OUP user menu

Programmed cell death in the developing heart

Maurice J.B van den Hoff, Stefan M van den Eijnde, Szabolcs Virágh, Antoon F.M Moorman
DOI: http://dx.doi.org/10.1016/S0008-6363(99)00401-0 603-620 First published online: 1 February 2000
  • Apoptosis
  • Development biology
  • Embryology
  • AIF, apoptosis-inducing factor
  • AP, aorticopulmonary
  • APAF, apoptotic protease activating factor
  • AVC, atrioventricular canal
  • CAD, caspase-activated Dnase
  • CARD, caspase recruitment domain
  • cIAP, cellular inhibitor of apoptosis
  • Daxx, death domain associated protein
  • DD, death domain
  • DED, death effecter domain
  • DM, dorsal mesocardium
  • DR, death receptor
  • DcR, decoy receptor
  • ED, embryonic day
  • ERK, extracellular signal-regulated kinases
  • FADD, Fas-associated death domain
  • FLIP, FADD-like interleukine-1 converting enzyme inhibitory proteins
  • H/H, Hamburger and Hamilton stage
  • IAS, interatrial septum
  • ICAD, inhibitor of caspase-activated DNase
  • IVS, interventricular septum
  • Jak/Stat, Janus kinase/signal transducer and activator of transcription
  • JNK, stress-activated c-Jun NH2-terminal kinase
  • LA, left atrium
  • LV, left ventricle
  • MAP, mammalian mitogen-activated protein
  • ND, neonatal day
  • NF-AT, nuclear factor of activated T-cells
  • OFT, outflow tract
  • PIF, primary interventricular foramen
  • PS, phosphatidylserine
  • RAIDD, RIP associated ICH1/CED3 homologous protein with death domain
  • RA, right atrium
  • RAR, retinoic acid receptors
  • RIP, receptor-interacting protein
  • RV, right ventricle
  • R×R, retinoid×receptors
  • SODD, silencer of death domain
  • TNF, tumor necrosis factor
  • TNF-R, tumor necrosis factor receptor
  • TRADD, TNFR-associated death domain
  • TRAF, TNFR-associated factor
  • TUNEL, terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling

1 Introduction

In humans every minute millions of cells die by programmed cell death and by the end of their life almost 99.9% of all cells once made have undergone this fate [1]. The functionality of cell death is different during development and adult life. During development apoptosis serves three major functions: deleting vestigial structures, i.e. phylogenetic cell death; controlling cell numbers, i.e. histogenetic cell death, and remodeling structures, i.e. morphogenetic cell death [2]. In adult life, apoptosis mainly serves to maintain homeostasis by counterbalancing mitosis and deleting cells, which are potentially autoimmunoreactive, malignant, or virus infected [3,4].

In contrast to adult life cell death, where cell death is generally considered a pathological event, during cardiac development cell death is a physiological event that is highly regulated in space and time. It took till 1968 when cell death was first recognized in the chicken ventricular myocardium [5]. In the next decade, the spatiotemporal pattern of cell death in the developing heart was extensively studied, in particular by the groups of Pexieder [6–10], Okamoto [11–14] and Ojeda [15–18]. These studies accurately described cell death patterns, but the underlying mechanisms and functions were not addressed. Despite the large body of data regarding the spatiotemporal patterns of cell death in the developing heart, a comprehensive overview of cell death in this developing organ is difficult to obtain. This is mainly because the original data were not published in regular journals and/or in the German language. Moreover, the nomenclature of the various structures of the developing heart has changed significantly during the last decades, which adds to the imperviousness of the data. The objective of this review is to bring together data derived from the earlier as well as from recent studies, and to generate a map of cardiac cell death. Such a map may provide a valuable tool in the analysis of the developing heart in genetically modified mice.

To appreciate the strengths and weaknesses of such a map of embryonic cardiac cell death, it is important to evaluate the cell death detection techniques used in the various studies. These methods include classical cell morphological and histological techniques, and molecular assays based on recent advances in our understanding of the molecular mechanism of cell death, including: (i) the detection of (internucleosomal-) DNA cleavage into 180–200 bp fragments [19–21]; (ii) the detection of cell surface exposure of the phospholipid phosphatidylserine (PS) [22–26], and (iii) immunohistochemical detection of activated cysteine proteases called caspases and/or their various targets [27–30].

Finally, the steadily increasing knowledge on the molecular control of the apoptosis machinery has allowed the construction of genetically modified animals in which a single component of the apoptotic machinery is functionally disrupted, sometimes resulting in a cardiac phenotype. The analysis of these animals is discussed as it might give clues to the molecular mechanisms that underlie the regulation of cell death and differentiation in the developing heart.

2 Detection of cell death

Soon after the invention of the compound light microscope developmental cell death was recognized on the basis of specific cell morphological changes, which were many years later dubbed apoptosis [2,3,31,32]. In the last decade, techniques have been developed to detect cell death. Each of these techniques has its own advantages and pitfalls regarding the sensitivity, selectivity, and convenience of cell death detection. These characteristics have to be considered when comparing quantitative and qualitative data on developmental patterns of cell death.

2.1 Morphological methods

2.1.1 Histological analysis

In the histological analysis, thin paraplast- or plastic-embedded tissue sections are stained using Hematoxylin-Eosin, Toluidine Blue Staining, Feulgen Reaction, PAS Reaction and/or Methyl Green-Pyronine Staining [8,10,11]. Cells in cultures or in suspension are most frequently stained using the fluorescent dye Hoechst 33342 [33]. Each of these staining procedures does not specifically stain dying and dead cells; they do allow the identification of apoptosis by one of its hallmarks, i.e. nuclear and cellular condensation (pyknosis) and the discrimination from necrosis or pathological cell death where cell swelling, also dubbed ballooning, is the most remarkable feature. In addition, the pattern and intensity of staining of dying cells changes during the progression of apoptosis, which may help to get insight in the dynamics of the process of apoptosis.

2.1.2 Histochemical analysis

In histochemical analyses, zones of cell death were identified in frozen sections that were stained using acid-phosphatase staining and two adjacent control sections, making this analysis technique laborious and prone to errors. One of the control sections has to be incubated without staining substrate and the other control section with the assay mixture, supplemented with sodium fluoride, an inhibitor of phosphatases, to exclude false positives. Acid-phosphatase is present in lysosomes of macrophages and dying or dead cells, irrespective whether this is due to pathological or programmed cell death. Although this method has been frequently used in the past, it is not a valid method to detect programmed cell death due to its lack of specificity. This lack of specificity was already noticed analyzing cell death in the developing heart because not all zones of cell death identified in the histological analysis could be recognized by the acid-phosphatase staining procedure [6,8]. Thus, the data on cell death reported in literature based on the analysis of acid-phosphatase staining procedure should be regarded with caution because not all zones of cell death can be identified and when staining is observed the numbers of cells in apoptosis are most probably over-estimated because besides death due to programmed cell death also death due to pathological cell death and macrophages are identified. The absence of staining might be due to the fact that apoptotic debris is removed by acid-phosphatase negative macrophages or other cell types. Electron microscope studies showed that phagocytes can be divided into two major groups (Fig. 1). The first group consists of big phagocytizing cells with large phagosomes, being most probably macrophages, and the second group of phagocytes are small neighboring cells, being most often mesenchymal cells [6]. Finally, also phagocytized apoptotic debris has been found in cardiomyocytes in electron microscopical studies [15,16].

Fig. 1

Transmission electron micrograph of the proximal portion of the OFT of an ED16 rat heart (panel A). Panel B is a schematic representation of the electromicrograph shown in panel A. The light gray cells indicate the mesenchymal cells and the dark gray the cardiomyocytes. The stars point to apoptotic debris within cells. Note at the right site of the picture the large phagocyte, most probably a macrophage, that is engulfing a dying mesenchymal cell that contains a very electron dense, pyknotic nucleus and vesicle containing cellular debris. Furthermore, a cardiomyocyte is shown that protrudes in the endocardial ridges mesenchyme, contributing to the formation of the muscular outlet septum.

2.1.3 Supravital staining procedures

In supravital staining procedures, embryos are perfused with Ringer's solution supplemented with Nile blue sulfate or Neutral Red [9,10,17]. These supravital stains color dying and dead cells, and phagocytes that have taken up cellular debris. After incubation with supravital stains the hearts have to be directly inspected under a dissecting microscope, because fixation and embedding results in loss of Nile blue and/or Neutral red staining [9,10]. Thus, this method identifies relative large areas with a relative high frequency of dying and dead cells and phagocytes, without detailed morphological or high-resolution information at the cellular level. So far, Nile blue staining has only been reported for the efficient demonstration of dying or dead cells in chicken embryonic hearts [10].

2.1.4 Ultrastructural analysis

Transmission electron microscopy allows high-resolution analysis of dying and dead cells although it is difficult to generate an overview of the spatiotemporal pattern using this technique. Electron microscopical analysis of various cell types showed that the distinct morphologies present during the viable phase of cellular existence are to some degree also reflected in the morphological characteristics of dying and dead cells. This insight may be particularly important for the detection of cell death in the heart, since during development of the heart cardiomyocytes, mesenchymal as well as endothelial cells become deleted. Ultrastructural analyses of various tissues of developing rat and mice embryos have revealed that there are three main types of cell death in developing tissues [34]. The first type of dying cells, degenerates without any detectable role of the cells own lysosomes, but their fragments are degraded in lysosomes of other cells, phagocytes. This type of cell death is regarded as the classical apoptosis described by Kerr and co-workers in 1972, and seems to apply to the cardiac endothelial and mesenchymal cell population. The second type of dying cells degrades their cell content to a large extent in their own lysosomes. This type of cell death is also dubbed autophagic degeneration and resembles the phenotype described for dying ventricular cardiomyocytes [6,10,34,35]. The third type of dying cell is destroyed without an apparent role of lysosomes and is, therefore, dubbed non-lysosomal vasiculate degradation. This type occurs in cartilage cells during bone development [36], but has not been described in the heart.

2.2 Biochemical and/or molecular methods

Morphological analysis not always unambiguously assesses physiological cell death. The different morphological faces of dying and dead cells necessitate an independent method to discriminate these cells, preferentially based on well-defined biochemical and/or molecular characteristics.

2.2.1 Detection of apoptosis-associated DNA fragmentation

The first method used to detect apoptosis-associated DNA fragmentation, was by ethidium bromide staining of total genomic DNA that was size fractionated on an agarose gel [21]. The obtained results show a characteristic ladder of DNA bands with increasing size as a result of internucleosomal DNA fragmentation. In non-apoptotic controls only high molecular weight DNA is present whereas a smear is observed when the DNA is broken down due to the isolation method or another process. Although this method, in principle, allows the analysis of apoptosis in the developing heart according to developmental stage, it does not provide any spatial information at the histological level.

However, in the early nineties [19,20] a molecular method for detecting apoptotic cells at the cellular level became available, by in situ labeling of fragmented DNA in apoptotic cells using terminal deoxynucleotidyl transferase (TUNEL assay). The TUNEL assay is a widely used method to detect apoptosis due to its apparent simplicity. However, the TUNEL-assay has limitations in sensitivity and specificity that are mainly caused by tissue processing. To increase sensitivity a pre-treatment with Triton X100, proteinase K, the microwave, alone or in combination are introduced. These pretreatments can easily lead to general labeling of all cell nuclei, therefore, the resulting effects should be evaluated carefully. Secondly, DNA breaks in non-apoptotic cells, like for instance due to DNA replication, DNA repair, tissue fixation and handling can result in false–positive staining in the TUNEL-assay and thus reduces the specificity of the assay [37].

2.2.2 Exposure of phosphatidylserine on the plasma membrane

Phospholipids are the major constituents of plasma membranes and several phospholipid species have been reported to be asymmetrically distributed across the inner and outer leaflet of the lipid bilayer of the plasma membrane [38–40]. The most profound asymmetric distribution is observed for phosphatidylserine (PS), which is confined to the membrane leaflet that faces the cytoplasm in most viable cell types [38,40], and is equally distributed across both leaflets of the plasma membrane during apoptosis [23–25,41].

Based on this difference of distribution of PS between viable cells and apoptotic cells, an assay was developed using FITC- or biotin-tagged human recombinant annexin-V. Annexin-V is a protein which binds PS in the presence of physiological or higher levels of Ca++ [42–44]. This marker has been shown to bind apoptotic cells relatively early, before the onset of TUNEL-detectable DNA fragmentation or profound cell morphological characteristics such as pyknosis both in vitro and in vivo [24,45,46].

In developmental studies on apoptosis, tagged Annexin-V was injected into the blood stream, like in the supravital staining procedures (see Section 2.1.3). However, the annexin-V assay is superior to supravital stainings because (i) annexin-V only identifies apoptotic cells, (ii) human recombinant annexin-V is able to interact with PS in all species tested so far, and (iii) the annexin-V binding can not only be evaluated in whole mounts, but also in sections at both the light microscopical and electron microscopical level [45]. A disadvantage of the annexin-V assay is that this assay has to be applied to fresh, preferentially living organisms.

2.2.3 Detection of apoptosis-associated antigens

With increasing knowledge of the molecular pathways regulating cell death, it is to be expected that new biochemical markers will become available that can be used to identify apoptotic cells. Recent examples of such markers include activity assays for various caspases. In these activity assays, homogenates or single cells are incubated with a modified-tetrapeptide as substrate that upon cleavage yields fluorescent molecules that is, within limits, proportionate with the activity in the sample. Although caspases can be subdivided into three groups based on their substrate specificity, (I) caspase-1, -4 and -5 prefer the sequence WEHD, (II) caspase-2, -3 and -7 DExD and (III) caspases-6, -8, -9 and -10 (I/L/V)ExD, at least in vitro multiple redundancies are observed as a result of the concentration and accessibility of the reaction components and the reaction time [47,48]. As these assays are performed on homogenates or single cells, they do not provide information at the histological level. Based on such an assay it has been shown that activated caspases, that cleave the tetrapeptide DEVD are present in the chicken outflow tract (OFT) during shortening of the OFT myocardium [49]. Furthermore, using different tetrapeptides is was observed that in staurosporine-induced apoptosis of cultured rat neonatal cardiomyocytes only active caspases belong to group II are present and to group-I and -III are absent [50]. To overcome, or at least reduce, the problem of substrate redundancy, the caspase activity assays have recently been modified such that the caspase of interest is captured in a microtiter plate that is coated with a monoclonal antibody directed against the respective caspase, prior to the activity assay (Roche). New antisera become rapidly commercially available for the immunohistochemical detection of apoptosis or associated processes (e.g. caspase cleaved actin [29] and cytokeratin 18 [28], and activated caspase 3 [27,30]). The value of each in the analysis of apoptosis in the developing heart has to be assessed as yet.

2.3 Concluding remarks

Neither of the above-described analysis methods (Table 1) is suitable to be used as the sole assay to identify dying and dead cells in the developing heart. Therefore, we consider, at present, two independent assay methods necessary to identify cell death in the developing heart. From Table 1 one can appreciate that the type of analysis and the material available dictates which assays can be used to identify dying and dead cells. If one has the opportunity to prepare new specimens, we consider, at present, the combination of an Annexin V- and TUNEL assay, to detect early- and late-apoptosis, the best combination to identify dying and dead cells in the developing heart.

View this table:
Table 1

Overview of frequently used apoptosis detection methods with some of their characteristics. A minus (−) indicates that the assay is not suitable, a plus (+) that the assay is suitable, and a plus/minus (±) that the assay could be suitable depending on the fixation and preservation of the specimens, to identify dying and dead cells

Routine histologyVital dyesTUNEL assayAnnexinV assayImmuno-histochemistry
Whole mounts+++
Light microscopy++++
Electron microscopy++++
Archived material+±±
Primary detectspyknosislysosomesDNA fragmentationPS exposureantigen

3 Spatiotemporal distribution of cell death in the developing heart

An inventory of the pattern of cell death in the developing heart was made for chicken [7–10,14,17,18], rat [11,14] and human [11–13]. These inventories were based on the combination of the results of four different morphological analysis methods, i.e. histology, histochemistry, supravital staining and electron microscopy. Each of the separate methods does not unambiguously identify dying or dead cells during development but the combination does so usefully. Recently, this inventory of zones of cell death based on morphological criteria was complemented with cell death determined by the TUNEL-assay [46,49,51–54]. Most of these studies identified in almost all parts of the heart and in each developmental stage analyzed a low percentage (1–3 tissue–volume percentages) of dying cells. Pexieder designated this overall low level of cell death as “background cell death” and suggested that it is associated with histogenetic apoptosis. Interestingly, also areas with temporally and locally high quantities (15–20% tissue–volume percentage) of dying and dead cells were identified. These areas are dubbed “zones of cell death” or “cell death foci”.

As the spatiotemporal inventory made for cell death in the developing chicken heart is most extensive, it will be described in detail in the next section followed by a section indicating the difference relative to humans and rats.

3.1 Spatiotemporal pattern of apoptosis in the developing chicken heart based on morphological studies (Fig. 2)

The spatiotemporal map of cell death in the developing chicken heart consists of a total of 32 zones combining the morphological analyses reported by the groups of Pexieder and Ojeda [7–10,14,17,18].

Fig. 2

Cartoon of the looping embryonic chicken heart at about four days of incubation. This cartoon shows the ballooning right and left atrium (RA, LA), and left and right ventricle (LV, RV) from the initial straight primary heart tube, comprising the dorsal mesocardium (MD), the atrioventricular canal (AVC), the primary interventricular foramen (PIF), and the outflow tract (OFT). In the primary heart tube the endocardial swellings, being the atrioventricular canal cushions and outflow tract ridges, are indicated. In between the right and left atria the forming interatrial septum (IAS I) is shown and in between the left and right ventricle the forming interventricular septum (IVS). The arrows indicate the already separated left and right blood flows. The numbers point to the following zones of cell death that are correlated with: 1. rupturing of the dorsal mesocardium, 2. entering of the pulmonary veins into the atria, 3. formation of pores in the atrial septum to maintain the fetal communication between the left and right atrium, 4. septation of the atrioventricular canal, 5. septation of the outflow tract, 6 and 7 shortening of the OFT myocardium to the future level of the valves.

At one day of incubation, the first zone of cell death is observed in the midline of the fusing heart fields during formation of the primary straight heart tube.

At two days of incubation, a zone of cell death is found at both the venous and arterial end of the heart. These two zones are considered to be located at the two ends of the rupturing dorsal mesocardium.

At four days of incubation the zone of cell death identified at the venous pole of the heart at two days of incubation still persists (Fig. 2, position 1). At the venous pole of the heart an additional zone of cell death is identified that is associated with the position where the pulmonary veins enter the atria (medinastinal mesenchyme, regio mesocardialis dorsalis, pneumocardial stalk) (Fig. 2, position 2). The other zones of cell death are related to septation and remodeling. Going from the venous to the arterial pole of the heart the following zones of cell death are observed. With septation of the atria a zone of cell death is found in the developing atrial septum (septum primum) that can be related to the formation of pores to maintain the fetal communication between the left and right atrium (Fig. 2, position 3). A low but significant level of dying and dead cells is observed in the mesenchyme of the not yet fused cushions of the atrioventricular canal (AVC; Fig. 2, position 3), whereas extensive cell death is observed in the mesenchyme of the not yet fused ridges of the OFT (Fig. 2, position 5). Apoptosis of the mesenchymal cells in the OFT ridges and AVC cushions precedes the invasion of cardiomyocytes from the surrounding myocardium, dubbed myocardialization. Due to the process of myocardialization the muscular outlet septum, the muscular atrioventricular septum and the lower part of the muscular interatrial septum are formed in the adult chicken heart [46,52,53,55,56]. The observation that the frequency of cell death in the mesenchyme of the cushions in the AVC is lower than in the ridges of the OFT, suggests that cell death is strictly regulated during development. In addition, two other zones of cell death are reported in the arterial pole of the heart. One zone is observed in the myocardium at the right side of the OFT (Fig. 2, position 6) and the other in the myocardium at the transition of the ventricle to the OFT (bulboventricular groove; conoventricular sulcus; Fig. 2, position 7). The OFT is defined as the myocardial part of the heart at the arterial pole that is not trabeculated and is covered with ridges, and that has a distinct functional and molecular phenotype [46,57–59]. These two zones of cell death in the myocardium of the OFT are suggested to be involved in the regression of the myocardium from the level of the pharyngeal arches to the future adult level of the valves. Finally, on the surface of the heart in the forming epicardium apoptosis is found associated with blood islets.

At six days of incubation the number of zones of cell death is largest. Although the zones observed in the previous stage are not as prominent anymore, they can still be identified in some embryos. However, also new zones were identified. A zone of cell death is found at the side of fusion of the atrial septum with the interventricular septum and the septum of the OFT. Interestingly, a zone of cell death is reported in the myocardium enclosing the primary interventricular foramen (Fig. 2, PIF). Although, this zone is not further specified, it needs further evaluation, because the myocardium of this area has a unique molecular phenotype. In the chicken this area is characterized by the expression of Msx2 [60] and in the human by GlN2, Leu7, and/or HNK1 [61,62]. With development, the primary interventricular foramen takes origin to the inlet of the right ventricle and the left ventricular outlet. The surrounding myocardium form the lower rim of the right atrium, the subaortic portion of the outlet septum, and the atrioventricular bundle of conduction tissue at the top of the interventricular septum [55,57,62,63]. Compared to four days of incubation the frequency of death and dying cells has substantially decreased in the mesenchyme of the OFT ridges, whereas extensive cell death is observed in the mesenchyme of the now fused cushions of the AVC. In the arterial pole of the heart three new zones of cell death could be identified, one is located at the caudal tip of the down-growing aorticopulmonary septum that is associated with the fusion of this septum and the fused OFT ridges. The other zones of cell death in the arterial pole are located outside the heart in the walls of the developing aorta and pulmonary arteries. Finally, a zone of cell death is found in the caudal part of the interventricular septum that is suggested to be correlated with the insulation of the developing conduction system from the surrounding working myocardium. However, the role and function of these latter two zones of cell death has not been established yet.

By eight days of incubation, when remodeling and septation of the heart has almost been completed, the frequency of dying and dead cells has decreased but the prominent zones of cell death identified at six day of incubation can still be recognized.

At ten days of incubation, only two zones of cell death have remained one in the mesenchyme of the right atrioventricular junction and the other in the developing valves of the aorta and pulmonary artery.

Finally, at sixteen days of incubation, no zones of cell death could be identified anymore, because the frequency and distribution of dying cells has decreased to the level of background cell death.

In the zones of cell death the dying and dead cells were reported to be almost exclusively mesenchymal cells, with only few cardiomyocytes. These data are difficult to reconcile with the supposed absence of significant numbers of mesenchymal cells in the primary atrial septum, the interventricular septum, the area of transition from the ventricle into the OFT, and the myocardium of the right side of the OFT. Hence, one would anticipate that substantial numbers of cardiomyocytes have to undergo cell death in these zones. Dying or dead endocardial cells have rarely been described in literature [10]. However, cell death of the endocardial lining of the ridges would be expected to precede and facilitate fusion of the ridges, as in an analogous process of apoptosis occurs in the ectoderm covering the face at specific spatiotemporal coordinates during fusion of facial swellings and closure of the secondary palatum [64,65]. The very low frequency of reported dying and dead endocardial cells might partly be due to swallowing of early dying endocardial cells by the blood stream, that also seems to apply to yolk sac derived circulating erythroblasts at a specific stage of mouse development [26]. However, annexin V has permitted the identification of dying endocardial cells at site of fusion of the endocardial ridges of the OFT (Fig. 3).

Fig. 3

Annexin V-biotin labeling of phosphatidylserine exposing endocardial cells in the fusing OFT ridges (panel A, B) and epithelial cells in the fusing facial swellings (panel C, D) of an ED11.5 mouse embryo. Panels B and D show an enlargement of the boxed area indicated in panels A and C. The arrows point to an endocardial cell (panel B) or an ectodermal cell (panel D) exposing phosphatidylserine as identified by Annexin V-biotin labeling. Abbreviations: ec: epithelial cells, end: endocardial cells, cm: cardiomyocytes, mes: mesenchymal cells, er: endocardial ridges, lu: lumen of the OFT.

3.2 Spatiotemporal pattern of apoptosis in the developing chicken heart based on molecular studies

Recently, the TUNEL assay has been used to study the spatiotemporal distribution of cell death in the developing chicken heart [49,51–53]. Overall, the reported results are consistent with the results of the morphological analyses of cell death, although generally the frequency of dying and dead cell is lower and the onset of apoptosis seems to be slightly later using the TUNEL-assay. These differences are not surprising. (i) In the TUNEL-assay cells are only identified very late during apoptosis when internucleosome fragmentation of the genomic DNA has occurred, whereas in the morphological methods, in principal all stages of dying and dead cells are identified as well as the phagocytizing cells. (ii) Pexieder's group did not stage the embryos according to Hamburger and Hamilton [66], but according to the period of incubation. Using the period of incubation is, especially, for young embryos unreliable.

However, in a study of Cheng et al. [51] also some differences can be observed when the spatiotemporal pattern of cell death determined with the TUNEL-assay as compared to the pattern described above. The zone of cell death in the myocardium at the circumference of the primary interventricular foramen was not shown in this study, suggesting that differential growth might be more important than cell death in remodeling of this part of the heart (see also Section 3.1 “six days of incubation”). Also, two new zones of cell death were reported. The first new zone of cell death was identified at H/H32–33 (about 7.5 days of incubation) in the trabeculations of the ventricles. Although the appearance of this zone is quite diffuse due to the organization of the trabeculations, the number of apoptotic cells is enormous due the size of the trabeculated compartment. The second novel zone of cell death was found between H/H40 and 44 (about 14–18 days of incubation) in the wall of the coronary vessels and the aorta. The identification of this latter zone of cell death suggests morphological/phenotypical changes in the developing vessels, which needs further analysis.

3.3 Interspecies differences in the zones of cell death

Data on interspecies differences are rare. Some differences in the number of zones of cell death between chicken, rat and human have been reported. In rat embryos between day 13 and 17 of gestation only 21 zones of cell death and in human embryos with a crown-crump length between 4 and 14 mm only 16 zones of cell death were identified. Although, it has been reported that the frequency of dying and dead cells and the developmental timing of the occurrence of the zones of cell death is different between these species, the individual zones of cell death have not been described [10,13,14,67], which makes it difficult to appreciate these differences.

A number of reasons may account for the observed lower number of zones of cell death in the rat and human heart compared to the chicken heart. Firstly, concerning the analysis of cell death in human embryos the difference is trivial because the number of available human embryos did not allow an extensive inventory of the zones of cell death [10]. Secondly, the developmental periods studied in the different species were different. Thirdly, the number of zones of cell death is lower in mammalian than in chickens due to developmental differences. For example, in the mammalian heart the atria are septated by the formation of two septa, the primary and secondary atrial septum, whereas in the chicken only the primary atrial septum is formed. As a result the communication between the right and left atria is possible via the foramen secundum in the mammalian heart and through (secondary) pores in the primary atrial septum in the chicken. Nevertheless, the foramen secundum in mammals is also formed in the primary atrial septum and could be considered as being the result of fusion of small pores to one large hole. As yet no apoptosis has been described in the mammalian primary interatrial septum, one has thus to propose that another mechanism underlies the formation of the foramen in mammals. A second example concerns the different morphology of the mammalian and chicken OFT. The chicken OFT has distinct proximal and distal outflow tract ridges, whereas distinct proximal and distal ridges are not obvious in the mammalian heart. This difference in morphology leads to the identification of four zones of cell death in the chicken OFT and only two zones in the rat and human OFT. Other striking examples are found in the arterial pole of the developing heart. In the chicken heart a zone of cell death is identified in the tip of the aorticopulmonary (AP) septum and one in the myocardium at the right side of OFT, whereas both are not found in the human or rat heart [10,14,67,68]. The absence of the zone of cell death in the tip of the AP septum in rat and human is hard to imagine as the down growing AP-septum has to fuse with the septum formed by fusion of the OFT ridges. At all other spots in the heart where structures fuse they are accompanied by cell death. Finally, the interspecies difference in the zone of cell death in the OFT myocardium was recently also confirmed by different molecular analysis methods (rat: [46,54]; chicken: [49,51,52]). These observations suggest that two different processes underlie the regression of the OFT myocardium. In chicken this regression of the OFT myocardium would be due to disappearance of the cardiomyocytes by apoptosis at H/H25–28 [49], whereas in rats due to trans-differentiation from cardiomyocytes into mesenchymal cells at ED12 [46]. Early electron microscopical analysis by Arguello and coworker [69] suggested that myocardial cells at the most distal myocardial border of the outflow tract trans-differentiate into connective tissue cells at H/H27 in chicken. Lineage marking studies of the myocardial cells at the distal border of the OFT are needed to resolve whether the fate of these cells is changing or whether they die.

4 Teratogens, zones of cell death and cardiac malformations

With the identification of zones of cell death at positions that are supposed to be crucial in remodeling and septation of the developing heart, it was hypothesized that experimental manipulation of cell death would lead to cardiac malformations. In order to test this hypothesis pregnant rats or chicken embryos were treated with teratogens just prior to or during development of the cardiovascular system. As teratogenic agents, cyclophosphamid, dexamethason, β-aminopropionitrile, trypan blue, Janus green, and X-ray irradiation were used. These treatments resulted in a reproducible scala of cardiac septational defects depending on the onset, duration and dose of the teratogen. Analyses of treated embryos showed in several cases that the spatio-temporal pattern of zones of cell death had changed only slightly. Although the induced changes in the zones of cell death are highly suggestive for a role of cell death in cardiac morphogenesis, a causal relation between apoptosis and altered cardiac morphogenesis has never been shown [10,12].

5 The molecular mechanism of apoptosis in the developing heart

After the extensive description of zones of cell death in the developing heart, the question remains how apoptosis is initiated and regulated in the different zones of cell death and what the underlying molecular mechanisms are. Although these questions were already raised in the seventies, they are not yet answered. In the last five years, the signalling pathways that tell a cell when it is time to die start to become apparent (Fig. 4). It is now generally accepted that a variety of stimuli can activate the program of cell death via many transduction mechanisms that eventually all convert on the final executing pathway [70]. Which and how these signalling pathways are involved in cardiac morphogenesis, is not at all clear. In the last years several extensive reviews have been published discussing the different molecules and molecular pathways that are involved in cell death [1,22,71–83]. Below, we have summarized the different apoptotic signalling pathways and their potential roles in cardiac development.

Fig. 4

Schematic representation of the potential apoptosis signalling transduction pathways involved in the regulation of cell death in the developing heart. The box indicated the following domains in the depicted proteins: box with horizontal lines is a death domain, the box with diagonal line is a death effecter domain, the open box is a caspase recruitment domain, the black balls the proteolytic domain in the initiator caspases, and the light gray box highlights the effector caspases. Arrows indicate an interaction that is considered pro-apoptotic, whereas a blocked line indicates an interaction that is considered anti-apoptotic.

5.1 Alteration of the membrane composition (Fig. 4)

One of the first molecular changes detectable in apoptotic cells, is cell surface exposure of phosphatidylserine (PS). In most healthy cells PS is predominantly found in the inner leaflet of the plasma membrane, whereas its distribution becomes equal over the inner and outer leaflet of the plasma membrane during cell death. Exposure of PS has been shown to result from a decrease in aminophospholipid translocase activity, which flips PS from the outer leaflet to the inner leaflet of the plasma membrane [84]. Secondly, a calcium-dependent scramblase, which randomizes the distribution of PS over both the inner- and outer leaflet of the plasma membrane, is activated [85]. The consequence of PS exposure is the recognition and removal of the apoptotic cell by phagocytes before the integrity of the plasma membrane becomes disrupted, and the cell leaks its potentially toxic and inflammatory contents to the surroundings [86,87]. Exposure of PS seems to be common to the apoptotic cells from all species tested so far, including the apoptotic cells in the developing heart [24–26,46]. Although the signal transduction pathway from the instruction to die to exposure of PS on the cell is not yet known, a role for caspase-3, and/or AIF is suggested [22].

5.2 Death receptor mediated signalling (Fig. 4)

Most probably the best-studied molecular pathway executing death is mediated by death signals and death receptors [72,73,79,80,83]. Death receptors belong to the tumor necrosis factor receptor gene superfamily and share a cysteine-rich extracellular domain and a cytoplasmic “death domain” (DD). The best-characterized death receptors are CD95 (also called Fas or Apo1) and TNF-R (also called p55 or CD120a). This superfamily includes four additional death receptors (DR), being DR3, DR4, DR5, and DR6 and three decoy receptors, being DcR1, DcR2, and DcR3. Interestingly, the decoy receptors are able to interact with their ligand, but are not able to execute the cell death program. Upon binding of the ligand (CD95L, TNF, Apo3L or Apo2L), the death receptors activate caspases. Caspases have been extensively characterized in mammals and were found to belong to a family with at least fourteen members [74,77,78,81,88–92]. Caspases exist as dormant proenzymes in healthy cells and are activated through proteolysis. Caspases can be divided into three classes based on their biological function. The first class of caspases (caspase-1, -4, -5, -11 and -12) do not play a primary role in cell death, but in cytokine maturation and inflammation. The other classes are primarily involved in the regulation of cell death. The second class of caspases are initiator caspases (caspase-2, -8, -9 and -10) that are activated by self-cleavage. The downstream effector caspases (caspase-3, -6 and -7) are the third class of caspases that are activated through proteolytic cleavage by the initiator caspases. These effector caspases, in turn, cleave a variety of cellular substrates after aspartate residues, leading to an irreversible commitment to cell death and the morphological hallmarks of apoptosis, including condensation of cellular organelles and DNA fragmentation. All four initiator caspases are expressed in the heart [92–96] as well as the three effector caspases [92]. Recently two new caspases have been identified. The role of these caspases is not yet clear. Caspase-13 [90] and caspase-14 [88,89] play a potential role in receptor mediate apoptosis, are expressed in the heart and seem to be downstream of the initiator caspase-8.

The activation of the initiator caspases is not due to a direct interaction of the procaspase with the death receptor, but mediated via adapter molecules. These adapter molecules interact via their “death domain” (DD) with the DD in the death receptor and via their “death effecter domain” (DED) or “caspase recruitment domain” (CARD) with the respective DED or CARD in the initiator caspase. Due to the recruitment and resulting clustering of initiator caspases to the death receptors, the initiator caspases activate themselves through self-cleavage [5,72,77,81,83].

The adapter molecules seem to play a crucial role as mediators in the propagation of the apoptosis signal from the environment to the interior of the cell. Not only do the different initiator caspases interact with different adapter molecules, but also the interactions between the adapter molecule and death receptor at one side and the adapter molecule and initiator caspase at the other side are modulated by anti-apoptotic molecules. The next three examples illustrate this carefully regulated balance between cell survival and cell death. (i) SODD (silencer of death domain), which is expressed in the heart, contains a DD that can interact with the death receptors but SODD is not able to recruit initiator caspases. Thus, by competing with the other adapters for the DD in the death receptors, SODD can inhibit intracellular signalling of the death receptors [97]. (ii) FADD-like interleukine-1 converting enzyme inhibitory proteins (FLIPs), which are expressed in the heart, have been observed to compete with procaspase-8 for the DED in FADD adapter molecule, inhibiting the transduction of the death signal. Generalizing, FLIPs inhibit apoptosis by competing for the DED or CARD in the adapter molecules such that initiator caspases are not activated, allowing a further fine-tuning the apoptotic machinery [98,99]. (iii) Thirdly, adapter molecules are not only observed to interact with initiator caspases but also with other factors that are able to activate other signal transduction pathways. An extensively characterized adapter molecule is TNFR-associated death domain (TRADD) that can bind a variety of molecules and as such acts either pro- or anti apoptotic. (a) TRADD was found to be pro-apoptotic via the adapter FADD with the associated initiator caspase-8 and/or via the adapter RIP-2 (receptor-interacting protein-2) with the associated initiator caspase-1, -2 and/or -9 [100]. (b) The anti-apoptotic activity of TRADD is mediated via TRAF (TNFR-associated factor) -1 or -2 that allows cross talk with the JNK (stress-activated c-Jun NH2-terminal kinase) signalling pathway, via RIP-1 or -3 that allows cross talk with NF-κB, and/or via an interaction with cIAP (cellular inhibitor of apoptosis) that, in turn, inhibits the biological function of the initiator caspase-8 [99,101–107].

As can be appreciated from Table 2, only mice homozygously deficient in either caspase 8 or FADD (Fas-associated death domain) die in mid-gestation from congested accumulation of erythrocytes and impaired heart development [108,109]. The hearts of these mice were not appreciably larger than normal but the ventricular musculature was thin and in some cases mesenchyme-like. The trabeculae were thin and disorganized [108]. Surprising both the Caspase-8 and FADD knockout mice have a phenotype of hypoplastic ventricle, whereas one would have expected a hyperplastic ventricle when cell death in the ventricles was abolished. This may indicate that this molecular pathway is linked with apoptosis and cell proliferation, depending on the cellular context [110]. Furthermore, it will be interesting to see what the phenotype of functional disruption of the recently cloned caspase-10 will be, because caspase-10 (i) has, like caspase-8, a DED that can interact via FADD with CD95 and (ii) is expressed at relative high levels in the heart [95,96].

View this table:
Table 2

Overview of the different molecules involved in execution of death signals. “A/P” indicates whether the biological action of the molecule is anti- or pro-apoptotic. “H?” indicates whether the molecule is expressed in the heart. “KO?” Indicates whether a knock out mouse has been prepared and the accompanying cardiac phenotype (column: phenotype). The final column identifies the corresponding references. It should be noted that many of the knock out mice had other phenotypic abnormalities, that are not included in this table. Secondly, when “no cardiac phenotype” is printed in the column phenotype this indicates that in the description of the knock out phenotype no cardiac abnormalities were reported, whereas subtle cardiac abnormalities could have been missed. A plus indicates yes and a minus no

Adaptor molecules
FADDP++hypoplastic ventricle, thin disorganized trabeculae, ED12.5 lethal[108,109,131]
RIP1A++no cardiac phenotype, postnatal lethal[132,133]
Initiator caspases
Caspase-2P++no cardiac phenotype, viable[135]
Caspase-8P++hypoplastic ventricle, thin disorganized trabeculae, ED12.5 lethal[108]
Caspase-9P++no cardiac phenotype, postnatal lethal[93,113]
Effector caspases
Caspase-3P++no cardiac phenotype, ED12 or ND21 lethal[94]
Novel caspases
Associated molecules
APAF-1P++no cardiac phenotype, ED16.5 lethal[112]
Bcl-2A++no cardiac phenotype, postnatal lethal[136,137]
Bcl-wA++no cardiac phenotype, viable[138,139]
Bcl-xA++no cardiac phenotype, ED13 lethal[136,140]
BaxP++no cardiac phenotype, viable[141–143]
CAD/ICADP++no cardiac phenotype, viable[109]
TRAFA++no cardiac phenotype, perinatal lethal[107]
Growth factor pathways
Endothelin-1++ventricular septal defect, malformations of[144]
great vessels
gp130++conventional KO: hypoplastic ventricle,[128]
prenatal lethal
conditional KO: no cardiac phenotype, viable[129]
Neurotrophin-3++ventricular and atrial defects, Perinatal lethal[145]
TGFβ2++ventricular septal defect, perinatal lethal[127]
PDGF-Rα++OFT septational defects[126]
RARβ1/3+RXRα++ventricular septal defect, perinatal lethal[35]
  • a Indicates that AIF was immunohistochemically detected in many tissues, but that it was not indicated in which tissues.

5.3 Cell death mediated via the mitochondria (Fig. 4)

Induction of apoptosis by cytotoxic agents is mediated via a death receptor independent signalling pathway [71,75,76,83]. In this signalling pathway a complex is formed that consists of APAF1 (apoptotic protease activating factor 1), cytochrome-c, dATP and procaspase-9. Upon the induction of apoptosis both cytochrome-c and dATP are released from the inter-membrane space of the mitochondria through pores into the cytosol. The release from the mitochondria is regulated by Bax- and Bcl2-family members that are pro- and anti-apoptotic, respectively. As a result of the formation of the complex of APAF1, cytochrome-c, dATP and the initiator caspase-9, caspase-9 is activated allowing the activation of the downstream effector caspases, including caspase-3. These, in turn, cleave downstream effector caspases and a variety of cellular substrates, like in death receptor mediated signalling (see Section 5.2). Recently, another factor was found to be released from the mitochondria that accumulates in the nucleus during induction of apoptosis, being apoptosis-inducing factor (AIF). AIF was found to mediate chromatin condensation and gross chromosomal cleavage [111]. Together with caspase-activated DNase (CAD/ICAD) that is activated by the effector caspases [109] and some yet unknown factors, pyknosis and internucleosomal cleavage of the genomic DNA is obtained.

Homozygous knockouts (Table 2) of APAF1 [112] die at ED16.5 and display in their phenotype amongst others craniofacial malformations. These malformations are most probably due to an absence of fusion of the facial swelling. Although the mechanism of fusion of the facial swelling is in essence comparable to fusion of the endocardial cushions and ridges it is surprising that cardiac abnormalities are absent in these mutant mice. Furthermore, mice in which ICAD (DFF45) is homozygously disrupted are resistant to chromatin condensation and fragmentation, which is the hallmark of apoptosis. However, these mice do not show any reported phenotypical abnormalities, suggesting that DNA fragmentation and condensation is not required for normal development [109]. Mice in which caspase-9 [93,113] or caspase-3 [94] is homozyguously disrupted die due to overgrowth of the brain, but without any reported cardiac abnormalities. Taken together, these data strongly suggest that the cell death transduction mediated via the mitochondria is not of primary importance in the regulation of cell death during cardiac development.

The fact that the different participants of the cell death transduction pathway are expressed during cardiac development [92–94,109,113] suggests an role in the regulation of cell death. The following mechanism have been proposed. (i) Cross talk between the signal transduction pathway of cell death via the mitochondria and via the death receptors was observed. The activated initiator caspase-8 was found to cleave Bcl-2, hence, shifting the Bax-Bcl2 balance in the pro-apoptotic direction. This cross talk was proposed to serve as a way to amplify the death signal mediated via the death receptors [106]. (ii) In healthy cells Bcl2 was found to be associated at the plasma membrane with calcineurin, of which one proposed function is facilitating translocation of NF-AT (nuclear factor of activated T-cells) into the nucleus. Over-expression of Bax in these cells, shifting the Bax-Bcl2 balance into the pro-apoptotic direction, resulted in a destabilization of the calcineurin-Bcl2 interaction and translocation of NF-AT to the nucleus [77,114–116]. Interestingly, homozygous deletion of NF-ATc resulted in absence of fusion of the OFT ridges and of formation of valves [117,118]. In wild type mice NF-ATc is expressed in the endocardium overlaying the OFT ridges [117,118]. It will be of interest to evaluate whether this Bax-Bcl2-calcineurin-NF-ATc signalling pathway leads to cell death of the endocardial cells and the subsequent fusion of the OFT ridges or whether this pathway regulates differentiation of the endocardial ridges.

5.4 Growth factor withdrawal induced cell death

Mammalian mitogen-activated protein (MAP) kinases can be subdivided into three signalling pathways with as key-regulators, (i) the extracellular signal-regulated kinases (ERK), (ii) the Jun N-terminal kinases (JNK) and (iii) p38 MAP kinases, that are each activated by different upstream signalling molecules. The delicate balance between these three pathways determines whether a cell survives or undergoes apoptosis. Neural growth factor (NGF) withdrawal of neural PC12 cells results in apoptosis and has shown to correlate with a down-regulation of the ERK pathway and up-regulation of the JNK and p38 MAP kinase pathways. Using inhibitors, constitutive active and dominant negative regulators of these pathways, showed that p38 MAP kinase pathway plays a key role in neural cell apoptosis [119,120]. In cardiomyocytes cardiotrophin-I (CT-1) was found to promote proliferation and survival, via the MAP kinases and via the Jak/Stat (Janus kinase/signal transducer and activator of transcription) signalling pathways [121]. The involvement of the MAP kinase signalling pathways might suggest that the delicate control as observed in neurons is also applicable to cardiomyocytes. The Jak/Stat signalling has as downstream target Bcl-xL, that functions anti-apoptotic in the apoptosis signalling pathway mediated via the signalling pathway involving the mitochondria (see Section 5.3). How the dichotomy in the signalling from the growth factor receptors is achieved, leading to the stimulation of proliferation and differentiation versus cell death is not clear at present.

Nevertheless, growth factors were found to be key-regulators in the onset and regulation of the population of the endocardial cushions and ridges by mesenchymal cells from various origins [56,122]. Intriguingly, none of the functionally disrupted genes involved in apoptosis (Table 2) resulted in cardiac malformations that correlate to the zones of cell death. One might speculate that local differences or changes in the biological available amounts of growth factors might underlie the spatiotemporal regulation of cell death in the developing heart. Two recent observations underscore this hypothesis. (i) Retinoid signalling is transduced by homo- and heterodimers of retinoic acid receptors (RARα, β and γ) and retinoid X receptors (RXRα, β and γ). Disruption of one allele of the RARβ locus, decreasing the expression level of all four RARβ isoforms, in a RXRα negative background, resulted in an increase in apoptosis in the OFT ridge mesenchyme and as a consequence absence of fusion of the ridges [35]. (ii) Functional disruption of endothelin-1 results in mice in complete absence of septation of the OFT, and a number of other abnormalities. In these mice apoptosis of neural crest cells was observed in the pharyngeal arches. Hence, the neural crest cells do not arrive in the OFT, with as consequence that the ridges are hypoplastic and that the aorticopulmonary septum does not develop [123,124]. The potential underlying molecular mechanism is regulated by the balance between Msx1 and Msx2 that are co-expressed by the neural crest cells in the pharyngeal arch. Endothelin-1 is secreted by the pharyngeal arch epithelium and via the endothelin receptor-A and dHand the expression of Msx1 is stimulated, whereas Msx2 promotes cell death via BMP4 [124,125]. It will be important to evaluate whether this signalling pathway is also responsible for the spatiotemporal pattern of cell death of the mesenchymal cells in the endocardial ridges during normal development.

Functional disruption of other growth factor signalling pathways, such as platelet derived growth factor [126] or transforming growth factor β [127] have also been found to lead to OFT septational defects. Whether alterations in the spatiotemporal pattern of cell death underlie the cardiac abnormalities remains an important issue to be tested.

It should be noted that each functional disruption of a growth factor system that results in cardiac abnormalities should not be interpreted as being due to alterations in the spatiotemporal pattern of cell death. The proliferative and survival responses elicited by cardiotrophin-I involve an initial low-affinity interaction with the leukemia inhibitory factor receptor (LIFR), followed by recruitment of the gp130 cytokine receptor into a high-affinity heterotrimeric complex, that results in intracellular signalling [121]. The conventional knock out of the gp130 receptor showed multiple organ defects among which is hypoplastic ventricle without septational and trabecular defects [121,128], whereas the conditional knock out of gp130, in which gp130 is only removed in the ventricular myocardium, does develop normal. Upon induction of aortic overload, these mice show massive apoptosis of the ventricular myocytes, whereas the controls show compensatory hypertrophy [129]. Apparently, the hypoplastic ventricle observed in the conventional knock out is a secondary effect rather than a primary effect of the disruption of the target gene. It is important to know how this secondary effect resulting in hypoplastic ventricle is achieved, as various knock out mice show this phenotype.

6 Concluding remarks and future perspective

In the last two decades a vast quantity of work has been done in order to elucidate the molecular pathways underlying the already centuries ago identified process of cell death or apoptosis. Maybe slightly disappointing but to date only the transduction pathway via CD95, FADD and caspase-8 has been shown to be critical in ventricular development, whereas cardiac septation and remodeling was normal. The in Section 3 described zones of cell death in the developing heart are suggestive for an involvement in proper formation of the heart. The elucidation of the molecular mechanisms underlying the spatiotemporal control of apoptosis in the zones of cell death has just been started. It is to be expected that understanding the regulation of apoptosis in the developing heart and the consequences for cardiac morphogenesis, will lead to new insights in the formation of the adult four-chambered double-circuited heart from the single-circuited straight heart tube. Taken together the findings that functional disruption of the genes involved in apoptosis signal transduction do not lead to septational and remodeling abnormalities, whereas functional disruption of the genes of growth factor signalling pathways do, suggests that growth factors play a more dominant role in the regulation of cell death in the developing heart.


Dr Maurice J.B. van den Hoff is supported by the Netherlands Heart Foundation (Grant no.: 96.002).


  1. [1]
  2. [2]
  3. [3]
  4. [4]
  5. [5]
  6. [6]
  7. [7]
  8. [8]
  9. [9]
  10. [10]
  11. [11]
  12. [12]
  13. [13]
  14. [14]
  15. [15]
  16. [16]
  17. [17]
  18. [18]
  19. [19]
  20. [20]
  21. [21]
  22. [22]
  23. [23]
  24. [24]
  25. [25]
  26. [26]
  27. [27]
  28. [28]
  29. [29]
  30. [30]
  31. [31]
  32. [32]
  33. [33]
  34. [34]
  35. [35]
  36. [36]
  37. [37]
  38. [38]
  39. [39]
  40. [40]
  41. [41]
  42. [42]
  43. [43]
  44. [44]
  45. [45]
  46. [46]
  47. [47]
  48. [48]
  49. [49]
  50. [50]
  51. [51]
  52. [52]
  53. [53]
  54. [54]
  55. [55]
  56. [56]
  57. [57]
  58. [58]
  59. [59]
  60. [60]
  61. [61]
  62. [62]
  63. [63]
  64. [64]
  65. [65]
  66. [66]
  67. [67]
  68. [68]
  69. [69]
  70. [70]
  71. [71]
  72. [72]
  73. [73]
  74. [74]
  75. [75]
  76. [76]
  77. [77]
  78. [78]
  79. [79]
  80. [80]
  81. [81]
  82. [82]
  83. [83]
  84. [84]
  85. [85]
  86. [86]
  87. [87]
  88. [88]
  89. [89]
  90. [90]
  91. [91]
  92. [92]
  93. [93]
  94. [94]
  95. [95]
  96. [96]
  97. [97]
  98. [98]
  99. [99]
  100. [100]
  101. [101]
  102. [102]
  103. [103]
  104. [104]
  105. [105]
  106. [106]
  107. [107]
  108. [108]
  109. [109]
  110. [110]
  111. [111]
  112. [112]
  113. [113]
  114. [114]
  115. [115]
  116. [116]
  117. [117]
  118. [118]
  119. [119]
  120. [120]
  121. [121]
  122. [122]
  123. [123]
  124. [124]
  125. [125]
  126. [126]
  127. [127]
  128. [128]
  129. [129]
  130. [130]
  131. [131]
  132. [132]
  133. [133]
  134. [134]
  135. [135]
  136. [136]
  137. [137]
  138. [138]
  139. [139]
  140. [140]
  141. [141]
  142. [142]
  143. [143]
  144. [144]
  145. [145]
View Abstract