Reciprocal Temporospatial Patterns of Msx2 and Osteocalcin Gene Expression During Murine Odontogenesis


  • Miri Bidder,

    1. Departments of Medicine and Molecular Biology and Pharmacology, Divisions of Endocrinology, Diabetes, and Metabolism and Bone and Mineral Diseases, Washington University School of Medicine, St. Louis, Missouri, U.S.A.
    Search for more papers by this author
  • Tammy Latifi,

    1. Departments of Medicine and Molecular Biology and Pharmacology, Divisions of Endocrinology, Diabetes, and Metabolism and Bone and Mineral Diseases, Washington University School of Medicine, St. Louis, Missouri, U.S.A.
    Search for more papers by this author
  • Dr. Dwight A. Towler M.D., Ph.D.

    Corresponding author
    1. Departments of Medicine and Molecular Biology and Pharmacology, Divisions of Endocrinology, Diabetes, and Metabolism and Bone and Mineral Diseases, Washington University School of Medicine, St. Louis, Missouri, U.S.A.
    • Washington University School of Medicine, Department of Molecular Biology and Pharmacology, Box 8103, 660 South Euclid Street, St. Louis, MO 63110 U.S.A.
    Search for more papers by this author

  • Presented in part as an abstract at the 19th Annual Meeting of the American Society for Bone and Mineral Research, Cincinnati, Ohio, U.S.A., 1997.


Msx2 is a homeodomain transcription factor that regulates craniofacial development in vivo and osteocalcin (Osc) promoter activity in vitro. Msx2 is expressed in many craniofacial structures prior to embryonic day (E) E14 but is expressed at later stages in a restricted pattern, primarily in developing teeth and the calvarium. We examine Osc expression by in situ hybridization during murine development, detailing temporospatial relationships with Msx2 expression during preappositional and appositional odontogenesis and calvarial osteogenesis. Osc expression at E14–14.5 is very low, limited to a few perichondrial osteoblasts in the dorsal aspect of developing ribs. At E16.5 and E18.5, Osc expression is much higher, widely expressed in skeletal osteoblasts, including calvarial osteoblasts that do not express Msx2. No Osc is detected in early preappositional teeth that express Msx2. In incisors studied at an early appositional phase, Msx2 is widely expressed in the tooth, primarily in ovoid preodontoblasts and subjacent dental papilla cells. Osc is detected only in a small number of maturing odontoblasts that also express α1(I) collagen(Col1a1) and that are postproliferative (do not express histone H4). Msx2 expression greatly overlaps both histone H4 and Col1a1 expression in ovoid preodontoblasts and dental papilla cells. By the late appositional phases of E18.5 and neonatal teeth, Osc mRNA is highly expressed in mature columnar odontoblasts adjacent to accumulating dentin. In appositional bell-stage molars, reciprocal patterns of Msx2 and Osc are observed in adjacent preodontoblasts and odontoblasts within the same tooth. Osc is expressed in mature columnar odontoblasts, while Msx2 is expressed in adjacent immature ovoid preodontoblasts. In less mature teeth populated only by immature ovoid preodontoblasts, only Msx2 is expressed-–no Osc is detected. Thus, Msx2 and Osc are expressed in reciprocal patterns during craniofacial development in vivo, and Msx2 expression in preodontoblasts clearly preceeds Osc expression in odontoblasts. In functional studies using MC3T3-E1 calvarial osteoblasts, Msx2 suppresses endogeneous Osc, but not osteopontin, mRNA accumulation. In toto, these data suggest that Msx2 suppresses Osc expression in the craniofacial skeleton at stages immediately preceeding odontoblast and osteoblast terminal differentiation.


The transcriptional heirarchy that regulates skeletal gene expression in any particular mineralized tissue is dependent upon the morphogenetic, metabolic, and mechanical demands placed upon that ossicle.(1,2) Biochemical and genetic analyses are now providing significant insight into how the osteoblast-specific transcriptional machinery integrates multiple combinatorial cues conveying morphogenetic demands. For example, a Runt domain transcription factor, Cbfa1(3–5) was recently determined to globally control the osteoblast differentiation program, including expression of genes such as Osteocalcin (Osc, inhibits mineralization) and α1(I) collagen (Col1a1, promotes mineralization). By contrast, the homeodomain transcription factors Dlx1, Dlx2, Mhox, Msx1, and Msx2 are expressed in specific mineralizing morphogenetic fields of the head (e.g., teeth, calvarial bones, palate, mandible), regulating odontogenesis and craniofacial osteogenesis.(6–10)

Msx2 is a homeodomain gene expressed in neural crest–derived calvarial osteoblasts and odontoblasts and in several osteoblastic cell lines.(11,12)Msx2 is also transiently expressed during BMP2-induced heterotopic bone formation at the onset of endochondral mineralization.(13) In cotranfection studies carried out in cultures of phenotypically immature calvarial osteoblasts, Msx2 is a transcriptional repressor that suppresses Osc gene expression.(12,14,15) In differentiating cultures of primary rat calvarial osteoblasts, Msx2 and Osc genes are expressed in a temporally reciprocal fashion; Msx2 mRNA decreases with time in culture, while Osc mRNA increases with time, associated with withdrawal from the cell cycle and the onset of terminal differentiation and matrix mineralization.(16,17) However, the temporospatial relationships between Msx2 action and Osc gene expression in vivo during normal development remain to be determined.

Osc is an ∼6 kDa gamma-carboxylated protein that binds Ca2+ and inhibits mineralization.(18–20) Expression of Osc is limited almost exclusively to bones and teeth,(20) although both Osc mRNA and protein expression have been detected in aortic valves undergoing dystrophic calcification.(21) The crucial role of Osc in normal bone physiology was recently demonstrated by Karsenty and coworkers.(19) Mice homozygous for disruption of the Osc gene cluster develop a diffuse hyperostotic syndrome postnatally, characterized by accelerated bone formation and mineralization.(19) The mineral deposited in Osc -/- mice is altered and may exaggerate osseous responses to ovariectomy.(19,22) A landmark study of Osc expression as determined by in situ hybridization analysis of neonatal rat bone has been published,(23) and a systematic analysis of Osc gene expression during murine development by a semiquantitative reverse transcription polymerase chain reaction (RT-PCR) assay has been presented.(24) However, no detailed analysis of Osc expression with respect to craniofacial tissue morphogenesis and the expression of a known transcriptional regulator has been reported to date.

We now provide such an analysis using in situ hybridization and establish temporospatial patterns of Msx2 and Osc gene expression in mineralizing developmental fields where they potentially interact, namely teeth and certain craniofacial bones. By this analysis, we extend earlier observations that identified Msx2 as a transcriptional repressor the Osc gene.(12,15,16)Osc and Msx2 expression were compared with each other and with reference to matrix deposition phases of bone and tooth development.(17) The matrix deposition/apposition phase of developing structures was monitored by trichrome histochemical staining for dense regular type I collagen deposition and in situ hybridization for elevated Col1a1 expression. Proliferative phases of developing structures were also monitored by in situ hybridization for histone H4 (expressed in S phase(17,25)). As observed in neonatal rats,(23) murine prenatal Osc expression is limited to a subset of osteoblasts immediately adjacent to bone and odontoblasts depositing dentin. Osc expression is first observed by in situ hybridization at embryonic day (E) 14–14.5, in dorsal regions of ribs undergoing matrix apposition(26); no Osc gene expression is observed in cartilage or hypertrophic chondrocytes. At E16.5 and E18.5, Msx2 expression is readily noted in proliferating dental papilla cells and preodontoblasts of the developing tooth. Although Osc is highly expressed in E16.5 calvarium and alveolar bone, no Osc is detected in preappositional E16.5 or E18.5 teeth. However, at E18.5, Osc is highly expressed in mature odontoblasts of teeth undergoing apposition, first observed in cusp apices.(27) In the E18.5 face, Msx2 expression is detected in anlage of the developing nasal septum, turbinates, and cartilage underlying the internasal suture.

By contrast, Osc expression is limited to osteoblasts adjacent to frontal bones overlying these structures. At E18.5 and later stages, Msx2 expression becomes progressively more restricted to a subset of maturing preodontoblasts. Some odontoblasts and dental papilla cells that express Msx2 also express Col1a1. However, in the same tooth, reciprocal patterns of Msx2 and Osc expression are observed in differentiating odontoblasts undergoing the transition from the preappositional to the appositional phases at the cusp apex; Msx2 expression is down-regulated as preodontoblasts mature, with concomitant up-regulation of Osc expression in maturing columnar odontoblasts that deposit dentin. Stable expression of Msx2 suppresses endogeneous Osc mRNA accumulation in the MC3T3-E1 calvarial osteoblasts. These observations are consistent with the observation that Msx2 suppresses Osc promoter activity in cell cultures.(12,14,15) In toto, these data suggest that Msx2 may serve two functions during development: to presage and regulate craniofacial morphogenesis in specific mineralizing structures, including the calvarium and dentition(9,11,26,27); and to regulate the onset of odontoblast differentiation prior to matrix apposition, including repression of odontoblast terminal differentiation markers such as Osc.(10,15,27)


Specimen preparation for in situ analyses

All reagents were purchased from Sigma Chemical Company (St. Louis, MO, U.S.A.) unless otherwise indicated. Solutions for in situ procedures were treated with diethyl pyrocarbonate prior to use. All glassware was rinsed with diethylpyrocarbonate-treated water and oven-dried. Protocols were approved by the Washington University Animal Studies Committee. Time pregnant female ICR mice were purchased from Harlan Bioproducts (Indianapolis, IN, U.S.A.). Embyros were dissected in ice-cold phosphate buffered saline (PBS: 0.13 M NaCl, 7 mM Na2HPO4, 3 mM NaH2PO4, pH 7.4) prior to fixation. Embyros were fixed in 4% paraformaldehyde in PBS (freshly prepared and filtered) for 0.5–1 h at 25°C. Larger specimens were fixed overnight at 4°C. After fixation, specimens were washed with fresh PBS for 30 minutes, then decalcified with Decalcifier I (Surgipath Medical Industries, Inc., Richmond, IL, U.S.A.) for 20 (E14–14.5) to 60 minutes (E18.5 and older) at 25°C. After washing again with fresh PBS, samples were dehydrated by a graded ethanol series (30%, 50%, 70% ×2, 95% ×3, then 100% ethanol ×3) for 0.5–2 h at each step, with longer times used for larger specimens. Following dehydration, samples were cleared twice with xylene. Specimens were then embedded in Tissue Prep-2 (Fisher Scientific, St. Louis, MO, U.S.A.) at 65°C, oriented by placement in plastic moulds. Subsequently, 5-μm paraffin sections were cut, floated in a waterbath containing STA-ON (Surgipath Medical Industries) to relax compression, then transferred onto either polylysine-coated slides or Superfrost Plus slides (Fisher Scientific). Sections were air dried for several hours, incubated at 50°C for 30 minutes, and then allowed to cool overnight at room temperature. Slides were then stored at 4°C until used for hybridization.

Plasmids and cDNA templates for riboprobe synthesis

Templates for generating Msx2 riboprobes were obtained by PCR using the mouse Msx2 cDNA as a template. The C-terminal Msx2 antisense probe is generated in pKS from the T7 promoter; the N-terminal Msx2 antisense probe is generated in pcDNA3 from the Sp6 promoter. Templates for generating Osc and Col1a1 riboprobes were obtained by RT-PCR using RNA extracted from differentiated MC3T3-E1 calvarial osteoblasts. The probe for histone H4 was generated by PCR using human genomic DNA as a template. PCR fragments were gel purifed, ligated into pUC19 for sequencing, then directionally subcloned into pcDNA3 for generating antisense riboprobes from the T7 promoter. All plasmids were purified by Qiagen column chromatography (Qiagen, Chatsworth, CA, U.S.A.), sequenced to verify insert orientation and identity (ABI Prism Dye Terminator Cycle Sequencing Kit, Perkin Elmer, Foster City, CA, U.S.A.) prior to use in riboprobe generation. The specific sequences used for riboprobes were: Osc, Genbank #X04142, nucleotides 9–376; Msx2 N terminus, Genbank #X59252, nucleotides 367–909; Msx2 C terminus, Genbank #X59252, nucleotides 748–1173; Col1a1 3′-untranslated region, Genbank #X06753, nucleotides 1248–1517; histone H4, Genbank #M16707, nucleotides 610–897.

Preparation of nonradioactive and radioactive riboprobes

The plasmids containing cDNA insert fragments were linearized, phenol extracted, ethanol precipitated, and resuspended in 10 mM Tris, pH 8, at a final DNA concentration of 0.2–0.5 μg/μl. Digoxigenin-labeled riboprobe was then synthesized with the DIG-RNA Labeling Kit by in vitro transcription in the presence of digoxigenin-UTP, as detailed in the maufacturer's instructions (Boehringer Mannheim, Indianapolis, IN, U.S.A.). [35S]-labeled riboprobes were synthesized using a standard in vitro transcription system (Boehringer Mannheim) and α-[35S]dUTP (Amersham Life Science Inc., Arlington Heights, IL, U.S.A.). Sense and antisense riboprobes were synthesized using either SP6 or T7 RNA polymerases from appropriately linearized templates. Transcription reactions were performed for 75 minutes. Radiolabeled riboprobes were purified by treatment of reactions with RNAse-free DNAse I (20 minutes), adjustment to 100 μg/ml yeast tRNA (carrier), extraction with RNAzol and CHCl3 (TEL-TEST Inc., Friendswood, TX, U.S.A.), isopropanol precipitation, resuspension, and gel filtration (MicroSpin S-200 HR Columns, Pharmacia Biotech, Piscataway, NJ, U.S.A.). Purified radiolabeled riboprobes were immediately used for in situ hybridization.

Nonradioactive in situ hybridization

Paraffin was removed by two 10-minute incubations in xylene and one 1-minute treatment with ethanol:xylene (1:1, v/v). Specimens were subsequently rehydrated in serially decreasing ethanol solutions (100%, 95% ×2, 70% ×2, 50% EtOH, then 30% ethanol in water). After rinsing in PBS, rehydrated specimens were postfixed in 4.0% paraformaldehyde in 2× SSC (1× SSC is 0.15 M NaCl, 0.015 M sodium citrate, pH 7.0) for 10 minutes at 25°C, then incubated for 30 minutes in 2× SSC at 68°C, followed by a quick rinse in water. Specimens were digested for 15 minutes at 40°C with proteinase K (2 μg/ml in 10 mM Tris, pH 8.0, 1 mM EDTA). Digestion was terminated by placing the slides in 2 mg/ml glycine in PBS, followed by several washes in PBS, a second postfixation in 4% paraformaldehyde in PBS for 15 minutes at 25°C, and several washes with PBS. Specimens were acetylated with 0.25% acetic anhydride in 0.1 M triethanolamine, pH 8.0, for 10 minutes at 25°C. Acetylated specimens were rinsed three times in PBS, then dehydrated in serially increasing ethanol solutions as described above. Samples were allowed to air dry for at least 2 h before proceeding with hybridization. Hybridization was carried out at 50°C for 16 h in a humidified chamber saturated with 50% formamide, with indicated probes dissolved in hybridization solution (50% formamide, 5× SSC, 0.1 M phosphate buffer, pH 7.0, 0.25 mg/ml heat-denatured salmon sperm DNA, 1× Denhardt's solution, 0.25 mg/ml phenol-extracted yeast tRNA, 10% dextran sulfate, 10 mM DTT).

Immediately prior to use, the hybridization solution with probe was heated at 95°C for 3 minutes. The final concentration of probe utilized was calculated as detailed.(28) Subsequently, after dislodging coverslips with 2× SSC/10 mM DTT, specimens were washed three times for 10 minutes at 50°C in 50% formamide, 5× SSC, 10 mM DTT. Samples were pre-equilibrated for 5 minutes at 25°C with TNE buffer (10 mM Tris, pH 7.4, 1 mM EDTA, 500 mM NaCl, 10 mM DTT), followed by treatment with 1 μg/ml RNAse in TNE for 15 minutes at 37°C. Subsequently, samples were washed twice in TNE (25°C, 20 minutes), once in 2× SSC (25°C, 15 minutes), and twice in 0.2× SSC (60°C, 15 minutes). The slides were then incubated in buffer I (100 mM Tris, pH 7.5, 150 mM NaCl) overnight at 4°C. The following morning, slides were rinsed again with buffer I, then incubated for 1 h in blocking solution containing 1.5% Boehringer Mannheim Blocking Reagent and 2% normal sheep serum (ICN Biomedicals, Aurora, OH, U.S.A.). Samples were then incubated for 3 h with 1.5–7.5 U/ml alkaline phosphatase (ALP)–conjugated antidigoxigenin Fab fragments diluted in the blocking solution in a sealed chamber. Subsequently, samples were rinsed twice in buffer I (25°C, 15 minutes), and then incubated with Buffer II (100 mM Tris HCl, pH 9.5, 100 mM NaCl, 50 mM MgCl2) for 5 minutes. Hybridized riboprobe was then visualized using the ALP substrate 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) with nitrobluetetrazolium (NBT) in buffer II, prepared as follows. Ten milligrams of NBT (Sigma) was dissolved in a microfuge tube in 0.2 ml of dimethylformamide and 1 ml of buffer II subsequently added. This solution was then transferred to a fresh tube containing 30 ml of buffer II at 37°C. Five milligrams of BCIP (Sigma) was dissolved in 0.2 ml dimethylformamide, and then added slowly to the NBT solution while stirring. The final substrate mix was stored at –20°C in aliquots until required. Levamisole (10 mM) was added to the color solution just prior to mounting gasketted specimens with a coverslip (to suppress background tissue ALP signal). Specimens were incubated at 25°C until signal was evident (routinely 16 h). The reaction was then stopped by rinsing the slides with 10 mM Tris, pH 8.2, 2 mM EDTA, followed by a quick rinse in water. Slides were then mounted with Crystal Mount (Biomedia, Foster City, CA, U.S.A.) for photography. For some specimens, colorimetric detection of ALP-conjugated antidigoxigenin Fab was achieved by using Sigma Fast Fast Red TR/Naphthol AS-MX (Sigma). This latter procedure generally follows the method outlined by Chiu et al.,(29) but has been modified by the addition of 10 mM levamisole to the color detection buffer and counterstaining of specimens with 0.5% toluidine blue in distilled water immediately prior to application of Crystal Mount.

Radioactive in situ hybridization

The procedures used for sections treatment, hybridization, and washes were based upon those used by Angerer and Angerer(30) with some modifications. Briefly, sections were rehydrated, washed once in PBS for 5 minutes at 25°C, and digested with proteinase K (1 μg/ml; Boehringer Mannheim) in 10 mM Tris, pH 8.0, 1 mM EDTA for 15 minutes at 37°C. Sections were then washed, acetylated, dehydrated, and air dried as described above. Subsequently, 20 μl of hybridization solution containing 106 cpm of labeled probe was added to each specimen (hybridization solution is 50% deionized formamide, 20 mM Tris, pH 7.5, 335 mM NaCl, 1 mM EDTA, pH 8.0, 1× Denhardt's solution, 10% dextran sulfate, 150 μg/ml yeast tRNA, 300 μg/ml salmon sperm DNA, 40 mM DTT). Coverslips were placed on slides containing the hybridization mix and the probe. Hydridization was performed for 16 h in a humidified chamber as described above. After hybridization, coverslips were removed in 4× SSC containing 0.25% sodium thiosulfate pentahydrate (STS). Hybridized specimens were then washed in 4× SSC/0.25% STS for 15 minutes at 25°C, 2× SSC/0.25% STS for 15 minutes at 25°C, 0.5× SSC/0.25% STS for 5 minutes at 25°C, then two washes with 0.2× SSC/0.25% STS for 10 minutes at 60°C. After rinsing with RNAse buffer (10 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 1 mM EDTA) for 5 minutes at 25°C and 5 minutes at 37°C, samples were treated with RNAse A (10 μg/ml) for 15 minutes at 37°C. Sections were then washed twice with 2× SSC for 25 minutes at 25°C, followed by two washes with 0.1× SSC for 10 minutes at 60°C. Slides then rapidly dehydrated in graded ethanol solutions (50% for 3 minutes, 70% for 5 minutes, and 95% ethanol for 2 minutes each). Dehydrated specimens were air dried for several hours then exposed to Biomax MR film (Eastman Kodak, Rochester, NY, U.S.A.) for 3 days to estimate the length of time required for autoradiography. Slides were processed for standard autoradiography using Kodak NTB-2 nuclear track emulsion. The emulsion was diluted 1:1 and incubated at 45°C for 45 minutes. Sections were then dipped in diluted emulsion and air dried upright in the dark for 3 h. Slides were placed in a light tight box with a dessicant and exposed for 2–6 weeks. Subsequently, slides were developed with Kodak D19 Developer diluted 1:1 in water for 4 minutes at 15°C and fixed with Kodak Fixer for 5 minutes, counterstained, and mounted. Analysis was carried out using both light- and dark-field optics on a Zeiss Axiophot microscope (Carl Zeiss, Inc., Thornwood, NY, U.S.A.). Adjacent 5 μm sections were stained either using Gomori's one-step trichrome to reveal dense regular type I collagen matrix accumulation(31) or by routine hematoxylin and eosin as indicated.

Stable transfection of MC3T3-E1 cells and RT-PCR analyses

MC3T3-E1 cells were maintained as previously described.(15) After plating in 35 mm diameter culture dishes (5 × 105 cells/well), cells were transfected the next day with either pcDNA3 (Invitrogen, San Diego, CA, U.S.A.) or pcDNA3-Msx2 expression vector(15) by calcium phosphate–glycerol shock. After recovery for 3 days, the entire transfected culture was subcultured into 100 mm diameter tissue culture dishes and grown for 1 week in culture media(15) supplemented with 400 μg/ml active G418 (pcDNA3 also encodes a neomycin resistance gene). Fresh media was added every other day. Subsequently, cells from these cultures were plated onto 150 mm diameter dishes and grown for an additional 3 weeks in culture media supplemented with 400 μg/ml G418 and 50 μg/ml ascorbic acid; fresh media was added every other day. At the end of the culture period, total RNA was extracted from cultures transfected with either pcDNA3 or pcDNA3–Msx2 and then analyzed for Osc and (Osteopontin)Opn mRNA accumulation using a radiactive semiquantitative RT-PCR reaction previously described.(32) Amplimers used for mouse Osc were: 5′-AAGTCCCACACAGCAGCTTG-3′ and 5′-AGCCGAGCTGCCAGAGTTTG-3′. Amplimers used for mouse Opn were: 5′-TGCCCTTTCCGTTGTTGTCC-3′ and 5′-ACACTTTCACTCCAATCGTCC-3′.


Osc expression is first detected at mouse embryonic stage E14–14.5 in a subset of osteoblasts in the dorsal rib perichondrium

Previous analyses of Osc gene expression have established that Osc mRNA accumulates in terminally differentiated osteoblasts and odontoblasts undergoing mineralization.(1) However, Osc expression has not been compared with developmental expression patterns of known transcriptional regulator such as Msx2. To compare Osc and Msx2 expression during development, we first wished to establish the general pattern of Osc expression in murine embryo development beginning at the onset of skeletal mineralization (E14–14.5), subsequently emphasizing craniofacial development. Consistent with published semiquantative RT-PCR analyses of embyro RNA,(24) very little Osc is expressed at E14–14.5; only a subset of osteoblasts in dorsal aspects of rostral ribs undergoing endochondral ossification express marginably detectable levels of Osc (Figs. 1A and 1C); ventral portions of these same ribs that are not ossifying do not express Osc (data not shown). These Osc-expressing osteoblasts are actively elaborating type I collagen–based extracellular matrix (revealed by trichrome staining in Figs. 1B and 1D; dense regular type I collagen stains with aniline blue); this pattern of Osc expression corresponds to the earliest known ossification in the developing mouse embryo as detected by alizaran red staining.(26) No Osc expression is observed in chondrocytic cells of the rib (Fig. 1A), even when intentionally overdeveloped (Fig. 1C), and essentially no background signal is observed in these specimens in the absence of digoxigenin–Osc probe (Fig. 1E). Very little Osc expression is detected in the developing head at this stage, although low levels of Osc expression can be appreciated in the occipital, zygoma, and maxillary bones of the developing head within the following 0.5 days (data not shown), paralleling the reported progression of ossification centers at this stage of development.(26) Thus, Osc expression revealed by in situ hybridization in the developing mouse embryo faithfully recapitulates the earliest ossification centers observable by histochemical staining.(26)

Figure FIG. 1.

Expression of Osc mRNA in E14–14.5 mouse embryo dorsal rib perichondrial osteoblasts visualized by nonradioactive in situ hybridization. (A) Only a subset of osteoblasts (ob) in the developing dorsal rib periosteum express detectable levels of Osc. (C) The visualization reaction is intentionally overdeveloped to demonstrate that little if any Osc in situ signal can be detected in the hypertrophic chondrocytes (ch). (B,D) trichrome histologic analyses reveals dense regular type I collagen matrix accumulation (demonstrated with aniline blue), indicative of endochondral bone formation by osteoblasts (ob). (E) Little if any background is observed in control in situ visualization reactions carried out in the absence of digoxigenin-Osc probe.

Osc is first detected in developing teeth at mouse embryonic E18.5 in mature columnar odontoblasts

By E16.5, Osc expression is now readily detected in calvarial osteoblasts (Figs. 2A and 2C). Again, osteoblasts at this stage of calvarial osteogenesis are actively elaborating extracellular matrix, revealed by trichrome staining (Figs. 2B and 2D). However, Osc expression is not detected in nonappositional E16.5 teeth (data not shown). Similarly, at E18.5, nonappositional bell stage teeth still do not express Osc, even though adjacent alveolar bone expresses detectable Osc (Figs. 2E and 2F). However, Osc is expressed in mature columnar odontoblasts of other, appositional E18.5 bell stage teeth (Figs. 2G, 2H, and 2I), corresponding to the earliest prenatal onset of mineral apposition for murine bell stage molars.(27) Again, mature columnar odontoblasts expressing Osc are actively elaborating type I collagen–based extracellular matrix, revealed by trichrome staining (Figs. 2H and 2I; dense regular type I collagen stains with aniline blue). Osc is not expressed in adjacent ameloblasts, immature odontoblasts, or dental papilla cells (Figs. 2G, 2H, and 2I). Thus, as observed in neonatal rat bone,(23)Osc expression in the developing embryonic mouse is expressed in osteoblasts and odontoblasts immediately adjacent to nascent bone and dentin, respectively.

Figure FIG. 2.

Expression of Osc mRNA in craniofacial structures of E16.5 and E18.5 mouse embryos. Radioactive in situ hybridization analyses were performed on frontal sections utilizing an Osc antisense probe. (A,C) At E16.5, high-level Osc mRNA expression is detected in osteoblasts of calvarial bone (cb). (B,D) Trichrome staining on adjacent sections. (C,D) Enlargements of (A) and (C), respectively. (E) to (I) Reflect the transition from preappositional to appositional phases as revealed by Osc in situ hybridization and trichrome staining. Two different E18.5 teeth are shown that represent the preappositional (E,F) and appositional (G,H,I) stages of bell stage molar odontogenesis.(27) (E,F) No Osc mRNA expression is detected in the preappositional tooth, apposition confirmed by trichrome staining; however, Osc expression is detected in the same section in nearby osteoblasts of alveolar bone. (G,H) Osc is highly expressed in odontoblasts (od) of appositional teeth. (E,H) Trichrome staining confirms that mature columnar odontoblast and osteoblasts are expressing Osc and are actively elaborating type I collagen–based extra cellular matrix. (I) represents an enlargement of the square area in (H).

Msx2 and Osc expression in bone and teeth during craniofacial development

Prior to the onset of mineralization at E14–14.5, Msx2 is expressed in multiple structures of the developing mouse head.(11,32) However, at later stages, Msx2 is expressed in the developing tooth and calvarium.(9,11) We(13,15) and others(14,16) have identified that Msx2 suppresses Osc expression in calvarial osteoblasts. To detail the temporospatial relationships between Msx2 and Osc gene activation in vivo, we contrasted and compared expression patterns of Msx2 and Osc in adjacent 5 μm sections of developing craniofacial structures. We related these expression patterns to collagenous matrix synthesis by examining Col1a1 expression. As shown in Fig. 3A, at E18.5 Msx2 is readily detected in the cartilagenous anlage and perichondrium of the nasal bones and septum, but not in the overlying calvarial bone or suture (compare with hematoxylin- and eosin-stained section; Fig. 3D). By contrast, Osc is expressed in calvarial bone osteoblasts and a few perichondrial cells (Fig. 3B). Col1a1 is highly expressed in calvarium bone, suture cells, and dermal fibroblasts, but not in the cartilagenous anlage of the septum (Figs. 3C and 3D). In the neonatal incisor, Msx2 expression is readily detected in odontoblasts and dental papillae cells, but not in alveolar bone (Figs. 4A and 4C). By contrast, Osc expression is limited to a subset of mature columnar odontoblasts and adjacent alveolar bone osteoblasts (Figs. 4B, 4D, 4F, and 4H). Col1a1 expression is found in bone immature ovoid preodontoblasts and mature columnar odontoblasts (Figs. 4C and 4F). We related these expression patterns to the proliferative stage of cells by examining histone H4 expression (identifies cells in S-phase(17,25)). As observed during cultured calvarial osteoblast differentiation,(17) odontoblasts that express Osc in vivo are postproliferative, indicated by the down-regulation of histone H4 (Fig. 4G and data not shown).

Figure FIG. 3.

Comparison of expression of Msx2, Osc, and Col1A1 in E18.5 mouse developing craniofacial structures by radioactive in situ hybridization on frontal sections of 18.5-day-old mouse embryo. (A) Msx2 is readily detected in the cartilagenous anlage (nc) and perichondrium (pc) of the nasal bones but not in the overlying calvarial bone or suture. (B) Osc is expressed in osteoblasts of calvarial bone (cb) and a few perichondrial cells (pc). (C) Col1a1 is highly expressed in calvarial bone (cb), suture cells (su), and dermal fibroblasts (d.f.). (D) Hematoxylin and eosin histology.

Figure FIG. 4.

Expression of Msx2, Osc, Col1a1, and histone H4 in newborn mouse incisors as detected by radioactive in situ hybridization. (A,E) Msx2 expression is clearly detected in ovoid preodontoblasts (po) and dental papilla cells (dp). (B,F) Osc expression is limited to mature columnar odontoblasts (od) and adjacent alveolar bone osteoblasts (ab). (C) Col1a1 expression is found in both immature ovoid preodontoblasts (po) and mature columnar odontoblasts (od). (G) Odontoblasts expressing Osc in vivo are post-proliferative as indicated by down-regulation of histone H4 expression. (D,H) Histology revealed by hematoxylin and eosin staining.

Msx2 and Osc exhibit reciprocal temporal and spatial patterns of expression during odontogenesis

These data (Fig. 4) suggested that at best Osc and Msx2 are only transiently coexpressed during odontoblast maturation, and that Osc and Msx2 expression patterns may be mutually exclusive in the developing tooth. To confirm this, we contrasted and compared Msx2 and Osc expression in neonatal appositional molars, but using nonradioactive visualization of in situ hybridization probes, a technique that identifies expressing cells with better resolution than autoradiography.(28–30) Two adjacent bell stage molars were examined in parasagittal sections at preappositional and appositional stages of odontogenesis (Fig. 5). As shown in Fig. 5, nonradioactive in situ hybridization again reveals Msx2 expression in subsets of immature ovoid preodontoblasts (Figs. 5A and 5D; compare with hematoxylin- and eosin-stained section in Fig. 5C). As previously described,(11)Msx2 expression is also observed in cells of the substantia intermedia and cervical loop (Figs. 5A and 5C). In an adjacent 5 μm section, Osc expression is again limited to mature columnar odontoblasts (Figs. 5B and 5E; compare with Fig. 5C). Importantly, very little if any overlap exists between Osc and Msx2 expression within the same tooth, with high level Msx2 expression preceeding Osc expression during odontoblast maturation (Fig. 5). Thus, these results confirm that as occurs during time-dependent calvarial osteoblast differentiation in vitro,(16)Msx2 and Osc express temporally and spatially reciprocal patterns of expression during odontogenesis in vivo.

Figure FIG. 5.

Msx2 and Osc are expressed in reciprocal patterns during odontogenesis. (A,B,D,E) Gene expression in parasaggital sections of newborn mouse detected by nonradioactive in situ hybridization. (A,D) Msx2 expression. (B,E) Osc expression. (C) Cytodifferentiation revealed by hematoxylin eosin staining. Msx2 is expressed in a subset of immature ovoid preodontoblasts (po). Note that only slight overlap, if any, exists between Osc and Msx2 expression within the same tooth. Msx2 expression preceeds Osc expression during odontogenic maturation. Msx2 expression is also detected in the cervical loop (cl) and in the stratum intermedium (si).(11)

Transfection of MC3T3-E1 calvarial osteoblasts with the pcDNA3–Msx2 expression construct suppresses endogenous Osc mRNA accumulation

In transient cotransfection assays, we previously demonstrated that Msx2 suppresses Osc promoter–luciferase reporter activity,(15) consistent with the reciprocal patterns of expression we observe in vivo (vide supra). To provide additional functional data that support the notion that Msx2 suppresses Osc expression, we contrasted and compared effects of pcDNA3 versus the expression construct pcDNA3-Msx2(15) on the accumulation of endogenous Osc mRNA in pools of stably transfected MC3T3-E1 calvarial osteoblasts. MC3T3-E1 cells were chosen for these analyses since: like odontoblasts,(27) calvarial osteoblasts are neural crest–derived cells that mineralize and express osteocalcin(1); and no reliable differentiating odontoblast cell culture line is currently available. As shown in Fig. 6, MC3T3-E1 cells transfected with the pcDNA3 vector alone express readily detectable levels of Osc mRNA. In comparison, Osc message is markedly decreased in cells transfected with pcDNA3-Msx2 (Fig. 6). By contrast, no differences are observed in mRNA accumulation for another extracellular matrix protein, Opn, between osteoblasts transfected with pcDNA3 or pcDNA3-Msx2 (Fig. 6). Thus, these data provide further functional evidence that Msx2 can suppress Osc expression during craniofacial skeletal development.

Figure FIG. 6.

Transfection of MC3T3-E1 osteoblasts with the pcDNA3-Msx2 expression construct suppresses endogenous Osc mRNA accumulation. MC3T3-E1 cells stably transfected with either pcDNA3 or pcDNA3-Msx2 expression construct(15) were analyzed for accumulation of Osc or Opn mRNA by RT-PCR as described in the text. Note that pcDNA3-Msx2 down-regulates Osc, but not Opn, mRNA accumulation.


In this study, we extend earlier observations that identified Msx2 as a transcriptional repressor of the Osc gene(12,15,16) by delineating temporospatial relationships between Msx2 and Osc expression in the developing head. Early in tooth development, Msx2 presages both epithelial and mesenchymal components of this complex tissue.(10,11,15) These cells are rapidly proliferating as indicated by abundant histone H4 expression, and Osc is not expressed in these cells. At later stages of odontogenesis, Msx2 expression becomes limited to subsets of proliferating immature ovoid odontoblasts, a.k.a. preodontoblasts.(27) Significant overlap of Msx2 and Col1a1 expression is observed in mature and immature odontoblasts. By contrast, expression of Osc is limited to mature columnar odontoblasts, a.k.a. functional odontoblasts,(27) adjacent to accumulating dentin, and is not observed in cells expressing Msx2. Enhanced spatial resolution of gene expression is achieved using nonradioactive in situ hybridization, since signal is confined to the cell body with this technique.(28–30) The data indicate that maturing odontoblasts at most only transiently coexpress both Msx2 and Osc, and that gene expression is reciprocally regulated during odontogenesis in vivo. This is entirely consistent with the observation that Msx2 represses Osc promoter activity,(12,15,16) and that Msx2 and Osc are reciprocally regulated during calvarial osteoblast differentiation in vitro.(16) Transgenic mouse models(33,34) and our data(12,15) (this study) suggest that as observed in vitro,(16) temporospatial regulation of Msx2 expression is important for normal craniofacial skeletogenesis in vivo. However, the effects of Msx2 depend upon the specific cellular background in which it is expressed,(12,14,33,34) most likely reflecting differences in developmentally entrained coregulatory proteins that influence Msx2 action.(15,35)

The sterotypical spatial relationships between sequential elaboration of the odontoblast differentiation program and tooth morphology(10,27) permit temporal resolution of stage-specific gene expression in vivo. Terminal differentiation of odontoblasts at cusps is heralded by the concomitant up-regulation of phenotypic markers, vectorial secretion, and accumulation of type I collagen–based extracellular matrix and dramatic change in cellular morphology from an ovoid to columnar shape.(27) In this regard, nonendochondral ossification (dentin formation) in the developing tooth recapitulates the attractive experimental features afforded by the growth plate during endochondral ossification.(25) While this work was in progress, Sommer et al.(36) published an analysis of matrix protein gene expression in the developing mouse skeleton that also emphasizes this point. Thus, conclusions concerning the relationships between transcriptional regulatory programs, stage-specific gene expression, and terminal differentiation during nonendochondral ossification in vivo may be more accurately drawn from studies of teeth than from studies of calvarial bone. Since calvarial osteoblasts and odontoblasts are both neural crest–derived cells that mineralize via nonendochondral mechanisms,(1,10,27) some mechanistic features of their respective terminal differentiation programs may in fact prove to be similar.

The transcriptional mechanisms that control gene expression in mineralizing tissues are just beginning to be understood. The emerging regulatory complexity reflects the underlying biological complexity. For example, orthotopic tissue mineralization occcurs via overlapping yet distinct routes of endochondral and nonendochondral (e.g., tooth, clavicle, calvarial bone) ossification.(1,27) Morever, during development, mineralized skeletal tissues are controlled by regional morphogenetic cues that impinge upon intrinsic cellular differentiation programs. Finally, postnatal metabolic and mechanical demands influence expression of hormones, cytokines, prostanoids, and growth factors that in turn regulate skeletal cell growth and differentiation at every stage. However, elegant analyses of human hereditary skeletal abnormalities, murine genetics, and biochemical studies provide mechanistic insights that outline the transcriptional heirarchy that regulates osteoblast and odontoblast gene expression.(3–5,15–17,37–39) The estrogen receptor-α and vitamin D receptor affect metabolic regulation of skeletal economy at multiple levels, largely during postnatal development.(40,41) By contrast, the Runt domain transcription factor, Cbfa1 (a.k.a., PEBP2a1), globally controls osteoblast gene expression, terminal differentiation, and mineralization throughout prenatal skeletal development.(2–5) In comparison, homeodomain transcription factors such as Msx1, Msx2, MHox, Dlx1, and Dlx2 control mineralized tissue formation in highly specific morphogenetic fields during craniofacial development.(6–11) Although genetically defined as transcription factors that control skeletogenesis, few specific skeletal gene targets are known for these transcription factors. Transcriptional regulation of the Opn gene by the vitamin D receptor and the Osc gene by the vitamin D receptor, Cbfa1, and Msx2 represent the best characterized osteoblast transcription factor–promoter interactions documented to date.(15,16,20,37–42) It is interesting to note that Msx2 “knockout” mice are reported to exhibit brittle teeth,(10) that Msx2 suppresses Osc expression,(12,14–16) and that Osc inhibits mineralization.(17,18) It is intriguing to speculate that Msx2 gene disruption may dysregulate and precociously increase Osc expression by preodontoblasts, which perturbs tooth mineralization at subsequent stages of odontogenesis. Future experiments will address whether systematic perturbation of Msx2 expression in tooth and calvarial bone in vivo predictably alters Osc expression and other characteristic features of craniofacial mineralization, including high level expression of the Col1a1 gene(43); and test the notion that antagonistic interactions between Osc promoter transcriptional activators and Msx2 coordinate the onset of terminal differentiation and mineralization in neural crest–derived skeletal tissues.


The authors thank Elvie Taylor of the Departmental Histology Core Facility for her expert technical assistance. This work was supported by National Institutes of Health grant AR43731 (to D.A.T.) and the Charles E. Culpeper Foundation. D.A.T. is a Charles E. Culpeper Scholar in Medical Science.