Methylation-Associated Gene Silencing of RARB in Areca Carcinogens Induced Mouse Oral ...

14 Pages·2015·3.46 MB·English

Hindawi Publishing Corporation
BioMed Research International
Volume 2014, Article ID 378358, 13 pages
http://dx.doi.org/10.1155/2014/378358
Research Article
RARB
Methylation-Associated Gene Silencing of in Areca
Carcinogens Induced Mouse Oral Squamous Cell Carcinoma
Zi-LunLai,1Yung-AnTsou,1,2,3Shin-RuFan,4Ming-HsuiTsai,2,3Hsiao-LingChen,5
Nai-WenChang,6Ju-ChienCheng,4andChuan-MuChen1,7
1DepartmentofLifeSciences,AgriculturalBiotechnologyCenter,NationalChungHsingUniversity,
No.250Kao-KuangRoad,Taichung402,Taiwan
2DepartmentofOtolaryngologyHeadandNeckSurgery,ChinaMedicalUniversity,Taichung404,Taiwan
3SchoolofMedicine,CollegeofMedicine,ChinaMedicalUniversity,Taichung404,Taiwan
4DepartmentofMedicalLaboratoryScienceandBiotechnology,ChinaMedicalUniversity,No.91Hsueh-ShihRoad,
Taichung404,Taiwan
5DepartmentofBioresources,Da-YehUniversity,Changhwa515,Taiwan
6DepartmentofBiochemistry,CollegeofMedicine,ChinaMedicalUniversity,Taichung404,Taiwan
7RongHsingResearchCenterforTranslationalMedicine,iEGGCenter,NationalChungHsingUniversity,Taichung402,Taiwan
CorrespondenceshouldbeaddressedtoJu-ChienCheng;[email protected]
andChuan-MuChen;[email protected]
Received7May2014;Revised8June2014;Accepted10June2014;Published17August2014
AcademicEditor:CalvinYu-ChianChen
Copyright©2014Zi-LunLaietal.ThisisanopenaccessarticledistributedundertheCreativeCommonsAttributionLicense,
whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited.
Regardingoralsquamouscellcarcinoma(OSCC)development,chewingarecaisknowntobeastrongriskfactorinmanyAsian
cultures.Therefore,weestablishedanOSCCinducedmousemodelby4-nitroquinoline-1-oxide(4-NQO),orarecoline,orboth
treatments,respectively.ThesearethemaintwocomponentsofthearecanutthatcouldincreasetheoccurrenceofOSCC.We
examined the effects with the noncommercial MCGI (mouse CpG islands) microarray for genome-wide screening the DNA
methylationaberrantininducedOSCCmice.Themicroarrayresultsshowed34hypermethylatedgenesin4-NQOplusarecoline
inducedOSCCmicetonguetissues.Theexaminationsalsousedmethylation-specificpolymerasechainreaction(MS-PCR)and
bisulfitesequencingtorealizethemethylationpatternincollectedmousetonguetissuesandhumanOSCCcelllinesofdifferent
grades,respectively.Theseresultsshowedthatretinoicacidreceptor𝛽(RARB)wasindicatedinhypermethylationatthepromoter
regionandthelossofexpressionduringcancerdevelopment.Accordingtotheresultsofreal-timePCR,itwasshownthatde
novoDNAmethyltransferaseswereinvolvedingeneepigeneticalternationsofOSCC.Collectively,ourresultsshowedthatRARB
hypermethylationwasinvolvedintheareca-associatedoralcarcinogenesis.
1.Introduction ThemainriskfactorfordevelopingOSCCischewingareca,
especiallyinmanyAsiancultures.
Throughouttheworld,oralsquamouscellcarcinoma(OSCC)
In a clinical study, the incidence of oral cancer was
isoneofthemostcommontypesofcancers.Ithasahighcure elevated28timesforbetelquidusersascomparedtononbetel
rateforsmallprimarytumorsandinvolvesthedevelopment quidusers[3].Cigarettesmokinghassynergisticeffectwith
of second primary tumors and the long-term survival rate areca chewing, and such users have an 89 times higher
is <60% [1]. Furthermore,in Taiwan, accordingto statistics incidenceratethannonusers.Ifonehasthehabitofdrinking,
from the Department of Health, Executive Yuan, Taiwan, smoking, and betel quid chewing combined, there will be
OSCCranksasfourthamongthetenleadingcausesofcancer a 123 times higher incidence rate of having oral cancer
among males and is the fourth leading cancer in the male than those average individuals in the general population
populationandthenumberofdeathsincreaseseveryyear[2]. that are nonusers. There is the longitudinal cohort study
2 BioMedResearchInternational
on the alcohol, betel quid, and smoking, to the oral cancer genomicalterationsinOSCC[32,33].Consequently,weused
risk. The betel quid partook the significant higher hazard home-made mouse CpG island microarray to understand
risk to the oral carcinogenesis [4]. The most tumorigenic aberrantmethylationprofileduringOSCCtumorigenesisin
part of betel quid is the Piper longum L. and the calcium thisstudyandvalidatedthemethylationstatusbyMS-PCR,
hydroxide (slaked lime) which will cause the oral cavity bisulfitesequencing,andreal-timePCR.
to develop into an alkaline condition which will promote In many previous studies, retinoid acid suppresses car-
the tumorigenic effect of the Safrole in the Piper longum cinogenesisandinhibitsthegrowthofhumanheadandneck
L. The fibers of areca also cause oral mucosa damage and squamouscellcarcinoma(HNSCC)[34,35].Lossofretinoids
increasedmucosatobeexposedtothetumorigenicmaterial and their receptors has been associated with malignant
inthebetelquid.Inaclinicalsurvey,oralcancerwithareca progression in HNSCC [36]. Their receptors (RAR) are
chewing had far more incidence of oral submucosa fibro- centralregulatorstothenormalgrowthanddifferentiationof
sis and erythroplakia; furthermore, the pathologic findings avarietyofepithelialcells.RARchangeshavebeenassociated
also show severe hyperkeratosis, caries, and gingivitis as with cell immortalization, and re-expression of RAR-beta
compared to nonareca users. The oral cancer patients who (RARB) leads to growth inhibition in some circumstances
had the habit of betel quid chewing were also found to [19].LossofRARBexpressionisassociatedwithachangein
have a higher percentage of dysplastic change surrounding proliferativelifespanpotentialfrommortalitytoimmortal-
the tumor margin, and skip cancer lesion was frequently ity in HNSCC [37–39]. The promoter hypermethylation of
notedintheupperaerodigestivetract(tongue,hypopharynx, RARBcouldinhibitthegeneexpressionwhenaddedtothe
󸀠
and esophagus). The condemned mucosa even reached the methylationinhibitoranddeacetylationinhibitorlike5 -aza-
󸀠
esophagus. In an average clinical survey, 18% of cases were 2 -deoxycytidine(5-aza-dC)andtrichostatinA(TSA)which
foundtohaveesophaguscancerdiagnosedatthesametime could recover the gene expression and inhibit tumor cells
whenoralcancerpresented[5].Synchronousdoublecancer growth[36,39].
(second primary cancer) in the upper aerodigestive tract is In the present study, we investigated the role of RARB
frequentlynoted[6]. hypermethylationofCpGislandsinOSCCmousemodeland
Therewasastimulatingeffectwhenarecanutischewed its association with RARB expression in human oral cancer
along with betel leaf [7]. Furthermore, in Chiang et al. [8], celllines.Inaddition,weexaminedwhethertherepressionof
they used areca nut extract (ANE) and saliva-reacted ANE RARBtranscriptioncouldbereversedby5-aza-dCinhuman
(sANE) to treat three oral carcinoma cell lines, KB (epi- oral cancer cell lines, and finally, we evaluated the three
dermoid carcinoma), SAS (tongue carcinoma), and Ca9-22 main DNA methyltransferases that were involved in RARB
(gingivalcarcinoma).The higher cytotoxiceffects involving hypermethylation.
cellmorphologicchangesandupregulationofinflammatory
signalinginmRNAexpressionlevelswereobservedinthese
2.MaterialsandMethods
treatments. In addition, arecoline is the major alkaloid
in areca nut extracts and betel quid. It is the primary
2.1.MouseModelforOralCancer. Themousemodeldevel-
activeingredientresponsibleforthecentralnervoussystem
opment was modified as highlighted by Chang et al. [15,
simulation that is roughly comparable to that of nicotine,
40]. Briefly, the OSCC model was established by treating
which has a similar chemical structure [9–12]. There is
arecoline (Sigma, St. Louis, MO), as well as in combina-
alsoanothercarcinogen,4-nitroquinoline1-oxide(4-NQO),
tion with 4-NQO (Fluka, St. Louis, MO) in 4-5week age
which effectively induces oral and esophageal cancers that
old of C57BL/6JNarl male mice. The conditions for OSCC
closelyresembleearlyhumanlesionsinmiceandrats[13,14]. formation are 500𝜇g/mL arecoline (A), 200𝜇g/mL 4-NQO
Inthisstudy, wefollowedChangetal.[15]whoestablished
(N), and 4-NQO (200𝜇g/mL) combined with arecoline
aneffectivemousemodeloforalcancerandusedthismodel
(500𝜇g/mL) (NA) in the drinking water for 8 weeks. The
to identify potential markers of oral tumor progression by
drinkingwaterwaschangedeveryday,andmicewereallowed
utilizinganoncommercialmethylationmicroarray.
accesstothedrinkingwateratalltimeswhilereceivingtreat-
The promoter hypermethylation now has a key role for
ment.Afterthetreatment,thedrinkingwaterwaschangedto
research in the area of human multistage carcinogenesis.
Silencing of certain tumor suppressor genes may occur in ddH2Oandmiceweresacrificedatweek8,12,14,18,20,26,
and28,respectively.Thetongueswerecollectedandclassified
the absence of genetic change, via aberrant methylation of
into tumor parts (T) and nontumor parts (NT) for mouse
CpG islands [16–18]. OSCC is believed to arise through
CpGislandmicroarrayanalysis.
theaccumulationofnumerousgeneticandepigeneticalter-
ations [19–21]. There are several methods to determine
whetherpromotermethylationhasbeendevelopedincluding 2.2. Cell Culture and 5-aza-dC Treatment. Normal human
combination of bisulfite restriction assay (COBRA) [22], oral keratinocytes (NHOK) were cultured in Keratinocyte
genomicbisulfatesequencing[23],methylation-specificPCR Growth Medium (KGM, GIBCO, CA, USA). Oral cancer
(MS-PCR)[24],andmicroarray-basedmethylationanalysis celllines,DOK,OC2,andCa9-22,wereculturedinDMEM
[25]. Methylation microarray is a high throughput tool (GIBCO, CA, USA). HSC3 and TW2.6 were cultured in
for genome-wide methylation analysis [26–31]. To identify DMEM-F12(GIBCO,CA,USA).Allcellssupplementedwith
and characterize potential targets for treating oral cancer, a 10% fetal calf serum and 1% penicillin-streptomycin and
∘
genome-wideapproachwastakentoquantitativelymeasure culturedat37 Cwith5%CO2.Oralcancercellsweretreated
BioMedResearchInternational 3
Table1:PrimersetsusedforRT-PCR,MS-PCR,andbisulfitesequencing.
Primersets Senseprimer(5󸀠 →3󸀠) Antisenseprimer(5󸀠 →3󸀠) Tm(∘C) PCRsize(bp)
RARB-Hu-RT AGGAGACTTCGAAGCAAG GTCAAGGGTTCATGTCCTTC 60 771
Dnmt1-Hu-RT TACCTGGACGACCCTGACCTC CGTTGGCATCAAAGATGGACA 60 102
Dnmt3a-Hu-RT TATTGATGAGCGCACAAGAGAGC GGGTGTTCCACCCTAACATTGAG 64 110
Dnmt3b-Hu-RT GGCAAGTTCTCCGAGGTCTCTG TGGTACATGGCTTTTCGATAGGA 62 112
RARB-Mo-M GGATTAGAGTTTTCGTGCGTCG TACCCCGCCGATACCCAAACG 65 90
RARB-Mo-U GGATTAGAGTTTTTGTGTGTTG TACCCCACCAATACCCAAACA 62 90
RARB-Mo-BS CCACCCAACTCCATCAAACTC CCATACAATCAAACATAATCTC 58 476
RARB-Hu-M ATGTCGAGAACGCGAGCGATTC CTCGACCAATCCAACCGAAACG 64 151
RARB-Hu-U GGATGTTGAGAATGTGAGTGATTT TACTCAACCAATCCAACCAAAACA 62 155
RARB-Hu-BS GTGTGATAGAAGTAGTAGGAAG GTGATAGAAGTGGTAGGAAG 55 401
∗
Hu:human,Mo:mouse,RT:real-timeRT-PCR,M:methylatedset,U:unmethylatedset,BS:bisulfitesequencing.
by 5-aza-dC at 2𝜇M to reverse the methylation status as (Perkin-Elmer Life Sciences, NJ, USA) fluorescent dyes
describedin[36,41]. were coupled to tumor (T) and normal (NT) amplicons,
respectively, and cohybridized to the microarraypanel. The
combinedtumor/normalcontrolpair,with8𝜇gDNA,more
2.3.DNAandRNAExtraction. ThegenomicDNAextraction
than 180pmol Cy5, and 150pmol Cy3, would give strong
was conducted as noted in our previous report [42]. We
hybridization signals. The hybridization of 4,608 spots is
usedcollectednontumorparts(NT)andtumorparts(T)of
carried out under a 24 × 50mm cover glass sealed tightly
tonguesforDNAandRNAextractions.
within a moistened hybridization chamber, GeneMachines
∘
HybChambers (Genomic Solutions, MI, USA), in a 65 C
2.4. RT-PCR and Real-Time PCR. Total RNA was prepared
water bath from 12 to 16h. The posthybridization wash-
usingtheTRIREAGENT(Invitrogen,CA,USA).Onemicro-
ing steps are essentially those described by UltraGAPS
gramoftotalRNAwastreatedwith10unitsofRQ1RNase-
CoatedSlidesinstructionManual.Thehybridizedslideswere
FreeDNase(Promega,WI,USA)andextractedwithphenol-
scannedwiththeGenePix4000Bscanner(Axon,CA,USA)
chloroform.DNasethattreatedtotalRNA(1𝜇g)wasreverse
and the acquired images were analyzed with the software
transcribedwiththeImPromIIReverseTranscriptionSystem
GenePix Pro 4.0 (Axon, CA, USA). The microarray data
(Promega, WI, USA). For RT-PCR amplification was per-
wasanalyzedasdescribedpreviously[43–46,48,49].Briefly,
formed with 2720 thermal cycler (Applied Biosystems Inc.,
the Cy5/Cy3 ratio and the hybridization intensity from the
CA,USA)andReal-TimePCRamplificationwasperformed
tumor amplicons to the hybridization intensity from the
withRotor-Gene6000(Corbett,CA,USA).Theamplification
normal amplicons, from each image, are normally guided
∘
ofRT-PCRwasrepeatedfor28cyclesasfollows:95 C,30sec
by both the average global Cy5/Cy3 ratio from each image
for denature of the annealing temperature depending on
and the Cy5/Cy3 ratios from 9 internal controls (clones
thepairofgenespecificprimersets(Table1)for30secand
withoutrestrictioncuttingsiteswhosecopynumbersremain
∘
72 C, 30sec for extension. PCR reactions were performed
the same in tumors and normal samples). Yellow spots
in triplicate and the transcription level was normalized
(normalizedCy5/Cy3=1)representequalamountsofbound
with the GAPDH. For Real-Time PCR, the calculated gene
DNAfromeachamplicon,indicatingnomethylationdiffer-
expression fold fromCT valuewas performedaccordingto
encesbetweentumor(T)andnontumor(NT)genomes.The
the previously mentioned study, with a 𝑃 value of less than
analyzed data were using hierarchical clustering to classify
0.05exhibitinganobviouslysignificantdifference.
the relationships of all genes between collected T and NT
samples. A hierarchical clustering algorithm was used to
2.5.PreparationofMouseCpGIslandMicroarrayandAmpli- investigaterelationshipsamongtumorsamples.Thecomplete
con Generation. The mouse CpG island microarray was linkage and the dissimilarity measure (1 minus the Pearson
based on previously described human CpG island microar- correlation coefficient of the log-adjusted Cy5:Cy3 ratios)
rays [43–46]. A total 2,304 mouse CpG islands library wereusedfortheanalysis.Theresultantdendrogramshowed
(mCGI) clones were spotted on UltraGAPS Coated Slides linked closely related colorectal tumors into a phylogenetic
(Corning,MA,USA)bytheBioDotAD1500(BIODOT,CA, treewhosebranchlengthsrepresentedthedegreeofsimilar-
USA).Theampliconsformethylationanalysiswereprepared itybetweenthesetumors.
aspreviouslydescribed[47,48].
2.7. Bisulfite Sequencing and Methylation-Specific PCR
2.6.MicroarrayHybridizationandDataAnalysis. Thepuri- (MS-PCR) for Methylation Status Analysis. Genomic DNA
fied amplicons (5𝜇g) were conducted using the BioPrime (∼0.5𝜇g)wastreatedwithsodiumbisulfiteaccordingtothe
DNA labeling system (Invitrogen, CA, USA). Cyanine 5- manufacture’srecommendations(EZDNAMethylationKit;
ddUTP(Cy5-ddUTP)andCyanine3-ddUTP(Cy3-ddUTP) Zymo Research, CA, USA). All selected genes methylation
4 BioMedResearchInternational
100
%) 80
esis ( 60
n
e
g
no 40
ci
ar
C 20
0
8 12 14 18 20 26 28
Week
4-NQO/arecoline Arecoline
4-NQO
(a)
18weeks 18weeks 26weeks 28weeks
Control 4-NQO
Arecoline 4-NQO+arecoline
(b)
Figure1:TheprogressionofmousemodeldevelopmentforOSCC.(a)TheratioofcarcinogenesisinmouseOSCCmodel.Therewerethree
treatments,4-NQO/arecoline,4-NQO,andarecoline.Miceweresacrificedatweeks8,12,14,18,20,26,and28,respectively.Thescoringcriteria
formouseOSCCmodelaredescribedinSection2.(b)OSCCtonguetissueswithtumorswereexcised,fixed,embedded,andsectionedfor
H&Estaining.Themicethatweretreatedwith4-NQO+arecolinewouldinducemoreseriousOSCCformationthan4-NQOonlyand
arecolineonly.Theorderofseveritywasfollowedthetimeoftreatment.
statuses were examined by methylation-specific PCR (MS- 3.Results
PCR) and sodium bisulfite genomic sequencing. The PCR
∘ 3.1.OSCCMouseModelInducedbyArecolineand4-NQO. To
reaction was as follows: 95 C for 5min, followed by 45
∘ ∘ evaluatetheefficiencyofmousemodelinvolvingcotreating
cycles of 95 C, 30sec, Tm for 30sec (Table1), 72 C for
∘ witharecolineand4-NQOthatmimictheetiologyforOSCC
45sec and ended with an extension of 72 C for 5min and
∘ tumorgrowth,thepercentageofmiceexhibitingcarcinogen-
quickchillto4 ConaGeneamp2400PCRsystem(Applied
esis was calculated in Figure1(a). Tumor development was
Biosystems, CA, USA). For bisulfate sequencing analysis,
assessed when treated with 4-NQO (N) and 4-NQO plus
each PCR product was subcloned into the pGEM-T Easy
arecoline(NA)butnotinarecoline.Micewerealsosacrificed
Vector (Promega, WI, USA) and performed 5–10 clones
at 18, 26, and 28 weeks, and tongues with tumors were
in each selection, respectively. Each colony was sequenced
excised, fixed, embedded, and sectioned for H&E staining
usingtheBigDyeTerminatorv3.1CycleSequencingKitand
(Figure1(b)).TheH&Estainingalsoshowedthatthetumor
theautomatedABIPRISM3100GeneticAnalyzed(Applied
progresses were dealing with time and according to the
Biosystems,CA,USA).
treatments.Accordingtotheresults,theincidenceoftongue
carcinogenesisinNAgroupwassignificantlyhigherthanN
2.8. Statistical Analysis. Statistical analysis was performed group and arecoline group. Taken together, the treatment
using𝑡-testtoexaminetheassociationbetweenNHOKand of NA at week 28 was much more serious than other
othercelllines.One-sidedtestingwasusedtocalculatethe𝑃, weeks.Theseresultssuggestthatarecolinepromotes4-NQO
and𝑃<0.05wasconsideredstatisticallysignificant. carcinogenesisindamagedoralepitheliacells.
BioMedResearchInternational 5
4-NQO (28weeks) 4-NQO/arecoline (26weeks) 4-NQO/arecoline (28weeks)
(a)
4-NQO/arecoline 4-NQO
26NA3 26NA6 28A 28C 28NA3 28N7 28N8
Nit1
MDR1
Hoxd4
BRCA2
ELYS
p21
RARB
(b)
Figure2:PresentingofMCGImicroarrayhybridizationresultsandhierarchicalclusteringofmethylationdataofOSCCmodelmice.(a)
MCGImicroarrayhybridizationpanelcontained4,608duplicatedCpGislandtags.Theexpandedhybridizationviewsshowedtheusefulness
oftheMCGImicroarrayscohybridizedwithfluorescentlylabeledT(tumorpart)andNT(nontumorpart)withNat28weeksandNAat
26and28weeks.Spotshybridizedpredominantlywithtumorampliconbutnotwithnormalampliconwouldappearredandareindicative
ofhypermethylatedCpGislandlocipresentinthetumorgenome.(b)HierarchicalclusteringofN(4-NQO)andNA(4-NQO/arecoline)
samples.Atthetopliststhe5NAand2Nstudied.Therowcorrespondstoeachof109CpGislandlociselectedformethylationanalysis.CpG
islands(thenormalizedCy5:Cy3ratiosare≧2)arethosewithhypermethylationintumorDNA.Theselectedhypermethylatedcandidates
wereshowedontherightsides.TheyaredescribedingreaterdetailinTable2.
3.2.IdentificationofHypermethylationGenesfromOSCCTis- red and are indicative of hypermethylated CpG island loci,
suesbyCpGIslandMicroarray. Accordingtothetumorpro- present in the tumor genome. The hybridization results
gression percentages and H&E staining results, we selected showed that treatment with NA at week 28 represented
the tongues tissues to be targets from OSCC mouse model muchmorespotsthatwereobviouslymorehypermethylated
with N and NA at 26 and 28 weeks for MCGI microarray thanothertreatments.Yellowspots(Cy5:Cy3=1)represent
screening. Figure2(a) depicts representative data from the equal amounts of bound DNA from each amplicon, an
OSCC study. The expanded hybridization views showed indicationofnomethylationdifferencesbetweentumorand
the usefulness of the MCGI microarrays cohybridized with nontumor genomes. Selection of genes was based on the
fluorescently labeled T (tumor part) and NT (nontumor criteria described in the Materials and Methods. Figures
part) with N at week 28 and NA at weeks 26 and 28, 6(b) and 6(c) showed specific genes of selection results
respectively. Spots hybridized predominantly with tumor inhypermethylationandhypomethylation,respectively.We
amplicon, but not with nontumor amplicon, would appear conducted a confirmation study to determine whether the
6 BioMedResearchInternational
Table2:Theselectedgenelistofhypermethylation.
AccessionnumberGenesymbol Description Functions Relatedcarcinomatypes
DNAbinding, Nonsmallcelllungcancer,hepatoma,bladder
S80555 RARB Retinoicacidreceptor-beta ligand-dependentnuclear cancer,rectalcancer,breastcancer,headand
receptoractivity necksquamouscancer
AF069985 Nit1 NitrilasehomologI Hydrolaseactivity Oralsquamouscellcarcinoma(inthisstudy)
AL355176 BRCA2 BreastcancerIIgene Proteinbinding Bladdercancer,breastcancer,ovariancancer
AL928664 Hoxd4 HomeoboxD4gene DNAbinding Breastcancer,leukemia,neuroblastoma
Nonsmallcelllungcancer,bladdercancer,
Cyclin-dependentkinase Cyclin-dependentprotein
AF457187 p21 breastcancer,ovariancancer,
inhibitorIA kinaseinhibitoractivity
medulloblastoma,hepatoma
Pancreaticcancer,breastcancer,colorectal
Multidrug-resistance ATPbinding,ATPase
M60348 MDR1 cancer,glioblastoma,leukemia,laryngeal
proteingene activity
cancercell
Embryoniclargemolecule
AB081498 ELYS DNAbinding Oralsquamouscellcarcinoma(inthisstudy)
derivedfromyolksac
cutoffratio(≧2)couldaccuratelyidentifyhypermethylation. 3.4. Methyltransferase Expressions in Human OSCC Cell
The hierarchical clustering presented the 109 gene loci of Lines. DNA methylationis catalyzed by the familyof DNA
hypermethylationintheclassifierinNandNA(Figure2(b)). methyltransferases(DNMT)includingDnmt1,Dnmt3a,and
Thismethylationprofileanalysishasledtotheidentification Dnmt3b. Figure5 shows the measurement of gene expres-
of CpG island clusters that could evaluate many new genes sionsviareal-timePCRonDnmt1,Dnmt3a,andDnmt3bin
correlating with OSCC progression in mouse model. These humanoralcancerlines.ForDnmt1,therewerenoexpression
newly collected genes are shown in detail in Table2. Upon differencesbetweenNHOKandothercelllines(Figure5(a)).
furtherexamination,weselectedtheRARBgenetoexamine However, Dnmt3a and Dnmt3b showed significantly higher
expression levels in TW2.6 and Ca922 than others (Figures
ingreaterdetailbyMS-PCRbisulfitesequencingandsemi-
5(b)and5(c)).
quantitativeRT-PCR.ThelocationsofCpGislandsinmouse
and human RARB genes were predicted using MethPrimer
(http://www.urogene.org/methprimer/index1.html), respec-
tively(Figures3(a)and4(a)). 4.Discussion
OSCCisthemostcommonheadandneckneoplasm,affect-
3.3.VerificationofMethylatedGenesbyMS-PCRandBisulfite ing270,000peopleworldwideeachyear[20,32].According
Sequencing. TheRARBgenewasacandidatetargettoverify to related research, patients who smoke, drink, and chew
themethylationstatusinthesemouseOSCCtonguestissues betel quid experience a 5.32-fold increased likelihood of
andinhumanoralcancercelllinesthatwerealsotreatedby death as compared to those without any oral habits [50].
󸀠 󸀠
5 -aza-2 -deoxycytidine(5-aza-dC).Interestingly,RARBwas In Taiwan and other Southeast Asian countries, betel quid
hypermethylatedinmouseOSCCandreexpressionby5-aza- chewing is one of the most important risk factors for oral
dCtreatmentinhumanoralcancercelllines.InFigure3(b), cancer patients and associates as the main cause between
RARB was hypermethylated in mouse OSCC tumor parts betel quid chewing and oral cancer development [51].
of N (4-NQO) and NA (4-NQO + Arecoline) compared to 4-NQOisquinolinederivativeandatumorigeniccompound
normal parts. Furthermore, the methylation ratio of RARB which can induce DNA lesions. Quinone oxidoreducatase
in NA treatment at 28 week was 60%. It was much higher is one of the major enzymes that convert 4-NQO to the
than the 4-NQO treatment (20%) at 28 week. These results more active metabolite, 3-hydroxyaminoquinoline 1-oxide
correlated with the microarray analysis data. RARB also [52].Thisoxidoreducatasecanbeproducedfromthemucosal
investigated the methylation status in human oral cancer ofthesublingualinhumansandmice.Arecolineisanatural
lines (Figure4(b)). The results showed that Ca922, TW2.6, alkaloidproductfoundinthearecanut.Inthisstudy,these
andHSC3werehypermethylatedthaninNHOK.Italsolost two compounds were used to induce oral carcinogenesis in
expressionsinCa922,TW2.6,andHSC3butnotinNHOK a mouse model. When we added 4-NQO plus Arecoline to
(Figure4(c)).InFigure4(d),thebisulfitesequencingshowed thedrinkingwaterofmice,theresultsshowedthatmicewere
the hypermethylation in TW2.6 (92.5%), Ca922 (96.3%), shown to have induced carcinogenesis at 100% at week 26
OC2(90%),andHSC3(93.8%),butintheNHOKandDOK, and28(Figure1(a)).InH&Estaining,resectedoraltissueat
the normal and precancer cell lines were not methylated the end of week 28 (Figure1(b)) also showed that NA and
in bisulfite sequencing results. Taken together, these results Ntreatmentcouldinducesquamouscellcarcinoma,respec-
showed that the promoter methylation of RARB plays the tively.However,treatmentwitharecolinealonerevealedthat
mainroleinOSCCprogression. itdidnotinducetumorigenesisafter28weeks.
BioMedResearchInternational 7
CDS
Exon1 Exon2 Exon3 Exon4 Exon5 Exon6
Probe 1170bp
MS-PCR analytic region
(a)
4-NQO 4-NQO/arecoline
B S N T N T N T N T N T N T
M
18weeks
U
Methylation (%) 0(0/3) 0(0/3)
M
20weeks
U
Methylation (%) 33.3(1/3) 33.3(1/3)
M
26weeks
U
Methylation (%) 33.3(1/3) 33.3(1/3)
M
28weeks
U
Methylation (%) 20 (1/5) 60 (3/5)
(b)
Figure3:MethylationstatusofRARBgeneinmouseOSCCmodel.(a)TheCpGislanddiagramofRARBgene.Thehybridizedprobeusedto
microarraylocatedfromexononetoexonthree.WedesignedtheMS-PCRprimersetslocatedwithinthisregion.(b)ThedesignedMS-PCR
analysisresultsin4-NQOand4-NQO/arecolineat18,20,26,and28weeks,respectively.TheRARBgeneshowedhypermethylation(60%)in
4-NQO/arecolineat28week.M:methylatedset,U:unmethylatedset,B:bloodDNAformethylationnegativecontrol,S:SssItreatedDNA
formethylationpositivecontrol,and%methylation:thepercentagesofmethylation.Theredinvertedtriangleshowedthatthemethylated
RARBgenecouldbeamplified.
Methylation is important in the development of OSCC themethylationstatusandmRNAexpressionlevelscouldbe
and many tumor suppressor genes targeted by promoter verifiedbyMCGIarray.
methylation will by no doubt be described in the future. Accordingtothedepictedrepresentativedatafrompre-
Thetechniquesusedatpresenttodetectmethylationprovide vious studies, RARB expression is thought to be associated
goodsensitivity,specificity,andspeed.Therearemanytypes withcellularsensitivitytoretinoidinnumerouscancercells,
ofmethylationarraysforagenome-wideapproachtorealize including HNSCC cells, breast cancer cells, lymphocytic
methylation profile. In this study, we applied a home-made leukemia,andlungcancercells[53–57].MethylationofRARB
high throughput MCGI array to analyze DNA methylation was identified which had a correlation with primary oral
acrosstheentiregenomeintheOSCCmousetonguetissues. malignant diseases [58, 59]. The methylation array of 4,608
Thishome-madeMCGIarraynotonlycouldidentifymethy- genes(duplicateonchip)thatweusedinthisstudyincluded
lationprofilebutalsocoulddetectmRNAexpression(inour themajorityofgeneswhichhavepreviouslybeenassociated
previous studies). The chip was cohybridized with mouse withheadandneckcancer(e.g.,DAPK,MGMT,andCDH1,
Cot-I DNA and total RNA mixture to evaluate the quality etc.). However, only RARB in previously studied genes
and the exon-containing portions can be used to measure were shown to be positive for methylation. To explain the
levelsofgeneexpression.Accordingtothepreviousresearch, possibilityofthisdiscrepancy,thehybridizedprobesmaybe
8 BioMedResearchInternational
CDS
Promoter +1
Human RARB
CpG island
R T
MS-PCR analytic region 1469bp
Bisulfite sequencing region
R RA response element (RARE)
T TATA box
(a)
NHOK DOK TW2.6 Ca922 OC2 HSC3 Blood SssI
M
U NHOK DOK TW2.6 Ca922 OC2 HSC3
RARB
M
5-aza-dC
U 𝛽-Actin
(b) (c)
NHOK DOK
0% 5%
TW2.6 Ca922
92.5% 96.3%
OC2 HSC3
90% 93.8%
(d)
Figure4:MethylationstatusinhumanoralcancercelllinesofRARB.(a)TheschematicforCpGislandpresentationofRARB.Therewas
a CpG island located in promoter region. We designed the MS-PCR and bisulfite sequencing primer sets in this region. (b) The RARB
󸀠
methylation status in different human oral cancer cell lines. The cell lines were also treated with 5-aza-dC. The methylation status was
󸀠
recoveredbecauseof5-aza-dCtreatment.(c)TheRARBmRNAexpressionindifferenthumanoralcancercelllines.TheRARBwasnot
expressedinCa922,OC2,andHSC3obviously.(d)TheRARBbisulfitesequencingindifferenthumanoralcancercelllines.Thereweremore
methylatedRARBinTW2.6,Ca922,OC2,andHSC3thaninNHOKandDOK.ThehollowcircleistheunmethylatedCpGsiteandthefull
circleisthemethylatedCpGsite.
locatedondifferentregionsbetweenourarrayandprevious were higher expressions in TW2.6 and Ca922, the primary
methylation studies. However, we found some interesting oralcancercelllines,butnotinOC2andHSC3.Ascompared
genes and described this in greater detail in Table2 as they to other tumors, no correlation was seen between DNMT
were shown to have hypermethylation on the chip, but this upregulationandpromoterhypermethylation-inducedinac-
wasnotreportedbeforeinOSCC. tivation of tumor-related genes. The exact mechanisms of
In our study, DNMT1 does not affect the expression in DNMTupregulationremainunclear,butitissuggestedthat
human OSCC cell lines. Nevertheless, DNMT3a and aberrantDNMTactivity,especiallywithregardtoDNMT1,
DNMT3bdidaffecttheexpressionsinhumanOSCCcelllines isduetoarapidproliferationofcancercellsbecauseDNMT1
(Figure5).TheseresultsshowedthatDNMT3aandDNMT3b binds to proliferating cell nuclear antigen (PCNA) [60].
BioMedResearchInternational 9
4 12
d) Dnmt1 d) Dnmt3a
n (fol 3 n (fol 10 ∗
o o
essi essi 8
pr pr
x 2 x 6
e e
A A
N N 4
R R
m 1 m
d d 2
e e
at at
Rel 0 Rel 0
NHOK DOK TW2.6 Ca922 OC2 HSC3 NHOK DOK TW2.6 Ca922 OC2 HSC3
Cell lines Cell lines
(a) (b)
30 Dnmt3b ∗
25
∗
d) 20
ol
n (f 15
o
pressi 4
x
e
A
N
R
m
d 2
e
at
el
R
0
NHOK DOK TW2.6 Ca922 OC2 HSC3
Cell lines
(c)
Figure5:DNMTsRNAexpressioninhumanoralcancercelllines.ThemeasurementofDNMTsexpressionlevelsbyreal-timePCRwas
showedin(a),(b),and(c),respectively.(a)TheDnmt1expressionwasnotchangedinallhumanoralcancercelllines.(b)and(c)TW2.6and
Ca922werehigherexpressioninDnmt3aandDnmt3bthanothers.Thefiguresshownarethemeanofthreeexperimentswhereallofthe
sampleswereanalyzedintriplicate.Thestartsignshowedthestatisticallysignificant(𝑃<0.05).
Overexpression of all the DNMTs at the mRNA level havesecondprimarycancersandlocalregionalrecurrence,
has been shown for several cancers [45, 61, 62]. However, thus causing the results to be less significant even though
DNMT3aandDNMT3bwerethedenovomethyltransferases. there was trend to have reduced second primary cancers
In this result (Figures 5(b) and 5(c)), we thought DNMT3a developedandlesslocalregionalrecurrenceinpriorstudies
andDNMT3bwereaddedtonewmethylgroupstoDNAto [63]. However, there was another clinical research using
causeDNAmethylationaberrantattheearlystageofOSCC the isotretinoin (13-cis-retinoic acid) (50 to 100mg per
progression. square meter of body-surface area per day) as compared
We used some different approaches to deal with the with placebo, to be taken daily for 12 months. They offer
sparsenessofdata.Tobeginwithweusedamethylationarray significantly reduced second primary cancer development
of 4,086 genes (duplicated on chip), which provided more after 32 months of follow-up, and multiple second primary
comprehensive data. Here we also suggest that this study tumorsdevelopedintheplacebogroup[64].Therefore,they
couldofferabasicevidencethatpromoterhypermethylation suggestthatisotretinoinstillhastheabilitytopreventsecond
of RARB is correlated with the occurrence of betel-related primary cancers, but there was less use in preventing the
OSCC. primary site recurrence [64]. There was another human
Isotretinoin (13-cis-retinoic acid) is considered to have cohort study using an in situ hybridization (ISH) analysis
the effect of preventing second primary cancers and local checkerwith38pairsofsurgicalspecimensofprimaryOSCC
orregionalrecurrenceafterheadandneckcanceristreated. and noncancerous matched normal control to compare the
However,inapreviousclinicaltrial,chemopreventionther- cellular expression level of RARB. They found the loss of
apy involving retinoid acid does not cause significant dif- RARB in the advanced OSCCs especially when they are
ferencesinearlyheadandneckcancers.Thenonsignificant betel quid users [65]. This prominent result also give us
benefit result is due to the small number of patients. The an understanding that RARB is really an important factor
prospectivestudyrenderedasmallpercentageofpatientsthat forchemopreventionfortumorprogression.Therefore,betel
10 BioMedResearchInternational
Mouse
4-NQO/arecoline induced
oral cancertumorigenesis
Amplicons preparations
45s rRNA,3
CpG island microarray 18s rRNA,3
26and28week normal/tumor part
Nonblast,13
Selective ratiomedian≥2 Others,8
34genes upregulated in mouse OSCC
RARB
ELYS Hosd4 Nit1
Rule out the ratio of minus and≥10000 MDR1 p21 RRCA2
15genes upregulated in mouse OSCC
Selective ratio ofmedian≥2
34genes upregulated in mouse OSCC
(a) (b)
Others,8 45s rRNA
18s rRNA
Nonblast
p21
RARB BRCA2
Hoxd4
Rule out the ratio of minus and≥10000
15genes upregulated in mouse OSCC
(c)
Figure6:Outlineofthisstudy.(a)Theflowchartofresearchproject.(b)TheselectionfromMCGImicroarrayhybridizationresultsbythe
ratiohigherorequaltwo.Therewere34genesthatwerehypermethylatedinOSCCmousemodel.(c)TheselectionfromMCGImicroarray
hybridizationresultsbyrulingouttheratiowasminusandhigherorequalto10,000.Therewere15genesthatwerehypermethylatedinOSCC
mousemodel.
quid related hypermethylation of RARB will really increase oral cancer to reduce the incidence of oral cancer and to
thetumorigenesisandpoortreatmentoutcomeoforalcancer. provideabettertreatmentoutcome.
Concerning the mechanisms of the chemoprevention func-
tion,therewasanotherstudythatrevealedthattheretinoids
ConflictofInterests
could suppress basal expression of Cox-2 or EGF-mediated
inductionofCox-2inhumanoralsquamouscarcinomacells
The authors declare that there is no conflict of interests
[66].Thus,RARBnotonlyhascellcycleinhibitionandtumor
regardingthepublicationofthispaper.
suppressant effects, but also anti-inflammatory effects that
ceaseCOX-2relatedcancerizationeffectsontheoralmucosa.
This is pilot study that talked about the detailed mech- Authors’Contribution
anisms of retinoid acid function affected by areca. In the
future,weshouldconsiderretinoidaciduseorRARBrelated Zi-Lun Lai, Yung-An Tsou, and Shin-Ru Fan contributed
drugsforarecauserswhoarefoundtohaveoraltumorsor equallytothiswork.