RNF43 and PWWP2B inhibit cancer cell proliferation and are predictive or prognostic biomarker for FDA-approved drugs in patients with advanced gastric cancer

Background: Abnormal regulation of genes has been closely related to gastric cancer. The characterization of gastric cancer has necessitated the development of new therapeutics as well as the identification of prognostic markers to predict the response to novel drugs. In our study, we used RNA sequencing analyses to show that on gastric cancer tissues to identification of gastric cancer prognostic markers. We specifically chose to study RNF43 because it inhibits gastric cancer-related Wnt/β-catenin signaling by interacting with Wnt receptors. PWWP2B was chosen because it is a gene which is downregulated in gastric cancer. Methods: Utilizing RNA sequencing analysis, we evaluated the mRNA expression profile in gastric cancer patients. Also, we used HAP1 cells which is a human near-haploid cell line derived from the male chronic myelogenous leukemia cell line KBM-7. These cell line has one copy of each gene, ensuring the edited allele will not be masked by additional alleles. We investigated the screening of 1,449 FDA-approved drugs in HAP1, HAP1 RNF43 KO and HAP1 PWWP2B KO cells. RNA sequencing data reveals that RNF43 and PWWP2B expression were down-regulated in recurrence gastric cancer patients. Next, we investigated the anti-cancer effects of selected drugs in RNF43 and PWWP2B down-regulated MKN45 gastric cancer cells and xenograft model. Results: Among these FDA-approved drugs, three drugs (docetaxel trihydrate, pelitinib and uprosertib) showed strong inhibitory effects in RNF43 KO cells and PWWP2B KO cells. In MKN45 xenograft model, tumor volumes were significantly reduced in the docetaxel trihydrate, uprosertib or pelitinib-treated group. Our data demonstrated that RNF43 and PWWP2B are a biomarker that predict recurrence of gastric cancer. Conclusions: Our findings suggest that docetaxel trihydrate, uprosertib and pelitinib could be used as novel therapeutic agents for the prevention and treatment of gastric cancer with a decrease in RNF43 and PWWP2B expression.

cancers are loss-of-function events that defy standard inhibitor-based drug screening strategies. One strategy for overcoming the loss of function is to perform high-throughput screening using target gene knockout (KO) cell lines.
RNF43 encodes a transmembrane ubiquitin E3 ligase and is a tumor-suppressing gene that suppresses the Wnt/β-catenin signaling pathway, which is often the cause of cancer [8][9][10]. Loss-of-function mutations in RNF43 promote reinforced cell proliferation and result in neoplastic transformation [11].
Recent studies revealed that RNF43 suppresses proliferation and induces apoptosis in gastric carcinoma cells [8,12].
In this study, we applied an RNA sequencing (RNA-seq) approach to identify RNF43 (i.e., verify a known marker) and PWWP2B (i.e., explore a novel marker) genes differentially expressed in gastric cancer and adjacent normal tissues from 34 patients. To identify Food and Drug Administration (FDA)-approved drugs that selectively target cancer cells with inactivated RNF43 and PWWP2B genes, we performed a high-throughput screening of 1,449 drugs in HAP1, HAP1 RNF43 KO, HAP1 PWWP2B KO, and SNU620 cells. This study was conducted to identify anti-cancer drugs in gastric cancer with an aberrant decrease in RNF43 and PWWP2B expression.

Methods
Study subjects and gastric tissue specimen collection Gastric cancer and adjacent normal tissues obtained from 34 patients who underwent initial surgery at Hallym University Sacred Heart Hospital from March 2014 to July 2015, were selected as the discovery cohort for RNA-seq. All cases were prospectively followed up for at least 3 year. Table 1 summarizes the discovery sets. This study was approved by the Ethics Committee of Hallym University Sacred Heart Hospital (2015-I078). Written informed consent was obtained from all of the participants.
RNA-seq and differentially expressed gene (DEG), single-nucleotide polymorphism (SNP), and insertions/deletions (indels) analyses Gastric cancer and adjacent normal tissues from 34 patients were subjected to RNA-seq. Total RNA was extracted with TRIzol Reagent. Beads containing oligo (dT) were used to isolate poly(A) mRNA from total RNA. mRNA was fragmented, and rst-strand cDNA was synthesized using random hexamer primers. Second-strand cDNA was synthesized using dNTPs, RNase H, and DNA polymerase I. Next, short doublestranded cDNA fragments were ligated to Illumina sequencing adaptors. DNA fragments were gel-puri ed and ampli ed by polymerase chain reaction (PCR). The ampli ed library was sequenced on an Illumina HiSeq 2500 sequencing machine. The raw reads were saved in the FASTQ format, and the dirty raw reads were removed before analyzing the data. Reads that could be uniquely mapped to a gene were used to calculate gene expression levels, which were measured based on the number of reads per kilobase of transcript per million mapped reads. We identi ed DEGs between paired tumor and normal samples and considered P ≤ 0.001 as signi cant. The comprehensive detection of SNPs and small indels followed the method described by Yoon et al. (2013) [13] and Lee et al (2014) [14], with slight modi cations.

Cell lines and cell culture
The male and female human cell lines used in this study. HAP1, HAP1 RNF43 KO, and HAP1 PWWP2B KO cells (derived from 40year male chronic myelogenous leukemia) were obtained from Horizon. SNU620 (derived from 59year female gastric adenocarcinoma), MKN28 (derived from 37year male gastric tubular adenocarcinoma), MKN45 (derived from 62year female gastric adenocarcinoma), and KatoIII (derived from 57year male Signet ring cell gastric adenocarcinoma) cells were obtained from the Korean Cell Line Bank. The cells were grown and maintained under conditions of 100% humidity and 5% CO 2 at 37°C in Iscove's modi ed Dulbecco's medium (IMDM) or RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) and 1% streptomycin and penicillin (Invitrogen Life Technologies, Rockville, MD, USA). The cells were plated onto tissue culture asks (T-75 cm 2 ) at a density of 1 × 10 7 /mL in hormonally de ned IMDM or RPMI 1640 medium as described previously. The medium was changed every 3 days until the cells reached 80-90% con uence, at which point they were used in the experiments.

Morphological evaluation
Microscopical studies of the HAP1, HAP1 RNF43 KO, and HAP1 PWWP2B KO cells were carried out with a Microscope. The magni cation was 400 fold. Immunohistochemical analysis Immunohistochemical (IHC) analysis of RNF43 was performed. Tissue sections were treated with 3 % hydrogen peroxide, and nonspeci c binding sites were blocked. The sections were incubated with anti-RNF43 antibodies (AP13204B, ABGENT, San Diego, CA, USA). An automatic immunostainer (BenchMark XT; Ventana Medical Systems, Inc, Tucson, AZ, USA) and UltraView Universial DAB detection kit (Ventana Medical Systems) were used for immunostaining.

Cell proliferation assay
The proliferation of HAP1, HAP1 RNF43 KO, and HAP1 PWWP2B KO cells was assessed using the MTT assay. Cells were cultured in 96-well plates (2 × 10 4 /mL) for 0, 24, 48, and 72 h. MTT solution (5 mg/mL) was added at the end of incubation, which was continued for 4 h. The reaction was terminated by adding a detergent reagent. The absorbance was read at 570 nm using a microplate reader (Synergy 2 Multi-Mode Microplate Reader; BioTek, Winooski, VT, USA).
Colony forming assay HAP1, HAP1 RNF43 KO, and HAP1 PWWP2B KO cells were diluted and seeded at a density of approximately 5,000 cells per well in 6-cm plates. After incubation for 7 days, colony formation and growth were visualized with crystal violet staining. After the wells were photographed, the dye was solubilized with methanol and the optical density was measured at 570 nm using a microplate reader.
Cell migration assay HAP1, HAP1 RNF43 KO, and HAP1 PWWP2B KO cells were diluted and seeded at a density of approximately 1 × 10 5 cells per well in 6-cm or 24-well plates. After incubation for 1 day, a straight scratch was made on the cells using a P200 pipette tip. The cells were then washed with PBS and further cultured with or without docetaxel trihydrate, pelitinib, and uprosertib in IMDM. After incubation for 0, 48, and 72 h, the gap width of the scratch re-population was photographed and then compared with the initial gap size at 0 h.

High-throughput drug screening
An FDA-approved compound library of 1,448 drugs was purchased from Selleck Chemicals. INC280 was supplied from Novartis (Basel, Switzerland). HAP1, HAP1 RNF43 KO and HAP1 PWWP2B KO cells were seeded at a density of 2,000 or 3,000 cells per well in 384-well, clear-bottom culture plates with 20 µL IMDM or RPMI 1640 medium containing 10% FBS for 24 h. Then, 10 µM of FDA-approved drug was added to the wells, and the cells were incubated for an additional 48 h. Control cells were not exposed to drugs. On the day of the proliferation assay, the medium was removed, 20 µL of fresh medium was added to each well of the 384-well plates, followed by 5 µL of MTS solution (Cell Titer 96 Aqueous One Solution Cell Proliferation Assay Kit; Promega, Madison, WI, USA), and the plates were incubated at 37°C for 1 h in a humidi ed environment with 5% CO 2 . The absorbance was read at 490 nm using a PerkinElmer (Waltham, MA, USA) EnVision luminescence microplate reader. Data were validated using the Z factor analysis. The percentage inhibition was expressed as [viability level of test samples/viability level of control)] × 100.

Con rmatory growth inhibition assays
The half maximal inhibitory concentrations of the selected drugs of HAP1, HAP1 RNF43 KO and HAP1 PWWP2B KO cells were measured using the MTS assay for selected drugs at concentrations of 40, 20, 10, 5, 2.5, 1.25, 0.625, and 0.3125 µM for 48 h. On the day of the proliferation assay, medium was removed, and 100 µL of fresh medium was added to each well of 96-well plates, followed by 20 µL of MTS solution, and the plates were incubated at 37°C for 1 h in a humidi ed environment with 5% CO 2 .
The absorbance was read at 490 nm using a microplate reader (Synergy 2 Multi-Mode Microplate Readers; BioTek, Winooski, VT, USA). The IC 50 values were determined after tting growth inhibition curves to dose-response curves using GraphPad Prism software (GraphPad Software Inc., CA, USA).

In vivo tumor growth inhibition studies
All the experiments and animal handling procedures in this study were approved by the Animal Experimental Ethics Committee of the Asan Medical Center, Seoul, Korea. Six-week-old male BALB/cnu/nu mice (Joongang Laboratory Animal Inc., Seoul, Korea) were housed in cages, and maintained at 23ºC with a 12-h light/dark cycle under speci c pathogen free conditions. Each mouse was inoculated subcutaneously (s.c.) into the right ank with either 1 x 10 7 cells/mouse of human gastric cancer cell line MKN45. When the average s.c. tumor volume reached 100 mm 3 , the mice were randomly divided into various treatment and control group (5 mice per group). Tumor size was measured twice every week with caliper (calculated volume = shortest diameter 2 x longest diameter/2). Body weight and tumor size were recorded twice every week. After three weeks, the mice were sacri ced.

Statistical analysis
The data were statistically analyzed using Prism 5 (GraphPad Software Inc.). All values are presented as means ± the standard error of the mean. Statistical signi cance was examined with the Mann-Whitney test or Fisher's exact test. A value of P < 0.05 was considered statistically signi cant.

Baseline characteristics
A total of 34 subjects were enrolled in this study to gain insight into the molecular pathogenesis of gastric cancer in a Korean population. We searched for genetic alterations using RNA-seq in gastric cancer samples and their matched adjacent normal tissues. By comparing the transcriptome sequences of the cancer tissues with their matched normal tissues, we identi ed differentially expressed genes, including RNF43 and PWWP2B. The associations of RNF43 and PWWP2B expression with clinicopathological characteristics are shown in Table 1. The subjects included 19 males (55.9%) and 15 females (44.1%), with a median age of 68.6 years (range: 44-87 years). Eleven patients had recurrence.
Low expression of RNF43 and PWWP2B (100%, P < 0.001) was signi cantly associated with recurrence ( Table 1). We found RNF43 and PWWP2B mutations in 7 of 34 tumors and 6 of 34 tumors, respectively (Fig. 1a). In addition, 34 tumors carried germline RNF43 mutations. The IHC analysis of RNF43 showed not detect of RNF43 in gastric cancer tissue with low or high expression of RNF43 gene (Fig. 1b).

Cell proliferation and migration regulation by RNF43 and PWWP2B
In light microscopy we rst investigated the cell size effect of RNF43 KO or PWWP2B KO compared to WT cells (Fig. 2a). The KO of RNF43 and PWWP2B show only visible in uence on the cell size of PWWP2B KO cells. The morphological parameters of the FACS analysis shows only altered cell size in PWWP2B KO cells (Fig. 2a). The effects of RNF43 KO and PWWP2B KO on cell proliferation, we investigated cell proliferation using MTT and colony formation assays ( Fig. 1b and 1c). RNF43 KO and PWWP2B KO increased the proliferation and colony formation of HAP1 cells. Especially, PWWP2B KO cell line formed colonies much larger than WT and RNF43 KO cells, suggesting that the PWWP2B KO might endow gastric cancer cells with accelerated tumorigenic capability. Then we investigated cell migration using the wound-healing assay. RNF43 KO and PWWP2B KO increased migration in HAP1 cells (Fig. 2d). These results suggest that the loss of RNF43 and PWWP2B may modulate cell migration.

FDA-approved drug library screen
To evaluate the effect of the 1,449 FDA-approved drugs on HAP1, HAP1 RNF43 KO and HAP1 PWWP2B KO cell viability, cells were treated with 10 µM of FDA-approved drugs for 48 h, and the inhibitory effect of the drugs was evaluated using the MTS assay. After 48 h, the cell viability of the drug-treated cells was lower than that of the untreated control cells, with 9 drugs resulting in cell viability less than 40% in HAP1 cells. The cells were treated with different concentrations of each drug for 48 h, and the optimal dose was determined by evaluating cell viability using MTS assays. Treatment with the nine drugs (aprepitant, docetaxel trihydrate, ethinyl estradiol, griseofulvin, INC280, pelitinib, pimobendan, tepotinib, and uprosertib) decreased cell viability in a dose-dependent manner (n = 3). The IC50 values of the nine drugs were determined using non-linear regression analysis (Table 2). Among these drugs, docetaxel trihydrate, pelitinib, and uprosertib showed the best inhibition rates.

Effects of three drugs on cell migration
To determine the inhibitory effects of docetaxel trihydrate, pelitinib, and uprosertib on HAP1, HAP1 RNF43 KO, and HAP1 PWWP2B KO cells, cell migration was examined by performing a wound healing assay with the respective IC50 values of the three drugs (Fig. 3a-c). The wound gaps in the cells treated with each of the three drugs were signi cantly wider than those of the untreated groups at 48 and 72 h. IC50 values of these drugs, pelitinib and uprosertib showed the best inhibitory effect.

Effects of three drugs on cell apoptosis
To evaluate the effects of docetaxel trihydrate, pelitinib, and uprosertib on cell death in HAP1, HAP1 RNF43 KO, HAP1 PWWP2B KO, SNU620, Kato III, MKN28, and MKN45 cells, apoptosis was examined by staining with annexin V-APC/PI followed by ow cytometry (Fig. 4-5). Cells were stained with annexin V-APC and PI, which assess early apoptosis and the rate of apoptosis in a cell population. Docetaxel trihydrate showed the best cell death rates in HAP1, RNF43 KO, PWWP2B KO, SNU620, and Kato III cells (Fig. 4-5). Pelitinib showed the best cell death rates in MKN45 cells (Fig. 5d). RNF43 and PWWP2B genes were weakly expressed in MKN45 cells compared with the other cell types (Additional le 1). The percentage of apoptotic cells was 10%, 15%, and 6% after exposure to docetaxel trihydrate, pelitinib, and uprosertib, respectively, while that of control cells was only 3% (Fig. 5). In contrast, these drugs were not effects MKN28 cells, which highly express PWWP2B (Fig. 5 and Additional le 1).

In vivo anti-tumor e cacy of three drugs in tumor xenografts
Prompted by the in vitro data supporting a potential anti-tumor activity of docetaxel trihydrate, pelitinib, and uprosertib, we examined the in vivo e cacy of 3 drugs on the growth of MKN45 xenograft models. As demonstrated in Fig. 6, mice bearing s.c. MKN45 tumors were treated with docetaxel trihydrate (blocking tubulin), pelitinib (blocking EGFR), and uprosertib (blocking AKT). The docetaxel trihydrate, pelitinib, and uprosertib could signi cantly inhibit tumor growth at 3 weeks with the inhibition rate 49%, 31%, and 27% in KMN45 xenografts, respectively. The docetaxel trihydrate and pelitinib were well tolerated as demonstrated by the weight gain of treatment groups over the treatment period (Fig. 6).

Discussion
Unsatisfactory treatment outcomes occur in Asian countries due to differences in the intrinsic biological factors and rate of diagnosis of gastric cancer between Western and Eastern countries, which represents a major impediment in this eld [6,7]. Therefore, in the present study, 1,449 FDA-approved drugs were screened according to the tumor characteristic status in Korea.
Aberrant regulation of Wnt/β-catenin signaling is observed in colon, ovarian, lung, prostate, liver, breast, and gastric cancers [12,[15][16][17][18][19][20]. Wnt/β-catenin signaling mediates the epithelial-to-mesenchymal transition in gastric cancer [21], a process whereby epithelial cells are converted into migratory and invasive cells [22,23]. RNF43 is encoded by Wnt target genes, and the loss of expression of this E3 ligase has been predicted to result in hyper-responsiveness to endogenous Wnt signals [11]. RNF43 is a tumor suppressor gene in mucinous ovarian cancers, mucinous pancreatic precancerous cysts, and gastric cancer [8,[24][25][26][27]. The PWWP domain is an essential component of DNMT3B that promotes tumorigenesis and contributes to aberrant DNA methylation in carcinogenesis [28]. Using a gastric cancer cohort, we retrospectively evaluated the relationship between RNF43 and PWWP2B expression levels and clinical characteristics. We found that gastric cancer recurrence patients showed downregulated RNF43 and PWWP2B (100%). In addition, we found RNF43 and PWWP2B mutations in 7 and 6 of 34 gastric cancer patients, respectively. In addition, 34 gastric cancer patients carried germline RNF43 mutations. The RNF43 and PWWP2B genes are potential biomarkers candidates for patients with advanced gastric cancer. Therefore, we examined the action of RNF43 and PWWP2B in RNF43 KO and PWWP2B KO cell lines. Our results showed that the cellular capacity for proliferation and migration was markedly increased in RNF43 KO and PWWP2B KO cell lines compared with the wild type control. These data suggest that downregulation of RNF43 and PWWP2B might promote proliferation of gastric cancer cells and could be a condition for the conversion of normal gastric epithelial cells into cancerous cells.
Used simultaneously, cell-and target-based screening procedures might be the optimal methods for promoting cancer drug discovery. They can be used to screen large numbers of drugs to determine their therapeutic potential, and correlations between target genes and clinical characteristics. Therefore, in this study, 1,449 FDA-approved drugs were screened to determine whether they could be used as therapeutic agents for the treatment of gastric cancer using growth inhibition assays of HAP1, HAP1 RNF43 KO and HAP1 PWWP2B KO cells. Among the 1,449 FDA-approved drugs tested, nine (aprepitant, docetaxel trihydrate, ethinyl estradiol, griseofulvin, INC280, pelitinib, pimobendan, tepotinib, and uprosertib) showed high inhibitory activity; therefore, these drugs were selected for further study. Table 2 presents the IC50 values and effective doses of these drugs. Docetaxel trihydrate, pelitinib, and uprosertib, which showed the highest inhibition and apoptotic rates of the tested drugs, have been shown to have therapeutic utility.
For example, AKT signaling is responsible for development of resistance in cancer to various chemotherapeutics drugs [29]. The downstream protein GSK3ß is major AKT substrate. Docetaxel trihydrate is used widely to treat recurrent or gastric cancer [30,31]. Docetaxel treatment inhibit phosphorylation state of GSK3ß [32]. Pelitinib is used to treat lung cancer [33]. Pelitinib is also an inhibitor of epidermal growth factor receptor (EGFR), making it active against putative EGFR-dependent tumor types [34,35]. In addition, Pelitinib inhibit EGF-induced activation of AKT and ERK1/2 in cancer cells [36]. Uprosertib is used in recurrent or persistent ovarian cancer, endometrial cancer, and melanoma [37,38]. Uprosertib is a broad AKT inhibitor used not only for the treatment of gastric cancer, but also for AKT-dependent cancers [39]. It is well-known that active AKT phosphorylates Gsk3β on serine 9, thereby inactivating it [40]. Gsk3β inhibits Wnt/β-catenin signaling by inhibiting AKT in gastric cancer.

Conclusion
The results of this study indicate that RNF43 and PWWP2B are differentially expressed in gastric cancers compared with normal adjacent gastric mucosa. Furthermore, a correlation between low RNF43 and PWWP2B expression and tumor recurrence was seen. Docetaxel trihydrate, pelitinib, and uprosertib have signi cant inhibitory activity in cancers with low RNF43 and PWWP2B expression; therefore, further studies are needed to elucidate their mechanisms of action should to aid in the discovery of new therapeutic agents for the treatment of gastric cancer.