enhanced targeting of the egfr network with mm-151, an...

Large Molecule Therapeutics

Enhanced Targeting of the EGFR Network withMM-151, an Oligoclonal Anti-EGFR AntibodyTherapeuticJeffrey D. Kearns, Raghida Bukhalid, Mark Sevecka, Gege Tan, Nastaran Gerami-Moayed,Shannon L.Werner, Neeraj Kohli, Olga Burenkova, Callum M. Sloss, Anne M. King,Jonathan B. Fitzgerald, Ulrik B. Nielsen, and Beni B.Wolf

Abstract

Although EGFR is a validated therapeutic target acrossmultiplecancer indications, the often modest clinical responses to currentanti-EGFR agents suggest the need for improved therapeutics.Here, we demonstrate that signal amplification driven by high-affinity EGFR ligands limits the capacity ofmonoclonal anti-EGFRantibodies to block pathway signaling and cell proliferation andthat these ligands are commonly coexpressed with low-affinityEGFR ligands in epithelial tumors. To develop an improvedantibody therapeutic capable of overcoming high-affinityligand-mediated signal amplification, we used a network biologyapproach comprised of signaling studies and computationalmodeling of receptor–antagonist interactions.Model simulationssuggested that an oligoclonal antibody combination may over-come signal amplification within the EGFR:ERK pathway drivenby all EGFR ligands. Based on this, we designed MM-151, a

combination of three fully human IgG1 monoclonal antibodiesthat can simultaneously engage distinct, nonoverlapping epitopeson EGFR with subnanomolar affinities. In signaling studies, MM-151 antagonized high-affinity EGFR ligandsmore effectively thancetuximab, leading to an approximately 65-fold greater decreasein signal amplification to ERK. In cell viability studies, MM-151demonstrated antiproliferative activity against high-affinity EGFRligands, either singly or in combination, while cetuximab activitywas largely abrogated under these conditions. We confirmed thisfinding both in vitro and in vivo in a cell line model of autocrinehigh-affinity ligand expression. Together, these preclinical studiesprovide rationale for the clinical study of MM-151 and suggestthat high-affinity EGFR ligand expression may be a predictiveresponse marker that distinguishes MM-151 from other anti-EGFR therapeutics. Mol Cancer Ther; 14(7); 1625–36. �2015 AACR.

IntroductionEpidermal growth factor receptor (EGFR; ErbB1) is the proto-

typical member of the ErbB family of receptor tyrosine kinases(RTK), which also comprises ErbB2 (HER2), ErbB3, and ErbB4.Dysregulation of EGFR/ErbB signaling and the resulting signalamplification to the downstream MAPK (ERK) and AKT effectorpathways in cancer is well described (1–3). Currently approvedEGFR-targeted therapies, including small-molecule tyrosinekinase inhibitors (TKI; e.g., erlotinib and gefitinib) and ligand-blocking monoclonal antibodies (e.g., cetuximab and panitumu-mab), have demonstrated clinical activity in KRAS wild-typecolorectal cancer, squamous cell carcinoma of the head and neck(HNSCC), and non–small cell lung cancer (NSCLC). However,the often modest response to these therapies motivates thedevelopment of predictive diagnostics, design of optimal drugcombinations, and improved therapeutics (4). Although a num-ber of newer approaches to EGFR targeting, including antibody

combinations (Sym004; bi-clonal; refs. 5, 6), dual-specificityantibodies (MEHD7945a; ref. 7), and ADCC-enhanced antibo-dies (imgatuzumab; ref. 8), have recently entered the clinic, theseagents have not shown enhanced clinical activity in early testing.

A family of seven ligands, dichotomized by affinity into twoclasses, activate homo- and heterodimers of EGFR and ErbB2/3/4receptors (Fig. 1A; refs. 9–12). The low-affinity ligands includeamphiregulin (AREG), epiregulin (EREG), and epigen (EPGN)and have approximate affinities between 100 and 3,000 nmol/L.The high-affinity ligands include betacellulin (BTC), EGF, hepa-rin-binding EGF-like growth factor (HB-EGF), andTGFa andhaveapproximate affinities between 1 and 10 nmol/L. The ligandsshare a high degree of functional redundancy in that each pro-motes cell proliferation, and both double (AREG and EGF) andtriple (AREG, TGFa, EGF) ligand knockout mice are viable anddevelop to adulthood (13, 14). However, the ligands can bedifferentiated by their abilities to activate distinct ErbB homo-and heterodimers (9, 15) and to affect EGFR trafficking (16, 17),which may enable divergent biologic responses dependingupon the composition and dynamics of the ligand microenvi-ronment (18).

EGFR and other RTK ligands also appear to play divergent rolesin determining sensitivity to EGFR-targeted therapies. For exam-ple, preclinical studies show that upregulation of TGFa or HB-EGF can maintain EGFR signaling and promote cetuximab resis-tance (19–21), whereas heregulin and hepatocyte growth factorcan induce cetuximab resistance via bypass of EGFR (22, 23). Incontrast, AREG upregulation correlates with sensitivity to cetux-imab and EGFR TKIs in vitro (24, 25). In colorectal cancer tumors,

Merrimack Pharmaceuticals, Cambridge, Massachusetts.

Note: Supplementary data for this article are available at Molecular CancerTherapeutics Online (http://mct.aacrjournals.org/).

J.D. Kearns and R. Bukhalid contributed equally to this article.

Corresponding Author: Jeffrey D. Kearns, Merrimack Pharmaceuticals, OneKendall Square, Suite B7201, Cambridge, MA 02139. Phone: 617-441-1000;Fax: 617-491-1386; E-mail: [email protected]

doi: 10.1158/1535-7163.MCT-14-0772

�2015 American Association for Cancer Research.

MolecularCancerTherapeutics

www.aacrjournals.org 1625

on June 19, 2019. © 2015 American Association for Cancer Research. mct.aacrjournals.org Downloaded from

Published OnlineFirst April 24, 2015; DOI: 10.1158/1535-7163.MCT-14-0772

http://mct.aacrjournals.org/

upregulation of the low-affinity EGFR ligands AREG and EREGoccurs, and several clinical studies suggest that high levels of theirmRNAs predict clinical benefit with cetuximab treatment (26–29). Less is known about the possible predictive nature of high-affinity EGFR ligand expression; however, several groups havereported high-affinity ligand expression (RNA and protein) intumors and serum (19, 26, 30–34). Together, these studieshighlight the capacity of EGFR ligands to modulate sensitivityto EGFR-targeting antibodies and potential as predictive markersin the clinic.

In this study, we measure the activities of existing EGFR anti-bodies to inhibit signal amplification and proliferation stimulat-edby low- andhigh-affinity EGFR ligands.Weobserve incompleteinhibition of high-affinity ligands, both alone and in combina-tion, and demonstrate through a bioinformatics prevalence anal-ysis that these ligands are expressed across EGFR-related cancerindications. Motivated by these findings, we describe develop-ment and characterization of MM-151, a novel oligoclonal com-bination of three fully human IgG1 antibodies targeted againstnonoverlapping epitopes on EGFR, designed specifically to

TGFα (high-affinity) EGF (high-affinity)

pER

K, fraction +

lig control

AREG (low-affinity)

MEHD7945a

Nimotuzumab

Inhibitor, molar

1.1e

-11

3.2e

-11

9.6e

-11

2.9e

-10

8.7e

-10

2.6e

-9

7.8e

-9

3.1e

-8

1.3e

-7

3.2e

-11

9.6e

-11

2.9e

-10

8.7e

-10

2.6e

-9

7.8e

-9

3.1e

-8

1.3e

-7

5.0e

-7

3.2e

-11

9.6e

-11

2.9e

-10

8.7e

-10

2.6e

-9

7.8e

-9

3.1e

-8

1.3e

-7

5.0e

-7

Cetuximab

Panitumumab1.5

1.25

1.0

0.75

0.5

0.25

0.0

−14 −12 −10 −80.00

0.25

0.50

0.75

1.00

Ligand, Log(Molar)

Fra

ctio

n ac

tivity EGF

AREG

pERK pEGFR B

A

D

1 4411 1 1131 311 221EGFR dimers

Low- and high-affinityEGFR ligands

AREGEREGEPGNBTC EGF

HB-EGFTGFα

AREGEGF

pEGFR14388.0

61.7

233.3x

pERK8.31.5

5.5xΔ

EC50 (pmol/L)

E

< 5% signal

signal amplification

100% signalProliferation

EGFR dimersCRAS

RAF

MEK

ERK

Cel

l via

bilit

y,F

ract

ion

+Li

g co

ntro

l

Control

EPGNAREGEREGTGFαEGFBTCHB-EGF

−12 −10 −8 -60.25

0.50

0.75

1.00

Cetuximab, Log(Molar)

−10 −8 −6

0.25

0.50

0.75

1.00


−10 −8 −6

0.25

0.50

0.75

1.00


SCC-25 (HNSCC) H322M (NSCLC)CCK-81 (CRC)

−12 −12

Anti-EGFR monoclonal antibodies

EGFR downregulation

Immune-effectorfunction (ADCC,CDC)

Ligand antagonism

Figure 1.High-affinity EGFR ligands initiate signal amplification that limits the capacity ofmonoclonal EGFR antibodies to block EGFR pathway signaling and cell proliferation.A, the schematic shows ligand redundancy within the EGFR network. Multiple low- (AREG, EREG, EPGN) and high-affinity (BTC, EGF, HB-EGF, TGFa) ligands caninitiate formation of EGFR-containing ErbB dimers. Receptor homo- and heterodimers are shown at the cell surface with 1 ¼ EGFR/ERBB1, 2 ¼ Her2/ERBB2,3 ¼ HER3/ERBB3, and 4 ¼ HER4/ERBB4. Anti-EGFR monoclonal inhibitors are described to inhibit EGFR-expressing tumors via cell-intrinsic (ligand antagonism,receptor downregulation) and -extrinsic (immune-effector activity) mechanisms. B, the high-affinity EGFR ligand EGF elicits signal amplification to a greater extentthan the low-affinity EGFR ligand AREG. NCI-H292 cells were stimulated with AREG (blue) or EGF (orange) at the indicated concentrations for 10 minutes andpEGFR (solid lines) and pERK (stippled lines) measured by ELISA. Results for each analyte are normalized to the maximum observed value. The EC50 value and foldchange (D) between the EC50 of each ligand are indicated beside the graph. C, the schematic summarizes signal amplificationwithin the EGFR:ERK network whereineven low fractional EGFR activation can stimulate high ERK activities through robust signal amplification. D, monoclonal EGFR antibodies poorly block signalamplification by high-affinity EGFR ligands. NCI-H322M cellswere treatedwith inhibitor for 1 hour followed by 10-minute stimulationwith 0.8 nmol/L AREG, TGFa, orEGF. pERKwasmeasured by ELISA, and the results normalized to untreated control. Values below the limit of detection are represented on the heatmapwith a blue/white dotted pattern. E, cell proliferation stimulated by high-affinity ligands is partially or completely resistant to cetuximab treatment. Relative cell viabilitieswere measured in multispheroid culture of CCK-81, SCC-25, and NCI-H322M cell lines after 3-day cetuximab treatment in the presence of the indicated EGFR ligands(HB-EGF: 0.8 nmol/L; all other ligands: 8 nmol/L).

Kearns et al.

Mol Cancer Ther; 14(7) July 2015 Molecular Cancer Therapeutics1626




prevent EGFR activation and downstream signal amplification byfully inhibiting all EGFR ligands.

Materials and MethodsCell culture and reagents

Human cancer cell lines were obtained from the AmericanType Culture Collection (A-431, Cal 27, FaDu, NCI-H292, NCI-H358, SCC-25, SCC-4, SK-MES-1), Japan Health Foundation(CaR-1 and CCK-81), the National Cancer Institute (NCI-H322M), and Sigma-Aldrich (LIM125). All cell lines wereauthenticated by short tandem repeat profiling by the manu-facturer and passaged in our laboratory for less than 3 monthsaccording to manufacturers' instructions (Supplementary TableS1) in monolayer culture in a humidified incubator at 37�C in a5% CO2 atmosphere. Cetuximab was obtained from Myoderm.Panitumumab, nimotuzumab, MEHD7945a, and the singlecomponent antibodies comprising the Sym004 combination(Sym992 and Sym1024) were expressed and purified usingsequence information from the patent literature. The MM-151 antibodies were derived from a fully human IgG libraryin yeast. Isolated binders were binned by nonoverlappingepitopes on EGFR (Adimab LLC) and expressed in CHO cells.The recombinant EGFR ligands EGF and HB-EGF were obtainedfrom Peprotech. Recombinant TGFa, BTC, AREG, and EREGwere obtained from R&D Systems.

Enzyme-linked immunosorbent analysisTo study phosphorylated levels of EGFR and ERK under

ligand and drug treatments, cells were grown under low-serumconditions overnight, pretreated with inhibitors for 2 hours,stimulated with EGFR ligands for 10 minutes, and lysed undernative conditions in Mammalian Protein Extract Reagent (M-PER; Pierce)–based lysis buffer. To measure levels of total EGFRupon drug treatments, cells were grown under low-serumconditions, treated with inhibitors for 18 hours, and lysed inM-PER–based lysis buffer. To measure levels of HB-EGF proteinin lysates and conditioned medium of engineered H322M cells,cells were grown under full-serum conditions for 72 hours,supernatant medium was collected, and adherent cells werelysed in M-PER-based lysis buffer. Levels of each target proteinwere then measured by sandwich enzyme-linked immunosor-bent analysis (ELISA) assays, using commercially availableantibodies (pEGFR: Thermo Scientific/R&D Systems; totalEGFR: R&D Systems; pERK: Cell Signaling Technology; HB-EGF: R&D Systems) and a conventional chemiluminescentreadout (Pierce; pEGFR, total EGFR, and HB-EGF assays), oran electrochemiluminescent detection system (Meso Scale Dis-covery; pERK assay). A detailed description is contained in theSupplementary Methods.

Western blottingA-431, Cal 27, NCI-H292, or NCI-H322M cell lines were

cultured and treated in the same manner as for ELISA analysis,then lysed in M-PER buffer. Lysate protein concentrations weredetermined thereafter by bicinchoninic acid assay. Equal amountsof lysates were then separated electrophoretically on sodiumdodecyl sulfate polyacrylamide copolymer gels, transferred tonitrocellulosemembranes, blocked and probed with primary anddye-labeled secondary antibodies, and finally scanned on a near-infrared fluorescent imager.

Epitope binning experiment with surface plasmon resonanceOne of the MM-151 component antibodies was immobilized

on the surface of a CM5 Sensor Chip (GEHealthcare) using aminecoupling according to the manufacturer's instructions (GEHealthcare). The chipwas inserted into a Biacore 3000 instrument(GE Healthcare) and the baseline signal intensity measured asBiacore Response Units. 0.5 mmol/L each of recombinant humanEGFR extracellular domain (EGFR-ECD) and individual MM-151component antibodies were then sequentially injected at a flowrate of 20 mL/min, and the increase in resonance signal wasmeasured over time.

Assessment of ligand antagonism by the individual MM-151component antibodies

A 96-well plate was seeded with A-431 cells suspended in 100mL flow cytometry buffer (2%FBS, 0.1% sodium azide in PBS) at adensity of 30,000 cells/well. Cells were incubated at room tem-perature for 1 hour with an EC99 concentration of antibody, ascalculated using a computational binding model (P1X: 0.96nmol/L, P2X: 2.00 nmol/L, P3X: 4.70 nmol/L; ref. 35), followedby 10 minutes with a dilution series (0.003 to 200 nmol/L) ofbiotin-XX-EGF ligand (Invitrogen). Cells were washed with 100mL of flow cytometry buffer at 4�C and resuspended in 100 mLstreptavidin Alexa Fluor 647 conjugate (Invitrogen) diluted 1:500in flow cytometry buffer. Following a 30-minute incubation, cellswere washed twice as before, resuspended in 80 mL fixationbuffer (2% FBS, 2% paraformaldehyde in PBS), and transferredto a U-bottom 96-well plate. Cells were analyzed by flow cyto-metry on aBectonDickinson FACSCalibur instrumentwith gatingfor live cells (FSC/SSC) and for Alexa Fluor 647–positive cells.Data were analyzed using WinList v6.0 software.

Cell proliferation studiesCell lines were seeded into 96-well low-binding multispheroid

culture plates (Scivax) at a density of 5,000 cells/well in basemedium supplemented with 1%Matrigel (BD Biosciences). At 24hours, ligands and inhibitors were added concurrent with reduc-tion of serum to 2%. Relative cell viabilities at 72 hours weredeterminedusing theCellTiter-Glo assay (Promega) as a surrogatemeasure of cell density.

Engineering of an autocrinemodel of high-affinity EGFR ligandexpression

The NCI-H322M cell line was transfected with a plasmid thatexpresses full-length human HB-EGF precursor (NCBI referencesequence NM_001945.2) from an EF1a promoter, as well as apuromycin resistance gene. Seventy-two hours after transfection,selection for stable integrationwas initiated using escalating dosesof puromycin, which resulted in the generation of a stable cell lineafter several weeks of selection. A GFP-expressing control cell linewas derived from NCI-H322M cells in the same manner.

Murine xenograft studiesFemale Fox Chase SCID-Beigemice (CB17.Cg-PrkdcscidLystbg-J/

Crl; Charles River Labs), weighing 16 � 0.5 g, were inoculatedwith 0.2mLofH322M-HB-EGF cell suspension in PBS containinggrowth factor–reduced Matrigel (BD Biosciences) at a densityof 5 � 106 cells per mouse. Once tumors had reached approxi-mately 200 mm3 in volume, mice were randomized into treat-ment groups (10 mice/group) to receive PBS, cetuximab at

Enhanced Inhibition of EGFR Signal Amplification by MM-151

www.aacrjournals.org Mol Cancer Ther; 14(7) July 2015 1627




6.25 mg/kg, 12.5 mg/kg, or 25 mg/kg, or the MM-151 toolcompound "25E-Trio" (consisting of 25E, P2X, and P3X) at dosescalculated to result in summed serum exposure for the combi-nation equivalent to the three cetuximab dose levels. Theantibody 25E substitutes for P1X because it does not engagemurine EGFR and thus allows for a weekly administrationschedule that matches the cetuximab comparator. P1X is anaffinity matured variant of 25E (11 vs. 145 pmol/L), and theantibodies engage overlapping epitopes on EGFR. Treatmentswere administered by IP injection, with cetuximab, 25E, andP3X on a Q7D schedule on Mondays (the three antibodies haveequivalent half-life values of approximately 4–5 days), and P2Xon Mondays, Wednesdays, and Fridays (owing to an additionalclearance route due to target-mediated drug disposition viamurine EGFR). Tumor dimensions were measured three timeseach week by calipers, and the tumor volumes were calculatedusing the formula: p/6 � L � W2, where L and W, respectively,represent the larger and smaller tumor diameter.

Analysis of EGFR ligand expression in primary tumorsEGFR ligand mRNA expression in 1,792 treatment-na€�ve, pri-

mary, solid tumors from colorectal cancer, HNSCC, and NSCLCindications was measured by The Cancer Genome Atlas (TCGA)Research Network (http://cancergenome.nih.gov) using the Illu-mina HiSeq RNAseqv2 platform. RSEM normalized data weredownloaded in January 2014 from the Broad Institute GenomeData Analysis Center (http://gdac.broadinstitute.org) and alsoinclude a small number (�10% per indication) of matchednormal tissue samples. MATLAB R2014a (MathWorks) was usedto calculate relative expression versus the pooled normal sampleswithin each indication. Visualization and correlation analyseswere performed in MATLAB and JMP 11.1 (SAS).

Computational modeling of combinations of EGFR ligandantagonists

A mechanistic computational model was constructed todescribe the competitive association of EGFR with ligand orligand-antagonists on the surface of a cell. The reactions in themodel are represented as a system of ordinary differential equa-tions. Model construction and simulation were performed withMATLAB SimBiology (MathWorks) and are fully described in theSupplementary Methods (Supplementary Tables S2–S4).

ResultsMonoclonal antibodies poorly inhibit high-affinity EGFRligands

Mechanistic, biochemical studies show that the high-affinityligands EGF and TGFa readily activate EGFR and initiate signalamplification through the ERK pathway (36–39). Whether low-affinity ligands initiate similar network dynamics and comparablesignal amplification is unclear. We therefore performed a signal-ing experiment with NCI-H292 NSCLC cells to ascertain whetherlow- and high-affinity EGFR ligands elicit similar (redundant) ordistinct pathway activation (Fig. 1B). As expected, the represen-tative ligands AREG and EGF activate EGFR, as determined byinduction of EGFR autophosphorylation (pEGFR), with a shift inEC50 (233-fold) nearly proportional to the difference in affinitybetween the two ligands. Unexpectedly, the large difference inEGFR activation with low- and high-affinity EGFR ligands did notconfer a large difference in downstream signaling to ERK, with

only an approximately 5.5-fold EC50 difference observed betweenAREG (low-affinity) and EGF (high-affinity). This suggests thatthe network topology of the EGFR:ERK pathway, previouslydescribed as negative feedback amplifier (40), enables robustsignal amplification that compensates for the wide differences inEGFR ligand affinities (Fig. 1C).

Given theweak induction of pEGFRby AREG,we surmised thatcetuximab would potently inhibit ERK signaling induced by low-affinity ligands and suboptimally inhibit signaling induced byhigh-affinity ligands. To test this, a pERK signaling experimentwasperformed with cetuximab and three representative ligands in theNSCLC cell lines NCI-H322M (Fig. 1D) and NCI-H292 (Supple-mentary Fig. S1). Indeed, cetuximab fully inhibits pERK signalingbelow the limit of the assay in cells stimulated with the low-affinity ligand AREG, but elicits weaker inhibition of signalinginduced by the high-affinity ligands TGFa and EGF. To furthercharacterize the interplay between ligand and antagonist affini-ties, the experiment included three additional a-EGFR antibodieswith approximate monovalent affinities of 50 pmol/L (panitu-mumab; fully-human IgG2; ref. 41), 10 nmol/L (MEHD7945a;dual-specificity EGFRþErbB3 IgG1; ref. 7), and 45 nmol/L (nimo-tuzumab; humanized IgG1; ref. 42). Cetuximab is a chimericmouse/human IgG1 with an approximate monovalent affinity of100 pmol/L (43). Strikingly, all therapeutic anti-EGFR monoclo-nal antibodies examined showed diminished inhibition of high-affinity ligand-driven ERK signaling regardless of antibody affin-ity, suggesting a class effect.

To determine whether this trend scales to phenotypic response(cell fate), we assessed the inhibition of proliferation driven byindividual ligands in representative cell lines from three indica-tions—CCK-81 (KRAS wild-type colorectal cancer), SCC-25(HNSCC), and NCI-H322M (NSCLC; Fig. 1E). Detailed mecha-nistic signaling studies by Albeck and colleagues identified athreshold of 90% inhibition of ERK activity required for EGFRinhibitors to substantively affect cell proliferation (39). Consis-tent with this report and our ERK signaling results, cetuximabstrongly inhibits proliferation driven by low-affinity ligands.High-affinity ligands induced complete or partial cetuximabresistance even at a drug concentration (1,000nmol/L)well abovethe steady-state serum trough concentration (360 nmol/L)reported in the clinic (44).

Primary tumors expressmixtures of low- andhigh-affinityEGFRligands

Our in vitro signaling data suggest that cetuximab and otherEGFR ligand antagonists may have reduced activity in tumorsthat express high-affinity ligands. We therefore evaluated theprevalence of the EGFR ligands across four cancer indications inwhich EGFR-targeting agents have shown clinical benefit—colorectal cancer, HNSCC, lung adenocarcinoma, and lungsquamous cell carcinoma (LUSC; refs. 45–47). To this end,mRNA expression data for the seven EGFR ligands wereobtained from TCGA for 1,792 treatment-na€�ve primary tumorsand 178 matched normal tissue samples (SupplementaryTables S5 and S6).

We observe distinct expression patterns for six EGFR ligandsboth in absolute and relative scales (Fig. 2A; Supplementary Fig.S2). EPGN was infrequently expressed in these indications andacross other TCGA indications (data not shown) and was there-fore excluded from further analysis. The expression levels of theremaining six ligands are log-normally distributed. Of note,

Kearns et al.





colorectal cancer and HNSCC tumors overexpress AREG andEREG, whereas lung adenocarcinoma and LUSC tumors over-express TGFa and EGF and underexpress HB-EGF. To confirmthese results, we additionally profiled 64 primary colorectalcancer tumors and 10 normal colon samples for EGFR ligandexpression using a multiplex qPCR platform. In agreement withthe TCGA colorectal cancer analysis, we observed robust over-expression of AREG and EREG and a similar ligand mixtureexpression profile (Supplementary Fig. S3 and SupplementaryTable S7).

To our knowledge, this is the first population analysis of EGFRligand expression across cancer indications. As orthogonal vali-dation, we performed a correlation analysis with the TCGAexpression data (Supplementary Table S8) to determine concor-dance with published clinical studies that have reported elevatedand correlated expression of AREG and EREG in colorectal tumors(26–29). We note the contiguous location of both genes onchromosome 4 and similar transcriptional regulatory mechan-isms (48, 49). Our colorectal cancer results are consistent withthese findings and, additionally, we observe correlated expressionin the remaining three indications.

Given the redundant activities of the EGFR ligands in stimu-lating the EGFR pathway and potential for different ligandexpression profiles to impact response to anti-EGFR therapies(Fig. 1), we sought to understand the coexpression of EGFRligands across indications (Fig. 2B). Indeed, an expanded analysisof the relative expression of EGFR ligands across 14 TCGA indica-tions reveals that distinct mixtures are present in the majority ofindications (Supplementary Fig. S4). The majority of tumorsamples in the four EGFR-related indications overexpress two ormore EGFR ligands above the median of the indication-specificnormal tissue (Fig. 2B). The converse also holds true, as only aminority of samples harbor overexpression of no ligands. When amore stringent threshold of overexpression above the 75th per-centile of the normal tissue is used, themajority of tumor samplesare still defined as coexpression positive (a plurality in HNSCC).

Development of MM-151: an oligoclonal combination ofa-EGFR antibodies

Inspired by a systems biology analysis from our group thatdescribed synergistic inhibition of a cell surface receptor tyrosinekinase by a combination of two or more targeted agents (50), we

# ligands012345

Colorectal (n = 333)

Head & neck (n = 424)

Lung, adeno(n = 490)

Lung, squamous (n = 482)

Nor

mal

tiss

ue50

th p

erce

ntile

Nor

mal

tiss

ue75

th p

erce

ntile

Co

exp

ress

ion

, # li

gan

ds

abo

ve t

hre

sho

ld

−2

−1

0

1

2

3

4

−4

−3

−2

−1

0

1

2

Colorectal (n = 333)

Head & neck (n = 424)

Lung, adeno(n = 490)

Lung, squamous (n = 482)

AR

EG

ER

EG

AR

EG

ER

EG

AR

EG

ER

EG

AR

EG

ER

EG

HB

EG

FT

GF

A

EG

FB

TC

HB

EG

FT

GF

A

EG

FB

TC

HB

EG

FT

GF

A

EG

FB

TC

HB

EG

FT

GF

A

EG

FB

TC

Abs

olut

eR

elat

ive

vs N

orm

al

Low High

B

A

Exp

ress

ion

, lo

g10

(RS

EM

no

rmal

ized

)

Affinity : Low High Low High Low High

overunder

01

2

3 45

Figure 2.Primary colon, head and neck, andlung tumors upregulate andcoexpress multiple low- and high-affinity EGFR ligands. mRNAexpression of EGFR ligands wasassessed in primary, treatment-na€�vesolid tumors as measured by RNAseq(Illumina HiSeq). Datawere generatedby TCGA Research Network anddownloaded from the Broad Institute.A, density plots of absolute (top) andfold-versus-normal tissue (bottom)expression in four EGFR-relatedindications. The stippled line in thebottom plot denotes the medianexpression of the normal tissuesamples within each indication.Expression above this line indicatesupregulation. B, pie charts denotingthe number of tumor samples withinan indication that coexpress 0–5ligands above either the median (top)or the 75th percentile (bottom) of thenormal tissue samples within thatindication. Note that tumorsupregulate and coexpress multipleEGFR ligands.






applied a similar approach to ascertain whether a combination ofligand antagonists could inhibit high-affinity ligand signaling. Amechanistic computational model was constructed to describethe association of ligand and EGFR on the cell surface andthe inhibition of this complex by combinations of ligand antago-nists. As expected from our earlier work (50), synergistic inhibi-tion is observed in simulations with two or three ligand antago-nists (Fig. 3A). The improved inhibition by the combination ofligand antagonists is most pronounced in the presence of high-affinity ligand and is observed with both monovalent (e.g., Fab)and bivalent (e.g., IgG) inhibitors (Supplementary Fig. S5). Addi-tion of a third antagonist enables increased inhibition of high-affinity ligands (Fig. 3B).

Based on the simulation results, we sought to develop anoligoclonal combination of anti-EGFR monoclonal antibodiescapable of providing superior ligand antagonism. A library of 65fully human IgG1 antibodies was generated using a yeast-basedantibody discovery platform consisting of a unique library design

that recapitulates the humanpreimmune diversity (Adimab LLC).The antibodies were clustered into five nonoverlapping epitopebins and assessed individually and in combination for affinity toEGFR and inhibition of ligand-driven signaling (data not shown).The affinity of each of the lead antibodies from each round ofselection was then further increased through affinity maturation.

Three antibodies from nonoverlapping bins were selected forfurther characterization and the kinetics of EGFR binding fullydescribed via an integrated study of binding experiments andmechanistic computational modeling in Harms and colleagues(35). In that study, the antibodies Ab#1, Ab#2, and Ab#3 refer toP3X, P1X, and P2X, respectively. In brief, all three antibodies werefound to have subnanomolar affinities (P1X¼ 11 pmol/L; P2X¼71 pmol/L; P3X ¼ 300 pmol/L), but vary in their ability tocrosslink receptor—P2X and P1X are strongly avid, whereasP3X is weakly avid.

We introduceMM-151 as an oligoclonal therapeutic consistingof a combination of these three fully human IgG1 antibodies. P1X

C P1X

P3X

E

biotin-EGF, Log(Molar)

Bou

nd E

GF

, M

FI

−11 −10 −9 −8 −7 −6

0

200

400

600 EGF control F

−12 −10 −8 −60

15

Antibody, Log(Molar)

Pho

spho

-EG

FR

, MF

I

10

5

×107

D

Res

pons

e un

its

400

800

1,200

0 EGFR-ECD

P3X

P2X

Time (seconds)200 400 600 800 1,0000

I

IIIII

IV

P2X

A B

EG

F:E

GF

R c

ompl

ex,

Fra

ctio

n +

Lig

cont

rol

−11 −10 −9 −8 −7 −60

0.25

0.5

0.75

1

Individual antagonist, Log(Molar)

# of antagonistsSingleDuoTrio

IC98

(nm

ol/L

)

0

15

30

45

60

Single Duo Trio

Low-affinity (100 nmol/L)High-affinity (1 nmol/L)

538 ≈

P1XP2X

P3X

P1XP2X

P3X

Figure 3.Development of MM-151: an oligoclonal combination of three fully human EGFR antibodies. A, computational modeling of ligand:receptor interactions predictsenhancedEGFR antagonismwith three noncompeting ligand antagonists. Shown are the simulated dose–response curves for the formation of EGF:EGFR complex inthepresenceof one (blue), two (green), or three (red) nonoverlapping antagonists. Stippled lines denote thedose–response curves calculated byBliss Independencefor the two (green) or three (red) inhibitor combinations, demonstrating that both combinations are synergistic. B, modeling predicts enhanced antagonism ofhigh-affinity EGFR ligandswith a trio combination of antagonists. Calculated IC98 values frommodel simulations, performed as in A, with either low-affinity (blue) orhigh-affinity (orange) ligand stimulation. C, MM-151 oligoclonal design showing the three component IgG1 antibodies—P1X, P2X, and P3X—that bind tononoverlapping epitopes on domains I and III of EGFR, as indicated. D, a binning experimentwas performed via surface plasmon resonance (Biacore) to demonstratesimultaneous engagement of EGFR by the three MM-151 antibodies. In the experiment shown, P1X was conjugated to a CM5 chip and EGFR-ECD, P3X, P2X, and P1Xwere then sequentially injected. E, P1X, P2X, and P3X block ligand binding to cell surface EGFR. Ligand antagonismwas assessed for each antibody alone via a ligandcompetition flowcytometry experiment at saturating antibody concentrations (respective EC99 of on-cell binding). F, P1X, P2X, andP3Xare EGFR ligand antagonists.A-431 cells were treated with P1X, P2X, or P3X for 1 hour followed by 10-minute stimulation with 8 nmol/L EGF ligand. pEGFR was measured by ELISA, andresults were plotted in units of mean fluorescence intensity (MFI).

Kearns et al.





and P3X engage epitopes on Domain III of the EGFR extracellulardomain, whereas P2X engages an epitope on Domain I (Fig. 3C),and all are capable of simultaneously binding receptor, as dem-onstrated by an epitope binning assay (Fig. 3D; SupplementaryFig. S6).

Ligand antagonism for each antibody was measured in vitrousing EGF and the A-431 human epidermoid carcinoma cell linewhich highly expresses EGFR (�2e6/cell). P1X and P2X stronglyinhibited EGF binding and potently inhibited subsequent EGFRactivation (Fig. 3E and F) through engagement of the well-described cetuximab-like epitope on Domain III and a novelligand-blocking epitope on Domain I, respectively. P3X alsoblocked EGF binding and EGFR activation, but less potently thanthe other two antibodies. The MM-151 combination is thuscomprised of two potent and one partial ligand antagonists.Relative to currently approved and investigational anti-EGFRmonoclonal antibodies, MM-151 has the three unique character-istics of being the only tri-clonal antibody formulation, the onlycombination that contains multiple antibodies that are individ-ually ligand antagonists, and the only combination that targetsthe complete ligand binding epitope that spans EGFR domains I(P2X) and III (P1X, P3X).

MM-151 overcomes high-affinity ligand-driven signalamplification and ligand redundancy

The mechanistic computational modeling that guided thedevelopment of MM-151 predicted that superior inhibition ofpEGFR is an emergent property of an oligoclonal combinationof ligand antagonists (Fig. 3A). To test whether MM-151achieved this design criterion, phospho-signaling assays wereperformed with NCI-H292 NSCLC cells and demonstrate that,

indeed, MM-151 elicits near-complete inhibition of EGF-drivenpEGFR (Fig. 4A) at antibody concentrations � 1 nmol/L. Assuggested by the ligand activation experiments in Fig. 1, inhi-bition of the profound signal amplification to pERK by MM-151 leads to a disproportionately large increase in pERK inhi-bition (Fig. 4B). In comparison, cetuximab is unable to over-come signal amplification to ERK, as even strong inhibition ofpEGFR activity (84%) elicits only a modest decrease in pERK(8%) at an antibody concentration (500 nmol/L) that is abovethe 360 nmol/L steady-state trough concentration reported inclinical studies (44).

We extended this comparison to assess inhibition of keyeffector proteins induced by single administration of low- orhigh-affinity EGFR ligands. Analysis of pERK signaling drivenby AREG, TGFa, and EGF reveals expected comparable inhibi-tion of AREG-driven signaling, but maintained potency forMM-151 in the presence of high-affinity ligands in bothNCI-H322M (Fig. 4C) and NCI-H292 cell lines (Supplemen-tary Fig. S7), unlike cetuximab and the other single monoclonalantibodies (Fig. 1E). The same profile is observed for keyphospho-proteins in the MAPK/AKT/mTOR signaling pathwaysand demonstrates that the superior EGFR inhibition by MM-151 similarly affects the main effector pathways downstream ofEGFR (Fig. 4D).

The oligoclonal combination enables robust receptordownregulation and ADCC/CDC activities

Next, we characterized the ability of MM-151 to elicittwo additional mechanisms of action that have been describedas a class effect for combinations of noncompetitive antibodiesagainst cell surface receptors—receptor downregulation (5, 51–

A

B

C D

pERK

pp90RSK

pAKT

pPRAS40

pS6

pSTAT3

PCNA

−+

−−

+−

−+

+−

−+

+−

EGF

CTX MM-151

AREG

Fra

ctio

n +

EG

F c

ontr

ol

−11 −10 −9 −8 −7 −60.0

0.2

0.4

0.6

0.8

1.0

1.2

pEGFR

MM-151Cetuximab

Total antibody, Log(Molar)

} ~2x


pERK

Fra

ctio

n +E

GF

con

trol

−11 −10 −9 −8 −7 −60.0

0.2

0.4

0.6

0.8

1.0

1.2

~65x}

AREG

TGFα

EGF

AREG

TGFα

EGF

00.2

0.4

0.6

0.8

11.2

Cet

uxim

abM

M-1

51

pERK, Fraction +Lig control

Inhibitor, Molar

3.2e

-11

9.6e

-11

2.9e

-10

8.7e

-10

2.6e

-9

7.8e

-9

3.1e

-8

1.3e

-7

5.0e

-7

“On” “Off”

Figure 4.MM-151 elicits superior ligand antagonismversus high-affinity ligands. NCI-H292 cellswere incubatedwithMM-151 or cetuximab for 1 hour at indicated concentrationsfollowed by 10-minute stimulation with ligand. A and B, MM-151 blocks signal amplification by the high-affinity EGFR ligand EGF (8 nmol/L) more effectively thancetuximab. pEGFR and pERK were measured by ELISA, and the results normalized to treatment with EGF alone. C, MM-151 inhibits ERK activation by high-affinityligands TGFa and EGF (0.8 nmol/L) more effectively than cetuximab. pERK was measured by ELISA, and values below the limit of detection are representedon the heat map with a blue/white dotted pattern. D, MM-151 elicits superior inhibition of key phospho-signaling nodes in the EGFR pathway. Following 1-hourincubation with 1 mmol/L inhibitor, cells were stimulated with 0.8 nmol/L AREG or EGF ligand and phospho-signaling nodes downstream of the ERK (ERK, p90RSK,STAT3), AKT (AKT, PRAS40), or both ERK and AKT (S6) pathways were measured by Western blot.






53) and the immune-effector activities of antibody-dependentcytotoxicity (ADCC) and complement-dependent cytotoxicity(CDC; refs. 5, 54, 55).

In agreement with these studies, the MM-151 oligoclonalcombination promoted both mechanisms. MM-151 stronglydownregulated EGFR in three representative EGFR-expressing celllines (Supplementary Fig. S8A), and thiswas associatedwith rapidlocalization to internalized vesicles (Supplementary Fig. S8E).Neither cetuximab nor the combination of monovalent MM-151Fab fragments induces EGFR downregulation (SupplementaryFig. S8B).

To demonstrate the necessity of oligoclonal targeting ofEGFR to elicit potent receptor downregulation, H358 cells weretreated with MM-151 component antibodies alone and incombination (duos and trio), the monoclonal antibodiescetuximab and panitumumab, or an analogue of the bi-clonalantibody combination Sym004 (refs. 5, 6; Supplementary Fig.S8C and S8D). The three component antibodies are observed todownregulate EGFR with P1X eliciting activity beyond thatachieved by the monoclonal comparators. The oligoclonalcombinations of two antibodies are additive (P2XþP3X) orsynergistic (P1XþP2X, P1XþP3X), as calculated by Bliss Inde-pendence. Notably, the MM-151 combination of three anti-bodies reduced total EGFR levels to a greater extent (94.4%)than the best MM-151 duo combination (P1XþP3X, 89.1%) orthe Sym004 combination (85.7%).

In vitro chromium release assays were performed to assessADCC and CDC activities. While MM-151 and cetuximabachieve similar ADCC activities (Supplementary Fig. S9A), onlyMM-151 elicited CDC activity (Supplementary Fig. S9B),likely through presence of a high cell surface antibody density(54–56).

MM-151 inhibits cellular proliferation driven by clinicallyrelevant multiligand mixtures

Having demonstrated the superiority of MM-151 to inhibitsignaling driven by both low- and high-affinity EGFR ligands,we next studied the capacity of MM-151 to block cell prolif-eration. We first compared the antiproliferative activity of MM-151 with two monoclonal (cetuximab, panitumumab) and onebi-clonal (Sym004 analogue) anti-EGFR antibodies in thepresence of the two representative high-affinity ligands EGFand HB-EGF in the NCI-H322M and LIM1215 cell lines,respectively (Fig. 5A). The resulting data show that oligoclonalinhibition is necessary to block high-affinity ligand-driven cellproliferation, and further show that the tri-clonal MM-151combination has superior activity over the bi-clonal Sym004combination. A comparison of MM-151 with cetuximab acrossa broader panel of cell lines from the colorectal cancer, HNSCC,and NSCLC indications (Fig. 5B and Supplementary Fig. S10)indicates that the superior antiproliferative activity of MM-151is not restricted to particular model cell lines.

As demonstrated in our bioinformatics analysis, primaryhuman tumors express heterogeneous mixtures of low- andhigh-affinity EGFR ligands (Fig. 2; Supplementary Figs. S2 andS3).We hypothesized that the presence of a high-affinity ligand ina ligand mixture would rescue cells from the signaling inhibitionand antiproliferative effects of cetuximab. We therefore firstperformed a signaling experiment inNCI-H322Mcells tomeasureinhibition of pERK stimulation by pairwise dose titrations of low-affinity AREG and high-affinity EGF (Supplementary Fig. S11),

and observed that MM-151 fully inhibits all ligand combinationswhile cetuximab fails to strongly inhibit signal amplificationdriven by combinations containing EGF. We next carried out aproliferation experimentwith a high,fixed concentration of AREG(18.2 nmol/L) alone or in combination with four increasingconcentrations of EGF (Fig. 5C). Cetuximab inhibited AREG-mediated cell proliferation, but addition of EGF at levels �32.3 pmol/L induced partial or complete cetuximab resistance,indicating a dominant role for the high-affinity ligand. Incontrast, MM-151 blocked both AREG-mediated proliferationand AREGþEGF–mediated proliferation when EGF was in the32.3 pmol/L to 3.23 nmol/L range. Thus, MM-151 elicitssuperior antagonism across the range of ligand concentrations.The inclusion of a supra-physiological concentration of EGF(32.3 nmol/L) demonstrates that high ligand burden canreduce sensitivity to both inhibitors.

Lastly, we examined the activities of MM-151 and cetuximabto inhibit cell proliferation in the context of the clinicallyrelevant ligand mixtures identified in the bioinformatics anal-ysis (Supplementary Fig. S4). One cell line was selected for eachindication—CCK-81 (KRAS wild-type colorectal cancer), SCC-25 (HNSCC), and NCI-H322M (NSCLC)—and a proliferationexperiment performed with a clinically relevant concentrationof 100 nmol/L inhibitor and pairwise combinations of twoindication-specific ligands (Fig. 5D). Inhibition of singleligands, denoted in the separate heat maps along each axis, isconsistent with the earlier findings (Fig. 5B; Supplementary Fig.S9A) demonstrating the greater activity of MM-151 againstsingle high-affinity ligands. Of note, the high-affinity ligand(s) present in the mixtures dictated the overall response tocetuximab and MM-151. For example, in the colorectal cancerhigh- and low-affinity ligand mixture (HB-EGF þ EREG), HB-EGF concentrations greater than 1 nmol/L induced cetuximabresistance, but this was largely overcome by MM-151 (Fig. 5D,top row). Similarly, in the HNSCC high- and low-affinity ligandmixture (TGFa þ EREG), TGFa dictated the antiproliferativeresponse to both drugs (Fig. 5D, middle). Finally, mixtures oftwo high-affinity ligands (TGFa þ EGF) in the NSCLC modelinduced partial or complete cetuximab resistance at ligandconcentrations > 10 nmol/L (Fig. 5D, bottom row). In contrast,MM-151 effectively blocked TGFaþEGF–induced proliferationover all concentrations examined. Thus, MM-151 elicits super-ior inhibition of ligand mixtures that include one or more high-affinity ligands in three model systems.

MM-151 inhibits tumor growth driven by expression of a high-affinity EGFR ligand

Our discovery of signaling driven by high-affinity ligands asan in vitro resistance mechanism to EGFR-directed antibodiesmotivated us to further explore this phenomenon in an in vivoxenograft model. For this, we engineered the EGFR-drivenNCI-H322M cell line to stably express autocrine HB-EGF.Compared with parental and GFP-expressing control celllines, the engineered cell line ("H322M-HB-EGF") producedhigher levels of HB-EGF protein in both lysate and condi-tioned medium, indicating intact membrane routing andproteolytic cleavage of the ligand proprotein at the cell surface(Fig. 6A).

We characterized the intrinsic sensitivity of the H322M-HB-EGF cell line to saturating concentrations (1 mmol/L) of cetux-imab, panitumumab, Sym004 (analogue), and MM-151 under

Kearns et al.





endogenous ligand conditions in a cell proliferation assay. Whileboth the parental and GFP-expressing control cell lines werehighly sensitive to all inhibitors, autocrine HB-EGF expressionrendered H322M cells largely resistant to cetuximab and panitu-mumab, while sensitivity to Sym004 was reduced. In contrast,MM-151 maintained its antiproliferative activity in the HB-EGF–expressing cell line (Fig. 6B).

We next compared the in vivo efficacy of cetuximab and anMM-151–equivalent tool compound ("25E-Trio") againstH322M-HB-EGF xenograft tumors. Remarkably, while all threedose levels of cetuximab elicited modest inhibition of tumorgrowth relative to PBS control, 25E-Trio exhibited rapid andsustained tumor regression at all dose levels (Fig. 6C). Thisresult suggests that treatment with an oligoclonal mixture ofpotent ligand antagonists is able to overcome high-affinity

ligand-mediated resistance to monoclonal EGFR antibodies intumors.

DiscussionA rich body of literature has described how a complex network

involving ligand redundancy, receptor homo- and heterodimers,signal amplification, and feedback interactions between effectorproteins controls signal transduction through the EGFR:ERKpathway (1–3, 11, 36, 37). Strikingly, few studies have investi-gated how EGFR network complexity mediates sensitivity orresistance to EGFR antibody therapeutics. The principal findingof this study is that the high-affinity EGFR ligands, EGF, HB-EGF,BTC, and TGFa, initiate EGFR signaling to ERK and consequentcellular proliferation that is largely resistant to antagonism by

B

No drug

Cetuximab

MM-151

Cetuximab

MM-151

EREG0.0

1.0

0.5

TGFα0.0

1.0

0.5

CRC HNSCC NSCLC

CCK-81

Lim12

15

CAR-1FaD

u

SCC-4

SCC-25

H322M

H358

SK-MES-1

Cel

l via

bilit

y,fr

actio

n of

no

drug

HB-EGF0.0

1.0

0.5

C


Cel

l via

bilit

y,fr

actio

n of

no

drug

−10 −8 −6 −10 −8 −6 −10 −8 −6 −10 −8 −6−10 −8 −6

18.2+0 18.2+0.0323 18.2+0.323 18.2+3.23 18.2+32.3

0.25

0.50

0.75

1.00

DMM-151Cetuximab

H322M(NSCLC)T

GF

α

SCC-25(HNSCC)T

GF

α

CCK-81(CRC)

HB

-EG

F

EREG

EREG

EGF

Ligand notblocked

Ligand fullyblocked

Scaled cell viability0 0.5 1

AREG+ EGF (nmol/L)

Media controlLigand control

MM-151

CetuximabPanitumumabSym004

0.4

0.6

0.8

1.0

1.2

0.4

0.6

0.8

1.0

1.2

Cel

l via

bilit

y,fr

actio

n +

Liga

nd c

ontr

ol

−10 −8 −7−9 −6 −8 −7−9 −6

Total antibody, Log(Molar) Total antibody, Log(Molar)

ALIM1215 (CRC)H322M (NSCLC)

Figure 5.MM-151 blocks tumor cell proliferation more effectively than cetuximab across cancer indications and ligand environments. Relative cell viabilities were measured inmultispheroid culture after 3-daydrug/ligand treatments. A, NCI-H322Mcellswere treatedwith 8 nmol/L of EGF, andLIM1215 cellswere stimulatedwith0.8 nmol/L ofHB-EGF. Each cell line was simultaneously treated with dose-titrations of cetuximab, panitumumab, Sym004 (analogue), or MM-151, or medium alone. B, ninerepresentative cell lines from three EGFR-related indications—KRAS wild-type colorectal cancer (CRC), HNSCC, and NSCLC—were treated with medium alone (nodrug), cetuximab or MM-151 in the presence of 10 nmol/L EREG (100 nmol/L inhibitors), 10 nmol/L TGFa (1 mmol/L inhibitors), or 1 nmol/L HB-EGF (1 mmol/Linhibitors). C, NCI-H332M cells were treatedwith MM-151 or cetuximab in the presence of a combination of the low- and high-affinity ligands AREG (18.2 nmol/L) andEGF (10-fold titration series, as indicated). The stippled line indicates cell viability in the absence of exogenous ligand. D, three representative cell lines (CCK-81,colorectal cancer; SCC-25, HNSCC; NCI-H322M, NSCLC) were treated with MM-151 or cetuximab in the presence of pairwise dose titrations of the two ligands mostexpressedwithin each indication (titrations of each ligand alone are shown in the small adjacent panels). Ligandswere titrated from 10 pmol/L to 31.6 nmol/L in semi-log intervals (10, 31.6, 100, 316, 1,000, 3,160, 10,000, 31,600 pmol/L).






monoclonal anti-EGFR antibodies such as cetuximab (Fig. 1).Ourcomprehensive analysis of how both low- and high-affinity EGFRligands and ligand mixtures impact monoclonal antibody antag-onism confirms and extends prior work with single ligands andcetuximab (19–21).

To overcome resistance to EGFR-directed antibodies, we devel-oped MM-151, a next-generation oligoclonal antibody therapeu-tic that overcomes signal amplification driven by both low- andhigh-affinity EGFR ligands. Relative to monoclonal anti-EGFRantibodies, preclinical characterization of MM-151 showssuperior signal inhibition (Fig. 4), antiproliferative activity(Fig. 5), and inhibition of tumor growth (Fig. 6) againsthigh-affinity ligands. We further demonstrate that MM-151additionally elicits receptor downregulation (SupplementaryFig. S8) and immune-effector (Supplementary Fig. S9) activitiesthat are each superior to that of monoclonal EGFR antibodies.The potential clinical relevance of our findings to therapeutictargeting of EGFR was demonstrated via expression analyses(Fig. 2) which showed that primary tumors express multipleEGFR ligands, including high-affinity ligands potentially capa-ble of inducing resistance to monoclonal anti-EGFR antibodies.We hypothesize that high-affinity EGFR ligand expression maybe a predictive response marker that distinguishes MM-151from other anti-EGFR therapeutics.

Our results build on and contribute to the emerging field ofoligoclonal inhibitors of ErbB and other cell-surface receptors(51–53), which notably includes the approved combination oftrastuzumab (Herceptin) and pertuzumab (Perjeta) targetingErbB2 in metastatic breast cancer (57) and the early-stage clinicaldevelopment of Sym004 targeting EGFR (5, 6). Our observationsof near-complete downregulation of EGFR and enhanced CDC

activity by MM-151 are consistent with these studies and providean exciting prospect to elicit additional EGFR inhibition andpathway-extrinsic cytotoxicity beyond that achievable by singlemonoclonal antibodies.

The differential cell proliferation and EGFR downregulationresponses betweenMM-151 and Sym004 (analogue) suggest thatthe nascent development of oligoclonal inhibitors is ripe for asystem engineering approach to design optimal therapeutics toachieve both pathway inhibition and cell-extrinsic activities.Indeed, a combination of three anti-ErbB2 antibodies is under-development to achieve improved activity beyond that achiev-able by two-antibody combinations (58). Future developmentin this field must overcome the challenges of characterizing andaccounting for complexities beyond simply the specificity andaffinity of a monoclonal antibody to its receptor—includingdevelopment of relevant in vitro and in vivo model systems thatassess multiple mechanisms of action simultaneously insteadof in isolation. We posit that successful development of oligo-clonal antibody inhibitors will require a more forward-lookingand complete understanding of preclinical, translational, andclinical issues early on in the development process, even beforethe lead selection phase.

Overall, the data presented herein demonstrate that MM-151elicits three mechanisms of action—ligand antagonism, receptordownregulation, and immune-effector function—that are eachsuperior to that of existing single EGFR antibodies. Based on thesepreclinical studies, MM-151 has advanced to clinical testing, andphase I results todatedemonstrate an acceptable safety profile andobjective clinical activity in refractory cancer patients, includingthose failing cetuximab therapy (ClinicalTrials.gov Identifier:NCT01520389; ref. 59).

A

B

C

0.0

0.5

1.0

Parental GFP HB-EGF

Cel

l via

bilit

y,F

ract

ion

of n

o dr

ug

In vitro phenotype

Basal HB-EGF expression

No drugCetuximabPanitumumabSym004MM-151

c(H

B-E

GF

) (p

g/µ

g to

tal p

rote

in)

Protein in lysate

c(H

B-E

GF

)(n

g/m

L)Par

enta

lGFP

HB-EGF

Protein in medium

Xenograft tumor growth (H322M - HB-EGF)

0

100

200

300

400

500

600

700

800

900

10 14 16 18 21 23 25

Tum

or v

olum

e (m

m3 )

Days after cell injection

PBS

Cetuximab

25E-Trio (MM-151 tool)

6.25 12.5 25.0Dose levels, mg/kg {

0

2

4

Paren

tal

GFP

HB-EGF

0

1

2

3

Figure 6.MM-151 is superior to cetuximab at blocking tumor growth of a high-affinity EGFR ligand–expressing murine xenograft model. NCI-H322M NSCLC cells wereengineered to stably express full-length humanHB-EGF. The resulting cell line ("H322M-HB-EGF")was then characterized in relation to parental andGFP-expressingH322M cells. A, HB-EGF protein expression was measured in both lysates and cell-free conditioned medium of H322M-HB-EGF cells and controls using an ELISAassay. B, H322M-HB-EGF and control cell lines were treated with 1 mmol/L each of cetuximab, panitumumab, Sym004 (analogue), or MM-151, or medium only for 72hours in multispheroid culture, and relative cell viabilities were then measured. C, xenograft tumors of H322M-HB-EGF cells were established in immune-compromisedmice and treated with PBS, indicated doses of cetuximab, or indicated cetuximab-equivalent doses of 25E-Trio (MM-151 tool compound, see Materialsand Methods), and tumor volumes were measured by calipers.

Kearns et al.





Disclosure of Potential Conflicts of InterestJ.B. Fitzgerald and U.B. Nielsen have ownership interest (including patents)

in Merrimack Pharmaceuticals. No potential conflicts of interest were disclosedby the other authors.

Authors' ContributionsConception and design: J.D. Kearns, R. Bukhalid, M. Sevecka, G. Tan, N. Kohli,J.B. Fitzgerald, U.B. Nielsen, B.B. WolfDevelopment of methodology: J.D. Kearns, R. Bukhalid, M. Sevecka, G. Tan,N. Gerami-Moayed, N. Kohli, J.B. Fitzgerald, B.B. WolfAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): M. Sevecka, G. Tan, N. Gerami-Moayed, S.L. Werner,N. Kohli, C.M. Sloss, A.M. KingAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): J.D. Kearns, M. Sevecka, S.L. Werner, N. Kohli,C.M. Sloss, B.B. WolfWriting, review, and/or revision of the manuscript: J.D. Kearns, M. Sevecka,N. Gerami-Moayed, S.L. Werner, A.M. King, U.B. Nielsen, B.B. Wolf

Administrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): J.D. Kearns, O. Burenkova, B.B. WolfStudy supervision: J.D. Kearns, R. Bukhalid, B.B. Wolf

AcknowledgmentsThis work would not have been possible without the efforts of numerous

current and former colleagues at Merrimack. The authors thank Mike Briskin,Birgit Schoeberl, DaneWittrup, and CemElbi for their guidance and leadership,Brian Harms and Jaeyeon Kim for their computational modeling efforts duringthe early development ofMM-151, andClaire Bonal for technical assistance. Thecolleagues who generated the MM-151 candidate and lead antibodies, espe-cially the protein expression and process development groups atMerrimack andcolleagues at Adimab LLC, deserve enthusiastic thanks.

The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received September 10, 2014; revised April 2, 2015; accepted April 17, 2015;published OnlineFirst April 24, 2015.

References1. Yarden Y. The EGFR family and its ligands in human cancer. signalling

mechanisms and therapeutic opportunities. Eur J Cancer 2001;37 Suppl 4:S3–8.

2. Dhomen NS, Mariadason J, Tebbutt N, Scott AM. Therapeutic targeting ofthe epidermal growth factor receptor in human cancer. Crit Rev Oncog2012;17:31–50.

3. Yarden Y, Pines G. The ERBB network: at last, cancer therapymeets systemsbiology. Nat Rev Cancer 2012;12:553–63.

4. Dienstmann R, De Dosso S, Felip E, Tabernero J. Drug development toovercome resistance to EGFR inhibitors in lung and colorectal cancer. MolOncol 2012;6:15–26.

5. Koefoed K, Steinaa L, Søderberg JN, Kjær I, Jacobsen HJ, Meijer P-J, et al.Rational identification of an optimal antibody mixture for targeting theepidermal growth factor receptor. MAbs 2011;3:584–95.

6. Pedersen MW, Jacobsen HJ, Koefoed K, Hey A, Pyke C, Haurum JS, et al.Sym004: a novel synergistic anti-epidermal growth factor receptorantibody mixture with superior anticancer efficacy. Cancer Res 2010;70:588–97.

7. Schaefer G, Haber L, Crocker LM, Shia S, Shao L, Dowbenko D, et al. Atwo-in-one antibody against HER3 and EGFR has superior inhibitoryactivity compared with monospecific antibodies. Cancer Cell2011;20:472–86.

8. Gerdes CA, Nicolini VG,Herter S, Van Puijenbroek E, Lang S, RoemmeleM,et al. GA201 (RG7160): a novel, humanized, glycoengineered anti - EGFRantibody with enhanced ADCC and superior in vivo efficacy comparedwith cetuximab. Clin Cancer Res 2013;19:1126–38.

9. Jones JT, AkitaRW, SliwkowskiMX. Binding specificities and affinities of egfdomains for ErbB receptors. FEBS Lett 1999;447:227–31.

10. Kochupurakkal BS, Harari D, Di-Segni A, Maik-Rachline G, Lyass L, GurG, et al. Epigen, the last ligand of ErbB receptors, reveals intricaterelationships between affinity and mitogenicity. J Biol Chem 2005;280:8503–12.

11. Harris RC, Chung E, Coffey RJ. EGF receptor ligands. Exp Cell Res 2003;284:2–13.

12. Sanders JM, Wampole ME, Thakur ML, Wickstrom E. Molecular determi-nants of epidermal growth factor binding: a molecular dynamics study.PLoS One 2013;8:e54136.

13. Troyer KL, Luetteke NC, Saxon ML, Qiu TH, Xian CJ, Lee DC. Growthretardation, duodenal lesions, and aberrant ileumarchitecture in triple nullmice lacking EGF, amphiregulin, and TGF-alpha. Gastroenterology 2001;121:68–78.

14. Luetteke NC, Qiu TH, Fenton SE, Troyer KL, Riedel RF, Chang A, et al.Targeted inactivation of the EGF and amphiregulin genes reveals distinctroles for EGF receptor ligands in mouse mammary gland development.Development 1999;126:2739–50.

15. Yarden Y, SliwkowskiMX.Untangling the ErbB signalling network. Nat RevMol Cell Biol 2001;2:127–37.

16. Roepstorff K, Grandal MV, Henriksen L, Knudsen SLJ, Lerdrup M, GrøvdalL, et al. Differential effects of EGFR ligands on endocytic sorting of thereceptor. Traffic 2009;10:1115–27.

17. Henriksen L, Grandal MV, Knudsen SLJ, van Deurs B, Grøvdal LM.Internalization mechanisms of the epidermal growth factor receptor afteractivation with different ligands. PLoS One 2013;8:e58148.

18. Wilson KJ, Mill C, Lambert S, Buchman J, Wilson TR, Hernandez-GordilloV, et al. EGFR ligands exhibit functional differences in models of paracrineand autocrine signaling. Growth Factors 2012;30:107–16.

19. Hatakeyama H, Cheng H, Wirth P, Counsell A, Marcrom SR, Wood CB,et al. Regulation of heparin-binding EGF-like growth factor by miR-212and acquired cetuximab-resistance in head and neck squamous cell car-cinoma. PLoS One 2010;5:e12702.

20. Troiani T, Martinelli E, Napolitano S, Vitagliano D, Ciuffreda LP, Costan-tino S, et al. Increased TGF-a as a mechanism of acquired resistance to theanti-EGFR inhibitor cetuximab through EGFR-MET interaction and acti-vation of MET signaling in colon cancer cells. Clin Cancer Res 2013;19:6751–65.

21. Hobor S, Van Emburgh BO, Crowley E, Misale S, Di Nicolantonio F,Bardelli A. TGF and amphiregulin paracrine network promotes resistanceto EGFR blockade in colorectal cancer cells. Clin Cancer Res 2014;20:6429–38.

22. Liska D, Chen C-T, Bachleitner-Hofmann T, Christensen JG, Weiser MR.HGF rescues colorectal cancer cells from EGFR inhibition via MET activa-tion. Clin Cancer Res 2011;17:472–82.

23. Yamada T, Takeuchi S, Kita K, Bando H, Nakamura T, Matsumoto K, et al.Hepatocyte growth factor induces resistance to anti-epidermal growthfactor receptor antibody in lung cancer. J Thorac Oncol 2012;7:272–80.

24. Yonesaka K, Zejnullahu K, Lindeman N, Homes AJ, Jackman DM, Zhao F,et al. Autocrine production of amphiregulin predicts sensitivity to bothgefitinib and cetuximab in EGFR wild-type cancers. Clin Cancer Res2008;14:6963–73.

25. Ashraf SQ, Nicholls AM, Wilding JL, Ntouroupi TG, Mortensen NJ,Bodmer WF. Direct and immune mediated antibody targeting of ERBBreceptors in a colorectal cancer cell-line panel. Proc Natl Acad Sci U S A2012;109:21046–51.

26. Baker JB, Dutta D, Watson D, Maddala T, Munneke BM, Shak S, et al.Tumour gene expression predicts response to cetuximab in patientswith KRAS wild-type metastatic colorectal cancer. Br J Cancer2011;104:488–95.

27. Jonker DJ, Karapetis CS, Harbison C, O'Callaghan CJ, Tu D, Simes RJ, et al.Epiregulin gene expression as a biomarker of benefit from cetuximab in thetreatment of advanced colorectal cancer. Br J Cancer 2014;110:648–55.

28. Khambata-Ford S, Garrett CR, Meropol NJ, Basik M, Harbison CT, Wu S,et al. Expression of epiregulin and amphiregulin and K-ras mutation statuspredict disease control in metastatic colorectal cancer patients treated withcetuximab. J Clin Oncol 2007;25:3230–7.






29. Jacobs B, De Roock W, Piessevaux H, Van Oirbeek R, Biesmans B, DeSchutter J, et al. Amphiregulin and epiregulinmRNA expression in primarytumors predicts outcome in metastatic colorectal cancer treated withcetuximab. J Clin Oncol 2009;27:5068–74.

30. Pentheroudakis G, Kotoula V, De Roock W, Kouvatseas G, Papakostas P,Makatsoris T, et al. Biomarkers of benefit from cetuximab-based therapy inmetastatic colorectal cancer: interaction of EGFR ligand expression withRAS/RAF, PIK3CA genotypes. BMC Cancer 2013;13:49.

31. Yoshida M, Shimura T, Sato M, Ebi M, Nakazawa T, Takeyama H, et al. Anovel predictive strategy by immunohistochemical analysis of four EGFRligands in metastatic colorectal cancer treated with anti-EGFR antibodies. JCancer Res Clin Oncol 2013;139:367–78.

32. Han S-W,OhD-Y, Im S, Park SR, Lee K-W, SongHS, et al. Phase II study andbiomarker analysis of cetuximab combined with modified FOLFOX6 inadvanced gastric cancer. Br J Cancer 2009;100:298–304.

33. Mutsaers AJ, Francia G, Man S, Lee CR, Ebos JML, Wu Y, et al. Dose-dependent increases in circulating TGF-alpha andother EGFR ligands act aspharmacodynamic markers for optimal biological dosing of cetuximaband are tumor independent. Clin Cancer Res 2009;15:2397–405.

34. Vollebergh MA, Kappers I, Klomp HM, Buning-Kager JC, Korse CM,Hauptmann M, et al. Ligands of epidermal growth factor receptor andthe insulin-like growth factor family as serum biomarkers for response toepidermal growth factor receptor inhibitors in patients with advanced non-small cell lung cancer. J Thorac Oncol 2010;5:1939–48.

35. Harms BD, Kearns JD, Su SV, KohliN,NielsenUB, Schoeberl B.Optimizingproperties of antireceptor antibodies using kinetic computational modelsand experiments. Methods Enzymol 2012;502:67–87.

36. Schoeberl B, Eichler-Jonsson C, Gilles ED, M€uller G. Computationalmodeling of the dynamics of the MAP kinase cascade activated by surfaceand internalized EGF receptors. Nat Biotechnol 2002;20:370–5.

37. Resat H, Ewald JA, Dixon DA,Wiley HS. An integrated model of epidermalgrowth factor receptor trafficking and signal transduction. Biophys J2003;85:730–43.

38. Bairy S, Wong CF. Influence of kinetics of drug binding on EGFR signaling:a comparative study of three EGFR signaling pathway models. Proteins2011;79:2491–504.

39. Albeck JG,Mills GB, Brugge JS. Frequency-modulated pulses of ERK activitytransmit quantitative proliferation signals. Mol Cell 2013;49:249–61.

40. SturmOE,OrtonR,Grindlay J, BirtwistleM, VyshemirskyV,GilbertD, et al.The mammalian MAPK/ERK pathway exhibits properties of a negativefeedback amplifier. Sci Signal 2010;3:ra90.

41. Yang X-D, Jia X-C, Corvalan JRF, Wang P, Davis CG. Development of ABX-EGF, a fully human anti-EGF receptor monoclonal antibody, for cancertherapy. Crit Rev Oncol Hematol 2001;38:17–23.

42. Ramakrishnan MS, Eswaraiah A, Crombet T, Piedra P, Saurez G, Iyer H,et al. Nimotuzumab, a promising therapeuticmonoclonal for treatment oftumors of epithelial origin. MAbs 2009;1:41–8.

43. Vincenzi B, Schiavon G, Silletta M, Santini D, Tonini G. The biologicalproperties of cetuximab. Crit Rev Oncol Hematol 2008;68:93–106.

44. Tan AR, Moore DF, Hidalgo M, Doroshow JH, Poplin EA, Goodin S, et al.Pharmacokinetics of cetuximab after administration of escalating singledosing and weekly fixed dosing in patients with solid tumors. Clin CancerRes 2006;12:6517–22.

45. Cohen MH, Chen H, Shord S, Fuchs C, He K, Zhao H, et al. Approvalsummary: cetuximab in combination with Cisplatin or Carboplatin and 5-Fluorouracil for the first-line treatment of patients with recurrent locor-egional or metastatic squamous cell head and neck cancer. Oncologist2013;18:460–6.

46. Argiles G,Dienstmann R, Elez E, Tabernero J. Panitumumab: a summary ofclinical development in colorectal cancer and future directions. FutureOncol 2012;8:373–89.

47. Broadbridge VT, Karapetis CS, Price TJ. Cetuximab in metastatic colorectalcancer. Expert Rev Anticancer Ther 2012;12:555–65.

48. Katoh Y, Katoh M. Canonical WNT signaling pathway and human AREG.Int J Mol Med 2006;17:1163–6.

49. Oliveras-Ferraros C, Cufí S, Queralt B, Vazquez-Martin A, Martin-CastilloB, de Llorens R, et al. Cross-suppression of EGFR ligands amphiregulin andepiregulin and de-repression of FGFR3 signalling contribute to cetuximabresistance in wild-type KRAS tumour cells. Br J Cancer 2012;106:1406–14.

50. Fitzgerald JB, Schoeberl B, Nielsen UB, Sorger PK. Systems biology andcombination therapy in the quest for clinical efficacy. Nat Chem Biol2006;2:458–66.

51. Ben-Kasus T, Schechter B, Lavi S, Yarden Y, SelaM. Persistent elimination ofErbB-2/HER2-overexpressing tumors using combinations of monoclonalantibodies: relevance of receptor endocytosis. Proc Natl Acad Sci U S A2009;106:3294–9.

52. Spangler JB, Neil JR, Abramovitch S, Yarden Y, White FM, LauffenburgerDA, et al. Combination antibody treatment down-regulates epidermalgrowth factor receptor by inhibiting endosomal recycling. Proc Natl AcadSci U S A 2010;107:13252–7.

53. Friedman LM, Rinon A, Schechter B, Lyass L, Lavi S, Bacus SS, et al.Synergistic down-regulation of receptor tyrosine kinases by combinationsofmAbs: implications for cancer immunotherapy. Proc Natl Acad Sci U S A2005;102:1915–20.

54. DechantM,WeisnerW, Berger S, PeippM, Beyer T, Schneider-Merck T, et al.Complement-dependent tumor cell lysis triggered by combinationsof epidermal growth factor receptor antibodies. Cancer Res 2008;68:4998–5003.

55. Fogler WE, Klinger MR, Abraham KG, Gottlinger HG, Riethmuller G,Daddona PE. Enhanced cytotoxicity against colon carcinoma by combina-tions of noncompeting monoclonal antibodies to the 17-1A antigen.Cancer Res 1988;48:6303–8.

56. Velders MP, van Rhijn CM, Oskam E, Fleuren GJ, Warnaar SO, Litvinov SV.The impact of antigen density and antibody affinity on antibody-depen-dent cellular cytotoxicity: relevance for immunotherapy of carcinomas. Br JCancer 1998;78:478–83.

57. Wong DJL, Hurvitz SA. Recent advances in the development of anti-HER2antibodies and antibody-drug conjugates. Ann Transl Med 2014;2:1–14.

58. Pedersen MW, Jacobsen HJ, Koefoed K, Dahlman A, Kjaer I, Poulsen TT,et al. Targeting three distinct HER2 domains with a recombinant antibodymixture overcomes trastuzumab resistance. Mol Cancer Ther 2015;14:669–80.

59. Lieu CH, BeeramM, HarbWA, Kearns JD, Sloss CM, Nering R, et al. Safety,pharmacology, and preliminary clinical activity ofMM-151: an oligocolnalanti-EGFR therapeutic in patients with cetuximab-resistant CRC and otherrefractory solid tumors. J Clin Oncol 2015;33 Suppl 3:abst 647.


Kearns et al.




2015;14:1625-1636. Published OnlineFirst April 24, 2015.Mol Cancer Ther Jeffrey D. Kearns, Raghida Bukhalid, Mark Sevecka, et al. Oligoclonal Anti-EGFR Antibody TherapeuticEnhanced Targeting of the EGFR Network with MM-151, an

Updated version

10.1158/1535-7163.MCT-14-0772doi:

Access the most recent version of this article at:

Material

Supplementary

http://mct.aacrjournals.org/content/suppl/2015/04/25/1535-7163.MCT-14-0772.DC1

Access the most recent supplemental material at:

Cited articles

http://mct.aacrjournals.org/content/14/7/1625.full#ref-list-1

This article cites 56 articles, 21 of which you can access for free at:

Citing articles

http://mct.aacrjournals.org/content/14/7/1625.full#related-urls

This article has been cited by 7 HighWire-hosted articles. Access the articles at:

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

Subscriptions

Reprints and

[email protected]

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

Permissions

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

.http://mct.aacrjournals.org/content/14/7/1625To request permission to re-use all or part of this article, use this link



http://mct.aacrjournals.org/lookup/doi/10.1158/1535-7163.MCT-14-0772

http://mct.aacrjournals.org/content/suppl/2015/04/25/1535-7163.MCT-14-0772.DC1

http://mct.aacrjournals.org/content/14/7/1625.full#ref-list-1

http://mct.aacrjournals.org/content/14/7/1625.full#related-urls

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

mailto:[email protected]

http://mct.aacrjournals.org/content/14/7/1625


enhanced targeting of the egfr network with mm-151, an...

Documents