JNK Pathway Activation Modulates Acquired Resistance to ......Bliss synergy calculation The...

Integrated Systems and Technologies

JNK Pathway Activation Modulates AcquiredResistance to EGFR/HER2–Targeted TherapiesSimin Manole, Edward J. Richards, and Aaron S. Meyer

Abstract

Resistance limits the effectiveness of receptor tyrosine kinase(RTK)-targeted therapies. Combination therapies targeting resis-tance mechanisms can considerably improve response, butwill require an improved understanding of when particular com-binations will be effective. One common form of resistance isbypass signaling, wherein RTKs not targeted by an inhibitor candirect reactivation of pathways essential for survival. Althoughthis mechanism of resistance is well appreciated, it is unclearwhich downstream signaling events are responsible. Here, weapply a combined experimental- and statistical modeling–basedapproach to identify a set of pathway reactivation essential for

RTK-mediated bypass resistance. Differences in the downstreampathway activation provided byparticular RTKs lead toqualitativedifferences in the capacity of each receptor to drive therapeuticresistance. We identify and validate that the JNK pathway isactivated during and strongly modulates bypass resistance. Theseresults identify effective therapeutic combinations that blockbypass-mediated resistance and provide a basic understandingof this network-level change in kinase dependence that willinform the design of prognostic assays for identifying effectivetherapeutic combinations in individual patients.CancerRes; 76(18);5219–28. �2016 AACR.

IntroductionTherapies targeting aberrant receptor tyrosine kinase (RTK)

signaling are effective in treating a subset of multiple malignan-cies, including breast carcinoma and lung adenocarcinoma (1, 2).Despite transient effectiveness, the resulting survival benefit ofthese therapies is limited by resistance mechanisms that allowtumor cells to escape the effect of therapy. Resistance etiologyvaries widely, including mutation of the drug target to block theeffect of therapy, amplification of the drug target to overcomeinhibition, pharmacokinetic barriers that block trafficking of drugto tumor cells, and "bypass" switching to alternative pathways nottargeted by therapy (3, 4). In the case of RTK-targeted therapies,many nontargeted RTKs may become activated to provide bypassresistance (5). Two well-studied combinations are the ability ofHER3 to provide resistance to HER2-targeted therapy in breastcarcinoma and the ability of Met to provide resistance to EGFR-targeted therapies in lung carcinoma (6–10). In each case theresistance-conferring receptors may contribute to innate or adap-tive resistance and can become activated by multiple means,including ligand-mediated autocrine or paracrine induction,amplification, or mutation (8, 11, 12).

Combination therapy can effectively combat resistance but willrequire accurate identification of relevant combinations for indi-vidual tumors. However, many receptors exist that can causeresistance and are activated through both tumor cell-intrinsic

and -extrinsic means. This dictates that a more fundamentalunderstanding of network-level bypass, and methods to identifyeffective combinations for individual tumors,will be required (6).An improved understanding of exactly which pathways mustbecome reactivated to provide resistancemay also identify widelyeffective therapeutic combinations not dependent upon the par-ticular RTK providing bypass signaling.

Here, we have undertaken a combined experimental andcomputational approach to understand bypass resistance. Weexamine four cell lines in which activation of a noninhibited RTKcan provide bypass resistance to develop a multipathway under-standing of the process (6). Through modeling and validationexperiments, we identify a core set of pathway activation thatdetermines whether cells are ultimately viable. Individual RTKsactivate these pathways to varying extents, and thus have varyingability to drive bypass resistance. This observation in turn explainswhy RTK expression alone poorly predicts resistance capacity. Wesubsequently apply this model to identify therapeutic combina-tions that can block bypass resistance from multiple drivingreceptors simultaneously. This information will be valuable fora basic understanding of bypass resistance, development of prog-nostic tools to identify resistance mechanism, and design ofeffective therapeutic combinations.

Materials and MethodsCell culture

BT474, SKBR3, and HCC827 were obtained from ATCC in2015, confirmed by STR profiling by the source, and have beenpassaged for fewer than 6months. PC9was obtained from Sigma-Aldrich in 2015, confirmedby STRprofiling by the source, andhasbeen passaged for fewer than 6 months. HOP-62, HOP-92,H322M, and H522 confirmed by STR profiling were obtainedfrom Merrimack Pharmaceuticals in 2016. BT474 and PC9 iden-tities were additionally confirmed in 2016 through RNA sequenc-ing experiments. BT474 and SKBR3 were grown in DMEM,whereas PC9, HOP-62, HOP-92, H322M, H522, and HCC827

Koch Institute for Integrative Cancer Research at MIT, Cambridge,Massachusetts

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

Corresponding Author: Aaron S. Meyer, Massachusetts Institute of Technology,77 Massachusetts Avenue, Cambridge, MA 02139. Phone: 617-324-4404;E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-16-0123

�2016 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 5219

on May 26, 2021. © 2016 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 22, 2016; DOI: 10.1158/0008-5472.CAN-16-0123

http://cancerres.aacrjournals.org/

were grown in RPMI1640, both supplemented with 10% FBS and1% penicillin/streptomycin, at 37�C and 5% CO2.

Viability measurements were performed using CellTiter Gloreagent according to the manufacturer's protocol (Promega).Cells were seeded in 96-well plates at a density of 200 per well,and treated the next day with the indicated GFs (Peprotech) andinhibitors (LC Labs). After 72 hours, CellTiter Glo reagent wasmixed with each well and luminescence was quantified.

Cell signaling and protein abundance quantitationCells were treated for 4 hours with the indicated GFs and

inhibitors in serum-free medium without prior stimulation, thenlysed in 10mmol/L Tris-HCl pH 8.0, 1mmol/L EDTA, 1% Triton-X 100, 0.1%Na deoxycholate, 0.1% SDS, and 140mmol/L NaCl,with protease and phosphatase inhibitor supplemented beforeuse (Bio-Rad). Protein concentration was normalized by bicinch-oninic acid assay.

A bead-based multiplexed ELISA-based assay was used forsignaling measurement (Bio-Rad). In all cases, pErk is Erk1/2(pT185/pY187, pT202/pY204), pMek is Mek pS217/221, pGSK3 isGSK3a/b pS21/9, pAkt is Akt pS473, pP38 is P38 pT180/pY182, pcJunis c-Jun pS63, pSTAT3 is STAT3 pY705, and pJNK is JNKpT183/pY185. RTK abundancewas quantified bymultiplexed ELISA(Millipore). Absolute quantification of receptor abundance wasperformed by comparison to recombinant standards of eachreceptor (R&D Systems). Lysates were incubated with beads over-night, and then the beads were washed with 0.1% (v/v) Tween-20in TBS. Detection antibody and streptavidin–phycoerythrin wereincubated for 60 and 10 minutes, respectively. Signal was quan-tified fromeachbead set usingaMagPix Luminex reader (Bio-Rad).

For quantification of total cJun, cells were treated and lysedidentically to those used in ELISA measurement. Lysates werenormalized according to protein concentration, and then sepa-rated by SDS-PAGE and transferred to a nitrocellulosemembrane.Eachmembrane was incubated overnight with antibodies againstcJun (Cell Signaling Technology, cat. no. 9165) or paxillin (BDBiosciences, cat. no. 610052).

Partial least squares regression and generalized linearmodeling

Partial least squares regression was performed using MatLab(MathWorks). Phosphosignaling time courses for each conditionwere summarized either by calculating the AUC via trapezoidalintegration, or the steady-state levels by averaging the 2- and 4-hour timepoints.Z-scored phosphosignalingmeasurementswereregressed against z-scored viability measurements. The SD of theloadings were calculated by jackknife, leaving out individual GF/drug combination treatments (13).

Generalized linear regression (GLM) models were developedrelating cell signaling as measured at 4 hours to viabilitymeasurements at 72 hours using R. Before regression, eachphosphorylation and viability measurement was z-score nor-malized. GLMmodels were fit using glm using the formula Viab� pAkt þ pErk þ pGSK þ pcJun þ pJNKþ pP38 þ pAkt� pErk.For sampling the model parameters posterior distribution,MCMCglmm from the package of the same name was usedwithout thinning, a burn-in of 1,000 iterations, and samplingover 10,000 iterations. Markov chain mixing was achievedaccording to the Geweke diagnostic. A default model prior ofeach parameter normally distributed around 0 with a varianceof 1 was used.

Bliss synergy calculationThe interaction between Mek and JNK inhibition was calcu-

lated according to the Bliss independence model. Viability mea-surements were normalized to the value with no inhibitor treat-ment and then inverted to be in the form of percent inhibition:

Ex;y ¼ 1� Vx;y=V0� �

where Ex;y is the percent inhibition effect of a particular combi-nation, Vx;y is the viability measurement for a particular combi-nation, and V0 is the viability measured in the absence ofinhibitor. Thus, for a condition where the measured viability was70%ofwhat wasmeasuredwithout inhibitor, this would becomean effect of 30%. The results of combination inhibitor treatmentwere then calculated by Bliss independence, using the equation:

Ex;y ¼ Ex þ Ey � ExEy

where Ex, Ey, and Ex;y are the effects of the individual inhibitorsand combination, respectively. For example, if one inhibitordecreased viability by 10% and the other by 20% when admin-istered alone, the predicted combined effect would be0:1þ 0:2� ð0:1� 0:2Þ ¼ 0:28, or 28%. The measured and pre-dicted effects were then plotted on an inverted scale, so thatpositive effects (inhibition of viability) were negative, to aidcomparison with the untransformed viability measurements. Theeffects and difference between that predicted and measured wereplotted on the same color scale for ease of comparison.

Flow analysisAXL and a kinase-dead variant K562R were amplified from

previously used vectors (14). Met and PDGFRb were amplifiedfrom pDONR223-PDGFRb and pDONR223-MET from WilliamHahn (Dana-Farber Cancer Institute, Boston, MA) and David Root(Broad Institute, Cambridge, MA; Addgene plasmids 23893 &23889; ref. 15). Each receptor was then cloned into IRES-EGFP2(Clontech). Cells were seeded densely on 10-cm plates, and thentransfected the next day with 5 mg of each plasmid with a cor-responding amount of 5-mL Lipofectamine 2000 in OptiMEMaccording to manufacturer's instructions. After 4 hours, the mediawere exchanged into full serum media lacking antibiotics.

For evaluation of resistance-mediated selection, the day aftertransfection, cells were split into six-well plates and, after adher-ing, were placed in serum-free media with inhibitor and GF asindicated. The next and following day, wells were trypsinized,spun down, and then resuspended in PBS. Fluorescence, forwardscatter, and side scatter were then immediately quantified on anAccuri C6 (BD Biosciences).

For evaluating receptor overexpression, cells were trypsinizedthe day after transfection and fixed in 4% PFA in PBS for 1 hour,blocked in Odyssey Blocking Buffer (LI-COR) for 1 hour, thenstained with antibodies against Met, AXL, or PDGFRb overnight.The next day, cells were repeatedly washed and stained with AlexaFluor 594–conjugated antimouse antibodies for 1 hour. Afteradditional washing, fluorescence, forward scatter, and side scatterwere immediately quantified on an Accuri C6 (BD Biosciences).

ResultsRTK expression is essential but not sufficient for bypassresistance

To better understand the process of bypass resistance andits relationship to signaling network state we first selected two

Manole et al.

Cancer Res; 76(18) September 15, 2016 Cancer Research5220




HER2-overexpressing breast carcinoma cell lines, both studiedbefore for their sensitivity to the HER2-targeted inhibitor lapati-nib and their capacity for bypass resistance in the presence ofHRG(6). We treated each cell line with 0, 1, or 5 mmol/L of lapatinibeither alone or in the presence of 50 ng/mL of different growthfactors (GF; Fig. 1A). Cells exhibited a dose-dependent decreasein viability measured at 72 hours strongly counteracted by simul-

taneous addition of HRG and partially by other GFs (Fig. 1B).To expand our perspective beyond HER2-dependent signalingdysregulation, we additionally selected two EGFR-dependentlung adenocarcinoma cell lines, PC9 and HCC827, that areaccordingly sensitive to the EGFR inhibitor erlotinib. We mea-sured viability with 0 or 1 mmol/L erlotinib in the presence of50 ng/mL of different GFs (Fig. 1C). Particular GFs counteracted

B

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Ce

llT

ite

r L

um

ine

sce

nce

(R

U)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Ce

llTit

er

Lu

min

esc

en

ce (

RU

)

Co

ntr

ol

EG

F

TGF

α

IGF

HR

G

HG

F

PD

GF

HB

EG

F

SKBR3 BT474

Co

ntr

ol

EG

F

TGF

α

IGF

HR

G

HG

F

PD

GF

HB

EG

F

PC9

0.0

0.5

1.0

1.5

0.0

0.5

1.0

1.5 HCC827C

o

ntr

ol

EG

F

IGF

HR

G

HG

F

PD

GF

FG

F

Co

ntr

ol

EG

F

IGF

HR

G

HG

F

PD

GF

FG

F

Ce

llTit

er

Lu

min

esc

en

ce (

RU

)

Ce

llTit

er

Lu

min

esc

en

ce (

RU

)

C

PC9

0 1 2 30

0.50

1.00

IGF1R

HER2-HRG

EGFR

MET

HER3-HRG

None

RTK (106/cell)

BT474

0 3 6 9 12 150.00

0.25

0.50

0.75

1.00

1.25

IGF1R

HER2

EGFRMET

HER3-HRG

RTK (106/cell)

HCC827

0 2.5 5.0 7.5 100

0.33

0.67

1.00

IGF1R

HER2-HRG

EGFR

MET

HER3-HRGNone

RTK (106/cell)

SKBR3

0 2 40.00

0.25

0.50

0.75

1.00

1.25

IGF1R

HER2

EGFRMET

HER3-HRG

RTK (106/cell)

Via

bil

ity

(FI)

Via

bil

ity

(FI)

Via

bil

ity

(FI)

Via

bil

ity

(FI)

ED

Correlation P < 0.05 Correlation n.s.

Lap

Erlot

4 6 8 10 12

MET Probe value

4 6 8 10 12

IGF1R Probe value

None

Partial

Full

None

Partial

GF HPC9

HCC827

SKBR3

BT474

RTK (cell-1)

Re

sist

an

ce p

rom

oti

on

0.25 0.5 1 2 4 8 16 320.25

0.5

1

2

4

8

16

HER2 HER3

MET EGFR

IGF1R

Time (hours)

No

rma

lize

d r

ece

pto

r

ab

un

da

nce

Control Erlotinib

PC9

103 104 105 106 107

0.0

0.5

1.0

PDGFRCorrelation P < 0.05 Correlation P < 0.05PDGFRPDGFR

PDGFRNone

None

EGFR

HER2

HER3

HER4

IGF1R

MET

FGFR1

PDGFR

EGF

TGFα

HRG

HBEGF

IGF

HGF

FGF

PDGF

PC9

HCC827BT474

SKBR3

Erlotinib

Lapatinib

Ligands

RTKsCells

Inhibitors

A

Lap

Erlot

Figure 1.

RTK abundance does not fully predict bypass signaling capacity. A, schematic of the relevant RTKs, GFs, cell lines, and inhibitors. B, luminescence-based measurementof SKBR3 and BT474 cell viability 72 hours after treatment with 0, 1, or 5 mmol/L lapatinib and 50 ng/mL of the indicated GFs. C, luminescence-based measurementof PC9 and HCC827 cell viability 72 hours after treatment with 0 or 1 mmol/L erlotinib and 50 ng/mL of the indicated GFs. Gray and red horizontal shaded regionsindicate SE of control conditions in the absence or presence of drug, respectively. Bar colors indicate conditions with partial (blue) or full (red) resistance. Green indicatesviability in the absence of inhibitor or GF. Error bars, SE of biological replicates (N ¼ 5). D and E, correlation between RTK abundance and capacity to providebypass resistancewith additionof the cognateGF. "None" point is viability in the presence of drugwithout exogenousGF. Correlation significance calculated using Studenttdistribution for a transformationofPearsoncorrelation.Gray arrow, erlotinib-inducedchangeobserved inErbB3abundance.F,normalized receptor expression inPC9cellsafter the indicated period of cell starvation in the presence or absence of 1 mmol/L erlotinib. Error bars, SE of biological replicates (N ¼ 2). Gray region, changes ofless than 2-fold.G, combined plot of RTK abundance and resistance promotion across four cell lines. The viabilitymeasurements for each cell linewere normalized such thatthe viability in the absence of inhibitor or GF was equal to 1.0 and the viability in the presence of inhibitor and absence of GF was equal to 0.0. Dotted line, threshold of2� 106 per cell.H, threshold effect for MET and IGF1R resistance promotion across cell lines. Probe values for gene expressionweremeasured previously (18). Each cell linewas binned into whether resistance occurred with cognate ligand treatment by (6). The cell lines and raw data used are listed in Supplementary Table S1.

JNK Modulates EGFR/HER2–Targeted Therapy–Acquired Resistance

www.aacrjournals.org Cancer Res; 76(18) September 15, 2016 5221




the erlotinib-induced decrease in viability in a consistent patternto that observed previously (6).

Because ofmanifold bypass possibilities among RTKs and theircritical function in diverse biological processes, prospective diag-nostic measurements are needed to identify which among manytargeted therapies would be most effective to apply in precisecombination to overcome bypass resistance. A first notion mightbe to measure and focus on treating those receptors that are mostabundant or most abundantly activated within a tumor (16).Therefore, we next sought to directly determine the relationshipbetween receptor abundance and capacity to provide bypassresistance. We measured the absolute abundance of seven RTKsacross four cell lines using a recombinant standard. RTK abun-dances varied across several orders of magnitude, and for manycases, the RTK that provided primary drug sensitivity was not themost abundant receptor (Supplementary Fig. S1A). To test therelationship between RTK abundance and resistance-promotingcapacity (viability with inhibitor and GF normalized to that withjust inhibitor), we plotted viability in the presence of inhibitoragainst the abundance of different GF-activated RTKs (Fig. 1Dand E). Each data point indicates RTK abundance matched to theviability promoting effect upon treatment with each cognate GF(Fig. 1A). The viability of cells in the absence of erlotinib orlapatinib was predominantly dependent on EGFR or HER2 sig-naling, respectively. Therefore, we plotted in green the abundanceof the inhibitor-targeted RTK and cell viability in the absence ofinhibitor. Although abundance significantly correlated with resis-tance capacity in the case of three cell lines, no significantcorrelation was observed for the HCC827 (Fig. 1D and E).Strikingly, in HER2-dependent cells, HER3-HRG was relativelypotent in its resistance-conferring capacity (Fig. 1D), while it wasless potent than predicted in EGFR-dependent cell lines (Fig. 1E).

We considered that drug treatment or sustained serumwithdrawalmay influenceRTKexpression and thus the correlationsweobserved.For example, ifHER2/HER3expression is lost byeithermanipulationin PC9 cells, the change would explain the relative resistance-pro-moting inefficiencyofHRG. Thus,we examined the influenceof eachover 24 hours in PC9 cells (Fig. 1F). Only HER3 changed inabundance more than two-fold, increasing up to seven-fold at 24hours. This change in fact exacerbates the discrepancy and removesthe correlationbetweenviability andabundance (Fig. 1E, tan line; r¼0.84,P < 0.05 to r¼ 0.62,P > 0.05). Althoughmoremodest in effect,EGFR andMET displayed a simultaneous decrease of roughly 2-foldinabundanceby24hourswith serumwithdrawal,or increaseof50%in the presence of 1 mmol/L erlotinib (Supplementary Fig. S1B).Therefore, inhibitor-induced changes in RTK abundance onlyreduced the correlation between RTK abundance and resistance-promoting capacity in the case of PC9 cells.

Despite mixed correlation between RTK abundance and resis-tance capacity within cell lines, we did observe a consistentthreshold effect across all four cell lines, wherein all receptorsthat promoted resistance more than 30% were more abundantthan 2 � 106 per cell (Fig. 1G). We defined resistance promotionas the GF-induced increase in viability, scaled between 0.0 and1.0, to compare across cell lines. To examine whether somethreshold of expression might be necessary but not sufficient forresistance, we explored the results of two previous studies exam-ining expression and ligand-induced resistance across a largenumber of cell lines (6, 17). Similarly, we observed that expres-sion below a certain level correctly predicted that ligand-inducedbypass resistance would not occur, but also that high receptor

expression could not predict resistance-promoting capacity (Fig.1H). Therefore, the abundance of a particular RTK is unlikely toaccurately predict whether it is driving resistance.

RTK overexpression identifies qualitative differences inresistance-promoting capacity

The four cell lines of this studywere selectedon thebasis of theirsensitivity to an RTK-targeted therapy and ability to displayresistance upon the addition of one or more GFs (6). Thesecriteriamay artificially select for an improved correlation betweenRTK abundance and resistance-conferring capacity because celllines lacking a resistance-promoting bypass receptor would nothave been included. Therefore, we took an orthogonal approach,overexpressing individual RTKs in PC9 cells and then determiningthe extent to which each receptor could promote resistance toerlotinib when highly abundant (Fig. 2).

Many cell lines exhibit clonal heterogeneity that is reflected intheir response to targeted inhibition (18, 19). To account for this,we took a short-term overexpression approach, transiently over-expressing RTKs within an IRES-EGFP vector, to minimize clonalselection within the population of cells (Fig. 2A). If overexpres-sion of a particular RTK improved the resistance capacity of thecells, we expected to observe selection of transiently transfectedcells in the presence of inhibitor. We first selected overexpressionof the RTK AXL as an orthogonal control, as the receptor's role inresistance to EGFR-targeted treatment, often without exogenousligand, is well appreciated (11, 20). Selection of transfected cellswas quantified by flow analysis (Fig. 2B). We observed selectionfor AXL-transfected cells in the presence of erlotinib, dependentupon kinase activity of the receptor, providing confidence in ourapproach (Fig. 2C).

Our previous observations indicate that Met is inefficient (pro-motes viability less than its abundancewould suggest) but capableof EGFR-associated bypass resistance (Fig. 1C and E). In contrast,in previous work across a panel of cell lines, PDGFRb was neverobserved to promote resistance, although very few cell linesshowed expression of the receptor (6). Overexpressing each ofthese receptors and then selecting among amixed populationwitherlotinib, we observed selection of Met-expressing cells in thepresence of erlotinib, but no selection of PDGFRb-positive cellseven when overexpressed (Fig. 2D and E). We verified as a controlthat PDGFRbwas indeedbeingoverexpressed inGFP-positive cells(Supplementary Fig. S2). Thus, receptor expression alone is insuf-ficient to predict mechanism of resistance, and receptors qualita-tively differ in their resistance-promoting capacity. Furthermore, inthis cell line, PDGFRb is incapable of promoting resistance at anyexpression level obtained by transient overexpression.

Conserved RTK-specific signaling exists despite differences inbypass capacity

To examine the role of downstream pathway activation, wemeasured a panel of phosphorylation sites for each of the four celllines, 4 hours after treatment with inhibitor and GFs, to capturethe signaling consequences of each treatment (Fig. 3A and B).Basic analysis of these signaling measurements captured knownpathway associations. For example, pAkt andpGSKwere clusteredtogether in three of the four cell lines. For SKBR3, BT474, andHCC827, resistant conditions clustered with those in the absenceof erlotinib (Fig. 3A and B). For PC9 cells, all of the conditionswith erlotinib clustered separately from those without (Fig. 3B).Overall clustering profiles were still quite distinct between cell

Manole et al.





lines, likely a reflection of different receptor and intracellularcomponent expression levels (Fig. 3A and B).

We applied GLM modeling, regressing our phosphorylationmeasurements against viability, to understand which phosphor-ylation sites could predict the particular GF conditions leading toresistance. To capture dependencies on coordinate pathway acti-vation, we included an interaction termof Akt and Erk.We electedto not include other interaction terms, or more complex relation-ships, due to less prior evidence implicating the other pathwaysmeasured. In the EGFR-dependent cell lines PC9 and HCC827,pcJun (and pErk for HCC827) was identified as particularlyimportant (Fig. 3C and Supplementary S3). A model combiningmeasurements from both EGFR-dependent cell lines, with onlysignificantly nonzero parameters shown, and 95% confidenceintervals shown in parentheses, was: Viability ¼ 0.17 (0.01 �0.33) pErk þ 0.75 (0.57 � 0.92) pcJun. Positive weighting ofpcJun and pErk suggests that further phosphorylation contributespositively to resistance (Fig. 3C and Supplementary S3). Incontrast, the regressionmodels for the HER2-dependent cell linesSKBR3andBT474positivelyweightedpAkt but negativelyweight-edpcJun (Fig. 3C andSupplementary S3), indicating greater pcJunmeasurement corresponded to conditions with less viability. Amodel combining measurements from both HER2-dependentcell lines, with only significantly nonzero parameters shown, and95% confidence intervals shown in parentheses, was Viability ¼1.00 (0.69� 1.32) Akt� 0.30 (0.51� 0.09) cJun. The parametersidentified as important by each model correlated with whichphosphorylation sites weremost inhibited by inhibitor treatment

(Fig. 3D, Wilcoxon signed-rank test, P < 0.01). Model parameterswere similar among but not between EGFR- or HER2-dependentcell lines, indicating that this influencedwhichpathway activationchanges were essential for resistance (Supplementary Fig. S3). Intotal, this regression modeling identified that multiple pathwayactivation measurements are essential to predicting resistanceconditions, and formed novel hypotheses as to the contributionof JNK pathway activation.

EGFR bypass resistance requires sustained JNK pathwaysignaling

Complex signaling dynamics can govern cellular phenotypicresponse to extracellular cues (21). Therefore, we wished toexamine whether the inclusion of pcJun in our models of EGFRinhibitor bypass resistance might be due to incompletely captur-ing the complete dynamic response to ligand and inhibitortreatment. To determine this, we measured a dense time-courseof pErk, pMek, pcJun, and pAkt response in PC9 cells withouterlotinib treatment, with simultaneous inhibitor treatment, orwith 4-hour pretreatment (Fig. 4A). These phosphosites werechosen as Akt and Mek/Erk reactivation are widely implicated inbypass resistance (6), and were the four phosphosites repeatedlyselected in models based on a single time point (Fig. 3C). Asexpected, the responses at shorter timescales were often distinctfrom the sustained responses observed. Similar to the measure-ments at 4 hours, these dynamic responses showed that eachpathway was inhibited by erlotinib treatment, usually quiterapidly. Although the responses to HGF stimulation changed

RTK-IRES-EGFP

A

0.0 0.5 1.0 1.5 2.00.0

0.5

1.0

1.5 AXL

kinase dead AXL

Days

Fra

ctio

n G

FP

+ vs.

No

Dru

g

Day 1 Day 20.0

0.5

1.0

1.5

Fra

ctio

n G

FP

+ v

s. N

o D

rug

Day 1 Day 20.0

0.5

1.0

1.5

2.0

2.5

Fra

ctio

n G

FP

+ v

s. N

o D

rug

Log10

(FSC)

Log

10(G

FP

/ S

SC

)

–4

–3

–2

–1

0

1

2

Log10

(FSC)

Log

10(G

FP

/ S

SC

)

–4

–3

–2

–1

0

1

2

54 76 54 76

AXL

Inhibitor

B

C ED

cMETcMET/erlotinibcMET/erlotinib/HGF

PDGFRbPDGFRb/erlotinibPDGFRb/erlotinib/PDGF

7

GFP+

GFP-

GFP+

GFP-

AXL + erlotinib

Figure 2.

Validation of variation in RTK bypass resistance capacity. A, schematic of the selection assay to quantify resistance modulation. B, plot of FACS quantificationfor AXL-transfected PC9 cells 60 hours after transfection and 48 hours after addition of 1 mmol/L erlotinib. Gating used for EGFP expressing or nonexpressingcells is indicated. FSC, SSC, and GFP indicate forward scatter, side scatter, and GFP fluorescence values, respectively. The number of GFPþ cells was matchedto aid visualization of GFP� population depletion. C, quantification of selection for AXL-expressing PC9 cells in the presence of erlotinib with or withoutAXL kinase activity. D, quantification of selection for Met-overexpressing PC9 cells in the presence of erlotinib. E, quantification of selection for PDGFRb-overexpressing PC9 cells in the presence of erlotinib. Error bars, SEM (N ¼ 3).






little with addition of erlotinib, the bypass effects of other GFsweremore complex. For example, additionof IGFentirely restoredpAkt levels, but not pMEK or pcJun. FGF restored pcJun levels, butnot pAkt or pMEK. Also, although HRG strongly activated allpathways in the absence of erlotinib, its response was dependentupon EGFR activity. These differences all highlight a potentialsource of variation in the resistance capacity of particular RTKs.

To examine the relationship between pcJun, Erk pathwayactivation, and bypass resistance over differing timescales,we plotted both pathways with viabilitymeasurements from eachcondition (Fig. 4B and C). Regardless of whether pMek or pErkmeasurements were used, conditions separated by their eventualviability only at later time points (Fig. 4B and C; 2–4 hours). Inboth cases pcJun was required for separation, again demonstrat-ing that pcJun is required to predict viability. To further evaluatethe contribution of signaling at different timescales, we summa-rized the response of each condition using the AUC of eachphosphorylation time course, capturing early response, as wellas the sustained (>2 hours) measurement. Using partial least

squares regression, models could accurately predict the responseof PC9 cells with the sustained (>2 hours) measurement (Fig. 4Dand E). pcJun was required for predictive capacity, but all earliertime points (AUC) were dispensable (Fig. 4D). Sustained pMek/pErk was positively weighted with viability along principal com-ponent 1, whereas pcJun was weighted with viability along bothprincipal components (Fig. 4E), consistent with our earliermodeling (Fig. 3C). In total, these results indicate that measure-ment of sustained pcJun and Erk pathway activation is necessaryfor accurate prediction of bypass resistance in PC9 cells.

Validation of the bypass resistance model identifiescombination therapy approaches

With confidence that pcJun and Erk pathway activation werecoordinately necessary for predicting bypass resistance develop-ment in PC9 cells, we sought to determine whether this interac-tion might suggest effective therapeutic combinations. We firstmeasured the pathways influenced by treatment with a Mekinhibitor U0126 to ensure that pcJun was not simply responding

pG

SK

3α

/β

pA

kt

pcJ

un

pE

rk1

/2

pJN

K

pP

38

ControlIGFTGFαHBEGFEGFPDGFHGFHGFTGFαHRGIGFHRG

PDGF

EGF

Control

HBEGF

HGF

EGF

TGFα

PDGF

Control

IGF

HBEGF

ControlHBEGF

HGF

HRG

EGF

TGFα

IGF

HRG

PDGF

HRG

EGF

PDGF

FGF

Control

IGF

FGF

IGF

HGF

HGF

HRG

Control

PDGF

EGF

pcJ

un

pG

SK

3α

/β

pA

kt

pP

38

pJN

K

pE

rk1

/2

pA

kt

pcJ

un

pE

rk1

/2

pG

SK

3α

/β

pP

38

pJN

K

pA

kt

pG

SK

3α

/β

pcJ

un

pE

rk1

/2

pJN

K

pP

38

FGF

PDGF

Control

EGF

HRG

HGF

IGF

HGF

EGF

Control

PDGF

FGF

HRG

IGF

A BSKBR B3 T474 PC9 HCC827

3 2 1 0 –1 –2 –3

z-Score(Log2(Signal))

pErkpcJ

un

pP38pJN

K

–2.0

–1.5

–1.0

–0.5

0.0

0.5

SKBR3 (lapatinib)

BT474 (lapatinib)

PC9 (erlotinib)

HCC827 (erlotinib)Log

10(P

ho

sph

osi

te f

old

ch

an

ge

)

pAkt

pGSK3α/βD

pA

kt

pE

rk

pG

SK

3b

pcJ

un

pJN

K

pP

38

k

HCC827

PC9

BT474

SKBR3

HCC827/PC9

BT474/SKBR3

C

P < 0.05

P < 0.01

P < 0.001

Coefficient:Positive

Negative

Figure 3.

GLM identifies core bypass resistance pathway components. A, phosphorylation measurement of SKBR3 and BT474 cells at 4 hours after treatment with either0 (no bar) or 1 mmol/L (black bar) lapatinib and 50 ng/mL of the indicated GFs. B, phosphorylation measurement of PC9 and HCC827 cells at 4 hoursafter treatment with either 0 (no bar) or 1 mmol/L (black bar) erlotinib and 50 ng/mL of the indicated GFs. Conditions are colored according to the criteriain Fig. 1A and B. C, GLM models of signaling and viability in each and combinations of the cell lines. D, fold change in each phosphorylationmeasurement observed upon treatment with indicated inhibitor. Data are derived from A and B. Error bars indicate SEM propagated from each individualmeasurement in biological triplicate.

Manole et al.





to Erk activation. Although Erk1/2 phosphorylation wascompletely abrogated by Mek inhibition, cJun phosphorylationwas onlymodestly reduced in PC9, and not at all inHCC827 (Fig.5AandSupplementary S4A). pcJunwasmost reliably and strongly

abrogated by JNK inhibition (Supplementary Fig. S4A); however,total cJun levels also varied, indicating this phosphorylationmeasurement is likely a measure of both protein abundance andphosphorylation changes (Supplementary Fig. S5). Thus, pcJun is

0

20

40

60

80

100

120

0

200

400

600

800

1,000

0

20

40

60

80

100

0 50 100 150 200

Time (min)

0

5

10

15

20

25

30

0 50 100 150 200

Time (min)0 50 100 150 200

Time (min)0 50 100 150 200

Time (min)0 50 100 150 200

Time (min)0 50 100 150 200

Time (min)

pA

KT

(A

U)

pM

EK

(A

U)

pcJ

un

(A

U)

pE

RK

1/2

(A

U)

EGF (No resistance) IGF (Partial resistance) HGF (resistance) FGF (resistance) PDGF (No resistance) HRGβ (No resistance)

0 500 1,000pMEK

0

10

20

30

40

50

60

pcJ

UN

Early (5–10 min)

0 500 1,000pMEK

0

10

20

30

40

50

60

pcJ

UN

Middle (15–60 min)

0 500 1,000pMEK

0

10

20

30

40

50

60

pcJ

UN

Late (2–4 hr)

0 5 10 15 20

pErk1/2

0

10

20

30

40

50

60

pcJ

UN

Early (5–10 min)

0 5 10 15 20

pErk1/2

0

10

20

30

40

50

60

pcJ

UN

Middle (15–60 min)

0 5 10 15 20

pErk1/2

0

10

20

30

40

50

60

pcJ

UN

Late (2–4 hr)

1 2 3Components

0

20

40

60

80

100

Pe

rce

nt

va

ria

nce

exp

lain

ed

in v

iab

ility

on

cro

ssv

alid

ati

on

Reduced models

–5 50

Principal component 1

–4

–2

0

2

4

Pri

nci

pa

l co

mp

on

en

t 2

Reduced model loading

Sustained pMek

Sustained pErk1/2

Sustained pcJun

A

B

C

AUC & sustainedJust sustained

Just AUC

AUC & sustained, no cJun

D

E

Viability

Viability

EGF

HRGβ

IGF

PDGF

FGF

HGF

None

ed

k

Sustained pAkt

No drug

Simultaneous

Pre-treated

Figure 4.

EGFR bypass resistance relies upon sustained JNK pathway signaling. A, time-course measurement (0, 5, 10, 30, 60, 120, and 240 minutes) of pathway activation inPC9 cells. The red, blue, and green lines indicate growth factor stimulation without drug, with simultaneous administration of 1 mmol/L erlotinib, or with4-hour drug pretreatment, respectively. B, pcJun and pMek with respect to viability at distinct timescales. Shapes indicate the administered growth factor, andcolors indicate the viability measurement of the corresponding treatment. C, pcJun and pErk1/2 with respect to viability at distinct timescales. D, percentvariance in viability explained by reduced PLSR models based on area under the curve or sustained pathway measurement. The sustained measurement wascalculated as the mean of the 2- and 4-hour time points. E, loadings of each phosphorylation measurement (X) and viability (Y) for the two-component modelwith the sustained phosphosite measurements. Error bars, SD of each loading as calculated by bootstrap.






most prominently modulated by JNK activity in these cells but isalso influenced by Erk activity.

A first prediction of our model is that Erk and JNK inhibitionwould have additive effectiveness in decreasing PC9 viability (Fig.3D). Indeed, we observed that combination treatment with a JNK(SP600125) and Mek inhibitor coordinately reduced PC9 viabil-ity in the absence (Fig. 5B) or presence (Fig. 5C) of erlotinib. Weobserved the same coordinate reduction in viability with JNK andMek inhibition across a panel of HCC827 and four other lungcarcinoma cell lines, aswell aswith another JNK inhibitor JNK-IN-7 (Supplementary Fig. S4B and S4C). We used the Bliss indepen-dence model to examine the nature of combinationMek and JNKinhibition on viability (Supplementary Fig. S6A). This modelrecapitulated the results of the combination inhibitor treatments,consistent with an additive interaction between both inhibitors(Supplementary Fig. S6B and S6C).

Next, we predicted that bypass resistance viaMet is coordinatedthrough activation of JNK and Erk. Previous studies suggested that

activationof JNKbyMetoccurs via Src family kinases (22). Indeed,Met-mediated pcJun, but not pMek, was blocked by Src inhibition(dasatinib; Fig. 5D). Consistent with this and our model's pre-diction of the coordinate importance of Erk/JNK activation tobypass resistance, Src (Fig. 5E and Supplementary S4D) or JNK(Fig. 5F and Supplementary S4E) inhibition cooperated withMekinhibition to block increases in Met-induced viability.

In contrast to the case with EGFR-dependent cells, our modelsof signaling–viability relationship for HER2-dependent celllines indicated that JNK inhibition should in fact increase viability(Fig. 3D). To test this, wefirst examined the effect of JNK inhibitionon viability in the presence or absence of lapatinib, and as pre-dicted observed a dose-dependent increase (Fig. 5G). We won-dered if this effect could modulate the ability of particular GFs toprovide resistance in HER2-dependent cell lines. In particular,although lapatinib is thought to be an EGFR/HER2 dual-targetinginhibitor, we observed potent STAT3 phosphorylation suggestiveof EGFR activation with EGF treatment in each HER2-dependent

0.0 0.5 1 2 4 8 16 32 640

50

100

150

200

0 µmol/L U0126 (Mek inh.) 0 µmol/L U0126 (Mek inh.)3 µmol/L U0126

0 µmol/L U0126 (Mek inh.)3 µmol/L U0126


3 µmol/L U0126

SP600125 (JNK inh.) (µmol/L) SP600125 (JNK inh.) (µmol/L)

SP600125 (JNK inh.) (µmol/L)


Via

bil

ity

(LU

)

0

10

20

30

40

50

Via

bil

ity

(LU

)

0

1

2

3

0.0 0.5 1 2 4 8 160

1

2

3

Dasatinib (Src inh.) (nmol/L)

Fo

ld v

iab

ilit

y

incr

ea

se w

ith

HG

F

B C

ED

Fo

ld v

iab

ilit

y

incr

ea

se w

ith

HG

F

0.0 0.5 1 2 4 8 16 32 64

0.0 0.5 1 2 4 8 16 32 64

0

1

2

3

Ph

osp

ho

ryla

tio

n (

AU

)

0.0

0.5

1.0

1.5

Ph

osp

ho

ryla

tio

n (

AU

)

0.25 0.5 1 2 4 8 16 32 640.0

0.5

1.0

1.5

2.0

2.5

Lapatinib (L)

L + HRG

L + EGF

No

rma

lize

d v

iab

ilit

y

0.25 0.5 1 2 4 8 16 32 640

20,000

40,000

60,000

80,000


Via

bili

ty (

LU)

PC

9

AB

T4

74

Control

EGF

0 1.0SKBR3 pSTAT3 (AU)

Control

EGF

10 .0BT474 pSTAT3 (AU)

Contro

lHRG

EGF0.5

1248

163264

128pA

kt (A

U)

0 µmol/L Lapatinib1 µmol/L Lapatinib

F

HG JI


Control 1 μmol/L Erlotinib (E)E/HGF E/HGF/10 nmol/L Dasatinib (Srcinh.)

+ 1 µmol/L erlotinib

pAkt pMek pErk1/2 pcJun

pAkt pMek pcJun

0 µmol/L Lapatinib 1 µmol/L Lapatinib

0 μmol/L Lapatinib 1 μmol/L Lapatinib

Figure 5.

Model validation identifies points of effective combination therapy. A, pAkt, pMek, pErk1/2, and pcJun measurement of PC9 cells with 3 mmol/L U0126 treatment.B, viability measurement of PC9 cells with combination Mek (U0126) and JNK (SP600125) inhibition. C, viability measurement of PC9 cells with combinationMek and JNK inhibition in the presence of 1 mmol/L erlotinib. The y-axis is the same as that in B. D, pAkt, pMek, and pcJun measurement with erlotinib,erlotinib/HGF, and erlotinib/HGF/dasatinib treatment. E,measurement of the viability increase conferred by HGF stimulation in the presence of 1 mmol/L erlotinibcombined with Src/Abl (dasatinib) and Mek inhibition. F,measurement of the viability increase conferred by HGF stimulation in the presence of 1 mmol/L erlotinibcombined with Mek and JNK inhibition. G, viability measurement of BT474 cells with JNK inhibition in the presence and absence of 1 mmol/L lapatinib.H, pSTAT3 measurement in SKBR3 (top) and BT474 (bottom) cells with 1 mmol/L lapatinib and/or EGF treatment. I, pAkt measurement with EGF or HRG,with or without 1 mmol/L lapatinib treatment. J, measurement of the viability increase conferred by EGF or HRG in the presence of 1 mmol/L lapatinib combinedwith JNK inhibition. Error bars, SEM (N � 3).

Manole et al.





cell line, and an increase in STAT3 phosphorylation with lapatinibtreatment (Fig. 5H). This indicated that some amount of EGFRsignaling must be maintained in the presence of lapatinib. Withrespect to Akt phosphorylation, althoughmore modest than HRGtreatment, EGF treatment increased Akt phosphorylation roughly4-fold in the presence of lapatinib (Fig. 5I). We predicted that thisincrease in Akt phosphorylation may have more pronouncedeffects in the presence of JNK inhibition. Although EGF treatmentonly leads to a modest increase in viability in the absence of JNKinhibition, with JNK inhibition the GF was able to make BT474cells completely resistant to lapatinib (Fig. 1B and 5J). Thisindicates that JNK pathway activity strongly modulates the bypasssignaling response of cells, and that our model accurately capturesthese effects through measurement of pcJun.

DiscussionAlthough targeted therapies lead to survival benefits in patients,

thesebenefits areusually short-liveddue to resistancemechanisms.Many studies have focused on various mechanisms of resistancewith the notion that identifying particular resistance mechanismswould lead to widely efficacious treatment combinations. Thesehave left us with panoply mutations, expression changes, andtumor-extrinsic factors that can contribute to resistance, but littlein the way of a unified understanding of these processes.

Here, we have focused on a family of resistance mechanismsthat arise due to redundancy in the cellular signaling machinery.By a combined experimental and computational approach, weobserve that bypass resistance can be predicted by coordinatemeasurement of multiple pathways. The pathways that must bereactivated for resistance in each case, however, depend on whichRTK, and thus signals, have been therapeutically targeted. Becauseof the pathway dependencies, and particular RTKs being biased intheir relative pathway activation, not all RTKs are similar in theircapacity to provide resistance to a particular therapy (Figs. 1and 6). Previous works have similarly recognized that pathwayreactivation must occur for bypass resistance, but have notattempted to capture what those signaling changes might bebeyond Erk and Akt activation (5, 6, 11). Contribution of other

pathways such as JNKwould explainwhy Erk/Akt activation is notalways entirely predictive of resistance (6).

Our results indicate that the effects of JNK pathway activation,whether promoting or inhibiting resistance, depend upon thecellular context and/or the original RTK inhibited. The effects ofthe JNK pathway have similarly shown opposing effects on cellsurvival, apoptosis, and proliferation dependent upon cellularcontext, aswell as activation duration and intensity, inmanyothercases (23). For example, surviving melanoma cells after Raf/Mekinhibitor treatment are enriched for higher JNK activation, andRaf/JNK dual inhibitor treatment is synergistic (24). However,positive feedback in JNK signaling response predicts neuroblas-toma patient survival, and JNK inhibition drives tumor growth ina MYCN-driven spontaneous neuroblastoma model (25). Inbreast cancer, our observed negative relationship between cellsurvival/proliferation and JNK activity is supported by the onco-genic role of dominant-negative MAP2K4 mutants (26). Howev-er, dissecting the exact contexts in which JNK activity promotes orinhibits resistance will require more detailed study of the path-way's diverse effects.

The widespread nature of redundancy-mediated resistancemeans we need precise and prognostic ways of identifying com-bination therapies. Just among RTK-targeted therapies, inhibitorsfor cMET, IGF1R, PDGFRb, EGFR, HER3, AXL, and HER2 haveentered clinical trials. How will patients be matched to effectivecombinations of these therapies, while avoiding toxicity? A mostbasic approach might be to quantify the abundance of variousRTKs within tumor cells, and selectively target those with highestabundance or activation (16). Our results, however, indicate thatreceptors display qualitative differences in their resistance capacity(Fig. 1 and 2), suggesting this approachwould be poorly predictiveof effective therapeutic combinations. Rather, a more effectiveapproach might be to identify receptors providing the particularpathwayactivationessential to conferring resistance. Itwas recentlyshown that selective measurement of Grb2–EGFR interactioncould prognostically predict therapeutic response, more so thanabundance orphosphorylationof EGFR (27). Similarly, our resultsindicate that if different receptors rely on the same adapter proteinsfor essential pathway reactivation, assays might be developed to

No resistance conferred

Survival&

X

Jnk Erk Akt

Survival&

Jnk Erk Akt

Survival&

Jnk Erk Akt

X

EGFR-Dependent cell lines

HER2-Dependent cell linesNo resistance conferred

Survival

X

Jnk Erk Akt

Survival

Jnk Erk Akt

Survival

Jnk Erk Akt

Original tumor

Original tumor

EGFR FGFR/cMet IGF1R/PDGFR

Compensatory resistance

Compensatory resistance

EGFRHER3

? XHER2/HER3

Figure 6.

Pathway bias underlies differences in bypassresistance capacity. In RTK-driven tumors,essential signals are transduced from thereceptor to various kinases. Upon blocking theoriginal cancer driver, resistance can beconferred by an untargeted receptor. Somereceptors, however, do not provide the fullcomplement of essential resistance signals orsimultaneously activate pathways that inhibitcell viability. Differences in the requiredcomplement of downstream pathway activationresult in particular receptors being more or lesscapable of driving bypass resistance.Understanding these pathway dependencieswillidentify points for therapeutic intervention andbetter methods to identify which receptor isdriving resistance.






evaluate which receptor is interacting with a particular adapter,identifying the particular RTK-mediating resistance.

In which cases does resistance in essence represent reestab-lishment of the same signaling state as opposed to fundamen-tally different requirements for cancer cell survival? Our resultssuggest that RTK bypass resistance often corresponds to reacti-vation of the particular pathways lost by inhibitor treatment,resembling intracellular signaling in the absence of inhibitor (e.g., HRG in BT474 and SKBR3; Fig. 3A). Particular RTKs canprovide measurable signaling responses and yet fail to produceresistance due to mismatch between the requisite and providedsignaling changes (Fig. 4A). An outstanding question is whethermore global transcriptional mechanisms of resistance, such asepithelial–mesenchymal transition, operate through funda-mental changes in the pathways relied on for survival, or ifother factors such as RTK expression changes still lead to acti-vation of the same survival pathways (28). The many pathwaysinvolved in the development of resistance will necessitatemodeling approaches such as those undertaken here to addressthese mechanisms.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: S. Manole, A.S. MeyerDevelopment of methodology: S. Manole, A.S. MeyerAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): S. Manole, E.J. Richards, A.S. MeyerAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): S. Manole, E.J. Richards, A.S. MeyerWriting, review, and/or revision of the manuscript: S. Manole, E.J. Richards,A.S. MeyerAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): S. Manole, A.S. MeyerStudy supervision: A.S. Meyer

AcknowledgmentsThe authors would like to thank Douglas Lauffenburger, Forest White,

Allison Claas, Sarah Schrier, and Annelien Zweemer for helpful discussions.

Grant SupportThis work was supported by NIH 1-DP5-OD019815-01 (A.S. Meyer) and in

part by the Koch Institute Support (core) grant P30-CA14051 from the NCI.The costs of publication of this articlewere defrayed inpart by the payment of

page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received January 18, 2016; revised May 27, 2016; accepted June 16, 2016;published OnlineFirst July 22, 2016.

References1. Shepherd FA, Rodrigues Pereira J, Ciuleanu T, Tan EH, Hirsh V, Thong-

prasert S, et al. Erlotinib in previously treated non-small-cell lung cancer. NEngl J Med 2005;353:123–32.

2. Baselga J, Cort�es J, Kim S-B, Im S-A, Hegg R, Im Y-H, et al. Pertuzumab plustrastuzumab plus docetaxel for metastatic breast cancer. N Engl J Med.2012;366:109–19.

3. ErcanD, Xiao Y, Capelletti M, Rogers A, Brown A, Lee C, et al. Amplificationof EGFR T790M causes resistance to an irreversible EGFR inhibitor. Onco-gene 2010;29:2346–56.

4. Provenzano PP, Cuevas C, Chang AE, Goel VK, Hoff Von DD, HingoraniSR. Enzymatic targeting of the stroma ablates physical barriers to treatmentof pancreatic ductal adenocarcinoma. Cancer Cell 2012;21:418–29.

5. Niederst MJ, Engelman JA. Bypass mechanisms of resistance to receptortyrosine kinase inhibition in lung cancer. Sci Signal 2013;6:re6.

6. Wilson TR, Fridlyand J, Yan Y, Penuel E, Burton L, ChanE, et al.Widespreadpotential for growth-factor-driven resistance to anticancer kinase inhibi-tors. Nature 2012;487:505–9.

7. Tao JJ, Castel P, Radosevic-RobinN, ElkabetsM, AuricchioN, AcetoN, et al.Antagonism of EGFR and HER3 enhances the response to inhibitors ofthe PI3K-Akt pathway in triple-negative breast cancer. Sci Signal 2014;7:ra29–9.

8. Turke AB, Zejnullahu K, Wu Y-L, Song Y, Dias-Santagata D, Lifshits E, et al.Preexistence and clonal selection of MET amplification in EGFR mutantNSCLC. Cancer Cell 2010;17:77–88.

9. Wilson TR, Lee DY, Berry L, Shames DS, Settleman J. Neuregulin-1-medi-ated autocrine signaling underlies sensitivity toHER2 kinase inhibitors in asubset of human cancers. Cancer Cell 2011;20:158–72.

10. Hutcheson IR, Knowlden JM, Hiscox SE, Barrow D, Gee JMW, Robertson JF,et al. Heregulin beta1 drives gefitinib-resistant growth and invasion in tamox-ifen-resistant MCF-7 breast cancer cells. Breast Cancer Res 2007;9:R50.

11. Zhang Z, Lee JC, Lin L,Olivas V, AuV, LaFramboise T, et al. Activation of theAXL kinase causes resistance to EGFR-targeted therapy in lung cancer. NatGenet 2012;44:852–60.

12. Szerlip NJ, Pedraza A, Chakravarty D, Azim M, McGuire J, Fang Y, et al.Intratumoral heterogeneity of receptor tyrosine kinases EGFRandPDGFRAamplification in glioblastoma defines subpopulationswith distinct growthfactor response. Proc Natl Acad Sci U S A 2012;109:3041–6.

13. Efron B, Gong G. A leisurely look at the bootstrap, the jackknife, and cross-validation. Am Stat 1983;37:36–48.

14. Meyer AS, Zweemer AJM, Lauffenburger DA. The AXL receptor is a sensor ofligand spatial heterogeneity. Cell Syst 2015;1:25–36.

15. Johannessen CM, Boehm JS, Kim SY, Thomas SR, Wardwell L, Johnson LA,et al. COT drives resistance to RAF inhibition throughMAP kinase pathwayreactivation. Nature 2010;468:968–72.

16. Kothari V, Wei I, Shankar S, Kalyana-Sundaram S, Wang L, Ma LW, et al.Outlier kinase expression by RNA sequencing as targets for precisiontherapy. Cancer Discov 2013;3:280–93.

17. Barretina J, Caponigro G, Stransky N, Venkatesan K, Margolin AA, Kim S,et al. The cancer cell line encyclopedia enables predictive modelling ofanticancer drug sensitivity. Nature 2012;483:603–7.

18. Fallahi-Sichani M, Honarnejad S, Heiser LM, Gray JW, Sorger PK. Metricsother than potency reveal systematic variation in responses to cancer drugs.Nat Chem Biol 2013;9:708–14.

19. Bhang H-EC, Ruddy DA, Krishnamurthy Radhakrishna V, Caushi JX, ZhaoR, HimsMM, et al. Studying clonal dynamics in response to cancer therapyusing high-complexity barcoding. Nat Med 2015;21440–8.

20. Meyer AS, Miller MA, Gertler FB, Lauffenburger DA. The receptor AXLdiversifies EGFR signaling and limits the response to EGFR-targeted inhi-bitors in triple-negative breast cancer cells. Sci Signal 2013;6:ra66–6.

21. BeharM, BarkenD,Werner SL, HoffmannA. The dynamics of signaling as apharmacological target. Cell 2013;155:448–61.

22. Garcia-Guzman M, Dolfi F, Zeh K, Vuori K. Met-induced JNK activation ismediated by the adapter protein Crk and correlates with the Gab1-Crksignaling complex formation. Oncogene 1999;18:7775–86.

23. Wagner EF, Nebreda AR. Signal integration by JNK and p38 MAPK path-ways in cancer development. Nat Rev Cancer 2009;9:537–49.

24. Fallahi-Sichani M, Moerke NJ, Niepel M, Zhang T, Gray NS, Sorger PK.Systematic analysis of BRAFV600Emelanomas reveals a role for JNK/c-Junpathway in adaptive resistance to drug-induced apoptosis. Mol Syst Biol2015;11:797–7.

25. FeyD,HalaszM,DreidaxD,Kennedy SP,Hastings JF, RauchN, et al. Signalingpathway models as biomarkers: patient-specific simulations of JNK activitypredict the survival of neuroblastoma patients. Sci Signal 2015;8:ra130–0.

26. Kan Z, Jaiswal BS, Janakiraman V, Bhatt D, Stern HM, Yue P, et al. Diversesomatic mutation patterns and pathway alterations in human cancers.Nature 2010;466:869–73.

27. Smith MA, Hall R, Fisher K, Haake SM, Khalil F, Schabath MB, et al.Annotation of human cancers with EGFR signaling-associated proteincomplexes using proximity ligation assays. Sci Signal 2015;8:ra4–ra4.

28. Thomson S, Petti F, Sujka-Kwok I, Mercado P, Bean J, MonaghanM, et al. Asystems view of epithelial-mesenchymal transition signaling states. ClinExp Metastas 2011;28:137–55.


Manole et al.




2016;76:5219-5228. Published OnlineFirst July 22, 2016.Cancer Res Simin Manole, Edward J. Richards and Aaron S. Meyer

Targeted Therapies−EGFR/HER2 JNK Pathway Activation Modulates Acquired Resistance to

Updated version

10.1158/0008-5472.CAN-16-0123doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2016/09/24/0008-5472.CAN-16-0123.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/76/18/5219.full#ref-list-1

This article cites 28 articles, 7 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/76/18/5219.full#related-urls

This article has been cited by 4 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://cancerres.aacrjournals.org/content/76/18/5219To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-16-0123

http://cancerres.aacrjournals.org/content/suppl/2016/09/24/0008-5472.CAN-16-0123.DC1

http://cancerres.aacrjournals.org/content/76/18/5219.full#ref-list-1

http://cancerres.aacrjournals.org/content/76/18/5219.full#related-urls

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

mailto:[email protected]

http://cancerres.aacrjournals.org/content/76/18/5219


JNK Pathway Activation Modulates Acquired Resistance to ......Bliss synergy calculation The...

Documents

Transcript of JNK Pathway Activation Modulates Acquired Resistance to ......Bliss synergy calculation The...