Live Imaging to Study Sequential Segmentation in Vertebrates
The extraordinarily dynamic nature of axial elongation and segment formation that take place in all vertebrates analyzed so far has been revealed paradoxically by decades of experiments based mainly on static images. Since the first studies at the end of the 1980s (Wilson et al. 1989; Thorogood and Wood 1987), a modest number of publications have incorporated time-lapse imaging in their analysis. Although the boom of the use of this technique to understand the elongation and segmentation processes is still yet to come, some recent studies in vertebrates (and also arthropods) have demonstrated that live imaging is a powerful tool for investigating the complex relationship between cellular behavior and the molecular machinery underlying germband extension and the ordered formation of body segments.
Despite the small contribution of time-lapse analysis (in relation to the relative number of papers), there are some key points during axial extension and segmentation in vertebrates where the use of live imaging was essential or at least very influential to reach our current understanding of these processes. These include: (1) cellular dynamics during segmental border formation, (2) the segmentation clock including the control of segment size and number, and (3) the migratory behavior of presomitic mesoderm (PSM) cells during axial elongation, especially the comparison between posterior and anterior PSM.
Cell Dynamics during Somite Border Formation
During somitogenesis, the sequential formation of regularly sized segments implies the passage from an unsegmented tissue composed of undifferentiated cells to an ordered and epithelialized somite. Moreover, at the transition zone, a segment boundary is formed that separates the PSM from the newly formed somite. Although the molecular mechanisms involved in this process are well documented, the molecular dynamics and their cellular context are less understood.
Time-lapse recording using Nomarski differential interference contrast (DIC) microscopy, without any labeling but taking advantage of the transparency of the embryos of the teleost fish Barbus conchonius, allowed Peter Thorogood and Andrew Wood to describe for the first time the modest individual cell rearrangements that take place during the formation of a new somite boundary (Thorogood and Wood 1987; Wood and Thorogood 1994). Similar results were obtained years later in zebrafish, when the analysis of time-lapse imaging showed that interso- mitic borders are formed both in wild type (WT) as well as in convergent extension mutant embryos, based on local epithelial movements (Henry et al. 2000). The separation between the new somite and the posteriorly located PSM occurred despite the absence of mediolateral convergence in these mutants, maintaining the cellular behaviors observed in the WT. Additionally, the lack of internal mesenchymal cells within mutant somites (who only have border cells) revealed that these cells are dispensable for somite boundary formation (Henry et al. 2000).
When cellular movements were analyzed during segmentation in the chicken embryo, in ovo time-lapse imaging showed that far from being an ordered division of a pre-patterned PSM, as had been inferred from fish and Xenopus embryos (Thorogood and Wood 1987; Wilson et al. 1989; Henry et al. 2000), avian somite border formation was a dynamic process that involves high cell motility (Kulesa and Fraser 2002). Chick eggs were injected with the vital dye Bodipy ceramide in order to fluorescently label cell membranes and track them at the site of the presumptive somite border formation. The high spatiotemporal resolution that was achieved (10— 15 pm thick z-stacks captured every 2 minutes) revealed a “ball and socket” separation of the recently formed somite from the anterior part of the PSM (forming a hollow structure of cells that contains the ball-shaped somite). Here, different groups of cells move in opposed directions, exchanging positions in the anterior-posterior axis. In the end, original posterior cells become integrated into the forming somite and cells that begin at a more anterior position end up in the anterior PSM, leaving a gap between the new' somite and the presomitic mesoderm, which creates the somite border (Kulesa and Fraser 2002).
At the molecular level, the vertebrate segmental boundary is also defined at the anterior PSM, where posterior-to-anterior traveling waves of gene expression (known as the segmentation clock; see Chapter 5) are arrested by a determination front mainly defined by posterior Wnt/FGF gradients. Since segmentation is coupled to axis elongation, this “wavefront” moves posteriorly as the embryo grows, converting the temporal progression of oscillations into spatially patterned segments (Palmeirim et al. 1997; Dubrulle et al. 2001; Aulehla et al. 2003; Dubrulle and Pourquie 2004). Just anterior to this front, the dynamic expression of a key transcription factor, Mesoderm Posterior 2 (Mesp2 in mice; related to cMeso-1 in chickens, Thylacine 1 in Xenopus and Mesp-a/-b in zebrafish), is necessary to arrest oscillations and define the segmental border, as well as to establish the rostrocaudal somite orientation (Saga et al. 1997; Buchberger et al. 1998; Sparrow' et al. 1998; Sawada et al. 2000).
In an attempt to reproduce the endogenous expression pattern of Mesp2 and, thus, correlate its localization w'ith the presumptive new somite border, Morimoto and colleagues generated a Mesp2-venus knock-in mouse (Morimoto et al. 2005). They visualized the expression of Mesp2-venus at the anterior border of the future somite in vivo, before a morphologically recognizable limit is formed at the PSM. This evidence, combined with Mesp2 and Notch 1 activity (based on the presence of the Notch intracellular domain) double immunostaining, allowed them to propose that new' somite boundaries are formed at the border between Notch 1 activity and Mesp2 expression. However, they only managed to reproduce one cycle of Mesp2-venus expression in their PSM culture conditions, losing part of the dynamic behavior of the Mesp2 endogenous protein (Morimoto et al. 2005).
Years later, by using double destabilized luciferase reporters of Mesp2 and Fgf signaling (Mesp2-UbEluc and Dusp4-UbSLR, respectively) and live imaging, Niwa et al. (2011) showed the periodic onset of Mesp2 expression at the prospective limit between newly formed somites in vivo and that this dynamic expression was the consequence of the cyclic cutoff in Fgf-Erk activity oscillation (Niwa et al. 2011). This last part w'as well supported by time-lapse analysis of the expression of both reporters in mice where oscillations were not operating. For example, in Hes7-null mice, w'here the oscillatory expression of Hes7 (a Notch target gene) is abolished, the authors found no sign of phosphorylated Erk (the active form of the Fgf effector Erk) oscillations and a desynchronized onset of Mesp2. A similar effect was found after blocking the Fgf signaling pathway with SU5402 (Fgf receptor inhibitor): no Fgf-Erk activity and a premature and continuous expression of Mesp2 (Niwa et al. 2011). Interestingly, Erk activity was then uncovered in zebrafish as an earlier marker of the future somite boundary than mesp-b expression, showing that segmentation was established at least three segments before it was expected (Akiyama et al. 2014). For this they used time-lapse images to track photoconverted cells at the presumptive position of Erk activity, in a transgenic line where the photoconvertible fluorescent protein KikGR is expressed under the herl promoter. Several rounds of segmentation later, photoconverted cells were found at the anterior border of the formed somite, confirming p-Erk as an earlier somite boundary marker in zebrafish (Akiyama et al. 2014).
Regarding the segmentation clock, the relationship between oscillations and somite border formation is established at the aforementioned wavefront, where cyclic gene expression is arrested. It was known that oscillations gradually slow down from the posterior-to-anterior PSM, but it was not until live imaging allowed for the measurement of the dynamic expression of an oscillator that the molecular behavior at the determination front was analyzed (Lauschke et al. 2013; Shih et al. 2015). Using the herl:herl-venus transgenic zebrafish line, where the regulatory region of the herl gene drives the cyclic expression of the yellow fluorescent protein Venus fused to Herl (Delaune et al. 2012; see later), Shih and colleagues were able to track the oscillation period of the Herl-Venus reporter across the PSM, with single-cell resolution (Shih et al. 2015). They confirmed the gradual slowing of oscillations along the PSM, estimating that anterior cells have a period of around 1.5 times longer than posterior cells. They also found that the oscillator at the anterior PSM is expressed in alternate segments, showing a peak at the presumptive new somite position. Since they did not find an abrupt cessation in oscillations at the determination front, as was expected, they proposed a new model of segmentation based on phase (and antiphase) gradients instead of the transition from permissive (clock) to restrictive (wave- front) phases, as the clock and wavefront model proposes (Lauschke et al. 2013; Shih et al. 2015; Pais-de-Azevedo et al. 2018).