BMP4 regulates embryonic development of taste papillae [41, 42], but its expression and role in adult taste buds has not been investigated. We show here that BMP4, via a genetically engineered LacZ reporter mouse line, is expressed in a subset of cells associated with mature taste buds, but that the pattern of expression differs between the circumvallate and fungiform taste papillae. In the circumvallate, BMP4-ß-gal is expressed within taste buds, by a subset of cells of each taste cell type, as well as by likely immature taste cells. Perigemmal epithelial cells within the circumvallate trenches are also BMP4-ß-gal positive. By contrast, BMP4-ß-gal intragemmal cells are absent from fungiform taste buds; instead, only a small number of perigemmal epithelial cells are BMP4-ß-gal positive. These differing expression patterns suggest that BMP4 expression marks the same perigemmal cell population in both fungiform and circumvallate taste epithelia, but hints at an additional role for this gene product in within taste buds of the circumvallate papilla.
Throughout our study we have equated BMP4-ß-gal expression with that of BMP4 protein. This reporter allele comprises a lacZ knockin to the BMP4 coding region and regulation of transcription of this locus should thus be unaltered. Further, ß-galactosidase expression in this mouse line has been shown repeatedly and in a variety of tissues and developmental stages to accurately reflect BMP4 mRNA expression [44, 45, 48], including faithfully replicating BMP4 mRNA expression in the developing taste bud progenitor cells of mouse embryos [41, 60]. Nonetheless, we were unable to confirm this concordance in adult mouse taste epithelium. We used 3 different BMP4 antisense probes and were never able to detect signal above background in the lingual epithelium, even though one of these probes worked well in our positive control tissue, i.e., hair follicles (data not shown). One explanation is that BMP4 mRNA is expressed at very low levels in adult taste tissue, and is undetectable via in situ hybridization. Given the persistence of ß-gal protein, we suspect that the BMP4LacZ line reports subtle, perhaps more short-lived, BMP4 protein expression patterns that would otherwise be missed. Alternatively, albeit less likely, BMP4 mRNA may be specifically degraded in taste epithelium resulting in the absence of protein, while LacZ transcripts escape this suppression due to differences in mRNA structure. However, given the documented conformity of BMP4 mRNA expression patterns with those of ß-galactosidase expression in this knockin line, we are reasonably confident that the reporter reveals BMP4 expression in adult taste epithelium.
One explanation for the differences in expression patterns and potential roles of BMP4 between circumvallate and fungiform papillae may be due to differences in their embryonic origins . Although in all vertebrate, taste buds arise from local epithelium [62, 63], fungiform and circumvallate taste buds may derive from epithelia of different origins. In rats, keratin 20-IR is distributed in the posterior third of the tongue, suggesting that this region, including circumvallate papillae, derives from endoderm , while the anterior two-thirds of the tongue, in which fungiform papillae reside, is derived from ectoderm (keratin 20-immunonegative) . We have observed rare BMP4-ß-gal expressing cells inside fungiform taste buds, but only in the most posterior fungiform papillae (data not shown), which may possess this expression pattern due to an endodermal origin. These ideas remain to be tested until we obtain a clear understanding of the contributions of endoderm and ectoderm to the epithelium lining the oral cavity.
In fungiform and circumvallate taste buds, BMP4-ß-gal is expressed by a relatively small number of perigemmal epithelial cells located adjacent to taste buds, in the same general location as the progenitor population for taste cells. Stone , using X-inactivation mosaic mice, reported that multiple progenitors give rise to individual taste buds, and these progenitors are assumed to comprise the basal cells located at the base of taste buds, and/or perigemmal cells situated adjacent to taste buds proper [24–26]. Most recently, Okubo  have shown that cells within taste buds arise from K14-expressing basal keratinocytes, which sit along the basement membrane of the lingual epithelium, including that of taste papillae. Taste cells within buds do not express K14, but rather express cytokeratin 8 [52, 64, 65]. In our studies, while BMP4-ß-gal perigemmal cells are located in a position consistent with that of proposed progenitors and many of BMP4 cells were also K14 positive, we could find no evidence that these cells are actively dividing, via both BrdU birthdating and immunostaining for known markers of proliferation. However, it is possible that these BMP4-ß-gal cells divide so infrequently that we simply did not examine enough taste buds from enough mice at a larger range of times, although we did assay 12 mice ranging in age from 2 months to 6 months, and in no case encountered a single BMP4-ß-gal expressing cell that was in any state of proliferation. Intriguingly, another set of BMP4-ß-gal cells in the taste epithelium does not express K14; thus, while these singly labeled cells do reside within the perigemmal taste epithelium, they are not within the progenitor pool identified by Okubo and colleagues . In conclusion, our expression data indicate that the perigemmal BMP4-ß-gal cells in both the circumvallate and fungiform papillae are a heterogeneous epithelial population potentially comprising: (i) a slowly cycling taste bud stem cell population (the K14-immunopositive, BMP4-ß-gal expressing cells); and (ii) a niche population or a signaling center for taste bud stem cells in combination perhaps with nearby BMP4-ß-gal cells of the lamina propria, which together may regulate taste cell genesis from taste bud stem cells.
In circumvallate taste buds, BMP4 is expressed in early differentiating taste cells
Our birthdating analysis with BrdU is consistent with the model that taste cells are born outside of taste buds proper, and then, within a day, move into the taste bud at the margins, and ultimately come to occupy the taste bud core [24, 26]. BrdU is incorporated into dividing cells outside taste buds within 1-6 hours after injection, but BrdU-IR cells are not evident in taste buds until 12-18 hours, mainly in the basal compartment. At later time points, as has been described by others, we find labeled taste cells become more centrally located, progressing toward the taste bud core within 24-72 hours post BrdU injection [27, 33, 65, 66]. This timeframe is consistent with reports where taste cells typically express markers of differentiation by 2.5-3.5 days after birth. For example, expression of BrdU in gustducin-IR cells is first detected 2.5 days after injection [30, 31].
To test the idea that BMP4 marks immature taste cells in the circumvallate papilla, we labeled newly dividing cells with BrdU and followed these cells for 3 days to assess when BMP4-ß-gal expressing fusiform cells are born. As type II cells differentiate at 2.5-3 days after birth, we predicted that, if BMP4 marks a transitional stage from transit amplifying cells to mature taste cells, we would detect BMP4-ß-gal and BrdU co-labeled cells within 2.5 days of BrdU injection. In fact, BrdU is detected in BMP4-expressing fusiform cells beginning at 24 hours post-injection, suggesting that BMP4-expressing fusiform cells are indeed immature taste cells. It is possible, however, that ß-galactosidase in taste buds persists longer than native BMP4 protein would, which in vitro, is known to decay within 6 hours , while ß-galactosidase has a much longer half-life (approximately 13 hours) [44, 49–51]. Thus, BMP4-ß-gal inside taste buds might not accurately reflect BMP4 expression, but rather would represent residual ß-galactosidase that was intially driven by the BMP4 promoter in cells outside of taste buds. If this were the case, the number of BMP4-ß-gal and BrdU-IR double labeled cells inside taste buds should peak within 24 hours after BrdU injection, as the half-life of ß-gal is 13 hours [44, 49–51]. However, we find that the number of BrdU and X-gal double positive cells inside taste buds peaks at 48 hours post-injection, well after the expected decay of ß-gal, indicating that BMP4-ß-gal inside taste buds must instead be due to active transcription and translation of lacZ under the BMP4 promoter within newly generated taste cells in the circumvallate papilla.
There are at least 3 models for cell lineage relationships in taste buds: (i) Taste cells derive from a single lineage, progressing from basal cells to type I, then to type III, and finally to type II cells during maturation ; (ii) Taste buds are composed of at least two cell lineages, each of which produces a subset of differentiated taste cell types [28, 29, 68]; or (iii) all 3 taste cell types arise via distinct lineages . To determine if BMP4 expression is limited to one cell type, and potentially marks one or a subset of taste cell lineages, we examined which cell types within circumvallate taste buds expressed BMP4-ß-gal. We found that a small percentage of each taste cell type (I, II, and III) co-express BMP4-ß-gal, suggesting that intragemmal BMP4 is expressed by differentiating taste cells regardless of cell type. Except for the subset of type II cells immunopositive for gustducin, the percentages of double labeled BMP4-ß-gal and type I, II, III cells (NTPDase2, PLCß2-IR, PGP9.5-IR, serotonin-IR, NCAM-IR) are not statistically different (Tukey test; p > 0.05), tending to support the hypothesis that BMP4 is broadly expressed by differentiating taste cells, regardless of cell type. However, this may not be the case for gustducin-IR cells, which are the subset of type II cells within the circumvallate that transduce bitter taste [11, 12]. As very few double-labeled gustducin-IR cells were detected, it may be that BMP4 is not involved in differentiation of the bitter-sensing cell lineage. The fact that the number of BMP4-ß-gal-expressing cells per taste bud is higher at the bottom portion of the circumvallate trenches suggests the possibility of more rapid turnover of cells in taste buds in the deeper regions of the papilla epithelium, but this idea remains to be tested.
To investigate further our hypothesis that BMP4-expressing cells within buds are immature taste cells, we assessed the expression of a proposed marker for immature taste cells, Sox2 , and compared this pattern to that of BMP4-ß-gal. High levels of Sox2 expression have been found in the progenitor cells that are committed to taste cell lineage . We, too, found that in circumvallate taste buds, a large number of BMP4-ß-gal intragemmal cells are co-immunoreactive for Sox2, providing another piece of evidence that BMP4-ß-gal marks immature taste cells.
What is the nature of the mitotically quiescent BMP4-ß-gal expressing epithelial cells in taste papillae?
It has been proposed that basal and/or perigemmal cells adjacent to taste buds comprise the taste progenitor population, responsible for continual generation of adult taste cells [26, 28, 29, 69, 70]. As in generalized epithelium, taste bud stem cells are thought to undergo asymmetric division to produce a progenitor daughter that goes through transit amplifying divisions to produce a number of immature taste cells [52, 71]. Thus, we hypothesized initially that the BMP4-ß-gal cells were stem cells and/or transit amplifying cells of the taste bud lineage. If so, these cells should express markers of actively cycling cells, including Ki-67 and BrdU incorporation during S phase DNA synthesis. However, double labeling with Ki-67, which labels mitotically active cells in all phases of the cell cycle , showed that these BMP4-ß-gal expressing cells were not actively cycling. We further confirmed this lack of mitotic activity with additional markers of cell proliferation (PCNA and pH3), and in short duration birthdating studies employing BrdU. In sum, these data indicate that BMP4-ß-gal expressing cells are not rapidly dividing transit amplifying cells.
That BMP4-expressing cells found outside of taste buds are not mitotically active suggests at least two possibilities:
1) BMP4-ß-gal-expressing epithelial cells are taste bud stem cells. In general, stem cells in adult tissues divide very infrequently, occupy protected microenvironments or niches, and are predominantly in the resting phase (G0) of the cell cycle [72, 73]. Cells in G0 typically do not express gene products of cell proliferation, and our failure to detect mitotic BMP4-ß-gal cells is consistent with this criterion. In skin, quiescent stem cells are normally intermingled with the transit amplifying population (see review, Alonso and Fuchs, 2003) ; similarly, BMP4-ß-gal cells are located in the basal epithelium surrounding taste buds, where we find that the majority of cells are actively cycling. Finally, Okubo  have recently suggested that multipotent stem cells located in the basal epithelium of fungiform papillae give rise to transit-amplifying daughters, which contribute to both taste buds and keratinocytes of the gustatory papillae. However, these latter studies involve genetic mapping of cells that express K14, while the precise identity of the initially labeled cells was not assessed. In our study, while some BMP4 cells express K14, another subset of perigemmal BMP4-ß-gal cells are not immunoreactive for K14, yet are located in the taste papilla epithelium immediately adjacent to taste buds. This heterogeneity in basal keratinocyte marker expression may reflect different functions for these cell subpopulations. Regardless of K14-immunoreactivity, however, we failed to detect proliferation of BMP4-positive cells.
2) A second possibility is that, instead of a taste bud stem cell population, BMP4-ß-gal-expressing epithelial cells represent a signaling center or compartment of the stem cell niche, which controls stem cell and/or transit amplifying cell division and/or differentiation. Our double-staining experiments for BMP4-ß-gal and proliferation markers revealed that the BMP4-expressing cells in the subepithelial lamina propria are also mitotically inactive, suggesting that in concert, these epithelial and mesenchymal BMP4 cells adjacent to taste buds may maintain the niche, or signaling center for taste bud stem cells. Interestingly, Miura  report that Shh is expressed exclusively in basal cells within taste buds, whereas Ptch1, a Shh receptor (reviewed by Ingham and McMahon, 2001) , is expressed in the epithelial cells outside of taste buds, but adjacent to intragemmal Shh expressing cells. Moreover, mitotic cells, as detected via short term BrdU incorporation are mainly in the Ptch1 expressing region, raising the likelihood that Ptch1 expressing cells comprise the transit amplifying population, and possibly the taste bud stem cells . In both fungiform and circumvallate papillae, BMP4-ß-gal-expressing cells may lie adjacent to the Ptch1 expressing cells, or may express Ptc1 suggesting the possibility that BMP4 and SHH may coordinate taste cell turnover, as has been demonstrated for regeneration of a number of other epithelial appendages, including hair follicles [76, 77].