FGF10 maintains stem cell compartment in developing mouse incisors

During embryogenesis, a single fertilized oocyte gives rise to a multicellular organism whose cells and tissues adopt differentiated characteristics or fates to perform the specified functions of each organ of the body. Many tissues and organs maintain a process known as homeostasis: as cells die, either by apoptosis or by injury, they are replenished. Additionally, the epidermis, hair, small intestine and hematopoietic system are all examples of adult tissues that exist naturally in a state of dynamic flux. Even in the absence of injury, these structures continually give rise to new cells, and are able to divide transiently, terminally differentiate and die. These regenerative tissues are maintained by adult stem cells, which have both the capacity to self-renew, to divide and create additional stem cells, and also to differentiate along a specified molecular pathway. In recent years tremendous advances have been made in our understanding of identification of adult stem cells (reviewed by Fuchs and Segre, 2000). However, molecular mechanisms for establishing adult stem cell compartments during embryogenesis have not been examined in detail.

The mouse incisor tooth is an excellent model for analyzing certain aspects of stem cell regulation and function (Harada et al., 1999). The continuous eruption of mouse incisors throughout the lifetime of an animal is maintained by the division of germs localized in the cervical loop of the apical region. We have previously shown that stem cells divide rarely and asymmetrically, and one daughter cell remains as an undifferentiated stem cell in the cervical loop, whereas the other cell moves to the incisal end, giving rise to transit-amplifying cells. Dental epithelial stem cells in the cervical loop give rise to four cell-lineages – inner enamel epithelium (ameloblast cell-lineage), stratum intermedium, stellate reticulum and outer enamel epithelium.

The ability to select among multiple terminal differentiation pathways and to generate cells that can migrate in the selected direction are characteristics that parallel those of the hair follicle bulge or of crypt cells of the small intestine (Rochat et al., 1994; Fuchs and Segre, 2000). These stem cells seem to rely upon mesenchymal cues for their survival and differentiation (Korinek et al., 1998; DasGupta and Fuchs, 1999; Millar et al., 1999). Epithelial-mesenchymal interactions play an essential role in regulating a wide variety of developmental processes, including those of the teeth (Thesleff and Sharpe, 1997; Peters and Balling, 1999; Hogan, 1999). Tissue recombination experiments have shown that the mesenchymal tissue controls advancing morphogenesis after initiation of tooth development and regulates the continuous growth of epithelium in the mouse incisor (Koller and Baird, 1969). Fibroblast growth factors (FGFs) play a crucial role in the developing tooth (Kettunen and Thesleff, 1999). Our recent studies have shown that mouse incisor mesenchymal cells express Fgf3 and Fgf10, and their receptors, Fgfr1b and Fgfr2b, are expressed throughout the dental epithelium (Kettunen et al., 1998; Harada et al., 1999). Furthermore, bead-implantation assays have shown that FGF10 signaling from the mesenchyme indirectly regulated cell division and fate decision of epithelial stem cells by modulating the Notch pathway in the cervical loop via stimulation of lunatic fringe expression. The involvement of FGF10 in tooth morphogenesis was further indicated by the hypoplastic tooth organ in FGF10 null mice (Ohuchi et al., 2000).

In the present study, which focuses on mesenchymal molecular cues for forming the stem cell compartment during organogenesis, we used the developing mouse incisor germ cell as a model of regenerative tissues. To further examine the role of FGF10 during incisor development, we have designed two loss-of-function experimental lines. Analysis of developing incisor germ of Fgf10-deficient mice showed that FGF10 is not involved in the early signaling networks that regulate tooth initiation at early morphogenesis, but is involved in the establishment of adults stem cells. Other experiments using neutralizing antibody provided direct evidence that FGF10 prevents apoptosis in the stem cell compartment. We propose a model that demonstrates the formation of adult stem cells during organogenesis and shows that stem cells are maintained as undifferentiated cells by mesenchymal signals.

Heads from Fgf10^–/– mice were obtained at desired embryonic stages from intercrosses of heterozygous breeding pairs. Genotyping was performed as described previously (Sekine et al., 1999). Tissues were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS), dehydrated in graded ethanol, embedded in paraffin and sectioned. The sections were stained with Hematoxylin and Eosin.

The sections were incubated with rabbit anti-keratin (L-1824, DAKO, Carpinteria, CA) as a primary antibody for 20 minutes at room temperature. The anti-rabbit Vectastain ABC-kit (PK-6101, Vector Laboratories) and the chromogen DAB (Vector Laboratories, Burlingame, CA) were used for antibody detection. The sections were then counterstained with Hematoxylin.

Culture methods were as described previously (Harada et al., 1999). Dissected incisors from the mandible of Fgf10^+/+ or Fgf10^+/– (n=8) and Fgf10^–/– mice (n=7) at E19 were cultured for 20 days. All pictures were digitized and the growth length of dental epithelium was estimated at intervals of about 5 days by using NIH image 1.62 (public domain program, US National Institute of Health). Averaged growth rate was plotted using Microsoft Excel 98 (Microsoft Corporation, Redmond, WA).

For culture of the cervical loop, the mesenchyme or tissue recombinants the incisors were incubated for 4 hours in 2% collagenase (GibcoBRL) in DMEM at 4°C, and the epithelia were mechanically separated from mesenchyme in PBS. The cervical loops were cut from the epithelial sheets with 26G needle. On tissue recombination experiments the cervical loop was placed in contact with mesenchyme. These explants were cultured for 48 hours on the filters coated matrigel (BD Biosciences, MA). The cervical loops were cultured in the medium supplemented with human recombinant FGF10 (R&D systems Inc, Minneapolis, MN) varying in concentration (0-10 ng/ml). Heparin acrylic beads (Sigma) that released FGF10 (25 ng/μl) were placed in cultures of the cervical loop or the apical end of the mesenchyme. Control beads were incubated in bovine serum albumin (BSA, Sigma). The preparation of the beads has been described in detail (Kettunen et al., 1998). For antibody inoculation, anti-FGF3 antibody (SC-135, Santa Cruz Biotechnology, Santa Cruz, CA), anti-FGF10 antibody (SC-7375, Santa Cruz Biotechnology, Inc) and control goat IgG (SC-2028, Santa Cruz Biotechnology) were added to the culture medium at a concentration of 50 μg/ml.

Cell proliferation was localized by mapping the distribution of S-phase cells. BrdU (1.5 ml/100g, 5-bromo-2′-deoxyuridine, Roche Diagnostics, Mannheim, Germany) was administered to ether-treated mice, which were sacrificed 2 hours later, and incorporated BrdU was detected immunohistochemically by the BrdU Staining Kit (Oncogene Research Products, Darmstadt, Germany).

For in situ hybridization analysis, digoxigenin-labeled antisense riboprobes for Fgf3 and Fgf10 mRNA were synthesized according to the manufacturer’s instructions (Roche Diagnostics). Paraffin sections were made as described previously (Kettunen and Thesleff, 1998; Kettunen et al., 1998). The chromogen 4-nitro blue tetrazolium chloride/5-bromo-4-chloro-3-indolyl-phosphate (NBT/BCIP, Roche Diagnostics) was used for color detection.

We used Alexa 488-conjugated Annexin V (Molecular Probes, Eugene, OR), which binds to phosphatidylserine moieties that become exposed on the outer surface of the cell membrane at the onset of apoptosis, to detect apoptotic cells in the explants. The explants were incubated with Hepes buffer (10 mM Hepes, 150 mM NaCl, 5 mM KCl, 1 mM MgCl, 1.8 mM CaCl₂) containing 1 μg/ml Annexin V-Alexa 488 for 20 minutes. After rinsing with Hepes buffer, fluorescence was detected by fluorescent microscopy.

A mouse incisor exhibits special developmental systems that allow for regeneration, a capacity not seen in molars. The first morphological sign of incisor development is a thickening of the oral epithelium (Fig. 1, E11), which subsequently buds into the underlying mesenchyme (E13, bud stage). The mesenchymal cells condense around the bud, and during the following cap stage the tooth bud is lain down longitudinally in the mandible (E14, cap stage). The labial epithelium, which produces differentiated ameloblasts, grows longer in length than the lingual epithelium and the cervical loop, which is a region thought to contain a population of self-renewing epithelial stem cells (Harada et al., 1999) formed at the end of the epithelium (E16, early bell stage). Subsequently, unlike in the molar, the incisor germs continue to grow permanently without forming a root (E19.5, late bell stage). Elucidating the molecules that orchestrate specific developmental programs for stem cells is important for determining the key components that regulate stem cells.

We studied the role of FGF10 in mesenchymal signaling by analyzing the development of mouse incisors in Fgf10-deficient mice. Morphological defects were not found before E14 (Fig. 2A,C). Tooth germs in Fgf10^+/+, Fgf10^+/– and Fgf10^–/– mice developed normally until bud, cap and bell stage. However, at E16, the cervical loop of FGF10 null mice was smaller than that of wild type (Fig. 2B,D). At later stages, the cervical loop was missing (Fig. 2E-H). We examined the cervical loop defect in detail by immunostaining with anti-cytokeratin (CK) antibody. The cervical loop consists of a core of stellate reticulum cells surrounded by basal epithelial cells contacting the dental mesenchyme. Within the cervical loop of Fgf10^+/+ or Fgf10^+/– mice, the center of the stellate reticulum was CK positive, whereas the remaining cells were CK negative. Notably, the peripheral stellate reticulum cells (putative stem cells), which are located in close proximity to basal epithelial cells, were CK negative (Fig. 2I,J, a smaller asterisk). However, in FGF10 null mice, although a CK-positive stellate reticulum remained (Fig. 2L, arrowheads), the basal epithelium and CK-negative stellate reticulum cells were missing, verifying the absence of the incisor stem cell population. However, a normal cell-differentiation gradient of enamel-forming ameloblasts and dentin-forming odontoblasts was observed in both incisors and molars of FGF10 null mice (Fig. 2E,G,M,N).

To characterize the timing of the morphological inconsistency between the mutant and wild type, we examined the gene expression pattern of Fgf3 and Fgf10 in the developing incisors by in situ hybridization. At E14, Fgf3 and Fgf10 mRNA were co-expressed in the dental papilla (Fig. 3A,B) suggesting a functional redundancy between the two FGFs in early incisor morphogenesis. After E16, however, cells expressing Fgf10 mRNA spread from the zone neighboring the inner enamel epithelium to the apical end and surrounded the cervical loop, and this expression pattern appeared to correlate with the growth of labial epithelium (Fig. 3C). By contrast, cells expressing Fgf3 mRNA remained only in dental papilla neighboring the inner enamel epithelium (Fig. 3D). Fgf7 mRNA was weakly expressed in the dental follicle at the apical end and not seen in dental papilla (Kettunen et al., 2000). Fgfr1b and Fgfr2b are expressed throughout the dental epithelium (Kettunen et al., 1998; Harada et al., 1999). These characteristic patterns do not change after birth (Harada et al., 1999). Thus, the timing of cervical loop formation correlates well with the divergence of Fgf10 and Fgf3 gene expression patterns.

To determine the role of existing stem cells, we examined the growth of epithelium that lacked a stem cell compartment. As Fgf10-deficient mice die soon after birth, owing to an absence of lung development (Sekine et al., 1999), we cultured incisor germs dissected from the mandible of E19 mice. The dental epithelium of both mutant and wild-type mice seemed to grow steadily for 20 days (Fig. 4A). However, as shown in Fig. 4B, the epithelial growth rate in mutant mice began to decrease and reached a maximum of approximately 400 μm (Fig. 4B). By contrast, the epithelium of wild-type mice continued to grow. When these explants were cultured beyond 20 days, the explants of mutant type showed disruption of the dental epithelium. Consequently, the length could not be estimated precisely.

To investigate difference of epithelial growth pattern between wild type and mutant or between incisor and molar, BrdU labeling analysis was used to detect the dividing cells of dental epithelium in E18 mice (Fig. 5A) and the percentage of BrdU-positive cells was quantified (Fig. 5B). Labeled cells were detected in the dental epithelium of wild type mice in the cervical loop and extended into the inner enamel epithelial zone (Fig. 5A, parts a and b). The percentage of BrdU-positive cells was 24% on average in the cervical loop and 68% in the inner enamel epithelium (Fig. 5B). In the cervical loop the positive cells were located in peripheral stellate reticulum showing putative stem cell population (Harada et al., 1999). The characteristic scattering pattern of positive cells observed in the cervical loop was clearly distinct from the pattern of dense labeled cells in the inner enamel epithelium. In the mutant incisor, the percentage was high (63%) in the apical end of dental epithelium and was nearly identical (68%) to that in inner enamel epithelium of wild type. But, the labeling pattern seen in the cervical loop of the wild type was missing in the epithelium of mutant type (Fig. 5A, parts c and d). The inner enamel epithelium of molar germs showed high percentages (88%, 67.5%) in both wild type and mutant (Fig. 5B), and their patterns of labeled cells were dense (Fig. 5A, parts e and f). These data suggest that a stem cell population does not exist in the dental epithelium of FGF10 null mice and molar germs. As the growth of epithelium that lacks stem cells depends only on proliferation of transit-amplifying cells, which undergo terminal differentiation after several rounds of cell division, the growth of mutant incisors was expected to be limited (Fig. 4B). Furthermore, cell death in the dental epithelium of the mutant involved limit of cell division of transit-amplifying cells (Fig. 4C). Thus, continuous growth would depend on stem cell division induced by mesenchymal FGF10.

To clarify the role of FGF10 in the dental epithelium of incisors, we designed an in vitro loss-of-function experiment using anti-FGF10 neutralizing antibody. When the incisor explants were cultured for 48 hours in the presence of anti-FGF10 antibody, destruction of the cervical loop was observed (Fig. 6A,B). By contrast, in the culture with anti-FGF3 antibody, anti-FGF10 antibody and human recombinant FGF10, control goat IgG at an equal concentration to anti-FGF10 antibody, the cervical loop epithelium maintained a morphological feature (Fig. 6E,H,K). To identify the extent of cell death in the explants, we visualized apoptotic cells using whole-mount Annexin V staining. In the culture with anti-FGF10 antibody, most epithelial cells in the cervical loop reacted with Annexin V (Fig. 6C). Interestingly, apoptotic cells were observed in the neighboring mesenchyme, suggesting that the survival of mesenchymal cells was associated with that of epithelial cells. In the presence of anti-FGF3 antibody, apoptosis occurred in the inner enamel epithelial and the neighboring mesenchymal cells (Fig. 6F). Notably, few apoptotic cells were seen in the cervical loop or the surrounding mesenchyme. Furthermore, human recombinant FGF10 rescued the explants from the destruction of the cervical loop, which was caused by the FGF10 neutralizing antibody (Fig. 6G-I). In the presence of control goat IgG, apoptotic cells were rarely seen (Fig. 6L).

We also determined whether FGF10 protein could prevent cell death in the cervical loop epithelium. The isolated cervical loops were cultured in the medium supplemented with recombinant human FGF protein varying concentration (0-10 ng/ml). The cervical loops were disrupted within 48 hours (Fig. 7B,F) in a concentration of 0.1 ng/ml or less and Annexin V bound to many of the cells (Fig. 7B′,F′). By contrast, in the culture supplemented with FGF10 (1 or 10 ng/ml), the epithelium maintained its tight appearance (Fig. 7D,G) and few cells bound Annexin V (Fig. 7D′,G′). FGF10-releasing beads also rescued the destruction of the cervical loop epithelium (Fig. 7H′). Control beads had no effect on maintaining the morphology or escaping apoptosis (Fig. 7I′). The neutralizing anti-FGF10 antibody (50 ng/ml) disturbed the effects of FGF10 protein (1 ng/ml) as a survival factor. These results indicated that FGF10 plays a role in the maintenance of the stem cell compartment in the developing incisors.

The apoptosis of some mesenchymal cells in the presence of anti-FGF10 antibody suggested that it might be due to absence of epithelial signaling (Fig. 6C). To examine whether absence of the cervical loop cause apoptosis of the neighboring mesenchyme, we had the tissue recombination experiments and the culture of mesenchyme in absence of the dental epithelium. When the isolated mesenchyme and cervical loop were placed in contact on the filter and cultured for 2 days, apoptotic cells were rarely seen in the mesenchyme around the cervical loop (Fig. 7L, asterisk). However, the mesenchymal cells distant from the cervical loop underwent apoptosis. A few apoptotic cells were also seen in the cervical loop. When the isolated mesenchyme was cultured, a large number of apoptotic cells were seen (Fig. 7N,N′). FGF10-releasing beads were not effective in preventing the isolated mesenchyme from undergoing apoptosis (Fig. 7O,O′).

Mouse incisors erupt throughout the animal’s lifetime through the renewal of dental epithelium produced from the cervical loop located in the tooth apex. Slowly and asymmetrically dividing Notch1-expressing stellate reticulum cells, located near the lunatic fringe-expressing basal epithelial cells, have previously been identified as putative stem cells. As molecular signaling during development has been well studied (http://bite-it.helsinki.fi), the incisor presents an excellent model for the study of stem cell niche formation, stem cell division and fate decision mechanisms.

The cervical loop, including putative stem cells, is not formed in developing incisors of FGF10-null mice. The present study indicates that mutant incisors lacking the cervical loop lose the ability to grow continuously. In situ hybridization of developing incisor germs has revealed differential expression of Fgf10 and Fgf3. After E16 and in conjunction with the growth of dental epithelium, cells expressing Fgf10 mRNA spread out from a region underlying the inner enamel epithelium to the apical end of the tooth germ and overlaid the cervical loop. By contrast, cells expressing Fgf3 remained restricted to the regions adjacent to the inner enamel epithelium. Thus, FGF10 rather than FGF3 is responsible for the formation of the cervical loop. Furthermore, our loss-of-function experiments using neutralizing antibodies verify that FGF10 plays a critical role in formation and maintenance of the stem cell population during organogenesis of mouse incisors.

In contrast to molars, mouse incisors exhibit specific developmental systems that allow for regeneration. The putative stem cell compartment of incisors, the cervical loop, is not found in molar germs. In molars, the dental epithelium (Hertwig’s epithelial root sheath) migrates downwards to the root base. At this point, the dental epithelium loses the ability to grow (Kaneko et al., 1999). By contrast, the continuously growing molars of rabbits show small epithelial bulges containing cells that behave like stem cells at the apical end of the germs (Starkey, 1963). Our experiments on incisor explants showed that epithelium-lacking stem cells exhibited limited growth. Taken together, these results suggest that formation of the cervical loop during incisor morphogenesis is crucial for developing the capacity to regenerate.

Because the expression pattern of Fgf10 mRNA is identical to that of Fgf3 during molar development (Kettunen et al., 2000), FGF10 null mice do not exhibit striking molar abnormalities. Likewise, Fgf3-deficient mice do not exhibit gross morphological changes in their teeth (Mansour et al., 1993). These data support the results from our loss-of-function experiments in which antibodies that neutralize FGF3 did not affect incisor explants perniciously. Considering that Fgfr1b and Fgfr2b are expressed throughout dental epithelium (Kettunen et al., 1998; Harada et al., 1999) (http://bite-it.helsinki.fi), these results suggest that the FGFs function somewhat redundantly. As expected from the phenotype of Fgfr2(IIIb)-deficient mice (Moerlooze et al., 2000), in Fgf3 and Fgf10-double knockout mice, tooth germs would fail to progress beyond the bud stage. In the present study, in situ hybridization in the incisor germs revealed that after E16, the Fgf10-expression zone differed from that of Fgf3. Expression of Fgf3 mRNA was restricted to the mesenchyme underlying the inner enamel epithelium, but cells expressing FGF10 extended to the zone covering the cervical loop. Fgf7 mRNA was weakly expressed in the dental follicle at outside of the apical end, but not in the dental papilla (Kettunen et al., 2000), suggesting that FGF7 is associated with growth of outer enamel epithelium. Consequently, only FGF10 appears to function in maintaining cervical loop. Kettunen et al. (Kettunen et al., 2000) have found that Fgf10, Fgf3 and Fgf7 mRNA are not expressed in the mesenchyme after the later bell stage of molar development. These results suggest that continuous expression of Fgf10 in the mesenchyme of the apical end is crucial for the formation of stem cells during incisor development.

FGF signaling plays an important role in the mediation of epithelial-mesenchymal interactions in a variety of organs. Our previous studies of Fgf10 expression patterns and in vitro application of FGF10-soaked beads suggested that FGF10 acts as a regulatory signal from the mesenchyme to the epithelium during tooth development (Harada et al., 1999; Kettunen et al., 2000). In the present study, when FGF10 neutralizing antibodies caused apoptosis in the cervical loop, some neighboring mesenchymal cells were also apoptotic (Fig. 6C). Furthermore, many of cells in the isolated mesenchyme from the epithelium were apoptotic regardless of the presence of FGF10 protein (Fig. 7N,N′,O,O′). Thus, the mesenchymal cell death appears to be attributable to the absence of signaling from the cervical loop. The development of tooth germs is governed by reciprocal epithelial-mesenchymal interaction and the molecular mechanism is very similar to that seen in limbs and hair (Tickle, 1995). The growth of limbs and genital tubercles is governed by the reciprocal interaction between FGF8 and FGF10 (Ohuchi et al., 1997; Xu et al., 1998; Haraguchi et al., 2000). Although Fgf8 mRNA was observed in the dental epithelium from its thickening to early bud stage (E12), it was not seen in the developing incisor at a later stage (Kettunen and Thesleff, 1998). Kettunen et al. (Kettunen et al., 2000) have shown that the expression of both Fgf3 and Fgf10 in the mesenchyme depends on dental epithelium, and that FGF4 and FGF8 induce Fgf3 expression in the mesenchyme. In addition, Bei and Maas (Bei and Maas, 1998) have also shown that FGFs (FGF1, FGF2 and FGF8), but not bone morphogenetic protein 4 (BMP4), regulate Fgf3 expression in the dental mesenchyme. The supernatant from an odontogenic epithelial stem cell line originating from the incisor cervical loop stimulates cell proliferation and maintains expression of FGF10 in primary culture of odontogenic mesenchyme (S. Kawano, personal communication). However, FGFs, sonic hedgehog, BMP2, transforming growth factor β1 and Wnt6 expressed in the early dental epithelium did not stimulate Fgf10 expression in the dental mesenchyme. Although the epithelial signals regulating Fgf10 expression in teeth are still unknown, we speculate that a regulatory cascade exists between putative cervical loop molecules and FGF10 expressed in the mesenchyme. Adult stem cells in the mouse incisor seem to maintain an undifferentiated early embryonic stage by epithelial-mesenchymal interaction.

The persistence of stem cell populations through adulthood probably depends on the survival of quiescent cells, as well as on the ability of cycling cells to self-renew. Quiescent stem cells have been identified in the liver, brain, bone marrow (reviewed by Morrison et al., 1997) and hair bulge (Taylor et al., 2000). However, it is not clear whether these apparently quiescent cells are in G₀ or whether they are moving very slowly through G₁. We have previously identified slow-cycling cells of the incisor cervical loop as putative stem cells in the dental epithelium (Harada et al., 1999). In the present study, BrdU-labeled cells in the cervical loop showed a sparse distribution pattern. Cells escaping from the cell cycle appear to behave like quiescent cells. Although previously we have shown that FGF10 stimulates division of stem cells, we do not yet know whether FGF10 plays a role in the survival of quiescent cells.

Increased apoptotic cells are found in hair follicles of FGF10 null mice (Ohuchi et al., 2000). These data are consistent with the present study, which demonstrates that the FGF10-neutralizing antibodies cause apoptosis in the incisor cervical loop epithelium. Furthermore, recombinant human FGF10 could rescue the isolated cervical loop epithelium from apoptosis. The evidence indicates that FGF10 plays a role in stem cell survival and stimulates cell-proliferation; however, other, more complex regulatory mechanisms in the stem cell compartment remain to be studied.

We thank Prof. Irma Thesleff (University of Helsinki, Finland) for suggestions and critical reading of the manuscripts. This work was partly funded by grants from the Ministry of Education, Science, Sports and Culture (No.11771118, No.13470387, No.13877308), the Kato Memorial Bioscience Foundation in Japan and Scandinavia Japan Sasakawa Foundation (No.01-18).