Background
Based on incidence and mortality, head and neck squamous cell carcinoma (HNSCC) is the 6
th leading cancer worldwide with an estimated 900,000 newly diagnosed cases each year [
1]. The major risk factors for HNSCC include tobacco and alcohol, which account for at least 75 % of all cases diagnosed in Europe, the United States, and other industrialized regions [
2]. Tobacco use and alcohol consumption are two habits strongly linked with each other that have been shown to act synergistically in HNSCC, but many studies have also demonstrated each substance to be an independent risk factor [
2‐
5]. Despite this knowledge, there is little evidence that people have modified their alcohol intake, as alcohol consumption has remained high throughout the years whereas tobacco usage has steadily declined [
3,
4]. Meanwhile, a molecular understanding of the pathogenesis of alcohol-related HNSCC remains elusive and poorly understood. Since ethanol alone is not considered a carcinogen, most studies involving alcohol and cancer focus on its ability to increase penetration of carcinogens, interfere with DNA repair mechanisms, or cause DNA damage through acetaldehyde, the first metabolite of ethanol and an established carcinogen [
6,
7].
Recently, numerous studies have revealed that microRNAs (miRNAs) play a role in the initiation and progression of many cancers, including HNSCC. miRNAs are a class of non-coding RNAs 19–25 nucleotides in length that can regulate gene expression by binding to the partially complementary regions of messenger RNA targets. It is proposed that miRNAs regulate up to a third of protein coding genes, in turn impacting myriad cellular processes including cellular growth, differentiation, self-renewal, apoptosis, survival, and migration [
8,
9]. Analyzing the expression of miRNAs in cancer compared to normal tissue and investigating the functional significance of their dysregulation may be pivotal for improving early diagnosis, predicting prognosis, and establishing specific therapeutics. Several studies have previously revealed aberrations in miRNA expression in HNSCC tumors [
10‐
12], but conclusive biomarkers have yet to be established [
8]. Furthermore, there is little evidence about how miRNA dysregulation relates to many HNSCC clinical risk factors, which may prove vital for a clearer molecular understanding of how these factors play a role in carcinogenesis. One previous study analyzed the association between four pre-selected miRNAs and certain clinical features and risk factors, including alcohol consumption. Using multivariate comparisons, it found that
miR-375 was associated with tumor site, stage, and alcohol consumption in HNSCC [
13]. Our study aimed to expand current understanding of the link between alcohol consumption and miRNA expression by using next-generation RNA-sequencing data from 136 HNSCC patients to identify differentially expressed candidates among 1,046 annotated miRNAs. We subsequently investigated the alcohol-associated miRNAs in normal oral keratinocyte cell cultures in order to establish if these miRNAs may be crucial for their malignant transformation. After selecting the miRNAs that were the most dysregulated by alcohol and acetaldehyde, we evaluated how modulation of their expression would induce changes in cellular proliferation, sensitivity to cisplatin, and invasion, and finally sought to identify their mRNA targets The results of this study demonstrate that alcohol consumption regulates several miRNAs that likely play significant roles in the alcohol-associated carcinogenesis of HNSCC.
Discussion
Although alcohol consumption has been identified as a potent risk factor for HNSCC, knowledge regarding its functional role as an independent etiological agent for the disease remains significantly limited. Multiple studies have proposed that the action of alcohol in HNSCC pathogenesis is mediated by its first metabolite acetaldehyde, a well-established carcinogen [
15], but others have postulated that alcohol may exert adverse effects directly, either by increasing the cellular uptake and potency of carcinogens such as tobacco [
16], or by inducing DNA hypomethylation [
17]. The relevance of these proposed models to alcohol-associated HNSCC and the molecular mechanisms by which they may occur have remained obscure; previous investigations of the effects of environmental factors on genetic and epigenetic alterations in HNSCC have largely focused on characterizing tobacco, HPV exposure, or tobacco in association with alcohol [
1,
18,
19].
We are the first to globally profile alcohol-induced miRNA dysregulation in HNSCC and to functionally characterize the roles that these miRNAs may play in HNSCC pathogenesis and progression. Through RNA-sequencing analysis of miRNA expression in 136 patients, we identified a panel of 8 miRNAs (miR-3178, miR-675, miR-934, miR-101, miR-1266, miR-30a, miR-3164, and miR-3690) exhibiting significant upregulation in alcohol-associated HNSCCs, suggesting their role as oncomirs. By subdividing our patient cohort into smokers and non-smokers, and comparing the drinking and non-drinking populations within both groups for unifying miRNA aberrations, we ensured that these miRNA candidates were dysregulated due to alcohol consumption alone, removing the synergistic and likely confounding influence of tobacco.
None of the 8 miRNAs have been previously identified in studies of tobacco or HPV-induced miRNA dysregulation in HNSCC [
1]. Furthermore,
miR-3164 and
miR-3690 demonstrate no previous connections to cancer, establishing them as potentially fruitful targets for future investigation. Such findings underscore existing evidence of considerable heterogeneity in the genetic and transcriptional landscapes of HNSCCs associated with different etiologies [
1,
18], and pose implications for future distinctions in the diagnostic strategies and clinical management of alcohol-associated HNSCC.
Interestingly, previous studies of
miR-1266 and
miR-3678 characterize their roles as tumor suppressor miRNAs in gastric cancer.
miR-1266 was downregulated in gastric cancer tissues and was shown to inhibit tumor growth and invasion by targeting telomerase reverse transcriptase (hTERT) [
20], while
miR-3678 was found to be significantly downregulated in lymphatic metastases [
21]. Meanwhile, there exists conflicting evidence regarding the roles of
miR-675 and
miR-101 in various malignancies.
miR-675, along with its precursor long non-coding RNA
H19, was found to inhibit invasion and metastasis in hepatocellular carcinomas [
22], but was shown to promote the pathogenesis and metastasis of gastric cancer and target the tumor suppressor retinoblastoma in colorectal cancer [
23,
24]. Similarly, studies in a number of cancers have found
miR-101 to function as a tumor suppressor by targeting
EZH2 [
24,
25], yet it was also implicated as a promoter of growth in estrogen receptor (ER)-positive breast tumors by inducing the activation of Akt [
25]. While further investigation of these discrepancies may be necessary, there is mounting evidence that miRNAs may assume tissue-specific or cancer-specific targets and functions via distinct pathways [
26,
27]. Therefore, future characterizations of the miRNAs we have identified may find that they indeed exert unique effects in the contexts of HNSCC and alcohol-induced dysregulation.
We found
miR-30a,
miR-934,
miR-3164, and
miR-3178 to be consistently upregulated in oral keratinocytes and HNSCC cell lines following treatments with ethanol and acetaldehyde, demonstrating that dysregulation of these miRNAs is directly mediated by exposure to alcohol and is likely involved in both the early stages of malignant transformation and the progression to increased malignancy in HNSCC. Notably, acetaldehyde alone also induced aberrant expression of the miRNAs, implying that its carcinogenic properties are not only the result of DNA cross-links and chromosomal aberrations as previously characterized [
6], but also due to its modulation of miRNA-mediated pathways. These findings also confirm that ethanol-to-acetaldehyde metabolism is responsible, at least in part, for alcohol-related HNSCC pathogenesis. Interestingly, however, the 4 miRNAs failed to exhibit proportional increases in expression in response to increased ethanol or acetaldehyde dosage, with strongest upregulation generally occurring at the lowest ethanol (0.1 %) and acetaldehyde (75 μM) concentrations. This may simply indicate that alcohol-mediated miRNA dysregulation is governed by factors other than intake levels; a previous study of the relationship between carcinogen exposure and homozygous deletion of
p16
INK4A
in HNSCCs found that the duration, rather than intensity, of tobacco and alcohol use significantly predicted deletion frequency [
28]. Further investigation of the relationship between length of alcohol consumption and miRNA expression, in conjunction with the
in vitro experiments presented here simulating long-term alcohol exposure, may therefore be valuable in characterizing more precisely the conditions under which alcohol-induced miRNA dysregulation occurs.
In particular, we show that enforced expression of
miR-30a and
miR-934 in HNSCC cell lines promotes the induction of anti-apoptotic gene
BCL-2, as well as increased cellular proliferation in the HNSCC cell lines. Conversely, inhibition of the miRNAs in HNSCC resulted in reduced invasion, as well as decreased resistance to cisplatin. The potential of these alcohol-dysregulated miRNAs to confer proliferative and invasive properties to HNSCC cells may significantly account for the elevated risks of developing HNSCC and experiencing tumor recurrence that have been observed in drinkers [
2‐
5]. Furthermore, the ability of miRNA suppression to sensitize cells to cisplatin and decrease invasive capabilities establishes
miR-30a and
miR-934 as potential targets of future therapies specifically tailored to alcohol-related HNSCC. Our findings are consistent with previous studies reporting the ability of
miR-934 and
miR-30a to promote metastasis as well:
miR-934, along with host gene
VGLL1, was found to be upregulated in ER-negative
BRCA1-positive breast cancers and was associated with the acquisition and maintenance of luminal progenitor characteristics [
29]; similarly,
miR-30a was shown to promote invasion in nasopharyngeal carcinoma (NPC) via the repression of E-cadherin and induction of epithelial-to-mesenchymal transition (EMT) [
30].
The putative target genes we have shown to be modulated by both miRNAs further corroborate their proposed roles in HNSCC.
BNIP3L, potential target of
miR-30a, is a proapoptotic gene activated by p53 under hypoxic conditions [
31,
32]. Meanwhile,
SEPT7 is a cell cycle regulator involved in kinetochore localization and apoptosis [
33,
34], and
PRDM1 is a transcriptional repressor and inhibitor of the Wnt signaling pathway [
35], which has been shown to be centrally associated with stemness and invasiveness in HNSCCs [
36].
Both
SEPT7 and
PRDM1 have been found to be targeted by
miR-30a in gliomas [
35,
37]. The proposed targets for
miR-934 are also potent tumor suppressors:
HOXA4 has been shown to inhibit cell motility and invasion in ovarian carcinoma [
38];
MLL3 is involved in a p53 coactivator complex, with knockout resulting in epithelial tumor formation in mice [
39]; and
HIPK2 mediates apoptosis and activates p53 via phosphorylation at Ser46 [
40].
Methods
miRNA-sequencing datasets and differential expression analysis
Level 3-normalized miRNA expression datasets and clinical data for 136 HNSCC patients were obtained from The Cancer Genome Atlas (TCGA) (
https://tcga-data.nci.nih.gov/tcga).These patients were separated into four cohorts based on reported alcohol and tobacco use, in order to minimize the confounding influence of smoking on HNSCC prevalence and pathogenesis: (1) nonsmoking non-drinkers, (2) nonsmoking drinkers, (3) smoking non-drinkers, and (4) smoking drinkers (Table
1). Statistical analyses were performed using IBM SPSS Statistics v.22. To account for unequal variance, two-tailed Welch’s t-tests were used for the differential expression analyses. Only p-values below 0.05 were considered statistically significant.
Survival analysis
42 HNSCC patients in the TCGA database had available data for both alcohol consumption and length of survival. All of these patients were smokers and were subdivided into non-drinkers, light drinkers (1–3 drinks per day) and heavy drinkers (more than 3 drinks per day). Kaplan-Meier plots were graphed using IBM SPSS Statistics v.22 and a log-rank test was used to calculate the p-value. Only p-values <0.05 were considered statistically significant.
Cell culture and treatment with ethanol and acetaldehyde
Normal, early passage, oral epithelial cell linesOKF4 and OKF6 (derived from the floor of the mouth) were gifts from the Rheinwald Lab at Harvard Medical School. The cells were cultured in keratinocyte serum-free media (Life Technologies) supplemented with EGF, bovine pituitary extract, 2 % L-glutamine, 2 % penicillin/streptomycin, and CaCl2. These cells were either exposed to ethanol for 1 week or 4 weeks, or to acetaldehyde (Sigma Aldrich) for 48 h.
The HNSCC cell lines UMSCC-10B and UMSCC-22B (derived from laryngeal and hypopharyngeal tumors, respectively) were gifts from Dr. Tom Carey, University of Michigan. The cells were cultured in DMEM supplemented with 10 % fetal bovine serum, 2 % penicillin/streptomycin, and 2 % L-glutamate (GIBCO) and maintained at 37 °C in a humidified 5 % CO2/95 % air atmosphere. These cells were either exposed to ethanol for 1, 2, or 4 weeks, or to acetaldehyde for 48 h.
The doses used for ethanol treatment were 0.1 %, 0.3 %, and 1 % by volume (approximate concentrations 17 mM, 51 mM, and 170 mM, respectively). We chose the 0.1 % (17 mM) dose to represent social drinking habits, as 0.1 % is the blood alcohol level constituting legal intoxication in the U.S. [
41]. The 0.3 % (51 mM) ethanol dose was used to simulate binge drinking habits, as it is representative of the blood alcohol levels of moderate to heavy drinkers [
42]. The 1 % (170 mM) ethanol dose, while potentially lethal in humans, was employed as an upper limit control as it was the highest concentration that was minimally toxic to the oral keratinocytes (Additional file
1: Figure S1). For all ethanol-culture experiments, treatment media was replaced every 24 h with fresh media containing the stated ethanol concentration. The tissue culture plates were also sealed with paraffin film to reduce evaporative loss of ethanol from the media. It has been shown that sealed culture vessels are able to maintain ethanol concentrations over significantly longer incubation periods [
43].
The doses used for acetaldehyde treatment were 75 μM, 150 μM, and 300 μM, to represent a range of light to heavy drinking as determined by the saliva concentrations of alcohol consumers [
14]. To account for the short evaporation half-life of acetaldehyde, treatments were performed every 4 h and the tissue culture plates were also sealed with paraffin film.
microRNA expression profiling by qRT-PCR
Total RNA was isolated (Fisher Scientific) from cultured cells after their respective treatments with ethanol or acetaldehyde. cDNA was synthesized using the QuantiMiRTM RT kit (System Biosciences, Mountain View, CA) as per the manufacturer’s instructions. Real-time PCR reaction mixes were created using FastStart Universal SYBR Green Master Mix (Roche Diagnostics), and run on a StepOnePlusTM Real-Time PCR System (Applied Biosystems) using the following program: 50 °C for 2 min, 95 °C for 10 min, 95 °C for 30 s, and 60 °C for 1 min, for 40 cycles. Experiments were analyzed using the ddCt method. U6 primers and a Universal Reverse Primer were used from the QuantiMiRTM RT kit, and custom primers (Eurofins MWG Operon) were ordered using the following sequences: miR-3178: 5’-GGGGCGCGGCCGGATCG-3’, miR-675: 5’-TGGTGCGGAGAGGGCCCACAGTG-3’, miR-934: 5’-TGTCTACTACTGGAGACACTGG-3’, miR-101: 5’-CAGTTATCACAGTGCTGATGCT-3’, miR-1266: 5’-CCTCAGGGCTGTAGAACAGGGCT-3’, miR-30a: 5’-TGTAAACATCCTCGACTGGAAG-3’, miR-3164: 5’-TGTGACTTTAAGGGAAATGGCGAA-3’, and miR-3690: 5’-ACCTGGACCCAGCGTAGACAAAG-3’.
microRNA plasmid and siRNA transfections
Expression plasmids for human miR-30a and miR-934 (OriGene Technologies, Rockville, MD) were transiently transfected using Lipofectamine 2000 (Invitrogen, Carlsbad, CA), following the manufacturer’s specifications. The pCMV-MIR empty vector was used as control and transfection efficiency for all three plasmids was monitored using GFP as a reporter.
Knockdown of both miRNAs was performed using Anti-miRTM miRNA inhibitors (Ambion, Austin, TX) that were specific for miR-30a-5p and miR-934. These inhibitors were also transfected with Lipofectamine 2000.
qRT-PCR for BCL-2 gene expression
Total RNA was collected 48 h after transfection with miRNA expression plasmids or inhibitors (Fisher Scientific). cDNA was synthesized using Superscript III Reverse Transcriptase (Invitrogen, Carlsbad, CA) as per the manufacturer’s instructions. Real-time PCR reaction mixes were created using FastStart Universal SYBR Green Master Mix (Roche Diagnostics), and run on a StepOnePlusTM Real-Time PCR System (Applied Biosystems) following the previously described program. Results were analyzed using the ddCt method, with GAPDH expression as the endogenous control. Primers were custom ordered (Eurofins MWG Operon, Huntsville, AL) using the following sequences: GAPDH forward: 5′-CTTCGCTCTCTGCTCCTCC-3′, reverse: 5′-CAATACGACCAAATCCGTTG −3′. BCL-2 forward: 5’-CCTGTGGATGACTGAGTACC-3’, reverse: 5’-CCTGTGGATGACTGAGTACC-3’.
Alamar blue cell viability assay
UMSCC-10B and UMSCC-22B cells were plated into a 96-well flat-bottom tissue culture plate (Falcon) at a density of 2,000 cells per well. After a 24 h plating period, these cells were transfected with the miRNA expression plasmids. After a 48 h incubation period, cellular proliferation was analyzed using Alamar blue reagent (ThermoFisher) in accordance with the manufacturer's protocol. All assays were performed in triplicate wells and experiments were individually performed at least twice.
MTS cell proliferation assay
UMSCC-10B and UMSCC-22B cells were plated into a 96-well flat-bottom tissue culture plate (Falcon) at a density of 5,000 cells per well. After a 24 h plating period, these cells were transfected with the miRNA expression plasmids or inhibitors. For cisplatin sensitivity experiments, the control and transfected cells were subsequently exposed to one of several doses of cisplatin ranging from 0–5 μg/mL. After a 48 h incubation period, cellular proliferation was analyzed using an MTS proliferation assay (Promega) in accordance with the manufacturer's protocol. All assays were performed in triplicate wells and experiments were individually performed at least twice.
Matrigel invasion assay
Invasion of UMSCC-10B and UMSCC-22B cells was measured using a Matrigel invasion assay (Becton Dickinson, Bedford, MA). 48 h post transfection with either the expression plasmid or inhibitor, cells were trypsinized, and 500 μL of cell suspension (1 × 105 cells/mL) was added in triplicate wells. The lower chamber of the transwell was filled with 750 μl of culture media containing 0.5 % serum as a chemoattractant and allowed to incubate at 37 °C for 24 h. Invading cells on the lower surface that passed through the filter were fixed and stained using crystal violet in gluteraldehyde and photographed. The number of the stained nuclei was counted in a predetermined and consistent section of each well.
miRNA target gene identification and qRT-PCR
Total RNA was collected from UMSCC-10B and UMSCC-22B 48 h after transfection with miR-30a and miR-934 inhibitors. cDNA was synthesized using SuperScript III Reverse Transcriptase and qRT-PCRs were performed as previously described, with GAPDH serving as endogenous control. Primers for the target genes were custom synthesized (Eurofins MWG Operon) using the following sequences: GAPDH forward: 5′-CTTCGCTCTCTGCTCCTCC-3′, reverse: 5′-CAATACGACCAAATCCGTTG −3′. BNIP3L forward: 5’- GGACTCGGCTTGTTGTGTTG-3’, reverse: 5’-AGACTGCTCATTTTCCTCGCA-3’. PRDM1 forward: 5’-TAACACAGACAAAGTGCTGCC-3’, reverse: 5’-CTACCCAGTCCACATTCTCCC-3’. SEPT7 forward: 5’-ATTCACGCTTATGGTAGTGGG-3’, reverse: 5’-AGCAACTGAACACCACCTTCT-3’. HIPK2 forward: 5’- CACAGGCTCAAGATGGCAGA-3’, reverse: 5’-GGGATGTTCTTGCTCTGGCT-3’. HOXA4 forward: 5’-AGAAGATCCATGTCAGCGCC-3’, reverse: 5’-TGTTGGGCAGTTTGTGGTCT-3’. MLL3 forward: 5’- CTGCCAAAGGAGACTCAGGG-3’, reverse: 5’- GTCCGTTTGCTTCGCTGTTT-3’.
Competing interests
The authors declare that there are no conflicts of interest.
Authors’ contributions
MAS performed cell culture and treatment with ethanol and acetaldehyde, microRNA expression profiling by qRT-PCR, microRNA and siRNA transfections, qRT-PCR for putative mRNA targets, MTS proliferation assay, and assisted in drafting the manuscript. SZK performed miRNA sequencing and survival analysis, plasmid and siRNA transfections, and MTS proliferation assay. ER performed miRNA sequencing analysis and assisted in designing and coordinating in-vitro experiments. AEZ analyzed miRNA sequencing data and analyzed putative mRNA targets, designed appropriate primers, and assisted in drafting the manuscript. AK performed microRNA expression profiling by qRT-PCR. MR performed qRT-PCR for putative mRNA targets. EK completed matrigel invasion assay. HZ performed miRNA sequencing analysis. MAY performed cell culture and ethanol treatments, and MTS proliferation assay. JWR conceived andcoordinated the study. WMO conceived the study, designed and coordinated the experiments, and wrote the manuscript. All authors read and approved the final manuscript.