Summary
The early Xenopus laevis embryo is replete with dynamic spatial waves. One such wave, the cell division wave, emerges from the collective cell division timing of first tens and later hundreds of cells throughout the embryo. Here we show that cell division waves do not propagate between neighboring cells, and do not rely on cell-to-cell coupling to maintain their division timing. Instead, intrinsic variation in division period autonomously and gradually builds these striking patterns of cell division. Disrupting this pattern of division by placing embryos in a temperature gradient resulted in highly asynchronous entry to the mid-blastula transition and misexpression of the mesodermal marker Xbra. Remarkably, this gene expression defect is corrected during involution, resulting in delayed yet normal Xbra expression and viable embryos. This implies the existence of a previously unknown mechanism for normalizing mesodermal gene expression during involution.
Graphical abstract
Introduction
The frog Xenopus laevis must solve a common problem after fertilization: how does a single, large (1.2 mm diameter) cell become thousands of somatic-sized cells that are ready to perform gastrulation and form an adult animal? Diverse animal embryos have found similar solutions to this problem. Zebrafish, Drosophila, and some species of frog all undergo multiple rounds of extremely rapid cleavages following fertilization (Keller et al., 2008; Olivier et al., 2010; Tomer et al., 2012) (Movies S1-2). The first cell division arrives around 95 min post-fertilization (mpf) at 18°C, and regular rounds of cell divisions follow roughly every 35 min thereafter. At first, these divisions occur approximately synchronously in all cells in the embryo, but by division 5-6, a spatial wave of division is visible, with rounds of cell divisions progressing from one side of the embryo to the other (Boterenbrood et al., 1983). Zebrafish and Drosophila embryos likewise display waves of cell division (Tomer et al., 2012; Keller et al., 2008), suggesting that cell division waves may play a conserved role in early embryogenesis.
Cell division waves are not the only spatial waves in the early frog embryo (Ubbels et al., 1983). Within minutes after fertilization, a wave of intracellular calcium spreads from the sperm entry point across the egg (Figure 1), and it contributes to the block to polyspermy and to the coordinated resumption of the cell cycle (Fontanilla and Nuccitelli, 1998; Stricker, 1999; McIsaac et al. 2011; Gelens et al. 2015). Fifteen to twenty minutes after fertilization at 18°C (a common temperature for cultivating X. laevis), another wave follows the same path, but more slowly. This is the post-fertilization wave (Hara et al., 1977), which coincides with enlargement of the sperm aster (Geertje et al., 1983). Around 70 min after fertilization, a wave emanates from the animal pole, the top of the embryo when oriented with respect to gravity, and travels toward the vegetal pole, the bottom. This is called the first surface contraction wave (Hara et al.,1980; Rankin and Kirschner, 1996). This wave marks the entry of the embryo into mitosis 1, and it is thought to be generated by the interaction of a spherical trigger wave of Cdk1 activation with the cortical cytoskeleton (Chang and Ferrell, 2013). Trigger waves can effectively propagate over large distances, coordinating biological processes along the way (Gelens et al., 2014) and the mitotic wave of Cdk1 activation may serve to synchronize mitotic entry across the fertilized embryo. A second surface contraction wave follows about 10 min later; it also proceeds from the top of the embryo to the bottom, and is thought to be due to the interaction of a spherical trigger wave of Cdk1 inactivation with the cortical cytoskeleton (Hara et al., 1980; Rankin and Kirschner, 1996; Chang and Ferrell, 2013). Each of these waves has a suggested role in coordinating developmental processes.
Figure 1. Waves in Early Xenopus laevis Development.
The Xenopus laevis embryo undergoes multiple spatially-organized and dynamic events in its early development. SEP denotes the sperm entry point, and mpf is minutes post-fertilization. Adapted from Nieuwkoop and Faber, 1994.
The subsequent cell division waves are perhaps the easiest developmental waves to observe, yet little is known about their origin and biological function. Using time-lapse microscopy and the ability to perturb division timing with temperature, we sought to understand the roll of cell division waves in early embryogenesis. We found that these waves arise through intrinsic differences in cell division timing, surprisingly without an active coupling mechanism. Perturbing the cell division waves resulted in a transient defect in mesoderm induction, which was corrected during involution. This points to the existence of a previously-unknown mechanism that corrects problems due to desynchronization prior to gastrulation, thereby contributing to robust embryonic development.
Results
We first set out to quantitatively characterize normal cell division waves in the X. laevis embryo. To accomplish this, we observed fertilized embryos in the top view (with the animal pole up and vegetal pole down) using a dissecting microscope and time-lapse video microscopy. We scored individual cell divisions by eye, marking the centroid of the dividing parent cell at the time that the cleavage plane just began to clearly form. We also kept track of the lineages of dividing cells (Movie S3).
As previously shown (Satoh, 1977; Boterenbrood et al., 1983; Newport and Kirschner, 1982a), the first cell cycle is long (∼95 min), the subsequent 11 cycles are comparatively short (∼35 min), and cell divisions are relatively synchronous within each cycle (Figure 2A-D). Once many cells were formed, not every division could be scored because some cells were not on the surface of the embryo, and others were obscured from view; thus Figure 2 (as well as the subsequent figures) includes only the subset of divisions that could be scored. Division periods decreased through cell cycle six, then increased beginning at divison nine (Figure 2B). A similar trend has been reported in zebrafish (Olivier et al., 2010).
Figure 2. Cell Division Waves in Three Dimensions.
The first 12 divisions of the X. laevis embryo are regular in time and nearly-synchronous within a round of division. Deviations from synchrony take the form of a wave of divisions.
(A) Top view of X. laevis embryos at the 1-, 2-, 4-, and 8-cell stage, 18°C.
(B) Cell cycle periods in the top view as a function of time at 18°C. The first cell cycle is much longer and was omitted for clarity. Cell cycle periods shortened slightly through division six, then began lengthening around division 9 or 10, followed by an increase in period at divisions 11 and 12. Error bands are ± one standard deviation.
(C) Top-down view of cell division waves at 18°C. Cell division waves originated opposite the sperm entry point (SEP) and terminated near the SEP. Color scale denotes the timing of cell division with cooler colors being earlier divisions. Contours are at two minute intervals.
(D) Top view of cell division as a function of position and time at 18°C. Lines were fit to rounds of division to illustrate their progression across the surface of the embryo in spatial waves. Centroids of parent cells were projected on to a line that runs along the direction of the division wave and plotted on the y axis.
(E) Side view of X. laevis embryos at the 1-, 2-, 4-, and 8-cell stage.
(F) Cell cycle periods in the side view as a function of time at 18°C. The second division is difficult to accurately score in the side view and is omitted along with the first for clarity. Trend of cell cycle periods is similar to the top view in B, but with more variation. Error bands are ±one standard deviation.
(G) Side view of cell division waves at 18°C. Color scale and contours as in (C).
(H) Side view of cell division as a function of position and time at 18°C. Centroids of parent cells are projected on to a line that runs along the animal-vegetal axis.
(I) The direction of the first cleavage plane correlates with the direction of the post-fertilization wave.
(J) Cell division waves anti-correlate with the direction of the post-fertilization wave. The post-fertilization wave begins near the sperm entry point and progresses away from it. The cell division wave begins opposite the SEP and progresses toward it. A total of 47 embryos were analyzed in panel I and 92 in panel J.
To visualize cell division waves, we needed spatial information about divisions. We plotted the onset of cell divisions as a function of time and cell position. By the time of the fifth cell cycle, waves of cell division were consistently observed. They progressed across the top surface of the embryo toward the sperm entry point (SEP), which usually corresponds to the dorsal-ventral axis (Figure 2C). To visualize several rounds of cell division waves at once, we found it was useful to condense the two-dimensional spatial information into one dimension, and plot it against time (Figure 2D). To accomplish this, we projected cell centroids onto a line in the direction of the cell division wave (Movie S3). The wave of cell divisions constituted roughly 10 min out of a 35 min cell cycle period at 18°C (Figure 2C and 2D). This corresponded to an apparent speed of ∼2 μm/s.
To characterize the cell division wave in the animal-vegetal direction, which is obscured in the top view, we used optical-quality mirrors mounted on a 45° bias to horizontal in order to observe fertilized embryos from the side (Figure 2E). We found the same trend of decreasing followed by increasing cell cycle period in the side view, but the cell-to-cell period heterogeneity was much greater than in the top view (Figure 2B and 2F). It was difficult to accurately score the first division in this view, so we omit both the first and second periods in Figure 2F. The side view allowed us to observe an orthogonal component of the cell division wave (Movie S2), which again typically became clearly visible around division 5 (Figure 2G). Cells near the bottom, unpigmented vegetal pole divided later than cells near the top, pigmented animal pole, and divisions ran in a smooth wave from top to bottom (Figure 2G and 2H). The cell division wave progressed more slowly along the animal-vegetal axis than along the dorsal-ventral axis, taking up 30 min out of a 35 min cell cycle at 18°C. This corresponded to an apparent speed of 0.7 μm/s.
It has been reported that the first cleavage plane strongly associates with the gray crescent, a feature that in turn forms with respect to the SEP and the post-fertilization wave (Klein, 1987). We verified this, finding a strong correlation between the first cleavage plane axis and the axis of the post-fertilization wave (Figure 2I).
Again using the origin of the post-fertilization wave as a proxy for SEP location, we found that the cell division wave almost always began on the side opposite the sperm entry point, typically the dorsal side, and terminated near the sperm entry point, typically the ventral side (Figure 2C and 2J). Thus, the post-fertilization wave and the cell division wave travel in opposite directions. This is the reverse of the cell division wave direction described in one early report (Boterenbrood et al., 1983), but it agrees with the findings of Satoh, 1977, and it was a consistent finding (Figure 2J).
Does the Cell Division Wave Propagate via Cell-Cell Coupling?
One possible mechanism for the cell division waves is suggested by the discovery that mitosis can propagate through cytoplasm via trigger waves at a speed of ∼1 μm/s (Chang and Ferrell, 2013). This mitotic trigger wave propagates from the animal pole to the vegetal pole (Chang and Ferrell, 2013; Perez-Mongiovi et al., 1998), similar to the predominantly top-to-bottom direction of the cell division wave. If a mechanism exists for transmitting trigger waves between cells in an embryo, then trigger waves could potentially account for the observed cell division waves.
We therefore set out to test whether cell cycles are coupled in a multicellular embryo. To this end, we elected to transiently delay the division timing of one cell at the two-cell stage. If cell cycles are coupled, the descendants of the delayed and undelayed cells should come back toward synchrony after the delaying perturbation is terminated.
We assessed a number of possible means for creating a delay, and the most suitable proved to be transient application of a temperature gradient. Gradients of temperature have been used in frogs (Huxley, 1927; Black, 1989) and other organisms (Gilchrist, 1928; Niemuth and Wolf, 1995; Jiang et al., 2000; Lucchetta et al., 2005) to preferentially speed up and slow down cell cycles and other developmental phenomena. We reasoned that a temperature gradient would supply the strong but transient change to cell cycle period required to produce a division timing delay.
To this end we built a device that controls the temperature of two sides of a chamber that is just wide enough to hold a row of X. laevis embryos (Figure 3A). Two Peltier-effect heat pumps allowed for the addition or removal of heat from each side of the chamber as necessary to rapidly control a temperature gradient across embryos.
Figure 3. A Temperature Gradient Reveals Lack of Coupling in Cell Divisions.
(A) Temperature gradient device.
(B) Desynchronizing cell divisions. Embryo was initially maintained at 23°C, then a temperature gradient of 11°C - 25°C was applied during the time marked by the horizontal red bar, from 70 mpf to 107 mpf. The second round of cell divisions was desynchronized as a result, with the two divisions occurring approximately 15 min apart. Temperature was then uniformly set to 18°C for the remainder of time. Subsequent divisions were labeled red (descendants of the warmed cell) and blue (descendants of the cooled cell). Vertical lines indicate average division time of each group, and gray regions indicate the difference in average division time between the groups.
(C) Cell divisions in a mock-treated control embryo at 18°C.
(D) The difference in average division time between descendants of cooled cells and descendants of warmed cells. These values correspond to the width of gray regions in panel B. The error band is ± 1 standard deviation of the timing difference.
(E) Average periods in lineages descended from the cooled cell (blue) and warmed cell (red) at the two-cell stage. Error bards are ± 1 standard deviation of the period.
(F) Comparison of early and late division timing. Average timing differences measured at divisions two and three were compared with average timing differences measured at divisions seven and eight. For gradient embryos, timing differences were between descendants of warmed and cooled cells, and the temperature gradient was applied with different durations, ranging from 37 to 57 min (always starting at 70 mpf). For control embryos, timing differences were between descendants of the two-cell stage. Red point is the embryo in panels B, D, and E.
To test the cell-to-cell coupling hypothesis, we applied an 11°C to 25°C temperature gradient to embryos from the one-cell stage to the two-cell stage. A 37 min application of this temperature gradient created a ∼15 min timing difference between the divisions of the embryo's two cells (Movie S4 and Figure 3B). As a control, embryos were placed in the gradient chamber but maintained at a constant temperature of 18°C (Figure 3C).
We then followed the descendants of the transiently warmed cell and of the transiently cooled cell through the next 10 cell divisions. As indicated by the gray regions in Figure 3B, the time between the divisions of the two lineages did not decrease after the temperature gradient was removed. Instead, the average difference in division timing increased slightly as periods increased in the later divisions (Figure 3D). Rather than observing shorter “catching up” periods in the descendants of the cooled cell or longer periods in the descendants of the warmed cell, we saw that average periods in the two lineages were indistinguishable (Figure 3E). Thus, there is no evidence for cell-to-cell coupling; once two cells are out of phase, they remain out of phase indefinitely.
Next we explored the behavior of embryos that experienced different timing delays, to see whether some particular phase differences might be more amenable to resynchronization. As shown in Figure 3F, this was not the case; no matter the initial phase difference, the phase difference was maintained over time and not corrected toward synchrony. Similar results were found when applying the temperature gradient later (after several cell divisions) and/or for longer durations.
Therefore we have found no evidence for cell-to-cell coupling in creating or maintaining cell division waves. This suggests that cell division waves are instead generated by a cell-autonomous mechanism.
A Simple Model of Cytoplasmic Pre-Patterning Can Account for Cell Division Waves
Cell cycle periods in the vegetal half of the embryo were consistently longer than periods in the animal half (Figure 4A), which could be due to differences in yolk content, differences in cyclin mRNA localization (Bowes et al., 2010), or a combination of factors. The period difference was first observable at division four, the first division after the animal and vegetal cytoplasm are separated by cleavage, and was similar in subsequent divisions (Figure 4A). This observation suggested a simple model of autonomous cell division wave generation. In this model, constant differences in cell cycle period that are intrinsic to different parts of the fertilized egg's cytoplasm cause cell division waves to gradually build.
Figure 4. A Simple Model Accounts for Cell Division Waves.
(A) Cell cycle periods for cells in the animal half (blue) and vegetal half (red) of nine unperturbed embryos at 18°C. Black bars are medians, and the shaded boxes indicate 25th and 75th percentiles.
(B) The first six simulated divisions of a cubic space representing the embryo.
(C-F) C and E are simulated cell division waves in the side view (C) and top view (E). D and F are experimentally measured cell division waves at 18°C for comparison.
See also Figure S1.
To test whether these period differences alone could give rise to the observed cell division waves, we created a model of the cleaving embryo, making simple assumptions. First, we assumed that period varied linearly throughout the embryo along both the animal-vegetal axis and the dorsal-ventral axis. Then we assumed that cells respond to the local period at their midpoint. The alternate assumption, that cells respond to an average period over their volume, yielded similar results. Finally, we let the model segment a cubic volume similarly to how cells in the X. laevis embryo divide (Figure 4B), and calculated the period in each of the daughter cells. The cubic geometry is of course an approximation, but was sufficient to capture the essence of the partitioning process. The model produced waves of divisions that appeared remarkably similar to experimentally measured divisions in the top view and side view (Figures 4C and 4F). The best match to experimentally observed cell division waves was obtained by assuming that the period intrinsic to the fastest part of the embryo was 18% shorter than the period in the slowest part (Figure S1A). This corresponded to a predicted period difference of three minutes at the hemisphere midpoints, similar to measured period differences for cycles 4, 5, and 6 (Figures S1B and S1C). Therefore the development of cell division waves could occur through the gradual and steady accumulation of timing differences pre-patterned in the fertilized egg.
Developmental Consequences of Asynchronous MBT Entry
The lack of evidence for coupling in the cell division wave led us to wonder whether the timing of early embryonic cell divisions is important for subsequent developmental events. As mesoderm is specified during these early rounds of cell division, we tested whether the normal near-synchronous divisions were necessary for proper mesoderm induction. To this end, we subjected embryos to a sustained 11°C-25°C temperature gradient orthogonal to the animal-vegetal axis (side-to-side) or along the animal-vegetal axis (top-to-bottom), from the one-cell stage to the time just before the first cells entered the midblastula transition (7-8 hours post-fertilization (hpf), Nieuwkoop and Faber (NF) stage 9). This resulted in embryos with drastic timing differences between the warmed and cooled sides and drastically different sizes of cells: large cells that had completed fewer divisions on the cooled side and small cells that had completed more divisions on the warmed side (Figure 5A, Movies S5-S10).
Figure 5. A Side-to-Side Temperature Gradient Leads to Asynchronous MBT Entry.
(A) Snapshots of embryo at 23°C at three different time points after it has experienced a side-to-side temperature gradient (11°C-25°C) from 1:10 hpf – 5:45 hpf. Two regions of equal size have been selected on the previously cold and previously warm side.
(B) Number of visible cells in the previously cold region (blue) and the previously warm region (red) as indicated in (A). The number of visible cells have been calculated by taking the initial number of cells and then increasing it by one every time a cell division is observed. The black lines show a smoothed fit using the “robust LOESS” (quadratic fit) option in Matlab.
(C)The average division rate of the visible cells as calculated from the fitted curves in (B). The time of maximal average division rate is taken as a measure for MBT onset. MBT is found to occur approx. 60 min later in the previously cold region than in the previously warm region (with a standard deviation of 16 min - 3 analyzed embryos).
In order to estimate the timing of MBT in embryos that were held at fixed temperature but previously experienced a side-to-side gradient (Figure 5), we kept track of cell divisions in a selected region on the previously hot side (smaller cells) and the previously cold side (larger cells). We then used this to determine the number of visible cells in those regions (Figure 5B), and analyzed the corresponding average rate of cell division in each region. This division rate initially increased as more dividing cells accumulate inside the region, but then started to decrease as cells entered MBT, which was accompanied by much slower cell divisions (Figure 5C). Moreover, the onset of MBT was also characterized by an increased motility of the cells (Movie S6). This analysis shows that cells on the previously cold side entered MBT approx. 60 min (standard deviation: 16 min, n = 3) later than cells on the previously hot side, a time difference that corresponds to about two full cycles. Figure 5 thus argues that all cells that were slowed down due to the cooling continue to divide until they have finished the same number of divisions as the fast, warmed cells had.
After the onset of MBT, mesoderm induction can be evaluated by in situ hybridization for the mesodermal marker Xbra. In control embryos, Xbra expression was absent at 8 hpf, but a faint symmetrical ring of Xbra expression could be seen as soon as 8:15 hpf (Figure S2A). This ring of Xbra expression became stronger and thicker in the following hours (measured at 8:30 hpf, 9:30 hpf, 10 hpf, 12 hpf), but always maintained its characteristic symmetric ring shape (Figure 6A and Figure S2A and S2B). In contrast, in the embryos desynchronized with a side-to-side temperature gradient (Figure S2C, Movie S5), Xbra was expressed in a highly asymmetrical arc (Figure 6A and Figure S2B and S2E). The arc coincided spatially with the concentration of pigmented bottle cells and the slight lip that marks the blastopore at the beginning of involution (Hardin and Keller, 1988; Black, 1989; Nieuwkoop and Faber, 1994; Keller, 2005). Pigmented bottle cells, in turn, appear first on the previously-heated side of temperature gradient-treated embryos (Movie S7), linking the orientation of the temperature gradient with the orientation of the arc of Xbra expression. These findings suggest that disrupting the endogenous early cell division timing using a temperature gradient, which leads to an asynchronous entry into MBT, affects the organization of the mesoderm during gastrulation.
Figure 6. A Long Temperature Gradient Induces a Mesodermal Induction Defect and Reveals a Resynchonizing Mechanism.
(A) Xbra expression after MBT. Time course of Xbra expression in gradient embryos treated in the side-to-side and top-to-bottom directions. Unstained and mock-treated control embryos are shown for comparison.
(B-C) Phenotype and survival of embryos two weeks after gradient treatment. “Alive” includes embryos that survived with a generally normal phenotype, and “abnormal” includes embryos that did not survive and embryos that experienced clear developmental defects such as dorsalization, ventralization, and bent body axes.
After observing this effect of a side-to-side temperature gradient, we wondered whether a temperature gradient in a different direction might produce different effects. We rotated the gradient device 90° and applied a temperature gradient along the animal-vegetal axis (the top-to-bottom direction) of embryos from the first cell cycle to just before MBT. 45-degree mirrors allowed us to visualize embryonic development during this top-to-bottom gradient. We warmed the vegetal pole and chilled the animal pole in order to reverse the normal animal-vegetal component of the cell division wave (Figure S2D, Movie S8). As shown in Fig. 6A, Xbra was expressed normally and symmetrically in embryos that experienced a top-to-bottom temperature gradient (see also Figure S2B). Therefore, the observed asymmetric Xbra pattern in side-to-side gradient experiments was not due simply to the presence or creation of the temperature gradient itself. Instead, this Xbra pattern resulted from the application of a temperature gradient specifically along axes orthogonal to the animal-vegetal axis.
Consequences of Asynchrony for Embryonic Survival
Finally, we asked whether embryos would survive to develop into normal tadpoles following highly asynchronous MBT entry and mesoderm induction. We performed the side-to-side and top-to-bottom gradient treatments as previously described, then removed embryos from the gradient chamber and observed their development for two weeks. Remarkably, despite the Xbra misexpression that resulted from the side-to-side gradient, we found no significant difference between the rate of survival with generally normal phenotype for embryos that experienced a gradient and those that were simply placed in a gradient chamber at uniform 18°C for the same amount of time (Movies S9-10 and Figures 6B and 6C). These results are consistent with the findings of Black (1989), and Huxley (1927), where it was reported that temperature gradients can change the orientation of the dorsal lip, but do not reset the dorsal-ventral axis.
To determine how the gradient-treated embryos were able to recover from their aberrant Xbra expression pattern, we repeated the gradient treatment and examined Xbra staining at times later in development to see what became of the misexpressed pattern. By 12 hpf, gradient-treated embryos expressed a symmetric ring of Xbra around their vegetal pole, just like control embryos at 10 hpf. Following this time point, gradient-treated embryos seemed to display normal but delayed Xbra expression, performing normal neurulation and expressing Xbra in a characteristic shape (Figure 6A). Strikingly, the Xbra expression defect in embryos that entered MBT asynchronously was corrected by 12 hpf, NF stage 10.5. This recovery of a normal Xbra expression pattern after a side-to-side temperature gradient suggests the existence of a corrective process that is carried out at approximately the time of involution. The corrective process is able to take thousands of autonomously-dividing cells and coordinate them to express a symmetric ring of Xbra.
Discussion
Newport and Kirschner (1982a) showed that X. laevis embryonic cells continue to divide with near-normal periods when dissociated from their embryonic context. This implied that each cell possesses an independent and accurate cell cycle clock, yet the result left open the possibility that in the intact embryo, cells are coupled to actively maintain near-synchrony of cell division timing in the face of perturbation.
Here we found that X. laevis embryos do not rely on cell-to-cell coupling to maintain cell division timing. Transiently desynchronized embryos fail to become more synchronous, and instead remain desynchronized, dividing in an autonomous and apparently uncoupled manner. Thus we ruled out mechanisms for producing a cell division wave that rely on a trigger wave or intercellular coupling.
The cell cycle period is shortest in the animal half and longest in the vegetal half. This period difference could be accounted for by the heterogeneous distribution of cyclin B1 and cyclin B2 mRNA, with greatest concentration in the animal hemisphere (Bowes et al., 2010). Similarly, the yolk material is strongly concentrated in the vegetal hemisphere. Additionally, it has been shown that animal yolk material (and perhaps period determinants) are also rearranged during the first cell cycle as a result of the growing sperm aster (Ubbels et al., 1983, Brown et al., 1993, Kikkawa et al., 1996). Likewise, cell division waves begin near the animal pole and terminate near the vegetal pole. An orthogonal component of division waves travels from opposite the sperm entry point toward the sperm entry point. These waves develop gradually over multiple cell cycles into smooth patterns of division, and a model that assumes autonomous divisions and linearly-varying cell cycle periods recapitulates the waves and their evolution.
Despite the lack of active coordination in early embryonic cell division timing, normal near-synchrony is required for the proper initiation of mesoderm induction, as read out by the transcription factor Xbra. Similarly, the timing of MBT was found to be strongly asynchronous, which points to a relationship between cell size and the onset of MBT and cell differentiation, as also suggested before (Newport and Kirschner 1982b; Farrell and O'Farrell 2014; Amodeo et al., 2015). However, near-synchrony along the animal-vegetal axis is not required for proper Xbra patterning, which suggests that the observed mesodermal patterning defect is only caused by the developmental delay induced by the side-to-side temperature gradient.
Remarkably, asynchrony-induced Xbra misexpression was corrected during involution, resulting in a delayed but otherwise normal, spatially symmetric Xbra pattern. Embryos recovered from asynchrony with generally normal development and normal survival rates. Perhaps this secondary resynchronization during involution explains why unlike other waves in the early embryo, the cell division wave lacks active regulation.
Thus the X. laevis embryo displays extreme robustness to environmental perturbation. It uses simple mechanisms to prepare a fertilized zygote for the complex rearrangements of gastrulation. Drosophila and the wasp Pimpla turionellae also possess robustness to temperature gradient perturbation, developing normally after desynchronization of embryonic cell lineages (Lucchetta et al., 2005; Niemuth and Wolf, 1995).
Why are embryos built to survive perturbations to division timing they would likely never encounter in nature? Are these perturbations simply more extreme versions of the cell division asynchrony embryos are already selected to tolerate? (Cells along the animal-vegetal axis of an unperturbed X. laevis embryo enter MBT over a window of 30 min at 18°C, for example.) Or is robustness to asynchrony an exaptation – a beneficial trait that evolved in response to an unrelated need, perhaps the need to globally coordinate cell lineages during involution?
Although we have identified the approximate moment when resynchronization occurs, the mechanism remains elusive. One intriguing idea is that after MBT there exists some unknown checkpoint that pauses differentiation and further development until all cells have reached the same required status. Such a checkpoint that ensures symmetrical morphogenesis might have had a high priority in fitness during evolution. Making precise measurement of cell cycle periods after MBT but before involution, paying special attention to abnormally short or long periods in gradient-treated embryos, could help to answer this question.
The rounds of cell cleavage following fertilization of the X. laevis embryo have surprised us with their minimal complexity. Because this period of development prepares the embryo for the complex rearrangements of gastrulation, we initially expected the timing of cell division waves to be highly regulated. We have instead come to expect elegant robustness from the early embryo.
Experimental Procedures
Embryos
Xenopus laevis adult females were primed with 67 IU of pregnant mare serum gonadotropin at least 3 days prior to induction, and induced with 500 IU of human chorionic gonadotropin 16 hours prior to ovulation. Eggs were squeezed from ovulating females and fertilized using dissected testes, then dejellied by gently swirling a dish of embryos in 2% cysteine, pH 7.8. After 3-4 min in cysteine, fertilized embryos were washed ∼10× in 0.1× MMR (Murray, 1991). Development was observed in 0.1× MMR.
Time-Lapse Microscopy
Throughout this study, we collected data by time-lapse microscopy on a dissection microscope with a 0.5× objective and a variable zoom. An attached camera recorded images for later analysis. We used frame rates of once every 10 s and once every 30 s.
Mirrors
Fertilized and dejellied X. laevis embryos orient with respect to gravity, such that one can image only their gravity-up animal pole with a standard dissecting microscope. To examine the sides and the gravity-down vegetal pole of the embryo, we made use of an optical-quality mirror manufactured to sit at 45 degrees from horizontal, the half-cube 4.2mm mirror from Edmund Optics. These mirrors were placed near an embryo, and reflected light 90 degrees upward to a dissecting microscope. Placing an embryo between two such mirrors allowed us to view two sides of a single embryo. With this setup, we were able to record a majority of embryonic surface divisions.
Mirror Imaging Chamber
We used the large temperature-controlled chamber described in Gelens et al. (2015) to observe unperturbed embryos from the top-down and in the side view using 45-degree mirrors.
Timing of divisions within an embryo
Because X. laevis embryos are opaque, we could not use fluorescence-based markers to visualize cell divisions. Instead, we scored bright field movies for the timing and position of cell divisions by eye. To assist in the task, we developed custom software using MATLAB to record scored division information. Our software allows replay of a movie frame by frame so the user can carefully identify divisions. We scored division time as the first frame during which any part of the dark cleavage plane is clearly visible, and we scored division location as the centroid of the visible portion of a dividing parent cell. Both time and position information are recorded with a click, then the user scrolls backward in time to find the parent of the scored cell. In this way, the user works backward division-by-division to either the first frame of the movie, or to a previously-scored parent. The software then adds the recently-scored lineage to a tree of cell relationships within the embryo.
Temperature Gradient Device
The temperature gradient device is composed of a chamber, temperature-controlling machinery, and aluminum heat sinks. The chamber is formed by a Lexan insert sandwiched between two aluminum blocks. Each block contains an internal thermistor to monitor its temperature. A Peltier cooler/heater is glued to the outside edge of each aluminum block with thermally conductive glue, and aluminum heat sinks attach to the Peltier devices. The thermistors and Peltier devices are linked to a digital temperature controller, which allows us to precisely (within 0.1°C) and independently control the temperature of each aluminum block. A row of embryos rests on the Lexan insert between the two aluminum blocks, and is covered from above with a glass coverslip sealed with a mixture of equal masses melted petroleum jelly:lanolin:paraffin wax (VLP). For side-to-side gradients in the direction orthogonal to the animal-vegetal axis, the chamber is filled with 0.5% low-melting point agarose in 0.1× MMR, to prevent convection from influencing the development of a linear temperature gradient. In top-to-bottom gradients along the animal-vegetal axis, the chamber was filled with 0.1× MMR and no agarose to allow embryo reorientation, but the gradient device still produced comparable timing differences.
To prepare embryos for a gradient experiment, we fertilized them, removed their jelly coats (Murray, 1991), and placed them in the gradient device's chamber along with 0.1× MMR. If adding agarose, we removed the MMR to a level just above the embryos, then added 0.5% low-melting point agarose dissolved in 0.1× MMR, previously held at 36°C, to fill the chamber. We maintained embryos at a uniform and constant 23°C while filling the chamber with warm agarose.
The Short Gradient Experiment
Next we needed to pick appropriate times to apply the temperature gradient with the goal of creating a large difference in timing between adjacent cells while still leaving plenty of time remaining to observe divisions after gradient termination. To satisfy these requirements, we applied temperature gradients at the one- and two-cell stages, terminating in time to observe the third division onward at uniform temperature.
Fertilization and dejellying were carried out at room temperature. Accordingly, embryos were maintained at 23°C before beginning the gradient. During the gradient, we used a variety of temperatures that spanned a range including the highest and lowest temperatures that embryos could tolerate, 11°C - 25°C. Gradients were applied before the first division, typically at 70 mpf, and were terminated before the second division was complete, typically at 1:47 hpf. The gradient device required about a minute to come to a new temperature after being set. For example, it took 73 s to chill one block to 9°C while maintaining the other at 23°C.
After gradient termination, there is a challenge in temperature selection. Too cold, and the remaining divisions take days to complete. Too hot, and experimental and scoring noise obscures division timing. We compromised and brought embryos to a uniform 18°C after terminating the gradient.
The Long Gradient Experiment
For long gradient experiments, we prepared embryos identically to the short gradient, then simply waited to terminate the gradient until division number 10. In these experiments, we removed embryos from the chamber immediately after gradient termination and observed their development in a dish filled with 0.1× MMR, then either fixed at time points for in-situ staining or transferred to 24-well plates filled with 0.1× MMR for long-term survival and phenotype observation.
Because of the gravity-orientation of fertilized X. laevis embryos, our gradient device typically applies a temperature gradient along an axis orthogonal to gravity. To apply the gradient along the axis of gravity, we simply sealed fertilized dejellied embryos in the gradient chamber, without removing MMR or adding agarose, then turned the gradient device 90° so that one aluminum block faced down and the other faced up. The embryos reoriented with respect to gravity, so their bottom vegetal pole was in contact with one aluminum block, and their top animal pole was in contact with the other. We then placed 45° mirrors outside the chamber, such that they directed light emanating horizontally from the chamber upwards into the microscope.
RNA In-Situ Hybridization
Staining was performed as in (Harland, 1991).
Regulatory Approval
All experiments on live vertebrates were performed in accordance with relevant institutional and national guidelines and regulations; the Stanford University Administrative Panel on Laboratory and Animal Care approved the experiments.
Supplementary Material
Movie S1. Cell Division Wave Viewed from the Top, Related to Figure 2, A wave of cell divisions emerges after roughly six divisions and travels from one side of the embryo to the other. 20 min real time = 1 s movie time.
Movie S2. Cell Division Wave Viewed from the Side, Related to Figure 2, 45-degree mirrors provide a side view of the embryo in Movie S1, which makes visible a striking wave of cell divisions, traveling from the pigmented animal pole on the top to the unpigmented vegetal pole on the bottom. 20 min real time = 1 s movie time.
Movie S3. Scoring Cell Divisions, Related to Figure 2, X. laevis cell divisions filmed by time-lapse microscopy then scored by eye. Centroids of dividing parent cells, time of division, and a tree of cell relationships were recorded. Two-dimensional centroids were projected onto one dimension, here the direction of the eventual wave of cell divisions.
Movie S4. Side-to-Side Gradient, Related to Figure 3, Top view of embryos, side-to-side temperature gradient, 25°C at the bottom to 11°C at the top. Gradient length is 37 minutes beginning at 70 mpf, late in the first cell cycle. 20 min real time = 1 s movie time.
Movie S5. Side-to-Side Long Gradient, Related to Figure 5, Top view of embryos, side-to-side long temperature gradient, 25°C at the bottom to 11°C at the top. Gradient begins at 2:00 hpf, between 2nd and 3rd division. 20 min real time = 1 s movie time.
Movie S6. Post-MBT Cell Divisions Following a Side-to-Side Long Gradient, Related to Figure 5, Top view of a side-to-side long gradient-treated embryo. Embryo is oriented with large cells, which correspond to the previously-cooled side, to the top, and small cells, which correspond to the previously-heated side, to the bottom. Movie starts 6:45 hpf. 40 min real time = 1 s movie time.
Movie S7. Dorsal Lip Formation after Side-to-Side Long Gradient, Related to Figure 6, Top and side views of a single side-to-side long gradient-treated embryo. Embryo is oriented with small cells, which correspond to the previously-cooled side, to the top, and large cells, which correspond to the previously-heated side, to the bottom. Dorsal lip formation is visible in the left side view. 80 min real time = 1 s movie time.
Movie S8. Top-to-Bottom Long Gradient, Related to Figure 6, Side view of embryos, top-to-bottom long temperature gradient. Unpigmented vegetal side (bottom) heated to 24°C, pigmented animal side (top) chilled to 18°C. Gradient on at 1:10 hours post fertilization, just before the first division. 20 min real time = 1 s movie time.
Movie S9. Side-to-Side Long Gradient Recovery, Related to Figure 6, Top view of embryos, side-to-side long gradient-treated embryos and controls are mixed. Gradient-treated embryos have uneven cell sizes. 80 min real time = 1 s movie time.
Movie S10. Top-to-bottom Long Gradient Recovery, Related to Figure 6, Top view of embryos, top-to-bottom long gradient-treated embryos and controls are mixed. Gradient-treated embryos have larger cell sizes than control embryos. 80 min real time = 1 s movie time.
Acknowledgments
We would like to thank Jessica Chang for in-situ instruction and sharing her bench, Tek Hyung Lee, Sarah Trosin and Julia Kamenz for comments, and all the members of the Baker and Ferrell Labs for comments, advice, and being wonderful people to work with. L.G. thanks Guy Verschaffelt and Lars Keuninckx of the Applied Physics Group for their help with the initial design of the gradient embryo chamber. This work was supported in part by grants from the National Institutes of Health (R01 GM110564 and P50 GM107615 to J.F., and R01 GM103787 and R01 HD076839 to J.B.), the National Science Foundation Graduate Research Fellowship Program to G.A., and the Research Foundation-Flanders (FWO-Vlaanderen), the Belgian American Educational Foundation (BAEF), the KU Leuven Junior Mobility Programme (JuMO), and the Research Council of the Vrije Universiteit Brussel to L.G.
Footnotes
Author Contributions: G.A., L.G., J.B, and J.F. designed the studies; G.A. and L.G. carried out the experiments and the modeling; and G.A., L.G., J.B, and J.F. wrote the paper.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- Amaya E, Stein PA, Musci TJ, Kirschner MW. FGF signaling in the early specification of mesoderm in Xenopus. Development. 1993;118:477–487. doi: 10.1242/dev.118.2.477. [DOI] [PubMed] [Google Scholar]
- Black SD. Experimental reversal of the normal dorsal-ventral timing of blastopore formation does not reverse axis polarity in Xenopus laevis embryos. Dev Biol. 1989;134:376–381. doi: 10.1016/0012-1606(89)90109-7. [DOI] [PubMed] [Google Scholar]
- Boterenbrood EC, Narraway JM, Hara K. Duration of cleavage cycles and asymmetry in the direction of cleavage waves prior to gastrulation in Xenopus laevis. Roux Arch Dev Biol. 1983;192:216–221. doi: 10.1007/BF00848652. [DOI] [PubMed] [Google Scholar]
- Bowes JB, Snyder KA, Segerdell E, Jarabek CJ, Azam K, Zorn AM, Vize PD. Xenbase: gene expression and improved integration. Nucleic Acids Res. 2010;38:D607–D612. doi: 10.1093/nar/gkp953. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Brown EE, Denegre JM, Danilchik MV. Deep cytoplasmic rearrangements in ventralized Xenopus embryos. Dev Biol. 1993;160:148–156. doi: 10.1006/dbio.1993.1293. [DOI] [PubMed] [Google Scholar]
- Chang JB, Ferrell JE., Jr Mitotic trigger waves and the spatial coordination of the Xenopus cell cycle. Nature. 2013;500:603–7. doi: 10.1038/nature12321. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fontanilla RA, Nuccitelli R. Characterization of the sperm-induced calcium wave in Xenopus eggs using confocal microscopy. Biophys J. 1998;75:2079–2087. doi: 10.1016/S0006-3495(98)77650-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Geertje UA, Hara K, Koster CH, Kirschner MW. Evidence for a functional role of the cytoskeleton in determination of the dorsoventral axis in Xenopus laevis eggs. J Embryol Exp Morph. 1983;77:15–37. [PubMed] [Google Scholar]
- Gelens L, Anderson GA, Ferrell JE., Jr Spatial trigger waves: positive feedback gets you a long way. Mol Biol Cell. 2014;25:3486–93. doi: 10.1091/mbc.E14-08-1306. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gelens L, Huang KC, Ferrell JE., Jr How does the Xenopus laevis embryonic cell cycle avoid spatial chaos? Cell Rep. 2015;12:892–900. doi: 10.1016/j.celrep.2015.06.070. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gilchrist FG. The effect of a horizontal temperature gradient on the development of the egg of a Urodele, Triturus torosus. Physiol Zool. 1928;1:231–268. [Google Scholar]
- Hara K, Tydeman P, Hengst RTM. Cinematographic observation of “Post-Fertilization Waves” (PFW) on the Zygote of Xenopus laevis. Roux Arch Dev Biol. 1977;181:189–192. doi: 10.1007/BF00848442. [DOI] [PubMed] [Google Scholar]
- Hara K, Tydeman P, Kirschner M. A cytoplasmic clock with the same period as the division cycle in Xenopus eggs. Proc Natl Acad Sci USA January. 1980;77:462–466. doi: 10.1073/pnas.77.1.462. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hardin J, Keller R. The behavior and function of bottle cells during gastrulation of Xenopus laevis. Development. 1988;103:211–230. doi: 10.1242/dev.103.1.211. [DOI] [PubMed] [Google Scholar]
- Harland RM. In situ hybridization: an improved whole-mount method for Xenopus embryos. Methods Cell Biol. 1991;36:685–695. doi: 10.1016/s0091-679x(08)60307-6. [DOI] [PubMed] [Google Scholar]
- Hartley RS, Rempel RE, Maller JL. In vivo regulation of the early embryonic ell cycle in Xenopus. Dev Biol. 1996;173:408–419. doi: 10.1006/dbio.1996.0036. [DOI] [PubMed] [Google Scholar]
- Huxley JS. The modification of development by means of temperature gradients. Roux Arch Dev Biol. 1927;112:480–516. doi: 10.1007/BF02253775. [DOI] [PubMed] [Google Scholar]
- Jiang YJ, Aerne BL, Smithers L, Haddon C, Ish-Horowicz D, Lewis J. Notch signaling and the synchronization of the somite segmentation clock. Nature. 2000;408:475–479. doi: 10.1038/35044091. [DOI] [PubMed] [Google Scholar]
- Katsumoto K, Arikawa T, Doi JY, Fujii H, Nishimatsu SI, Sakai M. Cytoplasmic and molecular reconstruction of Xenopus embryos: synergy of dorsalizing and endo-mesodermalizing determinants drives early axial patterning. Development. 2003;131:1135–1144. doi: 10.1242/dev.01015. [DOI] [PubMed] [Google Scholar]
- Keller RE. An experimental analysis of the role of bottle cells and the deep marginal zone in gastrulation of Xenopus laevis. J Exp Zool. 2005;216:81–101. doi: 10.1002/jez.1402160109. [DOI] [PubMed] [Google Scholar]
- Keller PJ, Schmidt AD, Wittbrodt J, Stelzer EHK. Reconstruction of zebrafish early embryonic development by scanned light sheet microscopy. Science. 2008;322:1065–1069. doi: 10.1126/science.1162493. [DOI] [PubMed] [Google Scholar]
- Kikkawa M, Takano K, Shingawa A. Location and behavior of dorsal determinants during first cell cycle in Xenopus eggs. Development. 1996;122:2687–3696. doi: 10.1242/dev.122.12.3687. [DOI] [PubMed] [Google Scholar]
- Klein SL. The first cleavage furrow demarcates the dorsal-ventral axis in Xenopus embryos. Dev Biol. 1987;120:299–304. doi: 10.1016/0012-1606(87)90127-8. [DOI] [PubMed] [Google Scholar]
- Lucchetta EM, Lee JH, Fu LA, Patel NH, Ismagilov RF. Dynamics of Drosophila embryonic patterning network perturbed in space and time using microfluidics. Nature. 2005;434:1134–1138. doi: 10.1038/nature03509. [DOI] [PMC free article] [PubMed] [Google Scholar]
- McIsaac RS, Huang KC, Segupta A, Wingreen NS. Does the potential for chaos constrain the embryonic cell-cycle oscillator? PLoS Comput Biol. 2011;7:e1002109. doi: 10.1371/journal.pcbi.1002109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Murray AW. Cell Cycle Extracts. Methods Cell Biol. 1991;36:581–605. [PubMed] [Google Scholar]
- Newport J, Kirschner M. A major developmental transition in early Xenopus embryos: I. characterization and timing of cellular changes at the midblastula stage. Cell. 1982a;30:675–86. doi: 10.1016/0092-8674(82)90272-0. [DOI] [PubMed] [Google Scholar]
- Newport J, Kirschner M. A major developmental transition in early Xenopus embryos: II. Control of the onset of transcription. Cell. 1982b;30:687–96. doi: 10.1016/0092-8674(82)90273-2. [DOI] [PubMed] [Google Scholar]
- Niemuth J, Wolf R. Developmental asynchrony caused by steep temperature gradients does not impair pattern formation in the wasp, Pimpla turionellae. Roux Arch Dev Biol. 1995;204:444–452. doi: 10.1007/BF00360852. [DOI] [PubMed] [Google Scholar]
- Nieuwkoop PD, Faber J. Normal table of Xenopus laevis (Daudin) Garland Publishing Inc; New York: 1994. [Google Scholar]
- Olivier N, Luengo-Oroz MA, Duloquin L, Faure E, Savy T, Veilleux I, Solinas X, Débarre D, Bourgine P, Santos A, Peyiéras N, Beaurepaire E. Cell lineage reconstruction of early zebrafish embryos using label-free nonlinear microscopy. Science. 2010;329:967–971. doi: 10.1126/science.1189428. [DOI] [PubMed] [Google Scholar]
- Perez-Mongiovi D, Chang P, Houliston E. A propagated wave of MPF activation accompanies surface contraction waves at first mitosis in Xenopus. J Cell Sci. 1998;111:385–393. doi: 10.1242/jcs.111.3.385. [DOI] [PubMed] [Google Scholar]
- Rankin S, Kirschner MW. The surface contraction waves of Xenopus eggs reflect the metachronous cell-cycle state of the cytoplasm. Curr Biol. 1997;7:451–454. doi: 10.1016/s0960-9822(06)00192-8. [DOI] [PubMed] [Google Scholar]
- Sakai M. The vegetal determinants required for the Spemann organizer move equatorially during the first cell cycle. Development. 1996;122:2207–2214. doi: 10.1242/dev.122.7.2207. [DOI] [PubMed] [Google Scholar]
- Satoh N. ‘Metachronous’ cleavage and initiation of gastrulation in amphibian embryos. Dev Growth Differ. 1977;19:111–117. doi: 10.1111/j.1440-169X.1977.00111.x. [DOI] [PubMed] [Google Scholar]
- Stricker SA. Comparative biology of calcium signaling during fertilization and egg activation in animals. Dev Biol. 1999;211:157–176. doi: 10.1006/dbio.1999.9340. [DOI] [PubMed] [Google Scholar]
- Tomer R, Khairy K, Amat F, Keller PJ. Quantitative high-speed imaging of entire developing embryos with simultaneous multiview light-sheet microscopy. Nat Methods. 2012;9:755–763. doi: 10.1038/nmeth.2062. [DOI] [PubMed] [Google Scholar]
- Tsai TYC, Theriot JA, Ferrell JE., Jr Changes in oscillatory dynamics in the cell cycle of early Xenopus laevis embryos. PLoS Biol. 2014;12:e1001788. doi: 10.1371/journal.pbio.1001788. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ubbels GA, Hara K, Koster CH, Kirschner MW. Evidence for a functional role of the cytoskeleton in determination of the dorsoventral axis in Xenopus laevis eggs. J Embryol Exp Morphol. 1983;77:15–37. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Movie S1. Cell Division Wave Viewed from the Top, Related to Figure 2, A wave of cell divisions emerges after roughly six divisions and travels from one side of the embryo to the other. 20 min real time = 1 s movie time.
Movie S2. Cell Division Wave Viewed from the Side, Related to Figure 2, 45-degree mirrors provide a side view of the embryo in Movie S1, which makes visible a striking wave of cell divisions, traveling from the pigmented animal pole on the top to the unpigmented vegetal pole on the bottom. 20 min real time = 1 s movie time.
Movie S3. Scoring Cell Divisions, Related to Figure 2, X. laevis cell divisions filmed by time-lapse microscopy then scored by eye. Centroids of dividing parent cells, time of division, and a tree of cell relationships were recorded. Two-dimensional centroids were projected onto one dimension, here the direction of the eventual wave of cell divisions.
Movie S4. Side-to-Side Gradient, Related to Figure 3, Top view of embryos, side-to-side temperature gradient, 25°C at the bottom to 11°C at the top. Gradient length is 37 minutes beginning at 70 mpf, late in the first cell cycle. 20 min real time = 1 s movie time.
Movie S5. Side-to-Side Long Gradient, Related to Figure 5, Top view of embryos, side-to-side long temperature gradient, 25°C at the bottom to 11°C at the top. Gradient begins at 2:00 hpf, between 2nd and 3rd division. 20 min real time = 1 s movie time.
Movie S6. Post-MBT Cell Divisions Following a Side-to-Side Long Gradient, Related to Figure 5, Top view of a side-to-side long gradient-treated embryo. Embryo is oriented with large cells, which correspond to the previously-cooled side, to the top, and small cells, which correspond to the previously-heated side, to the bottom. Movie starts 6:45 hpf. 40 min real time = 1 s movie time.
Movie S7. Dorsal Lip Formation after Side-to-Side Long Gradient, Related to Figure 6, Top and side views of a single side-to-side long gradient-treated embryo. Embryo is oriented with small cells, which correspond to the previously-cooled side, to the top, and large cells, which correspond to the previously-heated side, to the bottom. Dorsal lip formation is visible in the left side view. 80 min real time = 1 s movie time.
Movie S8. Top-to-Bottom Long Gradient, Related to Figure 6, Side view of embryos, top-to-bottom long temperature gradient. Unpigmented vegetal side (bottom) heated to 24°C, pigmented animal side (top) chilled to 18°C. Gradient on at 1:10 hours post fertilization, just before the first division. 20 min real time = 1 s movie time.
Movie S9. Side-to-Side Long Gradient Recovery, Related to Figure 6, Top view of embryos, side-to-side long gradient-treated embryos and controls are mixed. Gradient-treated embryos have uneven cell sizes. 80 min real time = 1 s movie time.
Movie S10. Top-to-bottom Long Gradient Recovery, Related to Figure 6, Top view of embryos, top-to-bottom long gradient-treated embryos and controls are mixed. Gradient-treated embryos have larger cell sizes than control embryos. 80 min real time = 1 s movie time.