OVCA1/DPH1 (OVCA1) encodes a component of the diphthamide biosynthesis pathway and is located on chromosome 17p13.3. Deletions in this region are associated with Miller–Dieker syndrome (MDS). Ovca1/Dph1 (Ovca1)-null mice exhibit multiple developmental defects, including cleft palate, growth restriction and perinatal lethality, suggesting a role in the craniofacial abnormalities associated with MDS. Conditional ablation of Ovca1 in neural crest cells, but not in cranial paraxial mesoderm, also results in cleft palate and shortened lower jaw phenotypes, similar to Ovca1-null embryos. Expression of transgenic myc-tagged Ovca1 in craniofacial structures can partially rescue the cleft palate and shortened mandible of Ovca1-null embryos. Interestingly, Ovca1-null mutants are resistant to conditional expression of diphtheria toxin subunit A in both neural crest cell and paraxial mesoderm derivatives. However, OVCA1-dependent diphthamide biosynthesis is essential for neural crest cell-derived craniofacial development but that is dispensable for paraxial mesodermal-derived craniofacial structures in mammals. These findings suggest that OVCA1 deficiency in the neural crest contributes to the craniofacial abnormalities in patients with MDS. Also, our findings provide new insights into the molecular and cellular mechanisms that lead to the craniofacial defects of MDS.
Miller–Dieker syndrome (MDS; OMIM 247200) is a human disease with severe developmental abnormalities, including lissencephaly (1), mental retardation, growth delay and cranial facial dysmorphism (2,3). The MDS critical deletion region spanning from PAFAH1B1 (coding for the beta subunit of platelet-activating factor acetyl hydrolase, also known as LIS1) to YWHAE (coding for 14-3-3ε) on chromosome 17p13.3 has been defined (Fig. 1) (4,5). Compared with individuals with larger deletions of chromosome 17p13.3, heterozygous point mutations and intragenic deletions in PAFAH1B1 result in an isolated lissencephaly sequence (OMIM 601545) with aberrant neural migration phenotypes, showing clinically non-syndromic milder lissencephaly (6). These clinical findings suggest that other genes within the MDS deletion region may additively or synergistically collaborate with PAFAH1B1 to enhance the neural migration deficits, resulting in severe lissencephaly of MDS. Using a gene-targeting approach in mice, Pafah1b1 and Ywhae mutants indeed exhibited dosage-dependent neuronal migration and cortical developmental defects consistent with the notion that other genes interact with PAFAH1B1, leading to more severe lissencephaly in MDS individuals (5,7–9).
In addition to lissencephaly, the main clinical manifestations of MDS patients also include craniofacial dysmorphism. The typical craniofacial abnormalities in MDS patients are prominent forehead, bitemporal hollowing, short upturned or broad nose, thickened upper lip and small jaw (2,3,10,11). It remains unclear whether facial abnormalities are a consequence of brain developmental defects or a separate phenotype of MDS caused by additional gene deletions within the MDS critical region. Although Pafah1b1 and Ywhae mutants show cerebral cortical developmental defects, no craniofacial abnormalities in these mice were reported (5,7–9). Several genes located between Pafah1b1 and Ywhae (Fig. 1), including Mnt (encoding an MYC-class basic helix–loop–helix leucine zipper transcription factor), Hic1 (encoding a pox virus zinc-finger domain-containing transcription factor) and Ovca1, also known as Dph1 (encoding a diphthamide biosynthesis enzyme), are mutated in mice (12–14). Interestingly, Mnt-, Hic1- and Ovca1-knockout embryos exhibit growth restriction and craniofacial abnormalities with cleft palate (12–14), suggesting that Mnt, Hic1 and Ovca1 are required for embryonic growth and craniofacial development, and that loss of these genes contributes to MDS. In Mnt-deficient mice, no overt phenotype in the brain has been reported in addition to craniofacial abnormalities (14). However, forebrain and midbrain protrusions have been described for Hic1- and Ovca1-deficient embryos, respectively (12,13). It is still unclear whether the craniofacial abnormalities are caused by the brain developmental defects in the Hic1 or Ovca1-knockout mice.
OVCA1 is an evolutionarily conserved gene, indicative of a fundamental biological role in a variety of species (15). In humans, OVCA1 was initially mapped within the deletion region of 17p13.3 involved in early ovarian cancer (16,17). Our initial study generating Ovca1-null mice, mainly addressed its tumor suppressive role and potential cooperative role in modifying p53-deficient tumorigenesis (12). OVCA1 acts as a component of an enzyme complex with DPH2, DPH3 and DPH4 that functionally catalyzes the first step of the diphthamide biosynthesis pathway, to posttranslationally modify a conserved histidine residue (His699 in yeast and His715 in human) of eukaryotic elongation factor 2 (eEF-2) (18–22). Diphtheria toxin subunit A (DTA) and Pseudomonas exotoxin A belong to a family of ADP-ribosyl transferases that target the modified diphthamide residue of eEF-2 for ADP ribosylation, thereby inhibiting protein translation (23–26). A recent report demonstrated that mouse embryonic fibroblasts (MEFs) with a Gly717-to-Arg (G717R) mutation of eEF-2, in which the first step of diphthamide modification is prevented, exhibit ribosomal −1 frameshifts during protein translation. Interestingly, Ovca1-null MEFs also exhibit a −1 translational frameshift (27). In addition, the Ovca1-null but not eEF-2 (G717R) mutant cells have a deficiency in protein elongation function (27). Strikingly, eEF2G717R/G717R;Ovca1−/− compound mutant mice exhibit less embryonic lethality compared with the 100% embryonic lethality of Ovca1−/− mutants, suggesting that diphthamide deficiency is the main cause of embryonic lethality in Ovca1/Dph1-null mice (27). To investigate whether Ovca1-mediated diphthamide biosynthesis plays a role in the craniofacial defects exhibited by individuals with MDS, we examined craniofacial development using both Ovca1-null allele (12) and newly generated Ovca1 conditional alleles. Conditional ablation of Ovca1 in neural crest cells, but not in cranial paraxial mesoderm, resulted in cleft palate and shortened mandible phenotypes similar to Ovca1-null mice. Transgenic Ovca1 expression could partially rescue the craniofacial defects in Ovca1-null mice. In addition, Ovca1-null embryos were resistant to DTA-induced cell death. These studies suggest that OVCA1 deficiency in neural crest-derived tissues contributes to the craniofacial dysmorphisms of MDS, and that diphthamide biosynthesis is essential for craniofacial development and embryonic growth.
Abnormal palatogenesis in Ovca1-null embryos
Previously, we found that Ovca1-null mice died perinatally and exhibited growth restriction and developmental defects (12). Here, we characterized the craniofacial structures of the Ovca1 mutants. In control and Ovca1-null fetuses, the lips and primary palate developed normally; however, all Ovca1-null fetuses displayed severe cleft palate at birth (Fig. 2A and B). Analysis of the Ovca1-null skull revealed that it was smaller and the palatine bone was absent compared with controls (Fig. 2C and D). Palate formation in the mutants compared with that in control embryos was examined histologically in coronal sections from E13.5 to E15.5 (Fig. 2E–P). The distance of palate shelves in control and mutant embryos was quantified (Fig. 2Q). Also, the sizes of palatal shelves in controls and mutants were measured (Fig. 2R). At E13.5, the palatal shelves of Ovca1-null embryos extended downward on either side of the tongue (Fig. 2G, H and R) and were smaller than the palatal shelves of control embryos (Fig. 2E, F and R). At E14.5, control palatal shelves extended medially above the tongue and were touching each other at the midline (Fig. 2I and J). However, the palatal shelves of Ovca1-null embryos remained on the lateral sides of the tongue (Fig. 2K and L). At E15.5, the control palatal shelves were completely fused at the midline (Fig. 2M and N), whereas the Ovca1-null palatal shelves were still developing but remained lateral to the tongue (Fig. 2O and P). Thus, the distance between palate shelves in the Ovca1-null embryos at E14.5 and E15.5 significantly differed from that in the control embryos (Fig. 2Q). Our data suggest that these palatal shelf morphogenesis defects are likely the cause of the cleft palate phenotype in Ovca1-null embryos (Fig. 2O and P).
Shortened mandible in Ovca1-null embryos
In addition to the absence of palatal shelf fusion at the midline, oral anatomic structures may also affect palatal fusion. For example, migration and condensation of cranial mesenchyme cells, growth of Meckel's cartilages, descent of the tongue and establishment of the oral cavity are all essential for the elevation and fusion of palatal shelves. As shown in Figure 2, the tongues of E14.5 and E15.5 Ovca1-null embryos failed to descend and this could have physically hindered the elevation and fusion of the palatal shelves (Fig. 2K, L, O and P). We next explored the development of the forming skeleton to identify potential craniofacial skeletal defects which might affect palatal fusion in Ovca1-null embryos. Although the overall size of Ovca1-null embryos was smaller than control embryos, Ovca1-null embryos displayed normal cartilage formation, including skull, rib, limbs and long bones, except for additional digits on the hindlimbs (Fig. 3A; Supplementary Material, Fig. S1), which resembled the polydactyly phenotype that occurs in rare individuals with MDS. Interestingly, Meckel's cartilages in the lower jaws of Ovca1-null embryos were shorter than those of the controls (Fig. 3A). We then measured the length of Meckel's cartilage that was normalized with crown-rump length. Meckel's cartilages of Ovca1-null embryos were significantly shorter than controls at E15.5 and E16.5 (Fig. 3B). Importantly, the small lower jaw phenotype (micrognathia) is one of the characteristic facial dysmorphologies in individuals with MDS. Underdeveloped jaws may also contribute to the development of cleft palate during palatogenesis. Collectively, these data suggest that OVCA1 contributes to MDS.
Palatal shelf elevation in ex vivo culture in the Ovca1-null head without tongue and lower jaw
To determine the contribution of the tongue and lower jaw to the palatal shelf defects in Ovca1-null embryos, we exploited an ex vivo organ culture system. E13.5 embryonic heads were isolated, and the mandible and tongue were removed for subsequent culture. Palate shelves at posterior portion of both Ovca1+/− control (Fig. 4A–C) and Ovca1−/− mutant (Fig. 4D–F) explants were in a vertical position prior to culture. On the next day, the control palatal shelves could elevate horizontally and start to extend toward midline without the mandible and the tongue (Fig. 4H and I). Although the size of cultured Ovca1−/− explants was smaller (Fig. 4K and L), the posterior portion of the palatal shelves was elevated and extended medially (Fig. 4L). No statistical significance was found in palate elevation angles between Ovca1+/− and Ovca1−/− explants after ex vivo culture (Fig. 4M). To ensure the explants remain healthy after cultured, TUNEL assay was performed to detect potential cell death. We found only a few of TUNEL-positive cells in both Ovca1+/− (Fig. 4N) and Ovca1−/− (Fig. 4O) explants. These data suggest that Ovca1-null palatal shelves can initiate elevation in the absence of the tongue and mandible. However, smaller palatal shelves (Fig. 2) and shorter lower jaw (Fig. 3) may combinatorially contribute to the defects of palatogenesis in Ovca1-null mutants.
Expression of exogenous Ovca1 partially rescues palate defects
We generated an Ovca1 transgenic mouse line to directly demonstrate that the palate abnormalities of Ovca1-null mice were primary defects caused by Ovca1 deficiency. The conditional expression cassette, which is driven by the cytomegalovirus enhancer/chicken actin promoter (CAG), is composed of a loxP-flanked β-geo gene with a polyadenylation sequence followed by a myc-tagged Ovca1 complementary DNA (cDNA) [Tg(CAG-β-geo-Ovca1-myc), Figure 5A]. The β-geo gene driven by the CAG promoter can be expressed ubiquitously. However, whole-mount X-gal staining of CAG-βgeo-Ovca1-myc transgenic embryos revealed β-galactosidase activity in some organs, but not ubiquitously. In E10.5 and E12.5 transgenic embryos, β-galactosidase activity was detected in the heart and neural tube (Fig. 5B and C). As development proceeded, β-galactosidase activity was detected in the nervous system, including the neural tube and forebrain (Fig. 5B–D). We also observed mosaic β-galactosidase expression in the body walls of E14.5 transgenic embryos. To examine the β-galactosidase activity in the craniofacial structures of E13.5 transgenic embryos, we removed the central nervous system and separated the lower parts of the oral cavity from the palatal shelves. No obvious β-galactosidase activity was detected in the palatal shelves (Fig. 5E), in contrast to intensive X-gal staining in the tongue (Fig. 5F). β-Galactosidase immunostaining showed strong fluorescence in the mandibles compared with mosaic expression in the tongue, and few scattered β-galactosidase-positive cells were noted in the palatal epithelium, indicating that the β-geo transgene was expressed mainly in the mandible and tongue among craniofacial structures of CAG-β-geo-Ovca1-myc transgenic embryos (Fig. 5G). To overexpress CAG-Ovca1-myc in mice, Tg(CAG-β-geo-Ovca1-myc) transgenic mice were bred with the Prm1-Cre transgenic mouse line that allows deletion of the loxP-flanked β-geo gene cassette in male germ cells (Fig. 5A). Because no overt phenotypic difference was observed in CAG-Ovca1-myc transgenic mice compared with controls, we then introduced the CAG-Ovca1-myc allele onto the Ovca1-null background (referred to as Ovca1−/−;CAG-Ovca1-myc). Although mosaic expression of the Ovca1 transgene failed to rescue the lethality of Ovca1-null mutants at weaning, the Mendelian ratios of Ovca1−/−;CAG-Ovca1-myc embryos were higher than the ratio of Ovca1−/− embryos at E14.5–16.5 (Supplementary Material, Table S1). Most importantly, compared with the palates of control embryos (Fig. 5H and K), 60% (three of five) of Ovca1−/−;CAG-Ovca1-myc embryos showed normal palatogenesis, including the elevation and fusion of palatal shelves at E15.5 (Fig. 5J) and the complete formation of a palatal plate at E16.5 (Fig. 5M). One of five histologically analyzed Ovca1−/−;CAG-Ovca1-myc embryos showed advanced palate elevation and a persistent epithelial seam (data not shown). However, we also found that another one of the five Ovca1-2−/−;CAG-Ovca1-myc embryos was not rescued and showed cleft palate phenotypes like that of Ovca1−/− embryos (Fig. 5I and L). Because the transgene was expressed in developing lower jaws, we examined the length of Meckel's cartilages in E14.5 embryos. Meckel's cartilages of Ovca1−/− lower jaws are significantly shorter than that of controls. Although the length of Meckel's cartilages of the Ovca1−/−;CAG-Ovca1-myc embryos was not completely restored compared with controls, the length was significantly longer than the Meckel's cartilages of the Ovca1−/− embryos (Fig. 5N and O). These data suggest that the expression of the Ovca1 transgene in mandibles can partially rescue cleft palate and shortened mandibles of Ovca1−/− embryos. Taken together, these results suggest that the palatal shelf elevation defect observed in Ovca1−/− mutants is secondary to spatial hindrance by the short mandibles and tongue.
Loss of Ovca1 in the neural crest cell lineage results in craniofacial defects with hypoplastic nasal bone and lower jaw
Most cell types in the palatal shelves and mandibles are originally derived from cranial neural crest cells and paraxial mesodermal cells (28–31). Previously, we have demonstrated that Ovca1 was expressed ubiquitously (32). In this study, we confirmed ubiquitous Ovca1 expression in E9.5 embryos and E14.5 head sections using RNA in situ hybridization (ISH). Using whole-mount RNA ISH, the intense signal could be detected by an Ovca1 antisense probe compared with a background signal detected by a sense probe on the E9.5 head structures including frontonasal prominence and mandibular arch (Supplementary Material, Fig. S2A and B). In the frontal sections of E14.5 embryonic heads, Ovca1 antisense RNA ISH revealed ubiquitous patterns on all cell types in palatal shelves, tongue and mandible (Supplementary Material, Fig. S2D–G) compared with that detected by a sense RNA probe (Supplementary Material, Fig. S2C). We next explored the cell-autonomous functions of Ovca1 in specific cell lineages that contribute to craniofacial structure. We generated an Ovca1 conditional null allele (referred to as Ovca1fx; Supplementary Material, Fig. S3A–C), in which exons 4–9 were flanked by loxP sites for subsequent Cre-mediated excision. We had conditionally ablated Ovca1 in the epiblast and subsequent cell lineages using the Sox2-Cre transgenic mouse line (33,34) as described in the Supplementary Material. Importantly, embryonic phenotypes of Sox2-Cre;Ovca1fx/fx were identical to that of Ovca1−/− (Supplementary Material, Fig. S4). Also, reverse transcriptase-polymerase chain reaction (RT-PCR) analysis revealed that the expression of Ovca1 was completely lost in the presence of Cre recombinase driven by Sox2 promoter (Supplementary Material, Fig. S3D).
Furthermore, we hypothesized that Ovca1 plays a role in the neural crest cells or paraxial mesodermal cell lineage to form proper craniofacial structures. To test this hypothesis, we utilized Mesp1-Cre mice (30,31,35) and Wnt1-Cre mice (30,31,36) to delete Ovca1 in paraxial mesodermal-derived head mesenchymal cells and neural crest cells, respectively. We examined Ovca1 expression in sorted paraxial mesodermal cells and neural crest cells by introducing Cre reporter, Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo (referred to as R26RmTmG) allele (37), to Mesp1-Cre and Wnt1-Cre mice, respectively. In the presence of Mesp1- or Wnt1-Cre, the GFP reporter could be sorted (Fig. 6A) for examining Ovca1 expression using RT-PCR. We found that Ovca1 was expressed in all sorted GFP-positive and -negative cell types from Mesp1-Cre;R26RmTmG/+ or Wnt1-Cre;R26RmTmG/+ embryonic heads (Fig. 6B), indicating that Ovca1 is expressed in both paraxial mesodermal cells and neural crest cells. We further generated conditional Ovca1 mutants and found that 14.5% of Mesp1-Cre;Ovca1fx/fx mice and 11.1% of Wnt1-Cre;Ovca1fx/fx mice were viable at weaning, which was lower than the expected ratio (25%) (Supplementary Material, Tables S2 and S3). At E17.5, the Mesp1-Cre;Ovca1fx/fx embryos showed no overt palatal phenotypes compared with their littermate controls (Fig. 6C and D). The Wnt1-Cre;Ovca1fx/fx embryos displayed a mucous cleft palate phenotype (Fig. 6F) with a separation between the palate and the nasal septum (Fig. 6G) when compared with control embryos (Fig. 6E). Skeletal analyses showed no overt difference in cartilages of the conditional Ovca1-depleted embryos compared with that of control embryos at E14.5 (Fig. 6H–K). However, the length of Meckel's cartilages of the Wnt1-Cre;Ovca1fx/fx embryos (Fig. 6K) was significantly shortened compared with the Mesp1-Cre;Ovca1fx/fx and control embryos (Fig. 6L and M).
At P0, the Mesp1-Cre;Ovca1fx/fx mice and the Wnt1-Cre;Ovca1fx/fx mice exhibited similar patterns of head skeleton compared with control mice (Fig. 7A–D). However, the skull length of Wnt1-Cre;Ovca1fx/fx mice was significantly shorter, albeit subtle (∼5% reduction), than that of control and Mesp1-Cre;Ovca1fx/fx mice (Supplementary Material, Fig. S5). We further specifically measured the areas of Mesp1-Cre-derived parietal bones and Wnt1-Cre-derived nasal and frontal bones traced by the GFP reporter expression in controls (Mesp1- or Wnt1-Cre;R26RmTmG/+;Ovca1fx/+) and Ovca1 conditional knockouts (Mesp1- and Wnt1-Cre;R26RmTmG/+;Ovca1fx/fx), respectively, at P0. In consistent with the skeletal patterns (Fig. 7A–D), the areas of Mesp1-Cre-derived parietal bones and Wnt1-Cre-derived frontal bones of the mutant skull vaults (Fig. 7F, H, J and L) were comparable to the controls (Fig. 7E, G, I and K). However, the areas of nasal bones were reduced by ∼20% in the Wnt1-Cre;R26RmTmG/+;Ovca1fx/fx mice (Fig. 7H, L and U) compared with that in the controls (Fig. 7G and K). Moreover, the ossification of palatine bones appeared normal in the control and Mesp1-Cre;Ovca1fx/fx mice (Fig. 7M and N). However, the ossified palatine bones with a gap, consistent with a cleft palate, were observed in the Wnt1-Cre;Ovca1fx/fx mice compared with that in the controls (Fig. 7P versus O). Also, the lengths of the mandibles were significantly shorter in the Wnt1-Cre;Ovca1fx/fx mice (Fig. 7T) compared with the mandibles in the Mesp1-Cre;Ovca1fx/fx mice and control mice (Fig. 7Q–S). The areas of frontal, parietal and nasal bone (Fig. 7U), the distances of two palatine bones at anterior and middle portions (Fig. 7V) and the length of mandible were measured and quantified (Fig. 7N). Significant impact on the size of nasal bone, palatine bone and mandible was specifically found in the Wnt1-Cre;Ovca1fx/fx mice, but not in the Mesp1-Cre;Ovca1fx/fx mice, compared with the control (Fig. 7U–W).
Nevertheless, some Wnt1-Cre;Ovca1fx/fx mice that survived to adulthood revealed craniofacial abnormalities with a round facial structure when compared with the Mesp1-Cre;Ovca1fx/fx or control mice (Fig. 8A and B). In addition, lateral views of the heads showed that the skull of the Wnt1-Cre;Ovca1fx/fx mutant was obviously shorter in length than the control (Fig. 8C). Micro-computed tomography (micro-CT) analyses of the skulls also confirmed the shorter skulls of the Wnt1-Cre;Ovca1fx/fx mice compared with the Mesp1-Cre;Ovca1fx/fx and control mice (Fig. 8D–G). Although the ossified palatine bones with a gap in the P0 Wnt1-Cre;Ovca1fx/fx mice were noted (Fig. 7H), the morphology of the palatine bones in survived Wnt1-Cre;Ovca1fx/fx mice was comparable to the Mesp1-Cre;Ovca1fx/fx and the control mice at 4 weeks of age (Fig. 8H–K). In contrast, the mandibles were still shorter in the Wnt1-Cre;Ovca1fx/fx mice at P0, in comparison with the Mesp1-Cre;Ovca1fx/fx and control mice (Fig. 8L–O). Therefore, loss of Ovca1 in the neural crest cell lineage resulted in shortening of the skull, likely due to shortening of the nasal bone (Fig. 7H, L and U) and mandible length (Fig. 8Q); whereas, the conditional deletion of Ovca1 in the cranial paraxial mesodermal cells displayed no overt phenotype in the skulls and mandibles (Fig. 8P). Taken together, these data indicate that Ovca1 is required for neural crest-derived craniofacial development and is dispensable in cranial paraxial mesodermal-derived head mesenchymal cells.
In vivo analysis of diphthamide deficiency in Ovca1-deficient Wnt1-Cre and Mesp1-Cre derivatives
A comparison of the phenotypes of the Wnt1-Cre;Ovca1fx/fx, Mesp1-Cre;Ovca1fx/fx and previously described Ovca1−/− mouse models enabled us to better delineate the role of this gene in craniofacial development. Loss of Ovca1 in the neural crest cell lineage, but not in the paraxial mesodermal-derived mesenchymal cells, is responsible for the craniofacial defects in vivo. Because OVCA1 is required for diphthamide biosynthesis (27,38), we therefore test whether the diphthamide modified function of OVCA1 is required in the neural crest cells in vivo. In an Ovca1+/+ background, the eEF-2 His715 residue is posttranslationally modified to form a diphthamide through an OVCA1-dependent diphthamide biosynthesis pathway (Fig. 9A). Loss of Ovca1 in cell, rendering the eEF-2 His715 unmodified, results in −1 ribosomal frameshifting during translation and DTA resistance. Since the DTA-mediated cell ablation is an effective analysis for evaluating diphthamide biosynthesis pathway in vivo, we therefore examined DTA resistance in either Wnt1-Cre or Mesp1-Cre derivatives under Ovca1+/− and Ovca1−/− conditions, respectively.
By introducing an ROSA26-loxP-EGFP/PGK-neo-loxP-DTA (referred to as R26-EGFP-DTA) allele in a wild-type background, EGFP is expressed in the absence of Cre recombinase, whereas DTA is only expressed upon Cre-mediated excision of a loxP-flanked EGFP/PGK-Neo cassette (Fig. 9A). Subsequently, ADP ribosylation by DTA on the diphthamide of eEF-2 blocks translation and leads to cell death. In contrast, the DTA cannot ADP-ribosylate eEF-2 His715 due to loss of OVCA1-dependent diphthamide modification in an Ovca1−/− background (Fig. 9A), resulting in the survival of EGFP-negative DTA-expressing cells after Cre-mediated recombination. To examine this hypothesis in vivo, Wnt1-Cre or Mesp1-Cre transgenic mice were utilized to excise the EGFP/PGK-neo cassette and induce the expression of DTA in both Ovca1+/+ (Fig. 9A, top) and Ovca1−/− (Fig. 9A, bottom) backgrounds. In the absence of the Cre transgene, R26-EGFP-DTA;Ovca1+/− and R26-EGFP-DTA;Ovca1−/− embryos expressed the EGFP ubiquitously and displayed normal appearances except for the growth restriction phenotype of the Ovca1−/− embryo at E10.5 (Fig. 9B–E). When Cre recombinase was expressed to excise the EGFP cassette in the neural crest lineage, the DTA-mediated cell ablation resulted in the loss of head structures in the Wnt1-Cre;R26-EGFP-DTA;Ovca1+/− embryos (Fig. 9F and G). The remaining portion of the Wnt1-Cre;R26-EGFP-DTA;Ovca1+/− embryo still expressed EGFP (Fig. 9F). Strikingly, the Wnt1-Cre;R26-EGFP-DTA;Ovca1−/− embryos retained their head structures (Fig. 9I). Moreover, compared with the R26-EGFP-DTA;Ovca1−/− embryos, the remaining head structures of the Wnt1-Cre;R26-EGFP-DTA;Ovca1−/− embryos did not express the EGFP that was due to the excision of the EGFP/PGK-neo cassette, indicative of Wnt1-Cre-mediated DTA expression and resistance in Ovca1-null mutants (Fig. 9H and I).
Similarly, Mesp1-Cre transgenic mice were introduced to excise the EGFP/PGK-neo cassette from R26-EGFP-DTA allele and induce the expression of DTA in Ovca1+/+, Ovca1+/− and Ovca1−/− embryos. In this study, we were unable to obtain Mesp1-Cre;R26-EGFP-DTA;Ovca1+/+ and Mesp1-Cre;R26-EGFP-DTA;Ovca1+/− embryos at E10.5 (Supplementary Material, Table S4). The presence of Mesp1-Cre activity in the heart promodium (31) for inducing the DTA expression from five-somite stage might explain the embryonic lethality in Mesp1-Cre;R26-EGFP-DTA;Ovca1+/+ and Mesp1-Cre;R26-EGFP-DTA;Ovca1+/− embryos. Interestingly, Mesp1-Cre;R26-EGFP-DTA;Ovca1−/− embryos could be obtained (Fig. 9J and K) because the DTA-expressing cells (EGFP-negative, ventral side of the neural tube and the developing heart; Fig. 9J and L–O) could survive in the Ovca1−/− background.
Taken together, these data suggest that Ovca1, a newly identified MDS craniofacial gene, is important for neural crest-derived mandible development through OVCA1-dependent diphthamide posttranslational modification of eEF-2. Although OVCA1-dependent diphthamide biosynthesis is also present in the Mesp1-Cre-derived tissues, OVCA1 appears to be dispensable for the paraxial mesodermal-derived mandible development.
In the present study, we show that Ovca1 plays an important role in neural crest cell-derived nasal bone and mandible development. This finding implies that inactivation of OVCA1 in human may cause craniofacial dysmorphisms such as small upturned nose and small jaw phenotypes, which are often described in MDS patients. Our results also show that conditionally expressed DTA causes DTA-mediated cell cytotoxicity in control neural crest cells and paraxial mesodermal cells, but not in an Ovca1-null background. Our findings suggest that OVCA1-dependent diphthamide biosynthesis of eEF-2 is crucial for neural crest cell-derived craniofacial structures, consistent with a recent report (27) showing that diphthamide deficiency is the cause of embryonic lethality in Ovca1-null mice. Liu et al. (27) also demonstrated reduced protein elongation and increased −1 ribosomal frameshifts in Ovca1-null MEFs. Thus, OVCA1 loss causes diphthamide deficiency and aberrant ribosomal frameshifts, which may generate incorrect or truncated proteins that consequently affect craniofacial organogenesis. However, our study demonstrated that OVCA1-dependent diphthamide biosynthesis appears to be differentially required in different tissues. In this study, our genetic data show that OVCA1-dependent diphthamide biosynthesis is present in paraxial mesoderm but it seems to be dispensable for paraxial mesodermal-derived craniofacial structures. Additional studies to address aberrant truncated proteins or frameshift translation products in Ovca1-null tissues will be needed for further exploration of the molecular alterations that are involved in craniofacial development.
Forward genetic screens in yeast and Chinese Hamster Ovary (CHO) cells identified mutations in Ovca1 that conferred the resistance to ADP-ribosylating toxins (18–20) without overt growth disadvantage. In contrast, the Ovca1-null MEFs and Ovca1−/− embryos exhibit severe growth disadvantages resulting in embryonic developmental delay and lethality (12). Similarly, inactivation in other components, Dph3 and Dph4, of the first step of diphthamide biosynthesis in mice also leads to embryonic lethality (39,40). Notably, Dph4-null embryos exhibit preaxial polydactyly in the hindlimb similar to Ovca1-null embryos (12,40), indicating that diphthamide deficiency also results in aberrant digit patterning. Polydactyly is also a clinical manifestation of MDS (4). Nonetheless, diphthamide deficiency seems to exhibit severe developmental defects in embryonic organogenesis and patterning, which are dependent upon cellular morphogenesis through tissue–tissue interaction and coordination, unlike single cell propagation, without affecting growth in diphthamide deficient yeast or CHO cells.
In this study, cell type-specific ablation of Ovca1 exhibits milder craniofacial abnormalities in the cranial neural crest cell lineage or no overt phenotype in the cranial paraxial mesodermal lineage compared with severe craniofacial abnormalities in conventional Ovca1-null embryos. Perhaps, differential cell lineage contributions of Wnt1- versus Mesp1-Cre derivatives in the craniofacial structures result in the distinct phenotypes observed in Wnt1-Cre;Ovca1fx/fx versus Mesp1-Cre;Ovca1fx/fx mice. Previously, Yoshida et al. (31) reported that Wnt1-Cre-derived cells contributed to almost entire frontonasal prominence and mandibular arch, whereas Mesp1-Cre-derived cells contributed to adjacent mesenchyme of cranial neural tube and gave rise to endothelial and myogenic lineages, which intermingled within neural crest cells derivatives in mandibular arch. In the developing palate, the major mesenchymal cells arise from Wnt1-Cre-derived neural crest cells, although a few of mesenchymal cells are derived from non-Wnt1-Cre lineage (41). In our R26RmTmG reporter tracing experiments, we observed that Mesp1-Cre lineage only gave rise to a few of cells with a scattering pattern, which was similar to previously identified non-Wnt1-Cre derivatives (41), and separated from the majority of neural crest cells derived mesenchyme in the developing palate (unpublished observations). In the skull vault, the cell lineage contributions of Wnt1-Cre neural crest cells and Mesp1-Cre mesodermal cells are mutually exclusive (31). The neural crest cells are contributed to the frontal bone and the medial portion of interparietal bone, whereas the mesodermal-derived cells are contributed to the parietal bone and peripheral portion of interparietal bone (31). Our data suggest that Ovca1 is dispensable for the development of frontal and parietal bones. In contrast, Ovca1-null neural crest cells impact on the nasal and mandibular bones. Possibly, compensation of Ovca1-null cranial paraxial mesodermal cells by wild-type neural crest or other cells may occur and result in normal phenotypes in Mesp1-Cre;Ovca1fx/fx mice. Also, Ovca1-null neural crest cells may be partially compensated by other wild-type tissues, resulting in milder phenotypes in craniofacial dysmorphism and lethality. Moreover, the differential requirements of Ovca1 in different craniofacial cell lineages may partly imply that Ovca1 is involved in multiple tissues for proper tissue–tissue interactions or for coordination during craniofacial development.
Our current assessment of DTA-mediated cell cytotoxicity is based on conditionally expressed DTA in an Ovca1-null background, in which the diphthamide deficiency of eEF-2 occurs prior to DTA expression. We also analyzed DTA-mediated cell cytotoxicity, when conditionally expressed DTA and Ovca1 deletion occurred simultaneously in Wnt1-Cre;R26-EGFP-DTA;Ovca1fx/fx embryos. However, in these embryos, the head structures were eliminated and failed to resist DTA-mediated toxicity (unpublished observations), indicating that DTA-mediated ADP ribosylation is a rapid process that is sufficient to trigger cytotoxicity if diphthamide preexists on eEF-2. Thus, the current approach, using single Cre-loxP-mediated tissue-specific ablation of Ovca1 and expression of DTA, has technical limitations for understanding the cell-autonomous role of OVCA1 in diphthamide biosynthesis in vivo. To our knowledge, our findings provide the first evidence of DTA resistance in an Ovca1-null background in neural crest cells during craniofacial development.
MDS is known to be a contiguous gene deletion syndrome. However, the aberrant craniofacial development is only observed in the mice carrying homozygous Hic1, Mnt and Ovca1-null alleles and mice with conditional ablation of Ovca1 in the neural crest cell lineage. Based on the craniofacial abnormalities of null mutants for these genes, our findings add OVCA1 with MNT and HIC1 as that are the MDS craniofacial genes. The homozygous mutants of these genes commonly exhibit cleft palate, although cleft palate is not commonly reported in MDS patients. Perhaps, cleft palate is a common phenotype when MDS craniofacial genes are ablated in mice. Nevertheless, the causal mechanisms of cleft palate may be varied in different mouse mutants. In the present study, we showed that Wnt1-Cre-mediated Ovca1 ablation is crucial for neural crest-derived mandible development, whose defects can affect palate elevation and fusion. In Mnt−/− mice, cleft palate is also considered to be a secondary phenotype due to the small lower jaw (14). Conditionally ablated Mnt in the ectodermal-derived epithelial components of the oral cavity and the first branchial arch in Pitx1-Cre+/MntCKO/CKO mice show normal palate fusion without significant craniofacial defects (14). This is an indirect evidence hints the important function of Mnt in mesenchymal components of the oral structures and the mandibular/maxillary processes during palate fusion. Meanwhile, no tissue-specific Hic1 knockout mouse model has been reported but conditional Hic1 allele is currently available (42) that can be used for further deciphering craniofacial phenotypes of Hic1 mutants. While acrania was observed in Hic1−/− embryos, smaller skulls were observed in Mnt−/−, Ovca1−/− embryos and Wnt1-Cre;Ovca1fx/fx mice, suggesting that these genes contribute to skull development in different degree. Nonetheless, heterozygous Mnt+/−, Hic1+/− and Ovca1+/− mice do not show overt mutant phenotypes. Apparently, the remaining wild-type alleles of Mnt, Hic1 and Ovca1 in the corresponding heterozygous mice are sufficient to support embryonic growth and craniofacial morphogenesis. It remains unclear whether MNT, HIC1 and OVCA1 residing within the MDS deletion region have reduced or absent expression in MDS patients. However, direct examination of MNT, HIC1 and OVCA1 gene products in the craniofacial tissues of rare MDS individuals is not feasible. It is also unclear whether combined haploinsufficiency of MNT, HIC1, OVCA1 and other genes can cooperatively involve in craniofacial development. Thus, current genetically modified mouse models at least partly provide the possible molecular explanations for the clinical manifestations of MDS. Efforts to engineer contiguous gene deletions within mouse chromosome 11B5 may benefit our future understanding of the causal mechanisms involved in the clinical manifestations of MDS patients.
In summary, this study identifies the neural crest cell lineage as the target tissue for the action of OVCA1 in craniofacial development. The diphthamide deficiency in cranial neural crest cells is likely to be the main causal mechanism aberrantly affecting craniofacial development that is correlated with manifestations of MDS.
MATERIALS AND METHODS
Ovca1-2+/− mice were generated and maintained on a mixed C57BL/6J;129/SvEv background as described previously (12). Because the phenotypes of Ovca1-2−/− embryos were identical to Ovca1-specific knockouts in our previous study (12), in this study the Ovca1-2 mutant mice were therefore referred to as Ovca1. Wnt1-Cre (43), Sox2-Cre (33,34), R26RmTmG (37) and R26-EGFP-DTA (44) mice were obtained from the Jackson Laboratories (Bar Harbor, ME, USA). Mesp1-Cre mice (35) were obtained from RIKEN BioResource Center (Ibaraki, Japan). To generate CAG-βgeo-Ovca1-myc transgenic mice, the PmeI-digested 1.4 kb Ovca1-myc DNA fragment from pcDNA4/TO/Ovca1-myc-His (32) was inserted into an XhoI-digested Klenow enzyme-fill-in pCCALL1 vector (obtained from Dr. Richard R. Behringer, MD Anderson Cancer Center, Houston, TX, USA). The total size of the ∼11 kb DNA fragment (Fig. 5A) was excised by SfiI/XmnI from the vector backbone and was utilized for pronuclear microinjection (45). The CAG-βgeo-Ovca1-myc transgenic mice were maintained on a C57BL/6J background. To generate the Ovca1 floxed allele, we used a recombination-based method (46). The essential materials were generously provided by Dr. Neal Copeland (Methodist Hospital Research Institute, Houston, TX, USA). Briefly, a 13 kb DNA fragment, containing the region from exons 1–12 of the Ovca1 gene, was retrieved from a 129 strain-derived bacterial artificial chromosome (BAC) clone (bMQ-275d19) into pL253 with A1 and A6 homologous arms amplified by the primers (A1: 9072-F:GCT CAC TCC CAG TCT GCC TTC CTC HindIII-9469-R:AAG CTT CCC TAA ACC TCC TGG GTC T and A6: 21707-F:CTG CGA GTG CTG TGT CTA GCC GG BamHI-22074-R:GGA TCC ACA GGC CCA GGG TTA GAG). The resulting construct was used as a backbone for subsequent insertion of the first loxP sequence from pL452 with A2 and A3 homologous arms amplified by the primers (A2: SalI-15385-F:GTC GAC TGG GCT GCA TTC TAA GGA G, ScaI-15652-R:AGT ACT GAA TCA CTG GGG GAC AGG G and A3: BamHI-15653-F:GGA TCC TGG CAG ATC TCA AAA, 15908-R:GCG TAC CCA GGG CAC GTA AAG AT) into intron 3. The frt-PGK-neo-frt-loxP cassette from pL451 with A4 and A5 homologous arms amplified by the primers (A4: SalI-20270-F:GTC GAC TCT CTT TTC CCC TGA AGC A, 20476-R:AAA GCA AAA GTG GAG CTG GAC AC and A5: BamHI-20477-F:GGA TCC TAA TGC AAT GGT TTT GGG AG, 20761-R:TAC CTC GTA CGG TGT CAG TGG) was placed into intron 9. All the homologous recombination events were conducted in DY380 competent cells. The following ES cell targeting was performed by homologous recombination in R1 ES cells (129X1/SvJ × 129S1/Sv) as described previously (45) and carried out by the Transgenic Mouse Models Core Facility of National Core Facility Program for Biotechnology (National Taiwan University Hospital, Taipei, Taiwan). Correctly targeted ES cell clones were verified by Southern blotting with a 5′ probe amplified by the primers (8202-F:CTG ACT TGC TCA GCA GCA TC, 8629-R:TGC AAA ACC ATC GTG ACT GT) and injected into blastocysts of C57BL/6JNarl origin to establish chimeric mice that were bred with C57BL/6J females. Germline transmission of the floxed Ovca1-neo allele (Ovca1fx-neo) was obtained from the agouti progeny. The Ovca1fx-neo/+ mice were then bred with the CAG-FLPe (Tg(CAG-FLPe)36) transgenic mouse line (47) (obtained from RIKEN BioResource Center) to remove the frt-PGK-neo-frt cassette and generate the Ovca1fx/+ allele. Genotypes could be determined by Southern blots with a 5′ probe, and polymerase chain reaction (PCR) analyzes with P1 (Ov20269-F: 5′ TCT CTT TTC CCC TGA AGC AAT C 3′) and P2 (5′ Ov20761-R: TAC CTC GTA CGG TGT CAG C 3′) primers (Supplementary Material, Fig. S3B and C). The Ovca1fx/fx mice were normal, fertile and maintained on a mixed C57BL/6J;129 background. Excision of the Ovca1 floxed allele by the Sox2-Cre transgene was characterized (Supplementary Material). All experiments with mice were performed with the approval of the Institutional Animal Care and Use Committee of National Yang-Ming University.
Genotyping of mice
Genomic DNA was extracted from 2-week-old mouse toes and genotyped using PCR methodology. The Ovca1 allele was detected by the primers Ov10H3-1T3out (5′-CAA CTT CAT AGA GAT TCC CTT GC-3′), Ovca1E13-R (5′-CAC TGT GGA CTC TTC CAG AGC-3′) and Ovca2intron-F (5′-CTG GCT GCA CTT TCC CAA GC-3′), yielding a 200 bp product corresponding to the Ovca1 wild-type allele and a 300 bp product corresponding to the Ovca1-null allele. The Ovca1 transgenic alleles of CAG-βgeo-Ovca1-myc and CAG-Ovca1-myc were detected by the primers pCCALL1-F (5′-GCC TCT GCT AAC CAT GTT CAT GC-3′), βgeo-R (5′-ATT CAG GCT GCG CAA CTG TTG GG-3′) and Ovca1-552 (5′-GGT TGA CAC AAA CTG AAT GGT G-3′), yielding a 450 bp product corresponding to the CAG-βgeo-Ovca1-myc allele and a 720 bp product corresponding to the CAG-Ovca1-myc allele. The Ovca1 floxed allele was detected by the P1 and P2 primers (see above; Ov20269-F and Ov20761-R), yielding a 500 bp product corresponding to the Ovca1 wild-type allele and a 600 bp product corresponding to the Ovca1fx allele. The Cre allele of all the Cre line mice was detected by the primers Cre-1 (5′-GGA CAT GTT CAG GGA TCG CCA GGC G-3′) and Cre-β (5′-CGA CGA TGA AGC ATG TTT AGC TG-3′), yielding a 220 bp product. The R26RmTmG allele was detected by the primers oIMR7318 (5′-CTC TGC TGC CTC CTG GCT TCT-3′), oIMR7319 (5′-CGA GGC GGA TCA CAA GCA ATA-3′) and oIMR7320 (5′-TCA ATG GGC GGG GGT CGT T-3′), yielding a 330 bp product corresponding to the wild-type allele and a 250 bp product corresponding to the R26RmTmG allele. The R26-EGFP-DTA allele was detected by the primers DTA-F (5′-AAA GTC GCT CTG AGT TGT TAT-3′), DTA-R1 (5′-GGA GCG GGA GAA ATG GAT ATG-3′) and DTA-R2 (5′-GCG AAG AGT TTG TCC TCA ACC-3′), yielding a 650 bp product corresponding to the wild-type allele and a 340 bp product corresponding to the R26-EGFP-DTA allele.
For the examination of embryonic cartilaginous skeletons, E14.5–E16.5 embryos were collected and fixed in Bouin's solution (Sigma-Aldrich, St Louis, MO, USA) for 2 h at room temperature, followed by washing for 24 h with 0.1% NH4OH/70% ethanol (EtOH). Then, samples were equilibrated in 5% acetic acid and stained in 0.05% Alcian blue (Sigma-Aldrich) in 5% acetic acid/70% EtOH for 3 h (E14.5), 3.5 h (E15.5) or 4 h (E16.5). Finally, samples were washed with 5% acetic acid, dehydrated in 100% methanol and cleared in 1 : 2 benzyl alcohol–benzyl benzoate (Sigma-Aldrich). For staining and visualization of whole skull morphology, heads of P0 fetuses were collected and stained with Alizarin red S and Alcian blue 8G (Sigma-Aldrich), as previously described (48). For the skeleton preparation of adult mice, heads of 4-week-old mice were collected and skin and muscles of heads were removed, followed by fixation with 95% EtOH for 5 days. Samples were then transferred to 2% KOH solutions until all tissues were removed, except for bones. Finally, bones were bleached with 3% H2O2 for 1 h, photographed and stored in distilled water.
Embryos were fixed in 4% paraformaldehyde (PFA)/1× phosphate-buffered saline (PBS) at 4°C overnight, washed with 1× PBS and dehydrated in a graded series of EtOH. The samples were then cleared in xylene and infiltrated with paraffin. The procedures of embedding and sectioning were described previously (49,50). Paraffin sections were deparaffinized, and then stained with hematoxylin and eosin (H&E) (Sigma-Aldrich), as described previously (49,50).
Whole-mount β-galactosidase staining
Immunofluorescence staining of β-galactosidase was performed on 10 μm cryosections of E13.5 CAG-βgeo-Ovca1-myc embryos with primary antibody against β-galactosidase (β-gal) (rabbit IgG, 1:5000; MP Biomedicals, Solon, OH, USA), at 4°C overnight. The sections were further incubated with an Alexa Fluor 488-conjugated secondary antibody (1:200; Invitrogen, Carlsbad, CA, USA) at room temperature for 1 h, followed by counterstaining with 0.5 ng/ml DAPI (4',6-diamidino-2-phenylindole) for 3 min and mounting in fluorescence mounting medium (DakoCytomation, Carpeinteria, CA, USA). Detailed procedures were described previously (49,52–54).
Head ex vivo culture
E13.5 embryonic heads, without mandibles and tongues, were dissected and collected in sterile 1× PBS. The samples were cultured in 20% fetal bovine serum/Dulbecco's modified Eagle's medium at 37°C in a 5% CO2 incubator for 1 day, and then processed for histological analyses.
FragEL™ DNA Fragmentation Detection Kit (Calbiochem, Darmstadt, Germany) was used for detecting apoptotic cells on 5 μm sections of ex vivo cultured Ovca1+/− and Ovca1−/− heads. Nuclei were stained with DAPI.
E9.5 whole embryos and E14.5 embryonic heads were fixed in 4% PFA in PBS at 4°C overnight. E9.5 embryos were dehydrated through a graded series of methanol in PBS with 0.1% Tween-20 and stored at −20°C for whole-mount RNA ISH as described previously (52,55). PFA-fixed E14.5 embryonic heads were then prepared for paraffin-embedded tissue sections. The detailed procedures of RNA ISH on tissue sections were described previously (52). The portion of Ovca1 cDNA spanning from exons 7–9 was amplified by the primers OvExon7-F (5′-GCT TTC ATC TGG AGT CTG TC-3′) and OvExon9-R (5′-TCC ACC TCA GGA AGT AGA C-3′) and subcloned into pGEM-T Easy Vector. The digoxigenin-labeled sense and antisense riboprobes were generated from BglII- and SacI-digested pGEM-T/Ovca1(Exon7–9) plasmid and transcribed in vitro by SP6 and T7 RNA polymerases, respectively.
Cell isolation and fluorescence-activated cell sorting
Mesp1-Cre;R26RmTmG/+ and Wnt1-Cre;R26RmTmG/+ embryonic heads at E11.5 and E13.5 were dissected and further digested by 0.05% Trypsin-EDTA (Gibco, Grand Island, NY, USA) and DNaseI (1:1000; Sigma-Aldrich) at 37°C for 10 min. Dissociated cells were washed and suspended in 1× PBS followed by passing through cell strainers with 70 and 40 µm pore size (BD Falcon, San Jose, CA, USA) consecutively. GFP+ and GFP− cells were analyzed and sorted by a BD FACSAria sorter.
Reverse transcription-polymerase chain reaction
Total RNA was extracted from E10.5 embryos and fluorescence-activated cell sorting (FACS)-sorted cells by TRIzol reagent (Invitrogen). The cDNA templates were generated from extracted total RNAs (1 μg RNA from each E10.5 embryo and 0.5 μg RNA from each set of sorted cells), which were reverse transcribed using the SuperScript III First-Strand Synthesis System (Invitrogen). The primers, OvExon7-F and OvExon9-R, were used for RT-PCR to amplify a 300 bp product of Ovca1. The primers, Hprt-F (5′-TCC TCC TCA GAC CGC TTT T-3′) and Hprt-R (5′-CCT GGT TCA TCA TCG CTA ATC-3′), were used for RT-PCR to amplify a 100 bp product of hypoxanthine phosphoribosyltransferase (Hprt) as an internal control.
Skeletal samples were fixed and stored in 70% EtOH, and then scanned with a micro-CT scanner (Skyscan-1076, Skyscan/Bruker, Kontich, Belgium) at 50 kV, 150 µA, 1° of rotation step, 0.5 mm Al filter and 35 µm/pixel of scan resolution. All three-dimensional analyses and images were reconstructed by Skyscan software conducted by the Taiwan Mouse Clinic, National Phenotyping and Drug Testing Center (Academia Sinica, Taipei, Taiwan).
All statistical data were analyzed using the paired Student's t-test for statistical comparisons. Data were expressed as mean ± SEM. The two-tailed unpaired Student's t-test was used to determine statistical significance. P-values of <0.05 were considered statistically significant.
This work was supported by a grant from the National Health Research Institutes (grant number NHRI-EX103-10240BI), a grant from the Ministry of Education ‘Aim for the Top University Plan’ and grants from the Ministry of Science and Technology to C.-M.C. (grant numbers NSC 102-2325-B-010-015 and MOST 103-2325-B-010-002) and to the Taiwan Mouse Clinic (grant numbers NSC 102-2325-B-001-042 and MOST 103-2325-B-001-015).
We thank Drs Richard R. Behringer and Ming-Ji Fann for comments on this manuscript. We also thank technical support from the Taiwan Mouse Clinic and Transgenic Mouse Models Core Facility of National Core Facility Program for Biotechnology supported by the Ministry of Science and Technology. We also thank Wann-Chun Yu for technical help to generate the Ovca1fx targeting vector.
Conflict of Interest statement. None declared.