MUC1 alters oncogenic events and transcription in human breast cancer cells

MDA-MB-468 and BT-20 cells (American Type Culture Collection) were cultured in Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA, USA) plus 10% fetal calf serum, 1% Glutamax (Invitrogen) and 1% penicillin/streptomycin. Stable cell lines (468.Neo and 468.MUC1Δ8) were selected with 0.5 mg/ml G418. For epidermal growth factor (EGF) stimulation, MDA-MB-468 cells were serum-starved overnight and treated for 10 minutes at 37°C with 100 ng/ml EGF. Transient siRNA transfection was performed with Lipofectamine2000 (Invitrogen) and 100 nM siRNA oligonucleotides. The commercially available siRNA constructs (all from Dharmacon, Lafayette, CO, USA) were scrambled (siCONTROL non-targeting siRNA #1), or directed against firefly luciferase (siCONTROL non-targeting siRNA #2) or MUC1 (siGENOME smartpool). The independent oligonucleotides designed in our laboratory target sequences beginning at MUC1 codons 882 and 956, and have been described previously [29]. The scrambled siRNA construct was used only in BT-20 cells as it causes a non-specific knockdown of MUC1 in the MDA-MB-468 line.

Two silent mutations (G891A and T894C) were introduced into the MUC1 cDNA (called MUC1Δ8) to make it resistant to the 882 siRNA that targets that region of the mRNA. The mutant cDNA was cloned into the pLNCX.1 vector with neomycin resistance (gift of Joseph Loftus, Mayo Clinic, Arizona, USA). Stable transfection was performed with Lipofectamine2000; cells were selected beginning 24 hours post-transfection and maintained as a polyclonal population.

Cells were lysed in buffer (20 mM HEPES pH 8.0, 150 mM sodium chloride, 1% Triton X-100, 2 mM EDTA) with commercial protease (Complete inhibitor cocktail, Roche, Pleasanton, CA, USA) and phosphatase inhibitors (10 mM sodium fluoride, 2 mM sodium vanadate, 50 μM ammonium molybdate). Protein concentration was determined by BCA (Pierce, Rockford, IL, USA); 50 μg of lysate were loaded on SDS-PAGE gels for each experiment, except for the pMEK1/2 blot in MDA-MB-468, where 150 μg were used. Non-commercial antibodies used were: BC2, a mouse monoclonal to the MUC1-EX (gift of Dr McGuckin, Queensland University, Queensland, Australia), and CT2, an Armenian hamster monoclonal to the MUC1-CT developed in our lab [12]. Antibodies to pMEK1/2, MEK1/2, ERK1/2, Myc, pAKT, AKT, β-tubulin (all Cell Signaling, Danvers, MA, USA), β-actin and dpERK1/2 (both Sigma, St. Louis, MO, USA) were used according to manufacturers' recommendations. All antibodies except β-actin (1:2,500) and dpERK1/2 (1:10,000) were used at 1:1,000 dilution for western blots. Flow cytometric analysis of MUC1 was done with HMPV-FITC, which recognizes the core peptide of the MUC1-EX tandem repeats (Pharmingen, San Diego, CA, USA). Bromodeoxyuridine (BrdU) staining was performed with a fluorescently conjugated antibody to BrdU (BrdU-PE, BD Biosciences, San Diego, CA, USA) as described below. Densitometry was performed using the public domain ImageJ program (developed at the NIH and available at [30]. Each band was measured in three places; the results were averaged and normalized to tubulin to control for loading.

Cells were serum-starved beginning 24 hours post-siRNA transfection. Cells were re-plated in serum-free medium 48 hours post-transfection at 50,000 cells per insert (sized for 24-well plates), with serum-containing medium in the bottom of the growth well as an attractant. Transwell inserts (BD Biosciences) pre-coated with laminin, fibronectin, collagen IV, or control (no matrix) were used, and cells were permitted to invade for 48 hours. At this point (96 hours post-transfection), visual inspection of the growth wells confirmed that negligible numbers of cells went through to the bottom of the plate. Non-invaded cells were swabbed from the tops of half of the inserts ('samples', containing only invaded cells), and retained in the others ('controls', all cells). Inserts were stained for 10 minutes with crystal violet (0.5% in 20% methanol) and washed with water. Membranes were cut out and destained for 10 minutes in 10% acetic acid in a 96-well plate; membranes were removed and absorbance was read at 570 nm. Percent invasion is defined as (absorbance of samples/absorbance of controls) × 100.

Cells were re-plated in quadruplicate 24 hours post-siRNA transfection at 15,000 cells/well (96-well plate) with [³H]thymidine (1 μCi/well), then incubated in normal conditions for 24 hours. At this time (48 hours post-siRNA transfection) excess radioactivity was washed off and the cells were harvested and read on a TopCount plate reader. Statistical analysis was performed using JMP 5.1.2 software (SAS Institute, Inc., Cary, NC, USA); the student's t test was used to determine p values and significance was confirmed with Wilcoxon rank sum and Pearson chi squared analyses.

BrdU (50 μM) was given to cells 48 hours post-siRNA transfection and permitted to incorporate for 1.5 hours. Cells were then washed with PBS, trypsinized, and washed again. BrdU staining was performed according to an adaptation of the manufacturer's protocol: cells were re-suspended in PBS, mixed 1:1 with -20°C neat ethanol, and incubated 1 hour at -20°C to fix. Fixed cells were then washed gently and denatured in 2 M HCl for 20 minutes at room temperature. Following washing and 2 minute's incubation with 0.1 M Tris to neutralize the acid, cells were re-suspended in FACS buffer (0.5% fetal calf serum in PBS) and stained with Phycoerythrin (PE)-conjugated anti-BrdU according to the manufacturer's protocol for flow cytometry analysis on a FACScan instrument.

Apoptosis was measured using a kit (BD Biosciences) containing propidium iodide (PI) and FITC-conjugated annexin V. Cells were stained according to the manufacturer's protocol and the level of apoptosis determined by flow cytometry. Quadrants are: early apoptosis (annexin V⁺/PI^-, lower right) late apoptosis (annexin V⁺/PI⁺, upper right) and non-apoptotic cell death (annexin V^-/PI⁺, upper left). Treatments for the stress panel were: no treatment (control); DMSO as a control for celecoxib; 20 mM celecoxib, brand name Celebrex™ (dissolved in DMSO) [31]; 0.2 mM H₂O₂ [32]; or 1 mg/mL G418 (Pfizer, New York, NY, USA).

Transcriptional analysis using Cancer PathwayFinder RT² profiler arrays (SuperArray, Frederick, MD, USA) was performed according to the manufacturer's protocol. Briefly, total RNA was isolated using an RNeasy extraction kit (Qiagen, Valencia, CA, USA); 1 μg of RNA was reverse transcribed with the cDNA synthesis kit (SuperArray) and cDNA was subjected to real-time PCR using SYBR green to detect product. Arrays were performed independently at least twice for each cell line; all PCR products were checked on agarose gels. Values were obtained for the threshold cycle (C_t) for each gene and normalized using the average of four housekeeping genes on the same array (HPRT1, RPL13A, GAPD, ACTB). C_t values for housekeeping genes and a dilution series of ACTB were monitored for consistency between arrays. Change (ΔC_t) between MUC1 siRNA and control siRNA was found by:

ΔC_t = C_{t(MUC1 siRNA)} - C_{t(control siRNA)}

and fold change by:

Fold change = 2^(-ΔCt)

Values are given as fold change; only genes showing two-fold or greater change were considered. Both luciferase and scrambled siRNA controls were used in BT-20; only genes showing consistent alteration with both controls were included in the results reported here. The scrambled siRNA could be not used in MDA-MB-468 as these cells decrease MUC1 expression in response to this construct.

Two human breast cancer cell lines, MDA-MB-468 and BT-20, were transiently transfected with a pool of four siRNA oligonucleotides directed against the MUC1 mRNA (468.siMUC1 and BT.siMUC1), or a control oligonucleotide directed against luciferase (468.siLuc and BT.siLuc). Both cell lines express high levels of MUC1, making them promising targets for this analysis. Western blots (Figure 1a) show successful knockdown of both the extracellular domain and cytoplasmic tail fragments of MUC1; luciferase siRNA does not substantially change the level of MUC1 compared to parental cells. 468.siMUC1 show a substantial decrease in the amount of MUC1-CT, while BT.siMUC1 show slightly less knockdown of MUC1-CT. Both MDA-MB-468 and BT-20 display a less dramatic decrease of MUC1 extracellular domain compared to MUC1-CT (Figure 1a); this likely represents protein synthesized prior to transfection, and may reflect differences in the turnover rates of the two subunits.

Analysis of the MUC1 extracellular domain by flow cytometry confirms that both cell lines substantially decrease MUC1 expression after siRNA (Figure 1b). By flow cytometry, 468.siMUC1 averaged 75% knockdown of MUC1 compared to 468.siLuc; and BT.siMUC1 averaged 50% knockdown relative to BT.siLuc. These effects could be titrated with the concentration of siRNA, were seen as early as 24 hours post-transfection (data not shown) and lasted to at least 96 h post-transfection (Figure 1b). All experiments were conducted within 48 to 96 hours after siRNA transfection. Similar results were obtained using two independent oligonucleotides designed in our lab (data not shown), designated '882' and '956' for the initial codon recognized by each.

Recent work indicates that MUC1 may affect transcription both directly via interaction with transcription factors and indirectly (for example, through modulating signaling). To study the effects of MUC1 knockdown in breast cancer cell lines, real-time PCR arrays were used to analyze transcription of 84 genes implicated in cancer. Only genes with greater than two-fold change were considered. Three genes (MAP2K1, VEGF, PDGFA) were altered two-fold or more after MUC1 siRNA in both MDA-MB-468 and BT-20 cells (Figure 2); two genes (ITGAV, MMP2) changed only in 468.siMUC1; and five genes (TIMP3, RAF1, JUN, TNF, CDC25A) only in BT.siMUC1. This list represents all genes affected greater than two-fold after MUC1 siRNA, rather than a select group. Three genes whose transcription was changed by less than two-fold are shown, two of which (PDGFB and ITGB1) are listed because they relate closely to genes altered by two-fold (PDGFA and ITGAV). The third, MYC, is included because western blots confirmed a substantial change at the protein level (Figure 3a) that may reflect both transcriptional and post-transcriptional regulation.

Interestingly, transcription of MAP2K1 was decreased in both cell lines after MUC1 siRNA. This gene encodes MEK1, one of the primary regulators of the ERK1/2 MAPK pathway [33], a network that has been linked several times to MUC1 [12, 34‐36]. We examined MEK1 and MEK2 levels by western blot to confirm decreased protein in MUC1 siRNA-treated cells (Figure 3a), and found that not only were total MEK1/2 levels lower in 468.siMUC1 and BT.siMUC1 compared to controls (0.48 and 0.68 relative to siLuc, respectively), but so were the basal amounts of active (phosphorylated) MEK1/2 (pMEK1/2; 0.12 and 0.42 relative to siLuc, respectively). Both 468.siMUC1 and BT.siMUC1 also showed reduced activation of ERK1/2 (dpERK1/2; 0.21 and 0.27 relative to siLuc, respectively), as would be expected with diminished signaling through MEK1/2; total ERK1/2 levels remain unchanged.

As both lines have high levels of EGFR and thus activate the MEK-ERK cascade intensely when stimulated with EGF [37], siRNA-transfected cells were treated with EGF. Notably, MUC1 siRNA impairs this important oncogenic pathway in MDA-MB-468 cells, as 468.siMUC1 display less pMEK1/2 in response to EGF than do 468.siLuc (Figure 3b). Interestingly, EGF treatment of BT-20 cells results in slightly higher pMEK1/2 levels in BT.siMUC1 compared to BT.siLuc. Though this result seems paradoxical in light of decreased MAP2K1 transcription in BT.siMUC1, it likely results from differential functions of Raf isoforms in combination with the increased RAF1 transcription (Figure 2) and protein level (Figure 3a) in these cells. Specifically, B-Raf is thought to be the main activator of MEK under normal conditions; Raf-1 activates MEK in response to stimulus [38]. Thus, it appears that basal pMEK1/2 levels are not greatly affected by Raf-1 overexpression in BT.siMUC1 cells, likely because MEK is regulated primarily by B-Raf under normal growth conditions. In contrast, when the cells are stimulated (EGF), increased Raf-1 levels in BT.siMUC1 leads to heightened pMEK1/2 (Figure 3b).

We next examined whether MUC1 knockdown and its associated transcriptional alterations would affect overall cellular events. As several of the genes shown in Figure 2 are important in regulating proliferation and survival, and because of the recently described role of MUC1 in modulating apoptosis in response to cellular stresses [20, 21, 24], we first analyzed whether MUC1 siRNA would alter apoptosis in these lines. Although there was no change in basal apoptosis in either line (Figure 4a), we observed that the cell lines responded differently when trypsinized for re-plating 24 hours after transfection (Figure 4b). Interestingly, 468.siMUC1 cells show greater apoptosis after trypsinization than do 468.siLuc (49.8% versus 34.0%, respectively), while BT-20 cells from both siRNA treatments display similar levels of apoptosis (around 22%).

To examine whether this phenomenon is specific to trypsin treatment or part of a general stress response involving MUC1, we subjected cells to a panel of stresses and measured cell death. In agreement with the patterns seen with trypsinization, BT.siLuc and BT.siMUC1 respond similarly to all treatments (data not shown), while 468.siMUC1 die more readily than 468.siLuc in response to trypsin, G418, hydrogen peroxide, or celecoxib, a chemotherapeutic that targets the cyclooxygenase-2 (COX-2) pathway (Figure 4c); these data were confirmed with two independent siRNA constructs (data not shown).

Like the MAPK pathway, AKT signaling has been linked to MUC1 in cancer. Although transcription of AKT was not altered in MUC1 siRNA-treated cells, the results of our apoptosis studies prompted us to investigate levels of AKT further. As expected, the total AKT protein level is not greatly changed after MUC1 siRNA in either cell line, though the active form (pAKT) is increased in both 468.siMUC1 and BT.siMUC1 compared to controls (Figure 3a). This result disagrees with MUC1 activation of the AKT pathway in rat 3Y1 cells [18], and may reflect regulation more appropriate to breast cancer cells; this is supported by activation of AKT in response to MUC1 siRNA in other lines [21]. In addition, there is a striking difference in the relative amounts of AKT and pAKT in the two cell lines (Figure 4d). When lysates from both lines are exposed to film for the same length of time (overexposure masks the differences between BT.siLuc and BT.siMUC1 that are apparent in Figure 3a), it is clear that pAKT levels are much higher in BT-20 than in MDA-MB-468, despite lower total AKT expression. This difference in AKT activation between MDA-MB-468 and BT-20 likely contributes to the disparity in their sensitivity to the increased apoptosis expected with loss of MUC1.

As MUC1 is involved in apoptosis, we next analyzed its effects on proliferation. BrdU and [³H]thymidine incorporation were used to analyze proliferation after MUC1 siRNA. 468.siMUC1 cells show a significant decrease in [³H]thymidine incorporation compared to 468.siLuc, while intriguingly, BT.siMUC1 cells show a significant increase in proliferation (Figure 5a). Growth curves mirror these results, as do experiments with the two independent MUC1 siRNA oligonucleotides (data not shown). Note that these assays require trypsinizing cells 24 hours post-transfection; therefore, the results in the MDA-MB-468 line could stem from the changes in apoptosis described in the previous section, rather than a true effect on proliferation. To control for this, we incubated non-trypsinized, siRNA-transfected cells at similar confluence with BrdU to measure incorporation. The 'clumped' profile of cells (contrast to Figure 4b) is likely a result of the acid denaturation (recommended by the antibody manufacturer), as it occurs uniformly in these experiments. BrdU incorporation (Figure 5b) confirms that the [³H]thymidine results are not solely due to alterations in apoptosis, as 468.siMUC1 cells incorporate less BrdU than 468.siLuc; once again, BT.siMUC1 cells show increased proliferation over BT.siLuc.

Given the role of MUC1 in adhesion, we examined whether MUC1 siRNA affects cellular invasion. In transwell assays, BT-20 cells invaded poorly, regardless of the siRNA used (data not shown). However, MDA-MB-468 cells invade more readily, and were analyzed on a panel of three different extracellular matrix proteins. Interestingly, 468.siMUC1 cells display somewhat decreased invasion on collagen IV, laminin, and fibronectin matrices, and on a no-matrix control (Figure 5c), which is in agreement with the trend towards decreased metastasis observed in Muc1^-/- × MMTV-PyV MT mice [8].

To determine if the above effects are specific to MUC1, we created stable transfectants of the MDA-MB-468 line using empty vector (468.Neo) or a full-length MUC1 construct (468.MUC1Δ8) that is resistant to one of the independent MUC1-directed oligonucleotides ('882'). These cells were maintained in G418-containing medium to retain transgene selection. As expected, 468.MUC1Δ8 cells show higher levels of both the MUC1 extracellular domain and the MUC1-CT than do 468.Neo (Figure 6a). Note that 468.Neo have MUC1 expression comparable to parental MDA-MB-468; the exposures in Figure 6a are lighter than those in Figure 1a, in order to clearly show the relative levels of MUC1 in the stable transfectants. After MUC1 siRNA, 468.MUC1Δ8 lose some MUC1 (likely endogenous protein, which is not siRNA-resistant) but retain high-level expression, while 468.Neo show a decrease in MUC1 levels similar to parental 468.siMUC1 cells (Figures 6a,b). The difference in the amount of MUC1 knockdown between 468.Neo and 468.MUC1Δ8 is highlighted by the purple shading in Figure 6b.

BrdU incorporation (Figure 6c) indicates that 468.Neo show decreased nucleotide incorporation after MUC1 siRNA compared to control (3.3% versus 25.0%, respectively); this is not seen in 468.MUC1Δ8 cells, which show similar levels of BrdU incorporation regardless of the siRNA used (21.5% for luciferase, 23.9% for MUC1). 468.Neo cells display a more dramatic decrease in BrdU incorporation after MUC1 siRNA than what is seen in parental 468.siMUC1 cells, which may reflect the additional stress of being maintained in G418-containing medium. Similarly, analysis of apoptosis in trypsinized cells indicates that the increased apoptosis seen in parental 468.siMUC1 cells is also present in the 468.Neo line after MUC1 siRNA (Figure 6d; 43.6% in control versus 59.6% in MUC1 siRNA). However, in 468.MUC1Δ8 cells, the level of apoptosis after luciferase siRNA (34.1%) is lower than that in 468.Neo cells; MUC1 siRNA increases the amount of apoptosis slightly (42.8%), restoring it to a level similar to that seen in luciferase siRNA-treated 468.Neo cells. Together, these studies suggest that the above-described results are specific to MUC1, as stable transfection of an siRNA-resistant MUC1 rescues the phenotype seen in 468.siMUC1 cells.