Scaling of left ventricle cardiomyocyte ultrastructure across development in the kangaroo Macropus fuliginosus

The four-chambered mammalian heart completely separates the high-pressure systemic system that supplies oxygenated blood to the body from the low-pressure pulmonary system that sends deoxygenated blood to the lungs. By operating two cardiac circuits at vastly different pressures, and by preventing oxygenated and deoxygenated blood from mixing, mammals are equipped with an efficient, high-output, oxygen delivery system. The effectiveness of the mammalian heart as a pump is evident in that the same fundamental design meets the requirements of mammals spanning eight orders of magnitude in body mass, from a 1.9 g Etruscan shrew to a 190,000 kg blue whale. Over this size range, volume loads on the heart vary significantly, with absolute cardiac output and whole-body metabolic rate increasing systematically with body mass. In terrestrial mammals, pressure loads on the heart also increase with body size, as larger and, particularly, taller species support a higher vertical column of blood above the heart (Seymour and Blaylock, 2000; White and Seymour, 2014). According to the principle of Laplace, as volume loads and pressure loads on the heart increase with body mass, so too must the thickness of the ventricular wall, ensuring that wall stress (N m⁻²) is conserved and the load is spread (Grande and Taylor, 1965). This increase in wall thickness is ultimately realised as an increase in heart mass, which has been the subject of numerous allometric investigations.

An allometric analysis of heart mass (M_h) against body mass (M_b) is generally expressed in the form of a power equation, M_h=a×M_b^b, where a is the scaling coefficient (elevation of log-transformed equation) and b is the scaling exponent (slope of log-transformed equation). If b=1 then heart mass increases in direct proportion to body mass (isometric), if b>1 then heart mass increases faster than body mass (positive allometry or hyperallometry), and if b<1 then heart mass increases slower than body mass (negative allometry or hypoallometry). Phylogenetic analysis across a range of different-sized mammalian species shows that heart mass increases with body mass more-or-less isometrically, with an exponent (b) between 0.96 and 1.06 (Bishop, 1997; Brody, 1945; Holt et al., 1968; Hoppeler et al., 1984; Lindstedt and Schaeffer, 2002; Prothero, 1979; Seymour and Blaylock, 2000; Stahl, 1965). Analyses restricted to marsupial species also trend toward isometry, with an exponent between 0.94 and 1.05 (Dawson and Needham, 1981; Dawson et al., 2003). However, the scaling coefficient (a) for marsupials is approximately 30% larger than that for placental mammals, presumably owing to the marsupial heart being approximately 30% larger.

Ontogenetic analyses of heart mass at different stages of development within a single species of mammal, in contrast, show more variable results, with exponents ranging from 0.74 to 0.82 in rats (Stewart and German, 1999; von Bertalanffy and Pirozynski, 1952) and 1.00 to 1.02 in humans (de Simone et al., 1998; Mühlmann, 1927; Thompson, 1942). This variation is partly due to species-specific differences in the trajectory of heart growth after parturition, and partly due to the narrow postnatal body mass range of placental mammals (ca. 20- to 50-fold), over which most ontogenetic analyses are restricted. The range of body mass can be increased by incorporating both prenatal and postnatal phases of development (ca. 600-fold in placental mammals), and the few studies that have done this report a biphasic scaling pattern in cardiac growth. For instance, human heart mass increases with body mass during prenatal development with an exponent of 1.19, but following parturition, heart mass increases with body mass with an exponent of 0.89 (Hirokawa, 1972). Similarly, prenatal heart mass in the giraffe Giraffa camelopardalis scales with an exponent of 1.03, whereas postnatal heart mass scales with a shallower exponent of 0.90 (Mitchell and Skinner, 2009). A broader ontogenetic body mass range can also be obtained from marsupial mammal species, where neonates are highly altricial, and much of development occurs within the maternal pouch (Renfree, 2006; Tyndale-Biscoe and Janssens, 1988). In a recent study, we found that heart mass in the marsupial western grey kangaroo Macropus fuliginosus also follows a biphasic allometric pattern across ontogeny, scaling with a relatively steep exponent of 1.10 during in-pouch development, before switching to a shallower exponent of 0.77 during post-pouch development (Snelling et al., 2015). The biphasic allometry of cardiac growth in the marsupial kangaroo before and after leaving the pouch and the biphasic allometry reported for placental species before and after parturition suggest that key life history events that involve a change in the animal's behaviour, physiology and activity levels are reflected in the scaling of the heart.

Clearly, the allometry of heart mass provides important insight into the blood supply requirements of the body at different stages of development. However, another important factor likely to significantly affect the functional performance of the heart, but not evident from simple allometry of heart mass, is the gradual maturation of the cardiac ultrastructure during development. While species-specific differences exist in the level of maturity of the myocardium at parturition, there is nonetheless a clear increase in the volume-density of key organelles within the cardiac muscle, much of which occurs during the fetal and early postnatal periods (Canale et al., 1986; Smolich, 1995). For instance, the volume-density of the muscle's myofibril contractile machinery increases significantly across this period of development in the rat, guinea pig, hamster, rabbit and sheep (Colgan et al., 1978; Friedman, 1972; Hirakow and Gotoh, 1976,, 1980; Smith and Page, 1977). Presumably to support the growing demand for ATP, the volume-density of mitochondria also increases during the fetal and early postnatal periods, as shown in the rat, guinea pig, hamster and rabbit (Colgan et al., 1978; Hirakow and Gotoh, 1976,, 1980; Hirakow et al., 1980; Smith and Page, 1977). Similarly, the sarcoplasmic reticulum, which allows for myofibril contraction by the release of its Ca²⁺ stores into the cytoplasm, increases in volume-density during the fetal and early postnatal periods, as evident from studies on the rat, hamster and rabbit (Colgan et al., 1978; Hoerter et al., 1981; Olivetti et al., 1980). Finally, the T-tubule elements, which function in the propagation of action potentials deep within the cardiomyocyte, increase in volume-density as cardiomyocyte diameter increases and the diffusion distance to the interior increases (Hirakow and Gotoh, 1976,, 1980). It is important to understand that the increase in the volume-density of these key organelles occurs in addition to the absolute increase in myocardial mass (absolute volume). The changing cardiomyocyte ultrastructure and cardiac tissue mass during ontogeny would therefore influence cardiac performance on a cellular level, and as a functioning organ. In support of this, measurements of cardiac function in rat, rabbit and sheep show maturation of the myocardial ultrastructure during development is associated with a concomitant increase in the contractile performance of isolated myocardium and of the organ as a whole (Anderson, 1996; Friedman, 1972; Hopkins et al., 1973; Nakanishi and Jarmakani, 1984; Romero et al., 1972).

Despite all the work done to quantify the change in cardiac ultrastructure at different stages of development, a thorough allometric analysis across key life history stages within a species has not yet been undertaken for any mammal. An allometric examination is crucial because it is the only means by which the overwhelming influence of body mass can be properly accounted (Calder, 1996). Because the changes in body mass between birth and adulthood span over two orders of magnitude in kangaroos, it is possible to define shifts in developmental trajectories more clearly than in placental species. For example, in our recent study, we showed that the exponent for heart mass in the western grey kangaroo decreases significantly around the time of pouch vacation (Snelling et al., 2015). In the present study, we used a subset of these individuals, encompassing a wide range of developmental stages and body masses, to analyse the development of the ultrastructure of the left ventricle. The left ventricle is responsible for the delivery of oxygen around the body via the systemic circuit. Given that significant cardiomyocyte maturation occurs during the fetal and early postnatal life of placental mammals, we present the hypothesis that the volume-density of key organelles within cardiomyocytes of the left ventricle – the myofibrils, mitochondria, sarcoplasmic reticuli and T-tubules – increases with hyperallometry during in-pouch development, but scales independently of body mass across post-pouch development. In accordance with the principle of Laplace, we also hypothesise that the left ventricle's wall-to-lumen volume ratio, and myofibril volume-density, might scale in a manner that ensures wall stress is constant across development, which would agree with a phylogenetic study of wall stress in the left ventricle of mammals (Seymour and Blaylock, 2000).

Western grey kangaroos selected for this study varied 190-fold in body mass (0.355−67.5 kg, N=16). Detailed regression and ANCOVA statistics are provided in Table 1. The left ventricle exhibits biphasic allometric growth, whereby myocardial mass (and volume) scales hyperallometrically across pouched young development (hereafter termed ‘in-pouch’; N=6), with an exponent of 1.18±0.05, before switching to a hypoallometric trajectory in free-roaming young-at-foot, juvenile and adult kangaroos (hereafter ‘post-pouch’; N=10), with an exponent of 0.78±0.13 (Fig. 1). The left ventricle inner and outer radii, and wall thickness, show the same biphasic scaling pattern (Table 1). Breakpoint analysis (see Materials and methods) confirms that the in-pouch and post-pouch allometric functions separate at approximately 5–6 kg body mass, which corresponds to the body mass at which individuals of this species typically vacate the pouch permanently (Dawson, 2012).

A number of conspicuous changes also occur at the cellular and subcellular levels of the left ventricle during development (Fig. 2). During in-pouch growth, the volume-density (%) of cardiomyocytes increases, scaling with an exponent of 0.07±0.03 (Table 1), and this contributes to a rather steep exponent for the scaling of the absolute volume (ml) of cardiomyocytes, 1.25±0.06 (Fig. 3A). The volume-density (%) of key components of the cardiomyocyte cell also increases to varying degrees during in-pouch development, with exponents of 0.13±0.06 for myofibrils, 0.04±0.03 for mitochondria, 0.20±0.11 for sarcoplasmic reticuli and 0.62±0.25 for T-tubules (Table 1) (ANCOVA, F_3,16=24.5, P<0.0001). The increasing volume-density of these organelles within the cardiomyocytes, combined with the increasing volume-density of cardiomyocytes within the left ventricle, and the increasing relative mass of the left ventricle, results in steep exponents across in-pouch development for the absolute volume (ml) of the myofibrils 1.38±0.10 (Fig. 3B), mitochondria 1.29±0.08 (Fig. 3C), sarcoplasmic reticuli 1.45±0.13 (Fig. 3D) and T-tubules 1.87±0.31 (Fig. 3E) (ANCOVA, F_3,16=15.8, P<0.0001). The increasing prominence of these organelles across in-pouch development is in contrast to that of other cellular elements (primarily nuclei and cytosol), which decrease in volume-density (%) with an exponent of −0.44±0.12 (Table 1), such that absolute volume (ml) increases with a relatively shallow exponent of 0.80±0.06 (Fig. 3F).

During post-pouch development, the volume-density (%) of cardiomyocytes within the left ventricle tissue does not vary, scaling with an exponent of 0.00±0.04 (Table 1), and so absolute cardiomyocyte volume (ml) scales in parallel with left ventricle mass, with an identical exponent of 0.78±0.15 (Fig. 3A). The volume-density (%) of key organelles within the cardiomyocytes also scales invariantly with post-pouch body mass, with exponents of 0.01±0.03 for myofibrils, −0.01±0.06 for mitochondria, 0.01±0.09 for sarcoplasmic reticuli and 0.01±0.21 for T-tubules (Table 1) (ANCOVA, F_3,32=0.026, P=0.99). Thus, the allometric exponents for the absolute volume (ml) of the myofibrils 0.79±0.15 (Fig. 3B), mitochondria 0.77±0.17 (Fig. 3C), sarcoplasmic reticuli 0.79±0.20 (Fig. 3D) and T-tubules 0.79±0.30 (Fig. 3E) are all statistically indistinguishable from one another across post-pouch development (ANCOVA, F_3,32=0.0082, P=1.00). The combined volume-density (%) of other cellular elements also scales with post-pouch body mass with an exponent not significantly different from zero, −0.05±0.32 (Table 1), and the absolute volume (ml) scales with an exponent of 0.73±0.27 (Fig. 3F). Given that organelle volume-densities do not vary systematically with body mass in post-pouch kangaroos, we can provide mean values, which are: 63.7±1.3% for myofibrils, 22.7±0.8% for mitochondria, 2.9±0.1% for sarcoplasmic reticuli, 4.9±0.6% for T-tubules and 5.8±1.0% for the combined sum of other less abundant cellular elements.

All animals experience life history events during development that have important consequences for physiology, activity and behaviour. Inevitably, such events also have important consequences for anatomy. For instance, birth in placental mammals necessitates a shift from placental to pulmonary gas exchange, which requires significant cardiovascular remodelling so that the left and right ventricles can transform from working in-parallel at similar pressures, to working in-series at vastly different pressures owing to different resistances of the systemic and pulmonary circuits (Stopfkuchen, 1987). This event is reflected in allometric studies of heart mass growth, with placental mammals tending to show different scaling exponents before and after parturition (Hirokawa, 1972; Mitchell and Skinner, 2009).

In marsupial mammals, cardiovascular remodelling also takes place at birth, but this happens much earlier in development than it does in placental mammals (Runciman et al., 1995). While birth in marsupials involves the energetic demands of crawling from the birth canal to the pouch, it has been suggested that gas exchange is predominantly via the skin at this stage (Mortola et al., 1999). Perhaps the most important cardiovascular life history event for marsupials occurs when the emergent young vacates the maternal pouch. In kangaroos and other marsupials, the period leading up to permanent pouch exit is associated with the development of endothermy and an increase in metabolic rate (Hulbert, 1988; Rose et al., 1998). Around the same time, kangaroos make brief excursions from the pouch, which continues until the animal is capable of sustained and independent locomotion (Dawson, 2012). Reflecting this change in physiology, activity and behaviour, we recently showed that heart mass growth in the kangaroo shifts allometric trajectory around the time of pouch exit (Snelling et al., 2015). A biphasic scaling pattern is further apparent in the subset of animals selected for the present study, where left ventricle mass increases with a relatively steep exponent of 1.18 during in-pouch development, before shifting to a relatively shallow exponent of 0.78 across post-pouch development (Fig. 1).

In addition to the biphasic scaling of left ventricle mass across kangaroo development, it is apparent that cardiac growth is associated with significant changes at the cellular and subcellular levels (Figs 2, 3). In particular, in-pouch development is accompanied by significant changes to cardiomyocyte composition as the cells mature to a post-pouch-like state. The volume-density of myofibrils, mitochondria, sarcoplasmic reticuli and T-tubules within the cardiomyocytes increases significantly during in-pouch growth (Table 1), which is entirely consistent with the increasing volume-densities of these organelles in placental mammals during the fetal and early postnatal periods of development (Canale et al., 1986; Smolich, 1995). The volume-density and absolute volume of the muscle's contractile machinery, the myofibrils, increases with steep hyperallometry across in-pouch development (exponents of 0.13 and 1.38, respectively), which suggests that the left ventricle's contractile performance, and capacity to generate pressure and blood flow, also increases across this period. Consistent with this increase in myofibril volume-density during early development in kangaroos and placental mammals, the tension (=stress; N m⁻²) developed by isolated cardiac muscle performing isometric (fixed muscle length) contraction is greater in adult sheep compared with fetal lambs within a range of pre-determined sarcomere lengths, over which the extent of actin–myosin cross-bridging varies. In addition, maximum isometric tension, elicited at sarcomere lengths that maximise cross-bridging, is approximately 60% greater in the adult compared with the fetal myocardium (Friedman, 1972). Likewise, measurements taken from isolated sheep myocardium performing isotonic (changing muscle length) contraction shows that both the extent and velocity of shortening are greater in the adult than the fetal heart over a range of applied loads (Friedman, 1972). From a whole-organ perspective, the increasing volume-density of myofibrils is consistent with the observation that stroke volume is significantly depressed in neonate lambs at afterloads that are modest for adults (Downing et al., 1965). From a whole-animal perspective, the hyperallometry of myofibril volume-density and absolute volume, and the associated increase in relative cardiac contractile performance, would augment cardiac output during in-pouch development and in the lead up to pouch vacation, and thus almost certainly facilitate the kangaroo's transition from ectothermy to endothermy (Hulbert, 1988; Loudon et al., 1985; Rose et al., 1998) and independent locomotion (Dawson, 2012).

During contraction–relaxation cycles in skeletal muscle, approximately 70% of the total energy harnessed from ATP hydrolysis is used by myofibril ATPase, while the remaining 30% is used by the sarcoplasmic reticulum for Ca²⁺ uptake and by the sarcolemma for the maintenance of ion gradients (Hoppeler and Billeter, 1991; Rall, 1985). Presumably, similar fractions apply to cardiac muscle. The bulk of this ATP energy is generated from oxidative phosphorylation, which takes place within the cell's mitochondria. Thus, the increasing volume-density of myofibrils during in-pouch development will only translate to improved contractile performance provided sufficient mitochondria are present to supply the ATP. Thus, concomitant with the increasing myofibril density, the volume-density and absolute volume of mitochondria also increase with hyperallometry during in-pouch growth (exponents of 0.04 and 1.29, respectively). Nonetheless, the exponent for mitochondria is not quite as steep as that for myofibrils, such that the mitochondria:myofibril ratio decreases across in-pouch body mass, with an exponent of −0.09. This means that younger in-pouch individuals have more mitochondria relative to myofibrils compared to older in-pouch individuals. It is feasible that in the younger in-pouch kangaroos, other energy-intensive processes might be operating that demand mitochondrial ATP, in addition to those involved in the contraction–relaxation cycle. During early development, particularly when muscle contractions are weak and growth is rapid, an important role of mitochondria would be to supply ATP energy for cardiomyocyte growth. The decreasing mitochondria:myofibril ratio during in-pouch development might also be explained by the apparent increase in mitochondrial cristae surface-density over this period; however, the myofibrils also exhibit structural maturation early in development, whereby the myofilaments become organised and aligned (Canale et al., 1986). Measurements of myofilament anisotropy and mitochondrial cristae surface-density across development are an avenue for future research.

The increasing volume-density of myofibrils and mitochondria across in-pouch development is accompanied by an even steeper increase in the volume-density and absolute volume of the sarcoplasmic reticulum (exponents of 0.20 and 1.45, respectively). The steep hyperallometry of the sarcoplasmic reticulum probably arises in part to function alongside the increasing myofibril component, with which there is an intimate functional association, and in part to counter the increase in diffusion distance associated with increasing cardiomyocyte diameter with growth. The in-pouch period of kangaroo development is also associated with a very steep increase in the volume-density and absolute volume of the T-tubule system (exponents of 0.62 and 1.87, respectively). T-tubules are invaginations of the sarcolemma that function to send action potentials deep within the cardiomyocyte. In placental mammals, T-tubules are absent in very early fetal cardiomyocytes, and only begin to form later in fetal and postnatal development (Canale et al., 1986; Smolich, 1995). There is a close association between the T-tubule system and the sarcoplasmic reticulum, reflecting their functional association in the release of Ca²⁺ that then triggers cardiomyocyte contraction. Sarcolemma depolarisation activates L-type channels that permit the passive movement of Ca²⁺ into the cardiomyocyte, which then triggers Ca²⁺ release from stores within the sarcoplasmic reticulum (Fabiato, 1989; Näbauer et al., 1989). The cytosolic Ca²⁺ binds with troponin, which alters the tropomyosin arrangement, so that the myosin head and actin filament can bind, and myofibrillar contraction can proceed. When cardiomyocytes are small and immature, trans-sarcolemmal influx of Ca²⁺ alone appears largely sufficient to enable the relatively modest tensions (N m⁻²) to develop. However, as cardiomyocytes mature, increasing cardiomyocyte size and diffusion distance, and greater developed tensions, necessitate an increasing reliance on the T-tubule system to activate the increasingly necessary intracellular Ca²⁺ stores of the sarcoplasmic reticulum (Anderson, 1996; Fisher, 1994). In support of this, experimental evidence shows that decreased extracellular [Ca²⁺] has a more significant effect on the contractile function of neonatal rabbit cardiomyocytes compared with adults, while ryanodine-interference of sarcoplasmic reticuli reduces contraction amplitude in adults more than it does in neonates (Chin et al., 1990). It therefore seems likely that both the increase in myofibril volume-density and associated calcium requirements, as well as the increase in cardiomyocyte diameter, contribute to the steep hyperallometry of the sarcoplasmic reticulum and T-tubule system across in-pouch development.

The hyperallometry of myofibril, mitochondrial, sarcoplasmic reticulum and T-tubule volume-densities during in-pouch development ensures that cardiac architecture is fully developed by the time of permanent pouch exit as the kangaroo must now support its own locomotion. It is clear that ultrastructural maturity has been attained in the emergent young, because the volume-density of these key organelles does not continue to increase beyond the in-pouch stage, but rather scales invariantly with body mass after pouch exit (Table 1). This invariant scaling means that their respective absolute volumes scale in parallel with left ventricle mass, with a relatively shallow exponent of approximately 0.78. Hypoallometry of heart mass, and presumably left ventricle mass, across postnatal development in placental mammals is not uncommon; for instance, heart mass in the rat scales with an exponent between 0.74 and 0.82 (Stewart and German, 1999; von Bertalanffy and Pirozynski, 1952). Nonetheless, we show for the first time, at least in the post-pouch western grey kangaroo, that no ultrastructural changes occur on the subcellular level that in any way compensate for the shallow exponent. Thus, it appears that a decrease in relative cardiac requirements occurs as body mass increases in these free-roaming kangaroos. This could be related to the fact that hopping locomotion is energetically more efficient as body size increases, as larger macropods have a greater capacity to store energy in the tendons, ligaments and muscles of the hindlimbs and tail between each hop (Baudinette et al., 1992; Bennett, 2000; Dawson and Taylor, 1973). While pentapedal locomotion is expensive relative to the low speeds over which the gait is employed, the total metabolic cost of hopping is nevertheless higher, despite its apparent efficiency (Dawson and Taylor, 1973). It has been hypothesised that if the efficiency of hopping locomotion increases as body mass increases ontogenetically, then it could unload some of the work of the heart as body mass increases in growing post-pouch western grey kangaroos (Snelling et al., 2015). The shallow post-pouch exponent might also reflect generally lower maximum activity levels in larger kangaroos due to old age, because growth appears persistent across the lifetime of medium- and large-sized species of macropod (Jarman, 1983; Sadleir, 1965). The shallow exponent might also be related to the decreasing surface area-to-body mass ratio associated with growth, which could reduce the cardiac output required for the maintenance of body temperature, especially when ambient temperatures drop well below the thermal neutral zone.

The principle of Laplace has been used by Seymour and Blaylock (2000) to explain the allometric scaling of left ventricle mass and geometry across 24 species of mammal. Their study showed that pressure loading on the left ventricle increases in a systematic manner with body mass (volume loading increases in direct proportion), which appears to necessitate a graded increase in relative wall thickness to ensure wall stress is conserved at approximately 8–20 kPa. Applying the principle of Laplace to the present ontogenetic situation is made complicated because we do not know how pressure loading on the left ventricle changes throughout development in a marsupial. Nonetheless, we do know how the volume-density of the myofibril contractile machinery in the left ventricle tissue changes across development in the kangaroo (Table 1) and we know that the left ventricle of a mature post-pouch kangaroo (myofibril volume-density of tissue Vv_mf,t=55%) generates a mean arterial pressure of 12.3 kPa (92 mmHg) under light anaesthesia (Maxwell et al., 1964). Myocardial contractile force is dependent on myofibril volume-density (Friedman, 1972), and so if the radius of curvature and wall thickness are held constant, and a linear relationship is assumed between mean arterial pressure (P; kPa) and myofibril volume-density of the ventricle tissue (Vv_mf,t; %), an adjusted pressure value can be calculated using the function P=0.221×Vv_mf,t. The adjusted mean arterial pressure, together with left ventricle wall volume (V_w; ml) and lumen volume (V_l; ml), can then be used to calculate mean fibre stress across the left ventricle wall (σ_f; kPa) according to the principle of Laplace, using the spherical model presented by Arts and colleagues, σ_f=P×[1/3 ln(1+V_w/V_l)]⁻¹ (Arts et al., 1991). The results of this analysis are presented in Fig. 4. Because left ventricle wall and lumen volume scale more-or-less in parallel with one another across both in-pouch and post-pouch stages of development, wall stress varies as a function of mean arterial pressure and myofibril volume-density. Thus, wall stress increases significantly with body mass (M_b; kg) during in-pouch development, 13.0M_b^0.18±0.15, concomitant with the increasing myofibril fraction of the myocardium. In contrast, wall stress scales independent of body mass across post-pouch development, 19.0M_b^{−0.01±0.12}, as the invariant scaling of myofibril volume-density keeps the calculated pressure loading constant.

In summary, the growth and development of the left ventricle in the western grey kangaroo is characterised by two distinct allometric patterns that shift at approximately 5–6 kg body mass, which coincides with pouch vacation in this species. During in-pouch development, the mass of the left ventricle and the absolute volume of the myofibrils, mitochondria, sarcoplasmic reticuli and T-tubules all increase with steep hyperallometry. Furthermore, calculations of wall stress show that it too increases significantly during this period of development. These changes are associated with a significant increase in relative contractile performance, which probably facilitates the development of endothermy and independent locomotion in emergent young as they prepare to leave the pouch. Upon leaving the pouch, cardiac ultrastructure has reached maturity, and the volume-density of the key organelles together with wall stress scale independent of post-pouch body mass. Nonetheless, the size of the left ventricle and the absolute volume of these organelles scale with relatively shallow hypoallometry. This implies that post-pouch kangaroos have reduced cardiac requirements as body mass increases, which could be because the energy required for hopping is independent of speed, and the capacity for energy storage during hopping could increase as the kangaroo grows. This idea requires formal testing. We therefore validate our hypothesis that the volume-density of key organelles within the left ventricle's cardiomyocytes increases with hyperallometry during in-pouch development, but scales independently of body mass across post-pouch development. However, despite the left ventricle's wall-to-lumen volume ratio remaining constant throughout development, the increasing myofibril volume-density across in-pouch growth is probably associated with an increasing capacity to produce pressure, which leads to increasing stress across the left ventricle wall as the young develop in the pouch.

We joined a planned management cull of western grey kangaroos Macropus fuliginosus melanops (Desmarest 1817), 80 km south of Adelaide, Australia. A licensed marksman shot all animals according to Australian legal requirements and code of practice (www.environment.gov.au). Approval to scavenge organ and tissue samples was provided by the University of Adelaide Animal Ethics Committee (S-2011-223). An allometric analysis of whole-heart and chamber masses in these animals has been published (Snelling et al., 2015). In the present study, a subset of these animals was selected for a more detailed investigation into the allometry of left ventricle geometry and cardiomyocyte ultrastructure, comprising six ‘in-pouch’ kangaroos (pouched young) and 10 free-roaming ‘post-pouch’ kangaroos (young-at-foot, juveniles and adults). The animals were selected so that a comprehensive range of development and body mass was obtained.

Each carcass was weighed by spring balance (2.0, 5.0, 20 or 100 kg capacity; Salter, Australia) before the thoracic cavity was opened, and the whole heart removed, emptied of blood, and rinsed clean with an isotonic saline solution (0.90% w/v NaCl). The left ventricle, defined as left ventricular free wall+interventricular septum (Fulton et al., 1952; Joyce et al., 2004; Keen, 1955), was separated from the remainder of the heart and weighed to either 0.1 mg (ventricles<30 g; AE163, Mettler, Greifensee, Switzerland) or 1.0 mg (>30 g; 1265 MP, Sartorius, Göttingen, Germany). The left ventricle was bisected along the mid-equator, and random linear measurements of transmural wall thickness (×4), and inner (×2) and outer (×2) diameters were taken with digital callipers to 0.01 mm. Left ventricle lumen volume (ml) was calculated as 4/3×π×R_i³, where R_i is mean inner radius (cm).

Unbiased estimates of the relative composition of the left ventricle's ultrastructure were obtained by generating isotropic uniform random images for stereological analysis (Howard and Reed, 1998; Mühlfeld et al., 2010). Soon after the heart was extracted from the thoracic cavity, ∼1 mm³ samples of cardiac muscle were excised from two random locations within the left ventricle wall. The samples were immediately immersed in a chemical fixative solution of 3% glutaraldehyde and 3% formaldehyde in 0.1 mol l⁻¹ phosphate buffer (pH 7.4), and left overnight at 6°C. The following day, each sample was given a series of buffer rinses (6×10 min) followed by secondary fixation in a 2% aqueous solution of osmium tetroxide (4 h). Next, each sample was given a series of distilled water rinses (3×20 min) before progressive dehydration in ethyl alcohol in 10% incremental steps starting from 50% to 80% (10 min each) followed by consecutive immersions in 90% ethanol (2×10 min), 100% ethanol (2×10 min) and finally pure propylene oxide (2×10 min). Samples were then incrementally infiltrated with embedding resin (Durcupan, Fluka, Switzerland) at ratios of 3:1, 2:2 and 1:3 (propylene oxide:resin), each for a duration of 2 h, and then left overnight at room temperature in pure resin. Each sample was then placed in a random orientation in individual embedding moulds where they were covered with pure embedding resin and left to polymerise in a 70°C oven for 48 h.

One randomly oriented 70 nm ultrathin section was cut at a random distance into each sample using 8 mm glass knives, a 2.4 mm diamond knife (Ultra 45°, Diatome, Switzerland) and an ultramicrotome (EM UC6, Leica Microsystems, Germany). Ultrathin sections were placed onto 3 mm copper mesh grids, stained with uranyl acetate (15 min) and lead citrate (10 min), and viewed with a 120 kV transmission electron microscope (Tecnai G2, FEI, USA) coupled to an in-column CCD digital camera (Veleta, Olympus, Japan) running the bundled TIA imaging platform. From each section, 10 low magnification images (×1100–2550) of random left ventricle tissue and 10 high magnification images (×7900–20,500) of random cardiomyocyte ultrastructure were captured and digitally stored. Magnification was adjusted to account for differences in cardiomyocyte size across development. In total, 640 random images were taken from 32 randomly oriented sections, cut from 32 random samples sourced from the left ventricle of 16 kangaroos.

where Vv_org,cm is the fraction of cardiomyocyte occupied by the organelle (or the sum of other cellular elements) determined from point grid counts and V_cm is the total left ventricle cardiomyocyte volume (ml) calculated in Eqn 1.

All mean values are presented with 95% confidence intervals. Body mass is expressed in kg, heart and left ventricle mass are in g, linear dimensions are in mm, volumes are in ml and volume-densities are expressed as percentages. Allometric relationships were derived by taking the log₁₀ of the variable and of the body mass, and calculating ordinary least-squares linear regressions. The slopes and intercepts of the regressions were compared with ANCOVA, with the morphometric variable of interest set as the dependent, and body mass as the covariate (Zar, 1998), using statistical software (GraphPad Prism 6, GraphPad Software, USA). To determine whether biphasic allometry existed in the dataset, a breakpoint analysis was performed by fitting a series of two-phase linear regressions to the log₁₀-transformed data in which the intersection point was shifted along consecutive readings. The breakpoint was identified as the intersection that minimised the sum of the regressions' residual sums of squares (Mueller and Seymour, 2011; Yeager and Ultsch, 1989).

The authors acknowledge exceptional and humane marksmanship exhibited during the management cull and we thank the marksman for allowing us access to carcasses. We thank Dr David McLelland from Zoos South Australia who engaged in discussions on the scaling of heart size. We are indebted to volunteers who assisted in the field. We thank John Snelling from the University of Adelaide for donation of a utility vehicle to collect carcasses. Two reviewers provided valuable feedback on the manuscript.