Abstract

The vertebrate caudal proteins, being upstream regulators of the Hox genes, play a role in establishment of the body plan. We describe analysis of two orthologous caudal genes (chick cdx-A and mouse cdx-1) by use of lacZ reporters expressed in transgenic mouse embryos. The expression patterns show many similarities to the expression of endogenous mouse cdx-1. At 8.7 days, cdx/lacZ activity within neurectoderm and mesoderm forms posterior-to-anterior gradients, and we discuss the possibility that similar gradients of cdx gene expression may function as morphogen gradients for the establishment of Hox gene expression boundaries. Our observations suggest that gradients form by decay of cdx/lacZ activity in cells that have moved anterior to the vicinity of the node. The cdx-A/lacZ expression pattern requires an intron enhancer that includes two functional control elements: a DR2-type retinoic acid response element and a Tcf/β-catenin binding motif. These motifs are structurally conserved in mouse cdx-1.

Author Keywords: cdx-A; cdx-1; Transgenic mice; lacZ; Expression gradient; Tcf/β-catenin; Retinoic acid response element; Hox

Article Outline

1. Introduction

2. Results

2.1. cdx-A/lacZ reporter is expressed in gradients in transgenic embryos

2.2. cdx-A/lacZ reporters require an intron element for cdx-like expression in transgenic embryos

2.3. Conserved Tcf/β-catenin binding sites and a RARE within the cdx-A reporter construct

2.4. Formation of the gradients detected by chick cdx-A/lacZ expression

2.5. Formation of the gradients detected by expression of a mouse cdx-1/lacZ transgene

2.6. cdx-A/lacZ and cdx-1/lacZ activity in 9–10.5 day embryos

2.7. cdx-1 and cdx-1/lacZ expression compared by in situ hybridization

3. Discussion

3.1. Gradients of cdx-A/lacZ and cdx-1/lacZ expression

3.2. The enhancer elements in chick cdx-A and mouse cdx-1

3.3. Upstream regulators of the cdx genes

4. Experimental procedures

4.1. Preparation of chick cdx-A/lacZ and mouse cdx-1/lacZ reporter constructs

4.2. Transgenic embryo production and staining

4.3. In situ hybridization

References

1. Introduction

The mouse embryo develops in an anterior-to-posterior temporal sequence by the progressive addition of new parts to the growing posterior end. The posterior region can therefore be considered as a ‘founder zone’. For paraxial and lateral plate mesoderm, the founder cells reside within the primitive streak, which terminates anteriorly at the level of the node. For neural tube, founder cells lie within the ectoderm germ layer. These migrate towards the midline to join the neural tube at its growing posterior end.

Whilst this is in progress, the Hox gene expression patterns become laid down, with the various genes acquiring their characteristic, and often unique, anterior boundaries of expression. In accordance with spatial colinearity, the order of their expression domains (anterior to posterior) corresponds with the order of the genes (3′–5′) along their clusters (e.g. Gaunt et al., 1988).

In recent years, the caudal (cdx) genes have become strong candidates as regulators of Hox gene expression patterns. The evidence is as follows. (1) cdx binding motifs are present in the enhancers of several different vertebrate Hox genes (first shown by Shashikant et al., 1995 and Subramanian et al., 1995), and mutation of these motifs disrupts expression (Charité et al., 1998; Shashikant et al., 1995). (2) cdx gene products have been found to stimulate Hox gene expression (Charité et al., 1998; Epstein et al., 1997; Isaacs et al., 1998 and Subramanian et al., 1995). (3) Knockout of cdx-1 results in posterior shifts of several Hox genes within mesoderm, together with accompanying homeotic mutations in the resultant vertebral structures (Subramanian et al., 1995). Knockout of cdx-2 also results in posterior homeotic changes (Chawengsaksophak et al., 1997).

The way in which cdx proteins might regulate Hox patterns is unclear. Two simple models have been proposed in an attempt to explain how Hox expression patterns become established. The ‘timing’ model (Dollé et al., 1989), based upon the progress zone model ( Summerbell et al., 1973), proposes that Hox genes become sequentially activated, 3′–5′, within cells that remain in the founder zone, and that once cells move out of this zone their Hox expression patterns become fixed. Here, time of exposure to a Hox inducer is the important factor in setting up the patterns. In this model, the Hox inducers could be the cdx proteins since these are expressed in greatest abundance in the posterior founder zones of the embryo. Alternative ‘morphogen gradient’ models (e.g. Gaunt et al., 1999 and Gaunt, 2000) propose that embryos contain a graded concentration of a Hox inducer, and that different Hox genes become activated at different threshold concentrations. Here, concentration of a Hox inducer is a key factor in determining pattern. In such models, the cdx proteins could be the graded morphogen, since several authors have reported that both cdx mRNAs (Gamer and Wright, 1993; Marom et al., 1997 and Morales et al., 1996) and proteins ( Gamer and Wright, 1993 and Meyer and Gruss, 1993) are expressed in gradients along the embryo: cdx-A (in chick; cdx-1 in mouse) being most anterior, and cdx-B (cdx-4) most posterior.

Two lines of evidence suggest that vertebrate cdx proteins might indeed function as morphogen gradients. The first is by analogy with Drosophila, where caudal (cad) protein forms a morphogen gradient in the syncytial blastoderm (Rivera-Pomar et al., 1995). The cad gradient has its highest concentration at the posterior end. It sets the position of the knirps expression stripe, which is activated only within a threshold range of cad concentration. A second line of evidence lies in experiments reported by Charité et al., 1995 and Charité et al., 1998. These authors showed that multimerization of the enhancer within a Hoxb-8/lacZ transgene results in a forward shift in its expression. A plausible interpretation is that multiple copies of the enhancer results in the construct becoming more sensitive to an embryonic inducer of Hox genes, and that this inducer must normally be present at higher amounts in prospective posterior cells than anterior cells. Function of the enhancer depends upon an intact cdx binding motif (Charité et al., 1998). These observations, together with the graded expression pattern of cdx, suggest that caudal proteins may form morphogen gradients for the establishment of Hox expression boundaries. An enhancer multimerization experiment was similarly used in Drosophila to show the significance of the bicoid morphogen gradient (Struhl et al., 1989).

In the present paper, we describe both chick cdx-A/lacZ and mouse cdx-1/lacZ reporter constructs that are graded in their expression along both neurectoderm and mesoderm of transgenic mouse embryos. Chick cdx-A is the orthologue of mouse cdx-1 (Marom et al., 1997). The expression patterns appear to be very similar to that of endogenous mouse cdx-1 protein ( Meyer and Gruss, 1993). The initial objective was to provide clearer evidence for cdx expression gradients along the embryo. The lacZ technique, although indirect, provides a simple and highly reproducible method of detecting gene expression, uncomplicated by the background labelling that is associated with in situ hybridization or antibody staining. Our time course studies suggest that the gradient forms by a time-dependent decay of cdx activity in cells that have left the posterior founder zones of the embryo.

We note that an upstream T cell factor (Tcf)/β-catenin binding motif in mouse, responsive to Wnt signalling in cell culture assays (Lickert et al., 2000 and Prinos et al., 2001), is conserved in the chick. This upstream region taken from mouse DNA is able to generate a pattern of lacZ reporter expression that is similar to that of endogenous cdx-1 (Lickert and Kemler, 2002). However, this upstream region taken from chick cdx-A was insufficient to generate lacZ expression in ectoderm and mesoderm of early mouse embryos, and we found it necessary to include a second enhancer located at the 5′ end of the cdx-A first intron. The intron enhancer contains motifs, conserved between chick and mouse, for both a Tcf/β-catenin binding site and a DR2-type retinoic acid response element (RARE). Mutation of either the RARE or the Tcf/β-catenin within the cdx-A/lacZ reporter caused posterior shift in lacZ expression.

2. Results

2.1. cdx-A/lacZ reporter is expressed in gradients in transgenic embryos

The principal cdx-A/lacZ reporter construct that we used in order to obtain a cdx-like pattern of lacZ expression (construct 1, Fig. 1B; Fig. 2 and Fig. 5) includes 1.4 kb of cdx-A upstream sequence, the first exon, the first 2.1 kb of the first intron, and the first part of the homeobox in exon 2. A second construct (construct 2; Fig. 1C) resembles construct 1 but includes an additional 4.3 kb of sequence from the cdx-A first intron. Construct 2 was expressed with an apparently identical pattern to construct 1 (not shown).

2.2. cdx-A/lacZ reporters require an intron element for cdx-like expression in transgenic embryos

As a first step towards locating the cdx-A enhancer element within constructs 1 and 2, we prepared a new construct (construct 3; Fig. 1D) which contains the same upstream non-coding region as construct 1, but only part of the first exon, and non of the cdx-A intron. The intron used instead in construct 3 was the second intron from human β-globin. Four transgenic embryos were generated. One of these expressed lacZ, but expression was confined to a few cells in the posterior embryo (Fig. 3A). Sections showed that these were located in the hindgut, and that no labelled cells were present within neurectoderm or mesoderm (Fig. 3A′). An additional construct, construct 4 ( Fig. 1E), resembles construct 3 but includes an additional 6 kb of cdx-A upstream sequence. Of eight independent transgenic embryos produced with this construct, non expressed lacZ (not shown).

2.3. Conserved Tcf/β-catenin binding sites and a RARE within the cdx-A reporter construct

We sequenced the 4 kb cdx-A gene component of the reporter construct 1, and have submitted this to the EMBL database (accession no. AJ534305). When this is compared, over non-coding regions, with mouse cdx-1 (Hu et al., 1993) only two substantial regions of conservation are found ( Fig. 4A, B). Both contain conserved transcription factor binding sites and are therefore putative enhancer elements. The positions of these are indicated by white boxes in Fig. 1.

The second conserved element is located within the intron (Fig. 4B). We found that this includes complementary sequences to both a Tcf/β-catenin binding site and a DR2-type RARE. Since we observed a cdx-like pattern of lacZ expression only in constructs that contain intron sequences, we considered that this is likely to be an essential element in our reporter construct. It is already known that mouse cdx-1 is stimulated by both Wnt proteins, which use Tcf/β-catenin as a second messenger (Lickert et al., 2000 and Prinos et al., 2001), and by retinoic acid ( Houle et al., 2000 and Prinos et al., 2001).

To test the role of the intron RARE, we produced a new construct (construct 5; Fig. 1F) in which the RARE of construct 1 is mutated to disrupt all of the consensus sequence ( Fig. 1F; Chambon, 1994; legend to Fig. 4). Of six 8.7-day transient transgenic embryos generated with construct 5, three were seen to express lacZ. In each of these, expression was detected only in the neural tube (Fig. 3B, C), and in comparison to the expression of construct 1 (e.g. Fig. 2A–C), was confined mainly to posterior parts. Thus, while construct 1 is expressed in neural tube up to the level of somite 1, construct 5 is expressed mainly posterior to the level of somite 7.

To test the role of the intron Tcf/β-catenin binding motif, we produced construct 6 (Fig. 1G) in which this site is mutated to destroy the consensus sequence ( Lickert et al., 2000; legend to Fig. 4). Construct 6 is otherwise identical to construct 1. Of three transient transgenic 8.7-day embryos produced with construct 6, two expressed lacZ ( Fig. 3D). In both, staining was confined to a minority of cells in the posterior neurectoderm. No labelled cells were seen anterior to the level of somite 13, or in the mesoderm.

These findings demonstrate functional roles for both the RARE and the Tcf/β-catenin binding motif in the conserved intron enhancer of chick cdx-A. Since both motifs are conserved in the mouse, it seems likely that they play a similar role in this species.

2.4. Formation of the gradients detected by chick cdx-A/lacZ expression

Since the range of Hox genes activated in any cell may depend upon the dosage of cdx (e.g. Charité et al., 1998), we considered that the cdx gradient might be responsible for the establishment of Hox expression patterns along the embryo, and we attempted to elucidate how the gradient becomes formed.

During vertebrate development, newly organized neurectoderm and mesoderm tissues emerge from posterior founder regions. These are apparently the same regions as those that show high levels of cdx-A/lacZ activity (Fig. 5). Two simple mechanisms could establish the cdx gradient along the embryo. First, there may be a progressive increase in the concentration of cdx in both the posterior region and the newly emergent cells, thereby generating a concentration gradient along the embryo. Second, there may be a constant concentration of cdx in posterior parts, but cells that emerge from here may be less stimulated to express cdx, and so a gradient might be formed by progressive decay in cdx activity levels. To distinguish between these alternatives, it seemed important to test whether or not the cdx concentration displays a time-dependent increase over the period of gastrulation, from 7.5 to 9 days. To permit comparison of expression levels at different embryo stages, all the embryos tested in this section (Fig. 5) were from the same (high expressing) transgenic line, and all embryos were stained for 40 min under identical conditions.

Over the following 1.5 days (Fig. 5A–I), the lacZ-expressing tissues elongate as new neural and mesoderm tissue leaves the posterior founder zones. The neural plate begins to roll up at 8.25 days to reveal, below, an apparent gradient in lacZ staining within the underlying mesoderm (Fig. 5E). Over the period 7.5–8.25 days ( Fig. 5A–F), there is no detectable increase in the maximal levels of cdx-A/lacZ activity, and indeed, levels become reduced over the period 8.25–9 days (Fig. 5E–I). This reduction is apparently more rapid in mesoderm than in neurectoderm. Graded expression in these layers does not therefore arise by progressive increase in the cdx-A/lacZ activity within the posterior founder tissues. We therefore suggest that the gradients, which become apparent in neurectoderm and mesoderm over the period 8.25–8.7 days ( Fig. 5E–H), are more likely due to a time-dependent decay of cdx-A/lacZ (and cdx-1) activity in cells that have left the posterior founder zone. This conclusion is supported by the observation that at 8 days, there is little evidence of graded expression along either neurectoderm or mesoderm, and lacZ expression remains strong even in anteriormost parts of the expression domains ( Fig. 5D). By 8.5–8.7 days, however, anteriormost expression has weakened relative to posterior expression, thereby generating the gradients (cf. Fig. 5D with Fig. 2 and Fig. 5). As the domain of cdx-A/lacZ activity in neural tube decays and recedes along the embryo, it is seen that activity tends, over the period 8.7–9 days ( Fig. 5H, I), to remain as a more persistent domain within the region adjacent to somites 3–5.

2.5. Formation of the gradients detected by expression of a mouse cdx-1/lacZ transgene

To test whether a mouse cdx-1/lacZ transgene establishes an expression gradient in a similar way, we prepared a construct that utilizes mouse cdx-1 upstream and intron gene sequences (construct 7; Fig. 1H). Two independent lines of transgenic mice gave similar patterns of expression for construct 7, and the most strongly expressing line is shown in Fig. 6.

The results provide further evidence that the gradients arise by time-dependent decay of cdx/lacZ activity as cells move away from the vicinity of the node. Thus, somite 5 at 8 days is labelled as strongly as more posterior parts of the paraxial mesoderm (Fig. 6B), yet somite 5 at 8.7 days is only weakly labelled ( Fig. 6D).

A notable difference between chick and mouse constructs is that chick cdx-A/lacZ activity in somites declines more rapidly than does mouse cdx-1/lacZ activity. Thus, although the ratio of somite to neural staining intensity is similar for both constructs at 8 days (Fig. 5 and Fig. 6), this ratio at 8.7 days is greatly reduced for cdx-A/lacZ activity ( Fig. 2 and Fig. 5), though not for cdx-1/lacZ ( Fig. 6D). Somite expression of cdx-A/lacZ at 8.7 days is detected only after prolonged staining ( Fig. 2B).

2.6. cdx-A/lacZ and cdx-1/lacZ activity in 9–10.5 day embryos

By 9 days, rhombomeres have become distinct within the hindbrain, permitting localization of the anterior boundary of cdx-A/lacZ activity to the level of the junction between rhombomeres 7 and 8 (Fig. 7A). Many authors (e.g. Lumsden, 1990) regard this position as lying within the hindbrain, with the junction between the hindbrain and spinal cord lying adjacent to the junction of the last occipital and first cervical somites (that is, somites 4 and 5; large arrow in Fig. 7A). The anterior boundary of cdx-1/lacZ expression is similar to that of cdx-A/lacZ activity (level of somite 1; Fig. 7D), but cdx-1/lacZ activity appears to decay more rapidly in anterior neural tissue.

The forelimb bud at 9.5 days shows widespread labelling within both mesoderm and the apical ectodermal ridge (AER) (Fig. 7B for chick and Fig. 7E, E′ for mouse constructs). By 10.5 days ( Fig. 7C, C′, F, F′) forelimb expression becomes more focussed within the AER. Occasional labelled cells are also detected in dorsal, but not ventral, forelimb bud ectoderm ( Fig. 7C′). Weaker expression at this stage is seen in the AER of the hindlimb bud and within the tailbud ( Fig. 7C, F).

At 9 days, lacZ-expressing cells are seen to be streaming laterally from the neural tube (Fig. 7A, A′). These cells are apparently the same as those detected earlier using the anti-cdx-1 antibody ( Meyer and Gruss, 1993). We believe that they are neural crest cells, since later, by 9.5 days, they are organizing to form the cervical trunks and the spinal ganglia ( Fig. 7B, E, E′). As already noted for somite expression, expression in these neural structures persists longer for the mouse construct than the chick ( Fig. 7C, F).

2.7. cdx-1 and cdx-1/lacZ expression compared by in situ hybridization

Nearby sections of cdx-1/lacZ transgenic embryos were hybridized either to a probe for detection of endogenous cdx-1 mRNA (not reactive with transgene), or else to a probe for detection of lacZ mRNA (transgene). At 8 days (Fig. 8A–C), cdx-1 and lacZ mRNAs are expressed with the same anterior boundaries, both within mesoderm and neurectoderm, along the P–A axis. These boundaries are apparently the same as those for cdx/lacZ staining (Fig. 5 and Fig. 6). However, unlike the results obtained for lacZ protein activity at 8 days ( Fig. 5 and Fig. 6), the anterior regions of mRNA expression are, most notably in neurectoderm, less intense than more posterior regions. In terms of the time-dependent decay model considered above, this suggests that the lacZ (and cdx-1) mRNA may decay more rapidly than the lacZ protein.