The initiation knot is a signaling center required for molar tooth development

ABSTRACT Signaling centers, or organizers, regulate many aspects of embryonic morphogenesis. In the mammalian molar tooth, reiterative signaling in specialized centers called enamel knots (EKs) determines tooth patterning. Preceding the primary EK, transient epithelial thickening appears, the significance of which remains debated. Using tissue confocal fluorescence imaging with laser ablation experiments, we show that this transient thickening is an earlier signaling center, the molar initiation knot (IK), that is required for the progression of tooth development. IK cell dynamics demonstrate the hallmarks of a signaling center: cell cycle exit, condensation and eventual silencing through apoptosis. IK initiation and maturation are defined by the juxtaposition of cells with high Wnt activity to Shh-expressing non-proliferating cells, the combination of which drives the growth of the tooth bud, leading to the formation of the primary EK as an independent cell cluster. Overall, the whole development of the tooth, from initiation to patterning, is driven by the iterative use of signaling centers.


Introduction
while remaining IK cells stayed in G1/G0. Quantification of cell cycle phases from E12.5+12h showed 148 that number of molar IK G1/G0 cells decreased slightly (Fig.2E). The bud S/G2/M population 149 continued to expand, leveling out after six hours. This coincided with the appearance of the first G1/G0 150 cells contributing to the pEK. Also, at this stage, proliferation was specific to the tooth bud (Fig.S2D). 151 Shh and Fgf signaling have been suggested to induce proliferation in tooth buds (Hardcastle et al., 152 1998, Cobourne et al., 2009; however, inhibition of Shh signaling at the placode stage was reported  162 We, therefore, conclude that the molar IK is a functional signaling center that regulates proliferation 163 in tooth bud invagination and growth. The molar bud is formed by localized cell proliferation, with 164 the involvement of Shh and FGF signaling.

IK ablation arrests progression of tooth development 166
To confirm that the IK drives molar bud growth and is necessary for progression of tooth 167 development, we ablated the IK at different developmental stages by microsurgery and laser ablation. 168 When the placode was microsurgically removed at E11.5 and the tissue cultured for 24h, no G1/G0 169 condensate was observed in the diastema and the epithelium remained flat (Fig.S4A,C).
Microsurgical removal of the IK at E12.5 similarly arrested tooth growth, while development on the 171 untreated side proceeded to bud stage with the emerging pEK present (Fig.S4B,C). 172 For a more targeted approach we next removed the IK G1/G0 cells at E11.5, E12.5 and E12.75 with 173 laser ablation, followed by 24h culture, in K17-GFP and Fucci whole mount mandibles. Laser

186
These experiments demonstrate that the molar IK is a functional signaling center that drives cell 187 proliferation, thereby regulating tooth bud growth. The IK is, therefore, necessary for the progression 188 of tooth development. Our live imaging experiments showed that IK cells reorganize dynamically during placode/bud maturation. To define the significance of this to IK maturation we quantified IK cell condensation 218 and analyzed if the movements involve active cell migration. 219 We first measured cell density in EpCam stained fixed whole-mount samples. Initially, at E11.5, 220 G1/G0 cells were more dispersed and at E12.5, they had condensed and retained this density until 221 E13.5 (Fig.5A,B). Oral epithelial cells did not show a similar condensation. Quantification of cell 222 density showed that condensation was specific to IK cells, with a significant increase in density from 223 E11.5 to E12.5 compared to the oral epithelium (Fig.5B).

224
To study if IK condensation is achieved through active cell migration, we followed the movement of 225 individual cells by live imaging at E11.5/E12.5+12h. Tracking showed active migration of the molar 226 IK G1/G0 cells at both time points. We quantified the overall track length and net displacement in the 227 different cell populations, and at E11.5+12h, a significant difference in IK G1/G0 cells was observed:

228
They migrated more compared to both oral epithelial G1/G0 and tooth bud S/G2/M cells (Fig.5C). At 229 E12.5+12 h, IK G1/G0 cells still showed a longer mean track length in the bud compared to S/G2/M 230 cells (Fig.5C). Quantification of IK G1/G0 cell displacement angles at E11.5+12h showed a distinct 231 orientation towards the mesio-lingual side of the forming placode/bud, whereas oral epithelial cells 232 showed a random orientation (Fig.5D). We confirmed active migration by following pairs of IK G1/G0 233 and bud S/G2/M cells that were initially in close proximity (distance≤15µm). The pairwise 234 comparison revealed that while many IK G1/G0 cells remained neighbors, 30% of cells switched their 235 partners; most S/G2/M pairs remained neighbors (Fig.5E). Tracing groups of cells that were initially, 236 at E11.5, located next to each other in different areas of the IK, and defining cell positions 12h later 237 showed that some cells remained close to their original neighbors but several cells ended up with a 238 different group. (Fig.5F). Pharmacological inhibition of acto-myosin based motility, with the 239 inhibitor blebbistatin, repressed IK G1/G0 condensation and abrogated progression of tooth 240 development (Fig.5G).

Dynamics between Wnt HI and Shh cell populations regulate the maturation and maintenance 242 of the IK 243
Our live imaging analysis suggested that TCF/Lef:H2B-GFP reporter expressing Wnt Hi cells were 244 closely juxtaposed to Shh expressing G1/G0 IK cells but they seemed to comprise two different cell 245 populations that remained in close contact with each other through bud development (Fig.6 Wnts and also an inhibitor of Wnt signaling via a negative feedback loop (Sarkar, et al., 2000;Sarkar 249 and Sharpe, 1999). We next investigated the behavioral dynamics and the molecular identity of the 250 two populations.

Modulation of canonical Wnt signaling affects IK cell dynamics and tooth bud shape 289
Wnt Hi -Wnt10b cells directing the movement and condensation of the G1/G0-Shh cell population in 291 molars. Wnt10b has been implicated as a paracrine chemotactic factor in epithelial cancer contexts 292 (Chen, et al., 2017;Aprelikova, et al., 2013). To explore if this dynamic occurs in developing molars, 293 we modulated canonical Wnt signaling levels by stimulation with Wnt3a or a Wnt10b releasing bead, 294 and by inhibition with a Wnt antagonist, XAV939 that acts via stimulation of β-catenin degradation 295 and stabilization of axin. E11.5 explants were treated either with Wnt3a/XAV939 in the growth 296 medium for 24h or by placing a recombinant Wnt10b soaked/control bead next to the placode at 297 E11.5 and followed the explants for 16h. 298 We used K17-GFP to visualize the shape of the epithelium and Fucci G1/G0 for IK cell distribution

308
To dissect the role of Wnt10b on IK condensation, we placed a Wnt10b releasing or control bead 309 close to the IK distally on the lingual side of the placode and imaged the explants at E11.5, after 8h, 310 and 16h. While morphogenesis and IK condensation proceeded normally in both the control bud and 311 the bud with the control bead, the bud with the Wnt10b bead showed a loss in condensation of the 312 G1/G0 IK cells (Fig.7F). Instead, the G1/G0 IK cells were spread out toward the Wnt10b bead.
Wnt10b bead buds together with a decrease in G1/G0 IK cell density (Fig.S6A). The changes in IK 315 cell distribution and bud shape were also accompanied with decrease in proliferation (Fig.S6B, C).

317
The reiterative genetic regulation of tooth development via signaling centers is conserved across tooth 318 types, but it is less understood how it is interpreted into different cellular behaviors to regulate tooth 319 shape and size. In the present study we have identified a molar IK signaling center that is necessary

333
Tissue recombination studies have shown that the instructive potential in the tooth first resides in the 334 epithelium and shifts only later to the mesenchyme (Lumsden, 1988;Mina and Kollar, 1987). EKs 335 require inductive signals from the mesenchyme, but it is plausible that the IK inducing signal comes 336 from planar epithelial signaling. Wnts7b/3 and Shh have a mutually exclusive expression already at 337 E10.5 in the presumptive oral and dental ectoderm (Sarkar, et al., 2000;Sarkar and Sharpe, 1999) so 338 it appears that different Wnt expression patterns and Shh determine the ectodermal boundaries of 339 competence at a very early stage. Shh is possibly a downstream target of Wnts and also a negative 340 feedback inhibitor. Spatial inhibition of Wnt10b by Shh has been reported in teeth: Shh coated beads repressed Wnt10b but no other epithelial markers (Dassule and McMahon, 1998). Constitutive activation of epithelial Wnt/β-catenin, somewhat later from E12.5, induced multiple patches of 343 signaling center markers, including Shh and Wnt10b at E13-E14, and ectopic teeth (Liu, et al., 2008;344 Jarvinen, et al., 2006). Our work evidences that Wnt10b and Shh are differentially expressed during 345 molar initiation and that these cell populations remain functionally separate. However, close 346 interaction between the G1/G0-Shh and Wnt Hi -Wnt10b expressing cells is crucial in the positioning 347 and maintenance of the molar IK. Wnt10b has been implicated as a paracrine chemotactic factor in 348 cancer contexts (Chen, et al., 2017;Aprelikova, et al., 2013). The migration of G1/G0-Shh IK cells 349 toward the canonical Wnt gradient and specific area of endogenous Wnt10b expression and 350 distribution of IK cells toward exogenous source of recombinant Wnt10b, suggest that Wnt10b carries 351 an instructive role in signaling center condensation. 352 We show that molar invagination and growth take place through cell proliferation in both basal and 353 suprabasal bud cell populations, driven by the non-proliferative IK. Several signaling pathways 354 regulate behaviors in these cell populations, but the role of Hh signaling in this context has been 355 debated. Shh has been interpreted to be a primary inducer of proliferation in some experimental 356 settings, whereas other studies suggest a role in bud cell rearrangement (Li, et al., 2016b;Prochazka, 357 et al., 2015;Cobourne, et al., 2009;Hardcastle, et al., 1998). Shh expression is a hallmark of signaling 358 centers, and while autocrine signaling cannot be ruled out, most of the responsive cells seem to reside 359 elsewhere: at later stages, from E14.5 onwards, the pEK expresses Shh but receptor Ptch and Shh signaling at the placode stage resulted in loss of proliferation in the bud. Thus, it seems that in the IK, Shh expressing cells are different from Shh responsive cells and signaling can induce proliferation. In agreement, early findings from conditional Shh mutants show smaller teeth and 368 posteriorly misplaced buds (Dassule, et al., 2000;Dassule and McMahon, 1998). In the limb bud, 369 Shh has been reported to affect proliferation both directly and indirectly via induction of Fgfs in the 370 AER (Prykhozhij and Neumann, 2008;Towers, et al., 2008). In the tooth, other factors downstream 371 of the initial placodal inducer Shh, such as Fgfs, likely contribute to bud invagination via proliferation 372 mainly in basal cells, and possibly concurrent with stratification (current study, (Li, et al., 2016b).

373
Expression levels of Shh may specifically affect signaling center identity, differentiation, and 374 maintenance. Shh overexpression arrests development at the bud stage due to lack of proliferation, 375 and multiple superficial invaginations are induced in the epithelium but still fail to develop further 376 (Cobourne, et al., 2009). It is tempting to hypothesize that exogenous Shh expression would drive 377 cells into abnormal cell cycle exit and/or a change in fate into a signaling center. Signaling center 378 maintenance may also be associated with Shh expression levels. Shh has been shown to be protective 379 of early apoptosis in the tooth (Cobourne, et al., 2001). Apoptosis is a mechanism used to silence 380 signaling centers in teeth, the AER of the limb, and in embryonic brain development (Nonomura, et 381 al., 2013;Matalova, et al., 2004;Vaahtokari, et al., 1996b). The interplay between Wnt Hi and Shh+ 382 cells may serve as a feedback mechanism regulating the timing of apoptosis in the IK. 383 We demonstrate that the pEK in the molar is formed de novo without clonal contribution from the 384 IK, and that the IK is apoptotically silenced upon pEK appearance. This differs mechanistically from 385 signaling centers later in molar development, where the pEK contributes cells to sEKs (Du,et al.,386 2017). The development of teeth is conserved, however, in being driven by the iterative use of 387 signaling centers. We have shown functionally here that the progression of early molar 388 morphogenesis is dependent on the IK signaling center that arises in the placode and exhibits many 389 hallmarks of ectodermal signaling centers. What differentiates the IK from the later signaling centers 390 on a transcriptomic level will be of special interest to future studies.           specific for Shh or Wnt10b performed as described previously (Shirokova, et al., 2013;Fliniaux, et al., 2008;Wang and Shackleford, 1996). Imaging of the hybridization signal was done with the same Zeiss Lumar microscope and Zeiss AxioCam ICc1 CCD camera and images were transposed. 663 For 3D time-lapse imaging dissected tissues were allowed to recover for a minimum of 2h prior to 664 imaging. The explants were imaged as described previously (Ahtiainen, et al., 2016;Ahtiainen, et al.,