Cancer Therapy: Preclinical

In vivo Antitumor Activity of MEK and Phosphatidylinositol 3-Kinase Inhibitors in Basal-Like Breast Cancer Models Klaus P. Hoeflich,2 Carol O’Brien,1Zachary Boyd,1Guy Cavet,3 Steve Guerrero,3 Kenneth Jung,3 To m Ja n u a r i o , 1 Heidi Savage,1Elizabeth Punnoose,1Tom Truong,2 Wei Zhou,2 Leanne Berry,2 Lesley Murray,2 Lukas Amler,1Marcia Belvin,4 Lori S. Friedman,4 and Mark R. Lackner1

Abstract Purpose: The pathways underlying basal-like breast cancer are poorly understood, and as yet, there is no approved targeted therapy for this disease.We investigated the role of mitogen- activated kinase kinase (MEK) and phosphatidylinositol 3-kinase (PI3K) inhibitors as targeted therapies for basal-like breast cancer. Experimental Design: We used pharmacogenomic analysis of a large panel of breast cancer cell lines with detailed accompanying molecular information to identify molecular predictors of response to a potent and selective inhibitor of MEK and also to define molecular mechanisms underlying combined MEK and PI3K targeting in basal-like breast cancer. Hypotheses were confirmed by testing in multiple tumor xenograft models. Results:We found that basal-like breast cancer models have an activated RAS-like transcriptional program and show greater sensitivity to a selective inhibitor of MEK compared with models rep- resentative of other breast cancer subtypes.We also showed that loss of PTEN is a negative pre- dictor of response to MEK inhibition, that treatment with a selective MEK inhibitor caused up-regulation of PI3K pathway signaling, and that dual blockade of both PI3K and MEK/extracel- lular signal ^ regulated kinase signaling synergized to potently impair the growth of basal-like breast cancer models in vitro and in vivo. Conclusions: Our studies suggest that single-agent MEK inhibition is a promising therapeutic modality for basal-like breast cancers with intact PTEN, and also provide a basis for rational combination of MEK and PI3K inhibitors in basal-like cancers with both intact and deleted PTEN.

Recent molecular profiling studies have suggested that breast poor prognosis if untreated, and significant clinical benefit cancers represent at least three distinct diseases with a common from the HER2 targeting monoclonal antibody trastuzumab tissue of origin but different underlying biological aberrations. (Herceptin, Genentech; ref. 3). Basal-like breast cancers expression studies have shown that primary breast tumors typically lack expression of HER2, ER, and progesterone may be classified into at least three major subtypes by receptor and hence overlap with tumors immunohistochemi- unsupervised hierarchical clustering analysis and that the cally defined as ‘‘triple-negative’’ tumors (4). Basal-like breast subtypes have different prognostic outcomes in terms of patient cancers have a poor prognosis and currently have not been survival (1). Luminal breast cancers are typically estrogen shown to respond to any targeted therapy and thus represent a receptor (ER) positive and characterized by coordinate expres- particularly pressing unmet medical need (5). Interestingly, the sion of a number of epithelial specific , a relatively good majority of basal-like breast cancers express high levels of the prognosis, and good response rates to targeted hormonal epidermal growth factor receptor (EGFR) and have an EGFR- therapies (2). HER2-positive breast cancers are characterized by associated signature (6). Because growth HER2 high-level gene amplification of the oncogene, relatively factor stimulation of EGFR results in potent activation of the signaling pathway composed of RAS, RAF, mitogen-activated protein kinase (MAPK) kinase (MEK), and extracellular Authors’ Affiliations: 1Development Oncology Diagnostics, 2Translational 3 4 signal–regulated kinase (ERK; ref. 7), we sought to determine Oncology, Bioinformatics, and Cancer Signaling Pathways, Genentech, Inc., South San Francisco, California whether the downstream arm of this pathway might be a target Received2/9/09;revised3/19/09;accepted3/24/09;publishedOnlineFirst6/30/09. of therapeutic intervention in breast cancer. The costs of publication of this article were defrayed in part by the payment of page The protein kinase MEK occupies a crucial signaling node advertisement charges. This article must therefore be hereby marked in accordance downstream of RAS and RAF and directly upstream of ERK and with 18 U.S.C. Section 1734 solely to indicate this fact. Note: Supplementary data for this article are available at Clinical Cancer Research as such has been the subject of intensive drug discovery Online (http://clincancerres.aacrjournals.org/). activities (8). Genetic and biochemical analyses of MEK K. Hoeflich, C. O’Brien, and Z. Boyd contributed equally to this work. function have suggested that MEK activity is necessary for the Requests for reprints: Mark R. Lackner, Genentech, Inc., 1DNA Way, South San transforming and proliferative effects of this pathway, suggest- Francisco, CA 94080. Phone: 650-467-1846; Fax: 650-225-7571; E-mail: [email protected]. ing that therapeutics that completely inhibit MEK function may F 2009 American Association for Cancer Research. have utility in the treatment of cancers driven by activation of doi:10.1158/1078-0432.CCR-09-0317 the RAS/RAF axis. Such efforts have led to the development of

www.aacrjournals.org 4649 Clin Cancer Res 2009;15(14) July 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. Cancer Therapy: Preclinical

cancer based on in vitro analyses (16). Our studies are consistent Translational Relevance with those findings and, moreover, offer extensive in vivo confirmation of the important role MEK signaling plays in We find that basal-like breast cancer models are sensitive basal-like breast cancer. In addition, we also show that activation in vitro in vivo and to the inhibitory effects of mitogen- of the PI3K/Akt pathway by loss of the tumor suppressor PTEN activated protein kinase kinase (MEK) inhibition relative to occurs in a subset of basal-like breast cancer cell line models and luminal and HER2-amplified tumors and identify the tumor is associated with diminished response to MEK inhibition, and suppressor PTEN as a negative predictor factor for that siRNA ablation of PTEN in responsive models blocks cell response to MEK inhibition.We also find that a combination cycle and proliferative changes triggered by MEK inhibition. of MEK and phosphatidylinositol 3-kinase inhibitors Based on the apparent role of PTEN and activation of the PI3K/ results in enhanced efficacy in both PTEN wild-type and Akt pathway in mediating resistance to MEK inhibition, we use PTEN-null basal-like tumors and provide a mechanistic both in vitro and in vivo combination studies with potent and explanation for this observation. Thus, our studies provide selective MEK and PI3K inhibitors to show that simultaneous a rational hypothesis for biomarker evaluation and patient treatment with a MEK inhibitor and a potent and selective stratification in clinical trials seeking to evaluate the efficacy inhibitor of PI3K results in synergistic decreases in cell viability, of MEK and phosphatidylinositol 3-kinase inhibitor enhanced down-regulation of cyclin D1, and increased apoptosis treatment in this poor-prognosis tumor type. specifically in basal-like breast cancer models. molecules such as PD0325901 and AZD6244 that act through a Materials and Methods novel noncompetitive allosteric mechanism of inhibition by binding to a site adjacent to the ATP-binding site and locking Cell lines and cell viability experiments. Details and sources of all the kinase in a closed and catalytically inactive conformation cell lines used in this study have been described previously (11, 17) and (9). This pathway is frequently activated by mutations in RAS are also detailed in Supplementary Materials and Methods. All studies family members or BRAF in a number of solid tumor types, and in this report describing MEK inhibition have used the potent and elegant studies in cell lines and mouse tumor models have selective MEK1/2 inhibitor PD0325901 (Pfizer) to evaluate the effects suggested that mutations in BRAF and, to a lesser extent, KRAS of MEK inhibition in breast cancer preclinical models, with the exception of a small number of confirmatory studies described in may be useful to stratify patients enrolled in clinical trials of Supplementary materials that use a second potent and selective MEK MEK inhibitors (10). We sought to address whether MEK inhibitor (9), purchased from Synthesis MedChem.5 For cell viability inhibition might have therapeutic value in breast cancer studies, cells were plated in quadruplicate at a density of 3,000 per well through evaluation of a large panel of breast cancer cell lines in 384-well plates in normal growth medium and allowed to adhere and tumor xenograft models that have been characterized in overnight. Compounds were added in 10 concentrations based on a detail at the molecular level and shown to reflect many of the 3-fold dilution series. Cell viability was measured 72 h later using the genomic alterations commonly found in breast tumors (11). CellTiter-Glo Luminescent Cell Viability Assay (Promega). The concen- We also set out to examine the relationship between RAS/ tration of drug resulting in 50% of maximal inhibition (EC50) was RAF/MEK/ERK pathway activity and phosphatidylinositol calculated from a four-parameter curve analysis (XLfit, IDBS software) 3-kinase (PI3K) pathway activity in breast cancer. PI3K signals and was determined from a minimum of three experiments. For cell lines where minimal inhibition was observed and the software was through Akt and other downstream effectors and represents a unable to calculate an EC50, the value was nominally assigned as the key mechanism of transducing proliferative and antiapoptotic highest concentration screened (20 Amol/L) and these cell lines were signals (12), and this pathway is activated to some extent in all classified as resistant. Caspase-3/caspase-7 activity was assessed by three breast cancer subtypes. Activating mutations in the plating 5,000 cells per well in quadruplicate in 384-well plates catalytic subunit of p110-a (PIK3CA) occur in 25% to 30% and incubation in the presence of DMSO, 1 Amol/L MEK inhibitor, of breast tumors and are associated with hormone receptor 1 Amol/L PI3K inhibitor, or a combination of MEK and PI3K inhibitors positive (i.e., luminal subtype) status (13), and it has recently each at 1 Amol/L. Activation of caspase-3/caspase-7 was assessed been shown that loss of the tumor suppressor gene PTEN is 24 h later by the Caspase-Glo 3/7 assay kit (Promega). common in basal-like tumors (14). Finally, the HER2 oncogene High-content analysis of cell proliferation. High-content screening signals through the PI3K pathway, and Herceptin has been assays were carried out on an Arrayscan VTI (Cellomics, Inc.) using standard bromodeoxyuridine incorporation and staining protocols, as shown to down-regulate antiapoptotic signaling mediated by described in Supplementary Experimental Procedures. this pathway (15). Given the role of both the RAS and PI3K In vitro combination experiments. In vitro synergy was assessed pathways in cancer signaling, a crucial question is whether using the combination index of Chou and Talalay method using these pathways are activated concomitantly in breast tumors CalcuSyn software.6 Generally, combination index values of <0.5 are and how activation of one pathway can affect the response to taken to indicate synergistic interaction between compounds, and therapeutics that target the other. combination index values near 1.0 indicate no interaction (18). A central finding of this study is that basal-like breast cancer Additional details are provided in Supplementary Experimental cell lines and tumor models manifest an oncogenic Ras-like Procedures. gene expression signature and are substantially more sensitive Three-dimensional cell culture. For three-dimensional cultures, in vitro and in vivo to MEK inhibition than luminal or HER2- BT474M1 and MDA-MB-231 cells were trypsinized from monolayer cultures and plated on top of commercially available laminin-rich amplified models, suggesting that MEK inhibition represents a promising therapeutic modality for this aggressive tumor type with currently limited treatment options. A recent independent 5 http://www.synmedchem.com study showed a role for MEK inhibition in basal-like breast 6 www.biosoft.com

Clin Cancer Res 2009;15(14) July 15, 2009 4650 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. TargetingMEKandPI3KinBasal-likeBreastCancer

Fig. 1. In vitro response of breast cancer cell lines to MEK inhibition. A, sensitivity to a MEK inhibitor in a panel of breast cancer cell lines as determined by EC50 in a cell viability assay. Note that the y axis is graphed on a logarithmic scale to better show distinctions between cell lines with submicromolar sensitivity. Responses below 1 Amol/L (indicated by the red dashed line) may represent off-target effects of the MEK inhibitor and should be interpreted with caution. Cell lines are classified into major subtypes of breast cancer, basal-like, HER2-amplified, and luminal, based on gene expression profiling. The text below the graph indicates cell lines determined to have detectable expression of EGFR protein (+) or the presence of activating mutations in BRAF, HRAS, RRAS2, or KRAS; activating mutations in PIK3CA; or loss of PTEN (NULL), by Western blot.The heat map indicates phosphorylation levels of Akt (S473) as assessed by quantitative analysis using reverse-phase protein microarrays and is color coded according to z-scores (-2, low/absent; 0, moderate; +2, high). B, Western blots showing modulation of phospho-ERK (pERK), p27, and cyclin D1after treatment with a MEK inhibitor in five breast cancer cell lines. HDQ-P1, CAL85-1, and MDA-MB-231are sensitive to MEK inhibition of cell viability, but BT474 and MCF-7 are not. Note that pERK is suppressed in all cell lines regardless of subtype and efficacy, whereas decreased cyclin D1and increased p27 expression are seen only in basal-like cell lines that are sensitive to MEK inhibition. matrix produced from Englebreth-Holm-Swarm tumors (Matrigel, BD duplicate plates were made for each time point. Cells were incubated in Biosciences) using a previously described protocol (19). Cell lines were the compound for 6 or 24 h, then washed with cold PBS and processed maintained in modified DMEM and 10% fetal bovine serum for Western blotting using standard protocols, as described in conditioned with 5% Matrigel. Cells were treated in triplicate with Supplementary Experimental Methods. CAL85-1 cells were cultured in 0.25 Amol/L GDC-0941, 0.5 or 1 Amol/L PD0325901, or a medium containing 10 ng/mL EGF to assess phospho-Akt (pAkt) levels combination of the two agents on day 0. Culture medium was because basal levels of pAkt are undetectable in this cell line when refreshed every 3 d, and acinar growth and morphogenesis were cultured in 10% fetal bovine serum. For analysis of protein expression analyzed on day 10. Cell number was assayed on day 7 (MDA-MB-231) in tumor xenografts, lysate from tumor samples was collected by adding or day 10 (BT474) using CellTiter-Glo (Promega) according to the DNase Lysis Buffer to frozen tissue and pulverized using the TissueLyser manufacturer’s instructions. Data are representative of three indepen- (Qiagen) as described by the manufacturer. Primary blotting antibodies dent experiments. used were p27 (C-19; Santa CruzBiotechnology), cyclin D1 (DCS-6; Protein analyses. For in vitro pharmacodynamic studies, BT474, Santa CruzBiotechnology), ERK, pERK (Thr202/Tyr204) (Cell Signal- HDQ-P1, CAL85-1, and MDA-MB-231 cells were plated in 12-well ing Technology), total AKT, and pAkt (S473) (Cell Signaling plates and allowed to grow until cells reached 60% to 80% confluence. Technology). Secondary blotting antibodies used were polyclonal goat Cells were dosed with 0, 0.1, or 1.0 Amol/L MEK inhibitor and anti-mouse IgG horseradish peroxidase and polyclonal goat anti-rabbit

www.aacrjournals.org 4651 Clin Cancer Res 2009;15(14) July 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. Cancer Therapy: Preclinical

Fig. 2. AMEKinhibitorhasin vivo antitumor activity in basal-like breast cancer xenograft models. A and B, 21-d daily dosing of flank-xenografted MDA-MB-231basal-like cells substantially reduced tumor growth in a dose-dependent manner, whereas 21-d daily dosing of flank-xenografted BT474 HER2-amplified cells with doses of up to 25 mg/kg did not result in notable reduction of tumor growth compared with vehicle.Tumor volumes represent means from measurements in10 tumor-bearing mice. C and D, daily dosing of basal-like human primary tumor explant xenograft models MAXF-401and MAXF-1384 with either 12.5 mg/kg MEK inhibitor or vehicle for 20 d resulted in substantial reduction of tumor growth in animals dosed with the MEK inhibitor.

IgG horseradish peroxidase (both from DAKO or Cell Signaling 200 to 250 mm3 were grouped into treatment cohorts of 10 mice each. Technology). Quantitation of Western blot bands was accomplished For BT474M1 tumor xenografts, beige nude Xid mice (Harlan) were using Image J software (NIH) and band intensities were normalized to injected in the right flank with 5  106 human BT474M1 cells h-actin or total ERK. Quantitative analysis of protein expression using resuspended in 200 AL PBS and Matrigel (BD Biosciences). Mice with reverse-phase protein arrays was done at Theranostics Health (Rock- tumors of a mean volume of 200 to 250 mm3 were grouped into ville, MD) as described previously (20) and in Supplementary Materials treatment cohorts. MEK inhibitor was administered daily by oral gavage and Methods. at a final concentration of 2.5 to 25 mg/kg, as indicated. The compound siRNA experiments. Individual siRNA duplexes specific to human was prepared fresh weekly in 0.5% methylcellulose and 0.2% Tween 80 PTEN or EGFR (Dharmacon) or a control siRNA that does not target in water and stored at 4jC. Body weights and caliper measurements were any sequence in the (nontarget control; Dharmacon) taken twice per week during the study. Tumor volumes were calculated was used in transient transfection experiments. Transfection conditions by the following formula: tumor volume = 0.5  (a  b2), where a is the are described in detail in Supplementary Materials and Methods. largest tumor diameter and b is the perpendicular tumor diameter. In vivo studies. For in vivo studies at Oncotest, tumor fragments Tumor volume results are presented as mean tumor volume F SE and obtained from MAXF-401 and MAXF-1384 xenografts in serial passage data were analyzed by Student’s t test and Hsu’s MCB. Tumor samples were cut into 1- to 2-mm3 pieces and placed in RPMI 1640 until s.c. from additional cohorts of mice were collected for analysis of pathway implantation into NMRI nu/nu mice (Taconic). When tumors reached activation following 8 d of treatment (2 h after the last dose). an average of 100 mm3, animals were stratified into two equivalent Flowcytometry. For cell cycle analysis, cell suspensions were fixed in groups of 10 tumor-bearing mice each. MX-1 tumors were maintained 70% ethanol and stained with propidium iodide (Invitrogen) and by serial s.c. transplantation of 1-mm3 fragments in nu/nu mice DNase-free RNase A (Sigma-Aldrich) at final concentrations of 50 and (Harlan). For efficacy studies, mice with tumors of a mean volume of 20 Ag/mL, respectively. Apoptotic cells were detected using a protocol

Clin Cancer Res 2009;15(14) July 15, 2009 4652 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. TargetingMEKandPI3KinBasal-likeBreastCancer

Fig. 3. Oncogenomic signature determinants of MEK inhibitor response in basal-like breast cancer. A, gene expression signature derived from supervised analysis to identify genes differentially expressed between MEK-sensitive and MEK-resistant cell lines. Hierarchical clustering of cell lines was done with the top 100 differentially expressed genes defined by max/min ratio across the cell line panel. Cell lines are shown on the horizontal axis, and select gene identities on the vertical axis.The two rows at the top of the figure show MEK inhibitor ^ sensitive (S) and MEK inhibitor ^ resistant (R) cell lines as well as luminal (L) or basal-like (B) phenotype as determined by overall gene expression. Select genes are indicated to the right of the heat map. B, an activated Ras/MEK signature from cell lines is enriched in basal-like tumors. HRAS or MEK gain-of-function mutant were expressed in MCF-10A cells and supervised analyses were used to identify the 50 most differentially expressed genes in common between these experiments. Probes representing these genes were used to perform unsupervised hierarchical clustering on independent breast cancer tumor samples that had been classified into different molecular subtypes (luminal, basal-like, or HER2 amplified) based on independent gene expression analysis. The subtype of each sample is indicated by color coding in the top row according to the key in the figure. In the heat map, red indicates high expression and blue indicates low expression of a given probe. C, basal-like and MEK inhibitor ^ sensitive cell lines have an activated RAS signature. Predictive classifiers of RAS pathway activation were developedfromgene expression data derived from MCF-10A cells expressing mutant HRAS and used to classify breast cancer cell lines.The y axis indicates predicted RAS pathway activation based on a scalar predictor.The x axes show cell lines divided into molecular subtypes (top) and into MEK inhibitor ^ sensitive and MEK inhibitor ^ resistant classes (bottom).

www.aacrjournals.org 4653 Clin Cancer Res 2009;15(14) July 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. Cancer Therapy: Preclinical

Fig. 4. Functional analysis of EGFR signaling and PTEN in basal-like breast cancer. A, the MEK(gf) signature (y axis) is correlated with a published EGFR signature (x axis). Overall gene signature levels were calculated as described in Supplementary Experimental Procedures. Each point represents a single cell line. B, cell lines of the basal-like subtype show greater in vitro sensitivity to the EGFR inhibitor erlotinib compared with luminal or HER2-amplified cell lines. Cell lines are classified into major subtypes as described for Fig. 1, and cell viability was assessed in a 3-d ATP-based viability assay. C, EGFR siRNA knockdown and effects on MEK inhibitor sensitivity in a basal-like line without known RAS or RAF mutations (CAL85-1), a basal-like cell line with a KRAS mutation (MDA-MB-231), and a HER2-amplified cell line (JIMT-1).The y axis indicates percent cell viability relative to untreated nontarget control siRNA ^ treated cells, and the x axis indicates MEK inhibitor concentration. Light gray columns, nontarget control siRNA ^ treated cells; dark gray columns, EGFR siRNA ^ treated cells. EGFR siRNA affects viability and enhances the effects of MEK inhibition in CAL85-1, but not MDA-MB-231orJIMT-1, cells (*, P < 0.05, Student’s t test). D, effects of a MEK inhibitor on basal-like cell lines with intact PTEN or where PTEN has been ablated by RNA interference. The cell lines are labeled above each graph. The y axis of each chart shows percent cell viability normalized to nontarget control siRNA ^ treated cells in the absence of MEK inhibitor, whereas the x axis indicates the concentration of MEK inhibitor. Dark gray columns, cells treated with control siRNA; light gray columns, cells treated with two independent PTEN siRNA duplexes as indicated in the key. Bars, SDs calculated from duplicate experimental wells. PTEN siRNA treatment significantly reduced inhibition in response to treatment with two different concentrations of MEK inhibitor in CAL-85-1and MDA-MB-231cells (*, P < 0.05, Student’s t test). As a control for specificity of the siRNA duplexes, PTEN siRNA was done in the PTEN-null cell line MDA-MB-436 and had no effect on cell viability compared with control siRNA.

Clin Cancer Res 2009;15(14) July 15, 2009 4654 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. TargetingMEKandPI3KinBasal-likeBreastCancer provided with the APO-BrdU terminal deoxyribonucleotidyl classified as basal-like were significantly more sensitive to MEK transferase–mediated dUTP nick end labeling (TUNEL) assay inhibition relative to luminal and HER2-amplified lines. (Invitrogen). Flow cytometric data analysis and visualization were Specifically, 13 of 21 basal-like lines had an EC50 of conducted using FlowJo v8.4 software (Tree Star, Inc.). <1 Amol/L, whereas only 2 of 29 luminal and HER2 lines Breast tumor samples. Primary breast tumors from 29 independent had an EC of <1 Amol/L (P < 0.0001, Fisher’s exact test). breast cancer patients obtained from the Genentech Tumor Bank were 50 used to make mRNA for analysis on Human Genome U133P 2.0 Array PD0325901 has been reported to have exquisite specificity for f chips (Affymetrix). Tumors were assigned to molecular subtypes using MEK1/2 based on profiling against a panel of 80 kinases the method of subtype centroids described by Hu et al. (21). In brief, (besides MEK1/2, only MKK5 was substantially inhibited each tumor was assigned to the subtype whose gene expression centroid starting at a concentration of 2 Amol/L; ref. 25); thus, EC50 had the highest Spearman correlation to the expression profile of the values <1 Amol/L are likely to reflect on-target effects of tumor (22). inhibiting MEK1/2. In addition, we note that basal-like lines Somatic mutation data in breast cell lines and tumors. Exonic regions such as CAL85-1 typically show 50% to 60% inhibition in a were amplified by PCR of genomic DNA and sequenced in both 3-day viability assay (Supplementary Fig. S1), thus showing a directions using Big Dye Terminator Kit reagents and ABI PRISM 3730xl similar magnitude of inhibitory response to that described for sequencing machines (Applied Biosystems). Sequence variants were RAS-mutant cell lines and somewhat less than the dramatic identified using Mutation Surveyor software (SoftGenetics) and confirmed by manual trace review and a second round of sequencing. hypersensitivity and 90% to 95% inhibition described for DNA from all cell lines, Oncotest tumors, and breast tumor samples BRAF-mutant cell lines (26). Of the two non–basal-like lines were analyzed for PIK3CA, PTEN, BRAF, KRAS, HRAS, NRAS, and that showed submicromolar sensitivity, the HER2-amplified RRAS2 sequence alterations. line HCC1954 is reported to have an overall basal-like gene Gene expression microarray studies. Gene expression analysis of expression profile (27), in contrast to the majority of HER2- breast cancer cell lines was carried out on RNA extracted from amplified cell lines, which cluster with luminal cell lines in subconfluent cell cultures using Qiagen RNAeasy kits. RNA quality unsupervised analyses. The luminal cell line MDA-MB-175VII was verified by running samples on an Agilent Bioanalyzer 2100, and has been reported to have an autocrine growth loop mediated samples of sufficient quality were profiled on Affymetrix HGU133- by g-heregulin that results in constitutive signaling through the Plus_2.0 chips. Preparation of complementary RNA, array hybrid- MAPK pathway (28). To show that this selective growth izations, scanning, and subsequent array image data analysis were done using the manufacturer’s specified protocol. Detailed descriptions of inhibition of basal-like cell lines is not a compound-specific the bioinformatic analyses described are presented in Supplementary or off-target effect of PD0325901, we determined that a second Experimental Procedures. All gene expression data described in this MEK inhibitor (9) also had significant antiproliferative effects study have been deposited in the Gene Expression Omnibus database in several basal-like cell lines but not in luminal or HER2- under accession no. GSE12790. amplified cell lines (Supplementary Fig. S1B). We also Identification of activated RAS and MEK signatures. We developed confirmed PD0325901 inhibition of proliferation as deter- pathway activation signatures for HRAS (G12V) and MEK (S217E221E) mined by reduced bromodeoxyuridine incorporation in an using a variation of the strategy used by Bild et al. (23). Detailed immunofluorescence assay for a subset of lines and saw more descriptions of the protocols and statistical analyses are provided in dramatic effects in basal-like than luminal cell lines in this assay Supplementary Experimental Procedures. Recombinant adenoviruses format as well (Supplementary Fig. S1C and data not shown). expressing GFP, HRAS (G12V), and MEK1 (S217E S221E) transgenes, as well as null control vectors, were used at optimal multiplicity of In addition, we determined PTEN and EGFR protein expression infection. Expression on transfection for MEK1(gf) and HRAS(gf) was for each of the cell lines (Fig. 1A; Supplementary Fig. S2) confirmed by Western blot, and isolated RNA was reversed transcribed because PTEN loss and EGFR overexpression have been to cDNA and then run on Human Genome U133P 2.0 Array chips described as common events in basal-like breast cancer (6). (Affymetrix). At least five independent replicates were profiled for each Indeed, we found that basal-like cell lines were more likely to expression construct. Differentially expressed genes between cells harbor loss of PTEN protein and express detectable EGFR than infected with control vector and MEK(gf) or RAS(gf) vectors were luminal or HER2-amplified cell lines (Fisher’s exact test, P < identified with the Cyber-T algorithm (24). Pathway activation 0.001 for EGFR, P < 0.01 for PTEN). The majority of the cell signatures were developed using training and test data sets as described lines exhibiting loss of PTEN protein expression do not harbor in detail in Supplementary Experimental Procedures, and predicted coding sequence mutations in the PTEN gene (14), consistent pathway activation levels derived from the model for each cell line are shown in Supplementary Table S1. with previous studies suggesting that PTEN loss in breast cancer is primarily due to promoter methylation, loss of heterozygos- ity, or regulation at the RNA or protein level (29). Interestingly, Results we found that within the basal-like lines, PTEN protein loss was associated with reduced response to MEK inhibition because In vitro sensitivity to MEK inhibition in breast cancer cell 10 of 11 lines with intact PTEN had EC50 values of <1 Amol/L, lines. We screened 50 breast cancer cell lines for in vitro whereas only 3 of 10 lines with deleted PTEN had EC50 values sensitivity to the selective MEK inhibitor PD0325901. Cell lines of <1 Amol/L (P = 0.022, Fisher’s exact test). We also used were classified into molecular subtypes (basal-like, HER2, or quantitative reverse-phase protein microarrays to assess levels luminal) based on unsupervised hierarchical clustering using of pAkt (S473) across the cell line panel and observed gene expression data as well as HER2 copy number determined significantly higher pAkt (S473) in basal-like cell lines resistant by quantitative reverse transcription-PCR on genomic DNA, as to MEK inhibition (P < 0.001, Student’s t test; Fig. 1A). This is described previously (11). Figure 1A shows the in vitro consistent with recent findings that loss of PTEN is strongly sensitivity to MEK inhibition for each cell line as determined associated with high levels of pAkt in breast cancer cell by a standard ATP-based viability assay. Sample cell viability lines and tumors (30), and suggests a model in which assay curves are shown in Supplementary Fig. S1A. Cell lines activation of Akt signaling via PTEN loss may desensitize a

www.aacrjournals.org 4655 Clin Cancer Res 2009;15(14) July 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. Cancer Therapy: Preclinical

Fig. 5. Effects of dual treatment with MEK and PI3K inhibitors on pathway signaling and cell viability in basal-like breast cancer. A, Western blot analysis of ERK and Akt phosphorylation and cyclin D1levels in the presence of single agent MEK inhibitor, single agent PI3K inhibitor, or a combination. Compounds were addedat either0.1or1.0 Amol/L as indicated and lysates were prepared after 6 h of treatment. Further quantitation of cyclin D1levels is provided in Supplementary Fig. S9. B, combination of MEK and PI3K inhibitors has synergistic effects on viability in two basal-like cell lines (Hs578t and CAL85-1)but not luminal (MCF-7) or HER2-amplified (JIMT-1) cell lines. Inhibitors were used either singly or in combination in a fixed dose ratio and cell viability was assessed in a luminescence-based viability assay.The y axis indicates viability relative to vehicle-treated control cells. Actual concentrations for each compound are givenin Supplementary Experimental Procedures. A measure of synergy, the combination index (CI), is also shown for the MEK-PI3K inhibitor combination for each cell line. Combination index values of <0.5 generally indicate synergistic effects on cell viability. C, caspase activation in response to MEK, PI3K, or combined inhibition in a panel of cell lines. Strong combination effects are limited to basal-like cell lines. Pathway status as assessed by quantitative protein array analysis of pAkt (S473), PTEN protein status, or oncogenic mutations are indicated below the figure (*, P < 0.05, Student’s t test). subset of basal-like breast cancer cells to MEK inhibition. In (G464V)], and CAL-51 (RRAS2), but not in any of the luminal addition, we sequenced core members of the RAS and RAF or HER2-amplified lines. Mutations in RRAS2 have been family of genes from all of the cell lines (Fig. 1A) and found previously described in CAL-51 (31) and shown to result in mutations in four basal-like cells lines, DU4475 (BRAF), transformation in NIH-3T3 cells (32). Finally, we sequenced Hs578t (HRAS), MDA-MB-231 [KRAS (G12V) and BRAF PIK3CA, the gene encoding the catalytic subunit of PI3K

Clin Cancer Res 2009;15(14) July 15, 2009 4656 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. TargetingMEKandPI3KinBasal-likeBreastCancer

(Fig. 1A), in all of the cell lines and found that PIK3CA directly into nude mice and are reported to have maintained alterations are common in HER2-amplified and luminal cell their typical tumor histology, including a stromal component lines but rare in basal-like breast cancer cell lines, and thus are and vasculature (35), and hence may be somewhat more not likely to play a significant role in MEK inhibitor response in representative of actual patient tumors. We used Sanger basal-like cells. sequencing to determine that these tumors are wild-type at Pathway modulation by MEK inhibition in breast cancer cell the BRAF, HRAS, and KRAS loci and that they express PTEN lines. To determine the cellular consequences of MEK inhibi- protein based on Western blot analysis and have low levels of tion in breast cancer, we performed Western blotting experi- pAkt (data not shown). Daily dosing of the basal-like models ments before and after treatment with two different MAXF-1384 and MAXF-401 with 12.5 mg/kg MEK inhibitor concentrations of PD0325901 on three cell lines that are resulted in significant reduction of tumor growth (75% and sensitive to MEK inhibition (HDQ-P1, CAL85-1, and MDA-MB- 67%, respectively) and near stasis compared with vehicle- 231) and two non–basal-like cell lines that are insensitive to treated animals (Fig. 2C and D). To confirm pathway MEK inhibition (BT474 and MCF-7; Fig. 1B). In all five cases, inhibition in these models, we ran Western blots on end-of- we found that treatment with the MEK inhibitor resulted in study tumors from both vehicle- and MEK inhibitor–treated reduction of pERK, a direct readout of MEK1 and MEK2 activity, animals harboring MAXF-1384 and MAXF-401 tumors (Sup- at both 6 and 24 hours (Fig. 1B). A number of studies have plementary Fig. S4). In both cases, we saw significant down- shown that activation of the MEK/ERK pathway is linked to regulation of pERK and up-regulation of p27 in the tumors expression of cyclin D1 and that this is an important from MEK inhibitor–treated animals compared with vehicle- component of progression through the G1 phase of the cell treated animals, confirming both proximal pathway effects on cycle (33). The MEK/ERK pathway is also thought to regulate the MEK substrate ERK and distal cell cycle effects that can be cell cycle progression and proliferation by degradation of the attributed to MEK inhibition. We were unable to assess cyclin cyclin D1–dependent kinase inhibitor p27 (34). Based on D1 down-regulation in these tumors due to cross-reactivity of these observations, we determined the effects of MEK inhibition the antibody with mouse cyclin D1. BT474 tumors were also on cyclin D1 and p27 expression in breast cancer cell lines. analyzed for pERK and p27 levels and showed substantial Notably, we did not see any substantial differences in the levels reduction of pERK but no change in p27 levels after treatment of the cell cycle regulators p27 or cyclin D1 on treatment in (Supplementary Fig. S4), consistent with the in vitro results insensitive BT474 or MCF-7 cells but observed elevated p27 from Fig. 1, and again suggesting that pERK knockdown levels and decreased cyclin D1 levels in all three responsive cell may be necessary but is not sufficient for efficacy in response in vivo lines (Fig. 1B). In addition, we observed G1 arrest in sensitive to MEK inhibition. Altogether, these studies support HDQ-P1 cells treated with MEK inhibitor (Supplementary the in vitro findings that MEK inhibition has in antitumor Fig. S1D), suggesting that the antiproliferative mechanism of activity specifically in breast cancers with a basal-like pheno- action is likely similar to that reported in BRAF- and KRAS- type, including explants derived directly from patients with mutant models. Together these results suggest that pERK basal-like breast cancer. knockdown is not sufficient for efficacy in response to a MEK Gene expression markers of response to MEK inhibition in inhibitor, that alterations in the abundance of cell cycle basal-like breast cancer. We set out to understand the regulators may serve as early pharmacodynamic indicators of underlying mechanistic basis for sensitivity of basal-like breast efficacy, and that the mechanism of action of MEK inhibitors in cancers to MEK inhibition through analysis of gene expression basal-like breast cancer may involve cyclin D1 down-regulation signatures. It is notable that only 4 of the 15 breast cancer cell and accumulation of the negative regulator p27. lines that are sensitive to MEK inhibition in vitro harbor known MEK inhibitor activity in breast cancer xenograft models. To mutations in RAS or RAF family members, suggesting that other confirm that the sensitivity to MEK inhibitors in basal-like factors must contribute to sensitize these cells to respond to breast cancer cell lines is not specific to in vitro growth in MEK inhibition. We first took an unbiased approach to shed monolayer culture, we sought in vivo confirmation of this light on these other factors by examining genes that are finding. We evaluated antitumor activity of PD0325901 in differentially expressed between sensitive and resistant cell lines flank-implanted xenografts of the in vitro sensitive cell line (Supplementary Table S1; Fig. 3A). Notable genes identified MDA-MB-231 (basal-like), as well as the in vitro resistant cell through this approach include the receptor tyrosine kinases line BT474 (HER2 amplified). As shown in Fig. 2A, daily dosing fibroblast growth factor receptor 2 and c-Met and the negative for 21 consecutive days in MDA-MB-231 xenograft mice regulator of RAS signaling, Sprouty2 (36). We also used gene set resulted in significant reduction of tumor size relative to enrichment analysis on a collection of more than 1,000 vehicle-treated controls at all four doses examined, ranging publicly available gene sets to identify biological modules that from 68% inhibition at a dose of 2.5 mg/kg to 84% inhibition were enriched among the differentially expressed genes (37). at a dose of 25 mg/kg. In comparison, daily dosing of mice The 10 most significant gene sets are listed in Supplementary harboring xenografts of the HER2-amplified line BT474 at Table S2. The top gene set (P < 0.0001) was the Broad Institute/ levels up to 25 mg/kg did not show any significant inhibition of MIT ‘‘Breast Cancer ER negative’’ data set of 475 genes whose tumor growth relative to vehicle (Fig. 2B). We also sought expression is consistently correlated with ER status (37). This confirmation of MEK inhibitor activity in primary tumor finding is consistent with the observation that basal-like, explant models that we determined have basal-like breast hormone receptor–negative cell lines are the most responsive cancer characteristics such as lack of expression of ER, PR, and to MEK inhibition. Interestingly, the second gene set identified HER2 and high expression of vimentin, EGFR, and other basal- (P < 0.0001) was a collection of genes whose expression has like markers (Supplementary Fig. S3). These tumor models are been shown to be induced by HRAS expression in IMR90 derived from patient tumors that have been transplanted s.c. fibroblasts (38). Together these results reinforce the association

www.aacrjournals.org 4657 Clin Cancer Res 2009;15(14) July 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. Cancer Therapy: Preclinical

Fig. 6. Antitumor effects of combined MEK and PI3K inhibition in three-dimensional cultures and basal-like in vivo models. A, images of basal-like MDA-MB-231cells and HER2-amplified BT474 cells on laminin-rich, reconstituted basement membrane in the presence of MEK inhibitor, PI3K inhibitor, or a combination. B, quantitation of three-dimensional culture cell viability in triplicate wells using an ATP-based cell viability assay. A strong combination effect on cell viability is seen in MDA-MB-231basal-like cells but not in HER2-amplified BT474M1cells (*, P < 0.005, Student’s t test). C, combined MEK and PI3K inhibitor 28-d daily dosing in flank-xenografted MDA-MB-231 basal-like cells showed superior antitumor activity to dosing with either single agent (**, P < 0.0001, Student’s t test). Points, mean tumor volume over time of 10 treated animals; bars, SD. Percent tumor growth inhibition is shown to the right of each treatment group and values >100% indicate tumor regression. D, combined MEK and PI3K inhibitor daily dosing in flank-xenografted MX-1PTEN-null basal-like cells showed superior antitumor activity to dosing with either single agent (**, P < 0.0001, Student’s t test). A single dose level of PI3K inhibitor (100 mg/kg) was administered either alone or in combination with two different doses of MEK inhibitor (1.5 or 3 mg/kg). between subtype and sensitivity and also suggest that basal-like expression data was first used to identify genes that showed tumors have similarities at the transcriptional level to tumor differential expression between control virus–infected cells and models harboring activated RAS mutations and that this may RAS(gf)- or MEK(gf)-infected cells. The list of the top 50 sensitize them to respond to MEK inhibition. differentially expressed genes is shown in Supplementary To further explore the possibility that basal-like breast Table S3. Notably, this list includes several members of the cancers tend to have activated RAS/MEK pathway signaling cyclin family (CCNE1, CCNE2, and CCND2) and also includes and that this activity leads to MEK inhibitor sensitivity, we used genes that have been linked to MEK signaling in other contexts. a variation of the method described by Bild et al. (23) to For instance, a-basic-crystallin (CRYAB) has been shown identify genomic predictors of oncogenic pathway activation. previously to be expressed in basal-like breast carcinomas and The strategy was to independently identify expression signa- to predict shorter survival, as well as to mediate transformation tures of RAS and MEK activation and use those signatures to in a MEK-dependent manner (41). To ascertain the relevance of assess the status of RAS/MEK pathway activity in a given cell this signature to the biology of breast cancer, we used probes line or tumor. As described in detail in Supplementary for the top 50 differentially expressed genes to perform Experimental Procedures, we expressed gain of function (gf) unsupervised hierarchical clustering on 29 breast tumor versions of HRAS (39) or MEK1 (40) in the immortalized but samples. We found that the genes that constitute the RAS/ nontransformed cell line MCF-10A. Supervised analysis of gene MEK(gf) signature from MCF-10A cells are coordinately

Clin Cancer Res 2009;15(14) July 15, 2009 4658 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. TargetingMEKandPI3KinBasal-likeBreastCancer expressed in tumor samples and that significant up-regulation signaling and MEK activation, we performed RNA interference of RAS and MEK target genes is preferentially seen in tumors studies comparing the effect of EGFR knockdown with that of independently classified as basal-like (Fig. 3B). EGFR-targeting siRNAs on sensitivity to MEK inhibition in We next used Affymetrix microarray data from control cells breast cancer cell lines of differing subtypes and genetic and from RAS(gf) or MEK(gf) cells to train a predictive model backgrounds. To show that the results obtained with the siRNA that could distinguish high from low RAS/MEK pathway pools were not due to off-target effects of the siRNA mix, we activation. We applied this model to predict pathway activity showed similar effects with two individual siRNA duplexes and in our independent collection of breast cancer cell lines also determined that transcript levels were reduced by 80% to (Supplementary Table S4). This enabled us to examine the 85% on siRNA transfection (Supplementary Fig. S6). CAL85-1 relationships between predicted pathway activity, subtype, and express high levels of EGFR and are responsive to erlotinib, and sensitivity to MEK inhibition. To see whether these relation- RNA interference–mediated knockdown of EGFR resulted in ships extended beyond cell lines with known RAS and RAF reduced cell viability even in the absence of MEK inhibition in activating mutations, we excluded those lines from this these cells (Fig. 4C). Importantly and perhaps due to the partial analysis. As shown in Fig. 3C, basal-like cell lines without knockdown observed, the EGFR siRNAs only reduced cell RAS/RAF mutations are predicted to have significantly higher viability by 50%, allowing us to determine whether MEK probability of RAS/MEK pathway activation compared with inhibition could enhance the effects of reduction of EGFR luminal cell lines (P = 0.0002, two-sided Wilcoxon rank sum function. Notably, we found that both 0.1 and 1.0 Amol/L MEK test). In addition, a key finding from these studies is that cell inhibitor treatment resulted in significantly reduced cell lines with submicromolar sensitivity to MEK inhibitor are viability compared with EGFR siRNA alone (e.g., 80% predicted to have elevated predicted RAS/MEK pathway reduction for EGFR siRNA and 0.1 Amol/L MEK inhibitor activation (P = 0.0002, two-sided Wilcoxon rank sum test; compared with a 50% reduction for EGFR siRNA alone). In Fig. 3C). Because we excluded cell lines harboring known RAS contrast, analogous studies in the KRAS-mutant cell line MDA- and RAF mutations from these analyses, these results offer MB-231 and the HER2-amplified cell line JIMT-1 revealed that independent confirmation that basal-like cell lines have EGFR knockdown alone had little or no effect on cell viability characteristics of RAS/MEK activation despite not harboring and did not enhance the inhibitory effects of MEK inhibition known oncogenic mutations, and that gene expression–based on cell viability in these cell lines (Fig. 4C), suggesting a lack of models of RAS/MEK activation may have utility in predicting dependency on EGFR signaling in these cell lines. These results sensitivity to MEK inhibitors. are consistent with a model in which a subset of basal-like cell Role of EGFR in MEK inhibitor sensitivity. One possible lines without known oncogenic mutations (e.g., HDQ-P1, explanation for the presence of an activated RAS signature in HCC1806, and CAL85-1) are dependent on high levels of EGFR basal-like cancers is that constitutively high expression of EGFR signaling through RAS and MEK and thus are sensitive to or other receptor tyrosine kinases in basal-like lines leads to inhibition at key nodes in this pathway. However, the relative elevated signaling through the RAS/RAF/MEK/MAPK pathway lack of potency of erlotinib compared with MEK inhibition, and renders these cells dependent on the pathway and, thus, coupled with the fact that not all basal-like cell lines express sensitive to MEK inhibition. Consistent with this hypothesis, EGFR, suggests a model in which signaling via other upstream we found that our predictions of RAS/MEK pathway activity in receptors may also stimulate RAS activity and activate the MEK breast cancer cell lines were highly correlated with those from pathway. a previously published EGFR signature (ref. 6; P < 0.0001, PTEN knockdown reduces the effects of a MEK inhibitor on cell Fig. 4A). To test the functional significance of EGFR signaling in viability and cell cycle progression. Given the association breast cancer, we screened our breast cancer panel for in vitro observed between lack of PTEN protein and reduced sensitivity sensitivity to the selective EGFR inhibitor erlotinib (Tarceva, to MEK inhibition in the cell line screen, we sought to Genentech) and observed that basal-like cell lines are more determine whether ablation of PTEN by RNA interference sensitive to EGFR inhibition than are luminal cell lines (P < would alter the inhibitory effects of MEK inhibition on cell 0.001, Fisher’s exact test; Fig. 4B). However, only three cell lines viability. We selected one basal-like cell line with a KRAS (HDQ-P1, CAL85-1, and HCC1806) showed submicromolar (G12V) alteration (MDA-MB-231), one basal-like line with no sensitivity to erlotinib, suggesting that the dramatic sensitivity apparent mutational activation of the pathway (CAL85-1), and observed in response to MEK inhibition likely cannot be fully a control basal-like cell line that lacks expression of PTEN explained by EGFR signaling through the RAS pathway. HER2- mRNA and protein (MDA-MB-436). We then performed RNA amplified cell lines also showed sensitivity to erlotinib in the 2 interference–mediated knockdown of PTEN in all of these cells to 5 mmol/L range, although this may possibly be due to off- followed by treatment with two concentrations of MEK target effects on HER2 because the majority of these cell lines inhibitor (Fig. 4D). Knockdown of PTEN with two independent do not express EGFR (Supplementary Fig. S2) and it has been siRNA duplexes at the mRNA level was confirmed by shown that erlotinib can directly inhibit HER2 kinase activation quantitative reverse transcription-PCR (Supplementary in cells lacking EGFR expression (42). To evaluate whether Fig. S6). In both MDA-MB-231 and CAL85-1 cells, PTEN EGFR signals through MEK and ERK in sensitive basal-like cells, knockdown resulted in an increase in cell viability compared we treated cells for 6 hours with erlotinib and assessed with control cells in the absence of treatment with the MEK phospho-ERK levels before and after treatment. Sensitive inhibitor, consistent with the role of PTEN as a tumor CAL85-1 cells showed a dramatic reduction in pERK, whereas suppressor gene. In addition, PTEN knockdown significantly MDA-MB-231 cells that harbor a mutation in the downstream blunted the inhibitory effects of MEK inhibition on cell effector KRAS showed no reduction in pERK (Supplementary viability in these cell lines. Specifically, cells with PTEN Fig. S5). To further test the functional link between EGFR knockdown treated with 1 Amol/L MEK inhibitor showed, on

www.aacrjournals.org 4659 Clin Cancer Res 2009;15(14) July 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. Cancer Therapy: Preclinical average, 60% viability compared with 40% in control siRNA PI3K and MEK inhibitors resulted in a highly synergistic cells for CAL85-1 and 44% compared with 24% for MDA-MB- reduction in cell viability in both cell lines with a combination 231 cells (Fig. 4D). As a control for specificity of the siRNA index of <0.2, manifested as a significantly lower combination duplexes to PTEN, we performed siRNA knockdown in a PTEN- EC50 as well as greater inhibition of cell viability across a range null line and showed no effect on cell viability relative to cells of concentrations (Fig. 5B). In contrast, the luminal cell line transfected with a control siRNA, suggesting that the observed MCF-7 and the HER2-amplified cell line JIMT-1 both showed effects are due to PTEN knockdown. To address specific cell substantially reduced viability in response to the PI3K cycle effects of PTEN ablation in RAS-driven cancers, we inhibitor, no effect in response to the MEK inhibitor, and no conducted flow cytometric DNA content analyses in MDA- synergistic or additive effects in response to the combination of MB-231 KRAS-mutant cells. We compared cell cycle profiles compounds (Fig. 5B). Both JIMT-1 cells and MCF-7 cells show before and after treatment with 1 Amol/L MEK inhibitor and strongly reduced cyclin D1 protein levels in response to PI3K found that PTEN siRNA significantly reduced G1 arrest induced inhibition, modest reduction in response to MEK inhibition, by the MEK inhibitor (Supplementary Fig. S7) from 73.9% in and slight enhancement with the combination (Supplementary control siRNA–treated cells to 56.9% in cells transfected with a Fig. S10). Thus, cyclin D1 levels seem to be more tightly linked PTEN-targeting siRNA. This result suggests that loss of PTEN to MEK signaling in basal-like lines and to PI3K signaling in renders RAS-driven cancer cells relatively less responsive to the luminal and HER2-amplified lines. inhibitory effects of MEK inhibition and may do so by To further explore the nature of this apparent synergy, we decreasing the fraction of cells undergoing G1 arrest. investigated the effects of combination MEK and PI3K inhibitor Effects of combined MEK and PI3K inhibition on cell viability treatment on induction of apoptosis in cell lines of differing and apoptosis. Because activation of the PI3K pathway by genetic backgrounds and levels of activation of the RAS and PTEN loss reduces the effects of MEK inhibition in cell lines, we PI3K/Akt pathways. First, we used a TUNEL assay to determine sought to investigate whether dual inhibition of both MEK and induction of apoptosis in response to single-agent and PI3K might result in synergistic effects on cell growth inhibition combination treatment in a BRAF-mutant cell line (DU4475), or apoptosis induction in breast cancer cells. To inhibit PI3K, basal-like cell lines with low (CAL85-1 and MDA-MB-231) and we used the selective PI3K inhibitor GDC-0941, a novel high activity of the Akt pathway (Hs578t and BT549), and a compound with low nanomolar IC50s against all four isoforms luminal cell line (MCF-7; Supplementary Fig. S10). DU4475 of class I PI3K that is currently in phase I clinical development cells showed strong induction of apoptosis in response to single (43). We first examined the effects of single-agent MEK or PI3K agent MEK inhibitor and this effect was not significantly inhibition as compared with combination treatment on the enhanced by PI3K inhibition, suggesting that in BRAF-mutant state of pathway activation (as determined by phosphorylation models, single-agent MEK inhibition may be enough to trigger of ERK and Akt) in the resistant, high-pAkt cell line Hs578t and apoptosis in the absence of inhibition of the PI3K pathway, as the sensitive, low-pAkt cell line CAL85-1. Intriguingly, we has been reported previously (10). In addition, we found that found that MEK inhibitor treatment alone not only reduced basal-like cells with both low and high levels of Akt signaling ERK phosphorylation but also markedly increased pAkt levels showed enhanced induction of apoptosis in response to the both cell lines (Fig. 5A), consistent with the existence of a combination, although the effect was modest in BT549 cells. In feedback loop whereby inhibition of the MEK/ERK pathway luminal MCF-7 cells, we did not see significant induction of results in concomitant up-regulation of signaling through the TUNEL staining in response to either single agent or the PI3K/Akt pathway. To confirm that increased pAkt in response combination despite the fact that this cell line harbors a to MEK inhibition was not unique to these cell lines, we mutation in the PIK3CA gene. We confirmed and extended assessed pAkt (S473) levels in response to 6-hour MEK these findings in additional cell lines using a cell-based assay inhibitor treatment in a range of cell lines of differing subtypes for the presence of cleaved caspase-3 and caspase-7, a hallmark on reverse-phase protein arrays, and found that all of the cell of activation of apoptotic signaling pathways (Fig. 5C). We lines showed elevated levels of pAkt (S473) compared with found that three of four basal-like lines with low pAkt showed control-treated cells (Supplementary Fig. S8). In contrast, GDC- significant enhancement of apoptosis in the combination 0941 treatment resulted in diminished pAkt levels but did not treatment when compared with either single agent. In addition, appreciably affect pERK levels in these cell lines. Notably, the three of four basal-like cell lines with high pAkt (including two combination of PI3K and MEK inhibitors resulted in near- PTEN-null cell lines) also showed significant enhancement of complete blockade of both pERK and pAkt. In addition, we apoptosis relative to either single agent. CAL-148 (PTEN null, examined the effects of the treatments on cyclin D1 levels and PIK3CA mutant) cells did not show enhanced effects with the found that in both cell lines, treatment with 1 Amol/L MEK combination but showed significant induction of apoptosis in inhibitor resulted in substantial reduction in cyclin D1 levels, response to single-agent PI3K inhibition. In contrast, none of treatment with 1 Amol/L PI3K inhibitor had a modest effect, the HER2-amplified and luminal cell lines showed significant and cotreatment with the two agents resulted in a more enhancement of apoptosis with the combination treatment. dramatic reduction in cyclin D1 levels (Fig. 5A; Supplementary Taken together, these results suggest that basal-like lines with Fig. S10). This effect was particularly strong in CAL85-1 cells. both high and low levels of Akt pathway activation, but not We next examined the consequences of dual inhibition of HER2-amplified or luminal cell lines (which have low RAS these pathways on cellular viability in these four cell lines using pathway activation), have the potential for enhanced apoptotic the Chou and Talalay fixed dose ratio method of assessing the response to combined targeting of MEK and PI3K. combined effects of compounds on cell viability (18). In the Combined MEK and PI3K inhibition shows superior efficacy in basal-like cell lines Hs578t and CAL85-1, each compound basal-like models in three-dimensional cultures and xenograft alone had a significant effect on viability; the combination of models. Our in vitro studies suggest a role for rational

Clin Cancer Res 2009;15(14) July 15, 2009 4660 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. TargetingMEKandPI3KinBasal-likeBreastCancer combination of MEK and PI3K inhibitors in basal-like tumor with intact PTEN, consistent with the hypothesis that PTEN is a models with both high and low levels of basal Akt signaling. To negative predictive marker for MEK inhibitor response. In extend these findings and ensure they were not artifacts of addition, both combination arms showed significantly greater cellular growth in monolayer cultures on plastic, we explored reduction in tumor growth (93% and 98% for the 1.5 and 3 the effects of combined targeting of MEK and PI3K in three- mg/kg groups, respectively) than the single-agent treatment dimensional cultures of cell lines of different subtypes. Growth groups, where the best observed response was 60% tumor of mammary epithelial cells in three-dimensional cultures is growth inhibition with 3 mg/kg MEK inhibitor (P < 0.0001, thought to more closely recapitulate tumor biology in that that Student’s t test). In addition, tumor growth was monitored cells can organize into structures that more closely resemble following cessation of treatment and durable responses were in vivo tumor architecture (44), and thus may represent a useful observed for the combination treatment group given 3 mg/kg surrogate for investigating physiologic and pharmacologic MEK inhibitor and 100 mg/kg PI3K inhibitor (30% survival, effects on tumor biology. We grew basal-like MDA-MB-231 10% complete regression), but not for other cohorts, at day 40. cells and HER2-amplified BT474M1 cells on laminin-rich, Given that a concern with combining therapeutics that target reconstituted basement membrane under conditions in which these fundamental signaling pathways is the occurrence of they proliferated rapidly and formed solid acinar structures overlapping toxicities, it is encouraging to note that in this (Fig. 6A), then compared the effects of single agent inhibitors of experiment, the combination was well tolerated and resulted in MEK and PI3K or a combination of the two on cell viability minimal weight loss (Supplementary Fig. S11). Together, these under these growth conditions. In MDA-MB-231 cells, we studies provide strong preclinical support that basal-like breast observed that MEK inhibition potently inhibited cell prolifer- cancer may provide an opportunity for rational combination ation both visually and based on an ATP-based viability assay; therapy with MEK and PI3K inhibitors based on the knowledge PI3K inhibition had modest effects; and the combination was of underlying pathway activation status. significantly more potent than either single agent. In contrast, BT474M1 cells showed minimal loss of viability in response to Discussion MEK inhibition, potent killing by the PI3K inhibitor, but no enhancement over PI3K inhibition alone when the drugs were A key finding of this study is that MEK inhibition selectively administered in combination (Fig. 6B). Thus, these results are affects the viability of basal-like cell lines and primary tumor consistent with a model in which basal-like cell lines with high models compared with other breast cancer subtypes. We have RAS signaling and the potential to up-regulate Akt signaling in also shown that basal-like cell lines can be further stratified into response to MEK inhibition show enhanced response to the a subset with low levels of Akt pathway signaling and intact combination, but HER2-amplified and luminal cell lines lack PTEN and a distinct subset with high levels of Akt pathway dependence on the RAS pathway and hence may show good activation and loss of PTEN, with the latter subset being, on response to PI3K inhibition but no additional benefit from average, less responsive to single-agent MEK inhibition. concomitant MEK inhibition. We extended this finding from a Previous studies have identified BRAF and KRAS as putative three-dimensional culture to a true in vivo setting by challeng- predictive biomarkers for stratifying patients for treatment with ing flank-implanted tumor xenografts of the basal-like KRAS- selective inhibitors of MEK (10). Our results extend these mutant cell line MDA-MB-231 with single agent PD0325901, findings by showing that basal-like breast tumors have single agent GDC-0941, or a combination of the two agents. similarities at the transcriptional level with RAS-mutant tumors Consistent with the in vitro studies, we observed modest despite an apparent lack of oncogenic mutations and that reduction of tumor growth with GDC-0941 (43%), substantial patients with such tumors may also benefit from treatment with inhibition (97%) in the presence of PD0325901, and a a MEK inhibitor. statistically significant increase in tumor growth inhibition Our studies suggest a model in which basal-like cell lines are (104%) in the combination arm and noted that the average preferentially sensitive to MEK inhibition due to activation and animal showed slight tumor regression in the combination addiction to the RAS pathway specifically in this subtype. KRAS, treatment arm (Fig. 6C). RRAS2, HRAS, and BRAF mutations were associated with A second model that emerges from our in vitro studies is that sensitivity and found only in basal-like breast cancer cell lines, basal-like cell tumors with high levels of Akt signaling driven by but at a relatively low prevalence (4 of 20 basal-like lines), and PTEN loss have dual activation of both RAS and PI3K/Akt thus cannot account for all of the activity seen in basal-like signaling and thus may show enhanced antitumor activity with breast cancer. Mutations in RAS and RAF family members also regimens combining selective MEK and PI3K inhibitors. We occur at a very low frequency in human breast tumors tested this hypothesis in flank-implanted tumor xenografts of according to data in the Sanger COSMIC database (BRAF 2%, the basal-like PTEN-null MX-1 cell line challenged with single KRAS 5%, and HRAS 1%), although an examination of the agent PD0325901, single agent GDC-0941, or a combination occurrence of mutations across different breast cancer subtypes of the two agents. We used a single dose level of GDC-0941 has not been reported. We used several different bioinformatic (100 mg/kg) and two different dose levels of PD0325901 (1.5 approaches that converge on the idea that basal-like breast and 3 mg/kg). We first noted that tumor growth inhibition in cancer cell lines and tumors have an overall gene signature that response to 3 mg/kg 28-day daily dosing in PTEN-null MX-1 is related to other RAS and EGFR signatures, and our functional cells was only 60%, compared with 97% with this same dose studies suggest that, in some cases, RAS activation may be due and schedule in MDA-MB-231 cells with intact PTEN (compare to signaling through EGFR. This finding agrees with previous Fig. 6C and D). Thus, whereas these models have similar in vitro results showing that basal-like breast cancers express high levels sensitivity to MEK inhibition, the PTEN-null line is substan- of EGFR and have an EGFR-like gene expression signature that tially less responsive to in vivo MEK inhibition than the model is predictive of poor prognosis in sporadic tumors (6).

www.aacrjournals.org 4661 Clin Cancer Res 2009;15(14) July 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. Cancer Therapy: Preclinical

However, the overall modest nature of erlotinib response in that PTEN loss does not completely block response, but rather basal-like breast cancer and our specific functional studies shifts cells upward on a response continuum and renders them suggest that EGFR alone likely does not account for all of the less sensitive than they would presumably be without activation MEK inhibitor sensitivity observed. These observations are of the Akt/PI3K pathway. A hypothesis arising from these consistent with recent results from a clinical trial (TBCRC 001) results would be that basal-like PTEN-null tumors would be an in which women with histopathologically confirmed triple- ideal setting for combining inhibitors that target both MEK and negative breast cancer received the EGFR-targeting monoclonal PI3K, and indeed, we observed a strong combination effect in antibody cetuximab plus carboplatin, where it was observed our in vivo experiments with xenografted MX-1 PTEN-null that EGFR targeted showed clinical benefit in a subset of tumors. In addition, our findings may have relevance for patients but was insufficient in the majority of women (45). rational combination of these agents even in cases of basal-like Thus, the exact nature of all of the upstream signals that activate cancers with intact PTEN and low levels of resting Akt/PI3K RAS specifically in basal-like cells remains to be determined. pathway activation because we have found that all cell lines Other candidates would include receptor tyrosine kinases that examined show evidence of feedback up-regulation of the Akt are expressed highly in basal-like cell lines and tumors and pathway in response to MEK inhibitor treatment. This compose part of the MEK sensitivity signature such as c-Met and phenomenon may suggest an explanation for the lack of fibroblast growth factor receptor 2, as well as members of the apoptosis observed in response to single-agent MEK inhibitor SRC family of nonreceptor tyrosine kinases, because previous treatment in these cell lines because activation of the Akt studies have shown that the selective SRC inhibitor dasatinib pathway mediates antiapoptotic signaling (12), as well as preferentially kills basal-like cell lines (46) and that phosphor- provide a basis for observation that in vitro inhibition of basal- ylation of SRC family members is associated with in vitro like cell lines does not approach the extreme hypersensitivity response to a MEK inhibitor (20). Future studies should aim to observed in BRAF-mutant cell lines. Such feedback regulation dissect the relative contribution of different upstream pathways and enhanced Akt phosphorylation in response to the MEK to MEK activation, as well as determine the activating events inhibitor U0126 has been described previously (47) and shown that lead to receptor tyrosine kinase and RAS activation in to involve ERK-driven regulation of the interaction between basal-like breast cancer. Ultimately, however, because RAS may EGFR and Gab1 serving to down-regulate PI3K signaling. Thus, represent a signal integration point for multiple upstream both the presence of constitutive Akt activation due to PTEN receptors, the RAS activation signature may provide a more loss and the feedback up-regulation of pAkt by single-agent generic picture of pathway activation in basal-like cells, and MEK inhibition suggest a rationale for combined targeting of therapeutics targeting key downstream nodes such as MEK may MEK and PI3K in basal-like breast cancer. The synergistic effects have broader activity than agents targeting particular upstream on cell viability in basal-like cell lines may also involve effects receptors. on cell cycle progression as well as enhanced apoptosis because Mechanistically, MEK inhibition in basal-like tumors has we found that cotreatment with both MEK and PI3K inhibitors elements in common with the described mechanism of action resulted in more substantial down-regulation of cyclin D1 than in BRAF-mutant tumors (10) because we find that MEK did either single agent in CAL85-1 basal-like cells that showed inhibition results in G1 arrest, ERK-mediated cyclin D1 down- in vitro synergy. Thus, one facet of the synergistic in vitro effects regulation, and p27 up-regulation in models that show of MEK and PI3K inhibitors may be the convergence of the two sensitivity to the MEK inhibitor. This observation supports a pathways on cyclin D1 regulation. Somewhat surprisingly, hypothesis that these cell cycle events are mediating at least because it might be assumed that combining MEK and PI3K some of the antiproliferative effects observed after MEK inhibitors would always be superior to single-agent treatment inhibition and that MEK and ERK activities are required for based on the fundamental importance of these signaling G1 progression in basal-like breast cancer. These preclinical pathways, our in vitro experiments and three-dimensional findings also have potential utility in the clinical development culture experiments suggest that dramatic combination effects of MEK inhibitors because they suggest that biomarker assays may be limited to cell lines of the basal-like subtype. Luminal for cyclin D1 and p27 could have utility as early indicators of and HER2-amplified cell lines do not have apparent activation drug efficacy and facilitate clinical decision making. It is also and dependence on the RAS pathway or potential to respond to notable that we did not observe induction of apoptosis or MEK inhibition by engaging the cell cycle machinery, and thus activation of cleaved caspase in response to single-agent do not seem to benefit from the addition of a MEK inhibitor to MEK treatment in any of the basal-like cell lines except the a PI3K inhibitor. However, based on the studies reported here BRAF-mutant cell line DU4475, suggesting that the mechanism and by others (48), as well as the frequent occurrence of of action of single-agent MEK inhibition in basal-like breast oncogenic mutations in PIK3CA we observe in cell lines, cancer is predominantly cytostatic and not cytotoxic and may luminal and HER2-amplified tumors may respond well to be mediated by G1 arrest in response to cyclin D1 down- single-agent PI3K inhibition. A recent study suggested that a regulation. combination regimen with MEK and PI3K inhibitors might PTEN loss and an activated Akt pathway constitute an effectively be used to treat KRAS-mutant lung tumors (49), apparent intrinsic resistance mechanism to MEK inhibition in and our studies suggest that such an approach might be basal-like breast cancer because cell lines where PTEN is absent similarly fruitful in basal-like breast cancers with both intact or ablated by siRNA are relatively less sensitive to MEK and deleted PTEN. inhibition than are cells with intact PTEN. Importantly, PTEN Breast cancer has emerged as a model for personalized loss is common not just in the cell line panel but has also been medicine based on molecular tumor characterization because reported to be frequent and relatively subtype specific in basal- clinical management of disease involves routine evaluation of like breast tumors (14). Our PTEN siRNA experiments suggest diagnostics tests before treatment with hormonal therapy or

Clin Cancer Res 2009;15(14) July 15, 2009 4662 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. TargetingMEKandPI3KinBasal-likeBreastCancer

HER2 targeting agents such as trastuzumab (Herceptin) or rational combinations of agents (50). Our results suggest that lapatinib (Tykerb). The results reported here as well as previous the combination of MEK and PI3K inhibitors in basal-like studies (11) suggest that large panels of breast cancer cell lines breast cancer may represent a useful avenue for such future represent much of the genetic and transcriptional diversity seen clinical investigation. in primary tumors and hence may be useful surrogates to develop preclinical hypotheses around molecular determinants of drug response that can then be tested during clinical Disclosure of Potential Conflicts of Interest development. The results reported here show that a selective inhibitor of MEK kinase primarily has activity in the basal-like All authors are either current or former employees of Genentech. subtype of breast cancer and hence provides a strategy for patient stratification during clinical trials for MEK inhibitors. Moreover, clinical studies focusing on single-agent MEK Acknowledgments inhibitor activity have thus far not shown much evidence of clinical benefit resulting from targeting MEK alone (26), and We thank David Dornan, Richard Neve, David Stokoe, David Davis, Ellen Filvaroff, and Elaine Storm for critically reviewing the manuscript, Somasekar Seshagiri, the clinical challenges experienced in the development of MEK Fred de Sauvage, and the Cancer Genome project for mutational analyses inhibitors have led others to suggest that the future develop- and support, and the Genentech Medicinal Chemistry Group for providing ment of MEK inhibitors as anticancer agents may require GDC-0941.

References 1. SorlieT, Perou CM,Tibshirani R, et al. Gene expression TRAIL-induced apoptosis. FEBS Lett 2005;579: cer cell line, MDA-MB-175. Oncogene 1997;15: patterns of breast carcinomas distinguish tumor sub- 4149 ^ 58. 13 8 5 ^ 9 4. classes with clinical implications. Proc Natl Acad Sci 16. Mirzoeva OK, Das D, Heiser LM, et al. Basal subtype 29. Hennessy BT, Smith DL, Ram PT, Lu Y, Mills USA2001;98:10869^74. and MAPK/ERK kinase (MEK)-phosphoinositide 3-ki- GB. Exploiting the PI3K/AKT pathway for cancer 2. CiglerT, Goss PE. Breast cancer adjuvant endocrine nase feedback signaling determine susceptibility of drug discovery. Nat Rev Drug Discov 2005;4: therapy. Cancer J2007;13:148 ^55. breast cancer cells to MEK inhibition. Cancer Res 988^1004. 3. Shepard HM, Jin P, Slamon DJ, Pirot Z, Maneval DC. 2009;69:565 ^ 72. 30. Stemke-Hale K, Gonzalez-Angulo AM, Lluch A, Herceptin. Handb Exp Pharmacol 2008;181:183 ^ 17. Hu XSH, Ge L, O’Brien C, et al. Genetic alterations et al. An integrative genomic and proteomic analysis 219. and oncogenic pathways associated with breast can- of PIK3CA, PTEN, AKT mutations in breast cancer. 4. BidardFC,ConfortiR,BouletT,MichielsS,Delaloge cer subtypes. Mol Cancer Res 2009;7:511^ 22. Cancer Res 2008;68:6084^91. S, Andre F. Does triple-negative phenotype accurately 18. ChouTC,Talalay P. Quantitative analysis of dose-ef- 31. Barker KT, Crompton MR. Ras-related TC21is acti- identify basal-like tumour? An immunohistochemical fect relationships: the combined effects of multiple vated by mutation in a breast cancer cell line, but infre- analysis based on 143 ‘‘triple-negative’’ breast can- drugs or enzyme inhibitors. Adv Enzyme Regul 1984; quently in breast carcinomas in vivo.BrJCancer1998; cers. Ann Oncol 2007;18:1285 ^ 6. 22:27 ^ 55. 78:296 ^ 300. 5. Reis-Filho JS,Tutt AN. Triple negative tumours: a crit- 19. Debnath J, Muthuswamy SK, Brugge JS. Mor- 32. Saez R, Chan AM, MikiT, Aaronson SA. Oncogenic ical review. Histopathology 2008;52:108 ^ 18. phogenesis and oncogenesis of MCF-10A mamma- activation of human R-ras by point mutations analo- 6. Hoadley KA,WeigmanVJ, Fan C, et al. EGFR associ- ry epithelial acini grown in three-dimensional gous to those of prototype H-ras oncogenes. Onco- ated expression profiles vary with breast tumor sub- basement membrane cultures. Methods 2003;30: gene 1994;9:2977^82. type. BMC Genomics 2007;8:258. 256 ^ 68. 33. Roovers K, Assoian RK. Integrating the MAP kinase 7. Crews CM, Alessandrini A, Erikson RL. ERKs: their 20. Boyd ZS, Wu QJ, O’Brien C, et al. Proteomic analy- signal into the G1 phase cell cycle machinery. Bio- fifteen minutes has arrived. Cell Growth Differ 1992; sis of breast cancer molecular subtypes and bio- essays 2000;22:818^ 26. 3:135 ^ 42. markers of response to targeted kinase inhibitors 34. Motti ML, De Marco C, Califano D, et al. Loss of 8. Thompson N, Lyons J. Recent progress in targeting using reverse-phase protein microarrays. Mol Cancer p27 expression through RAS->BRAF->MAP kinase- the Raf/MEK/ERK pathway with inhibitors in cancer Ther 2008;7:3695 ^ 706. dependent pathway in human thyroid carcinomas. Cell drug discovery. Curr Opin Pharmacol 2005;5:350 ^ 6. 21. Hu Z, Fan C, Oh DS, et al.The molecular portraits of Cycle 2007;6:2817^ 25. 9. Adjei AA, Cohen RB, Franklin W, et al. Phase I phar- breast tumors are conserved across microarray plat- 35. Fiebig HH, Schuler J, Bausch N, Hofmann M, Metz macokinetic and pharmacodynamic study of the oral, forms. BMC Genomics 2006;7:96. T, Korrat A. Gene signatures developed from patient small-molecule mitogen-activated protein kinase ki- 22. SorlieT,Tibshirani R, ParkerJ, et al. Repeated obser- tumor explants grown in nude mice to predict tumor nase 1/2 inhibitor AZD6244 (ARRY-142886) in vation of breast tumor subtypes in independent gene response to11cytotoxic drugs. Cancer Genomics Pro- patients with advanced cancers. JClin Oncol 2008; expression data sets. Proc Natl Acad Sci U S A 2003; teomics 2007;4:197 ^209. 26:2139 ^ 46. 100 :8 418 ^ 23. 36. Guy GR, Wong ES, Yusoff P, et al. Sprouty: how 10. Solit DB, Garraway LA, Pratilas CA, et al. BRAF mu- 23. Bild AH,Yao G, Chang JT, et al. Oncogenic pathway does the branch manager work? JCell Sci 2003;116: tation predicts sensitivity to MEK inhibition. Nature signatures in human cancers as a guide to targeted 3061^8. 2006;439:358 ^ 62. therapies. Nature 2006;439:353 ^ 7. 37. Subramanian A, Tamayo P, Mootha VK, et al. 11. O’Brien C, Cavet G, Pandita A, et al. Functional 24. Baldi P, Long AD. A Bayesian framework for the Gene set enrichment analysis: a knowledge-based genomics identifies ABCC3 as a mediator of taxane analysis of microarray expression data: regularized approach for interpreting genome-wide expression resistance in Her2 amplified breast cancer. Cancer t test and statistical inferences of gene changes. Bio- profiles. Proc Natl Acad Sci U S A 2005;102: Res 2008;68:5380^9. informatics 2001;17:509 ^ 19. 155 4 5 ^ 5 0. 12. Cantley LC.The role of phosphoinositide 3-kinase in 25. Bain J, Plater L, Elliott M, et al. The selectivity of 38. Mason DX, Jackson TJ, Lin AW. Molecular signa- human disease. Harvey Lect 2004;100:103 ^22. protein kinase inhibitors: a further update. Biochem J ture of oncogenic ras-induced senescence. Oncogene 13. SaalLH,HolmK,MaurerM,etal.PIK3CAmutations 2007;408:297 ^ 315. 2004;23:9238^46. correlate with hormone receptors, node metastasis, 26. Friday BB, Adjei AA. Advances in targeting the 39. Colby WW, Hayflick JS, Clark SG, Levinson AD. and ERBB2, and are mutually exclusive with PTEN Ras/Raf/MEK/ERK mitogen-activated protein kinase Biochemical characterization of polypeptides loss in human breast carcinoma. Cancer Res 2005; cascade with MEK inhibitors for cancer therapy. encoded by mutated human Ha-ras1genes. Mol Cell 65:2554 ^ 9. Clin Cancer Res 2008;14:342 ^ 6. Biol 1986;6:730 ^ 4. 14. Saal LH, Gruvberger-Saal SK, Persson C, et al. Re- 27. NeveRM,ChinK,FridlyandJ,etal.Acollection 40. Brunet A, Pages G, Pouyssegur J. Constitutively current gross mutations of the PTEN tumor suppres- of breast cancer cell lines for the study of function- active mutants of MAP kinase kinase (MEK1) in- sor gene in breast cancers with deficient DSB repair. ally distinct cancer subtypes. Cancer Cell 2006;10: duce growth factor-relaxation and oncogenicity Nat Genet 2008;40:102^ 7. 515 ^ 27. when expressed in fibroblasts. Oncogene 1994;9: 15. Dubska L, Andera L, Sheard MA. HER2 signaling 28. Schaefer G, Fitzpatrick VD, Sliwkowski MX. 3379^87. downregulation by trastuzumab and suppression of g-Heregulin: a novel heregulin isoform that is an 41. Moyano JV, Evans JR, Chen F, et al. aB-crystallin the PI3K/Akt pathway: an unexpected effect on autocrine growth factor for the human breast can- is a novel oncoprotein that predicts poor clinical

www.aacrjournals.org 4663 Clin Cancer Res 2009;15(14) July 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. Cancer Therapy: Preclinical

outcome in breast cancer. JClin Invest 2006;116: 44. Weaver VM, Fischer AH, Peterson OW, Bissell MJ. the epidermal growth factor-mediated interaction of 261 ^ 70. The importance of the microenvironment in breast Gab1 and the phosphatidylinositol 3-kinase. JBiol 42. Schaefer G, Shao L,Totpal K, Akita RW. Erlotinib di- cancer progression: recapitulation of mammary tu- Chem 2002;277:19382^ 8. rectly inhibits HER2 kinase activation and down- morigenesis using a unique human mammary epitheli- 48. Torbett NE, Luna-Moran A, Knight ZA, et al. A stream signaling events in intact cells lacking al cell model and a three-dimensional culture assay. chemical screen in diverse breast cancer cell lines epidermal growth factor receptor expression. Cancer Biochem Cell Biol 1996;74:833^51. reveals genetic enhancers and suppressors of sensi- Res 2007;67:1228^ 38. 45. Schneider BP, Winer EP, Foulkes WD, et al. tivity to PI3K isoform-selective inhibition. Biochem J 43. FolkesAJ,AhmadiK,AldertonWK,etal.The Triple-negative breast cancer: risk factors to poten- 2008;415:97 ^ 110. identification of 2-(1H-indazol-4-yl)-6-(4-methane- tial targets. Clin Cancer Res 2008;14:8010^ 8. 49. Engelman JA, Chen L, Tan X, et al. Effective use of sulfonyl-piperazin-1-ylmethyl)-4-morpholin-4-yl- 46. Huang F, Reeves K, Han X, et al. Identification of PI3K and MEK inhibitors to treat mutant Kras G12D thieno[3,2-d]pyrimidine (GDC-0941) as a potent, candidate molecular markers predicting sensitivity in and PIK3CA H1047R murine lung cancers. Nat Med selective, orally bioavailable inhibitor of class I PI3 ki- solid tumors to dasatinib: rationale for patient selec- 2008;14:1351 ^ 6. nase for the treatment of cancer. JMed Chem 2008; tion. Cancer Res 2007;67:2226 ^ 38. 50. Price S. Putative allosteric MEK1and MEK2 inhibi- 51:5522^ 32. 47. YuCF,LiuZX,CantleyLG.ERKnegativelyregulates tors. Exp OpinTher Patents 2008;18:603 ^ 27.

Clin Cancer Res 2009;15(14) July 15, 2009 4664 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research. In vivo Antitumor Activity of MEK and Phosphatidylinositol 3-Kinase Inhibitors in Basal-Like Breast Cancer Models

Klaus P. Hoeflich, Carol O'Brien, Zachary Boyd, et al.

Clin Cancer Res 2009;15:4649-4664.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/15/14/4649

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2009/07/21/1078-0432.CCR-09-0317.DC2

http://clincancerres.aacrjournals.org/content/suppl/2009/06/16/1078-0432.CCR-09-0317.DC1

Cited articles This article cites 50 articles, 21 of which you can access for free at: http://clincancerres.aacrjournals.org/content/15/14/4649.full#ref-list-1

Citing articles This article has been cited by 86 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/15/14/4649.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/15/14/4649. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2009 American Association for Cancer Research.