Trastuzumab (Herceptin) is a recombinant humanized monoclonal antibody directed against HER-2. The objective response rate to trastuzumab monotherapy is 12–34% for a median duration of 9 months, by which point most patients become resistant to treatment. We created two trastuzumab-resistant (TR) pools from the SKBR3 HER-2-overexpressing breast cancer cell line to study the mechanisms by which breast cancer cells escape trastuzumab-mediated growth inhibition. Both pools maintained her-2 gene amplification and protein overexpression. Resistant cells demonstrated a higher S-phase fraction by flow cytometry and a faster doubling time of 24–36 h compared with 72 h for parental cells. The cyclin-dependent kinase inhibitor p27kip1 was decreased in TR cells, and cyclin-dependent kinase 2 activity was increased. Importantly, exogenous addition of p27kip1 increased trastuzumab sensitivity. Additionally, resistant cells displayed heightened sensitivity to the proteasome inhibitor MG132, which induced p27kip1 expression. Thus, we propose that trastuzumab resistance may be associated with decreased p27kip1 levels and may be susceptible to treatments that induce p27kip1 expression.
The her-2 (erbB2/neu) gene is amplified and/or overexpressed in approximately 20–30% of invasive breast carcinomas and is associated with increased metastatic potential and decreased overall survival (1) . Trastuzumab (Herceptin; Genentech, South San Francisco, CA) is a recombinant humanized monoclonal antibody directed against the extracellular domain of the HER-2 tyrosine kinase receptor (2) . Clinical studies established that trastuzumab is active against HER-2-overexpressing metastatic breast cancers (3 , 4) , leading to its approval in 1998 by the United States Food and Drug Administration. However, the objective response rates to trastuzumab monotherapy are actually quite low, ranging from 12% to 34% for a median duration of 9 months (4) . Currently trastuzumab is administered in combination with chemotherapies such as paclitaxel (5 , 6) or docetaxel (7) , which increase response rates, time to disease progression, and overall survival compared with trastuzumab monotherapy.
Although the mechanisms by which trastuzumab induces regression of HER-2-overexpressing tumors are incompletely defined, several molecular and cellular effects have been observed in vitro (reviewed in Ref. 8 ). Trastuzumab and the murine monoclonal antibody 4D5, from which trastuzumab is derived, induce HER-2 receptor internalization and degradation in a dose-dependent manner in the BT474 and SKBR3 HER-2-overexpressing breast cancer cell lines (9 , 10) . Cells treated with trastuzumab undergo arrest during the G1 phase of the cell cycle, with a concomitant reduction in proliferation due in part to induction of the cyclin-dependent kinase (cdk) inhibitor p27kip1 and increased formation of p27kip1-cdk2 complexes (10, 11, 12) .
Trastuzumab clearly offers clinical benefit to a subset of HER-2-positive breast cancers. However, the majority of patients who achieve an initial response generally acquire resistance within 1 year (13) . Elucidating mechanisms by which tumors escape the cytotoxic properties of trastuzumab is critical to improving the survival of metastatic breast cancer patients whose tumors overexpress HER-2. In this study we found that trastuzumab-resistant (TR) breast cancer cells derived from the SKBR3 HER-2-overexpressing line have decreased p27kip1 levels in association with increased cdk2 activity and increased proliferation rate. Furthermore, we demonstrate that ectopic expression of p27kip1 increases trastuzumab sensitivity of resistant cells and that resistant cells display heightened sensitivity to pharmacological induction of p27kip1. Our findings suggest that a subset of TR breast cancer cells may be susceptible to treatments that induce p27kip1 expression.
MATERIALS AND METHODS
Trastuzumab (Genentech) was dissolved in sterile water at 20 mg/ml. The 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxy-phenyl)-2-(4-sulfonyl)-2H-tetrazolium CellTiter 96 Aqueous One Solution proliferation reagent (Promega, Madison, WI) was used in accordance with manufacturer guidelines. MG132 (Calbiochem, EMD Biosciences, Inc., San Diego, CA) was dissolved as a 10 mm stock in DMSO.
SKBR3 breast cancer cells were purchased from the American Type Culture Collection (Manassas, VA) and maintained in DMEM supplemented with 10% FCS. TR SKBR3 pools were developed as described previously (14) by continuously exposing cells to trastuzumab (4 μg/ml for pool 1 and 8 μg/ml for pool 2) for 3 months, at which point cells regained morphology similar to the parental line. Cells per plate were then pooled together and tested for dose response to trastuzumab as described below. Pools are now maintained in 4 μg/ml trastuzumab, a concentration at which parental cells are not viable. For serum starvation experiments, cells were maintained overnight in OPTIMEM I reduced serum medium (Life Technologies, Inc., Gaithersburg, MD).
Dose-Response Studies and Growth Curves.
SKBR3 parental cells and TR pools were seeded at 5 × 104 cells/well in 12-well dishes. After 24 h cells were treated in triplicate with 2-fold serial dilutions of trastuzumab at doses ranging from 1 μg/ml to 32 μg/ml. Cells were trypsinized after 5 days, stained with trypan blue dye, and viable cells were counted by microscopic examination. For MG132 studies, cells were exposed to 2-fold concentrations ranging from 100 nm to 800 nm for 48 h. The results of trypan blue assays were confirmed by exposing parallel cultures to the 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxy-phenyl)-2-(4-sulfonyl)-2H-tetrazolium reagent and measuring absorbance in a microplate reader as directed by the manufacturer. All of the experiments were performed in triplicate and repeated at least three times. Growth inhibition for all of the experiments is expressed as the percentage of viable cells compared with untreated cultures. For growth curves, SKBR3 parental cells and TR pools were seeded at 5 × 104 cells/well in 12-well dishes. After 24, 48, or 72 h cells were trypsinized, stained with trypan blue dye, and counted by microscopic examination.
Fluorescent in Situ Hybridization.
Cytospin slides of parental and resistant cells were prepared. The PathVysion HER-2 DNA Probe kit (Vysis, Inc., Downers Grove, IL) was used as directed by the manufacturer. Briefly, a DNA probe that spans the entire her-2 gene labeled in SpectrumOrange and a cep-17 probe (specific for the α satellite DNA located at the centromere of chromosome 17, 17p11.1-q11.1) labeled in SpectrumGreen were hybridized onto slides. Inclusion of the cep-17 probe allows for the relative copy number of the her-2 gene to be determined. Twenty nuclei were assessed for her-2 and cep-17 copy number, and the numbers were averaged for each gene. The ratio of average her-2 to average cep-17 copy number was then calculated. A normal average her-2:cep-17 ratio is generally defined as 2; a ratio >2 indicates gene amplification.
Cell Cycle Analysis.
Untreated parental SKBR3 cells, parental cells treated with 4 μg/ml trastuzumab for 72 h, and TR pools maintained in 4 μg/ml trastuzumab were fixed overnight in 70% ethanol and resuspended in propidium iodide (50 μg/ml) supplemented with RNase A (1 μg/ml). DNA content was measured using a FACScan cytometer (Becton Dickinson, Franklin Lakes, NJ).
Protein lysates were obtained using 1% NP40 lysis buffer [150 mm NaCl, 50 mm Tris (pH 8.0), and 1% NP40] and immunoblotted (50 μg) for p27kip1 (monoclonal SX53G8; DAKO, Carpinteria, CA), HER-2 (Ab-3; Oncogene Research Products, EMD Biosciences, Inc., San Diego, CA), or β-actin (Santa Cruz Biotechnology). Each primary antibody was used at a 1:1000 dilution in 5% nonfat milk in PBS-Tween. Secondary antibodies were chosen according to the species of origin of the primary antibody and detected using enhanced chemiluminescence (Amersham-Pharmacia Biotech, Piscataway, NJ).
Immunoprecipitation and Kinase Assays.
For immunoprecipitation-immunoblot analyses, parental and resistant cells were lysed with 1% NP40 buffer. Total protein extracts (200 μg) were precleared with protein G-agarose for 1 h, and cdk2 was immunoprecipitated (polyclonal-agarose conjugate; Santa Cruz Biotechnology) overnight, washed with 1% NP40 buffer, and immunoblotted for cdk2 (monoclonal; Santa Cruz Biotechnology) and p27kip1 (monoclonal SX53G8; DAKO). Incubating lysates with normal rabbit IgG overnight produced immunoprecipitation controls. Because the nuclear protein histone H1 is a substrate of phosphorylation by the cyclin E/cdk2 complex, the kinase activity of the complex can be evaluated by the detection of radiolabeled histone H1 protein (histone H1 kinase assay). In the current study, to measure the phosphorylation activity of p27-associated cdk2, histone H1 kinase assay was performed. Cdk2 was immunoprecipitated from total protein extracts (200 μg) overnight and incubated with [32P]dATP, ATP, and histone H1 (Sigma) in kinase buffer. Samples were analyzed by SDS-PAGE and autoradiography, and relative cdk2 kinase activity was measured by quantitating [32P]histone H1 using the NIH Image program. Experiments were performed in duplicate.
Parental and resistant cells were transiently transfected with 25 ng or 100 ng of pCMV-FLAG-p27kip1 (provided by Dr. Mong-Hong Lee, M.D. Anderson Cancer Center) or 100 ng of an empty pCMV vector as a control for 48 h using Lipofectamine (Invitrogen). At that time, cells were either lysed for protein and immunoblotted for p27kip1 (monoclonal SX53G8, DAKO) or else incubated with trastuzumab (1, 5, or 10 μg/ml) for 72 h. Trastuzumab-treated p27kip1-transfected and trastuzumab-treated control vector-transfected cells were counted by trypan blue exclusion assay. Cell viability is reported as a percentage of control cells transfected with the pCMV-FLAG-p27kip1 or the empty pCMV vector but not treated with trastuzumab.
Development of TR SKBR3 Pools.
HER-2-overexpressing SKBR3 breast cancer cells were continuously exposed to trastuzumab for 3 months, at which point cells regained morphology similar to the parental line. Two different doses of trastuzumab (4 μg/ml for pool 1 and 8 μg/ml for pool 2) were used in separate cultures to ensure that cells were exposed to a high enough dose to kill all of the sensitive cells. Cells per plate were then pooled together. Sensitivity to trastuzumab was determined by treating pools with 2-fold serial dilutions of the drug and performing trypan blue exclusion assays after 5 days. Growth inhibition is expressed as the percentage of viable cells compared with untreated cultures (Fig. 1A) ⇓ . A dose of 4 μg/ml trastuzumab was lethal to parental cells, but the majority of pool 1 and 2 cells remained viable. Thus, these pools demonstrated resistance to trastuzumab and are now maintained in 4 μg/ml trastuzumab. Trastuzumab resistance was also confirmed by 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxy-phenyl)-2-(4-sulfonyl)-2H-tetrazolium colorimetric assay (Promega; data not shown). To ensure that cells were specifically resistant to trastuzumab, we treated resistant and parental cells with an anti-epidermal growth factor receptor (EGFR) antibody (data not shown). Resistant and parental cells were similarly sensitive to EGFR blockade, suggesting that there is not cross-resistance to another antibody, but that the resistance is specific for trastuzumab.
To eliminate the possibility that we had selected a population of cells that do not possess her-2 gene amplification, fluorescent in situ hybridization was performed for both her-2 and cep-17 using the PathVysion HER-2 DNA Probe kit. The ratio of the average her-2 gene copy number to the average cep-17 gene copy number was similar among all three of the samples (Table 1) ⇓ , demonstrating that both TR pools possess her-2 amplification at the same level as parental cells. Immunoblotting additionally demonstrated that resistant pools that have been maintained in the absence of trastuzumab for 96 h express HER-2 at levels equivalent to parental cells (Fig. 1B) ⇓ . Additionally, HER-2 levels were down-regulated when the resistant pools were maintained in trastuzumab (Fig. 1C) ⇓ , suggesting that the molecular mechanisms contributing to resistance lie downstream of the receptor.
TR Cells Have an Increased Proliferation Rate.
Cell cycle analysis of untreated parental cells, parental cells treated with 4 μg/ml trastuzumab for 72 h, and TR pools maintained in 4 μg/ml trastuzumab was performed by flow cytometry. Representative cell cycle profiles are shown for each cell line (Fig. 2A) ⇓ . Trastuzumab produced a 1.5-fold decrease in the percentage of proliferating parental cells (Fig. 2B) ⇓ with a modest increase in G1-phase cells (Table 2) ⇓ . The percentage of G1 cells for each pool was similar to the untreated parental cells. However, the S-phase fraction in the TR pools was three times higher than that of untreated parental cells. Concomitant with increased proliferation was a 3-fold decrease in G2-M cells versus untreated parental cells. Growth curves (Fig. 2C) ⇓ additionally demonstrated that the doubling time for TR pools is 2–3-fold faster than for parental cells, as TR pools double in 24–36 h compared with 72 h for the parental cells. These results indicate that TR SKBR3-derived cells have an increased S-phase fraction and reduced doubling time, and confirm that these pools are resistant to trastuzumab-mediated inhibition of proliferation.
TR Cells Express Reduced Levels of p27kip1.
The growth-inhibitory properties of trastuzumab depend in part on effects on the cdk-inhibiting protein p27kip1. Trastuzumab increases the half-life of p27kip1 by decreasing cyclin E/cdk2-mediated phosphorylation of p27kip1 and blocking subsequent ubiquitin-dependent degradation (15) . Trastuzumab also increases association between p27kip1 and cdk2 complexes resulting in G1 arrest (12) . Importantly, small interfering RNA that reduced p27kip1 expression levels also blocked trastuzumab-mediated growth arrest in SKBR3 cells (15) . Thus, we examined expression levels of p27kip1 in our TR SKBR3 model. Parental cells treated with 4 μg/ml trastuzumab over a 72 h time course demonstrated induction of the p27kip1protein (Fig. 3A) ⇓ . Importantly, both of the TR pools expressed reduced levels of p27kip1 in comparison with untreated parental cells, with pool 2 showing the greatest reduction. To ensure that the reduction in p27kip1 was not merely a reflection of the increase in S phase, cell proliferation was blocked by incubating overnight in reduced serum medium. Immunoblotting serum-starved cells for p27kip1 confirmed that TR cells express lower levels of p27kip1 than parental cells (Fig. 3B) ⇓ .
Cdk2 Kinase Activity Is Elevated in TR Cells.
The p27kip1 protein binds and inhibits cdk2 complexes. Thus, we examined the interaction between p27kip1 and cdk2 and measured cdk2 kinase activity in the TR pools. Cdk2 was immunoprecipitated from parental and resistant cells and immunoblotted for p27kip1 (Fig. 4A) ⇓ . Incubating lysates with normal rabbit IgG produced immunoprecipitation controls. Levels of cdk2 were quantitated and normalized, and the ratio of p27kip1 to cdk2 was then determined. In comparison with parental cells, pools 1 and 2 displayed ∼30% less p27kip1 bound to cdk2. To determine whether this correlated with decreased inhibition of cdk2 activity, immunoprecipitation-cdk2 radioactive kinase assays were performed, using histone H1 as a substrate (Fig. 4B) ⇓ . Pool 1 displayed a very modest 1.7-fold increase in phosphorylated histone H1, and pool 2 showed a 2.9-fold elevated cdk2 kinase activity versus parental cells. These results were consistent with p27kip1expression data, which showed reduced levels in both pools, but a greater reduction in pool 2. Thus, both TR pools harbor elevated cdk2 kinase activity in association with reduced p27kip1levels, with pool 2 showing the greatest difference versus parental cells.
Expression of p27kip1 Restores Trastuzumab Sensitivity to TR Pools.
To determine whether reduced p27kip1 contributes to trastuzumab resistance in our cell model, a mammalian expression vector containing a FLAG-p27kip1 construct was transiently transfected into parental and resistant cells. A representative immunoblot of parental cells transfected with 25 ng or 100 ng of CMV-FLAG-p27kip1 or with 100 ng of the empty vector control for 48 h demonstrated that FLAG-p27kip1, which migrates higher than endogenous p27kip1, is expressed efficiently in cells receiving pCMV-FLAG-p27kip1 (Fig. 5A) ⇓ . Immunoblotting also indicated that 25 ng and 100 ng of CMV-FLAG-p27kip1 encompass the level of endogenous p27kip1 found in parental cells. Cells transfected with 100 ng control vector or 25 ng or 100 ng of pCMV-FLAG-p27kip1 for 48 h were then treated with 1, 5, or 10 μg/ml trastuzumab for 72 h, and counted by trypan blue exclusion (Fig. 5B) ⇓ . Expression of p27kip1 increased the sensitivity of parental and resistant cells to trastuzumab. Parental cells transfected with 25 ng FLAG-p27kip1 were similarly sensitive to trastuzumab as control transfectants, probably because these cells already express relatively high levels of endogenous p27kip1, at least compared with the resistant cells. FLAG-p27kip1 at 100 ng concentration increased the sensitivity of parental cells. The trastuzumab sensitivity of both TR pools 1 and 2 dramatically increased when p27kip1 was transfected at either 25 ng or 100 ng. These data suggest that reduced levels of p27kip1 may be a major mechanism of trastuzumab resistance in these cells. However, it is not likely to be the only mechanism. The slightly different responses of pool 1 versus pool 2 support the concept that mechanisms contributing to trastuzumab resistance are likely to vary among different cancer cell populations but that reduced p27kip1 expression may be one important mechanism.
TR Pools Demonstrate Increased Sensitivity to Pharmacological Induction of p27kip1.
The peptide aldehyde MG132 (Cbz-leu-leu-leucinal) inhibits proteasome-mediated degradation of multiple cell cycle regulating proteins including p27kip1 (16) . Parental and resistant pools were treated with MG132 for 48 h and counted by trypan blue exclusion (Fig. 6A) ⇓ . Both pools demonstrated increased sensitivity to MG132 relative to parental cells, with ∼50% of resistant cells viable at 400 nm MG132 versus ∼90% of parental cells. This dose of MG132 induces p27kip1 expression to a similar degree in resistant and parental cells (Fig. 6B) ⇓ . Hence, resistant cells appear to be more sensitive to induction of p27kip1 versus parental cells. These results support the finding that p27kip1 expression inhibits the survival of these TR cells.
In the clinical setting, trastuzumab achieves an objective response rate of up to 34% for 9 months when administered as a first-line agent in HER-2-overexpressing metastatic breast cancers (4) . The remaining cases do not demonstrate tumor regression (17) . Furthermore, tumors that initially respond to trastuzumab begin to progress again within 1 year (4) . We have created TR cells to determine potential mechanisms by which tumors escape trastuzumab-mediated growth inhibition and to identify novel agents that may overcome this resistance.
Trastuzumab-mediated growth inhibition requires expression of p27kip1, as small interfering RNA against p27kip1 blocked growth arrest of SKBR3 cells exposed to trastuzumab (15) . Thus, we examined whether reduced p27kip1 expression contributed to trastuzumab resistance in our SKBR3 cell model. We demonstrated here that SKBR3 TR cells have an increased S-phase fraction associated with reduced p27kip1 levels and increased cdk2 kinase activity relative to parental cells. Expression of p27kip1 restored trastuzumab sensitivity in the resistant cells, suggesting that p27kip1 contributes to trastuzumab resistance. Sensitivity of TR pools to trastuzumab was restored close to parental levels upon transfection with p27kip1. Additionally, both pools responded similarly to the proteasome inhibitor MG132, which induces p27kip1 expression, exhibiting increased sensitivity versus parental cells.
p27kip1 is a distal downstream effector of multiple converging growth factor receptor pathways including EGFR, HER-2, and IGF-IR. Hence, down-regulation of p27kip1 in our resistant cells is possibly a reflection of aberrant signaling from one of these receptor pathways or from a middle signaling molecule such as phosphatidylinositol 3′-kinase or mitogen-activated protein kinase. Studies are ongoing to evaluate these possibilities. Nonetheless, the finding of reduced p27kip1 levels in these cells provides a possible marker of trastuzumab resistance and a putative therapeutic target.
Because the TR pools used in this study were derived from one cell line, it is likely that multiple mechanisms of resistance exist and that these may vary among cell lines. Additionally, mechanisms contributing to acquired versus inherent resistance may differ. The pools developed during this study may harbor resistance acquired over time or may represent a selected subpopulation of cells with inherent resistance. Interestingly, discontinuation of trastuzumab treatment for 4 months did not restore sensitivity to the TR pools, suggesting that the aberrant molecular mechanisms that contribute to resistance in these pools are either inherent or are acquired and maintained long-term. Additionally, the resistant cells used in this study are pools of multiple clones and, thus, represent multiple genetic or molecular alterations. Another approach to our research question would have been to use clones, which represent one isolated alteration. However, because pools represent the alterations that are represented in the majority of cells, we chose to use this approach. The drawback to our approach is that pools may contain cells with various degrees of resistance.
An important consideration is whether cells developed resistance specifically to trastuzumab or whether cross-resistance to other antibodies exists. Thus, we examined the response of resistant pools to an anti-EGFR antibody (data not shown). Resistant and parental cells were similarly sensitive to EGFR blockade, suggesting that there is not cross-resistance, but that resistance is specific for trastuzumab. These results open up the possibility that other HER family-targeted antibodies or kinase inhibitors may be effective treatments against cancers that have progressed while on trastuzumab. Experiments addressing this possibility are ongoing.
Our results suggest that a subset of TR breast cancers express reduced p27kip1 levels with increased cdk2 activity. These findings support p27kip1 as a potential therapeutic target in TR breast cancers. The p27kip1 protein is degraded via ubiquitin-dependent mechanisms (18) . Thus, the potential efficacy of proteasome inhibitors, which are known to induce p27kip1 expression, is being tested against the TR pools. In addition, these resistant cells may demonstrate sensitivity to cdk2 inhibiting drugs, as their levels of endogenous cdk2 kinase activity are slightly elevated. In conclusion, our data suggest that p27kip1 may be a relevant therapeutic target in TR breast cancers and support the development and testing of treatments that induce p27kip1 expression, such as proteasome inhibitors, or therapies that block cdk2 activity in these cancers.
We thank W. Fraser Symmans for interpreting FISH data. We also thank Ann Sutton and Scientific Publications for editorial assistance.
Grant support: AACR-Amgen, Inc. Fellowship in Clinical/Translational Cancer Research (R. Nahta), NIH Grants P01 CA099031 (M-C. Hung) and K23 CA82119 (F. J. Esteva), and Nellie B. Connally Breast Cancer Research Fund.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Rita Nahta, Department of Breast Medical Oncology, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Unit 424, Houston, TX 77030-4009. Phone: (713) 792-2817; Fax: (713) 745-5768; E-mail:
- Received December 12, 2003.
- Revision received February 26, 2004.
- Accepted March 16, 2004.
- ©2004 American Association for Cancer Research.