Cancer Genes and Networks

Epigenetic Regulation of Dlg1, via Kaiso, AltersMitotic Spindle Polarity and Promotes IntestinalTumorigenesisMadeleine A. Young1, Stephanie May1, Angelos Damo1, Young So Yoon2,Man-Wook Hur2,Wojiech Swat3, and Lee Parry1

Abstract

Both alterations to the epigenome and loss of polarity havebeen linked to cancer initiation, progression, and metas-tasis. It has previously been demonstrated that loss ofthe epigenetic reader protein Kaiso suppresses intestinaltumorigenesis in the Apcþ/min mouse model, in whichaltered polarity plays a key role. Thus, we investigatedthe link between Kaiso deficiency, polarity, and suppres-sion of intestinal tumorigenesis. We used Kaiso-deficientmice to conditionally delete Apc within the intestinalepithelia and demonstrated upregulation of the spindlepolarity genes Dlg1 and Dlgap1. To understand the roleof Dlg1, we generated Villin-creApcþ/minDlg1flx/flx Kaiso�/y

mice to analyze gene expression, survival, tumor bur-den, and spindle orientation. In vivo analysis of theDlg1-deficient intestine revealed improper orientation ofmitotic spindles and a decreased rate of cellularmigration.

Loss of Dlg1 decreased survival in Apcþ/min mice, validating itsrole as a tumor suppressor in the intestine. Significantly, theincreased survival of Apcþ/minKaisoy/� mice was shown to bedependent on Dlg1 expression. Taken together, these dataindicate that maintenance of spindle polarity in the intestinalcrypt requires appropriate regulation of Dlg1 expression. AsDlg1 loss leads to incorrect spindle orientation and a delay incells transiting the intestinal crypt. We propose that thedelayed exit from the crypt increase the window in whichspontaneous mutations can become fixed, producing a"tumor-permissive" environment, without an increase inmutation rate.

Implications: Loss ofmitotic spindle polarity delays the exit ofcells from the intestinal crypt and promotes a tumorigenicenvironment.

IntroductionRecent work recognizing the importance of the link between

polarity and epigenetic pathways in cancer onset and progression(1) has highlighted a need to increase our understanding of theserelationships. DNA methylation of CpG dinucleotides is a fun-damental epigenetic modification. Within gene promoters' meth-ylation leads to transcriptional silencing and is exploited by arange of different cancer types, including colorectal cancer, tosilence tumor suppressor genes (2, 3). There is now a growingbody of evidence that aberrant epigenetic silencing of polaritygenes influences tumorigenesis in a variety of settings (4–6).Mechanistically, methylated CpG dinucleotides are recognizedand bound to by proteins from the methyl binding family, whichthen recruit transcriptional repressor complexes (7). These pro-

teins play a key role in disease by identifying tumor suppressorgenes that have been aberrantly methylated (2, 3, 8). The methylbinding proteinKaiso, a POZ/BTB (BroadComplex Tramtrak, Brica brac/pox virus, and zinc finger) family zinc finger transcriptionfactor protein, has been shown to play a key role in intestinaltumorigenesis. POZ/BTB proteins are a well conserved proteinfamily (9), which have been indicated to play a role in bothdevelopment and cancer (10, 11). Despite the requirement ofKaiso for amphibian development (11),Kaiso nullmice are viableand do not show any notable abnormalities in either develop-ment or reproduction (12). In cancer cells, Kaiso appears to have apro-tumorigenic function. Its deficiency relieved repression oftumor suppressor and DNA repair genes (13, 14) in colon cancer.Although its nuclear localization is indicative of higher grade andmetastatic breast and prostate cancers (15). In addition to its rolesas a transcriptional repressor, it can also bind to d-Catenin andbehave as a transcriptional activator (16, 17). Further, in vivostudies have shown that Kaiso overexpression promotes intestinalinflammation and intestinal tumorigenesis (18, 19), whereas itsabsence increased survival of Apcþ/min mice by delaying the onsetof intestinal tumorigenesis (12). The Apcþ/min mouse is a modelof the human disease familial adenomatous polyposis (FAP).InbothApcþ/minmice andpatientswith FAP,mutationof thewild-type Apc allele (or `second hit') is required for adenoma forma-tion (20). APCmutation and subsequent activation of the canon-ical Wnt pathway is an early, if not the first step, in the develop-ment ofmore than 80%of all sporadic colorectal cancers (21, 22).Because of its role in mediating epigenetic silencing of tumorsuppressor gene in human colorectal cancer (13) and the low

1European Cancer Stem Cell Research Institute, Cardiff School of Biosciences,Cardiff University, Cathays, Cardiff, United Kingdom. 2Department ofBiochemistry and Molecular Biology, Yonsei University School of Medicine,Yonsei-Ro, Seodaemoon-Ku, Seoul, Republic of Korea. 3Department ofPathology & Immunology, Washington University School of Medicine, Missouri.

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

Corresponding Author: Lee Parry, Cardiff University, Hadyn Ellis Building,Maindy Road, Cardiff CF24 4HQ, Wales, UK. Phone: 44-029-2068-8017; E-mail:parryl3@cardiff.ac.uk

doi: 10.1158/1541-7786.MCR-18-0280

�2018 American Association for Cancer Research.

MolecularCancerResearch

Mol Cancer Res; 17(3) March 2019686

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280

frequency ofmutations observed in the disease (0%observed in 72cases; refs. 23, 24), Kaiso represents a potentially druggable target.However, the mechanism of intestinal tumor suppression due toKaisodeficiency in vivohas yet tobeexplored.As thedividing cells inthe precancerous and cancerous intestinal tissue of Apcþ/min micedisplay alterations to the spindle polarity (25), we have investi-gated whether alterations to spindle polarity play a role in theintestinal tumor suppression observed in the Apcþ/minKaisoy/�

mouse model. We have demonstrated that the polarity and tumorsuppressor geneDlg1 is mis-regulated in the absence ofKaiso. Dlg1forms part of the Scribble/Lgl/Dlg polarity complex, which hasbeen shown to play an important role in the maintenance ofepithelial integrity (26).Mutations in this complex associatedwiththe progression of a range of different epithelial cancer types (27–29) andmutations within theDlg family have been observed in 12of 72 (16.7%) of patients with colorectal cancer (23, 24). Wefurther demonstrate that the absenceofDlg1alters spindle polarity,delays migration, and promotes tumorigenesis and progression inthe Apcþ/min and Apcþ/minKaisoy/� setting. Taken together, thisincreases the evidence demonstrating the importance of epigeneticregulation inmaintenance of spindle polarity in intestinal homeo-stasis and disease.

Materials and MethodsMice

All animal procedures were conducted in accordance withinstitutional animal care guidelines and UK Home Office regula-tions. Mice were maintained in a SPF barrier facility in conven-tional open top cages on Eco-Pure Chips 6 Premium bedding(Datesand) under a 12-hour light cycle, with IPS 5008 diet(Labdiet-IPS Ltd.) provided for nutritional support. To enrichenvironment, we provided irradiated sunflower seeds (at weaningonly), Techniplast mouse houses (Techniplast) and small chew-sticks (Labdiet-IPS Ltd.). All mice were from amixed backgroundand were homozygous with respect to the C57Bl/6 Pla2g2a (alsocalled Mom-1) allele. Experimental animals were between 10and 15weeks old with siblings used as controls. The alleles for theAh-cre (30), Apcflx (31), Apcþ/min (32), Villin-Cre (33), Dlg1flx (34),Kaisoy/� (12) have been described previously. Genotyping condi-tions available upon request. Induction of the Ah-cre transgenewas performed by administering three intraperitoneal injectionsof b-naphthoflavone (BNF; Sigma) at 80 mg/kg in a 24-hourperiod. For BrdU labeling, mice were injected intraperitoneallywith 0.15 mL of BrdU. Following cre induction loxP flankedalleles are referred to as deleted in intestinal epithelial cells(DIEC).

Microarray expression analysisTen-week-old male mice were used for the array analysis.

Three mice for each of the desired genotypes were used. Three-centimeter portions of the SI located 5 cm from the stomach wereplaced in RNAlater (after removing any mesentery and ensuringthat no Peyer's patches were present). The tissue was then homog-enized in Trizol reagent and RNA extracted using standard phe-nol–chloroformmethodologies. Samples were sent to the Molec-ular Biology Core facility at the Cancer Research UK MolecularBiology Core Facility at The University of Manchester. Wherebiotinylated target cRNA was generated and hybridized to Affy-metrix Mouse430A_2 gene expression chips. The raw data fromthese microarrays is available at http://www.ebi.ac.uk/arrayex

press/. The data were normalized and analyzed to generate dif-ferential expression tables using the AffylmGUI package for linearmodeling of microarray data (35).

Tissue isolation, reporter visualization, IHCMice were euthanized at specified time points and the small

intestine (SI) removed and flushed with water. Intestines weredissected as follows: The proximal 7 cm was mounted, fixedovernight in 10% formalin, and paraffin embedded. The follow-ing 3 cmwas opened and placed into RNA later (Sigma), ensuringthat all mesentery and Peyers patches were removed. The follow-ing 5 cm was divided into 1-cm lengths, bundled using surgicaltape, and then fixed in 4% formaldehyde at 4�C for no morethan 24 hours before processing into wax blocks by conventionalmeans. Section were cut at 5 mm thickness, dewaxed, and rehy-drated into PBS. Staining was performed using the Envisionþmouse or Rabbit Kit (Dako; Agilent Ltd.) according to manufac-turer's instructions. To identify cells which had lost Dlg1 westained using a validated rabbit polyclonal anti-Dlg1 antibody(Catalog No. #STJ111281; St John's Antibodies) at 1:200. Cellswhich had incorporated BRDUwere identifiedusing an anti-BrdUantibody conjugate at 1:50, cells positive for DNAdamagemarkergH2AX were identified using anti-gH2AX (Upstate 05636) Cel-lular analysis was performed on >25 whole crypts from at leastthree mice of each genotype.

Cellular analysisCellular analysis was performed on >25 whole crypts from at

least three mice of each genotype. Apoptotic and mitotic indexwere scored from hematoxylin-and-eosin (H&E)-stained sectionsas previously described (36). The cells between the base ofthe crypt and the junction with the villus was designated as theproliferative zone. Formigration analysis,mice of 60 to 80 days ofage were given an intraperitoneal injection of BrdU 2 or 24 hoursprior to culling and dissection. IHC analysis for BrdU incorpo-rationwas performed on formalin-fixed small intestinal rolls, andthe number BrdU-positive cells and their location (with 0 beingthe bottom of the crypt) was measured on 50 half-crypts permouse, minimum of four mice. Statistical analysis of the cumu-lative frequency of positive cells was performed using a two-tailedKolmogorov–Smirnov test, on graphs P values are indicated asfollows: �, P < 0.05; ��, P < 0.01; ���, P < 0 0.001. Measurement ofmitotic angle was performed on H&E slides stained sections. Theangles of 200 mitosis in which the two spindle poles were easilyobserved weremeasured permouse, with the angle of a direct linedrawnbetween the two spindle poles and the line of the basementmembrane measured. Measurements of 0� to 45� and 135� to180� were classified as "planar," being parallel to the basementmembrane, whereas measurements of 45� to 135� were classifiedas being "Apico-basal," being perpendicular to the basementmembrane anddividing in the orientation of apico-basal polarity.The Chi-squared test was performed to assess statistical differ-ences between the two genotypes, on graphsP values are indicatedas follows: �, P < 0.05; ��, P < 0.01; ���, P < 0.001.

Chromatin immunoprecipitation assaysChromatin immunoprecipitation (ChIP) assays were per-

formed to assess Kaiso binding to the promoters of Dlgap1 andDlg1 in murine samples from aged Apcþ/min mice, using polypsand adjacent normal tissue from the SI and the large intestine.Anti-Kaiso antibody (sc-23871) was used to identify DNA

Dlg1 Suppresses Tumors by Maintaining Spindle Polarity

www.aacrjournals.org Mol Cancer Res; 17(3) March 2019 687

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280

binding partners of Kaiso, and primers for Dlgap1 and Dlg1 wereused to confirm presence of these genes (primers detailed inSupplementary Fig. S2).

siRNA against ApcmRNAwas purchased from Bioneer. The ApcsiRNA (sense 50-GCAGGCGUAGAGUAUUCAU-30, antisense 50-AUGAAUACUCUACGCCUGC-30) was transfected into CT-26cells using LipofectamineRNAiMax (Invitrogen), and knockdownlevels of were confirmed using qRT-PCR (available upon request).CT-26 cells at passage 86 were purchased from the Korean CellLine Bank and certified mycoplasma free and verified by STRprofiling. The cells were expanded to create a cell stock and frozen.For each experiment, cells were thawed and used at <3 passages orwithin 2 weeks. After transfection, the cells were harvested, andChIP analysis performed.

qRT-PCR analysisThe following methods were all performed according to man-

ufacturer's instructions unless otherwise stated. For analysis ofgene expression in the intestine epithelia, three to five mice fromeach control and experimental group were harvested. RNA wasextracted froma0.5 cmportionof SI taken�5 cmdistally from thestomach and stored at �80 in RNAlater. Total RNA was extractedusing the RNeasy Kit (Qiagen) and DNase treated using theTurbo DNase Kit (Ambion). Complimentary DNA (cDNA) wastranscribed from 1 mg of RNA using random hexamers (Promega)and Superscript III (Invitrogen) kits. For relative quantitation,all samples were run in duplicate on the StepOnePlus PCRmachine using Fast Sybr Green master mix (Applied Biosystems).The threshold cycle (Ct) values of each gene analyzedwere normalized to a reference gene. For expression analysis(qRT-PCR) Ct values were normalized against the ActB gene.Differences between groups were assessed using the 2-DDCTmethod [47]. Two-tailed Mann–Whitney U tests were performedon the DCt values to determine significance (P� 0.05) differencesbetween groups. Oligonucleotide sequences are available uponrequest.

SurvivalAminimum of 14mice per cohort were used to assess survival.

Mice were aged until they showed symptoms of intestinal tumorburden (pale feet, bloating, prolapse or piloerection). Survivaldata were analyzed using the Kaplan–Meier test. If not indicatedotherwise, the statistical mean is presented, and error bars repre-sent standard deviation, on graphs P values are indicated asfollows: �, P < 0.05; ��, P < 0.01; ���, P < 0.001.

Tumor burden and gradingTumor burdenwasmeasured at point of deathwith the number

of macroscopic lesions and their size from fresh tissue. Gradingwas performedmicroscopically using H&E slides of SI taken fromagedmice, tumors were microscopically identified as; single cryptlesion (grade 1), microadenoma (grade 2), adenoma (grade 3),adenocarcinoma with stromal invasion (grade 4), and adenocar-cinoma with smooth muscle invasion (grade 5).

Intestinal organoid cultureIntestinal organoid culture was performed using the method

described by Sato and colleagues (37). Organoids were grown inNunc LabTek Chamber Well slides for whole mount immuno-fluorescence. Organoids were fixed overnight in 4% PFA, thenpermeabilized using 1% Triton-X100 for 1 hour at 37�C. Alexa

Fluor Phalloidin at 1:150 and neomarkers anti-lysozyme(Rb-327-A1) at 1:300 were used as primary antibodies.

ResultsApc-deficient SI crypts are resistant to Kaiso deficiency

Thedelayed onset of adenomas in theApcþ/minKaisoy/�mouse ispotentially due to a role for Kaiso in the Wnt signaling pathway.To investigate the direct effects of Kaiso status upon geneexpression and Wnt signaling, we analyzed the phenotype of aconditional deletion of Apc in the context of Kaiso deficiency.Kaisoy/�-deficient mice were crossed with mice carrying an Ah-cretransgene and a loxP-flanked Apc allele, following induction withBNF Apc is deleted in the SI crypts. We generated and comparedcohorts of Ah-creþ Kaisoþ/þ (WT), Ah-creþ Kaiso�/y (Kaiso�/y),Ah-creþ Apcflx/flx Kaisoþ/þ(ApcDIEC/DIEC), and Ah-creþ Apcflx/flx

Kaisoy/� (ApcDIEC/DIECKaiso�/y) mice. At 4 days post-induction(d.p.i.), comparing WT and Kaiso�/y indicated the absence ofKaiso does not grossly alter the crypts (Fig. 1A and B). Similarly,the ApcDIEC/DIEC mice, which are characterized by enlarged, aber-rant crypts (Fig. 1C; as previously reported; ref. 38), appearedunaltered inApcDIEC/DIECKaisoy/� setting (Fig. 1D). To examine celldivision and migration, mice were killed 2 or 24 hours afterinjection with BrdU. BrdU is incorporated into the DNA of cellsat S phase and is bioavailable for less than 2 hours. Therefore, bykilling mice at 2 and 24 hours after exposure, we could track themovement of BrdU-positive cells from the proliferation zone inthe crypt on to the villus by IHC. The position of BrdU-positivecells was plotted as the cumulative frequency of cells 2 and 24hours after labeling (Fig. 1E–H).Migration patterns of the SI werenot significantly altered by the absence ofKaiso; however, the dataindicated a trend towards an increase in the speed of migration ofKaiso-deficient cells from the proliferative zone onto the villi. InApcDIEC/DIECmice, cell proliferation occurs independently of posi-tion andmigration of cells along the crypt–villus axis is perturbed(Fig. 1F), as previously reported (27), the additional loss of Kaisofailed to alter this phenotype (Fig. 1F).

Further, it has been previously shown that Paneth cell locali-zation is altered in the inducedApcDIEC/DIEC SI (27), the additionalloss of Kaiso had no effect on this phenotype (Fig. 1I). Finally,analysis ofWnt target genes demonstrated that the increase inWnttarget gene expression synonymous with Apc deletion was notsignificantly altered in the absence of Kaiso (Fig. 1J; Supplemen-tary Fig. S1). In conclusion, the lack of any gross phenotype ineither the Kaisoy/� or the ApcDIEC/DIECKaisoy/� mice suggested thatrepression of intestinal tumorigenesis in theApcþ/minKaisoy/�miceis Wnt independent.

The polarity genes Dlg1 andDlgap1 are upregulated within theKaiso-deficient SI

AsKaisodeficiency iswell toleratedby the intestine,we aimed toidentify novel roles for Kaiso in the normal and Apc-deficientintestine which may influence tumorigenesis. To identify genesmis-regulated in the absence of Kaiso, we generated microarraygene expression profiles on SI from WT and Kaiso�/y mice (rawdata are available from https://www.ebi.ac.uk/arrayexpress/express under accession number E-MTAB-6009). After normali-zation genes differentially expressed were identified and rankedaccording to their statistical significance (Supplementary TableS1). As expected, the genes Kaiso and Hprt which were deleted inthe generation of the model were highly downregulated and the

Young et al.

Mol Cancer Res; 17(3) March 2019 Molecular Cancer Research688

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280

most significantly altered. As Kaiso is a mediator of epigenetictranscriptional repression, we identified the genes that wereupregulated by its loss and candidates for a role in tumorigenesis.The analysis identified the gene Discs Large Homologue Associ-ated Protein 1 (Dlgap1; Supplementary Table S1), which wassignificantly 4.9-fold upregulated in theKaiso�/y intestine. Dlgap1is a highly conserved protein, which interacts with the guanylatekinase-like domain of the important cancer regulatory proteinDiscs Large Homolog 1 (Dlg1), a binding partner of the keyWnt regulator Apc. To support the evidence that these genesare regulated by Kaiso, we analyzed whether the sequence ofthe promoter regions of these genes contained the Kaisobinding sequences TCCTGCNA (minimum binding sequence ofCTGCNA) and CGCG (39). In silico analysis of the 5kb promoterregions upstream of the Dlg1 and Dlgap1 ATG start codonsconfirmed the presence of five minimum Kaiso binding motifsand one complete Kaiso bindingmotif inDlg1 and oneminimum

Kaiso binding motif in theDlgap1 promoter region, indicating itspotential to regulate expression of these genes (SupplementaryFig. S2). To confirm the upregulation of gene expression, we usedqRT-PCR analysis to demonstrate a significant 1.5- and 3.3-foldupregulation of Dlgap1 and Dlg1 expression, respectively(Fig. 2A), in the Kaiso-deficient intestine. This was further con-firmed by IHC which showed a dramatic increase in Dlg1 proteinlevels within the intestinal epithelium of Kaiso-deficient mice(Fig. 1B and C). Dlg1 is important in maintenance of spindlepolarity and is a binding partner of Apc. As spindle polarity isaltered in the Apcþ/min intestine (25), the upregulation of Dlg1suggested that it may play a functional role in suppression ofintestinal tumorigenesis and the increased survival of Apcmin

Kaisoy/� mice. To confirm a direct role for Kaiso to regulate Dlg1and Dlgap1 expression, ChIP was utilized to pull-down DNAassociated with the Kaiso protein in both normal and polyptissue from the small and large intestine of Apcmin mice. Kaiso

Figure 1.

Loss of Kaiso does not alter the gross phenotype ofWT and Apc-deficient SI crypts. H&E staining showing no difference betweenWT (A) and Kaiso�/y (B) orAh-cre ApcDIEC/DIEC (C) and Ah-creApcDIEC/ DIEC Kaiso�/y (D) SI, note the enlarged crypts in the ApcDIEC/DIECmice are retained in ApcDIEC/DIECKaiso�/y ([indicatesheight of crypts). Scale bars represent 50 mm. E and F,Mice were injected with BrdU to mark cells in S phase at 2 hours and track migration onto villi 24 hourslater. E,WT 2h (——), Kaiso�/y 2h (---), WT 24h (���), Kaiso�/y 24h (-�-�) and (F) ApcDIEC/DIEC 2h (——), ApcDIEC/DIEC Kaiso�/y 2h (---), ApcDIEC/DIEC 24h (���),ApcDIEC/DIEC Kaiso�/y 24h (-�-�) BrdU-positive cells indicated no alteration to the position and size of the proliferative zones or rate of migration (absent in theApcDIEC/DIEC SI), n� 4, with representative images of 24 hours BrdU inWT and Kaiso�/y SI (G and H), scale bar represents 50 mm. I, The absence of Kaiso does notalter the normal position of Paneth cells or rescue their mislocalization in the ApcDIEC/DIEC SI; WT (——), Kaiso�/y (---), ApcDIEC/DIEC (���), and ApcDIEC/DIEC Kaiso�/y

(-�-�), n� 4. J, Relative expression analysis indicated no significant alteration toWnt target genes in the absence of Kaiso (WT-black, Kaiso�/y-white, ApcDIEC/DIEC

-dark gray, and ApcDIEC/DIEC Kaiso�/y-light gray), error bars indicate SD, n� 4.

Dlg1 Suppresses Tumors by Maintaining Spindle Polarity

www.aacrjournals.org Mol Cancer Res; 17(3) March 2019 689

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280

binding to either Dlgap1 or Dlg1 was not observed in 10 normalsmall intestinal samples or 5 normal large intestinal samples (Fig.2D). In contrast within Apcmin polyps, Dlgap1 binding wasobserved in 3/10 small intestinal tumors and 3/5 colonic polyps,withDlg1bindingobserved in 1/10 small intestinal tumors and3/5 colonic polyps (Fig. 2D). These data indicate that Kaiso has theability to bind to Dlgap1 and Dlg1, but does so only an Apc-deficient environment. To confirm this finding, ChIP was alsoused to assess Kaiso binding toDlgap andDlg1 in themouse coloncarcinoma cell line CT-26 in which Apc is intact. Using siRNAmethods to generate a greater than 8-fold knock-down Apc(Fig. 2E), demonstrated that binding of Kaiso to the promotersof Dlgap1 and Dlg1 was significantly increased in the in an Apc-deficient setting (Fig. 2F). Taken together these data indicate thatregulation of expression of Dlgap1 and Dlg1 in the normalintestine is most likely indirect; however, in a tumor permissiveenvironment, Kaiso can directly bind to the promoters andregulate expression.

Dlg1 loss disrupts cell division polarity and migration rates inintestine

Dlg1 is known to regulate cell polarity and is a key part of thescribble/dlg/lgl polarity complex (40) and the Apc destruction

complex. To investigate the role of Dlg1 in the intestine, weanalyzed villin-creDlg1flx/flx mice (34), in which exon 4 of theDlg1 gene is flanked by LoxP and driven by endogenous expres-sion of Cre from the Villin promoter to initiate loss of Dlg1 atembryonic E9 within the intestinal epithelium (41). Survivalanalysis was performed by aging cohorts of villin-cre Dlg1þ/þ (WT)and villin-cre Dlg1flx/flx (Dlg1DIEC/DIEC) mice. Overall there was nosignificant difference in survival between WT and Dlg1DIEC/DIEC

and at death the intestinal epithelium of Dlg1DIEC/DIEC miceappeared functionally normal (Fig. 3A). Histologic assessmentof Dlg1DIEC/DIEC intestinal tissue revealed that Dlg1 deficiencydoes not alter crypt length (number of cells), apoptosis (numberof apoptotic bodies), or proliferation (Supplementary Fig. S3A,S3B, and S3C). In summary, it appears that Dlg1 loss is welltolerated by the intestinal epithelium and alone is not sufficientto induce colorectal tumorigenesis. As Dlg1 is known to formpart of the important Scribble/Lgl/Dlg polarity complex, wenext explored the role of these gene in theDlg1-deficient intestine.Expression qRT-PCR analysis of the polarity genes associatedwith Dlg1 indicated a significant reduction in expression of theScribble (Scrib1) gene, which forms part of the polaritycomplex; Fig. 3B). The lack of any gross phenotype as a resultof loss of Dlg1 and significant reduction in Scrib expression

Figure 2.

Upregulation of Dlg1 and Dlgap1 in the Kaiso-deficient intestine.A, qRT-PCR analysis indicating downregulation of Kaiso and upregulation of Dlg1 and Dlgap1 inthe Kaiso�/y intestine (WT, black bars; Kaiso�/y, gray bars), error bars indicate standard deviation, n� 4. IHC comparingWT (B) to Kaiso�/y SI (C) demonstratingan increase in Dlg1 staining in the intestinal epithelial of the Kaiso-deficient intestine, scale bar represents 100 mm. ChiP analysis was used to assess the ability ofKaiso to bind Dlgap1 and Dlg1 in murine SI, large intestine, small intestinal polyps, and large intestinal polyps (D), Apc SiRNA enabled a more than eight-foldknockdown of Apc within the murine cell line CT-26 (E), and ChIP analysis was used to assess the ability of Kaiso to bind to Dlgap1 and Dlg1when CT-26 cellswere treated with either control scrambled siRNA (light gray) or Apc siRNA (dark gray) (F). Error bars represent SD.

Young et al.

Mol Cancer Res; 17(3) March 2019 Molecular Cancer Research690

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280

suggested that other cues from the extracellular matrix (ECM) orredundancy amongst Dlg family members may compensatefor the absence of Dlg1 in order to maintain the correct architec-ture of the intestinal epithelia in vivo (42). However, furtherinvestigation revealed no expressional compensation by otherDlg familymember genes (Supplementary Fig. S4), indicating thatextremely low levels of Dlg1 expression as observed in theDlg1DIEC/DIEC model are sufficient to maintain intestinal cryptmorphology.

To investigate the role of Dlg1 loss on polarity, specificallywithin the intestinal epithelium in the absence of ECMcues, smallintestinal organoid cultures were made from both WT andDlg1DIEC/DIEC tissue. Whole mount immunofluorescence for themarker of the apical tip/ brush border of IECs revealed no obviouscell polarity defects as a result of Dlg1 deficiency within theintestinal epithelium (Fig. 3C and D). As well as playing a keyrole in maintaining normal apical basal polarity Dlg1 has arole in spindle orientation during mitosis. Within the WT crypt86% of cell divisions occurred in a "planar" orientation, with

mitotic spindles running parallel to the basement membranewhile the remaining 14% of mitotic spindles occurred in a"apico-basal" direction (Fig. 3E). Analysis of the Dlg1DIEC/DIEC

intestine demonstrated a significant shift in spindle orientationto 79% planar to 21% apico-basal (Fig. 3F; Chi-squared test P ¼0.0437). As planar cell divisions are in linewith normalmigrationalong the crypt–villus axis any alterations may lead to delayedexit of cells from the crypt. Further BrdU analysis at a 2-hourtime point indicated that Dlg1 deficiency did not alter theposition of cells undergoing mitosis (Fig. 3G). However, at24h the Dlg1-deficient BrdUþ cells were significantly lower downthe crypt–villus axis compared with wild-type controls (Fig. 3G–I;Kolmogorov–Smirnov P < 0.0001). Potentially, the delay inDlg1-deficient cells exiting the crypt may impact on tumorigenesis as itcould increase the time for oncogenic mutations to become fixed,creating a tumor-permissive environment. Levels of cell DNAmutation within the intestine were assessed by qRT-PCR for arange of DNA repair pathway genes, and the protein localizationof the DNA damage marker gH2AX was assessed by IHC and

Figure 3.

Absence of Dlg1 is tolerated by the SI but disrupts mitotic spindle orientation andmigration. A, Kaplan–Meier survival analysis of agedWT (——) andvilCre-Dlg1DIEC/DIEC (. . .) cohorts indicates no difference in overall survival. B, Scrib is significantly (Mann–Whitney; P¼ 0.002) downregulated in the Dlg1DIEC/DIEC

intestine, error bars indicate SD, n� 4. C and D,Whole mount immunofluorescence ofWT (C) and Dlg1DIEC/DIEC (D) organoids demonstrating the maintenance ofcellular polarity (green-phalloidin, red-lysozyme expression, and blue-DAPI). Analysis of mitotic angles inWT (E) and Dlg1DIEC/DIEC (F) indicates a significantdecrease in divisions occurring in a planar orientation (white segments; E, right panel red circle indicating planal mitosis) and an increase in apico-basal divisions(gray segments; F, right panel red circle indicating apico-basal mitosis) as a result of Dlg1 deficiency, n� 6. G, Comparison of the position of BrdUþ cells at 2hours (——) and 24 hours (. . .) in WT (black) and Dlg1DIEC/DIEC (gray) indicating a delay in migration in the Dlg1DIEC/DIEC crypt (Kolmogorov–Smirnov test P <0.0001), with representative images of 24 hours BrdU, n� 4 (H and I).

Dlg1 Suppresses Tumors by Maintaining Spindle Polarity

www.aacrjournals.org Mol Cancer Res; 17(3) March 2019 691

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280

counted, showing no difference (Supplementary Fig. S5).This wasexpected as it is not thought that the rate ofmutation is controlledby Dlg1, simply the rate at which mutated cells migrate.

Dlg1 loss promotes intestinal tumorigenesis in Apcþ/min miceTo explore the role of Dlg1 in intestinal tumorigenesis,

VillinCreþ Dlg1flx/flx mutant mice were crossed with Apcþ/min

mice to generate Apcþ/min and Apcþ/min VillinCreþ Dlg1fl/fl

(ApcminDlg1DIEC/DIEC) cohorts. To confirm our previous findingfor the role ofDlg1 in spindle orientation, we repeated the analysison the normal crypts in these mice. As predicted we observeda significant increase in apico-basal mitotic spindles from 14%in WT (Fig. 3E) to 19% in Apcþ/min (Fig. 4A; Chi-squared testP < 0.001) to 32% in Apcþ/minDlg1DIEC/DIEC mice (Fig. 4B; Chi-squared test P > 0.001). The mice were aged and harvested untilthey reached a humane endpoint indicating an intestinal tumorburden. Survival analysis indicated a significant decrease in sur-vival of ApcminDlg1DIEC/DIEC in comparison to Apcþ/min mice[Fig. 4C, log-rank(Mantel–Cox) P ¼ 0.0309]. At point of harvest

there was no significant difference to the tumor number ormedian tumor burden between the cohorts, indicating mice wereanalyzed at equivalent stages (Fig. 4D and E). However, it wasnoted that even though it did not achieve statistical significancethe median number of tumors in the SI increased from 17 in theApcþ/min mice to 30 in the ApcminDlg1DIEC/DIEC cohort (Fig. 4D),with theApcminDlg1DIEC/DIECmice showing a significant increase ininter-mouse variability. To understand why ApcminDlg1DIEC/DIEC

reached the endpoint sooner than Apcþ/min mice, we examinedthe grade of tumors in the mice, as loss of spindle polarity isassociated with EMT and progression. Microscopic grading ofthe observed lesions identified a significant increase in the tran-sition of single crypt lesions to more advance tumor types,including adenocarcinomas with stromal and smooth muscleinvasion rarely observed in the Apcmin mice (Fig. 4F and G). Inconclusion, the data indicate that that Dlg1 acts as a tumorsuppressor in the murine intestine by preventing loss ofspindle polarity, supporting cell migration rates and preventingprogression of lesions.

Figure 4.

Dlg1 is a tumor suppressor in the murine intestine. Analysis of mitotic spindle angles in crypts of Apcþ/min (A) and Apcþ/minDlg1DIEC/DIEC (B) crypts indicates lossof Dlg1 leads to a significant decrease in the proportion of cells dividing in a normal planar orientation (Chi-squared test P > 0.001; gray). C, Kaplan–Meier analysisdemonstrating a significant reduction in overall survival of Apcþ/minDlg1DIEC/DIEC (black line) mice compared with Apcþ/min [gray line; log-rank (Mantel–Cox)P¼ 0.0309]. The total number of tumors (D) and their burden (E) were not significantly altered in Apcþ/minDlg1DIEC/DIECmice, although they both displayed anincreased inter mouse variability (Levene's test for equality of variances P¼ 0.018). F, Lesions observed within Apcþ/minDlg1DIEC/DIEC (black bars) SI at endpointwere significantly more severe than those observed in Apcþ/min (gray bars, error bars indicate standard deviation; Chi-squared test). SI, stromal invasion; SMI,smooth muscle invasion, n� 28 lesions counted from four mice. G, Lesion from an Apcþ/minDlg1DIEC/DIECmouse indicating progression of tumor toadenocarcinomawith submucosal ( ) and SMI ( ), scale bar represents 500 mm.

Young et al.

Mol Cancer Res; 17(3) March 2019 Molecular Cancer Research692

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280

Tumor suppression due to Kaiso deficiency is Dlg1 dependentPreviously we reported that loss of Kaiso suppressed tumori-

genesis in theApcþ/minmodel. Here we have identified thatDlg1 isupregulated in the Kaiso null intestinal epithelium and acts as atumor suppressor in the Apcþ/minmodel. To assess if the increasedsurvival of Apcmin Kaisoy/�mice is dependent on the presence of afunctional Dlg1 gene and/or related to spindle polarity, wegenerated Apcmin Kaisoy/�Dlg1DIEC/DIEC mice. Analysis of spindlepolarity indicated that the proportion of aberrantly orientatedspindles was significantly reduced from 19% in the Apcþ/minmiceto 12% in the Apcmin Kaiso�/� (Fig. 5A and B; Chi-squared testP < 0.001). A similar level to the WT setting (Fig. 3E) and apotential reason for the subtle increase in cell migrationfrom villus to crypt (Fig. 1E). We next examined the Apcþ/min

Kaiso�/�Dlg1flx/flx mice and demonstrated that the improvementin percentage ofmis-orientated divisions due toKaiso losswas lostwith the additional absence of Dlg1 (Fig. 5A and B; Chi-squaredtest P < 0.001). Significantly, there was no difference in the

percentage of aberrant mitosis in the apico-basal orientationbetween the ApcminDlg1DIEC/DIEC and Apcmin Kaisoy/�Dlg1DIEC/DIEC

intestines (Fig. 5A), and no change in total levels of cell prolif-eration between the cohorts (Fig. 5C). To investigate whetherthe improvement in the proportion of mitotic spindles in theplanar orientated in Apcþ/min Kaisoy/� intestine was importantin the tumor suppression, we generated cohorts of Apcmin

Kaisoy/�Dlg1DIEC/DIEC mice and aged until humane endpoint wasreached. As previously observed, the loss of Kaiso increasedsurvival (Fig. 5D; Apcþ/min Kaisoy/� mice) were only aged to 200days to confirm increased survival and then sacrificed. As reportedearlier (Fig. 4C), there is a significant decrease in survival ofApcmin

Dlg1DIEC/DIEC, which was not rescued by the additional lossof Kaiso (Fig. 5D). Further, the increased aggression oflesions observed in ApcminDlg1DIEC/DIEC again presented in theApcmin Kaisoy/�Dlg1DIEC/DIEC (Fig. 5E). In conclusion, these datasuggest that the tumor resistance observed in the intestineApcminKaisoy� mice requires expression of the Dlg1 gene. We

Figure 5.

The Kaiso-deficient intestine requires an intact Dlg1 to manifest its tumor resistance.A, The proportion of aberrantly orientatedmitotic spindles is significantlyreduced in the Apcþ/minKaisoy/� but increased in the Apcþ/minDlg1DIEC/DIEC, and Apcþ/minKaisoy/�Dlg1DIEC/DIEC, when compared with Apcþ/min intestine. Statisticalanalyis were performed using Mann–Whitney, � indicates P� 0.05, n� 6. B, Representative images of Apcþ/min, Apcþ/minDlg1DIEC/DIEC, Apcþ/minKaisoy/�, andApcþ/minKaisoy/�Dlg1DIEC/DIEC intestine, with indicated aberrant apico-basally orientated cell divisions ( ) and planar cell divisions ("). Scale bars represent 100mm. BrdU analysis (C) indicated there were no differences in total levels of proliferation within the intestine, n� 4. D, Kaplan–Meier survival analysisdemonstrating a decreased survival of Apcþ/minDIEC/DIEC (gray dotted line) and Apcþ/minKaisoy/�Dlg1DIEC/DIEC [gray solid line; log-rank (Mantel–Cox) P¼ 0.0441]compared with Apcþ/min (black solid) and Apcþ/minKaisoy/� (black dashed) mice, vertical mark indicate mouse disease-free survival, n� 9. E, Adenocarcinomawith submucosal invasion ( ) (from an Apcþ/minKaisoy/�Dlg1DIEC/DIECmouse, scale bar represents 500 mm.

Dlg1 Suppresses Tumors by Maintaining Spindle Polarity

www.aacrjournals.org Mol Cancer Res; 17(3) March 2019 693

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280

propose that the appropriate expression of Dlg1 prevents mitosisoccurring at an aberrant apico-basal orientation, maintains therates of cell migration onto the villus, and therefore may decreasethe window of opportunity for tumorigenesis (Fig. 6).

DiscussionIn summary, here we have demonstrated that in the absence of

the epigenetic regulator Kaiso there is an increase in expression ofthe polarity proteinsDlg1 andDlgap1. The absence ofDlg1 resultsin an increase in the number of baso-lateral cell divisions withinthe intestinal crypt and a decrease in cell migration rates. On theApcþ/min background, the absence ofDlg1 decrease survival due toan increase in the number and aggressiveness of tumors. Further,the loss of Dlg1 overcomes the suppression of intestinal tumor-igenesis observed in the Apcþ/minKaiso�/y mouse model. Suggest-ing that themechanisms that maintain planal mitotic orientationand migration rates have important antitumorigenic roles.Although there are many potential mechanisms for the controlof migration rates along the crypt–villus axis, including cell:cellsignaling and negative draw of cells through the sloughing of cellsat the villus tip, there ismuch evidence to support the concept thatmitosis supports normal cellular migration along the crypt–villusaxis through mitotic pressure (43, 44). This leads to the assump-tion that the angle of mitosis within the crypt could play a part innormal migration patterns. The rapid cell migration along thecrypt–villus axis is one of the driving factors which controls thequick turn over of cells within the intestinal epithelium and is

essential tominimize the risk of tumorigenesis. It achieves this byflushing undetected mutant cells from the epithelium before theycan disrupt intestinal homeostasis. Through delayed migration itcan be hypothesized that Apcþ/min cells which have lost the wildtype copy of Apc (through random mutations, loss of heterozy-gosity or epigenetic silencing) may remain within the intestinalepithelium for a longer period of time, increasing the opportunityfor lesion formation. An alternative hypothesis is that theincreased crypt residence time of a cell increases the likelihoodthat it will accumulate additional oncogenicmutations, for exam-ple K-ras, Smad4, or p53. This may account for the increasedaggressiveness of the tumors observed in the Apcþ/min Dlg1�/�

model. The role of cell migration speed within the intestinalepithelium in creating a tumor permissive environment has beenhypothesized elsewhere (45, 46), and in fact many previouslyestablished chemo-preventative agents play a role in increasingcellular migration (47), which may contribute to their efficacy indelaying tumor onset.

Normal planar cell divisions within the crypt can supportmitotic pressure for migration to occur in a straight line, thequickest route from crypt base to villus tip, whereas apico-basaldivisions could logically result in "side-to-side" cell movementand "upward," which could be responsible for the increasedmigration time observed in Dlg1DIEC/DIEC mice (Fig. 6). This hasbeen previously shown in astrocyte cells inwhich deficiency of theDlg1:Dlgap1 complex, also identified in this study, leads toinefficient migration (40). Further, although severely abnormalangles of mitosis may lead to delamination of cells they would

Figure 6.

Model of tumor suppression by Dlg1.A, In the normal intestine Dlg1 plays an important role in maintaining the planal orientation of mitosis to ensure that divisionoccurs in the same direction as migration. B, In the Dlg1-deficient intestine alteration to the angles of mitosis conflict with the direction of cell migration leading tolonger transit times from the crypt-base to the villus. Potentially increasing the window of opportunity for fixation of a mutant cell, which may then migrate inmultiple directions and increasing the chance of invasion into the submucosa.

Young et al.

Mol Cancer Res; 17(3) March 2019 Molecular Cancer Research694

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280

normally undergo apoptosis. Any further suppression of apopto-sis leads to formation of disorganized masses of aberrant cellswith characteristics of tumorigenesis and EMT, potentiallyaccounting for the increase in more advanced lesion observedhere. However, it must be noted that during mitosis within theintestinal crypt, cells step out of the epithelial sheet but maintaincontact with surrounding cells and divide more luminallybefore the daughter cells return to the intestinal epitheliumproper. It is possible that the orientation of cell division makesless difference to the final position of the daughter cells within theintestinal epithelium than the process by which cells return to theepithelial sheet.

Previous reports have linked epigenetic modification to alteredexpression of genes associated with cell polarity (48), therebyenabling transcriptional repressors such as Kaiso to regulate cellpolarity. However, further work using chromatin immunoprecip-itation techniques is required to confirm that these genes are directtargets of Kaiso. The work we present here supports the idea thattranscriptional repression of polarity associated genes plays a rolein tumorigenesis and tumor invasion (49), and that inhibition ofsuch epigenetic regulators may be of therapeutic value. However,it should be noted that the phenotype associated with the loss ofKaisomay also reflect yet unidentified changes in genes associatedwith its transcriptional activation functions. At the very least,identification and exploration of the targets of such epigenetictranscriptional regulators whichmay influence cell polarity couldprove to be of value to the field of cancer research. As Kaisoinhibition has been suggested as a potential therapeutic strategyfor the treatment of colorectal cancer, and inhibitors for thistranscriptional repressor are currently being investigated (50).The work presented here indicates that the tumor-suppressiveeffect ofKaiso loss is limited to systems inwhich functionalDlg1 ispresent at the initiation stage of a tumor. It remains to bedetermined whether targeting Kaiso in a higher-grade tumor will

have any beneficial effect and whether it would be dependent onDlg1 remaining unmutated within the patient's tumor. As such,the efficacy of such a treatment would be limited to lower gradetumors where Dlg1 remains functional, and tumor resistance totreatment could develop through additional mutations in Dlg1.

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

Authors' ContributionsConception and design: W. Swat, L. ParryDevelopment of methodology: M.A. Young, W. Swat, L. ParryAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): M.A. Young, S. May, A. Damo, L. ParryAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): M.A. Young, Y.S. Yoon, M.-W. Hur, L. ParryWriting, review, and/or revision of the manuscript: M.A. Young, L. ParryAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): M.A. Young, S. May, Y.S. Yoon, L. ParryStudy supervision: M.A. Young, M.-W. Hur, L. Parry

AcknowledgmentsThe authors are grateful to Elaine Taylor, Matthew Zverev, and Derek

Scarborough for technical support. L. Parrywould like to thankDavid&DeborahPhilpott and LiamHurley for assistance with space. This work was supported byCancer Research UK (to M. Young; L. Parry; program grant C1295/A15937);Moorhouse Foundation award (to S. May); Cardiff University, School of Bio-science Seedcorn Award (to M. Young); Cardiff University Fellowship (toL. Parry).

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

ReceivedMarch 23, 2018; revised September 28, 2018; accepted December 4,2018; published first December 14, 2018.

References1. Atrian F, Leli�evre SA. Mining the epigenetic landscape of tissue polarity

in search of new targets for cancer therapy. Epigenomics 2015;7:1313–25.

2. Baylin SB. DNA methylation and gene silencing in cancer. Nat Clin PractOncol 2005;2:S4–S11.

3. Esteller M. Epigenetic gene silencing in cancer: the DNA hypermethylome.Hum Mol Genet 2007;16:R50-9.

4. Gloss B, Moran-Jones K, Lin V, Gonzalez M, Scurry J, Hacker NF, et al.ZNF300P1 encodes a lincRNA that regulates cell polarity and is epige-netically silenced in type II epithelial ovarian cancer. Mol Cancer2014;13:3.

5. Guo YL, Shan BE, Guo W, Dong ZM, Zhou Z, Shen SP, et al. Aberrantmethylation of DACT1 and DACT2 are associated with tumor progressionand poor prognosis in esophageal squamous cell carcinoma. J Biomed Sci2017;24:6.

6. Papageorgis P, Lambert AW, Ozturk S, Gao F, Pan H, Manne U, et al. Smadsignaling is required to maintain epigenetic silencing during breast cancerprogression. Cancer Res 2010;70:968–78.

7. Parry L, Clarke AR. The roles of themethyl-CpG binding proteins in cancer.Genes Cancer 2011;2:618–30.

8. Kondo Y, Shen L, Issa JP. Critical role of histone methylation in tumorsuppressor gene silencing in colorectal cancer. Mol Cell Biol 2003;23:206–15.

9. Stogios PJ, Downs GS, Jauhal JJ, Nandra SK, Priv�e GG. Sequence andstructural analysis of BTB domain proteins. Genome Biol 2005;6:R82.

10. Prokhortchouk A, Hendrich B, JørgensenH, Ruzov A,WilmM,GeorgievG,et al. The p120 catenin partner Kaiso is a DNA methylation-dependenttranscriptional repressor. Genes Dev 2001;15:1613–8.

11. Ruzov A, Dunican DS, Prokhortchouk A, Pennings S, Stancheva I,Prokhortchouk E, et al. Kaiso is a genome-wide repressor of transcriptionthat is essential for amphibian development. Development 2004;131:6185–94.

12. Prokhortchouk A, Sansom O, Selfridge J, Caballero IM, Salozhin S,Aithozhina D, et al. Kaiso-deficient mice show resistance to intestinalcancer. Mol Cell Biol 2006;26:199–208.

13. Lopes EC, Valls E, Figueroa ME, Mazur A, Meng FG, Chiosis G, et al. Kaisocontributes to DNAmethylation-dependent silencing of tumor suppressorgenes in colon cancer cell lines. Cancer Res 2008;68:7258–63.

14. Koh DI, Han D, Ryu H, Choi WI, Jeon BN, Kim MK, et al. KAISO, a criticalregulator of p53-mediated transcription of CDKN1A and apoptotic genes.Proc Natl Acad Sci U S A 2014;111:15078–83.

15. Jones J, Wang H, Zhou J, Hardy S, Turner T, Austin D, et al. NuclearKaiso indicates aggressive prostate cancers and promotes migrationand invasiveness of prostate cancer cells. Am J Pathol 2012;181:1836–46.

16. Rodova M, Kelly KF, VanSaun M, Daniel JM, Werle MJ. Regulation ofthe rapsyn promoter by kaiso and delta-catenin. Mol Cell Biol 2004;24:7188–96.

17. Pierre CC, Longo J, Bassey-ArchibongBI,Hallett RM,Milosavljevic S, BeattyL, et al. Methylation-dependent regulation of hypoxia inducible factor-1alpha gene expression by the transcription factor Kaiso. Biochim BiophysActa 2015;1849:1432–41.

18. Pierre CC, Longo J, Mavor M, Milosavljevic SB, Chaudhary R, Gilbreath E,et al. Kaiso overexpression promotes intestinal inflammation and potenti-ates intestinal tumorigenesis in Apc(Min/þ) mice. Biochim Biophys Acta2015;1852:1846–55.

Dlg1 Suppresses Tumors by Maintaining Spindle Polarity

www.aacrjournals.org Mol Cancer Res; 17(3) March 2019 695

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280

19. Chaudhary R, Pierre CC, Nanan K,Wojtal D, Morone S, Pinelli C, et al. ThePOZ-ZF transcription factor Kaiso (ZBTB33) induces inflammation andprogenitor cell differentiation in the murine intestine. PLoS One 2013;8:e74160.

20. Su L, Kinzler K, Vogelstein B, Preisinger A, Moser A, Luongo C, et al.Multiple intestinal neoplasia caused by amutation in themurine homologof the APC gene. Science 1992;256:668–70.

21. Powell S, Zilz N, Beazer-Barclay Y, Bryan T, Hamilton S, Thibodeau S, et al.APC mutations occur early during colorectal tumorigenesis. Nature1992;359:235–7.

22. Cancer Genome Atlas Network. Comprehensive molecular characteriza-tion of human colon and rectal cancer. Nature 2012;487:330–7.

23. Cerami E,Gao J,DogrusozU,Gross BE, Sumer SO, Aksoy BA, et al. The cBiocancer genomics portal: an open platform for exploring multidimensionalcancer genomics data. Cancer Discov 2012;2:401–4.

24. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, et al.Integrative analysis of complex cancer genomics and clinical profiles usingthe cBioPortal. Sci Signal 2013;6:pl1.

25. Quyn AJ, Appleton PL, Carey FA, Steele RJ, Barker N, Clevers H, et al.Spindle orientation bias in gut epithelial stem cell compartments is lost inprecancerous tissue. Cell Stem Cell 2010;6:175–81.

26. SuWH,Mruk DD,Wong EW, Lui WY, Cheng CY. Polarity protein complexScribble/Lgl/Dlg and epithelial cell barriers. Adv Exp Med Biol 2012;763:149–70.

27. Humbert PO, Grzeschik NA, Brumby AM, Galea R, Elsum I, RichardsonHE. Control of tumourigenesis by the Scribble/Dlg/Lgl polarity module.Oncogene 2008;27:6888–907.

28. Wodarz A, N€athke I. Cell polarity in development and cancer. Nat Cell Biol2007;9:1016–24.

29. Dow LE, Humbert PO. Polarity regulators and the control of epithelialarchitecture, cell migration, and tumorigenesis. Int Rev Cytol 2007;262:253–302.

30. Ireland H, Kemp R, Houghton C, Howard L, Clarke AR, Sansom OJ, et al.Inducible Cre-mediated control of gene expression in the murine gastro-intestinal tract: effect of loss of beta-catenin. Gastroenterology 2004;126:1236–46.

31. Shibata H, Toyama K, Shioya H, Ito M, Hirota M, Hasegawa S, et al. Rapidcolorectal adenoma formation initiated by conditional targeting of the Apcgene. Science 1997;278:120.

32. Moser AR, Pitot HC, Dove WF. A dominant mutation that predis-poses to multiple intestinal neoplasia in the mouse. Science 1990;247:322–4.

33. El Marjou F, Janssen KP, Hung Junn Chang B, Li M, Hindie V, Chan L, et al.Tissue specific and inducible Cre mediated recombination in the gutepithelium. Genesis 2004;39:186–93.

34. Stephenson LM, Sammut B, GrahamDB, Chan-Wang J, Brim KL, Huett AS,et al.DLGH1 is anegative regulator of T-lymphocyte proliferation.MolCellBiol 2007;27:7574–81.

35. Wettenhall JM, Simpson KM, Satterley K, Smyth GK. affylmGUI: a graph-ical user interface for linear modeling of single channel microarray data.Bioinformatics 2006;22:897–9.

36. Merritt AJ, Allen TD, Potten CS, Hickman JA. Apoptosis in small intestinalepithelial from p53-null mice: evidence for a delayed, p53-independentG2–M-associated cell death after gamma-irradiation. Oncogene 1997;14:2759–66.

37. Sato T, Vries RG, Snippert HJ, Van De Wetering M, Barker N, Stange DE,et al. Single Lgr5 stem cells build crypt villus structures in vitro without amesenchymal niche. Nature 2009;459:262–5.

38. Sansom OJ, Reed KR, Hayes AJ, Ireland H, Brinkmann H, Newton IP, et al.Loss of Apc in vivo immediately perturbs Wnt signaling, differentiation,and migration. Genes Develop 2004;18:1385–90.

39. Daniel JM, Spring CM, Crawford HC, Reynolds AB, Baig A. The p120(ctn)-binding partner Kaiso is a bi-modal DNA-binding protein that recognizesboth a sequence-specific consensus and methylated CpG dinucleotides.Nucleic Acids Res 2002;30:2911–9.

40. Manneville JB, Jehanno M, Etienne-Manneville S. Dlg1 binds GKAP tocontrol dynein association with microtubules, centrosome positioning,and cell polarity. J Cell Biol 2010;191:585–98.

41. El Marjou F, Janssen KP, Chang BHJ, Li M, Hindie V, Chan L, et al.Tissuespecific and inducible Cremediated recombination in the gutepithelium. Genesis 2004;39:186–93.

42. Lee JL, Streuli CH. Integrins and epithelial cell polarity. J Cell Sci2014;127:3217–25.

43. Meineke FA, Potten CS, Loeffler M. Cell migration and organizationin the intestinal crypt using a lattice-free model. Cell Prolif 2001;34:253–66.

44. Parker A, Maclaren OJ, Fletcher AG, Muraro D, Kreuzaler PA, Byrne HM,et al. Cell proliferationwithin small intestinal crypts is the principal drivingforce for cell migration on villi. FASEB J 2016;31:636–49.

45. Sansom OJ, Mansergh FC, Evans MJ, Wilkins JA, Clarke AR. Deficiencyof SPARC suppresses intestinal tumorigenesis in APCMin/þ mice.Gut 2007;56:1410–4.

46. Reed KR, Korobko IV, Ninkina N, Korobko EV, Hopkins BR, Platt JL, et al.Hunk/Mak-v is a negative regulator of intestinal cell proliferation.BMC Cancer 2015;15:110.

47. Fenton JI, Wolff MS, Orth MW, Hord NG. Membrane-type matrix metal-loproteinases mediate curcumin-induced cell migration in non-tumori-genic colon epithelial cells differing in Apc genotype. Carcinogenesis2002;23:1065–70.

48. Parfitt DE, Zernicka-GoetzM. Epigenetic modification affecting expressionof cell polarity and cell fate genes to regulate lineage specification in theearly mouse embryo. Mol Biol Cell 2010;21:2649–60.

49. Moreno-Bueno G, Portillo F, Cano A. Transcriptional regulation of cellpolarity in EMT and cancer. Oncogene 2008;27:6958–69.

50. Chikan NA, Vipperla B. KAISO inhibition: an atomic insight. J BiomolStruct Dyn 2015;33:1794–804.

Mol Cancer Res; 17(3) March 2019 Molecular Cancer Research696

Young et al.

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280

2019;17:686-696. Published OnlineFirst December 14, 2018.Mol Cancer Res   Madeleine A. Young, Stephanie May, Angelos Damo, et al.   Polarity and Promotes Intestinal Tumorigenesis

, Alters Mitotic SpindleKaiso, via Dlg1Epigenetic Regulation of

  Updated version

  10.1158/1541-7786.MCR-18-0280doi:

Access the most recent version of this article at:

  Material

Supplementary

  http://mcr.aacrjournals.org/content/suppl/2018/12/13/1541-7786.MCR-18-0280.DC1

Access the most recent supplemental material at:

   

   

  Cited articles

  http://mcr.aacrjournals.org/content/17/3/686.full#ref-list-1

This article cites 49 articles, 19 of which you can access for free at:

  Citing articles

  http://mcr.aacrjournals.org/content/17/3/686.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

  .pubs@aacr.org

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/17/3/686To request permission to re-use all or part of this article, use this link

on August 31, 2021. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst December 14, 2018; DOI: 10.1158/1541-7786.MCR-18-0280