Axial skeletal and Hox expression domain alterations induced by retinoic acid, valproic acid, and bromoxynil during murine development

Retinoic acid (RA) alters the developmental fate of the axial skeletal anlagen. "Anteriorizations" or "posteriorizations," the assumption of characteristics of embryonic areas normally anterior or posterior to the affected tissues, are correlated with altered embryonal expression domains of Hox genes after in utero RA treatment. These "homeotic" changes have been hypothesized to result from alterations of a "Hox cod" which imparts positional identity in the axial skeleton. To investigate whether such developmental alterations were specific to RA, or were a more general response to xenobiotic exposure, CD-1 pregnant mice were exposed to RA, valproic acid (VA), or bromoxynil (Br) during organogenesis. Additionally, the expression domains of two Hox genes, Hoxa7 and Hoxa10, were examined in gestation day (GD) 12.5 embryos obtained from control, RA, VA, or Br, treated gravid dams exposed on GD 6, 7, or 8. The anterior expression boundary of Hoxa7 is at the level of the C7/T1 vertebrae and that of Hoxa10 is at L6/S1. Compound-induced changes in the incidence of skeletal variants were observed. These included supernumerary cervical ribs (CSNR) lateral to C7, 8 vertebrosternal ribs, supernumerary lumbar ribs (LSNR) lateral to L1, extra presacral vertebrae, and the induction of vertebral and/or rib malformations. RA and VA administration on GD 6 caused posteriorization in the cervico-thoracic region (CSNR) while GD 8 exposure to any of the three compounds resulted in anteriorizations in the thoraco-lumbar area (LSNR and an increase in the number of presacral vertebrae). These effects occurred across regions of the axial skeleton. Analysis of gene expression demonstrated changes in the anterior boundaries of Hoxa7 expression domains in embryos treated on GD 6 and 8 with RA. VA and Br did not induce any statistically significant alterations in Hoxa7 and none of the compounds caused alterations in Hoxa10 expression domains. The studies indicate that RA GD 6 treatment-induced Hoxa7 shifts were rostral (posteriorization) while the RA-induced GD 8 anterior expression boundary shift was caudal (anteriorization), correlating with the axial skeletal changes noted. These data suggest that xenobiotic compounds such as VA and Br may induce similar axial skeletal changes by affecting different components of the developmental processes involved in the patterning of the axial skeleton.     

Hox patterning of the vertebrate rib cage

Unlike the rest of the axial skeleton, which develops solely from somitic mesoderm, patterning of the rib cage is complicated by its derivation from two distinct tissues. The thoracic skeleton is derived from both somitic mesoderm, which forms the vertebral bodies and ribs, and from lateral plate mesoderm, which forms the sternum. By generating mouse mutants in Hox5, Hox6 and Hox9 paralogous group genes, along with a dissection of the Hox10 and Hox11 group mutants, several important conclusions regarding the nature of the ;Hox code' in rib cage and axial skeleton development are revealed. First, axial patterning is consistently coded by the unique and redundant functions of Hox paralogous groups throughout the axial skeleton. Loss of paralogous function leads to anterior homeotic transformations of colinear regions throughout the somite-derived axial skeleton. In the thoracic region, Hox genes pattern the lateral plate-derived sternum in a non-colinear manner, independent from the patterning of the somite-derived vertebrae and vertebral ribs. Finally, between adjacent sets of paralogous mutants, the regions of vertebral phenotypes overlap considerably; however, each paralogous group imparts unique morphologies within these regions. In all cases examined, the next-most posterior Hox paralogous group does not prevent the function of the more-anterior Hox group in axial patterning. Thus, the ;Hox code' in somitic mesoderm is the result of the distinct, graded effects of two or more Hox paralogous groups functioning in any anteroposterior location.     

Cdx1 and Cdx2 have overlapping functions in anteroposterior patterning and posterior axis elongation

Mouse Cdx and Hox genes presumably evolved from genes on a common ancestor cluster involved in anteroposterior patterning. Drosophila caudal (cad) is involved in specifying the posterior end of the early embryo, and is essential for patterning tissues derived from the most caudal segment, the analia. Two of the three mouse Cdx paralogues, Cdx 1 and Cdx2, are expressed early in a Hox-like manner in the three germ layers. In the nascent paraxial mesoderm, both genes are expressed in cells contributing first to the most rostral, and then to progressively more caudal parts of the vertebral column. Later, expression regresses from the anterior sclerotomes, and is only maintained for Cdx1 in the dorsal part of the somites, and for both genes in the tail bud. Cdx1 null mutants show anterior homeosis of upper cervical and thoracic vertebrae. Cdx2-null embryos die before gastrulation, and Cdx2 heterozygotes display anterior transformations of lower cervical and thoracic vertebrae. We have analysed the genetic interactions between Cdx1 and Cdx2 in compound mutants. Combining mutant alleles for both genes gives rise to anterior homeotic transformations along a more extensive length of the vertebral column than do single mutations. The most severely affected Cdx1 null/Cdx2 heterozygous mice display a posterior shift of their cranio-cervical, cervico-thoracic, thoraco-lumbar, lumbo-sacral and sacro-caudal transitions. The effects of the mutations in Cdx1 and Cdx2 were co-operative in severity, and a more extensive posterior shift of the expression of three Hox genes was observed in double mutants. The alteration in Hox expression boundaries occurred early. We conclude that both Cdx genes cooperate at early stages in instructing the vertebral progenitors all along the axis, at least in part by setting the rostral expression boundaries of Hox genes. In addition, Cdx mutants transiently exhibit alterations in the extent of Hox expression domains in the spinal cord, reminding of the strong effects of overexpressing Cdx genes on Hox gene expression in the neurectoderm. Phenotypical alterations in the peripheral nervous system were observed at mid-gestation stages. Strikingly, the altered phenotype at caudal levels included a posterior truncation of the tail, mildly affecting Cdx2 heterozygotes, but more severely affecting Cdx1/Cdx2 double heterozygotes and Cdx1 null/Cdx2 heterozygotes. Mutations in Cdx1 and Cdx2 therefore also interfere with axis elongation in a cooperative way. The function of Cdx genes in morphogenetic processes during gastrulation and tail bud extension, and their relationship with the Hox genes are discussed in the light of available data in Amphioxus, C. elegans, Drosophila and mice.     

Specification of vertebral identity is coupled to Notch signalling and the segmentation clock

To further analyse requirements for Notch signalling in patterning the paraxial mesoderm, we generated transgenic mice that express in the paraxial mesoderm a dominant-negative version of Delta1. Transgenic mice with reduced Notch activity in the presomitic mesoderm as indicated by loss of Hes5 expression were viable and displayed defects in somites and vertebrae consistent with known roles of Notch signalling in somite compartmentalisation. In addition, these mice showed with variable expressivity and penetrance alterations of vertebral identities resembling homeotic transformations, and subtle changes of Hox gene expression in day 12.5 embryos. Mice that carried only one functional copy of the endogenous Delta1 gene also showed changes of vertebral identities in the lower cervical region, suggesting a previously unnoticed haploinsufficiency for Delta1. Likewise, in mice carrying a null allele of the oscillating Lfng gene, or in transgenic mice expressing Lfng constitutively in the presomitic mesoderm, vertebral identities were changed and numbers of segments in the cervical and thoracic regions were reduced, suggesting anterior shifts of axial identity. Together, these results provide genetic evidence that precisely regulated levels of Notch activity as well as cyclic Lfng activity are critical for positional specification of the anteroposterior body axis in the paraxial mesoderm.     

Plasticity of axial identity among somites: cranial somites can generate vertebrae without expressing Hox genes appropriate to the trunk

Classic studies have shown that the presomitic mesoderm is already committed to a specific morphological fate, for example, the ability to generate a rib. Hox gene expression in the paraxial mesoderm has also been shown to be fixed early and not susceptible to modulation by an ectopic environment. This is in contrast to the plasticity of Hox expression in neuroectodermal derivatives. We reexamine here the potential of somites for morphological plasticity by transplanting the cranial (occipital) somites 1-4, that normally produce small contributions to the skull, to the trunk of avian embryos. Surprisingly, the transposed cranial somites are able to form reasonably normal vertebral anlage. In addition, the cranial somitic mesoderm produces intervertebral disks, structures not normally found in the skull. These somites are however unable to generate some elements of the vertebrae, such as the costal process. In contrast to the morphogenetic plasticity of the occipital somites, their characteristic inability to support survival of dorsal root ganglia was not significantly modified by posterior transplantation. Dorsal root ganglia initially developed and then degenerated with the same morphological stages as normally observed. In striking contrast to the plasticity of morphology, we found that all four members of the of the fourth paralogous group of Hox genes that are expressed endogenously at the level of the graft are not upregulated in the caudad-transposed cranial mesoderm. It therefore appears that genes other than those of the Hox family normally expressed at this axial level control the position-specific morphogenesis of ectopic vertebrae formed from cranial somites. In evolutionary terms, the present results imply that occipital somites that were incorporated into the "New Head" retain the ability to develop according to their original morphogenetic fate, into vertebrae.     

Why do almost all mammals have seven cervical vertebrae? Developmental constraints, Hox genes, and cancer.

Mammals have seven cervical vertebrae, a number that remains remarkably constant. I propose that the lack of variation is caused by developmental constraints: to wit, changes in Hox gene expression, which lead to changes in the number of cervical vertebrae, are associated with neural problems and with an increased susceptibility to early childhood cancer and stillbirths. In vertebrates, Hox genes are involved in the development of the skeletal axis and the nervous system, among other things. In humans and mice, Hox genes have been shown also to be involved in the normal and abnormal (cancer) proliferation of cell lines; several types of cancer in young children are associated with abnormalities in Hox gene expression and congenital anomalies. In these embryonal cancers the incidence of a cervical rib (a rib on the seventh cervical vertebra, a homeotic transformation of a cervical vertebra towards a thoracic-type vertebra) appears to be increased. The minimal estimate of the selection coefficient acting against these mutations is about 12%. In birds and reptiles variations in the number of cervical vertebrae have frequently occurred and there is often intraspecific variability. A review of the veterinary literature shows that cancer rates appear lower in birds and reptiles than in mammals. The low susceptibility to cancer in these classes probably prevents the deleterious pleiotropic effect of neonatal cancer when changes in cervical vertebral number occur. In mammals there is, thus, a coupling between the development of the axial skeleton and other functions (including the proliferations of cell lines). The coupling of functions is either a conserved trait that is also present in reptiles and birds, but without apparent deleterious effects, or the coupling is new to mammals due to a change in the functioning of Hox genes. The cost of the coupling of functions in mammals appears to be an increased risk for neural problems, neonatal cancer, stillbirths, and a constraint on the variability of cervical vertebral number.     

Growth differentiation factor 11 signaling controls retinoic acid activity for axial vertebral development

Mice deficient in growth differentiation factor 11 (GDF11) signaling display anterior transformation of axial vertebrae and truncation of caudal vertebrae. However, the in vivo molecular mechanisms by which GDF11 signaling regulates the development of the vertebral column have yet to be determined. We found that Gdf11 and Acvr2b mutants are sensitive to exogenous RA treatment on vertebral specification and caudal vertebral development. We show that diminished expression of Cyp26a1, a retinoic acid inactivating enzyme, and concomitant elevation of retinoic acid activity in the caudal region of Gdf11^−/− embryos may account for this phenomenon. Reduced expression or function of Cyp26a1 enhanced anterior transformation of axial vertebrae in wild-type and Acvr2b mutants. Furthermore, a pan retinoic acid receptor antagonist (AGN193109) could lessen the anterior transformation phenotype and rescue the tail truncation phenotype of Gdf11^−/− mice. Taken together, these results suggest that GDF11 signaling regulates development of caudal vertebrae and is involved in specification of axial vertebrae in part by maintaining Cyp26a1 expression, which represses retinoic acid activity in the caudal region of embryos during the somitogenesis stage.     

Mice doubly deficient for the Polycomb Group genes Mel18 and Bmi1 reveal synergy and requirement for maintenance but not initiation of Hox gene expression

Polycomb group genes were identified as a conserved group of genes whose products are required in multimeric complexes to maintain spatially restricted expression of Hox cluster genes. Unlike in Drosophila, in mammals Polycomb group (PcG) genes are represented as highly related gene pairs, indicative of duplication during metazoan evolution. Mel18 and Bmi1 are mammalian homologs of Drosophila Posterior sex combs. Mice deficient for Mel18 or Bmi1 exhibit similar posterior transformations of the axial skeleton and display severe immune deficiency, suggesting that their gene products act on overlapping pathways/target genes. However unique phenotypes upon loss of either Mel18 or Bmi1 are also observed. We show using embryos doubly deficient for Mel18 and Bmi1 that Mel18 and Bmi1 act in synergy and in a dose-dependent and cell type-specific manner to repress Hox cluster genes and mediate cell survival of embryos during development. In addition, we demonstrate that Mel18 and Bmi1, although essential for maintenance of the appropriate expression domains of Hox cluster genes, are not required for the initial establishment of Hox gene expression. Furthermore, we show an unexpected requirement for Mel18 and Bmi1 gene products to maintain stable expression of Hox cluster genes in regions caudal to the prospective anterior expression boundaries during subsequent development.     

Segmental relationship between somites and vertebral column in zebrafish

The segmental heritage of all vertebrates is evident in the character of the vertebral column. And yet, the extent to which direct translation of pattern from the somitic mesoderm and de novo cell and tissue interactions pattern the vertebral column remains a fundamental, unresolved issue. The elements of vertebral column pattern under debate include both segmental pattern and anteroposterior regional specificity. Understanding how vertebral segmentation and anteroposterior positional identity are patterned requires understanding vertebral column cellular and developmental biology. In this study, we characterized alignment of somites and vertebrae, distribution of individual sclerotome progeny along the anteroposterior axis and development of the axial skeleton in zebrafish. Our clonal analysis of zebrafish sclerotome shows that anterior and posterior somite domains are not lineage-restricted compartments with respect to distribution along the anteroposterior axis but support a 'leaky' resegmentation in development from somite to vertebral column. Alignment of somites with vertebrae suggests that the first two somites do not contribute to the vertebral column. Characterization of vertebral column development allowed examination of the relationship between vertebral formula and expression patterns of zebrafish Hox genes. Our results support co-localization of the anterior expression boundaries of zebrafish hoxc6 homologs with a cervical/thoracic transition and also suggest Hox-independent patterning of regionally specific posterior vertebrae.     

Exogenous retinoic acid rapidly induces anterior ectopic expression of murine Hox-2 genes in vivo.

Exogenous retinoic acid (RA) has teratogenic effects on vertebrate embryos and alters Hox-C gene expression in vivo and in vitro. We wish to examine whether RA has a role in the normal regulation of Hox-C genes, and whether altered Hox-C gene expression in response to RA leads to abnormal morphology. The expression of 3' Hox-2 genes (Hox-2.9, Hox-2.8, Hox-2.6 and Hox-2.1) and a 5' gene (Hox-2.5) were examined by whole-mount in situ hybridization on embryos 4 hours after maternal administration of teratogenic doses of RA on embryonic day 7 to 9. The expression of the 3' Hox-2 genes was found to be ectopically induced in anterior regions in a stage-specific manner. The Hox-2.9 and Hox-2.8 genes were induced anteriorly in the neurectoderm in response to RA on day 7 but not at later stages. Expression of Hox-2.6 and Hox-2.1 was ectopically induced anteriorly in neurectoderm in response to RA on day 8. Hox-2.1 remained responsive on day 9, whereas Hox-2.6 was no longer responsive at this stage. The expression of the 5' gene Hox-2.5 was not detectably altered at any of these stages by RA treatments. We also examined the response of other genes whose expression is spatially regulated in early embryos. The expression of En-2 and Wnt-7b was not detectably altered by RA, whereas RAR beta expression was induced anteriorly by RA on day 7 and 8. Krox-20 expression was reduced in a stage- and region-specific manner by RA. The ectopic anterior expression of Hox-2.8 and Hox-2.9 induced by RA on day 7 was persistent to day 8, as was the altered expression of Krox-20. The altered pattern of expression of these genes in response to RA treatment on day 7 may be indicative of a transformation of anterior hindbrain to posterior hindbrain, specifically, a transformation of rhombomeres 1 to 3 towards rhombomere 4 identity with an anterior expansion of rhombomere 5. The ectopic expression of the 3' Hox-2 genes in response to RA is consistent with a role for these genes in mediating the teratogenic effects of RA; the rapid response of the Hox-C genes to RA is consistent with a role for endogenous RA in refining 3' Hox-C gene expression boundaries early in development.     

Wnt signaling is a key mediator of Cdx1 expression in vivo

In the mouse, Cdx1 is essential for normal anteroposterior vertebral patterning through regulation of a subset of Hox genes. Retinoic acid (RA) and certain Wnts have also been implicated in vertebral patterning, although the relationship between these signaling pathways and the regulation of mesodermal Hox gene expression is not fully understood. Prior work has shown that Cdx1 is a direct target of both Wnt and retinoid signaling pathways, and might therefore act to relay these signals to the Hox genes. Wnt and RA are believed to impact on Cdx1 through an atypical RA-response element (RARE) and Lef/Tcf-response elements (LRE), respectively, in the proximal promoter. To address the roles of these regulatory motifs and pathways, we derived mice mutated for the LRE or the LRE plus the RARE. In contrast to RARE-null mutants, which exhibit limited vertebral defects, LRE-null and LRE+RARE-null mutants exhibited vertebral malformations affecting the entire cervical region that closely phenocopied the malformations seen in Cdx1-null mutants. Mutation of the LRE also greatly reduced induction of Cdx1 by RA, demonstrating a requirement for Wnt signaling in the regulation of this gene by retinoids. LRE and LRE+RARE mutants also exhibited vertebral fusions, suggesting a defect in somitogenesis. As Wnt signaling is implicated in somitogenesis upstream of the Notch pathway, it is conceivable that Cdx1 might play a role in this process. However, none of the Notch pathway genes assessed was overtly affected.     

Cdx4 is a direct target of the canonical Wnt pathway

There is considerable evidence that the Cdx gene products impact on vertebral patterning by direct regulation of Hox gene expression. Data from a number of vertebrate model systems also suggest that Cdx1, Cdx2 and Cdx4 are targets of caudalizing signals such as RA, Wnt and FGF. These observations have lead to the hypothesis that Cdx members serve to relay information from signaling pathways involved in posterior patterning to the Hox genes. Regulation of Cdx1 expression by RA and Wnt in the mouse has been well characterized; however, the means by which Cdx2 and Cdx4 are regulated is less well understood. In the present study, we present data suggesting that Cdx4 is a direct target of the canonical Wnt pathway. We found that Cdx4 responds to exogenous Wnt3a in mouse embryos ex vivo, and conversely, that its expression is down-regulated in Wnt3a(vt/vt) embryos and in embryos cultured in the presence of Wnt inhibitors. We also found that the Cdx4 promoter responds to Wnt signaling in P19 embryocarcinoma cells and have identified several putative LEF/TCF response elements mediating this effect. Consistent with these data, chromatin immunoprecipitation assays from either embryocarcinoma cells or from the tail bud of embryos revealed that LEF1 and beta-catenin co-localize with the Cdx4 promoter. Taken together, these results suggest that Cdx4, like Cdx1, is a direct Wnt target.