Identification and biochemical characterization of small molecule inhibitors of ERK2 that target the D-recruitment site

Abstract

Extracellular signal-regulated kinase (ERK) is the culmination of a mitogen-activated protein kinase cascade that regulates cellular processes like proliferation, migration, and survival. Consequently, abnormal ERK signaling often plays a role in the tumorigenesis and metastasis of numerous cancers. ERK inhibition is a sought-after treatment for cancers, especially since clinically approved drugs that target signaling upstream of ERK often induce acquired resistance. Furthermore, the ERK2 isoform may have a differential role in various cancers from the other canonical isoform, ERK1. We demonstrate that small molecules can inhibit ERK2 catalytic and noncatalytic functions by binding to the D-recruitment site (DRS), a protein-protein interaction site distal to the enzyme active site. Using a fluorescence anisotropy-based high-throughput screening, we identify compounds that bind to the DRS and exhibit dose-dependent inhibition of ERK2 activity and ERK2 phosphorylation. We characterize the dose-dependent potency of ERK2 inhibitors using fluorescence anisotropy-based binding assays, fluorescence-based ERK2 substrate phosphorylation assays, and in vitro ERK2 activation assays. In our example, the binding of a DRS inhibitor can be prevented by mutating the DRS residue Cys-159 to serine, indicating that this residue is essential for the interaction. Resulting inhibitors from this process can be assessed in cellular and in vivo experiments for inhibition of ERK signaling and can be evaluated as potential cancer drugs.

Keywords: D-recruitment site; Docking site; Extracellular signal-regulated kinase 2 (ERK2); Fluorescence anisotropy; High-throughput screening; Protein–protein interactions; Small molecule inhibitors.

Fig. 1

Fluorescence anisotropy values are plotted for three concentrations of FITC-X-Lig-D (10, 60, and 100 nM) in binding equilibrium with inactive ERK2 concentrations of 0–12 μM.

Fig. 2

Using the parameter inputs from Table 2, the fluorescence anisotropy data for 10 nM FITC-X-Lig-D in binding equilibrium with inactive ERK2 concentrations of 0–12 μM are fit to Eq. (11) in Prism.

Fig. 3

The fluorescence anisotropy data are plotted for the assay condition of 10 nM FITC-X-Lig-D, 1 μM ERK2, and 0–200 μM Lig-D(Dap). Lig-D(Dap) induces the competitive displacement of FITC-X-Lig-D from ERK2. The data are fit to approximate the $I{C}_{50}$ concentration using Eq. (12) (dashed curve). The data are also fit to a reversible equilibrium binding model (Eqs. (9), (17), and (18)) in GraphPad Prism (solid curve) using the parameter inputs from Table 3. The independent variable is the logarithm of the Lig-D(Dap concentration), so the data for the concentration of 0 μM Lig-D(Dap) is approximated as 10⁻¹⁰ μM Lig-D(Dap).

Fig. 4

The tolerance of the fluorescence anisotropy assay to a functional concentration range of buffer additives must be evaluated. Here, we tested the solvent DMSO at different volume percentages % (v/v) in the assay. The FA signal change is shown as a percentage of the maximum FA that occurs when no solvent is present. FA = fluorescence anisotropy.

Fig. 5

The design of the plate layout for the screening is shown. Columns (Col) 1–2 contain negative controls (−), Col 3–22 contain test compounds from the screening library, and Col 23–24 contain positive controls (+). First, 3 μL of 0.5mM compounds (CPD) at 5% (v/v) DMSO in FA(−) buffer are dispensed to wells in Col 3–22, and 3 μL of 5% (v/v) DMSO in FA(−) buffer is dispensed to wells in Col 1–2 and Col 23–24. Then, 27 μL of 11.1 nM FITC-X-Lig-D in FA(+) buffer is dispensed to wells in Col 23–24. Similarly, 27 μL of 11.1 nM FITC-X-Lig-D and 1.1 μM ERK2 in FA(+) buffer is dispensed to wells in Col 1–22. FA = fluorescence anisotropy.

Fig. 6

A sampling of data from the screen is shown, consisting of 2506 negative controls (x′s), 2528 positive controls (crosses), and 2561 samples (circles). The y-axis depicts normalized percent (%) change in fluorescence anisotropy signal, where the mean positive control value is set as 0% and mean negative control value is 100% (Eq. (20)). The dotted line indicates the hit threshold of ≥50% signal reduction. The dashed lines represent ± standard deviation for the positive and negative controls.

Fig. 7

The fluorescence anisotropy data are plotted for the assay condition of 10 nM FITC-X-Lig-D, 1 μM ERK2, and 0–100 μM of the compound NSC 194308. The data are fit to a reversible equilibrium binding model (Eq. (9), Eq. (17), and Eq. (18)) in GraphPad Prism using the model from Section 3.2.2.4 and parameter inputs from Table 3. The independent variable is the logarithm of NSC 194308 concentration, so the data for the concentration of 0 μM NSC 194308 is approximated as 10⁻¹⁰ μM NSC 194308.

Fig. 8

The fluorescence anisotropy data are plotted for 10 nM FITC-X-Lig-D in binding equilibrium with ERK2 concentrations 0–12 μM of ERK2 that possesses a single Cys-159 to Ser mutation (ERK2-C159S). The data are fit to Eq. (11) in Prism. This procedure is done as described for ERK2 in Section 3.2.2.2.

Fig. 9

The fluorescence anisotropy data are plotted for the assay condition of 50 nM FITC-X-Lig-D, 5 μM ERK2, and 0–200 μM of (a) Lig-D(Dap) and (b) the compound NSC 194308. The data in (a) are fit to a reversible equilibrium binding model (Eqs. (9), (17), and (18)) in GraphPad Prism using the model from Section 3.2.2.4, adjusting for the different parameter constraints required for ERK2-C159S compared to ERK2. The independent variable is the logarithm of Lig-D(Dap) or NSC 194308 concentration, so the data for the concentration of 0 μM Lig-D(Dap) or NSC 194308 is approximated as 10⁻¹⁰ μM NSC 194308.

Fig. 10

The Michaelis–Menten kinetics are assessed for the phosphorylation of a fluorescent substrate peptide (Sox-STE7) by ERK2. (a) The ability of active ERK2 to phosphorylate 0–50 μM Sox-STE7 after the initiation of the reaction with 1mM MgATP is measured by recording fluorescence intensity (RFU = relative fluorescence units) over time (s = seconds). Fluorescence intensity data is shown over a 240 s time range that comprises the initial rate period for the reaction. (b) A plot of the slopes of the initial rates in (a) is fit to Eq. (21) to evaluate the kinetic parameters of the reaction.

Fig. 11

The ability of ERK2 to phosphorylate the fluorescent Sox-STE7 peptide is measured after preincubating 2 nM enzyme with 0–50 μM of NSC 194308. The reactions are initiated by adding 1mM MgATP and 2 μM Sox-STE7. The linear initial rates of the reactions are converted to percent activity (%) by normalizing to the rate where NSC 194308 is 0 μM. The plot of percent activity as a function of NSC 194308 concentration is fit to Eq. (12) to evaluate the $I{C}_{50}$. The independent variable is the logarithm of NSC 194308 concentration, so the data for the concentration of 0 μM NSC 194308 is approximated as 10⁻¹⁰ μM NSC 194308.

Fig. 12

Compounds are evaluated for inhibition of ERK2 phosphorylation by the kinase MKK1G7B. In this example, ERK2 (1 μM) is incubated with varied concentrations of the compound 2507-1 (0–100 μM) for 10 min, and reactions are initiated with 20 nM MKK1G7B and 0.5mM MgATP. (a) Resulting ERK2 phosphorylation at Thr183/Tyr185 (pTpY) is imaged and measured by a Western blot assay (Section 4.2.2.1). Controls include detecting total ERK protein per lane and (b) doubly phosphorylated 100% active ERK2 and unphosphorylated 0% active ERK2. (c) Quantification of the Western blot in (a) according to Section 4.2.2.2 where the pTpY-ERK signal is normalized to the total ERK signal to obtain % phosphorylation. (d) An $I{C}_{50}$ curve is generated by fitting the initial rate measurements of % ERK2 phosphorylation per minute at different 2507-1 concentrations to Eq. (12). The independent variable is the logarithm of 2507-1 concentration, so the data for the concentration of 0 μM 2507-1 is approximated as 10⁻¹⁰ μM 2507-1. Copyright 2019, American Chemical Society. This figure is reprinted with permission from (Sammons, Perry et al., 2019).