Research ArticleDevelopmental Biology

IRF6 and TAK1 coordinately promote the activation of HIPK2 to stimulate apoptosis during palate fusion

See allHide authors and affiliations

Science Signaling  06 Aug 2019:
Vol. 12, Issue 593, eaav7666
DOI: 10.1126/scisignal.aav7666

HIPK2 promotes palate fusion

The palate arises from bilaterally symmetric primordia that migrate toward one another and fuse during embryonic development. Fusion is not complete until the seam cells along the border where the two palatal shelves meet are eliminated, a process that depends on transforming growth factor–β3 (TGF-β3) and the transcription factor IRF6. Ke et al. found that IRF6 directly promoted expression of the gene encoding the kinase HIPK2 and that this was required for seam cell apoptosis during palate fusion in mice. Contact between palatal shelves induced the phosphorylation of the TGF-β–activated kinase TAK1, which in turn promoted the phosphorylation of HIPK2. Thus, IRF6 and TAK1 cooperatively promote the activity of HIPK2 to drive the final stage of palate fusion.

Abstract

Cleft palate is a common craniofacial defect caused by a failure in palate fusion. The palatal shelves migrate toward one another and meet at the embryonic midline, creating a seam. Transforming growth factor–β3 (TGF-β3)–induced apoptosis of the medial edge epithelium (MEE), the cells located along the seam, is required for completion of palate fusion. The transcription factor interferon regulatory factor 6 (IRF6) promotes TGF-β3–induced MEE cell apoptosis by stimulating the degradation of the transcription factor ΔNp63 and promoting the expression of the gene encoding the cyclin-dependent kinase inhibitor p21. Because homeodomain-interacting protein kinase 2 (HIPK2) functions downstream of IRF6 in human cancer cells and is required for ΔNp63 protein degradation in keratinocytes, we investigated whether HIPK2 played a role in IRF6-induced ΔNp63 degradation in palate fusion. HIPK2 was present in the MEE cells of mouse palatal shelves during seam formation in vivo, and ectopic expression of IRF6 in palatal shelves cultured ex vivo stimulated the expression of Hipk2 and the accumulation of phosphorylated HIPK2. Knockdown and ectopic expression experiments in organ culture demonstrated that p21 was required for HIPK2- and IRF6-dependent activation of caspase 3, MEE apoptosis, and palate fusion. Contact between palatal shelves enhanced the phosphorylation of TGF-β–activated kinase 1 (TAK1), which promoted the phosphorylation of HIPK2 and palate fusion. Our findings demonstrate that HIPK2 promotes seam cell apoptosis and palate fusion downstream of IRF6 and that IRF6 and TAK1 appear to coordinately enhance the abundance and activation of HIPK2 during palate fusion.

INTRODUCTION

Cleft palate is a common craniofacial defect that is caused by a failure in the fusion of the bilaterally symmetric embryonic palatal shelves. In mammals, the development of the secondary palate initially starts with the caudal growth of the two palatal shelves on either side of the tongue. The palatal shelves subsequently migrate over the tongue and grow medially toward one another, make contact, and form the medial edge seam (MES) at the midline (1). The medial edge epithelial (MEE) cells present at the MES subsequently disappear, which leads to palate fusion (1, 2). Three major mechanisms have been proposed for the disappearance of the MEE cells: epithelial-mesenchymal transition (EMT) (35), apoptosis (610), and migration of the MEE cells away from the seam (7, 11, 12). Failure of one or more of these processes would result in the formation of a cleft palate.

The proapoptotic protease caspase 3 is activated in MEE cells at the MES during palate fusion (9). Many studies, in both humans and mice, have shown that signaling initiated by transforming growth factor–β3 (TGF-β3) is required for the disappearance of the MEE cells and palatal shelf fusion by promoting EMT and apoptosis (1318). The transcription factor interferon regulatory factor 6 (IRF6) also plays an important role in palate closure and interacts with TGF-β signaling in this context. TGF-β signaling increases Irf6 expression in the developing palate through both a SMAD-dependent pathway and the p38 mitogen-activated protein kinase (MAPK) pathway (18). Knocking out the gene encoding TGF-β receptor type 2 (Tgfbr2) in the basal epithelium of mice results in decreased expression of Irf6 and Cdkn1a, which encodes the protein cyclin-dependent kinase inhibitor 1 (also called p21), and reduced apoptosis of MEE cells, which leads to failure of MES degeneration within the palate (18). Map3k7, which encodes mitogen-activated protein kinase kinase kinase 7 [also called TGF-β–activated kinase 1 (TAK1)], an important mediator of the TGF-β pathway, also plays an important role in the TGF-β3–induced apoptosis of MEE cells and palate fusion (19).

Mutations in IRF6 lead to van der Woude syndrome (VWS) and popliteal pterygium syndrome, both of which are characterized by cleft palate (2022), and single-nucleotide polymorphisms (SNPs) in IRF6 are associated with nonsyndromic cleft lip and palate (23, 24). Furthermore, Irf6-null and Irf6R84C mutant mice have defects in skin, limbs, and craniofacial development (25, 26). Ectopic expression of Irf6 rescues p21 expression, MEE apoptosis, and palate fusion defects in basal epithelium–specific Tgfbr2 knockout mice. Knockdown of Irf6 in palatal shelf organ culture enhances the expression of the ΔN isoform of Tumor protein 63 (Tp63), which encodes the p53-related transcription factor ΔNp63, and reduces the expression of the ΔNp63 target p21, leading to a lack of apoptosis during palate fusion (18). Thus, during palate fusion, IRF6 promotes MEE apoptosis by inhibiting ΔNp63-mediated repression of p21 (18).

IRF6 promotes ΔNp63 protein degradation in both mouse palatal shelves and human keratinocytes (27, 28). In cancer cells, phosphorylation of ΔNp63 at Thr397 by homeodomain-interacting protein kinase 2 (HIPK2) is required for ΔNp63 protein degradation (29). Chromatin immunoprecipitation-sequencing (ChIP-seq) and microarray analysis have shown that IRF6 binds to an intron of the HIPK2 gene and may promote HIPK2 expression in human normal keratinocytes (30). We therefore hypothesized that IRF6 promotes ΔNp63 degradation through activation of HIPK2 and investigated the role of HIPK2 during palate fusion. Our findings indicate that HIPK2 mediated IRF6-induced ΔNp63-p21 signaling to promote MEE apoptosis. Whereas IRF6 promoted the transcription of Hipk2, contact between palatal shelves stimulated the activity of TAK1, which in turn promoted HIPK2 phosphorylation. Thus, IRF6 and TAK1 coordinately promote the activity of HIPK2 at the MES, resulting in apoptosis of MEE cells and palate fusion.

RESULTS

HIPK2 is a downstream target of IRF6 in palatal shelves

To understand whether HIPK2 functions downstream of IRF6 during palate fusion, we first examined the distribution of HIPK2 in palatal shelves from embryonic day 13.5 (E13.5) to E15.5 mice. HIPK2 was present in the MEE cells of palatal shelves at E14.5 but was not detected in either epithelial or mesenchymal cells at E13.5 or E15.5 (Fig. 1A). The highest amount of Hipk2 mRNA during palate development was detected in palatal shelves at E14.5 (Fig. 1B). We previously reported that infecting cultured palatal shelves with a lentivirus expressing a short hairpin RNA (shRNA) targeting Irf6 (shIrf6) abolishes the production of IRF6 in MES cells (31). We examined Hipk2 expression after knockdown of IRF6 in palatal shelves in organ culture. In cultured normal palatal shelf pairs, the abundance of Hipk2 mRNA gradually increased, reaching an eightfold increase after 24 hours. IRF6 knockdown significantly reduced the abundance of Hipk2 mRNA in the palatal shelves after 12 hours of lentivirus infection (Fig. 1C and fig. S1, A to D). Compared to the shLuc control, the abundance of Hipk2 mRNA and HIPK2 protein was reduced to 2 and 20%, respectively, of the control value after 24 hours of lentivirus infection (Fig. 1, C and D). Furthermore, adenovirus-mediated overexpression of IRF6 increased the abundance of HIPK2 (Fig. 1D). These results placed IRF6 upstream of Hipk2 expression in mouse palatal shelves.

Fig. 1 IRF6 promotes the expression of Hipk2 during palate development.

(A) Immunofluorescence images showing HIPK2 in palatal shelves from E13.5, E14.5, and E15.5 C57BL/6 mouse embryos. Nuclei were counterstained with Hoechst 33258. Scale bar, 20 μm. (B) Results of quantitative RT-PCR showing the relative abundance of Hipk2 transcripts in palatal shelves dissected from E13.5, E14.5, and E15.5 C57BL/6 mouse embryos (n = 3 independent experiments with four pairs of palatal shelves per time point in each experiment). (C) Results of quantitative RT-PCR showing the relative abundance of Hipk2 and Irf6 transcripts in E13.5 palatal shelves at the indicated time intervals after infection with lentiviruses carrying plasmids encoding a control shRNA (shLuc) or shRNA targeting Irf6 (shIrf6) (n = 3 independent experiments with four pairs of palatal shelves in each treatment group). (D) Representative Western blot and quantification of HIPK2 and IRF6 in E13.5 palatal shelves infected with lentiviruses carrying shLuc (control) or shIrf6 plasmids or adenoviruses carrying empty vector (AdIE) or an IRF6 expression construct (AdIRF6). GAPDH is a loading control (n = 3 independent experiments with four pairs of palatal shelves in each treatment group). (E) Schematic diagram of intron 1 of the mouse Hipk2 gene showing the locations of enhancers 1 to 3 (E1 to E3) and putative IRF6 binding sites (BS-1, BS-2, and BS-3). (F) Each enhancer was fused to luciferase (Luc) and used for reporter assays in HEK293T cells expressing an empty vector control (Vec), wild-type IRF6 (IRF6-WT), or the R84C mutant form of IRF6 (IRF6-R84C) (n = 3 independent experiments). (G) ChIP analysis of DNA fragments immunoprecipitated with a control IgG, an IRF6-specific antibody, or a histone 3–specific antibody and PCR-amplified with primers flanking the putative IRF6 binding sites (data are representative of three independent experiments with seven to nine pairs of palatal shelves for each experiment). Statistical analyses were performed by one-way ANOVA. Error bars represent SD. *P < 0.05.

IRF6 enhances the expression of Hipk2

By ChIP-seq analysis, an IRF6-binding peak in intron 1 of the human HIPK2 gene is detected in primary normal human epidermal keratinocytes (30), which suggests that IRF6 binds to this element and may promote human HIPK2 expression. We identified several potential IRF6-binding elements in intron 1 of the mouse Hipk2 gene. Using clustering analysis of the UCSC (University of California, Santa Cruz) genome browser ChIP-seq densities of the activating histone mark H3K4me1, an indicator of active enhancers, in mouse embryonic craniofacial tissue with the aforementioned potential IRF6-binding elements within mouse Hipk2 intron 1, we identified three candidate enhancer regions in Hipk2 intron 1 (E1, E2, and E3) including the IRF6 binding sites 1, 2, and 3 (BS-1, BS-2, and BS-3), respectively (Fig. 1E). To test whether these are enhancers of mouse Hipk2, we individually cloned the E1, E2, and E3 elements into a luciferase reporter vector with a minimal promoter to test their enhancer activity. In 293T cells, ectopic expression of IRF6 significantly increased E1 and E2 enhancer activity but had no effect on the activity of the E3 element (Fig. 1F). The IRF6 R84C mutant, which lacks DNA binding activity, did not activate any of the candidate enhancers. ChIP analysis was performed to test whether IRF6 could bind to Hipk2 enhancer regions in palatal shelves. Chromatin was immunoprecipitated from tissue with an IRF6 antibody and subjected to polymerase chain reaction (PCR) analysis using primers specific for BS-1, BS-2, and BS-3 (Fig. 1E). Our results showed that IRF6 protein bound to the predicted binding sites in the E1, E2, and E3 regions of mouse Hipk2 in palatal shelf tissue (Fig. 1G). The results suggest that IRF6 directly promotes Hipk2 expression in mouse palatal shelves.

HIPK2 contributes to IRF6-promoted apoptosis during palate fusion

To address the role of HIPK2 in IRF6-promoted palate fusion and apoptosis, we tested whether knockdown of HIPK2 blocked IRF6-promoted palate fusion. Compared to shLuc controls, shHipk2 decreased the amount of fusion (AOF) observed in pairs of palatal shelves, as assessed by the disappearance of the seam, to 74.3 ± 15.7% at 48 hours after lentivirus infection (fig. S2). As expected, adenovirus-mediated IRF6 overexpression (AdIRF6) promoted palatal shelves fusion (Fig. 2A), but fusion was significantly decreased at 24 hours by coinfection with shHipk2 lentivirus (from 35.3 ± 4.9% to 14.9 ± 2.8%) and 48 hours (from 92.1 ± 3.3% to 78.4 ± 5.7%) after infection (Fig. 2, A and B, and fig. S2). Furthermore, shHipk2 lentivirus infection of palatal shelves decreased the staining intensity of p21 by 88% and the staining intensity of active caspase 3 by 83% (Fig. 2, A and C). Ectopic expression of IRF6 increased the staining intensity of p21 by 2.1-fold and the staining intensity of active caspase 3 by 1.7-fold (Fig. 2, A and C). In addition, IRF6-promoted p21 induction and caspase 3 activation were blocked by shHipk2 lentivirus infection (Fig. 2, A and C). Furthermore, knockdown of HIPK2 did not affect the abundance of ΔNp63 mRNA (Fig. 2D) but significantly increased the abundance of the ΔNp63 protein (Fig. 2E). Moreover, shHipk2 blocked the IRF6-induced reduction in ΔNp63 protein abundance (Fig. 2E). These results placed HIPK2 downstream of IRF6 and upstream of ΔNp63 protein degradation during palate fusion.

Fig. 2 Knockdown of HIPK2 blocks IRF6-enhanced apoptosis and palate fusion.

(A) Immunohistochemical stains showing IRF6 and immunofluorescence showing HIPK2, p21, and active caspase 3 in E13.5 palatal shelves at 24 hours after infection with shLuc (control), shHipk2, AdIRF6, or shHipk2 + AdIRF6 viruses. Nuclei were counterstained with Hoechst 33258 (n = 5 independent experiments). Scale bar, 20 μm. (B) Quantification of the AOF of the palatal shelves at 24 hours after infection with the indicated viruses (n = 5 independent experiments). (C) Quantification of the staining intensities for HIPK2, p21, and active caspase 3 in palatal shelves at 24 hours after infection (n = 5 independent experiments). (D) Quantification of Hipk2, ΔNp63, and p21 transcripts in virus-infected palatal shelves by quantitative RT-PCR (n = 3 independent experiments). (E) Representative Western blot showing ΔNp63, p21, active caspase 3, HIPK2, and IRF6 in palatal shelves at 24 hours after infection with the indicated viruses. Quantification data are from three independent experiments. Statistical analyses were performed by one-way ANOVA. Error bars represent SD. *P < 0.05.

To further investigate the importance of HIPK2 in the IRF6-promoted apoptosis pathway, we examined whether HIPK2 could rescue the fusion delay caused by IRF6 depletion (shIrf6). Consistent with our previous study (31), most MEE cells had disappeared in the shLuc controls at 36 hours after lentivirus infection (Fig. 3A), but the disappearance of the MEE cells was significantly decreased by IRF6 knockdown (AOF from 71.3 ± 3.4% to 37.8 ± 3.9%) (Fig. 3, A and B). Overexpression of HIPK2 rescued the delay in palate fusion induced by IRF6 knockdown (from 37.8 ± 3.9% to 77.7 ± 5.4%) at 36 hours after infection. However, ectopic expression of the kinase-dead HIPK2 K221R mutant (fig. S3) (32) did not rescue the delay of palate fusion induced by shIrf6. At the protein level, IRF6 depletion significantly decreased the abundance of p21 as well as that of active caspase 3 in palatal shelves, and these effects were rescued by HIPK2 overexpression (AdHipk2) (Fig. 3, A and C). In addition, shIrf6-induced ΔNp63 protein accumulation was also blocked by HIPK2 overexpression but not by HIPK2-K221R (Fig. 3, A and C).

Fig. 3 HIPK2 rescues the shIrf6-induced delay of palate fusion.

(A) Immunofluorescence showing HIPK2, p21, and active caspase 3 in E13.5 palatal shelves at 36 hours after infection with shLuc (control), shIrf6, shIrf6 + AdHipk2, or shIrf6 + AdHipk2-K221R viruses. Nuclei were counterstained with Hoechst 33258. Scale bar, 20 μm (n = 4 independent experiments). (B) Quantification of the AOF of the palatal shelves at 36 hours after infection with the indicated viruses (n = 4 independent experiments). (C) Representative Western blot showing HIPK2, ΔNp63, p21, active caspase 3, and IRF6 in palatal shelves at 24 hours after infection with indicated viruses. Quantification data are from three independent experiments. Statistical analyses were performed by one-way ANOVA. Error bars represent SD. *P < 0.05.

HIPK2 promotes apoptosis through p21 in the MEE cells of palatal shelves

We next investigated the role of p21 in HIPK2-promoted apoptosis of MEE cells during palate fusion. We knocked down p21 and examined apoptosis activation and the degree of palate fusion. Ectopic expression of HIPK2 increased p21 and active caspase 3 and enhanced palate fusion (Fig. 4A). Knockdown of p21 in epithelial cells of palatal shelves not only significantly inhibited caspase 3 activation but also blocked caspase 3 activation stimulated by the overexpression of HIPK2 (Fig. 4, A and B). In addition, knockdown of p21 significantly inhibited AdHipk2-promoted palate fusion at 24 hours (from 54.3 ± 2.5% to 9.3 ± 2.7%) (Fig. 4, A and C) and at 48 hours (from 90.35 ± 1.7% to 51.35 ± 9.7%) after virus infection (fig. S4, A and B). It has been reported that, in cancer cells, HIPK2 phosphorylates ΔNp63 at Thr397 and that this promotes ΔNp63 degradation (29). Our results show that ectopic expression of HIPK2 decreased the abundance of ΔNp63 and increased the amount of p21 (Fig. 4D). Real-time PCR showed that AdHipk2 significantly increased the expression of p21 but not the expression of ΔNp63 (Fig. 4E). Thus, our findings suggest that HIPK2 decreases ΔNp63 protein stability rather than ΔNp63 expression during palatal shelves fusion.

Fig. 4 Knockdown of p21 inhibits HIPK2-promoted apoptosis and palate fusion.

(A) Immunofluorescence showing the abundance of HIPK2, p21, and active caspase 3 at 24 hours after infection with shLuc (control), shp21, AdHipk2, or shp21 + AdHipk2 viruses. Nuclei were counterstained with Hoechst 33258. Scale bar, 20 μm (n = 5 independent experiments). (B) Quantification of staining intensities for HIPK2, p21, and active caspase 3 in palatal shelves at 24 hours after infection with indicated viruses (n = 5 independent experiments). (C) Quantification of the AOF of the palatal shelves at 24 hours after infection with the indicated viruses (n = 5 independent experiments). (D) Representative Western blot showing the abundance of ΔNp63, active caspase 3, p21, and HIPK2 in palatal shelves at 24 hours after infection with indicated viruses. Quantification data are from three independent experiments. (E) Results of quantitative RT-PCR showing the relative mRNA abundance of Hipk2, ΔNp63, and p21 transcripts in virus-infected palatal shelves (n = 3 independent experiments). Statistical analyses were performed by one-way ANOVA. Error bars represent SD. *P < 0.05.

HIPK2 is involved in EMT in cancer cells (3335). To verify the possible role of HIPK2 in EMT regulation in palatal shelves, we ectopically expressed HIPK2 in palatal shelves and examined the expression of EMT regulators and markers. Overexpression of HIPK2 in palatal shelves did not affect the amounts of the EMT-promoting transcription factor Snai2 or the epithelial-specific calcium-dependent cell adhesion molecule E-cadherin, encoded by Cdh1 (fig. S5A). The expression of EMT transcription factors Snai2 and Twist and the epithelial-specific markers Cdh1, ZO-1, and Plakopilin was not affected by overexpression of HIPK2 (fig. S5B). These findings suggested that HIPK2 is not involved in EMT during palate fusion.

TAK1 phosphorylates HIPK2 in a contact-dependent manner during palate fusion

Previous studies indicate that phosphorylation of HIPK2 at Tyr354 is essential for HIPK2 catalytic activity (36, 37). We found that the abundance of HIPK2 phosphorylated at Tyr354 was decreased by IRF6 knockdown but increased by IRF6 overexpression (Fig. 5A). However, the ratio of phosphorylated HIPK2 to total HIPK2 was not changed, which suggests that IRF6 may affect the abundance of HIPK2 rather than by enhancing HIPK2 phosphorylation (Fig. 5A). Phosphorylated HIPK2 was only detected in MEE cells (Fig. 5B), supporting the observation that ectopic expression of HIPK2 decreased the ΔNp63 protein amount in MEE cells only and not in all epithelial cells of the palatal shelves (fig. S6). Therefore, we wondered whether, in addition to HIPK2, other kinases might be involved in apoptosis activation during IRF6-HIPK2–induced palatal epithelial fusion. TAK1 has been reported to phosphorylate HIPK2 at Tyr354 and to increase the stability and activity of HIPK2 (38). Moreover, TAK1 is involved in TGF-β3–induced palate fusion during development (19). Thus, we hypothesized that TAK1 may contribute to apoptosis by phosphorylating HIPK2 during palate fusion. Consistent with other studies, we detected TAK1 in the epithelial and mesenchymal cells of palatal shelves at E13.5, E14.5, and E15.5 (fig. S7A) (39). Furthermore, phosphorylated TAK1 was present in MEE cells and colocalized with phosphorylated HIPK2 (fig. S7B). Knockdown of TAK1 decreased the abundance of HIPK2 and phosphorylated HIPK2 and caused a decrease in caspase 3 activation in MEE cells and a blockage of palate fusion (Fig. 5, B and C). Similar protein changes were also observed by Western blot analyses (Fig. 5D). These observations suggested a role for TAK1 upstream of HIPK2 in palate fusion, perhaps in phosphorylating HIPK2 to induce ΔNp63 degradation and activation of the p21-dependent apoptosis pathway.

Fig. 5 TAK1 increases phosphorylation of HIPK2 during palate fusion.

(A) Representative Western blot showing the abundance of phosphorylated HIPK2 (p-HIPK2) and HIPK2 in palatal shelves at 24 hours after infection with shLuc, shIrf6, AdIE (control), or AdIRF6. The asterisk indicates a nonspecific band. Quantification data showing the abundance of p-HIPK2 relative to HIPK2 are shown (n = 3 independent experiments). (B) Immunofluorescence showing p-HIPK2 and active caspase 3 in palatal shelves at 48 hours after infection with the indicated viruses. Nuclei were counterstained with Hoechst 33258. Scale bar, 20 μm. (C) Quantification of the AOF of the palatal shelves at 48 hours after infection with the indicated viruses (n = 4 independent experiments). (D) Representative Western blot and quantification data showing p-HIPK2, HIPK2, and active caspase 3 in palatal shelves at 48 hours after infection with the indicated viruses (n = 3 independent experiments). (E) Representative Western blot showing phosphorylated TAK1 (p-TAK1), TAK1, p-HIPK2, HIPK2, and IRF6 in pairs of E13.5 palatal shelves that were placed in culture so that they touched each other (contact) or so that they did not touch each other (no contact) and infected with either AdIE control or AdIRF6 adenovirus for 24 hours. Quantification data are from three independent blots. Statistical analyses were performed by one-way ANOVA. Error bars represent SD. *P < 0.05.

IRF6 did not affect the amount of TAK1 or its phosphorylation status in palatal shelves (Fig. 5E); thus, other signaling molecules are likely to be needed for stimulating TAK1 phosphorylation. Because phosphorylated TAK1 was present in MEE cells and HIPK2 overexpression only brought about a decrease in the abundance of ΔNp63, an increase in p21, and an increase in activated caspase 3 in MEE cells, we hypothesized that MES and MEE formation is required for TAK1 activation during palate fusion. We cultured E13.5 palatal shelves in contact with one another or without contact. When we compared no-contact controls with pairs of palatal shelves in contact, there was a 3.8-fold increase in the amount of phosphorylated TAK1, but this did not affect the total abundance of TAK1 (Fig. 5E). This observation suggested that TAK1 was activated by contact between MEE cells of neighboring palatal shelves. In addition, contact between the paired palatal shelves also increased the amounts of total HIPK2 and phosphorylated HIPK2 (Fig. 5E). Overexpression of IRF6 statistically significantly increased total and phosphorylated HIPK2 amounts in both the contact and noncontact pairs of palatal shelves. However, IRF6 did not affect the abundance of TAK1 or phosphorylated TAK1. Together, our findings show that contact between two palatal shelves appears to play an important role in the activation of TAK1 phosphorylation, which then results in the phosphorylation of HIPK2, leading to an increase in HIPK2 stability. Taken as a whole, our findings suggest that IRF6 and TAK1 act coordinately to induce apoptosis during palate fusion by increasing the amount of phosphorylated HIPK2 through transcriptional and posttranslational mechanisms, respectively (Fig. 6).

Fig. 6 IRF6 promotes apoptosis through HIPK2 during palate fusion.

A schematic model outlining the IRF6-HIPK2-apoptosis pathway that eliminates seam cells at the border between fusing palatal shelves. Physical contact between the two palatal shelves stimulates the phosphorylation of TAK1 and HIPK2 during palate fusion.

DISCUSSION

We have demonstrated that IRF6 and TAK1 coordinately enhanced the amount and phosphorylation of HIPK2 during palate fusion. We have also shown that Hipk2 is a direct target of IRF6 and that HIPK2 is involved in the IRF6-ΔNp63-p21-apoptosis signaling pathway by inducing ΔNp63 protein degradation during palate fusion. We have demonstrated that physical contact between the palatal shelves induced TAK1 phosphorylation, which subsequently resulted in the phosphorylation of HIPK2, increased HIPK2 stability, and stimulated MEE cell apoptosis (Fig. 6). These findings are consistent with previous reports of phosphorylation of HIPK2 at Tyr354 being essential for its catalytic activity, subcellular localization, and protein stability (3638).

IRF6 induces ΔNp63 protein degradation through posttranslational modification, and the transcriptional regulatory activity of IRF6 is important for ΔNp63 protein degradation (27). In this study, we show that phosphorylated HIPK2 is important for IRF6-induced ΔNp63 degradation (Fig. 3C) and that IRF6 enhances Hipk2 expression through direct binding to enhancers in a Hipk2 intron in mouse palatal shelves (Fig. 1, F and G). There are four potential IRF6-binding elements in each of the enhancer candidate regions E1, E2, and E3. These binding elements contain the sequence “GAAAC,” which conforms to the IRF family recognition minimal core sequence 5′-AANNGAAA-3′ identified by Fuji et al. (40). Our results showed that IRF6 bound to these putative binding elements in Hipk2 intron 1 in vivo (Fig. 1G). Thus, we propose that the GAAAC sequences within the putative core binding sequence for IRF6 mediate IRF6-stimulated Hipk2 expression.

Although knockdown of HIPK2 significantly inhibited MEE cell apoptosis within the palatal shelves in organ culture, palate fusion was not completely blocked 48 hours after infection (fig. S2). Similar to the effect of shHipk2 on palate fusion, knockdown of p21 also did not completely block palate fusion at 48 hours (fig. S4A). Our findings suggest that p21-dependent apoptosis is essential, but not sufficient, for palate fusion. In addition to p21-dependent apoptosis, TGF-β3 induces p21-independent apoptosis in MEE cells during palate fusion through the activation of the extracellular matrix protein TGF-β–induced (TGFBI; also known as Bigh3) and the FasL-Fas-caspase cell death pathway (41, 42). Moreover, apoptosis and EMT are both important for IRF6-promoted MEE disappearance and palate fusion during palate development (31). Thus, the remaining MEE cells, after knockdown of HIPK2 or p21, may disappear through either the EMT or another p21-independent apoptosis process.

TAK1 has been suggested as a candidate gene for the cleft palate phenotype in the syndromic condition Pierre Robin sequence (PRS) because inactivation of TAK1 in the neural crest lineage in mice causes failed palate elevation associated with a malformed tongue cleft palate (39). This phenotype is similar to that seen in human PRS-associated clefting. Both noncanonical and canonical pathways of TGF-β signaling control the disappearance of MEE cells during palate fusion. Mutant mice lacking Tgfbr2 in cranial neural crest (CNC) cells show cleft palate with reduced CNC cell proliferation in the palatal mesenchyme, and a SMAD-independent TAK1-p38 MAPK pathway is responsible for the CNC cell proliferation defect and failure of palate fusion (43). TAK1 protein is present in both the epithelium and the mesenchyme of the developing palatal shelves. Deletion of Tak1 specifically in the palatal epithelium does not result in cleft palate (19, 39), but simultaneous deletion of both Tak1 and Smad4 in the palatal epithelium results in defects in the anterior and posterior secondary palate, suggesting that Tak1 and Smad4 interact genetically in palatal epithelial fusion (19, 39). The E3 ubiquitin ligase and regulator of TGF-β signaling TRIM33 (also known as Tif1γ or ectodermin) also cooperate with Smad4 during palatogenesis, and simultaneous deletion of Trim33 and Smad4 in the palatal epithelium also causes palate fusion defects (19). SMAD4, TAK1, and TRIM33 cooperate to regulate palatal epithelial fusion in vivo, and they are functionally redundant in TGF-β signaling during palatogenesis (19). Here, we showed that IRF6 did not increase the abundance of either total or phosphorylated TAK1 during palate fusion (Fig. 5); however, phosphorylated TAK1 was present in high amounts in MEE cells, and contact between two palatal shelves enhanced TAK1 phosphorylation (Fig. 5E and fig. S7, A and B). Our results imply that physical contact is essential for the phosphorylation of TAK1 during palate fusion. These findings are also supported by the observation that phosphorylated HIPK2 is present in a high abundance in MEE cells.

It has been reported that Notch is present in palate epithelial cells and may play an important role in palate development (44). Notch and IRF6 functionality involves convergent molecular pathways, and both seem to play key roles in ensuring appropriate palate adhesion (45). Moreover, cross-talk and integration between the Notch and TGF-β pathways are important for the functioning of TGF-β pathways (4648). These studies and our findings support the hypothesis that cross-talk between the Notch and TGF-β pathways may play a key role in TAK1 activation when there is physical contact between the two palatal shelves and MES formation. It is worth mentioning that reverse signaling through the receptor Eph and its ligand, ephrin-B1, activates TAK1 in 293T cells (49). In addition, ephrin-B2 is expressed in MEE cells during palate fusion, and ephrin reverse signaling has been shown to promote palate fusion in mouse embryos (50, 51). The results of these studies and our findings, when taken together, suggest another possibility, namely, that TAK1 may be activated by ephrin reverse signaling in MEE cells during palate fusion (49, 50). In addition to the Notch and ephrin-Eph signaling pathways, TAK1 can also be activated by the Toll-like receptor (TLR), tumor necrosis factor receptor (TNFR), or focal adhesion kinase-Src (FAK-Src) signal cascade in other cell types (49, 52, 53). The possibility that these signaling pathways are also involved in TAK1 activation during palate fusion requires further investigation.

Because of the functional redundancy between HIPK1 and HIPK2, Hipk1 (54) or Hipk2 (55) single-knockout mice are grossly normal. Hipk1 Hipk2 double-mutant mice show severe growth retardation that leads to early embryonic lethality between E9.5 and E12.5 (56). Thus, there are no previous reports showing cleft palate or other craniofacial phenotypes in Hipk2 knockout mice. The human HIPK2 gene is located on chromosome 7q34 (57, 58). Several reports have found that patients with distal 7q deletions (between 7q32-7qter) have some common phenotypes, including developmental delay, speech impairment, hearing loss, and craniofacial abnormalities, with some patients presenting with a cleft lip and/or palate (5961). These findings suggest that HIPK2 may be a candidate gene for the mutations that bring about human cleft lip and/or palate as well as other lower lip anomalies. Furthermore, mutations affecting the IRF6 gene have been associated with sensorineural hearing loss (62), and HIPK2 and IRF6 are both present in neurons and seem to be involved in regulating neuron survival and development (55, 63). These studies suggest that the IRF6-HIPK2 pathway not only may play an important role in palate development but also may be involved in neural development.

In cancer cells, studies of HIPK2 function have focused on its role in DNA damage–induced apoptosis and as a tumor suppressor (29, 34, 35, 6466). In this study, we demonstrated a role for HIPK2 in apoptosis following contact of the palatal primordia during palate fusion. Using palatal shelves organ culture, we demonstrated that HIPK2 is involved in an IRF6-ΔNp63-p21-apoptosis signaling pathway during palate fusion by promoting ΔNp63 protein degradation. We also provide evidence demonstrating that Hipk2 is a direct target of IRF6 within palatal shelves. Our findings revealed the importance of TAK1 in this process and how contact between palatal shelves affects HIPK2 phosphorylation and apoptosis during palate fusion. These findings provide new insights into the molecular and cellular mechanisms that lead to nonsyndromic craniofacial defects and VWS in humans.

MATERIALS AND METHODS

Immunostaining

Immunohistochemistry was performed by following the manufacturer’s instructions (DAKO Company). Briefly, after deparaffinization and rehydration, tissue sections were placed in antigen retrieval buffer (pH 6.0 or 9.0; DAKO) and heated for 20 min by microwave. After the retrieval buffer cooled down to room temperature, any endogenous peroxidase activity of samples was quenched using 3% hydrogen peroxide (Merck) for 5 min. Then, the samples were incubated with blocking buffer containing 5% bovine serum albumin (BSA; Sigma-Aldrich) and 0.1% cold water fish gelatin (Aurion) for 1 hour at room temperature. Incubation with primary antibodies against IRF6 (GeneTex), ΔNp63 (Calbiochem), HIPK2 (Cell Signaling Technology), phospho-HIPK2-Tyr354 (Aviva Systems Biology), active caspase 3 (cleaved caspase3 D175; Cell Signaling Technology), p21 (Abcam), TAK1 (Abcam), or phopho-TAK1-Thr184/187 (Cell Signaling Technology) overnight at 4°C followed the procedure. For immunohistochemistry staining, sections were incubated with biotinylated secondary antibodies (DAKO) for 30 min at room temperature. Subsequently, the slides were incubated with streptavidin–horseradish peroxidase (HRP; DAKO) for 10 min at room temperature, which was followed by staining with 3,3′-diaminobenzidine (DAB) (DAKO). Next, the sections were counterstained with hematoxylin (Sigma-Aldrich) for 8 s. For immunofluorescence staining, anti-rabbit immunoglobulin G (IgG) or anti-mouse IgG secondary antibodies conjugated to Alexa Fluor 568 or Alexa Fluor 488 (Invitrogen), respectively, were used. Hoechst 33258 was used for nuclei staining. Images of samples were captured using an Olympus BX51 image system. The immunostaining intensity of HIPK2, phospho-HIPK2-Tyr354, p21, ΔNp63, and active caspase 3 proteins was quantified by pixel analysis using ImageJ software (31, 67). The antibodies and dilutions used for immunostaining are listed in table S3.

Quantitative real-time PCR

For quantitative real-time PCR (RT-PCR), total RNAs were extracted from the palatal shelves using TRIzol Reagent by following the manufacturer’s instructions (Invitrogen). In total, 5 μg of total RNA was used to synthesize complementary DNA (cDNA) using a RevertAid First strand cDNA Synthesis kit (Thermo Scientific) according to the manufacturer’s instructions. Real-time PCRs using SYBR Green PCR Master Mix (Applied Biosystems) were then performed using the ABI StepOnePlus process. The primers used for quantitative RT-PCR in this study are listed in table S1.

Lentivirus production

Lentivirus was prepared following the protocol provided by the National RNAi Core Facility (68). To generate lentivirus, pLKO.1-shRNA, pCMV-∆R8.91, and pMD.G were cotransfected to human embryonic kidney (HEK)–293T cells. The virus suspension was collected at 24 and 36 hours after transfection and pooled. Next, the virus suspension was filtered through a 0.22-μm filter (Merck Millipore). For lentivirus titration, HEK293T cells were infected for 24 hours with lentivirus supernatant using serial dilution, and the infected cells were subsequently selected using puromycin (2 μg/ml; Sigma-Aldrich) for 3 days. pCMV-∆R8.91 plasmid, pMD.G plasmid, and the shRNA expression plasmids (pLKO.1 vector) targeting the mouse Irf6, Hipk2, p21 (Cdkn1a), or Tak1 (Map3k7) genes were obtained from the National RNAi Core Facility (Taiwan). Mouse embryonic palatal shelves were infected with lentivirus carrying different gene shRNAs to test the knockdown efficiencies or coinfected with different adenoviruses to examine the rescue effect and as off-target tests.

Adenovirus production

Adenovirus was produced by following the AdEasy protocol (69). Human IRF6, human p21 (CDKN1A) cDNA, and mouse Hipk2, TAK1 (Map3k7) cDNA clones from the Mammalian Gene Collection (MGC) were cloned into the pAdTrack-CMV vector (Agilent Technologies) separately. The pAdTrack-CMV-IRF6-R84C and pAdTrack-CMV-Hipk2-K221R mutants were obtained by site-directed mutagenesis PCR. AdIRF6, Adp21, AdHipk2, and AdTAk1 adenovirus expression plasmids were generated by recombination between the pAdEasy-1 vector and pAdTrack-CMV-IRF6, pAdTrack-CMV-p21, pAdTrack-CMV-Hipk2, or pAdTrack-CMV-TAK1 in BJ5183 competent cells (Agilent Technologies). AdIRF6, AdHipk2, Adp21, AdTAK1, or AdHipk2-K221R plasmid (40 μg) was linearized by Pac I restriction enzyme and subsequently transfected separately into AD293 cells using Lipofectamine 2000 reagent (Invitrogen). The adenoviruses were harvested at 2 to 3 weeks after transfection.

Organ culture and virus infection

Organ culture and the virus infection model using the mouse palatal shelves were established and used a submerged system as previously described (13, 31, 70). Briefly, palatal shelves were dissected from E13.5 C57BL/6 mouse embryos, and they were placed together on a polycarbonate membrane filter with a 0.8-μm pore size (Merck Millipore) in a 35-mm culture dish. Subsequently, they were cultured in 0.5 ml of serum-free Dulbecco’s modified Eagle’s medium/F12 (Gibco) containing penicillin-streptomycin (100 U/ml; Gibco), 2 mM l-glutamine (Gibco), and 0.1 mM nonessential amino acid (Biological Industries) for 3 hours in a 37°C incubator with 5% CO2; this was to allow the palatal shelves to attach to the filters. Next, the filters with the attached palatal shelves were transferred to a 48-well culture plate and incubated with medium containing 5.4 × 107 relative infection units/ml (RIU/ml) of an appropriate adenovirus or 3.3 × 106 RIU/ml of an appropriate lentivirus for 24 hours. Subsequently, the palatal shelves still on the filter were fixed with 4% paraformaldehyde (PFA) (Sigma-Aldrich) in phosphate-buffered saline, or alternatively, the medium was changed and the membrane and palatal shelves were cultured for another 12 or 24 hours (a total of either 36 or 48 hours after infection). Fixed palatal shelves were transferred into Spin Tissue Processor 120 (MICROM) for tissue processing. Next, the samples were embedded in paraffin and sectioned using a HM315 microtome (MICROM) set at a thickness of 4 μm. All animal experiments were performed with the approval of the Institutional Animal Care and Use Committee of Chang-Gung Memorial Hospital and National Yang-Ming University.

Histological examination and scoring of palatal fusion

Coronal serial sections (4 μm) were collected and numbered in sequence from the anterior to the posterior of organ culture palatal shelves samples and embryos. The AOF (%) for each section was calculated as the length of mesenchymal confluence divided by total length of adherence at the MES region × 100% (31). The AOF was individually analyzed for anterior, middle, and posterior sections of each sample, and on average, three sections of each sample were analyzed.

Western blotting

Total lysates of the palate shelves were extracted using 2× sample buffer [100 mM tris (pH 6.8), 0.1 M MgCl2.6H2O, 4% SDS, 5% glycerol, 2.5% β-mercaptoethanol, 2.5% bromophenol blue] on ice for 15 min, which was followed by vortexing for 5 min, and then the samples were heated at 95°C for 5 min. Subsequently, the total lysates were centrifuged at 4°C, 14,000 rpm for 1 hour, and the supernatant was collected to a new microtube. Total proteins from the palatal shelves were separated by 8, 10, or 12% SDS–polyacrylamide gel electrophoresis (PAGE). Proteins separated by SDS-PAGE were transferred to 0.22 μm of polyvinylidene difluoride (PVDF) membrane. Western blotting was performed by incubating the membranes individually with primary antibodies against IRF6 (GeneTex), HIPK2 (Cell Signaling Technology), phospho-HIPK2-Tyr354 (Aviva Systems Biology), active caspase 3 (cleaved caspase3 D175; Cell Signaling Technology), p21 (Abcam), p53 (Cell Signaling Technology), phospho-p53 (Ser46) (Cell Signaling Technology), TAK1 (Abcam), phopho-TAK1-Thr184/187 (Cell Signaling Technology), or GAPDH (glyceraldehyde phosphate dehydrogenase; Merck Millipore) overnight at 4°C. Subsequently, the membranes were washed and then incubated with goat anti-mouse IgG-HRP or goat anti-rabbit IgG-HRP secondary antibodies (GE Healthcare) at room temperature for 1 hour. The signals were detected using a Western Lightning ECL Pro kit (PerkinElmer). Relative abundance as changes in the folds of proteins present was analyzed by ImageJ. The antibodies and dilutions used for Western blotting are listed in table S3.

Promoter search for the IRF6 responsive elements and the luciferase assay

By analysis of the mouse Hipk2 sequence data from the UCSC genome browser and the enhancer markers available in ENCSR645ETR and ENCSR258AED of the ENCODE (Encyclopedia of DNA Elements) project, we selected three enhancer candidates associated with the mouse Hipk2 gene. Mouse Hipk2 enhancer candidates 1 to 3 in intron 1 (Hipk2-E1: GRCm38/mm10, chr6: 38,868,986-38,873,293; Hipk2-E2: GRCm38/mm10, chr6: 38,846,439 38,850,311; Hipk2-E3: GRCm38/mm10, chr6: 38,831,491-38,833,753) were cloned into pGL4 vector (pGL4.24 vector was purchased from Promega). To verify whether these elements are IRF6 dependent, these reporter plasmids were cotransfected with pAdTrack-CMV-IRF6 or pAdTrack-CMV-IRF6-R84C plasmid into 293T cells using jetPEI reagent (Polyplus transfection) according to the manufacturer’s instructions. Luciferase activities were measured by the Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s instructions.

Chromatin immunoprecipitation

Nine pairs of palatal shelves harvested from E14.5 C57BL/6 wild-type mouse were used for ChIP. After 1.4% PFA fixation for 15 min and stopped by 227 mM glycine, tissue was homogenized and washed three times with A1 buffer [150 mM KCl, 15 mM NaCl, 15 mM Hepes (pH 7.5), 0.5% Triton X-100, 0.5 mM EDTA] and one time with A2 buffer [140 mM NaCl, 15 mM Hepes (pH 7.5), 1 mM EDTA, 0.5 mM EGTA, 0.5% NP-40] using dounce homogenizer on ice. Lysate was processed by chromatin fragmentation with Ultrasonic Processor XL (Misonix) for 8 to 10 cycles of 30-s pulse with 30-s interval at 15% power in ice. After preclearing with BSA (1 mg/ml), 0.1% normal goat serum, and rabbit anti-IgG antibody (Abcam, ab171870), the samples were incubated overnight with anti-IgG antibody, anti-IRF6 antibody (GeneTex, GTX104862), or anti–histone H3 antibody (Abcam, ab1791) previously conjugated with Dynabeads protein A (Invitrogen), and the immunocomplexes were washed four times with A3 buffer (A2 buffer, 0.05% SDS, 0.01% Tween 20) and two times for 5 min with TE buffer [10 mM tris-HCl (pH 8), 1 mM EDTA]. The immunocomplexes were extracted with extraction buffer [50 mM tris-HCl (pH 8), 10 mM EDTA, 1% SDS] and reextracted with TE buffer containing 0.67% SDS. Last, combined elute complexes were processed by reverse cross-linking, proteinase K treatment, and precipitation of DNA. DNA was analyzed by PCR using the specific primers for each potential IRF6 binding sites, respectively. The primers used for ChIP assay in this study are listed in table S2.

Statistical analysis

One-way analysis of variance (ANOVA) was used to analyze all statistical data of Western blotting, quantitative RT-PCR, AOF, and immunostaining results. The results were presented as mean ± SD and were considered to be significant at P < 0.05.

SUPPLEMENTARY MATERIALS

stke.sciencemag.org/cgi/content/full/12/593/eaav7666/DC1

Fig. S1. Knockdown efficiency of shRNAs and off-target tests in palatal shelves organ culture.

Fig. S2. Knockdown of HIPK2 decreases palate fusion at 48 hours after infection.

Fig. S3. Loss of kinase activity of the HIPK2-K221R protein.

Fig. S4. Knockdown of p21 inhibits HIPK2-promoted apoptosis and palate fusion at 48 hours after infection.

Fig. S5. Ectopic expression of HIPK2 does not affect the expression of EMT markers during palate fusion.

Fig. S6. Ectopic expression of HIPK2 represses p63 accumulation only in MEE cells during palate fusion.

Fig. S7. TAK1 and phosphorylated TAK1 in palatal shelves during development.

Table S1. Primers for RT-PCR.

Table S2. Primers for ChIP-PCR.

Table S3. Antibodies and dilutions.

REFERENCES AND NOTES

Acknowledgments: We thank the National Yang-Ming University Genome Research Center for the human IRF6 and mouse Hipk2 cDNA clone (pCMV-SPORT6-IRF6 and pCMV-SPORT6-Hipk2). We are grateful to L.-H. Hwang (Institute of Microbiology and Immunology, National Yang-Ming University) for the AD293 cells, the pAdEasy-1 vector, and the pAdTrack-CMV vector. We thank C.-M. Chen and M.-J. Fann (Department of Life Sciences and Institute of Genome Sciences, National Yang-Ming University) for their technical support regarding mouse handling. Funding: This work was supported by grants from the Chang-Gung Memorial Hospital (CMRPG381271-3, CRRPG5C0161-3, and CMRPG3F0651-653) and a grant from the Ministry of Education, Aiming for the Top University Plan (105AC-P667 and 106AC-P667). RNA interference (RNAi) reagents were obtained from the National Core Facility for Manipulation of Gene Function by RNAi, miRNA, miRNA sponges, and the CRISPR/Genomic Research Center, Academia Sinica, supported by the National Core Facility Program for Biotechnology Grants of MOST (MOST 104-2319-B-001-001). Author contributions: C.-Y.K. conceived, designed, and performed the experiments and wrote the manuscript. H.-H.M. performed the experiments. L.-J.L. and F.-H.W. conceived and designed the experiments and helped write the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials. The plasmids require a material transfer agreement from National Yang-Ming University and National Core Facility Program.
View Abstract

Stay Connected to Science Signaling

Navigate This Article