The role of Tcfap2c in tumorigenesis and cancer growth in an activated Neu model of mammary carcinogenesis

Abstract

TFAP2C/AP-2γ influences development of the mammary gland and regulates patterns of gene expression in luminal and HER2-amplified breast cancer. The roles of TFAP2C in mammary gland tumorigenesis and in pathways critical to cancer progression remain poorly understood. To gain greater insight into oncogenic mechanisms regulated by TFAP2C, we examined mammary tumorigenesis in MMTV-Neu transgenic female mice with or without conditional knockout (KO) of Tcfap2c, the mouse homolog of TFAP2C. Loss of Tcfap2c increased the latency of tumorigenesis and tumors that formed demonstrated reduced proliferative index and increased apoptosis. In addition, tumors formed in Tcfap2c KO animals had a significant reduction in Egfr levels without a change in the expression of the Neu oncogene. The MMneu-flAP2C cell line was established from tumor tissue derived from MMTV-Neu/Tcfap2c^L/L control animals and parallel cell lines with and without expression of Tcfap2c were created by transduction with adenovirus-empty and adenovirus-Cre, respectively. KO of Tcfap2c in vitro reduced activated phosphorylated-Erk, decreased cell viability, repressed tumor growth and was associated with attenuation of Egfr expression. Chromatin immunoprecipitation and direct sequencing and expression analysis confirmed that Egfr was a Tcfap2c target gene in murine, as well as human, mammary carcinoma cells. Furthermore, decreased viability of mammary cancer cells was directly related to Egfr functional blockade. We conclude that TFAP2C regulates tumorigenesis, cell growth and survival in HER2-amplified breast cancer through transcriptional regulation of EGFR. The findings have important implications for targeting the EGFR pathway in breast cancer.

INTRODUCTION

The molecular subtypes of breast cancer are defined by characteristic patterns of gene expression, which are predictive of outcome and response to therapy.^1,2 The luminal breast cancer subtype accounts for ~ 70% of all breast cancers and is identified by the expression of a set of estrogen receptor-alpha (ERα)-associated genes.^3,4 ERα-negative cancer with amplified expression of the epithelial growth factor receptor HER2/ERBB2 characterizes the HER2 breast cancer subtype. Growth and cell survival of the HER2 breast cancer subtype is dependent upon HER2/ERBB2 signaling and blocking this signaling pathway with drugs such as trastuzumab has been developed as directed therapy for this subclass of breast cancers.^5,6 Cancers that lack ERα, progesterone receptor and HER2-amplified expression are characterized as triple-negative breast cancer, most of which are classified as basal-like breast cancers.^7,8 The molecular characterization of breast cancer based on patterns of gene expression has expanded our understanding of breast cancer biology and guided treatment decisions related to the use of chemotherapy. Investigating the transcriptional mechanisms that contribute to defined patterns of gene expression of the different breast cancer subtypes has begun to provide valuable insight into key mechanisms driving breast cancer oncogenesis and progression.

The activating enhancer binding protein 2 alpha (TFAP2A/ AP-2α) and gamma (TFAP2C/AP-2γ) transcription factors are highly homologous and play a critical role in mammary gland development. Studies of the mouse AP-2 homologs, Tcfap2a and Tcfap2c, have helped to clarify the role of the AP-2 proteins in mammary gland development. Both AP-2 family members are expressed in human and mouse mammary epithelial cells (MMECs) early in development and during mammary gland maturation.^9,10 Overexpression of TFAP2A driven by the MMTV promoter stunted mammary gland development with a reduction in alveolar budding, although mature transgenic animals demonstrated normal lactation.¹⁰ However, overexpression of a Tcfap2c transgene using the MMTV promoter resulted in mammary gland epithelial hypoplasia and lactation failure.¹¹ Whole animal knockout (KO) of Tcfap2c is embryonic lethal due to its critical role in the development of extra-embryonic membranes.¹² Conditional KO of Tcfap2c has been accomplished using SOX2-Cre and MMTV-Cre and loss of Tcfap2c in the mammary gland epithelial cells resulted in impaired ductal branching and a reduction in the luminal cell compartment with a concomitant increase in the basal cell population at maturity.^13,14 Importantly, SOX2-Cre mediated loss of Tcfap2c leads to impaired mammary gland development into adulthood, while the MMTV-Cre/Tcfap2c^L/L mouse mammary glands appear to be normal in mature animals.¹⁴ The differences are likely due to MMTV-Cre resulting in KO of Tcfap2c expression MMEC, while SOX2-Cre leads to deletion of Tcfap2c in the whole animal.^13,14

The AP-2 factors have a critical role in breast cancer oncogenesis and progression. In luminal breast cancer cell lines, TFAP2C regulates the expression of ERα and many ERα-associated genes.¹⁵ Loss of TFAP2C in luminal breast cancer cell lines induced epithelial-mesenchymal transition characterized by repression of luminal gene expression and induction of basal-associated genes with an expansion of cells expressing cancer stem cell markers.¹⁴ Interestingly, the transcriptional activity of TFAP2A at luminal gene promoters was blocked by sumoylation and inhibiting SUMO conjugation of TFAP2A allowed this AP-2 family member to acquire TFAP2C-like transcriptional activity.¹⁶ In addition, AP-2 factors have been implicated in the transcriptional regulation of the HER2/ERBB2 promoter.^17–20 Further, the HER2 breast cancer subtype has been reported to demonstrate dependency on TFAP2C, with knockdown inducing apoptosis.²¹ Knockdown of TFAP2C in breast cancer cell lines partially downregulated expression of HER2/ERBB2, though the effects were not uniform for all siRNAs or cell lines.^19,21 Of particular note, the effects on cell survival with knockdown of TFAP2C were not reversed by re-expression of HER2/ERBB2 via a heterologous promoter indicating that TFAP2C regulates the expression of additional genes that mediate cell survival.²¹ An analysis of clinical specimens has shown that the expression of HER2/ERBB2 demonstrated a significant correlation with TFAP2C expression in primary breast cancer.^22,23 These studies established a central role for TFAP2C in regulating gene expression in the HER2 breast cancer subtype.

There have been limited investigations into the role of TFAP2C in HER2/Neu-driven breast cancer oncogenesis. Tumorigenesis in mice expressing MMTV-Neu has been examined in female mice that were bitransgenic for MMTV-Tcfap2c.²⁴ Overexpression of Tcfap2c only slightly prolonged tumor latency by ~ 1 week. In contrast, early-stage tumors with Tcfap2c overexpression demonstrated increased proliferation and a higher tumor grade, leading to the conclusion that overexpression of Tcfap2c promoted tumor progression. Although the findings indicate that Tcfap2c influenced oncogenesis of Neu-derived mammary tumors, conclusion based on forced overexpression may be due to non-physiologic effects. In an attempt to clarify the role of TFAP2C in HER2/Neu-activated oncogenesis, we developed a tumorigenesis model based on KO of the Tcfap2c gene with MMTV-Cre in Tcfap2c-floxed animals expressing the MMTV-Neu transgene. This system also offers the potential of defining Tcfap2c target genes that are involved in tumorigenesis and cancer progression.

RESULTS

Conditional KO of Tcfap2c delays tumorigenesis

To investigate the role of Tcfap2c in mammary tumorigenesis, we utilized a mouse model of mammary oncogenesis based on overexpression of the rat activated Neu gene with and without conditional KO of the Tcfap2c gene in MMECs.¹⁴ MMTV-Neu/ MMTV-Cre double transgenic mice were crossed with Tcfap2c-floxed animals (Tcfap2c^L/L) to develop female progeny with homozygous deletion of Tcfap2c with the MMTV-Neu transgene. The animals were genotyped and assessed for onset of spontaneous palpable tumor compared to tumors that were found in MMTV-Neu/Tcfap2c^L/L littermate controls. Mice from both groups (25 control and 21 KO) formed tumors and were killed when tumors reached 2 cm. As seen in Figure 1a, KO of the Tcfap2c gene significantly delayed tumor formation according to Kaplan–Meier analysis. Median age of tumor formation in control mice was 27 weeks vs 39 weeks in KO mice (P<0.0001). The average time to tumor formation in mice that formed tumors in the MMTV-Neu/Tcfap2c^L/L animals was 28 weeks vs 36 weeks for MMTV-Neu/MMTV-Cre/Tcfap2c^L/L KO animals (P<0.003; Figure 1b). There were also differences noted in the number of mammary tumors that developed in KO vs control animals; KO animals had an average of one tumor per mouse compared to 1.8 in control mice (P<0.0002; Figures 1c). Tumors were assessed for histologic grade; no differences were identified in tumor grade comparing tumors in control and KO animals (data not shown). Interestingly, while all control mice developed tumors, 19% of the KO mice failed to develop any tumors by one-year of age (P<0.04; Figure 1d). Hence, while tumor grade was not affected, tumorigenesis was altered.

Figure 1.

Conditional KO of Tcfap2c increased tumor latency and decreased tumor burden. (a) Tumor-free survival analysis of mice with homozygous KO of Tcfap2c demonstrated a significant delay (~12 weeks, P<0.0001) in median time for tumor formation compared to control mice that retained Tcfap2c expression. Data include 25 control mice and 21 KO mice. (b) Control mice developed tumors at an average age of 28 weeks vs 36 weeks in KO mice (*P<0.003). (c) KO had an average of one tumor per animal compared to 1.8 in control mice (*P<0.0002). (d) All animals in the control group developed tumors, whereas, 19% of the KO mice failed to develop tumors by 52 weeks of age (*P<0.04, Fisher's exact test).

Tumors in KO and control animals were further analyzed for expression of Tcfap2c, Ki-67 and cleaved caspase-3. Tumors that developed in KO animals were confirmed to have a loss of Tcfap2c protein, though some scattered tumor cells retained Tcfap2c expression (Figure 2, top panel). The decrease in Tcfap2c protein expression in tumors from KO animals was also demonstrated by western blot (Figure 2, top panel). Since tumor outgrowth was significantly delayed, we sought to clarify whether this was due to a change in proliferation, apoptosis or both. When tumors from genetically engineered KO and littermate control mice reached 2 cm, tumors were harvested and immunohistochemistry (IHC) analysis was performed for Ki-67, a nuclear marker of proliferation, and cleaved caspase-3, an executioner caspase and known driver of apoptosis. Figure 2, middle panel shows that there was a 43% decrease in Ki-67 positivity in tumors from KO animals compared to control animals (P<0.05), thus suggesting that proliferation was attenuated by loss of the Tcfap2c gene. Interestingly, Figure 2, bottom panel shows there was also a concomitant 3.3-fold increase in cleaved caspase-3 staining in tumors from KO animals compared to control (P<0.04). Hence, the data suggested that the delay in tumor formation with KO of Tcfap2c likely resulted from both a reduction in proliferation and an increase in apoptosis in the tumors.

Figure 2.

Loss of Tcfap2c decreased proliferation and increased apoptosis. Top panel: IHC and western blot demonstrate a significant reduction in Tcfap2c protein expression in KO compared to control (Ctl) animals. Neu expression by western blot was not altered with KO of Tcfap2c. Middle panel: IHC for Ki-67 shows a significant reduction in proliferative index in KO animals; bar graph shows quantitative analysis demonstrating a 43% decrease in Ki-67 staining in KO compared to control mice (*P<0.05). Bottom panel: IHC for cleaved caspase-3 (caspase) shows increased staining in tumors from KO animals; bar graph shows quantitative analysis demonstrating a 3.3-fold increase in CC3 staining in tumors from KO compared to control animals (*P<0.04).

Egfr expression is attenuated with loss of Tcfap2c in mouse mammary tumors

Loss of TFAP2C expression in ERα-positive luminal breast carcinoma cell lines (MCF-7 and T47-D) led to epithelialmesenchymal transition characterized by a reprograming from a luminal phenotype to a basal-like phenotype.¹⁴ Expression of luminal markers such as CDH1, ESR1 and KRT8 decreased with a concomitant increase in expression of basal-associated genes such as VIM, CD44 and CDH2. Hence, it was reasonable to determine whether KO of Tcfap2c altered the molecular phenotype of HER2-positive mammary cancer in the genetically engineered mouse model. Most HER2-positive mouse mammary cancers, including the ones utilizing MMTV-Neu, express luminal markers, such as cytokeratin 8 (CK8), and lack basal markers, such as cytokeratin 5 (CK5).²⁵ IHC of tumors derived from control and KO genetically engineered mouse model were stained for both of these markers. As seen in Figure 3, tumors from control animals were CK5-negative and CK8-positive confirming that Neu-derived tumors had a luminal phenotype. In contrast to previous studies in luminal development and breast cancer cell line models, loss of Tcfap2c failed to reprogram the Neu-positive mouse tumors, and expression of the makers CK5 and CK8 were not significantly different comparing tumors from control and KO animals (P>0.1 and P>0.4, respectively).

Figure 3.

Loss of Tcfap2c attenuates EGFR expression in mammary cancer. IHC of tumors developing in control and KO mice lack the basal marker CK5 and retain the luminal marker, CK8. Expression of CK5 and CK8 was not significantly different between tumors derived from control (Ctl) and KO mice (CK5, P>0.1; CK8, P>0.4). NS, not significant. Neu expression was demonstrated in nearly 100% of tumor cells from both control and KO mice (P>0.4). Egfr was found to be significantly decreased in tumors derived from the KO mice relative to control (0.69 vs 1.0, *P<0.04).

Since we utilized the MMTV promoter to drive overexpression of the Neu oncogene, we found it unlikely that loss of Tcfap2c would lead to changes in Neu expression through direct regulation of the promoter. However, in human breast cancer cells, TFAP2C regulates expression of ERα/ESR1, which also affect levels of progesterone receptor; hence, it was possible that progesterone receptor, may alter expression from the MMTV promoter. Furthermore, the MMTV long terminal repeat contains a functional AP-2 site.²⁶ Because of the potential effect of Tcfap2c on expression from the MMTV promoter, we examined the expression of Neu protein in tumors developed in control and KO animals. There was no significant difference in Neu expression as determined by western blot (Figure 2) or IHC (Figure 3) comparing tumors from control vs KO animals (P>0.4). To determine the basis for altered tumorigenesis with KO of Tcfap2c, we sought to clarify what Tcfap2c target genes might contribute to increased latency of tumorigenesis. To further investigate the role of Tcfap2c in controlling cell growth, we generated a cell line from a tumor arising in a control animal with the MMTV-Neu/Tcfap2c^L/L genotype. The resulting cell line (MMneu-flAP2C) was transduced with either adenovirus-Cre (ad-Cre) or adenovirus-empty (ad-E). Using array profiling, we identified genes with significant change in expression with KO of Tcfap2c and performed chromatin immunoprecipitation and direct sequencing (ChIP-seq) to identify primary gene targets of Tcfap2c in the mouse tumor cells. A full list of potential target genes defined by changes in expression with KO of Tcfpa2c and peaks present within the regulatory regions of the genes are provided in Supplementary Materials. We determined that the expression of another HER family member, Egfr, was reduced with KO of Tcfap2c. IHC for Egfr was performed in tumors from control and KO animals (Figure 3). We found a 31% reduction in expression of the Egfr/Her1/Erbb1 gene in tumors from KO compared to control animals (P<0.04). This finding suggested that the delay in tumorigenesis identified in the KO animals might be due to reduced expression of Egfr mediated by a loss of Tcfap2c.

Tcfap2c potentiated cell viability and tumor growth

Viability and tumor growth was assessed in the MMneu-flAP2C cell line with and without KO of Tcfap2c. Western blot confirmed appropriate loss of Tcfap2c with Cre expression (Figure 4a). Despite the fact that the MMTV long terminal repeat is known to contain a functional AP-2 element, we show that KO of Tcfap2c had no effect on Neu RNA and protein levels (Figure 4a); however, loss of Tcfap2c resulted in a significant reduction in phosphorylated Erk1/2 (p-Erk) without a change in total Erk1/2 (Figure 4a). Cell viability, measured by MTT assay, showed that loss of Tcfap2c resulted in a 25% reduction in cell viability compared to ad-E transduced cells (P<0.006). Propidium-iodide staining confirmed a significant reduction in cells in S-phase with KO of Tcfap2c (Figure 4c). In addition, there was a noted significant increase in the percentage of cells in G2 (16.20% to 65.87%) upon loss of Tcfap2c, further solidifying the results. To show human relevance and to avoid a bias of using a single cell line, we repeated parallel experiments using the HER2-amplified human breast cancer cell line BT-474. Knockdown of TFAP2C by siRNA was confirmed by western blot, as seen in Figure 4d. In contrast to earlier reports,^19,21 Neu protein levels were not affected by knockdown of TFAP2C in BT-474 (Figure 4d). Following knockdown of TFAP2C, ERK1/2 activation was significantly reduced (Figure 4d) and was associated with a 46% decrease in cell viability (P<0.005, Figure 4e).

Figure 4.

Loss of Tcfap2c decreased Erk activation, cell viability and growth of xenografts. (a) Western blots show that MMneu-AP2Cfl cells tranduced with ad-Cre have loss of Tcfap2c expression, no change in Neu protein and RNA (qPCR) and reduction in P-Erk activation. (b) Relative cell viability was reduced 25% following loss of Tcfap2c (*P<0.006). (c) Flow cytometry demonstrated that KO of Tcfap2c significantly reduced S-phase (34.12% vs 5.26%) and increased cells in G2/M (16.2% vs 65.87%). (d) Parallel experiments performed with knockdown of TFAP2C in BT-474 cells show no change in Neu expression with knockdown of TFAP2C (C) compared to non-targeting siRNA (NT) and a reduction in p-ERK activation. (e) Knockdown of TFAP2C (C) in BT-474 cells induced a 46% decrease in cellular viability compared to non-targeting (NT) siRNA (*P<0.005). (f) Athymic nude mice xenografted with mammary cancer cells with KO of Tcfap2c (ad-Cre) exhibited a significant prolongation of overall survival (tumors reaching 2 cm) compared to control cells with ad-empty (20 days vs 28 days, P<0.003). Five mice were xenografted per cell line.

The effect of Tcfap2c on tumorigenesis was assessed by performing allogeneic transplants into nude mice using MMneu-flAP2C cells with and without KO of Tcfap2c (five control and 5 KO). Approximately, 4 × 10⁶ cells were xenografted in parallel into the right flank of athymic nude mice. Tumors were allowed to grow until they reached a diameter of 2 cm in the longest dimension, when euthanasia was required due to tumor burden. Figure 4f shows that the median time to reach 2 cm with cells that retained Tcfap2c was 20 days. In contrast, mice xenografted with cells with KO of Tcfap2c had a mean overall survival of 28 days (20 vs 28 days, P<0.003). This suggested that loss of Tcfap2c inhibited tumor growth in vivo. Furthermore, the findings indicate that Tcfap2c regulated cell growth in Neu-activated mouse mammary cancer cells through mechanisms that involve Erk activation. Hence, the potential role of Egfr was examined in greater detail.

Egfr is a direct target of tcfap2c

As mentioned previously, gene expression profiling and IHC suggested that tumors lacking Tcfap2c had decreased levels of EGFR expression compared to tumors in control animals (Figure 3). Thus, we compared Egfr expression in MMneu-flAP2C cells after ad-E or ad-Cre by western blot. As seen in Figure 5a, loss of Tcfap2c significantly reduced expression of Egfr mRNA and protein expression (P<0.05). As noted above, we subjected the mouse cell line to ChIP-seq analysis performed with anti-Tcfap2c antibody and identified genomic binding sites for Tcfap2c in the mouse cell line (GSE61394). As seen in Figure 5b, a significant Tcfap2c peak was identified associated with the transcriptional start site of the Egfr promoter. ChIP-seq was also performed for RNA pol II and the binding peak was similarly associated with the peak for Tcfap2c in the Egfr promoter. Parallel experiments were performed with BT-474 cells (Figures 5b and c). ChIP-seq identified a TFAP2C peak associated with the homologous region of the transcriptional start site of the human EGFR promoter. Similarly, knockdown of TFAP2C by siRNA repressed EGFR mRNA and protein expression (P<0.05). These data suggest that TFAP2C and its mouse homolog directly regulate the expression of EGFR in both human and murine HER2-amplified mammary carcinoma cells. Hence, the role of Egfr in accounting for the effects on cellular viability mediated by Tcfap2c was further investigated.

Figure 5.

Egfr is a direct target gene of Tcfap2c. (a) Loss of Tcfap2c through Cre-mediated excision leads to decreased levels of Egfr RNA (top graph) and protein (lower western blot) in mouse mammary cancer cells (*P<0.05). (b) ChIP-seq performed in mouse cells demonstrated peaks for Tcfap2c and RNA pol II near the TSS of the Egfr gene promoter (top) and similar peak for TFAP2C associated with the EGFR gene in human BT-474 cells (bottom). (c) Knockdown with TFAP2C (C) or control non-targeting (NT) siRNA attenuated expression of EGFR RNA (top graph) and protein (bottom western blots) in BT-474 cells (*P<0.05).

The Egfr pathway contributes to cellular viability

We sought to confirm that the Egfr pathway contributed to cell viability in the murine cancer cell line. Although EGFR and Neu/ HER2/ERBB2 are both members of the HER family of epithelial receptor tyrosine kinases and known to potentiate proliferation, we asked the question whether stimulation and inhibition of Egfr in the murine carcinoma cells would lead to an increase and decrease in cellular viability, respectively. After stimulation with the growth factor Egf, MMneu-flAP2C cells exhibited an increased level of p-Erk activation by western blot (Figure 6a) and a 24% increase in cell viability (P<0.05, Figure 6b). Unlike other growth factors such as HB-EGF and the family of neuregulins,²⁷ EGF preferentially binds to EGFR as opposed to other receptor tyrosine kinases of the HER family. Even so, we examined the effects of pharmacologically inhibiting Egfr with PD153035, a small molecular inhibitor, which blocks Egfr phosphorylation but has no effect on HER2/Neu and ERBB3/HER3 phosphorylation at the concentrations used.²⁸ As seen in Figure 6a, Egf-induced Erk activation was effectively blocked by pharmacological inhibition of Egfr with PD153035 treatment. As seen in Figure 6b, cell viability was similarly affected by blockade of the Egfr pathway with PD153035 treatment (P<0.05).

Figure 6.

Egfr pathway-mediated cell viability through Tcfap2c. Experiments with MMneu-flAP2C cells were used to confirm Tcfap2c-regulated activation of Erk was mediated through activation of Egfr. (a) Western blots show that Egf activated P-Erk and PD153035 (PD) blocked activation. (b) Relative cell viability was increased 24% by Egf stimulation, whereas, PD153035 treatment reduced relative viability to 41% compared to untreated control cells (*P<0.05). (c) Western blots show that knockdown of Tcfap2c (C) or Egfr by siRNA reduced Egfr and P-Erk activation to a similar extent compared to NT control transfection. (d) Cell viability was reduced by knockdown of Egfr and Tcfap2c (C) by 28% and 41%, respectively (*P<0.02 and **P<0.003).

To further demonstrate that the effects were mediated through Tcfap2c, knockdown of Tcfap2c by siRNA diminished Egfr expression and p-Erk activation (Figure 6c). In addition, knockdown of Egfr with siRNA similarly reduced Erk activation, suggesting that the effects of Tcfap2c are mediated, at least in part, through regulation of Egfr expression. Cell viability was examined in parallel. As seen in Figure 6d, knockdown of Egfr and Tcfap2c both reduced cell viability by 28% and 41%, respectively (P<0.02 and P<0.003). The fact that knockdown of Tcfap2c had a greater effect than knockdown of Egfr is not surprising considering that Tcfap2c exhibits pleiotropic effects, controlling expression of a wide number of genes that regulate multiple pathways contributing to growth and survival of breast cancer. The results confirm that the effects on Erk activation and cell viability are directly related to activation of the Egfr signaling pathway.

DISCUSSION

Findings from the current study add to the expanding evidence that the AP-2 family of transcription factors regulate the expression of genes that have a significant influence on breast cancer tumorigenesis and progression. By using the complementary approaches of genetically engineered mouse model, xeno-graft models with Neu-activation and complementary studies in HER2-amplified human breast cancer cell lines, additional evidence for a critical role of TFAP2C in the HER2 subtype of breast cancer has been established. KO of Tcfap2c in MMECs significantly delayed tumorigenesis and tumors that formed demonstrated decreased proliferation and increased apoptosis. These effects were, at least in part, accountable through the regulation of Egfr, which was shown to be a direct target of Tcfap2c. Where they can be compared, the findings have been confirmed in HER2-amplified human breast cancer cells, supporting the clinical relevance of the model. Hence, the genetically engineered mouse model presented herein may offer a valuable animal model to test additional TFAP2C target genes that may contribute to oncogenesis and cancer progression.

Effect on Tumorigenesis Related to Oncogenesis vs Tumor Growth

The finding that KO of Tcfap2c increased the latency of tumorigenesis could result from effects on the ‘cell of origin’ of oncogenesis or from a direct effect on the growth or viability of cancer cells. Previously reported effects of the KO of Tcfap2c in the luminal MMEC compartment may be mediated, in part, through its role in differentiation of the luminal progenitor (LP) population.¹⁴ Recent findings have provided evidence that even basal-like breast cancers are derived from the LP population.²⁹ Though controversy exists, there appears to be a consensus developing that mammary cancer developing in MMTV-Neu transgenic animals are derived from multipotent alveolar progenitors,³⁰ which are derived from the LP population.^31,32 In addition to the luminal B (ERα+/HER2+) breast cancer subtype, many cancers of the HER2 (ERα − /HER2+) subtype have a luminal expression profile.⁴ Therefore, since KO of Tcfap2c alters the development of luminal MMEC, the increased latency observed with loss of Tcfap2c might be due to a reduction or functional inhibition of the LP population. Though this possibility cannot be formally excluded, findings from our xenograft experiments revealing significantly reduced cell growth and tumor outgrowth with KO of Tcfap2c suggest that the observations can be accounted for by a direct effect of Tcfap2c on the growth and survival of mammary cancer cells. In addition, the tumors that formed in KO animals maintained a luminal phenotype based on cytokeratin profile expression and is consistent with development from an intact LP population. Additional investigations into the role of Tcfap2c in the mammary stem cell population will be needed to clarify alternate mechanisms through which TFAP2C may influence oncogenesis.

Cancer growth and progression mediated by EGFR

EGFR is a member of the human epidermal growth factor receptor family of receptor tyrosine kinases and has been an attractive target for treatment of breast cancer.³³ EGFR is more highly expressed in triple-negative breast cancer and targeting EGFR offers the potential of directed therapy in this aggressive breast cancer subtype.³⁴ EGFR is also expressed in the HER2 breast cancer subtype and expression is predictive of worse outcome.³⁵ The associated increased aggressiveness of HER2 and EGFR dual expression profile may further relate to the formation of HER1/ HER2 heterodimers,³⁶ although other mechanisms cannot be ruled out and requires further future investigation. These finding have implicated EGFR as a potential target for therapy but tyrosine kinase inhibitors with anti-EGFR activity and anti-EGFR monoclonal antibodies have not demonstrated a significant clinical response.³³ One potential reason for the disappointing results may be due to an incomplete understanding of the molecular basis for EGFR signaling in breast cancer. Although we cannot exclude additional possible molecular mechanisms, the current findings support a model for EGFR signaling contributing to breast cancer proliferation and cell survival in the HER2 cancer subtype.

Previous studies have demonstrated the dependency of HER2 breast cancer on TFAP2C with knockdown of TFAP2C inducing apoptosis.²¹ Although TFAP2C was shown to regulate the expression of HER2/ERBB2, the apoptotic effects induced by a loss of TFAP2C was not directly related to HER2. In these reported studies, induction of apoptosis also correlated with sensitivity to the drug lapatinib, an inhibitor of the ErbB family of growth factor receptors. The findings suggested that TFAP2C promoted cell survival via regulation of other ErbB family members. One possibility was that in the HER2 cell lines, ERBB3 maintains cell survival. Indeed, previous studies have shown that loss of ErbB3 reduced luminal MMEC proliferation and survival.³⁷ Similar to findings for Tcfap2c, loss of ErbB3 was also associated with a luminal to basal transition in mammary epithelial cells. Further studies are needed to determine whether Tcfap2c regulates ErbB3 expression; however, in light of the present findings, it is interesting to question whether regulation of Egfr by Tcfap2c may also account for some of the sensitivity to lapatinib, which is known to suppress EGFR signaling.³⁸ Interestingly, previous studies examining a Tcfap2a/AP-2α KO mouse model reported that Tcfap2a repressed Egfr expression in skin epidermis.³⁹ This raises the question of whether Tcfap2a and Tcfap2c have opposing effects or whether AP-2 factors have different mechanisms of action in regulation of the Egfr promoter in breast vs skin epithelial cells. In breast cancer models, we have previously demonstrated significant differences and, in some cases, opposing effects comparing transcriptional activity of TFAP2A vs TFAP2C.^15,16 The current study confirms that Egfr is an AP-2 target gene but further study is needed to clarify the role of the various AP-2 family members in different tissues.

On balance, the data support the conclusion that Tcfap2c regulates the expression of several HER family members and regulation of these pathways accounts for some of the physiologic effects that Tcfap2c has on HER2-positive breast cancer growth and progression. This regulatory mechanism and its influence on tumor growth mediated by TFAP2C in HER2-amplified breast cancer contrasts with the effects exhibited in ERα-positive and triple-negative breast cancer (Figure 7). Previous studies have demonstrated the role of TFAP2C in maintaining the luminal cancer phenotype by inducing the expression of ERα-associated luminal gene and repressing basal-associated genes.^14–16 In the triple-negative basal cancer subtype, TFAP2C represses key basal markers including the cancer stem cell marker, CD44^14,16,40 and the sumoylation pathway is critical in regulating AP-2 transcriptional activity.¹⁶ The current data expand the role of TFAP2C in HER2-amplified breast cancer by showing that the effects of TFAP2C on cell growth and survival is mediated through regulation of EGFR. In the context of other studies, TFAP2C appears to regulate the expression of all members of the HER family in HER2 breast cancer. From a clinical perspective, approaches to alter TFAP2C activity may offer the potential of altering cell growth and survival by altering expression of several members of the HER family.

Figure 7.

The role of TFAP2C in regulation of breast cancer subtypes. The TFAP2C transcription factor plays a critical role in gene regulation in multiple breast cancer subtypes. TFAP2C is critical for maintenance of the luminal breast cancer phenotype by regulating the expression of ERα-associated genes, which control luminal cell growth and differentiation. TFAP2C represses basal-associated genes, such as CD44, and the sumoylation pathway plays a significant role in regulating AP-2 activity thus maintaining the basal phenotype. Herein, the current data demonstrate an important role for TFAP2C regulation of the HER family of receptor tyrosine kinases, which control tumorigenesis, cell growth and survival of HER2-amplified breast cancer subtype.

MATERIALS AND METHODS

Cell lines

BT-474 cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and maintained as previously described.⁴¹ Briefly, the MMneu-flAP2C cell line was harvested from a mammary gland tumor from a mouse that carried both the MMTV-Neu and Tcfap2c loxP-flanked alleles, but had never been exposed to Cre. The tumor was mechanically minced, followed by incubation with collagenase/hyaluronidase (Stem Cell Technologies, Vancouver, BC, Canada) according to manufacturer's suggestions to create a single cell suspension passaged in cell culture. The mouse mammary cancer cells were maintained in IMDM (Iscove's Modified Dulbecco's Medium, Life Technologies, Grand Island, NY, USA) media with 10% fetal bovine serum. EGF was purchased from BD Biosciences (Bedford, MA, USA) and stimulation was carried according to manufacturer's suggestions. PD153005 was purchased from Selleck Chemicals (Houston, TX, USA) and was used at a concentrations of <350 nm.

Chromatin immunoprecipitation and direct sequencing

ChIP-seq was accomplished as previously described using mouse tumor cells and anti-Tcfap2c and anti-RNA pol II antibodies (Millipore, Billerica, MA, USA).⁴² ChIP-Seq data are available in the GEO database (National Center for Biotechnology Information, Bethesda, MD, USA) GSE36351 and GSE61394.

Adenovirus infection

Ad-Cre and ad-E were acquired from the University of Iowa DNA Core Lab. Infection was carried out at a multiplicity of infection of 100. All experiments were performed with cells at an early passage number (<20).

RNA isolation, complementary DNA synthesis and real-time PCR

Total RNA was harvested from cell lines using the Rneasy Mini Kit (Qiagen, Valencia, CA, USA). Complementary DNA was synthesized from 1 mg of total RNA using the High Capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, CA, USA). Gene expression was measured by qPCR. TaqMan primers and probes (Life Technologies) were used for the following mouse genes: Tfap2c Mm00493473_m1, Egfr Mm00433023_m1 and Gapdh Mm00484668_m1. TaqMan primers and probes (Life Technologies) were used for the following human genes: TFAP2C Hs00231476_m1, EGFR Hs01076078_m1 and GAPDH Hs02758991_g1. Expression values were normalized to average Gapdh, the endogenous control.

Western blots

Protein was isolated in RIPA buffer (Millipore), supplemented with Halt protease inhibitor (Thermo Scientific, Rockford, IL, USA). Primary antibodies were used according to the manufacturer's instructions: TFAP2C #sc-12762, NEU #sc-284, ERK #sc-93, GAPDH #sc-32233 and EGFR sc-03 (Santa Cruz Biotechnology, Santa Cruz, CA, USA); pERK #4370 (Cell Signaling, Danvers, MA, USA). Secondary antibodies were used according to manufacturer's instructions: anti-rabbit HRP #sc-2030 and anti-mouse HRP #sc-2005. Protein was visualized with SuperSignal West Dura extended duration substrate (Thermo Scientific) and SuperSignal West Femto maximum sensitivity substrate (Thermo Scientific).

Microarray

Two hundred nanograms of total RNA was isolated. Hybridization and microarray analysis were carried out with help from the University of Iowa DNA Core Facility and Bioinformatics Core Facility. Arrays were performed in triplicates. Analysis was carried out with GenomeStudio software (Illumina, San Diego, CA, USA).

Genetically engineered mice and IHC

All animal protocols adhered to guidelines by the University of Iowa Institutional Animal Care and Use Committee. Mice harboring loxP-flanked Tcfap2c alleles were obtained from Dr Trevor Williams and maintained as described.¹⁴ Mice harboring the MMTV-Neu allele were obtained from Dr Bill Muller. Mice harboring the MMTV-Cre allele were purchased from Jackson Laboratories (Bar Harbor, ME, USA). F4 progeny carrying Tcfap2c allele carrying or lacking MMTV-Cre were crossed with female mice harboring MMTV-Neu. These progeny were used for analysis. Genotyping was carried out with tail snips and analyzed by Transnetyx (Cordova, TN, USA). Tumors (n = 3 tumors) were harvested from mice, formalin fixed and paraffin embedded. Hematoxylin and eosin and IHC analyses were completed with the aid from the University of Iowa Pathology Core Lab with the aid from a veterinary pathologist. Categorical data were analyzed by Fisher's exact test. IHC was analyzed by t-test. Survival curves were analyzed by long-rank test.

Tumor xenografts

Six-week-old female athymic mice were purchased from Jackson Laboratory (Bar Harbor, ME, USA). Approximately, 4 × 10⁶ cells in a volume of 200 ul (50% media, 50% BD Bioscience Matrigel (Bedford, MA, USA)) were injected subcutaneously in the right flank of each mouse. Survival curves were analyzed by long-rank test.

Small interfering RNA transfections

siRNA constructs were obtained for the following genes: TFAP2C (D-005238-01, Dharmacon (Lafayette, CO, USA)), Tcfap2c (4390771, Life Technologies), Egfr (4390771, Life Technologies) and NonTargeting (D-001210-01-05, Dharmacon). siRNA transfection was performed according to manufacturer's protocols using Lipofectamine RNAiMAX (Life Technologies). Following 72-h incubations, cells were harvested and used for study.

MTT Cell viability assay

Cells were plated in at least triplicates and allowed to adhere overnight. MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) was acquired from Life Technologies and the assay was carried out according to the manufacturer's protocol. Plates were analyzed by Infinite 200 Pro plate reader (Tecan, Zurich, Switzerland) at an absorbance wavelength of 590 nm with a reference wavelength of 630 nm. Statistical analysis was performed by t-test.

Flow cytometry

Cells were isolated and 70% ethanol was gradually added while vortexing. Cells were permeabilized with 0.5% Triton X-100. Cells were then treated with RNAse and propidium iodide (Life Technologies) and subjected to flow cytometry.