Coordinating Etk/Bmx Activation and VEGF Upregulation to Promote Cell Survival and Proliferation

Coordinating Etk/Bmx activation and VEGF upregulation to promote cell survival and proliferation

Cindy H Chau1, Kai-Yun Chen7, Hong-Tao Deng1, Kwang-Jin Kim1,2,3,4,5, Ken-ichi Hosoya8, Tetsuya Terasaki9, Hsiu-Ming Shih7 and David K Ann*,1,6

1Department of Molecular Pharmacology and Toxicology, University of Southern California, Los Angeles, California, CA 90033, USA; 2Department of Medicine, University of Southern California, Los Angeles, California, CA 90033, USA; 3Department of Physiology and Biophysics, University of Southern California, Los Angeles, California, CA 90033, USA; 4Department of Biomedical Engineering, University of Southern California, Los Angeles, California, CA 90033, USA; 5Will Rogers Institute Pulmonary Research Center, University of Southern California, Los Angeles, California, CA 90033, USA; 6Center for Craniofacial Molecular Biology, University of Southern California, Los Angeles, California, CA 90033, USA; 7Division of Molecular and Genomic Medicine, National Health Research Institute, Taipei 11529, Taiwan; 8Faculty of Pharmaceutical Sciences, Toyama Medical and Pharmaceutical University, 2630 Sugitani, Toyama 930-0194, Japan; 9Department of Molecular Biopharmacy and Genetics, Graduate School of Pharmaceutical Sciences, Tohoku University, Aoba, Aramaki, Aoba-ku, Sendai 980-8578, Japan

Etk/Bmx, a member of the Tec family of non-receptor that converge on the Etk-VEGF axis, causally associat- tyrosine kinase, is characterized by an N-terminal PH ing Etk-mediation of VEGF induction with enhanced domain and has recently been shown to be involved in the cellular processes in both epithelial and endothelial cells. regulation of various cellular processes, including Oncogene (2002) 21, 8817 – 8829. doi:10.1038/sj.onc. proliferation, differentiation, motility and apoptosis. 1206032 Since VEGF and the activation of its signaling pathway have been implicated in modulating a variety of Keywords: Etk; VEGF; autocrine; survival; prolifera- biological responses, we characterized the role of Etk- tion dependent signaling pathways involved in the upregulation of VEGF expression, and explored the functional implications of this enhancement in sustaining cell Introduction proliferation and survival. Using Northern and Western analyses, transient transfections, and pharmacological Cellular homeostasis is a process regulated by an agents, we demonstrate that Etk activation alone is integrative and coordinated balance among differentia- sufficient to transcriptionally induce VEGF expression, tion, proliferation and apoptosis. Alterations to this independent of the previously identified hypoxia response complex circuitry can lead to progressive conversion of element (HRE), in both Pa-4 epithelial and TR-BBB normal cells into prototypic cancer cells, which are endothelial cells under normoxia. In addition, Etk characterized by deregulated (i.e., autonomous) cell utilizes both MEK/ERK and PI3-K/Pak1 signaling growth. Results from biochemical and genetic studies pathways in concert to activate VEGF transcription. have revealed that members of the non-receptor Functionally, Etk activation elicits a profound stimula- tyrosine kinases, such as the Btk/Tec family, play a tory effect on TR-BBB cell proliferation and formation pivotal role in the regulation of cellular responses and of capillary-like networks in Matrigel containing reduced gene expression (Ekman et al., 1997; Qiu et al., 1998; levels of growth factors. Finally, antisense oligonucleo- Tamagnone et al., 1994; Tsai et al., 2000). tides against either endogenous VEGF or Etk abrogate Members of this kinase family consist of Btk, Itk, the proliferation of Etk-activated TR-BBB cells, and Tec, and Etk/Bmx (reviewed in Qiu and Kung, 2000), exogenous VEGF treatment stimulates endogenous Etk which share conserved structural motifs including an tyrosine phosphorylation in HUVECs. Taken together, N-terminal pleckstrin homology (PH) domain, these results indicate that VEGF is both an Etk followed by Src homology 3 (SH3) and 2 (SH2), and downstream target gene and an Etk upstream activator, a carboxyl-terminal tyrosine kinase domain. While Btk/ constituting a reciprocal Etk-VEGF autoregulatory loop. Tec family kinases are preferentially expressed in These findings, to our knowledge, are the first delineation hematopoietic cells and have lineage-specific functions, of a network of positive feedforward signaling pathways Etk/Bmx is expressed in epithelial, endothelial, and granulomonocytic cells (Qiu et al., 1998; Tamagnone et al., 1994). Etk activation was reported to be involved in Src-mediated cell transformation (Tsai et al., 2000) *Correspondence: DK Ann, University of Southern California, and IL-6 induced neuroendocrine differentiation of Health Science Campus, PSC-210B, 1985 Zonal Avenue, Los Angeles, California, CA 90033, USA; E-mail: [email protected] prostate cells (Qiu et al., 1998). Several lines of Received 21 May 2002; revised 30 August 2002; accepted 5 evidence have implicated Btk family members in September 2002 modulating apoptosis pathways under different experi- Etk/VEGF feedforward mechanism CH Chau et al 8818 mental paradigms. For example, overexpression of Etk induction of VEGF expression under hypoxic condi- renders LNCaP cells more resistant to photodynamic tions is mainly mediated by the hypoxia response therapy- or thapsigargin-induced apoptosis (Xue et al., element (HRE)/hypoxia inducible factor-1 (HIF-1) 1999), whereas Etk sensitizes mast cell line 32D toward (Forsythe et al., 1996; Ryan et al., 1998), the apoptosis upon treatment with G-CSF (Ekman et al., transcriptional hierarchy governing VEGF expression 2000). Wu et al. (2001) have demonstrated that Etk is during normoxia is only partially delineated. Because an enzyme substrate for caspase 3 during apoptosis, Etk enhances cell survival and overexpression of both implicating that Etk may serve as an apoptotic switch. VEGF and Etk are involved in tumor cell progression Recently, results from experiments using transient (Bagheri-Yarmand et al., 2001; Ferrara, 1999), we transfection in MCF-7 and MDA-MB435 cells hypothesize that VEGF may be a downstream target suggested that Etk enhances the proliferation and gene of the Etk signaling pathway. Moreover, as anchorage-independent, tumorigenic growth of upregulation of VEGF expression by PI3-K (Jiang et mammary epithelium-derived cells (Bagheri-Yarmand al., 2000) promotes cell survival and tumor growth et al., 2001). In addition, our laboratory has demon- (Blume-Jensen et al., 1998; Murakami et al., 1999; strated that Etk activation renders hypoxia resistance Shaw, 2001; Toker and Cantley, 1997; Vanhaesebroeck to Pa-4 cells by preserving their epithelial tight and Alessi, 2000; Zheng et al., 2000) in hypoxia, we junctions and barrier function during chronic hypoxic also postulate that PI3-K might also provide a link exposure (Hamm-Alvarez et al., 2001). Together, these between the Etk signaling cascade and VEGF gene data are consistent with a thesis that Etk/Bmx regulation under normoxia. Since VEGF governs cell activation is involved in the cytoprotection and survival through activation of a distinct ERK1/2 adaptive responses against environmental cues. pathway (Kanno et al., 2000) and previous studies However, the precise mechanism of how Etk assumes from our laboratory demonstrated that Etk activation such an essential role in the modulation of signal is sufficient to stimulate ERK1/2 kinases (Wen et al., pathways underlying cell survival and growth or 2000), we further propose that ERK1/2 provides an evading apoptosis remains to be elucidated. additional link between Etk and VEGF. Taken Cytokines and growth factors have also been shown together, Etk may utilize both ERK and PI3-K to mediate many aspects of multicellular communica- signaling pathways, acting in concert, to stimulate tions. Among them, vascular endothelial growth factor VEGF expression via an HRE-independent transcrip- (VEGF) is particularly adept in evoking an extremely tional activation of the VEGF promoter. broad spectrum of biological responses (Ferrara, 1999). Additionally, recent studies have suggested VEGF to For example, increased VEGF levels are associated be an autocrine growth factor in various tumor cell with stimulating cell proliferation, inducing cell migra- lines that express both VEGF and its receptor (Dias et tion, promoting cell survival, and regulating al., 2002; Masood et al., 2001; Strizzi et al., 2001). physiological and pathological angiogenesis (Neufeld However, the mechanism underlying this observation et al., 1999; Veikkola and Alitalo, 1999). The has yet to be established. Since our endothelial model, angiogenic action of VEGF involves an anti-apoptotic TR-BBBDEtk : ER, exhibits both properties, we also effect that promotes endothelial cell proliferation, predict that the enhancement in survival and prolifera- mediated through its tyrosine kinase receptor, Flk-1/ tion of endothelial cells occurs, at least in part, as a KDR/VEGFR-2, and the PI3-K/Akt signaling cascade consequence of an autocrine mechanism via the (Gerber et al., 1998a,b). Moreover, VEGF was found activation of endogenous Etk by Etk-stimulated VEGF to stimulate DNA synthesis of endothelial cells expression. In this study, we provide direct evidence preferentially through the same receptor, activating a that Etk activation alone is sufficient to promote the distinct ERK1/2 pathway (Kanno et al., 2000), which expression of endogenous VEGF in both epithelial and was considered indispensable for the survival of endothelial cells, contributing to their subsequent endothelial cells (Gupta et al., 1999). Furthermore, phenotypic manifestations via a reciprocal feedforward these observations have recently been extended to renal mechanism. tubular epithelial cells, suggesting that VEGF also promotes the survival of certain epithelial cells Results (Kanellis et al., 2000). Consequently, alterations in the expression and activity of VEGF have been Induction of VEGF expression by Etk activation in postulated to play a crucial role in acquisition of the Pa-4DEtk : ER cells migrating and invasive tumorigenic properties of human malignant cells (Bagheri-Yarmand et al., 2001). We have previously reported that Etk activation is able Mechanisms underlying the induction of VEGF to provide cytoprotection against hypoxia in Pa- expression involve multiple signaling pathways that 4DEtk : ER cells overexpressing an estradiol-inducible are rather complex and not clearly understood. form of Etk, DEtk : ER (Hamm-Alvarez et al., 2001). Transcriptional regulators for VEGF include cytokines, However, the specific signaling pathway and down- growth factors, oncogenes (Grugel et al., 1995; Rak et stream target genes activated by Etk are still elusive. al., 1995) and hypoxia (Shweiki et al., 1992), acting Since VEGF has been shown to be a survival factor for through either Src (Mukhopadhyay et al., 1995) or epithelial cells (Kanellis et al., 2000) and is also PI3-K signaling cascade (Jiang et al., 2000). While the upregulated by hypoxia (Jiang et al., 2000), we

Oncogene Etk/VEGF feedforward mechanism CH Chau et al 8819 examined herein whether VEGF could be an Etk Hck are all 110-fold higher than the IC50 determined downstream target gene and whether Etk is involved in for the recombinant Btk (Vassilev et al., 1999). Since the signaling cascade that leads to the activation of Etk is closely related to Btk and belongs to the same VEGF expression in both normoxia and hypoxia family of cytoplasmic protein tyrosine kinases, we conditions. Hence, Northern blot analyses were tested the possibility that LFM-A13 is capable of performed on total RNAs prepared from Pa-4 and attenuating Etk activity. As shown in Figure 1b, LFM- Pa-4DEtk : ER cells. The steady-state level of VEGF A13-treatment inhibited Etk tyrosine phosphorylation mRNA in Pa-4DEtk : ER cells was substantially higher in a dose-dependent manner (lanes 2 and 3 vs lane 1). than that detected in the parental Pa-4 cells under We then examined the effect of LFM-A13 treatment on normoxia; however, this was to a lesser extent when VEGF expression in the parental and Etk-activated Pa- compared to hypoxia (Figure 1a, lane 5 vs lanes 1 and 4 epithelial cells. In addition, since the PI3-K pathway 2, respectively). As expected, hypoxia and Etk activation has been demonstrated to augment VEGF activation together induced a much higher steady-state expression under hypoxia (Jiang et al., 2000), Wort- level of VEGF mRNA than that detected under either mannin, a PI3-K inhibitor, was used as a positive Etk activation or hypoxia alone (Figure 1a, lane 6 vs control. The steady-state b-actin mRNA levels were lanes 5 and 2, respectively). used to assure the quality and quantity of mRNAs We then sought to use pharmacological reagents to loaded onto each lane and the relative specificity of explore the role of Etk in modulating VEGF individual experimental paradigms. As shown in Figure expression under both experimental paradigms. LFM- 1a, both Wortmannin and LFM-A13 treatments A13, a leflunomide metabolite analog, has been rendered a marked attenuation of hypoxia-induced reported to be a selective inhibitor of Btk, but a poor VEGF induction in Etk-activated cells (lanes 7 and 8 vs inhibitor of EGFR, Hck, JAK1, JAK3A, and IRK lane 6). The downregulation of VEGF induction by the (Mahajan et al., 1999). The reported IC50 values for same treatments, to a lesser extent, was also observed EGFR, insulin receptor, JAK1, JAK2, JAK3A, Syk, in the parental Pa-4 cells under hypoxic condition (Figure 1a, lanes 3 and 4 vs lane 2), implicating that both PI3-K and endogenous Etk pathways might also be involved, following hypoxia. Together, we hypothesize that the increase in steady-state levels of VEGF expression can be achieved by a combined Etk- and hypoxia-dependent mechanism. Since Etk augments VEGF expression regardless of hypoxia, we herein focused our efforts on delineating the Etk-dependent pathways involved in VEGF gene regulation and investigating the role of Etk activation in governing cellular processes under normoxia.

Etk induces VEGF transcription via a HRE-independent mechanism Next, we directly determined whether Etk utilizes a hypoxia response element (HRE)-dependent and/or -independent mechanism to upregulate VEGF expression at the transcriptional level under normoxia. A series of 5’-VEGF promoter/enhancer deletion reporter constructs were engineered and examined by transient transfection assays (Figure 2). We ﬁrst conﬁrmed that a HRE-dependent pathway is operative in our cells, evident from the fact that transfected HIF-1a/1b supports hypoxia-mediated 71.2VEGF-luc induction (Chau CH and Ann DK unpublished observation). The Figure 1 Etk activation increases the steady-state level of relative abilities of these individually transfected VEGF mRNA. (a) Northern analyses were performed to analyse VEGF-reporter constructs to be stimulated by Etk the changes in VEGF mRNA expression in both Pa-4 and under normoxia were then analysed and compared in Pa-4DEtk : ER cells (see Materials and methods). Lanes 1 – 4: order to determine the DNA region required for Pa-4 cells treated with vehicle or either 100 nM Wortmannin (lane 3) or 50 mM LFM-A13 (lane 4); lanes 5 – 8: Pa-4DEtk : ER inducing VEGF expression by Etk signaling under cells treated with either 100 nM Wortmannin (lane 7) or 50 mM normoxia. As shown in Figure 2b, the DNA sequences LFM-A13 (lane 8). (b) Cells were transfected with T7-Etk and within the (71175/7790) DNA fragment were clearly LFM-A13 (50 or 100 mM) was added at 48 h post-transfection dispensable for Etk-mediated VEGF promoter activa- for 1 h. Immunoprecipitation and Western analyses were performed as described in Materials and methods. One representative tion (constructs I and IV vs constructs II and V, Northern and Western analysis from at least three independent respectively), supporting the notion that Etk utilizes an experiments is shown HRE-independent mechanism to potentiate VEGF

Oncogene Etk/VEGF feedforward mechanism CH Chau et al 8820

Figure 2 Schematic representation of various VEGF-luc reporter constructs and the induction of VEGF expression by Etk activation in Pa-4DEtk : ER cells. (a)ASacI–SacI fragment (1511 bp) from human VEGF gene, including the 1175 bp of promoter/enhancer region upstream of the transcription initiation site and the 336 bp of the 5’-untranslated region, was PCR-ampliﬁed and cloned in a sense orientation into the SacI site of pGL2-basic to create constructs I and IV. Constructs II, III, and IV were made by a double digestion of construct I with KpnI (vector)/PstI, SacII and ApaII, respectively. (b) The relative activities of the resultant reporter constructs were assayed and compared in the absence (basal activity) or presence (Etk-induced activity) of Etk activation. (c) Etk-stimulated VEGF-reporter activation is attenuated by a dominant negative kinase-dead Etk expression construct, EtkKQ. The luciferase assays and data analyses were carried out as described in Materials and methods. Each value shown is the mean+s.d. based on at least three independent transfection experiments

promoter activity. The inclusion of DNA sequences (Jui et al., 2000). As shown in Figure 2c, co- within the 7790/7268 region conferred approximately transfection with EtkKQ resulted in a substantial 2 – 3-fold stimulation of their basal activities, while inhibition of Etk-stimulated VEGF-reporter activation deletion of these sequences unequivocally reduced Etk under normoxia. Taken together, our results clearly inducibility (Figure 2b, constructs II and VI vs provide the ﬁrst demonstration that Etk activation constructs III and V). Moreover, Etk inducibility was alone is capable of upregulating VEGF expression at also partially retained in its minimal (7131/+336) the transcriptional level via an HRE-independent promoter sequences (Figure 2b, construct IV). mechanism in normoxia. In order to ascertain that this VEGF-reporter induction is indeed mediated by Etk activation alone, Activation of VEGF enhancer/promoter activity by Etk is we co-transfected Pa-4DEtk : ER cells with either conferred via both ERK and PI3-K pathways 71.2VEGF-luc or 70.13VEGF-luc and a dominant negative kinase-dead Etk expression construct, EtkKQ. In order to delineate Etk-mediated downstream EtkKQ harbors a point mutation (K445Q) in its ATP signaling events that stimulate VEGF promoter/ binding pocket, rendering the loss of its kinase activity enhancer activity, transient transfections assays with

Oncogene Etk/VEGF feedforward mechanism CH Chau et al 8821 reporter constructs devoid of the aforementioned HRE Treatment with 50 or 100 mM LFM-A13 attenuated the were utilized. To further conﬁrm that Etk activation is Etk-mediated VEGF-promoter induction in a dose- absolutely required for inducing VEGF expression in dependent manner (Figure 3a, lanes 4 and 5 vs lane 2), normoxia, Pa-4DEtk : ER cells were transiently trans- consistent with the observation in Figure 1a that LFM- fected with reporter constructs, 70.79VEGF-luc and A13 treatment downregulates VEGF expression. 70.13VEGF-luc, followed by Etk stimulation in the Since Etk activates the ERK pathway (Wen et al., presence or absence of LFM-A13, (Figure 1b), prior to 2000), we also investigated the contribution of the luciferase assays. As expected, Etk activation resulted ERK pathway to VEGF expression using Ras in approximately 9.7- and 4.1-fold inductions over their expression plasmids and appropriate pharmacological respective basal activities (Figure 3a, lane 2 vs lane 1). agents. First, expression plasmids encoding a constitu-

Figure 3 Stimulation of VEGF enhancer/promoter activity by Etk is conferred via both ERK and PI3-K pathways in two distinct DNA regions. (a) Reporter constructs II and IV (see Figure 2) were transiently co-transfected with expression plasmids encoding either a constitutively active Ras (Ras V12) or PI3-K (myr-p110), a dominant negative Ras (Ras A15), p21 activated kinase (Pak1DN), or EtkKQ; or treated with pharmacological agents U0126 (10 or 50 mM), LFM-A13 (50 or 100 mM), or LY 294002 (10 mM) for 16 h. The luciferase assays and data analyses were carried out as described in Figure 2. (b) Immunoprecipitation with anti-ER and Western analyses with either the 4G10 (anti-Tyr(P)) or an anti-ER were performed as described in Figure 1b on cells treated with vehicle, E2 alone, E2+10 mM LY 294002, E2+100 nM Wortmannin, and E2+50 mM U0126, respectively. These experiments have been repeated at least three times with comparable results. One representative Western analysis is shown

Oncogene Etk/VEGF feedforward mechanism CH Chau et al 8822 tively activated Ras, Ras V12, or a dominant negative expression of the Pak1DN (a dominant negative p21 Ras, Ras A15, were transiently co-transfected with activated kinase), a downstream effector of PI3-K, led 70.79VEGF-luc or 70.13VEGF-luc reporter con- to the attenuation of Etk-stimulated 70.79VEGF-luc structs to test whether Ras/ERK pathway is utilized activation, but not 70.13VEGF-luc activation (Figure by Etk to modulate VEGF expression. As shown in 3a, lane 12 vs lane 2). The activation of PI3-K clearly Figure 3a, our results clearly demonstrated that co- rendered VEGF transcriptional activation via the transfection with the constitutively activated Ras V12 (7790/7131), but not the (7131/+336), region. The recapitulated Etk-dependent 70.79VEGF-luc activa- lack of effect by co-transfected myr-p110 or Pak1DN, tion, albeit to a lesser extent (4.6-fold induction vs 9.7- and by LY 294002 pretreatment on the activities of fold induction; lane 6 vs lanes 1 and 2). By contrast, 70.13VEGF-luc served as a negative control, demon- Ras V12-mediated pathway was sufficient to mimic Etk strating the specificity of individual signaling pathways for 70.13VEGF-luc activation. Moreover, this Etk- on distinct regions located within the VEGF promoter/ mediated activation was inhibited by co-expression of enhancer. This observation provided a direct link the dominant negative Ras A15 (Figure 3a, lane 7). between PI3-K/Pak1 activation and Etk-mediated Thus, Etk-mediated enhancement of VEGF expression VEGF induction. To rule out the possibility that PI3- occurs, at least in part, via stimulating Ras-dependent K or MEK/ERK pathway acts upstream of Etk signaling pathways. Ras V12 was less efficacious in activation, we investigated whether LY 294002, stimulating 70.79VEGF-luc reporter, suggesting that a Wortmannin, or U0126 interferes with Etk activation. Ras-independent signaling pathway via the (7790/ As illustrated in Figure 3b, the pretreatment of each 7131) region is also critical for mediating VEGF inhibitor had very little, if any, effect on DEtk : ER transcriptional activation upon Etk stimulation. We activation. Taken together, our results conceivably further determined whether the activation of MEK is suggest that both PI3-K/Pak and MEK/ERK pathways required for Ras-dependent VEGF induction by Etk. are the downstream effectors of Etk, albeit via different The pharmacological inhibitor U0126 was chosen, as it DNA motifs, in modulating the transcriptional upre- inhibits MEK1/2 directly by blocking its catalytic gulation of VEGF expression under normoxic activity (Favata et al., 1998). Clearly, treatment with condition. Based on these observations, we propose 10 or 50 mM U0126 of Etk-activated and 70.13VEGF- that there are at least two distinct upstream and luc transfected cells resulted in a complete repression of downstream Etk response regions, independent of VEGF promoter activities (Figure 3a, lanes 8 and 9). HRE, acting in concert to modulate VEGF expression Therefore, activation of the MEK/ERK pathway is a under normoxia (see Discussion for details). necessary step for the transcriptional regulation of VEGF by Etk. However, inhibition of MEK activa- Etk induces VEGF expression in endothelial cells and tion, even with 50 mM U0126, did not completely enhances cellular differentiation on matrigel abolish the Etk-mediated 70.79VEGF-luc induction, implicating the involvement of at least one additional Since Etk is also expressed in endothelial cells and pathway other than MEK/ERK, activated by Etk to VEGF is a known endothelial cell mitogen, we regulate VEGF expression via the (7790/7131) examined the biological effect of Etk activation on region. endothelial cellular processes. To achieve this, rat Since Ras is also known to stimulate intracellular blood-brain-barrier microvascular endothelial TR- signaling via the PI3-K pathway, we next determined BBB cells (Terasaki and Hosoya, 2001) were stably whether PI3-K activation is also involved in this Etk- transfected with a DEtk : ER expression construct. A dependent signaling cascade. We used LY 294002, a clone (clone 4) exhibiting the highest DEtk : ER more selective PI3-K inhibitor than Wortmannin, to activation upon ER stimulation, as assessed by inhibit PI3-K activity in our transient transfection Western analyses with an anti-Tyr(P) antibody (data assays. Treatment with this inhibitor of 70.79VEGF- not shown), was selected for further studies and luc transfected cells resulted in a dose-dependent (data hereafter referred to as TR-BBBDEtk : ER cells. not shown) reduction of Etk-mediated 70.79VEGF- To ascertain if the Etk-mediated signaling pathways luc reporter activation. To further ascertain that the that we demonstrated above in epithelial Pa-4 cells also PI3-K pathway is involved in the induction of VEGF dictate overall VEGF expression level in endothelial by Etk, we co-transfected 70.79VEGF-luc and cells, we assessed the effect of Etk on VEGF promoter/ 70.13VEGF-luc with an expression construct, myr- enhancer activity using transient transfection and p110, harboring constitutively activated PI3-K. Ectopic reporter assays carried out in TR-BBBDEtk : ER cells. expression of the myr-p110 subunit in Pa-4 cells As illustrated in Figure 4a, Etk activation in TR-BBB resulted in a substantial enhancement of the Etk- cells also markedly induced the expression of trans- stimulated 70.79VEGF-luc activity, but not the Etk- fected 71.2VEGF-luc and 70.79VEGF-luc reporters, stimulated 70.13VEGF-luc reporter activity (Figure while a very modest induction on the transfected 3a, lane 10). Pretreatment of myr-p110 co-transfected 70.13VEGF-luc was elicited by Etk. Thus, the HRE is cells with 10 mM LY 294002 had an inhibitory effect on also not required for Etk-mediated VEGF induction Etk/p110-stimulated activities of 70.79VEGF-luc, under normoxia in endothelial cells. Furthermore, a while this combined regimen had little effect on the dominant negative form of Etk, EtkKQ, markedly induction of 70.13VEGF-luc (Figure 3a, lane 11). Co- repressed Etk-stimulated reporter activation when co-

Oncogene Etk/VEGF feedforward mechanism CH Chau et al 8823

Figure 4 Etk induces VEGF expression in endothelial cells and enhances cellular differentiation on Matrigel. (a) The effect of Etk on VEGF promoter/enhancer activity using reporter assays in TR-BBBDEtk : ER cells and co-transfected with dominant negative kinase-dead Etk expression construct, EtkKQ. The results represent the mean+s.d. of three separate experiments performed in duplicate. (b) TR-BBB or (c) TR-BBBDEtk : ER cells were plated on growth factor-reduced Matrigel in the absence of exogenous growth factors, serum, or phenol red for 9 h. Cells were photographed and assessed at 106 magniﬁcation using phase microscopy. Five random ﬁelds were examined and the experiment was repeated twice additionally

transfected in TR-BBB cells. These observations in Etk enhances the diﬀerentiation of endothelial cells on microvascular TR-BBB cells are reminiscent of the the basement membrane, we plated parental TR-BBB identiﬁcation of Etk as a positive stimulator of VEGF and TR-BBBDEtk : ER cells on growth factor-reduced gene expression via a HRE-independent mechanism in Matrigel. All cells were incubated in the absence of epithelial cells, as shown Figures 1 and 2. Hence, we exogenous growth factors, serum, or phenol red. As conclude that Etk could function to induce VEGF shown in Figure 4b, TR-BBBDEtk : ER cells formed expression in an analogous fashion in endothelial cells much more elaborate, branching, cord-like capillary to that in epithelial cells. networks at 9 h post-plating of cells than the parental Since VEGF is known to stimulate angiogenesis in TR-BBB cells. Thus, TR-BBBDEtk : ER cells retained vivo, we tested the ability of Etk to promote in vitro their endothelial characteristics of forming the intricate formation of capillary-like networks on Matrigel, a capillary network on Matrigel at a faster rate and to a protein substratum mixture that supports the organiza- greater extent than the parental TR-BBB cells, most tion of endothelial cells into tubes. To examine whether likely due to its Etk activation.

Oncogene Etk/VEGF feedforward mechanism CH Chau et al 8824 (Chen et al., 2001). As shown in Figure 5a,b, TR- Etk promotes the survival and proliferation of endothelial BBBDEtk : ER cells exhibited a more cuboidal cells via an autocrine mechanism morphology compared to the spindle shape of the To confirm the role of VEGF and Etk in promoting prototypical microvascular cell. Treatment of TR- proliferation and survival of TR-BBBDEtk : ER cells, BBBDEtk : ER cells with any of these two antisense two antisense oligonucleotides, AS-3 (anti-VEGF) oligonucleotides exhibited a marked growth inhibitory (Masood et al., 2001) and anti-Etk (Chen et al., effect (Figure 5d,e vs 5c), while a lesser effect was 2001), were utilized to downregulate the expression of observed on parental TR-BBB cells (data not shown). VEGF and Etk, respectively. These two antisense Western analyses were performed to assure the validity oligonucleotides have previously been tested extensively of our antisense approach. As shown in Figure 5d,e, for its specificity against their respective targets and anti-VEGF substantially attenuated VEGF expression, demonstrated to be effective reagents in many cell types while anti-Etk elicited a less pronounced effect on

Figure 5 Etk promotes the survival and proliferation of endothelial cells. Morphological differences between parental TR-BBB cells (a) and stably transfected TR-BBBDEtk : ER cells (b). Treatment of TR-BBBDEtk : ER cells with vehicle (c), 10 mM antisense VEGF (d), or 10 mM antisense Etk (e) for 96 h (see Materials and methods). The endogenous levels of VEGF in (d) and Etk in (e)as well as the level of ectopically expressed DEtk : ER in (e), compared to each respective level in the vehicle control (c), are also shown. Western analysis (f) for the presence of VEGF receptor, KDR/Flk-1, in TR-BBB (lane I) and TR-BBBDEtk : ER cells (lane II). The recombinant KDR (SC-4058 WB, Santa Cruz Biotechnology, Inc.) was included as a positive control (lane III)

Oncogene Etk/VEGF feedforward mechanism CH Chau et al 8825 endogenous Etk level and no effect on DEtk : ER VEGF. Taken together, these results clearly establish transgene expression in TR-BBBDEtk : ER cells. The that VEGF induces Etk activation in endothelial cells lack of robust inhibition on endogenous Etk expression and the signal transduction pathway mediated by Etk by anti-Etk may be due to the degradation of the is indeed indispensable for rendering the enhanced antisense oligonucleotides during our prolonged 48 h survival of these cells. incubation time in a gene-specific and cell context- dependent manner. Together, it is conceivable to propose that the endogenous VEGF/Etk expression is Discussion involved in maintaining the survival of these cells. As we have shown that Etk upregulates VEGF expression This report, for the first time, provides evidence that and promotes the proliferation of endothelial cells, we the transcription of VEGF, an important mitogen and postulate that this enhanced proliferative state might survival factor for both epithelial and endothelial cells, occur via a positive feedforward mechanism, i.e., a is directly induced by Etk activation. The subsequent reciprocal VEGF-induced activation of endogenous secretion of VEGF and binding to its receptor in an Etk is in place. To examine the effect of VEGF on autocrine fashion are able to further activate Etk, Etk, it is necessary to first determine if these cells ultimately rendering an autoregulatory loop to rein- express VEGF receptors. Hence, Western analysis was force its effect on cellular processes. We have identified performed on total cell lysates to indicate the presence Etk as one of the major key regulators in governing of the VEGF receptor, Flk-1/KDR/VEGFR-2, as epithelial/endothelial cell proliferation and survival. shown in Figure 5f. Such an autocrine manner of promoting cell survival In order to test the hypothesis that Etk increases the by VEGF has recently been investigated by several production of VEGF, which subsequently binds to its groups (Dias et al., 2001; Dunk and Ahmed, 2001; receptor after secretion into the culture media, leading Masood et al., 2001; Vega-Diaz et al., 2001). However, to the increased survival and enhanced cellular our data on the reciprocal stimulation between Etk processes of these cells, cultured HUVECs, known to activation and VEGF induction represent the first express Flk-1/KDR/VEGFR-2 (Kanno et al., 2000) demonstration, to our knowledge, on the roles of and Etk, were treated with 20 ng/ml VEGF and signaling pathways utilized by Etk in the transcrip- assayed for Etk activation. As Etk activation is known tional control of the VEGF gene and its subsequent to lead to its own tyrosine autophosphorylation (Jui et biological responses, especially in normoxia. al., 2000; Wen et al., 1999), we also examined whether We elucidated the mechanisms by which Etk/Bmx VEGF treatment results in a change in the tyrosine achieves specific transcriptional induction of endogen- phosphorylation of Etk. For this, Etk was immuno- ous chromosomal genes, in particular VEGF, to precipitated by anti-Etk antibody from lysates stimulate cell growth and survival by examining the prepared at various time intervals post-treatment, roles of individual pathways of Etk, MEK/ERK, and followed by Western analysis with an anti-Tyr(P) PI3-K/Pak1 signaling and their cross-talk in the antibody to detect Etk activation, while an anti-Etk induction of VEGF expression and activation of its antibody was used independently to demonstrate the downstream effector. The major findings of the present comparable loading of protein samples. As illustrated study are as follows: (1) Etk activation alone, utilizing in Figure 6, VEGF treatment alone was sufficient to a HRE-independent mechanism, is capable of inducing induce tyrosine phosphorylation of Etk within 30 min VEGF expression at the transcriptional level in both following VEGF treatment and further elevation was epithelial and endothelial cells under normoxia observed with a longer exposure of HUVECs to (Figures 1a, 2, and 4a); (2) either LFM-A13, an inhibitor for Btk/Tec tyrosine kinase family members, or a dominant negative form of Etk, EtkKQ, inhibits the Etk-dependent induction of VEGF promoter/ enhancer activation (Figures 1a and 3a), consistent with Etk-mediated upregulation of VEGF expression in normoxia; (3) both MEK/ERK and PI3-K/Pak1 signaling pathways are utilized by Etk, acting in concert to activate VEGF transcription (Figures 3 and 4a); (4) Etk activation also markedly potentiates their ability to form capillary networks in vitro (Figure 4b); (5) exogenous addition of antisense oligonucleotides against VEGF or Etk results in a dramatic Figure 6 Autophosphorylation of Etk upon treatment with inhibition of proliferation of these endothelial cells VEGF. VEGF stimulates tyrosine phosphorylation of Etk in HU- (Figure 5d,e); and (6) treatment with exogenous VEGF VECs. Confluent cultures of HUVEC were stimulated with 20 ng/ activates endogenous Etk in human endothelial ml VEGF for different time periods. Cells were lysed and sub- HUVECs (Figure 6). These data taken together suggest jected to immunoprecipitation experiments (see Materials and methods) with an anti-Etk antibody and further immunoblotted a feedforward mechanism to reinforce Etk-mediated with the anti-phosphotyrosine antibody (4G10) or anti-Etk anti- signaling events, providing direct evidence that there is body an auto-regulation dictating the reinforcement between

Oncogene Etk/VEGF feedforward mechanism CH Chau et al 8826 Etk activation and the upregulation of VEGF expres- dependent, pathways to stimulate VEGF expression; sion, ultimately leading to the observed phenotypic (2) MEK/ERK and PI3-K, two Ras downstream manifestations. signaling pathways, and a to-be-identified Ras-inde- We purposely set out to determine the mechanisms pendent signaling pathway are all utilized by Etk to utilized by Etk under the non-hypoxic condition to mediate the VEGF induction; (3) while MEK/ERK induce VEGF expression in both epithelial Pa- exerts its inducibility via both upstream and down- 4DEtk : ER and microvascular TR-BBBDEtk : ER cells. stream Etk response regions, PI3-K/Pak1 and Ras- Previous studies by others have shown the major role independent pathways activate VEGF promoter via the that HIF-1 plays in VEGF expression and angiogenesis upstream region exclusively; (4) the induced secretion in hypoxia (Forsythe et al., 1996; Ryan et al., 1998). of VEGF not only promotes cell proliferation and While hypoxia remains the undisputed, ubiquitous survival, but also activates Etk-mediated signaling inducer for VEGF expression, our results suggest that events further. Based on the results presented herein, activation of other signaling pathways, such as Etk, we conclude that both VEGF and Etk are involved in also increases VEGF expression in normoxia to an auto-regulatory loop that enhances cellular constitute a novel and additional regulatory network, processes and biological responses in both epithelial providing a timely and precise modulation of a variety and endothelial cells, such as cell proliferation and of physiological and pathological responses. Recently, capillary-like tube formation on growth factor-reduced Bagheri-Yarmand et al. (2001) reported that Etk Matrigel. activation enhances the proliferation, anchorage-inde- We would like to point out that Etk regulation of pendent growth, and tumorigenicity of human breast VEGF gene expression seems to be more dependent on cancer cells; however, the precise mechanism by which the MEK/ERK pathway, since inhibition of MEK1/2 Etk exerts its profound stimulatory effect on cancer with U0126 completely abrogated 70.13VEGF-luc cells is still elusive. We believe that our studies reported induction, while reducing 70.79VEGF-luc induction herein provide the first mechanistic evidence to by more than 65%. PI3-K, on the other hand, appears potentially link Etk activation/VEGF upregulation to play a somewhat lesser role, since treatment with a and increased cancer cell activity in vitro and in vivo, pharmacological inhibitor of PI3-K led to about 30% as an underlying mechanism. reduction in the stimulated 70.79VEGF-luc activities As illustrated in Figure 7, our results suggest that: (Figure 3a). Although stimulation of the PI3-K path- (1) Etk activation and hypoxia may function through way plays a less prominent role, regulation of VEGF at least two distinct, Etk-dependent and HRE- by PI3-K may be independent of MEK/ERK, since they appear to function in an additive manner (Figure 3a). Moreover, Etk-dependent VEGF induction can be achieved via both Ras-dependent and -independent mechanisms, since Ras V12 does not exactly reproduce the fold-induction achieved by Etk activation, and Ras A15 does not completely ameliorate Etk-mediated inducibility, either (Figure 3a). A direct role for the downstream effector(s) of Ras-independent pathway in the activation of VEGF expression, however, has yet to be determined. It is also possible that in addition to these identified effectors, there are other unidentified Etk downstream effectors that contribute to the observed phenotypic changes. While the mechanism to accomplish Etk-mediated VEGF induction in endothelial cells could be different from those identified in epithelial cells, as cellular contexts may impart different specificity on the VEGF promoter/enhancer, we have unequivocally demonstrated that Etk activation alone is sufficient to induce both 71.2VEGF-luc and 70.79VEGF-luc activities in TR-BBB cells (Figure 4a). It should be pointed out, however, that the utilization of these proposed Etk-mediated pathways by hypoxia also remains to be established. Considerable evidence from previous reports has established that VEGF signaling through Flk-1/KDR/ VEGFR-2 or Flt-1/VEGFR1 induces activation of ERK1/2, p38 MAPK, Akt, phospholipase Cg, and focal adhesion kinase (Abedi and Zachary, 1997; Figure 7 Putative mechanisms underlying an auto-regulation of Etk-dependent signaling pathways involved in modulating cell Gerber et al., 1998b; Kroll and Waltenberger, 1997; proliferation and angiogenesis. See Discussion for details. ERE: Mao et al., 2001; Wu et al., 2000). While focal Etk Response Element; HRE: Hypoxia Response Element adhesion kinase is associated with cell migration (Lu

Oncogene Etk/VEGF feedforward mechanism CH Chau et al 8827 et al., 2001), PI3-K/Akt and ERK1/2 are involved in 358C. Pa-4DEtk : ER cells were established and cultured as promoting cell survival (Parrizas et al., 1997). More- described previously (Wen et al., 1999). The conditionally over, ERK1/2 activation also promotes endothelial cell immortalized rat brain microvascular endothelial cells, TR- proliferation and morphological changes (Gotoh et al., BBB (Terasaki and Hosoya, 2001), were plated on type I 1999; Wu et al., 2000). Our results further implicate collagen-coated dish or flask in DMEM/F12 (1 : 1) medium (supplemented with 10% fetal bovine serum, endothelial cell that Etk has an anti-apoptotic/pro-proliferative role in growth factor (15 mg/ml), glutamate (5 mM), and kanamycin endothelial cells (Figure 6). The expression of Flk-1/ monosulfate (60 mg/ml), and maintained in a humidified KDR/VEGFR-2 is abundant in proliferating endothe- atmosphere of 5% CO2 and 95% air at 338C. For lial cells of embryonic vascular sprouts and early establishing TR-BBBDEtk : ER cells, TR-BBB cells were postnatal brain, but is reduced in adult brains (Mill- transfected with pLNCX-DEtk : ER construct (Wen et al., auer et al., 1993). Together with the data of Flk-1/ 1999) by the lipofectAMINETM 2000-mediated method KDR/VEGFR-2 being present in TR-BBB cells and according to the standard procedure suggested by the HUVEC (this report and Zeng et al., 2001), it is manufacturer (Invitrogen Corporation - Life Technologies, conceivable to conclude that Etk induces VEGF Carlsbad, CA, USA) and selected with G418 (800 mg/ml) for expression and that subsequently the secreted VEGF neomycin resistant colonies. Different colonies of stably transfected TR-BBBDEtk : ER cells were isolated by limiting binds to Flk-1/KDR/VEGFR-2 to modulate endothe- dilution. The expression and activation of DEtk : ER in TR- lial cellular processes in an autocrine manner. Studies BBBDEtk : ER cells were evaluated by Western analyses of with Etk/Bmx knockout mice demonstrated that Etk/ cell lysates using an anti-ER antibody or sodiumdodecylsul- Bmx-null mice are viable and without an obvious fate-polyacrylamide gel electrophoresis (SDS – PAGE) of phenotype, at least in the context of DBA/2 back- immunoprecipitates using an appropriate antibody. ground (Rajantie et al., 2001). Although this observation suggests that Etk-mediated signaling may Plasmid constructs, transient transfections and antisense be a redundant pathway, it does not rule out the treatment possibility that enhanced Etk-dependent signaling exerts a positive modulatory effect on various cell All the VEGF-reporter constructs used in transient transfection assays contain sequences derived from the human VEGF behaviors, rendering adaptive responses to noxious gene (Tischer et al., 1991). A SacI–SacI fragment (1511 bp) environment. Taken together, our results clearly from the human VEGF gene, including 1175 bp of the support the notion that Etk-dependent signal transduc- promoter/enhancer region upstream of the transcription tion and Etk-induced VEGF regulation have initiation site and 336 bp of the 5’-untranslated region was physiological and/or pathophysiological relevance, in PCR-amplified and cloned in a sense orientation into the the context of an autocrine-based mechanism. SacI site of pGL2-basic, to create 71.2VEGF-luc (Figure 2, In conclusion, the data from our current studies construct I). The individual 70.79VEGF-luc, 70.27VEGF- suggest that both Etk and VEGF are involved in an luc, and 70.13VEGF-luc (Figure 2, constructs II, III and IV) autocrine regulatory loop, which plays an important were made by a double digestion of 71.2VEGF-luc with role in governing cell proliferation and cellular KpnI (vector)/PstI, SacII and ApaII, respectively. Vector- containing fragments were isolated, blunt-ended by T4 DNA adaptive responses. Moreover, our results presented polymerase (Promega, Madison, WI, USA), and self-ligated. herein constitute the first direct evidence that Etk In addition, 0.66 kb (PstI/SacII) and 0.16 kb (SacII/ApaII) activation alone, independent of hypoxia, is sufficient fragments of VEGF were excised from 71.2VEGF-luc to to stimulate VEGF expression, utilizing both PI3-K/ construct (71175/7790)70.13VEGF-luc and (71175/ Pak1 and ERK pathways. Hence, Etk activation is 7268)70.13VEGF-luc reporter plasmids, respectively likely to be necessary for modulating these and other (Figure 2, constructs V and VI). Expression plasmids various cellular processes. Understanding the signaling harboring myr-p110 and Pak1DN were generous gifts from pathways mediated by Etk for the activation of VEGF Dr Yun Qiu (University of Minnesota) and Dr Gary Bokoch expression and the design of novel inhibitor molecules (Scripps Research Institute), respectively. effective against Etk and Etk-dependent cascades to Pa-4DEtk : ER cells or TR-BBBDEtk : ER cells were transiently transfected with 0.5 mg of reporter plasmids. disrupt this autoregulatory loop may have important One-tenth mg of the Renilla luciferase pRL-TK plasmid was implications in tumor progression, metastasis and co-transfected as an indicator for normalization of transfec- angiogenesis. tion efficiency. Six hours after the start of transfection, cells were recovered overnight in 0.1% serum-stripped medium and subsequently treated with vehicle (-Etk) or 1 mM b- estradiol (E2) to activate DEtk : ER (+Etk). Cells were Materials and methods treated for 16 h, before harvesting cell lysates for luciferase assays. The relative luciferase activity from the firefly- Cell culture luciferase reporter gene was determined and normalized with The rat parotid epithelial cell line, Pa-4, was plated on Renilla-luciferase activity using the Dual Luciferase Reporter Primaria culture dishes (BD Labware, Franklin Lakes, NJ, Assay System (Promega). All fold-induction data reported in USA) in Dulbecco’s modified Eagle’s/Ham’s F12 (1 : 1) this manuscript were calculated after normalizing the reporter medium (DMEM/F12), supplemented with 2.5% fetal bovine activities with the activities from the indicator plasmid. In serum, insulin (5 mg/ml), transferrin (5 mg/ml), epidermal general, there was no obvious change on the basal activity in growth factor (25 ng/ml), hydrocortisone (1.1 mM), glutamate our co-transfection and inhibitor studies. Sometimes, a (5 mM), and kanamycin monosulfate (60 mg/ml), and main- particular treatment may elicit a modulating effect on the tained in a humidified atmosphere of 5% CO2 and 95% air at general transcriptional machinery; however, these general

Oncogene Etk/VEGF feedforward mechanism CH Chau et al 8828 effects were taken into account after the normalization with with RIPA buffer supplemented with a cocktail of phospha- the activities of the indicator plasmid. tase and protease inhibitors. A comparable number (*10 000) of TR-BBB or TR- BBBDEtk : ER cells was seeded in 24-wells on Day 0. On Day Northern analysis 1, cells were treated with antisense oligonucleotides of either VEGF or Etk in serum-reduced (2.5% stripped FBS) Pa-4 and Pa-4DEtk : ER cells were cultured in a regular medium, as described previously (Masood et al., 2001). The medium prior to exposure to 20 or 1% O2 for 4 h. Cells were treatment was repeated on Day 3. On Day 5, cells were fixed exposed to either normoxic or hypoxic condition in 0.05% with paraformaldehyde and stained with Coommassie blue. serum-stripped medium prior to total RNA isolation. RNA Five random fields were photographed at 46 under the light blots were prepared and hybridization was carried out with microscope and the experiment was repeated twice addition- 32P-labeled VEGF probe prepared from an isolated rat ally. VEGF cDNA fragment (bases 1 – 560). All blots, after stripping, were re-probed with 32P-labeled rat b-actin cDNA probe to ensure that the quality and quantity of RNA Western analysis and immunoprecipitation between lanes were comparable. Blots were exposed to films TR-BBB and TR-BBBDEtk : ER cells were treated with 1 mM at 7808C overnight. E2 for 16 h and lysed with 16 loading buffer for SDS – PAGE. Approximately 20 mg of whole cell extract from each Matrigel assays for capillary-like network formation sample was subjected to SDS – PAGE on a 10% gel. The immunoblot analyses were carried out as described previously TR-BBB or TR-BBBDEtk : ER cells were plated on growth (Li et al., 1997). Anti-VEGF, anti-Flk-1/KDR/VEGFR-2 and factor-reduced Matrigel for 9 h at 338C. Cells were anti-Etk/Bmx antibodies were purchased from Santa Cruz photographed and assessed at 106 magnification using phase Biotechnology, Inc. (Santa Cruz, CA, USA) and Cell microscopy. Five random fields were examined and the Signaling Technology, Inc. (Beverly, MA, USA). For experiment was repeated twice. immunoprecipitation studies, cells were lysed in RIPA buffer containing 16PBS, 1% Nonidet P-40, 0.5% sodium Statistical analysis deoxycholate, 0.1% SDS, 100 mg/ml phenylmethylsulphonyl fluoride, 30 mg/ml aprotinin and 1 mM sodium orthovana- Experiments were carried out in duplicate at least three times. date. Subsequently, the endogenous Etk was precipitated Representative data from these three independent experi- from cell lysates with an anti-Etk/Bmx antibody as described ments were presented. Significance (P50.05) was determined previously (Li et al., 1997), followed by immunoblot analyses relative to the corresponding control by Student’s t-test. The using an anti-phosphotyrosine (anti-Tyr(P)) antibody (4G10, reporter activity shown is the mean+s.d. based on at least Upstate Biotechnology, Inc., Lake Placid, NY, USA). three independent transfection experiments. Human endothelial cells from umbilical veins (HUVEC), passages 2 – 5, were cultured in Medium 199 supplemented with 16.6% fetal bovine serum, 14 U/ml heparan sulfate, 20 mg/ml endothelial cell growth supplement (ECGS, Upstate Biotechnology), and 2 mML-glutamine. HUVEC was Acknowledgments cultured for 48 – 72 h to reach confluence and then starved We thank Drs Rizwan Masood, Parkash Gill, Yun Qiu and for 6 h with Medium 199 supplemented with 1.5% FBS, Gary Bokoch for useful reagents and Dr Wei-Chiang Shen 14 units/ml heparan sulfate and 2 mML-glutamine. For for helpful suggestions on the Matrigel assay. This work VEGF stimulation, the cells were pretreated with 0.5 mM was supported in part by National Institute of Health vanadate for 1 h, followed by adding 20 ng/ml of human Research grants R01 DE 10742 and R01 DE 14183 (to DK recombinant VEGF165 (Sigma, St. Louis, MO, USA) to Ann), R01 HL 38658 and R01 HL 64365 (to K-J Kim), and media for the indicated time periods and subjected to lysis Predoctoral Fellowship DE 07211 (to CH Chau).

References

Abedi H and Zachary I. (1997). J. Biol. Chem., 272, 15442 – Ekman N, Arighi E, Rajantie I, Saharinen P, Ristimaki A, 15451. Silvennoinen O and Alitalo K. (2000). Oncogene, 19, Bagheri-Yarmand R, Mandal M, Taludker AH, Wang RA, 4151 – 4158. Vadlamudi RK, Kung HJ and Kumar R. (2001). J. Biol. Ekman N, Lymboussaki A, Vastrik I, Sarvas K, Kaipainen A Chem., 276, 29403 – 29409. and Alitalo K. (1997). Circulation, 96, 1729 – 1732. Blume-Jensen P, Janknecht R and Hunter T. (1998). Curr. Favata MF, Horiuchi KY, Manos EJ, Daulerio AJ, Stradley Biol., 8, 779 – 782. DA,FeeserWS,VanDykDE,PittsWJ,EarlRA,Hobbs ChenR,KimO,LiM,XiongX,GuanJL,KungHJ,Chen F, Copeland RA, Magolda RL, Scherle PA and Trzaskos H, Shimizu Y and Qiu Y. (2001). Nat. Cell Biol., 3, 439 – JM. (1998). J. Biol. Chem, 273, 18623 – 18632. 444. Ferrara N. (1999). J. Mol. Med., 77, 527 – 543. Dias S, Choy M, Alitalo K and Raﬁi S. (2002). Blood, 99, ForsytheJA,JiangBH,IyerNV,AganiF,LeungSW,Koos 2179 – 2184. RD and Semenza GL. (1996). Mol. Cell. Biol, 16, 4604 – DiasS,HattoriK,HeissigB,ZhuZ,WuY,WitteL,Hicklin 4613. DJ,TatenoM,BohlenP,MooreMAandRaﬁiS.(2001). Gerber HP, Dixit V and Ferrara N. (1998a). J. Biol. Chem., Proc. Natl. Acad. Sci. USA, 98, 10857 – 10862. 273, 13313 – 13316. Dunk C and Ahmed A. (2001). Am.J.Pathol.,158, 265 – 273.

Oncogene Etk/VEGF feedforward mechanism CH Chau et al 8829 Gerber HP, McMurtrey A, Kowalski J, Yan M, Keyt BA, Rajantie I, Ekman N, Iljin K, Arighi E, Gunji Y, Kaukonen Dixit V and Ferrara N. (1998b). J. Biol. Chem., 273, J, Palotie A, Dewerchin M, Carmeliet P and Alitalo K. 30336 – 30343. (2001). Mol. Cell. Biol., 21, 4647 – 4655. Gotoh I, Fukuda M, Adachi M and Nishida E. (1999). J. Rak J, Mitsuhashi Y, Bayko L, Filmus J, Shirasawa S, Biol. Chem., 274, 11874 – 11880. Sasazuki T and Kerbel RS. (1995). Cancer Res., 55, 4575 – Grugel S, Finkenzeller G, Weindel K, Barleon B and Marme 4580. D. (1995). J. Biol. Chem., 270, 25915 – 25919. Ryan HE, Lo J and Johnson RS. (1998). EMBO J., 17, Gupta K, Kshirsagar S, Li W, Gui L, Ramakrishnan S, 3005 – 3015. Gupta P, Law PY and Hebbel RP. (1999). Exp. Cell. Res, Shaw LM. (2001). Mol. Cell. Biol, 21, 5082 – 5093. 247, 495 – 504. Shweiki D, Itin A, Soffer D and Keshet E. (1992). Nature, Hamm-Alvarez SF, Chang A, Wang Y, Jerdeva G, Lin HH, 359, 843 – 845. Kim KJ and Ann DK. (2001). Am. J. Physiol. Cell. Physiol, Strizzi L, Catalano A, Vianale G, Orecchia S, Casalini A, 280, C1657 – C1668. Tassi G, Puntoni R, Mutti L and Procopio A. (2001). J. Jiang BH, Zheng JZ, Aoki M and Vogt PK. (2000). Proc. Pathol., 193, 468 – 475. Natl. Acad. Sci. USA, 97, 1749 – 1753. Tamagnone L, Lahtinen I, Mustonen T, Virtaneva K, Jui HY, Tseng RJ, Wen X, Fang HI, Huang LM, Chen KY, Francis F, Muscatelli F, Alitalo R, Smith CI, Larsson C Kung HJ, Ann DK and Shih HM. (2000). J. Biol. Chem, and Alitalo K. (1994). Oncogene, 9, 3683 – 3688. 275, 41124 – 41132. Terasaki T and Hosoya K. (2001). Biol. Pharm. Bull., 24, Kanellis J, Fraser S, Katerelos M and Power DA. (2000). 111 – 118. Am. J. Physiol. Renal Physiol., 278, F905 – F915. TischerE,MitchellR,HartmanT,SilvaM,Gospodarowicz Kanno S, Oda N, Abe M, Terai Y, Ito M, Shitara K, D, Fiddes JC and Abraham JA. (1991). J. Biol. Chem., 266, Tabayashi K, Shibuya M and Sato Y. (2000). Oncogene, 11947 – 11954. 19, 2138 – 2146. Toker A and Cantley LC. (1997). Nature, 387, 673 – 676. Kroll J and Waltenberger J. (1997). J. Biol. Chem., 272, Tsai YT, Su YH, Fang SS, Huang TN, Qiu Y, Jou YS, Shih 32521 – 32527. HM, Kung HJ and Chen RH. (2000). Mol. Cell. Biol., 20, Li D, Lin HH, McMahon M, Ma H and Ann DK. (1997). J. 2043 – 2054. Biol. Chem., 272, 25062 – 25070. Vanhaesebroeck B and Alessi DR. (2000). Biochem. J., 346, Lu Z, Jiang G, Blume-Jensen P and Hunter T. (2001). Mol. 561 – 576. Cell. Biol., 21, 4016 – 4031. Vassilev A, Ozer Z, Navara C, Mahajan S and Uckun FM. Mahajan S, Ghosh S, Sudbeck EA, Zheng Y, Downs S, (1999). J. Biol. Chem., 274, 1646 – 1656. Hupke M and Uckun FM. (1999). J. Biol. Chem., 274, Vega-Diaz B, Herron GS and Michel S. (2001). J. Invest. 9587 – 9599. Dermatol., 116, 525 – 530. Mao CD, Hoang P and DiCorleto PE. (2001). J. Biol. Chem., Veikkola T and Alitalo K. (1999). Semin. Cancer Biol., 9, 276, 26180 – 26188. 211 – 220. Masood R, Cai J, Zheng T, Smith DL, Hinton DR and Gill Wen X, Lin HH and Ann DK. (2000). Adv. Dent. Res., 14, PS. (2001). Blood, 98, 1904 – 1913. 76 – 80. Millauer B, Wizigmann-Voos S, Schnurch H, Martinez R, Wen X, Lin HH, Shih HM, Kung HJ and Ann DK. (1999). J. Moller NP, Risau W and Ullrich A. (1993). Cell, 72, 835 – Biol. Chem., 274, 38204 – 38210. 846. Wu LW, Mayo LD, Dunbar JD, Kessler KM, Baerwald MR, Mukhopadhyay D, Tsiokas L, Zhou XM, Foster D, Brugge Jaffe EA, Wang D, Warren RS and Donner DB. (2000). J. JS and Sukhatme VP. (1995). Nature, 375, 577 – 581. Biol. Chem., 275, 5096 – 5103. Murakami H, Iwashita T, Asai N, Shimono Y, Iwata Y, Wu YM, Huang CL, Kung HJ and Huang CY. (2001). J. Kawai K and Takahashi M. (1999). Biochem. Biophys. Res. Biol. Chem., 276, 17672 – 17678. Commun., 262, 68 – 75. Xue LY, Qiu Y, He J, Kung HJ and Oleinick NL. (1999). Neufeld G, Cohen T, Gengrinovitch S and Poltorak Z. Oncogene, 18, 3391 – 3398. (1999). Faseb J., 13, 9 – 22. Zeng H, Dvorak HF and Mukhopadhyay D. (2001). J. Biol. Parrizas M, Saltiel AR and LeRoith D. (1997). J. Biol. Chem., 276, 26969 – 26979. Chem., 272, 154 – 161. Zheng DQ, Woodard AS, Tallini G and Languino LR. Qiu Y and Kung HJ. (2000). Oncogene, 19, 5651 – 5661. (2000). J. Biol. Chem., 275, 24565 – 24574. Qiu Y, Robinson D, Pretlow TG and Kung HJ. (1998). Proc. Natl. Acad. Sci. USA, 95, 3644 – 3649.

Oncogene