Cellular and Subcellular Structure of Neoproterozoic Animal Embryos

See allHide authors and affiliations

Science  13 Oct 2006:
Vol. 314, Issue 5797, pp. 291-294
DOI: 10.1126/science.1133129


Stereoblastic embryos from the Doushantuo Formation of China exhibit occasional asynchronous cell division, with diminishing blastomere volume as cleavage proceeded. Asynchronous cell division is common in modern embryos, implying that sophisticated mechanisms for differential cell division timing and embryonic cell lineage differentiation evolved before 551 million years ago. Subcellular structures akin to organelles, coated yolk granules, or lipid vesicles occur in these embryos. Paired reniform structures within embryo cells may represent fossil evidence of cells about to undergo division. Embryos exhibit no evidence of epithelial organization, even in embryos composed of ∼1000 cells. Many of these features are compatible with metazoans, but the absence of epithelialization is consistent only with a stem-metazoan affinity for Doushantuo embryos.

Fossilized embryos in the 635- to 551-million-year-old Doushantuo Formation of South China provide a record of embryology during the emergence of animal phyla (1, 2). Based on interpreted cellular layers and polar lobe–like structures (37), a suite of putative blastulae, gastrulae, larvae, and minute adults has been described from the deposit. These fossils may also house cellular- and organelle-scale details of cleavage patterns of early animals (8).

Unfortunately, taphonomic and diagenetic artifacts are abundant in Doushantuo embryos. These include postmortem evidence for differential cell shrinkage, cytoplasmic decay, loss of cells and embryo envelopes, patchy organic matter degradation, multiple episodes of botryoidal void-filling cementation and dissolution, clotted fabric development, geopetal infilling, ambient pyrite trail formation, and mineral over-growth (912). These features are rarely visible on the surface of embryos and are difficult to resolve in individual thin sections or in embryos lacking complete embryo envelopes. To constrain interpretations of metazoan affinity and resolve internal structures in these embryos, we identified a suite of 162 relatively pristine envelope-bound and spheroidal embryos in which recurrent biological structures and cleavage patterns could be distinguished from inorganic artifacts. We then used x-ray computed tomography (CT) in concert with scanning electron microscopy (SEM), transmission electron microscopy (TEM), and thin-section petrography to examine their internal morphology in three dimensions. Because variations in x-ray attenuation within embryos typically correspond to mineralogical or density variations associated with cell boundaries, subcellular structures, and diagenetic cements (1315), it was possible to screen samples and identify recurrent biological features. The cellular and subcellular architecture of each embryo was volume-rendered to determine the number, arrangement, shape, and volume of cells in each embryo, and then compared with SEM, TEM, and thin-section petrography of similar samples [see supporting online material (SOM)].

Several interpretations of Doushantuo embryos have assumed that they follow the canonical pattern of progressive and synchronous cleavage (1, 16, 17). Using image stacks of x-ray attenuation through each embryo, we extracted three-dimensional (3D) isosurface models of each cell and counted the number of cells, including unexposed cells (Fig. 1). Embryos composed of more than 64 cells were difficult to count owing to diagenetic homogenization of some cell boundaries; these and embryos exhibiting shrinkage or breakage were excluded from our analysis. Although many specimens follow the expected 2n pattern (n = 42 out of 57 specimens), as many as one in four deviate from this pattern (n ≥ 9), being composed instead of 3, 5, 7, 9, 15, 24, and 31 cells (Fig. 2, fig. S1, and movie S1). Aberrant numbers of cells have been noted in previous studies of Doushantuo embryos, and the “extra cell” has been interpreted as a polar lobe (6, 7). However, this explanation becomes less likely in later-stage embryos, in embryos that have fewer than the expected number of cells, and in embryos that deviate more substantially from the canonical pattern.

Fig. 1.

(A) Scanning electron micrograph of a 16-cell embryo. (B) Iso-surface model of the exterior of a 16-cell embryo (specimen DOU-10). (C) Schematic drawing with cells labeled to correspond to their models [(E) to (T)]. (D) An x-ray section showing very faint cell boundaries and subcellular spheroidal structures (arrowheads). Grayscale values in image have been inverted and the image placed on a black background to match (A). (E to T) Models of all 16 cells. Cells with intracellular structures (shaded dark green) are rendered transparent and those without are opaque. Corrugation on cell faces is an isocontouring artifact. (M) shows the internal cell. Scale bar in (A), 200 μm; in (E), 85 μm.

Fig. 2.

Aberrant embryos. (A) Reflected-light photomicrograph of a 3-cell embryo (specimen DOU-25). (B) Exterioriso-surface model. (C) Volume-rendered and extracted cell models, with the left cell rendered transparent to show the only sub-cellular structure (shaded green) in this embryo. (D and E) Seven-cell embryo (specimen MPLa). (D) Exterior isosurface model. (E) Volume renderings of each cell in (D). Cells in (C) and (E) are oriented in same position as they occur in (B) and (D). Scale bar in (A), 200 μm; in (D), 130 μm.

Deviation from the canonical 2n pattern can result from taphonomical, pathological, or developmental processes. Taphonomic loss of cells and pathologic production of unequalsized cells can result from degradation, anoxia, and toxicity, but such patterns are irregular and probably variable among individuals of the same species (18). Thus, pathologic and taphonomic causes may be excluded at least for Doushantuo embryos with regularly shaped cells, leaving developmental control as a possible explanation. Perhaps metazoan cleavage was less well organized and constrained in the Neoproterozoic, in comparison to living metazoans. Noncanonical cell numbers in marine embryos can result from several developmental processes (see SOM). Many are peculiar to unequally cleaving embryos, resulting from differences in cell division machinery that produce a large and a small daughter cell rather than two equal-sized ones. In addition, noncanonical cell numbers can result from asynchronous timing as a normal part of equally cleaving cell lineages. Asynchronous cleavage of cell lineages is widespread among metazoans. In such marine embryos, a 2n cleavage pattern is followed, but because not all cell lineages divide at the same rate, the numbers of cells at any time do not necessarily match this count sequence.

We examined serial x-ray attenuation slices through the x, y, and z axes of a number of 2- to 32-cell embryos in order to identify cells that have distinct subcellular structures (6, 8, 12, 14). The structures are typically characterized by a different degree of x-ray attenuation than the surrounding cellular material, suggesting that they differ in composition or density from the matrix (Fig. 3). There are two types of intracellular structures. The first are small spheres seen in various kinds of embryos, including two- and three-cell embryos that preserve distinct membranes (Fig. 3A and movie S2). The spheres differ from clotted and botryoidal fabrics and are coated by layers with relatively high x-ray absorption. Some spheres exhibit evidence of collapse and deflation (Fig. 3C). Internally, they are either empty or have a lower degree of x-ray attenuation than the surrounding matrix, and thin-sectioned samples indicate that their interiors are more organic-rich than the matrix. Together these features are consistent with apatitic encrustation of degrading organic spheres, which are comparable in shape and size to membranebound cytoplasmic vesicles, lipids, or yolk granules in the cleaving blastomeres of metazoans such as demosponges (19) and echinoids (Fig. 3B).

Fig. 3.

(A) A two-cell embryo (specimen MPKxiv). (A1) Surface rendering of the embryo based on tomographic scans, showing the position of the orthoslice (A2), which reveals internal preservation of spheroidal subcellular structures analagous to modern lipid vesicles or yolk granules. (A3) High magnification of an orthoslice at the boundary between the two cells. (B1 and B2) Two-cell embryo of the sea urchin Heliocidaris erythrogamma, including an enlarged view of the lipid vesicles. Such vesicles are spheroidal and can be even larger in large-egged embryos of other modern invertebrates. (C) Orthosliced volume rendering of a possible embryo (specimen SB0604Clea), illustrating intact, deflated, and collapsed spheroidal and ovoidal interior structures, coated by several generations of cement. The left hemisphere is dominated by a clotted fabric. Representative deflated (white arrowhead) and collapsed (black arrow) spheres are indicated. Slices in (A) and (C) are negatives: Darker grayscale values represent lower x-ray attenuation. (D and F) Photomicrographs of four-cell embryos with subcellular structures (arrowheads). (F) Extracted, embedded, and thin-sectioned specimen. (E1) X-ray section through a four-cell embryo (specimen DOU-8), showing three of the cells with paired subcellular structures (one pair is indicated by arrowheads), each with slightly greater attenuation (dark regions in image). (E2 to E4) Isosurface models of the four tetrahedrally arranged cells, with two of the cells extracted. All cells have paired subcellular structures, but only the upper two cells are rendered transparent to show intracellular structures (shaded green). (E4) Embryo from (E2), rotated horizontally 180°. (G) TEM images of reniform (G1) to spheroidal (G2) structures like those in (D) to (F). Although the exact orientation of the embryo in (G1) is poorly constrained because of difficulties inherent in sample embedding and microtoming, it is hypothesized that this represents a tangential section through a reniform subcellular structure, so that each kidney-shaped structure is represented by two subcircular, more electron-dense areas (dark regions in image). (H) Transmitted-light photomicrograph of much smaller algal or cyanobacterial cells, each with one or more intracellular structures that occupy much of each cell. Relative scale bar size is as follows: for (A) and (B), 170 μm; for (C), 270 μm; for (D), 150 μm; for (E), 150 μm; for (F), 190 μm; for (G1), 8 μm; for (G2), 5 μm; for (H), 20 μm.

The second group (n = 10) consists of spheroidal-to-reniform structures, present as one or two per cell, as revealed by isosurface modeling (Fig. 3, D to G, and movies S3 and S4). In contrast to the first type, these show a slightly higher x-ray attenuation than the surrounding matrix. Mirror-image pairs of these structures have been observed in all cells of a four-cell embryo (Fig. 3E), but in only one cell of another (fig. S2), and single intracellular structures exist in some cells of 3-, 4-, 8- and 16-cell embryos (Figs. 1 and 2C, figs. S1 and S3, and movie S5). Thin-section petrography and TEM of epoxy-embedded four-cell specimens bearing these structures were performed (Fig. 3, F and G) and reveal that the reniform structures are more electron-dense, which is consistent with relative x-ray attenuation values. Together, these observations suggest early diagenetic replacement of a compositionally distinct zone in the center of the cell, rather than later-stage cavity infilling.

There are two possible interpretations for these paired structures. First, they may represent inorganically precipitated diagenetic or taphonomic features. Intracellular structures are well known in coccoidal cyanobacterial cells preserved in cherts of the Bitter Springs Formation of Australia, and these have been interpreted as degraded cytoplasmic bodies because of their highly variable size and morphology (2022). Indeed, some of the much smaller Doushantuo algal cells also contain intracellular structures with variable morphologies (Fig. 3H), and some Doushantuo embryo cells contain eccentrically located internal bodies covered with botryoidal encrustation [for example, see figure 3.7 of (16)] that are best interpreted as degraded cytoplasm. However, the reniform subcellular structures of Doushantuo embryos have consistent size, occurrence, and location among the studied population and have diffuse margins that grade into surrounding matrix. They may represent fossilized nuclei, spindle bundles, or other organelles (8, 14). The size, shape, and mirrored orientation of paired intracellular structures in four-cell embryos are similar to those of spindle bundles or divided nuclei formed during karyokinesis in extant metazoan embryos (23). The long axis of the reniform bodies is parallel to an anticipated cell division plane, which might pass through the junction of the two largest polyhedral cell faces.

Rotation of each embryo's polyhedral cell models (movie S4) reveals the packing topology of cells. Two-cell embryos typically have hemispherical cells. In four-cell embryos, each cell typically has one convex face (the exposed exterior surface) and three planar faces that correspond to its three neighboring cells (Fig. 3E and fig. S2). Three-, five-, seven-, and nine-cell embryos reveal evidence of the rearrangement of cell topology between adjacent 2n stages (Fig. 2E), bearing out inferences of flexibility in cell membranes (17). In eight-cell embryos, each cell has four or five neighbors and planar interfaces (fig. S3). In 16-cell embryos, one of the cells sits in the center of the embryo and is surrounded by the remaining 15 cells, and thus has 15 planar faces (Fig. 1 and movie S5). The remaining exposed cells each have one convex exterior surface and six planar faces, which correspond to the five adjacent exposed cells and the internal cell. In embryos with ∼32 cells, there are abundant many-faceted cells toward the exterior of the embryo, and many of these are more spheroidal than in earlier-stage embryos (movie S6). There are also several cells in the interior of each of these embryos, some of which are more elongate, pyramidal, or hourglass-shaped.

In many embryos composed of ∼64 or more cells, the center is structureless, and embryos composed of hundreds to more than ∼1000 cells still show no evidence of blastocoel formation or the organization of blastomeres into epithelia (fig. S4 and movie S7). Epithelialization, by whatever mechanism of gastrulation, should be underway in modern embryos with >100 cells.

The affinity of the embryos remains problematic because evidence of later developmental stages, juveniles, and adults is lacking, and because features in early developmental stages are often more widely distributed among animals than are those arising later in development (24). Despite hypotheses that Doushantuo embryos are unusual in comparison to most known metazoans (25), the patterns of cleavage and cell topology are compatible with a range of animal groups. For instance, in embryos composed of eight or more cells, the offset arrangement of successive tiers of cells, strong cell cohesion, and a stereoblastic cell topology are comparable to early cleavage embryos of many arthropod groups. Stereoblastulae are also particularly common among sponges (26) and scyphozoan cnidarians (27). Doushantuo embryos composed of many hundreds of cells resemble the purported gastrulae of demosponges, before the development of parenchymella larvae, although at this stage demosponges exhibit evidence of gastrulation, with a differentiated superficial layer of micromeres surrounding a core of macromeres (19).

None of the 162 embryos that we analyzed show any evidence of epithelial development. This absence of evidence is striking, given that the presence of subcellular structures in these fossils demonstrates a level of preservational fidelity comparable to that of the most celebrated fossil deposits (28). The absence of this 3D hallmark of sponge- and higher-grade metazoans (29, 30) may indicate that they did not yet exist. Germ layers and tissue-grade structures suggested in Doushantuo microfossils (35) may instead be diagenetic artifacts (911). Together with topologic evidence and the absence of cell walls (17), which falsify an algal affinity, the combined observations suggest that the Doushantuo embryos are probably stem-group metazoans.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S4

Movies S1 to S7

References and Notes

Stay Connected to Science

Navigate This Article