ARLTS1 and prostate cancer risk : analysis of expression and regulation

ARLTS1 and P os a e Cance Risk - Analysis o Exp ession and Regula ion Sanna Sil anen 1 , Daniel Fische 2 , Tommi Ran ape o 1 , Vi pi Lai inen 1 , John Pa ick Mpindi 3 , Olli Kallioniemi 3 , Tiina Wahl o s 1 , Johanna Schleu ke 1,4 * 1Ins i u e o Biomedical Technology/BioMediTech, Uni e si y o Tampe e and Fimlab Labo a o ies, Tampe e, Finland, 2School o Heal h Sciences, Uni e si y o Tampe e, Tampe e, Finland, 3Ins i u e o Molecula Medicine (FIMM), Uni e si y o Helsinki, Helsinki, Finland, 4Depa men o Medical Biochemis y and Gene ics, Ins i u e o Biomedicine, Uni e si y o Tu ku, Tu ku, Finland Abs ac P os a e cance (PCa) is a he e ogeneous ai o which se e al suscep ibili y loci ha e been implica ed by genome-wide linkage and associa ion s udies. The genomic egion 13q14 is equen ly dele ed in umou issues o bo h spo adic and amilial PCa pa ien s and is consequen ly ecognised as a possible locus o umou supp esso gene(s). Dele ions o his egion ha e been ound in many o he cance s. Recen ly, we showed ha homozygous ca ie s o he T442C a ian o he ARLTS1 gene (ADP- ibosyla ion ac o -like umou supp esso p o ein 1 o ARL11, loca ed a 13q14) a e associa ed wi h an inc eased isk o bo h unselec ed and amilial PCa. Fu he mo e, he a ian T442C was obse ed in g ea e equency among malignan issue samples, PCa cell lines and xenog a s, suppo ing i s ole in PCa umou igenesis. In his s udy, 84 PCa cases and 15 con ols we e analysed o ARLTS1 exp ession s a us in blood-de i ed RNA. A s a is ically signi ican (p = 0.0037) dec ease o ARLTS1 exp ession in PCa cases was de ec ed. Regula ion o ARLTS1 exp ession was analysed wi h eQTL (exp ession quan i a i e ai loci) me hods. Al oge he ou een signi ican cis-eQTLs a ec ing he ARLTS1 exp ession le el we e ound. In addi ion, epis a ic in e ac ions o ARLTS1 genomic a ian s wi h genes in ol ed in immune sys em p ocesses we e p edic ed wi h he MDR p og am. In conclusion, his s udy u he suppo s he ole o ARLTS1 as a umou supp esso gene and e eals ha he exp ession is egula ed h ough a ian s localised in egula o y egions. Ci a ion: Sil anen S, Fische D, Ran ape o T, Lai inen V, Mpindi JP, e al. (2013) ARLTS1 and P os a e Cance Risk - Analysis o Exp ession and Regula ion. PLoS ONE 8(8): e72040. doi:10.1371/jou nal.pone.0072040 Edi o : Amanda Ewa Toland, Ohio S a e Uni e si y Medical Cen e , Uni ed S a es o Ame ica Recei ed Ap il 4, 2013; Accep ed July 3, 2013; Published Augus 5, 2013 Copy igh : ß2013 Sil anen e al. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License, which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal au ho and sou ce a e c edi ed. Funding: This wo k was suppo ed by he Tampe e G adua e P og am in Biomedicine and Bio echnology sala y and O ion-Fa mos Resea ch Founda ion g an o S.S. and by he Sig id Juselius Founda ion, he Academy o Finland (251074), he Finnish Cance O ganisa ions, and he Compe i i e Resea ch Funding o he Pi kanmaa Hospi al Dis ic (9N069) g an s o J.S. The unde s had no ole in s udy design, da a collec ion and analysis, decision o publish, o p epa a ion o he manusc ip . Compe ing In e es s: The au ho s ha e decla ed ha no compe ing in e es s exis . * E-mail: Johanna.Schleu k[email p o ec ed] In oduc ion P os a e cance (PCa) is a he e ogeneous ai , and i is he mos common malignancy among men in wes e n coun ies, including Finland. I is a mul i ac o ial disease, and de ini i e isk ac o s include age, e hnic o igin and amily his o y. In Finland, he incidence o PCa is 89.4/100,000, and in 2010, 4697 new p os a e cance cases we e diagnosed (h p://www.cance . i/ syopa ekis e i/en/). Despi e ex ensi e esea ch o e he las decade, he e iological isk ac o s and genes ha cause gene ic suscep ibili y emain la gely unknown. This lack o knowledge has hampe ed e ec i e cance p e en ion and de elopmen o be e ea men s. Mu a ions in he known high-pene ance PCa p edisposi ion genes explain only a small ac ion o PCa cases. A polygenic model o amilial agg ega ion o cance has been p oposed whe e se e al low-pene ance alleles may ha e a mul iplica i e and/o modi ying e ec . One low-pene an candida e gene is ARLTS1 (ARL11), ADP- ibosyla ion ac o like umou supp esso p o ein 1, a pu a i e umou -supp esso gene on ch omosome 13q14, which has been shown o unc ion in many human cance s [1–5]. ARLTS1 is a membe o he ADP- ibosyla ion ac o amily ha plays a ole in apop o ic signalling. The same ch omosomal a ea on 13q has been indica ed in a mul i-cen e genome-wide linkage s udy in amilies wi h a leas i e a ec ed membe s [6]. In ano he ecen s udy, he immedia e adjacen egion 13q13 showed a sugges i e linkage o PCa, wi h a HLOD.1.9 [7]. In addi ion, he locus 13q14 is among he mos equen ly dele ed ch omosomal egions in soma ic umou issues in bo h unselec ed and he edi a y p os a e cance s [8,9], sugges ing ha ARLTS1 could be a a ge o bo h ge mline and soma ic mu a ions. We p e iously epo ed a signi ican associa ion o ARLTS1 T442C ( s3803185) homozygo e ca ie s wi h PCa [10]. This isk geno ype was also associa ed wi h dec eased ARLTS1 exp ession in he lymphoblas oid cell line samples o PCa pa ien s, and ARLTS1 co-exp ession signa u es om da a mining e ealed ha ARLTS1 exp ession was s ongly associa ed wi h immune sys em p ocesses. These p ocesses a e o in e es because a link be ween ch onic in lamma ion and PCa p og ession has been epea edly p oposed, and mul iple genes ac ing in in lamma o y pa hways ha e been linked o PCa suscep ibili y [11–13]. Complex diseases, such as cance , a e caused by a combina ion o mul iple gene ic in e ac ions and en i onmen al ac o s, which a e ha d o de ec and link o each o he . To de e mine ue causal associa ions using s a is ical me hods and pheno ype in o ma ion, i is ad isable o o ganise indi idual ma ke s in o g oups acco ding PLOS ONE | www.plosone.o g 1 Augus 2013 | Volume 8 | Issue 8 | e72040 o meaning ul biological c i e ia, such as in lamma ion. By composing a ian se s, i is possible o educe he numbe o hypo heses being es ed, which allows he associa ion be ween a genomic ea u e and a pheno ype o be mo e easily de ec ed. Epis asis in geno ype le el is de ined as he in e ac ion among mul iple genes o loci, and his join gene ic e ec may be he ac o behind ‘‘missing he i abili y’’, a phenomenon linked o he unexplained po ion o he edi a y cance suscep ibili y, which is obse ed in PCa. The genome-wide exp ession quan i a i e ai loci, eQTL, analysis is a me hod o s udying epis asis in complex ai s ha is able o de ec associa ions be ween geno ypic and exp ession da a. The eQTL analysis is a widely used app oach o gain insigh in o he ole o single nucleo ide polymo phisms (SNPs) a ec ing ansc ip le els. In his s udy, we in es iga ed he mechanisms behind he p e iously obse ed associa ion o ARLTS1 and PCa by ocusing on inding gene/exp ession in e ac ions ha dispose pa ien s o PCa, including in e ac ions in ol ing he ARLTS1 gene. To u he in es iga e he ARLTS1 exp ession di e ences seen p e iously in umo samples, p os a e cance and lymphoblas oid cell lines [10], we pe o med unc ional eQTL analysis om whole blood de i ed o al RNA om PCa pa ien s. The p e iously epo ed ARLTS1 co-exp ession wi h immune sys em p ocesses [10] was es ed by MDR analysis. To ou knowledge, his is he i s s udy epo ing he indings o ARLTS1 in e ac ing a ian s and p os a e cance eQTLs a he 13q14 egion. Ma e ials and Me hods S udy popula ion All he samples we e o Finnish o igin. The iden i ica ion and collec ion o he Finnish HPC amilies has been desc ibed elsewhe e [14]. The amilial samples analysed in his s udy had a leas wo a ec ed i s o second deg ee ela i es. Al oge he , 102 p os a e cance cases and 33 heal hy male amily membe s belonging o 31 amilies we e ini ially aken in o he s udy popula ion. The clinical cha ac e is ics o he amilial pa ien s used in RNA sequencing (n = 84) a e e e ed o in Table 1. A e age age a diagnosis was 63.0 y. Pa ien in o ma ion and samples we e ob ained wi h ull w i en in o med consen . The s udy was pe o med unde app op ia e esea ch pe missions om he E hics Commi ees o he Tampe e Uni e si y Hospi al, Finland, as well as he Minis y o Social A ai s and Heal h in Finland. Genome-wide SNP Geno yping Genome-wide SNP geno yping was pe o med using he HumanOmniExp ess BeadChip mic oa ay (Illumina, Inc., San Diego, CA, USA) by he Technology Cen e, Ins i u e o Molecula Medicine Finland (FIMM), Uni e si y o Helsinki. This a ay co e s mo e han 700,000 ma ke s wi h an a e age spacing o 4 kb ac oss he en i e genome. DNA Sequencing The ARLTS1 a ian s a us o he PCa pa ien s used in Illumina geno yping was examined by di ec sequencing. Sequencing was pe o med in an Applied Biosys ems 3130xl Gene ic Analyze (Li e Technologies Co po a ion, Ca lsbad, CA, USA) acco ding o he manu ac u e ’s ins uc ions. P ime s and PCR condi ions used in he mu a ion sc eening a e a ailable upon eques . RNA Ex ac ion and sequencing To al RNA was ex ac ed om 84 PCa cases and 15 heal hy male ela i es. All subjec s belonged o he 31 Finnish HPC amilies men ioned abo e. To al RNA was pu i ied om whole blood collec ed in PAXgeneHBlood RNA Tubes (P eAnaly iX GmbH, Swi ze land/Qiagen/BD) using he MagMAX TM o S abilized Blood Tubes RNA Isola ion Ki (AmbionH/Li e Technologies, Ca lsbad, CA, USA) and he PAXgene Blood miRNA Ki (P eAnaly iX GmbH, Swi ze land/Qiagen/BD). The RNA quali y was assessed using he Agilen 2100 Bioanalyze and he Agilen RNA 6000 Nano Ki (Agilen Technologies, San a Cla a, CA, USA). Lib a y p epa a ion, a ge en ichmen and massi ely pa allel pai ed-end sequencing o exp essed ansc ip s was pe o med by Beijing Genomics Ins i u e (BGI Hong Kong Co., L d., Tai Po, Hong Kong) using Illumina HiSeq2000 echnology (Illumina Inc., San Diego, CA, USA). eQTL analysis To ob ain ARLTS1 exp ession alues, i s , he eads om RNA sequencing we e aligned wi h opha 2 using hg19 as he e e ence genome [15]. The aw ead coun o ARLTS1 was calcula ed using HTseq, and ead coun s we e ans o med o no malised exp ession alues using DESeq (www-hube .embl.de/use s/ ande s/HTSeq/, [16]). The linea eg ession model implemen ed in PLINK was used o de ec ansc ip speci ic a ian s. Associa ions in cis we e delinea ed by a 1 Mb window ups eam o downs eam o he ARLTS1 SNP s9526582, as mos o he cis- eQTLs a e loca ed wi hin o close o he gene o in e es [17]. In addi ion, we applied a new me hod based on p obabilis ic indices o es o eQTL. The new me hod is a non-pa ame ic di ec ional es (simila o he well-known Jonckhee e-Te ps a es ), imple- men ed in ou R-Package Gene icTools ha is a ailable on he Comp ehensi e R A chi e Ne wo k (h p://c an. -p ojec .o g/ package = Gene icTools), and a package desc ip ion is unde Table 1. Clinicopa hologic indings a diagnosis o he PCa pa ien s used in RNA sequencing (n = 84). n(%) Age a diagnosis ,65 yea s 48 (57.1) $65 yea s 35 (41.7) S age T s age T1 (clinically unde ec able) 31 (36.9) T2–T4 (clinically de ec able) 48 (57.1) M s age M0 (no e idence o me as asis) 53 (63.1) M1 (bone me as asis) 0 (0) MX (bone me as asis canno be assessed) 27 (32.1) PSA alue ,20 ng/ml 60 (71.4) $20 ng/ml 14 (16.7) G ade Gleason sco e ,7 44 (52.4) 7 10 (11.9) .7 0 (0) doi:10.1371/jou nal.pone.0072040. 001 Exp ession and Regula ion o ARLTS1 wi h PCa PLOS ONE | www.plosone.o g 2 Augus 2013 | Volume 8 | Issue 8 | e72040 de elopmen (unpublished da a). P- alues we e calcula ed using pe mu a ion es s and we e adjus ed o mul iple es ing by applying he Benjamini-Hochbe g co ec ion. We also calcula ed o each es size ain [0,0.1] he a io o he amoun o expec ed es ejec ions and he amoun o obse ed ejec ions. The a o which he a io o hese wo alues was maximal, we chose also as an op imal es size. Gene exp ession da ase All mic oa ay gene exp ession da a on cell lines (n = 1445) included in hese analyses a e publicly a ailable ia he Gene Exp ession Omnibus (GEO) (h p://www.ncbi.nlm.nih.go /geo/, accession numbe s; GSE36133, GSE7127, GSE8332, GSE10843, GSE10890, GSE12777, GSE15455, GSE18773, GSE20126, GSE21654 and GSE24795) and GSK Cance Cell Line Genomic P o iling Da a (h ps://cabig.nci.nih.go / ools/ caA ay_GSKda a) (2008) [18] (GlaxoSmi hKline). The bulk o he gene exp ession da a was acqui ed om he Cance Cell Line Encyclopedia (CCLE) (GSE36133) [19]. We used samples om he mos ecen and widely ci ed A yme ix mic oa ay pla o m, HGU133_plus 2.0, o pe o m hese analyses. Gene exp ession da a No malisa ion Gene exp ession da a no malisa ion was pe o med om he aw CEL iles using he A oma A yme ix (Ve sion 1.3.0) R package (h p://www.a oma-p ojec .o g) based on cus om CDF iles ( e sion 16) ound a h p://b aina ay.mbni.med.umich.edu [20]. We p ocessed exp ession o 19,003 dis inc genes. All compu a ions we e pe o med in he R s a is ical en i onmen , employing he BioConduc o sui e o packages. Co-exp ession analysis o ARLTS1.The co-exp ession analysis me hod was desc ibed p e iously [10]. The me a cell line da a (n = 1445) o igina ed om o e 48 human ana omical pa s. A co ela ion alue .0.30 and a p- alue ,0.05 we e used as de e minan s o a s a is ically signi ican associa ion. We pe - o med mul iple es co ec ions using Benjamini Hochbe g and Bon e oni me hods. We decided o use he Benjamini Hochbe g (BH) me hod o selec ing signi ican ly co-exp essed genes because i was mode a ely s ic a calling a gene pai co ela ion a alse posi i e. In silico unc ionali y p edic ion RegulomeDB (h p:// egulome.s an o d.edu/) and HaploReg (h p://www.b oadins i u e.o g/mammals/haplo eg/haplo eg. php) [21,22] da abases we e used o u he elucida e he ole o eQTLs in gene egula ion. RegulomeDB allows he ea u es o DNA and egula o y elemen s o non-coding egions o be assessed, and HaploReg is a ool o de eloping mechanis ic hypo heses o he impac o candida e egula o y non-coding a ian s on clinical pheno ypes and no mal a ia ion. MDR analysis The mul i ac o dimensionali y educ ion (MDR) p og am is publicly a ailable om he in e ne (www.epis asis.o g). We used e sion 2.0_be a_8.4 o examine gene-gene in e ac ions. MDR de ec s in e ac ions in ela i ely small sample sizes. MDR is a non- pa ame ic, model- ee da a mining app oach cons uc i e induc ion algo i hm ha ans o ms he high-dimensional da a in o one-dimensional a iables by pooling geno ypes in o high and low isk g oups based on he a io o cases o con ols ha ha e he geno ype in ques ion [23]. MDR selec s one gene ic model (a one, wo, h ee o ou s age locus) ha mos success ully p edic s he pheno ype o disease s a us, e.g., cance . Da a a e hen di ided in o en equal pa s o pe o m a 10- old c oss- alida ion. The model c ea es a aining se (9/10 o he da a) and es ing se (1/10 da a) o e alua e he p edic ion abili y. The p ocedu e epea s his p o ocol en imes and calcula es he c oss- alida ion consis ency (CVC). CVC depic s he numbe o imes ha pa icula model is chosen as he bes one o hose en in e als. F om he MDR esul s sec ion, es ing balanced accu acy (TBA) shows how many ins ances a e co ec ly classi ied. The ARLTS1 geno ypes om 102 PCa cases and 33 con ols we e combined wi h he GWAS da a o 700,000 SNPs. Selec ion o genes and SNPs o MDR The genes unc ioning in immune sys em p ocesses and in lamma ion pa hways we e selec ed om a p e iously published comp ehensi e collec ion made by Loza MJ e al. [24] because MDR is no able o un da a se s o housands o SNPs wi hin easonable ime limi s. In sho , he SNPs selec ed included hose in ol ed in apop osis, cy okine signalling, Toll-like ecep o signalling, leukocy e signalling, complemen , adhesion and na u al kille cell signalling. We also an MDR wi h a subse o genes ga he ed om ou p e ious ARLTS1 co-exp ession s udies (e.g., wel e genes wi hin B-cell ecep o [BCR] signalling pa hway). The o al amoun o SNPs was 12,011 o which 4,764 we e ound in ou Illumina Human OmniExp ess GWAS da a. Resul s RNA exp ession The ARLTS1 RNA exp ession le els we e analysed om o al RNA o 84 PCa cases and 15 con ols. A signi ican dec ease o he ARLTS1 exp ession le el in PCa cases was de ec ed (p = 0.0037, Fig. 1). This is in conco dance wi h ou p e ious esul s om cell line, benign p os a ic hype plasia (BPH) and umou specimen RNA exp ession da a [10]. eQTL analysis To iden i y possible cis-ac ing gene ic a ian s associa ed wi h ARLTS1 ansc ip le els, we pe o med an eQTL analysis wi hin a special a ea o he 13q14 egion. By a linea eg ession model (PLINK), we we e able o de ec 5 eSNPs a ec ing ARLTS1 exp ession (Table 2). When he calcula ion window was dimin- ished om 1 Mb o 200 kb, only one SNP, s7997377, emained. Wi h he di ec ional es pe o med by he R-Package analysis ool, 11 s a is ically signi ican eSNPs we e ound (Table 2), and wo we e in conco dance wi h he PLINK linea eg ession model Figu e 1. Rela i e ARLTS1 RNA exp ession om PCa pa ien s and heal hy con ols. Rela i e ARLTS1 RNA exp ession was de e - mined by RNA sequencing analysis. Columns ep esen means o indi iduals; ba s ep esen SD. doi:10.1371/jou nal.pone.0072040.g001 Exp ession and Regula ion o ARLTS1 wi h PCa PLOS ONE | www.plosone.o g 3 Augus 2013 | Volume 8 | Issue 8 | e72040 esul s. Fo bo h es p ocedu es a es size o 0.01 was chosen. The eSNPs ound by he di ec ional es we e loca ed mainly in non- coding egions (Table 2). The genomic loca ions o he eSNPs ound in he 13q14 egion a e isualised in Figu e 2. Al oge he , 468 genomic a ian s wi hin he 1 Mb egion o igina ing om ARLTS1 SNP s9526582 we e es ed by a linea eg ession model and di ec ional es . A e adjus ing o mul iple es ing, a FDR o 39% in he linea model and app oxima ely 32% in he di ec ional es has o be accep ed o keep he signi ican es esul s om he ma ginal p- alues. Fo a mo e common FDR le el o 10% one signi ican eSNP om he di ec ional es emained signi ican ( s9568354). When we conside ed he a io o expec ed and obse ed signi ican es s, a maximum a io o a= 0.011 in di ec ional es was iden i ied. Fo ha aapp oxima ely 2.5 imes mo e signi ican es esul s appea ed han expec ed. Hence, we epo also he abo e men ioned non-adjus ed p- alues o a signi icance le el 0.01. Wi h linea eg ession model, he amoun o obse ed signi ican es s ma ched he amoun o expec ed es ejec ions unde he null hypo hesis. The associa ion o he eSNP geno ypes wi h he ARLTS1 exp ession le el was calcula ed. A s a is ically signi ican co ela- ion was na u ally obse ed be ween all he 14 SNPs and ARLTS1 ansc ip le els. The eSNP geno ype - ARLTS1 exp ession associa ion box plo s a e depic ed in Figu e 3. The unc ionali y and possible ansc ip ional egula o y e ec o he 14 eSNPs wi hin genes ound by eQTL was e alua ed using ENCODE-da a in he RegulomeDB and HaploReg da abases. Al oge he nine o he ou een eQTLs a e epo ed in RegulomeDB. The indings in he GM12878 lymphoblas oid cell line a e emphasized below because his cell line esembles he issue ype om which he RNA sequencing da a was e ie ed. The mos subs an ial e idence o he egula ion o ARLTS1 was ound o SNP s2532975. I s egula o y ole is suppo ed by i s loca ion in a egula o y ac i e egion. Acco ding o he Chip-Seq da a om he GM12878 cell line, s2532975 esides in a BATF (basic leucine zippe ansc ip ion ac o ) binding si e. In addi ion, using posi ion weigh ma ix (PWM) ma ching, a CDC5 (cell cycle se ine/ h eonine-p o ein kinase) binding mo i has been iden i ied ha spans he genomic posi ion o his a ian . Fu he mo e, a p omyelocy ic leukemia zinc inge (PLZF) mo i is epo ed in HaploReg. As epo ed by HaploReg, CDC5 binding e iciency dec eases while PLZF binding inc eases. Addi ionally, he ch oma in s a e in he egion su ounding s2532975 migh adop weak enhance cha ac e is ics. This p edic ion is u he s eng h- ened by he p esence o wo his one ma ks in his egion, H3k4me1 and H3k4me2, iden i ied in he GM12878 cells. Ano he possible candida e o ARLTS1 egula ion is s9562905. In a Chip-Seq s udy, a POLA2 binding si e was iden i ied in a human emb yonic s em cell line, H1-hESC, in he egion su ounding s9562905. HaploReg p edic s ac i e enhance cha ac e is ics in he ch oma in su ounding s9562905 in he lymphoblas oid GM12878 cells, simila o s2532975. This in e p e a ion is suppo ed by he p esence o wo enhance Figu e 2. Schema ic diag am showing he genomic loca ions o eSNPs ga he ed by eQTL analysis in PCa pa ien s. The gene symbols a e as ollows: CYSLTR2 (cys einyl leuko iene ecep o 2), FNCD3A ( ib onec in ype III domain con aining 3A), MLNR (mo ilin ecep o ), CDADC1 (cy idine and dCMP deaminase domain con aining 1), CAB39L (calcium binding p o ein 39-like), SETDB2 (SET domain, bi u ca ed 2/CLLD8), PHF11 (PHD inge p o ein 11/NY-REN-34 an igen), RCBTB1 ( egula o o ch omosome condensa ion [RCC1] and BTB [POZ] domain con aining p o ein 1/CLLD7), ARLTS1 (/ARL11, ADP- ibosyla ion ac o -like 11), EBPL (emopamil binding p o ein-like), KPNA3 (ka yophe in alpha 3, impo in alpha 4), SPRYD7 (SPRY domain con aining 7/C13o 1, ch omosome open eading ame 1), MIR3613 (mic oRNA 3613), TRIM13 ( ipa i e mo i con aining 13), KCNRG (po assium channel egula o ), MIR-15A and MIR16-1 (mic oRNA genes 15a and 16-1) and DLEU2 (dele ed in lymphocy ic leukemia 2). doi:10.1371/jou nal.pone.0072040.g002 Exp ession and Regula ion o ARLTS1 wi h PCa PLOS ONE | www.plosone.o g 4 Augus 2013 | Volume 8 | Issue 8 | e72040 associa ed his one ma ks, H3k4me1 and H3k4me2. In addi ion, a FAIRE-sequencing s udy conduc ed o he GM12878 cells also implies ha his egion is in an open ch oma in s a e and is he e o e likely o ha e egula o y ac i i y. The a ian s s1543513, s7997737 and s9568354 sha e simila ch oma in s uc u al ea u es in he GM12878 cells. Acco ding o HaploReg, he ch oma in s a e associa es wi h he weakly ansc ibed egion. This p edic ion is con i med by he elonga ion o his one ma k H3k36me3 in he su ounding egions o he a ian s s1543513, s7997737 and s9568354. Acco ding o RegulomeDB, s1543513 is loca ed wi hin he I x3 and I x6 mo i s. Howe e , in HaploReg, he p esence o hese mo i s is no epo ed. Simila ly, he ansc ip ion ac o binding mo i o AIRE (au oimmune egula o ) is epo ed o s1262781 in RegulomeDB bu no in HaploReg. A MZF1 (myeloid zinc inge 1) mo i su ounding s7997737 is epo ed in bo h da abases. HaploReg epo s a nega i e LOD sco e di e ence o s7997737, which can be in e p e ed as lowe ed binding e iciency o MZF. Compa ed o he h ee a ian s men ioned abo e, s9568232 sha es simila cha ac e is ics ega ding i s ch oma in s a e. Acco ding o HaploReg, his ch oma in s a e is ela ed o elonga ed ansc ip- ion. The a ian s s2075610 and s1262774 a e loca ed wi hin egions mos likely unde he con ol o epigene ic egula ion by he polycomb-g oup p o eins in he GM12878 cells. In addi ion, DNase-Seq s udies pe o med o se e al cell lines indica e an open ch oma in s a e su ounding s2075610. Howe e , wo ep essed s a e ch oma in his one ma ks, H3k27me3 and H3k9me3, ha e also been iden i ied. Va ian s1262774 is no epo ed in he RegulomeDB. The emaining a ian s a e loca ed in he e och oma in egions, acco ding o HaploReg. Fo s7995192, s2580189 and s1262781, his p edic ion is u he con i med by he p esence o H3k27me3. In addi ion, ano he ep essi e s a e associa ed his one ma k, namely H3k9me3, is p esen in he s7995192 and s2580189 egions. Al hough he e is e idence o ep essed s a e ch oma in, RegulomeDB epo s ha ansc ip ion ac o binding mo i s do in ac span he genomic posi ions o s2580189 and s1262781. Two mo i s o XBP-1 (X-box binding p o ein 1) and ATF6 (ac i a ing ansc ip ion ac o 6) span he egion o s1262781, acco ding o RegulomeDB. HaploReg con i ms he p esence o XBP-1 and ATF6 mo i s and an addi ional mo i o SOX-17 (SRY [sex de e mining egion Y] box 17). HaploReg epo s lowe p edic ed binding e iciencies o XBP-1 and ATF6 and a sligh ly inc eased binding e iciency o SOX-17. RegulomeDB epo s a ZNF143 (zinc inge p o ein 143) mo i in he egion su ounding SNP s2580189. Howe e , HaploReg does no epo he p esence o his mo i . Taken oge he , he esul s ga he ed om RegulomeDB and HaploReg indica e ha he ARLTS1 eQTLs a e loca ed wi hin egula o y a eas o he 13q14 egion. Co-exp ession analysis The GeneSapiens mRNA exp ession da abase da a, including ARLTS1 exp ession, om 1445 cell lines (48 cance sub ypes) and p os a e cance umou s was a ailable o in e ac ion s udies. Al oge he 1381 genes wi h co ela ion alue .0.30 and p- alue ,0.05 was ound o be posi i ely co ela ing wi h he ARLTS1 gene. Using DAVID GO unc ional clus e ing, (h p://da id.abcc. nci c .go / ools.jsp) [25,26] a s ong associa ion wi h nucleus and zinc- inge p o ein p ocesses was illus a ed when all he genes posi i ely co ela ing wi h ARLTS1 exp ession (n = 36) in he PCa cell line coho we e aken in o accoun (wi h a mo e s ingen co ela ion alue .0.50). The g oup o nuclea p ocesses (nucleus, in acellula o ganelles, ansc ip ion and DNA-binding) was en iched when da a o PCa cell lines was s udied. The en ichmen sco e was 13.49 wi h a p- alue 1.1E-32. The adjus ed Benjamin sco e was 5.1E-30, and he clus e o zinc- inge binding p o eins e ealed a p- alue o 9.0E-22. The same phenomenon was Table 2. Ch omosomal egion 13q14 isk a ian s (eSNPs) associa ed wi h di e en ial ARLTS1 exp ession. SNP Gene Posi ion Allele1 Allele2 P- alue Adjus ed P- alue Linea eg ession model RS1886014 N/A 49321044 A G 0,007 0,369 RS7997737 SETDB2 50033188 G A 0,006 0,322 RS7337547 N/A 50443527 C A 0,008 0,384 RS7995192 N/A 50782599 G A 0,008 0,331 RS2532975 N/A 50945011 G A 0,005 0,322 Di ec ional es RS2075610 MLNR 49795705 G A 0,010 0,322 RS7997737 SETDB2 50033188 G A 0,008 0,322 RS1543513 SETDB2 50034684 A C 0,007 0,322 RS9568232 PHF11 50089844 A G 0,000 0,322 RS9562905 N/A 50210212 A C 0,008 0,322 RS9568354 SPRYD7 50487993 A G 0,002 0,000 RS2580189 N/A 50806640 A G 0,009 0,322 RS2532975 N/A 50945011 G A 0,001 0,322 RS1262781 N/A 51066171 A G 0,010 0,322 RS1262774 N/A 51068896 A G 0,006 0,322 RS17074618 N/A 51153475 A G 0,006 0,322 doi:10.1371/jou nal.pone.0072040. 002 Exp ession and Regula ion o ARLTS1 wi h PCa PLOS ONE | www.plosone.o g 5 Augus 2013 | Volume 8 | Issue 8 | e72040 Exp ession and Regula ion o ARLTS1 wi h PCa PLOS ONE | www.plosone.o g 6 Augus 2013 | Volume 8 | Issue 8 | e72040 obse ed wi hin he whole da a o cell lines (me a coho ) wi h genes showing a co ela ion alue .0.30. ARLTS1 co-exp ession genes (wi h co ela ion alue .0.50) om he me a cell line da a e ealed a ca ego y o immune sys em p ocesses (B-/T-cell ac i a ion, leukocy e/lymphocy e di e en ia ion and ac i a ion), wi h an en ichmen sco e o 2.94 (p- alue 5.57E-7, adjus ed Benjamin sco e 2.9E-5). ARLTS1 co-exp ession da a o genes nega i ely co ela ed o ARLTS1 iden i ied a s ong gene on ology o glycop o ein and plasma memb ane p o ein genes in PCa cell lines (n = 2722, co ela ion alue,20.50, en ichmen sco e 52.13, p- alue 3.4E- 72 and 38.35, p- alue 7.9E-32, espec i ely). Wi hin he nega i ely co ela ing genes, a clus e o immunoglobulin domain con aining p o eins ha bou ed an en ichmen sco e o 12.23 wi h a p- alue o 5.4E-24. The GO e m ‘‘cy okine ac i i y’’ e ealed an en ich- men sco e o 10.43 wi h a p- alue o 6.5E-10 wi hin nega i ely co ela ing genes. In addi ion o he esul o ARLTS1 nega i ely co ela ing genes, we also iden i ied clus e s o G-p o ein coupled ecep o s and cell-cell signalling. Top i e posi i ely and nega i ely ARLTS1 co ela ing genes in cell line da a a e p esen ed in Table 3 (uppe panel). Addi ionally, esul s o he ARLTS1 co-exp ession wi h genes (SETDB2, PHF11, SPRYD7, MLNR) ha ha bo ed eSNPs a e p esen ed in he lowe pa o he Table 3, om he co-exp ession analysis pe o med in he me a cell line, p os a e cell line da a and p os a e umo da a. The exp ession o ARLTS1 was posi i ely co ela ed wi h he exp ession o SETBD2, in bo h he whole da a o cell lines (me a cell line coho ) (co ela ion alue 0.64, p- alue 0.000) and in he Figu e 3. Co ela ion o he di e en geno ype g oups o eSNPs o ARLTS1 RNA exp ession le els. A, s2075610; B, s7997737; C, s1543513; D, s9568232; E, s9568354; F, s1886014; G, s7337547; H, s7995192; I, s9562905; J, s2580189; K, s2532975; L, s1262781; M, s1262774 and N, s17074618. doi:10.1371/jou nal.pone.0072040.g003 Table 3. ARLTS1 Co-exp ession signa u es om umo specimens and cell lines. Gene Co ela ion alue P- alue Samples (n) p al_co ec ed* p al_co ec ed# Top i e genes Co-exp ession in me a cell line da a Posi i e co ela ion BTK 0,69 0 2818 0 0 GPR18 0,66 0 2818 0 0 CXo 21 0,66 0 2818 0 0 P2RY8 0,66 0 2818 0 0 PIK3CG 0,65 0 2818 0 0 Nega i e co ela ion NCKAP1 20,62 1,69E-295 2818 3,22E-291 7,70E-295 CDC42BPB 20,59 1,64E-263 2818 3,11E-259 7,43E-263 GIPC1 20,57 1,10E-239 2818 2,10E-235 5,01E-239 PTMS 20,53 2,68E-208 2818 5,09E-204 1,22E-207 KIAA0284 20,53 1,04E-203 2818 1,97E-199 4,71E-203 Genes ha bo ing eSNPs Co-exp ession in me a cell line da a MLNR 20,15 1,35E-16 2818 2,57E-12 2,96E-16 PHF11 0,44 0,000 2818 0,000 0,000 SETDB2 0,64 0,000 2818 0,000 0,000 SPRYD7 0,28 0,000 2818 0,000 0,000 Co-exp ession in PCa cell lines MLNR 2 0,35 0,0352 36 1 0,081 PHF11 0,09 0,608 36 1 0,704 SETDB2 0,61 9,01E-05 36 1 0,001 SPRYD7 0,17 0,336 36 1 0,451 Co-exp ession in p os a e umo samples MLNR 0,03 0,821 75 1 0,934 PHF11 20,07 0,576 75 1 0,806 SETDB2 20,07 0,547 75 1 0,788 SPRYD7 2 0,34 0,003 75 1 0,045 *Bon e oni co ec ion. # Benjamini Hohchbe g mul iple es ing co ec ion. doi:10.1371/jou nal.pone.0072040. 003 Exp ession and Regula ion o ARLTS1 wi h PCa PLOS ONE | www.plosone.o g 7 Augus 2013 | Volume 8 | Issue 8 | e72040 speci ic PCa cell lines (co ela ion alue 0.61, p- alue 0.00009) (Table 3). Exp ession o he PHF11 gene was posi i ely co ela ed wi h ARLTS1 exp ession in he me a cell line da a (co ela ion alue 0.44, p- alue 0.000). Exp ession o MLNR and SPRYD7 was nega i ely co ela ed wi h ARLTS1 exp ession. ARLTS1 and MLNR had a co ela ion alue o 20.35 (p- alue 0.04) in PCa cell lines, and ARLTS1 and SPRYD7 had a co ela ion alue o 20.34 (p- alue 0.003) in p os a e umou specimens (Table 3). The s ong nega i e co ela ion o ARLTS1 and SPRYD7 exp ession le els was also alida ed in ou ansc ip ome da a o 84 PCa cases and 15 con ols. MDR analysis By di ec sequencing, we we e able o de ec six ARLTS1 a ian s a he same amplicon om PCa pa ien s included in he Illumina geno yping (n = 135). All he a ian s we e p e iously known ( s117251022, s3803186, s147120792, s3803185, s138452698 and G446A [T p149S op]). To elucida e he geno ypic ARLTS1 in e ac ions we used mul i ac o dimensionali y educ ion (MDR). Gene-gene in e ac ion s a us be ween ARLTS1 and genes unc ioning in immune sys em p ocesses wi hin 102 PCa cases and 33 con ols was calcula ed, bu we we e no able o ind any s a is ically signi ican ARLTS1 in e ac ions in his s udy coho (da a no shown). Discussion ARLTS1 is a cance -p edisposing gene wi h p o en umou supp esso p ope ies. Howe e , e y li le e idence on unc ion, especially on pa hways, is cu en ly a ailable. In his s udy, we we e able o e i y down egula ed ARLTS1 exp ession in he blood-de i ed RNA o PCa pa ien samples, demons a ing o he i s ime he e ec o ge mline al e a ion on ARLTS1 exp ession le els. Tumou supp esso unc ion o he ARLTS1 gene has been p e iously p o en by Calin GA e al., who ound ha ansduc ion o ull-leng h ARLTS1 o A549 cells in Nu/Nu mice dec eased umou g ow h when compa ed o emp y ec o [1]. The abili y o ARLTS1 o supp ess umou o ma ion in p eclinical models has also been obse ed wi h o a ian [27] and lung cance cells [28]. Howe e , hese esul s a e based on soma ic mu a ions in cance ous cell lines, whe eas ou esul e eals a no el exp ession di e ence a he ge mline le el, suppo ing he ole o ARLTS1 as a umou supp esso gene. In he u u e, hese ypes o indings could be used o enable sc eening and de ec ion o a - isk pa ien s e en be o e clinical diagnoses. We ha e p e iously geno yped ARLTS1 a ian s in p os a e, b eas and colo ec al cance [29] and p oduced a p os a e cance ollow-up s udy [10]. In he i s s udy [29], we epo ed a s a is ically signi ican associa ion wi h ARLTS1 a ian s T442C, G194T and p os a e cance isk. Howe e , a e adjus ing o mul iple es ing, none o he esul s we e signi ican . In he ollow- up s udy wi h la ge sample size, we epo ed a s a is ically signi ican associa ion wi h T442C a ian and p os a e cance isk [10]. We epo ed also a dec eased o los ARLTS1 RNA o p o ein exp ession in clinical p os a e umo s, p os a e cance cell lines and xenog a s, suppo ing he ole o ARLTS1 as a umo supp esso gene. Thus, he e is no con lic be ween he p e ious publica ions and his s udy ha is con i ming he umo supp esso ole o ARLTS1. He e, 14 eSNPs loca ed a he 13q14 egion we e shown o signi ican ly in luence ARLTS1 ansc ip le els in PCa pa ien s. The eQTL analysis was pe o med wi h wo me hods, a linea model app oach and a di ec ional es . One ad an age o ou di ec ional es me hod o e he commonly used linea model app oach is i s obus ness agains ou lie s and he weake model assump ions. In he linea model app oach, he exp ession alues o he di e en geno ype g oups a e conside ed o ollow he same no mal dis ibu ion up o a loca ion shi pa ame e . I he loca ion shi pa ame e is no equal o ze o, i gene a es a es ing p oblem. The di ec ional es only assumes ha he di e en exp ession alues ollow ce ain dis ibu ion unc ions, so a es ing p oblem only occu s i he dis ibu ions o he geno ype g oups a e s ochas ically o de ed. In e es ingly, 5 o he eSNPs a e loca ed wi hin he p o ein- coding genes SETDB2,PHF11,SPRYD7 and MLNR. Two eQTLs we e posi ioned in he SETDB2 (SET domain, bi u ca ed 2) gene, also known as he CLLD8 (ch onic lymphocy ic leukemia dele ion egion gene 8 p o ein) gene, which unc ions mainly in epigene ic egula ion [30]. The PHF11 gene, ano he hi o he eSNPs, is a posi i e egula o o Th1- ype cy okine gene exp ession h ough nuclea ac o kappa B, (NF-kB) [30,31]. The hi d eSNP gene, SPRYD7, (SPRY domain con aining 7/ch omosome 13 open eading ame 1 [C13o 1]) also known as ch onic lymphocy ic dele ion egion gene 6 p o ein, (CLLD6) [32], is p oposed o ha e a umou supp esso unc ion because i has been de ec ed o be down egula ed in B-CLL pa ien s [33]. The SPRY/B30.2 p o ein domain occu s in a a ie y o cellula p o eins, media es p o ein- p o ein in e ac ions and nega i ely egula es cy okine ac i i ies [34,35]. The ou h eSNP gene, MLNR/MTLR1, (G-p o ein- coupled ecep o 38, GPR38) is a membe o he G-p o ein coupled ecep o 1 amily and is iden i ied as he mo ilin ecep o [36]. The ound eSNPs we e no a ec ing he exp ession o he gene hey esided in, sugges ing he ole o hese eSNPs as speci ic egula o s a ge ing ARLTS1 exp ession. Howe e , no explici conclusions abou he ac ual causal in e ac ion be ween hese a ian s and ARLTS1 can be made wi hou u he unc ional alida ion. The eQTL esul ga he ed in his s udy was om a ela i ely small sample se in which analyses encompassed 1 Mb up- and downs eam o he ARLTS1 gene. Thus, u he alida ion in la ge sample se s and genomic a eas a e wa an ed. Howe e , in ou esul s, one eSNP emained signi ican e en a e adjus men s o mul iple es ing. Those adjus men s a e in ica e in he case o eQTL analysis and s udies wi h small sample sizes o en su e om a low de ec ion a e a e mul iple es ing adjus men [37]. One possible s a egy o deal wi h his is o accep a ela i ely high FDR o he sake o la e alida ion in a la ge popula ion. He e, no po en ial candida e eSNPs we e excluded om u he s udies because he awi h la ges a io be ween obse ed and ejec ed es s was conside ed as op imal es size. Lymphoblas oid cell lines (LCLs) a e widely used in eQTL s udies because hey a e an easily accessible sou ce o pa ien samples o a single cell ype. Pa icula ly wi h p os a e cance , he umou specimens o mul iple oci a e ha d o collec in la ge amoun s making i di icul o eliably de ec disease-associa ed ai s. By using p os a e cance specimens, i is possible o ind issue-speci ic gene ic e ec s, bu he use o lymphoblas oid cell lines o whole blood o amilial p os a e cance pa ien s enables one o de ec he possible he i able ge mline di e ences which may con ibu e o p os a e cance suscep ibili y. I has been a gued ha SNP- ansc ip app oaches using LCLs o small sample se s a e unde powe ed [38], bu many s udies ha e been able o ind o e lapping eQTLs in cell lines and p ima y issues [39–41]. Ou inding o he eSNP in he SPRYD7 gene was endo sed in he co- exp ession da a, in which he ARLTS1 exp ession le els we e signi ican ly co ela ed wi h SPRYD7 exp ession, and in he p os a e umou specimens (Table 3). Addi ionally, in he da a om he ENCODE conso ium, he signi icance and usage o LCLs a e jus i ied because he lymphoblas oid cell line (GM12878) Exp ession and Regula ion o ARLTS1 wi h PCa PLOS ONE | www.plosone.o g 8 Augus 2013 | Volume 8 | Issue 8 | e72040 is one o he h ee cell lines in he highe p io i y Tie 1 coho [42]. The e may be a join e ec be ween he 13q14 genes. Fo example, one la ge ansc ip a ian ha consis s o mo e han one gene has been p oposed wi h SETDB2 and PHF11 [30]. Ve y li le is known abou he in e ac ions in his a ea, pa icula ly he genomic collabo a o s o he ARLTS1 gene. A ecen s udy iden i ied cellula e inoic acid binding p o ein 2 (CRABP2) and phosphoglyce a e mu ase 1 (PGAM1) as no el ARLTS1-binding p o eins using he in- ame cDNA lib a y echnique [43]. These p o eins we e no obse ed in he exp ession o geno ype le el o ou da a, so u he s udies a e needed o elucida e he ac ual ARLTS1 in e ac ing genes and p o eins. Conside ing he in e ac ions be ween ARLTS1 and in lamma- ion pa hway genes, i was ou hypo hesis ha ARLTS1 would ha e in e ac ions wi h in lamma o y genes, as ch onic in lamma- ion has been p oposed as a p os a e cance isk ac o . Wi h MDR analysis, we aimed o in es iga e he p e iously obse ed ARLTS1 connec ion wi h immune sys em p ocesses [10]. Ano he aim was o es he hypo hesis o genes unc ioning in in lamma ion p ocesses a ec ing p os a e cance isk. Howe e , we ailed o subs an ia e ou hypo hesis, and s a is ically signi ican associa ion be ween ARLTS1 T442C SNP and he in lamma o y/immune sys em SNPs was no ound wi hin he 700,000 SNPs analysed. Fo he analysis, we ex ac ed dis inc SNP g oups om 4,764 ma ke s ound om ou da a o 700,000 a ian s in o al. One possible eason o he nega i e esul s may be he ela i ely limi ed sample se o 135 cases and con ols because we we e no able o gene alise he esul s o he es da a. We we e also conce ned abou o e i ing he da a wi h MDR. Ou p e iously ound associa ion be ween ARLTS1 T442C and p os a e cance isk may be caused by he in e ac ion ne wo k o di e en SNPs in he 13q14 egion. Va ian T442C may be included in a e y a e haploblock, o he e may s ill be a missing a ian because he egion has been connec ed o o he cance s in addi ion o PCa. In conclusion, in his s udy we ha e shown ha he ARLTS1 exp ession le el is in luenced by changes in ge mline exp ession le els and ha he ARLTS1 gene is a quan i a i e ai locus o 14 eSNPs in he 13q14 egion. Thus, ou esul s indica e a mo e complica ed ne wo k o changes ha inc ease PCa isk han me ely one gene ic a ian in ARLTS1. Acknowledgmen s We hank all o he pa ien s who pa icipa ed in ou s udy and acknowledge he con ibu ion o Riina Liikala and Rii a Vaala uo o echnical assis ance. Au ho Con ibu ions Concei ed and designed he expe imen s: SS TW DF JPM JS. Pe o med he expe imen s: SS VL DF JPM TR. Analyzed he da a: SS DF TR JPM. Con ibu ed eagen s/ma e ials/analysis ools: JS OK. W o e he pape : SS TW JS. Re e ences 1. Calin GA, T apasso F, Shimizu M, Dumi u CD, Yendamu i S, e al. (2005) Familial cance associa ed wi h a polymo phism in ARLTS1. N Engl J Med 352: 1667–1676. 2. F ank B, Hemminki K, B enne H, Ho meis e M, Chang-Claude J, e al. (2006) ARLTS1 a ian s and isk o colo ec al cance . Cance Le 244: 172– 175. 3. F ank B, Hemminki K, Meindl A, Wappenschmid B, Klaes R, e al. (2006) Associa ion o he ARLTS1 Cys148A g a ian wi h amilial b eas cance isk. In J Cance 118: 2505–2508. 4. F ank B, Meye P, Boe ge MB, Hemminki K, S apelmann H, e al. (2006) ARLTS1 a ian s and melanoma isk. In J Cance 119: 1736–1737. 5. Sellick GS, Ca o sky D, Houls on RS (2006) Rela ionship be ween ARLTS1 polymo phisms and isk o ch onic lymphocy ic leukemia. Leuk Res 30: 1573– 1576. 6. Xu J, Dimi o L, Chang BL, Adams TS, Tu ne AR, e al. (2005) A combined genomewide linkage scan o 1,233 amilies o p os a e cance -suscep ibili y genes conduc ed by he in e na ional conso ium o p os a e cance gene ics. Am J Hum Gene 77: 219–229. 7. C opp CD, Simpson CL, Wahl o s T, Ha N, Geo ge A, e al. (2011) Genome- wide linkage scan o p os a e cance suscep ibili y in inland: E idence o a no el locus on 2q37.3 and con i ma ion o signal on 17q21–q22. In J Cance 129: 2400–2407. 8. Visako pi T, Kallioniemi AH, Sy anen AC, Hyy inen ER, Ka hu R, e al. (1995) Gene ic changes in p ima y and ecu en p os a e cance by compa a i e genomic hyb idiza ion. Cance Res 55: 342–347. 9. Rokman A, Koi is o PA, Ma ikainen MP, Kuukasja i T, Pou iainen M, e al. (2001) Gene ic changes in amilial p os a e cance by compa a i e genomic hyb idiza ion. P os a e 46: 233–239. 10. Sil anen S, Wahl o s T, Schindle M, Sa amaki OR, Mpindi JP, e al. (2011) Con ibu ion o ARLTS1 Cys148A g (T442C) a ian wi h p os a e cance isk and ARLTS1 unc ion in p os a e cance cells. PLoS One 6: e26595. 11. S anos KS, De Ma zo AM (2012) P os a e cance and in lamma ion: The e idence. His opa hology 60: 199–215. 12. Kwon EM, Salinas CA, Kolb S, Fu R, Feng Z, e al. (2011) Gene ic polymo phisms in in lamma ion pa hway genes and p os a e cance isk. Cance Epidemiol Bioma ke s P e 20: 923–933. 13. De Ma zo AM, Pla z EA, Su cli e S, Xu J, G onbe g H, e al. (2007) In lamma ion in p os a e ca cinogenesis. Na Re Cance 7: 256–269. 14. Schleu ke J, Ma ikainen M, Smi h J, Koi is o P, Ba oe-Bonnie A, e al. (2000) A gene ic epidemiological s udy o he edi a y p os a e cance (HPC) in inland: F equen HPCX linkage in amilies wi h la e-onse disease. Clin Cance Res 6: 4810–4815. 15. T apnell C, Pach e L, Salzbe g SL. (2009) TopHa : Disco e ing splice junc ions wi h RNA-seq. Bioin o ma ics 25: 1105–1111. 16. Ande s S, Hube W. (2010) Di e en ial exp ession analysis o sequence coun da a. Genome Biol 11: R106-2010-11-10- 106. Epub 2010 Oc 27. 17. Vey ie as JB, Kuda a alli S, Kim SY, De mi zakis ET, Gilad Y, e al. (2008) High- esolu ion mapping o exp ession-QTLs yields insigh in o human gene egula ion. PLoS Gene 4: e1000214. 18. Geo ge RE, A iyeh EF, Li S, Mo eau LA, Neube g D, e al. (2007) Genome- wide analysis o neu oblas omas using high-densi y single nucleo ide polymo - phism a ays. PLoS One 2: e255. 19. Ba e ina J, Caponig o G, S ansky N, Venka esan K, Ma golin AA, e al. (2012) The cance cell line encyclopedia enables p edic i e modelling o an icance d ug sensi i i y. Na u e 483: 603–607. 20. Hu R, Qiu X, Glazko G, Klebano L, Yako le A (2009) De ec ing in e gene co ela ion changes in mic oa ay analysis: A new app oach o gene selec ion. BMC Bioin o ma ics 10: 20-2105-10-20. 21. Wa d LD, Kellis M (2012) HaploReg: A esou ce o explo ing ch oma in s a es, conse a ion, and egula o y mo i al e a ions wi hin se s o gene ically linked a ian s. Nucleic Acids Res 40: D930–4. 22. ENCODE P ojec Conso ium. (2011) A use ’s guide o he encyclopedia o DNA elemen s (ENCODE). PLoS Biol 9: e1001046. 23. Hahn LW, Ri chie MD, Moo e JH (2003) Mul i ac o dimensionali y educ ion so wa e o de ec ing gene-gene and gene-en i onmen in e ac ions. Bioin o - ma ics 19: 376–382. 24. Loza MJ, McCall CE, Li L, Isaacs WB, Xu J, e al. (2007) Assembly o in lamma ion- ela ed genes o pa hway- ocused gene ic analysis. PLoS One 2: e1035. 25. Huang da W, She man BT, Lempicki RA (2009) Sys ema ic and in eg a i e analysis o la ge gene lis s using DAVID bioin o ma ics esou ces. Na P o oc 4: 44–57. 26. Huang da W, She man BT, Lempicki RA. (2009) Bioin o ma ics en ichmen ools: Pa hs owa d he comp ehensi e unc ional analysis o la ge gene lis s. Nucleic Acids Res 37: 1–13. 27. Pe occa F, Iliopoulos D, Qin HR, Nicoloso MS, Yendamu i S, e al. (2006) Al e a ions o he umo supp esso gene ARLTS1 in o a ian cance . Cance Res 66: 10287–10291. 28. Yendamu i S, T apasso F, Fe acin M, Cesa i R, Se ignani C, e al. (2007) Tumo supp esso unc ions o ARLTS1 in lung cance s. Cance Res 67: 7738– 7745. 29. Sil anen S, Sy jakoski K, Fage holm R, Ikonen T, Lipman P, e al. (2008) ARLTS1 ge mline a ian s and he isk o b eas , p os a e, and colo ec al cance . Eu J Hum Gene 16: 983–991. 30. Zhang Y, Lea es NI, Ande son GG, Pon ing CP, B oxholme J, e al. (2003) Posi ional cloning o a quan i a i e ai locus on ch omosome 13q14 ha in luences immunoglobulin E le els and as hma. Na Gene 34: 181–186. Exp ession and Regula ion o ARLTS1 wi h PCa PLOS ONE | www.plosone.o g 9 Augus 2013 | Volume 8 | Issue 8 | e72040