Real-Time Imaging of the Segmentation Clock

The theoretical proposition that the process of somitogenesis (and also segment formation in arthropods) was founded on an interacting “clock” and “wavefront” (the clock and wavefront model; Cooke and Zeeman 1976), occurred more than 20 years before the experimental demonstration by Olivier Pourquie’s group in 1997 (Palmeirim et al. 1997). Based on static images of bisected embryos and comparing temporally delayed halves, they showed by in situ hybridization that c-hairyl gene expression oscillates in the PSM of the chicken embryo for a period of 90 minutes and that each oscillation correlates with the formation of a new somite. Such a segmentation clock operating at the PSM was live-imaged for the first time in the mouse in 2006, almost 10 years after Palmeirim and colleagues’ experiments, using a luciferase-based reporter of the cyclic gene Hesl, a Notch signaling effector and the mammalian homolog of the Drosophila hairy gene (Masamizu et al. 2006).

Since the oscillator period is short, ranging from 25 minutes in zebrafish to approximately 120 minutes in mice (Hubaud and Pourquie 2014), and Hes proteins have short half-lives (around 20 minutes in mice), to visualize the dynamic expression of cyclic genes it is necessary to use reporters with rapid turnover, avoiding the accumulation of the reporter mRNA or protein as somite formation proceeds. For this reason, transgenic animals made for real-time live imaging of cyclic genes had to incorporate sequences that destabilize their mRNA and/or induce the degradation of their protein.

Masamizu and colleagues made use of a mutated Ubiquitin tag resistant to hydrolysis that directed rapid degradation of the Luciferase reporter (Ub-Luc). They replaced the Hesl coding region with the Luciferase coding region, but maintained the endogenous Hesl gene 3'UTR in order to destabilize the reporter mRNA (Masamizu et al. 2006; Soroldoni and Oates 2011). Using this unstable reporter (Hesl-Ub-Luc), they generated transgenic animals and performed time-lapse imaging of ex utero cultured explants of the caudal region of mouse embryos. During the 15-hour movie, several (at least 5) successive waves of bioluminescence were propagated from posterior-to-anterior along the PSM. These in vivo oscillations showed to be stable in both period and amplitude, in contrast to what was observed in vitro with dissociated PSM cells, where period and amplitude varied between cells. Nevertheless, time-lapse analysis of isolated PSM single cells allowed them to demonstrate that oscillations are cell-autonomous and that intercellular communication is important for their stability (Masamizu et al. 2006).

Years later the first fluorescently labeled transgenic mouse was developed by Aulehla and coworkers, who directed the expression of the Venus reporter, an improved version of the yellow fluorescent protein with faster maturation, under the regulation of the Notch cyclic gene Lunatic fringe (Lfng) promoter region. The LuVeLu transgenic mouse, as they named it, also carried a modified PEST domain and the Lfng 3'UTR in order to increase protein and mRNA instability, respectively (Aulehla et al. 2008). Using two-photon microscopy and taking advantage of the brightness of the fluorescent Venus protein, the authors obtained higher temporal resolution in vivo (8.5 minutes) than what was obtained with the Hesl-Ub-Luc transgenic mouse (20 minutes), while maintaining a tissue-level resolution (Masamizu et al. 2006; Aulehla et al. 2008; Soroldoni and Oates 2011).

Mutant mice carrying a stabilized (3-catenin protein that disrupts its posterior- to-anterior PSM expression gradient reported the absence of somite boundaries, as well as the expansion of the posterior PSM. When these mutants were crossed with LuVeLu transgenic mice, time-lapse imaging analysis showed that Lfng-Venus cyclic expression was not abrogated. On the contrary, multiple oscillations traversed the extended PSM, highlighting that the arrest of oscillations and consequently the formation of the new somite border requires the diminution of p-catenin levels (Aulehla et al. 2008).

The transgenic approach described heretofore includes two kinds of reporters: the bioluminescent activity of the luciferase enzyme and the fluorescent emission of a mutated variant of the jellyfish Aequorea Victoria green fluorescent protein (GFP). Both models were developed in mice and their expression was driven by the regulatory region of two genes belonging to the Notch signaling pathway, Hesl and Lfng.

The first and so far only live reporter of clock oscillations constructed in another model organism was the herl.herl-venus transgenic zebrafish line (Delaune et al. 2012; an improved version was then generated by Soroldoni et al. 2014). In order to obtain this cyclic reporter line, Delaune and colleagues tested several constructs and strategies, some of them based on the live cyclic reporters already generated in mice (and commented earlier; Masamizu et al. 2006; Aulehla et al. 2008), but their Her!-Venus reporter showed to be the best combination between dynamic instability and detectable expression using confocal microscopy. For real-time imaging of individual PSM cells, they incorporated important features that improved resolution, bringing it to a cellular level (Delaune et al. 2012). They measured and calculated the phase of the oscillations in individual PSM cells for 4-6 hours, starting at the 8-somite stage and acquiring about 30 z-stack images every 4 minutes, from crosses between the herl.herl-venus line and Notch pathway mutants. They found that cells continued to oscillate but that phase-synchronization between neighbors was lost, indicating that Notch signaling is necessary for neighboring cell synchronization and not for single-cell oscillations. They also discovered by time-lapse analysis that newly divided cells in both WT and mutant backgrounds were more in phase (low phase-shift) with each other than with their neighbor cells. However, only WT sibling cells progressively resynchronized with their neighbors, in contrast to Notch mutant cells that were always desynchronized. In addition, they observed that PSM cells had a marked preference to divide at a certain oscillation phase (off phase in the case of Herl-Venus). Taking into account that 10-15% of the PSM cells are found in M-phase and that each mitosis lasts for a minimum of 15 minutes, as was measured by time-lapse imaging during zebrafish segmentation (Horikawa et al. 2006), this cellular process is considered an important source of noise that can affect cell-to-cell oscillation coupling. Delaune et al. (2012) argued that the tendency of a PSM cell to enter into mitosis at a particular phase would collaborate to minimize the expected developmental noise caused by cell proliferation.

In addition to the crucial role of Notch signaling in cell-cell synchronization during vertebrate segmentation, genes coding for Hes/Her transcription factors that act downstream of the Notch pathway appear to be involved in the establishment of single-cell oscillations by regulating the pace of the clock. Based on mathematical simulations and previous empirical and theoretical work, Julian Lewis (2003) proposed the relationship between the delay in the autoinhibitory feedback loop of Hes/ Her proteins and the length of the oscillation period. After that, great progress has been made to uncover the mechanisms underlying the “pacemaker” of the segmentation clock. Time-lapse imaging highly contributed to these advances, particularly by the Kageyama group, who generated transgenic mice carrying the promoter region of the Hes7 gene fused to the Luciferase gene and destabilized by the non-hydrolyz- able human ubiquitin variant (G76V) (Masamizu et al. 2006). What they evaluated in vivo was the effect of the transcriptional delay on the timing (period) of cyclic gene expression, by removing one, two, or three (all) intron sequences from their Hes7-Ub-Luciferase construct (Takashima et al. 2011; Harima et al. 2013). When the caudal part of transgenic mouse embryos in culture were imaged by time-lapse microscopy, complete elimination of introns (3070 bp transcript) showed the absence of oscillatory expression compared with the transgenic line that contained all three introns (4913 bp transcript) (Takashima et al. 2011), whereas reducing the number of introns to only one accelerates the oscillations and shortens the period of the segmentation clock in 8.8% of cases (Harima et al. 2013).

Using a similar approach, the Kageyama group also generated two mutant versions of a Deltal (Dill) knock-in mouse expressing the Dlll-luciferase fusion protein from the endogenous gene locus by the control of a light-inducible expression system; one mutant with only exon sequences (type 1) and the other with all 10 introns plus an extra sequence (type 2) in order to extend the transcript (Shimojo et al. 2016). Time-lapse imaging analysis of the spatiotemporal expression of the Dlll-Luc reporter showed dampened oscillations with smaller amplitudes in both mutants. In addition, type 1 and type 2 mutants exhibited shorter and longer periods, respectively.

As we can see, live imaging of transgenic embryos has clearly shown the role of the Notch signaling pathway in the control of the segmentation clock at two different levels (Oates et al. 2012): single-cell oscillations (Masamizu et al. 2006; Takashima et al. 2011; Harima et al. 2013) and synchronization between neighboring cells (Masamizu et al. 2006; Delaune et al. 2012; Shimojo et al. 2016). Interestingly, the importance of intron delays in the cyclic expression in individual cells seems to be extended to neighboring synchronization, given that the oscillatory expression of the Dill protein (also demonstrated by time-lapse imaging by Shimojo and colleagues), a membrane-bound ligand of the Notch signaling pathway, must be related to intercellular communication and the maintenance of the oscillatory rhythm.

Beyond the local control of oscillations, segment formation at the anterior part of the PSM depends on the arrest of the traveling waves of cyclic gene expression that move from posterior-to-anterior regions along the PSM. Real-time imaging analysis has allowed researchers to visualize differences in the oscillating period between posterior and anterior regions within the zebrafish PSM. They found more oscillations near the arrest front (anterior PSM) than close to the posterior PSM and only the anterior cell oscillating period matches the timing of segment formation (Soroldoni et al. 2014). These findings were explained by the existence of a Doppler effect modulating the period of segmentation. To do this, the authors generated a new transgenic line (Herl-Venus or Looping), showing oscillatory expression along the entire PSM, improving the transgene construct designed by Delaune and colleagues (2012) that mimicked endogenous oscillations only at anterior and intermediate positions along the PSM. Moreover, Soroldoni and colleagues (2014) used a time- lapse setup developed to perform multidimensional imaging of 20 zebrafish embryos simultaneously with real-time resolution.

Considering the reduction in the number of traveling waves that they also found as the PSM shortens, Soroldoni and coworkers concluded that in zebrafish, both the Doppler effect together with this dynamic wavelength effect (as they call it) cannot account for the scaling of the segment length to the size of the PSM that Lauschke et al. (2013) visualized ex vivo in the mouse. Using a monolayer culture of PSM cells (mPSM) from the tail bud mesoderm of LuVeLu transgenic mice (Aulehla et al. 2008), the authors analyzed the periodic oscillations of the Lfng-Venus reporter activity by time-lapse imaging. Compared to in vivo experiments, ex vivo PSM cultured cells showed equivalent oscillation periods, with traveling waves progressing from the center to the periphery (12-15 oscillations), even forming segment boundaries (Lauschke et al. 2013). In order to visualize the scaling process, they measured the oscillation phase of each cell (as Delaune et al. 2012 did in vivo) forming part of PSM monolayers of different lengths over time, and thus they could determine the spatial phase changes along the mPSM. First, they found that at the center of the monolayer the period was constant, independent of the mPSM length, and that smaller segments arose from shorter mPSMs in a linear relationship. The quantified slope of the phase gradient (the oscillation phase differences between single mPSM cells) was found to be inversely proportional to the segment size. They concluded that the scaling property of the mPSM relies on the conservation of the phase gradient amplitude independent of the mPSM length (Lauschke et al. 2013).

Taking advantage of the same ex vivo culture approaches, two studies performed by the same lab (Alexander Aulehla’s group) addressed on the one hand the emergence of the collective synchronization within the PSM (Tsiairis and Aulehla 2016) and on the other the dynamic relationship between Wnt and Notch signaling oscillations (Sonnen et al. 2018), both using time-lapse microscopy.

Tsiairis and Aulehla imaged reaggregated PSM explants where cells were dissociated and mixed in order to lose their original cell-cell interactions and position along the anterior-posterior axis. PSM explants were obtained from LuVeLu transgenic embryos (Aulehla et al. 2008) and LuVeLu reporter activity was used to monitor the appearance of de novo traveling expression waves. They visualized the formation of regularly spaced multiple foci, or emergent PSM (ePSM) as the authors call them, of synchronized cells oscillating from the center to the periphery. Moreover, the various ePSM formed in a single reaggregated PSM are also synchronized. In addition, Tsiairis and Aulehla performed FACS (fluorescence activated cell sorting) to separate PSM cells from several embryos in two groups (high and low), based on their LuVeLu expression level (intensity). Thus, one cell population was out of phase with respect to the other at their original positions in the intact PSM. When cells were separately mixed, they allowed them to reaggregate and tracked them by time-lapse imaging. The authors found that formed ePSMs retained their oscillatory phases and both were in antiphase with respect to each other (Tsiairis and Aulehla 2016). Together with more interesting and clever experiments combined with the use of time-lapse microscopy, the authors concluded that PSM oscillators have the ability to self-organize forming wave patterns that reflect the integration between the single-cell and the systemic levels.

In the other study, Sonnen and colleagues (2018) generated the first oscillatory Wnt signaling transgenic reporter mouse line, using the promoter of Axin2 (a negative regulator and target of the Wnt/fi-catenin pathway). Using this in vivo reporter combined with the Notch signaling reporter LuVeLu (Aulehla et al. 2008) they performed simultaneous imaging of monolayer PSM cells. Both signaling reporters oscillated out of phase at the center of the mPSM and in phase at the periphery of the monolayer, preceding the region of Mesp2 expression, where Axin2-mediated fluorescence abruptly decreases, which they could visualize by combining the expression of the Axin2 line with a Mesp2-GFP line (Sonnen et al. 2018). In order to study the relative timing of Wnt and Notch signaling oscillations, Sonnen and coworkers set out to control and visualize the rhythm of each oscillator by entraining their cyclic expression on a microfluidic system using pharmacological manipulation combined with time-lapse imaging analysis. Using this approach they found that Notch and Wnt signaling oscillations are coupled, which means that when altering the rhythm of one, the oscillations of the other are also altered, maintaining their relative synchronization. On the other hand, when the authors simultaneously entrained both signaling oscillations, resulting in antiphase oscillations at the “anterior” mPSM, LuVeLu expression arrest was delayed and no sign of a morphological segment appeared, indicating that Wnt/ Notch phase shift is crucial for segment formation (Sonnen et al. 2018).

The molecular nature of the segmentation clock forced the use of biolumines- cent or fluorescent dynamic reporter systems to visualize and analyze the dynamic expression of cyclic genes, first at the tissue level and later with single-cell resolution (Table 8.1). By using only two animal models, the mouse and zebrafish, some


Oscillatory Reporters Created and Used to Elucidate the Underlying Mechanisms of the Segmentation Clock in Vertebrates

In Mice




Hes 1 -Ub-Luciferase


Masamizu et al. 2006*


(Four variants: one containing all three introns and the others lacking one. two, or all three introns)


Takashima et al. 2011 * Niwa et al. 2011 Gonzalez et al. 2013 Harima et al. 2013

LuVeLu (Lunatic-fringe-Venus/YFP)


Aulehla et al. 2008* Lauschke et al. 2013 Tsiairis and Aulehla 2016 Sonnen et al. 2018

Delta 1 -Ub-Luciferase

(And two mutant types: one containing only exon sequences and the other having all ten introns plus an extra sequence)


Shimojo et al. 2016*




Sonnen et al. 2018*




Sonnen et al. 2018*


(Lunatic-fringe-mVenus-PEST) In Zebrafish


Sonnen et al. 2018*

Her 1-Venus


Delaune et al. 2012* Shihet al. 2015


(improved version of Herl-Venus)


Soroldoni et al. 2014*

^Original article where the transgenic reporter was first described.

research groups have analyzed long-standing questions in vivo about the onset of oscillations, their synchronicity, and how cells maintain their rhythm along segmentation. Interestingly, the use of time-lapse imaging in the study of segmentation was not reduced to cell tracking but has been combined with very different approaches, such as intensity kymograph analysis (Aulehla et al. 2008; Takashima et al. 2011; Lauschke et al. 2013; Soroldoni et al. 2014; Tsiairis and Aulehla 2016; Sonnen et al. 2018), FACS (Tsiairis and Aulehla 2016), ex vivo and two-dimensional monolayer PSM cultures (Lauschke et al. 2013; Sonnen et al. 2018), mutant backgrounds (Aulehla et al. 2008; Niwa et al. 2011; Takashima et al. 2011; Delaune et al. 2012; Harima et al. 2013; Shimojo et al. 2016) and pharmacological manipulation of signaling pathways in real time and at desired time windows (Gonzalez et al. 2013; Sonnen et al. 2018), among others.

< Prev   CONTENTS   Source   Next >