Inhibition of triple-negative breast cancer models bycombinations of antibodies to EGFRDaniela A. Ferraroa, Nadège Gaborita, Ruth Maronb, Hadas Cohen-Dvashia, Ziv Poratc, Fresia Parejaa, Sara Lavia,Moshit Lindzena, Nir Ben-Chetrita, Michael Selab,1, and Yosef Yardena,1

Departments of aBiological Regulation, bImmunology, and cBiological Services, Weizmann Institute of Science, Rehovot 76100, Israel

Contributed by Michael Sela, December 10, 2012 (sent for review August 23, 2012)

Breast tumors lacking expression of human epidermal growthfactor receptor 2 (HER2) and the estrogen and the progesteronereceptors (triple negative; TNBC) are more aggressive than otherdisease subtypes, and no molecular targeted agents are currentlyavailable for their treatment. Because TNBC commonly displaysEGF receptor (EGFR) expression, and combinations of monoclonalantibodies to EGFR effectively inhibit other tumor models, weaddressed the relevance of this strategy to treatment of TNBC.Unlike a combination of the clinically approved monoclonal anti-bodies, cetuximab and panitumumab, which displaced each otherand displayed no cooperative effects, several other combinationsresulted in enhanced inhibition of TNBC’s cell growth both in vitroand in animals. The ability of certain antibody mixtures to removeEGFR from the cell surface and to promote its intracellular degra-dation correlated with the inhibitory potential. However, unlikeEGF-induced sorting of EGFR to lysosomal degradation, the anti-body-induced pathway displayed independence from the intrinsickinase activity and dimer formation ability of EGFR, and it largelyavoided the recycling route. In conclusion, although TNBC clinicaltrials testing EGFR inhibitors reported lack of benefit, our resultsoffer an alternative strategy that combines noncompetitive anti-bodies to achieve robust degradation of EGFR and tumor inhibition.

cancer therapy | signal transduction

Growth factors and their transmembrane receptor tyrosinekinases play critical roles in tumor progression (1). One re-

markable example entails a large family of growth factors, allsharing an EGF motif, and their respective receptor tyrosinekinases of the EGFR family. Consistent with essential roles intumor progression, strategies able to interfere with ERBB func-tions, such as mAbs and tyrosine kinase inhibitors (TKIs), haveyielded in the past decade several oncology drugs (2). For example,two genetically engineered mAbs to EGF receptor (EGFR),cetuximab and panitumumab, are approved for treatment of co-lorectal cancer. Unlike TKIs’ well-understood mode of action, themechanisms underlying therapeutic activities of mAbs are lessunderstood. In general, potential mechanisms can be divided intoimmune-mediated cell killing, such as antibody-dependent cellularcytotoxicity, and diverse neutralizing effects, such as inhibition ofligand binding, prevention of receptor dimerization, and inductionof receptor internalization (3). Early animal studies that testeda set of mAbs to the rodent form of ERBB2/human epidermalgrowth factor receptor 2 (HER2) indicated that individual mAbscause partial tumor eradication, whereas the administration ofcertain mixtures of antibodies resulted in synergistic effects (4).Similar effects on the human HER2 protein were later confirmed(5, 6). In vitro, the more effective mAb mixture was also moreeffective than the single mAbs in inducing receptor degradation(6) and antibody-dependent cellular cytotoxicity (5). Synergisticantitumor effects were confirmed and associated with receptordegradation using another set of mAbs to HER2 (7). Importantly,a mixture of two mAbs to HER2, trastuzumab and pertuzumab,in combination with chemotherapy, significantly prolonged pro-gression-free survival of breast cancer patients whose tumors over-express HER2 (8). Similar to anti-HER2 mAbs, cetuximab inducesdown-regulation of EGFR, and this effect appears important forgrowth inhibition (9). Experiments that used a radiolabeled cetux-imab confirmed endocytosis of the mAb, but the internalized mAb

recycled more effectively than internalized EGF (10). In similarityto the synergistic internalizing effects of combinations of HER2-directedmAbs, we noted that certain pairs of anti-EGFR antibodiescan accelerate receptor endocytosis and degradation (11). To en-hance endocytosis, themAbsmust engage nonoverlapping antigenicepitopes of EGFR. Another study showed that highly potent mAbcombinations reduced surface receptor levels through a mechanismconsistent with mAb-mediated inhibition of EGFR recycling (12).The ability of certain mAb mixtures to enhance EGFR deg-

radation raised the possibility that such a strategy would inhibitEGFR-driven tumors, including the most aggressive fraction ofbreast cancer, which is defined by absence of estrogen receptor,progesterone receptor, and HER2 (13). A subclass of triple neg-ative breast cancer (TNBC) overexpresses EGFR (14), and ex-haustive gene expression profiling identified several EGFR-associated poor prognostic signatures (15). Contrary to otherbreast cancer subtypes, for which therapy targeting biologicaldrivers proved to be successful, no molecular targeted agents areapproved for TNBC. Importantly, kinase inhibitors and anti-EGFR therapy using a single mAb did not improve outcome ofTNBC (16). For these reasons, it is imperative to develop newstrategies able to control TNBC and delay the onset of patientresistance to chemotherapy. Here we characterize the mecha-nism of mAb-induced EGFR internalization and demonstratethat a cooperative mAb mixture can inhibit growth of TNBCin animals.

ResultsCertain Combinations of Anti-EGFR mAbs Enhance Receptor Down-Regulation. Certain combinations of epitope-distinct anti-HER2/ERBB2 antibodies induce receptor endocytosis and synergisticallyinhibit growth of HER2-overexpressing tumor cells (7). To identifysimilar pairs of anti-EGFR mAbs, we incubated MDA-MB-468TNBC cells with a fluorescent derivative of EGF in the presence ofanti-EGFR mAbs 111, 565, panitumumab, and cetuximab. Threeof the four antibodies effectively displaced EGF, but mAb565behaved as a relatively weak competitor (Fig. 1A). Next, we radi-olabeled panitumumab and found that the two clinically approvedantibodies share antigenic epitopes and therefore they may notsynergize; however, mAb565 engages a distinct epitope of EGFR(Fig. S1 A and B). Similar analyses revealed that mAb111 bindsEGFR independently from the binding of cetuximab and pan-itumumab (Fig. 1 B and C), and according to our previous report,mAbs 111 and 565 are noncompetitive (11). These lines of evi-dence excluded the panitumumab plus cetuximab pair and haveidentified several combinations (e.g., 111/panitumumab and 111/cetuximab) as potential collaborative pairs. A standard EGFRdown-regulation assay, performed with HeLa cells, validated thesepredictions (Fig. 1D): long-term exposure to the noncompetitivemAb combinations achieved receptor down-regulation. This was

Author contributions: D.A.F., M.S., and Y.Y. designed research; D.A.F., N.G., R.M., H.C.-D.,Z.P., F.P., S.L., M.L., and N.B.-C. performed research; D.A.F., R.M., M.S., and Y.Y. analyzeddata; and D.A.F., N.G., R.M., M.S., and Y.Y. wrote the paper.

The authors declare no conflict of interest.1To whom correspondence may be addressed. E-mail: michael.sela@weizmann.ac.il oryosef.yarden@weizmann.ac.il.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1220763110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1220763110 PNAS | January 29, 2013 | vol. 110 | no. 5 | 1815–1820

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

June

14,

202

0

also confirmed by immunoblot analyses showing that receptordown-regulation closely reflected disappearance of the respectiveprotein band (Fig. 1E). Interestingly, mAb565 induced more ex-tensive down-regulation of EGFR, an activity we attribute to afaint ability to stimulate auto-phosphorylation (Fig. 1E). Similarly,exposure ofHeLa cells to eithermAb111 or panitumumab inducedweak auto-phosphorylation, but their combination strongly in-duced both phosphorylation and down-regulation (Fig. 1F). Inconclusion, the ability of mAbs to down-regulate EGFR correlateswith the engagement of nonoverlapping sites.

When Combined, Noncompetitive mAbs Induce EGFR Ubiquitination.Because EGF-induced ubiquitination of EGFR by c-CBL E3 ligaseis tightly associated with receptor degradation (17), we tested re-ceptor ubiquitination after treatment with the panitumumab plusmAb111 pair. Unlike single mAbs, the combined treatment in-creased EGFR ubiquitination (Fig. 2A). As expected, EGF-inducedubiquitination appeared earlier and, interestingly, yielded a mixtureof higher molecular weight species. Another difference emergedfrom a coimmunoprecipitation experiment: Unlike the well-char-acterized EGF-induced formation of EGFR-CBL complexes, nei-ther a single mAb nor a combination resulted in detectablecomplexes (Fig. 2B). Nevertheless, both EGF- and mAb-inducedEGFR degradation were inhibited by a blocker of lysosomalhydrolases (bafilomycin), but an antagonist of the 26S protea-some (bortezomib) was ineffective (Fig. 2C). Conceivably, mAbmixtures accelerate lysosomal degradation by inducing relativelyslow ubiquitination of EGFR and recruiting an E3 ligase distinctfrom c-CBL.The wealth of information on EGF-induced receptor degra-

dation contrasts with the paucity of data on mAb-induced en-docytosis. Hence, we screened a library of siRNAs arranged in180 pools, each targeting a single endocytic mediator (Materialsand Methods). HeLa cells were transfected with individual pools,along with control siRNAs. Thereafter, cells were treated with amAb mixture and incubated for 12 h before analysis of EGFRdegradation. A validation test and a list of positive hits (in two

cycles) are depicted in Fig. S2. Remarkably, two proteins thatplay critical roles in clathrin-mediated endocytosis, Huntingtininteracting protein 1 (HIP1) and dynamin 2, a GTPase, scoredpositively. In addition to the dynamin’s partner, amphiphysin,and the ubiquitin binder EPS15L1, the screen identified severalproteins involved in late stages of the endocytic pathway, such asRAB7L1 (18) and TSG101. These observations are consistentwith a model assuming that mAb combinations divert EGFRfrom the recycling route to late endosomes and lysosomes (12).

Antibody-Induced Endocytosis of EGFR Avoids the RecyclingCompartment. To examine a scenario of mAb-mediated inhibi-tion of recycling, we analyzed subcellular distribution of EGFR.As predicted, treatment with a mAb mixture translocated EGFRof HeLa cells into intracellular vesicles (Fig. S3). Although EGFinduced a similar translocation, treatment with single mAbs wasineffective. These differences were quantified using flow cytom-etry (Fig. 3 A and B): incubation with either EGF or two mAbsincreased the intracellular fraction. To directly follow the recy-cling route, we used an antibody to RAB11, a GTPase of recyclingendosomes (18). As expected, in untreated cells, EGFR andRAB11 displayed nonoverlapping patterns, but EGF translocatedthe receptor to the perinuclear recycling compartment (Fig. 3C).Importantly, although the mAb mixture extensively translocatedEGFR into intracellular puncta, these structures were devoid ofRAB11, which is in line with the possibility that mAb mixturessort EGFR for degradation while avoiding the recycling route.

mAb-Induced Degradation of EGFR Is Independent of the KinaseActivity and the Intrinsic Dimerization Ability of EGFR. Because oftheir ability to force EGFR dimerization, antibodies might act aspartial agonists. To examine this, we probed extracts of stimulatedHeLa cells. As expected, EGF enhanced phosphorylation of EGFRas well as phosphorylation of two effectors, AKT andERK (Fig. 4A).Remarkably, although the agonist antibody 565 and mixtures ofnoncompetitive mAbs enhanced EGFR auto-phosphorylation aswell as accelerated receptor degradation, this did not translate to

A B

D C

E

F

Fig. 1. Specific combinations of anti-EGFR mAbs enhance receptor down-regulation and degradation. (A) MDA-MB-468 cells were incubated for 45 min at4 °C with increasing concentrations of growth factors or mAbs. Thereafter, Texas Red-EGF (300 nM) was added for an additional 45 min. Fluorescence in-tensity was determined after washing using a microplate reader. The results represent averages (± SD) from three experiments. (B and C) Serum-starved MDA-MB-468 cells were incubated at 4 °C with the indicated mAbs (or with EGF). Cetuximab (B; ctx) or panitumumab (C; pan), each at 1 μg/mL, was added 15 minlater and incubated for 45 min. The cells were then washed and incubated for 60 min at 4 °C with HRP-labeled anti human IgG. Light absorbance (415 nm) wasdetermined using an ELISA microplate reader after washing and incubating for 15 min with 2,2′-Azino-bis (3-ethylbenzothiazoline-6-sulfonic acid). (D) HeLacells were incubated for 12 h with the indicated mAbs (20 μg/mL) or for 60 min with EGF (10 ng/mL). Bound ligands were acid-stripped, and surface EGFR wasmeasured by FACS analysis. Values are the average ± SD of triplicates. (E) HeLa cells were treated for 12 h with the indicated anti-EGFR mAbs (20 μg/mL) or for60 min with EGF (10 ng/mL). Afterward, whole cell extracts were subjected to immunoblotting (IB) with the indicated antibodies. (F) HeLa cells were in-cubated for the indicated intervals with mAbs (20 μg/mL), or with EGF (10 ng/mL), and lysates were probed with the indicated antibodies.

1816 | www.pnas.org/cgi/doi/10.1073/pnas.1220763110 Ferraro et al.

Dow

nloa

ded

by g

uest

on

June

14,

202

0

ERK or AKT activation. Hence, we assumed that mAb-mediatedand EGF-induced dimerization differ. Next, we used a previouslydescribed EGFR mutant that is defective in both homodimerformation and EGF-induced internalization (19). Using this de-letion mutant (ΔCR1), we confirmed that dimerization is essentialfor both EGF-induced auto-phosphorylation and degradationof EGFR (Fig. 4B). Strikingly, neither mAb-mediated auto-phosphorylation nor mAb-mediated degradation of EGFR dis-played dependence on the intrinsic dimer-forming ability ofEGFR, implying that mAb-induced clustering preempts the in-trinsic dimer-forming feature of EGFR.The EGFR-specific TKI, AG1478, was used in an effort to

clarify the contribution of the intrinsic kinase activity to mAb-induced endocytosis. Although treatment with AG1478 completelyblocked both EGF and mAb-induced auto-phosphorylation, onlythe ligand-induced degradation of EGFR was inhibited (Fig. 4C).Conceivably, EGFR is passively recruited by mAbs: neither kinase

activity nor dimer-forming competence is needed, but all threefunctions are essential for EGF-induced degradation of EGFR.

EGFR Down-Regulation Impairs the Ability of TNBC Cells to Migrate.Node-positive TNBC is associated with higher EGFR expressioncompared with node-negative lesions (20), suggesting that EGFRdrives tumor progression. As a prelude to testing the applicabilityof mAb-mediated down-regulation, we surveyed EGFR status ina series of TNBC cell lines (Fig. S4A). This analysis confirmedEGFR overexpression in TNBC cell lines; lines that belong toother subtypes, such as HER2+ (SKBR3 and T47D) and the lu-minal type (MCF7), displayed lower EGFR levels. Next, wetransfected BT-549 TNBC cells with EGFR-specific siRNAs (Fig.S4B) and found that loss of >80% of the receptor associatedwith 60% or 80% reduction in the ability of cells to migrate orinvade through an extracellular barrier, respectively (Fig. S4 CandD). Next, we asked whether a similar effect would accompany

A

B

C

Fig. 2. Anti-EGFR antibodies enhance receptor ubiquitination and degradation.(A) Serum-starved HeLa cells were incubated with mAbs (10 μg/mL), a combina-tion (each at 5 μg/mL), or with EGF (10 ng/mL), and lysates analyzed using im-munoprecipitation (IP) and immunoblotting (IB). (B) HeLa cells transfected witha plasmid encoding an MYC peptide-tagged CBL, or an empty vector, were se-rum-starved and treated for 3 h with the indicated mAbs (20 μg/mL) or a com-bination (each at 10 μg/mL). Alternatively, cells were treated for 10 min with EGF(10 ng/mL). Lysates were probed as indicated. (C) HeLa cells that were pre-incubated (12 h) with bortezomib (2 μM) or bafilomycin (10 nM) were incubated(60 min) with EGF (10 ng/mL) or for 6 h with the indicated combination of mAbs(each at 10 μg/mL). Lysates were subjected to IB and signal quantification.

A

B

C

Fig. 3. A combination of anti-EGFR antibodies enhances receptor in-ternalization and impairs recycling. (A and B) Serum-starved HeLa cells wereincubated with EGF (10 ng/mL; 30 min) or mAbs (20 μg/mL total; 4 h).Thereafter, cells were permeabilized and EGFR was localized using flowcytometry. EGFR distribution between the plasma membrane (scored as 1)and the cell’s center (scored as 0) was assessed. (C) HeLa cells were treatedwith EGF (10 ng/mL; 30 min) or mAbs (10 μg/mL total; 4 h). Following fixation,permeabilization, and incubation with anti-EGFR and anti-Rab11 antibodies,cells were incubated with fluorescent secondary antibodies, and imageswere acquired with a confocal fluorescent microscope. Colocalization of twomarkers appears as yellow (Merge).

Ferraro et al. PNAS | January 29, 2013 | vol. 110 | no. 5 | 1817

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

June

14,

202

0

mAb-induced endocytosis. Using a set of TNBC cell lines andmAb mixtures, we observed EGFR down-regulation (Fig. 5A),similar to the effects on HeLa cells (Fig. 1D). When singly ap-plied, mAbs other than the partial agonist, 565, were unable todown-regulate EGFR, but mixtures of noncompetitive mAbscaused EGFR degradation in all tested TNBC lines. Notably,EGF was unable to down-regulate EGFR of MDA-MB468 cells,probably because of saturation of the endocytic pathway by thehighly expressed receptor (21). Nevertheless, reproducible down-regulation was achieved by mAbs, in line with distinct routes ofendocytosis engaged by mAbs and EGF.Next, we tested the prediction that mAb combinations would

inhibit migration of TNBC cells. For this, we used BT-549 cellsand a mAb combination that effectively down-regulated EGFR(Fig. 5B). Under these conditions, we observed almost 50% re-duction in cell migration, but neither mAb, when singly applied,was effective (Fig. 5 C and D). These findings are consistent withthe observations we made using siRNAs specific to EGFR (Fig.

S4). In conclusion, EGFR appears to play an important role inthe migratory behavior of the TNBC model we used, such thatmAb-induced endocytosis and degradation of EGFR can sig-nificantly limit cell migration in vitro.

Mixtures of mAbs Arrest TNBC Cells at G1 and Inhibit Their TumorigenicGrowth in Animals. Prolonged exposure of HCC70 cells to cetux-imab, or to cetuximab plus mAb565, exerted only minor effects oncell survival, as reflected by the unchanged sub-G1 fraction (Fig.S5A). In contrast, exposure to an apoptosis-inducing drug, puro-mycin, increased sub-G1 and erased the S-phase fraction, con-sistent with up-regulation of the cleaved form of caspase-3 (Fig.S5B). Although the mAbs induced no detectable apoptosis, theyclearly arrested cells at G1 (Fig. S5A). To measure proliferation,HCC70 cells were incubated for 5 d with another pair of non-competitive mAbs (Fig. 6A). This prolonged incubation tookplace in serum-free medium that was supplemented, or not, withEGF. Although EGF strongly promoted proliferation of HCC70cells, this effect was almost completely inhibited by the combi-nation of mAbs. Notably, each antibody exerted only a small in-hibitory effect when singly applied, in line with the proposed effectof antibody-induced endocytosis of EGFR.Animal studies using HCC70 cells examined the relevance of

our observations to tumor growth. Cells were injected intrader-mally; once tumors became palpable, we injected into the peri-toneum either single mAbs or mAb combinations. Fig. 6 B and Cpresents the results: differences emerged following 4 wk of tumorgrowth—each mAb induced a partial inhibitory effect, but thecombinations more effectively regressed tumors (see also Fig. S6).Importantly, the inhibitory effects progressively increased, suchthat the cetuximab plus mAb111 combination reached statisticalsignificance (P < 0.01) at week 8, and the other combinationclearly showed a similar trend at the time of trial termination.In summary, although TNBC clinical trials using EGFR inhib-

itors, including cetuximab, reported lack of clinical benefit (22), ourresults offer an alternative strategy. This strategy combines non-competitive mAbs to achieve robust EGFR degradation. Similar toEGF, noncompetitive mAbs target EGFR to degradation in lyso-somes and engage ubiquitination and the clathrin-mediated route.Nevertheless, mAb-induced degradation is unique and identifiesoligoclonal mixtures as a viable alternative to the singly usedtherapeutic mAbs: this relatively slow process is independent of c-CBL and the intrinsic kinase activity, or dimer-forming ability, ofEGFR. As a result, mAb mixtures inhibit motility of TNBC cells aswell as arrest them at G1, attributes that translate to effective in-hibition of tumor growth in an animal model.

DiscussionSeveral lines of evidence support the possibility that EGFR playsa driver role in a large fraction of TNBC. For example,EGFR geneamplification is commonly identified in metaplastic breast carci-noma, a basal-like fraction of tumors (14). Likewise, gene ex-pression signatures correlated TNBC with modules comprisingEGF-like ligands, EGFR, and several downstream effectors (15).In an effort to examine the proposition that agents interceptingEGFR bear therapeutic potential for TNBC, we examined mix-tures of anti-EGFR mAbs. Because their antigenic epitopes areoverlapping, combining cetuximab and panitumumab, the EGF-competitive anti-EGFR monoclonals routinely used to treat co-lorectal cancer patients, did not improve receptor degradation.Importantly, the percentage of patients responding to these anti-bodies is low and many patients experiencing an initial responseeventually relapse. Whether or not mAbs selected on the basis oftheir ability to accelerate EGFR degradation will increase re-sponse rate or delay the onset of patient resistance remains anopen issue. An answer to this question might be provided bycurrent clinical trials applying Sym004, a mixture of two non-competitive anti-EGFR antibodies, on both squamous cell carci-noma of the head and neck and metastatic colorectal cancer (23).Additional combinations of anti-EGFR antibodies can ac-

celerate receptor degradation, and the underlying mechanismappears related to the sorting of internalized EGFRs to eitherrecycling or degradation (12). In the case of EGF-stimulated

A

B

C

Fig. 4. A combination of anti-EGFR mAbs down-regulates a dimerization-defective mutant of EGFR but cannot trigger downstream signaling. (A)Serum-starved HeLa cells were incubated with EGF (10 ng/mL, 1 h) or withanti-EGFR mAbs (20 μg/mL total, 12 h) and lysates probed as indicated. (B)HeLa cells were transfected with a plasmid encoding a dimerization-defective mutant of EGFR (ΔCR-EGFR-YFP). Forty-eight hours later, serum-starved cells were treated with EGF (10 ng/mL) or with mAbs (20 μg/mL total)before lysate immunoblotting. (C) HeLa cells were pretreated for 4 h witha selective EGFR kinase inhibitor (AG1478; 10 μM). Subsequently, cells werewashed and treated with EGF (10 ng/mL), or with a combination of mAbs(each at 10 μg/mL). Lysates were probed with the indicated antibodies.

1818 | www.pnas.org/cgi/doi/10.1073/pnas.1220763110 Ferraro et al.

Dow

nloa

ded

by g

uest

on

June

14,

202

0

EGFRs, sorting requires conjugation of multiple ubiquitins,which mark the receptor for degradation (24). By applyinga mixture of mAbs, we detected relatively slow EGFR ubiquiti-nation and degradation (Fig. 2). Despite similarities, the mech-anisms underlying sorting of EGFR by mAb mixtures and byEGF remarkably differ: EGF robustly increases receptor phos-phorylation, which is necessary for recruitment of an E3 ubiq-uitin ligase of the CBL family (17), but antibody mixtures behaveas very weak agonists of auto-phosphorylation, and we could notdetect recruitment of c-CBL. Importantly, mAb-induced re-ceptor ubiquitination is associated with avoidance of the recy-cling route (Fig. 3D); consistent with this observation, siRNAscreens implicated two components of late endosomes, RAB7and TSG101, along with early mediators such as clathrin anddynamin, in mAb-induced receptor sorting. Although this en-semble of endocytic players outlines a clathrin-mediated route, itis interesting noting that knocking down HIP1, an endocyticactin-binding protein, severely impaired antibody-induced deg-radation of EGFR (Fig. S2).Conceivably, the robust antibody-induced down-regulation of

EGFR employs more than one intracellular route to target surfaceEGFRs to lysosomes, such as actin-dependent micropinocytosisor a clathrin-independent route (25). Our subsequent experi-ments, which used several TNBC lines, indicated that down-

regulation of EGFR can retard motility, signaling, and pro-liferation of different TNBC cell lines. These observations weresupported by animal studies demonstrating that antibody mix-tures can reduce the tumorigenic growth of TNBC cells. Thus,regardless of the exact mechanisms underlying the cooperativeeffect of antibody mixtures, the studies we presented offer atherapeutic scenario to treat a rather heterogeneous and ther-apy-resistant type of breast cancer (13).

Materials and MethodsCell Lines and Transfections. Cell lines were obtained from the American TypeCulture Collection and grown in Roswell ParkMemorial InstitutionmediumorDMEM, supplemented with FCS [10% (vol/vol)] and 1 mM sodium pyruvate.For transient plasmid and siRNA transfections, we used JetPEI (Polyplus) andeither Dharmafect from Dharmacon or HiPerfect from Qiagen, respectively.

EGF Displacement Assays Using ELISA. Cells were plated in 96-well plates andthe next day they were incubated for 45 min at 4 °C with increasing con-centrations of mAbs in Krebs buffer. Afterward, EGF-Texas Red was addedfor an additional 45 min. After washing, we determined fluorescent signalsusing a microplate reader.

Invasion and Migration Assays. BT549 cells (10 × 105/well) were seeded in theupper compartment of Transwell (Corning) or Matrigel-coated invasion

A

B

C

DFig. 5. A combination of antibodies down-regulatesEGFR and inhibits invasion of TNBC cells. (A) Wholeextracts were prepared from the indicated cell linesafter treatment with EGF (10 ng/mL, 1 h) or with themAbs (20 μg/mL total, 6 h). Lysates were immuno-blotted as indicated. (B) BT-549 cells were treated for48 h with mAbs (20 μg/mL total) and lysates immu-noblotted for EGFR and ERK2. (C and D) BT-549 cellswere treated with mAbs as in (B). Thereafter, cellswere plated in the upper compartment of invasionchambers. The lower compartments were filled withthe respective mAb-containing media. Eighteenhours later, the filters were removed, fixed, per-meabilized, and stained with methyl violet (0.3%).Cells growing on the upper side of the filter wereremoved and cells on the bottom side were photo-graphed and quantified.

Ferraro et al. PNAS | January 29, 2013 | vol. 110 | no. 5 | 1819

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

June

14,

202

0

chambers. The lower compartment was filled with serum-containing me-dium. Eighteen hours later, cells on the lower side of the filter were fixed,permeabilized with Triton X-100 (0.1%), and stained with crystal violet.Images were quantified by using ImageJ.

Tumorigenic Growth in Animals. All animal studies were approved by theWeizmann Institute’s Review Board (IRB). CD1/nude mice were divided ingroups of five mice and injected intradermally with HCC70 cells (7 × 106 permouse). mAbs were injected intraperitoneally at 160 μg per mouse perinjection on days 7, 14, 21, 28, and 35 after grafting. Tumor volume wasevaluated once per week.

ImageStream Analysis. Following treatments, cells were trypsinized, fixed,permeabilized, and stained with an anti-EGFR antibody and a secondary anti-rabbit antibody. A total of 10,000 events from each sample were collectedusing the ImageStreamX instrument (Amnis Corp.) and IDEAS 4.0 software.Single cells were gated using the area and aspect ratio features as well as theGradient RMS feature (26). Cells were then gated to select only positively

stained cells using their pixel intensity values. The localization of EGFR wascalculated using the Max Contour Position feature.

Statistical Analysis. Two-way ANOVA and Bonferroni multiple comparisontests were used to analyze differences between groups. P values < 0.01 wereconsidered significant.

ACKNOWLEDGMENTS. We thank Dr. Siena (University of Milan) for panitu-mumab and cetuximab, Dr. Wang (University of Alberta) for the ΔCR1-EGFRplasmid, and Mrs. Abramovitch-Elhanati for initial analyses. We also thankDr. Bilha Schechter for her insightful help in analyzing and interpreting theexperimental data. Our work is supported by Grant CA072981 from the USNational Cancer Institute, and by the European Research Council, the SeventhFramework Programme of the European Commission, the German-Israeli Pro-ject Cooperation, the Dr. Miriam and Sheldon G. Adelson Medical ResearchFoundation, the M. D. Moross Cancer Institute, the Julius Baer Trust, anda Dukler Mudy Grant. Y.Y. is a research professor of the Israel Cancer Re-search fund and the incumbent of the Harold and Zelda Goldenberg Pro-fessorial Chair. M.S. is the incumbent of the W. Garfield Weston Chair.

1. Witsch E, Sela M, Yarden Y (2010) Roles for growth factors in cancer progression.Physiology (Bethesda) 25(2):85–101.

2. Ciardiello F, Tortora G (2008) EGFR antagonists in cancer treatment. N Engl J Med358(11):1160–1174.

3. Ben-Kasus T, Schechter B, Sela M, Yarden Y (2007) Cancer therapeutic antibodiescome of age: Targeting minimal residual disease. Mol Oncol 1(1):42–54.

4. Drebin JA, Link VC, Greene MI (1988) Monoclonal antibodies reactive with distinctdomains of the neu oncogene-encoded p185 molecule exert synergistic anti-tumoreffects in vivo. Oncogene 2(3):273–277.

5. Spiridon CI, et al. (2002) Targeting multiple Her-2 epitopes with monoclonal anti-bodies results in improved antigrowth activity of a human breast cancer cell line invitro and in vivo. Clin Cancer Res 8(6):1720–1730.

6. Kasprzyk PG, Song SU, Di Fiore PP, King CR (1992) Therapy of an animal model ofhuman gastric cancer using a combination of anti-erbB-2 monoclonal antibodies.Cancer Res 52(10):2771–2776.

7. Ben-Kasus T, Schechter B, Lavi S, Yarden Y, Sela M (2009) Persistent elimination ofErbB-2/HER2-overexpressing tumors using combinations of monoclonal antibodies:Relevance of receptor endocytosis. Proc Natl Acad Sci USA 106(9):3294–3299.

8. Baselga J, et al.; CLEOPATRA Study Group (2012) Pertuzumab plus trastuzumab plusdocetaxel for metastatic breast cancer. N Engl J Med 366(2):109–119.

9. Fan Z, Masui H, Altas I, Mendelsohn J (1993) Blockade of epidermal growth factorreceptor function by bivalent and monovalent fragments of 225 anti-epidermalgrowth factor receptor monoclonal antibodies. Cancer Res 53(18):4322–4328.

10. Jaramillo ML, et al. (2006) Effect of the anti-receptor ligand-blocking 225 monoclonalantibody on EGF receptor endocytosis and sorting. Exp Cell Res 312(15):2778–2790.

11. Friedman LM, et al. (2005) Synergistic down-regulation of receptor tyrosine kinases bycombinations of mAbs: Implications for cancer immunotherapy. Proc Natl Acad SciUSA 102(6):1915–1920.

12. Spangler JB, et al. (2010) Combination antibody treatment down-regulates epidermalgrowth factor receptor by inhibiting endosomal recycling. Proc Natl Acad Sci USA107(30):13252–13257.

13. Foulkes WD, Smith IE, Reis-Filho JS (2010) Triple-negative breast cancer. N Engl J Med363(20):1938–1948.

14. Reis-Filho JS, et al. (2006) EGFR amplification and lack of activating mutations inmetaplastic breast carcinomas. J Pathol 209(4):445–453.

15. Hoadley KA, et al. (2007) EGFR associated expression profiles vary with breast tumorsubtype. BMC Genomics 8:258.

16. Nogi H, et al. (2009) EGFR as paradoxical predictor of chemosensitivity and outcomeamong triple-negative breast cancer. Oncol Rep 21(2):413–417.

17. Levkowitz G, et al. (1999) Ubiquitin ligase activity and tyrosine phosphorylation un-derlie suppression of growth factor signaling by c-Cbl/Sli-1. Mol Cell 4(6):1029–1040.

18. Zerial M, McBride H (2001) Rab proteins as membrane organizers. Nat Rev Mol CellBiol 2(2):107–117.

19. Wang Q, Villeneuve G, Wang Z (2005) Control of epidermal growth factor receptorendocytosis by receptor dimerization, rather than receptor kinase activation. EMBORep 6(10):942–948.

20. Sutton LM, et al. (2010) Intratumoral expression level of epidermal growth factorreceptor and cytokeratin 5/6 is significantly associated with nodal and distant me-tastases in patients with basal-like triple-negative breast carcinoma. Am J Clin Pathol134(5):782–787.

21. Wiley HS (1988) Anomalous binding of epidermal growth factor to A431 cells is dueto the effect of high receptor densities and a saturable endocytic system. J Cell Biol107(2):801–810.

22. Burness ML, Grushko TA, Olopade OI (2010) Epidermal growth factor receptor intriple-negative and basal-like breast cancer: Promising clinical target or onlya marker? Cancer J 16(1):23–32.

23. Koefoed K, et al. (2011) Rational identification of an optimal antibody mixture fortargeting the epidermal growth factor receptor. MAbs 3(6):584–595.

24. Goh LK, Huang F, Kim W, Gygi S, Sorkin A (2010) Multiple mechanisms collectivelyregulate clathrin-mediated endocytosis of the epidermal growth factor receptor.J Cell Biol 189(5):871–883.

25. Sigismund S, et al. (2005) Clathrin-independent endocytosis of ubiquitinated cargos.Proc Natl Acad Sci USA 102(8):2760–2765.

26. George TC, et al. (2006) Quantitative measurement of nuclear translocation eventsusing similarity analysis of multispectral cellular images obtained in flow. JImmunol Methods 311(1-2):117–129.

A

B C

Fig. 6. A combination of anti-EGFR mAbs interfereswith EGF-dependent proliferation and with in vivogrowth of TNBC cells. (A and B) HCC70 cells (3,000cells per well) were seeded in 12-well plates andallowed to adhere overnight. Afterward, cells wereincubated for 5 d in serum-free medium containingmAbs (20 μg/mL total) in the absence or presence ofEGF (10 ng/mL). After fixation and staining withGiemsa (0.2% in saline), cells were photographedand quantified. Shown are the results of one of threeexperiments. (C and D) Groups of five CD1/nude micewere injected intradermally with 7 × 106 HCC70 cells.Antibodies (total: 160 μg/animal/injection) wereweekly injected intraperitoneally once tumors be-came palpable. Tumor volumes were assessed onceper week. **P < 0.01 by two-way analysis of variancewith Bonferroni’s multiple comparison posttests. Theaverage tumor size in each group ± SEM is presented.

1820 | www.pnas.org/cgi/doi/10.1073/pnas.1220763110 Ferraro et al.

Dow

nloa

ded

by g

uest

on

June

14,

202

0