Caudal (Cdx) and Hox proteins as regulators of the embryonic body plan
During the early development of a mouse or human embryo, the correct location of body parts along the head-to-tail axis is specified by the expression boundaries of 39 special genes known as Hox genes. Each Hox protein is expressed in a unique spatial domain along the body where, functioning as a transcription factor, it instructs embryonic cells on a particular route of morphogenesis. Hox expression domains thus provide a pre-pattern to the body plan. How are these domains established? The three caudal (Cdx) proteins are upstream activators of Hox genes, and they are expressed in posterior-to-anterior gradients along the embryonic axis. Could these gradients function as instructional (morphogen) gradients for the setting of Hox domains?
Fig 1a and 1b. (Click to enlarge)
Structure of transgenes 1 to 3 (A) and their expressionpatterns in 10.5 day mouse embryo (B)
Expression of the Hoxa-7 gene can be detected experimentally by looking at Hoxa-7/lacZ transgenes expressed in transgenic mice. Wherever the Hoxa-7/lacZ protein is produced it can be detected as a blue stain. When the transgene, like the normal gene, contains one copy of the enhancer element (red box in Figure 1A) the anterior boundary of the expression domain lies at the levels of spinal ganglion 5 in neural tissue and vertebra 13 in mesoderm (transgene1; Figure 1B). The addition of three extra copies of the enhancer causes a forward shift in the anterior boundary of the expression domain to the levels of spinal ganglion 3 in neural tissue and vertebra 8 in mesoderm (transgene 2; Figure 1B).
The Hoxa-7 enhancer element contains the motif TTTATG, which is known to be the binding site for Cdx transcription factors. To test the significance of this in the difference between the expression of transgenes 1 and 2 (Figure 1), we produced transgene 3 which is identical to transgene 2 except that the three additional copies of the enhancer are each mutated in the binding motif (green boxes; Figure 1A). The expression boundaries of transgene 3 are located at the levels of spinal ganglion 4 and vertebra 11 (Figure 1B). Thus, much, although not all, of the anteriorizing effect of additional enhancers (transgene 2 relative to transgene 1) is nullified by destruction of the additional Cdx protein binding sites (transgene 3).
These results can be explained if firstly: there is normally an instructional (morphogen) gradient of Cdx proteins along the length of the embryo to regulate each Hox gene's expression boundaries within specific threshold limits, and secondly: multiple copies of the cdx binding motif make a Hox gene more sensitive to Cdx concentration, thereby enabling activation at lower (more anterior) positions along the Cdx gradient.
Gradients of Cdx gene activity formed by time-dependent decay
The three vertebrate caudal (Cdx) proteins are upstream activators of the Hox genes. The Hox genes become expressed in a series of partially overlapping domains along the embryo. Hox proteins, functioning as transcription factors, then coordinate development of the embryo along its anteroposterior axis.
Figure 2. (Click to enlarge)
8.7 day (A) and 8 day (B) mouse embryos from a transgenic line expressing a cdx-1/lacZ reporter construct
In an attempt to unravel the mechanism of Cdx gene function, particularly with regard to Hox gene activation, we have studied the mouse Cdx1 gene and its chick orthologue, CdxA, by use of lacZ reporter constructs expressed in transgenic mouse embryos. Both chick and mouse constructs give similar results: at 8.7 days, Cdx/lacZ reporter activity is seen as a clear gradient over somites 5 to 11, and over the anterior neural tube (Figure 2A, C-F).
We have suggested that such gradients of Cdx gene activity may function as morphogen gradients to specify the expression domains of the Hox genes.
In this model, different Hox genes would become activated at different threshold concentrations of Cdx gene activity, and each Hox gene would therefore acquire its own anterior boundary of expression along the embryo. The range of somites over which we observe the gradient (somites 5 to 11, contributing to neck vertebrae 1 to 7) is the same as those over which there is derangement of Hox expression, and accompanying homeotic mutation, in Cdx1 knockout and over-expresser mice.
The mouse embryo forms by the progressive emergence of somites from posterior tissues (the primitive streak). We found that the gradient forms by a time-dependent decay of Cdx/lacZ activity within somites once they have emerged from the streak. Newly formed (more posterior) somites therefore have a higher activity than older (more anterior) somites. Evidence for this is seen by comparing Figure 2A and B. Thus, it is seen that at 8 days newly formed somite 5 is labelled as strongly as more posterior parts, yet by 8.7 days somite 5 is only weakly labelled.
By deletion of sequence within the Cdx/lacZ reporter constructs we found that an essential Cdx enhancer element lies within the first intron. Comparison of mouse and chick sequences over this region shows conserved motifs, potentially responsive to both retinoic acid and Wnt/β-catenin signalling. We found that specific mutation of either of these two binding motifs results in a posterior shift in the position of the Cdx/lacZ activity along the embryo, indicating roles for both retinoic acid and Wnt proteins in the upstream activation of Cdx1.
Increased Cdx protein dose effects upon axial patterning.
A prediction of the Cdx morphogen gradient model is that Hox boundary positions must be sensitive to dosage of Cdx proteins. To test this we have constructed transgenic mouse lines each of which over-expresses one of the Cdx genes under control of its own promoter and enhancer elements and within the normal expression domain of the tailbud.
We find that over-expressions of the Cdx proteins result in forward shifts in 1) the Cdx protein gradients along the embryonic axis, 2) the expression boundaries of Hox genes, and 3) vertebral morphologies along the axial skeleton (Figure 3). For example, in the normal, non-transgenic (NT) mouse the 8th vertebra bears the first rib (Figure 3A), the 20th vertebra the last rib (Figure 3D), and the 26th vertebra is the first sacral vertebra (attached to pelvis) (Figure 3D). A mouse over-expressing Cdx1 has the first rib on the 5th vertebra (Figure 3B), the 1st vertebra attached to the skull (Figure 3B), the last rib on the 18th vertebra (Figure 3E), and the first sacral vertebra is the 24th (Figure 3E). A mouse over-expressing Cdx2 has the first rib on the 7th vertebra and the 2nd vertebra attached to the first (Figure 3C).
Figure 3. (click to enlarge)
Effect of Cdx over-expression on the skeletons of newborn mice A-C, neck region viewed from the side; D, E, axial skeleton viewed from above. Numbers are vertebral addresses. A, D, normal (non-transgenic) mice; B, E, Cdx1 over-expresser; C, Cdx2 over-expresser. Blue, cartilage; red, bone.
The term homeotic mutation is used to describe such a shift of normal anatomical structures to abnormal positions along the axis. We suggest, as the most likely explanation of these findings, that homeotic mutations result from the shifts in Hox expression domains, and that these result in turn from shifts in Cdx morphogen gradients. The three Cdx genes clearly overlap with each other in their functions but they differ in the anteriormost limits of their effects. We find that these limits are at the levels of vertebra 1 for Cdx1 (Figure 3B), vertebra 2 for Cdx2 (Figure 3C) and vertebra 7 for Cdx4. Cdx over-expression defects are found most commonly in the neck and anterior thoracic regions leading us to believe that other gene products may form more posterior morphogen gradients along the axis.
Origins of Cdx1 retinoic acid response elements suggest roles in vertebrate evolution
The major groups of vertebrates diverged at different evolutionary times, although little is known about the genes responsible for this and the accompanying acquisition of new body forms. Hox and Cdx genes, being regulators of body design, are likely candidates. Retinoic acid regulates Cdx1 expression in amniotes (mammals, birds and reptiles) though not in amphibia (Xenopus). We found that one retinoic acid response element (RARE 1) arose in the Cdx1 gene at the split between amphibia and amniotes, whilst a second (RARE 2) arose at the split between marsupial and eutherian mammals (Figure 4). To our knowledge this is the first evidence that origins of regulatory elements in a homeotic gene coincide with evolutionary transitions between vertebrate groups.
Cdx1 mutant mice display two major abnormalities, one in the neck vertebrae (Figure 3B) and the other in the female urogenital system. We have proposed that origins of RAREs 1 and 2 (Figure 4) contributed to the considerable morphological changes that occurred at these transitions in, respectively, the neck and the female urogenital anatomies.
Figure 4. (click to enlarge)
Origins of Cdx1 retinoic acid response elements in vertebrate evolution.(mya; millions of years ago)
Evolutionary shifts of vertebrate Hox gene expression
During evolution, animals may change their body shape by shift (or 'transposition') of anatomical structures up or down their body axes. We earlier showed how this shift in mesodermal structures (e.g. the vertebral column) could be attributed to a prior shift in the embryonic expression patterns of Hox genes. More recently, we presented evidence that shifts in embryonic Hox expression can similarly account for evolutionary transposition in neural components. Mouse and chick display a relative transposition in the position of their forelimb and accompanying brachial plexus.
Figure 5 (click to enlarge)
How this is presaged by similar transposition in the neural expression of the Hoxa-7 gene. The brackets depict the embryonic expression domains of various Hoxa genes. Spinal ganglia, vertebrae and somites are shown as circles, squares and lines respectively.
Cervical, thoracic, lumbar and sacral structures are coloured pale blue and yellow alternately along thebody. Arrows show anterior boundaries of Hox gene expression in vertebrae.
Updated 23 August, 2011