SEARCH

SEARCH BY CITATION

Keywords:

  • coupled oscillator;
  • segmentation clock;
  • zebrafish

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. DISTINCT OSCILLATION MODES IN THE PSM IN ZEBRAFISH
  5. COUPLED OSCILLATION IN FLASHING FIREFLIES AND PSM CELLS
  6. NOISE AND NOISE-RESISTANCE IN THE SEGMENTATION CLOCK
  7. PERSPECTIVE
  8. Acknowledgements
  9. REFERENCES

A unique feature of vertebrate segmentation is its strict periodicity, which is governed by the segmentation clock consisting of numerous cellular oscillators. These cellular oscillators, driven by a negative-feedback loop of Hairy transcription factor, are linked through Notch-dependent intercellular coupling and display the synchronous expression of clock genes. Combining our transplantation experiments in zebrafish with mathematical simulations, we review how the cellular oscillators maintain synchrony and form a robust system that is resistant to the effects of developmental noise such as stochastic gene expression and active cell proliferation. The accumulated evidence indicates that the segmentation clock behaves as a “coupled oscillators,” a mechanism that also underlies the synchronous flashing seen in fireflies. Developmental Dynamics 236:1416–1421, 2007. © 2007 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. DISTINCT OSCILLATION MODES IN THE PSM IN ZEBRAFISH
  5. COUPLED OSCILLATION IN FLASHING FIREFLIES AND PSM CELLS
  6. NOISE AND NOISE-RESISTANCE IN THE SEGMENTATION CLOCK
  7. PERSPECTIVE
  8. Acknowledgements
  9. REFERENCES

Somites are formed one by one from the anterior end of the un-segmented mesoderm (presomitic mesoderm, PSM) in an anterior-to-posterior (A-P) direction. One of the most important aspects of somitogenesis is its strict periodicity; a pair of somites is formed every 30 min for zebrafish, 90 min for chick, and 120 min for mouse (Saga and Takeda,2001; Pourquie,2003). Because of this, it has long been speculated that a segmentation clock creates this periodicity during somite formation (Cooke and Zeeman,1976). A recent breakthrough was the finding of a cyclic gene linked to periodic somite formation in the PSM (Palmeirim et al.,1997). Since then, genetic and experimental evidence has accumulated to place a hairy negative-feedback loop at the core of the cellular clock in PSM cells (unit oscillator) (reviewed in X in this special issue). A widely accepted idea is that the segmentation clock is an ensemble of numerous cellular oscillators connected through the Notch pathway (Fig. 1a) (Holley and Takeda,2002; Rida et al.,2004).

thumbnail image

Figure 1. The segmentation clock in the zebrafish embryo. a: Schematic representation of the zebrafish presomitic mesoderm (PSM) and the PSM oscillators. Details are described in the text. b: Representative images and a graphic depiction of the her1 expression profile. The positions of the her1-positive cells in 23 embryos are indicated by a solid bar. The her1 expression domain appears every 30 min in the posterior PSM, and travels anteriorly. The mode of her1 expression changes from a synchronized to traveling mode in the intermediate zone (dotted line). c: High-resolution ISH detection of the subcellular localization of her1 mRNA. The PSM cells display a cycle of no signal (c1), nuclear dots (c2), and cytoplasmic signals (c3). d: Representative high-resolution ISH images of her1 expression (green). Numbered insets are enlarged in d1–d4. Nuclei are counter-stained with propidium iodide (magenta). d1–d3 indicate the posterior PSM, which exhibits synchronous oscillation, whereas d4 from the intermediate zone shows a gradual transition of transcription toward the anterior PSM. Scale bar = 20 μm.

Download figure to PowerPoint

In spite of the recent identification of clock components and oscillatory genes (Dequeant et al.,2006), the collective behavior of PSM oscillators remains largely unknown. One reason for this is that current genetic and molecular analyses at either the whole-animal or cellular level alone cannot fully address the operating principles that underlie a collection of mutually interacting unit oscillators. We, thus, need to step back and deal with the whole segmentation clock normally operating in vivo (or in a near in vivo situation such as organ culture). System-level properties can be examined when the responses of normal system to external stimuli are analyzed. The zebrafish is an ideal experimental platform to perform these kinds of analyses. The other problem we would face is a limitation to our imagination. Unfortunately, our minds are not so good at grasping the collective behavior of numerous unit oscillators. In this situation, mathematics, which most bench biologists never pick up, comes into play. Fortunately, many mathematical considerations have been made by theoretical biologists to understand animal segmentation (Meinhardt,1986; Collier et al.,2000; Lewis,2003; Baker et al.,2006). Having these tools in hand, we are now beginning to understand the basic principles of the segmentation clock, particularly in relation to how it maintains synchrony in the presence of biological noise. In this process, Notch-dependent intercellular communication was found to be essential. In our present report, we review our recent analysis and other related studies that focus on the collective oscillation of the clock population, and discuss possible future directions in achieving a better understanding of the segmentation clock.

DISTINCT OSCILLATION MODES IN THE PSM IN ZEBRAFISH

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. DISTINCT OSCILLATION MODES IN THE PSM IN ZEBRAFISH
  5. COUPLED OSCILLATION IN FLASHING FIREFLIES AND PSM CELLS
  6. NOISE AND NOISE-RESISTANCE IN THE SEGMENTATION CLOCK
  7. PERSPECTIVE
  8. Acknowledgements
  9. REFERENCES

In the developing zebrafish embryo, a new wave of her1 (zebrafish hairy1) expression appears in the tailbud every 30 min (the period of one-somite formation in zebrafish), becomes narrower as it moves rostrally, and finally arrests at the future segmentation point in the anterior PSM before decaying (Holley et al.,2000; Sawada et al.,2000). As shown in Figure 1b, in the posterior domain (0–30% of the relative PSM length), her1-positive cells are simultaneously detected during the limited period of the cycle, suggesting that these cells oscillate synchronously. In contrast, in the intermediate and anterior domains (30–100% of the relative PSM length), her1-positive cells form a traveling wave with a gradual anterior shift. Thus, as reported also in chick (Dale et al.,2003), zebrafish hairy expression displays a transition from a synchronous to a traveling mode, in which a traveling wave appears to bud off from the anterior of the synchronous expression domain.

The conventional technique for in situ hybridization (ISH) can detect the accumulation of transcripts in the cytoplasm, but cannot resolve the time-sequence of transcription, such as initiation and termination. To determine the phase of individual oscillators with a higher time resolution, a recently developed fluorescent high-resolution ISH has been shown to be a very powerful tool (Kosman et al.,2004; Julich et al.,2005). An example of the usefulness of this new technique can be seen in Figure 1c. When the zebrafish embryo PSM was hybridized with a her1 probe, cells display one of the following patterns of mRNA localization; no signal, nuclear dots, or cytoplasmic localization, representing states with no transcript, with active transcription, or with a ready-to-translation state, respectively.

In a series of embryos fixed at different phases of the segmentation cycle, almost every PSM cell displays one of these three patterns of mRNA localization, confirming that the oscillatory expression of her1 is synchronized in the posterior PSM (Fig. 1d). In the intermediate PSM, however, a gradual transition is observed from the nuclear signals in the anterior few rows to the cytoplasmic signals in successive rows along the A-P axis (Fig. 1d). This indicates that the timing of transcription is delayed toward the traveling direction. As the wave reaches the anterior PSM, the number of both nuclear-positive “forerunners” and cytoplasmic-positive “followers” is gradually decreased, reflecting the fact that each wave arrests prior to segmentation. Thus, the transcriptional activity of individual hairy oscillators is differentially regulated along the A–P axis, which would cause a transition from synchronized oscillation to a traveling state.

The expression of multiple stripes in the intermediate PSM, characteristic of zebrafish clock genes, is not observed in other vertebrates such as the medaka (Gajewski et al.,2006), suggesting that different timing or distinct mechanisms function during mode transition in zebrafish. In contrast, the synchronized oscillation that is evident in the posterior PSM is the earliest and also the most common mode of oscillation in all of the vertebrate species examined so far (Bessho et al.,2003; Maroto et al.,2005), and thus should be the first target of clock oscillation studies.

COUPLED OSCILLATION IN FLASHING FIREFLIES AND PSM CELLS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. DISTINCT OSCILLATION MODES IN THE PSM IN ZEBRAFISH
  5. COUPLED OSCILLATION IN FLASHING FIREFLIES AND PSM CELLS
  6. NOISE AND NOISE-RESISTANCE IN THE SEGMENTATION CLOCK
  7. PERSPECTIVE
  8. Acknowledgements
  9. REFERENCES

In general, persistent synchrony is a common feature in the universe and nature (Winfree,1967). A well-known and classical example of this in the natural world is fireflies blinking on and off in unison. Recent experimental and mathematical studies have suggested a mechanism for this enigmatic phenomenon (for a comprehensive description, see Winfree,2000; Strogatz,2003). There are two crucial cues in the behavior of the fireflies; one is that the fireflies alone, when isolated, keep to a steady beat at more or less regular time intervals, and the other is that they communicate with light by adjusting their rhythms in response to the flashes of others. The current accepted model is that in a congregation of flashing fireflies, every one is repeatedly sending and receiving signals, shifting the rhythms of others and being shifted by them in turn, which maintains synchronous oscillation. This kind of system is thus referred to as “coupled oscillators” and two or more oscillators are said to be coupled if some physical or chemical process allows them to influence one another. Indeed, the mechanisms underlying coupled oscillators have been implicated in other biological systems such as the circadian clock and the beating heart (Winfree,2000; Strogatz,2003).

In a similar manner to groups of fireflies, PSM cells oscillate in synchrony. At first glance, the segmentation clock appears to share similar features with the fireflies. Single PSM cells, when isolated and cultured, manage to keep cycling for a certain period of time, although their oscillation is unstable (Masamizu et al.,2006), and can communicate with signaling molecules in the PSM (Maroto et al.,2005). Furthermore, recent molecular and genetic analyses of both mutant mice and zebrafish have been used to propose a model that hairy-negative feedback lies at the core of the oscillator and also that Hairy-regulated (i.e., oscillator-linked) intercellular communication couples the individual oscillators (Fig. 1a) (Jiang et al.,2000; Pourquie,2003). If this model really works in vivo, PSM oscillators will be able to influence one another. Indeed, the segmentation clock behaves as a coupled oscillators, as described below.

PSM Oscillators Utilize Notch Signaling to Communicate With One Another

One may speculate that if PSM oscillators are functionally coupled, they would respond to external stimuli in a non-cell-autonomous manner. This prediction was tested in our zebrafish system in which non-oscillating cells with high levels of Delta proteins were placed onto the host PSM (Fig. 2a) (Horikawa et al.,2006). These donor cells were obtained from embryos in which the translation of her1 and her7 had been inhibited by morpholino anti-sense oligonucleotides (her-MO cells) (Fig. 2b) (Oates and Ho,2002). In this case, the oscillation of the segmentation clock was affected by these signaling cells in a non-cell-autonomous manner, indicating the presence of intercellular coupling of the PSM oscillators. Interestingly, the segment positions were found to always shift anteriorly and accordingly, the segment size was locally reduced around the explants (Fig. 2c). High-resolution ISH further confirmed the local acceleration of her1 oscillation in the cells around the transplant (Fig. 2c). Moreover, this acceleration activity of her-MO cells was found to depend upon the function of Delta, confirming that Notch signaling plays a role in the intercellular communication of the segmentation clock.

thumbnail image

Figure 2. Notch-dependent intercellular communication in the segmentation clock. a: Cell-transplantation assay at the blastula stage. b: Donor cells that constitutively expresses DeltaC due to a morpholino-knockdown of Her. c: Effects of her-MO cells on the transcription of her1 (left) and on the segmentation points (right). The timing of her1 transcription is locally advanced in the area near to the explants in the posterior PSM. The nuclei, her1 mRNA, and explants are stained in red, green, and blue, respectively. Arrowheads indicate nuclear her1 cells, the phase of which is advanced by transplantation. This effect results in a segment-shift phenotype. The segment positions are anteriorly shifted on the transplanted side (arrows). Donor cells are in red and the dashed line indicates the last normally formed segment border. Scale bar = 20 μm. d: 1-D simulation of oscillating PSM cells (dots). Snapshots of the calculated results in the 1st (up) and 10th-round of oscillation (down) are shown. An actively signaling cell is represented by a red arrow. The active signal from the transplanted cell influences the adjacent cell to accelerate the oscillating. After 10 rounds of oscillation, the oscillation phase of the red PSM cell advances by 13.2%, as compared with the green PSM cell that is located far from the signaling cell. This effect is transmitted in succession, and results in the phase-shift of relatively distant cells, although the effect is still locally limited. e: The smaller somite (red solid box) is formed by the accelerated oscillation (red line). The spatial pattern of clock oscillation (lines at the bottom) is translated according to the clock-and-wavefront mechanism.

Download figure to PowerPoint

These experiments were the first to experimentally change the oscillation period of the clock in the normal PSM, and thereby to show that the oscillation period controls the somite size, a prediction of the clock-and-wavefront mechanism (Cooke and Zeeman,1976; Pourquie,2003). A more intriguing finding in terms of clock synchronization was that the effects of these activating signals do not influence the entire clock but are limited to a range of one to two somites. Furthermore, one may wonder why the segmentation clock was always advanced on the transplanted side, as in general, biological oscillators can either be delayed or advanced upon stimulation, depending on their phase. Since the results of our transplantation experiments reflect the long-term consequences of the interaction between the host coupled oscillators and non-oscillating signaling cells, a correct interpretation goes beyond our imagination and thus requires mathematical simulation.

Mathematical Simulations

Among the mathematical models of the segmentation clock that have been proposed previously, the model of Lewis (2003) has been tested for its ability to account for the results of earlier in vivo experiments (Hirata et al.,2002). This model was constructed based on the proposed molecular network shown in Figure 1a and can be used to explain both the autonomous oscillation of cycling genes at the single-cell level and the Notch-dependent synchronized oscillation at the two-cell level. A one-dimensional simulator was subsequently constructed, based on mathematical equations that are similar to those used in the Lewis model (Horikawa et al.,2006). An example is shown in Figure 2d, which represents the posterior PSM, comprising 10 linearly connected cells along the A-P axis. In this simulator, the transcription of her1 is assumed to be a consequence of combinatorial regulation by both Her-dependent repression (cell-autonomous) and Notch-dependent activation (non-cell-autonomous), whereas that of delta is solely regulated by Her in each oscillator. Under appropriate conditions, the levels of her1 mRNA oscillate synchronously, as seen in vivo (open circles in Fig. 2d).

The collective behavior of PSM oscillators in the above transplantation experiment is perfectly predicted by this 1-D simulator. When a continuously signaling cell is placed at the posterior end of the virtual PSM (arrows in Fig. 2d), the synchrony is gradually affected in the region near the actively signaling cell (red dots in Fig. 2d). Moreover, when the oscillation dynamics of each cell are translated into a spatial pattern, according to the clock-and-wave front mechanism, the formation of a smaller somite is predicted (Fig. 2e). The point of this simulation result is that the signaling cell always advances the host oscillation but that its effect is limited to a short range, even after 10–20 rounds of cycles. Hence, although the 1-D simulator neglects molecular details, it is a very helpful tool in interpreting the results of in vivo experiments, and provides important insights into how PSM oscillators are coupled (Notch-dependent local coupling) and how this coupling activity is regulated (linked by internal hairy oscillator).

NOISE AND NOISE-RESISTANCE IN THE SEGMENTATION CLOCK

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. DISTINCT OSCILLATION MODES IN THE PSM IN ZEBRAFISH
  5. COUPLED OSCILLATION IN FLASHING FIREFLIES AND PSM CELLS
  6. NOISE AND NOISE-RESISTANCE IN THE SEGMENTATION CLOCK
  7. PERSPECTIVE
  8. Acknowledgements
  9. REFERENCES

The coupled oscillators is known to be a robust system that is resistant to internal and external noise. One can easily imagine that biological systems function in “noisy environments” (Raser and O'Shea,2005) and of course the segmentation clock is no exception. Indeed, both the period and phase of individual oscillators fluctuate in the segmentation clock. One reason could be the stochastic nature of rhythm-generating reactions such as the transcription, translation and translocation of key factors (McAdams and Arkin,1997; Blake et al.,2003; Shav-Tal et al.,2004). More profound noise can be caused by the active cell-proliferation that occurs in the entire PSM; in the synchronized zone, during one cycle of oscillation, 10–15% of cells experience a mitotic phase (M-phase) of at least 15 min in duration. During mitosis, the transcription and translation of nearly all genes are arrested (Prescott and Bender,1962). As a result, a proportion of PSM cells (about 10%) are always out of phase, even in the synchronized zone of the posterior PSM (Horikawa et al.,2006), and this tendency is often evident for other biological oscillators (Nagoshi et al.,2004; Welsh et al.,2004). Given the remarkably short period (30 min) of her1 oscillation, the presence of numerous dividing cells with the 15-min M-phase may not be negligible for coherent oscillation.

The effects of noise must be minimized to maintain coherent oscillation. According to the basic nature of coupled oscillators, mutual coupling can endow a system with noise resistance (Winfree,1967). This is, indeed, the case in the segmentation clock, where Notch-dependent intercellular communication again plays a crucial role in a noise reduction process. In a 1-D simulation in which noise observed in vivo is introduced as a variable, cells with different free-running frequencies mutually entrain and acquire a common frequency in the presence of Notch-Delta communication, whereas this synchrony is gradually lost without intercellular coupling. Significantly, this is observed in zebrafish embryos whereby the synchronous oscillation of her1 is easily disrupted after brief treatment with DAPT (Geling et al.,2002), a gamma-secretase inhibitor that reduces Notch activity (Fig. 3a) (Horikawa et al.,2006). Thus, PSM cells are always watching one another and adjusting their asynchronous neighbors to the consensus oscillation of the majority. The question then arises of what the outcome would be if two independently oscillating populations are juxtaposed under normal conditions. This situation was again achieved in zebrafish embryos by cell transplantation (Fig. 3b–d), and similar results were obtained by both in vivo and in silico experiments; transplanted asynchronous oscillators quickly become synchronized with the phase of the host oscillation (Fig. 3b–e). Phase synchronization must occur everywhere during clock oscillation, and, thus, these results illustrate how biological noise can be absorbed in the normally operating clock in order to maintain synchrony.

thumbnail image

Figure 3. Notch-dependent phase synchronization. a: Disrupted synchrony in 10-somite-stage embryos transiently treated with DAPT for 2 hr. A large number of cells go out of phase, showing both delayed (arrows) and advanced (arrowheads) phases, in the synchronized zone of the posterior PSM (compare with 1d1-3 in the normal PSM). Scale bar = 20 μm. b–d: Wild-type cells of the posterior PSM were directly transplanted into normal embryos at the same axial level. her1 expression is shown (purple) in these embryos just after transplantation (arrow in c) and at 1.5 hr later (d). Donor cells in d are encircled with white dots. Scale bar = 50 μm. e: Interaction between two oscillating groups calculated by 1-D simulation. A cell with a delayed phase (1/4-cycle) is transplanted at t = 140 min (arrow). The phase of the explants is gradually synchronized to that of the host over several rounds of oscillation. The levels of her1 mRNA in the host (green) and explant cells (magenta) are traced as a function of time. For details, see Horikawa et al. (2006).

Download figure to PowerPoint

PERSPECTIVE

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. DISTINCT OSCILLATION MODES IN THE PSM IN ZEBRAFISH
  5. COUPLED OSCILLATION IN FLASHING FIREFLIES AND PSM CELLS
  6. NOISE AND NOISE-RESISTANCE IN THE SEGMENTATION CLOCK
  7. PERSPECTIVE
  8. Acknowledgements
  9. REFERENCES

The evidence that we and others have accumulated so far supports the idea that the segmentation clock basically follows the principle of coupled oscillators, in which synchronous oscillation is maintained through mutual coupling in a “democratic” manner. This mechanism assures the robustness of the segmentation clock and its resistance against inherited biological noise. Indeed, developing biological systems such as vertebrate embryos need to satisfy two conflicting requirements. Embryos must grow quickly but at the same time undergo precise pattern formations. The former relies on high levels of mitosis, which potentially endangers the correct transcriptional regulation required for the latter. Thus, these developing systems must have self-contained noise-reduction machinery. Recent studies of the segmentation clock demonstrate that Notch-dependent intercellular coupling is one example of such machinery, but is not, of course, the only such mechanism (Bokel et al.,2006). Since developmental biologists have paid little attention to developmental noise in the past, future studies will need to uncover the different types of inherent noise-reduction systems working during embryonic pattern formation.

Despite recent advances in our understanding of the segmentation clock, many questions remain unanswered. We have assumed that PSM oscillators are essentially identical in nature and that they are all coupled equally. However, this may not, in fact, be the case in vivo because the cellular oscillators must alter their characteristics as their relative positions move along the A-P axis, or reach maturation in the PSM. These changes may cause the transition between the synchronized and traveling mode, an essential step for translating temporal information into spatial patterning. Fgf signaling has been implicated in this transition, as the level of Fgf activation (highest in the posterior PSM) serves as a positional cue within the PSM that regulates the progression of the cyclic wave (Dubrulle et al.,2001; Sawada et al.,2001). We do not yet know how the level of Fgf activation influences the behavior of the individual oscillators and/or the intercellular coupling. In this context, Her 13.2, downstream of Fgf signaling and forming a complex with Her1, could be a key player (Kawamura et al.,2005). Simulation experiments that incorporate these factors will, thus, be helpful in increasing our understanding of this process. It will also be interesting to ascertain whether the segmentation clock can generate synchrony from artificially randomized situations. Since under appropriate conditions, order can spontaneously emerge from randomness in the coupled oscillators, hairy synchronized oscillation could be gradually formed, stating with pairs, trios, and then patches of cells in phase. Because of the likely technical difficulties, no attempt has yet been made to address this issue. Finally, we do not currently know how the initial synchrony is established in the presumptive PSM. However, detailed expression analyses combined with mathematical simulations in the future is likely to provide insights into the mechanisms that start the clock synchronization events.

We are still far from a complete understanding of the segmentation clock. However, this system will remain a good model for the study of biological oscillation in the near future as we already have a much better idea about the nature of individuals (PSM cells and clock components), their behavior (oscillatory gene expression), and their interactions (oscillator-linked intercellular communication through Notch-Delta) compared with other biological oscillation systems.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. DISTINCT OSCILLATION MODES IN THE PSM IN ZEBRAFISH
  5. COUPLED OSCILLATION IN FLASHING FIREFLIES AND PSM CELLS
  6. NOISE AND NOISE-RESISTANCE IN THE SEGMENTATION CLOCK
  7. PERSPECTIVE
  8. Acknowledgements
  9. REFERENCES

The 1-D oscillator presented here was constructed in collaboration with the laboratory of Shigeru Kondo at Nagoya University. We thank Dr. Atsuko Takamatsu of Waseda University for useful discussions and critical reading of the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. DISTINCT OSCILLATION MODES IN THE PSM IN ZEBRAFISH
  5. COUPLED OSCILLATION IN FLASHING FIREFLIES AND PSM CELLS
  6. NOISE AND NOISE-RESISTANCE IN THE SEGMENTATION CLOCK
  7. PERSPECTIVE
  8. Acknowledgements
  9. REFERENCES