adaptive upregulation of egfr limits attenuation of tumor growth

Microenvironment and Immunology

Adaptive Upregulation of EGFR LimitsAttenuation of Tumor Growth by Neutralizing IL6Antibodies, with Implications for CombinedTherapy in Ovarian CancerCarla S. Milagre1, Ganga Gopinathan1, Gemma Everitt1, Richard G. Thompson1,Hagen Kulbe1, Haihong Zhong2, Robert E. Hollingsworth2, Richard Grose1,David D.L. Bowtell3, Daniel Hochhauser4, and Frances R. Balkwill1

Abstract

Excess production of the proinflammatory IL6 has both localand systemic tumor-promoting activity in many cancers, includ-ing ovarian cancer. However, treatment of advanced ovariancancer patients with a neutralizing IL6 antibody yielded littleefficacy in a previous phase II clinical trial. Here, we report resultsthat may explain this outcome, based on the finding that neu-tralizing antibodies to IL6 and STAT3 inhibition are sufficient toupregulate the EGFR pathway in high-grade serous and otherovarian cancer cells. Cell treatment with the EGFR inhibitorgefitinib abolished upregulation of the EGFR pathway. Combin-ing neutralizing IL6 antibodies and gefitinib inhibited malignant

cell growth in 2D and 3D culture. We found that ErbB-1 waslocalized predominantly in the nucleus of ovarian cancer cellsexamined, contrastingwithplasmamembrane localization in lungcancer cells. Treatment with anti-IL6, gefitinib, or their combina-tion all led to partial restoration of ErbB-1 on the plasma mem-brane. In vivo experiments confirmed the effects of IL6 inhibitionon the EGFR pathway and the enhanced activity of a combinationof anti-IL6 antibodies and gefitinib on malignant cell growth.Taken together, our results offer a preclinical rationale to combineanti-IL6 and gefitinib to treat patients with advanced stage ovariancancer. Cancer Res; 75(7); 1–10. �2015 AACR.

IntroductionAbnormal regulation of IL6 and its major downstream tran-

scription factor STAT3 is a feature of many human cancers (1).Constitutive production of IL6 and STAT3 occurs downstream ofsome oncogenic mutations in malignant cells and there is strongevidence that IL6 is tumor promoting in many different experi-mental cancer models (2).

IL6 is implicated in the pathophysiology of high-gradeserous ovarian cancer (HGSC) and clear cell ovarian cancer(3–5). We previously demonstrated that constitutive IL6 pro-duction by malignant cells is a major regulator of cancer-relatedinflammation and cytokine networks in HGSC, having impor-tant local and systemic tumor-promoting actions (3, 4, 6).

In mouse models, anti-IL6 antibodies have antitumor activity,inhibiting communication between malignant cells and stroma,reducing the leukocyte infiltrate and angiogenesis, with evidenceof vessel normalization (3). We reported some activity in a phaseII clinical trial of an anti-IL6 antibody in patients with advancedHGSC, but sustained responses were not achieved (3). One of the17 HGSC patients treated had a partial response to anti-IL6therapy, 7 others had periods of disease stabilization, and sys-temic levels of some cytokines, inflammatory, and tumor bio-markers were reduced during the therapy but rose as the patientsregressed.

There could be many reasons for the low efficacy of IL6blockade in patients with HGSC. Our previous research wouldsuggest that a major factor might be the complexity of malig-nant cell cytokine production. As we have shown that inflam-matory cytokines such as IL6 interact in networks with otherinflammatory mediators and growth factors in ovarian cancercells (4, 7, 8), we hypothesized that inhibiting constitutiveIL6 production by malignant cells may induce reciprocalfeedback regulation in other signaling pathways that compen-sates for their action and reduces efficacy of neutralizinganti-IL6 antibodies.

To investigate this hypothesis, we treated ovarian cancer cellswith neutralizing anti-IL6 antibodies and studied changes inintracellular signaling pathways. We found that inhibiting IL6signaling in these cells and ovarian cancer xenografts upregulatedEGFR signaling and ERK activation. A combination of EGFRinhibition by gefitinib and neutralizing anti-IL6 antibodies hadenhanced anticancer activity.

1Barts Cancer Institute, Queen Mary University of London, Charter-house Square, London, United Kingdom. 2MedImmune, One MedIm-mune Way, Gaithersburg, Maryland. 3Cancer Genomics and GeneticsProgram, Peter MacCallum Cancer Centre, Research Division, PeterMacCallum Cancer Centre, Melbourne, Australia. 4UCL Cancer Insti-tute, University College London, United Kingdom.

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

CorrespondingAuthor:FrancesR. Balkwill, Barts Cancer Institute, CharterhouseSquare, London, EC1M 6BQ, United Kingdom. Phone: 44-207-882-3851; Fax:44-207-882-3885; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-14-1801

�2015 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org OF1

Research. on April 14, 2018. © 2015 American Association for Cancercancerres.aacrjournals.org Downloaded from

Published OnlineFirst February 10, 2015; DOI: 10.1158/0008-5472.CAN-14-1801

http://cancerres.aacrjournals.org/

Materials and MethodsOvarian cancer cell lines

The IGROV-1 linewas recently characterized as a hypermutatedline but does have TP53 and BRCA2 mutations typical of HGSC(9). The AOCS1 cell linewas established fromapatient diagnosedwith HGSC, Silverberg grade 3, with <1cm residual disease afterprimary surgery. The patient had 18 months progression-freesurvival after six cycles carboplatin and paclitaxel adjuvant che-motherapy but showed no response to Line 2 liposomal doxo-rubicin. The cell line AOCS1 was established frommaterial takenat second relapse. AOCS1 stains with antibodies to EPCAM andPax8. The G33 cell line was established in our laboratory fromomental metastases of a patient with HGSC after chemotherapy.It has a p53 mutation W146� and is positive for EPCAM andPax8. Quality control of all cell lines was carried out by frequentSTR analysis (Eurofins MWG), mycoplasma testing (InvivoGen)and cell lines were used for 4 to 5 passages before new cellswere recovered from frozen master stocks. Cells were cultured inRPMI-1640 supplemented with 10% FCS and 1% penicillin/streptomycin. Cells were counted using a Vi-cell cell counter(Beckman Coulter) on days 3 and 7. All technical and experi-mental replicates were repeated in triplicate.

Primary cancer cellsAll the tissue sampleswere obtained under the guidelines of the

HumanTissueAuthority Act 2004, and all patients had givenpriorconsent under the Research Ethical Committee Project reference10/H0304/14.

TreatmentsCells were treatedwith 10 mg/mL anti-IL6 antibody (MEDI5117;

Medimmune), and 1 mmol/L gefitinib (AstraZeneca AZD1839).MEDI5117 is a human IgG1k monoclonal antibody thatbinds to IL6 with sub-pM affinity and neutralizes it by pre-venting the binding to IL6R. MEDI5117 was generated usingphage display technology. It bears a triple mutation (referredto as YTE) in the Fc domain of the heavy chain that extends itsstability in circulation. MEDI5117 is active in several preclin-ical cancer models, including non–small cell lung cancer,prostate cancer, breast cancer, and ovarian cancer xenograftsin mice (Manuscript in preparation, Medimmune). An IgG1isotype control antibody was generated by MedImmuneand was used at the same concentration as anti-IL6 antibody(10 mg/mL).

Protein extraction from mouse tumorsOf note, 75 mg of tumor tissue was lysed with 1 mL of ice-

cold lysis buffer (150 mmol/L NaCl 20mmol/L Tris, pH 7.5,1 mmol/L EDTA 1 mmol/L EGTA 1% Triton X-100) withprotease and phosphatase inhibitors. Samples were then dis-sociated using gentleMACS Dissociator. After dissociation,samples were centrifuged at 1,500 rpm for 2 minutes. Sampleswere always kept on ice between procedures. Next, using aprobe sonicator set at 40% amplitude, tissues were sonicatedfor 5 to 15 seconds bursts. Sonicated samples were thenrotated for 30 minutes at 4�C followed by a centrifugationfor 15 minutes at 13,200 rpm at 4C. The pellet was discardedand protein concentration measured. Lysates were frozen at�80�C until loaded on gels.

Western blottingCells were washed with PBS and harvested using RIPA Buffer

(R0278, Sigma) with 1� proteinase inhibitors. Protein quantifi-cation was performed using the Bradford reagent (Sigma-Aldrich),according to the manufacturer's instructions. Cell extracts (25 mg)were run on a NuPAGE Novex 4% to 12% Bis-Tris Gel, 1.5 mmand transferred to a nylon membrane. The membrane wasblocked overnight (4�C in PBS with 0.1% Tween and 5% milkpowder) and probed using the following antibodies: Phospho-Stat3 (Tyr705; 1:1,000 #9145, Cell Signaling Technology), Stat3(1:1,000 #4904, Cell Signaling Technology), p-ERK (1:1,000sc-7383, Santa Cruz Biotechnology), ERK (1:1,000 #9102, CellSignaling Technology), phospho-ErbB-1 (1:1,000 #2220, CellSignaling Technology), ErbB-1 (1:1,000 2232, Cell SignalingTechnology), phospho-ErbB-4 (1:500 #3790, Cell Signaling Tech-nology), ERbB-4 (1:1,000 #4795, Cell Signaling Technology),TNFa (1:1,000 ab6671, Abcam), Jagged1 (1:1,000 SC-8303,Santa Cruz Biotechnology), a-tubulin (1:2,000 sc-8035, SantaCruz Biotechnology), lamin A/C (1:2,000 #2032, Cell SignalingTechnology), b-actin (1:5,000 A5316, Sigma). A rabbit or mousehorseradish peroxidase-conjugated secondary antibody (GEHealthcare) incubation allowed visualization using enhancedchemiluminescence (ECL; GE Healthcare). Protein concentrationequivalence was confirmed by anti–b-actin antibody.

Quantifying Western blot analysesWestern blot analyses were quantified using ImageJ analysis.

The images were set to grayscale 8-type bit and a rectangular boxwas drawn to enclose a single lane. The box was then selected asthe first lane and the same process was repeated for the remainingof the lanes. The relative density of the contents of each lane wasthen plotted as histograms and a straight line was drawn under-neath where the peak ends to subtract any background. The wandtool was then used to get the measurements of each profile. Theratios were obtained by normalizing the density of each band toits total protein or loading control (10).

ImmunoprecipitationA total of 2 � 107 cells were homogenized in 1 mL of RIPA

buffer (R0278, Sigma) containing phosphate inhibitor cocktail(P-5726, Sigma) and incubated on ice for 10 minutes. Sampleswere centrifuged at 10,000 g for 10 minutes at 4�C, and super-natant was then transferred to a new tube. Protein concentrationof the supernatant was determined. Lysates were cleaned usingRabbit IgG control and 20 mL of Protein A/G PLUS-Agarose beads(sc-2003; Santa Cruz Biotechnology) for 30minutes at 4�C. Beadswere pelleted by centrifugation at 2,500 rpm for 5 minutes andlysates transferred to a clean tube. Samples were immunopreci-pitated for 2 hours at 4�C using 500 mg of total lysate and 2 mg ofappropriate antibody (ErbB-1 or STAT3). Lysates were then rotat-ed overnight at 4�C with 20 mL of A/G PLUS Agarose beads. Thenext day, the Agarose beadswerewashed four times in RIPA bufferfollowed by a 5 minute centrifugation at 1,000�g at 4�C. Afterfinal wash, supernatant was discarded and pellet was resuspendedin 40mL of 1� electrophoresis buffer, boiled for 5minutes at 70�Cbefore separation using SDS-PAGE.

Immunofluorescence staining and nuclear quantificationA total of 1 � 104 cells were plated on coverslips (Nalge Nunc

International, ThermoFisher Scientific) and kept in culture with

Milagre et al.

Cancer Res; 75(7) April 1, 2015 Cancer ResearchOF2




RPMI-1640 medium supplemented with 10%FCS and 1% pen-icillin/streptomycin. Next day, mediumwas replacedwith serum-free medium. Cells were treated with 10 mg/mL of anti-IL6antibody and/or 1 mmol/L gefitinib and incubated at 37�C, 5%CO2 for 24 hours. Supernatants were discarded and cells werefixed with 4% paraformaldehyde for 30 minutes, permeabilizedin 0.5% Triton X-100 for 30 minutes, and washed three times in0.1%TritonX-100. Stainingwas performedovernightwith ErbB-1(1:50 Alexa Fluor 488, Santa Cruz Biotechnology) or isotypecontrol (1:200, R&D systems). Primary antibody was washedwith PBS. Slides were counterstained with DAPI Prolong Gold(Invitrogen) and images captured by Nikon Eclipse 80i micro-scope and Image-Pro Plus software (Media Cybernetics).

Nuclear and cytoplasmic quantification of ErbB-1 immunoflu-orescent data was done using MetaMorph Microscopy Automa-tion & Image Analysis Software following the instructions pro-vided by the manufacturer.

siRNA transfectionA total of 2 � 104 cells were plated in a 6-well plate and

transfected with 100 pmol of siRNA pool of four different target-ing sequences or nonspecific scramble control (stock concentra-tion at 50 pmol/mL) in 250 mL serum-free medium. Of note, 4 mLof lipofectamine 2000 (Life technologies) was also added to themix. Samples were vortexed and incubated at room temperaturefor 15 minutes. After 6 to 8 hours, medium was removed andreplaced with fresh RPMI-1640 supplemented with 10% FCS.Twenty-four hours later, protein quantification was accessed and20 mg of cell lysate was analyzed by Western blot analysis.

IHCParaffin-embedded sections of tumors collected from mouse

xenograft were stained with antibodies for Phospho-p44/42MAPK (Erk1/2; Thr202/Tyr204; Cell Signaling Technology,9106S), Tyr705 phospho-STAT3 (Cell Signaling Technology,9145), and Y1068 phospho-EGFR (Cell Signaling Technology,8543S). Slides were counterstained with hematoxylin. Negativecontrols were isotype matched. The conditions used for stainingwith individual antibodieswere in accordwith themanufacturers'recommendations.

Growth of tumors in mice and bioluminescence imagingA total of 10 � 106 luciferase-expressing IGROV-1 cells

(IGROV-1luc) were injected intraperitoneally (i.p.) into 20 g6- to 8-week-old female BALB/c nu/nu mice (purchased fromCharles River, UK Ltd). These cell lines were generated aspreviously published (11). Mice were observed daily for tumorgrowth and killed once they reached end points as defined inour Home Office Project Licence PPL 70/7411 (20% increasein abdominal distension). All observable peritoneal tumorswere dissected and weighed at the end of the experiment. Therewere 8 mice in each group and two separate experiments wereconducted.

Statistical analysisStatistical analyses were carried out using Prism GraphPad

software. Statistical significance was calculated using ANOVA andthe Student t test. Findings are presented as SEM.

ResultsIL6 inhibition activates EGFR signaling in ovarian cancer cells

We initially tested the effect of neutralizing antibodies to IL6 onnewly established high-grade serous ovarian cancer cell lines(AOCS1, G33), on freshly isolated cells from HGSC ascites andon IGROV-1, a previously characterized IL6 producing ovariancancer line of uncertain histotype but with TP53 and BRCA2mutations (9). We first asked whether treating cells with a neu-tralizing antibody to IL6 had an impact on other intracellularsignaling pathways that were likely to be active in these cells, witha focus on ErbB signaling. Cells were cultured for 3 and 7 days inthe presence of an anti-IL6 antibody (MEDI5117) or an IgGisotype control (Fig. 1A, for AOCS1 and primary cells; Supple-mentary Fig. S1A for IGROV-1 cells).

Western blot analysis and quantification showed, as expected,that inhibition of IL6 reduced STAT3 phosphorylation (Supple-mentary Fig. S1A–S1C,). However, in the presence of IL6 inhibi-tion, the cells activated alternative signaling pathways through

Figure 1.IL6 inhibition activates the EGFR signaling pathway. A, 2 � 105 HGSC cells(AOCS1) and primary cells were treatedwith anti-IL6 antibody (10 mg/mL) for3 and 7 days and lysates were analyzed byWestern blotting. Data shown aretypical of three independent experiments. B, 2 � 105 AOCS1 and G33 HGSCcells were treated with anti-IL6 antibody (10 mg/mL), gefitinib (1 mmol/L)alone, or with a combination of both inhibitors.

IL6 Inhibition Upregulates EGFR Signaling

www.aacrjournals.org Cancer Res; 75(7) April 1, 2015 OF3




ErbB-1, ErbB-4, and ERK phosphorylation (Fig. 1A and Supple-mentary Fig. S1). The results in Fig. 1 and Supplementary Fig. S1were consistent in repeated experiments. To demonstrate this, weshowmean results of the key finding from three experiments withthe AOCS1 cell line (See Supplementary Fig. S2). We alsoobserved that IL6 inhibition leads to a modest increase in releaseof EGFR ligands (HB-EGF, TGFa, and EGF) by the cells (data notshown).

EGFR family members are overexpressed in approximately70% of ovarian cancers and this is associated with a poor prog-nosis (12, 13). However, single-agent targeting of ErbB-1 hasshown minimal activity in patients with ovarian cancer (13, 14).Our results suggested that combination therapies targeting boththe EGFR and IL6 pathways might have therapeutic potential. Toinvestigate this, cells were treated with gefitinib, a selective inhib-itor of the EGFR tyrosine kinase bybinding to theATP-binding siteof the enzyme (15). Gefitinib treatment alone decreased phos-phorylation of ErbB-1 and ERK but induced pSTAT3 activation(Fig. 1B), as previously described (16). In AOCS1, a combinationof gefitinib and the anti-IL6 antibody also decreased ErbB-1

phosphorylation and pERK levels, but in this instance, pSTAT3activation decreased as well (Fig. 1B). Similar results wereobtained with the G33 cell line (Fig. 1B, right) and IGROV-1cells (data not shown). Quantification of these blots can be foundin Supplementary Fig. S3.

Combined therapy that targets IL6 and EGFR pathways reducesproliferation in 2D and 3D cultures

To investigate the biologic relevance of these observations, westudied growth of the cell lines in the presence of anti-IL6 andgefitinib. AOCS1 and IGROV-1 cells were treatedwith the anti-IL6antibody, gefitinib, or their combination. In 2D culture (Fig. 2Aand B), IL6 inhibition had no effect on cell proliferation as wehave previously reported (3). Gefitinib significantly reduced cellcounts and this growth inhibition was greater in the combinationtreatment group (Fig. 2A and B).

We next used a 3D cell culture system adapted from (17) thatmimics the microenvironment of the human omentum, a majorsite of peritoneal cancer metastasis. The model consisted offibroblasts embedded in a collagen matrix covered by a layer of

Figure 2.Anti-IL6 combined with gefitinibreduces proliferation in vitro andtumor growth area on 3D culturemodels. A and B, AOCS1 and IGROV-1cells were treated with an anti-IL6antibody (10 mg/mL) and/or gefitinib(1 mmol/L) for 7 days. C and D, H&Estaining sections of 3D cultures ofAOCS1 cells. AOCS1 cells were seededon top of a layer of mesothelial cellsand fibroblasts embedded in collagen.Data in A, B, and D are shown as mean� SEM of three independentexperiments. Statistical analysis wasperformed and is shown as � , P�0.05;��, P � 0.01; �� , P � 0.001.

Milagre et al.





primary human mesothelial cells. AOCS1 cells were grown for 2weeks in this system and then the area of malignant cell growthwas measured using ImageJ (Fig. 2C and D). Both anti-IL6 andgefitinib reduced malignant cell growth in the organotypic cul-tures and the greatest inhibition was seen in the combinationtreatment group (Fig. 2D).

ErbB-1 has a nuclear localization in ovarian cancer cells andthis is altered by IL6 inhibition

We also investigated the effects of anti-IL6 on EGFR in thecells using immunofluorescence and confocal microscopy.Confocal images and z-stack analysis of the images revealedthat ErbB-1 was primarily located in the nucleus (nErbB-1) inall three lines (Fig. 3A–C and Supplementary Fig. S4A, z-stacktop). This observation has been previously reported in someovarian cancer cell lines and human ovarian cancer biopsies(18). Upon treatment with anti-IL6 antibody for 24 hours,nErbB-1 staining was reduced and in some of the cells analyzed,relocalized to the plasma membrane (Fig. 3A–C and Supple-mentary Fig. S4A, z-stack bottom). As an experimental control,we also studied A549 cells, a lung adenocarcinoma cell linewith well-characterized cell surface expression of ErbB1. ErbB-1remained at the plasma membrane whether or not the cellswere treated with anti-IL6 (Fig. 3D). In a control for thespecificity of antibody staining, IGROV-1 cells were treatedwith siRNA to ErbB-1 or a scramble control, resulting in analmost complete removed staining in both nucleus and cyto-plasm (Fig. 3E and F). We then carried out nuclear and cyto-plasmic fractionation of cells treated either with IgG controlantibody or the anti-IL6 antibody. Anti-IL6 treatment resultedin a reduction of ErbB-1 in the nuclear fraction of IGROV-1 cellsat 24 hours and 3 days treatment with anti-IL6 (Fig 3G and Hand Supplementary Fig. S4B). Similar results were obtainedwith AOCS1 (Supplementary Fig. S5A). Tubulin and nuclearlamin antibodies were used to test separation of cytoplasmicand nuclear fraction, respectively, confirming the purity of thefractions (Fig. 3G and H and Supplementary Fig. S5A).

Anti-IL6 antibodies and b-importin silencing both disruptErbB-1 nuclear localization

Next, we tested the effects of gefitinib plus/minus anti-IL6 onnErbB-1 (Fig. 4A). Gefitinib, anti-IL6 treatment, and their com-bination relocalized ERB-1 expression to the membrane.

The presence of growth factor receptors in the nucleus iswell documented in several human cancers and nuclear trans-location of EGFR family members has been extensivelydescribed (19–21), but the mechanism of translocation andthe role of EGF receptor family members in the nucleus ofmalignant cells is still unclear (22–24). Others have reportedthat ErbB-1 colocalizes and interacts with b-importin in thy-roid cells to be translocated into the nucleus (19, 22, 24). Wetherefore treated AOCS1 cells with b-importin siRNA. Knock-down of b-importin reduced nErbB-1; this was not observedwhen AOCS1 cells were treated with scrambled control siRNA(Fig. 4A).

Data from the above experiments were quantified using meta-morph analysis software and correspond to an average of fivemicroscopefields per condition (Fig. 4B). The intensity of nErbB-1was reduced after treating the cells with either anti-IL6 antibody,gefitinib, or the combination of both agents (Fig. 4B). Studies

have suggested that ErbB-1 interacts with other transcriptionfactors, such as STAT3, in the nucleus (25). Thus, we next per-formed coimmunoprecipitation to seewhether ErbB1 colocalizedwith STAT3. Figure 4C shows that immunoprecipitation of ErbB-1pulls down STAT3 in AOCS1 cells, thus suggesting that IL6inhibition and consequent STAT3 reduction in the nucleus mightoffer a possible explanation for the reduction of ErbB-1 in thenucleus.

Our work so far has shown that ovarian cancer cells are able tocompensate for reduced autocrine IL6 signaling by upregulatingthe EGFR signaling pathway. In addition, it appears that in HGSCcell lines, ErbB1 is located to the nucleus and that treatment witheither anti-IL6 antibody or gefitinib, or their combination, leadsto reinstatement of ErbB1 on the plasma membrane of AOCS1,G33, and IGROV-1 cells.

To test whether treatment of ovarian cancer cells with bothgefitinib and anti-IL6 antibodies was able to enhance antitumoractivity, we conducted experiments in peritoneal xenografts.

Gefitinib in combination with anti-IL6 reduces tumor weightin mouse xenografts

We grew IGROV-1 cells as intraperitoneal xenografts in nudemice and treated the mice with the anti-human IL6 antibody,gefitinib, or both in combination. Results of two independentexperiments were combined and data are shown in Fig. 5.

Twenty-one days after treatment, tumors were extracted andweighed. IL6 inhibition reduced tumor weight as previouslyshown (Fig. 5A; ref. 3). Gefitinib treatment alone was able tosignificantly reduce tumor weight compared with the IgGcontrol group. However, the combination of anti-IL6 antibo-dies and gefitinib was the most effective in reducing tumorweight (Fig. 5A).

Histologic analysis of tumors collected from mice reveal-ed a reduction of pSTAT3 with IL6 inhibition, which wasaccompanied by an increase in pERK and pErbB-1 (Fig. 5B).We next analyzed the cellular localization of ErbB-1 in thexenograft sections. Results show that tumors from controlmice have nuclear ErbB-1. Tumors from mice treated withanti-IL6, gefitinib, or the combination have mainly plasmamembrane ErbB-1 (Fig. 5C) as previously suggested in Fig. 3.To assess further the effect of anti-IL6 on the EGFR andIL6 pathways in vivo, we extracted protein from the mousetumors and assessed the expression of pSTAT3, pErbB-1, andpERK. IL6 inhibition in vivo effectively reduced pSTAT3 acti-vation and led to an increase in pErbB-1 and pERK (Fig. 5D).In addition, we used these lysates to demonstrate that in vivo,IL6 inhibition reduces TNFa and Jagged1 levels (Fig. 5E), twocomponents of the IL6 tumor cytokine network that wepreviously described (3).

A summary diagram of our results is shown in Fig. 6.

DiscussionAlthough many preclinical studies and clinical observations

suggest that IL6 inhibitors would be of therapeutic benefit inpatients with advanced cancers (1, 3, 6, 8), clinical trials havenot shown conclusive evidence of activity, apart from inpatients with Castleman disease, which is primarily driven byIL6. In our clinical trial of a therapeutic anti-IL6 antibody inpatients with HGSC, we found some evidence of disease sta-bilization but this was not sustained (3). All our evidence to






date would suggest that the effects of IL6 in this disease aredriven by constitutive malignant cell production of IL6 (3, 6–8). Thus we investigated the possibility that IL6 inhibitionmight induce some reciprocal feedback regulation of otherpathways.

In our study, treatment with the anti-IL6 antibody did notcompletely abrogate STAT3 phosphorylation. IL6 is not theonly cytokine/growth factor that activates STAT3, hence neu-tralizing IL6 would not always be expected to completelyabrogate STAT3 phosphorylation. However, experimental data

Figure 3.ErbB-1 is located in the nucleus ofHGSC cells. A–C, confocal analysis ofHGSC cell lines (AOCS1, G33) andIGROV-1 cells seeded on coverslipsand treated with IgG control(10 mg/mL) or anti-IL6 antibody(10 mg/mL) for 24 hours. Top panelshows the nuclear localization ofErbB-1 when cells are treated with anIgG control (see green and red lines inC highlighting nuclear localization ofErbB-1 after a z-stack image merge).After treatment with an anti-IL6antibody, the amount of ErbB-1fluorescence in the nucleus wasreduced and appeared moreconcentrated at the plasmamembrane. D, A549 cells treated withan IgG control (top) or anti-IL6(bottom). Confocal imaging showsErbB-1 located at the plasmamembrane in both groups. E, confocalanalysis of IGROV-1 cells transfectedwith scramble siRNA (top) or siErbB-1(bottom). F, percentage of nErbB-1–positive cells after transfection withscramble or ErbB-1 siRNA. Resultsrepresented on graph correspond tothe average obtained in four fieldsanalyzed. G and H, IGROV-1 cells weretreated with an IgG control or an anti-IL6 antibody for 24 hours and 3 daysand subjected to nuclear andcytoplasmic fractionation. Tubulin andlamin A/C were used for a qualitycontrol of the cytoplasmic/nuclearfractionation. Results shown aretypical of three independentexperiments.

Milagre et al.





in the literature would suggest that partial STAT3 abrogation istherapeutically significant. For example, tumor cells thatbecome dependent on persistent STAT3 signaling are moresensitive to STAT3 inhibition than normal cells (26). In experi-ments with a lymphoma model, partial downregulation ofSTAT3 (40%) was responsible for strong antitumor effects invivo (27) and in colon cancer xenografts, partial inhibition ofJAK/STAT3 signaling pathway is enough to suppress the growthof colon cancer xenografts (28).

In addition, our previous research on cytokine networks inmalignant cells (7) would suggest that rational combination oftherapies targeting the malignant cells and their interaction withthe microenvironment is worthy of investigation.

Links between the IL6 and EGFR pathway have already beenreported (29–32). For instance, EGFR stimulation of ovariancancer cell lines increased IL6 production (29), but none of thecell lines used in the latter paper are now thought to be ofHGSC origin (9). In addition, in human lung adenocarcinomas,mutations in the EGFR kinase domain stimulated STAT3 activa-tion via IL6 production (30).

Hence we decided to study the influence of IL6 inhibition onEGFR signaling. The cell lines that we used in our experimentsdid not have EGFR mutations (data not shown). In this paper,we have used cell lines and primary tumor cells from peritonealcancers that are collectively termed "ovarian" cancers. Asexplained in the methods section, AOCS1 and G33 are newcell lines recently established from HGSC patient biopsies. Theprimary cancer cells used in our experiments were from path-

ologically confirmed HGSC patient ascites. The cancer of originof the ovarian cancer cell line IGROV-1 is uncertain (33). Wehave included it in our experiments as an example of an IL6producing cell line that forms peritoneal xenografts in nudemice (3).

Our results led us to test combinations of anti-IL6 andgefitinib and our experiments suggested that this may havetherapeutic potential. However, our recent increased under-standing of the natural history and genetic drivers of HGSC(34–36) means that some of the cell lines and syngeneicmodels (e.g., ID8 cells) are not relevant models of the disease.A new genetic model of HGSC has recently been published (37)and in the next few years as this, or transplantable cell linesderived from such mice, become more widely used, we shouldhave more appropriate models of HGSC for preclinical experi-ments. IL6 produced by malignant cells has paracrine actionsnot only on angiogenesis, but also on the extent and phenotypeof the leukocyte infiltrate (3, 7). Hence, a combination of IL6and EGFR inhibition may be more powerful in a syngeneicmodel or in patients with HGSC compared with the resultsdescribed here using immunocompromised mouse xenograftmodels.

Nuclear EGFR signaling is thought to confer resistance tovarious anticancer therapies (23) and may explain the poorresponse of patients with ovarian cancer to gefitinib (38). All thecells used in this paper were resistant to the anti-EGFR antibodycetuximab (data not shown) and there was no interactionbetween cetuximab and anti-IL6. The nuclear expression of

Figure 4.Gefitinib and anti-IL6 reduces nuclearErbB-1 in HGSC cells. A, confocalanalysis of AOCS1 cells treated withIgG control, anti-IL6 antibody,gefitinib, gefitinibþanti-IL6 antibody,scramble siRNA or b-importin siRNA.After 24 hours, cells were fixed andstained for ErbB-1. Results observedwere validated in five independentexperiments. B, nErbB-1 intensity wasquantified usingmethamorph analysissoftware. Data correspond to theaverage of 5HPF analyzed per group.Statistical analysiswas performed andis shown as �� , P � 0.001. C, STAT3Western blot analysis after ErbbB-1immunoprecipitation in AOCS1 cells.






ErbB-1 may explain this finding (20). The partial restoration ofmembrane-localized ErbB-1 by anti-IL6 antibody treatment maynot be sufficient to permit inhibition by cetuximab.

As both gefitinib and therapeutic anti-IL6 antibodies havebeen given safely to patients with ovarian cancer, and bothtargets are implicated in this disease, specifically HGSC, we

Figure 5.Effects of inhibition of IL6 and/or ErbB-1 in mouse ovarian cancer xenografts. A, weight of dissected peritoneal tumors collected from mice injectedintraperitoneally with IGROV-1 and treated with either an IgG antibody, an anti-IL6 antibody, gefitinib, or their combination. Data in A are shown as mean � SEMof two independent experiments, with a total of 8 mice per experimental group in each experiment. Statistical analysis was performed and is shown as � , P �0.05; ��, P � 0.01; ��, P � 0.001. B, pSTAT3, pERK, and pErbB-1 staining on mouse tumor sections from the above experiment. C, ErbB-1 staining on mousetumor sections. D, Western blot analysis and quantification of pSTAT3, pErbB-1, and pERK in protein extracts from mouse xenografts treated with an anti-IL6antibody. E, Western blot analysis and quantification of TNFa and Jagged1 in protein extracts from mouse xenografts treated with an anti-IL6 antibody.Quantification was done using ImageJ software and is always relative to the total protein. Results represented on graphs correspond to fold increase.

Milagre et al.





suggest that our data provide a rationale for testing theircombination.

Disclosure of Potential Conflicts of InterestR.E. Hollingsworth is the senior director, oncology research inMedImmune.

No potential conflicts of interest were disclosed by the other authors.

Authors' ContributionsConception and design: C.S. Milagre, H. Kulbe, H. Zhong, R.E. Hollingsworth,D. Hochhauser, F.R. BalkwillDevelopment of methodology: C.S. Milagre, G. Gopinathan, G. Everitt,R.G. ThompsonAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): C.S. Milagre, G. Everitt, R.G. Thompson, H. Zhong,D.D.L. BowtellAnalysis and interpretation of data (e.g., statistical analysis, biosta-tistics, computational analysis): C.S. Milagre, G. Gopinathan, R. Grose,D.D.L. BowtellWriting, review, and/or revision of the manuscript: C.S. Milagre, H. Zhong,R.E. Hollingsworth, R. Grose, D.D.L. Bowtell, D. Hochhauser, F.R. Balkwill

Administrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): C.S. Milagre, R.G. Thompson, H. Kulbe,R.E. HollingsworthStudy supervision: C.S. Milagre, D. Hochhauser, F.R. BalkwillOther (P.I. on Cancer Research UK Programme Grant that was the majorfunder of this study): F.R. Balkwill

AcknowledgmentsThe authors thank Darius Etemadmoghadam and Chris Mitchell for pro-

viding the AOCS1 cells.

Grant SupportD. Hochhauser received a commercial seed grant from Merck Serono.The costs of publication of this article were defrayed in part by the

payment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received June 16, 2014; revised January 15, 2015; accepted January 16, 2015;published OnlineFirst February 10, 2015.

References1. Koji Taniguchia MK. IL6 and related cytokines as the critical lynchpins

between inflammation and cancer. Semin Immunol 2014;26:54–74.2. He G, Dhar D, Nakagawa H, Font-Burgada J, Ogata H, Jiang Y, et al.

Identification of liver cancer progenitors whose malignant progressiondepends on autocrine IL6 signaling. Cell 2013;155:384–96.

3. Coward J KH,Chakravarty P, LeaderD, Vassileva V, LeinsterDA, ThompsonR, et al. Interleukin-6 as a therapeutic target in human ovarian cancer.Clin Cancer Res 2011;15:6083–96.

4. Kulbe H, Chakravarty P, Leinster DA, Charles KA, Kwong J, Thompson RG,et al. A dynamic inflammatory cytokine network in the human ovariancancer microenvironment. Cancer Res 2012;72:66–75.

5. Anglesio MS, George J, Kulbe H, Friedlander M, Rischin D, Lemech C, et al.IL6-STAT3-HIF signaling and therapeutic response to the angiogenesisinhibitor sunitinib in ovarian clear cell cancer. Clin Cancer Res 2011;17:2538–48.

6. Stone RL, Nick AM, McNeish IA, Balkwill F, Han HD, Bottsford-Miller J,et al. Paraneoplastic thrombocytosis in ovarian cancer. N Engl J Med2012;366:610–8.

7. Kulbe H, Thompson R, Wilson JL, Robinson S, Hagemann T, Fatah R, et al.The inflammatory cytokine tumor necrosis factor-alpha generates anautocrine tumor-promoting network in epithelial ovarian cancer cells.Cancer Res 2007;67:585–92.

8. Kulbe H, Hagemann T, Szlosarek PW, Balkwill FR, Wilson JL. The inflam-matory cytokine tumor necrosis factor-alpha regulates chemokine receptorexpression on ovarian cancer cells. Cancer Res 2005;65:10355–62.

9. Domcke S, Sinha R, Levine DA, Sander C, Schultz N. Evaluating cell lines astumour models by comparison of genomic profiles. Nat Commun 2013;4:2126.

10. Gassmann M, Grenacher B, Rohde B, Vogel J. Quantifying Western blots:pitfalls of densitometry. Electrophoresis 2009;30:1845–55.

Figure 6.Model for mechanism of action of IL6 and ErbB-1 signaling inhibition. In control conditions, EGFR and IL6 signaling pathways interact. STAT3 translocates tothe nucleus where it interacts with ErbB-1. When IL6 signaling is inhibited, there is a reduction in available STAT3 in the nucleus. Cells compensate forIL6 inhibition by activating EGFR family members. nErbB-1 levels are decreased because there is less STAT3. As a result, ErbB-1 expression becomes moreprominent at the plasma membrane and this is sustained when inhibition of IL6 is combined with ErbB-1 signaling inhibitors.






11. Benard J, Da Silva J, De Blois MC, Boyer P, Duvillard P, Chiric E, et al.Characterization of a human ovarian adenocarcinoma line, IGROV1, intissue culture and in nude mice. Cancer Res 1985;45:4970–9.

12. Kandala PK, Wright SE, Srivastava SK. Blocking epidermal growthfactor receptor activation by 3,30-diindolylmethane suppresses ovariantumor growth in vitro and in vivo. J Pharmacol Exp Ther 2012;341:24–32.

13. Colomiere M, Findlay J, Ackland L, Ahmed N. Epidermal growth factor-induced ovarian carcinoma cell migration is associated with JAK2/STAT3signals and changes in the abundance and localization of alpha6beta1integrin. Int J Biochem Cell Biol 2009;41:1034–45.

14. Sheng Q, Liu J. The therapeutic potential of targeting the EGFR family inepithelial ovarian cancer. Br J Cancer 2011;104:1241–5.

15. Erjala K, Sundvall M, Junttila TT, Zhang N, Savisalo M, Mali P, et al.Signaling via ErbB2 and ErbB3 associates with resistance and epidermalgrowth factor receptor (EGFR) amplification with sensitivity to EGFRinhibitor gefitinib in head and neck squamous cell carcinoma cells.Clin Cancer Res 2006;12:4103–11.

16. Wu K, Chang Q, Lu Y, Qiu P, Chen B, Thakur C, et al. Gefitinib resistanceresulted from STAT3-mediated Akt activation in lung cancer cells. Onco-target 2013;4:2430–8.

17. Kenny HA KT, Yamada SD, Lengyel E. Use of a novel 3D culture model toelucidate the role of mesothelial cells, fibroblasts and extra-cellular matri-ces on adhesion and invasion of ovarian cancer cells to the omentum. Int JCancer 2007;1:1463–72.

18. Xia W, Wei Y, Du Y, Liu J, Chang B, Yu YL, et al. Nuclear expression ofepidermal growth factor receptor is a novel prognostic value in patientswith ovarian cancer. Mol Carcinog 2009;48:610–7.

19. Wang YN, Lee HH, Lee HJ, Du Y, Yamaguchi H, Hung MC. Membrane-bound trafficking regulates nuclear transport of integral epidermalgrowth factor receptor (EGFR) and ErbB-2. J Biol Chem 2012;287:16869–79.

20. Li C, Iida M, Dunn EF, Ghia AJ, Wheeler DL. Nuclear EGFR contributes toacquired resistance to cetuximab. Oncogene 2009;28:3801–13.

21. LoHW, Ali-SeyedM,Wu Y, Bartholomeusz G, Hsu SC, HungMC. Nuclear-cytoplasmic transport of EGFR involves receptor endocytosis, importinbeta1 and CRM1. J Cell Biochem 2006;98:1570–83.

22. Lo HW, Hung MC. Nuclear EGFR signalling network in cancers: linkingEGFR pathway to cell cycle progression, nitric oxide pathway and patientsurvival. Br J Cancer 2006;94:184–8.

23. Wei-Chien Huanga Y-JC, Mien-Chie Hunga. Implication of nuclear EGFRin the development of resistance to anticancer therapies. BioMedicine2011;1:2–10.

24. Brand TM, Iida M, Li C, Wheeler DL. The nuclear epidermal growth factorreceptor signaling network and its role in cancer. Discov Med 2011;12:419–32.

25. Lo HW, Hung MC. Nuclear EGFR signalling network in cancers: linkingEGFR pathway to cell cycle progression, nitric oxide pathway and patientsurvival. Br J Cancer 2007;96:R16–20.

26. Yu H, Jove R. The STATs of cancer–newmolecular targets come of age. NatRev Cancer 2004;4:97–105.

27. Scuto A, Kujawski M, Kowolik C, Krymskaya L, Wang L, Weiss LM, et al.STAT3 inhibition is a therapeutic strategy for ABC-like diffuse large B-celllymphoma. Cancer Res 2011;71:3182–8.

28. Phesse TJ, Buchert M, Stuart E, Flanagan DJ, Faux M, Afshar-Sterle S, et al.Partial inhibition of gp130-Jak-Stat3 signaling prevents Wnt-beta-catenin-mediated intestinal tumor growth and regeneration. Sci Signal 2014;7:ra92.

29. Alberti C, Pinciroli P, Valeri B, Ferri R, Ditto A, Umezawa K, et al. Ligand-dependent EGFR activation induces the co-expression of IL6 and PAI-1 viathe NFkB pathway in advanced-stage epithelial ovarian cancer. Oncogene2012;31:4139–49.

30. Gao SP, Mark KG, Leslie K, Pao W, Motoi N, Gerald WL, et al.Mutations in the EGFR kinase domain mediate STAT3 activation viaIL6 production in human lung adenocarcinomas. J Clin Invest 2007;117:3846–56.

31. Aaronson DS, Muller M, Neves SR, Chung WC, Jayaram G, Iyengar R, et al.An androgen-IL6-Stat3 autocrine loop re-routes EGF signal in prostatecancer cells. Mol Cell Endocrinol 2007;270:50–6.

32. Wang Y, van Boxel-Dezaire AH, Cheon H, Yang J, Stark GR. STAT3activation in response to IL6 is prolonged by the binding ofIL6 receptor to EGF receptor. Proc Natl Acad Sci U S A 2013;110:16975–80.

33. Domcke S, Sinha R, LevineDA, Sander C, Schultz N. Evaluating cell lines astumour models by comparison of genomic profiles. Nature Commun2013;4:2126.

34. Bashashati A, Ha G, Tone A, Ding J, Prentice LM, Roth A, et al. Distinctevolutionary trajectories of primary high-grade serous ovarian cancersrevealed through spatial mutational profiling. J Pathol 2013;231:21–34.

35. Vaughan S, Coward JI, Bast RC Jr, Berchuck A, Berek JS, Brenton JD, et al.Rethinking ovarian cancer: recommendations for improving outcomes.Nat Rev Cancer 2011;11:719–25.

36. Berns EM, Bowtell DD. The changing view of high-grade serous ovariancancer. Cancer Res 2012;72:2701–4.

37. Perets R, Wyant GA, Muto KW, Bijron JG, Poole BB, Chin KT, et al.Transformation of the fallopian tube secretory epithelium leads to high-grade serous ovarian cancer in Brca;Tp53;Pten models. Cancer Cell2013;24:751–65.

38. Murphy M, Stordal B. Erlotinib or gefitinib for the treatment of relapsedplatinum pretreated non-small cell lung cancer and ovarian cancer: asystematic review. Drug Resist Updat 2011;14:177–90.


Milagre et al.




Published OnlineFirst February 10, 2015.Cancer Res Carla S. Milagre, Ganga Gopinathan, Gemma Everitt, et al. Combined Therapy in Ovarian CancerGrowth by Neutralizing IL6 Antibodies, with Implications for Adaptive Upregulation of EGFR Limits Attenuation of Tumor

Updated version

10.1158/0008-5472.CAN-14-1801doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2015/02/10/0008-5472.CAN-14-1801.DC1

Access the most recent supplemental material at:

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

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://cancerres.aacrjournals.org/content/early/2015/03/20/0008-5472.CAN-14-1801To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-14-1801

http://cancerres.aacrjournals.org/content/suppl/2015/02/10/0008-5472.CAN-14-1801.DC1

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/early/2015/03/20/0008-5472.CAN-14-1801


adaptive upregulation of egfr limits attenuation of tumor growth

Documents