An Enhancer Element in the EphA2 (Eck) Gene Sufficient for Rhombomere-specific Expression Is Activated by HOXA1 and HOXB1 Homeobox Proteins*

In the hindbrain of the mouse embryo, there is often coincident rhombomere-restricted expression of Eph receptor tyrosine kinases and Hox homeobox genes, raising the possibility of regulatory interactions. In this paper, we have identified cis-acting regulatory sequences of the EphA2(Eck) gene, which direct node and hindbrain-specific expression in transgenic embryos. An 8-kilobase region of mouse genomic DNA element was sufficient to drive rhombomere 4 (r4)-specific expression while conferring patchy expression in the node. Further analysis localized the rhombomere-specific enhancer to a 0.9-kilobase sequence. This element contains multiple Hox-Pbx consensus binding sites that bind to both HOXA1/Pbx1 and HOXB1/Pbx1 proteins in vitro. Co-expression of either HOXA1 or HOXB1 with Pbx1 transactivated EphA2 enhancer-dependent reporter gene expression. These results, together with observations of reduced EphA2 expression in hoxa1 andhoxb1 double mutant mice, suggest that expression ofEphA2 gene in rhombomere 4 is directly regulated by Hoxa1 and Hoxb1 homeobox transcription factors.

Development of the vertebrate hindbrain involves specification of the neural epithelium into lineage-restricted compartments known as rhombomeres (r, r1 to r8 in mouse) (1,2). Cells within each compartment can mix freely with each other but not with those of adjacent rhombomeres. After rhombomere formation, cells in each compartment acquire specific patterns of neuronal differentiation, axon outgrowth, and neural crest cell migration. Thus, the segmentation of the hindbrain into rhombomeres plays a critical role in generating specific neuronal cell types and regional architecture. Coupled to these morphogenetic events, signaling molecules, receptor tyrosine kinases, and transcription factors regulate multiple aspects of hindbrain development at the molecular level (reviewed in Ref. 3). Components of the regulatory networks include Hox homeobox genes and, more recently, the Eph receptor tyrosine kinases and their ligands, designated ephrins.
Hox genes encode homeobox-containing transcription factors that are homologous to Drosophila homeotic genes. During hindbrain development, many members of the Hox gene family exhibit rhombomere-restricted expression patterns. Among the first expressed Hox genes, Hoxa1 and Hoxb1 are activated during early gastrulation in the primitive streak; by the head fold stage, their expression domains reach a sharp anterior boundary in the neuroectoderm coinciding with the presumptive r3/r4 border. Whereas Hoxa1 is subsequently down-regulated, Hoxb1 expression becomes restricted to r4 and persists until the disappearance of rhombomere boundaries (4 -8). Genetic analysis of hox mutants revealed that precise anterior domains of Hox expression correlate with their functional roles in the hindbrain (reviewed in Ref. 9). r4 is partially deleted in the absence of Hoxa1 but maintains its identity (10,11), whereas the absence of Hoxb1 causes r4 to lose identity without affecting its size (12,13). Loss of both genes does not further reduce the size of r4 but creates an r4-like territory with unknown identity, suggesting synergistic roles between Hoxa1 and Hoxb1 in rhombomere patterning (14,15). Eph receptor tyrosine kinases (RTK) 1 belong to a distinct class of RTKs that play important roles in many tissues, including the hindbrain. Structurally, each receptor of this family consists of a single polypeptide chain containing two fibronectin III repeats and a cysteine-rich region in the predicted extracellular domain. To date, at least 14 members of the Eph receptor family and a family of 8 ligands have been identified. Ligands of Eph family receptors are structurally related membrane-bound proteins that can be subdivided into two major subclasses (16), ephrin-A and ephrin-B. Ligands in the ephrin-A subclass, including the prototype family member ephrin-A1 (B61), are membrane associated through glycosylphosphatidylinositol linkages, whereas ephrin-B subclass consists of ligands with transmembrane domains. During hindbrain patterning, Eph receptors and their ligands exhibit complementary rhombomere-restricted expression patterns : Eph receptors EphA4 (Sek-1) (17) and EphA2 (Eck/Sek-2) (18 -22) are up-regulated prior to rhombomere boundary formation in pre-r3/ pre-r5 and in pre-r4, respectively, whereas EphB2 (Nuk/Sek-3) (18,23) and EphB3 (Sek-4) (18) are expressed in r3 and r5 after segmentation. All the ephrin ligands examined to date, Ephrin-B1 (Elk-L), Ephrin-B2 (Elf-2), and Ephrin-B3 (Elk-L3), are expressed in r2, r4, and r6 (22) (for nomenclature see Ref. 40 the name used in the original reference is in brackets). However, mice carrying null mutations in individual Eph receptor genes seem to have normal hindbrain development (21,28), 2 probably reflecting functional redundancy among the Eph family signaling pathways. Clues to the role of Eph receptors in the developing hindbrain came from the study of dominant negative EphA4 (sek-1) receptors. Injection of RNA encoding truncated EphA4 receptor into zebrafish and Xenopus embryos results in failure to establish sharp rhombomere boundaries (24), indicating that EphA4 and its ligands restrict cell intermingling between oddand even-numbered rhombomeres. These findings are consistent with a general role of the Eph family in mediating repulsive cell-cell interaction, as suggested by studies of axonal guidance (25)(26)(27)(28) and neural crest cell migration (29 -31).
The ordered and nested expression patterns of Eph receptors and their ligands along the neuraxis raise questions how these restricted domains of expression are achieved. The segmentrestricted expression patterns of Eph receptors are reminiscent of those of Hox genes, which are expressed at similar developmental stages, raising the possibility of regulatory interactions. For example, like Hoxa1 and Hoxb1, EphA2 is expressed in the primitive streak during gastrulation and becomes restricted to prospective r4 during the head fold stage. At the 4 -8 somite stage, EphA2 expression coincides with that of Hoxb1 in r4 and is subsequently down-regulated before Hoxb1 (19 -21). In contrast to extensive studies on Hox gene regulation, relatively little is known about the regulation of Eph receptors in the hindbrain. As a first step toward understanding the coupling between cell-cell interaction and transcriptional control in the hindbrain, we undertook an analysis of the regulatory elements of the murine EphA2 gene. Here, we report the identification of an enhancer element that drives EphA2 expression in r4 and show that human HOXA1 and HOXB1 can activate reporter gene expression through this enhancer element. These data suggest a direct mechanism for control of EphA2 expression by the HOX paralogy group 1.

EXPERIMENTAL PROCEDURES
Transgene Constructs and DNA Sequencing-A 12-kb genomic DNA fragment (construct 1) upstream of the EphA2 gene was originally isolated from sequences surrounding an inserted U3␤geo (lacZ-neo) gene trap retrovirus (21). For constructs 2-7, genomic DNA fragments isolated from EphA2 genomic clones were ligated into the SmaI site of ASShsplacZpA cassette (gift of Sasaki and Hogan (32)). Both strands of the 1.8-kb SmaI fragment (construct 5) were sequenced following cycle sequencing with fluorescent dideoxy chain terminators and analyzed on an ABI 377 DNA sequencer.
Generation of Transgenic Embryos-NotI fragments containing the transgene expression cassette were isolated from agarose gels by the Gelase protocol (33). DNA concentration was measured by either agarose gel electrophoresis or fluorometry. Transgenic embryos were produced by pronuclear injection as described (33). Briefly, 6-week-old (C57BL/6 ϫ DBA/2) F1 female mice were superovulated by intraperitoneal injection of 5 IU of gonadotropin from pregnant mare serum 48 h prior to injection of 5 IU of human chorionic gonadotropin. Female mice were then mated with (C57BL/6 ϫ DBA/2) F1 male mice. The pronuclei of fertilized eggs were microinjected with transgene DNA at a concentration of 3-6 ng/l. Injected eggs were transferred to pseudopregnant recipient ICR mice. The genotype of embryos was determined by polymerase chain reaction amplification of a lacZ sequence. Primers used were 5Ј-TTGCCGTCTGA-ATTTGACCTG and 5Ј-TCTGCTTCAATCTGCGTGCC.
Protein Production and DNA Binding Assays-Pbx and HOX proteins were produced in vitro from the corresponding pCDNA3-derived expression vectors using a T7 polymerase-based coupled transcriptiontranslation reticulocyte lysate system (Promega) according to manufacturer's instructions. HOX and Pbx proteins were translated separately in the presence of [ 35 S]methionine and normalized for the methionine content of each protein.
Cell Culture and Transfection-COS7 and P19 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% calf serum or fetal bovine serum, respectively. COS7 cells were transfected by the standard DEAE-dextran method. In a typical transfection experiment, 8 g of reporter construct, 4 -8 g of expression construct, and 0.4 g of pCH110-␤-gal as an internal control were used per 10-cm dish. P19 cells were transfected by LipofectAmine (Life Technologies, Inc.) reagent in a 6-well dish. Cells were harvested 48 h after transfection and assayed for luciferase and ␤-galactosidase activity.

A 12-kb DNA Element from the epha2 i Locus Regulates EphA2
Expression during Early Embryogenesis-The expression of EphA2 during hindbrain patterning is very dynamic (19 -21). Initially, EphA2 is detected at the head fold stage in a narrow band corresponding to presumptive r4. As somitogenesis begins, EphA2 expression becomes restricted to the presumptive r4, which includes both dorsal neural ectoderm and the floor plate. By the 10-somite stage, EphA2 expression is down-regulated. In addition to its hindbrain expression, transcripts of EphA2 become restricted in the node during gastrulation.
To define upstream signaling pathways and transcriptional events that control this dynamic expression pattern of EphA2, we mapped regulatory elements of the EphA2 gene in transgenic embryos. Our prior studies (21) have shown that integration of a gene trap provirus in the 5Ј end of the EphA2 gene disrupted the expression of the EphA2 protein. Moreover, the expression of the reporter gene carried by the provirus recapitulated the endogenous EphA2 expression between 6.5 and 10.5 days of embryonic development (21). Therefore, we reasoned that tissue-specific regulatory elements responsible for EphA2 gene expression might be in the regions flanking the provirus. To test this possibility, a 12-kb genomic DNA insert containing a single 4-kb provirus LTR and 8-kb flanking cellular DNA (Fig. 1A, construct 1) was isolated from a genomic DNA library constructed from epha2 i (eck i ) heterozygous mice (21). This 12-kb genomic DNA fragment carries a promoterless LacZ-neo fusion gene in the U3 region of the pro-viral LTR (viral enhancer deleted (34)) and was microinjected to generate transgenic embryos. As shown in Fig. 1B (construct 1), the spatial and temporal patterns of lacZ expression in transgenic embryos largely coincided with that of the endogenous EphA2 gene (19 -21). LacZ expression was detected in the node during gastrulation in E7.5 embryos (Fig. 1B, left panel) and in the presumptive r4 in E8.5 embryos (Fig. 1B, right panel). In E9.5 embryos, lacZ expression was down-regulated in r4 and regressed toward tail bud (data not shown). In the node, lacZ expression was only seen in some cells (data not shown), indicating that additional elements are required to recapitulate complete node expression of EphA2. Taken together, these results suggest that the 8-kb EphA2 genomic DNA flanking the provirus is sufficient to direct r4-specific expression of EphA2 gene during early embryogenesis.
Identification of a 0.9-kb Enhancer Element Specifying the r4 Expression of the EphA2 Gene-To further define cis-acting EphA2 regulatory elements, an 8-kb BglII fragment was isolated from a wild-type EphA2 genomic clone. Judging by re-striction mapping, this 8-kb fragment is identical to the 12-kb fragment except that it does not contain the 4-kb viral LTR. Restriction fragments were isolated from the 8-kb BglII genomic DNA, placed upstream of an hsp68promoter-lacZpoly(A) cassette (32), and tested for enhancer activity in transgenic embryos at E7.5 to E8.5, stages where EphA2 is expressed in presumptive r4. With such an enhancer assay, patterns and levels of gene expression may vary between transgenic embryos, depending on the integration site of the transgene. Therefore, only reproducible expression patterns reflecting enhancer activity of the test fragment were scored.
Expression from the proximal 3.8-kb XhoI-BglII fragment (Fig. 2, construct 4) was similar to the original 12-kb fragment, i.e. expression in the rhombomere and spotty expression in the node. In contrast, the distal 5.2-kb XhoI-XhoI fragment (Fig. 2,  construct 2) and the 4.5-kb BglII-XhoI fragment (Fig. 2, construct 3) gave only variable ectopic staining. Because the hsp68promoter-lacZ-poly(A) cassette is devoid of any enhancer sequences and, on its own, does not give a reproducible expression pattern in transgenic embryos (data not shown, see Ref. 32), these data indicate that the proximal 3.8-kb XhoI-BglII fragment (construct 4) contains sufficient enhancer activity to confer rhombomere-specific expression to a heterologous promoter and that the distal 5.2-kb XhoI-XhoI fragment (construct 2) and the 4.5-kb BglII-XhoI fragment (construct 3) do not contain any essential regulatory elements for expression in these domains. The 3.8-kb XhoI-BglII fragment was further divided into several fragments. Of these, a 1.8-kb SmaI fragment (Fig. 2, construct 5) recapitulated the expression pattern of the original 12-kb fragment, whereas the 1.5-kb SmaI-BglII fragment (Fig. 2, construct 6) gave only ectopic staining in embryos. Further division of the 1.8-kb fragment revealed that the 0.9-kb SacI-SmaI fragment (Fig. 2, construct 7) retained rhombomere-specific staining but did not give node staining. Taken together, these results demonstrated that the 0.9-kb SacI-SmaI fragment contains the rhombomere-specific enhancer and that enhancers for the rhombomere and the node reside in separate DNA elements.
The 0.9-kb Enhancer Element Contains Multiple Consensus Hox-Pbx Binding Sites-Localization of the EphA2 r4 enhancer element to a sequence of 0.9 kb enabled us to search for potential transcription factors that might mediate enhancer activity. As a first step, to examine whether any of the known transcription factor binding sites are present in the EphA2 enhancer regions, the 1.8-kb SmaI fragment (construct 5) was sequenced. Sequence analysis revealed that the 1.8-kb SmaI fragment contained many potential transcription factor binding sites, including the Hox-Pbx consensus binding repeat identified in the Hoxb1 r4 autoregulatory element (ARE) (35). However, most binding sites are present in both r4-specific enhancer element (3Ј half of the 1.8-kb fragment, construct 7) and nonspecific genomic DNA (5Ј half of the 1.8-kb fragment). In contrast, five Hox-Pbx binding sites were found in the 0.9-kb r4-specific enhancer element (Fig. 3), whereas no such sites were present in the 5Ј half of the 1.8-kb fragment. Because the up-regulation of EphA2 in r4 is very similar to the early expression patterns of Hoxa1 and Hoxb1 in the hindbrain, we focused our studies on the Hox-Pbx binding sites. These Hox-Pbx binding sites, 5Ј-(T/A)GAT(T/G)GA(T/A)G-3Ј, were shown to bind specifically to Hoxb1/Pbx1 and Hoxa1/Pbx1 complexes (35,36). In addition, the Hoxb1 r4 ARE contains three such sites (repeat 1 to repeat 3), which are necessary and sufficient to drive r4-specific expression in transgenic embryos (35). These putative Hox-Pbx binding sites in the EphA2 enhancer have been named repeat A to repeat E. Four of these repeats are clustered in the 3Ј end of the enhancer, whereas repeat A is located more centrally. All of the repeats share identical core nucleotides (bold in Fig. 3), with repeat D containing two overlapping repeats. These results, together with the similarity of expression patterns for EphA2, Hoxa1, and Hoxb1, raise the question whether transcriptional activation of EphA2 is directly regulated by Hoxb1 and/or its paralog Hoxa1.
HOXA1 and HOXB1 Bind to Hox-Pbx Binding Sites in EphA2 Enhancer in Conjunction with Pbx1 Protein-To investigate the possible involvement of Hoxa1 and Hoxb1 in mediating EphA2 enhancer activity, we tested the ability of HOXA1 and HOXB1 proteins to bind to a series of oligonucleotides, each of them spanning one repeat motif. Prior studies showed that specific interactions between Hox proteins and their target sequences require Pbx cofactors (35)(36)(37). Therefore, in vitro translated Pbx1, HOXA1, and HOXB1 proteins were used in EMSAs. As repeats B, C, D, and E are clustered in the 3Ј portion of the enhancer, we tested whether these four repeats can bind to HOXA1/Pbx1 or HOXB1/Pbx1 in vitro. The ability of HOX/Pbx protein to bind to repeats B-E (lanes 3-10) was also compared with binding to repeat 3 of the Hoxb-1 r4 ARE (lanes 1 and 2) (35). As shown in Fig. 4A (lanes 3-10), HOXA1/ Pbx1 and HOXB1/Pbx1 complexes can bind to all four repeats, although apparently with different affinities. Repeat E showed the strongest binding to HOX/Pbx1 complexes (lanes 9 and 10), which is comparable with the binding of repeat 3 in the Hoxb1 r4 ARE (lanes 1 and 2). A close examination of these repeat sequences revealed that although all four repeats are identical in their core sequence 5Ј-GATGGA-3Ј, repeat E in EphA2 enhancer and repeat 3 in Hoxb1 ARE carried a nucleotide T flanking the 5Ј-G, which might contribute to the tighter binding of these repeats to HOX/Pbx complexes. EMSAs were also used to test whether cooperative interactions between HOXA1 or HOXB1 and Pbx1 are essential for binding to repeat E from the EphA2 enhancer region. As shown in Fig. 4B, HOXA1 and HOXB1 alone did not bind to repeat E   FIG. 3. EphA2 r4 enhancer sequence. A, Hox-Pbx consensus binding site from the Hoxb-1 r4 ARE. Invariant core sequence within the repeats is shown in bold. B, DNA sequence of the 0.9-kb SacI-SmaI fragment containing r4 enhancer activity. The location of the five Hox-Pbx binding sites are underlined and named as repeats A to E. The Genbank accession no. is AF069295. This sequence was scanned against the data base and did not match with any sequences with significant relatedness. (lanes 2 and 3), whereas Pbx-1 alone bound only weakly (lane 4). However, under the same conditions, efficient binding was observed when Pbx1 was present together with either HOX protein (lanes 5 and 8). This binding was specific, as the complexes were competed by an excess of unlabeled normal (lanes 6 and 9) but not mutant (lanes 7 and 10) oligonucleotides containing repeat E. Similar results were obtained with repeat B and with Hoxb1 r4 ARE repeat 3 (data not shown; specific and cooperative binding to repeats C and D was not tested). These results showed that potential Hox-Pbx binding sites within the EphA2 enhancer can specifically bind to HOXA1/ Pbx1 or HOXB1/Pbx1 proteins in vitro.
HOXA1/Pbx1 and HOXB1/Pbx1 Activate Transcription through the EphA2 r42B Enhancer Element-In light of our binding results, the ability of the HOX/Pbx complexes to transactivate reporter gene expression through the EphA2 enhancer was tested in transient cotransfection experiments. Because the hsp68promoter-lacZ-poly(A) cassette gives rise to high background activity in cell lines, we subcloned a 200-base pair fragment (EphA2-r42B, Fig. 5A) containing repeats B to E from the EphA2 enhancer into pML, a luciferase reporter vector that exhibits minimal background activity in both COS7 and P19 cells (36). pML-ARE, a luciferase reporter under the control of the Hoxb-1 r4 ARE, was used as a positive control. As shown in Fig. 5B, cotransfection of HOXA1 and Pbx1, or HOXB1 and Pbx1, with the pML-EphA2-r42B reporter led to significant transactivation of the reporter activity in COS7 cells, whereas cotransfection of pcDNA3 control expression vector with pML-EphA2-r42B reporter gave only background activity. This level of transactivation of pML-EphA2-r42B by HOX/Pbx proteins is comparable with that of the pML-ARE positive control reporter. Transactivation of pML-EphA2-r42B reporter activity  1 and 2) and repeats B, C, D, and E of the EphA2 r4 enhancer (lanes 3-10). The Pbx1 and HOX proteins were produced and labeled by coupled transcription-translation in rabbit reticulocyte lysates. Lysates were mixed with the radiolabeled oligonucleotides in a binding reaction and subjected to EMSA. Lys, nonspecific complexes arising from endogenous reticulocyte lysate binding activities (see panel B, lane 1); R, repeat. Panel B, cooperative and specific binding of Pbx1 and HOX proteins to repeat E. HOXA1, HOXB1, or Pbx1 alone did not result in significant binding (lanes 2-4), but a specific complex formed (arrow, lanes 5 and 8) when either HOXA1 and Pbx1 or HOXB1 and Pbx1 were included. The specificity of the HOX-dependent complexes was demonstrated by competition with an unlabeled excess of repeat E (lanes 6 and 9) or mutated version of repeat E (lanes 7 and 10). Nonspecific complexes (Lys) arising from endogenous reticulocyte lysate binding activities (lane 1) were also observed, and an excess of either wild-type (lanes 6 and 9) or mutant (lanes 7 and 10) oligonucleotides competed the formation of some of these complexes. R, competitor oligonucleotides containing repeat E; M, competitor oligonucleotide containing mutant version of repeat E (for sequences, see "Experimental Procedures"). 5. HOXA1 and HOXB1 activate transcription from the EphA2 r42B enhancer in combination with Pbx1. A, schematic representation of EphA2 r42B enhancer, which contains four Hox-Pbx binding sites (repeat B to E). B, luciferase activity, in arbitrary units, assayed from extracts of transiently transfected COS7 cells. Cells were transfected with 4 g of HOXA1 or HOXB1, and 8 g of Pbx-1 expression constructs, together with 8 g of pML (luciferase reporter construct), pMLARE (pML containing Hoxb-1 r4 ARE), and pML-EphA2-r42B reporter constructs. 0.4 g of the pCH110␤-gal plasmid were cotransfected as an internal standard. C, luciferase activity assayed from extracts of transiently transfected P19 cells. Cells were transfected and analyzed as in B. Empty bars, pcDNA3; filled bars, HOXA1/ Pbx1; shaded bars, HOXB1/Pbx1. by HOX/Pbx proteins was also observed in a different cell type, i.e. the murine embryonal carcinoma P19 cell line (Fig. 5C). These data suggest that HOXA1 or HOXB1 in conjunction with Pbx1 protein can work in trans to activate enhancer-dependent expression of the EphA2 gene. DISCUSSION Despite extensive investigation of the transcriptional regulation of Hox genes during hindbrain development, relatively few studies (38,39) have addressed rhombomere-specific expression of Eph receptors. Here, we report that an 8.0-kb genomic fragment 5Ј of the EphA2 coding region directs reporter gene expression to r4 of transgenic mouse embryos. Further analysis showed that a 0.9-kb fragment within the region can recapitulate this r4-specific expression pattern. This activity is independent of the orientation of the 0.9-kb fragment, suggesting that it contains a transcriptional enhancer.

FIG.
r4-specific expression of the EphA2 gene is reminiscent of Hoxb1 and its paralog Hoxa1, raising the possibility that EphA2 expression is regulated by Hox genes. Here, we show that 1) the 0.9-kb enhancer of EphA2 contains multiple binding sites for HOXA1/Pbx1 and HOXB1/Pbx1 proteins, 2) oligonucleotides spanning the binding sites bind to these proteins in vitro, and 3) co-expression of HOXA1 and Pbx1 or HOXB1 and Pbx1 proteins transactivates reporter gene expression from the EphA2-r42B enhancer. These results are consistent with recent genetic studies in which EphA2 expression is substantially decreased in hoxa1/hoxb1 double null mutant mice, although an r4-like territory with unknown identity is present in the mutant (14,15). 3 Taken together, these data suggest that expression of EphA2 in r4 is, at least in part, directly regulated by Hoxa1 and Hoxb1.
r4 enhancer activity was first detected in sequences surrounding a U3␤geo gene trap retrovirus inserted into the EphA2 gene. Although the provirus inserted over 8 kb upstream of the 5Ј end of the previously published EphA2 cDNA sequence, the provirus completely abolished EphA2 protein expression, and expression of the inserted lacZ reporter recapitulated expression of the endogenous EphA2 gene (21). These studies illustrate the utility of gene entrapment as a mean to identify functional elements involved in tissue-specific gene regulation.
Although our evidence suggests a direct role of Hoxa1/Hoxb1 in regulating EphA2 expression, Hoxa1/Hoxb1 alone may not be sufficient to account for the complete dynamic temporal and spatial restricted expression pattern of EphA2 gene in r4 (19 -21). For example, EphA2 expression is substantially reduced, but not abolished, in the hoxa1/hoxb1 double null mutant, suggesting the existence of additional factors regulating residual EphA2 expression. Furthermore, EphA2 expression is down-regulated in r4 by embryonic day E8.75, whereas Hoxb-1 expression persists until at least E10.5 (8). Although these findings suggest the existence of specific transcriptional factors, further studies will be required to identify such additional factors regulating EphA2 expression in r4.
Members of the Eph receptor family are known to mediate repulsive cell-cell interactions in axonal guidance during neural development (27,28). Because of their segmental-restricted expression patterns in the hindbrain, this ligand/receptor family is likely to restrict cell intermingling between odd-and even-numbered rhombomeres. Experimental evidence to support this hypothesis came from a study involving truncated EphA4 receptors, in which injection of RNA encoding a dominant negative EphA4 into zebrafish and Xenopus embryos interfered with the establishment of sharp rhombomere boundaries (24). The role of EphA2 in hindbrain is less clear. Mutant mice carrying mutations generated either by retroviral insertion (21) or by gene targeting 2 seem to have normal hindbrain development, probably reflecting functional compensation by other receptors in the Eph family. The identification and characterization of this EphA2 r4 enhancer element permit targeting the expression of trans-dominant inhibitors of Eph receptor, e.g. dominant negative receptors or receptor-Ig, to be expressed at the endogenous locus in transgenic mice, allowing functional characterization of EphA2 during hindbrain development.