Aberrant DNA methylation of miR-219 promoter in long-term night shiftworkers

Authors


Correspondence to: Yong Zhu, Department of Environmental Health Sciences, Yale University School of Public Health, New Haven, CT. E-mail: yong.zhu@yale.edu

Abstract

The idea that shiftwork may be carcinogenic in humans has gained widespread attention since the pioneering work linking shiftwork to breast cancer over two decades ago. However, the biomolecular consequences of long-term shiftwork exposure have not been fully explored. In this study, we performed a genome-wide CpG island methylation assay of microRNA (miRNA) promoters in long-term night shiftworkers and day workers. This analysis indicated that 50 CpG loci corresponding to 31 miRNAs were differentially methylated in night shiftworkers compared to day workers, including the circadian-relevant miR-219, the expression of which has been implicated in several cancers. A genome-wide expression microarray assay was carried out in a miR-219-overexpressed MCF-7 breast cancer cell line, which identified 319 differentially expressed transcripts. The identified transcriptional targets were analyzed for network and functional interrelatedness using the Ingenuity Pathway Analysis (IPA) software. Overexpression of miR-219 in MCF-7 breast cancer cells resulted in accentuated expression of apoptosis- and proliferation-related anti-viral immunodulators of the Jak-STAT and NF-κβ pathways. These findings suggest that long-term night shiftwork exposure may lead to the methylation-dependent downregulation of miR-219, which may in turn lead to the downregulation of immunomediated antitumor activity and increased breast cancer risk. Environ. Mol. Mutagen. 54:406–413, 2013. © 2013 Wiley Periodicals, Inc.

Abbreviations
miRNAs

microRNAs

GEO

Gene Expression Omnibus

HCC

hepatocellular carcinoma

IPA

Ingenuity Pathway Analysis

INTRODUCTION

Shiftwork exposure has been hypothesized to increase the risk of chronic diseases, including cancer, cardiovascular disease, metabolic syndrome, and diabetes [Wang et al., 2011]. Multiple epidemiological studies have reported an increased risk of breast cancer in women working night or evening shifts [Pronk et al., 2010; Reed, 2011; Hansen and Stevens, 2012]. Although several theories have been put forth to explain the link between shiftwork and chronic diseases, including suppression of melatonin production and deregulation of circadian genes involved in cancer-related pathways, the influence of epigenetic modulators of gene expression, including DNA methylation and microRNA-induced gene silencing, remains to be elucidated.

MicroRNAs (miRNAs) are a class of endogenous small noncoding RNAs that negatively regulate gene expression by inducing degradation or translational inhibition of target mRNAs. miRNAs are involved in the control of several cellular processes altered in cancer, including proliferation, differentiation, and apoptosis [Croce and Calin, 2005]. More than 100 miRNAs have been found to be regulated by epigenetic mechanisms, and about half of them are known to be methylated in a cancer-specific manner in 23 different tumor types [Kunej et al., 2011]. There is also compelling evidence to suggest that miRNAs have essential roles to play in the fine tuning of rhythmic outputs from the central circadian clock [Kadener et al., 2009; Liu and Wang, 2012; Luo and Sehgal, 2012].

Alterations of the patterns of normal DNA methylation have been linked to many human diseases, including cancer [Hinoue et al., 2012; Rauch et al., 2012; Xu et al., 2012]. Moreover, there is new evidence implicating shiftwork-induced methylation changes at the core circadian genes CLOCK and CRY2, as well as a host of other genes implicated in carcinogenesis [Zhu et al., 2011; Jacobs et al., 2013]. It thus stands to reason that exposure to prolonged shiftwork may also lead to the epigenetic dysregulation of miRNAs with functional relevance in cancer- and circadian-related processes. In the current pilot study, we test the hypothesis that aberrant methylation events in the promoter regions of specific miRNA precursors can be induced by long-term night shiftwork and lead to the development of circadian-related diseases.

MATERIALS AND METHODS

Study Population

Our study is based on the Danish “Diet, Cancer and Health” prospective cohort, established between December 1993 and May 1997. All female participants were born in Denmark, free of any cancer and aged between 50 and 64 years at time of invitation (now between 65- and 79-years-old). The baseline questionnaire included questions on food consumption, folate intake, other lifestyle factors (e.g., tobacco smoking, alcohol habits, sun exposure, physical activity, and medical anamnesis), reproductive factors, education and occupation. All participants gave written informed consent. All collected biological samples were frozen and kept at −150°C [Tjonneland et al., 2007]. The season of blood collection did not differ between night shiftworkers and day workers (P = 0.162). Information on the characteristics of the study subjects are summarized in Table 1.

Table 1. Description of Subject Characteristicsa, b
CharacteristicDay workers (n = 10)Shiftworkers (n = 10)P-valuec
  1. a

    Tables values are given as mean ± SD.

  2. b

    All subjects are female and of non-Hispanic ethnicity.

  3. c

    P-value of independent t-test.

Age (years)54.0 ± 3.354.8 ± 3.60.610
Folate intake (µg/day)373.1 ± 108.4349.9 ± 125.10.664
Tobacco smoking (years)17.7 ± 15.218.0 ± 17.60.968
Cumulative alcohol intake (g)93,877.4 ± 68,517.9114,622.9 ± 62,902.00.490
Hormone replacement therapy (years)2.8 ± 4.83.65 ± 4.00.673
Years of shiftwork0 ± 0.021.2 ± 8.8<0.001

Collection of Shiftwork Information

Night work was defined as having worked fulltime between 7 pm and 9 am. Information on night shiftwork status was obtained by telephone interview. All study subjects were classified into 3 groups: women with 10+ years of night shiftwork, women with <10 years of night shiftwork, and daytime workers.

Collection of Folate Intake Information

Folate intake data were collected at baseline using a 192-item food-frequency questionnaire mailed to each participant. 12 categories, from “never” to “eight times or more per day,” were used to estimate the average intake of each food item during the last 12 months. Open-ended questions were designed for supplement use on brands and doses. Information on folate content in each supplement brand was obtained from producers or distributors. For each participant, average daily folate intake was calculated using the software program Food Calc (www.ibt.ku.dk/jesper/foodcalc) as total, dietary, and supplemental intake.

Genome-Wide CpG Island Methylation Assay of microRNA Promoters

10 pairs of long-term night shiftworkers and day workers were selected from the study population matched by age (±2 years) and total folate intake (±55 μg/day). Genomic DNA was isolated and purified from blood samples using the QIAamp DNA Blood Mini Kit (Qiagen) according to the manufacturer's protocol. 50 ng of genomic DNA per sample were bisulfite treated using the EZ DNA Methylation Kit (Zymo research) according to the manufacturer's protocol. CpG island methylation of miRNA promoter regions was determined using the Illumina Infinium Methylation Assay, covering 255 CpG sites of 110 microRNAs. A methylation index (β) was used to estimate the methylation level of each CpG locus using the ratio of intensities between methylated and unmethylated cytosines, which is a continuous variable between 0 and 1. 0 corresponds to a completely unmethylated site, while 1 corresponds to a completely methylated site. The Illumina Custom Model, as implemented in the Illumina GenomeStudio software, was used to assess the statistical significance of methylation differences between long-term shiftworkers and day workers at each CpG site. CpG loci were considered to exhibit differential methylation if they fit the criteria of P <  0.05, as derived from DiffScore values generated by the Illumina Custom Model.

Cell Culture and Construction of miR-219-Overexpressed Cell Model

Genome-wide expression microarray analysis was carried out in a miR-219-overexpressed breast cancer cell line. MCF-7 cells were cultured in DMEM supplemented with 10% fetal bovine serum, 1% penicillin-streptomycin (Invitrogen) and 0.01 mg/ml insulin (Sigma-Aldrich). Cells were incubated at 37°C and 5% CO2 and passaged every three days. Transfection of the miR-219 mimic (Qiagen) was performed using Lipofectamine™ RNAiMAX (Invitrogen). AllStars Negative Control siRNA (Qiagen) was used to ensure minimal nonspecific effects on gene expression and phenotype as it exhibits no homology to any known mammalian gene. Transfection efficiencies were determined using Cy3 DS Transfection Control (Integrated DNA Technologies) followed by fluorescence microscopy (Supporting Information Fig. S1). After incubation at 37°C and 5% CO2 for 48 hr, total RNA samples were isolated using the miRNeasy Mini Kit (Qiagen) with on-column DNA digestion.

Quantitative RT-PCR Detection of miR-219

Quantitative reverse transcriptase PCR analysis of miR-219 expression was carried out to evaluate the transfection effect of the miR-219 mimic in MCF-7 cells using the NCode™ miRNA First-Strand cDNA Synthesis Kit (Invitrogen) and KAPA™ SYBR FAST qPCR Kit (Kapa Biosystems) on a 7500 Real Time PCR System (Applied Biosystems). miR-16 was used as an internal control to normalize RNA input. The primers used for detection of miR-219 and miR-16 were CAATCACTAACTCCACTGCCAT and TAGCAGCACGTAAATATTGGCG, respectively. Relative miR-219 abundance was assessed using a comparative Ct method as described elsewhere [Livak and Schmittgen, 2001]. All quantitative real-time PCRs were done in triplicate.

Genome-Wide Expression Microarray in miR-219-Overexpressed Breast Cancer Cell Line

Gene expression differences between transfected cells and the negative control were determined using whole-genome expression microarray (Illumina). Fold changes in gene expression were determined for each gene in miR-219-overexpressed cells compared with the negative control. Transcripts were identified as significantly influenced by miR-219 treatment if they fit the criteria of fold change >|2.0|. All array data have been uploaded to the Gene Expression Omnibus (GEO) database, and can be accessed via their website (http://www.ncbi.nlm.nih.gov/geo/; accession number pending).

Network Analysis of Identified Differential Transcripts

Identified transcripts were investigated for network and functional interrelatedness using the Ingenuity Pathway Analysis software tool (Ingenuity Systems). The set of genes and their fold change values were inputted into the program and used to identify experimentally verified interactions using the Ingenuity Knowledge Base, a manually curated database of gene-gene interactions extracted from peer reviewed publications [Calvano et al., 2005]. P-values for individual networks were obtained by comparing the likelihood of obtaining the same number of transcripts or greater in a random gene set as are actually present in the input set using a Fisher's exact test, based on the hypergeometric distribution. Additionally, genes identified in the highest confidence interaction network were interrogated for both conserved and poorly conserved binding sites using the TargetScan online algorithm (http://www.targetscan.org).

RESULTS

MicroRNA Promoter Methylation Associated with Long-Term Night Shiftwork

As folate is involved in the synthesis of a methyl donor group essential for DNA methylation [Kim, 2004], folate intake may act as a potential confounder in the association between CpG methylation and shiftwork. The analysis of miRNA methylation profiles was thus performed on long-term night shiftworkers (n = 10) and day workers (n = 10) pair-matched on age (±2 years) and total folate intake (±55 μg). The results indicate that 50 CpG loci of 31 miRNAs were significantly differentially methylated in long-term night shiftworkers compared to day workers, including the circadian-relevant miR-219. In all, 48 CpG loci of 29 miRNAs were significantly hypermethylated while two loci of two miRNAs were significantly hypomethylated (Fig. 1).

Figure 1.

Differentially methylated miRNAs in long-term shiftworkers. 48 CpG loci of 29 miRNAs were significantly hypermethylated and two CpG loci of two miRNAs were significantly hypomethylated in long-term shiftworkers compared to day workers. Where two or more CpG loci corresponded to the same miRNA, methylation differences were averaged.

Interaction Network of Molecules Affected by miR-219 Overexpression

Long-term night shiftworkers exhibited significant miR-219 promoter hypermethylation compared to day workers, suggesting reduced miR-219 expression in shiftwork-exposed women. In congruence with this observation, miR-219 was previously identified as a potential regulator of circadian length [Liu and Wang, 2012]. To better understand the relationship between shiftwork-altered miRNAs and carcinogenesis, miR-219 was overexpressed in MCF-7 breast adenocarcinoma cells and resulting gene expression patterns were interrogated using a genome-wide expression microarray followed by network-based analysis. Three hundred and nineteen differentially expressed transcripts were identified from the microarray and analyzed for network enrichment using the Ingenuity Pathway Analysis (IPA) software. The network identified by IPA as being most significantly associated with the set of input genes was designated as having functional relevance in “Antimicrobial Response, Inflammatory Response, and Infectious Disease” (P = 10−59), and is punctuated by a small core of molecules consisting of the immunomodulatory signaling and transcription factors IRF1, INF-α, INF-β, TNFSF10, NF-κβ, STAT1, and IL29 (Fig. 2).

Figure 2.

The top network of genes exhibiting altered expression following transfection of the miR-219 mimic. This network was identified by the Ingenuity Pathway Analysis software as being the most significantly associated with the set of input genes with altered expression following the introduction of the miR-219 mimic into MCF-7 cells. According to the manually curated Ingenuity Knowledge Base, molecules within this network have functional relevance in “Antimicrobial Response, Inflammatory Response and Infectious Disease” (P = 10−59). Transcripts that were upregulated are shaded in red, while transcripts that were downregulated are shaded in green (none), with color intensity signifying the relative magnitude of change. Each interaction is supported by at least one literature reference identified in the Ingenuity Pathway Knowledge Base, with solid lines representing direct interactions, and dashed lines representing indirect interactions. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

IRF1 (2.48-fold increase) encodes a transcriptional activator of INF-α and INF-β, as well as a host of other immunomodulatory factors, including TNFSF10 (TRAIL; 2.28-fold increase), which encodes a cytokine belonging to the tumor necrosis factor ligand family that preferentially induces apoptosis in transformed and tumor cells, OAS2 (3.80-fold increase), an essential component in the innate anti-viral immune response and a possible mediator of cellular apoptosis and growth [Chebath et al., 1987; Rusch et al., 2000], and DDX58 (2.47-fold increase), which encodes an innate immune receptor involved in viral dsRNA recognition and activation of downstream anti-viral immune responses. TRAIL is also a known activator of NF-κβ, the misregulation of which is associated with cancer cell proliferation and survival. INF-α and INF-β, belonging to the type I class of immunomodulators, are known to exhibit antitumor activities and are widely used in anti-cancer therapies [Qin et al., 1998; Belardelli et al., 2002]. INF-α and INF-β act via the Jak-STAT pathway by inducing the heterodimerization of STAT1 (2.39-fold increase) with STAT2 (2.59-fold increase) to promote the expression of interferon-stimulated genes (ISGs) and an anti-viral state [de Veer et al., 2001]. Likewise, IL29 (2.58-fold increase) encodes a type III interferon (IL-29) that promotes localized monocyte- and macrophage-mediated cytokine release at viral infection sites via the Jak-STAT signaling pathway [Jordan et al., 2007]. Of these genes, STAT2 and OAS2 were predicted by TargetScan to contain poorly conserved binding sites for miR-219. Representative hypermethylated and hypomethylated genes from the top network, along with fold changes, are included in Table 2.

Table 2. Genes in Top Network Displaying Differential Methylation in miR-219-Overexpressed MCF-7 Cells
SymbolAccession no.Brief relevant functionsFold change
BST2NM_004335.2Involved in the growth and development of B-cells2.40
DDX58NM_014314.2Involved in innate immune defense against viruses2.47
GBP1NM_002053.1Binds guanine nucleotides2.04
HLA-BNM_005514.5Provides instructions for making a protein that plays a critical role in the immune system2.55
IFI6NM_022872.2Regulation of apoptosis2.03
IFIT1NM_001548.2Inhibition of viral replication and translational initiation3.42
IFIH1NM_022168.2Alteration of RNA secondary structure2.75
IFIT2NM_001547.3Restricts virus infection through alteration of cellular protein synthesis6.53
IFIT3NM_001549.2Viral defence7.68
IFITM1NM_003641.2Transduction of antiproliferative and homotypic adhesion signals3.39
IFNB1NM_002176.2Anti-viral, antibacterial and anticancer activities2.40
IL29NM_172140.1Anti-viral immunity2.58
IRF1NM_002198.1Regulates apoptosis and tumor–suppression2.48
IRF7NM_004030.1Transcriptional activator2.33
ISG20NM_002201.4Anti-viral defense, maturation of snRNAs and rRNAs2.38
MX1NM_002462.2Regulates the calcium channel activity of TRPCs3.09
OAS2NM_001032731.1Control of cell growth, differentiation, and apoptosis3.80
OAS3NM_006187.2Control of cell growth, differentiation, and apoptosis3.05
PSMB9NM_002800.4Cleaves peptides in an ATP/ubiquitin-dependent process in a non-lysosomal pathway3.35
RARRES3NM_004585.2Inhibiton of cell growth and differentiation4.51
STAT1NM_139266.1Upregulation genes in response to a signal by either type I, type II or type III interferons2.39
STAT2NM_005419.2Signal transducer and activator of transcription that mediates signaling by type I IFNs2.59
TAP1NM_000593.5Degradation of cytosolic peptides4.41
TAP2NM_018833.2Antigen presentation2.21
TMEM140NM_018295.1Transmembrane protein 1402.36
TNFSF10NM_003810.2Induces apoptosis2.28
TRIM21NM_003141.3Formation of ubiquitin ligase complex, cell cycle progression2.01
UBA7NM_003335.2Activation of ubiquitin2.08
USP18NM_017414.3Maintains a critical cellular balance of ISG15-conjugated proteins2.88

DISCUSSION

As a result of the large percentage of the working population engaged in shiftwork, circadian disruption resulting from exposure to ill-timed light has aroused widespread concern as an occupational risk factor. According to the United States Bureau of Labor Statistics, almost 15% of full-time salaried workers regularly worked the night shift (16.7% of men and 12.4% of women) in 2004 (Occupational Outlook Handbook). Epidemiologic studies have confirmed that shiftwork is probably related to a wide range of health problems ranging from cadiovascular disease to cancer. But knowledge on the pathways and mechanisms mediating the effect of shiftwork on health remains in its infancy.

Our finding that prolonged shiftwork is associated with miR-219 hypermethylation is consistent with the idea that aberrant epigenetic patterns can be induced by shiftwork and modify the risk of circadian-related disease. miR-219 is purported to play an important role in circadian length regulation via the modulation of CLOCK- and BMAL1-dependent PER transcription [Cheng et al., 2007]. Moreover, miR-219 hypermethylation is associated with early stage tumor development in hepatocellular carcinoma (HCC) patients [Gao et al., 2008] and has been found to exhibit methylation-dependent downregulation in HCC cell lines compared with normal liver tissue [Furuta et al., 2010]. Downregulation of miR-219 has also been observed in lung adenocarcinoma [Yanaihara et al., 2006] and squamous cell carcinoma of the tongue [Wong et al., 2008], although it is unclear whether these changes are methylation-dependent. Furthermore, a single nucleotide polymorphism (rs213210) within the primary miR-219 sequence was found to be associated with increased risk of esophageal cancer, suggesting that miR-219 precursor processing efficiency may be reduced in the process of esophageal cancer development [Ryan et al., 2010].

The idea that cancer-relevant biological pathways may be subject to circadian control has gained steady traction since the circadian gene hypothesis of cancer development was first proposed years ago. This clock-cancer connection has been exemplified in numerous studies reporting genetic associations between circadian genes and breast cancer [Zhu et al., 2008; Hoffman et al., 2010, 2010; Fu et al., 2012], and may partly explain why shiftworkers seem to be at greater risk of breast cancer. miR-219's role in circadian regulation and implied role in cancer development prompted us to identify possible mechanisms by which its epigenetic dysregulation may lead to breast carcinogenesis. Overexpression of miR-219 in MCF-7 breast cancer cells resulted in parallel upregulation of a number of anti-viral immunomodulators with known regulatory roles in cellular proliferation and apoptosis. Functional interactions among this group of molecules suggests that miR-219 may act to induce IRF-1-mediated activation of the Jak-STAT and NF-κβ pathways. Upon induction by IRF-1, INF-α and INF-β are released from leukocytes and fibroflasts, which may, in turn, lead to heterodimerization of STAT1 and STAT2 in surrounding tissue and increased expression of ISGs with anti-viral and antitumor functions. Likewise, IRF-1-induced expression of TRAIL may in turn lead to the activation of NF-κβ, which has been associated with enhanced sensitivity to Fas-mediated apoptosis and senescence in tumor cells [Liu et al., 2012].

Dysregulation of STAT1 and STAT2 activity following miR-219 overexpression is wholly consistent with the observation that the Jak-STAT pathway is subject to miRNA-mediated circadian control in Drosophila [Luo and Sehgal, 2012]. The understanding that the activity of specific immune functions fluctuates in a circadian manner is an old one. In this context, miR-219 may help to relay signals from the central regulatory clock to fine-tune the circadian rhythmicity of immunomediated apoptotic and proliferative activity. Of additional interest is the prominence of anti-viral immunomodulators in affected pathways. Emerging research on the effectiveness of anti-cancer oncolytic virotherapies have revealed that certain viruses are able to trigger cancer cell death by inducing apoptosis [Chiocca, 2002]. miR-219's possible role as an immunostimulant of the anti-viral response suggests that its methylation-dependent downregulation in long-term shiftworkers may serve to dampen cancer cell sensitivity to apoptosis. Another intriguing possibility is that miR-219 silencing serves to weaken the body's resistance against virus-induced breast cancer, a phenomenon that is well-documented in mice but has yet to be established in humans [Zapata-Benavides et al., 2007; Mok et al., 2008; Lawson et al., 2010].

The validity of the observations we have made rests upon a number of key assumptions. First, we expect that miR-219 promoter hypermethylation in long-term night shiftworkers can result in a biologically meaningful reduction in miR-219 expression. Although no RNA was available to test the degree of phenotypic agreement between miR-219 methylation and transcript levels, methylation-dependent downregulation of miR-219 has previously been documented in HCC [Furuta et al., 2010]. Second, we expect that miR-219 hypermethylation in blood to be reflective of similar changes in breast tissue. Although, as a phenotypic biomarker, DNA methylation is known to vary between tissue types, a number of studies have demonstrated that methylation changes in blood can be reliably used as surrogate biomarkers of target tissue-level changes [Cui et al., 2003; Widschwendter et al., 2008]. Lastly, we assume that the in vitro molecular relationships we observed can be extrapolated and applied to a functional disease model at the organismal level. However, as in any in vitro model, all extrapolated data must be interpreted with caution.

Of further note, miR-219 was only one of two miRNAs with an established track record in circadian regulation found to be hypermethylated in night shiftworkers, the other being miR-34b. Interestingly, miR-34b and miR-219 were found to affect many of the same immunological pathways, suggesting that they may act synergistically to target important upstream immunoregulators. Additionally, although STAT2 and OAS2 are predicted to carry poorly conserved binding sites for miR-219, both genes were found to be upregulated following miR-219 transfection, lending credence to the possibility that miR-219-mediated induction of the innate antiviral response may be propagated indirectly via interactions with other miR-219 targets. A potential point of concern is our decision to use MCF-7 cells over other breast cancer cell lines. As MCF-7 cells were originally isolated from early stage breast cancer, they may better represent the physiological conditions present in primary tumors when compared to other commonly used cell models (e.g., MDA-MB-231). Moreover, the characteristics of the original host (69-year-old Caucasian female) closely match the racial background and expected age of breast cancer onset of our study subjects. It would, nonetheless, be interesting to see whether our experimental results could be replicated in primary breast tumor cells, which should be the subject of a future investigation. Finally, it should be noted that natural dark-light cycles of northern latitude countries differ significantly from those of countries nearer the equator. Although seasonal variations in time of blood collection between our comparison groups have been taken into account, extrapolations of our findings to more central latitudes should be interpreted with care.

In summary, our epigenetic association study found differential promoter methylation in 31 miRNAs, including the circadian-relevant miR-219, the expression of which has previously been implicated in several cancer types. Overexpression of miR-219 in MCF-7 breast cancer cells resulted in accentuated expression of apoptosis- and proliferation-related anti-viral immunodulators of the Jak-STAT and NF-κβ pathways. These findings suggest that long-term night shiftwork exposure may lead to the methylation-dependent downregulation of miR-219, which may in turn lead to the downregulation of immunomediated antitumor activity and increased breast cancer risk.

AUTHOR CONTRIBUTIONS

FS analyzed the data and prepared the manuscript. XC helped with data interpretation. AF helped with data analysis and manuscript editing. JH, RGS, AT, UBV, TZ and YZ designed the study and were involved in data analysis and interpretation, and manuscript preparation.

Ancillary