Regulation of proliferation and invasion by the IGF signalling pathway in Epstein‐Barr virus‐positive gastric cancer

Abstract Several carcinomas including gastric cancer have been reported to contain Epstein‐Barr virus (EBV) infection. EBV‐associated gastric cancer (EBVaGC) is classified as one of four molecular subtypes of gastric cancer by The Cancer Genome Atlas (TCGA) group with increased immune‐related signatures. Identification of EBV‐dependent pathways with significant biological roles is needed for EBVaGC. To compare the biological changes between AGS gastric epithelial cells and EBV‐infected AGS (AGS‐EBV) cells, proliferation assay, CCK‐8 assay, invasion assay, cell cycle analysis, RT‐PCR, Western blot and ELISA were performed. BI836845, a humanized insulin‐like growth factor (IGF) ligand‐neutralizing antibody, was used for IGF‐related signalling pathway inhibition. AGS‐EBV cells showed slower proliferating rate and higher sensitivity to BI836845 compared to AGS cells. Moreover, invasiveness of AGS‐EBV was increased than that of AGS, and BI836845 treatment significantly decreased the invasiveness of AGS‐EBV. Although no apoptosis was detected, entry into the S phase of the cell cycle was delayed in BI836845‐treated AGS‐EBV cells. In conclusion, AGS‐EBV cells seem to modulate their proliferation and invasion through the IGF signalling pathway. Inhibition of the IGF signalling pathway therefore could be a potential therapeutic strategy for EBVaGC.

EBV, the focus of this study, was first detected in a Burkitt's lymphoma cell line. Since then, many studies have found EBV to be associated with several well-known human malignancies such as nasopharyngeal carcinoma (NPC), Hodgkin's lymphoma and Epstein-Barr virus-associated gastric cancer (EBVaGC). 3,4 Incidence of EBVaGC is reported to be approximately 10% of globally detected gastric cancers. In addition, clinicopathological characteristics such as undifferentiated type 5 and CpG island hypermethylation 6 are reported to be high in EBVaGC. Prognosis of EBVaGC is also better than that of EBV-negative cases because lymph node metastasis in EBVaGC is significantly less frequent than in EBV-negative gastric cancer. 7 Many studies characterizing EBVaGC have been reported since the 1990s. 8 Most studies have focused on finding EBV-specific genes and their biological functions, 9,10 related microRNAs 11,12 and chemo-resistance mechanisms. 13,14 Recently, high-throughput assays were attempted to uncover regulatory mechanisms in EBVaGC. 15 However, only a few studies on target-specific pathways in EBVaGC have been reported. One report has suggested that sequential combination is needed to improve the sensitivity of 5fluororacil, one of adjuvant therapy agents in solid tumour, sensitivity when using PI3K inhibitors in EBV-positive gastric cancer cell lines. 16 Several studies have suggested that the insulin-like growth factor-1 receptor (IGF-1R) pathway is an essential target of EBV-positive cancer such as NPC. 17 IGF-related ligands (IGF-1, IGF-2 and insulin) and insulin-like growth factor binding proteins (IGFBPs) are produced the liver that are stimulated by growth hormone (GH).
Mainly, a complex of circulating IGFBPs and acid-labile subunit (ALS) prolongs the half-life of IGF-related ligands and delivers them to a complete receptor. Activated IGF-1R by IGF-1 and IGF-2 plays a critical role in proliferation, migration and invasion.
Many cancers express IGF-1R and 75.2% of stomach cancer tissue expresses IGF-1R, which may be the cause of poor prognosis. 18 Several IGF-1R-targeted drugs were developed based on the importance of the IGF-1R signalling pathway, but few proved to be effective. Considerable biological restrictions of IGF-1R are the reasons why development of IGF-1R pathway-targeting drugs has been difficult. 19 The major problem of targeting IGF-1R lies in its sequence similarity to the insulin receptor, which may lead to metabolic dysfunction. Mainly, insulin receptor isoform A (IR-A) activated by IGF-2 is known to promote oncogenic signalling. To overcome blocking that action without interfering the insulin axis insulin receptor activation, BI836845 (Xentuzumab), a drug that targets both IGF-1 and IGF-2, was developed and there are many ongoing clinical trials on various type of tumour. 20 In this study, we investigated the IGF-1R pathway and related biological roles in EBVaGC using BI836845. We also compared biological changes after IGF-1R signalling pathway inhibition in AGS and AGS-EBV cell lines.

| Proliferation assay
For proliferation assays, 2.0 × 10 3 cells were seeded in 24-well plates. The wells were filled with fresh media and 10 μg/mL BI836845 the day after seeding and repeated every 3 days. Samples from triplicate wells were harvested every day, and cells were counted after Trypan blue staining. Growth curves were plotted as cell numbers versus time.
Next, 10 μL of CCK-8 solution was added to each well and incubated for 2 hours at 37°C. Absorbance was determined at 450 nm using microplate reader (TECAN, Männedorf, Switzerland).

| Invasion assay
Trans-well invasion chambers were pre-coated with 500 ng/mL Matrigel (Corning, Corning, USA) for 6 hours. The medium at the bottom of the 24-well plate was replaced with fresh medium containing 10% FBS, and 2 × 10 4 cells in serum free media were added to the top chamber. After incubation for 24 hours at 37°C, the cells were then fixed with 4% formaldehyde for 2 minutes. The chambers were then washed with PBS and stained with 1% crystal violet for 10 minutes. After cleaning and drying the chamber membrane, cells were counted using an inverted microscope (Zeiss, Oberkochen, Germany).

| ELISA
Concentration of IGF-1 and IGF-2 was measured using human ELISA kit (Abnova, Taipei City, Taiwan) according to the manufacturer′s instructions. For the sample preparation, 2.5 μg of total lysate protein and 5 μg of total secreted protein were quantified by BCA assay. Finally, concentration of IGF-1 and IGF-2 was calculated at 450 nm using a microplate reader.

| Western blotting
For Western blotting, control and BI836845-treated cells were collected 24 hours after treatment. A total of 30 μg of whole cell protein extracts in RIPA lysis buffer were size-fractionated using 10% SDS polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane (Bio-Rad, Hercules). Membranes were blocked with 5% non-fat dry milk/TBS-T for 1 hour at room temperature and then incubated overnight with primary antibodies at 4°C. Membranes were washed five times with TBS-T and incubated with horseradish peroxidase-conjugated secondary antibody for 1 hour at room temperature. Blots were washed five times again, and protein signals were enhanced and detected using a chemiluminescence detection kit (Santa Cruz, Dallas). The intensity of the bands was normalized using α-tubulin with ImageJbased quantification.

| Flow cytometry
For apoptosis analysis, 5 × 10 5 cells were seeded in 6-well plates and incubated for 24 hours at 37°C in an incubator. BI836845 was directly added to the wells, and cells were incubated for 24 hours and 48 hours and collected in 15 mL tubes. After PBS washing, cells were double-stained with recombinant Pacific blue-conjugated Annexin V and propidium iodide (PI). Cells were gently vortexed and incubated for 15 minutes at room temperature in the dark. Flow cytometry analysis was performed using a FACS LSRII (BD biosciences, Franklin Lakes) with CellQuest software. Ten thousand single cells were gated and analysed for apoptosis.
Cell cycle analysis was performed in two different conditions. In the first condition, ordinary cell cycle analysis was performed. Cells were seeded in 60-mm 2 dishes and incubated in media with or without BI836845 after 24 hours. After further incubation for 24 and 48 hours, cells were harvested. In the second condition, cells were synchronized in G0/G1 phase after culturing in confluent monolayers under serum starvation during 48 hours. After starvation, wells were incubated with 10% FBS-containing media, and 10 μg/mL BI836845 was added to each well. Cells harvested immediately after the 48 hours of serum starvation were labelled 0 hours; additional aliquots were harvested at the indicated time-points. Cells harvested at either the first or the second condition were fixed in cold 70% ethanol for more than 24 hours and stained with PI (BD biosciences, Franklin Lakes. DNA contents were determined using FACS LSRII (BD biosciences), and gated 20,000 events/sample were collected for cell cycle analysis.

| Statistical analysis
Continuous data were analysed using the Student t test. Differences were considered statistically significant when P < 0.05. The significance of dose-or time-dependent change was calculated by twoway ANOVA with the use of IBM SPSS statistics.

| Expression of IGF-related genes and proteins in EBVaGC
To evaluate IGF-related gene and protein expression, the baseline expression levels in AGS and AGS-EBV cells were first determined.
Comparing AGS and AGS-EBV cells, no significant differences in the mRNA levels of IGF-1R, IGF-1, IGF-2 and IGFBP-6 were observed.
Western blot analysis showed that, although total IGF-1R protein levels in AGS-EBV were 17.4 ± 28.8% lower than those in AGS cells, phospho-IGF-1R levels were 38.9 ± 28.1% higher than those in AGS JEONG ET AL.
To compare the expression levels of ligands, lysate and secreted IGF-1 or IGF-2 were measured using ELISA ( Figure 1C

| Effect of BI836845 on proliferation, sensitivity and invasion of AGS and AGS-EBV cells
To evaluate the effect of EBV infection on AGS cells, proliferation assay was first performed. On days 6 and 7, proliferating AGS cells were at significantly higher number than AGS-EBV cells (P < 0.01 and P < 0.001 at days 6 and 7, respectively; Figure 2A and IGFBP-6 was observed ( Figure 3A).
As can be seen in Figure 3B

| Apoptosis or cell cycle arrest was not observed after BI836845 treatment in AGS-EBV
We performed Annexin V/PI double-staining assay to determine whether the reduction in cell viability was induced by apoptosis. As shown in Figure 4A

| BI836845 inhibits progression of AGS-EBV cells into S phase
To further investigate the possible mechanism of proliferation inhibition, we performed cell cycle analysis with G0/G1 synchronization using serum starvation, because no cell cycle arrest could be observed in normal culture conditions. After synchronization, the G0/G1 proportion in AGS and AGS-EBV cells was 62.0% and 58.2%, respectively. Interestingly, cell cycle progression of AGS-EBV cells was faster than that of AGS cells ( Figure 5A).

| DISCUSSION
To optimize genome replication, most viruses manipulate the host cell environment and the cell cycle. Many of the EBV-specific viral factors were identified due to their functional impact on various cell types. 22,23 Most of these studies, however, limited their focus to viral gene expression and their function. In our study, we extend our focus to the biological changes in gastric cancer cells after EBV infection and also on possible target signalling pathways.
The EBV-infected gastric cancer cell line (NUGC-3) and EBVpositive NPC cell line showed increased IGF-1 mRNA and secreted IGF-1 levels compared to parental cell lines, which suggest that IGF-1 mediates cell proliferation. 17,24 In our results, IGF-1 mRNA levels did not increase, although secreted IGF-1 levels increased in AGS-EBV cells ( Figure 1C). As well known, secreted IGF-1 stimulates IGF-1R phosphorylation. Although the mRNA and protein expression levels of most genes we examined were down-regulated in AGS-EBV cells, phospho-IGF-1R levels were increased in AGS-EBV cells. When EBV infects cells, it regulates growth of the host cell and activates selective pathways to increase the efficiency of viral factor synthesis. 25,26 Our results suggest that EBV may use the IGF-1R pathway to adapt to the host environment in the AGS cell line. IGFBPs are known to increase IGF-1 and IGF-2 stability and transport in the tissue, and IGFBP-3 and IGFBP-6 stabilize IGF-1 and IGF-2, respectively. 27 In our experiments, resistance to BI83685 in AGS cells was observed in concordance with increased IGFBP-3 and IGFBP-6 secretion.
Levels of mRNA or cytosol IGFBPs are associated with tumour aggressiveness, and the underlying mechanisms are complex and include both IGF-dependent and IGF-independent pathways. 28 In our results, we observed that IGFBP-3 mRNA levels were greatly increased in AGS-EBV cells ( Figure 1A). IGFBP-3 is known to have various binding partners and is associated with different cellular functions. 29 Several studies have reported an association of elevated mRNA expression levels of IGFBP-3 with cell growth inhibition, namely in breast and prostate cancer cell lines. These mechanisms are mediated by IGF-independent pathways. [30][31][32] In our result, IGFBP-3 mRNA expression was reduced only in the AGS-EBV treatment group. IGF-independent cell death by IGFBP-3 mRNA regulation may be increased by EBV infection, which in turn increases IGF dependence and BI836845 sensitivity on AGS-EBV treatment group.
We also observed that while cell cycle progression was faster in AGS-EBV cells compared to that of AGS cells, cell proliferation was in fact slower in AGS-EBV (Figures 2A and 4A). The number of necrotic AGS-EBV cells was increased by fourfold compared to AGS  Figure 3D). EBV infection of AGS cells led to suppression of most baseline protein expression, a common characteristic of virus-infected cells. Particularly, expression of EMT-related molecules has been shown to be regulated by EBV-specific molecules. 35 In our result, mRNA levels of IGFBP-6 in AGS cells increased when treated with BI836845 (Figure 3A), and AGS cells treated with 10 μg/mL of BI836845 showed 3 times higher ability for invasion than the control group ( Figure 2D). In contrast, mRNA level of IGFBP-6 in AGS-EBV was decreased after BI836845 treatment. Interestingly, high mRNA level of IGFBP-6 promotes cancer migration and invasion in an IGF-independent manner. 36 It has been shown that IGF-1-dependent secretion of IGFBP-3 induces angiogenesis and positively regulates the expression of proangiogenic molecules. 37 In our results, IGFBP-3 secretion was only observed in treated AGS cells and their invasion to the microenvironment was increased compared to AGS-EBV. IGFBP-3 and IGFBP-6 have very complex IGF-dependent and IGF-independent functions, which remains mostly unclear. EBV-infected AGS cells may regulate their growth and invasiveness using IGFBP-3 and IGFBP-6 inside and outside cells. Inhibition of vimentin expression, independent of other EMT proteins Snail and E-cadherin, was observed in the AGS-EBV treatment group ( Figure 3D). This suggests that vimentin might play an independent role in the regulation of AGS-EBV cell invasiveness.
As mechanistic studies of IGFBP-3 and IGFBP-6 were not accomplished in our study, future studies will be needed to evaluate the intra-and extracellular functional activity.

| CONCLUSIONS
In our study, we demonstrated that the IGF-1R-related pathway dependency of both proliferation and invasion was changed by EBV infection in a gastric cancer cell line. Moreover, EBV-infected cancer cells became sensitive to BI836845. Also we suggested that regulation of IGFBP-3 and IGFBP-6 had important roles on proliferation and invasion of EBVaGC. Our results suggest that although limited in scope due to validation in a single EBV-infected gastric cell line, IGF-1R pathway inhibition might be an effective therapeutic target in EBVaGC.