Emergence of traveling waves in the zebrafish segmentation clock

The periodic formation of vertebrate somites is governed by the segmentation clock, a congregation of cells that exhibit oscillating expression of hairy-related transcriptional repressors (her1 and her7 in zebrafish) through a negative-feedback loop. In zebrafish, a somite is formed every 30 minutes, and the segmentation clock cyclically exhibits a dynamic spatial pattern of hairy gene expression in the presomitic mesoderm (PSM) with the same periodicity (Fig. 1A, left). Previous reports have demonstrated that the phases of cellular oscillation are highly coordinated in a spatial manner, which results in traveling waves that repeatedly bud off from the synchronized oscillation zone in the posterior PSM, sweep anteriorly, and stop around the future segmentation point (Fig. 1A, middle) (Gajewski et al., 2003; Henry et al., 2002; Holley et al., 2000; Holley et al., 2002; Horikawa et al., 2006; Jiang et al., 2000; Oates and Ho, 2002; Palmeirim et al., 1997; Pourquie, 2003; Saga and Takeda, 2001; Sawada et al., 2000). The traveling waves are the result of phase differences among the oscillating cells in a posterior-to-anterior direction; cells with nuclear dots, with cytoplasmic signals and with no signal are detected sequentially from anterior to posterior, corresponding to transcribing cells, translating cells and non-transcribing cells, respectively (Fig. 1A, right). This pattern of oscillation is maintained throughout segmentation and is referred to as the ‘mature’ pattern. Although many studies have addressed the mechanisms of the mature segmentation clock, very few investigations have focused on the initiation of the segmentation clock (Riedel-Kruse et al., 2007) and little is known about how the traveling wave emerges. In the present study, we investigate this early mechanism through in vivo and in silico experiments using zebrafish as a model system.

Whole-mount in situ hybridization (Nikaido et al., 1997) and fluorescent in situ hybridization (Horikawa et al., 2006) were performed as previously described.

Fgf8b-soaked bead transplantation and SU5402 treatment were performed as described (Sawada et al., 2001). Bead transplantation was performed at the 30% epiboly stage (~5.0 hpf, as indicated in Fig. S1A in the supplementary material). SU5402 (Calbiochem) treatment was performed at 0.2 mg/ml for 8 minutes (strong treatment in Fig. 2B) or 4 minutes (weak inhibition in Fig. 3A). Then, the embryos were incubated in Ringer's solution for 2 hours.

Embryos transplanted with Alexa488-injected cells were mounted in 1% agarose. Images were collected at 10-minute intervals with a MZFLIII fluorescence microscope (Leica). Cell tracking and quantification were performed with ImageJ (NIH).

Antisense morpholino oligonucleotides for her1/7 were designed as described previously (Henry et al., 2002). Morpholinos were injected at 0.25 mM into single-cell stage embryos.

The traveling wave in an anterior direction first appears at ~70% epiboly [7.8 hours post-fertilization (hpf)], long (~3 hours) before segmentation takes place (Fig. 1B-E; see Fig. S1A in the supplementary material) (Riedel-Kruse et al., 2007). Until this point, her1 oscillation is restricted to the blastoderm margin and is synchronized among cells in this area as reported previously (Riedel-Kruse et al., 2007). At the initiation of the traveling wave, phase delay in an anterior direction is observed, indicating that the traveling wave is created mainly by sequential initiation of her1 oscillation from posterior to anterior. During the synchronous oscillation in the margin, gastrulation movement towards the anterior occurs (see Fig. S1B in the supplementary material) (Warga and Kimmel, 1990), indicating that the marginal cells, which transiently expressed her1, attain a transcription-negative state when the cells exit the marginal area and form the hypoblast layer. This was further supported by a comparison between expression of the PSM marker papc (pcdh8 – Zebrafish Information Network) and her1 (see Fig. S1B in the supplementary material). her1/7 morpholino injection also revealed that the suppression of her1 in the hypoblast depends on its own negative feedback: the her1 expression domain was found to be broadened in morpholino-injected embryos, whereas it was restricted to the marginal area in control embryos (7.0 hpf) (Fig. 1F,G). We did not observe the horizontal waves from ventral to dorsal previously reported (Riedel-Kruse et al., 2007), although we performed precise time-course analysis of her1 expression using embryos of strictly controlled stages (see Fig. S1A in the supplementary material). In the following experiments we focus on the mechanism underlying the emergence of the first traveling wave.

Intercellular communication through the Notch pathway is dispensable for the initiation of traveling waves as revealed both by Notch mutants (Jiang et al., 2000) and inhibitor experiments (see Fig. S2A-C in the supplementary material). Furthermore, other types of communication might not be required as our preliminary data suggest that wild-type cells initiate her1 expression when they are surrounded by non-oscillating mutant cells (see Fig. S2D,E in the supplementary material). It is also unlikely that the anterior movement of the mesoderm plays any role, as it would result only in a phase advance towards the anterior and not in a phase delay. These facts led us to speculate that the traveling wave is generated by an extracellular factor that triggers her1 oscillation. Previous studies have shown that Fgf signaling is necessary for clock oscillation in mouse and that Fgf actually induces Hes1 oscillation in murine cultured cells (Nakayama et al., 2008; Niwa et al., 2007; Wahl et al., 2007). Fgf signaling is also known to control the spatial pattern of hairy gene oscillation in zebrafish and chick (Dubrulle et al., 2001; Sawada et al., 2001). However, the role of Fgf in initiating the traveling wave has not been studied. We therefore examined its role in the emergence of the traveling wave in zebrafish embryos.

We found that fgf8a is expressed at the expected stages and location (Fig. 2A). We first confirmed in zebrafish that Fgf is necessary for her1 expression. Treatment of whole embryos with SU5402, a chemical inhibitor of Fgf signaling, abolished the intrinsic traveling waves as well as the marginal her1 oscillation at early stages (Fig. 2B), consistent with previous findings in mouse (Niwa et al., 2007; Wahl et al., 2007). We then asked whether ectopic activation of Fgf signaling would induce her1 oscillation in the embryo. We transplanted Fgf8b-soaked beads into embryos at the stage before the emergence of the traveling wave (see Fig. 2C for ectopic Fgf activity), and then examined their effect on her1 expression 0.5-1.5 hours after transplantation. The results were striking in that the ectopic her1 expression gradually expanded with time, forming a band of her1 expression around the bead in 9 out of 22 dorsally transplanted samples (Fig. 2D), whereas the remainder exhibited ectopic non-band-like expression of her1 around the beads. We confirmed that this ectopic her1 stripe was actually a traveling wave by observing a spatial transcriptional delay in the direction away from the bead: the cells at the transcription-negative, translational and transcriptional states were detected in sequence starting from the site of bead transplantation (Fig. 2E,F). The induced traveling waves were however restricted to the dorsal side and to the mesodermal layer, forming an incomplete circle (Fig. 2G). This suggests the presence of other permissive factors for her1 expression in the future PSM, before the endogenous wave emerges.

One possible mechanism for creating traveling waves by Fgf signaling is that Fgf activity gradually expands in the anterior direction so that it induces her1 oscillation in sequence with a phase delay. To monitor the activity of Fgf signaling, we examined the expression of sprouty 4 (spry4), one of the well-known downstream genes of Fgf signaling, and found that Fgf activity indeed gradually expands anteriorly from the restricted region of the margin (Fig. 2H; see Fig. S1B in the supplementary material). The timing of spry4 expansion correlated well with the emergence of the first traveling wave (Fig. 2H-J; see Fig. S1B in the supplementary material). The mechanism that expands Fgf activity at this stage remains elusive, although endocytosis could regulate this process (Scholpp and Brand, 2004). Taken together, we conclude that the anterior expansion of Fgf activity triggers the first traveling wave that leaves the marginal expression domain.

In the mature segmentation clock, the stripes of her1 become narrower as they move anteriorly. This characteristic oscillation pattern has been considered to result from a gradient in the oscillation period along the anterior-posterior (AP) axis (Giudicelli et al., 2007; Gomez et al., 2008; Morelli et al., 2009; Palmeirim et al., 1997; Uriu et al., 2009), which is shorter in the posterior and longer in the anterior PSM. We also confirmed this idea using the phase oscillator model (see Fig. S3 in the supplementary material). This idea appears to hold true at the emergence of the traveling wave because the first traveling wave also becomes narrower as it moves. Furthermore, the Fgf activity gradient (higher in the posterior PSM) is known to affect the spatial oscillation pattern of her1 (Dubrulle et al., 2001; Sawada et al., 2001). These facts raise the possibility that early Fgf signaling regulates the period, as well as the timing of initiation, of her1 oscillation at the emergence of the traveling wave. To directly test this possibility, we weakly inhibited gastrulating embryos with SU5402, in which the gradient of Fgf activity was expected to be uniformly decreased. When the stripe first appeared (for control, t=60; for SU5402 treated, t=90), the her1 stripe in treated embryos became narrower and was formed at a more posterior site than that in control embryos (Fig. 3A). Under our experimental conditions, the mesoderm was maintained after SU5402 treatment, but the PSM was reduced in size along the AP axis (see Fig. S4A in the supplementary material). The reduction in PSM size, however, might not account for the narrower and posteriorly shifted her1 expression band in SU5402-treated embryos because the her1 expression pattern is thought to depend primarily on the oscillation dynamics of unit oscillators (e.g. the period) (Giudicelli et al., 2007; Gomez et al., 2008; Morelli et al., 2009; Palmeirim et al., 1997; Uriu et al., 2009). Furthermore, the effect of SU5402 treatment on cellular movement was limited (Fig. 3B; see Fig. S4B in the supplementary material). Taken together, we conclude that the observed change in her1 stripes induced by SU5402 treatment was mainly attributable to the altered dynamics of her1 oscillation.

Since we were unable to determine directly whether the results obtained by SU5402 treatment were the consequence of an altered oscillation period, we instead assessed the effect of altered period on the spatial oscillation pattern in a mathematical simulation, using the phase oscillator model (see Fig. S3 in the supplementary material). For control embryos, we set the intrinsic frequency (the frequency intrinsically given to each virtual oscillator) as the green line in Fig. 3Ca, so that the observed frequency (the actual frequency in simulation, which is not identical to the intrinsic frequency as a result of cellular coupling) also forms a similar gradient as indicated by the green line in Fig. 3Cb. This reproduced the normal emergence of traveling waves in silico (Fig. 3Cc, left). We then examined the spatial oscillation pattern when the gradient of the intrinsic frequency is uniformly decreased, a situation expected to occur when embryos are treated with SU5402 (Fig. 3Ca,b). Under this condition, the phenotype obtained in vivo is clearly reproduced in the simulation, resulting in a posteriorly shifted and narrower stripe (Fig. 3Cc, right). Furthermore, the opposite phenotype was obtained with Fgf bead transplantation (see Fig. S5 in the supplementary material). All these data support the idea that the oscillation period is controlled by Fgf signaling.

Taken together, our findings suggest that at the emergence of the traveling wave, the initiation timing and the oscillation period of her1 are simultaneously controlled by Fgf and co-operatively establish the posterior-to-anterior phase delay in the oscillatory field. In wild-type embryos, the traveling waves appear in a highly coordinated manner despite dynamic morphogenesis during gastrulation. The global control of cellular oscillators by secreted factors, such as Fgfs, could be crucial in this situation. The mechanism by which Fgf signaling modulates her1 negative feedback to induce her1 oscillation and alter the her1 oscillation period remains to be established. Further effort aimed at direct measurements of each reaction step in the her1 negative-feedback loop in response to Fgf will help to identify the Fgf target step, and improve our current understanding of the basic principles governing the emergence of the traveling wave.

We thank Dr Sharron Amercher for providing her1/7 deletion mutants; Dr Julian Lewis for advice and helpful discussions; and the National BioResource Project, Zebrafish, NBRP/Brain Science Institute, RIKEN for zebrafish supply. This work was supported in part by Grants-in-Aid for Scientific Research Priority Area Genome Science and Scientific Research (A and B) and Global COE Program (Integrative Life Science Based on the Study of Biosignaling Mechanisms) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan. K.I. was supported by Research Fellowships for Young Scientists from the Japan Society for the Promotion of Science.