WO2012073047A2 - Compositions and methods - Google Patents

Abstract

An inhibitor of BCL11Afor use in the prevention or treatment of cancer.

Description

Compositions and Methods

The present invention relates to breast cancer, including methods and compounds for diagnosing and treating breast cancer.

Background

Breast cancer is a heterogeneous disease with various prognostic outcomes. Recent research efforts have focused on identifying the cellular origin for these different types of breast cancer. One prevailing hypothesis is that different subtypes of breast cancer arise from distinct types of cells within the breast.

There is still a need to understand breast cancer development, in particular for aggressive cancers, to allow for better diagnosis, prognosis and treatment.

Statements of invention

The present invention relates to:

An inhibitor of BCL1 1 A for use in the prevention or treatment of cancer.

An inhibitor of BCL1 1 A for use in the prevention or treatment of breast cancer.

A method of prevention or treatment of breast cancer in an individual in need thereof, the method comprising delivery of an effective amount of an inhibitor of BCL1 1 A.

Use of an effective amount of an inhibitor of BCL1 1A in the preparation of a medicament for prevention or treatment of breast cancer in an individual in need thereof.

A method for assessing the severity of breast cancer, the method comprising determining the level of expression or activity of BCL1 1 A in a breast cell.

A method for diagnosing basal breast cancer, the method comprising determining the level of expression or activity of BCL1 1 A in a breast cell.

A kit for use in assessing the presence or severity of breast cancer, the kit comprising a detection reagent for determining the expression or activity of BCL1 1 A in a breast cell.

An activator of the BCL1 1 A gene or gene product in the growth of, and/or conversion of cells to, mammary stem cells (MaSCs)

An activator of BCL1 1 A gene or gene product for use in breast tissue regeneration. An inhibitor of BCL1 1A in combination with p53, or an activator of p53, for use in the prevention or treatment of breast cancer.

A method for the prevention or treatment of breast cancer, the method comprising delivery of a combination of an inhibitor of BCL1 1A and p53, or an activator of p53, to an individual in need thereof.

Figures

Figure 1 : Bcl11a is expressed in mammary stem cells. Figure 2: BCL1 1 A is required for mammary stem cells.

Figure 3: Bcl1 1a enhances sternness in mouse mammary epithelial cells by regulating key EMT and Notch signalling genes.

Figure 4: BCL11A is highly expressed in basal breast cancer and is essential for cancer stem cells.

Figure 5. BCL1 1A is Highly Expressed Specifically in Basal Breast Cancer

Figure 6. High Levels of BCL1 1A Enhance Self-Renewal of Mammary Epithelial Cells and promote Tumourigenesis

Figure 7. BCL1 1A Regulates Genes Implicated in Stem Cell Function and Tumourigenesis Figure 8. BCL1 1 A Is a Negative Regulator of p53 via MDM2 Supplementary Figure 1 : Bcl11a expression in the mammary gland.

Supplementary Figure 2: Schematic diagram of the Bcl11a conditional knockout allele used in this study.

Supplementary Figure 3: Loss of both basal and luminal epithelial cells in the Bc/77a-deficient mammary gland.

Supplementary Figure 4: Severe defects in the Bcl17a-deficient virgin mammary gland.

Supplementary Figure 5: over-expression leads to more CK14⁺ and fewer CK18⁺ EpH4 cells.

Supplementary Figure 6: Over-expression of BCL11A in human breast cell enhances sternness and detection of BCL11A expression in human breast cancer cell lines.

Supplementary Figure 7: Bcl11a is expressed in and essential for luminal progenitors.

Supplementary Figure 8. Expression of BCL1 1 A in Breast Cancer Supplementary Figure 9. BCL1 1 A Overexpression and Knockdown in mammary epithelial cells.

Supplementary Figure 10. High Levels of BCL1 1A promote Tumourigenesis whereas BCL1 1A knockdown induces differentiation.

Supplementary Figure 1 1. Bcl1 1a Negatively Regulates P53 Signalling

Supplementary Figure 12. SNP285C in MDM2 intron 1 Overlaps with the BCL1 1A Binding Site

Supplementary Figure 13: Figure 1. Bcl1 1a is expressed in hematopoietic stem cells (HSCs).

Supplementary Figure 14: Rapid depletion of long-term HSCs upon Bcl1 1a deletion.

Supplementary Figure 15 Bcl1 1a has non-cell autonomous essential roles in HSCs.

Supplementary Figure 16. Analysis of Bcl11a expression in tissues of an adult mouse using a lacZ reporter mouse.

Table 1. Oncogenes And Tumour Suppressors Differentially Regulated in BCL1 1A Overexpressing HMLE Cells.

Supplementary Table 1. Genotyping primers.

Supplementary Table 2. Primers and PCR conditions for RT-PCR analysis.

Supplementary Table 3. Primers for Chromatin Immuno-precipitation PCR analysis

Detailed description

In the present invention we demonstrate that BCL1 1A expression is a marker of mammary stem cells and a marker of basal breast cancer, that BCL1 1A over-expression correlates with low survival and that BCL1 1A knockdown induces cell death, reduces proliferative capacity and drives the terminal differentiation of basal breast cancer cells. Thus we demonstrate that the BCL1 1A gene and gene product can be targeted for prevention and treatment of breast cancer, and used in breast cancer diagnosis and prognosis.

BCL1 1A is a known protein, a C2H2 zinc finger transcription factor. It is essential for B-Cell development and controls the switch between foetal and adult haemoglobin. It is expressed in many epithelial tissues including the mammary gland, implicated in lymphoma pathogenesis. Human BCL1 1A has three isoforms; NM_018014.3 (aka, BCL1 1A-L), NM_022893.3 (aka BCL1 1A-XL) OR NM_138559.1 (aka BCL1 1A-S). The invention is not limited to a specific isoform, nor to human bcl1 1A and the methods and other aspects of the invention are contemplated for use in other mammals with equivalent genes. In a first aspect the invention relates to an inhibitor of BCL1 1A for use in the prevention or treatment of breast cancer.

Reference herein generally to an inhibitor, such as a BCL1 1A inhibitor, includes an inhibitor of gene expression and/or an inhibitor of gene product (e.g. protein) activity.

In one aspect reference to a BCL1 1 A gene product is a reference to the BCL1 1 A protein.

Any of a number of different approaches can be taken to inhibit BCL1 1A expression or activity. An inhibitor includes any chemical or biological entity that, upon treatment of a cell, results in inhibition of the amount and/or biological activity of BCL1 1A, for example biological activity caused by activation of BCL1 1 A in response to cellular signals.

WO2010/030963 discloses a number of different approaches to inhibition of the BCL1 1A, both at the level of the gene or gene product, related to modulation of BCL1 1 A activities for treatment of hemoglobinopathies, and such approaches can be adopted herein for prevention and/or treatment of breast cancer.

The inhibitor may be an inhibitor of gene expression, such as an RNA, for example an inhibitor utilising RNA interference RNAi, for example miRNA, or SiRNA or shRNA.

The inhibitor may also be an inhibitor of protein activity, such as a monoclonal or polyclonal antibody or fragment thereof, including a domain antibody, and fully or partially humanised antibodies.

The inhibitor may also be a small molecule which binds to, and inhibits, protein function.

The inhibitor, activator or other modulator of the invention, may be a Transcription activator-like effectors (TALEs), for example a polypeptide capable of binding to specific DNA sequence to regulate gene expression (eg see Proceedings of the National Academy of Sciences 107 (50): 21617-21622 and Nature Biotechnology 29 (2): 135-6.).

The inhibitor may be a complete or partial inhibitor, preferably sufficient to reduce the proliferative capacity of breast cancer cells.

Bcl11a deficiency caused depletion of MaSCs in the mouse mammary gland and loss of luminal progenitors. Thus the inhibitors of the invention are also disclosed for use in depletion of MaSCs and luminal progenitors, for example those that give rise to ERo luminal cells.

As described in WO2010030963, antibodies that specifically bind BCL1 1A can be used for the inhibition of the factor in vivo. Antibodies to BCL1 1 A are commercially available and can be raised by one of skill in the art using well known methods. The BCL1 1 A inhibitory activity of a given antibody, or, for that matter, any BCL1 1 A inhibitor, can be assessed using methods known in the art or described herein, such as reduction in cellular proliferative capacity. Antibody inhibitors of BCL11 A can include polyclonal and monoclonal antibodies and antigen- binding derivatives or fragments thereof. Well known antigen binding fragments include, for example, single domain antibodies (dAbs; which consist essentially of single VL or VH antibody domains), Fv fragment, including single chain Fv fragment (scFv), Fab fragment, and F(ab')2 fragment. Methods for the construction of such antibody molecules are well known in the art.

In more detail, reference to antibodies includes complete immunoglobulins, antigen binding fragments of immunoglobulins, as well as antigen binding proteins that comprise antigen binding domains of immunoglobulins. Antigen binding fragments of immunoglobulins include, for example, Fab, Fab', F(ab')2, scFv and dAbs. Modified antibody formats have been developed which retain binding specificity, but have other characteristics that may be desirable, including for example, bispecificity, multivalence (more than two binding sites), and compact size (e.g., binding domains alone). Single chain antibodies lack some or all of the constant domains of the whole antibodies from which they are derived. Therefore, they can overcome some of the problems associated with the use of whole antibodies. For example, single-chain antibodies tend to be free of certain undesired interactions between heavy-chain constant regions and other biological molecules. Additionally, single-chain antibodies are considerably smaller than whole antibodies and can have greater permeability than whole antibodies, allowing single-chain antibodies to localize and bind to target antigen- binding sites more efficiently. Furthermore, the relatively small size of single-chain antibodies makes them less likely to provoke an unwanted immune response in a recipient than whole antibodies. Multiple single chain antibodies, each single chain having one VH and one VL domain covalently linked by a first peptide linker, can be covalently linked by at least one or more peptide linker to form multivalent single chain antibodies, which can be monospecific or multispecific. Each chain of a multivalent single chain antibody includes a variable light chain fragment and a variable heavy chain fragment, and is linked by a peptide linker to at least one other chain. The peptide linker is composed of at least fifteen amino acid residues. The maximum number of linker amino acid residues is approximately one hundred. Two single chain antibodies can be combined to form a diabody, also known as a bivalent dimer. Diabodies have two chains and two binding sites, and can be monospecific or bispecific. Each chain of the diabody includes a VH domain connected to a VL domain. The domains are connected with linkers that are short enough to prevent pairing between domains on the same chain, thus driving the pairing between complementary domains on different chains to recreate the two antigen- binding sites. Three single chain antibodies can be combined to form triabodies, also known as trivalent trimers. Triabodies are constructed with the amino acid terminus of a VL or VH domain directly fused to the carboxyl terminus of a VL or VH domain, i.e., without any linker sequence. The triabody has three Fv heads with the polypeptides arranged in a cyclic, head-to- tail fashion. A possible conformation of the triabody is planar with the three binding sites located in a plane at an angle of 120 degrees from one another. Triabodies can be monospecific, bispecific or trispecific. Thus, antibodies useful in the methods described herein include, but are not limited to, naturally occurring antibodies, bivalent fragments such as (Fab')2, monovalent fragments such as Fab, single chain antibodies, single chain Fv (scFv), single domain antibodies, multivalent single chain antibodies, diabodies, triabodies, and the like that bind specifically with an antigen. Antibodies can also be raised against a polypeptide or portion of a polypeptide by methods known to those skilled in the art. Antibodies are readily raised in animals such as rabbits or mice by immunization with the gene product, or a fragment thereof. Immunized mice are particularly useful for providing sources of B cells for the manufacture of hybridomas, which in turn are cultured to produce large quantities of monoclonal antibodies. While both polyclonal and monoclonal antibodies can be used in the methods described herein, it is preferred that a monoclonal antibody is used where conditions require increased specificity for a particular protein.

A powerful approach for inhibiting the expression of selected target polypeptides is through the use of RNA interference agents. RNA interference (RNAi) uses small interfering RNA (siRNA) duplexes that target the messenger RNA encoding the target polypeptide for selective degradation. siRNA-dependent post-transcriptional silencing of gene expression involves cleaving the target messenger RNA molecule at a site guided by the siRNA. "RNA interference (RNAi)" is an evolutionally conserved process whereby the expression or introduction of RNA of a sequence that is identical or highly similar to a target gene results in the sequence specific degradation or specific post-transcriptional gene silencing (PTGS) of messenger RNA (mRNA) transcribed from that targeted gene (see Coburn, G. and Cullen, B. (2002) J. of Virology 76(18):9225), thereby inhibiting expression of the target gene. In one embodiment, the RNA is double stranded RNA (dsRNA). This process has been described in plants, invertebrates, and mammalian cells. In nature, RNAi is initiated by the dsRN A- specific endonuclease Dicer, which promotes processive cleavage of long dsRNA into double- stranded fragments termed siRNAs. siRNAs are incorporated into a protein complex (termed "RNA induced silencing complex," or "RISC") that recognizes and cleaves target mRNAs. RNAi can also be initiated by introducing nucleic acid molecules, e.g., synthetic siRNAs or RNA interfering agents, to inhibit or silence the expression of target genes.

Reference to an inhibitor of gene expression or inhibition of gene expression herein includes any decrease in expression or protein activity or level of gene or protein encoded by the gene as compared to a situation wherein no inhibition, for example by RNA interference, has been induced. The decrease may be of at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95% or 99% or more as compared to the expression of a target gene or the activity or level of the protein encoded by a target gene which has not been targeted by the inhibitor.

The terms "RNA interference agent" and "RNA interference" as they are used herein are intended to encompass those forms of gene silencing mediated by double-stranded RNA, regardless of whether the RNA interfering agent comprises an siRNA, miRNA, shRNA or other double-stranded RNA molecule. "Short interfering RNA" (siRNA), also referred to herein as "small interfering RNA" is defined as an RNA agent which functions to inhibit expression of a target gene, e.g., by RNAi. An siRNA may be chemically synthesized, may be produced by in vitro transcription, or may be produced within a host cell. In one embodiment, siRNA is a double stranded RNA (dsRNA) molecule of about 15 to about 40 nucleotides in length, preferably about 15 to about 28 nucleotides, more preferably about 19 to about 25 nucleotides in length, and more preferably about 19, 20, 21 , 22, or 23 nucleotides in length, and may contain a 3' and/or 5' overhang on each strand having a length of about 0, 1 , 2, 3, 4, or 5 nucleotides. The length of the overhang is independent between the two strands, i.e., the length of the overhang on one strand is not dependent on the length of the overhang on the second strand. Preferably the siRNA is capable of promoting RNA interference through degradation or specific post- transcriptional gene silencing (PTGS) of the target messenger RNA (mRNA). siRNAs also include small hairpin (also called stem loop) RNAs (shRNAs). In one embodiment, these shRNAs are composed of a short ( e.g., about 19 to about 25 nucleotide) antisense strand, followed by a nucleotide loop of about 5 to about 9 nucleotides, and the analogous sense strand. Alternatively, the sense strand may precede the nucleotide loop structure and the antisense strand may follow. These shRNAs may be contained in plasmids, retroviruses, and lenti viruses and expressed from, for example, the pol III U6 promoter, or another promoter ( see, e.g., Stewart, et al. (2003) RNA Apr;9(4):493-501 , incorporated by reference herein in its entirety). The target gene or sequence of the RNA interfering agent may be a cellular gene or genomic sequence, e.g. the BCL1 1 A sequence. An siRNA may be substantially homologous to the target gene or genomic sequence, or a fragment thereof. As used in this context, the term "homologous" is defined as being substantially identical, sufficiently complementary, or similar to the target mRNA, or a fragment thereof, to effect RNA interference of the target. In addition to native RNA molecules, RNA suitable for inhibiting or interfering with the expression of a target sequence include RNA derivatives and analogs. Preferably, the siRNA is identical to its target. The siRNA preferably targets only one sequence. Each of the RNA interfering agents, such as siRNAs, can be screened for potential off-target effects by, for example, expression profiling. Such methods are known to one skilled in the art and are described, for example, in Jackson et al. Nature Biotechnology 6:635-637, 2003. In addition to expression profiling, one may also screen the potential target sequences for similar sequences in the sequence databases to identify potential sequences which may have off-target effects. 15, or perhaps as few as 1 1 , contiguous nucleotides of sequence identity are sufficient to direct silencing of non-targeted transcripts. Therefore, one may initially screen the proposed siRNAs to avoid potential off-target silencing using the sequence identity analysis by any known sequence comparison methods, such as BLAST. siRNA sequences are chosen to maximize the uptake of the antisense (guide) strand of the siRNA into RISC and thereby maximize the ability of RISC to target human GGT mRNA for degradation. This can be accomplished by scanning for sequences that have the lowest free energy of binding at the 5'- terminus of the antisense strand. The lower free energy leads to an enhancement of the unwinding of the 5'- end of the antisense strand of the siRNA duplex, thereby ensuring that the antisense strand will be taken up by RISC and direct the sequence-specific cleavage of the human BCL1 1 A mRNA. siRNA molecules need not be limited to those molecules containing only RNA, but, for example, further encompasses chemically modified nucleotides and non- nucleotides, and also include molecules wherein a ribose sugar molecule is substituted for another sugar molecule or a molecule which performs a similar function. Moreover, a non- natural linkage between nucleotide residues can be used, such as a phosphorothioate linkage. The RNA strand can be derivatized with a reactive functional group of a reporter group, such as a fluorophore. Particularly useful derivatives are modified at a terminus or termini of an RNA strand, typically the 3' terminus of the sense strand. For example, the 2'-hydroxyl at the 3' terminus can be readily and selectively derivatizes with a variety of groups. Other useful RNA derivatives incorporate nucleotides having modified carbohydrate moieties, such as 2Ό- alkylated residues or 2'-0-methyl ribosyl derivatives and 2'-0-fluoro ribosyl derivatives. The RNA bases may also be modified. Any modified base useful for inhibiting or interfering with the expression of a target sequence may be used. For example, halogenated bases, such as 5- bromouracil and 5- iodouracil can be incorporated. The bases may also be alkylated, for example, 7- methylguanosine can be incorporated in place of a guanosine residue. Non-natural bases that yield successful inhibition can also be incorporated. The most preferred siRNA modifications include 2'-deoxy-2'-fluorouridine or locked nucleic acid (LAN) nucleotides and RNA duplexes containing either phosphodiester or varying numbers of phosphorothioate linkages. Such modifications are known to one skilled in the art and are described, for example, in Braasch et al., Biochemistry, 42: 7967-7975, 2003. Most of the useful modifications to the siRNA molecules can be introduced using chemistries established for antisense oligonucleotide technology. Preferably, the modifications involve minimal 2'-0-methyl modification, preferably excluding such

modification. Modifications also preferably exclude modifications of the free 5'- hydroxyl groups of the siRNA. The Examples of WO2010030963 provide specific examples of RNA interfering agents, such as shRNA molecules that effectively target BCL1 1 A mRNA.

In a preferred embodiment, the RNA interference agent is delivered or administered in a pharmaceutically acceptable carrier. Additional carrier agents, such as liposomes, can be added to the pharmaceutically acceptable carrier. In another embodiment, the RNA interference agent is delivered by a vector encoding small hairpin RNA (shRNA) in a pharmaceutically acceptable carrier to the cells in an organ of an individual. The shRNA is converted by the cells after transcription into siRNA capable of targeting, for example, BCL1 1 A.

In one embodiment, the vector is a regulatable vector, such as tetracycline inducible vector. Methods described, for example, in Wang et al. Proc. Natl. Acad. Sci. 100: 5103-5106, using pTet-On vectors (BD Biosciences Clontech, Palo Alto, CA) can be used. In one embodiment, the RNA interference agents used in the methods described herein are taken up actively by cells in vivo following intravenous injection, e.g., hydrodynamic injection, without the use of a vector. One method to deliver the siRNAs is catheterization of the blood supply vessel of the target organ. Other strategies for delivery of the RNA interference agents, e.g., the siRNAs or shRNAs used in the methods of the invention, may also be employed, such as, for example, delivery by a vector, e.g., a plasmid or viral vector, e.g., a lentiviral vector. Such vectors can be used as described, for example, in Xiao-Feng Qin et al. Proc. Natl. Acad. Sci. U.S.A., 100: 183-188. Other delivery methods include delivery of the RNA interfering agents, e.g., the siRNAs or shRNAs of the invention, using a basic peptide by conjugating or mixing the RNA interfering agent with a basic peptide, e.g., a fragment of a TAT peptide, mixing with cationic lipids or formulating into particles. The RNA interference agents, e.g., the siRNAs targeting BCL1 1A imRNA, may be delivered singly, or in combination with other RNA interference agents, e.g., siRNAs, such as, for example siRNAs directed to other cellular genes. BCL1 1 A siRNAs may also be administered in combination with other pharmaceutical agents which are used to treat or prevent diseases or disorders associated with oxidative stress, especially respiratory diseases, and more especially asthma. Synthetic siRNA molecules, including shRNA molecules, can be obtained using a number of techniques known to those of skill in the art. For example, the siRNA molecule can be chemically synthesized or recombinantly produced using methods known in the art, such as using appropriately protected ribonucleoside phosphoramidites and a conventional DNA RNA synthesizer (see, e.g., Elbashir, S.M. et al. (2001 ) Nature 41 1 :494-498; Elbashir, S.M., W.

Lendeckel and T. Tuschl (2001 ) Genes & Development 15: 188-200; Harborth, J. et al . (2001 ) J. Cell Science 1 14:4557-4565; Masters, J.R. et al. (2001 ) Proc. Natl. Acad. Sci., USA 98:8012- 8017; and Tuschl, T. et al . (1999) Genes & Development 13:3191- 3197). Alternatively, several commercial RNA synthesis suppliers are available including, but not limited to, Proligo (Hamburg, Germany), Dharmacon Research (Lafayette, CO, USA), Pierce Chemical (part of Perbio Science , Rockford, IL , USA), Glen Research (Sterling, VA, USA), ChemGenes (Ashland, MA, USA), and Cruachem (Glasgow, UK). As such, siRNA molecules are not overly difficult to synthesize and are readily provided in a quality suitable for RNAi. In addition, dsRNAs can be expressed as stem loop structures encoded by plasmid vectors, retroviruses and lentiviruses (Paddison, PJ. et al. (2002) Genes Dev. 16:948-958; McManus, M.T. et al. (2002) RNA 8:842-850; Paul, CP. et al. (2002) Nat. Biotechnol. 20:505- 508; Miyagishi, M. et al. (2002) Nat. Biotechnol. 20:497-500; Sui, G. et al. (2002) Proc. Natl. Acad. ScL, USA 99:5515-5520; Brummelkamp, T. et al. (2002) Cancer Cell 2:243; Lee, N.S., et al. (2002) Nat. Biotechnol. 20:500-505; Yu, J.Y., et al. (2002) Proc. Natl. Acad. ScL, USA 99:6047-6052; Zeng, Y., et al. (2002) Mol. Cell 9: 1327-1333; Rubinson, D.A., et al. (2003) Nat. Genet. 33:401-406; Stewart, S.A., et al. (2003) RNA 9:493-501 ). These vectors generally have a pol III promoter upstream of the dsRNA and can express sense and antisense RNA strands separately and/or as a hairpin structures. Within cells, Dicer processes the short hairpin RNA (shRNA) into effective siRNA. The targeted region of the siRNA molecule of the present invention can be selected from a given target gene sequence, e.g. a BCLI IA coding sequence, beginning from about 25 to 50 nucleotides, from about 50 to 75 nucleotides, or from about 75 to 100 nucleotides downstream of the start codon. Nucleotide sequences may contain 5' or 3' UTRs and regions nearby the start codon. One method of designing a siRNA molecule of the present invention involves identifying the 23 nucleotide sequence motif AA(NI9)TT (where N can be any nucleotide) and selecting hits with at least 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70% or 75% G/C content. The "TT" portion of the sequence is optional. Alternatively, if no such sequence is found, the search may be extended using the motif NA(N21 ), where N can be any nucleotide. In this situation, the 3' end of the sense siRNA may be converted to TT to allow for the generation of a symmetric duplex with respect to the sequence composition of the sense and antisense 3' overhangs. The antisense siRNA molecule may then be synthesized as the complement to nucleotide positions 1 to 21 of the 23 nucleotide sequence motif. The use of symmetric 3' TT overhangs may be advantageous to ensure that the small interfering ribonucleoprotein particles (siRNPs) are formed with approximately equal ratios of sense and antisense target RNA-cleaving siRNPs (Elbashir et al., (2001 ) supra and Elbashir et al., 2001 supra). Analysis of sequence databases, including but not limited to the NCBI, BLAST, Derwent and GenSeq as well as commercially available oligosynthesis companies such as

OLIGOENGINE®, may also be used to select siRNA sequences against EST libraries to ensure that only one gene is targeted.

Methods of delivering RNA interference agents, e.g., an siRNA, or vectors containing an RNA interference agent, to the target cells, e.g., lymphocytes or other desired target cells, for uptake include injection of a composition containing the RNA interference agent, e.g., an siRNA, or directly contacting the cell with a composition comprising an RNA interference agent, e.g., an siRNA. In another embodiment, RNA interference agent, e.g., an siRNA may be injected directly into any blood vessel, such as vein, artery, venule or arteriole, via, e.g., hydrodynamic injection or catheterization. Administration may be by a single injection or by two or more injections. The RNA interference agent is delivered in a pharmaceutically acceptable carrier. One or more RNA interference agent may be used simultaneously. In one preferred embodiment, only one siRNA that targets human BCL1 1 A is used. In one embodiment, specific cells are targeted with RNA interference, limiting potential side effects of RNA interference caused by non-specific targeting of RNA interference. The method can use, for example, a complex or a fusion molecule comprising a cell targeting moiety and an RNA interference binding moiety that is used to deliver RNA interference effectively into cells. For example, an antibody-protamine fusion protein when mixed with siRNA, binds siRNA and selectively delivers the siRNA into cells expressing an antigen recognized by the antibody, resulting in silencing of gene expression only in those cells that express the antigen. The siRNA or RNA interference-inducing molecule binding moiety is a protein or a nucleic acid binding domain or fragment of a protein, and the binding moiety is fused to a portion of the targeting moiety. The location of the targeting moiety can be either in the carboxyl-terminal or amino-terminal end of the construct or in the middle of the fusion protein. A viral-mediated delivery mechanism can also be employed to deliver siRNAs to cells in vitro and in vivo as described in Xia, H. et al. (2002) Nat Biotechnol 20(10): 1006). Plasmid- or viral-mediated delivery mechanisms of shRNA may also be employed to deliver shRNAs to cells in vitro and in vivo as described in Rubinson, D.A., et al. ((2003) Nat. Genet. 33:401-406) and Stewart, S.A., et al. ((2003) RNA 9:493-501 ). The RNA interference agents, e.g., the siRNAs or shRNAs, can be introduced along with components that perform one or more of the following activities: enhance uptake of the RNA interfering agents, e.g., siRNA, by the cell, e.g., lymphocytes or other cells, inhibit annealing of single strands, stabilize single strands, or otherwise facilitate delivery to the target cell and increase inhibition of the target gene, e.g., BCLI IA. The dose of the particular RNA interfering agent will be in an amount necessary to effect RNA interference, e.g., post translational gene silencing (PTGS), of the particular target gene, thereby leading to inhibition of target gene expression or inhibition of activity or level of the protein encoded by the target gene. The invention also relates to a vector for delivery of an inhibitor of the present invention, such as a vector for delivery of an shRNA, such as a piggyBac transposon, for example the PB-H1- shRNA-GFP construct as described herein. The vector may be an expression vector, from which an inhibitor can be expressed.

The invention also relates to cells comprising said vectors.

In one aspect the BCL1 1 A inhibitor of the invention is able to cause an increase in miR-200c expression in the cell to which it is delivered, in one aspect at least 10 fold, in one aspect at least 20 fold, in one aspect at least 3 fold, in one aspect at least 40 fold, in one aspect at least 50 fold, in one aspect at least 60 fold, in one aspect at least 70 fold, in one aspect at least 80 fold, in one aspect at least 90 fold, in one aspect at least 100 fold.

An inhibitor of BCL1 1A (e.g. expression or activity) may be a direct inhibitor of the BCL1 1A gene or gene product, for example it may bind directly to the BCL1 1 A protein or directly regulate the bell 1 A gene. Alternatively an inhibitor may act upon a downstream gene or gene product whose activity is regulated by the BCL1 1 A gene or gene product. Alternatively it may act upon an upstream gene or gene product which regulates the BCL1 1 A gene or gene product. Suitably such an inhibitor produces a downstream modulation in a biological pathway in which said BCL1 1A protein is involved.

The present invention discloses that over-expression of BCL1 1 A led to an up-regulation of Notch3 and the Epithelial to Mesenchymal Transition (EMT) inducers; ZEB1 , ZEB2, FOXC2 and TWIST2 expression. Accordingly the invention relates to a modulator of Notch expression or activity for use in the treatment of breast cancer. It also relates to a modulator of EMT for use in the treatment of breast cancer, such as a modulator of the expression or activity of an EMT inducer such as ZEB1 , ZEB2, FOXC2 and TWIST2 for use in the treatment of breast cancer. In particular the invention relates to inhibitors of the expression or activity of Notch 3 and/or an EMT inducer for use in this way.

BCL1 1A is thought to interact with binding partners: GATA-I, FOG-I, components of the NuRD complex, matrin-3, MTA2 and RBBP7. Accordingly, any antibody or fragment thereof, small molecule, chemical or compound that can block any of these interactions is considered an inhibitor of BCL1 1 A activity.

In another aspect the invention relates to miR-200c for use in the prevention or treatment of breast cancer. Reference to miR-200c includes functional equivalents thereof, capable of behaving in substantially the same way as miR-200c in the context of cell differentiation.

Functional equivalents may include variations of 1 , 2, 3, 4, 5 or more nucleotides. In a further aspect the invention relates to a modulator of the activity of notch 1 for use in treatment or prevention of breast cancer. Bcl1 1 A has been shown to modulate the expression of Notch 1 herein, with increased expression of Notchl in a bell 1 A deletion.

The compounds of the invention, such as a modulator including an inhibitor or activator, may be formulated into pharmaceutical compositions prior to administration to a patient by an appropriate route. Accordingly, in another aspect, the invention provides pharmaceutical compositions comprising an inhibitor of the invention and one or more pharmaceutically-acceptable excipients.

The pharmaceutical compositions of the invention typically contain one inhibitor of the invention. However, in certain embodiments, the pharmaceutical compositions of the invention contain more than one inhibitor of the invention. In addition, the pharmaceutical compositions of the invention may comprise one or more additional pharmaceutically active compounds.

As used herein, "pharmaceutically-acceptable excipient" means any pharmaceutically acceptable material present in the pharmaceutical composition or dosage form other than inhibitor(s) of the invention. Suitable pharmaceutically-acceptable excipients include the following types of excipients: diluents, fillers, binders, disintegrants, lubricants, glidants, granulating agents, coating agents, wetting agents, solvents, co-solvents, suspending agents, emulsifiers, sweeteners, flavouring agents, flavour masking agents, colouring agents, anticaking agents, humectants, chelating agents, plasticizers, viscosity increasing agents, rate modifying agents, antioxidants, preservatives, stabilizers, surfactants and buffering agents. The skilled person will appreciate that certain pharmaceutically-acceptable excipients may serve more than one function and may serve alternative functions depending on how much of the excipient is present in the formulation and what other ingredients are present in the formulation.

The skilled person possesses the knowledge and skill in the art to enable them to determine suitable pharmaceutically-acceptable excipients in appropriate amounts for use with the compounds of the invention. In addition, there are a number of resources that are available to the skilled person which describe pharmaceutically-acceptable excipients and may be useful in selecting suitable pharmaceutically-acceptable excipients. Examples include Remington's Pharmaceutical Sciences (Mack Publishing Company), The Handbook of Pharmaceutical Additives (Gower Publishing Limited), and The Handbook of Pharmaceutical Excipients (the American Pharmaceutical Association and the Pharmaceutical Press). The pharmaceutical compositions of the invention may be prepared using techniques and methods known to those skilled in the art. Some of the methods commonly used in the art are described in Remington's Pharmaceutical Sciences (Mack Publishing Company).

In another aspect, the invention provides dosage forms comprising modulators of the invention such as inhibitors, activators, or pharmaceutical compositions of the invention. Each discrete dosage form contains from 1 ng - 500 mg of a inhibitor of the invention. In another aspect, each discrete dosage form contains from ^g to 100 mg of the inhibitor, such as 10 μg to 100mg, such as 10μg to 50mg.

The compositions of the invention will typically be formulated into dosage forms which are adapted for administration to the patient by the desired route of administration. For example, dosage forms include those adapted for (1 ) oral administration such as tablets, capsules, caplets, pills, lozenges, powders, syrups, elixirs, suspensions, solutions, emulsions, sachets and cachets; (2) parenteral administration such as sterile solutions, suspensions, implants and powders for reconstitution; (3) transdermal administration such as transdermal patches; (4) rectal and vaginal administration such as suppositories, pessaries and foams; (5) inhalation and intranasal such as dry powders, aerosols, suspensions and solutions (sprays and drops); (6) topical administration such as creams, ointments, lotions, solutions, pastes, drops, sprays, foams and gels; (7) ocular administration such as drops, ointment, sprays, suspensions and inserts; (8) buccal and sublingual administration such as lozenges, patches, sprays, drops, chewing gums and tablets.

The present invention also relates to a method of prevention or treatment of breast cancer in an individual in need thereof, the method comprising delivery of an effective amount of an inhibitor of BCL1 1A.

The invention also relates to use of an effective amount of an inhibitor of the BCL1 1A in the preparation of a medicament for prevention or treatment of breast cancer in an individual in need thereof.

In one aspect the breast cancer is basal breast cancer.

In one aspect individuals have not been diagnosed with cancer. These individuals may be considered at risk of breast cancer, for example may be those individuals in whom there is a familial history of cancer.

In one aspect patients have been diagnosed with breast cancer but not the specific subtype of cancer (basal, HER2, luminal A, luminal B or normal).

In one aspect patients have been diagnosed with basal breast cancer.

Thus in one aspect the invention relates to a 'test and treat' approach in which individuals are tested for the presence of basal breast cancer, and treated with an inhibitor of the invention where basal breast cancer is detected.

It will be appreciated that the inhibitors of BCL1 1 A may be used across all breast cancer patients, irrespective of type, where screening to determine cancer subtype may result in a delay that is clinically not acceptable. Thus the invention relates to the use of an inhibitor of the invention in a population of breast cancer patients, or individuals suspected of having breast cancer, or at risk from breast cancer. Basal breast cancers are generally negative for expression of the estrogen receptor, progesterone receptor, and human epidermal growth factor receptor (HER)-2, and this phenotype can be used to identify such cases.

The inhibitors of the invention may be delivered by any suitable route. Some are discussed above. Generally they may be delivered by injection, infusion, instillation, or ingestion. "Injection" includes, without limitation, intravenous, intramuscular, intra-arterial, intra-thecal, intra-ventricular, intra-capsular, intra-orbital, intra-cardiac, intradermal, intraperitoneal, transtracheal, subcutaneous, subcuticular, intra-articular, sub capsular, subarachnoid, intra-spinal, intracerebro spinal, and intra-sternal injection and infusion.

It will be recognised by one of skill in the art that the optimal quantity and spacing of individual dosages of compounds of the invention, e.g. inhibitors or activators, will be determined by the nature and extent of the condition being treated, the form, route and site of administration, and the particular mammal being treated, and that such optimums can be determined by conventional techniques. It will also be appreciated by one of skill in the art that the optimal course of treatment, i.e., the number of doses of compounds of the invention given per day for a defined number of days, can be ascertained by those skilled in the art using conventional course of treatment determination tests.

The invention also relates to a method for assessing the severity and/or prognosis of a breast cancer, the method comprising determining the level of expression or activity of the BCL1 1 A gene or gene product in the breast cancer cell.

In one aspect the breast cancer is basal breast cancer cell.

The invention further relates to a method for screening for and/or detecting breast cancer, the method comprising determining the level of expression or activity of BCL1 1A gene or gene product in a breast cell, or expression of activity of a gene or gene product whose expression or activity is controlled by the BCL1 1 A gene. The screening may be carried out in vitro.

The level of expression may be compared to a control non-cancerous cell, or a predetermined threshold indicative of cancerous activity or progression suitably standardised by reference to the expression of a housekeeping gene.

BCL1 1A gene or gene product expression or activity may be assessed at the level of protein or nucleic acid, such as by determination of mRNA levels.

The present invention also relates to a kit for use in the methods of the invention, the kit comprising a detection reagent for BCL1 1 A gene or gene product expression or activity. Suitably the kit also comprises one or more of: instructions for use; diagnostic reagents such as detection reagents, buffers and water. The present invention also relates to a kit comprising a detection reagent for BCL1 1A gene or gene product expression or activity, combined with an inhibitor of BCL1 1 A as defined above.

In a further aspect the invention relates to an imaging agent for imaging /screening for breast cancer, the imaging agent suitably being specific for the detection of BCL1 1 A. In one aspect the imaging agent is capable of detection of cancerous cells. In one aspect the agent may be used to distinguish levels of bell 1 A gene or protein in normal and cancerous cells. In one aspect the agent is capable of distinguishing cells which over-express BCL1 1 A from those with normal levels of BCL1 1 A. In one aspect the agent is for detection of the BCL1 1 A protein. In one aspect the agent is for detection of nucleic acid encoding, or transcribed from, bell 1A. In one aspect the agent is detectable by MRI, CT or ultrasound. The agent may be labelled with a detectable marker. The agent may be an inhibitor of bell 1 A, for example, an antibody or fragment thereof.

In a yet further aspect the invention relates to a method of treatment for breast cancer comprising delivery of a cytopathic agent specifically to cancer cells such as basal cancer cells. The invention also relates to a cytopathic agent for treatment of a cancer cell, such as a breast cancer cell, the agent comprising a component capable of killing a cell and a component having specificity for a cell with over-expression of BCL1 1 A or an effector thereof. In one aspect specificity may be achieved by delivery of the cytopathic agent to cells which over-express the BCL1 1 A protein. The cytopathic agent may comprise an antibody against BCL1 1 A or fragment thereof. The cytopathic agent may comprise a radioactive agent or nano- or microparticles that can be used to generate local heating to kill cells, or comprise ligands or molecular complexes carrying or containing drugs effective to kill the cancer cell.

In a yet further aspect the means for inhibition of bell 1 a may be delivered in an amount that has an effect on cancer cells which over-express BCL1 1a, but which amount does not adversely affect cells which exhibit normal BCL1 1a expression.

In a further aspect the invention relates an activator of bcl1 1A gene or protein (e.g. of gene expression or gene product activity) for use in the growth of, and/or conversion of cells to, mammary stem cells (MaSCs), and to the use of an activator of BCL1 1 A gene or gene product in breast tissue regeneration.

In a further aspect the invention relates to a method for generating patient specific mammary stem cells by expressing and /or activating the BCL1 1A gene or polypeptide (for example, by inducible or transient expression). In this way patient specific MaSCs can be generated which can be stored and used in the future for drug screens using large libraries of inhibitors or other modulators. Stored MaSCs can be frozen and re-grown and driven to generate normal mammary tissue or breast cancer (in case of cancer patients) in humanised mice or in vitro. In this way new drugs may be screened on breast tissue. Thus, the invention further relates to a method for screening a drug comprising contacting a drug with a cell derived from a MaSC, obtained or obtainable by modulation of the expression of BCL1 1 A.

We have presented data herein that demonstrates that wild type p53 is effective when BCL1 1 A is knocked-down in cells.

Thus in a further aspect the present invention relates to: a combination of (i) an inhibitor of BCL1 1A and (ii) p53, or an activator of p53, for use in the prevention or treatment of breast cancer; a combination of (i) an inhibitor of BCL1 1A and (ii) p53, or an activator of p53, optionally in combination with a pharmaceutically acceptable excipient or carrier, such as any described herein; use of a combination of (i) an inhibitor of BCL1 1A and (ii) p53, or an activator of p53, in the preparation of a medicament for use in the prevention or treatment of breast cancer; and a method for the prevention or treatment of breast cancer, the method comprising delivery of a combination of (i) an inhibitor of BCL1 1A and (ii) p53, or an activator of p53 to an individual in need thereof.

Combinations may be provided as mixtures of active components, or may be provided as individual components for simultaneous or substantially simultaneous delivery, or sequential delivery, as appropriate.

The activator of p53 may be an indirect activator, for example may be an inhibitor of an inhibitor of p53, such as an inhibitor of mdm2.

Suitable activators of p53 include cis-imidazoline analogs Nutlins which inhibit the interaction between MDM2 and p53 and thus activate p53 (Vassilev et al., 2004), PRIMA-1 and related compounds, (Lambert et al., 2009).

The disclosure herein relating to the parameters relating to inhibitor of BCL1 1 A applies equally to the combination with p53 or an activator thereof.

We have also now identified transcription factor Bcl1 1 a to be an essential factor in HSCs. Hematopoietic stem cells (HSCs) are multipotent stem cells in the bone marrow that give rise to all blood cell types in the mouse and human. They have self-renewal capability and are differentiated to functional blood cells in a stepwise manner. On the other hand, leukemia stem cells have similar properties of normal HSCs. Therefore, studying HSCs has been an intensive and competitive research field for many years, and a number of genes have been implicated in producing or maintaining HSCs in the bone marrow. Bcl1 1a is expressed in long-term HSCs. Deletion of Bcl1 1a in the mouse bone marrow causes complete depletion of long-term HSCs and loss of all blood cell types. These results are similar to the essential role of Bcl1 1a in mouse mammary stem cells. Since inactivating BCL1 1 A causes breast cancer cell death, we expect that inactivating BCL1 1 A will likely cause cell death of leukemia or lymphoma which express BCL1 1 A. Bcl1 1a is also expressed in other tissues (Figure 4). Developing inhibitors for BCL1 1A, or for the pathways it is involved in, or for the factors acting upstream or downstream of BCL1 1 A, could be potentially beneficial to not only breast cancer patients but patients with other types of cancer.

Accordingly the present invention also relates to uses of bcl1 1a inhibitors as described herein in cancer applications more generally, beyond breast cancer, and includes in particular cancers in which bell 1a is over-expressed with respect to normal cellular levels, and may include, inter alia, leukemias or lymphomas.

It will be understood that particular embodiments described herein are shown by way of illustration and not as limitations of the invention. The principal features of this invention can be employed in various embodiments without departing from the scope of the invention. Those skilled in the art will recognize, or be able to ascertain using no more than routine study, numerous equivalents to the specific procedures described herein. Such equivalents are considered to be within the scope of this invention and are covered by the claims. All publications and patent applications mentioned in the specification are indicative of the level of skill of those skilled in the art to which this invention pertains. All publications and patent applications are herein incorporated by reference to the same extent as if each individual publication or patent application was specifically and individually indicated to be incorporated by reference. The use of the word "a" or "an" when used in conjunction with the term "comprising" in the claims and/or the specification may mean "one," but it is also consistent with the meaning of "one or more," "at least one," and "one or more than one." The use of the term "or" in the claims is used to mean "and/or" unless explicitly indicated to refer to alternatives only or the alternatives are mutually exclusive, although the disclosure supports a definition that refers to only alternatives and "and/or."

Throughout this application, the term "about" is used to indicate that a value includes the inherent variation of error for the device, the method being employed to determine the value, or the variation that exists among the study subjects.

As used in this specification and claim(s), the words "comprising" (and any form of comprising, such as "comprise" and "comprises"), "having" (and any form of having, such as "have" and "has"), "including" (and any form of including, such as "includes" and "include") or "containing" (and any form of containing, such as "contains" and "contain") are inclusive or open-ended and do not exclude additional, unrecited elements or method steps

The term "or combinations thereof as used herein refers to all permutations and combinations of the listed items preceding the term. For example, "A, B, C, or combinations thereof is intended to include at least one of: A, B, C, AB, AC, BC, or ABC, and if order is important in a particular context, also BA, CA, CB, CBA, BCA, ACB, BAC, or CAB. Continuing with this example, expressly included are combinations that contain repeats of one or more item or term, such as BB, AAA, AB, BBC, AAABCCCC, CBBAAA, CABABB, and so forth. The skilled person will understand that typically there is no limit on the number of items or terms in any combination, unless otherwise apparent from the context.

All of the compositions and/or methods disclosed and claimed herein can be made and executed without undue experimentation in light of the present disclosure. While the compositions and methods of this invention have been described in terms of preferred embodiments, it will be apparent to those of skill in the art that variations may be applied to the compositions and/or methods and in the steps or in the sequence of steps of the method described herein without departing from the concept, spirit and scope of the invention. All such similar substitutes and modifications apparent to those skilled in the art are deemed to be within the spirit, scope and concept of the invention as defined by the appended claims.

Reference to any agent herein for use in the treatment of breast cancer, or other disease, relates also to the use of that agent in the preparation of a medicament for the prevention or treatment of that disease.

Example 1 : Essential function of BCL11A in mammary stem cells

Breast cancer affects an estimated one million women worldwide each year with more than 400,000 dying from the disease [1]. Identification of mouse mammary stem cells (MaSCs) has facilitated understanding of the mammary epithelial lineages and the cellular origin of breast cancer [2] [3]. Recent studies have implicated Notch signalling [4, 5], steroid hormone signalling[6, 7] and the Epithelial to Mesenchymal Transition (EMT)[8-10] in MaSC function. However, transcriptional control in adult MaSCs still remains poorly defined. Bcl1 1 a is a C2H2 zinc finger transcription factor that plays critical roles in lymphocyte development [1 1] and in silencing fetal hemoglobin expression in adult erythrocytes [12]. Here, we report that Bcl1 1a has essential roles in both normal mouse MaSCs and human breast cancer stem cells. Deletion of Bcl11a in the virgin female mouse abolishes the ability of MaSCs to reconstitute a cleared mammary fat-pad and depletes MaSCs. Bcl1 1a transcriptionally regulates Notchl , Notch3 and key EMT-inducing transcription factors. In the human, BCL11A is highly expressed in basal breast cancer cell lines, and in the basal subset of the NKI-295 patient cohort[13]. Knock-down of BCL11A causes either apoptosis or differentiation of cancer stem cells with dramatic down- regulation of key EMT inducers and over-expression of miR-200c, a known sternness inhibitor [14], and reduces the development of tumour. Besides its critical roles in MaSCs, Bcl1 1a is required for the luminal progenitors from which some basal breast cancers are thought to originate[15, 16]. The essential roles of BCL1 1A in mammary cells thus provide novel insights into breast cancer and have important implications for therapies.

The mammary epithelium is composed of two main cell types: luminal cells, which line the ductal and alveolar lumena, and the myoepithelial or basal cells, which line the basal surface of the luminal cells and interact with the stroma [17]. Both types of cells are thought to arise from a multi-potent progenitor or stem cell population that has recently been characterized [2, 3, 18]. The developmental hierarchy of the mammary epithelium resembles hematopoiesis whereby stem cells give rise to progressively restricted progenitors that ultimately form various functional hematopoietic cells. Indeed, several studies have already demonstrated that genes involved in lymphoid lineage development are important in the mammary gland [19-21]. The C2H2 zinc finger transcription factor family Bcl1 1 has critical roles in several hematopoietic lineages. Bcl1 1a is essential for B cell development [1 1] and for repressing fetal hemoglobin expression in adult erythocytes [12], whereas Bcl1 1 b was recently found to be required for T cell development and identity maintenance [22]. These observations, together with the fact that BCL11A mutations were identified in human breast cancer [23], prompted us to investigate the functions of Bcl1 1a in the mammary gland.

We first examined whether Bcl1 1a was expressed in the mammary gland using the lacZ knock-in mouse line, Bcl11a^{lacZ +} (Supplementary Fig. 1 a). X-gal staining of the /acZ-reporter embryos revealed that Bcl1 1 a was among the earliest transcription factors that were specifically expressed in the mammary placodes (Supplementary Fig. 1 b-c). At puberty, Bcl11a was expressed in the cap cells of the terminal end buds (TEBs), a region thought to harbour stem cells [17] (Fig. 1a, panel i). In the mature virgin gland, expression of Bcl11a was found in the subtending ducts and in a small number of both luminal and basal cells (Fig. 1a, panel ii, and Supplementary Fig. 1d-e). The mammary epithelial cells of the Bcl11a^{lacZ +} mice were subsequently stained with fluorescein di" -D-galactopyranoside (FDG), which detects β-galactosidase activity in flow cytometry at the single cell level [24]. FDG cells (Bc/77a-expressing) were found in both luminal and basal compartments (Fig. 1 b), which was confirmed by quantitative (q) RT-PCR analysis (Supplementary Fig. 11). During mammary gland ontogeny, Bcl1 1 a exhibited a dynamic expression pattern (Fig. 1 c). Expression of Bcl11a was first markedly increased in early gestation (Supplementary Fig. 1f-i) and was then decreased in the lactating gland (Supplementary Fig. 1j- k). The MaSCs-enriched population is the CD24^medCD49f^hi basal epithelial cells [2]. Remarkably, Bc/77a-expressing basal cells (FDG⁺) were almost exclusively CD24^medCD49f^hi (Fig. 1 b), which was confirmed by qRT-PCR using purified CD24^medCD49f^hl basal cells from the wild type mouse (Fig. 1 d) and was consistent with the X-gal staining in the TEBs (Fig. 1 a). Therefore, Bcl1 1a was expressed in the MaSCs-enriched mammary epithelial cell population.

We next made the Bcl11a conditional knockout (CKO) mice (Supplementary Fig. 2) since the conventional Bcl11a knockout allele caused neonatal lethality [1 1]. The CKO mice were crossed to the Rosa26-CreERT2 mice for inducible Cre recombinase activation in almost all mouse cells [25]. The CreERT2; Bcl11a^f'^{o /f}'°^x mice (referred to as flox/flox in this work) appeared overtly normal and produced normal litter sizes. To investigate Bcl1 1a function in the mammary gland, we injected the virgin female mice with tamoxifen to induce Bcl11a deletion and harvested mammary epithelial cells 1-3 weeks later. Injection of tamoxifen to the control mice (flox/+) did not cause any obvious defects in mammary epithelial cells (Fig. 2a). However, an obvious reduction of total mammary epithelial cells was noticed in the flox/flox mice (data not shown). At the single cell level, basal cells in the Bc/77a-deficient gland appeared to be depleted (Fig. 2a); in particular, the MaSC fraction (CD24^medCD49f^hi) was essentially absent (Fig. 2a). Although some luminal cells were still present in Bc/77a-deficient gland detected by flow cytometry (Fig. 2a), immuno- histological analysis as well as qRT-PCR revealed a drastic reduction in expression of both basal (CK14 and p63) and luminal genes (CK18, Gata-3, Elf5, Mud) (Supplementary Fig. 3, 4d-g). Besides depletion of MaSCs, the mutant gland had distended primary ducts and reduced secondary branching in the virgin gland (Supplementary Fig. 4a-b). Histological examination further revealed that unlike the bi-layered structure in the control gland, ductal architecture was perturbed with intercalation of the luminal and basal/myoepithelial layers in the mutant gland (Supplementary Fig. 4c).

To confirm the MaSC depletion in the Bc/77a-deficient mammary gland and to demonstrate that the defect was cell-autonomous, we harvested basal cells from flox/flox mice one week after tamoxifen injection when the MaSC population was still present (Supplementary Fig 4j), and transplanted them at limiting dilution into cleared fat pads of NOD/SCID/IL2rY^_/" female mice [26]. The injected fat pads were examined for re-constitution of the gland after the recipients were made pregnant between 3-6 weeks later. Transplantation of Bc/77a-deficient basal cells did not result in any outgrowths while the same numbers of control cells (wild-type or flox/+ mice treated with tamoxifen) gave rise to robust outgrowths (Fig. 2b). On two occasions, outgrowths at the 1000 cell dose of Bc/77a-deficient basal cells were found. Genotyping of these outgrowths demonstrated that the mammary cells did not have a complete Bcl11a deletion (Supplementary Fig. 4k), due to incomplete Cre-/oxP recombination in the tamoxifen-treated flox/flox mice. This result thus confirmed that only the MaSCs which expressed Bcl11a were able to engraft the fat pads and that Bcl1 1a was required for MaSC survival and self-renewal.

To probe global gene expression changes in the MaSCs following Bcl11a deletion, we performed microarray analysis of the MaSC-enriched population from the control and the Bc/77a-deficient mammary gland one-week post tamoxifen injection. PCR genotyping of the MaSCs from the Bc/77a-deficient mammary gland did not detect the cko allele, which confirmed complete Bcl11a deletion in these cells (Supplementary Fig. 4I). Statistical analysis identified 4800 genes that had significant expression changes upon Bcl11a deletion. We compared this gene list to a common signature of genes that are enriched in and shared between mouse and human MaSCs [10]. Many of these, for instance Itgb4, Mmp2, Apoe, C0H6AI, were down-regulated in the Bcl11a- deficient MaSCs (Fig. 2c), consistent with loss of MaSC activity. On the other hand, expression of genes implicated in differentiation such as Glycaml, Lmo4, Ndrgl was increased upon Bcl11a deletion.

Several recent studies have demonstrated that transcription factors involved in the EMT promote production of stem cell-like cells from both primary and immortalized mammary cells [8, 10]. Interestingly, many of the most down-regulated genes in the Bc/77a-deficient MaSCs, such as Snai3, Twist2, Mmp2, Mmp3, Mmp14, are implicated in the EMT (Fig. 2c-d). qRT-PCR analysis of individual genes further confirmed that expression of two key EMT transcription factors, Zeb1 and Zeb2 (Sip1 ), which repress E-Cadherin expression [27, 28] and promotes sternness of cancer cells [29], and of Vimentin, was reduced upon Bcl11a deletion (Fig. 2d). Therefore, loss of expression of key EMT genes was associated with Bcl11a deficiency and the depletion of MaSCs.

Notch signalling is suggested to have important roles in the MaSC compartment[4, 30]. Specifically, lower canonical Notch signalling causes expansion of the MaSC population, while higher Notchl signalling promotes differentiation of stem cells to the luminal lineage[4, 30]. On the other hand, Notch3 is expressed and appears to be required in MaSCs[5, 31]. We showed previously that Bcl11a deficiency caused abnormal Notch signalling in T cells [1 1]. Therefore, we analyzed whether Bcl11a deletion in the mammary gland would affect Notch expression. A substantial up-regulation of Notchl receptor and its ligand Jaggedl (Fig. 2d, Supplementary Fig. 4h-i) was found upon Bcl11a deletion. This was concomitant with the dramatic down-regulation of Notch3 (Fig. 2d). Therefore, similarly to in T cells, Bc/77a-deficiency also caused abnormal Notch signalling in the mammary epithelial cells, which also likely contributed to the defects in the mutant mammary gland. qRT-PCR on various epithelial compartments isolated by FACS sorting using previously describe cell surface markers (Stingl et al., 2006) revealed that Bcl1 1a expression was higher in the luminal progenitor, basal and MaSC-enriched populations (Figure 2E).

To investigate the consequence of over-expressing Bcl11a, we made the MSCV-Bcl1 1a-IRES- EGFP vector, where Bcl11a expression was controlled by the LTR of MSCV, and transfected this vector into the mouse luminal mammary epithelial cell line EpH4 [32] that only expresses low endogenous levels of Bcl11a. The control and Bc/77a-overexpressing cells (GFP⁺) were sorted and embedded at a low density in matrigel to assay for mammosphere formation. Over- expression of Bcl11a led to a significant increase in the number of mammospheres, which were noticeably larger (Fig. 3a). Remarkably, in the cells over-expressing Bcl11a, basal cell markers such as CK14 and p63 were induced in these otherwise luminal cells (Fig. 3b). Immuno-staining confirmed that fewer cells were CK18⁺ luminal cells (Supplementary Fig. 5). Moreover, in contrast to Bcl11a deficiency where EMT genes were down, in the Bc/77a-expressing GFP⁺ cells, key EMT transcription factors, Foxc2, Twist2, Zeb1 and Zeb2, as well as Notch3 were dramatically up-regulated. Furthermore, Bcl11a expression reduced expression of E-Cadherin and Notchl (Fig. 3b). Loss and gain of function studies in the mouse mammary epithelial cells thus clearly suggest that Bc1 1 1a differentially affects expression of Notch genes and key EMT transcription factors. Moreover, in silico examination of the Notchi, Notch3 Foxc2, Twist2 and Zeb1 genomic loci identified putative Bcl1 1a binding sites (GGCCGG) [33] (Fig. 3c). We thus performed ChIP assays to determine whether Bcl1 1a directly regulates these genes. Binding of Bcl1 1a to several of the consensus-binding sites at Notchi (regions i and ii), Notch3 (regions iii and iv), Foxc2 (region v), and Zeb1 (region viii) loci is shown in Fig. 3d, where the genomic DNA fragments harbouring these putative binding sites were preferentially pulled down by the Bcl1 1a antibody. The specific binding of Bcl1 1a to these sites was further confirmed using two additional Bcl1 1a antibodies (data not shown). Interestingly, despite the clear down-regulation of Twist2 when Bcl11a was deleted and the dramatic up-regulation of Twist2 in the Bcl11a over-expressing cells, no detectable Bcl1 1a binding to the Twist2 locus (Region vi) could be found (Fig 3d). To further dissect regulation of these genes by Bcl1 1a, we assessed methylation state of two Histone H3 lysine (K) residues at these genomic loci either in the control or in Bc/77a-overexpressing EpH4 cells. H3K27-trimethylation is strongly associated with regions of gene silencing while H3K4 dimethylation marks active gene expression [34]. Bcl11a expression caused an increase in H3K4 dimethylation at the Notch3 (region iii), and Foxc2 (region v) loci, and a decrease of H3K27- trimethylation at the Foxc2 (region v) and Zeb1 (region viii) loci (Fig. 3e). Conversely, H3K27- trimethylation at the Notchi locus (region ii) was increased upon Bcl11a overexpression, indicating transcriptional repression at this locus (Fig. 3e). Although Bcl1 1a did not directly bind the Twist2 locus, an increase of H3K4-methylation at this locus was obvious in the Bcl1 la-over- expressed cells. This could be due to the EMT switch in these cells or perhaps failure to identify the true Bcl1 1a binding sites at this locus (Fig. 3e). These results thus demonstrated that Bcl1 1a directly regulates Zeb1, Foxc2, Notchi and Notch3 in mouse mammary epithelial cells.

Mouse and human Bcl1 1a are highly conserved proteins. To investigate the roles of BCL1 1A in human mammary epithelial cells, we first over-expressed BCL1 1A in immortalized human mammary epithelial cells (HMLE) [35]using a doxycycline inducible piggyback (PB) transposon vector to facilitate its integration into the genome by transposition [36]. Human mammary epithelial cells that have the CD447CD24^" antigenic phenotype are proposed to have stem cell properties [37] [8]. Over-expression of BCL1 1A increased CD447CD24^" cells from 1.5% to 6.0% (Supplementary Fig. 6a) and decreased CK18⁺ luminal cells (Supplementary Fig. 6b). Similar to that in the mouse cells, over-expression of BCL1 1A caused a 3-fold increase of Colony Forming Cells (CFC) and produced larger colonies, compared to using the vector control (Fig. 4a). At the molecular level, BCL1 1A over-expression increased expression of basal gene CK14, and the EMT inducers ZEB1, ZEB2, FOXC2, as well as VIMENTIN and NOTCH3 (Fig. 4b). miR-200c is a sternness inhibiting microRNA in pancreatic, colorectal and breast cancer cells [14, 29]. ZEB1 and miR-200c appear to negatively regulate each other[29, 38]. Concomitant with increased ZEB1 expression and more cancer stem cells upon BCL1 1A over-expression, miR-200c expression was reduced by about 50% (Fig. 4b).

We next studied possible involvement of BCL1 1A in breast cancer by examining BCL1 1A expression in 22 human breast cancer cell lines. qRT-PCR analysis showed that BCL1 1A expression was at much higher levels in the majority of basal breast cancer cell lines (Supplementary Fig. 6c). Encouraged by this finding, we analysed BCL1 1A expression in primary patient samples using the NKI-295 breast cancer dataset [13]. High levels of BCL1 1A expression were significantly correlated with ER negative status of the primary tumours (Fig 4c), with the "basal" subtype expression profile, and to a lesser extent, with the "normal" subtype expression profile (Fig. 4d).

We performed shRNA knock-down (KD) analysis to investigate whether BCL1 1A expression in basal breast cancer was functionally relevant. Delivery of shRNAs into cancer cells was achieved by using a piggyBac transposon vector (PB-H 1 -shRNA-GFP) into which the BCL1 1A shRNA oligos were cloned . KD of BCL11A using two independent shRNA vectors increased apoptosis in CAL120 and BT549 cells within 24-48 hours of transfection compared to the scrambled control shRNA (Fig. 4e). The cells which survived the expression of the BCL1 1A shRNAs (GFP⁺) appeared to have lost their mesenchymal morphology and gained a more differentiated cuboidal/epithelial morphology as soon as 6 days after transfection (Fig. 4f). The BCL11A-KD cells that survived were further analysed. Compared to the cells expressing the control shRNAs, the BCL11A-KD culture had many fewer CD447CD24^" stem-like cells and more differentiated cells (Fig. 4g). Molecular analysis confirmed that BCL11A KD substantially reduced expression of several key EMT transcription factors such as ZEB1, ZEB2, FOXC2 and TWIST2, and dramatically up-regulated the differentiation markers GATA-3, FOXA1, ERct and CK18 (Fig. 4h). Remarkably, in the BCL11A-KD cells, miR-200c expression was increased over 100-fold (Fig. 4h). Depletion of cancer stem cells and appearance of differentiated cells were confirmed by immunostaining as many more BCL11A-KD cells were stained positively for E-CADHERIN and CK18 (Fig. 4f). Similarly, knock-down of BCL11A in the normal breast epithelial cell line HMLE also led to more cells expressing E-CADHERIN and CK18 (Supplementary Fig. 6d). To demonstrate the effect of reducing BCL11A levels on tumor development, we injected GFP⁺ CAL120 cells transfected with either the control shRNA vector or the BCL11 A-shRNA vector into the immuno-compromised NSG mice. In vivo imaging of recipient mice showed that BCL11A-KD reduced the development of tumours (Supplementary Fig. 6e). Apoptosis or differentiation of breast cancer stem cells upon BCL11A knock-down thus demonstrated that BCL1 1A was required for breast cancer stem cell survival and self-renewal. The different outcomes of cancer cells, either apoptosis or differentiation, are likely due to the remaining levels of BCL1 1 A after the knock-down since the knock-down efficiency varied in individual cancer cells.

ER negative and basal classifications of breast cancer are highly correlated with poor prognosis[13, 39]. Two recent studies proposed that BRC \ 7-deficient basal breast cancers originate from luminal progenitor cells[15, 16]. Severe defects in Bc/77a-deficient luminal cells (Supplementary Fig. 3-4) indicate a likely role for Bcl1 1 a in luminal progenitors. We thus used CD49b and Seal expression to subdivide luminal cells into progenitors (CD49b⁺) and differentiated cells (CD49b^~) (Shehata, Stingl and Watson, manuscript in preparation), and analysed Bcl 1 1 a expression in luminal cells in the /acZ-reporter mouse. About 80% of the Bcl11a- expressing luminal cells (FDG⁺) were in the progenitor fraction with a majority of them being localized within the CD49⁺Sca1 ^" progenitor compartment (Supplementary Fig . 7a). Expression of Bcl1 1 a in luminal progenitors was confirmed by RT-PCR using FACS-purified cells. Differentiated luminal cells expressed lower levels of Bcl11a while the progenitors expressed higher levels of Bcl11a (Supplementary Fig. 7b). Mammary progenitor cells can be detected by their ability to form clonal progeny in vitro [2]. Consistent with its expression in the progenitors, a 6-fold enrichment for Mammary-Colony Forming Cells (Ma-CFCs) was found in Bc/77a-expressing epithelial cells (CD24^hlFDG⁺) compared to the non-expressing cells (Supplementary Fig. 7c). The role of Bcl 1 1 a in luminal progenitors was further investigated in a loss of function assay. Total mammary epithelial cells from the flox/flox mice were assayed for Ma-CFCs one week after tamoxifen injection. Deletion of Bcl11a significantly diminished Ma-CFCs (Supplementary Fig. 7d).

Gata-3 and Elf5 are two key transcription factors in luminal progenitors [19, 20, 40], and Elf5 is considered to be a master regulator of the alveolar switch [40]. Elf5 was expressed at higher levels in the CD49⁺Sca1^" luminal progenitors, similar to Bcl11a (Supplementary Fig. 7b). Bcl11a deletion caused a complete loss of Elf5 expression, and to a slightly less extent, of Gata3 (Supplementary Fig. 3). Careful examination of the Bc/77a-deficient mammary gland revealed that Bcl11a deletion had differential effects on ERoc⁺ and ERoc^" luminal cells. The Bc/77a-deficient mammary gland had relatively higher percentage of ERoc⁺ cells (49.0 + 8.8% vs 25.0 + 7.1 % in the control) (Supplementary Fig. 7e-f), suggesting that Bcl11a deficiency preferably depleted ERoc^" luminal cells, similar to deletion of Elf5 [40]. Therefore, Bcl1 1a has key roles in luminal progenitors, in particular in those that gives rise to ERoc^" luminal cells.

We report here that Bcl1 1a is a key transcription factor required in both MaSCs and breast cancer stem cells. Bcl11a deficiency caused depletion of MaSCs in the mouse mammary gland and loss of luminal progenitors. Bcl1 1a functions in MaSCs and the mammary epithelium by regulating EMT genes and Notch signalling. Two recent studies suggest that some basal tumours could have originated from luminal progenitors[15, 16]. Based on the critical roles of Bcl1 1a in MaSCs and in Elf5-expressing luminal progenitors, its ability to enhance sternness and to induce basal gene expression in luminal cells, it is tempting to speculate that some Bc/77a-overexpressing luminal progenitors might represent the abnormal cells from which basal tumours eventually develop, and that consequently, the basal cancer stem cells would still require BCL1 1A for their survival and self-renewal. This scenario is reminiscent of transformation from committed myeloid progenitor to leukaemia stem cell [41]. Further investigation of BCL1 1A in breast cancer is thus required, which would ultimately lead to better understanding of this disease and more effective therapies.

Example 2

BCL11A is Specifically Highly Expressed in Basal Breast Cancer

To identify genes that are important in breast cancer, we compiled a list of 32 genes encoding transcription factors known to be important in hematopoiesis, and examined their expression across different subtypes of breast cancer. Using a publically available microarray dataset [42], we found that out of these 32 chosen genes, BCL11A was the only one differentially and highly expressed in Basal and Claudin-low subtypes of breast cancer (Figure 5A). This is in sharp contrast to GATA3, which is highly expressed in luminal subtypes (Figure 5A) and is a known prognosis marker for these subtypes [43]. Following this initial discovery, we subsequently reexamined five additional published microarray datasets [13, 44-47] which are available in the ICR ROCK database [48]. Again, in all these published datasets, a similar pattern of BCL1 1A expression was observed (Figure S8A). We next analysed BCL1 1A expression using quantitative RT-PCR (qRT-PCR) in 22 established breast cancer cell lines. High levels of BCL1 1A were primarily found in basal breast cancer cell lines (Figure 5B).

High Levels of BCL11A Enhance Self-Renewal of Mammary Epithelial Cells

The above loss of function studies in the mouse showed that Bcl1 1 a has essential functions in mammary stem cells/progenitors. To address whether high levels of Bcl1 1a are sufficient to confer or enhance self-renewal of mammary epithelial cells, we made a piggyBac transposon- based BCL1 1A overexpression vector, PB-TRE-BCL1 1A, which carries a puromycin selection cassette and also allows for tight and inducible control of BCL1 1A cDNA expression (Figure S9A). piggybac DNA transposition enables convenient and efficient delivery of genetic material to the genome of cultured mammalian cells [36]. Mammaosphere assays act as a surrogate readout for stem cell self-renewal and function in vitro [49]. We transfected immortalised mouse (EpH4) [50] and human (HMLE) mammary cell lines [35] with the PB-TRE-BCL1 1A vector and a PB transposase transient-expressing plasmid. The stable transgenic lines were established using puromycin selection. BCL1 1A expression was induced in these cells using doxycycline. EpH4 cells were generated from spontaneously immortalised mid-pregnancy mammary epithelial cells [50], whereas HMLE cells were immortalised by expression of SV40 large T antigen and hTert [35]. EpH4 cells overexpressing Bcl1 1a (EpH4-1 1A) induced higher expression levels of basal genes (K14, p63) (Figure S9B). Compared to the control EpH4 cells, these transgenic cells formed larger mammospheres in Matrigel and at a higher rate (Figure 6A). Similarly, compared to the parental cells, HMLE cells expressing the transgenic BCL1 1A (HMLE-1 1A) also showed almost 100% increase in the number and size of primary and secondary mammospheres (Figure 6B).

HMLER cells are KRAS transformed HMLE cells [35] and have higher mammosphere forming ability in culture. We thus asked the question whether similar to deleting Bcl11a in mouse mammary epithelial cells, knockdown of BCL1 1A in HMLER (HMLER-sh-1 1A) cells would affect mammosphere formation. For this purpose, we generated a piggyBac shRNA knockdown vector (PB-sh-BCL1 1A-GFP) using previously published shRNA sequences [12]. The GFP⁺ cells were harvested for further analysis by flow sorting and showed successful knock down of BCL1 1A (Figure S1 1 D). Decreased levels of BCL1 1A in HMLER-sh-1 1A cells substantially reduced primary and secondary mammosphere numbers (Figure 6C). Besides the numbers, the mammospheres from HMLER-sh-1 1 A cells were much smaller in size and appeared to be disorganised (Figure 6C). These in vitro mammosphere formation and self-renewal results, as well as the in vivo loss of function studies in the mouse, demonstrate that Bcl1 1a is a master regulator in mammary stem cells.

Overexpression of BCL11A Causes Mammary Tumour Development

BCL1 1A was discovered as a proviral intergration site in mouse myeloid leukaemia and has the ability to transform NIH-3T3 cells [51]. In human B cell lymphoma patients BCL1 1A was ectopically expressed due to chromosomal translocation [52]. To explore the possibility that BCL1 1A overexpression can lead to mammary tumour development, we injected EpH4-1 1A and EpH4-control cells orthotopically in contralateral cleared mammary fat-pads of NSG mice. Mice were fed with doxycycline containing diet and periodically monitored for tumours. Within 6 weeks of the injections, palpable tumours appeared (Figure 6D). The tumours from EpH4-1 1 A cells were much larger compared to those formed from control cells (Figure 6D) and displayed more basal characteristics (Figure 6E). These EpH4-1 1A tumour cells stained strongly for vimentin, CK14 and weakly for CK18 compared to the control tumours (Figure 6E). This is consistent with the ability of Bcl1 1a to induce expression of K14 and p63 expression in cultured EpH4 cells (Figure S9B).

The immortalised human breast epithelial HMLE cells injected subcutaneously in immune compromised mice do not form tumours [35]. To assess if BCL1 1 A overexpression is sufficient to transform HMLE cells, we injected HMLE-1 1A cells subcutaneously in NSG mice. Three out of four mice injected with HMLE-1 1A cells formed tumours within 8 weeks of injection compared to zero out of four for the HMLE-control cells (Figure 6F). Immunohistological analysis revealed that HMLE-1 1A tumours also displayed a basal tumour characteristics because many tumour cells stained strongly for Vimentin and CK14 but weakly for CK18 (Figure 6G). To further confirm the basal breast cancer nature of these tumours induced by BCL1 1A, we performed global gene expression analysis, and compared their expression profiles to the tumours in the METABRIC dataset. As shown in Figure S10A, these three tumours were clearly classified into the Basal-like subtype.

Unlike HMLE, HMLER (KRAS transformed) can develop tumours when injected subcutaneously in immune compromised mice [35]. We have shown in the above experiments that knocking down BCL1 1A in these cells greatly reduced mammosphere formation (Figure 6C). To find out whether BCL1 1A knockdown would also affect tumour formation of these cells, we injected one million HMLER-sh-control and HMLER-sh-1 1 A cells subcutaneously into NSG mice and measured tumour size periodically. Initially, HMLER-sh-control and HMLER-sh-1 1 A cells developed a small mass but the HMLER-sh-1 1 A tumours gradually diminished in size to become essentially undetectable within 25 days (Figure 6H), consistent with loss of mammosphere formation capability of these cells. In contrast, the tumours from HMLER-sh-control cells continued to grow for 3 months till the experiment was terminated (Figure 6H).

To further demonstrate the important roles of BCL1 1 A in cancer cells, we transfected the PB-sh-BCL1 1A-GFP and the control PB-sh-scramble-GFP vectors into breast cancer cell lines CAL120 which expressed high levels of BCL1 1A (Figure 5B). GFP⁺ cells were examined and a higher rate of apoptosis was found within 48hrs in both cell lines when BCL1 1A was knocked down (Figure S10B). The CAL120 cells that survived the knock-down continued to proliferate but appeared to gradually gain a more differentiated epithelial-like phenotype (Figure S10C). Moreover, genes associated with luminal cells, such as GAT A3, K18, FOXA1 , ERa and E- cadherin, were now highly expressed in these more differentiated BCL1 1A knockdown CAL120 breast cancer cells (Figure S10D). Although the CAL120-sh-1 1A knockdown cells kept proliferating, they had an attenuated capability to form mammospheres. This is consistent with the fact that survived CAL120-sh-1 1A KD cells lost their stem cell ability and became differentiated (Figure S10E).

Tumour development from overexpression of BCL11A in both mouse and human mammary epithelial cells and the suppression of tumour development when BCL11A is knocked down demonstrate that BCL11A is likely a breast cancer gene.

BCL11A Regulates Genes Implicated in Stem Cell Function and Tumourigenesis

The critical roles of BCL1 1A in mouse mammary stem cells and in breast cancer cells prompted us to investigate how Bcl1 1 a mediates its effects in these cells. We first performed a microarray analysis using FACS-purified MaSC populations (CD24med/CD49hi) harvested from the control and flox/flox mice one-week after tamoxifen injection. Statistical analysis of the expression data identified 4,800 genes that had significant expression changes upon Bcl11a deletion. We compared this gene list to a common signature of genes that are enriched in and shared between mouse and human MaSCs [10]. In the Bc/77a-deficient MaSCs, many of these genes, for instance Itgb4, Mmp2, Apoe, Col16A1, were down-regulated (Figure 7A), consistent with loss of

MaSC activity. On the other hand, expression of genes implicated in differentiation such as

Glycaml, Lmo4, Ndrgl was increased. Conversely, overexpression of BCL1 1A in HMLE cells upregulated many MaSC signature genes (Figure 7A). Expression of several known key stem cell genes was also increased in HMLE-1 1A cells such as CD47 and MLL, which are essential in hematopoiesis and hematopoietic stem cells (HSCs) [53, 54]. Additionally, HMLE-1 1A cells had increased expression of several proto-oncogenes for instance GPR87, MIR21 , KRAS and

NEDD9 [55-58] but reduced expression of tumour suppressor genes such as CASP8, RASSF4 and AIM2 [59-61] (Tablel ).

Closer examination of microarray expression data in the Bc/77a-deficient MaSCs revealed that several key genes involved in EMT were down-regulated. These genes included

Snai3, Twist2, Mmp2, Mmp3, Mmp14. qRT-PCR analysis of additional genes further confirmed that Bcl1 1 a deficiency caused lower expression of other key EMT transcription factors namely,

Foxc2, a known basal breast cancer gene [62], Zeb1 and Zeb2, which repress E-Cadherin expression and promote sternness of cancer cells [29, 63] (Figure 7B). It is known that expression of EMT transcription factors promotes production of stem cell-like cells from both primary and immortalized mammary cells [8]. Direct correlation of loss of Bcl1 1a in MaSCs and decrease expression of EMT transcription factors indicated that Bcl1 la's function in MaSCs might be partly through regulating these EMT genes. We thus examined the expression of EMT genes in EpH4- 1 1A and HMLE-1 1A cells. In both cases, high levels of Bcl1 1a increased expression of EMT inducing genes including Foxc2, Zeb1 and Zeb2 (Figure 7C-D). We next examined the consequence of BCL1 1A knockdown in the breast cancer cell line CAL120 for the EMT changes. As described above, BCL1 1A knockdown caused cells death of many cells. However, the surviving cells (CAL120-sh-1 1A) still proliferated but appeared to be differentiated and had much lower mammosphere-forming potential (Figure S5). CAL120-sh-1 1A cells stained strongly for E- cadherin and the luminal marker CK18, indicating a possible mesenchymal to epithelial transition (MET) (Figure 7F). Consistent with the notion of MET in CAL120-sh-1 1A cells, expression of FOXC2, ZEB1 and ZEB2 was drastically reduced (Figure 7G). In addition, Mir-200c plays a negative regulatory role in MaSC [14] and inhibits expression of EMT inducers [38]. We found that BCL1 1A expression in HMLE-1 1A cells suppressed Mir-200c expression whereas BCL1 1A knockdown in CAL120-sh-1 1A cells increased Mir200c expression by a 100-fold (Figure 7D,G).

BCL1 1 A is a C2H2 zinc finger transcription factor with a consensus binding site proposed as GGCCGG [33]. In silico analysis identified BCL1 1A binding sites on the human and mouse Foxc2 and Zeb1 loci (Figure 7H). To confirm if these sites are bona fida BCL1 1A binding sites, we performed chromatin immunoprecipitation (ChIP) using an anti-BCL1 1 A antibody to pull down genomic DNA fragments from cell extracts of EpH4-1 1 A cells cultured in the presence or absence of doxycyclin. An approximate 3-fold enrichment was found in the 5^'UTR of Zeb1 and in the Foxc2 promoter (-1 , 100bp) compared to isotype controls (Figure 7I), indicating the direct regulation of these two genes by BCL1 1A. Chromatin modifications play an important role in the control of gene expression [34]. To further decipher regulation of Zeb1 and Foxc2 genes by Bcl1 1a, we studied methylation state of two Histone H3 lysine (K) residues at Zeb1 and Foxc2 loci in EpH4-1 lAcells by repeating the ChIP experiments using an antibody to either H3K27-3me or H3K4-3me. H3K27-trimethylation (H3K27-3me) is strongly associated with regions of gene silencing while H3K4 trimethylation (H3K4-3me) marks actively transcribed loci [34]. Expression of Bcl1 1a substantially increased H3K4-3me at the Zeb1 locus concomitant with a decrease of H3K27-3me. Although Bcl1 1a expression did not alter the levels of H3K4-3me at the Foxc2 locus, H3K27-3me levels were reduced (Figure 7JK). These results thus further support that BCL1 1A binds to these two loci and induces chromatin modifications to promote EMT.

Taken together, BCL1 1A functions in mammary stem cells and in breast cancer cells at least partly through modulating the EMT pathway.

[Figure 7 I, J and K show groups of 4 columns that reflect the order of the legend of the graph, i.e .Fig 7I:

IgG - dox;

IgG + dox;

bell 1a - dox;

bcl; 1 1a + dox. The same principle is applied in Figure 8]. BCL11A Negatively Regulates p53 Activities

Besides EMT genes, global gene expression analysis of the HMLE-1 1A cells also showed that several factors that negatively regulate p53, namely, MDM2 and PSMD10, were up-regulated (Figure 8A). This observation was subsequently confirmed by qRT-PCR (Figure S1 1A). On the other hand, in Bc/77a-deficient mouse basal cells, expression of Mdm2 and Mdm4 was reduced by approximately 50% (Figure 8B). p53 is a critical player in cell cycle, apoptosis and genome integrity and is estimated to be mutated in more than 50% of all tumours [64]. Moreover, many of the of BLBC cases have p53 mutations [65]. The E3 ubiquitin ligase MDM2 is a key negative regulator of p53 levels and activity. A delicate balance between these two proteins is critical for normal tissue homeostasis [66]. In the mammary epithelium, p53 suppresses MaSC activities [67] and negatively regulates EMT [68]. It is unclear, however, how p53 itself is regulated in MaSCs. Decreased expression of Mdm2 and Mdm4 upon Bcl11a deletion hints that Bcl1 1a could regulate p53 activities in MaSCs and in breast cancer.

To address this possibility, we asked whether Bcl11a overexpression could compromise p53 function in response to cellular stress. EpH4-1 1A cells and EpH4-Control cells were exposed to ultraviolet (UV) light which induces DNA damage and hence p53 activation. In the EpH4- control cells, total p53 protein levels began to accumulate within 6 hours of UV treatment and most cells were dead within 48 hours (Figure S1 1 B). Five days after UV irradiation, no surviving cells were observed (data not shown). In contrast, EpH4-1 1A cells failed to accumulate p53 protein and many more cells survived the UV irradiation (Figure 8C and Figure S1 1 B). Similar observations were made when EpH4-1 1A and HMLE-1 1A cells were treated with the DNA damaging agent Doxorubicin (Figure 8D and Figure S1 1 C). Moreover, accompanied with high levels of BCL1 1A expression in both EpH4-1 1A and HMLE-1 1A cells, MDM2 expression increased (Figure 8D and Figure S1 1 C). In contrast to BCL1 1A overexpression, BCL1 1A knockdown in HMLER cells had drastically reduced MDM2 protein levels (Figure S1 1 D). In these cells, expression of p21 , a target gene of p53 responsible for cell cycle arrest, was substantially increased (Figure S1 1 D).

As described above, BCL1 1A knockdown in breast cancer cell lines, CAL120 and caused apoptosis and cell differentiation (Figure S10). However, knocking-down BCL11A using the same vector did not cause obvious changes in the mammosphere assay in another breast cancer cell line, MDA-MB-231 (MDA231 ), which has high levels of BCL1 1A (Figure 5B). One possibility is that p53 is mutated in MDA231 cells so that modulating BCL1 1 A was no longer able to affect p53 activities. We thus investigated the p53 status in this cell line from the COSMIC database [69], which shows MDA231 cells have a point mutation in TP53 gene causing Argenine 280 to be replaced with Lysine (R280K). This mutation significantly hinders p53 transcriptional activity and promotes invasion and metastasis [70]. To confirm that the ineffectiveness of BCL1 1A knockdown in MDA231 cells was due to the p53 mutation, we transfected MDA231 with a wild- type (WT) p53-expressing plasmid, and co-transfected with either the control shRNA vector or the PB-sh-BCL1A-GFP vector. The GFP⁺ cells were first analysed in mammosphere assays. Only in cells where BCL1 1A was knocked down and also expressed WT p53 did we see a significant reduction in the number of mammospheres forming (Figure 8E). In cells treated with Doxorubicin, high levels of p21 were found only in those cells that had both BCL1 1A knockdown and expressed WT p53, thus confirming the unhindered p53 activity in these cells (Figure 8F). Therefore, p53 appears to act epistatically downstream of BCL1 1A in breast cancer cells, possibly through BCL1 1 A regulation of MDM2.

To confirm the regulation of MDM2 by BCL1 1A, we performed in silico analysis of the Mdm2 locus and identified a cluster of Bcl1 1a binding sites in the 5^'UTR and intron 1 which are conserved between the human and mouse MDM2/Mdm2 loci (Figure 8G). We then performed a ChIP assay using EpH4-1 1A cells cultured in the presence or the absence of doxycyclin using a Bcl1 1a antibody to pull down genomic DNA fragments. A 5-fold enrichment for the Mdm2 promoter region was shown in Bcl1 1a-expressing cells vs. the control cells (Figure 8H). Moreover, EpH4-1 1a cells displayed a 4-fold increase of the H3K4-3me level and slightly lower levels of H3K27-3me (Figure 8I-J). Therefore, Bcl1 1a negatively regulates p53 in mammary epithelial cells by transcriptionally activating Mdm2.

Discussion: BCL1 1 A is a Basal Breast Cancer gene.

BLBC accounts for 50-70% of all Triple Negative Breast Cancer cases. Such cases are characterised by negative staining for hormone receptors and normal or low expression levels of the Her-2 receptor [65, 71 , 72]. As such, these types of tumours do not respond to conventional therapeutic regimes such as, the anti-hormonal drug, tamoxifen, or the monoclonal antibody against the Her-2 receptor, Trastuzumab. This leaves conventional chemotherapy as the only therapeutic option for these tumours. Consequently, BLBC has the worst prognostic outcome compared to other breast cancer subtypes.

We have shown here that the C2H2 transcription factor BCL1 1A is highly expressed in BLBC. In the mouse, Bcl1 1a was expressed predominantly in MaSCs in the basal compartment, and also in many luminal progenitors. Bcl1 1 a deletion depleted MaSCs and caused a loss of luminal progenitors cells. By contrast, overexpression of BCL1 1 A increased the number of stem-like cells in culture and was sufficient to transform mouse and human mammary epithelial cells in xenograft transplantation models. Conversely, BCL1 1A knockdown caused basal breast cancer cell death or differentiation and abolished these cells' tumour-forming potential. Importantly, approximately 17% of BLBC cases have copy number gains at the BCL1 1A locus. We have therefore identified BCL1 1A as a master regulator in mammary stem cells and an important basal breast cancer gene.

Implications for Cell of Origin of Basal Breast Cancer

Patients with germline BRCA1 mutations account for 5% of all breast cancer patients and often develop BLBC [65]. BRCA1 mutation carriers have a larger luminal progenitor population compared with WT carriers. This is proposed to be the target candidate population for basal tumour development in these patients [15]. In the mouse model, conditional deletion of Brcal in luminal cells gave rise to BLBC more frequently compared to deletion of Brcal in basal cells [16]. BCL1 1A is also expressed in luminal progenitors, likely in those generating ERoc^" luminal cells. As well as playing a critical role in mammary stem cells, BCL1 1A is able to induce basal gene expression in cultured mouse and human mammary epithelial cells, and transform these cells to form tumours with basal characteristics/gene expression signature. It is therefore possible that abnormal BCL1 1A expression in aberrant luminal progenitors gives advantages in survival and self-renewal, which may account for development of some basal breast cancers. This scenario is reminiscent of transformation from committed myeloid progenitor to leukaemia stem cell by the Mixed lineage leukemia (MLL) fusion protein MLL-AF9 [41].

BCL11A Negatively Regulates p53 in Mammary Cells

The p53 protein orchestrates stress responses to various types of damage to the cell and is frequently mutated in cancer [64]. Besides its critical role in DNA damage repair, p53 signalling has important roles in regulating stem cell function. It is proposed that p53 regulates hematopoietic stem cell quiescence [73] and asymmetric stem cell divisions in the mammary gland [67]. One recent role of p53 in stem cells is that it suppresses the EMT by transcriptionally activating the Mir-200 family members, which in turn negatively regulates key EMT transcription factors [68]. An intact EMT programme is critical in the MaSC [8].

We show here that BCL1 1A activates EMT inducer expression and negatively regulates p53 protein. When BCL1 1A is overexpressed in mammary cells, p53 response to genotoxic stress is attenuated. Conversely, decreased BCL1 1A activities in breast cancer cell lines potentiate a stronger p53 response. Such response was characterised by the up-regulation of p21 , a well studied inducer of cell cycle arrest, differentiation and senescence [74]. It was independently suggested that BCL1 1A is a negative regulator of p21 via BCLHA's direct regulation of SIRT1 [75], which deacetylates p53 rendering it inactive [76]. However, in the BCL1 1A-overexpressing cells, we did not find changes of SIRT1 expression nor levels of acetylated p53 (data not shown). Instead, we found that BCL1 1A regulates key negative regulators of p53, MDM2 and MDM4.

Although MDM2's role in mammary stem cells is still not clear, a recent study showed that MDM2 has a critical role in hematopoietic stem ceils [77]. Moreover, a human genetics study shows that SNP285C in intron 1 of the MDM2 locus is strongly associated with a reduced risk of developing breast and ovarian cancer in Caucasians [78]. It is speculated that SNP285C affects SP1 binding which in turn tunes down MDM2 expression. Interestingly, SNP285C is located only one base away from a core BCL1 1 A binding site in intron 1 of the MDM2 locus. It is possible that SNP285C may prevent BCL1 1A from effectively inducing MDM2 gene expression and thus lead to a greater protection against tumourigenesis. We therefore propose that high levels of BCL1 1A promote self-renewal and survival of aberrant cells by lowering p53 protein levels and facilitating the EMT. These cells subsequently accumulate additional mutations, and eventually become malignant cells which require BCL1 1A for cancer cell self-renewal and survival (Figure S12).

BCL11A and Therapeutic Implications

Given the importance of p53 in cancer, restoration of WT p53 function in human tumours is a valid therapeutic approach. Drugs targeting p53 are currently in clinical trials. For example, cis- imidazoline analogs Nutlins inhibit the interaction between MDM2 and p53 and thus activate p53 [79], and covalent modification of mutant p53 by compounds converted from PRIMA-1 is sufficient to induce apoptosis in some tumor cells [80]. However, in some cases, restoring wild type p53 activities is not sufficient to kill tumour cells. One example shown in this work comes from the MDA231 cells: neither BCL1 1A knockdown nor introduction of WT p53 into these cells was sufficient to suppress cell growth. WT p53 was effective only when BCL1 1A was also knocked- down in these cells. This finding is directly relevant to basal breast cancer therapies particularly in light of the recent attempts to re-activate mutant p53 in tumour cells using small molecule compounds [81]. Future investigation on how to modulate BCL1 1A activities is required to fully realise the therapeutic potential of BCL1 1 A in breast cancer treatment.

Methods Summary Mouse strains

All animals were generated and treated in accordance to the UK's Animals (Scientific Procedure) Act 1986 and the local ethical committee and UK Home Office guidelines. Bcl11a-lacZ reporter and Bcl11a conditional knockout mice were produced as detailed in Methods. Rosa26-CreERT2 mice were obtained from Dr. David Adams and Prof. Allan Bradley (Sanger Institute, UK), flox/flox mice were injected intraperitoneal^ with 1.0 mg of tamoxifen (Sigma) dissolved in sunflower oil for 3 consecutive days. Mice were analyzed 1-3 weeks after the 1^st injection.

Mammary cell preparation, cell sorting and cell culture

Mammary glands from 5-12-week-old virgin female mice were dissected and mammary epithelial cell suspensions were prepared as previously described {Stingl, 2006 #31}. A list of antibodies and cell culture conditions is detailed in Methods. Fluorescein di- -D-galactopyranoside (FDG; Sigma) was used for flow cytometric analysis to identify Bc/77-expressing mammary epithelial cells. For colony-formation assays, freshly sorted cells were cultured as described previously [2].

Transplantation of mammary epithelium Mammary glands from tamoxifen injected and non-injected flox/flox mice were dissected and mammary epithelial cell suspensions were as described elsewhere [2]. Sorted mammary epithelial cells were transplanted into the inguinal glands of NOD/SCI D/IU-rv '^" females in limiting dilutions as described previously [2]. The females were impregnated 3-6 weeks after transplantation and outgrowths were stained with carmine and scored.

Immunohistochemistry

Paraffin-embedded mammary sections were de-paraffinized and antigen retrieval was performed as described previously [21]. Primary antibodies are listed in Methods.

Chromatin Immunoprecipitation

ChIP assay was performed using the ChlP-IT express kit (Active Motif). Untransfected, EpH4 cells transfected with wither MSCV-IRES-EGFP or MSCV-Bcl1 1a-IRES-EGFP vectors were used for ChIP assay.

Figure legends

Figure 1. Bcl11a is expressed in mammary stem cells, (a) X-gal staining of (i) wholemount of mammary gland from 4-5 weeks old Bcl11a^{lacZ +} mice. TEB: Terminal end bud; (ii) Immunohistochemistry of ductal sections for Bcl1 1a and SMA (basal) expression, (b) Flow cytometric analysis of mammary epithelial cells from the wild type control or Bcl11a-lacZ reporter mice using FDG, in combination with antibodies to CD24 and CD49f. The left panel is the typical profile of mammary epithelial cells. Expression of Bcl11a was detected in both luminal (CD24^hiCD49f⁺) and basal (CD24⁺CD49f^hi) subpopulations. Numbers indicate percentage of FDG⁺ epithelial cells, (c) Bcl11a expression during an adult mammary gland developmental cycle detected by qRT-PCR. Error bars indicate standard deviations (n=3). Numbers on the X-axis indicate mammary gland developmental time points (days). Cyclophillin A is used as the expression reference, (d) qRT-PCR analysis in FACS purified MaSCs and Myoepithelial cells (Myo) showing an almost exclusive expression of Bcl1 1a in MaSCs compared to other known mammary epithelial markers. Error bars display standard deviation (n=3)

Figure 2. Bcl11a is required for mammary stem cells, (a) Flow cytometric analysis of mammary epithelial cells. MaSCs-enriched population (CD24^medCD49f^hl) is depleted in the Bc/77a-deficient gland, (b) Bc/77a-deficient mammary stem cells fail to engraft the cleared mammary fat pad in a limiting dilution transplant. 2-6 mice were transplanted per cell dosage, and the control and Bc/77a-deficient mammary cells were injected in the contra-lateral gland, respectively, (c) Bcl11a deficiency significantly affects expression of genes implicated in mammary stem cells. Control: wild type MaSCs. [These genes are, from the top of the figure, ETF1 , XBP1 , ETS2, MPP6, GLYCAM1 , HERPUD1 , SYVN1 , UNC5B, GSK3B, LDHA, TNSRSF1 1A, VEGFA, PD1A3, KRT15, CITED2, MYSM1 , ZBTB7A, TCFAP2C, NFAT5, NDRG1 , LM04, EIF5, TRIM32, BCL1 1A, TRIM29, TIMP2, S100A6, COL16A1 , OXCT1 , FXYD3, APOE, ITGB4, MMP3, S100A1 , ITM2B, SERPING1 , MMP2, ITGB5, LY6E, CTSB, TIMM8B, TWIST2, AOP5, MMP14, S100A1 1 , TIMP1 , FOXA1 , LY6A, HES1 , SOCS3, BCL1 1 B, ADAMTS4, SNA13, NOTCH3]. (d) qRT-PCR analysis of gene expression changes. RNA samples were from FACS purified basal cells of wild type (control) and Bc/77a-deficient mammary glands. Error bars display standard deviation (n=3). (E) Quantitative RT-PCR (qRT-PCR) for Bcl11a in different mammary epithelial cell compartments which were FACS-purified using antibodies for CD24, CD49f and CD49b.

Figure 3. Bcl11a enhances sternness in mouse mammary epithelial cells by regulating key EMT and Notch signalling genes, (a) Over-expression of Bcl11a in EpH4 cells significantly increases the number of mammospheres compared to the control (n=3). T-Test *p<0.03. (b) Over-expression of Bcl11a in EpH4 up-regulates p63, CK14, Notch3 and the EMT genes Zeb1, Zeb2, Twist2 and Foxc2 in qRT-PCR analysis, (c) Putative Bcl1 1a binding regions (i-viii) at Zeb1, Foxc2, Twist2, Notchl and Notch3 genomic loci, (d) Bcl1 1a binds directly to some of the putative binding sites as shown by chromatin immunoprecipiation (ChIP) with a Bcl1 1a antibody using cell extracts from the control and Bcl1 1 a over-expressing EpH4 cells. MIG: retroviral vector, (e) Bcl11a expression induces changes of H3K27 tri-methylation and H3K4 di-methylation at several genomic loci in ChIP analysis.

Figure 4. BCL11A is highly expressed in basal breast cancer and is essential for cancer stem cells, (a) Over-expression of BCL11A in HMLE increases CFCs compared to the control (n=3). T-Test *p<0.01. Cells were transfected with the PB vectors and BCL11A expression was induced by adding 1.C^g/ml Doxcycline in the media, (b) Over-expression of BCL1 1A up- regulates expression of the EMT marker VIMENTIN and EMT inducers ZEB1, ZEB2 and FOXC2, but represses miRNA-200c in HMLE cells, (c-d) BCL11A expression is significantly correlated with the ER negative status (c) and basal subtype of breast cancer (p<1 e-16) (d) in the NKI-295 dataset. (e) shRNA-mediated knockdown of BCL11A increases apoptosis in CAL120 and BT549 basal breast cancer cells 24-48 hours post transfection compared to the scrambled control shRNA (n=3). T-Test *p<0.05 and **p<0.08. (f-g) The BCL1 1A-KD CAL120 cell culture has many more epithelial-like, CK18⁺ and E-CADHERIN⁺ cells (f), and many fewer CD447CD24^" cancer stem cells (g). (h) Gene expression analysis in the BCL1 1A-KD CAL120 cells by qRT-PCR reveals a significant up-regulation of the differentiation markers GATA-3, FOXA1, ERct, CK18, E- CADHERIN and miRNA-200c, concomitant with down-regulation of the EMT inducers TWIST2, FOXC2, ZEB1 and ZEB2.

Figure 5. BCL11A is Highly Expressed Specifically in Basal Breast Cancer

(A) Analysis of expression of 32 key hematopoiesis genes in 337 breast cancer patients [42] of different breast cancer molecular subtypes. (B) Expression of BCL1 1A in 22 breast cancer cell lines. BCL1 1A expression in primary mammary epithelial cells (HMEC) was used as the reference. C) IHC using an anti-BCLUA antibody on normal breast tissue (top) and invasive ductal carcinoma (bottom). The sections were counter stained with Hematoxylin to visualise the nuclei (blue). These two images are enlarged ones taken from tissue microarray data (A2 and D6) shown in Figure S8B.

Figure 6. High Levels of BCL11A Enhance Self-Renewal of Mammary Epithelial Cells and promote Tumourigenesis

(A) Bcl1 1a overexpression in EpH4 cells led to an increase in the number of colonies formed by cells seeded in Matrigel (n=3). (B) Increase of both the number and size of floating mammospheres formed from human HMLE cells overexpressing BCL11A (n=3). (C) Knockdown of BCL11A in HMLER cells reduced the number as well the size of mammospheres formed from these cells (n=3). (D) Image and graph depicting the size difference between tumours at 6 weeks after injection of EpH4-control and EpH4-1 1A orthotopicallly into contralateral mammary fatpads (n=6). (E) H&E and IHC analysis of the EpH4control and EpH4-1 1A tumours showing more intense staining for CK14 and Vimentin, two of the markers for basal-like tumours (F) Image depicting tumours at 10weeks after subcutaneous injection of HMLE-Control and HMLE-1 1A (n=4). (G) H&E and IHC analysis of the HMLE-1 1A tumours showing stronger staining for CK14 and Vimentin and weak staining for CK18. (H) Image and graph depicting tumour sizes of HMLER-sh-control and HMLER-sh-1 1A cells after subcutaneous injected (n=4). Error bars display standard deviation (n=3). T-Test *p<0.05 and **p<0.08.

Figure 7. BCL11A Regulates Genes Implicated in Stem Cell Function and Tumourigenesis

(A) Gene expression changes by either Bcl11a deletion or overexpression in Bcl11a deficient MaSCs and HMLE-1 1A cells. The gene list for this supervised clustering are chosen from the MaSC gene signature [10]. 2. (B) Downregulation of the EMT genes Foxc2, Zeb1, Zeb2 and Vimentin in Bcl11a deficient MaSCs detected by qRT-PCR. (C-D) BCL1 1A overexpression in either mouse EpH4-1 1A cells (C) or human HMLE cells (D) increased expression levels of several EMT genes (Foxc2, Zeb1, Zeb2) in qRT-PCR analysis. BCL1 1A also downregulated Mir- 200c in human cells (D). (F) Intense IHC staining of differentiated epithelia markers CK18⁺ and E- CADHERIN in CAL120-sh-1 1A cells. (G) BCL1 1A knockdown in CAL120-sh-1 1A cells also downregulated expression of EMT genes FOXC2, ZEB1, and ZEB2 concomitant with the upregulation of Mir-200c. (H) Diagrams showing the putative Bcl1 1a binding sites at the Zeb1 and Foxc2 loci. Bold lines indicate the region amplified by PCR after the ChIP. Arrows mark transcription start sites. (I-K) ChIP experiments to detect Bcl1 1 a binding to the Zeb1 and Foxc2 loci, and to examine histone H3 methylation status, in EpH4-1 1A cells (-Dox or +Dox). qRT-PCR was used to detect specific genomic DNA fragments DNA pulled down using specific antibodies. In all ChIP experiments, non-specific rabbit IgG was used as the negative control. (H) Binding of Bcl1 1 a to the Zeb1 and Foxc2 loci using a Bcl1 1a antibody. An antibody to either histone H3K4- 3me (I) or histone H3K27-3me (J) was used to detect chromatin changes when Bcl1 1 a was overexpressed.

Figure 8. BCL11A Is a Negative Regulator of p53 via MDM2

(A) Overexpression of BCL1 1A in human HMLE-1 1A cells upregulated several p53 negative regulators. The supervised clustering was performed based on microarray data (data not shown).

(B) qRT-PCR analysis also detected downregulation of p53 negative regulators Mdm2 and Mdm4 in Bc/77a-deficient mouse MaSCs. (C) Overexpression of BCL1 1A prevented p53 accumulation in EpH4 cells in response to UV irradiation. Vinculin was used as the loading control for the Western blots. (D) Treatment with Doxorubicin in Bc/77a-expressing EpH4 cells also failed to accumulate p53. Note the increase of Mdm2 in EpH4-1 1A cells. (E) Expression of WTp53 in MDA231 cells, together with BCL1 1A knockdown reduced Mammosphere formation. MDA231 cells were transfected with either a control plasmid (-WT53) or one expressing the WT p53 (+WTp53). BCL1 1A knockdown shRNA (black bars) or the control shRNA (white bars) were expressed in these cells. Asterisk: p<0.05. (F) Increase of p21 expression in MDA231 cells expressing the WT form of p53 and BCL1 1A knockdown on Western blots. All cells expressed the WT form of p53. The control knockdown did not change p21 upon Doxorubicin treatment whereas BCL1 1A knockdown upreulated p21 even without Doxorubicin treatment (0 hour). (H) Diagram depicting putative Bcl1 1a binding sites at the Mdm2 Locus. Bold lines indicate the region amplified by PCR after the ChlP. Arrow shows the transcription start site. (I-K) ChIP experiments to detect Bcl1 1a binding to the Mdm2 locus, and to examine histone H3 methylation status, in EpH4-1 1A cells (-Dox or +Dox). qRT-PCR was used to detect specific genomic DNA fragments DNA pulled down using specific antibodies. In all ChIP experiments, non-specific rabbit IgG was used as the negative control. (I) Binding of Bcl1 1a to the Mdm2 locus using a Bcl1 1a antibody. An antibody to either histone H3K4-3me (J) or histone H3K27-3me (K) was used to detect chromatin changes when Bcl1 1a was overexpressed.

Table 1. Oncogenes And Tumour Suppressors Differentially Regulated In BCL11A Overexpressing HMLE Cells.

Gene ID pvalue FOLD change HMLE-1 1A vs HMLE-control

GPR87 0.0171 330

FGFR1 0.0392 79.25

MIR21 0.0400 33.8039

MLL 0.01 10 30.9608

ABCC4 0.0328 24.175

MYCBP2 0.0180 18.6731

FGFR10P 0.0006 14.3259 NEDD9 0.0433 14.125

KRAS 0.0005 1 1.3297

WNT6 0.0345 10.1 163

EVI 1 0.0205 8.35841

BIRC3 0.0282 6.01757

IL6 0.0039 4.59227

CD47 0.0012 3.27205

FOXC2 0.0084 3.16084

ITGA6 0.0232 3.06806

IGF1 R 0.0461 3.05854

PIK3CB 0.0122 3.04781

TCF12 0.0530 2.45495

MDM2 0.0466 2.20475

CDK8 0.0019 2.20361

HIF1A 0.0400 2.10022

IDH1 0.0060 2.0217

ITGB1 0.0046 1 .83951

CDKN1 C 0.0192 -3.29297

TP53AIP1 0.0733 -4.60759

AIM2 0.0769 -5.60345

ZBTB2 0.0009 -5.69892

RASSF4 0.0790 -6.55072

CASP8 0.0231 -12.1316

BMP6 0.0032 -58.04

Supplementary Figures and Tables Legends

Supplementary Figure 1. Bcl11a expression in the mammary gland, (a) Schematic diagram of the Bcl11a-lacZ allele. The SA-IRES-lacZ cassette was targeted to intron 3 of the Bcl11a locus, (b-k) Whole mount and sections of X-gal stained Bcl11a'^aoZ/+ embryonic and adult mammary glands at various developmental stages, (b) 12.5 days post coitum (dpc) and (c) 14.5 dpc embryos. Arrows point to the mammary placodes that express Bcl11a. (d-e) 8-12 weeks virgin gland. The insert shows expression of Bcl1 1a in alveolar buds during estrus). (f-g) Early gestation (4-5 dpc), (h-i) Late gestation (14-15 dpc) and (j-k) Day 3 lactation mammary glands. Sections were counter-stained with carmine alum stain. (I) Expression of Bcl11a in sorted luminal (CD24^hiCD49f⁺) and basal (CD24⁺CD49f^hi) cells from the virgin and gestation mammary glands by qRT-PCR. Error bars display standard deviation (n=3)

Supplementary Figure 2. Schematic diagram of the Bcl11a conditional knockout allele used in this study. Exon 1 of Bcl11a was flanked by loxP sites for deletion. Southern analysis on the right shows detection of the conditional knockout allele in mice. Wt: wild type allele, fl: conditional knockout allele.

Supplementary Figure 3. Loss of both basal and luminal epithelial cells in the Bc/77a-deficient mammary gland. Analysis of gene expression changes in the Bc/77a-deficient virgin mammary gland by both regular qRT-PCR and qRT-PCR. Note the drastic decrease of expression of basal (CK14 and p63) and luminal (CK18, Mud, Gata-3 and ElfS) genes. Importantly, expression of Elf5, which encodes one of the key transcription factors in the luminal progenitors, is undetectable in the mutant mammary gland in regular RT-PCR analysis.

Supplementary Figure 4. Severe defects in the Bcl17a-deficient virgin mammary gland.

(a-b) Whole mount carmine alum-stained and (c) H&E stained sections of the control (flox/+) and the flox/flox mammary glands 3 weeks after tamoxifen administration, (d-j) Immunostaining of the mammary epithelium with antibodies to various markers: (d) Aquaporin 5 and CK14, (e) SMA; (f) p63; (g) Gata-3 (note very few Gata-3 positive cells); (h) Notchl , and (i) Jagged! (j) FACS profiles of the control (flox/+) and the mutant (flox/flox) mammary epithelial cells one week post tamoxifen injection, (k) Genomic DNA PCR analysis of the outgrowth from tamoxifen treated flox/flox mammary epithelial cells. Note the compete absence of the deletion band from this outgrowth indicating that these re-constituted cells were originated from some undeleted donor stem cells in the tamoxifen treated flox/flox mammary gland. (I) Confirmation of Bcl11a deletion in MaSCs for gene expression analysis. The genomic DNA fragment across the floxed region (the CKO band) and the fragment generated by Cre excision (the deletion band) are amplified by PCR. Note the complete absence of the CKO fragment in the tamoxifen treated cells, demonstrating a near complete deletion of Bcl11a in the MaSCs-enriched population.

Supplementary Figure 5. Bcl11a over-expression leads to more CK14⁺ and fewer CK18⁺ EpH4 cells. Immunostaining of the control or the Bc/77a-overexpressing EpH4 cells with CK14 and CK18 antibodies.

Supplementary Figure 6. Over-expression of BCL11 A in human breast cell enhances sternness and detection of BCL11A expression in human breast cancer cell lines, (a) BCL11A over- expressing in HMLE cells increases the percentage of CD447CD24^" stem cells, (b) Immunostaining of the control and the Bc/77a-overexpressing HMLE cells with CK18 antibody showing drastic reduction of CK18⁺ cells. Cells were counter-stained with DAPI. (c) qRT-PCR analysis of 22 human breast cancer cell lines showing much higher levels of BCL1 1A in basal cell lines compared to luminal cell lines, (d) HMLE cells transfected with the BCL11A-KD vector and the control shRNA vector. Cells were stained with antibodies for E-CADHERIN and CK-18. (e) In vivo imaging of mice injected with CAL120 cancer cells. Both the control CAL120 cells and the BCL1 1a-KD CAL120 cells expressed GFP from the KD vector, which was detected by MS Lumina imager 10days after injection. Tumour sizes were measured using Image J (NIH).

Supplementary Figure 7. Bcl11a is expressed in and essential for luminal progenitors.

(a) The FDG⁺ (Bc/77a-expressing) luminal cells (CD24^high) are gated out for expression of CD49b and Seal . Most FDG⁺ luminal cells are progenitors (CD49b⁺, 76%). And most Bc/77a-expression luminal progenitors are Seal^". Numbers indicate percentages of FDG⁺ cells, (b) RT-PCR analysis in sorted mammary luminal cells shows that expression of Bcl11a is highest in the luminal CD49b⁺Sca1^" progenitors, similar to that of Elf5. (c) The Bc/77a-expressing mammary epithelial cells (FDG⁺) have more Mammary colony-forming cells (Ma-CFCs). Error bars denote standard deviation. P value represents Student's t-test between FDG⁺ and FDG^" cells (n=3). (d) Significant decrease of Ma-CFCs upon Bcl11a deletion in the virgin gland. Error bars denote standard deviation. P value represents Student's t-test between the control (flox/+) and the flox/flox mice one week after induction of Bcl11a deletion (n=3). (e) Immunostaining of the control or the Bc/77a-deficient virgin gland for ERa. (f) Deletion of Bcl11a in the virgin gland results in approximately two-fold increase in the percentage of ERa⁺ cells in the luminal cell population. ERa⁺ luminal cells were calculated based on immunostaining. Error bars denote standard deviation. P value represents Student's t-test between control and Bc/77a-deficient glands (n=3).

Supplementary Figure 8. Expression of BCL11A in Breast Cancer

(A) Re-examination of BCLI1A expression using published datasets on the ICR ROCK website as indicated. Box plots showing High levels of BCL11A in the basal subtype of breast cancer are shown.

(B) IHC staining for BCL1 1A on tissue microarray from 48 breast cancer patients (full details of the array can be found at www.biomax.us). The table below shows the ERa status of the corresponding tissue sections. The dashed square (samples A1 and A2) are normal breast tissues from reduction mammoplasty. Samples A3-A8 are classified as benign plasma cell mastitis and stained strongly for BCL1 1 A. This is expected given the expression of BCL1 1A in B- cells.

Supplementary Figure 9. BCL11A Overexpression and Knockdown in mammary epithelial cells.

(A) Schematic diagram of the BCL1 1A overexpression vector used in this study. The expression cassette is delivered to cells by the PB transposition. The Western blot confirms regulation of BCL1 1A expression in EpH4 cells is doxycycline-dependent. (B) BCL1 1A expression in EpH4 cells induces basal gene expression such as CK14 and p63. (C) The BCL1 1 A shRNA knockdown vector in the PB transposon.

Supplementary Figure 10. High Levels of BCL11A promote Tumourigenesis whereas BCL11A knockdown induces differentiation.

(A) Unsupervised clustering of the HMLE-1 1A tumours in the mouse with human tumours from the METABRIC study based on the PAM50 gene expression. Black rectangle highlight the three HMLE-1 1A tumours clustered within the basal subgroup (enlarged on the right). (B) shRNA- mediated knockdown of BCL11A increases apoptosis in CAL120 basal breast cancer cells 24-48 hours post transfection. (C) Change of cell morphology of CAL-120 cells when BCL1 1A is knocked down. These cells now have an epithelial morphology rather than mesenchymal one. (D) BCL11A knockdown also causes higher expression levels of differentiation markers CK18, GATA3, FOXA1 and ERa in CAL120-sh-1 1A cells compared to CAL120-sh-control cells. (E) CAL120-sh-1 1A cells have a decreased capability to form mammospheres compared to CAL120- sh-control cells. Supplementary Figure 11. Bcl11a Negatively Regulates P53 Signalling

(A) Validation of MDM2 and PSMD10 upregulation in the microarray data (Figure 6A) by qRT- PCR. (B) Cell death in EpH4-1 1A and EpH4-control cell cultures at various time points following UV treatment. (C-D) Western blot analysis on the effect of either BCL11A overexpression (C) or knockdown (D) on p53, MDM2 and p21 protein levels in HMLE and HMLER cells, following Doxorubicin treatment.

Supplementary Figure 12. SNP285C in MDM2 intron 1 Overlaps with the BCL11A Binding Site

(A) Alignment of the mouse and human BCL1 1A binding site in BCL11A intron 1. Red box highlights SNP285C just outside a BCL1 1 A binding site (dashed line).

(B) A working model of BCLHA's function in maintaining normal mammary stem cell and tissue homeostasis and in tumour development.

Supplementary Figure 13. Bcl1 1 a is expressed in hematopoietic stem cells (HSCs). We made a reporter mouse where the EGFP gene was targeted to the 3' UTR of the Bcl1 1a locus. Consequently, expression of Bcl1 1a could conveniently be monitored in flow cytometry as GFP+. (A) Almost all long-term HSCs (Lin-Kit+Sca1 +CD48-CD150+) are GFP+ and thus express Bcl1 1 a. (B) qRT-PCR analysis of Bcl1 1a expression in GFP+ HSCs compared to in GFP- bone marrow cells. Numbers above columns indicate relative expression.

Supplementary Figure 14. Rapid depletion of long-term HSCs upon Bcl1 1a deletion. The Bcl1 1a conditional knockout mice were crossed to Rosa26-CreERT2 mice to produce Bcl1 1a flox/flox;CreERT2 mice. Once Tamoxifen is administrated to these mice, Cre recombinase is activated which deletes the floxed Bcl1 1a in almost all cells in the mouse. (A) Reduced bone marrow cellularity (left panel) and depletion of Lin-Kit+Sca1 + cells (gated) (middle panel) and long-term self-renewing Lin-Kit+Sca1 +CD150+CD48- (gated) (right panel) once Bcl1 1 a is deleted. Number above gates are percentage gated of total bone marrow cells. (B) Cellularity of Lin-Kit+Sca-1 + and long-term HSCs (Lin-Kit+Sca1 +CD150+CD48-) in the control and the Bcl1 la- deleted bone marrow. Flow cytometric analysis was performed 7 days after Tamoxifen treatment. The control mice were Bcl1 1a flox/+;CreERT2 heterozygotes.

Supplementary Figure 15 Bcl1 1a has non-cell autonomous essential roles in HSCs. To further demonstrate the essential roles of Bcl1 1a in HSCs, we transplanted 1 x 106 bone marrow cells from Bcl1 1aflox/flox;CreERT2 or the control mice to lethally irradiated wild type recipients. Following a period of 6-8 weeks reconstitution, the primary recipient mice were treated with 2 rounds of three injections of Tamoxifen over a period of 4 weeks to delete Bcl1 1a. Flow cytometric analysis was performed to investigate HSCs in the bone marrow of the recipient mice (A) Reduced bone marrow cellularity (left panel) and depletion of Lin-Kit+Sca1 + cells (gated) (middle panel) and long-term self-renewing Lin-Kit+Sca1 +CD150+CD48- (gated) (right panel) once Bcl1 1 a is deleted. Number above gates are percentage gated of total bone marrow cells. (B) Representative cellularity and absolute cell number of Lin-Kit+Sca1 + cells and long-term HSCs (Lin-Kit+Sca1 +CD150+CD48-) in the control and the Bcl1 1a-deleted bone marrow. Once Bcl1 1 a is deleted, all long-term HSCs are depleted in these mice

Supplementary Figure 16: Analysis of Bcl1 1a expression in tissues of an adult mouse using a lacZ reporter mouse. The lacZ expression cassette was targeted to the Bcl1 1 a locus. Consequently, expression of Bcl1 1 a could be tracked by X-gal staining. (A) Expression of Bcl1 1 a in the brain. (B) Expression of Bcl1 1a in the intestine villi.

Supplementary tables

Supplementar Table 1 Genotyping primers.

Supplementary Table 2. Primers and PCR conditions for RT-PCR analysis.

Su lementary Table 3. Primers for Chromatin Immuno-precipitation PCR analysis

S1_F

m-MDMX-

AAATGCAGTGCAGGCCTTAG GACCCCAACACCACACCTTA (SEQ ID

S1_F (SEQ ID NO: 116) NO: 117)

Additional materials and methods Mouse strains

Bcl11a bacterial artificial chromosomes (BACs) were identified from the 129/SvJ mouse BAC library (Sanger Institute) and used to generate the Bcl11a-lacZ and Bcl11a conditional knockout targeting vectors. For Bcl11a-lacZ tagged conditional knockout mouse, targeting construct (Supplementary Fig. 1a) was generated based on the recently published strategy[82]. For the conventional Bcl11a conditional knockout mouse, targeting construct (Supplementary Fig. 1 b) was generated based on the original recombineering strategy[83]. Gene targeting in ES cells and chimera production were performed according to standard procedures. Genotyping primers are listed in Supplementary Table 1.

Whole mount staining

For whole mount X-gal staining of embryos, embryos were dissected and fixed in 4% paraformaldehyde for 30 min at 4°C. Embryos were then washed thrice in a large volume of PBS at 4°C. Next, embryos were incubated in X-gal staining solution [5 mM K₃Fe(CN₆), 5 mM K₄Fe(CN₆).3H₂0, 0.02% IGEPAL CA-630, 0.01 % sodium deoxycholate, 2 mM MgCI₂ and 1 mg/ml X-gal in PBS] for 24-48 hours at 4°C. For whole mount X-gal staining of tissues, mammary glands were spread out on glass slides and fixed in 4% paraformaldehyde for 2 hours at 4°C. Glands were then washed thrice with PBS (with 2 mM MgCI₂). Next, the glands were incubated in permeabilization solution (0.2% NP40, 0.01 % sodium deoxycholate, 2 mM MgCI₂ in PBS) for 1 hour. Finally, the glands were incubated in X-gal mixer [25 mM K₃Fe(CN₆), 25 mM K₄Fe(CN₆).3H₂0, 0.2% NP40, 0.01 % sodium deoxycholate, 2mM MgCI₂ in PBS] for 1.5 hours at 37°C before adding 1 mg/ml X-gal (Invitrogen) and incubating for 1-3 days at 37°C. For whole mount analysis, mammary glands were spread out on glass slides and stained in carmine alum stain as detailed previously[21].

Mammary cell preparation, cell sorting and cell culture

Primary antibodies used: biotinylated anti-CD45 (clone 30-F1 1 ; eBioscience), anti-Ter1 19 (clone Ter1 19; eBioscience) and anti-CD31 (clone 390; eBioscience); anti-CD24-R-phycoerythrin (PE; clone M1/69, eBioscience), anti-CD49f-Alexa Fluor 647 (AF647; clone GoH3, eBioscience), anti- CD49b-Alexa Fluor 488 (AF488; clone HMa2; eBioscience) and Sca1-Alexa Fluor 647 (AF647; clone D7, eBioscience). Secondary antibodies used: Strepavidin-PE-Texas Red (PE-TR; Molecular Probes). Apoptotic cells were excluded by elimination of propidium iodide (PI) positive cells. For FDG staining, cells were first stained with primary and secondary antibodies. Samples and FDG stock (2 mM in DMSO) were pre-warmed at 37°C for 5 min. Next, an equal volume of FDG stock was added and the cell mixture was incubated at 37°C for 1 min before adding 2 ml of HBSS supplemented with 2% fetal bovine serum (FBS; StemCell) and incubating on ice for 1 hour in the dark. Flow cytometric analysis was done using CyAN ADP (DakoCytomation) and all sorts were performed using MoFlo (DakoCytomation) and gates were set to exclude >99.9% of cells labelled with isoform-matched control antibodies conjugated with the corresponding fluorochromes. For colony-formation assays, the medium used was (human) NeuroCult NS-A Proliferation Medium (StemCell) supplemented with 5% FBS, 10 ng/ml epidermal growth factor (Sigma), 10 ng/ml basic fibroblast growth factor (Peprotech) and N2 Supplement (Invitrogen) and the cultures were maintained at 37°C/5% C0₂ for 7 days and then fixed using ice cold acetone: methanol (1 :1 ) and visualized using Giemsa staining (Merck).

Ma-CFC assay

Mammary cell preparation and cell sorting were performed as detailed in experimental procedures. Lin^"CD24^hlCD49b⁺Sca1 ^+or" luminal progenitors from the flox/flox mammary gland were sorted and plated with irradiated feeders in NSA media (human) NeuroCult NS-A Proliferation Medium (StemCell) supplemented with 5% FBS, 10 ng/ml epidermal growth factor (Sigma), 10 ng/ml basic fibroblast growth factor (Peprotech) and N2 Supplement (Invitrogen) for 24 hours before 1 μΜ of 4-hydroxytamoxifen (4-OHT) was added to induce deletion of Bcl11a. After 2 hours, fresh NSA media was added and cells were maintained at 37°C/5% C0₂ for another 6 days before the number of Ma-CFCs was enumerated

Transplantation of mammary epithelium

Mammary epithelial cells (Basal fraction) from tamoxifen injected flox/flox and non-injected flox/flox or flox/+ mice were sorted based on CD24/CD49f and transplanted in limiting doses (500/750/1000/2000 cells) into cleared fat-pads of 3 week-old NOD/SCID/IL^rv '^" females. In each case, non-injected and tamoxifen-injected epithelial cells were engrafted into contralateral glands of the same recipient mice. The recipient mice were impregnated 3-6 weeks after transplant and outgrowths produced were dissected, stained with carmine and scored.

Cell lines, Transfection and Mammosphere assays

EpH4, CAL120 and MDA231 cells were cultured to confluence in 1 : 1 DMEM:F12 (Invitrogen) media containing 10% FCS (Fetalclonelll, Clonetech). HMLE and HMLER cells were a gift from Prof. Robert Weinberg, Whitehead Institute and were cultured in complete HuMEC media (Invitrogen). The control or the Bcl11a overexpression piggybac vectors were delivered into cells using the using Amaxa basic nucleofactor kit for primary mammalian epithelial cells (Lonza) according to the manufacturer's recommendations. Transfected cells were maintained at 37°C/5% C0₂ for 48 hours. Cells were then cultured in puromycin (1-5 μg/ml) for 48 hours to allow for selection. Doxycycline (Clonetech) was used at a final concentration of 1.0 μg/ml. Mammosphere were cultured and passaged as previously described (Dontu et al., 2003) in ultra- low attachment plates (Corning). Transfection

EpH4 cells were cultured to confluence in 1 :1 DMEM:F12 (Invitrogen) media containing 10% FCS (Fetalclonelll, Clonetech). The control or the Bcl11a MSCV-IRES-EGFP vectors were transfected into EpH4 cells using Lipofectamine LTX reagent (Invitrogen) according to the manufacturer's recommendations. Transfected cells were maintained at 37°C/5% C0₂ for 48 hours. GFP positive cells were sorted and collected for analysis. HMLE cells were maintained in HuMEC media (Invitrogen) and were transfected with 4.C^g of the control or Bcl11a piggyback vector using Amaxa basic nucleofactor kit for primary mammalian epithelial cells (Lonza) and puromycin (1.C^g/ml) was added 24 hours later for selection. Doxycycline (clonetech) was used at a final concentration of 1.C^g/ml

RNA knockdown

BCL1 1A shRNA sequences were described previously[84] and were cloned into piggyBac transposon vector (PB-H 1 -shRNA-GFP). CAL120 and BT-549 were transfected with 4.(^g of piggybac Vector using Amaxa basic nucleofactor kit for primary mammalian epithelial Cells and green cells were sorted/analysed 24-48hours later.

RNA Extraction and Real-time PCR analysis

RNA from sorted cells was extracted using PicoPure RNA isolation kit (Molecular Devices) according to the manufacturer's instructions. RNA from mammary tissue was extracted using Tri- reagent (Invitrogen) according to the manufacturer's instructions. cDNA was synthesized from 1- 2μg of total RNA using the Transcriptor reverse transcription cDNA synthesis kit (Roche). RT- PCR was performed using Hi-Fidelity Extensor mix (Thermo) using primers listed in Supplementary Table 2. Quantitative real-time PCR detection of cDNA was performed using SYBR Green Master Mix (Sigma) according to supplier's recommendations. The real-time PCR reactions were run in ABI-7900HT (Applied Biosystems) in triplicate. Primers used for real-time PCR were designed using PrimerBank[85] website (http://pqa.mqh.harvard.edu/primerbank ). All primers were purchased from Sigma-Aldrich

UV treatment and western blotting

Cells were exposed to 25J/m² using a UV Stratalinker 2400 (Stratagene) and collected at the indicated timepoints. Samples were prepared as described previously (Khaled et al., 2007) and probed using anti-total p53, (Abeam), anti MDM2, anti-p21 , anti-vinculin and anti-Bcl1 1a (SCBT). miR200c detection miR200c expression levels were detected using TaqMan® MicroRNA assays according to the manufacturer's instructions (Applied Biosystems). Apoptosis assay

Apoptosis was measured using Annexin-V-PE (BD Biosciences) according to the manufacturer's instructions and analysed using BD LSRII (BD Biosciences).

Immunohistochemistry

Ki67 (Abeam; 1 :50); CK14 (Abeam; 1 : 100); CK18 (Progen; 1 : 100); SMA (NeoMarkers; 1 : 1000); Aquaporin 5 (Calbiochem; 1 :50); p63 (Abeam; 1 :50); Notchl (Santa Cruz; 1 :100), Vimentin (Abeam; 1 :100), E-cadherin (Cell signalling; 1 :100) and Jaggedl (Santa Cruz; 1 :100) were used. Staining was detected using AF488- or Cy3-conjugated secondary (Sigma) and bisbenzimide- Hoechst 33342 (Sigma). Fluorescence microscopy was carried out using a Zeiss Axiophot microscope equipped with a Hamamatsu Orca 285 camera, with images visualized, captured and manipulated using Simple PCI 6 (C imaging systems). The H&E stains were visualized on a LEICA light microscope while the mouse mammary gland whole mounts were visualized using the LEICA MZ75 light microscope.

In silico analysis

In silico analysis and images of Bcl1 1a binding sites on the Notchl, Notch3, Foxc2 and Twist2 loci were generated using the freeware Genepalete 1.2.

Control or BCL1 1A overexpressing EpH4 cells were harvested, fixed and lysed according to manufacturer's instructions. DNA sheering optimization was performed using a Bioruptor Next Gen sonicator (Diagenode). Antibody incubation was performed overnight at 4°C with the following antibodies; Normal Rabbit IgG (Cell Signalling, 2729), Tri-Methyl-H3-K27 (Cell Signalling, 9733), Tri-Methyl-H3-K4 (Cell Signalling, 9751 ), Bcl1 1a (Bethyl Labs, A300-382A). qPCR was performed using 2 μΙ of purified input and pulldown DNA , primers are listed in Table S3.

Chromatin Immunoprecipitation

Untransfected, MSCV-IRES-EGFP or MSCV-IRES-Bcl1 1a-EGFP vectors stably transfected EpH4 cells were harvested, fixed and lysed according to manufacturer's instructions. DNA sheering and optimization was performed using a Soniprep sonicator with a fine tip at 30% of output and three pulses of 20 seconds each. Antibody incubation was performed overnight at 4°C with the following antibodies; Normal Rabbit IgG (Cell Signalling, 2729), Tri-Methyl-H3-K27 (Cell Signalling, 9733), Di-Methyl-H3-K4 (Cell Signalling, 9726), Bcl1 1a (Bethyl Labs, A300-382A). Standard PCR was performed using 5 ng of normalised DNA , primers are listed in Supplementary table 3.

Microarray analysis

The intensity value for each probe set was calculated and the average of each gene was computed before the data analysis. For the QC step, a set of intensity value of control genes were examined. All data were normalized and scaled by Partek Genomic Suite 6.4, PCA Principle components analysis was performed to show the distribution of samples, eliminating outliers. Differentially expressed genes were selected by One way ANOVA analysis by the factor of KO vs WT, p value < 0.08. Hierarchical Clustering of selected genes was performed to show the expressed pattern. The resulting genes then underwent a pathway analysis (GeneGO: http://www.genego.com) in order to determine the biological significance of the data.

NKI 295 data analysis

Gene expression and survival data was obtained from http://microarray- pubs.stanford.edu/wound_NKI/explore.html. The significance of associations between survival, gene expression, ERBB2 status and histological subtypes were determined with Cox hazards regression. The significance of associations between gene expression and histology or ERBB2 status was determined with an ANOVA. Box plots were generated with Matlab.

Tumour injection and imaging

150,000 cells were suspended in 25% Matrigel (BD biosciences) and HBSS (Invitrogen) and injected subcutaneously in 6-12 weeks old female NSG recipient mice. For live imaging, mice were anaesthetised and images were captured using the IVIS Lumina imager (Caliper life sciences) using the GFP filter and 1 minute exposure.

Tumourigenesis assays

One million EpH4, HMLE or HMLER cells were suspended in 25% Matrigel (BD Biosciences) and HBSS and injected into either cleared contralateral number 4 mammary fat pads of 3week old NSG female mice OR subcutaneously in 6-12 weeks old female NSG recipient mice.

Coughlin, S.S. and D.U. Ekwueme, Breast cancer as a global health concern. Cancer Epidemiol, 2009. 33(5): p. 315-8.

Stingl, J., et al., Purification and unique properties of mammary epithelial stem cells. Nature, 2006. 439(7079): p. 993-7.

Shackleton, M., et al., Generation of a functional mammary gland from a single stem cell. Nature, 2006. 439(7072): p. 84-8.

Bouras, T., et al., Notch signaling regulates mammary stem cell function and luminal cell- fate commitment. Cell Stem Cell, 2008. 3(4): p. 429-41.

Sansone, P., et al., IL-6 triggers malignant features in mammospheres from human ductal breast carcinoma and normal mammary gland. J Clin Invest, 2007. 117(12): p. 3988-4002.

Asselin-Labat, M.L., et al., Control of mammary stem cell function by steroid hormone signalling. Nature, 2010.

Joshi, P.A., et al., Progesterone induces adult mammary stem cell expansion. Nature. Mani, S.A., et al., The epithelial-mesenchymal transition generates cells with properties of stem cells. Cell, 2008. 133(4): p. 704-15.

Morel, A.P., et al., Generation of breast cancer stem cells through epithelial- mesenchymal transition. PLoS One, 2008. 3(8): p. e2888.

Lim, E., et al., Transcriptome analyses of mouse and human mammary cell subpopulations reveal multiple conserved genes and pathways. Breast Cancer Res, 2010. 12(2): p. R21.

Liu, P., et al., Bcl11a is essential for normal lymphoid development. Nat Immunol, 2003. 4(6): p. 525-32. Sankaran, V.G., et al., Human fetal hemoglobin expression is regulated by the developmental stage-specific repressor BCL11 A. Science, 2008. 322(5909): p. 1839-42. van de Vijver, M.J., et al., A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med, 2002. 347(25): p. 1999-2009.

Shimono, Y., et al., Downregulation of miRNA-200c links breast cancer stem cells with normal stem cells. Cell, 2009. 138(3): p. 592-603.

Lim, E., et al., Aberrant luminal progenitors as the candidate target population for basal tumor development in BRCA1 mutation carriers. Nat Med, 2009. 15(8): p. 907-13.

Molyneux, G., et al., BRCA1 Basal-like Breast Cancers Originate from Luminal Epithelial Progenitors and Not from Basal Stem Cells. Cell Stem Cell, 2010. 7(3): p. 403-417.

Hennighausen, L. and G.W. Robinson, Information networks in the mammary gland. Nat Rev Mol Cell Biol, 2005. 6(9): p. 715-25.

Kordon, E.C. and G.H. Smith, An entire functional mammary gland may comprise the progeny from a single cell. Development, 1998. 125(10): p. 1921-30.

Asselin-Labat, M.L., et al., Gata-3 is an essential regulator of mammary-gland morphogenesis and luminal-cell differentiation. Nat Cell Biol, 2007. 9(2): p. 201-9.

Kouros-Mehr, H., et al., GATA-3 maintains the differentiation of the luminal cell fate in the mammary gland. Cell, 2006. 127(5): p. 1041-55.

Khaled, W.T., et al., The IL-4/IL-13/Stat6 signalling pathway promotes luminal mammary epithelial cell development. Development, 2007. 134(15): p. 2739-50.

Li, P., et al., Reprogramming of T cells to natural killer-like cells upon Bcl11b deletion. Science, 2010. 329(5987): p. 85-9.

Wood, L.D., et al., The genomic landscapes of human breast and colorectal cancers. Science, 2007. 318(5853): p. 1 108-13.

Fiering, S., et al., Targeted deletion of 5'HS2 of the murine beta-globin LCR reveals that it is not essential for proper regulation of the beta-globin locus. Genes Dev, 1995. 9(18): p. 2203-13.

Hameyer, D., et al., Toxicity of ligand-dependent Cre recombinases and generation of a conditional Cre deleter mouse allowing mosaic recombination in peripheral tissues. Physiol Genomics, 2007. 31(1 ): p. 32-41.

Ishikawa, F., et al., Development of functional human blood and immune systems in NOD/SCID/IL2 receptor {gamma} chain(null) mice. Blood, 2005. 106(5): p. 1565-73. Comijn, J., et al., The two-handed E box binding zinc finger protein SIP1 downregulates E-cadherin and induces invasion. Mol Cell, 2001. 7(6): p. 1267-78.

Taube, J.H., et al., Core epithelial-to-mesenchymal transition interactome gene- expression signature is associated with claudin-low and metaplastic breast cancer subtypes. Proc Natl Acad Sci U S A, 2010. 107(35): p. 15449-54.

Wellner, U., et al., The EMT-activator ZEB1 promotes tumorigenicity by repressing sternness-inhibiting microRNAs. Nat Cell Biol, 2009. 11(12): p. 1487-95.

Raouf, A., et al., Transcriptome analysis of the normal human mammary cell commitment and differentiation process. Cell Stem Cell, 2008. 3(1 ): p. 109-18.

Visvader, J.E., Keeping abreast of the mammary epithelial hierarchy and breast tumorigenesis. Genes Dev, 2009. 23(22): p. 2563-77.

Montesano, R., et al., Isolation of EpH4 mammary epithelial cell subpopulations which differ in their morphogenetic properties. In Vitro Cell Dev Biol Anim, 1998. 34(6): p. 468- 77.

Avram, D., et al., COUP-TF (chicken ovalbumin upstream promoter transcription factor)- interacting protein 1 (CTIP1) is a sequence-specific DNA binding protein. Biochem J, 2002. 368(Pt 2): p. 555-63.

Kouzarides, T., Chromatin modifications and their function. Cell, 2007. 128(4): p. 693- 705.

Elenbaas, B., et al., Human breast cancer cells generated by oncogenic transformation of primary mammary epithelial cells. Genes Dev, 2001. 15(1 ): p. 50-65.

Wang, W., et al., Chromosomal transposition of PiggyBac in mouse embryonic stem cells. Proc Natl Acad Sci U S A, 2008. 105(27): p. 9290-5.

Al-Hajj, M., et al., Prospective identification of tumorigenic breast cancer cells. Proc Natl Acad Sci U S A, 2003. 100(7): p. 3983-8. Gregory, P.A., et al., The miR-200 family and miR-205 regulate epithelial to mesenchymal transition by targeting ZEB1 and SIP1. Nat Cell Biol, 2008. 10(5): p. 593- 601.

Sorlie, T., et al., Repeated observation of breast tumor subtypes in independent gene expression data sets. Proc Natl Acad Sci U S A, 2003. 100(14): p. 8418-23.

Oakes, S.R., et al., The Ets transcription factor Elf5 specifies mammary alveolar cell fate. Genes Dev, 2008. 22( 5): p. 581-6.

Krivtsov, A.V., et al., Transformation from committed progenitor to leukaemia stem cell initiated by MLL-AF9. Nature, 2006. 442(7104): p. 818-22.

Prat, A., et al., Phenotypic and molecular characterization of the claudin-low intrinsic subtype of breast cancer. Breast Cancer Res, 2010. 12(5): p. R68.

Mehra, R., et al., Identification of GATA3 as a breast cancer prognostic marker by global gene expression meta-analysis. Cancer Res, 2005. 65(24): p. 1 1259-64.

Miller, L.D., et al., An expression signature for p53 status in human breast cancer predicts mutation status, transcriptional effects, and patient survival. Proc Natl Acad Sci U S A, 2005. 102(38): p. 13550-5.

Hess, K.R., et al., Pharmacogenomic predictor of sensitivity to preoperative chemotherapy with paclitaxel and fluorouracil, doxorubicin, and cyclophosphamide in breast cancer. J Clin Oncol, 2006. 24(26): p. 4236-44.

Pawitan, Y., et al., Gene expression profiling spares early breast cancer patients from adjuvant therapy: derived and validated in two population-based cohorts. Breast Cancer Res, 2005. 7(6): p. R953-64.

Farmer, P., et al., A stroma-related gene signature predicts resistance to neoadjuvant chemotherapy in breast cancer. Nat Med, 2009. 15(1 ): p. 68-74.

Sims, D., et al., ROCK: a breast cancer functional genomics resource. Breast Cancer Res Treat, 2010. 124(2): p. 567-72.

Dontu, G., et al., In vitro propagation and transcriptional profiling of human mammary stem/progenitor cells. Genes Dev, 2003. 17(10): p. 1253-70.

Reichmann, E., et al., New mammary epithelial and fibroblastic cell clones in coculture form structures competent to differentiate functionally. J Cell Biol, 1989. 108(3): p. 1 127- 38.

Nakamura, T., et al., Evi9 encodes a novel zinc finger protein that physically interacts with BCL6, a known human B-cell proto-oncogene product. Mol Cell Biol, 2000. 20(9): p. 3178-86.

Satterwhite, E., et al., The BCL11 gene family: involvement of BCL11A in lymphoid malignancies. Blood, 2001. 98(12): p. 3413-20.

Ernst, P., et al., Definitive hematopoiesis requires the mixed-lineage leukemia gene. Dev Cell, 2004. 6(3): p. 437-43.

Jaiswal, S., et al., CD47 is upregulated on circulating hematopoietic stem cells and leukemia cells to avoid phagocytosis. Cell, 2009. 138(2): p. 271-85.

Zhang, Y., et al., The G protein-coupled receptor 87 is necessary for p53-dependent cell survival in response to genotoxic stress. Cancer Res, 2009. 69(15): p. 6049-56.

Selcuklu, S.D., M.T. Donoghue, and C. Spillane, miR-21 as a key regulator of oncogenic processes. Biochem Soc Trans, 2009. 37(Pt 4): p. 918-25.

Malumbres, M. and M. Barbacid, RAS oncogenes: the first 30 years. Nat Rev Cancer, 2003. 3(6): p. 459-65.

Izumchenko, E., et al., NEDD9 promotes oncogenic signaling in mammary tumor development. Cancer Res, 2009. 69(18): p. 7198-206.

Cox, A., et al., A common coding variant in CASP8 is associated with breast cancer risk. Nat Genet, 2007. 39(3): p. 352-8.

Eckfeld, K., et al., RASSF4/AD037 is a potential ras effector/tumor suppressor of the RASSF family. Cancer Res, 2004. 64(23): p. 8688-93.

Chen, I.F., et al., AIM2 suppresses human breast cancer cell proliferation in vitro and mammary tumor growth in a mouse model. Mol Cancer Ther, 2006. 5(1 ): p. 1-7.

Mani, S.A., et al., Mesenchyme Forkhead 1 (FOXC2) plays a key role in metastasis and is associated with aggressive basal-like breast cancers. Proc Natl Acad Sci U S A, 2007. 104(24): p. 10069-74. 63. Aigner, K., et al., The transcription factor ZEB1 ([delta]EF1) promotes tumour cell dedifferentiation by repressing master regulators of epithelial polarity. Oncogene, 2007. 26(49): p. 6979-6988.

64. Vousden, K.H. and X. Lu, Live or let die: the cell's response to p53. Nat Rev Cancer, 2002. 2(8): p. 594-604.

65. Carey, L, et al., Triple-negative breast cancer: disease entity or title of convenience? Nat Rev Clin Oncol, 2010. 7(12): p. 683-92.

66. Chene, P., Inhibiting the p53-MDM2 interaction: an important target for cancer therapy.

Nat Rev Cancer, 2003. 3(2): p. 102-9.

67. Cicalese, A., et al., The tumor suppressor p53 regulates polarity of self-renewing divisions in mammary stem cells. Cell, 2009. 138(6): p. 1083-95.

68. Chang, C.J., et al., p53 regulates epithelial-mesenchymal transition and stem cell properties through modulating miRNAs. Nat Cell Biol, 201 1 . 13(3): p. 317-23.

69. Forbes, S.A., et al., The Catalogue of Somatic Mutations in Cancer (COSMIC). Curr Protoc Hum Genet, 2008. Chapter 10: p. Unit 10 1 1.

70. Junk, D.J., et al., Different mutant/wild-type p53 combinations cause a spectrum of increased invasive potential in nonmalignant immortalized human mammary epithelial cells. Neoplasia, 2008. 10(5): p. 450-61.

71. Foulkes, W.D., I.E. Smith, and J.S. Reis-Filho, Triple-negative breast cancer. N Engl J Med, 201 1. 363(20): p. 1938-48.

72. Blows, F.M., et al., Subtyping of breast cancer by immunohistochemistry to investigate a relationship between subtype and short and long term survival: a collaborative analysis of data for 10, 159 cases from 12 studies. PLoS Med. 7(5): p. e1000279.

73. Liu, Y., et al., p53 regulates hematopoietic stem cell quiescence. Cell Stem Cell, 2009.

4(1 ): p. 37-48.

74. Abbas, T. and A. Dutta, p21 in cancer: intricate networks and multiple activities. Nat Rev Cancer, 2009. 9(6): p. 400-14.

75. Yin, B., et al., A retroviral mutagenesis screen reveals strong cooperation between Bcl11a overexpression and loss of the Nf1 tumor suppressor gene. Blood, 2008.

76. Luo, J., et al., Negative control of p53 by Sir2alpha promotes cell survival under stress.

Cell, 2001. 107(2): p. 137-48.

77. Abbas, H.A., et al., Mdm2 is required for survival of hematopoietic stem cells/progenitors via dampening of ROS-induced p53 activity. Cell Stem Cell, 2010. 7(5): p. 606-17.

78. Knappskog, S., et al., The MDM2 promoter SNP285C/309G haplotype diminishes Sp1 transcription factor binding and reduces risk for breast and ovarian cancer in Caucasians. Cancer Cell, 201 1. 19(2): p. 273-82.

79. Vassilev, L.T., et al., In vivo activation of the p53 pathway by small-molecule antagonists of MDM2. Science, 2004. 303(5659): p. 844-8.

80. Lambert, J.M., et al., PRIMA-1 reactivates mutant p53 by covalent binding to the core domain. Cancer Cell, 2009. 15(5): p. 376-88.

81. Wiman, K.G., Pharmacological reactivation of mutant p53: from protein structure to the cancer patient. Oncogene, 2010. 29(30): p. 4245-52.

82. Chan, W., et al., A recombineering based approach for high-throughput conditional knockout targeting vector construction. Nucleic Acids Res, 2007. 35(8): p. e64.

83. Liu, P., N.A. Jenkins, and N.G. Copeland, A highly efficient recombineering-based method for generating conditional knockout mutations. Genome Res, 2003. 13(3): p. 476- 84.

84. Sankaran, V.G., et al., Developmental and species-divergent globin switching are driven by BCL11A. Nature, 2009. 460(7259): p. 1093-7.

85. Wang, X. and B. Seed, A PCR primer bank for quantitative gene expression analysis.

Nucleic Acids Res, 2003. 31(24): p. e154.

Claims

Claims

1 An inhibitor of BCL1 1 A for use in the prevention or treatment of cancer.

2 An inhibitor of BCL1 1A according to claim 1 for use in the prevention or treatment of breast cancer.

3 An inhibitor according to claim 1 or 2 wherein the inhibitor is an RNA.

4 An inhibitor according to claim 1 or 2 wherein the inhibitor is an antibody or fragment thereof.

5 An inhibitor according to any preceding claim which has a direct inhibitory effect on the BCL1 1 A gene or gene product expression or activity.

6 An inhibitor according to any preceding claim which acts on a target upstream or downstream of BCL1 1A in a biological pathway, and which is able to cause an effect on the downstream which is equivalent to direct inhibition.

7 An inhibitor according to claim 6 which is an inhibitor of Notch 3, or an inhibitor of EMT or an inhibitor of an EMT inducer such as ZEB1 , ZEB2, FOXC2 and TWIST2.

8 A pharmaceutical composition comprising an inhibitor according to any preceding claim and a pharmaceutically acceptable excipient.

9 A method of prevention or treatment of breast cancer in an individual in need thereof, the method comprising delivery of an effective amount of an inhibitor of BCL1 1 A.

10 Use of an effective amount of an inhibitor of BCL1 1A in the preparation of a medicament for prevention or treatment of breast cancer in an individual in need thereof.

1 1 A method or use according to claim 10 or 1 1 wherein the individual has not been diagnosed with cancer.

12 A method or use according to claim 10 or 1 1 wherein the individual has been diagnosed with breast cancer but not the specific subtype of cancer.

13 A method or use according to claim 10 or 1 1 wherein the individual has been diagnosed with basal breast cancer.

14 An inhibitor, use, composition or method according to claims 1-10 wherein the breast cancer is basal breast cancer.

15 A method for assessing the severity and/or prognosis of a breast cancer, the method comprising determining the level of expression or activity of the bcl1 1A gene or BCL1 1A gene product in the breast cancer cell. 16 A method for screening for and/or detecting breast cancer, the method comprising determining the level of expression or activity of the BCL1 1A gene or gene product in a breast cell.

17 A diagnostic kit comprising a detection reagent for assay of the BCL1 1A gene or gene product expression or activity.

18 miR-200c, for use in the prevention or treatment of breast cancer, such as basal breast cancer.

19 An activator of bell 1A gene expression or gene product activity for use in the growth of, and/or conversion of cells to, mammary stem cells (MaSCs)

20 An activator of bcl1 1A gene expression or gene product activity for use in breast tissue regeneration.

21 A method for generating mammary stem cells specific for an individual by expressing and /or activating the BCL1 1 A gene or polypeptide in cells from that individual.

22 A method for screening drugs that inhibit breast cancer, the method comprising contacting a cell derived from a MaSC obtained by modulation of the expression of BCL1 1A with a drug, to identify a drug which cause inhibition of cancerous cell development or growth.

23 An imaging agent for imaging or screening for breast cancer, the agent being specific for the detection of BCL1 1A, either nucleic acid or protein or activity thereof, optionally able to distinguish cells which over-express BCL1 1 A from those with normal levels of BCL1 1 A.

24 A cytopathic agent for treatment of a cancer cell, such as a breast cancer cell, the agent comprising a component capable of killing a cell and a component having specificity for a cell with over-expression of BCL1 1 A or a downstream effector thereof.

25 An inhibitor of BCL1 1A in combination with p53, or an activator of p53, for use in the prevention or treatment of breast cancer.

26 An inhibitor of BCL1 1 A in combination with p53, or an activator of p53.

27 An inhibitor of BCL1 1 A in combination with p53, or an activator of p53, in the preparation of a medicament for use in the prevention or treatment of breast cancer,

28 A method for the prevention or treatment of breast cancer, the method comprising delivery of a combination of an inhibitor of BCL1 1A and p53, or an activator of p53, to an individual in need thereof.