Free Radical Biology & Medicine 52 (2012) 993–1002

Contents lists available at SciVerse ScienceDirect

Free Radical Biology & Medicine

j ourna l homepage: www.e lsev ie r .com/ locate / f reeradb iomed

Original Contribution

Overexpression of GRP94 in breast cancer cells resistant to oxidative stress promoteshigh levels of cancer cell proliferation and migration: Implications fortumor recurrence

Nicolas Dejeans a, Christophe Glorieux a,1, Samuel Guenin b,1, Raphael Beck a, Brice Sid a, Rejane Rousseau c,Bettina Bisig b, Philippe Delvenne b, Pedro Buc Calderon a,d,⁎, Julien Verrax a

a Université Catholique de Louvain, Louvain Drug Research Institute, Toxicology and Cancer Biology Research Group, Belgiumb Department of Pathology, GIGA Cancer, University of Liege, B23 CHU Sart Tilman, 4000 Liège, Belgiumc Institut de Statistique, Biostatistique et Sciences Actuarielles, Université Catholique de Louvain, Louvain-la-Neuve, Belgiumd Departamento de Ciencias Quimicas y Farmaceuticas, Universidad Arturo Prat, Iquique, Chile

Abbreviations: ROS, reactive oxygen species; UPR, uMen, ascorbate/menadione; ER, endoplasmic reticulum.⁎ Corresponding author at: Avenue E. Mounier 7

Fax: +32 2 764 73 59.E-mail address: pedro.buccalderon@uclouvain.be (P

1 These authors contributed equally to this work.

0891-5849/$ – see front matter © 2011 Elsevier Inc. Alldoi:10.1016/j.freeradbiomed.2011.12.019

a b s t r a c t

a r t i c l e i n f o

Article history:Received 2 March 2011Revised 21 October 2011Accepted 2 December 2011Available online 2 January 2012

Keywords:GRP94Cancer recurrenceCancer resistanceOxidative stressEndoplasmic reticulum stress

Targeting the altered redox status of cancer cells is emerging as an interesting approach to potentiatechemotherapy. However, tomaximize the effectiveness of this strategy and define the correct chemotherapeuticassociations, it is important to understand the biological consequences of chronically exposing cancer cells toreactive oxygen species (ROS). Using an H2O2-generating system, we selected a ROS-resistant MCF-7 breastcancer cell line, namely Resox cells. By exploring different survival pathways that are usually induced duringoxidative stress, we identified a constitutive overexpression of the endoplasmic reticulum chaperone,GRP94, in these cells, whereas levels of its cytoplasmic homolog HSP90, or GRP78, were not modified. This over-expression was not mediated by constitutive unfolded protein response (UPR) activation. The increase in GRP94is tightly linked to an increase in cell proliferation and migration capacities, as shown by GRP94-silencingexperiments. Interestingly, we also observed that GRP94 silencing inhibits migration and proliferation of thehighly aggressiveMDA-MB-231 cells. By immunohistochemistry, we showed that GRP94 expression was higherin recurrent human breast cancers than in their paired primary neoplasias. Similar to the situation in the Resoxcells, this increase was not associated with an increase in UPR activation in recurrent tumors. In conclusion, thisstudy suggests that GRP94 overexpressionmay be a hallmark of aggressiveness and recurrence in breast cancers.

© 2011 Elsevier Inc. All rights reserved.

Introduction

Exploiting vulnerabilities of cancer cells is a key strategy for thedevelopment of new anticancer drugs. One such vulnerability is intra-cellular redox alteration, consecutive to a decrease in antioxidantcapacities and an increased formation of reactive oxygen species(ROS) via several mechanisms including oncogenic signaling [1–3].Since the discovery of this susceptibility, scientists have developedstrategies that specifically target cancer redox sensitivity [1,3].Candidate drugs designed for this purpose can be separated intotwo main categories: the prooxidants, for example, β-lapachone [4]or the association of ascorbate/menadione (Asc/Men) [5], and theinhibitors of antioxidant capacity, for instance, phenylethyl

nfolded protein response; Asc/

3, 1200 Bruxelles, Belgium.

. Buc Calderon).

rights reserved.

isothiocyanate [6] or As2O3 [7]. Despite initial promise with theseapproaches, adaptation of cancer cells to oxidative stress may resultin unwanted side effects. Indeed, oxidants can induce activation ofprosurvival pathways, such as NF-E2-related factor 2 (Nrf2) [8] orthe unfolded protein response (UPR) [9], which have been associatedwith chemoresistance [8,10]. Cancer adaptation to redox equilibriumdysregulation is also supported by the recent discovery that cancerstem cells contain a high degree of ROS defense resulting in radio-resistance [11]. Consequently, there is a special need to better under-stand the physiological and molecular consequences of chronicallyexposing cancer cells to oxidative stress in order to improve anticanceroxidative strategies.

Within the tumor,malignant cells face restricted supplies of nutrientsand oxygen. Therefore, to survive, they need to activate prosurvivalpathways, like Nrf2 and UPR [12]. Under these stress conditions,activation of UPR leads to the upregulation of prosurvival proteinsinvolved in angiogenesis, folding capacity, redox protection, or degrada-tion of unfolded proteins. However, when activation of this response isprolonged, it can also result in cell death. Factors controlling the balancebetween proapoptotic and protective functions of the UPR are of critical

994 N. Dejeans et al. / Free Radical Biology & Medicine 52 (2012) 993–1002

importance for tumorigenesis [10,13]. A major mechanism of the UPRprotecting arms is an increase in endoplasmic reticulum (ER) molecularchaperones and there is a large body of evidence to support a role ofthese chaperones in cancer features, including proliferation rate, drugresistance, and metastasis [14]. For example, GRP78 (also called BIP),one of the best-characterized ER chaperones, has been shown to benecessary for tumorigenesis of fibrosarcoma cells injected into mice[15] and to protect cancer cells from topoisomerase inhibitors [16,17],microtubule-interfering agents [18], and doxorubicin [17]. This, togetherwith the fact that GRP78 has been found to be overexpressed in a widerange of cancers [10,14], clearly supports a role of ER chaperones incancer biology. As for GRP78, an elevated level of GRP94 (gp96,HSP90B1) has also been reported in several cancers, including esophageal[19], lung [20], colon [21], gastric [22], and breast [23] carcinoma. Never-theless, the biological significance of the increase in GRP94 remainsunclear.

The present study was designed to identify survival pathwaysassociated with redox protection, with a particular focus on putativechanges in ER chaperones and their potential effects on cancer cellresistance, proliferation, and migration capacities.

Materials and methods

Cell lines and chemicals

The breast cancer cell line MCF-7 was purchased from ECACC(Salisbury, UK) and maintained in DMEM supplemented with 10%fetal calf serum, penicillin (10,000 U/ml) and streptomycin (10 mg/ml). TheMDA-MB-231 cell linewas a kind gift from Pr Akeila Bellahcene(Metastasis Research Laboratory, Giga Cancer, Liege, Belgium). Thesecells were maintained in DMEM supplemented with 10% fetal calfserum, penicillin (10,000 U/ml), and streptomycin (10 mg/ml) and 1%of non-essential amino acids. An MCF-7 cell line resistant to oxidativestresswas established by constantly exposing cells to increasing concen-trations of the pro-oxidant association of Asc/Men for 6 months, startingwith 0.5 mM ascorbate/5 μM menadione to a final concentration of1.5 mM ascorbate/15 μM menadione. Cells were first treated at 50%confluence by replacing their media with fresh media containing Asc/Men. When surviving cells reached 50% confluence, they were washedwith warm PBS and treated again (every 1 to 7 days, depending on thelevel of resistance). To avoid the development of islets of resistance,which could arise from cooperation between cells, the cells weretrypsinized approximately every 2 weeks and subcultured into newflasks. After selection, the cell line was stabilized in drug-free mediumfor 1 month. We will refer to these MCF7 cells resistant to oxidativestress as Resox cells. Menadione sodium bisulfite, sodium ascorbate,dimethyl sulfoxide, glucose oxidase, hydrogen peroxide, cisplatin,doxorubicin, paclitaxel, 5-fluorouracil, DCFH-DA (2′,7′-dichlorofluores-cein diacetate), tert-butylhydroquinone, N-acetylcysteine, and 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide (MTT) werepurchased from Sigma (St. Louis, MO). Thapsigargin was purchasedfrom Santa Cruz Biotechnology (Santa Cruz, CA). Small interfering RNA(siRNA) and Dharmafect 1 reagent were purchased from Dharmacon(Lafayette, CO). All other chemicals were of ACS reagent grade.

Cell culture, functional assays, and siRNA transfection

Cell viability was evaluated using a classical MTT reduction assay[24]. In all cell viability experiments, MTT assay was being completedin serum-containing media. For the cell proliferation assay, cells wereseeded into 96-well plates at a density of 5000 cells per well for 24 to96 h and the relative cell density was then measured using the MTTreduction assay. To evaluate proliferation under reducing conditions,cells were cultured in a medium containing 1 mM DTT, and mediawith or without DTT were changed every 24 h. For doubling timedetermination, 100,000 cells were seeded in 6-well plates and

counted every 24 h with an automated cell counter (TC10, Bio-RadLaboratories, Hercules, CA). Doubling time was then determinedusing the doubling-time.com website (Roth V. 2006; http://www.doubling-time.com/compute.php). Clonogenic assays were performedby seeding cells (500) in 6-well plates at a single-cell density and allowingthem to grow for 14 days. Clonogenicity was determined by stainingcolonies using crystal violet. Migration was tested in wound-healingassays using Culture Inserts (Ibidi, München, DE) according to the manu-facturer's instructions. After removal of culture inserts, pictures of theremaining gaps were taken at different time periods. Dharmafect 1 wasused for transfection of siRNAs into MCF-7 and Resox cells according toprotocols provided by Dharmacon. Transfection was performed on cellsat 1/2 confluence for 24 h, with a 0.1 μM siRNA solution. The siRNAsolution was then washed with PBS and replaced with DMEM.

Immunoblotting

Immunoblotting was performed as previously described [25]. ForNrf2 immunoblotting, cytoplasmic and nuclear extracts wereprepared as previously described [26]. β-Actin, HSP70 (cytoplasmiccontrol), or TBP (nuclear control) levels were used to control proteinloading amounts in each sample. Rabbit antibodies to ERp72, ERp57,ERp44, phospho-eIF2α (Ser51), ERO1-Lα, PDI, and anti-GRP78 werefrom Cell Signaling Technology (Danvers, MA); mouse antibodies toHSP90, Nrf2, SOD, and NQO1 were from Santa Cruz Biotechnology(Santa Cruz, CA); mouse antibody to β-actin and TBP were fromAbcam (Cambridge, UK); rabbit antibodies to GRP94 and catalasewere from Chemicon International (Temecula, CA).

Biochemical measurements

The content of reduced glutathione was determined using theGSH-Glo glutathione assay (Promega, Madison, WI, USA) accordingto the procedures described by the supplier. Catalase activity wasmeasured using the TiSO4 method [27]. NQO1 (DT-diaphorase) activitywas assayed spectrophotometrically by measuring cytochrome creduction in the presence of NADH as previously described [28].

Immunohistochemistry on human breast tissues

Twenty nine formalin-fixed, paraffin-embedded breast cancersamples were retrieved from the Tumor Bank of the University ofLiege (Liège, Belgium). These samples included 15 primary invasiveductal breast carcinomas and their related secondary tumors (including5 local recurrences and 9metastases). For the immunohistochemistry onnormal human breast tissue, 10 formalin-fixed, paraffin-embeddedbreast healthy tissue samples originating from reductionmammoplastieswere retrieved from the Tumor Bank of the University of Liege (Liège,Belgium). The protocol was approved by the Ethics Committee of theUniversity Hospital of Liege. Clinical and pathological datawere availablefor each patient and are summarized in supplementary Table 1. Serialsections of the breast tumor specimens were stained using antibodiesdirected against GRP94 (Abcam, Cambridge, UK) and P-eIF2 (CellSignaling). Immunoperoxidase staining was performed using theLSAB-2 kit (Dako, Glostrup, Denmark) for GRP94 and CSA-II kit (Dako)for P-eIF2 according to the supplier's recommendations. Positive cellswere visualized using a 3,3′-diaminobenzidine (DAB) substrate and thesections were counterstained with hematoxylin. The immunostainingof GRP94 and P-eif2 was evaluated on labeled tissues using a semiquan-titative scoring scale based on staining. For staining intensity, score 0represented undetectable staining in tumor cells , and scores 1+, 2+,and 3+ denoted staining of low, moderate, and strong intensity, respec-tively. For staining extent, score 0 represented 0% of stained tumor cells,and scores 1, 2, and 3 corresponded to 6–33, 34–66, and 67–100%,respectively. Finally, the intensity score and the extent score were

995N. Dejeans et al. / Free Radical Biology & Medicine 52 (2012) 993–1002

multiplied, thus providing a unique global score for each immunostain-ing, ranging between 0 and 9.

Data analyses

All experiments (excepting immunohistochemistry) wereperformed at least 3 times. MTT assays, Western blot, enzymaticactivities, or proliferation assays were analyzed using a paired orunpaired t test or ANOVA as appropriate, performed with GraphPadPrism software (GraphPad Software, San Diego, CA, USA). For immuno-histochemistry on the breast cancer samples, the P values wereobtained from a paired Wilcoxon signed rank test with continuitycorrection, a Welch's t test, or a t test for unequal sample sizes withequal variance. P values less than 0.05 were considered statisticallysignificant. All the results were computed in R language.

Results and discussion

Establishment of a MCF-7 cell line resistant to oxidative stress

To investigate the relationship between resistance to oxidativestress and biological cancer properties (chemoresistance, proliferationand migration), we established an MCF-7 cell line, namely the Resoxcell line, resistant to the pro-oxidant association of ascorbate andmenadione (Asc/Men). This treatment was chosen because ascorbatedriven menadione-redox cycling generates mainly H2O2 [2] andrepresents a potential candidate treatment for future prooxidative-based anticancer therapies [5,29]. After 6 months of repeated exposureto Asc/Men, we obtained a quite homogeneous population (as deducedfrom karyotype analyses, see supplementary Figs. 1 and 2): the Resoxcells. The resistance to Asc/Men was confirmed by a cell survival assay(Fig. 1A). Most MCF7 cells were not viable when they were exposedto Asc/Men (1 mM/10 μM) whereas about 60% of Resox cells were stillviable. Two other pro-oxidant systems were also used to confirmtolerance of the Resox cells to oxidative stress (Fig. 1B). The first,glucose/glucose oxidase, is another H2O2-generating system that

Fig. 1. Evaluation of MCF-7/Resox cell resistance. (A) MCF-7 and MCF-7/Resox cellswere exposed to different concentrations of Asc/Men for 24 h and then cell viabilitywas measured. (B) Cells were exposed to GLOX (10 mU/ml) or 1 mM H2O2 for 24 hbefore cell viability measurements. Results represent means±SD. * Pb0.05, *** Pb0.001.

mimics chronic exposure to H2O2, and the second was acute exposureto H2O2 itself. With both systems, the Resox cells had increasedtolerance compared to their parental cell line (Fig. 1B). Using thefluorescent probeDCFH,we also observed a lower increase of intracellularlevel of ROS in Resox cells following Asc/Men treatment, as compared toMCF-7 cells (supplementary Fig. 3).

Analysis of antioxidant elements in Resox cells

To characterize the Resox cell line, a first set of experiments wasperformed to identify survival mechanisms or constitutively activatedpathways triggered by acute oxidative stress, focusing on the antioxidantcapacities. When looking at the expression of antioxidant enzymes,protein abundance (Fig. 2A) and enzyme activity (Figs. 2B and C) ofNAD(P)H:quinone oxidoreductase 1 (NQO1) and catalase (CAT) wereenhanced in Resox cells. Finally, intracellular GSH, another critical factorassociated with redox protection and chemoresistance [8], was notmodified in Resox cells compared to MCF-7 cells (Fig. 2D). Similarly, nochanges in the activities of glutathione peroxidase and superoxidedismutase were observed in Resox cells compared to MCF-7 cells (datanot shown). These results show that some antioxidant enzymes (CATand NQO1) were upregulated in Resox cells.

Analysis of survival pathways associated with resistance to oxidativestress

To identify other survival pathways likely activated in the resistantcells, a second set of experimentswas performed. First, since ER chaper-ones (particularly GRP78) have been shown to be important factors forcancer resistance and recurrence, we evaluated the expression of theseproteins in Resox cells. Fig. 3A shows that GRP94 was upregulated 3-fold in Resox cells, but not its cytoplasmic homologue HSP90, nor

Fig. 2. Increase of antioxidant defenses in Resox cells. (A) Comparison of abundance ofantioxidant proteins in MCF-7 and MCF-7/Resox cells. Western blots againstDT-diaphorase (NQO1), catalase (CAT), and superoxide dismutase (SOD) in MCF-7and Resox cells are shown. Numbers 1, 2, and 3 represent 3 independent proteinextractions. Numbers below immunoblots represent means±SEM of proteinabundances normalized to actin. (B and C) Catalase and NQO1 activities in both celllines (means±SEM). (D) Measurement of GSH levels in MCF-7 and MCF-7/Resoxcells (means±SD). * Pb0.05, **, Pb0.01.

Fig. 3. Resistance to oxidative stress is associated with Grp94 overexpression. (A) Western blot against GRP94, GRP78, HSP90, and actin in MCF-7 andMCF-7/Resox cells. Numbers 1,2, and 3 represent 3 different independent protein extractions. Protein abundances were normalized to actin and represent the mean±SEM of three independent experiments.(B) GRP94 silencing was not associated with HSP90, HSP70, or GRP78 protein variations. Cells were exposed to GRP94 (siGRP94) or nonrelevant (siCONT) siRNA. A classical Westernblot against the indicated proteins was performed on the proteins extracted from cells exposed to siRNAs for 96 h. (C) Phosphorylation of eIF2 after Asc/Men (1 mM/10 μM, 2 h) orthapsigargin (TG, 1 μM, 6 h) treatment in cells pretreated with the indicated siRNAs for 96 h. A Western blot against the phosphorylated form of eIF2 (P-eIF2) is shown. ** Pb0.01.

Fig. 4. Relationship between GRP94 upregulation and MCF-7 resistance. (A) MCF-7 andMCF-7/Resox cells treated with the indicated siRNA for 72 h were exposed to differentconcentrations of Asc/Men for 24 h. Then cell viability was measured by MTT reductionassay, as described under Materials and methods. (A) Cells pretreated with the indicatedsiRNAs for 72 h were exposed to cisplatin (Cis, 50 μM) or doxorubicin (Doxo, 1 μM),5-fluorouracil (5-FU, 100 μM) or paclitaxel (PAC, 0.1 μM) for 24 before cell viabilitymeasurement. Results represent means±SEM.

996 N. Dejeans et al. / Free Radical Biology & Medicine 52 (2012) 993–1002

GRP78. A similar increase in GRP94 protein abundance was alsoobserved in 8 Resox subclones established from the whole cell popula-tion (data not shown), thus confirming the relative homogeneity of theResox cell population revealed by the karyotype analysis. AlthoughGRP94 is one of the most commonly expressed ER chaperones and amember of (ER)-localized multiprotein complexes [30], its functionsremain poorly characterized [31]. This is partly due to the compensatoryregulation of multiple chaperones that has usually been observed whenGRP94 (or another chaperone) expression is suppressed [31,32]. Indeed,Mao et al. [32] showed that only complete knockout of GRP94, but notpartial depletion, was associated with ER chaperone upregulation. Tostudy the role of GRP94 overexpression in Resox cells, we thereforereduced GRP94 protein abundance using a siRNA, to about 70% of theMCF-7 basal level in the Resox cell line and 50% in MCF-7 cells after96 h (Fig. 3B). Silencing of GRP94 was not accompanied by changes inthe expression of other chaperones, namely GRP78, HSP70, or HSP90(Fig. 3B). Then, because GRP78 is able to sense ER protein unfolding/misfolding leading to induction of UPR, we also studied whetherGRP94 overexpression was associated with a modified ability to inducethis pathway. Fig. 3C shows that Resox cells did not significantly activatethe UPR during oxidative stress (as indicated by the inability to phos-phorylate eIf2 after Asc/Men treatment) but the UPR was activatedafter ER calcium depletion by thapsigargin (TG). This result shows thatResox cells still had the capacity to respond to an ER stress by activatingUPR but were unable to activate this pathway when it was triggered byoxidative stress. Finally, silencing of GRP94 did not restore UPR activa-tion after Asc/Men treatment (Fig. 3C), demonstrating that GRP94 wasnot associated with the inability of Resox cells to induce the UPR duringoxidative stress.

Secondly, we assessed whether the activation of Nrf2, a redox-sensitive factor, was modified in the Resox cells. No significant differ-ence in the activation of this factor was observed in these cells underbasal or stress conditions (Supplementary Fig. 4). This rules out aputative acquisition of resistance by a constitutive activation of Nrf2

leading to activation of several antioxidant genes. Indeed, no differ-ences were observed in the induction of NQO1, a well characterizedNrf2 target, after A/M treatment (data not shown).

997N. Dejeans et al. / Free Radical Biology & Medicine 52 (2012) 993–1002

Relationship among GRP94 expression, cell survival, and resistance

We investigated whether GRP94 upregulation had any effect oncell survival and Resox cell resistance to standard chemotherapy.Fig. 4A shows that there were no changes in the pattern of cellularresistance against oxidative stress in either cell line (MCF-7 andResox) whether GRP94 was silenced or not. This indicates thatGRP94 overexpression in Resox cells was not linked to the resistanceto oxidative stress.

We then assessed whether GRP94 overexpression, as well as resis-tance against oxidative stress, could be related to a putative resistanceto standard chemotherapeutic drugs. Four drugs, structurally andmechanistically unrelated, were selected: cisplatin (Cis), 5-fluorouracil(5-FU), doxorubicin (Doxo), and paclitaxel (PAC). Fig. 4B shows thatResox cells partially resisted Cis and Doxo treatments (Pb0.0001 forMCF-7 versus Resox, and for both treatments) for 24 h, but GRP94 didnot appear to be involved in this resistance. Indeed, the silencing ofGRP94 did not result in a survival decrease after doxorubicin exposure

Fig. 5. GRP94 is important for MCF-7/Resox increased proliferation and migration capacitieimage of MCF-7 andMCF-7/Resox clonogenicity is shown in the lower panel. (B) Role of GRPthe indicated siRNAs for 24 h. Cells were then trypsinized and incubated for different timeMaterials and methods. (C) MCF-7 and MCF-7/Resox cells were incubated with the indicateCell migration was assessed following removal of culture inserts. Typical pictures of the remSEM. * Pb0.05, ** Pb0.01, (ns) not significant.

(Fig. 4B; P=0.62) and appeared to induce resistance to cisplatin(Pb0.001). This is in agreement with previous reports showing thatinduction of ER chaperones is associatedwith potentiation of cytotoxicityinduced by platinating agents [33]. These results indicate that acquisitionof resistance against oxidative stress may be accompanied by crosseddrug resistance (i.e., cisplatin and doxorubicin), indicating theimportance of selecting the right association and schedule to effectivelyeradicate cancer cells.

Role of GRP94 in Resox proliferation and migration capacities

Although GRP94 overexpression was not involved in the survivalof Resox cells against oxidative stress or in the partial resistance tosome chemotherapy agents, we decided to explore the consequencesof its upregulation on two other hallmarks of cancer aggressiveness,i.e., cell proliferation andmigration. Twomethods were used to assessthe Resox cell proliferation rate, namely the colony forming unit(CFU) and the MTT reduction assay. Fig. 5A shows that, compared to

s. (A) Clonogenic assays performed on MCF-7 and MCF-7/Resox cells. A representative94 upregulation in the increase in MCF-7/Resox proliferation. Cells were incubated withs under normal growth conditions. Cell proliferation was assessed as described underd siRNAs for 24 h. Cells were then trypsinized and incubated in culture insert systems.aining gaps are shown 0, 3, or 7 days after insert removal. Results represent means±

998 N. Dejeans et al. / Free Radical Biology & Medicine 52 (2012) 993–1002

their parental cell line, the Resox cells had a higher proliferation ratewhich was partly mediated by GRP94 overexpression, as revealed bythe use of siRNA against GRP94 (Fig. 5B). Furthermore, we comparedMCF-7 versus Resox migration capacity using a culture-insert systemas described under Materials and methods. This experiment showedthat MCF-7 cells were virtually unable to migrate 7 days after insertremoval whereas the Resox cells almost completely recovered thegap during this time (Fig. 5C). This capacity to migrate is tightly asso-ciated with GRP94 overexpression since silencing GRP94 almostcompletely suppressed Resox migration capacity (Fig. 5C).

Fig. 6. GRP94 is important for the high proliferation and migration rates of MDA-MB-231. (A)blot. Numbers 1, 2, and 3 represent three different independent protein extractions. (B)MDA-MWestern blot against the indicated proteinswas performed on the proteins extracted from cellsindicated, and represent themean±SD of three independent experiments. (C) The amount of 1calculated as described under Materials and methods for each cell line. (D) MDA-MB-231 ceincubated in culture insert systems. Cell migration was assessed following removal of culture i(E) Clonogenic assays performed on MDA-MB-231 cells preincubated 24 h with the indicated

To confirm these results, we sought to establish the role of GRP94in the proliferation and migration capacity of a more aggressivebreast cancer cell line, the MDA-MB-231 cells. As shown in Fig. 6A,GRP94 was expressed in MDA-MB-231 to a level close to that of theResox cells. Interestingly, Resox and MDA-MB-231 cells presented analmost similar doubling time (33 and 31 h, respectively), considerablyshorter than that of MCF-7 cells (47 h, Fig. 6C). Finally, as observed forResox cells, GPR94 silencing (Fig. 6B) completely suppressed migration(Fig. 6D), and largely reduced proliferation of MDA-MB-231 cells(Fig. 6E). Taken together, these results support the existence of a link

Comparison of GRP94 protein abundance in MCF-7, Resox, and MDA-MB-231 by WesternB-231 cells were exposed to GRP94 (siGRP94) or nonrelevant (siCONT) siRNA. A standardexposed to siRNAs for 72 h. For (A) and (B), protein abundances normalized to actinwere00,000 cells was incubated for different time periods and counted. The doubling timewaslls were incubated with the indicated siRNAs for 24 h. Cells were then trypsinized andnserts. Typical pictures of the remaining gaps are shown 0, 6, or 24 h after insert removal.siRNA. Results represent means±SEM. * Pb0.05, ** Pb0.01.

999N. Dejeans et al. / Free Radical Biology & Medicine 52 (2012) 993–1002

between high GRP94 expression and increased aggressiveness ofhuman breast cancers. It should be noted that GRP94 is essential forthe correct processing of insulin-like growth factor [34], integrins [35],and Toll-like receptors [36]. Given the major role of these proteins incancer cell migration, this provides new perspectives for understandingthe role of GRP94 overexpression on the high migration capacity ofResox cells. Studies are now in progress to explore this particular issue.

Cell proliferation under reducing conditions

Increased ROS production has generally been associated withelevated cell growth. Therefore, it was rather unexpected that Resoxcells, although having high antioxidant capacities, had a higher prolif-eration rate than their parental cell line. We therefore hypothesizedthat GRP94 is overexpressed in resistant cells to promote proliferationunder an unfavorable redox context. To challenge this hypothesis, theintracellular redox homeostasis of Resox cells wasmodified by silencingcatalase (CAT), an antioxidant enzyme which is upregulated in thesecells (Fig. 2A). Fig. 7A shows that CAT silencing partially counteractedthe antiproliferative effect of GRP94 silencing in resistant cells. Thisindicates that GRP94 overexpression is not simply the consequence ofa reducing environment but that this chaperone is required by cells toproliferate under a reducing basal condition. Another element in favorof this hypothesis is the importance of GRP94 for cell proliferationunder reducing conditions (1 mM DTT) in both MCF-7 and Resox celllines (Fig. 7B). Indeed, exposing cells to DTT leads to an impairment ofprotein folding and to redox alterations in the ER [37]. Moreover, the

Fig. 7. Relationship between GRP94 expression and MCF-7/Resox proliferation rate and redutrypsinized and incubated for different times under normal growth conditions. Cell proliferawith the indicated siRNAs for 24 h. Cells were then trypsinized and incubated for different tunder Materials and methods. (C) Abundance of proteins involved in disulfide bond forma(PDI), and endoplasmic reticulum stress proteins 72, 57, and 44 (ERp72, ERp57 and ERp44)blotting was used as a loading control. Results represent means±SEM. * Pb0.05, ** Pb0.01

identification of genetic mutations responsible for DTT hypersensitivityhas been used as a method to identify genes involved in the cellularredox machinery required for disulfide-linked protein folding in theER [38]. Given the known association of GRP94 with the proteinmachinery dedicated to disulfide bond formation in the ER [30], itmay be suggested that GRP94 could help to maintain an optimalredox state in the ER or to transfer reducing equivalents through theER folding machinery. In this context, since hydrogen peroxide hasbeen proposed to directly mediate the formation of disulfide bonds inthe ER [39,40], it is tempting to speculate that the ER foldingmachinerywas increased in Resox cells in order to compensate for a high ERreducing state due to antioxidant enzyme overexpression. On thisbasis we decided to evaluate whether the ER chaperones involvedin disulfide bond formation were also elevated in Resox cells. Westernblot analyses revealed an increased abundance of the three majorprotein disulfides isomerases PDI, ERp72, and ERp57 in resistant cells(Fig. 7C).

GRP94 expression and breast cancer recurrence

Our previous results showed that GRP94 overexpression isinvolved in high cell proliferation and cell migration, suggestingthat it is an important mediator of cancer cell aggressiveness. Wetherefore wondered whether this was also the case in human recurrentbreast tumors. To this end, GRP94 expression was studied by immuno-histochemistry in primary invasive ductal carcinomas and in recurrent

cing state. (A) Cells were incubated with the indicated siRNAs for 24 h. Cells were thention was assessed as described under Materials and methods. (B) Cells were incubatedimes in media with or without 1 mM DTT. Cell proliferation was assessed as describedtion. Endoplasmic reticulum oxidoreductin 1-like (ERO1), protein disulfide isomerasewere evaluated in MCF-7 and MCF-7/Resox cells by Western blotting. Actin immuno-, (ns) not significant.

1000 N. Dejeans et al. / Free Radical Biology & Medicine 52 (2012) 993–1002

tumors. We first observed that GRP94 was overexpressed in tumors, ascompared to normal tissues (supplementary Figs. 5A and B). Then, toavoid the effects of interindividual variability, the analysis wasperformed in matched primary and secondary neoplasias derivedfrom the same patients. As exemplified in Fig. 8A and illustrated inFig. 8C, immunostaining revealed significantly higher levels of GRP94expression in recurrences compared to the corresponding initial breasttumors. An increase in GRP94 score was observed in the second cancerevent in 87% of the patients, whereas for the remaining 13% the GRP94scores were equivalent in the first and second events (supplementaryTable 1). In tumors, GRP94 can be overexpressed as a consequence ofnutrient and oxygen deprivation via UPR activation [41,42]. However,in our resistant cell model, the increase in GRP94 expression was asso-ciated with a reduced ability to phosphorylate eIF2 and, subsequently,to reduce UPR activation. We, therefore, decided to evaluate whetherUPR activation was correlated to GRP94 expression in recurrent breastcancers. This experiment revealed that, in contrast to GRP94, P-eIF2staining was not modified in secondary compared to primary tumors(Figs. 8B and C). Moreover, there was no statistical correlation betweenGRP94 expression and P-eIF2 staining (not shown). This suggested that,as in Resox cells, GRP94 overexpression in recurrent tumors was notdue to UPR activation. These data are also supported by the bidimen-sional hierarchical clustering of patients and tumor features (includingtumor GRP94 and P-eIF2 staining and number of chemotherapeuticclasses received by the 15 patients of the study), which is presented inFig. 8D. In this clustering, P-eIF2 staining was not grouped with GRP94

Fig. 8. Overexpression of GRP94 in recurrent breast carcinomas. Representative immunostairecurrence (2nd event) from a single patient. Original magnifications: ×200 and ×400 fortumors. (D) Bidimensional hierarchical clustering of patients (horizontal axis) and tumor fthe color scale. Red cells indicate a score or value equal to 9 and white cells equal to 0anthracyclines, pyrimidines, and alkylating agents). A, 1st cancer event; B, 2nd cancer even

staining in the first or second event of cancer, also suggesting a lack ofassociation between these two variables. It is noteworthy that patientsare essentially grouped according to their modulation of P-eIF2 stainingbetween the first (A) and second (B) event of tumors. Remarkably, itcan be also deduced from the clustering that most of the patients thatreceived chemotherapies presented the highest values of GRP94 levelsin their corresponding recurrent tumor. This observation wasconfirmed statistically, as samples from patients treated with chemo-therapy had higher GRP94 levels than nontreated patients (Pb0.05).Finally, with regard to other indexes, no correlation was observedbetweenGRP94 expression and tumor grade, KI67 staining, age, hormo-notherapy, radiotherapy, HER2/neu, estrogen receptor or progesteronereceptor status (supplementary Table 1 and data not shown). Takentogether, these findings suggest that GRP94 expression is increased inchemoresistant tumors.

Conclusions

This study highlights a clear association between GRP94 expressionand cancer aggressiveness. Indeed, GRP94 appeared to promote a highlevel of proliferation and migration in a cell line resistant to oxidativestress and some chemotherapy drugs. In addition, GRP94 expressionwas increased in recurrent breast cancers. GRP94 upregulation could,therefore, represent a central mechanism by which cancer cells acquirean aggressive phenotype after chemoresistance. A model summarizingour results and explaining the potential relationship between

ning of GRP94 (A) and P-eIF2 (B) in the initial tumor (1st event) and the correspondinginset. (C) Comparison of GRP94 and P-eIF2 expression score in primary and recurrenteatures (vertical axis). Individual data for the clinicopathological factors are shown by. Chemo.: number (from 0 to 5) of chemotherapeutic classes used (taxanes, platins,t. *** Pb0.001, (ns) not significant.

Fig. 9. Hypothetical scheme of the relationships between ROS resistance and ER folding capacity in cancer cells.

1001N. Dejeans et al. / Free Radical Biology & Medicine 52 (2012) 993–1002

chemoresistance and ER folding capacity in cancer cells is depicted inFig. 9. Thus, many types of cancer cells exhibit increased intracellularlevels of reactive oxygen species (ROS), which can promote a highlevel of cell proliferation and an increased adaptation capacity [1].However, this can also sensitize cancer cells to prooxidative-basedchemotherapies.We have shown in this study that targeting the alteredredox state of cancer cells results in an increase of proteins of the ERmachinery dedicated to protein folding, including GRP94 and PDIs.These proteins allowed ROS-resistant cells to maintain a high prolifera-tion rate despite their elevated antioxidant capacity. GRP94 overexpres-sion also leads, probably throughmodulation of secreted proteins, to anacquisition of a highmigratory capacity. In agreementwith this hypoth-esis, previous reports have shown that GRP94 was particularlyabundant in cancer metastases [21,22,43,44], disseminated tumorcells [45], and radioresistant cell lines [46]. Moreover, targeting GRP94with a vaccination strategy suppressed cancer metastasis whencombinedwith radiation therapy [47]. GRP94may, therefore, constitutean important and specific potential therapeutic target to avoid cancerrecurrence. Taken together, our findings provide new therapeuticperspectives in the fight against cancer recurrence.

Supplementary materials related to this article can be foundonline at doi:10.1016/j.freeradbiomed.2011.12.019.

Acknowledgments

The authors thank Isabelle Blave, Véronique Allaeys, and AgnèsDelga for their excellent technical assistance. This work wassupported by grants from Télévie, Wallonie-Bruxelles International,and the Belgian Fonds National de la Recherche Scientifique (FNRS).Raphaël Beck is an FRIA recipient, Julien Verrax is an FNRS postdoctoralresearcher, Nicolas Dejeans is an FNRS-Télévie postdoctoral researcher.Christophe Glorieux and Bettina Bisig are FNRS-Télévie recipients.

References

[1] Trachootham, D.; Alexandre, J.; Huang, P. Targeting cancer cells by ROS-mediatedmechanisms: a radical therapeutic approach? Nat. Rev. Drug Discov. 8:579–591;2009.

[2] Verrax, J.; Pedrosa, R. C.; Beck, R.; Dejeans, N.; Taper, H.; Calderon, P. B. In situmodulation of oxidative stress: a novel and efficient strategy to kill cancer cells.Curr. Med. Chem. 16:1821–1830; 2009.

[3] Pelicano, H.; Carney, D.; Huang, P. ROS stress in cancer cells and therapeutic im-plications. Drug Resist. Updat. 7:97–110; 2004.

[4] Bey, E. A.; Bentle, M. S.; Reinicke, K. E.; Dong, Y.; Yang, C. R.; Girard, L.; Minna, J. D.;Bornmann, W. G.; Gao, J.; Boothman, D. A. An NQO1- and PARP-1-mediated celldeath pathway induced in non-small-cell lung cancer cells by beta-lapachone.Proc. Natl. Acad. Sci. U. S. A. 104:11832–11837; 2007.

[5] Verrax, J.; Stockis, J.; Tison, A.; Taper, H. S.; Calderon, P. B. Oxidative stress byascorbate/menadione association kills K562 human chronic myelogenous leukae-mia cells and inhibits its tumour growth in nude mice. Biochem. Pharmacol. 72:671–680; 2006.

[6] Trachootham, D.; Zhou, Y.; Zhang, H.; Demizu, Y.; Chen, Z.; Pelicano, H.; Chiao,P. J.; Achanta, G.; Arlinghaus, R. B.; Liu, J.; Huang, P. Selective killing of oncogeni-cally transformed cells through a ROS-mediated mechanism by beta-phenylethylisothiocyanate. Cancer Cell 10:241–252; 2006.

[7] Lu, J.; Chew, E. H.; Holmgren, A. Targeting thioredoxin reductase is a basis for can-cer therapy by arsenic trioxide. Proc. Natl. Acad. Sci. U. S. A. 104:12288–12293;2007.

[8] Landriscina, M.; Maddalena, F.; Laudiero, G.; Esposito, F. Adaptation to oxidativestress, chemoresistance, and cell survival. Antioxid. Redox Signal. 11:2701–2716;2009.

[9] Harding, H. P.; Zhang, Y.; Zeng, H.; Novoa, I.; Lu, P. D.; Calfon, M.; Sadri, N.; Yun, C.;Popko, B.; Paules, R.; Stojdl, D. F.; Bell, J. C.; Hettmann, T.; Leiden, J. M.; Ron, D. Anintegrated stress response regulates amino acid metabolism and resistance to ox-idative stress. Mol. Cell 11:619–633; 2003.

[10] Ma, Y.; Hendershot, L. M. The role of the unfolded protein response in tumour de-velopment: friend or foe? Nat. Rev. Cancer 4:966–977; 2004.

[11] Diehn, M.; Cho, R. W.; Lobo, N. A.; Kalisky, T.; Dorie, M. J.; Kulp, A. N.; Qian, D.;Lam, J. S.; Ailles, L. E.; Wong, M.; Joshua, B.; Kaplan, M. J.; Wapnir, I.; Dirbas,F. M.; Somlo, G.; Garberoglio, C.; Paz, B.; Shen, J.; Lau, S. K.; Quake, S. R.; Brown,J. M.; Weissman, I. L.; Clarke, M. F. Association of reactive oxygen species levelsand radioresistance in cancer stem cells. Nature 458:780–783; 2009.

[12] Spiotto, M. T.; Banh, A.; Papandreou, I.; Cao, H.; Galvez, M. G.; Gurtner, G. C.;Denko, N. C.; Le, Q. T.; Koong, A. C. Imaging the unfolded protein response in pri-mary tumors reveals microenvironments with metabolic variations that predicttumor growth. Cancer Res. 70:78–88; 2010.

[13] Moenner, M.; Pluquet, O.; Bouchecareilh, M.; Chevet, E. Integrated endoplasmicreticulum stress responses in cancer. Cancer Res. 67:10631–10634; 2007.

[14] Fu, Y.; Lee, A. S. Glucose regulated proteins in cancer progression, drug resistanceand immunotherapy. Cancer Biol. Ther. 5:741–744; 2006.

[15] Jamora, C.; Dennert, G.; Lee, A. S. Inhibition of tumor progression by suppressionof stress protein GRP78/BiP induction in fibrosarcoma B/C10ME. Proc. Natl. Acad.Sci. U. S. A. 93:7690–7694; 1996.

[16] Reddy, R. K.; Mao, C.; Baumeister, P.; Austin, R. C.; Kaufman, R. J.; Lee, A. S. Endo-plasmic reticulum chaperone protein GRP78 protects cells from apoptosis in-duced by topoisomerase inhibitors: role of ATP binding site in suppression ofcaspase-7 activation. J. Biol. Chem. 278:20915–20924; 2003.

[17] Ranganathan, A. C.; Zhang, L.; Adam, A. P.; Aguirre-Ghiso, J. A. Functional couplingof p38-induced up-regulation of BiP and activation of RNA-dependent proteinkinase-like endoplasmic reticulum kinase to drug resistance of dormant carcino-ma cells. Cancer Res. 66:1702–1711; 2006.

[18] Wang, J.; Yin, Y.; Hua, H.; Li, M.; Luo, T.; Xu, L.; Wang, R.; Liu, D.; Zhang, Y.; Jiang, Y.Blockade of GRP78 sensitizes breast cancer cells to microtubules-interfering agentsthat induce the unfolded protein response. J. Cell. Mol. Med. 13:3888–3897; 2009.

[19] Chen, X.; Ding, Y.; Liu, C. G.; Mikhail, S.; Yang, C. S. Overexpression of glucose-regulated protein 94 (Grp94) in esophageal adenocarcinomas of a rat surgicalmodel and humans. Carcinogenesis 23:123–130; 2002.

[20] Wang, Q.; An, L.; Chen, Y.; Yue, S. Expression of endoplasmic reticulum molecularchaperon GRP94 in human lung cancer tissues and its clinical significance. Chin.Med. J. (Engl.) 115:1615–1619; 2002.

[21] Wang, X. P.; Qiu, F. R.; Liu, G. Z.; Chen, R. F. Correlation between clinicopathologyand expression of heat shock protein 70 and glucose-regulated protein 94 inhuman colonic adenocarcinoma. World J. Gastroenterol. 11:1056–1059; 2005.

[22] Wang, X. P.; Wang, Q. X.; Ying, X. P. Correlation between clinicopathology and ex-pression of heat shock protein 72 and glycoprotein 96 in human gastric adenocar-cinoma. Tohoku J. Exp. Med. 212:35–41; 2007.

1002 N. Dejeans et al. / Free Radical Biology & Medicine 52 (2012) 993–1002

[23] Bini, L.; Magi, B.; Marzocchi, B.; Arcuri, F.; Tripodi, S.; Cintorino, M.; Sanchez, J. C.;Frutiger, S.; Hughes, G.; Pallini, V.; Hochstrasser, D. F.; Tosi, P. Protein expressionprofiles in human breast ductal carcinoma and histologically normal tissue. Elec-trophoresis 18:2832–2841; 1997.

[24] Mosmann, T. Rapid colorimetric assay for cellular growth and survival: applica-tion to proliferation and cytotoxicity assays. J. Immunol. Methods 65:55–63; 1983.

[25] Dejeans, N.; Tajeddine, N.; Beck, R.; Verrax, J.; Taper, H.; Gailly, P.; Calderon, P. B. En-doplasmic reticulum calcium release potentiates the ER stress and cell death causedby an oxidative stress in MCF-7 cells. Biochem. Pharmacol. 79:1221–1230; 2010.

[26] Horion, J.; Gloire, G.; El Mjiyad, N.; Quivy, V.; Vermeulen, L.; Vanden Berghe, W.;Haegeman, G.; Van Lint, C.; Piette, J.; Habraken, Y. Histone deacetylase inhibitortrichostatin A sustains sodium pervanadate-induced NF-kappaB activation bydelaying IkappaBalpha mRNA resynthesis: comparison with tumor necrosis fac-tor alpha. J. Biol. Chem. 282:15383–15393; 2007.

[27] Baudhuin, P.; Beaufay, H.; Rahman-Li, Y.; Sellinger, O. Z.; Wattiaux, R.; Jacques, P.;De Duve, C. Tissue fractionation studies. 17. Intracellular distribution of mono-amine oxidase, aspartate aminotransferase, alanine aminotransferase, D-aminoacid oxidase and catalase in rat-liver tissue. Biochem. J. 92:179–184; 1964.

[28] Fitzsimmons, S. A.; Workman, P.; Grever, M.; Paull, K.; Camalier, R.; Lewis, A. D.Reductase enzyme expression across the National Cancer Institute Tumor cellline panel: correlation with sensitivity to mitomycin C and EO9. J. Natl. CancerInst. 88:259–269; 1996.

[29] Beck, R.; Pedrosa, R. C.; Dejeans, N.; Glorieux, C.; Leveque, P.; Gallez, B.; Taper, H.; Eec-khoudt, S.; Knoops, L.; Calderon, P. B.; Verrax, J. Ascorbate/menadione-induced oxida-tive stress kills cancer cells that express normal or mutated forms of the oncogenicprotein Bcr-Abl. An in vitro and in vivo mechanistic study. Invest. New Drugs 29:891–900; 2011.

[30] Meunier, L.; Usherwood, Y. K.; Chung, K. T.; Hendershot, L. M. A subset of chaper-ones and folding enzymes form multiprotein complexes in endoplasmic reticu-lum to bind nascent proteins. Mol. Biol. Cell 13:4456–4469; 2002.

[31] Eletto, D.; Dersh, D.; Argon, Y. GRP94 in ER quality control and stress responses.Semin. Cell Dev. Biol. 21:479–485; 2010.

[32] Mao, C.; Wang, M.; Luo, B.; Wey, S.; Dong, D.; Wesselschmidt, R.; Rawlings, S.; Lee,A. S. Targeted mutation of the mouse Grp94 gene disrupts development and per-turbs endoplasmic reticulum stress signaling. PLoS One 5:e10852; 2010.

[33] Chatterjee, S.; Hirota, H.; Belfi, C. A.; Berger, S. J.; Berger, N. A. Hypersensitivity toDNA cross-linking agents associated with up-regulation of glucose-regulatedstress protein GRP78. Cancer Res. 57:5112–5116; 1997.

[34] Wanderling, S.; Simen, B. B.; Ostrovsky, O.; Ahmed, N. T.; Vogen, S. M.; Gidalevitz,T.; Argon, Y. GRP94 is essential for mesoderm induction and muscle developmentbecause it regulates insulin-like growth factor secretion. Mol. Biol. Cell 18:3764–3775; 2007.

[35] Staron, M.; Yang, Y.; Liu, B.; Li, J.; Shen, Y.; Zuniga-Pflucker, J. C.; Aguila, H. L.;Goldschneider, I.; Li, Z. gp96, an endoplasmic reticulum master chaperone for

integrins and Toll-like receptors, selectively regulates early T and B lymphopoi-esis. Blood 115:2380–2390; 2010.

[36] Randow, F.; Seed, B. Endoplasmic reticulum chaperone gp96 is required for innateimmunity but not cell viability. Nat. Cell Biol. 3:891–896; 2001.

[37] Braakman, I.; Helenius, J.; Helenius, A. Manipulating disulfide bond formation andprotein folding in the endoplasmic reticulum. EMBO J. 11:1717–1722; 1992.

[38] Pollard, M. G.; Travers, K. J.; Weissman, J. S. Ero1p: a novel and ubiquitous proteinwith an essential role in oxidative protein folding in the endoplasmic reticulum.Mol. Cell 1:171–182; 1998.

[39] Hatahet, F.; Ruddock, L. W. Protein disulfide isomerase: a critical evaluation ofits function in disulfide bond formation. Antioxid. Redox Signal. 11:2807–2850;2009.

[40] Karala, A. R.; Lappi, A. K.; Saaranen, M. J.; Ruddock, L. W. Efficient peroxide-mediated oxidative refolding of a protein at physiological pH and implicationsfor oxidative folding in the endoplasmic reticulum. Antioxid. Redox Signal. 11:963–970; 2009.

[41] Reddy, R. K.; Dubeau, L.; Kleiner, H.; Parr, T.; Nichols, P.; Ko, B.; Dong, D.; Ko, H.;Mao, C.; DiGiovanni, J.; Lee, A. S. Cancer-inducible transgene expression by theGrp94 promoter: spontaneous activation in tumors of various origins andcancer-associated macrophages. Cancer Res. 62:7207–7212; 2002.

[42] Paris, S.; Denis, H.; Delaive, E.; Dieu, M.; Dumont, V.; Ninane, N.; Raes, M.;Michiels, C. Up-regulation of 94-kDa glucose-regulated protein by hypoxia-inducible factor-1 in human endothelial cells in response to hypoxia. FEBS Lett.579:105–114; 2005.

[43] Wang, Q.; He, Z.; Zhang, J.; Wang, Y.; Wang, T.; Tong, S.; Wang, L.; Wang, S.; Chen,Y. Overexpression of endoplasmic reticulum molecular chaperone GRP94 andGRP78 in human lung cancer tissues and its significance. Cancer Detect. Prev. 29:544–551; 2005.

[44] Chiu, C. C.; Lin, C. Y.; Lee, L. Y.; Chen, Y. J.; Lu, Y. C.; Wang, H. M.; Liao, C. T.; Chang,J. T.; Cheng, A. J. Molecular chaperones as a common set of proteins that regulate theinvasion phenotype of head and neck cancer. Clin. Cancer Res. 17:4629–4641; 2011.

[45] Bartkowiak, K.; Effenberger, K. E.; Harder, S.; Andreas, A.; Buck, F.; Peter-Katalinic,J.; Pantel, K.; Brandt, B. H. Discovery of a novel unfolded protein response pheno-type of cancer stem/progenitor cells from the bone marrow of breast cancer pa-tients. J. Proteome Res. 9:3158–3168; 2010.

[46] Lin, T. Y.; Chang, J. T.; Wang, H. M.; Chan, S. H.; Chiu, C. C.; Lin, C. Y.; Fan, K. H.;Liao, C. T.; Chen, I. H.; Liu, T. Z.; Li, H. F.; Cheng, A. J. Proteomics of the radioresis-tant phenotype in head-and-neck cancer: Gp96 as a novel prediction marker andsensitizing target for radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 78:246–256;2010.

[47] Liu, S.; Wang, H.; Yang, Z.; Kon, T.; Zhu, J.; Cao, Y.; Li, F.; Kirkpatrick, J.; Nicchitta,C. V.; Li, C. Y. Enhancement of cancer radiation therapy by use of adenovirus-mediated secretable glucose-regulated protein 94/gp96 expression. Cancer Res.65:9126–9131; 2005.