Flt1, an “fms-like tyrosine kinase” receptor, has been suggested to play an active role in vascular endothelial growth factor (VEGF)–mediated autocrine signaling of tumor growth and angiogenesis. Here, we used a neuroblastoma model to investigate the role of VEGF/Flt1 signaling in hypoxia-mediated tumor cell survival, drug resistance, and in vivo angiogenesis. SK-N-BE(2), a highly malignant neuroblastoma cell line resistant to hypoxia-induced apoptosis expresses active Flt1 but lacks VEGFR2 expression. We found that 24-hour hypoxia (<0.1% O2) alone (no serum deprivation) showed sustained activation of extracellular signal-regulated kinase 1/2 (ERK1/2) associated with bcl-2 up-regulation and resistance to etoposide-induced (5 μmol/L) apoptosis. Treatment with anti-VEGF and anti-Flt1 antibodies inhibited ERK1/2 activation, down-regulated bcl-2, and reversed the hypoxia-mediated drug resistance to etoposide. Similar results were obtained with U0126 and ursolic acid, specific and nonspecific inhibitors of ERK1/2, respectively. We confirmed the protective role of Flt1 receptor by small interfering RNA knockout and Flt1 overexpression studies. Subsequently, we found that inhibition of VEGF/Flt1 autocrine signaling led to reduced hypoxia-inducible factor-1α (HIF-1α) phosphorylation. Furthermore, the reduced phosphorylation was associated with down-regulation of basic fibroblast growth factor, a downstream target of the HIF-1α and VEGF pathways. Our findings suggested an expanded autocrine loop between VEGF/Flt1 signaling and HIF-1α. We investigated the angiogenic activity of the loop in an in vivo Matrigel plug assay. The hypoxia-treated conditioned medium induced a strong angiogenic response, as well as the cooption of surrounding vessels into the plugs; ursolic acid inhibited the angiogenesis process. We also found that three other Flt1-expressing neuroblastoma cell lines show hypoxia-mediated drug resistance to etoposide, melphalan, doxorubicin, and cyclophosphamide. Taken together, we conclude that a hypoxia-driven VEGF/Flt1 autocrine loop interacts with HIF-1α through a mitogen-activated protein kinase/ERK1/2 pathway in neuroblastoma. The interaction, in the form of an autocrine loop, is required for the hypoxia-driven cell survival, drug resistance, and angiogenesis in neuroblastoma.
Tumor hypoxia is associated with metastasis, angiogenesis, and resistance to radiation and some chemotherapeutic agents. In tumor cells, hypoxia-responsive pathways modulate well-orchestrated signaling mechanisms that drive cell survival, drug resistance, and angiogenesis (1, 2).
One of the principal regulators of hypoxia-responsive pathways is the hypoxia-inducible factor (HIF-1), which plays a pivotal role in hypoxia-mediated angiogenesis, migration, and invasion. HIF-1 is a transcription factor comprised of α and β subunits, the former being regulated by oxygen concentration. In a hypoxic environment, HIF-1α is activated by phosphorylation and then binds to the HIF-1α-binding site of putative target genes (2). HIF-1α activates expression of a number of genes related to growth and survival that include insulin-like growth factor-2 (IGF-2) and transforming-growth factor-β (TGF-β). IGF-2 and TGF-β bind to the IGF receptor (IGFIR) and epidermal growth-factor receptor (EGFR), respectively, to activate signaling cascades that lead to HIF-1α expression. This initiates a HIF-1α autocrine loop, which may be involved in tumor cell survival and growth in a hypoxic environment (3). HIF-1α also induces vascular endothelial growth factor (VEGF) and its receptor Flt1 (Fms-like receptor 1 or VEGFR1; refs. 2, 3). Recent investigations suggest that Flt1 may play an active role in VEGF-mediated autocrine growth and survival of a diverse group of tumors (4–13). It is possible that similar to TGF-β/EGFR and HIF-1α autocrine loop, VEGF/Flt1 may also form an autocrine loop with HIF-1α. Because VEGF is a highly potent growth factor involved in tumor growth and angiogenesis (14), such an autocrine VEGF/Flt1 and HIF-1α loop may contribute to cell survival, drug resistance, and angiogenesis during hypoxia.
Neuroblastoma is a childhood tumor of neural crest origin in which expression of VEGF and its receptors (Flt1 and VEGFR2) is correlated with high-grade biology (15). Autocrine VEGF/Flt1 activity was reported in a neuroblastoma cell line SK-N-BE, derived from a stage IV neuroblastoma patient (15). We found that the SK-N-BE(2) cell line, which is derived from the same patient after relapse, is resistant to cisplatin compared with several other neuroblastoma cell lines (16). We also found that SK-N-BE(2) cells were resistant to hypoxia-mediated apoptosis.7
B. Das, unpublished data.
Here we report that VEGF/Flt1 autocrine signaling is required for the hypoxia-mediated sustained activation of extracellular signal-regulated kinase 1/2 (ERK1/2), and subsequent survival, drug resistance, and in vivo angiogenesis. Inhibition of VEGF or ERK1/2 reduced HIF-1α activity. Taken together, these findings implicate the self-regulated survival pathway of the VEGF/Flt1 autocrine loop as an important component of the HIF-1α mediated hypoxia-responsive pathway.
Materials and Methods
Cell culture and induction of hypoxia. The human neuroblastoma cell line, SK-N-BE(2) was obtained from American Type Culture Collection (Manassas, VA). NUB-7 cell line was developed and characterized in our laboratory(18). IMR-5 was kindly provided by Dr. Meredith Irwin (Division of Hematology and Oncology, The Hospital for Sick Children); NGP and the LAN15N cell lines were kindly provided by Dr. David Kaplan (Cancer Research Program, Sickkids). Other neuroblastoma cell lines were obtained from American Type Culture Collection. Neuroblastoma cell lines were maintained in α-MEM (Life Technologies, Gaithersburg, MD) supplemented with 10% fetal bovine serum and antibiotics, in a humidified atmosphere of 5% CO2 at 37°C. Radiobiological hypoxic conditions (<0.1% O2) were established in a sealed chamber using the BBL GasPak Plus anaerobic system envelopes with a palladium catalyst (Becton Dickinson, Cockeysville, MD) as previously described (19, 20). Moderate hypoxic conditions (2% O2) were established in a hypoxia incubator (Forma Scientific, Inc., Marietta, OH) where N2 was used to compensate for the reduced O2 level.
Reagents and drugs. Cisplatin [cis-platinum (II) diamine dichloride], etoposide, doxorubicin, vinblastine, melphalan, and ursolic acid were obtained from Sigma (St. Louis, MO). Alamar blue reagent for the cytotoxic assay was obtained from Biosource International (Camarillo, CA). Caspase-3/7 substrate, N-acetyl-DEVD-7-amino-4-methylcoumarin (DEVD-AMC) peptide was obtained from BIOMOL Research Laboratories (Plymouth Meeting, PA). Recombinant human VEGF165 (rhVEGF165), nonspecific goat IgG, and a neutralizing antibody against VEGF and Flt1 (anti-human mouse monoclonal antibody) were purchased from R&D Systems (Minneapolis, MN). Source of other reagents, antibodies, and kits are noted along with the assay descriptions.
Reverse transcriptase-PCR for vascular endothelial growth factor–related receptors. Total RNA was isolated from neuroblastoma cell lines using Trizol (Invitrogen, Burlington, Ontario, Canada), and first-strand cDNA was synthesized using reverse transcriptase M-MLV (Promega, Madison, WI) followed by PCR using a TGradient Thermocycler (Whatman Biometra, Goettingen, Germany) with the following cycling conditions: 35 cycles of 94°C for 60 seconds, 52°C for 60 seconds, and 72°C for 60 seconds for Flt1, NRP-1, NRP-2, and VEGFR3. Thirty-five cycles of 94°C for 30 seconds, 56.0°C for 30 seconds, 72°C for 55 seconds for VEGFR2. Two rhabdomyosarcoma cell lines RH4 and RH28 were used as a positive control for Flt1 and VEGFR2, respectively, as reported previously (21). HeLa cell line was used as a negative control. The quality of the RNA was evaluated by reverse transcriptase-PCR (RT-PCR) for glyceraldehyde-3-phosphate dehydrogenase. Primers were used as previously described (22).
Alamar blue cytotoxicity assay. The assay was done as previously described (16, 23). Briefly, cells grown in 24-multiwell plates (Costar, Cambridge, MA) were treated with each drug and incubated with 10% Alamar blue solution. Relative fluorescence units (RFU) were obtained using the Gemini Spectra MAX microplate reader (emission, 540 nm; excitation, 590 nm; cutoff, 570 nm). Cell survival was expressed as the percent of RFU normalized to cells grown in vehicle (DMSO or ethanol). Experiments were done in triplicates.
Western blotting. Protein extracts were obtained using the cell lysis buffer (Cell Signaling Technology, Beverly, MA) mixed with protease inhibitor cocktail (Mini Pill, Roche, Nutley, NJ). For whole cell lysate preparation and HIF-1α detection, additional chemicals were used as described (26). Fifty micrograms of the clear protein lysate was resolved on 8% to 12% SDS-PAGE gels and transferred to 0.45-μm nitrocellulose membranes. The blocked membranes were incubated overnight with the following antibodies: bcl-2, Bax, XIAP, Survivin (Cell Signaling Technology); human HIF-1α (BD Transduction Laboratory, San Diego, CA); β-actin (Sigma); active mitogen-activated protein kinase (MAPK)/ERK1/2, ERK1/2 (Promega); Flt1 (Chemicon International, Temecula, CA); basic fibroblast growth factor (bFGF), phospho-tyrosine kinase, and vinculin (Santa Cruz Biotechnology, Santa Cruz, CA). Immunoblotting was detected by enhanced chemiluminescence (Amersham Bioscience, Buckinghamshire, United Kingdom) according to the manufacturer's instructions. Densitometry study was done using the TotalLab software (Amersham Scientific, Oakville, Ontario, Canada).
Immunohistochemistry. Immunohistochemistry was done using a Ventana diaminobenzidine kit (Ventana Medical Systems, Tucson, AZ) in an automated stainer according to manufacturer's instructions. Flt1 antibody (Santa Cruz Biotechnology) was used in 1:100 dilution.
Human vascular endothelial growth factor immunoassay. The assay was done as described (29). Briefly, after each treatment, the medium was collected and VEGF was measured by quantitative sandwich enzyme immunoassay technique using a Quantikine kit (R&D Systems) according to the manufacturer's instructions and normalized to cell number (ng/106 cells) using the trypan blue exclusion method of viable cell counting.
Small interfering RNA. All the small interfering RNA (siRNA) reagents including transfection reagents were obtained from Dharmacon Research, Inc. (Chicago, IL). The Flt1 siRNA SMARTpool reagent was used to knockdown Flt1 expression in neuroblastoma cells. In vitro transfection was done by using the DharmaFECT transfection reagent. Nontargeting siRNAs (siCONTROL) was used as a siRNA control. Forty percent confluent cells in 24-well plates were treated with 100 nmol/L siRNA, according to manufacturer's instructions, and the hypoxia experiments were conducted 48 hours after transfection.
Flt1 overexpression. The VEGFR-1 expression vector, human full-length VEGFR-1 cDNA in BCMGS neo vector (30) and pcDNA3/GFP plasmid were used to transfect the IMR-5 cell line using LipofectAMINE 2000 transfection reagent (Invitrogen Canada, Inc., Burlington, Ontario, Canada).
In vivo angiogenesis assessment by the Matrigel plug assay. Tumor angiogenesis was quantified in a Matrigel plug by the FITC-dextran perfusion method as previously described (31) with some modifications. Briefly, the conditioned medium (200 μL) of normoxia/hypoxia-treated cells was mixed with 600 μL aliquot of Matrigel (BD Bioscience, San Jose, CA) and immediately injected s.c. into both flanks of 6- to 8-week-old female BALB/c nude mice (Charles River Laboratory, Montreal, Canada). On day 10, 0.25 mL of 25 mg/mL FITC-dextran (Sigma) in PBS was injected, and an hour later, blood samples (300 μL) and Matrigel plugs were collected. The right flank plugs were homogenized, and the supernatant read for fluorescence as described (31). Angiogenic response was expressed as a ratio of Matrigel plug fluorescence/plasma fluorescence. The left flank plugs were fixed in formalin, paraffin embedded, and 5-μm sections stained with H&E for histology. Experiments were conducted according to the CCAC guidelines and approved by the Hospital for Sick Children's Animal Research Committee.
Statistical analysis. The results of the Alamar blue assay were analyzed by nonlinear regression analysis. The results of Caspase 3/7, VEGF and Matrigel-FITC assays were analyzed using Student's t test. Values were expressed as mean ± SE unless otherwise specified. Statistical significance was set at P < 0.05.
Vascular endothelial growth factor/Flt1 autocrine pathway is active in SK-N-BE(2) cells. A RT-PCR study reported that SK-N-BE cells [the parent cell line of SK-N-BE(2)] express Flt1 receptor (15). We did RT-PCR and immunoprecipitation to investigate the presence of active VEGF receptors in the SK-N-BE(2) cell line. Flt1 receptor is expressed, and upon rhVEGF treatment, the receptor is autophosphorylated (Fig. 1A-C). VEGF-Flt1 autocrine signaling has been found to activate the MAPK/ERK pathway (32, 33). We showed that activation of ERK1/2 by rhVEGF is inhibited by a neutralizing antibody against VEGF (Fig. 1D). The RT-PCR analysis also confirmed that VEGFR2 (Fig. 1A) is absent in SK-N-BE(2) cells. However, Neuropilin-1, Neuropilin-2, and VEGFR3 mRNA were present (data not shown).
An in vitro model of hypoxia-induced drug resistance to study vascular endothelial growth factor/Flt1 system. To evaluate the role of VEGF/Flt1 autocrine signaling in hypoxia response pathway, we first developed an in vitro model of hypoxia-induced drug resistance, where cells were exposed to hypoxia (<0.1% or 2% O2) in the presence of serum to avoid glucose deprivation or cell cycle inhibition. We found that SK-N-BE(2) cells treated with 24 hours of severe hypoxia (<0.1% O2 with 10% serum) did not experience cytotoxicity/cell cycle arrest (data not shown) or apoptosis (Fig. 2A). We then treated the hypoxic cells with three cytotoxic drugs, etoposide, melphalan, and cisplatin for the subsequent 24 hours under normoxia. Hypoxia treatment reduced the etoposide (5 μmol/L) cytotoxicity by 26% (P = 0.004; Fig. 2B) and caspase-3/7 activity by 40% (P < 0.05; Fig. 2C). Western blot analysis of cleaved poly(ADP-ribose) polymerase confirmed the etoposide resistance in hypoxia (Fig. 2D). Because antiapoptotic proteins XIAP (X-linked inhibitor of apoptosis) and survivin inhibit caspase-3/7 (34) and etoposide-mediated caspase-3/7 activity (35), respectively, we examined the possible up-regulation of these two proteins during hypoxia. The expression of XIAP and survivin did not change significantly in normoxia versus hypoxia (data not shown).
Moderate hypoxia (2% hypoxia) of 24 hours duration without serum deprivation failed to show drug resistance (data not shown). In addition, we did not find hypoxia-induced resistance against cisplatin and melphalan (Fig. 2B). Nonetheless, our results suggest that severe hypoxia alone could induce resistance to etoposide-induced apoptosis independent of glucose deprivation or cell cycle arrest (data not shown). Hence, we used this model of hypoxia-induced resistance to apoptosis to investigate the VEGF/Flt1 autocrine system during hypoxia.
Vascular endothelial growth factor/Flt1 system is involved in hypoxia-mediated cell survival and drug resistance. Etoposide induces apoptosis in the HepG2 cell line (a hepatocellular carcinoma cell line) through bcl-2/Bax-dependent mechanisms (36). Erler et al. found that hypoxia (0.5% O2) can induce resistance against etoposide by down-regulating bcl-2/Bax ratio in the HepG2 cell line (37). Beierle et al. found that VEGF165 up-regulates bcl-2 protein levels in neuroblastoma resulting in resistance to tumor necrosis factor-α–mediated apoptosis (38). Hypoxia up-regulates HIF-1α, which in turn up-regulates VEGF and Flt1 (2). Therefore, we hypothesized that HIF-1α-mediated induction of VEGF/Flt1 autocrine signaling led to bcl-2 up-regulation and subsequent survival and etoposide resistance during hypoxia. To investigate this possibility, SK-N-BE(2) cells were treated with the neutralizing anti-VEGF antibody and hypoxia followed by exposure to etoposide. The antibody treatment completely reversed the hypoxia-mediated drug resistance (Fig. 3A). In addition, anti-VEGF antibody treatment reduced survival of SK-N-BE(2) cells during hypoxia by 28% (P = 0.008; Fig. 3A). We then investigated the bcl-2/Bax ratio in normoxia versus hypoxia. Bcl-2 protein level was significantly up-regulated in hypoxia versus normoxia, whereas the level of Bax protein did not change significantly. Densitometry revealed a several fold increase of bcl-2/Bax ratio in hypoxia versus normoxia (Fig. 3B). The anti-VEGF treatment significantly down-regulated bcl-2 level under hypoxia (Fig. 3C). We also confirmed that hypoxia treatment induced HIF-1α and up-regulated Flt1 (data not shown). The level of VEGF165 in the conditioned medium of hypoxia-treated cells increased by >7-fold reaching ∼10 ng/million cells/mL (Fig. 3D). Because SK-N-BE(2) cells expressed VEGR3 and NRP-1, we used Flt1 siRNA knockdown to investigate the exclusive role of Flt1 in VEGF-mediated survival and drug resistance during hypoxia. siRNA knockdown of Flt1 resulted in severe toxicity during hypoxia (64% increase of toxicity; P < 0.0126; Supplementary Fig. S1A-C). We then investigated the hypoxia-mediated resistance against etoposide in a Flt1-overexpressing IMR-5 cell line. Flt1-overexpressed cells were resistant to hypoxia-induced toxicity (46% reduction of toxicity compared with non-Flt1 transfected cells; P < 0.03). Flt1 overexpression resulted in 35% reduction of etoposide toxicity during hypoxia (P < 0.05; Supplementary Fig. 1D-E).
The above results suggest that hypoxia up-regulates VEGF/Flt1 signaling leading to bcl-2 up-regulation and subsequent cellular survival during hypoxia. Hence, treatment with rhVEGF should recapitulate the hypoxia-mediated bcl-2 up-regulation and drug resistance. SK-N-BE(2) cells were treated with rhVEGF (100 ng/million cells/mL) for 24 hours followed by etoposide treatment. rhVEGF treatment failed to induce resistance against etoposide-induced cytotoxicity (Fig. 3E). Moreover, 24 hours of rhVEGF treatment did not change bcl-2 levels (data not shown). We then increased the duration of rhVEGF treatment for 2 to 3 days but failed to see changes in bcl-2 expression or etoposide-induced toxicity (data not shown). These results suggest that VEGF/Flt1 signaling alone was not sufficient to induce hypoxia-mediated up-regulation of bcl-2 and resistance against etoposide.
A hypoxia-driven vascular endothelial growth factor/Flt1 autocrine loop that sustains extracellular signal-regulated kinase 1/2 activity during hypoxia. Our findings that hypoxia alone and not rhVEGF was able to up-regulate bcl-2 in SK-N-BE(2) cells suggest differential signaling of VEGF/Flt1 in normoxia versus hypoxia. Recently, hypoxia and/or rhVEGF treatment has been found to up-regulate bcl-2 by activating MAPK/ERK1/2 pathway in serum-deprived hepatocellular carcinoma cells (39). Sustained activation of MAPK/ERK has been suggested to play an important role in cell survival (40). Yokoi et al. reported hypoxia-induced sustained activation of ERK1/2 in a pancreatic cell line where hypoxia induced resistance against gemcitabine (1). We investigated the activation of phospho-ERK1/2 (p-ERK1/2) by rhVEGF or hypoxia. rhVEGF treatment resulted in a transient activation of p-ERK1/2 that lasted for <10 minutes (Fig. 4A), whereas hypoxia resulted in a sustained increase of p-ERK1/2 levels that reached its peak within the first 3 hours of hypoxia and then continued for >6 hours after hypoxia (Fig. 4B). In addition, anti-VEGF antibody markedly reduced hypoxia-mediated ERK1/2 activation (Fig. 4B). To further investigate the role of sustained ERK1/2 activation on bcl-2 up-regulation and drug resistance, we used a specific inhibitor of MAPK/ERK, U0126, and a nonspecific inhibitor, ursolic acid, a dietary triterpenoid found to inhibit ERK1/2 (41). Both U0126 and ursolic acid inhibited hypoxia-mediated sustained ERK1/2 activation, reduced the cellular survival during hypoxia, and reversed the hypoxia-induced drug resistance in a dose-dependent manner (Fig. 4C-E). Furthermore, both U0126 and ursolic acid reduced the hypoxia-induced up-regulation of bcl-2 protein (Fig. 4F). Our findings on ursolic acid are interesting as it is a dietary compound with potential antitumor therapeutic applications.
Vascular endothelial growth factor/Flt1-induced sustain extracellular signal-regulated kinase 1/2 activation regulates hypoxia-inducible factor-1α activity. We found that anti-VEGF antibody treatment led to reduction of hypoxia-mediated bcl-2 up-regulation (Fig. 3C) and ERK1/2 phosphorylation (Fig. 4B). We also found that anti-Flt1 antibody treatment reduces hypoxia-mediated ERK1/2 phosphorylation (data not shown). In addition, we found that hypoxia up-regulated HIF-1α, Flt1, and VEGF (Fig. 3D). These results suggest that HIF-1α up-regulated VEGF/Flt1 signaling leading to sustain activation of ERK1/2. Interestingly, ERK1/2 has been found to directly phosphorylate the COOH-terminal domain of HIF-1α, the latter being required for the transcriptional activity of HIF-1α (26). Therefore, we investigated the HIF-1α phosphorylation status in the presence of ERK1/2 inhibitors, ursolic acid and U0126, by Western blots as described previously (26). As expected, the ursolic acid and U0126 treatment groups showed marked reduction of HIF-1α phosphorylation (Fig. 5A). Because HIF-1α activation is upstream of VEGF and Flt1 up-regulation, the HIF-1α phosphorylation should be independent of anti-VEGF treatment. However, we found that anti-VEGF antibody treatment also reduced HIF-1α phosphorylation (Fig. 5A). This result suggests that the VEGF/Flt1 autocrine loop may regulate HIF-1α activity. Therefore, we hypothesized that during the initial phase of hypoxia, VEGF/Flt1 signaling activates ERK1/2 that activates HIF-1α resulting in increased activity of VEGF/Flt1, eventually forming an autocrine loop (Fig. 6). In such a case, inhibition of ERK1/2 should lead to the down-regulation of VEGF. As expected, the ursolic acid treatment group showed marked reduction of VEGF secretion (Fig. 5B). Most importantly, ursolic acid did not inhibit the basal VEGF secretion, suggesting that ursolic acid–mediated inhibition of VEGF secretion was directly related to HIF-1α-related induction of VEGF secretion. To further study the loop (Fig. 6), we investigated the expression of bFGF, a potent angiogenic growth factor and one of the important downstream products of both HIF-1α and VEGF systems (42, 43). We found that hypoxia induced an increase in bFGF protein level, which was down-regulated by ursolic acid (Fig. 5C). The normoxia level of bFGF remained unaffected by ursolic acid treatment (data not shown). These results suggest that during hypoxia, VEGF/Flt1 autocrine system may cooperate with HIF-1α activity in an autocrine loop that up-regulates important survival and angiogenic factors such as bcl-2 (Fig. 3B-C) and b-FGF (Fig. 5C), respectively. To analyze the potential angiogenic effects of up-regulated VEGF/Flt1 and HIF-1α autocrine loop in vivo, we applied a murine Matrigel plug assay to hypoxia-treated conditioned media (details in Materials and Methods). As expected, hypoxia markedly increased in vivo Matrigel plug perfusion compared with normoxic medium (Fig. 7A). Interestingly, microscopic examination of the plugs revealed numerous host capillaries spreading into the plug (Fig. 7C). In its initial stage of progression, a tumor mass remains in a severely hypoxic microenvironment, where, as in the early stage of wound healing, new capillaries sprout from adjacent vessels to invade the tumor mass, a process called cooption (44). Therefore, we did H&E and Factor VIII staining of the Matrigel plugs to study vessel cooption. In the H&E staining, the plugs containing normoxic-conditioned medium showed an intact margin and no cooption (Fig. 7E), whereas the plugs containing hypoxic-conditioned medium showed a broken margin and invasion of numerous capillary-like structures deep into the plugs (Fig. 7F). Factor VIII staining confirmed the presence of endothelial cells lining these capillaries (data not shown). The ursolic acid–treated group showed dramatic reduction of capillary infiltration into the plug (Fig. 7D and G) and a dramatic reduction (>80% reduction) in the perfusion (Fig. 7A).
Evidence of vascular endothelial growth factor/Flt1 autocrine loop in other neuroblastoma cell lines. SK-N-BE(2) is a post-chemotherapy cell line expressing mutant p53. Because tumor hypoxia may select p53 mutant cells (45), it was possible that the VEGF/Flt1 and HIF-1α autocrine loop may be a p53-dependent phenomenon. In addition, we did not find hypoxia-induced resistance against cisplatin and melphalan in the SK-N-BE(2) cell line. Therefore, we investigated hypoxia-mediated survival in 15 neuroblastoma cell lines. We found that only five cell lines (NBL-S, NUB-7, SH-5YSY, SHEP, and SKNDZ) other than SK-N-BE-2 could tolerate 24 hours of hypoxia (0.1% oxygen). RT-PCR reveals that all these lines except SHEP express Flt1 receptor (Supplementary Fig. 2A). We investigated hypoxia-mediated drug resistance in NBL-S, NUB-7, and SH-5YSY. The results are summarized in a table (Supplementary Table 1). All three cell lines show relative resistance to cylcophosphamide (4-hydroxy cyclophosphamide), doxorubicin, melphalan, and etoposide. To investigate the role of a VEGF/Flt1 autocrine system, NBL-S and NUB-7, both expressing wild-type p53 were treated with anti-Flt1 antibody. The antibody treatment significantly reduced survival during hypoxia by >40% (P < 0.05; data not shown). To investigate whether Flt1 expression is a common feature of neuroblastoma, we examined 10 primary neuroblastoma cases for Flt1 expression by immunohistochemistry. Five of 10 tumors expressed Flt1 (Supplementary Fig. 2B).
Flt1, a “fms-like tyrosine kinase” receptor was initially thought to play the role of a decoy receptor of VEGF (14). However, subsequent research suggested that Flt1 may play an active role in VEGF-mediated autocrine signaling of tumor growth and invasion (5, 9, 12, 13). Here we report a hypoxia-driven VEGF/Flt1 autocrine signaling and its interaction with HIF-1α in neuroblastoma.
VEGF/VEGFRs signaling have been primarily studied in endothelial cells. In this context, VEGFR2 has been defined as the dominant signaling receptor for VEGF-induced mitogenesis and vascular permeability because of its strong tyrosine kinase activity. In contrast, Flt1 has weak tyrosine kinase activity, yet a 10-fold higher binding affinity for VEGF. This led to the suggestion that Flt1 acts as a decoy receptor, capable of being a positive or negative regulator of angiogenesis, through its ability to sequester VEGF from signaling through VEGFR2 (14). However, recent studies of Flt1 in several tumor cells, including prostate, colon, pancreas, glioblastoma, lymphoma, leukemia, and mesothelioma suggest that VEGF/Flt1 signaling has autocrine survival activity (4–13). In several colon carcinoma cell lines, autocrine VEGF/Flt1 promotion of cell survival was dominant over any other soluble growth factor signal. In addition, Flt1 expression correlated with malignant transformation and subsequent progression of human colon carcinoma (12). In the prostate carcinoma cell line FB2, which is the malignant counterpart of the rat prostate epithelial cell line NbE-1, Flt1 expression was correlated with malignant transformation and lung metastasis (13). In stage I and II neuroblastoma, Flt1 mRNA was detected in two of nine tumors, whereas in stage IV tumors, Flt1 mRNA was detected in 7 of 10 samples suggesting the role of Flt1 in neuroblastoma progression (15). We found that Flt1 mRNA is present in 8 of 15 cell lines and 5 of 10 primary neuroblastoma tumors. We found that VEGF/Flt1 signaling plays a significant role in neuroblastoma cell survival during hypoxia. The inhibition of Flt1 activity by siRNA knockdown significantly reduced SK-N-BE(2) cell survival during hypoxia, whereas overexpression of Flt1 in the IMR-5 cell line significantly increased survival during hypoxia. The VEGF/Flt1 activity was associated with up-regulation of bcl-2, an important antiapoptotic protein, and bFGF, a potent angiogenic factor. Most importantly, we found that inhibition of VEGF/Flt1 system led to the reduction of HIF-1α phosphorylation. This suggests a strong interaction between the VEGF/Flt1 autocrine loop and HIF-1α activation during initial hours of hypoxia. Based on our findings, here we propose an autocrine loop between VEGF/Flt1 signaling and HIF-1α. In this loop. VEGF/Flt1 first activates ERK1/2 during the initial period of hypoxia. ERK1/2 in turn activates HIF-1α leading to the up-regulation of VEGF, and Flt1, which further stimulates the ERK1/2 system (Fig. 6). Flt1 is the only receptor associated with VEGF that contains a binding site for HIF-1α (2), which further supports the recruitment of VEGF/Flt1 system in HIF-1α-driven tumor progression. Recently, Tang et al. showed the interaction between a hypoxia-driven VEGF/VEGFR2 autocrine loop and HIF-1α in endothelial cells to regulate early angiogenic process including cooption-like events (46). Our results suggest that the VEGF/Flt1 and HIF-1α loop may be involved in cooption in neuroblastoma (Fig. 7F). In addition, Flt1, being involved in endothelial cell migration (47), may also help to stimulate migration of tumor cells towards the nearby capillaries. Resistance to apoptosis and increased angiogenesis are two important hallmarks of tumor progression (48), and tumor hypoxia has been known to play an important role in both angiogenesis and the selection of apoptosis resistant cells (2). In this context, our findings of a hypoxia-driven VEGF/Flt1 autocrine survival pathway and its close cooperation with HIF-1α is significant in terms of tumor angiogenesis and drug resistance.
Studies have shown that in vitro exposure of cancer cell lines to hypoxia (24-72 hours) may lead to resistance against multiple chemotherapeutic agents (49). Earlier studies focused on hypoxia-induced cell cycle arrest and subsequent decreased sensitivity to chemotherapeutic agents as the mechanism of hypoxia-mediated drug resistance (50). However, recent investigations suggest that hypoxia-driven resistance may involve specific cellular survival pathways independent of cell cycle arrest. Erler et al. found that colon carcinoma cells develop resistance against etoposide-induced apoptosis via up-regulation of HIF-1α and down-regulation of Bid and Bax (37). We found that hypoxia decreases toxicity of several chemotherapeutic agents (etoposide, melphalan, cyclophosphamide, and doxorubicin) to neuroblastoma (Supplementary Table 1). We also found that HIF-1α and VEGF/Flt1 autocrine loop may play an important role in the hypoxia-mediated drug resistance. Neuroblastoma is the most common solid tumor of childhood, and a subset of older patients frequently succumb to metastatic disease despite even the most intensive chemotherapy (51). Metastatic and drug resistance neuroblastoma secretes high level of VEGF and other angiogenic factors (52). Our findings suggest that targeting VEGF-mediated autocrine activity may reduce drug resistance and thus increase the chemotherapeutic efficacy in metastatic neuroblastoma.
We previously showed that the combination of low-dose vinblastine and VEGFR2 antibody led to significant antiangiogenic response and an increase in neuroblastoma cell apoptosis (31). Effective blockade of the VEGF pathway by the neutralizing antibody, bevacizumab (avastin) has been shown to reduce tumor progression in colorectal cancer (53) and leukemia (6). Findings of the present work suggest that apart from its antiangiogenic activity, the anti-VEGF treatment strategy may also target hypoxia-mediated survival and drug resistance thus increasing the efficacy of the chemotherapeutic agents and reducing tumor burden. Anti-VEGF therapy may be combined with novel agents that target the hypoxia pathway to achieve greater tumor regression. Interestingly, ursolic acid has been shown to exert inhibitory activity against multiple kinase pathways including nuclear factor-κB, activator protein, protein kinase C, and ERK1/2 (41, 54). We found that ursolic acid reduces cellular survival during hypoxia by targeting VEGF/Flt1 related ERK1/2 activation. Ursolic acid, being a dietary compound, may be further explored for such uses.
In summary, we report a novel autocrine loop between the hypoxia-driven VEGF/Flt1 signaling and HIF-1α activity. We suggest that the loop serves the tumor cells to survive hypoxia, cytotoxic drugs such as etoposide, and recruits surrounding vessels (cooption) for the early phase of angiogenesis. We have shown that a naturally occurring molecule, ursolic acid, can inhibit this autocrine loop. Further understanding of the loop may help to decipher the early stages of tumor progression/relapse in neuroblastoma and could lead to more effective treatment strategies.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Grant support: James Birrell Neuroblastoma Research Fund; B.R.A.I.N Child (Meagan Bebenek) fund; Toronto and the Andrew Mizzoni Cancer Research Fund, The Hospital for Sick Children, Toronto, Canada; Hospital for Sick Children's Research Training Centre, Canada (B. Das, R. Tsuchida, and M.F.W. Gee); and National Cancer Institute of Canada Fellowship with funds from the Terry Fox Foundation (R. Tsuchida).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Drs. Robert Kerbel, David Kaplan, Meredith Irwin, Maria Zielenska, and Diana Stempak for critical review of the article; Michael Ho and Jean-François Lavoie for technical assistance; and the staff of the animal facility of Hospital for Sick Children for their technical support.