficiently small that these loops were approximately linear. Fig. 13A shows that the storage modulus E’ increased continuously with the frequency of oscillation for each level of pre-strain which was tested. In addition, the proportional change in E′ over the range of frequencies used increased with strain, i.e. the ratio of E′ at 10Hz to E′ at 0·1 Hz was 1·33 at ϵ_C = 0·3 (curve d), 1·43 at ϵ_C = 0·6 (curve c), 1·88 at ϵc = 1·0 (curve b) and 1·96 at ϵc = 1·2 (curve a).

In all experiments, the phase angle δ was positive, i.e. the change in pressure (or force) always led the change in radius (or extension), and tan δ varied between 0·1l and 0·27 (Fig. 13B). For these values of tan δ, E′ is approximately equal to E* and E” is relatively small. Unlike E′, tanδ.did not seem to change significantly with different levels of pre-strain, except perhaps at the highest frequencies when curves for higher pre-strain (a and b) appear to diverge from curves for lower pre-strain (c and d). With increasing frequency there was a steady rise in tan δ, suggesting that the loss modulus E” was increasing relative to E′ (since tan δ = E”/E′). However, the large scatter in data at frequencies above 1 Hz may indicate that errors were being introduced due to inertial effects of fluid acceleration in the vessel. We were unable to assess to what extent this problem influenced our measurements.

The results of this study show that the octopus aorta is a highly distensible and resilient elastic vessel which is well suited to function as an elastic reservoir in the central circulation. A very important feature of the aorta is that the elasticity is highly non-linear. Over the physiological pressure range the passive distensibility of the vessel wall changed dramatically as the circumferential elastic modulus increased from a relatively low value of about 10⁴ up to about 2x10⁵Nm⁻². Thus, a twenty-fold increase in stiffness occurs over the range of strain imposed on the artery wall by a resting blood pressure pulse. Similarly, in other cephalopods we found large changes in the aortic wall stiffness with extension in the physiological pressure range. In the nautilus (Nautilus pompilius) Ec increased from 2X10⁴ to 2x10⁵Nm⁻² between pressures of 2 and 6 kPa respectively, while in the squid (Nototodarus sloani) E_c increased from 10⁵ to 10⁶Nm⁻² between pressures of 10 and 20 kPa respectively (Gosline & Shadwick, 1982).

Non-linear elasticity is also characteristic of vertebrate arteries. For example, Bergel (1961a) showed that the luminal volume of the canine thoracic aorta increased by about four times when inflated from 0 to 40 kPa pressure, and the circumferential elastic modulus increased from 2X10⁵ to about 10⁶N m⁻² between 11 and 22 kPa, the approximate in vivo blood pressure. All blood vessels that have been studied exhibit ‘J-shaped’ stress-strain curves and a progressive increase in wall stiffness with extension, or pressure (Bergel, 1961a; Cox, 1979; Fung, 1981). Apparently this is a design feature of highly distensible pressure vessels which is necessary in order to prevent instability and rupture (Burton, 1954; Gordon, 1975; Bogen & McMahon, 1979).

Burton (1954) showed that for a pressurized elastic tube the volume distensibility D (=dV/VdP), is inversely proportional to [(Eh/R) —P], where E is the elastic modulus, h is the wall thickness, R is the radius and P is the pressure. As the pressure is raised, R will increase and h will decrease. If E remains constant with extension, then D will become infinitely large as P approaches Eh/R, and the tube will rupture. This can be avoided either by having E very large relative to P, or by having E increase with radius, such that Eh/R is always greater than P. In the first case the tube would be non-compliant (e.g. a steel water pipe), but this would not be useful as an elastic reservoir. The second case is what has been observed for the arteries of vertebrates as well as cephalopods. This allows the vessel to have large compliance at low pressures, thereby smoothing the pulse and offering low resistance to flow from the heart, and yet still be protected from ‘blowout’ at higher pressures. The range of modulus which is necessary to meet the above criteria will depend on the physiological pressures normally experienced, and on the dimensions of the vessel. The elastic modulus of the aorta is lower in the relatively low pressure circulatory systems of octopus and nautilus than in the high pressure systems of squid and dog.

In our experiments the artery segments were free to lengthen when inflated, so as to mimic the situation in the animal. The octopus has no rigid skeleton, so the aorta cannot be tethered to a fixed length in vivo. In contrast, the vertebrate aorta is restricted longitudinally by the axial skeleton, and retracts by about 30% when excised (Dobrin, 1978). The length change that occurs when the vessel is inflated depends on the Poisson’s ratios and the relative stiffness in the circumferential and longitudinal directions (equations 8, 9). Ec was greater than EL at all pressures in the octopus aorta. The vessel lengthened when inflated, although the longitudinal strain was always less than the circumferential strain. Mechanical anisotropy has also been demonstrated in many mammalian arteries. Patel, Janicki & Carew (1969) found that E_L was greater than E_c in the dog thoracic aorta. Others have shown that Ec is greater than EL in the dog carotid artery (Dobrin & Doyle, 1970; Cox, 1975).

The passive hysteresis for slow inflation of the octopus aorta was slightly greater than that typically observed in mammalian arteries. Vascular muscle is the main factor contributing to hysteresis in the artery wall (Dobrin, 1978), although a small amount of hysteresis is also found in isolated elastin and collagen when tested at low strain rates (Fung, 1981). The octopus aorta has more muscle, and less collagen and elastic fibres than have mammalian arteries (Shadwick & Gosline, 1983). These differences in composition of the tissue components, as well as their structural organization, probably account for the lower elastic modulus and greater hysteresis in the aorta of the octopus versus the mammal.

Hysteresis is increased by activation of the muscle component in mammalian arteries (Dobrin, 1973; Busse, Bauer, Sattler & Schabert, 1981) and apparently in the octopus aorta as well. We do not fully understand the role of the muscle in the octopus vessel wall. Fibroblast cells must be responsible for the deposition of the extracellular connective tissues. Muscle cells may serve this purpose, as well as forming a mechanical linkage between fibres in the elastic network (Shadwick & Gosline, 1983). Vascular muscle may be responsible for active propulsion of blood through the circulation, but this has not been clearly demonstrated (Johansen & Martin, 1962; Wells, 1978). It seems unlikely that muscle activation in the aorta could be important in regulating blood flow resistance since this parameter must be influenced predominantly by the peripheral vessels.

The non-linear elastic properties of the mammalian artery wall are attributed to the parallel arrangement of elastin and collagen fibre systems. Elastin itself has non-linear elasticity because of its non-Gaussian behaviour (Aaron & Gosline, 1981). However, the increase in stiffness of isolated elastin fibres with extension is not nearly as great as the increase in stiffness which occurs with comparable strains in the whole artery wall. Thus, the ‘J’ shape of the stress-strain curve for the mammalian artery arises, not from the non-linear properties of elastin, but from the recruitment of more elastin fibres and the transfer of load to the collagen network, as the artery is inflated (Roach & Burton, 1957; Wolinsky & Glagov, 1964).

In contrast, the non-linear elasticity of the OAE fibres may be important in determining the shape of the stress-strain curve for the octopus aorta. It was shown in the previous paper (Shadwick & Gosline, 1985) that the molecular chains of the OAE protein are more restricted in their flexibility than are those of elastin. Consequently, the stiffness of the OAE fibres increases rapidly with stretch in the region of non-Gaussian extensions. In Fig. 14 the shapes of the stress-strain curves for the octopus artery wall and for the OAE fibres are compared. Here, the stress scales have been adjusted arbitrarily so that the initial portions of the curves coincide (up to ϵ = 0·2). When this is done, it can be seen that the curves are almost identical in form up to strains of 1-2. By this semi-quantitative method it appears that the non-Gaussian behaviour of the elastic fibres gives rise to an increase in stiffness, with extension, which is similar to the non-linear elasticity of the whole artery wall, up to and including the normal range of physiological extensions.

We found that the elastic fibre network comprises about 3% of the volume of the aorta, and that the fibres are present as a sheet lining the vessel lumen, and as a multi-directional array throughout the muscle layers (Shadwick & Gosline, 1983, 1985). Fig. 14 shows that up to ϵ = l·2 the stress required to stretch the elastic fibres is about 100 times greater than the circumferential stress in the whole artery wall (i.e. the stress scales in Fig. 14 differ by a factor of about 100). If as little as one-third of the total fibre content (i.e. about 1% of the wall material) is orientated to support the circumferential stress when the artery is inflated, then the presence of elastic fibres alone can account for the stiffness of the artery wall through the physiological range of pressures. Collagen, which is present as a loose adventitia, may not be required to take a significant portion of the load until very large extensions.

The results of our dynamic mechanical tests show that the octopus aorta is a viscoelastic structure which becomes stiffer and less resilient with increasing rates of deformation. E′ and tan δ both increased continuously with frequency over the range of 0·05 to 10 Hz. This type of behaviour is typical of other biomaterials and polymeric solids at frequencies below the glass transition (Ferry, 1970; Fung, 1981). In the mammalian artery wall E′ increases by up to twice the quasi-static value, and tan δ5 generally rises from about 0·09 to 0·20 with frequencies from 0·2 to 20 Hz respectively (Bergel, 1961b; Gow, 1970; Patel, Janicki, Vaishnav & Young, 1973; Busse et al. 1981). The magnitude of viscosity in mammalian arteries can be correlated with the amount of vascular muscle at different anatomical sites, as can hysteresis and stress relaxation properties (Dobrin, 1978). In terms of tan δ and the relative change in E′ between 0 and 10 Hz, the octopus aorta is comparable to the carotid artery of mammals. No data are available for the dynamic mechanical properties of any other cephalopod arteries.

Viscous energy losses in the artery wall decrease the efficiency of elastic energy storage in the system. However, some degree of wall viscosity may be haemodynamically favourable. The impedance to flow which the heart must work against is determined by a complex interaction between incident and reflected pressure waves in the arterial tree; wall viscosity causes attenuation of pressure waves, and this can reduce the impedance (Taylor, 1966). Since viscosity in the artery wall arises primarily from the muscle component, then the amount of muscle present in an artery may be determined, at least in part, by a mechanical requirement to optimize elastic recoil and pressure wave damping in the vessel wall.

In the physiological range of frequencies for Octopus dofleini (about 0·1 to 1 Hz) the resilience of the aorta varied from 70% to 62%. This is only slightly less than the resilience of the mammalian aorta at the relevant physiological frequencies (approx. 2-20Hz in the dog; Bergel, 1961b). Thus the octopus aorta appears to have mechanical properties which would allow it to function effectively as an elastic reservoir in the circulatory system. We are currently investigating the relationship between arterial mechanical properties and the fluid dynamics of blood flow in the living octopus.

This research was funded by an operating grant to JMG from NSERC of Canada.