RUNX1 positively regulates the ErbB2/HER2 signaling pathway through modulating SOS1 expression in gastric cancer cells

The dual function of runt-related transcriptional factor 1 (RUNX1) as an oncogene or oncosuppressor has been extensively studied in various malignancies, yet its role in gastric cancer remains elusive. Up-regulation of the ErbB2/HER2 signaling pathway is frequently-encountered in gastric cancer and contributes to the maintenance of these cancer cells. This signaling cascade is partly mediated by son of sevenless homolog (SOS) family, which function as adaptor proteins in the RTK cascades. Herein we report that RUNX1 regulates the ErbB2/HER2 signaling pathway in gastric cancer cells through transactivating SOS1 expression, rendering itself an ideal target in anti-tumor strategy toward this cancer. Mechanistically, RUNX1 interacts with the RUNX1 binding DNA sequence located in SOS1 promoter and positively regulates it. Knockdown of RUNX1 led to the decreased expression of SOS1 as well as dephosphorylation of ErbB2/HER2, subsequently suppressed the proliferation of gastric cancer cells. We also found that our novel RUNX inhibitor (Chb-M’) consistently led to the deactivation of the ErbB2/HER2 signaling pathway and was effective against several gastric cancer cell lines. Taken together, our work identified a novel interaction of RUNX1 and the ErbB2/HER2 signaling pathway in gastric cancer, which can potentially be exploited in the management of this malignancy.

RUNX1, a member of RUNX family transcription factors (RUNX1, RUNX2 and RUNX3), is an essential transcription factor mediating diverse functions in mammalian cells and modulates the transcription of its target genes through recognizing the core consensus DNA binding sequences, classically 5′-TGTGGT-3′ [18][19][20] . We have previously reported that RUNX1 is strongly required for the maintenance and progression of acute myeloid leukemia (AML) and RUNX cluster inhibition would be a novel strategy to control AML [21][22][23][24] . We have also discovered that PI polyamides which could specifically recognize and bind to RUNX binding sites strongly inhibit the proliferation of various types of cancers including gastric cancer, suggesting that RUNX1 inhibition could be a legitimate therapeutic choice in the management of gastric cancer 22 . On the other hand, Boregowda et al. have recently demonstrated that RUNX2-deficient melanoma cells display a significant decrease in three receptor tyrosine kinases, EGFR, IGF-1R and PDGFRβ 25 . This report suggests a possible existence of undiscovered roles of RUNX transcription factors in the regulation of receptor tyrosine kinases in some types of tumor cells, which prompted us to investigate the potential interaction of RUNX1 and receptor tyrosine kinases in the gastric cancer cells.
In this report, we show evidence that RUNX1 plays a key role in the maintenance of gastric cancer cells by regulating the ErbB2/HER2 signaling pathway. Our findings suggest that RUNX1 inhibition therapy potentially constitutes a novel therapeutic strategy toward HER2-positive gastric cancer.

Results
RUNX1 confers cell proliferative advantage via ErbB2/HER2 signaling. We first investigated whether depletion of RUNX1 could have an anti-tumor effect on gastric cancer cells by using the tetracycline-inducible short hairpin RNA-mediated RUNX1 knockdown system. As shown in Fig. 1a,b, silencing of RUNX1 inhibited the growth of NUGC4 and MKN45 cells and induced apoptotic cell death. NUGC4 cells were originally established from a metastatic paragastric lymph node of a 35-year-old female with signet ring cell gastric adenocarcinoma and have significantly-upregulated expression of HER2. MKN45 cells were derived from a poorly-differentiated adenocarcinoma of the stomach of a 62-year-old woman and are known for MET amplification. These results prompted us to explore the association of RUNX1 expression levels and prognosis among gastric cancer patients. We thus examined it in a gastric cancer cohort from Gene Expression Omnibus (GEO) dataset (GSE62254, n = 300). We divided the patients into the following groups; RUNX1 high (top 10% of all patients; n = 30) and RUNX1 low (bottom 10% of all patients; n = 30) according to their RUNX1 expressions and compared their clinical outcomes. Intriguingly, as shown in Fig. 1c, we found that RUNX1 high-expressing gastric cancer patients exhibited significantly worse clinical outcomes than RUNX1 low-expressing patients. To investigate the underlying molecular mechanisms of RUNX1 in the tumorigenesis of gastric cancer cells, we next conducted human phospho-RTK array in MKN45 cells transduced with shRNA targeting RUNX1 or control luciferase and screened the relative phosphorylation levels of 49 RTKs in these samples. Interestingly, as shown in Fig. 1d and Supplementary Fig. 1a, the level of the phosphorylation of ErbB2/HER2 was specifically and most profoundly decreased upon RUNX1 knockdown. To confirm our finding, we performed immunoblot experiment and validated that RUNX1-silencing indeed dephosphorylated ErbB2/HER2 in the gastric cancer cells (Fig. 1e). These data collectively indicate that RUNX1 functions as an oncogene in gastric cancers possibly through enhancing the activity of the ErbB2/HER2 signaling pathway. SOS1 is up-regulated in RUNX1 high-expressing gastric cancers. To explore the underlying molecular mechanisms of RUNX1-mediated regulation of the ErbB2/HER2 signaling pathway, we first extracted the top one-third genes (6,000 genes) that are highly-expressed in the RUNX1 high-expressing gastric cancer patients in 5 independent previously-published microarray data sets (GSE62254, GSE29272, GSE35809, GSE34942 and GSE22377). As shown in Fig. 2a,b, we found 16 genes which were consistently highly-expressed in the RUNX1 high-expressing primary gastric cancer cells. Among them, we found SOS1, a known adaptor protein associated with RTKs. Intriguingly, we found that SOS1 protein was also highly-expressed in the gastric cancer cell lines relative to the normal gastric tissue (Fig. 2c). Moreover, knockdown of RUNX1 in NUGC4 and MKN45 cells significantly suppressed the expression of SOS1 both at the protein and mRNA levels (Fig. 2d,e). These data led us to hypothesize that RUNX1 possibly controls the ErbB2/HER2 signaling cascade through modulating the expression of SOS1 in gastric cancer cells.
Indispensable role of SOS1 in the ErbB2/HER2 signaling pathway. Although SOS family proteins have been shown to take pivotal roles in the RTKs signaling cascades 10,11 and higher expressions of SOS1 in solid tumors relative to their normal counterpart tissues have been repeatedly reported [15][16][17] , its role in gastric cancer cells has remained elusive so far. We thus investigated the oncogenic role of SOS1 in gastric cancer cells. For this purpose, we prepared shRNAs specifically targeting SOS1 (Fig. 3a) and examined their effects on the maintenance of gastric cancer cells. As shown in Fig. 3b, HER2 was significantly dephosphorylated in SOS1-depleted MKN45 cells relative to the control. In addition, SOS1 knockdown slowed down the proliferation speed of MKN45 cells (Fig. 3c). Together with our finding that HER2-depletion indeed decelerates the growth rate of MKN45 cells (Fig. 3d-f), these data collectively underpin the importance of SOS1 expression in the ErbB2/HER2 signaling cascade and in the maintenance of gastric cancer cell lines. Furthermore, as shown in Fig. 3g,h, immunoprecipitation assay proved the physiological interaction of SOS1 and HER2 in MKN45 cells, indicating that SOS1 probably functions as an adaptor protein in the RTKs signaling pathways like other SOS family proteins do 10,11 . Together with the results obtained in the RUNX1 knockdown experiments, we were convinced that RUNX1 controls the ErbB2/HER2 signaling cascade through modulating SOS1 expression in gastric cancer cells.
ScieNTific RepoRts | (2018) 8:6423 | DOI:10.1038/s41598-018-24969-w SOS1 expression is directly transactivated by RUNX1. To further elucidate how RUNX1 controls the expression of SOS1, we next performed chromatin immunoprecipitation (ChIP) assay using the promoter region of SOS1. As shown in Fig. 4a,b, we observed the actual binding of RUNX1 in this region. Luciferase reporter assay with SOS1 promoter (−1000 bp to +200 bp of transcription start site (TSS) of SOS1) showed statistically-significant elevation of reporter signals upon additive expression of RUNX1 (Fig. 4c). These results indicate that RUNX1 directly interacts with SOS1 promoter and positively regulates it. As shown in Fig. 5a,b, restoring the expression of SOS1 in RUNX1-knocked down NUGC4 or MKN45 cells consistently reverted the RUNX1-depletion-mediated growth inhibition. Taken together, these data collectively suggested that RUNX1 positively regulates the ErbB2/HER2 signaling pathway through directly transactivating SOS1 expression in the gastric cancer cells.

RUNX inhibitor Chb-M' induces dose-dependent apoptosis in gastric cancer cells.
We have previously reported a potent RUNX inhibitor Chb-M' and its efficacy in the various types of tumor cell lines including gastric cancer 22 . In this study, we examined whether its anti-tumor potency on the gastric cancer cell lines is mediated by the RUNX1-SOS1-ErbB2/HER2 axis. As in the RUNX1 knockdown experiments in Fig. 1, we  Supplementary Fig. 2a). Besides, Chb-M' suppressed the expression of SOS1 dose-dependently both at the protein and mRNA levels, and dephosphorylated HER2 in MKN45 and NUGC4 cells (Fig. 6b,c). Chb-M'-mediated down-regulation of SOS1 was also observed in the in vivo samples resected from the tumor, which originated from the xenotransplanted MKN45 cells in immunodeficient NOD/Shi-scid/ IL-2Rγ null (NOG) mice (Fig. 7a,b). Of note, while Chb-M' showed significant anti-tumor effects on NUGC4 and MKN45 cells at nanomolar to low micromolar levels, chlorambucil (Chb) itself or Chb-conjugated PI polyamides that target scrambled DNA sequence (5′-WGGCCW-3′; Chb-S) were ineffective against these cells (Fig. 7c). In these cell lines, we also found the superior potency of Chb-M' to lapatinib, a selective HER2 inhibitor currently used in the clinics (Fig. 7d). Moreover, as we have expected, SOS1-overexpressed NUGC4 and MKN45 cells became relatively resistant to the Chb-M' treatment ( Fig. 8a-c). These data collectively indicate that RUNX inhibition by Chb-M' could be a novel therapeutic strategy in HER2-positive gastric cancer cases.

Discussion
RUNX1 transcription factor plays a pivotal role in the development and maintenance of various cancers 20,22,[26][27][28][29] . Although we have previously reported that the Cluster Regulation of RUNX (CROX) is a promising therapeutic strategy against various types of tumors including gastric cancer 22 , precise molecular mechanisms of RUNX-inhibition-mediated anti-tumor effect on these malignant cells have poorly been elucidated. We herein revealed that RUNX1 potentially participates in the maintenance of gastric cancer cells through enhancing the activity of the ErbB2/HER2 signaling pathway by directly transactivating SOS1 expression. We do acknowledge that RUNX transcription factors have diverse targets in vivo and they potentially have multiple other targets than SOS1-ErbB2/HER2 in the gastric cancer cells. In the gastric cancer cell lines tested in this study, increased activity of HER2 was seen in 4 out of 9 cell lines (44.4%). Chb-M' treatment down-regulated the SOS1 protein expression in 7 out of 9 cell lines (77.8%) and significant anti-tumor efficacy was observed in 5 out of 9 cell lines (55.6%). All these 5 cells lines had the decreased expression of SOS1 upon Chb-M' treatment ( Supplementary Fig. 3a-c). These data suggest that SOS1 is probably one of the most important oncogenic targets of RUNX1 in the gastric cancer cells.
We do also acknowledge that MKN45 cells, one of the gastric cancer cell lines used in this study, have amplified-MET and more likely to dependent on this pathway than ErbB2/HER2 signaling for their survival. Yet, the results obtained in our shRNA-mediated HER2 knockdown experiments suggest that this cell line is somewhat dependent on this signaling pathway, and ErbB2/HER2 signaling pathway can still be targeted to control the proliferation of this gastric cancer cell line. To further validate our results, we admit that our findings should be reexamined in a larger set of HER2-amplified gastric cancer cell lines in future studies.
Given the recent introduction of trastuzumab for the treatment of patients with advanced gastric cancer, assessment of HER2 status is now mandatory for selecting patients eligible for this treatment 6,7 . Yet, trastuzumab therapy is not curative, with only modest improvement of overall survival in the latest trials with gastric cancer patients. Though not fully elucidated, the mechanism of resistance to trastuzumab therapy seems to be based on the compensatory up-regulation of other RTKs in response to HER2 inhibition 8,9 . Though pure speculation, Chb-M' therapy may potentially overcome the resistance of current specific HER2 inhibition therapies and provide alternative therapeutic choice for these patients, since Chb-M' regulates the expression of SOS1, an adaptor protein that is generally-required for the RTKs to function properly. Indeed, the phosphorylation of several RTKs other than ErbB2/HER2 was simultaneously and significantly suppressed upon shRNA-mediated RUNX1 depletion or Chb-M' treatment in the phospho-RTK arrays (Figs 1d and 6a). Cell proliferation. To assess cell proliferation, 5 × 10 4 cells of the indicated cells were seeded in 6-well plate.
For the tetracycline inducible shRNA expressions, doxycycline was added to the culture at a final concentration of 3 μM. Trypan blue dye exclusion assays were performed every other day.
Apoptosis assay. Apoptotic cells were detected by Annexin V Apoptosis Detection Kit APC (eBioscience Inc.). In brief, approximately 8 × 10 5 cells of the indicated control and experimental groups were washed in phosphate-buffered saline (PBS), suspended in annexin-binding buffer, and then mixed with 5 μL of annexin V. The reaction mixtures were incubated for 30 min. After incubation, cells were diluted, stained with DAPI and processed for flow cytometric analysis.

Real-time quantitative PCR (qRT-PCR).
Total RNA was isolated with RNeasy mini kit (Qiagen) and reverse transcribed with Reverse script kit (TOYOBO) to generate cDNA. Real-time PCR System (Applied Biosystems) according to the manufacturer's instructions. The results were normalized to GAPDH levels. Relative expression levels were calculated using the 2 −ΔΔCt method. Primers used for qRT-PCR were listed in Supplementary Table 1. ChIP-qPCR. Chromatin immunoprecipitation assay (ChIP) was performed using SimpleChIP ® Plus Enzymatic Chromatin IP Kit (Cell Signaling Technology, USA) according to the manufacturer's instructions. In brief, cells were cross-linked in 1% formaldehyde in PBS for 10 min at room temperature. After glycine quenching, cell pellets were collected, lysed and then subjected to sonication with Q55 sonicator system (QSONICA, USA). The supernatant was diluted with the same sonication buffer, and processed for immunoprecipitation with anti-RUNX1 antibody (ab23980, abcam) at 4 °C overnight. The beads were then washed and DNA was  Expression plasmids. We have amplified cDNAs for human RUNX1 and SOS1, and then inserted them into CSIV-TRE-RfA-UbC-KT expression vector. All of the PCR products were verified by DNA sequencing.

Production and transduction of lentivirus.
For the production of lentivirus, HEK293T cells were transiently co-transfected with lentivirus vectors such as psPAX2 and pMD2.G by polyethylenimine (PEI, Sigma-Aldrich). Forty-eight hours after transfection, viral supernatants were collected and immediately used for infection, and then successfully transduced cells were sorted by flow cytometer Aria III (BD Biosciences).
Luciferase reporter assay. Putative region of SOS1 (−1000 bp to +200 bp of transcription start site) was cloned from the genomic DNA of K562 cells using the following primers; F 5′-CTCCCTGCAGAAATCCCGAG-3′ and R 5′-TCTTGGGACAAAACAAAACACCT-3′, and then subcloned into pGL4.20 [luc2/Puro] vector (Promega). Both pGL4.20 SOS1 promoter vector and pRL-CMV control vector (TOYOBO B-Net Co., LTD.) were co-transfected into HEK293T cells that are stably-expressing shRNA of sh_Luc or expression vector of RUNX1. Promoter activities were measured using PicaGene Dual Sea Pansy Luminescence Kit (TOYOBO B-Net Co., LTD.) and detected by ARVO X5 (Perkin Elmer) according to the manufacturer's instructions. Data are mean ± SEM values. *P < 0.05, **P < 0.01, ***P < 0.001, by two-tailed Student's t-test. time-of-flight mass spectrometer by using positive ionization mode and proton nuclear magnetic resonance ( 1 H NMR) spectra were recorded with a JEOL JNM ECA-600 spectrometer operating at 600 MHz and in parts per million (ppm) downfield relative to tetramethylsilane used as an internal standard to verify the quality of synthesized PI polyamides. Study approval. All animal studies were properly conducted in accordance with the Regulation on