Hepatocellular carcinoma (HCC) is one of the most common tumor-related causes of death worldwide for which there is still no satisfactory treatment. We previously reported the antiangiogenic effect of gefitinib, a selective epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor that has been used successfully to treat lung cancer. In this study, we investigated the effects of gefitinib on tumor-induced angiogenesis by using HCC cell lines (HCC3, CBO12C3, and AD3) in vitro as well as in vivo. Oral administration of gefitinib inhibited angiogenesis induced by HCC3 and CBO12C3, but not by AD3 in the mouse dorsal air sac model. Production of both vascular endothelial growth factor (VEGF) and chemokine C-X-C motif ligand 1 (CXCL1) by EGF-stimulated HCC was more markedly inhibited by gefitinib in HCC3 and CBO12C3 cells than in AD3 cells. EGF stimulated the phosphorylation of EGFR, Akt, and extracellular signal-regulated kinase 1/2 (ERK1/2) in HCC3 and CBO12C3 cells, whereas EGF stimulated phosphorylation of EGFR and ERK1/2, but not Akt in AD3 cells. In fact, Akt was constitutively activated in the absence of EGF in AD3 cells. Gefitinib inhibited Akt phosphorylation in all three cell lines, but it was about five times less effective in AD3 cells. The concentration of PTEN in AD3 cells was about a half that in HCC3 and CBO12C3 cells. Transfection of HCC3 cells with PTEN small interfering RNA reduced their sensitivity to gefitinib in terms of its inhibitory effect on both Akt phosphorylation and the production of VEGF and CXCL1. In conclusion, effect of gefitinib on HCC-induced angiogenesis depends on its inhibition of the production of angiogenic factors, probably involving a PTEN/Akt signaling pathway. (Cancer Res 2006; 66(10): 5346-53)
Hepatocellular carcinoma (HCC) is the fifth most common malignancy worldwide (1) and the prognosis for HCC patients is still very poor. Members of the epidermal growth factor receptor (EGFR) family have emerged as critical factors in the development and growth of various types of cancer, including HCC (2). These receptors are part of a complex network of signal transduction cascades that modulate tumorigenic processes, such as proliferation, adhesion, migration, differentiation, angiogenesis, and escape from apoptosis (3–6). High EGFR expression in human cancers is often correlated with advanced disease, metastasis, and poor clinical outcome, for example in non–small cell lung carcinoma (NSCLC), breast, cervical, and head and neck carcinomas (7–9).
Gefitinib (Iressa, ZD1839) is an orally active and selective EGFR tyrosine kinase inhibitor that blocks EGFR-mediated signal transduction pathways involved in cancer growth (10, 11). Gefitinib has antiproliferative activity in various human cancers in vivo as well as in vitro (12, 13). It is now approved as a monotherapy for patients with locally advanced or metastatic NSCLC (14–16) and is under investigation for the treatment of prostate, breast, head and neck, gastric, and colorectal cancer (17). An important recent discovery is the close association between a clinical response to gefitinib in patients with NSCLC and somatic mutations in the EGFR gene (18, 19). In NSCLC cells carrying such mutations, gefitinib treatment markedly inhibited phosphorylation of EGFR and its downstream signaling kinases, Akt and extracellular signal-regulated kinase 1/2 (ERK1/2). Consistent with this observation, we have reported that the sensitivity of NSCLC cell lines to gefitinib under basal growth condition is closely correlated with their dependence for survival and proliferation on Akt and ERK1/2 activation in response to EGFR signaling (20). Cappuzzo et al. (21) have reported that patients with phosphorylated Akt–positive tumors who received gefitinib had a better response rate and time to progression. Moreover, increased EGFR gene copy number evaluated by fluorescent in situ hybridization (FISH) was significantly associated with higher response rates and lower progression rates in lung cancer (22) and in colon cancer (23).
HCC is a typical hypervascular tumor and tumor angiogenesis is a prerequisite for both its growth and metastasis (24, 25). Angiogenesis and vascular invasion are common characteristics of malignant tumors in patients with HCC (26, 27). Enhanced expression of vascular endothelial growth factor (VEGF) and chemokine C-X-C motif ligand 1 (CXCL1) is often seen both in HCC cells in culture and at sites of angiogenesis in the livers of HCC patients (26, 28–33). Ishikawa et al. (34) have previously reported antitumor effect by antiangiogenesis gene therapy using angiostatin gene on HCCs in the xenograft model. A recent study by Liu et al. (35) has also shown that administration of a VEGFR tyrosine kinase inhibitor induced both antitumor and antiangiogenesis effects against HCCs in the xenograft model. On the other hand, we have previously reported the inhibition of EGF-induced migration of vascular endothelial cell and neovascularization in mouse corneas by gefitinib, suggesting that its antitumor effect may be mediated in part through its antiangiogenic activity (36, 37). Overexpression of EGFR was also observed in 60% to 85% of tumor tissue of HCC (38), and a phase II trial of gefitinib has been proceeding against patients with advanced unresectable HCC in the United States. In this study, we asked whether EGFR inhibition could modulate tumor angiogenesis induced by HCC cells and examined possible mechanisms underlying the effects of gefitinib.
Materials and Methods
Cell culture and reagents. Three murine HCC cell lines HCC3, CBO12C3, and AD3 were established as described previously (39, 40) and were cultured in William's E (Life Technologies, Inc., Grand Island, NY) supplemented with 10 ng/mL EGF, 10% fetal bovine serum (FBS), 320 mg/L l-glutamine, and 2 g/L glucose, at 37°C in 5% CO2 in a humid environment. Gefitinib was provided by Astra Zeneca Pharmaceuticals (Macclesfield, United Kingdom) and was dissolved in DMSO for in vitro studies as described previously (20, 27). LY294002, a selective inhibitor of phosphatidylinositol 3-kinase (PI3K), and U0126, an ERK inhibitor, were purchased from Sigma Chemical Co. (St. Louis, MO). Recombinant EGF, anti-phospho-EGFR (Tyr1173), anti-EGFR, and anti-PI3K p85 antibodies were purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). Anti-phospho-Akt (Ser473), anti-Akt, anti-phospho-ERK1/2 (Thr202/Tyr204), and anti-ERK1/2 antibodies were purchased from Cell Signaling Technology (Beverly, MA). Anti-PTEN and anti-PI3K p110 antibodies were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-β-actin antibody was from Abcam, Ltd. (Cambridge, United Kingdom).
Cell proliferation assay. HCC cells were suspended at 5.0 × 103/mL in medium with or without 10 ng/mL EGF and seeded into 48-well plates for the indicated periods. The following day, cell numbers were counted using a Z1 Series Coulter Counter (Beckman Coulter, Inc., Fullerton, CA).
Mouse dorsal air sac assay. This assay was carried out in 7- to 10-week-old male mice as previously described (41). Male BALB/c mice were obtained from Kyudo Co., Ltd. (Fukuoka, Japan). HCC cells (2 × 106) were suspended in 150 μL PBS and injected into a chamber that consisted of a ring (Millipore Corp., Bedford, MA) covered with Millipore filters (0.45 μm pore size) on each side. This was implanted into an air sac produced by injecting 10 mL of air s.c. on the back of an anesthetized mouse (50 mg/kg pentobarbital, i.v.) on day 0. Gefitinib was administered p.o. from day 0 to day 4. On day 5, the chambers were removed from the s.c. fascia and replaced with black rings of the same inner diameter as the chambers. Photographs of these sites were assessed by counting the number of newly formed vessels >3 mm in length within the area of the rings.
Quantification of VEGF and CXCL1 in conditioned medium. The concentrations of VEGF and CXCL1 in conditioned medium from HCC cells were measured using ELISA kits (R&D Systems, Minneapolis, MN) as described previously (42). Briefly, HCC cells were seeded in 24-well dishes at 2.5 × 104/2 mL/well and, when subconfluent, the medium was replaced with serum-free medium for 24 hours, containing different concentrations of kinase inhibitors, with or without 10 ng/mL EGF at 37°C. Results are presented as means ± SD.
Western blot analysis. After culture for 12 hours in serum-free medium without EGF, HCC cells were treated with the indicated concentration of gefitinib for 3 hours and then stimulated with 10 ng/mL EGF for 15 minutes at 37°C. After rinsing with ice-cold PBS, the cells were lysed in Triton X-100 buffer [50 μmol/L HEPES (pH 7.4), 150 mmol/L NaCl, 1% Triton X-100, 10% glycerol, 1 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin, 10 μg/mL leupeptin, and 1 mmol/L Na3VO4]. Lysates were subjected to SDS-PAGE and transferred to Immobilon membranes (Millipore). Blots were incubated with primary antibodies and visualized with a secondary antibody coupled to horseradish peroxidase (Amersham, Piscataway, NJ), and enhanced chemiluminescence system, as described previously (20). Bands on Western blots were analyzed densitometrically using Scion Image software (version 4.0.2; Scion Corp., Frederick, MD).
PTEN gene silencing using small interfering RNA. The ability of 25-nucleotide duplexes of RNA to interfere with PTEN expression was tested. The specific small interfering RNA (siRNA) sequence used was nucleotide 58 to 82, relative to the first nucleotide of the start codon, of PTEN (Genbank accession number NM_008960). Mock siRNA duplexes were obtained from Invitrogen (Carlsbad, CA). siRNA duplexes were transfected using LipofectAMINE 2000 and Opti-MEM medium (Invitrogen) according to the recommendations of the manufacturer. Reduction in PTEN expression was confirmed by Western blot analysis.
FISH analysis. FISH analysis was done as described previously (43). Mouse EGFR cDNA (BC-023729-EGFR) and Mere Mouse BAC clone 43L24 were obtained from Open Biosystems (Huntsville, AL). The probes for EGFR and chromosome 11 were labeled with digoxigenin and biotin by nick translation methods, respectively. The slides were incubated at 75°C for 10 minutes to codenature the EGFR and chromosome 11 probes and allowed to hybridize overnight at 37°C. After stringency wash, digoxigenin-labeled EGFR and biotin-labeled chromosome 11 probes were detected using antidigoxigenin-Cy3 and anti-biotin-Cy5. The chromatin was counterstained with 4′,6-diamino-2-phenylindole. Analysis was done with a Leica DMRA2 fluorescence microscope (Deerfield, IL) equipped with the Leica CW4000 FISH software. Average numbers of chromosome were determined by scoring 30 to 40 metaphase spreads.
Statistical analysis. Statistical analysis used the Mann-Whitney U test and Student's t test, with P < 0.05 considered statistically significant, and was done using JMP 5.01 software (SAS Institute, Inc. Cary, NC).
Effect of EGF on the cell growth of HCC cell lines. The growth rates of HCC cell lines were measured under normal growth conditions with 10% FBS, in the presence or absence of 10 ng/mL EGF. Doubling times for HCC3, CBO12C3, and AD3 were 13.5, 12.1, and 12.6 hours, respectively, in the presence of EGF and 22.4, 14.6, and 14.6 hours, respectively, in the absence of EGF (Fig. 1). Growth rates for the three lines were comparable, although EGF affected the growth of HCC3 cells rather more than that of the other two cell lines.
Gefitinib inhibits tumor-induced angiogenesis in vivo. We next investigated whether gefitinib could inhibit angiogenesis induced by HCC cells in vivo, using the dorsal air sac assay. In the absence of gefitinib, implantation of chambers containing each of the three HCC cell lines resulted in the development of microvessels with thin curled structures and tiny bleeding spots, in addition to the preexisting vessels (Fig. 2A), consistent with our previous studies (41). The oral administration of gefitinib at 150 mg/kg/d markedly reduced the development of microvessels induced by HCC3 and CBO12C3 cells, but not by AD3 cells (Fig. 2A). About a 60% reduction in angiogenesis was seen in mice implanted with HCC3 cells and given 75 mg/kg/d gefitinib and a similar reduction was seen in mice implanted with CBO12C3 cells and given 150 mg/kg/d gefitinib (Fig. 2B).
Inhibitory effect of gefitinib on production of VEGF and CXCL1 in HCC cells. To understand how gefitinib modulates angiogenesis, we examined the effect of this EGFR-selective drug on the production of two potent inducers of angiogenesis, VEGF and CXCL1 [interleukin-8 (IL-8) homologue/KC/Gro-α] in HCC cell lines (Fig. 3A). EGF increased the production of both VEGF and CXCL1 by 1.6- to 3.2-fold in all three cell lines. At 0.1 to 5.0 μmol/L, gefitinib inhibited EGF-stimulated production of VEGF and CXCL1 by HCC3, CBO12C3, and AD3 to different extents. At 1 μmol/L, gefitinib blocked the production of CXCL1 by >80% in HCC3 and CBO12C3 cells and by 60% in AD3 cells in the presence of EGF. At the same concentration, gefitinib also blocked VEGF production by HCC3 and CBO12C3 cells by 60% to 80%, whereas VEGF production by AD3 cells was only inhibited by 20% in the presence of EGF (Fig. 3A). The production of VEGF and CXCL1 in response to exogenous EGF was rather more sensitive to inhibition by gefitinib in HCC3 and CBO12C3 cells compared with AD3 cells.
Effect of gefitinib on phosphorylation of EGFR, Akt, and ERK1/2 in HCC cells. When EGFR binds EGF or transforming growth factor-α, it is autophosphorylated and activates a number of downstream signaling molecules, such as Akt and mitogen-activated protein kinase (ERK1/2; ref. 42). We examined the effect of gefitinib on the phosphorylation of EGFR, Akt, and ERK1/2 in the three cell lines in vitro. Exogenous EGF enhanced the phosphorylation of EGFR, Akt, and ERK1/2 in HCC3 and CBO12C3 cells. By contrast, EGF enhanced the phosphorylation of EGFR and ERK1/2, but not Akt, in AD3 cells; Akt was constitutively phosphorylated in AD3 cells when EGF was added exogenously (Fig. 3B). Phosphorylation of Akt was also inhibited by gefitinib to different extents in the three cell lines: 0.1 μmol/L gefitinib inhibited Akt phosphorylation in HCC3 and CBO12C3 by 50% compared with EGF alone, but had no effect at this concentration in AD3 cells (Fig. 3B). Although Akt was constitutively phosphorylated in the absence of EGF in AD3 cells, phospho-Akt was rather less sensitive to the inhibitory effect of gefitinib in AD3 than in the other two cell lines.
Involvement of Akt and ERK1/2 in EGF-induced production of angiogenic factors, VEGF and CXCL1. To identify the EGF/EGFR–activated signal pathways involved in the production of angiogenic factors, we looked at the effects of LY294002, a selective inhibitor of PI3K, and U0126, an inhibitor of ERK1/2, on the production of VEGF and CXCL1 by HCC3 and AD3 cells. LY294002 inhibits Akt activation by specifically blocking PI3K, a positive regulator of Akt kinase. EGF increased VEGF and CXCL1 production in both these cell lines 2- to 3-fold (see Fig. 3A) and treatment with LY294002 resulted in a significant decrease in the production of both factors in HCC3 cells compared with EGF alone (Fig. 4A and B). However, LY294002, tested up to a concentration of 5 μmol/L, had no effect on the EGF-induced production of VEGF and CXCL1 in AD3 cells (Fig. 4A and B). Treatment with U0126 also affected cellular production of VEGF in both cell lines but there was no significant difference in the inhibitory effect of U0126 on cellular production of VEGF between two cell lines (Fig. 4C). U0126 showed only a slight, if any, inhibition of CXCL1 production, but there was no significant difference in its inhibition between two cell lines (Fig. 4D).
Role of PTEN in gefitinib-mediated inhibition of Akt activation and production of angiogenic factors. Akt activation is regulated by both phosphorylation at serine-473 and PIP3 binding to pleckstrin homology domain of Akt. The level of cellular PIP3 is controlled by PI3K and the lipid phosphatase, PTEN (44). We compared the expression of PI3K and PTEN in the three cell lines in the presence or absence of EGF. In the absence of EGF, all three cell lines expressed similar levels of the PI3K catalytic subunit, p110, but HCC3 and CBO12C3 expressed rather higher levels of the regulatory subunit, p85, than AD3 (Fig. 5A). Quantitative analysis showed that PTEN was expressed in AD3 at about half the level in HCC3 or CBO12C3, both in the absence and presence of EGF (Fig. 5A). This difference in PTEN expression between AD3 and the other two cell lines was consistently found in repeated experiments with independently cultured cell lines (data not shown).
Because gefitinib had been found to be a much less effective inhibitor of Akt activation and angiogenic factor production in AD3 cells than in HCC3 cells (Fig. 3B), we next examined whether the lower expression of PTEN in AD3 could be related to this. In HCC3 cells transfected with PTEN siRNA, the PTEN gene was very effectively silenced (see Fig. 5B); Akt was found to be phosphorylated in the absence of EGF, but was not in mock-transfected cells; and Akt phosphorylation was much less sensitive to inhibition by gefitinib (Fig. 5C). We then looked at cellular levels of CXCL1 and VEGF. The cellular levels of CXCL1, but not VEGF, in PTEN siRNA-transfected HCC3 cells were about half those in mock-transfected cells (Fig. 5D) and the production of both VEGF and CXCL1 in response to EGF became insensitive to inhibition by gefitinib.
Increased EGFR gene copy number in HCC cells. Cappuzzo et al. (22) reported that gene amplification and copy number of EGFR gene in association with Akt activation is closely associated with drug sensitivity on therapeutic efficacies of gefitinib in NSCLC. We examined if EGFR gene was amplified in liver cancer cell lines, HCC3 and AD3 by FISH analysis. HCC3 and AD3 cells showed increased copy number of EGFR gene compared with normal spleen cells (Fig. 6). However, EGFR gene amplification was not observed in both HCC3 and AD3. HCC3 cells were near triploid with 2.7 ± 0.9 copies of chromosome 11 per cells and 61.3 ± 18.4 of chromosome per cells. AD3 cells were near tetraploid with 3.7 ± 0.9 copies of chromosome 11 per cells and 77.8 ± 18.3 of chromosome per cells. We found that both cell lines increased copy number of EGFR gene; however, there was no apparent difference in both gene amplification and copy number of EGFR in HCC3 and AD3.
We have previously reported that gefitinib inhibits EGF-induced angiogenesis both in vitro and in vivo (36, 37). In this study, we have shown for the first time that tumor-induced angiogenesis in the mouse dorsal air sac assay can be blocked by gefitinib treatment. However, when different HCC cell lines were used in this model to induce angiogenesis, its susceptibility to inhibition by this drug was found to vary. Angiogenesis induced by HCC3 cells was most susceptible to inhibition by gefitinib. Angiogenesis induced by CBO12C3 cells was also inhibited by gefitinib, but at higher doses of the drug. In contrast, angiogenesis induced by AD3 was relatively resistant to inhibition by gefitinib.
We first examined expression of which angiogenic factor was augmented by exogenous addition of EGF in HCC cell lines by membrane-bound antibody array technology. Of various angiogenesis-related factors, we observed up-regulation of VEGF and CXCL1 genes by EGF (data not shown), and we focused on these two potent angiogenic factors. In all HCC lines, EGF was found to enhance cellular production of both VEGF and CXCL1. The production of these potent angiogenic factors was also more effectively inhibited by gefitinib in HCC3 and CBO12C3 cells than that in AD3 cells. This suggests that the inhibitory effect of gefitinib on HCC-induced angiogenesis might be due to its effect on the production of such angiogenic factors. The production of angiogenic factors, including VEGF and IL-8, is often enhanced by EGF, TGF-α, and other cytokines in tumor cell lines (25, 38, 45–48). Consistent with these previous studies, the production of both VEGF and CXCL1 increased 1.6- to 3.2-fold with EGF treatment in all three HCC cell lines. The fact that gefitinib was a much less effective inhibitor of VEGF and CXCL1 production in AD3 than in HCC3 and CBO12C3 cells suggests that the EGF induction mechanism in AD3 cells may be refractory to gefitinib. Our previous studies suggested two pathways by which gefitinib could exert its antiangiogenic effects: (a) EGF/TGF-α up-regulates the expression of angiogenic factors by cancer cells themselves, resulting in the induction of angiogenesis by a paracrine mechanism, and this process is inhibited by gefitinib; or (b) EGF/TGF-α induces angiogenesis through direct interaction with EGFRs on endothelial cells and this process is inhibited by gefitinib (36, 37). In our studies of EGF-induced neovascularization in the mouse cornea, we obtained evidence supporting the latter pathway as gefitinib inhibited activation of EGFR in the new vessels themselves (37). The results of our present study suggest that the former pathway is a more likely candidate because gefitinib inhibited the production of both VEGF and CXCL1 by HCC cells.
While studying the mechanism underlying the different inhibitory effects of gefitinib on VEGF and CXCL1 production by HCC cell lines, we found that the drug had differential effects on Akt activation. Whereas EGF treatment of HCC3 and CBO12C3 cells resulted in the phosphorylation of EGFR, Akt, and ERK1/2, and this was highly susceptible to inhibition of gefitinib. In AD3 cells, gefitinib also inhibited the phosphorylation of EGFR and ERK1/2 in response to EGF, but Akt seemed to be constitutively active in the absence of EGF, and phospho-Akt levels in AD3 cells were less susceptible to the effects of gefitinib. Interestingly, LY294002, a selective inhibitor of PI3K, blocked production of VEGF and CXCL1 in HCC3, but not that in AD3, whereas U0126, an ERK1/2 inhibitor, did not have this selective inhibitory effect. Taken together, these different effects of gefitinib on the induction of angiogenesis and the production of angiogenic factors suggest an important role for the Akt pathway. However, as ERK1/2 and PI3K inhibitors also reduced the production of angiogenic factors, these signaling pathways may also play a role in the antiangiogenic effects of gefitinib.
Recent studies have shown that a mutation in exons 18 to 21 of the EGFR gene is closely associated with a clinical response to gefitinib in NSCLC patients (18, 19) and that the proliferation of NSCLC cells carrying such EGFR mutations is highly sensitive to gefitinib in vitro (18). We, however, could not observe any mutation in exons 18 to 21 of the EGFR gene in three lines used in this study (data not shown). Cappuzzo et al. (22) have reported that EGFR gene amplification and increased copy number are important to limit therapeutic efficacy of gefitinib in lung cancer. However, there was no difference in gene amplification and copy number of EGFR in both HCC3 and AD3 cells. Therefore, altered expression of angiogenic factors and Akt activations between two lines might not be directly associated with gene amplification of EGFR.
In this study, AD3 cells showed both constitutive Akt activation and reduced levels of PTEN compared with HCC3 and CBO12C3 cells (Figs. 3B and 5A). It has been shown that phospho-Akt was recognized as a risk factor for early disease recurrence and poor prognosis of HCC (49). We have previously reported that sensitivity to gefitinib in NSCLC cell lines is associated with EGFR signaling pathways involving Akt and ERK1/2 and is closely linked with cell growth and apoptosis; activation of Akt and/or ERK1/2 was most susceptible to inhibition by gefitinib in one of the nine NSCLC cell lines, which carries a mutation in the EGFR catalytic domain (20). Both PI3K and PTEN are closely associated with activation of Akt, and reducing PTEN levels might be expected to change both Akt activation in response to EGF and its sensitivity to gefitinib. Reducing PTEN gene expression in HCC3 cells using PTEN siRNA indeed decreased the sensitivity of both Akt activation and the production of VEGF and CXCL1 to gefitinib (Fig. 5). Another study has shown that loss of PTEN expression contributed to resistance to gefitinib by permitting Akt activation independently of receptor kinase signaling, in EGFR-expressing tumor cells (50). These findings support our current evidence for a role of constitutive Akt activation in the gefitinib resistance shown by angiogenesis induced by AD3 cells. However, further studies will be required to understand exactly how constitutive Akt activation is associated with the sensitivity or resistance of EGFR signaling to gefitinib.
Grant support: Ministry of Education Culture, Sports Science, and Technology of Japan (M. Ono and M. Kuwano) and the Second Comprehensive 10-Year Strategy for Cancer Control from the Ministry of Health, Welfare and Labor (M. Kuwano).
We thank Naomi Shinbaru (Kyushu University) for editorial help.