1 Introduction

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common human malignancy in the world, with high mortality and dismal prognosis. More than half a million new cases are reported per year, and the 5-year survival rate is approximately 50%. Many internal and external factors have been reported to be associated with head and neck cancer (HNC). DNA methylation is an internal factor that relates to HNC.

DNA methylation is a kind of epigenetic modification that refers to changes in the level and pattern of DNA methylation while maintaining normal DNA sequences. DNA methylation can be affected by many external factors, such as tobacco abuse, alcohol intake, betel quid chewing, and pharmaceutical treatment. Enzymes including DNMTs and TETs contribute largely to this process.

In this review, we summarize DNA methylation associated with head and neck cancer to illustrate the relationships between them and shed light on the clinical potential of DNA methylation, such as diagnostic assistance and therapeutic effect prediction in HNC.

2 Hypermethylation of tumor suppressor genes in HNC

CpG dinucleotides are aggregated at 5’ promoter regions, also known as CpG islands (CGIs), which are commonly sparsely methylated in transcriptionally active genes [1]. In some pathological conditions, CGI methylation levels increase, gene transcription is inhibited, and gene dysfunction occurs [2]. Once the hypermethylation of CGIs occurs in tumor suppressor genes, their transcriptional silencing follows, consequently leading to malignancies [3]. Here, we review recent progress in studies of p16, PTEN, DAPK, MGMT, ECAD and RASSF1A, which are widely researched in HNSCC.

2.1 P16 hypermethylation in HNC

The gene p16 (CDKN2A), belonging to the INK4 family of genes, is the second most common tumor suppressor gene after p53 [4]. The physiological functions of the p16 product are inhibiting cyclin-dependent kinase and negatively regulating the cell cycle at the G1/S phase [5, 6]. p16 gene inactivation caused by promoter hypermethylation may result in uncontrolled cell proliferation, which may then contribute to carcinogenesis [6]. Many studies have confirmed that hypermethylation of the p16 promoter frequently occurs in HNSCC [7, 8]. A meta-analysis including 804 thyroid cancer cases and 487 controls showed that the frequency of p16 promoter methylation in cancer tissue/blood was significantly higher than that in normal tissue/blood, which led to the conclusion that p16 promoter methylation may be associated with tumorigenesis [9]. Studies among 67 consecutive oral squamous cell carcinoma (OSCC) patients and 59 normal individuals suggested a relationship between methylation rate and the repressed expression of p16, which prompted p16 status as a potential classifying factor and a promising therapeutic target for OSCC, which accounts for 17.8% of HNC [7, 10]. Researchers have also revealed that p16 methylation and expression are affected by smoking and HPV infection [7]. A case‒control study found a significantly higher occurrence rate for p16 hypermethylation in HPV 16-positive oral epithelial dysplasia samples than in HPV 16-negative samples [11]. Systematic network analysis carried out in 104 OSCC patients revealed consistently downregulated expression of p16 in recurrent cases. Array comparative genomic hybridization revealed a copy number deletion, and next-generation sequencing confirmed the deletion and reported it as a commonly occurring nonsense mutation with stop/loss of function of the gene in recurrent patients. Downregulation of p16 also suggested a dismal prognosis and low survival rate in OSCC [12]. In the field of prognosis prediction, researchers believe that p16 hypermethylation is generally associated with poor survival [13] For recurrence, in an Indian cohort of 116 patients, scientists found that cases with hypermethylation of p16 had a threefold higher risk of disease recurrence [13].

2.2 PTEN hypermethylation in HNC

The phosphatase and tensin homolog deleted on chromosome 10 gene (PTEN) is also a widely known tumor suppressor gene. The PTEN protein possesses both protein and lipid activities, regulating many cellular processes, including proliferation, survival, and energy metabolism. PTEN proteins can dephosphorylate proteins and peptides phosphorylated on tyrosine residues. Insulin/PI3K pathways play an important role in regulating cell growth and metabolism by PTEN [14]. Kim discovered that hypermethylation of PTEN induced by EPHA3 (a member of the EPH family), which leads to inhibition of PTEN expression, is commonly found in radioresistant HNC cells. Understanding the resistance mechanisms related to EPHA3 and PTEN can benefit patients by optimizing clinical treatment plans [15]. Examination of fifty biopsy samples from HNC patients confirmed that both PTEN and p16 promoter hypermethylation occurred in HNC, leading to low expression of PTEN and p16. Studies have also suggested that the low expression of PTEN and p16 may contribute to tumor progression and could be a target for predicting the prognosis of HNC [16]. PTEN deficiency was also confirmed to contribute to the development and progression of HNC [17]. CpG island hypermethylation of PTEN was verified in nasopharyngeal carcinoma (NPC) in both tissues and cell lines compared to nontumor nasopharyngeal epithelial tissues, which suggested that hypermethylation of PTEN may be an early stage and a candidate biomarker of NPC [18]. A case‒control study carried out in patients with HNC and adjacent normal-tissue control also confirmed the correlation between downregulation of PTEN and the initiation and progression of HNC. Furthermore, upregulation of the progression biomarker Ki-67 was also revealed to be related to the downregulation of PTEN, which hinted at the biomarker potential of PTEN [19].

2.3 DAPK hypermethylation in HNC

The death-associated protein kinase (DAPK) gene is also a tumor suppressor gene, carrying out the function of responding to various death stimuli. DAPK-induced cell death most prominently features as an effect on the cytoskeleton, including loss of matrix attachment and membrane blebbing [20]. Promoter hypermethylation of DAPK mainly leads to suppressed expression of DAPK, which may then cause tumorigenesis [21]. Studies have revealed that DAPK hypermethylation also occurs in HNC [8, 22]. Furthermore, DAPK hypermethylation was also determined to be significantly related to lymph node involvement and advanced disease stage [8]. As calculated by a meta-analysis, the estimated prevalence of DAPK promoter methylation in OSCC was 39.7% [23], while another meta-analysis including eighteen studies showed that DAPK promoter methylation in HNSCC patients was 4.09-fold higher than that in normal controls [24]. Strzelczyk and colleagues also concluded that DAPK1 exhibited higher methylation frequency in tumors than in surgical margins. Additionally, patients with family histories and older age exhibited more frequently methylated DAPK1, while DAPK hypermethylation was also closely related to lymph node metastasis and decreased risk of death, which made DAPK1 methylation status a potential diagnostic biomarker of OSCC [25]. Arantes et al. [26] further validated that DAPK methylation was highly specific for OSCC samples, and the combination of DAPK and several other genes reached an even higher sensitivity and specificity. Moreover, DAPK hypermethylation was also related to clinical T1 and T2 stages. A systematic review carried out by Pall et al. [27] discussed the potential of circulating tumor DNA (ctDNA) in diagnosis or recurrence monitoring tests. Sixteen studies were included in the systematic review, of which 3 studies evaluated the methylation of the DAPK1 gene. The study revealed that increasing the number of ctDNA genetic methylations resulted in an increase in diagnostic sensitivity and accuracy.

2.4 MGMT hypermethylation in HNC

MGMT (O6-methylguanine-DNA methyltransferase) gene produces a DNA repair protein that functions mainly by transferring methyl and alkyl lesions from the O6 position of guanine to a cysteine in its structure [28]. Recently, many researchers have reported MGMT hypermethylation in HNSCC [8], here we summarize recent progress of MGMT hypermethylation researches in HNSCC. A meta-analysis including 20 studies (1,030 cases and 775 controls) unveiled that the frequency of MGMT promoter methylation was significantly higher in HNSCC patients than normal controls, which indicates the relation between aberrant methylation of MGMT promoter and the rising risks of HNSCC [29]. The conclusion was confirmed in another meta-analysis of OSCC, as well [23]. Meanwhile, the expression of MGMT is also lower in methylated samples than unmethylated ones [30]. A pyrosequencing was carried out to determine the methylation pattern of MGMT in laryngeal cancer, regarding the methylation of MGMT common in laryngeal cancer, but couldn’t ensure the relation between MGMT methylation status and clinicopathological parameters, including age, tumor stage, histopathological differentiation, recurrence or disease-free survival [31]. Somewhat controversially, a meta-analysis including 17 studies (1,368 patients and 1,489 normal controls) in esophageal cancer ensured the relation between MGMT promoter methylation and tumorigenesis, but the study also concluded the MGMT promoter methylation correlated with age, lymph node status and clinical stage [32]. To evaluate the prognostic value of MGMT, Scesnaite et al. [33] detected the MGMT promoter methylation, MGMT expression and clinicopathological parameters in salivary gland cancer patients, showing the significant ability of MGMT loss to predict poor clinical outcome of salivary gland cancer patients, holding the potential of predicting overall survival. We can conclude that promoter hypermethylation of MGMT leads to lower expression, which disturb the physical tumor suppression effect of MGMT, and finally associated with carcinogenesis.

2.5 ECAD hypermethylation in HNC

E-cadherin (ECAD) is a calcium-dependent cell‒cell adhesion molecule with important roles in the suppression of cancer [34] and is frequently hypermethylated in HNSCC patients [22]. Promoter hypermethylation of ECAD is an important cause of ECAD repression, which may then switch E-cadherin into mesenchymal cadherin, such as N-cadherin. N-cadherin did not enhance carcinogenesis but promoted tumor progression by potentiating ERK oncogenic signaling involving MMP-9 upregulation [34]. A meta-analysis including 13 studies showed a significant association between ECAD promoter hypermethylation and oral cancer risk [35]; additionally, such promoter hypermethylation was likely to lead to lower expression of ECAD [36]. Another meta-analysis including more than 4 studies revealed a significant association between ECAD promoter methylation and thyroid tumorigenesis [37]. A meta-analysis of nasopharyngeal cancer confirmed this conclusion and regarded ECAD methylation as an effective biomarker for the early detection of NPC [38]. In the field of clinical use, ECAD hypermethylation contributes to cancer diagnosis and prognosis prediction. Righini et al. [39] used methylation-specific PCR to determine the methylation status in primary tumors, normal adjacent mucosa, and saliva specimens. The results showed that ECAD was one of the most frequently observed methylated genes in tumors and paired saliva samples, while 90% of normal adjacent mucosa and all normal saliva samples were negative. Generally, ECAD methylation in saliva exhibited strong potential for follow-up and early detection in HNSCC patients. Both saliva and oral rinse were also used in the detection of OSCC. The aberrant methylation of ECAD in oral rinse had high sensitivity (>75%) and specificity for detecting oral cancer. Additionally, using the combination of ECAD and several other genes reached even higher sensitivity (97.1%) and specificity (91.7%) [40].

2.6 RASSF1A hypermethylation in HNC

Ras association domain family 1 (RASSF1) contains two main variants (RASSF1A and RASSF1C); RASSF1A is a scaffold protein communicating with the Ras pathway, estrogen receptor signaling pathway, and Hippo pathway and functions as a tumor suppressor gene through apoptotic signaling, microtubule stabilization and mitotic progression [41, 42]. Similar to the other tumor suppressor genes mentioned above, promoter hypermethylation of RASSF1A leads to loss of expression in HNC [43]. In thyroid cancer, Schagdarsurengin et al. [44] detected complete methylation of the RASSF1A promoter CpG island in nine cell lines and absent expression of RASSF1A. Additionally, a DNA methylation inhibitor could reactivate RASSF1A transcription. Methylation status was also analyzed in 38 primary thyroid tumors, showing that 71% of thyroid carcinomas had hypermethylated RASSF1A CpG islands. Higher methylation of RASSF1A was also confirmed by means of methylation-specific multiplex ligation-dependent probe amplification in sinonasal adenocarcinoma and squamous cell carcinoma [45]. A meta-analysis containing 55 articles discovered that promoter methylation of RASSF1 was a risk factor for thyroid cancer, with an odds ratio of 4.16 [37]. However, the subgroup meta-analysis also determined that the fresh-frozen tissue and paraffin-embedded tissue were similar, while blood samples were less reliable due to limited study quantity. Meng et al. [46] performed a similar meta-analysis, including 12 studies (550 HNSCC patients and 404 controls), and came to a similar conclusion that there was a significant association between aberrant RASSF1A promoter methylation and HNSCC. A systematic review carried out by Pall et al. [27] revealed a significant association between ctDNA methylation of RASSF1 and HNSCC and indicated that multiple gene methylation detection of HNSCC may increase diagnostic sensitivity accuracy. Circulating free-cell DNA (cfDNA) was also used as a diagnostic source in cancer. The methylation status of the RASSF1 promoter region was analyzed by methylation-specific high‐resolution melting curve analysis, and the results showed that hypermethylation in the more proximal promoter regions to the RASSF1 and SLC5A8 genes expressed higher and acceptable specificity for discrimination between papillary thyroid cancer and thyroid nodules [47].

3 Hypomethylation in HNC patients

DNA hypomethylation occurs frequently in head and neck neoplasm patients, including genome-wide hypomethylation and hypomethylation of gene promoters, retrotransposons, CpG islands or gene deserts, which may decrease the stability of genes [48]. In the field of gene expression, in contrast to DNA hypermethylation, hypomethylation usually affects oncogenes and induces their expression or overexpression instead of downregulating tumor suppressor genes via DNA hypermethylation [49].

3.1 Global or genome-wide DNA hypomethylation in HNC

Global or genome-wide DNA hypomethylation mostly influences tumor development and prognosis by inducing chromosomal instability [48]. Global DNA methylation status could be assessed by the methylation level of LINE-1, as it makes up approximately 17% of the human genome [50]. In esophageal squamous cell carcinoma, LINE-1 hypomethylation was significantly associated with lymph node involvement, lymphovascular invasion and poor survivability. The results suggested that genome-wide hypomethylation may initiate carcinogenesis of esophageal squamous cell carcinoma (ESCC) through chromosomal instability, which may also correlate with the progression of ESCC [51]. However, studies have also revealed that when reduced by DNMT1 hypomorphic alleles, genomic hypomethylation inhibits carcinogenesis in the tongue and esophagus [52]. In tongue squamous cell carcinoma (TSCC), researchers analyzed 248 surgically resected TSCC and 202 corresponding tumor adjacent normal (TAN) tissues, discovering that the global 5mC level in TSCC was significantly lower than that in TAN tissues. Moreover, global hypomethylation of TSCC was also associated with poor disease-specific survival [53]. Assessed by the level of 5-methylcytidine (5-mc), the global methylation status has been proven to be a diagnostic biomarker for thyroid tumors. Compared with benign tumors or adjacent normal tissues, computerized image analysis showed a significantly lower level of 5-mc immunostaining in thyroid cancer. Combined with other valuable markers, such as galectin-3, the diagnostic accuracy of 5-mc could even reach a high level of 96% [54].

To evaluate the function of global DNA hypomethylation in the development and prognosis of HNSCC, Wu et al. measured the expression of Dicer 1, DNMT1, DNA-dependent protein kinase catalytic subunit (DNA-PKcs) and let-7b miRNA in 150 advanced-stage ESCC patients treated with preoperative radiotherapy. The results suggest that global DNA hypomethylation is correlated with the sensitivity of ESCC to radiotherapy and the prognosis of patients with ESCC [55]. The genome-wide methylation features in Barrett’s esophagus (BE) and esophageal adenocarcinoma (EAC), including DNA hypomethylation, were used to group BE and EAC into four subtypes, and each subtype has different outcomes and potential responses to therapy [56]. To explore the more specific mechanisms of DNA hypomethylation, Liu et al. revealed that HNSCC patients with high APOBEC3H levels tended to exhibit a genome-wide hypomethylation pattern, and APOBEC3H was also believed to demethylate and upregulate CXCL10 and contribute to CD8 + T-cell immune infiltration, which made it a potential biomarker for immunotherapy in HNSCC [57]. Overexpression caused global hypomethylation in ESCC cell lines, and the UHRF1 expression level was also significantly associated with global methylation levels in ESCC tissues, which confirmed it as an epigenetic regulator of DNA methylation [58].

3.2 Hypomethylation of the promoter of genes in HNC

Hypomethylation of gene promoters can modulate gene expression and consequently influence tumor biological behaviors. In esophageal cancer, a study revealed that the silencing of long intergenic noncoding RNA 184 (LINC00184) could demethylate the promoter of PTEN and subsequently inhibit glycolysis, proliferation, migration, invasion and colony formation of esophageal cancer cells and tumor growth while restoring mitochondrial oxidative phosphorylation [59]. 5-Aza-2′-deoxycytidine DNA demethylation of claudin-4 (CLDN4), a component of epithelial tight junctions, suppressed the migration and invasion of Hep-2 cells, while silencing CLDN4 restored the migration and invasion of Hep-2 cells, which indicates the potential function of DNA demethylation in LSCC [60]. In nasopharyngeal carcinoma, led by LMP2A ectopic expression, demethylation of the promoter of S100 calcium binding protein A4 (S100A4) was found in LMP2A-positive NPC tissues. Moreover, S100A4 was also overexpressed in CNE-1 cells ectopically expressed LMP2A, showing highly increased invasion ability. The results above demonstrated the function of hypomethylation and the activation of related genes in cancer metastasis and progression [61]. S100A4 expression could also be upregulated by hypomethylation of the c-Myb motif, one of the four transcription factor binding motifs, in laryngeal cancer [62].

3.3 Hypomethylation of retrotransposon elements in HNC

Retrotransposon elements, including long interspersed elements (LINEs) and short interspersed elements (SINEs), have been proven to be associated with many types of cancer. DNA methylation of the LINE-1 promoter is important for inhibiting transposition and consequently contributes to genome stability [63]. In ESCC, a pyrosequencing assay of LINE-1 methylation levels revealed that ESCC tissues showed significantly lower LINE-1 methylation levels than normal mucosa [64], and the hypomethylation of LINE-1 in ESCC was also proven to be associated with shorter survival, suggesting its potential as a prognostic biomarker [65]. In EAC, researchers assessed hypomethylation of LINE-1 in cell-free DNA (cfDNA) in blood. Furthermore, longitudinal studies in Barret esophagus patients showed that the methylation status of LINE-1 was associated with the progression of EAC, indicating LINE-1 methylation analysis as a novel molecular assay to monitor EAC patients [66]. In OSCC, Foy et al. discovered that patients with LINE-1 hypomethylation had significantly worse oral cancer-free survival and that LINE-1 hypomethylation may be a risk factor for OSCC development in patients with oral premalignant lesions [67]. In addition to LINEs, SINEs are also surrogate markers of global DNA methylation and have been widely studied. Alu, a SINE, is believed to be associated with thyroid cancer progression and dedifferentiation. Researchers have determined that distant metastatic differentiated thyroid cancer, poorly differentiated thyroid cancer, and anaplastic thyroid cancer are increasingly affected by global Alu hypomethylation, indicating the correlation between Alu hypomethylation and thyroid cancer progression and dedifferentiation [68].

4 Unhealthy living habits and HPV infection related to DNA methylation in HNC

4.1 Effects of tobacco use in DNA methylation

Studies have confirmed that tobacco use is among the main risk factors for developing OSCC, and over one-third of oral cancers are attributed to tobacco use [69]. The risk of oral cancer for smokers is two to three times higher than that for nonsmokers [70]. Researchers believe that tobacco-related covalent DNA adducts are major DNA damage caused by tobacco, which exhibits the ability to alter DNA methylation by various mechanisms associated with the formation of DNA lesions [71]. Specifically, promoter/regulatory region DNA hypermethylation and altered methylation patterns in the gene body and/or introns are both related to tobacco use and are associated with cancer progression, lymph node invasion, metastasis and other processes [72,73,74].

4.2 Effects of alcohol abuse in DNA methylation

For alcohol abuse, individual-level pooled data from 17 European and American case‒control studies (11,221 cases and 16,168 controls) showed that the population attributable risk (PAR) for tobacco and alcohol was 72% for head and neck cancer, of which 35% was tobacco and alcohol combined [75]. Recent studies have confirmed that the epigenetic changes induced by alcohol, including DNA methylation, are associated with tumorigenesis [76]. Regarding the mechanisms of the tumorigenic effect of alcohol, DNA methylation is affected by disturbing folate metabolism and transmethylation reactions [71].

4.3 Effects of betel quid chewing in DNA methylation

Betel quid chewing is among the three main risk factors for oral cancer and throat cancer [77]. Lower expression of DNA methyltransferase and gene promoter hypermethylation have both been confirmed in betel quid chewing-related tumors [78]. Toxic chemicals such as 3-(methylnitrosamino) propionitrile (MNPN) were also found in the saliva of betel quid chewers and can promote DNA methylation in the nasal mucosa, esophagus, and liver and thus become a strong carcinogen [79]. More precisely, arecoline in betel quid can induce the hypomethylation of H3K9 and consequently affect the stability of the chromatin structure [80] (Fig. 1).

Fig. 1
figure 1

How unhealthy living habits, including tobacco use, alcohol abuse, and betel quid chewing lead to cancer through DNA hyper/hypomethylation and chromatin instability

Full size image

4.4 HPV and DNA methylation associated with HNC

In the last two decades, HPV-associated HNC, especially oropharyngeal squamous cell carcinoma (OPSCC), has been increasingly drawing public attention, not only because of the rising incidence of HPV-associated OPSCC [81, 82] but also because of the lack of screening instruments for HPV-associated OPSCC, unlike cervical cancer [83]. DNA methylation status and HPV infection status were applied to stratify OPSCC into four epigenotypes, namely, HPV(+) high-methylation (OP1), HPV(+) intermediate-methylation (OP2), HPV(−) intermediate-methylation (OP3) and HPV(−) low-methylation (OP4), showing significantly different prognoses, distinguishing the most favorable OPSCC subgroup (OP1) among generally favorable HPV(+) cases and the most unfavorable OPSCC subgroup (OP3) among generally unfavorable HPV(−) cases [84] Sujita Khanal et al. also revealed that promoter hypermethylation of EREG suppresses its expression, which is associated with the development of HPV-associated HNSCC [85].

5 Methylation enzymes in DNA methylation of HNC patients

The methylation level of DNA is closely related to the expression and activity level of DNA methylation enzymes. Herein, we report two main enzymes, DNA methyltransferases (DNMTs) and ten-eleven translocation (TET), and the mechanisms by which they affect the occurrence and/or progression of HNC (Table 1).

Table 1 Different DNA methylation enzymes and their physical functions
Full size table

5.1 Effects and mechanisms of DNMTs in DNA methylation of HNC

The human genome encodes five DNMTs: DNMT1, DNMT2, DNMT3A, DNMT3B and DNMT3L [88]. Traditional DNMT1, DNMT3A and DNMT3B carry out their functions by catalyzing the transformation of cytosine to 5-methylcytosine, which mostly occurs in CpG islands [89] (Fig. 2). Researchers discovered through immunohistochemical evaluation that the overexpression of DNMT3A may be associated with downregulation of the Klotho gene, which is known as an antiaging gene, in oral cancer tissues [90]. Studies in photoinduced carcinogenesis revealed that DNMT3A could be crucial in the methylation process of UV carcinogenesis present in actinic cheilitis (a lip precancerous lesion), while DNMT3B may play a key role in de novo methylation in established lip cancer [91]. In vitro experiments revealed that, targeted by miR-30a and miR-379, DNMT3B was believed to increase the methylation level of the alcohol dehydrogenase, iron containing 1 (ADHFE1) and aldehyde dehydrogenase 1 family member A2 (ALDH1A2) genes, consequently promoting oral carcinogenesis [92]. DNMT3B also affects the metabolism of cancer cells. Aerobic glycolysis, also known as the Warburg effect, combined with augmented glucose intake and lactate accumulation, is widely involved in the metabolism of cancer cells [93]. To reverse the Warburg effect in HNC cells, DNMT3B was inhibited by employing a dietary phytochemical, thus reducing the methylation level at exon 10, leading to the pyruvate kinase M (PKM, a rate-limiting enzyme in glycolysis)-splicing switch from cancer-specific PKM2 to normal PKM1 [94]. In the field of metastasis of head and neck cancer, researchers revealed that DNMT3B might control the 5’ region of E-cadherin methylation to inhibit E-cad expression and then promote the epithelial-mesenchymal transition (EMT) to contribute to cancer metastasis [95]. DNMT1 mRNA expression analysis revealed that overexpression of DNMT1 was highly associated with the overall survival and relapse-free survival of OSCC patients. Patients with DNMT1 overexpression showed dismal clinical outcomes, with a 2.385-fold higher risk of relapse than those with lower expression [96].

Fig. 2
figure 2

The relationship among tumor suppressor genes (TSGs), cytosine and DNMTs. Cytosines in the promoter region of TSGs transfer to 5-methylcytosine through the catalysis of DNMTs, while inhibitors like 5-aza-2′-deoxycytidine could inhibit the effect of DNMTs

Full size image

5.2 Effects and mechanisms of TET in DNA methylation of HNC

The TET family contains three proteins, TET1, TET2 and TET3. In the family, TET1 was first discovered to be able to modulate methylcytosine and possibly reverse DNA methylation [97] (Fig. 3a) Gradually, researchers revealed that TET1, TET2 and TET3 catalyze the transformation of 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) through an Fe(II) and alpha-ketoglutarate (α-KG)-dependent oxidation reaction [87] (Fig. 3b). These products were regarded as intermediates in the process from 5mC to unmethylated cytosines [98]. The C-terminal region of TET carries out catalyzing functions, while the N-terminal region of TET1 and TET2 is composed of a conserved CXXC region, which recognizes cytosine [99]. More precisely, the catalytic region in the C-terminal is composed of a conserved double-stranded β-helix (DSBH) domain, a cysteine-rich domain, and binding sites for cofactors Fe(II) and 2-oxoglutarate (2-OG) [98]. Researchers revealed that among three TET proteins, only TET1 is downregulated in nasopharyngeal carcinoma (NPC) cells [100]. Ectopic expression of TET1 exerted its tumor suppressor gene activity, including suppressing the growth of NPC cells, inducing apoptosis, and inhibiting cell division, migration and invasion. Further study confirmed that TET1 can demethylate Wnt antagonists (DACT2 and SFRP2) promoters to restore their expression in NPC cells [101]. Retrospective analysis confirmed that TET1 and TET2 expression was significantly reduced in papillary thyroid carcinoma (PTC) tissue. Additionally, the downregulation of the TET2 gene was also confirmed to account for the hypermethylation of the TET2 gene. The studies mentioned above indicated the use of TET1 and TET2 hypermethylation as biomarkers for thyroid carcinogenesis with good sensitivity and specificity [102]. In the field of therapeutics for OSCC, chemotherapy has been expanding significantly in the past few years [103]. However, cancer stem cells may be resistant to chemotherapy drugs, leading to chemotherapy failure [104]. Researchers revealed that the low expression of TET1 induced by TET1-siRNA in OSCC stem cells contributes to the promoter methylation of o6-methylguanine-DNA methyltransferase (MGMT), triggering MGMT mRNA expression inhibition, which ultimately increases the chemotherapeutic sensitivity of OSCC stem cells [105]. Both TET1 and TET2 also share this characteristic. Song. et al. [106] found that TET2 recruited by promyelocytic leukemia (PML) can regulate DNA modification and cell proliferation in response to chemotherapeutic drugs, ultimately affecting overall survival in patients with HNSCC.

Fig. 3
figure 3

A The different structures, effects, and outcomes of TET1, TET2 and TET3. B The process of methylation and demethylation of Cytosine and enzymes involved. BER refers to base-excision repair, TDG refers to thymine DNA glycosylase

Full size image

6 Clinical potentials of DNA methylation

6.1 Diagnosis assistance of DNA methylation status

As a detectable and widely occurring type of pathological change, DNA methylation holds the potential to assist in the diagnosis of head and neck cancer. The methylation status of several target genes combined with other diagnostic examinations may aid in the early diagnosis of head and neck cancer. As mentioned above, hypermethylation of the promoters of p16, PTEN, DAPK, MGMT, RASSF1A and other tumor suppressor genes is believed to be associated with tumorigenesis in head and neck cancer [9, 18, 25, 29, 38, 46]. Meanwhile, genome-wide hypomethylation was thought to influence chromosomal stability and thus may initiate carcinogenesis [51]. Consequently, the detection of methylation status may be able to help distinguish head and neck cancer from tumor-like lesions in the early stage.

6.2 Therapeutic effect prediction of DNA methylation status

By far, multidisciplinary, sequential treatments, including surgery, radiotherapy, chemotherapy, molecular targeted therapy, and integrative immunotherapy approaches, have been carried out clinically [107, 108]. However, the 5-year survival rate for patients with head and neck cancer is only approximately 50%, and locoregional recurrence and metastatic disease still threaten the lives of patients. According to first-line phase II trials exploring taxane/cetuximab combinations, the overall response rate varies from 36 to 54%, which indicates that many patients benefit little from molecularly targeted drugs [109]. In the field of immunotherapy, the low response rate and lack of predictive biomarkers still hamper large-scale clinical applications [109, 110]. Without suitable biomarkers, tumor localization and staging mainly influence therapeutic decisions. Under such circumstances, DNA methylation might provide a novel predictive method, helping screen out the dominant population of molecularly targeted therapy or immunotherapy. Researchers have carried out studies that integrate data across several omic profiles, including DNA methylation (methylome), conducive to patient molecular subgroup identification, therapeutic response prediction and prognosis prediction [111]. The expression patterns of immune checkpoint receptors (including TIM-3, LAG-3, PD-L1 and CTLA-4) were also analyzed to predict therapeutic responses during nimotuzumab therapy and the prognosis of OSCC patients [112]. These studies remind us that single biomarkers are usually insufficient to predict therapeutic response or prognosis, and epigenetic and genetic alterations usually affect the course of diseases together [113]. DNA methylation combined with other prediction biomarkers, such as DNA copy number and miRNA expression, might have higher predictive value and benefit the progress of individual-based treatment (Fig. 4).

Fig. 4
figure 4

The clinical potentials and mechanisms of DNA methylation, including diagnosis assistance as a novel biomarker and therapeutic effect prediction of molecular targeted therapy and immune therapy

Full size image

DNA methylation holds promising potential in head and neck cancer diagnosis and therapy but is still a long way from clinical use. First, the DNA methylation detection method includes methylation-specific PCR (MSP), methylation-specific multiplex ligation-dependent probe amplification, pyrosequencing and other techniques. A sensitive, convenient, and economical detection method needs to be chosen as a standard clinical detection method. Second, a suitable sample also needs to be selected from biopsies, oral rinse, circulating tumor DNA or other sources. Most importantly, the sensitivity and specificity of diagnosis, therapeutic response prediction, and prognosis prediction require further clinical investigation.

6.3 DNA methylation as therapeutic target at present

Previously, we summarized studies that focus on DNA methylation as a therapeutic effect prediction marker or prognosis prediction marker, but there are also some studies that examine therapeutic methods that target DNA methylation directly. To choose a specific drug, single nucleotide polymorphism (SNP) arrays were used to identify genes with aberrant DNA methylation in squamous cell carcinoma as candidate drug targets. Then, candidate drug compounds from the DrugBank database were screened. ChooseLD was used to perform docking between the candidate drug compounds and the proteins related to the genes [114]. Later, silencing DNMT1 in ESCC was conducted both in vivo and in vitro. An in vitro study demonstrated that silencing DNMT1 inhibited the proliferation, metastasis and invasion of three different ESCC cell lines, K150, K410 and K450. An in vivo study showed that silencing DNMT1 suppressed tumor growth in nude mice. Moreover, silencing DNMT1 decreased methylation in the promoter of RASSF1A and DAPK and increased the expression of RASSF1A and DAPK [115]. For patients with thyroid carcinoma refractory to radioiodine treatment, researchers revealed that epigenetic treatments, including 5-azacytidine and valproic acid, can restore radioiodine uptake in anaplastic thyroid carcinoma cells through an ectopic sodium iodine cotransporter, which shed light on the treatment of such thyroid carcinomas [116].

7 Conclusion

DNA methylation is an epigenetic modification that is widely associated with head and neck neoplasms. When DNA hypermethylation occurs in the promoter of tumor suppressor genes, such as p16, PTEN, DAPK, MGMT, ECAD and RASSF1A, the expression of the genes mentioned above decreases significantly and subsequently leads to cancer development or poor prognosis. In contrast, when DNA hypomethylation occurs, chromosomal stability may be influenced, or some metabolic processes, such as glycolysis, may be affected, consequently leading to cancer progression. During the process of DNA methylation, DNA methylation enzymes carry out important functions. DNMTs can catalyze the transformation of cytosine to 5-methylcytosine, while TETs modulate methylcytosine and reverse DNA methylation. Researchers have also determined that external factors, including HPV status, smoking, drinking, and betel quid use, may also influence cancer progression through DNA methylation. Hopefully, elucidation of the interaction between DNA methylation and head and neck neoplasms will provide new therapeutic targets or diagnostic biomarkers, contributing to better and more individualized treatment.