Vinylpyridine as a Tunable Covalent Warhead Targeting C797 in EGFR

Abstract

To further facilitate the discovery of cysteine reactive covalent inhibitors, there is a need to develop new reactive groups beyond the traditional acrylamide-type warheads. Herein we describe the design and synthesis of covalent EGFR inhibitors that use vinylpyridine as the reactive group. Structure-based design identified the quinazoline-containing vinylpyridine 6 as a starting point. Further modifications focused on reducing reactivity resulted in substituted vinyl compound 12, which shows high EGFR potency and good kinase selectivity, as well as significantly reduced reactivity compared to the starting compound 6, confirming that vinylpyridines can be applied as an alternative cysteine reactive warhead with tunable reactivity.

Targeted covalent inhibitors (TCI) have had a significant impact in drug discovery over the last 20 years.¹ Some of the most successful examples to showcase this include the approved kinase inhibitors: afatinib,² osimertinib,³ and ibrutinib⁴ which all form a covalent bond with a noncatalytic cysteine in the ATP binding site. Recently, the nonkinase inhibitor sotorasib was approved and this has further established the importance of TCIs as the covalent mode of action allowed targeting KRAS G12C, a target previously viewed as undruggable.⁵

Despite the success of cysteine targeting covalent inhibitors the majority of approved and emerging inhibitors utilize acrylamide or structurally related Michael acceptors, such as butynamides, to react with the targeted cysteine, and there are relatively few examples of inhibitors that contain other types of reactive motifs.⁶ Considering the importance and potential of TCIs, there is a need to identify novel warheads with tunable reactivity that can present vectors and properties not accessible by the traditional acrylamide-type motif. This has been a growing research area,⁷ and recent noteworthy alternatives targeting noncatalytic cysteines include cyanamide,⁸ alkynylbenzoxazine,⁹ alkynyl-substituted heteroarenes,¹⁰⁻¹² and 2-chloropyridine.¹³ In this study, we report the use of vinylpyridine (VP) as a cysteine reactive group with tunable reactivity that has been applied to obtain potent and selective covalent epidermal growth factor receptor (EGFR) inhibitors.

Both 2- and 4-VP regioisomers have been shown to undergo conjugate addition with a range of nucleophiles, and the reactivity of VPs toward cysteine is well-known and has been utilized for labeling and bioconjugation of cysteines in proteins.¹⁴ Keeley et al. included 2- and 4-VP fragments in their study focused on characterization of a cysteine-targeted covalent library and showed that unsubstituted 2- and 4-VPs reacted rapidly with glutathione.¹⁵ There are also a few examples in the literature where vinyl-substituted heteroarenes have been applied as warheads to target cysteines.^12,16 However, we have not been able to identify examples where VP or related vinyl-substituted heteroarenes have been shown to react selectively with a targeted cysteine and where reactivity has been tuned to such a level that this becomes relevant for use in drug design. Considering that heteroaromatic rings such as pyridine are a common motif in small molecule drugs, the VP could present a versatile option as a covalent warhead, especially in cases where the targeted cysteine is close to an aromatic ring in a reversible ligand, as the reactive β-carbon in the vinyl is closer to the aromatic motif than the corresponding reactive carbon in an acrylamide.

Thus, the VP could offer benefits when there is not enough space for an acrylamide or the higher polar surface area of the acrylamide is less compatible with the target protein or target profile, for example, projects where high CNS exposure is required. For evaluating the use of VP as a covalent warhead, we wanted to identify a target containing a cysteine with known reactivity and where a chemotype with a suitable vector toward the cysteine could be identified. One target that has been central in the field of TCIs over the past decade is EGFR. The quinazoline-containing EGFR inhibitor afatinib 1 was one of the first approved covalent kinase inhibitors in 2013 (Figure 1). This second-generation EGFR inhibitor covalently modifies cysteine 797 situated in the solvent channel of the EGFR ATP-binding site. The quinazoline core is present in the majority of first- and second-generation EGFR inhibitors and is also present in poziotinib 2 (Figure 1). The binding mode of quinazoline-type compounds in EGFR is well characterized and the majority of published covalent inhibitors extend the reactive motif toward cysteine 797 from the 6-position of the quinazoline core. Further examining the X-ray structure of afatininb (PDB: 4g5j)¹⁷ suggested that introduction of a VP could be accommodated by extending from a heteroatom at the 6-position of the quinazoline (Figure 2A). Docking of compound 6 extending the VP from an ether linkage looked promising as the reactive vinyl could be placed in close proximity to cysteine 797 (Figure 2B).¹⁸ This together with the known reactivity of cysteine 797 suggested that the quinazoline core would be a good place to start for evaluating VP as a covalent warhead. The quinazoline-type EGFR inhibitors are often very potent inhibitors of wild type (WT) EGFR. This has not been the desired selectivity profile for later generations of EGFR inhibitors which mainly target activating mutants of EGFR and selectivity over WT is desired, as it can lead to dose-limiting toxicities. We did not see this as an issue in this initial work, which mainly focused on evaluating VP as a covalent warhead; throughout this work, compounds were profiled as EGFR WT inhibitors.

Figure 1.

Quinazoline-based irreversible covalent EGFR inhibitors.

Figure 2.

(A) Crystal structure of afatinib binding to EGFR (pdb: 4g5j).¹⁷ (B) Compound 6 docked in EGFR crystal structure showing a similar binding mode as afatinib. The docking has been generated using a minimization of compound 6 in the rigid pocket of pdb: 4g5j.¹⁸

Our efforts started with introducing the unsubstituted regioisomeric 4- and 2-VPs on the 3-chloro-2-fluoroaniline substituted quinazoline core giving compounds 4–6. Gratifyingly, all three compounds showed good potency in the EGFR enzyme assay and potency was in a region similar to reference covalent inhibitors 1 and 2 (Table 1). 2-VP 6 was the most potent in the cellular assay and was, therefore, further profiled in a glutathione assay to evaluate the reactivity of the vinyl warhead. In the presence of glutathione, compound 6 had a half-life of 22 min. To put this in context, the clinical covalent inhibitors 1 and 2 had half-lives of 34 and 938 min, respectively, so thus reactivity of the unsubstituted vinyl was considered to be high. The risk with highly reactive compounds is that these can react unselectively and bind to cysteines in other proteins which could potentially give rise to idiosyncratic toxicity. Therefore, our focus was to evaluate if reactivity of the 2-VP could be reduced while maintaining high on-target potency.

Table 1. Structure, EGFR Enzyme and Cell Inhibitory Potencies, and Pharmacokinetically Relevant Properties^a.

^aAll biological values are the mean of ≥3 replicates except ^‡n = 2 and ^†n = 1.

^bEGFR WT biochemical assay using ADP-Glo assay technology performed with an ATP concentration of 12 μM.

^cEGFR WT phosphorylation cell assay using NCI-H2073 cells.

^dIntrinsic clearance in rat hepatocytes ((μL/min)/10⁶ cells).

^eHuman microsomal Clint (μL/min/mg).

^fSelected examples profiled in reactivity assay measuring half-life (minutes) in the presence of 4.6 mM glutathione in 0.1 M PBS at pH 7.4 and 37 °C.

^gFormation of glutathione adducts not observed

There are different tactics that one could imagine applying to tune the reactivity; one option could be to make changes to the pyridine ring to modulate electronic or steric properties. Alternatively, further substituting the vinyl moiety to change its steric/electronic properties would also be expected to reduce the reactivity. The latter option seemed most attractive in this case as we envisioned that changes to the vinyl group would have a greater effect on reducing reactivity, and these analogues were also assumed to be more synthetically accessible. Compounds 7–12 with further substitution on the vinyl moiety were prepared and screened. All of these analogues showed high potency in the EGFR enzyme and cell assay. Many of these analogues were quite lipophilic with log Ds > 4 and as a result had low solubility. However, compound 12 containing a basic amine had a lower log D and showed high solubility and lower in vitro Cl_ints values than the other analogues and was selected for further profiling in the glutathione assay. The data showed that reactivity had been considerably reduced compared to the unsubstituted VP 6 as compound 12 had a half-life of 404 min which was considered to be in a reasonable region. Overall, the properties of this compound looked attractive and were comparable to those of the covalent clinical inhibitors 1–3; VP 12 was thus selected for further profiling. It is worth to note that the methyl-substituted analogue 7 also showed a significant drop in reactivity with a half-life >1000 min; for the dimethyl substituted compound 9, no reactivity with glutathione could be observed in this assay (Table 1).

Compound 12 was profiled further in another enzymatic assay format at low (10 μM) and high (1 mM) ATP concentration, which corresponds to 1 × K_m and 100 × K_m ATP, respectively. Reversible ATP-competitive inhibitors are expected to show a 50-fold increase in IC₅₀ value going from 1 × K_m to 100 × K_m ATP concentration. A modest shift in IC₅₀ value was observed for VP 12 and for the covalent acrylamide inhibitors 1 and 2 (Table 2). The reversible inhibitor gefitinib 3 showed a more significant 10-fold drop. While reversible ATP-competitive tight-binding inhibitors such as gefitinib (Table 2) can also show less than the theoretical drop in potency at the high ATP concentration, this behavior is more often observed with covalent inhibitors, particularly when the covalent step is allowed to proceed to near completion.

Table 2. Further Profiling of Compounds 1–3 and 12.

compound	EGFR enz. ATP conc: 10 μM pIC50	EGFR enz. ATP conc: 1000 μM pIC50
1 (afatinib)	9.7	9.3
2 (poziotinib)	9.6	9.4
3 (gefitinib)	9.6	8.6
12	9.6	9.4

Covalent inhibitors normally exhibit time-dependent inhibition, that is, the progress curves deviate from linearity and slow down over time. This was indeed observed for compound 6 (Figure 3) and compounds 1, 2, 7, and 12 (Figure S1). The inhibition constants, k_inact and K_i, can normally be obtained from analysis of the progress curves as described.¹⁹ However, the high potency of the compounds in our study precluded these values from being determined from the data due to tight-binding considerations.

Figure 3.

Progress curves of WT-EGFR inhibition by compound 6 using PhosphoSens technology.

To unambiguously confirm that VPs 6 and 12 are covalent inhibitors, mass spectrometry (MS) data was generated. Compound 1, VPs 6 and 12, and also reversible inhibitor 3 were incubated together with EGFR WT protein for 1 h before analysis by MS. These data clearly demonstrate a covalent mode of action for both the positive control 1 and VPs 6 and 12, while the reversible inhibitor 3 included as a negative control did not show any modification of the protein (Figure 4). The mass shift observed shows a major new peak that suggests a single cysteine modification, i.e., the entire protein population converted to a covalent 1:1 complex suggesting that the compounds have mainly reacted with one cysteine in the protein.

Figure 4.

MS analysis of intact protein alone (red spectrum) or upon exposure to compound (blue spectrum) demonstrate covalent modification for compounds 1 (positive control), VP 6, and VP 12, but not for reversible inhibitor 3 (negative control).

Encouraged by these data, VPs 6 and 12 together with acrylamide inhibitors 1 and 2 were further profiled at 1 μM in a panel of 127 kinases (ThermoFisher). Interestingly, VPs 6 and 12 inhibited fewer kinases in the panel than 1 and 2 (Figure 5). Top hits (>75% inhibition at 1 μM) in the kinase panel for VPs 6 and 12 were mainly EGFR and EGFR mutants (Table S1 in Supporting Information). VP 6 also inhibited the kinase BLK which contains a cysteine analogous to cysteine 797 in EGFR. VP 12 showed slightly lower activity vs BLK (70 vs 90% respectively; see Table S1). It was noticeable that VPs 6 and 12 did not show high activity vs BMX or BTK, which also contain a cysteine in this position, while these kinases were among the top hits for acrylamide 2.

Figure 5.

Kinase selectivity of reference acrylamide containing compounds 1 (A), 2 (B), and VPs 6 (C) and 12 (D) all profiled at a concentration of 1 μM against a panel of kinases (ThermoFisher). Each circle represents a kinase, and the colors show percent inhibition: red > 75%, orange > 50%, and gray < 50%. Only kinases with more than 50% inhibition are annotated with the name of the kinase. Kinases represented by a triangle contain a cysteine in the equivalent position of cys797 in EGFR. Further details can be found in Table S1 in the Supporting Information. The kinome tree was generated using KinMap,²⁰ and the illustration was reproduced courtesy of Cell Signaling Technology, Inc. (www.cellsignaling.com).

Compound 12 was prepared in two steps starting from commercially available quinazoline intermediate 13. The 2-bromopyridine was introduced by reacting 13 with 2-bromo-4-fluoropyridine in a nucleophilic aromatic substitution reaction. The obtained intermediate 14 was converted to the desired VP 12 by a one-pot Suzuki/substitution reaction using dimethylamine combined with (E)-(3-chloroprop-1-en-1-yl)boronic acid (Scheme 1).

Scheme 1. Synthesis of Representative Example VP 12^a.

Reagents and conditions: (a) Cs₂CO₃, 2-bromo-4-fluoropyridine, NMP, 80 °C, 57%; (b) Cs₂CO₃, dimethylamine, (E)-(3-chloroprop-1-en-1-yl)boronic acid, PdCl₂(dppf), DMSO, 90 °C, 21%.

In summary, VP has been applied as a covalent warhead to target cysteine 797 in EGFR. Initial leads were obtained by structure based design. Compound 6 showed high EGFR potency in biochemical and cellular assays but this unsubstituted 2-VP showed very high reactivity toward glutathione with a half-life of 22 min. Design was focused on reducing the reactivity by further substitution of the vinyl motif. Compound 12 containing a N,N- dimethylaminomethane substituent on the vinyl motif combined high potency improved physicochemical properties and lower reactivity as the half-life had been increased to 404 min in the glutathione assay. Covalent mode of action was confirmed for both VPs 6 and 12 by incubating with WT EGFR protein and detecting adducts via MS. In addition, both compounds showed good kinase selectivity as they mainly hit EGFR and EGFR mutants in a panel consisting of 127 kinases. Overall these data show that VP can be an interesting alternative to the acrylamide type warheads. This initial work focused on pyridines but it is very likely that the same strategy of using activated vinyls can be extended to other 5- and 6-membered heteroarenes, such as pyrimidine, pyridazine, thiazole, and oxazole. Furthermore, the vinyl motifs are readily available from halogenated intermediates using transition metal-catalyzed reaction. This suggests that reactive vinyls can be a valuable addition to the toolbox for drug designers targeting cysteine with covalent inhibitors.

Experimental Procedures

The purity of compounds for biological testing were assessed by HPLC to be >95%. The synthetic protocol for compound 12 is included below; protocols for all other examples are included in the SI. No unexpected or unusually high safety hazards were encountered.

6-((2-Bromopyridin-4-yl)oxy)-N-(3-chloro-2-fluorophenyl)-7-methoxyquinazolin-4-amine (14)

Cs₂CO₃ (5.10 g, 15.64 mmol) and 2-bromo-4-fluoropyridine (0.713 mL, 6.88 mmol) were added to a solution of 4-((3-chloro-2-fluorophenyl)amino)-7-methoxyquinazolin-6-ol (2 g, 6.26 mmol) 13 in NMP (40 mL) at room temperature. The resulting mixture was heated to 80 °C. After 18 h, the reaction mixture was diluted water and DCM. The desired compound was obtained by filtration and dried under vacuum to afford the title compound (1.7 g, 57%) as an off-white solid. ¹H NMR (500 MHz, DMSO-d₆): 3.92 (s, 3H), 7.04 (dd, J = 5.7, 2.3 Hz, 1H), 7.23–7.29 (m, 2H), 7.42–7.52 (m, 3H), 8.27 (d, J = 5.7 Hz, 1H), 8.33 (s, 1H), 8.48 (s, 1H), 9.79 (s, 1H). MS m/z: calcd for C₂₀H₁₃BrClFN₄O₂ [M + H]⁺, 475.0; found, 475.0/477.0.

(E)-N-(3-Chloro-2-fluorophenyl)-6-((2-(3-(dimethylamino)prop-1-en-1-yl)pyridin-4-yl)oxy)-7-methoxyquinazolin-4-amine (12)

Cs₂CO₃ (137 mg, 0.42 mmol), dimethylamine (0.42 mmol), 14 (100 mg, 0.21 mmol), (E)-(3-chloroprop-1-en-1-yl)boronic acid (50.6 mg, 0.42 mmol), and PdCl₂(dppf) (15 mg, 0.02 mmol) were heated in DMSO (2 mL) at 90 °C for 5 h. The reaction mixture was allowed to attain room temperature. EtOAc and water were added, the organic phase was separated, and the aqueous phase was extracted with DCM. The combined organic layers were filtered through a phase separator and concentrated. The residue was purified by preparative HPLC on a Kromasil C8 column (10 μm 250 × 20 ID mm) using a gradient of 15–55% acetonitrile in H₂O/ACN/FA 95/5/0.2 buffer to afford the formic acid salt of the title compound (21 mg, 21%) as a white solid. ¹H NMR (500 MHz, DMSO-d₆): 2.25 (s, 6H), 3.17 (d, J = 6.5 Hz, 2H), 3.91 (s, 3H), 6.62 (d, J = 15.7 Hz, 1H), 6.74 (dt, J = 6.5, 15.7 Hz, 1H), 6.80 (dd, J = 2.4, 5.6 Hz, 1H), 7.03 (d, J = 2.4 Hz, 1H), 7.26 (td, J = 1.0, 8.1 Hz, 1H), 7.43–7.53 (m, 3H), 8.23 (s, 1H) formic acid proton, 8.32 (s, 1H), 8.40 (d, J = 5.6 Hz, 1H), 8.50 (s, 1H), 9.79 (s, 1H). ¹³C NMR (126 MHz, DMSO) δ 164.2, 158.1, 157.4, 156.5, 155.4, 152.6, 151.6, 149.9, 141.3, 133.1, 131.3, 128.4, 127.9, 127.4, 125.4, 120.8, 120.1, 116.2, 110.0, 109.6, 109.3, 109.3, 60.1, 55.3, 45.5. HRMS m/z: calcd for C₂₅H₂₃ClFN₅O₂ [M + H]⁺, 480.1602; found, 480.1603.

Enzyme Screening Assay

WT EGFR Biochemical Assay

The inhibitory activity of compounds against WT EGFR was determined using an ADP-Glo kinase assay (Promega V9102). The assay monitors the phosphorylation activity of the recombinant N-terminal GST-tagged EGFR cytoplasmic domain (residues 669–1210; Carna, 08–115) acting on the poly(Glu, Tyr) peptide (4:1 random copolymer, Sigma81357). Compounds were serially diluted in 100% (v/v) DMSO, before being acoustically dispensed from an Echo 555 (Labcyte) into white 384-well assay plates (Greiner 784075). 2 μL of the enzyme reaction [10 nM WT EGFR in kinase base buffer (20 mM HEPES pH 7.5, 5 mM MgCl2, 0.01% Brij-35, 1 mM DTT, 100 mM NaCl)] was preincubated with the compound for 15 min. 2 μL of the substrate reaction [12 μM ATP, 50 μg/mL poly(Glu, Tyr) in kinase base buffer] was added, giving a reaction volume of 4 μL. Following 50 min of incubation at room temperature, the reaction was stopped with 2 μL of the ADP-Glo reagent and incubated for 40 min. 4 μL of the ADP-Glo kinase detection reagent was added and incubated for 30 min. Luminescence was detected by using a PHERAstar FS instrument (BMG).

Cellular Assay

WT EGFR Phosphorylation Cell Assay

The assay is using NCI-H2073 (H2073) obtained from American Type Culture Collection (ATCC). Cells were cultured in RPMI-1640 (Sigma-Aldrich) supplemented with 1% Glutamax and 10% fetal calf serum (FCS). EGFR phosphorylation was measured using a modified Pan phospho-EGFR Cellular Assay Kit (Cisbio). NCI H2073 cells were detached using TrypLE Express Enzyme (Life Technologies) from culture flasks and resuspended in RPMI supplemented with 1% Glutamax and 3% charcoal stripped FCS. 5 μL of cell suspension at a density of 600 cells per well were dispensed into a low volume 384 proxi plates (Greiner) predosed with titrations of test compound and incubated for 2 h at room temperature. The NCI H2073 cells were stimulated for the final 10 min with 20 ng/mL of EGF. Following the 2 h incubation, 2 μL of the XL665 and Cryptate-labeled antibodies diluted in lysis buffer were added to the cells and incubated for 2 h at room temperature. Plates were then read on a Pherastar with a HTRF module.

Glossary

Abbreviations

EGFR

epidermal growth factor

TCI

targeted covalent inhibitor

VP

vinylpyridine

WT

wild type

MS

mass spectrometry

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.3c00425.

• Preparation and characterization for compounds 4–17; Biochemical kinase selectivity data for compounds 1, 2, 6, and 12; Method protocols for MS and kinetic profiling (PDF)

The authors declare no competing financial interest.

Special Issue

Published as part of ACS Medicinal Chemistry Lettersvirtual special issue “Exploring Covalent Modulators in Drug Discovery and Chemical Biology”.