Endothelial cells (EC) express several members of the Homeobox (Hox) gene family, suggesting a role for these morphoregulatory mediators during angiogenesis. We have previously established that Hox D3 is required for expression of integrin αvβ3 and urokinase plasminogen activator (uPA), which contribute to EC adhesion, invasion, and migration during angiogenesis. We now report that the paralogous gene, Hox B3, influences angiogenic behavior in a manner that is distinct from Hox D3. Antisense against Hox B3 impaired capillary morphogenesis of dermal microvascular EC cultured on basement membrane extracellular matrices. Although levels of Hox D3-dependent genes were maintained in these cells, levels of the ephrin A1 ligand were markedly attenuated. Capillary morphogenesis could be restored, however, by addition of recombinant ephrin A1/Fc fusion proteins. To test the impact of Hox B3 on angiogenesis in vivo, we constitutively expressed Hox B3 in the chick chorioallantoic membrane using avian retroviruses that resulted in an increase in vascular density and angiogenesis. Thus, while Hox D3 promotes the invasive or migratory behavior of EC, Hox B3 is required for the subsequent capillary morphogenesis of these new vascular sprouts and, together, these results support the hypothesis that paralogous Hox genes perform complementary functions within a particular tissue type.

Introduction

The Homeobox (Hox) morphoregulatory genes encode transcription factors that play an essential role in organogenesis during development and, more recently, have been linked to both hormonal and pathologically induced tissue remodeling in adults (for review Boudreau and Bissell 1998; Stelnicki et al. 1998; Chen and Capecchi 1999; for review Cillo et al. 1999). Although the identity of many genes whose expression is modulated by binding of Hox proteins to target DNA sequences remains to be firmly established, a growing body of evidence indicates that genes associated with cell–cell and cell–extracellular matrix (ECM) interactions are putative targets of Hox activity ( Jones et al. 1992; Li and Gudas 1996; Boudreau et al. 1997; Lorentz et al. 1997).

To this end, we have been investigating a potential role for homeobox genes during angiogenesis, a process which involves extensive, coordinated changes in cell–cell and cell–ECM interactions. In response to angiogenic stimulators, including the tumor or wound microenvironment, normally quiescent endothelial cells (EC) upregulate expression of proteinases, including matrix metalloproteinases and urokinase plasminogen activator (uPA), which degrade the surrounding basement membrane (BM) ECM (for review see Werb et al. 1999). As EC reenter the cell cycle, they also upregulate expression of adhesion molecules, including αvβ3 integrin, which allows the cells to adhere and migrate into the adjacent stromal matrix environment ( Brooks et al. 1994). After this, EC must then resynthesize and deposit a new BM ECM and undergo morphological reorganization into tubular structures complete with a lumen ( Stromblad and Cheresh 1996). Maturation of the newly formed capillaries follows via recruitment of pericytes, which strengthen and stabilize the vascular wall ( Hirschi and D'Amore 1996; Maisonpierre et al. 1997).

We previously identified several class I Homeobox genes expressed in cultured EC and showed that one of these, namely Hox D3, is required for expression of the β3 subunit of the αvβ3 integrin, as well as for expression of the uPA. When constitutively expressed in EC in vivo, Hox D3 produced endothelioma-like structures, consistent with a role for this gene in mediating the invasive and migratory behavior of EC during the early stages of neovascularization ( Boudreau et al. 1997). Recent work by others has also shown that EC upregulate the expression of several members of the Hox B cluster or splice variants of Hox A9 in response to a variety of angiogenic cytokines ( Belotti et al. 1998; Patel et al. 1999). The role of these genes during angiogenesis, however, remains to be established. Given the complex multistep nature of angiogenesis and the coordinate changes in cell–cell and cell–ECM interactions, it is likely that many of the Hox genes expressed in EC contribute to this process. Having established a role for Hox D3 during the early stages of angiogenesis, we are particularly interested in establishing a role for Hox genes that may act at later stages of angiogenesis, which involve resynthesis of BM ECM and/or acquisition of a tubular three-dimensional capillary morphology.

Evidence gathered from single and compound Hox null mice have suggested that paralogous Hox genes (i.e., those with similar numerical designations, but located on different chromosomes such as Hox D3, Hox B3, and Hox A3) may play complementary, overlapping, or synergistic roles in a particular tissue ( Condie and Capecchi 1994; Manley and Capecchi 1997). For example, whereas deletion of a single member of Hox 9 paralogous group did not significantly impact postnatal mammary gland development, deletion of Hox A9, B9, and D9 dramatically impaired the normal branching morphogenesis and differentiation of this tissue ( Chen and Capecchi 1999). Therefore, we wished to investigate whether paralogues of Hox D3 also contribute to angiogenic behavior in EC and whether these genes acted at the same or later stages of neovascularization. We had previously observed that the paralogous gene Hox B3 is highly expressed in adult EC ( Boudreau et al. 1997) and thus wished to establish whether this gene acted in concert with Hox D3 and/or via distinct mechanisms to influence endothelial cell behavior.

Materials and Methods

Cells, Culture Conditions, and Cytokines

An immortalized human dermal microvascular endothelial cell line (HMEC-1; Ades et al. 1992), was a gift from T. Lawley (Emory University, Atlanta, GA). These cells previously have been shown to maintain many properties of primary dermal microvascular cells in culture, including the ability to undergo capillary morphogenesis when cultured on BM (Matrigel) and maintain expression of a number of endothelial cell surface markers ( Xu et al. 1994). Cells were maintained in media MCBD 131 supplemented with 10% FCS, gentimicin, and 1% hydrocortisone (Sigma Chemical Co.) and passaged using calcium- and magnesium-free PBS supplemented with 0.053 mM EDTA. Primary cultures of HMEC were purchased from Clonetics. Recombinant mouse ephrin A1/human IgG fusion proteins, recombinant human VEGF, and basic fibroblast growth factor (bFGF) were purchased from R&D Systems. Recombinant human TNF α and Matrigel were obtained from Collaborative Research. Endothelial cell culture on BM (Matrigel) was performed as previously described ( Boudreau et al. 1997). To release cells from Matrigel for protein or mRNA determination, cultures were suspended in PBS without Ca2+ or Mg2+, containing 0.5 mM EDTA and incubated on ice for 1 h to allow the Matrigel to disperse. Cells were then pelleted by centrifugation at 500 g for 5 min at 4°C. For the ephrin add back experiments, 250 ng/ml recombinant mouse ephrin A1 fused to human IgG was diluted in serum-free MCDB 131 and preclustered for 1 h at room temperature using 25 ng/ml of an antibody against the Fc region of human IgG (Sigma Chemical Co.) as described ( Wang and Anderson 1997). Clustered ephrins were added to cells at the time of plating on Matrigel.

RNA Isolation and Northern Blot Analysis

Total RNA was isolated from HMEC-1 using the Qiagen RNEasy kit. For Northern blot analysis, a total of 10 or 20 μg total RNA was electrophoresed through 1% agarose formaldehyde gels as previously described using standard methods ( Boudreau et al. 1997). Ribosomal RNA was visualized by staining with 1% ethidium bromide. 32P{dCTP} probes were prepared using the Ambion Decaprime Kit and purified using Sephadex G-25 columns (Boehringer Mannheim Corp.). Blots were probed with 1 × 106 cpm/ml of hybridization buffer (Hybridsol I; Oncor) and exposed to Kodak M5 X-Omat film at −70°C. cDNA for integrin β3 was a gift from David Cheresh (The Scripps Research Institute, La Jolla, CA).

Reverse Transcriptase PCR Measurement of Hox B3 in EC

1 μg of total RNA was reverse transcribed using MMuLV RT for 1 h in total volume of 25 μl. 0.1, 1, 2, and 10 μl of this RT reaction was then amplified for 20, 30, or 35 cycles of 95, 58, and 72°C for 30, 30, and 90 s, respectively, with the following primers: forward primer 5′ cgatgcagaaagccacctactacgac 3′ corresponding to bp 362–387 of human Hox B3, and reverse primer 5′ cgccgaccccggggggctcttct 3′ corresponding to nucleotides 1,666–1,682 of the published sequence. The expected 1.32-kb PCR product was visualized by electrophoresis on 1% agarose gels containing ethidium bromide. From this analysis, it was determined that amplification of 1 μl of the total 25 μl RT reaction for 30 cycles gave optimal, reproducible results within the linear range for amplification. To normalize for total RNA, 1 μl of the same RT reaction was diluted 1:800 in water and amplified under the same conditions with commercially available primer sets for human GAPDH or β-actin (Stratagene). The 1.32-kb PCR product corresponding to Hox B3 was subsequently ligated into the TOPO II TA cloning vector (Invitrogen Corp.) and the identity confirmed as Hox B3 by dideoxy sequencing using the USB Sequenase 2 kit (Nycomed Amersham, Inc.).

Construction of Hox B3 Sense and Antisense Expression Vectors

The insert containing the entire cDNA encoding human Hox B3 was subcloned into the expression vector PCR 3.1 containing a CMV promoter (Invitrogen Corp.). The orientation was confirmed by restriction digest and the ends were resequenced to confirm the identity and orientation. Clones in the antisense orientation were directly transfected into HMEC-1 using a CaP04 method and stable transfectants selected in the presence of 50 μg/ml G418. To achieve high levels of translation and expression of Hox B3 in the sense orientation, we introduced a Kozak consensus sequence on the 5′ end by reamplifying the cDNA with the primer 5′ ggaattcggccaccatgcaga 3′. The resulting cDNA was religated into the same PCR 3.1 expression in the sense orientation, stably transfected into HMEC-1, and stable pools were selected as described above. To allow us to distinguish transgene expression from endogenous Hox B3, we have also added a COOH-terminal 6×His epitope tag by deleting the stop codon and cloning into the PCR 3.1 mycHis vector (Invitrogen Corp.). This construct was subsequently cloned into the CK proviral vector described below.

Western Blotting

Total protein was isolated from EC using ice-cold 10 mM Tris/150 mM NaCl in the presence of 10 μg/ml aprotinin, pepstatin, and leupeptin, and 0.02 M PMSF. Protein concentration was determined using a protein assay kit (BioRad). For Western analysis, a total of 20 or 50 μg protein lysates were run on SDS-PAGE and transferred to PVDF membranes (Immobilon; Nycomed Amersham Inc.). Membranes were blocked in 5% milk protein in 10 mM TBS, pH 7.6. Polyclonal antibodies against Hox B3 were purchased from Berkeley Antibody Co. Polyclonal rabbit anti–human antibodies against ephrin A1 were purchased from Santa Cruz Biotechnology (sc-911). For Western Blot analysis, a 1:2,000 or 1:500 dilution was used as indicated, followed by incubation with HRP-conjugated goat anti–rabbit antibodies (Nycomed Amersham Inc.). To reduce nonspecific background, blots were washed in 1 M NaCl followed by several washes in TBS, pH 7.6. Bands were visualized by enhanced chemiluminescence (ECL Plus; Nycomed Amersham Inc.).

cDNA Probes for Ephrin A1 and Ephrin B1

A cDNA probe corresponding to ephrin A1 was isolated using the following primer pairs: forward 5′ ggaacccagacccataggagac 3′′ corresponding to nucleotides 17–38, and the reverse primer 5′ ttcagcctgtccctctttaaga 3′ corresponding to nucleotides 711–732 of the human sequence (GenBank/EMBL/DDBJ accession numbers M57730 and M37476). The resulting 0.7-kb PCR product was ligated into the TOPO cloning vector and the identity confirmed by sequencing as described. The entire 700-bp insert was used for Northern blot analysis. A probe for ephrin B1 was also generated by RT-PCR from 1 μg total RNA for 30 cycles at 95, 58, and 72°C for 30, 30, and 90 s, respectively, using the following primer pairs: forward 5′ ttg gtg agg agg cgc caa gg 3′ corresponding to nucleotides 653–672, and the reverse primer 5′′ ggg cac tca gac ctt gta gta ga 3′ corresponding to nucleotides 1,748–1,726 of the published sequence of human ephrin B1 (GenBank/EMBL/DDBJ accession number NM_004429). The 1,095-bp PCR product was ligated into the TOPO II PCR cloning vector (Invitrogen Corp.) and the identity of the PCR product confirmed as ephrin B1 by dideoxy sequencing as described above.

Immunoprecipitation of Eph A2 Receptor

EphA2 receptor was measured by immunoprecipitation of equal amounts of protein lysates of HMEC-1 using a polyclonal antibody against human Eph A2 (sc-924; Santa Cruz Biotechnology). HMEC-1 were lysed in ice-cold 10 mM Tris-HCl/150 mM NaCl, pH 7.6, in the presence of 2 mM NaF, 2 mM orthovanadate, 10 μg/ml aprotinin, pepstatin, and leupeptin, and 0.02 M PMSF. After determination of protein concentrations, 300 μg of cellular lysate was diluted in RIPA containing 2 mM NaF and 2 mM orthovanadate, and immune complexes precipitated using 25 μl of a 10% wt/vol solution of protein A–Sepharose (Sigma Chemical Co.). Pellets were washed three times in RIPA, followed by two additional washes in PBS, and resuspended in 2× Laemmli sample buffer. After separation on 7.5% SDS-PAGE and transfer to PVDF membranes, blots were probed with antibodies against Eph A2 or antibodies against total phosphotyrosine (clone 4G10; Upstate Biotechnology).

Migration Assays

Endothelial cell migration assays were performed using a modification of the method described by Klemke et al. 1997. In brief, migration assays were performed using tissue culture-treated Transwell chambers of 6.5-mm diam, with 8-μm pores (Costar Corp.). The bottom surfaces of the chambers were coated with 20 μg/ml of bovine fibrinogen (Sigma Chemical Co.) or 1% BSA for controls. Before the assay, wells were rinsed twice with migration buffer (fibroblast basal medium + 0.5% BSA; Clonetics). EC were serum-starved for 18 h and removed from the culture dishes by harvesting with Ca2+- and Mg2+-free PBS containing 0.53 mM EDTA, and resuspended in migration buffer. 50,000 cells were added to each well and the assay was allowed to proceed for 5 h at 37°C. Nonmigratory cells were removed from the upper chambers with a cotton swab and cells that had migrated to the bottom of the membrane were visualized by fixing and staining with Diff-Quick (VWR Scientific Products). The total number of cells that had migrated was determined by counting at least five different fields using a Nikon inverted TMS microscope. Alternatively, quantitative analysis of migration was performed by measuring absorbance of methylene blue stained cells at 600 nm. Each determination represents the average of three individual wells and error bars represent SD.

Construction and Application of Avian Replication-defective Retroviruses Expressing Hox B3 in Chick Chorioallantoic Membrane

To produce replication defective retroviruses encoding Hox B3, the cDNA encoding the entire human Hox B3 sequence was excised with HindIII and PmeI and inserted into the proviral vector CK at the HindIII site and an XbaI site that was blunted with Klenow polymerase. The viral packaging cell line, Q4dh, derived from the QT6 quail fibrosarcoma cells line ( Stoker and Bissell 1988), was maintained in M199 containing 4% FCS, 1% chicken serum, and 1× tryptose phosphate broth. Stable transfectants expressing either human Hox B3 proviral vectors or empty vector (CK) were performed using CaP04, and pools of stable transfectants producing infectious virus were selected in the presence of 200 μg/ml G418 as described ( Stoker and Bissell 1988; Boudreau et al. 1997). To induce angiogenesis, we grafted 5 × 106 Q4dh cells in a volume of 50 μl of medium M199 onto the chick chorioallantoic membranes (CAMs) of 10-day chick SPAFAS pathogen-free chick embryos as previously described ( Boudreau et al. 1997). After 72 h, CAMs were harvested and vascular density, morphology, and immunohistochemical analysis were performed. Angiogenesis was quantitated by counting the number of branch points arising from the tertiary vessels in a 6-mm square area on the underside of the tumors. Measurements were made in 12 samples infected with control virus and 12 samples infected with Hox B3 expressing virus from three separate experiments. Statistical significance was assessed using a paired t test.

Tissue Fixation and Immunofluorescence

CAMs were fixed in situ in 4% paraformaldehyde, embedded in OCT medium, and frozen in a dry ice/ethanol bath and stored at −70°C. 7-μm sections of the CAMS were used for immunohistochemistry. After brief acetone fixation, sections were air-dried and subsequently blocked in PBS containing 2% BSA for 1 h, followed by staining with appropriate antibodies. A 1:200 dilution of a monoclonal anti–his-6 antibody (Boehringer Mannheim Corp.) diluted in PBS containing 0.025% Triton X-100, 0.1% BSA, or a 1:200 dilution of a polyclonal rabbit anti–human antibody against von Willebrand factor were used, followed by incubation for 1 h with FITC-conjugated goat anti–mouse (Cappel Laboratories) or Texas red-conjugated goat anti–rabbit secondary antibody (Calbiochem-Novabiochem Corp.).

Results

Expression of Hox B3 in Cultured Endothelial Cells

As the levels of Hox mRNA and protein expression in mammalian cells are normally low, we established conditions to measure Hox B3 using semiquantitative RT-PCR. Using the RT-PCR conditions described in Materials and Methods, we assessed expression levels of Hox B3 in both primary human dermal microvascular cells or the immortalized HMEC-1 cultured under various conditions. Although EC continue to proliferate and form a cobblestone monolayer when plated on untreated tissue culture dishes, culturing on dishes coated with reconstituted BM induces EC to become quiescent and adopt a three-dimensional capillary-like morphology ( Kubota et al. 1988; Boudreau et al. 1997). Therefore, we compared expression of Hox B3 in cells cultured in the presence or absence BM ( Fig. 1 A). In contrast to our previous observations showing that BM suppressed expression of Hox D3, we observed that Hox B3 was expressed at slightly higher levels in HMEC-1 cultured on BM for 24 h, as compared with cells maintained in the absence of BM ( Fig. 1 A). Similar results were observed in primary cultures of dermal microvascular EC (not shown). We also treated EC with the angiogenic cytokine, bFGF, and observed that levels of Hox B3 were not increased by this treatment ( Fig. 1 B). This distinct expression profile of Hox B3 suggested that, perhaps, Hox B3 and Hox D3 contributed to different endothelial cell phenotypes.

Suppression of Hox B3 Impairs Capillary Morphogenesis of EC Cultured on BM ECM

Because levels of Hox B3 are maintained when EC are cultured on BM, we reasoned that Hox B3 may contribute to the capillary-like morphogenesis that EC undergo when they are cultured on reconstituted BM. Therefore, we stably transfected an immortalized line of dermal HMEC-1 with CMV-driven plasmids expressing antisense against Hox B3 or with empty plasmid (control). Western blot analysis of stable pools of cells transfected with antisense against Hox B3 displayed a decrease in Hox B3 protein, as compared with control transfected cells ( Fig. 2 A). In contrast to controls, we also observed that when cells with reduced levels of Hox B3 were cultured on BM ECM, many formed aggregates and were unable to branch or form capillary-like tubules, indicating that Hox B3 is necessary for BM-dependent capillary morphogenesis ( Fig. 2 B). Furthermore, we also observed that BM-induced capillary morphogenesis of EC is not influenced by blocking Hox D3 or its putative target, integrin αvβ3. In addition, we also observed that blocking either Hox D3 or Hox B3 using antisense did not influence expression of the paralogous genes ( Boudreau et al. 1997; data not shown). Together, these results indicate that the paralogous Hox B3 and Hox D3 genes have distinct effects on EC behavior.

Impaired Capillary Morphogenesis in the Absence of Hox B3 Correlates with Decreased Levels of Ephrin A1, but Not the EphA2 Receptor in EC

Although many factors have been reported to influence capillary morphogenesis in cultured EC, several recent reports have implicated interactions between ephrin ligands and their corresponding Eph receptors in the morphogenesis of EC cultured on BM and during angiogenesis in vivo ( Pandey et al. 1995; Daniel et al. 1996; Adams et al. 1999). In addition, as Hox genes have been linked to regulation of expression of the EphA2 receptor during development, we wished to further explore a potential link between Hox B3 and ephrin or Eph receptor-expression in EC ( Chen and Ruley 1998; Studer et al. 1998). We initially compared the levels of the EphA2 receptor by immunoprecipitation of lysates from control transfected HMEC-1 or HMEC-1 transfected with antisense against Hox B3 ( Fig. 3 A). Although the levels of the Eph A2 receptor were similar, subsequent blotting revealed that phosphorylation of the EphA2 receptor was dramatically reduced in cells lacking Hox B3 ( Fig. 3 A). These findings further suggested that the levels of an ephrin ligand that may bind to and activate phosphorylation of the EphA2 receptor may be reduced in cells lacking Hox B3. Therefore, we performed a Western blot analysis with antibodies against the ephrin A1 ligand and observed that its expression was significantly reduced in EC transfected with antisense against Hox B3 ( Fig. 3 A). Furthermore, the reduction of ephrin A1 protein and mRNA ( Fig. 3A and Fig. B) appeared to be somewhat selective, since levels of another ephrin ligand, ephrin B1, were not affected ( Fig. 3 B). Importantly, the level of integrin β3 ( Fig. 3 B) and uPA (not shown), two targets of Hox D3 activity, were unaffected. These results indicate that paralogous Hox genes have distinct target genes within a particular tissue type.

Capillary Morphogenesis Can Be Enhanced by Addition of Ephrin A1 in Cells Expressing Antisense against Hox B3

To determine whether restoring ephrin A1 ligand to EC lacking Hox B3 could help these cells regain a three-dimensional capillary-like morphology, cells were treated with preclustered, recombinant mouse ephrin A1/human Fc fusion proteins. Fig. 4 A shows that addition of clustered mouse ephrin A1/Fc fusion protein was capable of eliciting high levels of phosphorylation of the human EphA2 receptor in both control transfected HMEC-1 and those transfected with antisense against Hox B3. More significantly, addition of recombinant ephrin A1/Fc fusion proteins partially restored the ability of cells lacking Hox B3 to elongate and form branching networks resembling the three-dimensional capillary networks produced by untreated control transfected cells ( Fig. 4 B). This ephrin A1-enhanced capillary morphogenesis was observed as early as four hours after plating on reconstituted BMs and persisted for at least 18 h. Thus, the Hox B3-dependent expression of ephrin A1 contributes to the normal capillary-like morphogenesis acquired by EC cultured on reconstituted BM.

Overexpression of Hox B3, but Not Hox D3, Results in Increased Levels of Ephrin A1 Expression

To further establish a relationship between Hox B3 and ephrin A1 expression, we transfected HMEC-1 with vectors containing CMV-driven human Hox B3. Western blot analysis of stable pools of cells transfected with Hox B3 showed increased levels of Hox B3 protein, as compared with HMEC-1 transfected with empty vector (control; Fig. 5 A). Cells overexpressing Hox B3 also showed the predicted increase in expression of ephrin A1 protein ( Fig. 5 A). Immunoprecipitation with an anti-EphA2 antibody of lysates from HMEC-1 transfected with Hox B3 plasmid or empty vectors indicated that, although levels of the EphA2 receptor were not increased, subsequent blotting with antiphosphotyrosine antibodies revealed an increased level of phosphorylation of the receptor in cells overexpressing Hox B3 and ephrin A1 ligand ( Fig. 5 B). HMEC-1 overexpressing Hox B3 also show the expected increase in ephrin A1 mRNA levels ( Fig. 5 C), whereas HMEC-1 transfected with the paralogous gene Hox D3 do not upregulate expression of ephrin A1, but do show increased expression of β3 integrin mRNA ( Fig. 5 D). Furthermore, we directly compared the ability of EC overexpressing Hox B3 or Hox D3 to migrate into a fibrinogen-rich environment. We observed that, whereas the Hox B3 transfected cells migrated at a rate similar to control transfected cells, Hox D3 transfected EC that express high levels of integrin αvβ3, show a significantly enhanced ability to migrate in this environment ( Fig. 5 E). These results again emphasize that Hox B3 and Hox D3 specifically modulate expression of different angiogenic effector molecules in cultured EC, which in turn, contribute to different aspects of the angiogenic process.

High Levels of Hox B3 Promote Angiogenesis In Vivo

Based on the above results, we predicted that constitutive overexpression of Hox B3 in vivo would yield a phenotype that would be distinct from the hemangiomas that arise in response to constitutive expression of Hox D3. In addition, based on our observations that Hox B3 was required for EC to acquire a tubular three-dimensional capillary morphology, we also reasoned that constitutive expression of Hox B3 in vivo may promote capillary morphogenesis of invasive vascular sprouts induced in response to an angiogenic stimuli. To this end, we induced 10-d-old chick CAMs to undergo angiogenesis by grafting tumorigenic viral packaging cells onto the CAM. The packaging cells were transfected with either empty proviral vector (CK) or retrovirus expressing human Hox B3. After 72 h, we observed that CAMs infected with retrovirus expressing Hox B3 displayed a significant increase in vascular density and capillary branching, as compared with CAMs infected with empty virus (18 ± 6.5 branch points/6 mm2 vs. 9.33 ± 3.78 branch points/6 mm2; n = 12, P < 0.05; Fig. 6). Similar increases in capillary branching/vascular density was observed in at least 24 different embryonic membranes examined from four different experiments. Furthermore, we did not observe any evidence of hemangioma-like structures characteristic of EC constitutively expressing Hox D3 ( Boudreau et al. 1997). Thus, both in culture and in vivo, Hox D3 and Hox B3 appear to differentially influence endothelial gene expression and angiogenesis (Table).

Finally, to demonstrate that the increase in angiogenic activity induced in the CAM was related to the presence of the retrovirally transduced Hox B3, we generated a replication defective retroviral vector expressing Hox B3 fused to a 6×His epitope tag on the COOH terminus. Similar to the results seen with the nontagged Hox B3, infection with the His-tagged Hox B3 also produced increase in endothelial cell density and capillary branches in CAMs after 72 h. Staining of serial 7-μm sections showed that the increased endothelial cell density, indicated by positive staining with von Willebrand factor, colocalized with staining for the His epitope present in the virally expressed Hox B3 ( Fig. 7).

Discussion

We have shown that the Homeobox gene, Hox B3, which is expressed by HMEC-1, helps promote capillary morphogenesis in culture and angiogenesis in vivo. Furthermore, Hox B3 influences angiogenesis in a manner that is distinct from the previously described proangiogenic effects of the paralogous gene, Hox D3 ( Boudreau et al. 1997). Specifically, whereas Hox D3 upregulated expression of both uPA and integrin αvβ3, which in turn facilitate endothelial cell migration, adhesion, and invasion, Hox B3 induced expression of ephrin A1 and promoted capillary morphogenesis of sprouting EC. Furthermore, this phenotype was reiterated in vivo as Hox B3 promoted branching and angiogenic behavior, in contrast to the hemangioma-like structures generated by Hox D3 ( Boudreau et al. 1997). Together, these findings support earlier observations made in mice lacking two or more paralogous Hox genes, which suggested that, although individual Hox genes performed distinct functions in a given tissue type, together, paralogous Hox genes acted in concert to more dramatically influence overall tissue phenotype and morphology ( Condie and Capecchi 1994; Manley and Capecchi 1997; Chen and Capecchi 1999).

The requirement for Hox B3 during BM-induced capillary morphogenesis appears to stem in large part from regulation of the expression of the ephrin A1 ligand. Although previous studies have linked both Hox A1 and Hox B1 to expression of the ephA2 receptor ( Chen and Ruley 1998; Studer et al. 1998), this is the first report that expression of the ephrin ligands may also be regulated by homeobox genes. It is of interest to note, however, that another ephrin ligand, ephrin B1 (Lerk 2), was identified previously in a screen for retinoic acid-inducible genes ( Bouillet et al. 1995), as retinoic acid is amongst the most potent inducers of anterior class I Homeobox genes known ( Langston and Gudas 1994).

Recent work has established morphoregulatory roles for the large family of ephrin ligands and their tyrosine kinase Eph receptors in both neural pathfinding and vascular development, and angiogenesis (for review see Holder and Klein 1999). In cultured adult EC, the type of ephrin ligand required for BM-induced capillary morphogenesis depends upon the vascular site of origin of the EC. For example, branching morphology in renal microvascular cells requires ephrin B1, whereas capillary morphogenesis of umbilical vein EC depended on the addition of ephrin A1 ( Daniel et al. 1996). Our results using EC derived from the dermal microvasculature indicate that ephrin A1 is a potent mediator of capillary morphogenesis in EC derived from this tissue.

Although interaction with, and subsequent phosphorylation of, a corresponding Eph receptor are required for the morphological effects induced by ephrins, it is not clear whether these interactions enhance EC migration or adhesion, or alternatively induce cellular repulsion (and perhaps promote branching of vascular sprouts), as has been observed with cultured neurons ( Wang and Anderson 1997). One recent report also suggested that ephrin B1 may stimulate migration of renal microvascular EC via activation of αvβ3 integrin-mediated attachment ( Huynh-Do et al. 1999). However, as αvβ3 is not required for adhesion to BM ECM, and blocking αvβ3 does not influence BM-induced capillary morphogenesis in culture, it is unlikely that the ephrin A1-induced morphological changes we observed arose via activation of αvβ3. It is intriguing, however, to consider the possibility that ephrin A1 may activate integrins that mediate EC adhesion to laminin, the major component of BMs required for capillary morphogenesis ( Kubota et al. 1988).

Our results suggest that, in addition to ephrin A1, Hox B3 may also induce expression of other genes that promote capillary morphogenesis. For example, although addition of recombinant ephrin A1 helped restore normal capillary-like morphology in many of the EC lacking Hox B3, a small proportion of cells consistently remained in clusters and failed to elongate or form tubules, despite the intense phosphorylation of the ephA2 receptor that followed addition of ephrin A1. Although the incomplete responsiveness to exogenous ephrin A1 may be attributed to heterogeneity amongst the EC population, it is also worth noting that since Hox genes are considered to be master regulatory mediators ( Botas 1993; Duboule 1998), expression of several, different morphoregulatory genes may be regulated by Hox B3 in EC. Our preliminary results indicate that EC lacking Hox B3 also show decreased expression of laminin β2 (LAMB2; data not shown). A role for this laminin isoform in BM-induced capillary morphogenesis is currently under investigation.

Other potential targets of Hox B3 include the thyroid transcription factor, TTF-1. Studies in HeLa and NIH 3T3 cells showed that TTF-1 expression was specifically activated by Hox B3, but not the paralogous Hox D3 ( Guazzi et al. 1994). Whether TTF-1 is also a target of Hox B3 in EC is not known. Studies that examined the effects of Hox B3 during hematopoietic differentiation indicated that Hox B3 can exhibit distinct effects on different cell populations. For example, whereas overexpression of Hox B3 impaired differentiation of B cells, it also promoted expansion of a subset of granulocyte/macrophage cells ( Savageau et al. 1997). Thus, the Hox genes may exert tissue-specific influences on gene expression. Indeed, our previous work using Hox D3 expressing retrovirus indicated that upregulation of αvβ3 integrin was restricted to endothelial and blood cells, as infected epithelial and fibroblast cells did not express αvβ3 ( Boudreau et al. 1997).

Although compound mutants of the Hox 3 paralogous group have also been generated, the embryonic lethal phenotype makes it impossible to study the contribution of these genes to EC-specific gene expression or angiogenesis in adult organisms ( Manley and Capecchi 1997). Furthermore, as evidence from compound mutants for Hox 9 members indicates, embryonic and adult tissues are differentially effected by the same Hox genes ( Chen and Capecchi 1999) and embryonic and adult angiogenesis may be subject to different modes of regulation ( Bader et al. 1998). These genetic models may not help clarify the role of Hox genes in pathologically induced angiogenesis in differentiated adult EC.

Nonetheless, our findings suggest that the multistep process of neovascularization requires participation from at least two paralogous members of the Hox 3 gene family. Whereas Hox D3 is necessary to initiate vascular sprouting and migration in response to angiogenic stimuli such as bFGF, Hox B3 is subsequently required by sprouting EC to undergo capillary morphogenesis. Whether the combined expression of Hox D3 and Hox B3 will also reveal additional synergistic interactions is currently not known, but is the subject of current investigations in our laboratory.

Acknowledgments

The authors gratefully acknowledge the excellent technical assistance of Irene Cheung. The authors would also like to thank Drs. Peter L. Jones and W.J. Welch for critical reading of the manuscript.

N. Boudreau is recipient of a March of Dimes Basil O'Connor Scholar Award.

References

References
Adams
R.H.
,
Wilkinson
G.A.
,
Weiss
C.
,
Diella
F.
,
Gale
N.W.
,
Deutsch
U.
,
Risau
W.
,
Klein
R.
Roles of ephrin B1 and EphB receptors in cardiovascular developmentdemarcation of arterial venous domains, vascular morphogenesis and sprouting angiogenesis
Genes. Dev
18
1999
2165
2173
Ades
W.
,
Candal
F.J.
,
Swerlick
R.A.
,
George
V.G.
,
Summers
S.
,
Bosse
D.C.
,
Lawley
T.J.
HMEC-1–establishment of an immortalized human microvascular endothelial cell line
J. Invest. Dermatol
99
1992
683
690
[PubMed]
Bader
B.L.
,
Rayburn
H.
,
Crowley
D.
,
Hynes
R.O.
Extensive vasculogenesis, angiogenesis and organogenesis precede lethality in mice lacking all αv integrins
Cell.
95
1998
507
520
[PubMed]
Belotti
D.
,
Clausse
N.
,
Flagiello
D.
,
Alami
Y.
,
Daukandt
M.
,
Deroanne
C.
,
Malfoy
B.
,
Boncinelli
E.
,
Faiella
A.
,
Castronovo
V.
Expression and modulation of homeobox genes from cluster B in endothelial cells
Lab. Invest
78
1998
1291
1299
[PubMed]
Botas
J.
Control of morphogenesis and differentiation by Hom/Hox genes
Curr. Opin. Cell Biol
5
1993
1015
1022
[PubMed]
Boudreau
N.
,
Bissell
M.J.
Extracellular matrix signalingintegration of form and function in normal and malignant cells
Curr. Opin. Cell Biol
10
1998
640
646
[PubMed]
Boudreau
N.
,
Andrews
C.
,
Srebrow
A.
,
Ravanpay
A.
,
Cheresh
D.A.
Hox D3 induces an angiogenic phenotype in endothelial cells
J. Cell Biol
139
1997
257
264
[PubMed]
Bouillet
P.
,
Oulad-Abdelghani
M.
,
Vicaire
S.
,
Garnier
J.-M.
,
Schubaur
B.
,
Dolle
P.
,
Chambon
P.
Efficient cloning of cDNAs of retinoic acid-responsive genes in P19 embryonal carcinoma cells and characterization of a novel mouse gene Stra1 (mouse Lerk2)
Dev. Biol
170
1995
420
433
[PubMed]
Brooks
P.C.
,
Montgomery
A.M.P.
,
Rosenfeld
M.
,
Riesfeld
R.A.
,
Hu
T.
,
Klier
G.
,
Cheresh
D.A.
Integrin αvβ3 antagonists promote tumor regression by inducing apoptosis of angiogenic blood vessels
Cell.
79
1994
1157
1164
[PubMed]
Chen
F.
,
Capecchi
M.R.
Paralogous mouse Hox genes, Hoxa9, Hoxb9, Hoxd9, function together to control development of the mammary gland in response to pregnancy
Proc. Natl. Acad. Sci. USA.
96
1999
541
546
[PubMed]
Chen
J.
,
Ruley
H.E.
An enhancer element in the EphA2 (Eck) gene sufficient for rhombomere-specific expression is activated by Hox A1 and Hox B1 homeobox proteins
J. Biol. Chem
273
1998
24670
24675
[PubMed]
Cillo
C.
,
Faiella
A.
,
Cantile
M.
,
Boncinelli
E.
Homeobox genes and cancer
Exp. Cell Res
248
1999
1
9
[PubMed]
Condie
B.G.
,
Capecchi
M.R.
Mice with targeted disruptions in the paralogous genes hoxa-3 and hoxd-3 reveal synergistic interactions
Nature.
370
1994
304
307
[PubMed]
Daniel
T.O.
,
Stein
E.
,
Cerretti
D.P.
,
St
P.L.
,
John
B.
,
Robert
,
Abrahamson
D.R.
ELK and LERK-2 in developing kidney and microvascular endothelial assembly
Kidney Int. Suppl.
57
1996
S73
S81
[PubMed]
Duboule
D.
Vertebrate hox gene regulationclustering and/or colinearity?
Curr. Opin. Genet. Dev
8
1998
514
518
[PubMed]
Guazzi
S.
,
Lonigro
R.
,
Pintonello
L.
,
Boncinelli
E.
,
Di Lauro
R.
,
Mavillo
F.
The thyroid transcription factor-1 gene is a candidate target for regulation by Hox proteins
EMBO (Eur. Mol. Biol. Organ.) J
13
1994
3339
3347
Hirschi
K.K.
,
D'Amore
P.A.
Pericytes in the microvasculature
Cardiovasc. Res
32
1996
687
698
[PubMed]
Holder
N.
,
Klein
R.
Eph receptors and ephrinseffectors of morphogenesis
Development.
126
1999
2033
2044
[PubMed]
Huynh-Do
U.
,
Stein
E.
,
Lane
A.A.
,
Liu
H.
,
Cerretti
D.P.
,
Daniel
T.O.
Surface densities of ephrin B1 determine EphB-coupled activation of cell attachment through αvβ3 and α5β1 integrins
EMBO (Eur. Mol. Biol. Organ.) J.
18
1999
2165
2173
Jones
F.S.
,
Chalepakis
G.
,
Gruss
P.
,
Edelman
G.M.
Activation of the cytotactin promoter by the homeobox containing gene Evx-1
Proc. Natl. Acad. Sci. USA.
89
1992
2091
2095
[PubMed]
Klemke
R.L.
,
Cai
S.
,
Giannini
A.L.
,
Gallagher
P.J.
,
De Lanerolle
P.
,
Cheresh
D.A.
Regulation of cell motility by mitogen-activated protein kinase
J. Cell Biol.
137
1997
481
492
[PubMed]
Kubota
Y.
,
Kleinman
H.K.
,
Martin
G.R.
,
Lawley
T.J.
Role of laminin and basement membrane in the morphological differentiation of human endothelial cells into capillary-like structures
J. Cell Biol
107
1988
1589
1596
[PubMed]
Langston
A.W.
,
Gudas
L.J.
Retinoic acid and homeobox gene regulation
Curr. Opin. Genet. Dev.
4
1994
550
555
[PubMed]
Li
C.
,
Gudas
L.J.
Murine laminin B1 gene regulation during the retinoic acid and dibutyrl cAMP induced differentiation of embryonic F9 teratocarcinoma stem cells
J. Biol. Chem
271
1996
6810
6818
[PubMed]
Lorentz
O.
,
Duluc
I.
,
De Arcangelis
A.
,
Simon-Assman
P.
,
Kedinger
M.
,
Freund
J.-N.
Key role of the CDX2 homeobox gene in extracellular matrix-mediated intestinal cell differentiation
J. Cell Biol
139
1997
1553
1565
[PubMed]
Maisonpierre
P.C.
,
Suri
C.
,
Jones
P.F.
,
Bartunkova
S.
,
Wiegand
S.J.
,
Radziejewski
C.
,
Compton
D.
,
McClain
J.
,
Aldrich
T.H.
,
Papadopoulos
N.
Angiopoietin-2, a natural antagonist for Tie2 that disrupts in vivo angiogenesis
Science.
277
1997
55
60
[PubMed]
Manley
N.R.
,
Capecchi
M.R.
Hox group 3 paralogous genes act synergistically in the formation of somitic and neural crest-derived structures
Dev. Biol
192
1997
274
288
[PubMed]
Pandey
A.
,
Shao
H.
,
Marks
R.M.
,
Polverini
P.
,
Dixit
V.M.
Role of B61, the ligand for the Eck receptor tyrosine kinase in TNFα-induced angiogenesis
Science.
268
1995
567
569
[PubMed]
Patel
C.V.
,
Sharangpani
R.
,
Bandyopadhyay
S.
,
DiCorleto
P.E.
Endothelial cells express a novel, tumor necrosis factor–alpha-regulated variant of HOXA9
J. Biol. Chem
274
1999
1415
1422
[PubMed]
Savageau
G.
,
Thorsteindottir
U.
,
Hough
M.R.
,
Hugo
P.
,
Lawrence
H.J.
,
Largman
C.
,
Humphries
R.K.
Overexpression of Hox B3 in hematopoietic cells causes defective lymphoid development and progressive myeloproliferation
Immunity
6
1997
13
22
[PubMed]
Stelnicki
E.J.
,
Arbeit
J.
,
Cass
D.L.
,
Saner
C.
,
Harrison
M.
,
Largman
C.
Modulation of the human homebox genes PRX-2 and HOXB13 in scarless fetal wounds
J. Invest. Dermatol
111
1998
57
63
[PubMed]
Stoker
A.W.
,
Bissell
M.J.
Development of avian sarcoma and leukosis virus-based vector packaging cell lines
J. Virol
62
1988
1008
1015
[PubMed]
Stromblad
S.
,
Cheresh
D.A.
Cell adhesion and angiogenesis
Trends Cell Biol
6
1996
462
468
[PubMed]
Studer
M.
,
Gavalas
A.
,
Marshall
H.
,
Ariza-McNaughton
L.
,
Rijli
F.M.
,
Chambon
P.
,
Krumlauf
R.
Genetic interactions between Hox A1 and Hox B1 reveal new roles in regulation of early hindbrain patterning
Development.
125
1998
1025
1036
[PubMed]
Wang
H.U.
,
Anderson
D.J.
Eph family transmembrane ligands can mediate repulsive guidance of trunk neural crest migration and motor axon outgrowth
Neuron.
18
1997
383
396
[PubMed]
Werb
Z.
,
Vu
T.H.
,
Rinkenberger
J.L.
,
Coussens
L.M.
Matrix-degrading proteases and angiogenesis during development and tumor formation
APMIS.
107
1999
11
18
[PubMed]
Xu
Y.
,
Swerlick
R.A.
,
Sepp
N.
,
Bosse
D.
,
Ades
E.A.
,
Lawley
T.J.
Characterization of expression and modulation of cell adhesion molecules on an immortalized dermal microvascular endothelial cell line (HMEC-1)
J. Invest. Dermatol
102
1994
833
837
[PubMed]

Abbreviations used in this paper: bFGF, basic fibroblast growth factor; BM, basement membrane; CAM, chick chorioallantoic membrane; EC, endothelial cells; ECM, extracellular matrix; HMEC, human microvascular endothelial cells; HMEC-1, human dermal microvascular endothelial cells; Hox, Homeobox; RT, reverse transcriptase; uPA, urokinase plasminogen activator.