JNK pathway plays a critical role for expansion of human colorectal cancer in the context of BRG1 suppression

Tumor stem cells (TSCs), capable of self‐renewal and continuous production of progeny cells, could be potential therapeutic targets. We have recently reported that chromatin remodeling regulator Brg1 is required for maintenance of murine intestinal TSCs and stemness feature of human colorectal cancer (CRC) cells by inhibiting apoptosis. However, it is still unclear how BRG1 suppression changes the underlying intracellular mechanisms of human CRC cells. We found that Brg1 suppression resulted in upregulation of the JNK signaling pathway in human CRC cells and murine intestinal TSCs. Simultaneous suppression of BRG1 and the JNK pathway, either by pharmacological inhibition or silencing of c‐JUN, resulted in even stronger inhibition of the expansion of human CRC cells compared to Brg1 suppression alone. Consistently, high c‐JUN expression correlated with worse prognosis for survival in human CRC patients with low BRG1 expression. Therefore, the JNK pathway plays a critical role for expansion and stemness of human CRC cells in the context of BRG1 suppression, and thus a combined blockade of BRG1 and the JNK pathway could be a novel therapeutic approach against human CRC.


| INTRODUC TI ON
Colorectal cancer is the third most common type of cancer worldwide and the second most common cause of cancer-related death. 1 Chemotherapy is a main therapeutic option against inoperable CRCs. However, they finally acquire resistance to conventional cytotoxic chemotherapy. Thus, there is an unmet need to invent a novel therapeutic approach. In this context, drug discovery related to epigenetic modifications represents a promising field for the treatment of chemoresistant CRC. Epigenetic modifications including methylation and histone modifications and chromatin remodeling play important roles in initiation and progression of CRC. 2,3 Switch/sucrose nonfermentable (SWI/SNF) chromatin remodeling complexes play important roles in transcriptional regulation, DNA replication, and damage repair in an ATP-dependent manner. 4,5 They contain either one of the two mutually exclusive catalytic ATPase subunits: SMARCA4 (BRG1, also known as Brahma-related gene 1) or SMARCA2 (BRM, also known as Brahma). 6 Although human BRG1 negative cancers exist and harbor malignant behavior, 7-9 BRG1 has recently been shown to be overexpressed in many tumor types and is associated with tumor aggressiveness and poor prognosis, including hCRC. [10][11][12][13] We previously showed that Brg1 is essential for acinar cell-derived pancreatic cancer formation by inhibiting apoptosis. 14 Previous reports, including our report, have shown that hCRCs strongly express BRG1. 12,15,16 As in other types of cancer, expression of BRG1 in hCRC is associated with recurrence, metastasis, and poor prognosis. 12,13,16 We have recently reported that Brg1 plays an essential role for maintenance of murine intestinal TSCs and for cell survival and stemness features of hCRC cells. 16 Continuous ablation of Brg1 in intestinal TSCs maintains suppression of intestinal tumors accompanied by increased apoptosis and loss of their capacity for self-renewal in mice. However, it is still unclear how BRG1 suppression changes the underlying intracellular mechanisms of hCRC cells. In this study, we investigated the underlying molecular mechanism by which BRG1 suppression affects stemness of hCRC cells to develop a novel therapeutic approach against hCRC.

| Animals
The following mouse strains were used. Dclk1 CreERT2-IRES-EGFP mice were generated by our group, as previously described. 17

| Human subjects
Specimens of surgically resected CRC were obtained from patients at Kyoto University Hospital. Clinicopathologic data were collected from medical records and pathological reports of Kyoto University Hospital. The TNM classification was decided in accordance with the UICC 8th classification.

| Histological analysis
Mice tissues were fixed with 4% buffered paraformaldehyde solution overnight at 4°C. They were then paraffin-embedded and sectioned (5 μm thickness). Sections were deparaffinized and rehydrated. Antigen retrieval for all primary Abs was achieved by boiling in 10 mM citrate buffer at pH 6.0 for 15 min. For immunohistochemistry, sections were incubated with primary Abs overnight at 4°C, followed by incubation with biotinylated secondary Ab for 1 h at room temperature. Immunoperoxidase labeling was undertaken with Vectastain ABC kit (Vector Laboratories; catalog no. PK-6102), and sections were then colored with diaminobenzi-

| RNA extraction and qRT-PCR
Total RNA was extracted from tissues or cells using the RNeasy Mini Kit (Qiagen). cDNA was synthesized using the ReverTra Ace qPCR RT Kit (Toyobo). Quantitative RT-PCR was carried out using FastStart SYBR Green Master (Roche Applied Science) and Light Cycler 96 (Roche Applied Science). The Cq values were measured in triplicate.
The expression levels were standardized by comparing to the levels of GAPDH. Primers were designed using Primerbank and are listed in Table S1.

| Spheroid establishment and culture
The extracted mice intestines were washed several times with PBS and the intestinal tumors were dissected. Tumor cells were dissociated using 2.5 mg collagenase from Clostridium histolyticum For induction of Cre-mediated recombination in vitro, 1 μmol/L 4-OHT (Sigma-Aldrich) was added to the culture medium.

| Small interfering RNA transfection
Human CRC cell lines were transfected with 10 nmol/L siRNA targeting BRG1 and JUN (SMARTpool: ON-TARGETplus SMARCA4 siRNA; Dharmacon), and siRNA nontargeting control (ON-TARGETplus nontargeting pool; Dharmacon) using Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher Scientific) and Opti-MEM (Thermo Fisher Scientific). We added this mixture into dishes or wells overnight and changed to fresh medium the next day. Due to unexpected toxicity, penicillin and streptomycin were excluded from the culture medium with siRNA.

| Crystal violet staining
Twenty thousand cells were seeded in 35-mm plates overnight prior to intervention. After removal of the culture medium of cell lines, cells were gently washed with PBS. They were fixed and stained with a mixture of 6.0% v/v glutaraldehyde (Sigma-Aldrich) and 0.4% w/v crystal violet (Merck) for 30 min. Plates were carefully washed with water so as not to tear off the fixed cells on the plates. Images were taken using an optical microscope.

| Western blot analysis
Cell lines were homogenized in Cell Lysis Buffer (Cell Signaling Technology). We measured the protein concentration using a Bio-Rad protein assay (Bio-Rad) and added sample buffer solution with reducing reagent (6×) for SDS-PAGE (Nacalai Tesque) to each sample after matching the concentration. Proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes. After blocking with blocking buffer (Nacalai Tesque), membranes were incubated with primary Abs overnight at 4°C. The next day, membranes were incubated with HRP-conjugated secondary Abs (1:5000 anti-rabbit, 7074; 1:5000 anti-mouse, 7076; both Cell Signaling Technology).

| Cell proliferation assay
Cells were seeded in 96-well plates (1000 cells/well) overnight prior to intervention and were then maintained in the presence of 100 μl culture medium. We added 20 μl CellTiter 96 Aqueous One Solution (Promega) per well. Plates were incubated for 1 h before absorbance was measured at 490 nm with a Sunrise microplate reader (Tecan).

| Microarray analysis
We collected three samples with Control and three samples with siBRG1 per cell line. We used two cell lines (DLD-1 and HCT 116).
The quality of RNA extracted from cell lines was examined using a Nanodrop (Thermo Fisher Scientific). The RNA samples were hybridized using a Clariom S Assay (Thermo Fisher Scientific). Normalization was undertaken using Affymetrix Power Tools Software (Thermo

| Kaplan-Meier curve
For Figure 5A,B, the RNA sequencing dataset of colon adenocarcinomas combining data (n = 380) in the TCGA dataset was downloaded from cBioportal. For Figure 5A, the highest 25% and the lowest 25% of c-JUN expression data with colon adenocarcinomas were extracted from the TCGA dataset (n = 95). For Figure 5B, the high 50% and the low 50% of c-JUN expression data from colon adenocarcinomas with lower BRG1 expression (n = 95) was extracted from the TCGA dataset (n = 47, respectively). The analysis including the log-rank test was carried out using EZR version 1.51.

| Statistics
All values are presented as mean ± SEM, unless otherwise described.
A two-tailed unpaired Student's t-test after F-test was used for statistical analysis of continuous data. The χ 2 -test was used for statistical analysis of categorical data. All statistical analyses were undertaken using GraphPad Prism version 6.07 for Windows (GraphPad Software). Those p values <0.05, <0.01, and <0.001 were considered statistically significant.

| Study approval
All mouse experiments were approved by the animal research  HCT 116) and that BRG1 silencing results in impaired cell growth, loss of stemness feature, and increased apoptosis in hCRC cells. 16 In the report, we undertook a microarray analysis of control and MAP3K14, which were known to be JNK pathway-related genes, and DDIT3 (CHOP) and HMOX1, which were known to be ER stressrelated genes. 16 Gene Set Enrichment Analysis revealed that there was a positive enrichment of gene sets related to the JNK pathway ("ST_JNK_MAPK_PATHWAY") in BRG1-silenced cells ( Figure 1D).
These findings indicate that the JNK pathway was upregulated in the context of BRG1 suppression in hCRC cells. The qRT-PCR analysis verified the upregulated expression of c-JUN in BRG1-silenced hCRC cells ( Figure 1E). We found that the activated level of p-c-JUN was significantly increased in BRG1-silenced DLD-1 and HCT 116 cells ( Figure 1F). Protein level of c-JUN was elevated in BRG1-silenced DLD-1 and HCT 116 cells ( Figure 1F). Therefore, these findings indicate that the JNK pathway was activated to some extent at the basal level in hCRCs and the activation of the JNK pathway was further reinforced by BRG1 suppression.

| Activation of the JNK pathway reinforced by Brg1 ablation in murine intestinal tumors
We next investigated whether activation of the JNK pathway was augmented by Brg1 ablation in murine intestinal tumors. Hnf1b is expressed in intestinal tumor cells and spheroids generated from intestinal tumors in Apc Min mice. 16 Therefore, we first generated HAB mice and established spheroids from intestinal tumors of HAB mice (Figure 2A,B). Through this model, we could ablate Brg1 in tumor cells of spheroids by adding 4-OHT in the medium. Regarding Brg1 KO efficacy in HAB spheroids, we had verified the low expression of Brg1, which was shown in our previous report. 16 We determined the expression level of c-Jun mRNA in spheroids from intestinal tumors from HAB mice compared to controls. Expression of c-Jun was significantly upregulated in Brg1 KO spheroids from HAB mice compared to controls ( Figure 2C). Immunohistochemistry for p-c-JUN also revealed that the JNK pathway was upregulated in Brg1 KO spheroids from HAB mice ( Figure 2D).
To further determine whether activation of the JNK pathway was reinforced by Brg1 ablation in murine intestinal tumors in mice, we next generated DAB mice along with Dclk1 CreERT2-IRES-EGFP/+ ; Apc Min/+ ; Brg1 wt/wt or Dclk1 CreERT2-IRES-EGFP/+ ; Apc Min/+ ; Brg1 wt/wt (DA) mice ( Figure 2E). Because we previously reported that Dclk1 is an intestinal TSC marker, 17 we could ablate Brg1 specifically in intestinal TSCs by injecting tamoxifen intraperitoneally. 16 We previously reported that most intestinal tumor cells were Brg1 negative in DAB mice on day 5 after tamoxifen injection, resulting in the collapse of intestinal tumors. 16 Immunostaining for p-c-Jun revealed that p-c-Jun was highly expressed in intestinal tumor cells in DAB mice on day 5 after tamoxifen injection, whereas p-c-Jun expression was rarely detected in intestinal tumor cells in DA mice ( Figure 2F). Brg1 negative intestinal tumor clusters were positive for p-c-Jun expression in DAB mice ( Figure 2G).
These findings suggests that the JNK pathway is activated by Brg1 ablation also in murine intestinal tumors.

| JNK pathway crucial for expansion and stemness of BRG1-silenced hCRC cells
Given that the JNK pathway was upregulated in the context of BRG1 suppression in hCRC cells, we sought to clarify its role in BRG1suppressed hCRC cells in the context of BRG1 suppression. To this end, we first examined whether treatment with SP600125, a selective JNK pathway inhibitor, 21 affects the expansion of siRNA-mediated BRG1silenced hCRC cells. We validated the efficiency of siBRG1 by qRT-PCR ( Figure S1). Western blot analysis showed that c-JUN activation was sufficiently inhibited by SP600125 treatment ( Figure 3A). We carried out an MTS assay and found that SP600125 does not influence cell viability or proliferation, or expression of stemness-related genes in hCRC cells without treatment of siBRG1 ( Figure 3B,C). We next   (Table S2). Overall survival did not significantly differ between p-c-JUN positive hCRC and p-c-JUN negative hCRC (p = 0.825; Figure 5C). These results also suggested that activation of the JNK pathway alone is not a prognostic factor for hCRC. These data further strengthen the human relevance of our data, providing evidence that the JNK pathway plays a crucial role for CRC in the context of BRG1 suppression. Therefore, a combined blockade of BRG1 and the JNK pathway could be a novel therapeutic approach against hCRC.

| DISCUSS ION
In this study, we found that the JNK pathway was upregulated in both hCRC cells and murine intestinal tumors by BRG1 suppression.
Furthermore, we showed for the first time that the JNK pathway plays a critical role in the expansion and stemness of hCRC cells in the absence, but not in the presence, of BRG1. It is notable that suppression of BRG1 combined with the inhibition of the JNK pathway led to even stronger inhibition of expansion of hCRC cells than did Brg1 suppression alone. Therefore, our findings suggest that a combined blockade of BRG1 and the JNK pathway could be a novel therapeutic approach for hCRC.
Our data showed that JNK pathway blockade even strengthened the phenotype of BRG1-suppressed hCRC cells, which includes loss of cell expansion and stemness features. The JNK pathway has been shown to regulate stemness in normal murine intestinal epithelial cells and human pancreatic cancer cells. 22,23 In this study, we showed that simultaneous suppression of BRG1 and the JNK pathway resulted in downregulated expression of stemness-related genes in hCRC cells compared to Brg1 suppression alone, whereas JNK inhibition alone did not affect the expansion of hCRC cells or the expression of stemnessrelated genes. This suggests that increased expression of JNK pathway genes is a compensatory mechanism to survive the fatal phenotypes (i.e., loss of expansion and stemness) induced by BRG1 suppression.
In this study, our findings provide important clinical relevance for the development of therapeutic strategies for hCRC. We showed that suppression of BRG1, combined with the inhibition of the JNK pathway, led to even stronger inhibition of the expansion of hCRC cells compared to Brg1 suppression alone. Consistently, it is notable that high c-JUN expression was correlated with worse prognosis for survival in hCRC patients with low BRG1 expression. Therefore, enhancing the antitumor effect of BRG1 inhibition through the use of combination therapy involving both BRG1 suppression and JNK inhibitors could be a new therapeutic strategy for hCRC. In particular, inhibition of the JNK pathway could be effective for hCRC patients with low BRG1 expression.
In conclusion, activation of the JNK signaling was augmented by contributed reagents, materials, and analytic tools. T.Y. wrote the manuscript, and A.F. and H.S. revised the manuscript.

ACK N OWLED G M ENTS
We thank Shoko Yokoyama for mouse bleeding and technical support, and all members of the A.F. laboratory for technical assistance and helpful discussions. We also thank D. Reisman of the University of Florida, with permission from P. Chambon, for sharing

E TH I C S S TATEM ENT
Approval of the research protocol by an institutional review board: Analyses for human subjects were approved by the ethical committee of Kyoto University Hospital (#G1200-1, R2904) and conducted in accordance with the Declaration of Helsinki.

A N I M A L S TU D I E S
All mouse experiments were approved by the animal research