Modeling Edar expression reveals the hidden dynamics of tooth signaling center patterning

Abstract

The generation of patterns during development is generally viewed as a direct process. In the mouse jaw, however, the sequential patterning of molars initiates with abortive tooth signaling centers called MS and R2, thought to be vestiges of the lost rodent premolars. Moreover, the mature signaling center of the first molar (M1) is formed from the fusion of two signaling centers (R2 and early M1). Here, we report that Edar expression reveals the hidden dynamics of signalling centers patterning. First, Edar expression evidenced a hidden two-step patterning process that we modelled with a single activator-inhibitor pair: the epithelium is initially broadly activated, then activation becomes restricted in space to give rise to the signalling centers. Second, Edar expression unveils successive phases of pattern making and pattern erasing events, a phenomenon that we called a developmental palimpsest. MS is erased by a broad activation for the benefit of R2, which itself is erased before it recovers when the first molar signaling center forms. In the lower but not the upper jaw, the two neighboring signaling centers then fuse into a single elongated center. Our model recapitulated the erasure of the R2 signaling center by the wave of activation that precedes the formation of M1 signaling center, and predicted the surprising rescue of R2 in the context of an Edar mutant with reduced activation. It suggested that R2 was not intrinsically defective, but actively outcompeted by M1 formation. We confirmed this by cultivating R2 separately from the posterior tissue and showing it could then generate a tooth. Finally, by introducing chemotaxis as a secondary process of tooth germ maturation, we recapitulated the fusion of R2 and M1 in the lower jaw only, and the loss of fusion when Edar function is impaired in organ cultures. In conclusion, we have uncovered a highly indirect and dynamic nature of pattern formation in the molar field that could nevertheless be simulated with simple mathematical models. Our study argues for viewing embryonic patterns as dynamical objects rather than as fixed endpoints, where dynamics is essential to the outcome of the patterning process.