Chemoproteomics-Aided Medicinal Chemistry for the Discovery of EPHA2 Inhibitors

Accepted Article

Title: Chemoproteomics-aided medicinal chemistry for the discovery of

EPHA2 inhibitors

Authors: Stephanie Heinzlmeir, Jonas Lohse, Tobias Treiber, Denis

Kudlinzki, Verena Linhard, Santosh Lakshmi Gande, Sridhar

Sreeramulu, Krishna Saxena, Xiaofeng Liu, Mathias Wilhelm,

Harald Schwalbe, Bernhard Kuster, and Guillaume Médard

This manuscript has been accepted after peer review and appears as an

Accepted Article online prior to editing, proofing, and formal publication

of the final Version of Record (VoR). This work is currently citable by

using the Digital Object Identifier (DOI) given below. The VoR will be

published online in Early View as soon as possible and may be different

to this Accepted Article as a result of editing. Readers should obtain

the VoR from the journal website shown below when it is published

to ensure accuracy of information. The authors are responsible for the

content of this Accepted Article.

To be cited as: ChemMedChem 10.1002/cmdc.201700217

Link to VoR: http://dx.doi.org/10.1002/cmdc.201700217

A Journal of

www.chemmedchem.org

10.1002/cmdc.201700217

ChemMedChem

Chemoproteomics-aided medicinal chemistry for the discovery of EPHA2

inhibitors

Stephanie Heinzlmeir^{[a],[d],[e],[f]}, Jonas Lohse^[a], Tobias Treiber^[a], Dr. Denis Kudlinzki^[b],[d],[e], Verena

Linhard^[b], Dr. Santosh Lakshmi Gande^[b],[d],[e], Dr. Sridhar Sreeramulu^[b], Dr. Krishna Saxena^[b], Dr.

Xiaofeng Liu^[c], Dr. Mathias Wilhelm^[a], Prof. Dr. Harald Schwalbe^[b],[d],[e],*, Prof. Dr. Bernhard

Kuster^{[a],[d],[e],[f],[g],}*, Dr. Guillaume Médard^[a],*

[a] S. Heinzlmeir, J. Lohse, T. Treiber, Dr. M. Wilhelm, Prof. Dr. B. Kuster, Dr. G. Médard

Chair of Proteomics and Bioanalytics

Technical University of Munich

Emil-Erlenmeyer-Forum 5 , 85354 Freising, Germany

g.medard@tum.de, kuster@tum.de

[b] Dr. D. Kudlinzki, V. Linhard, Dr. S. L. Gande, Dr. S. Sreeramulu, Dr. K. Saxena, Prof. Dr. H. Schwalbe

Center for Biomolecular Magnetic Resonance

Johann Wolfgang Goethe-University

Max-von-Laue-Str. 7, 60438 Frankfurt am Main, Germany

schwalbe@nmr.uni-frankfurt.de

[c] Dr. X. Liu

Shanghai Key Laboratory of New Drug Design, School of Pharmacy, East China University of Science

and Technology, Shanghai, China

[d] S. Heinzlmeir, Dr. D. Kudlinzki, Dr. S. L. Gande, Prof. Dr. H. Schwalbe, Prof. Dr. B. Kuster

German Cancer Consortium (DKTK), Heidelberg, Germany

[e] S. Heinzlmeir, Dr. D. Kudlinzki, Dr. S. L. Gande, Prof. Dr. H. Schwalbe, Prof. Dr. B. Kuster

German Cancer Research Center (DKFZ), Heidelberg, Germany

[f] S. Heinzlmeir, Prof. Dr. B. Kuster

Bavarian Center for Biomolecular Mass Spectrometry (BayBioMS), Technical University of Munich,

Freising, Germany

[g] Prof. Dr. B. Kuster

Center for Integrated Protein Science Munich (CIPSM), Freising, Germany

1

10.1002/cmdc.201700217

ChemMedChem

Abstract

The receptor tyrosine kinase EPHA2 has gained interest as therapeutic drug target in cancer and

infectious diseases. However, EPHA2 research and EPHA2-based therapies have been hampered by the

lack of selective small molecule inhibitors. Here, we report on the synthesis and evaluation of dedicated

EPHA2 inhibitors based on the clinical BCR-ABL/SRC inhibitor Dasatinib as a lead structure. We designed

hybrid structures of Dasatinib and the previously known EPHA2 binders CHEMBL249097, PD-173955 and

a known EPHB4 inhibitor in order to exploit both the ATP pocket entrance as well as the ribose pocket as

binding epitopes in the kinase EPHA2. Medicinal chemistry and inhibitor design was guided by a

chemical proteomic approach allowing for early selectivity profiling of the newly synthesized inhibitor

candidates. Concomitant protein crystallography of 17 inhibitor co-crystals delivered detailed insight

into the atomic interactions that underlie the structure-affinity-relationship. Finally, the anti-

proliferative effect of the inhibitor candidates was confirmed in the glioblastoma cell line SF-268. In this

work, we thus discovered a novel EPHA2 inhibitor candidate 4a featuring an improved selectivity profile

while maintaining potency against EPHA2 and anti-cancer activity in SF-268 cells.

2

10.1002/cmdc.201700217

ChemMedChem

Introduction

The receptor tyrosine kinase EPHA2 has found increasing interest as a therapeutic target over the past

ten years. ^[1]In 1994, EPHA2 was linked to carcinogenesis for the first time ^[2], and, since, has been

implicated in a variety of different pathologies including cancer, infectious diseases and cataract

formation. ^{[1, 3]}In oncology, EPHA2 overexpression has been discovered in a wide variety of different

cancer entities (e.g. head and neck ^[4], prostate ^[5], breast ^[6], and non-small cell lung cancer ^[7]) and is a

known driver for carcinogenesis, metastasis, angiogenesis and resistance ^[8]of tumors ^[1]– ultimately

leading to a poor prognosis for affected patients. ^{[7, 9]}In infectious diseases, EPHA2 mediates target cell

binding and cell entry via endocytosis of pathogens, such as viruses (KHSV ^[10]and Hepatitis C virus ^[11]),

bacteria (C. trachomatis ^[12]and M. tuberculosis ^[13]) or even single cell parasites (P. falciparum ^[14]).

Inhibition of EPHA2 catalytic activity has proven to be beneficial in both disease areas. ^{[15, 16]}

Several therapeutic strategies have been developed to interfere with EPHA2 signaling. These include

agonists or antagonists targeting the extracellular ligand binding site and driving endocytosis of the

receptor, or inhibition of the intracellular kinase domain by small molecule kinase inhibitors. ^{[15, 17]}

Despite the rising relevance of EPHA2 as a therapeutic drug target, medicinal chemistry efforts aiming at

the development of dedicated EPHA2 small molecule inhibitors are still in their infancy. ^[18]Owing to this

shortage of dedicated chemical inhibitors, the dual BCR-ABL/SRC inhibitor Dasatinib – which also

potently targets EPHA2 – has been employed as a surrogate to investigate EPHA2 function and to

abrogate EPHA2 signaling in cells. ^[19]In fact, Dasatinib is the only drug which is already administered to

patients in a number of clinical trials in order to assess the clinical value of EPHA2 inhibition in treating

cancer (clinicaltrials.gov). Recently, we reported on 24 known clinical kinase inhibitors that bind EPHA2

as an unintended off-target which substantially expanded the landscape of known EPHA2 inhibitors. ^[20]

This collection of compounds should enable research to select more appropriate EPHA2 targeting

compounds and foster drug repurposing of clinically evaluated molecules for application in EPHA2-

dependent diseases. We found Dasatinib to be the most potent known EPHA2 inhibitor but its very

promiscuous target profile renders unambiguous interpretation of cell biological and clinical data

difficult.

Here we report on a medicinal chemistry effort to develop EPHA2-inhibitors with improved target

profiles. We systematically employed chemical proteomics ^{[21, 22]}alongside the medicinal chemistry

campaign to profile the potency and selectivity of each and every synthesized molecule in lysates of

cancer cells. We hypothesized that this new methodology, which can be described as

3

10.1002/cmdc.201700217

ChemMedChem

“chemoproteomics-aided drug discovery”, could facilitate lead selection and lead optimization by

establishing, in a timely manner, the affinity of the molecules for around 250 native kinases and many

other ATP-binding proteins as well as potentially toxic off-targets. ^[23]This synchronous approach indeed

provided valuable guidance during the drug-discovery process and protein crystallography added

detailed molecular insight into drug-protein interactions and structure-affinity-relationships at the

atomic level. We further demonstrate that the priority compounds discovered in this way show cellular

activity against the glioblastoma cell line SF-268 and displayed a favorable selectivity profile over other

EPHA2 inhibitors, notably Dasatinib.

4

10.1002/cmdc.201700217

ChemMedChem

Results and Discussion

Hybridization of known EPH receptor binders yields new EPHA2 inhibitor designs

Dasatinib features a very broad target profile with low nanomolar affinities mainly towards tyrosine

kinases (TK) such as the ABL family, SRC family or EPH family (Figure 1a). In total, Dasatinib targets 44

protein kinases with sub-micromolar affinity, including many proteins of the TKL, AGC, CAMK subclasses

of kinases. We hypothesized that a judicious choice of substituents could improve selectivity while

maintaining potency against EPHA2. We therefore first sought to gain a detailed understanding of the

molecular interactions governing the binding of Dasatinib to EPHA2. We identified a minimal

pharmacophore as a scaffold on which to append relevant chemical moieties found in other EPH

receptor binders. Our previously reported crystal structure that determined the binding mode of

Dasatinib to EPHA2 (PDB: 5I9Y) established that the pyrimidine moiety of Dasatinib is not engaged in

EPHA2 binding. By docking experiments, we discovered that two other known EPHA2 inhibitors,

CHEMBL249097 ^[24]and PD-173955 ^{[20, 25]}(Figure 1b), position a phenyl moiety in lieu of the pyrimidine

(Supplementary Figure 1). We therefore excluded hetero substitutions from this aromatic ring in our

design. The X-ray structure further highlighted two sites amenable for chemical modification – the

ribose pocket and the ATP pocket entrance which can both accommodate spatially demanding groups

(schematic representation in Figure 1c). On the basis of this analysis, we designed a hybrid structure of

Dasatinib and previously described EPHB4 inhibitors ^[26](Figure 1b) to investigate the size of both

pockets by introduction of bulky morpholino and methylsulfonyl substituents. In contrast to the ribose

pocket, the pocket entrance is rich in selectivity residues, which are defined as side chain exposed and

chemically targetable residues with low conservation across the kinome (notably Lys702, Glu706 and

Lys617). ^[20]Direct interactions of a compound with these charged residues could potentially drive

selectivity towards EPHA2 and away from many other kinases. Accordingly, we introduced chemical

moieties such as primary and secondary amines, hydroxyl groups or carboxylic acid groups into the

molecules. Docking of CHEMBL249097 also revealed a putative interaction between its amide and the

backbone of Glu696 located in the hinge region. Since this residue was not engaged by Dasatinib, we

probed this interaction by including an amide bond at the meta-position of the aryl. This led us to

consider 4 sub-series of the N-(2-chloro-6-methylphenyl)-2-(arylamino)thiazole-5-carboxamide series

(Figure 2): series 1, to interact with Glu696 and to probe for selectivity residues interactions by

5

10.1002/cmdc.201700217

ChemMedChem

introducing an amine; series 2, to engage the same residues at the pocket entrance by other chemical

moieties such as fluorinated piperidines, hydroxyl groups or carboxylic acids; series 3, to examine the

spatial properties of the ribose pocket by introduction of sterically demanding substituents, and series 4

to probe for dual exploitation of the pocket entrance and the ribose pocket by bulky substituents.

Figure 1. EPHA2 inhibitor design by hybridization of known EPH receptor binders. a) The dual BCR-

ABL/SRC inhibitor Dasatinib comprises a very broad selectivity profile targeting 44 proteins with sub-

micromolar affinities across several kinase families. Dasatinib is the most potent known EPHA2 inhibitor

and was chosen as molecular scaffold for the development of dedicated EPHA2 inhibitors. b)

Hybridization of known EPH binders (Dasatinib, EPHB4 inhibitors, CHEMBL249097, PD-173955)

motivated the introduction of morpholino and methylsulfonyl moieties, the substitution of Dasatinib’s

pyrimidine by an aryl moiety, and the introduction of an amide bond in meta-position. c) Inhibitor design

comprised a N-(2-chloro-6-methylphenyl)-2-(arylamino)thiazole-5-carboxamide scaffold binding the

nucleotide binding pocket and chemical modifications which were introduced at positions R’ and R’’ to

engage interactions within the ribose pocket and/or the ATP pocket entrance, respectively.

6

10.1002/cmdc.201700217

ChemMedChem

To obtain these four series, we synthesized the common intermediate 2-chloro-N-(2-chloro-6-

methylphenyl)thiazole-5-carboxamide (II) following the procedure reported for the synthesis of

Dasatinib ^[27]: 2-chlorothiazole (I) was regioselectively metalated at position C₅by n-butyllithium and the

intermediate trapped with 2-chloro-6-methylphenyl isocyanate with 70-80% yield. II was further

derivatized by acid-catalyzed nucleophilic heteroaromatic substitution with either (+)-camphor-10-

sulfonic-acid (general methods A and B) or hydrogen chloride (general method C); reaction with ethyl 3-

aminobenzoate yielded III (Method A; 66-76% yield); reaction with 5-substituted 3-aminobenzoic acids

yielded V (Method B; 28-92% yield); reaction with 3,5-disubstituted anilines yielded VII (Method C; 26-

[28]

34% yield). IV was obtained from III by reaction with a primary amine in the presence of DABAL-Me₃

in a microwave reactor at 100-140 °C ^[29](inhibitors 1a-m; 3-91 % yields). The amides VI were obtained

from the carboxylic acids V by reaction with a primary amine and PyBroP (inhibitors 2a-i, 3a, 3b; 4-74%

yields). Reacting V with 4-aminopiperidine and HATU led to the formation of both regioisomers and

reduced the yield of the desired product (3c, 22% yield). Compound V was reacted with tributyl(vinyl)tin

in a palladium-catalyzed coupling reaction according to Stille to afford the carboxylic acid precursor of

inhibitor 3d. The final inhibitor was produced by coupling this precursor to 4-aminopiperidine (27%

yield). Another inhibitor, 3e, was synthesized by reacting the bromine containing inhibitor 3a with 2-

aminobenzene boronic acid in a palladium-catalyzed Suzuki coupling reaction (32% yield). Compound VII

was afforded in an HCl-catalyzed coupling reaction by linking intermediate II to a 3,5-substituted aniline

derivative. This aniline derivative was generated from 3,5-halogenated nitrobenzene by nucleophilic

aromatic substitution at C₃and C₅, followed by reduction of the nitro group to the primary amine.

Coupling to intermediate II afforded the inhibitors 4a (26% yield) and 4b (34% yield). According to this

synthesis routes, we obtained a total of 29 EPHA2 inhibitor candidates.

7

10.1002/cmdc.201700217

ChemMedChem

Figure 2. Synthesis scheme for EPHA2 inhibitors. a) Compound I, n-butyllithium, THF, 78 C, 15 min,

argon; 2-chloro-6-methylphenyl isocyanate, THF, 78 C, 2h, argon (70-80% yield). b) Compound II, ethyl

3-aminobenzoate, (+)-camphor-10-sulfonic acid, 2-propanol, MW, 120 C, 3h, argon (66-76% yield). c)

DABAL-Me₃, primary amine, THF, 40 C, 30 min; Compound III, THF, MW, 100-140 C, 2-6 h (3-91 %

yields). d) Compound II, (+)-camphor-10-sulfonic acid, 5-substituted 3-aminobenzoic acids, tert-butanol,

MW, 120 C, 3.5 h (28-92% yield). e) Compound V, primary amine, DMF, 0 C; DIEA, TEA, PyBroP, DMF,

30-140 min (4-74% yield). f) Compound V, tributyl(vinyl)tin, dioxane, toluene, Pd(PPh₃)₄, 110 C, 4h,

argon (59% yield). g) Inhibitor 3a, 2-aminobenzenboronic acid, Pd(PPh₃)₄, DMF; K₂CO₃, MW, 115 C, 5h,

argon (32% yield). h) 3,5-substituted aniline, 2-propanol, compound II, HCl, ON, 95 C (26-43%).

EPHA2 inhibitors bind EPHA2 with high affinity

We determined the affinity of all 29 inhibitor candidates by single dose competitive Kinobeads ^{[21, 22]}

pulldowns (Supplementary Figure 2) coupled to either quantitative Western Blotting (10 M compound

concentration, Supplementary Figure 3) or mass spectrometry read out (3 M compound concentration)

to calculate the relative binding inhibition of EPHA2 compared to a DMSO control (Table 1). The results

reveal that inhibitors targeting charged residues at the ATP pocket entrance by an amine (1a-m; 40-95%

binding inhibition), a hydroxyl group (2d-f; 43-67% binding inhibition) or a carboxyl group (2g-I; 40-93%

binding inhibition) showed varying binding inhibition of EPHA2. The most promising candidates

8

10.1002/cmdc.201700217

ChemMedChem

presenting binding inhibitions of more than 90% contained either a secondary amine (1g: 95%, 1l: 94%)

or a carboxylic acid functional group (2h: 93%, 2i: 93%). For further inhibitor development, the

piperidine moiety of inhibitor 1g was favored over the other modifications as it was the only one

without chiral center. The modifications introduced at R’ (Figure 1c) allowed us to investigate the spatial

properties of the ribose pocket using groups of increasing steric hindrance (inhibitors 3a-e, 4a-b).

Inhibitor 3e comprising an aniline moiety showed high binding inhibition (96%) which indicated that

larger hydrophobic groups potentially establishing hydrophobic interactions are tolerated at this

position. Inhibitors 4a and 4b contain a morpholine or methylsulfonyl group as substituents pointing

towards the ribose pocket and showed similar behavior with very high binding inhibition of 98% and

95%, respectively. The inhibitory activity of selected compounds (2c, 2f, 2i, 4b) was confirmed by

recombinant activity assays, establishing that the Kinobeads binding assay can indeed aid in discovering

functional inhibitors of EPHA2 (Table 2). ^[30]

Novel EPHA2 inhibitors display improved selectivity

To investigate whether the new inhibitors would be more selective for EPHA2 than the parent lead

compound Dasatinib, we subjected 16 compounds to full dose response measurements using Kinobeads

with quantitative mass spectrometry readout (Supplementary Table). The addition of increasing

concentrations of an inhibitor to lysates of cancer cells in a competitive pulldown setup will lead to a

concentration-dependent decrease of compound binding to the beads and thus the mass spectrometric

intensity. This approach allows determination of an apparent dissociation constant K_d^appfor all drug-

protein interactions. As mentioned earlier, Dasatinib bound 44 targets with sub-micromolar affinity

(Figure 3a). In contrast, all novel EPHA2 inhibitors had fewer sub-micromolar targets ranging from 13

(inhibitor 2g) to 31 (inhibitor 4a). Closer inspection of the target profiles revealed decreased affinities

and fewer targets in all kinase groups, but particularly in the TK, TKL, STE and AGC branches.

Interestingly, for many of our inhibitors we observed substantially reduced binding affinities for

members of the SRC family of kinases (SRC, YES1, FYN, LCK, LYN, HCK, FRK) which are among the main

targets of the designated dual ABL/SRC inhibitor Dasatinib. We then sought to quantify the selectivity of

the inhibitors. Different metrics have been developed for this purpose including the selectivity score ^[31]

the gini coefficient ^[32]and the selectivity entropy ^[33]. However, they all suffer from different drawbacks

,

9

10.1002/cmdc.201700217

ChemMedChem

such as dependence on the number of kinases tested in the experiment, negligence of binding affinities

or non-protein centric selectivity calculation. Therefore, we developed a novel scoring system called

concentration and target dependent selectivity (CATDS; Supplementary Figure 4a) which enabled us to

overcome these limitations and to calculate drug selectivity for a certain target of interest (here: EPHA2)

in a concentration-dependent manner (here: at the K_d^appof EPHA2). This approach is of benefit in a

medicinal chemistry program to evaluate the difference in selectivity of several inhibitors in the light of

a particular target protein. CATDS adopts values between 0 (unselective) and 1 (selective) and can be

interpreted as the percentage of an applied compound that binds to a particular target at a particular

concentration. Here, simple target counting without consideration of the affinities towards EPHA2 or

the other targets would identify inhibitor 2g (13 targets below 1 M) as most selective inhibitor (Figure

3b, upper panel). This interpretation would, however, be misleading because the compound is also the

weakest of all the inhibitors (K_d^app(EPHA2): 3.4 M). In contrast the CATDS score highlighted inhibitors 4a

(CATDS_EPHA2: 0.176, K_d^app(EPHA2): 0.8 nM) and 2c (CATDS_EPHA2: 0.141, K_d^app(EPHA2): 2 nM) to be the most

selective EPHA2 inhibitors in the panel (Figure 3b, lower panel, Table 2, Supplementary Figure 4b for

comparison with other selectivity metrics). Even though inhibitor 2g only bound 13 proteins and

inhibitor 4a targeted 31 proteins at a concentration arbitrarily set to 1 M, the latter is still the more

selective compound as it engages fewer drug-protein interactions at the K_d^appof EPHA2 (Figure 3c,

Supplementary Figure 5 for radar plots of all inhibitors). In our opinion, it is important to consider the

dose dependence of compound selectivity because it can guide the choice of dose in a biological system.

10

10.1002/cmdc.201700217

ChemMedChem

Figure 3. Selectivity profiles of EPHA2 inhibitors. a) Heatmap showing selectivity profiles and apparent

binding affinities (pK_d^app[M]) of Dasatinib and EPHA2 inhibitor candidates. Target kinases are sorted

according to kinase phylogeny. EPHA2 is highlighted by a black box. b) Selectivity of EPHA2 inhibitors as

obtained by simple target counting (upper panel) or the EPHA2-specific CATDS_EPHA2score (concentration

and target dependent selectivity, lower panel). Inhibitor 4a was found to be the most potent and most

selective EPHA2 inhibitor in our panel. c) Radar plots depicting the target space and binding affinities of

inhibitor 2g which showed the lowest number of targets but also lowest affinity for EPHA2 (upper panel),

and inhibitor 4a that has the highest affinity and selectivity for EPHA2 (lower panel).

11

10.1002/cmdc.201700217

ChemMedChem

Protein crystallography validates two major pockets for interaction with EPHA2 inhibitors

We performed co-crystallization experiments of 17 EPHA2 inhibitors and recombinant EPHA2 kinase

domain (Asp596-Gly900) expressed in Sf9 cells (Table 2). ^[34]As anticipated, all inhibitors bound to the

ATP binding site in the typical type 1 (DFG-in) conformation, and the N-(2-chloro-6-methylphenyl)-2-

(methylamine)thiazole-5-carboxamide part of the molecule  common to all inhibitors, including

Dasatinib  was equally positioned in all inhibitor-EPHA2 co-crystals (Figure 4a). The direct interactions

to residues Met667, the gatekeeper Thr692 and Met695 (hinge region) as previously reported for

Dasatinib were also maintained. More interestingly, an overlay of all obtained structures showed that

the inhibitors engaged the pocket entrance as well as the ribose pocket as originally intended (Figure

4b). We noticed that residues R’ and R’’ do not adopt similar conformations in the different crystal

structures but distribute across the pockets. As observed e.g. for inhibitor 2c, the aryl moiety could

adopt different rotational poses leading to the R’ moieties to target either of the two pockets. Inhibitors

1a-m, 2a-i and 3a-e contain an amide linkage and we observed that this additional moiety, as compared

to Dasatinib, was engaged in binding interactions with the target in the majority of inhibitors. The

interaction was mostly established with Glu696 located in the hinge region, as intended by our hybrid

design (Figure 4c, 11 molecules out of 15). In addition to Glu696, the amide bond also engaged Tyr694

via direct side chain interaction (e.g. inhibitors 2f-g). Unexpectedly, we found that the amide bond

orientation was very flexible and could also form backbone interactions with Ile619 and Ala699 within

the ribose pocket (e.g. inhibitors 2e, 1l). Targeting what we expected to be “selectivity residues” also

proved successful, since some amine, hydroxyl or carboxyl functional groups, which we positioned in

different spatial orientations, were able to extend the molecular interactions to the charged residues

Lys617, Glu706 and Lys702 located at the entrance of the ATP pocket (Figure 4d). Inhibitors carrying an

amine often engaged Glu706 and Lys702 in a water-mediated interaction (e.g. inhibitors 1g, 2a, 3e, 3b,

3d). Lys617 located in the N-lobe was targeted by a primary amine (inhibitor 1j). The carboxyl containing

inhibitor 2g looped back towards the ATP pocket to interact with Tyr694. Molecules 2d and 2e

comprising a hydroxyl group unexpectedly engaged the ribose pocket which positioned the hydroxyl

group in an orientation distant to the target residues, whereas the hydroxyl group of molecule 2f

directly interacted with Glu706 with a distance of 2.5 Å. As expected, direct interactions within the

ribose pocket were scarce and inhibitor-EPHA2 interaction in this binding site was mainly driven by

hydrophobic contacts and advantageous space-filling with sterically demanding groups (e.g. inhibitors

3e, 4a). Interestingly, inhibitor 3b features a precise symmetric orientation of its trifluoro group towards

the carbonyl oxygen of the Ile619 backbone. Due to the strong electronegativity of oxygen, CF--O

12

10.1002/cmdc.201700217

ChemMedChem

interactions are typically known to be of rather repulsive than attractive nature. ^[35]However, the

carbonyl oxygen directly points towards the positively polarized C atom of the CF₃unit. It is therefore

tempting to speculate that this might positively affect inhibitor binding. Inhibitor 4a was found to be the

most affine (K_d^app: 0.8 nM) and most selective inhibitor for EPHA2 (CATDS_EPHA2: 0.176, i.e. 17.6% of 4a

bound to EPHA2, which is 4 times more than Dasatinib). Somewhat surprisingly, it did not engage in

additional direct interactions but occupied both the ATP entrance as well as the ribose pocket. We

hypothesize that occupying both pockets with the same residue reduces degrees of freedom and

therefore favors inhibitor binding due to entropic contribution.

13

10.1002/cmdc.201700217

ChemMedChem

Figure 4. Inhibitor-EPHA2 co-crystal structures validate medicinal chemistry concept. a) All 17 EPHA2

inhibitor candidates bound in the type 1 conformation (DFG-in) in the ATP binding site located between

the N lobe and the C lobe. b) The inhibitors engaged both, the ATP pocket entrance and the ribose

pocket. c) In most inhibitors, the introduced amide bond formed direct side chain interactions with

Tyr694 and Glu696 located towards the ATP pocket entrance. In some inhibitors, another conformation

involving backbone interactions to Ile619 and Ala699 in the ribose pocket was observed. d) As intended,

selectivity residues Lys617, Lys702, Glu706 were often engaged in direct or water-mediated interactions

by inhibitors carrying an amine, hydroxyl or carboxyl group. Surprisingly, inhibitor 2g carrying a

carboxylic moiety adapted an unexpected bent conformation and interacted with Tyr694 (indicated by

dashed line).

Cellular activity of EPHA2 inhibitors in SF-268 cells

We examined the anti-proliferative potential of the EPHA2 inhibitor candidates in a human cancer cell

line that is growth-dependent on EPHA2. In order to identify a cellular system that would be specifically

suitable for testing our inhibitor panel, we used kinome-wide protein expression profiling data of the

NCI60 panel of cancer cell lines previously obtained using the Kinobeads technology. ^[36]A suitable cell

line was defined as featuring high EPHA2 expression and low abundance of other kinases targeted by

the EPHA2 inhibitors. As mass-spectrometry intensities are biased by the affinity of the Kinobeads for

the different kinases, we chose to determine the expression change of the proteins across the NCI60

panel relative to the mean protein intensity level (z-score transformation). The resulting value depicts

the number of standard deviations by which the protein intensity in a particular cell line differs from the

mean intensity across the panel and indicates relative high (positive values) or low (negative values)

abundance of a protein. In the literature, the prostate cancer cell line PC-3 is often used as EPHA2

dependent cellular system. However, for testing our EPHA2 inhibitors this cell line was found to be non-

advantageous as it revealed high expression levels of several other prominent targets of our inhibitors,

such as ABL1, SRC, YES1 and other EPH inhibitors (e.g. EPHB4, EPHB2, EPHA5). The glioblastoma cell line

SF-268 showed the highest EPHA2 expression level among the NCI60 panel and also featured an

advantageous overall kinome profile (Figure 5a). We combined all 77 proteins identified as targets of all

EPHA2 inhibitors generated in this study as well as Dasatinib and sorted them according to the

14

10.1002/cmdc.201700217

ChemMedChem

maximum observed affinity, regardless of which inhibitor showed this affinity (pK_d^app, Figure 5a, black

dots). Target proteins that bound with high affinity were targeted by a large proportion (> 80 %) of the

dedicated EPHA2 inhibitors (Figure 5a, bar color). The distribution of relative intensities (Figure 5a, bar

chart) revealed that almost all potently inhibited targets other than EPHA2  as for example the main

targets of Dasatinib ABL and SRC  are relatively weakly expressed in this cell line compared to all other

cell lines of the NCI60 panel. Within a therapeutic window of 10 (indicated by the dark grey box), only

EPHA4 and FYN show slightly higher expression and may thus be candidates for relevant off-targets in

this cell line. Within a therapeutic window of 100 (indicated by the light grey box) only DDR2 was

another such candidate. Western blot analysis of SF-268 cells showed that this cell line expresses

phosphorylated EPHA2 (at S897) and AKT1 (at S473) which is indicative for aberrant and tumor-

promoting ligand-independent EPHA2 signaling (Supplementary figure 6). ^[37]Next, protein knockdown

experiments by siRNA were performed to demonstrate a functional dependence of SF-268 cell viability

on EPHA2 expression (Figure 5b). Four independent siRNAs were examined separately and as a mixture

and revealed a reduction of cell viability of about 60% after 6 days of treatment. From these results, we

deduced that SF-268 cells are suitable for testing EPHA2-directed inhibitors and that cell viability

reduction is an appropriate readout for assessing the response of the cell to EPHA2 inhibition.

15

10.1002/cmdc.201700217

ChemMedChem

Figure 5. Inhibitor 4a shows anti-proliferative effects in the SF-268 glioblastoma cell line. a) Kinome

profiling of the NCI60 panel of cancer cell lines revealed that SF-268 cells feature high EPHA2 expression

and low abundance of other targets (barchart) frequently hit by our EPHA2 inhibitors (bar color).

Inhibitor targets were sorted according to the maximum observed affinity (pK_d^app[M]) within our dataset

and theoretical therapeutic windows were set at 10x and 100x K_d^appof EPHA2. b) Protein knockdown of

EPHA2 by four independent siRNAs confirmed that SF-268 cell viability is dependent on EPHA2

expression. Experiments were performed in three biological replicates each containing three technical

replicates. Significance was calculated by a two-sided unpaired student’s t-test. c) SF-268 cells were

treated with increasing concentrations of inhibitor 4a resulting in potent reduction of cell viability similar

to Dasatinib.

16

10.1002/cmdc.201700217

ChemMedChem

EPHA2 inhibitor 4a potently reduces SF-268 cell viability

Having established the glioblastoma cell line SF-268 as a suitable biological model, we measured the

anti-proliferative effect of the inhibitor candidates synthesized in this study. Cellular activity was

examined utilizing a resazurin-based cell viability assay where reduced metabolic activity of cells leads to

decreased reduction of resazurin to resorufin. Inhibitor treatment was performed in full dose response

experiments with eight inhibitor concentrations ranging from 3 nM to 30 M. Most of the compounds

inhibited cell viability with low micromolar potency (2c: 1.4 M – 1g: 4.4 M, Table 2), whereas inhibitors

2g-i did not show any anti-proliferative effect. We attribute the latter to the presence of a charged

carboxyl moiety that likely impairs cellular uptake. Future work on these compounds will therefore

include derivatization into ester prodrugs with enhanced permeability. Inhibitor 1g bearing a piperidine

at position R’ affected SF-268 cell viability with a very low potency of 4.4 M. This result can be

rationalized by comparison to meta-fluorinated versions of this compound (inhibitors 2a-2c). Cellular

potency increased 2- to 3-fold for the singly fluorinated (EC₅₀(2a): 1.99 M, EC₅₀(2b): 1.54 M) and

doubly fluorinated inhibitor 2c (EC₅₀: 1.4 M). This indicates that the positive charge of the piperidine at

physiological pH (pK_aof around 11) could be responsible for the diminished cellular effect of inhibitor

1g. Indeed, fluorination in meta-position to the secondary amine should reduce the pK_aof piperidine

from 11 to 9 (singly fluorinated) and 7 (doubly fluorinated) and the obtained molecules showed more

activity in the viability assay. Interestingly, inhibitors 4a and 4b were much more potent (4a: 92 nM, 4b:

506 nM) and similar to the potency of the lead compound Dasatinib (97 nM) (Figure 5c). While we cannot

entirely rule out that this potency can only be attributed to the inhibition of EPHA2, this excellent

activity designated 4a as the most promising EPHA2 inhibitor in our current compound set with a strong

demonstrated anti-proliferative effect in a relevant EPHA2 overexpressing cell line.

17

10.1002/cmdc.201700217

ChemMedChem

Conclusions

In this study, we have demonstrated the utility of a chemoproteomics-aided medicinal chemistry

approach for the successful discovery of EPHA2 inhibitors based on a clinically used drug as a lead

structure. Our currently best compound inhibitor 4a shows an improved selectivity profile compared to

Dasatinib making it a more suitable chemical tool for the study of EPHA2 related disease biology. We will

continue to work on 4a and related compounds but will also provide the compound to interested

researchers in order to better characterize its properties and to further develop its potential not only for

oncology but also in infectious diseases. We also point out here that the chemoproteomic approach to

guiding medicinal chemistry presented here should be more generally applicable and could develop into

a valuable strategy for drug discovery in the future.

18

10.1002/cmdc.201700217

ChemMedChem

Experimental section

Cell lines and reagents. K-562, MV-4-11 and COLO 205 cells were cultivated in RPMI medium 1640

(RPMI1640, Biochrom GmbH) and SK-N-BE(2) was grown in Dulbecco’s modified Eagle medium/Ham’s

F12 (DMEM/F12, Biochrom GmbH), all supplemented with 10% (v/v) fetal bovine serum (Biochrom

GmbH) and 1% antibiotic/antimycotic solution (Sigma Aldrich). SF-268 cells were grown in Iscove's

Modified Dulbecco's Medium (IMDM, Biochrom GmbH) supplemented with 10% (v/v) fetal bovine

serum. Cell lines were kindly provided by the NCI-Frederick Cancer DCTD Tumor/Cell line repository,

were not found in the database of commonly misidentified cell lines (ICLAC) and were tested negatively

for mycoplasm contamination. Cell lysis and Bradford assay were performed as described

previously. ^{[22, 23]}Kinase affinity matrices were prepared in house as published elsewhere. ^{[22, 23]}

Selectivity profiling with kinase affinity matrices. 96-well plate Kinobeads competition assays have

been performed as described previously. ^{[22, 23]}Briefly, 5 mg mixed protein lysate (K-562, MV-4-11, SK-N-

BE(2) and COLO 205) was pre-incubated with compound dilutions in DMSO (0, 3, 10, 30, 100, 300, 1000,

3000, 30000 nM final concentrations) for 45 min at 4 °C. Subsequently, lysate was incubated with 35 l

settled and equilibrated Kinobeads for 30 min at 4 °C. DMSO control lysate was recovered for the

pulldown of pulldown experiment. Beads were washed and bound proteins were eluted with 40 l 2x

LDS sample buffer (NuPAGE, Invitrogen) containing 50 mM DTT. Reduced disulfide bridges were

alkylated using 4 l chloroacetamide (55 mM). Proteins were concentrated and desalted by a short SDS-

PAGE using a 4-12% gel (NuPAGE, Invitrogen) and in-gel digested according to standard procedures.

Immunoblot analysis. Antibodies against EPHA2 (C20, sc-924), EPHA2 pS897 (#6347), AKT1 (#4691),

AKT1 pS473 (#4060), α-tubulin (E-19, sc-12462), β-Actin (C4, sc-47778) were purchased from Santa Cruz

or Cell Signaling technology. Half of the obtained Kinobead eluate or 120 g cell lysate was separated by

4-12% NuPAGE gel electrophoresis and transferred onto PVDF membranes (Novex, Life Technologies).

Membranes were blocked for 1 h in 2% BSA in 1x Tris buffered saline at room temperature and probed

over night at 4 C with the respective primary antibody. Antibody binding was detected using

fluorophore-conjugated secondary antibodies (LI-COR) in an Odyssey scanner (LI-COR Biosciences).

LC-MS/MS analysis. Peptides generated by in-gel trypsin digestion were analyzed via LC-MS/MS on a

nanoLC-Ultra 1D+ (Eksigent) coupled to a LTQ-Orbitrap Elite mass spectrometer (Thermo Scientific) as

described previously. ^{[22, 23]}Peptides were delivered to a trap column (100 µm x 2 cm, packed in house

19

10.1002/cmdc.201700217

ChemMedChem

with Reprosil-Pur C18 AQ 5 µm resin, Dr. Maisch) for 10 min at a flow rate of 5 l/min in 100% solvent A₀

(0.1% v/v FA in HPLC grade water). Peptides were then separated on an analytical column (75 m x 40

cm, packed in-house with Reprosil-Gold C18, 3 µm resin, Dr. Maisch) using a 100 min gradient ranging

from 4-32% solvent B (0.1% v/v FA and 5% v/v DMSO in acetonitrile) in A (0.1% v/v FA and 5% v/v DMSO

in HPLC grade water) at a flow rate of 300 nl/min. The mass spectrometer was operated in data

dependent mode, automatically switching between MS and MS2 spectra. Up to 15 peptide precursors

were subjected to fragmentation by higher energy collision-induced dissociation (HCD) and analyzed in

the Orbitrap. Dynamic exclusion was set to 20 s. A kinase peptide inclusion list (3 most intense peptides

per kinase) was enabled.

Peptide and protein identification and quantification. Label free quantification was performed using

MaxQuant (version 1.5.3.30) ^[38]by searching MS2 data against all canonical protein sequences as

annotated in the Swissprot reference database (human proteins only, 20,193 entries, downloaded

22.03.2016, internally annotated with PFAM domains) using the embedded search engine

Andromeda ^[39]as described previously. ^{[22, 23]}Carbamidomethylated cysteine was used as fixed

modification; variable modifications included phosphorylation of serine, threonine or tyrosine, oxidation

of methionine, and N-terminal protein acetylation. Trypsin/P was specified as proteolytic enzyme with

up to two allowed missed cleavage sites. Precursor and fragment ion tolerances were 10 ppm and

20 ppm, respectively. Label-free quantification ^[40]and match-between-runs options were enabled and

results were filtered for a minimal length of seven amino acids, 1% peptide and protein FDR as well as

common contaminants and reverse identifications. For consistent peptide identification and protein

grouping, the MS data for each compound was supplemented with 15 standard DMSO controls. Each

compound was analyzed separately.

Data analysis and visualization. LFQ intensities were normalized to DMSO controls and EC₅₀values were

deduced by a four-parameter log-logistic regression using an in-house pipeline based on the drc add-on

package in R. ^[41]The depletion factor was calculated and applied to transform EC₅₀to K_D^appvalues as

described previously. ^{[22, 23]}Binding affinities are reported as pK_D^appvalues which is the negative

logarithm (base 10) of the K_D^appvalue in mol/l. For the calculation of the selectivity score S a threshold

was set at 10 times K_D^app(EPHA2), the number of kinase targets was counted and divided by all identified

protein kinases as a reference. ^[31]Gini coefficient was calculated using the K_D^app(EPHA2) of the

respective compound as threshold concentration. ^[32]Figures and tables were produced in GraphPad

Prism 5 (version 5.01) and Excel.

20

10.1002/cmdc.201700217

ChemMedChem

NCI60 kinome profiling. NCI60 kinome data were obtained from Gholami et al ^[36]and processed in

MaxQuant (version 1.4.0.5) applying the same settings as above and including iBAQ quantification.

MaxQuant results were analyzed using the MaxQuant associated software suite Perseus

(version1.5.6.0). ^[42]iBAQ intensities were log2 transformed and normalized by protein-wise Z-score

transformation.

Protein knockdown by small interfering RNA (siRNA). SF-268 cells were seeded in 96-well plates (1000

cells per well), incubated for 24 h and medium was exchanged before addition of siRNA (Qiagen:

SI00063553, SI0030081, SI00300188, SI02223508). siRNA was diluted in Opti-MEM medium (Gibco, Life

Technologies) to a concentration of 70 nM. Upon addition of INTERFERin^TM(PolyPlus, peqlab), the

mixture was vortexed for 10 s and incubated at room temperature for 10 min before addition to the

cells at a final concentration of 20 nM. The different siRNA preparations were tested individually and as

mix of all four siRNA. Controls were performed using 1 nM AllStars scrambled (Qiagen, SI03650318) and

CellDeath siRNA (Qiagen, SI04381048), INTERFERin^TMonly and Opti-MEM medium only. Cell viability was

determined according to the manufacturer after 6 days by fluorescence readout of alamarBlue (Pierce)

in a FluoStar Omega plate reader (excitation: 544 nm, emission: 590 nm). All experiments were repeated

in technical triplicates and biological triplicates. Relative cell viability was normalized to the INTERFERin

control. Statistical analysis was performed using an unpaired two-tailed t-test (significance level

p<0.001).

Kinase inhibitor treatment. SF-268 cells were seeded in 96-well plates (1000 cells per well) and

incubated for 24 h before treatment. Drug dilutions were prepared in DMSO (0, 3, 10, 30, 100, 300,

1000, 3000, 30000 nM final concentrations) and added to the cells to a final DMSO concentration of

0.3%. Cells were incubated with drug for 72 h and cell viability was determined according to the

manufacturer by fluorescence readout of alamarBlue (Pierce) in a FluoStar Omega plate reader

(excitation: 544 nm, emission: 590 nm). Drug treatments were performed in biological replicates each

containing three technical replicates.

Kinase activity assay. Dose dependent inhibition of EPHA2 kinase activity was validated for selected

inhibitors by a radioactive filter binding assay provided by Reaction Biology as published previously. ^[43]

Docking. The binding poses of Dasatinib, CHEMBL249097and PD-173955 in the ATP-binding site of

EPHA2 were predicted by molecular docking using Glide (Schrödinger, Inc.). The crystal structure of

EPHA2 in complex with ANP was retrieved from the Protein Data Bank (PDB: 1MQB) and prepared using

the Protein Preparation Wizard in Maestro (Schrödinger, Inc.) to remove non amino acid molecules, add

21

10.1002/cmdc.201700217

ChemMedChem

hydrogen atoms, and assign the protonation states for the polar residues. The scoring grid was

generated by enclosing the residues 14 Å around Dasatinib in the binding site. The docking was

performed in Glide SP mode and the top ranked pose by GScore of each inhibitor was retained for visual

analysis of interactions.

Recombinant expression and purification of EPHA2 kinase domain. The detailed protocols for isotopic

labeling procedures, expression and purification conditions of EPHA2 were published elsewhere. ^[34]In

brief, the gene encoding the catalytic domain of human EPHA2 (residues D596-G900) was synthesized at

GenScript USA Inc., USA and was optimized for its expression in insect cells. Synthesized EPHA2 gene

was sub-cloned in pTriEx 1.1 with cleavable N-terminal Flag-His tag. Recombinant baculovirus

incorporating the kinase cDNA construct was generated by homologous recombination with BacMagic

DNA (Novagen), in Spodoptera frugiperda (Sf9) cells. High titer viral stocks were generated by infection

of Sf9 cells, which were then used for the expression of EPHA2 by infection of Sf9 cells in large-scale.

Recombinantly expressed EPHA2 was isolated from the cellular extracts by passage over NiNTA resin,

followed by removal of Flag-polyhistidine tags with TEV protease (in-house) and inverse NiNTA. The

protein was further purified by size-exclusion chromatography with 20 mM Tris, 0.2 M NaCl, 5 mM

MgCl₂, 3 mM TCEP, the peak fractions were concentrated to 9.4 mg/ml and stored at –80 °C. Purified

EPHA2 was not phosphorylated during the expression in Sf9 cells and isolation procedures, as

determined by mass spectrometry.

Protein crystallization and structure determination. Purified EPHA2 protein was concentrated up to 6–

10 mg/ml in a buffer containing 20 mM Tris-HCl pH 8.0, 200 mM NaCl, 5 mM MgCl₂and 3 mM TCEP. After

the addition of different ligands up to a final concentration of 1 mM, the protein sample was diluted 1:1

with crystallization buffer in crystallization drops of 500–800 nl volume on a 96-well microplate. Crystals

were grown as sitting drops at 291 K against 50 µl reservoir solution. Rod-shaped crystals (0.1–0.2 mm)

or plate-like cuboids (0.05–0.1 mm) appeared after 1–2 weeks and grew to their final size within 4

weeks. Crystallization conditions were composed from three different building blocks and varied for

each ligand between the following ranges: 30–40% precipitant solution (stock solution consisting of 25%

PEG 1000, 25% PEG 3350, 25% MPD), 0.05–0.3 M amino acids solution (stock solution consisting of 0.2

M glutamate, 0.2 M glycine, 0.2 M serine, 0.2 M alanine, 0.2 M lysine) or 0.05–0.3 M carboxylic acids

solution respectively (stock solution consisting of 0.2 M sodium formate, 0.2 M ammonium acetate, 0.2

M sodium citrate, 0.2 M sodium/potassium tartrate, 0.2 M sodium oxamate) and 0.1 M buffer pH 5.5-

8.5 (Bis-Tris, Tris, Hepes, MES, Bicine or Tris/Bicine). Prior to flash-cooling some crystals had to be

22

10.1002/cmdc.201700217

ChemMedChem

soaked with 20% ethylene glycol in mother liquor for cryoprotection. Diffraction data at resolutions

between 1.10–1.89 Å had been collected at beamline BL14.1 operated by the Helmholtz-Zentrum Berlin

(HZB) at the BESSY II electron storage ring (Berlin, Germany) ^[44]at beamline P13 operated by the EMBL

Hamburg at the PETRA III synchrotron source (DESY, Hamburg Germany) ^[45], at X06DA (PXIII) beamline

operated by the Swiss Light Source (PSI, Villigen, Switzerland) and at PROXIMA-1 beamline operated by

the Synchrotron SOLEIL (Saint-Aubin, France). The data was processed using XDSAPP. ^[46]The structure

was determined by molecular replacement with Phaser ^[47]using the crystal structure of the human

EPHA2 kinase domain (PDB: 5I9Y) ^[20]as a search model. Model building was performed using Coot ^[48]

and the structure was refined and validated using PHENIX. ^[49]Drug-protein interaction analysis was

performed using LigPlot+. ^[50]Figures containing molecular graphics were prepared using PyMOL

(Schroedinger). Structure-solution and refinement statistics can be found in Supporting Information.

Data deposition. Structural data is accessible via PDB protein data bank: inhibitor 1g (pdb: 5NJZ),

inhibitor 1j (pdb: 5NK0), inhibitor 1k (pdb: 5NK1), inhibitor 1l (pdb: 5NK3), inhibitor 1m (pdb: 5NK5),

inhibitor 2a (pdb: 5NK7), inhibitor 2c (pdb: 5NK4), inhibitor 2d (pdb: 5NK6), inhibitor 2e (pdb: 5NK9),

inhibitor 2f (pdb: 5NK8), inhibitor 2g (pdb: 5NKA), inhibitor 3a (pdb: 5NKE), inhibitor 3b (pdb: 5NKF),

inhibitor 3d (pdb: 5NKG), inhibitor 3e (pdb: 5NKH), inhibitor 4a (pdb: 5NKB), inhibitor 4b (pdb: 5NKI).

The mass spectrometry data and all obtained dose response curves for each inhibitor selectivity profiling

have been deposited to the ProteomeXchange Consortium

(http://proteomecentral.proteomexchange.org) via the PRIDE partner repository ^[51]with the dataset

identifier PXD006193.

23

10.1002/cmdc.201700217

ChemMedChem

Acknowledgements

Diffraction data were collected on BL14.1 at BESSY II (Helmholtz-Zentrum Berlin), on beamline P13 at

PETRA III (EMBL Hamburg), on X06DA (PXIII) beamline at the Swiss Light Source (Paul Scherer Institute)

and at PROXIMA-1 beamline operated by the Synchrotron SOLEIL (Saint-Aubin, France). The presented

research was in parts funded by the German Consortium for Translational Cancer Research (DKTK), the

German Cancer Research Center (DKFZ), Heidelberg, Germany and the European Community's Seventh

Framework Programme (FP7/2007-2013) under the BioStruct-X (grant agreement N283570). H.S. is

member of the DFG cluster of excellence: macromolecular complexes. BMRZ is supported by the state of

Hessen. B.K. is a member of the DFG cluster of excellence: Center for Integrated Protein Science Munich

(CIPSM). The authors want to thank Susan Klaeger, Runsheng Zheng and Dongxue Wang for mass-

spectrometric measurements and Huichao Qiao for production of cell lysates. The authors also want to

thank Andrea Hubauer, Michaela Krötz-Fahning and Andreas Klaus for technical assistance and Jonas

Goldstein for LigPlot+ analysis. The authors thank the members of the Chair of Food Chemistry and

Molecular Sensory Science (Prof. Hofmann) and the Chair of Biochemistry (Prof. Eisenreich, Prof. Groll)

at the Technical University of Munich for NMR measurement.

Keywords: chemical proteomics, drug discovery, EPH receptor, inhibitors, selectivity profiling

24

10.1002/cmdc.201700217

ChemMedChem

References

[1]

[2]

A. Barquilla, E. B. Pasquale, Annu Rev Pharmacol Toxicol 2015, 55, 465-487.

A. C. Andres, H. H. Reid, G. Zurcher, R. J. Blaschke, D. Albrecht, A. Ziemiecki, Oncogene 1994,

9(5), 1461-1467.

[3]

[4]

J. E. Park, A. I. Son, R. Zhou, Genes (Basel) 2013, 4(3), 334-357.

Z. Wu, J. B. Doondeea, A. M. Gholami, M. C. Janning, S. Lemeer, K. Kramer, S. A. Eccles, S. M.

Gollin, R. Grenman, A. Walch, S. M. Feller, B. Kuster, Mol Cell Proteomics 2011, 10(12), M111.011635.

[5] M. L. Taddei, M. Parri, A. Angelucci, B. Onnis, F. Bianchini, E. Giannoni, G. Raugei, L. Calorini, N.

Rucci, A. Teti, M. Bologna, P. Chiarugi, Am J Pathol 2009, 174(4), 1492-1503.

[6]

[7]

E. Tsouko, J. Wang, D. E. Frigo, E. Aydogdu, C. Williams, Carcinogenesis 2015, 36(9), 1051-1060.

J. M. Brannan, W. Dong, L. Prudkin, C. Behrens, R. Lotan, B. N. Bekele, I. Wistuba, F. M. Johnson,

Clin Cancer Res 2009, 15(13), 4423-4430.

[8]

[9]

H. Koch, M. E. Busto, K. Kramer, G. Medard, B. Kuster, J Proteome Res 2015, 14(6), 2617-2625.

P. D. Dunne, S. Dasgupta, J. K. Blayney, D. G. McArt, K. L. Redmond, J. A. Weir, C. A. Bradley, T.

Sasazuki, S. Shirasawa, T. Wang, S. Srivastava, C. W. Ong, K. Arthur, M. Salto-Tellez, R. H. Wilson, P. G.

Johnston, S. Van Schaeybroeck, Clin Cancer Res 2015, 22(1), 230-242.

[10]

[11]

C. Boshoff, Nat Med 2012, 18(6), 861-863.

C. C. Colpitts, J. Lupberger, C. Doerig, T. F. Baumert, Biochim Biophys Acta 2015, 1854(10 Pt B),

1657-1662.

[12]

P. Subbarayal, K. Karunakaran, A. C. Winkler, M. Rother, E. Gonzalez, T. F. Meyer, T. Rudel, PLoS

Pathog 2015, 11(4), e1004846.

[13]

M. Khounlotham, S. Subbian, R. Smith, 3rd, S. L. Cirillo, J. D. Cirillo, J Infect Dis 2009, 199(12),

1797-1806.

[14]

A. Kaushansky, A. N. Douglass, N. Arang, V. Vigdorovich, N. Dambrauskas, H. S. Kain, L. S. Austin,

D. N. Sather, S. H. Kappe, Science 2015, 350(6264), 1089-1092.

[15]

[16]

M. Tandon, S. V. Vemula, S. K. Mittal, Expert Opin Ther Targets 2011, 15(1), 31-51.

S. Chakraborty, M. V. Veettil, V. Bottero, B. Chandran, Proc Natl Acad Sci U S A 2012, 109(19),

E1163-1172.

[17]

[18]

R. Noberini, I. Lamberto, E. B. Pasquale, Semin Cell Dev Biol 2012, 23(1), 51-57.

C. J. Lim, K. S. Oh, J. D. Ha, J. H. Lee, H. W. Seo, C. H. Chae, D. G. Kim, M. J. Lee, B. H. Lee, Bioorg

Med Chem Lett 2014, 24(17), 4080-4083.

[19]

[20]

Q. Chang, C. Jorgensen, T. Pawson, D. W. Hedley, Br J Cancer 2008, 99(7), 1074-1082.

S. Heinzlmeir, D. Kudlinzki, S. Sreeramulu, S. Klaeger, S. L. Gande, V. Linhard, M. Wilhelm, H.

Qiao, D. Helm, B. Ruprecht, K. Saxena, G. Medard, H. Schwalbe, B. Kuster, ACS Chem Biol 2016, 11(12),

3400-3411.

[21]

M. Bantscheff, D. Eberhard, Y. Abraham, S. Bastuck, M. Boesche, S. Hobson, T. Mathieson, J.

Perrin, M. Raida, C. Rau, V. Reader, G. Sweetman, A. Bauer, T. Bouwmeester, C. Hopf, U. Kruse, G.

Neubauer, N. Ramsden, J. Rick, B. Kuster, G. Drewes, Nat Biotechnol 2007, 25(9), 1035-1044.

[22]

T. Kuehne, Z. Wu, A. Dittmann, C. Hopf, K. Kramer, B. Kuster, J Proteome Res 2015, 14(3), 1574-1586.

[23] S. Klaeger, B. Gohlke, J. Perrin, V. Gupta, S. Heinzlmeir, D. Helm, H. Qiao, G. Bergamini, H. Handa,

M. M. Savitski, M. Bantscheff, G. Medard, R. Preissner, B. Kuster, ACS Chem Biol 2016, 11(5), 1245-1254.

G. Medard, F. Pachl, B. Ruprecht, S. Klaeger, S. Heinzlmeir, D. Helm, H. Qiao, X. Ku, M. Wilhelm,

[24]

P. Bamborough, D. Drewry, G. Harper, G. K. Smith, K. Schneider, J Med Chem 2008, 51(24), 7898-

7914.

[25]

M. I. Davis, J. P. Hunt, S. Herrgard, P. Ciceri, L. M. Wodicka, G. Pallares, M. Hocker, D. K. Treiber,

P. P. Zarrinkar, Nat Biotechnol 2011, 29(11), 1046-1051.

25

10.1002/cmdc.201700217

ChemMedChem

[26]

C. Bardelle, B. Barlaam, N. Brooks, T. Coleman, D. Cross, R. Ducray, I. Green, C. L. Brempt, A.

Olivier, J. Read, Bioorg Med Chem Lett 2010, 20(21), 6242-6245.

[27]

J. Das, P. Chen, D. Norris, R. Padmanabha, J. Lin, R. V. Moquin, Z. Shen, L. S. Cook, A. M.

Doweyko, S. Pitt, S. Pang, D. R. Shen, Q. Fang, H. F. de Fex, K. W. McIntyre, D. J. Shuster, K. M. Gillooly, K.

Behnia, G. L. Schieven, J. Wityak, J. C. Barrish, J Med Chem 2006, 49(23), 6819-6832.

[28]

5769.

[29]

[30]

[31]

A. Novak, L. D. Humphreys, M. D. Walker, S. Woodward, Tetrahedron Lett 2006, 47(32), 5767-

D. Glynn, D. Bernier, S. Woodward, Tetrahedron Lett 2008, 49(39), 5687-5688.

A. F. Rudolf, T. Skovgard, S. Knapp, L. J. Jensen, J. Berthelsen, PLoS One 2014, 9(6), e98800.

M. W. Karaman, S. Herrgard, D. K. Treiber, P. Gallant, C. E. Atteridge, B. T. Campbell, K. W. Chan,

P. Ciceri, M. I. Davis, P. T. Edeen, R. Faraoni, M. Floyd, J. P. Hunt, D. J. Lockhart, Z. V. Milanov, M. J.

Morrison, G. Pallares, H. K. Patel, S. Pritchard, L. M. Wodicka, P. P. Zarrinkar, Nat Biotechnol 2008, 26(1),

127-132.

[32]

[33]

[34]

P. P. Graczyk, J Med Chem 2007, 50(23), 5773-5779.

J. C. Uitdehaag, G. J. Zaman, BMC Bioinformatics 2011, 12, 94.

S. L. Gande, K. Saxena, S. Sreeramulu, V. Linhard, D. Kudlinzki, S. Heinzlmeir, A. J. Reichert, A.

Skerra, B. Kuster, H. Schwalbe, Chembiochem 2016, 1439-7633.

[35]

[36]

P. Zhou, J. Zou, F. Tian, Z. Shang, J Chem Inf Model 2009, 49(10), 2344-2355.

A. Moghaddas Gholami, H. Hahne, Z. Wu, F. J. Auer, C. Meng, M. Wilhelm, B. Kuster, Cell Rep

2013, 4(3), 609-620.

[37]

H. Miao, D. Q. Li, A. Mukherjee, H. Guo, A. Petty, J. Cutter, J. P. Basilion, J. Sedor, J. Wu, D.

Danielpour, A. E. Sloan, M. L. Cohen, B. Wang, Cancer Cell 2009, 16(1), 9-20.

[38]

[39]

J. Cox, M. Mann, Nat Biotechnol 2008, 26(12), 1367-1372.

J. Cox, N. Neuhauser, A. Michalski, R. A. Scheltema, J. V. Olsen, M. Mann, J Proteome Res 2011,

10(4), 1794-1805.

[40]

J. Cox, M. Y. Hein, C. A. Luber, I. Paron, N. Nagaraj, M. Mann, Mol Cell Proteomics 2014, 13(9),

2513-2526.

[41]

[42]

C. Ritz, J. C. Streibig, J. Statist. Software 2005, 12(5).

S. Tyanova, T. Temu, P. Sinitcyn, A. Carlson, M. Y. Hein, T. Geiger, M. Mann, J. Cox, Nat Methods

2016, 13(9), 731-740.

[43]

T. Anastassiadis, S. W. Deacon, K. Devarajan, H. Ma, J. R. Peterson, Nat Biotechnol 2011, 29(11),

Röwer, K. Sparta, M. Steffien, M. Ühlein, P. Wilk, M. C. Weiss, Eur. Phys. J. Plus 2015, 130, 141-150.

[45]

M. Cianci, G. Bourenkov, G. Pompidor, I. Karpics, J. Kallio, I. Bento, M. Roessle, F. Cipriani, S.

Fiedler, T. R. Schneider, J Synchrotron Radiat 2017, 24(Pt 1), 323-332.

[46]

[47]

M. Krug, M. Weiss, U. Heinemann, U. Mueller, J. Appl. Cryst. 2012(45), 568-572.

A. J. McCoy, R. W. Grosse-Kunstleve, P. D. Adams, M. D. Winn, L. C. Storoni, R. J. Read, J Appl

Crystallogr 2007, 40(Pt 4), 658-674.

[48]

[49]

P. Emsley, K. Cowtan, Acta Crystallogr D Biol Crystallogr 2004, 60(Pt 12 Pt 1), 2126-2132.

P. V. Afonine, R. W. Grosse-Kunstleve, N. Echols, J. J. Headd, N. W. Moriarty, M. Mustyakimov, T.

C. Terwilliger, A. Urzhumtsev, P. H. Zwart, P. D. Adams, Acta Crystallogr D Biol Crystallogr 2012, 68(Pt 4),

[51]

J. A. Vizcaino, R. G. Cote, A. Csordas, J. A. Dianes, A. Fabregat, J. M. Foster, J. Griss, E. Alpi, M.

Birim, J. Contell, G. O'Kelly, A. Schoenegger, D. Ovelleiro, Y. Perez-Riverol, F. Reisinger, D. Rios, R. Wang,

H. Hermjakob, Nucleic Acids Res 2013, 41(Database issue), D1063-1069.

26

10.1002/cmdc.201700217

ChemMedChem

Legends for figures and schemes

Figure 1. EPHA2 inhibitor design by hybridization of known EPH receptor binders. a) The dual BCR-

ABL/SRC inhibitor Dasatinib comprises a very broad selectivity profile targeting 44 proteins with sub-

micromolar affinities across several kinase families. Dasatinib is the most potent known EPHA2 inhibitor

and was chosen as molecular scaffold for the development of dedicated EPHA2 inhibitors. b)

Hybridization of known EPH binders (Dasatinib, EPHB4 inhibitors, CHEMBL249097, PD-173955)

motivated the introduction of morpholino and methylsulfonyl moieties, the substitution of Dasatinib’s

pyrimidine by an aryl moiety, and the introduction of an amide bond in meta-position. c) Inhibitor

design comprised a N-(2-chloro-6-methylphenyl)-2-(arylamino)thiazole-5-carboxamide scaffold binding

the nucleotide binding pocket and chemical modifications which were introduced at positions R’ and R’’

to engage interactions within the ribose pocket and/or the ATP pocket entrance, respectively.