Background
Colorectal cancer (CRC) is the third most common malignant neoplasms in the world. Each year almost 1.4 million new cases were diagnosed and 0.7 million patients died of this disease [
1]. Aberrant DNA methylation is an important driver mechanism in tumorigenesis [
2] and ever-growing number of genes showed abnormal methylation in CRC [
3‐
5]. Because aberrant methylation alteration can begin very early in tumor progression, especially earlier than protein expression changes and malignant cell proliferation [
6], such genes are promising to be good indicators for early diagnosis and prognosis of CRC.
The
p16 gene (also named as
CDKN2A, INK4A, CDK4I) is one of the most studied epigenetic markers in CRC. As a tumor suppress gene,
p16 inactivation results in loss of the cellular capacity to block cell cycle and has been widely reported in human malignancy [
7,
8]. The
p16 hypermethylation is a frequent event in CRC and acts as a major mechanism leading to
p16 inactivation [
7]. Since the methylation change of
p16 gene in cancer was firstly identified at promoter-associated region [
9], previous most research of
p16 aberrant methylation focus on its promoter and/or upstream-exon1 regions. Concerning the change of
p16 promoter methylation in CRC, most investigators observed that the tumor tissues were more highly methylated than adjacent normal mucosae [
10‐
15]. But a recent large-scale research found that nearly 10% of CRC cases had greater methylation at
p16 promoter region in the adjacent non-neoplastic tissues than in the carcinoma [
16]. Regarding the impacts of
p16 promoter hypermethylation (PHM) on CRC, some investigations revealed a correlation between it and some clinicopathological parameters or poor prognosis [
10‐
14,
16], such as
p16 PHM with larger tumor size, more frequent recurrence and shortened survival. But others did not observe statistically correlation [
17‐
19]. Even some reported CRC patients with
p16 PHM had a better survival [
20]. Due to those inconsistent results, the clinicopathological significance of
p16 PHM remains controversial. More optimal methylation loci within
p16 gene are still to be explored.
Recently, gene body methylation (GbM) was found that frequently occurred in the transcribed regions of many oncogenic regulated genes and actively involved in multiple regulation processes [
21,
22]. More detailed genome-wide studies have demonstrated that GbM can alter gene expression by silencing alternative promoters or effecting transcription elongation or regulating splicing [
23‐
25]. Thereby GbM is suggested as a novel biomarker or therapeutic target in cancer [
26]. However, the intragenic DNA methylation of
p16 gene received less attention and is poorly understood up to date. A few studies explored the methylation status of
p16 exon 2 region and found it was frequently methylated in head and neck squamous carcinoma [
27], oesophageal cancer [
28] and breast cancer [
29], and its methylation changes were associated with breast carcinogenesis. Whether hypermethylation of
p16 exon 2 also occurs in other cancers remains unclear.
To explore the methylation changes of
p16 gene body in CRC, we focused on CpG-rich regions in
p16 gene body, namely exon 1 and exon 2. Their methylation levels were evaluated in paired CRC and adjacent normal tissues by LC-MS/MS method, which can quantify the average methylation level of target genomic region [
30]. Statistical analysis was carried out to find more reliable methylation biomarkers. Moreover, we analyzed the relationship between methylation status of each region and clinicopathological parameters of CRC patients, such as gender, age, differentiation and T/N/Dukes stage, to investigate whether they were associated.
Methods
Chemicals and reagents
Cytosine (Cyt), Adenine (Ade) and Protease K were purchased from Sigma (St. Louis, USA). Isotopes Cyt13C15N2 and Adenine-2-13C were purchased from Toronto Research Chemicals Inc. (Toronto, Canada) and C/D/N Isotopes Inc. (Quebec, Canada), respectively. PCR reagents were purchase from TAKARA Bio Inc. (Dalian, China). Ammonium formate, methanol, acetonitrile, formic acid (chromatographic grade) were purchased from Merck (Darmstadt, Germany). The monoclonal antibody against p16 protein and the Streptavidin-Peroxidase Detection Kit for immunohistochemistry were purchased from ZSGB Bio (Beijing, China).
CRC tissue samples
Thirty pairs of colorectal cancer tissue and corresponding para-carcinoma tissue were collected from Department of gastrointestinal surgery, Affiliated Hospital of Guangdong Medical University from 2014 to 2015. The 30 patients comprised 18 males and 12 females, with a mean age of 56.5 years (range 20–77). The mean tumor size was 4.5 cm3. The adjacent tissues were about 10 cm distant from tumors. All CRC samples were confirmed by pathological diagnosis. Fresh tissues were snap frozen in liquid nitrogen and stored at − 80 °C until further protocols. Clinical data were collected prospectively. The collection of tissue samples for this project was approved by Ethic Censor Committee of Affiliated Hospital of Guangdong Medical University and manipulated fully in accordance with its guidelines.
DNA extraction and bisulfite conversion
Genomic DNA was extracted from tissue samples using Tissue Genomic DNA Extraction Kit (Tiangen, Beijing, China) following the manufacture’s protocols. The concentration and purity of genomic DNA were determined using Nanodrop2000 Ultramicro Spectrophotometer (Thermo Scientific, Massachusetts, USA). 200 ng DNA was used for bisulfite conversion with accordance to the specification of EZ DNA Methylation-Gold Kit (ZYMO, Irvine, USA).
PCR amplification and purification of target regions
The whole CpGs islands lying in the exon 1 and exon 2 within
p16 gene body were targeted and amplified from bisulfite-converted genomic DNA via nested PCR using specific modified primers (Table
1). The outer PCR amplification was conducted in a 25 μL total reaction volume containing 1.0 μL of 10 μM of each primer, 1 U
ExTaq DNA polymerase, and approximately 75 ng bisulfite-treated genomic DNA. The inner PCR was performed in a 50 μL total reaction volume including 2.0 μL of 10 μM of each primer, 2 U
ExTaq DNA polymerase and about 30 ng outer PCR products. PCR amplification was implemented in Veriti gradient thermal cycler (Applied Biosystem, Carlsbad, USA). The inner PCR products were evaluated using 1.5% agarose gel electrophoresis and bidirectionally sequenced with inner PCR primers to ensure the sequence correctness. Acquired PCR products of two target regions were purified according to the instruction of EZ gene Gel/PCR Extraction Kit (ZYMO, Irvine, USA), and their concentration was measured by Ultramicro Spectrophotometer. At the same time, three specific DNA samples with known methylation level (0%, 47% and 100%) were also prepared as controls according to our prior work [
30].
Table 1
The primer sequences and PCR conditions
p16 exon 1 |
Outer | TTAGAGGATTTGAGGGATAGGGT | TACAAACCCTCTACCCACCTAAAT | 56 °C | 324 bp |
Inner | GGATTTGAGGGATAGGGT | CCCTCTACCCACCTAAAT | 56 °C | 313 bp |
p16 exon 2 |
Outer | TGGTAGGTTATGATGATGGGTAG | ATCCTCACCTAAAAAACCTTCC | 54 °C | 321 bp |
Inner | GGTTATGATGATGGGTAG | TTACTACCTCTAATACCCC | 53 °C | 273 bp |
Methylation level determination of target regions by a LC-MS/MS method
The NQ-E (Nucleobases Quantitation of bisulfite amplicon coupled with an Equation) method described in a recent publication [
30], was applied to determine the methylation levels of target regions. Briefly, 100 ng purified PCR products of
p16 exon 1/2 region were added into 100 μL of 100 ng/mL mixed internal standard solution including Cyt
13C
15N
2 and Adenine-2-
13C and mixed evenly, then dried at 60 °C. The residue was mixed with 200 μL of 88% formic acid (
v/v) and hydrolyzed at 140 °C for 90 min. Hydrolyzed product was dried and dissolved in 200 μL acetonitrile - 0.7 mM aqueous ammonium formate (93, 7, v/v), and then centrifuged at 12,000 g for 5 min. The final supernatant was extracted for LC-MS/MS analysis.
After LC separation and MS detection, the quantification of cytosine (Q
Cyt-M) and adenine (Q
Ade-M) in target amplicons were accomplished in multiple reactions monitoring mode. Based on the LC-MS/MS data and Genebank data (P
Gua-D and P
Cyt-CpG-D in target genomic region), the average methylation level of target region was calculated using the following two formulas as described previously [
30]:
$$ {\mathrm{P}}_{\mathrm{GuaCyt}\hbox{-} \mathrm{M}}={\mathrm{Q}}_{\mathrm{Cyt}\hbox{-} \mathrm{M}}/\left({\mathrm{Q}}_{\mathrm{Cyt}\hbox{-} \mathrm{M}}+{\mathrm{Q}}_{\mathrm{Ade}\hbox{-} \mathrm{M}}\right) $$
(1)
And
$$ \%\mathrm{Methylation}=\left({\mathrm{P}}_{\mathrm{Gua}\mathrm{Cyt}-\mathrm{M}}-{\mathrm{P}}_{\mathrm{Gua}-\mathrm{D}}\right)/{\mathrm{P}}_{\mathrm{Cyt}-\mathrm{CpG}-\mathrm{D}}\times 100\% $$
(2)
Immunohistochemistry assay
Immunohistochemical analyses for p16 protein were performed in 30 CRC samples. The paraffin-embedded tissue sections were deparaffinized with xylene and rehydrated. For antigen retrieval, sections were immersed in 10 mM citrate buffer and microwaved for 5 min. Endogenous peroxidase and non-specific protein binding was blocked by incubating with 3% H
2O
2 and then with 10% goat serum. Then sections were incubated, in turn, with the anti-p16 primary antibody (dilution 1:200) at 4 °C overnight, with the biotin-labelled secondary antibody for 15 min and with HRP-labelled streptavidin for 15 min. Signals were visualized with DAB for 1 min, with slight counterstaining using hematoxylin. In each experiment, the primary antibody was omitted as negative controls. The sections were evaluated independently by two investigators as described previously with slight modification [
31]. The degree of immunohistochemical staining was evaluated by the sum of the staining intensity score (0, no, 1: light yellow, 2: yellow, 3: brown yellow) and the staining proportion score (0, < 25%, 1: 25–50%, 2: 51–75%, 3: > 75%). The p16 protein expression was assessed by the final score (0 ~ 6) of immunohistochemical staining.
Statistical analysis
SPSS 17.0 software was used throughout. The difference of methylation level between CRC tissue and adjacent normal tissue at each region was analyzed by paired samples T-test (
p16 exon1A-T, exon2A-T) and the difference between
p16 exon 1 and exon 2 was analyzed by independent samples T-test (exon1T - exon2T). Heat-map analysis was carried out by HemI 1.0 software (
http://hemi.biocuckoo.org/index.php). The association of
p16 exon 1/2 methylation level and p16 protein expression level in 30 CRC tissues were analyzed using Spearman test. The relationship between the methylation status of each region and the clinicopathological features of 30 CRC patients were analyzed using Fisher’s exact test. A
p value < 0.05 was considered statistically significant.
Discussion
It is largely accepted that
p16 promoter hypermethylation occurs frequently in CRC. However, its clinicopathological significance remains controversial because of the inconsistent research results. In previous studies, the methods commonly used to quantify DNA methylation included methylation specific PCR (MSP) [
9,
11,
13,
17,
28] and quantitative MSP [
10,
20], MethyLight [
12,
19] and methylation-sensitive high resolution melting (MS-HRM) [
29], BSP [
15] and pyrosequencing [
16]. Some of them analyzed one or several CpG sites, and others analyzed a genomic region with some length limitation about < 200 bp. Moreover, different studies targeted different CpG sites or genomic regions. These methodological factors resulted in large differences and non-comparability, which may be one of important reasons for inconsistent results of
p16 PHM in previous researches.
To address this issue, we adopted a LC-MS/MS approach in present study, which was recently reported [
30,
32]. Although the LC-MS/MS method can’t distinguish the methylation status of single CpG, it can provide an average methylation level across all CpG sites of a target region and make it easy to compare the methylation alteration between different samples. A significant characteristic of this method is no limitation on fragment length and CpG density/number, which permits to detect a whole CpG island (usually 200 ~ 3000 bp) and to analyze more CpG sites at one time. The detection results of three methylation controls demonstrated it covered a wide detection range (from 0 to 100% methylation) and had a high accuracy.
Therefore, we adopted this LC-MS/MS approach to determine methylation levels of the whole CpG islands within
p16 gene body in CRC tissues. We found that the overall methylation levels of two CpG-rich regions were both significantly higher in tumors than in adjacent normal tissues. Comparing with
p16 exon 1, higher and more frequent hypermethylation occurred at
p16 exon 2 in tumors. It’s worth noting that there were 2 cases showed higher methylation in adjacent normal tissue than carcinoma tissue at
p16 exon 1 region, being in line with previous pyrosequencing result of
p16 promoter-exon1 region [
16]. This fact might be an important reason for the controversy of
p16 promoter hypermethylation as CRC biomarker. ROC curve analysis revealed that
p16 exon 2 had a high sensitivity and specificity for distinguishing adjacent normal tissue and CRC tissue, and the combination use of both indicators could further improve the sensitivity and specificity. These results suggested that longer genomic region covering more CpG sites could better reflect the overall methylation status of some specific gene, and therefore could be better indicators.
In our research, aberrant hypermethylation of
p16 exon 1 or exon 2 were observed in only about 50% of the CRCs, which were similar with previous studies [
27,
29]. This frequency is relatively modest compared to some other loci such as ADAMTS19 [
33], and it may be a limitation as biomarkers. However, combining two loci, it reached 73.3% when either exon 1 or exon 2 was aberrant hyper-methylated. ROC curve also showed the combined use of two loci could promote the sensitivity and specificity. Concerning the importance of
p16 gene in tumorigenesis, our findings supported that the combination of
p16 exon 1 and exon 2 could be an effective methylation marker of CRC.
We further explored the relationship between the aberrant hypermethylation of
p16 gene body and clinicopathological features of CRC patients. Because
p16 methylation may occur in non-neoplastic tissues, a threshold should be set to confirm positive hypermethylation. Our results showed the average methylation differences between tumors and adjacent normal tissues of exon 1 and exon 2 both reached 10% (Fig.
4c). Therefore, we set a threshold value of 20% (two fold of 10%) to ensure that only aberrant hypermethylation cases were assigned as positive. This threshold value was in line with an previous systematic study [
16]. Considering cases with at least 20% methylation difference between tumor and normal tissue as positive, all clinical cases were classified into two categories. Subsequent statistical analysis uncovered a significant correlation between
p16 exon 1 and N/Dukes staging, also between
p16 exon 2 and T staging, which suggested the hypermethylation of
p16 gene body was associated with CRC invasion and metastasis. These findings further supported that the combination of
p16 exon 1 and exon 2 could be an effective methylation marker of CRC.
Currently, methylation alteration of exon-based gene body has attracted more attention since GbM was found that frequently occurred in some oncogenic genes and DNA methylation in transcribed regions were also correlated with gene expression [
25,
26]. It would provide more detectable loci and may be novel biomarkers or therapeutic targets in cancer [
26,
32]. Here, our study observed that the methylation level of
p16 gene body had high sensitivity and specificity as potential CRC biomarker, and the hypermethylation of
p16 exon 1 or exon 2 was associated with N/Dukes or T staging. The immunohistochemistry assay demonstrated a negative correlation between
p16 exon 1/2 methylation level and p16 protein expression. These results suggested that the gene body methylation could affect
p16 gene expression, possibly by preventing aberrant transcription initiation or effecting transcription elongation, and thus be associated with CRC progression. These findings will promote the application of
p16 gene body methylation as biomarker for CRC diagnosis.