Basonuclin-1 Modulates Epithelial Plasticity and TGF-&Beta

ORIGINAL ARTICLE Basonuclin-1 modulates epithelial plasticity and TGF-β1-induced loss of epithelial cell integrity

A Feuerborn1,4, D Mathow1, PK Srivastava2, N Gretz3 and H-J Gröne1

Transforming growth factor-β1 (TGF-β1) is a multifunctional cytokine and critically involved in the progression of a variety of cancers. TGF-β1 signaling can impair tumor development by its anti-proliferative and pro-apoptotic features. In contrast, it may actively promote tumor progression and cancer cell dissemination by inducing a gradual switch from epithelial towards mesenchymal-like cell features (EMT-like), including decreased intercellular adhesion. Here, we show that expression of the transcription factor Basonuclin-1 (Bnc1) modulates TGF-β1-induced epithelial dedifferentiation of mammary epithelial cells. RNAi-mediated repression of Bnc1 resulted in enhanced intercellular adhesion and strongly impaired TGF-β1-dependent sheet disintegration and cell scattering. In contrast, forced expression of Bnc1 modiﬁes plasma membrane/cytoskeletal dynamics and seemingly interferes with the initiation of sustainable cell–cell contacts. Follow-up analyses revealed that Bnc1 affects the expression of numerous TGF-β1-responsive genes including distinct EMT-related transcription factors, some of which modulate the expression of Bnc1 themselves. These results suggest that Bnc1 is part of a transcription factor network related to epithelial plasticity with reciprocal feedback-loop connections on which Smad-factors integrate TGF-β1 signaling. Our study demonstrates that Bnc1 regulates epithelial plasticity of mammary epithelial cells and inﬂuences outcome of TGF-β1 signaling.

Oncogene (2015) 34, 1185–1195; doi:10.1038/onc.2014.54; published online 24 March 2014

INTRODUCTION types and organs, including corneal epithelial cells, testis and 10,12,13 Transforming growth factor-β1 (TGF-β1) affects several processes mammary gland cells. 14 including differentiation, cell scattering and cell death1 in Moreover, Bnc1 expression is required for spermatogenesis epithelial cells. and Bnc1-mutant mice reveal impaired wound-healing of corneal 15 TGF-β1 may initially function as tumor suppressor by inhibiting epithelial cells due to impaired proliferation. proliferation and/or inducing cell death, whereas at later stages of Bnc1 affects RNApol-I and RNApol-II-dependent gene 16 cancer progression TGF-β1 may foster tumor dissemination by expression. Known targets provide circumstantial evidence that promoting epithelial dedifferentiation (EMT-like progression).1–3 Bnc1 may affect a broad range of cellular functions, including 17,18 Several TGF-β1-responsive transcription factor-families modify cell-adhesion, intracellular transport and ion-channels. epithelial cell integrity and affect the expression of junction- Various cancer cells harbor promoter methylation and tran- 19,20 proteins (for example, E-cadherin, Occludin and others).4,5 scriptional repression of Bnc1 suggesting that reduced Bnc1 However, many of them do so in the absence of TGF-β1 (for expression confers a survival advantage, though this awaits further proof.19,21 In contrast, Bnc1 expression is elevated in breast cancer example, FoxQ1, Snail1, Zeb1). Hence, the potential of those 22 fi β cells with increased invasive and metastatic capacity and directly factors to affect epithelial plasticity is not speci ed by TGF- 1. 23 On the other hand, transcription factors integrate TGF-β1 regulated by p63 in squamous carcinomas of the head and neck. signaling into cell-specific transcriptional contexts.6,7 Comprehen- Hence, Bnc1 may have diverse roles in epithelial cells and tumor sive identification and understanding of putatively relevant progression, depending on cellular context and disease stage. Further understanding of Bnc1 and its putative interaction with transcription factors remains challenging. Evidently, Smad-factors β β 7 TGF- 1 may reveal important mechanistic insights into epithelial are central mediators of TGF- 1 signaling and whether the so- cell physiology and associated pathophysiological conditions. called ‘Smad-independent’ TGF-β1 signaling is truly separate from 8 Our study implicates that Bnc1 affects epithelial plasticity of Smads requires further proof. mammary epithelial cells and modifies outcome of TGF-β1 signaling. We studied involvement of transcription factor Basonuclin-1 (Bnc1) in TGF-β1 signaling, which we found transcriptionally induced on the level of mRNA upon TGF-β1 stimulation in RESULTS mammary epithelial cells. Expression of Bnc1 is induced by canonical TGF-β1 signaling and These cells represent a well-characterized and relevant model affected by several transcription factors related to epithelial for TGF-β1-dependent processes and EMT-like dedifferentiation.3,9 plasticity Bnc1 was discovered in keratinocytes and has been associated Our previous work suggested that Bnc1 is transcriptionally 10,11 with proliferation but is also expressed in several other cell induced by TGF-β1.24 Consistently, induction was blocked by

1Department of Cellular and Molecular Pathology, German Cancer Research Centre (DKFZ), Heidelberg, Germany; 2Physiological Genomics and Medicine, MRC Clinical Sciences, Imperial College, London, UK and 3Department of Medical Research, Faculty of Medicine, Medical Research Centre (ZMF), University of Heidelberg, Mannheim, Germany. Correspondence: Dr A Feuerborn or Professor Dr H-J Gröne, Department of Cellular and Molecular Pathology, German Cancer Research Centre, Im Neuenheimer Feld 280, Heidelberg 69120, Germany. E-mail: [email protected] or [email protected] 4Current address: Sir William Dunn School of Pathology, University of Oxford, South Parks Road, Oxford, OX1 3RE, UK Received 16 December 2012; revised 3 January 2014; accepted 6 January 2014; published online 24 March 2014 Bnc1 inﬂuences outcome of TGF-β1-signaling A Feuerborn et al 1186 treating cells with transcription inhibitor 5,6-dichlorobenzimida- not result in elevated Bnc1 mRNA expression (Figures 1e zole 1-β-D-ribofuranoside (Figure 1a). and f). To analyze whether induction is related to canonical TGF-β1 Recruitment and binding of Smad-factors to genomic regions signaling, we assayed mRNA and protein expression of Bnc1 in involves several complex aspects, including binding site accessi- Smad3 and Smad4 knockdown cells. bility, chromatin remodeling and expression and binding of other Knockdown of Smad3 or Smad4 impaired TGF-β1-mediated transcription factors.7 induction of Bnc1 mRNA (Figure 1b). However, Bnc1 In addition to Smad-binding regions, the prediction of protein expression was not noticeably increased by TGF-β1 transcription factor binding sites within Bnc1-promoter regions (Figures 1c and 2c). using three different prediction tools25–27 revealed numerous Chromatin-IP indicated that Smad3 associates with the transcription factors that may putatively inﬂuence Bnc1 expres- Bnc1 promoter in response to TGF-β1 (Figure 1d). However, sion. These include (TGF-β1-responsive) factors relevant for overexpression of wild-type or constitutively active Smad3 did modulating epithelial plasticity: for example, FoxA2 (HNF-3B),

Figure 1. Bnc1 mRNA expression is transcriptionally induced by TGF-β1 in a Smad-dependent manner and constitutively affected by transcription factors related to epithelial plasticity. (a) Quantitative RT–PCR-based analyses of Bnc1 mRNA expression in cells treated as indicated. TGF-β1 and DRB were applied simultaneously for 24 h (n = 3). (b) qRT-PCR-based analyses of mRNA expression of Bnc1 in Smad3- (left) or stable Smad4-knockdown cells (right) treated with TGF-β1 for 24 h (n = 3). (c) Western blot analyses of Bnc1, Smad3 in Smad3 siRNA- transfected cells − /+ TGF-β1 (left panel), and analyses of Bnc1, Smad4 in stable Smad4 knockdown cells − /+ TGF-β1 (right panel). Gapdh- staining conﬁrms equal input of total protein. (d) Chromatin-IP experiments using NM18 cells − /+TGF-β1 for 1 h. DNA was ampliﬁed using primers for promoter region (898–776 upstream of the transcriptional start site) of Bnc1 before chromatin pull-down (input) or after pull-down using antibodies against Smad3 or control IgG antibodies. M = Marker (100 bp ladder). (e) Western blot analysis of Smad3 of cells transfected (48 h) with control-plasmid (pIRESpuro3) or plasmids encoding wild-type- (pCMV2-Flag-WT-Smad3) or constitutively active-Smad3 (pCMV2- Flag-CA-Smad3). (f) Expression of Bnc1 mRNA in cells transiently transfected as in Figure 1e. (g) qRT–PCR-based analyses of mRNA expression of FoxQ1, c-Jun, Zeb1 and FoxA2 in cells transfected with indicated siRNAs. Treatment with TGF-β1 (24 h) is omitted in the case of FoxA2 as FoxA2 is decreased upon TGF-β1 stimulation (see Figures 2c and d and Supplementary File 4). (h) Expression of Bnc1 mRNA in cells shown in Figure 1g. (n = 3). RT–PCR, PCR with reverse transcription.

Figure 2. Bnc1 inﬂuences numerous TGF-β1-responsive genes and affects expression of a distinct set of transcription factors related to epithelial plasticity. (a) Number of differentially expressed genes in dependence of TGF-β1 and Bnc1 depicted as Venn-diagram. Out of 2305 genes differentially expressed within 24 h of TGF-β1 stimulation, 1/3 (767) are affected by repression of Bnc1. (b) Activity proﬁles for Jun, FoxA2 and Zeb1 in dependence of TGF-β1 and Bnc1. Unprocessed microarray data were analyzed via MARA to predict differences in transcription factor activity between respective settings (ns-siRNA//ns-siRNA+TGF-β1//Bnc1 siRNA+TGF-β1). (c) Time-resolved protein expression analyses of Zeb1, Bnc1, FoxA2 and c-Jun in response to TGF-β1 stimulation. (d) Quantitative RT–PCR-based analyses of mRNA expression of c-Jun, FoxA2 and Zeb1 in cells transfected with Bnc1 siRNA − /+TGF-β1 (24 h). (e) Protein expression of c-Jun and activated (phosphorylated on serine 63 and 73) c-Jun, FoxA2, Zeb1 and Bnc1 in Bnc1 knockdown cells − /+ TGF-β1 stimulation (24 h). Note that repression of Bnc1 impairs the TGF-β1-mediated induction of Zeb1 proteins with an observed molecular weight of ~ 130 kDa and ~ 180 kDa. (f) Expression and nuclear localization of FoxA2 (green) in Bnc1 knockdown cells − /+ TGF-β1 treatment (24 h) using confocal microscopy. Scale bar, 50 μM; red, phalloidin; blue, DAPI. RT–PCR, PCR with reverse transcription.

FoxQ1 (HFH-1), Zeb1 (AREB6/deltaEF1) and c-Jun (Supplementary In contrast, cells with an impaired Bnc1 expression maintained Files 1, 2a and b). sheet integrity and remained unscattered (Figure 3b, Remarkably, all these factors constitutively modified expression Supplementary Movie 1), suggesting alterations in intercellular of Bnc1 (Figures 1g and h) and influenced the outcome of TGF-β1- adhesion strength. This was assessed using hanging-drop assays. induced Bnc1 expression. As shown, TGF-β1-treated control-transfected cells lose the These results reveal that numerous factors are constitutively ability to form stable aggregates, whereas repression of Bnc1 balancing Bnc1 expression and that Smad3 recruitment in significantly impaired TGF-β1-dependent loss of cell–cell adhesion response to TGF-β1 may rely on multiple components. (Figure 3c). This was paralleled by increased expression of the adherens- junction proteins E-cadherin and Lmo7 (Lim-domain-only 7) in Bnc1 affects intercellular adhesion independent of the adherens- Bnc1 knockdown cells (Figures 3d and e). We therefore tested if junction proteins E-cadherin and Lmo7 repression of E-cadherin or Lmo7 would antagonize enhanced To test if expression of Bnc1 is functionally relevant for the intercellular adhesion of Bnc1 knockdown cells. However, neither outcome of TGF-β1 signaling, we inhibited expression of Bnc1 repression of E-cadherin nor Lmo7 sufficed to overcome this (Figure 3a). Consistent with previous work,9 TGF-β1 treatment phenomenon (Figures 3f–i, Supplementary Figure 1a). Moreover, resulted in the dissolution of epithelial sheet structures. double-labeling of E-cadherin and Lmo7 in E-cadherin- or

Figure 3. Repression of Bnc1 impairs TGF-β1-induced loss of intercellular adhesion and induction of cell scattering independent of E-cadherin and Lmo7. (a) Quantitaive RT–PCR-based analyses of mRNA expression of Bnc1 in Bnc1 knockdown cells untreated (left) or treated with TGF-β1 (24 h) (right), (n = 3). (b) Repression of Bnc1 impairs TGF-β1-induced cell scattering. Cells were transfected as indicated and treated with TGF-β1 for 24 h, ﬁxed in methanol and stained with Giemsa. Scale bar, 200 μm. (c) Hanging-drop assays for control- and Bnc1-knockdown cells − /+ TGF-β1 stimulation. Cells were transfected and treated as indicated. Cells were harvested and further cultured as single-cell suspensions in hanging drops and allowed to re-aggregate for 2 h. Re-aggregated cell clusters were mechanically challenged in a deﬁned manner and evaluated for dispersion. The assays were evaluated as previously described and results expressed as ‘Aggregation Index’53 (see also Materials and methods). (d) Confocal microscopy staining of E-cadherin (green) in control- and Bnc1 knockdown cells − /+TGF-β1. Bnc1 knockdown cells reveal increased membranous and cytoplasmic E-cadherin expression compared with control cells in conjunction with TGF-β1- stimulation. Scale bar, 20 μm; blue, DAPI. (e) Staining of Lmo7 (green) of control- and Bnc1 knockdown cells. Bnc1 knockdown results in a pronounced expression of membranous Lmo7. Scale bar, 20 μm. (f) E-cadherin expression (green) in stably-infected control (ns-shRNA) or E-cadherin shRNA (Ecad-shRNA) cells transiently transfected and treated as indicated. Scale bar, 20 μm. (g) Hanging-drop assay for stably infected ns-shRNA and E-cadherin shRNA cells transiently transfected with control or Bnc1 siRNA − /+TGF-β1 (24 h). (h) Lmo7 expression (green) in control siRNA-, Bnc1 siRNA-, and Lmo7/Bnc1 siRNA-transfected cells − /+ TGF-β1 treatment (24 h). Scale bar, 20 μm; blue, DAPI. (i) Hanging drop assay for cells reverse transfected with ns-siRNA, Bnc1 siRNA and Lmo7/Bnc1-siRNA − /+ TGF-β1 stimulation (24 h). RT–PCR, PCR with reverse transcription.

Oncogene (2015) 1185 – 1195 © 2015 Macmillan Publishers Limited Bnc1 influences outcome of TGF-β1-signaling A Feuerborn et al 1189 Lmo7-knockdown cells suggested that membranous expression Protein expression of Zeb1 revealed more complex patterns. and localization of Lmo7 was not markedly affected TGF-β1 stimulation noticeably induced Zeb1 protein expression by knockdown of E-cadherin and vice versa (Supplementary (∼130 kDa and ∼ 180 kDa) after 24 h, which was again decreased Figures 1b and c). after 48 h of TGF-β1-treatment. In contrast, we detected a Of note, cells revealed reduced intercellular adhesion within 24 continuous increased expression of Zeb1-proteins with a mole- h of TGF-β1 treatment (Figure 3c) but overall E-cadherin cular weight of ∼ 90 kDa and ~ 200 kDa, which were expressed at expression seemed only marginally affected (Supplementary lower levels in TGF-β1-treated cells. Finally, a signal of ∼ 110 kDa Figure 1a, lanes 1–2 and 5–6, Supplementary File 3). This may remained rather consistently expressed and unaffected by TGF-β1 suggest that an overall decrease in E-cadherin protein expression induction (Figure 2c). is not a requisite for loss of intercellular adhesion. In agreement, Bnc1 protein expression did not differ between control- and effective repression of E-cadherin did not decrease intercellular TGF-β1-stimulated cells compared with the aforementioned adhesion (Figure 3g, Supplementary Figure 1a). factors. Gene expression profiling (see next section) revealed that Lmo7 These data demonstrate distinct expression patterns of the mRNA expression is decreased upon TGF-β1 stimulation in considered transcription factors over time, including their different control-transfected cells, whereas it is not reduced in Bnc1 responsiveness towards TGF-β1 stimulation on the level of protein knockdown cells. In contrast, E-cadherin mRNA remains repressed expression. Further, they document significant expression differ- in Bnc1 knockdown cells upon TGF-β1 stimulation (Supplementary ences within untreated cells (for example, FoxA2, Zeb1) but File 3, Supplementary Figure 1a, lanes 3–4 and 7–8). Therefore, were not explicitly instructive to infer a regulatory connection elevated protein levels of E-cadherin in Bnc1 knockdown cells are between c-Jun, FoxA2, Zeb1 and Bnc1 in untreated and TGF-β1- not on the basis of increased E-cadherin mRNA expression. treated cells. Further, it suggests that Bnc1 is not a direct regulator of Analyses of mRNA expression of c-Jun, FoxA2 and Zeb1 in Bnc1 E-cadherin expression. Consistently, in face of stable repression knockdown cells in conjunction with TGF-β1 stimulation showed of E-cadherin expression, knockdown of Bnc1 slightly restored that repression of Bnc1 antagonized mRNA expression of c-Jun, E-cadherin protein expression (Supplementary Figure 1a, compar- Zeb1 and FoxA2 (Figure 2d). Moreover, c-Jun and Zeb1 protein ing lanes 5–6 with lanes 7–8). expressions were decreased upon knockdown of Bnc1 (Figure 2e). In summary, our results imply that E-cadherin and Lmo7 are In contrast, FoxA2 protein levels remained substantially reduced in associated with alterations in intercellular adhesion and restruc- Bnc1 knockdown cells in response to TGF-β1 despite impaired turing of adhesion complexes but seemingly functionally neutral TGF-β1-mediated decrease of FoxA2 mRNA in Bnc1-repressed cells to increased cellular aggregation of Bnc1 knockdown cells. (Figures 2d and e). However, confocal microscopy (Figure 2f), which confirmed reduced expression of FoxA2 in TGF-β1-treated cells, suggested elevated nuclear expression in Bnc1 knockdown Bnc1 regulates a subset of transcription factors related to cells compared with control-transfected cells during TGF-β1 epithelial plasticity and TGF-β1 signaling signaling. This supports the notion of some FoxA2-activity We performed gene expression profiling and identified 2305 transcripts functionally restored in TGF-β1-treated Bnc1-repressed cells as modulated by TGF-β1 treatment. Out of these, 767 transcripts were suggested by MARA (Figure 2b). affected in expression upon Bnc1 knockdown (Figure 2a). Such influence on TGF-β1-dependent gene expression changes suggested that Bnc1 exerts its impact on TGF-β1 signaling partially by c-Jun, FoxA2 and Zeb1 modulate cell adhesion and cell scattering potentially affecting other (TGF-β1-responsive) transcription factors. of NM18 cells In support of this, we identified 82 differentially expressed We next tested if repression of c-Jun, FoxA2 and Zeb1 may alter genes annotated within the GO-term ‘Transcription factor activity’ intercellular adhesion strength and cell scattering, features in TGF-β1-treated cells. Out of these, 30 genes were altered in affected by repression of Bnc1. expression upon Bnc1 knockdown (Supplementary File 4). Therefore, we inhibited their expression (Figures 1g and 4a) and We confined these results by employing two additional and subjected cells to hanging-drop assays (Figure 4b) and monitored complementary approaches: (a) we predicted activity changes of alterations in cell scattering and epithelial sheet formation transcription factors using Motif Activity Response Analysis (Figures 4c–e and Supplementary Movies 2–4). (MARA) (Figure 2b, Supplementary File 4) and (b) employing Repression of c-Jun was sufficient to partially prevent TGF-β1- oPOSSUM, we searched for overrepresented transcription factor induced cell dispersion and loss of intercellular adhesion binding motifs within TGF-β1-responsive genes affected by Bnc1- (Figures 4b and c and Supplementary Movie 2) which supports repression (Supplementary File 5a). previous reports.29,33 Taken together, these approaches suggested c-Jun, FoxA2 and Knockdown of FoxA2 resulted in disintegration of epithelial Zeb1 to participate in partially mediating Bnc1-dependent sheets without notably modifying intercellular adhesion strength phenotypes in TGF-β1-stimulated cells (Figure 2b, Supplementary (Figures 4b and d). This may suggest that additional (FoxA2- Files 4 and 5a). Interestingly and as shown above (Figures 1g and affected) mechanisms that do not primarily modulate intercellular h), those factors endogenously affect Bnc1-expression, implying adhesion may contribute to Bnc1-dependent cell scattering in that regulation between them may occur in reciprocal manners. response to TGF-β1, consistent with findings that loss of epithelial Also, c-Jun, FoxA2 and Zeb1 have previously been suggested to sheet integrity requires the modulation of various cellular features, control EMT-like transitions, including alterations in epithelial including motility.36 Indeed, time-lapse microscopy showed that junctions and cell scattering.28–33 FoxA2-repressed cells detaching from epithelial sheets are highly In contrast, other EMT-related transcription factors (for example, mobile compared with respective controls (Supplementary Movie JunB, Zeb2 and FoxQ1)28,34,35 were not affected by Bnc1 3), which suggests that these cells disintegrate from epithelial repression. sheets by overcoming cell–cell adhesion through increased Time-resolved protein expression analyses of c-Jun, FoxA2, Zeb1 motility forces.37 and Bnc1 in response to TGF-β1 revealed early and consistent However, we also observed that phenotypic alterations of expression changes of c-Jun and FoxA2. Interestingly, FoxA2 FoxA2-repressed cells depend on cell density (Supplementary protein expression was increasingly elevated in untreated cells Movie 3) and we cannot exclude that a subtle decrease in across the analyzed time points, correlating with establishment of intercellular adhesion of cells repressed in FoxA2 expression cell–cell adhesion and cell density (Figure 2c). escape detection due to limited sensitivity of hanging-drop assays.

Figure 4. Bnc1-affected transcription factors are functionally involved in intercellular adhesion and cell scattering. (a) Protein expression of c-Jun, FoxA2 and Zeb1 in cells transfected and treated as indicated. FoxA2 knockdown cells are not treated with TGF-β1, as TGF-β1 stimulation decreases FoxA2 expression. Hence, FoxA2 knockdown mimics this aspect. Note, that transfection of Zeb1 siRNAs reduces the expression of Zeb1 proteins with an observed molecular weight of ~ 130 kDa and ~ 180 kDa. (b) Hanging-drop assays of c-Jun, FoxA2 and Zeb1 knockdown cells. Cells were transfected and treated as indicated and processed as described for hanging-drop assays. (c) Light microscopy images of Giemsa stained c-Jun knockdown cells − /+TGF-β1. Repression of c-Jun partially impaired dissolution of epithelial sheets upon TGF-β1- stimulation. (d) Repression of FoxA2 is partially sufﬁcient to macerate epithelial sheets. Cells were transfected as indicated, ﬁxed in methanol, stained with Giemsa and imaged via light microscopy. (e) Light microscopy images of Giemsa stained Zeb1 knockdown cells − /+ TGF-β1. Repression of Zeb1 strongly impaired dissolution of epithelial sheets upon TGF-β1-stimulation. Scale bars, 200 μm. (f) Confocal images of E-cadherin expression (green) in Zeb1 knockdown cells − /+ TGF-β1. Scale bar, 20 μm. (g) Confocal images of Lmo7 expression (green) in Zeb1 knockdown cells − /+ TGF-β1. Scale bar, 20 μm

As in the case of Bnc1, repression of Zeb1 strongly impaired These results may imply that alterations in intercellular adhesion TGF-β1-dependent dispersion of epithelial sheets (Figures 4b and and scattering imposed upon repression of Bnc1 may rely on the e and Supplementary Movie 4), consistent with a role for Zeb1 in combination of several factors, including c-Jun, FoxA2 and Zeb1. EMT-like progression.28,31 However, E-cadherin and Lmo7 expression clearly differed Stable expression of Bnc1 and Zeb1 modiﬁes epithelial sheet between Bnc1 and Zeb1 knockdown cells (comparing Figure 3d formation with Figure 4f, Figure 3e with Figure 4g). Zeb1 knockdown cells Our previous results have shown that repression of Bnc1 results did not accumulate cytoplasmic E-cadherin, nor membranous in enhanced cell aggregation and impairs TGF-β1-mediated Lmo7 compared with Bnc1 knockdown cells. epithelial dedifferentiation.

Figure 5. Bnc1-overexpression cells form less-sustainable cellular sheets and reveal dynamic cellular protrusions. (a) Bnc1 mRNA expression analyses of cells stably-infected with control (pMSCV) or pMSCV-Bnc1. Cells were infected in triplicate, maintained as puromycin-resistant cell pools (clones 1–3) and analyzed for the expression of Bnc1 mRNA. (b) Light microscopy images of control- (pMSCV) and Bnc1-overexpressing cells (pMSCV-Bnc1) grown at lower density (shown for clone 2). Two individual areas of the same plate are depicted for each clone. Scale bar, 40 μm. (c) Confocal images of E-cadherin (green) in cells stably expressing Bnc1 (shown for clones 1 and 2). Cells were ﬁxed in acetone and stained for E-cadherin. As expected, single cells that are not integrated into sheet structures reveal lower expression of E-cadherin. Scale bar, 50 μm, please also refer to Supplementary Movie 5 which reveals some dynamics of sheet formation and disintegration. (d) Confocal images of Lmo7 (green) in Bnc1-overexpressing cells (shown for clones 1 and 2). Scale bar, 50 μm; blue, DAPI, please also refer to Supplementary Movie 5, which reveals some dynamics of sheet formation and disintegration. (e) Overexpression of Bnc1 does not decrease intercellular adhesion as assessed by hanging-drop assays. Cells were plated in six-well plates one day before hanging-drop assays and further processed as previously mentioned. (f) Western blot analyses of Bnc1, Zeb1, c-Jun and FoxA2 in cells overexpressing Bnc1. Cells were plated 24 h before harvesting and processes for Western blot analyses as described. (g) Protein expression analyses of cells stably overexpressing Zeb1. Overexpression of full-length Zeb1 induces the expression of Zeb1 of molecular weights observed at ~ 130 kDa and ~ 180 kDa. (h) Analyses of relative intercellular adhesion strength of Zeb1-overexpression and control cells via hanging-drop assays. Cells were plated one day before hanging-drop assays and processed as previously mentioned. (i) Light microscopy images of cells stably overexpressing Zeb1. Zeb1- overexpression induces aberrant sheet formation in conjunction with the formation of long spindle-shape protrusions. Scale bar, 40 μm.

© 2015 Macmillan Publishers Limited Oncogene (2015) 1185 – 1195 Bnc1 influences outcome of TGF-β1-signaling A Feuerborn et al 1192 To test if overexpression of Bnc1 antagonizes epithelial sheet TGF-β1-responsive genes being constitutively affected by formation, we established NM18 cells stably expressing Bnc1 repression of EMT-related transcription factors. Thus, protein (Figures 5a and f). concentration of Bnc1 is not and need not change appreciably. Forced expression of Bnc1 resulted in morphological alterations Further, epithelial-related transcription factors regulate each of NM18 cells and an incoherent formation of epithelial sheets other in reciprocal manners (for example, Bnc1, c-Jun, Zeb1, (Figures 5b–d). FoxA2). Consequently, this suggests that similarities in the In contrast to control cells, Bnc1-overexpressing cells revealed outcome of TGF-β1 signaling upon individual repression of, for dynamically-switching cellular protrusions dispersed around the example, c-Jun, Zeb1 or Bnc1 are partially on the basis cell body (Figure 5b and Supplementary Movie 5). Furthermore, of overlapping consequences at the level of TGF-β1-induced these cells form only transiently sustained cell sheets in contrast to Smad-integration (for example, promoter-binding). control-infected cells which maintain sheet integrity Moreover, recent data using chromosome confirmation (Supplementary Movie 5). However, increasing cell density caused capture-related techniques42 suggest that gene–gene interactions phenotypes to become increasingly masked. Furthermore, over- dynamically evolve in ‘transcription factories’, subnuclear foci, expression of Bnc1 did not reduce intercellular adhesion, as which specialize in transcribing Smad-dependent genes in assayed by hanging-drop assays (Figure 5e). Moreover, expression response to TGF-β1.43 Consequently, transcription factors related of c-Jun, FoxA2 and Zeb1 remained unaltered in Bnc1- to epithelial plasticity (for example, Bnc1 and others) may specify overexpressing cells (Figure 5f). the outcome of TGF-β1 signaling by organizing gene/chromatin We compared these results with the results obtained from cells configuration with functional consequences for integration of overexpressing Zeb1 (Figure 5g), a known inducer of epithelial Smad-signaling. In this regard it is noteworthy that, for example, plasticity and EMT-like progression. Although knockdown of Zeb1- Zeb1 and c-Jun may interact directly with Smads.44,45 impaired TGF-β1-mediated loss of intercellular adhesion, forced Protein expressions of EMT-related transcription factors reveal expression of full-length Zeb1 was insufficient to decrease cell–cell considerable heterogeneity and dynamics in the absence of TGF- adhesion strength (Figure 5h). In addition, also Zeb1- β1 which refutes the perception of defining factors potentially overexpressing cells revealed aberrant epithelial sheet formation relevant for the outcome of TGF-β1 signaling by induced/reduced and formed characteristic, spindle-shaped cell protrusions protein expression levels (only). Again, protein expression changes (Figure 5i), which were morphologically distinct from cell may not even be required (for example, Bnc1), as in many protrusions formed by Bnc1-overexpression cells. instances target genes are already primed before stimulation.41 Taken together, these results show that overexpression of Bnc1 Consequently, any transcription factor that interferes with (like Zeb1) modifies epithelial cell plasticity. Smad-binding and function (in a direct or indirect manner) has the potential to affect the outcome of TGF-β1 signaling. Many such factors may escape recognition as they are not noticeably DISCUSSION affected in expression or modification by TGF-β1. Expression of Bnc1 is sensitive to different signaling pathways.38,39 Finally, putative Smad-binding (for example, Smad3) motifs are Koinuma et al.39,40 detected Bnc1-regulatory regions in overrepresented within genomic regions of Bnc1-affected genes genome-wide binding studies of Smad2/3 and Smad4 in in conjunction with TGF-β1 signaling (Supplementary File 5b). keratinocytes but did not further assess regulatory details or any However, it will be crucial to further strengthen these notions. potential functional relevance. However, their findings pointed to As previously revealed for other transcription factors (for a possible connection between TGF-β1 signaling and Bnc1. example, Oct4), genome-wide binding of Bnc1 (− /+TGF-β1), Here we identified Bnc1 mRNA expression as transcriptionally correlation with Smad-binding and especially Smad-patterns in induced in mammary cells undergoing TGF-β1-mediated epithelial Bnc1-repressed cells could provide critical insights. dedifferentiation. Supposedly, the immense impact of Bnc1 on TGF-β1-mediated Smad-factors are essential for TGF-β1 signaling. Recruitment gene expression reported here supports a multilayered scenario and binding of Smad-complexes to DNA-motifs involves many with regard to Bnc1 expression and Smad-signaling integration. more different facets than previously anticipated. These include Phenotypically, repression of Bnc1 abolished TGF-β1-mediated chromatin remodeling, accessibility to Smad-binding elements cell scattering and loss of intercellular adhesion. An interesting and goes beyond a simplistic model of cell-type-specific observation is the ambivalent expression of E-cadherin protein transcription factors recruiting Smad-complexes and determining and mRNA expression in dependence of Bnc1 and TGF-β1. their actions.7 Though, Bnc1 knockdown cells reveal elevated protein levels of Of note, repression of Smad4 renders NM18 cells inert to TGF- E-cadherin, mRNA expression remained suppressed upon TGF-β1 β1-induced phenotypic changes,9 indicating that integration of stimulation. This demonstrates that the repressive potential of TGF-β1 signaling is exclusively Smad-dependent in these cells. TGF-β1 signaling on E-cadherin expression remains in Bnc1 Importantly, this does not refute relevance of signaling pathways knockdown cells. The TGF-β1-responsive transcription factors also engaged upon TGF-β1 stimulation and diverging from Smads. FoxQ1, which is unaffected by Bnc1 expression (Supplementary However, it means the loss of Smad4 cannot be compensated for File 4) might have a role for this observation.24,35,46 by other factors, making Smad4 (Smad-factors) the central and It has previously also been noted that E-cadherin is dispensable immediate node for integration of TGF-β1 signaling. for the maintenance of cell–cell adhesion in several other cell Following this notion, we propose that EMT-related transcrip- types47,48 and decreased expression of E-cadherin is not a tion factors (in context of TGF-β1 signaling) significantly influence requisite for cell scattering or invasion.37,49 Instead, traction forces how Smads integrate TGF-β1 signaling by largely defining the pinpointed to adherens-junctions significantly contribute to the nuclear/genomic context on which Smad-factors are imposed on dissolution of adhesion complexes.37 upon TGF-β1 stimulation. On the basis of our findings, we speculate that increased This interpretation takes account of several experimental protein expression of E-cadherin in Bnc1-repressed cells may observations, including some made here for Bnc1. partially result from fostered intercellular adhesion, which in turn Bnc1 (like other factors, for example, FoxQ1, JunB, Zeb1) is not supports establishment and stabilization of adherens-junctions. specific for TGF-β1 signaling, yet expression specifies the response This notion would also be consistent with staining patterns of and largely determines TGF-β1-induced gene expression changes. Lmo7. On the other hand, E-cadherin-accumulation in the Transcription factors often prime target genes for induction before cytoplasm presumably indicates a range of alterations occurring signaling initiation (for example, JunB),41 consistent with many in Bnc1 knockdown cells (for example, turnover, transport or

Oncogene (2015) 1185 – 1195 © 2015 Macmillan Publishers Limited Bnc1 influences outcome of TGF-β1-signaling A Feuerborn et al 1193 recycling of E-cadherin). Moreover, despite increased intercellular buffered saline (PBS, including 1 mM PMSF, 1 μg/ml pepstatin A, 1 μg/ml adhesion of Zeb1 knockdown cells, Lmo7 is not markedly aprotinin). recruited to the membrane as compared with Bnc1-repressed Cleared supernatants were incubated with antibodies against Smad3 cells. This implies that intercellular adhesion may not comprehen- (1:50, #9523, Cell Signaling Technologies, Danvers, MA, USA) or rabbit-IgG sively account for these observations and imply additional antibody (#2729, Cell Signaling Technologies) overnight at 4 °C. Chromatin was subjected to PCR using primers: Bnc1-prom forward 5′- alterations distinctly occurring in Bnc1-repressed cells. ′ fl cactttggcctctcttcagg and Bnc1-prom reverse 5 -aaaagaatccttcggcccta. Finally, overexpression of Bnc1 also in uences epithelial Chromatin-immunoprecipitation experiments were performed twice inde- plasticity. We speculate that the aberrant and dynamic pendently with reproducible outcome. cell protrusions in these cells interfere with initiation of stable cell–cell interactions. This is consistent with an influence of cytoskeletal dynamics on formation and promotion of cell–cell adhesion.50 Confocal microscopy and live cell imaging fi − These observations reveal that also enhanced expression of Cells were transfected and treated on coverslips, acetone- xed ( 20 °C) for Bnc1 modifies epithelial plasticity. However, overexpression 5 minutes and air dried. fi Slides were blocked in 3% (w/v) bovine serum albumin-PBS and approaches are inevitably arti cial (for example, expression is incubated with antibodies against E-cadherin (BD Transduction Labora- not physiologically controlled) and may limit solid conclusions at tories, Franklin Lakes, NJ, USA, clone 36, 1:3000), Lmo7 (Santa Cruz, CA, this point. USA, sc-98422, 1:500) and FoxA2 (Cell Signaling Technologies, #3143, In summary, our data suggest that Bnc1 is a relevant effector of 1:500). Alexa-fluor 488- or 546-conjugated secondary antibodies (Molecular epithelial plasticity in mammary epithelial cells. In addition, this Probes, Invitrogen, Carlsbad, CA, USA, A-21202, A-21206 and A-11003, study suggests that Bnc1 may be a component of a transcription 1:1000) were used to detect primary antibodies. Antibodies were diluted in factor network associated with epithelial plasticity and that 1% (w/v) bovine serum albumin-PBS. efficient TGF-β1-dependent epithelial dedifferentiation depends Cell nuclei were counterstained with DAPI (100 ng/ml). Cells were on the expression of Bnc1. embedded using Fluoromount (Sigma) and imaged using the LSM700 laser confocal microscope (Zeiss, Jena, Germany). For live cell imaging, cells were reverse transfected on CELLview Glass Bottom Slides (four compartments, Advanced, Greiner Bio-One, Solingen, MATERIALS AND METHODS Germany). Selected spots were imaged using a motorized inverted IX81 Cell culture, treatments and retroviral infections microscope coupled to a CCD Hamamatsu Orca-ER camera. Pictures were taken every 6 minutes for around 24 h. Recording was NM18, Smad4 knockdown (NM18-S4kd) and control cells (NM18-pRS)9 initiated 10–15 minutes after cells were induced with TGF-β1. Cells were were cultured in Dulbecco’s Modified Eagle’s medium/high glucose media kept at a temperature (37 °C) and CO (5%)-controlled environment (ATCC, Manassas, VA, USA), 10 μg/ml insulin (Sigma, St Louis, MO, USA) and 2 throughout recording. Images were converted into avi-files using Fiji 10% (v/v) fetal calf serum (Gibco, Carlsbad, CA, USA). software. Movies are shown with 15 frames per second. Cells were treated with 5 ng/ml TGF-β1 (PeproTech, Rocky Hill, NJ, USA) for indicated time points. 5,6-dichlorobenzimidazole 1-β-D-ribofuranoside (Sigma) was used to Cloning of Bnc1and Zeb1 block transcription (40 μM). fi ’ fi ’ Coding sequence of Bnc1 was ampli ed from pDEST26-Bnc1 (Source Plat-E cells were cultured in Dulbecco s Modi ed Eagle s medium/high BioScience LifeSciences, Nottingham, UK) using Phusion High fidelity glucose media containing 10% (v/v) fetal calf serum. polymerase (Thermo Fisher Scientific, Waltham, MA, USA) with primers: Plat-E cells were transfected with respective plasmids using Lipofecta- Bnc1 forward 5′-gatgcggcggtcgccgag and Bnc1 reverse 5′-cttgttactggagg mine 2000 according to the supplier’s instructions. fi μ tggctt. Virus-containing supernatants were ltered (0.45 m, Millex, Millipore, Products were purified (Qiaquick or PCR-purification kits, Qiagen), Basel, Switzerland) and used to transduce NM18 cells in the presence of μ μ phosphorylated (T4 kinase, NEB, Beverly, MA, USA) and ligated into 1.5 g/ml polybrene (Sigma). Infected cells were selected with 5 g/ml pMSCVpuro (T4 ligase, NEB). For generating pMSCV-Zeb1, a fragment puromycin (Sigma). containing the coding sequence of Zeb1 was released from pCR-BluntII- TOPO (BioCat, Heidelberg, Germany) by EcoRI-digest and ligated into Transfection of small-interfering RNAs (siRNAs), transient pMSCVpuro. fi overexpression and establishment of stable E-cadherin Sequence integrity of inserts was veri ed by restriction digest and sequencing (GATC, Konstanz, Germany). knockdown cells Following siRNAs (Qiagen, Hilden, Germany) were used—Bnc1 (SI00930160), Lmo7 (SI01091034), Zeb1 (SI01476293), FoxA2 Hanging-drop assays (SI01004521), c-Jun (SI01079862), Smad3 (SI00210266), FoxQ1 After transfection and TGF-β1 induction, cells were harvested using Citric (SI01005375) and non-silencing siRNA (ns-siRNA, 1027310). Cells were Saline (135 mM KCl, 15 mM NaCl) and seeded as single cells in hanging reverse transfected using HiPerFect (Qiagen) with a final concentration of drops in regular growth media (30 000 cells/20 μl) underneath a lid of a 10 12.5 nM siRNA for Bnc1, Zeb1; 20 nM for FoxA2, c-Jun, Smad3, 8.5 nM for cm dish. The dish was filled with 10 ml PBS to limit evaporation during the Lmo7 and 25 nM for FoxQ1. Cells were transfected for 24 h before 2-hour incubation. stimulation with TGF-β1. Cell clusters were mechanically challenged by pipetting each drop 20 For transient overexpression of Smad3, cells were transfected using times through yellow pipette tips and cells counted with a hemocyt- Lipofectamine 2000. Smad3-encoding plasmids have previously been ometer. Four drops were counted per condition and experiments were at described,51 pIRESpuro3 (Clontech, Heidelberg, Germany) was used as least repeated twice. Aggregation index was determined as previously control. described.53 Cell aggregates consisting of >4 cells were considered true For generating stable E-cadherin knockdown cells, an E-cadherin shRNA- aggregates and counted as one particle whereas aggregates with ⩽ 4 cells encoding cassette,52 was subcloned into pRISC-Stuffer. A non-silencing were considered random cell associations and cells were counted as shRNA cassette (sequence corresponds to ns-siRNA) ligated into pRISC- individuals. Significance was calculated using Student’s t-test (two groups) Stuffer served as negative control. Sequences of shRNA cassettes were or one-way ANOVA followed by post hoc Tukey’s test (more than two confirmed by sequencing. groups).

Chromatin-immunoprecipitation Western blot Chromatin-immunoprecipitation experiments were performed using the Lysates were prepared using RIPA-buffer, centrifuged at ~ 20 000 g for Chromatin-immunoprecipitation-Assay kit (Millipore) according to the 20 minutes (4 °C) and protein concentrations of the supernatants were supplier’s instructions. determined using Bradford reagent (Sigma). Proteins were separated on DNA was sheared (100–1500 bp) at 5 °C (Covaris S2 sonicator: 30 cycles; SDS-polyacrylamide-gels and transferred onto Immobilon PVDF mem- duty factor 20%; intensity 5; cycles/burst 200) in 500 μl ice-cold phosphate branes (Millipore). Membranes were blocked in 5% (w/v) non-fat dry milk

© 2015 Macmillan Publishers Limited Oncogene (2015) 1185 – 1195 Bnc1 influences outcome of TGF-β1-signaling A Feuerborn et al 1194 or bovine serum albumin in PBS-Tween20 (0.1% v/v) or TBS-Tween (0.1% v/ Activity changes of transcription factors were predicted using Motif v) for 1 h at RT and incubated with primary antibodies against E-cadherin Activity Response Analysis (MARA)62 using raw Cel-files. Motif activities are (BD Transduction Laboratories, clone 36, 1:10 000), Gapdh (sc-25778, depicted as decreased if the transcription factor functions as repressor (for 1:1000), Actin (sc-1616-R, 1:1000) Smad4 (sc-7966, 1:1000) (Santa Cruz), example, Zeb1). c-Jun and activated phospho-c-Jun (serine 63 and serine 73) (#9165, #9261 and #9164, all 1:1000), FoxA2 (#3143, 1:1000), Smad3 (#9513, 1:1000) and Zeb1 (#3396, 1:1250) (all Cell Signaling Technologies) and Bnc1 (Aviva Statistics Systems Biology, San Diego, CA, USA), ARP33283_P050, 1:500. Details are mentioned in respective Materials and methods section. Error Membranes were incubated with HRP-conjugated secondary antibodies bars represent ± s.e.m.; in case of quantitative RT–PCR data and depending (Santa Cruz, sc-2004, 1:5000 or sc-2005, 1:2500) and visualized using the on the experimental setup, the upper and lower limit of expression ECL detection system (GE Healthcare Bio-Sciences, Piscataway, NJ, USA). according to Livak et al.54 Signals were captured onto Super RX films (Fuji). A P-value of Po0.05 was considered significant.

RNA isolation and qRT-PCR CONFLICT OF INTEREST RNA was isolated using Trizol Reagent (Life Technologies, Invitrogen) or The authors declare no conflict of interest. RNeasy Mini Kit (Qiagen) following the manual guidelines and treated with TURBO DNAse (Life Technologies, Invitrogen). cDNA was generated using SuperScript II (Life Technologies, Invitrogen) using Oligo(dT)-primer and ACKNOWLEDGEMENTS diluted 10-fold in RNAse-free water (Qiagen) before quantitative RT-PCR. Experiments contained no RT-controls. We kindly acknowledge Peter ten Dijke, David Danielpour, Sebastian Diecke and For quantitative RT–PCR, LightCycler Fast Start DNA SYBR Green I system Roderick Beijersbergen for generously providing cells and plasmids. Support by the (Roche, Basel, Switzerland) was used according to the supplier’s DKFZ Light Microscopy Facility is gratefully acknowledged. We thank Maria Muciek recommendations. Primer sequences and annealing temperatures for excellent assistance with regard to microarray expression profiling and Ann Na are available upon request. RT–PCR data was analyzed according to the Tan for excellent technical support. This study was supported by a grant of the DFG 2−(ΔΔC(T)) method.54 (SFB-938) to HJG. Significance was calculated comparing normalized C(T) values using unpaired or paired Student’s t-test, depending on the experimental setting. For the analyses of more than two groups one-way analysis of REFERENCES ’ variance (ANOVA) followed by post hoc Tukey s tests were performed. 1 Kalluri R, Weinberg RA. The basics of epithelial-mesenchymal transition. J Clin Invest 2009; 119: 1420–1428. Gene expression profiling 2 Drabsch Y, Ten Dijke P. TGF-beta signalling and its role in cancer progression and metastasis. Cancer Metastasis Rev 2012; 31:553–568. fi Gene expression pro ling was performed using GeneChip MOE430A arrays 3 Heldin CH, Vanlandewijck M, Moustakas A. Regulation of EMT by TGFbeta (Affymetrix, Santa Clara, CA, USA). RNAs were isolated using the RNeasy in cancer. FEBS Lett 2012; 586: 1959–1970. Mini Kit (Qiagen), including on-column DNAse digest. RNA quantity and 4 Peinado H, Olmeda D, Cano A. Snail, Zeb and bHLH factors in tumour progression: quality was checked using Agilent 2100 Bioanalyzer (Agilent Technologies, an alliance against the epithelial phenotype? Nat Rev Cancer 2007; 7: 415–428. Santa Clara, CA, USA). 5 Feuerborn A, Kuffer S, Grone HJ. Forkhead factors regulate epithelial plasticity: ′ RNA was processed using the GeneAtlas 3 IVT Express Kit (Affymetrix) impact on cancer progression. Cell Cycle 2011; 10: 2454–2460. ’ fl according to manufacturer s recommendations. Brie y, 100 ng total RNA 6 Mullen AC, Orlando DA, Newman JJ, Lovén J, Kumar RM, Bilodeau S et al. Master were reverse transcribed using T7-Oligo (dT) primer and converted into transcription factors determine cell-type-specific responses to TGF-beta signaling. double-stranded cDNA. Double-stranded cDNA was used to generate Cell 2011; 147: 565–576. biotin-modified aRNA which was purified, fragmented and hybridized on 7 Morikawa M, Koinuma D, Miyazono K, Heldin CH. Genome-wide mechanisms of the mentioned arrays. Smad binding. Oncogene 2013; 32: 1609–1615. 24,55 NIAC-NTR-based gene expression profiling is detailed elsewhere. 8 Moustakas A, Heldin CH. Non-Smad TGF-beta signals. J Cell Sci 2005; 118: NIAC-NTR relies on 4 sU (4-thio-uridine)-incorporation into nascent RNA 3573–3584. and provides analyses of differentially expressed genes of total RNA and 9 Deckers M, van Dinther M, Buijs J, Que I, Löwik C, van der Pluijm G et al. The tumor insights into transcriptionally regulated genes within respective time suppressor Smad4 is required for transforming growth factor beta-induced epi- 55 frames. thelial to mesenchymal transition and bone metastasis of breast cancer cells. Here, we complement our previous data-release of NMuMG cells treated Cancer Res 2006; 66: 2202–2209. 24 with TGF-β1 and 4 sU for 2 h with gene expression data of NMuMG cells 10 Vanhoutteghem A, Bouche C, Maciejewski-Duval A, Herve F, Djian P. Basonuclins β treated with TGF- 1 for 24 h and incorporation of 4 sU for the last 2 h of and disco: Orthologous zinc finger proteins essential for development in verte- β TGF- 1-treatment. Gene expression data are available at Gene Expression brates and arthropods. Biochimie 2011; 93: 127–133. Omnibus database (GEO) via accession number GSE54717. 11 Tseng H. Basonuclin, a zinc finger protein associated with epithelial expansion and proliferation. Front Biosci 1998; 3: D985–D988. 12 Mahoney MG, Tang W, Xiang MM, Moss SB, Gerton GL, Stanley JR et al. Trans- Bioinformatic analyses, prediction of transcription factor binding location of the zinc finger protein basonuclin from the mouse germ cell nucleus sites, motif activity response analysis (MARA) to the midpiece of the spermatozoon during spermiogenesis. Biol Reprod 1998; Affymetrix Cel-files were imported into R Statistical package version 2.14 59:388–394. using package R Affy (version 1.34.0).56 13 Tseng H, Matsuzaki K, Lavker RM. Basonuclin in murine corneal and lens epithelia Probes were annotated using custom Cel-Definition-File (version 14) for correlates with cellular maturation and proliferative ability. Differentiation 1999; ENTREZ genes.57 Signal intensities were normalized using VSN 65:221–227. normalization.58 14 Zhang X, Chou W, Haig-Ladewig L, Zeng W, Cao W, Gerton G et al. BNC1 is Differentially expressed genes were identified using Significance required for maintaining mouse spermatogenesis. Genesis 2012; 50:517–524. Analysis of Microarrays (SAM) via R package SAMR59 (FDR o5%). 15 Zhang X, Tseng H. Basonuclin-null mutation impairs homeostasis and wound Genes with fold-change of ⩾ ± 1.30 were considered for further analyses. repair in mouse corneal epithelium. PLoS One 2007; 2: e1087. We extracted differentially expressed genes annotated within the GO- 16 Zhang S, Wang J, Tseng H. Basonuclin regulates a subset of ribosomal RNA genes term ‘Transcription factor activity’ or ‘Adherens Junctions’ using Cytoscape in HaCaT cells. PLoS One 2007; 2: e902. and BINGO.60,61 17 Ma J, Zeng F, Schultz RM, Tseng H. Basonuclin: a novel mammalian maternal- Transcription factor binding sites and overrepresentation of binding effect gene. Development 2006; 133: 2053–2062. motifs were predicted using PROMO,25 TFbind26 and oPOSSUM27 using 18 Wang J, Zhang S, Schultz RM, Tseng H. Search for basonuclin target genes. Bio- default settings, with the exception of PROMO where ‘Maximum matrix chem Biophys Res Commun 2006; 348: 1261–1271. dissimilarity rate’ was set to 5 (default 15), thereby increasing stringency. 19 Shames DS, Girard L, Gao B, Sato M, Lewis CM, Shivapurkar N et al. A genome- Sequences of genomic regions of Bnc1 were retrieved using UCSC wide screen for promoter methylation in lung cancer identifies novel methylation (assembly GRCm38/mm10). markers for multiple malignancies. PLoS Med 2006; 3: e486.

Oncogene (2015) 1185 – 1195 © 2015 Macmillan Publishers Limited Bnc1 influences outcome of TGF-β1-signaling A Feuerborn et al 1195 20 Morris MR, Ricketts C, Gentle D, Abdulrahman M, Clarke N, Brown M et al. Iden- reveals roles of ETS1 and TFAP2A in transforming growth factor beta signaling. tification of candidate tumour suppressor genes frequently methylated in renal Mol Cell Biol 2009; 29:172–186. cell carcinoma. Oncogene 2010; 29:2104–2117. 41 Garber M, Yosef N, Goren A, Raychowdhury R, Thielke A, Guttman M et al. A high- 21 Brena RM, Plass C, Costello JF. Mining methylation for early detection of common throughput chromatin immunoprecipitation approach reveals principles of cancers. PLoS Med 2006; 3: e479. dynamic gene regulation in mammals. Mol Cell 2012; 47:810–822. 22 Guo L, Fan D, Zhang F, Price JE, Lee JS, Marchetti D et al. Selection of brain 42 Dekker J, Rippe K, Dekker M, Kleckner N. Capturing chromosome conformation. metastasis-initiating breast cancer cells determined by growth on hard agar. Am J Science 2002; 295: 1306–1311. Pathol 2011; 178: 2357–2366. 43 Papantonis A, Kohro T, Baboo S, Larkin JD, Deng B, Short P et al. TNFalpha signals 23 Boldrup L, Coates PJ, Laurell G, Nylander K. p63 Transcriptionally regulates BNC1, through specialized factories where responsive coding and miRNA genes are a Pol I and Pol II transcription factor that regulates ribosomal biogenesis and transcribed. EMBO J 2012; 31: 4404–4414. epithelial differentiation. Eur J Cancer 2012; 48: 1401–1406. 44 Nakahata S, Yamazaki S, Nakauchi H, Morishita K. Downregulation of 24 Feuerborn A, Srivastava PK, Kuffer S, Grandy WA, Sijmonsma TP, Gretz N et al. The ZEB1 and overexpression of Smad7 contribute to resistance to TGF-beta1- Forkhead factor FoxQ1 influences epithelial differentiation. J Cell Physiol 2011; mediated growth suppression in adult T-cell leukemia/lymphoma. Oncogene 226: 710–719. 2010; 29:4157–4169. 25 Messeguer X, Escudero R, Farre D, Nunez O, Martinez J, Alba MM. PROMO: 45 Sundqvist A, Zieba A, Vasilaki E, Herrera Hidalgo C, Söderberg O, Koinuma D et al. detection of known transcription regulatory elements using species-tailored Specific interactions between Smad proteins and AP-1 components determine searches. Bioinformatics 2002; 18: 333–334. TGFbeta-induced breast cancer cell invasion. Oncogene 2013; 32: 3606–3615. 26 Tsunoda T, Takagi T. Estimating transcription factor bindability on DNA. Bioin- 46 Qiao Y, Jiang X, Lee ST, Karuturi RK, Hooi SC, Yu Q. FOXQ1 regulates epithelial- formatics 1999; 15:622–630. mesenchymal transition in human cancers. Cancer Res 2011; 71: 3076–3086. 27 Kwon AT, Arenillas DJ, Worsley Hunt R, Wasserman WW. oPOSSUM-3: advanced 47 Kumper S, Ridley AJ. p120ctn and P-cadherin but not E-cadherin regulate cell analysis of regulatory motif over-representation across genes or ChIP-Seq data- motility and invasion of DU145 prostate cancer cells. PLoS One 2010; 5: e11801. sets. G3 (Bethesda) 2012; 2: 987–1002. 48 Capaldo CT, Macara IG. Depletion of E-cadherin disrupts establishment but not 28 Shirakihara T, Saitoh M, Miyazono K. Differential regulation of epithelial and maintenance of cell junctions in Madin-Darby canine kidney epithelial cells. Mol mesenchymal markers by deltaEF1 proteins in epithelial mesenchymal transition Biol Cell 2007; 18:189–200. induced by TGF-beta. Mol Biol Cell 2007; 18: 3533–3544. 49 Nieman MT, Prudoff RS, Johnson KR, Wheelock MJ. N-cadherin promotes motility 29 Xie L, Law BK, Aakre ME, Shyr Y, Bhowmick NA, Moses HL et al. Transforming in human breast cancer cells regardless of their E-cadherin expression. J Cell Biol growth factor beta-regulated gene expression in a mouse mammary gland epi- 1999; 147: 631–644. thelial cell line. Breast Cancer Res 2003; 5: R187–R198. 50 Vasioukhin V, Fuchs E. Actin dynamics and cell-cell adhesion in epithelia. Curr 30 Tang Y, Shu G, Yuan X, Jing N, Song J. FOXA2 functions as a suppressor of tumor Opin Cell Biol 2001; 13:76–84. metastasis by inhibition of epithelial-to-mesenchymal transition in human lung 51 Chipuk JE, Cornelius SC, Pultz NJ, Jorgensen JS, Bonham MJ, Kim SJ et al. The cancers. Cell Res 2011; 21: 316–326. androgen receptor represses transforming growth factor-beta signaling through 31 Aigner K, Dampier B, Descovich L, Mikula M, Sultan A, Schreiber M et al. interaction with Smad3. J Biol Chem 2002; 277:1240–1248. The transcription factor ZEB1 (deltaEF1) promotes tumour cell dedifferentiation 52 Redmer T, Diecke S, Grigoryan T, Quiroga-Negreira A, Birchmeier W, Besser D. E- by repressing master regulators of epithelial polarity. Oncogene 2007; 26: cadherin is crucial for embryonic stem cell pluripotency and can replace OCT4 6979–6988. during somatic cell reprogramming. EMBO Rep 2011; 12:720–726. 32 Aigner K, Descovich L, Mikula M, Sultan A, Dampier B, Bonné S et al. The tran- 53 Nagafuchi A, Takeichi M. Cell binding function of E-cadherin is regulated by the scription factor ZEB1 (deltaEF1) represses Plakophilin 3 during human cancer cytoplasmic domain. EMBO J 1988; 7: 3679–3684. progression. FEBS Lett 2007; 581: 1617–1624. 54 Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time 33 Giampieri S, Manning C, Hooper S, Jones L, Hill CS, Sahai E. Localized and quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 2001; 25:402–408. reversible TGFbeta signalling switches breast cancer cells from cohesive to single 55 Kenzelmann M, Maertens S, Hergenhahn M, Kueffer S, Hotz-Wagenblatt A, Li L cell motility. Nat Cell Biol 2009; 11:1287–1296. et al. Microarray analysis of newly synthesized RNA in cells and animals. Proc Natl 34 Gervasi M, Bianchi-Smiraglia A, Cummings M, Zheng Q, Wang D, Liu S et al. JunB Acad Sci USA 2007; 104: 6164–6169. contributes to Id2 repression and the epithelial-mesenchymal transition in 56 Gautier L, Cope L, Bolstad BM, Irizarry RA. Affy--analysis of Affymetrix GeneChip response to transforming growth factor-beta. J Cell Biol 2012; 196:589–603. data at the probe level. Bioinformatics 2004; 20: 307–315. 35 Zhang H, Meng F, Liu G, Zhang B, Zhu J, Wu F et al. Forkhead transcription factor 57 Dai M, Wang P, Boyd AD, Kostov G, Athey B, Jones EG et al. Evolving gene/ foxq1 promotes epithelial-mesenchymal transition and breast cancer metastasis. transcript definitions significantly alter the interpretation of GeneChip data. Cancer Res 2011; 71: 1292–1301. Nucleic Acids Res 2005; 33: e175. 36 Loerke D, le Duc Q, Blonk I, Kerstens A, Spanjaard E, Machacek M et al. Quanti- 58 Huber W, von Heydebreck A, Sultmann H, Poustka A, Vingron M. Variance sta- tative imaging of epithelial cell scattering identifies specific inhibitors of cell bilization applied to microarray data calibration and to the quantification of dif- motility and cell-cell dissociation. Sci Signal 2012; 5: rs5. ferential expression. Bioinformatics 2002; 18:S96–S104. 37 de Rooij J, Kerstens A, Danuser G, Schwartz MA, Waterman-Storer CM. Integrin- 59 Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the dependent actomyosin contraction regulates epithelial cell scattering. J Cell Biol ionizing radiation response. Proc Natl Acad Sci USA 2001; 98:5116–5121. 2005; 171: 153–164. 60 Maere S, Heymans K, Kuiper M. BiNGO: a Cytoscape plugin to assess over- 38 Cui C, Elsam T, Tian Q, Seykora JT, Grachtchouk M, Dlugosz A et al. Gli proteins up- representation of gene ontology categories in biological networks. Bioinformatics regulate the expression of basonuclin in Basal cell carcinoma. Cancer Res 2004; 64: 2005; 21:3448–3449. 5651–5658. 61 Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D et al. Cytoscape: a 39 Koinuma D, Tsutsumi S, Kamimura N, Imamura T, Aburatani H, Miyazono K. software environment for integrated models of biomolecular interaction Promoter-wide analysis of Smad4 binding sites in human epithelial cells. Cancer networks. Genome Res 2003; 13: 2498–2504. Sci 2009; 100: 2133–2142. 62 Suzuki H, Forrest AR, van Nimwegen E, Daub CO, Balwierz PJ, Irvine KM et al. 40 Koinuma D, Tsutsumi S, Kamimura N, Taniguchi H, Miyazawa K, Sunamura M et al. The transcriptional network that controls growth arrest and differentiation in a Chromatin immunoprecipitation on microarray analysis of Smad2/3 binding sites human myeloid leukemia cell line. Nat Genet 2009; 41: 553–562.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)