Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of...

98
Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein by Rahul Rana A thesis submitted in conformity with the requirements for the degree of Masters of Science in Chemistry Department of Chemistry University of Toronto © Copyright by Rahul Rana 2016

Transcript of Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of...

Page 1: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor

and Design of Peptidomimetic Inhibitors of STAT5 Protein

by

Rahul Rana

A thesis submitted in conformity with the requirements for the degree of Masters of Science in Chemistry

Department of Chemistry University of Toronto

© Copyright by Rahul Rana 2016

Page 2: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

ii

Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor

and Design of Peptidomimetic Inhibitors of STAT5 Protein

Rahul Rana

Masters of Science in Chemistry

Department of Chemistry

University of Toronto

2016

Abstract

Genes encoding for tyrosine kinases that undergo rearrangement at chromosomal

breakpoints have been associated with constitutively activated transcript products. Repressor of

silencing 1 (ROS1) tyrosine kinase fusions have been shown to drive cellular proliferation and

survival signaling pathways in numerous human cancers. First generation kinase inhibitors of

ROS1 suffer from acquired resistance due to a point mutation in the ROS1 kinase domain. This

work looks to determine the structural features of cabozantinib, a potent inhibitor of both wildtype

and mutant ROS1. Inhibitor binding to the ROS1 kinase domain and specific scaffold contributions

with synthesized analogues will be covered.

Additionally, a peptidomimetic strategy was employed to design inhibitors of the SH2

domain of the signal transducer and activator of transcription 5 (STAT5) protein, a transcription

factor constitutively phosphorylated in hematological malignancies and inflammatory diseases.

Native receptor peptide sequences that interact with the STAT5 SH2 domain were used for the

initial library.

Page 3: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

iii

Table of Contents

Chapter 1: An introduction to Repressor of Silencing 1 (ROS1)

1.1 ROS1 tyrosine kinase gene fusions............................................................................... 1

1.2 Oncogenic Activity of ROS1 ........................................................................................ 7

1.3 Inhibiting ROS1 fusion kinases .................................................................................. 10

1.4 Conclusion .................................................................................................................. 17

Chapter 2: Resistance and selectivity profiles of cabozantinib and derivatives with structural

insight

2.1 Computational studies of TKI interaction with ROS1................................................ 19

2.2 Proposed library of cabozantinib derivatives.............................................................. 26

2.3 Synthesis of small-molecule TKIs of ROS1 ............................................................... 31

2.4 In vitro evaluation of preliminary library ................................................................... 37

2.5 Binding constant determination using biophysical assay .......................................... 38

2.6 Concluding remarks .................................................................................................... 44

Chapter 3: Design of peptdiomimetic inhibitors of signal transdcuer and activator of transcription

factor 5 (STAT5)

3.1 Introduction to signal transducer and activator of transcription factor 5 (STAT5) .... 46

3.2 Structure and signaling of STAT5 .............................................................................. 47

3.3 Role of STAT5A and STAT5B in disease.................................................................. 51

3.4 Therapeutic strategies towards the STAT5A/B signaling pathway ........................... 53

3.5 Proposed isoform-selective peptidomimetic inhibitors of STAT5A/B ...................... 57

Page 4: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

iv

3.6 Initial FP analysis of proposed phosphopeptides ....................................................... 64

3.7 Conclusion ................................................................................................................. 70

Chapter 4: Conclusions and Future Directions ................................................................. 71

References ......................................................................................................................... 73

Page 5: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

v

List of Abbreviations

Akt protein kinase B

ALL acute lymphoblastic leukemia

Alk anaplastic lymphoma kinase

AML acute myelogenous leukemia

Bcl-XL B-cell lymphoma-extra large

Bcr-abl Philadelphia chromosome

CCD coiled-coil domain

CDK cyclin-dependent kinase

CML chronic myelogenous leukemia

C-Met hepatocyte growth factor receptor kinase

EGFR Epidermal growth factor receptor

EPO erythropoietin

EpoR Epo receptor

Erk Extracellular signal-regulated kinases

FI fluorescence intensity

FIG fused in glioblastoma

FISH fluorescence in situ hybridization

FLT3 Fms-like tyrosine kinase

FP fluorescence polarization

FRET fluorescence resonance energy transfer

GAS gamma-activated sequence

GH growth hormone

GM-CSF granulocyte macrophage colony stimulating factor

gp-130 glycoprotein 130 HTS high-throughput screen

IL interleukin

IFN interferon

ITC isothermal titration calorimetry

Jak Janus kinase

K562 a CML cell line

Lck lymphocyte-specific protein tyrosine kinase

MAPK mitogen-activated protein kinase

Mcl-1 induced myeloid leukemia cell differentiation protein

MV-4-11 an AML cell line

NGF nerve growth factor

NSCLC Non-small cell lung cancer

NTD N-terminal domain

PI3K Phosphatidylinositol 3 kinase

PIAS protein inhibitors of activated STAT

PLC phospholipase C

PKB Protein kinase B

PRL prolactin PTP

phosphatase pY

Page 6: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

vi

phosphotyrosine

RET rearranged

during transfection

kinase

ROS1 repressor of silencing 1

RSK ribosomal S6 kinase

RTK receptor tyrosine kinase

SAR structure-activity relationship

SCCHN squamous cell carcinoma of the head and neck

SH2 Src-homology 2

Shp Src homology-2 protein phosphatase

SOCS suppressors of cytokine signaling

SPPS solid-phase peptide synthesis

SPR surface plasmon resonance

STAT signal transducer and activator of transcription

TAD transactivation domain

TKI tyrosine-kinase inhibitor

TPO thrombopoietin

TRF time resolved fluorescence

TR-FRET time resolved fluorescence resonance energy transfer Y

tyrosine

Page 7: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

vii

List of Schemes

Scheme 2-1. Amide coupling synthesis of cyclopropylcarboxylic acid and aniline: (i) SOCl2,

reflux, 2 h; then aniline, THF, DIPEA, rt, 16 h; (ii) aniline, HBTU, DIPEA, DMF, rt, 18 h; (iii)

LiOH.H2O, THF/H2O, 0 oC to rt, 2 h.

Scheme 2-2. Alkylation of catechol with numerous alkyl halides and subsequent preparation

of aniline: (i) 1,3-dibromopropane, K2CO3, ethylene glycol, 75-110 oC for 8-10 h; (ii) H2(g), Pd/C

(10%), THF/MeOH.

Scheme 2-3. Cyclization of quinolone using microwave-assisted conditions, chlorination

and subsequent coupling: (i) 90 oC, 10 min, microwave assisted; then aniline 6a-l, EtOH,

reflux, 2h; (ii) Ph2O, 230 oC, 10-20 min, microwave assisted; (iii) POCl3, reflux, 2 h; then cold

H2O, Na2CO3; (iv) NaH, 4-aminophenol, DMSO, 10 minutes; then chloroquinoline 12a-l

DMSO, 100 oC, 12h.

Page 8: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

viii

List of Tables

Table 2.1- Summarized KD values for cabozantinib and analogues determined by LanthaScreen

TR-FRET assay.

Table 3.1- Summary of pY-containing ligands and negative control analogs

Table 3.2- Calculated Ki values for EpoR and GM-CSFR derived peptide ligands

Page 9: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

ix

Acknowledgements

Page 10: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

x

Contribution of Authors

The work reported in this dissertation includes contributions from several workers. The initial

computational work and method development was performed by Dr. Nadeem Vellore (Huntsman

Cancer Institute, University of Utah). Dr. Thomas O’Hare’s laboratory conducted the biological

work, specifically the cell cytotoxicity assay. Finally, Daniel Ball contributed significantly

towards the computational methods, docking simulations, chemical synthesis. He also developed

the LanthaScreen assay parameters for the inhibitor library and optimized the protocol.

Furthermore, he carried out the calibration and competition assays with data analysis.

Page 11: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

xi

List of Figures

Figure 1.1- Postulated extracellular domain of ROS1 determined from sequencing and homology

comparison. Three β-propeller domains (purple) are dispersed over β-sandwich structures (blue).

(Image from ref 7).

Figure 1.2- Tertiary structure of ROS1 kinase domain (PDB code: 3ZBFf) with amino residues

in the catalytic site highlighted red and represented by stick models.[PT1] [RR2] Image

generated using MacPyMol 2009-2010.

Figure 1.3- Possible downstream signaling cascades transduced by ROS1. (Figure reproduced

from reference 6)

Figure 1.4- ROS1 fusion genes with corresponding receptor partner and breakpoint regions

indicated. Rearranged genes are categorized by disease prominence.[PT1] (Figure taken from

Stumpfova, M.; Janne, P.A. Zeroing in on ROS1 Rearrangements in Non-Small Cell Lung Cancer.

Clinical Cancer Research. 2012. 18, 4222-4224.)

Figure 1.5- A) Sanger sequencing of RT-PCR products from various tissue samples of the NSCLC

patient indicating the G2032R mutation. B) Western blot analysis of cell lysates prepared from

293T cells expressing CD74-ROS1 or CD74-ROS1G2032R post crizotinib and TAE684 treatment.

C) Crystal structure of native ROS1 with crizotinib on the left. The right panel illustrates the

predicted steric clash between crizotinib and R2032. D) First-generation small molecule TKI,

crizotinib and E) TAE684. (Figure taken from reproduced from reference 39)

Figure 1.6- Second-generation TKI inhibitors, foretinib and cabozantinib, both originally

developed by Exelixis.

Figure 2.1- A) Active and inactive dynamic equilibrium of RTK enzymes controlled by

phosphorylation of the DFG motif in the A-loop; B) Anatomy of kinase domain with specific

regions highlighted. ‘Hydrophobic region I’ is the cleft adjacent to the adenine regions which gets

exposed upon dephosphorylation of the kinase to the inactive state. (Figure from reference 50).

Figure 2.2- Overlap of the active and inactive conformations of the ROS1 kinase domains with

the DFG motif phenylalanine 2103 represented by a stick model. Image generated using

MacPyMol 2009-2010.

Figure 2.3- Docking score histograms for active ROS1-cabozantinib (top), inactive

ROS1cabozantinib (bottom left) and inactive ROS1-cabozantinib (bottom right).

Figure 2.4- Lowest energy conformation of cabozantinib bound to ROS1 kinase domain. Panel A

shows exposed methoxy groups and the quinoline ring occupying the ATP binding site (green).

The tethering 4-aminophenol and cyclopropyl carboxamide motifs reside in the hinge region,

highlighted yellow. Panel B shows a few of the interacting amino acids. Panel C illustrates

Page 12: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

xii

occupancy of the hydrophobic groove (adjacent to the adenine site) by the 4-fluorobenzene

appendage. Image generated using MacPyMol 2009-2010.

Figure 2.5- Alternative, reversed binding pose of foretinib with ROS1 compared to predicted

cabozantinib-ROS1 complex. Image generated using MacPyMol 2009-2010.

Figure 2.6- Proposed structural modifications to cabozantinib for fragment-binding computational

studies.

Figure 2.7- Tabulated Ki values (nM) for alkyl group derivatives of 6,7-dimethoxyquinoline

position from docking screen.

Figure 2.8- Tabulated Ki values (nM) for alkyl group derivatives of 6,7-dimethoxyquinoline

position from docking screen.

Figure 2.9- Proposed mechanisms for decarboxylation-mediated cyclization reactions of imines

9k and 9h. Mechanisms A and B illustrate how 9k, can give two regioisomers, 10k(i) and 10k(ii).

Similarly, diverging mechanisms C and D outline the formation of 10h(i) and 10h(ii).

Figure 2.10- Preliminary library of synthesized analogues of cabozantinib, 13a-t.

Figure 2.11- Graphical representation of IC50 values of preliminary library of small-molecule

inhibitors 13a-t and crizotinib in Ba/F3 CD74-ROS1 and CD74-ROS1G2032R cells.

Figure 2.12- Illustration of principle of LanthaScreen (Invitrogen) TR-FRET assay. (Figure taken

from reference 74).

Figure 2.13- Calibration curves of Alexa Fluor-647 tracer titration with ROS1 kinase with

100μs, 50 μs, 25 μs and 20 μs time delay between excitation and measurement of emission, in the

absence (black trace) and presence (red trace) of staurosporine.

Figure 2.14- Emission ratio plot with increasing concentrations of tracer titrated with

Eu(III)chelate antibody in the absence of ROS1 protein.

Figure 3.1- Schematic illustration of STAT5A/B domains with critical Y694 (STAT5A) and Y699

(STAT5B) highlighted. (Image generated using MacPyMol 2009-2010)

Figure 3.2- Brief schematic of normal JAK/STAT5 signaling with cytokine/growth factor

receptors dimerized and bound to activated JAK2. Latent STAT5 is recruited via the SH2-pY

interaction and undergoes phosphorylation. Dimerization in the cytosol is followed by nuclear

translocation and regulation of transcription of target genes. Negative regulators, SOCS, PIAS and

PTP are represented as well. (Image generated using ChemBioDraw 14)

Figure 3.3- First generation BCR-ABL TKI imatinib for CML treatment. Second-generation TKIs

from imatinib, nilotinib and dasatinib are shown below.

Figure 3.4- First-generation FLT-3 TKI lestaurinib and following second-generation derivative

sorafenib. Both drugs are used for FLT-3 mutation positive AML. Ruxolitinib is the widely used

Page 13: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

xiii

JAK2 inhibitor in AML treatment, given its efficacy against JAK2V617F

Figure 3.5- Traditional SH2 domain tertiary structure ribbon illustration. The central anti-parallel

β sheet is flanked by two α-helices. Interacting pY-containing peptide is shown in black, stick

figure representation, with pTyr, +1, +2 and +3 sites labeled. (figure from ref 60).

Figure 3.6- A) Overlap of STAT5A (orange) and STAT5B (blue) ribbon figures. The five

differing residues in the SH2 domain in the vicinity of the pY binding motif are represented by

stick figures. B) Shared Ser622, Arg618 and Lys600 in STAT5A/B SH2 domains. On comparison

with conventional SH2 domain structure, these three residues contribute to recognition of pY

functional group. (Image generated using MacPyMol 2009-2010)

Figure 3.7- Calibration curve for EpoR derived 5-FAM-GpYLVLDKW fluorescent probe with

STAT5A protein.

Figure 3.8- Calibration curve for EpoR derived 5-FAM-GpYLVLDKW fluorescent probe with

STAT5B protein.

Figure 3.9- Normalized FP inhibition curve for EpoR derived peptide, QDTpYLVLDKWL for

STAT5A. Ki = 522.3 nM

Figure 3.10- Normalized FP inhibition curve for EpoR derived peptide, QDTpYLVLDKWL for

STAT5B. Ki = 426.0 nM

Figure 3.11- Normalized FP inhibition curve for GM-CSFR derived peptide, QQDpYLSLPPWE

for STAT5B. Ki = 876.6 nM

Figure 3.12- Normalized FP inhibition curve for GM-CSFR derived peptide, QQDpYLSLPPWE

for STAT5A. Ki = 652.4 nM

Page 14: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

1

Chapter 1: An introduction to repressor of silencing 1 (ROS1)

1.1 ROS1 tyrosine kinase gene fusions

Structural chromosomal aberrations have been studied extensively for their effects on the

genome at the cellular level and the resultant biological phenotype. This specific category of

chromosomal aberrations is defined as the physical breakage of chromosomal segments in the

DNA.1 This results in numerous possible mutations including, deletions, duplications,

translocations, inversions, and insertions.2 Specifically, translocations involve partitions in two

different chromosomes and eventual exchange of chromatin material. Chromosomal translocations

of exon DNA segments ultimately compromise the function and regulation of the natural

geneencoding product. Concurrently, it has been established that several key translocation genetic

events contribute towards the initiation of carcinogenesis. The first chromosomal translocation

identified in human cancer was in chronic myeloid leukemia (CML) where the fragment exchange

resulted in the fusion gene product, BCR-ABL. The fusion was characterized as the association of

the ABL1 and BCR genes where the expressed product is a tyrosine kinase (TK). The discovery of

the BCR-ABL fusion gene represents an important breakthrough in recognition of TK fusion genes

as a result of chromosomal translocations and their oncogenic potential.3

There are approximately 90 TK genes in the human genome, out of which, 58 encode for

receptor tyrosine kinases (RTKs).4 Repressor of Silencer 1 (ROS1) is a RTK of the insulin receptor

family that has been characterized as a partner of multiple fusion oncoproteins responsible for

driving numerous epithelial cancers including glioblastoma, cholangiocarcinoma, gastric cancer,

ovarian cancer, and non-small cell lung cancer (NSCLC).5

The human ROS1 gene was initially discovered as the homolog of the UR2 avian sarcoma

virus.6 The ROS1 protein shares 49% sequence homology with the anaplastic lymphoma kinase,

ALK, with more than 64% identical residues in the kinase domain.7, 8 ROS1 is a 264 kDa protein

which spans over 44 exons on chromosome 21.7 The tertiary structure of ROS1 protein follows the

conventional RTK arrangement: an N-terminus extracellular receptor region (residues 1-1861), a

single transmembrane-spanning domain (residues 1862-1882) and a C-terminus region (residues

Page 15: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

2

1882-2346) which contains the catalytic active site.9 The extracellular segment consists of three β-

propeller domains distributed along nine β-sandwich domains,7 shown in figure 1.1.

Figure 1.1- Postulated extracellular domain of ROS1 determined from sequencing and homology

comparison. Three β-propeller domains (purple) are dispersed over β-sandwich structures (blue).

(Image from ref 7).

Despite possessing an extensive extracellular domain, a ligand for the human ROS1 protein

has yet to be identified. This ‘orphan RTK’ status has impeded investigations concerning the ROS1

activation and resultant signaling pathways. However, sequence comparison with known receptor

sequences suggests that the extracellular domain is involved in mediating cell-cell adhesion. These

peptides resemble fibronectin type III proteins, which exhibit high affinity for extracellular

adhesion molecules.6 Thus, cellular attachment and similar processes are postulated to be trigger

events for the activation of the tyrosine phosphorylation activity of the kinase with following

intracellular signaling cascades.10 The kinase domain, situated at the intracellular domain of ROS1

comprises of the traditional tyrosine kinase elements: an ATP biding pocket, a substrate binding

pocket, an activation loop (A loop) and finally, the catalytic loop (C loop)11 with the catalytic

K1980 residue, figure 1.2.

Page 16: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

3

Figure 1.2- Tertiary structure of ROS1 kinase domain (PDB code: 3ZBFf) with amino

residues in the catalytic site highlighted red and represented by stick models. Image generated

using MacPyMol 2009-2010.

Unfortunately, the definitive function of wild type ROS1 in humans remains elusive. Northern blot

studies of various human organs conducted by Acquaviva and co-workers in 2009 indicated

highest ROS1 RNA expression occurs in the lungs.10 Reitchmacher et al. were able to study the

biological effects upon loss of the receptor in vivo. They discovered that male ROS1knockout mice

were viable but suffered from loss of normal reproduction. The infertility was attributed to defects

in the epididymis epithelium.12

Through the use of its tyrosine phosphorylation activity, ROS1 kinase is responsible for the

regulation of signaling pathways that control cell proliferation and survival. The precise series of

events that switch ROS1 into the active conformation from its own inactive state are currently

unknown. This can be attributed to an unidentified native ligand as well as the difficulty in

expressing the full-length extracellular domain of ROS1 in vitro. However, similarities can be

drawn from analogous RTKs, including ALK and RET (rearranged during transfection). In the

case of ALK, ligand–receptor binding induces dimerization of the receptor via oligomerization

motifs. This orients the intracellular kinase domains in an ideal positionfor an autophosphorylation

event. In the inactive conformation, the Asp-Phe-Gly triad in the activation loop block substrate

L 2 0 8 6 &

K 1 9 8 0 &

F 2 1 0 3 &

L 2 0 2 8 & E 2 0 2 7 &

D 2 1 0 2 &

Page 17: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

4

binding. Upon phosphorylation, however, the kinase A loop undergoes a conformational shift

forcing the same amino acid residues to protrude outwards. This phosphorylation-triggered

structural change permits binding of the substrate and subsequent ATP transfer. In human ROS1,

Tyr2274 has been shown to be the site of phosphorylation and mutation of this residue to Phe

abolished ROS1 activity.13 Like all tyrosine kinases, the C-terminal tail serves as a docking site for

native substrates. With the limitations mentioned earlier, detailing the ROS1 downstream signaling

mechanisms have been hindered. One approach to circumvent these challenges of studying this

orphan RTK is the use of artificially engineered ROS1 fusions with known ligand binding

receptors. Examples include the insulin receptor14, epidermal growth factor receptor (EGFR)15,

nerve growth factor (NGF) receptor16, and CD74 receptor.17 Utilizing known receptor partners

with the ROS1 kinase domain allows one to regulate enzyme activity in a liganddependent manner

and potentially study the interacting proteins. These chimeric ROSI fusion models have shed some

light on the ambiguous, yet diversified set of signaling cascades stimulated upon ROS1 activation.

The proteins activated upon ROS1 phosphorylation include: the mitogenactivated protein kinases

(MAPKs), ERK1 and ERK2, Src-homology region 2 domain phosphatase 2 (SHP-2),

phosphatidylinositol 3-kinase (PI3K), protein kinase B (PKB or Akt), and signal and transducer

and activator of transcription 3 (STAT3). All of the mentioned proteins function as signaling

factors that regulate cellular growth, differentiation, mitosis, and apoptosis. The signaling

mechanisms are shown in figure 1.3 below.

Page 18: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

5

Figure 1.3- Possible downstream signaling cascades transduced by ROS1. (Figure reproduced

from reference 6)

The MAPK/ERK pathway is an established oncogenic signaling cascade. The pathway begins

with binding of mitogens to cellular receptors which are eventually translated to regulation in

transcription of oncogenic genes. The pathway involves the phosphorylation of multiple proteins

with MAPKs, ERK1, and ERK2 as pivotal contributors. These extracellular signal-related kinases

phosphorylate factors such as ribosomal s6 kinase (RSK), c-MYC, and c-FOS. Upon

phosphorylation, these oncoproteins are rendered active, and upregulate the transcription of genes

involved in cell proliferation and survival. Xiong and co-workers observed activation of ERK1/2

in mouse fibroblasts upon treating EGFR-ROS1 fusions with ligand.15 Additionally, Davies et al.

reported increase in ERK phosphorylation in Ba/F3 cells expressing with SD-C4-ROS1 protein.17

Alternatively, Gu and colleagues conducted experiments with Ba/F3 cells transfected with other

ROS1 fusions and noted a decrease in ERK phosphorylation levels upon treating with potential

ROS1 inhibitors.18

SHP-2 (encoded by PTPN6 gene) is a member of the protein tyrosine phosphatase (PTP)

family. This enzyme contains two independent SH2 domains that recognize phosphotyrosine (pY)

Page 19: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

6

motifs. It is believed that recruitment of SHP-2 to ROS1 is mediated by the pY-SH2 domain

interaction.19 Phosphorylated SHP-2 has been shown to activate MAPKs and thus, the subsequent

MAPK/ERK pathway.20 Charest et al. studied glioblastoma mouse derived cells harbouring ROS1

fusions which resulted in the increase levels of SHP-2 protein phosphorylation.19 Further evidence

of SHP-2 mediated ROS1 signaling was reported by Rikova and co-workers as they successfully

isolated peptide sequences containing the tyrosine-phosphorylation motifs of SHP-2 from solid

NSCLC tumors.21

Protein kinase B (PKB) or Akt, is a serine/threonine kinase that phosphorylates key

regulators of cell apoptosis and mitosis. Akt is known to disrupt the Bcl-2/Bcl-X multimeric protein

in the mitochondrial membrane, which in turn compromises the cell’s pro-apoptosis mechanism.

Cyclin-dependent kinases 1 and 2 (CDK1 and CDK2) regulate the cell-cycle check phases, G1 and

G2. Activation of Akt via phosphorylation is known to overcome cell cycle arrest junctures, hence

increasing the chances of mutagenesis and uncontrolled cellular division. Nguyen and co-workers

demonstrated inhibition of Akt significantly diminished the ability of ROS1 fusions to promote

colony growth in NIH3T3 fibroblasts.22 Glioblastoma mouse models incorporating ROS1 fusions

also exhibited increased levels of phosphor-Akt.19

The PI3K signaling pathway is a crucial link between several cellular trafficking events

spanning from metabolism to ligand-independent proliferation. PI3K is responsible for the

phosphorylation and subsequent activation of Akt, whose oncogenic kinase activity is described

above. Validating the direct interaction between ROS1 and PI3K has been difficult, given the

complexity of the involved signaling mechanisms. Also, with Akt being an important protein in

PI3K signaling pathway, all evidence indicating increase in phospho-Akt levels can be correlated

to ROS1 activation of PI3K. One interesting study conducted by Uttamsingh and colleagues

suggested that knockdown of the PI3K pathway reduced anchorage independent growth in ROS1

activated fibroblasts.

STAT3 belongs to a family of cytoplasmic transcription factors that undergo

phosphorylation by kinases stimulated by cytokine binding to cell surface receptors.

Phosphorylated STAT3 monomers (pSTAT3) dimerize via reciprocal SH2 domain-pY interactions

and translocate into the nucleus. After recognizing specific DNA sequences, the STAT3 dimer

promotes transcription of genes that induce angiogenesis, metastasis, and proliferation. In tandem

Page 20: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

7

with their study of ROS1 induced PI3K/Akt signaling, Gu et al. also studied effects on STAT3.

They noted a marked increase in pSTAT3 protein levels when Ba/F3 cells where transformed with

a chimeric ROS1 fusion. In a separate finding, Zong and co-workers observed fibroblast colony

growth was markedly reduced when cells expressed dominant negative STAT3 allele. Rimkuna and

colleagues also conducted experiments trying to solve the signaling cascades elicited upon ROS1

activation in 2012 and reported decrease in pSTAT3 levels in Ba/F3 cells transfected with ROS1

fusion when treated with ROS1 inhibitor.

1.2 Oncogenic activity of ROS1

Overall, numerous studies indicate that ROS1 can engage multiple signaling pathways upon

activation. One common theme across these cascades is the phosphorylation of a protein substrate,

usually a kinase, and eventual downstream transcription of genes promoting cell growth and

survival. The specific targets mentioned above could be constitutively activated by aberrant ROS1

activity. On a molecular level, this is the case when the RTK is in the active conformation, which

is considered the “switched on’ state. RTK fusions are susceptible to adapting their catalytic

conformations through binding of cytokines and growth factors specific to the fused receptor.

Ideally, the downstream tyrosine containing protein targets of ROS1 fusions are identical to that

of the wild type protein mentioned above, and tyrosine-phosphorylating activity is independent of

the receptor motif. Thus, ligand sensitive ROS1 fusions that force the intracellular kinase domain

into a constitutively active mode have the potential to deregulate conventional cell growth

mechanisms and inhibit apoptosis. These ligand-transmitted events, which would otherwise be

absent with wild type ROS1, are believed to be responsible for transformation to the oncogenic

phenotype.

To date, seven ROS1 fusion genes have been reported and sequenced (figure 1.4).

Birchmeier et al. reported the first instance of a ROS1 fusion in a tumor model in 1987. The workers

isolated rearranged ROS1 cytogenically with the use of fluorescence in situ hybridization (FISH)

and RT-PCR from the U118MG glioblastoma cell line.23 FISH is a powerful technique used to

identify location of breakpoints in chromosomal rearrangements with the help of fluorescently

labeled oligonucleotide probes. Two different coloured DNA probes that are complimentary to the

Page 21: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

8

regions flanking the gene of interest are designed. In the absence of translocation events involving

the target gene, the two probes remain adjacent to one another, resulting in a fusion colour. When

chromosomal translocation disrupts the gene of interest, the probes are separated and consequently,

the primary coloured probes are observed individually.

This specific application of FISH, often referred to as “break-apart” FISH offers high specificity

with minimal false positive outcomes.

Figure 1.4- ROS1 fusion genes with corresponding receptor partner and breakpoint regions

indicated. Rearranged genes are categorized by disease prominence. (Figure taken from

Stumpfova, M.; Janne, P.A. Zeroing in on ROS1 Rearrangements in Non-Small Cell Lung Cancer.

Clinical Cancer Research. 2012. 18, 4222-4224.)

The genetic fusion partner was hence termed ‘fused in glioblastoma’ (FIG), and the

resultant hybrid kinase was eventually discovered in cholangiocarcinoma, ovarian cancer, and

NSCLC patient samples.6 Furthermore, the FIG-ROS1 fusion has been expressed in murine cell

lines, murine epithelium tissues and even mouse models to study its role to drive oncogene

character independently. Consistently, the experiments resulted in uncontrolled cellular

proliferation, carcinogenesis, and rapid tumor growth. Since then, numerous ROS1 fusion partners

have been characterized and sequenced in various human cancers. In all ROS1 rearrangements

Page 22: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

9

known to date, the kinase domain of ROS1 is conserved and the fusion partner contributes its

extracellular receptor domain.

Interestingly, there are numerous instances of ROS1 fusions in NSCLC throughout

academic literature and clinical data wherein the ROS1 derived oncogene most frequently reported

is the CD74 rearrangement. The first CD74-ROS1 hybrid was discovered recently in 2007 by

Rikova et al. in a non-smoking, female NSCLC patient.21 The fusion kinase profiling investigation

was part of a comprehensive RTK signaling study in 41 NSCLC cell lines and more than NSCLC

150 Chinese patient tumor samples. Additionally, transforming capability of CD74-ROS1 has now

been confirmed both in vitro and in vivo.17 Fused CD74-ROS1 occurs when exon 6 of CD74

(chromosome 18) is juxtapositioned to exon 34 of ROS1 resulting in the N-terminal extracellular

domain of CD-74 fused to the transmembrane and C-terminal kinase domain of ROS1. As a result,

CD-74 functions as the extracellular motif which relays ligand binding into protein structural

changes that prompt the constitutively active conformation of the ROS1 kinase. More specifically,

CD74 recognizes the macrophage migration inhibitory factor (MIF), an inflammatory cytokine

involved in inflammation responses. Through MIF signaling, CD-74 plays a crucial role in the

expression and transport of the major histocompatibility complex II (MHCII).24

In a subsequent study, Bergethon and co-workers screened 1073 NSCLC patients and noted

18 occurrences of ROS1 gene rearrangements. However, only 6 samples were sufficient for RT-

PCR analysis.25 Five CD74-ROS1 fusions were sequenced successfully, while other samples were

inconclusive in identifying the receptor partner. Interestingly, in their panel of NSCLC tumors, the

breakpoint of ROS1 was observed to occur at exon 32, highlighting another rearrangement pattern.

Three more groups reported the frequencies of CD74-ROS1 fusions while probing ALK and ROS1

translocations in a large panel of lung cancer patient-derived cell lines and tumors. The independent

studies consistently found approximately 1.2-2.4% of NSCLC patient tumors with CD74-ROS1

rearrangements.26,27,28 Generally, ROS1 rearranged positive patients are neversmokers and tended

to be younger than the median age of patients with NSCLC. In addition, NSCLC patients

harbouring translocated ROS1 are devoid of other oncogenic RTK-receptor hybrids. Overall, ROS1

rearrangements have been identified in approximately 2% of patients with NSCLC. Although this

is seemingly a low percentage, with an estimated 1.5 million new cases of NSCLC worldwide each

year, this corresponds to more than 20,000 patients with oncogenic ROS1 fusions.29,30 This

evidence undoubtedly highlights the prevalence of ROS1 rearrangements as oncogenic drivers in

Page 23: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

10

NSCLC and new data continues to support this trend. As a result, this RTK is considered in the

class of oncoproteins that can independently drive transformation in NSCLC both in vitro and in

vivo.

1.3 Inhibiting ROS1 fusion kinases

Given the rise of ROS1 fusion positive NSCLC cases in the clinic, there has been substantial

effort in therapeutic intervention of the ROS1-relevant pathways. A traditional strategy is to inhibit

the aberrant kinase activity of the protein itself. This strategy involves preventing the constitutive

tyrosine phosphorylating activity of ROS1. Directly inhibiting the enzyme would diminish the

aberrant anti-apoptotic signaling and potentially take advantage of the ‘oncogene addiction’ state

of the tumor. The principle of oncogenic addiction refers to the dependency of tumors on specific

oncoproteins to sustain malignant growth.31 Gene mutations that switch on a certain oncogene

typically drive aberrant survival processes in the cell that would otherwise be controlled under

biological cues. Once the function of the oncoprotein is constitutively activated, it plays a much

more involved role in survival, promoting conversion to the neoplastic state. The cancer cells then

become reliant on the translated product of the same oncogene to sustain its tumor phenotype.32

From a molecular perspective, the output of the hyperactive oncoprotein tends to outcompete

natural pro-apoptosis cascades and cell cycle checks. Furthermore, the identical mutated genome

and resulting ‘addictive’ oncogenic mechanism is present in progeny cells. Therapeutic

intervention abolishing the aberrant function of the oncoprotein in question can result in cancer

cell death and eventual tumor regression or ‘reversal’ of the malignant phenotype. With oncogene-

specific inhibition, the anti-survival and cell division checks are no longer dominated by the

hyperactive function of the oncoprotein. This notion is supported by experimental data where

therapeutics targeting an individual activated oncoprotein have led to apoptosis in cancer cells.33

This specific hypothesis is termed “oncogene shock” and has formed the fundamental principle

underlying target-specific therapeutics. Appropriately, ROS1-targeted inhibition is the primary

approach for therapeutic treatment in the relevant set of diseases, including NSCLC. Previous and

contemporary strategies with the aim of inhibiting the ROS1 protein are discussed below.

In the 1990’s, significant strides were made in identifying tyrosine kinases as major role

Page 24: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

11

players in oncogenic transformation. The human kinome was thoroughly examined and potential

oncokinases were labeled as important targets for drug development programs. Genetically

rearranged kinases became a crucial part of the kinase inhibitor discovery umbrella because of the

growing information on their role in cancer. Approaches to inhibit RTK fusions include: ligand

modulation, monoclonal antibodies to inhibit the receptor domain, siRNA therapeutics, anti-sense

oligonucleotides, peptidomimetics, and direct small molecule inhibitors.34 The category of small

molecule tyrosine kinase inhibitors (TKIs) comprises of natural and synthetic chemicals that inhibit

the kinase activity through a variety of mechanisms, all of which involve direct interaction with

the protein.35 Possible mechanisms include: allosteric inhibition, substrate competitive inhibition,

and ATP competition. TKIs targeting the ATP binding site in the catalytic domain have

substantially stood out as the most well-studied and efficacious subset of inhibitors with

considerable success in the clinic. Given these small molecules were designed to compete for the

well-defined ATP pocket, most TKIs possess a scaffold that mimics ATP. Initially, the challenge

of achieving selective inhibition of RTK fusions with ATP competitive binders seemed

monumental as the ATP binding site was universal amongst all proteins in the human kinome.

Nonetheless, advances in small molecule screening, computational chemistry, protein X-ray

crystallography, and in silico guided structure-based drug design have contributed considerably in

the discovery of selective and potent TKIs. Amino acids in the catalytic site of the kinase are now

easily identifiable and interactions with ATP on the molecular level can be elucidated. Binding

clefts surrounding the sugar, phosphate, and adenine binding sites of the target RTK are now easy

to characterize and serve as the basis for inhibitor scaffold design. An important result of TKI drug

design was imatinib, the first case of a selective, RTK small molecule inhibitor with an outstanding

clinical profile. Imatinib (trade name: Gleevec) has been approved as the first-line treatment of

BCR-ABL positive CML by the F.D.A.36

The first series of small molecule inhibitors of ROS1 fusion proteins originated from the

ALK drug discovery program. As mentioned earlier, ROS1 and ALK share approximately 64%

overall sequence homology in the kinase domain and 84% in the ATP binding site. This high

degree of similarity in the linear amino acid residue sequence can be extrapolated to considerable

overlap in the tertiary structure of the ALK and ROS1 kinase domains. Thus, small molecule

inhibitors that bind appreciably to the ALK kinase domain should exhibit a similar inhibitory

Page 25: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

12

profile against ROS1. McDermott and coworkers were the first to recognize the sensitivity of

ROS1 to ALK TKIs in vitro in 2008.37 They observed potent activity of the ALK inhibitor,

TAE684, in the HC778 NSCLC cell line, which was positive for only ROS1 rearrangements.

Promising results followed this work with Gu et al. reporting reduction in phospho-ROS1 (pROS1)

and subsequent apoptosis of Ba/F3 cells harbouring FIG-ROS1, when treated with TAE684.18 In

addition, when performing western blot analysis, there was concomitant with decrease in levels of

p-STAT3, p-ERK, p-Akt and p-SHP-2 in a dose-dependent manner. In parallel with the

breakthrough work validating use of ALK inhibitors to inhibit ROS1 fusion proteins, clinical trials

with the ALK inhibitor, crizotinib (developed by Pfizer), as a NSCLC therapeutic were underway.

The impressive inhibitory activity of crizotinib against rearranged ALK kinases warranted

investigations probing the efficacy of crizotinib in ROS1 fusion positive NSCLC tumors. In 2012,

three separate studies by Yasuda, Davies, and Bergethon concluded that crizotinib was indeed a

potent ROS1 inhibitor. Their collective work showed crizotinib induced apoptosis in ROS1 fusion

harbouring NSCLC cell lines with concomitant reduceed levels of pROS1 and downstream protein

targets. More importantly, tumor growth in NSCLC patient tissue samples was reduced and upon

dosing a ROS1 positive NSCLC patient, tumor burden decreased with no symptoms of recurrence.

Critzotinib (figure 1.5.D) became the first line treatment for patients diagnosed with ROS1

translocation positive NSCLC. NSCLC patients can be easily assessed for the presence of ROS1

fusions using break-apart FISH. Following this, RT-PCR is employed for gene amplification and

Sanger sequencing to identify the fusion partner. Initial stages of a Phase I/II clinical trial in 50

NSCLC patients carried out by Awad et al. (funded by Pfizer; ClinicalTrials.gov number,

NCT00585195) looked promising and ROS1 tumors (with the CD74 being the most prominent

fusion partner) showed extreme sensitivity to crizotinib treatment. Unfortunately, tumor growth

and associated symptoms redeveloped in a subset of patients, after initial response. Continued drug

administration proved to be ineffective and it was revealed the patients had acquired resistance to

crizotinib.39

Page 26: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

13

Figure 1.5- A) Sanger sequencing of RT-PCR products from various tissue samples of the NSCLC

patient indicating the G2032R mutation. B) Western blot analysis of cell lysates prepared from

293T cells expressing CD74-ROS1 or CD74-ROS1G2032R post crizotinib and TAE684 treatment.

C) Crystal structure of native ROS1 with crizotinib on the left. The right panel illustrates the

predicted steric clash between crizotinib and R2032. D) First-generation small molecule TKI,

crizotinib and E) TAE684. (Figure taken from reference 39)

Tissue biopsy of a resistant tumor from a patient confirmed a point mutation in the kinase

domain of ROS1. Sanger sequencing of RT-PCR products from tissues from the chest wall, right

and left lungs, pleural fluid, and lymph nodes showed mutation of guanine to adenine at position

6094, which in turn, corresponded to a Gly2032Arg (G2032R) substitution in the kinase domain

illustrated in figure 1.5.A. This specific mutation was not detected in the tumor prior to crizotinib

administration and was the only such case retrieved from the resistant specimen. The investigators

conducted a series of experiments to determine the molecular basis of the acquired mutation in

vitro. First, cell lysates were prepared from 293T human embryonic kidney cells transfected with

CD74-ROS1 and G2032 mutant CD74-ROS1 after crizotinib treatment. Western blot analysis of

pROS1 levels confirmed crizotinib was ineffective against the G2032 mutant with IC50 values >

Page 27: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

14

1000 nM, compared to IC50 = 30-50 nM for the non-mutant fusion, figure 1.5.B. An enzymatic

activity assay showed an increase in the Ki value of the G2032 mutant kinase by a factor of 270

with respect to non-mutant ROS1 (Ki = 570 nM cf. 2.1 nM). Finally, to delve into the specific

inhibitor-protein interaction, the workers crystallized the phosphorylated ROS1 kinase domain

bound to crizotinib. As expected, the crystal structure showed crizotinib bound to the ATP binding

site where Gly2032 is solvent exposed, positioned towards the distal end of the kinase hinge. In

terms of molecular interactions, this specific glycine is situated such that it can participate in van

der Waal’s interactions with the pyrazole ring crizotinib. Modeling studies that replace the G2032

residue with arginine in silico, place the substituted side chain guanidinium substituent in close

proximity to the piperidine functional group in crizotinib, figure 1.5.C. It is predicted that the

bulkier amino residue sterically clashes with the inhibitor to a considerable extent, thus decreasing

its binding affinity for ROS1.

Despite initial responses in TKI chemotherapy regimes, acquired drug resistance is a

recurring theme universally recognized in the clinic, not only limited to NSCLC.40 Targeted

therapy using small molecule inhibitors eventually results in genetic alterations that confer loss of

sensitivity to inhibitor activity.41 Continued therapy is ineffective as the resistant cells successfully

survive and pass on their mutated genetic material and outcompete non-resistant cells. In essence,

interference with TKI targeted therapy serves as an artificial selection pressure which allows the

tumor to ‘evolve’. In the case of crizotinib inhibition of ROS1, the molecular basis of acquired

resistance was a mutation in the oncoprotein of interest. Rapidly dividing cancer cells are highly

susceptible to point mutations. These single codon modifications operate by reducing the binding

affinity of the TKI for the target kinase domain. A weak binding inhibitor does not effectively

abrogate the catalytic activity of the kinase, hence allowing constitutive phosphorylation of

substrates. Alternate mechanisms of drug resistance include amplification of the target oncogene

or activation of independent signaling cascades that restore the hyperactivity of the downstream

signaling pathways while the original oncoprotein is inhibited.40,42

Initial strategies to overcome the acquired resistance paradigm observed in clinical TKI

therapy revolve around the design of next-generation kinase inhibitors. This requires designing

ATP-mimetic small molecules that do not suffer loss in binding interactions with the mutant

oncoprotein. This strategy has yielded success in other secondary cancer treatments, most notably,

Page 28: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

15

in CML.43 Upon continuous exposure to imatinib treatment, it was observed the BCR-ABL fusion

kinase developed point mutations that confer resistance to against small molecule inhibitors. The

most common mutation was observed with Thr315, the gatekeeper residue to Ile. New TKIs were

developed with the aim of improved binding interactions with the mutant ABL kinase domain from

the previous generation. A series of second and third generation inhibitors including dasatinib,

nilotinib, bosutinib, and ponatinib had an improved activity profile against imatinib resistant

BCRABL while sustaining desirable inhibition against the non-mutated fusion.40 This has seen

successfully translated to the clinic where next generation TKIs are now part of the chemotherapy

regime for imatinib resistant CML.

Drawing inspiration from the progress of leukemia therapy using next-generation TKIs,

several groups screened a series of ALK, RET, and MET multi-kinase inhibitors for their activity

against rearranged ROS1 positive NSCLC and the reported G2032 mutant. A breakthrough was

made when Davare et al. identified foretinib (developed by Exelixis, XL-880, figure 1.6) as a more

potent and selective ROS1 and ROS1G2032 inhibitor than crizotinib in 2013.44 In vitro cell growth

assays coupled with immunoblot analysis pROS1 and downstream targets in Ba/F3 transformed

wild type CD74-ROS1 and CD74-ROS1G2032 cell lines confirmed the mutant kinase is insensitive

to crizotinib. Ba/F3 CD74-ROS1G2032 cells were highly resistant to crizotinib treatment (IC50 =

2200 nM) when compared with wild type CD74-ROS1 (IC50 = 14 nM). Foretinib also decreased

phosphorylation levels of ROS1 and its downstream targets, ERK1/2 and SHP-2 in a

dosedependent fashion across both wild type and G2032 mutant cell lines.

Figure 1.6- Second-generation TKI inhibitors, foretinib and cabozantinib, both originally

developed by Exelixis.

Another high-throughput screen of established TKIs (approved by the F.D.A. or under

clinical trials) by Katayama and colleagues in 2014 led to the discovery of cabozantinib as a potent

inhibitor of cell survival in CD74-ROS1 and CD74-ROS1G2032 transformed Ba/F3 cell lines.45

Page 29: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

16

Originally developed by Exelixis, cabozantinib (XL-184, figure 1.6) is a multikinase domain

inhibitor of cMET and VEGFR-246 and is currently approved for treatment of refractory medullary

thyroid cancer.47,48 Firstly, the workers examined viability of Ba/F3 cells either expressing

CD74ROS1 and CD74-ROS1G2032 fusion protein. Ba/F3 CD74-ROS1 cells exhibited significant

reduction in cellular growth when exposed to either crizotinib or cabozantinib with IC50’s of 2 nM

and 2.12 nM, respectively. However, CD74-ROS1G2032 Ba/F3 cells did not respond to crizotinib

treatment (IC50 = 253.7 nM, 126-fold increase), whereas cabozantinib was quite potent in reducing

cell viability with IC50 = 13.53 nM. Next, they compared the extent of ROS1 autophosphorylation

in Ba/F3 cell lines expressing either CD74-ROS1 or CD74-ROS1G2032. Upon incubation with

increasing concentrations of cabozantinib and crizotinib, immunoblotting revealed that both

inhibitors were efficient in suppressing pROS1 protein levels in non-mutated CD74-ROS1

constructs. Not surprisingly, only cabozantinib retained its anti-pROS1 activity in the

G2032mutated CD74-ROS1 Ba/F3 cells. Immunoblotting studies were extended to known

downstream substrates of ROS1 phosphorylation. Treatment with either TKI suppressed pSTAT3,

pAkt and pERK proteins, but this trend failed to carry over to CD74-ROS1G2032 mutants with

crizotinib remaining inactive against the G2032 mutated fusion kinase. It is important to note the

encouraging toxicity profile of these inhibitors as negative control experiments involving parent

Ba/F3 cells treatment with both TKIs did not inhibit cell growth (IC50 = 10,000 nM). The same

workers assessed the efficacy of both inhibitors in crizotinib resistant NSCLC patient-derived cell

lines harbouring the G2032 mutation (MGH047 cells). As expected, cabozantinib potently

inhibited growth of MGH047 cells whereas crizotinib did not show any anti-proliferative potency.

Finally, a comparison of pROS1 levels and its associated signaling partners replicated the

observations seen in the Ba/F3 cell lines where only crizotinib effectively suppressed

phosphorylation of ROS1, ERK and Akt. In light of these findings, strong emphasis was placed on

cabozantinib’s selectivity and resistance profile against the CD74-ROS1 fusion kinase and the

G2032 variant. Following from their initial work on foreitinib in CD74-ROS1 driven NSCLC,

Davare and colleagues also found cabozantinib to be a highly potent inhibitor of Ba/F3 CD74-

ROS1 and CD74-ROS1G2032 cell proliferation and viability with IC50’s of 1.1 nM and 15.3 nM

respectively. Both foretinib and cabozantinib demonstrated high selectivity for CD74-ROS1 as

Ba/F3 cells expressing rearranged EML4-ALK fusion kinase were insensitive to inhibition when

treated with concentrations of up to 2,500 nM for both TKIs. This degree of selectivity is

Page 30: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

17

considered remarkable considering the selectivity profile of other kinase inhibitors and their loss

in potency upon acquired resistance. The same workers also observed foretinib suffers from a

greater loss in activity against ROS1G2032R (IC50 = 50.1 nM) compared to cabozantinib (IC50 = 15.3

nM). In conclusion, it is clear that cabozantinib can overcome the point mutation of the solvent-

front Gly2032 to Arg in the kinase domain of ROS1 following initial crizotinib treatment. This

provides a promising alternative therapeutic strategy for CD74-ROS1 positive NSCLC patients

suffering from acquired crizotinib resistance.

1.4 Conclusion

Both foretinib and cabozantinib have shown encouraging results in both wild type and

genetically modified G2032 CD74-ROS1 models. The structural variations between these

inhibitors and crizotinib must contribute to the observed differences in CD74-ROS1G2032R

inhibition in vitro. More specifically, the overall scaffold of the molecule offers a specific binding

conformation within the ATP recognition pocket, and this particular binding event is evidently

perturbed in the case of crizotinib-CD74-ROS1G2032. The specific interacting residues in the ROS1

kinase domain can help elucidate which functional groups/substituents contribute towards the

second-generation ROS1 TKIs in retaining their binding potency for the G2032R mutated kinase.

Furthermore, we can even look to gain more insight of the structural effects of the G2032R

substitution. This direction would involve designing cabozantinib/foretinib analogues with

incorporation of functional groups that interact with specific residues in the G2032R mutated

ROS1 kinase domain. Such a study would help decipher the exact mechanism of TKI binding to

the ROS1 kinase domain, an important question that has yet to be answered. We can even take a

step back and try to identify the structure-based resistance liabilities in crizotinib. The clinically

relevant G2032R mutation in CD74-ROS1 arises only upon exposing ROS1 fusion positive

NSCLC patients to crizotinib, suggesting the resistance pathway of CD74-ROS1 is induced upon

exposure to this specific TKI. Analyzing differences between the inhibitor-catalytic site interaction

for crizotinib and second generation TKIs will pinpoint if substituents on these molecules

contribute to amino acid substitutions in the first place. This will definitely help improve future

Page 31: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

18

efforts in TKI design as we can seek to chemically modify drug scaffolds that offer less resistance

liabilities.

The work presented in the following chapter details our efforts in identifying structural

contributions into the selectivity and resistance profiles of ROS1 TKIs, with special focus on

cabozantinib. We have employed computational modeling to study the lowest energy binding

conformation of relevant TKIs and the interactions they generate with the ATP pocket of both wild

type and G2032 mutated ROS1. We then designed a diverse library of analogues and performed

docking simulations with the ATP binding site of ROS1 to probe for any trends in predicted

binding affinities. From these results, we deduced a representative set of inhibitors which were

taken forward and synthesized. In vitro evaluation was conducted using biophysical and cell-based

assays.

Page 32: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

19

Chapter 2: Resistance and selectivity profiles of cabozantinib and

derivatives with structural insight

2.1 Computational studies of TKI interaction with ROS1

The X-ray crystal structure of the ROS1 kinase, PDB ID: 3ZBF, reported by Awad et al.

in 2013 was used for our computational studies (obtained from the Research Collaboratory for

Structural Bioinformatics Protein Data Bank). The reported crystal structure was incomplete and

required several in silico modifications. Firstly, there were 5 and 11 amino residues omitted from

the P- and A-loops, respectively. These key residues were introduced into the protein using the

Prime program of Schrödinger’s Suite (2012, Maestro 9.3). This protein-structure prediction

program allows insertion or substitution of residues and translates the modified linear sequence

into an accurate 3-dimensional structure. All the coordinates of the backbone atoms and side chains

of conserved residues in the protein were retained. Newly installed residue side chains were

inserted from a library of known peptide dihedral angles and side chain residues. In any tertiary

protein, however, the structural orientation, ionization state, and conformational rigidity of every

amino acid are governed by its microenvironment. Thus simple addition of the newly inserted

residues was not sufficient. Loop refinement is a task in Prime that calculates all local contacts and

generates multiple conformations of the loop.49 The conformations were clustered and energy

minimization was conducted using the OPLS_2005 force field. The output with the lowest total

energy of the system was selected as the final model.

All ATP competitive TKIs are known to interact with their target kinase either in a type I

or type II orientation.50 Type I kinase inhibitors recognize the catalytically active conformation of

the kinase where the DFG amino residues in the activation loop adopt the ‘in’ state. On the other

hand, type II TKIs engage the kinase in the inactive state where the DFG motif out from the kinase

ATP pocket, as shown in figure 2.1A.51 When the active loop is not phosphorylated, type II kinase

inhibitors are able to take advantage of the unique hydrophobic site that is adjacent to the ATP

binding region, highlighted in blue in figure 2.1.B.

Page 33: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

20

Figure 2.1- A) Active and inactive dynamic equilibrium of RTK enzymes controlled by

phosphorylation of the DFG motif in the A-loop; B) Anatomy of kinase domain with specific

regions highlighted. ‘Hydrophobic region I’ is the cleft adjacent to the adenine regions which gets

exposed upon dephosphorylation of the kinase to the inactive state. (Figure from reference 50).

Upon examining the inhibitor-bound ROS1 crystal structure, (see Chapter 1, figure 1.5)

crizotinib can be classified as a type I TKI. As expected, the 3-isopropoxypyridin-2-amine rests in

the adenine site, lined with A1978, E2027, and L2028. Hydrophobic contacts are predicted

between the 1,3-dichloro-4-fluorobenzyl ring of the inhibitor and V1959, L2086, and L2010 of the

protein. Most significantly, the piperidine substituent faces the solvent, surrounded by G2032,

D2033, and T2036 and the bridging pyrazole is positioned flat with respect to the

3isopropoylpyridin-2-amine ring, lying below L1951.

In the absence of an inactive ROS1 crystal structure, Prime was employed to produce a

suitable ROS1 model where the DFG triad in the activation loop is projecting out from the kinase

domain. The inactive ALK crystal structure (PDB code: 4FNY) was used as a template and

homology modeling was performed to construct the inactive ROS1 protein. The deviation in the

Page 34: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

21

position of the DFG motif in inactive ROS1 from its active state was comparable to that of ALK

(approximately ~10 Å with respect to the alpha carbon in Phe2103), see figure 2.2.

Figure 2.2- Overlap of the active and inactive conformations of the ROS1 kinase domains with

the DFG motif phenylalanine 2103 represented by a stick model. Image generated using

MacPyMol 2009-2010.

With the inactive and active states of the native ROS1 kinase domain model in hand, we

then proceeded to study the lowest energy binding poses of cabozantinib and foretinib in the ROS1

catalytic site using molecular docking simulations. The ligand-receptor docking program, GLIDE

(Grid-Based Ligand Docking with Energetics, version 6.1, Schrödinger Suite 2014) was employed

to run the docking simulations for both TKIs. Docking simulations required three steps before

running the docking algorithm: 1) preparation of the protein, 2) preparation of the ligand, and 3)

establishing docking parameters.

First, all waters excluding bridging between a co-factor and protein were deleted. The

ROS1 kinase structures were then individually optimized using restrained minimization, which

reorients any amino acid side-chain hydroxyl groups and readjusts steric clashes. Formal charges,

bond orders, all hydrogen atoms, and protonation states were added sequentially. An OPLS_2005

I n c $ v e ' R O S 1 '

A c $ v e ' R O S 1 '

Page 35: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

22

force field was used for optimization. Secondly, LigPrep, a ligand preparation module available

through Maestro (version 9.2, Schrödinger Suite 2014)52 produced single low-energy 3D structures

of cabozantinib and foretinib. From a 2D structural input, multiple functions within Ligprep were

utilized. Relevant chiralities and possible tautomers were inserted. In addition, ionization states of

the TKIs were generated within pH range of 7.0 ± 0.4. Finally, low-energy ring conformation

sampling and geometry optimization, using OPLS_2005 force field was performed to relax the

cabozantinib and foretinib 3D structures. A docking grid with dimensions of 15 Å for the ROS1

receptor was generated using the binding site residues defined from the crizotinib-bound ROS1

crystal structure. This precise location in the kinase domain fully encompassed the ATP binding

site with specific residues L1951, A1978, K1980, E1997, M2001, L2028, G2032, L2086, and

D2102. The dimensions of the grid box were selected such that all possible conformations and

rotations of the ligands could be accommodated within the ATP cleft.

Docking simulations were then carried out with a rigid protein structure with a flexible

ligand (includes acyclic torsion bonds, pyramidal nitrogen inversions of amides and sample ring

conformations) using a Lamarckian Genetic Algorithm (LGA).53,54 The GLIDE extra-precision

(XP) mode is a comprehensive sampling function that offers many advantages over the GLIDE

standard precision (SP) module. Not only does the XP mode account for all the energetically

favourable and non-favourable contacts, it incorporates a much more rigorous treatment of

solvation and hydrophobic interactions with the protein.55 GLIDE XP interprets interactions

between the ligand and protein involving charged and/or polar groups by comparing the overall

energy if these same functional groups were solvated, as in a natural biological system. By

assessing the difference in energies, GLIDE XP is able to assess any solvation inadequacies and

assign corresponding penalties in the scoring algorithm. Furthermore, the GLIDE XP tool looks to

reward contacts between the ligand and receptor involving hydrophobic substituents.

Lipophiliclipophilic interactions, pi-pi stacking, and even potential pi-cation interactions are

recognized and rewarded.55 This component in the GLIDE XP scoring algorithm includes the

significant non-polar contributions towards ligand-protein binding. Incorporation of the buried

hydrophobic sites in a receptor and its enclosure of lipophilic atoms in a ligand are crucial for

binding and simple nonpolar surface area analysis is not sufficient. We assessed the validity of the

molecular docking protocol by conducting an initial docking study with crizotinib and the

ROS1G2032 and comparing the obtained results with the published inhibitor-protein bound crystal

Page 36: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

23

structure (PDB code: 3ZBF). Fittingly, the conformation of the most populated conformation was

almost identical with that of the crystal structure. The superimposed crizotinib structures revealed

minimal root-meansquare deviation (RMSD) between the atoms. This control computational

analysis confirmed the accuracy and precision of our docking protocol.

Figure 2.3- Docking score histograms for active ROS1-cabozantinib (top), inactive

ROS1cabozantinib (bottom left) and inactive ROS1-cabozantinib (bottom right).

Analysis of the lowest energy binding poses for foretinib and cabozantinib revealed both

TKIs exhibited type II binding with the ROS1 kinase domain, with predicted binding energies of -

10.10 kcal/mol and -12.00 kcal/mol, respectively for the highest docking score, figure 2.3. In the

case of cabozantinib, the active state of ROS1 was not preferred as the energy for the highest

ranked binding pose was approximately -7.5 kcal/mol. As shown in in figure 2.4.A, the

6,7dimethoxyquinoline ring occupied the adenine pocket (coloured green) forming hydrogen

bonds with residues E2027 and M2029 amide backbone (figure 2.4.B). Interestingly, both

dimethoxy groups are solvent exposed and the rest of the scaffold projects in towards the tunnel-

shaped sugaradenine-linker/hinge-hydrophobic series of pockets. The tethering 4-aminophenol

aromatic ring participates in pi-pi stacking with F2013 of the DFG motif. Positioning of the

quinoline ring follows typical type II TKI binding profile wherein numerous inhibitor-RTK crystal

a ct i ve - co n f

R O S 1 ' : ' i n a c $ v e '

R O S 1 ' : ' a c $ v e '

R O S 1 ' : ' i n a c $ v e '

Page 37: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

24

structures confirm nitrogen-containing aromatic groups occupy the adenine-binding site. The

catalytic K1980 residue forms a hydrogen bond with the dicarboxamide functional group.

Importantly, the hydrophobic cleft adjoining the ATP recognition pocket captures the terminal 4-

fluorobenzene substituent through T-shaped and pi-pi aromatic interactions with F2004 and F2075,

respectively (figure 2.4.C). This specific set of molecular interactions involving ROS1 and a type

II TKI has never been reported and hence presents a focal point of our study. Finally, the

cyclopropyl appendage provides hydrophobic contacts with M2001 and L2070 in the linker/hinge

region (figure 2.4.B, coloured yellow), which is slightly above the plane of the quinoline and

4fluorobenzene.

Figure 2.4- Lowest energy conformation of cabozantinib bound to ROS1 kinase domain. Panel A

shows exposed methoxy groups and the quinoline ring occupying the ATP binding site (green).

The tethering 4-aminophenol and cyclopropyl carboxamide motifs reside in the hinge region,

highlighted yellow. Panel B shows a few of the interacting amino acids. Panel C illustrates

occupancy of the hydrophobic groove (adjacent to the adenine site) by the 4-fluorobenzene

appendage. Image generated using MacPyMol 2009-2010.

Foretinib is a very similar structural analog of cabozantinib with the fundamental quinoline,

carboxamide linker, and 4-fluorobenzene-containing scaffold retained.Foretinib has an additional

morpholine ring attached to the 3-phenoxy of the quinoline through a 3-C linker along with a

2fluoro-4-aminophenol bridging system. Analysis of the lowest-energy docked pose of foretinib

E 2 0 2 7 *

M 2 0 2 9 * *

D 2 1 0 2 * L 2 0 8 6 *

L 2 0 7 0 *

B *

A *

C *

Page 38: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

25

revealed a molecular conformation analogous to cabozantinib. Similarly, the morpholine-branched

6,7-dimethyoxyquinoline sits in the ATP site with the morpholine lying solvent exposed and

interacting with residues K1976 and E2030. The remainder of the inhibitor traverses the narrow

ROS1 kinase domain tunnel in the identical cabozantinib-manner without much variance in the

interacting protein amino residues. Closer examination of the distribution cluster of the

foretinibROS1 docking simulation however showed a possible reverse binding conformation,

figure 2.5. The type II inhibitor model is retained, however, the orientation of the molecule in 32%

of the poses, is completely inverted with respect to the anatomy of the ROS1-cabozantinib

complex. The morpholine group engages the hydrophobic cavity whereas the 4-fluorobenzene ring

recognizes the adenine-binding region.

ROS1@CabozanBnib* ROS1@ForeBnib*

Figure 2.5- Alternative, reversed binding pose of foretinib with ROS1 compared to predicted

cabozantinib-ROS1 complex. Image generated using MacPyMol 2009-2010.

With a well-characterized inhibitor-protein interaction blueprint, both cabozantinib and

foretinib were good primary candidates to use in our study of the molecular mechanism of mutation

and subsequent acquired resistance in the ROS1 kinase domain. While we were conducting our

docking studies, it was reported that the ongoing clinical trial for foretinib as a therapeutic for

NSCLC was withdrawn. Novartis had acquired GlaxoSmithKline’s oncology portfolio and

terminated development of foretinib partially due to possible its toxicity issues in the trial and

Page 39: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

26

cabozantinib’s F.D.A. approved status. With foretinib being dropped as an inhibitor for

CD74ROS1 positive NSCLC in the clinic, only cabozantinib was taken forward in the project.

There would be no potential drug resistance due to inhibitor-induced point mutations in the case

of foretinib and all therapeutic efforts for ROS1 were being directed towards cabozantinib.

Additionally, as the docking studies indicated, approximately one-third of the foretinib binding

conformations adopted a reversed orientation, suggesting a dual binding mode in presence of the

G2032R mutation. This system could be further resolved with molecular dynamic simulations to

probe for any induced binding effects and their corresponding energy calculations. Unfortunately,

such studies were out of the scope of this project, and thus, cabozantinib was the ideal candidate

given its affirmed interaction with the ROS1 kinase domain and higher four-fold higher potency

against ROS1G2032 relative to foretinib.

2.2 Proposed library of cabozantinib derivatives

We segmented cabozantinib based on the different pockets recognized by the TKI and its

corresponding interactions. The 6,7-dimethoxyquinoline and 4-fluorobenzene groups provided

maximum scope for a structure-based inspection on the inhibitor binding mechanism. As

previously mentioned, the methoxy groups are solvent-exposed, protruding from the

adeninebinding site. We postulated this could be due to hydrophobic driven effect since the

methoxy groups are surrounded by G2031 and L2034 and most importantly, G2032. This

hypothesis could certainly be tested by comparing the docking and accompanied in vitro binding

activity of cabozantinib analogues with various alkyl groups on the 6,7-diphenoxyquinoline motif.

We decided to replace the methyl groups with ethyl, propyl, iso-propyl, and tert-butyl along with

the parent non-substituted 6,7-diphenoxyquinoline. Installing bulkier alkyl groups and comparing

their binding profiles with cabozantinib would allow us to deduce the impact of the hydrophobic

effect and any trends derived from sterics. Furthermore, probing the substituents in close proximity

to the crucial 2032 residue will lend insight in to its role in inhibitor binding.

With regards to the 4-fluorobenzene appendage, numerous substituents are synthetically

feasible including: different sized-aromatic rings, numerous heteroaromatic systems, variable

Page 40: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

27

regioisomeric, and atomic substitutions of the parent 4-fluorobenzene group. The cavity

surrounding this motif is quite spacious, lined with non-polar residues M2001, F2004, L2070,

F2075, H2077, and I2010. Our proposed derivatives included furan, pyrrole, pyridine, 5-fluoro-

2pyridinyl, and 6-fluoro-3-pyridinyl rings. For the furan and pyrrole, connectivity to the

carboxamide through C2 and C3 were considered. We generated a comprehensive first round

library of cabozantinib analogues with each permutation of both ends of the scaffold. Every

molecule was subjected to molecular docking simulations with the inactive ROS1G2032R kinase

domain. The derivatives proposed are summarized in figure 2.6 below.

Figure 2.6- Proposed structural modifications to cabozantinib for fragment-binding computational

studies.

At the time of proposing the library of cabozantinib structural derivatives, Schrödinger

Suite was unavailable to conduct the computational studies. We opted to use the AutoDock 4.2

(Scripps Research Institute, 2013) program to study the receptor-ligand interactions of our system.

Just like Glide, AutoDock 4.2 utilizes the LGA for its search methods and follows an identical

protocol where the protein, ligand and receptor grid are prepared independently.56

Page 41: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

28

First, all the polar and non-polar hydrogens of the protein were introduced with partial

charges (using Gasteiger method) applied to each atom. Following this, three-dimensional

structures of cabozantinib and its derivatives were prepared using Chem3D Pro 12.0 with

energyminimization using an MM2 force field calculation with a minimum RMS gradient of 0.000.

The coordinates of the receptor grid box used for Glide was duplicated and implemented for

AutoDock 4.2 purposes. Docking simulations were then executed using global and adaptive local

search parameters through 100 trials of the “long” LGA runs. All calculations were performed with

a rigid protein structure and a flexible ligand. Clusters of conformations were populated based on

their free energy of binding with ROS1.

To validate the consistency of Autodock 4.2, we first docked cabozantinib with the

ROS1G2032R kinase domain and looked to compare the output with the results from Glide. To our

satisfaction, Autodock 4.2 produced a lowest energy conformation with a structural orientation and

corresponding predicted binding energy (ΔG = -11.20 kcal/mol) very similar to that of Glide (ΔG

= -12.00 kcal/mol).

From the docking results, we observed that increasing steric bulk on the

6,7diphenoxyquinoline confers more favourable binding affinity (Ki). The predicted Ki for

cabozantinib was approximately 1 nM whereas for the 6,7-di-tert-butoxyquinoline analogue, Ki <

1 x 10-2 nM. The ethyl, propyl, and iso-propyl containing alkyl derivatives had similar Ki’s to that

of the original inhibitor. With regards to the lowest energy conformation of the proposed

molecules, no significant deviations were observed. The 4-fluorobenze retained its occupancy of

the adjacent hydrophobic site while the quinoline ring remained in the ATP pocket. This trend can

only be supported by entropically driven hydrophobic effects during binding. The interaction of a

lipophilic group or surface in the receptor with a bulky substituent in the ligand is entropically

favoured by the displacement of ordered water molecules, hence an overall beneficial contribution

to ΔG of binding. Calculated Ki values (nM) for representative compounds are summarized in

figure 2.7 below.

Page 42: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

29

R&Group&

Figure 2.7- Tabulated Ki values (nM) for alkyl group derivatives of 6,7-dimethoxyquinoline

position from docking screen.

Docking results of molecules containing the 4-fluorobenzene substitutions did not indicate

any meaningful trends in terms of ring size, identity of heteroatom, nor presence of fluorine. The

predicted Ki values of all the derivatives were within 4-9 nM, with cabozantinib (4-fluorobenzene)

exhibiting the most favourable binding profile, Ki ≈ 1 nM figure 2.8. Similar to the trend with the

analogues mentioned above, the aromatic rings did not affect the conformation of the lowest energy

pose. All of the appendages occupied the cavity adjacent to the ATP site while the

6,7dimethoxyquinoline motif recognized the adeninecleft..

110$

1 0.1$ R&Group&

Page 43: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

30

10$ 1$

Figure 2.8- Tabulated Ki values (nM) for alkyl group derivatives of 6,7-dimethoxyquinoline

position from docking screen.

Since our first library of proposed molecules did not provide any promising candidates for

examining the ROS1 binding mechanism, we looked to other derivatives for our purposes. We

continued our focus on the two terminal ends of the structure as all lowest energy conformations

consistently showed the bridging aminophenol and cyclopropylcarboxamide residing in their

respective well-defined cavities. We looked to substitute the para-fluorine group with other

halogens: Cl, Br, and I along with other isosteres: H, CH3, and CF3. This initial small set of

derivatives could potentially help delineate any specific contribution of the para substituted

benzene ring towards cabozantinib’s unprecedented type II inhibition of ROS. In addition, the

revised 6,7-dimethoxyquinoline-structure activity relationship (SAR) would encompass the

originally proposed ethyl, propyl, iso-propyl, and tert-butyl, as well as new substituents including,

iso-butyl, n-butyl, bridging 2C, 3C and 4C alkyl chains and finally, the monomethoxy analogues.

With this library, we intended to looktowards second-generation analogues and compare selectivity

and resistance profiles for both wt ROS1 and ROS1G2032R through in vitro assays.

O

R O

F

F

O

O

N

O H N H

N

O O

H N

H N

N

N F

N

Page 44: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

31

2.3 Synthesis of small-molecule TKIs of ROS1

We embarked on the synthesis of cabozantinib and its select derivatives using the synthetic

procedure outlined in the original patent (U.S. Patent No. 8,067,436 B2, Nov. 29, 2011)57 with

appropriate deviations to incorporate necessary changes. We approached the molecule through a

convergent synthesis by preparing the 4-fluorobenzene modified cyclopropane-1,1-dicarboxylic

acid and the 6,7-dimethoxyquinoline synthon incorporating the bridging 4-aminophenol. The

commercially available monomethylester of cyclopropane-1,1-dicarboxylic acid 1, was

transformed to its corresponding acid chloride, using excess thionyl chloride and reflux conditions.

Amide coupling (THF with DIPEA) with various para-substituted anilines afforded amides 2a-g

and subsequent saponifaction using LiOH.H2O in THF/H2O produced the modified carboxylic

acids 3a-g, scheme 2. We later discovered that simple amide coupling using (2-(1H-benzotriazol1-

yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU) proved to be equally efficient but

could be performed with milder conditions than acid-chloride activation. The addition step with

numerous para-substituted anilines provided a synthetically feasible divergence point necessary

for the library.

Scheme 2-1. Amide coupling synthesis of cyclopropylcarboxylic acid and aniline: (i) SOCl2,

reflux, 2 h; then aniline, THF, DIPEA, rt, 16 h; (ii) aniline, HBTU, DIPEA, DMF, rt, 18 h; (iii)

LiOH.H2O, THF/H2O, 0 oC to rt, 2 h.

Synthesis of the remaining half of the inhibitor scaffold required a more elaborate

procedure as a majority of the alkyl derivatives, the corresponding amines were unavailable. We

Page 45: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

32

looked to synthesize the desired amines by first alkylating 4-nitrocatechol (4), followed by

reduction of the nitro group to produce the amine substrate. Rigorous method development was

conducted to obtain a facile and robust alkylation protocol. We initially used 1,3 dibromopropane

as the substrate to survey ideal reaction conditions. Attempts included using DIPEA (2.2 equiv.),

and 1.5 equivalents of the dibromide in numerous solvents: CH3CN, DMSO, DMF and THF. With

no significant conversion to the desired product, we opted to use Cs2CO3 base in the same set of

solvents. However, product yields were quite low and further synthetic improvements were

required. We considered a much stronger base, NaH, however the alkylation did not proceed in

desirable yields at room temperature and even upon increasing the temperature to 80 oC. Addition

of AgNO3 to force the reaction forward via formation of AgBr salt also proved to be futile.

Ultimately, we were able to successfully carry out the alkylation of the 4-nitrocatechol with

1,3dibromopropane using K2CO3 in ethylene glycol at 110 oC for 8 h. This protocol was applied

to various alkyl bromides to synthesize the corresponding alkylated intermediates, 5a-h.

Page 46: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

33

Scheme 2-2. Alkylation of catechol with numerous alkyl halides and subsequent preparation

of aniline: (i) 1,3-dibromopropane, K2CO3, ethylene glycol, 75-110 oC for 8-10 h; (ii) H2(g), Pd/C

(10%), THF/MeOH.

Next, we explored two sets of reduction conditions for conversion of the nitro group in 5ah

to the amine. First, we attempted SnCl2 reduction in methanol with refluxing conditions.

Unfortunately, there was minimal conversion to product even with addition of concentrated

HCl/AcOH forcing us to resort to another reduction procedure. As a test reaction, we discovered

catalytic hydrogenation of 4-methoxynitrobenzene with palladium on carbon proved to be

promising, with a convenient work-up and purification (92% yield). Upon extending these

conditions to all nitro-substrates, we were pleased to see efficient 75%-96% yields conversion to

the corresponding anilines, 6a-h. For the 6,7-dimethoxy, 6-methoxy, 7-methoxy, and

nonsubstituted quinolone systems, the commercially available anilines 6i, 6j, 6k and 6l were

purchased.

Entry! R1! R2! Entry R1 R2

a! !O

O !

h(ii)!

!

O

b!

c!

i! H ! H !

d!

j! O ! H !

k(i)!

H O

e!

H3CO f!

N O O

O ! O

! O

!

O !

O

! O

! O

! O

! O

N

Page 47: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

34

O O

k(ii)!

H O

g! O O

CH O

h(i)! O O

O

l O O

Scheme 2-3. Cyclization of quinolone using microwave-assisted conditions, chlorination and

subsequent coupling: (i) 90 oC, 10 min, microwave assisted; then aniline 6a-l, EtOH, reflux, 2h;

(ii) Ph2O, 230 oC, 10-20 min, microwave assisted; (iii) POCl3, reflux, 2 h; then cold H2O, Na2CO3;

(iv) NaH, 4-aminophenol, DMSO, 10 minutes; then chloroquinoline 12a-l DMSO, 100 oC, 12h.

The remainder of the quinoline ring was installed in a three-step procedure. First, enolate

alkylation of 2,2-dimethyl-1,3-dioxane-4,6-dione (7), with triethylorthoformate (8), under

microwave-assisted conditions (90 oC for 10 min) afforded the 1,3-dioxane-4,6-dione intermediate.

This crude reaction mixture was taken forward to imine condensation with anilines 6a-l in refluxing

EtOH. For all the imine products 9a-l, filtration of the reaction mixture and subsequent EtOH

washes were sufficient as further purification was not required. The third and final step was a

decarboxylation-mediated ring closure performed in diphenyl ether at 230 oC for 10 min in

microwave-accelerated conditions. In some cases, if there wasn’t sufficient conversion to the

quinolin-4-ols product (10a-l), the reaction was setup again with identical conditions for 20

minutes.

Interestingly, we detected the presence of two regioisomeric products when subjecting 9h

and 9k to the quinoline ring closure conditions mentioned above. The presence of two isomers can

be attributed to the first step in the mechanism of the reaction, where attack at the 1,3-dioxane-

4,6dione ester group can proceed either through C8, as shown in mechanism A, or through C6,

indicated in mechanism B for imine 9h, figure 2.9. Para-directed mechanism A yields quinolin-

4ol 10h(i), whereas the ortho-selective attack in mechanism B produces 10h(ii). Similarly, in the

case of 9k, nucelophilic attack is plausible through either C2 (see mechanism C) or C6 (mechanism

D). The two regioisomers 8-methoxyquinolin-4-ol 10h(i), and 5-methoxyquinolin-4-ol 10h(ii),

were easily separated using silica gel chromatography. Unfortunately, complete purification of

!

!

!

!

N

3

!

! O N

Page 48: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

35

10k(i) and 10k(ii) from the product mixture was not possible as each intermediate retained slight

impurity of its regioisomeric partner (~10%). This was carried forward in the remainder of the

synthesis with the possibility that the resulting final compounds could be separated with the use of

much more comprehensive purification techniques such as HPLC.

Mechanism*D* O

Figure 2.9- Proposed mechanisms for decarboxylation-mediated cyclization reactions of imines

9k and 9h. Mechanisms A and B illustrate how 9k, can give two regioisomers, 10k(i) and 10k(ii).

Similarly, diverging mechanisms C and D outline the formation of 10h(i) and 10h(ii).

With the set of substituted quinolin-4-ols 10a-l in hand, we looked to add the bridging

4aminophenol motif using nucleophilic aromatic substitution (SNAr). The 4-position alcohol was

chlorinated in the presence of excess POCl3 and reflux conditions, furnishing 4-chloro-

6,7dialkoxyquinolines, 11a-l, in essentially quantitative yields. To drive the chemoselectivity of

the SNAr for the free -OH, 4-aminophenol was initially treated with NaH, followed by addition of

the 4-chloroquinoline substrate. For all intermediates 12a-l, this methodology proved to be

successful in selecting the phenol as the nucleophile in the SNAr. The final coupling step was

attempted using traditional peptide coupling reagents such as HBTU or N,N′-

Page 49: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

36

Diisopropylcarbodiimide (DIC) but with little success as significant quantities of both starting

materials were recovered. We then elected to pursue convert the acids 3a-g to produce their

corresponding acid chlorides using thionyl chloride. Cabozantinib (13a) was successfully

synthesized by coupling the acid chloride of 3a with aniline 12l under mild conditions (1.5 equiv.

DIPEA, 1.1 equiv. K2CO3, THF, room temperature).

This final coupling procedure was adapted to synthesize the library of cabozantinib analogues 13at,

shown in figure 2.10. Due to synthetic challenges associated with the alkylation of the

4nitrocatechol in the early stages of the scheme, 19 final molecules were prepared. Synthesis of

the 6,7-di-tert-butyl derivative proved to be quite challenging and hence was omitted.

F N

O

H N

O

H N

O F

O O

H N

O

H N

O O

O

N

Page 50: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

37

O 13s 13t

Figure 2.10- Preliminary library of synthesized analogues of cabozantinib, 13a-t.

2.4 In vitro evaluation of preliminary library in cell models

The preliminary cabozantinib derivatives were screened for anti-proliferative activity in

Ba/F3 cells expressing native CD74-ROS1 or CD74-ROS1G2032. Cabozantinib and crizotinib were

used in the MTS cell viability assay as a control. Parental Ba/F3 cells were cultured in RPMI

medium 1640 with additional 10% (v/v) FBS, L-glutamine, penicillin/streptomycin. Cells were

supplemented with 15% (vol/vol) WEHI-3– conditioned media as a source of IL-3 and maintained

at densities of 0.5 × 106 to 1.5 × 106 per mL. The cell cultures were infected with retrovirus either

encoding native or G2032R mutant human CD74-ROS1. Stable cells were confirmed using

puromycin-based selection by GFP expression using a FACS Aria flow cytometer. Stable cell lines

were then washed to remove exogenous IL-3. Cells exhibiting healthy proliferation and survival

post IL-3 removal were maintained and carried forward for cell proliferation assays.

All inhibitors were prepared as 1 mM stocks in DMSO before each experiment. Ba/F3 cells

expressing CD74-ROS1 and CD74-ROS1G2032R constructs were seeded (800 cells per well; 25 μL)

into 384-well plates. Final compounds were added with 25 μL per well of complete medium.

Twofold dilution series format was setup with maximum concentration of 500 nM of inhibitor (12

concentrations per compound). The plates were incubated for 72 h at 37 °C, 5% CO2. Viability was

measured using a methanethiosulfonate-based assay (CellTiter96 Aqueous One Solution;

Promega) read on a Biotek Synergy 2 plate reader (emission wavelength = 490 nm). All

experiments were performed at least two independent times in triplicate. Data were normalized

using Microsoft Excel, and absolute IC50 values were calculated for each inhibitor with GraphPad

Prism software using a nonlinear curve fit equation and results are tabulated below, figure 2.11.

Page 51: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

38

Figure 2.11- Graphical representation of IC50 values of preliminary library of small-molecule

inhibitors 13a-t and crizotinib in Ba/F3 CD74-ROS1 and CD74-ROS1G2032R cells.

Unfortunately, the preliminary library of TKIs did not reveal any noticeable trends in terms

of cell viability inhibitory activity. Most notably, there were no inhibitors that were more potent

than cabozantinib. For the majority of the compounds tested, 50% of cell death was observed only

at concentrations close to the maximum drug dose, 500 nM, for Ba/F3 CD74-ROS1 cells. Activity

against Ba/F3 CD74-ROS1G2032R cells was significantly diminished with only inhibitors 13b and

13f having IC50 values comparable to cabozantinib, 13a. We are currently pursuing analysis of

levels of pROS1 and its downstream signaling proteins, pSTAT3, pAkt, pERK and pSHP-2 by

Western blot in both cell lines.

2.5 Binding constant determination using biophysical assay

We considered a variety of biophysical assays that would allow us to characterize the

potency and selectivity of our compounds for the wild type ROS1 kinase domain and its G2032

mutant in vitro. Ideal experiments would provide inhibitory constant (Ki) values that would define

the concentration of TKI required in order to inhibit the maximal enzymatic activity by 50%. Direct

modulation of an enzyme’s activity by an inhibitor is an important approach to study the system of

interest without interfering biological cascades. We can effectively adapt such an assay for a high-

throughput screening (HTS) format and use this as a benchmark to rank inhibitors in terms of

1 3 f * 1 3 a * 1 3 g * 1 3 c * 1 3 e * 1 3 b * 1 3 d * 1 3 h * 1 3 i * 1 3 j * 1 3 k * 1 3 l * 1 3 m * 1 3 n * 1 3 o * 1 3 p * 1 3 q * 1 3 r * 1 3 s * 1 3 t *

Page 52: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

39

potency. Academic groups and industry have extensively utilized a host of kinase enzymatic assays

given the popularity of TKI drug development programs.58 These kinase assay technologies

generally require quantification of a species in the phosphoryl transfer reaction catalyzed by the

enzyme. Methods include detection of ATP consumption, measuring ADP accumulation or

tracking the formation of the phosphopeptide/phosphoprotein product.59 Fluorescence-based

detection assays are the most commonly used approach for TKI profiling given their HTS

compatibility, relative ease of use and wide applicability to numerous kinases.60,61 Techniques

relying on fluorescence measurements, however, are liable to false-positive and false-negative

results due to interfering or non-specific fluorescence output. The contribution of fluorescent

tracers, labeled substrates or fluorescent compounds to high background and interfering signals is

an important consideration for assay design including FI, FP and even FRET experiments.71

In light of the susceptibilities of the previously mentioned fluorescence-based procedures,

we looked to another assay to attain valuable binding data for our TKI-ROS1 kinase domain study.

Time-resolved-fluorescence resonance energy transfer (TR-FRET) provided a suitable alternative

as it offers the advantages of both FRET and time-resolved fluorescence (TRF) spectroscopy. TRF

is considered an extension of fluorescence spectroscopy where the emission of a sample is

monitored over time after the initial excitation. TR-FRET looks to eliminate the hampering

background fluorescence observed in conventional FRET by introducing a time delay between

excitation of the acceptor chromophore and signal acquirement.72 Given the background

fluorescence signals have a typical lifetime in the range of nanoseconds, time-resolved

measurements this short-lived emission can be cleared from the measured fluorescence. To achieve

suitable FRET with a time delay, the donor species must have a fluorescence lifetime that is much

higher than that of any background or undesirable emission. Lanthanide ion complexes such as

europium (III) and terbium (III) chelates are commonly used for these purposes, given their long

emission lifetimes (10-6 to 10-3 seconds). 73 A comparatively longer lifetime will allow the decay

of transient signals and the observed emission to be a result of only energy transfer from the donor

to the acceptor. The combination of FRET and TRF results in a much more sensitive and reliable

assay. Specifically, we elected to use the LanthaScreen TR-FRET assay developed by Invitrogen.

A brief representation of the assay is presented below in figure 2.12.

Page 53: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

40

Figure 2.12- Illustration of principle of LanthaScreen (Invitrogen) TR-FRET assay. (Figure taken

from reference 74).

The principle of this assay is based on FRET occurring between a europium(III) labeled

antibody bound to the kinase and a proprietary chromophore (Alexa Fluor 647) conjugated

ATPcompetitive tracer.74 The ROS1 protein is modified with a GST tag recognized by the Eu(III)

containing anti-GST antibody in solution. Binding of the Alexa Fluor 647 containing tracer to the

ROS1 kinase domain is preferred at the ATP recognition pocket since the tracer is derived from

scaffolds of various ATP mimetic TKIs. Excitation of the Eu(III) chelate complex with light is

succeeded by energy transfer to the proximal Alexa Fluor 647-tracer species and its associated

emission. Thus, concurrent binding of the Eu(III)-antibody and Alexa Fluor 647 labeled tracer to

ROS1 results in a high degree of FRET since both species would be in ideal molecular proximity

when bound. A TKI specific for the ROS1 catalytic site would compete for binding with the tracer,

thus disturbing the proximity of the FRET pair. As the inhibitor is titrated in solution at increasing

concentrations, the Alexa Fluor 647 labeled tracer would continue to get displaced, thus decreasing

overall FRET signal. In further support, Invitrogen’s LanthaScreen TR-FRET assay is appropriate

for both type I and type II TKIs with high sensitivity. We attempted to assess the in vitro potency

of our inhibitors and cabozantinib for the native ROS1 protein using the LanthaScreen assay.

Prior to any inhibitor competition experiments, a calibration experiment had to be

performed to calculate the KD of the tracer. The KD value calculated from this binding experiment

would be used to approximate the concentration of the tracer required in the inhibitor titrations.

Ideally, we would like to select tracer concentrations that extract the maximum dynamic range

from the experiment from completely bound tracer (high FRET) to free, unbound tracer (low

Page 54: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

41

FRET). The calibration experiment required a final concentration of 3 nM ROS1 kinase protein

and 6 nM of the Eu(III) chelate antibody in each well. Titration of increasing concentrations of

tracer would lead to increases in concentration of the antibody-ROS1-tracer complex, thereby

generating a dose-response curve with respect to observed FRET. The concentration of tracer

ranged from 0.0153 to 500 nM (2 fold dilutions). An additional experiment was performed in the

presence of 0.150 nM of staurosporine, a universal kinase inhibitor. In this negative control, excess

staurosporine occupies the ATP site in ROS1, thus negating any binding of the tracer. Any potential

energy transfer from the excited Eu(III) antibody to unbound tracer in solution can be monitored

with this setup and accounted for in the calibration curve and subsequent inhibitor experiments.

GST-tagged ROS1 protein, 5X kinase buffer A (1X buffer contains: 50mM HEPES pH 7.5, 10

mM MgCl2, 1 mM EGTA, 0.01% Brij-35), kinase tracer 236 (50 μM stock in DMSO),

LanthaScreen Eu-anti-GST antibody and staurosporine were obtained from Invitrogen. Assays

were conducted in triplicate in clear low-volume 384-well plates (Greiner). Addition of all the

components was followed by a 60-minute incubation at room temperature. All fluorescence

measurements were performed with the BioTek Cytation3 plate reader instrument. Acceptor

LanthaScreen Eu(III)-anti-GST antibody excitation was at 340 nm, emission of the Eu(III)-chelate

bound antibody was 615 nm and kinase tracer emission was measured at 665 nm. A delay time of

100 μs was inserted between excitation and fluorescence measurement. The emission ratio was

calculated by dividing the emission of the Alexa Fluor 647 conjugated tracer (acceptor, 665 nm)

by the emission of the Eu(III)-tagged antibody (donor, 615 nm). Concentration of tracer was plotted

against emission ratio in the presence and absence of staurosporine.

We were able to retrieve an ideal calibration curve for binding of the tracer with ROS1 protein

with complete saturation of the kinase, figure 2.13. Based on the tracer-ROS1 saturation

experiment, the KD was determined to be 1.272 nM with 95% confidence interval of 1.123 to 1.421

nM.

Page 55: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

42

Figure 2.13- Calibration curves of Alexa Fluor-647 tracer titration with ROS1 kinase with 100μs,

time delay between excitation and measurement of emission.

Competitive inhibitor experiments were then performed to determine the KD values for the

preliminary library based on the quadratic mathematical model describing competitive binding of

two different ligands to a protein molecule described by Wang, 1994.66 Experimental procedure

was identical to that of the calibration experiment described earlier, except each well had a final

10 nM of tracer, 1 nM of ROS1 protein and 4-fold inhibitor concentration range from 0.024 nM to

100 𝛍M. Emission ratios obtained were plotted against concentration of inhibitor and were fit using

GraphPad Prism 5.0 to obtain dissociation constants. The KD values obtained are summarized in

table 2.1 and figure 2.14 below (binding curves are included in supplemental information).

Table 2.1- Summarized KD values for cabozantinib and analogues determined by

LanthaScreen TR-FRET assay.

Compound ROS1 KD (nM)

13a 21.43 ± 1.642

13b 16.8 ± 1.563

13c 30.5 ± 4.467

13d 341.2 ± 25.29

13e 196.7 ± 19.44

13f 56.22 ± 3.829

Page 56: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

43

13g 412.92 ± 38.37

13h 413.8 ± 43.62

13i 966.3 ± 162.8

13j 5433 ± 1832

13k 7359 ± 1907

13l 63.88 ± 7.324

13m 237 ± 29.27

13n 61.43 ± 7.247

13o 2850 ± 374.6

13p 8056 ± 1273

13q 5095 ± 611.4

13r 4518 ± 585.6

13s 941.5 ± 89.28

13t 1526 ± 249.9

Figure 2.14- Graphical representation of calculated KD values for initial library in addition with

crizotinib and staurosporine controls.

Page 57: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

44

Cabozantinib had a KD value of 21.43 ± 1.642 nM, whereas there was no significant improvement

in binding potency with the analogues. Interestingly, in the MTS cell proliferation assay,

cabozantinib’s IC50 value of 8.4 nM (in Ba/F3 transformed CD74-ROS1 cells) is lower than its KD

for the protein, indicative of off-target binding and slight non-selective profile. Trends between the

inhibitor binding potency correlate strongly with that of the cell cytotoxicity results in section 2.5.

Small isostere replacements of the para-fluorine position rendered a similar in vitro activity profile

both in CD74-ROS1 transformed cells and the LanthaScreen assay with the chloroderivative, 13b

(KD = 16.8 ± 1.563 nM), bromo-derivative, 13c (KD = 30.5 ± 4.467 nM) and methyl group, 13f,

(KD = 56.22 ± 3.829 nM). There was eventual loss in activity with the bulkier iododerivative (13d)

and simple phenyl derivative 13e (KD = 30.5 ± 4.467 nM). An incremental loss in affinity for ROS1

kinase was seen with bulkier alkyl substituents on the quinoline ring (13l-p) and activity was

completely lost with the cyclized analogues bearing bridged alkyl units (13q-t). The trends

observed in both the cellular and biophysical in vitro results for the bulky alkyl analogues such as

the ethyl, propyl, iso-propyl, n-butyl, iso-butyl heavily contrast with the Ki values predicted by the

docking model in GLIDE (section 2.3). This disparity cannot be directly attributed to the entropic

contribution towards binding as water molecules are removed from the protein structure for the

docking simulations. As a result, displacement of water molecules from the binding pocket cannot

be deemed as a contributor to ΔGbinding for the more hydrophobic analogues in the series. It is

possible that the favourable hydrophobic contacts and corresponding surface area calculated by

GLIDE in silico do not translate in vitro when considering a dynamic protein free in solution or

even in the context of a cellular environment.

Concluding remarks

The in vitro evaluation of our library of cabozantinib analogues is still in its early stages. We are

currently looking to use the LanthaScreen assay first and rank compounds based on their binding

affinity with the native ROS1 protein. Furthermore, we have considered additional experiments

such as a thermal-shift assay to gain further information on inhibitor and protein interaction. Other

typical experiments that monitor a binding event including surface plasmon resonance (SPR)

spectroscopy and isothermal calorimetry (ITC) are being considered as well.

Page 58: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

45

From the thermodynamic and kinetic binding data generated from these experiments, we can

investigate any potential trends and elucidate the substituents that contribute to recognition of the

ATP site in the native ROS1 kinase domain and its G2032 mutant. Ideally, these results can

corroborate with our computational modeling work and highlight the scaffold features of

cabozantinib that contribute to its unique in silico predicted binding mode and in vitro potency and

selectivity for ROS1 and ROS1G2032. Additionally, we are also interested in the effect on the levels

of pROS1 and downstream signaling proteins by western blot analysis in Ba/F3 CD74-ROS1 and

Ba/F3 CD74-ROS1G2032 cells to ensure target selectivity.

Page 59: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

46

Chapter 3: Design of peptdiomimetic inhibitors of signal transdcuer and

activator of transcription factor 5 (STAT5)

3.1 Introduction to Signal Transducer and Activator of Transcription Factor 5

(STAT5)

Signal transduction pathways serve as the basis of biological communication to efficiently

regulate cellular cycle growth, survival, and apoptosis. These molecular pathways begin with the

interaction of an external molecule with its specific target followed by the succeeding downstream

cascade. Generally, the event of exogenous ligand-receptor binding is relayed to protein-protein

interactions (PPIs) that eventually modulate the level of transcription of specific genes.75 One

specific protein mediated signalling network of physiological interest is the signal transducer and

activator of transcription (STAT) associated pathway. The study of this network of signaling

molecules is of significance due to its relevance with numerous human cancers: blood, prostate,

breast, pancreas, lymphatic system, and liver.76,77,78 The presence of excessive growth factors,

hyperactive kinase activity, oncogenic mutations or a decrease in activity of the negative regulators

of the STAT cascade leads to over-expression of anti-apoptotic and proliferation inducing genes.

Collectively, these factors lead to the aberrant activity of the STAT transcription factors and

eventually drive the oncogenic phenotype.

The STAT family of signal transducers consists of seven members, namely: STAT1,

STAT2, STAT3, STAT4, STAT5A, STAT5B, and STAT6. STATs have been characterized as

90115 kDa proteins consisting of approximately 750-900 amino acids.79 Genes encoding for the

various STAT amino acid sequences map to three different chromosomes. STAT3, STAT5A, and

STAT5B map to chromosome 17; STAT1 and STAT4 map to chromosome 2; finally, STAT2 and

STAT6 map to chromosome 12. Out of the seven members of STAT transcription factors, STAT5A

and STAT5B have been profoundly associated with cellular transduction mechanisms that mediate

hormone-related responses and cellular growth events. More importantly, STAT5A and STAT5B

signaling is heavily involved in the regulation the growth and maintenance of the hematopoietic

and lymphoid systems. 80

Page 60: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

47

3.2 Structure and signaling of STAT5

The STAT5 protein was originally isolated in mouse mammary glands in 1991 by

SchmittNey et al. and was identified as a mammary gland factor (MGF), responsible for the

transcription of the β-casein gene.81 Further studies conducted by Gouilleux et al. identified STAT5

as a signaling molecule which mimicked the effects of prolactin.82 Upon finding a significant

degree of sequence homology with the STAT transcription factor family, MGF was renamed,

STAT5. In 1995, an isoform of STAT5 was characterized in murine mammary glands and was

confirmed as a separate gene product from MGF.83 This homolog was named STAT5B whereas

the original MGF/STAT5 was termed as STAT5A. Consequently, the proteins share a high degree

of similarity in amino acid sequence (approximately 93%). The structural and functional units of

both homologs are shown below, figure 3.1:

Figure 3.1- Schematic illustration of STAT5A/B domains with critical Y694 (STAT5A) and

Y699 (STAT5B) highlighted. (Image generated using MacPyMol 2009-2010)

The STAT5A and STAT5B proteins consist of six well-defined domains that are

structurally and functionally conserved amongst the STAT family. The N-terminal domain (NTD)

contributes to stabilization of the STAT-dimer and even the STAT-DNA complex. Neighbouring

the NTD, is the coiled-coil domain (CCD), a four α-helical motif rich in hydrophilic amino

acids.84,85 The CCD recognizes native binding partners including nuclear transport proteins,

receptor α-helix motifs, chaperones, and intranuclear co-transcription proteins.86,87 Adjacent to the

CCD is the DNA-binding domain (DBD), an arrangement of β-sheet units that bind to the

N " t e r m i n a l + + N H 2 " + c o i l e d " c o i l + + D N A + b i n d i n g + L i n k e r + S H 2 + T r a n s a c : v a : o n + " C O O H +

Y 6 9 4 % %

S T A T 5 A + ( 7 9 4 + a a , + 9 4 + k D a ) + +

Y 6 9 9 % %

S T A T 5 B + ( 7 8 8 + a a , + 9 2 + k D a ) + +

Page 61: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

48

TTCNNNGA γ-interferon activation sequence (GAS) sites in the regulatory elements of the target

genes.88 The linker domain is a short sequence that stabilizes the multimeric proteinprotein/protein-

DNA complexes and connects the DBD to the Src-homology 2 (SH2) domain. In both STAT5A/B

isoforms, the SH2 domain of STAT5A/B is an significant site contributing to receptor recruitment

and protein dimerization by recognizing conserved phosphotyrosine (pY) residues. In the signaling

mechanism, STAT5A and STAT5B dimerization occurs in a reciprocal fashion, wherein the SH2

domain of one partner recognizes the phosphorylated tyrosine of the other.84,89 Finally, the C-

terminal transactivation domain (TAD) is highly specialized for STAT5A (aa 722- 794) and

STAT5B (aa 727-787), owing to the association with isoform-specific transcription co-regulators.

Furthermore, the TAD region is heavily involved in co-regulatory localization. Structural

divergence originates at the NTD and TAD, whereby the difference in residue sequence and

resulting tertiary structure may contribute to interaction with different transcription factors and

recognition of varying DNA sequences.84 Interestingly, there are no differences in the transcription

promoter elements recognized by both isoforms. Thus, any nonredundant functions of STAT5A

and STAT5B are a result of difference in tissue distribution or divergence in their TAD regions.90

In healthy cells undergoing normal division and differentiation, the signaling activity of

STAT5A/B is transiently activated by the JAK/STAT pathway and tightly regulated by multiple

negative regulatory mechanisms. The activation of the both STAT5 isoforms signaling is initiated

by binding of the cytokines or growth factors to their respective receptors. The interleukin family

(IL-2, IL-3, IL-5, IL-7, IL-12 and IL-15), interferons (IFNs), granulocyte-macrophage colony

stimulating factor (GM-CSF), erythropoietin (EPO) and thrombopoietin (TPO) have been reported

ligands that induce STAT5A/B signaling.91,92 Glycopeptides including Growth Hormone (GH),

prolactin, epidermal growth factor (EGF) and even insulin have been discovered as ligands that

regulate STAT5 activity upon binding to their extracellular receptors.84

Cytokine receptors are devoid of kinase activity themselves, and therefore are dependent

on the JAK (Janus kinase) family of tyrosine kinases for activation and subsequent protein target

phosphorylation. In mammals, four enzymes constitute the JAK family, JAK1, JAK2, JAK3, and

TYK2.93 These kinases are ubiquitously associated with cytokine receptors on the membrane

proximal domain and are activated on the receptor following ligand association. Other cellular

Page 62: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

49

tyrosine kinases have been associated with STAT5 signaling such as the Src kinase family and

Etk/Bmx (from the Tec family kinases). However, conclusive studies establishing their respective

signalling mechanisms remain elusive. Upon extra-cellular cytokine binding, the receptors undergo

homo- or hetero-dimerization and adopt an activated conformation. This conformational change

forces the receptor associated JAKs within close proximity to allow transphosphorylation,

producing a catalytically ‘active’ kinase. Next, active JAK kinases then catalyze phosphorylation

of conserved tyrosine residues on the intracellular domain of the receptor.16,6 Latent cytoplasmic

STAT5 monomers are recruited to these receptor pY motifs via their SH2 domains. Following

STAT5-receptor binding, JAK kinases phosphorylate specific tyrosine residues in the C-terminus

of STAT5A and STAT5B. In the case of STAT5A, the phosphoryl group is transferred to Y694,

whereas in STAT5B, the tyrosine acceptor is Y699. Phosphorylated STAT5A/B dissociate from

the receptor and undergo hetero- or homo-dimerization in the cytosol through reciprocal pY-SH2

recognition. Homo-dimerization of STAT5A/B isoform units is favoured over heterodimerization

of STAT5A and STAT5B. Active STAT5-dimer complex dimers are able to translocate into the

nucleus through the nuclear pore complex (NPC) in an energy dependent process. Importin

proteins in the nuclear membrane mediate transportation of phosphorylated STAT5 dimers across

the NPC by recognizing a specific nuclear localization signal (NLS) sequence on the STAT5A/B

TAD domains.94

Once in the nucleus, phosphorylated STAT5 dimers recognize specific oligonucleotide

GAS regions (TTCNNNGA) through the DBD and induce transcription of target genes. Numerous

operational transcription factors and chromosomal-associated proteins are involved in regulation

of gene expression and thus interact with intranuclear STAT5. The STAT5A/B transcription

machinery is responsible for increase in transcription levels of genes controlling cell proliferation

such as BCL-XL (B-cell lymphoma-extra large; member of apoptotic Bcl-2 proteins), pim-1

(protooncogene serine/threonine kinase Pim-1)80 and C-MYC (proto-oncogene promoting cyclin

production, downregulates Bcl-2).95 STAT5 activity also controls the transcription elements

responsible for inflammation and differentiation include OSM (encodes Oncostatin-M, an

interleukin) and c-fos (encodes Fos protein) respectively.95 STAT5 signaling has also been reported

to activate expression of the MCL-1 pro-survival gene.96

Page 63: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

50

The principle inhibitory mechanisms associated with the regulation of the JAK/STAT

pathway include protein tyrosine phosphatase (PTP) enzymes, suppressor of cytokine signalling

(SOCS) proteins and the protein inhibitors of activated STAT (PIAS) family of regulators.97 The

direct route of deactivation of activated STAT5 is conducted by PTP enzymes, which catalyze the

dephosphorylation of pY694 and pY699 in STAT5A and STAT5B, respectively. Known PTPs that

cleave the phosphate group of pSTAT5A/B are PTP1B98 and SH2 domain containing

phosphatases, SHP-1 and SHP-2.99,100 Protein tyrosine phosphatase non-receptor 2 (PTPN2 or TC-

PTP) is believed to carry out the dephosphorylation of STAT5A/B in the nucleus.101

PIAS proteins are known transcription co-regulators that inhibit transcription by either

binding to the transcription factor directly, disrupting binding with DNA or labelling such factors

for protein degradation.102 More specifically, PIAS proteins have been shown to conjugate SUMO

(small ubiquitin related modifier) to target transcription factors such as STAT5 with their E3 ligase

activity.103 Upon SUMO modification, STAT5A/B are labelled for degradation by the 26s

proteasome.

Finally, the SOCS family consists of CIS (cytokine-inducible SH2 domain protein) and

members SOCS1 – 7. Collectively, they modulate the JAK/STAT pathway via a negative feedback

loop. SOCS indirectly decreases the phosphorylation of STAT5 monomers by binding to the active

receptor associated JAKs, inhibiting the enzyme activity.104 Alternatively, CIS proteins deactivate

STAT5 signaling by interacting with the cytokine/growth-factor receptors, to inhibiting access to

the STAT5 binding site.105 These binding interactions are mediated through the SOCS SH2 domain

and pY residues on the target proteins. The overall signaling cascade of STAT5 is presented below,

figure 3.2.

Page 64: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

51

Figure 3.2- Brief schematic of normal JAK/STAT5 signaling with cytokine/growth factor

receptors dimerized and bound to activated JAK2. Latent STAT5 is recruited via the SH2-pY

interaction and undergoes phosphorylation. Dimerization in the cytosol is followed by nuclear

translocation and regulation of transcription of target genes. Negative regulators, SOCS, PIAS and

PTP are represented as well. (Image generated using ChemBioDraw 14)

3.3 Role of STAT5A and STAT5B in disease

Given STAT5’s key role as a transcription factor for genes that control crucial cell

processes such as proliferation, survival and apoptosis, the signaling cascade needs to be tightly

regulated to ensure normal cell function and morphology. An in vivo study conducted by Grimley

et al. revealed genetic knockout of STAT5A, STAT5B or both isoforms in mice did not

compromise viability. Both STAT5A and STAT5B murine knockout models showed deficiencies

in immune responses.106 In addition, STAT5A null adult mice (STAT5-/-) exhibited defects in

mammary gland development and lactogenesis.107 On the other hand, STAT5B -/- mice showed

reduced growth and other morphological defects characteristic of insufficient growth hormone

levels.108

Page 65: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

52

Contrastingly, constitutive STAT5 activity has been linked to several diseased states

including inflammation, auto-immunity and most importantly, tumorgenesis and metastasis. 109,110

The resulting phenotype is dependent on the specific tissue, cell types and even the isoform of the

STAT5 protein. The role of STAT5 in hematological malignancies such as CML and AML has

been extensively documented. In the CML myeloproliferative disease, the BCR-ABL fusion kinase

constitutively phosphorylates STAT5 even in the absence of cytokine or growth-inducing

stimuli.111 This leads to hyperactivity of STAT5 protein, whereby transcription of downstream

anti-apoptotic and pro-survival genes is elevated. On the other hand, in the case of AML which is

characterized by excessive immature white blood cells, mutations in receptor-associated JAK2 and

FLT3 (Fms-like tyrosine kinase 3) kinases lead to increased levels of active STAT5.112 These

kinase mutations that drive hematological cancers through constitutive activation of STAT5 do not

show any isoform specificity and are dependent only on the hematopoietic system development.

As previously mentioned, the non-redundant physiological functions of STAT5A and STAT5B are

attributed primarily due to difference in expression levels in varying tissues. Specific cell types

exhibit varying levels of expression of each isoform and thus maintain their required thresholds.

High levels of STAT5A are found in mammary tissue whereas STAT5B is more prevalent in the

muscle, prostate, and hepatic tissues. Thus, deregulations in the form of isoform specific point

mutations that contribute to hyperactivity of STAT5A and STAT5B individually can lead to

isoform-driven tumors. More specifically, hyperactivated STAT5A drives proliferation of human

mammary carcinoma cells, promoting tumor development and progression

113 Constitutively active STAT5B has been shown to increase in tumor volume in

breast cancer.

and invasion in squamous cell carcinoma of the head and neck114 (SCCHN) and colorectal

cancer.115 Other diseased phenotypes associated with sustained STAT5B signaling include growth

hormone insensitivity116, IL-2 based immunodeficiency and autoimmune disease.117 Mutations

within the STAT5A/B isoforms which themselves confer to its oncogenic activity, independent of

up-stream mutated kinases are also possible. Several in vitro studies have established

isoformspecific point mutations in STAT5A and STAT5B that render the proteins capable of

malignant transformation on their own.

Page 66: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

53

Kitamura and co-workers discovered a STAT5A mutant through a mutagenesis screen

identifying transformed Ba/F3 cell lines capable of IL-3 cytokine independent growth.118 They

identified the point mutation N642H in the SH2 domain, in close proximity to the pY-recognition

site. It is believed this particular mutation provides extra stability to the STAT5-dimer complex,

particularly at the pY-SH2 domain site, hence sustaining its signaling in the absence of IL-3.

Interestingly, prolonged pSTAT5 activity upon IL-3 stimulation was observed. The same workers

discovered another potential oncogenic STAT5A mutation using the same mutagenesis screen. The

amino acid substitution reported was S710F in the TAD domain. Expression of the STAT5AS710F

mutant in STAT5 null Ba/F3 cells led to transformation with increase in transcription levels of

STAT5 target genes inducing proliferation.119

Numerous STAT5B variants harbouring point mutations have been proposed to promote

leukemogenesis in T-lymphocytes in patients who relapsed from lymphocytic leukemias. The most

commonly occurring mutation observed in patient samples was substitution of Asn642 by His,

similar to STAT5A.120 In vitro, it was observed that leukemia cell populations expressing

STAT5BN642H protein were able to induce cytokine-independent survival with the STAT5B

transcriptional machinery retaining sustained activity.121

Given the growing body of evidence supporting the significance of the aberrant STAT5

signaling pathway in multiple diseases, with strong clinical emphasis on human malignancies, such

as CML and AML, therapeutic intervention is highly critical. In particular, effective inhibition of

the STAT5-mediated cascade can supress transcription of genes that promote immature

differentiation and uncontrolled proliferation ultimately leading to malignant growth and

maintenance. Furthermore, with studies revealing the oncogenic capacity of STAT5A and

STAT5B in several isoform specific malignancies and other hormone-related diseases, strategies

that can achieve isoform-selective inhibition would be of utmost importance as diagnostic tools or

potential therapeutics.

3.4 Therapeutic strategies towards the STAT5A/B signaling pathway

The majority of therapeutic approaches to combat hyperactive signal transduction have

revolved around the identification and optimization of small molecule inhibitors that target

Page 67: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

54

upstream kinases. Tyrosine kinase inhibitor (TKI) drug programs have yielded substantial success

in the treatment of multiple cancers by increasing survival rate, and prolonging patient remission.

The BCR-ABL TKI, imatinib has been successful in the clinic for CML treatment with high rates

of complete patient remission. Imatinib effectively induces apoptosis in leukemic cells harbouring

the BCR-ABL chromosomal rearrangement and decreases overall tumor burden. However, it was

later discovered that prolonged treatment of imatinib eventually resulted in acquired resistance

through point mutations in the BCR-ABL fusion kinase.122 The most frequently identified mutation

was the gatekeeper mutation, T315I. These amino acid substitutions occur in the BCR-ABL kinase

domain and compromise imatinib binding. It is also suggested that cells are capable of bypassing

inhibition of BCR-ABL kinase activity via a compensatory mechanism whereby other kinases

phosphorylate STAT5, thus maintaining constitutive expression of downstream pro-survival

genes.123 Increased dosing regimes become difficult to tolerate due to toxic side effects and the

multi-targeted nature of most TKIs. As a result, second-generation TKIs have been developed as a

means to address the acquired resistance observed in the clinic. Nilotinib and dasatinib were

second-line TKIs developed for imatinib resistant CML patients with higher potency than

firstgeneration ABL inhibitors, figure 3.3. Unfortunately, these small-molecules are not able to

diminish leukemic cell viability against all BCR-ABL mutant variants, thus limiting their use to a

small class of CML patients.124

Figure 3.3- First generation BCR-ABL TKI imatinib for CML treatment. Second-generation TKIs

from imatinib, nilotinib and dasatinib are shown below.

Page 68: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

55

Based on the success of BCR-ABL TKIs as effective CML therapeutics, a similar drug

development method was adapted for AML treatment wherein mutated FLT3 kinase activity results

in constitutive phosphorylation of STAT5A/B. Notable first generation FLT3 TKIs include

nanomolar inhibitors lestaurtinib and sorafenib, both with promising initial success in clinical

trials, figure 3.4.125 One common theme amongst FLT3 small molecule inhibitors is poor

pharmacodynamic and pharmacokinetic profiles. This serves as a major limitation to advancement

of potential candidates towards clinical application. Another concern is potential loss of sensitivity

to TKIs by FLT-3 kinase domain mutations.

Therapeutic intervention of the JAK2/STAT5 pathway has also been attempted in treatment

of myeloproliferative neoplasms where JAK2 reportedly harbours somatic mutations. The most

commonly occurring mutation is a B1859T base pair inversion, resulting in the V617F substitution.

This mutation is localized at the autoinhibitory domain of JAK2, and disrupts the modulation of

the catalytic activity of the kinase.126 The loss of the autoinhibitory function of JAK2 leads to

constitutive phosphorylation of STAT5A/B, inducing increased levels of antiapoptotic

transcription products. The JAK2 inhibitor ruxolitinib has faired quite successful in the clinic with

potent inhibition of JAK2V617F at therapeutic concentrations, figure 3.4. Similar to BCR-ABL and

FLT-3 TKIs, small molecule inhibitors of JAK2 lack specificity, given their reactivity against

multiple off-targets.127

Figure 3.4- First-generation FLT-3 TKI lestaurinib and following second-generation derivative

sorafenib. Both drugs are used for FLT-3 mutation positive AML. Ruxolitinib is the widely used

JAK2 inhibitor in AML treatment, given its efficacy against JAK2V617F

In summary, therapeutic strategies aimed at inhibition of STAT5 signaling have focused

on the discovery and optimization of inhibitors targeting upstream mutant kinases. Aberrant kinase

Page 69: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

56

activity, as seen in BCR-ABL, FLT-3 and JAK2-driven disease, leads to upregulation of cell

growth and survival via hyperactive STAT5A/B signal transduction. Although TKIs have had the

most success, they have several drawbacks, including adverse toxic side effects and acquired drug

resistance. Cardiac toxicity, hepatotoxicity, hemorrhaging, nausea, and immunodeficiency are a

few examples of side effects associated with TKI administration in patients. Generally, these

symptoms are a result of drug interaction with off-target proteins, a pronounced trend with TKIs,

given their multikinase nature.128,129 Acquired drug resistance is a multifaceted drawback

extensively covered in Chapter 1. Reduction in TKI binding affinity as a result of upstream kinase

mutations diminishes inhibitor efficacy, requiring increased dosing. Additionally, the cell can

resort to complimentary bypass mechanisms through other signaling cascades or kinases to

constitutively activate STAT5. Most importantly, since TKIs work against an upstream protein,

there is no possibility of inhibition of STAT5A/B isoform selective function. Reduction in the

hyperactive signaling of one specific isoform in particular diseases is not attainable with current

therapies. Furthermore, given that mutations in STAT5A and STAT5B can render the proteins as

the primary source of oncogenic transformation themselves, upstream TKIs will be ineffective in

suppressing tumor growth and survival. One strategy to resolve both shortcomings is the targeted

inhibition of STAT5A and STAT5B proteins directly.

To date, little progress has been made towards STAT5A/B isoform selective inhibitors.

Only Berg et al. have identified a small molecule capable of isoform selective inhibition of

STAT5A/B proteins.130 The workers reported Stafib-1, as the first small molecule which inhibits

the STAT5B SH2 domain, (Ki = 44 ± 1 nM determined by FP) with more than 50-fold selectivity

over STAT5A. Treatment of K562 CML cell lines with a prodrug analogue of Stafib-1 showed

dose-dependent decrease in pSTAT5B levels with minimal suppression of pSTAT5A across the

same range of concentrations. Despite the high affinity binding profile of Stafib-1 and its

derivatives, there is an opportunity to determine structural differences in the SH2 domains of

STAT5A and STAT5B that contribute to recognition of the isoform selective inhibitors.

STAT5A/B specific amino residues and their interaction with specific functional groups in the

proposed inhibitors is an interesting investigation in its own.

Page 70: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

57

3.5 Proposed isoform-selective peptidomimetic inhibitors of STAT5A/B

We decided to approach the challenge of designing direct STAT5 isoform selective

inhibitors with the use of phosphopeptides and subsequently derived peptidomimetics.

Peptidomimetic inhibitors represent an important field in medicinal chemistry and have had

monumental impact in the discovery and advancement of numerous drugs targeting protein-protein

interactions.131 Peptidomimetic design originates form the identification of native peptide

sequences that interact with the target protein. Then, an alanine scan mutagenesis study is usually

conducted wherein each amino acid in the inherent peptide is substituted for an alanine. By

comparing the difference in binding affinity of each alanine-derivative in the combinatorial library

with the original sequence, residues important for interaction with the target are identified.132 The

alanine scan screening allows a series of logical truncations and chemical modifications are

implemented to remove the pharmacokinetic liabilities of peptides while retaining the original

pharmacophore. Peptidomimetic inhibitors seek to incorporate drug-like properties into the

scaffold of a biologically active peptide with minimal loss in activity. Overall, a peptidomimetic

strategy encompassing the use of high affinity STAT5A/B phosphopeptides with subsequent

chemical modification to potent and isoform selective peptidomimetic inhibitors presents a novel

therapeutic route to suppress aberrant STAT5A/B signaling.

As previously mentioned, the SH2 domain is an important module as it mediates critical binding

events. Recruitment to cytokine and growth factor receptors and homo/heterodimerization

represent critical junctures in STAT5A/B signal transduction. Accordingly, both phosphorylation

of STAT5 monomers and formation of the dimeric species that regulates transcription depend on

the SH2 domain’s ability to recognize selective pY protein substrates. Given its role in recognizing

specific pY containing motifs in various cellular receptors through PPIs, the SH2 domain is an

ideal hotspot region in both STAT5A and STAT5B. There is an abundance of information

concerning the interaction of STAT5 with several phosphorylated cellular receptors, which can be

used to select the preliminary library of SH2 domain-binding phosphopeptides. McMurray et al.

have successfully employed this approach to inhibit the STAT3 SH2 domain, using a peptide

sequence derived from the STAT3-native cellular receptor, gp130.133 The SH2 assembly is the

largest class of domains dedicated to selective recognition of pY motifs with 111 proteins in the

Page 71: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

58

human proteome containing at least one SH2 domain.134 Usually consisting of 100 amino acids,

SH2 domains possess an evolutionary conserved tertiary structure that must recognize a pY residue

generally located on signal transduction factors participating in PPIs.135 They must derive

significant binding energy from the phosphorylated tyrosine residue so that binding is dictated by

the pY/Y (“active/inactive”) state. Furthermore, SH2 domains gain substrate specificity from the

residues flanking the pY site. The recognition of adjacent residues not only provides additional

energetically favourable contacts, but also dictates the selectivity of the SH2 domain-

phosphopeptide interaction. For example, studies investigating the Src and Lck SH2 domains have

indicated that although 50% of binding affinity is obtained from the phosphate group of the pY

residue, residues -2 to +4 relative to the phosphotyrosine contribute towards binding specificity

through maximizing their interactions with the surface of the SH2 domain.136 Additionally,

numerous co-crystal structures of SH2 domains bound to their respective ligands suggest a larger

contact surface is plausible.137 The traditional SH2 domain fold comprises of a central anti-parallel

β-sheet flanked by two α-helices providing two chemically unique environments on each side of

the β-sheet, figure 3.5.134 On one side, there is a positively charged pocket consisting of arginine,

serine, lysine/arginine and histidine that recognize the pY moiety. This group of amino acids is

highly invariable across all SH2 domains and participates in coordination of the oxygen atoms on

the phosphate functionality. On the other hand, the secondary pocket adjacent to the central β-sheet

provides an extended cleft for the interaction of residues Cterminal to the pY. We are particularly

interested in the molecular structure of this extended surface as it can provide a desirable degree

of selectivity for pY-containing peptide ligands.

Page 72: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

59

Figure 3.5- Traditional SH2 domain tertiary structure ribbon illustration. The central anti-parallel

β sheet is flanked by two α-helices. Interacting pY-containing peptide is shown in black, stick

figure representation, with pTyr, +1, +2 and +3 sites labeled. (figure from ref 60).

Despite sharing overall 93% sequence homology, STAT5A and STAT5B differ in 6 amino acids

in their respective SH2 domains. This lends further support of targeting the SH2 domain as this

specific interface of the STAT5A/B transcription factors has differences in molecular structure that

can govern binding affinity for phosphopeptide ligands and generate selectivity. Unfortunately,

only the STAT5A protein crystal structure has been reported.138 To conduct a thorough comparison

of the SH2 domains of both isoforms, we engineered a STAT5B protein 3D model using homology

modeling with STAT5A as the sequence template. The STAT5A crystal structure was retrieved

from the Research Collaboratory for Structural Bioinformatics Protein Data Bank (PDB ID code:

1Y1U) and the methodology described in chapter 2 used to generate the inactive ROS1 structure

was employed. Briefly, using the Prime 3.9 module from Schrödinger Suite, the STAT5A protein

sequence was used as a template and the STAT5B amino acid sequence was inserted. The differing

amino residues were replaced using a library of side chain residues and was subjected to energy

minimization using the OPLS_2005 force field. Finally, loop refinement and side chain

optimization generated the final STAT5B model. Upon careful examination, we observed the

conventional SH2 domain arrangement in both isoforms. We proposed that the positively charged

Page 73: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

60

pY recognition site encompasses residues Arg618, Lys600 and Ser622 allowing electrostatic and

hydrogen bond interactions with the phosphotyrosine group originating from either a receptor or a

STAT5 binding partner, see figure 3.6. The six variable residues between STAT5A and STAT5B

were in moderate neighbouring distance to the pY-binding cleft. Amino acids Pro636, Leu639 and

Asn640 occupying the α-helix flanking the antiparallel β-sheet in

STAT5A were replaced by Gln, Phe and Met respectively in STAT5B. Located on the terminal

βstrand, the residue within closest proximity to the critical Arg618 was Lys644 in STAT5A

whereas the identical position was occupied by Met in STAT5B as shown in figure 3.6. Thus, there

was some promising variability between the molecular architectures of the STAT5A and STAT5B

SH2 domains that could be utilized by different pY-containing peptide ligands. The remaining

substitution in STAT5A, for Asn in STAT5B at position 654 is situated out of the scope of the

pYbinding site.

B+

Figure 3.6- A) Overlap of STAT5A (orange) and STAT5B (blue) ribbon figures. The five differing

residues in the SH2 domain in the vicinity of the pY binding motif are represented by stick figures.

B) Shared Ser622, Arg618 and Lys600 in STAT5A/B SH2 domains. On comparison with

conventional SH2 domain structure, these three residues contribute to recognition of pY functional

group. (Image generated using MacPyMol 2009-2010)

With a hot-spot target in hand, we then explored receptor sequences known to bind to

STAT5. In the early stages of peptidomimetic design, it is always preferable to use native binding

peptides for the target given their naturally high binding affinity and selectivity.

P r o 6 3 6 + G l n 6 3 6 +

P h e 6 7 9 + T y r 6 7 9 +

L y s 6 4 4 + M e t 6 4 4 +

A s n 6 3 9 + M e t 6 3 9 +

L e u 6 4 0 + P h e 6 4 0 +

S T A T 5 A % S T A T 5 B %

A +

A r g 6 1 8 +

L y s 6 0 0 +

S e r 6 2 2 +

Page 74: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

61

Erythropoietin (Epo) is a cytokine whose signaling is necessary for the proliferation,

differentiation, and survival of erythrocyte precursor cells in erythropoiesis.139 Epo elicits its

cellular response by binding to the Epo receptor (EpoR), triggering receptor dimerization and

subsequent phosphorylation by receptor associated JAK2.140 STAT5 is then recruited to the

intracellular domain of EpoR, triggering the STAT5 signaling pathway. Silva et al. first reported

the survival function of Epo, through its induced expression of Bcl-xL. Upon Epo stimulation in

Ba/F3 cells transfected with EpoR, increased levels of Bcl-xL expression were concurrent with

activation of STAT5, along with expression of another STAT5 target gene, oncostatin M.141 In

1996, Klingmüller and colleagues verified the role of EpoR pY in STAT5 recruitment and were

able to deduce the exact site in the 96-amino acid intracellular domain. Through mutation studies

of all eight tyrosine residues in the intracellular domain of Epo, immunoblotting for pSTAT5

suggested that Y343 or Y401 are sufficient for maximal activation of STAT5 whereas Y429 and

Y431 can partially activate STAT5.142 A following investigation by Quelle and co-workers in 1996

confirmed the significance of residue Y343 in localization of STAT5 and its activation to EpoR.143

Cells transfected with Y343 truncated-EpoR could not induce STAT5 phosphorylation mediated

signaling. A competition gel-based assay showed that pY343 containing peptide successfully

prevents the binding of DNA:STAT5 by disrupting the formation of STAT5 dimers. In this

competition assay, the peptide was inclusive of residues flanking pY343 on the N- and C-terminus,

with an overall sequence: QDTpY343LVLDKWL. Berg et al. used this EpoR derived sequence to

design a fluorescence polarization assay for STAT5B.144 Through their method development, they

discovered that amino acids at the N-terminus of the pY should be omitted and the insertion of a

glycine linker between the pTyr and the fluorophore, 5-carboxyfluorescein (5-FAM) prevented any

interference with binding to STAT5B. Using the 5-FAM modified peptide as their fluorescent

probe, a Ki value of 0.21 ± 0.03 μM for the QDTpYLVLDKWL ligand was obtained. Thus, the

EpoR phosphopeptide is an appropriate candidate to screen for potency for STAT5A, followed by

a thorough combinatorial alanine scanning.

Another ligand known to activate the STAT5 signaling pathway is the

granulocytemacrophage colony-stimulating factor (GM-CSF). The GM-CSF glycoprotein was

first reported for its capacity to modulate the production of specific leukocytes including

granulocytes, macrophages, eosinophils and megakaryocytes in mice, through process termed as

Page 75: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

62

myelopoiesis.145 The sequencing of the human GM-CSF factor revealed its pivotal role in regulate

the survival, proliferation and differentiation of monocytes and macrophages.146 The GM-CSF

receptor (GM-CSFR), is a heterodimer, consisting of a ligand-binding α-subunit and the βc-subunit

which is a chain common to the cytokine receptors, IL-3 and IL-5. Upon GM-CSF binding, the

two subunits undergo dimerization, prompting the transphosphorylation of two βc-subunits via

receptor-associated JAK2.147 The GM-CSF induced activation of STAT5 is know to upregulate the

transcription of target gene elements related to inflammation and survival, pim-1 and

cytokineinducible SH2-containing protein (CIS).146 Numerous investigations in monocytes,

macrophages, and Ba/F3 cells transfected with GM-CSFR have pointed out STAT5 activation is a

key event in GM-CSF signaling.148 All evidence strongly suggests that pSTAT5 levels and

downstream transcriptional markers increase upon GM-CSF stimulation. In terms of a specific pY

containing sequence responsible for STAT5 recruitment, van Dijk et al. found Y612 of the βc-

subunit was independently capable of inducing activation of endogenous STAT5 upon GM-CSF

stimulation.149 While there were two other tyrosine residues reported for GM-CSF mediated

STAT5 activation150,

May and co-workers confirmed that Y612 of the βc-subunit, or Y882 of the entire GM-CSFR was

responsible for STAT5 recruitment using hybrid receptors.151 Their chimeric receptor construct

composed of the extracellular Epo domain and the gp130 transmembrane and intracellular regions.

The intracellular domain of gp130 presents a motif negative for STAT5 localization. Hence,

addition of various tyrosine modules to the gp130 fold can provide insight on their ability to recruit

STAT5, upon receptor stimulation. An EMSA readout was used to determine the extent of pSTAT

levels after stimulating COS-7 cells expressing chimeric receptors containing the QDYLSLP

sequence in the intracellular domain representative of the module QDpY882LSLP. We decided to

expand the phosphopeptide sequence for our studies to maintain consistency with EpoR with

regards to sequence length flanking the pY motif. Upon exploring the receptor sequence of the

GM-CSF, we obtained QQDpYLSLPPWE.

The STAT5 signaling cascade is known to transmit horomone-related responses to the

nucleus. Prolactin is a peptidic hormone that has been established as a ligand responsible for

STAT5 phosphorylation.152 Not only does prolactin balance lactogenesis, it also plays a profound

role in lymphocyte development and proliferation. Prolactin is believed to be involved in positively

effecting the uncontrolled division of immature granulocytes, activation of malignant B

Page 76: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

63

lymphocytes and lymphoma cells.153,154 In addition, several autoimmune diseased states are

believed to be caused by prolactin activity. Specific transcription products of prolactin-STAT5

signaling include c-myc in lymphoid tissue.155 Prolactin relays external stimuli to the

phosphorylation of STAT5 using the prolactin receptor (PRL-R) and receptor associated JAK2.

Prolactin ligand binding to PRL-R follows the exact same cascade mentioned for EpoR and

GMCSFR. Pezet and colleagues conducted selective mutation studies of all tyrosine groups in the

PRL-R intracellular domain. Immunoprecipitation studies with 293 fibroblasts expressing rat

PRLR revealed upon substitution of all tyrosines to phenylalanine residues, STAT5 activity was

completely lost. Upon narrowing down the mutation studies to individual tyrosine motifs, three

positions were determined as potential STAT5 docking sites: Y473, Y479 and Y580.156 However,

when comparing the rat PRL-R sequence to the human analog, only Y479 was conserved,

translating to Y509. Thus, we proceeded with this phosphotyrosine site and ensured that the amino

acids spanning from Y-3 to Y+7 were identical. Similar to their observation of GM-CSFR-STAT5,

May et al. observed similar activation of STAT5 with the Y509 module. The final phosphopeptide

taken forward was PLDpYVEIHKVN.

Finally, we looked into deriving a high affinity pY containing ligand from the IL-2 receptor.

Interleukin 2 (IL-2) is a signaling cytokine molecular utilized by the innate immune system to

promote the differentiation of immature T cells into helper T cells, memory T cells and suppressor

T cells. These lymphocytes also rely on cytokine signaling for proliferation, including IL-2. IL-2

uses transactivation of the JAK/STAT5 mechanism to induce C-MYC, BCL-2 and BCLXL gene

expression.157 The IL-2 receptor (IL-2R) consists of three distinct chains, IL-2Rα, IL-2Rβ and IL-

2Rγ. While IL-2Rα is not involved in signaling, IL-2 binding initiates the heterodimerization of

the β and γ chains.158 The IL-2Rγ subunit is shared amongst other IL receptors, whereas the IL-

2Rβ polypeptide is unique to IL-2 and IL-5 as it is associated with JAK1.159,160 Amongst relevant

tyrosine sites prone to phosphorylation, Y338 is required for activation of SHC-1 protein while

phosphorylation of Y392 and Y510 is known to activate STAT5. Lord et al. demonstrated complete

loss of STAT5 activation levels when all tyrosines except Y510 in IL-2Rβ were deleted in CTLL-

2 T cells and Ba/F3 transformed cells.160 Activated STAT5 levels and T cell proliferation were

comparable with the non-mutant receptor constructs, suggesting Y510 is the most potent activating

residue. The expression levels of downstream STAT5 target genes were also rescued upon using

Page 77: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

64

the Y510 containing IL-2Rβ receptor chain. Given the strong activity of Y510 in IL-2Rγ in the

phosphorylation of STAT5, this sequence was an ideal choice for our initial pY peptide screening.

The overall sequence considered was TDApYLSLQELQ.

3.6 Initial FP analysis of proposed phosphopeptides

The biologically active peptides known to recruit STAT5 upon external stimulation are

summarized below. We first decided to screen these pY peptides and their non-phosphorylated

counterparts against both STAT5A and STAT5B using FP. Berg and co-workers had initially

reported a high-throughput FP assay for STAT5B in 2008 and a complementary assay for STAT5A

in 2015 using the EpoR derived 5-FAM-GpYLVLDKW as their fluorescent reporter. We first

conducted saturation experiments with the fluorescent probe with both STAT5A and STAT5B to

determine KD values for both proteins.

Fluorescence assays were performed in black, flat bottomed, non-treated, 384-well plates

(Corning #3573) and FP measurements were taken with the Infinite M1000 machine (Tecan,

Crailsheim, Germany). The excitation wavelength was set at 475 nm and the emission wavelength

was observed at 525 nm, characteristic of 5-carboxyfluorescin. Measurements were taken with an

optimal gain, 50 flashes and a G-factor of 1. The buffer conditions for all assays were 10 mM

HEPES, 25 mM NaCl, 1mM EDTA, 2 mM dithiothreitol, pH 7.5 and the final DMSO

concentration in the wells was kept constant at 10%. DMSO was used to improve the solubility of

the inhibitor compounds in the buffer system, as well as the fluoresceinated-phosphopeptide. A

calibration curve for STAT5A and STAT5B protein was derived by incubating a 10 nM final

concentration of the EpoR derived fluorescent-phosphopeptide, 5-FAM-GpYLVLDKW, which is

known to bind STAT5A/B’s SH2 domains. Increasing concentrations of STAT5 protein were

titrated ranging from 0 to 2.5 µM. The raw data was normalized in excel and fitted with a 1:1

binding model using GraphPad Prism 6.0. Saturation curves are shown below, figures 3.7 and 3.8

for STAT5A and STAT5B respectively.

Page 78: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

65

Figure 3.7- Calibration curve for EpoR derived 5-FAM-GpYLVLDKW fluorescent probe with

STAT5A protein.

Figure 3.8- Calibration curve for EpoR derived 5-FAM-GpYLVLDKW fluorescent probe with

STAT5B protein.

The 5-FAM-EpoR probe showed saturation with both STAT5A and STAT5B as expected.

C a l i b r a ti o n C u r v e ST A T 5 B

Page 79: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

66

The KD values obtained for both proteins were comparable with reported literature values, KD =

123 nM cf. 130 nM for STAT5A and KD = 126 nM cf. 125 nM for STAT5B. Next, we conducted

competition experiments with all four phosphopeptides and their non-phosphorylated analogs

listed in Table 1. However, peptides were ordered from CanPeptide, and only the EpoR and GM

CCSFR peptides were synthesized and delivered. The non-phosphorylated negative controls of

EpoR and GM-CSFR and remaining peptide library will be synthesized using solid-phase peptide

synthesis (SPPS).

Table 3.1- Summary of pY-containing ligands and negative control analogs

Receptor pY Derived Sequence Non-phosphorylated analog

EpoR QDTpYLVLDKWL QDTYLVLDKWL

GM-CSFR QQDpYLSLPPWE QQDYLSLPPWE

PRL-R PLDpYVEIHKVN PLDpYVEIHKVN

IL-2Rβ TDApYLSLQELQ TDAYLSLQELQ

For the STAT5A and STAT5B competitive-binding fluorescence polarization assays, the

5-FAM-GpYLVLDKW peptide and wildtype STAT5A/B protein (purchased from SignalChem)

were first incubated for 20 minutes at room temperature. Inhibitors were titrated at concentrations

ranging from 1 nM – 500 µM and incubated for a further 15 minutes. The fluorescence polarization

measurements were then taken in triplicate (λex = 470 nm, λem = 525 nm). The final well

concentration of the fluoresceinated-phosphopeptide was 10 nM after addition of the inhibitor

component. Final concentrations of STAT5A and STAT5B proteins were 130 nM and 125 nM,

respectively. The resulting fluorescence polarization measurements were normalized and plotted

against inhibitor concentration. The raw data was fitted with a standard dose response inhibition

curve with four parameters using GraphPad Prism 6.0 software. The IC50 values were converted to

Ki values using Equation 1, the Nikolovska-Coleska equation.161

Equation 1

There were major discrepancies in the initial set of data obtained, particularly pertaining to the

experiments with STAT5A. Fluorescence polarization values observed were not consistent with

Page 80: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

67

any binding event under the given experiment conditions. The overall change in fluorescene

polarization (mP units) for the STAT5A experiments (~100 units) was much smaller relative to

STAT5B (~300 units). Upon retaking measurements after 1 minute and 5 minutes delay after the

initial measurement, the observed fluorescence anisotropy values drastically changed across all

concentrations ranging from no inhibitor to expected complete displacement of probe. We

attributed this to the quality or the stability of the STAT5A protein given the consistency of the

results from the experiments with STAT5B protein. Competition binding experiments for STAT5A

were then altered by removing the incubation times so that protein stays at 0 oC. This gave much

more reliable data with a larger range of fluorescence anisotropy (~150 units). The competition FP

curves for EpoR and GM-CSF derived peptides against STAT5A (0 oC, no incubation) and

STAT5B (room temperature) are shown below with Ki values summarized in Table 2.

Figure 3.9- Normalized FP inhibition curve for EpoR derived peptide, QDTpYLVLDKWL for

STAT5A. Ki = 522.3 nM

Ep o R ST A T 5 A

Log[ Ep o R i n µ M]

Page 81: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

68

Figure 3.10- Normalized FP inhibition curve for EpoR derived peptide, QDTpYLVLDKWL for

STAT5B. Ki = 426.0 nM

Figure 3.11- Normalized FP inhibition curve for GM-CSFR derived peptide, QQDpYLSLPPWE

for STAT5B. Ki = 876.6 nM

Ep o R ST A T 5 B

Log[ Ep o R i n µ M]

G M-C SF R ST A T 5 A

L o g [ G M-C SF R i n µ M]

Page 82: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

69

Figure 3.12- Normalized FP inhibition curve for GM-CSFR derived peptide, QQDpYLSLPPWE

for STAT5A. Ki = 652.4 nM

Table 3.2- Calculated Ki values for EpoR and GM-CSFR derived peptide ligands

pY Peptide STAT5A Ki STAT5B Ki

QDTpYLVLDKWL (EpoR) 522.3 nM 426.0 nM

QQDpYLSLPPWE (GM-CSFR) 876.6 nM 652.4 nM

Both EpoR and GM-CSFR derived phosphopeptides bind with much higher affinity to the

STAT5B SH2 domain in comparison to the STAT5A SH2 domain. However, given the observed

discrepancy in the FP experiments with the STAT5A protein and pending Ki values from other

ligands, there are no conclusive trends. The data also indicates incomplete displacement of the

fluorescent tag from the STAT5A SH2 domain for both EpoR and GM-CSFR derived peptides.

The measurements taken at the higher concentration of the inhibitor are not completely reliable

given the temporal instability of STAT5A and may consequently affect the calculated Ki values.

To circumvent this, we will possibly look to another source for STAT5A protein given its lack of

application for a high-throughput assay. Complete binding saturation of the fluorescent probe was

observed since there was no incubation time, but the range of anisotropy values are still not

consistent with those reported literature by Berg et al. The workers utilized a different plasmid

vector for expression of STAT5A and an N-terminal maltose protein binding tag. Upon obtaining

functional STAT5A protein, we will look to redo the calibration curves and inhibitor experiments.

G M-C SF R ST A T 5 B

L o g [ G M-C SF R i n µ M]

Page 83: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

70

3.7 Conclusion

The development of STAT5A/B isoform selective peptidomimetic inhibitors is still in its primitive

stages. We are attempting to narrow down the SH2 domain as the “hot-spot” of STAT5A/B and

examine the structural differences in this region between the two isoforms in silico. Through

exhaustive literature research, we deduced a novel peptidomimetic strategy involving

pYcontaining peptides that are known to recruit STAT5A/B transcription factors and cause their

eventual activation. Using FP as our initial assay to determine the binding affinities of the proposed

peptides, we will look to study the efficacy of each receptor-derived ligand. Important trends and

contribution of each amino acid in the phosphopeptide sequence will be confirmed with an alanine

scan mutagenesis study.

Page 84: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

71

Chapter 4 Conclusions and Future Directions

Chromosomal aberrations involving structural changes in the original genome lead to

dysregulation of affected genes or a hybrid genetic product. In the case of cellular receptor tyrosine

kinases, fusion genes often result in a kinase domain juxtapositioned to a foreign extracellular

receptor. ROS1 is an RTK that has been reported as a fusion partner with several receptors. ROS1

rearrangements are capable of enforcing oncogene addiction and numerous clinical studies have

highlighted ROS1 fusions as molecular drivers in numerous malignancies, most notably, NSCLC

(CD74-ROS1). Conventional therapy involves small-molecule TKIs that recognize the ATP site of

the enzyme. Despite initial success with the first-line therapeutic crizotinib, cases of acquired drug

resistance have been reported. Specifically, in ROS1 fusion positive NSCLC patients, a G2032R

point mutation in the ROS1 kinase domain reduces sensitivity to crizotinib and has warranted the

use of other TKIs. With the emergence of cabozantinib and foretinib as highly potent and selective

ROS1 and ROS1G2032 inhibitors, we were interested in the structural properties of the scaffolds that

contribute to their privileged resistance profiles. We designed in silico models of the

ROS1and ROS1G2032 kinase domains and conducted computational docking to study how

cabozantinib and foretinib interact with the relevant binding pockets. Our initial work suggests that

both TKIs occupy an unprecedented binding conformation with ROS1, with access to a

hydrophobic region adjacent to the ATP site. Furthermore, cabozantinib exhibited a much more

consistent binding profile with ROS1 over foretinib, prompting us to explore the effect of the major

functional groups that distinguish cabozantinib as a unique ROS1 inhibitor. We screened a large

library of cabozantinib derivatives encompassing iterative changes in the buried 4-fluorobenzene

appendage with numerous fluorine isosteres, 5 and 6-membered aromatic rings including

heteroatoms. We even probed the solvent-exposed 6,7-dimethoxyquinoline region of cabozantinib

with bulkier alkyl groups and bridged cyclic structures. With no initial trends, we focussed on the

4-fluorobenzene by preparing a preliminary library of cabozantinib analogues with various

substituents at the para position. Along with this, we synthesized another library of molecules in

which the 4-fluorobenzene group was kept constant and the 6,7-dimethoxyquinoline was modified

to investigate the entropic contributions to cabozantinib-ROS1 binding. The initial library was

found to be not as active as cabozantinib when screened for anti-proliferation activity against Ba/F3

cells transfected with CD74-ROS1 or CD74-ROS1G2032R. More interested in the inhibitor

Page 85: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

72

interaction with the ROS1 kinase domain, we decided to employ a FRET-based LanthaScreen

assay to derive KD values for our library. The assay results were in corroboration with the trends

observed from the cellular viability experiment. The KD value of cabozantinib was calculated to be

21.43 ± 1.642 nM with no marked improvement in ROS1 binding affinity for the synthesized

derivatives. Collectively, the results obtained from both in vitro experiments contradict the binding

affinities predicted by GLIDE in silico. With our continued efforts, we hope to study the structural

and resistance profile of cabozantinib for ROS1 and the clinically relevant ROS1G2032 mutant and

decipher how this unique TKI interrogates the kinase domain.

This thesis also covered the work towards the design of peptidomimetic inhibitors for a

transcription factor downstream of aberrant kinase activity. We focussed specifically on STAT5,

a signaling protein critical in the JAK/STAT pathway driving cell differentiation, growth and

apoptosis in lymphoid and hematopoietic systems. Considering the drawbacks of conventional TKI

chemotherapy, we explored the therapeutic potential of STAT5 with emphasis on the two isoforms,

STAT5A and STAT5B. The pY-recognizing SH2 domain is the major hot-spot for

STAT5’s protein-protein interactions. We designed an in silico model to investigate the differing

residues in the STAT5A and STAT5B SH2 domains and concluded that peptidomimetic based

inhibitors developed from the critical pY would be an effective strategy for isoform-selective

inhibition. Native receptor sequences that interact strongly with STAT5A/B’s SH2 domains were

deduced and taken forward for competitive binding experiments against STAT5A and STA5B. We

performed FP experiments with each receptor-derived sequence to determine Ki values that would

allow us to elucidate any trends or critical residues amongst the peptide ligands. The initial set of

inhibitor competition experiments revealed the instability of STAT5A and other sources the protein

are being investigated. Finally, we will look to examine the contribution of each amino acid in the

most potent pY-peptide through alanine mutagenesis combinatorial screening. Selective

truncations and eventual ‘drug-like’ modifications will be imposed after the alanine scan. We hope

to develop potent and isoform selective STAT5A/B peptidomimetic inhibitors for use in the study

and potential treatment of isoform specific STAT5 associated diseases.

Page 86: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

73

References

(1) Moore, C.M.; Best, R.G. Chromosomal Genetic Disease: Structural Aberrations.

Encyclopedia of Life Sciences; Wiley: New York, 2001; p1–8.

(2) Genetic Alliance, District of Columbia Department of Health. Understanding Genetics;

A District of Columbia Guide for Patients and Health Professionals: Washington (DC),

2010, pp 1–100.

(3) Mitelman, F.; Johansson, B.; Mertens, F. The Impact of Translocations and Gene Fusions

on Cancer Causation. Nature Reviews Cancer. 2007, 7, 233–245.

(4) Robinson, D.; Wu, Y.; Lin, S. The protein tyrosine kinase family of the human

genome.Oncogene, 2000, 19, 5548-5557.

(5) Solomon, B. Validating ROS1 Rearrangements as a Therapeutic Target in Non-SmallCell

Lung Cancer. Journal of Clinical Oncology. 2015, 33, 972–974.

(6) Davies, K. D.; Doebele, R. C. Molecular Pathways: ROS1 Fusion Proteins in Cancer.

Clinical Cancer Research. 2013, 19, 4040–4045.

(7) Ou, S.; Tan, J.; Yen, Y.; Soo, R. ROS1 as a ‘druggable’ receptor tyrosine kinase: lessons

learned from inhibiting the ALK pathway. Expert Review of Anticancer Therapy. 2012,

12, 447-456.

(8) Davare, M. A.; Vellore, N. A.; Wagner, J. P.; Eide, C. A.; Goodman, J. R.; Drilon, A.;

Deininger, M. W.; O’Hare, T.; Druker, B. J. Structural Insight Into Selectivity and

Resistance Profiles of ROS1 Tyrosine Kinase Inhibitors. Proceedings of the National

Academy of Sciences. 2015, 112, 5381–5390.

(9) Zwick, E. Receptor tyrosine kinase signalling as a target for cancer intervention strategies.

Endocrine Related Cancer. 2001, 8, 161-173.

(10) Acquaviva, J.; Wong, R.; Charest, Al. The Multifaceted Roles of the Receptor Tyrosine

Kinase ROS in Development and Cancer. Biochimica et Biophysica Acta. 2009, 1795,

37– 52.

(11) Hubbard, S.; Till, J. Protein Tyrosine Kinase Structure and Function. Annu. Rev. Biochem.

2000, 69, 373-398.

(12) Sonnenberg-Riethmacher, E.; Walter, B.; Riethmacher, D.; Godecke, S.; Birchmeier, C.

The c-ros tyrosine kinase receptor controls regionalization and differentiation of epithelial

cells in the epididymis. Genes & Development. 1996, 10, 1184-1193.

(13) Charest, A.; Kheifets, V.; Park, J.; Lane, K.; McMahon, K.; Nutt, C.; Housman, D.

Oncogenic targeting of an activated tyrosine kinase to the Golgi apparatus in a

glioblastoma. Proceedings of the National Academy of Sciences. 2003, 100, 916-921.

(14) Ellis, L.; Morgan, D.; Jong, S.; Wang, L.; Roth, R.; Rutter, W. Heterologous

transmembrane signaling by a human insulin receptor-v-ros hybrid in Chinese hamster

ovary cells. Proceedings of the National Academy of Sciences. 1987, 84, 5101-5105.

(15) Xiong, Q.; Chan, J.; Zong, C.; Wang, L. Two chimeric receptors of epidermal growth

factor receptor and c-Ros that differ in their transmembrane domains have opposite effects

on cell growth. Molecular and Cellular Biology. 1996, 16, 1509-1518.

(16) Sachs, M. Motogenic and morphogenic activity of epithelial receptor tyrosine kinases.

The Journal of Cell Biology, 1996, 133, 1095-1107.

Page 87: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

74

(17) Davies, K. D.; Le, A. T.; Theodoro, M. F.; Skokan, M. C.; Aisner, D. L.; Berge, E. M.;

Terracciano, L. M.; Cappuzzo, F.; Incarbone, M.; Roncalli, M.; Alloisio, M.; Santoro, A.;

Camidge, D. R.; Varella-Garcia, M.; Doebele, R. C. Identifying and Targeting ROS1 Gene

Fusions in Non-Small Cell Lung Cancer. Clinical Cancer Research. 2012, 18, 4570–4579.

(18) Gu, T.-L.; Deng, X.; Huang, F.; Tucker, M.; Crosby, K.; Rimkunas, V.; Wang, Y.; Deng,

G.; Zhu, L.; Tan, Z.; Hu, Y.; Wu, C.; Nardone, J.; MacNeill, J.; Ren, J.; Reeves, C.;

Innocenti, G.; Norris, B.; Yuan, J.; Yu, J.; Haack, H.; Shen, B.; Peng, C.; Li, H.; Zhou, X.;

Liu, X.; Rush, J.; Comb, M. J. Survey of Tyrosine Kinase Signaling Reveals ROS Kinase

Fusions in Human Cholangiocarcinoma. PLoS ONE. 2011, 6,15640–15649.

(19) Charest, A. ROS Fusion Tyrosine Kinase Activates a SH2 Domain-Containing

Phosphatase-2/Phosphatidylinositol 3-Kinase/Mammalian Target of Rapamycin

Signaling Axis to Form Glioblastoma in Mice. Cancer Research. 2006, 66, 7473–7481.

(20) Shi, Z.; Yu, D.; Park, M.; Marshall, M.; Feng, G. Molecular Mechanism for the Shp-2

Tyrosine Phosphatase Function in Promoting Growth Factor Stimulation of Erk

Activity. Molecular and Cellular Biology. 2000, 20, 1526-1536.

(21) Rikova, K.; Guo, A.; Zeng, Q.; Possemato, A.; Yu, J.; Haack, H.; Nardone, J.; Lee, K.;

Reeves, C.; Li, Y.; Hu, Y.; Tan, Z.; Stokes, M.; Sullivan, L.; Mitchell, J.; Wetzel, R.;

MacNeill, J.; Ren, J. M.; Yuan, J.; Bakalarski, C. E.; Villen, J.; Kornhauser, J. M.; Smith,

B.; Li, D.; Zhou, X.; Gygi, S. P.; Gu, T.-L.; Polakiewicz, R. D.; Rush, J.; Comb, M. J.

Global Survey of Phosphotyrosine Signaling Identifies Oncogenic Kinases in Lung

Cancer. Cell. 2007, 131, 1190–1203.

(22) Nguyen, K. T.; Zong, C. S.; Uttamsingh, S.; Sachdev, P.; Bhanot, M.; Le, M. T.; Chan, J.

L. K.; Wang, L. H. The Role of Phosphatidylinositol 3-Kinase, Rho Family GTPases, and

STAT3 in Ros-Induced Cell Transformation. Journal of Biological Chemistry. 2002, 277,

11107–11115.

(23) Birchmeier, C.; Sharma, S.; Wigler, M. Expression and rearrangement of the ROS1 gene

in human glioblastoma cells. Proceedings of the National Academy of Sciences.

1987, 84, 9270-9274.

(24) Calandra, T.; Roger, T. Macrophage Migration Inhibitory Factor: a Regulator of Innate

Immunity. Nature Reviews Immunology. 2003, 3, 791–800.

(25) Bergethon, K.; Shaw, A. T.; Ignatius Ou, S. H.; Katayama, R.; Lovly, C. M.; McDonald,

N. T.; Massion, P. P.; Siwak-Tapp, C.; Gonzalez, A.; Fang, R.; Mark, E. J.; Batten, J. M.;

Chen, H.; Wilner, K. D.; Kwak, E. L.; Clark, J. W.; Carbone, D. P.; Ji, H.; Engelman, J.

A.; Mino-Kenudson, M.; Pao, W.; Iafrate, A. J. ROS1 Rearrangements Define a Unique

Molecular Class of Lung Cancers. Journal of Clinical Oncology. 2012, 30, 863–870.

(26) Song, A.; Kim, T. M.; Kim, D. W.; Kim, S.; Keam, B.; Lee, S. H.; Heo, D. S. Molecular

Changes Associated with Acquired Resistance to Crizotinib in ROS1-Rearranged Non-

Small Cell Lung Cancer. Clinical Cancer Research. 2015, 21, 2379–2387.

(27) Soda, M.; Togashi, Y.; Suzuki, R.; Sakata, S.; Hatano, S.; Asaka, R.; Hamanaka, W.;

Ninomiya, H.; Uehara, H.; Choi, Y. L.; Satoh, Y.; Okumura, S.; Nakagawa, K.; Mano,

H.; Ishikawa, Y.; Takeuchi, K. RET, ROS1 and ALK Fusions in Lung Cancer. Nature

Medicine. 2012, 18, 378–381.

Page 88: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

75

(28) Cai, W.; Li, X.; Su, C.; Fan, L.; Zheng, L.; Fei, K.; Zhou, C.; Manegold, C.;

SchmidBindert, G. ROS1 Fusions in Chinese Patients with Non-Small-Cell Lung Cancer.

Annals of Oncology. 2013, 24, 1822–1827.

(29) Shaw, A. T.; Ou, S.-H. I.; Bang, Y.-J.; Camidge, D. R.; Solomon, B. J.; Salgia, R.; Riely,

G. J.; Varella-Garcia, M.; Shapiro, G. I.; Costa, D. B.; Doebele, R. C.; Le, L. P.; Zheng,

Z.; Tan, W.; Stephenson, P.; Shreeve, S. M.; Tye, L. M.; Christensen, J. G.; Wilner, K.

D.; Clark, J. W.; Iafrate, A. J. Crizotinib in ROS1-Rearranged Non–Small-Cell Lung

Cancer. New England Journal of Medicine. 2014, 371, 1963–1971.

(30) Gainor, J.; Shaw, A. Novel Targets in Non-Small Cell Lung Cancer: ROS1 and RET

Fusions. The Oncologist. 2013, 18, 865-875.

(31) Weinstein, I. B.; Joe, A.; Felsher, D. Oncogene Addiction. Cancer Research. 2008, 69,

pp 3077–3080.

(32) Torti, D.; Trusolino, L. Oncogene Addiction as a Foundational Rationale for Targeted

Anti-Cancer Therapy: Promises and Perils. EMBO Molecular Medicine. 2011, 3, 623–

636.

(33) Sharma, S. V.; Settleman, J. Oncogene Addiction: Setting the Stage for Molecularly

Targeted Cancer Therapy. Genes & Development. 2007, 21, 3214–3231.

(34) Madhusudan, S.; Ganesan, T. S. Tyrosine Kinase Inhibitors in Cancer Therapy. Clinical

Biochemistry. 2004, 21, 618–635.

(35) Al-Obeidi, F.; Lam, K. Development of inhibitors for protein tyrosine kinases. Oncogene.

2000,19, 5690-5701.

(36) Arora, A.; Scholar, E.M. Role of Tyrosine Kinase Inhibitors in Cancer Therapy. Journal

of Pharmacology and Experimental Therapeutics. 2005, 315, 971–979.

(37) McDermott, U.; Iafrate, A. J.; Gray, N. S.; Shioda, T.; Classon, M.; Maheswaran, S.;

Zhou, W.; Choi, H. G.; Smith, S. L.; Dowell, L.; Ulkus, L. E.; Kuhlmann, G.; Greninger,

P.; Christensen, J. G.; Haber, D. A.; Settleman, J. Genomic Alterations of Anaplastic

Lymphoma Kinase May Sensitize Tumors to Anaplastic Lymphoma Kinase Inhibitors.

Cancer Research. 2008, 68, 3389–3395.

(38) Yasuda, H.; de Figueiredo-Pontes, L. L.; Kobayashi, S.; Costa, D. B. Preclinical Rationale

for Use of the Clinically Available Multitargeted Tyrosine Kinase Inhibitor Crizotinib in

ROS1-Translocated Lung Cancer. Journal of Thoracic Oncology. 2012, 7, 1086–1090.

(39) Awad, M. M.; Katayama, R.; McTigue, M.; Liu, W.; Deng, Y.-L.; Brooun, A.; Friboulet,

L.; Huang, D.; Falk, M. D.; Timofeevski, S.; Wilner, K. D.; Lockerman, E. L.; Khan, T.

M.; Mahmood, S.; Gainor, J. F.; Digumarthy, S. R.; Stone, J. R.; Mino-Kenudson, M.;

Christensen, J. G.; Iafrate, A. J.; Engelman, J. A.; Shaw, A. T. Acquired Resistance to

Crizotinib From a Mutation in CD74– ROS1. New England Journal of Medicine. 2013,

25, 2395–2401.

(40) Sierra, J. R.; Cepero, V.; Giordano, S. Molecular Mechanisms of Acquired Resistance to

Tyrosine Kinase Targeted Therapy. Molecular Cancer. 2010, 9, 75–88.

(41) Camidge, D. R.; Pao, W.; Sequist, L. V. Acquired Resistance to TKIs in Solid Tumours:

Learning From Lung Cancer. Nature July 1, 2014, 11, 473–481.

(42) Niederst, M.; Engelman, J. Bypass Mechanisms of Resistance to Receptor Tyrosine

Kinase Inhibition in Lung Cancer. Science Signaling. 2013, 6, 1–7.

Page 89: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

76

(43) Gainor, J. F.; Shaw, A. T. Emerging Paradigms in the Development of Resistance to

Tyrosine Kinase Inhibitors in Lung Cancer. Journal of Clinical Oncology. 2013, 31,

3987–3996.

(44) Davare, M.; Saborowski, A.; Eide, C.; Tognon, C.; Smith, R.; Elferich, J.; Agarwal, A.;

Tyner, J.; Shinde, U.; Lowe, S; Druker, B. Foretinib is a potent inhibitor of oncogenic

ROS1 fusion proteins. Proceedings of the National Academy of Sciences. 2013, 110,

19519-19524.

(45) Katayama, R.; Kobayashi, Y.; Friboulet, L.; Lockerman, E. L.; Koike, S.; Shaw, A. T.;

Engelman, J. A.; Fujita, N. Cabozantinib Overcomes Crizotinib Resistance in ROS1

Fusion-Positive Cancer. Clinical Cancer Research. 2015, 21, 166–174.

(46) Yakes, F. M.; Chen, J.; Tan, J.; Yamaguchi, K.; Shi, Y.; Yu, P.; Qian, F.; Chu, F.;

Bentzien, F.; Cancilla, B.; Orf, J.; You, A.; Laird, A. D.; Engst, S.; Lee, L.; Lesch, J.;

Chou, Y. C.; Joly, A. H. Cabozantinib (XL184), a Novel MET and VEGFR2 Inhibitor,

Simultaneously Suppresses Metastasis, Angiogenesis, and Tumor Growth. Molecular

Cancer Therapeutics. 2011, 10, 2298–2308.

(47) Kurzrock, R.; Sherman, S. I.; Ball, D. W.; Forastiere, A. A.; Cohen, R. B.; Mehra, R.;

Pfister, D. G.; Cohen, E. E. W.; Janisch, L.; Nauling, F.; Hong, D. S.; Ng, C. S.; Ye, L.;

Gagel, R. F.; Frye, J.; Muller, T.; Ratain, M. J.; Salgia, R. Activity of XL184

(Cabozantinib), an Oral Tyrosine Kinase Inhibitor, in Patients with Medullary Thyroid

Cancer. Journal of Clinical Oncology. 2011, 29, 2660–2666.

(48) Durante, C.; Russo, D.; Verrienti, A.; Filetti, S. XL184 (Cabozantinib) for Medullary

Thyroid Carcinoma. Expert Opinion on Investigational Drugs. 2011, 20, 407–413.

(49) Jacobson, M. P.; Pincus, D. L.; Rapp, C. S.; Day, T. J. F.; Honig, B.; Shaw, D. E.; Friesner,

R. A. A Hierarchical Approach to All-Atom Protein Loop Prediction. Proteins: Structure,

Function, and Bioinformatics. 2004, 55, 351–367.

(50) Liu, Y.; Gray, N. S. Rational Design of Inhibitors That Bind to Inactive Kinase

Conformations. Nature Chemical Biology. 2006, 2, 358–364.

(51) Zhang, J.; Yang, P. L.; Gray, N. S. Targeting Cancer with Small Molecule Kinase

Inhibitors. Nature Reviews Cancer. 2009, 9, 28–39.

(52) Shelley, J. C.; Cholleti, A.; Frye, L. L.; Greenwood, J. R.; Timlin, M. R.; Uchimaya, M.

Epik: a Software Program for pK a Prediction and Protonation State Generation for Drug-

Like Molecules. Journal of Computer-Aided Molecular Design. 2007, 21, 681–691.

(53) Friesner, R. A.; Banks, J. L.; Murphy, R. B.; Halgren, T. A.; Klicic, J. J.; Mainz, D. T.;

Repasky, M. P.; Knoll, E. H.; Shelley, M.; Perry, J. K.; Shaw, D. E.; Francis, P.; Shenkin,

P. S. Glide: a New Approach for Rapid, Accurate Docking and Scoring. 1. Method and

Assessment of Docking Accuracy. Journal of Medicinal Chemistry. 2004, 47, 1739–1749.

(54) Morris, G.; Goodsell, D.; Halliday, R.; Huey, R.; Hart, W.; Belew, R.; Olson, A.

Automated docking using a Lamarckian genetic algorithm and an empirical binding free

energy function. J. Comput. Chem. 1998, 19, 1639-1662.

(55) Friesner, R. A.; Murphy, R. B.; Repasky, M. P.; Frye, L. L.; Greenwood, J. R.; Halgren,

T. A.; Sanschagrin, P. C.; Mainz, D. T. Extra Precision Glide: Docking and Scoring

Incorporating a Model of Hydrophobic Enclosure for Protein−Ligand Complexes.

Journal of Medicinal Chemistry. 2006, 49, 6177–6196.

Page 90: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

77

(56) Morris, G. M.; Huey, R.; Lindstrom, W.; Sanner, M. F.; Belew, R. K.; Goodsell, D. S.;

Olson, A. J. AutoDock4 and AutoDockTools4: Automated Docking with Selective

Receptor Flexibility. Journal of Computational Chemistry. 2009, 30, 2785–2791.

(57) Exelixis Inc. C-Met modulators and methods of use. US 8,067,436B2, Nov 29, 2011.

(58) Ma, H.; Deacon, S.; Horiuchi, K. The Challenge of Selecting Protein Kinase Assays for

Lead Discovery Optimization. Expert Opinion on Drug Discovery. 2008, 3, 607–621.

(59) Glickman, J. Assay Development for Protein Kinase Enzymes. Eli Lilly & Company and

the National Center for Advancing Translational Sciences 2012.

(60) Li, Y.; Xie, W.; Fang, G. Fluorescence Detection Techniques for Protein Kinase Assay.

Analytical and Bioanalytical Chemistry. 2008, 390, 2049–2057.

(61) Berezin, M.Y.; Achilefu, S. Fluorescence Lifetime Measurements and Biological

Imagning. Chemical Reviews. 2010, 110, 2641-2684.

(62) Yan, Y. Analysis of Protein Interactions Using Fluorescence Technologies. Current

Opinion in Chemical Biology. 2003, 7, 635–640.

(63) Degorce, F.; Card, A.; Soh, S.; Trinquet, E.; Knapik, G.P.; Xie, B. HTRF: A

Technology Tailored for Drug Discovery - A Review of Theoretical Aspects and Recent

Applications. Current Chemical Genomics. 2009, 3, 22-32.

(64) Handl, H. L.; Gillies, R. J. Lanthanide-Based Luminescent Assays for Ligand-Receptor

Interactions. Life Sciences. 2005, 77, 361–371.

(65) Lebakken, C.; Riddle, S.; Singh, U.; Frazee, W.; Eliason, H.; Gao, Y.; Reichling, L.;

Marks, B.; Vogel, K. Development and Applications of a Broad-Coverage, TR-

FRETBased Kinase Binding Assay Platform. Journal of Biomolecular Screening. 2009,

14, 924-935.

(66) Wang, Z.X. An exact mathematical expression for describing competitive binding of two

different ligands to a protein molecule. FEBS Letters. 1994, 360, 111-114.

(67) Nash, P.; Pawson, T. Protein–protein interactions define specificity in signal transduction.

Genes and Development. 2000, 14, 1027-1047.

(67) Song, J.I.; Grandis, J.R. STAT signaling in head and neck cancer. Oncogene. 19,

24892495.

(68) Dutta, P.; Li, W.X. Role of the JAKSTAT Signalling Pathway in Cancer. John Wiley &

Sons, Ltd: Chichester, 2013.

(69) O'Shea, J. J., Schwartz, D. M., Villarino, A. V., Gadina, M., McInnes, I. B., and Laurence,

A. The JAK-STAT Pathway: Impact on Human Disease and Therapeutic

Intervention. Annu. Rev. Med. 2015, 66, 311–328.

(70) Paukku, K., Silvennoinen, O. STATs as critical mediators of signal transduction and

transcription: lessons learned from STAT5. Cytokine & Growth Factor Reviews, 2004,

15, 435–455.

(71) Nosaka, T.; Kawashima, T.; Misawa, K.; Ikuta, K.; Mui, A.L.F.; Kitamura, T. STAT5 As

A Molecular Regulator Of Proliferation, Differentiation And Apoptosis In

Hematopoietic Cells. The EMBO Journal,1999, 19, 4754-4765.

(72) Schmitt-Ney, M.; Doppler, W.; Ball, R.; Groner, B. Beta-casein gene promoter activity

is regulated by the hormone-mediated relief of transcriptional repression and a mammary-

gland-specific nuclear factor. Molecular and Cellular Biology, 1991, 11, 3745-3755.

Page 91: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

78

(73) Gouilleux, F.; Wakao, H.; Mundt, Maren, Groner, B. Prolactin induces phosphorylation

of Tyr694 of Stat5 (MGF), a prerequisite for DNA binding and induction of transcription.

The EMBO Journal. 1994, 13, 4361-4369.

(74) Liu, X.; Robinson, G.; Gouilleux, F.; Groner, B.; Hennighausen, L. Cloning and

expression of Stat5 and an additional homologue (Stat5b) involved in prolactin signal

transduction in mouse mammary tissue. Proceedings of the National Academy of Sciences

1995, 92, 8831-8835.

(75) Tan, S. H.; Nevalainen, M. T. Signal transducer and activator of transcription 5A/B in

prostate and breast cancers. Endocrine Related Cancer, 2008, 15, 367–390.

(76) Strehlow, I.; Schindler, C. Amino-terminal Signal Transducer and Activator of

Transcription (STAT) Domains Regulate Nuclear Translocation and STAT Deactivation.

Journal of Biological Chemistry. 1998, 273, 28049-28056.

(77) Gewinner, C. The Coactivator of Transcription CREB-binding Protein Interacts

Preferentially with the Glycosylated Form of Stat5. Journal of Biological Chemistry.

2003, 279, 3563-3572.

(78) Dagvadorj, A.; Kirken, R. A.; Leiby, B.; Karras, J.; Nevalainen, M. T. Transcription

Factor Signal Transducer and Activator of Transcription 5 Promotes Growth of Human

Prostate Cancer Cells In vivo. Clinical Cancer Research, 2008, 14, 1317–1324.

(79) Soldaini, E.; John, S.; Moro, S.; Bollenbacher, J.; Schindler, U.; Leonard, W. DNA

Binding Site Selection of Dimeric and Tetrameric Stat5 Proteins Reveals a Large

Repertoire of Divergent Tetrameric Stat5a Binding Sites. Molecular and Cellular

Biology, 2000, 20, 389-401.

(80) Becker, S.; Groner, B.; Müller, C. Three-dimensional structure of the Stat3β homodimer

bound to DNA. Nature, 1998, 394, 145-151.

(81) Ehret, G.; Reichenbach, P.; Schindler, U.; Horvath, C.; Fritz, S.; Nabholz, M.; Bucher, P.

DNA Binding Specificity of Different STAT Proteins. Comparision of in vitro specificity

of natural binding sites. Journal of Biological Chemistry, 2000, 276, 66756688.

(82) Darnell, J.; Kerr, I.; Stark, G. Jak-STAT pathways and transcriptional activation in

response to IFNs and other extracellular signaling proteins. Science, 1994, 264,

14151421.

(83) Johnston, J.; Bacon, C.; Finbloom, D.; Rees, R.; Kaplan, D.; Shibuya, K.; Ortaldo, J.;

Gupta, S.; Chen, Y.; Giri, J. Tyrosine phosphorylation and activation of STAT5, STAT3,

and Janus kinases by interleukins 2 and 15. Proceedings of the National

Academy of Sciences, 1995, 92, 8705-8709.

(84) Paukku, K.; Silvennoinen, O. STATs as critical mediators of signal transduction and

transcription: lessons learned from STAT5. Cytokine & Growth Factor Reviews, 2004,

15, 435–455.

(85) Reich, N.; Liu, L. Tracking STAT nuclear traffic. Nature Reviews Immunology, 2006, 6,

602-612.

(86) Basham, B.; Sathe, M.; Grein, J.; McClanahan, T.; D'Andrea, A.; Lees, E.; Rascle, A. In

vivo identification of novel STAT5 target genes. Nucleic Acids Research 2008, 36, 3802-

3818.

(87) Malin, S.; McManus, S.; Busslinger, M. STAT5 in B cell development and leukemia.

Current Opinion in Immunology, 2010, 22, 168–176.

Page 92: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

79

(88) Chen, W.; Daines, M. O.; Khurana H.G.K. Turning off signal transducer and activator of

transcription (STAT): The negative regulation of STAT signaling. Journal of Allergy and

Clinical Immunology, 2004, 114, 476–489.

(89) Johnson, K. J.; Peck, A.R.; Liu, C.; Tran, T.H.; Utama, F.E.; Sjolund, A. B.; Schaber, J.

D.; Witkiewicz, A. K.; Rui, H. PTP1B Suppresses Prolactin Activation of Stat5 in Breast

Cancer Cells. The American Journal of Pathology, 2010, 177, 2971–2983.

(90) Chen, Y.; Wen, R.; Yang, S.; Schuman, J.; Zhang, E.E.; Yi, T.; Feng, G.-S.; Wang, D.

Identification of Shp-2 as a Stat5A Phosphatase. Journal of Biological Chemistry, 2003,

278, 16520–16527.

(91) Yu, C.; Jin, Y.; Burakoff, S. Cytosolic Tyrosine Dephosphorylation of STAT5:

POTENTIAL ROLE OF SHP-2 IN STAT5 REGULATION. Journal of Biological

Chemistry, 2000, 275, 599-604.

(92) Aoki, N.; Matsuda, T. A Nuclear Protein Tyrosine Phosphatase TC-PTP Is a Potential

Negative Regulator of the PRL-Mediated Signaling Pathway: Dephosphorylation and

Deactivation of Signal Transducer and Activator of Transcription 5a and 5b by TC-PTP

in Nucleus. Molecular Endocrinology, 2002, 16, 58-69.

(93) Chung, C. Specific Inhibition of Stat3 Signal Transduction by PIAS3. Science, 1997, 278,

1803-1805.

(94) Palvimo, J. PIAS proteins as regulators of small ubiquitin-related modifier (SUMO)

modifications and transcription. Biochemical Society Transactions, 2007, 35, 14051408.

(95) Alexander, W.S., Hilton, D.J. The Role of Suppressors of Cytokine S ignaling (SOCS)

Proteins in Regulation of the Immune Response. Annual Review of Immunology, 2004,

22, 503–529.

(96) Yoshimura, A.; Naka, T.; Kubo, M. SOCS proteins, cytokine signalling and immune

regulation. Nature Reviews Immunology. 2007, 7, 454-465.

(97) Yao, Z.; Cui, Y.; Watford, W.; Bream, J.; Yamaoka, K.; Hissong, B.; Li, D.; Durum, S.;

Jiang, Q.; Bhandoola, A.; Hennighausen, L.; O'Shea, J. Stat5a/b are essential for normal

lymphoid development and differentiation. Proceedings of the National Academy of

Sciences. 2006, 103, 1000-1005.

(98) Liu, X.; Robinson, G.; Wagner, K.; Garrett, L.; Wynshaw-Boris, A.; Hennighausen, L.

Stat5a is mandatory for adult mammary gland development and lactogenesis. Genes &

Development. 1997, 11, 179-186.

(99) Teglund, S.; McKay, C.; Schuetz, E.; van Deursen, J.; Stravopodis, D.; Wang, D.; Brown,

M.; Bodner, S.; Grosveld, G.; Ihle, J. Stat5a and Stat5b Proteins Have Essential and

Nonessential, or Redundant, Roles in Cytokine Responses. Cell. 1998, 93, 841-850. (100)

Wagner, K.-U.; Rui, H. Jak2/Stat5 Signaling in Mammogenesis, Breast Cancer Initiation

and Progression. Journal of Mammary Gland Biology and Neoplasia, 2008, 13, 93–103.

(101) Han, L.; Wierenga, A.T.J.; Rozenveld-Geugien, M.; van de Lande, K.; Vellenga, E.;

Schuringa, J. J. Single-Cell STAT5 Signal Transduction Profiling in Normal and

Leukemic Stem and Progenitor Cell Populations Reveals Highly Distinct Cytokine

Responses. PLoS ONE, 2009, 4, 7989(1–10).

(102) Baskiewicz-Masiuk, M.; Machalinski, B. The role of the STAT5 proteins in the

proliferation and apoptosis of the CML and AML cells. European Journal of

Haematology, 2004, 72, 420-429.

Page 93: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

80

(103) Saba, H.; Mufti, G. Advances in malignant hematology; Wiley-Blackwell:

Chichester, UK, 2011.

(104) Walker, S. R.; Xiang, M.; Frank, D. A. Distinct roles of STAT3 and STAT5 in the

pathogenesis and targeted therapy of breast cancer. Molecular and Cellular

Endocrinology, 2014, 382, 616–621.

(105) Leong, P.; Xi, S.; Drenning, S.; Dyer, K.; Wentzel, A.; Lerner, E.; Smithgall, T.;

Grandis, J. Differential function of STAT5 isoforms in head and neck cancer growth

control. Oncogene, 2002, 21, 2846-2853.

(106) Du, W.; Wang, Y.-C.; Hong, J.; Su, W.-Y.; Lin, Y.-W.; Lu, R.; Xiong, H.; Fang, J.-

Y.

STAT5 isoforms regulate colorectal cancer cell apoptosis via reduction of

mitochondrial membrane potential and generation of reactive oxygen species. Journal

of Cellular Physiology, 2012, 227, 2421–2429.

Page 94: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

)

81

(107 Kofoed, E.; Hwa, V.; Little, B.; Woods, K.; Buckway, C.; Tsubaki, J.; Pratt, K.;

Bezrodnik, L.; Jasper, H.; Tepper, A.; Heinrich, J.J.; Rosenfeld, R.G. Growth Hormone

Insensitivity Associated with a STAT5b Mutation. New England Journal of Medicine,

2003, 349, 1139-1147.

(108) Moriggl, R.; Topham, D.; Teglund, S.; Sexl, V.; McKay, C.; Wang, D.; Hoffmeyer, A.;

van Deursen, J.; Sangster, M.; Bunting, K.; Grosveld, G.C.; Ihle, J.N. Stat5 Is Required

for IL-2-Induced Cell Cycle Progression of Peripheral T Cells. Immunity, 1999, 10, 249-

259.

(109) Onishi, M.; Nosaka, T.; Misawa, K.; Mui, A.; Gorman, D.; McMahon, M.; Miyajima,

A.; Kitamura, T. Identification and Characterization of a Constitutively Active STAT5

Mutant That Promotes Cell Proliferation. Molecular and Cellular Biology, 1998, 18,

3871-3879.

(110) Ariyoshi, K.; Nosaka, T.; Yamada, K.; Onishi, M.; Oka, Y.; Miyajima, A.; Kitamura, T.

Constitutive Activation of STAT5 by a Point Mutation in the SH2 Domain. Journal of

Biological Chemistry, 2000, 275, 24407–24413.

(111) Rajala, H.L.M.; Eldfors, S., Kuusanmaki, H.; van Adrichem, A.J.; Olson, T., Lagstrom,

S.; Andersson, E. I., Jerez, A.; Clemente, M.J.; Yan, Y., Zhang, D.; Awwad, A.;

Ellonen, P.; Kallioniemi, O.; Wennerberg, K.; Porkka, K.; Maciejewski, J.P.; Loughran,

T. P.; Heckman, C. Mustjoki, S. Discovery of somatic STAT5b mutations in large

granular lymphocytic leukemia. Blood, 2013, 121, 4541–4550.

(112) Bandapalli, O.; Schuessele, S.; Kunz, J.; Rausch, T.; Stutz, A.; Tal, N.; Geron, I.;

Gershman, N.; Izraeli, S.; Eilers, J.; Vaezipour, N.; Kirschner-Schwabe, R.; Hof, J.; von

Stackelberg, A.; Schrappe, M.; Stanulla, M.; Zimmermann, M.; Koehler, R.; Avigad, S.;

Handgretinger, R.; Frismantas, V.; Bourquin, J.; Bornhauser, B.; Korbel, J.;

Muckenthaler, M.; Kulozik, A. The activating STAT5B N642H mutation is a common

abnormality in pediatric T-cell acute lymphoblastic leukemia and confers a higher risk

of relapse. Haematologica. 2014, 99, 188-192.

(113) Bhamidipati, P.; Kantarjian, H.; Cortes, J.; Cornelison, A.; Jabbour, E. Management of

imatinib-resistant patients with chronic myeloid leukemia. Therapeutic Advances in

Hematology, 2012, 4, 103-117.

(114) Warsch, W.; Kollmann, K.; Eckelhart, E.; Fajmann, S.; Cerny-Reiterer, S.; Holbl, A.;

Gleixner, K.; Dworzak, M.; Mayerhofer, M.; Hoermann, G.; Herrmann, H.; Sillaber, C.;

Egger, G.; Valent, P.; Moriggl, R.; Sexl, V. High STAT5 Levels Mediate Imatinib

Resistance And Indicate Disease Progression In Chronic Myeloid Leukemia. Blood,

2011, 117, 3409-3420.

(115) O’Hare, T.; Eide, C. A.; Deininger, M. W. New Bcr-Abl inhibitors in chronic myeloid

leukemia: keeping resistance in check. Expert Opinion on Investigational Drugs, 2008,

17, 865–878.

(116) Fathi, A.T.; Chen, Yi-B. Treatment of FLT3-ITD acute myeloid leukemia. Am. J. Blood.

Res. 2011, 1, 175-189.

(117) Gnanasambandan, K.; Magis, A.; Sayeski, P.P. The Constitutive Activation of

Jak2V617F Is Mediated by a π Stacking Mechanism Involving Phenylalanines 595 and

617. Biochemistry, 2010, 49, 9972–9984.

Page 95: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

)

82

(118) Ganetsky, A. Ruxolitinib: A New Treatment Option For Myelofibrosis.

Pharmacotherapy: The Journal of Human Pharmacology and Drug Therapy, 2013, 33,

84-92.

(119 Orphanos, G. S.; Ioannidis, G. N.; Ardavanis, A.G. Cardiotoxicity induced by tyrosine

kinase inhibitors. Acta Oncologia. 2009, 48, 964–970.

(120) Shah, D.R.; Shah, R.R.; Morganroth, J. Tyrosine Kinase Inhibitors: Their On-Target

Toxicities as Potential Indicators of Efficacy. Drug Saf. 2013, 36, 413–426.

(121) Elumalai, N.; Berg, A.; Natarajan, K.; Scharow, A.; Berg, T. Nanomolar Inhibitors of the

Transcription Factor STAT5b with High Selectivity over STAT5a. Angew. Chem. 2015,

127, 4840–4845.

(122) Avan, I.; Hall, C. D.; Katritzky, A.R. Peptidomimetics via modifications of amino acids

and peptide bonds. Chemical Society Reviews. 2014, 43, 3575–3784.

(123) Morrison, K.L.; Weiss, G.A. Combinatorial alanine-scanning. Current Opinion in

Chemical Biology. 2001, 5, 302-307.

(124) Ren, Z.; Cabell, L.A.; Schaefer, T.S.; McMurray, J.S. Identification of a High-Affinity

Phosphopeptide Inhibitor of Stat3. Bioorganic & Medicinal Chemistry Letters. 2003, 13,

633–636.

(125) Liu, B.A.; Engelmann, B.W.; Nash, P.D. The language of SH2 domain interactions

defines phosphotyrosine-mediated signal transduction. FEBS Letters. 2012, 586, 2597–

2605.

(126) Russell, R.; Breed, J.; Barton, G. Conservation Analysis And Structure Prediction Of The

SH2 Family Of Phosphotyrosine Binding Domains. FEBS Letters. 1992, 304, 1520.

(127) Grucza, R.A.; Bradshaw, J.M.; Fütterer, K.; Waksman, G. SH2 Domains: From Structure

to Energetics, A Dual Approach to the Study of Structure– Function Relationships.

Medicinal Research Reviews. 1999, 19, 273–293.

(128) Pawson, T.; Gish, G.; Nash, P. SH2 Domains, Interaction Modules And Cellular Wiring.

Trends in Cell Biology. 2001, 11, 504-511.

(129) Neculai, D.; Neculai, A.; Verrier, S.; Straub, K.; Klumpp, K.; Pfitzner, E.; Becker, S.

Structure Of The Unphosphorylated Stat5a Dimer. Journal of Biological Chemistry.

2005, 280, 40782-40787.

(130) Constantinescu, S.; Ghaffari, S.; Lodish, H. The Erythropoietin Receptor: Structure,

Activation And Intracellular Signal Transduction. Trends in Endocrinology &

Metabolism. 1999, 10, 18-23.

(131) Yoshimura, A.; Lodish, H. In Vitro Phosphorylation Of The Erythropoietin Receptor And

An Associated Protein, Pp130. Molecular and Cellular Biology. 1992, 12, 706715.

(132) Silva, M.; Benito, A.; Sanz, C.; Prosper, F.; Ekhterae, D.; Nunez, G.; Fernandez-Luna,

J. Erythropoietin Can Induce The Expression Of Bcl-Xlthrough Stat5 In

ErythropoietinDependent Progenitor Cell Lines. Journal of Biological Chemistry. 1999,

274, 2216522169.

(133) Klingmuller, U.; Bergelson, S.; Hsiao, J.; Lodish, H. Multiple Tyrosine Residues In The

Cytosolic Domain Of The Erythropoietin Receptor Promote Activation Of STAT5.

Proceedings of the National Academy of Sciences, 1996, 93, 8324-8328.

Page 96: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

)

83

(134) Quelle, F.; Wang, D.; Nosaka, T.; Thierfelder, W.; Stravopodis, D.; Weinstein, Y.; Ihle,

J. Erythropoietin Induces Activation Of Stat5 Through Association With Specific

Tyrosines On The Receptor That Are Not Required For A Mitogenic Response.

Molecular and Cellular Biology. 1996, 16, 1622-1631.

(135 Müller, J.; Schust, J.; Berg, T. A high-throughput assay for signal transducer and

activator of transcription 5b based on fluorescence polarization. Analytical

Biochemistry, 2008, 375, 249–254.

(136) Francisco-Cruz, A.; Aguilar-Santelises, M.; Ramos-Espinosa, O.; Mata-Espinosa, D.;

Marquina-Castillo, B.; Barrios-Payan, J.; Hernandez-Pando, R. Granulocyte–

macrophage colony-stimulating factor: not just another haematopoietic growth factor.

Med Oncol, 2013, 31, 774–789.

(137) Lehtonen, A.; Matikainen, S.; Miettinen, M.; Julkunen, I. Granulocyte-macrophage

colony-stimulating factor (GM-CSF)- induced STAT5 activation and target-gene

expression during human monocyte/macrophage differentiation. Journal of Leukocyte

Biology, 2002, 71, 511–519.

(138) Carr, P.; Gustin, S.; Church, A.; Murphy, J.; Ford, S.; Mann, D.; Woltring, D.; Walker,

I.; Ollis, D.; Young, I. Structure Of The Complete Extracellular Domain Of The

Common Β Subunit Of The Human GM-CSF, IL-3, And IL-5 Receptors Reveals A

Novel Dimer Configuration. Cell. 2001, 104, 291-300.

(139) Sakurai, Y.; Arai, K.; Watanabe, S. In Vitro Analysis Of STAT5 Activation By

Granulocyte-Macrophage Colony-Stimulating Factor. Genes to Cells. 2000, 5, 937-947.

(140) van Dijk, T.B.; Caldenhoven, E.; Raaijmakers, J.A.M.; Lammers, J.W.J.; Koenderman,

L.; de Groot, R.P. Multiple tyrosine residues in the intracellular domain of the common

β subunit of the interleukin 5 receptor are involved in activation of STAT5. FEBS

Letters, 1997, 412, 161–164.

(141) Argetsinger, L.S.; Kouadio, J.L.K.; Steen, H.; Stensballe, A., Jensen, O.N.; Carter-Su,

C. Autophosphorylation of JAK2 on Tyrosines 221 and 570 Regulates Its Activity.

Molecular and Cellular Biology, 2004, 24, 4955–4967.

(142) May, P.; Gerhartz, C.; Heesel, B.; Welte, T.; Doppler, W.; Graeve, L.; Horn, F.;

Heinrich, P. Comparative Study On The Phosphotyrosine Motifs Of Different Cytokine

Receptors Involved In STAT5 Activation. FEBS Letters. 1996, 394, 221-226.

(143) Carr, P.; Gustin, S.; Church, A.; Murphy, J.; Ford, S.; Mann, D.; Woltring, D.; Walker,

I.; Ollis, D.; Young, I. Structure Of The Complete Extracellular Domain Of The

Common Β Subunit Of The Human GM-CSF, IL-3, And IL-5 Receptors Reveals A

Novel Dimer Configuration. Cell. 2001, 104, 291-300.

(144) Walker, A.; Montgomery, D.; Saraiya, S.; Ho, T.; Garewal, H.; Wilson, J.; Lorand, L.

Prolactin-Immunoglobulin G Complexes From Human Serum Act As Costimulatory

Ligands Causing Proliferation Of Malignant B Lymphocytes. Proceedings of the

National Academy of Sciences. 1995, 92, 3278-3282.

(145) Gout, P.W.; Beer, C.T.; Noble, R.L. Prolactin-stimulated growth of cell cultures

established from malignant Nb rat lymphomas. Cancer Research. 1980, 40, 2433-2436.

(146) Berczi, I.; Nagy, E.; De Toledo, S.M.; Matusik, R.J.; Friesen, H.G. Pituitary hormones

regulate c-myc and DNA synthesis in lymphoid tissue. Journal of Immunology. 1991,

Page 97: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

)

84

146, 2201-2206.

(147) Pezet, A.; Ferrag, F.; Kelly, P.; Edery, M. Tyrosine Docking Sites Of The Rat Prolactin

Receptor Required For Association And Activation Of Stat5. Journal of Biological

Chemistry 1997, 272, 25043-25050.

(148) Lord, J.D.; McIntosh, B.C.; Greenberg, P.D.; Nelson, B.H. The IL-2 Receptor

Promotes Lymphocyte Proliferation and Induction of the c-myc, bcl-2, and bcl-x Genes

Page 98: Resistance and Binding Profile of Cabozantinib, a …...ii Resistance and Binding Profile of Cabozantinib, a ROS1 Inhibitor and Design of Peptidomimetic Inhibitors of STAT5 Protein

85

Through the trans-Activation Domain of Stat5. The Journal of Immunology, 2000, 164,

2533–2541.

(149) Nakamura, Y.; Russell, S.; Mess, S.; Friedmann, M.; Erdos, M.; Francois, C.; Jacques,

Y.; Adelstein, S.; Leonard, W. Heterodimerization Of The IL-2 Receptor Β- And Γ-

Chain Cytoplasmic Domains Is Required For Signalling. Nature. 1994, 369, 330-333.

(150) J G Giri, S Kumaki, M Ahdieh, D J Friend, A Loomis, K Shanebeck, R DuBose, D

Cosman, L S Park, D M Anderson. Identification and cloning of a novel IL-15 binding

protein that is structurally related to the alpha chain of the IL-2 receptor. The EMBO

Journal. 1995, 14, 3654-3663.

(151) Friedmann, M.; Migone, T.; Russell, S.; Leonard, W. Different Interleukin 2 Receptor

Beta-Chain Tyrosines Couple To At Least Two Signaling Pathways And Synergistically

Mediate Interleukin 2-Induced Proliferation. Proceedings of the National Academy of

Sciences. 1996, 93, 2077-2082.

(152) Nikolovska-Coleska, Z.; Wang, R.; Fang, X.; Pan, H.; Tomita, Y.; Li, P.; Roller, P. P.;

Krajewski, K.; Saito, N. G.; Stuckey, J. A.; Wang, S. Development and optimization of

a binding assay for the XIAP BIR3 domain using fluorescence polarization. Analytical

Biochemistry. 2004, 332, 261–273.