Autocrine BMP-4 Signaling Is a Therapeutic Target in Colorectal Cancer

Published OnlineFirst June 13, 2017; DOI: 10.1158/0008-5472.CAN-17-0112 Cancer Molecular and Cellular Pathobiology Research

Autocrine BMP-4 Signaling Is a Therapeutic Target in Colorectal Cancer Yuichiro Yokoyama1,2, Toshiaki Watanabe2, Yusuke Tamura1, Yoshinobu Hashizume3, Kohei Miyazono1, and Shogo Ehata1

Abstract

Poor prognoses for colorectal cancer patients with meta- LDN-193189 elevated expression of the phosphatase DUSP5 static lesions have driven demand for the development of in colorectal cancer cells, inducing apoptosis via dephosphor- novel targeted therapies. Here, we demonstrate that expres- ylation of Erk MAPK. Administering LDN-193189 to mice sion of bone morphogenetic protein 4 (BMP-4) is universally diminished tumor formation of colorectal cancer cells. Our upregulated in human colorectal cancer cells and tissues, ﬁndings suggest inhibition of autocrine BMP-4 as a candidate resulting in activated BMP signaling. Inhibition of endoge- treatment strategy for colorectal cancer. Cancer Res; 77(15); 1–13. nous BMP signaling by the BMP type I receptor inhibitor 2017 AACR.

Introduction their ability to bind certain type I receptors. Upon binding to type I and type II receptors, BMPs form heterotetrameric com- Colorectal cancer is the third most common cancer and the plexes; the protein kinase of the type II receptor activates the fourth most common cause of cancer-related death worldwide protein kinase of the type I receptor, which in turn phosphor- (1). Although surgical resection can cure early-stage colorectal ylates the BMP-specific receptor-regulated Smads (R-Smads), cancer, a combination of surgery and chemotherapeutic agents is Smad1 and Smad5. Phosphorylated R-Smads induce a hetero- recommended in advanced stages of colorectal cancer (1). In meric assembly with common-partner Smad (Co-Smad; addition to conventional cytotoxic agents, new agents targeting Smad4) and translocate into the nucleus, regulating the tran- VEGF signaling and EGFR signaling have been introduced during scription of target genes. BMPs can also activate non-Smad the last decade (2). Although colorectal cancer prognoses have signaling pathways, including the MAPK pathway (3). steadily improved, the 5-year survival rate remains low, especially Divergent roles of BMPs have been reported during cancer in patients with metastatic lesions (1). Thus, the development of progression (4–6). BMPs inhibit proliferation of gastric cancer, new molecular targets for treatment of advanced colorectal cancer breast cancer, and prostate cancer cells, induce differentiation of is critical for improving patient outcomes. glioma-initiating cells, and inhibit glioblastoma tumor formation Bone morphogenetic proteins (BMP) are members of the (5), indicating a tumor-suppressive role of BMPs. In contrast, TGF-b family and are multifunctional cytokines (3–5). BMPs BMPs have been reported to enhance the motility and invasive- recognize two distinct receptors, termed type I and type II ness of various types of cancer cells, such as breast cancer, prostate receptors, with serine/threonine and tyrosine kinase activities. cancer, and malignant melanoma cells, suggesting that BMPs also Type I BMP receptors include activin receptor-like kinase (ALK)- function as tumor-promoting factors (4). 1, -2, -3, and -6, and type II receptors include BMP type II In this study, the role of BMP-4 produced by colorectal cancer receptor (BMPR-II), activin type II receptor (ActR-II), and cells in cancer progression was investigated. We reveal for the first activin type IIB receptor (ActR-IIB). BMPs are classified into time that inhibition of BMP-4 induces the apoptosis of colorectal several subgroups, including the BMP-2/4 group, BMP-5/6/7/8 cancer cells through the attenuation of MAPK activity in culture group, BMP-9/10 group, and growth and differentiation factor and that the small-molecule BMP inhibitor LDN-193189 (GDF)-5/6/7 group, according to structural similarities and diminishes colorectal cancer formation in vivo.

1Department of Molecular Pathology, Graduate School of Medicine, The Univer- Materials and Methods sity of Tokyo, Tokyo, Japan. 2Department of Surgical Oncology, The University of Cell culture and reagents 3 Tokyo, Bunkyo-ku, Tokyo, Japan. RIKEN Program for Drug Discovery and Human colon adenocarcinoma cells HT29 and DLD-1 (Japa- Medical Technology Platforms, Wako, Saitama, Japan. nese Cancer Research Resource Bank) were cultured in RPMI Note: Supplementary data for this article are available at Cancer Research containing 10% FBS, penicillin, and streptomycin. Human colon Online (http://cancerres.aacrjournals.org/). adenocarcinoma SW480 cells (ATCC) were cultured in DMEM Corresponding Authors: Shogo Ehata, Graduate School of Medicine, The containing 10% FBS, penicillin, and streptomycin. Routine Myco- University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Phone: plasma testing was performed by PCR regularly on these cells. The 813-5841-3356; Fax: 813-5841-3354; E-mail: [email protected]; and Kohei cells were bought in 2002 and have been stocked as cryopreserved Miyazono, [email protected] aliquots in liquid N2. The cells were used within 8 passages after doi: 10.1158/0008-5472.CAN-17-0112 thawing and reauthenticated by short tandem repeat proﬁling in 2017 American Association for Cancer Research. 2017. LDN-193189 was obtained from Wako or RIKEN.

www.aacrjournals.org OF1

Yokoyama et al.

Quantitative real-time RT-PCR analyses and chromatin NCI-60 cell line panel indicated that, among various BMPs, immunoprecipitation–qRT-PCR analyses expression of BMP4 was commonly elevated in colon cancer cells Quantitative real-time RT-PCR (qRT-PCR) analysis and chro- (Fig. 1A). NCBI GEO database GSE14258 revealed that expression matin immunoprecipitation (ChIP)–qRT-PCR analysis were of BMP4, but not of other BMPs, was significantly higher in performed as previously described (7, 8). Primer sequences colon cancer tissues than in normal colon tissues (Fig. 1B). Next, are described in Supplementary Table S1. Anti-TCF4 antibody the correlation between BMP4 expression and colorectal cancer (sc-8631) was purchased from Santa Cruz Biotechnology. patient prognosis was examined. GSE14333 showed that elevated expression of BMP4 was associated with poor prognosis in Immunoblotting patients with stage II colorectal cancer (Fig. 1C). Furthermore, Immunoblotting was performed as previously described (7). multivariate analysis demonstrated that BMP4 expression was an Antibodies are described in Supplementary Materials and Meth- independent prognostic factor in stage II colorectal cancer (Sup- ods. ImageJ (NIH) was used to quantify blot band intensities. plementary Table S2). Expression of BMP-4 and activation of Smad-dependent BMP signaling was then examined using human Apoptosis assay colorectal cancer tissues and cells. IHC analysis revealed that Terminal deoxynucleotidyl transferase-mediated dUTP expression levels of BMP-4 and phosphorylated Smad1/5 were nick end labeling (TUNEL) assay was performed as previously upregulated in colorectal cancer tissues compared with those in described (7). Fluorescence was examined using a Leica corresponding normal tissues (Fig. 1D). ELISAs demonstrated DMI6000 B. that these colorectal cancer cells produced BMP-4, whereas, with siRNA the exception of SUIT-2 pancreatic cancer cells, other cancer cells fi Stealth RNAi Pre-Designed siRNAs targeting CTNNB1, BMP4 examined did not (Fig. 1E). These ndings suggest that colorectal and DUSP5 were synthesized by Thermo Fisher Scientific. Cells cancer cells produce BMP-4, which may act in an autocrine were transfected in the presence of 30 nmol/L siRNA or control manner, and that BMP-4 expression may be related to colorectal siRNA in a 500 mL volume with 3 mL RNAiMAX reagent (Thermo cancer progression. BMP4 Fisher Scientific) per well in a 6-well plate. Next, we sought to clarify the mechanism by which mRNA was elevated in colorectal cancer. Mutations in the APC Lentiviral production and infection gene occur in the early phase of colorectal cancer progression, We used a lentiviral vector system to induce specific gene which in turn increases the stability of b-catenin (15). Because introduction and knockdown as previously described (9, 10). BMP4 expression is reported to be regulated by the Wnt/b-catenin The target sequences for shRNA are described in the Supplemen- pathway (16), the involvement of Wnt/b-catenin in the induction tary Materials and Methods. of BMP4 expression in colorectal cancer cells was assessed. b-Cate- nin protein levels were elevated in colorectal cancer cells but not in RNA-sequence analyses non-colorectal cancer cells, such as pancreatic cancer cells (SUIT- RNA-sequence (RNA-seq) analysis was performed as described 2) and breast cancer cells (MDA-231-D; Fig. 2A). Knockdown of previously (11). Raw and processed data are available at GEO the CTNNB1 gene (encoding b-catenin) in colorectal cancer cells (GSE96914). Gene Ontology analysis was performed using CLC by siRNAs suppressed BMP4 mRNA expression and BMP-4 pro- Genomics Workbench (Qiagen Bioinformatics). tein production, as well as expression of a direct downstream target of the Wnt/b-catenin pathway, AXIN2 (Fig. 2B and C). Subcutaneous xenograft model Similar to colorectal cancer cells, introduction of siCTNNB1 to BALB/c nu/nu female mice (4–5 weeks) were obtained from SUIT-2 cells also decreased BMP4 mRNA, indicating that regula- Sankyo Labo Service Corporation. A total of 5 106 cells in 100 mL tion of BMP4 by the Wnt/b-catenin pathway was not restricted to of culture medium were subcutaneously inoculated. Tumor vol- colorectal cancer cells (Fig. 2B). Although colorectal cancer cells ume was estimated as previously described (12). All animal were stimulated with Wnt-3a, increase of AXIN2 was not observed experiments were performed under the policies of the Animal (Fig. 2D), suggesting that signaling activity of Wnt/b-catenin Ethics Committee of The University of Tokyo (approval number: might have already been saturated in colorectal cancer cells. 12312). The bioavailability of administered LDN-193189 was Likewise, stimulation of pancreatic cancer cells with Wnt-3a examined as described previously (13). increased the expression of both AXIN2 and BMP4. Pretreatment of BxPC-3 cells with cycloheximide did not suppress Wnt-3a– IHC induced BMP4 expression (Supplementary Fig. S1). ChIP–qRT- fi fi Formalin- xed, paraf n-embedded human clinical samples PCR analysis revealed a binding of TCF4 to the previously were obtained from patients at The University of Tokyo Hospital reported enhancer region of BMP4 gene, which was enhanced by with informed consent. The protocol was approved by the Wnt-3a in SUIT-2 cells and attenuated by introduction of Research Ethics Committee at The University of Tokyo, Graduate siCTNNB1 in HT29 cells (Fig. 2E; ref. 16). These findings suggest School of Medicine (approval number: 10475). IHC was per- that BMP-4 is directly regulated by the Wnt/b-catenin pathway. formed as previously described (14). Antibodies are described in the Supplementary Materials and Methods. Colorectal cancer cells undergo apoptosis following the inhibition of autocrine BMP-4 signaling Results To examine the role of BMP-4 on colorectal cancer tumor Colorectal cancer produces BMP-4 through aberrant activation growth, shRNA-targeting BMP4 was introduced into HT29 and of the Wnt/b-catenin pathway SW480 cells. As a result, expression of inhibitor of DNA binding 1 To investigate the expression of BMP mRNAs in colorectal (ID1), a direct target gene of BMPs, and production of BMP-4 cancer, data from several public databases were re-analyzed. The protein in culture supernatants were significantly diminished

OF2 Cancer Res; 77(15) August 1, 2017 Cancer Research

BMP-4 as a New Therapeutic Target in Colorectal Cancer

A C Lung Stage II CRC CNS Renal tumor Colon cancer cancer Breast cancer cancer cancer Ovarian Leukemia Melanoma 1.00 Prostate cancer 0.80 * BMP2

BMP4 )

2 0.60 BMP5 0.40 BMP4 low (n = 48) BMP6 BMP4 high (n = 46) BMP7 0.20 Disease-free survival BMP8A 0.00 BMP8B 0204060 GDF2 Month BMP10 Relative expression (log GDF5 0.0 6.0 12.0

B 2PMB 4PMB 6PMB 7PMB 01PMB 300 1,200 200 250 50 *** 1,000 250 200 40 150 Normal colon tissue 200 800 (n = 54) 150 30 150 600 100 Colon cancer tissue 100 20 (n = 186) 100 400 Relative expression 50 10 50 200 50

0 0 0 0 0

D E BMP-4 Protein H&E BMP-4 pSmad1/5 700 600 500 400 300 (pg/mL)

Normal colon tissues Normal colon 200

Protein concentration 100

Patient #1 0 A549 HeLa HT29 DLD-1 SUIT-2 SW480 CRC tissues MDA-231-D

CRC Non-CRC Normal colon tissues Normal colon Patient #2 CRC tissues

Figure 1. BMP-4 is produced in colorectal cancer cells and colorectal cancer tissues. A, Comprehensive gene-expression analysis from NCI-60 cell line panels showing gene expression proﬁles of BMP mRNA in leukemia (n ¼ 6), breast cancer (n ¼ 5), central nervous system (CNS) tumor (n ¼ 6), colon cancer (n ¼ 7), renal cancer

(n ¼ 8), lung cancer (n ¼ 9), ovarian cancer (n ¼ 7), prostate cancer (n ¼ 2), and melanoma (n ¼ 9) cells. Color indicates distance from log2 6. B, Gene expression analysis from the NCBI GEO database (GSE14258). Box plot reveals expression of BMP mRNA in normal colon epithelium (n ¼ 54) and colon cancer (n ¼ 186) tissues. C, Kaplan–Meier plot of disease-free survival of patients with stage II colorectal cancer (CRC; n ¼ 94) stratiﬁed by median BMP4 mRNA expression based on data from NCBI GEO database (GSE14333). D, IHC of colorectal cancer tissues and corresponding normal colon tissues from two colorectal cancer patients (Patients #1 and #2), stained with hematoxylin and eosin (H&E), anti–BMP-4 antibody, and antiphospho-Smad1/5 antibody (pSmad1/5); scale bars, 50 mm. E, Concentrations of BMP-4 proteins in cancer cell culture supernatants (48 hours) determined by ELISA (n ¼ 2). Figure data are shown as box whisker plots (B) or as means SD (E). , P < 0.05, , P < 0.001, as determined by Student t test (B) or by log-rank test (C).

www.aacrjournals.org Cancer Res; 77(15) August 1, 2017 OF3

Yokoyama et al.

A B 1.5 CTNNB1 1 AXIN2 1 BMP4

CRC Non-CRC ) 0.8 0.8 1

GAPDH 0.6 0.6 HT29 DLD-1 SUIT-2 SW480 0.4 0.4 MDA-231-D 0.5 β-Catenin 0.2 0.2 Relative expression

(normalized by (normalized by 0 0 0 α-Tubulin HT29 (CRC) 2.5 3 CTNNB1 2 AXIN2 1 BMP4 2 ) 0.8 1.5 siNTC 1.5 2 0.6 siCTNNB #1 1 GAPDH -Catenin 1 siCTNNB #2 β 0.4

(arbitrary units) 0.5 1 0.5 0.2 0 Relative expression

(normalized by (normalized by 0 0 0

DLD-1 (CRC) HT29 DLD-1 SUIT-2 SW480

MDA-231-D 7 CTNNB1 0.1 AXIN2 0.8 BMP4

) 6 0.08 C 5 0.6 BMP-4 Protein BMP-4 Protein GAPDH 4 0.06 600 600 0.4 3 0.04 500 500 2 0.2 0.02 400 400 1 Relative expression (normalized by (normalized by 300 300 0 0 0 (pg/mL) 200 200 SUIT-2 (Non-CRC)

Protein concentration 100 100 E ChIP: TCF4 ChIP: TCF4 0.7 0.5 0 0 0.6 0.4 0.5 siNTC (-) 0.3 siCTNNB #1 siNTC siNTC 0.4 siCTNNB #2 siCTNNB #1 Wnt-3a siCTNNB #2 siCTNNB #1 0.3 0.2 siCTNNB #2 % Input HT29 DLD-1 % Input 0.2 0.1 0.1 0 0 SOBP BMP4 SOBP BMP4 HT29 SUIT-2

) AXIN2 BMP4 D 1 AXIN2 1 BMP4 200 4 0.8 0.8 150 3 GAPDH 0.6 0.6 100 2 0.4 0.4 50 1 0.2 0.2

Relative expression 0 0 0 0 (normalized by by (normalized 0 10203040 0 1020304050 0 1020304050 0 1020304050 Hours after Wnt-3a treatment Hours after Wnt-3a treatment Hours after Wnt-3a treatment Hours after Wnt-3a treatment

HT29 (CRC) SUIT-2 (Non-CRC) (-)

) AXIN2 BMP4 AXIN2 BMP4 1.5 1.5 2 0.8 Wnt-3a

1.5 0.6 GAPDH 1 1 1 0.4 0.5 0.5 0.5 0.2

Relative expression 0 0 0 0 (normalized by by (normalized 0 1020304050 0 1020304050 0 1020304050 0 1020304050 Hours after Wnt-3a treatment Hours after Wnt-3a treatment Hours after Wnt-3a treatment Hours after Wnt-3a treatment

DLD-1 (CRC) BxPC-3 (Non-CRC)

Figure 2. Elevated expression of BMP4 in colorectal cancer cells is due to aberrant activation of the Wnt/b-catenin pathway. A, Top, immunoblotting of cell lysates with indicated antibodies. Bottom, relative expression of b-catenin protein in indicated cells. B, qRT-PCR analysis of CTNNB1, AXIN2, and BMP4 expression in cancer cells (n ¼ 2). Indicated cells were transfected with control siRNA (siNTC) or siRNA targeting CTNNB1 (siCTNNB1#1 and #2), cultured for 72 hours, and analyzed by qRT-PCR. C, Concentrations of BMP-4 proteins in cells in B. Cell culture supernatants (48 hours) were examined by ELISA (n ¼ 4). D, qRT-PCR analysis of AXIN2 and BMP4 expression in indicated cancer cells after Wnt-3a stimulation (200 ng/mL) at the indicated time points (n ¼ 2). E, ChIP-qRT-PCR analysis of TCF4- bound DNA using primers designed at the enhancer region of BMP4. SUIT-2 cells and HT29 cells were ﬁxed and harvested 1.5 hours after Wnt-3a (200 ng/mL) stimulation or 48 hours after introduction of siCTNNB1, respectively. Sine oculis binding protein homolog (SOBP) was used as negative control. Figure data are shown as means SD (B–E).

OF4 Cancer Res; 77(15) August 1, 2017 Cancer Research

BMP-4 as a New Therapeutic Target in Colorectal Cancer

A B D ** ** *** BMP4 ID1 BMP4 ID1 BMP-4 protein 2.5 * 6 n.s. 12 ** 2.5 1.5 0.5 0.3 600 50

) 10 ) 2 5 2 1.2 0.4 500 40 5 10 4 8

0.2 㽢

HPRT1 400 1.5 1.5 0.9 0.3 30 6 300 3 1 0.6 0.2 20 1 0.1 (pg/mL) 200 2 4 0.5 0.3 0.1 10 0.5

Relative expression 1 2 100 Cell number ( Protein concentration (normalized by (normalized by 0 0 0 0 0 0 0 0 0 shNTC shNTC shNTC shNTC shNTC shNTC siNTC siNTC siNTC shBMP4 shBMP4 shBMP4 shBMP4 shBMP4 shBMP4 siBMP4 #1 siBMP4 #2 siBMP4 #1 siBMP4 #2 siBMP4 #1 siBMP4 #2 HT29 SW480 HT29 SW480 HT29 DLD-1 SW480 C

shNTC (n = 5) shNTC (n = 5) shBMP4 (n = 5) shBMP4 (n = 5)

400 ) 600 shNTC 3 ) 3 shNTC 500 300 400 200 * 300 volume (mm volume 200 ** 100 100 Tumor Tumor volume (mm Tumor volume 0

shBMP4 0 0 5 10 15 0 5 10 15 20 shBMP4 Days after transplantation Days after transplantation

HT29 SW480

HT29 DLD-1 E siNTC siBMP4 #1 siBMP4 #2 F siBMP4 #2 siBMP4 #2 siBMP4 #1 siNTC siNTC siBMP4 #1 HT29 PARP

pSmad1/5 SYTOX Green TUNEL α-Tubulin

** * 4 4 ** *

3 3

2 2

1 1 (fold change) Cleavage of PARP Cleavage of 0 0 siNTC siNTC siBMP4 #2 siBMP4 #2 siBMP4 #1 siBMP4 #1

HT29 DLD-1

www.aacrjournals.org Cancer Res; 77(15) August 1, 2017 OF5

Yokoyama et al.

(Fig. 3A and B). When these cells were subcutaneously inoculated of cancer cells with BMP-4 attenuated DUSP5 mRNA expression, into nude mice, knockdown of BMP-4 inhibited tumor formation confirming that DUSP5 is regulated by BMP signaling (Fig. 5F). of both HT29 and SW480 cells in vivo (Fig. 3C). To investigate the Because DUSP5 acts as an inducible nuclear MAPK phosphatase role of autocrine BMP-4 on proliferation and survival of colorectal and specific dual phosphatase for extracellular signal-regulated cancer cells, BMP-4 expression was silenced with siRNAs. Knock- kinase (Erk) MAPK, the effects of LDN-193189 and noggin on down of BMP4 inhibited proliferation of colorectal cancer cells MAPK signaling in colorectal cancer cells were examined. under serum-free conditions (Fig. 3D). TUNEL assay revealed that Although different effects on the phosphorylation of p38 MAPK the number of TUNEL-positive cells increased following silencing and JNK were observed in a cell type-dependent manner, phos- of BMP4 (Fig. 3E). In addition, cleavage of PARP was enhanced in phorylation of Erk MAPK was commonly attenuated in the BMP4-silenced cells (Fig. 3F), suggesting that autocrine BMP-4 colorectal cancer cells examined (Fig. 5G; Supplementary Fig. prevents apoptosis of colorectal cancer cells, which in turn pro- S3A and S3B). Moreover, although siBMP4#2 exhibited a partial motes tumor growth. effect probably because of inefficient knockdown of BMP-4, We next examined whether LDN-193189, a BMP type I receptor siBMP4#1 caused a similar result on DUSP5 expression to inhibitor, induced apoptosis of colorectal cancer cells. Similar to LDN-193189 (Supplementary Fig. S3C), suggesting that regula- the extracellular BMP antagonist protein noggin, LDN-193189 tion of Erk MAPK by BMP signaling is mediated through DUSP5 attenuated the phosphorylation of Smad1/5 and the expression of in colorectal cancer cells. ID1 (Fig. 4A and B), demonstrating that LDN-193189 potently TodeterminewhethertheprosurvivaleffectofBMP-4is inhibits endogenous BMP signaling in colorectal cancer cells. mediated by DUSP5, DUSP5 was knocked down in colorectal LDN-193189 significantly suppressed proliferation of colorectal cancer cells with siRNAs (Fig. 6A). Knockdown of DUSP5 cancer cells under serum-free conditions (Fig. 4C). Apoptosis was attenuated LDN-193189-induced apoptosis (Fig. 6B and C). also induced by LDN-193189 (Fig. 4D and E). Because LDN- Furthermore, the enhancement of PARP cleavage and the atten- 193189 has been reported to exhibit off-target effects on various uation of Erk phosphorylation were diminished following protein kinases, such as AMP-activated protein kinase (AMPK) silencing of DUSP5 in colorectal cancer cells (Fig. 6D). Because and the tyrosine receptor kinases for platelet-derived growth expression levels of other DUSP members in colorectal cancer factor (PDGF) and VEGF (17, 18), we investigated whether noggin cells were upregulated by the transfection of siDUSP5 (data not could reproduce the apoptosis-inducing activity of LDN-193189 shown), siDUSP5 appeared to induce some off-target effects on on colorectal cancer cells. As shown in Supplementary Fig. S2A other phosphatases, which might have in fluence on phosphor- and S2B, apoptosis of colorectal cancer cells was induced by ylation of Erk. noggin under serum-free conditions. Moreover, LDN-193189 did To clarify the proapoptotic effect of DUSP5, DUSP5 was not inhibit proliferation of BMP-4-negative cancer cells (Supple- introduced into colorectal cancer cells using a lentiviral vector mentary Fig. S2C). These results suggest that LDN-193189 (Supplementary Fig. S4A). Under serum-free conditions, over- induces apoptosis of colorectal cancer cells through the inhibition expression of DUSP5 resulted in cell-number reduction and of BMP signaling. elevated apoptosis of colorectal cancer cells (Supplementary Fig. S4B and S4C). PARP cleavage was also enhanced and Erk phos- LDN-193189 induces apoptosis of colorectal cancer cells phorylation was attenuated by overexpression of DUSP5 (Sup- through the induction of DUSP5 plementary Fig. S4D). To determine the underlying mechanism by which LDN- Finally, we examined whether attenuation of Erk MAPK results 193189 induces apoptosis of colorectal cancer cells, downstream in apoptosis of colorectal cancer cells. Treatment of U0126, a MEK fi targets of BMP-4 were identi ed using RNA-seq analysis of HT29 inhibitor, abolished phosphorylation of Erk and enhanced cleav- and DLD-1 cells. As a result, 62 genes were commonly upregulated age of PARP and apoptosis (Supplementary Fig. S5A and S5B). and 24 were downregulated by LDN-193189 in HT29 cells and These results suggest that LDN-193189–induced apoptosis of DLD-1 cells (Fig. 5A). Gene ontology (GO) analysis indicated that colorectal cancer cells is regulated by DUSP5-mediated dephos- LDN-193189 diminished MAPK activity in colorectal cancer cells phorylation of Erk MAPK. (Fig. 5B), resulting in the elevation of genes encoding dual specificity phosphatases (DUSP; Fig. 5C). RNA-seq analysis and LDN-193189 inhibits tumor formation in vivo qRT-PCR analysis revealed that, among the DUSP members in On the basis of the above findings, the efficacy of LDN-193189 colorectal cancer cells, LDN-193189 had the greatest impact on as a new therapeutic agent for colorectal cancer was evaluated. the expression of DUSP5 (Fig. 5D and E). Conversely, stimulation Colony formation of colorectal cancer cells in soft agar was

Figure 3. Knockdown of BMP-4 inhibits tumor formation of colorectal cancer cells in vivo through induction of apoptosis. A, qRT-PCR analysis of ID1 and BMP4 expression in colorectal cancer cells (n ¼ 2). Colorectal cancer cells were transduced with control shRNA (shNTC) and shRNA targeting BMP4 (shBMP4) using lentiviral vector and analyzed by qRT-PCR. B, Concentrations of BMP-4 proteins in cells in A. Cell culture supernatants (48 hours) were examined by ELISA (n ¼ 4). C, Tumor-forming ability of BMP-4-silenced colorectal cancer cells. BALB/c nu/nu female mice received subcutaneous transplants of HT29-shNTC (n ¼ 5) and HT29-shBMP4 (n ¼ 5) cells or SW480-shNTC (n ¼ 5) and SW480-shBMP4 (n ¼ 5) cells. Left, representative photographs 19 days (HT29) and 15 days (SW480) after injection. Right, tumor volumes at the indicated time points. D, Proliferation of BMP-4-silenced colorectal cancer cells. Colorectal cancer cells were transfected with control siRNA (siNTC)– or siRNA-targeting BMP4 (siBMP4 #1 and #2). On the following day, cells were deprived of serum and cultured for 3 days. Cell numbers are indicated (n ¼ 2 for HT29 and DLD-1 cells and n ¼ 4 for SW480 cells). E, TUNEL staining of cells in D. Top, red, TUNEL; blue, SYTOX green. Bottom, the percentage of TUNEL-positive cells among SYTOX green-positive cells. Data represent the mean of six microscopic ﬁelds. F, Top, immunoblotting of lysates from cells in D with indicated antibodies. Bottom, cleavage of PARP protein in indicated cells. Data represent a fold increase compared with negative control (n ¼ 2 for HT29 and n ¼ 3 for DLD-1 cells). Data are shown as means SD (A, B, and D–F) or as means SEM (C). , P < 0.05; , P < 0.01; , P < 0.001; n.s., nonsigniﬁcant, as determined by Student t test (D–F) or by two-way ANOVA (C).

OF6 Cancer Res; 77(15) August 1, 2017 Cancer Research

BMP-4 as a New Therapeutic Target in Colorectal Cancer

A HT29 DLD-1 SW480 LDN -+- -+- -+- Noggin --+ --+ --+

Smad1

pSmad1/5

B ID1 ID1 ID1 C 0.4 1.4 0.2 5 ** 7 * 5 ** 6

) 1.2 4 4 0.3 ) 0.15 5 1 5

HPRT1 3 3 0.8 㽢 10 4 0.2 0.1 0.6 2 3 2 2 0.1 0.4 0.05 Relative expression 1 1 (normalized by (normalized by 0.2 Cell number ( 1 0 0 0 0 0 0 (-) (-) (-) LDN LDN LDN (-) (-) (-) LDN LDN LDN Noggin Noggin Noggin HT29 DLD-1 SW480 HT29 DLD-1 SW480

D HT29 DLD-1 SW480

16 ***50 *** 30 ***

(-) 40 25 12 20 30 8 15 20 10 4 10

Apoptotic cells (%) Apoptotic cells 5

LDN 0 0 0 (-) LDN (-) LDN (-) LDN

HT29 DLD-1 SW480

SYTOX Green TUNEL

E 3 *** 2.5 **2 HT29 DLD-1 SW480 2.5 2 LDN 1.5 -+ -+ -+ 2 1.5 PARP 1.5 1 1 1

(fold change) 0.5 pSmad1/5 0.5 0.5 Cleavage of PARP Cleavage of 0 0 0 α-Tubulin (-) (-) (-) LDN LDN LDN

HT29 DLD-1 SW480

Figure 4. LDN-193189 induces apoptosis of colorectal cancer cells. A, Immunoblot analysis of colorectal cancer cells treated with BMP inhibitors. Colorectal cancer cells were cultured with 0.2–0.3 mmol/L LDN-193189 (LDN) or 50 ng/mL noggin for 2 hours. Immunoblotting of cell lysates was conducted with indicated antibodies. B, qRT-PCR analysis of ID1 expression in colorectal cancer cells in A (n ¼ 2). C, Effects of LDN-193189 on proliferation of colorectal cancer cells. Cells were seeded in 6-well plates. On the following day, cells were deprived of serum and cultured with DMSO or 0.2–0.3 mmol/L LDN-193189 for 3 days. Cell numbers are indicated (n ¼ 2). D, TUNEL staining of cells in C. Left, red, TUNEL; blue, SYTOX green. Right, the percentage of TUNEL-positive cells among SYTOX green-positive cells. Data represent the mean of six microscopic ﬁelds. E, Left, immunoblotting of lysates from cells in C with indicated antibodies. Right, cleavage of PARP protein in indicated cells. Data represent a fold increase compared with untreated control (n ¼ 2). Data are shown as means SD (C–E). , P < 0.05; , P < 0.01; , P < 0.001, as determined by Student t test (C–E).

www.aacrjournals.org Cancer Res; 77(15) August 1, 2017 OF7

Yokoyama et al.

A HT29 DLD-1 DLD-1 C HT29 DLD-1 HT29 DUSP1 DUSP1 DUSP4 DUSP2 540 62 454 255 24 687 DUSP5 DUSP3 DUSP8 DUSP5 SPRED1 DUSP8 SPRED2 DUSP16 GPS2 Genes upregulated by LDN-193189 Genes downregulated by LDN-193189 B HT29 DLD-1 Gene ontology P Gene ontology P Ion transport 1.24E-11 Digestion 9.44E-05 Positive regulation of transcription from RNA polymerase II promoter 1.52E-10 Endoderm formation 0.000132 Amino acid transport 2.35E-08 Carbohydrate metabolic process 0.000156 Negative regulation of transcription from RNA polymerase II promoter 5.59E-08 Transmembrane transport 0.000167 Negative regulation of transcription, DNA-dependent 9.37E-07 Heterotypic cell-cell adhesion 0.000373 Chromatin silencing 3.38E-06 Notch signaling involved in heart development 0.000373 Amino acid transmembrane transport 4.4E-06 Skeletal muscle cell differentiation 0.000498 Response to cAMP 4.61E-06 Negative regulation of cell proliferation 0.000915 Inflammatory response 2.92E-05 Inactivation of MAPK activity 0.001274 Transcription, DNA-dependent 5.15E-05 Positive regulation of leukocyte chemotaxis 0.001462 Skeletal muscle cell differentiation 6.27E-05 Proteasomal protein catabolic process 0.001462 Positive regulation of NF-kappaBtranscription factor activity 0.000164 Negative regulation of protein autophosphorylation 0.001462 Cellular response to hormone stimulus 0.000196 G-Protein coupled receptor signaling pathway 0.001979 Negative regulation of fat cell differentiation 0.000258 Gluconeogenesis 0.002047 Mammary gland branching involved in thelarche 0.000303 Response to calcium ion 0.002047 Cellular response to insulin stimulus 0.000363 Positive regulation of cell migration 0.002959 Inactivation of MAPK activity 0.000405 Glycosaminoglycan metabolic process 0.00328 Response to progesterone stimulus 0.000405 Lung-associated mesenchyme development 0.003457 Liver development 0.000457 Cellular response to zinc ion 0.003457 Placenta blood vessel development 0.000542 Glycerol metabolic process 0.003457

D 80 160 E DUSP5 (-) (-) 40 70 20 60 LDN 120 LDN 60 30 50 15 40 80 40

RPKM 20 10 30 20 40

Relative expression 10 20 5 (normalized by GAPDH ) 10 0 0 0 0 0 (-) (-) (-) LDN LDN LDN DUSP1 DUSP6 DUSP7 DUSP2 DUSP4 DUSP5 DUSP8 DUSP9 DUSP1 DUSP2 DUSP5 DUSP6 DUSP8 DUSP9 DUSP10 DUSP16 DUSP4 DUSP7 DUSP10 DUSP16 HT29 DLD-1 SW480

HT29 DLD-1

F G HT29 DLD-1 SW480 ID1 DUSP5 ID1 DUSP5 1.2 * 1.2 ** 1.2 *** (-) (-) 2 8 1.2 2.5 (-) LDN LDN LDN

1 pErk 1 1 1 1.6 2 6 0.8 pp38 0.8 0.8 0.8 1.2 1.5 0.6 4 pJNK 0.6 0.6 0.6 0.8 1 0.4 2 DUSP5 0.4 0.4 0.4 Relative expression 0.5 0.4 0.2 (fold change) (normalized by GAPDH ) Erk 0.2 0.2 0.2 0 0 0 0 Erk Phosphorylation of pSmad1/5 0 0 0 (-) (-) (-) (-) BMP-4 BMP-4 BMP-4 BMP-4 (-) (-) β-Actin (-) LDN LDN LDN MDA-231-D A549 HT29 DLD-1 SW480

Figure 5. LDN-193189 inactivates MAPK through induction of DUSP5 in colorectal cancer cells. A, Identiﬁcation of genes regulated by LDN-193189 in colorectal cancer cells using RNA-seq analysis. Cells were seeded in 6-well plates. On the following day, cells were deprived of serum and cultured with DMSO or 0.2–0.3 mmol/L LDN-193189 for 3 days. All genes whose RPKM values were >3 were included in analysis. Left Venn diagrams, number of genes upregulated >1.5-fold by LDN-193189. Right Venn diagrams, number of genes downregulated >1.5-fold by LDN-193189. B, Gene ontology analysis of genes upregulated by LDN-193189 in indicated cells. C, Genes upregulated by LDN-193189 (>1.5-fold) that belong to the gene ontology "inactivation of MAPK activity." Red genes belong to DUSP family. D, Expression of DUSP family genes in colorectal cancer cells. Data were extracted from RNA-seq analysis. E, qRT-PCR analysis of DUSP5 expression in colorectal cancer cells (n ¼ 2). Cells were seeded in 6-well plates. On the following day, cells were deprived of serum and cultured with DMSO or 0.2–0.3 mmol/L LDN-193189 (LDN) for 3 days. F, qRT-PCR analysis of DUSP5 expression in cancer cells (n ¼ 2). Cells were seeded in 6-well plates. On the following day, cells were deprived of serum and cultured with BMP-4 (30 ng/mL) for 3 days (MDA-231-D cells) or 24 hours (A549 cells). G, Left, immunoblotting of lysates from cells in E with indicated antibodies. Right, expression of pErk MAPK protein in indicated cells. Data represent as a fold decrease compared with untreated control (n ¼ 2). Data are shown as means SD (E–G). , P < 0.05; , P < 0.01; , P < 0.001, as determined by Student t test (G).

OF8 Cancer Res; 77(15) August 1, 2017 Cancer Research

BMP-4 as a New Therapeutic Target in Colorectal Cancer

A DUSP5 B 5 4 12 3 3 3.5 ** * n.s. ** * n.s. ** n.s. * ) ) (-) 10

3 5 4 2.5 2.5 3 (-) 2.5 LDN 8 2 2 㽢 10 3 LDN GAPDH 2 2 6 1.5 1.5 1.5 2 4 1 1 1 1 1

Cell number ( 2 Relative expression 0.5 0.5 0.5 (normalized by (normalized by 0 0 0 0 0 0 siNTC siNTC siNTC siNTC siNTC siNTC siDUSP5 #2 siDUSP5 #3 siDUSP5 #1 siDUSP5 #1 siDUSP5 #2 siDUSP5 #2 siDUSP5 #1 siDUSP5 #3 siDUSP5 #1 siDUSP5 #2 siDUSP5 #2 siDUSP5 #2

HT29 DLD-1 SW480 HT29 DLD-1 SW480 C (-) LDN 16 25 25 ** ** * ** n.s. n.s. *** ** * 20 20 12 (-) 15 15 LDN siNTC 8 10 10 4 DLD-1 5 5 Apoptotic cells (%) Apoptotic cells

0 0 0 siDUSP5 #2 siNTC siNTC siNTC siDUSP5 #1 siDUSP5 #2 siDUSP5 #1 siDUSP5 #2 siDUSP5 #3 siDUSP5 SYTOX Green TUNEL #2 siDUSP5 HT29 DLD-1 SW480

D HT29 DLD-1

siDUSP5 siDUSP5 siDUSP5 siDUSP5 siNTC siNTC #1 #2 #1 #2 (-) (-) (-) (-) (-) (-) LDN LDN LDN LDN LDN LDN

PARP PARP

pErk pErk

Erk Erk

DUSP5 DUSP5

pSmad1/5 pSmad1/5

β-Actin β-Actin

2 2 1.5 1.5 (-) (-)

1.5 1.5 LDN LDN 1 1

1 1 0.5 0.5 (fold change)

(fold change) 0.5 (fold change) 0.5 (fold change) Cleavage of PARP Cleavage of Cleavage of PARP Cleavage of Phosphorylation of Erk Phosphorylation of 0 0 0 Erk Phosphorylation of 0 siNTC siNTC siNTC siNTC siDUSP5 #1 siDUSP5 #2 siDUSP5 #1 siDUSP5 #2 siDUSP5 #2 siDUSP5 siDUSP5 #1 siDUSP5 siDUSP5 #1 siDUSP5 #2 siDUSP5

Figure 6. Silencing of DUSP5 in colorectal cancer cells abolishes proapoptotic effect of LDN-193189. A, qRT-PCR analysis of DUSP5 expression in colorectal cancer cells (n ¼ 2). Colorectal cancer cells were transfected with control siRNA (siNTC)– or siRNA-targeting DUSP5 (siDUSP5 #1 and #2). On the following day, cells were deprived of serum and cultured with DMSO or 0.2–0.3 mmol/L LDN-193189 for 3 days, followed by qRT-PCR analysis. B, Number of cells in A (n ¼ 2). C, TUNEL staining of cells in A. Left, red, TUNEL; blue, SYTOX green. Right, the percentage of TUNEL-positive cells among SYTOX green-positive cells. Data represent the mean of six microscopic ﬁelds. D, Top, immunoblotting of lysates from cells in A with indicated antibodies. Bottom, cleavage of PARP and expression of pErk MAPK protein in indicated cells. Data represent a fold change compared with negative control. Data are shown as means SD (A–C). , P < 0.05; , P < 0.01; , P < 0.001; n.s., nonsigniﬁcant, as determined by Student t test (B and C).

www.aacrjournals.org Cancer Res; 77(15) August 1, 2017 OF9

Yokoyama et al.

A C HT29DLD-1 SW480

800 DMSO (n = 7) ) 3 LDN (n = 7) 600 (mm ** (-) 400 volume 200 Tumor 0

LDN 2 6 10 14 18 Days after transplantation

0.7 * 160 ** 500 ** 40 *** 0.6 )

2 140 0.5 400 DMSO

mm 120 30 0.4 3 100 10 300 0.3 80 20 0.2 LDN 60 200 Tumor weight (g) 0.1 40 10 100 0 20

Tumor area ( DMSO LDN 0 0 0 (-) LDN (-) LDN (-) LDN SW480

HT29 DLD-1 SW480

B D H&E pSmad1/5 pErk ID1 0.18

) 0.16 0.14 DMSO 0.12 GAPDH 0.1 0.08 0.06

elative expression 0.04 R

(normalized by 0.02 LDN 0 DMSO LDN

BMP-4 BMP-4

P P Type I P Type I P LDN-193189 receptors receptors

BMPR-II

Non-Smad Smad Non-Smad Smad pathway pathway pathway pathway

Wnt/b-catenin pathway DUSP5 DUSP5

APC APC Mutation P Mutation P Erk Erk Erk Erk

CRC Cells CRC Cells Survival Survival

Figure 7. (Continued on the following page.)

OF10 Cancer Res; 77(15) August 1, 2017 Cancer Research

BMP-4 as a New Therapeutic Target in Colorectal Cancer

significantly suppressed by LDN-193189 (Fig. 7A). To confirm the through ActR-II and ActR-IIB. On the basis of these findings, we bioavailability of administered LDN-193189, an ex vivo bioassay concluded that BMP signaling was activated in colorectal cancer was performed using DLD-1 cells. qRT-PCR analysis revealed that cells and tissues. serum from LDN-193189–treated mice successfully suppressed The role of BMP signaling during colorectal cancer progression ID1 expression (Fig. 7B). Body-weight loss was not observed in has not been fully elucidated. Some studies have revealed that LDN-193189–treated mice (data not shown), suggesting that BMPs promote invasiveness and tumor formation of colorectal LDN-193189 acted as a potent BMP inhibitor in vivo without cancer cells (4, 5), whereas other reports have demonstrated severe toxicity. Finally, the effect of LDN-193189 on the tumor growth-suppressive roles of BMPs in colorectal cancer (22, 23). formation of colorectal cancer cells was investigated in vivo. Smad4 and p53 are reported to be key molecules affecting the Notably, LDN-193189 inhibited tumorigenesis in mice bearing functional roles of BMPs in colorectal cancer progression. Voor- colorectal cancer cells (Fig. 7C). Furthermore, LDN-193189 atten- neveld and colleagues reported that loss of Smad4 switches uated the phosphorylation of Smad1/5 and Erk MAPK in vivo (Fig. BMPs from antitumorigenic to protumorigenic, whereas p53 7D). These results indicate that the inhibition of endogenous BMP mutations suppress the enhancement of chemosensitivity by BMP signaling by LDN-193189 may represent a potential strategy for signaling in colorectal cancer cells with wild-type Smad4 (24, 25). treatment of colorectal cancer. Because p53 was mutated in all colorectal cancer cells used in the current study (26), the association between p53 status and the protumorigenic role of BMP signaling in colorectal cancer could Discussion not be assessed in our study. However, we demonstrated that In this study, we demonstrated the protumorigenic role of inhibition of endogenous BMP signaling induced apoptosis of BMP-4 in colorectal cancer (Fig. 7E). Aberrant activation of the not only Smad4-null cells (HT29 and SW480) but also cells with Wnt/b-catenin pathway increases BMP4 expression in colorectal wild-type Smad4 (DLD-1). Although factors other than Smad4 cancer cells. Autocrine BMP-4 signaling protects cells from apo- and p53 may modulate the role of BMP signaling in colorectal ptosis through suppression of DUSP5-mediated dephosphoryla- cancer progression, our findings suggest that autocrine BMP-4 tion of Erk MAPK. When autocrine BMP-4 signaling is inhibited exerts a prosurvival effect on colorectal cancer cells regardless of by LDN-193189, Erk MAPK is dephosphorylated via induction of Smad4 status. DUSP5, resulting in colorectal cancer cell apoptosis. These results Erk enhances the proliferation, survival, and metastasis of suggest that autocrine BMP-4 represents a potential target for cancer cells and acts as an oncogenic signaling pathway (27). colorectal cancer treatment. Although BMPs have been known to phosphorylate Erk MAPK We demonstrated that endogenous BMP signaling was activat- and enhance its activity, the underlying molecular mechanism is ed by autocrine BMP-4 in colorectal cancer cells and tissues. not fully understood (28–30). Using RNA-seq, we identified Elevated BMP4 expression is unique to colorectal cancer, as other DUSP5 as a BMP target gene and determined that DUSP5 was BMPs are not elevated in colorectal cancer cells, and BMP4 appears important for the Erk-mediated prosurvival effect of BMP-4 on to be elevated only in colorectal cancer. Elevated expression of colorectal cancer cells. DUSP5 is a member of the four inducible BMP ligands in colorectal cancer was shown previously and was nuclear MAPK phosphatases and dephosphorylates Erk1/2 (31, reported to correlate with poor prognosis (4, 5, 19); however, 32). DUSP5 has been implicated as a tumor suppressor in various activation of endogenous signaling by autocrine BMPs in colo- types of cancer, including skin, gastric, prostate, colon, and lung rectal cancer remains controversial. Kodach and colleagues (20) cancer (33–36). In this study, we also showed that enhancement reported that Smad1/5 phosphorylation was not observed in of Erk MAPK phosphorylation by BMP-4 is mediated by a reduc- most colorectal cancer cases due to mutations in Smad4 or tion in DUSP5. Although we did not examine whether the BMPR-II. In contrast, Beck and colleagues (21) showed that regulation of DUSP family genes by BMPs occurs via a Smad- Smad1 phosphorylation was detected in colorectal cancer tissues dependent pathway, our findings may provide insight into the and observed in colorectal cancer cells with BMPR2 mutations. In mechanism of regulation of MAPK signaling by BMP signaling. this study, BMP-4 expression and Smad1/5 phosphorylation were Because BMPs provide a potential target for cancer treatment, observed in colorectal cancer tissues and correlated with each various BMP inhibitors have been developed. Calpe and col- other. Furthermore, Smad1/5 phosphorylation was detected in leagues (37) showed that neutralizing antibodies for BMP-4 DLD-1 cells, which are reported to carry mutations in BMPR-II increase the chemosensitivity of HT29 cells. Dorsomorphin was (21), suggesting that BMP signaling is transduced in these cells identified as a potent BMP inhibitor, and its analogue, DMH-1,

(Continued.) LDN-193189 attenuates tumor formation of colorectal cancer cells in vivo. A, Colony formation assay of colorectal cancer cells with or without LDN- 193189. Top, representative photographs. Bottom, box plot reveals tumor areas of eight microscopic ﬁelds (n ¼ 2). B, Ex vivo bioassay to conﬁrm the bioavailability of administered LDN-193189. Separated mouse serum was diluted 2-fold with RPMI containing 10% FBS. DLD-1 cells were treated with 200 mL of serum in 12-well plates. After 2 hours, ID1 expression was examined by qRT-PCR analysis (n ¼ 2). C, Tumor formation of colorectal cancer cells with or without LDN-193189. BALB/c nu/nu female mice received subcutaneous transplants of SW480 (n ¼ 7) cells. Vehicle (2.5% DMSO) or 6 mg/kg LDN-193189 in 2.5% DMSO was injected intraperitoneally twice a day, starting 2 days before inoculation of SW480 cells. Top, tumor volumes at the indicated time points. Bottom, representative photographs and tumor weights 21 days after injection. D, Tumor tissues in C stained with hematoxylin and eosin (H&E), antiphospho-Erk antibody (pErk), and antiphospho-Smad1/5 antibody (pSmad1/5). Representative photographs are shown. E, Scheme of autocrine BMP-4 signaling in colorectal cancer cells with or without BMP inhibitors. Left, in colorectal cancer cells, aberrant activation of the Wnt/b-catenin pathway induces expression of BMP4 mRNA, which in turn activates endogenous BMP signaling. This endogenous BMP signaling promotes phosphorylation of Erk through downregulation of DUSP5, which results in survival of colorectal cancer cells. Right, BMP inhibitors, including LDN-193189, inhibit endogenous BMP signaling. This results in dephosphorylation of Erk through elevation of DUSP5, which in turn induces apoptosis of colorectal cancer cells. Data are shown as means SD (B) or as means SEM (C, top). , P < 0.05; , P < 0.01; , P < 0.001, as determined by Student t test (A and C, bottom) or two-way ANOVA (C, top).

www.aacrjournals.org Cancer Res; 77(15) August 1, 2017 OF11

Yokoyama et al.

disturbs lung cancer growth and breast cancer metastasis was also effective against SW480 cells harboring KRAS mutations (38, 39). LDN-193189 was reported to inhibit growth of breast in vivo (46). Together, these results suggest that BMP inhibitors, and prostate cancers in vivo and prolong survival of mice especially small-molecule kinase inhibitors such as LDN-193189, bearing ovarian cancer cells (40–42). In colorectal cancer, may represent attractive new therapeutic strategies for colorectal Voorneveld and colleagues demonstrated that LDN-193189 cancer treatment. reduced the viability and enhanced the chemosensitivity of Smad4-silenced colorectal cancer cells in vitro (25). On the Disclosure of Potential Conflicts of Interest basis of these reports, we attempted to determine whether LDN- No potential conflicts of interest were disclosed. 193189 inhibits colorectal cancer tumor formation in vivo. Tumor formation in mice bearing colorectal cancer cells was Authors' Contributions significantly diminished by LDN-193189, suggesting that this Conception and design: Y. Yokoyama, K. Miyazono, S. Ehata therapeutic strategy may potentially be of use in colorectal Development of methodology: Y. Yokoyama, S. Ehata Acquisition of data (provided animals, acquired and managed patients, cancer treatment. However, the risk of intestinal carcinogenesis in vivo provided facilities, etc.): Y. Yokoyama, Y. Tamura, S. Ehata must be noted when LDN-193189 is used .Because Analysis and interpretation of data (e.g., statistical analysis, biostatistics, inhibition of BMP signaling in mice by transgenic expression computational analysis): Y. Yokoyama, Y. Tamura, S. Ehata of noggin under control of the villin promoter or by condi- Writing, review, and/or revision of the manuscript: Y. Yokoyama, tional knockout of Bmpr1a leads to intestinal polyposis (43, Y. Hashizume, K. Miyazono, S. Ehata 44), inhibition of BMP signaling in colon epithelial cells may Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Yokoyama, T. Watanabe, S. Ehata increase the risk of intestinal carcinogenesis. Whissell and Study supervision: T. Watanabe, K. Miyazono, S. Ehata colleagues (16) demonstrated that orally administered LDN- Other [preparation of chemical tool and evaluation of the properties; 193189 increased intestinal tumor formation in conditional designed the administration method (solvent, administration route, and Apc knockout mice; however, we did not detect intestinal tumor dosage) based on the above data]: Y. Hashizume formation in LDN-193189–treated mice (data not shown). One possible explanation for this discrepancy is a difference Acknowledgments in the route of LDN-193189 administration. Intermittent dos- We thank Yasuyuki Morishita (The University of Tokyo) for technical ing of inhibitors may be required to avoid the risk of tumor- assistance and Hiroyuki Miyoshi (Keio University) for providing lentiviral vectors. igenesis. Another possible explanation is the use of different strains of mice in the experiments. The balance between BMP signaling and Wnt/b-catenin pathways is important for main- Grant support fi tenance of homeostasis of intestinal epithelial regeneration This work was supported by a KAKENHI Grant-in-Aid for Scienti c Research Apc on Innovative Areas, Integrative Research on Cancer Microenvironment Net- (44). Because knockout mice exhibit activation of the work (22112002) from the Ministry of Education, Culture, Sports, Science and b Apc Wnt/ -catenin pathway in the intestine, knockout mice Technology of Japan (MEXT; to K. Miyazono); a KAKENHI Grant-in-Aid for are more sensitive to BMP inhibitors than wild-type mice. Scientific Research (C) (15K08393) from the Japan Society for the Promotion of In this study, we also demonstrated that LDN-193189 atten- Science (JSPS; to S. Ehata); a grant for Leading Advanced Projects for Medical uated phosphorylation of not only Smad1/5 but also Erk MAPK in Innovation (LEAP; 16am0001003h0003) from the Japan Agency for Medical fi colorectal cancer tissues in mice. Because the Ras–Raf–MEK–Erk Research and Development (AMED; to K. Miyazono); and Speci c Research Grants from The Cell Science Research Foundation (to S. Ehata), and the Yasuda signaling cascade is activated by mutations in the signaling Medical Foundation (to K. Miyazono). components of cancer cells, this signaling pathway is considered The costs of publication of this article were defrayed in part by the an important target for cancer treatment (27). Indeed, Ras muta- payment of page charges. This article must therefore be hereby marked tions are detected in 50% of colorectal cancer cases (45). Anti- advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate EGFR antibodies targeting this signaling, such as cetuximab and this fact. panitumumab, are effective against certain types of colorectal cancer; however, their usefulness is limited to colorectal cancer Received January 12, 2017; revised April 5, 2017; accepted June 5, 2017; cases without KRAS mutations (1). We showed that LDN-193189 published OnlineFirst June 13, 2017.

References 1. Brenner H, Kloor M, Pox CP. Colorectal cancer. Lancet 2014;383:1490– 7. Hoshino Y, Katsuno Y, Ehata S, Miyazono K. Autocrine TGF-b protects 1502. breast cancer cells from apoptosis through reduction of BH3-only protein, 2. Cercek A, Saltz L. Evolving treatment of advanced colorectal cancer. Curr Bim. J Biochem 2011;149:55–65. Oncol Rep 2010;3:153–9. 8. Hoshino Y, Nishida J, Katsuno Y, Koinuma D, Aoki T, Kokudo N, et al. 3. Miyazono K, Kamiya Y, Morikawa M. Bone morphogenetic protein recep- Smad4 decreases the population of pancreatic cancer-initiating cells tors and signal transduction. J Biochem 2010;1:35–51. through transcriptional repression of ALDH1A1. Am J Pathol 2015;185: 4. Ehata S, Yokoyama Y, Takahashi K, Miyazono K. Bi-directional roles of 1457–70. bone morphogenetic proteins in cancer: another molecular Jekyll and 9. Murai F, Koinuma D, Shinozaki-Ushiku A, Fukayama M, Miyaozono K, Hyde? Pathol Int 2013;6:287–96. Ehata S. EZH2 promotes progression of small cell lung cancer by 5. Davis H, Raja E, Miyazono K, Tsubakihara Y, Moustakas A. Mechanisms of suppressing the TGF-b-Smad-ASCL1 pathway. Cell Discov 2015;1: action of bone morphogenetic proteins in cancer. Cytokine Growth Factor 15026. Rev 2016;27:81–92. 10. Sakurai T, Isogaya K, Sakai S, Morikawa M, Morishita Y, Ehata S, et al. RNA- 6. Wakeﬁeld LM, Hill CS. Beyond TGFb: roles of other TGFb superfamily binding motif protein 47 inhibits Nrf2 activity to suppress tumor growth in members in cancer. Nat Rev Cancer 2013;13:328–41. lung adenocarcinoma. Oncogene 2016;10:5000–9.

OF12 Cancer Res; 77(15) August 1, 2017 Cancer Research

BMP-4 as a New Therapeutic Target in Colorectal Cancer

11. Mizutani A, Koinuma D, Seimiya H, Miyazono K. The Arkadia-ESRP2 axis pathways independent of BMPRII in PASMCs. PLoS ONE 2014;9: suppresses tumor progression: analyses in clear-cell renal cell carcinoma. e112695. Oncogene 2016;35:3514–23. 30. Gallea S, Lallemand F, Atfi A, Rawadi G, Ramez V, Spinella-Jaegle S, et al. 12. Kawabata KC, Ehata S, Komuro A, Takeuchi K, Miyazono K. TGF-b-induced Activation of mitogen-activated protein kinase cascades is involved in apoptosis of B-cell lymphoma Ramos cells through reduction of MS4A1/ regulation of bone morphogenetic protein-2-induced osteoblast differen- CD20. Oncogene 2013;32:2096–106. tiation in pluripotent C2C12 cells. Bone 2001;28:491–98. 13. Ehata S, Hanyu A, Fujime M, Katsuno Y, Fukunaga E, Goto K, et al. Ki26894, 31. Kidger AM, Keyse SM. The regulation of oncogenic Ras/ERK signalling by a novel transforming growth factor-b type I receptor kinase inhibitor, dual-specificity mitogen activated protein kinase phosphatases (MKPs). inhibits in vitro invasion and in vivo bone metastasis of a human breast Semin Cell Dev Biol 2016;50:125–32. cancer cell line. Cancer Sci 2007;98:127–33. 32. Caunt CJ, Keyse SM. Dual-specificity MAP kinase phosphatases 14. Shirai YT, Ehata S, Yashiro M, Yanagihara K, Hirakawa K, Miyazono K. Bone (MKPs): shaping the outcome of MAP kinase signalling. FEBS J 2013; morphogenetic protein-2/4 play tumor suppressive roles in human dif- 280:489–504. fuse-type gastric carcinoma. Am J Pathol 2011;179:2920–30. 33. Cai C, Chen JY, Han ZD, He HC, Chen JH, Chen YR, et al. Down-regulation 15. Fearon ER, Vogelstein B. A genetic model for colorectal tumorigenesis. Cell of dual-specificity phosphatase 5 predicts poor prognosis of patients with 1990;61:759–67. prostate cancer. Int J Clin Exp Med 2015;8:4186–94. 16. Whissell G, Montagni E, Martinelli P, Hernando-Momblona X, Sevillano 34. Shin SH, Park SY, Kang GH. Down-regulation of dual-specificity phos- M, Jung P, et al. The transcription factor GATA6 enables self-renewal of phatase 5 in gastric cancer by promoter CpG island hypermethylation and colon adenoma stem cells by repressing BMP gene expression. Nat Cell Biol its potential role in carcinogenesis. Am J Pathol 2013;182:1275–85. 2014;16:695–707. 35. Ueda K, Arakawa H, Nakamura Y. Dual-specificity phosphatase 5 (DUSP5) 17. Hao J, Ho JN, Lewis JA, Karim KA, Daniels RN, Gentry PR, et al. In vivo as a direct transcriptional target of tumor suppressor p53. Oncogene 2003; structure-activity relationship study of dorsomorphin analogues identifies 22:5586–91. selective VEGF and BMP inhibitors. ACS Chem Biol 2010;5:245–53. 36. Rushworth LK, Kidger AM, Delavaine L, Stewart G, van Schelven S, 18. Vogt J, Traynor R, Sapkota GP. The specificities of small molecule inhibitors Davidson J, et al. Dual-specificity phosphatase 5 regulates nuclear ERK of the TGFb and BMP pathways. Cell Signal 2011;23:1831–42. activity a nd suppresses skin cancer by inhibiting mutant Harvey-Ras 19. Motoyama K, Tanaka F, Kosaka Y, Mimori K, Uetake H, Inoue H, et al. (HRasQ61L)-driven SerpinB2 expression. Proc Natl Acad Sci USA 2014;111: Clinical significance of BMP7 in human colorectal cancer. Ann Surg Oncol 18267–72. 2008;15:1530–37. 37. Calpe S, Wagner K, El Khattabi M, Rutten L, Zimberlin C, Dolk E, et al. 20. Kodach LL, Wiercinska E, de Miranda NF, Bleuming SA, Musler AR, Effective inhibition of bone morphogenetic protein function by highly Peppelenbosch MP, et al. The bone morphogenetic protein pathway is specific llama-derived antibodies. Mol Cancer Ther 2015;14:2527–40. inactivated in the majority of sporadic colorectal cancers. Gastroenterology 38. Hao J, Lee R, Chang A, Fan J, Labib C, Parsa C, et al. DMH1, a small 2008;134:1332–41. molecule inhibitor of BMP type I receptors, suppresses growth and inva- 21. Beck SE, Jung BH, Fiorino A, Gomez J, Rosario ED, Cabrera BL, et al. Bone sion of lung cancer. PLoS ONE 2014;9:e90748. morphogenetic protein signaling and growth suppression in colon cancer. 39. Owens P, Pickup MW, Novitskiy SV, Giltnane JM, Gorska AE, Hopkins CR, Am J Physiol Gastrointest Liver Physiol 2006;291:135–45. et al. Inhibition of BMP signaling suppresses metastasis in mammary 22. Zhang Y, Chen X, Qiao M, Zhang BQ, Wang N, Zhang Z, et al. Bone cancer. Oncogene 2015;34:2437–49. morphogenetic protein 2 inhibits the proliferation and growth of human 40. Ali JL, Lagasse BJ, Minuk AJ, Love AJ, Moraya AI, Lam L, et al. Differential colorectal cancer cells. Oncol Rep 2014;32:1013–20. cellular responses induced by dorsomorphin and LDN-193189 in chemo- 23. Lee CW, Ito K, Ito Y. Role of RUNX3 in bone morphogenetic protein therapy-sensitive and chemotherapy-resistant human epithelial ovarian signaling in colorectal cancer. Cancer Res 2010;70:4243–52. cancer cells. Int J Cancer 2015;136:455–69. 24. Voorneveld PW, Kodach LL, Jacobs RJ, Liv N, Zonnevylle AC, Hoogen- 41. Balboni AL, Hutchinson JA, DeCastro AJ, Cherukuri P, Liby K, Sporn MB, boom JP, et al. Loss of SMAD4 alters BMP signaling to promote colorectal et al. DNp63a-mediated activation of bone morphogenetic protein sig- cancer cell metastasis via activation of Rho and ROCK. Gastroenterology naling governs stem cell activity and plasticity in normal and malignant 2014;147:196–208. mammary epithelial cells. Cancer Res 2013;73:1020–30. 25. Voorneveld PW, Kodach LL, Jacobs RJ, van Noesel CJ, Peppelenbosch MP, 42. Lee YC, Cheng CJ, Bilen MA, Lu JF, Satcher RL, Yu-Lee LY, et al. BMP4 Korkmaz KS, et al. The BMP pathway either enhances or inhibits the Wnt promotes prostate tumor growth in bone through osteogenesis. Cancer Res pathway depending on the SMAD4 and p53 status in CRC. Br J Cancer 2011;71:5194–203. 2015;112:122–30. 43. Haramis AP, Begthel H, van den Born M, van Es J, Jonkheer S, Offerhaus GJ, 26. Rodrigues NR, Rowan A, Smith ME, Kerr IB, Bodmer WF, Gannon JV, et al. De novo crypt formation and juvenile polyposis on BMP inhibition in et al. p53 mutations in colorectal cancer. Proc Natl Acad Sci USA mouse intestine. Science 2004;303:1684–6. 1990;87:7555–59. 44. He XC, Zhang J, Tong WG, Tawfik O, Ross J, Scoville DH, et al. BMP 27. Roberts PJ, Der CJ. Targeting the Raf-MEK-ERK mitogen-activated protein signaling inhibits intestinal stem cell self-renewal through suppression of kinase cascade for the treatment of cancer. Oncogene 2007;26:3291–310. Wnt-b-catenin signaling. Nat Genet 2004;36:1117–21. 28. Lai CF, Cheng SL. Signal transductions induced by bone morphogenetic 45. Malumbres M, Barbacid M. RAS oncogenes: the first 30 years. Nat Rev protein-2 and transforming growth factor-b in normal human osteoblastic Cancer 2003;3:459–65. cells. J Biol Chem 2002;277:15514–22. 46. Vecchione L, Gambino V, Raaijmakers J, Schlicker A, Fumagalli A, Russo M, 29. Zhang Y, Wang Y, Yang K, Tian L, Fu X, Wang Y, et al. BMP4 increases the et al. A vulnerability of a subset of colon cancers with potential clinical 2þ expression of TRPC and basal [Ca ]i via the p38MAPK and ERK1/2 utility. Cell 2016;165:317–30.

www.aacrjournals.org Cancer Res; 77(15) August 1, 2017 OF13

Autocrine BMP-4 Signaling Is a Therapeutic Target in Colorectal Cancer

Yuichiro Yokoyama, Toshiaki Watanabe, Yusuke Tamura, et al.

Cancer Res Published OnlineFirst June 13, 2017.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-17-0112

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2017/06/13/0008-5472.CAN-17-0112.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2017/07/19/0008-5472.CAN-17-0112. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.