OTX1 Expression in Breast Cancer Is Regulated by P53

A Terrinoni1, IS Pagani2, I Zucchi3, AM Chiaravalli4, V Serra1, F Rovera5, S Sirchia6, G Dionigi7, M Miozzo6, A Frattini3, A Ferrari8, C Capella9, F Pasquali2, FL Curto2, A Albertini3, G Melino1,10 and G Porta2

1Biochemistry Laboratory, IDI-IRCCS, c/o Department of Experimental Medicine and Biochemical Sciences, University of Rome ‘Tor Vergata’, Rome, Italy; 2Department of Experimental and Clinical Biomedical Sciences, Universita` dell’Insubria, Varese, Italy; 3Institute of Biomedical Technologies, National Research Council, Milan, Italy; 4Department of Pathology, Ospedale di Circolo, Varese, Italy; 5Department of Surgical Sciences, University of Insubria, Varese, Italy; 6Department of Medicine, Surgery and Dentistry, Medical Genetics Unit, Universita‘ degli Studi di Milano, Milano, Italy; 7Endocrine Surgery Research Center, Department of Surgical Sciences, University of Insubria, Varese, Italy; 8Fondazione I.R.C.C.S. Policlinico San Matteo, Pavia, Italy; 9Department of Human Morphology, Anatomic Pathology Unit, Centro Insubre di Biotecnologie per la Salute Umana, Ospedale di Circolo, University of Insubria, Varese, Italy and 10Medical Research Council, Toxicology Unit, Hodgkin Building, Leicester University, Leicester, UK

The p53 transcription factor has a critical role in cell At puberty ovarian hormones stimulate ductal elongation stress response and in tumor suppression. Wild-type p53 and the cells involved differentiate into lumen epithelial protein is a growth modulator and its inactivation is a cells and myoepithelial cells. The ducts remain quiescent critical event in malignant transformation. It has been until pregnancy. How this control is regulated is not known, recently demonstrated that wild-type p53 has develop- but probably involves the orchestration of hormones, mental and differentiation functions. Indeed an over- morphogens and growth factors (Lewis, 2000; Watabe expression of p53 in tumor cells induces asymmetrical and Miyazono, 2009). After its development, the mammary division avoiding self-renewal of cancer stem cells (CSCs) gland passes from linear to cyclical phases. A series of and instead promoting their differentiation. In this study, complex transitions lead to lobulo-alveolar differentia- 28 human breast carcinomas have been analyzed for expres- tion, lactation, involution and gland remodeling. Genetic sion of wild-type p53 and of a pool of non-clustered homeobox mutations, mainly in signal transduction genes, have been genes. We demonstrated that orthodenticle homolog 1 identified that block or delay these processes. gene (OTX1) is transcribed in breast cancer. We established Recently, new data demonstrate that p53 is implicated that the p53 protein directly induces OTX1 expression by in normal cell function, differentiation and development acting on its promoter. OTX1 has been described as a (Levantini et al., 2003; Lin et al., 2005; Riley et al., critical molecule for axon refinement in the developing 2008). It has a function in the control of asymmetric cerebral cortex of mice, and its activity in breast cancer cell division critical to generate alternative cell fates suggests a synergistic function with p53 in CSC differentia- (Smalley and Ashworth, 2003; Cicalese et al., 2009). p53 tion. Wild-type p53 may regulate cellular differentiation regulates polarity of cell division in mammary stem cells by an alternative pathway controlling OTX1 signaling only (SCs). The loss of p53 favors symmetric divisions of in breast cancer cells and not in physiological conditions. cancer stem cells (CSCs), contributing to tumor growth. Oncogene (2011) 30, 3096–3103; doi:10.1038/onc.2011.31; In mammals, homeobox genes have critical roles in a published online 11 April 2011 variety of developmental processes in the nervous and skeletal systems, in limb and digit formation, and in Keywords: p53; OTX1; human breast cancer; cancer organogenesis (Eizenberg et al., 1996; Lewis, 2000; stem cells Danilova et al., 2008). One difficulty in understanding the function of individual members of the homeobox family is that little is known about upstream genes that control their expression (Lewis, 2000). Furthermore, Introduction homeotic genes have received attention with respect to mammary gland organogenesis, functional differentiation The mammary gland, unlike most organs that form in a and cancer (Ford, 1998; Chen and Capecchi, 1999; Cillo linear way during embryogenesis, develops after puberty. et al., 1999; Lewis, 2000; Abate-Shen, 2002; Argiropoulos and Humphries, 2007). For these reasons, the expression of a pool of non-clustered homeobox genes belonging to Correspondence: Professor G Porta, Department of Experimental and the muscle segment homeobox (MSX), empty spiracles Clinical Biomedical Sciences, Universita` dell’Insubria, via Dunant 5, homeobox (EMX) and orthodenticle homeobox (OTX) Varese 21100, Italy. E-mail: [email protected] families was investigated in human breast cancer. Received 5 May 2010; revised 13 November 2010; accepted 11 January We showed a statistically significant level of 2011; published online 11 April 2011 co-expression of p53 and orthodenticle homolog 1 gene OTX1 expression in breast cancer A Terrinoni et al 3097 (OTX1) in all samples (r ¼ 0.864 significantly different Table 1 Tumor grade, p53 mutations and polymorphisms of patient from zero). We demonstrated with in vivo and in vitro samples experiments the direct regulation of OTX1 by p53. Samples Histological Tumor p53 p53 polymorphisms The role of p53 in directing SCs into symmetrical or type grade mutations asymmetrical division could involve the activation of the differentiation function of OTX1. 3 IDC G2 R72 4 IDC G2 R72P We detected the increase of OTX1 and p53 gene c.305G4A p.P36 in expression levels in rodent CSCs, when induced to heterozygous state differentiate. 5 IDC G2 R72P c.836A4G p.R213 7 IDC G3 c.775A4G R72P p.193H4R Results 8 IDC G2 P72 c.305G4A p.P36 in A pool of 33 ductal and lobular breast cancers collected heterozygous state by the pathology department of Varese Hospital 11 IDC G2 R72 c.836A4G p.R213 in (Varese, Italy) were sequenced to evaluate the presence heterozygous state of mutations in p53 (see primer sequences in Supple- 12 IDC G3 c.910A4T P72 mentary information 1). We found mutations in the p.238C4F p53 DNA-binding domain in three breast cancer samples 16 IDC G3 R72P 18 ILC G2 R72 with a frequency of 9.1 versus 20% reported in literature 20 IDC ND c.781T4C R72 (Borresen-Dale, 2003). The mutations found to be hetero- p.195I4T zygous were p.193H4R, p.195I4T and p.238C4F. 23 IDC G2 P72 We also found 17 out of 33 (51.5%) tumors to be 26 IDC G2 R72 homozygous for the polymorphism c.412G, which codes 27 IDC G3 R72P 29 IDC G3 R72 for the amino acid R72, 13 out of 33 (39.4%) to be P72R 31 ILC G3 R72 (c.412C4G) and 3 out of 33 (9.1%) to be homozygous 33 IDC G2 R72 for P72 (c.412C). Further analysis in patients’ cDNA 35 IDC G2 R72P sequences demonstrated that sample 8, homozygous for 37 IDC G3 R72 41 IDC G2 R72P P72, and sample 11, homozygous for R72, are respec- 43 IDC G3 R72 tively heterozygous for polymorphisms c.305G4A and 44 IDC G2 R72 c.836A4G adjacent to 72 on the same stretch of DNA 47 IDC G2 R72P sequence (Carbone et al., 1991; Papadakis et al., 2000; 49 IDC G2 R72 Whibley et al., 2009). P72 or R72 hemizygosity cannot 50 IDC G2 R72P 51 IDC G3 R72 be excluded in the other patients, as DNA samples are 53 IDC G2 R72P not available (Table 1). 54 IDC G2 R72 Quantitative real-time reverse transcriptase PCR 55 IDC G2 R72 (qRT–PCR) was used to evaluate the levels of gene 56 IDC G2 R72P 57 IDC G2 R72 transcription. A total of 28 cases of breast carcinoma 58 ILC G2 R72P were analyzed in parallel with surrounding normal tissue 62 IDC G2 R72 (Figure 1a). We excluded from our study three samples, 63 IDC G2 R72P because they were mutated for p53. Alterations in the expression of OTX1 (drosophila) in breast cancer were Abbreviations: G1, low-grade; G2, moderate or intermediate grade; demonstrated. Tumors showed a higher level of OTX1 G3, high-grade; IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma; ND, not determined. expression in comparison with the normal breast tissues, The numeration is referred to the cDNA p53 sequence accession in which we found a basal level of expression. We also number NM_000546. Mutation analysis of the p53 gene was observed a statistically significant co-expression of performed using direct DNA sequencing of PCR-amplified cDNA. OTX1 and p53 in tumor samples (Figure 1a), but not The coding region was amplified using the primer pairs indicated in the in normal breast tissues (data not shown). In tumor supplementary data section (Supplementary Table 1). tissues, OTX1 and p53 gene expression have a correlation coefficient of r ¼ 0.864 significantly different from zero (t ¼ 8.412), with a probability of P ¼ 1.284EÀ08 (**Po0.001) (Figure 1b). The highest and the lowest In order to demonstrate the molecular interaction values were excluded for statistical reasons. between p53 and OTX1 gene expression, we used an OTX1 expression was confirmed by immunohisto- SaOs2 cell line (p53 Tet-On) (Riley et al., 2008) that does chemical analysis. Using the OTX2/OTX1 antibody we not express endogenous p53. The p53 expression in this showed positivity both for the nuclear (Figure 1c) and/ cell line was induced by adding doxycyline to the culture or the cytoplasmic (Figure 1d) localization. As OTX2 medium. The expression of p53 at 12, 24 and 48 h (and mRNA has not been detected in any tumors we may its absence in control cells), was confirmed by western consider that the immunoreactivity was due to the blot using an anti-p53 antibody (Figure 2a). To evaluate presence of OTX1 protein. In p53-mutated cases, no OTX1 expression levels, we used qRT–PCR and western OTX1 positivity was detected (data not shown). blot analysis. Both techniques confirmed the expression

Oncogene OTX1 expression in breast cancer A Terrinoni et al 3098

Figure 1 Expression of orthodenticle homolog 1 gene (OTX1) and p53 in human breast cancer. (a) Gene expression levels of OTX1 and p53 were detected by quantitative real-time reverse transcriptase PCR (qRT–PCR) in ductal and lobular breast cancer. In the x-axis are the patient samples and the y-axis the relative gene expression. Tissues from breast carcinoma over two centimeters in diameter were obtained after informed consent from the patients for molecular and cellular analysis. The study was approved by the relevant local ethical committees. For each case, tumor and, when possible, non-neoplastic tissues, were immediately and carefully sampled. Both tissue blocks were divided in two mirror parts, one immediately preserved in RNA later until RNA extraction, the other fixed in formalin and embedded in paraffin to evaluate the suitability of the materials. The RNA was extracted with RNeasy Fibrous Tissue mini Kit from Qiagen (Valencia, CA, USA). Equal amounts (1 mg) of total RNA from each sample were reverse transcribed using random primers (hexamers), according to the protocol of High Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA, USA). qRT–PCR was performed with TaqMan technology in an ABI Prism 7000 apparatus (Applied Biosystems). Human p53 and human OTX1 gene expression were studied using Applied Biosystems assays. Beta-actin was used as a housekeeping gene. A relative quantitative analysis was performed using the 2ÀDDCt method. In adjacent normal tissues, there is no correlation between the expression levels of OTX1 and p53 (data not shown), (b) whereas in tumor tissues, OTX1 and p53 gene expression have a correlation coefficient of r ¼ 0.864, significantly different from zero (t ¼ 8.412), with a probability of P ¼ 1.284EÀ08 (Po0.001**). (c) Formalin- fixed, paraffin-embedded section of breast carcinoma showing nuclear localization (brown nuclei) of OTX1 protein. The immunohistochemical analysis was performed with a biotin-free detection system (Ultravision detection system kit, Thermo scientific, Fremont, CA, USA), according to the manufacturer’s instructions. Rabbit anti-OTX2 polyclonal antibody (Chemicon International, Temecula, CA, USA, cat. #AB9566) was applied overnight at a dilution of 1/2000. The antibody reacts also with OTX1 due to amino acid sequence homology between the two proteins. As OTX2 mRNA has not been detected in any tumors we can consider that the immunoreactivity was due to the presence of OTX1 protein. Specificity controls consisted of: (i) use of control tissues with or without the pertinent antigen, (ii) omission of the primary antibody and substitution with non-immune serum at the same dilution (DAB-Haematoxylin, scale bar 40 mm). (d) Formalin (4%)-fixed, paraffin-embedded sections of two cases of human ductal breast carcinomas showing nuclear (C1, brown nuclei) or cytoplasmic (C2) localization of OTX1 protein in tumor epithelial cells. A full colour version of this figure is available at the Oncogene journal online.

of the OTX1 gene at the transcriptional (Figure 2b) and approximately 3.6 Kb upstream of OTX1 transcription at the protein level (Figure 2a), after the induction of starting site, up to 1.2 Kb downstream to the transcrip- p53 ( þ doxycycline). The expression of OTX1 tran- tion starting site. The OTX1 transcript contains five scripts clearly increased after 12 and 24 h, after the exons: the first two represent the 50-UTR, whereas the addition of the antibiotic (Figure 2b), according to the ATG is located at the beginning of the third exon. expression of p53. The western blot analysis demon- It is known that regulatory elements could lie inside strated the presence of the OTX1 protein also at 48 h, introns and exons of UTR mRNA (Gressner et al., after p53 induction (Figure 2a). 2005); indeed the analysis was extended to include We performed an ‘in silico’ analysis of an approxi- the first two exons/introns. Bioinformatic analysis of mate 4.8 Kb CpG island in the 50 region of OTX1 gene, the region was performed using the MatInspector looking for p53 regulatory sequences of the OTX1 program by Genomatix (Genomatix Software GmbH, promoter (Figure 2c). The CpG island extends from freeware version), looking for transcription factors with

Oncogene OTX1 expression in breast cancer A Terrinoni et al 3099 core and matrix similarity above 0.8. The ‘in silico’ In order to understand if one of these two putative analysis showed the presence of two putative p53 RE sites represented a functional p53-binding site, responsive elements (RE). The ﬁrst is located at a chromatin immunoprecipitation analysis was carried À3435 bp from the ATG (p53RE-50), with matrix out (Figure 2d). SaOs2 p53 Tet-On cells, treated similarity of 0.942 and core similarity of 0.921. The and untreated with doxycyclin at 12 and 24 h were second is located at þ 1268 bp from the ATG into the immunoprecipitated using an anti-p53 antibody and second intron, with matrix similarity of 0.898 and a non-speciﬁc IgG as control. Two regions approximately core similarity of 1.0. of 160–170 bp surrounding the two RE sites were

SaOs Tet SaOs p53 OTX1 qPCR 5 Dox -+- ++ 4.5 4 Ab:p53 3.5 3 Ab:Actin 2.5 2 1.5

12 h24 h 48 h Fold over control 1 Ab:Otx1 0.5 0 Doxoh Dox12h Dox24h

OTX1 Gene structure

OTX1 promoter CpG_island : 5‘ UTR Ex3 Ex4 Ex5 mRNA PUTATIVE PROMOTER Ex1 Ex2 ATG 3‘ UTR CDS

2K 34684K 4899 6K 8K 10K

5’ p53 RE 3’ p53 RE

ChIP analysis

IP Input IP Input Hours 0 12 24 0122412 24 Hours 012240

500 300 IP: IP: 200 200 p53 p53 100 100

Hours 0 12 24 0 12 24 Hours 0122412 24 0

500 IP: 300 IP: 200 200 IgG IgG 100 100

Luciferase Assay

p53-RE 3’ 0.5

5‘ UTR Ex3 0.4 Ex1 Ex2 ATG 0.3

3468 4K 4899 0.2 pGL3-Basic f1 ori KpnI Vector 5 Fold Over Control XbaI 1742 SacI 11 0.1 MluI 15 NheI 21 28 XhoI 32 0.0 BgIII 36 Ctrl HindIII 53 p53 luciferase + NarI 121NcoI 86

Figure 2 For caption please see page 3100.

Oncogene OTX1 expression in breast cancer A Terrinoni et al 3100 amplified by PCR (see primers sequences in Supplemen- differentiate with DMSO (LA7D) (Zucchi et al., 2008), tary information 2). and we detected increased levels both of p53 and OTX1 No amplification was obtained for the p53RE-50,even in LA7D, implying a role of these genes in the using exhaustive PCR amplification of 40 cycles, asymmetric division of CSCs (Figure 3a). LA7D showed indicating that there was no binding of p53 on this site. increased immunohistochemical OTX1 positivity in A positive signal was obtained for the p53RE-30,at24h comparison with non-induced LA7 (data not shown). of p53 induction. No signal was obtained from the We also demonstrated changes in OTX1 gene expres- immunoprecipitation with IgG, indicating the specificity sion levels in mice during mammary gland development of the recognition by p53 antibody. These results clearly (Figure 3b). RNAs from breast tissue of virgin, pregnant demonstrated the physical interaction between p53 and and lactating mice were retro-transcribed and the the OTX1 promoter via 30p53RE. It is interesting that cDNAs were analyzed with qRT–PCR. OTX1 presents the putative functional p53 RE is located after the higher expression levels during lactation, when breast transcription-starting site, internal to a transcribed differentiates into ducts and alveoli. region. To investigate the possible role of p53 modulation of To clarify if the 30p53RE is able to drive the expres- OTX1 during the physiological breast remodeling in sion of the OTX1 gene, we cloned 1431 bp of the region rodent, we evaluated the level of OTX1 in p53À/À breast containing this RE in the pGL3 luciferase basic tissue in lactation (Jacks et al., 1994) Our results showed vector (pGL3-OTX1) to be used as minimal promoter. the presence of OTX1 gene expression also in p53 Luciferase assay was performed by co-transfecting deficient mice, demonstrating that although p53 could pGL3-OTX1 with increasing amounts of p53 expression drive the expression of OTX1 in cancer breast cells, it is vector, from a ratio of 1:1 through to 1:10. The results not involved in the physiological tissue remodeling indicate that this region is able to drive the luciferase during lactation (Figure 4b). transcription, behaving as a minimal promoter. This ability is directly proportional to the increase of p53 expression, confirming also the in vivo data from SaOs2 cells (Figure 2e). Discussion All these results lead to the conclusion that OTX1 is a target gene for p53, which can regulate its expression Breast cancer, one of the most common and feared through the p53RE-30. cancers, rank second after lung cancer among tumor- Furthermore, to test the interaction of other p53 related deaths (Ries et al., 2000). Indeed, one out of nine family members with the OTX1 promoter, we per- women are affected by breast cancer at some stage in formed a luc-assay experiment using the alpha isoforms life. The direct control of the tumor suppressor p53 over of both TA/DeltaNp63, and p73. We found that both the OTX1 homeobox gene raises the question on the were able to transactivate the OTX1 promoter, albeit functional role of this interaction in tumors. Although at lower levels than p53 (Melino et al., 2003; Murray- p53 has initially been studied for its tumor suppressor Zmijewski et al., 2006) (Figure 4a). function, recently it has been proposed as a central In order to investigate the molecular interaction molecule in the differentiation of CSCs (Cicalese et al., between OTX1 and p53 during cell differentiation, 2009). OTX1 has been described as a critical molecule LA7 CSCs were studied. qRT–PCR was performed for axon refinement in developing cerebral cortex, and on cDNA obtained from LA7 and LA7 induced to the developmental role of the OTX family genes have

Figure 2 Expression of orthodenticle homolog 1 gene (OTX1) by induction of p53 in SaOs2 cells. (a) p53 (upper panel) and OTX1 (lower panel) proteins were analyzed in the doxycycline inducible Tet-On/p53-SaOs2 cell line after 12, 24 and 48 h. The p53 and OTX1 protein levels were detected with specific antibodies. Beta-actin was used as a loading control. Primary antibodies were: monoclonal anti-OTX1 (ab58238 Abcam, Cambridge, UK, 1:200) and polyclonal anti-p53 (DO-1, catalog SC-126, Santa Cruz Biotechnology, Santa Cruz, CA, USA, 1:500). Western blotting was performed according to standard procedures. (b) Quantitative real-time reverse transcriptase PCR (qRT–PCR) shows increased levels of the OTX1 expression in Tet-On/p53-SaOs2 inducible cell line treated with doxycycline for 12 and 24 h (at this time OTX mRNA expression reach the plateau). Cells not induced (Dox0 h) were used as control, and beta-actin was used as internal standard for 2DDCt calculations. SaOs2-inducible cells were treated as previously described (Gressner et al., 2005). (c) OTX1 promoter analysis. The OTX1 gene structure shows the promoter carrying two putative p53 responsive elements (REs): 50 p53 RE (À3435 bp from the ATG) and 30 p53 RE ( þ 1268 bp from the ATG). In silico analysis has been performed using Genomatix MatInspector software (Genomatix Software GmbH, Mu¨nchen, Germany). (d) Chromatin immunoprecipitation analysis of the OTX1 promoter. Samples were assayed for the presence of 50 p53 RE (left panels) or the 30 p53 RE (right panels) of the OTX1 promoter using chromatin from SaOs2-p53TetON, untreated (0 h) or 12 and 24 h Dox-induced. Immunoprecipitation was carried out using p53 specific antibody (upper panels) or IgG-unspecific antibody (lower panels). Signal is only detectable in the sample at 24 h p53-induced cells, confirming a physical interaction between p53 and OTX1 promoter. 12 Â 106 cells were cross-linked and used for Chromatin immunoprecipitation analysis accordingly (Candi et al., 2006). (e) Schematic representation of luciferase reporter plasmid regulated by the putative OTX1 promoter (left). Transiently co-transfected SaOs2 cells were lysed 24 h after transfection and subjected to determination of luciferase activity, according to Promega Dual-Luciferase Reporter Assay System (E1910) (Promega Corporation, Madison, WI, USA) instructions. The graph clearly shows an increase of luciferase activity when increasing amounts of p53 expression vector are transfected (ratio 1:3; 1:5 and 1:10), in respect to the reporter plasmid. The empty vector has been used as a control. Data are presented as fold of induction over control. Three independent experiments were performed (mean þ /Às.d., n ¼ 3) (right).

Oncogene OTX1 expression in breast cancer A Terrinoni et al 3101

Figure 3 p53 and orthodenticle homolog 1 gene (OTX1) expression levels in cell differentiation. (a) Gene expression levels of OTX1 and p53. Quantitative real-time (qRT–PCR) was performed Figure 4 (a) Orthodenticle homolog 1 gene (OTX1) induction on cDNA obtained from LA7 and LA7, induced to differentiate by all p53 family members. Transiently co-transfected SaOs2. with dimethylsulfoxide (LA7D). Increase levels of both p53 and The graph shows an increase of luciferase activity, when activated OTX1 in LA7D is shown. The cell line LA7, a clonal derivative with p53 and TAp73-alpha. (see Figure 2e for protocols). Three from the Rama-25 line (Zucchi et al., 2007) recently sub-cloned, independent experiments were performed (mean þ /Às.d., n ¼ 3) was cultured in undifferentiating condition using Dulbecco’s (right). (b) OTX1 expression in p53À/À breast tissue during modified Eagle’s medium. For differentiation experiments in lactation. p53À/À OTX1 expression was evaluated in comparison adherent cultures, LA7 cells were grown at confluence and exposed with 8-month-old virgin mice by qRT–PCR. to 1.8% dimethylsulfoxide for 48 h. Cells were then harvested in trizol for RNA isolation. (b) Changes in OTX1 gene expression levels in mice during linear and cyclical breast development. RNA from breast tissue of virgin, pregnant and lactating CD1 mice have or symmetric division and the balance between these two been retro-transcribed and the cDNA have been analyzed with models is modulated according to developmental and real-time PCR. During lactation, when cells are differentiating into ductal-alveolar architecture, OTX1 shows higher expression levels, environmental signals (Morrison and Kimble, 2006). indicating its function in normal breast stem cell differentiation. Recent evidence suggests that symmetrical division is CD1 mice were from the Charles River Laboratories (Wilmington, necessary for neoplasia proliferation and that asym- MA, USA). Mice were maintained in accordance with the Italian metric division is a mechanism of tumor suppression in Ministry of Health and European Community guidelines. Expression drosophila neuroblasts (Gonzales, 2007). Recently, p53 studies were performed on two representative mice. All mice were deeply anesthetized and killed at 1 month and 8 months of age, and has been shown to regulate the polarity of cell division after 2 weeks of gestation, as well as 2 weeks of lactation. Mammary in mammary gland and its loss of function leads to gland tissue was preserved in RNA later (Ambion, c/o Applied symmetrical division resulting in tumor growth (Cicalese Biosystems, Foster City, CA, USA) until RNA extraction. et al., 2009). In order to verify the interaction between p53 and OTX1 in CSC differentiation, LA7 and LA7D cells were tested for p53 and OTX1 gene expression, and been well established in murine brain (Acampora et al., we found higher levels for both genes in the differ- 1996; Larsen et al., 2010). entiated cells. These data suggest that p53 and OTX1 The role of OTX1 could have a synergistic function overexpression represent an attempt to force the with p53 in the differentiation of CSCs in breast cancer. neoplastic cells to differentiate and this raises the Adult tissues like breast, colon and blood are main- hypothesis that OTX1 could be involved in the balance tained by proliferation of tissue SCs and the same seems between symmetrical and asymmetrical cells division. to apply to the tumors that originate from these tissues. Our study provides evidence that p53 controls OTX1 SCs are defined for their ability to make more SCs in tumors and that both genes are co-expressed during (self-renewal), as well as to produce cells able to the differentiation of breast cancer cells. OTX1 and differentiate. The balance between these two processes p53 are expressed during the embryonic development is determined in embryos and in adult tissues through when cells are subjected to rapid cell division and they the switch between asymmetric and symmetric cell are switched on again in breast cancer cells, suggesting divisions. Most SCs can divide either by asymmetric that their interaction could be involved in CSC

Oncogene OTX1 expression in breast cancer A Terrinoni et al 3102 differentiation as a tumor suppression mechanism. The CSCs and not in mammary gland physiological remo- OTX1 nuclear and/or cytoplasmic localization suggest deling. different functions of this molecule in cell differentiation (Zhang et al., 2002; Omodei et al., 2009). Additional data show different OTX1 gene expression levels in mice Conflict of interest breast tissues during the linear and the cyclical organ phases, suggesting its function in normal stem cell The authors declare no conflict of interest. differentiation. In p53 deficient mice, both lactation and involution appeared normal (Blackburn and Jerry, 2002). Our studies on p53À/À breast tissue in lactation Acknowledgements show the presence of OTX1 expression independent from p53, implying the function of other transcription We thank Centro Grandi Strumenti University of Insubria, factors in OTX1 regulation. Indeed, p63 and p73 Fondazione Comunitaria del Varesotto- ONLUS, Ministero members of the p53 family are expressed in p53À/À dell’Istruzione, dell’Universita` e della Ricerca DM25591 and breast tissue, and our luc-assay experiments show their Banca del Monte di Lombardia, Pavia, Italy to GP. This work has also been supported by the Medical Research Council, involvement in OTX1 regulation, in agreement with UK; ‘Alleanza contro il Cancro’ (ACC12-ACC6), MIUR/ other studies (Melino et al., 2003; Murray-Zmijewski PRIN (RBIP06LCA9_0023), AIRC (2008-2010_33-08), Italian et al., 2006). Human ProteomeNet RBRN07BMCT_007 and Telethon The significance of this work lies on the unexpected (GGPO9133) to GM, Ministero della Salute RF06-RF07 p53 control over the OTX1 gene expression only in to GM. GP thanks to AIL Varese.

References

Abate-Shen C. (2002). Deregulated homeobox gene expression in Larsen KB, Lutterodt MC, Mollgard K, Moller M. (2010). Expression cancer: cause or consequence? Nat Rev Cancer 2: 777–785. of the homeobox genes OTX2 and OTX1 in the early developing Acampora D, Mazan S, Avantaggiato V, Barone P, Tuorto F, human brain. J Histochem Cytochem 58: 669–678. Lallemand Y et al. (1996). Epilepsy and brain abnormalities in mice Levantini E, Giorgetti A, Cerisoli F, Traggiai E, Guidi A, Martin R lacking the Otx1 gene. Nat genet 14: 218–222. et al. (2003). Unsuspected role of the brain morphogenetic Argiropoulos B, Humphries RK. (2007). Hox genes in hematopoiesis gene Otx1 in hematopoiesis. Proc Natl Acad Sci USA 100: and leukemogenesis. Oncogene 26: 6766–6776. 10299–10303. Blackburn AC, Jerry DJ. (2002). Knockout and transgenic mice of Lewis MT. (2000). Homeobox genes in mammary gland development Trp53: what have we learned about p53 in breast cancer? Breast and neoplasia. Breast Cancer Res 2: 158–169. Cancer Res 4: 101–111. Lin T, Chao C, Saito S, Mazur SJ, Murphy ME, Appella E et al. Borresen-Dale AL. (2003). TP53 and breast cancer. Hum Mut 21: (2005). p53 induces differentiation of mouse embryonic stem cells by 292–300. suppressing Nanog expression. Nat Cell Biol 7: 165–171. Candi E, Terrinoni A, Ruﬁni A, Chikh A, Lena AM, Suzuki Y et al. Melino G, Lu X, Gasco M, Crook T, Knight RA. (2003). Functional (2006). p63 is upstream of IKK alpha in epidermal development. regulation of p73 and p63: development and cancer. Trends Biochem J Cell Sci 119: 4617–4622. Sci 28: 663–670. Carbone D, Chiba I, Mitsudomi T. (1991). Polymorphism at codon Morrison SJ, Kimble J. (2006). Asymmetric and symmetric stem-cell 213 within the p53 gene. Oncogene 6: 1691–1692. divisions in development and cancer. Nature 441: 1068–1074. Chen F, Capecchi MR. (1999). Paralogous mouse Hox genes, Hoxa9, Murray-Zmijewski F, Lane DP, Bourdon JC. (2006). p53/p63/p73 Hoxb9 and Hoxd9, function together to control development of the isoforms: an orchestra of isoforms to harmonise cell differentiation mammary gland in response to pregnancy. Dev Biol 96: 541–546. and response to stress. Cell Death Differ 13: 962–972. Cicalese A, Bonizzi G, Pasi CE, Faretta M, Ronzoni S, Giulini B et al. Omodei D, Acampora D, Russo F, De Filippi R, Severino V, (2009). The tumor suppressor p53 regulates polarity of self-renewing Di Francia R et al. (2009). Expression of the brain transcription divisions in mammary stem cells. Cell 138: 1083–1095. factor OTX1 occurs in a subset of normal germinal-center B cells Cillo C, Faiella A, Cantile M, Boncinelli E. (1999). Homeobox genes and in aggressive Non-Hodgkin Lymphoma. Am J Pathol 175: and cancer. Exp Cell Res 248: 1–9. 2609–2617. Danilova N, Sakamoto KM, Lin S. (2008). p53 family in development. Papadakis EN, Dokianakis DN, Spandidos DA. (2000). p53 codon 72 Mech Dev 125: 919–931. polymorphism as a risk factor in the development of breast cancer. Eizenberg O, Faber-Elman A, Gottlieb E, Oren M, Rotter V, Schwartz M. Mol Cell Biol Res Comm 3: 389–392. (1996). p53 plays a regulatory role in differentiation and apoptosis of Ries S, Biederer C, Woods D, Shifman O, Shirasawa S, Sasazuki T central nervous system-associated cells. MolCellBiol16: 5178–5185. et al. (2000). Opposing effects of Ras on p53: transcrip- Ford HL. (1998). Homeobox genes: a link between development, cell tional activation of mdm2 and induction of p19ARF. Cell 103: cycle, and cancer? Cell Biol Int 22: 397–400. 321–330. Gonzales G. (2007). Spindle orientation, asymmetric division and Riley T, Sontag E, Chen P, Levine A. (2008). Transcriptional tumour suppression in Drosophila stem cells. Nat Rev Genet 8: control of human p53-regulated genes. Nat Rev Mol Cell Biol 9: 462–472. 402–412. Gressner O, Schilling T, Lorenz K, Schulze Schleithoff E, Koch A, Smalley M, Ashworth A. (2003). Stem cells and breast cancer: a ﬁeld in Schulze-Bergkamen H et al. (2005). TAp63alpha induces apoptosis transit. Nature 3: 832–844. by activating signaling via death receptors and mitochondria. Embo Watabe T, Miyazono K. (2009). Roles of TGF-b family signal- J 24: 2458–2471. ing in stem cell renewal and differentiation. Cell Res 19: Jacks T, Remington L, Williams BO, Schmitt M, Halachmi S, 103–115. Bronson RT et al. (1994). Tumor spectrum analysis in p53-mutant Whibley C, Pharoah PD, Hollstein M. (2009). p53 polymorphisms: mice. Curr Biol 4: 1–7. cancer implications. Nat Rev Cancer 9: 95–107.

Oncogene OTX1 expression in breast cancer A Terrinoni et al 3103 Zhang YA, Okada A, Hong Lew C, McConnell SK. (2002). Regulated from a tumor generated by rat mammary cancer stem cells. Proc nuclear trafﬁcking of the homeodomain protein Otx1 in cortical Natl Acad Sci USA 105: 16940–16945. neurons. Mol Cell Neurosci 19: 430–446. Zucchi I, Sanzone S, Astigiano S, Pelucchi P, Scotti M, Valsecchi V Zucchi I, Astigiano S, Bertalot G, Sanzone S, Cocola C, Pelucchi P et al. (2007). The properties of a mammary gland cancer stem cell. et al. (2008). Distinct populations of tumor-initiating cells derived Proc Natl Acad Sci USA 104: 10476–10481.

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

Oncogene