Total m6A RNA levels and VIRMA expression as potential diagnostic and prognostic markers in oral squamous cell carcinoma

Abstract

Background

Oral squamous cell carcinoma (OSCC) is associated with poor prognosis due to extensive local invasiveness and lymph node metastasis, often leading to a significant decrease in aesthetics and function after surgery. Therefore, elucidation of the molecular mechanisms underlying OSCC is necessary for its early detection and treatment. N6-methyladenosine (m6A) modification of mRNA is the most common form of post-transcriptional RNA methylation and is often involved in the progression of cancer by regulating the expression of various genes. Recent studies reported the tumor-promoting effects of vir-like m6A methyltransferase associated (VIRMA, also termed KIAA1429), a novel molecule involved in m6A modification; however, its role in OSCC remains poorly understood.

Methods

In the present study, we determined the total m6A levels and VIRMA expression in OSCC using immunohistochemistry of tissue specimens and evaluated their association with clinicopathologic characteristics. We also performed gene expression analysis of VIRMA/KIAA1429 using public datasets.

Results

We found that the m6A levels were significantly higher in the OSCC specimens of patients with a more advanced clinical stage (P = 0.0063), lymph node metastasis (P = 0.0323), and venous invasion (P = 0.0380) compared to those without. The analysis of the public datasets revealed that VIRMA/KIAA1429 expression levels were higher in head and neck SCC than in normal mucosa, whereas immunohistochemistry revealed that VIRMA-expressing OSCC was associated with a significantly shorter disease-free survival (P = 0.0043) and was an independent poor prognostic factor (P = 0.0179).

Conclusions

These results highlight the potential utility of RNA methylation and VIRMA expression for the diagnosis and treatment of OSCC.

Introduction

Head and neck cancer, including oral squamous cell carcinoma (OSCC), is the sixth most common cancer worldwide, with 58,450 new cases and 12,230 deaths predicted in the USA in 2024 [1]. OSCC is a highly locally invasive cancer with a high rate of lymph node metastasis, explaining its frequent diagnosis in an advanced stage, and the 5-year survival of patients with OSCC has not significantly improved over the past several decades despite the recent, remarkable advances in molecular biology and oncology [2]. Postoperative functional and aesthetic impairment is also a major issue, emphasizing the need for the early detection and treatment of OSCC. However, useful diagnostic markers for OSCC are lacking and only a small number of molecular targeted drugs, such as cetuximab and pembrolizumab, are currently in use for OSCC [3]. The development of molecular diagnostic tools and therapeutic approaches is sorely needed to improve outcomes in patients with OSCC.

Post-transcriptional methylation of RNA regulates gene expression and biologic function, and the addition of N6-methyladenosine (m6A) to mRNA is the most common RNA modification that has been shown to frequently occurs in various cancers [4]. This chemical modification requires enzymes including writers such as methyltransferase-like proteins 3 and 14, which read the modification information, demethylases such as fat mass and obesity-associated protein and alpha-ketoglutarate-dependent dioxygenase alkB homolog 5, and proteins (i.e., readers) such as YT521-B homology (YTH) domain-containing proteins (e.g., YTHDF and YTHDC), which recognize RNA modifications [4, 5]. Studies suggest that the amount of m6A modification is closely associated with progression in a wide range of cancers and that m6A modification is extensively involved in the development, proliferation, and metastasis of OSCC [6–8]. Interestingly, RNA methylation is a regulatory mechanism that is independent on genetic abnormalities [5] and may be responsible for carcinogenesis and cancer heterogeneity that cannot be explained by known genetic abnormalities. Elucidating the role of RNA methylation in carcinogenesis may provide clues to the complex mechanisms underlying cancer and foster the development of new therapeutic approaches in cancer.

Recent studies reported that a novel m6A RNA methyltransferase, vir-like m6A methyltransferase associated (VIRMA, also termed KIAA1429) [9, 10], was involved in the development of various cancers, with some stories showing that the increased expression of VIRMA was associated with m6A modification and that VIRMA exerted a tumor-promoting effect [10–12]. In one study on OSCC, VIRMA was involved in the regulation of ferroptosis, i.e., cell death caused by the iron-dependent accumulation of lipid peroxides, and in aerobic glycolysis (Warburg effect) in a YTHDF1/m6A-dependent manner [11]. Moreover, VIRMA accelerated cell proliferation, invasion, and metastasis in in vitro analyses and in vivo tumor transplantation models of OSCC [13]. Furthermore, the upregulation of VIRMA expression was regulated by the m6A modification of inhibin A in OSCC cell lines [14].

However, high VIRMA expression levels were also reported to enhance the proliferative ability of tumor cells and worsen prognosis independent of the presence of m6A modifications in gastrointestinal [15, 16] and breast cancers [17]. Therefore, many aspects regarding the role of VIRMA in cancer remain unknown. Furthermore, few studies have investigated the significance of VIRMA expression in the clinical specimens of patients with OSCC.

The present study aimed to elucidate the significance of total m6A level and VIRMA expression in human OSCC specimens.

Materials and methods

Public databases

Gene expression data on VIRMA/KIAA1429 were obtained from The Cancer Genome Atlas (TCGA) dataset (https://portal.gdc.cancer.gov), which included the data of 519 head and neck squamous cell carcinoma (HNSCC) patients and 44 healthy individuals registered in the Gene Expression Profiling Interactive Analysis 2 web server (https://gepia2.cancer-pku.cn). The database contained the clinical information of 528 patients, including 382 (72.4%), 116 (22%), and 30 (5.6%) patients with oral cavity, laryngeal, and oropharyngeal cancer, respectively, and the RNA-seq data of 519 patients with at least 30 days of prognostic follow-up [18, 19].

Tumor specimens

In the present study, the formalin-fixed/paraffin-embedded (FFPE) tumor specimens of 69 patients with OSCC (mean age, 70 years; range, 29–91 years) who visited the Department of Oral and Maxillofacial Surgery at Kagoshima University Hospital in Kagoshima, Japan between 2014 and 2016 were used to determine the total m6A levels. Additionally, 64 of the 69 specimens with available data on prognostic were evaluated with immunohistochemistry. Tumor stage and histologic grade were determined according to the 8th edition of the Union for International Cancer Control TNM classification and the World Health Organization criteria, respectively. Medical records were obtained from the patient database managed by the hospital.

The present study was performed according to the ethical standards proposed in the Declaration of Helsinki and was approved by the Medical Ethics Committee of Kagoshima University in Kagoshima, Japan (approval no: 213223-1).

RNA m6A quantification

In FFPE specimens, areas with tumor cells were macrodissected with narrow margin to avoid the inclusion of nontumor cells and total RNA was isolated using the High Pure FFPET RNA isolation kit (Roche Diagnostics, Basel, Switzerland). The m6A modification levels of total RNA were determined using the EpiQuik m6A RNA methylation quantification kit (Epigentek, Farmingdale, NY, USA), according to the manufacturer’s instructions. Briefly, 200 ng total RNA from each specimen was bound to each well and m6A levels were colorimetrically quantified using antibodies in an enzyme-linked immunosorbent assay-like manner, and absorbance was measured at a wavelength of 450 nm. To quantify m6A, a standard curve was prepared and the absolute amount of m6A in each sample was measured. Each experiment was performed in duplicate.

Immunohistochemistry

Consecutive 3-µm-thick sections prepared from FFPE blocks were used for immunohistochemistry. Briefly, after deparaffinization, the endogenous peroxidase activity was blocked with the incubation of sections in methanol containing 0.3% H₂O₂ for 30 min. To block nonspecific antibody binding, the sections were treated with 1% bovine serum albumin (Wako Pure Chemicals, Osaka, Japan) for 10 min. Next, the sections were incubated with a primary antibody against VIRMA (Proteintech, Rosemond, IL, USA) diluted to 0.5 mg/mL in protein-blocking serum at 4 °C for 8 h, followed by incubation with a 0.5% horseradish peroxidase-conjugated secondary antibody (Medical & Biological Laboratories, Nagoya, Japan) at room temperature for 1 h. Color development was achieved using a diaminobenzidine solution (DAKO, Carpinteria, CA, USA), and the sections were counterstained with Mayer’s hematoxylin (Sigma-Aldrich, St. Louis, MO).

Evaluation of immunohistochemistry

Immunostaining for VIRMA was evaluated using proportional and intensity scores, as described in our previous studies [20, 21]. Briefly, the proportional scores were based on the percentage of immunopositive cells and categorized as follows: 0, 0%; 1, 1–10%; 2, 11–35%; 3, 36–70%; and 4, 71–100%. The intensity scores were based on the intensity of immunostaining and categorized as follows: 0, negative; 1, weak; 2, intermediate; and 3, strong. The total score was calculated as the sum of the proportional and intensity scores (range, 0, 2–7), and a total score of ≥ 5 was defined as positive VIRMA immunoreactivity [21].

Statistical analysis

All statistical analyses were performed using JMP version 13 (SAS Institute, Cary, NC, USA). The association of total m6A levels with specific clinicopathologic characteristics was evaluated using Welch’s t test, and the association of VIRMA expression with specific clinicopathologic characteristics was evaluated using Fisher’s exact test. Disease-free survival (DFS) and overall survival (OS) were analyzed using the Kaplan–Meier method, and the differences between groups were evaluated using the log-rank test. Multivariate analysis of prognosis was performed using the Cox proportional hazards model. For all tests, a P value of < 0.05 was considered statistically significant.

Results

Association of total m6A levels with clinicopathologic characteristics of OSCC

First, we determined the association of the total m6A levels in RNA with the clinicopathologic characteristics using the RNA from the FFPE specimens of 69 patients with OSCC. Table 1 shows the details of the study cohort. As shown in Fig. 1, total m6A levels in FFPE specimens were significantly elevated in patients with more advanced clinical stage (P = 0.0063; Fig. 1A), lymph node metastasis (P = 0.0323; Fig. 1B), and venous invasion (P = 0.0380; Fig. 1C). Additionally, m6A levels showed a trend toward being higher in patients with T4-grade disease (P = 0.0518; Fig. 1D) and those with perineural invasion (P = 0.0568; Fig. 1E). No significant associations were found between the total m6A levels and the remaining clinicopathologic characteristics evaluated in the present study.

Table 1.

Breakdown of 64 cases of OSCC

Parameter	Number of patients
Age
≤ 60 years	33 (47.8%)
> 60 years	36 (52.2%)
Gender
Male	35 (50.7%)
Female	34 (49.3%)
Tumor site
Tongue	34 (49.3%)
Others	35 (50.7%)
Histological grade
Well	51 (73.9%)
Mod/Por	18 (26.1%)
T grade
Tis-3	53 (76.8%)
T4	16 (23.2%)
Clinical stage
I, II	48 (69.6%)
III, IV	21 (30.4%)
Nodal metastasis
Negative	54 (78.3%)
Positive	15 (21.7%)

T classification and clinical staging were done according to the TNM classification

*Histological differentiation: Well, well-differentiated squamous cell carcinoma; Mod, moderately differentiated squamous cell carcinoma; Por, poorly differentiated squamous cell carcinoma

Fig. 1.

Association of m6A levels in oral squmous cell carcinoma (OSCC) specimens with clinical stage (A), nodal metastasis (B), venous invasion (C), T grade (D), and perineural invasion (E)

Association of VIRMA/KIAA1429 expression levels in tumor specimens with HNSCC outcomes

Next, we compared the VIRMA/KIAA1429 mRNA expression levels between the tumor specimens of 519 patients with HNSCC and the normal oral tissue specimens of 44 participants included in the TCGA database. As shown in Fig. 2A, the VIRMA/KIAA1429 levels were higher in the HNSCC specimens than in the normal oral tissue specimens. However, the VIRMA/KIAA1429 expression levels did not significantly differ among the different clinical stages of HNSCC (Fig. 2B). By Kaplan–Meier analysis, the DFS was significantly worse in patients with high VIRMA/KIAA1429 expression levels than in those with low VIRMA/KIAA1429 expression levels (P = 0.025) (Fig. 2C). However, no significant association was observed between the VIRMA/KIAA1429 expression levels and OS in the patients with HNSCC (Fig. 2D).

Fig. 2.

Gene expression levels of vir-like m6A methyltransferase associated (VIRMA/KIAA1492) in heand and neck SCC (HNSCC) specimens in a public dataset. (A) Expression levels of VIRMA/KIAA1492 in normal and HNSCC tissues. (B) VIRMA/KIAA1492 expression levels according to the clinical stage of HNSCC. (C, D) Disease-free survival (DFS) (C) and overall survival (OS) (D) according to the VIRMA/KIAA1492 expression level in HNSCC

Immunostaining of VIRMA in the OSCC specimens

Next, we evaluated the VIRMA expression using immunohistochemistry in 64 patients with OSCC. The immunoreactivity of VIRMA was negative or very weak in the noncancerous oral mucosa (Fig. 3A, B), including the minor salivary glands (Fig. 3C), whereas nuclear VIRMA expression was observed in OSCC cells (Fig. 3A, D). In summary, VIRMA positivity in OSCC was 42.2% (27/64) in the study cohort.

Fig. 3.

Immunositaining for VIRMA in the border region between the normal mucosa and OSCC (A), in normal oral mucosa (B), minor salivary glands (C), and OSCC (D). Scale bar, 100 μm

Table 2 shows the relationship between the expression of both molecules and the clinicopathologic characteristics. No significant differences were observed in the clinicopathologic characteristics between the patients with and without VIRMA immunoreactivity. Although the total m6A levels tended to be higher in those with VIRMA immunoreactivity than in those without VIRMA immunoreactivity, the difference was not statistically significant.

Table 2.

Relationship between VIRMA expression and clinicopathological parameters in 64 cases with OSCC

	VIRMA
Paramet	negative	positive	P value
Gender
Male	20	13
Female	17	14	0.8005
Age
<-65	19	10
> 65	18	17	0.3138
Site
Tongue	16	13
Others	21	14	0.8009
Histological grade*
Well	30	20
Mod, Por	7	7	0.5513
T grade
Tis-3	26	17
T4	11	10	0.5964
Clinical stage
I, II	24	12
III, IV	13	15	0.1229
Nodal metastasis
Negative	32	18
Positive	5	9	0.0724
Lymphatic invasion
Negative	35	25
Positive	2	2	1.0000
Venous
Invasion
Negative	29	20
Positive	8	7	0.7692
Perineural
Invasion
Negative	29	23
Positive	8	4	0.5374
Surgical marigin
Negative	30	21
Positive	7	6	0.7627
m6A	0.360 ± 0.241	0.468 ± 0.321	0.2341

Relationship between expression of VIRMA and parameters was calculated by Fisher’s exact test. Correlation between m6A levels and VIRMA expression was determined using Welch’s t test. T classification and clinical staging were done according to the TNM classification

*Histological grade: Well, well-differentiated squamous cell carcinoma; Mod, moderately differentiated squamous cell carcinoma; Por, poorly differentiated squamous cell carcinoma

Finally, the Kaplan–Meier analysis was performed to explore the association of VIRMA immunoreactivity with OS (range, 22–2184 days) and DFS (range, 22–2217 days) in patients with OSCC. During the follow-up period, 14 patients died and 20 patients experienced local or metastatic recurrence of OSCC. The DFS was significantly shorter in patients with VIRMA immunoreactivity than in those without VIRMA immunoreactivity (P = 0.0043, Fig. 4A); however, no significant association was found between VIRMA immunoreactivity and OS (Fig. 4B). By multivariate analysis using the Cox proportional hazards model, which included T classification, clinical stage, lymph node metastasis, surgical margin status, and VIRMA expression as confounding factors associated with prognosis in OSCC, only VIRMA expression was an independent predictor of poor prognosis in patients with OSCC (P = 0.0179, Table 3).

Fig. 4.

Association of VIRMA immunoreactivity with (A) DFS and (B) OS in patients with OSCC

Table 3.

Multivariate analysis of DFS

DFS
Parameters	HR	95% CI	P value*
T factor
Tis-3	1.00
T4	1.1820	0.3470–4.0262	0.7891
Clinical stage
I-II	1.00
III-IV	1.3033	0.3247–5.2313	0.7087
Nodal metastasis
Negative	1.00
Positive	1.7265	0.5421–5.4991	0. 3555
Surgical marigin
Negative	1.00
Positive	1.0620	0.3380–3.3373	0.9179
VIRMA
Negative	1.00
Positive	3.3222	1.3484–9.8451	0.0179

Multivariate analysis was calculated by means of Cox proportional hazard model. HR and 95% CI mean hazard ratio and 95% confidence intervals, respectively

*P value < 0.05 was regarded as statistically significant

Overall, these results suggested the utility of total m6A level and VIRMA expression as diagnostic and therapeutic markers in OSCC.

Discussion

Methylation of the N atom at the sixth position of adenine, i.e., m6A modification, is the most common post-transcriptional mRNA modification, accounting for more than 80% of all RNA methylation [6]. The m6A modification levels in tumors primarily depend on the expression levels of three groups of m6A methylation regulators: writers (methyltransferases), erasers (demethylases), and readers (reading proteins); the expression and functions of these regulators have been studied in many cancer types. However, few studies evaluated the total m6A levels in cancer, particularly in tumor cells isolated from FFPE specimens. In the present study, the m6A levels in the FFPE specimens of OSCC were significantly associated with clinical stage, lymph node metastasis, and venous invasion. Furthermore, the expression of VIRMA, a novel methyltransferase, was higher in OSCC than in normal oral mucosa whereas patients with VIRMA-positive OSCC experienced significantly worse prognosis than those with VIRMA-negative OSCC. Finally, VIRMA expression was an independent poor prognostic factor.

VIRMA contains an RNA-binding domain that allows its binding to target RNA and introduce m6A modifications, which have been shown to promote tumorigenesis by increasing the stability of several oncogenes while decreasing the stability of several tumor suppressor genes [22]. Specifically, in cancer cells, VIRMA upregulates the glycolytic enzymes hexokinase 1 [23], hexokinase 2 [24], and enolase 1 [25] in an m6A-dependent manner, thereby increasing glucose uptake and lactate production and promoting aerobic glycolysis, i.e., the Warburg effect [22]. Moreover, one study suggested that VIRMA acted in an m6A-dependent manner to induce the Warburg effect in OSCC [11]. Normal cells efficiently obtain energy in the form of 2 ATP molecules through anaerobic glycolysis and in the form of 2 ATP molecules through the tricarboxylic acid cycle and 34 ATP molecules through oxidative phosphorylation, both in mitochondria. However, cancer cells obtain energy through the less efficient aerobic glycolysis due to the Warburg effect [26], a hallmark that is considered to promote tumorigenesis [27]. The glycolytic enzyme pyruvate kinase M2 (PKM2) plays an important role in the Warburg effect within cancer cells; we previously reported that PKM2 activation played a role in OSCC proliferation and invasion through the Warburg effect [26]. Additionally, we previously demonstrated that PKM2 activation increased peroxidasin expression in OSCC, leading to an increase in ATP production in glycolysis and contributing to lymph node metastasis and poor prognosis [28]. However, the Warburg effect, which occurs early in cancer development, can also act in a tumor-suppressive manner [29]. In addition, the Warburg effect involving cancer-associated fibroblasts rather than cancer cells, i.e., the reverse Warburg effect, provides lactic acid and pyruvate as high-energy metabolites that are taken up by cancer cells [30]. Therefore, many aspects of the Warburg effect in cancer cells remain unknown. Future studies should evaluate the role of VIRMA and energy metabolism in OSCC, given that the Warburg effect is an important characteristic of cancer [27].

Recent advances in liquid biopsy have allowed the analysis of circulating tumor cells and exosomes in peripheral blood [31]. In particular, exosomes released from cancer cells contain important nucleic acid information, which is important for intercellular communication and plays a pivotal role in the formation of the tumor microenvironment and metastasis [32]. In this context, the m6A levels of exosomal mRNA in the peripheral blood may be a useful diagnostic and predictive marker for the progression and metastasis of OSCC, which requires further investigation.

A recent study demonstrated that the poly ADP-ribose polymerase inhibitor rucaparib downregulated VIRMA expression in chronic myeloid leukemia, suppressing cancer proliferation and enhancing sensitivity to imatinib [33]. VIRMA has also been reported to reduce radiosensitivity by inhibiting ferroptosis in colorectal cancer cells [34]. These emerging lines of evidence highlight the potential utility of VIRMA as a target for cancer therapy. Furthermore, given that m6A modifications are dynamic and reversible [6], it may be possible to normalize cancerous cells by removing the methylation state as a molecular targeted therapy approach with potential clinical applicability. As mentioned above, cetuximab and pembrolizumab are the only useful molecular targeting agents for OSCC in Japan [3], but new OSCC therapies targeting VIRMA and m6A modifications are expected to be developed. Large-scale global clinical trials and preclinical studies with appropriate animal models are necessary to develop clinical applications utilizing RNA methylation and VIRMA as diagnostic and therapeutic targets.

Conclusions

In conclusion, we demonstrated the significance of RNA methylation and VIRMA expression in OSCC specimens. The present study has several limitations. First, the quality of the RNA in the FFPE sample will likely affect the quantification of m6A, but one of the limitations of this study is that the degree of RNA degradation was not considered. Second, the nonsignificant difference in m6A levels observed between the VIRMA-positive and VIRMA-negative OSCC cases might be related to the limited RNA yield of the FFPE specimens. However, studies have also reported the VIRMA functions independently of m6A methylation [15–17]. In future studies, we will investigate the relationship between RNA methylation and VIRMA in total RNA extracted from frozen OSCC specimens. Further, it would be important to identify genes targeted by VIRMA, if it functions in an m6A-dependent manner in OSCC. Despite the many questions that should be explored, our findings suggest the potential utility of RNA methylation and VIRMA as markers for predicting the progression, metastasis, and prognosis of OSCC as well as targets for diagnosis and therapy.

Abbreviations

DFS

Disease-free survival

FFPE

Formalin-fixed/paraffin-embedded

HNSCC

Head and neck squamous cell carcinoma

m6A

N6-methyladenosine

OS

Overall survival

OSCC

Oral squamous cell carcinoma

TCGA

The Cancer Genome Atlas

VIRMA

Vir-like m6A methyltransferase associated

YTH

YT521-B homology

Author contributions

Kaori Shima: Conceptualization, Investigation, Data curation, Software. Yudai Shimojukkoku: Investigation, Data curation, Yasunobu Oku: Investigation. Kanako Higashimoto: Investigation. Takahiro Tsuchiyama: Investigation. Yuka Kajiya: Investigation, Data curation, Miyako Kurihara-Shimomura: Writing– review & editing, Funding acquisition, Supervision. Tomonori Sasahira: Writing– original draft, Funding acquisition, Conceptualization, Supervision, Project administration.

Funding

This study was partially supported by (1) Grant-in-Aid for Scientific Research from Japan Society for the Promotion of Science to T.S. (Grant Number: 20K09912, 24K02647) and M.K.S. (Grant Number: 21K10077); (2) Takahashi Industrial and Economic Research foundation to T.S.

Data availability

No datasets were generated or analysed during the current study.

Declartaions

Ethics approval and consent for participate

The present study was in accordance with the Ethics Committee on Epidemiological Studies at Kagoshima University [Approval No. 213223-1] and was performed in accordance with the Helsinki declaration.

Consent for publication

Informed consent was obtained from all individual participants included in the study.

Competing interests

The authors declare no competing interests.