Methylation of tumour suppressor genes associated with thyroid cancer

Abstract

BACKGROUND

: Thyroid carcinoma is the most common endocrine malignancy worldwide. Changes in DNA methylation can cause silencing of normally active genes, especially tumour suppressor genes (TSG) or activation of normally silent genes.

OBJECTIVE

: The aim of this study is to evaluate the degree of promoter methylation for a panel of markers for thyroid neoplasms and to establish their relationship with thyroid oncogenesis.

METHODS

: To generate a comprehensive DNA methylation signature of TSGs involved in thyroid neoplasia, we use Human TSG EpiTect Methyl II Signature PCR Array-Qiagen for 24 samples (follicular adenomas and papillary thyroid carcinomas) compared with normal thyroid tissue. We extended the evaluation for three TSGs (TP73, WIF1, PDLIM4) using qMS-PCR. Statistical analysis was performed with GraphPad Prism.

RESULTS

: We noted four important genes NEUROG1, ESR1, RUNX3, MLH1, which presented methylated promoter in tumour samples compared to normal. We found new characteristic of thyroid tumours: methylation of TP73, WIF1 and PDLIM4 TSGs, which can contribute to thyroid neoplasia. A significant correlation between BRAF V600E mutation and RET/PTC rearrangements with TIMP3 and CDH13, RARB methylation, respectively was observed.

CONCLUSIONS

: TSGs promoter hypermethylation is a hallmark of cancer and a test that uses methylation quantification method is suitable for diagnosis and prognosis of thyroid cancer.

1. Introduction

Thyroid cancer, the most common malignant endocrine disease (more than 95% cases worldwide) [1], that can affect both genders, although it is most common in women, three to four times higher than in men. Though thyroid cancer can occur at any age, most tumours are diagnosed after the age of sixty [2].

Thyroid neoplasia represents a highly heterogeneous entity that possesses different cytological, histological and genetic characteristics. Starting with benign lesions – follicular adenoma (FA), to the well differentiated carcinomas, such as papillary thyroid carcinoma (PTC) in 85–90% of thyroid cancer cases, or follicular thyroid carcinoma (FTC) in 5–10% of all thyroid cancer cases, to poorly/undifferentiated thyroid carcinoma and anaplastic thyroid carcinoma (ATC) in 2% of thyroid cancers, or medullary carcinoma (MTC) in about 5% of thyroid cases, most primary thyroid cancers are originated from thyroid follicular cells [2].

The most prevalent type of thyroid neoplasia is papillary thyroid cancer (PTC). The survival rate is higher compared with other cancers. Prognosis for nodules less than 20 mm in diameter is excellent, but a clinically significant thyroid nodule may develop in 5% to 10% of the general population during their lifetime. 99% of patients with nodules smaller than 20 mm will be alive for 20 years, in contrast with FTC (10–20% of will be alive for 10 years), MTC (25% of patients with will be alive for 10 years) and ATC (90% of patients with will be alive for 5 years) [3].

Thyroid cancer is a multifactorial disease, the patients presenting a genetic susceptibility, but more evidences showed the involvement of epigenetic deregulated mechanisms in thyroid oncogenesis [4]. The well-known point mutations and rearrangements found in thyroid cancer tend to affect the effector genes of the MAPK pathway, such as v-raf murine sarcoma viral oncogene homolog B1 (BRAF), the RAS family and the protooncogene RET. These alterations have been shown to be subtype specific. Almost all tumours presenting RAS mutations have a follicular pattern of growth (FA, FTC or PTC follicular variant [fvPTC]). Mutations in BRAF gene and rearrangements in the RET gene are characteristic of PTC and poor prognosis [5, 6, 7]. RAS mutations have diagnostic but not prognostic value.

DNA methylation is one of the most important epigenetic modifications along with histone modifications and silencing mediated by RNA interference [8]. The epigenetic mechanisms are deregulated in cancer and the hallmark for cancer cells is represented by hypermethylation of tumour-suppressor gene and hypomethylation of repetitive DNA sequences and/or oncogenes. DNA methylation is important for transcriptional regulation, being essential for the normal cellular development. In chromosomal regions of tumour-associated genes, the epigenetic modifications can change important associated regulatory mechanisms for pathogenic malignant transformation. With DNA methylation, a methyl group is added to the fifth carbon of the cytosine residue in a CpG dinucleotide. CpG dinucleotides’ rich regions (CpG islands) are usually located in the 50-flanking gene promoter areas. Gene promoter methylation, particularly near the transcription start site, is associated with chromatin remodelling, which affects gene silencing [9, 10].

Tumour suppressor gene inactivation by promoter hypermethylation is thought to be important in carcinogenesis. Thus, measuring DNA methylation in a genome-wide manner would be valuable for studying mechanisms of epigenetic control involved in gene expression regulation. For personalized therapy, molecularly-defined tumour subgroup identification appears promising. For example, gene expression panels associated with breast cancer are now clinically used to provide individualized therapy. These panels, which depict carcinogen exposure-associated differences in individual tumours, include DNA hypermethylation involved in gene silencing as well as DNA hypomethylation, which is associated with oncogene activation and genomic instability [11, 12].

It has been observed that methylation of the thyroid cancer genes, especially methylation of tumour suppressor genes, can be correlated with clinical parameters, risk factors and the prognosis of the disease. Despite the efforts, this mechanism is not completely known.

Recent studies focused on the global methylation pattern in thyroid cancer. Ceolin et al. proved differences between the global MTC and PTC methylation level, therefore, the overall methylation profile of DNA can be correlated with thyroid cancer histology [13].

Lee et al. reported decreased levels of MAPK signal-inhibiting gene SERPINA5 in papillary thyroid cancer, due to DNA methylation [14, 15].

The aim of this study is to evaluate the degree of promoter methylation for a panel of novel and known methylation markers for thyroid neoplasms and to establish their relationship with thyroid oncogenesis.

2. Materials and methods

Sample collection. Biopsy samples ( $n=$ 114), tumour and adjacent tissue were obtained from 57 patients who, undergo surgical resection. The samples originated from patients with thyroid neoplasms: thyroid adenomas (age range: 37–68 years median: 55), papillary thyroid cancer (age range: 23–79 years median: 47.5) and papillary thyroid cancer follicular variant (age range: 17–63 years median: 34). For nucleic acids preservation all sample were collected in a sterile tube with RNA later (Thermo Fisher Scientific, Massachusetts, USA). Tumours were immediately stored at $-$ 80 ${}^{\circ }$ C. Blood samples from patients diagnosed with thyroid neoplasms ( $n=$ 57) and from 20 subjects without thyroid case history (as controls) were collected in EDTA tubes. Informed consent was obtained from each subject prior to the study enrolment.

All the samples were provided with patient’s agreement and the study protocol was approved by the Ethic Committee.

DNA isolation. DNA was isolated from thyroid tissue samples and blood using High Pure PCR Template Preparation Kit (Roche Molecular Biochemicals, Mannheim, Germany). DNA samples were stored at $-$ 20 ${}^{\circ }$ . The DNA concentration and purity of all samples were evaluated with NanoDrop spectrophotometer (NanoDropTechnologies, Montchanin, DE, USA).

BRAF V600E mutation was detected using PCR-RFLP analysis. PCR primer sequences used were: forward primer: 5’-GCT TGC TCT GAT AGG AAA ATG AG-3’; reverse primer: 5’-GAT AGA CAG CAG CAT CTC AGG-3’. PCR reaction mix contains: 100 ng DNA, 1.5 U/ $\mu$ l Taq polymerase, 10 pmol/ $\mu$ l primer, 0.2 mM dNTP, 1.5 mM MgCl ${}_2$ , buffer up to 20 $\mu$ l reaction volume (Promega, Madison, WI, USA). Amplification protocol: 5 min 95 ${}^{\circ }$ C, followed by 40 X (95 ${}^{\circ }$ C, 30 sec, 58 ${}^{\circ }$ C, 30 sec, 72 ${}^{\circ }$ C 30 sec) and final elongation step 72 ${}^{\circ }$ C 30 sec. Enzyme restriction: 20U TspRI in 20 $\mu$ l reaction volume. Incubation for 16 h, 65 ${}^{\circ }$ C. Identification of restriction fragments: electrophoresis in agarose gel 2%, TBE 1x, using DNA ladder 50 bp DNA step Ladder (Promega). Restriction products: wildtype 3 fragments – 117 bp, 87 bp, 33 b; mutant homozygote, 2 fragments – 204 bp, 33 bp, mutant heteozygote, 4 fragments – 204 bp, 117 bp, 87 bp, 33 bp.

RET split detection was performed using FISH technique. The protocol is specific for formalin-fixed paraffin-embedded (FFPE) tissue sections and consists of successive stages of dewaxing, rehydration, deproteinization and pretreatment followed by in situ hybridization with cDNA probes double labelled with fluorochromes (Zytovision GmbH – Germany). The two ends of the gene were labelled with different fluorochromes, green and orange, respectively. If the gene is intact, one see one single yellow spot. If the gene is splitted there were two distinct green and orange spots on the image. 3 $\mu$ m sections displayed on the slide were fixed by exposure for 16 hours at 50 ${}^{\circ }$ C on histological thermostatic plate, then gradually dewaxed using xylene. After dewaxing the sections were rehydrated by successive baths of 100%, 90%, 70% ethanol and deionized water, respectively. The next step was an enzymatic deproteinization at 98 ${}^{\circ }$ C for the nuclear membrane permeabilization, then hybridization with cDNA double labelled probes, overnight (16–18 hours). After hybridization, the samples were washed and counterstained with DAPI.

RNA isolation. Isolation of total RNA was performed using Trizol technology (Invitrogen) according to manufacturer’s instructions. 50 mg of tissue was disrupted using a homogenizer and K1 cells were deposited after centrifugation and 1 ml Trizol reagent was added. Isolated RNA quality check was done by reading the concentration and purity NanoDrop spectrophotometer and Experion.

Human tumour suppressor genes DNA methylation PCR array, signature panel

The Human Tumour Suppressor Genes EpiTect Methyl II Signature PCR Array profiles the methylation status of 22 tumour suppressor gene promoters whose hypermethylation has been reported in the literature to occur frequently in a variety of cancers and tumours. Profiling tumour samples with these arrays may help correlate CpG island methylation status with biological phenotypes. The results may provide insights into the molecular mechanisms and biological pathways behind oncogenesis, as well as cancer pathology. The protocol is based on restriction enzyme digestion, followed by real-time PCR, for the analyse of the 22 different tumour suppressor genes methylation status.

Apoptosis: BRCA1, CDKN2A, DAPK1, GSTP1, MGMT (AGT), PTEN, RUNX3, TIMP3, TP73, VHL.

Cell adhesion: APC, CDH1, CDH13, CDKN2A.

Cell cycle: APC, BRCA1, CDKN2A, FHIT, MLH1, NEUROG1, PTEN, RASSF1, RUNX3, TP73, VHL.

DNA damage and repair: APC, BRCA1, MGMT (AGT), MLH1, TP73.

Signal transduction: APC, BRCA1, CDH13, ESR1, PTEN, RASSF1, SOCS1, WIF1.

Transcription factors: BRCA1, CDH1, ESR1, NEUROG1, RARB, RUNX3, TP73, VHL.

Other: PDLIM4 (RIL).

Unmethylated C residues conversion was realized with bisulphite treatment using EpiTect Bisulfite kit (Qiagen, Valencia, California, USA). 700 ng/ $\mu$ l of each isolated DNAs (tumours and normal adjacent tissue) were bisulphite treated along with positive (CpGenome Universal Methylated DNA) and negative (CpGenome Universal Unmethylated DNA) controls (Millipore, Billerica, MA, USA).

Table 1

Primers sequence used in qMS-PCR

Gene name	Gene ID	Primers		Genome location
PDLIM-4	NC_000005.10	M-F5’-TTTTAAAAGCGAT TTTAAAGGGTC-3	U-F5’-TTTAAAAGTGATT TTAAAGGGTTGG-3’	chr5 $+$ : 132256233-132257725;
		M-R5’-TTCGAAAAACTC TAAAAACCCG-3’	U-R 5’-CACTTCAAAAAACT CTAAAAACCCA-3’	TSS $=$ 132257658;
TP73	NC_000001.11	M-F 5’-CGCGTAGTAGGT TGTAATAGTCGA-3’	U-F 5’-TGGGTGTGTAGT AGGTTGTAATAGTTG-3’	chr1 $+$ : 3648800-3650601;
		M-R 5’-AAATCCCTAATC GAAAAAAACGTA-3’	U-R 5’-AAAATCCCTAAT CAAAAAAAACATA-3’	TSS $=$ 3650000;
WIF1	NC_000012.12	5’-CGAGTTAAGGTT AGCGTTTGTTATC-3’	5’-TTGAGTTAAGGTTAG TGTTTGTTATTG-3’	chr12-: 65122903-65121328;
		5’-AAAATCCCTAAT CAAAAAAAACATA -3’	5’-TAATAACTCCTAT TCCTCCTCCAAC-3’	TSS $=$ 65121566

Primers In this purpose the promoter sequence was retrieved from EMBL database (https://www.ebi.ac.uk/ research/publications/embl-nucleotide-sequence-data base-1). The methylation primers were selected using MethPrimer (http://www.urogene.org/methprimer/) which use an CpG islands prediction algorithm. By default a CpG island is defined as a 200 bp DNA sequence with a GC content $>$ 50% (Li and Dahiya, 2002). The primers distinguish between methylated and unmethylated DNA after bisulphite treatment of the template (Table 1) and were synthesized by Invitrogen Corporation (Carlsbad, California, USA). On the other hand, the primers are usually placed in the vicinity of the TSS, because the methylation in the region around the TSS (Transcription Start Site) must have a profound effect on the gene transcription [16].

WIF1 gene promoter presents 3 CpG islands (199 bp, 255 bp, 357 bp) and the transcription is on the minus strand. The TSS is situated at $-$ 238 nt, the primers being located around $-$ 533 nt, a critical region for transcriptional silencing according to Licchesi et al. [17]. PDLIM4 gene promoter presented one CpG island of 314 bp, the primers were situated upstream of TSS $+$ 294 nt, this region being found hypermethylated in colon cancer [18]. TP73 presents 5 CpG islands (357 bp, 382 bp, 176 bp, 560 bp, 822 bp), the primers were selected downstream of TSS around-800nt, the region found also hypermethylated in chondrosarcoma [19].

Direct Q-MSP of Genomic DNA is a quantitative method for evaluating the degree of sample methylation.

Standard curves were designed using positive controls (fully methylated) and negative controls (fully unmethylated). Positive and negative controls were diluted in order to obtain a serial dilution: 10 pg, 100 pg, 1 ng and 10 ng according to Absolute Quantitation Using Standard Curve Getting Started Guide (Applied Biosystems 7300/7500/7500 Fast Real-Time PCR System, Foster City, CA, USA). All DNA samples were diluted to a final concentration of 10 ng/ $\mu$ l. 25 $\mu$ l FastStart Universal SYBR Green Master (ROX) (Roche Molecular Biochemicals, Mannheim, Germany) 0.3 $\mu$ M primers and 5 $\mu$ l DNA were used. Direct Q-MSP was performed in 50 $\mu$ l final volume and the parameters we used were: 95 ${}^{\circ }$ C–10 min (1 cycle), and 40 cycles at 95 ${}^{\circ }$ C for 15 sec and 60 sec at the specific annealing temperature.

qPCR experiments were performed in duplicate ( $\mathrm{\Delta }$ Ct $\approx$ 1.2 between replicates) and mean values were used for calculations. In order to establish the quality of DNA samples we used ACTB gene as reference [20]. All umethylated samples for methylation presented a Ct for ACTB below 35, excluding false negative results.

Methylation percentage calculation method was done according to Fackler et al. [20]. The concentration of unmethylated and methylated DNA for each patient sample was extrapolated using the standard curves. The percentage of methylation was calculated according to the formula: % M $=$ 100 $\times$ [ng methylated gene A/(ng methylated gene A $+$ unmethylated gene A)], where total target gene DNA was taken as the sum of U $+$ M.

Cell culture treatment. In order to evaluate the role of methylation in gene expression suppression, K1 cell line culture (derived from a papillary carcinoma primary thyroid) was cultured for 24 hours in DMEM-Ham’s F12 medium supplemented with 2 mM glutamine and 10% foetal calf serum (FCS) to allow cell adherence to a minimum of 50%. Then the cells were treated with 5-azacytidine (AzaC) a DNA demethylating agent in a concentration of 5 $\mu$ M and 10 $\mu$ M for 24 h and 48 h. This treatment can free CpG islands promoters from methylation [21].

Table 2

Summary of the main clinical and pathological characteristics of samples used in this study

Gender	M	F
number	14	43
Age (years)
Median	54.5	51.5
Min-max	29–74	31–67
Histology (grade)
PTC	8	20
fvPTC	3	7
FA	3	16
Tumour size (cm)
Median	2.8	3.35
Min-max	1.1–6.5	0.1–13.4
Stage
T1a/b	1	4
T2	4	5
T3	5	16
T4b	1	2
Braf	1/10	5/18
Ret/PTC1	5/10	16/18

Apoptotic process evaluation by flow cytometry. Changes that occur in the plasma membranes are part of the early events in apoptosis. Apoptotic cells present phosphatidylserine translocated into the inner face of the plasma membrane allowing external annexin family members (proteins that bind calcium dependent phospholipids) to attach. Following the characteristic changes occurring during the apoptotic process, the test used allows a double staining: annexin V labelled with FITC (fluorescein isothiocyanate) which binds to phosphatidylserine exposed to the extracellular environment and PI (propidium iodide) – a dye used for the evaluation viability is tied intracellular DNA. For staining of cells an apoptosis kit was used (BD Bioscience, NJ, USA). More than 1 $\times$ 10 ${}^5$ cells suspended in 100 $\mu$ l binding buffer were added 5 $\mu$ l of Annexin V/FITC and 5 $\mu$ l PI solution, incubating the mixture for 15 minutes in the dark. In each tube were added 400 $\mu$ L of Annexin V binding buffer and the sample was collected using FACSCantoII flow cytometer. Data was analysed using Diva 6.2 software (Becton Dickinson) and shown as two-colour dot plot with FITC-Annexin V (green fluorescence, $X$ axis) vs. PI (red fluorescence, $Y$ axis) [22].

Table 3

Tumour suppressor methylation percentage evaluation in tumour samples. The results were presented in descending order, according to the AUROC values

Gene symbol	Median methylated DNA percentage in samples/ range %	Area under the ROC curve	Cut-off %	Sensitivity %/ specificity %	$p$ -value
TP73	36.35 (22.60–100)	1	21.23	100/100	$<$ 0.0001
WIF1	60.04 (55.67–66.20)	1	32.08	100/95.83	$<$ 0.0001
PDLIM4	31.51 (15.20–42.58)	0.9844	20.93	95.83/100	$<$ 0.0001
NEUROG1	30.02 (14.57–60.32)	0.9826	19.18	87.5/100	$<$ 0.0001
RARB	39.83 (12.24–50.21)	0.9826	18.99	83.33/100	$<$ 0.0001
ESR1	26.58 (16.58–41.07)	0.9757	20.89	91.67/100	$<$ 0.0001
RUNX3	35.60 (14.3–30.06)	0.9722	21.07	95.83/100	$<$ 0.0001
MLH1	25.39 (15.94–31.83)	0.9601	23.6	75/100	$<$ 0.0001
RASSF1	30.75 (14.35–51.93)	0.9375	22.23	83.33/100	$<$ 0.0001
APC	24.72 (17.53–29.59)	0.9236	22.01	79.19/100	$<$ 0.0001
PTEN	31.26 (14.12–55.84)	0.9167	28.79	66.67/95.83	0.0001
FHIT	23.69 (11.09–29.52)	0.8681	20.72	62.5/100	0.0001
DAPK1	21.32 (11.54–46.26)	0.8628	20.27	75/91.67	0.0002
MGMT	31.22 (12.30–50.46)	0.8611	27.8	62.5/100	0.0001
CDH13	29.71 (11.77–40.33)	0.842	29.12	54.17/95.83	0.0001
CDKN2A	28.30 (13.11–39.99)	0.829	27.25	58.33/91.67	$<$ 0.0001
BRCA1	27.33 (11.06–39.34)	0.8212	27.56	50/100	$<$ 0.0001
VHL	21.18 (9.13–28.57)	0.809	20.71	54.17/100	0.0002
SOCS1	26.02 (10.18–62.27)	0.7969	21.79	66.67/83.33	0.0001
GSTP1	16.84 (4.13–25.32)	0.7951	13.02	75/79.17	0.0002
TIMP3	25.07 (6.69–54.63)	0.7934	24.38	58.33/91.67	0.0002
CDH1	15.29 (4.89–25.63)	0.7222	15.1	54.17/79.17	0.003

Table 4

Standard curve parameters for all tested PCR primers (for methylated-M and unmethylated-U DNA sequences)

	WIF1-M	WIF1-U	TP73-M	TP73-U	PDLIM4-M	PDLIM4-U
Slope	$-$ 3.432	$-$ 3.599	$-$ 3.386	$-$ 3.312	$-$ 3.377	$-$ 3.176
R ${}^2$	0.991	0.998	0.990	0.991	0.990	0.994
Efficiency	0.956	0.900	0.974	1.004	0.977	1.065

3. Results

3.1 Patients clinical characteristic feature

We examined a cohort of 57 patients: 19 benign thyroid lesions (follicular adenomas) and 38 thyroid cancers (28 papillary-PTC and 10 follicular variant of papillary thyroid cancer-fvPTC). The clinicopathological characteristics and genetic tests are detailed in Table 2. The patients were grouped according to gender and the number of female patients was higher than the male. 75% from patients with thyroid tumours were positive for RET/PTC1 rearrangements, whereas 21.43% were positive for BRAF (V600E).

3.2 The human tumour suppressor genes epiTect methyl II signature PCR array

We examined 22 genes with diverse function, including cell cycle regulation, tumour suppression and DNA repair in thyroid tissues. We observed that the gene promoter methylation for each tested gene is higher in thyroid cancer tissue compared with normal adjacent tissue. The methylation for each evaluated gene is listed in Table 3. To assess the potential utility of hypermethylation of genes as molecular markers of thyroid cancer, we used ROC analysis. We evaluated the accuracy of the test measured by the area under the ROC (AUROC) curve and sensitivity/specificity. An area of 1 represents a perfect test; an area of 0.5 represents a worthless test. A rough guide for classifying the accuracy of a diagnostic test is the traditional academic point system:

•

0.90–1 $=$ excellent

•

0.80–0.90 $=$ good

•

0.70–0.80 $=$ fair

•

0.60–0.70 $=$ poor

•

0.50–0.60 $=$ fail

We found that the methylation percentage was significantly higher for TP73, WIF1 and PDLIM4 gene promoter methylation. These three genes presented also the highest values for sensitivity and specificity. Good results were obtained for a panel of 8 genes (NEUROG1, RARB, ESR1, RUNX3, MLH1, RASSF1, APC, PTEN), which correlates with thyroid cancer.

The sample positive for BRAFV600E mutation presented higher methylation values for MGMT, NEUROG1, RARB gene promoters and significant methylation values for TIMP3 (Braf ( $+$ ) median $=$ 28.03; Braf ( $-$ ) median $=$ 17.54; $p=$ 0.05).

The patients positive for RET/PTC traslocation presented higher methylation values for ESR1, NEUROG1, SOCS1 gene promoters and significant methylation values for RARB (RET/PTC ( $+$ ) median $=$ 41.03; RET/PTC ( $-$ ) median $=$ 25.64; $p=$ 0.0091) and for CDH13 (RET/PTC ( $+$ ) median $=$ 36.14; RET/PTC ( $-$ ) median $=$ 26.11; $p=$ 0.0271).

This array was realized in 24 patients on DNAs extracted from biopsy tissue (tumour vs normal). Further we evaluated the potential as biomarkers of the most important factors discovered through this type of analysis. For this purpose, in an extended patient group, we evaluated the role of WIF1, TP73 and PDLIM4 tumour suppressor genes in thyroid oncogenesis.

3.3 Methylation status of WIF1, TP73 and PDLIM4 gene promoters and gene expression levels in thyroid tumours samples

Validation of direct qMSP was performed for each primer pair specific amplification for methylated and unmethylated DNA in the qPCR reaction, according to our previous publications [21, 23]. The $\mathrm{\Delta }$ Cq $=$ CqM $-$ CqU formula was calculated as a function of sample dilution over the range specified for the specific standards. The differences between $\mathrm{\Delta }$ Cq values were similar for all dilutions ( $\mathrm{\Delta }$ CqWIF1 $\approx$ 0.542; $\mathrm{\Delta }$ CqTP73 $\approx$ 0.875, $\mathrm{\Delta }$ CqPDLIM4 $\approx$ 0.476). The slope values for all the standard curves were calculated along with efficiency Table 4.

Table 5

Statistical parameters of promoter methylation status for investigated genes in patients’ samples (control vs. tumoral)

Tissue type	Control	Tumour	Control	Tumour	Control	Tumour
Gene methylation	WIF1		TP73		PDLIM4
Median $\pm$ SE (range)	21.90 $\pm$ 1.068 0.007–46.20	60.56 $\pm$ 0.9128 38.22–87.08	25.70 $\pm$ 1.060 0.0002–39.18	41.96 $\pm$ 3.040 22.60–100.0	14.83 $\pm$ 0.8071 0.001–25.94	53.83 $\pm$ 3.428 0.00731–97.94
$p$ -value*	$<$ 0.0001		$<$ 0.0001		$<$ 0.0001
Gene expression**	WIF1		TP73		PDLIM4
Median $\pm$ SE (range)	$-$ 3.032 $\pm$ 0.172 $-$ 4.836–1.68	$-$ 6.102 $\pm$ 0.200 $-$ 7.23–( $-$ 0.78)	$-$ 3.970 $\pm$ 0.0107 $-$ 4.912–( $-$ 1.12)	$-$ 4.363 $\pm$ 0.126 $-$ 6.09–( $-$ 2.28)	$-$ 2.755 $\pm$ 0.137 $-$ 4.099–( $-$ 0.547)	$-$ 3.652 $\pm$ 0.1465 $-$ 5.99–( $-$ 1.32)
$p$ -value*	$<$ 0.0001		0.0010		$<$ 0.0001

${}^{*}$ for comparison $t$ -test was used; **relative gene expression method was used.

The slope values were between [ $-$ 3.176–( $-$ 3.599)], indicating a 2-fold increase in PCR product per cycle during the linear phase of real-time PCR. The average of the calculated efficiencies was between 0.9–1.065, taking into account that the optimal efficiency is between 0.8–1. R ${}^2$ values show evidence of linearity over the entire range of template concentrations.

The tumour samples presented significant higher methylation status than control samples ( $p<$ 0.0001). All the studied genes presented the methylation percentage significantly increased in tumour tissue ( $p<$ 0.001) Table 5. The WIF1 gene was found to present a hypermethylated promoter, whereas PDLIM4 presented lower methylation levels.

The mRNA expression levels of WIF1, TP73 and PDLIM4 were measured in samples from thyroid carcinomas and adjacent normal thyroid tissues. The relative PDLIM4 and WIF1 mRNA expression was significantly lower ( $p<$ 0.0001) in thyroid carcinomas. Similarly, the relative TP73 ( $p=$ 0.0010) mRNA expression was lower in thyroid carcinomas than in the paired normal thyroid tissues.

Spearman correlation was performed to establish the influence of gene promoter methylation on gene expression. The methylation was inversely correlated with gene expression (WIF1- $r^2=-$ 0.3722, $p=$ 0.0118; TP73- $r^2=-$ 0.3224, $p=$ 0.05; PDLIM4- $r^2=-$ 0.3705, $p=$ 0.0133).

Figure 1.

The analysis of 5-azacytidine (AzaC) treatment effect upon K1 cells (concentration of 5 $\mu$ M and 10 $\mu$ M for 24 h/48 h). Dot plots representation (A) shown the early stage of apoptosis (Q4 – the cells bind annexin V while still excluding PI) and the late stage of apoptosis (Q2) the cells bind annexin V-FITC and stain brightly with PI, the necrotic cells (Q1) and the live cells (Q3).

Further, we investigated the targeted gene expression and methylation promoter levels depending of the histopathological analysis.

As seen in Table 6, higher TP73 gene promoter methylation levels are characteristic for papillary carcinoma and papillary carcinoma follicular variant ( $p=$ 0.00146). Gene promoter methylation of PDLIM4 seems to be an event that characterizes the debut of the neoplastic process, being found in follicular adenoma and maintaining a high rate in papillary carcinoma and papillary carcinoma follicular variant ( $p=$ 0.0293). The level of WIF1 gene promoter methylation neoplasia is high in both benign and malignant tumour types ( $p=$ 0.0185).

Regarding the expression of the studied genes and histopathological type statistically significant difference was observed for WIF1 gene and PDLIM4. The expression PDLIM4 and TP73 genes decrease gradually from benign neoplasia to malign status. In case of WIF1 the lowest value of gene expression was found in PTC.

3.4 Methylation status of WIF1, TP73 and PDLIM4 gene promoters and gene expression level evaluation in blood samples of the studied group of patients

Furthermore, we decided to estimate WIF1, TP73 and PDLIM4 gene promoter’s methylation as well as their expression levels in peripheral blood samples. Results revealed a high percentage of methylation for both TP73 and PDLIM4 genes in thyroid carcinoma patients, both with statistically significant difference between groups ( $p=$ 0.0148, $p=$ 0.0094 respective). In the control group relative low levels of methylation were observed for WIF1 (median $=$ 2.85), TP73 (median $=$ 1.50) and PDLIM4 (median $=$ 2.54).

Regarding gene expression, it was observed that patients with thyroid carcinoma (PTC and fvPTC) exhibit reduced relative expression levels for all tested genes (Table 7).

Table 6

Statistical parameters for investigated genes promoter methylation grade and n-fold genes expression of in various neoplasms for studied patients’ group

	Follicular adenoma	Papillary carcinoma	Papillary carcinoma follicular variant	Follicular adenoma	Papillary carcinoma	Papillary carcinoma follicular variant	Follicular adenoma	Papillary carcinoma	Papillary carcinoma follicular variant
Gene methylation	WIF1			TP73			PDLIM4
Median $\pm$ SE (range)	59.63 $\pm$ 2.83 38.22–65.53	59.67 $\pm$ 3.894 50.35–79.57	62.74 $\pm$ 1.796 58.27–84.70	33.96 $\pm$ 1.96 22.6–40.36	38.27 $\pm$ 10.60 33.3–100	41.75 $\pm$ 5.54 34.83–100.0	36.92 $\pm$ 7.02 0.007–75.54	61.68 $\pm$ 5.24 24.48–96.29	67.86 $\pm$ 5.78 26.95–97.94
$p$ -value	0.0185			0.0146			0.0293
Gene expression	WIF1			TP73			PDLIM4
Median $\pm$ SE (range)	$-$ 2.21 $\pm$ 0.18 $-$ 3.27–( $-$ 0.78)	$-$ 3.03 $\pm$ 0.29 $-$ 4.77–( $-$ 1.42)	$-$ 3.59 $\pm$ 0.53 $-$ 7.23–( $-$ 1.57)	$-$ 1.46 $\pm$ 0.29 $-$ 2.75–0.003	$-$ 1.63 $\pm$ 0.17 $-$ 3.312–( $-$ 0.79)	$-$ 1.87 $\pm$ 0.16 $-$ 6.09–( $-$ 0.76)	$-$ 2.31 $\pm$ 0281 $-$ 3.46–0.75	$-$ 2.93 $\pm$ 0.24 $-$ 4.21–( $-$ 1.15)	$-$ 3.66 $\pm$ 0.43 $-$ 5.99–( $-$ 2.05)
$p$ -value	0.0037			ns			0.0049

Table 7

Statistical parameters for methylation grade and n-fold genes expression of selected genes promoters in blood samples of studied patients’ group

	Follicular adenoma	Papillary carcinoma	Papillary carcinoma follicular variant	Follicular adenoma	Papillary carcinoma	Papillary carcinoma follicular variant	Follicular adenoma	Papillary carcinoma	Papillary carcinoma follicular variant
Gene methylation	WIF1			TP73			PDLIM4
Median $\pm$ SE (range)	32.74 $\pm$ 3.147 0.657–40.64	36.62 $\pm$ 5.304 26.16–81.30	36.76 $\pm$ 4.848 27.91–80.32	1.473 $\pm$ 3.519 0.0–24.72	22.02 $\pm$ 6.388 4.328–77.83	25.93 $\pm$ 2.311 13.35–35.50	26.13 $\pm$ 1.95 15.20–39.78	28.48 $\pm$ 2.79 11.64–33.19	36.45 $\pm$ 2.35 12.07–42.58
$p$ -value	ns.			0.0148			0.0094
Gene expression	WIF1			TP73			PDLIM4
Median $\pm$ SE (range)	1.492 $\pm$ 0.461 $-$ 1.507–3.010	$-$ 0.295 $\pm$ 0.356 $-$ 2.946–1.639	$-$ 0.591 $\pm$ 0.229 $-$ 2.134–1.407	0.28 $\pm$ 0.28 $-$ 0.33–2.35	0.05 $\pm$ 0.40 $-$ 2.19–0.94	$-$ 0.22 $\pm$ 0.213 $-$ 2.22–0.95	0.090 $\pm$ 0.185 $-$ 1.055–1.349	0.2878 $\pm$ 0.391 $-$ 2.737–2.483	$-$ 0.441 $\pm$ 0.242 $-$ 1.993–1.026
$p$ -value	ns.			ns.			ns.

Table 8

The analysis results of 5-azacytidine (AzaC) treatment effect upon K1 cells

Time	24h				48h				Total apoptosis
Treatment	Q1	Q2	Q3	Q4	Q1	Q2	Q3	Q4	Q2 $+$ Q4
									24 h	48 h
Control	1.2%	7.1%	87.8%	3.9%	3.1%	8.2%	82%	6.7%	11.0%	14.9%
Aza 5 $\mu$ M	1.5%	8.3%	80.4%	9.8%	0.5%	10.3%	73%	16.2%	18.1%	26.5%
Aza 10 $\mu$ M	1%	11.6%	76.2%	11.2%	1%	11.8%	64.6%	22.6%	22.8%	34.4%
Camptothecin	3.7%	14.2%	67.7%	14.4%	3.3%	15.8%	62.0%	18.9%	28.6%	34.7%

3.5 Effect of 5-azacytidine (AzaC) treatment on the K1 cells apoptotic process

Standard K1 cells (derived from a papillary carcinoma primary thyroid) were cultured for 24 hours in DMEM-Ham’s F12 culture medium supplemented with 2 mM glutamine and 10% foetal calf serum (FBS) to allow cell adherence to a minimum of 50%. The cells were treated with 5-azacytidine (AzaC) that induces DNA demethylation in a concentration of 5 $\mu$ M and 10 $\mu$ M for 24 h and 48 h. This treatment can demethylate CpG islands from promoters.

3.5.1 Apoptotic process evaluation by flow cytometry

The untreated cell population was used as control to define the basal level of apoptotic and dead cells, while the cells incubated for 3 hr at 37 ${}^{\circ }$ C with an inducer of apoptosis 10 $\mu$ M Camptothecin was used as positive control.

As shown in Fig. 1 the treatment with 5-azacytidine of K1 tumour cell line reduces the cell viability (Q3) in a time and concentration depending manner.

The cell viability is reduced as a response to 5-azacytidine treatment correlated with an increase in K1 cells apoptotic process Table 8.

As shown in the Fig. 1, the treatment of K1 tumour cell line with 5-azacytidine determines an increase of both the early and late apoptosis, also depending on time and concentration.

5-azacytidine induces an increase of early apoptosis 2.5 times (9.8%) compared with untreated cells (3.9%), after 24 hours of exposure at 5 $\mu$ M 5-azacy- tidine; the increment is correspondingly maintained also at 48 h. Moreover, the effect on early apoptosis is even amplified in case of a concentration of 10 $\mu$ M 5-azacytidine (2.9 times-11.2%/24 h; 3.4 times-22.6%/48 h).

The data showed that 5-azacytidine, a DNA deme- thylating agent enhanced the apoptotic process of K1 cell line, suggesting a possible link between the DNA

Figure 2.

Effect of 5-azacytidine treatment on the target genes methylated promoters.

methylation status and the activation of apoptotic pathways.

3.5.2 Evaluation of WIF1, TP73 and PDLIM4 methylation status and gene expression levels in 5-azacytidine K1 treated cell line

Further, we evaluated the effect of 5-azacytidine upon the promoter of target genes. As shown in the Fig. 2 the promoter methylation percentage decreases considerably for all three genes. The promoter is demethylated especially after a 5 $\mu$ M 5-azacytine treatment for 48 h in all cases.

Consecutive, the gene expression increases significantly, the promoter becoming active. The highest increment was observed for TP73 gene ( $\sim$ 61.5-fold/5 $\mu$ M 48 h), followed by PDLIM4 gene ( $\sim$ 42-fold/5 $\mu$ M 48 h) and WIF1 gene ( $\sim$ 19.5-fold/5 $\mu$ M 48 h).

4. Discussion

Along with various molecular alterations found in thyroid neoplasia (mutations and/or gene rearrangements) that were described (RET germline mutation, and chromosomal rearrangements resulting in the fusion gene PAX8/PPAR $\gamma$ [24], epigenetic modifications are also common in thyroid tumours. Deciphering the genetic and epigenetic alterations that are involved in thyroid cancer, represents an essential step in understanding the disease pathogenesis, and help to improve the diagnosis methods for an individual appropriate disease management.

Like other human tumours, aberrant methylation of tumour suppressor genes is also a characteristic for thyroid tumours. Some reports mention the involvement in thyroid tumorigenesis of aberrant silencing trough promoter methylation of PTEN [25], RASSF1A [26, 27], tissue inhibitor of metalloprotei- nase-3 (TIMP3), SLC5 A8, death-associated protein kinase (DAPK) and retinoic acid receptor $\beta$ 2 (RAR $\beta$ 2) [28, 29] tumour suppressor genes.

Our study confirms the precedent reports regarding the gene promoter methylation of RARB, RASSF1, PTEN in thyroid cancer. According to gene panel studied we found new factors characteristic to thyroid tumours: methylation of TP73, WIF1 and PDLIM4 tumour suppressor genes, which can be a contributor to thyroid neoplasia.

The TP73 gene is a member of TP53 family involved in controlling cellular proliferation, differentiation and cell death, which includes also the TP63 gene [30]. TP73 role compensates for the loss of p53 function, and is rarely mutated compared to TP53 [31]. One interesting fact is that TP73 gene transcription is controlled by two different promoters (P1 and P2), leading to p73 protein isoforms with opposite functions, one containing (TA) or lacking ( $\mathrm{\Delta }$ N) the transactivation domain. While TAp73 possesses a pro-apoptotic tumour suppressor protein function, $\mathrm{\Delta }$ Np73 acts as an anti-apoptotic oncoprotein [32, 33]. It was observed that $\mathrm{\Delta }$ Np73 isoform expression increases in tumours which presented P1 promoter hypermethylated [34].

Hypermethylation of the TP73 gene promoter was also found in a variety of solid tumours such as squamous cell lung cancer [35], gastric carcinoma [36] and cervical cancer [37].

The current study showed that the methylation of TP73 gene promoter is significantly higher in tumour tissue compared to normal tissue ( $p<$ 0.0001), and a borderline significance ( $p=$ 0.05) was found when the different neoplasias are compared. The highest methylation values were characteristic for PTC samples. The expression level of TP73 gene without discrimination between isoforms is higher in tumour samples vs. normal, but no differences were observed between the thyroid neoplasia samples. Traces of TP73 methylation were detected in blood of patients with PTC and fvPTC, the amount being significant ( $p=$ 0.0148) compared to normal, having a good potential for diagnosis and disease management.

The WNT inhibitory factor-1 (WIF-1) gene is a secreted antagonist that binds to WNT proteins to suppress WNT/ $\beta$ -catenin signalling, which is known to be involved in oncogenesis process [38, 39].

WIF-1 gene expression was found to be down-regulated through promoter hypermethylation, geno- mic loss, or genomic rearrangement in many human neoplasia, including lung, colon, breast, bladder, kidney, prostate and salivary gland tumours [40, 41, 42, 43, 44, 45, 46].

We showed that WIF1 is hypermethylated in tumour samples, the percentage of methylation being significantly higher ( $p<$ 0.0001). The WIF1 methylation values were increased also in thyroid adenoma samples, indicating this factor could contribute in early stages of oncogenesis. The median values of WIF1 gene were lower in the thyroid neoplasm group. WIF1 gene promoter presented higher methylation percentage values in blood samples than controls.

PDLIM4 (reversion-induced LIM gene-RIL) was a potential tumour suppressor gene involved in cell growth regulation and modulates actin stress fibre dynamics through its association with alpha-actin [47]. Moreover, it has been shown that PDLIM4 promoter methylation could be the main mechanism to suppress its expression in prostate cancer tissues [48]. Other report showed also an association of PDLIM4 promoter methylation with breast cancer [49]. In colon cancer hypermethylation of PDLIM4 suppresses activation of Src (a regulatory protein that functions in several fundamental processes, including cell differentiation, proliferation, migration, and survival) [50]. PDLIM4 promoter is hypermethylated in tumour samples, especially in PTC samples. Follicular adenomas presented higher PDLIM4 methylation levels, suggesting the possibility to use this factor as a marker for diagnosis as long as significant methylation levels were quantified from blood of patients with thyroid neoplasms. Lowest PDLIM4 expression values were found in fvPTC samples.

Detection of methylation levels of the tested genes in blood could represent a potential method to evaluate the patient status. Even if PDLIM4 and TP73 genes expression decreases gradually from benign neoplasia to malign status, and WIF1 presented the lowest value in PTC samples; evaluation of gene expression in blood does not provides the same informational value.

We indicated also that the treatment with demethylating agents as 5-AzaC could restore the gene function as the expression increases significantly and the promoter becomes active. The increment was notable especially for TP73 gene, but also for PDLIM4 and WIF1 gene.

Among the tested genes we noted four more important genes NEUROG1, ESR1, RUNX3, MLH1, which presented the promoter with higher methylation values in tumour samples compared to normal.

The genetic tests for mutations in thyroid cancer are innapropiate due to the low incidence of mutations in thyroid oncogenesis [51] and no specifc inactivating mutations have been identifed in well-differentiated thyroid cancers [52, 53, 54]. However, BRAF V600E mutation is present in papillary thyroid cancer and is associated with an aggressive tumor phenotype and higher risk of recurrent and persistent disease [55].

Various studies suggested the involvement of BRAF V660E mutation in aberrant methylation processes of certain genes in colon cancer [56, 57] and in thyroid cancer [10, 29]. A high-throughput study demonstrated the effect of BRAF V600E signalling on aberrant methylation that may have a profound impact on cell functions and behaviors of PTC [58]. We found also a significant correlation between Braf V600E mutation presence and TIMP3 gene promoter methylation, but also with MGMT, NEUROG1, RAR-beta gene promoter methylation.

RET/PTC rearrangements were associated with a more aggressive phenotype, a larger tumor size, and it was shown to influence the methylome [59, 60]. The present study showed a significant correlation between RET/PTC rearrangement and CDH13, RARB genes promoter methylation. A correlation was also observed for this translocation with ESR1, NEUROG1 and SOCS1 gene promoter methylation.

5. Conclusions

The hypermethylation of tumour suppressor promoter is a hallmark of cancer and by inhibiting the genes expression, it alters the cell signalling pathways. Therefore, a test that uses as a method for methylation quantification of tumour suppressor genes is suitable for diagnosis and prognosis of thyroid cancer.