Effect of aestivation on long bone mechanical properties in the green-striped burrowing frog, Cyclorana alboguttata

Bone tissue is phenotypically plastic and is continually remodelled in mammals and birds to maintain a form appropriate to its mechanical function(Biewener and Bertram, 1993). Increases in bone usage (i.e. loads acting on the bone via activity or gravity) can lead to increases in size and overall strength of the bone(Umemura et al., 1995; Biewener and Bertram, 1994). Conversely, disuse following limb immobilisation, cage constraint, hindlimb suspension and space travel can elicit one or more of several responses,including demineralisation, decrease in size and corresponding loss of bone strength (Bagi and Miller,1994; Globus et al.,1984; Whedon and Heaney,1993; Turner,2000). For example, in a clinical setting, Hanssen et al.(1975) described a 2% spinal bone loss per week in humans during prolonged bed rest after scoliosis surgery. Furthermore, 12 weeks of space travel led to an 8% decrease in bone density in human astronauts (Holick,1998). These effects are directly attributable to disuse and are over and above the homogeneous crystallisation of calcium carbonate induced by microgravity (Liu, 2001).

In addition to skeletal disuse resulting from clinical or otherwise unnatural circumstances, many animals experience extended periods of immobilisation in nature that could also affect bone properties or size. Two prominent examples of natural immobilisation are hibernation and aestivation. Hibernation in endotherms is a dormancy strategy characterised by a reduction in body temperature near to ambient temperature, a markedly reduced metabolic rate and spontaneous arousals by activation of major heat-producing mechanisms(Bartholomew and Hudson, 1960; Kayser, 1961; Lyman et al., 1982). Similarly, aestivation is a state of reduced metabolism seen most commonly in organisms, such as anurans, inhabiting periodically dry habitats(Pinder et al., 1992). The green-striped burrowing frog, Cyclorana alboguttata, survives droughts through the excavation of an underground chamber, the formation of a waterproof cocoon and the storage of water in the bladder(Flanigan et al., 1993; Withers, 1993). The inactivity of hibernation is known to be correlated with decreases in bone size and strength (Doty and Nunez,1985; Whalen et al.,1971) although the response varies between species. However,whether aestivation also induces such changes in bone properties is unknown.

Limb immobilisation related to aestivation can be extreme. In such a capacity, the long bones of the hindlimbs can be effectively immobilised for several years at a time. C. alboguttata induced to aestivate on top of pressure-sensitive piezoelectric film in glass jars showed no limb movements in a 12-week experimental period (N. Hudson and C. Franklin,unpublished data). However, it has become apparent that the morphology of the semimembranosus capillary beds is unaffected by aestivation(Hudson and Franklin, 2003). Moreover, following nine months of aestivation in a laboratory setting, C. alboguttata have the immediate capability to resume muscle performance as measured by burst swimming and in vitro gastrocnemius force production (Hudson and Franklin,2002a).

Taken together, these data strongly suggest that C. alboguttataemerge from aestivation in the field with a fully competent locomotor system,and therefore within hours of surfacing can transmit forces through the long bones that are representative of an active specimen. This contrasts with the situation in hibernating mammals where there is some impairment in muscle performance. For example, Harlow et al.(2001) found a 23% deficit in the contractile performance of the tibialis anterior in the overwintering black bear Ursus americanus. Although overall locomotor performance has never been measured in a hibernator, it is likely to parallel the deficit in isolated muscle performance and thereby reduce the demands on the skeletal system until the muscle regains its prehibernation performance. Hibernation also induces loss of bone in bats (Whalen et al., 1971), hamsters(Kayser and Frank 1963) and squirrels (Mayer and Bernick,1959). In light of the empirically determined difference in muscle performance between emerging hibernators and aestivators, it seems possible that, in contrast to hibernators, aestivating frogs may preserve long bone size and strength in order to withstand the forces associated with jumping upon emergence.

To assess the impact of three and nine months of aestivation on the skeletal system of C. alboguttata, we tested the ex vivomechanical properties of the femur and tibiofibula using three-point bending and compared these results with those of frogs kept active in the laboratory. The femur and tibiofibula are the main skeletal elements in the hind limbs and are typically sensitive to usage-dependent remodelling in vertebrates. Although bending has never been experimentally demonstrated in frogs(Calow and Alexander, 1973), it is predicted on the grounds that the long bones (particularly the femur)display some curvature and are subject to compression and because the predominant direction of muscle and reaction forces rarely coincides with the bones' longitudinal axes (Biewener,1983).

We estimated an appropriate in vivo loading rate to apply to the long bones of C. alboguttata using force plate records of ground reaction forces developed during jumping. The mechanical data were supplemented by histological analysis to assess osteocyte activity, as it is known that in hibernating bats osteocytic osteolysis accounts for much of the associated bone loss (Whalen et al.,1971).

Adult green-striped burrowing frogs (C. alboguttata Günther 1867; body mass 7–30 g) were captured from flooded roadsides following summer rains in southeast and mid-eastern Queensland. Animals were transported back to the laboratory in individual plastic bags (containing water and plant matter) within 24 h. They were randomly assigned to one of three treatment groups (controls, 3-month aestivators and 9-month aestivators) with animals matched as closely as possible for sex and mass between the three groups. Control frogs were housed in individual plastic boxes (27 cm×14 cm×12 cm) containing wet paper towelling and fed ad libitumwith grasshoppers and cockroach nymphs. To induce aestivation, frogs were placed into individual plastic containers filled with wet clay (obtained from the locality of animal capture) that was allowed to dry naturally over a period of several weeks. Most frogs burrowed almost immediately and did not resurface. When the aestivation period was complete, frogs were extracted by breaking open the soil block across the burrow line. Euthanasia was achieved by double pithing.

To determine a biologically relevant loading rate for bending tests, we used force platform records of jumps to evaluate an in vivo rate of bone loading. Ground reaction forces represent one of the major forces experienced by the long bones during locomotion; therefore, the time to peak force was judged to be the most biologically relevant duration of loading to apply during tests of bone mechanical properties. Control frogs(N=35) of various masses were encouraged to leap off a custom-built force platform at 25°C (platform based on design of Katz and Gosline, 1993; see Wilson and Franklin, 2000 for details) that simultaneously measured the vertical, horizontal and lateral ground reaction forces exerted during a single jump. Net ground reaction force was calculated as the vector sum of the three force components. Changes in force in the three dimensions were detected with 5-mm aluminium foil strain gauges attached to the outer side of each spring blade. Each strain gauge,which corresponded to a separate dimension, formed a quarter of a bridge circuit that fed a signal directly into a Maclab bridge amplifier. Data were collected by a Maclab 4e (AD Instruments, Castle Hill, Australia)analog-todigital data acquisition system that sampled at 1000 Hz. Data were recorded and analysed using Chart 3.5 software. As it has previously been shown in swimming C. alboguttata that locomotor performance after nine months aestivation is indistinguishable from control levels(Hudson and Franklin, 2002a),it was assumed that the same would be true for jumping; thus, force traces from post-aestivating frogs were not obtained. For each individual, the longest of three jumps was considered to represent a maximal performance. However, we cannot exclude the possibility that variations in takeoff angle could confound the results, as Marsh(1994) has previously pointed out.

The femur and tibiofibula of control (N=8; mass=19.75±3.86 g; snout–vent length=5.60±0.22 cm, mean± s.e.m.), 3-month-aestivating (N=11;mass=23.70±4.94 g; snout–vent length=5.53±0.24 cm, data collected immediately post aestivation for both aestivation treatment groups)and 9-month-aestivating frogs (N=6; mass=19.2±5.06 g;snout–vent length=5.38±0.35 cm) were removed by disarticulation at the hip joint. The variation in sample size was a simple function of the availability of animals. The overlying soft tissues were left intact. The samples were wrapped in 0.9% salinemoistened tissue, placed in plastic vials within sealed polystyrene bags and stored frozen at –20°C. Prior to mechanical testing, the bones were thawed to room temperature (22°C),cleared of surrounding soft tissue and measured.

Each bone was rested (metaphysis to metaphysis) on rigid brass supports 19.5 mm apart and broken by three-point bending using an Instron 8722 servo-hydraulic materials testing machine (Instron, High Wycombe, UK). Three-point bending was achieved using a hydraulically driven actuator with a single point of load application equidistant between the supports. All bones were loaded at approximately mid-shaft with the posterior surface experiencing compressive loading (Fig. 1). This loading regime was chosen for practical reasons as the bones were stable in this orientation throughout the whole loading event. In vivo, the tibiofibula is most probably loaded in this way, with the femur experiencing anterior compressive loading primarily by the cruralis and gluteus magnus,coupled with a bending moment caused by the resultant ground reaction force. However, the femur is relatively cylindrical with little variation in cortical thickness; as such, the second moment of area is not influenced by the bone's orientation, assuming a neutral axis passing through the centre of the bone's cross-section. Consequently, the effect of orientation on femoral bending strength was considered minimal; either way, the orientation was the same in all tests. An actuator speed of 20 mm s^–1, used as a preliminary experiment, showed that it led to bone fracture after approximately 50 ms, comparable with the time-to-maximum force recorded for in vivo jumping of C. alboguttata. This loading rate is somewhat faster than has been used on organisms of comparable size. Correspondingly, values for stress and modulus of elasticity will be higher than those generated at less realistic loading rates(Currey, 1975).

The load cell output was pre-amplified (QUANTEC, Brisbane, Australia) and subsequently collected at 1000 Hz on a Maclab data acquisition system interfaced with a Macintosh 4e computer running Chart software (version 3.5). Load-displacement records were re-plotted using Sigmaplot 5.0 software (SPSS,Chicago, IL, USA). Values for the load and displacement at the yield and failure points were determined from these plots. The point of yield was defined as the intersection of tangents drawn to the linear elastic and plastic deformation portions of each curve. Areas under the curves, indicative of the energies of yield and of failure, were also determined from the plots.

The cross-sectional areas of the femur and tibiofibula were normalised to snout–vent length (cross-section/snout–vent length; mm²cm^–1) for the tabulated data to determine if there were any gross changes in tissue size with aestivation.

The tibiofibula and femur from two control frogs and two 9-month aestivators were fixed in 10% neutral buffered formalin, embedded in methyl methacrylate and mounted on glass slides. Serial sections (65 μm thick)were cut from the undecalcified midpoints of bones using a Leitz 1600 diamond saw microtome. For each bone, a representative section was qualitatively examined and photographed using an Olympus DP10 digital camera attached to an Olympus BX60 microscope.

Data are presented as means ± s.e.m. of Nmeasurements. The comparisons of cross-sectional area were made using analysis of covariance (ANCOVA) on the original (non-normalised) data with snout–vent length as the covariate. Comparisons of ultimate strength,moduli, second moment of area, energy of yield, energy of failure and yield displacement between the treatment groups were made using 1-way analysis of variance (ANOVA). Data were log transformed if they initially failed to meet the assumption of normality. All analyses were performed using Sigmastat and SPSS software. P<0.05 was considered significant.

The maximum jumping performance of 35 C. alboguttata weighing between 7.13 g and 30.84 g was determined at 25°C. A typical jump involved a rapid development of force resulting in a peak ground reaction force between 43 ms and 80 ms (mean ± s.e.m., 62±1.61 ms) after force development commenced (Fig. 2A). Time to peak ground reaction force scaled with body mass(r²=0.13, P<0.05, N=35) but was independent of snout–vent length (r²=0.09, P=0.081, N=35) within this size range, but it is probably inappropriate to attribute much difference between the two relationships given that the r² values are low for both comparisons(Fig. 3).

During a typical load-displacement on a control femur and tibiofibula,strain initially varied with stress linearly(Fig. 2B). When the bones yielded (after 53±3 ms for the femur and 54±1 ms for the tibiofibula), increments in strain led to smaller increases in stress. Soon thereafter (64±3 ms for the femur and 66±2 ms for the tibiofibula) the bone would fracture and stress dropped abruptly.

The ultimate strength and moduli of the tibiofibula and femur of C. alboguttata after 3-months and 9-months aestivation were compared with those of active frogs. Aestivation had no significant effect on these bending mechanical properties in the long bones(Table 1), with 9-month-aestivating frogs displaying the same resistance to bending forces as the controls (femoral ultimate strength, P=0.45; femoral modulus, P=0.32; tibiofibular ultimate strength, P=0.09; tibiofibular modulus, P=0.21). Energy of yield (femoral, P=0.91;tibiofibular, P=0.42) and failure (femoral, P=0.73;tibiofibular, P=0.59) and yield displacement (femoral, P=0.80; tibiofibular, P=0.96) were also compared, and again aestivation had no significant effect on these parameters(Table 1). Admittedly, the data do approach significance at the 0.05 level in some of the comparisons, and a higher sample size may have produced a significant difference. Having said this, previous studies in which differences following dormancy were found(e.g. Krook et al., 1977) used smaller sample sizes than the present study.

Second moment of area was compared, and aestivation had no significant effect (femoral, P=0.53; tibiofibular, P=0.68). Neither femoral nor tibiofibular cross-sectional area (normalised to snout–vent length) changed following the disuse associated with either 3- or 9-months aestivation (femoral, P=0.108; tibiofibular, P=0.063; Table 1).

Qualitative examination of the histological parameters indicated the absence of disuse osteoporosis despite 9 months of limb immobilisation. For example, the bone margins of aestivators had no ruffle border, remaining relatively smooth and bearing no evidence of osteoclastic activity(Fig. 4).Osteocytes were distributed similarly through the cortex in both controls and aestivators and there was no intracortical remodelling in any bone. Finally, the osteocytes in the aestivators were not enlarged compared with those of the controls and did not possess either basophilia or metachromasia.

Of further note was the presence of arrest/growth lines running parallel to the endosteal perimeter in the middle of the cortex in some of the specimens. This feature was evident in both the tibiofibula and femur of one of the aestivators and in the femur of one of the controls.

The biomechanical properties of a structure depend on both the materials in question and the way the material is distributed. The chronic bone unloading associated with 9-months aestivation had no measurable effect on C. alboguttata long bone morphology and mechanical properties. These data support the hypothesis that bone mechanical competence and morphology are preserved during aestivation in this species. This contrasts with both clinical systems of unloading and, more intriguingly, with the other natural disuse systems examined to date (hibernating mammals) in which bone remodelling is evident. For example, bone loss has been described in hibernating bats (Myotis lucifugus; Whalen et al., 1971), ground squirrels (Spermophilus undulatus; Mayer and Bernick, 1959) and hamsters (Cricetus cricetus; Kayser and Frank, 1963). In bats, much of the bone loss has been attributed to enhanced osteolysis(Whalen et al., 1971). However, the distribution and morphology of osteocytes in aestivating frogs indicate no osteocytic osteolysis. Given that the cross-sectional area of the bones was constant and the overall bending properties of the frog bones were unchanged, it is likely that only negligible changes occurred in the material properties of the frog bone tissue. The occasional presence of arrest lines in C. alboguttata femora and tibiofibula may indicate that the long bones undergo cyclical phases of growth and arrest similar to those found in the dentine bone of hibernating microchiroptera such as Nyctalus noctula (Klevezal and Kleinenberg,1969). If so, it is not clear whether the changes in bone histology are due to seasonal effects, aestivation or some internal physiological cycle (e.g. linked to reproduction).

The major findings of this study are consistent with the preservation of other components of the locomotor system in C. alboguttata, namely hind limb muscle masses (Hudson and Franklin, 2002a), capillary tortuosity(Hudson and Franklin 2003), in vitro contractile properties of the gastrocnemius and burst swimming performance (Hudson and Franklin,2002a).

The present study presents a comprehensive analysis of bone mechanical properties in an amphibian. The ultimate strength of the femur and tibiofibula in C. alboguttata (328 MPa and 253 MPa in the controls) corresponds favourably with values calculated for small mammals. For example, Biewener(1982) found that the chipmunk(Tamais striatus) femur and tibia had ultimate strengths of 263 MPa and 303 MPa, respectively, while the rat (Rattus norvegicus) femur and tibia showed values of 253 MPa and 233 MPa. Equally, the bending strengths of whole bones in birds are also broadly comparable, although in these cases there seems to be a greater disparity between femoral and tibiofibular data. Analysis of three-point bending in the painted quail (Excalfatoria chinensis) gave values of 311 MPa and 170 MPa for the femur and tibia,respectively, while those of the bobwhite quail (Colinus virginianus)were 193 MPa and 294 MPa (Biewener,1982).

In the present study, no unequivocal pattern emerges when comparing the strength of the femur with respect to the tibiofibula, which is consistent with some prior data on small mammals and birds(Biewener, 1982). However,greater femoral strength has previously been noted for alligators, iguana(Blob and Biewener, 1999) and horses (Currey, 1984). In these cases, one possible explanation was given in terms of the bone's position. The more distal the bone, the greater contribution it makes to the leg's moment of inertia and the more energy is required to move it(Currey, 1984). As such, in some cases the energy saved by the animal in having a lighter tibia (which is distal to the femur) might offset the `riskier' safety factor that follows the reduction in strength. This general pattern may be evident in the elongated metatarsals of frogs that are distal to the tibiofibula but were not examined in this study of C. alboguttata. Alternatively, the advantages of reducing distal limb segments may scale with mass and the lack of a pattern in these frogs may be a simple function of small body size.

Blob and Biewener (1999)have shown that magnitudes of peak ground reaction force acting on the limbs,relative to body weight, are lower in green iguanas (Iguana iguana)and American alligators (Alligator mississippiensis) than in mammals and birds. Force platform recordings indicate that peak ground reaction force magnitudes acting on a single limb range between 1.3× and 2.4×body weight for quadrupedal mammals but only 1.1× for iguanas and alligators. The peak ground reaction force during jumping for C. alboguttata is about 1.4× body weight (each limb supports half the 0.49 N force exerted during a jump; i.e. 0.49/2=0.245 N, the mass of the frog=0.018 kg or 0.17 N; thus, 0.245/0.17=1.4), falling well into the range for the ectotherms but very much at the lower end of the scale for the endotherms. Fundamentally, the ground reaction force data for C. alboguttata are in line with those found in the American alligator and green iguana, species that have high safety factors(Blob and Biewener, 2001). Moreover, a relatively high safety factor was shown experimentally by Calow and Alexander (1973) for a different species of anuran, the European common frog Rana temporaria. Accordingly, it seems probable that C. alboguttatahave a high safety factor, although one cannot be accurately calculated from the present data without detailed strain gauge data or frog bone strains in vivo. The fact that bone strength could presumably be lost in these frogs without undue risk of fracture, but is not, points to a simple,inherent preservation mechanism.

Bradymetabolic organisms are relatively quiescent, which means prior to immobilisation their bones have only a moderate loading history. Assuming that the rate of remodelling depends on the extent of unloading (sensuJee and Ma, 1999), bone tissue of relatively inactive organisms (such as amphibians) should be intrinsically more resistant to disuse osteoporosis than more active, tachymetabolic organisms (such as mammals) because the change in stimulus is that much weaker. In a sense, it could be argued that burrowing frogs are predisposed to withstand lengthy periods of disuse without losing bone mass. We have previously invoked this argument to partly explain the preservation of skeletal muscle structure in aestivating C. alboguttata(Hudson and Franklin, 2002b). In light of this predisposition, the pertinence of C. alboguttata as a model system for biomedical studies of disuse osteoporosis is reduced. This logic points clearly to the ideal biomedical system: a tachymetabolic mammal that suffers no disuse osteoporosis during natural disuse.

Nevertheless, in mammals, a period of disuse produces a fairly rapid response involving the loss of bone and there may be compensatory cellular mechanisms in operation in aestivating frogs to prevent bone loss. In hibernating mammals, parafollicular cells in the thyroid gland are enlarged and active compared with those in non-hibernators(Whalen et al., 1971). Parafollicular cells secrete calcitonin, an inhibitor of bone resorption. In hibernating bats, the parafollicular cells show striking seasonal changes in structure. During the first half of hibernation, the amount of granular endoplasmic reticulum is reduced and they lose their solid dense core,suggesting that calcitonin may be functionally inactive during hibernation(Whalen et al., 1971). This is consistent with the bone resorption observed in these animals. Intriguingly,Krook et al. (1977) showed that injections of calcitonin could prevent this bone loss in hibernating bats and argued that this was reflected in smaller femoral osteocytes compared with those in untreated hibernators. Unfortunately, the endocrinology of aestivating frogs has been neglected and its study may provide us with a better understanding of the effects of aestivation.

Aestivating C. alboguttata can emerge within 24 h of the first heavy summer rainfall (N.J.H., personal observation). Indeed, the emergence process itself is presumably demanding on the skeletal system as it involves digging through wet clay. The frogs immediately engage in predator avoidance strategies that require a fully competent locomotor system such as burst swims and powerful jumps. The latter transmits considerable forces through the long bones of the hindlimbs during the leg extension phase. Given that post-aestivation locomotor performance in swimming matches control levels(Hudson and Franklin, 2002a)and that frogs may actually gain mass during aestivation (N.J.H., unpublished data), the transmission of forces through the hindlimbs in frogs after dormancy is almost certainly equivalent to that in control frogs. The maintenance of bone mechanical properties throughout aestivation facilitates performance of these activities by ensuring that an adequate capacity for load bearing is maintained.

We thank Mark Forwood for advice and support in bone morphometry. C.E.F. wishes to acknowledge the ARC for a large research grant. N.J.H. would like to acknowledge the Northcote trust from which he received a generous postgraduate scholarship and Peter Kraft for technical advice.