HOXA13 directly regulates EphA6 and EphA7 expression in the genital tubercle vascular endothelia



Hypospadias, a common defect affecting the growth and closure of the external genitalia, is often accompanied by gross enlargements of the genital tubercle (GT) vasculature. Because Hoxa13 homozygous mutant mice also exhibit hypospadias and GT vessel expansion, we examined whether genes playing a role in angiogenesis exhibit reduced expression in the GT. From this analysis, reductions in EphA6 and EphA7 were detected. Characterization of EphA6 and EphA7 expression in the GT confirmed colocalization with HOXA13 in the GT vascular endothelia. Analysis of the EphA6 and EphA7 promoter regions revealed a series of highly conserved cis-regulatory elements bound by HOXA13 with high affinity. GT chromatin immunoprecipitation confirmed that HOXA13 binds these gene-regulatory elements in vivo. In vitro, HOXA13 activates gene expression through the EphA6 and EphA7 gene-regulatory elements. Together these findings indicate that HOXA13 directly regulates EphA6 and EphA7 in the developing GT and identifies the GT vascular endothelia as a novel site for HOXA13-dependent expression of EphA6 and EphA7. Developmental Dynamics 236:951–960, 2007. © 2007 Wiley-Liss, Inc.


Hypospadias, a defect affecting the growth and closure of the external genitalia, is highly prevalent in the birth populations of industrialized nations, including the United States, United Kingdom, Sweden, and Japan (Giwercman et al.,1993; Paulozzi et al.,1997; Gallentine et al.,2001). While the frequency of hypospadias ranges as high as 1 in 125 live births, the molecular mechanisms underlying this defect are poorly understood (Svensson et al.,1997; Paulozzi et al.,1997; Stadler,2003). One phenotype commonly associated with hypospadias is the enlargement of blood vessels supplying the glans or prepuce (Baskin et al.,1998; Baskin,2000). At present, no molecular link between hypospadias and vessel enlargement in the genitalia has been identified; however, studies examining perturbations in Eph–ephrin signaling may provide important clues toward understanding the pathology of hypospadias and its associated vascular malformations (reviewed by Eichmann et al.,2005a,b; Hinck,2004; Klagsbrun and Eichmann,2005; Davy and Soriano,2005). Indeed, Eph–ephrin signaling is essential for the patterning of multiple tissues and cell types, including vascular endothelial cell assembly, cell migration, mesenchymal cell condensation, vascular bed formation, tumor neovascularization, and the closure of the external genitalia (Wang et al.,1998; Ogawa et al.,2000; Stadler et al.,2001; Chan et al.,2001; Dravis et al.,2004; Davy et al.,2004; Marquardt et al.,2005; Egea et al.,2005).

Recently, we and others have shown that HOXA13 function is necessary for EphA7 expression in the developing limb (Stadler et al.,2001; Salsi and Zappavigna,2006). Recognizing that Hoxa13-deficient mice also exhibit hypospadias and capillary vessel enlargement (Morgan et al.,2003), we hypothesized that HOXA13 may regulate Eph receptor expression in the genital tubercle (GT) and its vasculature. Testing this hypothesis, we report that HOXA13 directly regulates EphA6 and EphA7 expression in the GT vascular endothelia. Analysis of the EphA6 and EphA7 promoter regions revealed a conserved series of DNA sequences bound with high affinity by the HOXA13 DNA binding domain (A13). In vivo, direct interactions between HOXA13 and the EphA6 and EphA7 promoter elements were detected in the GT using Hoxa13-directed ChIP. In vitro, HOXA13 can use the bound gene-regulatory elements in the EphA6 and EphA7 promoters to direct gene expression. Together these findings indicate that EphA6 and EphA7 are direct transcriptional targets of HOXA13 in the GT vascular endothelia, providing new insight into the cell-signaling mechanisms functioning during the growth and development of the external genitalia.


Vessel Expansion Is Present Throughout GT Development in Hoxa13–Green Fluorescent Protein Homozygous Mutants

Analysis of the distal GT from embryonic days (E) 12.5 to 15.5 revealed average vessel diameters of 15 μm (± 0.5 μm; n = 12 independent samples) in Hoxa13-GFP (GFP, green fluorescent protein) wild-type and heterozygous mutants. In contrast, the GT vasculature of age-matched homozygous mutants exhibited greatly enlarged vessels with average diameters of 85 μm (± 15 μm; n = 12 independent samples; Fig. 1). Of interest, embryonic sex did not influence the presentation of the vascular phenotype as homozygous mutant male and female embryos exhibited similar increases in GT vessel diameter (Fig. 1, compare E–H with I–L).

Figure 1.

Hoxa13-deficient mice display enlarged blood vessels in the developing genital tubercle. A–L: Hematoxylin and eosin staining of sectioned genital tubercles from wild-type (A–D), homozygous mutant male (E–H), and homozygous mutant female (I–L) embryos at embryonic days (E) 12.5–15.5. Arrows denote a typical example of normal vessel diameters in wild-type embryos (A–D) or enlarged vessels in homozygous mutants (E–L). UPE, urethral plate epithelium. Scale bar = 100 μm.

EphA6 and EphA7 Are Reduced in the GT Vascular Endothelia of Hoxa13 Homozygous Mutants

Recognizing that defects in vascular patterning are strongly associated with perturbations in Eph–ephrin signaling, we examined whether the affected GT vasculature exhibited changes in the expression of Eph receptors or their ephrin ligands. While no changes in EPHA2, EPHA4, EPHA5, EPHB2, EPHRIN A2, or EPHRIN A5 expression were detected in the Hoxa13-GFP homozygous mutants (data not shown), both EPHA6 and EPHA7 were consistently reduced in the GT vascular endothelia compared with heterozygous mutant controls (Fig. 2A–D), which do not exhibit a GT phenotype (Morgan et al.,2003). In the vasculature, HOXA13-GFP, EPHA6, and EPHA7 were strongly colocalized in the endothelial layer of heterozygous controls, whereas the reduced levels of EPHA6 and EPHA7 in the homozygous mutants minimized our detection of colocalization in the expanded GT vasculature (Fig. 2E–L). Platelet endothelial cell adhesion molecule-1 (PECAM-1) expression was also present in the affected GT vasculature, suggesting that endothelial cell identity was not affected by the loss of HOXA13 function (Fig. 2M–N).

Figure 2.

Immunohistochemical localization of EPHA6 and EPHA7 in the genital tubercle (GT) of E13.5 Hoxa13-green fluorescent protein (GFP) mice. A,C: EPHA6 and EPHA7 (red signal) are expressed in the GT mesenchyme (MES) as well as the endothelial cells lining the GT vasculature (arrows) of heterozygous Hoxa13-GFP embryos. B,D: EPHA6 and EPHA7 expression is reduced in the enlarged GT vessels of age-matched Hoxa13 homozygous mutants (arrows). E,G: Colocalization (yellow signal) of Hoxa13-GFP with EPHA6 and EPHA7 in the GT vascular endothelia of heterozygous embryos. Arrows denote colocalization in the GT endothelia. F,H: Reduced expression of EPHA6 and EPHA7 in the homozygous mutant vascular endothelia. Arrows denote the enlarged vasculature. I–L: Higher magnification images of the heterozygous control and homozygous mutant vessels depicted in E–H. M,N: Platelet endothelial cell adhesion molecule-1 (PECAM-1) expression (red) is maintained in the GT vascular endothelia of Hoxa13-GFP (green) heterozygous mutant controls and age-matched homozygous mutants. UPE, urethral plate epithelium; MES, mesenchymal tissue; R, red blood cells. For clarity, only male GTs are shown, although similar vascular defects were observed in mutant female GTs. Scale bars = 140 μm in A–H, 20 μm in I–L, 70 μm in M–P.

Semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) analysis confirmed the levels of EphA6 and EphA7 expression detected by immunohistochemistry. In particular, a uniform level of EphA6 and EphA7 expression was detected throughout the GT mesenchyme in Hoxa13-GFP heterozygous and homozygous mutant embryos (Fig. 3A). Next, because cell-specific changes in gene expression are often not detectable in whole tissue RNA samples, we examined whether endothelial cells purified GT vasculature exhibited reductions in EphA6 and EphA7 expression, as observed by immunohistochemistry. RT-PCR analysis of GT endothelial cell isolates confirmed that EphA6 and EphA7 are expressed in the endothelial component of the GT vessels along with Hoxa13 and PECAM-1 in Hoxa13-GFP heterozygous mutants (Fig. 3B). In contrast, the expression of EphA6 and EphA7 was consistently reduced (n = 3 independent assessments) in the purified GT vascular endothelia of Hoxa13-GFP homozygous mutants, although no difference in PECAM-1 expression was observed (Fig. 3B).

Figure 3.

Quantitation of Hoxa13, EphA6, and EphA7 expression in the genital tubercle (GT) mesenchyme and vascular endothelia. A: Semiquantitative RT-PCR of EphA6 and EphA7 transcripts revealed no differences in EphA6 and EphA7 expression in the whole distal GT. B:Hoxa13, EphA6, and EphA7 are coexpressed in the GT vascular endothelium as determined by semiquantitative reverse transcriptase-polymerase chain reaction on platelet endothelial cell adhesion molecule (PECAM) -positive endothelial cell isolates. Furthermore, both EphA6 and EphA7 expression levels were reduced in vascular endothelial cells of Hoxa13 homozygous mutant GT compared with Hoxa13 heterozygous controls. The glyceraldehydes-3-phosphate dehydrogenase (GAPDH) housekeeping gene was used as a cDNA control.

HOXA13 Binds Discrete Regions of the EphA6 and EphA7 Promoters

DNA sequence analysis of the EphA6 and EphA7 promoter regions revealed several A-T–rich regions matching sequences we previously identified as sites bound by HOXA13 (Knosp et al.,2004; Fig. 4A). Because key gene-regulatory regions are often conserved between species, we examined whether the EphA6 and EphA7 regions containing the clustered HOXA13 binding sites are conserved using the UCSC Genome Browser (Kent et al.,2002). For the EphA6 element, a high degree of conservation was observed between mouse (Mus musculus), human (Homo sapiens), rat (Rattus norvegicus), and opossum (Monodelphis domestica) with complete identity in several A-T–rich regions where HOXA13 is thought to bind (Knosp et al.,2004; McCabe and Innis,2005; Salsi and Zappavigna,2006; Fig. 4B). For the EphA7 fragment, strong sequence conservation was also detected between mouse, human, rat, opossum, and several additional mammalian species, including dog (Canis familiarus), armadillo (Asypus novemcinctus), and elephant (Loxodonta africana), with strong identity within the putative HOXA13 binding site (Fig. 4B).

Figure 4.

HOXA13 binds discrete regions of the EphA6 and EphA7 promoters. A: Sequences of the murine EphA6 and EphA7 promoter regions. Underlined nucleotides denote regions exhibiting high conservation among multiple species. Lower case nucleotides denote primer sites used to amplify the promoter region for electrophoretic mobility shift assay (EMSA) analysis and chromatin immunoprecipitation in Figure 5. B: Species conservation of the EphA6 and EphA7 promoter regions bound by the HOXA13 DNA binding domain peptide (A13). Sequence conservation and alignment were determined using BLAST analysis software and the UCSC Genome Browser Database (Kent et al.,2002,2005; Karolchik et al.,2003). C: EMSA analysis of the EphA6 and EphA7 promoter regions. At 0.2 μM, the HOXA13 DNA binding domain peptide (A13) exhibits nonsaturating levels of binding to several Hoxa13 binding sites present in the EphA6 and EphA7 promoter fragments, whereas at 2.0 μM of A13, all binding sites present in the EphA6 and EphA7 promoter fragments appear bound, producing a single reduced mobility DNA fragment. Note that the addition of 0.2 μM unlabeled EphA6 or EphA7 promoter fragments was sufficient to competitively displace the A13 peptide from the radiolabeled promoter fragments.

Figure 5.

HOXA13 associates with the EphA6 and EphA7 promoter fragments in vivo in the developing genital tubercle (GT). A,B: Positive polymerase chain reaction (PCR) amplification of the EphA6 and EphA7 conserved regions after HOXA13 chromatin immunoprecipitation (ChIP) confirms that wild-type HOXA13 binds the EphA6 (A) and EphA7 (B) promoter regions in the embryonic day (E) 12.5 GT. Note that parallel ChIP assays of GT chromatin from homozygous mutants did not detect these same DNA, suggesting that HOXA13′s DNA binding function is necessary for the in vivo binding of HOXA13 to the conserved EphA6 and EphA7 regions. NC, negative PCR control; IgG, immunoglobulin antibody control; Input, positive control confirming the presence of the EphA6 and EphA7 conserved regions in the chromatin samples before immunoprecipitation.

Electrophoretic mobility shift assay (EMSA) analysis of the EphA6 region (−2410 to −2067) revealed consistent binding by the HOXA13 DNA binding peptide, which could be competitively removed using identical nonlabeled DNA fragments (Fig. 4C). Similar binding was also observed for the 325-bp EphA7 promoter region (−839 to −514) as well as an additional HOXA13-regulated region previously characterized by Salsi and Zappavigna (2006; data not shown; Fig. 4C). Interestingly, both the EphA6 and EphA7 regions exhibited a staggered series of mobility-shifted bands when incubated with 0.2 μM A13 peptide, suggesting that multiple binding sites are present in the EphA6 and EphA7 conserved regions, which require a higher protein concentration to be completely saturated (Fig. 4C). Increasing the HOXA13 DNA binding peptide concentration to 2 μM produced a single higher molecular weight product for the EphA6 and EphA7 promoter elements, indicating saturation of all Hoxa13 binding sites (Fig. 4C).

Next, to establish whether the HOXA13 DNA binding domain peptide binds DNA as monomer or as higher order complex, we performed a sedimentation equilibrium analysis. Analysis of the sedimentation the HOXA13 DNA binding peptide bound to DNA revealed a single complex of approximately 16,900 daltons, which correlates to a 1:1 stoichiometric ratio of peptide (8,137 daltons) to DNA 8,100 (daltons), suggesting that the staggered bands detected by EMSA most likely reflect the concentration-dependent saturation of multiple binding sites by the monomeric protein, rather than higher order complexes.

HOXA13 Binds the EphA6 and EphA7cis-Regulatory Elements In Vivo

Next, to determine whether HOXA13 directly interacts with the EphA6 and EphA7 promoter sequences in vivo, we examined whether GT-specific chromatin containing the conserved EphA6 and EphA7 regions was immunoprecipitated with a HOXA13 antibody (αA13). Previous characterization of the HOXA13 antibody confirmed that it can bind both HOXA13 wild-type and mutant proteins and facilitate the immunoprecipitation of gene-regulatory elements directly bound by wild-type HOXA13 (Knosp et al.,2004). PCR amplification of wild-type αA13-chromatin immunoprecipitates consistently detected the EphA6 and EphA7 promoter fragments (n = 5 independent assessments; Fig. 5). In homozygous mutants, the EphA6 and EphA7 promoter fragments could not be detected in the αA13-immunoprecipitated chromatin (n = 5 independent assessments; Fig. 5), suggesting that HOXA13′s DNA binding function, which is absent in the mutant HOXA13-GFP protein is necessary for the immunoprecipitation of these gene-regulatory sequences.

In Vitro Utilization of the EphA6 and EphA7 Promoter Elements by Full-Length HOXA13

The capacity of full-length HOXA13 to regulate gene expression through the EphA6 and EphA7 promoter fragments was assessed in NG108-15 cells. In the absence of the EphA6 or EphA7 promoter fragments, the empty pGL4.1 luciferase plasmid produced insignificant amounts of luciferase when cotransfected with a Hoxa13 expression plasmid (pCMVHoxa13; Fig. 6). Cotransfection of pCMVHoxa13 with pGL4.1 plasmid containing the forward orientation of the EphA6 promoter fragment produced nearly a fourfold increase in luciferase expression (Fig. 6). Of interest, cotransfection of pCMVHoxa13 with the pGL4.1 plasmid containing the reverse orientation of EphA6 promoter fragment produced no increase in luciferase expression, suggesting the EphA6 gene-regulatory region functions in an orientation-specific manner. In contrast, activation of the EphA7 promoter fragment by HOXA13 facilitated luciferase expression at levels three- to fourfold higher than controls independent of its orientation, suggesting the sequence bound by HOXA13 may function as an enhancer (Fig. 6).

Figure 6.

HOXA13 activates gene expression through the conserved EphA6 and EphA7 region in vitro. Cotransfection of NG108-15 cells with a pGL4.1 luciferase reporter plasmids containing forward or reverse orientations of the EphA6 region bound by HOXA13 with a HOXA13 expression vector (pCMV-A13) resulted in increased luciferase expression (3.75-fold) only in the forward orientation compared with transfections with a control pCMV vector. Identical transfections using the conserved EphA7 region bound by HOXA13 resulted in a 2.75- to 4-fold increase in relative luciferase expression, suggesting that this gene-regulatory region may be functioning as a HOXA13-directed enhancer of EphA7 expression. Luciferase activity was normalized for transfection efficiency using a Renilla Luciferase control plasmid in all cotransfection assays. Bars represent the standard deviation of results derived from three independent assays.

Loss of EphA7 Is Not Sufficient to Cause Gross Enlargement of the GT Vasculature

The individual function of EphA7 was assessed in the developing GT vasculature using a null EphA7 allele (Holmberg et al.,2000). A comparison of the GT vessel diameters between wild-type and EphA7 homozygous mutant embryos revealed no significant expansions of the GT vessels (Fig. 7, page 3), suggesting that the expression of EphA6 in the GT vasculature may be sufficient to compensate for the loss of EphA7 function, or that EphA7 does not play a role in regulating vessel wall diameter. The effect of a combinatorial loss of EphA6 and EphA7 in the GT vasculature could not be assessed as EphA6 null mutations are early embryonic lethal (Brown et al.,2000).

Figure 7.

Genital tubercle (GT) vessel diameter and endothelial cell identity are maintained in EphA7 homozygous mutants. A: Typical GT vessel diameters and endothelial cell adhesion molecule-1 (PECAM-1) expression in embryonic day (E) 13.5 wild-type control embryos. B,C: The GT vasculature of EphA7 homozygous mutants (B) does not exhibit vessel enlargement or alteration in PECAM-1 expression when compared with the enlarged GT vasculature of Hoxa13-GFP homozygous mutants (C). Scale bar = 50 μm.


In humans, the loss of HOXA13 function causes hand-foot-genital syndrome (HFGS), an autosomal dominant disorder that profoundly affects the development of the external genitalia (hypospadias), uterus, vagina, cervix, bladder, and ureter (Stern et al.,1970; Mortlock and Innis,1997; Warot et al.,1997; Morgan et al.,2003; Stadler,2003). An important step toward understanding the developmental basis for the genitourinary defects associated with HFGS is the identification of the genes directly regulated by HOXA13. In this report, we identify EphA6 and EphA7 as direct transcriptional targets of HOXA13 in the GT. The in vivo association of HOXA13 with gene-regulatory elements present in EphA6 and EphA7, as well as their in vitro utilization by HOXA13 to direct gene expression, provides strong evidence that HOXA13 can regulate the tissue-specific expression of these receptor tyrosine kinases.

Studies examining the functional consequences of perturbations in Eph receptor signaling indicate that changes in cell migration, morphology, and adhesion are the major phenotypes in tissues lacking one or more Eph receptor or ephrin ligand (Shamah et al.,2001; Wahl et al.,2000; reviewed by Pasquale,2005; Cooke et al.,2005). In the developing vasculature, functional studies of EphA2, ephrin A1, ephrin B2, and EphB4 firmly establish a role for Eph–ephrin signaling throughout the angiogenic process, including endothelial cell proliferation, assembly, and extracellular matrix remodeling (McBride and Ruiz,1998; Gerety et al.,1999; Adams et al.,2001; Hunter et al.,2006).

A link between Hox proteins and the expression of ephrin ligands and Eph receptors has also been established. In the hindbrain, the loss of Hoxa1 and Hoxb1 directly affect the expression of EphA2 in rhombomere 4, whereas in the developing microvasculature, antisense oligos directed toward Hoxb3 also caused reductions in ephrin A1 in the vascular endothelia (Chen and Ruley,1998; Myers et al.,2000). Finally the loss of HOXA13 function has also been linked to reductions in EphA7 in the developing limb mesenchyme and EphA7 and EphA4 in the umbilical artery endothelia (Stadler et al.,2001).

While the characterization of the combinatorial functions of EphA6 and EphA7 in the GT vasculature await the production of a conditional EphA6 allele, the present body of evidence linking perturbations in Eph receptor signaling to angiogenic defects suggests that the combined reduction of EphA6 and EphA7 in the GT vasculature may cause a change in cell adhesion, which under vascular load could affect the overall diameter of the GT vessels. Functionally, the loss of EphA7 in the embryo causes defects in cell adhesion and anencephaly, although no defects in the developing vasculature were reported (Holmberg et al.,2000). Furthermore, the loss of HOXA13 function also affects the morphology of the umbilical artery endothelia, which exhibit reduced levels of EphA4 and EphA7 expression (Stadler et al.,2001).

In the limb, HOXA13 and HOXD13 function in a redundant manner to regulate EphA7 expression (Salsi and Zappavigna,2006). This finding provides a possible explanation for the maintenance of EphA6 and EphA7 expression in the GT mesenchyme of Hoxa13 homozygous mutants as both Hoxa13 and Hoxd13 are coexpressed strongly in this region (Warot et al.,1997, Scott et al., 2005). More importantly, among the group 13 HOX proteins, only HOXA13 has been reported to be expressed in vascular endothelia (this work; Warot et al.,1997; Stadler et al.,2001). Thus, in the GT vasculature, the loss of HOXA13 function would affect EphA7 expression more severely, due to the absence of functionally redundant factors such as HOXD13.

Of interest, our chromatin immunoprecipitation (ChIP) analysis in the GT did not detect HOXA13 binding to the site described by Salsi and Zappavigna (2006) in the limb, although we did verify that the A13 DNA binding domain peptide could bind this site in vitro. One possible explanation for this difference in binding site utilization is that HOX protein cofactors, such as the TALE class of DNA binding proteins, are also differentially expressed and can influence which DNA sequences are used by a particular HOX protein (reviewed by Moens and Selleri,2006; Villaescusa et al.,2004; Williams et al.,2005; Erickson et al.,2006). For HOXA13, interactions with its cofactor, MEIS-1B, appears to be specific to the genitourinary region and may influence which of the DNA sequences are regulated by HOXA13 in this tissue (Williams et al.,2005). Alternatively, DNA accessibility may also vary in limb versus genitourinary chromatin, which could also account for the differential binding of HOXA13 to gene-regulatory sequences in a tissue-specific manner (Vashee et al.,1998; Kodadek,1998).

Of interest, a common theme emerging from studies of HOXA13-deficient genitourinary tissues is that epithelial lineages appear to be affected in a similar manner. Indeed, the urethral plate epithelium (which serves as a signaling center for the GT), the vascular endothelia of the umbilical arteries, and the epithelia lining the developing bladder and ureter (data not shown) all exhibit changes in cell morphology and stratification with the loss of functional HOXA13 (present study, Perriton et al.,2002; Morgan et al.,2003). While alterations in epithelial cell morphology and stratification could reflect a loss in cellular identity, this possibility is unlikely as sonic hedgehog expression is maintained in the mutant urethral plate epithelium, and PECAM-1 expression is also maintained in the mutant vascular endothelia (this study; Morgan et al.,2003). Thus, altered epithelial morphology and stratification in the mutant GT vasculature is more likely to reflect a change in key matrix or cytoskeletal components. These factors are necessary for a tissue-specific morphological state, and it has been recently demonstrated that they are regulated by Eph–ephrin signaling (reviewed by Cheng et al.,2002; Harbott and Nobes,2005; Hunter et al.,2006; Marston and Goldstein,2006).


Mouse Strains

Hoxa13-GFP mutant embryos were derived from heterozygous intercrosses as described (Stadler et al.,2001; Morgan et al.,2003). The mutant HOXA13 allele encodes a fusion protein of HOXA13 and GFP where the last 34 amino acids of HOXA13, encoding the DNA contacting third helix (ISATTNLSERQVTIWFQNRRVKEKKVINKLKTTS), is removed and replaced with the enhanced GFP protein (Clontech). The nuclear localization, turnover, and tissue-specific expression of the HOXA13-GFP protein appeared similar to the wild-type protein (Stadler et al.,2001). Timed matings were used to establish embryonic gestational age in embryonic days, where E0.5 represents the first day of vaginal plug detection. EphA7 homozygous mutant embryos were produced by intercrosses of EphA7 heterozygous mutant mice, kindly provided by Jonas Frisén (Karolinska Institute, Stockholm, Sweden). EphA7 embryo genotypes were determined by PCR of yolk sac-derived DNA using the following primers: EphA7 mutant allele, 5′-CTAAGGTCCTATTTTGCCTG-3′, 5′-CATTACACTTCCAGACCTGGGAC-3′; EphA7 wild-type allele, 5′-CAGGAGTGGCCCGGGAA-3′, 5′-CATTACACTTCCAGACCTGGGAC-3′ and 40 cycles of 94°C (30 sec), 54°C (30 sec), 72°C (30 sec). All procedures using mice were done in accordance with an approved institutional animal protocol (A729 to H.S.S.).

GT Histology and Immunohistochemistry

Paraffin (Paraplast Plus, Fisher) -embedded Hoxa13 wild-type and homozygous mutant embryos (E12.5–E15.5) were sectioned at 7-μm intervals and placed sequentially onto Superfrost plus slides (Fisher) and stained with hematoxylin and eosin as described by Stadler and Solursh (1994). Sections containing the GT and its associated vasculature were photographed using a Leica DMLB2 microscope and a Q Imaging Digital camera.

E13.5 Hoxa13 heterozygous and homozygous mutant embryos were embedded in OCT (Tissue Tek) and sectioned as previously described (Morgan et al.,2003). Antibodies specific for EphA6 (R&D Systems MAB6071), EphA7 (R&D Systems MAB1495), and PECAM-1 (BD Pharmingen 553708) were used at dilutions of 1:200 and incubated on the sections at 4°C overnight. Secondary antibodies labeled with Cy5 were used as described by the manufacturer (Jackson Immunological). Imaging was performed on a Bio-Rad MRC1024 confocal microscope using Kalman filtering. Identical laser level, iris, and black level settings were used for all samples.

GT Endothelial Cell Isolation and RT-PCR

The distal half of the E13.5 GTs were isolated by microdissection in 1× phosphate buffered saline. E13.5 embryos were chosen because they strongly express Hoxa13, EphA6, and EphA7 within the GT vasculature as shown by immunohistochemistry and in situ hybridization. Approximately 5–8 GTs of identical Hoxa13 genotype were combined to gain adequate amounts of endothelial cells and isolated RNA. The tissues were pooled in individual Netwells (Costar) and treated with 0.2% Collagenase Type IV (Gibco) at 37°C for 30 min with occasional shaking. After collagenase treatment, the tissues were placed in digestion medium (0.1% trypsin/ethylenediaminetetraacetic acid [EDTA, Gibco], 0.2% Collagenase IV, in phosphate buffered saline [PBS]) for 15 min at 37°C, using gentle pipetting every 5 min to dissociate the tissue. Cell flow-through was collected in 15-ml tubes with 10 ml of quenching buffer (15% fetal bovine serum [#26140-087, Gibco] and 0.1% bovine serum albumin in D-PBS [BP1605-100, Fisher]). Finally, to collect any residual cells, the Netwell baskets were rinsed with 0.1% BSA/PBS and added to the cell flow-through. Cells were stored on ice for 5 min and spun at 3,000 rpm for 5 min, followed by an additional quenching wash and spin. Dynabeads were coated with PECAM-1 antibody (MEC 13.3, #553369 BD Pharmingen) as described by the manufacturer (Dynal). Cells (2.5 × 106) were combined with the Dynabeads–antibody complex (three times more beads than cells), and the mixture was incubated for 1 hr at 4°C on a rotating platform. The bead–antibody–cell complexes were isolated with a magnet, and the remaining PECAM-negative cells were collected as a control. The cells were gently washed with 0.1% BSA/PBS and collected for RNA extraction using RNA Stat-60 (CS-110, Tel-Test). GT mesenchyme RNA was isolated in a similar manner using dissected GTs from E13.5 embryos. RNA quality was analyzed by agarose gel electrophoresis and ultraviolet spectroscopy. One microgram of RNA was used for cDNA synthesis using the Superscript First-Strand Synthesis system (Invitrogen).

Semiquantitative RT-PCR

cDNAs derived from distal GT RNA or endothelial cell RNA was used for semiquantitative RT-PCR to detect the expression levels of EphA6, EphA7, Gapdh, Pecam-1, and Hoxa13 in Hoxa13 wild-type and homozygous mutants. For RT-PCR, each cDNA template was diluted 1:1 with water. The following gene-specific primer sequences were used: EphA6-For: 5′-GAGAGACCGTACTGGGAAATG-3′; EphA6-Rev: 5′-GCCTGTGGTTTCTCTCCTTC-3′ (NM007938, bp2901-3042); EphA7-For: 5′- CTCTTCGCTGCTGTTAGCAT-3′; EphA7-Rev: 5′-GTGATGACTCCATTGGGATG-3′ (BC026153, bp1433-1566); Pecam1-For: 5′-CCAGTGCAGAGCGGATAAT-3′; and Pecam1-Rev: 5′-GCACCGAAGTACCATTTCAC-3′ (NM008816, bp1487–1634); and Hoxa13-For: 5′-CTGGAACGGCCAAATGTACT-3′; Hoxa13- Rev: 5′-TATAGGAGCTGGCGTCTGAA-3′ (NM008264, bp952–1058).

Gapdh primers were previously described (Shou et al.,2005). All RT-PCR primer pairs were designed to flank an intronic sequence to distinguish any PCR products derived from genomic DNA contamination. Each RT-PCR analysis was tested from at least three independent RNA isolates.

EphA6 and EphA7 Promoter Analysis

Sequence analysis of the EphA6 promoter (Ensembl: ENSMUST00000068860) identified a single region (−2410 to −2067) containing multiple T-A-A motifs we previously identified as being bound by HOXA13 (Knosp et al.,2004). Analysis of the EphA7 locus (Ensembl: ENSMUSG00000028289) identified the HOXA13 binding site characterized by Salsi and Zappavigna (2006) as well as a novel region (−839 to −514) containing several T-A-A motifs that we examined in this report for HOXA13 binding and gene regulation. PCR amplification of the EphA6 and EphA7 promoter regions was performed using the following primers: EphA6P1F 5′-GATAGGCAGAATGCCAGGTG-3′; EphA6P1R 5′-GGAGCAAGGAAAGCTCAGAA-3′; EphA7P1F 5′-TGCCTCTCGAGTTACAGAACAG-3′; EphA7P1R 5′- GGGAGCACTTGGCTTTTAGC-3′. HOXA13 binding to the PCR-amplified EphA6 and EphA7 promoter fragments was determined using an EMSA as previously described (Knosp et al.,2004). Briefly, the amplified PCR products were radio-labeled with T4 Polynucleotide kinase and assessed for HOXA13 binding by incubation with a HOXA13 DNA binding domain peptide (A13) followed by nondenaturing acrylamide gel electrophoresis. Competitor DNA consisting of the identical unlabeled PCR product was used to assess the binding affinity for the EphA6 and EphA7 promoter sequences as described (Knosp et al.,2004). Interspecies comparisons of the EphA6 and EphA7 regions containing the HOXA13 binding sites were performed using the UCSC Genome Browser (Kent et al.,2002).

Oligomerization Analysis of the HOXA13 DNA Binding Peptide and DNA

Complexes of the A13 DNA binding domain peptide and DNA were assessed for their oligomerization state using sedimentation equilibrium ultracentrifugation on a Beckman Coulter XL-1A Protein Analysis System as described (Huffman et al.,2001). A self-annealing fluorescein-labeled oligonucleotide, 5′-6-FAM-CCCATAAACCCCCCCGGTTTATGGG-3′ (5 μM) was combined with the A13 DNA binding peptide (6 μM to 10 μM) in a buffer containing 80 mM KCl, 10 mM MgCl2, 0.2 mM EDTA, 1 mM dithiothreitol, and 20 mM Tris.HCl pH 7.8. The samples were centrifuged at 4°C at 26,000 or 32,000 rpm in an AN-60Ti rotor equipped with 12-mm path length Epon double sector cells. The absorbance was monitored at 480 nm as a function of radial distance. Molecular weights of the DNA and DNA + protein complexes were calculated from a nonlinear least squares fit of the absorbance data using the software supplied with the Model XL-1A system. A density value for the solvent of 1.005 g/ml was used for the calculations. The partial specific volumes of the peptide–DNA complexes were estimated by taking the mass-weighted average of the partial specific volumes for the free oligonucleotide (8,100 daltons) and free peptide (8,137 daltons). Using the additivity method of Cohn and Edsall (1943) the partial specific volume of the free peptide was estimated to be 0.745 ml/g. The partial specific volume of the free oligonucleotide was determined using a separate sedimentation equilibrium analysis in identical buffers without peptide. The partial specific volume of unbound oligonucleotide was determined to be 0.512 ml/g. Using the mass-weighted partial specific volume averages for the peptide and DNA, a partial specific volume of 0.629 g/ml was calculated for the samples containing peptide and DNA in a 1:1 ratio. Using this value, the molecular weight of the complex was determined to be 16,900 daltons, indicating that the HOXA13 DNA binding domain peptide associates with DNA in a 1:1 stoichiometric ratio, confirming monomeric binding to the EphA6 and EphA7 gene-regulatory elements.

ChIP Assays

ChIP was performed using a HOXA13 antibody and whole GTs dissected from E12.5 embryos. Multiple attempts to isolate sufficient quantities of endothelial cells for ChIP were not successful as each GT yielded less than 1,000 endothelial cells, whereas 106 cells are required to produce sufficient chromatin quantities for immunoprecipitation (Lavrrar and Farnham,2004). Therefore, all PCR amplification of the HOXA13 bound DNA regions were derived from whole GT chromatin isolates. The 300-bp fragments encompassing bases −3000 to +1 of the EphA6 (Ensembl: ENSMUST00000068860) and EphA7 (Ensembl: ENSMUSG00000028289) promoters were examined for in vivo association with HOXA13. The GTs were dissected in PBS containing 15 μl/ml protease inhibitor cocktail (PIC, Sigma).

Tissues were fixed in 1% formaldehyde/PBS and rocked at room temperature for 10 min. Protein–DNA cross-linking was stopped by the addition of glycine to a final concentration of 0.125 M for 5 min. Next, the samples were centrifuged at low speed, and the pellet was washed once with cold PBS containing PIC and centrifuged at low speed. The pellet was resuspended in 100 μL cell lysis buffer (5 mM PIPES, pH 8.0/85 mM KCl/0.5% NP40) plus PIC and incubated on ice for 10 min. Next, the lysate suspension was microcentrifuged at 5,000 rpm for 5 min at 4°C, followed by resuspension in 50 μl of nuclear lysis buffer (50 m Tris-HCl, pH 8.1, 10 mM EDTA, 1% sodium dodecyl sulfate) plus PIC and incubation for 10 min on ice. The lysed nuclei were sonicated for 20 periods of 30 sec ON and 1 min OFF at 4°C using a Bioruptor (Cosmo Bio) to produce sheared chromatin of an average length of 200–1,000 bp. The sheared chromatin was microcentrifuged at 13,000 rpm for 10 min at 4°C and the supernatant was transferred to a new tube.

ChIP was performed using a ChIP Assay Kit as described by the manufacturer (Upstate Biotechnologies/Millipore). Each chromatin supernatant was precleared with 40 μl of salmon sperm DNA/Protein A Agarose (Upstate Biotechnology). The chromatin samples were incubated with the HOXA13 antibody or IgG control antibody on a rotating platform at 4°C for 3 hr. Washes, DNA elution, and reverse cross-linking were performed as described in the Upstate ChIP Assay Kit. Samples were ethanol precipitated, resuspended in 100 μl of TE, and DNA purified using the Qiaquick PCR Purification kit (Qiagen).

The eluted DNA from the HOXA13 antibody, control IgG, or no antibody samples were assessed for the presence of the EphA6 and EphA7 promoter DNA using PCR. Primers used to amplify the ChIP-positive regions were identical to those used to amplify the candidate binding site regions described above.

Epha6 and Epha7 Luciferase Assays

NG108-15 cells (ATCC#HB-12317) were maintained and transfected as previously described (Knosp et al.,2004). Transfections were performed in 12-well plates (Costar) using 1.5 μg of a pGL4 plasmid containing either the forward (PGL4.1-EphA6F or PGL4.1-EphA7F) or reverse (PGL4.1EphA6R or PGL4.1-EphA7R) orientations of the EphA6 or EphA7 ChIP-positive regions as well as 0.25 μg of the normalization vector, pRL-CMV Renilla, and 0.5 μg pCMV-Hoxa13-HA or an empty pCMV control plasmid per well. Cell lysates were processed to detect luciferase activity using the Dual-Glo Luciferase Assay System (Promega) in OptiPlate-96F black plates (Packard) as described (Knosp et al.,2004). Luciferase activity was detected using a Packard Fusion Microplate Analyzer (Perkin Elmer). Sample wells were read three times for 1 sec each and averaged. Three replicates of each transfection were performed, and each transfection assay was repeated three times. Results were normalized for transfection efficiency using relative Renilla luciferase expression levels as described by the manufacturer (Promega) and plotted using SigmaPlot 9.0 (Systat).


The authors thank Eric Steele, Hans Peter Bächinger, and the Shriners Hospital for Children Analytical Core for their assistance with the sedimentation equilibrium analysis. W.M.K. was funded by Predoctoral Fellowships from the National Institutes of Health, and C.A.S. was funded by the American Heart Association.