Expression of vertebrate Hox genes is regulated by retinoids in cell culture and in early embryonic development. We have identified a 185-bp retinoid-responsive transcriptional enhancer 5′ of the human HOXD4 gene, which regulates inducibility of the gene in embryonal carcinoma cells through a pattern of DNA-protein interaction on at least two distinct elements. One of these elements contains a direct repeat mediating ligand-dependent interaction with retinoic acid receptors, and is necessary though not sufficient for the enhancer function. The HOXD4 enhancer directs expression of a lacZ reporter gene in the neural tube of transgenic mouse embryos in a timeregulated and regionally restricted fashion, reproducing part of the anterior neuroectodermal expression pattern of the endogenous Hoxd-4 gene. Administration of retinoic acid to developing embryos causes alterations in the spatial restriction of the transgene expression domain, indicating that the HOXD4 enhancer is also a retinoid-responsive element in vivo. The timing of the retinoic acid response differs from that seen with more 3′ Hox genes, in that it occurs much later. This shows that the temporal window of competence in the ability to respond to retinoic acid differs between Hox genes and can be linked to specific enhancers. Mutations in the direct repeat or in a second element in the enhancer affect both retinoid response in culture and developmental regulation in embryos, suggesting that co-operative interactions between different factors mediate the enhancer activity. These data provide further support for a role of endogenous retinoids in regulation and spatial restriction of Hox gene expression in the central nervous system.

Hox genes encode a family of transcription factors containing a highly conserved DNA binding homeodomain (Gehring et al., 1990), and play a key role in the specification of regional identities along the anteroposterior (AP) axis in embryonic development (reviewed by Krumlauf, 1993; McGinnis and Krumlauf, 1992). In vertebrates, Hox genes are organised in four homologous clusters, and are expressed according to characteristic, spatially restricted patterns (reviewed by Duboule, 1992; Krumlauf, 1993; McGinnis and Krumlauf, 1992). In all four Hox clusters, there is a linear relationship between the order of genes along the chromosome, the timing of their activation in development, and the position of their expression domains along the embryonic AP axis, according to a 3′-earlyanterior/5′-late-posterior colinearity rule (Duboule and Dollé, 1989; Graham et al., 1989).

Reconstruction of Hox expression patterns in transgenic mice using DNA sequences from within the clusters fused to a lacZ reporter gene has provided a convenient functional assay for identifying cis-regulatory elements and their associated factors. In some cases, Hox/lacZ transgenes with adjacent flanking regions retain most of the spatial and temporal expression patterns of the original genes, despite the fact that they are integrated outside their normal chromosomal locations (Behringer et al., 1993; Frasch et al., 1995; Gerard et al., 1993; Marshall et al., 1992, 1994; Püschel et al., 1990, 1991; Sham et al., 1992; Studer et al., 1994; Whiting et al., 1991). However, in other cases, the transgenes generate only limited subsets of the complete expression patterns, suggesting that for their normal regulation, multiple elements and long-range interactions are essential (Eid et al., 1993; Kress et al., 1990; Vogels et al., 1993).

Hox genes are specifically regulated, both in vivo and in cell culture by retinoids, vitamin A derivatives with pleiotropic effects on development and cell differentiation (Blomhoff et al., 1990; Saurat, 1990; Summerbell and Maden, 1990; Tabin, 1991). Retinoids are present at active concentrations in embryonic structures with a proven role in pattern formation, such as the zone of polarising activity (Thaller and Eichele, 1987, 1990), Hensen’s node or its mammalian equivalent (Chen et al., 1992; Hogan et al., 1992), and the floor plate of the neural tube (Horton and Maden, 1995; Wagner et al., 1992, 1990). Systemic or local administration of excess retinoic acid (RA) to developing vertebrate embryos has profound effects on axial patterning and specification of regional identities in a number of structures, including the CNS, axial skeleton and limbs (reviewed by McGinnis and Krumlauf, 1992; Summerbell and Maden, 1990; Tabin, 1991). This is invariably accompanied by re-patterning of Hox gene expression domains (Conlon and Rossant, 1992; Izpisúa-Belmonte et al., 1991; Kessel, 1992; Kessel and Gruss, 1991; Marshall et al., 1992; Nohno et al., 1991; Oliver et al., 1990; Simeone et al., 1995), suggesting a specific role for Hox gene products as molecular transducers of the morphogenetic (positional?) signals provided, or at least mimicked, by retinoids (McGinnis and Krumlauf, 1992; Tabin, 1991 and references therein).

In human and mouse embryonal carcinoma (EC) cells, Hox genes are activated by RA in a dose-dependent, 3′→5′ sequential pattern similar to that observed in embryonic development (Papalopulu et al., 1991; Simeone et al., 1990, 1991), providing an in vitro model to study the molecular mechanisms of Hox gene regulation. Studies carried out both in vitro and in vivo allowed the identification of RA-responsive elements at the 3′ end of the mouse or human Hoxa and Hoxb clusters, which confer to a lacZ reporter RA inducibility in EC cells (Langston and Gudas, 1992; Ogura and Evans, 1995a,b) or RA inducibility and regulated expression in transgenic mouse embryos (Frasch et al., 1995; Marshall et al., 1994; Studer et al., 1994). Retinoid-responsive elements are therefore likely to participate in the activation of Hox clusters, and are part of the regulatory mechanisms underlying temporal and spatial regulation of Hox gene expression in development. However, it is unknown if the colinear RA response is mediated by a single region at the 3′ end of the complexes, or involves multiple RAREs spread throughout the clusters.

The human HOXD4 gene is expressed in multiple transcripts, developmentally regulated in embryonic development (Cianetti et al., 1990; Mavilio et al., 1986), and induced in the human EC cell line NT2/D1 by exposure to retinoids in the nM to μM concentration range (Simeone et al., 1990). Hoxd-4 is activated early in mouse embryogenesis, and is expressed in 12.5-day embryos mainly in the limbs, paraxial mesoderm and CNS, with an anterior restriction at the level of rhombomere (r) 6/7 boundary (Gaunt et al., 1989; Hunt et al., 1991). We and others have previously reported the characterisation of a 185-bp transcriptional enhancer, upstream of the human and mouse Hoxd-4 genes. This enhancer regulates gene inducibility by all-trans- and 9-cis-RA in culture, through a complex pattern of DNA-protein interactions on at least four distinct elements (Moroni et al., 1993; Pöpperl and Featherstone, 1993). Although ligand-dependent activity of RA receptors (RARs) upon one of these elements has a key role in the enhancer function, RARs represent only a fraction of the transcription factors interacting with the RA-responsive enhancer, which appears to be a complex element requiring specific combinations of nuclear factors for its proper function (Moroni et al., 1993).

In this study, we show that the HOXD4 enhancer is able to mediate expression of a lacZ reporter gene in the neural tube of transgenic mice according to a time-regulated, regionally restricted pattern, which reproduces part of the expression pattern of the endogenous Hoxd-4 gene. Mutations in two of the four elements of the enhancer affect both RA-inducibility in EC cells and developmentally regulated expression in transgenic mice, establishing direct correlations between specific RA-dependent functions identified in vitro and their role in gene regulation in vivo.

Plasmids

A partial HincII/AseI digest spanning positions −2892 to +21 from the P2 proximal transcription start site in the HOXD4 upstream genomic sequence, was subcloned from the λ13G phage (Mavilio et al., 1986) in pBluescript SK (Stratagene). The fragment was recovered as a SalI/XbaI fragment and cloned in the EcoRI/XhoI sites of the pNASSβ vector (Clontech), to generate the pHOXD4-2892-lacZ vector (Fig. 1, construct no. 1). Before injection, plasmid vector sequences were removed by EcoRI/SalI digestion. The same fragment was subcloned in the BamHI site of plasmid pB4-lacZ (Whiting et al., 1991), in which the lacZ reporter gene is driven by the mouse Hoxb-4 promoter. Fragments for injection were isolated after digestion with NotI to remove vector sequences. The pHOXD4-2892Δ RARE vector (constructs no. 2) was generated by internal deletion of the NheI/NarI fragment, containing the entire RA-responsive enhancer (from position −2711 to −2333). The pHOXD4 RARE and pHOXD4 RARE rev. vectors (constructs nos. 3 and 4) were generated by cloning the NheI/NarI fragment in both orientations in the pB4-lacZ vector. Deletions of the X (pHOXD4-2892ΔX, construct no. 5) and 3 (pHOXD4-2892Δ3, construct no. 7) elements, and mutation of the element 2 (pHOXD4-2892m2, construct no. 6), were obtained by PCR-mediated splicing by overlap extension on the HindIII/XhoI fragment (−2720 to −2109), subcloned in pBSK as template, using SK and T7 as external primers, and specific internal primers for element X (direct: 5′-CGTTAAATATTCACTCCCGCTCTAGAGTCTC-TAGGGCTTTTTGTGC-3′, reverse: 5′-GCACAAAAAGCCCTA-GAGACTCTAGAGCGGGAGGTGAATATTTAACG-3′), element 3 (direct: 5′-CCTGTCTGAACAAGTCGAGCGATCGTCTAA-CAAATATGAAAATGTG-3′, reverse: 5′-CCTGTCTGAACA-AGTCGAGCGATCGTCTAACAAATATGAAAATGTG-3′) and element 2 (direct: 5′-TCAATAACTCGTTGGCTTGGGCCCGTC-TAGACCAGTCGAGCAATAAGGTG-3′, reverse: 5′-CACCT-TATTGCTCGACTGGTCTAGACGGGCCCAAGCC AACGAGT-TATTGA-3′). Each mutant was then recovered from the template plasmid as NheI/NarI fragment, and used to replace the wild-type NheI/NarI fragment into the pHOXD4-2892 constructs.

Fig. 1.

Left: map of the HOXD4-lac-Z reporter constructs. The symbols represent the footprinted elements X, 1, 2 and 3 of the 185-bp RA responsive enhancer (see Fig. 4C). Deletion or mutation (m) of an element is indicated. The horizontal arrow indicates the orientation of the enhancer with respect to the promoter. The short arrow represents the proximal promoter of the HOXD4 gene (Moroni et al., 1993) in the constructs used for cell transfection, or the murine Hoxb-4 promoter in the constructs used to generate transgenic mice. Constructs no. 3 and 4 contained only the Hoxb-4 promoter. Hc, HincII; Nh, NheI; Na, NarI. Middle: generation of neural tube-specific lacZ expression pattern in transient transgenic mouse embryos or lines, as number of positively expressing/total transgenic embryos. In the case of construct no. 6, yolk sac DNA sample were lost, so it was impossible to determine the total number of transgenic embryos (n.t., not tested). Right: response of the constructs to RA induction in vivo (+, anterior shift in the neural tube expression domain; −, no change; n.t., not tested) and in vitro (transfection in NT2/D1 cells). Activity of the lacZ reporter in transfected cells is expressed as β-gal arbitrary units (mean ± s.e.m. of at least three separate, duplicate experiments), after normalisation for the luciferase activity derived from a co-transfected pT81luc plasmid (see methods).

Fig. 1.

Left: map of the HOXD4-lac-Z reporter constructs. The symbols represent the footprinted elements X, 1, 2 and 3 of the 185-bp RA responsive enhancer (see Fig. 4C). Deletion or mutation (m) of an element is indicated. The horizontal arrow indicates the orientation of the enhancer with respect to the promoter. The short arrow represents the proximal promoter of the HOXD4 gene (Moroni et al., 1993) in the constructs used for cell transfection, or the murine Hoxb-4 promoter in the constructs used to generate transgenic mice. Constructs no. 3 and 4 contained only the Hoxb-4 promoter. Hc, HincII; Nh, NheI; Na, NarI. Middle: generation of neural tube-specific lacZ expression pattern in transient transgenic mouse embryos or lines, as number of positively expressing/total transgenic embryos. In the case of construct no. 6, yolk sac DNA sample were lost, so it was impossible to determine the total number of transgenic embryos (n.t., not tested). Right: response of the constructs to RA induction in vivo (+, anterior shift in the neural tube expression domain; −, no change; n.t., not tested) and in vitro (transfection in NT2/D1 cells). Activity of the lacZ reporter in transfected cells is expressed as β-gal arbitrary units (mean ± s.e.m. of at least three separate, duplicate experiments), after normalisation for the luciferase activity derived from a co-transfected pT81luc plasmid (see methods).

The pT81luc expression vector (Nordeen, 1988), containing the fruitfly luciferase gene under the control of the enhancerless HSV-TK promoter, was used as internal control in all transfection experiments.

Cell culture and transfection

The human embryonal carcinoma cell line NT2/D1 (Andrews et al., 1984) was cultured in high-glucose Dulbecco’s medium (DMEM) supplemented with 10% fetal calf serum and 20 mM Hepes buffer (all from GIBCO) in 5% CO2 in air humidified atmosphere. For transfection, 5×105 cells were seeded in 60-mm dishes, induced for 24 hours with 10 μM all-trans-retinoic acid (Sigma) as previously described (Simeone et al., 1990), and then transfected in calcium phosphate/Hepes-buffered saline solution (Sambrook et al., 1989) with a total of 10 μg of plasmid DNA (9 μg reporter + 1 μg pT81luc), always in the presence of RA. Total protein extracts were prepared after 36 additional hours of culture by three cycles of freezing and thawing, and assayed for luciferase or β-galactosidase activity as described (de Wet et al., 1987; Sambrook et al., 1989). Transfection of non-induced cells was carried out by the same protocol, without adding RA.

DNase I footprinting analysis

Crude nuclear extracts were prepared as described by Dignam et al. (1983) from 107-108 uninduced or RA-induced NT2/D1 cells, aliquoted and stored in liquid nitrogen. A 171-bp subcloned fragment corresponding to the −2720 to −2449 HOXD4 upstream region was 3′ end-labeled, purified on polyacrylamide gel, and incubated with 10 to 60 μg of NT2/D1 nuclear extracts and 1 μg of poly(dI-dC) for 10 minutes at room temperature (RT) in a binding reaction containing 50 mM MgCl2, 20 mM Tris-HCl pH 7.9, 1 mM DTT and 25% glycerol. 300 ng of DNase I (Boehringer) were added to the mixture and the incubation continued for 90 seconds at RT. Reaction was stopped with 0.2 M Tris-HCl pH 7.5, 0.2 M NaCl, 1% (w/v) SDS, 30 mM EDTA and 20 μg yeast tRNA, extracted with phenol-chloroform and pre-cipitated in ethanol. Pellets were resuspended in 8 μl of standard 80% formamide dye, denatured for 5 minutes at 95°C and run on a 6% urea-polyacrylamide gel. Gels were dried and exposed overnight at −70°C to Kodak X-AR5 films.

Whole-mount in situ hybridisation

Mouse embryos at 9.5-12.5 d.p.c. were fixed in 4% paraformaldehyde and then hybridised using a modification of the method of Wilkinson (1992). The digoxigenin-labeled RNA probe corresponded to the full-length Hoxd-4 cDNA (Featherstone et al., 1988).

Transgenic mice

Transgenic mice were produced using intercrosses of (CBA × C57BL10) F1 mice. Pronuclear injection was performed as described by Hogan et al. (1986). In timed pregnancies, the day of appearance of the vaginal plug was taken as 0.5 d.p.c. β-galactosidase staining was performed as previously described (Whiting et al., 1991). Embryos for sectioning underwent a further fixation in 4% paraformaldehyde overnight at 4°C, and following standard processing of wax embedded sections (6 μm) were counterstained with eosin. All-trans-retinoic acid (Sigma; 25 mg/ml in dimethylsulphoxide) was diluted 1/10 in sesame seed oil and approximately 200 μl of this given to pregnant mothers by gavage at 7.5, 8.5 or 9.5 d.p.c. for a final dose of 25 mg/kg as previously described (Conlon and Rossant, 1992; Marshall et al., 1992). Total number of transgenic embryos was determined by PCR analysis on yolk sac DNA from embryos harvested for staining, except in the case of construct no. 6, where DNA samples were not taken and efficiency could not be measured. The line ML19, containing the Hoxb-2 r3/5 lacZ transgene (Sham et al., 1993) was mated with lines carrying construct no. 1 to determine the relative position of the HOXD4 enhancers anterior boundary of expression with respect to rhombomere 5.

The HOXD4 upstream genomic region directs neural-specific expression in transgenic mice

We first tested the activity of the 2.9 kb SalI-XbaI HOXD4 genomic fragment spanning the −2892/+21 region of the human HOXD4 gene in transgenic mice to assay for its functional properties in vivo. This fragment, which contains a previously identified retinoid-responsive enhancer (Moroni et al., 1993) and two promoters (Cianetti et al., 1990), was placed in the context of the promoter-less pNASSβ vector, containing a lacZ reporter gene. Out of 5 transgenic founder embryos harvested at 9.5 days post coitum (d.p.c.) and identified by PCR, none showed any expression (not shown). The same construct was able to direct RA-inducible lacZ expression in human NT2/D1 cells (construct no. 1, Fig. 1) as well as in mouse P19 EC cells (not shown). This suggested that either the regulatory region was non-functional in the transgenic assay, or that the human HOXD4 promoter(s) used in transfected NT2/D1 cells are different from those used in embryonic tissues. Indeed, it has been shown for a number of Hox genes (including HOXD4) that in human and mouse EC cells only a subset of the embryonic transcripts are present and induced by RA (Papalopulu et al., 1991; Pöpperl and Featherstone, 1993; Simeone et al., 1990, 1991). Therefore, to assay for regulatory activity within the fragment in transgenic mice we generated constructs which were independent of the endogenous HOXD4 promoters. We placed the HOXD4 genomic fragment into a different vector (pB4-lacZ), which contains the basal promoter of the mouse Hoxb-4 gene driving a lacZ reporter. This basic mouse vector was originally generated in a transgenic deletion analysis of the Hoxb-4 gene, where it displayed no reporter activity on its own except for slight staining in the superior colliculi of the midbrain (Whiting et al., 1991). We have routinely used the pB4-lacZ vector to test for enhancer activity in transgenic mice (Aparicio et al., 1995; Marshall et al., 1994; Morrison et al., 1995; Sham et al., 1992, 1993; Whiting et al., 1991). Seven independent transgenic animals were obtained with the human HOXD4 fragment in this construct, and five of these expressed the transgene (two lines and three founder embryos). All five had an identical pattern of expression in the neural tube, restricted to the spinal cord just posterior to the hindbrain (Fig. 2). The expression pattern was consistent with that previously obtained with a HOXD4 upstream genomic fragment linked to the basal mouse Hoxa-5 promoter (Tuggle et al., 1990), showing that the HOXD4 enhancer element drives a consistent expression pattern on at least two different mouse basal promoters. Therefore, all HOXD4 genomic fragments tested in transgenic mice were in the context of the basal pB4-lacZ reporter, whereas those tested in parallel in NT2/D1 cells were in the context of the pNASSβ reporter (human promoters). The parallel constructs are listed in Fig. 1 with the same number and name, regardless of the promoter usage.

Fig. 2.

Time course of β-galactosidase expression from a transgenic line (AL15) containing construct no. 1. (A) 9.5 d.p.c., (B)10.5 d.p.c. and (C) 12.5 d.p.c. (D) Dorsal view of embryo in A, showing that the anterior limit of lacZ expression lies at a position more caudal to the otic vesicle (OV) than the limit of endogenous Hoxd-4 expression, at the r6/7 boundary of a 9.5 d.p.c. embryo (dorsal view (E) and lateral view (F)), as detected by in situ hybridisation. The open triangle in B indicates the midbrain staining mediated by the basal pB4-lacZ vector in approx. 50% of integration sites. (G) The expression at 11.5 d.p.c. of a second independent transgenic line (AL12) carrying construct no. 1. Expression extends posteriorly along the neural tube, eventually producing a biphasic distribution of lacZ staining which is clearly evident by 12.5 d.p.c., indicated in C by open arrows.

Fig. 2.

Time course of β-galactosidase expression from a transgenic line (AL15) containing construct no. 1. (A) 9.5 d.p.c., (B)10.5 d.p.c. and (C) 12.5 d.p.c. (D) Dorsal view of embryo in A, showing that the anterior limit of lacZ expression lies at a position more caudal to the otic vesicle (OV) than the limit of endogenous Hoxd-4 expression, at the r6/7 boundary of a 9.5 d.p.c. embryo (dorsal view (E) and lateral view (F)), as detected by in situ hybridisation. The open triangle in B indicates the midbrain staining mediated by the basal pB4-lacZ vector in approx. 50% of integration sites. (G) The expression at 11.5 d.p.c. of a second independent transgenic line (AL12) carrying construct no. 1. Expression extends posteriorly along the neural tube, eventually producing a biphasic distribution of lacZ staining which is clearly evident by 12.5 d.p.c., indicated in C by open arrows.

The two independent transgenic lines (AL-12 and AL-15) carrying construct no. 1 were used to generate a developmental time course of expression (Fig. 2A-D,G). Expression of the transgene was first detected between 9.0-9.5 d.p.c. in the neural tube in a gradient with a rostral limit in the spinal cord posterior to the otic vesicle (Fig. 2A,D) and posterior expression boundary mapping to the mid thoracic region. This was close to but caudal to the anterior expression limit of the endogenous Hoxd-4 gene, which lies at the r 6/7 boundary as revealed by in situ hybridisation (Fig. 2E,F). At 10.5 d.p.c., levels of lacZ expression had increased and expanded posteriorly along the neural tube. The distance between the rostral limit of expression and the otic vesicle had shortened, suggesting that the anterior extension domain of expression shifts rostrally with time (compare also Fig. 3A,B with Fig 4A,B). At later stages (11.0 d.p.c. onwards), the anterior boundary of expression becomes fixed at a level which appeared to be posterior to that of the endogenous Hoxd-4 gene (Gaunt et al., 1989; Hunt et al., 1991). By 12.5 d.p.c., transgene expression could be seen in two distinct domains within the neural tube, one extending from the hindbrain/spinal cord junction into the mid-thoracic region, and the other at the level of the posterior limb buds (Fig. 2C). These results show that the HOXD4 upstream region directs a subset of the neural-specific expression pattern of the endogenous Hoxd-4.

Fig. 3.

The effect of RA upon the lacZ staining pattern obtained using the line (AL15) containing construct no. 1.The position of the anterior expression boundary of the HOXD4 transgene is shown relative to r5, marked with a second transgene that uses the Hoxb-2 r3/5 specific enhancer to drive lacZ expression. The r3/5-specific lacZ transgene is contained in the line ML19. Embryos containing both transgenes were obtained by interbreeding male and female mice from these two transgenic lines. (A) Expression of the two transgenes in a 10.0 d.p.c. embryo which was not treated with RA. (B,C) Dorsal view and flat mount respectively of the embryo shown in A, illustrating the considerable distance between the anterior boundary of the HOXD4 transgene expression (arrowhead) and r5. (D) Expression of the two transgenes in a 10.0 d.p.c. embryo which was treated with RA in utero at 7.5 d.p.c. Note that the relative positions of the anterior boundary of the HOXD4 transgene expression and r5 are unchanged compared with the untreated control shown in A and B. (E) Dorsal view of the embryo shown in D. (F) Flat mount of non-RA-treated embryo shown in A, showing both the r3 and r5 stripes of lacZ expression. No expression from the HOXD4 transgene can be seen in this field of view. (G) Expression of the two transgenes in a 10.0 d.p.c. embryo which was treated with RA in utero at 9.5 d.p.c.. (H) Dorsal view of the embryo shown in G, note that the anterior boundary of the HOXD4 transgene has shifted rostrally towards r5 and the otic vesicle (OV). (I) Flat mount of embryo shown in G, showing that expression of the HOXD4 transgene extends up to the r6/7 boundary (dashed line). Compare this with a similar field of view (F) from the untreated control embryo shown in A.

Fig. 3.

The effect of RA upon the lacZ staining pattern obtained using the line (AL15) containing construct no. 1.The position of the anterior expression boundary of the HOXD4 transgene is shown relative to r5, marked with a second transgene that uses the Hoxb-2 r3/5 specific enhancer to drive lacZ expression. The r3/5-specific lacZ transgene is contained in the line ML19. Embryos containing both transgenes were obtained by interbreeding male and female mice from these two transgenic lines. (A) Expression of the two transgenes in a 10.0 d.p.c. embryo which was not treated with RA. (B,C) Dorsal view and flat mount respectively of the embryo shown in A, illustrating the considerable distance between the anterior boundary of the HOXD4 transgene expression (arrowhead) and r5. (D) Expression of the two transgenes in a 10.0 d.p.c. embryo which was treated with RA in utero at 7.5 d.p.c. Note that the relative positions of the anterior boundary of the HOXD4 transgene expression and r5 are unchanged compared with the untreated control shown in A and B. (E) Dorsal view of the embryo shown in D. (F) Flat mount of non-RA-treated embryo shown in A, showing both the r3 and r5 stripes of lacZ expression. No expression from the HOXD4 transgene can be seen in this field of view. (G) Expression of the two transgenes in a 10.0 d.p.c. embryo which was treated with RA in utero at 9.5 d.p.c.. (H) Dorsal view of the embryo shown in G, note that the anterior boundary of the HOXD4 transgene has shifted rostrally towards r5 and the otic vesicle (OV). (I) Flat mount of embryo shown in G, showing that expression of the HOXD4 transgene extends up to the r6/7 boundary (dashed line). Compare this with a similar field of view (F) from the untreated control embryo shown in A.

Fig. 4.

The expression of the HOXD4 and Hoxb-2 r3/5 transgenes in 11.5 d.p.c. embryos after either mock treatment (A) or RA treatment (C ) at 9.5 d.p.c.. (B) Dorsal view of embryo shown in A. (D) Dorsal view of RA-treated (9.5 d.p.c.) embryo shown in C. Again the anterior boundary of expression of construct no. 1 in the RA-treated embryo is more rostral than in the mock treated embryos.

Fig. 4.

The expression of the HOXD4 and Hoxb-2 r3/5 transgenes in 11.5 d.p.c. embryos after either mock treatment (A) or RA treatment (C ) at 9.5 d.p.c.. (B) Dorsal view of embryo shown in A. (D) Dorsal view of RA-treated (9.5 d.p.c.) embryo shown in C. Again the anterior boundary of expression of construct no. 1 in the RA-treated embryo is more rostral than in the mock treated embryos.

The HOXD4 enhancer is a retinoid-responsive element in vivo

Since the HOXD4 upstream region contains an enhancer mediating a response to RA in EC cells (Moroni et al., 1993), we wished to determine its ability to respond to exogenous RA administered in utero. Previously it had been shown that the Hoxb-1 gene contains a RARE which mediates a response to RA treatment in utero at 7.5-8.0 d.p.c., but not upon exposure in later periods (Conlon and Rossant, 1992; Marshall et al., 1992, 1994), so we initially used the same conditions to test for a response by the HOXD4 enhancer. Because the anterior boundary of transgene expression generated by the HOXD4 enhancer shifts slightly over time (Fig. 2), we mated a line carrying construct no. 1 to a second line which contained an r3/r5 specific enhancer of the mouse Hoxb-2 gene linked to a lacZ reporter (Sham et al., 1993) to create an internal standard. The r5 stripe of expression from the Hoxb-2 enhancer is a valid control, because it has previously been shown that this domain is not altered by RA treatment in utero (Marshall et al., 1992). Therefore, the anterior position of HOXD4 transgene expression, in both treated and untreated embryos, was always determined by proximity to the reference stripe of r5-specific lacZ gene expression. The dorsal and lateral views of control and RA-treated embryos clearly show that the anterior boundary of HOXD4 expression remains unchanged relative to r5 (Fig. 3A,B,D,E). Therefore, despite the presence of typical anterior phenotypic abnormalities induced by RA treatment at 7.5 d.p.c., there was no effect on lacZ expression in the lines carrying the HOXD4 construct no. 1 (Fig. 3A,B,D,E).

Mammalian Hox genes show a colinear, 3′→5′ response to retinoids in culture (Papalopulu et al., 1991; Simeone et al., 1990, 1991), and since HOXD4 is located in a more 5′ position in its cluster with respect to Hoxb-1, higher RA doses or a later period of exposure might be required for its induction in vivo. Higher RA concentrations administered at 7.5 d.p.c. had no effect on the lacZ expression driven by construct no. 1(not shown), however we did find stimulation when transgenic embryos were exposed to RA after 9.0 d.p.c. Pregnant females carrying embryos containing construct no. 1 and the Hoxb-2 r3/r5 lacZ transgene, were treated with all-trans-RA at 9.5 d.p.c., and embryos were harvested 12-24 hours (10.0-10.5 d.p.c.) after treatment to monitor transgene expression compared to untreated or mock-treated embryos of the same age. At both time points after treatment, construct no. 1 responded by shifting its expression in an anterior direction. At 10.0d.p.c. lacZ expression had clearly moved rostrally compared to that of untreated embryos, being closer to the r5 stripe of expression, with an anterior boundary of expression mapping at the r6/7 junction (se flat mounts Fig. 3C,F,I). We have also checked that the expression boundary of the endogenous Hoxd-4 gene moved anteriorly as a consequence of RA treatment under the same conditions (not shown), i.e. response to RA at late but not early stages of treatment.

Given that the normal time course of expression from the lines containing construct no. 1 showed a progressive anterior shift in expression up to 10.5-11.0 d.p.c., we wished to determine whether the RA induced changes initiated early (9.5 d.p.c.) persist or were even more extensive in later periods. We treated embryos carrying construct no. 1 and the Hoxb-2 r3/r5 lacZ transgene with RA at 9.5 d.p.c. and analysed the lacZ staining patterns at 11.5 d.p.c., 48 hours after treatment (Fig. 4A-D). It is not possible to observe the morphology of rhombomere boundaries in embryos older than 10.5 d.p.c., however using the r5 stripe of expression as a guide, the anterior boundary of expression from construct no. 1 still appears to lie at the r6/7 junction following RA treatment, while the boundary in untreated embryos maps more caudally, at the hindbrain/spinal cord junction (Fig. 4A,B). Therefore, the RA induced shifts in anterior expression initiated at 9.5 d.p.c. are stable and persist to later stages, and appear to be fixed at the r6/7 junction.

Together, these results indicate that the HOXD4 upstream fragment acts in vivo as a retinoid-responsive region, and is responsible for an anterior shift in the neural tube expression domain observed after exogenous administration of excess RA.

Retinoid response in vivo and in cell culture maps to a 185-bp conserved enhancer

Since the human HOXD4 upstream region mediates neuralrestricted gene expression and RA response in vivo in transgenic mice, as well as RA-dependent induction in NT2/D1 cells (Fig. 1, construct no. 1, and Moroni et al., 1993), we wanted to determine if these activities were related to each other, and under the control of similar factors interacting with similar cis-acting regions. It was previously shown that a 185-bp sequence, spanning position −2693 to −2508 from the proximal transcription start site of HOXD4, is highly conserved in the mouse Hoxd-4 upstream regulatory region, and can mediate response to all-trans- and 9-cis-RA in culture (Moroni et al., 1993; Pöpperl and Featherstone, 1993). Internal deletion of the conserved 185 bp sequence from the 2.9 kb HOXD4 upstream region (−2711 to −2333, construct no. 2, Fig. 1) completely abolished its activity in cell culture as well as in transgenic mice. Of eight transgenic founder embryos and one line obtained with this construct, none showed any expression in the neural tube. Two of these embryos however, did show some ectopic expression patterns, probably due to position effects at the site of integration, indicating that construct no. 2 was capable of activity in vivo (Fig. 5A).

Fig. 5.

The conserved 185-bp enhancer element is necessary and sufficient to produce the same expression pattern seen with the full-length HOXD4 upstream region. (A) A 10.5 d.p.c. transgenic embryo containing construct no. 2. No expression is observed in the neural tube although ectopic expression is seen in the superior colliculi and caudal mesoderm. This staining illustrates that the construct is capable of activity. (B) A comparison of the lacZ staining patterns obtained in 11.5 d.p.c. embryos carrying constructs no. 3 and no. 1 (indicated above each embryo). Expression is seen in the neural tube up to the same anterior boundary, again with a more posterior biphasic distribution indicated by the arrowheads. The additional expression in the lateral mesoderm and face of the embryo carrying construct no. 3 was not observed in other examples (see C), suggesting that this is due to position effects at the site of integration for this embryo. (C) Expression of a 10.5 d.p.c. embryo carrying construct no. 3. (D) A dorsal view of the embryo in C showing that the rostral limit to expression in the neural tube is caudal to the r6/7 boundary, which lies immediately behind the otic vesicle (OV). (E) Two 10.5 d.p.c. transgenic embryos carrying construct no. 4, showing an identical staining pattern to that seen with constructs no. 1 and no. 3.

Fig. 5.

The conserved 185-bp enhancer element is necessary and sufficient to produce the same expression pattern seen with the full-length HOXD4 upstream region. (A) A 10.5 d.p.c. transgenic embryo containing construct no. 2. No expression is observed in the neural tube although ectopic expression is seen in the superior colliculi and caudal mesoderm. This staining illustrates that the construct is capable of activity. (B) A comparison of the lacZ staining patterns obtained in 11.5 d.p.c. embryos carrying constructs no. 3 and no. 1 (indicated above each embryo). Expression is seen in the neural tube up to the same anterior boundary, again with a more posterior biphasic distribution indicated by the arrowheads. The additional expression in the lateral mesoderm and face of the embryo carrying construct no. 3 was not observed in other examples (see C), suggesting that this is due to position effects at the site of integration for this embryo. (C) Expression of a 10.5 d.p.c. embryo carrying construct no. 3. (D) A dorsal view of the embryo in C showing that the rostral limit to expression in the neural tube is caudal to the r6/7 boundary, which lies immediately behind the otic vesicle (OV). (E) Two 10.5 d.p.c. transgenic embryos carrying construct no. 4, showing an identical staining pattern to that seen with constructs no. 1 and no. 3.

A construct containing only the conserved 185 bp HOXD4 region linked to a basal Hoxb-4 minimal promoter/lacZ reporter was able to confer RA inducibility in EC cells in either orientation (construct no. 3 and 4, Fig. 1), at levels comparable to those obtained with the whole −2892/+21 region (Fig. 1). This confirms previous results obtained with larger fragments from either the human or the mouse gene, driving the herpes virus thymidine kinase or the adenovirus major late promoter respectively (Moroni et al., 1993; Pöpperl and Featherstone, 1993). The two minimal enhancer constructs were also sufficient to direct neural tube-specific lacZ expression in transgenic mice (Fig. 5B-E). Three out of seven transgenic founder embryos obtained with construct no. 3, and two out of four with construct no. 4, showed lacZ staining in the neural tube in a pattern indistinguishable from that seen with construct no. 1, containing the full-length 2.9-kb fragment. The minimal region contained in construct no. 3 and 4 contains neither of the human HOXD4 EC cell promoters, nor the region homologous to a putative autoregulatory element previously found in the mouse Hoxd-4 gene (Pöpperl and Featherstone, 1992). These results show that the conserved 185-bp enhancer is both necessary and sufficient to direct spatially restricted expression in the neural tube in vivo, and RA-inducibility in cell culture.

Molecular dissection of the HOXD4 enhancer

DNase I footprinting of the HOXD4 minimal enhancer with nuclear extracts from NT2/D1 cells identified four distinct elements, termed X, 1, 2 and 3, and several DNase I hypersensitive sites (Fig. 6). These footprints become evident after 3-7 days of RA induction, as compared to a wide, less distinct footprint covering the entire region observed in uninduced cells (Fig. 6A). Two of the four footprinted sequences contain direct repeats (DR) matching the PuG[G/T]TCA consensus sequence for nuclear hormone receptors (Leid et al., 1992), i.e., a TGTTCA(N3)AGGTCA (DR3) motif (in reversed orientation) in the footprinted element 2, and a AGGTGA(N5)AGGTCA (DR5) motif in element 3 (Fig. 6C). Furthermore, the footprinted element 1 contains a half-site TGTTCA motif, together with two sequences matching in 11 out of 13 nucleotides the 5′-GTTAATNATTAAC-3′ consensus binding site for the HNF1/LBF1 family of proteins (Courtois et al., 1988; De Simone et al., 1991) (Fig. 6C). Conversely, the element X contains no recognisable binding site for nuclear hormone receptors, nor for other known transcription factors. The factors actually binding element X, 1 and 2 are unknown, whereas the DR5 motif present in element 3 was previously shown to bind protein complexes containing RARs, and to be necessary for the RAR-dependent, RA-inducible activity of the HOXD4 enhancer in murine and human EC cells (Moroni et al., 1993; Pöpperl and Featherstone, 1993). Fig. 6A shows that the footprint on element 3 is generated by NT2/D1 nuclear extracts several days after RA exposure, indicating that availability of the ligand is not the only factor required for binding of the RAR-containing protein complexes on the HOXD4 enhancer.

Fig. 6.

(A) DNase I foot printing analysis of the HOXD4 upstream genomic region spanning nucleotides −2720 to −2449 from the proximal (P2) transcription start site. The 3′ end-labeled fragment was incubated with 12, 24, 36 or 54 μg of nuclear extracts from NT2/D1 cells, uninduced (−RA) or induced with 10 μM retinoic acid (+ RA 10 μM) for 1, 3, 7 or 10 days, before digestion with DNase I. Unprotected DNA is shown by −? A G+A Maxam and Gilbert sequence reaction is shown in the first and last lane. The regions protected by nuclear extracts of RA-induced NT2/D1 cells are indicated as X, 1, 2 and 3. Arrows indicate DNase I hypersensitive sites. (B) DNase I foot printing analysis of the same HOXD4 region, mutated in element 2, or deleted of element X or 3. Nuclear extracts were obtained from NT2/D1 cells induced by RA for 10 days, and used in the same amounts as in A. The point mutations in region 2 are indicated by arrowheads. The position of each deletion is indicated on the right side of the respective panels (ΔX or Δ3). A G+A sequence reaction is shown in the first and last lane in each panel. (C)Nucleotide sequence of the HOXD4 upstream region from position −2663 to position −2521 with respect to the proximal (P2) transcription start site (Cianetti et al., 1990). The regions footprinted in A and B are shaded, and indicated with X, 1, 2, and 3, and a symbolic notation (circle, triangle, diamond and square, respectively). The arrows indicate DNase I hypersensitive sites. The DR5 RARE in element 3, the DR3 site in element 2 and the hemisite in the element 1, are shown in bold face. Two putative HNF-1 binding sites are underlined in element 1. The sequence mutated in the pHOXD4-2892m2 constructs (see Fig. 1, construct no. 6) is indicated below the wild-type sequence. The regions deleted in the pHOXD4-2892ΔX and Δ3 mutants are indicated by brackets.

Fig. 6.

(A) DNase I foot printing analysis of the HOXD4 upstream genomic region spanning nucleotides −2720 to −2449 from the proximal (P2) transcription start site. The 3′ end-labeled fragment was incubated with 12, 24, 36 or 54 μg of nuclear extracts from NT2/D1 cells, uninduced (−RA) or induced with 10 μM retinoic acid (+ RA 10 μM) for 1, 3, 7 or 10 days, before digestion with DNase I. Unprotected DNA is shown by −? A G+A Maxam and Gilbert sequence reaction is shown in the first and last lane. The regions protected by nuclear extracts of RA-induced NT2/D1 cells are indicated as X, 1, 2 and 3. Arrows indicate DNase I hypersensitive sites. (B) DNase I foot printing analysis of the same HOXD4 region, mutated in element 2, or deleted of element X or 3. Nuclear extracts were obtained from NT2/D1 cells induced by RA for 10 days, and used in the same amounts as in A. The point mutations in region 2 are indicated by arrowheads. The position of each deletion is indicated on the right side of the respective panels (ΔX or Δ3). A G+A sequence reaction is shown in the first and last lane in each panel. (C)Nucleotide sequence of the HOXD4 upstream region from position −2663 to position −2521 with respect to the proximal (P2) transcription start site (Cianetti et al., 1990). The regions footprinted in A and B are shaded, and indicated with X, 1, 2, and 3, and a symbolic notation (circle, triangle, diamond and square, respectively). The arrows indicate DNase I hypersensitive sites. The DR5 RARE in element 3, the DR3 site in element 2 and the hemisite in the element 1, are shown in bold face. Two putative HNF-1 binding sites are underlined in element 1. The sequence mutated in the pHOXD4-2892m2 constructs (see Fig. 1, construct no. 6) is indicated below the wild-type sequence. The regions deleted in the pHOXD4-2892ΔX and Δ3 mutants are indicated by brackets.

The role of the single footprinted elements in the activity of the HOXD4 enhancer was first tested by transient transfection in NT2/D1 cells, in the context of the −2892/+21 genomic region with the two native promoters driving the lacZ reporter gene. Deletion of element X (−2654 to −2628, see Fig. 6C) had no significant effect on the basal activity of the reporter in uninduced NT2/D1 cells, whereas it caused a 70% reduction of its activity in RA-induced cells (construct no. 5, Fig. 1). Extensive mutation of the DR3 motif in element 2 (from TGACCTGTCTGAACA to GGGCCC GTCTAGACC, mutations in bold) had virtually no effect on the reporter activity in both uninduced and induced cells (construct no. 6, Fig. 1), while deletion of element 3 (−2556 to −2532, see Fig. 5C), abolished the RA response to the same extent as deletion of the whole enhancer (construct no. 7, Fig. 1). These results confirmed the essential role of the DR5-containing element 3 in mediating the ability of the HOXD4 enhancer to confer RA response in EC cells, and showed that element X, which contains no obvious RAREs nor binds RARs or RAR-containing complexes (Moroni et al., 1993), is nonetheless crucial for the enhancer function. Deletion of either element X or element 3, as well as mutation of the DR3 in element 2, had no effect on the relative spacing of the remaining elements, and caused no apparent effect on their ability to bind nuclear proteins in 7-day NT2/D1 extracts, as shown by DNase I footprinting (Fig. 6B). This suggests that binding of nuclear factors to the four elements occurs independently, and is not strictly based on co-operative interactions between each element to initiate binding.

We next tested the mutated constructs in transgenic mice. Of 13 transgenic founder embryos containing construct no. 5, deleted in the X element, none showed any expression in the neural tube. Again variable ectopic expression was observed in two embryos showing that the reporter gene was functional (Fig. 7A). Since deletion of the X element did not completely abolish the RA responsiveness of the enhancer in transfected EC cells, we tested whether construct no. 5 could still respond to exogenous RA in vivo. Seven of the 13 embryos obtained with construct no. 5 were also treated with RA in utero at 9.5 d.p.c.. None showed expression in the neural tube at 10.5 d.p.c., although ectopic expression was observed in one case (data not shown). This shows that the X element is essential for the enhancer activity in vivo. Thus while deletion of the X element leaves some residual enhancer activity in the transfection assay, it appears that this is insufficient to direct gene expression in the neural tube, even in the presence of exogenous RA.

Fig. 7.

The footprinted elements X and 3 are essential for lacZ expression in transgenic mice. (A)One of the 13 transgenic embryos carrying construct no. 5, none of which showed expression in the neural tube. The 10.5 d.p.c. embryo shown here has ectopic expression in the superior colliculi and around the otic vesicle, indicating that the reporter gene was capable of activity. (B) One of eight 10.5 d.p.c. embryos carrying construct no. 7. Again, no expression is observed in the neural tube. (C) A 10.5 d.p.c. embryo carrying construct no. 6 showing expression in the neural tube in a pattern identical to that seen with construct no. 1, suggesting that element 2 is not required for enhancer activity in transgenic mice.(D,E) Dorsal view and coronal section respectively of the embryo shown in A, showing that the anterior boundary of lacZ expression (arrowhead) relative to the otic vesicle (OV) and the r5 and 6 bulges. (F) The effect of RA treatment at 9.5 d.p.c. in utero on the expression pattern at 10.5 d.p.c. of a founder transgenic embryo carrying construct no. 6. (G,H) A dorsal view and coronal section respectively of the embryo shown in F. The RA-treated embryo containing construct no. 6 shows a rostral shift in the anterior lacZ expression boundary (arrowhead) towards the otic vesicle up to the r6/7 junction. This is the same response to RA seen with construct no. 1.

Fig. 7.

The footprinted elements X and 3 are essential for lacZ expression in transgenic mice. (A)One of the 13 transgenic embryos carrying construct no. 5, none of which showed expression in the neural tube. The 10.5 d.p.c. embryo shown here has ectopic expression in the superior colliculi and around the otic vesicle, indicating that the reporter gene was capable of activity. (B) One of eight 10.5 d.p.c. embryos carrying construct no. 7. Again, no expression is observed in the neural tube. (C) A 10.5 d.p.c. embryo carrying construct no. 6 showing expression in the neural tube in a pattern identical to that seen with construct no. 1, suggesting that element 2 is not required for enhancer activity in transgenic mice.(D,E) Dorsal view and coronal section respectively of the embryo shown in A, showing that the anterior boundary of lacZ expression (arrowhead) relative to the otic vesicle (OV) and the r5 and 6 bulges. (F) The effect of RA treatment at 9.5 d.p.c. in utero on the expression pattern at 10.5 d.p.c. of a founder transgenic embryo carrying construct no. 6. (G,H) A dorsal view and coronal section respectively of the embryo shown in F. The RA-treated embryo containing construct no. 6 shows a rostral shift in the anterior lacZ expression boundary (arrowhead) towards the otic vesicle up to the r6/7 junction. This is the same response to RA seen with construct no. 1.

Of 8 expressing embryos obtained with construct no. 6, in which the DR3 sequence of element 2 was mutated, all showed lacZ expression in the neural tube (Fig. 7C), in a pattern indistinguishable from that seen with constructs no. 1, 3 or 4. Founder embryos carrying construct no. 6 were also treated at 9.5 d.p.c. with RA, and analysed at 10.5 d.p.c. Fig. 6C-H shows, in whole-mount embryos and coronal sections, that treatment with RA induces a rostral shift in the anterior boundary of expression up to approximately the r6/7 junction, as with the wild-type enhancer. These data show that the DR3 in element 2 has no influence on the enhancer activity in vivo, in generating either the neural tube-specific expression or the response to exogenous RA.

Finally, we tested the role of the DR5-containing element 3 by generating transgenic mice with construct no. 7. Of a total of eight transgenic founder embryos containing this construct, none showed expression in the neural tube. One embryo did however show ectopic expression indicating the reporter was capable of functioning (Fig. 7B). Administration of exogenous RA in utero at 9.5 d.p.c. to these embryos had no effect on lacZ expression. Because we obtained no detectable expression in transgenics when either element 3 or X were mutated, unlike the basal level seen in EC cells when X is mutated (construct no. 5, Fig. 1), it is not possible to determine if elements X and 3 are both involved in the RA response. These experiments show that element 3 containing the DR5 RARE is also a critical component of the HOXD4 enhancer, and suggests that interactions between factors binding to element X and the RARE are important in directing neural-specific expression and RA response.

In this paper we show that a conserved 5′ flanking region of the HOXD4 gene active in EC cells is capable of directing spatially restricted neural expression in vivo. Expression of a lacZ transgene under the control of the 2.9-kb HOXD4 upstream region is activated in the neural tube with a graded distribution. The highest levels of staining appear in the anterior domain which has a border caudal to the otic vesicle, and the posterior boundary maps in the mid-thoracic region. Expression increases with time, and progressively spreads forward until it reaches an anterior boundary in the region of the hindbrain-spinal cord junction, which is just posterior to the anterior limit of expression of the endogenous Hoxd-4 gene at the r6/r7 junction (Gaunt et al., 1989; Hunt et al., 1991). Therefore, this pattern represents a major subset of the neural domain of endogenous Hoxd-4 expression (Fig. 2E,F), and suggests that this enhancer region normally regulates a component of Hoxd-4 neural expression.

In support of this idea, transgenic analysis of the paralogous gene, Hoxb-4, has shown that the domains of neural expression are not generated by a single enhancer element. Rather, the expression is the result of overlapping patterns produced by the combinatorial action of at least two different enhancers, one of which (Region A) specifies the correct anterior boundary of expression at the r6/r7 junction and the other (Region C) which produces the more posterior domain (Region C) (Aparicio et al., 1995; Whiting et al., 1991). The pattern of HOXD4 transgene expression closely resembles that generated by the Hoxb-4 region C enhancer, and we predict by analogy that the elements which regulate anterior expression at the r6/r7 boundary must reside elsewhere in the locus. Furthermore, analysis of the Hoxa-4 gene has also shown that multiple 5′ elements regulate its neural expression (Behringer et al., 1993), so the requirement for different enhancers in generating the full neural pattern may be a common property of the paralogous group 4.

We performed a detailed analysis of the ability of the HOXD4 regulatory region to respond to exogenous RA in utero, because of its RA sensitivity in EC cells (Moroni et al., 1993; Pöpperl and Featherstone, 1993). Our experiments clearly demonstrate that this region is able to mediate a rapid and stable anterior shift in neural-restricted HOXD4 transgene expression, up to the r6/r7 boundary (Figs 3 and 4). This response occurs when embryos are treated with RA at 9.5 d.p.c., but not when they are treated at 7.5 d.p.c.. This is interesting because other Hox genes, such as Hoxb-1 and Hoxb-2, have been shown to respond to RA treatments at 7.5 d.p.c., but not later stages (Conlon and Rossant, 1992; Marshall et al., 1992). This implies that there are temporal differences in the competence of specific Hox genes to respond to RA, and that these can be linked to particular enhancers.

Identical cis-acting elements are required for the HOXD4 enhancer activity in vivo and in EC cells

In a direct comparative analysis, we mapped sequences within the 2.9-kb HOXD4 upstream region which are necessary for its activity both in RA-induced EC cells and in transgenic mice. The region contains two previously identified domains which might be involved in the generation of the expression pattern we observed in vivo. These are a potential autoregulatory element at the 3′ end of the region (Pöpperl and Featherstone, 1992), and a conserved 185-bp transcriptional enhancer associated with RA response in EC cells (Moroni et al., 1993; Pöpperl and Featherstone, 1993). A possible reason why the entire HOXD4 upstream region functioned effectively in transgenic mice could have been that the autoregulatory element served as a readout for part of the endogenous Hoxd-4 expression. However, deletion of the conserved 185-bp block, leaving the autoregulatory element and the remaining sequences in the 2.9 kb region unaltered (construct no. 2 in Fig. 1), specifically abolished the spatially restricted expression even in the presence of exogenous RA in vivo. Hence, the autoregulatory motif is not sufficient for the expression we observed in transgenic mice, although it is able to direct regionspecific expression in transgenic flies (McGinnis et al., 1990). Furthermore, we found that the conserved 185-bp block alone could function as an enhancer in an orientation-independent manner linked to a heterologous promoter, and mediates all the regulatory activities observed in vivo and in cell culture with the full 2.9-kb fragment. Combined, these results indicate that the minimal 185-bp block is both necessary and sufficient for generating the spatially restricted expression pattern in the neural tube. It is possible that the autoregulatory element could be involved in controlling other aspects of HOXD4 expression, in conjunction with regions outside the 2.9-kb upstream fragment, but our analysis clearly shows that it has limited activity by itself and does not appear to work co-operatively with the 185-bp enhancer.

Previous foot printing analysis (Moroni et al., 1993), and work carried out in this study, show that nuclear extracts obtained from RA-induced EC cells protect four specific elements within the 185-bp enhancer. One of these (element 2) contains a DR3 consensus binding site of the type recognised by vitamin D or thyroid hormone receptors, but mutations abolishing the DNase I footprint in this element (Fig. 6B) have no effect on the enhancer activity or RA response in either transgenic mice or EC cells (construct no. 6; Figs 1 and 7C-H). Similarly, we have no evidence for a functional role of element 1, containing a consensus hormone receptor binding half-site, and this site appears not to be required in the homologous element of the mouse Hoxd-4 enhancer (Pöpperl and Featherstone, 1993). Conversely, mutation or deletion of either of the two remaining elements (X and 3) specifically abolished expression and RA response in both EC cells and transgenic mice (constructs no. 5 and no. 7 in Figs 1 and 7A,B). Therefore, an identical subset of elements within the conserved 185-bp enhancer, acting independently or co-operatively, are responsible for regulating the HOXD4 expression and RA response in both systems.

Requirement of a DR5 RARE for HOXD4 regulation

The response of the HOXD4 enhancer to exogenous RA in embryos and cell culture raised the question of whether the ectopic induction was mediated by the same or different elements from those required for spatially restricted expression in vivo. Our results demonstrated that footprint element 3, which contains a DR5 RARE that interacts with RARs (Moroni et al., 1993; Pöpperl and Featherstone, 1993), is absolutely required for the neural-restricted expression and RA response directed by the enhancer in transgenic mice and EC cells. There is no expression from constructs which specifically mutate this element (construct no. 7, Figs 1 and 7), implying that the DR5 RARE is likely to be essential in the normal regulation of part of the HOXD4 neural expression pattern.

In agreement with these findings in HOXD4, different RAREs have also been identified close to the most 3′ genes in two mouse complexes, Hoxa-1 (Frasch et al., 1995; Langston and Gudas, 1992) and Hoxb-1 (Marshall et al., 1994; Studer et al., 1994). The RAREs in Hoxb-1 are highly conserved in the human, mouse, chick and puffer fish homologous genes, implying that they are functionally important (Marshall et al., 1994; Ogura and Evans, 1995a,b; Studer et al., 1994). Indeed, the conserved DR2 RARE at the 3′ end of Hoxb-1 is required for establishing early neural expression in vivo and the early RA response (Marshall et al., 1994), which is similar to the role of the DR5 in HOXD4. Together these results of transgene expression show that retinoid-responsive elements of the DR2 and DR5 type are involved in the generation of neural-specific expression of multiple Hox genes. The fact that the sequences required for spatial aspects of Hox gene regulation are the same as those required for ectopic response to exogenous retinoids in both cultured cells and embryos, further supports the idea that endogenous retinoids are instrumental in establishment and spatial restriction of Hox gene expression.

The DR5 RARE requires additional factors for HOXD4 enhancer activity in vivo

The patterns of transgene expression we observed with the HOXD4 enhancer are not simply a readout of the DR5 ability alone to respond to endogenous retinoids. Multimerized RARE elements conferring an immediate-early response to ligand-activated RARs, such as the DR5 RARE from the RARβ gene, are capable of directing broad expression of a reporter gene in transgenic mice (Balkan et al., 1992; Colbert et al., 1993; Mendelsohn et al., 1991; Reynolds et al., 1991; Rossant et al., 1991). Furthermore, transgene expression directed by the RARβ retinoid-responsive element induced lacZ activity in the entire embryo upon exposure to excess RA in utero (Rossant et al., 1991). These types of observations previously reported for minimal RAREs contrast with those we obtained using the HOXD4 enhancer. In transgenic mice, deletion of the X element from the enhancer (construct no. 5 Fig. 7) completely abolishes spatially restricted expression, even though the DR5 RARE is still present. In addition, treatment with excess RA had no effect on the inability of construct no. 5 to direct expression in vivo. Hence, the HOXD4 DR5 RARE is not sufficient to mediate spatial expression in vivo or response to RA. In agreement with these findings, mutational analysis of the RAREs in Hoxb-1 has also shown that they are necessary but not sufficient for the function of their respective enhancers in transgenic mice (Marshall et al., 1994; Studer et al., 1994).

The essential role of element X in these experiments points to a critical requirement for additional factors in potentiation of the DR5-dependent activity in vivo. In cell culture, deletion of the X element significantly decreases, but does not abolish, the RA inducibility of the enhancer, suggesting that the X element may work by increasing the basal promoter activity upon which the DR5 element can impose an RA-dependent response. It is difficult to ascertain if the same is true in transgenic embryos, because the basal activity is undetectable in all (13) of the independent integrations with the same construct. The in vivo analysis suggests that the X element may, rather, be involved in co-operative interactions with the DR5 RARE, by affecting the binding specificity/affinity of RARs themselves or by recruiting other positive or negative acting proteins to the RARE sites. In fact, RARα, β and γ bind the isolated HOXD4 DR5 in vitro much less efficiently than the RARβ RARE (Moroni et al., 1993). In agreement with the requirement for additional factors, the ability of the human HOXB1 RAREs to confer an RA response in EC cells is also influenced by theunknown factor(s) (Ogura and Evans, 1995a,b). The X element has no homology with the accessory sequences surrounding the Hoxb-1 RAREs, nor does it contain binding sites for characterised transcription factors, so it will be important to identify these accessory factor(s) to understand the mechanism by which they modulate RARE activity.

Retinoid-responsive elements and activation of the Hox complexes

Hox genes are sequentially activated in development, display ordered or colinear domains of expression along the embryonic axis, and show a colinear response to retinoids in cell culture (reviewed by Krumlauf, 1994; McGinnis and Krumlauf, 1992). Understanding the basis of the sequential RA response in vitro could therefore be important in understanding the mechanisms which generate colinearity, a concept reinforced by the finding that the same RAREs required for RA response are also essential for spatial regulation of gene expression in vivo. However, multiple mechanisms are likely to be responsible for the retinoid-dependent activation of the Hox complexes (reviewed by Langston and Gudas, 1994). This could include indirect RA responses (Pöpperl et al., 1995), putative master 3′ RAREs involved in opening the complexes (Langston and Gudas, 1992), multiple RAREs with different potential (Frasch et al., 1995; Marshall et al., 1994; Studer et al., 1994), and a cascade of cross and autoregulatory interactions between the Hox genes (Faiella et al., 1994; Pöpperl et al., 1995; Pöpperl and Featherstone, 1992).

The results presented here show that in addition to the RAREs flanking the 3′-most genes, RARE-dependent enhancers capable of mediating restricted expression can also be present within a cluster. Since the 3′ genes are flanked by RARE-containing elements and show a most rapid and protein synthesis-independent response to RA treatment (La Rosa and Gudas, 1988; Simeone et al., 1991), it is logical to consider them as direct targets for the retinoid receptors. The late response in genes in the middle or at the 5′ end of the clusters could require indirect factors or progressive ‘opening’ of the cluster triggered by 3′ gene activation. We found that the endogenous Hoxd-4 gene, and the lacZ transgene under the control of the HOXD4 enhancer, do respond to RA treatment in utero, but under different conditions with respect to Hoxb-1. Ectopic expression of Hoxb-1 is activated by exposure to RA at 7.5 d.p.c., whereas Hoxd-4 and the Hoxd4-lacZ transgene respond only if embryos are treated two days later, at 9.0-9.5 d.p.c.. The fact that RAREs, and additional sequence elements, are required for both the early and late response, suggests that there is a restriction on the ability of the HOXD4 DR5 element to respond to ligand-activated RARs present during the earlier period. Similarly, additional factors in the Hoxb-1 enhancer may facilitate its early response and block activity later. Our in vitro binding studies show that the complex footprint observed on the HOXD4 enhancer in RA-induced EC cells is not generated immediately, as soon as the ligand is given to the cells. Rather there is a time lag (approx. 3 days) necessary for the factors binding to the four elements, including the DR5 RARE, to be assembled in the characteristic ‘induced’ configuration. This might entail de novo synthesis and/or post-translational modification of nuclear factors in addition to the RARs, required for binding and enhancer activity. Therefore, we favour the idea that specific trans-acting factors synergising with RARs could temporally regulate the activity of RAREs, and binding such factors could be one of the roles the X element plays in the HOXD4 enhancer.

In conclusion, this work has implicated a DR5 RARE and its co-operating components in the normal regulation of neural expression of the HOXD4 gene from a 5′ enhancer. It serves to provide further support for the role of endogenous retinoids in regulating normal domains and boundaries of Hox expression, particularly in the nervous system. Studying the induction of Hox genes by RA in EC cell culture appears to be a valuable means for investigating normal regulation of these genes, as we found that EC cells and mouse embryos require the same cis-elements for appropriate regulation. This is an important advantage as it provides a valid culture system for identifying factors, such as those interacting with region X, in the face of limited amounts of embryonic tissue.

This work was supported by grants from the European Community (Biotechnology Program) and the Italian National Research Council (CNR), Special Projects ‘Ingegneria Genetica’ and ‘ACRO’. A. M. was supported by a Medical Research Council Training Fellowship. We thank Mark Featherstone for the mouse Hoxd-4 cDNA probe for in situ hybridisation, Wendy Hatton for her help with the histology, and Zoe Webster and Lorraine Jones for help with animal husbandry. The expert assistance of Valeria Ganazzoli in typing the manuscript is hereby gratefully acknowledged.

Andrews
,
P. W.
,
Damjanov
,
I.
,
Simon
,
D.
,
Banting
,
D. S.
,
Carlin
,
C.
,
Dracopoli
,
N. C.
and
Fogh
,
J.
(
1984
).
Pluripotent embryonal carcinoma clones derived from the human teratocarcinoma cell line Tera-2. Differentiation in vivo and in vitro
.
Lab. Invest
.
50
,
147
162
.
Aparicio
,
S.
,
Morrison
,
A.
,
Gould
,
A.
,
Gilthorpe
,
J.
,
Chaudhurl
,
C.
,
Rigby
,
P.
,
Krumlauf
,
R.
and
Brenner
,
S.
(
1995
).
Detecting conserved regulatory elements with the model genome of the Japanese puffer fish, Fugu rubripes
.
Proc. Natl. Acad. Sci. USA
92
,
1684
1688
.
Balkan
,
W.
,
Colbert
,
M.
,
Bock
,
C.
and
Linney
,
E.
(
1992
).
Transgenic indicator mice for studying activated retinoic acid receptors during development
.
Proc. Natl. Acad. Sci. USA
89
,
3347
3351
.
Behringer
,
R.
,
Crotty
,
D. A.
,
Tennyson
,
V. M.
,
Brinster
,
R.
,
Palmiter
,
R.
and
Wolgemuth
,
D.
(
1993
).
Sequences 5′ of the homeobox of the Hox-1.4 gene direct tissue-specific expression of lacZ during mouse development
.
Development
117
,
823
833
.
Blomhoff
,
R.
,
Green
,
M. H.
,
Berg
,
T.
and
Norum
,
K. R.
(
1990
).
Transport and storage of vitamin A
.
Science
250
,
399
403
.
Chen
,
Y. P.
,
Huang
,
L.
,
Russo
,
A. F.
and
Solursh
,
M.
(
1992
).
Retinoic acid is enriched in Hensen′s node and is developmentally regulated in the early chicken embryo
.
Proc. Natl. Acad. Sci. USA
89
,
10056
10059
.
Cianetti
,
L.
,
Di Cristofaro
,
A.
,
Zappavigna
,
V.
,
Bottero
,
L.
,
Boccoli
,
G.
,
Testa
,
U.
,
Russo
,
G.
,
Boncinelli
,
E.
and
Peschle
,
C.
(
1990
).
Molecular mechanisms underlying the expression of the human HOX-5.1 gene
.
Nucl. Acids Res
.
18
,
4361
4368
.
Colbert
,
M. C.
,
Linney
,
E.
and
LaMantia
,
A. S.
(
1993
).
Local sources of retinoic acid coincide with retinoid-mediated transgene activity during embryonic development
.
Proc. Natl. Acad. Sci. USA
90
,
6572
6576
.
Conlon
,
R. A.
and
Rossant
,
J.
(
1992
).
Exogenous retinoic acid rapidly induces anterior ectopic expression of murine Hox-2 genes in vivo
.
Development
116
,
357
368
.
Courtois
,
G.
,
Baumhueter
,
S.
and
Crabtree
,
G. R.
(
1988
).
Purified hepatocyte nuclear factor 1 interacts with a family of hepatocyte-specific promoters
.
Proc. Natl. Acad. Sci. USA
85
,
7937
7941
.
De Simone
,
V.
,
De Magistris
,
L.
,
Lazzaro
,
D.
,
Gerstner
,
J.
,
Monaci
,
P.
,
Nicosia
,
A.
and
Cortese
,
R.
(
1991
).
LFB3, a heterodimer-forming homeoprotein of the LFB1 family is expressed in specialized epithelia
.
EMBO J
.
10
,
1435
1443
.
de Wet
,
J. R.
,
Wood
,
K. V.
,
DeLuca
,
M.
,
Helsinki
,
D. R.
and
Subramani
,
S.
(
1987
).
Firefly luciferase gene: structure and expression in mammalian cells
.
Mol. Cell. Biol
.
7
,
725
737
.
Dignam
,
J. R.
,
Lebovitz
,
R. M.
and
Roeder
,
R. G.
(
1983
).
Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei
.
Nucl. Acids Res
.
11
,
1475
1489
.
Duboule
,
D.
(
1992
).
The vertebrate limb: a model system to study the Hox/HOM gene network during development and evolution
.
BioEssays
14
,
375
384
.
Duboule
,
D.
and
Dollé
,
P.
(
1989
).
The structural and functional organization of the murine HOX gene family resembles that of Drosophila homeotic genes
.
EMBO J
.
8
,
1497
1505
.
Eid
,
R.
,
Koseki
,
H.
and
Schughart
,
K.
(
1993
).
Analysis of lacZ reporter genes in transgenic embryos suggests the presence of several cis-acting regulatory elements in the murine Hox-b6 gene
.
Dev. Dyn
.
196
,
205
216
.
Faiella
,
A.
,
Zappavigna
,
V.
,
Mavilio
,
F.
and
Boncinelli
,
E.
(
1994
).
Inhibition of retinoic acid-induced activation of 3′ human HOXB genes by antisense oligonucleotides affects sequential activation of genes located upstream in the four HOX cluster
.
Proc. Natl. Acad. Sci. USA
91
,
5335
5339
.
Featherstone
,
M. S.
,
Baron
,
A.
,
Gaunt
,
S. J.
,
Mattei
,
M. G.
and
Duboule
,
D.
(
1988
).
Hox-5.1 defines a homeo-box containing gene locus on mouse chromosome 2
.
Proc. Natl. Acad. Sci. USA
85
,
4760
4764
.
Frasch
,
M.
,
Chen
,
X.
and
Lufkin
,
T.
(
1995
).
Evolutionary-conserved enhancers direct region-specific expression of the murine Hoxa-1 and Hoxa-2 loci in both mice and Drosophila
.
Development
121
,
957
974
.
Gaunt
,
S. J.
,
Krumlauf
,
R.
and
Duboule
,
D.
(
1989
).
Mouse homeogenes within a subfamily Hox-1.4, -2.6 and -5.1 dispaly similar anteroposterior domains of expression in the embryo, but show stage-and time-dependent differences in their regulation
.
Development
107
,
131
134
.
Gehring
,
W. J.
,
Müller
,
M.
,
Affolter
,
M.
,
Percival-Smith
,
A.
,
Billeter
,
M.
,
Qian
,
Y. Q.
,
Otting
,
G.
and
Wüthrich
,
K.
(
1990
).
The structure of the homeodomain and its functional implications
.
Trends Genet
.
6
,
323
329
.
Gerard
,
M.
,
Duboule
,
D.
and
Zakany
,
J.
(
1993
).
Structure and activity of regulatory elements involved in the activation of the Hoxd-11 gene during late gastrulation
.
EMBO J
.
12
,
3539
3550
.
Graham
,
A.
,
Papalopulu
,
N.
and
Krumlauf
,
R.
(
1989
).
The murine and drosophila homeobox gene complexes have common features of organization and expression
.
Cell
57
,
367
378
.
Hogan
,
B. L. M.
,
Costantini
,
F.
and
Lacy
,
E.
(
1986
).
Manipulating the Mouse Embryo. A Laboratory Manual
.
Cold Spring Harbor, NY
:
Cold Spring Harbor Laboratory Press
.
Hogan
,
B. L. M.
,
Thaller
,
C.
and
Eichele
,
G.
(
1992
).
Evidence that the Hensen′s node is a site of retinoic acid synthesis
.
Nature
359
,
237
241
.
Horton
,
G.
and
Maden
,
M.
(
1995
).
Endogenous distribution of retinoids during normal development and teratogenesis in the mouse embryo
.
Dev. Dyn
.
202
,
312
323
.
Hunt
,
P.
,
Gulisano
,
M.
,
Cook
,
M.
,
Sham
,
M. H.
,
Faiella
,
A.
,
Wilkinson
,
D.
,
Boncinelli
,
E.
and
Krumlauf
,
R.
(
1991
).
A distinct Hox code for the branchial region of the vertebrate head
.
Nature
353
,
861
864
.
Izpisúa-Belmonte
,
J.-C.
,
Tickle
,
C.
,
Dollé
,
P.
,
Wolpert
,
L.
and
Duboule
,
D.
(
1991
).
Expression of the homeobox Hox-4 genes and the specification of position in chick wing development
.
Nature
350
,
585
589
.
Kessel
,
M.
(
1992
).
Respecification of vertebral identities by retinoic acid
.
Development
115
,
487
501
.
Kessel
,
M.
and
Gruss
,
P.
(
1991
).
Homeotic transformations of murine vertebrae and concomitant alteration of Hox codes induced by retinoic acid
.
Cell
67
,
89
104
.
Kress
,
C.
,
Vogels
,
R.
,
De Graaf
,
W.
,
Bonnerot
,
C.
,
Meijlink
,
F.
,
Nicolas
,
J. F.
and
Deschamps
,
J.
(
1990
).
HOX 2.3 upstream sequences mediate lacZ expression in intermediate mesoderm derivatives of transgenic mice
.
Development
109
,
775
786
.
Krumlauf
,
R.
(
1993
).
Hox genes and pattern formation in the branchial region of the vertebrate head
.
Trends Genet
.
9
,
106
112
.
Krumlauf
,
R.
(
1994
).
Hox genes in vertebrate development
.
Cell
78
,
191
201
.
La Rosa
,
G. J.
and
Gudas
,
L. J.
(
1988
).
An early effect of retinoic acid: cloning of an mRNA (ERA-1) exhibiting rapid and protein synthesis-indipendent induction during teratorcarcinoma stem cell differentiation
.
Proc. Natl Acad. Sci. USA
85
,
329
333
.
Langston
,
A. W.
and
Gudas
,
L. J.
(
1992
).
Identification of a retinoic acid responsive enhancer 3′ of the murine homeobox gene Hox-1.6
.
Mech. Dev
.
38
,
217
228
.
Langston
,
A. W.
and
Gudas
,
L. J.
(
1994
).
Retinoic acid and homeobox gene regulation
.
Curr. Opin. Genet. Dev
.
4
,
550
555
.
Leid
,
M.
,
Kastner
,
P.
and
Chambon
,
P.
(
1992
).
Multiplicity generates diversity in the retinoic acid signalling pathways
.
Trends Biochem. Sci
.
17
,
427
433
.
Marshall
,
H.
,
Nonchev
,
S.
,
Sham
,
M. H.
,
Muchamore
,
I.
,
Lumsden
,
A.
and
Krumlauf
,
R.
(
1992
).
Retinoic acid alters hindbrain Hox code and induces transformation of rhombomeres 2/3 into a 4/5 identity
.
Nature
360
,
737
741
.
Marshall
,
H.
,
Studer
,
M.
,
Pöpperl
,
H.
,
Aparicio
,
S.
,
Kuroiwa
,
A.
,
Brenner
,
S.
and
Krumlauf
,
R.
(
1994
).
A conserved retinoic acid response element required for early expression of the homeobox gene Hoxb-1
.
Nature
370
,
567
571
.
Mavilio
,
F.
,
Simeone
,
A.
,
Giampaolo
,
A.
,
Faiella
,
A.
,
Zappavigna
,
V.
,
Acampora
,
D.
,
Poiana
,
G.
,
Russo
,
G.
,
Peschle
,
C.
and
Boncinelli
,
E.
(
1986
).
Differential and stage-related expression in embryonic tissues of a new human homeobox gene
.
Nature
324
,
664
666
.
McGinnis
,
N.
,
Kuziora
,
M. A.
and
McGinnis
,
W.
(
1990
).
Human Hox 4.2 and Drosophila Deformed encode similar regulatory specificities in Drosophila embryos and larvae
.
Cell
63
,
969
976
.
McGinnis
,
W.
and
Krumlauf
,
R.
(
1992
).
Homeobox genes and axial patterning
.
Cell
68
,
283
302
.
Mendelsohn
,
C.
,
Ruberte
,
E.
,
LeMeur
,
M.
,
Morriss-Kay
,
G.
and
Chambon
,
P.
(
1991
).
Developmental analysis of the retinoic acid-inducible RAR-beta 2 promoter in transgenic animals
.
Development
113
,
723
734
.
Moroni
,
M. C.
,
Vigano
,
M. A.
and
Mavilio
,
F.
(
1993
).
Regulation of the human HOXD4 gene by retinoids
.
Mech. Dev
.
44
,
139
154
.
Morrison
,
A.
,
Chaudhuri
,
C.
,
Ariza-McNaughton
,
L.
,
Muchamore
,
I.
,
Kuroiwa
,
A.
and
Krumlauf
,
R.
(
1995
).
Comparative analysis of chicken Hoxb-4 regulation in transgenic mice
.
Mech. Dev
.
63
,
47
59
.
Nohno
,
T.
,
Noji
,
S.
,
Koyama
,
E.
,
Ohyama
,
K.
,
Myokai
,
F.
,
Kuroiwa
,
A.
,
Saito
,
T.
and
Taniguchi
,
S.
(
1991
).
Involvement of the Chox-4 chiken homeobox genes in determination of anteroposterior axial polarity during limb development
.
Cell
64
,
1197
1205
.
Nordeen
,
S. K.
(
1988
).
Luciferase reporter gene vectors for analysis of promoters and enhancers
.
Biotechniques
6
,
454
457
.
Ogura
,
T.
and
Evans
,
M. R.
(
1995a
).
Evidence for two distinct retinoic acid response pathways for HOXB1 gene regulation
.
Proc. Natl. Acad. Sci. USA
92
,
392
396
.
Ogura
,
T.
and
Evans
,
R. M.
(
1995b
).
A retinoic acid-triggered cascade of HOXB1 gene activation
.
Proc. Natl. Acad. Sci. USA
92
,
387
391
.
Oliver
,
G.
,
De Robertis
,
E. M.
,
Wolpert
,
L.
and
Tickle
,
C.
(
1990
).
Expression of a homeobox gene in the chick wing bud following application of retinoic acid and grafts of polarizing region tissue
.
EMBO J
.
9
,
3093
3099
.
Papalopulu
,
N.
,
Lovell-Badge
,
R.
and
Krumlauf
,
R.
(
1991
).
The expression of murine Hox-2 genes is dependent on the differentiation pathway and displays a colinear sensitivity to retinoic acid in F9 cells and Xenopus embryos
.
Nucl. Acids Res
.
19
,
5497
5506
.
Pöpperl
,
H.
,
Bienz
,
M.
,
Studer
,
M.
,
Chang
,
S.-K.
,
Aparicio
,
S.
,
Brenner
,
S.
,
Mann
,
R.
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
.
Pöpperl
,
H.
and
Featherstone
,
M. S.
(
1992
).
An autoregulatory element of the murine Hox-4.2 gene
.
EMBO J
.
11
,
3673
3680
.
Pöpperl
,
H.
and
Featherstone
,
M. S.
(
1993
).
Identification of a retinoic acid response element upstream of the murine Hox-4.2 gene
.
Mol. Cell. Biol
.
13
,
257
265
.
Püschel
,
A. W.
,
Balling
,
R.
and
Gruss
,
P.
(
1990
).
Position-specific activity of the Hox1.1 promoter in transgenic mice
.
Development
108
,
435
442
.
Püschel
,
A. W.
,
Balling
,
R.
and
Gruss
,
P.
(
1991
).
Separate elements cause lineage restriction and specific boundaries of Hox 1.1 expression
.
Development
112
,
279
287
.
Reynolds
,
K.
,
Mezey
,
E.
and
Zimmer
,
A.
(
1991
).
Activity of the beta-retinoic acid receptor promoter in transgenic mice
.
Mech. Dev
.
36
,
15
29
.
Rossant
,
J.
,
Zirngibl
,
R.
,
Cado
,
D.
,
Shago
,
M.
and
Giguere
,
V.
(
1991
).
Expression of retinoic acid response element-hsp lacZ transgene defines specific domains of transcriptional activity during mouse embryogenesis
.
Genes Dev
.
5
,
1333
1344
.
Sambrook
,
J.
,
Fritsch
,
E. F.
and
Maniatis
,
T.
(
1989
).
Molecular Cloning: A Laboratory Manual
. 2nd ed., pp.
Cold Spring Harbor, NY
:
Cold Spring Harbor Laboratory Press
.
Saurat
,
J. H.
(
1990
).
Retinoids: 10 years on
.
Basel
:
S. Karger
.
Sham
,
M. H.
,
Hunt
,
P.
,
Nonchev
,
S.
,
Papalopulu
,
N.
,
Graham
,
A.
,
Boncinelli
,
E.
and
Krumlauf
,
R.
(
1992
).
Analysis of the murine Hox-2.7 gene: conserved alternative transcripts with differential distributions in the nervous system and the potential for shared regulatory regions
.
EMBO J
.
11
,
1825
1836
.
Sham
,
M. H.
,
Vesque
,
C.
,
Nonchev
,
S.
,
Marshall
,
H.
,
Frain
,
M.
,
Das Gupta
,
R.
,
Whiting
,
J.
,
Wilkinson
,
D.
,
Charnay
,
P.
and
Krumlauf
,
R.
(
1993
).
The zinc finger gene Krox-20 regulates Hox-b2 (Hox-2.8) during hindbrain segmentation
.
Cell
72
,
183
197
.
Simeone
,
A.
,
Acampora
,
D.
,
Arcioni
,
L.
,
Andrews
,
P. W.
,
Boncinelli
,
E.
and
Mavilio
,
F.
(
1990
).
Sequential activation of HOX 2 homeobox genes by retinoic acid in human embryonal carcinoma cells
.
Nature
346
,
763
766
.
Simeone
,
A.
,
Acampora
,
D.
,
Nigro
,
V.
,
Faiella
,
A.
,
D′Esposito
,
M.
,
Stornaiuolo
,
A.
,
Mavilio
,
F.
and
Boncinelli
,
E.
(
1991
).
Differential regulation by retinoic acid of the homeobox genes of the four Hox loci in human embryonal carcinoma cells
.
Mech. Dev
.
33
,
215
228
.
Simeone
,
A.
,
Avantaggiato
,
V.
,
Moroni
,
M. C.
,
Mavilio
,
F.
,
Arra
,
C.
,
Cotelli
,
F.
,
Nigro
,
V.
and
Acampora
,
C.
(
1995
).
Retinoic acid induces stage-specific anteroposterior transformation of rostral central nervous system
.
Mech. Dev
.
51
,
83
98
.
Studer
,
M.
,
Pöpperl
,
H.
,
Marshall
,
H.
,
Kuroiwa
,
A.
and
Krumlauf
,
R.
(
1994
).
Role of a conserved retinoic acid response element in rhombomere restriction of Hoxb-1
.
Science
256
,
1728
1732
.
Summerbell
,
D.
and
Maden
,
M.
(
1990
).
Retinoic acid, a developmental signalling molecule
.
Trends Neurosci
.
13
,
142
147
.
Tabin
,
C. J.
(
1991
).
Retinoids, Homeoboxes and Growth Factors: toward molecular models for limb development
.
Cell
66
,
199
217
.
Thaller
,
C.
and
Eichele
,
G.
(
1987
).
Identification and spatial distribution of retinoids in the developing chick limb bud
.
Nature
397
,
625
628
.
Thaller
,
C.
and
Eichele
,
G.
(
1990
).
Isolation of 3,4-didehydroretinoic acid, a novel morphogenetic signal in the chick wing bud
.
Nature
345
,
815
819
.
Tuggle
,
C. K.
,
Zàkàny
,
J.
,
Cianetti
,
L.
,
Peschle
,
C.
and
Nguyen-Huu
,
C. M.
(
1990
).
Region-specific enhancers near two mammalian homeobox genes define adjacent rostrocaudal domains in the central nervous system
.
Genes Dev
.
4
,
180
189
.
Vogels
,
R.
,
Charite
,
J.
,
de Graaff
,
W.
and
Deschamps
,
J.
(
1993
).
Proximal cis-acting elements cooperate to set Hox-b7 (Hox-2.3) expression boundaries in transgenic mice
.
Development
117
,
71
82
.
Wagner
,
M.
,
Han
,
B.
and
Jessel
,
T. M.
(
1992
).
Regional differences in retinoid release from embryonic neural tissue detected by an in vitro reporter assay
.
Development
116
,
55
66
.
Wagner
,
M.
,
Thaller
,
C.
,
Jessel
,
T.
and
Eichele
,
G.
(
1990
).
Polarizing activity and retinoid synthesis in the floor plate of the neural tube
.
Nature
345
,
819
822
.
Whiting
,
J.
,
Marshall
,
H.
,
Cook
,
M.
,
Krumlauf
,
R.
,
Rigby
,
P.
,
Stott
,
D.
and
Allemann
,
R.
(
1991
).
Multiple spatially-specific enhancers are required to reconstruct the pattern of Hox-2.6 gene expression
.
Genes Dev
.
5
,
2048
2059
.
Wilkinson
,
D. G.
(
1992
).
Whole Mount in Situ Hybridisation of Vertebrate Embryos
. pp.
75
83
.
Oxford, UK
:
Oxford University Press
.