Main

Tooth development results from a cascade of permissive and instructive epithelio-ectomesenchymal interactions (1–3). After initiation and progressive invagination of dental epithelium in ectomesenchyme, tooth morphogenesis reaches successive developmental stages, from bud to cap and bell stages leading to a characteristic tooth shape (4). During tooth formation, numerous potential regulatory molecules are sequentially expressed, including transcription factors, growth factors, and extracellular matrix proteins (5, 6) (WWW-site Gene expression in tooth, http://honeybee.helsinki.fi/toothexp). The expression patterns of many of these molecules have been related to their role, explored during in vitro and in vivo tooth development, as shown for BMP (7, 8), Lef-1 (9), Msx-1 (10, 11), FGF (12, 13), Notch (14), and Sonic Hedgehog (15).

The Msx (Hox7/8) and Dlx genes constitute two subfamilies of divergent homeobox genes related to the muscle-segmentation gene (msh) and the DLX (Dll) gene in Drosophila, respectively (16, 17). Several studies have proposed that the determination of position and shape of each tooth germ arises from specific spatiotemporal combinations of gene expression (8, 18–20). During the early development of mouse craniofacial structures, the Msx and Dlx genes are expressed in distinct and overlapping regions (20, 21). Targeted mutations of Msx and Dlx gene expression in mice results in profound alterations of a number of mineralized tissues derived from the first and second branchial arches, including the arrest of tooth development (10, 13, 22, 23).

In humans, the role of MSX and DLX genes in craniofacial development has been highlighted by the identification of MSX-1 (24), MSX-2 (25), and DLX-3 (26) mutations associated with specific syndromes i.e., tooth agenesis, craniosynostosis, and tricho-dento-osseous syndrome, respectively. Eight members in the DLX family, as well as MSX-1 and MSX-2 genes, have been identified in the human genome (24–31). Furthermore, a polymorphism for MSX-1 and DLX-3 genes has been reported that may be related to the prevalence of human developmental defects (26, 32).

To establish some relationship between those reported human syndromes, previous experimental data in mice, and the expression patterns of MSX and DLX homeogenes in humans, in situ hybridization studies were performed to map the specific/overlapping territories of MSX-2, DLX-5, and DLX-7 expression in teeth during the early stages of human development.

METHODS

Tissue collection and preparation.

A restricted collection of human embryos was obtained from legally approved medical abortions in the Robert Debré Hospital Obstetric Department. The project was performed under approval of the French Ethic National Committee in the context of the bioethical laws that define precisely the human study protocols. The fetuses studied ranged from 7.5 to 9 wk of gestation. The precise stage of embryonic development was determined by medical assessment of the date of the latest period and embryo morphologic characteristics. Samples were fixed by immersion in 4% paraformaldehyde, 15% sucrose in 0.1 M PBS, pH 7.4 (Sigma Chemical Co., La Verpillière, France), for 12 h at 4°C and were rinsed in 15% sucrose, PBS, for 12 h at 4°C. Serial frozen sections (10-μm) were prepared and cut in a frontal or sagittal plane with a cryostat (Bright Instrument Company LDT, Huntingdon, GB) at −25°C. Cryostat sections were placed on 50 mg/mL poly L-lysine (Sigma Chemical Co.)-coated slides, dehydrated in a graded series of ethanol, and stored at 4°C.

Human DLX-5 and DLX-7 cDNA probes.

A human third molar cDNA library, constructed as previously described (33), was used to isolate a partial human DLX-5 probe. DlX-5 and Dlx-7 primers were constructed by using published mouse DlX-5 and DlX-7 sequences, respectively (Genbank accession number U67840 and U73338) by the DNA Core Laboratory (University of Texas Health Science Center at San Antonio). The sense primers corresponded to mouse DlX-5 552–570 5′-TGGTAAACCAAAGAAAGTTC-3′ and mouse DlX-7513–531 5′-TAATGAAGCTCCTGAAACAG-3′, and the antisense primer to DlX-51025–1009 5′-ATAAAGCGTCCCGGAGG-3′and DlX-7 725–708 5′-ACTCACATCATCT-GAGGC-3′ flanking the highly conserved homeodomain sequence (17, 29). The DNA products were subcloned in pCRScript (Stratagene) according to the supplier's instructions. DNA sequence was determined by using a Sequence 2.O kit (U.S. Biochemical Corp). BLAST searches of the Genbank/EMBL Nucleotide Sequence Database were performed. The Clustal Program (MacVector 6.0) was used for alignments with mouse DlX-5, DlX-7, and human DLX-5, DLX-7.

In situ hybridization.

For in situ hybridization, we used the PCR fragments of DLX-5 and DLX-7 subcloned into pCRScript plasmid linearized either with Sac I or Kpn I (Promega, Charbonnière, France). MSX-2 riboprobes were transcribed on a 1.9-kb human MSX-2 cDNA fragment subcloned into pBS plasmid. MSX-2 plasmid was linearized with either Hin dIII or Bam HI endonucleases (Promega). [35S]-UTP-labeled single-stranded antisense and sense probes were synthesized in vitro, using T7 and T3 polymerases (Promega). In situ hybridization was performed as described previously (34). All pretreatment and hybridization solutions were made with 1 mL/L diethyl pyrocarbonate (Sigma Chemical Co.) autoclaved water. Pretreatment was performed at room temperature. Cryostat sections were rehydrated for 5 min in PBS and were pretreated for 15 min in 7 μg/mL of proteinase K (Sigma Chemical Co.) in 50 mM Tris-HCl, 5 mM EDTA at pH 8. Sections were rinsed in PBS and fixed for 20 min in 4% paraformaldehyde, PBS. They were then rinsed for 5 min in water and treated for 10 min with 0.25% acetic anhydride added to 0.1 M triethanolamine. Sections were rinsed for 5 min with PBS and dehydrated in a graded series of ethanol. The hybridization solution contained 60% deionized formamide, 300 mM NaCl, 20 mM Tris-HCl at pH 8, 5 mM EDTA at pH 8, 10% dextransulphate, 0.5 mg/mL yeast tRNA, and 1x Denhardt's solution. Sections were hybridized with 20 μL of labeled probes containing 60,000 cpm/μl radioactivity in hybridization solution, 20% 1 M DTT, in a humid chamber overnight at 50°C. Sections were washed for 60 min in 5xSSC, 10 mM DTT at 50°C, and then for 30 min in 50% formamide, 2xSSC, and 20 mM DTT at 65°C. Sections were rinsed three times for 15 min in 500 mM NaCl, 10 mM Tris-HCl, and 5 mM EDTA (NTE) at 37°C, incubated for 30 min in ribonuclease A (Sigma Chemical Co.), 20 μg/mL NTE at 37°C, and rinsed for 15 min in NTE at 37°C. Sections were washed for 30 min in 50% formamide, 2xSSC, and 20 mM DTT at 65°C, rinsed for 15 min in 2 × SSC and for 15 min in 0.1 × SSC, at room temperature. Sections were then dehydrated in a graded series of ethanol. The slides were dipped into NTB2 autoradiographic emulsion (Kodak, Paris, France) and exposed for 5 wk at 4°C. After the film was developed, sections were stained with hematoxylin, dehydrated, and mounted under a coverslip. Sections were examined and photographed with a Leitz Orthoplan microscope.

RESULTS

MSX-2, DLX-5, and DLX-7 gene expression was first detected in the branchial arch tissues as early as at wk 7.5 of human embryo development (Fig. 1). Postnatal expression of DLX-5 and DLX-7 in tooth organ was also suggested, because DLX-5 and DLX-7 transcripts could be amplified from a human tooth cDNA library produced from 14-y-old third molars (data not shown).

Figure 1
figure 1

Expression of MSX-2, DLX-5, DLX-7, and transcripts in orofacial tissues of 7.5-wk human embryos. Views of frontal section of orofacial tissues (incisor and molar zones), vestibular aspect on the left. In situ hybridization with [35S]-labeled antisense riboprobes: dark field (a, c, and e) and bright field (b, d, and f) views. In maxilla (MX) and mandible (MB), a strong MSX-2 (a and b) and DLX-5 (c and d) in situ hybridization signal is observed in progenitor cells surrounding bone (B) and Meckel's cartilage (MC) and in tooth (T) germs. DLX-7 (e and f) mRNA labeling is detected only in vestibular lamina (VL). Bar = 200 μm.

Full size image

At 7.5 wk, MSX-2 was strongly expressed in progenitor cells of orofacial skeletal structures, including mandible and maxilla membranous bones, Meckel's cartilage, and tooth germs (Fig. 1, a and b). Such a gene expression pattern in mineralized tissue was observed for DLX-5 (Fig. 1, c and d). DLX-7 transcripts were detected only in oral and dental epithelium (Fig. 1, e and f). The respective expression patterns of MSX-2, DLX-5, and DLX-7 were similar in the different tooth types of the human temporary dentition.

In tooth germs, at bud stage, analysis of serial frontal sections showed that MSX-2 transcripts were continuously expressed from the lingual part of vestibular lamina to the vestibular part of dental lamina. A strong MSX-2 hybridization signal was also observed in the lingual part of dental mesenchyme (Fig. 2, a and b). The DLX-5 hybridization signal was intense and limited to a narrow band of epithelial cells, vestibular to dental epithelium thickening, whereas DLX-5 expression was weaker and focused in the central part of dental mesenchyme beneath dental epithelium (Fig. 2, c and d). DLX-7 mRNA labeling was detected only in the epithelial vestibular lamina (Fig. 2, e and f). No significant labeling was observed in control with sense probes (Fig. 2, g and h).

Figure 2
figure 2

Expression of MSX-2, DLX-5, DLX-7, and transcripts in tooth germs of 7.5-wk human embryos. Views of frontal section of tooth buds (incisor and molar), vestibular aspect on the left. In situ hybridization with [35S]-labeled antisense and sense riboprobes: dark field (a, c, e, and g) and bright field (b, d, f, and h) views. At bud stage, MSX-2 (a and b) in situ hybridization signal is seen in the lingual aspect of vestibular lamina (VL) and the vestibular aspect of dental epithelium (DE), and also in the lingual part of dental mesenchyme (DM). A strong DLX-5 (c and d) mRNA labeling is detected, especially in the vestibular aspect of DE and DM. Osteoprogenitor cells of mandibular bone (B) are also strongly labeled. The DLX-7 (e and f) mRNA signal is observed only in the epithelial VL. No significant labeling is observed in control (C) with MSX-2 sense riboprobe (g and h). Bar = 50 μm.

Full size image

At the bud stage, DLX-7, MSX-2 and DLX-5 were expressed in vestibular lamina and dental epithelium with a spatially ordered sequence along a vestibular/lingual axis. An inverted spatially ordered sequence was observed in dental mesenchyme for DLX-5 (vestibular) and MSX-2 (lingual) expression.

During cap stage at 9 wk, MSX-2 mRNAs were strongly expressed in the vestibular aspect of dental epithelium and also centrally in the enamel knot. MSX-2 hybridization signal was not detected in dental mesenchyme (Fig. 3, a and b). DLX-5 transcript signal was still restricted to the vestibular and apical aspect of dental epithelium invagination. DLX-5 expression was progressively located in a wider area of vestibular and lateral dental epithelium. The specialized enamel knot structure did not express DLX-5. DLX-5 was strongly expressed in dental mesenchyme, with its expression prominent as a sphere located beneath enamel knot (Fig. 3, c and d). DLX-7 mRNA labeling was observed only in the vestibular lamina and the most external cell layer of vestibular part of dental epithelium. DLX-7 expression increased progressively in the vestibular and apical aspect of dental epithelium (Fig. 3, e and f).

Figure 3
figure 3

Expression of MSX-2, DLX-5, and DLX-7 transcripts in orofacial tissues of 9-wk human embryos. Views of frontal section of orofacial tissues (maxilla canine zone), vestibular aspect on the left. In situ hybridization with [35S]-labeled antisense riboprobes: dark field (a, c, and e) and bright field (b, d, and f) views. At cap stage, MSX-2 labeling is observed in the vestibular aspect (arrow) and the enamel knot (asterisk) of dental epithelium (DE), whereas dental mesenchyme (DM) is not labeled (a and b). DLX-5 in situ hybridization signal is detected in the vestibular aspect (arrow) of DE and the vestibular and central aspect of DM adjacent to DE. The enamel knot (asterisk) is not labeled (c and d). DLX-7 labeling is located in the vestibular part (arrow) of DE. DLX-7 in situ signal is also detected in the lingual zone of vestibular lamina (VL) (e and f). Bar = 100 μm.

Full size image

At the cap stage, the expression patterns of DLX-5 and DLX-7 showed the same characteristics observed at the bud stage, whereas MSX-2 expression pattern varied, as shown by the disappearance of MSX-2 transcripts in dental mesenchyme. Furthermore, MSX-2, DLX-5, and DLX-7 expression patterns overlapped in the vestibular aspect of dental epithelium. Specific areas of homeogene expression also were observed, i.e., MSX-2 in enamel knot, DLX-5 in dental mesenchyme, and DLX-7 in vestibular lamina (Fig. 4, a and b).

Figure 4
figure 4

Schematic diagram summarizing the overlapping expression of MSX-2, DLX-5, and DLX-7 at bud (A) and cap (B) stage. VL, vestibular lamina;DE. dental epithelium;DM, dental mesenchyme;EK, enamel knot.

Full size image

DISCUSSION

The organization of orofacial skeletal structures, especially the dentition, is a species-specific trait in mammals. The genetic basis of the dentition may therefore diverge from rodents to humans (35). To address this point, and regarding the importance of homeogenes in human pathology (24–26, 35), we have initiated investigation of several homeobox genes during the morphogenesis of the different human tooth types.

The hypotheses with regard to the role of homeogenes during human orofacial development are based largely on both rodent experiments and analysis of human clinical phenotypes associated with homeogene mutations (36). The Boston type craniosynostosis syndrome is associated with an MSX-2 gene mutation resulting in MSX-2 protein dysfunction (37). This syndrome is characterized by skull defects related to premature suture fusion and, also, maxilla and mandible bone abnormalities (25). The presence of MSX-2 in orofacial bone- and cartilage-forming cells, shown in this study, again suggests a direct involvement of MSX-2 in the formation of human skeletal structures.

We observed, also, a strong expression of MSX-2 in the odontogenic cells of the various human tooth germs. During tooth development in mice, Msx-2 is continuously expressed in molars and incisor tooth germs (16) but is transiently expressed in dental lamina, which corresponds to putative tooth anlagen, such as canine (38). The regression of these tooth rudiments has been related to the disappearance of Msx-2 expression (38). The presence of MSX-2 in human molar, incisor, and also canine tooth germs suggests that MSX-2 is involved in the morphogenesis of each tooth type of the mammalian dentition, and it also supports the concept that the level and functionality of Msx gene expression may affect the dental patterning (39), as shown by the specific tooth agenesis associated with human MSX-1 mutation (24).

Dlx-7 expression pattern in mammals has never been investigated in situ, apart from the initiation stage (21). At that stage, Dlx-7 expression has been detected previously only in the mandibular arch of mice (21), as shown for Dlx-3, Dlx-5, and Dlx-6 expression (21, 22). During the later stages of tooth development, we observed that DLX-7 was expressed in both mandible and maxilla. Such a homeogene expression redistribution has been described for DLX-5 expression during orofacial development (40). At the bud stage, DLX-7 was specifically expressed in the vestibular lamina site. This prominent expression of DLX-7 in nondental epithelium may explain the relatively late detection of Dlx-7 by RT-PCR in developing mouse tooth germs (41). Furthermore, the specific presence of DLX-7 in the vestibular lamina primordia illustrated the fact that homeogenes could be involved in the morphogenic event of oral vestibule formation.

It has been shown that specific combinations of homeogene and growth factor expression control the homeogene expression in mandible and maxilla ectomesenchyme (12, 13, 15) and, consecutively, pattern the dentition (8). However, the factors that may control the expression of Msx/Dlx homeogenes in the epithelium at the initiation and bud stages of tooth development are questioned (12, 35). In our study, at the lamina-bud transition stage, DLX-7, MSX-2, and DLX-5 expression appeared to be spatially ordered in the epithelium from the vestibular lamina to the vestibular aspect of bud epithelium. At this stage, BMP4 is specifically expressed in the mesenchyme underlying this DLX-7-, MSX-2-, and DLX-5-expressing epithelium zone (23, 42, 43), suggesting that BMP4 controls the differential spatial restriction of homeogene expression in the epithelium, as shown for Msx-2 during the cap stage (19).

It has been proposed that during the later stages of tooth morphogenesis, the shape of the different tooth germs is under the control of various organizing centers, such as the enamel knot (4). The enamel knot corresponds to a transient morphologic event observed in dental epithelium during cap stage (19). This structure would control dental morphogenesis based on its regulation of odontogenic cell proliferation and of apoptosis via its strong expression of diffusible growth and differentiating factors such as FGF4, BMP2, BMP4, Sonic Hedgehog, and their regulatory roles on homeogene expression (15, 19, 44). The strong expression of MSX-2 observed here suggests the existence of such an organizing center—i.e., the primary enamel knot—not only during the development of molars but also during the formation of other tooth types, canine and incisors. Furthermore, the co-expression of DLX-5, MSX-2, and, to a lesser extent, of DLX-7 in the vestibular part of the dental epithelium allowed us to identify another area of dental epithelium, already observed here during bud stage and identical in the different human tooth types. It has been proposed that a functional antagonism observed in vitro between Msx-2 and Dlx-5 proteins controls homeogene action during limb development (45). Similarly, during tooth development, the coincident expression of MSX-2 and DLX-5 in the vestibular aspect of tooth germs could induce regulatory molecular interactions that specify the vestibular/lingual tooth polarity.

This study illustrates the potential roles of MSX and DLX homeogenes during human orofacial skeletal structure formation. Furthermore, the expression patterns of MSX-2, DLX-5, and DLX-7 extend the proposed site-specific combination of homeogene expression observed at the initiation stage (20, 21) to the later stages of tooth morphogenesis.