ERBB2-Mediated Transcriptional Up-Regulation of the A5B1 Integrin Fibronectin Receptor Promotes Tumor Cell Survival Under Adverse Conditions

Research Article

ERBB2-Mediated Transcriptional Up-regulation of the A5B1 Integrin Fibronectin Receptor Promotes Tumor Cell Survival Under Adverse Conditions

Christian Spangenberg,1 Ekkehart U. Lausch,1 Tatjana M. Trost,1 Dirk Prawitt,1 Andreas May,1 RomyKeppler, 1 Stephan A. Fees,1 Dirk Reutzel,1 Carolin Bell,2 Steffen Schmitt,3 Ilka B. Schiffer,4 Achim Weber,5 Walburgis Brenner,6 Matthias Hermes,7 Ugur Sahin,2 O¨zlem Tu¨reci,2 Heinz Koelbl,4 Jan G. Hengstler,7 and Bernhard U. Zabel1

1Children’s Hospital, 2Department of Internal Medicine, 3FACS Core Facility, 4Department of Obstetrics and Gynecology, 5Institute of Pathology, 6Department of Urology, University of Mainz, Mainz, Germany; 7Center for Toxicology, Institute of Legal Medicine and Rudolf-Boehm Institute of Pharmacology and Toxicology, University of Leipzig, Leipzig, Germany

Abstract Introduction Oncogenic activation of the receptor tyrosine kinase ERBB2 is The ERBB2 proto-oncogene is a member of the EGFR gene a key event in the development of a number of epithelial family consisting of ERBB1 (EGFR or HER1), ERBB2 (HER2 or malignancies. In these tumors, high levels of ERBB2 are NEU), ERBB3 (HER3), and ERBB4 (HER4). The genes encode type I strongly associated with metastatic disease and poor progno- receptor tyrosine kinases (RTK, except for ERBB3, which codes for sis. Paradoxically, an inherent cellular response to hyper- a receptor harboring an impaired kinase-domain; ref. 1) that are mitogenic signaling by ERBB2 and other oncogenes seems to involved in the transmission of proliferative as well as differenti- be growth arrest, rather than proliferation. Molecular ation signals (2, 3). ERBB2 plays an important role in the characterization of this yet undefined antiproliferative state development of human epithelial neoplasias and high levels of in independent cell lines overexpressing either wild-type ERBB2 expression correlate with poor prognosis in breast and ERBB2 or the mutationally activated receptor unveiled a ovarian carcinoma (4, 5). Overexpression of the receptor was not dramatic induction of the A5B1 integrin fibronectin receptor. only described in human breast and ovarian carcinomas (6), but A5 Integrin up-regulation is mainly a transcriptional response also in other types of cancer, e.g., bladder carcinoma (7), non–small mediated by the hypoxia-inducible transcription factors (HIF), cell lung cancer (8), in tumors of the gastrointestinal tract (9), as leading to a massive increase in membrane-resident receptor well as in other human malignancies (10). At least for breast molecules and enhanced fibronectin adhesiveness of the carcinogenesis, it is believed that oncogenic activation of ERBB2 is respective cells. Functionally, ERBB2-dependent ligation of an early event because it could be found in preinvasive lesions as fibronectin results in improved survival of mammary adeno- well as in advanced tumors (11, 12). Aberrant ERBB2 activation is carcinoma cells under adverse conditions, like serum with- implicated in the tumorigenic process through its influence on drawal, hypoxia, and chemotherapy. HIF-1A is an independent proliferation, survival of tumor cells, angiogenesis, invasive growth, predictor of poor overall survival in patients with breast metastatic behavior, and therapy resistance (12–15). However, as cancer. In particular, HIF-1A overexpression correlates for some other proto-oncogenes, like RAS and RAF, inherent significantly with early local relapse and distant metastasis, cellular anticarcinogenic mechanisms have been identified pre- a phenotype also highly characteristic of ERBB2-positive venting aberrant growth of affected cells and leading to a tumors. As HIF-1A is known to be stabilized by ERBB2 proliferative halt on oncogenic signaling (ref. 16; recent reviews: signaling under normoxic conditions, we propose that A5 refs. 17, 18). integrin is a major effector in this regulatory circuit and may Exploiting the tetracycline expression system, we have previously represent the molecular basis for the HIF-1A-dependent shown that inducible hypermitogenic ERBB2 signaling results in aggressiveness observed in ERBB2-overexpressing breast growth arrest of MCF-7 breast carcinoma cells, and that the cyclin- carcinomas. Hypermitogenic ERBB2 signaling and tumor dependent kinase inhibitor, P21, is an important mediator of hypoxia may act synergistically to favor the establishment of this cellular response (16). Expression screening focused on the chemoresistant dormant micrometastatic cells frequently regulation of integrin extracellular matrix (ECM) receptors by observed in patients with breast cancer. This new insight oncogenic ERBB2 signaling revealed a strong up-regulation of could be the basis for additional approaches complementing several a-integrin subunits, including a5 integrin, in response to current cancer therapy. (Cancer Res 2006; 66(7): 3715-25) ERBB2 induction. Integrin ligation influences signal transduction cascades that promote cell proliferation, cell migration, cell survival, as well as angiogenesis. Unlike growth factor receptors, they lack intrinsic enzymatic activity but participate in signaling by a h Note: C. Spangenberg and E.U. Lausch contributed equally to this work. coclustering with kinases and adaptor proteins. The 5 1 integrin Requests for reprints: Christian Spangenberg, Children’s Hospital, University of heterodimer constitutes the major cellular receptor for the ECM Mainz, Obere Zahlbacher Strasse 63, 55131 Mainz, Germany. Phone: 49-6131-393-3339; component fibronectin. Although its effect in tumorigenesis seems Fax: 49-6131-393-0227; E-mail: [email protected]. I2006 American Association for Cancer Research. to be cell type–specific, there are several studies providing evidence doi:10.1158/0008-5472.CAN-05-2823 for a pro-tumorigenic role of a5h1 integrin engagement (ref. 19 and www.aacrjournals.org 3715 Cancer Res 2006; 66: (7). April 1, 2006

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2006 American Association for Cancer Research. Cancer Research references therein). It has recently been shown that adhesion to epithelial cells derived from normal mammary tissue obtained from a fibronectin mediated by the heterodimer is responsible for reduction mammoplasty were obtained from ATCC (CRL-8799) and improved survival of different cell lines, including breast carcinoma cultured as described for MCF-7 cells. cells, under adverse conditions (20–23). Furthermore, a5h1 integrin The wild-type ERBB2 cDNA was amplified using the following oligonucleotides as primers (ERBB2/Nhe-For, 5V-AAAAAGCTAGCAGCCG- fibronectin receptor ligation is the key mediator of cell survival in CAGTGAGCACCAT-3V; ERBB2/Nhe-Rev, 5V-AAAAAGCTAGCGGTTCACACT- an in vitro bone marrow metastasis model in which breast GGCACGTC-3V; restriction sites introduced for cloning are underlined). The carcinoma cells undergo a transition to a nonproliferative, dormant resulting cDNA was digested and inserted into the NheI site of pINSpBI- state in response to fibroblast growth factor 2 (FGF2) treatment NeuT/EGFP following removal of the NeuT cDNA. (24). For patients with breast cancer, the presence of single Retroviral gene transfer was done using high-titer retroviral stocks dormant cells in their bone marrow is indicative of fatal relapse generated by transient calcium phosphate transfection of BING ampho- including metastatic disease in bone, and ERBB2 has been tropic packaging cells with a pBabe-puro vector containing the NeuT cDNA implicated in the survival of such disseminated cells (25, 26). (pBabe-NeuT) or vector without insert (pBabe) as previously described In this work, we explore the molecular characteristics and (16). Early-passage 184B5 cells were incubated with titered retroviral super- A consequences of a hypermitogenic state induced by stringently natants in the presence of 8 g/mL polybrene for 12 hours. Twenty-four hours after infection, cells were selected for 5 days in the presence of controlled overexpression of ERBB2 in MCF-7 breast carcinoma 2 Ag/mL puromycin. Individual cell lines emerging after 2 weeks were cells. In expression analyses, we observe a strong induction of analyzed for ERBB2 expression by immunoblotting. ITGA5 transcripts and protein in response to oncogenic ERBB2 Luciferase reporter gene constructs harboring relevant parts of the signaling. We characterize this regulatory interaction and investi- 5V-flanking region of the ITGA5 gene (as described in ref. 27) were generated gate the functional effect on tumor cell physiology. Our experi- inserting a proofreading PCR-amplified DNA fragment of the respective ments identify the regulation of cellular adhesion to fibronectin via region (the primers used for amplification were ITGA5prom-F, 5V-GTTAC- hypoxia-inducible factor (HIF)-dependent a5h1 integrin induction TCGAGGTCTTGAACTCCTGGCCTCA-3V and ITGA5prom-R, 5V-GTTACTC- as a novel mechanism by which oncogenic ERBB2 signaling GAGCCGCTCTTCCCTGTCCTG-3V; restriction sites introduced for cloning promotes cellular survival under adverse conditions (serum are underlined) into the XhoI site of pGL2-basic (Promega, Mannheim, deprivation, hypoxia, chemotherapeutic drugs) in vitro, thus adding Germany). All constructs were verified by sequence analysis. Transfections for reporter gene analysis were carried out in 24-well tissue insight to the processes of ERBB2-driven mammary carcinogenesis culture vessels. One day before transfection, 3 Â 104 cells per well were at the molecular level. This regulatory circuit helps to explain plated for duplicate analyses. The next day, 900 AL of fresh DMEM/10% fetal hitherto uncharacterized consequences of oncogenic ERBB2 bovine serum (with or without doxycycline) per well were added, followed activation, especially during the early stages of the carcinogenic by the addition of 1.5 Ag pGL2-DNA (pGL2-basic or pGL2-ITGA5), and process, in which strong ERBB2 signals may lead to nondividing, 0.5 Ag of the secreted alkaline phosphatase (SEAP2) control plasmid latent tumor cells exhibiting improved survival characteristics. (BD Clontech, Heidelberg, Germany), which had been complexed with 5 AL FUGENE 6 transfection reagent (Roche, Mannheim, Germany) according to the manufacturer’s instructions. To transfect human embryonic kidney Materials and Methods (HEK) cells, 1 Ag of pGL2-DNA (pGL2-basic or pGL2-ITGA5), 1 Ag of pBabe- Antibodies and reagents. ERBB2 (sc-284) and actin (sc-1616)-specific DNA (pBabe or pBabe-NeuT), and 0.5 Ag of SEAP2 control plasmid were antibodies were obtained from Santa Cruz (Heidelberg, Germany), the a5 treated with FUGENE 6 accordingly. Cotransfection experiments using von integrin-specific antibody (610633) used for immunoblotting was purchased Hippel-Lindau (VHL) tumor suppressor protein expression constructs were from BD PharMingen (Heidelberg, Germany), and the h1 integrin-specific carried out in analogy: 0.5 Ag pGL2-DNA (pGL2-basic or pGL2-ITGA5), antibody was purchased from BioGenex (San Ramon, CA). The a5 integrin- 0.5 Ag pcDNA3 or pcDNA3-VHL, and 0.1 Ag SEAP2 control plasmid specific antibody used for flow cytometric detection as well as for function complexed with 3 AL of FUGENE 6 transfection reagent were added to blocking experiments was purchased from Calbiochem (Schwalbach, NeuT48 cells 16 hours prior to doxycycline induction. Reporter gene assays Germany; integrin a5 fibronectin receptor Ab-1, clone P1D6). Doxycycline were done 48 hours after transfection (in the case of VHL cotransfection was used as the inducer of ERBB2/NeuT in all experiments. It was obtained experiments, reporter gene assays were done 48 hours after the addition of as the hydrochloride salt from Sigma (Munich, Germany) and stored as a doxycycline, i.e., 64 hours following transfection). 200 Ag/mL aqueous stock solution at À20jC. Human plasma fibronectin Luciferase assays were done using the firefly luciferase assay kit was purchased from Calbiochem. RGD (GRGDSPK, Sigma) and RGE (Promega). SEAP assays used the Great EscAPe SEAP chemiluminescence (GRGESP, Takara Bio Inc., distributed by Cambrex, Potsdam, Germany) assay kit (BD Clontech). Relative light units (RLU) were calculated as the peptides were reconstituted in PBS according to the manufacturer’s quotient of luciferase and SEAP activities. All transfection experiments were instructions. Inhibitors PD98059, SB203580, and wortmannin were pur- done at least thrice with similar results. chased from Sigma, SP600125 and GO¨6976 were obtained from Calbiochem. Quantitative real-time reverse transcription-PCR. RNA isolation, They were dissolved as stock solutions in DMSO and added to a final reverse transcription, and subsequent quantitative real-time reverse concentration of 10 Amol/L (Wortmannin), 50 Amol/L (PD98059), transcription-PCR experiments were done essentially as described (16). 20 Amol/L (SB203580), 10 Amol/L (SP600125), and 5 Amol/L (GO¨6976). The primers used for amplification of ITGA5 were ITGA5-F (5V-GGGCCAA- Solutions of chemotherapeutic drugs [cisplatin (CDDP), 5-fluorouracil GACTTTCTTGCAG-3V)andITGA5-R(5V-TGAAGAATCCAAGCTTGTA- (5-FU)] were prepared by the pharmacy of the Mainz University Hospital. GAGG-3V). Expression of ITGB1 was quantified using the following Cell lines, cell culture, and transfections. The MCF-7/pTet-NeuT and primers and cycling conditions: ITGB1-F (5V-TACTTGTGAAGCCAG- enhanced green fluorescent protein (EGFP) cell lines harboring doxycycline- CAACG-3V), ITGB1-R (5V-GGGGTAATTTGTCCCGACTT-3V). Cycling condi- inducible expression constructs respond to doxycycline administration with tions were: for ITGA5 amplification 95jC for 2 seconds, 61jC for 5 seconds, expression of EGFP alone (MCF-7/pTet-EGFP) or in combination with a 72jC for 10 seconds, 40 cycles; for ITGB1 amplification 95jC for 2 seconds, mutationally activated ERBB2-isoform isolated from rat glioblastoma (MCF- 61jC for 5 seconds, 72jC for 13 seconds, 40 cycles; for pyruvate dehydro- 7/pTet-NeuT). They have been described in detail in ref. (16). MCF-7/pTet- genase h (PDH) amplification 95jC for 2 seconds, 61jC for 5 seconds, 72jC ERBB2 cell lines expressing the human ERBB2 protein in response to for 8 seconds, 40 cycles. Both programs were preceded by an initial doxycycline treatment were generated similarly. All MCF-7-derived cell lines incubation for 10 minutes at 95jC. Relative quantification of cDNA were kept in DMEM/10% fetal bovine serum in humidified 5% CO2 concentrations was done using the Data analysis function of the Roche atmosphere at 37jC during routine culture. 184B5 transformed mammary Molecular Biochemicals Light Cycler Software (V. 3.5). Integrin expression

Cancer Res 2006; 66: (7). April 1, 2006 3716 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2006 American Association for Cancer Research. ERBB2 Induces the a5b1 Integrin Fibronectin Receptor levels were calculated in relation to the expression level of PDH as the preincubating the cells either with a function-blocking monoclonal reference gene [primers for PDH amplification were PDH-F (5V-GGAGTT- antibody against a5 integrin (clone P1D6), or nonspecific isotype control GAATGTGAGGTGATAAA-3V) and PDH-R (5V-ACGCAGGACCTTCCATGAT- antibody for 30 minutes at 37jC before plating. Fibronectin specificity was 3V)]. For quantification using real-time reverse transcription-PCR, at least documented by adhesion blockade, preincubating the cells either with an two independent experiments with duplicate amplifications were done. RGD- or with an RGE-containing peptide. Immunoblotting. Protein extract preparation, immunoblotting, and For the analysis of chemoresistance, experimental media containing antibody detection was done as described in ref. (16). a5- and h1 integrin- either specific chemotherapeutic agents dissolved in serum-reduced specific antibodies were used as a 1:1,000 dilution. OptiMEM I (Invitrogen, Karlsruhe, Germany) or control media without Flow cytometry. For flow cytometric detection of a5integrin, drugs were added after cell adherence. Serial dilutions were done to subconfluent cells that were grown for 48 hours in the presence or absence determine and span the effective dose range for each drug. Drugs in use for of doxycycline in 75 cm2 tissue culture flasks were harvested and washed clinical breast cancer were selected, including antimetabolites (5-FU) and twice with PBS/EDTA. After incubation in PBS/1% bovine serum albumin platinum compounds (CDDP). Following incubation for 72 hours at 37jC

(BSA) with the primary antibody specific for the a5 integrin protein (1:100 in a humidified atmosphere containing 5% CO2, plates were analyzed by dilution) for 30 minutes at room temperature, the cells were washed thrice adding the WST-1 reagent as described above. The fraction of surviving cells with PBS/1% BSA at 4jC. The secondary antibody (phycoerythrin- relative to control were plotted against the log of drug concentration, and conjugated goat anti-mouse IgG; obtained from Santa Cruz) was used in the IC50 was interpolated from the sigmoidal curve resulting from nonlinear a 1:200 dilution in PBS/1% BSA and the resulting cell suspension was kept regression applying a modified Hill model (28). All assays described were at room temperature for 20 minutes. Fluorescence-activated cell sorting done at least thrice with similar results. (FACS) analysis (FACSCalibur, Becton Dickinson, Heidelberg, Germany) was done after three additional washes in PBS/1% BSA at 4jC. The primary antibody was omitted in controls to check for background fluorescence in Results order to allow assessment of basal cell surface a5 integrin levels. Identification of A-integrins as targets of oncogenic ERBB2 Adhesion assay and quantification of living cells. Fibronectin-coating signaling. Specific interactions of tumor cells and surrounding of tissue culture vessels was done by adding 50 AL of a fibronectin solution ECM via cellular integrin receptors have been shown to have A (50 g/mL) per well of a 96-well tissue culture plate and a 16-hour profound effects on carcinogenesis due to the modulation of j incubation at 4 C. After removal of excess fibronectin solution, a 10 mg/mL cellular variables influencing tumor cell physiology, like prolifera- BSA in PBS solution was added, and the plates were kept at 25jC for an tion, survival, as well as invasion and metastatic spread. Recent additional hour. Subsequently, the wells were washed once with PBS and directly used for adhesion or survival assays. For profiling and adhesion evidence suggests that interactions between integrin ECM assays, subconfluent cells were cultured in the presence or absence of receptors and growth factor receptors play a substantial role in doxycycline for 48 hours, harvested by incubation in PBS/EDTA for f10 carcinogenesis (29–31). Because overexpression of the RTK ERBB2 minutes and washed once more in PBS/EDTA. Following an additional wash is causally involved in the development of f30% of mammary in serum-free DMEM, the cells were resuspended at a final concentration of carcinomas as well as several other epithelial malignancies, we 2 Â 105 cells per mL in DMEM. examined whether aberrant ERBB2 signaling results in the Initial profiling experiments determining the expression of a-integrin deregulation of integrin receptor expression in tumor cells. We a subunits in NeuT48 and EGFP control cells were done using the -integrin therefore analyzed MCF-7 cells conditionally expressing ERBB2 (16) mediated cell adhesion array (Chemicon, Hofheim, Germany) according to for an ERBB2-dependent response of a subset of relevant a-integrin the manufacturer’s instructions. Cells (2 Â 104; 100 AL) were added to the subunits. As shown in Fig. 1, doxycycline administration led to reconstituted wells and allowed to bind to the wells coated with different a a a a antibodies for 2 hours at 37jC. Following aspiration of the medium, wells selective up-regulation of 2, 3, 5, and V integrin proteins only were washed twice with assay buffer, 100 AL crystal violet containing Cell in NeuT48 cells, suggesting a role for oncogenic ERBB2 signaling Stain Solution were added to each well and bound cells were incubated for in regulating the tumor cell integrin receptor repertoire. Specific 5 minutes at room temperature. Subsequently, the staining solution was adhesion to fibronectin via the a5h1 integrin heterodimer has removed, the cells were washed four times with deionized water, air-dried, recently been shown to promote survival of nonproliferative, and cell-bound dye was extracted by the addition of 100 AL extraction dormant breast carcinoma cells under conditions of bone marrow buffer to each well. The amount of solubilized dye was quantified using a metastasis (24). Given this apparent analogy to our observations MRX II Elisa reader (Dynex Technologies, Inc., Chantilly, VA). regarding the effect of oncogenic ERBB2 signal transduction in A For adhesion assays, 100 L of the cell suspensions were plated as mammary adenocarcinoma cells (i.e., RTK-signaling provoking a quadruplicates in a 96-well tissue culture plate previously coated with proliferative arrest and a concomitant regulation of integrins) fibronectin, BSA, or uncoated, and left at 37jC for 1 hour. Subsequently, we looked more carefully for expression of the a5h1 integrin nonadherent cells were removed by three washings with PBS, and 100 AL DMEM per well were added. To quantify adherent cells, 10 ALofWST-1 fibronectin receptor. We therefore analyzed induced and unin- reagent (Roche) were added to the wells, and the plate was kept at 37jCfor duced NeuT48 as well as EGFP10 control cells by flow cytometry 1 hour, followed by photometric quantification in a MRX II Elisa reader for extracellular a5 integrin. As depicted in the histogram plot of (Dynex). Quantification of plated cells by omitting the washings controlled Fig. 1B, a strong increase in membrane-resident a5 integrin protein for differences in the number of respective cell populations. could be detected in NeuT48 cells in response to hypermitogenic To quantify the influence of ERBB2-dependent a5h1 integrin induction ERBB2 signaling: The mean relative fluorescence of doxycycline- on cell growth (as the combined effect of proliferation and survival) under induced NeuT48 cells was 127.67, nearly 4-fold higher compared adverse conditions, cells grown in the presence or absence of doxycycline with uninduced NeuT48 or EGFP10 control cells. Taken together, for 30 hours were harvested and washed as described for adhesion assays. these results clearly show that oncogenic ERBB2 signaling leads to 5 Â 103 cells (100 AL) per well were plated as triplicates in pretreated a selective up-regulation of specific a-integrin subunits in the 96-well tissue culture vessels (day 0) and kept in DMEM (with or without plasma membrane of breast carcinoma cells, including the a5h1 doxycycline) in a 5% CO2-humidified atmosphere or hypoxia (5% CO2, 0.1% integrin fibronectin receptor. O2, 94.5% N2). At the indicated time periods, 10 AL of WST-1 reagent (Roche) were added to quantify living cells. Wells of the plating control were Identification of ITGA5 and ITGB1 as target genes of quantified directly following plating on day 0. The specificity of a5h1 oncogenic ERBB2 signaling in mammary adenocarcinoma integrin ligation on cell growth kinetics was shown by adhesion blockade, cells. Given the increase of membrane-resident a5 integrin protein, www.aacrjournals.org 3717 Cancer Res 2006; 66: (7). April 1, 2006

Figure 1. Oncogenic ERBB2 signaling stimulates cell surface expression of selective a-integrin subunits in mammary adenocarcinoma cells. A, EGFP10 control cells and NeuT48 cells grown for 48 hours in the presence or absence of doxycycline were incubated in wells coated with different a-integrin monoclonal antibodies indicated below the diagram, followed by washing and subsequent staining, dye extraction and photometric determination of bound cells as indicated in Materials and Methods. Columns, OD at 570nm, which is proportional to the number of bound cells; results of a representative experiment. The negative control well (nc) contains the goat anti-mouse coating antibody. B, EGFP10 control cells (left) and NeuT48 cells (right) grown for 48 hours in the presence (black columns) or absence (open columns) of doxycycline were subjected to FACS analysis following incubation with an a5h1 integrin-specific primary and phycoerythrin-conjugated secondary antibody. Histograms, number of cells displaying the respective fluorescence intensity given on the horizontal axis (which is proportional to the a5h1 integrin expression level); dotted line, distribution of signals detected in control stainings (uninduced cells, primary antibody omitted) to allow assessment of the basal level of cell surface a5 integrin expression.

we next examined whether this up-regulation is the result of an increasing amounts of NeuT expression plasmid were cotransfected increase in transcripts and/or protein, and whether the hetero- (Fig. 2B); thus, demonstrating that the regulatory circuit medi- dimerization partner of a5 integrin, the h1 integrin subunit, is also ating ERBB2-stimulated a5 integrin expression is functional in a induced in an ERBB2-dependent manner. As shown in Fig. 2, we different cell type. detected a strong increase in transcripts of both genes on ITGA5 mRNA induction by hypoxia has previously been shown induction of oncogenic ERBB2 signaling in several independent in several cell lines, including normal and transformed keratino- MCF-7/pTet-NeuT cell lines (Fig. 2A). This increase in ITGA5 and cytes (32), colon cancer cells (33), as well as cell lines derived ITGB1 RNA abundance resulted in up-regulation of the respective from lung and breast cancers (34). HIF is a major transcriptional proteins, as shown in the immunoblots of Fig. 2B. In addition, regulator responsible for a5 integrin induction under conditions of induction of a5 integrin expression in response to ERBB2 signaling low oxygen (33). Furthermore, HIF is described to be an impor- was also observed in 184B5 immortalized mammary epithelial cells tant mediator of ERBB2-dependent transcriptional responses, constitutively overexpressing the oncogenically activated rat NeuT, even under normoxic conditions (35–37). We therefore wanted to as well as in MCF-7 cells inducibly expressing wild-type ERBB2 address the question of whether ERBB2 signaling uses HIF to protein (Fig. 2C). The NeuT48 cells were used to further char- induce transcription of the ITGA5 gene. The von Hippel-Lindau acterize the a5 integrin response following doxycycline-induction tumor suppressor (pVHL) is the adaptor protein of an E3 ubiquitin in a time course experiment. As shown in Fig. 2D, a significant ligase complex which, under normoxic conditions, efficiently elevation of a5 integrin protein expression could be detected 24 targets the a subunits of the HIF for proteasomal degradation. hours after induction of oncogenic ERBB2 signaling. Our data thus Hence, we used pVHL expression constructs to reduce endogenous clearly illustrates that ERBB2 signaling stimulates expression of the HIF activity in ERBB2 overexpressing cells and analyzed the ITGA5 and ITGB1 genes encoding the major cellular fibronectin consequences for ITGA5 promoter activity. As shown in Fig. 3C, receptor in mammary epithelial cell lines. expression of pVHL leads to a significant decrease of ITGA5 ERBB2 activates the A5 integrin promoter via HIF. To promoter activity in the absence, as well as in the presence, of determine whether the strong ERBB2-dependent increase of ERBB2 signaling. Moreover, most of the ITGA5 transcription ITGA5 transcripts is mediated through stimulation of the ITGA5 in response to ERBB2 seems to be controlled by HIF, as can be promoter, we analyzed the effect of oncogenic ERBB2 signaling on concluded from the reduction of ITGA5 promoter activity down an ITGA5 promoter/luciferase reporter construct (pGL2-A5). to the level of uninduced (ÀDox) NeuT48 cells when VHL is As shown in Fig. 3A, doxycycline-mediated induction of ERBB2 cotransfected (Fig. 3C). signaling leads to a massive (>13-fold) increase in luciferase activity In additional experiments, we examined the effect of small in NeuT48 cells, whereas no stimulation was observed when molecule inhibitors on ERBB2-dependent ITGA5 transcription to control cells (EGFP10) were transfected with the promoter- get a first impression about the pathways involved in HIF-mediated construct. Furthermore, we tested whether the influence of NeuT induction of a5 integrin following oncogenic ERBB2 signaling. As on the ITGA5 promoter is observed in a different cell line (293 HEK can be assessed from Fig. 3D, application of GO¨6976 prevented cells), and found increasing ITGA5 promoter activity when ITGA5 induction, whereas Wortmannin and SB203580 led to a

Cancer Res 2006; 66: (7). April 1, 2006 3718 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2006 American Association for Cancer Research. ERBB2 Induces the a5b1 Integrin Fibronectin Receptor partial decrease of ITGA5 transcripts, indicating a major contri- wanted to test whether ERBB2 signal transduction and low bution of protein kinase C (PKC) isoforms to ERBB2-mediated O2 tension act synergistically on ITGA5 transcription. For this ITGA5 induction. purpose, we first looked at whether hypoxic regulation of a5 Taken together, our data show conclusively that ERBB2 signaling integrin expression is functional in MCF-7 adenocarcinoma and uses the transcriptional regulator HIF to activate the promoter of 184B5 transformed mammary epithelial cells. As shown in Fig. 4A, the ITGA5 gene, with PKC, phosphoinositide-3-kinase (PI3K)/Akt, an increase of ITGA5 transcripts was observed by reducing oxygen and P38 being involved in this regulatory circuit. tension in both mammary cell lines. In contrast, VHL-deficient ERBB2 and hypoxia synergistically activate A5 integrin renal carcinoma cell lines (CaKi2, 786-O) did not respond to transcription. Given the convergence of growth factor receptor hypoxia by up-regulating the ITGA5 gene (data not shown). The (ERBB2) and hypoxic signal transduction pathways in HIF, we most likely explanation for this observation is that HIF-a is

Figure 2. Oncogenic ERBB2 signaling up-regulates the a5h1 integrin fibronectin receptor. A, independent MCF-7/NeuT cell lines (NeuT3, NeuT15, and NeuT48) and MCF-7/EGFP control cells (EGFP10) were grown for 48 hours in the absence (open columns) or presence (black columns) of doxycycline. Total RNA was subjected to quantitative real-time reverse transcription-PCR analysis of ITGA5 expression (left) and ITGB1 expression (right). Columns, means of relative integrin mRNA expression levels, results of a representative experiment; bars, SD. B, protein extracts isolated from the cell lines shown in (A) were subjected to SDS-PAGE, blotted and probed for a5 integrin (left) as well as h1 integrin expression (right). Actin-specific staining is shown below as loading control. C, protein extracts isolated from MCF-7 cell lines conditionally expressing human wild-type ERBB2 (left), and 184B5 human transformed mammary epithelial cells constitutively expressing activated NeuT (right) were subjected to SDS-PAGE and Western blotting. Immunodetection of ERBB2/NeuT- (top), a5 integrin- (middle), and actin expression (bottom). MCF-7/ ERBB2 cell lines were grown for 48 hours in the presence (+) or absence (À)of doxycycline. 184B5-V, extracts from empty vector-transfected control cells. D, protein extracts of NeuT48 cells cultured in doxycycline-containing DMEM/10% fetal bovine serum for the indicated time periods were blotted and probed for a5 integrin expression. As indicated, NeuT-expression starts around 6 hours following doxycycline induction (16). Actin-specific staining is shown as the loading control.

www.aacrjournals.org 3719 Cancer Res 2006; 66: (7). April 1, 2006

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2006 American Association for Cancer Research. Cancer Research constitutively stabilized in CaKi2 and 786-O cells, therefore a5 response to ERBB2 activation prompted us to test whether the integrin expression cannot be up-regulated further by reducing cells exhibit an increased adhesiveness towards fibronectin, and oxygen tension. Luciferase reporter gene assays analyzing the whether this has any functional consequences regarding tumor cell transcriptional hypoxia response of the ITGA5 promoter also pathophysiology. We therefore induced cells to express NeuT for showed a hypoxia-dependent up-regulation of reporter gene 2 days and quantified their adhesion on uncoated (standard tissue activity in MCF-7 cells (Fig. 4B). We then compared the effect of culture plastic), BSA-coated, as well as fibronectin-coated tissue ERBB2 signaling on a5 integrin induction under normoxic and culture vessels. As depicted in Fig. 5A, uninduced NeuT48 cells hypoxic conditions. The results of quantitative real-time reverse behave like uninduced and induced control cells in showing no transcription-PCR experiments depicted in Fig. 4C show that specific change in adhesiveness towards any of the surfaces. hypoxia (0.1% O2, 24 hours) in the absence of NeuT signaling, and However, when NeuT48 cells were induced to express oncogenic NeuT induction (36 hours, doxycycline) under normoxic conditions ERBB2, the fraction of cells adhering to fibronectin significantly led to a moderate increase of relative ITGA5 transcript abundance increased by 89% (from 37% to 70% of cells adhering to (f6-fold and f17-fold, respectively) when compared with the fibronectin-coated wells; Fig. 5A,right ). This increase in fibronectin uninduced (ÀDox) state under normoxic conditions. However, adhesiveness could be substantially blocked by pretreatment of when both stimuli were applied (36 hours, doxycycline; 0.1% O2 induced cells with an a5h1 integrin-specific antibody (reducing the for 24 hours starting 12 hours after doxycycline administration), a ERBB2-mediated increase in adhesiveness by about 60%; data not >50-fold elevation of relative ITGA5 transcript levels compared shown), suggesting that the a5h1 integrin fibronectin receptor is with normoxic/ÀDox culture conditions (Fig. 4C) was observed, the major cause of increased fibronectin adhesion of mammary demonstrating a synergistic effect of hypermitogenic ERBB2 signal adenocarcinoma cells on oncogenic ERBB2 signaling. transduction and low oxygen tension on expression of the a5 To test the influence of ERBB2-dependent increase in fibronec- integrin subunit. tin-adhesion on survival, we measured the relative number of living ERBB2-mediated up-regulation of the A5/B1integrin cells as the net effect of cell proliferation and cell death under fibronectin receptor results in increased fibronectin adhesion, conditions of serum withdrawal as well as hypoxia when cultured thereby promoting tumor cell survival under adverse con- on untreated or fibronectin-coated cell culture vessels. The growth ditions and resistance to chemotherapeutic drugs. Adhesion to curves depicted in Fig. 5B and C clearly show an ERBB2-dependent fibronectin has been shown to confer a selective survival advantage prosurvival effect that is dependent on the presence of a fibro- to tumor cells under various ‘‘adverse’’ conditions (20, 22–24, 38, 39). nectin matrix. As shown in Fig. 5B, the presence of fibronectin The up-regulation of the a5h1 integrin fibronectin receptor in leads to a significant increase in the number of oncogenic

Figure 3. ERBB2 signaling activates the ITGA5 promoter using HIF. A, EGFP10control cells and NeuT48 cells were transfected with the promoterless pGL2-basic vector (pGL2) or with a reporter gene construct carrying the proximal promoter region of the ITGA5 gene (pGL2-ITGA5). Reporter gene analysis after 48 hours of culture in the absence or presence of doxycycline reveals a strong induction of reporter gene activity (expressed as RLU) only in doxycycline-exposed NeuT48 cells transfected with the pGL2-ITGA5 plasmid. Columns, mean values of a representative experiment done in duplicate; bars, SD. B, cotransfection of HEK cells with pGL2-ITGA5 and increasing amounts of the pBabe-NeuT expression plasmid (as indicated below the histogram) yields increasing activity of the ITGA5 promoter. Columns, mean values of a representative experiment done in duplicate; bars, SD. C, NeuT48 cells were cotransfected with pGL2-ITGA5 and either empty (CMV-) or VHL-containing (CMV-VHL) expression plasmids, and the influence of VHL-expression on ERBB2-mediated activation of the ITGA5 promoter was examined in reporter gene analyses. The results show a significant effect of VHL expression on basal (ÀDox) as well as ERBB2-stimulated (+Dox) reporter gene activity, suggesting a prominent role of HIF in ERBB2-mediated regulation of a5 integrin expression. Columns, mean values of a representative experiment done in duplicate; bars, SD. D, quantitative real-time reverse transcription-PCR analysis (duplicate readings) of ITGA5 induction after treatment of NeuT48 cells with specific kinase inhibitors in the presence or absence of doxycycline. Inhibitors of PI3K (Wort), ERK1/2 (PD), P38 (SB), c-Jun-NH2-terminal kinase (SP), and PKC (GO¨ ), or DMSO (vehicle control) were added to the cells alone or together with doxycycline, and cells were cultured for 36 hours. Relative ITGA5 transcript levels are depicted with mean values in the absence of doxycycline set to 1. Columns, means of a representative experiment; bars, SD.

Cancer Res 2006; 66: (7). April 1, 2006 3720 www.aacrjournals.org

cells was also shown to have a significant effect on survival under conditions of low oxygen. Pretreatment of NeuT48 cells (+Dox) with RGD peptide reduces the fraction of living cells cultured in hypoxia for 2 days by 49% compared to pretreatment with a control peptide (data not shown). Although even patients with advanced stage breast cancer initially display responses to chemotherapy, metastatic disease almost invariably relapses, becomes resistant to treatment and therefore remains incurable. One of the main factors contrib- uting to the failure of chemotherapy is that the cancer cells are, or become, resistant to drug-induced apoptosis. Members of the human epidermal growth factor receptor family have a major role in the chemoresistance of breast cancer cells (40, 41). Although ERBB2 overexpression has been suggested to confer resistance to both hormonal therapy as well as to chemotherapy, and its expression may also predict the response to certain chemotherapy, the underlying molecular mechanisms remain unclear (42, 43). On the other hand, for small cell lung cancer it was previously shown that adhesion to ECM within the tumor microenvironment protects tumor cells from cytotoxic drugs commonly used to treat patients with small cell lung cancer (39). This mechanism of primary resistance is mediated via signaling through integrins and may be responsible for minimal residual disease and relapse. We therefore hypothesized that ligation of the fibronectin receptor would, at least partially, mediate chemoresistance in the course of hypermitogenic ERBB2 signaling and examined the sensitivity of our MCF-7 lines to agents currently used for breast cancer chemotherapy. Compared with EGFP10 control cells, NeuT48 cells showed a remarkably increased resistance to 5-FU and CDDP (Fig. 5D). For both drugs, the dose required to kill 50% of ERBB2- expressing cells increased f2-fold compared with the uninduced state. This resistance was further enhanced >2-fold, Figure 4. ERBB2 signaling and hypoxia act synergistically on a5 integrin expression. A, real-time quantitative reverse transcription-PCR analysis specifically in the ERBB2-positive cells when attached to (duplicate readings) of ITGA5 mRNA expression in different mammary cell lines fibronectin. In contrast, the resistance conferred to EGFP cultured under normoxic or hypoxic conditions (0.1% O2) for 24 hours. The control cells by attachment to fibronectin was minor, whether relative ITGA5 mRNA abundance under normoxia was set to 1 for each cell line. a Columns, means of a representative experiment; bars, SD. B, MCF-7 cells were induced or uninduced. ERBB2-induced up-regulation of 5 transfected with either empty (pGL2) or promoter-containing (pGL2-ITGA5) integrin may thus be able to confer resistance to apoptosis, a plasmids, left under routine culture conditions for 16 hours, and then kept either phenomenon referred to as cell adhesion–mediated drug under normoxia or hypoxia (0.1% O2) for an additional 24 hours. Luciferase activities (expressed as RLU) as measured in the resulting cellular extracts resistance. Although it has not yet been clearly established are depicted as mean values of duplicate transfections. Columns, means; whether such a mechanism might account for the acquired bars, SD. C, real-time quantitative reverse transcription-PCR analysis (duplicate readings) of ITGA5 mRNA-expression in NeuT48 cells grown in the resistance found in many tumors it provides a good explana- absence or presence of doxycycline for a total of 36 hours. Where indicated, cells tion for local recurrence often seen clinically after chemother- were put into hypoxic atmosphere (0.1% O2, 24 hours) 12 hours after the addition apy. In conclusion, our analyses provide direct experimental of doxycycline. The relative ITGA5 mRNA abundance in NeuT48 cells grown under normoxia in the absence of doxycycline was set to 1 to assess evidence for an important effect of the ERBB2/fibronectin transcriptional induction due to the two stimuli (ERBB2-signaling, hypoxia) alone receptor regulatory circuit on tumor cell survival under (compare open versus filled column under normoxic conditions, or open columns physiologically and clinically relevant adverse conditions like under normoxic versus hypoxic conditions, respectively) or in combination (compare open column under normoxic versus filled column under hypoxic lack of growth factors, low oxygen tension, and chemotherapy. conditions). Columns, means of a representative experiment; bars, SD.

Discussion ERBB2-expressing (i.e., doxycycline-treated) NeuT48 cells over 6 Overexpression of the ERBB2 proto-oncogene within solid days of culture without serum. In contrast, no such (pronounced) tumors is associated with poor prognosis (4, 44). Breast cancer effect could be observed when doxycycline-treated NeuT48 cells cell micrometastasis, the dissemination of single, nonproliferating were cultured on untreated plastic, or when uninduced NeuT48 or (or dormant) tumor cells even in the earliest phases of mammary EGFP10 control cells were analyzed. Again, specifically blocking the carcinogenesis, has a major effect on patient survival (45), and a5h1 integrin fibronectin receptor lead to a significant decrease ERBB2-positive epithelial tumor cells in bone marrow characterizes of ERBB2-mediated effects, reducing the fraction of living cells a clinically relevant subset of breast cancer micrometastases (46). cultured without serum for 6 days by 44% compared with the effect Originally identified as a potent stimulator of mitosis, oncogenic of an isotype control antibody (data not shown). In analogy, ERBB2 signaling has recently been found responsible for the ERBB2-dependent fibronectin adhesion of MCF-7 breast carcinoma activation of an initial antiproliferative response in tumor cells (16). www.aacrjournals.org 3721 Cancer Res 2006; 66: (7). April 1, 2006

Figure 5. ERBB2 signaling induces increased fibronectin adhesiveness of breast carcinoma cells resulting in improved survival under adverse conditions and chemoresistance. A, adhesion of EGFP10control cells ( left) and NeuT48 cells grown for 48 hours in the absence (open columns) or presence (black columns) of doxycycline to different substrates. Percent cells adhering relative to the plating control was calculated for the respective surfaces [untreated tissue culture (standard TC), BSA-treated (BSA), fibronectin-coated (FN)]. Columns, means of quadruplicate readings; bars, SD. B, EGFP10control cells ( left) and NeuT48 cells (middle) were grown in the presence (filled symbols) or absence (open symbols) of doxycycline for 30 hours, plated (day 0, d0) on untreated (E) or fibronectin-coated wells (n), and kept under serum-free conditions for the indicated periods of time. The relative number of living cells represents the portion of living cells at a given time point (determined as described in Materials and Methods) expressed as a multiple of the number of cells plated. Columns, means of triplicate readings; bars, SD. C, the same experimental setup as in (B), but the cells were plated (day 0, d0) and kept under hypoxic conditions (0.1% O2) for the indicated time periods. The relative number of living cells represents the portion of living cells at the respective time point (determined as described in Materials and Methods) expressed as a multiple of the number of cells plated. Columns, means of triplicate readings; bars, SD. D, EGFP10control cells ( left) and NeuT48 cells (right) were grown in the presence (filled symbols) or absence (open symbols) of doxycycline for 30 hours, plated (day 0, d0) on untreated (E) or fibronectin-coated wells (n), and exposed to the indicated concentrations of cisplatin (CDDP; top) or 5-FU (bottom)for 72 hours. The fraction of living cells relative to the drug-free control samples was determined for each concentration using the WST-1 reagent as described. Columns, means of triplicate readings; bars, SD. Survival curves were plotted applying nonlinear regression assuming a modified Hill model (28); dashed lines, standard tissue culture vessels; solid lines, fibronectin-coated vessels. Resistance factors (RF) are computed by dividing the IC50 of the respective experimental conditions and cell lines.

Thus, it was suggested that an inherent anticarcinogenic program adenocarcinoma cells with inducible ERBB2 for the expression of must be overcome in the early stages of carcinogenesis to progress a-integrin subunits of cellular integrin heterodimers because to malignancy (reviewed in refs. 17, 18). integrins have been shown to significantly contribute to the patho- In an attempt to unravel the molecular circuitry of early steps in genesis of malignancies with respect to tumor cell survival and ERBB2-driven tumorigenesis, we screened MCF-7 mammary proliferation (29–31). Among the integrins displaying regulation by

Cancer Res 2006; 66: (7). April 1, 2006 3722 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2006 American Association for Cancer Research. ERBB2 Induces the a5b1 Integrin Fibronectin Receptor oncogenic ERBB2 signal transduction in our cells, we selected the among those genes most profoundly up-regulated in bone constituents of the major cellular fibronectin receptor (a5 and h1 marrow–positive primary tumors. The consistent regulation of integrin proteins) for further in-depth investigation, mainly several known HIF-1 target genes in the samples analyzed suggests because of its known effect on tumor cell survival and that pathways connected with HIF impinge on bone marrow proliferation. By specifically adhering to the ECM component metastasis (60). Whether HIF-dependent regulation of ECM fibronectin, the receptor improves the survival of different cell receptors is responsible for the association of HIF-1a over- types under adverse conditions (see refs. 20–24; see refs. 29, 47–50 expression and unfavorable prognosis (shorter overall survival, for reviews). Thus, adhesion on fibronectin prevents the apoptosis shorter disease-free survival; refs. 61–63) in patients with advanced of Chinese hamster ovary cells upon serum withdrawal, and this stage breast cancer remains to be determined. Other integrin ECM prosurvival/antiapoptotic effect of fibronectin-adhesion has been receptors, like a6h4 and a6h1 integrin heterodimers which are attributed to the selective up-regulation of the antiapoptotic BCL-2 associated with breast cancer progression have been shown to play protein, involving SHC, FAK, and RAS subsequently activating the important roles in cellular responses to hypoxia (64, 65). As HIF-1a PI3K/AKT pathway in these cells (20, 51). In the recent work of is known to be stabilized by ERBB2 signaling under normoxic Korah et al., the interaction of a5h1 integrin and fibronectin conditions (35–37), we propose that the a5h1 integrin fibronectin maintains survival of different growth-arrested mammary adeno- receptor is a major effector in this regulatory circuit and may carcinoma cells in an in vitro model system of bone marrow represent the molecular basis for the HIF-1a-dependent aggres- metastasis. Comparable to results obtained in our inducible cell siveness observed in ERBB2-overexpressing breast carcinomas. system, the group elegantly shows that selective up-regulation of According to this hypothesis, HIF-1a-dependent up-regulation the a5h1 integrin fibronectin receptor by FGF2 treatment— of the a5h1 integrin fibronectin receptor in response to hypoxia thereby simulating the conditions of bone marrow metastasis— (as shown in refs. 33, 34; and in the present study) and/or ERBB2 leads to long-term survival of dormant clones (24). (this study) might be one decisive factor selecting for living but Tumor cell dormancy is a concept used to characterize nondividing, i.e., dormant chemoresistant cells frequently observed nonproliferating single tumor cells that have been shed by the in the bone marrow of patients with breast cancer. The functional primary tumor and extravasated in a distant target organ (refs. relevance of the cellular a5h1 integrin fibronectin receptor for 52, 53; see refs. 25, 26, 54–56 for reviews). Metastatic growth of breast cancer cell dormancy has recently been described (24). tumor cells at distant sites is generally considered a late step Hence, we suggest that the ERBB2-HIF-fibronectin receptor during tumorigenesis, but current metastasis models hypothe- regulatory interaction is involved in increasing the portion of size that dissemination of single tumor cells begins even in the surviving hypoxic cells overexpressing ERBB2 described recently by very early phases of malignancy (25, 26, 57). Such disseminated Dragowska and coworkers (66). This notion is supported by our micrometastases can be found in a number of organs, like the functional studies demonstrating a prosurvival effect of ERBB2 bone marrow, as solitary cells, where they may survive for a long signaling under hypoxia that is dependent on fibronectin adhesion period of time, eventually leading to metastatic disease even (Fig. 5C). decades later. On the molecular level, dormancy is not well Resistance to chemotherapeutic or radiotherapeutic interven- characterized. One prominent feature of dormant cells is the tion has been associated with ERBB2 overexpression in tumors altered spectrum of integrin cell surface receptors involved in (41, 44, 67–70). Given the stimulation of fibronectin adhesiveness survival signaling. Engagement of specific signaling pathways, following oncogenic ERBB2 signaling, we also examined the like the activity ratio of ERKMAPK to P38MAPK,seemstobean influence of this interaction on tumor cell chemoresistance and important determinant of tumor cell fate—i.e., proliferative found that fibronectin ligation lead to a 5-fold increase in growth versus dormancy (58, 59). Interestingly, the a5h1integrin resistance against drugs used in breast cancer chemotherapy fibronectin receptor, in concert with uPAR, is known to protocols (5-FU and CDDP), further supporting the clinical modulate the respective mitogen-activated protein kinase relevance of the ERBB2/a5 integrin regulatory interaction. One (MAPK) activities and therefore represents a critical relay in hallmark of tumor cell dormancy is the inherent resistance of the tumor cell with respect to proliferation/dormancy (58). nondividing cells against chemotherapeutic agents (26, 55, 71). As Given that a5andh1 integrins, as shown in the present study, suggested by the work of Korah et al. (24), as well as by our are targets of oncogenic ERBB2 signaling, it is intriguing to functional in vitro studies, the a5h1 integrin/fibronectin interac- speculate that this regulatory circuit represents an important tion may be a (perhaps essential) marker for dormant cells, and mechanism of breast cancer cell dormancy in vivo,thatmight therefore a very interesting target for breast cancer therapy. Given even be augmented by additional growth factor signals [e.g., its well accessible cellular location, peptides targeting fibronectin FGF2 as documented by Korah et al. (24)]. The observation that adhesion via the a5h1 integrin heterodimer, or small-molecule the ERBB2 conditional cells display a profound activation of inhibitors targeting downstream survival signaling could be used as both relevant MAPK signal transduction cascades in response to therapeutic tools—even in combination with strategies directed oncogenic ERBB2 signaling, with p38 MAPK being the pathway against the ERBB2 RTK—to prevent signal transduction essential responsible for transducing the growth arrest signal (16), further for long-term survival of dormant tumor cells (29, 72). Further- supports this concept. more, the identification of HIF as a central mediator of ERBB2- Expression profiling experiments have identified a molecular dependent regulation of tumor cells’ fibronectin adhesiveness adds signature associated with bone marrow micrometastasis in human an attractive drug target (73) that might be of special importance, breast cancer (60). By using cDNA arrays to comparatively monitor given the synergistic activation of fibronectin receptor expression primary tumors of patients with bone marrow–positive and bone by ERBB2-signaling and low oxygen tension observed in the work marrow-negative breast cancer, the authors showed that differen- presented. tially expressed genes include those involved in ECM remodeling, In summary, we identify a5 integrin, as well as its heterodi- adhesion, and signal transduction. In this study, HIF-1a was merization partner, h1 integrin, as targets of oncogenic ERBB2 www.aacrjournals.org 3723 Cancer Res 2006; 66: (7). April 1, 2006

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2006 American Association for Cancer Research. Cancer Research signaling. Using Tet-inducible epithelial breast adenocarcinoma following activation of the ERBB2 proto-oncogene. Hence, the cells, we show that ERBB2 signaling leads to the accumulation novel regulatory interactions described here could lead to of ITGA5 and ITGB1 transcripts and protein. Functionally, additional (complementary) therapeutic options for ERBB2- oncogene-induced expression of the a5h1 integrin fibronectin dependent malignancies. receptor on the surface of MCF-7 cells enhances their adhesiveness to fibronectin, and adhesion on fibronectin leads to chemo- Acknowledgments resistance and enhanced survival of ERBB2-overexpressing cells Received 8/9/2005; revised 1/20/2006; accepted 1/26/2006. under adverse conditions (specifically nutrient deprivation and Grant support: Stiftung Rheinland-Pfalz fu¨r Innovation (386261/618), Deutsche hypoxia). We discuss that increased fibronectin adhesiveness Forschungsgemeinschaft (SP 682-1/1), and Maifor (C. Spangenberg); Deutsche Forschungsgemeinschaft (LA 1381-3/2, E.U. Lausch). induced by oncogenic ERBB2 signaling plays an important The costs of publication of this article were defrayed in part by the payment of page role in the course of ERBB2-driven carcinogenesis, and that charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. this may be particularly relevant in the subsistence of long- The authors wish to thank Elke Deiss and Marcel Engel for excellent technical surviving, dormant micrometastatic cells in the early phase assistance.

References node-negative non-small cell lung cancer. Clin Cancer 37. Xu Q, Briggs J, Park S, et al. Targeting Stat3 blocks Res 2000;6:96–101. both HIF-1 and VEGF expression induced by multiple 1. Guy PM, Platko JV, Cantley LC, Cerione RA, Carraway 20. Zhang Z, Vuori K, Reed JC, Ruoslahti E. The a5h1 oncogenic growth signaling pathways. Oncogene 2005; KL III. Insect cell-expressed p180erbB3 possesses an integrin supports survival of cells on fibronectin and 24:5552–60. impaired tyrosine kinase activity. Proc Natl Acad Sci up-regulates Bcl-2 expression. Proc Natl Acad Sci U S A 38. Aoudjit F, Vuori K. Integrin signaling inhibits U S A 1994;91:8132–6. 1995;92:6161–5. paclitaxel-induced apoptosis in breast cancer cells. 2. Olayioye MA, Neve RM, Lane HA, Hynes NE. The 21. Varner JA, Emerson DA, Juliano RL. Integrin a5h1 Oncogene 2001;20:4995–5004. ErbB signaling network: receptor heterodimerization in expression negatively regulates cell growth: reversal by 39. Sethi T, Rintoul RC, Moore SM, et al. Extracellular development and cancer. EMBO J 2000;19:3159–67. attachment to fibronectin. Mol Biol Cell 1995;6:725–40. matrix proteins protect small cell lung cancer cells 3. Yarden Y, Sliwkowski MX. Untangling the ErbB 22. Nista A, Leonetti C, Bernardini G, Mattioni M, against apoptosis: a mechanism for small cell lung signalling network. Nat Rev Mol Cell Biol 2001;2: Santoni A. Functional role of a4h1 and a5h1 integrin cancer growth and drug resistance in vivo. Nat Med 127–37. fibronectin receptors expressed on Adriamycin-resistant 1999;5:662–8. 4. Slamon DJ, Clark GM, Wong SG, et al. Human breast MCF-7 human mammary carcinoma cells. Int J Cancer 40. Yu D, Hung MC. Overexpression of ErbB2 in cancer: correlation of relapse and survival with ampli- 1997;72:133–41. cancer and ErbB2-targeting strategies. Oncogene 2000; fication of the HER-2/neu oncogene. Science 1987;235: 23. Lee JW, Juliano RL. a5h1 Integrin protects intestinal 19:6115–21. 177–82. epithelial cells from apoptosis through a phosphatidy- 41. Knuefermann C, Lu Y, Liu B, et al. HER2/PI-3K/ 5. Tanner B, Kreutz E, Weikel W, et al. Prognostic linositol 3-kinase and protein kinase B-dependent Akt activation leads to a multidrug resistance in significance of c-erB-2 mRNA in ovarian carcinoma. pathway. Mol Biol Cell 2000;11:1973–87. human breast adenocarcinoma cells. Oncogene 2003; Gynecol Oncol 1996;62:268–77. 24. Korah R, Boots M, Wieder R. Integrin a5h1 promotes 22:3205–12. 6. Slamon DJ, Godolphin W, Jones LA, et al. Studies of the survival of growth-arrested breast cancer cells: an 42. Perez-Soler R, Kemp B, Wu QP, et al. Response and HER-2/neu proto-oncogene in human breast and in vitro paradigm for breast cancer dormancy in bone determinants of sensitivity to paclitaxel in human non- ovarian cancer. Science 1989;244:707–12. marrow. Cancer Res 2004;64:4514–22. small cell lung cancer tumors heterotransplanted in 7. Sauter G, Moch H, Moore D, et al. Heterogeneity of 25. Pantel K, Brakenhoff RH. Dissecting the metastatic nude mice. Clin Cancer Res 2000;6:4932–8. erbB-2 gene amplification in bladder cancer. Cancer Res cascade. Nat Rev Cancer 2004;4:448–56. 43. Leitzel K, Teramoto Y, Konrad K, et al. Elevated 1993;53:2199–203. 26. Wieder R. Insurgent micrometastases: sleeper cells serum c-erbB-2 antigen levels and decreased response 8. Weiner DB, Nordberg J, Robinson R, et al. Expression and harboring the enemy. J Surg Oncol 2005;89:207–10. to hormone therapy of breast cancer. J Clin Oncol 1995; of the neu gene-encoded protein (P185neu) in human 27. Birkenmeier TM, McQuillan JJ, Boedeker ED, et al. 13:1129–35. non-small cell carcinomas of the lung. Cancer Res 1990; The a5h1 fibronectin receptor. Characterization of the 44. Hengstler JG, Lange J, Kett A, et al. Contribution of 50:421–5. a5 gene promoter. J Biol Chem 1991;266:20544–9. c-erbB-2 and topoisomerase IIa to chemoresistance in 9. Ross JS, McKenna BJ. The HER-2/neu oncogene in 28. Gardner SN. A mechanistic, predictive model of ovarian cancer. Cancer Res 1999;59:3206–14. tumors of the gastrointestinal tract. Cancer Invest 2001; dose-response curves for cell cycle phase-specific and 45. Braun S, Pantel K, Muller P, et al. Cytokeratin-positive 19:554–68. -nonspecific drugs. Cancer Res 2000;60:1417–25. cells in the bone marrow and survival of patients with 10. Menard S, Casalini P, Campiglio M, et al. HER2 29. Jin H, Varner J. Integrins: roles in cancer development stage I, II, or III breast cancer. N Engl J Med 2000;342: overexpression in various tumor types, focussing on its and as treatment targets. Br J Cancer 2004;90:561–5. 525–33. relationship to the development of invasive breast 30. Danen EH. Integrins: regulators of tissue function 46. BraunS,SchlimokG,HeumosI,etal.ErbB2 cancer. Ann Oncol 2001;12:S15–19. and cancer progression. Curr Pharm Des 2005;11: overexpression on occult metastatic cells in bone 11. DiGiovanna MP, Chu P, Davison TL, et al. Active 881–91. marrow predicts poor clinical outcome of stage I-III signaling by HER-2/neu in a subpopulation of HER-2/ 31. Guo W, Giancotti FG. Integrin signalling during breast cancer patients. Cancer Res 2001;61:1890–5. neu-overexpressing ductal carcinoma in situ: clinico- tumour progression. Nat Rev Mol Cell Biol 2004;5: 47. Kumar CC. Signaling by integrin receptors. Oncogene pathological correlates. Cancer Res 2002;62:6667–73. 816–26. 1998;17:1365–73. 12. Menard S, Casalini P, Campiglio M, Pupa SM, 32. Denko NC, Fontana LA, Hudson KM, et al. Investi- 48. Ruoslahti E. Fibronectin and its integrin receptors in Tagliabue E. Role of HER2/neu in tumor progression gating hypoxic tumor physiology through gene expres- cancer. Adv Cancer Res 1999;76:1–20. and therapy. Cell Mol Life Sci 2004;61:2965–78. sion patterns. Oncogene 2003;22:5907–14. 49. Jones JL, Walker RA. Integrins: a role as cell signalling 13. Yu D, Hung MC. Role of erbB2 in breast cancer 33. Koike T, Kimura N, Miyazaki K, et al. Hypoxia molecules. Mol Pathol 1999;52:208–13. chemosensitivity. Bioessays 2000;22:673–80. induces adhesion molecules on cancer cells: a missing 50. Buttery RC, Rintoul RC, Sethi T. Small cell lung 14. Eccles SA. The role of c-erbB-2/HER2/neu in breast link between Warburg effect and induction of selectin- cancer: the importance of the extracellular matrix. Int J cancer progression and metastasis. J Mammary Gland ligand carbohydrates. Proc Natl Acad Sci U S A 2004; Biochem Cell Biol 2004;36:1154–60. Biol Neoplasia 2001;6:393–406. 101:8132–7. 51. Matter ML, Ruoslahti E. A signaling pathway from 15. Holbro T, Civenni G, Hynes NE. The ErbB receptors 34. Wykoff CC, Sotiriou C, Cockman ME, et al. Gene the a5h1 and a(v)h3 integrins that elevates bcl-2 and their role in cancer progression. Exp Cell Res 2003; array of VHL mutation and hypoxia shows novel transcription. J Biol Chem 2001;276:27757–63. 284:99–110. hypoxia-induced genes and that cyclin D1 is a VHL 52. Luzzi KJ, MacDonald IC, Schmidt EE, et al. Multistep 16. Trost TM, Lausch EU, Fees SA, et al. Premature target gene. Br J Cancer 2004;90:1235–43. nature of metastatic inefficiency: dormancy of solitary senescence is a primary fail-safe mechanism of ERBB2- 35. Laughner E, Taghavi P, Chiles K, Mahon PC, Semenza cells after successful extravasation and limited survival driven tumorigenesis in breast carcinoma cells. Cancer GL. HER2 (neu) signaling increases the rate of hypoxia- of early micrometastases. Am J Pathol 1998;153:865–73. Res 2005;65:840–9. inducible factor 1a (HIF-1a) synthesis: novel mecha- 53. Naumov GN, MacDonald IC, Weinmeister PM, et al. 17. Lowe SW, Cepero E, Evan G. Intrinsic tumour nism for HIF-1-mediated vascular endothelial growth Persistence of solitary mammary carcinoma cells in a suppression. Nature 2004;432:307–15. factor expression. Mol Cell Biol 2001;21:3995–4004. secondary site: a possible contributor to dormancy. 18. Dimri GP. What has senescence got to do with 36. Li YM, Zhou BP, Deng J, et al. A hypoxia-independent Cancer Res 2002;62:2162–8. cancer? Cancer Cell 2005;7:505–12. hypoxia-inducible factor-1 activation pathway induced 54. Naumov GN, MacDonald IC, Chambers AF, Groom 19. Adachi M, Taki T, Higashiyama M, et al. Significance by phosphatidylinositol-3 kinase/Akt in HER2 over- AC. Solitary cancer cells as a possible source of tumour of integrin a5 gene expression as a prognostic factor in expressing cells. Cancer Res 2005;65:3257–63. dormancy? Semin Cancer Biol 2001;11:271–6.

Cancer Res 2006; 66: (7). April 1, 2006 3724 www.aacrjournals.org

55. Chambers AF, Groom AC, MacDonald IC. Dissemi- Overexpression of hypoxia-inducible factor 1a is asso- 67. Yu D, Jing T, Liu B, et al. Overexpression of ErbB2 blocks nation and growth of cancer cells in metastatic sites. ciated with an unfavorable prognosis in lymph node- Taxol-induced apoptosis by up-regulation of p21Cip1, Nat Rev Cancer 2002;2:563–72. positive breast cancer. Clin Cancer Res 2002;8:1831–7. which inhibits p34Cdc2 kinase. Mol Cell 1998;2:581–91. 56. Cairns RA, Khokha R, Hill RP. Molecular mechanisms 62. Bos R, van der GP, Greijer AE, et al. Levels of hypoxia- 68. Tan M, Jing T, Lan KH, et al. Phosphorylation on of tumor invasion and metastasis: an integrated view. inducible factor-1a independently predict prognosis in tyrosine-15 of p34(Cdc2) by ErbB2 inhibits p34(Cdc2) Curr Mol Med 2003;3:659–71. patients with lymph node negative breast carcinoma. activation and is involved in resistance to taxol-induced 57. Riethmuller G, Klein CA. Early cancer cell dissemi- Cancer 2003;97:1573–81. apoptosis. Mol Cell 2002;9:993–1004. nation and late metastatic relapse: clinical reflections 63. Giatromanolaki A, Koukourakis MI, Simopoulos C, et 69. Liang K, Lu Y, Jin W, et al. Sensitization of breast and biological approaches to the dormancy problem in al. c-erbB-2 related aggressiveness in breast cancer is cancer cells to radiation by trastuzumab. Mol Cancer patients. Semin Cancer Biol 2001;11:307–11. hypoxia inducible factor-1a dependent. Clin Cancer Res Ther 2003;2:1113–20. 58. Aguirre-Ghiso JA, Liu D, Mignatti A, Kovalski K, 2004;10:7972–7. 70. Soderlund K, Perez-Tenorio G, Stal O. Activation of Ossowski L. Urokinase receptor and fibronectin regulate 64. Chung J, Yoon S, Datta K, Bachelder RE, Mercurio the phosphatidylinositol 3-kinase/Akt pathway prevents the ERK(MAPK) to p38(MAPK) activity ratios that AM. Hypoxia-induced vascular endothelial growth radiation-induced apoptosis in breast cancer cells. Int J determine carcinoma cell proliferation or dormancy factor transcription and protection from apoptosis are Oncol 2005;26:25–32. in vivo. Mol Biol Cell 2001;12:863–79. dependent on a6h1 integrin in breast carcinoma cells. 71. Naumov GN, Townson JL, MacDonald IC, et al. 59. Aguirre-Ghiso JA, Ossowski L, Rosenbaum SK. Green Cancer Res 2004;64:4711–6. Ineffectiveness of doxorubicin treatment on solitary fluorescent protein tagging of extracellular signal- 65. Yoon SO, Shin S, Mercurio AM. Hypoxia stimulates dormant mammary carcinoma cells or late-developing regulated kinase and p38 pathways reveals novel carcinoma invasion by stabilizing microtubules and metastases. Breast Cancer Res Treat 2003;82:199–206. dynamics of pathway activation during primary and promoting the Rab11 trafficking of the a6h4 integrin. 72. Paulhe F, Manenti S, Ysebaert L, et al. Integrin metastatic growth. Cancer Res 2004;64:7336–45. Cancer Res 2005;65:2761–9. function and signaling as pharmacological targets in 60. Woelfle U, Cloos J, Sauter G, et al. Molecular 66. Dragowska WH, Warburton C, Yapp DT, et al. HER-2/ cardiovascular diseases and in cancer. Curr Pharm Des signature associated with bone marrow micrometastasis neu overexpression increases the viable hypoxic cell 2005;11:2119–34. in human breast cancer. Cancer Res 2003;63:5679–84. population within solid tumors without causing changes 73. Semenza GL. Targeting HIF-1 for cancer therapy. Nat 61. Schindl M, Schoppmann SF, Samonigg H, et al. in tumor vascularization. Mol Cancer Res 2004;2:606–19. Rev Cancer 2003;3:721–32.

www.aacrjournals.org 3725 Cancer Res 2006; 66: (7). April 1, 2006

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2006 American Association for Cancer Research. ERBB2-Mediated Transcriptional Up-regulation of the α5β1 Integrin Fibronectin Receptor Promotes Tumor Cell Survival Under Adverse Conditions

Christian Spangenberg, Ekkehart U. Lausch, Tatjana M. Trost, et al.

Cancer Res 2006;66:3715-3725.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/66/7/3715

Cited articles This article cites 73 articles, 34 of which you can access for free at: http://cancerres.aacrjournals.org/content/66/7/3715.full#ref-list-1

Citing articles This article has been cited by 7 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/66/7/3715.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://cancerres.aacrjournals.org/content/66/7/3715. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.