AXL Inhibition Sensitizes Mesenchymal Cancer Cells to ...(TKI).Thus,drug-sensitivecancercelllines...

Therapeutics, Targets, and Chemical Biology

AXL Inhibition Sensitizes Mesenchymal CancerCells to Antimitotic Drugs

Catherine Wilson1, Xiaofen Ye1, Thinh Pham2, Eva Lin1, Sara Chan2, Erin McNamara3, Richard M. Neve1,Lisa Belmont1, Hartmut Koeppen2, Robert L. Yauch1, Avi Ashkenazi4, and Jeff Settleman1

AbstractMolecularly targeted drug therapies have revolutionized cancer treatment; however, resistance remains a

major limitation to their overall efficacy. Epithelial-to-mesenchymal transition (EMT) has been linked to acquiredresistance to tyrosine kinase inhibitors (TKI), independent of mutational resistance mechanisms. AXL is areceptor tyrosine kinase associated with EMT that has been implicated in drug resistance and has emerged as acandidate therapeutic target. Across 643 human cancer cell lines that were analyzed, elevated AXL was stronglyassociated with a mesenchymal phenotype, particularly in triple-negative breast cancer and non–small cell lungcancer. In an unbiased screen of small-molecule inhibitors of cancer-relevant processes, we discovered that AXLinhibition was specifically synergistic with antimitotic agents in killing cancer cells that had undergone EMT anddemonstrated associated TKI resistance. However, we did not find that AXL inhibition alone could overcomeacquired resistance to EGFR TKIs in the EMT setting, as previously reported. These findings reveal a novelcotreatment strategy for tumors displaying mesenchymal features that otherwise render them treatmentrefractory. Cancer Res; 74(20); 5878–90. �2014 AACR.

IntroductionEpithelial-to-mesenchymal transition (EMT) is a vital cel-

lular process during normal development, and contributes tothe invasive and metastatic properties of human tumors (1).EMT has also been implicated in resistance to multiple cancerdrug therapies, including several tyrosine kinase inhibitors(TKI). Thus, drug-sensitive cancer cell lines selected in culturefor acquired drug resistance can adopt a mesenchymal phe-notype (2–4). Moreover, EMT has been associated withacquired resistance to epidermal growth factor receptor(EGFR) TKIs in patients with EGFR-mutant lung cancer (2, 5).

The receptor tyrosine kinase (RTK) AXL is a member of theTAM family kinases, which also includes TYRO-3 and MER (6).AXL is overexpressed in many solid tumors (4, 7–10), althoughactivatingmutations have not been observed (11). AXL has oneknown ligand, Gas6 (12), and ligand binding promotes cellproliferation, survival, and migration through activation of thePI3K–AKT–S6K and ERK–MAPK pathways (9, 13–15). AXLexpression may be a negative prognostic factor for patients

with breast and pancreatic cancer (7, 9), and may be a uniqueEMT effector in breast cancer progression (7, 16).

The association of AXL with EMT has prompted interest inAXL as a therapeutic target. AXL has been associated with amesenchymal signature in non–small cell lung cancer (NSCLC)and has been proposed as a therapeutic target in EGFR TKIresistance (17), and in a model of acquired erlotinib resistance,EMT associated with elevated AXL appeared to underlieerlotinib resistance (4). Moreover, cotreatment of EGFRwild-type NSCLC xenografts with erlotinib and an anti-AXLantibody decreased tumor volume and metastasis (18). Sim-ilarly, lapatinib-resistant HER2-positive breast cancer cellsdemonstrated elevated AXL (19), and AXL knockdown report-edly re-sensitizes imatinib-resistant chronic myelogenous leu-kemia cell lines to imatinib (20).

We broadly surveyed AXL expression in a large panel ofhuman cancer cell lines, and assessed its role in drug resistanceassociated with a mesenchymal phenotype. AXL expressionwas well correlated with a mesenchymal phenotype in thecontext of both intrinsic and drug resistance–associated EMT.AXL inhibition did not detectably increase sensitivity to TKIsas previously reported; however, we observed a striking syn-ergistic interaction between AXL inhibition and antimitoticagents specifically in mesenchymal tumor cells.

Materials and MethodsCell lines

Cell lines were from American Type Culture Collection orDeutsche Sammlung von Mikroorganismenund Zelkulturen,and were obtained between 2010 and 2014. All cell lines wereauthenticated by short tandem repeat (STR) profiling and SNPfingerprinting, andwere confirmed to beMycoplasma-negative

1Department of Discovery Oncology, Genentech, South San Francisco,California. 2Department of Pathology, Genentech, South San Francisco,California. 3Department of Translational Oncology, Genentech, South SanFrancisco, California. 4Department of Cancer Immunology, Genentech,South San Francisco, California

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

Corresponding Author: Jeff Settleman, Department of Discovery Onco-logy, Genentech Inc., 1 DNAWay, South San Francisco, CA 94080. Phone:650-467-7140; Fax: 650-255-5770; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-14-1009

�2014 American Association for Cancer Research.

CancerResearch

Cancer Res; 74(20) October 15, 20145878

on August 3, 2021. © 2014 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst August 14, 2014; DOI: 10.1158/0008-5472.CAN-14-1009

http://cancerres.aacrjournals.org/

by the Genentech cell line banking facility (gCELL) before use(see Supplementary Data). Cells weremaintained in culture fora 6-week period and thereafter a fresh vial of cells was obtainedfromgCELL. To induceEMT, cellswere treatedwith 2 ng/mLofrh-TGFb1 every 3 days over a 2- to 3-week period.

InhibitorsErlotinib and cisplatin were from LC laboratories. Docetaxel

and doxorubicin were from Sigma. PF-03814735, BI-2536, MP-470, and PHA-739358 were from Selleck Chemicals. Paclitaxelwas from Tocris. Gemcitabine was from Toronto Research.R428 was from Synkinase. Recombinant human AXL–Fc, Gas6,and TGFb1 were from R&D Systems.

RNA-seq analysisRNA from cell lines and was used to prepare libraries using

the TruSeq RNA Sample Preparation Kit (Illumina). Librarieswere sequenced on the Illumina HiSeq 2000 platform. Expres-sion of AXL and Gas6 RNA was extracted from a databasederived from RNA-seq analysis (Klijn and colleagues; manu-script in preparation).

ImmunoblottingImmunodetection of proteins was performed using stan-

dard protocols. The AXL, MER, TYRO-3, E-cadherin, vimen-tin, phospho-CDC2 (Tyr 15), CDC2, phospho-AKT, total AKT,phospho-S6, total S6, and GAPDH antibodies were from CellSignaling Technology, and the PARP antibody was fromeBioscience.

Gas6 ELISACells were incubated overnight and replenished in fresh

RPMI-1640 media. Medium was collected 24 hours later andwas analyzed using theHumanGAS6QuantikineELISAKit fromQuantikine, R&D Systems. ELISA was performed as per themanufacturer's instructions and normalized to cell number.

Invasion assayCell suspensions were prelabeled with DilC12(3) from BD

Biosciences in serum-free RPMI media that was added to theapical chamber of the BD BioCoat Tumor Invasion System,8 mm from BD Biosciences. Invasion assays were performedas per the manufacturer's instructions.

Kinase profilingThe R428 kinase profile was performed at KINOMEscan.

R428 was screened at 1,000 nmol/L, and results for primaryscreen binding interactions are reported as percentage control,where lower numbers indicate stronger binding.

Cell viabilityCell viability was assessed as previously described (21).

RNA interferenceAXL knockdown was achieved by transfection using ON-

TARGETplus AXL siRNA (Dharmacon) and LipofectamineRNAiMax (Invitrogen). ON-TARGETplus Non-targeting PoolsiRNA (Dharmacon) served as control.

Time-lapse imagingCells were seeded in a glass-bottomed 24-well plate (Greiner

Bio-One). Hundred cells per treatment were tracked andmitotic fate and cell death were scored as described previously(22).

Xenograft studiesAll procedures were approved by and conformed to the

guidelines and principles of the Institutional Animal Care andUse Committee of Genentech and were carried out in anAssociation for the Assessment and Accreditation of Labora-tory Animal Care-accredited facility. Five million HeLa orMDA-MB-231 cells were inoculated in the right flank of Nu/Nu nude mice. When tumors reached 100 to 200 mm3, micewere treated with vehicle control, R428 (125 mg/kg five timesper week; oral gavage), docetaxel (10mg/kg three times aweek,intravenously for HeLa and 10 mg/kg once a week, intrave-nously for MDA-MB-231), or the combination. Tumors weremeasured three times weekly using digital calipers and tumorvolumes were calculated as [L � (W � W)]/2. Differencesbetween R428 and docetaxel combination groups and individ-ual-treated and control groupswere determined using the two-way ANOVA.

Statistical analysisBliss expectation was calculated as (Aþ B)� A� B, where A

and B are the fractional growth inhibitions of drug A and B at agiven dose. The difference between Bliss expectation andobserved growth inhibition of the combination of drugs A andB at the same dose is the "Delta Bliss excess." Delta Bliss valueswere summed across the dosematrix to generate the Bliss sum.Differences between two groups were determined using theStudent t test, P values are represented as � , P< 0.05, ��, P < 0.01,and ��, P < 0.001. Differences between the R428 and docetaxelcombination groups and individual-treated and control groupswere determined using the two-way ANOVA.

ResultsAXL expression correlates with a mesenchymalphenotype

UsingRNA-seq,we profiledmRNAexpression ofAXL (Fig. 1Aand Supplementary Dataset S1) and its ligand Gas6 (Supple-mentary Fig. S1A) in 643 human cancer cell lines. AXL expres-sion in cell lines from breast and lung demonstrated thegreatest dynamic range within a tissue subset, where expres-sion levels could readily classify most lines as "AXL-high" or"AXL-low." Gas6 mRNA was more uniformly expressed. Toidentify a potential correlation between AXL expression andthe mesenchymal phenotype, we examined expression of theepithelial marker E-cadherin and the mesenchymal markervimentin (Supplementary Fig. S1B). AXL-high cells generallyexpressed abundant vimentin, whereas AXL-low cells demon-strated higher E-cadherin expression. AXL-high cells showed,on average, a 4.6-fold increase in vimentin relative to AXL-lowcells (P ¼ 2.04e�24; Supplementary Fig. S1C). Thus, AXLexpression is strongly associated with a mesenchymal pheno-type in human cancer cell lines.

AXL Inhibition Sensitizes Cancer Cells to Antimitotics

www.aacrjournals.org Cancer Res; 74(20) October 15, 2014 5879




TNBC HER2-amp ER+/PR-2–10

2–5

20

25

210

AX

L R

PK

M

**

**

NS

Breast cancer cell linesAXL RNA Seq

Other

Blood

Brain

Breas

t

Colon

Liver

Lung

Lymph n

ode

Ovary

Pancr

eas

Skin

Stom

ach

2–10

2–5

20

25

210

AX

L R

PK

M

A B

C

D

E

Sensitivity to erlotinib

AXL

*PC

9

*HC

C82

7

H16

66

H32

2T

*HC

C40

06

SK

-ME

S-1

H29

2

H35

8

H21

26

H18

38

H14

35

H15

68

H59

6

H20

30

HO

P 1

8

H24

05

HLF

α

H52

2

H44

1

*H19

75

*H16

50

*H20

09

EK

VX

A54

9

H64

7 H

1651

H17

81

H17

93

HO

P 6

2 A

427

H46

0

H23

H66

1

SW

1573

H

OP

92

H17

03

H22

6 H

1299

CA

LU-1

H21

22

H11

55

H52

0

H65

0

H83

8

LXF

L529

*HC

C82

7 E

MT

*PC

9 E

MT

*P

C9

*HC

C82

7

GAPDH

Vimentin

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49

Lung cancer cell lines

Triple-negative ER-positive/ PR-negative Her2-amplified/overexpressed

AXL

AXL Longer exposure

TGFβ TGFβ

HC

C19

54

HC

C19

54

BT

-549

C

AL-

148

CA

L51

CA

L-85

-1

HC

C18

06

HC

C19

37

HC

C38

H

CC

70

HS

578

T

MD

A-M

B-4

36

MD

A-M

B-4

68

MD

A-M

B-2

31

EF

M19

C

AM

A-1

E

VS

A-T

HC

C14

28

HC

C15

00

KP

L1

MD

A-M

D-1

75

MC

F7

MD

A-M

B-4

15

T47

D

ZR

-75-

1

HC

C19

54

HC

C19

54

AU

565

EF

M-1

92A

B

T47

4 H

CC

2218

H

CC

202

HC

C15

69

HC

C14

19

EF

M-1

92B

K

PL4

M

DA

-MD

-453

M

DA

-MB

-361

S

K-B

R-3

U

AC

C81

2 U

AC

C89

3

GAPDH

GAPDH

AXL

TYRO-3

MER

*BT

-549

C

AL-

148

CA

L51

CA

L-85

-1

HC

C18

06

HC

C19

37

*HC

C38

H

CC

70

*HS

578

T

*MD

A-M

B-4

36

MD

A-M

B-4

68

*MD

A-M

B-2

31

*HD

Q-P

1 *H

CC

1143

*CA

L-12

0 *H

CC

1395

DU

4475

*S

W52

7

BT

-20

*SU

M 1

315M

02

MF

M-2

23

SU

M 1

49P

T

*MD

A-M

B-4

35

*MD

A-M

B-1

57

*SU

M 1

597T

Triple-negative breast cancer cell lines

– + – +

Wilson et al.

Cancer Res; 74(20) October 15, 2014 Cancer Research5880




Within breast cancer subtypes, AXL is relatively elevatedwithin the triple-negative breast cancer (TNBC) subset com-pared with estrogen (ER)-positive/progesterone (PR)-negativeandHER2-amplified subsets (Fig. 1B). GAS6 is similarly elevatedin the TNBC subset (Supplementary Fig. S1D). Correspondingly,AXL protein was highly elevated in TNBC relative to the othersubtypes (Fig. 1C).We expanded the TNBCpanel andobserved aparticularly strong correlation between AXL expression and themesenchymal subset of TNBC lines (Fig. 1D and SupplementaryFig. S1E). Expression of the AXL-related TYRO-3 RTK is variable,and is not correlated with any particular subset, and MERexpression is inversely correlated with AXL (Fig. 1D).Immunohistochemistry analysis of TNBC tumors revealed

that most expressed E-cadherin, with 20 of 26 cases exhibitingstrong membranous staining in all tumor cells and four dis-playing focal expression inmalignant cells (Supplementary Fig.S2). Significantly, seven of 26 samples displayed vimentintumor cell staining in varying proportions of malignant cells(Supplementary Fig. S2). In all samples, vimentin andAXLweredetected in normal stroma cells. AXL staining of a smallproportion of tumor cells was seen in three of 26 tumorsamples (Supplementary Fig. S2). The observed percentage ofAXL-positive TNBCs is consistent with recent previous reports(23, 24).In NSCLC, a previous report correlated the mesenchymal

phenotype with sensitivity to the EGFR TKI, erlotinib (25).Indeed, AXL tended to be elevated in erlotinib-insensitive cellswith relatively high vimentin (Fig. 1E). Conversely, cell lineswith greater erlotinib sensitivity tended to exhibit lowervimentin and AXL expression. This relationship was alsoobserved with RNA-seq analysis (Supplementary Fig. S1F).These results indicate a strong correlation between AXLexpression and a mesenchymal/drug–resistant phenotype inbreast cancer and NSCLC cells.

TGFb-induced EMT is associatedwith increased AXL andTKI resistanceTo explore a functional requirement for AXL in EMT-asso-

ciated drug resistance, we experimentally induced EMT incancer cell lines with previously established drug sensitivityby treatment with TGFb (1, 26, 27). Exposure of EGFR- orHER2-"addicted" cell lines to TGFb yielded the expected EMT,with transition from a compact uniform appearance to aspindle-like morphology (Fig. 2A), as well as loss of E-cadherinand gain in vimentin (Fig. 2B), and increased invasion capacity(Fig. 2C). EMT is a reversible process, and indeed, uponwithdrawal of TGFb for 2 weeks, the mesenchymal cellsreverted to an epithelial phenotype (Supplementary Fig. S3A).

Expression of AXL andGAS6was substantially induced uponEMT, consistent with AXL's association with a mesenchymalphenotype (Fig. 2B and D). Significantly, HER2-amplifiedHCC1954 cells, following EMT, exhibited 10-fold reduced sen-sitivity to the HER2/EGFR TKI lapatinib (parental: IC50, 0.456mmol/L; TGFb: IC50, 4.036 mmol/L; Fig. 2E). Similarly, EGFR-mutant NSCLC PC9 cells exhibited 4-fold reduced erlotinibsensitivity following EMT (parental: IC50, 0.05 mmol/L; TGFb;IC50, 0.214 mmol/L; Fig. 2F). In HCC1954 and PC9 cells, post-EMT, TKI treatment failed to promote apoptosis (Fig. 2G). TKIresistance in the mesenchymal cells did not reflect drug efflux,as TKIs induced strong suppression of downstream signaling(Supplementary Fig. S3B). Thus, upon TGFb-induced EMT inRTK oncogene-addicted cancer cells lines, the resulting mes-enchymal cells exhibit substantially increased AXL, andbecome drug resistant.

EMT-associated drug resistance is independent of AXLfunction

Recent reports have implicated AXL in acquired drugresistance (4, 19, 20). Therefore, we used these models toexamine a functional AXL requirement in EMT-associateddrug resistance. Initially, we used R428, an AXL kinaseinhibitor (Supplementary Dataset S2; ref. 28). As expected,R428 effectively suppressed GAS6-induced phospho-AXL(Supplementary Fig. S4A), phospho-AKT, and phospho-S6levels (Supplementary Fig. S4B and S4C), and decreasedinvasion capacity of AXL-expressing cells (SupplementaryFig. S4D). These effects were not observed in AXL-negativeBT-20 cells. We tested the ability of R428 to restore erlotinibsensitivity in a TGFb-induced resistance model and in twomodels of acquired erlotinib resistance demonstrating EMTand increased AXL (Fig. 3B and Supplementary Fig. S5A). Inboth parental and drug-resistant cell lines, we observedeither very weak R428/erlotinib synergy (positive delta Bliss)or antagonistic interaction (negative delta bliss; Fig. 3Aand C), with no change in cell viability (Fig. 3A). Similarly,there was no change in the IC50 value for erlotinib sensitivityin combination with R428 (1 mmol/L) in either model(Fig. 3D and E). Furthermore, AXL knockdown did notrestore erlotinib sensitivity in either of the resistance models(Fig. 3F–H and Supplementary Fig. S5B). In addition, inhi-bition of AXL using an AXL–Fc fusion protein, an anti-AXLantibody, YW327.6S2 (Supplementary Fig. S5C and S5D),doxycycline-inducible AXL knockdown (Supplementary Fig.S5E–S5G), or overexpression of AXL (Supplementary Fig.S6A–S6C), does not detectably affect sensitivity to erlotinib.Finally, cotreatment of resistant cells with either AXL–Fc

Figure 1. AXL expression correlates with amesenchymal signature. A, scatter plot representingAXLRNA expression for 643 human cancer cell lines. Red line,mean expression within each tissue type. B, scatter plot illustrating AXL expression in 46 breast cancer cell lines based on ER/PR-positive and HER2 status.Differences between TNBC subtype compared with the HER2-amp (P ¼ 0.0058) and ERþ/PR� (P ¼ 0.0081) were calculated using the Student t test. Nodifferences were observed between HER2-amp and ERþ/PR� (NS, nonsignificant). Black line, median expression within each subtype. C, immunoblotsindicating AXL protein in triple-negative, ER/PR-positive, andHER2-amplified breast cancer cells. D, immunoblots indicating AXL, MER, and TYRO-3 proteinin TNBC cells. We note reported controversy regarding the tissue origin of MDA-MB-435 cells, which clusters with cell lines of melanoma origin byexpression profiling. Asterisks, cell lines with high vimentin expression. E, immunoblots showing AXL and vimentin expression in NSCLC cell lines. Cell lineswere arrangedaccording to erlotinib sensitivity as IC50 (25). #1 indicates themost erlotinib-sensitive and#46 is the least sensitive. Lanes 47 and49 correspondto HCC827 and PC9 cell lines, respectively, treated with TGFb for 14 days. GAPDH levels demonstrate protein normalization. Asterisks, cell lines with EGFRmutation.






fusion protein or the AXL antibody with erlotinib did notprevent the emergence of drug-resistant clones (Supplemen-tary Fig. S6D).

We next examined R428 activity in an AXL-negative EMT-associated resistance model. EGFR-mutant HCC827 NSCLCcells, upon TGFb treatment, become mesenchymal (Fig. 1E,lane 48 and 49) and erlotinib resistant (Fig. 3D), but do notexhibit AXL induction (Fig. 1E). R428 did not sensitize thesecells to erlotinib (Fig. 3D). Furthermore, MP-470, another AXLinhibitor (4), did not synergize with erlotinib (SupplementaryFig. S7A). Notably, AXL expressionwas consistently suppressedby TKI treatment in the mesenchymal cells (Fig. 2G), furtherindicating that AXL is not required for the observed resistance.

These collective observations do not support a role for AXLinhibition in overcoming acquired resistance to TKIs in theEMT context.

AXL inhibition sensitizes erlotinib-resistantmesenchymal cells to docetaxel

TGFb-induced EMT or erlotinib-induced resistance yieldsmesenchymal cells that are cross-resistant to many antican-cer agents (Supplementary Table S1 and data not shown).We explored whether AXL inhibition could sensitize resis-tant mesenchymal cells to other anticancer agents. PC9 cellsand mesenchymal derivatives were screened for sensitivityto 100 anticancer agents. Mesenchymal cells were selectively

0

1

2

3

4

5

6

7

Parental

PC9HCC1954

TGFβ

Gas

6 ng

/mL

per

1×10

6 cel

ls

***

***

TGFβParental0

2

4

6

8

10

12

Fol

d ch

ange

in in

vasi

on

*

1010.10.0010.010

20

40

60

80

100

120

Lapatinib (μmol/L)

% C

ell v

iabi

lity

Parental

TGF-β

A C

D

BH

CC

1954

Parental TGFb

BT

474

PC

9

E F

E-cadherin

MER

GAPDH

TG

Fb

BT474

E-cadherin

Vimentin

CO

N

TG

Fb

HCC1954

GAPDH

AXL

GAPDH

E-cadherin

Vimentin

AXL

CO

N

TG

Fb

SUM149PT

CO

N

TG

Fb

GAPDH

E-cadherin

Vimentin

AXL

G

1010.10.0010.010

20

40

60

80

100

120

ParentalTGF-β

Erlotinib (μmol/L)

% C

ell v

iabi

lity

E-Cadherin

Vimentin

GAPDH

TGFb

HCC1954

Cleaved PARP

– + – + LAP

AXL

– + – + ERL

PC9

TGFb

PC9

Figure 2. AXL expression is increased upon TGFb-induced EMT. A, microscopically observed morphologic changes in breast cancer cell lines HCC1954 andBT474, and lung cancer cell line PC9 treated with TGFb for 14 days. B, immunoblot demonstrating loss of E-cadherin and MER and increased vimentinand AXL upon TGFb treatment of BT474, HCC1954, and SUM149PT, and PC9 cells. C, bar graph illustrating enhanced invasion capacity of TGFb-treated(day 14) HCC1954 cells. Differences between parental and TGFb-treated (P ¼ 0.033) cells were calculated by the Student t test. D, bar graphdemonstrating increased Gas6 secretion (by ELISA) in HCC1954 and PC9 cells following TGFb-induced EMT (day 14). Differences between parentalcomparedwith TGFb-treated cells (HCC1954, P < 0.0001 and PC9,P¼ 0.0003) were calculated by the Student t test. Error bars, mean�SEM. E, cell viabilityassay demonstrating the effect of lapatinib in TGFb-treated (day 14) HCC1954 cells. IC50 values for lapatinib were calculated in Prism, parental: IC50,0.456 mmol/L; TGFb: IC50, 4.036 mmol/L. F, cell viability assay demonstrating the effect of erlotinib in TGFb-treated (day 14) PC9 cells. IC50 values for erlotinibwere calculated in Prism, parental: IC50, 0.05 mmol/L; TGFb: IC50, 0.214 mmol/L. G, immunoblot showing the effect of erlotinib (ERL; 50 nmol/L) andlapatinib (LAP; 1 mmol/L) on PARP cleavage and AXL expression after 72 hours. All error bars, mean � SEM.

Wilson et al.





resistant to many agents—most notably, to docetaxel, anantimitotic agent (Supplementary Table S1). We extendedthis observation to other erlotinib resistance models. In both

the erlotinib-induced resistance (ERL-R) and TGFb-inducedmesenchymal resistance (MES) models, the mesenchymalcells were more docetaxel resistant (Fig. 4A). Cotreatment

0.001 0.01 0.1 1 100

20

40

60

80

100

120

% C

ell v

iabi

lity

NTC siRNA

Axl siRNA #2

Axl siRNA #4

HCC4006 ERL-R

Axl siRNA #1

Axl siRNA #3

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

120

140

160

Erlotinib (μmol/L)

Erlotinib (μmol/L)

Erlotinib (μmol/L)

Erlotinib (μmol/L) Erlotinib (μmol/L) %

Cel

l via

bilit

y

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

120

140

PARPAR + R428ERL-RERL-R + R428

HCC827 HCC4006 E

A

D

C

β-Actin

HCC4006

PAR

HCC827

AXL

β-Actin

AXL

ERL-

R

PAR

ER

L-R

Combination with R428 and erlotinibSum Bliss scoreCell line

117PC9 mesenchymal88PC9 parental43HCC4006–8HCC4006 ERL-R–13HCC827 ERL-R

–233HCC827 parental

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

120

PARPAR + R428MESMES+ R428

HCC827 PC9

0.001 0.01 0.1 1 100

20

40

60

80

100

120

% C

ell v

iabi

lity

F HG

PC9 mesenchymal Percentage inhibition

10

3

1

0.3

0.1

0.03

0.003

0.001

0

0.01

10

3

1

0.3

0.10.03

0.003

0.001

0

0.01

10

3

1

0.3

0.10.03

0.003

0.001

0

0.01

Delta Bliss PC9 parental

Percentage inhibition Delta Bliss

HCC4006 ERL-R Percentage inhibition Delta Bliss

HCC4006 parental Percentage inhibition Delta Bliss

Erlo

tinib

(μm

ol/L

)

R428 (μmol/L) R428 (μmol/L) R428 (μmol/L) R428 (μmol/L)

HCC827 ERL-R Percentage inhibition Delta Bliss

HCC827 parental Percentage inhibition Delta Bliss

Erlo

tinib

(μm

ol/L

)


Erlo

tinib

(μm

ol/L

)


B

Total AXL

GAPDH

NTC

SiR

NA

AX

L S

iRN

A #1

A

XL

SiR

NA

#2

AX

L S

iRN

A #3

AX

L S

iRN

A #4

Vimentin

Cleaved PARP

Vimentin

GAPDH

AXL#2 NTC AXL#2

MES

NTC

– + – + – + – + ERL 50nmol/L 72 h

PAR

0

0.1

0.25 0.

5 1 2 3

0

0.1

0.25 0.5 1 2 3

0

0.1

0.25 0.

5 1 2 3 0

0.1

0.25 0.

5 1 2 3 0

0.1

0.25 0.

5 1 2 3 0

0.1

0.25 0.

5 1 2 3

0

0.1

0.25 0.5 1 2 3 0

0.1

0.25 0.5 1 2 3 0

0.1

0.25 0.5 1 2 3

0

0.1

0.25 0

0.1

0.3 1

0.25 5

7.5 0

0.1

0.3 1

0.25 5

7.5

0.5 1 2 3

Figure 3. AXL inhibition does not re-sensitize erlotinib-resistant NSCLC cells. A, drug matrix heatmap grid illustrating percentage inhibition and delta Bliss forR428 in combinationwith erlotinib in PC9 parental andmesenchymal (MES) cell lines (top), HCC4006 parental and ERL-R cells (middle), andHCC827 parentaland ERL-R cells (bottom). Drug matrix heatmap grids correspond to one representative experiment out of three independent experiments. B, immunoblotdemonstrating increased AXL following prolonged erlotinib treatment (ERL-R) in HCC4006 and HCC827 cells. C, table illustrating the sum Bliss score forerlotinib in combination with R428. D, cell viability assay demonstrating the effect of R428 (1 mmol/L) in combination with erlotinib in PC9 parental andmesenchymal cells (left) andHCC827 parental andmesenchymal cells (right). E, cell viability assay illustrating the effect of R428 (1mmol/L) in combinationwitherlotinib in HCC4006 parental and ERL-R cells (left), and HCC827 parental and ERL-R cells (right). All error bars, mean� SEM. F, immunoblot demonstratingdecreased AXL expression following knockdown (10 nmol/L) for 72 hours with four different AXL siRNAs in PC9 mesenchymal cells. NTC siRNA serves as acontrol. G, immunoblot showing the effect of AXL knockdown (10 nmol/L) for 72 hours and in combination with erlotinib (50 nmol/L) for a further 72 hours onPARP cleavage and vimentin expression in PC9 parental (PAR) and mesenchymal (MES) cells. H, cell viability assay demonstrating the effect of AXLknockdown (20 nmol/L) for 24 hours and in combination with erlotinib for a further 72 hours in HCC4006 ERL-R cells. Error bars, mean � SEM.






with docetaxel and R428 was synergistic in both models (Fig.4A and B). Similarly, in the drug-resistant HCC827 ERL-Rand HCC4006 ERL-R cells, greater sum Bliss scores wereobserved (Fig. 4B). In contrast, we did not observe synergywith R428 and gemcitabine (Fig. 5A).

Cotreatment of drug-resistant mesenchymal cells with R428and docetaxel dramatically shifted docetaxel IC50 values. Thus,for mesenchymal PC9 cells, the IC50 for docetaxel alone was>300 nmol/L, whereas in combination with R428, it was 0.3nmol/L (Fig. 4C). Similarly, in the erlotinib-resistant mesen-chymal HCC4006 cells, the IC50 for docetaxel alone was >300nmol/L, whereas in combination with R428, it was 0.191nmol/L (Fig. 4D). Conversely, in mesenchymal HCC827 cells,R428 did not synergistically interact with docetaxel, suggestingthat the observed synergy reflects an AXL-targeted effect of

R428 (Fig. 4C). We observed similar synergy between the AXLinhibitor MP-470 and docetaxel (Supplementary Fig. S7A) andno change in IC50 values upon cotreatment with MP-470 anddocetaxel in the AXL-negative mesenchymal HCC827 cells(Supplementary Fig. S7B). Collectively, these findings suggestthat EGFR-mutant cancer cells that become resistant to erlo-tinib and cross-resistant to docetaxel through innate or drug-induced EMT, can be sensitized to docetaxel by AXL inhibition.

AXL inhibition sensitizes mesenchymal cancer cells toantimitotic agents

To establish that R428/docetaxel synergy is independentof TGFb or drug exposure, we evaluated this treatmentcombination in a panel of 100 cancer cell lines (Supplemen-tary Dataset S3). We observed strong synergy in a subset of

A

HCC4006 ERL-R


HCC4006 parental


DT

X(n

mol

/L)

R428 (µmol/L) R428 (µmol/L) R428 (µmol/L) R428 (µmol/L)

PC9 mesenchymal


DT

X(n

mol

/L)

PC9 parental



HCC827 ERL-R


HCC827 parental


DT

X(n

mol

/L)


C

D

BCombination with R428 and docetaxel

Cell line Sum Bliss scorePC9 Mesenchymal 2536

PC9 Parental 108HCC827 ERL-R 575

HCC827 Parental 11HCC4006 ERL-R 470

HCC4006 110

DTX (nmol/L) DTX (nmol/L)

DTX (nmol/L) DTX (nmol/L)

Figure 4. AXL inhibition synergizes with docetaxel in erlotinib-resistant mesenchymal NSCLC. A, drug matrix heatmap grid illustrating percentage inhibitionand delta Bliss for R428 in combination with docetaxel (DTX) for PC9 parental (PAR) and mesenchymal (MES) cells (top), HCC4006 parental and ERL-R celllines (middle), and HCC827 parental and ERL-R cells (bottom). Drug matrix heatmap grids correspond to one representative experiment out of threeindependent experiments. B, table illustrating the sum Bliss score for docetaxel in combination with R428. C, cell viability assay demonstrating the effect ofR428 (1 mmol/L) in combination with docetaxel. IC50 values for docetaxel single agent or in combination R428 (1 mmol/L) were calculated in Prism,PC9: PAR, 2.265 nmol/L; PAR þ R428, 0.535 nmol/L; MES, >300 nmol/L; MES þ R428, 0.3 nmol/L. D, cell viability assay illustrating the effect of R428(1 mmol/L) in combination with docetaxel. IC50 values for docetaxel single agent or in combination with R428 (1 mmol/L) were calculated in Prism, HCC4006(left): PAR, 0.429 nm; PAR þ R428, 0.223 nmol/L; ERL-R, >300 nmol/L; ERL-R þ R428, 0.191 nmol/L; HCC827 (right): PAR, 2.759 nm; PAR þ R428, 1.012nmol/L; ERL-R, >300 nmol/L; ERL-R þ R428, 100 nmol/L. All error bars, mean � SEM.

Wilson et al.






HeLa


MDA-MB-231


PH

A-7

3935

8( µ

mol

/L)

DT

X(n

mol

/L)

Gem

cita

bine

(µm

ol/L

)

A

DMSO Axl-FC R428

–DT

X+

DT

X

D DMSO MP-470

–DT

X+

DT

X

C

E NTC SiRNA Axl SiRNA

–DT

X+

DT

X

AXL

GAPDH

– + – + DTX

NTC AXL SiRNA

0.1 1 10 100 1,000−400

−200

0

200

400

600

Sum

Blis

ssc

ore

AXL RPKM (log10)

R2 = 0.5044B

Docetaxel PC9 mesenchymal 2,536antimitotic HCC827 ERL-R 575

HCC4006 ERL-R 470HeLa 371

CAL-120 292HCC1937 233HCC1143 133

MDA-MB-231 121HCC4006 110

PC9 parental 108HCC827 parental 11

PHA-739358 MDA-MB-231 528Aurora kinase inhibitor PC9 mesenchymal 524

antimitotic HeLa 250PC9 parental −568

HCC1143 −305PF03814735 PC9 mesenchymal 491

Aurora kinase inhibitor MDA-MB-231 333anti-mitotic PC9 parental 68

HCC1143 −45HeLa −124

BI-2536 PC9 mesenchymal 411PLK-1 inhibitor MDA-MB-231 368

HeLa 156PC9 parental 114

HCC1143 −127Gemcitabine MDA-MB-231 30

Nucleoside analog HCC1143 −304HeLa −169

PC9 parental −509PC9 mesenchymal −592

Cisplatin HeLa 2DNA synthesis inhibitor MDA-MB-231 −10

Doxorubicin HeLa −114DNA intercalator MDA-MB-231 −27

DTX (nmol/L) mol/L mol/L

mol/Lmol/L

Figure 5. AXL inhibition enhances the efficacy of antimitotic agents. A, table summarizing the sum Bliss score upon cotreatment with R428 and docetaxel;mesenchymal cell lines indicated in blue (left). Right, drugmatrix heatmapgrid illustratingpercentage inhibition anddeltaBliss forHeLa (left) andMDA-MB-231(right) cell lines upon cotreatment with R428 and docetaxel (DTX; top), PHA-739358 (middle), and gemcitabine (bottom). Drug matrix heatmapgrids correspond to one representative experiment out of three independent experiments. B, plot demonstrating the correlation between AXL expression(RPKM) and sum Bliss score of the R428 and docetaxel combination. AXL-negative cell lines were removed from analysis. R2 value equals 0.5044 andwas calculated using Pearson correlation coefficient. C, cell viability assay demonstrating the effect of R428 (1 mmol/L) in combination with docetaxel(left), PHA-739358 (middle), and gemcitabine (right). Error bars, mean � SEM. MDA-MB-231 IC50 values for docetaxel single-agent treatment or incombination with R428 were calculated in Prism, DMSO, 0.275 nmol/L; R428 (1 mmol/L), 0.147 nmol/L; R428 (3 mmol/L), 0.053 nmol/L. IC50 values forPHA-739358 single-agent treatment or in combination R428were calculated in Prism, DMSO, 1.626 mmol/L; R428 (1 mmol/L), 0.336 mmol/L; R428 (3 mmol/L),0.17 mmol/L. D, Syto 60 cell staining of HeLa cells treated for 72 hours with docetaxel (3 nmol/L), R428 (1 mmol/L), AXL–Fc (10 mg/mL), orMP-470 (1 mmol/L). E,left, Syto 60 cell staining ofHeLa cells upon siRNAknockdownofAXL (10 nmol/L) or nontargeting control (NTC) for 72hours and in combinationwith docetaxel(1 nmol/L) for a further 72 hours. Right, immunoblot demonstrating AXL protein expression following knockdown in combination with docetaxel (1 nmol/L).






AXL-expressing mesenchymal cell lines (Fig. 5A and Supple-mentary Dataset S3), and a correlation between AXL expres-sion and the sum Bliss score (Fig. 5B). Similarly, R428synergized with the related taxane paclitaxel to inhibitproliferation (Supplementary Fig. S8A–S8D), but not withgemcitabine, doxorubicin, or cisplatin (Fig. 5A and C andSupplementary Fig. S8F and S8G). We also observed similarsynergy between docetaxel and an AXL–Fc fusion protein,which blocks GAS6 binding to AXL (18), or with MP-470, inAXL-high cell lines (Fig. 5D and Supplementary Fig. S7C),and siRNAs targeting AXL further confirmed that AXLinhibition mediated the observed synergy (Fig. 5E).

Tubulin-binding taxanes cause a sustained mitotic block,resulting in cell death during mitosis (29). To determinewhether R428 synergistically interacts with antimitotic agentswith distinct mechanisms of action, we tested two aurorakinase inhibitors, PHA-739358 and PF-03814735, and the PLK1(polo kinase) inhibitor BI-2536, which activates the spindlecheckpoint, causing mitotic arrest (Fig. 5A). We observedsynergy between R428 and the aurora kinase and PLK1 inhi-bitors in AXL-expressing mesenchymal cells (Fig. 5A and C).

We also observed a 5-fold shift in the PHA-739358 IC50 uponcombination treatment with R428 for the MDA-MB-231 cells[PHA-739358 alone: IC50, 1.63 mmol/L and in combination withR428 (1 mmol/L), IC50, 0.34 mmol/L (Fig. 5C)]. Thus, AXLinhibition synergistically blocks cell proliferation in combina-tion with antimitotic agents in mesenchymal cancer cells.

AXL kinase inhibition enhances docetaxel activity tosuppress tumor growth

We extended the cell line findings to an in vivo contextusing tumor xenografts. First, we determined that R428 waspharmacologically active in xenografts (Fig. 6A). To assessefficacy, mice bearing 100 to 200 mm3 tumors were treatedwith R428 and/or docetaxel. R428 or docetaxel slightlyinhibited tumor growth; however, cotreatment caused sig-nificant tumor growth suppression (Fig. 6B). Similar effectswere seen with MDA-MB-231 xenografts (Fig. 6C). A modestdecrease in body weight appeared to be docetaxel-depen-dent (Fig. 6B and C). These observations support the poten-tial in vivo utility of AXL inhibitors combined with antimi-totic agents.

MDA-MB-231 xenograft

0 2 4 6 8 10 12 14 16

0 2 4 6 8 10 12 14 16

0

200

400

600

800

Time (days)

Tu

mo

rvo

lum

e(m

m3 )

Vehicle (n = 8)R428 (n = 9)DTX (n = 8)DTX + R428 (n = 9)

HeLa xenograft

0 5 10 15 200

200

400

600

800

Time (days)

Tu

mo

rvo

lum

e(m

m3 )


B C

–20

–10

0

10

20

Time (days)


0 5 10 15 20–20

–10

0

10

20

Time (days)

% B

od

y w

eig

ht

chan

ge

% B

od

y w

eig

ht

chan

ge


A

pS6 (Ser235/236)

Total S6

– – – + + + + + + + + R428

8 h 2 h

pAKT (ser473)

Total AKT

GAPDH

Figure 6. AXL inhibition incombination with docetaxel retardsthe growth of mesenchymal tumorcells in vivo. A, immunoblotsshowing phospho-AKT (pAKT),total AKT, phospho-S6 (pS6), andtotal S6, in individual MDA-MB-231xenograft tumors following R428treatment. B, tumor growth assayshowing the antitumor effect ofR428 in combinationwith docetaxelin HeLa xenografts. Differencesbetween the R428 and thedocetaxel combination groupcompared with the vehicle group(P < 0.0001), and individualtreatment with R428 (P ¼ 0.0001)and docetaxel (P < 0.0001) werecalculated using the two-wayANOVA. Error bars, mean � SEM.The percentage of body weightchange is shown at the bottom.C, tumor growth assay showing theantitumor effect of R428 incombination with docetaxel inMDA-MB-231 xenografts.Differences between the R428 andthe docetaxel combination groupcompared with the vehicle group(P ¼ 0.0033), and individualtreatment with R428 (<0.0001) anddocetaxel (P < 0.0001) comparedwith the combination group werecalculated using the two-wayANOVA. Error bars, mean � SEM.The percentage of body weightchange is shown at the bottom.

Wilson et al.





AXL inhibition enhances cell death during mitosis whencombined with antimitotic agentsTo establish whether AXL inhibition alters mitotic cell fate

when combined with antimitotic agents, we used time-lapsemicroscopy of individual cells to monitor duration of mitosisand cell fate during mitosis (Fig. 7A; Supplementary Fig. S9;Supplementary Movies S1–S4). The duration of completedmitosis in DMSO-treated HeLa cells ranged from 20 to 50minutes (Fig. 7B), and R428 did not affect duration (Fig. 7B andSupplementaryMovie S1 and S2). In contrast, docetaxel causedmitotic slippage generating three or four daughter cells (Fig. 7Aand SupplementaryMovie S3) and a longer duration ofmitosis,as expected (Fig. 7B). Upon cotreatment with docetaxel andR428, the percentage of cells that completed mitosis decreasedsubstantially, and the percentage of cells that died in mitosiswas greatly enhanced from 46% to 93% (Fig. 7A and E andSupplementary Movies S3 and S4).The duration of the mitotic cell death interval was compa-

rable between docetaxel treatment alone and cotreatmentwith R428 (Fig. 7C). However, cotreatment caused most cellsto enter mitosis and die without completing mitosis (Fig. 7A).In contrast, with docetaxel alone, the death in mitosis fateoccurred during the second mitosis; thus, cells had completedone mitotic event before experiencing death in mitosis(explaining why the combined cell fates total >100). To quan-tify this event, we recorded the timestamp of the mitotic deathevent, and observed a significant effect of combination treat-ment. Thus, the death in mitosis event occurred substantiallyearlier (25.14� 0.86 hours for combination versus 41.74� 1.98hours docetaxel alone; Fig. 7D). We similarly observedenhanced mitotic death in combination-treated MDA-MB-231 cells (Supplementary Fig. S9D–S9G). Moreover, theenhanced death in mitosis was dependent on the docetaxelconcentration (Supplementary Fig. S9D and S9G).As with R428, cotreatment with MP-470, or the AXL–Fc

protein, and docetaxel resulted in altered cell fate duringmitosis (Supplementary Fig. S9A and S9B). Similar results wereseen upon cotreatment with PHA-739358 and R428 (Fig. 7A). Incontrast, cotreatment with R428 and gemcitabine did not altermitosis or mitotic fate (Supplementary Fig. S9C). These find-ings support the specificity of AXL inhibition in potentiatingthe mitotic fate effect of antimitotic agents in mesenchymaltumor cells.

AXL's role in mitotic fate is associated with CDC2dysregulationCell death during mitosis is caspase-dependent (30), and

mitotic fate is determined by both caspase activation andcyclin B1 degradation. Enhanced caspase activation and slowercyclin B1 decay promotes death in mitosis, whereas, delay incaspase activation and enhanced cyclin B1 degradation causesmitotic slippage. Docetaxel induced caspase-3/7 activation;however, upon R428 cotreatment, caspase-3/7 activity wasinduced faster and more robustly than with docetaxel (Fig.7F), paclitaxel (Supplementary Fig. S8D), or PHA-739358 (Sup-plementary Fig. S8E) alone. In contrast, no effects on cyclin B1were observed with docetaxel, even in combination with R428(data not shown). Cotreatment with R428 and docetaxel

increased G2–M arrest from 36.2% (docetaxel alone) to51.2% (Fig. 7G). We observed enhanced PARP cleavage follow-ing cotreatment with R428 and either, docetaxel, paclitaxel, orPHA-739358 (Fig. 7H and Supplementary Fig. S8C), confirmingthat the observed death in mitosis event was apoptotic.

PI3K inhibition can enhance docetaxel efficacy in breastcancer models (31), and consistent with this observation,suppression of phospho-AKT and phospho-S6 by R428 wasfurther enhanced by docetaxel (Supplementary Fig. S10A). Acritical regulator of mitotic entry is the cyclin-dependentkinase-1 (CDC2; ref. 32), via a regulatory phosphorylation atTyr15 (33), and cotreatmentwithR428 anddocetaxel enhancedand sustained CDC2 dephosphorylation (Fig. 7I and Supple-mentary Fig. S10B). Furthermore, suppression of phospho-CDC2 was observed in R428-treated xenografts (Supplemen-tary Fig. S10C); however, CDC2 protein levels are not affectedby AXL inhibition (Supplementary Fig. S11). These observa-tions suggest that CDC2 is regulated by AXL, and uponexposure to antimitotic agents, cancer cells rapidly entermitosis and remain in mitosis until the death in mitosis event,implicating CDC2 regulation in sensitization to antimitoticagents in mesenchymal cancer cells by AXL inhibitors.

DiscussionEMT is an increasingly recognized driver of tumor progres-

sion, and accumulating evidence implicates EMT in bothinnate and acquired resistance to various anticancer agents.Our RNA-seq analysis confirmed that AXL expression is indeedwidely associated with amesenchymal phenotype, particularlywithin TNBCs and erlotinib-resistant NSCLCs. However, thelow percentage of AXL-positive tumors in TNBC does notcorrelate with the in vitro findings and remains to be fullyunderstood. Targeting AXL might therefore be an attractivetherapeutic approach to overcoming resistance associatedwith EMT.

Our observations do not support a role for AXL inhibition inovercoming acquired drug resistance in the EMT context. Thismay reflect, in part, the fact that many of the tool compoundsused in recent reports are multikinase inhibitors, lacking AXLselectivity (4, 17). Furthermore, there may be off-target cyto-toxic effects of AXL-targeted RNAi reagents. Notably, a recentlyreported analysis of NSCLC cells with acquired erlotinib resis-tance and increased AXL levels similarly excluded a functionalrole for AXL (34).

Although our findings do not support a rationale for AXLinhibition in the context of acquired TKI resistance, we foundthat AXL inhibition may be effective in mesenchymal tumors,specifically in combination with antimitotic agents. In TNBC,patients typically present a significant clinical challenge, asthey do not respond to the various targeted cancer therapiesdue to an apparent lack of RTK activation. However, patientswith TNBC do show some response to taxane-based chemo-therapy (35), and our studies suggest that combining antimi-totics with AXL inhibition may be an appropriate combinationtherapy in this disease setting.

Taxane-based therapies exert their cytotoxic effects bybinding and stabilizing microtubules, resulting in cell-cycle






Cou

ntC

ount

Cou

ntC

ount

FL2-A

FL2-A FL2-A

FL2-A

R428 +−+−0

10

20

30

40

50

DTX (3 nmol/L)DMSO

HeLa

Dur

atio

n of

mito

tic

deat

h in

terv

al (

h)+−+−+−

0

2

4

6

8

10

12 HeLa

R428

PHA (100 nmol/L)DTX (3 nmol/L)DMSO

Dur

atio

n of

com

plet

edm

itosi

s (h

)

0 10 20 30 40 50 60 70 80

DMSO

R428

DTX

DTX + R428

Time of mitotic death event (h)

P <

0.0

001

B DC

E

H

I

DM

SO

D

TX

−R428 +R428

DT

X

−R428 +R428 F

PARP

Cleaved PARP

GAPDH

R428

1 nm

ol/L

3 nm

ol/L

PHA DTX

A

pCDC2 (Tyr15)

pCDC2 (Tyr15)

pCDC2 (Tyr15)

pCDC2 (Tyr15)

Total CDC2

Total CDC2

Total CDC2

Total CDC2

DMSO

R428

DTX

COMBO

DMSO

R428

DTX

COMBO

0.5 1 2 3 4 5 6 7 8 (h)

DMSOR428

+−+−+−0

1

2

3

4

5

6

DTX

36 h24 h18 h

Cas

pase

-3/7

fold

cha

nge

lum

ines

cenc

e (

RF

U)

G

R428 (1 µmol/L)DMSOHeLa DTX (3 nmol/L) DTX + R428 PHA+ R428PHA (100 nmol/L)

100100100100100100Total cells

98%98%100%100%100%100%Enter Mitosis

6%74%6%46%99%99%Complete mitosis

73%15%1%8%0%0%Mitotic slippage

19%9%93%46%1%1%Die in mitosis

30%3%3%39%0%0%Die in interphase

0.5

μmol

/L

50 n

mol

/L

− + − + − + − + − +

300

200

100

0

300

200

100

0

0 200 400 600 800 1K 0 200 400 600 800 1K

0 200 400 600 800 1K0 200 400 600 800 1K

200

150

100

50

0

200

150

100

50

0

DMSO

DTX DTX +R428

R42865.4% 61.4%

50.7%

7.15%

36.2%

40.9%

7.01%

51.2%

6.43% 6.96%

25.1% 29.4%

Figure 7. Axl inhibition in combination with antimitotic agents promotes mitotic death. A, table representing the fate of 100 individual cells followingexposure to R428 (1 mmol/L), docetaxel (DTX; 3 nmol/L), PHA-739358 (PHA; 100 nmol/L), or the indicated combinations based on a 72 hoursmicroscopy assay. B, scatter plot demonstrating the duration of completed mitosis (hours) for 100 individual cells upon each drug treatment: R428(1 mmol/L), docetaxel, PHA-739358 (PHA), and combinations in HeLa cells in a 72-hour assay. Red line, mean. C, scatter plot demonstrating the durationof the mitotic death interval in hours of 100 individual cells upon each drug treatment: R428 (1 mmol/L), docetaxel, and the combination in HeLa cells in a72-hour assay. Red line, mean. D, scatter plot demonstrating a significant difference as assessed using the Student t test (HeLa P � 0.0001) ofthe time of mitotic death in hours of 100 individual cells upon each drug treatment: R428 (1 mmol/L), docetaxel (3 nmol/L), and the combination in HeLacells in a 72-hour assay. Red line, mean. E, microscopy images demonstrating mitosis following treatment with R428 (1 mmol/L), docetaxel (3 nmol/L),or the combination in HeLa cells at 18 hours (left) and 36 hours (right). F, bar graph representing caspase-3/7 activation following treatmentwith R428 (1 mmol/L), docetaxel (10 nmol/L), or the, combination in HeLa cells during a 18-, 24-, and 36-hour time course. Error bars, mean � SEM. G,histogram plots demonstrating DNA content upon exposure of HeLa cells to R428 (1 mmol/L) in combination with docetaxel (10 nmol/L) after 8 hours.H, immunoblot demonstrating apoptosis (cleaved PARP) in HeLa cells after PHA-739358 (PHA) or docetaxel treatment in combination with R428(1 mmol/L) for 72 hours. I, Immunoblots showing phospho-CDC2 (pCDC2) and total CDC2 in HeLa cells following treatment with R428 (1 mmol/L),docetaxel (3 nmol/L), or the combination (COMBO) during an 8-hour assay period.

Wilson et al.





arrest in mitosis. The response to antimitotic agents can varysignificantly among cancer cells. Thus, some cells undergodeath during prolongedmitosis, while others undergo slippageand die in interphase (29). The degree of caspase activation andthe level of cyclin B1 degradation can determine cell fate duringmitosis (30). Several treatments reportedly enhance the effi-cacy of taxane-based therapy, such as navitoclax (BCL-2/BCL-XL inhibitor), and the PI3K inhibitor GDC-0941 (22, 31, 36). Inthese studies, the enhanced efficacy of combination treatmentreflected a decrease in the duration of mitosis until mitotic celldeath, an enhanced rate of apoptosis, the downregulation ofcyclin D1 and pAKT, and altered kinetics of BCL-2 familyprotein stabilization. AXL engages some of these pathways,particularly the PI3K/AKT pathway (13, 14), which we foundwas suppressed by R428, particularly upon cotreatment withdocetaxel, which may contribute to the observed synergy. Ourfindings also indicate that the observed synergy with AXLinhibition reflects altered kinetics of dephosphorylation ofCDC2, a cell cycle–dependent kinase that governs mitoticentry. However, it remains unclear as to how AXL, relative toother RTKs that transduce signals to largely overlappingdownstream effectors, specifically regulates this signalingevent in the context of mitosis.

Disclosure of Potential Conflicts of InterestT. Pham has ownership interest (including patents) in Roche/Genentech.

R.M. Neve received other commercial research support from Genentech and

has ownership interest (including patents) in the same. H. Koeppen hasownership interest (including patents) in Roche Stock. No potential conflictsof interest were disclosed by the other authors.

Authors' ContributionsConception and design: C. Wilson, L. Belmont, A. Ashkenazi, J. SettlemanDevelopment of methodology: X. Ye, T. Pham, L. BelmontAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): X. Ye, T. Pham, E. Lin, S. Chan, E. McNamara,H. Koeppen, R.L. YauchAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): C. Wilson, R.M. Neve, H. Koeppen, A. Ashkenazi,J. SettlemanWriting, review, and/or revision of the manuscript: C. Wilson, T. Pham,R.M. Neve, H. Koeppen, A. Ashkenazi, J. SettlemanAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): X. Ye, E. Lin, S. Chan, J. SettlemanStudy supervision: J. Settleman

AcknowledgmentsThe authors thank members of the Settleman laboratory and Amitabha

Chaudhuri for helpful discussions, members of the Genentech cell line screeningfacility, and cell line bank. The authors thank Nancy Yu and Tom Januario forproviding cell line lysates, and the in vivo pharmacology group for guidance onxenograft studies.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicate thisfact.

Received April 8, 2014; revised August 1, 2014; accepted August 6, 2014;published OnlineFirst August 14, 2014.

References1. Thiery JP, Acloque H, Huang RY, Nieto MA. Epithelial–mesen-

chymal transitions in development and disease. Cell 2009;139:871–90.

2. Chung JH, Rho JK, Xu X, Lee JS, Yoon HI, Lee CT, et al. Clinicaland molecular evidences of epithelial to mesenchymal transitionin acquired resistance to EGFR-TKIs. Lung Cancer 2011;73:176–82.

3. Rho JK, Choi YJ, Lee JK, Ryoo BY, Na II, Yang SH, et al. Epithelial tomesenchymal transition derived from repeated exposure to gefitinibdetermines the sensitivity to EGFR inhibitors in A549, a non–small celllung cancer cell line. Lung Cancer 2009;63:219–26.

4. Zhang Z, Lee JC, Lin L, Olivas V, Au V, LaFramboise T, et al. Activationof the AXL kinase causes resistance to EGFR-targeted therapy in lungcancer. Nat Genet 2012;44:852–60.

5. Sequist LV, Waltman BA, Dias-Santagata D, Digumarthy S, Turke AB,Fidias P, et al. Genotypic and histological evolution of lung cancersacquiring resistance to EGFR inhibitors. Sci Transl Med 2011;3:75ra26.

6. Linger RM, Keating AK, Earp HS, Graham DK. Taking aim at Mer andAxl receptor tyrosine kinases as novel therapeutic targets in solidtumors. Expert Opin Ther Targets 2010;14:1073–90.

7. Gjerdrum C, Tiron C, Hoiby T, Stefansson I, Haugen H, Sandal T, et al.Axl is an essential epithelial-to-mesenchymal transition-induced reg-ulator of breast cancer metastasis and patient survival. Proc Natl AcadSci U S A 2010;107:1124–9.

8. ShiehYS, LaiCY,KaoYR,ShiahSG,ChuYW,LeeHS, et al. Expressionof axl in lung adenocarcinoma and correlation with tumor progression.Neoplasia 2005;7:1058–64.

9. Koorstra JB,Karikari CA, FeldmannG,BishtS,RojasPL,OfferhausGJ,et al. The Axl receptor tyrosine kinase confers an adverse prognosticinfluence inpancreatic cancer and represents anew therapeutic target.Cancer Biol Ther 2009;8:618–26.

10. Chen PX, Li QY, Yang Z. Axl and prostasin are biomarkers for prog-nosis of ovarian adenocarcinoma. Ann Diagn Pathol 2013;17:425–9.

11. Linger RM, Keating AK, Earp HS, Graham DK. TAM receptor tyrosinekinases: biologic functions, signaling, and potential therapeutic target-ing in human cancer. Adv Cancer Res 2008;100:35–83.

12. Nagata K, Ohashi K, Nakano T, Arita H, Zong C, Hanafusa H, et al.Identification of the product of growth arrest-specific gene 6 as acommon ligand for Axl, Sky, and Mer receptor tyrosine kinases. J BiolChem 1996;271:30022–7.

13. Hong CC, Lay JD, Huang JS, Cheng AL, Tang JL, Lin MT, et al.Receptor tyrosine kinase AXL is induced by chemotherapy drugs andoverexpression of AXL confers drug resistance in acute myeloidleukemia. Cancer Lett 2008;268:314–24.

14. Lee WP, Wen Y, Varnum B, Hung MC. Akt is required for Axl-Gas6signaling to protect cells from E1A-mediated apoptosis. Oncogene2002;21:329–36.

15. KeatingAK,KimGK, JonesAE,DonsonAM,WareK,MulcahyJM, et al.Inhibition ofMer and Axl receptor tyrosine kinases in astrocytoma cellsleads to increased apoptosis and improved chemosensitivity. MolCancer Ther 2010;9:1298–307.

16. Vuoriluoto K, Haugen H, Kiviluoto S, Mpindi JP, Nevo J, Gjerdrum C,et al. Vimentin regulates EMT induction by Slug and oncogenic H-Rasandmigration bygoverningAxl expression in breast cancer.Oncogene2011;30:1436–48.

17. Byers LA, Diao L, Wang J, Saintigny P, Girard L, Peyton M, et al. Anepithelial–mesenchymal transition gene signature predicts resistanceto EGFR and PI3K inhibitors and identifies Axl as a therapeutic targetfor overcoming EGFR inhibitor resistance. Clin Cancer Res 2013;19:279–90.

18. YeX, Li Y, Stawicki S,CoutoS, Eastham-Anderson J,KallopD, et al. Ananti-Axl monoclonal antibody attenuates xenograft tumor growth andenhances the effect of multiple anticancer therapies. Oncogene2010;29:5254–64.

19. Liu L, Greger J, Shi H, Liu Y, Greshock J, Annan R, et al. Novelmechanismof lapatinib resistance inHER2-positive breast tumor cells:activation of AXL. Cancer Res 2009;69:6871–8.






20. Dufies M, Jacquel A, Belhacene N, Robert G, Cluzeau T, Luciano F,et al. Mechanisms of AXL overexpression and function in imatinib-resistant chronic myeloid leukemia cells. Oncotarget 2011;2:874–85.

21. Sun Y, Wang BE, Leong KG, Yue P, Li L, Jhunjhunwala S, et al.Androgen deprivation causes epithelial–mesenchymal transition in theprostate: implications for androgen-deprivation therapy. Cancer Res2012;72:527–36.

22. Tan N, Malek M, Zha J, Yue P, Kassees R, Berry L, et al. Navitoclaxenhances the efficacy of taxanes in non–small cell lung cancermodels.Clin Cancer Res 2011;17:1394–404.

23. Lehmann BD, Bauer JA, Chen X, Sanders ME, Chakravarthy AB, ShyrY, et al. Identification of human triple-negative breast cancer subtypesand preclinical models for selection of targeted therapies. J Clin Invest2011;121:2750–67.

24. D'Alfonso TM, Hannah J, Chen Z, Liu Y, Zhou P, Shin SJ. Axl receptortyrosine kinase expression in breast cancer. J Clin Pathol 2014;67:690–6.

25. Yauch RL, Januario T, Eberhard DA, Cavet G, Zhu W, Fu L, et al.Epithelial versus mesenchymal phenotype determines in vitro sensi-tivity and predicts clinical activity of erlotinib in lung cancer patients.Clin Cancer Res 2005;11(24 Pt 1):8686–98.

26. Derynck R, Akhurst RJ, Balmain A. TGF-beta signaling in tumorsuppression and cancer progression. Nat Genet 2001;29:117–29.

27. Mani SA, Guo W, Liao MJ, Eaton EN, Ayyanan A, Zhou AY, et al. Theepithelial–mesenchymal transition generates cells with properties ofstem cells. Cell 2008;133:704–15.

28. Holland SJ, Pan A, Franci C, Hu Y, Chang B, Li W, et al. R428, aselective small molecule inhibitor of Axl kinase, blocks tumor spread

and prolongs survival in models of metastatic breast cancer. CancerRes 2010;70:1544–54.

29. Rieder CL, Maiato H. Stuck in division or passing through: whathappens when cells cannot satisfy the spindle assembly checkpoint.Dev Cell 2004;7:637–51.

30. Gascoigne KE, Taylor SS. Cancer cells display profound intra- andinterline variation following prolonged exposure to antimitotic drugs.Cancer Cell 2008;14:111–22.

31. Wallin JJ, Guan J, Prior WW, Lee LB, Berry L, Belmont LD, et al.GDC-0941, a novel class I selective PI3K inhibitor, enhancesthe efficacy of docetaxel in human breast cancer models byincreasing cell death in vitro and in vivo. Clin Cancer Res 2012;18:3901–11.

32. Santamaria D, Barriere C, Cerqueira A, Hunt S, Tardy C, Newton K,et al. Cdk1 is sufficient to drive the mammalian cell cycle. Nature2007;448:811–5.

33. Norbury C, Blow J, Nurse P. Regulatory phosphorylation of thep34cdc2 protein kinase in vertebrates. EMBO J 1991;10:3321–9.

34. Suda K, Mizuuchi H, Sato K, Takemoto T, Iwasaki T, Mitsudomi T. Theinsulin-like growth factor 1 receptor causes acquired resistance toerlotinib in lung cancer cells with thewild-type epidermal growth factorreceptor. Int J Cancer 2014;135:1002–6.

35. Hudis CA, Gianni L. Triple-negative breast cancer: an unmet medicalneed. Oncologist 2011;16(suppl 1):1–11.

36. ShaoN,Chen LH, YeRY, Lin Y,WangSM. The depletion of interleukin-8 causes cell cycle arrest and increases the efficacy of docetaxelin breast cancer cells. Biochem Biophys Res Commun 2013;431:535–41.


Wilson et al.




2014;74:5878-5890. Published OnlineFirst August 14, 2014.Cancer Res Catherine Wilson, Xiaofen Ye, Thinh Pham, et al. DrugsAXL Inhibition Sensitizes Mesenchymal Cancer Cells to Antimitotic

Updated version

10.1158/0008-5472.CAN-14-1009doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2014/08/21/0008-5472.CAN-14-1009.DC2

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/74/20/5878.full#ref-list-1

This article cites 36 articles, 13 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/74/20/5878.full#related-urls

This article has been cited by 29 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/74/20/5878To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-14-1009

http://cancerres.aacrjournals.org/content/suppl/2014/08/21/0008-5472.CAN-14-1009.DC2

http://cancerres.aacrjournals.org/content/74/20/5878.full#ref-list-1

http://cancerres.aacrjournals.org/content/74/20/5878.full#related-urls

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

mailto:[email protected]

http://cancerres.aacrjournals.org/content/74/20/5878


AXL Inhibition Sensitizes Mesenchymal Cancer Cells to ...(TKI).Thus,drug-sensitivecancercelllines...

Documents

Transcript of AXL Inhibition Sensitizes Mesenchymal Cancer Cells to ...(TKI).Thus,drug-sensitivecancercelllines...