Murine medial edge epithelium (MEE) is a subpopulation of palatal surface epithelium adjacent to and continuous with the oral and nasal epithelia. The MEE from opposing sides contact in the midline and form an epithelial seam when the paired palatal shelves are approximated. As palatal fusion proceeds, the MEE seam quickly breaks down and ultimately the epithelial cells disappear from the midline (Shuler, 1995). It has been demonstrated that there are three distinct fates for MEE, which include epithelial–mesenchymal transformation (Shuler et al., 1991, 1992; Griffith and Hay, 1992; Martinez-Álvarez et al., 2000), migration (Carette and Ferguson, 1992), and programmed cell death (Mori et al., 1994; Martinez-Álvarez et al., 2000). Despite these three different fates, the MEE cells all share a common feature, they cease DNA synthesis (Gehris and Greene, 1992) before the initial formation of the midline epithelial seam.
Endogenous transforming growth factor (TGF) -β3, but not TGF-β1 or -β2, is crucial in the regulation of disappearance of MEE during murine palatal fusion (Brunet et al., 1995; Kaartinen et al., 1995; Proetzel et al., 1995). TGF-β3 null mutant mice were born with cleft palate (Kaartinen et al., 1995; Proetzel et al., 1995). Studies on TGF-β3 null mutants revealed that the palatal shelf grew normally up to the stage when the opposing shelves contacted in the midline. MEE failed to be removed from the midline and, therefore, prohibited mesenchyme confluence. As a result, the palatal shelves that were in contact in the midline were pulled apart as the growing head expanded laterally. Failure of MEE disappearance could be rescued by exogenous recombinant TGF-β3 protein to the extent that complete palatal fusion occurred in vitro (Kaartinen et al., 1997; Taya et al., 1999).
TGF-β3, as well as other TGF-β isoforms, is a potent growth inhibitor of a wide variety of cell types, including epithelial cells. TGF-β inhibits cell proliferation by causing growth arrest in the G1 phase of the cell cycle (Massagué, 1990; Hartsough and Mulder, 1997). Members of the TGF-β family exert their function by triggering a series of serine/threonine kinase activities (Heldin et al., 1997; Massagué, 1998). An active TGF-β ligand may be presented by a membrane-anchored type III receptor (TβR-III) to the transmembrane signaling receptors TβR-I and -II, which subsequently activate the intracellular mediator SMAD proteins. Receptor-regulated SMAD2 and SMAD3 are closely related homologues. Upon activation through phosphorylation, SMAD2 and/or SMAD3 form a complex with the pathway common-mediator SMAD4 and translocate into the nucleus where it interacts in a cell-specific manner with numerous transcription factors to regulate the transcription of TGF-β–responsive genes (Heldin et al., 1997; Massagué, 1998). Absence of Smad2 or Smad3 gene expression resulting from targeted deletion of the respective Smad genes in mice has shown different developmental effects. Smad2-null mice were embryonic lethal due to failure to establish an anterior–posterior axis, gastrulation, and mesoderm formation. These developmental events were normally controlled by Smad2-dependent signals from the visceral endoderm (Nomura and Li, 1998; Waldrip et al., 1998; Weinstein et al., 1998). Because Smad2-null mice die at an early embryonic stage before emergence of palatal shelves, the effects of Smad2 null mutation on palatogenesis are unknown. Mice null for Smad3 were viable and survived for several months, indicating that Smad3 was dispensable for embryonic development and had no effect on palate development. Adult Smad3-null mutant mice develop immunologic defects and colon tumors (Zhu et al., 1998; Datto et al., 1999; Yang et al., 1999).
Our previous studies used immunohistochemistry to colocalize TGF-β3 ligand and TGF-β receptors in developing palates (Cui et al., 1998; Cui and Shuler, 2000). We found that the expression pattern of TGF-β3 and the signaling receptors TβR-I and -II were consistent. TGF-β3 and TβR-I and -II were localized predominantly to the palatal epithelium that included oral epithelium, nasal epithelium, and MEE before and during palatal shelf contact in the midline. All these molecules were lost from the midline when the MEE disappeared. Expression of TβR-III (betaglycan) was temporospatially restricted to the MEE during palatal fusion (Cui et al., 1998; Cui and Shuler, 2000). The results suggested that TGF-β3 exerted its developmental role in palatal fusion in an autocrine manner through TβR-I and -II binding. The role of TβR-III was to localize and mediate TGF-β3′s effect on MEE by specific presentation in the MEE. TβR-III might modulate TGF-β3 binding to TβR-II in the MEE cells to locally enhance TGF-β3 autocrine signaling through the TβR-I/-II receptor complex, which contributed to the selective disappearance of MEE (Cui and Shuler, 2000). In previous studies, evidence was not provided that the receptor-regulated SMADs had been spatially activated in the MEE cell in response to binding of the TGF-β3 and activation of the receptors.
In the present study, we characterized the normal expression pattern of SMAD2, phospho-SMAD2, SMAD3, and phospho-SMAD3 in developing palate tissue by Western blot analysis and immunohistochemistry. We analyzed changes in SMAD2 and phospho-SMAD2 in the absence of endogenous TGF-β3 by using TGF-β3 knockout mice. MEE proliferation was examined by BrdU incorporation analysis and correlated with phospho-SMAD2 in both TGF-β3 wild-type and TGF-β3 null mutant mice to correlate TGF-β3 signaling with cell proliferation. The results support our hypothesis that TGF-β3 is required for the phosphorylation of SMAD2 in MEE and for the inhibition of MEE proliferation during murine palatal fusion.
Identification of SMAD2/phospho-SMAD2 in the Developing Palate
By using Western blot analysis, we examined SMAD2 in the developing palate in Swiss-Webster mice between embryonic day (E) 12 to E15. To better analyze the MEE region, palatal tissues were dissected either close to the tip of the shelf or around the midline of the palate, depending on the development stage. Whole palate lysate of one fetus was used in each analysis. SMAD2 was present in the palatal shelves as early as the initial outgrowth from the maxillary prominence at E12 (Fig. 1A). SMAD2 was present from E13 to E15 during the period when the palatal shelves went through a series of morphologic changes and eventually completed palatal fusion (Fig. 1A). Comparison of SMAD2 protein levels was not possible due to the size and morphology variance of palatal shelves from E12 to E15. By using immunohistochemistry, SMAD2 was identified in the MEE as well as in the palatal oral and nasal epithelia but not in the palatal mesenchyme at the time of fusion (Fig. 1B).
To characterize the distribution pattern of activated SMAD2 in the normal developing palate, we examined specimens from Swiss-Webster mice between E12 and E15 by immunohistochemistry with the antibody against phospho-SMAD2. Phospho-SMAD2 was not detectable in any part of the palatal shelf during its initial growth into the oral cavity at E12. Phospho-SMAD2 was first found in the palate at E13 when the palatal shelves were vertical along the lateral sides of the tongue. Phospho-SMAD2 was present as a nuclear stain limited to the distal portion of the palatal nasal epithelium (Fig. 2A). Palatal shelves reoriented to a horizontal orientation above the tongue at E14 and phospho-SMAD2 was dispersed in the palatal epithelium. No phospho-SMAD2 was found in the mesenchyme immediately underlying the palatal epithelium (Fig. 2B). As the palate continued growth, the opposing shelves eventually contacted in the midline (E14.5), along which distribution of phospho-SMAD2 in the palate was concentrated to the tip of the palatal epithelium (Fig. 2C). Intense nuclear staining for phospho-SMAD2 was distinguished in the MEE and adjacent oral/nasal palatal epithelial cells (Fig. 2D). As the midline epithelial seam was disrupted, the distribution of phospho-SMAD2 in the midline appeared discontinuous, which was accompanied by an intense staining allocated to the oral/nasal triangle area as a result of MEE migration (Fig. 2E). No positive phospho-SMAD2 staining was observed in the palatal mesenchyme immediately underlying the MEE. Scattered mesenchymal-like cells were phospho-SMAD2 positive but were not adjacent to MEE (arrowhead in Fig. 2E). These mesenchymal-like cells were located in a position consistent with epithelial–mesenchymal transformed MEE cells. By late E15, the MEE had disappeared from the midline and the mesenchyme became confluent, at this point phospho-SMAD2 was completely absent in the palate tissues (Fig. 2F). The temporospatial distribution of phospho-SMAD2 in the MEE suggests involvement in the disappearance of the MEE during normal palatal fusion.
Phospho-SMAD2 Is Selectively Missing in the MEE in TGF-β3−/− Palate Tissue
To test the hypothesis that TGF-β3 was necessary for the phosphorylation of SMAD2 in the MEE, we examined the distribution of phospho-SMAD2 in palate tissue in TGF-β3 null mutant mice by immunohistochemistry. We focused on E14.5, the point at which TGF-β3−/− palatal fusion was arrested and MEE remained in the midline. We found no difference between TGF-β3+/+ control mice and the Swiss-Webster strain with regard to the distribution pattern of phospho-SMAD2 in palatal tissue. In TGF-β3−/− mice, there was an absence of phospho-SMAD2 in the MEE (Fig. 2G,H) in comparison to the TGF-β3+/+ littermates. Phospho-SMAD2 was infrequently identified in peridermal cells but clearly evident at the sites of blood vessel formation in the distal palatal mesenchyme (arrows in Fig. 2G,H) as well as in the tongue muscular (Fig. 2G,H) and epidermal cells (Fig. 2I) of the same section. These results indicate that SMAD2 phosphorylation does occur in the TGF-β3−/− mice but not in the MEE. Thus, a TGF-β signaling cascade initiated by TGF-β ligands other than TGF-β3 exists at these non-MEE sites. The results support our hypothesis that the temporospatial distribution of phospho-SMAD2 in MEE is regulated by TGF-β3.
To rule out the possibility that the absence of phospho-SMAD2 in MEE in TGF-β3−/− mice was due to an absence of SMAD2 in the null mutants, we conducted two experiments. First, we isolated the palate from TGF-β3−/− mice at E13 when the palatal shelves were in a vertical orientation and at E14.5 when the opposing palatal shelves touch in the midline and processed both sets of specimens for Western blot analysis. SMAD2 was present in palate tissue in TGF-β3−/− mice at the same levels as in the TGF-β3+/+ at both developmental stages (Fig. 3A). By using immunohistochemistry, SMAD2 remained in the basal MEE cells of the TGF-β3−/− palate at E14.5 (Fig. 3B). The distribution of SMAD2 in the TGF-β3−/− palate tissue supported the hypothesis that the absence of TGF-β3-triggered phosphorylation of SMAD2 is one of the components contributing to failure of palatal fusion in TGF-β3−/− mice.
Phospho-SMAD3 Is Not Detectable in MEE During Palatal Fusion
SMAD3 is another candidate to mediate TGF-β intracellular signaling (Heldin et al., 1997; Massagué, 1998). To characterize the distribution pattern of SMAD3 and its activated form, phospho-SMAD3, in the developing palate, we examined embryonic tissues from Swiss-Webster mice by immunohistochemistry. SMAD3 was present in palate, primarily the MEE and oral epithelium beginning at E14 (Fig. 4A). SMAD3 remained in MEE as palatal fusion proceeded (Fig. 4B). By E15 when MEE disappeared from the midline, SMAD3 was lost correspondingly but remained in the palatal oral epithelium (Fig. 4C). In contrast to a temporospatial distribution pattern of SMAD3 in MEE, phospho-SMAD3 was not detected in the palate during palatal fusion (Fig. 4D,E). Nuclear stain was intense in epidermal cells (Fig. 4F) and the maxilla region (Fig. 4D) of the same section. Absence of phospho-SMAD3 in MEE eliminates SMAD3 as the mediator of TGF-β3–regulated MEE fate during palatal fusion. Examination of SMAD3 in TGF-β3−/− palatal tissues did not provide evidence that there was a compensatory increase on SMAD3 in the null mutants (Fig. 5). The absence of SMAD3 phosphorylation during palatal fusion and the absence of compensatory change in TGF-β3−/− indicate SMAD2 as a primary regulator of MEE fate.
MEE Proliferation Is Inhibited by TGF-β3
TGF-β has been implicated in the inhibition of MEE proliferation. An in vitro study showed that the MEE stopped DNA synthesis prior to their disappearance from the midline (Gehris and Greene, 1992). TGF-β1, and to a lesser extent TGF-β2, administered in organ culture resulted in precocious cessation of MEE DNA synthesis followed by elimination of the MEE (Gehris and Greene, 1992). Addition of anti–TGF-β neutralizing antibody inhibited TGF-β–mediated MEE differentiation by blocking not only exogenous TGF-β1/-β2 but also endogenous TGF-βs, and the palatal shelf exhibited an intact MEE with active cell proliferation (Gehris and Greene, 1992). To specify the role of endogenous TGF-β3 in the inhibition of MEE proliferation during palatal fusion in vivo, we examined DNA synthesis of MEE in TGF-β3−/− mice at E14.5 by BrdU incorporation analysis and compared the patterns with TGF-β3+/+ littermate controls. TGF-β3−/− palate tissues had more BrdU-incorporated cells along the MEE than in the TGF-β3+/+ tissues. While the palatal shelf remained apart, BrdU-incorporated cells were easily identified on the tip of palatal epithelium of TGF-β3−/− (Fig. 6A) but not in the TGF-β3+/+ tissue (Fig. 6B). As the paired palatal shelves contacted in the midline, the BrdU-positive labeling remained in MEE cells in TGF-β3−/− mice (Fig. 6C) but few were present in the TGF-β3+/+ MEE (Fig. 6D). TGF-β3+/+ MEE remained a relative negative image of proliferating cells in the midline as long as the MEE seam was intact. We quantified the ratio of BrdU-labeled cells in MEE at a specific stage as showed in Fig. 6C,D and compared TGF-β3+/+ and TGF-β3−/− results. We found that the ratio of BrdU-labeled cells in MEE was significantly reduced in wild-type palate tissue compared with the null mutant (t-test, P < 0.05; Fig. 6E). The results provide evidence that MEE proliferation is inhibited by TGF-β3 expression in these cells.
Murine palatal fusion possesses two key features. One is the fate of MEE (Shuler, 1995) and the other is the requirement for TGF-β3 (Kaartinen et al., 1995; Proetzel et al., 1995). Previous studies suggested that the specific effects of TGF-β3 in MEE were controlled by TβR-III modulating the access of TGF-β ligands to signaling receptors (Cui and Shuler, 2000). Further studies were needed to demonstrate that the receptor substrate SMADs, the signaling effectors, were indeed expressed and phosphorylated in the MEE cells. SMAD2 and SMAD3 were both present in the MEE, whereas only SMAD2 was phosphorylated during normal palatal fusion. Phospho-SMAD2 was present in the MEE immediately before the opposing palatal shelves contacted and was restricted to the MEE until these cells disappeared from the midline. SMAD2 phosphorylation was temporospatially correlated with the disappearance of MEE, suggesting a role of SMAD2 in the regulation of this process. The nuclear staining of phospho-SMAD2 in the MEE cell is consistent with the role of this signaling molecule as transcription factor. Focusing on E14.5, we noticed that phospho-SMAD2 was selectively absent in the MEE in TGF-β3-null mutant mice, although nonphosphorylated SMAD2 was present at a wild-type level. The absence of phospho-SMAD2 was correlated with the persistence of MEE and failure of palatal fusion. These results support the hypothesis that TGF-β3 is a key initiator of phosphorylation of SMAD2, and subsequently, the onset of disappearance of MEE. Phospho-SMAD3 was absent in MEE during normal palatal fusion, thus ruling out SMAD3 as a mediator of TGF-β3–regulated MEE fate.
TGF-β has been shown to inhibit MEE proliferation in vitro (Gehris and Greene, 1992). The effect of endogenous TGF-β3 on MEE proliferation in vivo was unknown. The TGF-β3 knockout mouse represents an ideal model to address this issue. The MEE in TGF-β3+/+ palate tissue were in a nonproliferating state, whereas the MEE in TGF-β3−/− palatal tissue remained in active proliferation at developmental stage with opposing palatal shelves touching in the midline. The number of BrdU-labeled cells in MEE was significantly reduced in wild-type palate tissue compared with the null mutant (t-test, P < 0.05). This result is consistent with the reduction of cell death in MEE in TGF-β3–null palates compared with the wild-type at the time of fusion (Martinez-Álvarez et al., 2000). Attempts to rescue TGF-β3−/− palatal fusion have found that addition of recombinant TGF-β1 or TGF-β2 in culture were not sufficient to direct palatal fusion to the same degree as recombinant TGF-β3 (Taya et al., 1999). In summary, these results suggest that TGF-β3 is an essential inhibitor of MEE proliferation during palatal fusion. Cessation of DNA synthesis is likely the first and common step required by all MEE cells to progress to a differentiation fate. In the absence of TGF-β3, inhibition of MEE proliferation cannot be triggered efficiently and the MEE cells are not able to progress to either programmed cell death or epithelial–mesenchymal transformation/migration.
It has been implicated that the TGF-β effect in the inhibition of epithelial proliferation was mediated by participation of a TβR-I–activated phospho-SMAD (Hartsough and Mulder, 1997; Heldin et al., 1997; Massagué, 1998). SMAD2 phosphorylation coincided with the inhibition of MEE proliferation in TGF-β3+/+ mice in the process of MEE disappearance, whereas lack of phospho-SMAD2 coincided with active MEE proliferation in TGF-β3−/− mice at the developmental stage correlated with the persistence of the MEE. It is reasonable to believe that inhibition of MEE proliferation is mediated by TGF-β3–dependent phosphorylated SMAD2.
Two strains of timed pregnant mice were used in the present study. Swiss-Webster mice were used to examine the normal expression pattern of SMADs in the developing palate. TGF-β3 knockout mice (C57BL/6J-Tgfβ3tm1Doe, Jackson Laboratories, Bar Harbor, ME) were used to characterize the presumed relationship between TGF-β3, SMADs and cell proliferation in the MEE from homozygous mutant (TGF-β3−/−) vs. wild-type (TGF-β3+/+). TGF-β3 heterozygous mice were cross-mated to generate embryos. Mice were killed between embryonic days E12 and E15 as previously described (Cui et al., 1998). Each of the fetuses was carefully rinsed in phosphate buffered saline (PBS) and kept separately to avoid potential contamination from maternal tissue or littermates before and after removal from the amniotic sac. Fetal genotype was determined by polymerase chain reaction (PCR) using genomic DNA extracted from the embryonic body. Primers used for PCR were located in intron 5 and intron 6, respectively (Taya et al., 1999). PCR products were visualized as a 400-bp fragment for the wild-type allele and a 1,300-bp fragment for the mutated allele on an agarose gel stained with ethidium bromide. Two sets of DNA extraction followed by PCR were conducted for all individual samples for confirmation.
Western Blot Analysis
Palatal shelves were dissected from embryonic heads. To better analyze the MEE region, the palatal tissues were dissected either close to the tip of the shelf or around the midline of the palate, depending on the development stage. Tissues were lysed with a boiling solution containing sodium dodecyl sulfate and 0.1 M dithiothreitol. The total protein in each palatal sample was loaded in each well on a polyacrylamide gel. Electrophoresis was carried out in a modular mini-Protean II electrophoresis system (Bio-Rad, Hercules, CA). Protein was then transferred to a Millipore Immobilon-P membrane using a Bio-Rad mini-transblot electrophoretic transfer cell. The subsequent Western blot was performed by using a chemiluminescent Western blot kit (Boehringer Mannheim, Indianapolis, IN). The primary antibody against SMAD2 (mouse monoclonal IgG, Transduction Laboratories) or SMAD3 (rabbit polyclonal IgG, Upstate Biotechnology) was incubated with the membrane overnight at 4°C. The result was recorded on Kodak X-Omat AR film. The primary antibody (goat polyclonal IgG) against β-actin (Santa Cruz Biotechnology, Inc.) was used as internal control.
The embryonic heads were fixed in 4% paraformaldehyde–PBS at 4°C, followed by routine procedures for embedding in paraffin. Coronal sections (6 μm) were mounted in serial order on poly-L-lysine coated slides. The primary antibody was incubated on the tissues overnight at room temperature. The specific anti-SMAD2 (goat polyclonal IgG) and anti-SMAD3 (rabbit polyclonal IgG) antibodies were purchased from Santa Cruz Biotechnology, Inc., and Upstate Biotechnology, respectively. The anti–phospho-SMAD2 antibody (goat polyclonal IgG), which has been widely used in various studies (Nakao et al., 1999; Stopa et al., 2000; Esparza-López et al., 2001; Ito et al., 2001), was provided by Dr. C.-H. Heldin. The anti–phospho-SMAD2 antibody recognizes SMAD2 phosphorylated at serines 465 and 467. These amino acids are the phosphorylation target of the activated TβR-I kinase (Abdollah et al., 1997; Souchelnytskyi et al., 1997). The anti–phospho-SMAD3 (goat polyclonal IgG) antibody recognizes SMAD3 phosphorylated at serines 433 and 435, purchased from Santa Cruz Biotechnology, Inc. A Histostain kit (Zymed Laboratories, Inc., South San Francisco, CA) was used to carry out the immunohistochemical analysis as described previously (Cui et al., 1998). Normal serum was used as a negative control. Each of the experiments was repeated at least three times to demonstrate the consistency of the results.
Evaluation of DNA Synthesis Activity in MEE
TGF-β3 heterozygous pregnant mice were injected intraperitoneally with BrdU (Sigma) at 100 μg per gram body weight on E14.5. BrdU can be selectively incorporated into cellular DNA during S-phase and used to examine the DNA synthesis and cell proliferation of MEE. Mice were killed 2 hr after injection. The embryonic heads were processed for serial coronal sections (6 μm) and prepared for immunohistochemistry. Every fifth section was used for BrdU staining. Cells that have incorporated BrdU into DNA were detected by using a monoclonal antibody against BrdU and an enzyme-conjugated secondary antibody (HISTOMOUSE kit, Zymed Laboratories, Inc.). The number of BrdU-positive cells in the MEE and the total number of MEE cells per section were counted. The percentage of total BrdU-positive MEE cells over total number of MEE cells per head was then calculated and compared between TGF-β3+/+ and TGF-β3−/− groups. In total, 150 sections from eight heads (four from each group) were evaluated and BrdU-positive cells were counted.