Ripply3, a Tbx1 repressor, is required for development of the pharyngeal apparatus and its derivatives in mice

The pharyngeal apparatus is a transient structure that is formed ventrolateral to the hindbrain in vertebrate embryos. It consists of bilaterally segmented arches, between which ectodermal grooves and endodermal pouches are formed. The pharyngeal arches comprise mesodermal cells, neural crest-derived mesenchyme, an outer ectodermal cover, and an inner endodermal lining. Within the arches, pharyngeal arch arteries (PAAs) also develop. Components of the pharyngeal apparatus give rise to distinct tissues at later stages of development. For instance, the pharyngeal arteries and neural crest cells in the caudal pharyngeal arches contribute to cardiovascular development, whereas the endodermal cells located in the caudal pouches give rise to several organs, including the thymus and parathyroid gland. Thus, pharyngeal development is a key process in the generation of these organs.

Chromosome 22q11 deletion syndrome (22q11DS), which includes the DiGeorge syndrome (DGS), conotruncal anomaly face syndrome (CAFS) and velocardiofacial syndrome (VCFS), is characterized by the abnormal development of the pharyngeal apparatus in the form of thymic hypoplasia or aplasia, hypocalcemia arising from parathyroid hypoplasia, and defective cardiac outflow (Scambler, 2000). A number of mouse genetic studies and mutation analyses in human patients have indicated that Tbx1, which encodes a member of the T-box family of transcription factors, is most likely responsible for the phenotype of 22q11DS (Jerome and Papaioannou, 2001; Lindsay et al., 2001; Merscher et al., 2001; Yagi et al., 2003). During murine pharyngeal development, Tbx1 is first expressed in the mesoderm at E7.5. Between E8.5 and E11.5, Tbx1 expression appears in the pharyngeal endoderm, ectoderm and core mesoderm, but not in the neural crest cells (Chapman et al., 1996; Vitelli et al., 2002; Yamagishi et al., 2003). Cell type-specific inactivation and analysis of downstream targets of Tbx1 in mice indicate that Tbx1 plays multiple roles in endoderm, mesoderm and ectoderm cells during pharyngeal development (Arnold et al., 2006; Zhang et al., 2006; Calmont et al., 2009). However, the molecular mechanisms underlying the cell type-specific roles of Tbx1 have not been fully elucidated.

T-box transcription factors are characterized by their DNA-binding domain, known as the T-domain, and function as either activators or repressors depending on their association with transcriptional co-activator or co-repressor complexes. Tbx1 is known to act as a transcriptional activator as it activates the transcription of a reporter containing the T-box binding elements in vitro and in vivo (Xu et al., 2004; Hu et al., 2004; Stoller and Epstein, 2005; Paylor et al., 2006; Zweier et al., 2007). However, in addition to the `on DNA' manner, Tbx1 also functions in an `off DNA' manner by interfering with Smad1-Smad4 binding in the regulation of BMP signaling (Fulcoli et al., 2009). Despite an accumulation of studies on transgenic and mutant mouse lines, the molecular basis underlying Tbx1-mediated gene regulation in pharyngeal development remains to be elucidated.

Ripply proteins have recently been shown to modulate the transcriptional properties of T-box proteins (Kawamura et al., 2008; Kondow et al., 2007). Ripply associates with the transcriptional co-repressor Groucho/TLE and the T-box proteins through two distinct amino acid sequences: the WRPW motif, which is a highly conserved four amino acid stretch in the N-terminal half; and the Ripply homology (RH) domain, a conserved ∼50 amino acid stretch that interacts with the T-domain (Kawamura et al., 2005; Kondow et al., 2006). Therefore, Ripply is able to recruit the Groucho/TLE co-repressor to T-box proteins and control their intrinsic transcriptional properties. Ripply1 and Ripply2 have been shown to play roles in somite segmentation during development (Kawamura et al., 2005; Kondow et al., 2007; Morimoto et al., 2007; Chan et al., 2007; Moreno et al., 2008; Takahashi et al., 2010).

To reveal the developmental role of another member of the Ripply family, Ripply3 [also known as Down syndrome critical region 6 (Dscr6)] (Kawamura et al., 2005; Hitachi et al., 2009; Shibuya et al., 2000), we examined its expression and function in the early mouse embryo. On finding that both Ripply3 and Tbx1 are expressed in the pharyngeal endoderm, we examined the role of Ripply3 by generating Ripply3-deficient mice and investigated its relationship with Tbx1.

Whole-mount in situ hybridization was performed as described previously (Yoshikawa et al., 1997). In situ hybridization was also carried out on 7-μm paraffin sections collected from embryos. Paraffin sections (7 μm) were also incubated with antibodies specific for Tbx1 (ab18530, Abcam), Pax9 (clone 7C2, Sigma), AP-2α (3B5, DSHB), phospho-histone H3 (Ser10, Millipore Upstate), Nkx2.1 (TTF1; clone 8G7G3/1, Dako) and Pecam1 (CD31; clone MEC13.3, BD Pharmingen). The secondary antibodies and signal amplification used were anti-rat IgG-Alexa Fluor 488, anti-rabbit IgG-Alexa Fluor 546 (Molecular Probes) and the EnVision System-HRP (Dako). For the TUNEL assay, the In Vitro Cell Death Detection Kit-TM-Red was used (Roche).

For the generation of Ripply3-deficient mice, a mouse Ripply3 genomic clone was obtained from 129SV genomic DNA by PCR. An IRES-lacZ-PGK-neo cassette, in which the neomycin phosphotransferase gene is linked to the lacZ gene placed between the independent ribosomal entry sequence (IRES) and an SV40 polyadenylation signal (Ohbayashi et al., 2002), replaced a sequence covering the Ripply3 coding sequence of the first, second and third exons. The diphtheria toxin A (DTA) expression cassette was inserted at the 3′ end of the genomic DNA.

CJ7 ES cells were electroporated with linearized targeting vector and selected in media containing G418. Targeted clones were confirmed by PCR and Southern blot analysis. Heterozygous ES cells were injected into blastocysts of C57BL/6J mice to generate germline chimeras. Chimeric males were mated with C57BL/6J females, and heterozygous mice were subsequently backcrossed onto either C57BL/6J or 129SV strains. No obvious phenotypic differences were observed between the mice generated on the C57BL/6J or 129SV backgrounds. Tbx1 knockout mice were kindly provided by Dr Bernice Morrow [Albert Einstein College, NY, USA (Merscher et al., 2001)] and maintained as heterozygous lines.

For genotyping, the wild-type or mutant alleles were amplified by PCR using the following primers (5′ to 3′): Rpy3-geF1, AACCTGAGATCGACTACTGC; Rpy3-geR1, ATCCCTTAAGGTCTGTCTGC; lacZ-F1, TGTTTTGACCGCTGGGATCTGC; and lacZ-R1, CCAGACCAACTGGTAATGGTAGC. Amplification was performed for 32 cycles at 94°C for 30 seconds, 60°C for 30 seconds, and 72°C for 30 seconds. The presence of a 350 bp fragment, a 550 bp fragment or both fragments represented animals of wild-type, Ripply3 homozygous and Ripply3 heterozygous genotypes, respectively. PCR genotyping of Tbx1 alleles was as previously described (Merscher et al., 2001).

Total RNA was extracted from tissues or whole embryos using RNeasy (Qiagen). cDNA was synthesized using SuperScript III reverse transcriptase (Invitrogen) with oligo(dT) primers. Primer sequences (5′ to 3′) were as follows: beta-actin (Actb), TCGTACCACAGGCATTGTGATGG and GCAATGCCTGGGTACATGGTGG; Ripply3, F2 GTCGGTCTGAGAGATTCGCG and R4 CTTTATTCTGCCCTTTCCTCC.

For luciferase assays, COS-7 cells were cultured at 37°C in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Then, 5000 cells/well were seeded into 24-well plates (Iwaki). Following overnight incubation, the cells were transiently transfected with plasmid DNA, which included pcDNA3.1-mTbx1, pcDNA3.1-mRipply3, and a T-box reporter vector that has two T-box motifs (cgcgtAAATTCACACCTgggcccAAATTCACACCTc) inserted into the MluI and XhoI sites of the pGL3-Promoter vector (Promega). The total DNA concentration was made consistent in each transfection by supplementation with pcDNA3.1 empty vector. A 3.7 kb stretch of the 5′ upstream region of the mouse Pax9 gene was amplified from C57BL/6 genomic DNA using MluI or XhoI site-attached primers (5′-cgacgcgtcgGAAAACTTCCCCCAGACAGCTGC-3′, 5′-ccgctcgagcggTGCTCTGAGCAGTACACCAACC-3′). The PCR fragment was cloned into the pGL3-Basic vector (Promega) and the sequence checked. For measurement of luciferase activity, cells were harvested 24 hours after transfection and suspended in Dual-Luciferase Assay Solution (from the Dual-Glo Luciferase Assay System, Promega). For normalization of transfection efficiency, the cells were co-transfected with pRL-CMV carrying Renilla luciferase under the control of the CMV promoter. All experiments were undertaken in triplicate and statistical significance was evaluated.

COS-7 cells were transfected with pCS2-mRipply3-Flag, pcDNA3.1-mTbx1-myc or pcDNA3.1-mTbx1(ΔTbox)-myc using the FuGENE 6 transfection reagent (Roche). Cell lysates were incubated with anti-Flag M2-conjugated agarose gel (Sigma) or anti-Myc (4A6) agarose conjugate (Millipore) at 4°C overnight, and immunoblotting was performed with anti-myc (clone 4A6, Upstate) and anti-Flag D8 (Santa Cruz) antibodies and HRP-labeled anti-rabbit IgG (Jackson ImmunoResearch).

Antibodies used for flow cytometry included FITC-conjugated anti-mouse CD4 (L3T4; eBioscience) and PE-conjugated anti-mouse CD8a (Ly-2; eBioscience). For flow cytometry, thymocytes were isolated from the newborn thymus and the total cell number was calculated. A total of 1×10⁶ cells were then incubated with a combination of the CD4 and CD8 antibodies for 30 minutes on ice, and the cells were washed three times in PBS containing bovine serum albumin before the final addition of PE for the elimination of dead cells. Flow analysis was undertaken using an Epics ALTRA flow cytometer (Beckman Coulter).

To examine the spatiotemporal expression of Ripply3 during embryogenesis, we first performed in situ hybridization and found it to be dynamically expressed in the pharyngeal endoderm and ectoderm (Fig. 1A-E). In the mouse embryo, pharyngeal arches develop in a rostral-to-caudal manner. At E8.5, when the first arch formation is observed, Ripply3 signals were observed in the pharyngeal ectoderm (Fig. 1A,A′). After E8.5, in accordance with subsequent formation of more caudal pharyngeal arches, Ripply3 expression was evident in the ectoderm and endoderm cells of most of the pharyngeal pouches, although its expression in the anterior pouches had started to gradually decrease. By E10.5, strong Ripply3 expression only remained in the fourth pouch. Ripply3 and Tbx1 were strongly co-expressed in the endoderm around a newly forming pouch, although Ripply3 expression was more sustained in the anterior pouch, and Tbx1 expression was also identified in the mesodermal cores of the pharyngeal arches (Fig. 2A). These results suggest that Ripply3 might regulate Tbx1 function specifically in the developing caudal pharyngeal endoderm, but not in the mesodermal cells, of the arches.

It has been proposed that Ripply proteins repress the transcriptional activities of some T-box proteins by recruiting the Groucho/TLE-HDAC complex. To examine whether mouse Ripply3 was able to repress Tbx1 transcriptional activity, we performed luciferase reporter assays in COS-7 cells, in which the reporter gene was under the control of tandemly repeated T-box protein binding sites (Fig. 2B). We found that Tbx1 significantly activated the luciferase activity in a manner that was dependent on the T-domain, whereas Ripply3 inhibited it (Fig. 2B). Immunoprecipitation assays showed that Ripply3 physically interacted with Tbx1 in a T-domain-dependent manner (Fig. 2C). As zebrafish Ripply1 is known to interact with the transcriptional co-repressor Groucho/TLE via the WRPW motif of the former (Kawamura et al., 2005), the effect of removing the WRPW motif was also examined. Compared with the wild-type Ripply3 protein, Ripply3 lacking the WRPW tetrapeptide failed to efficiently repress Tbx1-mediated enhancement of luciferase activity (Fig. 2B). Thus, Ripply3 is able to repress the transcriptional activity of Tbx1, probably by recruiting the Groucho/TLE-HDAC complex via its WRPW motif.

Next, we asked whether Ripply3 could also repress the in vitro expression of a Tbx1 target gene that is known to be expressed in the pharyngeal endoderm. Microarray analyses have suggested a number of putative downstream genes of Tbx1 in the pharyngeal endoderm (Ivins et al., 2005; Liao et al., 2008). We found that the expression of a putative endodermal target, Pax9, substantially overlapped that of Tbx1 (Fig. 2D). Furthermore, expression from the Pax9 promoter was actually activated by Tbx1 in luciferase reporter assays (Fig. 2E) and Ripply3 repressed this Tbx1-mediated expression of Pax9. However, we found that even a 0.5 kb promoter fragment of Pax9, which does not contain a typical T-box binding site (see Fig. S1 in the supplementary material), was also regulated by Tbx1 and Ripply3. Therefore, in the case of Pax9 expression, Tbx1 and Ripply3 might regulate this expression in an atypical manner; for instance, through some DNA element that differs from the consensus sequence of the T-box binding site or through an `off DNA' interaction with some other component of the transcriptional machinery.

To reveal the roles of Ripply3 during embryogenesis, we generated a mutant allele of mouse Ripply3 by inserting the IRES-lacZ-polyA cassette and PGK-neo gene into its first exon (Fig. 3A-C). As most of the amino acid sequence of Ripply3 is deleted in the protein encoded by this allele, the resulting targeted allele was expected to be null. Mice heterozygous for this mutation were found to be viable, fertile and morphologically normal (Fig. 3D). Mice homozygous for this mutation were born, but died with cyanosis within 24 hours of birth (Fig. 3D; see Table S1 in the supplementary material).

This lethality with cyanosis prompted us to investigate cardiovascular development in the Ripply3-deficient embryos. The PAAs that pass through the pharyngeal arches to connect the aortic sac with the dorsal aorta are remodeled during development to form the mature aortic arch and great vessels (Srivastava and Olson, 2000). In wild-type embryos, the PAAs initially form as many pairs of symmetrical vessels. The first and second pairs of PAAs, as well as the right sixth PAA, regress during development. By contrast, the left fourth PAA contributes to the formation of the aortic arch, the third pair and the right fourth PAAs develop into the common carotid arteries and the right subclavian artery, respectively, and the left sixth PAA gives rise to a part of the pulmonary arteries and ductus arteriosus. Intracardiac ink injection and immunostaining with anti-Pecam1 (Baldwin et al., 1994) indicated that, in the Ripply3^–/– embryos, the third and fourth PAAs could not be identified (Fig. 4A-D; see Fig. S2 in the supplementary material). By contrast, the second PAAs, which normally disappear during mammalian development, persisted and remained connected to the dorsal aorta.

These abnormalities appear to have resulted in misshapen great vessels (Fig. 4E-J). The persisting second PAAs contributed abnormally to the development of the common carotid arteries (Fig. 4G,J, asterisks). Dorsolateral to these common carotid arteries, two additional ascending arteries were ectopically formed. These arteries are likely to have been generated from the dorsal aorta, a part of which regresses in normal development (Fig. 4H,J, #). Owing to the lack of the fourth PAAs, interruption of the aortic arch occurred. Subclavian arteries in Ripply3^–/– mutants were present, but they abnormally branched from the dorsal aorta in the retroesophageal region (Fig. 4H,J). The ductus arteriosus in Ripply3^–/– mutants persisted and anastomosed with the descending dorsal aorta (Fig. 4G,H,J). All of these structural defects appeared to have been caused by the inappropriate persistence and regression of PAAs in the mutants. In addition to aortic arch malformation, the ventricular septum was incompletely formed in the heart of Ripply3^–/– embryos. Ripply3^–/– embryos also exhibited a hypotrophic aorta in the conotruncus region, in spite of the formation of outflow septum (Fig. 4K,L; see Fig. S3 in the supplementary material). Based on these findings, we concluded that Ripply3 deficiency leads to severe defects in the development of the great arteries and heart.

In addition to the cardiovascular defects in Ripply3^–/– embryos, we also found defects in the development of the thymus, the parathyroid glands and the ultimobranchial bodies, all of which are derived from the pharyngeal pouch endoderm. The endodermal cells around the third pharyngeal pouches give rise to the thymus and parathyroid glands, and those of the fourth pouch give rise to the ultimobranchial bodies (Manley and Capecchi, 1998). These organs bud from the pharyngeal pouches and migrate ventrocaudally to their final position in the mature mouse. The thymus moves to a position just anterior to the heart (Fig. 5A), whereas the parathyroid glands and the ultimobranchial bodies move to their distinct positions either adjacent to, or within, the thyroid gland.

In the Ripply3^–/– embryos, these organs were not separated from the endodermal epithelia in the pharyngeal region. For instance, the thymus, defined by Foxn1 expression (Gordon et al., 2001), was formed but not separated from the pharyngeal endoderm at E13.5 (Fig. 5A,B,H,I). As a result, the thymus did not migrate caudally but developed ectopically in the pharynx region at E18.5 (see Fig. S4 in the supplementary material). In many cases, the ectopic thymus appeared to have evaginated into the oropharynx (Fig. 5B,D). Although the ectopic thymus was smaller than normal (Fig. 5E,F), the differentiation of thymocytes, as assessed by CD4 and CD8 expression, appeared to be normal at birth (Fig. 5G). Similarly, the primordia of the parathyroid glands, as defined by Gcm2 expression (Gordon et al., 2001), remained attached to the endodermal epithelium and were smaller than normal (Fig. 5J,K). In addition, the primordia of the ultimobranchial bodies, which were characterized by their modest expression of Nkx2.1 (Kusakabe et al., 2006), also failed to separate from the epithelium in the pharyngeal pouch (Fig. 5L,M). By contrast, no obvious abnormalities were identified in the derivatives of the first and second pharyngeal pouches, including the thyroid gland, which was defined on the basis of its strong Nkx2.1 expression (Fig. 5L,M). Thus, Ripply3 appears to be required for the development and separation of organs derived from the pharyngeal pouches.

The defects described above suggested that the development of the pharyngeal apparatus was defective in the Ripply3^–/– embryos. Therefore, we examined the morphology of the pharyngeal apparatus of the Ripply3^–/– embryos at E9.5 and E10.5. Although the formation of the first and second arches was normal in the Ripply3^–/– embryos, the third and fourth arches were severely reduced in size, and the ectoderm and endoderm layers were almost adjacent to each other in this region (Fig. 6A-F). In this caudal pharyngeal region, apoptotic cell death had increased dramatically (Fig. 6G,H), whereas the rate of cell proliferation had not obviously changed (Fig. 6I,J). Thus, the reduction in the size of the third and fourth arches in Ripply3^–/– embryos appeared to have been due to increased cell death. Because Ripply3 was expressed in the pharyngeal endoderm and adjacent ectoderm, but not in the mesoderm, of the pharyngeal arches, the defect in cell survival within the pharyngeal arches was likely to be a secondary consequence of defective Ripply3 function in the ectoderm and/or endoderm. Also, by monitoring AP-2α (Tcfap2α – Mouse Genome Informatics) and Hand2 expression (Fig. 6K-N) (Chazaud et al., 1996; Srivastava et al., 1995), we confirmed that the neural crest cells failed to populate the third and fourth pharyngeal arches. The abnormal development of the neural crest cells might have resulted in the failure of cardiac development in the Ripply3^–/– embryos.

Next, we examined the Ripply3^–/– embryos for expression of several marker genes that are specifically expressed around the pharyngeal region. Expression of Pax1 and Pax9, which is normally observed in the pharyngeal pouch endoderm (Wallin et al., 1996; Peters et al., 1998), was continuous rather than segmental in the endoderm posterior to the second pouch (Fig. 6O-R). Thus, most of the endoderm cells in the third and fourth arch regions had characteristics of the pharyngeal pouch in Ripply3^–/– embryos. By contrast, Gbx2 expression, a Tbx1 target in the pharyngeal ectoderm, was not changed in Ripply3^–/– embryos (see Fig. S5 in the supplementary material) (Calmont et al., 2009), suggesting that Ripply3^–/– embryos show no obvious defect in the pharyngeal ectoderm.

Enhancement of Pax9 expression was specifically observed in the third and fourth arch regions in Ripply3-deficient embryos at ∼E9.25-10.0 (Fig. 6Q,R; see Fig. S6 in the supplementary material). Thus, consistent with the in vitro luciferase assay indicating that Ripply3 can repress Tbx1-mediated Pax9 expression (Fig. 2F,G), Ripply3 is likely to be involved in the repression of transcriptional activation by Tbx1 in vivo. In contrast to Pax9, the expression levels of Pax1 and Tbx1, as well as those of Fgf-related genes including Fgf8, Fgf3, Etv4 and Etv5, were not significantly changed in the third and fourth arch regions of Ripply3-deficient embryos (Fig. 6S-V; see Figs S7 and S8 in the supplementary material), although the expression domains of some of these genes were altered, probably owing to morphological changes in this region.

Similar to Ripply3^–/– embryos, Tbx1^–/– embryos also lacked the characteristic segmental pattern of the caudal pharyngeal arches. However, we noted that the morphology of this region differed between these two mutant embryos. As shown above, in the third and fourth arch regions of Ripply3^–/– embryos, the entire ectoderm and endoderm layers were adjacent to each other, like those at the pharyngeal pouch, and endodermal pouch markers were expressed in almost all endoderm cells (Fig. 6B,D). By contrast, in Tbx1^–/– embryos the ectoderm and endoderm layers were completely separated, and no characteristics of the pharyngeal pouch with respect to morphology or expression of molecular markers, including Pax1, were observed in the caudal pharyngeal region of these embryos (Fig. 7E; see Fig. S9 in the supplementary material) (Vitelli et al., 2002; Jerome and Papaioannou, 2001). Given that Ripply3 functions oppositely to Tbx1, this difference in phenotype in the third and fourth arch regions might be explained by the difference in their functions. Furthermore, as predicted from the in vitro luciferase assay showing that the function of Ripply3 is dependent on Tbx1, Tbx1^–/–;Ripply3^–/– embryos exhibited a phenotype identical to that of Tbx1^–/– embryos, and not to that of Ripply3^–/– embryos, in the caudal pharyngeal region at E10.0-10.5 (Fig. 7E,F; see Fig. S9 in the supplementary material), as well as in the cardiovascular system at E17.5 (see Fig. S10 in the supplementary material).

Because Tbx1 heterozygous mice frequently exhibit hypoplasia of their fourth PAAs, and a reduction in the activity of certain genes enhances this phenotype (Lindsay et al., 2001; Calmont et al., 2009), we also examined this phenotype in Tbx1^+/–;Ripply3^+/+ versus Tbx1^+/–;Ripply3^+/– embryos. However, the penetrance and severity of this phenotype were not significantly changed by the dosage reduction to one copy of the Ripply3 allele (Fig. 7B,C; see Table S2 in the supplementary material), suggesting that any interaction between Ripply3 and Tbx1 is not involved in the fourth PAA hypoplasia phenotype.

A striking feature of Ripply3^–/– embryos is their almost complete lack of the third and fourth pharyngeal arches. In the area of these arches only a few mesodermal and neural crest cells were observed in the Ripply3^–/– embryos. As Ripply3 expression was not found in the mesoderm, nor in neural crest cells in the arches, the abnormal formation of the pharyngeal arches might be a consequence of the abnormal development of the pharyngeal endoderm, in which Ripply3 is expressed.

The disrupted development of the caudal pharyngeal arches appears to have resulted in two distinct cardiovascular defects in Ripply3^–/– embryos. One involved an abnormality in heart development. The reduction in neural crest cell number in the caudal pharyngeal arches appears to have led to the abnormal development of the outflow tract, including hypotrophy of the aorta and incomplete formation of the ventricular septum. However, in spite of the reduction in the number of neural crest cells in the third and fourth arches, the formation of the outflow septum, which is also orchestrated by neural crest cells, was achieved in Ripply3^–/– embryos. Thus, we speculate that the reduction in the number of neural crest cells might not have been so severe as to cause a separation defect in the outflow tract or that neural crest cells migrating through the sixth pharyngeal arches might have contributed to septum formation. The other defect involved the loss of the third and fourth PAAs. This defect appeared to result in abnormal development of the vascular system, including deletion of the aortic arch and misshapen major blood vessels.

Interestingly, the second PAAs, which normally disappear after E10.5, persisted until birth in Ripply3^–/– embryos. Persistence of the first and second PAAs has also been reported in endothelin 1-deficient mouse embryos, in which the fourth PAAs are also poorly developed, suggesting that the formation of PAAs proximal to the heart might be a prerequisite for the regression of the more distal PAAs (Kurihara et al., 1995). We speculate that development of the proximal PAAs resulted in a decrease in blood flow running through the existing distal PAAs, which in turn decreased the mechanical stress caused by blood flow, making it virtually impossible for the distal PAAs to be maintained (Yashiro et al., 2007).

Ripply3 expression in the developing pharyngeal apparatus was dynamic and was observed in endoderm and ectoderm cells in the pouches and in the caudal pharyngeal region, suggesting that Ripply3 plays specific roles in these cells. In the Ripply3^–/– embryos, phenotypic abnormalities were first identified in the pharyngeal pouches forming posterior to the second arch. In these embryos, Pax1 and Pax9, which are specifically expressed in the pharyngeal pouches of normal embryos (Wallin et al., 1996; Peters et al., 1998), are uniformly expressed throughout the entire pharyngeal endoderm region posterior to the second pouch. Thus, it appears that Ripply3 activity is required for the proper development of the posterior pharyngeal arches, although it remains unclear why only caudal pouches were affected in the Ripply3^–/– embryos.

In the later stages of mouse development the caudal pharyngeal pouches give rise to several organs, including the thymus, parathyroid glands and ultimobranchial bodies (Manley and Capecchi, 1998; Kusakabe et al., 2006). The primordia of these organs become specified at particular positions around the pouches, separate from the epithelial sheet of the pharyngeal pouch, and migrate ventrocaudally before reaching their final adult positions. The precise molecular mechanisms underling these processes remain to be elucidated. In Ripply3^–/– embryos, the thymus, parathyroid glands and ultimobranchial bodies did not separate from the pharyngeal endoderm but maturated ectopically. This suggests that Ripply3 is specifically required for the separation processes that result in the development of these organs from the pharyngeal epithelial sheets. By contrast, specification and differentiation of these primordia were not affected by the absence of Ripply3 activity, although some of these organs appeared smaller than those in normal embryos. Thus, Ripply3 is unlikely to be required for the specification and maturation of the organ primordium. These results suggest that the mechanisms underlying the movement of the organ primordium differ from those that are required for their specification and maturation.

The development of the thymus and parathyroid glands is also abnormal in mouse embryos exhibiting altered Tbx1 dosages (Liao et al., 2004; Zhang and Baldini, 2008). Whereas Tbx1 null mutants exhibit a complete lack of thymus and parathyroid glands, embryos with an increased Tbx1 dosage frequently exhibit ectopic thymus and parathyroid glands. Thus, Tbx1 appears to play an important role in several distinct aspects of the development of these organs, including their growth and migration. Given that the development of these organs in Ripply3-deficient embryos partly resembles that of mouse embryos overexpressing Tbx1 (Merscher et al., 2001; Liao et al., 2004), we presume that Ripply3-mediated repression of Tbx1 activity was involved in the Tbx1-mediated migration of these primordia from the pharyngeal endoderm.

Interestingly, ectopic location of the thymus has also been reported in several mouse mutants, as well as in humans. For instance, Pax9-deficient mouse embryos exhibit an ectopic thymus located in the larynx, as in the case of Ripply3-deficient mice, suggesting that there is interplay between Pax9 and Ripply3 in the movement of the thymus primordium (Hetzer-Egger et al., 2002). However, in contrast to Pax9-deficient embryos, we showed that expression of Pax9 was not reduced, but rather increased, in the pharyngeal endoderm in Ripply3-deficient embryos. Thus, too much or too little Pax9 appears to inhibit proper migration of the thymus primordium. One of the roles of Ripply3 might be the maintenance of an adequate expression level of Pax9 in the pharyngeal endoderm.

In addition to the resemblance in pharyngeal phenotype, many other results of this study support the idea that Ripply3 represses the function of Tbx1 in the pharyngeal endoderm. We showed by in situ hybridization that Ripply3 and Tbx1 are strongly co-expressed in the caudal pharyngeal endoderm. We also showed that the transcriptional activity of mouse Tbx1 is repressed by mouse Ripply3. In addition, the expression of Pax9, a downstream gene activated by Tbx1, was upregulated in Ripply3-deficient embryos. Thus, we speculate that Ripply3-mediated repression of Tbx1 activity is crucial for the normal development of the pharyngeal derivatives.

During pharyngeal development, Tbx1 is expressed in multiple cell types, including endodermal, mesodermal and ectodermal cells (Chapman et al., 1996; Vitelli et al., 2002; Yamagishi et al., 2003). Conditional gene knockout in the endoderm or mesoderm indicates that Tbx1 has specific roles in these tissues (Arnold et al., 2006; Zhang et al., 2006). In addition, Tbx1 is required for gene expression in the pharyngeal ectoderm (Calmont et al., 2009). Thus, Tbx1 plays multiple roles in endodermal, mesodermal and ectodermal cells during pharyngeal development. Because it is not expressed in the mesoderm, Ripply3 would appear to modulate the function of Tbx1 in a cell type-specific manner. In this study, we also showed that hypoplasia of the fourth PAAs, which is frequently observed in Tbx1 heterozygous mice, was not significantly changed by dosage reduction to one copy of the Ripply3 allele. Since Ripply3 is specifically expressed in the pharyngeal endoderm and ectoderm, but not in the mesoderm, this dosage reduction of Ripply3 was unlikely to affect the activity of Tbx1 in the mesoderm, which appears to be important for the development of PAAs. Thus, our finding of no significant enhancement of the PAA hypoplasia phenotype by dosage reduction of Ripply3 might represent cell type-specific regulation of Tbx1 by Ripply3.

In the pharyngeal endoderm, expression of Pax9, a target of Tbx1, was upregulated in Ripply3-deficient embryos. Thus, Ripply3 appears to control Tbx1 activity in the pharyngeal endoderm. Interestingly, alterations in the gene dosage of Tbx1, in addition to those in its mRNA level, have been shown to result in a wide range of phenotypes, suggesting that Tbx1 activity is tightly controlled during pharyngeal development (Liao et al., 2004; Zhang and Baldini, 2008). Given that Pax9 is expressed at some particular level in wild-type embryos even in the presence of Ripply3, we propose that Ripply3-mediated repression contributes to the stringent control of Tbx1 activity in the pharyngeal endoderm.

In summary, we have identified a key gene that acts cooperatively with Tbx1 in the development of the pharyngeal apparatus and its derivatives. Our results suggest that Tbx1 activity is modulated in the pharyngeal endoderm in a Ripply3 expression-dependent manner. Our findings provide new insight into the molecular mechanisms underlying the development of the pharyngeal apparatus.

We thank Dr N. Takeda of Kumamoto University for the blastocyst injection; Drs S. Kobayashi, C. Noda, M. Sasaki, A. Ohbayashi, N. Ohbayashi, S. Arai, R. Takada, Mr T. Yokota and Ms M. Furutani for technical support; Drs B. Morrow, B. Hogan, N. Manley, P. Gruss and A. Baldini for providing materials; and Dr A. Sato and members of the S.T. lab for helpful discussions. This work was supported in part by a grant-in-aid for scientific research (S.T. and T.O.) and by the Program for Promoting the Establishment of Strategic Research Centers, Special Coordination Funds for Promoting Science and Technology (R.M.) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.