Downregulation of DLC-1 Gene by Promoter Methylation during Primary Colorectal Cancer Progression

Purpose. DLC-1 is a tumor suppressor gene frequently silenced in human cancers. However, the pathogenicity of DLC-1 epigenetic silencing in the mucosa-adenoma-carcinoma transformation process of colorectal cancer (CRC) has not been studied. Methods. Promoter methylation status of DLC-1 was evaluated in 4 human CRC cell lines, 48 normal mucosa, 57 adenomas, and 80 CRC tissues with methylation-sensitive high-resolution melting analysis (MS-HRMA), while the mRNA expression was examined by qPCR. HRMA was utilized to detect the KRAS codon 12, 13 and BRAF V600Emutations. Results. Partial (1%–10%) and extensive (10%–100%) DLC-1 promoter methylations were observed in 10% and 0% of normal mucosa, 46% and 14% of adenomas, and 60% and 36% of CRCs, respectively. The promoter methylation of DLC-1 was related with the reduction of gene expression and the advanced Duke's stages (Stage C and D). DLC-1 promoter methylation and KRAS mutations are common concurrent pathological alternations. Conclusions. Epigenetic alternation plays a key role in the transcriptional silencing of DLC-1. It is also an independent risk factor related to the carcinogenesis of colorectal tumors and spans over its pathogenesis process. Therefore, DLC-1 promoter methylation quantitation may have a promising significance in the evaluation and management of CRC patients.


Introduction
Colorectal cancer (CRC) arises as a consequence of genetic and epigenetic alterations. Somatic mutations and epigenetic alterations have been frequently observed in CRC and are considered to be the driver factors of colorectal tumorigenesis [1]. e neoplastic progression of CRCs mostly follows a mucosa-adenoma-carcinoma step-by-step pattern, which is accompanied by an accumulation of successive genetic alterations (e.g., APC, KRAS/BRAF, and TP53) [2][3][4][5]. Widespread promoter CpG island methylation, also referred to as the CpG island methylator phenotype (CIMP), has been extensively studied in CRC. CIMP-high CRC signi�cantly correlates to microsatellite instability, BRAF mutation and low TP53 mutation rate, while CIMP-low tumors present a higher rate of KRAS mutation [6]. Despite these extensive researches, the molecular mechanisms underlying the colorectal mucosa-adenoma-carcinoma transformation process and the progression of CRC are still not fully understood.
Deleted in liver cancer-1, DLC-1 gene is a recently iden-ti�ed tumor suppressor gene. It is located on chromosome 8p21-8p22 and affected in multiple cancers. DLC-1 is a regulator of the Rho family of small GTPases [7][8][9][10][11][12]. e principal function of DCL-1 is to catalyze the conversion of active guanosine triphosphate-(GTP-) bound RhoA to the inactive guanosine biphosphate-(GDP-) bound form. DLC-1 is recurrently downregulated or inactivated by epigenetic mechanisms in the initiation and progression of cancers [13].
Ullmannova and Popescu [14] �rst reported the downregulation of DLC-1 mRNA expression in CRCs utilizing cancer-pro�ling arrays. Moreover, aberrant methylation of CpG islands in the promoter region of DLC-1 is a common mechanism leading to the transcriptional silencing [15][16][17][18], suggesting DLC-1 methylation is associated with the downregulation of this gene in CRC and DLC-1 may be a potential tumor suppressor gene. However, methylation and mRNA expression status of DLC-1 and its role in the adenomacarcinoma progression have not been de�ned.
To study the role of DLC-1 promoter methylation in the colorectal carcinogenesis routing down the adenomacarcinoma process, here we quanti�ed the methylation status and mRNA expression of DLC-1 and assessed its relation to various clinicopathological parameters and molecular features, especially the mutation status of KRAS and BRAF in 185 colorectal tissue samples from different disease stages.

Patients.
Eighty colorectal carcinomas, 57 adenomas, and 48 samples of adjacent histologically normal mucosa were collected from the Department of Gastroenterology at Shanghai Changning District Cental Hospital and Huashan Hospital. Tissue samples were frozen within 2 hours of removal and then stored at −80 ∘ C. All the tumors contained more than 80% tumor cells as con�rmed by histological examination of sequential sections. e patient's gender, age, Duke's stage, tumor differentiation, and tumor size and location were obtained from surgical and pathological records. In compliance with the Helsinki Declaration of 1975 as revised in 1996, this study was approved by the Institutional Review Board of Shanghai Changning District Center Hospital.

Cell Lines.
Four human colorectal carcinoma cell lines (SW480, LoVo, LS 174T, and COLO 320) were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) cell bank. e SW480 and LS 174T cells were cultured in RPMI 1640 medium; the LoVo and COLO 320 cells were cultured in F12K medium (Gibco, Grand Island, NY, USA). Both the mediums were supplemented with 10% fetal bovine serum (Sigma, St. Louis, MO, USA) and incubated in 5% CO 2 at 37 ∘ C.

Methylation-Sensitive High-Resolution Melting Analysis
(MS-HRMA) of DLC-1. Genomic DNA was extracted from each tumor tissue sample and cell line using QiaAmp NDA extraction kit (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions. Genomic DNA was then mod-i�ed by sodium bisul�te using EZ DNA Methylation kit (Zymo Research, Irvine, CA, USA) according to the manufacturer's instructions. A methylated reference sample, the CpGenome Universal methylated human male genomic DNA (Chemicon, Billerica, MA, USA), and genomic DNA isolated from the peripheral blood mononuclear cells of a healthy male individual were subjected to the bisul�te modi-�cation procedure and used as control standards. e methylated reference was then diluted with the unmethylated control in 0%, 1%, 10%, 30%, 50%, 80%, and 100% ratios to carry out the standard curve for the MS-HRMA. e PCR ampli�cation and HRMA were then performed using the methylation speci�ed primers (MS-HRM-F, MS-HRM-R, Table 1). e 20 L volume master mix contained 10 L Premix Taq Hot Start Version (2X) (TaKaRa BIO, Shiga, Japan), 1 L forward primer (5 M), 1 L reverse primer (5 M), 1 L SYTO 9 dye (30 M) (Invitrogen, Carlsbad, CA, USA), 6 L deionized distilled water, and 1 L DNA template (15-25 ng/ L). e mix was subjected to PCR on a Rotor-Gene Q real-time platform (Qiagen, Valencia, CA, USA). e samples were denaturated at 95 ∘ C for 2 min, followed by 40 cycles of 95 ∘ C for 30 sec, 58 ∘ C for 30 sec, and 72 ∘ C for 30 sec. Aer PCR, the products were again denaturated at 95 ∘ C for 2 min and cooled down to 40 ∘ C for 2 min to form the heteroduplex. HRMA was then performed at 0.2 ∘ C/s from 50 ∘ C to 95 ∘ C. Each sample was tested in duplicate. e HRMA and data interpretation were performed using Rotor-Gene Q 1.7 soware. A differential pro�le was then evaluated for each sample by comparing �uorescence at the melting point with the �uorescence of the unmethylated control. ere was a linear correlation between the differential �uorescence and the dilution of methylated DNA.
2.�. Bisul�te D�A Se�uen�ing. PCR primers were designed as previously described by Guan et al. [19] to amplify a 292 bp region of the DLC-1 promoter that encompasses 35 CpGs ( Figure 2(a)). e PCR product was then subcloned into the pMD19-T expression vector by using a TA Cloning Kit (TaKaRa BIO, Shiga, Japan). We selected four clones from each plate and sent the recombinant plasmids to MAP Biotech (Shanghai, China) to complete the bisul�te DNA sequencing procedure.  [20,21], producing small amplicons for the HRMA. e PCR mix preparation and the HRMA procedure were carried out with the identical methodology as described in the MS-HRMA of DLC-1 section. For each sample, the normalized melting curves were evaluated and then compared with the mutant and wild-type controls in a deduced difference plot. e samples with distinct melting curves compared with the wild-type allele were recorded as positive.

Statistical
Analysis. e association of DLC-1 methylation and mRNA expression with patients' clinical and genetic variables was analyzed with the 2 test. < was considered statistically signi�cant.  (Figure 1). When methylation level higher than >1% was considered methylation positive, DLC-1 methylation was observed in 5/48 (10%) of normal mucosa, 26/57 (46%) of adenomas, and 48/80 (60%) of CRCs. However, extensive methylation (methylation level of 10-100%) was observed in only 8/57 (14%) of adenomas, 29/80 (36%) of CRCs, and 0% of the normal mucosa ( 3.�. Bisul�te DNA Se�uencing. Bisul�te DNA sequencing was employed to determine the comprehensive methylation pattern of the 5 � -CpG islands in the DLC-1 promoter. Five CRCs and �ve adenoma samples that were identi�ed by MS-HRMA as extensive methylation and partial methylation, respectively, and two normal mucosa tissue samples were ampli�ed with a 292 bp fragment in the DLC-1 promoter, covering 35 CpG sites. Bisul�te DNA sequencing con�rmed the CpG islands in the normal sample unmethylated; all �ve CRC samples were frequently methylated. e tumor samples shared some common methylation sites, while the overall methylation patterns were distinct. Consistent with the MS-HRMA data, the CpG island exhibited that more CpG sites were methylated in clones obtained from CRCs than those from adenomas. We found the methylation of DLC-1 promoter more frequently methylated in the CRC samples (C1, 60% and C2, 31%) than in the adenomas (A1, 9% and A2, 4%) (Figure 2(b)).

Correlation of DLC-1 Methylation with Clinical and
Pathological Features. e correlations between the methylation status of DLC-1 and clinical features in the CRC and adenoma patients were presented in Table 3. DLC-1 methylation was signi�cantly higher in the CRC tissues with advanced Duke's stage (Stage C, versus Stage A + B, 0.025; Stage D, versus Stage A + B, 0.002). However, no signi�cant difference in the promoter methylation level was observed between Stages C and D ( 0. 02). Other than this, there were no signi�cant associations between DLC-1 methylation status and other clinical factors. Among adenomas, there were no signi�cant associations between DLC-1 methylation status and clinical parameters.

Correlation of DLC-1 Methylation with DLC-1 mRNA
Expression. We determined the mRNA expression of DLC-1 gene in 4 colon cancer cell lines, normal mucosa, adenomas, and CRC tissues. GAPDH mRNA was used as a housekeeper for cDNA integrity. DLC-1 mRNA expression was detected in Colo320 cells but not in SW480, LoVo, and LS 174T cell lines which harbor methylation in the promoter region of DLC-1. DLC-1 mRNA expression was observed in 41/48 (85%) of normal mucosa specimens. Of the 57 adenomas and 80 CRC tissues, DLC-1 mRNA was downregulated in 30/57 (52%) and 63/80 (79%) samples, respectively ( Table 2). Since very low level of methylation could not lead to signi�cant downregulation of gene expression, we considered 10% methylation of DLC-1 promoter region as the cutoff. ere was a correlation between DLC-1 mRNA expression reduction and extensive promoter methylation status (Table 4, 0.022). Our results suggest that the reduction or loss of DLC-1 mRNA expression was related to the extensive methylation in DLC-1 promoter.

Correlation of DLC-1 Methylation with KRAS and BRAF
Mutations. We found KRAS mutations in 4/48 (8%) normal mucosa, 10/57 (18%) adenomas, and 32/80 (40%) CRCs (Table 2). When we examined the association of KRAS mutations to DLC-1 methylation, a statistically signi�cant correlation was observed only in the CRCs ( 0.0 0), but not in adenomas ( 0. ). Hereaer, we investigated the association between extensive methylation of DLC-1 promoter and mutations in KRAS codon12, 13. In terms of CRCs, 62% (18/29) of tumors with extensive DLC-1 methylation showed KRAS mutations, while KRAS mutation alterations were present in only 32% (6/19) of tumors with partial methylation and 25% (8/32) of tumors with unmethylated DLC-1 promoter. ese data demonstrated that KRAS mutations signi�cantly correlated with extensive DLC-1 promoter methylation in CRCs. However, we did no further investigation in the correlation between BRAF mutation and DLC-1 methylation because only 4 of 185 tissues turned out to be BRAF V600E positive from the HRMA, limiting the statistical power of the data.

Discussion
e identi�cation of genes contributing to the development of colon cancer is critical to the understanding of molecular mechanisms of carcinogenesis and may provide new strategies for clinical therapy. A new candidate tumor suppressor gene, DLC-1, was �rst identi�ed as a rat p122RhoGA� homolog [10]. is gene is diminished or silenced in various types of human cancers as well as in metastatic cells compared to nonmetastatic cells [7-9, 11, 12]. DLC-1 has not been extensively studied in CRC. Moreover, the transcriptional regulation of DLC-1 gene expression through epigenetic mechanisms has not been investigated in the normal adenoma-carcinoma sequence. e relationships between DLC-1 methylation status and clinic pathological variables in CRC remain to be elucidated. In this study, we used MS-HRMA to detect DLC-1 promoter methylation in 48 normal mucosa, 57 adenomas, 80 CRC tissue samples, and 4 CRC cell lines. e methylation status of DLC-1 did not show a strong correlation with the widely accepted risk factors of CRC including age and sex (estrogen) [22]. We found the mRNA expression of DLC-1 was decreased when DLC-1 promoter was methylated in the cancerous tissues. Moreover, partial methylation was frequently observed in adenomas as well as CRC. Extensive methylation was primarily observed in CRCs but less prevalent in adenomas. In addition, by quantifying the methylation status in DLC-1 promoter, we also found an accumulation of aberrant methylation following the adenoma-carcinoma sequence. During this stepwise progression, the methylation level of the DLC-1 promoter region increased gradually and the existence of cytosine methylation expanded widely. None of the normal adjacent mucosa specimens showed extensive methylation in the DLC-1 promoter region. Extensive methylation is a characteristic of a more advanced CRC while partial methylation is the feature of an earlier stage of CRC. ese results indicate that the methylation of DLC-1 promoter runs through the whole course of colorectal tumorigenesis.
ese pieces of evidence indicate that the epigenetic mechanism is a driving factor of DLC-1 transcriptional silencing and may be involved in the tumorigenesis of CRC as an independent risk factor.
DLC-1 methylation levels were also found to be signi�cantly associated with Duke's stage. e methylation levels were signi�cantly higher in advanced stage (Stages C and D) tumors, which indicated the role of DLC-1 in the CRC progression. is result is consistent with previous �ndings, which showed the methylation status of DLC-1 was related to TNM stages [23]. Jin et al. [24] reported that the knocking down of DLC-1 transcriptional expression by RNAi resulted in the promotion of LoVo CRC cell proliferation, migration, and cell cycle progression, that is critical for tumor growth and metastasis. us, our results further imply that methylation of DLC-1 promoter is a potentially biomarker for prognosis evaluation of CRC.
KRAS oncogene is a guanine nucleotide-binding protein with GTPase activity that is involved in signal transduction. In this study, we con�rmed that the extensive DLC-1 meth- ylation was associated with the KRAS mutations in CRCs but not in adenomas. Our previous study showed that DLC-1 methylation signi�cantly correlated with PIK3CA mutations in Paget's disease [25]. ese �ndings highlighted the interaction between genetic and epigenetic alterations in CRC, although the mechanism underlying this phenomenon requires further study. On the other hand, overactivation of certain oncogenic pathways is known to affect the activity of methyltransferases and the regulation of gene transcription, thus possibly affecting components of the MAPK pathway, such as RAS, RAF, MEK, and ERK [26,27]. Data from KRAS transformation studies suggest that activated KRAS promotes P16 methylation [28]. Stable transformation of colon cancer cells with KRAS increased DNA methyltransferase activity and P16 gene methylation [29]. Collectively, the results from our study and previous work by others suggest epigenetic alterations of DLC-1 might occur as a consequence of overactivation of the oncogenic pathway in cancer.
In conclusion, the current data suggested that methylation-induced epigenetic silencing of DLC-1 is involved in the colorectal tumorigenesis and has a strong correlation with the Duke's stage, and with KRAS mutation. Quantitative detection of DLC-1 promoter methylation may have a promising clinical signi�cance in the evaluation of CRC patients and in the management of the disease.

Author's Contribution
Both Haixia Peng and Feng Long contributed equally to this study.

Con�ict o� �nterests
e authors declare that they have no con�ict of interests.