Human Molecular Genetics Advance Access originally published online on October 17, 2006
Human Molecular Genetics 2006 15(23):3394-3410; doi:10.1093/hmg/ddl416
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Cyp26 genes a1, b1 and c1 are down-regulated in Tbx1 null mice and inhibition of Cyp26 enzyme function produces a phenocopy of DiGeorge Syndrome in the chick
1 Molecular Medicine Unit and 2 Cardiac Unit, Institute of Child Health, 30 Guilford Street, London WC1N 1EH, UK and 3 Institute of Biosciences and Technology, Texas A&M University Health Sciences Center, 2121 W. Holcombe Blvd., Houston, TX 77030, USA
* To whom correspondence should be addressed. Tel: 44 207 905 2635; Fax: 44 207 905 2609; Email: p.scambler{at}ich.ucl.ac.uk
Received June 8, 2006; Accepted October 12, 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Cyp26a1, a gene required for retinoic acid (RA) inactivation during embryogenesis, was previously identified as a potential Tbx1 target from a microarray screen comparing wild-type and null Tbx1 mouse embryo pharyngeal arches (pa) at E9.5. Using real-time PCR and in situ hybridization analysis of Cyp26a1 and its two functionally related family members Cyp26b1 and c1, we demonstrate reduced and/or altered expression for all three genes in pharyngeal tissues of Tbx1 null embryos. Blockade of Cyp26 function in the chick embryo using R115866, a specific inhibitor of Cyp26 enzyme function, resulted in a dose-dependent phenocopy of the Tbx1 null mouse including loss of caudal pa and pharyngeal arch arteries (paa), small otic vesicles, loss of head mesenchyme and, at later stages, DiGeorge Syndrome-like heart defects, including common arterial trunk and perimembranous ventricular septal defects. Molecular markers revealed a serious disruption of pharyngeal pouch endoderm (ppe) morphogenesis and reduced staining for smooth muscle cells in paa. Expression of the RA synthesizing enzyme Raldh2 was also up-regulated and altered Hoxb1 expression indicated that RA levels are raised in R115866-treated embryos as reported for Tbx1 null mice. Down-regulation of Tbx1 itself was observed, in accordance with previous observations that RA represses Tbx1 expression. Thus, by specifically blocking the action of the Cyp26 enzymes we can recapitulate many elements of the Tbx1 mutant mouse, supporting the hypothesis that the dysregulation of RA-controlled morphogenesis contributes to the Tbx1 loss of function phenotype.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
TBX1, a T-box transcription factor, currently represents the best candidate gene for DiGeorge/Velo-Cardio-Facial syndrome (DGS/VCFS), a syndrome associated with large hemizygous deletions of chromosome 22q11. The human phenotype is variable and includes abnormalities of the aortic arch and outflow tract, such as common arterial trunk (CAT), interruption of the aortic arch between the left common carotid and left subclavian arteries (type B, IAA-B) and perimembranous ventricular septal defects (PM-VSD), thymic and parathyroid aplasia/hypoplasia, craniofacial defects and learning and behavioural anomalies.
Df1/+, a mouse model for the large hemizygous deletion (22 genes) of the region of mouse chromosome 16 syntenic to the most commonly observed DGS-associated deletion of 22q11, has been shown to produce a fully penetrant aplasia or hypoplasia of the fourth aortic arch artery. This early phenotype may subsequently recover before birth, with 25% of animals having characteristic DGS/VCFS-like aortic arch defects such as IAA-B at E18.5. Rescue of the Df1/+ phenotype was accomplished by genetic complementation with PACs or human BACs containing Tbx1 (1–3). Conventional Tbx1 null mutations in the heterozygous state share the Df1/+ phenotype and Tbx1 homozygous null animals have a poorly segmented pharynx, a hypo/aplastic pharyngeal arch 2 (pa 2) and complete loss of more caudal arches, which later produces aplasia of the thymus and parathyroid glands. There are a variety of craniofacial anomalies including a reduction in otic vesicle size and later ear malformations, both externally and internally, cleft palate and abnormal head mesenchyme development causing skull defects. All null embryos have a failure of septation of the outflow tract resulting in common arterial trunk by E12.5 (2–4). While neural crest migration is disrupted in Tbx1–/– embryos, this is believed to be secondary to the absence of the pharyngeal pouches and the signals normally emanating from these structures (5,6). Our recent data have suggested that disruption of Tbx1 function affects the expression of several genes with a role in embryonic RA metabolism (7). One of these genes, Cyp26a1, was shown to be down-regulated in Tbx1 nulls both Affymetrix and Codelink arrays set and by real-time quantitative PCR (RTQ-PCR) at E9.5.
Retinoic acid (RA) distribution is carefully controlled during embryogenesis by the combined action of synthesizing enzymes (particularly the retinaldehyde dehydrogenases Aldh1a1, Aldh1a2 and Aldh1a3 otherwise known as Raldh1, 2 and 3, respectively) (8–10), and catabolic enzymes of the Cyp26 family, which are cytochrome P450s that convert RA to more polar metabolites such as 4-hydroxy, 4-oxo and 5,8-epoxy all-trans RA (11–13). Both these classes of enzyme are expressed in a dynamic and spatially restricted manner during embryogenesis, such that they are often expressed in a complementary, but rarely overlapping fashion (11,14–24).
This exquisite and complex control of RA is necessary as it has long been known as a signalling molecule required for the normal development of a large number of embryonic tissues including the pharyngeal region and the heart (25–28), reviewed by Mark et al. (29) and Zile (30), the nervous system (31–33), lung (34), limb (35,36), kidney (37,38) and eye (39). Studies in both mammalian and non-mammalian models have shown that disruption of RA homeostasis via maternal diet or genetic/chemical modification (thus increasing or decreasing RA relative to normal endogenous levels) can result in a phenotype with strong similarities to DGS/VCFS (40–46). In addition, the pharyngeal and cardiovascular systems (in particular the outflow tract) have been shown to have an especially stringent requirement for proper regulation of RA. Work in the Raldh2 homozygous null mouse has shown that Raldh2 provides RA to the majority of embryonic tissues, as these embryos die at mid-gestation with a severe phenotype encompassing failure of axial rotation, a shortened anteroposterior axis and frontonasal process, abnormal somitogenesis, small otocysts, lack of limb buds and a single medial dilated heart cavity. Maternal RA administration was able to rescue much of this phenotype such that embryos survived until E13.5–14.5. However, in these ‘rescued’ embryos those structures affected in DGS and the Tbx1 null mouse, namely the pa, pharyngeal pouches and pharyngeal arch arteries (paa), had impaired development resulting in CAT and an aplastic/hypoplastic thymus and parathyroids (25–27).
All the three embryonic Cyp26 genes are expressed in regions of the embryo known to require careful RA regulation for normal development, notably in the neural plate and hindbrain, the tailbud, the heart and the pharyngeal tissues, including head and pa mesenchyme, pharyngeal pouch endoderm (ppe) and neural crest-derived mesenchyme. There is some variation in the expression of each specific gene between vertebrate species including between chick and mouse embryos. However, the combined domain of expression of all three genes is overall extremely similar between the two species (11,15–17,21,23,24,47).
Cyp26a1 null mouse embryos have a range of phenotypes at mid-gestation including: spina bifida, truncation of the tail and abnormalities affecting the kidneys, urogenital tract and hindgut; posterior transformation of the cervical ganglia; partial transformation and abnormal patterning of the hindbrain. Developmental arrest at E8.5 was observed in some embryos due to abnormal looping/heart oedema indicating cardiovascular anomalies (48). However, these mice do not have a DGS-like phenotype. Cypc1/a1 double null alleles have been described as having a severe neuroepithelial/craniofacial phenotype (49) and Cyp26b1 null embryos have micrognathia and a limb phenotype including meromelia and oligodactyly (50).
Cyp26 function can also be blocked using compounds originally designed with a view to enhancing RA treatment of skin cancers. R115866 ((B)-N-[2-ethyl-1-(1H-1,2,4-triazol-1-yl)butyl]phenyl]-2-benzothiazolamine) is a nanomolar inhibitor of Cyp26-dependent RA conversion (IC50=4 nM). While no compound is totally specific the selectivity for this Cyp is evidenced by trivial inhibition of Cyp-dependent synthesis of estradiol and testosterone (micromolar concentrations of R115866 are required to inhibit Cyp19, Cyp17, Cyp2c11, Cyp3a and Cyp2a1). In vivo, oral administration of R115866 to adult rats resulted in raised levels of RA in plasma, skin, fat, kidney and testis. These effects can all be reversed by administration of RA receptor antagonists suggesting that R115866 inhibition of Cyp26 results in an increased availability of endogenous RA in and around tissues expressing Cyp26s (51).
In this paper, we present RTQ-PCR and in situ hybridization data to show that in Tbx1–/– embryos, expression of all three embryonic Cyp26 genes is altered, most expression domains being either lost or down-regulated while others experience an ectopic shift or expansion of expression in pharyngeal and head tissues. If the reduced Cyp26 gene family expression we observed in Tbx1–/– embryos contributes to the phenotype, then a combined knockdown of all three genes should give DGS/VCFS-like abnormalities, or interact epistatically with Tbx1 loss of function, but this would be complicated to achieve using mouse genetics. Instead, we blocked Cyp26 function in chick embryos using R115866. The phenotype of treated chick embryos resembled Tbx1 null mice in several important respects including loss of caudal pa, abnormal ppe morphogenesis and heart defects characteristic of DGS.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Embryonic Cyp26 genes, Cyp26a1, b1 and c1 are down-regulated in the Tbx1–/– mouse embryos
We have recently shown the down-regulated expression of Cyp26a1 in Tbx1 mutant embryos by both microarray and RTQ-PCR, thus identifying this RA-degrading enzyme as a potential target of Tbx1 and raising the possibility that local RA toxicity may contribute to the Tbx1 null phenotype (7). Two further Cyp26 family members Cyp26b1 and Cyp26c1 are also strongly expressed in Tbx1-expressing pharyngeal tissues during embryogenesis and play a role in RA metabolism in this region; we therefore, examined their expression in Tbx1 mutants. As these genes were not represented on the chips used in the original microarray experiments, the initial analysis was performed using RTQ-PCR. The results showed that both of these genes were down-regulated in Df1/+;Tbx1/+ E9.5 mouse embryos, with Cyp26b1 reduced 1.6-fold and Cyp26c1 2.6-fold when compared with expression in wild-type embryos (Table 1).
|
Next, we carried out in situ hybridization on Tbx1–/– embryos in order to assess the changes in the expression of Cyp26a1, Cyp26b1 and Cyp26c1 in more detail. In the mouse, Cyp26a1 is expressed from early gastrulation stages, first in the extra-embryonic and embryonic endoderm and then a little later in embryonic posterior mesoderm and the primitive streak. Between E9.5 and 10.5, there is strong expression in cervical and pa mesenchyme, maxillo-mandibular cleft and tail-bud (11,23,47). In Tbx1–/– embryos at E10.5 expression of Cyp26a1 in the pharyngeal region was found to be considerably decreased when compared with stage-matched wild-type embryos, confirming the microarray and PCR results (Fig. 1A). Transverse sections showed that the areas of reduced expression included neural crest-derived cranial ganglia (Fig. 1B and C). Cyp26a1 was unaffected in the tail-bud region of Tbx1–/– embryos (not shown).
|
The pattern of Cyp26b1 expression in the pharyngeal region at E9.5–10.5 is quite distinct from that of Cyp26a1, with strong expression in the hindbrain and a weaker domain in the endoderm and ectoderm of the caudal pa (23, Fig. 1D and E). Wholemount in situ staining showed reduced expression in the pa of E9.5 Tbx1 mutant embryos, whereas hindbrain expression appeared unaffected (Fig. 1D). More specifically, coronal sections showed a loss of Cyp26b1 expression from the caudal portion of the pharyngeal endoderm (pe) and an anterior shift in the ectodermal expression in the mutants (Fig. 1E). The alteration in Cyp26b1 expression in the mutants therefore appeared more complex than simple down-regulation in the absence of Tbx1.
This was also the case for the final member of the Cyp26 family to be examined, Cyp26c1. Cyp26c1 is found in mouse embryos from E8, with strong expression at E9.5 in rhombomere 2 (r2), maxillary and mandibular tissues of pa1 including surface ectoderm, superficial arch mesenchyme and maxillo-mandibular cleft. Strong expression is also observed in epibranchial placodes as well as a domain encompassing the cervical mesenchyme of the head lateral and caudal to the otic vesicle (24), that we found also extends into the otic vesicle itself as well as into the caudal pe (Fig. 1H and I). Partially overlapping expression of Tbx1 and Cyp26c1 was observed in serial sections of E9.5 wild-type embryos, in pa and head mesenchyme and pe. The restricted region of Cyp26c1 expression in the otic vesicle also seemed to overlap with the broader domain of Tbx1 expression in the latero-ventral portion of the posterior otic vesicle (52,53). An overlapping domain of expression within the mesodermal core of pa1 was also observed (Fig. 1J and L). In Tbx1 null embryos we found dramatic expression changes. In addition to a complete loss of expression in the peri-otic mesenchyme, otic vesicle, lateral epibranchial placodes and caudal pe, there was an expanded domain of expressing cells in the first arch mesenchyme (Fig. 1F–I). This latter result may represent the expansion of expression in neural crest-derived mesenchyme at the expense of core mesoderm rather than a proliferation of core mesoderm. Previous work has documented the loss of core mesoderm myogenic markers in pa1 such as Myf5 and Tlx1 concomitant with the expansion of neural crest-derived mesenchymal markers including Tbx2 and Tbx3, Dlx7 and Msx2 in Tbx1 nulls, and experiments in Xenopus using a dominant negative Tbx1 combined with lineage tracing reveal a requirement for Tbx1 in pharyngeal mesoderm (54,55). Expression in the hindbrain and maxillo-mandibular cleft was unchanged.
Chick embryos treated with high doses of R115866, a Cyp26 inhibitor, partially phenocopy the Tbx1 null mouse
Embryos treated with a high dose of R115866 at stage 10 or 14 and cultured for a further 24–48 h displayed a variety of externally visible defects including decreased head mesenchyme, smaller otic vesicles and loss of pa, loss of anterior tissues such as the forebrain and heart defects such as abnormal looping and pericardial oedema (Fig. 2). Many of these defects are phenocopied in Tbx1 null mice and embryos treated with excess RA. Severity varied from embryo to embryo, with some displaying all these phenotypic characteristics, whereas others were normally apart from the loss of caudal pa. In common with the Tbx1 null mouse, the caudal arches were the more severely affected; most embryos retained pa 1, and a rudimentary arch 2 was sometimes visible but pa 3 and 4 were rarely seen in treated embryos (Fig. 2A and B). Analysis of histological sections of these embryos also reflected the phenotypic range. The majority of embryos had at least attempted to form pa 1 and the accompanying paa 1 (Fig. 2B3, B4 and B8–10) although in the most severe embryos no other pa could be distinguished (Fig. 2B5). In slightly less severe embryos, hypoplastic remnants of pa 2 and sometimes pa 3 were also visible, but the segmentation of the ppe was essentially lost (Fig. 2B3, B4 and B8–10) as compared with controls (Fig. 2B1, B2, B6 and B7). Asymmetry in the formation of the bilaterally paired pharyngeal tissues was also sometimes apparent (Fig. 2B4 and B8–10; Fig. 6A9 and A10). Embryos with the very mildest phenotypes did occasionally form more caudal pa 3 and 4, but these were hypoplastic, abnormally shaped and contained extremely small non-patent paa (Fig. 3).
|
|
Other tissues affected included the otic vesicle which showed an overall size reduction in many cases and was narrower in others (Fig. 2C1–C4; Fig. 3C3, C5 and C6). Neuroepithelial development was also affected; abnormal folding of the neuroepithelium was seen, more commonly rostral to the spinal cord and failure of neural tube closure was also observed (Fig. 2C4–C7). More caudally, the pseudostratified columnar epithelial nature of the neural tube was disrupted by what appeared to ectopic tissue of a mesenchymal nature which caused breaks in the integrity of the neural tube (Fig. 2C9; Fig. 3C3).
Cardiac development in treated embryos was compromised; 24 h after treatment affected embryos appeared to have much shorter and straighter outflow tracts than in controls, with the distal OFT being particularly affected (Fig. 2A1–A8; Fig. 3A1–A10). Cardiac looping was reduced in these treated embryos as a consequence (Fig. 3A1–A10) and there appeared to be reduced ballooning of the ventricular chambers. The inner curvature of these hearts reflected this, appearing deeper and more perpendicular than in controls 24 h after R115866 treatment (Fig. 3B1–B4), and 48 h after treatment the inner curvature appeared smaller in treated embryos than in controls (Fig. 3B5 and B6). The myocardial trabeculation of the ventricular chambers also seemed to extend more cranially in some of these embryos (Fig. 3B5 and B6). We also noted that the caudal movement of the OFT was altered with the OFT often remaining much more cranial and exiting the body just below pa 1, probably as a consequence of the loss of caudal pharyngeal structures (Fig. 2A1–A8; Fig. 3D1–D5). Territory encompassing the secondary heart field (SHF) including the splanchnic mesoderm underlying the caudal pharynx was also dysmorphic appearing to be thinner and more disorganized in R115866 embryos. The distinctive pseudostratified columnar morphology described by Waldo et al. (56), as part of the SHF continuous with the splanchnic mesoderm and outflow tract myocardium was not identifiable with cells having a more general mesodermal appearance (Fig. 3E1 and E2).
R115866 prevents caudal paa formation
In Tbx1 null mice, the paa posterior to paa 1 are not formed. Embryos were given either R115866 in ethanol or the equivalent volume of ethanol alone at st10 in ovo and cultured for a further 24–48 h to stages 14–18, when they were injected with Indian ink to visualize the paa. The results are presented in Table 2. Two-thirds of control embryos examined had clearly formed patent paa 2, 3 and 4 (n=25) (Fig. 4E). A further one-third of control embryos which appeared slightly younger had well-formed patent paa 2 and 3 alone (n=12) (Fig. 4A). Many of the R115866-treated embryos (n=40) exhibited elements of RA teratogenesis as well as disruption/loss of pa formation (n=33). In 45% of these embryos (n=15), ink injections and sections showed that no paa formation had taken place and that there was anterior and posterior truncation. No beating hearts were evident in these embryos although the dorsal aorta could be identified and these embryos were most likely in the process of dying, particularly as they were only seen at 24 h of culture, presumably having died after 48 h of culture. In a further 18% (n=6) formation of paa 1 had been attempted on at least one side of the embryo and in the remaining 36% of embryos paa 1 formation was seen after ink injection (n=12) (Fig. 4B–D and F). However, 17.5% of all R115866-treated embryos (n=7) had a much milder phenotype and did not display any sign of general RA teratogenicity; these embryos retained some form of pa1 but had lost pa caudal to pa1 (Fig. 4C and G). All of these embryos had also failed to form caudal paa; 86% (n=6) had formed paa 1 alone and one embryo had formed paa 2 on one side in addition to paa 1 (Fig. 4H). The phenotype of these embryos was reminiscent of Tbx1 null mice which also form only paa 1 normally. Using Fishers exact probability test, results showed that paa abnormalities resulting from R115866 treatment when compared with controls were statistically significant at P<3x10–8.
|
|
Vascular smooth muscle (VSM) has been shown to be lacking in the fourth paa of Tbx1 heterozygous embryos in which they are reduced in size and patency (57). Tbx1 null mice fail to form arch structures including paa below pa 2. The majority of R115866-treated chick embryos also failed to form caudal arch structures including patent paa as described above. We examined those embryos which were less affected by R115866 for the presence of VSM by immunohistochemistry for smooth muscle
-actin (SMA) in whole-mount and in sections to see if the loss of VSM could be contributing to the loss of paa in these embryos. In whole mount, staining appeared patchy and disorganized when compared with controls (not shown) confirming abnormal formation of the paa. Treated embryos with the least affected pharyngeal phenotype were selected for staining and some of these embryos had attempted to form hypoplastic abnormally shaped pa 3 and 4. However, even in these less-affected embryos arch artery size was reduced when compared with controls. Staining in sections for smooth muscle actin revealed that staining around the paa and the carotid arteries was reduced as compared with untreated embryos, although expression remained high in the dorsal aorta. Examination using high power microscopy showed that the number of SMA-positive cells surrounding the arch arteries was greatly reduced as compared with controls, and lack of these cells probably contributes to the small pa artery phenotype observed in R115866-treated embryos (Fig. 4I–N). The SHF has recently been shown to express SMA22 and to contribute cells to the VSM, the base of the great vessels as well as OFT myocardium (56). We examined this region in R115866-treated embryos and again found diminished numbers of SMA22-positive cells (Fig. 4O–Q). These cells could either be SHF cells or possibly SMA22 neural crest-derived cells which also contribute to the smooth muscle of the arterial pole (56,58). SHF markers Islet-1 and Mef2c were also down-regulated in R115866-treated embryos with loss of expression including that in the pharyngeal mesoderm and outflow tract (Fig. 4R–U).
Low-dose R115866 produces CAT and aberrant aortic arch artery patterning in chick embryos
Embryos given lower doses of R115866 could be cultured for much longer than high-dose embryos which generally die by E5 at the latest. The external development of the heart and great vessels was examined by ink injection at E8+. Seventeen of 32 surviving R115866-treated embryos (53%) exhibited abnormalities in the development of the cardiac outflow tract and/or paa derivatives. Five embryos (29%) had a small brachiocephalic artery (BCA) compared with controls. Four of those affected had a reduced right BCA and one the left. In some of the embryos with a small right BCCA, the branch point of the right subclavian artery (RSCA) appeared to be lower than normal (Fig. 5A1–A5). Defects in alignment of the great vessels, and outflow tract and/or ventricular size/shape were seen in the majority of these 17 embryos. Fifteen of 17 embryos with cardiac anomalies were otherwise normal on external observation. In two embryos with heart defects craniofacial malformations were also seen. Similarly, in a Tbx1 allelic series, the paa and outflow tract derivatives are more dose-sensitive than craniofacial tissues (59). None of the 64 control embryos examined appeared to have cardiac or other anomalies on external inspection. All 17 R115866-treated embryos with vascular defects were sectioned. Poor section quality led to the exclusion of two of these from further analysis.
|
One of the 15 R115866-treated hearts had normal morphology, but the remaining 14 hearts all displayed ventricular septal defects (VSD) of varying type; 11 of these had a perimembranous and doubly committed juxtarterial VSD, a type classically associated with DGS patients [(60) and references therein] and also seen in mice carrying Tbx1 mutant alleles (1–4). This type of VSD extends onto the membranous portion of the ventricular septum. There is also failure of the formation of the subpulmonary infundibulum with which it would normally fuse and consequently the VSD is related to both of the valves of the aorta and pulmonary trunk which forms the roof of the defect (Fig. 5C1–C5). In three embryos, reduced size of one of the brachiocephalic arteries was seen in conjunction with the VSD (Fig. 5B1–B3). Three embryos were more severely affected and they all exhibited CAT, in which there is failure of septation of the outflow tract not only below the level of the arterial valves (PM-VSD), but also at the level of the valves and above so that there is a common valve and arterial trunk (Fig. 5D1–D5). This defect is also a feature of DGS and the Tbx1 homozygous null mouse (1–4,61,62). One of these embryos had a brachiocephalic artery of reduced size and another exhibited abnormal craniofacial development. Two embryos with PM-VSD were also found to have double outlet right ventricle (DORV), in which both the aorta and pulmonary trunk exit from the morphologically right ventricle, in this case with the great arteries in the normal spiral relationship to each other (Fig. 5E2).
Three of 14 embryos had morphologically distinct septal defects; one embryo had a subaortic VSD, where VSD is more closely related to the aortic valve than the pulmonary valve. This embryo also had a DORV, this time with the great arteries in an abnormal parallel arrangement relative to each other (Fig. 5E4). DORV is thought to be the result of insufficient looping and remodeling of the inner curvature of the heart (63), and is also found in DGS patients and Tbx1 mice homozygous for a hypomorphic allele which transcribes a low level of Tbx1 [estimated to be 
25% of normal levels (59,64)]. Two embryos had an atrioventricular septal defect with failure of fusion of the atrioventricular cushions and both also had a CAT i.e. failure of OFT cushion fusion (Fig. 5F1 and F2). In both of these embryos the positioning of the heart was not normal with one embryo having ectopia cordis. This embryo also had a thin ventricular wall with disorganized myocardium and an abnormal ventricular shape. The atrial wall was also thicker than normal and disorganized.
Of the nine control hearts also sectioned, seven were normal and two abnormal. Fishers exact probability test indicated the incidence of cardiovascular defects resulting from R115866 treatment versus ethanol carrier that was statistically significantly greater with a P-value of 0.00075. These data are summarized in Table 3.
|
Molecular markers reveal altered gene expression and morphogenesis in the pe and modified expression of RA-responsive genes
To examine whether blocking Cyp26 function was raising endogenous RA levels as anticipated, we studied the expression of a number of genes suggested either to be targets of Tbx1, RA or in some cases both, in embryos in which Cyp26 function had been blocked by a high dose of R115866 at st10 and followed with 24–48 h culture. In the Tbx1–/– mouse, one of the major tissues affected is the pe. A number of genes normally expressed in this tissue are down-regulated in the Tbx1 null embryos. One of these is Pax9, a molecular marker for the endoderm of the pharyngeal pouches (7). In situ hybridization in the chick revealed that the organization of these structures was badly disrupted by Cyp26 inhibition. In Tbx1–/– embryos, the caudal pouches were the worst affected. Expression of Pax9 in the endoderm was apparent at the expected anterior–posterior level but the organization of the endoderm into regular loops of the pouches was almost entirely lost. A structure approximating to pouch 1 was often apparent and in some embryos there was an attempt to form pharyngeal pouch 2. However, in the majority of embryos the organization of the pharyngeal pouches was non-existent posterior to pouch 1 in a fashion reminiscent of Tbx1 null embryos at equivalent developmental stages (Fig. 6A1–A11), although Pax9 expression could be maintained in the absence of pharyngeal pouch formation (Fig. 6A11).
|
Tbx1 is known to be able to regulate the expression of Fgf8 and Tbx1+/–Fgf8+/– mouse embryos have a higher penetrance of cardiac defects than Tbx1+/– alone (59,65). In the pharyngeal tissues, the results were similar to those seen with Pax9: in some embryos, generally those showing seriously abnormal development, pharyngeal expression was nearly completely lost although expression in the forebrain and midbrain/hindbrain boundary was maintained; in others, expression in the pe was present in pouches 1 and 2 but lost in more caudally and those pouches still expressing Fgf8 seemed dysmorphic. In addition, the expression domain in the branchial groove ectoderm in many of these embryos was appreciably closer to the most anterior pe expression suggesting loss of tissue in pa 1 (Fig. 6B1–B5). Tbx1 itself provides a good marker for non-neural crest pharyngeal tissues and has also been shown to be reduced in the presence of exogenous RA (66). In most R115866-treated embryos, both the spatial domain and intensity of Tbx1 expression was reduced in both pe and mesoderm, the SHF and the otic vesicle and the remaining expression reflected the disorganization and abnormal development of the pharyngeal region in R115866 embryos (Fig. 6C1 and C2).
We also examined the expression of the RA-synthesizing enzyme Raldh2 in R115866-treated embryos. Raldh2 is altered in the Tbx1–/– mouse such that there is an apparent rostral shift of expression in the splanchnic mesoderm, and expression in a small patch of mesenchymal cells between the forebrain and dorsal aorta just rostral to pa 1 is also up-regulated (7,67). In the Cyp26-blocked chick embryos, the strong expansion of expression in the small patch of mesenchymal cells rostral to pa1 seen in Tbx1–/– mice was observed in over half of R115866-treated chick embryos (n=5/9). In these embryos, there was also an anterior extension of expression in the ventral mesoderm; ectopic expression was also observed in pa 1 and in some cases it seemed that overall expression was also up-regulated (Fig. 6D1–D4).
Sox10 was used as a marker for migrating neural crest and cranial ganglia in R115866 embryos (68). Results showed diminished intensity of Sox10 staining and a reduction in the number of positive migrating neural crest cells (Fig. 6E1–E3). Abnormal patterning of the cranial ganglia as observed in Tbx1 mutant mice was also seen with the trigeminal ganglion (X) being greatly reduced in size as were the facioacoustic (VII/VIII) and glossopharyngeal (IX) ganglia. Aberrant neuronal pathfinding was also observed (Fig. 6E4 and E5).
Hoxb1 is normally expressed in endoderm and mesoderm caudal to pharyngeal pouch 3/4 depending on the age of the embryo. Treatment with RA induces ectopic anterior Hoxb1 expression in both tissues (69) and this ectopic anterior Hoxb1 expression is also seen in Tbx1–/– embryos (67). Very similar anterior ectopic shifts of expression to a level just below pa 1 were present in the pe and mesoderm of R115866-treated chick embryos supporting the interpretation of a rise in the levels of RA in this region (Fig. 6F1 and F2).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
We have shown that all three embryonic Cyp26 genes have reduced/altered expression patterns in the Tbx1–/– mouse. When the activity of the Cyp26 enzymes is inhibited in vivo in the chick embryo using R115866, a potent and selective inhibitor of Cyp26 enzyme function (51), we can reproduce many of the phenotypes associated with the Tbx1 null mouse and DGS. The R115866 phenocopy includes the loss of caudal pa and paa and the specific heart defects common in DGS patients, including CAT and DORV in association with perimembranous and doubly committed juxtarterial VSDs. This data suggests a functional link between altered expression of the Cyp26 genes with secondary disruption of the RA regulatory pathway and the DGS/Tbx1 null phenotype.
The discovery of three embryonic Cyp26 RA-metabolizing genes as potential Tbx1 targets raised the possibility that part of the Tbx1 phenotype may be mediated by a rise in the levels of RA local to Tbx1-expressing domains due to the loss of RA-metabolizing capacity, particularly as there are strong phenotypic similarities between the effects produced by excess RA and those seen with Tbx1 null mutations. As previously reported in independent studies (7,67) mouse retinaldehyde dehydrogenase 2 (Raldh2), the enzyme responsible for synthesizing the majority of embryonic RA also appears to have an ectopic anterior shift of expression in the splanchnic/ventral pharyngeal mesoderm and to be upregulated in the head mesenchyme of Tbx1 nulls. This suggests a possible up-regulation of RA synthesis in addition to down-regulation of RA-metabolizing Cyp26 genes which would further alter local RA levels. Evidence in favour of this theory was provided by RARE-LacZ reporter transgenic mice. Colourimetric assays of ß-galactosidase demonstrated that the boundary of expression of the RARE-reporter construct is shifted anteriorly in the pe and head mesenchyme of Tbx1 homozygous null embryos (67). The authors of this study suggested that loss of Tbx1 alters Raldh2 with altered Cyp26a1 and b1 expression occurring as a subsequent event. However, our data suggest that lack of Tbx1, amongst other effects, causes dysregulation of the Cyp26 family and that Cyp26 knockdown (albeit likely to a greater extent than that seen in Tbx1 nulls) is sufficient to cause the shift in Raldh2 expression and a DGS/VCFS phenocopy. It is now becoming apparent that Tbx1 regulates and interacts with a number of genes and pathways required for formation of the structures affected in DGS/VCFS. Thus, we report that the Cyp26 dysregulation is likely to represent just one aspect of the relevant downstream effects, but our data strengthens the hypothesis that an RA toxicity local to embryonic regions normally expressing Tbx1 contributes to or exacerbates the malformations seen in Tbx1–/– mice.
Comparison of the effect of RA upon Cyp26 expression with the effect of null alleles of Tbx1 also suggests that the expression changes in the latter are not simply the result of exposure to excess RA. In the mouse oral administration of RA to pregnant mice at E8.5 and E9.0 which were then examined at E9.5 abolished or greatly reduced Cyp26a1 expression in the caudal neural plate, tail-bud mesoderm and hindgut. In the same embryos Cyp26a1 expression was greatly up-regulated and expanded in anterior expression domains such as the cervical neural crest mesenchyme destined to form the cranial ganglia and the otic vesicle (11). However, in the Tbx1–/– mouse, Cyp26a1 expression was considerably down-regulated in these anterior tissues, particularly in the cranial ganglial neural crest cells. In the chick, grafts of RA-soaked beads also up-regulated Cyp26a1 expression in this cervical mesenchyme (18). For Cyp26b1 no published data is available on the affect of RA application in the mouse but in the chick RA-bead grafts up-regulated expression in anterior and posterior tissues (18). RA also induced Cyp26b1 expression in three human cell culture lines (13). This is in contrast to the down-regulation of expression in caudal pe in combination with an anterior shift of ectodermal expression from arch 3 to arch 2 seen in Tbx1–/– embryos. Only with Cyp26c1 does the result of excess RA partially match the changes in expression seen in Tbx1–/– embryos with loss of expression in cervical mesenchyme in both these embryos and E8.5 mouse and chick embryos exposed to exogenous RA (18,70). However, even here there was an expansion of Cyp26c1 positive cells in first arch mesenchyme in Tbx1–/– embryos which was not observed in RA-treated embryos. In addition, Raldh2 was up-regulated in head mesenchyme in R115866-treated embryos (as observed in Tbx1–/– embryos) and displayed ectopic expression in the pharyngeal region, suggesting that at least some of the altered domains of Raldh2 expression in Tbx1–/– embryos may be subsequent to Cyp26 down-regulation. RA rescue of a Raldh2–/– mice at early stages, followed by a period of RA clearance, revealed sites of RA production which did not correspond to the expression pattern for Raldh1 and Raldh3, raising the possibility that novel RA sources exist in the embryo. However, these sources are insufficient to rescue the pharyngeal/cardiovascular phenotype of these ‘RA-rescued embryos’. However, by analogy, it does raise the possibility that there may be RA sinks active in the embryo other than the Cyp26 enzymes (10,26,27). A careful analysis of the expression patterns of all these different genes and the different RA receptor genes in Raldh2 and Cyp26 knockout mice might contribute useful data towards resolving the hierarchy of normal interactions between Tbx1, Raldh2, the Cyp26 genes, RA levels and the RA-receptor genes. This is likely to be an extremely complex set of feedback loops given that many of the genes involved, including Tbx1 itself are RA-responsive (18,20,66,71–75). Indeed, the finding that Tbx1 itself was down-regulated in high-dose R115866-treated embryos raises the question of to what extent the observed phenotype is due directly to blockade of RA catabolism versus an indirect affect caused by reduced Tbx1 levels. While the lower levels of Tbx1 expression due to the rise in RA levels may make a small secondary contribution to the observed phenotypes of R115866 embryos, this is more likely to be directly attributable to loss of Cyp26 function. First, Tbx1 expression is only partially repressed and yet the phenotype seen is as severe, or worse than, that seen in the Tbx1 null mouse. Secondly, we have observed a clear phenotype in embryos 16 h after R115866 treatment (data not shown). For this to be mostly due to RA-mediated Tbx1 repression, the drug would have to act to raise RA levels sufficient to repress Tbx1. The reduced levels of Tbx1 would then have to affect downstream transcription and translation to produce a phenotype within 16 h. Given that RA repression of Tbx1 requires 12 h to become apparent (66) this seems unlikely. Finally, we have shown that embryos treated with low doses of R115866 have significant defects of the cardiovascular system later in development, which are very similar to those seen in Tbx1 mutant mice and DGS patients. In situ analysis of Tbx1 at earlier stages revealed no difference in expression between control and treated embryos (data not shown) making it improbable that this phenotype is due to diminished Tbx1, but rather is due to abrogated Cyp26 function.
One question which arises is whether the influence of Tbx1 upon Cyp26 expression is cell-autonomous, or non-cell autonomous. In some tissues Tbx1 and the Cyp26 genes are co-expressed, e.g. Cyp26b1 in pe and ectoderm suggesting a possible cell autonomous effect. Microarray studies in this laboratory in which cells carrying a Tbx1–LacZ knock-in transgene were isolated from embryos using a fluorescent ß-galactosidase substrate found down-regulation of Cyp26b1 in Tbx1–/– cells compared to Tbx1+/– cells, adding further support for a cell-autonomous effect (K. Lammerts van Buren, personal communication). However, in other cell types, a non-cell autonomous effect must be taking place; for example the strong expression of Cyp26c1 in cervical neural crest is down-regulated in Tbx1 null embryos even though Tbx1 is not expressed in neural crest.
R115866 blockade of Cyp26 function in the chick embryo phenocopies both the Tbx1 null mouse and DGS to a substantial degree. When a high dose is given, pharyngeal development is severely dysmorphic. In the majority of embryos, pa caudal to arch 2 failed to form, as did the characteristic bulges of the endodermal pharyngeal pouches. In 63% of these embryos only paa 1, with an occasional paa 2, was patent and in the remainder did not form a fully patent paa 1. In addition, the otic vesicle was often hypoplastic. These features have all been described as characteristic of the Tbx1–/– null mouse at E10.5 (2,4). The embryos lacking paa 1 could be taken as evidence that RA catabolism is required for paa formation. There is data that RA is required for endothelial cell proliferation and vascular remodelling at early stages of mouse development (76,77). These stages are equivalent to chick stages before stage 10 and so lack of paa 1 could be due to embryos being younger than st10 when they received R115866. However, it is difficult to assess the role for RA catabolism in paa 1 formation based on such embryos as they were both developmentally delayed and very sick.
In these R115866-treated embryos, early heart development was also compromised with a hypoplastic outflow tract, reduced looping and inner curvature remodelling. Lower doses of R115866, which allowed embryos to survive longer produced severe outflow tract abnormalities of a type classically associated with DGS and also seen in a variety of Tbx1 mutant mice (2,4,59,64,78). These include CAT, DORV and doubly committed juxtarterial VSD, which is very closely related to CAT and frequently observed in DGS (60). Interestingly, at these lower doses, the cardiovascular system appeared to be more sensitive than craniofacial regions with very few embryos displaying additional abnormal phenotypes to the heart defects observed. This is also true of Tbx1 knockout mice, where in an allelic series, the aortic arch and cardiovascular system was affected at Tbx1 dosages where craniofacial systems remained normal (59) and in Raldh2 null embryos ‘rescued’ by maternal RA administration (26,27). This is a likely explanation for the fact that at low doses of R115866, which produced classic DGS-like cardiac abnormalities, we did not observe any thymic defects although the pe, which gives rise to the thymus, was severely abnormal in high-dose embryos.
The early heart defects seen are consistent with the type of outflow tract abnormalities detected at later stages; both OFT hypoplasia and abnormal looping and remodelling are associated with a spectrum of defects including CAT, DORV and the type of VSDs seen here (63,79–81). Contributions from two different cell types, external to the outflow tract, have been shown to be critical for its growth, correct alignment and proper septation. These two populations are the cardiac neural crest and the cells of the SHF, which consists of pharyngeal and splanchnic mesoderm cells and is molecularly delineated by the expression of genes such as Isl-1, Tbx1 and Fgf10 [(82) and references therein] and (83). Comparison of the expression of Cyp26 genes with Tbx1 in mouse and chick reveals possibly overlapping domains of expression in the SHF with Cyp26a1 and b1 at E8.25-8.5 in the mouse and Cyp26b1 at stages 11–16 in the chick (11,16,23). Using Tbx1 as a marker for the SHF in R115866-treated embryos revealed diffuse staining across a reduced area rather than the normal strong expression and loss of expression of other SHF markers Isl-1 and Mef2c was also seen. Histological sections revealed substantial dysmorphogenesis in this region, suggesting that SHF domain was not as extensive or as well organized as in controls and/or did not express molecular markers normally. Tbx1 appears to have a role in regulating the addition of cells of the SHF to the developing heart as revealed by the OFT hypoplasia of Tbx1 homozygote mutant embryos at E9.5 which later leads to the severe OFT defects such as CAT and DORV at E18.5 in these embryos (2,4,64,78).
Altered Cyp26 expression in the neural crest may also contribute to the specific DGS-like heart phenotype observed after R115866-treatment. Altered gene expression in neural crest cells has been shown to produce DGS-like heart defects including CAT, interrupted aortic arch type B, DORV in numerous mouse models [reviewed in (84)]. In the chick, neural crest ablation is well-documented to produce similar phenotypes including CAT, DORV, VSD and tetralogy of Fallot (56,58,85–88), and in both species neural crest has been shown to contribute to the aorticopulmonary septum and OFT cushions and subpulmonary infundibulum during normal septation and alignment of the OFT and ventricles (89–91). Recently, it has been shown that ablation of the neural crest produces a shortened and straighter OFT which inhibits subsequent looping, explaining the malalignment heart phenotypes produced by neural crest ablation (87). Neural crest ablation has a significant affect upon the SHF, preventing SHF cells destined to become OFT myocardium from entering the OFT, although SHF-derived VSM cells are unaffected (56,58).
We have observed a number of similar phenotypes to those of neural-crest ablation in the R115866-treated embryos including shorter straighter outflow tract, reduced looping and reduced caudal movement of the OFT. Staining for neural crest-derived VSM is greatly reduced in surviving paa and in the SHF. Sox10 expression showed a reduced number of positive migrating neural crest cells in Cyp26-inhibited embryos and variable transcript levels when compared with controls. Abnormal patterning of the neural-crest derived cranial ganglia was also seen as in Tbx1 mutant embryos.
Finally, as well as the proper contribution from the neural crest and SHF, normal patterning and development of both the pe and craniofacial mesoderm (92) are required for normal heart development. Both Foxg1-Cre and Mesp1-Cre driven Tbx1 conditional knockouts have a hypoplastic OFT at E10.5, hypoplastic pharynx lacking normal endodermal pouch formation and patterning at E11.5, abnormal craniofacial development, absent thymus and parathyroid glands and a characteristic cardiovascular defect of CAT with VSD (93,94). In Tbx1–/– mice we have shown that both Cyp26b1 and Cyp26c1 expression is down-regulated in the pe; we cannot be certain whether some of the mesenchymal dysregulation is in mesodermal or crest derivatives, or both. Tbx1 null embryos, Foxg1-Cre and Mesp1 conditional Tbx1 mutants and R115866-treated embryos have many developmental anomalies in common. It would be interesting to examine the expression of the Cyp26 genes in the Mesp1 and Foxg1-Cre Tbx1 conditional mice to determine what parts of the Cyp26 expression changes are attributable to the loss of Tbx1 in the pharyngeal mesoderm versus endoderm (although Foxg1 may also be expressed in mesoderm). Experiments to block Cyp26 function in specific cellular subsets using morpholinos and conditional Cyp26 mouse mutants will help to reveal exactly which Cyp26s in which tissues are responsible for the various elements of the phenotype described here, as it seems possible that there could be varying contributions from pe, mesoderm, neural crest and SHF.
In this respect, the Cyp26a1 knockout mouse has certain phenotypes which we have observed in the R115866 embryos; pericardial oedema, abnormal heart looping and caudal truncation (48,95). This latter phenotype which was the most prevalent in the mice was not common in our treated chick embryos. This may be due to species differences but although R115866 is an effective inhibitor of Cyp26 function it will not be as effective or long-lasting as a genetic knock-out. Interestingly, homozygous null mutants for cytochrome P450 reductase, the essential electron donor for Cyp26s, have a severe embryonic lethal phenotype which includes loss of the pa as well as more extreme phenotypes such as defective vasculogenesis and somitogenesis, heart oedema and anterior and caudal truncation (96,97).
Our hypothesis, based on the initial discovery of Cyp26a1 down-regulation in conjunction with the ectopic shift in Raldh2 expression and increased RARE-LacZ reporter activity in Tbx1–/– (7,67) was that Cyp26 family knockdown would recapitulate some aspects of the 22q11 deletion syndrome. Here, we have shown that all three embryonic Cyp26 genes have significantly altered expression in the absence of Tbx1 transcription and reduction of embryonic Cyp26 function produces a DGS-like phenotype. As mentioned earlier, published accounts of single gene Cyp26 knockouts do not suggest a DGS-like phenotype. While Cyp26b1–/– mice have been examined from the point of view of limb development they do demonstrate some mid-gestational lethality, and unsuspected cardiac anomalies may underlie this phenomenon (50). Together these data provide functional evidence for a role of Cyp26 enzymes in generating Tbx1–/– phenotypes. The availability of individual Cyp26, Cpr and Tbx1 heterozygous mutants will allow investigation of epistatic relationships between RA catabolism and reduced dosage of Tbx1.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Mouse strains, breeding and genotyping
Timed matings of Tbx1+/– mice with Df1/+ mice on a mixed C57Bl6 backgrounds were used to generate trans-heterozygote embryos dissected out at E9.5. Genotyping of embryos was carried out by PCR on yolk sacs or small pieces of embryonic tissue as described previously (1,2,7). Embryos were then processed for RTQ-PCR as described below. Timed matings of Tbx1+/– mice on a C57Bl6 background were used to obtain Tbx1–/– embryos which were dissected at E9.5 and E10.5 and processed for in situ hybridization along with wild-type and heterozygote litter mate controls.
Real-time quantitative PCR
The pharyngeal region of five Df1/+;Tbx1+/– E9.5 embryos and four stage matched wild-type controls was dissected out and RNA extracted and amplified as described previously (7). For RTQ-PCR wild-type and mutant samples aRNA were pooled. One microgram of wild-type or mutant aRNA was reverse-transcribed and RTQ-PCR performed using SYBRGreen for detection of fluorescence for amplification. Primer sequences for Cyp26A1, B1 and C1 were designed with annealing temperatures of 57–59°C and sequences are available on request. Amplification and analysis were performed on the Smart Cycler System (Cepheid) or ABI 7000 (Cyp26C1) with reactions being carried out at least three times in duplicate, all as detailed previously (7). Normalization to mGAPDH internal controls was carried out as described previously (66).
R115866 chick embryo treatment
Fertilized hens' eggs (White Leghorn, Henry Stewart and Co. Ltd., UK) were incubated in a humidified incubator at 38°C until stages 10 or 14 (E2 and E3) were reached. R115866, a chemical inhibitor of Cyp26 function was kindly donated by Janssen Pharmeceutica. A window was cut in the shell and either 2 µl of 5 mg/ml R115866 in ethanol (high dose) or a lower dose (0.5 µl of 5 mg/ml R115866/EtOH or less) added in ovo. In control embryos an equivalent volume of ethanol alone was added. High-dose embryos were then cultured a further 24 or 48 h and low-dose embryos were cultured to E8–10.
Embryos were dissected out in PBS and after fixation in 4% paraformaldehyde processed further for either in situ hybridization, histology or for antibody staining. To visualize the aortic arch artery system at different developmental stages embryos were injected with Indian ink via the outflow tract of the heart or the aorta.
Histology
Embryos were dehydrated, embedded in paraffin wax, sectioned at 10 µm and stained with haemotoxylin–eosin.
In situ hybridization
In situ hybridization was performed at 68°C using standard protocols as described previously (98) for both mouse and chick embryos. We thank Hiroshi Hamada for the mouse Cyp26a1 probe, Dr. Malcolm Maden for chick Raldh2 and Hoxb1 probes, Dr Phillippa Francis-West for chick Fgf8, Dr Paul Scotting for chick Sox10 and Dr Paul Riley for Isl-1 and Mef2c probes. Ch