A Targeted Rnai Screen of the Breast Cancer Genome Identifies KIF14 and TLN1 As Genes That Modulate Docetaxel Chemosensitivity in Triple-Negative Breast Cancer

Published OnlineFirst March 11, 2013; DOI: 10.1158/1078-0432.CCR-13-0082

Clinical Cancer Cancer Therapy: Preclinical Research

A Targeted RNAi Screen of the Breast Cancer Genome Identiﬁes KIF14 and TLN1 as Genes That Modulate Docetaxel Chemosensitivity in Triple-Negative Breast Cancer

Stina Mui Singel1,2, Crystal Cornelius1, Kimberly Batten1, Gail Fasciani1, Woodring E. Wright1, Lawrence Lum1, and Jerry W. Shay1

Abstract Purpose: To identify biomarkers within the breast cancer genome that may predict chemosensitivity in breast cancer. Experimental Design: We conducted an RNA interference (RNAi) screen within the breast cancer genome for genes whose loss-of-function enhanced docetaxel chemosensitivity in an estrogen receptor– negative, progesterone receptor–negative, and Her2-negative (ER,PR, and Her2, respectively) breast cancer cell line, MDA-MB-231. Top candidates were tested for their ability to modulate chemosensitivity in 8 breast cancer cell lines and to show in vivo chemosensitivity in a mouse xenograft model. Results: From ranking chemosensitivity of 328 short hairpin RNA (shRNA) MDA-MB-231 cell lines (targeting 133 genes with known somatic mutations in breast cancer), we focused on the top two genes, kinesin family member 14 (KIF14) and talin 1 (TLN1). KIF14 and TLN1 loss-of-function significantly enhanced chemosensitivity in four triple-negative breast cancer (TNBC) cell lines (MDA-MB-231, HCC38, HCC1937, and Hs478T) but not in three hormone receptor–positive cell lines (MCF7, T47D, and HCC1428) or normal human mammary epithelial cells (HMEC). Decreased expression of KIF14, but not TLN1, also enhanced docetaxel sensitivity in a Her2-amplified breast cancer cell line, SUM190PT. Higher KIF14 and TLN1 expressions are found in TNBCs compared with the other clinical subtypes. Mammary fat pad xenografts of KIF14- and TLN1-deficient MDA-MB-231 cells revealed reduced tumor mass compared with control MDA-MB-231 cells after chemotherapy. KIF14 expression is also prognostic of relapse-free and overall survival in representative breast cancer expression arrays. Conclusion: KIF14 and TLN1 are modulators of response to docetaxel and potential therapeutic targets in TNBC. Clin Cancer Res; 19(8); 2061–70. 2013 AACR.

Introduction may benefit from specific treatment regimens. Furthermore, Expression array analyses in breast cancer have revealed besides sequence mutations, there are numerous chromo- multiple subtypes of breast cancer, each with distinct clin- somal alterations, copy number variations, miRNA dysre- ical prognosis and response to treatments (1–4). Every gulations, and epigenetic events that are frequently found in tumor acquires a complex combination of somatic muta- human cancers (10–12). Successful therapy depends on the tions that contribute to the cancer phenotype. Large-scale identification of critical genes in the oncogenic network sequencing of multiple cancers has reported thousands of where pharmacologic inhibition can result in death of genes that have low frequency mutation rates in cancer (5– cancer cells while sparing normal cells. Clinical trials in 9). This poses a tremendous challenge for finding novel breast cancer so far have often shown that the most effective therapeutic targets and identifying patient subgroups that treatment is when chemotherapy is combined with targeted therapies rather than chemotherapy or targeted therapies alone (13–15). We used a combinatorial approach using RNA interfer- Authors' Affiliations: Departments of 1Cell Biology and 2Internal Medicine, University of Texas Southwestern Medical Center, Dallas, Texas ence (RNAi; short hairpin RNA; shRNA) against a cohort of candidate breast cancer genes identified via whole-genome Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/). cancer sequencing along with docetaxel to identify gene targets whose loss-of-function would augment chemosensi- Corresponding Author: Jerry W. Shay, Department of Cell Biology, Uni- versity of Texas Southwestern Medical Center, 5323 Harry Hines Boule- tivity. We conducted the chemosensitivity screen against a vard, Dallas, TX 75390. Phone: 214-633-1994; Fax: 214-648-5814; E-mail: well-characterized estrogen receptor–negative, progesterone [email protected] receptor–negative, and Her2-negative (ERPRHer2), doi: 10.1158/1078-0432.CCR-13-0082 "triple-negative," claudin-low breast cancer cell line, 2013 American Association for Cancer Research. MDA-MB-231, as it represents the clinical subtype that has

www.aacrjournals.org 2061

Singel et al.

diagnosed with Li-Fraumeni syndrome as previously Translational Relevance described (18). The missense p53 mutation (M133T) in Use of chemotherapies is often limited by side effects HME50 was sequence verified. All cancer cell lines were and intrinsic or acquired resistance of the tumor. Given cultured in basal medium supplemented with 10% fetal calf the vast heterogeneity of genetic aberrations found in serum. All benign cells were cultured in serum-free condi- breast cancer, biomarkers that may predict response or tions as described elsewhere (19). resistance to certain chemotherapy can be used not only to identify patients appropriate for certain treatment Chemicals regimens, but also to define patient populations who Doxorubicin and docetaxel were obtained from Sigma. may benefit from additional targeted therapies. Recent PP2 and SB203580 were obtained from Tocris Bioscience. acceleration of cancer genome sequencing efforts has Expression array analysis and statistics outpaced downstream molecular studies focused on the Five publically available breast cancer expression datasets identification of robust biomarkers and cancer drivers. (3, 20, 21) were separately normalized then pooled for Genes with known somatic mutations, compared with analysis (n ¼ 946). These datasets were chosen for having the entire genome, are more likely to include biomarkers clinical annotations that include ER status, axillary lymph that are not only prognostic but also encompass key node involvement, relapse-free and/or overall survival (OS) genes that drive or maintain the malignant phenotype. A information. Data were downloaded from Gene Expression chemosensitivity screen using RNA interference (RNAi) Omnibus (GEO; www.ncbi.nlm.nih.gov/geo/) with acces- against this subset of genes is a functional way to identify sion numbers GSE3494 (20), GSE6532 (21), GSE1456 biomarkers whose expressions are important from a (22), GSE7390 (23), and from Nederlands Kanker Instituut treatment response point of view. (http://bioinformatics.nki.nl/data.php; ref. 3). Neoadju- vant data were accessed with GSE25066 (24). Probes were mapped using Entrez identifiers and Human Genome Orga- the worst prognosis (16, 17). We used docetaxel as it is one of nization gene symbols and then averaged for gene level the most common chemotherapies given for breast cancer. analysis. Missing values were imputed using nearest neigh- Although response rates are high to taxanes, toxicities includ- bor averaging. All data analysis was conducted using tools ing neuropathy and myelosuppression often preclude use of in R/Bioconductor. Datasets were merged and standardized these drugs at high doses or for prolonged periods of time. by scaling the columns then the rows and checked using Identification of novel targets that would enhance docetaxel principal component analysis (PCA). Visualization of the chemosensitivity and enable lower effective dosages may PCA method for normalization is provided in Supplemen- allow patients a better quality of life and perhaps improved tary Fig. S1. Univariate and multivariate Cox proportional prognosis. hazard models were used to analyze prognostic capacity of KIF14 and TLN1, with and without clinical feature, to Materials and Methods relapse-free and OS. Using PCA, patients were split into 2 groups by the median value of the first component. Survival Cells curves for these groups were evaluated by Kaplan–Meier MDA-MB-231, HCC38, Hs578T, and MCF7 cells were estimators with log-rank P values reported. The Cancer kindly provided by M. White (Department of Cell Biology, Genome Atlas (TCGA) breast cancer expression data and University of Texas Southwestern Medical School, Dallas, clinical annotations were downloaded from the TCGA TX). T47D and HCC1428 cells were kindly provided by G. website (https://tcga-data.nci.nih.gov/tcga/), and expres- Pearson (Department of Pharmacology, Simmons Com- sion data with clinical annotations on 52 breast cancer cell prehensive Cancer Center, Dallas, TX). HME2424 cells were lines (25) were downloaded from http://cancer.lbl.gov/ a gift from D. Euhus and were originally immortalized by breastcancer/data.php. For the TCGA dataset and cell line retroviral infection with human telomerase reverse tran- data, the median expression of KIF14 was used to split each scriptase (hTERT) by D. Euhus (Department of Surgery, dataset into 2 cohorts to examine expression patterns in Simmons Cancer Center, University of Texas Southwestern samples that belong to the 4 clinical subtypes of breast Medical Center, Dallas, TX). The 2800delAA of BRCA1 in cancer: triple-negative, ERþ and/or PRþ Her2, HME2424 was sequence verified. SUM190PT cells were ERPRHER2þ,orERþ and/or PRþ HER2þ. Bootstrap- purchased from Asterand. HCC1937 cells were originally ping was confirmed by Kolmogorov–Smirnov test (D ¼ derived by A. Gazdar (University of Texas Southwestern 0.317; P < 0.0001). Medical Center, Dallas, TX) and are available from Amer- ican Type Culture Collection (ATCC) Cell Systems. Human Screen and cell viability assay mammary epithelial cells (HMEC; HME1) were originally A total of 328 shRNAs in pGIPZ against 133 candidate immortalized by retroviral infection with hTERT by J.W. breast cancer genes (5) were picked from Open Biosystems Shay (University of Texas Southwestern Medical Center, CSHL Hs shRNAmir 6.13 (Lenti) library (GE Healthcare Life Dallas, TX) and are available from ATCC Cell Systems Sciences). Clone IDs for each shRNA used in this study are (Gaithsburg, MD). HME50 cells were originally derived by provided in Supplementary Table S1 and could be used to J.W. Shay from the noncancerous breast tissue of a female retrieve target sequence from Open Biosystems. Of note,

2062 Clin Cancer Res; 19(8) April 15, 2013 Clinical Cancer Research

KIF14 and TLN1 Modulate Chemosensitivity in Breast Cancer

5,000 cells of each stably selected clone were seeded on a 96- IN; ref. 27). pEGFP-TLN1 was obtained from Addgene well plate using a BioMek FX Automation Workstation (plasmid #26724). Transfections were carried out with (Beckman Coulter Inc.). Five replicates of 12 doses of each Lipofectamine LTX reagent (Invitrogen) as per manufac- chemotherapy were arrayed on each 96-well plate. Cell turer’s protocol. viability was measured 72 hours later with CellTiter-Glo (Promega) as per manufacturer’s protocol. Docetaxel Quantitative PCR IC50 values were independently calculated for each cell All samples were prepared from cultured cells in log phase clone with GraphPad Prism 5 (GraphPad Software, Inc.). growth. RNA was extracted with RNeasy MiniKit (Qiagen) Nonsilencing shRNA control underwent 13 temporally and cDNA was made with First Strand cDNA Synthesis Kit distinct determinations of IC50 for docetaxel (Supplemen- (Roche) as per manufacturers’ protocols. Quantitative PCR tary Table S2) for determination of 95% confidence inter- (qPCR) were run in triplicates in Roche LightCyler 480 vals (95% CI) and z scores. For siRNA experiments, 100 System with DNA SYBR Green Master Mix (Roche). Refer- nmol/L pooled siRNAs (SMARTpool; Dharmacon) for ence genes HSP90, HRPT1, and GUSB were used in each KIF14, TLN1, CIT, ARRB2, PSTPiP1, PRC1, SVIL, ITGA2B, experiment for normalization. All data were analyzed with ITGB3, VCL, and PXN or siControl (D001206; Dharmacon) Biogazelle qBasePLUS. Primer efficiencies were determined were transfected into each cell line using RNAiMAX (Invi- empirically and were between 1.91 and 2.08. Primers for the trogen) as per manufacturer’s instructions. Docetaxel at genes are as follows: GUSB (forward-ctcatttggaattttgccgatt; 1 nmol/L and chemicals (PP2 and SB203580) or dimethyl reverse-ccgagtgaagatccccttttta); HSP90 (forward-ccaaaaag- sulfoxide (DMSO) as control were added 48 hours after cacctggagatca; reverse-tgtcggcctcagccttct); HPRT1 (for- siRNA infection and cell viability was determined 72 hours ward-tgacactggcaaaacaatgca; reverse-ggtccttttcaccagcaagct); later with CellTiter-Glo (Promega). Data are means from KIF14 (forward-tctgaaagggagcaagctaca; reverse-caccatca- 2 independent experiments carried out in triplicates. Cell caaacagcatcc); TLN1 (forward-caaaaaacggtgaagagagctgat; viability was also visualized with Crystal Violet stain (Sig- reverse-tgctgtacctcgatctgaatctg). ma) for confirmation. Viral transductions and stable selections Xenografts For lentivirus production, 1 mg of pGIPZ-shRNA plasmid Mammary fat pads were cleared from 3-week-old non- together with 1 mg of helper plasmids (0.4 mg pMD2G and 0.6 obese diabetic/severe combined immunodeficient (NOD/ mg psPAX2) were transfected into 293FT cells with Effectene SCID) mice, as previously described (26). One million cells reagent (Qiagen). Viral supernatants were collected 48 hours m in growth media suspension was injected in 10 L into a after transfections and cleared through 0.45-mmfilter.Cells cleared fat pad. Nonsilencing controls and knockdown cells were infected with viral supernatants containing 4 mg/mL were injected in contralateral mammary fat pads. Four polybrene (Sigma) and selected with puromycin for 7 days. weekly intraperitoneal injections of docetaxel (5 mg/kg) were administered starting at 7 days after xenograft place- Western blot analysis ment. Eight weeks after xenograft placement, primary Total cell lysates were prepared by harvesting cells in tumors were dissected out. Tumor volume was calculated Laemmli SDS reducing buffer. Protein concentrations were as (width2 length) mm2/2. All animal work was approved measured using a Pierce BCA protein assay kit (Thermo and conducted as per institutional guidelines. Scientific), resolved on an 8% to 10% polyacrylamide gel, and transferred to a polyvinylidine fluoride membrane. Flow cytometry Antibodies used are as follows: glyceraldehyde-3-phosphate Cells were trypsizined and suspended in PBS to a final cell dehydrogenase (GAPDH; Cell Signaling), KIF14 (Bethyl concentration of 10 million cells/mL. Annexin V Alexa Fluor Laboratories), and TLN1 (Millipore). Detection of peroxi- 647 (Life Technologies) was used as per manufacturer’s dase activity from horseradish peroxidase–conjugated anti- instructions. Flow cytometry was conducted on BD FACS bodies was done with SuperSignal West Femto Maximum Calibur Flow Cytometer (BD) and data analysis was con- Sensitivity Substrate (Thermo Scientific). Images were cap- ducted using FlowJo software (Tree Star, Inc.). tured with the G:BOX F3 with GeneSys software (SynGene). Kinase assay Inhibitor activity against RIP2 was conducted using a Results radioisotope assay developed and conducted by Signal- Identification of genes whose expressions most Chem. SB203580 and PP2 were provided at 10 mmol/L to correlate with chemosensitivity SignalChem. Profiling of the 2 compounds were done at 4 We used 328 shRNAs targeting 133 genes with known concentrations (4, 40, 400, and 4,000 nmol/L) in triplicate somatic mutations in breast cancer (ref. 5; Supplementary to determine the IC50 values. Table S1) to generate stable gene knockdown in MDA-MB- 231 cells. We arrayed each cell clone in quintuplicate on Plasmids and transfections 96-well microtiter plates for 12 doses of docetaxel to detect pEGFP KIF14 was kindly provided by T. Corson (Depart- the effect of each gene knockdown in chemosensitivity ment of Ophthalmology, Indiana University, Bloomington, as reflected in changes of IC50 from MDA-MB-231 cells

www.aacrjournals.org Clin Cancer Res; 19(8) April 15, 2013 2063

Singel et al.

Figure 1. A, summary of chemosensitivity (z scores) of each clone of shRNA cells compared with nonsilencing MDA-MB-231 cells from the primary screen. B, relative cell viability of loss-of- function KIF14 and TLN1 MDA- MB-231 cells along with transfected KIF14 and TLN1 cDNA plasmids compared with nonsilencing sh-control. Data represent average of 2 experiments done in triplicates SEM. Expression of top candidate shRNA clones by qPCR. Data represent average of triplicates SEM (n ¼ 6 for each group). C, relative KIF14 and TLN1 expressions by qPCR in sh- and si-KIF14 and TLN1 MDA-MB-231 cells. Data represent average of triplicates SEM (n ¼ 6 for each group). D, Western blot analyses showing relative expressions of loss- and reconstituted- expressions of KIF14 and TLN1 compared with nonsilencing sh- control. E, relative cell viability of various cell lines when transfected with si- KIF14 and TLN1 along with 1 nmol/L docetaxel for 72 hours. , P < 0.01 (two-tailed t test). F, Western blot analyses showing relative endogenous expressions of KIF14 and TLN1 in various cell lines.

containing a nonsilencing shRNA (Fig. 1A and Supplemen- To evaluate for potential off-target effects of the specific tary Table S2). We rank-ordered the shRNA cell lines shRNAs, we transfected KIF14 and TLN1 cDNAs into according to docetaxel chemosensitivity and focused on shKIF14 and shTLN1 MDA-MB-231 cells, respectively, and the top 2 genes whose knockdown on at least 3 separate IC50 found significant decrease in chemosensitivity, indicating a determinations resulted in significant sensitivity in MDA- specific role of KIF14 and TLN1 in chemosensitivity (Fig. MB-231 cells to docetaxel: KIF14 and Talin 1 (TLN1; Fig. 1A 1B). qPCRs and Western blot analyses of these clones and Supplementary Table S2). KIF14 is a microtubule- showed excellent amount of gene knockdown and recon- dependent molecular motor, containing a kinesin motor stituted expressions (Fig. 1C and D). To evaluate effects of domain and a forkhead-associated domain. Its function is gene knockdown on chemosensitivity of multiple cell lines, essential for the final phase of cytokinesis (27, 28). The we used siRNAs against KIF14 and TLN1 (Fig. 1E). Among mechanism of docetaxel chemosensitivity is unclear but available breast cancer cell lines, we chose 3 additional may include further disruption to microtubule dynamics. triple-negative breast cancer (TNBC) cell lines with basal KIF14 overexpression in many cancers including breast, or mesenchymal expression profiles (HCC38 and ovarian, and lung cancers has been described and has been HCC1937–basal, Hs478T–mesenchymal/stem-like) along tested successfully as an independent prognostic marker with MDA-MB-231 (mesenchymal/stem-like) representing (29–31). TLN1 is a focal adhesion complex protein that subtypes of breast cancer with poor prognosis and no regulates integrin interactions with the extracellular matrix molecularly targeted agents available (33) to evaluate for (ECM). Prostate cancer cell models have suggested TLN1 to changes in chemoresponsiveness when KIF14 and TLN1 are be involved in progression to metastasis (32). knocked down. KIF14 and TLN1 knockdown enhanced

2064 Clin Cancer Res; 19(8) April 15, 2013 Clinical Cancer Research

KIF14 and TLN1 Modulate Chemosensitivity in Breast Cancer

chemosensitivity in all 4 triple-negative cell lines tested to expression, when added to ER and node status, improves various degrees (P ¼ 0.0003–0.0054; Fig. 1E). KIF14 knock- relapse-free survival prognostic power compared with clin- down but not TLN1 knockdown affected docetaxel chemo- icopathologic features alone (P ¼ 6.9e-9 vs. P ¼ 4.1e-5, sensitivity in a Her2-amplified cell line, SUM190PT (P ¼ respectively). For patients with OS and clinicopathologic 0.0014 and 0.41, respectively). However, KIF14 and TLN1 annotations (n ¼ 493), KIF14 expression also improves OS loss-of-function in ERþ cell lines (MCF7, T47D, and prognostic power compared with clinicopathologic features HCC1428) and normal HMECs (HME1) or premalignant alone (P ¼ 3.6e-9 vs. P ¼ 6.2e-8, respectively). Boot-strap- cells including HME50 (p53 mutation carrier), and ping was conducted using 1,000 random genes selected HME2424 (BRCA1 mutation carrier) did not significantly from a pool of all genes in the combined dataset (n ¼ affect cell survival when treated with docetaxel at IC50 for 11,262) for relapse-free and OS to show the prognostic specific cell lines (Fig. 1E). KIF14 and TLN1 expressions capacity of KIF14 relative to a random sample, and showed seem to be upregulated in at least half of our panel of 8 that KIF14 is unlikely to be randomly associated with breast cancer cell lines compared with the 3 benign cell lines improved prognostic power (Fig. 3B). tested (Fig. 1F). We further evaluated if KIF14 and TLN1 may be prognostic for patients who received taxane-containing chemo- KIF14 and TLN1 are overexpressed more often in triple- therapy neoadjuvantly. We used the largest publicly avail- negative breast cancers than other clinical subtypes able dataset available with survival annotation (24) and To evaluate how elevated KIF14 and TLN1 expressions found that KIF14 expression is correlated with worse may be significant for breast cancer, we investigated relative distant relapse-free (DRF) survival but not TLN1 (KIF14: gene expressions within known clinical subtypes in a col- log-rank P ¼ 0.0028, HR, 1.8; TLN1: log-rank P ¼ 0.088, HR, lection of 52 breast cancer cell lines (Supplementary Table 1.4; n ¼ 508). However, if we only examine the triple- S3) previously described to model recurrent genomic and negative subgroup, both KIF14 and TLN1 are prognostic transcriptional characteristics of primary breast tumors of DRF survival (KIF14: log-rank P ¼ 0.035, HR, 1.9; TLN1: (24). Median values for KIF14 and TLN1 expressions were log-rank P ¼ 0.028, HR, 1.8; n ¼ 178). Neither gene is used to divide cases into high- or low-expression groups. prognostic in the ERþ subgroup (data not shown). Within TNBC cell lines, 66.7% and 75% of the cell lines have relatively high KIF14 or TLN1 expressions, respectively KIF14 and TLN1 confer survival advantage in vitro and (Fig. 2A). We then examined TCGA (34) to see if KIF14 in vivo and TLN1 overexpressions are more common among triple- Although MDA-MB-231 cells with KIF14 or TLN1 knock- negative primary breast cancers. TNBCs, even though repre- down do not have appreciable proliferation defects (Fig. senting only 15% (71 of459) of total cases, have a higher 4A), cell survival is significantly less when low dose che- proportion of high-KIF14 (91.6%) and TLN1 (62%) expres- motherapy (IC20) is given (Fig. 4B). Annexin V staining by sions relative to the other clinical subtypes (Fig. 2B). flow cytometry indicates that apoptosis contributes largely We next tested if KIF14 and TLN1 may be prognostic to the difference seen in cell numbers (Supplementary biomarkers. We chose 5 large, publicly available breast Fig. S2). cancer expression array datasets that have adequate clinical To assess chemosensitivity in vivo, xenografts of KIF14 annotations including ER status, axillary lymph node and TLN1 knockdown cells into cleared mammary fat pads involvement, relapse-free, and/or OS data (see Materials of NOD/SCID mice showed dramatic decrease in tumor and Methods). We then normalized and merged the data- mass after docetaxel treatments compared with nonsilen- sets (see Materials and Methods and Supplementary Fig. cing control MDA-MB-231 cells injected into contralateral S1), creating a combined breast cancer expression array cleared fat pads (Fig. 4C and D). dataset to evaluate if expressions of these genes correlated with prognosis in breast cancer (n ¼ 946 for relapse-free Interaction networks of KIF14 and TLN1 suggest survival and n ¼ 652 for OS). KIF14 expression is correlated further therapeutic targets to relapse-free survival and OS outcomes by univariate Cox Protein–protein interaction networks of KIF14 and TLN1 regression but TLN1 is not (P ¼ 2.2e-9 and 3.9e-7, respec- identify multiple genes, including 3 kinases (FAK, CIT, and tively for KIF14 and P ¼ 0.45 and 0.95, respectively for RIP2) that may participate in functional networks that affect TLN1). Kaplan–Meier analysis of relapse-free and OS out- chemosensitivity. Knockdown of RIP2, but not FAK, CIT, or comes shows significant prognostic value of KIF14 expres- a number of other known protein–protein interaction sion (see Fig. 3A), but not of TLN1 expression (log-rank P ¼ partners of KIF14 and TLN1, including ARRB2, PSTPiP1, 0.132; HR, 0.84 for relapse-free and log-rank P ¼ 0.957; HR PRC1, SVIL, ITGA2B, ITGB3, VCL, and PXN, significantly 0.99 for OS; data not shown). To evaluate the added value of altered docetaxel chemosensitivity in MDA-MB-231 cells using KIF14 expression in addition to key clinicopathologic (Fig. 5A). KIF14–RIP2 interaction was detected from a large- prognostic features such as ER status and axillary lymph scale mapping of human protein–protein interactions by node involvement, we used PCA and Cox regression anal- mass spectrometry (35). RIP2 is a serine/threonine/tyrosine ysis on the dataset of 919 patients (27 subjects were exclud- kinase that modulates innate and adaptive immune ed from our pooled total from statistical analysis due to responses and activates NF-kB (36). Functional significance incomplete clinical annotation) and found that KIF14 of KIF14–RIP2 interaction is unknown.

www.aacrjournals.org Clin Cancer Res; 19(8) April 15, 2013 2065

Singel et al.

A ER+/PR−/Her2− ER+/PR−/Her2+ ER−/PR+/Her2− ER−/PR+/Her2+ ER−/PR−/Her2− ER+/PR+/Her2− ER+/PR+/Her2+ ER−PR−Her2+ n = 24 n = 16 n = 4 n = 8

66.7% 37.5% 25% 37.5%

Median KIF14 expression

75% 25% 0% 50%

TLN1 expression Median

Figure 2. 0 10 20 30 40 50 Distribution of high- and low-KIF14 and TLN1 expressions Number of cell lines (total n = 52) (dichotomized about the median) among 52 breast cancer cell lines B ER+/PR−/Her2− ER+/PR−/Her2+ ER−/PR+/Her2− ER−/PR+/Her2+ (A) and TCGA (B) within each ER−/PR−/Her2− ER+/PR+/Her2− ER+/PR+/Her2+ ER−PR−Her2+ clinical subtype of breast cancer. n = 71 n = 294 n = 74 n = 20

91.6% 36.8% 52.7% 75%

Median KIF14 expression

62% 50.7% 48.6% 65%

Median TLN1 expression

0 100 200 300 400 Number of subjects (total n = 459) TCGA

Knockdown of both RIP2 and KIF14 further chemosen- Discussion sitizes MDA-MB-231 cells, which suggests a genetic inter- We aimed to find novel biomarkers of chemosensitivity action between KIF14 and RIP2 (Fig. 5B). RIP2 kinase from the breast cancer genome—a selected group of genes inhibitors, PP2 and SB203580 (IC50 for RIP2 is 51.36 and known to have somatic mutations in breast cancer. Anno- 99.79 nmol/L, respectively; Fig. 5C), enhanced chemosen- tation of the breast cancer genome to include functional sitivity of KIF14-deficient MDA-MB-231 cells to docetaxel consequence of gene expression on chemosensitivity is (Fig. 5D). RIP2 knockdown or inhibition by PP2 and important from a treatment perspective. From the screen, SB203580 at less than or equal to 10 mmol/L was not toxic we focused on 2 genes, KIF14 and TLN1, whose loss-of- to normal HMEC line HME1 (data not shown). function most significantly increased chemosensitivity of

2066 Clin Cancer Res; 19(8) April 15, 2013 Clinical Cancer Research

KIF14 and TLN1 Modulate Chemosensitivity in Breast Cancer

A KIF14 KIF14

n = 326 n = 473

n = 473 n = 326 Proportion of OS Figure 3. A, survival curves using Kaplan–Meier estimators using P = 7.1E-8 P = 7.2E-9 KIF14 as a biomarker (dichotomized HR 2.47 (95% CI, 1.80–3.39) Proportion of relapse-free survival HR 1.87 (95% CI, 1.49–2.36) 0.0 0.2 0.4 0.6 0.8 1.0 about the median) for relapse-free 0.0 0.2 0.4 0.6 0.8 1.0 P and OS with log-rank values 0 5 10 15 0 5 10 15 20 25 reported. Red denotes high- and Survival time (y) Survival time (y) black denotes low-KIF14 expressions. B, distribution of P B values following bootstrapping with Relapse-free survival (n = 919) OS (n = 493) 1,000 random genes as biomarkers P = 0.05 P = 0.05 for relapse-free and OS when using the genes alone versus genes along with clinical features (ER positivity KIF14 KIF14 − P = 7.2e−09 and axillary lymph node P = 7.1e 08 involvement).

Clinical features alone Clinical features alone P = 6.2e−08 P = 4.1e−05 KIF14 + clinical features KIF14 + clinical features P = 6.9e−09 P = 3.6e−09 PCA + Clinical PCA + Clinical

1e−11 1e−08 1e−05 1e−02 1e−11 1e−08 1e−05 1e−02 P value P value

MDA-MB-231 cells to docetaxel. KIF14 and TLN1 knock- the other clinical subtypes (37–39). We showed that down clearly enhanced chemosensitivity to docetaxel in KIF14 expression is prognostic of survival in breast cancer 4 triple-negative, mesenchymal/stem-like/basal breast can- in a large combined dataset (>900 patients with annota- cer cell lines (MDA-MB-231, HCC1937, HCC38, and tions for relapse-free survival and >400 patients for Hs578T) but not significantly in ERþ breast cancers, impli- OS, Fig. 3), corroborating previous report of KIF14 as a cating intrinsic differences in survival mechanisms or prognostic biomarker in breast cancer (29). Because we dependence on specific oncogenic pathways among differ- have found that high KIF14 expression is found predom- ent breast cancer subtypes. In addition, because KIF14 inately in TNBC (>90% of TNBC cases within TCGA have and TLN1 knockdown did not affect chemosensitivity of high KIF14 expression) and that decreasing KIF14 expres- normal HMECs, these genes may allow for pharmacologic sion or function would chemosensitize TNBC cells to intervention. docetaxel, KIF14 is an attractive target for therapeutic Besides their roles in chemosensitivity, KIF14 and TLN1 intervention. expressions are found to be upregulated more often within Via interrogation of protein–protein interaction part- the TNBC subtype than the other clinical subtypes. TNBCs ners with KIF14, we further identified RIP2 as a poten- convey a poor prognosis, insensitivity to adjuvant chemo- tial novel target for chemosensitization. Additive or therapy, and resistance to current targeted therapies (37). synergistic chemosensitization of RIP2 knockdown or We have found that KIF14 and TLN1 expressions correlate RIP2 chemical inhibition in a KIF14-deficient back- with docetaxel chemosensitivity in basal and mesenchy- ground in MDA-MB-231 cells also suggest a potential mal triple-negative cell lines in general better than other therapeutic combination. RIP2 is known to be an cell lines tested. Although a number of biomarkers have important mediator of inflammation (40, 41). Here, been identified for TNBC, none has seemed to be we show a previously unrecognized role of RIP2 in expressed in a significant proportion of TNBC cases or breast cancer and as a potential chemosensitizer and be specifically upregulated in the TNBCs compared with interactor with KIF14.

www.aacrjournals.org Clin Cancer Res; 19(8) April 15, 2013 2067

Singel et al.

In a search for speciﬁc druggable oncogenic depen- decreased KIF14 and TLN1 expression is a prognostic dence, we found that inhibition of 2 genes, KIF14 and marker for better outcome after cytotoxic chemotherapy, TLN1, enhances the therapeutic index of docetaxel in and that inhibition of these genes can sensitize KIF14 TNBC. These results suggest that KIF14 and TLN1 play and TLN1 overexpressing TNBC cells to therapeutic a role in response to cytotoxic chemotherapy, that intervention.

A MDA-MB-231 Hs478T HCC38

600 sh-control 600 600 sh-TLN1 sh-KIF14 400 400 400

200 200 200 # cells, x1,000 # cells, x1,000 # cells, x1,000

0 0 0 6420 6420 6420 Days Days Days

B IC20 docetaxel IC20 docetaxel 60 IC20 docetaxel 60 60

40 40 40

20 20 20 # cells, x1,000 # cells, x1,000 # cells, x1,000

0 0 0 0 2 4 6 0 2 4 6 0 2 4 6 Days Days Days

sh-control shKIF14 shTLN1 sh-control shKIF14 shTLN1 sh-control shKIF14 shTLN1 Docetaxel day 4 CD

shKIF14 sh-control P < 0.001

n = 6 1,500 P < 0.001 3 n = 5 1,000

shTLN1 sh-control

500 Tumor volume, mm 0

Control Control shKIF14 shTLN1

Figure 4. Functional consequences of KIF14 and TLN1 knockdown in MDA-MB-231, Hs478T, and HCC38 cells by (A) normal proliferation rates with no drug treatment, (B) proliferation rates when cells are treated with low dose docetaxel. Crystal violet stains of representative images from day 4 after docetaxel treatment. Data represent average of triplicates SEM. Data are representative of at least 2 independent experiments. C, representative mammary fat pad xenografts after treatment with docetaxel. Paired mammary fat pads are from contralateral sides of the same mice. Scale bar, 0.5 cm. D, summarizes average tumor volume SD for each group treated with docetaxel, pooled from 2 independent experiments.

2068 Clin Cancer Res; 19(8) April 15, 2013 Clinical Cancer Research

KIF14 and TLN1 Modulate Chemosensitivity in Breast Cancer

A * P < 0.0001 1.5

1.0

0.5 Surviving fraction

0.0 2 1 P1 -FAK VCL si-CIT PRC si si- si-PXN STPi si-SVIL si- si-RIP2 si-control si-ARRB si-ITGB3 si-P si-ITGA2B KIF14 interactors TLN1 interactors Figure 5. RNAi of various known protein–protein interaction partners B sh-control of KIF14 and TLN1 in control MDA- *P = 0.0007 sh-KIF14 MB-231 cells (A) and stably knocked 1.5 sh-TLN1 down KIF14 and TLN1 MDA-MB- 231 cells (B) and their effects on cell viability when docetaxel is given at 1 nmol/L for 3 days. Data represent 1.0 average of triplicates SEM. Data are representative of at least 2 independent experiments. C, in vitro 0.5 RIP2 kinase activity using PP2 and

SB203580 as inhibitors. D, relative Surviving fraction viability when MDA-MB-231 cells are treated with 1 nmol/L docetaxel 0.0 with or without IC50 of PP2 and SB203580 (see text) for 3 days. Data si-CIT si-SVIL si-RIP2 si-FAK si-VCL si-PXN represent average of triplicates si-control si-PRC1 si-ITGB3 si-ARRB2 si-ITGA2B SEM. Data are representative of at si-PSTPiP1 least 2 independent experiments. C D RIP2 kinase activity 1.5 sh-control sh-KIF14 100 PP2 SB203580 1.0

0.5 P = 0.037 50

Relative viability P = 0.022

0.0 % Inhibition of activity 0 0 1 2 3 4 No chemo Docetaxel Log concentration (nmol/L) Docetaxel + PP2

Docetaxel + SB203580

Disclosure of Potential Conﬂicts of Interest Acknowledgments No potential conﬂicts of interest were disclosed. The authors thank Michael White and Tim Corson for helpful discus- sions and materials and David Euhus for gift of cell culture cells. Authors' Contributions Conception and design: S.M. Singel, L. Lum, J.W. Shay Development of methodology: S.M. Singel, J.W. Shay Grant Support Acquisition of data (provided animals, acquired and managed patients, This study was supported by Cancer Center Support Grant (5P30 provided facilities, etc.): S.M. Singel CA142543-03), Susan G. Komen Foundation Postdoctoral Fellowship and Analysis and interpretation of data (e.g., statistical analysis, biosta- National Cancer Institute T32 CA136515 (to S.M. Singel), The Southland tistics, computational analysis): S.M. Singel, K. Batten, L. Lum, J.W. Shay Financial Corporation Distinguished Chair in Geriatric Research (to J.W. Shay), Writing, review, and/or revision of the manuscript: S.M. Singel, G. Cancer Prevention Research Institute of Texas grant (RP100119), Welch Foun- Fasciani, W.E. Wright, L. Lum, J.W. Shay dation (I-1665; to L. Lum), and National Cancer Institute Specialized Programs Administrative, technical, or material support (i.e., reporting or orga- of Research Excellence (P50 CA070907; to L. Lum and J.W. Shay). nizing data, constructing databases): C. Cornelius, K. Batten, G. Fasciani The costs of publication of this article were defrayed in part by the Study supervision: W.E. Wright, L. Lum, J.W. Shay payment of page charges. This article must therefore be hereby marked

www.aacrjournals.org Clin Cancer Res; 19(8) April 15, 2013 2069

Singel et al.

advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate Received January 10, 2013; revised February 14, 2013; accepted February this fact. 22, 2013; published OnlineFirst March 11, 2013.

References 1. Perou CM, Sorlie T, Eisen MB, van de Rijn M, Jeffrey SS, Rees CA, et al. tor-positive breast carcinomas through genomic grade. J Clin Oncol Molecular portraits of human breast tumours. Nature 2000;406:747–52. 2007;25:1239–46. 2. Sorlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H, et al. 22. Pawitan Y, Bjohle J, Amler L, Borg AL, Egyhazi S, Hall P, et al. Gene Gene expression patterns of breast carcinomas distinguish tumor expression profiling spares early breast cancer patients from adjuvant subclasses with clinical implications. Proc Natl Acad Sci U S A therapy: derived and validated in two population-based cohorts. 2001;98:10869–74. Breast Cancer Res 2005;7:R953–64. 3. van de Vijver MJ, He YD, van't Veer LJ, Dai H, Hart AA, Voskuil DW, 23. Desmedt C, Piette F, Loi S, Wang Y, Lallemand F, Haibe-Kains B, et al. et al. A gene-expression signature as a predictor of survival in breast Strong time dependence of the 76-gene prognostic signature for cancer. N Engl J Med 2002;347:1999–2009. node-negative breast cancer patients in the TRANSBIG multicenter 4. Paik S, Shak S, Tang G, Kim C, Baker J, Cronin M, et al. A multigene independent validation series. Clin Cancer Res 2007;13:3207–14. assay to predict recurrence of tamoxifen-treated, node-negative 24. Hatzis C, Pusztai L, Valero V, Booser DJ, Esserman L, Lluch A, et al. A breast cancer. N Engl J Med 2004;351:2817–26. genomic predictor of response and survival following taxane-anthracy- 5. Wood LD, Parsons DW, Jones S, Lin J, Sjoblom T, Leary RJ, et al. The cline chemotherapy for invasive breast cancer. JAMA 2011;305:1873–81. genomic landscapes of human breast and colorectal cancers. Science 25. Neve RM, Chin K, Fridlyand J, Yeh J, Baehner FL, Fevr T, et al. A 2007;318:1108–13. collection of breast cancer cell lines for the study of functionally distinct 6. Nik-Zainal S, Van Loo P, Wedge David C, Alexandrov Ludmil B, cancer subtypes. Cancer Cell 2006;10:515–27. Greenman Christopher D, Lau King W, et al. The life history of 21 26. Nik-Zainal S, Alexandrov Ludmil B, Wedge David C, Van Loo P, breast cancers. Cell 2012;149:994–1007. Greenman Christopher D, Raine K, et al. Mutational processes molding 7. Ding L, Ellis MJ, Li S, Larson DE, Chen K, Wallis JW, et al. Genome the genomes of 21 breast cancers. Cell 2012;149:979–93. remodelling in a basal-like breast cancer metastasis and xenograft. 27. Gruneberg U, Neef R, Li X, Chan EH, Chalamalasetty RB, Nigg EA, et al. Nature 2010;464:999–1005. KIF14 and citron kinase act together to promote efficient cytokinesis. 8. Pleasance ED, Stephens PJ, O'Meara S, McBride DJ, Meynert A, J Cell Biol 2006;172:363–72. Jones D, et al. A small-cell lung cancer genome with complex signa- 28. Carleton M, Mao M, Biery M, Warrener P, Kim S, Buser C, et al. RNA tures of tobacco exposure. Nature 2010;463:184–90. interference-mediated silencing of mitotic kinesin KIF14 disrupts cell 9. Shah SP, Morin RD, Khattra J, Prentice L, Pugh T, Burleigh A, et al. cycle progression and induces cytokinesis failure. Mol Cell Biol Mutational evolution in a lobular breast tumour profiled at single 2006;26:3853–63. nucleotide resolution. Nature 2009;461:809–13. 29. Corson TW, Gallie BL. KIF14 mRNA expression is a predictor of grade 10. Curtis C, Shah SP, Chin SF, Turashvili G, Rueda OM, Dunning MJ, et al. and outcome in breast cancer. Int J Cancer 2006;119:1088–94. The genomic and transcriptomic architecture of 2,000 breast tumours 30. Theriault BL, Pajovic S, Bernardini MQ, Shaw PA, Gallie BL. Kinesin reveals novel subgroups. Nature 2012;486:346–52. family member 14: an independent prognostic marker and potential 11. Thorsen SB, Obad S, Jensen NF, Stenvang J, Kauppinen S. The therapeutic target for ovarian cancer. Int J Cancer 2012;130:1844–54. therapeutic potential of microRNAs in cancer. Cancer J 2012;18:275–84. 31. Corson TW, Zhu CQ, Lau SK, Shepherd FA, Tsao MS, Gallie BL. KIF14 12. Huang Y, Nayak S, Jankowitz R, Davidson NE, Oesterreich S. Epige- messenger RNA expression is independently prognostic for outcome netics in breast cancer: what's new?Breast Cancer Res 2011;13:225. in lung cancer. Clin Cancer Res 2007;13:3229–34. 13. Guarneri V, Frassoldati A, Bottini A, Cagossi K, Bisagni G, Sarti S, et al. 32. Sakamoto S, McCann RO, Dhir R, Kyprianou N. Talin1 promotes tumor Preoperative chemotherapy plus trastuzumab, lapatinib, or both in invasion and metastasis via focal adhesion signaling and anoikis human epidermal growth factor receptor 2-positive operable breast resistance. Cancer Res 2010;70:1885–95. cancer: results of the randomized phase II CHER-LOB study. J Clin 33. Lehmann BD, Bauer JA, Chen X, Sanders ME, Chakravarthy AB, Shyr Oncol 2012;30:1989–95. Y, et al. Identification of human triple-negative breast cancer subtypes 14. Baselga J, Bradbury I, Eidtmann H, Di Cosimo S, de Azambuja E, Aura and preclinical models for selection of targeted therapies. J Clin Invest C, et al. Lapatinib with trastuzumab for HER2-positive early breast 2011;121:2750–67. cancer (NeoALTTO): a randomised, open-label, multicentre, phase 3 34. Network CGA. Comprehensive molecular portraits of human breast trial. Lancet 2012;379:633–40. tumours. Nature 2012;490:61–70. 15. Untch M, Loibl S, Bischoff J, Eidtmann H, Kaufmann M, Blohmer JU, 35. Ewing RM, Chu P, Elisma F, Li H, Taylor P, Climie S, et al. Large-scale et al. Lapatinib versus trastuzumab in combination with neoadjuvant mapping of human protein-protein interactions by mass spectrometry. anthracycline-taxane-based chemotherapy (GeparQuinto, GBG 44): a Mol Syst Biol 2007;3:89. randomised phase 3 trial. Lancet Oncol 2012;13:135–44. 36. Dufner A, Pownall S, Mak TW. Caspase recruitment domain protein 6 is 16. Prat A, Perou CM. Deconstructing the molecular portraits of breast a microtubule-interacting protein that positively modulates NF-kap- cancer. Mol Oncol 2011;5:5–23. paB activation. Proc Natl Acad Sci U S A 2006;103:988–93. 17. Prat A, Parker JS, Karginova O, Fan C, Livasy C, Herschkowitz JI, et al. 37. Podo F, Buydens LM, Degani H, Hilhorst R, Klipp E, Gribbestad IS, et al. Phenotypic and molecular characterization of the claudin-low intrinsic Triple-negative breast cancer: present challenges and new perspec- subtype of breast cancer. Breast Cancer Res 2010;12:R68. tives. Mol Oncol 2010;4:209–29. 18. Shay JW, Tomlinson G, Piatyszek MA, Gollahon LS. Spontaneous in 38. Glenisson M, Vacher S, Callens C, Susini A, Cizeron-Clairac G, Le vitro immortalization of breast epithelial cells from a patient with Li- Scodan R, et al. Identification of new candidate therapeutic target Fraumeni syndrome. Mol Cell Biol 1995;15:425–32. genes in triple-negative. Genes Cancer 2012;3:63–70. 19. Shay JW, Van Der Haegen BA, Ying Y, Wright WE. The frequency of 39. Sun T, Aceto N, Meerbrey KL, Kessler JD, Zhou C, Migliaccio I, et al. immortalization of human fibroblasts and mammary epithelial cells Activation of multiple proto-oncogenic tyrosine kinases in breast transfected with SV40 large T-antigen. Exp Cell Res 1993;209:45–52. cancer via loss. Cell 2011;144:703–18. 20. Miller LD, Smeds J, George J, Vega VB, Vergara L, Ploner A, et al. An 40. Yin X, Krikorian P, Logan T, Csizmadia V. Induction of RIP-2 kinase by expression signature for p53 status in human breast cancer predicts proinflammatory cytokines is mediated via NF-kappaB. Mol Cell Bio- mutation status, transcriptional effects, and patient survival. Proc Natl chem 2010;333:251–9. Acad Sci U S A 2005;102:13550–5. 41. Kobayashi K, Inohara N, Hernandez LD, Galan JE, Nunez G, Janeway 21. Loi S, Haibe-Kains B, Desmedt C, Lallemand F, Tutt AM, Gillet C, et al. CA, et al. RICK/Rip2/CARDIAK mediates signalling for receptors of the Definition of clinically distinct molecular subtypes in estrogen recep- innate and adaptive immune systems. Nature 2002;416:194–9.

2070 Clin Cancer Res; 19(8) April 15, 2013 Clinical Cancer Research

A Targeted RNAi Screen of the Breast Cancer Genome Identifies KIF14 and TLN1 as Genes That Modulate Docetaxel Chemosensitivity in Triple-Negative Breast Cancer

Stina Mui Singel, Crystal Cornelius, Kimberly Batten, et al.

Clin Cancer Res 2013;19:2061-2070. Published OnlineFirst March 11, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-13-0082

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2013/03/12/1078-0432.CCR-13-0082.DC1

Cited articles This article cites 41 articles, 13 of which you can access for free at: http://clincancerres.aacrjournals.org/content/19/8/2061.full#ref-list-1

Citing articles This article has been cited by 5 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/19/8/2061.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 Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/19/8/2061. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.