Building the Zebrafish Lateral Line System

Protoneuromasts are formed within the migrating primordium, starting from its trailing end, as clusters of cells apically constrict and form epithelial rosettes. Their formation is promoted by fibroblast growth factor (Fgf)–signaling centers, which form periodically in the wake of a shrinking Wnt–active domain that inhibits epithelial rosette formation and progressively shrinks toward the leading end of the primordium (Wnt and Fgf pathways are signaling pathways). However, the precise number and size of epithelial rosettes is not strictly dependent on a pre-pattern of Fgf–signaling activity, as it is broadly influenced by the balance of mechanical interactions that promote or oppose formation of epithelial rosettes. When chemokine-dependent migration of leading cells is compromised, the resulting slowing of the primordium is accompanied by the fusion of epithelial rosettes to form fewer larger rosettes. However, such fusion is not observed when Fgf signaling, responsible for migration of trailing cells, is inhibited to slow primordium migration. These observations can be accounted for by a mechanics-based model, in which local interactions associated with apical constriction and cell adhesion promote aggregation, while tension along the length of the primordium, influenced by the relative efficacy of leading and trailing cell migration, opposes such aggregation. We described the development of a computational Cellular Potts model, which allowed us to explore how the relative speed of leading versus trailing cells, as well as changes in cell adhesion and mechanical coupling, differentially regulated by Wnt and Fgf signaling, can influence the pattern of neuromast formation and deposition by the migrating primordium. Our studies illustrate how signaling and mechanics cooperate to coordinate self-organization of morphogenesis in the migrating primordium.

Neuromasts, the mechanosensory organs of the Lateral Line system, are periodically formed and deposited by the migrating pLLp from its trailing end, as it migrates under the skin from the near the ear to the tip of the tail, to spearhead formation of the posterior Lateral Line system in zebrafish. Understanding how they are formed and deposited in a robust and reproducible way serves as a model for understanding more broadly the steps that determine the reproducible self-organization of organ systems during development.

The pLLp forms as a group of about 140 cells separates from an epithelial placode located behind the ear. Prior to the initiation of migration, cells in the primordium are characterized by broad, relatively uniform Wnt–regulatory activity. However, as the primordium begins to migrate caudally towards a source of chemokine signals secreted by the horizontal myoseptum, Wnt activity becomes polarized such that it is highest at the leading end of the primordium and lowest at its trailing end. Such polarization provides the context for subsequent patterning events within the primordium that coordinate both polarized migration of the primordium along a relatively uniform stripe of chemokine expression and for the periodic formation and deposition of neuromasts.

Protoneuromast formation is initiated in Fgf–signaling domains that are periodically established in a trailing zone of the migrating primordium in response to Fgfs produced by Wnt–active cells in a leading zone. Local promotion of Wnt activity, coupled with longer range inhibition of Wnt by Fgf–dependent Dkk1b (a protein that inhibits Wnt) expression, is thought to constitute a reaction-diffusion system that determines self-organization of Fgf signaling centers, which initiate formation of nascent protoneuromasts. As previously formed protoneuromasts at the trailing end mature, they produce Wnt–signaling inhibiting factors, which contribute to progressive shrinking of the Wnt system. Progressive restriction of an initially broad Wnt–signaling domain to a smaller leading zone allows new Fgf signaling–dependent protoneuromasts to form in the wake of the shrinking Wnt system.

While initial studies focused on Dkk1b as the primary inhibitor of Wnt signaling produced by newly formed protoneuromasts, a question arose about what inhibits Wnt signaling in the most mature protoneuromasts at the trailing end of the primordium, where dkk1b is not expressed. We showed that Sox2 (a transcription factor that maintains pluripotency and neurogenesis) is expressed in nascent and maturing protoneuromasts in a pattern that is complementary to domains with Wnt signaling activity. Furthermore, Sox2 functions in a partially redundant manner with Sox1a and Sox3 to inhibit Wnt signaling. This helps keep Wnt activity restricted to a progressively smaller leading zone, which we showed is essential for effective maturation of trailing protoneuromasts and the timely deposition of stable neuromasts.

Fgfs secreted by Wnt–active cells in a leading zone determine Fgf signaling activity in an adjacent trailing zone, which initiates protoneuromast formation by promoting expression of factors that promote epithelialization of cells and their apical constriction to form epithelial rosettes. Furthermore, Fgf signaling promotes expression of the transcription factor atoh1a, which gives cells the potential to become sensory hair-cell progenitors, while lateral inhibition mediated by Notch signaling, which inhibits atoh1a expression during neurogenesis, ensures center-biased atoh1a expression, and sensory hair-cell progenitor fate becomes restricted to a central cell in the nascent protoneuromast. However, as the protoneuromast matures in a trailing zone of the primordium, it becomes increasingly separated from leading Wnt–active cells, which are the source of Fgfs that initiated its formation. As a result, effective maturation requires a switch to a new regulatory network that ensures sustained Fgf activity in the maturing protoneuromast, independent of the Fgfs from the leading Wnt–active zone that initiated its formation. This is achieved by the central Atoh1a–expressing sensory hair-cell progenitor, which becomes a new source of Fgfs that sustain Fgf signaling in surrounding protoneuromast cells. In addition, the central Atoh1a–expressing cell determines expression of Notch ligands such as DeltaD, which activate Notch in neighboring cells to promote expression of factors that ensure stabilization of epithelial rosette morphology. Importantly, however, as the Atoh1a–expressing cell in the maturing proneuromast becomes separated from leading Wnt–active cells, which are the source of Fgfs that initiate atoh1a expression, Atoh1a promotes atoh1b expression and sustains its own expression through autoregulation.

Our analysis of Sox2 function has now shown that this critical switch in the genetic regulatory network, where Atoh1a expression becomes self-sustaining through autoregulation with atoh1b, only occurs when Wnt activity is adequately suppressed in trailing cells by Sox2. Failure to adequately suppress and restrict Wnt activity to a leading zone prevents timely initiation of atoh1b expression in trailing protoneuromasts and failure to deposit mature neuromasts with stable epithelial rosettes. Together with past observations, our analysis of Sox2 function now defines three key steps in the periodic self-organization of neuromasts in the primordium: first, a step that polarizes Wnt activity in the primordium; second, a pattern-forming step that generates periodic Fgf–signaling centers in the context of polarized Wnt activity; and third, a step involving Sox2, which helps stabilize fate and morphology in neuromasts formed in the earlier pattern forming stage.

Although the steps outlined above are specific to the way the primordium develops, they define more broadly the sequence of events in the self-organization of development in many different contexts. A symmetry-breaking event that polarizes a tissue is a common first step in most developmental processes, which then provides the context for self-organizing mechanisms that determine patterning of cell fate, morphology, and cell migration within the tissue. Although self-organizing mechanisms could generate pattern spontaneously, independently of tissue polarization, their operation in the context of prior polarization ensures a more predictable reproducible outcome. Furthermore, while a chemical signaling–based reaction-diffusion system has been described above as an example of a pattern-forming system, pattern formation can operate at many scales with different modalities. For example, our previous studies showed that the number and size of rosettes in the primordium can also be understood as an outcome of mechanical interactions between cells, with periodic aggregates forming as a consequence of a balance of forces, such as apical constriction that pull cells together to form rosettes, while tension along the length of the primordium, associated with collective cell migration, pulls cells apart, opposing formation of rosettes. These modalities operate in parallel, interact, and typically work synergistically to determine robust reproducible outcomes. Patterns initiated by such self-organizing mechanisms are not necessarily stable and, as seen in the example of the Lateral Line primordium, are followed by changes in regulation that help consolidate and stabilize form and fate established in an earlier patterning phase.

Lateral inhibition ensures adjacent cells adopt different fates. In mindbomb mutants, loss of Delta-Notch–mediated Lateral inhibition results in a failure of central atoh1–expressing cells to inhibit atoh1a expression by their neighbors. Consequently, too many atoh1a–expressing cells are specified in the migrating pLLp. In addition, poor cohesion in mindbomb mutants results in deposition of unstable neuromasts, and the primordium eventually falls apart. This is related to reduced expression of adhesion molecules and junctional proteins, whose expression is also dependent on Notch signaling. While lateral inhibition ensures that adjacent cells are not alike, cohesion requires adjacent cells to be more like each other and express similar adhesion and junctional molecules. How does Notch ensure both? We showed that, while lateral inhibition, mediated by Delta-Notch, initiates a negative feed-back loop that amplifies differences between cells to ensure distinct fates, lateral induction, mediated by Jagged-Notch signaling, initiates a positive feedback loop that promotes expression of similar adhesion molecules in adjacent cells. Jagged1b, a ligand of the Notch receptor, initiated by Wnt signaling, is expressed broadly but is most prominent in a leading zone of the primordium, where Jagged-Notch signaling promotes cohesive collective migration. Jagged2b expression is initiated by Atoh1a/b–dependent expression of Notch ligands in developing protoneuromasts, where Jagged-Notch signaling promotes stability of epithelial rosettes. However, Jagged1b and Jagged 2b cooperate in their function and their expression is maintained by Notch signaling. A computational model integrating our experimental observations shows how lateral inhibition and lateral induction also play additional complementary roles in restricting atoh1a to a central protoneuromast cell in the context of center-biased atoh1a expression, initiated by Wnt and Fgf signaling interactions. Together our observations in vivo and in silico show how complementary roles of Notch in Lateral inhibition and Lateral induction determine self-organization of cell fate, morphogenesis, and collective migration in the zebrafish pLLp.