Tumor-Initiating Cells and FZD8 Play a Major Role in Drug Resistance in Triple-Negative Breast Cancer

Published OnlineFirst February 27, 2013; DOI: 10.1158/1535-7163.MCT-12-1090

Molecular Cancer Cancer Therapeutics Insights Therapeutics

Shuping Yin1, Liping Xu1, R. Daniel Bonﬁl1,2, Sanjeev Banerjee1, Fazlul H. Sarkar1, Seema Sethi1, and Kaladhar B. Reddy1

Abstract Triple-negative breast cancer (TNBC) studies have shown that neoadjuvant chemotherapy before surgery was effective in the minority of women, whereas the majority who had residual tumor had a relatively poor outcome. To identify the mechanism by which residual cancer cells survive chemotherapy, we initially conducted gene expression profiling using the CRL2335 TNBC cell line derived from a squamous breast carcinoma before and after treatment with cisplatin plus TRAIL. We found a significant increase in the expression of FZD8, one of Wnt receptors, and its downstream targets LEF1 and TCF in residual CRL2335 tumor cells after treatment with cisplatin plus TRAIL. Increased FZD8 levels were further confirmed in other TNBC cell lines. Inhibition of FZD8 by siRNA in CRL2335 cells in the presence of cisplatin plus TRAIL reduced b-catenin and survivin levels and increased apoptosis compared with scrambled siRNA–treated cells. In vivo data show that cisplatin plus TRAIL treatment significantly reduces tumor volume in NOD/SCID mice. However, we found that cisplatin plus TRAIL treatment predominantly eliminated non–tumor-initiating cells, as shown by whole-body fluorescent imaging of mice injected with mammosphere-forming CRL2335 cells stably transfected with DsRed. This led to TIC enrichment in residual tumors, as confirmed by immunostaining for TIC markers. Moreover, an increase in FZD8 expression was observed in residual tumors treated with cisplatin and TRAIL. Taken together, our findings suggest that FZD8-mediated Wnt signaling may play a major role in mediating resistance to chemotherapy, making it a potential target to enhance chemotherapeutic efficacy in patients with TNBCs. Mol Cancer Ther; 12(4); 491–8. 2013 AACR.

Introduction minority of women with TNBCs who show a complete In patients with breast cancer, the survival rates have pathologic response and an excellent outcome; however, improved steadily over the past 2 decades. However, triple- the outcome for the majority of them, who still have a negative breast cancer (TNBC) exhibits aggressive charac- residual disease after treatment, is relatively poor (6). teristics associated with shorter disease-free survival (1). Several alternative hypotheses have been proposed to TNBC tumors are characterized by the absence of expres- explain the treatment failure and recurrence. It has been sion of estrogen receptor (ER) and progesterone receptor suggested that tumor-initiating cells (TIC), also common- (PR), as well as HER-2 amplification. Therefore, patients ly called cancer stem–like cells (CSC), are resistant to with TNBCs do not benefit from commonly used anti- chemotherapy and radiotherapy, and surviving TICs can estrogen- and herceptin-based therapies (2, 3). Although reinitiate tumor growth after treatment (7–10). TICs can be enriched by fluorescence-activated cell sorting (FACS) chemotherapy is currently the mainstay of systemic treat- þ using specific cell surface markers, by either CD44 ment for breast cancer, patients with TNBC disease have a /low worse outcome after chemotherapy than patients with /CD24 staining and/or aldehyde dehydrogenase 1 (ALDH1) activity (8, 11, 12). In addition, it was shown that other subtypes of breast cancer (4, 5). Previous studies sug- þ /low gested that neoadjuvant treatment involving the adminis- human breast cancer cells that are CD44 /CD24 or tration of chemotherapy before surgery was effective in a ALDH1-positive could efficiently regenerate tumors containing an array of cell types similar to those found in the original tumor (8, 10, 13). Authors' Affiliations: Departments of 1Pathology and 2Urology. Wayne Defining the signaling pathways in residual tumors that State University School of Medicine, Detroit, Michigan survived chemotherapy could help to design new treat- Note: Supplementary data for this article are available at Molecular Cancer ments for patients with TNBCs. Here, we evaluated CRL- Therapeutics Online (http://mct.aacrjournals.org/). 2335 TNBC cells before and after treatment with cisplatin Corresponding Author: Kaladhar B. Reddy, Department of Pathology, plus TRAIL, with the assumption that changes seen in Wayne State University, 540 E. Canfield Avenue, Detroit, MI 48201. Phone: gene expression in residual tumors that survived chemo- 313-577-6191; Fax: 313-577-0057; E-mail: [email protected] therapy could provide new therapeutic targets for those doi: 10.1158/1535-7163.MCT-12-1090 patients with breast cancer who do not benefit from anti- 2013 American Association for Cancer Research. estrogen- and herceptin-based therapies.

www.aacrjournals.org 491

Yin et al.

Materials and Methods 4. FZD7: 0 0 Cell culture forward: 5 -ACAGACTTAGCCACAGCAGCAAGG-3 0 0 The human breast TNBC cell line CRL2335 (derived Reverse: 5 -TTTCCAAATCACCCCTCGCC-3 0 0 from squamous carcinoma of the breast), CRL-2336, 5. FZD8: forward: 5 -TGGAGTGGGGTTACCTGTTG-3 MDA-MB-468, and MDA-231 were obtained from the Reverse: 50-AGCGGCTTCTTGTAGTCCTC-30 American Type Culture Collection (ATCC). Cells were 6. Rspo1: Forward: 50-CAGCCATAACTTCTGCACCA-30 maintained in RPMI-1640 medium (GIBCO/Life Technol- Reverse: 50-AGAGCTGCTGCTTCTTGGAG-30 ogies) supplemented with FBS (GIBCO/Life Technolo- 7. Rspo2: forward: 50-GTACCAAGTGCAAAGTAGGC-30 gies) to a final concentration of 10%. The cells were Reverse: 50-AACTTCACATCCTTCCACAC-30. immediately expanded and frozen after being obtained from ATCC and restarted every 3 to 4 months from a The real-time PCR thermal profile was: 1 cycle of 10 frozen vial of the same batch of cells, and no additional minutes at 95C and then 40 cycles of 30 seconds at 95 C, 1 authentication was done on these cells. All cell lines were minute at 55C, and 1 minute at 72 C followed by 1 cycle free of mycoplasma infection tested by PCR. of 1 minute at 95C, and finally 41 cycles of increasing the temperature 1 C every 30 seconds starting at 55 C. Ct RT2 Profiler PCR array system values were obtained using the built-in Eppendorf pro- The expression of 84 genes related to human stem cell gram "realplex" of the Eppendorf Realplex2 Mastercycler signaling pathway was analyzed using RT2 Profiler PCR machine, and normalized with respect to average Ct array technology (SABiosciences) using CRL2335 TNBC values of GAPDH housekeeping gene. cells treated with 10 mg/mL cisplatin (Sigma) and 10 ng/mL TRAIL (R&D systems) for 24 hours compared RNA interference assay with vehicle-treated CRL2335 cells. Total RNA was CRL2335 cells were plated in 6-well tissue culture plates isolated using the RNeasy Mini Kit (Qiagen) according at a density of 3 105/well in complete growth culture to the manufacturer’s instructions. The RT2 Profiler PCR medium, and 24 hours later transfected with 100 pmol of Array Kit was used to convert mRNA into cDNA and siRNAs directed against FZD8 or random siRNA with then mixed with LightCycler 480 SYBR Green I Master scrambled sequence (Novus Biologicals) using Lipofecta- Mix (Stratagene) and dispensed to each well of the PCR mine transfection reagent (Invitrogen) according to the array plate containing prefilled gene-specific primer manufacturer’s instructions. Forty-eight hours after trans- sets, and PCR was carried out, according to the manu- fection, cells were treated with cisplatin (10 mg/mL) and facturer’s instructions. An MS Excel sheet from the TRAIL (10 ng/mL) for additional 24 hours. manufacturer’s web site was used to include the PCR data, which were analyzed using the PCR Array Data Western blot analysis Analysis Web Portal, and then normalized by correcting CRL2335 cells transiently transfected with FZD8- the C (threshold cycle) values obtained with respect to t siRNA or random siRNA and treated with or not with average C values of housekeeping genes present on the t cisplatin and TRAIL as explained above were grown near array to obtain fold change in expression for each gene. confluence. Cells were lysed and Western blotting was A 4-fold change in gene expression between vehicle- conducted as described previously (14) using a standard and cisplatin þ TRAIL–treated cells was used as a protocol. In brief, cell extracts were obtained by lysing the threshold. cells in radioimmunoprecipitation assay (RIPA) buffer (20 Real-time PCR was carried out to confirm the results mmol/L HEPES, 100 mmol/L NaCl, 0.1% SDS, 1% Non- for certain genes that revealed significant changes in idet P-40, 1% deoxycholate, 1 mmol/L Na VO , 1 mmol/L expression in the RT2 Profiler PCR array and to analyze 3 4 EGTA, 50 mmol/L NaF, 10% glycerol, 1 mmol/L EDTA, 1 additional genes that code for Wnt receptors and their mmol/L phenylmethylsulfonyl fluoride, and 1 protease ligand. Briefly, 1 mg of each total RNA sample was reverse- inhibitor mixture; all reagents from Sigma). Samples con- transcribed using iScript cDNA Synthesis Kit (BioRad). taining 100 mg of total protein were electrophoresed on 8% Brilliant SYBR Green QPCR master mix (Stratagene) was or 15% SDS-PAGE and transferred to polyvinylidene mixed with 1 ml of cDNA and 150 nmol/L of the following fluoride (PVDF) membrane by electroblotting. Mem- primers: branes were probed with specific antibodies against FZD8 (Aviva System Biology), b-catenin (Santa Cruz Biotech- 1. FZD1: forward: 50-GCCCATGAGCCCGGACTTCAC-30 0 0 nology), survivin (Cell Signaling Technology), or ALDH Reverse: 5 -TCAGACTGTAGTCTCCCCTTG-3 (BD Transduction Laboratories) followed by horseradish 2. FZD2: 0 0 peroxidase-conjugated mouse or rabbit secondary anti- forward: 5 -TACCCAGAGCGGCCTATCATTTTT-3 bodies (Amersham) accordingly. The specific bands 0 0 Reverse: 5 - ACGAAGCCGGCCAGGAGGAAGGAC-3 were developed on autoradiography films by treatment 3. FZD3: of the membranes with enhanced chemiluminescent sub- 0 0 forward: 5 -AAGGCTTCCACAGTGACACAAGG-3 strate (Pierce). Tissue homogenates from tumor xeno- Reverse: 50-AGAGGAGAGAAACCCCAACTACCAC-30 grafts obtained as described below were also used for

492 Mol Cancer Ther; 12(4) April 2013 Molecular Cancer Therapeutics

FZD8 and Drug Resistance

Western blot analysis of ALDH. As a loading control, treated with vehicle. The tumors were measured twice a b-actin expression levels were determined with an anti- week with calipers, and tumor volumes in mm3 were actin antibody (BD Biosciences). The intensities of specific calculated by the formula (width)2 length/2. The mice bands in immunoblots were measured using NIH free were fluoroimaged once every 10 days with Kodak software ImageJ and normalized to the corresponding IS4000MM Live Animal Multimodal Imager under anes- actin levels. Values for treated groups were presented as thesia to monitor growth of tumor areas formed by percentage compared with untreated group. DsRed-expressing CRL2335 cells. The quantification of DsRed expression in the tumors was an average of inte- Apoptosis assay grated intensity that was calculated by using built in Apoptosis was assessed using the Cell Death Detection software Metamorph (version 7.6.4.0.) associated with the ELISAplus Kit (Roche Applied Science) according to the Kodak image station. Mice were euthanized 30 days after manufacturer’s instructions. This kit quantitatively tumor injection, and their tumor xenografts harvested. detects cytosolic histone-associated DNA fragments. In Half of each tumor was fixed in 4% paraformaldehyde and brief, cells were treated with cisplatin and TRAIL for 16 then paraffin-embedded for immunohistochemical anal- hours after FZD8-siRNA or random siRNA treatment. ysis, whereas the other half was snap-frozen in liquid Apoptosis was quantified by ELISA and normalized to nitrogen and stored at 80C for further homogenate values measured in untreated cells. Data are mean SE of preparation. Briefly, tumor lysates were obtained by add- triplicate determination. ing 5 mg of tissue into RIPA buffer and homogenizing using an electric homogenizer on ice. The lysate was þ Isolation of ALDH1 cells by FACS centrifuged and the supernatant was used to determine ALDH activity of CRL2335 cells was measured using the protein concentration and used for Western blotting as the ALDEFLOUR Assay Kit (STEMCELL Technologies explained before. Inc.), according to the manufacturer’s instructions. Brief- ly, cells were incubated in an ALDEFLUOR assay buffer Immunohistochemical analysis m 6 containing ALDH substrate (1 mol/L per 1 10 cells). Adjacent 5-mm paraffin sections were used for expres- In each experiment, a sample of cells was stained under sion analysis of human CD44, CD24, ALDH1, survivin, identical conditions with 50 mmol/L of diethylaminoben- and FZD8. Briefly, tissue sections were deparaffinized, zaldehyde, a specific ALDH inhibitor, as a negative rehydrated, subjected to antigen retrieval using Ag Citrus control. Plus Retrieval Solution (BioGenex) and microwave treat- For FACS, CRL2335 cells were labeled with the ALDE- ment. Primary antibody detection was with Vectastain FLUOR Kit and sorted at the Imaging and Cytometry Elite ABC Systems (Vector Laboratories) accordingly cho- Resource’s Core at The Karmanos Cancer Institute, sen based on the species of origin for each antibody or M. Wayne State University (Detroit, MI). The sorting gates O.M. Immunodetection kit peroxidase (Vector Laborato- were established using propidium iodide–stained cells for ries) if mouse IgG. Color was developed with diamino- viability. benzidine peroxidase substrate SIGMAFAST DAB(Sig- ma), and sections were counterstained with hematoxylin Mouse experiments (Sigma). Primary antibodies and dilutions used were: Eight-week-old female NOD/SCID mice were pur- rabbit anti-CD44 (Abcam) at 1:50, 60 minutes at room chased from Taconic Farms and maintained under aseptic temperature, rabbit anti-CD24 (Abcam) at 1:100, over- conditions in accordance with Animal care and use guide- night at 4C, mouse anti-ALDH1 (BD Biosciences) at lines of Wayne State University. ADEFLUOR-negative 1:500, 60 minutes at room temperature, goat anti-FZD8 CRL2335 cells [non–tumor-initiating cells (non–TIC)] (Aviva System Biology) at 3 mg/mL, overnight at 4C, and sorted as explained above, were mixed at the ratio found rabbit anti-survivin (R & D System) at 10 mg/mL, over- in the original cell population ( 75% ADEFLUOR-nega- night at 4 C. TICs were identified in xenograft tumors by tive non-TICs and 25% ADEFLUOR-positive TICs) with þ low/ þ þ CD44 and CD24 staining and/or ALDH staining. ALDH1 CRL2335 cells stably transfected with pCMV- DsRed express vector (PDsRed2-N1, Clontech) using the Statistical analysis Lipofectamine 2000 Reagent (Invitrogen), as per manu- All data are expressed as mean SEM and statistically facturer’s recommendations. A total of 5 105 cells of the þ analyzed using unpaired Student t test. Differences were mixture of ALDH1 and DsRed-expressing ALDH1 considered statistically significant when P < 0.05. were suspended in 50 mL of cold liquid growth factor– reduced Matrigel (BD Biosciences) and injected into the inguinal mammary fat pad (presence of ALDH1 is an Results accepted marker for TICs; refs. 14–16). Once the tumors Gene expression profiles of residual cells after reached a mean size of about 50 mm3, a group of 4 mice cisplatin plus TRAIL treatment began to be treated once a week with cisplatin (4 mg/kg) To identify the major molecular pathways that are and TRAIL (15 mg/kg) intraperitoneally for 3 weeks. A involved in drug resistance to cisplatin plus TRAIL treat- control group consisting of the same number of mice was ment, gene expression profiling of CRL2335 cells before

www.aacrjournals.org Mol Cancer Ther; 12(4) April 2013 493

Yin et al.

Figure 1. FZD8 levels are increased in cisplatin plus TRAIL–treated CRL2335 cells. A, gene expression profiles were conducted to identify the major pathways involved in drug resistance using stem cell signaling PCR array. Note upregulation of Wnt receptor FZD8 and downstream targets LEF1 and TCF7 in cells treated with cisplatin plus TRAIL in heatmap. B, real-time PCR analysis show no significant change in gene expression levels for most Wnt receptors and ligands analyzed, with exception of FZD8 and RSPO-1 (, P < 0.05). C, FZD8 increase in residual cells after cisplatin plus TRAIL treatment was further confirmed at the protein level by Western blot analysis.

and after cisplatin plus TRAIL treatment was done using a FZD8 protein expression levels was further confirmed by stem cell signaling PCR array. We found that the expres- Western blot analysis in TNBC CRL2335, CRL2336, MDA- sions of FZD8, one of the Wnt receptors, and its down- 468, and MDA-231 cells after treatment with cisplatin plus stream targets LEF1 and TCF7 were increased in CRL2335 TRAIL (Fig. 1C). Taken together, these results suggest that cells treated by cisplatin plus TRAIL compared with the cells resistant to the combined treatment with cisplatin untreated controls (Fig. 1A, heatmap). Data from PCR and TRAIL showed increased expression of FZD8 and its array also showed that the treatment significantly inhib- ligand Rspo-1. ited other signaling pathways such as FGF, Hedgehog, Notch, and TGF-b (heatmap). Using the same samples FZD8 inhibition reduces b-catenin and survivin analyzed with the stem cell signaling PCR array, we levels leading to increased apoptosis validated the results obtained for some of the genes that The expression of FZD8 protein in CRL2335 cells was revealed significant changes in the expression and ana- knocked down using FZD8-siRNA to determine whether lyzed additional genes that code for Wnt receptors and inhibition of this protein could inhibit Wnt signaling and their ligand using real-time PCR. As shown in Fig. 1B, increases sensitivity to cisplatin plus TRAIL–induced apo- treatment did not change significantly the gene expression ptosis. The results presented in Fig. 2A suggest that the levels of most of the Wnt receptors and their ligands, inhibition of FZD8 by siRNA inhibited b-catenin and except for FZD8 and R-spondin-1 (RSPO-1), which is one survivin protein levels compared with random siRNA- of its ligands. It was previously shown that RSPOs bind to treated cells. However, maximum inhibition was observed FZD8 and low-density lipoprotein-related receptor (LRP)- in cells treated with FZD8-siRNA and cisplatin plus 5 and/or -6 to initiate downstream signaling that involves TRAIL. The results presented in Fig. 2B suggest that b-catenin, survivin, and c-myc, among others (17, 18). FZD8-siRNA in combination with cisplatin plus TRAIL Consistent with these observations, our data show an treatment significantly induced apoptosis, as compared increased expression of c-myc protein levels in cisplatin with cells treated with random siRNA and cisplatin plus plus TRAIL-treated TNBC cell lines compared with TRAIL. These results suggest that inhibition of FZD8 sen- untreated controls (Supplementary Fig. S1). Increase in sitizes the cells to cisplatin plus TRAIL–induced apoptosis.

494 Mol Cancer Ther; 12(4) April 2013 Molecular Cancer Therapeutics

FZD8 and Drug Resistance

Figure 2. FZD8 inhibition increases apoptosis in the presence of cisplatin plus TRAIL. A, Western blot analysis showing that FZD8 inhibition by siRNA reduces b-catenin and survivin protein levels in CRL2335 cells as compared with random siRNA–treated cells. Maximum inhibition was seen in cells transfected with FZD8 siRNA and treated with cisplatin plus TRAIL. B, FZD8 siRNA in combination with cisplatin plus TRAIL treatment signiﬁcantly increases apoptosis compared with random siRNA and cisplatin plus TRAIL–treated cells. Results are expressed as mean SE (, P < 0.05).

þ Cisplatin plus TRAIL treatment inhibits tumor transfected ALDH1 CRL2335 cells with cisplatin (4 mg/kg) growth as a consequence of non-TIC cell death and TRAIL (15 mg/kg) once a week for 3 weeks. We To determine the relevance of our in vitro data in a more confirmed the inhibitory effect of the combined treatment clinically related setting, we treated CRL2335 xenograft on the growth of the tumors, as shown in Fig. 3A and B. tumors formed by co-injection of ALDH1 and DsRed- However, this highly effective therapeutic effect, as docu- mented by a significant reduction in tumor growth, was not reflected in the reduction of the DsRed-labeled tumor area detected by whole-body fluorescent imaging. In fact, the red fluorescent areas in tumors from mice treated with cisplatin and TRAIL did not show a significant reduction with respect to fast-growing tumors in control mice (Fig. 4A and B). Homogenates obtained from the tumors showed no significant change in ALDH1 protein levels, one of the markers used for TICs, in cisplatin plus TRAIL– treated tumors compared with untreated tumors (Fig. 4C and D). As our previous studies have shown that cisplatin plus TRAIL treatment is very effective in eliminating TICs in vitro (19), we were interested in determining how this treatment could affect TICs in vivo. Therefore, we analyzed the expression of CD44 and CD24 in the tumors using immunohistochemistry, considering that the þ CD44 /CD24 /low phenotype has been associated with tumor-initiating properties. We observed a higher per- þ centage of CD44 /CD24 /low cells in tumors harvested from cisplatin plus TRAIL–treated animals than untreated animals (Fig. 5A and B). Moreover, the percentage of cells that immunostained positive for the anti-apoptotic protein and downstream target for Wnt/b-catenin signaling survivin in cisplatin plus TRAIL–treated tumors was significantly higher than in control tumors (Fig. 5A and B). Homogenates obtained from tumors were used to analyze the expression of FZD8 using Western blotting. As Figure 3. Efficacy of cisplatin plus TRAIL treatment on growth of tumor depicted in Fig. 6A and B, FZD8 protein levels were xenografts. A, significant inhibition of tumor xenografts formed by co- significantly increased in tumors treated with cisplatin þ injection of non-TICs (ALDH1 ) and DsRed-transfected TICs (ALDH1 ) plus TRAIL with respect to untreated tumors. To deter- and treated with cisplatin plus TRAIL with respect to those in mine whether FZD8 expression is localized to TICs, non- untreated control mice. Results are expressed as mean SE (, P < 0.05; , P < 0.005). Arrows indicate days of treatment. B, representative TICs, or both, we analyzed the expression of ALDH1 and pictures of untreated and cisplatin and TRAIL–treated tumor-bearing FZD8 in tumors harvested from cisplatin plus TRAIL– mice 20 after treatment initiation showing differences in tumor size. treated and untreated control tumors. We observed that

www.aacrjournals.org Mol Cancer Ther; 12(4) April 2013 495

Yin et al.

Figure 4. Cisplatin plus TRAIL reduces in vivo tumor volume by increasing non-TICs death. A, representative pictures of whole- body fluorescent imaging showing that despite the reduction in tumor volume reflected by dial caliper measurements (Fig. 3), DsRed- labeled tumor area is not reduced after treatment with cisplatin and TRAIL. B, quantification of DsRed expression in the tumors as integrated intensity in the untreated and treated groups. Results are expressed as mean SE. C, Western blot showing ALDH1 protein levels in untreated and treated tumors on day 20 of treatment. D, densitometric analysis of ALDH1 bands obtained with Western blotting showing relative levels in untreated and treated groups after normalization for equal protein loading using actin as loading control. Results are expressed as mean SE.

FZD8 was expressed both in TICs as well as non-TICs (Fig. and paclitaxel, many affected women still experience a 6C and D). Taken together, our results suggest that cis- relapse, and metastatic breast cancer remains a largely platin plus TRAIL treatment significantly enhances cell incurable disease. death in non-TICs, whereas the effects on DsRed-expres- We have previously shown that cisplatin plus TRAIL sing TICs are minimal. treatment was most effective in eliminating both TICs and non-TIC cells compared with PARP inhibitors, cisplatin, Discussion paclitaxel, and docetaxel in vitro (19, 20). To identify the TNBC is the most aggressive form of breast cancer and major pathways involved in drug resistance to cisplatin chemotherapy is currently the mainstay of systemic treat- plus TRAIL treatment, here we analyzed gene expression ment for breast cancer, patients with TNBC disease have a profiles using CRL2335 TNBC cells before and after treat- worse outcome after chemotherapy than patients with ment with cisplatin plus TRAIL. The results from the other subtypes of breast cancer (4, 5). Despite of exciting present investigation suggest that cisplatin plus TRAIL new developments in treatment options including PARP treatment leads to a significant increase in gene expression inhibitors, cisplatin, angiogenesis inhibitor bevacizumab, of FZD8, and its downstream targets LEF1 and TCF7. In,

Figure 5. Cisplatin plus TRAIL treatment increases TICs and survivin in vivo. A, percentage of þ CD44 /CD24 /low cells used as a marker for TICs in tumors harvested from cisplatin plus TRAIL–treated and untreated tumors. In addition to an increase in TICs, cisplatin plus TRAIL–treated tumors show an increase in the antiapoptotic protein survivin compared with untreated tumors. Results are expressed as mean SE (, P < 0.05). B, representative immunostaining for CD44, CD24, and survivin in tumor xenografts from cisplatin plus TRAIL–treated mice. Bars, 200 mm.

496 Mol Cancer Ther; 12(4) April 2013 Molecular Cancer Therapeutics

FZD8 and Drug Resistance

Figure 6. Cisplatin plus TRAIL treatment increases FZD8 expression in vivo. A, Western blot analysis of tumor homogenates show signiﬁcant increase in FZD8 protein levels in tumors from cisplatin plus TRAIL–treated mice compared with tumors from untreated mice. B, densitometric analysis of FZD8 bands obtained with Western blotting showing relative levels in untreated and treated groups after normalization for equal protein loading using actin as loading control. Results are expressed as mean SE (, P < 0.05). Immunodetection of ALDH1 (C) and FZD8 (D) in a representative tumor specimen harvested from a mouse treated with cisplatin and TRAIL. Red arrows show positive immunostaining for both ALDH and FZD8, whereas black arrows indicate cells that are positive for FZD8 but negative for ALDH1. Bars, 50 mm.

addition, we found that RSPO-1 gene expression is also shrinkage may not always correlate with clinical benefits increased in cisplatin plus TRAIL–treated CRL2335 cells, because TICs enriched in residual tumors after treatment suggesting the possibility that FZD8-mediated Wnt sig- may reconstitute the primary tumor leading to tumor naling may play a major role in drug resistance in this recurrence, drug resistance, and metastasis. Consistent system. It was previously shown that RSPOs can physi- with our observation, some studies have suggested that cally interact with the extracellular domains of LRP6/ there is only a modest overall survival advantage for LRP5 and FZD8 and activate Wnt reporter genes (17). patients with metastatic breast cancer even when tumor Consistent with these observations, our data showed that regression occurs (23–25). inhibition of FZD8 by siRNA significantly reduces b-cate- Previous studies have shown the involvement of sev- nin, TCF/LEF, and survivin protein expressions in the eral FZD receptors such as FZD2, FZD4, FZD7, and FZD10 presence of cisplatin plus TRAIL. We have previously in different types of cancers (26–30). It was recently shown shown that suppression of survivin significantly increases that FZD8 plays a major role in human lung cancer, and its TRAIL-induced cell death in CRL2335 TNBC cells (19). inhibition by FZD8-shRNA was shown to sensitize lung Different groups, in addition to our group, reported that cancer cells to taxotere chemotherapy (31). Recently, it increased survivin expression enhances tumor resistance was reported that Wnt signaling can be inhibited using the to various apoptotic stimuli, primarily through caspase- extracellular cysteine-rich domain of FZD8 fused to dependent mechanism and its inhibition increases apo- human Fc domain. This soluble receptor-fused protein ptosis (19, 21). was shown to inhibit tumor growth in xenograft models Our study using a new model that allows tracking (32). Our in vitro and in vivo results indicate that FZD8- of dsRed-expressing cells with TIC characteristics in mediated Wnt signaling is upregulated in residual tumors vivo showed that treatment with cisplatin plus TRAIL after cisplatin and TRAIL treatment in TNBCs, leading to reduces xenograft tumor volume by predominantly an increase in b-catenin and survivin levels, and reduced eliminating non-TICs while having a minimal effect on apoptosis. DsRed-expressing cells. Consistent with these findings, In conclusion, our results suggest that the concept of immunostaining studies showed an increase in cells with equating tumor shrinkage with clinical benefits should be þ CD44 /CD24 /low phenotype, which is typically associ- revisited, considering that a possible enrichment of cells ated with tumor-initiating properties, in residual tumor with tumor-initiating–like characteristics may occur after xenografts after treatment with cisplatin plus TRAIL. The treatment of TNBCs, leading to more malignant chemore- commonly accepted criteria for clinical efficacy in phase II sistant tumors. On the basis of our findings, the use of trials are tumor shrinkage, as defined by Response Eval- novel therapies targeting FZD8-mediated Wnt signaling uation Criteria in Solid Tumors (RECIS; ref. 22), being the in combination with chemotherapeutic agents could be premise that tumor regression equates with the clinical beneficial to improve the treatment outcome of patients benefits. However, our observations suggest that tumor diagnosed with TNBCs.

www.aacrjournals.org Mol Cancer Ther; 12(4) April 2013 497

Yin et al.

Disclosure of Potential Conflicts of Interests Grant Support No potential conflicts of interest were disclosed. This work was supported in part by Department of Pathology, Authors' Contributions WSU and the NIH grant RO1 CA 118089 (K.B. Reddy). The micros- Conception and design: S. Yin, K.B. Reddy copy, imaging and cytometry resources core are supported, in part, Development of methodology: S. Yin, L. Xu, R. Bonfil, S. Banerjee, F.H. by NIH center grant P30CA22453 (G. Bepler) to The Karmanos Cancer Sarkar, S. Sethi, K.B. Reddy Institute, WSU and Perinatology research branch of the National Acquisition of data (provided animals, acquired and managed patients, Institutes of Child Health and Development, Wayne State University. provided facilities, etc.): L. Xu, S. Sethi, K.B. Reddy The costs of publication of this article were defrayed in part by the Analysis and interpretation of data (e.g., statistical analysis, biostatistics, payment of page charges. This article must therefore be hereby marked computational analysis): S. Yin, L. Xu, F.H. Sarkar, S. Sethi, K.B. Reddy advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate Writing, review, and/or revision of the manuscript: S. Yin, R. Bonfil, F.H. this fact. Sarkar, S. Sethi, K.B. Reddy Administrative, technical, or material support (i.e., reporting or orga- nizing data, constructing databases): L. Xu, K.B. Reddy Study supervision: K.B. Reddy Received November 12, 2012; revised January 22, 2013; accepted In vivo animal experimentation: S. Banerjee February 7, 2013; published OnlineFirst February 27, 2013.

References 1. Foulkes WD, Smith IE, Reis-Filho JS. Triple-negative breast cancer. 17. Nam JS, Turcotte TJ, Smith PF, Choi S, Yoon JK. Mouse cristin/R- N Engl J Med 2010;363:1938–48. spondin family proteins are novel ligands for the Frizzled 8 and LRP6 2. Cleator S, Heller W, Coombes RC. Triple-negative breast cancer: receptors and activate beta-catenin-dependent gene expression. therapeutic options. Lancet Oncol 2007;8:235–44. J Biol Chem 2006;281:13247–57. 3. Reddy KB. Triple-negative breast cancers: an updated review on 18. Wei Q, Yokota C, Semenov MV, Doble B, Woodgett J, He X. R- treatment options. Curr Oncol 2011;18:e173–9. spondin1 is a high affinity ligand for LRP6 and induces LRP6 phos- 4. Liedtke C, Mazouni C, Hess KR, Andre F, Tordai A, Mejia JA, et al. phorylation and beta-catenin signaling. J Biol Chem 2007;282: Response to neoadjuvant therapy and long-term survival in patients 15903–11. with triple-negative breast cancer. J Clin Oncol 2008;26:1275–81. 19. Xu L, Yin S, Banerjee S, Sarkar F, Reddy KB. Enhanced anticancer 5. Tan DS, Marchio C, Jones RL, Savage K, Smith IE, Dowsett M, et al. effect of the combination of cisplatin and TRAIL in triple-negative Triple negative breast cancer: molecular profiling and prognostic breast tumor cells. Mol Cancer Ther 2011;10:550–7. impact in adjuvant anthracycline-treated patients. Breast Cancer Res 20. Yin S, Xu L, Bandyopadhyay S, Sethi S, Reddy KB. Cisplatin and TRAIL Treat 2008;111:27–44. enhance breast cancer stem cell death. Int J Oncol 2011;39:891–98. 6. Rodriguez-Pinilla SM, Sarrio D, Honrado E, Hardisson D, Calero F, 21. Altieri DC. Validating survivin as a cancer therapeutic target. Nat Rev Benitez J, et al. Prognostic significance of basal-like phenotype and Cancer 2003;3:46–54. fascin expression in node-negative invasive breast carcinomas. Clin 22. Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford Cancer Res 2006;12:1533–9. R, et al. New response evaluation criteria in solid tumours: revised 7. Al-Hajj M, Becker MW, Wicha M, Weissman I, Clarke MF. Therapeutic RECIST guideline (version 1.1). Eur J Cancer 2009;45:228–47. implications of cancer stem cells. Curr Opin Genet Dev 2004;14:43–7. 23. Abratt RP, Brune D, Dimopoulos MA, Kliment J, Breza J, Selvaggi FP, 8. Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF. et al. Randomised phase III study of intravenous vinorelbine plus Prospective identification of tumorigenic breast cancer cells. Proc Natl hormone therapy versus hormone therapy alone in hormone-refractory Acad Sci U S A 2003;100:3983–8. prostate cancer. Ann Oncol 2004;15:1613–21. 9. Creighton CJ, Li X, Landis M, Dixon JM, Neumeister VM, Sjolund A, 24. Huff CA, Matsui W, Smith BD, Jones RJ. The paradox of response and et al. Residual breast cancers after conventional therapy display survival in cancer therapeutics. Blood 2006;107:431–4. mesenchymal as well as tumor-initiating features. Proc Natl Acad Sci 25. Huff CA, Matsui WH, Douglas Smith B, Jones RJ. Strategies to U S A 2009;106:13820–5. eliminate cancer stem cells: clinical implications. Eur J Cancer 10. Li X, Lewis MT, Huang J, Gutierrez C, Osborne CK, Wu MF, et al. 2006;42:1293–7. Intrinsic resistance of tumorigenic breast cancer cells to chemother- 26. Fukukawa C, Nagayama S, Tsunoda T, Toguchida J, Nakamura Y, apy. J Natl Cancer Inst 2008;100:672–9. Katagiri T. Activation of the non-canonical Dvl-Rac1-JNK pathway by 11. Ginestier C, Liu S, Diebel ME, Korkaya H, Luo M, Brown M, et al. Frizzled homologue 10 in human synovial sarcoma. Oncogene 2009; CXCR1 blockade selectively targets human breast cancer stem cells in 28:1110–20. vitro and in xenografts. J Clin Invest 2010;120:485–97. 27. Jin X, Jeon HY, Joo KM, Kim JK, Jin J, Kim SH, et al. Frizzled 4 regulates 12. Hirsch HA, Iliopoulos D, Tsichlis PN, Struhl K. Metformin selectively stemness and invasiveness of migrating glioma cells established by targets cancer stem cells, and acts together with chemotherapy to serial intracranial transplantation. Cancer Res 2011;71:3066–75. block tumor growth and prolong remission. Cancer Res 2009;69: 28. Ueno K, Hazama S, Mitomori S, Nishioka M, Suehiro Y, Hirata H, et al. 7507–11. Down-regulation of frizzled-7 expression decreases survival, invasion 13. Al-Hajj M, Clarke MF. Self-renewal and solid tumor stem cells. Onco- and metastatic capabilities of colon cancer cells. Br J Cancer 2009; gene 2004;23:7274–82. 101:1374–81. 14. Charafe-Jauffret E, Ginestier C, Iovino F, Tarpin C, Diebel M, Esterni B, 29. Yang L, Wu X, Wang Y, Zhang K, Wu J, Yuan YC, et al. FZD7 has a et al. Aldehyde dehydrogenase 1-positive cancer stem cells mediate critical role in cell proliferation in triple negative breast cancer. Onco- metastasis and poor clinical outcome in inflammatory breast cancer. gene 2011;30:4437–46. Clin Cancer Res 2010;16:45–55. 30. Zeng G, Germinaro M, Micsenyi A, Monga NK, Bell A, Sood A, et al. 15. Croker AK, Goodale D, Chu J, Postenka C, Hedley BD, Hess DA, et al. Aberrant Wnt/beta-catenin signaling in pancreatic adenocarcinoma. High aldehyde dehydrogenase and expression of cancer stem cell Neoplasia 2006;8:279–89. markers selects for breast cancer cells with enhanced malignant and 31. Wang HQ, Xu ML, Ma J, Zhang Y, Xie CH. Frizzled-8 as a putative metastatic ability. J Cell Mol Med 2009;13:2236–52. therapeutic target in human lung cancer. Biochem Biophys Res Com- 16. Ginestier C, Hur MH, Charafe-Jauffret E, Monville F, Dutcher J, Brown mun 2012;417:62–6. M, et al. ALDH1 is a marker of normal and malignant human mammary 32. DeAlmeida VI, Miao L, Ernst JA, Koeppen H, Polakis P, Rubinfeld B. stem cells and a predictor of poor clinical outcome. Cell Stem Cell The soluble wnt receptor Frizzled8CRD-hFc inhibits the growth of 2007;1:555–67. teratocarcinomas in vivo. Cancer Res 2007;67:5371–9.

498 Mol Cancer Ther; 12(4) April 2013 Molecular Cancer Therapeutics

Tumor-Initiating Cells and FZD8 Play a Major Role in Drug Resistance in Triple-Negative Breast Cancer

Shuping Yin, Liping Xu, R. Daniel Bonfil, et al.

Mol Cancer Ther 2013;12:491-498. Published OnlineFirst February 27, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-12-1090

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2013/03/06/1535-7163.MCT-12-1090.DC1

Cited articles This article cites 32 articles, 12 of which you can access for free at: http://mct.aacrjournals.org/content/12/4/491.full#ref-list-1

Citing articles This article has been cited by 1 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/12/4/491.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://mct.aacrjournals.org/content/12/4/491. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.