|Home | About | Journals | Submit | Contact Us | Français|
Recently, we reported that the forkhead transcription factor, FKHR/FOXO1, is required for vascular endothelial growth factor (VEGF)-mediated upregulation of a number of genes in endothelial cells. Here, we tested the hypothesis that hepatocyte growth factor (HGF), a potent activator of PI3K-Akt in endothelial cells, is capable of depleting the nucleus of FKHR/FOXO1 and thus inhibiting VEGF induction of this class of genes.
Incubation of human coronary artery endothelial cells with HGF induced prolonged PI3K/Akt-dependent phosphorylation and nuclear exclusion of FKHR/FOXO1. HGF-mediated inhibition of FKHR/FOXO1 activity resulted in secondary attenuation of VEGF-induced expression of FKHR/FOXO1-dependent genes including vascular cell adhesion molecule-1, manganese superoxide dismutase, endothelial specific molecule-1, CBP/p300 interacting transactivator with ED-rich tail-2, bone morphogenetic protein-2, matrix metalloproteinase-10 and MGC5618. At a functional level, pre-incubation of HGF resulted in inhibition of VEGF-induced VCAM-1-mediated monocyte adhesion to endothelial cells. HGF-mediated inhibition of VEGF-inducible VCAM-1 expression and monocyte adhesion was reversed by overexpression of constitutively active phosphorylation-resistant triple mutant (TM)-FKHR.
These findings suggest that physiological agonists of PI3K-Akt signaling pathway may modulate VEGF-FKHR/FOXO1-dependent gene expression in endothelial cells. The data underscore the importance of the “set point” of the endothelial cell when considering mechanisms of signal transduction.
Vascular endothelial growth factor (VEGF) plays a critical role in endothelial cell survival, proliferation, migration, and is involved in wound repair, angiogenesis of ischemic tissue, tumor growth, microvascular permeability, vascular protection, and hemostasis 1–8. VEGF has been shown to activate a number of different intracellular signaling pathways, including PKC, PI3K and Akt, MEK1/2, p38 MAPK, and phospholipase Cγ 9–13. VEGF-mediated activation of signaling intermediates, in turn, results in altered activity of transcription factors, including NF-κB, Egr-1, NFAT-1, Ets-1, Stat-3/5, and forkhead transcription factors 14–20.
Hepatocyte growth factor (HGF) (also known as scatter factor) is a high-molecular weight multifunctional polypeptide growth factor. Although HGF originally was described for its ability to stimulate proliferation of liver cells 21, it is now recognized to interact with its receptor, c-MET, in other cell types. For example, in endothelial cells HGF has been shown to promote migration, proliferation, survival and barrier function 22–25. Under in vivo conditions, HGF has been implicated in angiogenesis 23. In cultured endothelial cells, HGF stimulates several signaling pathways and transcription factors including Rac, ERK1/2, p38 MAPK, PI3K, Src, AKT, FKHR/FOXO1, CREB, and ATF 25–28. However, in contrast to VEGF, HGF has not been shown to activate NF-κB 26, 29.
The mammalian members of the winged helix, or forkhead, transcription factors include FKHR (FOXO1), FKHRL1 (FOXO3a), and AFX (FOXO4) 19. Previous studies have demonstrated the presence of FKHR/FOXO1 and AFX in endothelial cells 19. Exposure of endothelial cells to several agonists, including VEGF, angiopoietin or angiotensin II resulted in PI3K-Akt-mediated phosphorylation and nuclear exclusion of FKHR/FOXO1, and subsequent downregulation of pro-apoptotic and anti-proliferative FKHR/FOXO1 target genes, such as GADD45A and p27kip1 19, 30, 31. More recently, we demonstrated that VEGF requires FKHR/FOXO1 activity for the induction of certain genes, including vascular cell adhesion molecule (VCAM)-1, manganese superoxide dismutase (MnSOD), endothelial specific molecule (ESM)-1, CBP/p300 interacting transactivator with ED-rich tail (CITED)-2, bone morphogenetic protein (BMP)-2, matrix metalloproteinase (MMP)-10 and MGC5618 20. These VEGF-inducible, forkhead-responsive genes are referred to as Class II genes, to distinguish them from the classical VEGF-repressible, forkhead-responsive transcripts (Class I) 20. Together with the observation that FKHR/FOXO1 is essential for embryonic vascular development 32, 33, these data suggest that FKHR/FOXO1 is an essential transcription factor that is involved in the coordinated regulation of a distinct set of genes in endothelial cells involved in cell maintenance and health.
The history of signal input, or the “set point” of the endothelial cell, is an important determinant of subsequent signaling. For example, the term “ischemic preconditioning” describes the cytoprotective effect of ischemia-reperfusion or hypoxia on tissues and/or endothelial cells 34. Repeated challenge of endothelial cells with lipopolysaccharide (LPS) leads to endotoxin tolerance characterized by temporary insensitivity to subsequent LPS challenge. Other preconditioning regimens that have been demonstrated to alter endothelial cell signaling include hypoxia 35, heat shock 36, and insulin-like growth factor (IGF)-1 37. As a final example, we recently demonstrated that pre-incubation of endothelial cells with tumor necrosis factor (TNF)-α results in altered thrombin-mediated gene expression by modulating nuclear-cytoplasmic trafficking of p65 NF-κB 38. In the present study, we tested the hypothesis that preconditioning of endothelial cells with the PI3K-AKT agonist, HGF would affect the ability of VEGF to activate Class II genes by limiting the availability of FKHR/FOXO1 in the nucleus (the functional equivalent of FKHR/FOXO1 knockdown). We show that HGF does indeed attenuate VEGF induction of FKHR-dependent genes. These findings add a new level of complexity to forkhead signaling in endothelial cells.
Human coronary artery endothelial cells (HCAEC) were grown in Endothelial Growth Medium-2-MV (EGM-2-MV) BulletKit (Clonetics, San Diego, CA) at 37°C and 5% CO2. Endothelial cells from passage 3 to 6 were used for all experiments. Cells were serum-starved in 0.5% FBS prior to treatment with 50 ng/mL human VEGF-A165 or human 20 ng/mL HGF (PeproTech Inc, Rocky Hill, NJ). Where indicated, cells were pre-incubated for 30 min with 20 ng/mL HGF and then washed with PBS prior to VEGF treatment. Monoclonal anti-hVCAM-1 neutralizing antibody was purchased from Chemicon (Millipore, Tokyo, Japan).
HCAEC were transduced with replication-deficient adenoviruses encoding the cDNAs of β-galactosidase (Adv) and triple mutant (TM)-FKHR/FOXO1 as previously described 20, 26. The triple mutant version of FKHR/FOXO1 contains T24A, S256A, and S319A and is resistant to agonist-induced phosphorylation and nuclear export. Adenovirus expressing DN-Akt was previously described 19. Briefly, HCAEC were plated at a density of 1 × 106 cells per 10-cm plate and adenoviruses were added to the cells at five multiplicity of infection (5 MOI) after 5 h of plating. Cells were then allowed to grow for 36 h followed by overnight serum starvation. Expression of adenovirus-based FKHR/FOXO1 and Akt was confirmed using RT-PCR analyses (Supplemental Figure I shows FKHR/FOXO1) and also by Western blots as described previously 19, 20.
HCAEC were grown to 70–80% confluency in 6-cm plates and transfected with siRNA against the following FKHR/FOXO1 target sequences: siFKHR2 CAG CGC CGA CTT CAT GAG CAA (Qiagen, Valencia, CA) in Opti-MEM containing Lipofectin (10 μg/mL) for 4 h as previously described 20. The cells were then incubated in EGM-2 medium for 24 h and serum-starved in 0.5% serum for 12–16 h before VEGF treatment for the times indicated.
Quantitative real-time PCR. See Supplemental Methods.
Western and Northern blot analyses. See Supplemental Methods.
Immunolocalization studies. See Supplemental Methods.
Nuclear and cytoplasmic cellular fractionation. See Supplemental Methods.
Monocyte adhesion assay. See Supplemental Methods.
All values are presented as mean ± SD where appropriate. Statistical significance between two groups was determined by use of a paired t-test, and values of p<0.05 were considered significant.
HCAEC were incubated in the absence or presence of VEGF and/or HGF for varying times and assayed for phosphorylation of Akt and FKHR/FOXO1 by Western blot analysis. In response to VEGF, Akt was phosphorylated at 5 min, returning to basal levels by 30 min (Figure 1A, upper panel). VEGF-mediated phosphorylation of FKHR/FOXO1 occurred at 15 min and returned to basal levels by 60 min (Figure 1B). In contrast, HGF-induced phosphorylation of Akt and FKHR/FOXO1 occurred at 5 min and did not return to their basal levels until after 2 h (Figures 1A-B and data not shown). VEGF- and HGF-mediated phosphorylation of FKHR/FOXO1 was inhibited by the PI3K-inhibitor LY294002 (data not shown) and dominant-negative (DN) Akt (Figures 1C–D). Together, these data indicate that while both VEGF- and HGF-mediated phosphorylation of FKHR/FOXO1 occurs through a PI3K-Akt signaling pathway, the effect of HGF on FKHR/FOXO1 phosphorylation in HCAEC occurs earlier and is more sustained.
Phosphorylation of FKHR/FOXO1 has been shown to promote translocation of the transcription factor from the nucleus to cytoplasm 19, 39. In immunofluorescence assays, incubation of HCAEC with VEGF resulted in cytoplasmic translocation of FKHR/FOXO1 by 15–30 min, an effect that was reversed by 60 min (Supplemental Figure II). HGF-mediated nuclear exclusion of FKHR/FOXO1 occurred by 15 min. However, in contrast to VEGF, the effect of HGF on cytoplasmic translocation persisted at 60 min, returning to basal levels only after 2 h (Figure 2). HCAEC were also processed for cytoplasmic and nuclear fractions and assayed for FKHR/FOXO1 by Western blot. In these assays, nuclear-to-cytoplasmic translocation of FKHR/FOXO1 by VEGF was maximal at 30 min and returned to the basal levels within 60 min, whereas HGF-induced cytoplasmic translocation was maximal from 30 to 60 min but did not return to the basal levels until after 2 h (Supplemental Figure III). In accordance with the phosphorylation data (please see Figs. 1C-D), LY294002 and DN-Akt inhibited VEGF- and HGF-mediated cytoplasmic translocation of FKHR/FOXO1 (Supplemental Figure IV shows the effect of DN-Akt), suggesting that PI3K-Akt-induced phosphorylation is critical for subcellular localization of FKHR/FOXO1 in HCAEC. Consistent with the early and sustained effect of HGF phosphorylation and subcellular localization of FKHR/FOXO1, pre-incubation of HCAEC with HGF for 30 min preempted the effect of VEGF (Supplemental Figure V). Together, these findings suggest that HGF induces sustained nuclear exclusion of FKHR/FOXO1 and that pre-incubation with HGF limits the availability of nuclear FKHR/FOXO1 in VEGF-treated cells.
Based on the above findings, we reasoned that HGF-mediated depletion of FKHR/FOXO1 from the nucleus would impair the ability of VEGF to induce those genes that depend on FKHR/FOXO1 as a positive transcription factor (i.e., Class II genes). To test this hypothesis, HCAEC were serum-starved, preincubated with HGF, treated in the absence or presence of VEGF, and harvested for total RNA. In real-time PCR analyses, pretreatment of HCAEC with HGF for 30 min blocked VEGF-mediated induction of FKHR/FOXO1-dependent genes including VCAM-1, MnSOD, BMP2, ESM-1, MMP10, CITED2 and MGC5618 (Table 1). Representative Northern blot analyses for MnSOD and VCAM-1 are shown in Figure 3A. To determine whether this inhibitory effect of HGF was due to unavailability of non-phosphorylated FKHR/FOXO1 in the nucleus, we transduced HCAEC with adenoviruses expressing either β-gal (Adv) or phosphorylation-defective constitutively active triple mutant (TM)-FKHR/FOXO1. As shown in Table 1 and Figure 3B, the ability of HGF to attenuate VEGF-mediated induction of VCAM-1, MnSOD, BMP2, ESM-1, MMP10, CITED2 and MGC5618 expression was reversed by nuclear-localized TM-FKHR/FOXO1. To rule out an effect of HGF preconditioning on VEGFR2 phosphorylation, we carried out immunoprecipitation of VEGFR2 followed by Western blotting with an anti-phospho-tyrosine antibody. Pre-incubation of HCAEC with HGF did not significantly alter VEGF-induced phosphorylation or mRNA expression of VEGFR2 (Supplemental Figure VI). A previous study demonstrated that HGF inhibits VEGF activation of NF-κB 40. However, consistent with our previous results 20, we failed to demonstrate an effect of HGF on VEGF-induced NF-κB signaling. Specifically, HGF had no effect on VEGF-mediated phosphorylation of IκB or NF-κB p65 DNA binding (Supplemental Figure VII). Together, these data suggest that HGF limits the availability of nuclear-localized FKHR/FOXO1, which in turn impairs VEGF-inducible/FKHR/FOXO1-dependent gene expression.
We next wished to determine whether the inhibitory effect of HGF on VEGF signaling was functionally relevant. To that end, we asked whether HGF-mediated attenuation of VCAM-1 expression in VEGF-treated endothelial cells influenced monocyte adhesion. As shown in Figure 4, VEGF resulted in more than 3-fold induction of U937 adhesion to HCAEC. Pre-incubation of HCAEC with HGF resulted in inhibition of VEGF-induced monocyte adhesion (Figure 4, upper panel). This effect was reversed by overexpression of a constitutively active TM-FKHR/FOXO1 (Figure 4, lower panel). In contrast, siRNA-mediated knockdown of siFKHR/FOXO1 in HCAEC resulted in a decrease in the VEGF-induced monocyte adhesion, irrespective of the presence of HGF (Supplemental Figure VIII A). Finally, neutralizing anti-VCAM-1 antibody inhibited VEGF-inducible monocyte adhesion to HCAEC (Supplemental Figure VIII B). These data suggest that HGF-mediated phosphorylation and nuclear exclusion of FKHR/FOXO1 and subsequent attenuation of VEGF-inducible VCAM-1 expression results in reduced monocyte adhesion to endothelial cells.
VEGF activation of PI3K-Akt results in phosphorylation and nuclear exclusion of the FKHR/FOXO1. Nuclear exclusion, in turn, leads to downregulation of genes that are primarily regulated by forkhead (e.g., p27kip1, BTG-1). However, there is a distinct set of forkhead-responsive genes that are actually upregulated by VEGF 20. VEGF-mediated induction of these genes requires FKHR/FOXO1 and at least one other positive acting factor, such as NF-κB or NFAT 20. The existence of the latter gene class has two implications. First, for VEGF to induce the expression of forkhead-responsive genes there must either be residual FKHR/FOXO1 in the nucleus, or timely reentry of the transcription factor into the nucleus. Second, pre-incubation of endothelial cells with an agonist that selectively alters PI3K-Akt-FKHR/FOXO1 signaling should function as a rheostat for VEGF effect on this gene class. Consistent with this hypothesis, we have demonstrated that HGF, a physiological agonist that activates PI3K-Akt, but not NF-κB 26, blocks the stimulatory effect of VEGF on Class II genes including VCAM-1 and secondary monocyte adhesion.
HGF shares certain properties with VEGF. For example, both growth factors promote endothelial cell proliferation and migration in vitro, and induce angiogenesis in vivo 41. However, HGF and VEGF also exert distinct functions in endothelial cells. For example, VEGF increases endothelial cell permeability, whereas HGF promotes barrier function 28. HGF and VEGF induce largely non-overlapping patterns of gene expression in endothelial cells, suggesting different signal transduction pathways 42. In combination, HGF and VEGF act additively or synergistically to promote endothelial cell survival, proliferation, angiogenesis and downstream gene expression 41. The results of the present study provide evidence for an additional level of cross-talk between HGF and VEGF.
Our data support the hypothesis that HGF inhibits VEGF-mediated gene expression by interfering with the FKHR/FOXO1 arm of the VEGF signaling pathway. Compared with VEGF, HGF induces early, intense and prolonged phosphorylation and nuclear exclusion of FKHR/FOXO1. By depleting the nucleus of a critical threshold of FKHR/FOXO1 and/or delaying nuclear reentry of the transcription factor, HGF may deprive the VEGF signal of a necessary transacting protein required for activation of Class II genes (Figure 5). In further support of our hypothesis is the observation that the effect of HGF on VEGF was reversed by constitutively active TM-FKHR/FOXO1.
Two alternative explanations for an inhibitory effect of HGF on VEGF-mediated gene expression must be considered. First, it is formally possible that HGF-c-Met signaling inhibits VEGFR2 activation. However, our finding that HGF pretreatment failed to alter VEGF-mediated VEGFR2 phosphorylation argues against this mechanism. Second, HGF may inhibit NF-κB, which functions with FKHR/FOXO1 to activate many of the Class II genes. Indeed, a previous study showed that when added together, HGF inhibited VEGF-mediated induction of IκB phosphorylation, p65 nuclear translocation and DNA binding, and VCAM-1/ICAM-1 expression 40. However, we have been unable to demonstrate an effect of HGF on NF-κB activation in the absence or presence of VEGF. The reason for the discrepancy between the two studies is unclear, but may relate to the use of different cell types (HUVEC in the published report, HCAEC in the current study) or the timing of HGF addition (concomitant with VEGF in the published report, prior to VEGF in the current study). Regardless of a potential contribution of NF-κB, our findings point to an important role of the FKHR/FOXO1 pathway in modulating the cross-talk between HGF and VEGF. The functional relevance of this effect is evidenced by the observation that HGF priming reduced VEGF-mediated VCAM-1 dependent adhesion of monocytes. Based on a survey of the HGF-inhibitable Class II genes (Table 2), we predict that HGF signaling may also attenuate VEGF-inducible superoxide dismutase activity, ESM-1 secretion and MMP10 synthesis.
In summary, we have demonstrated that cross-talk between agonists may influence forkhead-dependent signaling in endothelial cells. Stated another way, the capacity of VEGF to promote expression of these Class II genes is dependent on the “set point” of the cell. These findings raise interesting questions about (patho)physiology. HGF is expressed by many cell types, including endothelial cells and vascular smooth muscle cells. HGF levels are increased in certain diseases, such as cancer and atherosclerosis 43, 44. Moreover, recent studies support the therapeutic use of HGF as a means of augmenting angiogenesis. It will be important to consider the potential effects of endogenous and/or exogenous HGF on the VEGF-FKHR/FOXO1 signaling axis in the endothelium.
This work was supported by National Institutes of Health Grant HL077348 (to W.C.A. and M.R.A) and American Heart Association Grant SDG 0453284N (to M.R.A.). T.M. was supported by NIBIO, NEDO, the Science and Technology from Ministry of Education, Culture, Sports, Sciences and Technology, and the Takeda Science foundation, Japan.