Resistance to EGFR-TKI can be mediated through multiple signaling pathways converging upon cap-dependent translation in EGFR-wild type NSCLC

Abstract

INTRODUCTION

For the majority of patients with non-small cell lung cancer, response to epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKI) is suboptimal. In models of acquired resistance to EGFR-TKI, activation of Akt phosphorylation is frequently observed. Since Akt activation results in downstream initiation of cap-dependent protein translation, we hypothesized that a strategy of targeting cap-dependent translation in combination with erlotinib might enhance therapy.

METHODS

NSCLC cells that are wild-type for Egfr were assayed for sensitivity to erlotinib. Serum-starved NSCLC cells were assayed for EGFR signaling and downstream pathway activation by immunoblot after stimulation with EGF. EGFR signaling and signaling mediators of cap-dependent translation were assayed by immunoblot under serum replete conditions 24 hours after treatment with erlotinib. Finally, combination treatment with erlotinib and 2 different cap-dependent translation inhibitors were done to assess the effect on cell viability.

RESULTS

EGFR signaling is coupled to activation of cap-dependent translation in EGFR wild-type cells. Erlotinib inhibits EGFR phosphorylation in EGFR-TKI resistant cells, however, results in activation of downstream signaling molecules including Akt and ERK1/2 resulting in maintenance of eIF4F activation. eIF4F cap-complex formation is maintained in erlotinib resistant cells, but not in erlotinib sensitive cells. Finally, using an antisense oligonucleotide against eIF4E and a small-molecule inhibitor to disrupt eIF4F formation, we show that cap-dependent translation inhibition can enhance sensitivity to erlotinib.

Conclusions

The results of these studies support further clinical development of translation inhibitors for treatment of NSCLC in combination with erlotinib.

INTRODUCTION

Non-small cell lung cancer (NSCLC) remains the leading cause of cancer-related death worldwide¹. The outcomes with standard chemotherapy have reached a plateau in terms of efficacy and novel targeted therapies have made their way into the standard treatment of NSCLC. The epidermal growth factor receptor (EGFR) has emerged as a major target for NSCLC therapy with the seminal BR.21 trial demonstrating a modest but significant improvement in overall survival in patients treated with erlotinib, an EGFR tyrosine kinase inhibitor (TKI), in the second-line ². While every subset in that randomized trial showed some benefit to erlotinib, it is now clear that the small fraction of patients harboring an activating mutation in the EGFR tyrosine kinase domain get the most benefit with objective response rates of about 60%^3-5. Despite prevalent overexpression of EGFR, the response rates in Egfr wild-type (WT) patients is less than 10% with stable disease in about 50%. Therefore, while EGFR-directed therapy remains a viable option for patients with Egfr-WT tumors, the results are suboptimal.

Experimental models of EGFR-TKI acquired resistance demonstrate that activation of downstream pathways either through Kirsten rous sarcoma (K-ras) mutation or amplification of redundant pathways such as insulin-like growth factor receptor (Igfr) or mesenchymal-epithelial transition (c-Met) mediate such resistance^6-8. Based on these studies, two clinical trials have been completed to test the strategy of combination treatment of erlotinib in combination with c-Met and IGFR inhibitors⁹.

Still, other data show that there are potentially numerous other pathways that might result in resistance to erlotinib such as Her2 and Her3, among others¹⁰. Therefore, the best strategy for combination therapy with EGFR-TKI is not entirely straightforward and the correct biomarker to choose which combination to consider is not entirely clear at this time.

We and others have demonstrated that the 5′-mRNA cap-dependent protein translational machinery is vital to lung tumorigenesis and progression^11-14. However, this autonomous translational activity is not associated with any known mutations in the translational apparatus. Rather, the constitutive activity of the translational machinery appears to be mediated by active upstream signaling pathways that converge upon the eukaryotic initiation factor 4F (eIF4F) trimolecular complex^15-17. Therefore, we hypothesize that eIF4F serves as a major regulatory node downstream of EGFR as well as several other alternate signaling pathways, and maintained activation of cap-dependent protein translation due to redundant signaling may result in resistance to EGFR-TKI. In the following manuscript, experiments are described that show that most Egfr-WT NSCLC cells are primarily resistant to erlotinib treatment. Moreover, erlotinib treatment results in activation of Akt and maintenance of activated eIF4F complex formation. Finally, combination therapy with two different inhibitors of cap-dependent translation improved the efficacy of erlotinib against NSCLC cells in vitro. The result of this work supports further clinical development of translation inhibitors in combination with erlotinib.

MATERIALS AND METHODS

Cell lines and reagents

Cells were obtained from the ATCC or from the laboratory of Frederick Kaye (NCI). H2009, H522, H460, H520, H2030 were grown in RPMI 1640 (Gibco, Invitrogen) with 10% calf serum (R10). H838 and H2122 were grown in R10 and L-glutamine, HEPES, glucose, and sodium bicarbonate supplements. Erlotinib was obtained from LC laboratories. LY2275796 (Antisense oligonucleotide to eIF4E or 4E-ASO) and mismatch ASO (MM-ASO) were obtained from Jeremy Graff (Eli Lilly and Company, Indianapolis, Indiana). 4EGI-1 was purchased from Chembridge Corporation (San Diego, CA)¹⁸.

Cytotoxicity Assays

Cytotoxicity of erlotinib on NSCLC was performed by CCK-8 kit (Dojindo, Inc) as previously described ¹⁹. Briefly, 2000 to 5000 cells were seeded onto 96 well plates and allowed to adhere overnight. The following day, medium containing various concentrations of erlotinib were added to appropriate wells. After 72 hours, 10μL of CCK-8 reagent were added to the wells and incubated for 4 hours at 37°C. The color change was read on a 96-well plate reader at 405 nm of light. Experiments were performed in quadruplicate with untreated controls and additional wells were measured without cells as a background control.

EGF stimulation

Cells were seeded onto 10cm plates at 1.5-2.5 × 10⁶ cells and allowed to adhere overnight. The following night, cells were washed twice with PBS and serum-starved in RPMI overnight. The following morning, cells were stimulated with 100 ng/mL EGF with and without 1 μM erlotinib. Cell extracts were prepared at 20, 60, and 150 minutes post-stimulation. Cells were washed once with ice-cold 1× PBS. 1× cell lysis buffer (Cell Signaling) containing PMSF 1mM was added directly to the plate followed by scraping of the cells and the resulting lysate was immediately placed on ice. Cells were centrifuged to pellet nuclear material and cell debris and supernatants were stored at −80° C until use.

Immunoblots

25 to 100 μg of protein were subjected to SDS-PAGE and immunoblot as previously described ²⁰. Antibodies to p-EGFR^Tyr1068 (#2236), EGFR (#2646), p-IGFR^Tyr1135/1136 (#3024), p-c-Met^Tyr1003 (#3135), c-MET (#3127), p-JNK^{Thr183/Tyr185} (#9251), JNK (#9252), p-Akt^Ser473 (#9271), Akt (#9272), p-ERK1/2^{Thr202/Tyr204} (#9101), ERK1/2 (#9102), 4E-BP1 (#9452), p-eIF4E (#9741), and eIF4E (#9742) were obtained from Cell signaling and used at 1:1000 dilution in TBS-T unless otherwise mentioned. Anti IGFR-β (sc-713) was obtained from Santa Cruz Biotechnology, Inc. Anti-eIF4G antibody (1:5000 dilution) was kindly provided by Nahum Sonenberg. β-actin (Sigma, Cat.# A1978) was used as a loading control (1:10000 dilution). Briefly, cells were plated onto 10 cm culture plates overnight in R10. The following day, cells were treated with erlotinib 2μM or 5μM or equal volumes of drug vehicle (DMSO) as control. 24 hours later, cells were lysed and stored at −80°C until used. Protein concentrations were determined using Bradford assay and then loaded onto 8 to 15% SDS-PAGE gels, transferred to PVDF (GE Healthcare), and assayed with above antibodies to determine protein expression.

Cap-Affinity Assay

Binding of eIF4E and binding partners eIF4G and 4E-BP1 was determined using 7^m-GTP sepharose beads (Amersham) as previously described¹⁶. Briefly, 300μg of lysates were added to 50 μL of beads and rotated for 3 hours at 4°C. Beads were washed thrice with cell lysis buffer and binding proteins were eluted using 5x-Laemli’s buffer and boiling at 95°C for 5 minutes. These samples were then run on SDS-PAGE gels and subjected to immunoblot for eIF4E, eIF4G, and 4E-BP1.

RESULTS

A panel of NSCLC cell lines, all of which are Egfr-WT, were treated with various doses of erlotinib in vitro (Figure 1). The majority of NSCLC cells were found to be resistant to erlotinib with IC₅₀ that were greater than 5μM. Only H2122 was modestly sensitive with an IC₅₀ of ~1 μM. These results are consistent with previously published data²¹. Next, the effect of EGF stimulation on downstream signaling pathways leading to cap-dependent translation was examined. Cells were serum-starved prior to EGF stimulation to take away any potential effect of other growth factors in the serum replete medium. EGF stimulation led to activation of EGFR, which was inhibited if cells were pre-treated with erlotinib (Figure 2). Downstream phosphorylation of Akt and 4EBP-1 was observed after EGF stimulation of serum-starved cells. Erlotinib pre-treatment prevented Akt phosphorylation and subsequent hyperphosphorylation of 4E-BP1 in each of the cell lines after 20 minutes of EGF stimulation. Erlotinib pretreatment also inhibited ERK1/2 phosphorylation in all but H2009 cells. These data show that EGFR stimulation results in hyperphosphorylation of 4E-BP1 and that EGFR-TKI treatment can inhibit downstream signaling regulating 4E-BP1 under serum-starved conditions. Furthermore, under these conditions, there was little difference in the signaling characteristics mediated by EGF stimulation between EGFR sensitive and resistant cells.

Figure 1.

NSCLC cells that are wild-type for EGFR were treated with increasing concentrations of erlotinib in a 96-well format. Data are expressed as a percent of viable cells compared to untreated cells after 72 hours of erlotinib exposure. Cells were treated in triplicate. H2122 (black diamond) cells are the only cells that are sensitive to erlotinib. Error bars indicate standard deviation. * indicates statistical significance at p<0.05.

Figure 2.

Immunoblots of serum-starved NSCLCs that were then stimulated with exogenous EGF. Lysates were made at indicated time-points (minutes) and assessed for EGFR signaling and downstream pathway activation. Lane 2 (indicated with an arrow) were lysates prepared from cell pretreated with erlotinib 2 μM and then EGF stimulated for 20 minutes. Phosphorylation of 4E-BP1 is indicated by an upward shift of the bands as the hyperphosphorylated forms migrate slower on the gel. The hyperphosphorylated forms are β, and γ, while hypophosphorylated are α. β-actin is used as a loading control. Sfm = serum-free medium.

Since we hypothesized that eIF4F activation upon EGFR-TKI administration would be sustained by alternate signaling pathways, we examined the signaling properties of erlotinib-resistant (H460, H520, H2009 and H522) compared to erlotinib-sensitive (H2122) cells following erlotinib treatment under serum replete conditions (Figure 3). For both erlotinib sensitive and resistant cell lines, erlotinib administration resulted in inhibition of phosphorylated EGFR suggesting that the drug was able to inhibit its target. However, in 3 of 4 erlotinib-resistant cell lines, phosphorylation of either IGFR or c-MET was observed 24 hours after treatment. In all of the resistant cell lines, Akt phosphorylation was enhanced or maintained 24 hours after erlotinib administration compared to vehicle control treated cells. Of the resistant cell lines, all but H522 showed enhanced or maintained 4E-BP1 phosphorylation despite erlotinib treatment. H460 demonstrated activation of Akt and ERK1/2 after erlotinib treatment despite inhibition of IGFR and c-Met phosphorylation suggesting that another kinase pathway may play a role in maintaining activation of these pathways. In contrast, H2122 cells showed no increase in phosphorylated c-Met or IGFR in response to erlotinib treatment. Furthermore, downstream Akt and 4E-BP1 phosphorylation were markedly inhibited after erlotinib treatment in H2122 cells. For 3 of the 4 resistant cell lines, ERK and c-jun N-terminal kinase (JNK) phosphorylation was also enhanced following erlotinib treatment, but not for H2122. In both resistant and sensitive cells, phosphorylation of eIF4E was diminished after erlotinib treatment. While several reports indicate the phosphorylation of eIF4E contributes to enhanced eIF4F activity, there are conflicting data as to the influence of eIF4E phosphorylation^{20, 22, 23}. These results indicate that in resistant cell lines, erlotinib treatment results in the activation of numerous signaling pathways that maintain Akt phosphorylation and downstream formation of the eIF4F complex.

Figure 3.

Immunoblots of NSCLCs that were treated with erlotinib under serum replete conditions. Lysates were made 24 hours later and assessed for activation of downstream and parallel signaling. β-actin is used as a loading control. D = DMSO (vehicle). In erlotinib sensitive cell lines, erlotinib treatment results in inhibition of downstream phosphorylation in Akt and 4E-BP1. In resistant cells, erlotinib results in activation of Akt and maintenance of 4E-BP1 phosphorylation.

These cells were further analyzed by the 5′-cap affinity assay using 7^m-GTP sepharose beads to assay for binding of eIF4E to the cap and binding partners associated with it (Figure 4A). In H2122 (erlotinib-sensitive) cells, erlotinib treatment resulted in inactivation of the eIF4F complex as seen by dissociation of eIF4G from eIF4E. However, in H2009 cells, erlotinib treatment did not impact eIF4F complex formation. Thus, erlotinib sensitivity correlates with the ability to inhibit eIF4F activity. Therefore, we next analyzed the ability of specific translation inhibitors to restore sensitivity to erlotinib therapy. For these experiments, erlotinib resistant cells were used. Two different translational inhibitors were used for these experiments. Antisense oligonucleotide to eIF4E (LY2275796 or 4EASO) and 4EGI-1 were used either alone or in combination with erlotinib. Each of these drugs has been shown to inhibit eIF4F formation in vitro against NSCLC cells ^{18, 24}. The combination of 4EASO and erlotinib was additive for H2009 and significantly enhanced the cytotoxicity compared to erlotinib alone (Figure 4B). 4EGI-1 treatment, likewise, resulted in additive benefit in combination with erlotinib for H2009 cells. For H460 and A549 cells, the combination was greater than additive implying potential synergy with the combination (p<0.01 compared to either drug alone).

Figure 4.

A) Cap-affinity assay: Lysates of erlotinib treated NSCLC cells were exposed to 7^mGTP-sepharose beads to pull-down eIF4E and binding partners 4E-BP1 and eIF4G. Resulting eluates were applied to immunoblot to assess the relative binding of eIF4G and 4E-BP1 to eIF4E. Increased binding of eIF4G to eIF4E indicates activation of eIF4F cap-complex formation, whereas 4E-BP1 binding denotes repression of eIF4F formation. D = DMSO (vehicle). B) H2009 erlotinib resistant cells were grown in 96-well plates and exposed to LY2275796 (antisense oligonucleotide to eIF4E, denoted 4EASO) or mismatch ASO either alone or in combination with erlotinib. 72 hours after treatment cell viability was determined by CCK-8 assay. Untreated cells and cells treated with MM-ASO were used as controls. C) Erlotinib resistant H2030, H2009, A549, and H460 cells were grown in 96 well plates and exposed to 4EGI-1 alone or in combination with erlotinib. 72 hours after treatment, cell viability was determined as in B). Experiment was done in triplicate with error bars indicating standard deviation. * denotes p value <0.01.

DISCUSSION

Taken together this data shows that erlotinib sensitivity or resistance may be influenced by the ability to inhibit eIF4F formation and that primary resistance could be related to sustained eIF4F activation in EGFR WT NSCLC cells. These findings lend more credence to the observation of numerous other investigators that mRNA translation is a major regulatory checkpoint in cancer cells that is a legitimate target for cancer therapy^{25, 26}. Furthermore, in complex tumors with redundant signaling pathways, cap-dependent translation is an important downstream effector of these signaling pathways affecting cell survival and proliferation. While numerous clinical trials are investigating the utility of combined EGFR and IGFR or combination EGFR/c-Met inhibition, the attenuation of cap-dependent translation may be a more fruitful approach as it lays downstream of both of these important signaling pathways and likely also is downstream of other kinase pathways that might mediate resistance such as K-ras, Src, Her3, and JNK^27-30. The combination of erlotinib with inhibition of cap-dependent translation may also be useful in preventing the development of resistance to EGFR-TKI in EGFR mutated overexpression of eIF4E, and erlotinib-resistant EGFR mutant cells can become resensitized to EGFR-TKI treatment upon inhibition of cap-dependent translation³¹.

While the above results show a correlation between erlotinib resistance and maintenance of cap-dependent translation, other mechanisms may play a role as well. As was seen in the above data, several signaling pathways are activated upon erlotinib treatment and have pleiotropic effects on other pathways independent of translation. For example, Akt phosphorylation can have anti-apoptotic effects by regulating p53 and effects on cell cycle regulation in addition to translation control^{32, 33}.

Several inhibitors of cap-dependent translation are in early clinical development. Small molecule inhibitors such as 4EGI-1 have suffered from a lack of clinical potency. LY2275796 has recently completed a phase I clinical trial in patients with solid tumors³⁴. The treatment was safe and well tolerated. Of the 30 patients enrolled in the study, there were no dose-limiting toxicities and only two patients had grade 3 fatigue. Despite post-treatment biopsies showing inhibition of eIF4E expression in the tumor, there were no objective responders and stable disease was the best response. The authors concluded that future trials of LY2275796 be conducted in combination with other therapeutics. In preclinical studies, the combination of LY2275796 with chemotherapy results in enhanced cytotoxicity in NSCLC and mesothelioma (data to be published separately). Due to the excellent safety profile of LY2275796, the combination of translation inhibition in Egfr-WT combination with EGFR-TKI might be a promising approach to treating NSCLC.