Synthesis and Structural Characterization of a Monocarboxylic Inhibitor for GRB2 SH2 Domain

Abstract

A monocarboxylic inhibitor was designed and synthesized to disrupt the protein-protein interaction (PPI) between GRB2 and phosphotyrosine-containing proteins. Biochemical characterizations show compound 7 binds with the Src homology 2 (SH2) domain of GRB2 and is more potent than EGFR₁₀₆₈ phosphopeptide 14-mer. X-ray crystallographic studies demonstrate compound 7 occupies the GRB2 binding site for phosphotyrosine-containing sequences and reveal key structural features for GRB2–inhibitor binding. This compound with a −1 formal charge offers a new direction for structural optimization to generate cell-permeable inhibitors for this key protein target of the aberrant Ras-MAPK signaling cascade.

Graphical Abstract

Growth factor receptor-bound protein 2 (GRB2) is an adaptor protein that mediates activation of mitogenic Ras signaling pathways.^1-2 After growth factor stimulation, phosphotyrosine (pTyr)-containing sequences of active receptor tyrosine kinases (RTKs) bind to the central Src homology 2 (SH2) domain of GRB2, and the N- and C-terminal SH3 domains of GRB2 bring the nucleotide exchange factor SOS to the cell membrane to activate the Ras–mitogen-activated protein kinase (MAPK) signaling cascade for cell growth and differentiation. Aberrant GRB2-dependent Ras activation significantly contributes to cancer development and progression.^1-3 The SH2 domain of GRB2 can also bind with pTyr-containing sequences of adaptor proteins, such as the Src homologous and collagen (SHC) protein, and non-receptor tyrosine kinases, such as BCR-ABL, to mediate aberrant Ras–MAPK signaling.⁴ Hence, GRB2 is a key, convergent MAPK signaling node and an interesting protein target for anticancer drug development.

X-ray crystallographic^5-6 and NMR⁷ studies indicated that pTyr-containing sequences pYXNX (pY, X, and N represent phosphotyrosine, any residue, and asparagine, respectively) adopt a type I β-turn conformation to bind to GRB2 SH2 domain. This conformation is different from the extended conformation of pTyr-containing sequences observed in many other SH2 domain-containing proteins.^8-9 Highly appreciable medicinal chemistry efforts have been made to discover inhibitors that bind to GRB2 SH2 domain and disrupt the protein–protein interaction (PPI) between GRB2 and pYXNX-containing proteins.^10-14 Some interesting compounds are listed in Fig. 1. Compound 1 was extracted from an EGFR pTyr-containing sequence. This compound inhibited EGFR/GRB2 PPI with a half maximal inhibitory concentration (IC₅₀) of 8.64 μM.¹⁵ Compound 2 was evolved from 1 and exhibited an IC₅₀ two orders of magnitude lower than 1.¹⁶ The cyclization of 2 to mimic the β-turn binding conformation of pTyr-containing sequences with GRB2 led to 3.¹⁷ The introduction of a carboxymethyl group to 3 offered 4 with an IC₅₀ of 2 nM.¹⁸ The crystal and NMR structures of GRB2 SH2 domain in complexes with these inhibitors were reported.^5-6,19-30 However, despite excellent in vitro biochemical activities, these compounds showed overall poor cellular activities, although cell-based data were collected with some cell lines.^17-18,31-34 The phosphate and phosphonomethyl groups in these compounds carrying −2 charges at physiological pH were thought to cause poor cell permeability.^9-11,35-36 Phosphate mimicking bioisosteres were introduced to 2–4, and compounds 5 and 6 were reported as potent GRB2 inhibitors in biochemical assays.^31,37-43Again, studies indicated 5 and 6 displayed unsatisfying cell-based and in vivo activities for further development.^{31,38,40-41,44-47} The latter medicinal chemistry studies were switched to understand molecular recognition questions using GRB2 SH2 domain and inhibitors as a model system, and important knowledge was obtained about ligand conformational constraints, nonpolar surface area burial, and cation-π interaction.^27-30,48-49

Fig. 1.

(A) Reported GRB2 inhibitors. (B) Newly designed GRB2 inhibitors.

We decided to substitute the phosphonomethyl group in 3 with a carboxylic acid group and designed compound 7 in Fig. 1B to examine the binding affinity difference between 3, 7, and EGFR₁₀₆₈-containing phosphotyrosyl peptide. Out of all EGFR sequences, the EGFR₁₀₆₈-containing phosphopeptide is known to have the highest binding affinity with GRB2. The methyl ester 8 in Fig. 1B was also synthesized for comparison.

To synthesize 7 and 8, the commercially available Evans’ chiral auxiliary, (S)-(+)-4-phenyl-2-oxazolidinone, was first attempted to synthesize 12 in Scheme 1A. However, two diastereomers of 12 with a diastereometric ratio (dr) of 3:1 were inseparable by column chromatography. We then screened other Evans’ chiral auxiliaries, including (S)-4-benzyl-2-oxazolidinone, (S)-(−)-4-isopropyl-2-oxazolidinone, (S)-(−)-4-benzyl-5,5-dimethyl-2-oxazolidinone, and (S)-4-tert-butyl-2-oxazolidinone. The chiral auxiliary (S)-4-tert-butyl-2-oxazolidinone (14) offered the highest diastereomer selectivity for the desired product (dr = 100:6). The synthetic route for 7 is shown in Scheme 1B. The reaction between acrylic acid and acryloyl chloride gave acrylic anhydride 13, which was used to acylate 14 under the LiCl and Et₃N condition to afford 15 in 85% yield over two steps. The Heck reaction⁵⁰ between 9 and 15 gave 16 in 76% yield, which underwent 1,4-addition with vinylmagnesium bromide under the PhSCu condition⁵¹ to afford 17 with reproducible yield and high diastereoselectivity. The Evans’ chiral auxiliary in 17 was removed by hydrolysis to give an acid, which was then coupled with 18 to give 19. Key intermediate 18 was synthesized by a route slightly modified based on that reported previously⁵¹, as shown in Supplementary Scheme 1. Ruthenium-catalyzed ring-closing metathesis of 19 using the second generation of Grubbs catalyst and deprotection of the tert-butyl group under the acidic condition offered the final product 7 as a single E isomer. The configuration of the alkene was determined to be trans by ¹H NMR which showed the coupling constant (J) of 15 J Hz for two vicinal alkene protons.

Scheme 1.

The synthetic route for compounds 7 and 8.

Compound 8 was synthesized by esterification of 7 using CH₃I under the K₂CO₃/DMF condition. The positive control compound 3 was synthesized by following the literature procedure (Supplementary Scheme S2).⁵¹ The only difference was that (S)-4-tert-butyl-2-oxazolidinone was used to increase the diastereoselectivity of the 1,4-addition reaction with vinylmagnesium bromide.

Following the previously reported GRB2 fluorescence polarization (FP) assay,⁵² we overexpressed N-terminally His₆-tagged human full-length GRB2 (residues 1–217) in E. coll and purified GRB2. It is known that GRB2 is preferred to bind with pY₁₀₆₈ of EGFR.⁵³ N-terminally fluorescein (FITC)-labelled human EGFR₁₀₆₈ phosphopeptide, FITC-Ahx-PVPEpYINQSVPKRK-NH₂, (Ahx: 6-aminohexanoic acid) was synthesized and purified (HPLC purity >95%). The dissociation constant (K_D) of the EGFR/GRB2 PPI from fluorescence anisotropy binding experiments was around 191 nM, which is consistent with reported K_D, as shown in Fig. 2A.^54-55 The fluorescence anisotropy competitive inhibition assay was then used to evaluate the inhibitory activities of 3, 7 and 8.

Fig. 2.

(A) Fluorescence anisotropy binding experiments to determine the K_D of full-length human GRB2 with a fluorescently labeled EGFR₁₀₆₈ phosphopeptide, FITC-Ahx-PVPEpYINQSVPKRK-NH₂. The data was expressed as mean ± standard deviation (n = 3). (B) fluorescence anisotropy competitive inhibition assays to determine IC₅₀s of 3, 7, 8, and EGFR₁₀₆₈ phosphopeptide for disruption of the interaction between full-length GRB2 and FITC-Ahx-PVPEpYINQSVPKRK-NH₂. The K_i values were derived. The data was expressed as mean ± standard deviation (n = 3). (C) Thermal stabilization of GRB2 SH2 domain by 3 and 7. The melting point (T_m) of the SH2 domain was determined in the absence and presence of inhibitors by DSF.

As shown in Fig. 2B, the inhibition constant (K_i) of 7 for disruption of the EGFR/GRB2 PPI is 140 nM. This compound was 2-fold more potent than the EGFR₁₀₆₈ phosphopeptide 14-mer that displayed a K_i of 400 nM in the parallel assays, offering an exciting starting point for further inhibitor optimization. It is noted that the positive control, compound 3, displayed a K_i of 16 nM. On the other hand, the K_i of the methyl ester of 7, compound 8, was >20 μM, indicating the importance of the carboxylic acid group of 7 for the inhibitory activity. Assessment of binding potential by differential scanning fluorimetry (DSF) showed a significant increase in thermostability of GRB2 in the presence of compound 3 (ΔT_m = 16.5 °C) and compound 7 (ΔT_m = 8.6 °C) (Fig. 2C). The data confirmed the differential binding potential of 3 and 7 observed in fluorescence anisotropy assays (Fig. 2B).

Parallel artificial membrane permeability assays (PAMPAs) were performed to assess the permeability of 3 and 7 through the artificial membrane that was composed of 1% egg lecithin in n-dodecane, a useful system to examine compound cell permeability. Compounds 3 and 7 (500 μM) were placed on the donor side of the membrane. After 5-h incubation at room temperature, the amounts of 3 and 7 in the receiving solution were quantified by HPLC analyses. The percent transport (%T) and the apparent permeability coefficient (P_app) were calculated using the previously reported equations^56-57. As shown in Table 1, the PAMPA results demonstrate that 7 displays good permeability through the artificial membrane, while 3 has poor permeability in this assay.

Table 1.

The PAMPA results of compounds 3 and 7.

Compound	%T ± SD	Papp ± SD (cm • s−1, × 10−6)
3	0	0
7	24.3 ± 2.5	19.2 ± 2.0

To gain structural insights in GRB2 inhibition, a crystal structure of the GRB2 SH2 domain liganded with compound 7 was determined at 2.0 Å resolution (PDB code: 7MPH) (Fig. 3). The asymmetric unit is composed of two GRB2 trimers, and each monomer interacts with one molecule of 7 at full occupancy. The inhibitor establishes H-bonds with the main chain atoms of Lys109 and Leu120 through the carboxamide moiety, and with His107 through the N-cyclohexylacetamide moiety. Additional H-bonds occur between the phenylacetate oxygens and the side chains of Arg86 and Ser96, albeit not in all monomers, suggesting weaker interaction potential. Side chains of other surrounding residues stabilize the inhibitor through hydrophobic van der Waals interactions. The naphthalene moiety is solvent exposed and presents a potential exit vector for bivalent inhibitor design.

Fig. 3.

Cocrystal structure of GRB2 SH2 domain with compound 7 (PDB code: 7MPH). The asymmetric unit consists of 6 monomers (colored by chain ID), each liganded with one molecule of 7 (yellow). Binding pose and H-bonds of the inhibitor with surrounding residues are shown in spatial and planar views. The blue and red colored mesh shows the electron density of 7 upon refinement with (2Fo-Fc contoured at 1σ) and without ligand (Fo-Fc contoured at 3σ), respectively.

When compared to the crystal structures of GRB2 in complex with 6 (PDB IDs, 2AOA and 2AOB) and the NMR structure of GRB2 in complex with 3 (PDB ID, 1X0N), key H-bonding interactions with Lys109, Leu120, and His107 described above and van der Waals contacts with Phe108, Leu 111, His107, Lys109, and Ser96 were maintained in the crystal structure of GRB2 with 7. The H- and electrostatic bonding of GRB2 Arg86 and Ser96 with 3 and 6 seems stronger than that with 7, and the H- and electrostatic bonding of GRB2 Arg 67 with 3 and 6 is missing in the structure with 7, offering the structural base why 7 is a weaker inhibitor than 3 in biochemical assays.

In summary, we designed a new monocarboxylic inhibitor for GRB2 SH2 domain, compound 7 in Fig. 1B, and developed a robust synthetic route to synthesize this class of macrocyclic peptides. We also developed a fluorescence anisotropy competitive inhibition assay to assess disruption of the PPI between GRB2 and phosphotyrosine-containing peptides. Compound 7 exhibits a K_i of 140 nM for disruption of EGFR/GRB2 PPI and is 2-fold more potent than EGFR₁₀₆₈ phophopeptide 14-mer. The crystallographic studies demonstrate compound 7 binds with GRB2 SH2 domain and disclose key structural features for the noncovalent binding of compound 7 with GRB2. This compound carrying a −1 charge serves as an interesting lead compound for further optimization to generate a novel targeted therapy for anticancer treatment.

Highlights.

• A monocarboxylic GRB2 inhibitor, compound 7, was reported.

• The crystal structure of GRB2 in complex with 7 was determined.