Contribution of PGAP3 co‐amplified and co‐overexpressed with ERBB2 at 17q12 involved poor prognosis in gastric cancer

Abstract The locus at 17q12 erb‐b2 receptor tyrosine kinase 2 (ERBB2) has been heavily amplificated and overexpressed in gastric cancer (GC), but it remains to be elucidated about the clinical significance of the co‐amplification and co‐overexpression of PGAP3 gene located around ERBB2 in GC. The profile of PGAP3 and ERBB2 in four GC cell lines and tissue microarrays containing 418 primary GC tissues was assessed to investigate the co‐overexpression and clinical significance of the co‐amplified genes, and to evaluate the impact of the co‐amplified genes on the malignancy of GC. Co‐amplification of PGAP3 and ERBB2 accompanied with co‐overexpression was observed in a haploid chromosome 17 of NCI‐N87 cells with double minutes (DMs). PGAP3 and ERBB2 were overexpressed and positively correlated in 418 GC patients. Co‐overexpression of the PGAP3 and ERBB2 was correlated with T stage, TNM stage, tumour size, intestinal histological type and poor survival proportion in 141 GC patients. In vitro, knockdown of the endogenous PGAP3 or ERBB2 decreased cell proliferation and invasion, increased G1 phase accumulation and induced apoptosis in NCI‐N87 cells. Furthermore, combined silencing of PGAP3 and ERBB2 showed an additive effect on resisting proliferation of NCI‐N87 cells compared with targeting ERBB2 or PGAP3 alone. Taken together, the co‐overexpression of PGAP3 and ERBB2 may be crucial due to its significant correlation with clinicopathological factors of GC. Haploid gain of PGAP3 co‐amplified with ERBB2 is sufficient to facilitate the malignancy and progression of GC cells in a synergistic way.


| INTRODUC TI ON
Genomes instability, including gain at chromosome 17q12, is frequently observed in gastric cancer (GC). Amplified DNA can be organized as chromosomal homogeneously staining region (HSR) 1 or extrachromosomal double minutes (DMs) 2 or distributed at various locations in the genome. These amplifications are usually correlated with oncogene overexpression and clinical tumour aggressiveness. 3 Furthermore, amplicons pinpoint candidate oncogenes as they are a visible representation of the genes whose expression promotes tumour growth. 4 While some amplicons may harbour known oncogenes 'driving' amplification, amplicon rarely comprises single gene in tumour. The potential functional contribution of co-amplified genes remains largely unexplored. 5 Studies have found that multiple expressed genes presented usually in a single amplicon may provide a growth advantage to cells. 6 Therefore, the identification of genes with increased expression in a core amplified region could be essential.
A combined investigation of the genes refer to erb-b2 receptor tyrosine kinase 2 (ERBB2) could facilitate the identification of a target therapeutic regimen for GC patients. 7 The locus at 17q12 was well known because it is where ERBB2 gene located, which has been heavily amplificated and overexpressed in tumours (including GC). 8 PPP1R1B-ERBB2-GRB7 locus at 17q12 was frequently amplified in GC, 9 it is generally considered to be an oncogenomic recombination hotspot around the locus at human chromosome 17q12. 10 In GC, the frequency of ERBB2 amplification has been reported to be in the range from 7% to 27%, and this amplification has been established to correlate with an intestinal type histology and poor survival. 11 Studies demonstrated that the rate of ERBB2 overexpression is 6%-30% in GC. 12 Previous studies indicated that the ERBB2 amplicon included STARD3 and GRB7 genes, and the co-localization and co-amplification of these genes with ERBB2 suggested a region of genomic amplification in breast cancer that extends beyond ERBB2. 13,14 In addition to STARD3 and GRB7, post-GPI attachment to proteins phospholipase 3 (PGAP3) is also one of the genes which are located near ERBB2 around the locus at 17q12. PGAP3 and ERBB2 were found to be amplified and overexpressed in GC by array-based comparative genomic hybridization (aCGH) and gene expression microarray analyses. 9,15 However, the implication of PGAP3 co-amplified and co-overexpressed with ERBB2 at 17q12 in GC remains largely unknown.
Our previously unpublished result by aCGH indicated the highest amplification of ERBB2 amplicon on 17q12 in GC cell line NCI-N87 with DMs. In this study, we investigated the clinical and biological implications of PGAP3 co-amplification with ERBB2 in gastric tumorigenesis in vitro, and demonstrated that the aberrant expression of these genes is associated with the malignancy of GC cells.

| Cell lines and tissue chips
The GC cell lines NCI-N87 and AGS were obtained from American

| Metaphase chromosome preparation and karyotype analysis
Metaphase spread preparation was done following standard protocol. Briefly, cells were exposed to colchicine at the final concentration of 0.2 μg/mL for 1-1.5 h before harvesting. Supernatant was removed after centrifugation, then treated with 0.075 mol/L preheated KCl solution for about 14-15 min at 37°C, followed by 15 min fixation. After centrifugation, cells were resuspend and dropped onto slides. Before stained with Giemsa, the karyotypes of NCI-N87 cells were treated with trypsin at 37°C. Images were obtained using a microscope (NIKON YS100), the karyotypes of the cells were determined using standard cytogenetic techniques.

| Real-time PCR
Genomic DNA from GC cell lines was isolated with the QIAamp DNA mini kit (Qiagen, 51306, Germany) according to the manufacturer's protocol. Total RNA from GC cell lines was extracted using TRIzol reagent (Invitrogen, 66003, USA) and reverse transcribed into cDNA using the Transcriptor First Strand cDNA Synthesis Kit (Roche, 04897030001, Germany). The sequences of the PCR primers used in the study are listed in Table S1. All PCR reactions were performed in Roche Light Cycler 480 system (Roche, Switzerland). The relative level of each gene were normalized by β-actin, and was shown as fold differences (2 −ΔΔCT ).

| Western blotting
Cells were retrieved into RIPA buffer containing protease inhibitor.

| Cell cycle and apoptosis assay
Cells were seeded in a six-well plate with 1 × 10 4 cells/well, followed by transfected with lipofectamine RNAiMAX transfection reagent.
Three groups were set up in total. Negative control (NC) group was transfected with si-NC, siPGAP3 group was transfected with siPGAP3-189, and siERBB2 group was transfected with siERBB2-2.
Cells were digested 48 h after transfection, PI staining or Annexin V and PI double-staining were performed according to Annexin-V-FLUOS Staining Kit (Roche, 11988549001, Germany), and the percentage of cells was analysed by flow cytometry.

| Cell proliferation assay, colony formation and invasion assay
Cells proliferation were measured using CCK-8 solution (Dojindo, CK04, Japan) for 7 days over time. Cells were seeded into a 96-well cell culture plate with 3 × 10 3 cells/well. The dosage of Lipofectamine RNAiMAX was 0.3 μL/well, and the dosage of siRNA was 3 pmol/ well. The medium was removed 72 h after transfection, and cells were cultured with RPMI-1640 medium containing 10% CCK-8 kits for 2 h. The absorbance value was obtained at 450 nm by the microplate tester.
In colony formation assay, 1 × 10 4 cells were seeded into a sixwell plate after transfection. After 2 weeks, the number of cell colonies was determined by fixed staining with Giemsa to measure the ability of cells to proliferate.
In cell invasion assay, cells in each group were transfected with siRNA (siPGAP3-189, siERBB2-2), 4 × 10 3 cells/well were seeded into the upper layer of Transwell chamber (BD Biosciences, 353097, USA) precoated with 100 μL Matrigel. 600 μL medium containing 10% FBS was added to the lower layer of the chamber. After 48 h, the cells in the lower membrane were fixed in 4% paraformaldehyde, stained with DAPI and photographed under a fluorescence microscope.
The invasion ratio was calculated as the percentage of invasive cells across the chamber to the total number of seeded cells.

| Data collection and analysis
The expression levels of PGAP3 and ERBB2 were obtained from the Cancer Cell Line Encyclopedia (CCLE) database, which contains data on 45 cell lines involving GC. This information was accessed through the website https://sites.broad insti tute.org/ccle. The RNA levels of genes in 408 GC tissues were analysed using Pearson correlation analysis, along with 211 normal gastric tissues (36 normal and 175 stomach) through the Gene Expression Profiling Interactive Analysis (GEPIA) databases available at http://gepia2.cance r-pku.cn.  metastatic GC cell lines (62.22%) ( Figure S3), and after ERBB2 knockdown, PGAP3 decreases in intestinal (tubular) GC cell lines ( Figure S4).

| Statistical analysis
The co-amplification and co-overexpression of PGAP3 and ERBB2 implicated their potential role in gastric tumour pathogenesis.

| Co-overexpression of PGAP3 and ERBB2 in primary GCs
To assess the significance of PGAP3-ERBB2 co-amplification in primary GC, the expression of these genes were analysed by IHC in tissue microarrays (Figure 2A) Figure 2B). Pearson test revealed a significant positive correlation of expression between PGAP3 and ERBB2 (r = 0.264, ***p < 0.001) ( Figure 2C), which was consistent with the data of RNA sequencing (r = 0.75, ***p < 0.001) from GEPIA Database ( Figure 2D).
To further clarify the effect of co-overexpression of the PGAP3 and ERBB2 in GC patients, we assigned the GC patients into four groups according to gene expression in GC tissues as follows: PGAP3-low/ERBB2-low, PGAP3-low/ERBB2-high, PGAP3-high/ ERBB2-low and PGAP3-high/ERBB2-high ( Figure 2E). PGAP3-high/ ERBB2-high group means the co-overexpression of the PGAP3 and ERBB2 in GC tissues. Co-overexpression for PGAP3 and ERBB2 was found in 141 GC patients, in matched NG tissues (121 and 116) and IM tissues (9 and 10), respectively. Immunoreactivity of PGAP3 and ERBB2 showed stronger cytoplasm/membrane staining of the proteins in GC tissues (IHC score: 223.01 ± 4.607 and 194.50 ± 3.921), IM tissues (IHC score: 168.4 ± 7.937 and 117.6 ± 11.264) and NG tissues (IHC score: 146.67 ± 6.38 and 96.08 ± 6.357). The expression of PGAP3 and ERBB2 were gradually increased in NG tissues, IM tissues and GC tissues ( Table 2), with the expression of PGAP3 and ERBB2 in GC tissues significantly higher than that in NG tissues (***p < 0.0001 and ***p < 0.0001) and IM tissues (ERBB2, *p = 0.0154) ( Figure 2F) in a gradual way ( Table 2).

| Co-overexpression of ERBB2 and PGAP3 predicts poor survival in primary GCs
In all 418 GC tissues, the high expression rate for PGAP3 and ERBB2 was found in 322 (77.03%) and 160 (38.28%) GC tissues, respectively. Furthermore, the correlation analysis between gene expression and clinicopathological features indicated that the expression of PGAP3 was significantly correlated with Lauren type (Table S3, *p = 0.013), and the expression of ERBB2 was significantly correlated with T stage (*p = 0.015), tumour size (*p = 0.047) and Lauren type (*p = 0.016) (Table S4). Subsequently, both univariate and multivariate analyses indicated that PGAP3 and ERBB2 were independent prognostic factors for patients with GC (PGAP3: *p = 0.013 and *p = 0.042; ERBB2: **p = 0.008 and *p = 0.033).
Besides, tumour size, TNM stage, N metastasis and M metastasis were also significantly correlated with the prognosis of GC patients, as an independent factor of GC patients (Tables S5 and S6).
The expression of PGAP3 and ERBB2 was more frequently found in the intestinal histological type, and the expression of PGAP3 in intestinal type of GC was significantly higher than that in diffuse type of GC (***p < 0.001) ( Figure 3A). Kaplan-Meier survival analysis in 343 GC patients unravelled that the proportion of total survival is 35%, and overexpression of PGAP3 or ERBB2 has significant poor survival proportion compared with low expression in GC patients, respectively (32.3% and 44.6%, *p = 0.011; 27.4% and 40.6%, **p = 0.007) (Tables S7 and S8, Figure 3B).

| Phenotypic plasticity of PGAP3 and ERBB2 knockdown by siRNA in NCI-N87 cells
To determine whether the PGAP3 co-amplified with ERBB2 is a contributing factor to oncogenesis. SiRNA-mediated knockdown  Figure 4G). These data suggest that PGAP3 might be one of the driver genes within the 17q12 amplicon, and may constitute a therapeutic target for GC patients harbouring this amplification.

TA B L E 3
Correlation between PGAP3 and ERBB2 co-expression and clinicopathological characteristics in GC patients.

| Assessing additive effects by targeting both PGAP3 and ERBB2
To further investigate whether PGAP3 and ERBB2 might play an additive role. NCI-N87 cells were co-transfected with siRNAs targeting PGAP3 and ERBB2. Compared to NC, simultaneous silencing of PGAP3 and ERBB2 led to a clear additive inhibition of NCI-N87 cell proliferation ( Figure 5A). And compared to targeting PGAP3 or ERBB2 alone, simultaneous silencing of PGAP3 and ERBB2 begun to led a clear additive inhibition of cell proliferation in 6-7 days ( Figure 5B).

| DISCUSS ION
Amplification and/or deletion at chromosome 17 are frequently encountered in GC. In previously studies, we revealed the candidate tumour suppressor genes involved in GC pathogenesis by loss of heterozygosity (LOH) on chromosome 17 in primary GCs. 17,18 That our previous data of aCGH indicated the highest amplification around ERBB2 at chromosome 17q12 in GC cell line NCI-N87, has switched our focus to the amplification genes at 17q12 in GC.
Gene amplification represents one of the molecular mechanisms of oncogene overexpression in the tumour. 20 There has been much evidence that amplified genes in cancer cells usually reside on chromosomal HSR or extrachromosomal DM. 21 The co-overexpression and co-amplification of PGAP3 and ERBB2 in NCI-N87 cells suggests that the amplicon might be located in DMs in NCI-N87 cells, which carries DMs. 22   In our study, co-overexpression of PGAP3 and ERBB2 was found in 88.13% ERBB2 high expressed GC patients and the expression of two genes were positively correlated, suggesting that PGAP3-ERBB2 co-expression may be common in GC. Moreover, we also found that Alternatively, a novel molecular subtype could be established based on distinct tissue gene expression patterns involving PGAP3 and ERBB2 in GC, which could be novel target for personalized treatment. 34 Additionally, PGAP3 silencing alone has reversed the malignant phenotype of NCI-N87 cells in vitro, implying that PGAP3, not just a cooperator with ERBB2, may also play an important role in the progression of GC.
Co-amplification caused co-overexpression of PGAP3 and ERBB2 may have functional implication in GC pathogenesis. Our in vitro assays indicated that targeting PGAP3 or ERBB2 alone could significantly reduce cell proliferation and metastasis. Whether targeting combinations of PGAP3 and ERBB2 might provide additive effects? By targeting PGAP3 together with ERBB2, cell proliferation was significantly abrogated in an additive effect, compared to targeting PGAP3 or ERBB2 alone, suggesting that expression of coamplified genes is needed to sustain the growth of GC cells. Our data also first discovered that gain of the gene copies at 17q12 by chromosomal segments amplification event in haploid chromosome might be associated with GC progression, which was previously speculated. 9,15

| CON CLUS ION
On the whole, the enhanced co-expression of ERBB2 and PGAP3 in NCI-N87 cells implied their important function in GC cells.
Furthermore, the specific co-overexpression of ERBB2 and PGAP3 in intestinal type in primary GCs suggested that the co-overexpression of the co-amplified genes has a closely relationship with the clinical pathological types. Our finding underlines that therapy targeting PGAP3-ERBB2 amplicon may provide additional benefit over targeting ERBB2 alone for patients with tumours carrying 17q12 amplification. Further study is necessary to determine whether GC cells may become addicted to the amplification of several genes that reside in the ERBB2 amplicon, especially in vivo.

AUTH O R CO NTR I B UTI O N S
Dong Wang: Investigation (equal); methodology (equal); software (equal); writing -original draft (equal). Siyu Hao: Investigation (equal); methodology (equal); software (equal); writing -original draft (equal). Hongjie He: Software (equal); visualization F I G U R E 4 Effects of siRNA-mediated knockdown of PGAP3 and ERBB2 in NCI-N87 cells. (A, B) Validation of siRNA knockdown of PGAP3 or ERBB2 by Real-time PCR and Western blot, respectively. (C, D) Silencing of PGAP3 or ERBB2 led to significant decrease in cell viability. (E, F) Inhibition of PGAP3 or ERBB2 caused cell cycle arrest at G1 and increased apoptosis measured by flow cytometry. (G) Representative images of invading cells through the Matrigel-coated membrane. The invaded cells were quantified as a percentage of the original seeded cells. Magnification, ×200. (compared to NC, ns, no significance; *p < 0.05, **p < 0.01, ***p < 0.001).

F I G U R E 5
Effects of siRNA-mediated simultaneous knockdown of PGAP3 and ERBB2 on proliferation in NCI-N87 cells. (A) Simultaneous silencing of PGAP3 and ERBB2. (B) Silencing of PGAP3 and ERBB2 simultaneously and respectively (compared to NC, ns, no significance; *p < 0.05; ***p < 0.001).