Report

Role of Raf in Vascular Protection from Distinct Apoptotic Stimuli

See allHide authors and affiliations

Science  04 Jul 2003:
Vol. 301, Issue 5629, pp. 94-96
DOI: 10.1126/science.1082015

Abstract

Raf kinases have been linked to endothelial cell survival. Here, we show that basic fibroblast growth factor (bFGF) and vascular endothelial growth factor (VEGF) differentially activate Raf, resulting in protection from distinct pathways of apoptosis in human endothelial cells and chick embryo vasculature. bFGF activated Raf-1 via p21-activated protein kinase–1 (PAK-1) phosphorylation of serines 338 and 339, resulting in Raf-1 mitochondrial translocation and endothelial cell protection from the intrinsic pathway of apoptosis, independent of the mitogen-activated protein kinase kinase–1 (MEK1). In contrast, VEGF activated Raf-1 via Src kinase, leading to phosphorylation of tyrosines 340 and 341 and MEK1-dependent protection from extrinsic-mediated apoptosis. These findings implicate Raf-1 as a pivotal regulator of endothelial cell survival during angiogenesis.

Vascular remodeling and neovascularization can be induced by a wide variety of cytokines and growth factors. In addition to promoting endothelial cell (EC) proliferation and invasion, angiogenic growth factors protect ECs from both intrinsic and extrinsic inducers of apoptosis. The intrinsic pathway is activated at the mitochondria in response to stress, such as nutrient deprivation or DNA damage, whereas the extrinsic pathway is induced by receptor binding to proapoptotic death ligands such as tumor necrosis factor–α (TNF-α) and Fas.

bFGF and VEGF are EC survival factors that activate two distinct signaling pathways leading to angiogenesis (14). Because Raf kinases have been shown to be essential to this process (58), we evaluated the mechanisms underlying the antiapoptotic functions of bFGF and VEGF and the role played by Raf kinases in this response. We exposed ECs to the individual growth factors and induced the cells to undergo apoptosis through either the intrinsic (stress) or extrinsic (receptor) pathway. Surprisingly, bFGF preferentially protected ECs from stress-mediated death, whereas VEGF was primarily effective against receptor-mediated apoptosis (Fig. 1A). We also examined the effects of bFGF and VEGF on neovessel formation in 10-day-old chick chorioallantoic membranes (CAMs) in the presence of characterized inducers of angiogenic EC apoptosis via intrinsic (doxorubicin) or extrinsic [TNF-α/interferon-γ (IFN-γ)] death pathways (fig. S1). Consistent with the in vitro data, bFGF and VEGF preferentially promoted angiogenesis in the presence of doxorubicin and TNF-α/IFN-γ, respectively (Fig. 1B).

Fig. 1.

Differential protection of ECs in vitro and in vivo. (A) Human umbilical vein ECs (HUVECs) were starved in serum-free medium or treated overnight with 50 μM etoposide, 10 μM doxorubicin, TNF-α/IFN-γ cocktail (1 ng/ml each), or antibody to Fas (500 ng/ml) in 1% serum-containing medium in the presence or absence of bFGF or VEGF, each at 100 ng/ml. Apoptosis was scored by assessing cellular morphology or Annexin-V staining. Protection is expressed as percent increase in survival of growth factor–treated ECs relative to untreated controls. Data are representative of three independent experiments. Bars represent means ± SEM of 10 random fields of at least 30 cells. (B) Ten-day-old chick CAMs were exposed to filter paper disks saturated with 100 ng of bFGF or VEGF. After 48 hours, CAMs were treated with doxorubicin or TNF-α/IFN-γ together with 100 ng of VEGF or bFGF for 24 hours. De novo angiogenesis is expressed as the number of growth factor–induced branch points relative to untreated controls. Negative values represent disruption of preexisting vasculature. Data are representative of three independent experiments.

We next considered how Raf becomes activated during angiogenesis in response to these growth factors. The kinase activity of both B-Raf and Raf-1 (also known as c-Raf) depends not only on Ras, but also on growth factor–activated signaling pathways (9, 10). Raf-1 contains a central activation domain (Fig. 2A) that can be phosphorylated and activated by either p21-activated protein kinase (PAK) at amino acids Ser338 and Ser339 (SS338/339) (11) or Src kinase at amino acids Tyr340 and Tyr341 (YY340/341) (12), whereas B-Raf lacks the corresponding Src site. We investigated whether these upstream kinases are involved in Raf-1 activation in ECs after stimulation with bFGF or VEGF. Lysates from CAMs treated with bFGF or VEGF were evaluated by immunoblot analysis using phospho-specific antibodies directed to the PAK (SS338/339) or Src (YY340/341) sites within the activation domain of Raf-1. Interestingly, bFGF selectively promoted phosphorylation at the PAK site, whereas VEGF induced phosphorylation at the Src site (Fig. 2B). In parallel, we examined the activity of both of the EC Raf kinases, Raf-1 and B-Raf, in response to bFGF and VEGF. Although Raf-1 showed robust activation in response to both growth factors, B-Raf was not responsive to VEGF and was weakly activated by bFGF (fig. S2).

Fig. 2.

Differential Raf-1 activation by VEGF and bFGF. (A) Schematic showing key activation sites of Src. (B) Chick CAMs (10 days) were stimulated with VEGF or bFGF as in Fig. 1. After 20 hours, CAM tissue was excised and extracted with detergent. Lysates were probed with antibodies to Raf-1 specific for phosphorylated SS338/339 or YY340/341. (C) CAMs were transduced with RCAS-PAK83-149 and stimulated with bFGF or VEGF for 20 hours. Raf-1 was immunoprecipitated and subjected to an in vitro kinase assay using kinase-dead MEK as a substrate (22). (D) CAMs treated with the Src inhibitor PP1 (10 μM) or transduced with RCAS-Src251 (2, 3) were stimulated with bFGF or VEGF, and a Raf kinase assay was performed. Blots were probed with an antibody to Raf-1 as a loading control.

To evaluate whether the Src and PAK sites on Raf-1 are phosphorylated within intact blood vessels, we analyzed cryostat sections of CAMs stimulated with VEGF or bFGF by immunostaining with phospho-specific antibodies to the PAK or Src sites on Raf-1. Consistent with the immunoblotting results, intact blood vessels stimulated with VEGF stained preferentially with antibody to YY340/341, and those stimulated with bFGF stained only with antibody to SS338/339 (fig. S3). Interestingly, although Raf was differentially phosphorylated, both growth factors induced vascular staining with antibody to phospho-ERK in these tissues (fig. S4). Thus, both bFGF and VEGF induce ERK activity in blood vessels, but they likely do so via distinct pathways of Raf activation.

To evaluate whether PAK or Src was required for Raf activity in vascular tissue, we transduced growth factor–stimulated CAMs with replication-competent avian sarcoma leukosis virus long terminal repeat with a splice acceptor (RCAS) retrovirus encoding the PAK-1 autoinhibitory domain (PAK83-149), which potently inhibits PAK-1 activity by blocking substrate binding to its catalytic site (13). Alternatively, CAMs were transduced with kinase-deleted Src (Src251) (14) or treated with a pharmacological inhibitor of Src (PP1). PAK inhibition selectively suppressed bFGF-induced Raf activation (Fig. 2C), whereas Src251 or PP1 selectively blocked VEGF-mediated Raf activity (Fig. 2D).

Growth factor activation of Raf-1 leads to MEK/ERK activation (1517). To investigate the role of MEK1 and ERK1/2 in growth factor–mediated protection, we added the MEK1 inhibitor PD98059 to cultured ECs that had been treated with either bFGF or VEGF and then exposed the cells to inducers of the intrinsic or extrinsic pathways of apoptosis, respectively. PD98059 inhibits MEK1 but has also been shown to inhibit ERK5. Interestingly, MEK1 and ERK1/2 were required only for VEGF-mediated protection of ECs from the extrinsic pathway of apoptosis; bFGF protected cells against the intrinsic apoptotic cascade in the presence of the MEK1 inhibitor (Fig. 3A). The response was specific, because the concentration of PD98059 used (10 μM) completely blocked ERK1/2 phosphorylation yet had no effect on the phosphorylation of ERK5 (fig. S5).

Fig. 3.

Role of Raf in EC survival. (A) HUVECs cultured in the absence of serum or treated with TNF-α/IFN-γ were treated with bFGF or VEGF (100 ng/ml), respectively. Cells were treated with or without the MEK inhibitor PD98059 (10 μM) and analyzed for apoptosis as in Fig. 1. (B) HUVECs were electroporated with cDNAs (20 μg) encoding each Raf-1 mutant, starved overnight, and then exposed to bFGF or VEGF (100 ng/ml) for 10 min. Cell lysates were fractionated, electrophoresed, and probed with antibodies to Raf-1, HSP60 (mitochondrial loading control), and focal adhesion kinase (FAK, whole-cell loading control). (C) HUVECs were electroporated with cDNAs encoding each Raf mutant (20 μg) and green fluorescent protein (GFP) (10 μg). Twenty-four hours later, cells were starved overnight in the presence or absence of bFGF (100 ng/ml) or treated with TNF-α/IFN-γ (1 ng/ml) in the presence or absence of VEGF (100 ng/ml). Apoptosis was scored on the basis of cellular morphology (satellite formation and nuclear condensation). Protection is expressed as percent increase in survival of growth factor–treated ECs relative to untreated controls. Negative values represent possible sensitization of ECs to apoptosis. Bars represent means ± SEM from 10 fields containing at least 30 cells.

When Raf-1 is experimentally targeted to the mitochondrial membrane, it can protect cells from stress-mediated apoptosis independent of MEK/ERK activity (18). Therefore, we explored whether bFGF stimulation of ECs could promote Raf translocation to the mitochondria, leading to cell survival in response to stress. Serum-starved ECs were treated with either bFGF or VEGF, and lysates of these cells were subjected to subcellular fractionation to enrich for cytoplasmic or mitochondrial fractions, followed by immunoblot analysis to detect Raf-1. Notably, bFGF, but not VEGF, promoted Raf-1 translocation to the mitochondria (Fig. 3B). We next determined whether the bFGF-dependent phosphorylation of Raf-1 was required for mitochondrial translocation. Expression of a cDNA encoding point mutations at the PAK site on Raf-1 (Raf-SS338/339AA) in ECs completely blocked bFGF-mediated translocation of Raf to the mitochondria (Fig. 3B). In contrast, expression of Raf with point mutations at the Src site (YY340/341) did not prevent Raf translocation (Fig. 3B). Accordingly, expression of Raf-SS338/339AA, but not Raf-YY340/341FF, selectively blocked bFGF-mediated EC survival in response to stress (Fig. 3C). However, Raf-YY340/341FF blocked VEGF-induced survival of TNF-α/IFN-γ–treated cells, whereas expression of Raf-SS338/339AA had no effect (Fig. 3C). These findings demonstrate that the PAK site of Raf-1, which plays a key role in Raf coupling to the mitochondria, is critical for bFGF-mediated protection against intrinsic apoptosis, whereas the Src site of Raf-1 is critical for VEGF-mediated protection against extrinsic apoptosis.

Our results provide important insights into the mechanisms underlying EC survival signaling during normal and pathological neovascularization. Angiogenesis occurs in diverse proapoptotic microenvironments, and vascular growth and repair in these settings may require distinct angiogenic factors. For example, ECs at sites of inflammation may be exposed to high levels of death ligands (TNF and Fas) that compromise cell survival. In contrast, ECs at wound repair sites may be exposed to factors that initiate stress-mediated (19) or integrin-mediated apoptosis (20). These finding may be particularly relevant to the pathological angiogenesis associated with tumor growth, where multiple growth factors contribute to EC activation and angiogenesis. In this case, prosurvival signaling by VEGF and bFGF may cooperate to alter the threshold level of EC susceptibility to intrinsic and extrinsic inducers of apoptosis. This, in turn, may promote resistance of tumor-associated vasculature to cytotoxic agents such as irradiation and chemotherapeutic drugs (21). Delineation of these distinct signaling pathways for EC survival may facilitate the design of effective combination therapies for disruption of tumor-associated angiogenesis.

Supporting Online Material

www.sciencemag.org/cgi/content/full/301/5629/94/DC1

Materials and Methods

Figs. S1 to S5

References

References and Notes

View Abstract

Navigate This Article