Aberrant methylation of PCDH17 gene in high-grade serous ovarian carcinoma

Abstract

BACKGROUND:

Aberrant DNA methylation of protocadherins (PCDHs) has been associated with development and progression of various types of cancer. It could represent possible direction in the search for critically needed tumor biomarkers for ovarian cancer.

OBJECTIVE:

To investigate methylation of $\delta$2 group of non-clustered PCDHs in high-grade serous ovarian carcinoma (HGSOC) tissue in comparison with control tissue.

METHODS:

We used next-generation sequencing for detecting regions with the most altered methylation. For further confirmation of discovered alterations we used methylation-sensitive high-resolution melting analysis.

RESULTS:

PCDH17 methylation was detected in almost 70% of HGSOC patients without any methylation in the group of control samples and was found both in the late stage tumors as well as in the early stage ones. Other selected PCDHs did not show any relevant changes in methylation. Subsequent gene expression analysis of PCDH17 revealed decreased expression in all of the tumor samples in comparison to the control ones. Statistically significant negative correlation was found between methylation and levels of expression suggesting potentially methylation-based silencing.

CONCLUSIONS:

Methylation of PCDH17 could play an important role in development and progression of HGSOC and has potential to become a target in the search for new clinical biomarkers.

1. Introduction

Ovarian cancer (OC) is the 6th most common cause of cancer death in Europe for females. The estimated number of new cases in Europe in 2012 was 65,538 with 42,704 deaths [8]. OC is a nonspecific term for any cancerous growth that occurs in the ovary and covers heterogeneous group of tumors with distinct morphologic, prognostic, etiopathogenetic, and molecular characteristic. The majority of ovarian cancers arise from the epithelium of the ovary. The most common histological type accounting for up to $\sim$ 80% of advanced epithelial OC is invasive serous carcinoma, recently subdivided into two distinct disease entities, high-grade and low-grade serous carcinomas [15]. High-grade serous ovarian carcinoma (HGSOC) is characterized by an advanced stage at onset, nearly universal mutation of the TP53 gene, mutations in the homologous recombination DNA repair pathway (BRCA1 and BRCA2 genes) and widespread copy number alterations [7]. Originally, the ovary was thought to be the primary site of HGSOC tumorigenesis with the ovarian surface epithelium as the cell of origin. In recent years, however, there has been emerging evidence that the majority of HGSOC ($\sim$ 60%) originate in the fimbria of the fallopian tube and arise from serous tubal intraepithelial carcinomas (STIC) [25, 16]. Whether the ovarian surface epithelium can be considered the site of origin for the rest of HGSOC is still matter of debate [4].

Similarly to other malignancies, OC initiation and progression is also driven by epigenetics alterations [2]. It is well established that DNA methylation is one of the major epigenetic mechanisms regulating gene expression. In tumor cells, DNA methylation is usually redistributed between genomic hypomethylation with localized CpG island hypermethylation. Hypermethylation of CpG islands in the promoter region of genes involved in the control of the cell cycle, apoptosis and drug sensitivity as well as tumor suppressor genes results in transcriptional silencing [2]. Aberrant methylation of multiple CpG islands in the promoter region of various genes associated with OC has been observed in numerous studies [11, 12, 14].

Protocadherins (PCDHs) are group of transmembrane proteins belonging to the cadherin superfamily. They undergo cancer-related changes and their downregulation or absence in malignant cells has been associated with cancerogenesis and cancer progression [10]. The tumor suppressor role of PCDHs has been recently affirmed, and current studies also showed aberrant DNA methylation of various PCDH genes in human malignant tumors [9, 18, 27]. Based on their genomic structure they are subdivided into clustered and non-clustered groups. The clustered PCDHs, comprising $\alpha$, $\beta$, and $\gamma$ groups, are arranged in tandem on a single chromosome. The non-clustered PCDHs are located on multiple chromosomes at three different chromosomal loci and divided into $\delta$1, $\delta$2, and $\epsilon$ groups [5]. In this study, we focused on $\delta$2 group of non-clustered protocadherins PCDH8, PCDH10 and PCDH17. Different studies have confirmed the significance of altered methylation of these PCDHs in other types of cancers (PCDH8 in bladder or prostate cancer [21, 26], PCDH10 in colorectal cancer or lung cancer [22, 24] and PCDH17 in bladder or breast cancer [23, 28]). We used next-generation sequencing (NGS) for detecting regions with the most altered methylation status of selected PCDHs in HGSOC tissue in comparison with control tissue. To confirm discovered alterations, we further analyzed tissue from more extensive group of patients using methylation-sensitive high-resolution melting analysis (MS-HRM). We also investigated the PCDH17 expression level and its relationship to PCDH17 methylation.

2. Materials and methods

2.1. Study group

The study group consisted of 51 patients with HGSOC and 35 patients with a non-malignant diagnosis (such as descent of the uterus with adnexectomy, or uterine leiomyomas, etc.) surgically treated at the University Hospital Hradec Kralove between years 2001–2014. In addition to total number of 102 formalin-fixed, paraffin-embedded (FFPE) samples (51 tumors; 35 normal ovary samples and 16 samples from epithelium of the fallopian tube fimbriae used as control samples) fresh frozen samples were obtained from 20 patients (10 tumors and 10 control samples). All samples were reviewed by an experienced gynecopathologist (J.L.). The carcinomas were classified according to the current WHO classification of tumors of female reproductive organs [13]. Stage I or II was classified in 12 tumors, 39 tumors were stage III or IV. The median age of patients with carcinoma at the time of diagnosis was 57 years (40–79 years), median age at the time of surgery in control group was 57 years (43–84 years). The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of University Hospital Hradec Kralove (201611 S06P). Informed consent related to fresh frozen tissue samples was obtained from each concerned patient.

2.2. DNA extraction and bisulfite conversion

Genomic DNA was extracted using QIAmp DNA Mini Kit (Qiagen, Hilden, Germany) following the manufacturer’s instruction. The purity of extracted DNA was examined spectrophotometrically and then quantified using the Qubit${}^{\mathrm{\circledR }}$ Flourometer (Thermo Fisher Scientific, Waltham, MA, USA). DNA (500 ng) was bisulfite-converted with EZ DNA Methylation-Gold™ Kit according to the manufacturer’s protocol (Zymo Research Corporation, Irvine, CA, USA).

2.3. Next generation sequencing

Selected sites of the promoter regions and the adjacent exons of the PCDH8, PCDH10 and PCDH17 genes were sequenced in 20 fresh frozen ovarian tissue samples (10 tumors and 10 control samples). Specific primers for amplifying the regions of our interest were designed in methylation primer designing software (http://www.urogene.org/cgi-bin/methprimer/methprimer.cgi). Genomic coordinates based on GRCh 38/hg38 assembly were as follows: chr13: 52,848,432–52,849,262 (PCDH8), chr4: 133,149,215–133,149,575 (PCDH10) and chr13: 57,631,479–57,632,603 (PCDH17). Primer sequences and amplicons information along with used DNA polymerase for each amplicon are listed in Table 1. Genomic coordinates of all analyzed CpG sites are summarized in Supplement. Sequencing libraries were prepared using the Multiplicom approach. First PCRs were conducted for 40 amplification cycles using AmpliTaq Gold${}^{\mathrm{\circledR }}$ DNA Polymerase (Thermo Fisher Scientific), with annealing temperatures at 56/59${}^{\circ }$C or Platinum${}^{\mathrm{\circledR }}$ Taq DNA Polymerase (Thermo Fisher Scientific), with annealing temperatures at 56/60${}^{\circ }$C. All PCR amplifications were performed in Veriti™ Thermal Cycler (Thermo Fisher Scientific). Bisulfite treated universal methylated and unmethylated DNA (Zymo Research Corporation) were used as controls. Next steps in preparing sequencing libraries were carried out according to the protocol described in our previous study [6]. NGS was performed on MiSeq System (Illumina, San Diego, CA, USA) using Reagent Kit v2 at a 2 $\times$ 250 base pair read length configuration with paired-end reads following the manufacturer’s instructions. For further analysis of sequences acquired as FASTQ files and calculation of methylation status of analyzed CpG sites was used NextGENe${}^{\mathrm{\circledR }}$ software (Softgenetics, State College, PA, USA).

Table 1.

Primer sequences and amplicon information

Amplicon	Primer sequence 5’ $\to$ 3’	Amplicon size with	CpGs/Amplicon	Annealing	DNA polymerase
		adapters (bp)		temperature (${}^{\circ }$C)
PCDH8_1	Fw: *TTTTTTTGAAAGGGAAGTGGTAGT	404	12	59	AmpliTaq Gold
	Rv: *CAAAACTCCAAAAATAAAAAAAAC
PCDH8_2	Fw: *AGAAAGATTTTTTAATTTTTTTT	401	31	60	Platinum Taq
	Rv: *CTCATACCTCCAACCTCAAATAC
PCDH10	Fw: *GGTGGGTGGTGTTTTTGG	381	10	59	AmpliTaq Gold
	Rv: *ACTCTACAACTTAAAACTTTCATTCT
PCDH17_1	Fw: *AGTAAAATATTGTTTGAAAATAGAT	420	22	56	Platinum Taq
	Rv: *ACTAAAAATAAACCAAAAATTTC
PCDH17_2	Fw: *TTGTAGATTAATAGGTTTAGGGAATT	280	14	59	AmpliTaq Gold
	Rv: *CTTAAAAATAAAAACAAAAACCCATA

*adapter overhangs: Fw: AAGACTCGGCAGCATCTCCA. Rv: GCGATCGTCACTGTTCTCCA.

2.4. Methylation-sensitive high-resolution melting analysis

Based on results from NGS we selected CpG sites with most distinct changes in methylation between tumors and control samples for further analysis. To confirm hypermethylation of selected region in the PCDH17 gene, we analyzed 102 FFPE samples using MS-HRM analysis. The method is based on analysis of DNA melt curves following PCR amplification and has the ability to discriminate differences in base pairing and thus to determine the methylation status of bisulfite-converted DNA. MethPrimer was used to design primers for amplification of DNA, considering the fact that FFPE DNA is highly fragmented and amplicons over 200 bp in length result in lower melting resolution. Sequence of forward primer was 5’-AAAAGGATTTATAGATTTGTGGTT-3’, sequence of reverse primer 5’-AACAAATAAAAAAAT ACATCCCAAAC-3’, with amplicon length 144 bp. Amplicon included 11 out of 36 previously analyzed CpGs sites in PCDH17 gene. Genomic coordinates of selected CpG sites are listed in Supplement. PCR amplification and MS-HRM analysis were performed in Rotor Gene Q (Qiagen). PCRs were conducted for 40 amplification cycles using 2 $\times$ EpiTect${}^{\mathrm{\circledR }}$ HRM Master Mix (Qiagen), with annealing temperatures at 55${}^{\circ }$C. Following HRM step consisted of ramping between 62–87${}^{\circ }$C by 0.1${}^{\circ }$C with a hold of 2 s at each step. Each run included a no template control, a bisulfite-converted universal methylated and unmethylated DNA (Qiagen) and prepared standard containing 10% of universal methylated DNA which served as a cut-off for methylation status. HRM data were analyzed using Rotor Gene Q software 2.3 (Qiagen).

2.5. Gene expression

Fresh frozen tissue samples previously stored in RNAlater${}^{\mathrm{\circledR }}$ Stabilization Solution (ThermoFisher Scientific) were homogenized in TRIzol${}^{\mathrm{\circledR }}$ Reagent (ThermoFisher Scientific) in tubes filled with ceramic beads using MagNa Lyser Instrument (Roche, Basel, Switzerland). Total RNA from homogenized tissue was isolated using phenol:chloroform:isoamyl alcohol extraction. The concentration and quality of the RNA was determined by NanoDrop 1000 spectrophotometer (ThermoFisher Scientific). The total RNA (1 $\mu$g) was reverse-transcribed into cDNA using SuperScript${}^{\mathrm{\circledR }}$ VILO™ cDNA Synthesis Kit (ThermoFisher Scientific) with random primers and oligo (dT) according to the manufacter’s protocol using GeneAmp PCR System9700 (ThermoFisher Scientific). No template control and sample with deactivated reverse transcriptase were used as negative controls. Subsequent real-time PCR was performed in Rotor-Gene Q (Qiagen) with 20 ng cDNA using TaqMan Gene Expression Master Mix (2$\times$) and TaqMan Gene Expression Assay (20$\times$) (ThermoFisher Scientific) following the manufacter’s protocol. All reactions were performed in triplicates. Relative expression of PCDH17 was determined using the 2${}^{-\mathrm{\Delta }\mathrm{\Delta }\text{Ct}}$ method [17] with expression level of GAPDH for data normalization.

2.6. Statistical analysis

Categorical variables were compared by two-tailed Fisher’s exact test and/or Chi square test. The Kaplan Maier method and Logrank test were used to determine overall survival rate and significance. Mann-Whitney U Test and Spearman correlation analysis were used to analyze connection between expression level of PCDH17 and methylation status. All tests were two tailed and 0.05 was considered statistically significant. All statistical analyses were performed using STATISTICA Cz (data analysis software system) version 12 (StatSoft, Inc., Tulsa, OK, USA).

3. Results

3.1. Methylation analysis

Using NGS we examined methylation of 89 CpG sites in the promoter region and part of the adjacent exon of the PCDH8, PCDH10 and PCDH17 genes. Sequencing run produced 24.99 million reads with 24.16 millions passing filter. The percentage of bases with a quality score of 30 or higher was 85.31%. The average number of reads allocated to each sample was 35,227 (in total per 5 amplicons). The DNA methylation profiles of 20 fresh frozen samples (10 tumors, 10 controls) are depicted in Fig. 1.

Figure 1.

Next-generation sequencing methylation data. Each dash represents CpG without methylation (cut-off 15%). Methylated CpGs are displayed as circles: white 15–24.9%, grey 25–49.9% and black over 50% methylation. Grey band in the middle of the table marks CpGs clustered in CpG island. Black band at the bottom of the picture shows the gene region covered by HRM assay.

Analyzed region of PCDH8 containing 43 CpGs showed only sporadic methylation in both tumors and controls samples. Except one CpG in one tumor sample, there was no methylation detected in 10 analyzed CpGs of PCDH10. Statistically significant ( 0.01) site specific methylation was present in 10 of 36 analyzed CpGs in PCDH17 gene. In this region near the end of analyzed amplicon, high methylation was present in over 60% of tumor samples, with only minor methylation of one CpG in two control samples. We selected these sites for further analysis by HRM (see Fig. 1). To confirm hypermethylation detected by NGS, we analyzed 102 samples (51 tumors and 51 controls). Statistically significant ( 0.01) methylation-positive pattern was observed in 66.7% (34/51) of OC tissue samples. All of the control samples were methylation free. In the late stages tumors methylation was detected more frequently (in 71.8% of cases; 28/39), than in the early stage ones (50%; 6/12) ($p=$ 0.293).

3.2. PCDH17 expression and its correlation with DNA methylation

To investigate the relationship between PCDH17 hypermethylation and PCDH17 expression, we measured levels of mRNA transcripts in the same samples as used for NGS. In all of the tumor samples PCDH17 was downregulated in comparison to control samples with median fold change of $-$3.82 ($p=$ 0.019). As is evident from Fig. 2, in methylated samples more distinct downregulation was present than in unmethylated ones (median fold change $-$5.82 vs. $-$2.48; $p=$ 0.053).

Figure 2.

Gene expression levels of PCDH17 in the group of samples with methylated and unmethylated PCDH17.

We also examined correlation between PCDH17 expression in individual samples and their methylation status detected by NGS in the part of gene region that was then covered by HRM assay. Statistically significant negative correlation was found in 5 of 11 analyzed CpGs. Correlation coefficients between individual samples and every covered CpG are listed in part A of Fig. 3. The average methylation of analyzed region significantly correlated with levels of expression with correlation coefficient $r=$ 0.721 and $p=$ 0.019 (part B of Fig. 3).

Figure 3.

Analysis of methylation detected by NGS in comparison to relative expression in PCDH17. (A) Correlation of methylation in each of 11 CpGs in gene region covered by HRM assay to the gene expression. Spearman’s correlation coefficient in the bottom line followed by asterisk means p-value less than 0.05, two asterisks mean $p$-value less than 0.01. (B) Relative gene expression versus average methylation of analyzed region of PCDH17. Spearman’s correlation coefficient ($r$) and $p$-value are shown.

3.3. Follow-up

The patients were followed up in February 2017 and the progression free survival and the overall survival data were collected from 48 patients. We were unable to obtain follow-up data of 3 patients, as they were subsequently treated in another hospital. During the follow-up period, relapses occurred in 23/48 (47.9%) of patients and 25/48 (50.0%) patients died due to HGSOC. Overall survival of patients ranged from 2–194 months, with a median of 42 months. No significant correlation between PCDH17 methylation and recurrence or survival was observed.

4. Discussion

Cancer progression is a multi-step process in which adhesion molecules play a significant role. Specific signaling pathways activated by cell-cell interactions are primarily supported and regulated by cadherin-catenin complexes. Alterations in the structure of these molecules or their aberrant expression may lead to disruption of cell-cell connections and result in epithelial tumor aggressiveness, invasion and metastasis [19, 20]. Protocadherins, as a subgroup within cadherin superfamily, thus play crucial roles in the development and progression of cancer. Therefore, investigating the expression and promoter methylation of protocadherin genes can be of immense clinical value. In the present study we examined methylation patterns of PCDH8, PCDH10 and PCDH17 genes with the aim of determining their role in HGSOC, the most common and one of the most aggressive type of epithelial OC. Although different studies have confirmed the significance of altered methylation of above mentioned PCDHs in other types of cancers [21, 22, 23, 24, 26, 28], to our best knowledge this is the first study to investigate methylation status of these PCDHs in OC. In our study, however, using preliminary NGS scan there was no significant methylation observed in analyzed regions of PCDH8 and PCDH10 genes. Therefore, we excluded these genes from further analyses and focused on PCDH17, where significant hypermethylation in part of analyzed region was observed. High methylation was present in over 60% of tumor samples with only minor methylation of one CpG in two control samples. Subsequent HRM analysis of 102 samples (51 tumors and 51 controls) confirmed hypermethylation detected by NGS. Methylation-positive pattern was observed in 66.7% (34/51) of OC tissue samples, whereas all of the control samples were methylation free.

Since methylation of CpG islands is usually associated with loss of gene expression [1, 25], we examined the relationship between detected methylation and mRNA expression of PCDH17. We found that PCDH17 expression was lower in tumors in comparison to control samples (fold change of $-$3.82; $p=$ 0.019). Methylated samples showed more distinct downregulation than unmethylated ones (median fold change $-$5.82 vs. $-$2.48; $p=$ 0.053). Correlation between PCDH17 expression in individual samples and their methylation patterns in 11 CpGs (region covered by HRM assay) detected by NGS revealed 5 CpGs with statistically significant negative correlation ( 0.01 or 0.05; see Fig. 3). The average methylation of whole analyzed region was also in correlation with the levels of expression ($p=$ 0.019), demonstrating suitability of selected region range for further HRM analysis.

HGSOC has a high mortality rate due to the lack of any specific symptoms and molecular markers of early stage disease. Since aberrant DNA methylation occurs early in cancer it provides great potential for an early stage OC biomarker [2]. To investigate the presence of PCDH17 methylation in the early stage of OC we divided patients into two groups. In the early stage tumors, methylation was detected in 50% of cases (6/12), in the late stage ones detected methylation increased to 71.8% (28/39) ($p=$ 0.293). Although difference between early and late stage tumors was not considered statistically significant (probably due to the small number of samples in the first group), the presence of hypermethylation in early stage tumors suggests its potential for further examination as a part of biomarker panel for early detection, especially if detected in plasma.

In subsequent analysis of follow-up data we investigated correlation between detected PCDH17 methylation and recurrence or overall survival. No significant correlation was observed, which reflects the fact that methylation was detected in majority of tumor tissue samples.

In conclusion, in present study we detected PCDH17 methylation in almost 70% of HGSOC patients without any PCDH17 methylation in the group of control samples. Subsequent gene expression analysis revealed decreased expression of PCDH17 and the correlation between methylation and downregulated expression in tumor tissue samples was observed suggesting potentially methylation-based silencing. Our findings indicate that methylation of PCDH17 could play an important role in development and progression of HGSOC and has potential to become a target in searching for new clinical biomarkers. However, further studies on larger groups of patients are needed to confirm our novel results.

Supplementary data

Supplementary Table 1.

Genomic coordinates of analyzed CpG sites

CpG	Genomic coordinates (assembly GRCh38/hg38)	Strand	CpG	Genomic coordinates (assembly GRCh38/hg38)	Strand
	PCDH8_1			PCDH10
1	chr13:52,849,218	$-$	3	chr4:133,149,261	$+$
2	chr13:52,849,148	$-$	4	chr4:133,149,269	$+$
3	chr13:52,849,128	$-$	5	chr4:133,149,278	$+$
4	chr13:52,849,050	$-$	6	chr4:133,149,282	$+$
5	chr13:52,848,998	$-$	7	chr4:133,149,286	$+$
6	chr13:52,848,966	$-$	8	chr4:133,149,443	$+$
7	chr13:52,848,960	$-$	9	chr4:133,149,501	$+$
8	chr13:52,848,944	$-$	10	chr4:133,149,537	$+$
9	chr13:52,848,942	$-$		PCDH17_1
10	chr13:52,848,937	$-$	1	chr13:57,631,897	$+$
11	chr13:52,848,923	$-$	2	chr13:57,631,902	$+$
12	chr13:52,848,903	$-$	3	chr13:57,631,904	$+$
	PCDH8_2		4	chr13:57,631,924	$+$
1	chr13:52,848,737	$-$	5	chr13:57,631,929	$+$
2	chr13:52,848,729	$-$	6	chr13:57,631,968	$+$
3	chr13:52,848,726	$-$	7	chr13:57,631,998	$+$
4	chr13:52,848,724	$-$	8	chr13:57,632,005	$+$
5	chr13:52,848,713	$-$	9	chr13:57,632,031	$+$
6	chr13:52,848,699	$-$	10	chr13:57,632,073	$+$
7	chr13:52,848,693	$-$	11	chr13:57,632,090	$+$
8	chr13:52,848,677	$-$	12	chr13:57,632,099	$+$
9	chr13:52,848,674	$-$	13	chr13:57,632,103	$+$
10	chr13:52,848,656	$-$	14	chr13:57,632,128	$+$
11	chr13:52,848,653	$-$	15	chr13:57,632,134	$+$
12	chr13:52,848,651	$-$	16	chr13:57,632,180	$+$
13	chr13:52,848,646	$-$	17	chr13:57,632,197	$+$
14	chr13:52,848,643	$-$	18	chr13:57,632,209	$+$
15	chr13:52,848,625	$-$	19	chr13:57,632,228	$+$
16	chr13:52,848,617	$-$	20	chr13:57,632,235	$+$
17	chr13:52,848,604	$-$	21	chr13:57,632,239	$+$
18	chr13:52,848,598	$-$	22	chr13:57,632,248	$+$
19	chr13:52,848,596	$-$		PCDH17_2
20	chr13:52,848,594	$-$	1	chr13:57,632,393	$+$
21	chr13:52,848,581	$-$	2	chr13:57,632,400	$+$
22	chr13:52,848,570	$-$	3	chr13:57,632,407	$+$
23	chr13:52,848,557	$-$	4	chr13:57,632,446	$+$
24	chr13:52,848,544	$-$	5	chr13:57,632,448	$+$
25	chr13:52,848,533	$-$	6	chr13:57,632,451	$+$
26	chr13:52,848,524	$-$	7	chr13:57,632,453	$+$
27	chr13:52,848,516	$-$	8	chr13:57,632,455	$+$
28	chr13:52,848,499	$-$	9	chr13:57,632,457	$+$
29	chr13:52,848,495	$-$	10	chr13:57,632,461	$+$
30	chr13:52,848,460	$-$	11	chr13:57,632,466	$+$
31	chr13:52,848,455	$-$	12	chr13:57,632,485	$+$
	PCDH10		13	chr13:57,632,487	$+$
1	chr4:133,149,234	$+$	14	chr13:57,632,514	$+$
2	chr4:133,149,256	$+$

Supplementary Table 2.

Genomic coordinates of CpG sites included in HRM analysis

CpG	Genomic coordinates (assembly GRCh38/hg38)	Strand
1	chr13:57,632,446	$+$
2	chr13:57,632,448	$+$
3	chr13:57,632,451	$+$
4	chr13:57,632,453	$+$
5	chr13:57,632,455	$+$
6	chr13:57,632,457	$+$
7	chr13:57,632,461	$+$
8	chr13:57,632,466	$+$
9	chr13:57,632,485	$+$
10	chr13:57,632,487	$+$
11	chr13:57,632,514	$+$