Mesodermal expression of Tbx1 is necessary and sufficient for pharyngeal arch and cardiac outflow tract development

The pharyngeal apparatus is an important transient embryonic structure composed of a series of reiterated bulges, called pharyngeal arches, that give rise to a variety of craniofacial, cervical and thoracic organs and structures. Perturbation of normal pharyngeal development causes many congenital diseases, such as DiGeorge syndrome (reviewed by Lindsay, 2001) and branchio-oto-renal syndrome (Chen et al.,1995), Opitz syndrome (Robin et al., 1995). Pharyngeal morphogenesis requires precisely co-coordinated interaction of pharyngeal endoderm, surface ectoderm,pharyngeal and splanchnic mesoderm, and neural crest-derived cells. Classic avian embryonic transplantation studies(Noden, 1983) suggest that the neural crest may play a primary role in patterning the pharyngeal apparatus. However, Trainor et al. have suggested that the maintenance of transplanted neural crest cell fate in those experiments might be due to co-transplanted isthmic tissue (Trainor et al.,2002a). Other studies suggest that neural crest cells need environmental signals to maintain their identity(Saldivar et al., 1996; Trainor and Krumlauf, 2000). In addition, it has been shown that pharyngeal arch segmentation and the formation of the 2nd and 3rd pharyngeal pouches can occur in the absence of neural crest cells in chick (Veitch et al., 1999). Although these data do not support a primary role of neural crest cells in patterning the pharyngeal arches, other studies have shown that these cells have an instructive role in patterning facial structures (Helms and Schneider,2003; Schneider and Helms,2003). More recently, it has been suggested that pharyngeal endoderm may play an important role in pharyngeal patterning(Graham et al., 2005). For example, the zebrafish Tbx1 mutant vgo exhibited severe segmentation defects that could be partially corrected by transplantation of wild-type endoderm-fated cells (Piotrowski et al., 2003; Piotrowski and Nusslein-Volhard, 2000). However, whether or not the pharyngeal mesoderm has a role in pharyngeal segmentation is unclear.

Tbx1 encodes a transcription factor of the T-box gene family. The gene is haploinsufficient in humans, and is thought to play a major role in the pathogenesis of DiGeorge syndrome(Jerome and Papaioannou, 2001; Lindsay et al., 2001; Merscher et al., 2001; Yagi et al., 2003). Investigators have accumulated a substantial amount of data supporting interactions between Tbx1 and major signaling systems such as the fibroblast growth factor (FGF) (Hu et al.,2004; Vitelli et al.,2002b; Xu et al.,2004), hedgehog (Yamagishi et al., 2003), retinoic acid(Guris et al., 2006; Roberts et al., 2005) and vascular endothelial growth factor(Stalmans et al., 2003)signaling. These reports underscore the intricacy of the role of Tbx1 in mammalian embryonic development. We and others have initiated an extensive dissection of the mouse mutant phenotype using conditional time- and tissue-specific ablation and dose manipulation(Arnold et al., 2006; Hu et al., 2004; Liao et al., 2004; Xu et al., 2005; Xu et al., 2004; Zhang et al., 2005). Results indicated that at all developmental times and in most tissues tested there is a crucial role for Tbx1, and that different structures have different sensitivity to Tbx1 dose.

Tbx1 is mainly expressed in tissues that form the embryonic pharyngeal system, i.e. surface ectoderm, pharyngeal endoderm, head mesenchyme, core mesoderm, splanchnic mesoderm, but not neural crest-derived mesenchyme (Chapman et al.,1996; Vitelli et al.,2002a). The pharyngeal endoderm expression domain has stimulated considerable attention because of its dynamic nature, and because mutants have hypoplasia and defective segmentation of the pharynx(Jerome and Papaioannou, 2001; Lindsay et al., 2001; Vitelli et al., 2002a). Indeed, Tbx1 ablation reduces the proliferation of endodermal cells(Xu et al., 2005), while heterozygous ablation of the gene in pharyngeal epithelia (ectoderm and endoderm but not in the mesoderm) causes vascular abnormalities characteristic of the Tbx1 haploinsufficiency phenotype(Zhang et al., 2005). In addition, it has been shown that homozygous, conditional ablation of Tbx1 by the Foxg1^Cre driver, which induces recombination predominantly in the pharyngeal endoderm, causes a mutant phenotype similar to that of germ line null mutants. However, tissues that express Tbx1 interact closely during development, raising the issue of whether Tbx1 may be required in multiple tissues to contribute to morphogenesis of the pharyngeal system. In this study, we use a novel approach, i.e. tissue-specific re-activation of the gene in a mutant background, as well as classic tissue-specific gene ablation, to address this. Results show that most of the developmental defects generated by null mutation of Tbx1 are recapitulated by mesodermal-specific somatic deletion of the gene, while mesodermal reactivation of the gene in a mutant background rescues most of those defects. Our data revealed a previously unknown instructive role of mesoderm in patterning pharyngeal segmentation. In addition, the combination of tissue-specific ablation and tissue-specific re-activation revealed different requirements of Tbx1 expression in the development of different pharyngeal structures. Our in vivo analysis suggests that mesodermal Tbx1 expression supports proliferation and regulates the expression of Fgf8 in the splanchnic mesoderm/anterior heart field.

All the experiments involving mice were approved by the Institutional Animal Care and Use Committee of Baylor College of Medicine, and in compliance with the USA Public Health Service Policy on Humane Care and Use of Laboratory Animals.

The following mouse mutant alleles used in this study have been reported previously: Tbx1^flox, Tbx1^ΔE5, Tbx1^mcm(Xu et al., 2004), Tbx1^lacZ (Lindsay et al., 2001), Mesp1^Cre(Saga et al., 1999) and R26R(Soriano, 1999). All lines were backcrossed into the C57Bl/6 genetic background for at least two generations. PCR strategies for mouse genotyping have been described in the original reports.

The Tbx1^neo2 allele was generated by gene targeting in AB2.2 ES cells, as shown in Fig. 3A. A loxP-flanked PGKneo cassette (neomycin resistant gene driven by the PGK promoter) was inserted in an intron between exon 5 and exon 6 of Tbx1 by homologous recombination. ES cells were injected into C57BL/6 blastocysts and chimeric mice were crossed with C57BL/6 mice to obtain germ line transmission of the mutant allele. The Tbx1^neo2allele was genotyped with primer pair 5′-GCCAGAGGCCACTTGTGTAG-3′and 5′-AGGCTGGGATTCCAAAAGAC-3′.

Quantification of Tbx1 expression was carried out using the TaqMan Gene Expression Assay system (Applied Biosystem) on total RNA extracted from E9.5 wild-type, Tbx1^neo2/+, Tbx1^neo2/neo2 and Tbx1^neo2/- embryos. We used commercially available probe and primer sets (assay IDs are Mm01342800_gH for Tbx1 and Mm00607939_s1 for the β-actin control).

We used an inducible allele Tbx1^mcm/+ to map the distribution of Tbx1-expressing cells during embryogenesis as previously described (Xu et al.,2005). We injected tamoxifen (Sigma) intraperitoneally into pregnant female (75 mg/kg body weight) at E6.5 and half this dose on the following 2 days.

Aortic arch arteries and hearts of E18.5 embryos were isolated by manual dissection and photographed under a stereomicroscope. Hearts were embedded,sectioned and stained with Hematoxylin and Eosin. Earlier embryos were examined under the stereomicroscope, fixed and embedded in paraffin for histological analysis. Intracardiac ink injection was performed to visualize pharyngeal arch arteries at E10.5. Embryos were then fixed and dehydrated in ethanol:water:acetic acid:chloroform (95:3:1:1) solution and cleared in methyl salicylate:benzyl benzoate (50:50) solution. β-Galactosidase activity was revealed by staining paraformaldehyde-fixed embryos with the X-gal substrate,according to standard procedures. Stained whole-mount embryos were photographed and then embedded in paraffin and cut into 10 μm histological sections. Sections were counterstained with Nuclear Fast Red.

Whole-mount and tissue section RNA in situ hybridization with nonradioactive probes was performed as previously described(Albrecht et al., 1997), using probes for Pax1 (kindly provided by Dr R. Balling), Tbx1(from Dr V. Papaioannou), Crabp1(Giguere et al., 1990) and Fgf8 (from Dr G. Martin). The anti Hoxb1 antibody has been kindly provided by Dr N. Manley. The anti neurofilament-M monoclonal antibody 2H3 was obtained from the Developmental Studies Hybridoma Bank. Anti Hoxb1 and neurofilament-M were used in whole-mount immunohistochemistry. Proliferating cells were labeled by immunohistochemistry on paraffin-embedded sections using an anti phosphorylated histone H3 antibody (Upstate) and assayed on three controls and three conditional mutants. Tissue-specific deletion and restoration of Tbx1 expression was confirmed by immunofluorescence on cryosections with an anti Tbx1 antibody (Zymed).

Mesp1 is initially expressed in epiblast cells sorting through the primitive streak and is quickly downregulated after E7(Saga et al., 1999), i.e. before Tbx1 is turned on. The progeny of Mesp1-expessing cells contributes heavily to the cranial mesoderm(Saga et al., 1999; Zhang et al., 2005). Mesp1^Cre-induced recombination, as visualized by crosses with the R26R reporter (Soriano,1999), is restricted to mesodermally derived tissues and was not detected in endodermal or ectodermal cells(Fig. 1B). Based on fate mapping of MespI-expressing and Tbx1-expressing cells(Fig. 1A), we expected that Mesp1^Cre would ablate Tbx1 expression in all its mesodermal domains of the pharynx, head mesenchyme and splanchnic mesoderm. To conditionally ablate Tbx1 in these domains, we crossed Mesp1^cre/+;Tbx1^ΔE5/+mice with Tbx1^flox/flox mice. Immunohistochemistry using an anti Tbx1 antibody confirmed the loss of the protein in mesodermally derived tissues but not in the endoderm of Mesp1^cre/+;Tbx1^ΔE5/floxembryos at E9.5 (Fig. 1C,D),although we did detect some residual Tbx1 immunoreactivity in the core of the 1^st pharyngeal arch (arrow in Fig. 1D). At E9.0, we obtained a similar result and confirmed the preservation of the ectodermal signal(Fig. 1E-F). Thus, Mesp1^Cre-driven deletion is specific for the mesoderm and does not affect (directly or indirectly) epithelial expression of Tbx1.

We examined the morphological phenotype of Mesp1^cre/+; Tbx1^ΔE5/flox embryos (hereafter referred to as M-ko) at different developmental stages. At E18.5, M-ko embryos(n=15) exhibited a phenotype very similar to that observed in Tbx1^-/- embryos. In particular, M-ko embryos had hypoplastic external ears (15/15, Fig. 2A,A′), thymic aplasia or severe hypoplasia (12/15 and 3/15,respectively, Fig. 2B,B′), persistent truncus arteriosus (PTA) and ventricular septal defects (VSD) (15/15, Fig. 2C-D′), as well as aortic arch defects (15/15, Fig. 2C,C′). However,with one exception, M-ko embryos did not exhibit cleft palate, a common feature of Tbx1^-/- mutants (not shown). At E10.5, M-ko embryos (n=15) presented with hypoplasia of the 2nd pharyngeal arches(Fig. 2E,E′), loss of the 3rd, 4th and 6th pharyngeal arches and pharyngeal arch arteries (PAA)(Fig. 2F,F′), and severe hypoplasia of the pharynx (Fig. 2G,G′). RNA in situ hybridization with Pax1, a marker of pharyngeal pouch endoderm, showed no labeling of the 2nd and 3rd pharyngeal pouches of M-ko embryos (Fig. 2J,J′), while the 4th pouch, identified by immunohistochemistry with an anti Hoxb1 antibody, appeared to be smaller(Fig. 2K,K′) though not as severely affected as that of null mutants (data not shown). These data suggest that Tbx1 in the mesoderm regulates a signaling pathway required for pharyngeal endoderm development.

Neural crest cell distribution is disrupted in Tbx1 homozygous mutants (Vitelli et al.,2002a). To understand whether this phenotype depends upon loss of mesodermal Tbx1 expression, we used RNA in situ hybridization with a neural crest marker, Crabp1 on M-ko embryos. The results revealed abnormal distribution of this cell population. Specifically, we detected reduced labeling of the pre-otic stream directed to the 2nd pharyngeal arch,apparent interruption of the post-otic stream directed to the 3rd pharyngeal arch and abnormal distribution of the circumpharyngeal stream(Fig. 2H,H′). Consistent with these data, we observed abnormalities of cranial nerve pathways(Fig. 2I,I′). Cranial nerves are partially derived from neural crest cells and have a similar migratory pathway. In M-ko embryos, the mandibular branch of the trigeminal(V) was fused caudally with the facial (VII) nerve. The glossopharyngeal (IX)nerve appeared hypoplastic. Terminal projections of the glossopharyngeal, the vagus (X) and the accessory (XI) showed disarray and were fused with each other (Fig. 2I,I′). These findings were similar to those observed in Tbx1-null mutants(Vitelli et al., 2002a). These results indicate that mesodermal Tbx1 expression affects neural crest cell distribution.

Tissue-specific deletion can determine if an expression domain is necessary, but not if it is sufficient, for a particular developmental process. As Tbx1 is expressed in multiple interacting tissues during pharyngeal development, it is reasonable to hypothesize that the gene may be necessary in multiple tissues to contribute to the complex morphogenesis of pharyngeal derivatives. To address this hypothesis, we designed and generated a new allele of Tbx1 that has a low expression level but that can be reverted to wild-type level upon Cre recombination. To this end, we inserted a loxP-flanked PGKneo cassette into intron 5 by homologous recombination(Fig. 3A) and established the allele (Tbx1^neo2) in mice. We used quantitative real-time PCR to evaluate the amount of Tbx1 mRNA in Tbx1^neo2/- embryos at E9.5 and estimated that the allele produced ∼20% of the wild-type mRNA level (data not shown). To establish whether there may be tissue-specific differences in Tbx1mRNA or protein expression, we used RNA in situ hybridization (not shown) and immunofluorescence on frozen section (Fig. 3B). With both techniques, we observed homogeneous, strong reduction of the signal without any obvious tissue differences. We predicted that removal of the PGKneo cassette by Cre recombination (generating the allele Tbx1^neo2Δ) would revert this allele to a functionally `normal' allele. Indeed, Tbx1^{neo2Δ/neo2Δ}animals where viable, apparently normal and fertile (not shown). By contrast, Tbx1^neo2/+ and Tbx1^neo2/ΔE5 animals presented with abnormalities similar to those observed in Tbx1^ΔE5/+ and Tbx1^ΔE5/ΔE5animals, respectively, as detailed below. To test whether the neo2allele could be reactivated in a tissue-specific manner, we crossed Mesp1^Cre/+;Tbx1^ΔE5/+with Tbx1^neo2/+ animals and carried out immunofluorescence on Tbx1^neo2/ΔE5 and Mesp1^Cre/+;Tbx1^neo2/ΔE5E9.5 embryos, using an anti Tbx1 antibody on frozen sections. Results showed that although Tbx1^neo2/ΔE5 embryos had near-background signal levels, Mesp1^Cre/+;Tbx1^neo2/ΔE5embryos had a robust signal in mesodermal tissues(Fig. 3B,B′). Thus, Tbx1^neo2 is a hypomorphic allele that reverts to a functional allele upon Cre recombination in vivo. Because Mesp1^Cre-induced recombination precedes the onset of Tbx1 expression, it is predictable that the reactivation of the allele occurs from the onset of Tbx1 expression.

Tbx1^neo2/ΔE5 embryos exhibited phenotypic abnormalities of the same type as those seen in Tbx1^ΔE5/ΔE5 embryos, although in some cases with a milder expressivity. Specifically, the cardiovascular phenotype was very severe (n=20) and included PTA, aortic arch defects(Fig. 3E) and VSD(Fig. 3F). No thymus(n=17) could be observed except for three embryos that exhibited severe hypoplasia (Fig. 3D). The external ear (n=20) and the second pharyngeal arch(n=23) were hypoplastic (Fig. 3C,G). The latter two phenotypic abnormalities were present with different levels of severity but were never absent. The pharynx was hypoplastic and the 3rd, 4th, and 6th pharyngeal arches were not segmented(Fig. 3I). The 4th PAAs were absent in all the mutants (n=23), and the 3rd and/or 6th PAA were missing in one or both sides (Fig. 3H). Reactivation of mesodermal expression of Tbx1 in Mesp1^Cre/+;Tbx1^neo2/ΔE5embryos rescued completely the OFT defects (PTA and VSD, Fig. 3E′,F′), the formation and remodeling of the 3rd and 6th pharyngeal arch arteries(Fig. 3H′,I′), and the hypoplasia of the 2nd pharyngeal arch and of the external ear(Fig. 3C′,G′), but it did not rescue the thymic aplasia (Fig. 3D′) or the 4th pharyngeal arch and 4th PAA aplasia(Fig. 3H′,I′). The pharyngeal patterning defects were partially rescued. Tbx1^neo2/ΔE5 embryos had a hypoplastic 2nd pharyngeal pouch and no detectable 3rd pouch, as detected by Pax1expression (Fig. 3J). Mesodermal reactivation was associated with normalized 2nd pouch signal and partial 3rd pouch signal (Fig. 3J′, compare with Fig. 2J for wild-type pattern). The 4th pharyngeal pouch (as revealed by Hoxb1 immunohistochemistry), was very hypoplastic in Tbx1^neo2/ΔE5 embryos but was partially rescued by mesodermal reactivation of Tbx1(Fig. 3K,K′). Finally,the neural crest migration and cranial nerve pathway abnormalities observed in Tbx1^neo2/ΔE5 embryos were only marginally improved by mesodermal reactivation (Fig. 3L,L′,M,M′, compare with Fig. 2H,I for wild-type patterns).

Overall, these rescue experiments identified processes that are exquisitely dependent upon mesodermal expression of Tbx1. These are OFT development, 2nd pharyngeal arch morphogenesis, 3rd and 6th pharyngeal arch arteries. By contrast, 4th pharyngeal arch and PAA morphogenesis, thymic development, neural crest migration and cranial nerve guidance are dependent upon expression in the mesoderm and epithelia.

Having defined the mesoderm-specific developmental roles of Tbx1, we asked how these roles may be effected. Timed deletion of Tbx1 has shown that the structures affected by mesodermal-specific gene ablation require Tbx1 approximately between E8.0 and E9.0(Xu et al., 2005). We,therefore, focused our attention on E8.5 M-ko embryos. The Tbx1 gene is turned on at approximately E8 in the paraxial mesoderm, and in the adjacent ectoderm and endoderm (Fig. 4A,A′). At these stages, M-ko mutants were indistinguishable from controls. However, immunostaining with anti-phospho H3 antibody, which detects mitotic cells, showed that mutants had severely reduced mitotic activity (Fig. 4B,B′,E). By contrast, immunostaining with an anti-cleaved caspase 3 antibody did not reveal changes in apoptotic activity (data not shown). It has been shown that Tbx1 interacts with and may directly regulate Fgf8 gene expression (Hu et al., 2004; Vitelli et al., 2002b), and loss of Fgf8 can downregulate mitotic activity in pharyngeal mesoderm(Park et al., 2006). Therefore, we tested Fgf8 expression in M-ko mutants at E8.5, E9.0 and E9.5. At all stages tested, the pharyngeal epithelial expression of Fgf8 was conserved (with the caveat that the pharynx of M-ko embryos is hypoplastic, and, therefore, expression domains appear proportionally smaller) (Fig. 5). By contrast,the early splanchnic mesoderm/anterior heart field domain detectable at E8.5,was lost or strongly downregulated (Fig. 5A′-B′). This region also expresses Tbx1(Nowotschin et al., 2006; Zhang et al., 2005). Shh expression was maintained at all the stages tested (E8.5, E9.0 and E9.5, not shown). Thus, reduced proliferation of mesenchymal cells in M-ko embryos may be secondary to downregulation of autocrine Fgf8 signaling, but it is unlikely to be secondary to loss of epithelial signals.

The function of pharyngeal mesoderm in craniofacial myogenesis has been extensively investigated (Hacker and Guthrie, 1998; Kelly et al.,2004; Noden and Francis-West,2006), but little is known about its role in pharyngeal segmentation. Here, we show that loss of Tbx1 expression in the mesoderm suppresses pharyngeal pouch and posterior pharyngeal arch formation,and disrupts the distribution of neural crest-derived cells. These observations support a crucial role of the mesoderm in pharyngeal segmentation.

The pharyngeal mesoderm is composed of cells migrating from the cranial paraxial mesoderm (Trainor and Tam,1995; Trainor et al.,1994), which is loosely packed into meristic cell clusters called somitomeres (Tam and Trainor,1994). The somitomeristic mesoderm underlying the neural tube has a topographic relation to specific neural tube segments, and has regionalized cell fate along the craniocaudal axis. Cells migrate and colonize specific pharyngeal arches along with cranial neural crest cells from the same axial level (Trainor and Tam, 1995). Unlike the cranial neural crest streams, which are characterized by discrete Hox gene expression profiles(Hunt et al., 1991a; Hunt et al., 1991b),individual somitomeres do not appear to be identifiable by a gene expression code. Pharyngeal mesoderm was shown to provide permissive signals to maintain the identity of neural crest cell populations(Trainor and Krumlauf, 2000),and to generate `exclusion zones' separating streams of neural crest cells(Trainor et al., 2002b). However, it is unlikely that this function of the mesoderm is essential for pharyngeal segmentation, because neural crest cell ablation is not sufficient to disrupt pharyngeal segmentation (Veitch et al., 1999). Mesodermal Tbx1 is necessary but not sufficient for normal neural crest migration, whereas it is necessary and sufficient for, at least part of, the segmentation process. This suggests that the role of the mesoderm in segmentation may be cell-autonomous and/or dependent upon mesoderm-epithelial interactions. This is consistent with the findings that wild-type mesoderm can partially rescue pharyngeal segmentation defects in fgf8^-;fgf3-MO zebrafish mutants(Crump et al., 2004).

We have previously proposed that Tbx1 has a role in the expansion of cardiomyocyte precursors, but we were not able to exclude that at least part of the cardiac outflow tract phenotype may be due to reduced endodermal expression (Xu et al., 2005; Xu et al., 2004). This is because the key Cre driver used in those experiments, Nkx2.5^Cre, induced recombination in a region of the pharyngeal endoderm partially overlapping with Tbx1 expression(Xu et al., 2004). Here, we show that mesodermal ablation recapitulated the OFT abnormalities seen in Tbx1^-/- embryos, whereas mesodermal Tbx1restoration in a mutant background was sufficient to rescue those abnormalities. These results suggest that the role of endoderm expression in cardiac outflow development is marginal, if at all important, and assign the crucial role to mesodermal expression. However, Arnold et al.(Arnold et al., 2006) have shown that Tbx1 ablation using the Foxg1^Cre/+ driver results in a severe phenotype(including OFT defects) resembling that of Tbx1^-/-animals. The report showed that this driver induced recombination predominantly in the pharyngeal endoderm, suggesting a major role of Tbx1 in this tissue for OFT development. We have shown that the Foxg1^Cre/+ driver induces robust recombination not only in pharyngeal epithelia (endoderm and ectoderm) but also in pharyngeal mesoderm and secondary heart field(Zhang et al., 2005). Arnold et al. have used this driver in a different genetic background that clearly has attenuated extra-endodermal recombination. However, it cannot be excluded that residual recombination activity in the mesoderm might have reduced the dose of Tbx1 in this tissue to a level that affects normal OFT development.

Ablation of a gene from individual interacting tissues may potentially lead to similar morphological phenotypes without necessarily illuminating the role of the gene in a particular tissue. The strategy used here, i.e. tissue-specific gene reactivation in a mutant background, was designed to diminish this problem. Mesp1^Cre-driven deletion and restoration of Tbx1 expression showed that Tbx1 has several developmental roles that are confined to the mesoderm. However, the development of the thymus, of the 3rd and 4th pharyngeal pouches, and of the 4th pharyngeal arch require both mesodermal and epithelial expression. Thus,besides a cell-autonomous function in the mesoderm, Tbx1 may regulate interactions between different tissues. Exactly how this role is effected remains to be clarified, but considering that Tbx1 appears to interact with several of the major signaling systems that play fundamental roles in development, the candidate molecular pathways are manifold. Mechanisms could include transcriptional regulation of genes encoding extracellular ligands (Hu et al.,2004; Vitelli et al.,2002b; Xu et al.,2004) or of proteins involved in the catabolism of ligands(Guris et al., 2006; Ivins et al., 2005) or in signal transduction (Park et al.,2006).

The fact that the Tbx1^Neo2 allele is not null raises the issue of whether the rescue of some phenotypic abnormalities in the restoration experiments might have been `helped' by the residual expression of the neo2 allele in the endoderm and ectoderm. However, the observations that Tbx1^neo2/- embryos have essentially the same phenotype as Tbx1^-/- embryos and that Mesp1^Cre/+;Tbx1^flox/- animals also have a very severe and similar phenotype to Tbx1^-/-embryos, suggest that reduced or full expression of Tbx1 in non-mesodermal tissues, although required for several developmental processes,cannot rescue phenotypic abnormalities caused by loss of mesodermal Tbx1 expression.

To understand the earliest molecular consequences of Tbx1 loss of function in the mesoderm, we examined M-ko mutants at a stage previously shown to be crucial for Tbx1 function. At this stage (E8.5), the morphology of mutant embryos is normal but we found a strong reduction in cell proliferation in the region of the pharyngeal mesenchyme that normally expresses Tbx1. Such reduced proliferation could explain the strongly reduced cellularity in the pharyngeal arches and the cardiac outflow tract defects. Ataliotis et al. using lineage labeling experiments in Xenopus, suggested that Tbx1 has a cell-autonomous function in the pharyngeal mesoderm (Ataliotis et al., 2005).

We and others have proposed that the Tbx1 pro-proliferative activity may be effected, at least in part, by regulating FGF ligand gene expression. This view is supported by the demonstration that Tbx1 can indeed activate an Fgf8 enhancer in tissue culture assays(Hu et al., 2004), and by the genetic interaction between the two genes in vivo(Vitelli et al., 2002b; Vitelli et al., 2006). It has been shown that conditional ablation of Fgf8 in the mesoderm can cause OFT defects and reduce cell proliferation and survival(Ilagan et al., 2006; Park et al., 2006). We and others have previously shown that Tbx1 is required for endodermal,but not ectodermal, expression (Zhang et al., 2005), but Fgf8 expression in the mesoderm of Tbx1 mutants had not been tested before. Here, we show that mesodermal Tbx1 is required for Fgf8 expression in the splanchnic mesoderm/anterior heart field region. Whether or not the loss of Fgf8 in this tissue is sufficient to cause the OFT phenotype of M-ko mutants is unknown. Mesp1^Cre-driven deletion of Fgf8 causes early lethality in most embryos, preventing the full assessment of fetal consequences of this conditional mutation(Park et al., 2006). However, Fgf8 knock-in into the Tbx1 locus was unable to rescue or modify the OFT phenotype of Tbx1 mutants(Vitelli et al., 2006). In addition, Fgf8 loss in mesoderm reduces cell survival and proliferation, while Tbx1 mutation only affects proliferation. Further complication of the relationship between Tbx1 and Fgf8, is added by the report of downregulation of the FGF receptor Fgfr1 RNA expression in Tbx1^-/- mutants(Park et al., 2006). We propose that a reduction of FGF signaling, resulting from cell autonomous downregulation of ligand and receptor expression, has an important role in the cell proliferation phenotype of Tbx1 mutants. However, the role of other signaling systems should be investigated.

Overall, our data indicate a crucial role of the mesoderm in the pathogenesis of the DiGeorge-like phenotype and we show, for the first time,that the mesoderm has a direct influence on the proper morphogenesis of the pharyngeal pouches.

We thank A. Sobotka, G. Ji, P. Terrell, M. Woods and W. Yu for invaluable technical assistance; Y. Saga for Mesp1^Cre mice and N. Manley for anti-Hoxb1 antibody; and E. Lindsay and J. Martin for critical reading of the manuscript. This work was funded by grants HL051524, HL064832 and HL067155 from the National Institutes of Health (to A.B.).