Role of EIF4G1 network in non‐small cell lung cancers (NSCLC) cell survival and disease progression

Abstract Although the Eukaryotic Translation Initiation Factor 4 Gamma 1 (EIF4G1) has been found overexpressed in a variety of cancers, its role in non–small cell lung cancers (NSCLC) pathogenesis especially in immunoregulatory functions, its clinical relevance and therapeutic potential remain largely unknown. By using cancer patients tissue assays, the results indicate that EIF4G1 expressional levels are much higher in NSCLC tissues than in adjacent or normal lung tissues, which are also associated with NSCLC patient survival. By using an RNA‐Sequencing based pipeline, the data show that EIF4G1 has a significant association with immune checkpoint molecules such as PD‐1/PD‐L1 in NSCLC. EIF4G1 small‐molecule inhibitors effectively repress NSCLC growth in cell culture and xenograft animal models. Protein array results identify the signature of proteins controlled by EIF4G1 in NSCLC cells, in which new candidates such as MUC1 and NRG1 are required for NSCLC survival and tumorigenesis with clinical relevance. Taken together, these results have for the first time demonstrated the immunoregulatory functions, clinical relevance and therapeutic potential of the EIF4G1 network in NSCLC, which may represent a promising and novel target to improve lung cancer treatment.

in more than 50% of tumour cells, treatment with Pembrolizumab, which targets PD-1, leads to a superior progression-free and overall survival compared to platinum-doublet chemotherapy in the first-line setting. 4 Furthermore, the addition of Pembrolizumab to standard chemotherapy of pemetrexed and a platinum-based drug resulted in significant longer progression-free survival and overall survival irrespective to PD-L1 expression. 5 However, the resistance to immunotherapy and hyper-progressive disease of checkpoint inhibitors treatment has been recently reported in some NSCLC patients. 6,7 Thus, there is still an urgent need to better understand the mechanisms of lung carcinogenesis and to develop novel therapies (alone or combination of existing treatments) for lung cancer patients.
Dysregulation of mRNA translation is a frequent feature of neoplasia. Therefore, therapeutic agents that target components of the protein synthesis apparatus hold promise as novel anticancer drugs that can overcome intra-tumour heterogeneity. 8 Over the last two decades, the eukaryotic initiation factor 4F (EIF4F) complex has been shown to play important roles in oncogenesis. 9,10 As an important component of the EIF4F complex, EIF4G1 protein serves as a scaffold and interacts with several other initiation factors such as EIF4E and EIF4A and helps to initiate cap-dependent translation in mammalian cells by recruiting ribosomes to the capped end of mRNA. 11 In contrast to extensive studies of EIF4E and EIF4A components, much less attention has been paid to the function of EIF4G1. For example, several recent studies have indicated that EIF4E protein levels are associated with NSCLC cell proliferation, migration, invasion, epithelial-to-mesenchymal transition (EMT) and chemo-resistance. [12][13][14] Although EIF4G1 is overexpressed in a variety of cancers, its role in NSCLC pathogenesis especially immunoregulatory functions, clinical relevance and therapeutic potential remains largely unknown. One study showed that 4EGI-1, one of EIF4G1 inhibitors, enhanced the apoptotic effects of tumour necrosis factor-related apoptosis-inducing ligand (TRAIL) on NSCLC cell lines, through inducing CCAAT/enhancer binding protein homologous protein-dependent DR5 and ubiquitin/proteasome-mediated degradation of cellular FLICE-inhibitory protein (c-FLIP). 15 However, the investigators do not focus on the cellular functions of EIF4G1 including its controlled downstream proteins, and the efficacy of EIF4G1 targeted therapy in vivo remains unclear.
Our recently published data demonstrate for the first time that stable silencing of EIF4G1 by shRNA causes significant reduction of proliferation and anchorage-independent growth in NSCLC cell lines (eg A549, H460, H1299). Furthermore, EIF4G1 was potentially required for NSCLC metastasis through promoting tumour cell migration and invasion. 16,17 In the current study, we explore the clinical implications of EIF4G1 by using NSCLC tissue microarrays and other clinical databases, determine the efficacy of selective EIF4G1 inhibitors in NSCLC xenograft models and identify new EIF4G1-controlled cellular proteins in NSCLC cells as well as validating their functions.

| Immunohistochemistry
Immunohistochemistry was performed with the Avidin-Biotin-Peroxidase complex, according to the manufacturer's instructions (Vector Laboratories). Our modified protocol includes paraffin melting at 58°C in a regular oven for 20 minutes, deparaffination in xylene, re-hydration through descending grades of alcohol up to water, and non-enzymatic antigen retrieval in 0.01 M sodium citrate buffer, pH 6.0, heated to 95°C for 40 minutes in a vacuum oven. After a cooling period of 30 minutes, the slides were rinsed in PBS and treated with 3% H 2 O 2 in methanol for 25 minutes to quench endogenous peroxidase. Sections were then blocked with 5% normal horse serum (for mouse monoclonal antibodies), or normal goat serum (for rabbit polyclonal antibodies) in 0.1% PBS/BSA for 2 hours at room temperature. Primary antibodies were incubated overnight at room temperature in a humidifier chamber. Primary antibodies utilized in the present study included rabbit polyclonal anti-EIF4G1 (Abcam,

| NSCLC xenograft models
Cells were counted and washed once in ice-cold PBS. 5x10 5 H460 cells in 50 µL PBS plus 50 µL growth factor-depleted Matrigel (BD Biosciences) were injected subcutaneously into the flank of nude mice, 6-8-week old, male/female (Jackson Laboratory). Two days after this injection, the mice were randomly separated into different groups (4-6 mice per group) and received i.p. injection with either vehicle, 4EGI-1 or 4E1RCat (25 mg/kg of bodyweight, respectively), 3 days/week. The mice were observed and measured every 3 days for the size of palpable tumours for additional 3 weeks. At the end of experiment, the tumours were excised for subsequent histopathological analysis.

Haematoxylin & Eosin (H&E) and immunohistochemistry (IHC) for Ki67
were performed as described previously. 18,19 Images were collected using an Olympus BX61 microscope equipped with a high-resolution DP72 camera and CellSense image capture software. To test the roles of MUC1 and NRG1 in NSCLC tumorigenesis, 5x10 5 H460 stably MUC1-shRNA, NRG1-shRNA knockdown cells or NC-shRNA cells were injected subcutaneously into nude mice, respectively. After 4 weeks, the mice were killed and tumours were excised and compared. All the animal protocols were approved (# 3380) by the LSUHSC Animal Care and Use Committees in accordance with national guidelines.

| Statistical analysis
The protein expression of EIF4G1, PD-1, PD-L1, MUC1 and NRG1 in tumour tissues was measured as an ordinal variable with the four strength levels: negative (-), weak (+), intermediate (++) and strong (+++/++++). The pairwise correlations of proteins in NSCLC tumour tissues were tested using the Spearman correlations. The correlations of protein expression levels in NSCLC tumour tissues were evaluated using the Pearson correlations. Differences of protein expression levels between experimental and control groups were tested using the two-sided t test, and P values < .05 were considered significant. The 50% Inhibitory Concentrations (IC 50 ) were calculated by using SPSS v20.0.  Table 1. However, no significant correlation was found between EIF4G1 expression and tumour TNM, grade or stage (data not shown), which is probably because of the limited number of cases analysed here. By using NSCLC clinical data from

| Clinical implications of EIF4G1 in NSCLC patients
The Cancer Genome Atlas (TCGA) cohort, we found that the mRNA level of EIF4G1 were closely related to NSCLC patients' overall survival through Kaplan-Meier survival analyses, particularly in LUAD patients ( Figure 1D). Taken together, these data strongly support the important clinical relevance of EIF4G1 in NSCLC progression and pathogenesis.

| Regulation of immune checkpoint molecules by EIF4G1 in NSCLC patients
Recent studies have revealed that the EIF4F complex may have immunoregulatory functions (especially regulation of key immune checkpoint molecules) in the tumour microenvironment. For example, programmed cell death 1 (PD-1) can bind to EIF4E and promote its phosphorylation in hepatocellular carcinoma cells. 20 We recently

| EIF4G1 selective inhibitors effectively repress NSCLC cell growth
Currently, there are a limited number of EIF4G1 specific inhibitors commercially available. 4EGI-1 and 4E1RCat are two small-molecule competitive inhibitors that can prevent EIF4G:EIF4E interaction as well as EIF4F complex formation. 8 Our results indicated that both 4EGI-1 and 4E1RCat treatments effectively reduced the growth of NSCLC cell lines, including H460, A549, H1299, in a dose-dependent manner (IC 50 at ~ 8.0-11.0 μM), whereas they displayed almost no inhibitory effects on normal human bronchial epithelial cells (NHBEC) at the same range of concentrations (IC 50 » 100 μM) ( Figure 3A-B). This is probably because of the obvious elevation of EIF4G1 expression in NSCLC cell lines compared to normal lung cells as described previously. 16 By using

| Identification of the signature of EIF4G1controlled proteins in NSCLC cells
As the principal function of EIF4G1 is to control cellular gene translation, a protein array was used for the identification of the signature  Table 1. Most of these downstream proteins have never been reported to be regulated by EIF4G1, whereas they showed a close protein-protein interaction network and potential functional association ( Figure 4A). Interestingly, the Gene Ontology (GO) enrichment analysis indicated that many of EIF4G1-controlled proteins were involved in immune cell migration, chemotaxis, regulation of inflammatory responses, including cytokines and chemokines activities ( Figure 4B-C). KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis indicated that the NF-κB signalling pathway activity was potentially affected by EIF4G1 ( Figure S2A). The immunoblot results confirmed that knockdown of EIF4G1 by RNAi significantly reduced the phosphorylation of NF-κB p65 kinase from all the 3 NSCLC cell lines being tested (Fig. S2B). Notably, some recent studies have reported the regulation of immune checkpoint molecules by NF-κB signalling pathway in different types of cancer. [22][23][24] Taken together, these data indicate again the potential immunoregulatory function of EIF4G1 network in NSCLC tumour cells.

| Novel EIF4G1-controlled cellular proteins, MUC1 and NRG1, functional validation and clinical implications in NSCLC
To further confirm protein array results with functional validation, two of EIF4G1-controlled proteins newly identified, MUC1 (Mucin 1) and NRG1 (Neuregulin 1), were selected for subsequent investigation. MUC1 is a transmembrane glycoprotein that is aberrantly overexpressed in > 80% of NSCLC. 25 Moreover, the overexpression of MUC1 in NSCLC is associated with a poor disease-free and overall survival. 26,27 Another EIF4G1-controlled protein, NRG1, is a ligand for the HER3 and HER4 receptors. NRG1 autocrine signalling has been implicated in insensitivity of NSCLC to EGFR inhibitors. 28 Inhibition of NRG1-and other ligand-mediated HER4 signalling can consistently and significantly enhance the response to chemotherapy and delay tumour regrowth after cessation of treatment. 29 Our results first confirmed that silencing of EIF4G1 by siRNA dramatically reduced MUC1 and NRG1 expression from all the 3 NSCLC cell lines ( Figure 5A), which reflecting the accuracy of protein array data and the cell line relevance of these findings. In addition, directly silencing of MUC1 or NRG1 by siRNA significantly inhibited NSCLC cell growth, while almost not affecting NHBEC growth, which is probably because of the low basal levels of these proteins in normal lung cells ( Figure 5B-D). To further study the role of MUC1 and NRG1 in NSCLC tumorigenesis, the MUC1 or NRG1 stably knockdown H460 cell lines were created by using lentiviral vectors containing 2 shRNA specifically targeting each gene (KD1 and KD2, Figure S3). After being subcutaneously injected these cells or non-targeting control (NC) cells into nude mice, the results showed that direct knockdown of MUC1 or NRG1 significantly repressed NSCLC tumour growth in mice ( Figure 5E-F).
Next, the same NSCLC tissue arrays were used to detect the expressional levels of MUC1 and NRG1. The expression of both proteins was significantly higher in tumour tissues than paired adjacent TA B L E 1 The top 10 proteins significantly up-regulated and/ or down-regulated in NSCLC H1299 EIF4G1 stably knockdown cell line identified from protein array analyses

| D ISCUSS I ON
Our previous study reported that EIF4G1 was significantly up-regulated in NSCLC cell lines compared to normal lung cells, which makes these lung cancer cells more susceptible to EIF4G1-targeted therapy. 16,17 In the current study, we further demonstrate that EIF4G1 has much higher expressional levels in NSCLC tumour tissues than paired adjacent or normal lung tissues by using tissue microarrays, regardless of EIF4G1 IHC labelling scores for each case. The solid evidence has been provided that EIF4G1 inhibitors (eg 4EGI-1 and 4E1RCat) display effective inhibition of NSCLC cell lines growth and tumorigenesis in vitro and in vivo. In contrast, these compounds show much less inhibitory effects on normal lung cells in the same doses range, because of the significant elevation of EIF4G1 and related protein levels in tumour cells. However, there are currently limited EIF4G1 specific inhibitors available. In a recent study, highthroughput drug screening identified SBI-0640756 as a new firstin-class inhibitor that targets EIF4G1 through disrupting the EIF4F complex and attenuates the growth of clinically unresponsive melanomas. 30 Another recent study reported the design, synthesis, and in vitro characterization of a series of rigidified mimetic of 4EGI-1 in which the phenyl in the 2-(4-(3,4-dichlorophenyl)thiazol-2-yl) moiety was bridged into a tricyclic system. 31 They found some analogues in this series to be markedly more potent than the parent prototypic inhibitor in the disruption of EIF4E/EIF4G interaction, and inhibition of human cancer cell proliferation. Therefore, the efficacy of these new compounds will be tested and compared for NSCLC once they are commercially available.
As mentioned before, recent immunotherapy studies have shown promising results on some NSCLC patients, despite of many existing challenges. One of key questions is how to improve immunotherapy efficacy through combination with other therapies including targeted therapy. In fact, EIF4E has been looked on as a node for regulation of immune functions via different translational control pathways, 32 whereas much less known about EIF4G1 in this part. Here, we found that EIF4G1 has close associations with some immune checkpoint molecules such as PD-1 and PD-L1 in NSCLC by using RNA-Sequencing and tissue arrays data from cancer patients. However, a lower positive labelling for PD-1 than PD-L1 was noticed in NSCLC tissues (Supplemental Table 1). Interestingly, one recent study reports that immunoreactivity loss driven by humidity and temperature results in structural distortion of epitopes rendering them unsuitable for antibody binding following epitope retrieval, especially during PD-L1 immunohistochemistry in FFPE tissues. 33   Lavin Bernick faculty grant (to ZL). HYL is also supported by the same NIH COBRE grant above. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

CO N FLI C T O F I NTE R E S T
All the authors declare no conflicts of interest. Zhiqiang Qin: Conceptualization (equal); Formal analysis (equal).

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets used and/or analysed during the current study are available from the corresponding author on reasonable request.