This article is mentioned in:

Introduction

In the whole world, esophageal cancer is the sixth cause of death among various cancer types. Esophageal cancers are mainly classified into two histological types, esophageal squamous cell carcinoma (ESCC) and adenocarcinoma (1). Then, ESCC is thought to be the main histological type which accounts for more than 90% in Asian countries including Japan, Korea and China (2), and it is known that ESCC is a highly malignant malignancy among many cancer types (3–5). Esophagectomy remains the mainstay potential curative treatment for ESCC (6). However, esophagectomy is still a highly invasive surgical procedure with high morbidity and mortality (7). Although remarkable advances have been made in chemotherapy and chemotherapy with radiotherapy as cancer therapies, These therapeutic effects can be said to be extremely limited as curative treatment (8,9). Therefore, understanding the characteristics of ESCC and developing new therapeutic tools are urgently required.

Both genetic mechanisms and epigenetic alterations are thought to be closely involved in the development and progression of ESCC (10). Several epigenetic abnormalities have been reported, including DNA methylation, histone modifications and non-coding RNAs (11,12). In our studies, epigenetic modifications play crucial roles in the regulation of gene expression in ESCC (12–19). In particular, the methylation of lysine residues on histone proteins in the chromatin structure has received attention due to their potential regulatory ability on DNA-based nuclear processes such as transcription, replication and repair (20). The methylation of histone lysine residues was first reported in the 1960s and was considered an irreversible posttranslational modification (21). In 2004, however, a lysine demethylase was discovered, and the methylation of histone lysine residues is now regarded as a dynamic modulation (22).

Abnormalities in histone lysine methylation are frequently observed in various cancers (23–26). Lysine-specific histone demethylase 1 (LSD 1), a histone demethylase, is an amine oxidase that removes monomethyl and dimethyl moieties from Lys 4 of histone H 3 and produces a demethylated H3 tail (27). Identifying the key points of regulation in the histone methylation network for cancer development and progression can provide innovative targets for cancer therapies.

In the present study, we focused on the mechanisms underlying how demethylated Lys4 of H3 influences the gene expressions in ESCC cells. We investigated microarray and chromatin immunoprecipitation sequencing (ChIP-seq) in order to explore the effect of demethylated Lys4 of H3 on the transcriptional state of ESCC cells and identified genes affecting cancer growth.

Materials and methods

Cell culture and chemicals

The human esophageal cell lines T.Tn and TE2 were cultured in DMEM (Life Technologies, Grand Island, NY, USA) supplemented with 10% fetal calf serum. T.Tn cells were acquired from the Japanese Cancer Research Resources Bank (Tsukuba, Japan), and TE2 cells were obtained from Tohoku University (Sendai, Japan). NCL1, an LSD1 inhibitor, was provided by Kyoto Prefectural University of Medical Science (Graduate School of Medicine) (Kyoto, Japan) in cooperation. NCL1 showed higher inhibitory activity than the known LSD1 inhibitor, trans-2-phenylcyclopropylamine. Moreover, in the presence of NCL1, the methylation activity of H3K4 is observed and cell proliferation is inhibited in experiments using cancer cells (28–30). NCL1 was dissolved in dimethyl sulfoxide and used for in vitro studies.

Messenger RNA preparation and a cDNA microarray analysis

T.Tn or TE2 cells were seeded into a 225-cm2 flask, incubated for 48 h, treated with or without an IC80 concentration of LSD1 inhibitor and harvested at 24 h. Subsequently, the cells were washed with phosphate-buffered saline (PBS; cat. no: 14190-250, Invitrogen, Carlsbad, CA, USA) and total RNA was extracted using RNeasy Plus Mini kit (Qiagen, Inc., Chatsworth, CA, USA). Changes in gene expression were compared between 5.5 tor of total RNA extracted from cells cultured by exposure to NCL 1 and 5.5 tor of total RNA extracted from cells cultured in a control culture using an Affymetrix Human Exon 1.0ST array (Affymetrix, Santa Clara, CA, USA). Hybridization signals were detected with a GeneChip scanner 3000 7 G (Affymetrix), and the scanned images were analyzed using the GeneChip command console software (AGCC). All the processes were basically carried out according to the previous report (31). All experiments were done in duplicate and the averaged data were subjected to statistical analysis.

ChIP-seq analyses

ChIP-seq analyses were performed using the SimpleChIP plus enzymatic chromatin IP kit (Magnetic Beads; Cell Signaling Technology, Danvers, MA, USA). T.Tn or TE2 cells were cultured for 48 h in a 225-cm2 flask, then incubated under the condition with or without an IC80 concentration of LSD1 inhibitor and harvested at 24 h. The Cells were crosslinked with 1% formaldehyde for 10 min at room temperature, then washed twice with PBS containing 0.5 mM EDTA and collected. The cell pellet was lysed with 0.3 ml of cell lysis buffer (50 mM Tris-HCl [pH 8.1], 10 mM EDTA, 1% SDS, and protease inhibitor) and incubated on ice for 10 min. Lysates of the cells were sonicated to obtain DNA fragments of 150 to 900 base pair (bp) in size. About 50 µg of cross-linked sheared chromatin solution was then used for immunoprecipitation. The solution with the Anti-Histone H3 (di methyl K4) antibody-ChIP Grade (Abcam, Inc., Cambridge, UK; cat. no: ab7766) was incubated overnight at 4°C on a rotating shaker for immunoprecipitation. Magnetic beads were added to the solution, incubated at 4°C for 1 h, and then washed with washing buffer. The cross-linking was reversed by adding NaCl at a final concentration of 200 mM and heating at 65°C for 30 min. The DNA fragments were purified using a spin column. A sequencing library was prepared and massively parallel high throughput sequencing was performed with the Illumina HiSeq 2000 system (Illumina, Inc., San Diego, Calif., USA) and a 50-bp reads were aligned against the reference genome on a Burrows-Wheeler transform, and a minimum mapping quality filter 20 was applied (32). Enriched regions for each condition were detected and analyzed with MACS v1.4.0 (model-based analysis for ChIP-Seq) (33) and CEAS v1.0.2 (cis-regulatory element annotation system) (34,35). Peaks with overlaps in both cell lines were merged into a broad peak domain using BEDTools (36). All of the count data from the ChIP-Seq assays were analyzed with DESeq to normalize the peak signal (37).

The reverse transcription-quantitative PCR (RT-qPCR) for measuring the LDHB and AEG-1/MTDH mRNA expression

The mRNA expression of DUSP5, BHLHE40 and MXRA5 were examined by a RT-qPCR. T.Tn or TE2 cells were seeded into a 225-cm2 flask, incubated for 48 h, treated with or without an IC80 concentration of LSD1 inhibitor and harvested at 24 h. Subsequently, the cells were washed with phosphate-buffered saline (PBS) and total RNA was extracted using an RNeasy Plus Mini kit (Qiagen, Inc., Chatsworth, CA, USA). The cDNA templates for the qPCR were synthesized from 1 µg of total RNA using a High Capacity RNA-to-cDNA kit (Applied Biosystems).The Actin alpha 1 (ACTA1) gene served as an internal control. The PCR reaction consisted of Sso Fast Eva Green Supermix (BioRad; containing dNTPs, Sso7d fusion polymerase, MgCl2, EvaGreen dye, stabilizers), the primers (each 1 µM) and cDNA. All reactions were run in duplicate on the MyiQ2 Two-Color Real-Time PCR detection system (BioRad). The PCR processes were as follows: initial denaturation at 95°C for 30 sec, followed by 40 cycles of denaturation at 95°C for 5 sec, annealing at 55°C for 10 sec. The following primer sequences were used: DUSP5; 5′-CCTGCTAAAACTGGGATGGA-3′ and 5′-ACCTACCCTGAGGTCCGTCT-3′: BHLHE40; 5′-GGCATAGCACGGTAGTGGTT-3′ and 5′-TCAGACCTTGGTTTGGTTCC-3′: MXRA5; 5′-CTGTCCAGTCCTCAGGAAGC-3′ and 5′-TCCTGTGGAAACCTTTGTCC-3′: ACTA1; 5′-CCTTCATCGGTATGGAGTC-3′ and 5′-GTTGGCATACAGGTCCTT-3′.

The comparative quantitative cycle (Cq) method was applied to quantify the expression levels of mRNAs. The relative amount of DUSP5, BHLHE40 and MXRA5 to ACTA1mRNA was calculated using the following equation: 2q−ΔC, where ΔCq=(Cq DUSP5, BHLHE40 or MXRA5 or AEG-1/MTDH-Cq ACTA1).

Statistical analyses

The Student's t-test was performed to compare the differences in the mRNA expression levels. P<0.05 was considered to indicate a statistically significant difference. The SPSS v.16.0 (SPSS, Inc., Chicago, IL, USA) software program were used for the analyses.

Results

Effects of NCL1 on the expression of various genes in microarray analyses

We used a cDNA microarray to identify genes induced by LSD1 exposure in ESCC. We extracted genes with expression levels more than two-fold or greater compared to control, whether decreased or increased, as significant. In both T. Tn and TE2 cell lines, expression of 18 genes was increased, while expression of 9 genes was decreased (Table I).

Table I. List of genes up- or downregulated (>2-fold change) in T.Tn and TE2 cells.

Table I.

List of genes up- or downregulated (>2-fold change) in T.Tn and TE2 cells.

A, Upregulated (>2-fold changes)
		Maximum fold change
Genbank accession no.	Gene symbol	T.Tn	TE2	Gene description
BC005008	CEACAM6	2.84	3.62	carcinoembryonic antigen-related cell adhesion molecule 6 (non-specific cross reacting antigen)
BC012172	ACSS2	1.10	1.24	acyl-CoA synthetase short-chain family member 2
BC008723	ASNS	1.92	1.96	asparagine synthetase (glutamine-hydrolyzing)
L19501	CBS	1.15	1.18	cystathionine-β-synthase
M29540	CEACAM5	2.45	2.90	carcinoembryonic antigen-related cell adhesion molecule 5
BC019625	CHAC1	1.34	2.11	ChaC, cation transport regulator homolog 1 (E. coli)
U03688	CYP1B1	1.49	1.67	cytochrome P450, family 1, subfamily B, polypeptide 1
BC007333	ETV5	1.73	1.81	ets variant 5
AF110400	FGF19	1.17	1.26	fibroblast growth factor 19
EF152283	GCNT3	1.17	1.66	glucosaminyl (N-acetyl) transferase 3, mucin type
AF019770	GDF15	2.01	1.75	growth differentiation factor 15
BC033089	LCN2	1.62	2.78	lipocalin 2
BC004863	PSAT1	1.34	2.87	phosphoserine aminotransferase 1
AF539739	S100P	1.26	2.06	S100 calcium binding protein P
AF097514	SCD	1.25	2.38	stearoyl-CoA desaturase (∆-9-desaturase)
BC000658	STC2	1.95	1.59	stanniocalcin 2
BC011703	TMPRSS4	1.03	1.24	transmembrane protease, serine 4
AF022375	VEGFA	1.62	1.50	vascular endothelial growth factor A
B, Downregulated (>2-fold changes)
		Maximum fold change
Genbank accession no.	Gene symbol	T.Tn	TE2	Gene description
J04948	ALPPL2	−1.01	−1.69	alkaline phosphatase, placental-like 2
BC098561	EFEMP1	−1.59	−1.76	EGF-containing fibulin-like extracellular matrix protein 1
AB031548	GPR87	−1.11	−1.06	G protein-coupled receptor 87
M19154	TGFB2	−1.72	−1.11	transforming growth factor, β2
BC142678	PHLDB2	−1.14	−1.04	pleckstrin homology-like domain, family B, member 2
BC146868	COL12A1	−1.13	−1.22	collagen, type XII, α1
BC017782	WISP2	−1.12	−1.31	WNT1 inducible signaling pathway protein 2
AF098807	LHFP	−1.21	−1.26	lipoma HMGIC fusion partner
AK123348	C3orf57	−1.14	−1.72	chromosome 3 open reading frame 57

ChIP-seq analyses

To assess the functional significance of demethylated Lys4 of H3 in ESCC cells, we also analyzed the genome-wide modified targets of demethylation Lys4 of H3 using deep sequencing based on chromatin immunoprecipitation (ChIP-seq). When we compared the findings with control cells (without LSD1 inhibitor), we identified up-regulated peaks in 468 and 814 demethylated Lys4 of H3-specific modification sites in T.Tn and TE2 cells, respectively (Fig. 1). We also identified down-regulated peaks in 532 and 612 demethylated Lys4 of H3-specific modification sites in T.Tn and TE2 cells, respectively (Fig. 1).

Figure 1. Extraction peaks with a log2 fold change in each cell line. For both cell lines, peaks outside the ± 2 SD range were extracted. SD, standard deviation.

Identifying the relationship between histone modification states and the gene expression in ESCC cells

To clarify the gene expression change by the state of histone modification, the genes with up- or down-regulated expression were investigated using microarray data, and that the promoter region of these genes may be the targets of histone modification. The expression of some of these genes whose promoters were detected as candidates for targets of demethylated Lys4 of H3 were markedly changed according to the microarray data (Table II). The results showed that 17 genes were commonly up-regulated, while 16 genes were commonly down-regulated (Table III). These identified genes were categorized based on their function, referring to GENE ONTOLOGY™, and classified into 7 groups: apoptosis, cell cycles, defense and immunity, metabolism, signal transduction and transcription, structural protein, and unclassified. The frequencies of these functionally classified genes in each cluster are shown in Table IV.

Table II. List of numbers of genes up- or down-regulated in the chromatin immunoprecipitation-seq analysis.

Table II.

List of numbers of genes up- or down-regulated in the chromatin immunoprecipitation-seq analysis.

Cell line	Upregulated peaks, n	Downregulated peaks, n
T.Tn	468	532
TE2	814	612

Table III. List of gene symbols commonly up- or downregulated in both the microarray and ChIP-seq assay.

Table III.

List of gene symbols commonly up- or downregulated in both the microarray and ChIP-seq assay.

	Microarray
ChIP-seq	Upregulated	Downregulated
Upregulated	DUSP5, BHLHE40, TMC5 GNE, PMAIP1, TIMP3 C6orf223, PHLDA1,ERRFI1 MID1IP1, ULBP1, FGF19 GCNT3, HMOX1, TRIB3 VEGFA, CEACAM6	DIO2, RBMS3, LHFP PLK2, CP, TOM1L2 MXRA5, DKK1, EPHA4 EGLN3, CCDC80, MID1 SLC16A7, RAPGEF4, TOX VCL, MAP7D2, RASAL2 HAS2, TNS3, FLNA NEDD4L, KIAA1217, PSAPL1 SEMA3A, GPR126, EGFR
Downregulated	GDF15, CLGN	RHOB, KLHL13, ARID5B DIO2, MXRA5, THBS1 ALDH1A1, DKK1, C1orf116 SOX2, CACNG4, LHFP FGFR2, EPHA4, EFEMP1, PALMD

[i] ChIP-seq, chromatin immunoprecipitation-seq.

Table IV. Categorization of genes regulated by the LSD1 inhibitor based on their functions, referring to GENE ONTOLOGY™, and classified into 7 groups.

Table IV.

Categorization of genes regulated by the LSD1 inhibitor based on their functions, referring to GENE ONTOLOGY™, and classified into 7 groups.

	Microarray and ChIP-seq analysis
Function	Upregulated	Downregulated
Apoptosis	PMAIP1,TIMP3, PHLDA1, FGF19, TRIB3, CEACAM6	RHOB, FGFR2
Cell cycles	HMOX1, VEGFA	KLHL13, THBS1
Defense and immunity	GCNT3	–
Metabolism	MID1IP1	ALDH1A1
Signal transduction and transcription	DUSP5, BHLHE40, GNE, ERRFL1, ULBP1	ARID5B, DKK1, SOX2, EPH!4, EFEMP1
Structural protein	TMC5,	DIO2, MXRA5, C1orf116, CACNG4, LHFP, PALMD
Unclassified	C6orf223	–

[i] ChIP-seq, chromatin immunoprecipitation-seq.

Validation of the gene expression changes induced by NCL1

Among the mRNAs that showed altered expression levels in both microRNA and ChIP-seq experiments, changes in the expression levels of DUSP5, BHLHE40 and MXRA5 were confirmed by a RT-qPCR (Fig. 2). As expected, the expression of DUSP5 and BHLHE40 increased and the expression of MXRA5 decreased.

Figure 2. Among the mRNAs that exhibited altered expression levels in both microRNA and chromatin immunoprecipitation-seq experiments, changes in the expression of DUSP5, BHLHE40 and MXRA5 were confirmed by a reverse transcription-quantitative PCR. *P<0.05 and **P<0.01, as indicated. DUSP5, dual specificity phosphatase 5; BHLHE40, basic helix-loop-helix family member E40; MXRA5, matrix remodeling associated 5.

Discussion

In this study, we tried to clarify the changes in the gene expression due to histone demethylase LSD1 inhibitor using a microarray and ChIP-Seq analyses. Some LSD1 inhibitors have shown potent anti-cancer effects, and their pharmacological mechanisms have been elucidated (38,39). ORY-1001 is an LSD1 inhibitor that was shown to selectively inhibit KDM1A in clinical trials and is currently being assessed for its utility in treating patients with leukemia and solid tumors (40). Although clinical trials of LSD1 inhibitors are being conducted around the world, very few describe the mechanisms in detail (41,42).

We have already elucidated the anti-tumor effect of LSD1 inhibitors on ESCC, and this effect was shown to be caused by changes in the gene expression induced by the agent, with PHLDB2 reported to demonstrate a particularly enormous change in expression (19). In the present study, in addition to changes in the gene expression, genome-wide CHIP-Seq analyses were performed, and the histone methylation that occurred was evaluated.

DUSP5 is one of the nuclear localization members of the MKP/DUSP family and it is induced in response to the activation of ERK, specifically dephosphorylated, and has the function of anchoring the ERK in the nucleus (43). Furthermore, DUSP5 has been reported to increase RAF, MEK and ERK activities in the cytoplasm, in addition to its role in ERK nuclear inactivation. This activity has been shown to be caused by alleviation of upstream kinase inhibition and depends on its ability to sequester DUSP5 turnover rate and inactive ERK in the nucleus (44). Also, the expression of BRAFV 600 E oncoprotein, which has mutations in BRAF that are not sensitive to feedback inhibition, changes the function of DUSP 5 to become an inhibitor of the entire cell of ERK, and that the cell avoids hyperactivation and aging of ERK. These analysis results explain that DUSP5 functions as a tumor suppressor or a tumor promoter (45).

BHLHE 40 is an up-regulated gene and is a basic helix-loop-helix type transcription factor that has been shown to be involved in epithelial-mesenchymal transition (EMT). According to Asanoma et al (46), BHLHE 40 inhibited tumor cell invasion by suppressing the transcription of the EMT factors SNAI 1, SNAI 2 and TWIST 1. In addition, they showed an association between the transcription factor SP1 and the basal transcriptional activity of TWIST1 and BHLHE40 and competes with SP1 to regulate DNA transcription and control gene transcription. Therefore, BHLHE 40 is thought to function as a tumor suppressor.

It is thought that p53 reactivation and mass apoptosis induction (PRIMA-1), a low-molecular compound, restores the function of mutant TP53 to the function of wild-type TP53 and induces p53-mediated apoptosis (47). PRIMA-1 and its methylated form PRIMA-1 Met (APR-246) are thought to have antitumor effects and its effects are evident in several types of cancers such as osteosarcoma, multiple myeloma, lung cancer, breast cancer and colon cancer (48–52). Furthermore, several clinical trials using APR-246 have been performed, indicating its tolerability and clinical effects in hematologic malignancies and prostate cancer (53). Also in ESCC, Furukawa et al (47) reported that PRIMA-1 may restore the function of mutant TP 53 in ESCC with a TP 50 missense mutation, due to the enhanced expression of Noxa. Tissue inhibitor of metalloproteinase-3 (TIMP 3) which is one of the four members of the protein family is initially classified according to their function of inhibiting matrix metalloproteinases (MMP) (54–56). TIMP3 is thought to induce apoptosis in malignant cells, such as melanoma (57) human colon carcinoma (58), cervical carcinoma cells and breast cancer cells (59). The death domain of TIMP3, a region that inhibits the function of MMP, is localized at its N-terminus (60). TIMP3 has been reported in colon cancer cells and melanoma cells to increase susceptibility to apoptosis via stabilization of the TNF-α receptor on the cell surface (58,61). In ESCC, expression of TIMP-3 protein is correlated with depth of tumor infiltration, number of lymph node metastasis and stage of disease as a result of immunohistochemical analysis using clinical specimens (54). TIMP-3 protein was localizes in a shallow region of the tumor, and even in the same tumor, its expression was decreased in the deep part. Furthermore, the prognosis of cancer patients who lost TIMP-3 expression was significantly worse than that of TIMP-3-positive cancer patients.

PHLDA1 is a cell death mediator that induces cells into apoptosis and exerts antiproliferative activity (62–65). The overexpression of PHLDA1 inhibits cell proliferation and induces cell death in various malignant tumors, including breast cancer, melanoma and cervical carcinoma cells (66–68). Low expression of PHLDA1 mRNA and protein is strongly related to breast cancer and melanoma progression (63,64). Conversely, it has been found that in oral squamous cell carcinoma, high expression of PHLDA1 is associated with more advanced stage. Therefore, the function of PHLDA1 is still controversial, but from these results, it is possible that PHLDA1 functions as a tumor suppressor in ESCC.

The genes shown to be down-regulated on either the microarray analysis or ChIP-seq analysis in our study may have a potential oncogenic function in ESCC.

Rho protein belongs to the Ras superfamily and is a small molecule that functions as a binary switch in a wide range of signaling pathways (69,70). Rho proteins are a family of 20 intracellular signaling molecules, including RhoA, RhoB, RhoC, RhoG, RhoE, Rac1, Rac2, Cdc42Hs and TC10 (71). RhoB has the function of molecular switch, and it circulates between inactive GDP-bonded type and active GTP-bound type (72). RhoB was reported as a molecule that induces Ras-induced fibroblast transformation. New evidence suggests a potential role of RhoB in supporting the tumorigenic function. As an example, it is reported that RhoB protein expression is higher in T-acute lymphoblastic leukemia (T-All) cells compared to normal T cells, and it is shown to be significantly associated with leukemia cells (73). In the present study, the inhibition of RhoB increased cell apoptosis by ≥300%.

Matrix remodeling-associated protein 5 (MXRA5), also known as adhesion protein with leucine-rich repeats and immunoglobulin domains related to perlecan (Adlican), is one of matrix remodeling related proteins (74). The MXRA5 gene encodes a protein with a molecular weight of 312 kDa. The MXRA family contains three genes (MXRA5, MXRA7 and MXRA8), both of which are thought to be involved in cell adhesion and matrix remodeling (75). The increased expression of MXRA5 has been reported in many kinds of tumors, such as colorectal cancer, ovarian cancer and esophageal cancer (76,77). Furthermore, somatic mutation of MXRA5 has been reported, and this mutation has been confirmed in various malignant tumors such as lung, skin, brain, ovary and wall pleura (78).

Among thrombospondin, a family of extracellular matrix proteins, thrombospondin-1 (THBS 1) is the first member identified and its major roles are platelet aggregation, angiogenesis, and tumorigenesis (79). Also in ESCC, THBS 1 can activate the TGF-β signaling pathway, leading to the transcription of Cyr 61 and CTGF (78–80). In addition, its overexpression is thought to be significantly associated with TNM progression (P=0.029) and lymph node metastasis (P=0.026) in clinicopathologic studies. In the analysis of prognosis, it was shown that overexpression of THBS protein is a prognostic predictor in ESCC patients (P=0.042) (80).

The results in this study suggest that the large number of genes affected by demethylation of H3 in ESCC Lys4 may be greatly implicated in the development of ESCC cancer. The correlation between these gene groups and carcinogenesis and progression of ESCC needs to be verified in further studies, but the present results will be helpful for clarifying the mechanism.

The authors declare that they have no competing interests.

Acknowledgements

Not applicable.

Funding

The present study was supported by JSPS KAKENHI (grant no. JP 24689053).

Availability of data and materials

The datasets used and/or analyzed during the present study are available from the corresponding author on reasonable request.

Authors' contributions

IH designed, analyzed and conducted all of the experiments and wrote the manuscript. MT, AK and II performed and analyzed the results of the ChIP-seq experiments. YA, KM, YM, HS, NS, TS and HM contributed to the study conception and design, and the acquisition, analysis and interpretation of data. FI and YI contributed to the acquisition, analysis and interpretation of data, drafted the manuscript and revised it critically for important intellectual content. KI performed RT-qPCR experiments. All authors read and approved the final manuscript.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References