By continuing to browse this site you agree to us using cookies as described in About Cookies
Wiley Online Library is migrating to a new platform powered by Atypon, the leading provider of scholarly publishing platforms. The new Wiley Online Library will be migrated over the weekend of February 24 & 25 and will be live on February 26, 2018. For more information, please visit our migration page: http://www.wileyactual.com/WOLMigration/
The giraffe heart generates twice the pressure of similar-sized mammals under normal physiological conditions (Goetz and Budtz-Olsen O., 1955; Van Citters et al., 1966; Brøndum et al., 2009) with a relative ventricular mass no different from other mammals, that is, 0.5% of body mass (Fig. 1) (Brøndum et al., 2009; Mitchell and Skinner, 2009). The overall ventricular anatomy resembles other mammals (Brøndum et al., 2009; Mitchell and Skinner, 2009), but the left ventricle (LV) wall is significantly thicker compared to other mammals (Bie and Damkjær, 2011). These structural characteristics most likely serve to normalize wall tension despite the high pressure, but logically also results in smaller cavity volumes and consequently a smaller stroke volume and, as heart rate is normal for its body mass (Brøndum et al., 2009), also results in lower relative cardiac output (Bie and Damkjær, 2011; Smerup et al., unpublished data). The evolutionary adaptation of the giraffe LV to high pressure presumably differs from LV remodelling in hypertensive humans (Anversa et al., 1993; Kajstura et al., 1994; Richey and Brown, 1998; Anversa and Olivetti, 2011). Thus, while the LV hypertrophy of hypertensive humans also normalizes wall stress and hence appear adaptive, there are numerous concomitant pathological changes that impair normal ventricular function (Richey and Brown, 1998). In particular, when side-to-side slippage of myocytes occurs, myocardial scarring and collagen accumulation reduce compliance of the ventricular wall that decreases end diastolic volume and filling of the ventricle (Anversa et al., 1993; Beltrami et al., 1994; Weber et al., 1994). Hypertensive remodelling also alters structural properties of the coronary arteries, lowering perfusion and myocardial oxygen delivery (Anversa and Sonnenblick, 1990; Parodi et al., 1993). These structural changes increase the risk of cardiovascular diseases, often initiated with endothelial dysfunction in arteries, which is the first step of atherosclerosis (Kumar et al., 2004).
The high pressure in the giraffe is a normal physiological condition, in contrast to the pathophysiological changes during hypertension, and consequently, structural changes like myocardial fibrosis of the giraffe heart would not be expected. However, evolutionary structural changes would be expected.
While a few studies have characterized the overall morphology of the thick-walled giraffe ventricle (Goetz and Keen, 1957; Goetz et al., 1960; Badeer, 1986; Hargens et al., 1988; Brook and Pedley, 2002; Mitchell and Skinner, 2009), little is known about the morphology, number, and arrangement of cardiomyocytes. To understand the evolutionary adaptations of the giraffe heart, we used stereological methods to quantitatively characterize the LV morphology. Although previous studies in other mammals, including humans, have shown that the overall myocardial architecture in terms of orientation and linking of the myocytes varies greatly depending upon both myocardial depth and region (Smerup et al., ), there is no evidence so far to suggest that the individual LV myocytes vary in size and structure, regardless of position and orientation (Anderson et al., 2009). Using the stereological principles of systematic, uniform and random sampling in the entirety of the LV myocardial mass, we aimed to measure the volume, numerical density and total number of cardiomyocytes, the number of cardiomyocyte nuclei and non-cardiomyocyte nuclei, as well as the mean number of nuclei per myocyte. Number estimation is independent of orientation and using the disector for counting ensures unbiasedness (Sterio, 1984). These parameters are both relevant in clinical studies of development and diseases, and in comparative physiology (Tang et al., 2009). Furthermore, the volume of the coronary capillaries was measured, being functionally related to the volume of blood available for gas exchange within the myocardial tissue (Tang et al., 2009). These measurements provide knowledge of the development and efficiency of the heart and enable direct comparison to the human and other mammalian hearts.
Hearts were collected from 17 healthy 2- to 4-years-old male giraffes (Giraffa camelopardalis) euthanized for research purpose by the Danish cardiovascular Giraffe Research program, Gauteng province, South Africa. The experiments were approved by the Danish Animal Ethics Committee, the Animal Ethics Screening Committee (AESC) at The University of Witwatersrand, Johannesburg, South Africa and by the Animal Use and Care Committee, University of Pretoria, South Africa. In addition, hearts from four adult giraffes that were euthanized in European zoos due to chronic lameness were collected. Also from zoos, we obtained hearts from two still-born calves, two 4- to 6-months calves and one fetus. Mean body mass of the adult giraffes was 458 ± 91 kg and mean height was 344 ± 26 cm (Table 1). These data were not collected for the calves. After euthanasia, the hearts were immersion fixed in 4% phosphate-buffered formaldehyde for 24–48 hr and then stored in PBS + azid.
Table 1. Age, gender, and body composition of giraffes
Preparation of Tissue
The LV including the interventricular septum was isolated, the trabecular muscles and the chordae tendineae were removed, and the wet weight of the LV was measured (Fig. 2). The LVs were hereafter cut in 8–10 transverse sections perpendicular to the long-axis with a slice thickness of 1.0 or 1.5 cm depending on heart size. The procedure for tissue collection followed the principle of systematic, uniformly and random sampling (Nyengaard and Gundersen, 1992). Area sampling of the LV tissue was performed by placing a hole grid over each slice of the ventricle and all tissue visible in each hole was sampled (Fig. 2). A 3-mm biopsy punch was used to collect 10–12 samples from all specimens. Samples were cut in 3 mm thick blocks, perpendicular to the last cut. Of these blocks, four were systematically collected with a random start (Nyengaard and Gundersen, 1992). This sampling procedure gave four sets of tissue blocks: one was used for stereological analysis and three were stored. To ensure an isotropic cutting orientation of the tissue blocks, these were embedded in an agar sphere produced in a rubber mould with a spherical cavity—the isector method (Fig. 2) (Nyengaard and Gundersen, 1992). The agar spheres were randomly embedded in glycolmethacrylate (Technovit 7100, Ax-lab, Copenhagen, Denmark), except a subset for immunohistochemistry which were embedded in paraffin. From each block, one 30 µm thick and two 3 µm thick isotropic uniform random (IUR) sections were cut on a rotary microtome (Jung 2065 Supercut, Leica Instruments, Nussloch bei Heidelberg, Germany). The 30 µm sections were stained with Mayer's haematoxylin for nuclei counting. One set of 3 µm slices was stained with Mayer's hematoxylin and 0.15% Basic Fuchsine for myocyte volume counting and one set was stained with polyclonal antibody CD31 (Abcam®) for capillary volume counting. Further, 16 serial IUR sections of 2 µm-thick was cut from each tissue block and stained with PAS-M to visualize the intercalated discs for counting nuclei per myocyte.
Tissue shrinkage was estimated by determining the ratio between tissue volume before and after embedment. Tissue volume before embedment was determined by transforming the weight of each tissue block into volume using a density of 1.04 g/cm3 (Brüel and Nyengaard, 2005). The tissue blocks were then embedded in glycolmethacrylate, cut in 3 µm sections and the volume of the embedded tissue was estimated using the Cavalieri estimator (Gundersen and Jensen, 1987). Mean volume shrinkage was measured to be 6%; however, no corrections for shrinkage were made as it did not differ significantly from 0.
For stereological analysis the counting equipment consisted of an Olympus BX51 microscope with a motorized stage (Prior H148) and an electronic microcator (Heidenhein, Traunreut, Germany) for accurately measuring position in the z-axis. A digital camera (Olympus DP70) was mounted on top of the microscope and connected to a personal computer. The computer was equipped with the newCAST stereology software including the add-on module Autodisector (3.6.5 Visiopharm, Hørsholm, Denmark). The software was used for sampling, superimposing the stereological probes (counting frames, point counting grids, line grids) onto the live view from the microscope camera and for stereological analysis.
Heart Dimensions and LV Volume (Reference Volume)
The heart weight was measured before fixation of the tissue and the volume of the LV was estimated using the Cavalieri estimator (formula 1) (Gundersen and Jensen, 1987). The ventricle was sliced in 8–10 slices, with known thickness and a random starting point. A transparent counting grid was placed randomly on every slice and points hitting the ventricle were counted and volume was estimated:
Where V (LV) is the total volume of the LV, is the slice thickness, (a/p) is the area associated with each point on the counting grid and is the total number of point hitting the LV.
Volume Fraction and Total Volume of Cardiomyocytes and Non-cardiomyocyte Tissue
The volume fraction of myocytes and non-myocyte tissue (anything else than myocytes) was estimated by point counting on thin (3 µm) plastic sections at a magnification of 875×. On the computer, a point counting grid was superimposed onto the tissue image and every point hitting myocyte tissue was counted and every point hitting non-myocyte tissue was counted. To estimate the volume fraction we used Eq. (2):
P(myocyte) is the total number of points hitting myocytes in the LV. P(LV) is the total number of points hitting the LV including P(myocyte). Volume density of non-myocyte tissue was estimated by replacing P(myocyte) in Eq. (2) with P(non-myocyte)
Total volume of myocytes and non-myocyte tissue was estimated by multiplying Vv(myocyte, LV) by V(LV) which is the reference volume (LV volume).
Density and Total Number of Cardiomyocyte and Non-cardiomyocyte Nuclei
A 60× oil immersion objective with a numerical aperture of 1.4 and a final magnification of 2,158× was used for the optical disector. Systematic, uniform, and random sampling (SURS) of fields of view (FOV) was performed for each tissue section. An unbiased counting frame of 1,500 µm2 was used to estimate the numerical density of nuclei (Gundersen, 1977). The sampling volume was the area of the counting frame multiplied with the disector height or “thickness” (the distance between the first and the last optical plane) of the section measured. The cut thickness (z-axis) of the sections were 30 µm; however, as sections may have a wavy surface and may shrink to some extent, the thickness at every FOV was measured and a z-axis distribution was estimated (Dorph-Petersen et al., 2001). According to the z-axis distribution, a guard height was established by starting every measurement 5 µm from the top of the section, and the disector height was 15 µm. A nucleus was counted as a myocyte nucleus if it was located inside a myocyte and as a non-myocyte nucleus when located outside myocytes. The microscope was focused down the section and when a nucleus was completely inside the counting frame without touching the exclusion lines of the frame it was counted. The volume of ventricular tissue, in which the number of nucleoli was counted, was estimated using the four corners of the counting frame as a point grid. The nuclei were counted in 15–20 FOV for every tissue block. Equation (3) was used to estimate the numerical density of nuclei:
Where ti is the local height of the section at position i, qi– is the sampled number of myocyte nuclei at position i, i is the field of view, MA is the microtome advance, ∑Q– is thetotal number of myocyte nuclei counted in all disectors of one LV. ∑P(LV) is the total number of points falling on ventricular tissue (the reference space), p is the number of counting frame corner points used to count the reference space, h is the height of the dissector, and a is the area of the counting frame.
Number of Nuclei per Myocyte
Using the Section assembler of the software VIS (Visiopharm, Hørsholm, Denmark), areas of interest (longitudinal cut myocytes) on each of the 16 serial cut sections was marked. With these fixed points each area of interest could be related to the same area in the following sections. This method makes it possible to follow the same areas of interest throughout all 16 serial sections using the physical disector (Fig. 5), to ensure that all nuclei of the myocyte are counted. An unbiased counting frame was used and sections seven and eight of the 16 serial sections were the first to be used as a reference “look up” section in the physical disector. When myocyte nuclei were present in the reference section but not in the “look up” section, the myocyte was counted. All counted myocytes were followed in both directions in the serial sections until they disappeared. Between 50 and 100 myocytes were sampled per animal. The average number of nuclei per myocyte N (nuclei/myocyte) was obtained by the equation:
Where Q−(mono) is the number of myocytes with one nucleus, Q−(bi) is the number of myocytes with two nuclei, Q−(tri) is the number of myocytes with tree nuclei, Q−(quatro) is the number of myocytes with four nuclei and so forth up to Q−(eight) which is number of myocytes with eight nuclei.
Total Number and Numerical Density of Myocytes and Non-myocytes
By dividing the total number of myocyte nuclei with the average number of myocyte nuclei per myocyte we estimated the total number of myocytes in the LV. The numerical density of myocytes was estimated by dividing the total number of myocytes with the total volume of the LV.
Mean Cardiomyocyte Volume
The mean volume per cardiomyocyte was estimated by dividing the total volume of myocytes with the total number of myocytes in the LV.
Volume Density and Total Volume of Capillaries
The volume density of capillaries was estimated on 11 adult hearts by point counting these on 3 µm paraffin sections at a magnification of 875×. A point counting grid was superimposed onto the tissue image and points hitting capillaries as well as points hitting LV tissue were counted. To estimate the volume density we used Eq. (2), replacing P(myocytes) with P(capillary). Total volume of capillaries was estimated by multiplying VV(capillary/LV) by V(LV) which is the reference LV volume.
Data are presented as single values or as means ± SD in the figures. In the text data are presented as mean (CVtot), except the gross data of weight, height, and so forth, these are presented as mean ± SD. CVtot is the coefficient of variation defined as the SD divided by the mean. Where nothing else is noted N (adult) = 21 and N (young) = 5.
Heart Dimensions and Left Ventricular Volume (Reference Volume)
Mean adult heart mass was 2.5 ± 0.8 kg and mean relative heart mass was and 0.5% ± 0.06%. Mean heart weight of the young hearts was 0.28 ± 0.2 kg (Table 2). Mean volume of the LV (reference volume) measured using the Cavalieri estimator was 1,311 ± 400 cm3 for the adult and 22, 75.5, and 132 cm3 for the fetus, still-born, and 4–6 month, respectively. Data of the individual hearts from each giraffe are shown in Table 2.
Volume Fraction and Total Volume of Myocytes and Non-myocyte Tissue
The volume fraction (VV) of myocytes in LV was 0.89 (0.01) in the adult hearts and 0.90 (0.02) in the young LV collected in one group (Fig. 3A). Total volume (V) of myocytes in the adult LV was 1,153 cm3 (0.28), 68 cm3 (0.06) in the still-born, and 119 cm3 (0.06) in the 4–6 month LV. Volume fraction of non-myocyte myocardium was 0.11 (0.1) in the adult LV and 0.10 (0.19) in the young LV collected in one group (Fig. 3B). Total volume of non-myocyte myocardium in the adult LV was 135.9 cm3 (0.3), 8.42 cm3 (0.16) in the still-born, and 13 cm3 (0.01) in the 4–6 month LV (Table 3).
Table 3. Volume, densities, and numerical data of the giraffe LV
LV volume, cm3
V (myocyte/LV), cm3
V (non-myocyte/LV), cm3
Mean myocyte volume, um3
2.7 × 1010
Nv (myocyte nuclei/LV), mm−3
504 × 103
120 × 103
Nv (non-myocyte nuclei/LV), mm−3
72 × 103
74 × 103
N (myocyte nuclei)
4.9 × 1010
1.3 × 1011
N (non-myocyte nuclei)
5.9 × 109
7.7 × 1010
N nuclei per myocyte
Density and Total Number of Cardio Myocyte Nuclei and Non-myocyte Nuclei
In the adult LV the numerical density (NV) of myocyte nuclei was 120 × 103 mm−3 (0.2), whereas in the young LV (calves collected in one group) the numerical density was 504 × 103 mm−3 (0.17) (Fig. 4A). The total number (N) of myocyte nuclei was 1.3 × 1011 (0.21) in the adult LV and 4.9 × 1010 (0.26) in the young LV. There was a significant difference between adult and young nuclei number (P < 0.001). In the adult LV the numerical density of non-myocyte nuclei was 73.9 × 103 mm−3 (0.21) (Fig. 4B) and in the young LV it was 72 × 103 mm−3 (0.16). The total number of non-myocyte nuclei in the adult LV was 7.7 × 1010 and in the young LV the total number of non-myocyte nuclei was 5.9 × 109 (Table 3).
Number of Nuclei Per Myocyte
The mean number of nuclei per myocyte in the adult LV was 4.2 ± 0.07 (Fig. 5). The number of nuclei per myocyte ranged between 2 and 8 nuclei per myocyte (Fig. 5). Unfortunately, the PAS-M did not properly stain the young LV, rendering it impossible to count the cells.
Total Number and Numerical Density of Myocytes
The total number of myocytes in LV was estimated to 2.7 × 1010 (0.25) (Fig. 6) and the numerical density of myocytes in LV was 24.1 × 106 cm−3(0.25) (Fig. 7). The number of nuclei per myocyte is required to estimate the total number of myocytes and the numerical density of myocytes, thus unfortunately this was not possible to estimate for the young LV.
Mean Cardiomyocyte Volume
The mean cardiomyocyte volume was estimated using the total volume of myocytes in LV and the total number of myocytes in LV. On the basis of these values, the average volume of a myocyte of the adult LV was 39.5 × 103 µm3 (Fig. 8).
Capillary Volume Fraction and Total Volume
The mean capillary volume fraction of the adult giraffe LV was 0.054 (0.08) (Fig. 9) and the mean total volume of capillaries in LV was 65.6 cm3 (0.3).
This is the first study to quantitatively investigate the histological characteristics of the adult and young giraffe LV myocardium. First, we determined a relative heart mass of 0.5% of body mass, which is similar to most other mammals and consistent with recent studies on giraffes (Brøndum et al., 2009; Mitchell and Skinner, 2009). Second, volume fraction of myocyte volume and average myocyte volumes of the giraffe LV was comparable to those determined in rats and humans using stereological methods (Table 4; Olivetti et al., 1994; Brüel et al., 2007; Tang et al., 2009). Third, in the adult giraffe LV, we measured a higher number of nuclei per myocyte compared to other mammals, including humans (4.2 in our series vs. 1–2 in other species), as well as a higher overall nucleus density in the giraffe myocardium (120 × 103 mm−3 vs. 69.5 × 103 mm−3 in rats and 69 × 103 mm−3 in human). Finally and most importantly, the total number of myocyte nuclei in the adult LV was 62% higher than in the young LV. Altogether, these findings strongly suggest proliferation of cardiomyocytes during growth of the giraffe as an adaptation that allows the myocardium to overcome the increased systemic arterial afterload in the adult animal.
Table 4. Left ventricular volume- and numerical parameters
Human data from Brüel et al. (2002, 2005 and 2007) and rat data from Tang et al. (2009).
Nv(myocyte nuclei), 106 (cm−3)
V (capillary/LV), %
V per myocyte, 103 (cm−3)
Since the 1920s, it has been a dogma that the only manner of normalising LV wall stress, for example during systemic hypertension, is to increase cardiac mass by means of hypertrophy (Karsner et al., 1925; Petersen and Baserga, 1965; Morkin and Ashford, 1967; Morkin and Ashford, 1968), where the myocytes are thickening due to parallel replication of sarcomeres (Dorri et al., 2007). Since then, however, a number of recent investigations of both healthy and sick hearts have shown that hyperplasia can occur after the postnatal period (Linzbach, 1960; Olivetti et al., 1994; Beltrami et al., 1995; Anversa and Kajstura, 1998; Beltrami et al., 2001; Leri et al., 2002; Engel, 2005; Brüel et al., 2007b; Du et al., 2010). It is likely that the pathophysiological response to increased afterload elicits both a hypertrophic response, and a hyperplastic response with formation of- and parallel placement of new cardiomyocytes (Kajstura et al., 1994). Importantly, the combined response to chronically increased afterload is almost obligatorily accompanied by extracellular fibrosis, further impairing cardiac function (Lorell and Carabello, 2000). In our study, the finding that the relative heart mass of giraffes (0.5%) is similar to other mammals without heart disease, speaks strongly against the presence of hypertrophy in the classical sense. Moreover, the giraffe myocyte volume is comparable to that of other mammals, whereas the number of nuclei per myocyte and the absolute number of nuclei are significantly higher. Natural and pathological cell loss can mask proliferation of myocytes during aging, making it difficult to detect increases in myocyte number in morphometric studies. Several studies, however, indicate that an increase in nuclei number reflects increased myocyte number. For example, Du et al. (2010) showed that the ratio between mono- and bi-nucleated myocyte in rats changes only the first 7 days after cardiac overload and then returns to normal. These results exclude the possibility that the observed increase in nuclei reflects a change toward more binucleated myocytes. We find it reasonable to believe that the increase in nuclei reflects an increase in myocytes.
In the human heart, Olivetti et al. (1994) found the myocyte volume per nucleus to be 30–35·× 103 µm3, and as the human myocytes are mainly mono-nucleated (Baroldi et al., 1967; Korecky et al., 1979) this volume must be close to the volume per myocyte. According to these findings, the rat and human myocyte volume is comparable (Table 4). A difference of myocyte volume about 25% between giraffe and man is very modest comparing myocytes of severe hypertrophy due to aortic stenosis with normal ones there is a difference of more than 100%.
The mean capillary volume of 5 % within the LV volume in the adult giraffe heart is higher than the corresponding values of 3.6 and 2.9 % in humans and rats, respectively, obtained by similar methods (Table 3; Brüel et al., 2007; Brüel et al., 2002; Brüel et al., 2005; Tang et al., 2009). This could indicate that the myocardial oxygen delivery is increased in the giraffe LV either by a larger blood volume available or a smaller oxygen diffusion distance, in either case facilitating myocardial energy consumption in the face of a relatively low cardiac output (Bie and Damkjær, 2011). A study of the canine myocardial wall, however, showed a capillary volume of 4–6% (May-Newman et al., 1995) comparable with the capillary volume of the giraffe LV. A comparative study of the myocardial capillary volume in different mammals would be appealing, to elucidate if any correlations between workload, ventricular thickness and capillary volume exist.
We observed a high numerical density of left ventricular myocyte nuclei and a very high number of nuclei per myocyte compared to humans and rats. We hypothesize that the giraffe myocytes are optimized with many nuclei per cardiomyocyte to ensure a high synthesis of proteins and nucleic acid. Or, the myocyte regeneration and replication is optimized by having specialised nuclei playing various roles. This would be in line with Stephen et al. (2009), who showed that the nuclei of bi-nucleated cardiomyocytes may play different roles. They suggest that one nucleus is active in protein synthesis while the other “dormant” nucleus may be “activated” when cell replication and regeneration is necessary. This could be the mechanism maintaining a considerable proliferation of myocytes, which is indicated by the increasing number of myocyte nuclei as the giraffe ages. The high number of nuclei and possible myocyte proliferation may well be in line with the high pressure load exerted on the LV. This would also be consistent with the geometry of the ventricle with thick walls and small lumen, which according to Laplace would offset a high pressure load by lowering the stress per unit area of the ventricular walls. It is noteworthy that in the mammal with the highest known mean arterial blood pressure, the number of nuclei per cardiomyocyte differs considerably from all other known data from mammals including humans.
We are grateful to all members of the DaGiR team for making this study possible. We thank Paul R. Manger and Geoffrey Candy, for valuable contributions and Leith Meyer, Witwatersrand University, for ethical consulting. The staff at Wildlife Assignments International is acknowledged for expert animal handling during the study. Also Copenhagen Zoo, Givskud Zoo and Knuthenborg Safaripark are acknowledged for their contribution of animals. We are grateful to Maj-Britt Lundorf, Mette Skov Mikkelsen and Bente Wormstrup for tissue preparation.