Trisubstituted Imidazoles with a Rigidized Hinge Binding Motif Act As

May 8, 2017 - Trisubstituted Imidazoles with a Rigidized Hinge Binding Motif Act As Single Digit nM Inhibitors of Clinically Relevant EGFR L858R/T790M...
54 downloads 10 Views 4MB Size
Article pubs.acs.org/jmc

Trisubstituted Imidazoles with a Rigidized Hinge Binding Motif Act As Single Digit nM Inhibitors of Clinically Relevant EGFR L858R/ T790M and L858R/T790M/C797S Mutants: An Example of Target Hopping Michael Juchum,† Marcel Günther,† Eva Döring,† Adrian Sievers-Engler,† Michael Lam ̈ merhofer,† ,† and Stefan Laufer* †

Department of Pharmaceutical Chemistry, Institute of Pharmaceutical Sciences, Eberhard Karls Universität Tübingen, Auf der Morgenstelle 8, 72076 Tübingen, Germany S Supporting Information *

ABSTRACT: The high genomic instability of non-small cell lung cancer tumors leads to the rapid development of resistance against promising EGFR tyrosine kinase inhibitors (TKIs). A recently detected triple mutation compromises the activity of the gold standard third-generation EGFR inhibitors. We have prepared a set of trisubstituted imidazoles with a rigidized 7-azaindole hinge binding motif as a new structural class of EGFR inhibitors by a target hopping approach from p38α MAPK inhibitor templates. On the basis of an iterative approach of docking, compound preparation, biological testing, and SAR interpretation, robust and flexible synthetic routes were established. As a result, we report two reversible inhibitors 11d and 11e of the clinically challenging triple mutant L858R/T790M/C797S with IC50 values in the low nanomolar range. Furthermore, we developed a kinome selective irreversible inhibitor 45a with an IC50 value of 1 nM against the EGFR L858R/T790M double mutant. Target binding kinetics and metabolic stability data are included. These potent mutant EGFR inhibitors may serve as a basis for the development of structurally novel EGFR probes, tools, or candidates.



Besides different “escape pathways”9−11 and histological transformations,12,13 60% of the patients revealed a specific additional T790M point mutation.14 Second-generation inhibitors like afatinib were developed, carrying a Michael acceptor group to target a cysteine residue in the active cleft, acting as an irreversible inhibitor. Although this transformation led to a recovery of inhibition potency, a total loss of selectivity was recognized. Inhibition of the WT EGFR led to tremendous limitations in clinical application with respect to dose-limiting toxicities.9,15,16 These problems were circumvented by the development of third-generation EGFR inhibitors. N-(3-((5Chloro-2-((2-methoxy-4-(4-methyl-1-piperazinyl)phenyl)amino)-4-pyrimidinyl)oxy)phenyl)-2-propenamide, 46 (WZ4002),17 was the first reported mutant-selective EGFR inhibitor showing a 40-fold selectivity for EGFR L858R/ T790M over the WT. A few years later, the WT sparing inhibitors oimertinib18 and rociletinib19 were designed and studied in clinical trials, showing promising results.20,21 In 2015, osimertinib was FDA approved for NSCLC patients with proven T790M mutation who were treated with TKIs before but who had developed further tumor progression. Several recent studies have revealed the development of resistance to these new third-generation TKIs.19,22,23 One study exposed the

INTRODUCTION In recent decades, significant progress in the research fields of cell biology and cancer genetics has revealed the high instability of the tumor genome, leading to the knowledge that cancer is not a disease of an organ but rather of the genome. Around 30−35% of East Asian and 15% of Caucasian non-small cell lung cancer (NSCLC) patients bear mutations in the EGFR gene.1 Most of these mutations occur in the exons 18−21 that harbor the encoding genes for the tyrosine kinase domain of the receptor.2 The two most predominant types, covering 90% of these mutations, the so-called activating mutations, are exon 19 deletions or exon 21 L858R point mutations.3 In the case of L858R, the receptor shows a 50-fold increased catalytic activity, representing a crucial hub for the tumor cell in terms of increased downstream signaling, cell growth, and metastasis.4 The two approved amino quinazoline TKIs gefitinib and erlotinib revealed very potent and selective inhibition profiles with respect to these activating mutations. These mutants lose binding affinity toward ATP,4,5 whereas the TKIs bind stronger compared to the WT ones.6 Promising results were confirmed in clinical studies, and TKI sensitive patients showed high response rates and significant tumor shrinkage. Unfortunately, no benefit in overall survival was recognized because long-term TKI treatment led to acquired resistance and development of new tumor tissue.7,8 Genome analysis exposed different mechanisms of TKI resistance for these cancer cell lines. © 2017 American Chemical Society

Received: February 2, 2017 Published: May 8, 2017 4636

DOI: 10.1021/acs.jmedchem.7b00178 J. Med. Chem. 2017, 60, 4636−4656

Journal of Medicinal Chemistry

Article

Figure 1. (A) Molecular docking of lead structure 11g in the EGFR T790M kinase domain X-ray structure (PDB 2JIU). Hydrogen bonds are shown in orange dots. The angle between the imidazole and the fluorophenyl residue is visualized as blue dots. The Met790 gatekeeper is shown as spheres to visualize steric demands. Hydrophobic region I: area that is occupied by fluorophenyl residue, shown with purple coloration. (B) Addressed regions in our SAR study. HR I: hydrophobic region I, gatekeeper. PBS: phosphate binding site, much space, polar amino acids, Cys797 accessible. RP: ribose pocket, Cys797 accessible. HR II: hydrophobic region II, Cys797 accessible.

Scheme 1. Synthetic Route of Reversible EGFR Inhibitors of Table 1 (11a−f)a

(a) n-BuLi, Sn(Bu)3Cl, THF, −78 °C to rt (compound 9d: (1) NIS, MeCN, rt, (2) (CH3)2CHMgCl·LiCl, Sn(Bu)3Cl, THF, 0 °C to rt); (b) Pd(OAc)2:X-Phos (3:1), dioxane, reflux; (c) HCl (conc), MeOH, reflux (11e: (1) DCM, TFA, rt; (2) MeOH, K2CO3(aq), rt).

a

fluorophenyl residue in 4-position of the 7-azaindole occupied the hydrophobic region I (HR I). Earlier investigations of trisubstituted imidazoles as kinase inhibitors revealed the ideal angle and orientation of these compounds to interact with the HR I without any steric interference of the aromatic substituent in 4-position with the gate keeper side chain (Figure 1A).27 In more detail, the 4-fluorophenyl group in this position is an important structural element for inhibitors of the c-Jun Nterminal kinase 3.28 These compounds occupy the HR I by passing the Met gatekeeper. A methionine gatekeeper is also present in the clinically challenging EGFR mutant T790M. Besides the proven regain of ATP binding affinity (unlike the L858R mutant),29 the more sterically demanding methionine gatekeeper appears to clash with structural elements of firstgeneration EGFR inhibitors like gefitinib.8 We see the chance to occupy the HRI with our compounds without any gatekeeper interference in order to gain noncovalent binding affinity. On the basis of this promising lead structure, different synthetic strategies were developed to study the structure− activity relationships (SARs) of this new class of trisubstituted imidazoles as EGFR inhibitors (Figure 1B). On the basis of in silico docking studies, we created different substitution patterns. Together with the biological data, we identified the crucial structural elements of these trisubstituted imidazoles for a potent inhibition of the clinical challenging EGFR mutants. Kinetic binding studies, metabolic stability measurements, and a kinome selectivity screen of compound 45a were done to reveal both the effective formation of a covalent bond to the targetcysteine and stable, safe, and selective properties.

occurrence of a third point mutation, C797S, leading to a total loss of inhibitor quality in terms of the potential Michael acceptor cysteine reaction.24 The unstable genome of these tumors, accompanied by the fast development of resistance, calls for the design of structurally new potent EGFR inhibitors. Efforts to gain a greater understanding of inhibitor-active sitebinding properties are crucial and may lead to the discovery of new binding sites which could counteract the development of drug resistance. Herein, we report the development of a recently published target hopping strategy based on p38α MAPK inhibitor templates which resulted in the creation of potent EGFR L858R/T790M and EGFR L858R/T790M/C797S inhibitors.25 An off-target hit of a published p38α MAPK inhibitor (11g)26 (Figure 1A) revealed the efficacy of this class of ligands bound to the EGFR WT kinase domain. While inhibition of the WT (IC50 = 49.2 nM) and the activating mutant L858R (IC50 = 24.1 nM) was high, inhibition of the double mutant L858R/T790M (IC50 = 916 nM) was only moderate. The goal of our studies was to shift the kinase target of a known p38α MAPK inhibitor and transform it into an innovative class of potent EGFR inhibitors that have high binding affinities to the clinically challenging NSCLC mutants. In silico docking studies had shown a reasonable orientation of the mentioned lead structure 11g in the active site kinase domain of the T790M mutant (Figure 1A) in terms of hinge binding, nonpolar interactions, and possible attachment points for acrylamides to covalently target Cys797.25 Hinge binding was mediated via the 1H-pyrrolo[2,3-b]pyridine structure. The attached phenyl ring in the 2-position of this 7-azaindole towered into the hydrophobic region II (HR II). The p4637

DOI: 10.1021/acs.jmedchem.7b00178 J. Med. Chem. 2017, 60, 4636−4656

Journal of Medicinal Chemistry

Article

Scheme 2. Amide Coupling of the Benzylamines 18a,b to the Inhibitors 19a−ca

a

(a) Acrylic acid (19a,c)/(E)-4-(dimethylamino)but-2-enoic acid hydrochloride (19b), TBTU, DIPEA, THF, rt.

Scheme 3. Synthetic Route of the Building Blocks for the EGFR Inhibitors 27a,b and 33 of Table 2a

(a) NH4OAc, MeOH, rt; (b) NaH, MOM-Cl, THF, 0 °C to rt; (c) n-BuLi, Sn(Bu)3Cl, THF −78 °C to rt; (d) DIPA, n-BuLi, I2, THF, −78 °C to rt; (e) Pd(PPh3)2Cl2, MeCN, 1 M K2CO3(aq), 60 °C; (f) (1) mCPBA, EA, 0 °C to rt, (2) mesyl chloride, DMF, 70 °C; (g) n-BuLi, B(OMe)3, Et2O, −40 °C to rt.

a



RESULTS AND DISCUSSION Chemistry. The first set of reversible EGFR inhibitors (11a−f, Table 1) was synthesized using the optimized Stille cross-coupling route for trisubstituted imidazoles, published by Selig et al.26 (Scheme 1; for details, see Supporting Information, Scheme S1). We reported the synthesis of 18a previously.25 18b was designed in the analogous way (for detailed synthesis, see Supporting Information, Scheme S2). Amide coupling of the benzylamines 18a,b with acrylic acid/(E)-4-(dimethylamino)but-2-enoic acid hydrochloride (synthesized according to literature30) in the presence of TBTU and DIPEA led to the desired acrylamides (19a,c)/4-dimethylamino crotonic acid amide (19b) (Scheme 2). The series of inhibitors with the acrylamide attached to the phenyl at the 7-azaindole 2-position was synthesized starting from the imidazole building block 21a that was synthesized via Radziszewski synthesis (Scheme 3). 4-Chloro-1-(phenylsulfonyl)-1H-pyrrolo[2,3-b]pyridine (23*) was synthesized as described in literature.31 Selective deprotonation of the 2-position with n-BuLi (ortho-directing effect of the phenylsulfonyl group), followed by the addition of elemental iodine, led to the halogenated 7-azaindole 23. It was

fused with Boc-protected (3-aminophenyl)boronic acid (22) in a Suzuki cross-coupling reaction to obtain 24. The second 7azaindole building block (30) was synthesized starting with the commercially available 1H-pyrrolo[2,3-b]pyridine. Chlorination32 and MOM-protection yielded in compound 5b. Deprotonation and borylation at the 2-position formed the boronic acid 28. Suzuki cross-coupling reaction with Bocprotected (3-bromophenyl)methanamine (29) led to the building block 30. Stille cross-coupling with 21a* and deprotection yielded in 26 is described (Scheme 4). To obtain the acrylamide 27a, the aniline was coupled with 3-chloropropanoyl chloride. Elimination of HCl in the presence of NaOH(aq) created the necessary double bond. The reversible counterpart 27b was synthesized via the reaction of the free aniline with propionyl chloride in THF. Compound 31 was obtained by Stille cross-coupling reaction of 21a* with 30. Deprotection and amide coupling with acrylic acid in the presence of DIPEA and TBTU led to the final product 33. To obtain the simplified hinge binder motifs (Scheme 5), commercially available 1H-pyrrolo[2,3-b]pyridine was SEMprotected (34) and borylated at the 2-position selectively (35). Suzuki cross-coupling with 3-bromo benzylamine followed by amide coupling with acryloyl chloride yielded 36. Final 4638

DOI: 10.1021/acs.jmedchem.7b00178 J. Med. Chem. 2017, 60, 4636−4656

Journal of Medicinal Chemistry

Article

Scheme 4. Synthetic Route of the EGFR Inhibitors with a Michael Acceptor at the Phenyl Ring Attached to the 7-Azaindole (27a,b and 33) of Table 2a

(a) Pd(OAc)2:X-Phos (3:1), dioxane, reflux; (b) HCl (conc), MeOH, reflux; (c) (1) 3-chloropropanoyl chloride, THF, −15 °C, (2) NaOH, H2O, 60 °C; (d) propionyl chloride, THF, rt; (e) acrylic acid, TBTU, DIPEA, THF, 0 °C to rt.

a

Scheme 5. Synthetic Route of 7-Azaindole Hinge Binder Motifs to Study the Best Attachment Point for a Michael Acceptor (37, 41, Table 2)a

(a) NaH, SEM-Cl, DMF, 0 °C to rt; (b) n-BuLi, B(OMe)3, Et2O, −40 °C to rt; (c) (1) 3-bromo benzylamine, K2CO3, Pd(PPh3)4, dioxane, H2O, 80 °C, (2) acryloyl chloride, DIPEA, THF, 0 °C; (d) (1) TFA, DCM, rt, (2) MeOH, NaHCO3(aq), rt; (e) NIS, MeCN, 0−50 °C; (f) Pd(PPh3)2Cl2, 1 M Na2CO3(aq), MeCN, 60 °C; (g) (1) TFA, DCM, rt, (2) EtOH, NaHCO3(aq), rt.

a

deprotection with TFA (complete deprotection with MeOH and NaHCO3(aq)) led to the 7-azaindole with an Nbenzylacrylamido residue in position 2 (37). To obtain the 3substituted 1H-pyrrolo[2,3-b]pyridine, 4 was SEM-protected and iodinated at the 3-position (38). Suzuki cross-coupling with the prior synthesized acrylamide 39 yielded 40, whose deprotection with TFA (complete deprotection with EtOH and NaHCO3(aq)) resulted in compound 41. Stille cross-coupling of the SEM-protected, stannylated imidazoles with the chlorinated and SEM-protected 7-azaindole 5a led to compounds 10f−i (Scheme 6). Halogenation and Suzuki cross-coupling with 3-nitrophenyl boronic acid resulted in the coupled nitro derivatives. Slightly

different reaction conditions, according to the respective starting material, were used to perform the Suzuki crosscoupling reactions (see experimental section of materials and intermediates, Supporting Information). The nitro compounds were usually obtained as inseparable mixtures with the respective dehalogenated reactant. The mixture was treated with hydrogen on Pd/C to yield the amines 43a−d, which could be purified via column chromatography. Amide coupling with DIPEA in THF led to the acrylamides/propionylamides (44a−g). Final deprotection with TFA in DCM resulted in the particular irreversible inhibitors or their reversible counterparts (45a−g). 4639

DOI: 10.1021/acs.jmedchem.7b00178 J. Med. Chem. 2017, 60, 4636−4656

Journal of Medicinal Chemistry

Article

Scheme 6. Synthetic Route of EGFR Inhibitors with Michael Acceptor at the Phenyl Ring Attached to the 3-Position of 7Azaindole (Table 3)a

a 1 R −R3 analogous to Table 3. (a) NH4OAc, MeOH, rt (C3HBr2F3O: (1) NH4OAc, H2O, 100 °C; (2) NH4OAc, MeOH, rt); (b) NaH, SEM-Cl, DMF, 0 °C to rt; (c) n-BuLi, Sn(Bu)3Cl, THF, −78 °C to rt (9g: (1) NIS, MeCN, rt; (2) (CH3)2CHMgCl·LiCl, Sn(Bu)3Cl, THF 0 °C to rt); (d) Pd(OAc)2:X-Phos 3:1, dioxane, reflux; (e) NBS, THF, rt (10f: NIS, MeCN 0−50 °C); (f) (1) 3-nitrophenyl boronic acid, K2CO3 or K3PO4, Pd(PPh3)4, DMF or dioxane, H2O, 80 or 100 °C, (2) H2, Pd/C, EA, 1, 3, or 6 bar, rt or 75 °C; (f) acryloyl chloride or propionyl chloride, DIPEA, THF, 0 °C; (g) (1) TFA, DCM, rt, (2) K2CO3(aq) or NaHCO3(aq), MeOH or EtOH, rt.

Biological Evaluation and Rational Design. IC50 values of the inhibitors were determined via a radioisotope filter binding assay (Tables 1−4). The inhibitors were incubated with kinase, substrate, cofactors, and radioisotope-labeled ATP, followed by spotting the mixture onto filter papers that bind the radioisotope-labeled phosphorylated substrate. The Glide induced fit docking tool as well as the covalent docking tool of the Schrödinger software were used for the rational draft and design of our inhibitors. Visualization was performed via PyMOL. The p-fluorophenyl residue appeared to be the structural element of choice for the 4-position of the imidazole. A phenyl ring (11b) in this position showed a slight increase in activity (Table 1). In terms of metabolic stability and higher yields in the Radziszewski imidazole synthesis of the compounds (higher electron withdrawing effect), we insisted on the 4-fluorophenyl residue. Changes in the electron density of the aromatic ring (11a) led to a significant decrease in potency, particularly with respect to the L858R/T790M double mutant (Table 1). The first-generation amino quinazoline EGFR inhibitors (gefitinib, erlotinib) seem to clash with the methionine gatekeeper.8 Crystal structures and molecular modeling of the thirdgeneration aminopyrimidine inhibitors 46 (WZ4002),17 rociletinib,19 and osimertinib33 reveal that these compounds spare the gatekeeper area to avoid a sterical interference. Therefore, the 4-fluorophenyl group, in combination with an imidazole that has a hinge binder motif substituted in the 5position, should be a suitable structure for the clinically challenging EGFR T790M mutant, as it seems to pass the

gatekeeper methionine and reaches the HR I. This is an important additional increase in nonpolar binding affinity. The p-hydroxyphenyl ring in position 2 of the imidazole core is oriented toward a region with a high presence of polar amino acids (PBR). The physiological substrate ATP forms crucial interactions with these polar side chains via its phosphate groups. A crystal structure of ADP bound to EGFR (PDB 4RIW)34 revealed the importance of different polar interactions of the phosphate group with the amino acid side chains Lys721, Asp831, and Asn818 (Figure 2). Docking studies of 11h in the same crystal structure revealed the hydroxypropyl group to be in the same orientation as the phosphate group. Furthermore, docking predicted a polar interaction of the hydroxyl group with Asp831. The importance of these interactions is well-known, and they were targeted, among others, in the optimization of p38α MAPK inhibitors35 and might be transferred to EGFR, too. The exclusion of the hydroxyl group (11i) led to a significant loss of potency (Table 1), thus emphasizing its importance. Different derivatives were synthesized to study the perfect orientation of this OH group. Its migration to the 3-position (11c) led to a slight increase in potency. However, the replacement of the rigid aromatic phenol by an aliphatic alcohol chain increased the potency significantly (11d, 11e, 11h). Furthermore, the phenyl ring in the 2-position of 1H-pyrrolo[2,3-b]pyridine of 3-substituted imidazoles is an important potency increasing pattern for the inhibition of the p38α MAP kinase, especially in combination with a rigidized hinge binding motif found in the 7-azaindole.26 This increase in potency may be based on the occupation of a hydrophobic pocket that is also present in EGFR. Unfortu4640

DOI: 10.1021/acs.jmedchem.7b00178 J. Med. Chem. 2017, 60, 4636−4656

Journal of Medicinal Chemistry

Article

Table 1. Biological Data of the First Set of Reversible Inhibitors and Reference Compoundsf

4641

DOI: 10.1021/acs.jmedchem.7b00178 J. Med. Chem. 2017, 60, 4636−4656

Journal of Medicinal Chemistry

Article

Table 1. continued a

No data. bIC50 value below the lowest chosen inhibitor concentration in the assay. cCompounds published by Selig et al.26 as p38α MAPK inhibitors. dIC50 values and structure published by Günther et al.25 eNo inhibition. fWT = wild-type EGFR kinase; L858R, L858R/T790M, L858R/ T790M/C797S = clinical relevant EGFR mutants.

the clinically challenging triple mutant with IC50 values of 7.6 nM (11d) and 8.4 nM (11e). The FDA-approved osimertinib revealed a 35-fold weaker potency with an IC50 value of 278 nM. The first known third-generation inhibitor 46 demonstrated no significant inhibition of the aforementioned triple mutant (IC50 = 2050 nM), indicating the Achilles’ heel of actual third-generation EGFR inhibitors. The lead structure 11g was further applied as a template to create an irreversible inhibitor. Acrylamide (exception 19b) was used as the warhead template as this functional group has been widely used and studied by means of irreversible kinase inhibitors. Docking studies revealed position 2 of the imidazole as a suitable point to add an acrylamide via a benzyl linker (Figure 3A). The in silico model showed a suitable length and orientation to interact with the cysteine. An increase in potency of compound 19a (Table 2) could be seen compared to the lead structure 11g. The introduction of a 4-dimethylamino crotonic acid amide (19b) led to slightly improved IC50 values against WT and the L858R mutant but not to an increase in potency toward the T790M mutant. Furthermore, the substitution of the 4fluorophenyl residue with a cyclopropyl residue (19c) led to an almost complete loss of inhibitor potency, underlining the importance of the 4-fluorophenyl residue in this position. Although 19a and 19b provided higher potency compared to the lead structure 11g, no substantial increase in potency was recognized. This should be expected in terms of the potential covalent bond formation. The recently published precursor 18a25 revealed even lower IC50 values. Using a flexible alcohol chain in the 2-position of the imidazoles increased the potency substantially. Polar amino acid side chains of the PBS may “pull away” the acrylamide warhead via noncovalent interactions, thus preventing an optimal orientation toward the cysteine. Hence, we placed the acrylamide warhead to the phenyl ring attached to the 7-azaindole in the 2-position (Figure 3B). Unfortunately, compounds 27a and 33 displayed almost the same IC50 values compared to the lead structure and were

Figure 2. Molecular docking of 11h (light-gray sticks) in a crystal structure of ADP bound to EGFR (PDB 4RIW).34 ADP (black lines), covalent interactions (gray dots), and magnesium ion (cyan sphere) are adapted from the crystal structure. Polar interactions of 11h with hinge region and Asp831 (red dots) are postulated via Glide docking software.

nately, the exclusion of this phenyl ring (11f) led to a striking decrease in potency on EGFR (>130-fold for the double mutant). Further exchange of the hydroxypropyl group by an apolar phenyl ring in the 2-position of the imidazole (11j) resulted in a complete loss of potency toward the double/triple mutant. The trisubstituted imidazole 11h was recently published25 as a very potent reversible EGFR inhibitor with high potency against both clinically relevant mutations (IC50 (L858R) = 90% of the compound was seen after 1 h. Metabolic stability measurements in human liver microsomes revealed good stability. After 4 h of incubation, 87% of unmetabolized 45a was detected (stability graph and detected metabolites: see Supporting Information, Figures S1 and S2). While the reversible inhibitors 11d (7 mg/L) and 11e (not detectable) were almost insoluble in the kinase assay buffer (pH = 7.5), the irreversible compound 45a revealed good solubility 4645

DOI: 10.1021/acs.jmedchem.7b00178 J. Med. Chem. 2017, 60, 4636−4656

Journal of Medicinal Chemistry

Article

Figure 4. (A) Covalent docking of 45a in EGFR T790M kinase active site (PDB 2JIU). Predicted hydrogen bonds are indicated as red dots. (B) 2D schematic view of docked 3D structure.

Table 4. Crucial Structural Elements of Trisubstituted Imidazoles with a 7-Azaindole Hinge Binding Motif as Potent EGFR Inhibitors

a

High potency against clinically challenging EGFR T790M/L858R and T790M/L858R/C797S mutant. bHigh potency against clinically challenging EGFR T790M/L858R mutant.

(299 mg/L) as determined by HPLC with UV detection (see Supporting Information). The prior published inhibitor 11h as the most potent representative of the reversible inhibitors in the EGFR enzyme assay revealed no potency on EGFR cancer cell lines.25 However, the irreversible inhibitor 45a showed activity on the L858R/T790M cell line (739 nM) and a >40-fold selectivity for the double mutant compared to the WT cell line (30 μM) (see Supporting Information, Table S1). The observed selectivity in the cancer cell lines compared to the enzymatic IC50 data might be based on an “oncogen addiction”38 of the mutant cancer cell lines to the EGFR pathway. The EGFR WT cells may compensate the inhibition of this pathway via alternative ones. Furthermore, kinetic measurements were applied to examine the characteristics of irreversible binding of compound 45a (Table 5). The ability of the inhibitor to compete with an immobilized, active-site directed ligand was measured. The inhibitor showed Kd values in the low nM range, which confirmed the high

Table 5. Kd values of Compound 45a towards Different EGFR Mutants As Measured in Two Different Study Arms (B and C)a EGFR mutant

Kd [nM] without dilution (C)b

Kd [nM] with dilution (B)c

WT L858R L858R/T790M

2 0.93 13

3.4 1.8 14

a

WT = wild-type EGFR kinase; L858R, L858R/T790M = clinical relevant EGFR mutants. bCompound and kinase were combined and equilibrated for 1 h. ccompound and kinase were combined and equilibrated for 1 h, then diluted 30-fold in reaction buffer and equilibrated for 5 h.

binding affinity toward the ATP binding site. 45a effectively bound to the WT as well as to L858R and L858R/T790M. Obviously, the compound showed no significant differences in the Kd values without and after a 30-fold dilution, affirming an irreversible binding character. A recently published reversible 4646

DOI: 10.1021/acs.jmedchem.7b00178 J. Med. Chem. 2017, 60, 4636−4656

Journal of Medicinal Chemistry

Article

to be a very kinome selective compound with a selectivity score of 0.063 (Table 8).41

structurally related inhibitor revealed a massive shift to >10-fold higher Kd values after 30-fold dilution,25 manifesting the typical behavior of an inhibitor with noncovalent binding properties. Another nondiluted study arm (see Supporting Information, Table S2) with an equilibration time of overall 6 h show even 3−5-fold lower Kd values of 45a for all three kinases. This result may reflect the slow association kinetics of this compound, indicating that equilibrium had not been attained, after 1 h of incubation. Compound 45a was tested in a kinome selectivity screen against a set of 410 kinases (397 protein kinases and 13 lipid kinases). Beside the expected extensive potency in the ErbB family (Table 6), only 13 other kinases were addressed (Table 7).

Table 8. IC50 Values of the Most Potent EGFR Compounds in a p38α MAPK Assay compd 11a 11b 11c 11d 11e 11f 11gb 11h 11ib 11jb 45a

Table 6. EGFR Profile of 45a in a Panel of 13 EGFR Mutants at an Inhibitor Concentration of 1 μM EGFR family

residual activity (% of control)

EGFR d746-750 EGFR d747-749/A750P EGFR d747-752/P753S EGFR d752-759 EGFR G719C EGFR G719S EGFR L858R EGFR L861Q EGFR T790M EGFR T790M/L858R EGFR WT ERBB2 ERBB4

3 6 4 6 37 22 6 3 12 21 5 17 4

residual activity (% of control)a

BMX (Cys) BTK (Cys) CK1-δ JNK2 (Cys) JNK3 (Cys) MAP4K4 MAP4K5 MAPKAPK2 (Cys) MINK1 MST4 STK25 (Cys) TEC (Cys) TXK (Cys)

17 17 47 37 34 1 2 48 5 47 49 33 3

0.288 0.001 0.007 0.004 0.002 0.012 0.003 0.002 0.016 0.108 0.214

± ± ± ± ± ± ± ± ± ± ±

0.0283 0.0001 0.0006 0.0003 0.0002 0.0017 0.003 0.0002 0.0038 0.0012 0.0097

Mean ± SEM of three experiments. bData of compounds published by Selig et al.26

a

The reversible compounds 11d, 11e, and 11h showed no decrease in potency regarding p38α MAPK. Compound 11f is still a very potent p38α MAPK inhibitor despite of the lack of the HR II targeting phenyl ring. The loss of this interaction highly decreases the potency on EGFR double and triple mutant (IC50 = 893, 943 nM, Table 1). These data confirm the necessity of a hydrogen bond donating/accepting group in the 2-position of the imidazole plus a phenyl ring targeting the HR II to inhibit the EGFR double/triple mutant significantly. Nevertheless, this dual inhibition of the reversible compounds may have even positive effects because p38 MAPK is also known to be aberrantly triggered in lung cancer tissue.42,43 In contrast, compound 45a revealed an IC50 of 213 nM in an p38α MAPK assay, making it 140-fold less potent on this kinase compared to EGFR L858R/T790M (IC50 = 1.52 nM). Direct comparison with the lead compound 11g (IC50 value on p38 MAPK = 3 nM, IC50 value on EGFR T790M/L858R = 916 nM) exposed a successful switch in potency from p38 to EGFR. With respect to the potential reactivity of the acrylamide warhead, all final compounds were checked for structural elements which are present in pan-assay interference compounds (PAINS) using the latest version of ZINC15 database.44 No PAIN hits were found and good selectivity, metabolic stability, as well as a low GSH reaction rate were detected. Thus, no unpredictable reactions with random Cys side chains were found.

Table 7. Selectivity Screen of 45a in a Panel of 410 Kinases at an Inhibitor Concentration of 1 μM kinase name

IC50 [μM]a



CONCLUSION On the basis of an off-target hit of a p38α MAPK inhibitor and an iterative process of molecular docking, synthesis, and biological testing, we developed structurally novel EGFR inhibitors. Effective synthetic routes were evolved to obtain more flexible synthetic procedures that allow the modification at different parts of the molecule and tolerate less robust functional groups. Hence, a set of different trisubstituted imidazoles with a 7-azaindole hinge binding motif were synthesized to study the SARs of this class of compounds against different EGFR mutants. We revealed the possibility to address the HR I behind the Met790 gatekeeper with a combination of a 3-subsituted imidazole with a 7-azaindole hinge binding motif. By determining the crucial residues for high affinity binding to the active site, the development of the

a

Only hits with residual activity of