The RUNX Transcriptional Coregulator, Cbfβ, Suppresses Migration of ER+ Breast Cancer Cells by Repressing Erα-Mediated Express

Published OnlineFirst January 17, 2019; DOI: 10.1158/1541-7786.MCR-18-1039

Oncogenes and Tumor Suppressors Molecular Cancer Research The RUNX Transcriptional Coregulator, CBFb, Suppresses Migration of ERþ Breast Cancer Cells by Repressing ERa-Mediated Expression of the Migratory Factor TFF1 Henry J Pegg, Hannah Harrison, Connor Rogerson, and Paul Shore

Abstract

Core binding factor b (CBFb), the essential coregulator of anism revealed that the increase in migration is driven by the RUNX transcription factors, is one of the most frequently coregulation of Trefoil factor 1 (TFF1) by CBFb and ERa. þ mutated genes in estrogen receptor–positive (ER ) breast RUNX1–CBFb acts to repress ERa-activated expression of cancer. Many of these mutations are nonsense mutations and TFF1. TFF1 is a motogen that stimulates migration and we are predicted to result in loss of function, suggesting a tumor show that knockdown of TFF1 in CBFb / cells inhibits the suppressor role for CBFb. However, the impact of missense migratory phenotype. Our ﬁndings reveal a new mechanism þ mutations and the loss of CBFb in ER breast cancer cells have by which RUNX1–CBFb and ERa combine to regulate gene not been determined. Here we demonstrate that missense expression and a new role for RUNX1–CBFb in the prevention mutations in CBFb accumulate near the Runt domain–bind- of cell migration by suppressing the expression of the motogen ing region. These mutations inhibit the ability of CBFb to form TFF1. CBFb–Runx–DNA complexes. We further show that deletion þ of CBFb, using CRISPR-Cas9, in ER MCF7 cells results in an Implications: Mutations in CBFb contribute to the develop- increase in cell migration. This increase in migration is depen- ment of breast cancer by inducing a metastatic phenotype that dent on the presence of ERa. Analysis of the potential mech- is dependent on ER.

reported. However, the effects of these mutations have not been Introduction established. All CBFb mutations reported appear to been in – þ b Core binding factor b (CBFb) is the obligate binding partner estrogen receptor positive (ER ) breast cancer, with CBF muta- – þ of all three mammalian Runx transcription factors, Runx1, 2, tions estimated to occur in approximately 2.6% 4.7% of all ER – a þ and 3 (1). Knockout of CBFb in mouse phenocopies Runx- breast cancers (7 9). ER is a key driver of cell proliferation in ER loss and is embryonic lethal (2, 3). CBFb binding to Runx breast cancer, and it has recently been shown that RUNX1 pre- a proteins protects Runx from proteasomal degradation and vents ER -mediated repression of the Wnt pathway repressor, causes a conformational change in the runt domain, the AXIN1, suggesting one explanation for the role of RUNX1 and b DNA-binding domain, which increases its afﬁnity for DNA CBF in this context (10). a binding (4, 5). Trefoil factor 1 (TFF1) is a well-established ER target gene and a We have previously shown that CBFb has a proinvasive role in is often used as a robust marker of activated ER . TFF1 is essential triple-negative breast cancer cells (6). More recently, mutations in for normal function of the gastric mucosa; it is a small, secreted CBFb (and RUNX1) were reported in breast tumors and are protein that has been described as a "motogen" needed for repair among the most frequently reported for breast cancer tumors of the gastric epithelia following damage (11). In breast cancer, it (7–9). Many of these mutations are nonsense mutations predicted has been demonstrated to play a role in cell migration (12) and to result in loss of function, suggesting a tumor suppressor role for chemotaxis (13). b þ CBFb. In addition, numerous missense mutations have also been To investigate the role of CBF in ER breast cancer, we initially evaluated the mechanistic function of CBFb missense mutations that occur in patients with breast cancer. All of the missense mutations we studied suppress the formation of the CBFb– Runx–DNA complex. We then used CRISPR-Cas9 to delete CBFb þ Faculty of Biology, Medicine and Health, The University of Manchester, in ER MCF7 cells and we show that loss of CBFb leads to an Manchester, United Kingdom. increase in cell migration. Subsequent analysis revealed that the Note: Supplementary data for this article are available at Molecular Cancer increased migration is mediated, at least in part, by the derepres- Research Online (http://mcr.aacrjournals.org/). sion of TFF1 and we demonstrate that the increased TFF1 expres- Corresponding Author: Paul Shore, University of Manchester, Michael Smith sion is mediated by loss of ERa repression by CBFb. Importantly, Building, Oxford Road, Manchester M13 9PT, United Kingdom. Phone: 44-0161- this mechanism is distinct from the coactivator role of Runx1 2755978; Fax: 44-0161-2745978; E-mail: [email protected] previously reported for AXIN1 regulation and reﬂects a novel þ doi: 10.1158/1541-7786.MCR-18-1039 explanation for the occurrence of CBFb mutations in ER breast 2019 American Association for Cancer Research. cancer.

www.aacrjournals.org OF1

Pegg et al.

Materials and Methods Fluorescent electrophoretic mobility shift assays CBFb (WT and mutant) and Runt domain were expressed Cells and culture conditions and purified from Bl21-gold DE3 Cells (Agilent Technologies) Human HeLa, MCF7, MDA-MB-231, and T47D cell lines were using a GST tag that was cleaved off during purification. obtained from LGC Standards and were maintained in DMEM Electrophoretic mobility shift assays (EMSA) performed (Sigma-Aldrich) supplemented with 10% FBS (Biowest), 1% L- using the Odyssey Infrared EMSA Kit (LI-COR Biosciences) glutamine (Sigma-Aldrich), and 1% penicillin–streptomycin using a double-stranded fluorescently labeled DNA probe 50- (Sigma). Charcoal-stripped FBS (cs-FBS) was generated by incu- ATTO700-TCTGAAAACCACAGCCGCCA-30.CBFb (5 mg; WT or bation of FBS with 3% (m/v) dextran-coated charcoal (Sigma- mutant) was used per well alongside 0.6-mg Runt domain per Aldrich) at 4 C for 72 hours before filter (0.22 mmol/L) sterili- reaction. An SDS-PAGE gel was also run and stained with zation. For estrogen stimulation experiments, cells were grown in InstandBlue (Expedeon) to confirm equal loading of each phenol red–free DMEM (Life Technologies), supplemented with protein. 5% cs-FBS, 1% L-glutamine (Sigma-Aldrich), and 1% penicillin– streptomycin (Sigma-Aldrich) for 4 days prior to 10 nmol/L – b-estradiol stimulation (4–24 hours depending on the CRISPR-Cas9 mediated gene knockout b experiment). CBF gene knockout was performed using a double nickase CRISPR-Cas9 strategy as described previously (14). Guide RNA sequences were designed using E-CRISP (15) to minimize off- Preparation of conditioned media 6 2 target effects. Cells were fluorescence-activated cell sorted for GFP- A total of 1 10 cells were plated in a 10-cm dish and grown Cas9 expression 48 hours after transfection and grown up from in complete media supplemented with 10 nmol/L estrogen (17b- single colonies prior to genomic DNA PCR and Western blotting Estradiol, Sigma-Aldrich) for 3 days. Media were removed from screening. cells, spun at 4,000 g for 15 minutes to remove dead cells, and stored at 20C. Genomic DNA PCR CBFb plasmid mutagenesis Genomic DNA was isolated using a Purelink gDNA Extraction To introduce specific mutations into the CBFb sequence, a Kit (Invitrogen) as per the manufacturer's protocol. The genomic fi QuikChange II XL Site-Directed Mutagenesis Kit (Agilent Tech- region around the CRISPR-Cas9 target site was ampli ed using fi nologies) was used with primers designed using the complemen- nested PCR. Initially, a larger fragment was ampli ed using tary QuikChange Primer Design Program (Agilent Technologies, external-F: GAATGGTGCGTCTTGTTTGC; external-R: AAGACG- fi 2017). All plasmids were DNA sequenced by Manchester Uni- GCCCGAAATCAGG. DNA was puri ed and PCR repeated using fi versity DNA Sequencing Facility to confirm the presence of the primers inside the previously ampli ed piece; internal-F: mutation. GGCGGCAGGCAACGGCTGAG and internal-R: CCTCACACT- CGCGGCTCAGC. Cell lysis and Western blotting Briefly, cells were twice washed in ice-cold PBS and lysed in Quantitative reverse transcription PCR RIPA buffer (Sigma-Aldrich) supplemented with protease inhib- RNA was extracted using RNeasy Mini Kit (Qiagen) as per the fi itor cocktail (Roche) for 15 minutes on ice. Lysates were cold spun manufacturer's protocol. RNA was quanti ed using a Nanodrop fi for 15 minutes at 16,000 g and cleared supernatant was 2000 (Thermo Fisher Scienti c) and 50 ng used per reaction using transferred to new tubes. Protein concentration was determined Quantitect RT-PCR SYBR Green Kit (Qiagen). Primer sequences using the DC Protein Assay (Bio-Rad) and equal amounts of used: TFF1-F GTGTCACGCCCTCCCAGT; TFF1-R GGACCCCAC- protein run on 12% tris-glycine gels. Protein was transferred to GAACGGTG; RPLO-F GGCGACCTGGAAGTCCAA; and RPLO-R 0.45 mmol/L nitrocellulose membrane (GE Healthcare) and CCATCAGCACCACAGCCTT. membrane was blocked for 1 hour with 5% nonfat dry milk prior to incubation with primary antibodies overnight: AXIN1- Migration assay C76H11 (Cell Signaling Technology), CBFb- ab33516 (Abcam), Briefly, 5 104 cells in serum-free media were placed in a ERa- sc-8002 (Santa Cruz Biotechnology), FLAG- F1804 (Sigma- Boyden chamber insert (Corning) and allowed to migrate for 24 Aldrich), H3- 9715 (Cell Signaling Technology), RUNX1- sc-8563 hours at 37 C. Nonmigrating cells were removed from the top of (Santa Cruz Biotechnology), TFF1- ab92377 (Abcam), and Vin- the filter by scrubbing with a cotton swab. Cells that migrated were culin- ab129002 (Abcam). fixed and stained with 0.1% crystal violet. Cells were then imaged and cells/view counted. Cotransfection coimmunopreciptiation assay HeLa cells were plated and transfected for 48 hours with CBFb- Proliferation assay myc-Flag (OriGene), wild-type (WT) or mutant, alongside Proliferation rate was assessed using the thiazolyl blue tetra- RUNX1-Flag or RUNX2-Flag (generated in house) using Lipofec- zolium bromide (MTT) assay. Briefly, 1,000 cells were plated (in tamine 2000 (Thermo Fisher Scientific). Cells were lysed as above octuplicate) in a 96-well plate harvested after 24 hours growth. At and 1 mg of cell lysate was incubated per immunoprecipitate with which time, the cells were washed with PBS and incubated with 3-mg myc tag antibody- ab9132 (Abcam) and 30-mL PBS-Tween 0.4 mg/mL MTT solution in complete media for 3 hours at 37C/ washed protein G Dynabeads (Life Technologies) for 3 hours. 5% CO2. The cells were then washed twice with PBS and the Beads–protein complexes were washed five times with RIPA formazan crystals were dissolved in 200 mL absolute DMSO per buffer, prior to elution in 2 SDS loading buffer (95C for 10 well. The absorbance was read at 560 nm using a microplate minutes), followed by Western blotting. reader.

OF2 Mol Cancer Res; 2019 Molecular Cancer Research

Loss of CBFb Induces Migration through TFF1

siRNA transfections tion to the interaction with the Runt domain suggesting that the Dharmacon smartpool siRNA (CBFb, RUNX1, TFF1, and non- majority of mutations may directly affect the interaction of CBFb targeting, Horizon Discovery) was transfected using RNAimax with Runx proteins. (Thermo Fisher Scientific) as per the manufacturer's instructions for 72 hours prior to Western blot analysis and migration assay. Missense mutations in CBFb inhibit the formation of a CBFb– When conditioned media was to be isolated following siRNA Runx–DNA complex transfection, the siRNA was transfected as above for 48 hours prior To determine whether missense mutations in CBFb effect its to media change and further 48 hours incubation before isolation ability to bind the Runt domain, six missense mutations found of conditioned media as mentioned previously. near the Runx–CBFb binding interface were chosen for analysis (Fig. 2A). A cotransfection coimmunoprecipitation assay was Fulvestrant treatment used to assess the impact of the mutations on the ability of CBFb Cells were treated with 100 nmol/L fulvestrant (Sigma) for the to bind to RUNX1 and RUNX2. Surprisingly, only one of the time specified in the figure legend. mutations, N104S, robustly abrogated the binding of CBFb to RUNX1 and RUNX2 in comparison with WT CBFb (Fig. 2B and Statistical analysis C). To determine the effect of the mutations on DNA complex All bar graph data are presented as the mean SEM and all formation, we performed EMSAs using purified recombinant statistical analysis was done using Graphpad Prism 7.0 using proteins. CBFb and Runt domain proteins were expressed and unpaired or paired t tests depending on the experimental set up, , purified from bacterial extracts (Supplementary Fig. S1). EMSAs P < 0.05; , P < 0.01; and , P < 0.001. were performed using a fluorescently labeled RUNX-binding site and the Runt domain with either WT or mutant CBFb proteins. All Results six mutations significantly abrogated the formation of the DNA– RUNX–CBFb ternary complex (Fig. 2D). This demonstrates that Missense CBFb mutations occur more frequently near the Runt the defect in binding is only observable when CBFb–Runx is in domain–binding site complex with DNA. These data suggest that missense mutations, CBFb mutations that occur in patients with breast cancer are like the truncating mutations, are loss of function, supporting the found along the whole-protein sequence (Fig. 1A). Approximate- notion that CBFb is acting as a tumor suppressor gene. ly 60% of the mutations are truncating and are therefore likely to result in a loss of functional CBFb in these patients (Fig. 1B). However, the functional effect of the missense mutations has not Loss of CBFb in MCF7 cells leads to an increase in ERa-driven been reported. To analyze the effect of missense mutations, we cell migration þ first mapped them onto the crystal structure of the CBFb–Runt– All of the CBFb mutations that have been discovered are in ER DNA complex (16). This analysis revealed that there are "hot- patients. Given that CBFb appears to act as a tumor suppressor, we spots" where mutations occur with greater frequency (Fig. 1C). A decided to study the function of CBFb by deleting the gene in an þ x2 test performed on a frequency distribution of the mutations ER context. To this end, a CRISPR-Cas9 double nickase strategy along the primary structure determined that the mutations are not was used to target the start of CBFb exon1 in MCF7 cells (Fig. 3A; randomly distributed along the protein [x2 (8, N ¼ 52) ¼ 40.892, Supplementary Fig. S2A; ref. 14). Two clones were generated that P < 0.001; Fig. 1D]. One region in particular, between amino acids had changes in the genomic locus and complete loss of CBFb 100 and 120 harbors a large number of the most commonly protein (Supplementary Figs. S2B and S3B). To determine whether mutated amino acids. This region makes a significant contribu- the loss of CBFb prevented the function of the Runx transcriptional

Figure 1. Missense CBFb mutations, found in patients with breast cancer, accumulate near the Runt-binding interface. A, A lollipop diagram was generated detailing the position and number of missense and truncating (nonsense, frameshift, and splice) mutations in CBFb found in the METABRIC study and COSMIC studies (A) and the results as a pie chart (B). C, The amino acid positions of each of the missense mutations was displayed on the crystal structure of CBFb(cyan)– Runt(green)–DNA(blue) structure (16) as a heatmap. D, A frequency distribution was generated on which a x2 test was performed.

www.aacrjournals.org Mol Cancer Res; 2019 OF3

Pegg et al.

Figure 2. Missense mutations in CBFB inhibit the formation of the CBFB–Runt–DNA complex. A, Six missense mutations chosen for further study were annotated on the CBFb(cyan)–Runt(green)–DNA(blue) crystal structure (31). Cotransfection Coimmunoprecipitation (IP), followed by western blot analysis, was performed to assess the effect of CBFb mutation on the binding of RUNX1 (B) and RUNX2 (C). D, EMSAs were used to assess the formation of DNA–Runt–CBFb complexes with mutant CBFb and complex formation was quantiﬁed.

complexes, the expression level of MMP13, a well-characterized Runx–CBFb acts as an activator or a repressor of ERa target target gene of Runx–CBFb was measured (Fig. 3C). MMP13 genes in a gene-speciﬁc manner expression level was decreased by approximately 80% in Interplay between RUNX1 and ERa has been studied previ- CBFb / MCF7 cells, demonstrating that loss of CBFb resulted in ously (10). Analysis of RUNX1 in MCF7 cells has established a decrease in Runx protein–mediated gene transcription. RUNX1 as an inhibitor of ERa-driven cell proliferation. RUNX1 Previously, knockdown of CBFb in a triple-negative model of was shown to relieve the repression of AXIN1, a Wnt pathway breast cancer led to a dramatic decrease in cell migration (6). With repressor, by ERa (10). this in mind, the migratory capacity of CBFb / MCF7 cells was To assess whether CBFb is required for this function of determined using a transwell migration assay. Surprisingly, CBFb / RUNX1inMCF7cells,AXIN1 expression level was measured cells were found to be signiﬁcantly more migratory (Fig. 3D). This is in CBFb / cells. AXIN1 was found to be decreased at both in contrast to what is observed in CBFb-depleted ER models. As mRNA (Fig. 4A) and protein level (Fig. 4B) showing that CBFb MCF7 cells are dependent on the function of ERa,theroleofERa in is essential for this RUNX1-mediated gene activation. The this observed increase in cell migration was determined. This was regulation of AXIN1 does not explain the increase in migration determined by treating CBFb / MCF7 cells with fulvestrant, a drug observed in CBFb / cells, however, and we therefore sought to that targets ERa for degradation, or 72 hours prior to a migration determine the molecular mechanism by which loss of CBFb assay. The increased migratory capacity of the CBFb / cells was increases migration. To this end, public datasets were reana- attenuated following fulvestrant treatment, suggesting that this lyzed. Two previously reported RUNX1 ChIP-seq datasets, phenotype is dependent on ERa activity (Fig. 3E). which had used different RUNX1 antibodies, were merged to

OF4 Mol Cancer Res; 2019 Molecular Cancer Research

Loss of CBFb Induces Migration through TFF1

Figure 3. CBFb loss in MCF7 cells leads to an increase in ERa-dependent cell migration. A, To knockout CBFb,a double nickase CRISPR-Cas9 was targeted to the start of CBFb exon 1 generating two clones with complete CBFb loss by Western blot analysis (B). C, To determine the effect of CBFb on RUNX target gene regulation, the mRNA of a well-characterized Runx target gene, MMP13, was measured via RT- PCR. D, The effect of CBFb loss on cell migration in MCF7 cells was determined using a transwell migration assay. E, Incubation (72 hours) with the ERa inhibitor, fulvestrant, prior to transwell assay, and Western blot analysis, attenuated migration suggesting that ERa was required for this phenotype. Representative images were taken and number of cells in ﬁeld of view quantiﬁed.

þ generate high-confidence RUNX1-binding peaks (Supplemen- RUNX1 in T47D cells, another ER cell line, and elicited the same tary Fig. S4; refs. 10, 17). These peaks and an ERa dataset were TFF1 upregulation as seen in CBFb / MCF7 cells (Fig. 5B). then interrogated (10, 17, 18). From this meta-analysis, we Furthermore, reexpression of GFP-CBFb in CBFb / MCF7 cells ascertained that TFF1 was bound by both ERa and RUNX1 at rescued the repression of TFF1, confirming that this is a specific both the gene promoter and an enhancer region (Fig. 4C; effect (Fig. 5C). GFP-CBFb is known to be functional as it has been refs. 19, 20). To establish the role of CBFb in the regulation shown to rescue fetal liver hematopoiesis in CBFb / mice (21). of TFF1, TFF1 mRNA was then quantified in CBFb / cells in Depletion of ERa with fulvestrant for 24 hours prior to estrogen theabsenceandpresenceofestrogen.TFF1 was found to be stimulation for 24 hours led to a decrease in the amount of TFF1, upregulated in the presence of estrogen in comparison with the thus confirming ERa is required for the observed increase in TFF1 MCF7 control cells (Fig. 4D). These data suggest that Runx1– following loss of CBFb (Fig. 5D). CBFb complexes antagonize ERa function by repressing the expression of TFF1. Importantly, this is in contrast to the role of Upregulation of TFF1 contributes to the migratory phenotype of Runx1–CBFb as an activator of AXIN1. CBFb / cells Because TFF1 has previously been linked to cell migration, TFF1 CBFb suppresses ERa-mediated expression of TFF1 knockdown was performed in CBFb / and control MCF7 cells To establish the effects of CBFb loss on TFF1 protein expression, to investigate the role of TFF1 in the context of CBFb loss. Seventy- Western blot analysis of TFF1 was performed. TFF1 was signifi- two–hour siRNA-mediated TFF1 knockdown was efficient cantly increased in MCF7 CBFb \ cells in comparison with and led to a decrease in the migration of the MCF7 CBFb / control cells (Fig. 5A). Depletion of CBFb in ERa MDA-MB- cells to a level comparable with the control cells (Fig. 6A and B). 231 cells did not lead to an increase in TFF1, indicating that CBFb This demonstrated that TFF1 is indeed required for the loss alone is not sufficient to induce TFF1 upregulation if ERa is increased migration of the cells following loss of CBFb. not present (Fig. 5A). siRNA was used to knockdown CBFb and Furthermore, as TFF1 is a secreted protein and has previously

www.aacrjournals.org Mol Cancer Res; 2019 OF5

Pegg et al.

Figure 4. RUNX1–CBFB acts as an activator or a repressor in a gene-specific manner. The effect of CBFb knockout on AXIN1 expression level was assessed by RT-PCR (A)and Western blot analysis (B). C, Integrated genome viewer (32) snapshot of TFF1 promoter and enhancer with RUNX1 and ERa- binding peaks presented. Runx1 intersect ChIP-seq is a high- confidence peak file generated using the data from two publicly available Runx1 ChiP-seq datasets (10, 17) and compared with one ERa dataset (18). D, TFF1 mRNA levels in the absence and presence of estrogen (7b-estradiol) were measured via RT-PCR.

been shown to initiate chemotaxis of breast cancer cells, this tractant for a migration assay. When conditioned media derived possibility was investigated (Fig. 6C; ref. 12). Conditioned from CBFb / cells was used as the attractant in a migration media from control or CBFb / cells was used as a chemoat- assay, an increase in cell migration occurred (Fig. 6D).

Figure 5. TFF1 is antagonistically regulated by ERa and RUNX1. A, TFF1 protein levels in MCF7 and MDA-MB-231 cells (control or CBFb/) were measured by Western blot analysis. B, TFF1 protein expression following 72 hours CBFb and RUNX1 siRNA in T47D cells, another ERþ cell line. (C) Western blot analysis of TFF1 protein expression also performed using stable expressing GFP-CBFb rescue construct. D, Cells were pretreated with fulvestrant (or vehicle) for 24 hours prior to 24 hours vehicle/estrogen treatment followed by Western blot analysis to determine whether the TFF1 induction is dependent on ERa activity.

OF6 Mol Cancer Res; 2019 Molecular Cancer Research

Loss of CBFb Induces Migration through TFF1

Figure 6. Knockdown of TFF1 blocks CBFb/-induced cell migration. Seventy-two–hour TFF1 siRNA was used to knockdown TFF1 in control and CBFb/ MCF7 cells. Knockdown was conﬁrmed by Western blot analysis (A) and transwell migration assays (B) were performed and quantiﬁed. C, Western blot analysis of secreted TFF1 found in the media of MDA-MB-231, MCF7 control, and CBFb/ cells. D, Conditioned media (CM) was used as the chemotactant (50/50 with fresh serum- containing media) for transwell assays for MCF7 control or CBFb/ cells. Secreted TFF1 was knocked down with siRNA in control or CBFb/ cells (E)and conditioned media collected and also used as chemoattractant for transwell assay (F). G, Putative mechanism for the function of CBFb–Runx1 complexes in repressing ERa function at TFF1 and AXIN1.

Furthermore, when TFF1 was knocked down, prior to the Discussion generation of conditioned media, the migration was inhibited Because most CBFb mutations in patients with breast cancer (Fig. 6E and F). Taken together, these data demonstrate appear to result in loss of function, we set out to determine the role that CBFb represses ERa-mediated activation of TFF1,thereby þ þ of CBFb in ER breast cancer cells. We initially showed that suppressing the migratory phenotype of ER MCF7 cells missense mutations in CBFb abrogate binding to the runt domain (Fig. 6G).

www.aacrjournals.org Mol Cancer Res; 2019 OF7

Pegg et al.

and are therefore likely to result in loss of function. We subse- and enhancer, where it recruits HDACs to repress gene expression. quently modeled the loss of CBFb in a well-characterized estro- It is therefore possible that repression of TFF1 by RUNX1–CBFb is þ gen-dependent ER breast cancer cell line, MCF7. Our findings also mediated by recruitment of TLE proteins. However, what revealed an alternate mechanism by which Runx–CBFb and ERa determines whether RUNX1–CBFb activates or represses a specific combine to regulate gene expression. We also demonstrated a new gene is not presently understood. To date, no specific combina- role for Runx–CBFb in the prevention of cell migration by sup- tion or organization of RUNX-binding sites within the promoters pressing the motogen TFF1. and enhancers, that predict whether a gene is repressed by The firstpartofthisstudyfocusedonthemechanistic RUNX1–CBFb, have been identified. It will therefore be important function of CBFb missense mutations found in patients with to characterize other RUNX-repressed genes to establish whether breast cancer. It was evident that while missense and truncating there are sequence-specific determinants of RUNX-mediated tran- mutations were found along the CBFb primary sequence, a scriptional repression. largenumberofmissensemutations clustered in the 100–120 Finally, our findings suggest that therapies designed to inhibit þ amino acid region. N104S was the joint highest occurring formation of the RUNX1–CBFb complex in ER breast cancer missense mutation and caused an almost total abrogation of could actually contribute to disease progression, by relieving the CBFb-binding to RUNX1 and RUNX2 and a complete failure to repression on ERa (29, 30). Therefore, because the primary þ form a CBFb–RUNX–DNA heterocomplex. This was not treatment for ER patients is to inhibit ERa activity, it would be altogether surprising as it has previously been shown that important not to use inhibitors of the RUNX1–CBFb complex in mutating N104 to alanine prevented the binding of CBFb to these patients. Indeed, future work on therapies that stimulate the the Runt domain (22). repressive activity of RUNX1–CBFb might provide effective treat- þ The other five mutations, while not as severe as N104S, all ment in some ER patients. exhibited a reduced binding affinity for the Runt–DNA complex. This suggests that a reduction in affinity of CBFb for Runt–DNA is Disclosure of Potential Conflicts of Interest þ sufficient to contribute to development of breast cancer in ER No potential conflicts of interest were disclosed. patients. Previous analysis of AXIN1 regulation has shown that RUNX1 Authors' Contributions activates this gene by counteracting the repressive effect of ERa. Conception and design: H.J. Pegg, P. Shore Loss of RUNX1 function therefore leads to decrease in AXIN1 Development of methodology: H.J. Pegg, H. Harrison expression. In contrast, we have shown that loss of CBFb (and Acquisition of data (provided animals, acquired and managed patients, hence RUNX1 function) leads to an increase in the expression of provided facilities, etc.): H.J. Pegg, H. Harrison, P. Shore Analysis and interpretation of data (e.g., statistical analysis, biostatistics, TFF1 (Fig. 6G). In this case, CBFb acts to repress the activator a – b computational analysis): H.J. Pegg, C. Rogerson, P. Shore function of ER . Therefore, the mechanism by which Runx CBF Writing, review, and/or revision of the manuscript: H.J. Pegg, H. Harrison, and ERa combine to regulate gene expression is gene specific C. Rogerson, P. Shore (Fig. 6G). Administrative, technical, or material support (i.e., reporting or organizing þ Another important observation was that loss of CBFb in ER data, constructing databases): H.J. Pegg cellsleadtoanincreaseincellmigration and that this was due Study supervision: H. Harrison, P. Shore to an increase in TFF1 expression, because knockdown of TFF1 Acknowledgments blocked this effect. TFF1 is an archetypal ERa-activated gene, a – The Bioimaging Facility microscopes used in this study were purchased with frequently used as a readout of ER activity (23 25). However, grants from BBSRC, Wellcome Trust, and the University of Manchester Strategic surprisingly little is known about the role of TFF1 in breast Fund. The FACS Aria within the flow cytometry core was purchased with funding cancer, but it has previously been shown to be a chemoattrac- from the MRC. Thanks to Michael Jackson for guidance and assistance with þ tant for ER breast cancer cells (12). In the stomach, where sorting. The work was funded by a Cancer Research UK PhD Studentship (to H.J. TFF1 has been characterized as a "motogenic factor", leading to Pegg and C. Rogerson), Breast Cancer Now (to P. Shora and H. Harrison), and the disruption of cell–cell and cell–matrix contacts and wound the University of Manchester (to P. Shore). healing (11, 26). The costs of publication of this article were defrayed in part by the payment of RUNX1 is known to repress transcription by recruitment of the page charges. This article must therefore be hereby marked advertisement in corepressor, TLE1 via the VWRPY peptide located at its C-termi- accordance with 18 U.S.C. Section 1734 solely to indicate this fact. nus (27). Interestingly, TLE3 has been implicated in the repression of a subset of ERa target genes, including TFF1 in MCF7 cells (28). Received September 26, 2018; revised November 13, 2018; accepted January In this case, TLE3 is recruited by FOXA1 to both the TFF1 promoter 8, 2019; published first January 17, 2019.

References 1. Qin X, Jiang Q, Matsuo Y, Kawane T, Komori H, Moriishi T, et al. Cbfb 4. Huang G, Shigesada K, Ito K, Wee HJ, Yokomizo T, Ito Y. Dimerization with regulates bone development by stabilizing Runx family proteins. PEBP2beta protects RUNX1/AML1 from ubiquitin-proteasome-mediated J Bone Miner Res 2015;30:706–14. degradation. EMBO J 2001;20:723–33. 2. Chen W, Ma J, Zhu G, Jules J, Wu M, McConnell M, et al. Cbfb deletion in 5. B€ackstrom€ S, Wolf-Watz M, Grundstrom€ C, H€ard T, Grundstrom€ T, Sauer mice recapitulates cleidocranial dysplasia and reveals multiple functions of UH. The RUNX1 Runt domain at 1.25A resolution: a structural switch and Cbfb required for skeletal development. Proc Natl Acad Sci U S A 2014;111: speciﬁcally bound chloride ions modulate DNA binding. J Mol Biol 2002; 8482–7. 322:259–72. 3. Otto F, Thornell AP, Crompton T, Denzel A, Gilmour KC, Rosewell IR, et al. 6. Mendoza-Villanueva D, Deng W, Lopez-Camacho C, Shore P. The Runx Cbfa1,a candidategeneforcleidocranialdysplasiasyndrome,isessentialfor transcriptional co-activator, CBFbeta, is essential for invasion of breast osteoblast differentiation and bone development. Cell 1997;89:765–71. cancer cells. Mol Cancer 2010;9:171.

OF8 Mol Cancer Res; 2019 Molecular Cancer Research

Loss of CBFb Induces Migration through TFF1

7. Ellis MJ, Ding L, Shen D, Luo J, Suman VJ, Wallis JW, et al. Whole-genome leukemia 1 to initiate estrogen-dependent gene expression. Nucleic Acids analysis informs breast cancer response to aromatase inhibition. Nature Res 2014;42:2245–56. 2012;486:353–60. 21. Kundu M, Javed A, Jeon JP, Horner A, Shum L, Eckhaus M, et al. Cbfb 8. Banerji S, Cibulskis K, Rangel-Escareno C, Brown KK, Carter SL, Frederick interacts with Runx2 and has a critical role in bone development. Nat Genet AM, et al. Sequence analysis of mutations and translocations across breast 2002;32:639–44. cancer subtypes. Nature 2012;486:405–9. 22. Tang YY, Shi J, Zhang L, Davis A, Bravo J, Warren AJ, et al. Energetic and 9. Nik-Zainal S, Davies H, Staaf J, Ramakrishna M, Glodzik D, Zou X, et al. functional contribution of residues in the core binding factor beta Landscape of somatic mutations in 560 breast cancer whole-genome (CBFbeta) subunit to heterodimerization with CBFalpha. J Biol Chem sequences. Nature 2016;534:47–54. 2000;275:39579–88. 10. Chimge NO, Little GH, Baniwal SK, Adisetiyo H, Xie Y, Zhang T, et al. 23. Periyasamy M, Patel H, Lai CF, Nguyen VTM, Nevedomskaya E, Harrod A, RUNX1 prevents oestrogen-mediated AXIN1 suppression and b-catenin et al. APOBEC3B-mediated cytidine deamination is required for estrogen activation in ER-positive breast cancer. Nat Commun 2016;7:10751. receptor action in breast cancer. Cell Rep 2015;13:108–21. 11. Taupin D, Podolsky DK. Trefoil factors: initiators of mucosal healing. 24. Carroll JS, Meyer CA, Song J, Li W, Geistlinger TR, Eeckhoute J, et al. Nat Rev Mol Cell Biol 2003;4:819. Genome-wide analysis of estrogen receptor binding sites. Nat Genet 2006; 12. Prest SJ, May FE, Westley BR. The estrogen-regulated protein, TFF1, stimu- 38:1289–97. lates migration of human breast cancer cells. FASEB J 2002;16:592–4. 25. Bourdeau V, Deschênes J, Laperriere D, Aid M, White JH, Mader S. 13. Buache E, Etique N, Alpy F, Stoll I, Muckensturm M, Reina-San-Martin B, Mechanisms of primary and secondary estrogen target gene regulation in et al. Deficiency in trefoil factor 1 (TFF1) increases tumorigenicity of breast cancer cells. Nucleic Acids Res 2008;36:76–93. human breast cancer cells and mammary tumor development in TFF1- 26. Marchbank T, Westley BR, May FE, Calnan DP, Playford RJ. Dimerization of knockout mice. Oncogene 2011;30:3261–73. human pS2 (TFF1) plays a key role in its protective/healing effects. J Pathol 14. Ran FA, Hsu PD, Wright J, Agarwala V, Scott DA, Zhang F. Genome 1998;185:153–8. engineering using the CRISPR-Cas9 system. Nat Protoc 2013;8:2281–308. 27. Levanon D, Goldstein RE, Bernstein Y, Tang H, Goldenberg D, Stifani S, 15. Heigwer F, Kerr G, Boutros M. E-CRISP: fast CRISPR target site identifica- et al. Transcriptional repression by AML1 and LEF-1 is mediated by the TLE/ tion. Nat Methods 2014;11:122–3. Groucho corepressors. Proc Natl Acad Sci U S A 1998;95:11590–5. 16. Bravo J, Li Z, Speck NA, Warren AJ. The leukemia-associated AML1 28. Jangal M, Couture JP, Bianco S, Magnani L, Mohammed H, Gevry N. The (Runx1)–CBF beta complex functions as a DNA-induced molecular clamp. transcriptional co-repressor TLE3 suppresses basal signaling on a subset of Nat Struct Biol 2001;8:371–8. estrogen receptor a target genes. Nucleic Acids Res 2014;42:11339–48. 17. Barutcu AR, Hong D, Lajoie BR, McCord RP, van Wijnen AJ, Lian JB, et al. 29. Cunningham L, Finckbeiner S, Hyde RK, Southall N, Marugan J, Yedavalli RUNX1 contributes to higher-order chromatin organization and gene VR, et al. Identification of benzodiazepine Ro5-3335 as an inhibitor of regulation in breast cancer cells. Biochim Biophys Acta 2016;1859: CBF leukemia through quantitative high throughput screen against 1389–97. RUNX1-CBF interaction. Proc Natl Acad Sci U S A 2012;109:14592–7. 18. Welboren WJ, van Driel MA, Janssen-Megens EM, van Heeringen SJ, Sweep 30. Illendula A, Gilmour J, Grembecka J, Tirumala VSS, Boulton A, Kuntimaddi FC, Span PN, et al. ChIP-Seq of ERa and RNA polymerase II defines genes A, et al. Small molecule inhibitor of CBFb-RUNX Binding for RUNX differentially responding to ligands. EMBO J 2009;28:1418–28. transcription factor driven cancers. EBioMedicine 2016;8:117–31. 19. Jeltsch JM, Robert M, Schatz C, Garnier JM, Brown AM, Chambon P. 31. Zhang L, Lukasik SM, Speck NA, Bushweller JH. Structural and functional Structure of the human oestrogen-responsive gene pS2. Nucleic Acids Res characterization of Runx1, CBFb, and CBFb-SMMHC. Blood Cells Mol Dis 1987;15:1401–14. 2003;30:147–56. 20. Jeong KW, Andreu-Vieyra C, You JS, Jones PA, Stallcup MR. Establishment 32. Robinson JT, Thorvaldsdottir H, Winckler W, Guttman M, Lander ES, Getz of active chromatin structure at enhancer elements by mixed-lineage G, et al. Integrative genomics viewer. Nat Biotechnol 2011;29:24–6.

www.aacrjournals.org Mol Cancer Res; 2019 OF9

The RUNX Transcriptional Coregulator, CBFβ, Suppresses Migration of ER + Breast Cancer Cells by Repressing ERα -Mediated Expression of the Migratory Factor TFF1

Henry J Pegg, Hannah Harrison, Connor Rogerson, et al.

Mol Cancer Res Published OnlineFirst January 17, 2019.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-18-1039

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2019/01/17/1541-7786.MCR-18-1039.DC1

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/early/2019/02/12/1541-7786.MCR-18-1039. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.