The Direct Context of a Hox Retinoic Acid Response Element Is Crucial for its Activity*

During embryogenesis, target genes of retinoid signaling are able to respond differently to identical concentrations of retinoids. Small differences in the retinoic acid response elements (RARE) may be essential for these distinct responses. Recently, we identified a RARE in a Hox enhancer (dubbed distal element) that is active relatively late during mouse development. We now show that the RARE motif in the distal element is necessary and sufficient for the induction of gene expression by retinoic acid (RA) in P19 embryonic carcinoma cells. Furthermore, the significance of these results was established by RA treatment of transgenic mouse lines carrying the distal element containing the wild-type or a mutated RARE. We compared the in vitro activity of the distal element-RARE with that of the direct repeat with 5-bp spacer RARE of the RARβ2 gene, which is active during early during mouse development. We found that these RAREs, despite their similarity, responded differently to RA. By making single point mutations we show that the specificity resides in their retinoid X receptor-binding sites and is determined by base pairs located just outside the RARE consensus sequence. We suggest that the context of RARE motifs is important for the distinct transcriptional activities of genes under control of retinoid signaling.

During embryogenesis, target genes of retinoid signaling are able to respond differently to identical concentrations of retinoids. Small differences in the retinoic acid response elements (RARE) may be essential for these distinct responses. Recently, we identified a RARE in a Hox enhancer (dubbed distal element) that is active relatively late during mouse development. We now show that the RARE motif in the distal element is necessary and sufficient for the induction of gene expression by retinoic acid (RA) in P19 embryonic carcinoma cells. Furthermore, the significance of these results was established by RA treatment of transgenic mouse lines carrying the distal element containing the wild-type or a mutated RARE. We compared the in vitro activity of the distal element-RARE with that of the direct repeat with 5-bp spacer RARE of the RAR␤2 gene, which is active during early during mouse development. We found that these RAREs, despite their similarity, responded differently to RA. By making single point mutations we show that the specificity resides in their retinoid X receptorbinding sites and is determined by base pairs located just outside the RARE consensus sequence. We suggest that the context of RARE motifs is important for the distinct transcriptional activities of genes under control of retinoid signaling.
During vertebrate development, retinoic acid (RA) 1 signaling is essential for the correct patterning of embryonic structures (1)(2)(3). There has been a long-standing debate whether or not RA acts in a gradient and provides in such a way positional information that is needed for the proper patterning of the embryo (4 -6). Recently, evidence has been obtained that a graded distribution of RA in the hindbrain is essential for the specification of rhombomere identity (6). This process of rhombomere specification requires that distinct downstream gene targets of RA signaling must be able to respond differently to identical levels of retinoids.
RA-induced alterations in gene transcription are mediated by nuclear receptors, the RARs (␣, ␤, and ␥) and RXRs (␣, ␤, and ␥), that act as ligand-responsive transcriptional regulators (reviewed in Ref. 7). RAR/RXR heterodimers bind to DNA through sequence motifs known as retinoic acid response elements (RAREs; Ref. 8). Like many other hormone response elements (HREs), the RAREs consist of two half-sites arranged in tandem repeats. The specificity of the HREs for a given nuclear receptor is determined by sequence, spacing, and relative orientation of these half-sites (7,9). RAR/RXR heterodimers bind preferentially to direct repeats with the consensus sequence AGGTCA spaced by 1, 2, or 5 bp (DR1, -2, or -5; Refs. 7 and 10). Although naturally occurring RAREs have been identified that have divergent half-site configurations or sequence, the TCA motif is usually present (11,12). The binding of RAR/RXR heterodimers to the RAREs is co-operative. It is thought that the RARE motif aligns the RAR and RXR subunits and induces conformational changes that lead to stable interactions within this ternary complex (13,14). Moreover, protein binding to one of the half-sites may induce structural distortions in the DNA that facilitate protein binding to the second site (14). These co-operative interactions suggest that the heterodimers do not pre-assemble before DNA-binding, but rather assemble on the appropriate RAREs. Beside the interaction in the ternary complex response elements also affect the interactions of the receptors with their ligands and other protein factors important for gene regulation (10,15,16). These possible allosteric modifications by the DNAbinding sites may cause a RARE-dependent transcriptional response upon ligand binding (16).
Transgenic experiments in mice reveal a difference in response of two distinct RAREs present in the Hox cluster. When in the Hoxb4 early neural enhancer the DR5 RARE is replaced by a functional Hoxb1 DR2 RARE, the expression pattern of the reporter construct in the neural tube acquires the features of those of Hoxb1 (12). This experiment suggests that the RARE itself, and not its context, is crucial for the interpretation of the graded RA distribution.
Recently we discovered a novel functional DR5 RARE in a late neural enhancer, called the distal element (DE), which is able to mediate the expression of 5Ј Hoxb genes (17). This 533-bp DE drives reporter gene expression in a RAR␤2-like pattern (17)(18)(19). Interestingly, the RAR␤2 promoter is under control of a DR5 RARE as well (20,21). Although the late neural expression patterns are the same, the DE is not active in embryos younger than E10.0, whereas the RAR␤2 RARE starts to be active between E7.5 and E8.5 (17,22). The distinct temporal activities of both RAREs could reside in sequence differences in the RARE themselves as both half-sites of the RAREs differ in one base pair from each other.
In this study we directly compared the activities of the RAR␤2 and DE-RAREs. We demonstrate in P19 embryonic carcinoma (EC) cells and mouse embryos a RA sensitivity of the * This work was supported by Grants NWO-ALW-805-17.263 (to F. M.) and 809.38.05 (to J. D.) from the Netherlands Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Hubrecht Laboratory, Netherlands Inst. for Developmental Biology, Uppsalalaan 8, 3584 CT Utrecht, The Netherlands. Tel.: 31-30-212-1800; Fax: 31-30-251-6464; E-mail: frits@niob.knaw.nl. 1 The abbreviations used are: RA, retinoic acid; RAR, retinoic acid receptor; RARE, retinoic response element; RXR, retinoid X receptor (9-cis retinoic acid receptor); DE, distal element; DR5, direct repeat with 5-bp spacer; EMSA, electrophoretic mobility shift assay; HRE, hormone response element; EC, embryonic carcinoma. DE, which depends on the presence of an intact RARE. We found that in P19 cells the DE-RARE is far less sensitive to RA than the RARE of the RAR␤2 gene. By comparing 25 bp RAREcontaining elements derived from the RAR␤2 promoter and the DE, we found that the RXR-binding site and predominantly the direct context of these motifs account for the observed differences.

EXPERIMENTAL PROCEDURES
RA Exposure of Embryos in Utero, Whole Mount X-Gal Staining, and in Situ Hybridization-RA administration to pregnant mothers by oral gavage was performed as described (23). The procedures for the whole mount X-Gal staining and in situ hybridization with a Hoxb5 probe have been described in Ref. 17. All animal experiments were conducted under the approval of the animal care committee of the KNAW (Royal Dutch Academy of Arts and Sciences).
DNA Constructs-The basic luciferase reporter construct used was a modification of the p19Luc plasmid (24), which is named 19E1bluc and contains a fusion between the TATA box of the E1b promoter and the luciferase gene. Subfragments of the complete DE (25) and their mutated forms were cloned as EcoRV-HincII fragments into the HincII site of p19E1bluc. The subfragments of the DE studied were in most cases inserted in both orientations, but we found no influence of the orientation of these fragments on the activity of the resulting constructs. Therefore we only show the data corresponding to one orientation of the DE and its mutated versions. The DE with mutations in both half-sites was described previously (17), and the mutations made separately in each half-sites were made in a similar way and checked by sequencing. Two copies of the 25-bp elements containing the RARE (see "Results" section for exact sequence) were cloned as EcoRI-SalI fragments in the SalI site of the p19E1bluc vector. These constructs were subjected to sequence analysis for verification. Cell Culture and Transient Transfections-Culturing of P19 EC cells and cell transfections and luciferase and ␤-galactosidase assays were cultured as described by Folkers et al. (26,27), with minor modifica- tions. Cells were transfected in 24-well plates with 2 g of luciferase reporter construct and 1 g of pSV2lacZ. Lysates were prepared with 300 l of lysis buffer per well for 20 min at 4°C, and 50 and 25 l of the lysate was immediately used for the luciferase and ␤-galactosidase assay, respectively. Results are the mean (ϮS.D.) of at least three independent experiments of three to four transfections for each construct. Relative induction was obtained by dividing the mean of each reporter construct by the mean of the empty luciferase construct. Percentage induction was calculated by dividing the activity of a particular construct by the sum of activities of all constructs in one experiment.
Electrophoretic Mobility Shift Assays (EMSA)-The EMSAs were performed as described (28). The sequences of the used oligonucleotides, labeled by filling in a 5Ј AGG sticky end, were as follows:  (30 -33). Because the DE is particularly active in neural cells, we used this system to directly compare the activities of the RAREs derived from the RARE␤2 promoter and the DE.

RA Sensitivity of the Distal Element in P19 EC cells and Transgenic Mice Depends on the RARE-In P19 EC cells treated with RA, transcription of a luciferase reporter construct
under control of the DE is weakly but significantly (p ϭ 0.05) induced (Fig. 1A). To test whether the RA sensitivity was dependent on the RARE we made point mutations in either or both half-sites of the RARE. Mutations in both half-sites abolish RAR/RXR heterodimer binding (Fig. 1B, DEmut). EMSAs show that mutations in either of the half-sites (DE5Јmut and DE3Јmut) almost completely abolish RAR␥/RXR␣ binding (Fig.  1C). As expected, the introduction of these different mutations in the DE resulted in loss of RA sensitivity (Fig. 1A). Taken together, these data demonstrate that the RARE motif is essential for the RA sensitivity of the DE in P19 EC cells.
To further confirm the in vivo relevance, we used transgenic mouse lines carrying a Hoxb8-LacZ reporter construct driven by the DE or DEmut ( Fig. 2A; see Ref. 17). These constructs include an enhancer that drives the expression in more posterior embryonic regions (34) to allow us to verify that the integration site was permissive for expression. Previously we showed that lacZ expression in the hindbrain mediated by the DE depends on Raldh2 (17). Now we investigated its sensitivity in vivo to an excess of exogenous RA. RA treatment at E9.5 of transgenic embryos carrying construct 3 (carrying an intact DE; see Fig. 2A) led at E10.5 to strong activation of the reporter gene in the hindbrain (Fig. 2, B and C). In contrast, similar treatment of embryos carrying construct 4, which contains the mutated DE (DEmut) did not result in lacZ induction (Fig. 2, D  and E). Proper RA treatment was verified by analysis of Hoxb5 response, and the results were confirmed in a second transgenic line carrying the same construct (not shown). These in vivo data confirm the significance of the RA response of the DE observed in the cell transfection assays.
The DE-RARE Is Sufficient to Induce RA-dependent Gene Transcription-We compared the RA-induced activities of the RAREs derived from the DE and RAR␤2 in P19 EC cells. We used an oligonucleotide of 25 bp containing the DR5 RAR␤2-RARE and the flanking base pairs that are evolutionary conserved between mouse and human (18,22). This particular element is sensitive to RA starting early during mouse development (22). The RA-dependent activity of two copies of this 25 bp element was much higher than that of an identical configuration of a 25-bp element including the DE-RARE and the same number of its flanking nucleotides (Fig. 3). The activity of the DE-RARE was significantly increased upon RA treatment (p ϭ 0.05) but was lower than that of the complete DE both in the absence and presence of RA. The higher DE activity in the absence of RA is probably due to the presence of other (cryptic) binding sites for transactivators. These results show that other sequences in the DE may influence the activity of the RARE but that they are dispensable for its RA sensitivity.
The Distinct Activities of the RAREs of the RAR␤2 Gene and the DE Largely Depend on the Single-nucleotide Difference in the RXR-binding Site and on their Direct Context-In striking contrast with the large difference in RA sensitivity between the DE-RARE and the RAR␤2-RARE, the sequences of these DR5 RAREs differ by only one bp in each half-site. This should make it basically simple to pinpoint the nucleotides responsible for these distinct sensitivities. We therefore made a series of reporter constructs with two copies of the DE-RARE in which the corresponding RAR␤2-specific nucleotide change was introduced in either or both of the half-sites (Fig. 4). In addition, in one construct both of the half-sites of the RAR␤2-RARE were converted to a DE-RARE by two single point mutations.
As shown in Fig. 4, when the RAR-binding site of the DE was converted into that of the RAR␤2-RARE (construct DE␤A) the RA sensitivity of this mutated DE was reduced to levels not significantly higher than that of the control (Fig. 4, ϪRA). In contrast, the transformation of the RXR-binding site of the DE into that of the RAR␤2 (construct DE␤X) resulted in a very strongly enhanced RA sensitivity, to a level significantly (p ϭ 0.05) higher than that of the RAR␤2-RARE. Surprisingly, converting both binding sites of the DE to the corresponding ␤2-RARE sites (DE␤XA) led to a level of sensitivity even marginally higher that that of DE␤X, whereas conversion of the RAR site alone (DE␤A) resulted in an element that retained only about 20% of the activity of the DE-RARE (DE). Accordingly, changing both half-sites of the RAR␤2-RARE into that of the DE (␤2XADE) led to a decrease in RA sensitivity.
We noted that without additional RA the DE␤X and DE␤XA elements had higher basal activities than DE and the RAR␤2 RARE. One possibility could be that the point mutations introduced in the RXR-binding site create a binding site for another transactivator. Because these motifs mediate high RA sensitivity, a more likely explanation is that they sense the RA present in the fetal calf serum (see Ref. 35). To test if the DE␤X and DE␤XA motifs can detect lower concentrations of RA than the RAR␤2-RARE we exposed cells transfected with these constructs to different concentrations of RA. DE␤X and DE␤XA respond to a RA concentration of 10 Ϫ9 , whereas the RAR␤2-RARE respond only slightly to a RA concentration of 10 Ϫ8 (Fig. 5).
Taken together, these data demonstrate that: (1) the singlenucleotide difference in the RXR-binding site of the DE and RAR␤2 RARE accounts to a large extent for the distinct activities; (2) the function of certain RAREs may be affected by their immediate flanking sequences; and (3) the activity of a response element not only depends on the sequence of the halfsites but also on their specific combination.
The Influence of the Immediate Context on the Activity of a RARE-The sensitivity of the RAR␤2-RARE starts between E7.5 and E8.5, whereas the DE starts to be active during later stages of mouse development (E10.0). RA concentration cannot be the major determinant in the timing of activation, because at E8.5 it is not the limiting factor. RA treatment of E8.5 transgenic embryos carrying a DE reporter construct does not activate the reporter at E9.5 (17). The RARE and its direct context might determine the stage-specific activity of this element. Consistent with this, the activity of the DE-RARE in P19 Reporter assays show relatively high sensitivity of hybrid elements DE␤X and DE␤XA to low concentrations of RA. Changing the DE RXR-binding site alone or together with the DE RAR-binding site into that of the RAR␤2 RARE increases the RA sensitivity in comparison to the RAR␤2 RARE in the native context. Luciferase activities are given as percentage induction, obtained by dividing the means of the tested constructs by the mean activity of the empty luciferase construct. These values are plotted (y-axis) against the different RA concentrations applied (x-axis). Note that both x-and y-axes are log scales.
EC cells was only marginally lower when the rest of the DE was deleted.
How could the stage-specific activity of the DE-RARE depend on its direct context? Because there is no evidence that RAR/ RXR heterodimers contact the flanking nucleotides of the RARE half-sites (13), this influence might be indirect. It is known that binding of RAR/RXR induces helical distortion in the DNA (14). Nucleotides outside the consensus sequence could affect the degree of helical distortions in the RARE and thereby influence its affinity and/or activity. Interestingly, it has been shown for several transactivators, including nuclear hormone receptors, that binding to DNA can induce allosteric modifications (16). Such modifications are known to influence the docking of co-factors of RA receptors thereby mediating specific transcriptional responses (10, 15, 36 -38). Therefore, the degree of helical DNA distortions in the RARE motif, affected by nucleotides outside the consensus sequence, might result in allosteric modification of the RAR/RXR heterodimer. We propose that these modifications are necessary for the binding of particular proteins that are expressed during specific stages of development, which on their turn are essential for the activity of the RA receptors bound to the DE-RARE.