Transcription Factor KLF5 Binds a Cyclin E1 Polymorphic Intronic Enhancer to Confer ... · Genomics Transcription Factor KLF5 Binds a Cyclin E1 Polymorphic Intronic Enhancer to Confer

Genomics

Transcription Factor KLF5 Binds a Cyclin E1Polymorphic Intronic Enhancer to ConferIncreased Bladder Cancer RiskJillian M. Pattison1, Valeriya Posternak2, and Michael D. Cole1,2

Abstract

It is well established that environmental toxins, such as expo-sure to arsenic, are risk factors in the development of urinarybladder cancer, yet recent genome-wide association studies(GWAS)provide compelling evidence that there is a strong geneticcomponent associated with disease predisposition. A single-nucleotide polymorphism (SNP), rs8102137, was identified onchromosome19q12, residing 6kbupstreamof the important cell-cycle regulator and proto-oncogene, Cyclin E1 (CCNE1). Howev-er, the functional role of this variant in bladder cancer predis-position has been unclear because it lies within a non-codingregion of the genome. Here, it is demonstrated that bladdercancer cells heterozygous for this SNP exhibit biased allelicexpression of CCNE1 with 1.5-fold more transcription occur-ring from the risk allele. Furthermore, using chromatin immu-noprecipitation assays, a novel enhancer element was identified

within the first intron of CCNE1 that binds Kruppel-like Factor5 (KLF5), a known transcriptional activator in bladder cancer.Moreover, the data reveal that the presence of rs200996365, aSNP in high-linkage disequilibrium with rs8102137 residing inthe center of a KLF5 motif, alters KLF5 binding to this genomicregion. Through luciferase assays and CRISPR-Cas9 genomeediting, a novel polymorphic intronic regulatory element con-trolling CCNE1 transcription is characterized. These studiesuncover how a cancer-associated polymorphism mechanisti-cally contributes to an increased predisposition for bladdercancer development.

Implications: A polymorphic KLF5 binding site near theCCNE1 gene explains genetic risk identified through GWAS.Mol Cancer Res; 14(11); 1078–86. �2016 AACR.

IntroductionUrinary bladder cancer (UBC) is one of the most prevalent

cancers worldwide, ranking eighth in theUnited States for leadingcancer deaths in men (NCI, www.cancer.gov; ACS, www.cancer.org). This disease typically presents at the superficial, low-gradestage when surgical methods are the best treatment options;however, the odds for recurrence are alarmingly high even whenfull resections are successful (1, 2). Although the molecularmechanisms are not fully elucidated with either the superficialor more aggressive subtypes of UBC, the driving oncogenicmutations associated with low-grade tumors are often enhancedupon the inactivation of certain tumor suppressors (2). Thissynergism leads to the progression and recurrence of muscleinvasive tumors (3–6). Therefore, further understanding of theunderlying molecular pathways involved in bladder cancer path-ogenesis is crucial for better diagnosis and long-term treatment ofthis disease.

An emerging concept in bladder cancer is that genetic back-ground significantly contributes to tumor onset and invasiveness.Previously, the greatest contributors to UBC development wereenvironmental risk factors, including smoking and exposure toarsenic (7–10). Somatic activating mutations in H-Ras andFGFR3, as well as other large chromosomal deletions throughoutthe genome, also drive the development of this disease (2, 11–16). By contrast, recent genome-wide association studies (GWAS)performed in Icelandic, European, and Asian populationsrevealed a number of inherited polymorphisms within thegenome that lead to an increased susceptibility for developmentof UBC (17–21). These studies highlighted variants that are bothspecific toUBConset and associatedwith several proto-oncogenicloci, including Cyclin E1 (CCNE1), which harbors few otherdisease-associated SNPs. However, given that the CCNE1 variantand the other GWAS-identified polymorphisms lie in intergenicand non-coding regions, their biological significance is currentlyunknown.

CCNE1, a member of the cyclin family of proteins, has beenimplicated in the development of multiple cancer types, specif-ically bladder tumors (14, 20, 22–28). CCNE1 is also critical forfaithful centrosome duplication, initiation of DNA replication,and chromosomal stability (29–32). Tight transcriptional controlof CCNE1 guides cell-cycle regulation, such that inappropriateexpression confers rapid proliferation and genome instability, asis seen in low-grade, superficial UBC (27, 32, 33). This indicatesthat dysregulation of CCNE1 expression is likely an early path-ogenic event in the development of UBC, working synergisticallywith other driving oncogenic mutations. Interestingly, whileCCNE1 amplification is seen in UBC tumors (14, 27, 34), somaticmutations within the gene do not frequently occur, suggesting

1Department of Genetics, Geisel School of Medicine at Dartmouth,Norris Cotton Cancer Center, Lebanon, NewHampshire. 2Departmentof Pharmacology and Toxicology, Geisel School of Medicine at Dart-mouth, Norris Cotton Cancer Center, Lebanon, New Hampshire.

Note: Supplementary data for this article are available at Molecular CancerResearch Online (http://mcr.aacrjournals.org/).

Corresponding Author: Michael D. Cole, Geisel School of Medicine at Dart-mouth, Norris Cotton Cancer Center, One Medical Center Drive, Rubin 633HB7936, Lebanon, NH03756. Phone: 603-653-9975; Fax: 603-653-9952; E-mail:[email protected]

doi: 10.1158/1541-7786.MCR-16-0123

�2016 American Association for Cancer Research.

MolecularCancerResearch

Mol Cancer Res; 14(11) November 20161078

on August 21, 2021. © 2016 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst August 11, 2016; DOI: 10.1158/1541-7786.MCR-16-0123

http://mcr.aacrjournals.org/

that enhanced expression is associated with misregulation of thegene. It iswell established that E2F-binding sites andpolymorphicregions within the CCNE1 promoter contribute to transcriptionregulation (23, 29, 35); however, very little is known aboutadditional CCNE1 DNA regulatory regions, particularly thoseinvolved in UBC development.

In light of the GWAS in UBC patients and the implications ofCCNE1 in bladder cancer, we sought to mechanistically connectthe genetic variant near the CCNE1 locus to the increased risk ofUBC development. Fortunately, the presence of polymorphismswithin the genome allows for assessment of allele-specific con-tribution, expression, and physiology from genotypically distinctalleles. This provides a powerful tool to study genetic variants andto understand the altered activity occurring on different alleleswithin a cancer cell. Here we combined human UBC SNP datawithhigh-throughput sequencingmethods to characterize a novelenhancer active in bladder cancer cells to explain the functionalsignificance of one particular genetic component in the develop-ment of UBC.

Materials and MethodsCell culture

Bladder cancer cell lines were a generous gift from Dr. DavidRobbins (Sylvester Comprehensive Cancer Center, University ofMiami, Florida). All cells were cultured in Dulbecco's ModifiedEagle Medium (DMEM) supplemented with 10% fetal bovineserum (FBS) and 1% penicillin–streptomycin antibiotics.

SNP linkage disequilibrium (LD) analysisLD analysis was performed using the SNP Annotation and

Proxy Search (SNAP) algorithms from the Broad Institute (MIT,www.broadinstitute.org/mpg/snap) and an r2 � 0.8 was recog-nized as significant.

Heterogeneous nuclear RNA (hnRNA) extractionCells were harvested and lysed in 0.5% NP-40 lysis buffer to

isolate nuclei, and then nuclear RNA was extracted using theTRIzol protocol (Ambion). RT-PCR products were gel extractedand submitted for sequencing in parallel with PCR fragmentsfrom genomic DNA from the same cell line to control for dyeprofiles within each sequencing run. For allele-specific expressionanalysis, qRT-PCR was performed using the mismatch amplifica-tion mutation assay (MAMA; ref. 36).

Real-time PCR (qPCR) and reverse transcriptase-PCR (qRT-PCR)

qPCR and qRT-PCR were performed using the iQ SYBR GreenSupermix (BioRad) and analyzed on a C1000 Thermal Cycler(BioRad) using the BioRad CFXManager 2.0 software. Due to thehigh GC content within the CCNE1 intronic region, PCR wasperformed using HotStarTaq DNA Polymerase (Qiagen). TheHotStarTaq Master Mix together with Q solution (Qiagen) andappropriate locus-specific primers allowed for amplification oftheGC-rich intronic region. For quantitative analyses, theQuanti-Fast SYBR Green PCR Kit was utilized per the manufacturer'sinstructions using the Q solution.

Chromatin immunoprecipitation (ChIP)ChIPs were performed according to the Upstate ChIP Kit

protocol (Millipore) using antibodies against an IgG control

(sc-2027, Santa Cruz Biotechnology), H3K27ac (ab4729,Abcam), and KLF5 (07-1580, Millipore). Briefly, 1� 106 to 2 �106 cells were crosslinked at 37�C in 1% formaldehyde andsonicated twice after lysis using a water bath sonicator (BioRup-tor) on high intensity for 10minutes at 30-second intervals. qPCRwas used to measure enrichment compared with input relative toeither a no Antibody or IgG control. Graphs represent threeindependent experiments.

ChIP-sequencing analysisDNAwas pooled from two replicate H3K27ac ChIP assays, and

input DNA was run in parallel to the ChIP to account for anydiscrepancies in ploidy within the cell lines. ChIP DNA (30 ng)from three UBC cell lines (5637, T24, and J82) was submitted forlibrary preparation and sequencing in a 50 base pair, single-end,multiplexed run, yielding 20 to 30 million reads per sample.Upon completion of the sequencing run, sequencing reads werealigned against the human reference genome (hg19) using BOW-TIE (37). A MACS analysis was performed to determine peaks ofenrichment of H3K27ac throughout the genome (38). TheMACSanalysis was conducted as recommended by comparing the ChIPDNA reads to the Input reads using a tutorial (http://liulab.dfci.harvard.edu/MACS/; refs. 38, 39). Sequences and processed data-sets have been submitted to Gene Expression Omnibus underaccession number GSE75286.

siRNA-mediated knockdownTwo Silencer Select siRNAs (Life Technologies) against KLF5 or

a scrambled negative control were used for RNAi experiments.Reverse transfections were performed according to the manufac-turer's protocol using Lipofectamine RNAiMAX Reagent. Cellswere transfected with 10 nmol/L concentrations of siRNA, andRNA was harvested 48 hours after transfection. qRT-PCR andWestern blotting were performed to confirm efficient depletionusing the following antibodies: GAPDH (sc-47724, Santa CruzBiotechnology) and KLF5 (ab24331, Abcam). The data depictthree independent experiments.

Dual-luciferase assaysEnhancer capabilities of the CCNE1 intronic region were

assessed using the Dual-Luciferase expression assay (Promega)according to the manufacturer's instructions. Briefly, enhancerconstructs were cloned into the pGL3-promoter vector, and5637 cells were transiently transfected with a pGL3 vector, aRenilla control vector (using a 1:10 dilution), and with orwithout the pcDNA3 empty vector or the pcDNA3-KLF5 expres-sion vector (Addgene). Transfections were performed in 24-wellplates using Lipofectamine LTX and 0.5 mg DNA per well. After48 hours, cells were harvested and Firefly and Renilla luciferaseactivities were measured using a microplate luminometer. Dataare presented as relative luciferase units (RLU) indicating Fireflyactivity/Renilla activity and represent an average of three inde-pendent transfections.

Genome editing using CRISPR-Cas9Dual sgRNA/Cas9 CRISPR vectors were designed to target the

CCNE1 intronic region using lenticrispr v2 (40). Target sequenceswere GCGCAAAGGGGGAGGGGTAC and CGCAAAGGGG-GAAGGGGTAC. 5637 cellswere infectedwith a lentivirus carryingan sgRNA/Cas9 construct and infected cells were selected in 1 mg/mLpuromycinmedia. Single-cell cloneswere expanded, andRNA

KLF5 and Cyclin E1 Confer Bladder Cancer Risk

www.aacrjournals.org Mol Cancer Res; 14(11) November 2016 1079




andDNAwere harvested from each clone for subsequent analyseson genome editing efficiency. DNA surrounding the CCNE1intronwas PCR amplified from individual cell isolates and clonedto characterize individual alleles. Multiple clones were sequencedto determine the allelic complement of specific clonal cell lines.

Statistical analysisAll experiments were performed three independent times, and

error bars denote standard error of the mean (SEM) unlessotherwise noted. Student t testswere used to calculate significance.

ResultsBiased allelic expression of CCNE1 occurs in bladder cancercells heterozygous for the GWAS-identified SNP rs8102137

Wewere interested to address the connection betweenUBC riskand the rs8102137 SNP on chromosome 19q12 (17, 20). TheGWAS-identified polymorphism falls 6 kb upstream from thetranscription start site (TSS) of CCNE1 and exhibits an odds ratioof 1.13 (20). Upregulation of CCNE1 is typically associated withearly events in UBC pathogenesis, particularly within superficial

and low-grade lesions, undoubtedly affecting many cellular pro-cesses (27). Interestingly, rs8102137 falls within a linkage dis-equilibrium (LD) block that encompasses over 19 kb of codingand non-coding DNA (Fig. 1A), precluding ascribing any func-tional significance to the original identification. We used allele-specific expression analysis to explore this issue. First, we geno-typedbladder cancer cell lines and fortuitously found four that areheterozygous for the rs8102137 SNP and the LD block atCCNE1.These four UBC cell lines were all derived from transitional cellcarcinomas, varying from low to intermediate stages of disease(ATCC, www.atcc.org).

The heterozygosity of the bladder cancer cell lines allowedstudies of individual allelic expression within common cellularcontexts. The LD block encompassing the CCNE1 promoter andtranscribed regions contains an intronic SNP, rs3218036,which isstrongly linked to the GWAS SNP rs8102137with r2¼ 1.000 (Fig.1A). To determine if there were differences in allelic expression ofCCNE1 associated with the risk variant, we collected heteroge-neous nuclear RNA (hnRNA) from the heterozygous bladder celllines and assayed for expression of the intronic rs3218036 SNP inmRNA using allele-specific RT-PCR. This technique revealed the

Figure 1.

Allele-biased expression of CCNE1 isobserved in bladder cancer cell linesheterozygous for the GWAS SNPrs8102137. A, schematic of the CCNE1locus and the LD block encompassing theGWAS SNP rs8102137 (indicatedby the green plus) and two intronic SNPsin high LD. B, four heterozygousbladder cancer cell lines (J82, 5637,UMUC3, and T24) show biased allelicexpression of CCNE1, while a prostatecancer cell line (LNCaP) heterozygous atthis locus does not. Allelic expression ismeasured through qRT-PCR using MAMAprimers specific to each allele. The ratio ofthe two alleles in the hnRNA population isrelative to the ratio of the alleles in thegenomic DNA. Right, a representationof the chromatograms of genomic DNAand RNA from the heterozygous UBCline, 5637. Error bars indicate SEM,and asterisks indicate significance(� , P < 0.05).

Pattison et al.

Mol Cancer Res; 14(11) November 2016 Molecular Cancer Research1080




relative contribution of each individual allele (low or high risk)within the total unspliced RNA population (compared withgenomic DNA). We found that in all four UBC lines, there wasat least 1.5-fold more expression from the high-risk allele versusthe low-risk allele, a bias not seen in a non-UBC cell line (Fig. 1B).To further visualize the allelic differences (41), we performed RT-PCR followed by direct sequencing and compared the profiles togenomic DNA from the same cells (Fig. 1B, right). This indepen-dent approach confirmed that the higher risk allele is expressed ata 1.5-fold higher level than the lower risk allele in commoncellular backgrounds. We conclude that the rs8102137 SNP isassociated with elevated CCNE1 transcription.

Enhancer marks and linked SNP rs200996365 highlight a KLF5regulatory element within the first intron of CCNE1

The elevated levels of transcription associated with the higherrisk allele suggest that the presence of the risk variant directly orindirectly alters endogenous transcriptional regulation ofCCNE1,likely due to variation in a linked regulatory element. Thus, we

sought to understand the DNA landscape at chromosome 19q12in the bladder cancer genome. Initial studies compared the GWASSNP to the Genbank and ENCODE databases, which containextensive information on evolutionary sequence conservationand epigenetic modifications. We observed no peaks of enrichedsequence conservation or histone modifications characteristic ofregulatory elements over the rs8102137 SNP, which resides 6 kbupstream of CCNE1 (Fig. 2). There were no other regions withinthe 50 flanking region of CCNE1 that had any hallmarks ofregulatory elements and which also contained SNPs linked tothe rs8102137 bladder cancer risk GWAS SNP. Often, the GWAS-reported polymorphism serves as a chromosomal marker for arisk locus and does not necessarily contribute to disease pheno-types. Therefore, we expanded our analysis to a more thoroughexamination of CCNE1 and 19q12.

To identify active enhancers and regulatory elements withinthe CCNE1 locus of the bladder cancer genome, we performedhistone H3 lysine 27 acetylation (H3K27ac) chromatin immu-noprecipitation followed by high-throughput sequencing (ChIP-

Figure 2.

H3K27ac enrichment at the CCNE1 locus extends into the gene body. Top, a comparison of the ChIP-seq data generated from the bladder lines with ENCODE andother datasets available on the UCSC genome browser. DNase hypersensitivity (DHS) tracks from urothelial cells are shown below the ChIP-seq tracks andcorrespond to regions of H3K27ac enrichment. Bottom, H3K27ac enrichment peaks around the CCNE1 locus in 3 different bladder cancer cell lines. The yellowbar encompasses the UBC-specific peak of H3K27ac enrichment.






seq) in three bladder cancer cell lines (5637, J82, and T24 cells).The H3K27ac mark indicates active regulatory regions, includingpromoters and both distal and proximal enhancers (42). Wecompared our data to other previously published H3K27acChIP-seq datasets available from the ENCODE project (www.genome.ucsc.edu) to identify any bladder-specific enhancer-likeelements within theCCNE1 locus (Fig. 2). Interestingly, we founda bimodal peak of H3K27ac enrichment at the 50 end of the genethat corresponds to DNase1-Hypersensitivity Sites (DHS)highlighted in a urothelial cell line dataset (Fig. 2; ENCODE,UCSC Genome Browser, www.genome.ucsc.edu). The first peakimmediately upstream of theCCNE1 TSS ("Common") is presentin several ENCODE cell lines. However, the second peak ofH3K27ac ("UBC" in our ChIP-seq dataset) maps within the firstintron of the CCNE1 gene and is largely unique to the bladdercancer cell lines (Fig. 2). We therefore considered this putativeintronic enhancer for a role in CCNE1 regulation.

We next overlaid our ChIP-seq dataset with the disease-asso-ciated GWAS SNP and the LD block at CCNE1 to determine anyareas of overlap that could further highlight important regulatoryregions in bladder cancer. While noH3K27ac peak coincides withthe rs8102137 SNP 6 kb upstream of CCNE1 (Fig. 2), the nearbylinked SNP rs200996365 (A/�) falls within the "UBC" enhancersignal. We therefore examined this SNP for a potential causal rolein UBC risk.

The rs200996365 SNP (A/�) lies near the boundary of thesecond exon ofCCNE1 and is linked to the rs8102137GWAS SNPbased on allele frequencies within the LD block with r2 ¼ 0.890.Further analysis of this intronic DNA sequence revealed potentialtranscription factor binding sites in the vicinity. Specifically,motifanalysis searches using the JASPAR database showed thatrs200996365 falls in the middle of a consensus KLF5 motif (Fig.3A). Moreover, the higher risk allele (a 1-base pair deletion)predicts a very strong consensus KLF5 binding site (predictedbinding score of 12.471), whereas the lower risk A allele ispredicted to have less favorable binding, with a predicted bindingscore of 4.094 (Fig. 3A). Intriguingly, KLF5 is a transcriptionalactivator in the bladder epithelium, influences differentiation andproliferation of urothelial cells, and is also implicated in thetransformation of these cells to a malignant state (43, 44). KLF5is also a known downstream mediator of important signalingpathways such as H-Ras (45), a frequently altered pathway inUBC. Thus, we set out to determine the functional significance ofthis intragenic region and the rs200996365 SNP through a can-didate transcription factor approach beginning with KLF5.

We first assessed the occupancy of the KLF5 transcription factorwithin the CCNE1 intronic region through ChIP. We found thatKLF5 binds avidly to the predicted site in bladder cancer cells (Fig.3B; Supplementary Fig. S1). Notably, by assessing the allelicbinding specificity, we found that there was a >2-fold enrichmentin KLF5 binding to the higher risk allele (Fig. 3C). Furthermore, todetermine if KLF5 contributes to CCNE1 expression, we depletedKLF5 levels via transient transfection of siRNAs and found thatCCNE1 levels are reduced significantly by over 25% (Fig. 3D).These data strongly suggest that KLF5 contributes to CCNE1transcriptional activation.

The CCNE1 intronic region exhibits intrinsic enhancercapabilities

To further assess the activity of the CCNE1 intronic regulatoryelement, we conducted dual-luciferase assays with pGL3 vectors

containing the 350-bp regionwith the highest level ofH3K27ac inUBCs.We clonedboth the higher and lower risk intronic segmentsinto the pGL3-promoter vector and tested for intrinsic enhanceractivity. Normalized luciferase values revealed that both alleles(lower and higher risk genotypes) confer significant enhanceractivity (Fig. 4A). Additionally, upon exogenous overexpressionof KLF5, there was significant enrichment in luciferase activityfrom the higher risk allele (Supplementary Fig. S2).

The data above suggest that KLF5 regulates CCNE1 expressionin bladder cells via a novel polymorphic enhancer. However, wesought to validate the function of the intronic region in CCNE1gene regulation. To this end, we applied amore directed approachusing theCRISPR-Cas9–mediated genomeediting technique.Ourgoal was to mutate the enhancer in bladder cancer cells anddetermine any effect on CCNE1 expression. We designed twoCRISPR constructs that target the rs200996365 SNP and intro-duced them into 5637 cells.

We found that CRISPR-Cas9 targeting was efficient in inducingmutation of the CCNE1 intronic region. Cell lines were charac-terized by PCR and sequencing individual alleles. We found up tofive different alleles in clonal lines indicating that the cells arepolyploid for the CCNE1 gene region. Mutations varied from 2-base pair deletions and single-base-pairmismatches up to 63 basepair deletions (Fig. 4B).

Next, we analyzed CCNE1 mRNA levels in the mutated celllines to determine if any alterationwithin the intronic regulatoryregion resulted in a change in overall expression. Because eachCRISPR-mutated cell line harbored variable allelic genotypes,we observed varying levels of CCNE1 expression across theisolated clones (Fig. 4C). The A2 clone contained only twosingle-base-pair substitutions on two different alleles, neitherof which disrupted the predicted KLF5 binding site, and this lineexpressed wild-type levels of CCNE1 mRNA. In contrast, wefound that the expression levels in the CRISPR-mutated cell lineswere on average 30% lower than in the control, mimicking theobservations after KLF5 depletion with siRNA (Fig. 3D). TheCRISPR line D3 that contains no intact wild-type alleles demon-strates that this element contributes to almost 40% of CCNE1expression levels. Thus, CRISPR-mediatedmutation of the intro-nic enhancer supports its regulatory role for CCNE1 transcrip-tion in bladder cells.

DiscussionAlthough extensive studies have established a relationship

between UBC and environmental factors (7–9), the genetic com-ponents associated with disease progression still remain largelyunresolved. Our approach identified a novel regulatory elementwithin the bladder cancer genome and a putative causal geneticvariant involved in disease onset (Fig. 5). Ourmodel incorporatesChIP-seq analysis along with luciferase and CRISPR-Cas9 muta-genesis data that identify an intragenic enhancer that is crucial forproper CCNE1 transcriptional regulation. Our data also suggesthow the UBC-associated SNP at the CCNE1 locus contributes todisease predisposition. The recent focus on inter- and intra-genicDNA regulatory elements and how these regions contribute to amore complex network of gene regulation further emphasizes theputative importance of this intronic SNP within the "UBC" peakof enhancer marks (46, 47). While the GWAS-reported SNPappears to serve as a marker of this chromosomal location dueto the lack of sequence conservation and enhancer characteristics,

Pattison et al.





Figure 3.

The GWAS-linked SNP rs200996365 alters binding of KLF5 to this CCNE1 intronic region. A, rs200996365 maps to the center of a KLF5 binding site within the firstintron of CCNE1. The presence of the high-risk genotype at rs200996365 (�) predicts a higher affinity binding site—GGGAGGGG. Predicted binding scoresas calculated by JASPAR are shown beside each allele. B, ChIP analysis shows enrichment of KLF5 at the intronic enhancer region of CCNE1. The positivecontrol region is within the CCND1 promoter and the negative control region is in an intergenic region on chromosome 11 devoid of histone modifications. Data arepresented as a percentage of input and compared with IgG controls in this qPCR analysis. C, KLF5 preferentially binds the higher risk allele within CCNE1 asindicated by direct sequencing of the ChIP signal. Quantification of the ratio of sequencing dye traces at the bases indicated by a black star is depicted in the graph.Error bars represent SEM and asterisks indicate significance (� , P < 0.05). D, depletion of KLF5 results in significantly reduced CCNE1 expression. The 5637cells were transfected with two independent siRNAs targeting KLF5 compared with a scrambled control for 48 hours. KLF5 and CCNE1 mRNA levels were comparedwith Actin by qRT-PCR analysis. Error bars indicate SEM and asterisks indicate significance (�� , P < 0.005). The Western blot depicts the resulting proteinlevels of KLF5 upon siRNA knockdown with the two different siRNAs.






we find that a linked SNP 6 kb away from the GWAS variantappears to contribute to disease predisposition. We show thatthere is consistently more expression from the higher risk variantallele in different UBC cell lines, indicating that the regionharboring the higher risk polymorphism is a more potent tran-scriptional activator than the lower risk allele (Fig. 5). Upregula-tion of this cell-cycle regulatormay sensitize a normal cell to otheroncogenic hits, ultimately leading to cellular transformation. Thisis presumably the consequence of higher affinity binding of KLF5to the higher risk allele as shown by ChIP. Subtle enhancement ofCCNE1 expression from the higher risk variant could lead toenhanced cell proliferation or genomic instability and predispo-sition to UBC (29, 32).

It is well documented that KLF5 is a crucial downstreameffector of oncogenic H-Ras (45) and, specifically, KLF5 hasbeen implicated in bladder cancer development and transfor-mation, driving proliferation in urothelial cells (22, 43, 44).KLF5 interacts with chromatin modifiers and epigenetic med-iators, such as CBP and p300, both of which help promotetranscriptional activation and are also frequently mutated dur-ing oncogenesis (48, 49). Furthermore, in addition to beingoverexpressed in a variety of different cancer types, KLF5 isupregulated in 5% of bladder carcinomas and is frequentlyhypomethylated in cancers of the urinary tract (COSMICdatabase, http://cancer.sanger.ac.uk/cosmic; ref. 50). Thus, ele-vated levels of KLF5 in conjunction with increased binding to

Figure 4.

The CCNE1 intronic region exhibits enhancer capabilities, and mutation to this region results in a reduction in CCNE1 expression levels. A, the activity of the CCNE1enhancers containing either the low-risk allele (rs200996365/A) or the high-risk allele (rs200996365/�) was measured in a dual luciferase assay. Equalamounts of the pGL3 promoter only or pGL3 CCNE1-enhancer vectors were transfected into 5637 cells alongside a Renilla control. Firefly and Renilla activity weremeasured 48 hours after transfection and the graphs depict Firefly/Renilla measurements in relative luciferase units (RLUs). Error bars represent standarddeviation (SD), and asterisks indicate significance, � , P < 0.05 (NS, not significant). B, schematic of the CRISPR-Cas9 experimental design and a representation ofsequenced mutant alleles. WT denotes a wild-type allele (r, high-risk allele; n, low-risk allele). C, qRT-PCR analysis of CCNE1 mRNA levels compared withActin in genotypedCRISPR-mutated cell clones. A2 contains awild-type sequence; therefore, all expression levels are relative to this cell line. Amutant allele refers toa sequenced allele with at least a 2-base-pair deletion.

Pattison et al.





this polymorphic intronic region subsequently drive enhancedCCNE1 expression (29, 32).

In understanding the contribution of the 19q12 risk variants toUBC predisposition, our molecular studies augment the growingknowledge on specific genetic variations and how they confer anincreased risk for disease. Further correlation of these risk-asso-ciated variants with cis-regulatory mechanisms will provide aclearer image of the basic underpinnings of many cancers andother common diseases.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

DisclaimerThe content is solely the responsibility of the authors and does not neces-

sarily represent the official views of the NIH.

Authors' ContributionsConception and design: J.M. Pattison, M.D. ColeDevelopment of methodology: M.D. ColeAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): J.M. Pattison, V. Posternak

Analysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): J.M. Pattison, V. Posternak, M.D. ColeWriting, review, and/or revisionof themanuscript: J.M. Pattison, V. Posternak,M.D. ColeAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): M.D. Cole

AcknowledgmentsWe thank David Robbins for generously providing cell lines. We also thank

members of the Cole lab for thoughtful discussion and the Dartmouth Geno-mics Shared Resources core for help with the high throughput sequencingprocedures.

Grant SupportThis work was supported by a grant from the National Cancer Institute

(CA080320; M.D. Cole) and by a National Institutes of Health Training GrantAward (T32GM8704-12S1; J.M. Pattison).

The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received April 12, 2016; revised July 4, 2016; accepted July 28, 2016;published OnlineFirst August 11, 2016.

References1. Dinney CP, McConkey DJ, Millikan RE, Wu X, Bar-Eli M, Adam L, et al.

Focus on bladder cancer. Cancer Cell 2004;6:111–6.2. Rosenberg JE, HahnWC. Bladder cancer: modeling and translation. Genes

Dev 2009;23:655–9.3. Cordon-Cardo C, Wartinger D, Petrylak D, Dalbagni G, Fair WR, Fuks Z,

et al. Altered expression of the retinoblastoma gene product: prognosticindicator in bladder cancer. J Natl Cancer Inst 1992;84:1251–6.

4. Cote RJ, Dunn MD, Chatterjee SJ, Stein JP, Shi SR, Tran QC, et al. Elevatedand absent pRb expression is associated with bladder cancer progressionand has cooperative effects with p53. Cancer Res 1998;58:1090–4.

5. Presti JCJr, Reuter VE,GalanT, FairWR,Cordon-CardoC.Molecular geneticalterations in superficial and locally advanced human bladder cancer.Cancer Res 1991;51:5405–9.

6. Stein JP, GinsbergDA, Grossfeld GD, Chatterjee SJ, Esrig D, DickinsonMG,et al. Effect of p21WAF1/CIP1 expression on tumor progression in bladdercancer. J Natl Cancer Inst 1998;90:1072–9.

7. Fei DL, Li H, Kozul CD, Black KE, Singh S, Gosse JA, et al. Activation ofHedgehog signaling by the environmental toxicant arsenic may con-tribute to the etiology of arsenic-induced tumors. Cancer Res 2010;70:1981–8.

8. KaragasMR, Tosteson TD,Morris JS,Demidenko E,Mott LA,Heaney J, et al.Incidence of transitional cell carcinomaof the bladder and arsenic exposurein New Hampshire. Cancer Causes Control 2004;15:465–72.

9. Lesseur C, Gilbert-Diamond D, Andrew AS, Ekstrom RM, Li Z, Kelsey KT,et al. A case-control study of polymorphisms in xenobiotic and arsenic

metabolism genes and arsenic-related bladder cancer in New Hampshire.Toxicol Lett 2012;210:100–6.

10. Schwender H, Selinski S, Blaszkewicz M, Marchan R, Ickstadt K, Golka K,et al. Distinct SNP combinations confer susceptibility to urinary bladdercancer in smokers and non-smokers. PLoS One 2012;7:e51880.

11. Billerey C, Chopin D, Aubriot-Lorton MH, Ricol D, Gil Diez de Medina S,Van Rhijn B, et al. Frequent FGFR3 mutations in papillary non-invasivebladder (pTa) tumors. Am J Pathol 2001;158:1955–9.

12. Cappellen D, De Oliveira C, Ricol D, de Medina S, Bourdin J, Sastre-GarauX, et al. Frequent activating mutations of FGFR3 in human bladder andcervix carcinomas. Nat Genet 1999;23:18–20.

13. Czerniak B, Cohen GL, Etkind P, Deitch D, Simmons H, Herz F, et al.Concurrent mutations of coding and regulatory sequences of the Ha-rasgene in urinary bladder carcinomas. Hum Pathol 1992;23:1199–204.

14. Guo G, Sun X, Chen C, Wu S, Huang P, Li Z, et al. Whole-genome andwhole-exome sequencing of bladder cancer identifies frequent alterationsin genes involved in sister chromatid cohesion and segregation. Nat Genet2013;45:1459–63.

15. Zieger K, Dyrskjot L, Wiuf C, Jensen JL, Andersen CL, Jensen KM, et al. Roleof activating fibroblast growth factor receptor 3 mutations in the devel-opment of bladder tumors. Clin Cancer Res 2005;11:7709–19.

16. Linnenbach AJ, Pressler LB, Seng BA, Kimmel BS, Tomaszewski JE,Malkowicz SB. Characterization of chromosome 9 deletions in transi-tional cell carcinoma by microsatellite assay. Hum Mol Genet 1993;2:1407–11.

Figure 5.

A model for CCNE1 regulation by KLF5 in urothelialcells heterozygous for the GWAS SNP rs8102137.KLF5 binds both the low-risk and high-riskalleles of CCNE1 to drive gene transcription, yetthere is increased occupancy on the high-riskallele. This results in enhanced CCNE1 expressionfrom the risk allele.






17. Figueroa JD, Ye Y, Siddiq A, Garcia-Closas M, Chatterjee N, Prokunina-Olsson L, et al. Genome-wide association study identifies multiple lociassociated with bladder cancer risk. Hum Mol Genet 2014;23:1387–98.

18. Cortessis VK, Yuan JM, VanDenBergD, Jiang X,Gago-DominguezM, SternMC, et al. Risk of urinary bladder cancer is associated with 8q24 variantrs9642880[T] in multiple racial/ethnic groups: results from the LosAngeles-Shanghai case-control study. Cancer Epidemiol Biomarkers Prev2010;19:3150–6.

19. Kiemeney LA, Thorlacius S, Sulem P, Geller F, Aben KK, Stacey SN, et al.Sequence variant on 8q24 confers susceptibility to urinary bladder cancer.Nat Genet 2008;40:1307–12.

20. Rothman N, Garcia-Closas M, Chatterjee N, Malats N, Wu X, Figueroa JD,et al. A multi-stage genome-wide association study of bladder canceridentifies multiple susceptibility loci. Nat Genet 2010;42:978–84.

21. Wu X, Ye Y, Kiemeney LA, Sulem P, Rafnar T, Matullo G, et al. Geneticvariation in the prostate stem cell antigen gene PSCA confers susceptibilityto urinary bladder cancer. Nat Genet 2009;41:991–5.

22. Del Pizzo JJ, Borkowski A, Jacobs SC, Kyprianou N. Loss of cell cycleregulators p27(Kip1) and cyclin E in transitional cell carcinoma of thebladder correlates with tumor grade and patient survival. Am J Pathol1999;155:1129–36.

23. Fu YP, Kohaar I, Moore LE, Lenz P, Figueroa JD, Tang W, et al. The 19q12bladder cancer GWAS signal: association with cyclin E function andaggressive disease. Cancer Res 2014;74:5808–18.

24. Kitahara K, Yasui W, Kuniyasu H, Yokozaki H, Akama Y, Yunotani S, et al.Concurrent amplification of cyclin E and CDK2 genes in colorectal carci-nomas. Int J Cancer 1995;62:25–8.

25. Lin L, Prescott MS, Zhu Z, Singh P, Chun SY, Kuick RD, et al. Identificationand characterization of a 19q12 amplicon in esophageal adenocarcinomasreveals cyclin E as the best candidate gene for this amplicon. Cancer Res2000;60:7021–7.

26. MaroneM, Scambia G, Giannitelli C, Ferrandina G, Masciullo V, BellacosaA, et al. Analysis of cyclin E andCDK2 in ovarian cancer: gene amplificationand RNA overexpression. Int J Cancer 1998;75:34–9.

27. Richter J, Wagner U, Kononen J, Fijan A, Bruderer J, Schmid U, et al. High-throughput tissue microarray analysis of cyclin E gene amplification andoverexpression in urinary bladder cancer. Am J Pathol 2000;157:787–94.

28. Snijders AM, Nowee ME, Fridlyand J, Piek JM, Dorsman JC, Jain AN, et al.Genome-wide-array-based comparative genomic hybridization revealsgenetic homogeneity and frequent copy number increases encompassingCCNE1 in fallopian tube carcinoma. Oncogene 2003;22:4281–6.

29. GengY, YuQ, SicinskaE,DasM, Schneider JE, Bhattacharya S, et al. Cyclin Eablation in the mouse. Cell 2003;114:431–43.

30. OkudaM, Horn HF, Tarapore P, Tokuyama Y, Smulian AG, Chan PK, et al.Nucleophosmin/B23 is a target of CDK2/cyclin E in centrosome duplica-tion. Cell 2000;103:127–40.

31. Parisi T, Beck AR, Rougier N, McNeil T, Lucian L, Werb Z, et al. Cyclins E1and E2 are required for endoreplication in placental trophoblast giant cells.EMBO J 2003;22:4794–803.

32. Spruck CH, Won KA, Reed SI. Deregulated cyclin E induces chromosomeinstability. Nature 1999;401:297–300.

33. Kamai T, Takagi K, Asami H, Ito Y, Oshima H, Yoshida KI. Decreasing ofp27(Kip1)and cyclin E protein levels is associated with progression fromsuperficial into invasive bladder cancer. Br J Cancer 2001;84:1242–51.

34. Schraml P, Bucher C, Bissig H, Nocito A, Haas P, Wilber K, et al. Cyclin Eoverexpression and amplification in human tumours. J Pathol 2003;200:375–82.

35. Ohtani K, DeGregori J, Nevins JR. Regulation of the cyclin E gene bytranscription factor E2F1. Proc Natl Acad Sci U S A 1995;92:12146–50.

36. Cha RS, Zarbl H, Keohavong P, Thilly WG. Mismatch amplificationmutation assay (MAMA): application to the c-H-ras gene. PCR MethodsAppl 1992;2:14–20.

37. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome.Genome Biol 2009;10:R25.

38. Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, et al.Model-based analysis of ChIP-Seq (MACS). Genome Biol 2008;9:R137.

39. Feng J, Liu T, Zhang Y. Using MACS to identify peaks from ChIPSeq data.Curr Protoc Bioinformatics 2011;Chapter 2:Unit 2.14.

40. Sanjana NE, Shalem O, Zhang F. Improved vectors and genome-widelibraries for CRISPR screening. Nat Methods 2014;11:783–4.

41. Wright JB, BrownSJ, ColeMD.Upregulationof c-MYC in cis through a largechromatin loop linked to a cancer risk-associated single-nucleotide poly-morphism in colorectal cancer cells. Mol Cell Biol 2010;30:1411–20.

42. Heintzman ND, Hon GC, Hawkins RD, Kheradpour P, Stark A, Harp LF,et al. Histone modifications at human enhancers reflect global cell-type-specific gene expression. Nature 2009;459:108–12.

43. Bell SM, Zhang L, Mendell A, Xu Y, Haitchi HM, Lessard JL, et al. Kruppel-like factor 5 is required for formation and differentiation of the bladderurothelium. Dev Biol 2011;358:79–90.

44. Chen C, Benjamin MS, Sun X, Otto KB, Guo P, Dong XY, et al. KLF5promotes cell proliferation and tumorigenesis through gene regulationand the TSU-Pr1 human bladder cancer cell line. Int J Cancer 2006;118:1346–55.

45. Nandan MO, Yoon HS, Zhao W, Ouko LA, Chanchevalap S, Yang VW.Kruppel-like factor 5 mediates the transforming activity of oncogenic H-Ras. Oncogene 2004;23:3404–13.

46. KowalczykMS,Hughes JR, GarrickD, LynchMD, Sharpe JA, Sloane-StanleyJA, et al. Intragenic enhancers act as alternative promoters. Mol Cell2012;45:447–58.

47. OttCJ, BlackledgeNP,Kerschner JL, Leir SH,CrawfordGE,CottonCU, et al.Intronic enhancers coordinate epithelial-specific looping of the activeCFTR locus. Proc Natl Acad Sci U S A 2009;106:19934–9.

48. Gui Y, Guo G, Huang Y, Hu X, Tang A, Gao S, et al. Frequent mutations ofchromatin remodeling genes in transitional cell carcinoma of the bladder.Nat Genet 2011;43:875–8.

49. Zhang Z, TengCT. Phosphorylation of Kruppel-like factor 5 (KLF5/IKLF) atthe CBP interaction region enhances its transactivation function. NucleicAcids Res 2003;31:2196–208.

50. Forbes SA, Beare D, Gunasekaran P, Leung K, Bindal N, Boutselakis H, et al.COSMIC: exploring theworld's knowledge of somaticmutations inhumancancer. Nucleic Acids Res 2015;43(Database issue):D805–11.


Pattison et al.




2016;14:1078-1086. Published OnlineFirst August 11, 2016.Mol Cancer Res Jillian M. Pattison, Valeriya Posternak and Michael D. Cole Enhancer to Confer Increased Bladder Cancer RiskTranscription Factor KLF5 Binds a Cyclin E1 Polymorphic Intronic

Updated version

10.1158/1541-7786.MCR-16-0123doi:

Access the most recent version of this article at:

Material

Supplementary

http://mcr.aacrjournals.org/content/suppl/2016/08/11/1541-7786.MCR-16-0123.DC1

Access the most recent supplemental material at:

Cited articles

http://mcr.aacrjournals.org/content/14/11/1078.full#ref-list-1

This article cites 49 articles, 12 of which you can access for free at:

Citing articles

http://mcr.aacrjournals.org/content/14/11/1078.full#related-urls

This article has been cited by 1 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://mcr.aacrjournals.org/content/14/11/1078To request permission to re-use all or part of this article, use this link



http://mcr.aacrjournals.org/lookup/doi/10.1158/1541-7786.MCR-16-0123

http://mcr.aacrjournals.org/content/suppl/2016/08/11/1541-7786.MCR-16-0123.DC1

http://mcr.aacrjournals.org/content/14/11/1078.full#ref-list-1

http://mcr.aacrjournals.org/content/14/11/1078.full#related-urls

http://mcr.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://mcr.aacrjournals.org/content/14/11/1078


Documents

Transcription Factor KLF5 Binds a Cyclin E1 Polymorphic Intronic Enhancer to Confer ... · Genomics Transcription Factor KLF5 Binds a Cyclin E1 Polymorphic Intronic Enhancer to Confer