In euechinoid sea urchins, the fourth cleavage is unequal and gives rise to 4 micromeres, 4 macromeres and 8 mesomeres [1,2,3]. In sea stars, cleavage is thought to be equal and all blastomeres at the 16-cell stage are expected to have similar size [20]. However, cell size asymmetries can be observed in a proportion of sea star embryos (Additional file 1: Fig S1, [15]).
This raises the possibility that cleavage of sea star embryos is not necessarily equal: it might produce less obvious, yet consistent, cell size asymmetries, possibly involved in axis determination. To test if sea star embryos present differently sized blastomeres and at what stage unequal cleavage might occur, we used high-resolution live imaging. We analyzed embryos of two sea star species (Patiriella regularis and Patiria miniata) and one euechinoid sea urchin species (Lytechinus pictus) for comparison. We performed high-resolution live imaging of embryos expressing membrane and nuclear markers from the 4-cell stage to the 16-cell stage (Fig. 1A; Additional file 2: Fig S2, Additional file 3: Movie S1, Additional file 4: Movie S2, Additional file 5: Movie S3). The animal pole was assigned as opposite to the site of formation of the micromeres in L. pictus and as the side of polar body extrusion in the sea stars. We subsequently segmented individual cells in 3D and measured cell volumes (Fig. 1A, B; Additional file 2: Fig S2). To compare cell size asymmetries across species with embryos of different sizes, cell volumes were normalized on embryo volume (Fig. 1B).
As expected, we found clear cell size asymmetries in 16-cell stage L. pictus embryos, with high volume variations between macromeres, mesomeres and micromeres (Fig. 1B). Interestingly, we also found variation within mesomeres, which showed a wide range of sizes (Fig. 1B).
The P. regularis embryos have higher variation in cell size compared to both P. miniata and L. pictus embryos at the 4-cell stage and variation similar to P. miniata at the 8- and 16-cell stages. Interestingly, the smallest cells tend to be vegetal in P. regularis, while they are mostly animal in P. miniata (Fig. 1B). Notably, the differences between larger and smaller cells in sea star embryos at 16-cell stage are comparable to the differences between macromeres and mesomeres in the sea urchin (Fig. 1B), with the largest cells in P. miniata being on average twice the size of the smallest (max/min volume ratio: 2.05 ± 0.46; Fig. 1C). Taken together, these results suggest that cell size asymmetries are consistently produced during early cleavage of the sea star embryo. In P. miniata, even though there is no clear segregation between a group of large cells and a group of small cells, the majority of cells falling on the lower end of the spectrum of possible cell sizes are animal cells. This is similar to observations in P. pectinifera [15]. In P. regularis, instead, the smaller cells are mostly vegetal. This is different from the situation in echinoid species, in which the smallest cells, the micromeres, are always found on the vegetal side [2, 9, 10, 22, 23], where they are both necessary and sufficient to induce the AP axis [24]. The position of small cells between these two sea star species is, instead, variable with respect to the AP axis.
Next, we asked if the position of small cells is predictive of DV axis formation in the sea star embryo. To test this hypothesis we first analyzed the relationship between cleavage planes and embryonic axes in P. miniata. In sea stars, the embryo develops into a blastula and gastrulation consists of invagination of mesendodermal cells at the vegetal side (opposite to the polar bodies) [15]. During gastrula stages, the archenteron elongates within the hollow ectoderm tissues and eventually joins the anterior ectoderm to open the mouth [15, 19]. The opening of the mouth, which for P. miniata happens at around 72 h post fertilization (hpf), is the first clear morphological hallmark of DV axis formation: the larva is now a bipinnaria and both mouth and anus open on the ventral side [15, 25, 26]. To confirm that first and third cleavages predict the AP axis of P. miniata larvae, we performed lineage tracing at the 2- and 8-cell stages (Fig. 2). We injected one cell with a fluorescent dextran and raised the injected embryos up to the bipinnaria stage. As expected, we found that in embryos injected at the 2-cell stage, the labelled clone constituted roughly half of the larvae, including anterior ectoderm, posterior ectoderm and mesendoderm tissues (Fig. 2A). We then measured the angle formed by the labelled clone with the animal-vegetal axis at the beginning of gastrulation (26 hpf) and found that the first cleavage aligns consistently with the animal-vegetal axis (Fig. 2B). In embryos injected in one animal blastomere at the 8-cell stage, the labelled clone included only a portion of anterior ectoderm, while in those injected into one vegetal blastomere the labelled clone included a portion of posterior ectoderm and mesendoderm (Fig. 2A, C). Next, we sought to analyze the position of the labelled clones with respect to the DV axis (Fig. 3A–D). To this aim we imaged 72 hpf larvae in toto on a confocal microscope, rendered the acquired images in 3D and virtually oriented the larva to achieve an anterior view, with the ventral side facing up (Fig. 3A, C). This allowed us to faithfully measure the angle between the DV axis and the labelled clones (Fig. 3B, D). We found that both first (Fig. 3A, B) and second cleavage (Fig. 3C, D) are positioned randomly with respect to the DV axis.
Taken together, these results show that the first cleavage is aligned with the animal-vegetal axis but is not predictive of the future larval DV axis. The third cleavage separates the embryo into animal and vegetal halves, with animal cells giving rise to anterior ectoderm and vegetal cells giving rise to posterior ectoderm and mesendoderm. This confirms what was shown for P. pectinifera [15] and our preliminary observations in P. regularis (Additional file 6: Fig S3).
To test if the cell size asymmetries arising during cleavage stages are predictive of the DV axis in P. miniata, we turned to an imaging approach. The appearance of small cells is most obvious at the 16-cell stage in this embryo: given the difficulty of faithfully injecting those smaller cells we used photoconversion to label them. We raised embryos expressing the photoconvertible protein Kaede to the 16-cell stage and then used a UV laser on a confocal microscope to photoconvert either a random cell (control) or one small cell (Additional file 7: Fig S4). We aimed at photoconverting the smallest cells of an embryo, independently of its animal or vegetal position: we selected the smallest cell once on the confocal microscope, by (1) imaging the whole embryo, (2) identifying two or three small cells and (3) measuring the three major axes of those cells to make sure the smallest was photoconverted. We then raised the photoconverted embryos to the bipinnaria stage and assessed the position of the photoconverted clones (Fig. 3E, F; Additional file 8: Fig S5, Additional file 9: Fig S6, Additional file 10: Movie S4). We found that in larvae where a random cell had been photoconverted, the labelled clones were positioned randomly with respect to both the AP and DV axes (Fig. 3G, H; Additional file 8: Fig S5). In larvae where one small cell had been photoconverted, the position of labelled clones was biased toward the ventral half of the anterior ectoderm (Fig. 3I, L; Additional file 9: Fig S6).
It is possible that the observed ventral bias in the positioning of small cells may be due to a general bias of animal pole cells to give rise to ventral tissues, due to morphogenetic events. To test this possibility, we marked one random animal blastomere at the 8-cell stage and scored the position of the same labelled clone at three different stages: 26 hpf, early gastrula, when the apicobasal axis is most obvious; 50 hpf, late gastrula, when the bending of the anus allows identifying the ventral side of the gastrula but the mouth has not yet formed; and 72 hpf, bipinnaria, when the mouth, pre-oral hood and ciliary bands have formed (Fig. 4A; Additional file 11: Fig S7). We found that the clones derived from randomly labelled animal pole cells at the 8-cell stage are randomly positioned with respect to the dorsoventral axis (Fig. 4B; Additional file 11: Fig S7). Moreover, labelling of ventral clones shows that the ectodermal tissue forming the pre-oral hood is located on the future ventral side before mouth formation, and analysis of the position of several clones shows that the apical most tissues in the early gastrula correspond to the anterior-most tissues at the bipinnaria stage (Fig. 4A; Additional file 11: Fig S7). This experiment suggests that there is no bias in the positioning of animal pole clones due to morphogenesis.
Taken together, these results show that the smallest cells of P. miniata embryos at the 16-cell stage are more likely to arise in the animal side of the embryo, where the future ventral side will be specified. However, the association between the position of the smallest cell and the DV axis is broad, with clones deriving from the smallest cells found across the entire ventral half on the 3 dpf larvae, instead of exclusively at the site of mouth opening.
Next, we sought to understand if and how smaller cells may influence DV axis formation, as cell size asymmetries are often linked to cell differentiation due to asymmetric partitioning of both maternal determinants and cytoplasmic volumes during cell division (reviewed in [27]).
To test if asymmetric partitioning of maternal determinants is necessary to DV axis determination in P. miniata, we performed early embryo dissociations. We isolated blastomeres at the 2, 4 and 8-cell stages and raised them to the bipinnaria stage (Fig. 5; Additional file 12: Fig S8). As expected, control, whole embryos reached the bipinnaria stage at 72 hpf (Fig. 5A, G). Blastomeres isolated at the 2-cell stage formed half-sized larvae and reached the bipinnaria stage at 72 hpf in 79.6% of cases and by 120 hpf in 84.6% of cases (Fig. 5B, E, and G). Blastomeres isolated at the 4-cell stage formed smaller larvae and reached the bipinnaria stage at 72 hpf in 25% of cases and by 120 hpf in 86.9% of cases (Fig. 5B, E, and G). This suggests that in most cases all 4 blastomeres have the potential to form the DV axis.
To test if maternal determinants for DV axis formation are inherited by the animal cells at the 8-cell stage, we split embryos at the 8-cell stage into animal and vegetal quartets; we found that all the vegetal quartets developed into half-sized bipinnaria by 72 hpf (Fig. 5C, G), while animal quartets failed to form mesendoderm tissues and remained blastulae for 5 days (Fig. 5D, G). This suggests that the vegetal blastomeres retain the potency to establish a DV axis, even in the absence of animal cells. To test if all vegetal blastomeres are capable of establishing a DV axis, we isolated individual blastomeres at the 8-cell stage (Fig. 5F, F’). We found that 44.4% of the isolated blastomeres formed small gastrulae at 72 hpf (Fig. 5F) and 55.5% formed small blastulae (Fig. 5F’), indicating that isolated vegetal blastomeres can form mesendoderm tissues but fail to open a mouth by 72 hpf. However, 70.5% of those gastrulating mini-larvae reached the bipinnaria stage by 120 hpf, when the experiment was terminated (Fig. 5G; Additional file 12: Fig S8).
Taken together these results suggest that the vegetal portion of the embryo is necessary and sufficient for gut formation and establishment of the DV axis, although vegetal blastomeres isolated at the 8-cell stage establish a DV axis with considerable delay. Therefore, P. miniata and P. regularis (see Additional file 6: Fig S3) are similar to most other echinoderms analyzed so far in that differential allocation of maternal determinants is involved in the determination of the AP axis, but not necessary for the determination of the DV axis.
Zygotic cell fate determinants necessary for DV axis formation might accumulate in the small cells of P. miniata embryos and it is possible that cell fate determinant asymmetries are re-established after blastomeres dissociations. To test if zygotic determinants enriched in the small cells are necessary for DV axis formation, we used micropipette aspiration to remove one small cell from 16-cell stage P. miniata embryos (Fig. 5H–I). We found that perturbed embryos opened a mouth at 72 hpf, similar to CTRL non-perturbed embryos (Fig. 5H, I). Taken together these results suggest that enrichment of cell fate determinants in the small cells plays a negligible role in the determination of the DV axis in P. miniata.
Next, we sought to test if the relationship between smaller cells and DV axis formation in P. miniata may be due to cell size alone. If that were true, even after the smallest cells of an embryo were removed, either by aspiration or by dissociation, cell size asymmetries would still be in place, and the next smallest cell might instruct the position of the DV axis. To test this possibility, we artificially created a population of small cells on one side of the embryo, by removing cytoplasm from one of the animal blastomeres at the 8-cell stage by micropipette aspiration until that blastomere was the smallest in the embryo (Fig. 5J–M). We then scored the position of the clone formed by the progeny of that miniaturized blastomere at 72 hpf. We found that miniaturized clones were randomly positioned with respect to the DV axis (Fig. 5J, K), similar to CTRL embryos, in which one animal blastomere at the 8-cell stage was injected but not manipulated (Fig. 5L, M). Taken together these results indicate that manipulating cell size to introduce artificial cell size asymmetries is not sufficient to direct the positioning of the future DV axis in P. miniata embryos.