Epithelial-mesenchymal transition (EMT) is a well-established phenomenon during embryonic development, wound healing, and tumor metastasis (Thiery,2003; Kang and Svoboda,2005; Khew-Goodall and Wadham,2005; Thiery and Sleeman,2006). It encompasses a spectrum of intra- and inter-cellular events: loss of polarity and cell-cell junction in epithelia; degradation of basal lamina; rearrangement of cytoskeleton and mobility of epithelial cells (Hay,1995). Palatal fusion is a widely studied developmental EMT model. After the outgrowth of the two palatal shelves from the maxillary processes and apposition in the midline, the two epithelial sheets form a midline epithelial seam (MES) (Ferguson,1988; Thiery and Sleeman,2006). The medial edge epithelia (MEE) eventually disappear to achieve mesenchymal confluence. Although still controversial, EMT has been proposed as one of the mechanisms that may regulate the fate of MEE cells. EMT was demonstrated by cell tracking with dyes (CCFSE and DiI) (Griffith and Hay,1992; Sun et al.,1998,2000) or morphological appearance using transmission electron microscopy (Fitchett and Hay,1989).
EMT is orchestrated by a complex network of signaling molecules including various growth factors, transcription factors, and extracellular matrix (ECM) activators (Thiery and Sleeman,2006). TGFβ3 has been localized to the MEE and stimulated palatal fusion in chick embryos (Sun et al.,1998) through Smad-dependent or independent pathways (Nawshad and Hay,2003; Nawshad et al.,2004; Kang and Svoboda,2002). An alternative downstream effector of TGFβ3 signaling, PI-3K, has been identified in actin reorganization, matrix metalloproteinase (MMP) production, and cell mobility (Metzner et al.,1996; Sugiura and Berditchevski,1999) and delayed palatal fusion in mice (Kang and Svoboda,2002).
The hallmark of EMT in development and cancer is the down regulation of E-cadherin, which causes the loss of E-cadherin-dependent intercellular epithelial junctional complexes and E-cadherin-mediated β-catenin sequestering in the cytoplasm. Several EMT transcription factors down regulate E-cadherin expression including: Twist, Snail, and Slug. A correlation between Twist and tumor metastasis was identified in human breast cancer pathogenesis (Yang et al.,2004). Using a murine breast tumor model, it was determined that Twist plays an essential role in metastasis. In addition, suppression of Twist expression in highly metastatic mammary carcinoma cells specifically inhibited their ability to metastasize from the mammary gland to the lung. Ectopic expression of Twist resulted in loss of E-cadherin-mediated cell-cell adhesion, activation of mesenchymal markers, and induction of cell motility, suggesting that Twist contributed to metastasis by promoting EMT. In human breast cancers, high Twist expression correlated with invasive lobular carcinoma, a highly infiltrating tumor type associated with loss of E-cadherin expression. Furthermore, Twist mRNA was recently reported in the palatal tissue of normal mice by RNA analysis or in situ hybridization (Pungchanchaikul et al.,2005; Rice et al.,2005).
We hypothesize that Twist may function in palatal fusion by regulating MEE cell transdifferentiation. In this study, we examined Twist expression during mouse palatal fusion and analyzed the possible signaling pathways that regulate Twist expression, such as TGFβ3 and PI-3K. We demonstrated that Twist protein was expressed in the MEE before fusion. Twist-specific siRNA decreased the protein levels in palatal extracts and decreased palatal fusion. We also show that decreasing TGFβ3 and PI-3K activity decreased Twist mRNA expression in mice and, in contrast, exogenous TGFβ3 increased Twist mRNA expression in chicken palatal shelves. Our results suggest that Twist may be one regulator during palatal fusion.
Twist Protein Is Expressed in the Epithelium at Pre-Fusion Stages Both In Vivo and In Vitro
In vivo, Twist protein was detected in the epithelial layer of MEE and the future nasal and oral epithelium at different stages. Only a few Twist-labeled cells were in the epithelium at E13.5 when the palatal shelves were still open and oriented in the horizontal position (Fig. 1A,B). Twist protein increased at the pre-fusion stage (E14.5) when the two palatal shelves were approximated and had just made contact in the midline (Fig. 1C,D). As the fusion proceeded and the epithelial seam dispersed, scattered islands of labeled cells were present along the midline, as well as in the oral and nasal epithelium. The MEE and oral epithelium had more Twist than the nasal epithelium (Fig. 1E,F). After the mesenchyme converged completely, Twist decreased (Fig. 1G,H). However, throughout palatal fusion, the mesenchyme cells expressed Twist. Twist protein was located in the cytoplasm in a paranuclear pattern. Twist protein was also in some MEE cells at the fusion zone after 24-hr culture in vitro (see Supplemental Fig.1, which can be viewed at www.interscience.wiley.com/jpages/1058-8388/suppmat), which was equivalent in the pre-fusion stage (E14.5) in vivo. Twist protein was either closely related to or overlapped with the nuclei.
Palate Fusion Decreased in Palatal Shelves Cultured With Twist-Specific siRNAs
To determine the function of Twist protein in the process of palatal fusion, we decreased expression using specific siRNAs. The siRNA transfection efficiency was monitored with a fluorescent oligo (Block-iT™, Invitrogen). The oligo (200 nM) was transfected in the tissue, incubated for 24 hr, and viewed as whole mount with a confocal microscope. The oligo penetrated into the tissue (Suppl. Fig. 2) throughout the epithelium and mesenchyme. Twist siRNAs down regulated Twist protein over 80% in the first 24 hr compared to untreated controls or palates treated with a scrambled siRNA (Scr) (Fig. 2, Suppl. Fig.3). Histological analysis demonstrated palatal fusion after 72 hr of organ culture in controls. Due to the temporal-spatial delay in palatal fusion, there was variance in the fusion level from anterior to posterior of individual palatal cultures. In the control groups (controls and Scr controls), samples achieved complete fusion on most sections, with the highest complete fusion in the middle region of the palate. The mesenchyme was confluent, the basal lamina surrounding the midline epithelial seam (MES) was degraded, and the MEE disappeared from the midline with no epithelial islands (Fig. 3A). However, some sections from the posterior of the tissue had remaining MES in a few sections (data not shown). The palates treated with scrambled siRNA after 72 hr culture had similar fusion patterns as the control group (Fig. 3A). In palates treated with 100 nM Twist siRNA, a larger epithelial triangle remained and less than 1/3 of the mesenchyme achieved confluence (Fig. 3A). In the presence of 200 nM Twist siRNA, MEE cells were observed at the midline, forming a persistent midline seam (arrow) with the surrounding basal lamina remaining on most sections (Fig. 3A). However, we found that some siRNA-treated palates were completely fused, suggesting that Twist may not be the only factor that regulates MEE transition in palatal fusion.
The degree of palatal fusion was quantified with a scaling system, termed the mean fusion score (MFS). Briefly, every 20th section was mounted, stained with H&E, and scored. Fusion scores were assigned to the coronal sections from anterior to posterior according to the standard previously described with slight modifications (Kang and Svoboda,2002) where completely fused palate equals 5 and unfused palate equals 1 (Fig. 3B). The average score of all sections from one palate specimen was designated as the fusion score for that sample. The MFS for all palatal specimens in each group (Table 1, total n = 5–12) were compared. The MFS was compared between groups by using the Kruskal-Wallis test. P < 0.05 was considered statistically significant. There was no significant difference between control, scrambled siRNA, and 100 nM Twist siRNA groups. However, the 200 nM Twist siRNA group was significantly different than control groups.
Table 1. Mean Fusion Score (MFS) of Mouse Palate After 72-Hr Culture of siRNAa
P < 0.05.
This is the method for calculating the MFS and it is explained in the Experimental Procedures section. After 72-hr culture, the palatal shelves were harvested for histological analysis. Sections (8 μm) were cut and 1 of every 20 sections was mounted and examined with light microscopy. A fusion score was assigned to every section. An average score of all the sections (n = 20 sections/sample) from the same sample was designated as the fusion score of that sample. The mean fusion score was calculated by averaging all the scores from the same treatment group. Table 1 shows the treatment groups, MFS of each group, and the number of embryos examined in each group. A Kruskal-Wallis test was used to analyze the MFS. P < 0.05 was considered statistically significant. No statistic significance was found between control, scrambled siRNA control, and 100-nM Twist siRNA groups. However, MFS of 200-nM Twist siRNA group was significantly different from control group. Scr stands for control group with siRNA of scrambled sequence.
Subcellular Localization of β-Catenin Changed With Twist Down Regulation
As Twist regulates the expression of epithelial and mesenchymal markers in the EMT process including down-regulation of E-cadherin and β-catenin, we examined both markers in MEE after a 24-hr Twist siRNA treatment. The expression level of E-cadherin and β-catenin had no obvious change on Western blots (Fig. 4). After 24 hr in vitro, E-cadherin was between epithelial cells with no obvious differences in distribution between control and treatment groups (Fig. 5). However, β-catenin protein expression patterns changed in response to down regulating Twist (Fig. 6). In no treatment and scrambled siRNA controls and 100 nM Twist siRNA groups, the β-catenin had membrane (arrows), cytoplasmic and some nuclear staining in confocal single optical sections (arrowheads, Fig. 6A–I). However, in the palates treated with 200 nM Twist siRNA, there was less β-catenin with nearly no cytoplasmic distribution (Fig. 6J–L). Faint membrane-bound β-catenin (arrows, Fig. 6K) at the cell-cell adhesion sites was detectable. β-catenin was also located in some mesenchymal cells in a punctate paranuclear pattern (asterisks, Fig. 6B,C).
Twist mRNA Expression Levels Changed in Response to TGFβ3 and PI-3K Signaling
TGFβ3 is considered one of the “master genes” in palatal fusion (Nawshad et al.,2005; see review in this issue, Nawshad,2008). To determine if Twist is downstream of TGFβ3 in palatal fusion, two experimental approaches were used. First, we incubated mouse palates in media containing a neutralizing antibody specific for TGFβ3 and found that Twist mRNA expression levels were down-regulated in a dose-dependent manner in 24 hr (Fig. 7A). Second, exogenous TGFβ3 was added to chicken palate organ cultures, which fused as previously described (Sun et al.,1998). Tissues were harvested from 15 min to 72 hr after TGFβ3 stimulation. Twist mRNA expression increased in the 3- and 6-hr groups, and then dropped to baseline levels (Fig. 7B). We had previously shown that PI-3K activity was necessary for palatal fusion as the specific inhibitor, LY294002, delayed murine palatal fusion in organ culture (Kang and Svoboda,2002). This same PI-3K inhibitor also downregulated Twist mRNA expression in a dose-dependent manner (Fig. 7C).
The Expression Pattern of Twist in Palatal Development
Twist, a basic helix-loop-helix transcription factor was first described in Drosophila as a gene essential for dorsoventral polarity (Thisse et al.,1987,1988). Previous investigators reported Twist was required in head mesenchyme for cranial neural tube morphogenesis (Chen and Behringer,1995) and had a restricted expression pattern in mesenchymal cells during murine palatogenesis and tooth development (Rice et al.,2005). Twist was also detected in ectodermal and endodermal cells in specific stages during ventral furrow formation (Thisse et al.,1988; Leptin and Grunewald,1990). Ectopic expression of Twist in the ectoderm promoted a switch from epidermal and nervous system differentiation to a myogenic program, thus establishing a requirement for Twist in cell-fate choice (O'Rourke and Tam,2002). Twist was also expressed in wild type mammary epithelial cells (Howe et al.,2003) and up regulated in several epithelial cancers including breast, prostate, and gastric carcinomas (Rosivatz et al,2002; Watanabe et al,2004; Kwok et al,2005; Martin et al.,2005). These studies suggest that Twist is not restricted to specific cell types but, instead, the molecule may promote cell differentiation by modifying cell shape to facilitate mobility. Thus, there is controversy over the cell type and intracellular distribution of this protein in developing tissues. In many developing tissues, Twist is considered a mesenchymal marker. However, our data indicate that Twist protein is present in MEE cells during specific palatal developmental stages. The staining pattern appeared to be specific, as the palatal stages before and after the “first contact” stage had different distribution patterns. Our results not only show cell type differences, but also an intracellular location difference.
As a transcription factor, Twist would be expected to function in the nucleus, but our data indicate that, while there is expression of the molecule in the nucleus, much of the Twist immunolocalization is in the cytoplasm. The cytoplasmic form may represent molecules undergoing ubiquitination for ultimate degradation (Demontis et al.,2006); however, other investigators have reported cytoplasmic (and nuclear) Twist immunolocalization (Kwok et al.,2005; Kida et al.,2007). Alternatively, Twist may have cytoplasmic functions like β-catenin in the developing palate. It has also been suggested that Twist forms dimers with distinct developmental functions (Castanon et al.,2001). The immunostaining presented here is considered valid because it is specific in that Twist is only expressed in a particular temporal window in palatal epithelium. Further studies are needed to investigate the cytoplasmic role of Twist in the MEE.
The cellular mechanisms underlying midline seam degeneration have been a major focus in recent palatogenesis studies and also the topic of a review in this issue by Nawshad (2008). Three mechanisms have been proposed for palatal seam degeneration: lateral migration (Carette and Ferguson,1992), EMT (Sun et al.,1998; Kang and Svoboda,2002; Nawshad and Hay,2003; Martinez-Alvarez et al.,2004; Nawshad et al.,2004), and apoptosis (Cuervo et al.,2002; Cuervo and Covarrubias,2004). Morphological evidence for EMT in palatogenesis is based on identifying transforming MEE cells with extended filopodia in electron micrographs (Fitchett and Hay,1989). In addition, cell lineage dyes, such as DiI and CCFSE, were used to trace MEE cells. Labeled cells were found in the mesenchyme after palatal fusion (Griffith and Hay,1992; Shuler et al.,1992). Genetically labeling the epithelial cells with a cytokeratin 14 promoter driving a Lac Z reporter has produced confusing results. One study confirmed the occurrence of EMT (Jin and Ding,2006), and another concluded that EMT did not occur (Xu et al.,2006). Therefore, further studies to visualize the EMT process in living tissues need to be performed.
The Role of Twist During EMT in MEE Cells
Several transcription factors necessary for the EMT process are involved in palatal development. Lymphoid-enhancing factors1 (LEF1) is involved in transformation of epithelium to mesenchyme in palatal fusion by forming transcription complex with p-Smad2 and Smad4 (Nawshad et al.,2007). Snail, a zinc finger transcription factor, directly binds to the consensus sequences CAGGTG and suppresses E-cadherin expression (Peiro et al.,2006) and is expressed in MEE cells during palatal fusion (Martinez Alvarez et al.,2004). In this report, we show that Twist protein is also expressed in a selected population of cells within the fusion zone when the two palatal shelves make first contact, indicating that those cells are probably changing morphology and preparing to migrate. In experiments that down regulated Twist expression, palatal fusion is either blocked or delayed. There is a significant difference in palatal fusion as measured by the mean fusion scores of the 200 nM Twist siRNA group compared to controls (Table 1). The variability in palatal fusion in the Twist siRNA groups may be due to concentration effects. In our study, the siRNA concentrations were relatively low, compared to analogous studies (Nakajima et al.,2007). Our results also suggest that Twist may not be the only factor regulating EMT during palatogenesis.
Ectopic Twist expression in mammary cancer cells induces suppression of E-cadherin and β-catenin (Yang et al.,2004). We expected that E-cadherin and β-catenin would be sustained or increased in the MEE cells when Twist was down regulated. However, we did not detect an obvious change in either protein on Western blots or subcellular localization in the first 24 hr. Interestingly, while β-catenin amounts were not increased, a change in its distribution was observed. Twist suppression decreased cytosolic β-catenin after 24 hr, suggesting β-catenin did not get removed from cell-cell junctions when Twist was present in lesser amounts. β-catenin also serves as a transcriptional activator (Gottardi and Gumbiner,2004; Bienz,2005). The Twist promoter has been shown to respond to β-catenin when Twist is up regulated by Wnt activation in mouse mammary epithelial cell lines (Howe et al.,2003). Therefore, our results suggest that a reciprocal regulation between Twist and β-catenin may exist. Also, β-catenin was present in some mesenchymal cells just underneath the MEE cells (asterisks, Fig. 6). Since cytoplasmic β-catenin is usually considered evidence of active canonical Wnt signaling (Bienz,2005), this observation suggests that there may be crosstalk between TGFβ signaling in the MEE and Wnt signaling in the underlying mesenchyme. It was recently shown that Gsk3β conditional knockout mice have a secondary palatal cleft (Liu et al.,2007). Therefore, future experiments could explore possible crosstalk pathways between Twist and Wnt pathway proteins such as Gsk3β.
TGFβ3 and PI-3K Signaling Regulates Twist Expression in Palate
The palates of TGFβ3 null mice fail to fuse (Kaartinen et al.,1995), indicating an essential role for TGFβ3 in palatogenesis. Exogenous TGFβ3 induces chicken palate fusion by facilitating midline seam epithelium transformation (Sun et al.,1998). The signaling pathways downstream of the TGFβ receptors have been investigated by several groups. TGFβ3 signals through its receptors to the Smad pathway to up-regulate synthesis of LEF1 and to activate LEF1 transcription during palatal fusion (Nawshad and Hay,2003). In this study, we demonstrated that Twist mRNA expression was repressed in the presence of the TGFβ3 neutralizing antibody. In contrast, Twist mRNA was upregulated following TGFβ3 stimulation in chicken palates supporting the hypothesis that Twist is downstream of the TGFβ3 signaling.
In addition, PI-3K was identified in TGFβ-induced down regulation of cell-cell adhesion (Vogelmann et al.,2005). PI-3K is required for actin reorganization, MMP production, and cell mobility (Metzner et al.,1996) and necessary for palatal fusion (Kang and Svoboda,2002). Our mRNA results support the theory that Twist is also downstream of PI-3K signaling during mouse palatal fusion. Akt, a known downstream effector for PI-3K signaling, also regulates Twist mRNA expression during palatal fusion (personal communication, Dr. R. Spears).
In summary, our results demonstrate that Twist protein is transiently expressed in the palatal epithelium and involved in regulating the fate of MEE cells. We also provide conclusive evidence that Twist mRNA responds to TGFβ3 and PI-3K. We hypothesize that Twist acts downstream of the TGFβ3 signaling pathway by regulating the EMT of MEE.
Animal Preparation and Organ Culture
Timed-pregnant CD1 mice (Harlan Sprague-Dawley, Inc.) and fertile chicken eggs (Texas A & M Poultry Science Department) were used in these studies. Mice were maintained under standard conditions at a 10:14-hr light:dark cycle. Female mice were mated overnight, and the day of vaginal plug was timed as day 0. In CD1 mice, the palate shelves elevate above the tongue between embryonic day E13 and E13.5. The mouse embryos were dissected by cervical dislocation at E13.5, and the embryos were dissected from the amniotic sacs into a dish of Hanks' balanced saline solution (HBSS; Gibco) at E13.5. The chicken eggs were incubated for 8 days at 39°C before the embryos (Hamburger-Hamilton stages 27–34) were removed from the eggs and rinsed in Hanks' balanced saline solution (HBSS; Gibco). Palatal shelves were dissected from embryos and placed nasal side down on polycarbonate membranes (Nucleopore Corp.) with their medial edges in contact. The tissues were cultured with BGJb medium (Gibco) at the air–media interface on a triangular-shaped wire grid in an organ culture dish. Tissues were incubated at 37°C in a humidified gas mixture (5% CO2 and 95% air).
Transfection of siRNA.
The siRNA oligonucleotide specific for Twist (NM_011658) and Snail (NM_011427) was produced by and purchased from Ambion. The transfection efficiency of various combinations of siRNA concentrations and the delivery reagent, Lipofectamine 2000 (Invitrogen), were optimized by using a fluorescent tagged oligo (BLOCK-iT™, Invitrogen). RNA interference using siRNA was performed following the manufacturer's protocol (Invitrogen). Concentrations of 100 and 200 nM siRNA and 0.1% Lipofectamine were used in organ culture. A 21nt siRNA oligonucleotide A 21-nt siRNA oligonucleotide with scrambled sequence was used as a negative control. Tissues were exposed to siRNA treatment for up to 72 hr. The Twist siRNA sequence was 5′-GGUACAUCGACUUCCUGUAtt.
TGFβ3 neutralizing antibody (R&D Systems) of 1 and 10 μg/ml and LY294002 (Calbiochem®) of 1 and 10 μM were added to the culture medium. Tissues were cultured for 24 hr and mRNA from 3 pairs of whole palatal shelves was processed for mRNA expression analysis. TGFβ3 (50 ng/ml) was added to the chicken palatal organ culture for various times (15 min to 72 hr).
Histology and Immunofluorescent Staining:
Cultured palatal shelves were collected at 24, 48, and 72 hr and fixed in freshly prepared 4% paraformaldehyde/phosphate buffered saline (PFA/PBS; pH 7.4) for 30 min. After rinsing in PBS, the tissues were processed for paraffin embedding. Serial sections (8 μm) were collected and numbered in sequence. Every 1 out of 20 sections was mounted on the slides and stained with Hematoxilyn & Eosin (H&E). A fusion score was given to each section according to the standard previously described (Kang and Svoboda,2002). The average of 20 sections' scores was calculated as the fusion score of one sample. The MFS (mean fusion score) for each treatment group was calculated. Light microscope images were captured by Zeiss Axioplan.
For immunofluorescent staining, both in vivo and in vitro samples were used. Slides were deparaffinized and rehydrated. After blocking with 10% normal donkey serum/PBS, the tissues were incubated in 1:100 polyclonal antibodies for Twist (Santa Cruz), Snail (Abcam), E-cadherin (Cell signaling), and β-catenin (Abcam) overnight at 4°C or 1 hr at room temperature. After rinsing, the primary antibody was detected by secondary antibodies conjugated with Alex Fluoro 488 or 555 (Molecular Probes). The nuclei were counterstained with ToPro-633 (Molecular Probes). The images were collected and analyzed with a Leica TCS-SP2 confocal microscope and arranged with Adobe Photoshop.
Whole palatal shelves were lysed in the RIPA buffer (RadioImmunoPrecipitation Assay Buffer, Sigma) and used for Western blots. The concentrations of total protein in each sample were determined by BCA assay (Pierce). Total protein (10 μg) was loaded in each well on a 4–12% NuPage Bis-Tris gel (Invitrogen). Protein was transferred to PVDF membrane (Millipore). The membrane was incubated with polyclonal primary antibody against Twist (1:1,000, Santa Cruz), E-cadherin (1:2,000, Cell signaling), and β-catenin (1:2,000, Abcam) overnight at 4°C with agitation. Secondary antibodies conjugated with alkaline phosphatase were used at concentrations of 1:1,000–5,000. Signals were visualized with the chemiluminesecnt kit (Invitrogen), captured with the Kodak Image station 440CF (Kodak Digital Science™) or DAB (Vector Laboratories, Inc.).
RNA Extraction and Real-Time PCR
Tissues treated with TGFβ3 neutralizing antibody and PI-3K inhibitor were harvested after the 24-hr culture. Total RNA was extracted with RNeasy Mini Kit (Qiagen). The quantity and quality of mRNA were measured by Agilent 2100 Bioanalyzer. mRNA was reverse transcribed with SuperScriptII reverse-transcriptase (Invitrogen) and the resulting cDNA was diluted to 1:5 or 1:10 for quantified real-time PCR. The relative quantitation value was calculated by 2-ΔΔCt method. All quantifications were normalized to 18s (SuperArray). For each primer set, the running conditions were optimized. Experiments have been repeated 3 times. Primer pairs used for RT-PCR were obtained from Integrative DNA Technology (Table 2).
Table 2. Twist Primers for RT-PCR
We thank Sala Senkayi, Rebecca Kayne, and Cameron Cowan for their technical assistance. We thank Dr. Marion Gordon for critical review and editing the manuscript. We thank the Baylor Oral Health Foundation and March of Dimes Research grant (FY06-321) for the funds to support this work.