Prostanoid receptor genes confer poor prognosis in head and neck squamous cell carcinoma via epigenetic inactivation

Background Chronic inflammation is a risk factor for head and neck squamous cell carcinoma (HNSCC) and other diseases. Prostanoid receptors are clearly involved in the development of many types of cancer. However, their role is not simple and is poorly understood in HNSCC. Methods Methylation profiles of prostanoid receptor family genes were generated for tumour samples obtained from 274 patients with HNSCC, including 69 hypopharynx, 51 larynx, 79 oral cavity, and 75 oropharynx tumour samples, by quantitative methylation-specific PCR. Promoter methylation was then evaluated with respect to various clinical characteristics and patient survival. Results The mean number of methylated genes per sample was 2.05 ± 2.59 (range 0 to 9). Promoters of PTGDR1, PTGDR2, PTGER1, PTGER2, PTGER3, PTGER4, PTGFR, PTGIR, and TBXA2R were methylated in 43.8%, 18.2%, 25.5%, 17.5%, 41.2%, 8.0%, 19.3%, 20.4%, and 11.3% of the samples, respectively. Methylation indices for prostanoid receptor family genes tended to be higher as the number of TET methylation events increased. Patients with 5–9 methylated genes had a significantly lower survival rate than that of patients with 0–4 methylated genes (log-rank test, P= 0.007). In multivariate analyses, PTGDR1 methylation was most highly correlated with recurrence in patients with hypopharyngeal cancer (P = 0.014). A similar correlation was observed for PTGER4 in patients with laryngeal cancer (P = 0.046). Methylation of the PTGIR and TBXA2R promoters was positively correlated with recurrence in oropharyngeal cancer (P = 0.028 and P = 0.006, respectively). Moreover, Patients with 5–9 methylated genes were extremely lower of 5hmC levels (P = 0.035) and was correlated with increasing expression of DNMT3A and DNMT3B (P < 0.05 and P < 0.05, respectively). Conclusion We characterised the relationship between the methylation status of prostanoid receptor genes and recurrence in HNSCC. These results provide new perspectives for the development of molecular targeted treatment approaches.


Tumour samples
Surgical HNSCC tumour and matched adjacent nontumour tissues were obtained from 274 patients who underwent surgical resection at the Department of Otolaryngology/Head and Neck Surgery, Hamamatsu University School of Medicine (Hamamatsu, Shizuoka, Japan). Written informed consent was obtained from individual patients before surgery and the experimental protocol was approved by the Hamamatsu University School of Medicine (date of board approval: October 2, 2015, ethics code: . The ratio of males to females was 227:47. The mean age was 65.2 years (range, 32 to 90 years). Primary head and neck tumours included 69 hypopharyngeal carcinomas, 51 laryngeal carcinomas, 79 oropharyngeal carcinomas, and 75 oral cavity carcinomas (Additional file 1: Table S1).

DNA extraction and modification
DNA extraction from fresh tissue was performed using a QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). Sodium bisulphite conversion was performed using the MethylEasy Xceed Rapid DNA Bisulfite Modification Kit (TaKaRa, Tokyo, Japan) following the manufacturer's protocol.

Quantitative methylation-specific PCR analysis (Q-MSP)
Aberrant DNA methylation, which often occurs around the transcription start site (TSS) within a CpG island, was evaluated by Q-MSP. The sequences of primers used in this study are shown in Additional file 2: Table S2. Exon one or two and CpG sites within view of the promoter region relative to the TSS are presented in Additional file 3: Figure S1. A standard curve for Q-MSP was constructed by plotting five serially diluted standard solutions of EpiScope Methylated HeLa gDNA (TaKaRa). The normalized methylation value (NMV) was defined as follows: NMV = (prostanoid receptor gene-S/prostanoid receptor gene-FM)/(ACTB-S/ACTB-FM), where prostanoid receptor gene-S and prostanoid receptor gene-FM represent target gene methylation levels in the tumour sample and in the universal methylated DNA control, respectively. ACTB-S and ACTB-FM correspond to β-actin (ACTB) in the sample and the universally methylated DNA, respectively [15].

Detection of high-risk HPV DNA by PCR
To identify the HPV types, samples were also subjected to PCR using specific primers for HPV types 16,18,31,33,35,52, and 58. The prevalence of HPV DNA was examined using the PCR HPV Typing Set (TaKaRa).

ELISA for 5-hmC quantification
The 5hmC content of genomic DNA was determined with a Quest 5-hmC DNA ELISA Kit (Zymo Research, Irvine, CA, USA), according to the manufacturer's instructions. The amount of 5-hmC was calculated as a percentage based on a standard curve generated using kit controls.

RNA extraction and quantitative reverse transcription PCR (qRT-PCR)
Total RNA was isolated using an RNeasy Plus Mini Kit (Qiagen, Hilden, Germany); cDNA was synthesized using a ReverTra Ace qPCR RT Kit (Toyobo, Tokyo, Japan). DNMT3A, DNMT3B, and GAPDH mRNA expression levels were measured via qRT-PCR using SYBR Premix Ex Taq (Takara Bio Inc., Tokyo, Japan), the Takara Thermal Cycler Dice Real Time System TP8000 (Takara Bio Inc.), and the primer sets presented in previously reports [16].

Data analysis and statistics
A receiver operator characteristic (ROC) curve analysis of target genes was performed using the NMVs for 36 matched paired HNSCC and normal mucosal samples and the Stata/SE 13.0 system (Stata Corporation, College Station, TX, USA). To determine the area under the ROC curve, the true positive rate (Sensitivity) was plotted as a function of the false positive rate (1 − Specificity) for different cut-off points, and the NMV thresholds were calculated for each target gene. Cut-off values showing the greatest accuracy were determined based on sensitivity/specificity, as indicated in Additional file 4: Table S3. The MI was defined as the number of genes with promoter methylation [17]. Student's t-tests were performed to evaluate the associations between clinical variables and MI. Disease-free survival (DFS) was investigated using the Kaplan-Meier method and the log-rank test. The probability of survival can be evaluated by generating a Kaplan-Meier curve. A Cox's proportional hazards regression analysis that included age (≥ 65 vs. < 65 years), sex, alcohol intake, smoking status, and tumour stage (I-II vs. III-IV) and methylation status was used to identify the multivariate predictive value of prognostic factors. A value of P < 0.05 was considered statistically significant.

Analysis of methylation status in HNSCC tissue samples
A Q-MSP analysis of the methylation status of nine prostanoid receptor genes was performed using 274 primary HNSCC samples. The methylation frequencies were as follows: PTGDR1 (43.8%), PTGDR2 (18.2%), PTGER1 (25.5%), PTGER2 (17.5%), PTGER3 (41.2%), PTGER4 (8.0%), PTGFR (19.3%), PTGIR (20.4%), and TBXA2R (11.3%) (Fig. 1a). The average MI per sample was 2.05 ± 2.59 (range 0 to 9) (Fig. 1b). No significant differences in MI were observed with respect to the age at disease onset, sex, alcohol consumption, smoking status, tumour size, lymph node status, clinical stage, or HPV status (Fig. 1c). We analysed the relationships of the methylation status of each prostanoid receptor gene with the clinical features of patients with HNSCC. PTGFR methylation was significantly correlated with age at onset (P = 0.043). Methylation levels of PTGDR2, PTGER1, and PTGFR promoters were associated with alcohol exposure (P = 0.042, P = 0.04, and P = 0.049, respectively). There was an association between methylation of the PTGDR1 and PTGER3 promoters and HPV status (P = 0.004 and P = 0.005, respectively). We found that the promoter methylation of all prostanoid receptor genes with the exception of TBXA2R was associated with recurrence events ( Table 1).

Comparison of methylation frequencies between nine prostanoid receptor genes and ten-eleven translocation (TET) family genes
Mean differences in the MI of nine prostanoid receptor genes based on TET gene methylation events are illustrated in Fig. 2a. The MI was significantly higher in patients with methylation events at all TET genes (4.84 ± 2.73), two TET gene methylation events (2.92 ± 3.09), and one TET gene methylation event (1.94 ± 2.03) than in patients with no TET gene methylation events (0.39 ± 1.05; P < 0.01 for all comparisons) (Fig. 2b).

Site-specific analysis of the methylation status
Site-specific methylation frequencies across nine genes for the hypopharynx, larynx, oropharynx, and oral cavity are shown in Fig. 4a. MI levels were significantly higher in patients with hypopharyngeal cancer than in patients with oral cavity cancer (P = 0.020) (Fig. 4b).
Among 69 cases with hypopharyngeal cancer, the DFS rate in those with PTGDR1 methylation was similar to that in the unmethylated group (log-rank test, P = 0.011; Additional file 8: Fig. S4A). Patients with laryngeal cancer and methylated PTGER4 promoters had a relatively short DFS (log-rank test, P = 0.020; Additional file 8: Fig.  S4B). Additional analysis including only patients with oropharyngeal cancer (n = 79) revealed a shorter DFS for methylated than for unmethylated PTGIR and TBXA2R (log-rank test, P = 0.003 and P = 0.009, respectively; Additional file 8: Fig. S4C, D).

Stratification analysis
The relation between the methylation status and risk of recurrence was analysed by a multivariate analysis using a Cox proportional hazards regression model  5).

Comparison of methylation frequencies between nine prostanoid receptor genes and other epigenetic factors
The 5-hmC level showed the greatest decrease when prostanoid receptor genes were 9 to 5 methylation events (P = 0.035; Additional file 9: Fig. S5A). The DNMT3A mRNA expression levels were significantly higher in groups with 9 to 5 methylation and 4 to 1 methylation (P = 0.026 and P = 0.029, respectively; Additional file 9: Fig. S5B). The DNMT3B mRNA expression was significantly correlated with both 9 to 5 methylation and 4 to 1 methylation (P = 0.013 and P = 0.020, respectively; Additional file 9: Fig. S5C).

Analysis of methylation and expression data from TCGA
The methylation status of prostanoid receptor gene promoters was estimated in an additional 516 HNSCC samples and 50 normal samples from TCGA. The average β-values (indicating promoter methylation) for the nine genes were significantly higher in the HNSCC samples than in the normal samples (P < 0.05), except for PTGER4 and TBXA2R (Additional file 10: Fig. S6). The expression of prostanoid receptor genes were significantly higher in IL-6 mRNA high expression group (P < 0.05), except for PTGER1 gene. IL-11 expression was positively correlated with PTGDR1 and TBXA2R expression (P = 0.001 and P < 0.001, respectively). Expression of RANKL was concurrently associated with all prostanoid receptor genes expression (Additional file 11: Table S5).

Discussion
Epigenetic modifications of prostanoid receptor genes may contribute to tumour development and recurrence. We analysed the methylation statuses of genes encoding neuropeptide GPCRs in 274 HNSCCs originating in the hypopharynx, larynx, oropharynx, or oral cavity. We also compared the methylation status of genes in matched HNSCC and normal samples using data from TCGA. We found that the aberrant methylation of some prostanoid receptor gene promoters is positively correlated with recurrence in patients with HNSCC. In addition, a site-specific analysis revealed that abnormal CpG island hypermethylation was independently associated with aggressive clinical behaviour. Cancer may be related to chronic inflammation associated with persistent infections, immune-mediated damage, or prolonged exposure to irritants. Genetic and epigenetic alterations underlying carcinogenesis inevitably modify tissue homeostasis and may induce a chronic inflammatory response. Over 20 years ago, non-steroidal anti-inflammatory drugs (NSAIDs) were reported to have anti-colon cancer effects [18]. NSAIDs, which are potent inhibitors of COX, exert chemopreventive effects in cancer development [19]. Numerous epidemiological studies have shown that the regular intake of the NSAID aspirin, an inhibitor of COXs, substantially reduces both the incidence and progression of several prevalent cancers [20]. Abundant epidemiological and preclinical/clinical studies have demonstrated that celecoxib, a specific COX-2 inhibitor, is related to the suppression of cancer cell proliferation and a decrease in cancer incidence [21]. COX-1 is constitutively expressed in many tissues and regulates basal levels of prostaglandins [22]. COX-2 is responsible for the release of prostaglandins after an infection, injury, or in cancer development [23]. In HNSCC, IL-1 released by tumour cells plays a key role in inducing    the expression of COX-2 in fibroblasts [11]. Secretion of TGF-β and PGE2 by the HNSCC cells was increased following EGFR inhibition [24]. IL-6, TNF-a and PGE2 produced by primary oral keratinocytes and carcinoma cells may induce oral mucosal inflammation [25]. Recent studies continue to support the prostanoid pathway as a promising target for future HNSCC therapies. Prostanoids, including PGD2, PGE2, PGF2α, PGI2, and TXA2, activate nine GPCRs, namely PTGDR1, PTGDR2, PTGER1, PTGER2, PTGER3, PTGER4, PTGFR, PTGIR, and TBXA2R. PTGDR1 downregulation by DNA hypermethylation is correlated with colorectal cancer development [26,27]. PTGDR1 methylation from cervical scraping is a promising marker of endometrial cancer and ovarian cancer [28]. PTGER1 shows a strong association with DNA methylation in non-functioning adrenocortical adenoma [29]. The expression of PTGER2 is often silenced in neuroblastoma cell lines by epigenetic mechanisms [30]. Increased DNA methylation of PTGER2 is associated with the progression of neuroblastomas [30], non-small cell lung cancer [31], and cervical cancer tissue [32], suggesting that the aberrant methylation of this gene regulates cell proliferation. Cebola et al. detected PTGER3 and PTGFR hypermethylation in a high proportion of colorectal cancer cases, suggesting that DNA methylation is an important mechanism involved in the deregulation of this pathway [33]. The measurement of PTGER4 methylation in plasma DNA obtained by minimally invasive sampling can be used to detect malignant lung disease [34]. The loss of methylation and activation of PTGER4 can explain the acquisition of endocrine therapy resistance and is a therapeutic target for breast cancer [35].
Cancers of the upper aerodigestive tract account for the majority of squamous cell carcinomas, which develop in the epithelial linings of the oral cavity, pharynx, and larynx [36]. There are several subclassifications based on anatomic location, aetiology, and molecular findings [37]. Head and neck cancers arise from a multistep process involving the accumulation of genetic and epigenetic alterations [38]. DNA methylation is a frequent and key epigenetic mechanism underlying the regulation of processes associated with neoplastic transformation [38]. Recently, TET proteins have been identified as important epigenetic modifiers via their dioxygenase activity [39]. TET expression and activity are inhibited by genetic mutations and high methylation of their own promoters [40]. Our data showed that increased DNA methylation of TET genes is correlated with the accumulation of prostanoid receptor genes with aberrant methylation; this may be a meaningful DNA methylation event in HNSCC progression. Furthermore, the groups of high MI were extremely low of 5hmC levels and was correlated with increasing expression of DNMT3A and DNMT3B. To our knowledge, our study is the first to suggest that PTGDR1, PTGER4, PTGIR, and TBXA2R methylation is associated with worse DFS and may be a critical event in hypopharyngeal cancers, laryngeal cancers, oropharyngeal cancers, and oral cancers, respectively. However, our results obtained from human specimens and highthroughput profiling platforms may be susceptible to measurement bias from various sources. The current study continues to support the prostanoid receptor as a promising target for future HNSCC therapies.
We systematically evaluated the methylation status of the promoters of nine prostanoid receptor genes and the relationship between methylation and clinical characteristics in HNSCC samples. We identified a novel prognostic biomarker based on promoter DNA methylation changes in operable HNSCC to identify patients at high risk of recurrence and provide complementary epigenetic characterization of this tumour type. Collectively, these data demonstrate the functional roles of the epigenetic regulation of prostanoid receptors and show that these loci are potential targets for epigenetic therapies for inflammatory disorders, such as HNSCC.

Conclusion
We determined the relationship between the methylation status of prostanoid receptor genes and recurrence in HNSCC, providing new perspectives for the development of molecular targeted therapeutic approaches. To our knowledge, our study provides the first evidence for an association between PTGDR1, PTGER4, PTGIR, and TBXA2R methylation and worse survival in hypopharyngeal cancers, laryngeal cancers, oropharyngeal cancers, and oral cancers, respectively. This study involving human specimens and high-throughput profiling platforms may be susceptible to measurement bias from various sources; accordingly the use of methylation markers in clinical practice requires further testing in prospective studies with larger HNSCC cohorts.