Shh expression reveals a sequential order for ruga appearance
In late embryos (ED16.0), the Shh gene was expressed in rugae and its expression pattern closely mimicked the adult rugae pattern (compare panel C with panel A in Fig. 1). By looking at earlier stages, we found that the striped expression pattern of Shh appeared from very early during development (Fig. 2). As seen on sections, the stripes always corresponded to epithelial cells at the tip of rugae, that at least had reached stage I, i.e. thickened epithelium (for examples, see sections later in figure 5). We reasoned that we should be able to determine the order of appearance of Shh stripes by simply looking at a series of staged embryos. For this purpose we needed a reliable series of closely consecutive developmental stages, and to achieve this, we used the staging method proposed by Peterka et al. [15], in which embryo weight (in mg) allows a more precise specification of the developmental stage of embryos exhibiting the same chronological age (in ED = Embryonic Day). Then, starting with ED16.0 and looking back in developmental time to the first rugae initiation, we could identify Shh stripes according to numbering of rugae in adults. This method allowed us to trace the developmental fate for each ruga. For clarity however, we present here the results according to progressing developmental time (Fig. 2).
From very early stages and then on (ED12.0–16.0), the sharpest stripe of Shh expression corresponded to ruga 8 (shown in red, Fig. 2). Ruga 8 seemed to be the first ruga to appear, but it was very rapidly followed by two stripes corresponding to ruga 2 (in green, Fig. 2A) and ruga 9. From then on, new Shh stripes emerged between the last formed ruga and ruga 8. For example, ruga 3 emerged between rugae 2 and 8 (Compare Fig. 2B and 2C), ruga 4 between rugae 3 and 8 (Fig. 2D to 2F, note Shh expression was just beginning in Fig. 2E), ruga 5 between ruga 4 and 8 (Fig. 2F to 2H, note Shh expression was just beginning in Fig. 2G), and so on, until the last ruga was formed, i.e. ruga 7b (Fig. 2K to 2L; please note that we called this ruga "7b", instead of "8b" as in previous studies, as it was more consistent with order of appearance [3, 4]). The only exception was the stripe corresponding to ruga 1 which emerged anteriorly to ruga 2 and almost contemporary with ruga 3 (Fig. 2C to 2E).
In conclusion, this analysis of Shh expression allowed us to define a temporal sequence for ruga appearance that is 8-(2,9)-(1,3)-4-5-6-7-7b. Thus, rugae 3 to 7b are consecutively interposed between the last formed ruga and ruga 8. This process, that we call "interposition process", is accompanied by a clear anteroposterior extension of the region between ruga 2 and 8, by comparison to the region between ruga 8 and 9 (compare Fig. 2A and 2L). Moreover, it closely follows the anteroposterior elongation of the whole jaw (see drawings of embryo heads on Fig. 2). During the rest of our study, we chose to focus on the "interposition process" of rugae 3–7b in front of ruga 8, because of the very interesting reiterative aspect of this process and because of its implications for palatal A/P growth.
During the interposition process, new rugae are added in the region immediately posterior to the last formed ruga
As the increasing rugae number results from reiterated steps of ruga interposition, we thought that a detailed description of one of these steps should give us a general understanding of the process. We decided to focus on ruga 5 interposition at ED13.5. This allowed investigation of rugae at different stages (ruga 5 being the youngest) and avoiding potential interference with the horizontalization process (which starts at ED14.0).
We first chose three embryos at three close developmental stages, which corresponded to ruga 5 interposition (ED13.5; embryos I, II, III in Fig. 3A). By counting mitotic and non-mitotic epithelial cells and measuring epithelium thickness on every two second frontal sections of the palatal shelf, we could determine the mitotic index (i.e. a number of mitotic cells per 100 cells) along the A/P axis of the palatal shelf, that is, in gap areas (white in Fig. 3A), in rugae areas (dark grey) and in the region of interest, anterior to ruga 8 (light grey). We found a higher mitotic index in the gap than in the rugae areas (see the gap between ruga 2 and ruga 3 or between ruga 3 and ruga 4, Fig. 3A). Moreover, the mitotic index of ruga 4 progressively decreased from embryo I, where ruga 4 was only poorly developed, to embryo III, where it was fully developed, suggesting that proliferation is progressively decreasing as ruga outgrowth is completed.
In the region anterior to ruga 8 (light gray), the epithelium was only a little bit thicker than in gap areas, and curiously, the mitotic index was low, except for embryo II. In this case however, we noticed that the repartition of mitosis was strongly biased in favor of the anterior third of the region (see the mitotic index in red). This might correspond to the formation of ruga 5, since in the older embryo III, ruga 5 thickening was now discernable just next to ruga 4. In conclusion, ruga 5 seems to form from a burst of proliferation in the region immediately posterior to ruga 4, while the region immediately anterior to ruga 8 tends to proliferate at lower level. A true gap between ruga 4 and ruga 5 is only formed later on (see later at ED14.0 on figure 5).
In order to confirm these results, we used the Ki67 antibody to label head sagittal cryosections of embryos showing similar weight to the embryo III (Fig. 3B–C). This antibody is known to label cells engaged in the cell cycle, but not quiescent cells. Moreover it discriminates interphasic cells (discrete nucleolar signal) from mitotic cells (very sharp signal), and can thus be used as an indicator of proliferation activity. Results were consistent with the quantitative evaluation of mitotic index. Quiescent cells (non labeled cells) were found mostly in rugae, thus explaining their overall low mitotic index (Fig. 3B). The forming ruga 5 had more mitotic cells as compared with other rugae, in agreement with its higher mitotic index. Finally, it seems that the gap region just in front of ruga 8 had more quiescent cells than other gap regions (compare with the gap between ruga 3 and 4, white brackets), in agreement with a low mitotic index.
To complete these results obtained on sagittal sections, we used a complementary, whole mount approach. We performed whole-mount immunochemistry with Ki67 antibody and visualized the surface epithelium with confocal microscopy (Fig. 3D–F). Post-staining with DRAQ5TM allowed visualization of cell nuclei. Therefore the rugae could be well detected, possibly because cells are more densely packed there (Fig. 3D). This pattern closely resembled that seen through Shh expression in a littermate embryo of similar weight (Fig. 3F). As judged from the distance between ruga 4 and ruga 8, the stage seemed equivalent to that of "embryo II" in Fig. 3A. Results were again fully consistent with counting of mitotic index: unlabeled, quiescent, cells were mostly found in rugae and in the region next to ruga 8; on the other hand labeled, proliferating, cells were mostly found in gaps and in the region next to ruga 4, where ruga 5 will form (compare the two white arrows in Fig. 3E). We concluded that epithelial proliferation is high in all gaps, except in a quiescent zone just anterior to ruga 8. In contrast, proliferation is low in rugae, except in the new ruga, which seems to form from an active zone just posterior to lastly formed ruga.
Observations on wild type and EdaTamutant strains suggest that formation of a new ruga is only allowed at a distance from flanking rugae
In a previous study, we studied variations of the ruga pattern in EdaTa mice and in their wild type counterpart [3]. These variations had drawn our attention because they revealed morphological correlations between consecutive rugae.
First, in the wild-type sample we studied, ruga7b was highly polymorphic, as reported for other wild type backgrounds [4, 20]. It was either absent, short or fully formed (respectively in 9%, 56% and 35% of palatal halves Fig. 4A). As shown above, ruga 7b forms last, between ruga 7 and ruga 8 (Fig. 2). We found that its presence was associated with a higher distance between ruga 7 and 8 (Fig. 4A). A clear threshold in distance was systematically exceeded in half palates with ruga 7b and never reached in half palates without 7b (Fig. 4A). Moreover, whether ruga 7b was fully formed was also clearly correlated with higher distance between ruga 7 and 8 (Fig. 4A). Finally, variations in medio-lateral position and shape of ruga 7b also correlated with variations of the distance between ruga 7 and 8 (arrows in Fig. 4A). In summary, presence of ruga material correlated with exceeding a threshold distance between ruga 7 and 8.
These observations suggest a model where formation of a new ruga "n+1" somehow depends on the distance to its neighboring rugae at the time it forms, i.e. rugae "n" and 8. If true, then we expect that any mis-positioning of ruga "n" should be propagated to ruga "n+1" (Fig. 4B). We had seen this kind of correlations in the EdaTa mice, in which patterning anomalies are frequently found in adults [3]. These anomalies occur non symmetrically and include ectopic rugae, short rugae and mispositioning disrupting symmetry. From then on, our purpose was not to decipher how these anomalies occur, but whether and in which way an anomaly on ruga "n" is linked with an anomaly of ruga "n+1". According to our putative model, several predictions can be made depending on the type of anomaly in ruga "n" (Fig. 4B), which can be tested in EdaTa mice. In order to detect even subtle patterning anomalies, we chose to look at the Shh pattern in late embryos (ED14.5–15.0) rather than at the adult ruga pattern.
In a model where ruga formation is only allowed at a distance from preexisting rugae (i.e. out of grayed regions in Fig. 4B), an abnormal bending (towards posterior or anterior) of ruga "n" should be propagated to ruga "n+1". For example, ruga "n" bending towards posterior should repel ruga "n+1" formation (both in space and time: until ruga n to ruga 8 distance is wide enough); on the contrary, ruga "n" bending towards anterior should precipitate ruga "n+1" formation. Both cases were found in tabby mice (Fig. 4D–E; yellow and green arrows, compare with the normal pattern in 4C). Then, the model predicts that a shortening of ruga "n" should result in bending of corresponding extremity of ruga "n+1" towards the free place anterior (Fig. 4B). In EdaTa mice, shortening of ruga 5 was indeed followed by a bending of ruga 6 towards the anterior (Fig. 4F–G; blue arrows). Finally, a small patch of ectopic ruga posterior to ruga "n" should repel ruga "n+1" formation and result in ruga "n+1" posterior bending. Alternatively, such an ectopic patch may form after its flanking rugae, because at this site, the two of them are abnormally distant and consequently fail to prevent ruga formation. Whatever the case, ectopic patches should be found associated with ruga bending (Fig. 4B). In EdaTa mice, small ectopic patches were eventually found between ruga 2 and 3 (Fig. 4D, 4E, 4H; pink arrows) and ruga 4 and 5 (Fig. 4D, 4H, 4I; pink arrows). In each case, the lateral extremity of ruga 3 and ruga 5 were abnormally bent towards the posterior (compare with Fig. 4C which represents the normal pattern). In conclusion, all these observations are consistent with a model where ruga interposition is only allowed at a certain distance from the flanking rugae.
Ruga 8 marks a boundary of gene expression in the developing palate
As shown by the previous observations, ruga 8 has a very special position in the palatal shelf and we wondered if it may delimit different territories in the developing palate. We thus looked at expression of three genes, Shox2, Meox2 and Tbx22, that are known to be differentially expressed along the anterior/posterior axis [12, 14, 21, 22]. Interestingly, whatever the embryonic stage we looked at (ED12.5, ED13.5, ED14.0), Shox2 was found expressed in the mesenchyme anterior to ruga 8 (including the mesenchyme directly adjacent to ruga 8), while Meox2 and Tbx22 were always found expressed in the mesenchyme just posterior to ruga 8 (Fig. 5). Thus, the limit of expression of Shox2, Meox2 and Tbx22 genes in the underlying mesenchyme coincides with ruga 8 at consecutive embryonic stages, at least before palatal shelves elevation. Taken together, these data suggested that ruga 8 marks a developmental boundary delimiting two distinct parts of the palatal shelves.
The golden hamster ruga 7 is equivalent to mouse ruga 8 and marks the same boundary of gene expression
In order to further explore the possibility that mouse ruga 8 indeed represents an important developmental boundary, we looked at rugae development in a distant murid, the golden hamster (Mesocricetus auratus). Its posterior palate morphology is quite different from the mouse (Fig. 6A). This species harbors only 7 obvious rugae and a smooth eighth one bordering the soft palate. As for the mouse, we used a probe against the Shh gene (ma Shh) to determine the order of rugae appearance. Due to difficulties in detecting pregnant females at young embryonic stages, we could not determine the order of appearance for the earliest rugae, and our series started with embryos having already 5 Shh-stripes corresponding to rugae 1,2,3,7,8. From then on, rugae were progressively added by interposition between the latter formed ruga and ruga 7 (Fig. 6B), resulting in the partial sequence (1,2,3,7)-4-5-6. Ruga 7 in hamster is thus functionally equivalent to ruga 8 in mouse, and we called these very special rugae the "boundary rugae".
As compared with the mouse, the boundary ruga is found relatively more anteriorly than in the mouse (Fig. 6C). Nevertheless, in hamster as in mouse, the boundary ruga corresponded to the boundary between Shox2/Meox2 expression in embryos of weight class 190–200 mg (Fig. 6D, and see sections in Fig. 6E). We concluded that, since it is found in two species with different ruga number and pattern, the overlap between the boundary ruga and the Shox2/Meox2 boundary is unlikely to be fortuitous. Rather, it strongly suggests that the boundary ruga marks a conserved developmental boundary in the murid palate.