Up-regulation of miR-21 Mediates Resistance to Trastuzumab Therapy for Breast Cancer*

Trastuzumab resistance emerges to be a major issue in anti-human epidermal growth factor receptor 2 (HER2) therapy for breast cancers. Here, we demonstrated that miR-21 expression was up-regulated and its function was elevated in HER2+ BT474, SKBR3, and MDA-MB-453 breast cancer cells that are induced to acquire trastuzumab resistance by long-term exposure to the antibody, whereas protein expression of the PTEN gene, a miR-21 target, was reduced. Blocking the action of miR-21 with antisense oligonucleotides re-sensitized the resistant cells to the therapeutic activities of trastuzumab by inducing growth arrest, proliferation inhibition, and G1-S cell cycle checking in the presence of the antibody. Ectopic expression of miR-21 in HER2+ breast cancer cells confers resistance to trastuzumab. Rescuing PTEN expression with a p3XFLAG-PTEN-mut construct with deleted miR-21 targeting sequence at its 3′ UTR restored the growth inhibition of trastuzumab in the resistant cells by inducing PTEN activation and AKT inhibition. In vivo, administering miR-21 antisense oligonucleotides restored trastuzumab sensitivity in the resistant breast cancer xenografts by inducing PTEN expression, whereas injection of miR-21 mimics conferred trastuzumab resistant in the sensitive breast tumors via PTEN silence. Up-regulatin of miR-21 in tumor biopsies obtained from patients receiving pre-operative trastuzumab therapy was associated with poor trastuzumab response. Therefore, miR-21 overexpression contributes to trastuzumab resistance in HER2+ breast cancers and antagonizing miR-21 demonstrates therapeutic potential by sensitizing the malignancy to anti-HER2 treatment.

Trastuzumab resistance emerges to be a major issue in anti-human epidermal growth factor receptor 2 (HER2) therapy for breast cancers. Here, we demonstrated that miR-21 expression was upregulated and its function was elevated in HER2 ؉ BT474, SKBR3, and MDA-MB-453 breast cancer cells that are induced to acquire trastuzumab resistance by long-term exposure to the antibody, whereas protein expression of the PTEN gene, a miR-21 target, was reduced. Blocking the action of miR-21 with antisense oligonucleotides re-sensitized the resistant cells to the therapeutic activities of trastuzumab by inducing growth arrest, proliferation inhibition, and G 1 -S cell cycle checking in the presence of the antibody.

Ectopic expression of miR-21 in HER2 ؉ breast cancer cells confers resistance to trastuzumab. Rescuing PTEN expression with a p3XFLAG-PTEN-mut construct with deleted miR-21 targeting sequence at its 3 UTR restored the growth inhibition of trastuzumab in the resistant cells by inducing PTEN activation and AKT inhibition. In vivo, administering miR-21 antisense oligonucleotides restored trastuzumab sensitivity in the resistant breast cancer xenografts by inducing PTEN expression, whereas injection of miR-21 mimics conferred trastuzumab resistant in the sensitive breast tumors via PTEN silence. Up-regulatin of miR-21 in tumor
biopsies obtained from patients receiving pre-operative trastuzumab therapy was associated with poor trastuzumab response. Therefore, miR-21 overexpression contributes to trastuzumab resistance in HER2 ؉ breast cancers and antagonizing miR-21 demonstrates therapeutic potential by sensitizing the malignancy to anti-HER2 treatment.
Overexpression of human epidermal growth factor receptor 2 (HER2/NEU/c-ERBB2), 4 a member of the epidermal growth factor receptor (EGFR) family of 185 kDa, is found in 20 -35% of human breast cancers (1). Amplification of HER2 is linked to aggressive tumor behavior and poor clinical outcome with shorter disease-free intervals and overall survival in patients with early and advanced breast cancers (2). Tremendous efforts have been made to develop HER2-targeting cancer therapies, and a most successful strategy is the recombinant humanized anti-HER2 monoclonal antibody trastuzumab (Herceptin) that specifically binds to the extracellular domain of HER2 and blocks its function. Clinical application of trastuzumab in adjuvant and metastatic settings has been shown to prolong the survival of patients with HER2 ϩ breast cancers (3). However, the response rate to trastuzumab monotherapy is less than 35%, whereas ϳ60% of patients with HER2 ϩ cancers on regimens combining trastuzumab with microtubule stabilizing drugs do not respond to treatment (4). Moreover, most patients who achieve an initial response develop resistance to trastuzumab within 1 year (5). Therefore, identifying the mechanisms responsible for trastuzumab resistance is important for the development of new therapeutic strategies.
A number of mechanisms have been suggested for trastuzumab resistance, including dysregulation of downstream signaling pathways and compensated signaling by other EGF family members or through alternative pathways (6). Among them, reduced PTEN expression is a strong indicator to predict trastuzumab resistance in breast cancer patients. Further mechanistic study showed that loss of PTEN function in HER2 ϩ cancer cells leads to trastuzumab resistance by enhancing downstream PI3K/AKT phosphorylation and thus preventing trastuzumab-mediated growth arrest (7). However, how PTEN expression is silenced in trastuzumab resistance remains illusive.
In the present study, we screened for miRNAs that were differentially expressed in the trastuzumab-resistant breast cancer cells, and identified that miR-21 was up-regulated among the reported PTEN-targeting miRNAs. Antagonizing miR-21 expression by specific antisense oligonucleotides (ASOs) increased PTEN expression, and restored sensitivity to trastuzumab in vitro and in vivo. Vice versa, ectopic expression of miR-21 in HER2 ϩ breast cancer cells reduced PTEN protein expression, and resulted in trastuzumab resistance in vitro and in vivo.

EXPERIMENTAL PROCEDURES
Patients and Tissue Samples-Primary ductal carcinomas of the breast with HER2 gene amplification were obtained from 32 female breast cancer patients before and after pre-operative neoadjuvant Herceptin therapy in the breast tumor center, Sun-Yat-Sen Memorial Hospital, Sun-Yat-Sen University, from January 2008 to December 2009. All patients underwent preoperative neoadjuvant therapy with 3-4 cycles of the tri-weekly TCH regimen (Taxotere, 75 mg/m 2 , Carboplatin, AUC 6 mg/ml/min, and Herceptin, 8 mg/kg loading dose followed by 6 mg/kg every 3 weeks) according to NCCN guideline.
Breast tumor samples were obtained via core-needle biopsy prior to Herceptin therapy and from surgery after treatment. The collected tumor tissues were snap-frozen in liquid nitrogen for miRNA assay. In addition, the remaining tissues were embedded for studies of histology, immunohistochemistry, and in situ hybridization. All samples were collected with informed consent according to the internal review and ethics boards of the hospital.
In Situ Hybridization-miR-21 expression was examined by in situ hybridization on the formalin-fixed and paraffin-embedded sections of breast cancers. Briefly, after dewaxing and rehydration, samples were digested with proteinase K, fixed again in 4% paraformaldehyde, hybridized with the 5Ј digoxin-labeled LNA TM -modified miR-21 probe (Exiqon) at 49.5°C overnight, and then incubated overnight at 4°C with anti-digoxin monoclonal antibody (Roche Applied Science). Followed by nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate staining solution in the dark, the sections were mounted and observed.
Immunohistochemistry-PTEN expression was examined by immunohistochemistry on paraffin-embedded tissue sections. Briefly, rabbit anti-PTEN polyclonal antibody (1:125, Cell Signaling Technology, Danvers, MA) was used as primary antibody for overnight incubation at 4°C. The sections were subsequently treated with goat anti-rabbit secondary antibody, followed by further incubation with streptavidin-horseradish peroxidase complex. Diaminobenzidine (Dako, Carpinteria, CA) was used as a chromogen and sections were lightly counterstained with hematoxylin. The percentage of positively staining tumor cells was calculated per field of view, with at least 20 view fields per section evaluated at ϫ400 magnification.
Cell Cultures and Treatment-SKBR3, MDA-MB-453, and BT474 cell lines were obtained from the American Type Culture Collection. Cells were maintained in RPMI 1640 with 10% fetal bovine serum. Trastuzumab (Herceptin) was obtained from Genentech and dissolved in sterile water. Stable cell lines were generated by transfection of p3XFLAG-PTEN-mut or p3XFLAG constructs followed by selection in puromycin. Transfection of the cells with miRNA mimics or miRNA ASOs (Genepharma) was performed using Lipofectamine (Invitrogen) as previously described (18).
miRNA Microarray Analysis-miRNA microarray analysis was done in the parental and resistant BT474 lines. Briefly, total RNA is harvested using TRIzol (Invitrogen) and the RNeasy mini kit (Qiagen) according to the manufacturer's instructions. The samples are labeled using the miRCURY TM Hy3 TM /Hy5 TM Power labeling kit (Exiqon) and hybridized on the miRCURY LNA Array (Exiqon, version 11.0). Scanning was performed with the Axon GenePix 4000B microarray scanner. GenePix pro version 6.0 was used to read the raw intensity of the image. Unsupervised hierarchical clustering was performed on the miRNA expression profiling.
Real-time PCR Assay-cDNA was obtained by reverse transcription of total RNA using a TaqMan Reverse Transcription Kit (Applied Biosystems Inc.). Standard curves were generated and the relative amount of target miRNAs was normalized to U6 snRNA. Specially, the absolute copy number of mature miRNAs was determined by real-time quantitative reverse transcription PCR (qRT-PCR) using TaqMan Assays-on-Demand primer and probe sets along with TaqMan Universal PCR master mix (Applied Biosystems Inc.) for cDNA amplification. Amplification and analysis were performed on the ABI 7900 sequence detection system. Copies per cell were determined from total nanograms of RNA using an estimated 15 pg of total RNA per cell.
Northern Blot Analysis-An aliquot(10 g) of total RNA was subjected to denaturing (7 mol/liter of urea) polyacrylamide (15% acrylamide) gel electrophoresis, transferred to Zetaprobe GT membrane (Bio-Rad), immobilized to the membrane by ultraviolet cross-linking, and hybridized overnight at 42°C to 32 P-labeled deoxyoligonucleotide antisense to miR-21. The blots were reprobed with antisense oligo specific for U6 snRNA to show comparable RNA loading in each lane.
Luciferase Reporter Assay-To evaluate the function of miR-21, the 3Ј UTR of PTEN with a miR-21 targeting sequence was cloned into a pMIR-REPORT luciferase reporter vector (Ambion). The assay was conducted as described (18). Briefly, the reporter vector plasmid was transfected using Lipofectamine 2000. To correct transfection efficiency, a luciferase reporter vector without the miR-21 target was transfected in parallel. Luciferase activities in the cells were assayed using a luciferase assay kit (Promega, Madison, WI), and miR-21 function was expressed as percentage of the luciferase activity of the reporter vector with miR-21 target sequence over one without the miR-21 target sequence.
Western Blotting-Protein extracts were resolved through 10% SDS-PAGE, transferred to nitrocellulose membranes,
Measurements of Cell Viability, Proliferation, and Cell Cycle-Cell viability was measured using the [ 3 H]thymidine incorporation assay. Briefly, [ 3 H]thymidine (1 Ci) was added for 6 h to 2 ϫ 10 4 cells in octuplicate microtiter wells before harvesting and analysis by scintillation counting using a Top Count microplate reader (Packard). For colony-forming assay, cells were harvested and resuspended in medium supplemented with agar at a final concentration of 0.35%. Subsequently, 1 ml of the resulting cell suspension was plated in 12-well plates covered with medium supplemented with 0.6% agar, and cultured at 37°C with 5% CO 2 . The percentage of cells that formed into a spherical clone was calculated. For cell cycle analysis, cells were trypsinized, pelleted, and then resuspended in propidium iodide solution (Sigma), including 50 g/ml of propidium iodide, 0.1 mg/ml of RNase A, and 0.05% Triton X-100, for a 40-min incubation at 37°C and analyzed by flow cytometry using a FACScalibur instrument with CellQuest Software (BD Biosciences). Additionally, S phase cells were detected using 5-bromo-2Ј-deoxyuridine (BrdU) incorporation. Briefly, BrdU (Upstate) was added to the medium at 10 M for 8-or 24-h incubations. BrdU incorporation in cancer cells was evaluated by immunostaining with a phycoerythrin-conjugated anti-BrdU antibody (Upstate, Temecula, CA) and analyzed by flow cytometry analysis.
Animal Experiment-Athymic nude female mice, 3-4 weeks old, were subcutaneously implanted with 1.5 mg of 60-day release 17␤-estradiol pellets (Innovative Research, Sarasota, FL). Trastuzumab-resistant or parental BT474 cells (1 ϫ 10 7 ) in 100 l of PBS per 100 l of growth factor-reduced Matrigel (BD Biosciences) were injected into the mammary fat pads of the mice. When the xenografts reached Ϸ150 mm 3 , miR-21 ASO or control Lin4 ASO (20 g/injection, twice a week) was administered by multiple intratumor injection 72 h prior to intravenous injection of trastuzumab or control IgG (10 mg/kg, twice a week). Tumor volume was followed as volume ϭ width 2 ϫ length/2.
Statistics-The in vitro data were depicted as mean Ϯ S.D. of three independent experiments performed in triplicate or in penultimate. Statistical analysis was performed by one-way analysis of variance and comparisons among groups were performed by independent sample t test or Bonferroni's multiple comparison t test. To measure the association between variables, Spearman order correlations were run.

Trastuzumab-resistant Breast Cancer Cells
Overexpress miR-21-To obtain breast cancer cells with trastuzumab resistance, we cultured HER2 ϩ cell lines, SKBR3, MDA-MB-453, and BT474, in the presence of low-dose trastuzumab (5 g/ml) continuously for 6 months as reported elsewhere (19). The resulting cell lines grow at similar rates in the presence or absence of trastuzumab at 10 g/ml for 3 days (data not shown). We next examined the micro-RNA expression profile in the parental and resistant BT474 lines. Using a cutoff of Ͼ2-fold changes in miRNA expression, we identified that a set of 14 miRNAs were reduced and another set of seven miRNAs were up-regulated in the resistant line as compared with the parental one (supplemental Table S1). Among the reported PTEN-targeting miRNAs, only miR-21 and miR-141 were up-regulated in trastuzumab-resistant BT474 cells (supplemental Table S1). However, the levels of miR-216a, miR-217, miR-221, and miR-222 were comparable in the parental and resistant lines, and the rest of the PTEN-targeting miRNAs, including miR-17-92, miR-19a, miR-26a, and miR-494, were not detected in both lines (data not shown).
PTEN activation is involved in the anti-tumor mechanism of trastuzumab and its deficiency in tumor cells predicts primary resistance to trastuzumab (7). Herein, we examined PTEN expression in breast tumor lines with acquired trastuzumab resistance. Although PTEN mRNA remained unchanged in trastuzumab-resistant BT474, SKBR3, and MDA-MB-453 cells as compared with their parental lines (supplemental Fig. S2), its protein levels were significantly reduced (Fig. 2C). Furthermore, transfecting the trastuzumab-resistant breast cancer lines with a miR-21 ASO substantially increased PTEN protein, whereas miR-21 mimics significantly reduced the protein expression of PTEN in the parental lines (Fig.
Blockade of miR-21 Retrieves Trastuzumab Sensitivity-To investigate whether miR-21 overexpression may contribute to trastuzumab resistance in HER2 ϩ breast cancer cells, we inhibited miR-21 expression with ASOs. The tumor cells were treated with trastuzumab at increasing concentrations for 3 days, and growth inhibition induced by the drug was measured by [ 3 H]thymidine incorporation assay. Transfecting the resistant BT474, SKBR3, and MDA-MB-453 lines with miR-21 ASOs, but not Lin4 ASOs or mock transfection, markedly sensitized the tumor cells to trastuzumab, as the IC 50 decreased by roughly 3.5-5.3-fold, respectively (Fig. 3A). However, without trastuzumab treatment, transfection of miR-21 ASOs alone only slightly reduced the [ 3 H]thymidine incorporation in the resistant lines as compared with mock transfection (p Ͻ 0.05, supplemental Fig. S3). We next tested whether miR-21 ASOs may inhibit anchorage-independent growth of the resistant breast tumor cells in the presence of trastuzumab. The cells were treated with 10 g/ml of trastuzumab for 3 days. As shown by the quantifications of soft agar assay, transfection with miR-21 ASO resulted in 9.5-(p Ͻ 0.001), 7.0-(p Ͻ 0.001), and 4.5-(p Ͻ 0.001)-fold reduction in the number of colonies formed by the resistant BT474, SKBR3, and MDA-MB-453 lines, respectively (Fig. 3B). However, without trastuzumab, miR-21 ASOs alone reduced colony formation only by ϳ3.58 Ϯ 1.72% (p Ͻ 0.05, Fig. 3B). Therefore, blocking miR-21 with ASOs enhances trastuzumab sensitivity in the HER2 overexpressing breast cancer cells with acquired resistance to the drug.
We next asked whether miR-21 inhibition may reverse the refractoriness of the tumor cells to the anti-proliferative effect of trastuzumab. Under treatment with trastuzumab, the proliferation of breast tumor cells was measured by their ability to incorporate BrdU, and cell cycle profiles were examined by flow cytometric analysis for propidium iodide staining. In the presence of trastuzumab, the resistant BT474, SKBR3, and MDA-MB-453 breast cancer cells transfected with miR-21 ASO, instead of Lin4 ASO or mock transfection, showed a substantial (Ϸ24%) reduction in BrdU staining (p Ͻ 0.001, Fig. 3C). However, miR-21 ASO alone without trastuzumab results in a mild decrease in BrdU incorporation in the resistant lines (p Ͻ 0.05, Fig. 3C). Consistently, transfecting the resistant lines with miR-21 ASO markedly increased the percentage of G 1 population in the presence of trastuzumab (p Ͻ 0.001), but slightly affected the magnitude of G 1 -S arrest without trastuzumab (p Ͻ 0.05, Fig. 3D). These data suggested that miR-21 suppression resensitizes the resistant breast cancer cells to the antiproliferative and cell-cycle blocking effect of trastuzumab.
Ectopic miR-21 Expression Confers Trastuzumab Resistance on Breast Cancer Cells-Because our data have shown that miR-21 inhibition in trastuzumab-resistant breast cancer lines retrieves their sensitivity, we next investigated whether enforced miR-21 expression in the sensitive HER2-overexpressing cancer cells may lead to trastuzumab resistance. Contrary to miR-21 suppression in trastuzumab-resistant cells with HER2 overexpression, transfecting the parental BT474, SKBR3, or MDA-MB-453 cells with miR-21 mimics, but not with an irrelevant Lin4 mimics or mock transfection, significantly reduced their sensitivity to the growth inhibitory effect of trastuzumab, as the IC 50 of trastuzumab in these cells as determined by the [ 3 H]thymidine incorporation assay was 4.5-6.8fold higher than the mock transfected cells, respectively (p Ͻ 0.01, Fig. 4A). Additionally, when the trastuzumab-sensitive cells were treated with 10 g/ml of trastuzumab for 3 days, soft agar assay demonstrated that colony formation in the cells transfected with miR-21 mimics was 1.8 -2.0-fold higher than the mock transfected cells (p Ͻ 0.001, Fig. 4B). Moreover, trastuzumab treatment did not result in any changes of colony formation in the cells with enforced miR-21 expression (p Ͼ 0.05). However, in the absence of trastuzumab, miR-21 mimics increased [ 3 H]thymidine incorporation by nearly 7% (p Ͻ 0.05) and colony formation by ϳ8% (p Ͻ 0.01) in the parental lines (supplemental Fig. S4 and Fig. 4B). These data provided evidence that ectopic miR-21 expression confers trastuzumab resistance on HER2 ϩ breast cancer cells.
Furthermore, BrdU incorporation in the presence of 10 g/ml of trastuzumab for 3 days remained unchanged in the parental BT474, SKBR3, or MDA-MB-453 lines that were transfected with miR-21 mimics (p Ͼ 0.05, Fig. 4C), but was substantially reduced in control cells that were mock transfected or transfected with Lin4 mimics (p Ͻ 0.01, Fig. 4C). In agreement, under treatment with trastuzumab, the parental lines transfected with miR-21 mimics did not exhibit any significant enrichment of the G 1 population (p Ͼ 0.05) determined by propidium iodide staining followed by flow cytometric analysis, whereas a notable increase in the number of mock-transfected cells in G 1 was observed (p Ͻ 0.01, Fig. 4D). Taken together, these results indicate that ectopic miR-21 expression overcomes the G 1 -S arrest and proliferation inhibition medi-ated by trastuzumab treatment, and thus induces drug resistance.
miR-21 Induces Trastuzumab Resistance via PTEN Inhibition-To investigate whether miR-21-induced PTEN reduction is responsible for trastuzumab resistance in breast cancer cells, we retrieved PTEN expression in the resistant BT474 line using a miR-21 ASO or a p3XFLAG-PTEN-mut construct with three nucleotides mutated in the reported miR-21 seed sequence (20) induced by PCR site-directed mutagenesis (Fig. 5A). Transfection of the trastuzumab-resistant BT474 cells with p3XFLAG-PTEN-mut, instead of an empty p3XFLAG vector, restored PTEN expression to the level comparable with the parental line (supplemental Fig. S5),

miR-21 Mediates Trastuzumab Resistance
which was in line with transfection with miR-21 ASO as shown above.
It has been shown that trastuzumab exerts its anti-tumor effect by enhancing PTEN phosphatase activity and inhibiting AKT phosphorylation in HER2 ϩ breast cancer cells (7). Therefore, we evaluated whether retrieving PTEN expression in the trastuzumab-resistant cancer cells may recover their capability of PTEN activation upon treatment with the antibody. Interestingly, when PTEN expression was restored in the resistant BT474 cells with miR-21 ASO or p3XFLAG-PTEN-mut, treatment with trastuzumab for 1 h dramatically increased PTEN phosphatase activities, measured following immunoprecipitation of a similar amount of PTEN protein using an anti-PTEN antibody (Fig. 5B), and the activity remained stably high when treated for 24 or 72 h (data not shown). Aligned with restoration of PTEN expression and activity, the resistant cells transfected with miR-21 ASO or p3XFLAG-PTEN-mut, but not with Lin4 ASO or p3XFLAG vector, were re-sensitized to the inhibitory effect of trastuzumab on AKT phosphorylation (Fig. 5C), similar to their parental lines. However, trastuzumab-mediated AKT inhibition in the parental lines was independent of transfection with miR-21 ASO or p3XFLAG-PTEN-mut (Fig. 5C). These data suggest that retrieving PTEN expression in the resistant breast cancer cells restores trastuzumab-mediated AKT phosphorinhibition.
Furthermore, we evaluated whether restoring PTEN expression and activation may re-sensitize the resistant cancer cells to trastuzumab treatment in the context of miR-21 overexpression. Transfecting the resistant BT474 cells with p3XFLAG-PTEN-mut restored the inhibitory effect of trastuzumab on [ 3 H]thymidine incorporation, as the IC 50 decreased by ϳ2.6fold as compared with cells transfected with an empty p3XFLAG vector (p Ͻ 0.01, Fig. 5D). On the other hand, cotransfecting the parental BT474 cells with miR-21 mimics and p3XFLAG-PTEN-mut did not change [ 3 H]thymidine incorporation significantly upon treatment with various concentrations of trastuzumab (p Ͼ 0.05, Fig. 5E), whereas co-transfection of miR-21 mimics and the empty p3XFLAG vector significantly increased the IC 50 concentration of trastuzumab (p Ͻ 0.01, Fig. 5E). Therefore, retrieving PTEN expression may abolish trastuzumab resistance induced by miR-21 overexpression.
miR-21-mediated PTEN Silence Confers Trastuzumab Resistance on Breast Tumor Xenografts-To investigate whether miR-21 confers trastuzumab resistance in breast cancers in vivo, we treated the athymic nude mice inoculated with BT474 cells in their mammary fat pads with trastuzumab for 3 weeks and followed the growth of tumors. This regimen of trastuzumab did not influence miR-21 expression in the xenografts, as the miR-21 level in the xenografts determined by qRT-PCR was comparable in trastuzumab-and IgG-treated mice (p Ͼ 0.05, supplemental Fig. S6), which was not significantly different from that of the xenografts harvested before trastuzumab treatment (p Ͼ 0.05, supplemental Fig. S6).
When the xenografts of the trastuzumab-resistant BT474 line reached Ϸ150 mm 3 , miR-21 ASO (20 g/injection, twice a week) was injected into the tumors 72 h prior to intravenous injection of trastuzumab (10 mg/kg, twice a week). Without trastuzumab treatment, injection with miR-21 ASO only minimally inhibited tumor growth as compared with non-relevant Lin4 ASO (p Ͻ 0.05, Fig. 6A and supplemental Fig. S7A). Although trastuzumab did not influence the growth rate of the resistant BT474 xenografts, following intratumor injection with miR-21 ASO, the magnitude of tumor inhibition was significantly more profound in animals receiving trastuzumab dosing than those without treatment with the antibody (p Ͻ 0.01, Fig. 6A and supplemental Fig. S7A), suggesting that miR-21 inhibition may sensitize trastuzumab-resistant breast cancers to the antibody. On the other hand, treatment with trastuzumab inhibited the growth of tumor xenografts in nude mice inoculated with the parental BT474 line (p Ͻ 0.01, Fig. 6B and supplemental Fig. S7B), but was unable to retard tumor growth when miR-21 mimics (15 g/injection, twice a week) were injected into the xenografts 72 h prior to trastuzumab treatment (p Ͼ 0.05, Fig. 6B). However, without trastuzumab treatment, miR-21 mimics alone revealed a trend to accelerate the growth rate of parental BT474 xenografts, but did not reach statistical significance (p Ͼ 0.05, Fig. 6B).
In agreement with the changes in tumor growth rate, trastuzumab treatment reduced the percentage of PCNA ϩ cancer cells in the trastuzumab-resistant BT474 xenografts pretreated with miR-21 ASO (p Ͻ 0.01, Fig. 6C and supplemental Fig.  S7C), but failed to do so in those pretreated with Lin4 ASO (p Ͼ 0.05). Vice versa, trastuzumab was unable to reduce PCNA ϩ cancer cells in the parental BT474 xenografts pretreated with miR-21 mimics (p Ͼ 0.05), whereas it significantly reduced the one in those pretreated with Lin4 mimics (p Ͻ 0.01, Fig. 6D and supplemental Fig. S7D). However, intratumor injection with miR-21 ASO in trastuzumab-resistant BT474 xenografts or with miR-21 mimics in the parental BT474 xenografts without trastuzumab treatment only slightly changed PCNA expression in breast cancer cells (p Ͻ 0.05, Fig. 6, C and D, and supplemental Fig. S7, C and D). These data suggest that miR-21 overexpression in breast cancer cells enhances their resistance to the proliferation inhibitory effect induced by trastuzumab.
To further evaluate whether in vivo trastuzumab resistance in breast cancers mediated by miR-21 overexpression is related to PTEN gene silence, we examined PTEN expression in the xenografts using immunohistochemistry. As shown in Fig. 6D, PTEN expression in the tumor xenografts of the parental BT474 line was effectively inhibited (p Ͻ 0.01, supplemental  Table S3). In situ hybridization for miR-21 demonstrated that expression of the miRNA in tumor biopsies obtained prior to trastuzumab therapy was more abundant in the resistant tumors than the sensitive ones (Fig. 7, A and B). This was further confirmed by qRT-PCR showing that miR-21 was ϳ2.7-fold higher in the trastuzumab-resistant tumors than the sensitive ones (p Ͻ 0.01, Fig. 7C). Moreover, miR-21 expression was further up-regulated following trastuzumab therapy in both the resistant and sensitive tumors, but the difference did not reach statistical significance (Fig. 7, B and C), implying that given a prolonged period of trastuzumab treatment, acquired trastuzumab resistance due to induction of miR-21 may occur. In line with miR-21 up-regulation, the protein expression of PTEN, a miR-21 target, was significantly lower in trastuzumab-resistant tumors as compared with the sensitive ones, determined by immunohistochemistry before and after trastuzumab treatment (Fig. 7, D and E). Moreover, miR-21 expression in breast cancer cells was reversely correlated with PTEN expression (Fig. 7F), and the correlation was higher in tumors after trastuzumab therapy (Fig. 7F, p Ͻ

miR-21 Mediates Trastuzumab Resistance
0.01) than before (Fig. 7F, p Ͻ 0.05). These data suggest that miR-21 up-regulation is associated with PTEN reduction and trastuzumab resistance in breast cancers.

DISCUSSION
Trastuzumab is a model of highly specific targeted cancer therapy for HER2 overexpressing breast cancers, but the tumors may be initially resistant to the antibody or acquire the trait of resistance in the course of treatment (6). In the present study, we demonstrate that miR-21 up-regulation is induced in the process of acquired trastuzumab resistance in HER2 overexpressing breast cancer cells, and contributes to resistance by inhibiting PTEN expression. Retrieving PTEN expression in trastuzumab-resistant cells restores their sensitivity to the antibody, whereas ectopic expression of miR-21 in the previously sensitive cells confers trastuzumab resistance via PTEN inhibition. Clinically, up-regulation of miR-21, accompanied by reduced PTEN, is associated with poor trastuzumab response in breast cancers.
Altered expression and dysfunction of miRNAs have been well documented in nearly all types of human malignancies, and numerous miRNAs are involved in tumor formation and progression by regulating the expression and action of many oncogenes and tumor suppressor genes. Although most of the tumor-related miRNAs are down-regulated in cancer tissues as compared with their normal counterparts, miR-21 is one of the few miRNAs that are consistently up-regulated in malignancies of various tissue origins (21), including the breast (22), prostate (23), colon (24), lung (25), stomach (26), liver (27), tongue (18), and brain (28) etc. miR-21 is significantly up-regulated in a subgroup of breast cancers, and high miR-21 expression has been associated with unfavorable pathological and molecular features of the disease, including advanced tumor stage, high tumor grade, lymph node metastasis, negative hormone receptor status, and poor patient survival (29). In addition, miR-21 has been shown to promote oncogenesis (30) and progression (31) of breast cancers in vitro and in vivo. Intriguingly, the hypothesis of miR-21 as a central onco-miR in tumor formation is further strengthened by a recent finding demonstrating "onco-miR addition," in which mice with conditionally expressing miR-21 develop a pre-B malignant lymphoid-like phenotype, and inhibiting miR-21 alone induces tumor regression completely (32). Here, our study shows that overexpression of miR-21 is sufficient to induce trastuzumab resistance in HER2 ϩ expressing breast cancers in vitro and in vivo, whereas targeting the onco-miR alone is capable of restoring trastuzumab sensitivity in breast cancer cells with acquired resistance. Likewise, miR-21 overexpression also contributes to resistance of malignant tumors, including breast cancers, to chemotherapeutic agents, such as pacritaxel (33), docetaxel (34), and gemcitabine (35) etc. Therefore, miR-21 is not only crucial in the oncogenesis and development of breast cancer, but is also a key factor in the development of resistance to anti-cancer therapies.
In our present study, miR-21 expression is dramatically increased when HER2 overexpressing breast cancer cells are induced to acquire trastuzumab resistance by long-term exposure to the antibody in culture. This is inconsistent with the increased homodimerization of EGFR or heterodimerization of EGFR/HER3 and the elevated levels of phosphorylated AKT after continuous culture of previously sensitive cells in trastuzumab-containing medium, which are suggested as potential molecular mechanisms for acquired trastuzumab resistance (36 -37). Because the miR-21 promoter contains highly conserved regions with consensus binding sites for several transcription factors, including STAT3 (38, 39), AP-1 (40), and factors of the forkhead family (FOXO), such as FOXO3a (41), transcription of primary miR-21 (pri-miR-21) can be regulated via various pathways. Among them, activation of the RAS/ MEK/MAPK pathway due to increased homodimerization of EGFR or heterodimerization of EGFR/HER3 potentially may lead to enhanced transcription of pri-miR-21 via AP-1 (42,43). Additionally, phosphorylated activation of AKT may also upregulate miR-21 expression by relieving the repression of FOXO transcription factors on its promoter activity (41). On the other hand, miR-21 may result in EGFR overexpression and homodimerization or heterodimerization with other EGF receptors (44). Likewise, our data demonstrate that miR-21  MAY 27, 2011 • VOLUME 286 • NUMBER 21 contributes to sustain AKT phosphorylation in the presence of trastuzumab in the resistant cells. These may feed back to maintain miR-21 up-regulation in trastuzumab-resistant breast cancer cells. In these scenarios, miR-21-mediated trastuzumab resistance is in concert with other resistant mechanisms, and may form positive feed-forward loops with other pathways to exert the resistant effect. The role of miR-21 in acquired trastuzumab resistance can be reflected by our findings that blocking the action of miR-21 alone with ASO is sufficient to restore AKT inhibition by trastuzumab in the resistant cells, and is capable of sensitizing the resistant cells to trastuzumab in vitro and in vivo.

miR-21 Mediates Trastuzumab Resistance
The implications of the above findings spread past acquired trastuzumab resistance, as overexpression of miR-21 was observed in breast cancers that responded poorly to neoadjuvant trastuzumab therapy. Indeed, a subgroup of breast cancers were reported to express a high level of miR-21 and stimulation of HER2/NEU signaling may trigger miR-21 up-regulation (45). Moreover, enforced miR-21 expression in the sensitive breast cancer cells by transfecting its mimics confers trastuzumab resistance in these cells. Therefore, miR-21 up-regulation is not only crucial to acquired trastuzumab resistance, but may also play a critical role in the inherent primary resistance.
It is well established that miRNAs function via repression of target gene translation, and miR-21 has a variety of mRNA targets including PTEN (20), CDC25A (24), tropomyosin ␣ 1 (46), tissue inhibitor metalloproteinase 3 (47), PDCD4 (48), RHOB (31), and Maspin (49) validated in different cancer cell types. Evidence suggests that miR-21 exhibits distinct a biological function dependent on the target genes it silences in a given cancer cell type. Silence of PTEN and PDCD4 results in enhanced proliferation and reduced apoptosis (20,48), whereas CDC25A inhibition is related to cell cycle arrest (24), and tissue inhibitor metalloproteinases, RHOB, and Maspin are miR-21 targets for promoting invasion and metastasis (31,47,49). Among the miR-21 targets, PTEN deficiency is related to trastuzumab resistance in patients with HER2-overexpressing metastatic breast cancer (7). Apart from gene mutation, loss of heterozygosity at the gene locus and epigenetic down-modulation, miRNA dysregulation is also an important causal factor leading to loss of PTEN expression and function. Our results further validate that PTEN reduction due to elevating miR-21 is responsible for trastuzumab resistance in HER2 expressing breast cancer cells. PTEN is a tumor suppressing dual phosphatase that antagonizes the function of phosphatidylinositol 3-kinase (PI3K) and negatively regulates AKT activities, and PTEN phosphorylation is a crucial mechanism mediating the antitumor effect of trastuzumab by reducing and inhibiting the ErbB2 receptor-bound SRC (7). Here we show that rescuing PTEN expression in the trastuzumab resistant lines with a miR-21 ASO or a PTEN rescuing vector with the miR-21 targeting sequence deleted at its 3Ј UTR restores PTEN phosphorylation upon trastuzumab treatment, which ultimately leads to AKT inhibition and growth arrest induction. Moreover, applying the PTEN rescuing vector also abolishes trastuzumab resistance induced by miR-21 mimics. In light of this, blocking the action of miR-21 with ASO implies tremendous therapeutic value in combination with trastuzumab. Because enforced PTEN expression or direct inhibition of the AKT signaling pathway is not only technically challenging, but may interfere with normal cellular functions, specifically targeting the upregulated oncogenic miR-21 in cancer cells may turn out to be a more feasible therapeutic modality. However, this will require determining how to systemically deliver the ASO into HER2 expressing cancer cells in vivo and selecting delivering vectors with a targeting moiety against the HER2 receptor, such as a HER2 antibody fragment fused to protamine (50), may facilitate in vivo delivery of miR-21 ASO for therapeutic purposes.