-1 -0.5 0 0.5 1

Tumor A1

Tumor A2

Tumor A3

Tumor B1

Tumor B2

Tumor B3

Tumor C1

Tumor C2

Tumor C3

Tumor A1

Tumor A2

Tumor A3

Tumor B1

Tumor B2

Tumor B3

Tumor C1

Tumor C2

Tumor C3

Tumor A1

Tumor A2

Tumor A3

Tumor B1

Tumor B2

Tumor B3

Tumor B4

Tumor C1

Tumor C2

Tumor C3

Tumor A1

Tumor A2

Tumor A3

Tumor A4

Tumor B1

Tumor B2

Tumor B3

Tumor C1

Tumor C2

Tumor C3

Tumor D1

Tumor D2

Tumor D3

Tumor E1

Tumor E2

Tumor E3

Tumor F1

Tumor F2

Tumor F3

Tumor B1

Tumor B2

Tumor B3

Tumor B4

Tumor D1

Tumor D2

Tumor D3

Tumor E1

Tumor E2

Tumor E3

Tumor A1

Tumor A2

Tumor A3

Tumor B1

Tumor B2

Tumor B3

Tumor C1

Tumor C2

Tumor C3

Tumor E1

Tumor E2

Tumor E3

SNU

5 +

paz

op

anib

SNU

5 +

veh

icle

MK

N4

5 +

Paz

op

anib

Tiv

anti

nib

MK

N4

5 +

paz

op

an

ibM

KN

45

+ T

ivan

tin

ibM

KN

45

+ v

eh

icle

Vimentin:E-Cadherin (Log)

Quantitative immunofluorescence assessment of MET and epithelial-to-mesenchymal transition (EMT) biomarker modulation by antiangiogenic inhibitors in xenograft tumor tissues

Tony Navas1, Scott M. Lawrence1, Donna Butcher2, Lindsay M. Dutko2, Melinda G. Hollingshead3, Robert J. Kinders1, Ralph E. Parchment1, Donald P. Bottaro4, W. Marston Linehan4, Joseph E. Tomaszewski5, Apurva K. Srivastava1, and James H. Doroshow5

1Laboratory of Human Toxicology and Pharmacology, Applied/Developmental Research Directorate, Leidos Biomedical Research, Inc., Frederick National Laboratory for Cancer Research, Frederick, Maryland 21702; 2Pathology/Histotechnology Laboratory, Laboratory Animal Sciences Program, Leidos Biomedical Research, Inc., Frederick National Laboratory for Cancer Research, Frederick, Maryland 21702; 3Biological Testing Branch, Developmental Therapeutics Program, National Cancer Institute, Frederick MD 21702; 4Urologic Oncology Branch, Center for Cancer Research, National Cancer Institute, Bethesda MD 20810; 5Division of Cancer Treatment and Diagnosis and Center for Cancer Research, National Cancer Institute, Bethesda MD 20810

MET IFA • We have developed a specific and sensitive multiplex pY1235-MET and total c-Met IFA on FFPE tissues. • We have demonstrated assay “fit for purpose” application of the multiplex MET IFA using SNU5 xenograft tissues treated

with crizotinib as well as with pazopanib, tivantinib, or the drug combination. • We have shown by IFA that tivantinib had no activity on its own in inhibiting pY1235-MET in SNU5 xenografts, that pazopanib

induced pY1235-MET compared to vehicle; and that the drug combination inhibited pY1235-MET compared to pazopanib. • We have used Definiens analysis to enumerate % pY1235-MET/total Met levels in tissue sections and have shown good

concordance with ELISA data taken from correlated tumor samples from the same sets of experiments.

EMT IFA • EMT IFA analysis of gastric tumors showed that vehicle-treated MKN45 has a relatively epithelial phenotype, while

vehicle-treated SNU5 showed a mesenchymal phenotype. • Pazopanib-treated MKN45 (4/6 tumors) appears to show varying levels of epithelial-to-mesenchymal transition compared

to vehicle. • Pazopanib/tivantinib combination–treated groups (3/3 evaluable tumors) showed a generally more epithelial phenotype,

while 2/5 samples had no tumor content by H&E. 1. Sennino and Mcdonald (Oct 2012) Controlling escape from angioganesis inhibitors. Nature Reviews Cancer 12, 699-709 2. Srivastava et. al. (ASCO 2011) Development and validation of biomarker assays to assess pharmacodynamic modulation of MET 3. Srivastava et al. (AACR 2012) Preclinical assessment of MET modulation by a VEGFR inhibitor/MET inhibitor combination that shows additive antitumor efficacy 4. Srivastava et al. (2014) Development and Validation of MET Assays to Assess Pharmacodynamic Response in Tumor Tissues (submitted)

Introduction

Abstract # 1049

Summary and Conclusions

Results

GAPDH

Methods Reagents. Mouse monoclonal antibody against total MET (clone Met4) was from Van Andel Research Institute Technologies, Inc. Two rabbit monoclonal antibodies specific to pY1235-MET (clone 23111; with undetectable reactivity to pY1234-MET) and pY1356-MET were developed at Epitomics Inc., by using peptide antigens corresponding to the MET sequence surrounding these amino acids . E-Cadherin-AF488 mouse monoclonal antibody (clone 32) was from BD Biosciences, and Vimentin-AF647 (clone V9) was from Santa Cruz Biotechnology. Animal models and drug administration. Athymic nude mice (NCr; Animal Production Program, FNLCR) were implanted with the human cancer cell lines A549 (lung carcinoma), MDA-MB-231 (breast carcinoma), HT-29 (colon carcinoma), GTL-16 (gastric carcinoma; MET amplified), MKN 45 (gastric carcinoma; MET amplified), and SNU5 (gastric carcinoma; MET amplified) by subcutaneous injection. The NCI Animal Production Program, FNLCR, is accredited by Association for Assessment and Accreditation of Laboratory Animal Care International and follows the USPHS Policy for the Care and Use of Laboratory Animals. All the studies were conducted according to an approved animal care and use committee protocol.

MET inhibitors. PF 02341066 (NSC D749769-Y, crizotinib), tivantinib (NSC 758242), and VEGFR inhibitor pazopanib (NSC 737754) were synthesized at the Developmental Therapeutics Program (DTP), NCI. PF02341066 and pazopanib were administered by oral gavage in a saline vehicle, and PHA 665752 by intraperitoneal (IP) injections in a vehicle composed of 10% DMSO in saline. Tivantinib was administered orally in a PEG 400:20% vitamin E TPGS solution (60:40) vehicle. Xenograft biopsy and tumor quarter collection. Whole xenograft tumors were collected on the same schedule as tumor biopsies by standard dissection methods. Specimens were cut into two to four equal pieces with fine-point scissors and placed into Sarstedt tubes that were pre-cooled in liquid nitrogen and was immediately flash frozen in an o-ring-sealed, conical bottom screw-capped 1.5-mL Sarstedt cryovial by touching the biopsy to the inside of the tube within one minute of excision. Tubes were pre-cooled on liquid nitrogen. MET immunofluorescence analysis. HT29, GTL16, and A549 cancer cell lines (FNLCR Repository) were treated with DMSO (0.1% w/v), 100 nM crizotinib, or 100 nM sorafenib for 24 h at 37°C. Frozen tissues or cells were fixed in 10% neutral buffered formalin for 24 h (Sigma Aldrich), and then embedded in paraffin. Paraffin-embedded tissue was cut into 5 µm sections and placed on slides, and staining was performed in a Bond-max Autostainer (Leica Microsystems). Antigen retrieval was performed using Bond Epitope Retrieval Solution 2 (Leica) at 100°C for 10 min. For antigen detection, 10 g/mL rabbit monoclonal anti-pY1235–MET (clone 23111) was used as the primary antibody, followed by 10 µg/mL goat anti-rabbit AF488 (Life Technologies) as the secondary antibody. Multiplex immunofluorescence was performed on formalin-fixed paraffin-embedded tumor tissues; 10 g/mL rabbit monoclonal anti-pY1235–MET (clone 23111) and 5 g/mL mouse monoclonal anti-intact MET (clone MET4, Van Andel Research Institute) were used as primary antibodies, followed by 10 µg/mL goat anti-rabbit AF488 and 10 µg/mL anti-mouse AF660 (Life Technologies) secondary antibodies. Image acquisition and analysis were performed on a wide-field fluorescent and confocal microscope (Nikon 90i Andor Camera, NIS Elements Software).

This project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under Contract No. HHSN261200800001E. This research was supported [in part] by the Developmental Therapeutics Program in the Division of Cancer Treatment and Diagnosis of the National Cancer Institute. Frederick National Laboratory for Cancer Research is accredited by AAALAC International and follows the Public Health Service Policy for the Care and Use of Laboratory Animals. Animal care was provided in accordance with the procedures outlined in the "Guide for Care and Use of Laboratory Animals" (National Research Council; 1996; National Academy Press; Washington, D.C.).

Demonstration of pY1235-MET antibody specificity by Western Analysis

Anti-pY1235-MET (clone 23111)

145 kD, MET ~145 kD 145 kD, MET 145 kD, MET 170 kD, Pro-MET 145 kD, MET

0 5 25 crizotinib (nm)

23111 + pY1235 peptide

23111 + pY1234 peptide

23111 + rRON

23111

Figure 2. The rabbit monoclonal 23111 was highly specific for pY1235-MET, with undetectable cross-reactivity to pY1234-MET and to cytoplasmic RON. Specificity of pY1235-MET antibody clone 23111 demonstrated by Western blot staining of lysates from GTL-16 cells treated with increasing amounts of the MET inhibitor crizotinib (top panel). Similar blots pre-incubated with pY1235-MET peptide (2nd panel), pY1234-MET peptide (3rd panel), and recombinant cytoplasmic RON (4th panel). The bottom panel shows total intact MET staining with C-terminal antibody; similar total intact MET staining was observed with all blots (Ref 4).

Demonstration of pY1235-MET IFA staining specificity using specific MET and non-specific TK inhibitors on FFPE tissues by confocal microscopy:

GTL16 HT29 A549 pY1235-MET

GTL16

HT29 A549 Total MET

DMSO

100 nM crizotinib

(PF02341066)

100 nM sorafenib

Figure 3. Specificity of the pY1235-MET monoclonal antibody is demonstrated in formalin-fixed, paraffin-embedded GTL-16 (MET-amplified), A549, and HT-29 (paracrine or autocrine) cancer cells treated in vitro with 100 nM crizotinib (specific MET-kinase inhibitor) or 100 nM sorafenib (nonspecific TK inhibitor) for 4 h. Total MET staining of the same tissue sections using Met4 Mab shows differences in expression levels among the various cell lines, but no changes in expression were observed with drug treatment. Staining was performed in a Bond-max Autostainer (Leica Microsystems), and slides were imaged on a confocal microscope (Nikon 90i Andor Camera, NIS Elements Software).

HGF specifically induces pY1235-MET IFA staining, which was completely inhibited by crizotinib in MET-expressing cell lines in vitro

A549 + HGF + 100 nM crizotinib

A549 + HGF 15 min

A549

10 2ndary

only

pY1235-MET (ug/ml) 2.5 1.0

HT29

HT29 + HGF 24h

HT29 + HGF + 100 nM crizotinib

DAPI pY1235-MET Merge

Figure 4. Immunofluorescence staining of pY1235-MET on A549 and HT29 cell lines was specifically induced by HGF stimulation but completely blocked by treatment with crizotinib in vitro. A549 or HT29 cells were grown on chamber slides to 80% confluency before being serum starved for 24 h. Cells were incubated with or without 100 nM crizotinib for 4 h prior to stimulation with 20 ng/ml HGF for 15 min (A549) or for 24 h (HT29) in serum-free media. Cells were fixed in 10% NBF for 15 min, permeabelized in 0.3% Triton X-100, and stained with increasing concentrations of pY1235-MET antibody (A549), followed by 10 ug/ml anti-rabbit AF546 secondary. Slides were imaged with a Nikon 90i Andor Camera, and analyzed using NIS Elements Software.

Multiplex IFA staining and Definiens quantitation analysis of pY1235-MET and total MET on FFPE tissues

GTL16 HT29 MDA-MB-231 SNU1

pY1235-MET

AF488

Total MET AF660

Merge

GTL16 HT29 SNU1

Marker Masks

Nuclear Masks

pY1235-MET (Green) total Met (Red) Marker Masks

DAPI (Blue) Nuclear Mask and Segmentation

MDA-MB-231

0

200

400

600

800

1000

1200

GTL16 HT29 MDA-MB-231 SNU1

Mar

ker

Are

a (u

m2 )

/ 1

00

ce

lls

pY1235-METtotal MET

0

10

20

30

40

50

60

70

GTL16 HT29 MDA-MB-231 SNU1

% p

Y1

23

5-M

ET/t

ota

l MET

(M

arke

r A

rea

/ C

ell)

% pY1235-MET/total Met

Figure 5. Quantitative assessment of pY1235‐MET was determined in xenograft tissue models that demonstrate very high IFA expression of pY1235‐MET and total MET via gene amplification (GTL‐16), lower expression in HGF-induced paracrine or autocrine cell lines (A549 and HT29), and no detectable staining for pY1235‐MET in negative controls (SNU‐1 and MDA‐MB‐231).

pY1235-MET and total MET IFA staining of FFPE renal cell carcinoma tissue

Merge

pY1235-MET DAPI

total MET

40x

Renal Patient (20x) H1112R c-Met germline mutation

Figure 6. MET multiplex immunofluorescence staining of excised formalin-fixed, paraffin-embedded HPRCC tumor tissue from a patient harboring an H1112R c-Met germline mutation (left panel). Merged and individual fluorescent images are shown on the right, with total MET (pink; primarily membrane and cytoplasmic), pY1235-MET (green; primarily nuclear), and DAPI (blue; nuclear). Magnification, 40x.

Inhibition of VEGF signaling by pazopanib reduces tumor growth in some preclinical models. However, angiogenic inhibition is often accompanied by enhanced hypoxic conditions in tumors that lead to increased c-Met expression and activity as well as to an epithelial-to-mesenchymal transition (EMT)-like phenotype, which leads ultimately to tumor invasion and metastasis (Fig.1A, Ref 1). Co-inhibition of c-Met can block these effects, providing a potential mechanism to overcome increased invasion in the face of antiangiogenic therapy.

We have previously reported the development of ELISA assays for intact MET, pY1235-MET, and pY1356-MET in tumor lysates (Fig. 1B) (Ref. 2) and measured drug-related biomarker changes induced by the anti-VEGF and anti-MET combination therapy in gastric xenograft models (Ref 3).

Here, we report the development of a multiplex quantitative immunofluorescence assay (qIFA) for simultaneous detection of pY1235-MET and total MET in formalin-fixed paraffin-embedded (FFPE) tissues to determine whether pY1235-MET or total MET is induced by anti-VEGF inhibitors due to increased EMT transition and whether this effect could be reversed by the drug combination with a MET inhibitor, tivantinib (ARQ197) in FFPE tissue samples.

Preclinical study designs of MET inhibitor and VEGFR/MET inhibitor combination studies

A. SNU5 xenografts treated with crizotinib

B. SNU5 and MKN45 xenografts treated with pazopanib, tivantinib (ARQ197) or the drug combination

C. PD sample collection diagram

Doses

4 hr Post-dose 4

Biopsy collection

Day4

Predose

1………….4

Group # Mice Drug Dose Schedule Route

1 8 water 0.1 ml/10 grams BW QDx4 PO

2 8 crizotinib 25 mg/kg QDx4 PO

3 8 crizotinib 12.5 mg/kg QDx4 PO

4 8 untreated

tumors untreated

ELISA IFA

Tissue microarray multiplex staining of pY1235-MET and total MET on FFPE tissues from SNU5 xenografts treated with crizotinib in vivo

• pY1235-MET (Green) • total MET (Pink) • DAPI (Blue)

Definiens quantitation analysis of pY1235-MET and total MET on FFPE SNU5 xenograft tissues Vehicle (Water)

12.5 mg/kg crizotinib

25 mg/kg crizotinib

17

18

19

20

21

22

23

24

25

26

27

28

pY1235-MET (Green) total Met (Red) Marker Masks

***P<0.0001

Regression analysis showing concordance between % pY1235-MET/total MET IFA and ELISA data for SNU5 xenograft tissues treated with crizotinib in vivo

MET multiplex IFA vs ELISA analysis of %pY1235-MET/total MET in FFPE tissues from SNU5 xenografts treated with VEGF inhibitor (pazopanib), MET inhibitor (tivantinib; ARQ197) or the drug combination

ELISA

IFA

Definiens analysis of EMT IFA multiplex-stained SNU5 and MKN45 FFPE tissues treated with VEGF inhibitor (pazopanib), MET inhibitor (tivantinib) or the drug combination in vivo

EMT multiplex IFA staining and Definiens analysis of control FFPE xenograft tissues

b-Catenin- AF546

Vimentin AF647

E-Cadherin- AF488

DAPI

Direct-Conjugate Mab Source/Vendor Mab Clone Direct-

Conjugate Total mg

Total Vol ml

Direct Conjugate Mab Conc.

(mg/ml)

Moles Alexa-Fluor

per mole protein Catalog # Lot #

E-Cadherin-AF488 BD Bioscience 36

Mouse Mab 0.2 mg 4.0 ml 0.05 mg/ml 4.0 560061 39472

Betacatenin-AF546 Epitomics

Custom Conjugation E247

Rabbit Mab 3.69 mg 3.0 ml 1.23 mg/ml 7.0 C47215

96780A Lot #2

Vimentin-AF647 Santa Cruz

Biotechnology V9

Mouse Mab 5.0 mg 5 ml 1 mg/ml 3.0

sc-6260 AF647

KO813

EMT Multiplex IFA Assay Panel

Figure 8. Tissue microarray cores derived from SNU5 xenografts treated with vehicle (water), 12.5 mg/kg crizotinib, or 25 mg/kg crizotinib (QDx4, PO) 4 h after the last dose. Three TMA cores per tumor were collected from 4 different animals per treatment group. Tumors were stained with pY1235-MET (AF488) and total MET (AF647), and scanned by Aperio.

Figure 7. Preclinical study designs of (A) MET inhibitor (crizotinib) or (B) VEGFR (pazopanib)/MET inhibitor (tivantinib) single or combination studies in SNU5 and MKN45 gastric xenograft models for MET and EMT IFA assay “fit for purpose” biomarker studies. (C) PD sample collection diagram of tumor samples from preclinical studies for MET and EMT IFA and ELISA assays.

Figure 11. Scatter graphs (left) showing a comparison of % pY1235-MET/total MET IFA and ELISA values from 5-6 tumor quadrants from each treatment group of SNU5 xenografts treated with vehicle, pazopanib, tivantinib or the drug combination (QD or BID) based on the preclinical study plan shown in Figure 7B. Each point in the IFA graph represents the mean % pY1235-MET/total MET of 12 regions of interest (ROIs). Each ROI , containing 2–3,000 non-necrotic tumor cells, was randomly selected per tumor sample. Regression analysis (above) shows correlation between IFA values expressed as the ratio of marker area/# of nuclei vs. ELISA values expressed as pM/ug total protein, from each of the treatment groups

Vehicle (Water) 25 mg/kg crizotinib

17-1 17-2 17-3

18-1 18-2 18-3

19-1 19-2 19-3

20-1 20-2 20-3

12.5 mg/kg crizotinib

25-1 25-2 25-3

26-1 26-2 26-3

27-1 27-2 27-3

28-1 28-2 28-3

21-1 21-2 21-3

26-1 26-2 26-3

27-1 27-2 27-3

28-1 28-2 28-3

0

500

1000

1500

2000

2500

3000

3500

MB231 SNU5 SNU1 GTL16 HT29

Rela

tive

Ma

rke

r A

rea

p

er

RO

I

pe

r 1

00

0 c

ell

s

E-Cadherin

Vimentin

-4 -2 0 2 4

MB231

SNU5

SNU1

GTL16

HT29

Log10 V:E Ratio

Mesenchymal (Red) = Log (V:E) >0 Epithelial (Green) = Log (V:E) < 0

Vimentin and E-Cadherin marker area per ROI of 1000 tumor cells or nuclei

(LLQ < or 1 = 1 )

Vimentin

E-Cadherin

MDA-MB-231 SNU5 SNU1

GTL16 HT29

MDA-MB-231 SNU5 SNU1

GTL16 HT29

B. A. Epithelial Mesenchymal

Figure 13. EMT multiplex IFA staining and Definiens analysis of FFPE xenograft control tissues. (A) E-Cadherin (green) and Vimentin (red) IFA levels expressed as mean marker area per 1,000 cells for 3 ROIs examined per tissue. (B) Ratio of Vimentin and E-Cadherin (log scale). Mesenchymal phenotype (red) is shown as Log (V:E) > 0 and Epithelial phenotype (green) is expressed as Log (V:E) < 0. Left set of images shows RGB image files collected by Aperio, while Definiens-derived marker masked images of E-Cadherin (green) and Vimentin (red) are shown on the right.

Figure 14. EMT multiplex IFA staining and Definiens analysis of FFPE tissues from SNU5 and MKN45 xenografts treated with either vehicle, pazopanib, or the combination with tivantinib (QDx8) based on the preclinical study plan shown in Fig 7B. Representative images from different ROIs taken from each tumor in various treatment groups are shown in (A), while the Vimentin:E-Cadherin ratios (log scale) for 3 representative tumor ROIs per tumor selected from each treatment group are shown in (B). Based on the V:E log ratio (range from only -1 to 1 is shown ), vehicle-treated MKN45 shows a relatively epithelial phenotype, while vehicle-treated SNU5 shows a mesenchymal phenotype. These tissue controls were used to set the threshold for each biomarker channel that was applied to all tissue ROIs in each treatment group. ROIs for analysis were selected based on the H&E staining of non-necrotic tumor tissue and human mitochondrial-positive staining in adjacent tumor tissue sections. A total of 4/6 tumors in the pazopanib-treated MKN45 group appear to show varying levels of epithelial-to-mesenchymal transition compared to vehicle or the combination group. Pazopanib/tivantinib combination–treated groups (3/3 evaluable tumors) showed a generally epithelial phenotype, while 2/5 samples had no tumor content by H&E.

SNU5 + pazopanib

MKN45 + pazopanib

SNU5 + vehicle

MKN45 + vehicle

MKN45 + pazopanib

tivantinib QDx8

A1 A2 A3

A1

A2 A3

C1 C2 C3

A1 A2 A3

A1 A2 A3

D1 E1

F1 E2

D2

A1 C1 B1 C1

A3

References

Clone # AF276

Figure 1B

IFA vs. ELISA

% pY1234-35 MET/total MET ELISA (Ratio of pM / ug protein)

Left A Tumor Quadrant

% P

ho

s/To

tal M

ET I

FA

Rat

io o

f M

arke

r A

rea

/ #

Nu

clei

R

igh

t H

Tu

mo

r Q

uad

ran

t

IFA vs. ELISA

% p

Y1

23

5-M

ET/T

ota

l MET

IFA

R

atio

of

Mar

ker

Are

a /

# N

ucl

ei

Rig

ht

H T

um

or

Qu

adra

nt

% pY1235-MET/total MET ELISA (Ratio of pM/ug protein) Left A Tumor Quadrant

Figure 9. Definiens quantitation analysis showing marker masks for pY1235-MET (green) and total MET (red) (left panel) in representative tissue microarray cores derived from SNU5 xenografts treated with vehicle or crizotinib. Scatter graphs (right panel) show pY1235-MET (green) and total-MET (red) IFA levels expressed as mean marker area/# of nuclei for 3 representative cores per tumor, with 4 tumors examined per treatment group.

Figure 10. Scatter graphs show a comparison between IFA vs. ELISA values for % pY1235-MET/total MET measured from different tumor quadrants derived from SNU5 xenografts treated with vehicle or crizotinib. Regression analysis shows correlation between the IFA values expressed as the ratio of marker area/# of nuclei vs. ELISA values expressed as pM/ug total protein.

Figure 12. EMT multiplex IFA assay developed using directly conjugated monoclonal antibodies to E-Cadherin, b-catenin, and Vimentin.

Nature Reviews Cancer 12, 699-709 (October 2012)

Figure 1A

Epithelial Mesenchymal A. B.