Early ontogeny and placentation of the grey short-tailed opossum, Monodelphis domestica (Didelphidae: Marsupialia): contribution to the reconstruction of the marsupial morphotype

Introduction

This study is a contribution to the reconstruction of the morphotype of recent marsupials by providing current findings, a character polarization and a phylogenetic interpretation.

Marsupialia (Metatheria) and Placentalia (Eutheria) form the monophyletic Theria, which are the sistergroup of the Monotremata (Novacek et al. 1988; Shoshani and McKenna 1998; Zeller 1999b). Although marsupials and placentals are both viviparous (unlike the egg-laying monotremes) their modes of reproduction differ considerably (see Table 1). The polarization of most of these characters, as well as their evolutionary significance are still unresolved. For the reconstruction of the evolution of reproductive strategies in both groups, the morphotypes, which are the sum of characters present in the last common ancestors (cf. Hennig 1950; Königsmann 1975), of Theria, Marsupialia, Placentalia and internal groups must be reconstructed.

Table 1. Characters of reproduction in placentals and marsupials

The morphotype reconstruction is based on marsupial systematics as proposed by Luckett (1994). According to Luckett (1994), marsupials can be divided into the monophyletic sister-taxa Australidelphia and Ameridelphia. Ameridelphians comprise all American marsupials except Dromiciops australis, australidelphians consist of all Australasian marsupials including the South American Dromiciops australis. The monophyly of these groups is supported by the sperm pairing in the epididymis of ameridelphians (Temple-Smith 1987) and the fusion of two facets on the calcaneus (confluent lower ankle joint pattern, Szalay 1993) in australidelphians. This basal dichotomy of marsupials is essential for the morphotype reconstruction, based on character optimization and outgroup comparison. For character optimization and outgroup comparison the further division into syndactyls and diprotodonts is useful. The phylogeny of marsupials is represented by the cladogram in Fig. 1.

The differences in reproductive biology could have had important influence on the differences in extinction rates, radiation and migration between marsupials and placentals at the Cretaceous–Tertiary boundary (Archibald 1982; Case and Woodburne 1986; Caroll 1997). We regard the reconstruction of the marsupial morphotype as a first step toward an evolutionary scenario that explains the mammalian evolution at the Cretaceous–Tertiary boundary.

Materials and methods

Monodelphis domestica was bred in a colony established at the Institute of Anatomy of the University of Göttingen. This colony was later moved to the Institute of Zoology of the University of Mainz and then on to the Institute of Systematic Zoology at the Museum of Natural History, Berlin, where it is presently located. Estrus of female Monodelphis is induced by the presence of male pheromones (Fadem 1985; Hinds et al. 1992). Copulation takes place within the estrus (Baggott et al. 1987; Fadem and Rayve 1985). Ovulation occurs 20 (Mate et al. 1994), 22.5 (n=1, own observation) to 24 h (Baggott and Moore 1990) after copulation. Fertilization occurs 22 (Mate et al. 1994), 22.5 (n=1, own observation) to 24 h after copulation (Mate et al. 1994). Mating was monitored by direct observation. All embryos and stages mentioned in this study are dated according to the time of copulation. Thus, the first day of pregnancy coincides with the day of fertilization, that is, about 24 h after mating. Parturition occurs between the 14th and 15th day after copulation (Fadem et al. 1982; personal observation) after a 13- to13.5-day period of gestation (Mate et al. 1994).

The consistency between our findings and descriptions of Mate et al. (1994) confirmed the normality of embryos described in this study. Abnormal stages were not included in the description. In addition, the chronology of development in M. domestica as described here matches the timetable given by Mate et al. (1994).

Twenty-seven adult female M. domestica were killed with chloroform at different times of pregnancy, ranging between a few hours and 15 days after copulation. They were fixed by perfusion through the vascular system immediately after their death. Gravid uteri, embryos and pieces of yolk-sac placenta were prepared for histochemistry, light and electron microscopy (EM) according to the following procedures.

Histochemistry

Four methacrylate sections of specimen 82 [29 h post copulation (hpc)] and one frozen section of specimen 135 [7 days post copulation (dpc)] were stained with Kongo red (Romeis 1968). One frozen section of specimen 135 was also stained for keratin by Martinotti (Romeis 1968) to demonstrate the shell coat. Numerous frozen sections of specimen 135 were stained with Sudan red.

Twenty-four methacrylate sections of specimens 45 (22.5 hpc), four of specimen 85 (29 hpc) and eight of specimen 95 (4dpc) were stained with PAS. Six paraffin sections of specimen 45 (22.5 hpc) were stained with peroxidase-conjugated Ulex-europaeus-lectin (substrate: DAB).

Four frozen sections of specimen 135 were fixed in acetone (6 °C) for 10 min and treated with antibodies against keratin, laminin, collagen type IV and fibronectin (Table 2); the immunoreaction was visualized by the usage of biotinylated secondary antibodies and the biotin–avidin–peroxidase method according to Dako LSAB^®-Kit (DAKO Diagnostika GmbH, Hamburg, Germany). In addition, sections were treated with collagenase (Sigma no. C-0773; Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany). Antibodies against transferrin were applied to two frozen sections of specimen 132 and to two paraffin sections of specimen 28 (12 dpc), followed by the application of Dako LSAB^®-Kit.

Table 2. Female Monodelphis domestica investigated in this study

Control: sections treated with all chemicals used in immunohistochemistry without the primary antibody did not stain.

Taxonomy

Species names are used according to the nomenclature of Wilson and Reeder (1993). If authors cited in this paper used synonyms, the correct name is given in square brackets. Classification of species into higher taxonomic units also follows the taxonomy of Wilson and Reeder (1993). The species included in both the discussion and the morphotype reconstruction belong to the higher marsupial taxa as summarized in Table 3.

Table 3. Species mentioned in this study and their relation to higher marsupial taxa

Results

By 3 hpc, 11 tertiary follicles were present in the right ovary investigated. The oocytes contain large yolk vacuoles and are surrounded by follicular epithelial cells. Paired spermatozoa can be seen in the oviduct on SEM images. One hour later, the follicle epithelium degenerates.

By 22.5 hpc, the corpora lutea have developed. The eight oocytes passing each oviduct lack a corona radiata and are merely surrounded by the zona pellucida, which is PAS-reactive. They measure 170 μm in diameter. The oocyte nucleus is located in the peripheral, formative cytoplasm, which contains minimal yolk (Fig. 2a). The formative, yolk-free cytoplasm encloses the yolky deuteroplasm (Fig. 2a), which in turn encloses a small central zone that contains little yolk. Paired spermatozoa are visible inside the oviduct by light microscopy (LM) and SEM (24 hpc). Fertilization occurs in the oviduct (Fig. 2a). At this time, the second polar body is separated off. The surface of the zygote is in close association with the tubal epithelium (Fig. 2a) and binds the lectin Ulex europaeus agglutinin (UEA) like oviduct cells, but unlike the uterine epithelium. Mucoproteid layers covering the surface of the conceptus are secreted by non-ciliated oviductal cells. Corpora lutea are already present in the ovary.

By 29 hpc, the zygotes (n=5) have entered the uterus in the pronuclear stage. Thus, a maximum of 7 h is needed for the passage down the oviduct. Each zygote is covered by a concentric, lamellar mucoproteid layer (mucoid coat) (Fig. 2b), in which spermatozoa are trapped. A secondary yolk polarity is achieved by polar accumulation of the formative cytoplasm around the pronuclei (Fig. 2b), which fuse at 31 hpc. The uterine epithelium secretes the proteinaceous shell coat, which is Kongo red reactive but does not bind antibodies against keratin.

At the fourth day after copulation, the 16-cell stage (n=1) has been achieved. The yolk, which is not affected by cleavage, forms a compact mass opposite the blastomeres (Fig. 2c). This is due to deutoplasmolysis and apocrine yolk extrusion during the first and second cleavage at 42–50 hpc and 58–68 hpc, respectively (Fromm and Zeller 1997). The diameter of the conceptus measures 180 μm inside the zona pellucida and its entire diameter measures 440 μm. The mucoid coat surrounding the zona pellucida consists of two layers. The thinner outer one is strong PAS reactive, whereas the thicker inner one is less reactive to PAS. Pseudopodia-shaped processes of mucoid coat and shell coat project into the niches of the endometrium (Fig. 2d). The shell coat cannot be stained with PAS.

By 5 dpc, the blastomeres of the conceptus (n=7) contain large yolk vacuoles. One day later, the zona pellucida is partly broken down (n=9). The blastomeres containing yolky lipid droplets in vacuoles form the early unilaminar blastocyst. The mucoid coat is still visible.

By 7 dpc (bilaminar blastocyst, n=10), the mucoid coat and yolk have been incorporated into the blastocyst and the blastomeres are apposed to the peripheral shell coat (Fig. 2e). At this time, the diameter of the blastocyst measures 340 μm and almost doubles within one day (670 μm by 8 dpc). At this stage the pseudopodia-shaped projections are absent. The embryonic area partly consists of a thickened ectoderm with underlying endoderm, and it does not form an inner compact cell mass (Fig. 2e).

The shell coat surrounding the blastocyst cannot be stained with PAS or Alcian blue and does not bind antibodies against cytokeratin, laminin or fibronectin. It is resistant to collagenase, but can be stained for keratin by Martinotti, and Kongo red. Keratin was not detectable by immunohistochemistry.

By 10 dpc (embryo, five to six somites, n=6), the vesicle has formed a bulging sphere 5 mm in diameter. The embryo has a neural groove, neural crest with a developing mesenchyme, a primitive streak, a paraxial mesoderm with five to six somites, a lateral plate mesoderm with pericardium, extraembryonic mesoderm and a yolk sac cavity.

One-third of the yolk sac lying towards the embryonic pole forms the trilaminar omphalopleure, the abembryonic part is bilaminar. Ultrastructurally, the trophendoderm (yolk sac endoderm) consists of flat cells connected by cell projections and tight junctions. Wing-shaped processes of trophendoderm cells partly appose to the bases of trophectoderm cells in bilaminar omphalopleure and project with short processes into the basal folding, whereas extended parts of trophectoderm (ectoderm of the yolk sac) cells remain uncovered. The cells have abundant polyribosomes, mitochondria, endoplasmatic reticula (ER) and lipid droplets.

The trophectoderm (outer wall of the yolk sac) is formed of columnar to flattened distinct cells leaving small intercellular capillaries for transport in between. Apices are united by junctional complexes. Exocytotic vesicles secrete into wide intercellular gaps (about 15 μm in width) between trophendoderm and trophectoderm. The apical surface is densely covered with microvilli directed to the endometrium. The cells contain numerous mitochondria, smooth and rough ER, polyribosomes, vacuoles and lipid droplets. Endocytotic complexes and transport vesicles are present. The cell bases are folded. All this indicates active transepithelial transport of material.

The cells of the one-layered, high-columnar uterine epithelium are densely covered with microvilli on their luminal surface, their bases border a capillary net in the propria. A basal lamina could not be found. The propria contains only few cells. The endometrial glands consist of a thick pad of coiled tubular glands, which in some areas extend for 3 mm. They are devoid of glandular ducts and secrete histiotrophes directly into the uterine lumen. The myometrium does not exceed 0.2 mm. The shell coat is still intact and measures 1.8 μm.

By 11 dpc (4 mm total length), the differentiation of the frontal half of the embryo is highly advanced following rapid development. It shows a strong curvature of head and neck, a dorsally concave lumbar curvature, a foregut with visceral clefts and visceral arches, brain plate with eye anlagen, ear placodes, a heart anlage, and crest-shaped anlagen of upper extremities. The vitelline duct is still wide open. The neural tube, somites, notochord, aortae, coelom and nephrogenic crest can be seen on transverse sections through the embryo. The cranial and caudal ends of the neural tube are still open and form neuropores, respectively. Two-thirds of the anterior end of the embryo are sunk into the yolk sac and covered with the two-layered proamnion (ecto- and endoderm). The caudal third of the embryo is still part of the surface of the conceptus. A small extra-embryonic coelom can be seen between the trilaminar yolk sac and proamnion.

The yolk sac is trilaminar for over one-third of its extension. It measures 5–6 mm in diameter and is greatly expanded with elaborate folds increasing the surface area (cf. Fig. 2f). It is closely apposed to the endometrium (Fig. 3a). The endodermal cells of the bilaminar omphalopleure have long projections enclosing large intercellular spaces between trophendoderm and trophectoderm. In the trilaminar yolk sac wall the endodermal cells are flattened and closely connected to each other. Blood vessels with endothelium and haemocytoblasts are developed within the mesenchyme of the trilaminar yolk sac. The sinus terminalis appears at the border between the bi-and trilaminar omphalopleure.

The trophectoderm mainly consists of distinct cells united by desmosomes with intercellular transport capillaries in between. In particular, the cells of the trophectoderm of the bilaminar omphalopleure are flat and contain large vacuoles exceeding the diameter of the cell nucleus. Ultrastructurally, the trophectoderm is formed in most parts by distinct cells, which are apically united by junctional complexes (Figs 3b, 4a) with intercellular transport capillaries in between (Fig. 3b). Some trophectodermal cells of the trilaminar omphalopleure have already fused and intercellular borders can not be seen. The nuclei of those cells are close to each other. The trophectoderm is densely covered with microvilli (Figs 3a, 4a), and pinocytotic vesicles are present on its surface. The cells contain numerous transport vesicles, mitochondria, lipid droplets, rough and smooth ER, free ribosomes, and big euchromatic nuclei with a maximum diameter of 21 μm (Figs 3b, 4a). This indicates the contribution of the trophectoderm to both transport and synthesis.

The cells of the single-layered, high-columnar uterine epithelium contain often more than one large euchromatic nucleus (up to 15 μm in diameter) with prominent nucleoli (Fig. 3a). They are connected by apical junctional complexes, and are covered with a dense apical border of microvilli of 3 μm length (Fig. 3a). Secretory canaliculi are located between the cells of the apocrine secreting epithelium. The cells contain mitochondria, rER, polyribosomes, secretory granules and abundant lipid droplets (Fig. 4a). The uterine epithelium mainly serves as a pathway for materials from the maternal blood to the embryo. The basal plasmalemma is folded and borders on a capillary net in the propria, which is situated 0.5–2 μm beneath the epithelium. A basal lamina is lacking. The capillary endothelium is not fenestrated. Two layers of endometrial glands can be distinguished: The endometrial glands near the myometrium consist of a single-layered secreting epithelium with basal nuclei and cytoplasm that is comb shaped due to abundant inclusions. The endometrial glands near the uterine epithelium have mostly homogeneous cytoplasm and central nuclei, with inclusions and a comb-shaped cytoplasm that is restricted to their bases. The cells of the uterine glands contain ergastoplasm, dictyosomes, mitochondria, lipid droplets and secretory granules. The surface is covered partly with branching microvilli and partly with kinocilia. The uterine glands are the sites of synthesis and secretion of histiotrophes. The propria contains more free cells (histiocytes, plasmacytes, lymphocytes, polymorphonuclear granulocytes) on its subepithelial side than on its basal side.

The shell coat, which measures 2.3 μm, is unruptured at this stage (Figs 3a, 4a). It consists of narrow meshes, through which all material for the exchange between mother and embryo has to pass. Shell coats of adjacent embryos do not fuse.

By 12 dpc [5 mm crown–rump length (CRL)], the differentiation of the embryo is more advanced, especially in the facial and branchial regions as well as the upper extremities. The large heart supports the yolk sac circulation. The lumbar curvature is flattened out, the caudal amnion expanded and the embryo has completely sunk into the yolk sac. The exocoelom is expanded.

Only the yolk sac, which is now vascularized over two-thirds of its extension, contributes to the placenta (Fig. 2f). The endodermal cells of the vascular part are flat and in close contact with each other. The yolk sac vessels are the sites of blood formation by mitotic multiplication of proerythroblasts and myeloblasts. Immunohistochemistry demonstrates transferrin within the maternal epithelium as well as in the trophectoderm and the underlaying vascular layer of the omphalochorion, indicating transferrin-rich histiotrophe passing the meshes of the shell coat (Zeller 1999a).

The trophectoderm still consists in most parts of distinct, laterally interconnected cells, that are united by apical junctional complexes (compare Fig. 4a). The length of the microvilli is reduced. The cells contain large vacuoles and big nuclei mainly arranged in a regular pattern. At some areas of the trilaminar omphalopleure the nuclei of the trophectoderm are situated close to each other and between those areas the trophectoderm free of nuclei is flattened. The extent of folding of the endometrium is increased (Fig. 2f). The length of microvilli of the uterine epithelium is reduced (1.7 μm). Macro- and microapocrine secretion occurs. The endometrial glands secrete histiotrophes (Fig. 3c). The thickness of the glandular pad is reduced to about 1.5 mm. The former two-layered condition of the glandular pad has been lost. The shell coat, remnants of which can be seen between trophoblast and endometrium (Fig. 3c), is ruptured.

By 13 dpc (8.5 mm CRL), the neural tube is closed, peripheral nerves and ganglia, the eye cup, the labyrinth vesicle, an epithelial nasal sac, a muscular tongue, a chondrified endoskeleton with occipital pillar, an axial skeleton, Meckel’s cartilage and the anterior extremity, a gut canal, liver, heart, lung and mesonephros are differentiated. The frontal acropodium is pentarch, whereas the back one is still paddle-shaped.

The shell coat is completely broken down. Most of the trophectodermal cells of the trilaminar omphalopleure have fused to form a trophoblastic syncytium (see below).

By 14 dpc (10 mm CRL), the embryo has a two-layered head amnion lined by high microvilli on the side of the yolk sac. This indicates a transport of amniotic fluid between yolk sac and amnion. The exocoelom, which never covers the whole embryo, is located ventrolaterally to the embryo, from its navel to the caudal end. The allantois is small.

The trophendoderm of the trilaminar omphalopleure consists of flat, expanded cells united by junctional complexes. It is covered with short microvilli on the side of the yolk sac lumen. The yolk sac vessels around the navel are embedded in large, bulging cells. The trophendodermal cells of the bilaminar omphalopleure are also flattened and close to the base of the trophectodermal cells (Fig. 5d).

Most cells of the trophectoderm of the trilaminar omphalopleure have fused to form a syncytium (Figs 4b, 5a). Cell borders are rare. The thin basal lamina underlying the single layered trophectoderm measures about 150 nm. The syncytium is covered with microvilli and shows pinocytosis, intracellular vesicles and filaments (Fig. 3d), indicating active transepithelial transport of materials. It is rich in rER, polyribosomes and lipid droplets (Fig. 3d) and multivesicular bodies. Nuclei are located in syncytial knots (2–5 nuclei per knot) (Fig. 4b). Between the syncytial knots are thin lamellae of syncytial epithelium (Fig. 4b). Occasionally, cell borders were found within the syncytial lamellae. Lipid droplets and accumulations of glycogen are mainly located in the knots, cell organelles are rare. The trophectoderm is in close association with the uterine epithelium with an intimate interdigitation of trophoblastic and endometrial microvilli (Fig. 3d). The height of the microvilli does not exceed 1.6 μm. The invasive trophoblast penetrates the maternal epithelium at regular intervals, and reaches the maternal endothelium (Figs 3e, 5a). To the side of the invasive areas, the trophoblast is intimately interconnected with the uterine epithelium by newly formed cellular adhesions (Fig. 5a). The invasive trophoblast displaces the uterine epithelium at either small (Fig. 3e) or expanded areas (Fig. 5a). Thus, a mixed epithelio-endothelio-chorial yolk sac placenta is formed. At the invasive areas, the maternal endothelium, the trophoblast (with thin basal lamina underneath), interconnected cells of the connective tissue of the yolk sac, the extracellular matrix and the foetal endothelium form the layers, through which the foeto-maternal exchange of material and gases takes place. Owing to the thinning of foetal layers, the barrier between foetal and maternal blood does in some areas not exceed 2.5 μm. In these areas, the uterine epithelium is close to the cells of the foetal connective tissue near the vitelline vessels (Fig. 4b). Bilaminar parts of the yolk sac are rare, probably due to fusion of yolk sacs of adjacent embryos and subsequent loss of these areas (Freyer in prep.). In the bilaminar yolk sac wall, fusion of trophectodermal cells could not be found (Fig. 5b–d). Thinning of the trophoblast and accumulations of lipid droplets are similar to the trophoblastic syncytium (Fig. 5b–d). The trophoblast of the bilaminar yolk sac walls were closely attached to the surface of the maternal epithelium. However, invasive areas could not be found.

The endometrium is strongly folded. The largest folds form chambers for single embryos and their foetal appendices. As pointed out above, the height of the microvilli is reduced. The thickness of the endometrial glandular pad is reduced to 0.2–0.4 mm. The thickness of the myometrium remains unchanged at 0.1 mm.

At the 15th day after copulation, the propria is highly invaded by polymorphonuclear granulocytes, macrophages and lymphocytes (Fig. 3f). Granulocytes are also found in the glandular tubes. This is followed by the parturition of the placenta.

Discussion

In the present study of the grey short-tailed opossum, M. domestica, the early ontogeny and placentation from ovulation to parturition are described. To confirm and complement previously published detailed studies and timetables of early embryonic development (Harder et al. 1993; Mate et al. 1994; Selwood et al. 1997), the development of the oocyte and zygote, the embryonic development, as well as changes of the embryonic membranes and the endometrium during ontogeny are considered. Special emphasis is placed on the ultrastructural changes of the placenta, and the development of structures is interpreted functionally, where possible, by correlating data obtained from LM, EM, histochemistry and immunohistochemistry. The following discussion is aimed at reconstructing the marsupial morphotype based on the character distribution of own findings and literature data.

Placentation

Our observations agree with that of Harder et al. (1993) concerning both the close apposition of the trophectoderm to the uterus at 13 and 14 dpc and the microvillous surface of the endo- and trophectoderm in M. domestica. Our observed increase of foldings at 12 dpc is confirmed by the increase of the index of epithelial surface area from 1.66 at 9 dpc to 6.9 at 13 dpc measured by Harder et al. (1993). We can further confirm, that M. domestica possesses a choriovitelline placenta (Harder et al. 1993; Roberts and Breed 1994a) during the entire development from implantation to birth. The observed drop in uterine gland density in M. domestica between 9 and 14 dpc (Harder et al. 1993) can be correlated with the decreasing thickness of the uterine gland pad from 3 mm at 10 dpc, 1.5 mm at 12 dpc to 0.2–0.4 mm at 14 dpc. This indicates the shift from histiotrophic to haemotrophic nutrition of the embryo.

Invasiveness of the placenta

Occurrence and degree of invasion of uterine epithelium by the trophectoderm of the bilaminar and trilaminar yolk sac varies among marsupials as summarized in Table 4.

Table 4. Distribution of invasive bilaminar and trilaminar yolk sac areas among marsupials

Contrary to earlier reports given by Harder et al. (1993) and Roberts and Breed (1994a), who report the trophectoderm of M. domestica to be non-invasive, our electron-microscopic observations clearly demonstrate that the trophoblastic syncytium of the trilaminar yolk sac penetrates the uterine epithelium and is thus invasive (Fig. 3e, Fig. 4b). An invasive trophectoderm in a didelphid was only known for Philander opossum (Enders and Enders 1969).

The variation between invasive and non-invasive trophectoderm among closely related species may be explained by either interspecific variation or a lack of information. The latter suggestion is supported by the findings of an invasive trophectoderm in M. domestica, which was formerly described as possessing a non-invasive trophectoderm. Thus, this character needs to be re-examined during the entire ontogeny from implantation to birth, and by EM.

Formation of an allantochorion and yolk sac placenta

The different conditions of the allantois relative to the chorion found in different marsupial taxa were often mentioned and grouped into four placental types (Sharman 1959; Hughes 1984; Tyndale-Biscoe and Renfree 1987), which are summarized in Table 5.

Table 5. Types of marsupial placentation according to Tyndale-Biscoe and Renfree (1987)

However, it must be kept in mind that this classification is an artificial grouping of different characters, namely the contact between allantois and chorion, the size of allantois and yolk sac, their vascularization and the function of both allantois and yolk sac. This is the reason why a character polarization based on these types has failed. For a phylogenetic interpretation of these findings, a polarization of homologous characters is needed. To date, the insufficient and inconsistent data concerning the allantochorion in marsupials can only be polarized by comparison with the sistergroup (eutherians) of marsupials and the outgroup of the Theria (monotremes). Since a fusion between chorion and allantois was found in Dasyuridae, Vombatiformes and Perameloidea, and this is a character also found in eutherians (Wooding and Flint 1994) as well as in monotremes (Caldwell 1887), this may be a plesiomorphic character of the marsupial morphotype. Thus, the absence of a chorioallantois in didelphids on one side and australidelphian taxa (acrobatids, macropodids, phalangerids) on the other side is a convergent reduction. The parsimony of this assumption, however, is weakened by the absence of an allantochorion in the dasyurid S. crassicaudata (Roberts and Breed 1994a). In a morphotype reconstruction using parsimony (see Cunningham et al. 1998), the downpass reconstructed marsupial morphotype remains ambiguous, but the uppass reconstruction suggests an allantochorion to be present in the marsupial morphotype (Fig. 6). The absence of an allantochorion in phalangeroids (phalangerids, acrobatids) and macropodids may thus be synapomorphic. Since the allantochorion is vascularized in monotremes (Caldwell 1887), placentals (Wooding and Flint 1994), peramelids and vombatiforms, it is to date parsimonious to assume a reduction of this vascularization in D. viverrinus derived from a vascularized allantochorion in the marsupial morphotype. This preliminary polarization, however, demands a re-examination of recent findings, the examination of the ontogeny from implantation until birth in further species and the re-evaluation of marsupial systematics, which are in progress (Freyer, Zeller and Renfree in prep.). We are aware that our hypothesis appears to contradict with the interpretation of Luckett (1977; p. 490), who assumes a ‘retardation in expansion of the exocoelom and allantois’ in the marsupial morphotype.

Although the chorioallantoic placenta in marsupials never forms mesodermal villi and never replaces the yolk sac placenta as that of eutherians (Luckett 1977), it can be assumed that the allantochorion in vombatiforms and peramelids serves at least respiratory function (Luckett 1977; Semon 1894; Hughes 1984; Tyndale-Biscoe and Renfree 1987). A respiratory function of the vascularized allantochorion in the marsupial stem species seems likely. However, it is unknown whether this allantochorion formed a placenta which also allowed the exchange of material.

All marsupials so far examined possess a yolk sac placenta for most of the time after implantation (cf. Tyndale-Biscoe and Renfree 1987). Thus, a yolk sac placenta can be assumed for part of the ontogeny in the marsupial morphotype (Luckett 1977), irrespective of the formation of an allantois placenta at near-term stages.

Formation of a trophoblastic syncytium

Zeller (1999a) provides the first evidence of a trophoblastic syncytium ever observed in a marsupial by EM. Here, we give a detailed description of the ultrastructure of this syncytium by presenting original electron micrographs. We assume that the aggregations of nuclei in plasma accumulations are syncytial knots rather than plasmodia, i.e. nuclear division. This assumption is based on the constant nuclear sizes from 11 to 14 dpc (Fig. 4a versus 4b), the absence of evidence on nuclear divisions at any stage, and the great extension of the syncytial parts relative to former cell sizes and regular cell arrangements with tight junctions between single cells (Fig. 4a versus 4b). Hill (1900) reported a trophoblastic syncytium in D. viverrinus by LM. As the cell boundaries can hardly be seen by LM, this description needed to be verified by EM. However, the electron microscopic findings in M. domestica presented herein make a syncytium in D. viverrinus highly probable. There is no report of fusions between cells of the trophoblast or extended parts of it in any other marsupial species. If further examinations confirmed a trophoblastic syncytium to be present in D. viverrinus, a trophoblastic syncytium could be assumed for the marsupial morphotype.

The description of trophoblastic giant cells in M. domestica by Roberts and Breed (1994a) may represent syncytial knots of the trophoblastic syncytium not identified by the authors. The question whether the trophoblastic giant cells as observed in other marsupial species, e.g. S. crassicaudata (Roberts and Breed 1994a), are syncytical knots can only be answered through further examination.

Formation of a feto-maternal syncytium

Until now, there has been no evidence for a foeto-maternal syncytium in M. domestica. A foeto-maternal syncytium is known for the bilaminar yolk sac of Dasyurus sp. and P. cinereus (Flynn 1923). In S. crassicaudata, cells of the trilaminar omphalopleure partly fuse with uterine epithelial cells (Roberts and Breed 1994b). In peramelids (Isoodon macrourus), the trophoblast of the chorioallantois ‘may have fused with maternal homokaryons to create placental heterokaryons’ (Padykula and Taylor 1982; p.95). This data base is insufficient for a morphotype reconstruction.

Conclusions

This study is a contribution to the reconstruction of the morphotype of recent marsupials by providing current findings, a character polarization and a phylogenetic interpretation. This is based on marsupial systematics as proposed by Luckett (1994).

Our data support most of the previous examinations on the early ontogeny and placentation of M. domestica. We describe both the ultrastructure of a trophoblastic syncytium and an invasive trophoblast for the first time in M. domestica by presenting original electron micrographs. Table 6 summarizes characters that are concluded to be present in the marsupial morphotype and those that are derived, based on the character distributions within marsupials (Fig. 7).

Table 6. Characters of the marsupial morphotype (0) and apomorphic characters (1)

To date, polarization of most characters of the marsupial placenta is impossible due to insufficient and incongruent data. Obviously, a yolk sac placenta is established at early stages of placentation in the marsupial morphotype. In a preliminary parsimony analysis, a vascularized allantochorion can be assumed for the marsupial morphotype. The contribution of this allantochorion to placentation in the marsupial morphotype has not yet been established.

The early ontogeny and placentation in the stem species of recent marsupials may have occurred as follows: When released from the oviduct, the oocyte was covered by the zona pellucida and a corona radiata was lacking. The yolky cytoplasm was centrally located and surrounded by the formative, yolk-poor cytoplasm. Fertilization occurred in the oviduct. Prior to or shortly after fertilization, the yolky cytoplasm migrated to one pole (abembryonic pole) opposite the formative cytoplasm (embryonic pole). During the passage down the oviduct, the oocyte became surrounded by the mucoid coat, which was secreted by non-ciliated oviductal cells. The remaining spermatozoa were entrapped into the mucoid coat, preventing polyspermy. The mucoid coat also prevented direct contact between the oocyte and the tubal epithelium and may also have functioned as an osmotic stabilizer. With the assistance of ciliated oviductal cells, the conceptus reached the uterus within 1 day after ovulation. There was no cleavage during the tubal passage. Cells of both the utero-tubal junction and uterus secreted shell coat precursors, which polymerized into the shell coat on the mucoid coat. The shell coat may have consisted of keratin which differed in its chemical characteristics to known human keratins. The meshes of the shell coat could be passed by transferrin, nutrients and gases, but it also prevented direct contact of the conceptus with the uterine epithelium. The cleavage of the zygotes that contained moderate to large amounts of yolk started in the uterus. Most yolk was separated off at one pole during the first cleavage. Lesser quantities were subsequently extruded during the first two cleavages. A sequence of equatorial and latitudinal cleavages lead to the unilaminar blastocyst. Blastomere–zona adhesions preceded blastomere–blastomere connections. Thus, a morula stage was omitted.

Subsequently, the yolk was reabsorbed by the blastomeres and the blastocyst expanded. The zona pellucida and the mucoid coat disappeared during this process. The uptake of mucoid coat and entrapped molecules may have served as a nutritional source for the embryo. Before implantation the embryo was nourished by secretions (histiotrophe) of the uterine glands. Within the last third of pregnancy, the shell coat broke down and the trophectoderm apposed to the uterine epithelium. In certain areas the trophoblast may have penetrated the uterine epithelium. Early at implantation, the placenta was solely formed by the yolk sac. Histiotrophic nutrition may have been shifted toward an exchange of material and gases between foetal and maternal vessels (haemotrophic nutrition). Later in ontogeny, a vascularized allantochorion was formed, but did not replace the yolk sac placenta. It is unknown, whether this allantochorion served material transfer beside its respiratory function.

We regard the reconstruction of the marsupial morphotype as a first step toward an evolutionary (holistic) scenario that explains mammalian evolution arround the Cretaceous– Tertiary boundary.

Acknowledgements

We thank Dr H.-J. Kuhn for his support in maintaining the breeding colony of Monodelphis domestica at the Institute of Anatomy, University of Göttingen, until 1994 and for providing research facilities. We are indebted to the University of Mainz, Institute of Zoology, for providing research facilities during Dr Zeller’s stay there from 1994 to 1996. We are also grateful to the Humboldt-University Berlin for the substantial support in building new research facilities including animal housing at the Museum für Naturkunde since 1996. We thank Mrs. J. Zeller for technical assistance. We are indebted to Dr M. Ade, Dr S. Frahnert and Dr A. Mess for critical comments and fruitful discussions. We especially thank Dipl. Biol. P. Giere for his assistance with linguistic corrections. We thank Dr P. Luckett, Dr L. Selwood, Dr M. B. Renfree and Dr G. Shaw for critical comments and fruitful discussions. We especially thank Dr M. B. Renfree for her kind permission to use her research facilities during Dr Zeller’ s stay there from 3–16 October 2000.