BT474 wild-type, SKBR3 wild-type and MCF7 wild-type cells were obtained from the American Type Culture Collection (ATCC). They were maintained in a monolayer culture in DMEM/F12 (Dulbecco’s modified Eagle’s medium) with 10% fetal bovine serum, 2.5% L-glutamine and 0.5% penicillin/streptomycin. The MCF7-HER2 (MCF7 cells transfected with HER2) cell line was a generous gift of Dr C. Kent Osborne (Baylor College of Medicine, Houston, TX). MCF7-HER2 cells were maintained in a monolayer culture in DMEM 1X with 10% fetal bovine serum, 2.5% L-glutamine, 0.5% penicillin/streptomycin, and G418 (400 μg/ml).

Matrigel (BD, Cambridge, MA), 200 μl was spread as a thick layer on a 24-well plate and allowed to polymerize at 37°C for 15 min, then 2×10⁴ cells grown as monolayer were trypsinized to single cells and plated on top of the pre-coated Matrigel. Plates were incubated at 37°C to allow cells to fully settle down before media was replaced with appropriate growth media containing 5% Matrigel. Cells were grown for 15 days; fresh growth media with Matrigel was replenished every two days. Images of representative fields were taken.

The HER2 short hairpin RNA (shRNA) and control shRNA were purchased from the Ori Gene (Rockville, MD). The transfection was performed by using Lipofectamine 2000 (Invitrogen) reagent following the manufacturer’s instructions. The STAT3 shRNA was a kind gift of Dr Hyung-Gyoo Kang at the Children’s Hospital in Los Angeles. The HER2 and STAT3 shRNA knockdown expressions after shRNA transfection were determined by a Western blot analysis at 48–72 h of transfection.

Monolayer cultures of respective cell lines at an 80–90% confluence were lysed using 100 μl of RIPA (Thomas Scientific Co.) buffer. Tris-glycine (Bio-Rad) gels were loaded with 50–100 μg of lysates. After running gel electrophoresis, the gel was transferred to a nitrocellulose membrane for 2 h. The membrane was blocked for 1 h in 5% BSA or 5% skim milk at 4°C. The membrane was then washed 3 times with 1X TTBS and incubated overnight with the primary antibody at 4°C. Primary antibodies of Oct-4, Sox-2, STAT3, pSTAT3, E-cadherin, vimentin, slug and β-actin were purchased from Cell Signaling Technology (Danvers, MA).

Mouse fibroblasts (NIH-3T3) were used as a chemo-attractant, and grown in a 24-well plate in 2 ml of the same media as MCF7-HER2 cells. MCF7 WT and MCF7-HER2 experimental cells were synchronized to an equal number (125,000 cells) in a 6-well plate and were serum starved overnight. Boyden chambers were prepared with 25 μl of 1:6 diluted Matrigel and allowed to incubate for about 2 h to solidify. After cell synchronization, invasion was allowed to occur for 40 h. The cells were then fixed with 0.5% glutaraldehyde and stained with 5% toluidine blue for cell counting. Three different microscope fields of ×40 were used to quantify the invasion statistics when counting cells.

Approximately 500,000 cells of MCF7 WT or MCF7-HER2 cells were washed with 1X PBS, trypsinized and then transferred to a 15-ml aliquot tube. Cell suspensions were centrifuged down, re-suspended in 2 ml 1X PBS and then divided into two tubes 0.5 ml each for staining. One tube was used as an unstained control and the other one as stained with 10 μl CD44 antibody (FITC Green; BD Biotech). The tubes were vortexed briefly and incubated at room temperature for 15 min in the dark. Each tube was then washed with 3.5 ml 1X PBS and then centrifuged for 6 min. After aspirating the supernatant, the cells were re-suspended in 3 ml 1X PBS and subjected to the FACS profiling at UCLA FACS Core Laboratory.

All inhibitors were prepared as 200 mmol/l DMSO stocks for in vitro use. MCF7-HER2 cells that were at least 60–70% confluent were washed with 1X PBS and treated with 5 μM of Stattic (STAT3 Inhibitor V; Santa Cruz Biotechnology) dissolved in 4 ml of 1X DMEM media. Untreated Stattic MCF7-HER2 cells were used as a control. Treatment lasted 24 h and cells were harvested for further analysis.

MCF7-HER2 cells were cultured and counted and each test was conducted in triplicate. After mixing well, media/cell solution was diluted to 4,000 cells per 100 μl. Cells were then plated in a 96-well plate: 6 wells for each treatment dosage, 4,000 cells per well (100 μl). Cells were treated with Stattic (0 μM control, 1, 5 and 10 μM). Treatments lasted 48 h and then 50 μl of MTT buffer reagent was added into each well. After incubation for 4 h at 37°C, 100 μl of DMSO was added and gently agitated for 15 min. The plate was then read at 560 nm absorbance.

The initial set up was similar to the one described in MTT assay dose response curve. This time, however, 10,000 cells were plated per 100 μl in 8 wells per treatment. Cells were allowed to grow until about 50% confluence, and then treated with the following dosage treatments: untreated control, Stattic (5 μM), Herceptin (10 μg/ml, a gift of Genentech) and Stattic (5 μM) + Herceptin (10 μg/ml) combination. Treatments lasted 72 h and cell growth was monitored.

To monitor the protein-protein interactions among HER2, ER and STAT3 proteins, ER specific monoclonal antibody (Santa Cruz Biotechnology, sc-7207) was added to the cancer cell lysates and incubated overnight. The immune-pellets were spun at 3,000 rpm for 5 min and washed three times with RIPA buffer (Thomas Scientific Co). The pellets were then ran on the PAGE gel and immunoblotted for HER2 and STAT3 proteins, respectively.

Human breast cancer tissue micro-array slides (BRC961) were purchased from US Biomax Inc. (Ijamsville, MD, USA). Breast cancer tissue microarray slides were stained using standard IHC methods with an IHC-validated phospho-specific STAT3 antibody (Tyr705; Cell Signaling Technology, D3A7).

For determining statistical significance in all quantifications, Student’s t-test was used; the data are presented as the mean ± SD, and considered significant at p<0.05.

Initially we tested the effect of STAT3 in HER2-positive breast cancer. We wished to determine whether the co-expression of HER2 and ER induced STAT3 phosphorylation, and whether pSTAT3 promotes stem-like cell phenotype in the breast cancer model. To this end, the basal expression of STAT3 and stem cell markers were examined in various human breast cancer cell lines by western blot analysis and real-time PCR. STAT3 was phosphorylated in MCF7-HER2 cells, but not in MCF7 wild-type (Fig. 1A). In addition, STAT3 was phosphorylated in BT474 as well as at very low level in SKBR3. MCF7 wild-type lacks HER2 amplification and SKBR3 lacks ER. This is consistent with previous data in which MCF7 WT did not typically show phosphorylation of STAT3 (21,22). Moreover, we found that the stem cell markers, Oct-4 and Sox-2, were expressed in MCF7-HER2 and BT474 cells, but not in MCF7 wild-type and SKBR3 (Fig. 1A). Real-time PCR analyses confirmed the upregulation of stem cell markers in HER2-overexpressing, ER-positive cancer cells (Fig. 1B). Our results support the hypothesis that HER2 over-expression and ER positivity promote STAT3 phosphorylation and induces the stem cell-like phenotype.

Transcription factor slug, which has been implicated as a driver for the epithelial-mesenchymal transition (EMT) was upregulated in MCF7-HER2 cells (Fig. 1B), leading us to the hypothesis that, HER2 induced stem cell marker expression and slug upregulation promoted the EMT phenotype in MCF7 cells. Real-time PCR analyses showed the gene expression pattern matching with the mesenchymal microenvironment, showing significant upregulation in vimentin, slug and concurrent downregulation of E-cadherin in MCF7-HER2 compared with control MCF7 wild-type (Fig. 1C). In line with the PCR data, western blot analyses revealed that epithelial marker E-cadherin was downregulated, while mesenchymal maker vimentin was upregulated in HER2 transfected MCF7 cells (Fig. 1D). These results indicate that the sequential activations elicited by HER2 amplification converge into the HER2-pSTAT3-stem cell markers - early EMT characteristics in signal pathways.

To uncover Herceptin resistance mechanism in HER2-overexpressing cancer cells, HER2/ER-positive cancer cells were treated with STAT3 inhibitor and examined for cancer stem cell phenotype. Stattic is a non-peptidic small molecule that inhibits STAT3 activation by selectively inhibiting dimerization and nuclear translocation of STAT3 (23). For the purpose of proving the effect of Stattic on inhibiting STAT3 and to investigate its effects on the HER2-ER-STAT3, we treated cells with 5 μM Stattic for 24 h and monitored its effects on EMT and stem cell marker expression.

To begin with, dose response to Stattic was monitored in MCF7-HER2 cells (Fig. 2A). Stattic treatment of 5 μM resulted in 28% of cell survival. Western blots of the MCF7-HER2 cells treated with 5 μM of Stattic illustrated inhibition of STAT3 activation, as expected, but also showed significantly diminished Sox-2, Oct-4, and slug expression to the point of abrogation (Fig. 2B). While total STAT3 was almost intact with the Stattic treatment, pSTAT3 expression was also abolished simultaneously. For EMT marker expression, we monitored expression of E-cadherin, vimentin and slug protein. In agreement with cancer stem cell markers, the expression levels of EMT markers, vimentin and slug, were clearly decreased upon Stattic treatment (Fig. 2C). Our data suggest that STAT3 activation is responsible for the stem cell marker expression in HER2-overexpressing breast cancer cells.

Given Stattic’s effect on reducing stem cell-like characteristics, we designed a combined treatment regimen consisting of Stattic, STAT3 inhibitor, and Herceptin, HER2 monoclonal antibody. The goal was to demonstrate that HER2-positive, ER-positive cancer cells were more sensitive to Herceptin treatment when also exposed to Stattic, due to the inhibition of STAT3. This would also help prove the presence of cross-talk between HER2/ER-STAT3 pathways through the synergistic effect of the combined treatment. Thus, we hypothesized that Stattic could help delay or even reverse incurred resistance associated with HER2 overexpression.

Based upon the dose response curve, it was estimated that 5 μM would be the optimal dosage for the combined treatment. In a second MTT to determine the effectiveness of Stattic and Herceptin combined treatment, a control group, a solitary 10 μg/ml Herceptin treatment group, a solitary 5 μM Stattic treatment group and a combined 10 μg/ml Herceptin with 5 μM Stattic treatment group were compared to determine which treatment would provide the highest apoptosis rate. It was observed that while all treatments lead to significant reduction in the cell population based upon the control population, the Herceptin and Stattic combination treatment showed the highest rate of apoptosis (Fig. 2D). The combination treatment displayed significant effect (cell survival 11%) when compared to the solitary Herceptin and Stattic treatments.

STAT3 specific inhibitor Stattic treatment convincingly abolished the stem cell phenotype. To test whether STAT3 drives the stem cell phenotype in breast cancer cells, we knocked down STAT3 gene by shRNAs and examined the CD44⁺ subpopulation by fluorescence activated cell sorting (FACS). Targeted STAT3 knockdown was confirmed by western blot analysis and subsequently applied to FACS analysis (Fig. 3A). When STAT3 expression was knocked down with two independent shRNAs, the CD44(+) subpopulation has clearly reduced from 61.21% (CD44⁺ of the control RNA) to 37.22% and 38.23%, respectively (Fig. 3B). In agreement with FACS data, western blot analyses of STAT3 knockdown cells also showed downregulation of stem cell marker expressions (Fig. 3C). These results suggest that STAT3 plays a clear role in inducing the expression of stem cell markers Oct-4, Sox-2 and CD44.

Not only the expression level of stem cell markers of Oct-4 and Sox-2 was increased, but CD44(+) cell population was also increased in HER2-overexpressing, ER-positive breast cancer cells. To establish the potential link between HER2 amplifi cation and the stem cell phenotype, we knocked down the HER2 gene by shRNAs and examined CD44(+) FACS profile (Fig. 3D). The CD44(+) subcellular population was reduced from 60.77% to a range of 43.17-47.16%, when HER2 expression was knocked down by two independent HER2 shRNAs (Fig. 3E). This confirms that HER2 is also a driver for induction of the cancer stem cell-like phenotype in MCF7 cells.

Cancer stem cell characteristics are associated with enhanced cell invasiveness. To determine whether forced HER2 overexpression leads to STAT3 activation and functionally enhances the invasiveness of MCF7 cells, an in vitro Boyden chamber invasion assay was implemented. As expected, MCF-7 wild-type control cells showed minimal invasion (8% of invasion population) through the Matrigel, even in the presence of NIH3T3 mouse fibroblasts as a chemoattractant (Fig. 4A). In contrast, MCF7-HER2 cells were 4 times more invasive (31%, Fig. 4B), as they averaged 23 more invaded cells per high powered field (HPF, ×40). Although MCF7 is typically known for being a non-invasive cell line, HER2 overexpression leading to STAT3 activation resulted in an increase of the invasiveness of the cell line.

Formation of spheres by tumor cells is often directly correlated with cancer stem cell phenotype. Having considered Oct-4, Sox-2 and CD44 expression data, we wished to determine the tumorsphere formation capacity with HER2-overexpressing breast cancer. Approximately 20,000 cells were seeded on each Matrigel 3D well and tumorsphere formation was observed up to 15 days. While MCF7 WT did not form any spheres, HER2-overexpressing MCF7 cells formed stable tumorspheres at day 6 (Fig. 5A and B). To verify that the tumorsphere forming capacity is due to STAT3, we repeated the 3D culture with the targeted STAT3 knockdown cells. As shown in Fig. 5C, the figure, STAT3 knockdown cells did not form any spheres. The tumorspheres were counted for each well and presented as a graph (Fig. 5D). These results are consistent with the STAT3 knockdown effects on expression of stem cell markers Oct-4, Sox-2 and CD44. STAT3 is necessary for tumorsphere formation in HER2-overexpressing cancer.

Recently, we found that STAT3 was upregulated by 3.64-fold in MCF7 cells transfected with HER2 compared to a MCF7 wild-type (12). However, the relationship between STAT3 in primary tumors and HER2 overexpression was not described. Thus, we hypothesized that HER2-positive, ER-positive breast cancer cells expressed phosphorylated STAT3. To test this, we have examined tissue microarrays of 71 primary breast tumors with previously confirmed ER, PR and HER2 expressions. We observed the phosphorylated STAT3 expression from these breast tumors by immunohistochemistry. We found that pSTAT3 was expressed only in HER2(+) and ER(+) primary breast tumors (Fig. 6A). As shown in the figure, 9 out of 17 HER2⁺/ER⁺ tumors expressed pSTAT3 and 6 out of 9 HER2⁺⁺ or HER2⁺⁺⁺/ER⁺ tumors showed pSTAT3 expression (Table I), moreover, 2 out of 4 showed pSTAT3 in ER/PR/HER2 triple negative patient tissues. This pSTAT3 activation is possibly through other STAT3 activation signaling axis in triple negative breast tumors such as Src or non-RTK phosphorylation mechanisms.

We also determined whether there are physical connections between HER2, ER and STAT3. Our next hypo thesis was that ER is bound to both HER2 receptor and STAT3 to transduce the phosphorylation signals from HER2 to STAT3. We used the MCF7-HER2 cell line for our model system of HER2(+)/ER(+) breast cancer. We immunoprecipitated the cell lysates with antibody against ERα and subsequently immuno blotted for HER2 and STAT3 in MCF7 wild-type, MCF7-HER2, MCF7-HER2 STAT3 knockdown and MCF7-HER2 HER2 knockdown cells (Fig. 6B). ERα was bound to both HER2 and STAT3 simultaneously in MCF7-HER2 cancer cells. When STAT3 is knocked down, HER2 is still bound to ER, when HER2 is knocked down, ER is still bound to STAT3. The tissue microarray and immunoprecipitation data suggest, for the first time, the existence of HER2-ER-STAT3 signaling axis in HER2-positive breast cancers.