β-Catenin signaling is a critical event in ErbB2-mediated mammary tumor progression.

Although ERBB2 amplification and overexpression is correlated with poor outcome in breast cancer, the molecular mechanisms underlying the aggressive nature of these tumors has not been fully elucidated. To investigate this further, we have used a transgenic mouse model of ErbB2-driven tumor progression (ErbB2(KI) model) that recapitulates clinically relevant events, including selective amplification of the core erbB2 amplicon. By comparing the transcriptional profiles of ErbB2(KI) mammary tumors and human ERBB2-positive breast cancers, we show that ErbB2(KI) tumors possess molecular features of the basal subtype of ERBB2-positive human breast cancer, including activation of canonical β-catenin signaling. Inhibition of β-catenin-dependent signaling in ErbB2(KI)-derived tumor cells using RNA interference impaired tumor initiation and metastasis. Furthermore, treatment of ErbB2(KI) or human ERBB2-overexpressing tumor cells with a selective β-catenin/CBP inhibitor significantly decreased proliferation and ErbB2 expression. Collectively, our data indicate that ERBB2-mediated breast cancer progression requires β-catenin signaling and can be therapeutically targeted by selective β-catenin/CBP inhibitors.


Introduction
The progression of normal mammary epithelial cells to a malignant phenotype involves multiple genetic events including amplification and overexpression of proto-oncogenes, such as ERBB2 (Neu, HER2) [1]. ERBB2 amplification and subsequent overexpression strongly correlates with a negative clinical prognosis in both lymph node-positive and -negative breast cancer [2]. Direct evidence supporting a role for ERBB2 in human breast cancer derives from observations made with transgenic mice. Overexpression of activated ErbB2 in the mammary epithelium results in the rapid induction of metastatic multifocal mammary tumors (reviewed in [3]). One limitation of these studies is that ErbB2 expression is driven by the hormonally regulated mouse mammary tumor virus (MMTV) promoter, resulting in non-physiological expression of the transgene. In an attempt to more closely mimic the events involved in ErbB2-induced mammary tumor progression, we have generated transgenic mice that carry a Cre-inducible activated erbB2 allele under the transcriptional control of the endogenous erbB2 promoter (herein referred to as the ErbB2 KI model) [4]. In contrast to the rapid tumor progression observed in the MMTV-regulated ErbB2 strains, focal mammary tumors arose only after an extended latency period in the ErbB2 KI model, and this was further associated with a dramatic elevation of both ErbB2 protein and transcript. Remarkably, the elevated expression of ErbB2 was correlated with the selective genomic amplification of the activated erbB2 allele. Thus, as in human breast cancers, amplification of erbB2 appears to be a critical event in mammary tumor progression in this unique transgenic mouse model (reviewed in Wnt/β-catenin pathway triggers a series of events that inactivates a complex consisting of adenomatous polyposis coli (APC), axin, and glycogen synthase kinase 3 β (GSK3β). This complex is responsible for β-catenin degradation, and its inactivation promotes accumulation of β-catenin in the cytoplasm and translocation to the nucleus. Nuclear β-catenin forms a complex with T cell-specific transcription factor/lymphoid-enhancer binding factor (TCF/LEF) and recruits the transcriptional co-activators cAMP response element binding (CREB)-binding protein (CBP) and p300, along with other components of the basal transcription machinery, to regulate the expression of target genes such as Axin2, c-Myc, and Tcf7 [8][9][10]. Accumulation of cytosolic and nuclear β-catenin, due to stabilizing mutations in its N-terminal domain, is commonly observed in human cancers such as colorectal carcinoma. Although such mutations are uncommon in human breast cancer, aberrant expression and/or accumulation of β-catenin in the cytoplasm and nucleus have been observed and are associated with poor patient prognosis [11].
In this study, we have used the ErbB2 KI model to gain insight into the molecular and genetic events involved in ErbB2-induced mammary tumor progression. Immunohistological and transcriptional profiling revealed that in contrast to mammary tumors derived from a constitutive ErbB2 mouse model (MMTV/NIC), ErbB2 KI -derived tumors expressed markers characteristic of the basal and ERBB2 subtypes of human breast cancer and exhibited activation of the Wnt/β-catenin signaling pathway. Consistent with the ErbB2 KI mouse model, a cohort of human ERBB2-positive invasive ductal carcinomas also showed high cytoplasmic β-catenin and abundant expression of basal markers. To further validate the biological importance of β-catenin signaling in ErbB2-induced tumorigenesis, we used RNA interference to downregulate β-catenin in ErbB2 KI -derived mammary tumor cell lines.

Reduction of β-catenin levels significantly impaired mammary tumor initiation and
Research.
on September 9, 2017. © 2013 American Association for Cancer cancerres.aacrjournals.org Downloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
Author Manuscript Published OnlineFirst on ; DOI: 10.1158/0008-5472.  metastasis in these tumor cells, and was further associated with reduced expression of components of the erbB2 amplicon as well as ErbB3. Moreover, treatment of ErbB2 KI tumor cells with ICG-001, an inhibitor that specifically antagonizes β-catenin/CBP-mediated transcription which is critical for maintenance of a non-differentiated/proliferative state [12,13], resulted in a proliferative defect, which was accompanied with reduced expression of ErbB2. Similarly, treatment of human ERBB2-positive breast cancer cells with ICG-001 significantly decreased cell proliferation, which correlated with reduced levels of ErbB2.
These observations demonstrate that targeting of β-catenin-dependent signaling has potential therapeutic value in the treatment of the basal category of ERBB2-positive breast cancer.

Materials and Methods
Gene expression microarray data. Total RNA was isolated from five individual ErbB KI tumors. Given the homogeneity of the MMTV/NIC model, two RNA pools containing equal amounts of total RNA from five individual tumors were generated to obtain MMTV/NIC profiles that are representative of the general population [14]. Total RNA isolation and RNA sample preparations were performed as previously described [14]. Histology and immunostaining of tissue sections. Mammary tumors and lungs were harvested from mice at tumor endpoint. Tissue was fixed and embedded as described previously [16]. Paraffin sections of 5 μm were stained with hematoxylin and eosin (H&E).
Step sections of lungs (5 x 50 μm) were scanned using a Scanscope XT Digital Slide Scanner (Aperio) and analyzed at 5X magnification using Imagescope software (Aperio) to quantify the total number and surface area of lesions per lung. Immunostaining was performed as described [17] (SI Materials and Methods).
Immunoblotting. Tumor lysates were prepared from flash frozen tissue and immunoblot analyses were performed on 15μg of lysate as described [14]. For cell lines, extract was prepared in RIPA lysis buffer and 15μg of lysate was used for immunoblot analyses.
Antibodies for immunoblots include: c-Myc, Neu (C18), ErbB3 (C17), and Grb7 from Santa Cruz; β-catenin, survivin, Egfr, and tubulin from Cell Signaling, and β-actin from Sigma Aldrich. Horseradish peroxidase-conjugated secondary antibodies were obtained from Jackson Laboratories. Chromatin immunoprecipitation (ChIP) assay. ChIP was performed as described previously [18]. Additional details are available in Supplementary Methods. Hospital, Sydney, H00 036). A more detailed description of the clinicopathological characteristics of the cohort is published elsewhere [19,20] (SI Materials and Methods). βcatenin status has previously been determined for this cohort as described [19] (SI Materials and Methods).
Definition of intrinsic "molecular" phenotype of breast cancer. This was assessed immunohistochemically using criteria similar to those previously described [21] and FISH to determine ERBB2 status [20]. Four different subgroups were defined: basal, ERBB2 + /ER -/PR -, luminal A and B. Statistical evaluation was performed using Statview 5.0 Software (Abacus Systems, Berkeley, CA). A p-value of less than 0.05 was accepted as statistically significant. Analysis of variance was used to determine differences in expression of continuous variables across breast cancer subtypes as previously described [22]. Kaplan-Meier analysis and Cox-proportional hazard ratios were used to determine association of membrane to cytoplasmic (MTC) score < 0 with breast cancer specific death [19] (SI Materials and Methods). tumors with a phenotype that is comparable with MMTV/NDL2-5 animals, in which only activated ErbB2 is expressed [24]. Using cytokeratin markers that are reflective of the different epithelial cell lineages within the mammary gland, we observed uniform expression of the luminal marker cytokeratin 8 in the ErbB2 KI tumor (Fig. 1, upper panel). However, many of these tumors retained expression of basal cytokeratin markers such as Krt5, Krt6, and Krt14 (Fig. 1, Fig. S1A). A similar heterogeneous expression was observed for Trp63 and Egfr, which are also associated with a basal phenotype (Fig. 1, Fig. S1A). By contrast, MMTV/NIC tumors were exclusively Krt8-positive but negative for these basal markers ( immunohistopathological survey indicates that ErbB2 KI mammary tumor progression is associated with a high degree of intratumoral heterogeneity.

ErbB2 KI -induced mammary tumors possess molecular features of both the ERBB2 and basal-like subtypes of human breast cancer.
Previously, we demonstrated that ErbB2 KI mammary tumor progression is associated with distinct phenotypical and genomic properties compared to MMTV/ErbB2-derived tumors [25]. To further elucidate the molecular basis associated with these unique features of the ErbB2 KI model, we obtained transcriptional profiles from ErbB2 KI -induced tumors and compared them to profiles from MMTV/NIC tumors and normal mammary glands (FVB m. gl.). Using a class discovery approach with those genes that varied most across the dataset, the samples primarily cluster into two defined groups which completely separate the normal controls (FVB m. gl.) from the tumors (Fig. S3A). The analysis further segregates the two ErbB2 mouse models into two distinct subclusters (Fig. S3A). Notably, the ErbB2 KI tumorderived profiles displayed heterogeneous gene expression patterns. Among the genes showing increased transcriptional abundance uniquely in ErbB2 KI tumors were components and targets of the Wnt/β-catenin pathway and genes that are linked to the TGFβ signaling pathway ( Fig. 2A). Moreover, basal cytokeratins displayed altered transcript levels, supporting our initial immunohistochemical analyses ( Fig. 1).
To understand the molecular differences between tumors derived from the two ErbB2 models, genes systematically upregulated in each of the models were analyzed for enrichment in GO and KEGG pathways (Supplementary Table 1 Recently, a murine-specific tumor dataset and intrinsic gene list have been established from transgenic mouse models that resemble the intrinsic subtypes of human breast cancer based on their gene expression [26]. Using this murine-specific intrinsic gene list and tumor dataset, unsupervised hierarchical clustering assigned the ErbB2 KI and MMTV/NIC models to distinct molecular groups (Fig. 2B, Fig. S3B Next, we explored how the transcriptional profiles generated from two different ErbB2 breast cancer mouse models correspond to the human disease. ErbB2 tumor-derived gene expression profiles were compared to profiles from human breast cancers using a crossspecies intrinsic gene set as a basis [26]. Consistent with our results obtained from the intramouse model comparison, the MMTV/NIC tumors clustered with the human luminal subtypes while the ErbB2 KI tumors clustered within the human ERBB2 subtype (Fig. 3A). To preclude the influence of the erbB2 amplicon on human subtypes in this cross-species comparison, we removed Grb7 (the only erbB2 amplicon gene found in the original crossspecies gene set) from the intrinsic gene list. While the MMTV/NIC tumors still clustered with the human luminal subtypes, the ErbB2 KI tumors now grouped with the basal subtype (Fig. 3B). These findings substantiate the close resemblance of ErbB2 KI tumors to the human ERBB2 subtype, but indicate that ErbB2 KI mammary tumors contain molecular features of both the ERBB2 and basal subtypes of human breast cancer.
ErbB2 KI mammary tumor progression is associated with the activation of the Wnt/βcatenin signaling cascade.
As we had identified a Wnt/ȕ-catenin signaling axis in ErbB2 KI mammary tumors at the transcriptional level ( Fig. 2A), we next investigated whether ErbB2 KI tumors exhibited evidence of activation of this pathway. To test this possibility, tumor sections from either MMTV/NIC or ErbB2 KI mice were subjected to immunofluorescence analyses with antibodies against total β-catenin and activated β-catenin [27]. In contrast to the strict membrane localization of total β-catenin and the weak, diffused cytoplasmic signal of the activated form in the MMTV/NIC tumors, nuclear staining for both forms of β-catenin was detected in 75% (15/20) of ErbB2 KI samples (Fig. 4A). Consistent with these observations, ErbB2 KI mammary tumors with nuclear β-catenin also displayed partial loss of membranous E-cadherin, a protein that forms a complex with β-catenin in adherens junctions, whereas the MMTV/NIC samples showed robust membrane-associated expression of E-cadherin (Fig.   4A). These data suggest that the majority of ErbB2 KI tumors show activated Wnt/β-catenin signaling.
In support of this notion, significant upregulation of direct β-catenin transcriptional targets, including CyclinD1, Sox9, c-Myc, and two hallmarks of an active Wnt/β-catenin pathway, Axin2, and Tcf7, were detected in most of the ErbB2 KI mammary tumors (Fig. 4B-D) [8,9,[28][29][30]. In contrast, MMTV/NIC tumors showed only basal level expression of these β-catenin targets (Fig.4B-D). Consistent with the inherent intratumoral heterogeneity observed in ErbB2 KI mammary tumors, there was a high degree in variability in the expression of these β-catenin targets within the ErbB2 KI cohort.
Cytoplasmic and nuclear β-catenin accumulation has been commonly observed in poor outcome basal-like breast cancers that express ErbB2 [19,31]. Additionally, recent studies have shown that ERBB2-overexpressing tumors can be subdivided in different groups based on their estrogen receptor (ER) status and the expression of basal cytokeratins [23,32].
To establish whether subcellular β-catenin localization impacts on either ER status or the expression of basal cytokeratins, we performed further statistical analyses of our previously described patient cohort [19]. The results demonstrated that tumors with significantly higher cytoplasmic β-catenin (reflected by the "membrane to cytoplasmic" MTC score <0 [19]) segregated predominantly to the basal-like and ERBB2 + /ER -/PRsubtypes (Fig. 5A, B).
Additionally, ERBB2 + /ER -/PRtumors that displayed a cytoplasmic β-catenin expression pattern (MTC score <0, n=16) were associated with a higher rate of breast cancer-specific death (Hazard ratio = 2.5, p=0.02, Fig. 5C). Thus, our observations indicate that a distinct cohort of ERBB2-positive invasive ductal carcinomas display high cytoplasmic β-catenin, Although the above studies indicate that active β-catenin signaling can be detected in the ErbB2 KI tumors and a distinct group of human ERBB2-positive breast cancers, the biological significance of β-catenin signaling in ERBB2 mammary tumor progression is unclear. To directly test the importance of β-catenin in this process, RNA interference was used to downregulate endogenous β-catenin in ErbB2 KI -derived clonal mammary tumor cells.
One striking feature of these cell lines is the difference in their ability to metastasize to lung, with two clones being highly metastatic (TM15c7-2 and c10-2) [33]. To this end, stable c7-2 and c10-2 cell lines expressing either nonspecific shRNA or shRNAs targeting β-catenin (shCtnnb1) were generated. Although the in vitro growth rates of the pooled clones expressing β-catenin shRNA were comparable to the growth rates of the controls (Fig. S4A), the β-catenin-knockdown cells exhibited a pronounced defect in invasion in contrast to the control cells (Fig. S4B). Immunoblot analyses confirmed reduced β-catenin protein levels in both cell lines expressing β-catenin-specific shRNAs (Fig. S4C-D). Interestingly, in addition to the reduced levels of β-catenin, immunoblot analyses also detected a decrease in ErbB2 ( Fig. S4C-D). Similar data were obtained for a different set of β-catenin-specific shRNAs ( Fig. S4D), indicating that loss of β-catenin results in a reduction in ErbB2 protein levels.
To assess whether β-catenin-deficiency could alter the in vivo tumorigenic potential of these cells, the tumor cells were injected into the mammary fat pad of athymic mice and monitored for mammary tumor induction. While control mice exhibited rapid induction of mammary tumors, TM15c10-2 tumor cells expressing the β-catenin shRNA showed a substantial delay in tumor onset (Fig. 6A). In addition to the effect on tumor onset, the 2 shCtnnb1 cells formed lung lesions, with one lesion per lung on average (Fig. 6B). Similarly, the average total surface area of these lesions was dramatically reduced in the mice injected with the β-catenin-deficient tumor cells when compared to surface area of lesions in the control groups (Fig. S5A).
Since β-catenin-deficient ErbB2 KI cells exhibited low levels of ErbB2 protein (Fig.   S4C-D), we next examined the expression of ErbB2 in mammary fat pad outgrowths of ErbB2 KI cell lines lacking β-catenin and control groups. Consistent with our in vitro studies, β-catenin-deficient tumors showed a significant decrease in ErbB2 protein and transcript levels, while relative erbB2 amplification was unchanged across all samples (Fig. 6C, Fig.   S5B). Moreover, given the strong correlation between the expression of ErbB2, ErbB3 and components of the erbB2 amplicon, including Grb7, Stard3 and Perld1, we assessed their expression in these samples [34,35]. Like ErbB2 levels, the expression of these proteins was significantly reduced in β-catenin-deficient mammary tumors, (Fig. 6C).
To verify that β-catenin signaling was impaired in these tumors, the relative expression of several β-catenin target genes was measured by qRT-PCR analysis. Indeed, significant decrease in β-catenin, Axin2, and Tcf7 transcripts were detected in most of the βcatenin-deficient tumors, as well as reduced expression of CyclinD1 (Fig. S5C, D). These observations indicate that β-catenin is critical for both the initiation and metastatic phases of ErbB2 KI mammary tumor progression.

ERBB2-positive human mammary tumour cells.
To further address the role of β-catenin signaling in mammary tumorigenesis, we next prevents β-catenin/CBP-mediated transcription, which is critical for maintenance of a nondifferentiated/proliferative state, without affecting β-catenin/p300-dependent signaling, which is important for initiation of cellular differentiation [12,13]. Treatment of two independent ErbB2 KI -derived tumor cell lines (TM15c7-2, c10-2) with ICG-001 resulted in a clear proliferative defect in both cell lines without impacting the apoptotic status of the cells ( Fig. 6D left panel). In contrast, proliferation in two independent MMTV/NIC-derived tumor cell lines was not affected by ICG-001 treatment (Fig. S5E), supporting our data that βcatenin signaling is not activated in this MMTV-driven ErbB2 mouse model.
Like β-catenin-deficient ErbB2 KI tumor cells, specific inhibition of the β-catenin/CBP signaling resulted in reduced expression of ErbB2, Egfr, Grb7 and ErbB3 (Fig. 6D, right   panel). Collectively, these observations argue that inhibition of β-catenin/CBP function severely affected ErbB2 KI tumor cell proliferation accompanied by a reduction in the levels of Egfr family members (ErbB2, ErbB3, Egfr) and Grb7.
To confirm the importance of β-catenin /CBP signaling in human ERBB2-dependent breast tumor progression, we evaluated whether ICG-001 impacts on the proliferative status of a number of ERBB2-expressing human breast cancer cell lines. Given the association of βcatenin signaling with ERBB2-overexpressing breast cancer cells that are ER/PR-negative, we evaluated the β-catenin status in three such lines (HCC202, HCC1954, SKBR3) [37].
Immunofluorescence analysis using β-catenin and activated β-catenin antibodies revealed the presence of activated β-catenin in the nucleus in all three ERBB2 + /ER -/PRcell lines ( catenin/CBP-mediated transcription by incubating these human breast cancer cells with ICG-001. The results showed that all cell lines with evidence of activated β-catenin exhibited either a strong (SKBR3, MDAMB2310) or a moderate (HCC202 and HCC1954) growth inhibition (Fig. 7B, Fig. S6B). By contrast, ICG-001 had a modest effect on cell proliferation in the luminal control cancer cell line (MCF7) (Fig. S6B). The observed ICG-001-mediated proliferative defect was not associated with an increase in apoptosis as cell surface Annexin V staining was unaffected (Fig. S6C).
Consistent with impaired β-catenin/CBP signaling, biochemical analyses of ICG-001treated tumor cells in comparison with DMSO-and untreated cells exhibited significantly reduced levels of survivin in the basal and all three ERBB2-positive cell lines (Fig. 7C, Fig.   S6D). Moreover, in support with our data obtained from ErbB2 KI tumor cells, ERBB2overexpressing human cell lines displayed substantially lower levels of ERBB2, GRB7, ERBB3, and EGFR when treated with ICG-001 (Fig. 7C). Together, these data suggest that targeting β-catenin/CBP signaling results in the repression of ERBB2 and an associated decrease in cell proliferation amongst ERBB2-overexpressing breast cancer cell lines.
To this end our data indicate that antagonizing β-catenin signaling leads to repression of ErbB2/ERBB2 expression in both mouse and human mammary tumor cells. To explore whether β-catenin directly activated ERBB2 transcription, we performed ChIP analyses on the ERBB2 promoter with β-cateninand RNA polymerase II-specific antisera. Although we could not detect recruitment of β-catenin to the ERBB2 proximal promoter, we detected a 3fold enrichment in β-catenin and a 20-fold enrichment in RNA polymerase II recruitment to an intronic ERBB2 site (Fig 7D), which has been shown to be bound by several factors that are important in the transcriptional regulation of ERBB2 [18,38]. β-catenin and RNA polymerase II occupancy to this site was further enriched upon treatment with the Wnt3 ligand ( Fig 7D). We further exposed the cells to ICG-001 for 48 hours and there was no impact on recruitment of ȕ-catenin to the intronic ERBB2 site (data not shown) whereas RNA polymerase II occupancy was substantially decreased (Fig 7D). Importantly, activation by Wnt/β-catenin signaling in SKBR3 cells led to a 2-fold increase in the recruitment of RNA polymerase II to the promoter of ERBB2, while inhibition of ȕ-catenin/CBP by ICG-001 decreased RNA polymerase II recruitment by 2-fold ( Fig 7D). Interestingly, we observed a similar impact on RNA polymerase II occupancy on the proximal promoter of other components of the ERBB2 amplicon after Wnt3 and ICG-001 treatment (Grb7, Stard3, and PerlD1; Fig S6E). Together, these data indicate that β-catenin/CBP-mediated signaling directly regulates ERBB2 expression in ERBB2-positive breast cancer cells. Research.
on September 9, 2017. © 2013 American Association for Cancer cancerres.aacrjournals.org Downloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
Author Manuscript Published OnlineFirst on May 29, 2013; DOI: 10.1158/0008-5472.  Discussion A hallmark of human breast cancer is its intrinsic heterogeneity, reflecting the underlying molecular complexity of the disease. Based on comprehensive gene expression profiling, five major molecular subtypes have been identified (luminal A, luminal B, basallike, ERBB2 + , and normal-like) that are associated with differences in patient outcome and response to treatment [39]. A growing body of evidence established by transcriptional profiling, comparative genome hybridization, and immunohistochemical analyses, suggests that the biology of ERBB2-positive tumors is similarly heterogeneous. Indeed, this molecular subtype can be subdivided into ER-positive and ER-negative tumors, which may explain the variable response to ERBB2-targeted therapy [23,32,[40][41][42][43]. To address this issue, we employed the ErbB2 KI mouse model, which exhibits several unique features, including expression of ErbB2 from its endogenous promoter and spontaneous amplification of the erbB2 locus, resembling that of human ERBB2-positive breast cancers [4]. In this study, we showed that ErbB2 KI mammary tumor progression is associated with heterogeneous expression of distinct basal (Krt5, Krt6, Trp63, and Egfr), myoepithelial (Krt14) and luminal markers (Krt8), and the presence of cells exhibiting bipotential characteristics, whereas constitutive MMTV/NIC tumors are comprised of a uniform luminal epithelial cell type ( Fig.   1, S1, S2).
Consistent with these histopathological features, the gene expression signatures from ErbB2 KI mammary tumors differ from those of MMTV/NIC tumors ( Fig. 2A). While genes related to protein modification, cell growth and death, and FGF receptor signaling were associated with MMTV/NIC tumors, the transcriptional signature obtained from ErbB2 KI tumors comprised genes implicated in mesenchymal cell development, Wnt, TGFȕ, and Egfr signaling pathways ( Fig. 2A, Supplementary Table 1). Moreover, a comparison of these ErbB2-driven expression profiles with data derived from other murine models revealed that ErbB2 KI samples shared molecular features with murine models representative of the mesenchymal/basal phenotype (Fig. 2B). Using an intrinsic mouse/human signature [26], we further demonstrated that ErbB2 KI mammary tumors closely resemble both the basal-like and ERBB2-positive subtypes of human breast cancer (Fig. 3). By contrast, MMTV/NIC tumors exhibited transcriptional features typical of the solid/luminal subtype (Fig. 3). Since high ErbB2 expression is achieved much earlier in the MMTV/NIC strain in comparison to ErbB2 KI mice, it is likely that the tumor-initiating cell type may differ between the two models. In support of this argument, ErbB2 KI mammary tumors contain cells expressing Trp63 and co-expressing Krt8/14 and Krt8/5 ( Fig. 1, S1, S2), which are indicative of a multipotent phenotype [44,45].
Based on our gene expression data we found that ErbB2 KI mammary tumors overexpressed components of the Wnt/β-catenin signalling pathway, contained cellular regions with nuclear β-catenin, and abundantly expressed classical Wnt/β-catenin targets, including CyclinD1, c-Myc, Axin2, and Tcf7 (Fig. 2, 4). These findings indicate that ErbB2 KI mammary tumor progression is associated with Wnt/β-catenin activation. Consistent with this concept, recent genomic and immunohistochemical profiling of human ERBB2-amplified tumors identified an ERBB2 + /ERsubgroup that was associated with the accumulation of cytoplasmic β-catenin, overexpression of Wnt/β-catenin signaling components, and increased risk of recurrence [41]. We identified a subgroup within our ERBB2-positive patient cohort characterized as ERBB2 + /ER -/PR -, which predominantly displayed cytoplasmic β-catenin and was associated with poor clinical outcome (Fig. 5). Collectively, these observations suggest that mammary tumor progression in the ErbB2 KI mouse model recapitulates many of the molecular events observed in the human basal-ERBB2 subtype, including activation of a βcatenin signaling network. The importance of β-catenin signaling in ErbB2 mammary tumorigenesis is further highlighted by the observation that downregulation of endogenous β-catenin expression in ErbB2 KI -derived mammary tumor cells (TM15c7-2 and c10-2) severely impaired the invasive capacity of these cells in vitro (Fig. S4) and impacted both the initiation and metastatic phases of tumor progression in vivo (Fig. 6). Our observations are consistent with numerous reports where genetic ablation of β-catenin in various in vivo systems diminished metastatic progression [46,47], providing support for β-catenin as a critical mediator of the metastatic process.
The importance of β-catenin in ErbB2 tumor induction is further supported by studies with the ICG-001 small molecule inhibitor that specifically antagonizes the interaction of ȕcatenin with CBP but not with p300, facilitating downregulation of a subset of ȕcatenin/CBP-responsive genes. Treatment with ICG-001 caused a significant proliferative defect in mouse ErbB2 KI mammary tumor cells and in several human ERBB2-overexpressing cancer cells in vitro. Both cell systems showed reduced levels of survivin, which is indicative of antagonized ȕ-catenin/CBP signaling (Fig. 6D, Fig. 7C). This growth defect was not related to increased apoptosis but rather associated with a defect in cell proliferation. A body of evidence indicates that survivin function is not only limited to inhibition of apoptosis but also involves the regulation of cell division; its overexpression allows cancer cells to resume cell division (reviewed in [48]). importance of β-catenin in cellular viability, stable cell lines expressing β-catenin-specific shRNAs still retained basal level expression of β-catenin (Fig. 6C, S4C-D). Given the ability of β-catenin to recruit histone remodelling complexes such as SWI/SNF and Polycomb repressor complexes 1 and 2 (PCR1/2) [49][50][51], it is conceivable that downregulation of βcatenin in ErbB2 KI tumor cells results in genome-wide reprogramming due to an altered epigenetic landscape which compensates for loss of β-catenin. As a consequence these cells maintained their proliferative properties but lost their invasive potential. Further experiments are needed to enhance our understanding of β-catenin signal integration and crosstalk in ErbB2-overexpressing tumor cells.
The complexity by which ȕ-catenin regulates Wnt-dependent and -independent transcription has only emerged recently (reviewed by [52]). Aberrant activation of the Wnt pathway leads to nuclear localization of ȕ-catenin where it functions as a scaffold to link TCF factors with numerous complexes to regulate target genes. In this study, we have found that loss of endogenous ȕ-catenin in mouse ErbB2 KI tumor cells correlated with low expression of ErbB2, Grb7, Stard3, PerlD1, and ErbB3 (Fig. 6C). Similarly, inhibition of ȕ-catenin/CBP signaling by ICG-001 in this system significantly suppresses ERBB2, Grb7 and ERBB3 expression in a panel of human ERBB2-overexpressing tumor cells (Fig. 6D, 7C). Moreover, our ChIP data also illustrate a possible regulation of ERBB2 gene expression by ȕcatenin/CBP signaling (Fig. 7D). We found recruitment of ȕ-catenin and RNA polymerase II at an intronic ERBB2 site, which was further enhanced upon Wnt3 exposure while ICG-001 treatment disrupted RNA polymerase II occupancy at this site. This site is particularly important since transcription factors such as ERRα and ERα have been shown to compete for binding to this site to activate or repress ERBB2 transcription [18,38]. In addition, recent data has established that ERRĮ, ȕ-catenin, and Lef1 form a transcriptionally active complex to activate gene expression [53]. Taken together, the strong reduction of ERBB2 expression observed upon ablation of β-catenin activity either through RNA interference or IGC-001 is due to a disruption of the ERRα/ȕ-catenin/CBP-dependent recruitment of RNA polymerase II to an intronic ERBB2 site that regulates ERBB2 transcription. Consistent with this data, it has been previously demonstrated that ICG-001 specifically interferes with the ability of βcatenin to form complexes with CBP and Lef1 [54]. The ICG-001-dependent reduction in GRB7 transcription can also be attributed to the fact that the proximal promoters of GRB7 and other ERBB2 amplicon components are also regulated by these β-catenin-containing complexes (Suppl. Figure S6 D and E).
Collectively, this work highlights the relevance of the ErbB2 KI mouse strain as a pre-      taken at 400X. (B) Basal-like and ERBB2 + /ER -/PRsubtypes expressed the highest levels of cytoplasmic β-catenin, CK5/6 and CK14 when compared to luminal A and B subtypes. (C) Kaplan-Meier survival analysis of the breast cancer patient cohort showing separation of patients of the ERBB2 + /ER -/PRsubtype that have cytoplasmic β-catenin expression as indicated (n = 16, MTC score<0). These patients had a worse outcome in comparison to the rest of the cohort of invasive ductal carcinoma. Figure 6. Inhibition of β-catenin signaling in ErbB2 KI -derived tumor cells impairs tumor initiation and metastasis and is associated with reduced expression of erbB2 amplicon components as well as ErbB3. (A) ErbB2 KI tumor cells (TM15c10-2) and stable cells expressing non-silencing shRNA control (nonshRNA) or β-catenin-specific shRNA (shCtnnb1) were injected into the mammary fat pad of athymic mice and tissue was harvested at endpoint (2000 mm 3 ). Tumor volumes are presented for five independent mice for each group. (B) H&E-stained step sections of lungs were scored for the total number of lesions per lung as indicated. Ratios in parentheses indicate the number of mice with lung lesions relative to the total number of mice examined. Error bars represent standard error of mean (two-tailed t-test). (C) Left panel, tumor lysates (15μg) from each group were immunoblotted for the indicated proteins. Right panel, mean relative mRNA expression levels of Grb7, Stard3, PerlD1, ErbB2, and ErbB3 transcript measured by qRT-PCR in TM15c10-2, nonshRNA and shCtnnb1 tumors as indicated. mRNA levels were normalized to GAPDH. Error bars represent standard deviation of triplicates obtained in three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001; two-tailed t-test versus controls. (D) Left panel, graph showing the proliferation of ErbB2 KI cells (TM15c7-2, c10-2) treated with ICG-001 as indicated (MTS proliferation assay). Data are normalized to values from parental cells. Error bars represent standard error of mean of triplicates obtained in independent experiments; *P < 0.05, **P < 0.01, ***P < 0.001; two-tailed t-test versus controls. Right panel, ErbB2 KI cell lysates (15ȝg) were treated with ICG-001 for 48 hours and subjected to immunoblot analysis for the indicated proteins.