Studies of pattern formation in the vertebrate central nervous system indicate that anteroposterior positional information is generated in the embryo by signalling gradients of an as yet unknown nature. We searched for transcription factors that transduce this information to the Hox genes. Based on the assumption that the activity levels of such factors might vary with position along the anteroposterior axis, we devised an in vivo assay to detect responsiveness of cis-acting sequences to such differentially active factors. We used this assay to analyze a Hoxb8 regulatory element, and detected the most pronounced response in a short stretch of DNA containing a cluster of potential CDX binding sites. We show that differentially expressed DNA binding proteins are present in gastrulating embryos that bind to these sites in vitro, that cdx gene products are among these, and that binding site mutations that abolish binding of these proteins completely destroy the ability of the regulatory element to drive regionally restricted expression in the embryo. Finally, we show that ectopic expression of cdx gene products anteriorizes expression of reporter transgenes driven by this regulatory element, as well as that of the endogenous Hoxb8 gene, in a manner that is consistent with them being essential transducers of positional information. These data suggest that, in contrast to Drosophila Caudal, vertebrate cdx gene products transduce positional information directly to the Hox genes, acting through CDX binding sites in their enhancers. This may represent the ancestral mode of action of caudal homologues, which are involved in anteroposterior patterning in organisms with widely divergent body plans and modes of development.

Clustered Hox genes are present in many, conceivably all, metazoa, sharing many conserved features of organization, function and expression. They encode key regulators of regional diversification, primarily of the main anteroposterior axis of the body, and, accordingly, are expressed in the embryo in regionally restricted patterns with distinct expression boundaries. Genes located 3′ within the clusters are generally expressed more anteriorly than more 5′ genes (reviewed by McGinnis and Krumlauf, 1992). Relatively little is known about how these expression patterns are generated in different organisms.

In the Drosophila embryo, positional information is created in the form of gradients of maternal gene products set up from the anterior and posterior poles. This information is interpreted by a regulatory network of segmentation gene products, which subdivide the anteroposterior axis into discrete domains (reviewed by Pankratz and Jäckle, 1993). Hox gene expression in Drosophila comprises two distinct phases. In the initiation phase, different Hox genes are activated by specific combinations of segmentation gene products. In the subsequent maintenance phase, when segmentation gene expression subsides, these preestablished domains of expression are stably inherited throughout subsequent cell divisions by the opposing actions of the products of genes of the Polycomb- and trithorax-groups. This maintenance mechanism appears to be substantially conserved in vertebrates, involving homologues of Polycomb- and trithorax-group genes (reviewed by Schumacher and Magnuson, 1997; Gould, 1997). However, Drosophila, with its highly specialized mode of embryogenesis, in which most of the early axial patterning is achieved by transcription factors diffusing in a syncytium, has provided little guidance to our understanding of how positional information is generated in the cellularized vertebrate embryo, in which patterning is accompanied by extensive growth and cell migration. Likewise, the vertebrate equivalent of the segmentation gene network of Drosophila remains unknown.

Much work in amphibians (reviewed by Doniach, 1993; Ruiz i Altaba, 1993) as well as recent work in chick and mouse (Grapin-Botton et al., 1995, 1997; Itasaki et al., 1996) suggests that Hox gene expression in the vertebrate neural tube is regulated by a gradient of signals travelling within the neuroepithelium as well as by signals from the adjacent paraxial mesoderm, and that the strength of these signals gradually diminishes from posterior to anterior. However, what is the nature of these signals, and how they are transmitted in the embryo and transduced to the Hox genes, has not been determined.

A hypothetical regulatory pathway could involve translation of this gradient of positional signalling into a gradient of activity of a certain transcription factor that binds to Hox gene regulatory elements, with activation of different Hox genes occurring at distinct threshold levels of activity. The existence of such a transciption factor gradient was inferred from our studies on the regulation of Hoxb7 and Hoxb8: we identified several regulatory elements which confer overlapping but distinct, regionally restricted expression patterns upon the respective promoters, and which cooperate in setting the anterior boundaries of expression, transgenes containing more elements driving more anterior expression (Vogels et al., 1993; Charité et al., 1995). These observations suggested that the presence of multiple elements increases the sensitivity of the promoter towards a trans-acting factor necessary for activation, the expression or activity of which is differentially regulated along the anteroposterior axis. This factor thus seemed likely to be involved in the transduction of positional signals. Recently, the products of the vertebrate cdx genes, homologues of the caudal gene of Drosophila, have emerged as candidates for such a factor: like caudal, they are expressed in the embryo in gradients (e.g. Meyer and Gruss, 1993; Gamer and Wright, 1993; Beck et al., 1995), genetic (Subramanian et al., 1995; Chawengsaksophak et al., 1997) and other (Pownall et al., 1996; Epstein et al., 1997; Isaacs et al., 1998) experiments indicate that they are upstream of Hox genes, and potential CDX binding sites are present in Hox regulatory regions (Shashikant et al., 1995; Subramanian et al., 1995).

We have exploited the synergistic interaction between regulatory elements as an in vivo assay to functionally identify cis-acting sequences within a Hoxb8 regulatory element that respond to this putative differentially active factor, and present evidence that cdx gene products are the key effectors acting on these sequences and that CDX expression levels determine Hox gene expression boundaries.

Reporter and expression constructs

Reporter constructs 1, 2, 3 and 16 have been described previously (Charité et al., 1995). For constructs 4-15, fragments of BH1100, usually with some polylinker sequences, were cloned, singly or as head-to-tail multimers (retaining their original orientation) at the 5′ end of construct 1 or 2. Mutant versions were generated by PCR-based procedures. Plasmid sequences were removed prior to microinjection.

For constructs pSG5-cdx1, pSG5-cdx2 and pSG5-cdx4, a BstEII-BstYI cdx-1 cDNA fragment including 6 bp upstream and 27 bp downstream of the coding region, a HindIII-HindIII cdx-2 cDNA fragment (Beck et al., 1995), or a MaeIII-MaeIII cdx-4 cDNA fragment (Gamer and Wright, 1993), respectively, were cloned into the pSG5 expression vector (Stratagene).

To generate RARβ-cdx1, -2 and -4, the 350 bp SalI-StuI fragment containing the SV40 early promoter was removed from the respective pSG5 constructs and replaced with a 3.5 kb XhoI-BamH1 mouse RARβ2 promoter fragment (Charité et al., 1994). These constructs were linearized with SalI prior to microinjection.

Generation and analysis of transgenic embryos

Transgenic embryos carrying lacZ reporter constructs were generated and analyzed as previously described (Charité et al., 1995); however, embryos up to 8.75 days old were embedded in plastic, sectioned at 4 μm and counterstained with Nuclear Fast Red, except for the embryo shown in Fig. 4H-K, which was embedded in paraffin, cut at 10 μm and counterstained with Neutral Red.

Cdx expression constructs were injected into zygotes obtained from crosses of C57Bl/6 × CBA F1 females with males of transgenic lines 65 and 59, which carry one to a few copies (unpublished observations) of constructs 1 and 3, respectively, as randomly integrated transgenes. Transgenic founder embryos were recovered from foster mothers, genotyped by PCR analysis of yolk sac DNA, and analyzed as described above.

For combined β-galactosidase staining and whole-mount in situ hybridization, embryos were fixed for 2 hours at 4°C in 4% paraformaldehyde, rinsed with PBS and Xgal-stained for 1 hour, rinsed, postfixed in 4% paraformaldehyde for 24 hours, and processed for subsequent in situ hybridization, which was done according to Wilkinson (1992), with modifications (available on request). For detecting Hoxb8, a mixture of KpnI-SstI (Deschamps and Wijgerde, 1993) and SstI-SstI (Charité et al., 1994) probes was used.

Whole cell extracts

COS cells were transfected with expression constructs using the calcium phosphate procedure. 48 hours after transfection cells were washed with TBS (25 mM Tris-HCl, pH 7.4, 137 mM NaCl, 5 mM KCl, 0.7 mM CaCl2, 0.5 mM MgCl2, 0.6 mM Na2HPO4), scraped into TBS, pelleted, and resuspended in 3 volumes of WCE buffer (20 mM Hepes, pH 7.9, 400 mM KCl, 1 mM EDTA, 10% glycerol, 0.5 mM DTT, 1 μg/ml leupeptin, 2 μM pepstatin, 0.1 mM PMSF). Cells were disrupted by two freeze-thaw cycles and cell debris was removed by centrifugation.

8.5-day C57Bl/6 × CBA F2 embryos were dissected in TBS into anterior (head, including the first branchial arch), middle (from the second branchial arch to about the second most recently formed somite; for embryos older than about 6 somites heart and viscera were removed) and posterior (the rest of the embryo except the allantois) parts. Fragments from several embryos were pooled and extracts were prepared as described above.

Electrophoretic mobility shift assays

The EcMs79 probe fragment, comprising the EcoRI-MseI fragment from BH1100 and 8 bp of polylinker, was end-labeled to a specific activity of 107-108 cpm/μg, and purified from a polyacrylamide gel. Binding reactions, containing 2 ng of probe, 1.5 μl whole cell extract and 1 μg of poly(dI-dC) in 15 μl of 20 mM Hepes, pH 7.9, 60 mM KCl, 1 mM MgCl2, 0.5 mM DTT and 10% glycerol, were equilibrated for 10 minutes at room temperature and separated at 4°C on a 6% polyacrylamide (30:1) gel containing 0.5× TBE and 2.5% glycerol. For antibody supershifts, 0.5 μl of anti-CDX antibody (Gamer and Wright, 1993; Meyer and Gruss, 1993), diluted as indicated, was added after equilibration and incubation was continued for 20 minutes on ice.

Dissection of the BH1100 element

We have previously defined a 1.1 kb regulatory element (BH1100, see Fig. 1A) capable of driving robust Hox-like expression (Fig. 2C) when coupled to the 1-kb Hoxb8 promoter-containing region (construct 2, Fig. 1A), which does not drive regionally restricted expression by itself (Charité et al., 1995). Adding extra copies of BH1100 to a 4.4 kb reporter construct (construct 1, Fig. 1A), which includes BH1100 as well as some other region-specific elements, compensates for the absence of more upstream and downstream elements, anteriorizing the expression boundaries generated by this construct by several segments (Charité et al., 1995; construct 3, Fig. 1A,B). To identify the cis-acting sequences responsible for this effect, fragments of BH1100 were similarly multimerized, coupled to construct 1, and analyzed for their effect on the expression boundaries. Several 11.5-day F0 embryos were obtained for each construct, and their expression patterns were classified according to previously established criteria (see Table 1). For 1-4 embryos showing what appeared to be the maximal expression domain attainable by each construct, the exact expression boundaries in neural tube, spinal ganglia and prevertebrae were determined from histological sections; as in our previous studies, these varied somewhat within and between embryos. The most anterior boundary reached in each of the three tissues was then compared with the most anterior boundary reached by construct 1 (Fig. 1B). By testing progressively smaller fragments, proceeding each time with the fragment that had the greatest effect, we converged on a 79-bp EcoRI-MseI fragment (hereafter referred to as EcMs79), which strongly anteriorized expression in the neurectoderm (see Fig. 1B, construct 10; compare with C, construct 1, and D, construct 10; see also E,F). Adding an additional three copies of this fragment somewhat increased this effect (Fig. 1B, construct 12). Surprisingly, the expression boundaries in the paraxial mesoderm generated by constructs 10 and 12 were more posterior than the expression boundary of construct 1 (Fig. 1B,F).

Table 1.

Overview of expression patterns obtained with different reporter constructs

Overview of expression patterns obtained with different reporter constructs
Overview of expression patterns obtained with different reporter constructs
Fig. 1.

Identification of sequences responsive to positional information. (A) Organization of the Hoxb cluster surrounding Hoxb8, showing transcripts and coding regions. Regions underlined by black bars (BH1100 indicated) have been shown to drive regionally restricted expression with a Hox promoter (Vogels et al., 1993; Charité et al., 1995). Reporter constructs (1-16) are schematized underneath, aligned with the genomic map and showing the positions of the lacZ coding region (black triangle) and insertion of additional regulatory sequences (white arrowhead; see Table 1). (B) Deletion analysis of BH1100. BH1100 is depicted at the top, with restriction sites used to dissect it; markers indicate positions of TTTAT(A/T/G) motifs: closed circles, TTTATA; open circles, TTTATT; diamonds, TTTATG; markers with cross-bars, related motifs, identified using the MatInspector program (Quandt et al., 1995), which have a similarity greater than 0.85 to the sequence matrices based on the CdxA binding data of Margalit et al. (1993). Fragments were cut into two parts, as indicated by the arrowheads, and both parts were individually multimerized as indicated on the left (3×, 6×) and inserted at the 5′ end of construct 1 (bottom). The expression boundaries for the resulting constructs (construct numbers in circles) were determined in 1-4 embryos with expression patterns classified as maximal (see Table 1), according to criteria described earlier (Charité et al., 1995). The difference between the most anterior of these boundaries and those generated by construct 1 is indicated by the black bar graphs on each restriction fragment, for neural tube (NT), spinal ganglia (SG) and mesoderm (M), bars extending upwards indicate anteriorization (arrow A), bars pointing downwards (arrow P) posteriorization relative to the construct 1 boundaries. Grids are scaled in metameric units (distance between two prevertebrae or spinal ganglia). Inset, data for construct 10m, in which the extra copies of EcMs79 have all TTTAT(A/T) motifs mutated to CCCAT(A/T). Data for constructs 1 and 3 were taken from Charité et al. (1995). Abbreviations of restriction sites in A and B: Av, AvaI; Ba, BamHI; Bs, BstNI; Ec, EcoRI; H2, HindII; H3, HindIII; Kp, KpnI; Ms, MseI; Pv, PvuII; Sa, SalI. (C,D,G) 11.5-day embryos containing constructs 1, 10 and 10m, respectively. Arrows indicate the anterior boundary of expression in the neurectoderm. (E,F) Parasagittal sections through the embryo shown in D. (E) Expression in spinal ganglia; the expression boundary, not visible in the section, is anterior of C2 (indicated); green arrow indicates the most anterior boundary reached by construct 1. (F) Expression boundaries in neural tube (white arrowhead), at the level of the basioccipital anlage (bo), and prevertebrae, with the most anterior expression in pv14 (white arrow). Compare with construct 1 boundaries, green arrowhead and arrow, respectively. (H,I) Parasagittal sections of embryo shown in G. (H) Expression boundaries in spinal ganglia (in C5, white arrowhead) and prevertebrae (white arrow, in pv11), which are almost identical to the construct 1 boundaries (green arrow and arrowhead). (I) Expression boundary in the neural tube (white arrowhead), at the level of pv4. Green arrowhead indicates construct 1 boundary. In E, F, H and I, anterior is to the top, ventral to the left. Bars, 2 mm (C,D,G); 1 mm (E,F,H,I). Construct numbers are indicated in upper right corner. (J) Nucleotide sequence of EcMs79. EcoRI and MseI sites are underlined, and TTTAT(A/T) motifs, sites A to D, are shown bold and underlined.

Fig. 1.

Identification of sequences responsive to positional information. (A) Organization of the Hoxb cluster surrounding Hoxb8, showing transcripts and coding regions. Regions underlined by black bars (BH1100 indicated) have been shown to drive regionally restricted expression with a Hox promoter (Vogels et al., 1993; Charité et al., 1995). Reporter constructs (1-16) are schematized underneath, aligned with the genomic map and showing the positions of the lacZ coding region (black triangle) and insertion of additional regulatory sequences (white arrowhead; see Table 1). (B) Deletion analysis of BH1100. BH1100 is depicted at the top, with restriction sites used to dissect it; markers indicate positions of TTTAT(A/T/G) motifs: closed circles, TTTATA; open circles, TTTATT; diamonds, TTTATG; markers with cross-bars, related motifs, identified using the MatInspector program (Quandt et al., 1995), which have a similarity greater than 0.85 to the sequence matrices based on the CdxA binding data of Margalit et al. (1993). Fragments were cut into two parts, as indicated by the arrowheads, and both parts were individually multimerized as indicated on the left (3×, 6×) and inserted at the 5′ end of construct 1 (bottom). The expression boundaries for the resulting constructs (construct numbers in circles) were determined in 1-4 embryos with expression patterns classified as maximal (see Table 1), according to criteria described earlier (Charité et al., 1995). The difference between the most anterior of these boundaries and those generated by construct 1 is indicated by the black bar graphs on each restriction fragment, for neural tube (NT), spinal ganglia (SG) and mesoderm (M), bars extending upwards indicate anteriorization (arrow A), bars pointing downwards (arrow P) posteriorization relative to the construct 1 boundaries. Grids are scaled in metameric units (distance between two prevertebrae or spinal ganglia). Inset, data for construct 10m, in which the extra copies of EcMs79 have all TTTAT(A/T) motifs mutated to CCCAT(A/T). Data for constructs 1 and 3 were taken from Charité et al. (1995). Abbreviations of restriction sites in A and B: Av, AvaI; Ba, BamHI; Bs, BstNI; Ec, EcoRI; H2, HindII; H3, HindIII; Kp, KpnI; Ms, MseI; Pv, PvuII; Sa, SalI. (C,D,G) 11.5-day embryos containing constructs 1, 10 and 10m, respectively. Arrows indicate the anterior boundary of expression in the neurectoderm. (E,F) Parasagittal sections through the embryo shown in D. (E) Expression in spinal ganglia; the expression boundary, not visible in the section, is anterior of C2 (indicated); green arrow indicates the most anterior boundary reached by construct 1. (F) Expression boundaries in neural tube (white arrowhead), at the level of the basioccipital anlage (bo), and prevertebrae, with the most anterior expression in pv14 (white arrow). Compare with construct 1 boundaries, green arrowhead and arrow, respectively. (H,I) Parasagittal sections of embryo shown in G. (H) Expression boundaries in spinal ganglia (in C5, white arrowhead) and prevertebrae (white arrow, in pv11), which are almost identical to the construct 1 boundaries (green arrow and arrowhead). (I) Expression boundary in the neural tube (white arrowhead), at the level of pv4. Green arrowhead indicates construct 1 boundary. In E, F, H and I, anterior is to the top, ventral to the left. Bars, 2 mm (C,D,G); 1 mm (E,F,H,I). Construct numbers are indicated in upper right corner. (J) Nucleotide sequence of EcMs79. EcoRI and MseI sites are underlined, and TTTAT(A/T) motifs, sites A to D, are shown bold and underlined.

Fig. 2.

Autonomous function of regulatory elements. (A) 11.5-day transgenic embryo containing construct 14, showing regionally restricted expression in the neurectoderm starting at the level of C10 (white arrowhead), in addition to several ectopic sites of expression (black arrowheads), including the ventral part of the neural tube, in which expression extends into the hindbrain. (B) 11.5-day embryo containing construct 15, showing anterior boundary of expression in the neurectoderm at the level of C2 (arrowhead); pv19 was the most anterior prevertebra that expressed lacZ (not shown). (C) 11.5-day embryo containing construct 16, showing a Hox-like expression pattern (white arrowhead indicates anterior boundary in neurectoderm, at the level of C4) and characteristic expression in cranial sensory ganglia (black arrowheads) (D) Embryo containing construct 16m, which retains expression in cranial ganglia (arrowheads) but lacks regionally restricted expression; there is expression in the neural tube (arrow), essentially limited to the ventral part and extending into the hindbrain, weak expression in all spinal ganglia, and no expression in the mesoderm. Bars, 2 mm.

Fig. 2.

Autonomous function of regulatory elements. (A) 11.5-day transgenic embryo containing construct 14, showing regionally restricted expression in the neurectoderm starting at the level of C10 (white arrowhead), in addition to several ectopic sites of expression (black arrowheads), including the ventral part of the neural tube, in which expression extends into the hindbrain. (B) 11.5-day embryo containing construct 15, showing anterior boundary of expression in the neurectoderm at the level of C2 (arrowhead); pv19 was the most anterior prevertebra that expressed lacZ (not shown). (C) 11.5-day embryo containing construct 16, showing a Hox-like expression pattern (white arrowhead indicates anterior boundary in neurectoderm, at the level of C4) and characteristic expression in cranial sensory ganglia (black arrowheads) (D) Embryo containing construct 16m, which retains expression in cranial ganglia (arrowheads) but lacks regionally restricted expression; there is expression in the neural tube (arrow), essentially limited to the ventral part and extending into the hindbrain, weak expression in all spinal ganglia, and no expression in the mesoderm. Bars, 2 mm.

While expression boundaries were routinely assayed at 11.5 days, the difference in neural tube expression boundaries driven by constructs 3 and 10 versus that of construct 1 was evident at least as early as 8.5 days (7-10 somites; data not shown).

Autonomous function of EcMs79

To determine which aspects of the function of BH1100 are retained in EcMs79, it was coupled to construct 2, yielding construct 14. BH1100, when tested in this configuration, drives Hox-like expression (see Fig. 2C) with boundaries at the level of C4 in the neurectoderm, and pv10 in paraxial mesoderm (Charité et al., 1995). In contrast, embryos containing construct 14 showed expression in the neurectoderm only, in a relatively posterior domain lacking a sharp anterior boundary (Fig. 2A). While it is possible that this does not represent the maximal expression possible for this construct, it is not likely that the expression domain can be significantly more extensive, since the EcoRI-PvuII fragment in which EcMs79 is contained (see Fig. 1B), when coupled to construct 2, generated a well-defined, but only slightly more anterior expression boundary in the neural tube (at the C9 level; data not shown). When tested as a trimer (construct 15, see Fig. 1A) EcMs79 appeared more robust and was able to drive much more anterior expression in the neurectoderm, with a C2 boundary (Fig. 2B), as well as expression in the paraxial mesoderm, though with a relatively posterior boundary, at pv19 (data not shown).

Thus, EcMs79 responds to positional information and, like BH1100, can activate the Hoxb8 promoter in a region-specific manner, although in a much more limited domain, independently of the other regulatory elements present in construct 1.

Differentially expressed DNA binding proteins interact with EcMs79

We investigated whether embryos contained any detectable trans-acting factors that could be involved in transducing positional information to EcMs79. Since such factors might not persist in the embryo once a stable state of expression maintained by the Polycomb and trithorax-group proteins has been established, extracts were prepared from 8.5-day embryos (0-10 somites; essentially similar results were obtained with 7-

16 somite embryos); around this time Hoxb8 reaches its expression boundaries (Deschamps and Wijgerde, 1993), and several of the more 5′ Hox genes have yet to be expressed. To reveal differential expression of proteins along the anteroposterior axis, embryos were dissected into anterior (A), middle (M) and posterior (P) fragments as depicted in Fig. 3A. Electrophoretic mobility shift assays with these extracts and EcMs79 as probe revealed the formation of several complexes (Fig. 3B, lanes 1-3), some of which are differentially distributed along the embryonic axis: a slowly migrating complex S (Fig. 3B, lanes 1-3), which exhibits a shallow gradient throughout the embryo, and a number of faster-migrating complexes that show a steeper gradient of distribution: they are prominent in the P-extract (Fig. 3B, lane 3, P1 and P2) but undetectable in the A-extract, and several of them are detected at lower levels in the M-extract (Fig. 3B, lane 2, M). Given the anteroposterior differences in their expression, these proteins were candidate transducers of positional information, and we determined the sequences involved in their binding to EcMs79.

Fig. 3.

Detection of proteins binding to EcMs79. (A) Schematic representation of an 8.5-day embryo, showing dissection into anterior (A), middle (M) and posterior (P) fractions. (B) Electrophoretic mobility shift assay (EMSA) using embryo extracts A, M and P, as indicated. The probe for lanes 1-3 was the wild-type (wt) EcMs79, with sites A-D intact (ABCD), and the probe for lanes 4-6 was a mutant version in which all TTTAT(A/T) motifs were mutated to CCCAT(A/T) (xxxx). Differentially expressed complexes (S, M, P1, P2) are indicated; see main text for further description. (C) EMSA with embryo extracts A, M, P and wt EcMs79 (lanes 1-3), or mutant versions of EcMS79 that have one of the four TTTAT(A/T) motifs intact and the others mutated to CCCAT(A/T) (Axxx etc.; lanes 4-15), or a mutant version in which all four motifs have been changed to TTTATG (lanes 16-18). EMSA, using EcMs79 as probe, with embryo extracts (lanes 1-3) or extracts from COS cells (lanes 4-11), either untransfected (0), or transfected with empty pSG5 expression vector (SG), pSG5-cdx1 (c1) or pSG5-cdx4 (c4). Lanes 8 and 9, supershift with CDX1 antibody (a-C1), 10× diluted and undiluted, respectively. Lanes 10 and 11, supershift with CDX4 antibody (a-C4), 10× diluted and undiluted. Complexes in lanes 6 and 7 are numbered 1 to 4, as indicated between the two lanes. The asterisk refers to lane 7. EMSA with embryo extract P, using EcMs79 as probe. Lanes 2-5, antibody supershifts as in D, lanes 8-11. P-specific complexes which are supershifted are indicated by arrows (2,1,*).

Fig. 3.

Detection of proteins binding to EcMs79. (A) Schematic representation of an 8.5-day embryo, showing dissection into anterior (A), middle (M) and posterior (P) fractions. (B) Electrophoretic mobility shift assay (EMSA) using embryo extracts A, M and P, as indicated. The probe for lanes 1-3 was the wild-type (wt) EcMs79, with sites A-D intact (ABCD), and the probe for lanes 4-6 was a mutant version in which all TTTAT(A/T) motifs were mutated to CCCAT(A/T) (xxxx). Differentially expressed complexes (S, M, P1, P2) are indicated; see main text for further description. (C) EMSA with embryo extracts A, M, P and wt EcMs79 (lanes 1-3), or mutant versions of EcMS79 that have one of the four TTTAT(A/T) motifs intact and the others mutated to CCCAT(A/T) (Axxx etc.; lanes 4-15), or a mutant version in which all four motifs have been changed to TTTATG (lanes 16-18). EMSA, using EcMs79 as probe, with embryo extracts (lanes 1-3) or extracts from COS cells (lanes 4-11), either untransfected (0), or transfected with empty pSG5 expression vector (SG), pSG5-cdx1 (c1) or pSG5-cdx4 (c4). Lanes 8 and 9, supershift with CDX1 antibody (a-C1), 10× diluted and undiluted, respectively. Lanes 10 and 11, supershift with CDX4 antibody (a-C4), 10× diluted and undiluted. Complexes in lanes 6 and 7 are numbered 1 to 4, as indicated between the two lanes. The asterisk refers to lane 7. EMSA with embryo extract P, using EcMs79 as probe. Lanes 2-5, antibody supershifts as in D, lanes 8-11. P-specific complexes which are supershifted are indicated by arrows (2,1,*).

A potential binding site for any of these complexes was the sequence motif TTTAT(A/T), which occurs four times in EcMs79 (see Fig. 1J), and which resembles the prevalent binding site for the Caudal protein of Drosophila (Dearolf et al., 1989; Rivera-Pomar et al., 1995) and the in vitro binding sites determined for some of its vertebrate homologues (Margalit et al., 1993; Suh et al., 1994). Simultaneous mutation of all four TTTAT(A/T) motifs in EcMs79 to CCCAT(A/T) abolished formation of complex S as well as the P/M-specific complexes, leaving other complexes unaffected (Fig. 3B, lanes 4-6). When each of the four motifs was individually restored to its wild-type sequence, complexes P1 and M were formed, with different efficiencies, but complex S was not (Fig. 3C, lanes 4-15, compare to lanes 1-3). The diffuse complex P2 also did not appear to be formed with these mutant probes: although several different novel complexes were observed around the P2 position, most of these are not differentially expressed; some, however (Fig. 3C, lane 6), were differentially expressed and approximately comigrate with P2, though they are less diffuse. Mutation of all four motifs to TTTATG, a variation that occurs several times in BH1100 outside of EcMs79 (see Fig. 1B), essentially reproduced the wild-type EcMs79 pattern, with the exception of complex S (Fig. 3C, lanes 16–18).

Since all three cdx genes identified in the mouse are expressed at 8.5 days, predominantly in the posterior part of the embryo corresponding to the P-fraction in our dissections (Gamer and Wright, 1993; Meyer and Gruss, 1993; Beck et al., 1995), we determined whether any of the complexes formed with EcMs79 contain CDX proteins.

We first ascertained that CDX proteins can bind to EcMs79. Both CDX1 and CDX4 expressed in COS cells formed several complexes with EcMs79 (Fig. 3D, lanes 6 and 7, respectively), all of which could be supershifted by antibodies directed against the respective proteins (Fig. 3D, lanes 8-11). Complexes labeled 1-4 in Fig. 3D, lanes 6 and 7, in all likelihood represent one, two, three and four molecules of protein bound to the probe. Complexes 1 and 2 of both CDX1 and CDX4, as well as a minor complex of higher electrophoretic mobility (asterisk in Fig. 3D, lane 7), which might represent a differently modified form or a proteolytic product of CDX4, appeared to comigrate with P/M-specific complexes (Fig. 3D, lanes 2, 3). No complexes corresponding to complexes 3 and 4 were detectable with embryo extracts; we assume that under the conditions of the assay, formation of these higher-order complexes requires higher protein concentrations achieved only with the high expression levels reached in COS cells.

Antibody supershifts confirmed that the P-specific complexes formed with embryo extracts contain CDX proteins (Fig. 3E): although analysis of these results is somewhat complicated by the fact that CDX1 and CDX4, as well as CDX2, migrate close together (Fig. 3D, lanes 6 and 7, and data not shown), it is clear that CDX4 accounts for a major fraction of P-specific complexes (Fig. 3E, compare lanes 4, 5 with lane 1) and that CDX1 is present as well, as concluded from the smaller but reproducible decrease in intensity of the major, broad, P-specific band (arrow 1 in Fig. 3E), which presumably consists of several complexes, upon addition of CDX1 antibody (Fig. 3E, compare lanes 2 and 3 to lane 1). We have not yet conclusively identified CDX2 within these complexes, but we have determined that it can bind to EcMs79 and that this binding depends on the TTTAT(A/T) motifs (data not shown).

CDX binding sites are essential for function

Several mutant versions of EcMs79 were tested for their ability to influence the anterior expression boundaries in the multimerization assay. Mutation of all four TTTAT(A/T) motifs to CCCAT(A/T), which abolished binding of all differentially expressed proteins, almost completely abolished function of EcMs79 in this assay (Fig. 1B, construct 10m; G, compare to D; and H,I, compare to E,F). In contrast, mutation of all four motifs to TTTATG, which abolished formation of complex S but not P/M-specific complexes, appeared even to slightly increase the ability of EcMs79 to anteriorize neurectoderm expression (data not shown). Thus, formation of complex S does not appear to be relevant to the function of EcMs79 in vivo. We also tested versions of EcMs79 containing combinations of wild type and CCCAT(A/T) mutant motifs. Preliminary results (based on, for each construct, one or two embryos with expression patterns classified as maximal) indicate that one site (site A, see Fig. 1J) or two sites (A and C) are insufficient for function of EcMs79 in this assay, whereas EcMs79 with three intact sites (B, C and D) was at least partially functional (data not shown).

Introduction of all four CCCAT(A/T) mutations into BH1100 completely abolished its ability to autonomously drive regionally restricted, Hox-like expression from the Hoxb8 promoter (construct 16m, Table 1; Fig. 2D, compare to C), even though several other TTTAT(A/T/G) motifs are present elsewhere in BH1100 (see Fig. 1B).

Thus, CDX binding sites are essential for function of BH1100 as an autonomous regulator, as well as for that of EcMs79 in the, potentially more forgiving, multimerization assay.

Ectopic expression of cdx genes

At 8.5 days, all three CDX proteins exhibit a gradient of expression in the neural tube, and their expression domains form a nested set, overlapping in the posterior part and extending into the neural tube to different extents, CDX1 showing the most and CDX4 the least anterior expression (Gamer and Wright, 1993; Meyer and Gruss, 1993; Beck et al., 1995). If Hox gene expression boundaries are set in response to different levels of CDX activity, then increasing the level of CDX expression in the appropriate window of time, especially anteriorly, should anteriorize Hox gene expression.

A cdx4 cDNA was placed under the control of the mouse RARβ2 promoter (construct RARβ-cdx4), which would extend the cdx4 expression domain in the neural tube from just anterior of the unsegmented part of the embryo (Gamer and Wright, 1993) to well into the hindbrain (Charité et al., 1994; see Fig. 4A), and also drive expression in the somites (Charité et al., 1994), which normally do not express cdx4 (Gamer and Wright, 1993). Construct RARβ-cdx4 was introduced by pronuclear injection into zygotes derived from transgenic lines 65 and 59, which carry reporter constructs 1 and 3, respectively, as randomly integrated transgenes. Of seven embryos derived from line 65 that contained the RARβ-cdx4 construct, three showed a modest extension, of up to a few somite lengths, of the lacZ expression domain in the neural tube as compared to control littermates, although isolated patches of expression were present much more anteriorly (Fig. 4D). Expression in the somites appeared unchanged in these embryos.

In contrast, 16 of 40 8.5-8.75-day RARβ-cdx4-containing embryos derived from line 59 showed a large extension of the reporter gene expression domain in the neural tube (compare Fig. 4B and C, E and F), which in the most severe cases aproached the expression boundary of the RARβ2 promoter (compare Fig. 4A and C). Expression in the somites was also expanded in these embryos (Fig. 4G, compare to E). Overexpression of cdx1 and cdx2 in line 59 affected expression of the reporter gene in a similar manner (Fig. 4H-K, and data not shown).

Morphologically, overexpression of cdx genes frequently caused a mild malformation of the neural tube (Fig. 4F), and defects of neural tube closure were sometimes observed as well. In the most severely affected embryo, shown in Fig. 4H, the neural tube has failed to close and is extremely malformed, especially near the midline (Fig. 4K). In addition, the somites are irregular in shape and size, and several have fused together (Fig. 4I,J).

We subsequently determined the effect of cdx overexpression on the endogenous Hoxb8 gene. RARβ-cdx4 was microinjected into line-59 zygotes, and F0 embryos were stained very lightly with Xgal to reveal the effect on the reporter transgene, and then subjected to in situ hybridization with a Hoxb8 probe. Fig. 5A shows an embryo with a maximal extension of the lacZ expression domain; in situ hybridization revealed that expression of Hoxb8 is similarly antriorized, neural tube expression extending anteriorly of the normal boundary, at the level of somite 5, to the level of the otic pit (Fig. 5B). Expression in the somites was extended as well: histological sections revealed expression, though weak, in few cells in S7 through S9, and more ubiquitously in more posterior somites (data not shown; the normal expression boundary is somite 10/11).

Fig. 4.

Ectopic expression of cdx genes anteriorizes expression of Hoxb8 reporter transgenes. (A) 15-somite embryo showing lacZ expression driven by the RARβ2 promoter. Expression in the neural tube extends to the level of the otic pit (op); somites 1-12 also express the transgene. (B-K) Embryos derived from Hoxb8 reporter lines, with or without effector transgene (genotypes indicated at bottom of panels). (B) 13-somite embryo of line 59. Arrow indicates expression boundary in the neural tube. Expression in neural crest cells in the head is characteristic of construct 3. (C) 17-somite embryo of line 59 containing construct RARβ-cdx4. Expression in the neural tube extends to the level of the otic pit (op). (D) Parasagittal section through a line 65-derived embryo containing RARβ-cdx4. The expression domain in the neural tube is extended slightly (arrow) relative to the boundary in the parental line, which is at the level of S7/8. Small patches of expression (green arrowheads) are present more anteriorly in the neural tube. ov, otic vesicle. (E) Parasagittal section through embryo shown in B. Expression boundary in the neural tube (arrow) is at the S5/6 level, boundary in paraxial mesoderm is in S11. Sagittal section through embryo shown in C, showing the extension of the expression domain in the neural tube (arrow). Note the irregular morphology of the neural tube. More lateral section through the same embryo, showing ectopic expression in the somites, anterior of S11, up to S2. (H) Approximately 11-somite line 59-embryo containing RARβ-cdx1. Arrowhead indicates expression boundary in neural tube. (I) Parasagittal section through embryo shown in H, showing anteriorized expression boundaries in the neural tube (arrowhead; anterior to somite 1) and paraxial mesoderm, in somite 5 (S5). A few blue cells are present in somites 3 and 4 as well. (J) Detail of section shown in I, showing fusion of several somites (double-headed arrow). Somites 1-5, not shown here, are unfused, as is the most recently formed somite, estimated to be somite 11. (K) More median section through the same embryo, showing severe undulations of the, open, neural tube (double-headed arrow). Bars, 1 mm (A-C,H); 0.5 mm (D-G,I,K); 50 μm (J).

Fig. 4.

Ectopic expression of cdx genes anteriorizes expression of Hoxb8 reporter transgenes. (A) 15-somite embryo showing lacZ expression driven by the RARβ2 promoter. Expression in the neural tube extends to the level of the otic pit (op); somites 1-12 also express the transgene. (B-K) Embryos derived from Hoxb8 reporter lines, with or without effector transgene (genotypes indicated at bottom of panels). (B) 13-somite embryo of line 59. Arrow indicates expression boundary in the neural tube. Expression in neural crest cells in the head is characteristic of construct 3. (C) 17-somite embryo of line 59 containing construct RARβ-cdx4. Expression in the neural tube extends to the level of the otic pit (op). (D) Parasagittal section through a line 65-derived embryo containing RARβ-cdx4. The expression domain in the neural tube is extended slightly (arrow) relative to the boundary in the parental line, which is at the level of S7/8. Small patches of expression (green arrowheads) are present more anteriorly in the neural tube. ov, otic vesicle. (E) Parasagittal section through embryo shown in B. Expression boundary in the neural tube (arrow) is at the S5/6 level, boundary in paraxial mesoderm is in S11. Sagittal section through embryo shown in C, showing the extension of the expression domain in the neural tube (arrow). Note the irregular morphology of the neural tube. More lateral section through the same embryo, showing ectopic expression in the somites, anterior of S11, up to S2. (H) Approximately 11-somite line 59-embryo containing RARβ-cdx1. Arrowhead indicates expression boundary in neural tube. (I) Parasagittal section through embryo shown in H, showing anteriorized expression boundaries in the neural tube (arrowhead; anterior to somite 1) and paraxial mesoderm, in somite 5 (S5). A few blue cells are present in somites 3 and 4 as well. (J) Detail of section shown in I, showing fusion of several somites (double-headed arrow). Somites 1-5, not shown here, are unfused, as is the most recently formed somite, estimated to be somite 11. (K) More median section through the same embryo, showing severe undulations of the, open, neural tube (double-headed arrow). Bars, 1 mm (A-C,H); 0.5 mm (D-G,I,K); 50 μm (J).

Fig. 5.

Ectopic expression of cdx4 extends the Hoxb8 expression domain. (A) Line-59 embryo, approximately 11 somites, containing RARβ-cdx4, lightly stained for β-galactosidase activity, showing extension of neural tube expression to the level of the otic pit (arrowhead). The arrow indicates the approximate position of the normal expression boundary of the reporter gene. (B) Same embryo as in A, after in situ hybridization with a Hoxb8 probe. Hoxb8 expression also extends to the otic pit (arrowhead). The arrow indicates the approximate position of the Hoxb8 expression boundary in wild-type embryos, at the level of somite 5. Bars, 0.5 mm.

Fig. 5.

Ectopic expression of cdx4 extends the Hoxb8 expression domain. (A) Line-59 embryo, approximately 11 somites, containing RARβ-cdx4, lightly stained for β-galactosidase activity, showing extension of neural tube expression to the level of the otic pit (arrowhead). The arrow indicates the approximate position of the normal expression boundary of the reporter gene. (B) Same embryo as in A, after in situ hybridization with a Hoxb8 probe. Hoxb8 expression also extends to the otic pit (arrowhead). The arrow indicates the approximate position of the Hoxb8 expression boundary in wild-type embryos, at the level of somite 5. Bars, 0.5 mm.

Functional anatomy of BH1100

The molecular mechanisms involved in establishing Hox gene expression domains are largely unknown, but probably involve a complex interplay between factors carrying positional information, tissue-specific regulators and both positively and negatively acting maintenance factors. While the multimerization assay was set up to specifically detect responsiveness to differentially expressed factors, our data provide some insights into this complexity. The dissection of BH1100 reveals that factors differentially active along the anteroposterior axis interact with different tissue-specific factors, binding sites for which appear to be unevenly distributed within BH1100 (consider, for instance, constructs 5, 6 and 7, Fig. 1B). An unanticipated result was that EcMs79 posteriorizes the mesoderm expression boundary in this assay. Given our lack of knowledge of the molecular mechanisms involved, any explanation for this phenomenon is speculative. An interpretation we favor, not to the exclusion of other plausible hypotheses, is that (1) EcMs79 is relatively deficient in binding sites for mesoderm-specific factors, as suggested by the fact that, even as a trimer, it is a poor activator of mesoderm expression, and (2) it nevertheless interacts with the Hoxb8 promoter in the mesoderm, but in a non-productive way, as a result of which in the mesoderm the extra copies of EcMs79 effectively compete with BH1100 and decrease its interaction with the promoter. With these assumptions, the fact that mutation of the CDX binding sites in EcMs79 abolishes both the anteriorization of neurectoderm expression and the posteriorization of mesoderm expression, suggests that CDX proteins are necessary to establish a functional or dysfunctional interaction between regulatory elements and the promoter.

Although TTTAT(A/T) motifs are essential for EcMs79 to function in the multimerization assay, a few fragments lacking such motifs have some effect on the expression boundaries in this assay (e.g. BstNI-PvuII, Fig. 1B). We note that these fragments do contain related sequence motifs that may be functional CDX binding sites as well (see Fig. 1B; Margalit et al., 1993). However, alternative explanations for their effect on the expression boundaries are also possible.

Our results indicate that not all CDX binding sites are equally important: EcMs79, containing four binding sites, can autonomously drive regionally restricted expression, whereas construct 16m, in which BH1100 lacks these four sites but retains at least four others, is unable to do so. The importance of the sites contained in EcMs79 may derive from their close proximity to each other or to binding sites for other, as yet unidentified transcription factors that depend on binding of CDX at neighbouring sites. Similar observations were made by Shashikant et al. (1995), who found that only one of two TTTATG motifs in a Hoxc8 regulatory element is essential for its function, although both were later shown to bind CDX2 in vitro (Taylor et al., 1997).

cdx gene products transduce positional information

The observation that the fragment within BH1100, which is the most responsive in the multimerization assay, coincides with a pronounced clustering of potential CDX binding sites suggests that cdx gene products could be the sought-after differentially expressed factors, and this is supported by our in vitro binding studies and overexpression experiments. However, we cannot formally exclude that the actual effectors are unidentified proteins with similar specificity. One possibility we considered is that of autoregulation or cross-regulation by other Hox gene products, as HOX proteins of paralogue groups 6 to 13 can recognize TTTATG sites in vitro (Shen et al., 1997). While this possibility cannot be fully discounted, it seems very unlikely, for the following reasons. Most HOX proteins require cofactors of the PBX/EXD family for efficient DNA binding, and all HOX binding sites shown to be functional in vivo contain an overlapping PBX consensus site (Pöpperl et al., 1995; Gould et al., 1997; Maconochie et al., 1997). In contrast, no such site (or any recurrent motif) is associated with the TTTAT(A/T/G) motifs in BH1100 (data not shown). We tested whether a RARβ-Hoxb9 construct can affect reporter gene expression and did not observe any effect in seven line-59-derived embryos containing this construct (J. H. Roelfsema, W. dG. and J. D., unpublished observations).

HOX proteins of paralogue groups 11-13 do not bind DNA cooperatively with PBX (Shen et al., 1997). However, since nine copies of EcMs79 coupled to construct 2 drive expression in the neural tube up to the level of somite 4, anterior to the Hoxb8 expression boundary (J. H. Roelfsema, W. dG. and J. D., unpublished observations), the involvement of these more posteriorly expressed, PBX-independent Hox gene products, as well as of autoregulation by HOXB8, in establishing the anterior expression boundary of this construct can be excluded.

Our cdx gain-of-function experiments suggest that Hox expression domains are limited by CDX expression levels, and this seems to be supported by ectopic expression experiments in Xenopus (Epstein et al., 1997; Isaacs et al., 1998). Thus, while not providing absolute proof, the cumulative evidence strongly supports the hypothesis that cdx gene products transduce positional information. Whether their expression levels provide detailed positional information, or rather define relatively broad domains that are subsequently refined by other mechanisms, cannot be decided on the basis of our data.

Our results would imply that the sensitivity of Hoxb8 reporter constructs to the CDX activity gradient is a function of the number of regulatory elements interacting with the promoter, and this may explain the relatively modest anteriorizations obtained upon cdx overexpression in line 65 as compared to line 59. However, since cdx4 expression levels were not compared in these embryos, an alternative possibility is that expression of RARβ-cdx4 in the line-65 embryos happened to be mosaic and/or relatively low due to position effects. Given that lacZ expression was induced, though sparsely, at very anterior levels in the neural tube of the line 65 embryos, the latter possibility is perhaps more likely.

Finally, it is noteworthy that Hox gene expression in embryos lacking the trithorax group gene Mll is normal at 8.5 days, whereas defects in maintenance of expression are apparent at 9.5 days (Yu et al., 1998), indicating that a switch from an initiation phase to a maintenance phase occurs between these stages. This suggests that multimerization of regulatory elements and cdx overexpression affect Hox gene expression in the initiation phase. If functional CDX proteins with altered specificity can be designed it may be possible more rigorously to determine their role in Hox gene regulation in these experimental paradigms and in transplantation experiments.

Genetic studies have shown that caudal homologues are involved in anteroposterior patterning in organisms having such divergent modes of embryogenesis as insects (Macdonald and Struhl, 1986), nematodes (Waring and Kenyon, 1990, 1991) and vertebrates (Subramanian et al., 1995; Chawengsaksophak et al., 1997). In both Drosophila and vertebrate embryos, their gene products form an anteroposterior gradient. However, in the highly specialized Drosophila embryo, Caudal function has become deeply embedded in the network of maternal and segmentation genes that evolved to enable the rapid patterning of this embryo (Mlodzik and Gehring, 1987; Dearolf et al., 1989; Rivera-Pomar et al., 1995; Schulz and Tautz, 1995; La Rosée et al., 1997), and the absolute expression levels of Caudal do not appear to be relevant unless other layers of regulation are removed (Schulz and Tautz, 1995). The results presented here suggest that cdx gene products in vertebrates regulate Hox genes directly, and this may represent the ancestral situation.

We thank Christopher Wright for providing the cdx4 cDNA and antibodies, Peter Gruss for the cdx1 cDNA and antibodies, Robert James for the cdx2 cDNA, Peter Traber for CDX2 antibodies and Marga van Rooijen for help with preparation of the figures. This work was supported in part by an EEC Human Capital and Mobility program grant. The nucleotide sequence of BH1100 has been deposited into the EMBL database under accession number AF020540.

Beck
,
F.
,
Erler
,
T.
,
Russell
,
A.
and
James
,
R.
(
1995
).
Expression of Cdx-2 in the mouse embryo and placenta: Possible role in patterning of the extra-embryonic membranes
.
Dev. Dyn.
204
,
219
227
.
Charité
,
J.
,
De Graaff
,
W.
,
Shen
,
S.
and
Deschamps
,
J.
(
1994
).
Ectopic expression of Hoxb-8 causes duplication of the ZPA in the forelimb and homeotic transformation of axial structures
.
Cell
78
,
589
601
.
Charité
,
J.
,
De Graaff
,
W.
,
Vogels
,
R.
,
Meijlink
,
F.
and
Deschamps
,
J.
(
1995
).
Regulation of the Hoxb-8 gene: Synergism between multimerized cis-acting elements increases responsiveness to positional information
.
Dev. Biol.
171
,
294
305
.
Chawengsaksophak
,
K.
,
James
,
R.
,
Hammond
,
V. E.
,
Köntgen
,
F.
and
Beck
,
F.
(
1997
).
Homeosis and intestinal tumours in Cdx2 mutant mice
.
Nature
386
,
84
87
.
Dearolf
,
C. R.
,
Topol
,
J.
and
Parker
,
C. S.
(
1989
).
The caudal gene product is a direct activator of fushi tarazu transcription during Drosophila embryogenesis
.
Nature
341
,
340
343
.
Deschamps
,
J.
and
Wijgerde
,
M.
(
1993
).
Two phases in the establishment of HOX expression domains
.
Dev. Biol.
156
,
473
480
.
Doniach
,
T.
(
1993
).
Planar and vertical induction of anteroposterior pattern during the development of the amphibian central nervous system
.
J. Neurobiol.
24
,
1256
1275
.
Epstein
,
M.
,
Pillemer
,
G.
,
Yelin
,
R.
,
Yisraeli
,
J. K.
and
Fainsod
,
A.
(
1997
).
Patterning of the embryo along the anterior-posterior axis: the role of the caudal genes
.
Development
124
,
3805
3814
.
Gamer
,
L. W.
and
Wright
,
C. V. E.
(
1993
).
Murine Cdx-4 bears striking similarities to the Drosophila caudal gene in its homeodomain sequence and early expression pattern
.
Mech. Devel.
43
,
71
81
.
Gould
,
A.
(
1997
).
Functions of mammalian Polycomb group and trithorax group related genes
.
Curr. Opin. Genet. Dev.
7
,
488
494
.
Gould
,
A.
,
Morrison
,
A.
,
Sproat
,
G.
,
White
,
R. A. H.
and
Krumlauf
,
R.
(
1997
).
Positive cross-regulation and enhancer sharing: two mechanisms for specifying overlapping Hox expression patterns
.
Genes Dev.
11
,
900
913
.
Grapin-Botton
,
A.
,
Bonnin
,
M.-A.
,
McNaughton
,
L. A.
,
Krumlauf
,
R.
and
Le Douarin
,
N. M.
(
1995
).
Plasticity of transposed rhombomeres: Hox gene induction is correlated with phenotypic modifications
.
Development
121
,
2707
2721
.
Grapin-Botton
,
A.
,
Bonnin
,
M.-A.
and
Le Douarin
,
N. M.
(
1997
).
Hox gene induction in the neural tube depends on three parameters: competence, signal supply and paralogue group
.
Development
124
,
849
859
.
Isaacs
,
H. V.
,
Pownall
,
M. E.
and
Slack
,
J. M. W.
(
1998
).
Regulation of Hox gene expression and posterior development by the Xenopus caudal homologue Xcad3
.
EMBO J.
17
,
3413
3427
.
Itasaki
,
N.
,
Sharpe
,
J.
,
Morrison
,
A.
and
Krumlauf
,
R.
(
1996
).
Reprogramming Hox expression in the vertebrate hindbrain: Influence of paraxial mesoderm and rhombomere transposition
.
Neuron
16
,
487
500
.
La
Rosée
,
A.
,
Häder
,
T.
,
Taubert
,
H.
,
Rivera-Pomar
,
R.
and
Jäckle
,
H.
(
1997
).
Mechanism and Bicoid-dependent control of hairy stripe 7 expression in the posterior region of the Drosophila embryo
.
EMBO J.
16
,
4403
4411
.
Macdonald
,
P. M.
and
Struhl
,
G.
(
1986
).
A molecular gradient in early Drosophila embryos and its role in specifying the body pattern
.
Nature
324
,
537
545
.
Maconochie
,
M. K.
,
Nonchev
,
S.
,
Studer
,
M.
,
Chan
,
S.-K.
,
Pöpperl
,
H.
,
Sham
,
M. H.
,
Mann
,
R. S.
and
Krumlauf
,
R.
(
1997
).
Cross-regulation in the mouse HoxB complex: the expression of Hoxb2 in rhombomere 4 is regulated by Hoxb1
.
Genes Dev.
11
,
1885
1895
.
Margalit
,
Y.
,
Yarus
,
S.
,
Shapira
,
E.
,
Gruenbaum
,
Y.
and
Fainsod
,
A.
(
1993
).
Isolation and characterization of target sequences of the chicken CdxA homeobox gene
.
Nucl. Acids Res.
21
,
4915
4922
.
McGinnis
,
W.
and
Krumlauf
,
R.
(
1992
).
Homeobox genes and axial patterning
.
Cell
68
,
283
302
.
Meyer
,
B. I.
and
Gruss
,
P.
(
1993
).
Mouse Cdx-1 expression during gastrulation
.
Development
117
,
191
203
.
Mlodzik
,
M.
and
Gehring
,
W. J.
(
1987
).
Hierarchy of the genetic interactions that specify the anteroposterior segmentation pattern of the Drosophila embryo as monitored by caudal protein expression
.
Development
101
,
421
435
.
Pankratz
,
M. J.
and
Jäckle
,
H.
(
1993
). Blastoderm segmentation. In
The Development of Drosophila melanogaster
(ed.
M.
Bate
and A.
Martinez
Arias
), pp.
467
516
.
Cold Spring Harbor Laboratory Press
.
Pöpperl
,
H.
,
Bienz
,
M.
,
Studer
,
M.
,
Chan
,
S.-K.
,
Aparicio
,
S.
,
Brenner
,
S.
,
Mann
,
R. S.
and
Krumlauf
,
R.
(
1995
).
Segmental expression of Hoxb-1 is controlled by a highly conserved autoregulatory loop dependent upon exd/pbx
.
Cell
81
,
1031
1042
.
Pownall
,
M. E.
,
Tucker
,
A. S.
,
Slack
,
J. M. W.
and
Isaacs
,
H. V.
(
1996
).
eFGF, xcad3 and Hox genes form a molecular pathway that establishes the anteroposterior axis in Xenopus
.
Development
122
,
3881
3892
.
Quandt
,
K.
,
Frech
,
K.
,
Karas
,
H.
,
Wingender
,
E.
and
Werner
,
T.
(
1995
).
MatInd and MatInspector: new fast and versatile tools for detection of consensus matches in nucleotide sequence data
.
Nucl. Acids Res.
23
,
4878
4884
.
Rivera-Pomar
,
R.
,
Lu
,
X.
,
Perrimon
,
N.
,
Taubert
,
H.
and
Jäckle
,
H.
(
1995
).
Activation of posterior gap gene expression in the Drosophila blastoderm
.
Nature
376
,
253
256
.
Ruiz i
Altaba
, A
. (
1993
).
Induction and axial patterning of the neural plate: planar and vertical signals
.
J. Neurobiol.
24
,
1276
1304
.
Schulz
,
C.
and
Tautz
,
D.
(
1995
).
Zygotic caudal regulation by hunchback and its role in abdominal segment formation of the Drosophila embryo
.
Development
121
,
1023
1028
.
Schumacher
,
A.
and
Magnuson
,
T.
(
1997
).
Murine Polycomb-and trithorax-group genes regulate homeotic pathways and beyond
.
Trends Genet.
13
,
167
170
.
Shashikant
,
C. S.
,
Bieberich
,
C. J.
,
Belting
,
H.-G.
,
Wang
,
J. C. H.
,
Borbély
,
M.A.
and
Ruddle
,
F. H.
(
1995
).
Regulation of Hoxc-8 during mouse embryonic development: identification and characterization of critical elements involved in early neural tube expression
.
Development
121
,
4339
4347
.
Shen
,
W.-F.
,
Rozenfeld
,
S.
,
Lawrence
,
H. J.
and
Largman
,
C.
(
1997
).
The Abd-B-like Hox homeodomain proteins can be subdivided by the ability to form complexes with Pbx1a on a novel DNA target
.
J. Biol. Chem.
272
,
8198
8206
.
Subramanian
,
V.
,
Meyer
,
B. I.
and
Gruss
,
P.
(
1995
).
Disruption of the murine homeobox gene Cdx1 affects axial skeletal identities by altering the mesodermal expression domains of Hox genes
.
Cell
83
,
641
653
.
Suh
,
E.
,
Chen
,
L.
,
Taylor
,
J.
and
Traber
,
P. G.
(
1994
).
A homeodomain protein related to caudal regulates intestine-specific gene transcription
.
Mol. Cell. Biol.
14
,
7340
7351
.
Taylor
,
J. K.
,
Levy
,
T.
,
Suh
,
E. R.
and
Traber
,
P. G.
(
1997
).
Activation of enhancer elements by the homeobox gene Cdx2 is cell line specific
.
Nucl. Acids Res.
25
,
2293
2300
.
Vogels
,
R.
,
Charité
,
J.
,
De Graaff
,
W.
and
Deschamps
,
J.
(
1993
).
Proximal cis-acting elements cooperate to set Hoxb-7 (Hox-2.3) expression boundaries in transgenic mice
.
Development
118
,
71
82
.
Waring
,
D. A.
and
Kenyon
,
C.
(
1990
).
Selective silencing of cell communication influences anteroposterior pattern formation in C. elegans
.
Cell
60
,
123
131
.
Waring
,
D. A.
and
Kenyon
,
C.
(
1991
).
Regulation of cellular responsiveness to inductive signals in the developing C. elegans nervous system
.
Nature
350
,
712
715
.
Wilkinson
,
D. G.
(
1992
). Whole mount in situ hybridisation of vertebrate embryos. In
In Situ Hybridisation
(ed.
D. G.
Wilkinson
), pp.
75
83
.
Oxford
:
IRL Press
.
Yu
,
B. D.
,
Hanson
,
R.
,
Hess
,
J. L.
,
Horning
,
S. E.
and
Korsmeyer
,
S. J.
(
1998
). MLL, a mammalian trithorax-Group gene, functions as a transcriptional maintenance factor in morphogenesis. Proc.
Nat. Acad. Sci. USA
(in press).