(WT1) Have Distinct Effects on Mammary Epithelial Cells

Home , Estrogen receptor alpha

Oncogene (2007) 26, 3423–3430 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE Isoforms of Wilms’ tumor suppressor gene (WT1) have distinct effects on mammary epithelial cells

EA Burwell, GP McCarty, LA Simpson, KA Thompson and DM Loeb

Division of Pediatric Oncology, Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University, Baltimore, MD, USA

The role of WT1 (Wilm’s tumor suppressor gene) in breast tumors, and they observed a correlation between loss of cancer is controversial, with evidence for both tumor- WT1 expression and expression of the estrogen receptor promoting andtumor-suppressing activities. In orderto (ER). They, therefore, concluded that WT1 plays a address this question, we expressed different WT1 tumor suppressor role in breast cancer. Consistent with isoforms in the mammary epithelial cell line H16N-2, this was a report by Zhang et al. (2003) that WT1 which does not express endogenous WT1. Cells were suppresses the clonal growth of MDA-MB-231 cells in stably transfectedwith either WT1 ( ÀEx5/ÀKTS) or soft agar. In contrast, our group reported overexpres- WT1 ( þ Ex5/ þ KTS) under the control of the inducible sion of WT1 in breast cancer (Loeb et al., 2001). Using metallothionein promoter. Induction of WT1 (ÀEx5/ reverse transcription–polymerase chain reaction (RT– ÀKTS) upregulatedp21, causing a slowing of prolifera- PCR), we were only able to detect WT1 expression in tion anda G2-phase cell cycle arrest. In artificial one of 20 samples of mammary epithelium, but we basement membrane, the WT1 (ÀEx5/ÀKTS) isoform detected WT1 expression in 27 of 31 primary breast promotedthe appearance of highly organizedacinar carcinoma specimens. Subsequently, Miyoshi et al. cellular aggregates. In contrast, WT1 ( þ Ex5/ þ KTS) (2002) reported a correlation between increased WT1 hadno effect on p21 or proliferation, but rather causedan expression and poor long-term survival in women with epithelial–mesenchymal transition anda redistribution of breast cancer, and Zapata-Benavides et al. (2002) E-cadherin from the cell membrane to the cytoplasm. This reported a correlation between WT1 expression levels isoform also causes the cellular aggregates growing in and the proliferative rates of breast cancer cell lines. artificial basement membrane to appear significantly less Taken together, these data suggest that WT1 might organizedthan control cells. Thus, different WT1 iso- function as a tumor-promoting gene in breast cancer. forms have distinct effects in this cell line, suggesting that Complex transcriptional and translational regulation depending on the ratio of WT1 isoform expression in of the WT1 gene results in the expression of several mammary epithelial cells, WT1 couldfunction to either protein isoforms. The most abundant of these isoforms promote or suppress a transformedphenotype. result from two independent alternative splicing events Oncogene (2007) 26, 3423–3430. doi:10.1038/sj.onc.1210127; (Haber et al., 1991). One of these alternative splices published online 11 December 2006 includes or excludes exon 5 from the mature mRNA, resulting in proteins that contain or lack a 17 amino acid Keywords: breast cancer; oncogene; epithelial–mesenchymal segment in the N-terminal transcriptional regulatory transition;cellcycle;alternativesplicing domain. The other alternative splice involves the use of one of the two distinct splice acceptor sites at the N-terminus of exon 10, resulting in the inclusion of three additional amino acids (KTS) in the larger protein Introduction isoform, altering the spacing of the zinc-fingers that compose the DNA-binding domain. The presence or The WT1 (Wilm’s tumor suppressor gene) gene encodes absence of this so-called KTS insert has profound a zinc finger transcription factor that has been functional consequences for the mature protein, affect- implicated in the pathogenesis of breast cancer. Silber- ing both DNA-binding and subcellular localization, but stein et al. (1997) reported that WT1 is expressed in the functional importance of exon 5 remains unclear mammary epithelial cells, with higher levels of expres- (Larsson et al., 1995; Davies et al., 1998; Natoli et al., sion in myoepithelium than in luminal epithelium. They 2002; Reynolds et al., 2003). also reported a loss of WT1 expression in most breast Variable expression of WT1 isoforms in breast tumors has been reported (Silberstein et al., 1997; Loeb et al., Correspondence: Dr DM Loeb, Division of Pediatric Oncology, 2001). We, therefore, hypothesized that different WT1 Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins isoforms might have different effects on mammary University, Bunting-Blaustein Cancer Research Building, 1650 Orleans epithelial cells, thus accounting for the conflicting St., Baltimore, MD 21231, USA. E-mail: [email protected] reports of the role of this gene in the development of Received 16 December 2005; revised 13 September 2006; accepted 4 October breast cancer. To test this hypothesis, we expressed two 2006; published online 11 December 2006 distinct WT1 isoforms in the H16N-2 cell line. This Effects of WT1 isoforms on breast epithelial cells EA Burwell et al 3424 immortalized, but not transformed, mammary epithelial cell line does not express endogenous WT1 (Loeb et al., 2001). Because they are the most structurally divergent isoforms, we chose to investigate the role of the WT1 isoform containing both exon 5 and the KTS insert (designated WT1 ( þ Ex5/ þ KTS)) and the isoform lacking both these alternatively spliced sequences (designated WT1 (ÀEx5/ÀKTS)). We found that WT1 (ÀEx5/ÀKTS) inhibited proliferation and promoted an Figure 1 WT1 expression in stably transfected cell lines. Cells were treated overnight with ( þ ) or without (À) 100 mM ZnCl2 to acinar growth pattern in cells surrounded by extra- induce WT1 expression, and cell lysates were subjected to Western cellular matrix, characteristics of a tumor suppressor blotting with anti-WT1.(a) Cell lines expressing WT1 (ÀEx5/ gene, whereas WT1 ( þ Ex5/ þ KTS) caused a morpho- ÀKTS) and (b) Cell lines expressing WT1 ( þ Ex5/ þ KTS). Equal logical transition from an epithelial to a more mesench- protein loading was confirmed by Western blotting with antibody ymal phenotype, accompanied by alterations in the against tubulin. expression of E-cadherin and vimentin, consistent with a tumor-promoting function. Gene expression profiling hypothesis, we performed proliferation assays using cell revealed that these WT1 isoforms induced distinct gene lines stably transfected with the WT1 isoforms under the expression patterns consistent with the observed pheno- control of the metallothionein promoter. Cells were typic differences. These data support the hypothesis plated with or without ZnCl2 to induce WT1 expression, that different WT1 isoforms have distinct effects on and relative cell number was determined after 48 h. mammary epithelial cells. Induction of WT1 (ÀEx5/ÀKTS) resulted in a significant slowing of proliferation, induction of WT1 ( þ Ex5/ þ KTS) had no effect, and the addition of ZnCl2 to Results control cells did not affect their proliferation (Figure 2a). Cell cycle analysis revealed that WT1 Establishment of stable transfectants (ÀEx5/ÀKTS) causes an accumulation of cells in the H16N-2 cells were transfected with one of four different G2 phase of the cell cycle, but the addition of ZnCl2 to plasmids, and stable transfectants were isolated by the other cell lines had no effect on cell cycle distribution selection with G418. The plasmids designated (Figure 2b). Thus, WT1 (ÀEx5/ÀKTS) slows the pCB6WT(À/À) and pCB6WT( þ / þ ) direct the expres- proliferation of H16N-2 cells through induction of a sion of WT1 isoforms ((ÀEx5/ÀKTS) and ( þ Ex5/ G2-cell cycle arrest, whereas WT1 ( þ Ex5/ þ KTS) does þ KTS), respectively) under the control of the consti- not affect proliferation. tutively active cytomegalovirus (CMV) immediate-early WT1 (ÀEx5/ÀKTS) upregulates p21 expression promoter. Plasmids designated pMTWT(À/À) and in Saos2 cells (Englert et al., 1997), but not in 32D cl3 pMTWT( þ / þ ) direct the expression of the same cells (Loeb et al., 2002). Because p21 causes a cell cycle WT1 isoforms under the control of the zinc-inducible arrest and is a WT1 target gene in some cell lines, metallothionein promoter. Although cell lines expres- we investigated whether upregulation of p21 accompa- sing WT1 (ÀEx5/ÀKTS) under the control of the nied the WT1 (ÀEx5/ÀKTS)-induced cell cycle arrest metallothionein promoter were readily obtained, only in H16N-2 cells. As determined by western blot- one stably transfected clone that constitutively expressed ting, induction of WT1 (ÀEx5/ÀKTS), but not WT1 this isoform was obtained, despite numerous attempts. ( þ Ex5/ þ KTS), upregulated p21 expression In contrast, WT1 ( þ Ex5/ þ KTS) was easily expressed (Figure 2c). These data strongly suggest that WT1 both constitutively and in an inducible form. Cell lines (ÀEx5/ÀKTS) induces a G2-phase cell cycle arrest with inducible expression of WT1 (ÀEx5/ÀKTS) were through upregulation of p21. designated H16NMTWTA1, A2, and A3 and cell lines with inducible expression of WT1 ( þ Ex5/ þ KTS) were Effect of WT1 isoforms on morphology designated H16NMTWTD1, D2 and D3. Control cell It was difficult to generate clones constitutively expres- lines were designated H16NMT1, H16NMT2 and sing WT1 (ÀEx5/ÀKTS), and we also noticed that the H16NMT3. Western blotting was performed to verify cells expressing WT1 ( þ Ex5/ þ KTS) had a different expression of transfected WT1 proteins (Figure 1). morphology from control cells and from cells expressing Because there was only one clone constitutively expres- WT1 (ÀEx5/ÀKTS). Control cells grow with an sing WT1 ( Ex5/ KTS), further work was limited to À À epithelial morphology, tending to cluster together and the inducible expression system. Data obtained with form a regular, cobblestone pattern (Figure 3a). Cells H16NMTWTA1 and H16NMTWTD1 cells are shown, stably transfected with inducible WT1 (ÀEx5/ÀKTS) but all results were confirmed in multiple subclones. grow in a similar pattern (Figure 3b). In contrast, cells transfected with WT1 ( þ Ex5/ þ KTS) tend to grow Effect of WT1 isoforms on proliferation as individual cells, rather than in clusters, and tend to We hypothesized that it was difficult to obtain cell lines have a more spindle-like morphology (Figure 3c). The that constitutively express WT1 (ÀEx5/ÀKTS) because addition of ZnCl2 to cultures of the transfected cells did this isoform interferes with proliferation. To test this not alter their morphology, but this probably reflects

Oncogene Effects of WT1 isoforms on breast epithelial cells EA Burwell et al 3425

Figure 2 Effect of WT1 isoforms on proliferation. (a) Cells were treated with ( þ ) or without (À)50mM ZnCl2 for 48 h. Relative cell number was determined by performing an MTT assay. (b) Cells were grown for 48 h in the presence or absence of 50 mM ZnCl2, harvested in a lysis solution containing Hoechst 33258, and DNA content was analysed by flow cytometry. (c) H16NMT1 (control), H16NMTWTD1 ( þ E5/ þ K), and H16NMTWTA1 (ÀE5/ÀK) cells were treated overnight with ( þ ) or without (À)50mM ZnCl2. Cell lysates were subjected to Western blotting with anti-p21. Equal Figure 3 WT1 ( þ Ex5/ þ KTS) affects morphology of cells protein loading was confirmed by Western blotting with antibody growing in two-dimensional culture. Photomicrographs of against tubulin. H16NMT1 (a), H16NMTWTA1 (b) and H16NMTWTD1 cells (c) grown on tissue culture plates in 50 mM ZnCl2. leaky WT1 expression from the metallothionein promoter. It is unlikely that this morphologic distinction is bottom and grew on plastic surrounded by the dense a coincidence, unrelated to WT1 expression, because the extracellular matrix, and the others that grew suspended morphology was noted in all six clones expressing within the matrix. After 4–5 days in culture, there were inducible WT1 ( þ Ex5/ þ KTS) and in none of the distinct morphological differences among the cell lines. other cell lines. This suggests that expression of the The control cells that settled to the bottom grew in WT1 ( þ Ex5/ þ KTS) isoform might participate in an loose, disorganized clusters (Figure 4a). Cells expressing epithelial–mesenchymal transition. WT1 ( þ Ex5/ þ KTS), in contrast, grew with a spindle- In addition to tissue culture plastic, we also evaluated like morphology reminiscent of the morphology of these the morphology of transfected cells in Matrigel, a cells growing on plastic (Figure 4b). The cells expressing solubilized basement membrane preparation that more WT1 (ÀEx5/ÀKTS) grew in more tightly adherent closely mimics in vivo growth conditions. We observed clusters, with a flatter, more epithelial morphology two distinct populations of cells: one that settled to the (Figure 4c). Thus, cells growing on a flat surface but

Oncogene Effects of WT1 isoforms on breast epithelial cells EA Burwell et al 3426

Figure 4 WT1 isoforms affect the morphology of cells growing in three-dimensional culture in artiﬁcial basement membrane. H16NMT1 (a and d), H16NMTWTD1 (b and e) and H16NMTWTA1 cells (c and f) cells were grown in Matrigel supplemented with 50 mM ZnCl2. Photomicrographs of cells that settled to the bottom of the tissue culture dish (a, b and c) and cells suspended in Matrigel (d, e and f) are shown.

surrounded by extracellular matrix exhibited WT1 an epithelial–mesenchymal transition in mammary isoform-dependent morphologic differences just as the epithelial cells. One of the hallmarks of an cells growing in standard tissue culture did. epithelial–mesenchymal transition is the redistribution The morphologic differences of the cells growing in of E-cadherin from the cell membrane to the cytoplasm. suspension were quite similar. The control cells grew in Immunocytochemistry was, therefore, performed to loosely adherent clusters with little obvious organization determine if WT1 affects the subcellular distribution (Figure 4d). The WT1 ( þ Ex5/ þ KTS) transfectants of E-cadherin. Interestingly, whereas the subcellular grew in an even more disorganized manner, and many of distribution of b-catenin was unaffected by WT1 (data the clusters had spindle-like cells growing out from them not shown), there was a marked alteration in the (Figure 4e). In contrast, the cells transfected with WT1 localization of E-cadherin in cells expressing WT1 (ÀEx5/ÀKTS) grew in very tightly adherent clusters, ( þ Ex5/ þ KTS). In both control cells and cells expres- with a highly organized cellular arrangement remini- sing WT1 (ÀEx5/ÀKTS), E-cadherin is expressed scent of mammary acini (Figure 4f). Clusters such as almost exclusively at the cell surface (Figure 5a and b). these were never observed in cultures of control cells or In contrast, in cells expressing WT1 ( þ Ex5/ þ KTS), cultures of cells expressing WT1 ( þ Ex5/ þ KTS). Thus, immunofluorescence reveals a more diffuse expression the morphology of H16N-2 cells is profoundly affected pattern of E-cadherin, with much less distinct localiza- by WT1 expression in an isoform-specific manner. tion to the cell membrane (Figure 5c). We also evaluated E-cadherin expression levels by Western blotting and found that WT1 (ÀEx5/ÀKTS) modestly upregulated Molecular correlates of altered morphology expression, quantified by scanning densitometry as a The morphologic changes that we observed suggest 60% increase in protein level, whereas induction of WT1 the possibility that WT1 ( þ Ex5/ þ KTS) can initiate ( þ Ex5/ þ KTS) had no effect (data not shown).

Oncogene Effects of WT1 isoforms on breast epithelial cells EA Burwell et al 3427 Mesenchymal cells are also characterized by increased constitutively expressing WT1 (ÀEx5/ÀKTS) matched expression of the intermediate filament protein vimen- the profile of cells in which expression of this isoform tin. Neither WT1 isoform directly affected vimentin had been induced by ZnCl2 (data not shown). Genes expression, nor cells expressing WT1 ( þ Ex5/ þ KTS) whose expression was altered in only one cell line expressed markedly more vimentin than did control cells expressing a particular isoform, or whose expression was or cells expressing WT1 (ÀEx5/ÀKTS) (Figure 5d). altered by the addition of ZnCl2 to cultures of the These observations are consistent with the morphologic control cell lines, were not considered to be affected by changes we observed. Taken together, the increased WT1. Representative genes are reported in the Supple- expression of vimentin and redistribution of E-cadherin mentary Information, and a full description of gene support the idea that WT1 ( þ Ex5/ þ KTS) induces a expression profiling of WT1 isoforms in these cells will mesenchymal phenotype in these cells. be published separately. We used RT–PCR to indepen- To further evaluate the effects of different WT1 dently confirm the altered regulation of a number of the isoforms, we performed gene expression profiling. We genes identified in these experiments. For example, using isolated RNA from cell lines expressing WT (ÀEx5/ semiquantitative RT–PCR, we were able to confirm ÀKTS) or WT1 ( þ Ex5/ þ KTS) before and after upregulation of B-raf by both isoforms of WT1 induction of WT1. Control cells, with or without ZnCl2, (Figure 6a), and using quantitative RT–PCR we were used for comparison. Of the 440 genes evaluable, confirmed that WT1 ( þ Ex5/ þ KTS), but not WT1 36 (8.2%) were upregulated by both WT1 isoforms, 14 (ÀEx5/ÀKTS), upregulates ribosomal protein S6 kinase (3.2%) were upregulated by WT1 ( þ Ex5/ þ KTS) but expression more than the 2.5-fold threshold (Figure 6b). not WT1 (ÀEx5/ÀKTS), and one was upregulated by The largest subset of genes regulated by WT1 is those WT1 (ÀEx5/ÀKTS) but not WT1 ( þ Ex5/ þ KTS). In regulated by both isoforms. Twenty-four of these 36 addition, seven (1.6%) were downregulated by both genes (67%) have been firmly implicated in either isoforms, whereas two were downregulated by either one normal mammary gland biology, mammary carcino- isoform or the other. Genes were defined as upregulated genesis, or both the processes. Interestingly, the one or downregulated if their expression was altered by gene upregulated by WT1 (ÀEx5/ÀKTS) and not by WT1 at least 2.5-fold upon addition of ZnCl2 to cultures of ( þ Ex5/ þ KTS), albumin, is a gene associated with both cell lines expressing the indicated WT1 isoform mammary glad differentiation – albumin being a major under the control of the metallothionein promoter. In protein found in colostrum (Velona et al., 1999). One of addition, the gene expression profile of the cells the two genes downregulated by this isoform alone,

Figure 5 WT1 affects localization of E-cadherin and expression of vimentin. H16NMT1 (a), H16NMTWTA1 (b) and H16NMTWTD1 (c) cells were grown on glass slides in the presence of 75 mM ZnCl2. Immunocytochemistry was performed with an Figure 6 WT1 upregulates B-raf and ribosomal protein S6 kinase. antibody against E-cadherin. There was no staining with an isotype (a) RNA was isolated from the indicated cell lines treated with ( þ ) control (data not shown). In the cells expressing WT1 (ÀEx5/ or without (À)50mM ZnCl2 and reverse transcribed. The cDNA ÀKTS), there is increased staining that is localized to the periphery was used for PCR using primers specific for B-raf. After 30 cycles, of the cells. In the cells expressing WT1 ( þ Ex5/ þ KTS) there is PCR products were evaluated by agarose gel electrophoresis. Equal decreased staining and the most intense staining is cytoplasmic. For loading of cDNA in each reaction was confirmed by performing the experiment in (d), H16NMT1 (control), H16NMTWTD1 PCR using primers specific for the ribosomal RNA 36B4. (b) The ( þ Ex5/ þ KTS) and H16NMTWTA1 (ÀEx5/ÀKTS) cells were same cDNA was used for quantitative RT–PCR using primers treated overnight with ( þ ) or without (À)50mM ZnCl2. Cell specific for ribosomal protein S6 kinase. Reactions were nor- lysates were subjected to Western blotting with antibody against malized by co-amplification of the housekeeping gene GAPDH. vimentin. Equal protein loading was confirmed by Western blotting The dashed line indicates the 2.5-fold amplification that was used with antibody against tubulin. to define gene induction as discussed in the text.

Oncogene Effects of WT1 isoforms on breast epithelial cells EA Burwell et al 3428 HIV-1 Rev binding protein, has not been previously arrest and morphologic effects that we reported. The associated with breast cancer, but the other, transform- most striking effect of WT1 ( þ Ex5/ þ KTS) in H16N-2 ing growth factor-a, is thought to be an autocrine cells is on their morphology, which can be at least growth factor for breast cancer, acting by binding to and partially explained by a redistribution of E-cadherin in activating epidermal growth factor receptors (Lo et al., the WT1-expressing cells. WT1 might directly affect the 2005). Thus, the gene expression changes induced by expression of a gene whose product is critical for proper WT1 (ÀEx5/ÀKTS) are consistent with its phenotypic localization of E-cadherin in adherens junctions. Alter- effects opposing mammary carcinogenesis. In contrast, natively, WT1 ( þ Ex5/ þ KTS) could affect which among the genes upregulated by WT1 ( þ Ex5/ þ KTS) E-cadherin splice forms are expressed. This isoform of but not by WT1 (ÀEx5/ÀKTS) are genes encoding WT1 has been implicated in regulation of alternative hyaluronoglucosaminidase (Udabage et al., 2005), ribo- splicing (Davies et al., 1998), and perhaps this activity is somal protein S6 kinase (Barlund et al., 2000), vascular responsible for the observed effect on E-cadherin endothelial growth factor (Kolch et al., 1995) and erbB3 localization. This would be consistent with a lack of (Lemoine et al., 1992) that have been implicated in the effect of WT1 (ÀEx5/ÀKTS), which has not been pathogenesis of breast cancer. Thus, the gene expression implicated in the regulation of mRNA splicing. In changes we detected upon induction of individual WT1 addition to differential effects on transcription and isoforms reflect the phenotypic differences seen in cells RNA splicing, differential interactions with heterolo- expressing these isoforms. gous proteins might also contribute to the distinct phenotypic effects we observed. WT1 has been shown to physically interact with a number of other proteins, including estrogen receptor-a (Reizner et al., 2005), a Discussion transcriptional regulator with a profound impact on mammary epithelial cell biology, p53 (Maheswaran There are conflicting data in the literature regarding the et al., 1993), a critical gene in the development of breast role of WT1 in mammary carcinogenesis. Silberstein cancer, and others (Loeb, 2006). Future work will et al. (1997) reported loss of WT1 expression in breast investigate whether WT1 isoforms differ in these tumors, and Zhang et al. (2003) reported observations protein/protein interactions. consistent with a tumor suppressor role. In contrast, our We observed both qualitative and quantitative group reported overexpression of WT1 in breast tumors differences in function between the WT1 isoforms, and as compared with normal mammary epithelium (Loeb the combined effect of these differences probably et al., 2001), and Zapata-Benavides et al. (2002) underlies the distinct phenotypic changes they cause. reported data consistent with an oncogenic role. To The differential regulation of p21 protein expression is directly address these conflicting data, we expressed an example of a qualitative difference, and the effect on different WT1 isoforms in a mammary epithelial cell line RPS6K expression is a quantitative difference. Focusing that lacks endogenous WT1 expression. Our findings on these two genes, WT1 (ÀEx5/ÀKTS) causes a confirm our hypothesis that WT1 isoforms have distinct significant increase in p21 expression and only a modest functions in mammary epithelial cells. WT1 (ÀEx5/ increase in RPS6K expression, and the combined effect ÀKTS) slows proliferation, causes a cell cycle arrest in is a slowing of cell cycle progression. In contrast, WT1 G2, upregulates p21 and causes the cells to grow in an ( þ Ex5/ þ KTS) does not affect p21 expression but organized, acinar morphology in three-dimensional significantly upregulates RPS6K, accounting for the culture. In contrast, WT1 ( þ Ex5/ þ KTS) causes the lack of cell cycle arrest in cells expressing only this cells to assume a mesenchymal morphology, causes isoform. disorganized aggregation when grown with artificial How can these seemingly contradictory functions of basement membrane, and causes increased vimentin WT1 be integrated into a single model of WT1 function expression and a redistribution of E-cadherin from the in mammary carcinoma? We investigated the function of cell membrane into the cytoplasm, all characteristics individual WT1 isoforms expressed in isolation, but expected of an oncogene. Gene expression profiles of primary breast tumors frequently express multiple H16N-2 cells expressing different WT1 isoforms were isoforms (Silberstein et al., 1997; Loeb et al., 2001). consistent with biochemical and morphologic findings – Thus, the net effect of WT1 will probably reflect a WT1 ( þ Ex5/ þ KTS) differentially upregulated genes balance of the effects of different isoforms. In the breast like ribosomal S6 kinase that are associated with tumors our group studied, some tumors that expressed malignant transformation. Taken together, these data only isoforms containing exon 5 and the KTS insert strongly suggest that different WT1 isoforms have were found, and the remainder of the tumors expressed profoundly different effects on mammary epithelial cell both splice variants, but with a predominance of biology. isoforms containing these regions. Thus in all of the The mechanism by which WT1 (ÀEx5/ÀKTS) slows tumors, the major WT1 isoform, and the form that proliferation and upregulates differentiation-associated would be expected to dictate the phenotype of the cells, genes is most likely related to its role as a transcription was WT1 ( þ Ex5/ þ KTS) – the isoform with oncogenic factor. Both p21 and E-cadherin have previously been characteristics. shown to be WT1 target genes, and effects on these Our results echo findings reported by Hewitt and targets in H16N-2 cells could account for the cell cycle Saunders (1996), who noted that transfection of cells

Oncogene Effects of WT1 isoforms on breast epithelial cells EA Burwell et al 3429 with WT1 isoforms lacking exon 5 slowed their Immunofluorescence microscopy proliferation, whereas isoforms containing exon 5 Cells were incubated overnight with 75 mM ZnCl2, rinsed, fixed altered cellular morphology. Our results extend these and permeablized. After incubation with primary antibodies findings to a different lineage (mammary epithelial recognizing b-catenin or E-cadherin (both from BD Phar- cells), and begin to address the mechanism by which mingen, San Diego, CA, USA; catalog numbers 610153 and 610181), cells were rinsed and incubated with fluorescein- WT1 affects proliferation and morphology. The specific conjugated goat-anti-mouse secondary antibodies (Molecular role and importance of exon 5 in the function of WT1 is Probes Inc., Eugene, OR, USA). a matter of some debate. Natoli et al. (2002) demon- strated that mice engineered to only express WT1 isoforms lacking exon 5 develop normally and lactate Western blotting normally. Our findings do not directly address the Total cellular protein was run on 4–12% bis-tris gels using relevance of exon 5 for mammary carcinogenesis, MOPS buffer, blotted onto polyvinylidine difluoride (PVDF) membranes, and blocked overnight. Primary antibodies were because the isoforms we investigated differ at both diluted into blocking solution as follows: WT1 (Santa Cruz alternatively spliced sites. However, our results are Biotechnology, Santa Cruz, CA, USA, SC-192) 1:1000; p21 consistent with those of Natoli et al. (2002) – our data (BD Pharmingen, 556430) 1:500; tubulin (Sigma-Aldrich, St support a role of WT1 (ÀEx5/ÀKTS) in normal Louis, MO, USA, T9026) 1:10 000; Vimentin (Santa Cruz mammary development, including expression of albu- Biotechnology, SC 5565) 1:200. min, and suggest a role for WT1 ( þ Ex5/ þ KTS) in neoplastic transformation, but not normal breast development. Our findings are also in line with data Proliferation, cell cycle analysis and morphology MTT assays. MTT proliferation assays were performed regarding the role of WT1 in hematopoiesis, with according to manufacturer’s instructions (Chemicon, Temecu- reports that WT1 ( þ Ex5/ þ KTS) interferes with la, CA, USA). hematopoietic differentiation (Inoue et al., 1998), Cell cycle analysis. Cells were harvested in ‘Magic Solution’ whereas WT1 (ÀEx5/ÀKTS) potentiates this pathway (0.6% NP-40, 3.7% formaldehyde, 11 mg/ml Hoechst 33258 (Loeb et al., 2003). (Molecular Probes) in PBS), and absorbance quantified using a In summary, we found that in H16N-2 mammary FACScalibur (Becton Dickinson, Franklin Lakes, NJ, USA) epithelial cells, WT1 isoforms had distinct effects. WT1 and CellQuest Pro software (Becton Dickinson). (ÀEx5/ÀKTS) has some characteristics of a tumor suppressor gene: slowing proliferation and inducing a Gene expression profiling cell cycle arrest. In contrast, WT1 ( þ Ex5/ þ KTS) has cDNA was used to probe an array (OHS-802, SuperArray some characteristics of an oncogene: causing an Bioscience Corporation, Frederick, MD, USA) containing 440 epithelial–mesenchymal transition and decreasing the cancer-related genes. Reverse transcription, array hybridiza- cell membrane expression of E-cadherin. This difference tion, and data production were performed by SuperArray. in function of WT1 isoforms is consistent with the isoform expression pattern we observed in clinical Polymerase chain reaction et al samples (Loeb ., 2001). The concordance between Semiquantitative RT–PCR. The sequences of the primers our observations in primary breast tumors and the used to detect B-raf cDNA were as follows: 50-GAAGACCT functional differences seen in cell culture strengthens our CACAGTAAAAATAGGTGA-30 and 50-CCACAAAATG contention that WT1 ( þ Ex5/ þ KTS) contributes to GATCCAGACA-30. Other PCR sequences are as published. neoplastic transformation of mammary epithelial cells. Quantitative RT–PCR. Primers specific to RPS6K and to the housekeeping gene GAPDH were obtained from SuperArray Materials andmethods Bioscience (Fredrick, MD, USA). Amplification was performed in 40 cycles of 951C for 15 s and 601C for 1 minute. Complete details given in Supplementary materials.

Cell culture, plasmids and transfection Acknowledgements H16N-2 mammary epithelial cells were grown in mammary epithelial growth medium (MEGM) supplemented with 10% We are grateful to Venu Raman, Alan Friedman, and Robert fetal bovine serum. Cells were transfected with plasmids that Arceci for critical review of this article. This work was contain the cDNA for WT1 (ÀEx5/ÀKTS) or WT1 ( þ Ex5/ supported by grants from Flight Attendant Medical Research þ KTS) under the control of the CMV immediate-early Institute Young Clinical Scholar Award no. 012531 (DML), promoter or under the control of the sheep metallothionein Children’s Cancer Foundation (DML) and National Cancer promoter as described (Loeb, 2003, no. 1197). Institute T32CA60441 (LAS).

References

Barlund M, Forozan F, Kononen J, Bubendorf L, Chen Y, Davies RC, Calvio C, Bratt E, Larsson SH, Lamond AI, Bittner ML et al. (2000). Detecting activation of ribosomal Hastie ND. (1998). WT1 interacts with the splicing factor protein S6 kinase by complementary DNA and tissue U2AF65 in an isoform-dependent manner and can be microarray analysis. J Natl Cancer Inst 92: 1252–1259. incorporated into spliceosomes. Genes Dev 12: 3217–3225.

Oncogene Effects of WT1 isoforms on breast epithelial cells EA Burwell et al 3430 Englert C, Maheswaran S, Garvin AJ, Kreidberg J, Haber DA. suppressor gene potentiates granulocytic differentiation. (1997). Induction of p21 by the Wilms’ tumor suppressor Leukemia 17: 965–971. gene WT1. Cancer Res 57: 1429–1434. Maheswaran S, Park S, Bernard A, Morris JF, Rauscher FJd, Haber DA, Sohn RL, Buckler AJ, Pelletier J, Call KM, Hill DE et al. (1993). Physical and functional interaction Housman DE. (1991). Alternative splicing and genomic between WT1 and p53 proteins. Proc Natl Acad Sci USA 90: structure of the Wilms tumor gene WT1. Proc Natl Acad Sci 5100–5104. USA 88: 9618–9622. MiyoshiY,AndoA,EgawaC,TaguchiT,TamakiY,TamakiH Hewitt SM, Saunders GF. (1996). Differentially spliced exon 5 et al. (2002). High expression of Wilms’ tumor suppressor gene of the Wilms’ tumor gene WT1 modifies gene function. predicts poor prognosis in breast cancer patients. Clin Cancer Anticancer Res 16: 621–626. Res 8: 1167–1171. Inoue K, Tamaki H, Ogawa H, Oka Y, Soma T, Tatekawa T Natoli TA, McDonald A, Alberta JA, Taglienti ME, et al. (1998). Wilms’ tumor gene (WT1) competes with Housman DE, Kreidberg JA. (2002). A mammal-specific differentiation-inducing signal in hematopoietic progenitor exon of WT1 is not required for development or fertility. cells. Blood 91: 2969–2976. Mol Cell Biol 22: 4433–4438. Kolch W, Martiny-Baron G, Kieser A, Marme D. (1995). Reizner N, Maor S, Sarfstein R, Abramovitch S, Welshons WV, Regulation of the expression of the VEGF/VPS and its Curran EM et al. (2005). The WT1 Wilms’ tumor suppressor receptors: role in tumor angiogenesis. Breast Cancer Res gene product interacts with estrogen receptor-alpha and Treat 36: 139–155. regulates IGF-I receptor gene transcription in breast cancer Larsson SH, Charlieu JP, Miyagawa K, Engelkamp D, cells. J Mol Endocrinol 35: 135–144. Rassoulzadegan M, Ross A et al. (1995). Subnuclear Reynolds PA, Smolen GA, Palmer RE, Sgroi D, Yajnik V, localization of WT1 in splicing or transcription factor Gerald WL et al. (2003). Identification of a DNA-binding domains is regulated by alternative splicing. Cell 81: 391– site and transcriptional target for the EWS-WT1(+KTS) 401. oncoprotein. Genes Dev 17: 2094–2107. Lemoine NR, Barnes DM, Hollywood DP, Hughes CM, Silberstein GB, Van Horn K, Strickland P, Roberts Jr CT, Smith P, Dublin E et al. (1992). Expression of the ERBB3 Daniel CW. (1997). Altered expression of the WT1 Wilms gene product in breast cancer. Br J Cancer 66: 1116–1121. tumor suppressor gene in human breast cancer. Proc Natl Lo HW, Hsu SC, Hung MC. (2005). EGFR signaling pathway Acad Sci USA 94: 8132–8137. in breast cancers: from traditional signal transduction to Udabage L, Brownlee GR, Nilsson SK, Brown TJ. (2005). The direct nuclear translocalization. Breast Cancer Res Treat 95: over-expression of HAS2, Hyal-2 and CD44 is implicated in 1–8. the invasiveness of breast cancer. Exp Cell Res 310: 205–217. Loeb DM. (2006). WT1 influences apoptosis through tran- Velona T, Abbiati L, Beretta B, Gaiaschi A, Flauto U, scriptional regulation of Bcl-2 family members. Cell Cycle 5: Tagliabue P et al. (1999). Protein profiles in breast milk from 1249–1253. mothers delivering term and preterm babies. Pediatr Res 45: Loeb DM, Evron E, Patel CB, Sharma PM, Niranjan B, 658–663. Buluwela L et al. (2001). Wilms’ tumor suppressor gene Zapata-Benavides P, Tuna M, Lopez-Berestein G, Tari AM. (WT1) is expressed in primary breast tumors despite tumor- (2002). Downregulation of Wilms’ tumor 1 protein inhibits specific promoter methylation. Cancer Res 61: 921–925. breast cancer proliferation. Biochem Biophys Res Commun Loeb DM, Korz D, Katsnelson M, Burwell EA, Friedman AD, 295: 784–790. Sukumar S. (2002). Cyclin E is a target of WT1 trans- Zhang TF, Yu SQ, Guan LS, Wang ZY. (2003). Inhibition of criptional repression. J Biol Chem 277: 19627–19632. breast cancer cell growth by the Wilms’ tumor suppressor Loeb DM, Summers JL, Burwell EA, Korz D, Friedman AD, WT1 is associated with a destabilization of beta-catenin. Sukumar S. (2003). An isoform of the Wilms’ tumor Anticancer Res 23: 3575–3584.

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

Oncogene