Inhibition of Apoptosis by Amphiregulin via an Insulin-like Growth Factor-1 Receptor-dependent Pathway in Non-small Cell Lung Cancer Cell Lines*

Several abnormalities in the insulin-like growth factor-1 (IGF1) and erbB receptors pathways stimulate the growth and survival of lung cancer cells, but their mechanisms of action and cooperation are poorly understood. In this report, we have identified a new mechanism of apoptosis inhibition by amphiregulin through an IGF1-dependent survival pathway in non-small cell lung cancer (NSCLC) cells: amphiregulin activates the IGF1 receptor that in turn induces the secretion of amphiregulin and IGF1. In the absence of serum, the NSCLC cell line H358 resists apoptosis and secretes factors protecting the NSCLC cell line H322 from serum deprivation apoptosis. IGF1 receptor inhibitor AG1024 as well as epidermal growth factor receptor inhibitors AG556 and ZD1839 restore apoptosis in H322 cells cultured in H358-conditioned medium. Accordingly, the anti-apoptotic activity of H358-conditioned medium is completely abolished after incubation with anti-amphiregulin neutralizing antibody and only partially with anti-IGF1 neutralizing antibody. H358-conditioned medium and amphiregulin induce IGF1 receptor phosphorylation in H322 cells, which is prevented by anti-amphiregulin neutralizing antibody but not by AG556 or ZD1839. H358 cells secrete a high level of amphiregulin that, in combination with IGF1, prevents serum deprivation apoptosis. Finally, IGF1 receptor inhibitor blocks amphiregulin and IGF1 release by H358 cells.

In lung cancer, well characterized molecular abnormalities affect cell cycle control and apoptosis (1); in addition, a variety of peptide hormones and growth factors induce autocrine or paracrine stimulation of cell survival, proliferation, migration, and angiogenesis. Although tumor growth of small cell lung cancer is mainly mediated by neuropeptides such as bombesinrelated peptides that bind to G protein-coupled receptor, the major autocrine/paracrine loops for non-small cell lung cancer (NSCLC) 1 cells involve growth factors that bind to tyrosine kinase receptors (2).
The erbB receptor family contains four distinct members: the epidermal growth factor receptor (EGFR, erbB-1/HER1), erbB-2/neu (HER2), erbB-3 (HER3), and erbB-4 (HER4). These receptors consist of an extracellular domain containing the ligand-binding domain, a hydrophobic transmembrane domain, and an intracellular domain conferring tyrosine kinase activity (3). Binding of specific ligands induces homo-or heterodimerization of the receptors within members of the erbB family (4) and activates distinct intracellular signal transduction pathways and biological responses, including cellular proliferation, differentiation, migration, and survival. Five ligands of EGFR, the epidermal growth factor (EGF), transforming growth factor ␣ (TGF␣), amphiregulin (AR), heparin-binding EGF (HB-EGF) and betacellulin, can be produced by tumorigenic and, to a lesser extent, by non-tumorigenic human lung epithelial cells (5)(6)(7). As most cells of epithelial origin, NSCLC cells overexpress members of the erbB receptor family, in particular the EGFR and erbB-2/neu, but the prognostic impact of this overexpression is still controversial (5,6). Overexpression of TGF␣ and AR has been associated with shortened survival of patients with NSCLC (5,6,8,9).
Cross-talk between the IGF1-R, the EGFR, and their ligands has been reported. Stimulation of IGF1-R signaling interferes with the antitumoral activity of EGFR inhibitor (34) or of Trastuzumab, a monoclonal antibody used clinically to inhibit the growth of erbB-2/neu-expressing tumors (35). In addition, prolonged activation of the intracellular kinase ERK2 by EGF in mouse embryo fibroblasts requires a functional IGF1-R (36). Moreover, stimulation of the IGF1-R intracellular pathway was abolished by the blockade of EGFR in several cell systems (37)(38)(39). In keratinocytes, IGF1 induces the production and secretion of TGF␣ and AR (37), and, in COS-7 cells, stimulation of the IGF1-R transactivates the EGFR via a mechanism involving matrix metalloprotease-dependent release of HB-EGF (39). Thus, the IGF1/IGF1-R pathway transactivates the EGFR via an autocrine release of EGF-like factors (37,39).
Although the role of IGF1-R and EGFR pathways in growth and survival of lung cancer cells is well understood, their mechanism of action and cooperation are poorly understood. In this report, we analyzed the inhibition of serum deprivation apoptosis in two NSCLC cell lines, H358 and H322, and showed that a reciprocal activation of AR and IGF1 pathways induced cell survival. Indeed, we show here that AR phosphorylates the IGF1-R and that the subsequent activation of IGF1-R pathway induces the secretion of AR and IGF1. Transactivation of the IGF1R by AR is independent of EGFR. Subsequent cross-talk between those growth factors and receptors activates a survival pathway in NSCLC cell lines.
For serum deprivation, cells were rinsed once in PBS and cultured for at least 8 h in serum-free medium, which was then changed, and the culture continued in serum-free medium or in serum-free conditioned medium for the desired time. Conditioned media (CM) from cells were collected 72 h after serum-free culture, centrifuged (5000 ϫ g) at 4°C, and stored at Ϫ80°C until use.
Anti-human EGF or AR mouse monoclonal, anti-human TGF␣, betacellulin, or IGF2 goat polyclonal (R&D Systems Europe, Ltd., Abingdon, UK) or anti-human IGF1 goat polyclonal (Oncogene Research Products, Fontenay sous Bois, France) neutralizing antibodies were incubated overnight at 4°C in H358 CM. The immunoneutralized H358 CM was then added to H322 cells for 96 h.
Quantification of Apoptotic Cell Death-The morphological changes related to apoptosis were assessed by fluorescence microscopy after Hoechst 33342 (5 g/ml) staining of cells (Sigma-Aldrich), and the percentage of apoptotic cells was scored after counting at least 200 cells.
In some experiments, active caspase-3 was detected by flow cytometry using phycoerythrin-conjugated monoclonal active caspase-3 antibody kit (BD Pharmingen) following the manufacturer's instructions. Analysis was performed on a Becton Dickinson FACScan flow cytometer. Red fluorescence (phycoerythrin, FL-2) was detected at 575 nm. Cellular DNA was extracted by a salting-out procedure and detected after agarose gel electrophoresis.
AR and IGF1 Dosage-Conditioned media were obtained as previously described from serum-deprived H358 or H322 cells. Cells were plated at a density of 5 ϫ 10 6 cells/flask (75 cm 2 ) or 8 ϫ 10 5 cells/well (six wells plate). AR in the medium was analyzed using specific ELISA. Primary monoclonal anti-human AR antibody (2 g/ml, R&D Systems) was coated to the bottom of 96-well plates and incubated overnight at room temperature. The wells were washed with TPBS (PBS, 0.05% Tween 20), blocked by the addition of PBS containing 1% BSA and 5% sucrose for 1 h at room temperature and washed again. Solutions containing varying concentrations of biologically active recombinant amphiregulin or assay samples were diluted in buffer (20 mM Tris, pH 7.3, 150 mM NaCl, 0.1% BSA, 0.05% Tween 20) and incubated 2 h at room temperature. The wells were washed prior to incubation with a mixture of the secondary biotinylated anti-human amphiregulin antibody (25 ng/ml, R&D Systems) and horseradish peroxidase-conjugated avidin 1:250 in buffer. Following incubation for 2 h at room temperature, the wells were incubated with substrate reagents A (catalog number 2606KC) and B (catalog number 2607KC) (1:1 ratio, BD Pharmingen) 30 min in the dark. Adding 1 M H 2 SO 4 stopped the reaction, and optical density was measured at 450 nm with an ELISA plate reader. The IGF1 level in culture medium was analyzed by radioimmunoassay using the IGF1 extraction kit (Nichols Institute Diagnostics) following the manufacturer's instructions.  vealed by morphological change (Fig. 1A). These changes coincided with DNA fragmentation visualized after agarose gel electrophoresis (Fig. 1B) and proteolytic cleavage of procaspase-3 in its p12 active fragment (Fig. 1C). Conversely, no evidence of cell death was detected in H358 cells, even after 120 h of serum deprivation (Fig. 1, A-C).

H358 NSCLC Cells
To determine whether the resistance of H358 cells was due to the constitutive activation of a survival pathway or whether those cells secrete anti-apoptotic factors, serum-free conditioned medium from H358 cells (H358 CM) was collected and used for culturing H322 cells. H358 CM inhibited apoptosis of H322 cells, evaluated by the absence of morphological changes or cleavage of caspase-3, after 96 h of culture (Fig. 1D). This effect was dose-dependent (data not shown).
Inhibition of Serum Deprivation Apoptosis in H322 and H358 NSCLC Cell Lines by EGF and IGF1 Receptors-Human lung cancer cells are known to express EGFR, erbB-2/neu, IGF1-R, and platelet-derived growth factor receptors (PDG-FRs). We first considered the possible involvement of these receptors in the survival of H322 cells cultured in H358 CM. Serum-deprived H322 cells were incubated in medium supplemented with 10% serum, in serum-free medium or in H358 CM in the presence of several highly specific inhibitors (Fig. 2).
Tyrphostin AG825 and AG1295, which inhibited respectively the erbB-2/neu and the PDGFR, did not modify the survival of H322 cells cultured in the presence of serum, in serum-free medium, or in H358 CM (Fig. 2, A and B). The IGF1-R inhibitor tyrphostin AG1024 (1 M) did not modify apoptosis in H322 cells cultured in the presence or in the absence of serum, but restored, in a dose-dependent manner, apoptosis when cells are cultured in H358 CM (Fig. 2C). Both EGFR inhibitors tyrphostin AG556 (20 M) and quinazolin ZD1839 (500 nM) did not modify apoptosis of the cells in the presence of serum, increased the level of apoptosis of the cells when cultured in serum-free medium, and restored apoptosis in H322 cells cultured in H358 CM (Fig. 2, D and E).
We then analyzed the survival of H358 cells cultured in serum-free medium in the presence or absence of the same inhibitors (Table I). In the presence of serum none of the inhibitors induced apoptosis of H358 cells. In contrast, the EGFR inhibitor tyrphostin AG556 (10 M) as well as the IGF1-R inhibitor tyrphostin AG1024 (10 M) restored H358 cells apoptosis when cultured in serum-free medium (Table I). Conversely, concentration of up to 20 or 40 M of the erbB-2/ neu inhibitor tyrphostin AG825 and the PDGFR inhibitor tyrphostin AG1295 had no effect on H358 cell survival (Table I).

FIG. 4. AR and IGF1 recombinant proteins inhibit serum-deprived apoptosis in H322 cells. H322 cells were cultured in medium
supplemented with 10% serum (10), in serum-free medium (0), in conditioned medium from H358 cells (CM), or in serum-free medium supplemented with the indicated concentrations of AR or IGF1 recombinant proteins. Apoptosis was analyzed by immunolabeling of active caspase-3 and flow cytometry detection. Percentage of apoptosis was expressed as a rate of apoptosis detected in serum-free medium (0), normalized at 100. Data represent the mean Ϯ S.D. of at least three independent experiments. *, p Ͻ 0.05; **, p Ͻ 0.005; statistically less significant than serum-free medium condition (0). These results were confirmed by a Western analysis of proteolytic activation of pro-caspase-3 in its p12 active fragment (Fig.  3). Altogether, these data highly suggested that H358 cells secreted autocrine/paracrine factors activating the IGF1-R and EGFR in H358 and H322 cells. AR and IGF1 Inhibit Serum Deprivation Apoptosis of H358 and H322 Cells-To identify factors of the EGF and IGF families involved in the anti-apoptotic activity of H358 CM, we used neutralizing antibodies. The anti-apoptotic activity of H358 CM was completely abolished by a preincubation with the anti-AR neutralizing antibody and to a lower extent (about 50%) by the anti-IGF1 antibody (Table II). In contrast, preincubation of H358 CM with anti-EGF, anti-TGF␣, anti-betacellulin, anti-IGF2 antibodies, or with CRM197, which neutralized the HB-EGF, did not alter anti-apoptotic activity of H358 CM (Table II). These results were confirmed by treatment of H358 cells with an antisense oligonucleotide directed against the AR mRNA, which significantly restored apoptosis of H322 cells cultured in H358 CM (not shown).
We then quantified AR and IGF1 secreted by H358 cells cultured in the absence of serum (Table III). The level of IGF1 secreted by H358 cells was not significantly different than those secreted by other cancer cell lines from lung or other origins, but the level of AR was very high. Furthermore, H322 cells secreted IGF1, but AR was below detectable levels (Table  III).
We then used AR or IGF1 recombinant proteins to analyze inhibition of H322 cells in serum deprivation apoptosis. When AR or IGF1 were used as a single agent, the apoptosis inhibition was dose-dependant and observed only at high concentrations (Fig. 4). In contrast, when AR and IGF1 proteins were used in combination, the maximal apoptosis inhibition observed in H322 cells cultured in serum-free medium was reached with 5 ng/ml AR and 1 ng/ml IGF1, the concentrations present in H358 CM (Fig. 4). These data confirmed that both AR and IGF1 are involved in the anti-apoptotic activity of H358 CM and suggested a cooperation between the growth factors to inhibit serum deprivation apoptosis in the two cell lines.
AR Activates the IGF1-R in H322 Cells-We then analyzed the activation of the IGF1-R in H322 cells. The IGF1-R was unphosphorylated in H322 cells cultured in serum-free medium (Fig. 5, A and B). As expected, IGF1 induced the tyrosine phosphorylation of its receptor. Moreover, we observed the phosphorylation of the IGF1-R on H322 cells after 30 min of incubation in H358 CM (Fig. 5A). This phosphorylation was prevented when H358 CM was preincubated with the anti-AR neutralizing antibody (Fig. 5B). Moreover, incubation of H322 cells for 30 min with AR recombinant protein in serum-free medium induced the phosphorylation of the IGF1-R (Fig. 5B). Thus, in H322 cells, AR activated the IGF1-R. Accordingly, H358 cells cultured in serum-free medium, which secreted IGF1 and a high amount of AR, presented a constitutive phosphorylation of the IGF1-R (Fig. 5C).
Activation of the IGF1-R by AR Is Independent of the EGFR-We then analyzed whether or not the EGFR was involved in IGF1-R transactivation. Interestingly, we showed that IGF1-R phosphorylation by H358 CM or AR in H322 cells was not inhibited in the presence of the two EGFR inhibitors, AG556 and ZD1839 (Fig. 6). Altogether, these data demonstrated that the IGF1-R can be activated by AR independently of the activation of the EGFR.
Secretions of AR and IGF1 Are Induced by the IGF1-R-The release of soluble EGF-like growth factors involves proteolysis mediated by matrix metalloproteases (40,41) and the IGF1-R (37,39). To determine the mechanism by which a large quantity of AR is released by H358 cells, we assessed whether treatment with the IGF1-R inhibitor AG1024 or the metalloproteases inhibitor 1,10-phenanthroline affected the AR release by serum-deprived H358 cells. As shown in Fig. 7, incubation with AG1024 or 1,10-phenanthroline both markedly decreased the AR level detected in the serum-free culture medium. In contrast, incubation of H358 cells with 1,10-phenanthroline had no effect on IGF1 level, but AG1024 decreased IGF1 release in the serum-free culture medium (Fig. 7).
The data suggested that the IGF1-R pathway regulated the AR secretion through metalloproteases activation. Furthermore, activation of the IGF1-R might also induce IGF1 release. DISCUSSION We have shown here that the reciprocal activation of AR and IGF1 pathways induced inhibition of serum deprivation apoptosis in NSCLC cell lines H358 and H322. We demonstrated, for the first time to our knowledge, that AR activated the IGF1-R, which in turn induced the secretion of AR and IGF1. Transactivation of the IGF1R by AR is independent of its binding to EGFR. Thus, AR can inhibit apoptosis in NSCLC cells through an IGF1-R-dependent pathway.
AR, an EGF-related growth factor produced by epithelial cells, functions as an autocrine/paracrine cell proliferation factor (42,43). Several lines of evidence suggest that AR is an autocrine growth factor in human carcinoma cells (44 -51) and tumorigenicity of a human breast epithelial cell line is abolished by treatment with AR antisense (49). A protective role of the EGF-like growth factors against apoptosis has been demonstrated in several non-transformed and tumor cell types (52)(53)(54)(55)(56). Accordingly, anti-apoptotic activity of AR has been suggested by a study in human breast carcinoma cells (48). AR is overexpressed in a significant percentage of tumorigenic human lung epithelial cells (7). Moreover, an extensive analysis of the expression of EGF-related peptides and receptors in a large group of 195 stage I-IIIA NSCLC patients showed that enhanced AR expression was the only variable to be significantly correlated with a reduced overall survival (6). We confirm here that AR could be a potent inducer of NSCLC cells autonomous survival.
IGF1 and its receptor the IGF1-R contribute to carcinogenesis. The IGF1-R effects several growth-promoting functions: it stimulates mitogenesis (14,15), promotes cellular transformation (16,18), and protects cells from apoptosis (19 -33); a dominant-negative IGF1-R mutant induced extensive apoptosis in vivo and inhibited growth in vitro of rat glioblastoma C6 cells (57). We confirmed in this study that the activation of the IGF1/IGF1-R pathway inhibits NSCLC cell apoptosis.
Known EGFR ligands, including AR, are synthesized as cell surface precursor proteins, and soluble growth factors are produced by proteolysis (40,41). Activation of the IGF1-R pathway induces the secretion of soluble EGF-related peptide such as AR and TGF␣ in keratinocytes and HB-EGF in COS-7 cells that in turn activate the EGFR (37,39,58). We showed here that inhibition of the IGF1-R or the metalloproteases on H358 cell line decreased the level of secreted AR, suggesting that the IGF1-R increased the shedding of pro-AR from H358 cell surface by metalloprotease activation, as described for HB-EGF (39). Whether or not this mechanism is sufficient to explain the very high level of AR secreted by H358 cells is under investigation. Our results showed also here that AR stimulated the IGF1-R, which in turn induced the release of AR, an amplification phenomenon that could also explain the high levels of AR secreted by H358 cells. Furthermore, H358 cells express a mutated constitutively active Ras protein (59), and Ras activation has been shown to increase the production of EGF-related peptides (60,61). Accordingly, H322 cells, which did not secrete AR, express a wild-type Ras protein (59).
AR is known to elicit its biological activity through binding and activation of the EGFR. We demonstrated that AR can also induce the phosphorylation of IGF1-R. Furthermore, in H322 cells, the anti-apoptotic activity of H358 CM was strongly inhibited by both an IGF1-R inhibitor and anti-AR neutralizing antibody, although EGFR inhibitor and anti-IGF1 neutralizing antibody were less effective. Altogether, these data suggested that, in the presence of the two growth factors and their receptors, the anti-apoptotic effect is preferentially mediated by AR and the IGF1-R pathway. The activation of IGF1-R by EGFrelated peptides has never been demonstrated, although it was suggested in results from Swantek and Baserga (36) in mouse embryos fibroblastic cells. These authors showed that EGF induced a prolonged activation of ERK2 intracellular kinase in cells containing functional IGF1-R, but only a weaker and transitory activity of the kinase in cells that do not express IGF1-R or that express a mutated form of the receptor. Transfection of cells with IGF1-R restored the capacity of EGF to induce prolonged activation of ERK2 (36), thus suggesting that the IGF1-R was required to transduce the EGF-mediated signal.
Furthermore, we showed that AR transactivates the IGF1-R independently of its binding on its specific receptor, the EGFR. Whether interaction of AR with the IGF1-R is direct or due to release were measure in H358-conditioned medium using ELISA detection. 8 ϫ 10 5 cells were incubated for the times indicated in serum-free medium (0% serum), in serum-free medium supplemented with AG1024 1 (M), or 1,10-phenanthroline (100 M). Data were expressed as the mean Ϯ S.D. of three independent experiments. oligomerization of AR and IGF1 before binding to the receptor is under investigation. Although transactivation of IGF1-R by EGFR activation has never been described before, cases of transactivation of erbB family receptors by IGF1-R have been observed (37)(38)(39). Moreover, in glioblastoma, IGF1-R signaling is involved in resistance mechanisms to anti-EGFR therapy (34), and, in breast cancer cell models that overexpress erbB-2/neu, an increased level of IGF1-R signaling appears to interfere with the action of Trastuzumab, an anti-erbB2/neu antibody (35). By using combinations of antisense oligonucleotides directed against erbB-2 mRNA and IGF1-R, it has also been shown that IGF1-R directs erbB2 phosphorylation (62); this interaction involves physical association of both receptors and the formation of an heteromeric complex. Similar interaction between EGFR and IGF1-R has never been described.
Emergence of a tumoral clone resulted from accumulation of abnormalities resulting to a prolonged survival and a stimulation of growth and mobility. We previously showed that H358 NSCLC cells presented an increased resistance to p53-and BAX-mediated apoptosis (63). Resistance to apoptosis in the H358 cell line is linked to conformational abnormalities of the BAX protein and an inhibition of its translocation from the cytosol to the mitochondria; this phenomenon is reversed by treatment with caffeine. In this report, we characterized another original mechanism of apoptosis inhibition in NSCLC cells. It remains to be determined if there is a link between both these observations and, in particular, if the activation of the IGF1-R pathway by AR can modify the phosphorylation of the BCL2 family members and inactivate BAX.
In conclusion, secretion of growth factors, in particular EGFrelated peptides, by human lung epithelial cells is well established, but the consequences of the coexpression of EGFR and its ligands by malignant cells on lung tumor progression are still controversial as it is for IGF1-R expression. Our study, which demonstrates for the first time the activation of IGF1-R by AR, in addition to the stimulation of AR release by IGF1-R activation, is a very important step in the elucidating the involvement of AR and IGF1 in lung tumor development and progression. Further studies on the relevance of those growth factors in vivo on fresh tumor samples will allow development of a new therapeutic strategy.