|Home | About | Journals | Submit | Contact Us | Français|
Normal vascular development and angiogenesis is regulated by coordinated changes in cell-cell and cell-extracellular matrix (ECM) interactions. The Homeobox (Hox) family of transcription factors coordinately regulate expression of matrix degrading proteinases, integrins and ECM components and profoundly impact vascular remodeling. Whereas HoxA5 is downregulated in active angiogenic endothelial cells (EC), sustained expression of HoxA5 induces TSP-2 and blocks angiogenesis. Since HoxA5 is also lacking in EC in proliferating hemangiomas, we investigated whether restoring expression of HoxA5 could normalize hemangioma cell morphology and/or behavior. Sustained expression of HoxA5 in the murine hemangioma cell line (EOMA) reduced their growth in vivo and promoted branching morphogenesis in 3D BM cultures. Moreover, restoring HoxA5 expression increased the retention of β-catenin in adherens junctions and reduced permeability. In addition we also show that the HoxA5 mediated increase in stability of adherens junctions requires Akt1 activity and introduction of constitutively active myr-Akt in EOMA cells also increased retention of β-catenin in adherens junctions. Finally we show that HoxA5 increases Akt1 mRNA, protein expression and further enhances Akt activity via a coordinate down regulation of PTEN. Together these results demonstrate a central role for HoxA5 in coordinating a stable vascular phenotype.
Angiogenesis, the outgrowth of new capillaries from preexisting vessels, is essential for normal tissue repair as well as growth of solid tumors.1–3 To generate capillary sprouts, existing capillary endothelial cells (EC) proliferate, migrate and invade the host stroma toward the source of angiogenic stimuli. Angiogenic factors induce quiescent EC to re-enter the cell cycle, express extracellular matrix degrading proteinases and upregulate expression of adhesion molecules to allow migration and proliferation. Subsequently, vascular sprouts then resynthesize basement membranes (BM), undergo capillary morphogenesis and withdraw from the cell cycle and form mature quiescent vessels.4
Angiogenesis is a complex, multi-step process that involves temporally and spatially regulated changes in gene expression, and therefore it is not surprising that the Homeobox (Hox) master transcriptional regulators5 have been shown to play a role in coordinating these changes in EC both in vitro and in vivo.6–11 We previously identified several class I Hox genes expressed in cultured EC, which can act to enhance or inhibit the angiogenesis process. For example, HoxA3 increased endothelial cell migration by upregulation of matrix metalloproteinase-14 (MMP-14) and urokinase-type plasminogen activator receptor (uPAR) in vitro and in vivo,9 whereas HoxD3 increases expression of the angiogenic β3 and a5 integrins.6,7 The paralogous, HoxB3, has also been shown to promote angiogenesis by increasing expression of ephrin A1 (Efna1) and facilitates cell-cell interactions necessaries for capillary morphogenesis.10 Conversely, sustained expression of HoxD10 impaired EC migration and blocked angiogenesis by upregulation of tissue inhibitor of metalloproteinase-1 (TIMP-1) while simultaneously reducing the levels of uPAR and MMP-14, respectively.11
More recently we showed that another Hox gene, HoxA5 could also block angiogenesis.12 HoxA5 impaired endothelial cell migration in response to vascular endothelial growth factor (VEGF), and blocked angiogenesis in vivo. The anti-angiogenic activity of HoxA5 was associated with upregulation of the angiogenic inhibitor, thrombospondin 2 (TSP2)13–16 as well as a down regulation of Efna1.10,17
While normal angiogenesis is regulated by a balance between pro- and anti-angiogenic Hox genes, accumulating evidence suggests that inappropriate Hox expression may also contribute to aberrant vascular morphology including benign hemangiomas and brain arterial venous malformations. We previously showed that sustained expression of HoxD3 in the embryonic chick vasculature produced hemorrhagic lesions resembling vascular tumors.6 We also observed that high levels of HoxD3 were present in brain arterial venous malformations and in infantile hemangiomas.12,18 In contrast, expression of HoxA5, normally expressed in quiescent vessels, was conspicuously absent from infantile hemangiomas suggesting that loss of Hox genes expressed in the quiescent vasculature may also contribute to this phenotype.12
We therefore investigated whether reexpressing HoxA5 in endothelial cells derived from benign hemangiomas would restore normal vascular morphology and promote acquisition of a quiescent differentiated phenotype. We infected the murine hemangioma cell line (EOMA) with a retroviral vector expressing HoxA5 and evaluated growth and morphology relative to hemangioma cells lacking HoxA5 and normal endothelial cells which express endogenous HoxA5. We present evidence that HoxA5 plays an essential role in promoting and maintaining a stabile quiescent phenotype in endothelial cells.
The murine hemangioma cell line (EOMA), CRL-2586™, was obtained from the American Type Culture Collection (ATCC, Manassas, VA). EOMA cells were maintained in Dulbecco's Modified Eagle's Medium (DMEM) (Life Technologies, Rockville, MD) with 10% Fetal Bovine Serum (FBS) and 50 µg/ml gentamicin sulfate (Life Technologies). Immortalized human dermal microvascular endothelial cell line, HMEC-1,19 were cultured in MCDB131 media (Life Technologies), and fortified with 10% FBS, 2mM glutamine (Life Technologies) and 50 µg/ml gentamicin.
Three-dimensional (3D) cultures were prepared using a modified procedure (described in ref. 20) using 3D BM (Matrigel). Cultures were analyzed after 24 or 72 hours of cultivation. Morphology of the 3D cultures was assessed examining the degree of colony organization visually by phase contrast microscopy.
When indicated, four hours before release from the 3D BM, cells were preincubated at 37°C with 30 µM of LY294002 or 1 µM of MG-132 (Calbiochem, EMD Bioscience Inc., San Diego, CA) or DMSO as a control (Sigma Chemical CO, St. Louis, MO). After preincubation at 24 or 72 hours, cells were release as described below.
The pBabe HA-HoxA5 retroviral vector was constructed as described in reference 12. The myr-Akt pWZL retroviral construct myr a-129 was provided by Mina Bissell (Lawrence Berkeley National Laboratory, Berkeley, CA).
Retroviral vectors were transfected into the amphotropic packaging cell line Phoenix Ampho (ATCC) using the calcium/phosphate-DNA precipitation method, and 48 hours after transfection, viral supernatant was collected, passed through 0.45 µm filters and used for transduction of EOMA cells.
EOMA cells were transduced with control plasmid (pBabe), and pBabe HA-A5 using 2 ml of virus-containing media in the presence of 8 µg/ml of polybrene (Sigma) for 16 hours at 37°C. Twenty-four hours later cells were selected in 1 µg/ml puromycin (Sigma) for seven days, and the pooled cell population was used for subsequent experiments. EOMA cells transduced with pWZL and myr-Akt pWZL virus were selected in 60 µg/ml hygromycin (Sigma) for two weeks.
Cultures grown in 3D were isolated from 3DBM using a modified procedure (described in Weaver et al. ref. 20). Briefly, cells were washed with cold PBS without Ca2+ and Mg2+ and containing 5 mM EDTA, and scraped into a centrifuge tube with a minimum volume of 30 ml cold PBS/EDTA. Cells were incubated on ice for 45 min, until the Matrigel dissociated, and centrifuged at 115 g for 2 min. The pellet was resuspended in an appropriate amount of lysis buffer for mRNA or protein isolation. Soluble and insoluble subcellular fractions were obtained as described (ref. 21).
One microgram of RNA from cell lysates, extracted using the RNeasy isolation kit (Qiagen,) was reverse transcribed using MMuLV RT (Invitrogen) for one hour at 42°C in a total volume of 25 µl. 1/25 of this reaction was linearly amplified for 30 cycles (HoxA5 and TSP2), 27 cycles (Efna1) and 25 cycles (β-catenin and Akt), following denaturation (30 sec 95°C), annealing (30 sec at 55°C for Akt, 56°C for HoxA5, 58°C for TSP2 and β-catenin), and finally extension (30 sec at 72°C) in a thermal cycler (PTC-200 Peltier Thermal cycler, MJ Research). RNA was normalized using 18s internal standards at a 1:3 ratio (Ambion, Austin, TX). The expected products were visualized by electrophoresis on 1% agarose gels containing ethidium bromide.
The following primers were used:
The delivery of the siRNA into the HMEC-1 cells was done with a Nucleofector device and its corresponding kits (Amaxa, Inc. Cologne, Germany). Transfection protocols were performed following the manufacturer's instructions using the T23 program. The annealed predesigned siRNA against HoxA5 was purchased from Ambion (Austin, TX) and has the following sequence 5′gcugcacauaagucaugactt3′. As a control we used a siRNA against pMAXGFP (Amaxa, Inc). After the transfection, cells were plated into six well plates coated with 3D Matrigel as described above.
Cells were cultured in 3D Matrigel for 24 or 72 hours, and colonies were isolated using ice-cold PBS/EDTA as described (ref. 22). A small aliquot of pellets were smeared gently across a small area of a glass slide and were fixed in methanol: acetone (1:1) at room temperature for ten minutes. Immunofluorecence staining was performed as described (ref. 22). Antibodies used were β-catenin (1:1000) (BD Biosciences). The slides were mounted in fluoromount (Southern Biotechnology Associates, Inc. Birmingham, AL) and analyzed under a Nikon Eclipse TE300 fluorescence microscope and under a Nikon D-Eclipse C1 confocal microscope.
Total protein or soluble and insoluble fractions were isolated from lysates of HoxA5 or control-transfected EOMA cells at 24 or 72 hours after culturing on 3D Matrigel. Equal amounts of protein were separated on a 10% SDS-PAGE gels and transferred to PVDF membranes. After blocking for nonspecific binding, western-blots were probed with a monoclonal β-catenin antibody (1:1000, clone 14 BD Biosciences), monoclonal PTEN antibody (1:250) (BD Biosciences, San Jose, CA) or polyclonal phosphor-Akt (Ser473) (1:1000) (Cell Signaling Technology, Inc. Beverly, MA), followed by sheep anti-mouse-HRP at 1:2000 or donkey anti-rabbit-HRP at 1:5000. Excess of antibody was removed by extensive washing and blots were developed by ECL system (Amersham Biosciences, Piscataway, NJ). The membranes were then stripped and treated with polyclonal Akt antibody (1:1000) (Cell Signaling Technology, Inc. Beverly, MA), and polyclonal β-actin antibody (1:1000) (Abcam Inc. Cambridge, MA), followed by donkey anti-rabbit-HRP at 1:5000 and detected by ECL system.
Control or HoxA5 expressing EOMA cells were cultured for 72 hours and solubilized in ice-cold lysed buffer.23 Following removal of insoluble material by centrifugation at 14000 g for one min, at 4°C, aliquots of the lysates containing equal amounts of protein (500 µg) were precleared on Untralink Protein A/G beads (Pierce) for one hour at 4°C. Supernatant was incubated one hour with 2 µg of anti VE-cadherin antibody (C-19, Santa Cruz). Immune complexes were captured with 10 µl of Ultralink Protein A/G beads at 4°C for 1.5 hours under continuous mixing. This was followed by four washings with ice-cold PBS containing 300 µM sodium orthovanadate. Immunoprecipitates were recovered in Laemmli's buffer under reducing conditions, subjected to SDS-PAGE and transferred on to nitrocellulose membranes. Immunoblotting was carried out using antibodies against VE-cadherin and membranes were stripped and probed with antibodies against β-catenin or monoclonal phosphor-tyrosine (1:2000), or blotted with a polyclonal phospho-(Ser/Thr) Akt substrate (1:1000) antibody.
To measure transcriptional activity of β-catenin, control or HoxA5 expressing EOMA cells were grown to 70% confluence in six well dishes. Transfections were carried out in DMEM 10%FBS using the Effectene® transfection reagent (Qiagen) according to the manufacturer's instructions with a 1/25 ratio of DNA/ Effectene reagent. Cells were incubated with 0.4 µg of pTOPFLASH (reporter plasmid containing two sets TCF binding site, thymidine kinase minimal promoter and luciferase open reading frame) or 0.4 µg of pFOPFLASH (containing TCF mutant sites as a negative control) (both were a gift from Dr. Clevers). Luciferase assays were performed using the Dual-Luciferase reporter assay system (Promega, Madison, WI, USA) and their activities in the cell extracts were measured using a Wallac Victor2 1420 Multilabel Counter luminometer (PerkinElmer Life Sciences, Boston, MA). The transcriptional activity of each construct was evaluated by cotransfection with CMV- βgal to normalize transfection efficiency. All transfections were done on plastic and in some cases, after 24 hours cells were transferred and cultured on 3D Matrigel. Transcriptional activity was measured at 72 hours after transfection. Results were expressed as means ± s.d. of four to six independent experiments performed in triplicate.
Permeability across the endothelial cell monolayer was measured by using type I collagen-coated transwell units (6.5 mm diameter, 8 µm-pore-size PET filter, BD Falcon, NJ). HMEC-1 or HMEC-1 A5 expressing cells plated at 2 x 105 cells in each well were cultured for two days before the experiments and serum starved in Fibroblast Basal Media (FBM) (Lonza, MD) containing 0.5% BSA ON. Permeability was measured by adding 1 mg of FITC-labeled dextran (molecular weight 70,000)/ml in FBM with 0.5% BSA to the upper chamber and 500 µl FBM to the lower chamber. After incubation for different periods of time, between 30 min and three hrs, 50 µl of the sample from the lower compartment was diluted with 50 µl of PBS and measured for fluorescence at 535 nm when excited at 485 nm with a Wallac Victor Multilabel Counter.
A total of 2.5 x 106 of control or HoxA5 expressing EOMA cells were prepared in DMEM media and loaded into a 1 ml tuberculin syringe. Cells were injected subcutaneously in the dorsal midline region of nude mice (n = 6 mice/group) (The Jackson Laboratory, Bar Harbor, ME). All mice were euthanized at seven days post-injection and tumor tissue harvested by snap-freezing in liquid nitrogen.
In all experiments statistical analysis was performed using Student's t distribution with significance reported when p < 0.05.
In previous studies we observed that the anti-angiogenic HoxA5 gene was expressed in resting quiescent vessels, but expression of HoxA5 was markedly reduced in EC in proliferative hemangiomas.12 We evaluated HoxA5 expression in hemangioma derived endothelial cell lines. In contrast to our previous studies showing that HoxA5 was expressed in both primary and nontransformed endothelial cell lines (HMEC-1), neither of the murine hemangioma cell lines (bEND-1 and EOMA) expressed HoxA5 mRNA (Fig. 1A). We restored HoxA5 expression in EOMA or bEND cells with a retrovirus expressing an HA-epitope tagged HoxA5 or control (empty) viral vector and observed strong expression of HoxA5 (Fig. 1B). We also confirmed that restoring HoxA5 increases expression of TSP-2 mRNA and TSP-2 protein in EOMA cells (Fig. 1C) as well as in bEND cells (not shown).
We predicted that restoring HoxA5 would also block the growth of endotheliomas in vivo. We subcutaneously inoculated nude mice with 2.5 x 106 EOMA (control) or HoxA5 expressing EOMA cells. After seven days, large hemangiomas were observed in mice receiving control transfected EOMA cells, whereas tumors in mice receiving HoxA5 expressing EOMA cells were visibly smaller (Fig. 2A and B). Tumor volume in mice receiving HoxA5 expressing EOMA cells was significantly reduced (20 ± 5 mm2) compared with mice which received control transfected EOMA cells (160 ± 10 mm2) (mean ± sd, n = 6 /group, p < 0.001) (Fig. 2C). A similar reduction in endothelioma tumor volume was observed following engraftment of bEND cells expressing retrovirally transduced HoxA5 (not shown). RT-PCR analysis of RNA extracted from the mice hemangioma tissue after seven days showed an increase in expression of both HoxA5 and TSP-2 mRNA in mice which received HoxA5 expressing EOMA cells compared to control EOMA (Fig. 2D).
Previous studies showed that ectopic expression of the anti-angiogenic TSP-1 gene could restore the ability of endothelioma cells to undergo capillary network formation when cultured on 3D BM.24 We therefore examined whether restoring expression of HoxA5 and consequently, expression of the highly homologous TSP-2 gene would also promote branching and tubule formation by EOMA cells. We observed that in contrast to control EOMA cells which form cyst-like structures when cultured on 3D BM (Fig. 3A), EOMA cells expressing HoxA5 and TSP-2 formed branching structures which partially recapitulated the branching networks formed by nontransformed HMEC-1 that express endogenous HoxA5 (Fig. 3B and C).
To determine whether endogenous HoxA5 expressed in HMEC-1 was required for normal capillary tubule formation, we treated HMEC-1 with siRNA against HoxA5. We observed that transfection with siRNA against HoxA5 resulted in an approximately 80% reduction in endogenous HoxA5 and a corresponding decrease in TSP-2 (Fig. 3D). The reduction in HoxA5 and TSP-2 was accompanied by a marked reduction in both the number and length of capillary-like branches in HMEC-1 cells cultured on 3D BM (Fig. 3E and F).
To begin to evaluate how HoxA5 contributes to the branching morphology of endothelial cells, we first examined the distribution of β-catenin, as previous studies also showed that, compared to normal endothelial cells, EOMA cells also show a relative increase in nuclear, transcriptionally active β-catenin.25 We performed immunofluorescent analysis for β-catenin localization and observed that, in contrast to control EOMA cells, the majority of β-catenin in HoxA5 expressing EOMA cells was localized to junctional regions with a cortical membrane distribution (Fig. 4A and B). whereas in control EOMA cells, only a modest fraction of β-catenin was observed at cell-cell borders and the majority was found in nuclear and peri-nuclear regions. Western blot analysis of Triton X-100 soluble and insoluble fractions of cells, also confirmed that in control transfected EOMA cells, the majority of β-catenin was localized to the soluble, cytoplasmic and nuclear pools. In HoxA5 expressing EOMA cells however, the majority of β-catenin was present in the insoluble pellet containing cytoskeletal and junctional complexes. This distribution was similar to that found in normal human microvascular endothelial cells (HMEC-1) which express endogenous HoxA5 (Fig. 4C).
Further evidence that β-catenin in HoxA5 expressing EOMA cells was directly associated with adherens junctions was obtained by coimmunoprecipitation with VE-cadherin. We observed that increased amounts of β-catenin were found in association with VE-Cadherin in HoxA5 expressing EOMA cells compared to control EOMA cells (Fig. 4D).
To demonstrate that the HoxA5 mediated redistribution of β-catenin resulted in reduced transcriptional activity, we also compared β-catenin dependent transcriptional activity in control or HoxA5 expressing EOMA cells using the TOPFLASH vector which contains multiple β-catenin consensus sites linked to a luciferase reporter. Despite an increase in the overall levels of β-catenin protein in HoxA5 expressing EOMA cells (not shown), we observed a significant decrease in β-catenin dependent transcriptional activity compared to control EOMA cells (Fig. 4E), consistent with reduced nuclear β-catenin. Sustained expression of HoxA5 cells also decreased the permeability of endothelial cells to FITC labeled Dextran (Fig. 4F). Thus, restoring HoxA5 expression in EOMA cells increases retention of β-catenin in adherens junctions and stabilizes adherens junctions leading to reduced permeability.
A recent study showed that increased vascular stability and maturation and a corresponding increase in TSP-2 expression was dependent on Akt1 activity.26 To determine whether the HoxA5 mediated increase in β-catenin retention in adherens junctions also required Akt activity, we treated HoxA5 expressing EOMA cells with LY294002, an inhibitor of PI3Kinase and Akt activity. As expected, addition of LY294002 reduced levels of phospho-Akt in EOMA cells expressing HoxA5 cells but did not influence total Akt protein (Fig. 5A). Western blots revealed that addition of LY294002 selectively reduced the amount of β-catenin retained in the junctions (insoluble) fractions in HoxA5 transduced EOMA cells (Fig. 5B upper panel). Inhibition of Akt also reduced the amount of β-catenin retained in the junctional fraction of normal HMEC-1 which express endogenous HoxA5 (Fig. 5B lower panel).
In the absence of Akt signaling, β-catenin released from junctions is rapidly degraded by the ubiquitin pathway.27,28 We therefore treated HoxA5 expressing cells with the proteosome inhibitor MG132 and observed that the pool of β-catenin released from the junctions upon addition of LY294002 was stabilized (Fig. 5C). The increased pool of released β-catenin was reflected by an increase in β-catenin transcriptional activity which was now comparable to control transfected EOMA cells (Fig. 5D). Moreover, we also observed that HoxA5 reduced the relative degree of tyrosine phosphorylation of β-catenin, which would otherwise promote its dissociation from junctional complexes29 (Fig. 5E). Addition of LY294002 however, reversed this effect and tyrosine phosphorylation of junctional β-catenin in HoxA5 cells returned to levels seen in control EOMA cells (Fig. 5E).
As LY294002 inhibits both PI3K and AKT, we evaluated whether constitutively active Akt (myr-Akt) could phenocopy HoxA5 and stabilize endothelial cells by promoting β-catenin retention in junctions. EOMA cells were transfected with a constitutively active Akt (myr-Akt) retroviral vector. Sustained expression of AKT in EOMA cells (Fig. 6A) lead to increased amounts of β catenin present in the insoluble fraction (Fig. 6B and C) compared to control transfected EOMA cells. Subsequent immunofluorescent staining for β-catenin showed that in contrast to control EOMA cells where β-catenin was distributed throughout the cytoplasm, sustained Akt activity allowed β-catenin to be localized predominantly at the peripheral cell borders (Fig. 6D). Thus enhanced Akt activity was sufficient to mimic the effects of restoring Hox A5 expression in EOMA cells.
As Akt is required for and recapitulates the HoxA5 mediated effects on endothelial junctions, we investigated whether Akt activity was increased by HoxA5 in EOMA cells. RT-PCR analysis revealed an approximately two fold increase in mRNA levels of Akt1 in HoxA5 expressing EOMA cells compared to control EOMA cells (Fig. 7A). No change in Akt2 mRNA was detected (data not shown). Moreover, the increased Akt1 mRNA levels were also reflected by a similar 2-fold increase in Akt protein (Fig. 7B). Interestingly however, an even more striking increase in phosphorylated, active Akt was noted in HoxA5 expressing EOMA cells as compared to control. (Fig. 7B). We subsequently performed western blots to examine expression of the phosphatase, PTEN which dephosphorylates Akt and limits its activity. PTEN expression was markedly reduced in HoxA5 expressing cells (Fig. 7C). To further demonstrate the functional consequences of HoxA5 induced Akt activity, we examined the cellular distribution of Foxo1a, a transcription factor associated with vascular activation and whose activity is blocked via Akt mediated retention in the cytoplasm.30–32 Western blots of nuclear and cytoplasmic extracts revealed that in control EOMA cells, Foxo1a is equally distributed in the nucleus and cytoplasm (Fig. 7D). In HoxA5 expressing EOMA cells however, the majority of Foxo1 is retained in the cytoplasm, consistent with high Akt activity and vascular stability.
Our previous studies showed that although HoxA5 is normally downregulated in angiogenic environments, sustaining expression of HoxA5 could block neovascularization and HoxA5 increased expression of the anti-angiogenic, matricellular protein TSP-2.12 In the current study we have extended these findings to show that restoring HoxA5 expression, which is lacking in proliferating hemangiomas, blocks endothelioma growth in vivo and can partially restore a normal branching phenotype in endothelioma cells. Moreover, endogenous HoxA5 expression in normal microvascular endothelial cells is required for formation and maintenance of capillary networks. The morphological changes induced by HoxA5 are accompanied by increased retention of β-catenin in adherens junctions and reduction in β-catenin transcriptional activity and the impact of HoxA5 on junctional integrity arise via increased AKT expression and activity. Together out results indicate that HoxA5 contributes to normal vascular patterning and stabilization of a differentiated phenotype.
In EC, adherens junctions are comprised of a specific cadherin namely VE-cadherin, which is complexed to cytoplasmic proteins including β- and α-catenin and p120 which are also linked to the cytoskeleton.27,33–35 During angiogenesis, adherens junctions must first be disassembled and then subsequently reassembled during maturation of vascular sprouts.33,36,37 A critical role for adherens junctions in maintaining vascular stability is underscored by studies showing that interfering with VE-Cadherin function leads to loss of vascular integrity in the adult and loss of VE-Cadherin during development results in lethality of the embryo at day 9.5 due to impairment of vascular remodeling and maturation.36,37 Moreover, levels of VE-Cadherin are reduced in hemorrhagic vascular tumors underscoring its role in stabilizing the vasculature and guiding normal vascular morphogenesis.38 In addition, retention of β-catenin in these complexes is essential for vascular stability as targeted inactivation of β-catenin gene in ECs interferes with the cell-cell contacts, disrupts vascular patterning and increases vascular fragility, and essentially mimics similar to loss or disruption of VE-Cadherin.39 β-catenin which is not complexed in adherens junctions can translocate to the nucleus and act as a transcriptional coactivator for TCF/LEF family of transcription factors and can induce expression of genes associated with invasion and growth including MMPs and cyclinD1.40,41 Not surprisingly EOMA cells were found to exhibit increased nuclear localization of β-catenin compared to normal endothelial cells.25 However, by restoring HoxA5 expression in EOMA cells we could reduce the transcriptional activity of β-catenin by increasing the retention of β-catenin in adherens junctions to levels which mimicked normal differentiated endothelial cells. Moreover, constitutive expression of HoxA5 reduced permeability of endothelial cells, consistent with an increase in stable cell-cell junctions.
A number of studies have been performed using HoxA5 null mice.42–46 In each of these studies it was noted that branching in tissues including the breast, gastrointestinal tract and the lung were all impaired by the loss of HoxA5. Furthermore, loss of normal tissue morphology during tumorigenesis has been linked to changes in adherens junctions and increased cytoplasmic and nuclear localization of β-catenin.20,34–35 Indeed earlier studies also showed that many breast tumor cells lose expression of HoxA5, however, the impact on β-catenin or adherens junctions and tissue morphology was not examined.12,47 Our current findings that HoxA5 can promote vascular branching and stability are consistent with a role for HoxA5 in directing and establishing differentiated tissue architecture and our observations that HoxA5 increases retention of β-catenin in adherens junctions might suggest a common mechanism by which HoxA5 acts to maintain differentiated tissue function.
Our current results also show that the HoxA5 dependent increase in retention of β-catenin in junctions require Akt1 activity and can be mimicked by introduction of a constiutively active Akt. These findings are consistent with a recent report showing that loss of Akt resulted in increased vascular permeability and reduced maturation of vessels.26 Moreover additional studies indicate that sustained endothelial expression of Akt following arterial ligation reduced lesion formation and improved endothelial integrity and barrier function.48 Although previous studies have suggested that Akt can induce eNOS and NO expression, NO production alone could not completely account for the protective effects of Akt on the vasculature.48 Our findings that HoxA5 mediated induction of Akt1 leads to increased association of β-catenin in the adherens junctions may provide additional mechanistic insight as to how Akt might act to stabilize vessels.
Akt has also been linked to increased vascular stability through its central role in Angiopoietin (ANG)-1 mediated signaling. Ang-1, which binds to the endothelial specific receptor TIE-2, induces Akt activity49–51 which in turn phosphorylates and inactivates the transcription factor FHKR (FOXO1), by preventing its translocation to the nucleus.31,32 Constitutive Akt expression mimicked inhibition of FKHR and was accompanied by reduced expression of genes, which contribute to vascular destabilization including the antagonist ANG-2.32 We also observed that HoxA5 and the subsequent increase in Akt activity was reflected by increased retention of FOXO1 in the cytoplasmic compartment of cells, again consistent with promoting a more stabile vascular phenotype.
It should be noted however that strong evidence also supports a role for Akt1 in destabilizing vessels and promoting angiogenesis and vascular leakage in vivo.52–54 Studies by Ackah et al. using Akt1 null mice showed that Akt1 was required for VEGF induced vascular leakage.52 In genetic models in which constitutively active Akt1 was selectively expressed in endothelial cells, showed that sustained Akt1 activity lead to development of vascular malformations similar to microaneurysms, increased vessel size and induced chronic vessel permeability, which could be reversed when expression of the transgene was blocked.53,54 Moreover, Akt was found to be necessary for tumor induced angiogenesis, as addition of the mTOR inhibitor, rapamycin block angiogenesis in Akt transgenic mice. These studies are also supported by a number of other studies showing that Akt is also essential for angiogenesis and endothelial growth and survival through inhibition of caspases.49,55–58
Taken together this data might suggest that while increasing expression of Akt1 is necessary for endothelial survival and vascular remodeling during angiogenesis, baseline levels of Akt1 activity are also necessary for maintenance of vascular integrity and a quiescent, differentiated phenotype A recent report showed that Akt activity was in fact essential for differentiation of keratinocytes and Akt was also localized to adherens junctions in the differentiated cells.59 Importantly, despite the large body of evidence implicating Akt in promoting tumor growth, two recent reports showed Akt could block breast tumor invasion and metastasis by targeting NFAT or TSC2 for degradation.60,61 Thus the relative levels of Akt1 and the cellular microenvironment act to define a context-dependent role for this signaling pathway in cell growth, migration or differentiation.
Although it is well established that Akt can also prevent degradation of cytoplasmic β-catenin via phosphorylation and inhibition of GSK3β28,30,41,62 the stabilized pool of cytoplasmic β-catenin is not reincorporated into junctions, but instead translocates to the nucleus to activate transcription. Indeed, we observed that while inhibiting Akt allows degradation of free β-catenin in EOMA cells, restoring HoxA5 and Akt activity primarily acts to stabilize the association of β-catenin in adherens junctions and ultimately reduces the amount of free β-catenin and transcriptional activity.
Assembly of junctions is initiated by the association of β-catenin and cadherins in the endoplasmic recticulum and this association can be maintained until β-catenin is tyrosine phosporylated to promote its dissociation from this complex.29,63 We showed that reexpressing HoxA5 and the subsequent increase in Akt activity resulted in a reduction of tyrosine phosphorylation of β-catenin in adherens junctions. Morevoer, since Akt is a serine/threonine kinase, it would not directly influence β-catenin tyrosine phosphorylation, and neither β-catenin or cadherin contain an Akt consensus motif RxRxxS/T F/L.30,64 p120 however does contain an Akt phosphorylation consensus site spanning residues 315–32065,66 and recent studies have also indicated that serine phosphorylation can also stabilize that association of cadherin and β-catenin.65–67 We did observe an increase in an Akt phosphorylated substrate of approximately 120 Kd which coimmunoprecipitated with VE-Cadherin (N. Boudreau, unpublished observations), however the functional significance of this protein was not established in our current studies.
Nonetheless, in the present study we show that by restoring HoxA5 in hemangioma cells, we can activate a program of coordinate changes in gene expression culminating in reversion of the aberrant vessels to a more stabile and differentiated vascular phenotype. Moreover, since HoxA5 is reduced in angiogenic tumor vessels, ectopic HoxA5 may also provide a means to ‘normalize’ the hyper-permeable tumor vasculature to improve delivery of therapeutics.
This work was supported by funds from the NIH (NIH CA85249) and (PO1 NS44155) to Nancy Boudreau and Ministerio de Educación y Ciencia, Spain (EX2004-1181) for Gemma Arderiu.
Previously published online as a Cell Adhesion & Migration E-publication: http://www.landesbioscience.com/journals/celladhesion/article/5448