ErbB2-driven downregulation of the transcription factor Irf6 in breast epithelial cells is required for their 3D growth

The following compounds were used: lapatinib (Selleckchem, Houston, TX, USA), SCH772984 (Selleckchem), and trastuzumab (Roche, Mississauga, ON, Canada) (See Additional file 1: Supplemental Methods for addtional information).

The pEGFP-C1 plasmid encoding the wild-type green fluorescent protein (GFP) was obtained from Clontech (Mountain View, CA, USA). The pBABE-hygro expression vector was purchased from Addgene (Cambridge, MA, USA). The expression vector pcDNA-HA encoding the full-length human Irf6 cDNA (pcDNA-HAIrf6) was provided by Dr. Antonio Costanzo (University of Rome, Italy). The pcDNA expression vector encoding the full-length human ΔNp63α-FLAG was obtained from Addgene. Generation of Irf6- and ΔNp63α-encoding pBabe-hygro expression vectors is described in Additional file 1: Supplemental Methods. pHIT and pVSVG retroviral vectors were provided by Dr. P. Lee (Dalhousie University, Halifax, NS, Canada). pBABE-hygro retroviral expression vector was purchased from Addgene.

MCF10A cells and their derivatives MCF-ErbB2mut and MCF-ErbB2 were provided by Dr. M. Reginato (Drexel University, Philadelphia, PA, USA). The generation and use of these variants is described elsewhere [16, 17]. MCF10A cells were authenticated by the American Type Culture Collection (Manassas, VA, USA) by 17 short tandem repeat analysis. Lack of mycoplasma contamination in MCF10A, MCF-ErbB2mut, and MCF-ErbB2 cells was established by use of MycoFluor Mycoplasma Detection Kit (Molecular Probes, Eugene, OR, USA) according to the manufacturer’s instructions. MCF-ErbB2mut cells were referred to as pBabe-NeuT, MCF-ErbB2 cells as pBabe-NeuN, and MCF-MekDD as pBabe-MEK2-DD as reported previously [17]. BT-474 cells (American Type Culture Collection) were cultured in Hybri-Care medium (American Type Culture Collection), 10% FBS, 100 U/ml penicillin (Thermo Fisher Scientific, Waltham, MA, USA), 100 μg/ml streptomycin (Thermo Fisher Scientific), 0.29 mg/ml L-glutamine (Thermo Fisher Scientific). AU-565 cells (American Type Culture Collection) and HCC1419 cells (American Type Culture Collection) were cultured in RPMI 1640 medium (Thermo Fisher Scientific), 10% FBS (Sigma-Aldrich, St. Louis, MO, USA), 100 U/ml penicillin (Thermo Fisher Scientific), 100 μg/ml streptomycin (Thermo Fisher Scientific), and 0.29 mg/ml L-glutamine (Thermo Fisher Scientific). 293T cells (provided by Dr. A. Stadnyk, Dalhousie University) were cultured in DMEM (Thermo Fisher Scientific), 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 0.29 mg/ml L-glutamine. Primary human mammary epithelial cells (HMEC) (Lonza, Walkersville, MD, USA) were cultured in mammary epithelial growth medium (Lonza) supplemented with bovine pituitary extract, human epidermal growth factor, insulin, hydrocortisone, gentamicin (30 mg/ml), and amphotericin (15 mg/ml). To detach the cells from the ECM, they were plated in suspension culture above a layer of 1% sea plaque agarose polymerized in their respective culture medium not containing any additional ingredients.

BT-474 cells (1 × 10⁶) were cultured in suspension for 2 weeks in the presence of 5 μg/ml trastuzumab. The surviving cells were then maintained in the monolayer culture in the presence of 5 μg/ml trastuzumab for 4 months.

This assay was performed as described previously [10]. The following antibodies were used in this study: Anti-Irf6, anti-Blnk, anti-ErbB2, anti-RSK, and anti-phospho-RSK (all from Cell Signaling Technology, Danvers, MA, USA), anti-ΔNp63 and anti-TAp63 (BioLegend, San Diego, CA, USA), anti-CDK4 (Santa Cruz Biotechnology, Dallas, TX, USA), and anti-β-actin (Santa Cruz Biotechnology and Sigma-Aldrich).

RNA interference was performed as described previously [18]. The sequences of the sense strands of the RNAs used in this study were as follows: control RNA (siCONTROL nontargeting small interfering RNA [siRNA] 1), UGUUGUUUGAGGGGAACGGTT; Irf6 siRNA 5, GGAAACUCAUCUUGGUUCA; Irf6 siRNA 7, CGUUUGAGAUCUACUUAUG; p63 siRNA 14, CGACAGUCUUGUACAAUUU; p63 siRNA 15, GAUGAACUGUUAUACUUAC. All the RNAs were purchased from Dharmacon (Lafayette, CO, USA).

This procedure was performed as described previously [10]. Primers used to amplify the Irf6 messenger RNA (mRNA) were as follows: CAAAACTGAACCCCTGGAGATGGA and CCACGGTACTGAAAC TTGATGTCC. Primers used to amplify the 18S rRNA were as follows: ATAGTCAAGTTCG ACCGTCTTC and GTTGATTAAGTCCCTGCCCTT.

MCF-ErbB2 cells (5 × 10⁵) were cultured in a 60-mm dish in the presence of 4 μg of either pcDNA-HA or pcDNA-HA-IRF6, 0.8 μg of the pEGFP-C1 expression vector, and 4 μg of Lipofectamine 2000 reagent in 2.0 ml of Opti-MEM medium (Thermo Fisher Scientific). Six hours later, the medium was replaced with 4 ml of the fresh medium normally used for culturing these cells. The cells were cultured for 72 h, harvested, and used for subsequent assays.

293T cells (2 × 10⁶) were incubated with 5 μg of either control pBabehygro expression vector or pBabehygro-HA-Irf6 or pBabe-ΔNp63α FLAG or pBabe TAp63α expression vector and 2.5 μg of pHIT and 2.5 μg of pVSVG expression vectors encoding retroviral proteins in the presence of 20 μl of Lipofectamine 2000 reagent in 6 ml of Opti-MEM medium. The medium was changed 4 h later to DMEM containing 10% FBS. The medium was collected 48 h later and filtered through a 0.45-μm filter unit. The viral supernatant containing either the control virus or that encoding Irf6 or ΔNp63α was added to 2.5 × 10⁵ MCF-ErbB2 cells grown on a 60-mm dish in the presence of 8 μg/ml polybrene in the presence of 450 μg/ml hygromycin (in case of ΔNp63α), 300 μg/ml hygromycin (in case of ΤΑp63α) or 300 μg/ml hygromycin (in case of Irf6) for 48 h. The cells were then harvested and used for the assays described herein.

After transfection with respective RNAs, MCF10A cells were plated in monolayers either immediately or after being cultured in suspension. Cells were allowed to form colonies for 7–10 days. The colonies were then stained with crystal violet and counted.

After transfection with siRNAs or respective expression vectors in the presence of the pEGFP-C1 expression vector, cells were cultured in suspension, harvested, and washed with PBS followed by centrifugation at 1500 rpm for 5 min at room temperature. The cell pellet was resuspended and fixed in 50 μl of 4% paraformaldehyde solution for 30 min at room temperature. The cells were then washed and resuspended in 30–50 μl of 20 μg/ml Hoechst 33258 (Molecular Probes) dissolved in PBS. Tubes with cell samples were then coded to ensure that the samples were assayed blindly, and total cell population (in case of siRNA-transfected cells) or GFP-positive (green) cells (in case of the cells transfected with the expression vectors) were analyzed by use of a Zeiss fluorescence microscope (Carl Zeiss Microscopy, Toronto, ON, Canada) for the presence of Hoechst 33258-positive (blue) condensed or fragmented nuclei (characteristic features of apoptosis) using respective light filters.

An apoptosis detection kit (Chemicon, Temecula, CA, USA) was used. After harvesting, the cells were washed with PBS. Cells were further resuspended in the binding buffer provided by the manufacturer at a concentration of 10⁶ cells/ml. We then mixed 200 μl of the cell suspension with 3 μl of fluorescein isothiocyanate-conjugated annexin V and 2 μl of propidium iodide (PI). This solution was subsequently incubated for 15 min at room temperature. The FACSCalibur system (BD Biosciences, Franklin Lakes, NJ, USA) was used for detection of fluorescent cells.

Cells were suspended in 2 ml of their respective growth medium containing 0.3% of melted Bacto agar. Cell suspensions were added to a 60-mm plate covered with a 2-ml layer of solidified 0.5% Bacto agar polymerized in the same medium. Cell colonies were counted after 7–10 days.

Upon research ethics board approval from the local health authority, a list of patients with ErbB2-positive primary invasive breast cancer who underwent neoadjuvant chemotherapy and ErbB2-targeted therapy was obtained from the institution’s pharmacy information system (BDM Pharmacy; BDM IT Solutions, Saskatoon, SK, Canada) and a laboratory information system (Cerner Millennium; Cerner, North Kansas City, MO, USA). All patients were treated with neoadjuvant trastuzumab plus six cycles of fluorouracil, epirubicin, cyclophosphamide, and docetaxel prior to surgical resection of the tumor. A cohort of 11 consecutive cases was randomly selected from the 2011–2015 list for assessment of Irf6 expression in order to ensure an adequate follow-up interval. Representative tissue blocks from the diagnostic core biopsies and postneoadjuvant excisional specimens were selected in each case for Irf6 IHC after review of the H&E- as well as anti-ErbB2-stained IHC slides to confirm diagnosis and ErbB2 positivity. Formalin-fixed, paraffin-embedded core biopsies and excisional specimens underwent heat-induced epitope retrieval for 24 min in Cell Conditioning 1 (Ventana Medical Systems, Tucson, AZ, USA) followed by 32-min incubation in 1:100 dilution of primary Irf6 antibody (clone 2A12; MyBioSource, San Diego, CA, USA). The reaction was detected using the OptiView polymer detection system on a Ventana Benchmark Ultra platform (Ventana Medical Systems). Cells were also counterstained with hematoxylin. Nuclear positivity was scored on the preneoadjuvant core biopsies and postneoadjuvant excisional specimens following Irf6 IHC staining by manually counting positively stained nuclei as a percentage of all breast cancer cells. At least 400 cells were counted in each case, with the exception of one sample in which only 53 malignant cells were counted after neoadjuvant therapy (all cells were counted).

Statistical analysis of the data in Fig. 7a was performed by using the unpaired Student’s t test. Statistical analysis of all other data was performed by using the chi-square test for goodness of fit.

In an effort to identify the mechanisms by which ErbB2 blocks breast cancer cell anoikis, we used MCF10A cells, which are spontaneously immortalized highly anoikis-susceptible human nonmalignant breast epithelial cells [11]. They do not produce ErbB2 in 2D culture (when the cells are attached to the ECM) or in 3D culture (when they are detached from the ECM) (Fig. 1a, d). We also used a published anoikis-resistant derivative of MCF10A cells, MCF-ErbB2 cells, which were generated by infection of MCF10A cells with a retrovirus encoding the wild-type ErbB2 [16, 17] (Fig. 1a, d). We noticed that ErbB2 strongly downregulates Irf6 mRNA and protein in MCF10A cells in 3D culture (Fig. 1b, c).

Irf6 is a member of the interferon-regulatory factor family of transcription factors [19]. Irf6 upregulation in cells can kill them by apoptosis [19, 20]. Importantly, Irf6 is upregulated in the breast during mammary gland regression upon weaning, and such regression likely involves breast epithelial cell anoikis [21]. Moreover, Irf6 protein tends to be downregulated in breast, nasopharyngeal, and squamous cell carcinomas [22‐24]. Irf6 can be phosphorylated at several amino acid residues and is often detected as a doublet on Western blots [25]. At least two of these phosphorylation events are required for Irf6 transcriptional activity [26].

We observed that lapatinib, a small-molecule ErbB2/EGFR inhibitor used for treatment of ErbB2-positive breast cancer, strongly upregulates Irf6 in detached ErbB2-positive human breast cancer cells BT-474, AU-565, and HCC-1419 (Fig. 1d, e) [4]. Furthermore, we found that the anti-ErbB2 antibody trastuzumab used for ErbB2-positive breast cancer treatment upregulates Irf6 in BT-474 cells much more noticeably than in their variant selected for trastuzumab resistance by prolonged cell exposure to trastuzumab in 3D culture (Fig. 1f) [3]. Thus, ErbB2 downregulates Irf6 in detached human breast cancer cells.

We further found that detachment of MCF10A cells or that of anoikis-susceptible nonmalignant primary HMEC from the ECM upregulates Irf6 (Fig. 2a, b) [27]. Thus, detachment-induced Irf6 upregulation is not unique to MCF10A cells.

Others found that Irf6 knockdown in human keratinocytes by siRNAs downregulates several mRNAs, including that encoding the proapoptotic protein Blnk [24, 28]. We noticed that, similar to what was observed in the case of Irf6, detachment from the ECM strongly upregulates Blnk in MCF10A cells (Fig. 2c). In addition, we found that Irf6 knockdown in MCF10A cells by two different siRNAs (Fig. 2d) downregulates Blnk in 3D culture (Fig. 2e), indicating that Irf6 is capable of controlling gene expression under these conditions. Moreover, we observed that Irf6 knockdown increases survival of detached MCF10A cells (Fig. 2f) and substantially reduces their apoptosis in 3D culture (Fig. 2g). Hence, detachment-induced Irf6 upregulation contributes to anoikis of nonmalignant breast epithelial cells.

To reverse the effect of ErbB2 on Irf6, we transiently expressed ectopic hemagglutinin (HA)-tagged Irf6 in MCF-ErbB2 cells together with a GFP to visualize the transfected cells by fluorescence microscopy (Fig. 3a). Exogenous Irf6 noticeably increased apoptosis of MCF-ErbB2 cells in 3D culture (Fig. 3b).

To examine the role of Irf6 in the regulation of anoikis of ErbB2-producing cells by a complementary technique, we infected MCF-ErbB2 cells with a retrovirus encoding HA-tagged Irf6 (Fig. 3c). Exogenous Irf6 significantly increased apoptosis of MCF-ErbB2 cells in 3D culture (Fig. 3d) and noticeably reduced their clonogenicity without adhesion to the ECM in soft agar (the ability to grow in agar is a well-known consequence of cancer cell anoikis resistance [11]) (Fig. 3e). Hence, ErbB2-induced Irf6 downregulation is required for anoikis resistance of ErbB2-overproducing breast epithelial cells.

One protein that promotes Irf6 transcription by binding the Irf6 promoter is ∆Np63, a member of the p53 family of transcription factors [29]. The p63 gene is transcribed from two different promoters to yield the transcription factors TAp63 or ∆Np63 containing DNA-binding and oligomerization domains [29]. In addition, TAp63 has an N-terminal transactivation domain. Both TAp63 and ∆Np63 exist as α-, β-, or γ-isoforms generated via alternative splicing [29]. Importantly, ErbB2 causes loss of all p63 isoforms in a mouse model of breast cancer, and none of the p63 isoforms are expressed in human breast tumors [30]. ∆Np63 can trigger pro- and antiapoptotic signals [31, 32]. Enforced downregulation of ∆Np63α in MCF10A cells causes their epithelial-to-mesenchymal transition, whereas ErbB2- or Ras oncoprotein-induced ∆Np63α downregulation in MCF10A and other epithelial cell lines increases cell migration and metastatic potential [33, 34].

We investigated the status of ∆Np63 in MCF10A cells by using ∆Np63-specific antibody validated for the detection of the ∆Np63 isoforms [35]. Similar to published observations, we found that attached MCF10A cells produce only ∆Np63α (Fig. 4a) [34]. We further noticed that when MCF10A cells detach from the ECM, ∆Np63α levels remain unchanged for at least 6 h but decline at 24 h of 3D culture (Fig. 4a). The antibody validated by others for the detection of TAp63 did not detect any TAp63 species in MCF10A cells in 2D or 3D culture (Additional file 2: Figure S1) [35]. Because Irf6 is upregulated in these cells as early as 3 h of 3D culture (Fig. 2a), we reasoned that ∆Np63α could mediate detachment-induced Irf6 upregulation, while ∆Np63a levels in the cells in 3D culture are still high. Indeed, knockdown of p63 by two different siRNAs (Fig. 4b) caused noticeable Irf6 downregulation in MCF10A cells in 3D culture (Fig. 4c and Additional file 3: Figure S2). Importantly, although ∆Np63α levels are relatively elevated in the MCF10A cells in 2D culture (Fig. 4a), this is insufficient for Irf6 upregulation (Fig. 2a). Therefore, Irf6 is likely upregulated in these cells by yet unidentified detachment-induced signals that can upregulate Irf6 only in the presence of ∆Np63α.

The MAPKs Erk1 and Erk2 are major mediators of ErbB2 signaling [36]. ErbB2 induces MAPK signaling by activating a GTPase Ras, which then activates the protein kinase Raf [36]. Raf further phosphorylates and thereby activates the protein kinases Mek1 and Mek2. This allows the Mek kinases to phosphorylate and activate Erk1 and Erk2 [36]. Erk1 and Erk2 then phosphorylate and change the activity of various proteins [36]. We found that treatment with an Erk inhibitor, SCH772984, significantly reduced phosphorylation of the Erk substrate Rsk and noticeably upregulated Irf6 in MCF-ErbB2 cells in 3D culture (Fig. 5a) [37, 38].

We further found that a published derivative of MCF10A cells, MCF-MekDD, which we generated by infection of MCF10A cells with a retrovirus encoding an activated mutant of Mek (an Erk activator), displays significantly lower Irf6 levels than the parental MCF10A cells in 3D culture (Fig. 5b) [39]. Thus, Mek activation is sufficient for Irf6 downregulation in breast epithelial cells detached from the ECM. Collectively, our data indicate that ErbB2 downregulates Irf6 in breast epithelial cells in a MAPK-dependent manner.

Because ∆Np63α is required for detachment-induced Irf6 upregulation in nonmalignant breast epithelial cells (Fig. 4), we investigated the role of ∆Np63α in the effect of ErbB2 on Irf6. We noticed that both ErbB2 and an activated Mek mutant (Fig. 6a, b and Additional file 4: Figure S3) downregulate ∆Np63α in MCF10A cells in 3D culture. We further observed that the Erk inhibitor SCH772984 upregulates ∆Np63α in MCF10-ErbB2 cells in 3D culture (Fig. 6c). Moreover, when this upregulation was blocked by two different p63-specific siRNAs (Fig. 6c), the Erk inhibitor failed to upregulate Irf6 in the cells (Fig. 6d). As expected, the inhibitor blocked phosphorylation of the Erk substrate Rsk in all cases (Fig. 6e). Thus, ErbB2/MAPK downregulates Irf6 by suppressing ∆Np63α-dependent signals in cells detached from the ECM. It is noteworthy that in the case of nonmalignant breast epithelial cells, detachment-induced signals can upregulate Irf6 only in the presence of ∆Np63α (Figs. 4 and 6d). Hence, ErbB2/MAPK signaling could downregulate Irf6 in MCF-ErbB2 cells in 3D culture either by blocking the indicated signals or by downregulating ∆Np63α itself. To distinguish between these possibilities, we infected MCF-ErbB2 cells with a retrovirus encoding ∆Np63α (Fig. 6f). We found that ectopic ∆Np63α does not upregulate Irf6 in these cells in 3D culture (Fig. 6f). Thus, ErbB2/MAPK-driven signals block detachment-induced events that promote ∆Np63α-dependent Irf6 upregulation and, in addition, downregulate ∆Np63α itself in MCF-ErbB2 cells detached from the ECM (see Fig. 6g, h for models describing these scenarios). When Erk activity is blocked, both of these events are reversed, and Irf6 is upregulated (Fig. 6c, d). However, in the absence of the indicated detachment-induced signals, ectopic ∆Np63α by itself cannot upregulate Irf6.

One approach to testing whether ErbB2 downregulates Irf6 in human breast cancer is to examine whether therapies based on the use of a therapeutic anti-ErbB2 antibody trastuzumab upregulate Irf6 in patients’ tumors. Addressing this question requires access to tumor samples before and after the treatment. These samples can be derived from patients with ErbB2-positive, locally advanced breast cancer. Such patients normally receive neoadjuvant trastuzumab and chemotherapy for approximately 3 months, followed by tumor resection and further trastuzumab treatment for up to 9 months [40].

To begin to test the effect of trastuzumab-based therapies on Irf6 in human breast tumors, we used a pilot cohort of 11 patients with locally advanced breast cancers treated at the QEII Health Centre, Halifax, NS, Canada, with neoadjuvant trastuzumab and chemotherapy prior to surgical tumor resection. We obtained tissue sections from the diagnostic core biopsies and from the postneoadjuvant therapy excisional specimens and assessed Irf6 levels in these sections by IHC. In all 11 cases, breast cancer cells displayed very low cytoplasmic Irf6 staining. Likewise, the pretreatment samples showed poorly detectable nuclear Irf6 staining. In contrast, 9 of 11 (81%) of the posttreatment samples displayed various degrees of increased nuclear Irf6 staining (Fig. 7a). Overall, the percentage of Irf6 positive nuclei was increased approximately 4.5-fold after the treatment. Representative IHC data are shown in Fig. 7b–e (additional IHC data are shown in Additional file 5: Figure S4, Additional file 6: Figure S5, Additional file 7: Figure S6). Thus, Irf6 upregulation in patient-derived breast tumors is associated with ErbB2-targeted therapies.