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.

1. Introduction

Two extracellular signal-regulated kinase (ERK) isoforms, ERK1 and ERK2, comprise the bottleneck of the tiered Raf/MEK/ERK kinase signaling cascade (Wortzel & Seger, 2011). Various extracellular stimuli, such as growth factors, cytokines, and other stresses, initiate the ERK pathway (Raman, Chen, & Cobb, 2007; Roskoski, 2012). These stimuli can induce signaling events that activate the Raf family of kinases (Roskoski, 2010). Activated Raf then phosphorylates and activates the kinases MEK1/2 (also called MKK1/2 or mitogen-activated protein kinase kinase 1/2), which in turn phosphorylate threonine and tyrosine residues on ERK1 (T202/Y204) and ERK2 (T185/Y187) to activate them (Roskoski, 2012). Active ERK1/2 can phosphorylate many cellular substrates in different organelles that control various cell functions (Eblen, 2018; Wortzel & Seger, 2011).

Due to the crucial role of ERK signaling in cell proliferation and survival, ERK pathway signaling often emerges as a culprit in cancer, cardiac hypertrophy, neurodegenerative disorders, and other diseases marked by uncontrolled proliferation and inflammation (Yoon & Seger, 2006). ERK pathway signaling is abnormal in 85% of cancers (Yuan, Dong, Yap, & Hu, 2020). Our work focuses on ERK2, which has been shown to have isoform-specific functions in various cancers despite the high sequence similarity of ERK1 and ERK2. For example, studies indicate that ERK2, but not ERK1, has a role in the epithelial-mesenchymal transition of MCF-10A cells (Shin, Dimitri, Yoon, Dowdle, & Blenis, 2010). Evidence also shows that ERK2 is the predominant isoform involved in the metastasis of triple-negative breast cancer to the lungs (Gagliardi et al., 2020) and the proliferation of BRAF V600 mutant melanoma cells (Crowe et al., 2021).

Several binding sites on ERK2 mediate its functions, potentially offering numerous modes of drug targeting. Traditionally, kinase inhibitors target the ATP binding site, preventing the phosphorylation of substrates by competitively displacing ATP. This method has several drawbacks, such as high structural similarities among kinase ATP binding sites contributing to a lack of selectivity and the need for highly potent inhibitors to overcome high cellular ATP concentrations (Knight & Shokat, 2005; Scapin, 2006; Vulpetti & Bosotti, 2004).

Other druggable sites on ERK2 include the D-recruitment site (DRS) and F-recruitment site (FRS), and other putative docking interaction locations that have not yet been as extensively assessed as the DRS and FRS (Herrero et al., 2015; Kummer et al., 2012; Morris et al., 2013; Sammons & Dalby, 2020; Sammons, Ghose, Tsai, & Dalby, 2019; Yoshida, Nagao, Sugiyama, Sawa, & Kinoshita, 2022). Our work has focused on the DRS, mainly due to its interactions, functions, and targetable features. The DRS is posterior to the catalytic site of ERK2 and is composed of a hydrophobic groove bracketed by an acidic common docking (CD) domain composed of the residues D316/D319 and an ED domain composed of residues Thr157/Thr158. The ED domain is named after the analogous Glu and Asp residues on p38 mitogen-activated protein kinase (Lee et al., 2004). The ED domain is adjacent to a solvent-exposed cysteine residue (Cys-159), which we have shown can be selectively targeted by the covalent inhibitor BI-78D3 (Kaoud et al., 2019). BI-78D3 prevented ERK signaling and tumor growth in melanoma models and showed inhibition of ERK activity in melanoma cell lines resistant to upstream B-Raf inhibitors (Kaoud et al., 2019). The ED domain and the hydrophobic groove of the DRS have variable sequences among different mitogen-activated protein kinases, offering a potential basis for selectively targeted inhibition (Sammons, Ghose et al., 2019). Indeed, we have identified reversibly-binding inhibitors that target the DRS with selectivity over the mitogen-activated protein kinase JNK2 (Sammons, Perry et al., 2019).

The DRS mediates interactions between ERK2 and many other proteins, making it an attractive drug target. These proteins include activators (MKK1), scaffolds, and substrates (Sammons & Dalby, 2020). We have found that ERK2 inhibitors can block the activation of ERK2 by MKK1G7B, a constitutively active form of MKK1 (Waas & Dalby, 2002). Binding to different proteins via the DRS can also affect the subcellular localization of ERK2, which governs its spatiotemporal access to specific substrates (Sammons, Ghose et al., 2019).

Here, we illustrate a detailed strategy for identifying and characterizing small molecule inhibitors that target the D-recruitment site of ERK2. We first discuss the design, optimization, and execution of a high-throughput screening where we detect small molecules that displace a fluorescein isothiocyanate (FITC)-labeled peptide from the DRS using fluorescence anisotropy. Fluorescence anisotropy (FA) measures the degree of depolarization of fluorophore emission upon excitation with polarized light (Lakowicz, 2006). The FA signal depends on the rotational diffusion of the fluorophore, so smaller fluorophores will depolarize light faster than larger ones. When a small fluorescent ligand binds to a protein, its FA increases, and when an inhibitor displaces the ligand, the FA of the ligand decreases. We then detail the selection and validation of hit compounds, including determining inhibitor potency via FA and fluorescence-based ERK2 activity assays. We also illustrate a FA protocol to compare the binding of inhibitors to ERK2 and the DRS mutant ERK2-C159S and an assay to evaluate the activation of ERK2 by MKK1G7B in the presence of inhibitors.

2. Key resources

Table 1 shows key resources used in this chapter.

Table 1.

Table showing key resources used in this chapter. Products or applications are identified by their manufacturers and catalog numbers or website links. Proteins and peptides that were prepared in-house are accompanied by relevant references. X = 6-aminohexanoic acid. FITC = fluorescein isothiocyanate. Dap = 2,3-diaminopropionic acid. Sox-STE7 was synthesized by alkylation of Sox-Br (2-bromomethyl-8-tertbutyldiphenylsilyloxy-5-(N,N-dimethyl) sulfonamide quinolone) to the cysteine residue of the peptide.

Reagent or Resource	Source	Identifier and/or Notes
Peptides and recombinant proteins
Inactive, tagless ERK2	Kaoud, Devkota et al. (2011); Yan et al. (2011)	Note: All proteins used in this protocol were prepared in-house. Proteins may be purchased commercially as an alternative option.
Active, tagless ERK2	Waas and Dalby (2002); Yan et al. (2011)
ERK2-C159S mutant	Kaoud, Devkota et al. (2011)
MKK1G7B (constitutively active MKK1 mutant)	Waas and Dalby (2002)
FITC-X-Lig-D (FITC-X-FQRKTLQRRNLKGLNLNL)	Lee et al. (2011); Kaoud, Mitra et al. (2011)	Note: Peptides for the anisotropy assay were prepared and labeled in-house.
Lig-D(Dap) (FQRKTLQRRNLK(Dap)LNLNL)	Lee et al. (2011)
Sox-STE7 (FQRKTLQRRNLKGLN-LNL-XXX-TGPLSP-C(Sox)-PF)	Zamora-Olivares et al. (2014)	Note: Unlabeled peptide was purchased from Anaspec (#1570512) and labeled in-house.
Antibodies
Antiphospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) (E10) mouse mAb	Cell Signaling Technology	9106
Anti-p44/42 MAPK (Erk1/2) (137F5) rabbit mAb	Cell Signaling Technology	4695
IRDye 800CW Goat (polyclonal) anti-rabbit IgG	LI-COR	926-32211
IRDye 680RD Goat (polyclonal) anti-mouse IgG	LI-COR	926-68070
Chemicals
HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid)	Sigma-Aldrich	H3375
KCl (potassium chloride)	Sigma-Aldrich	P9541
EDTA (ethylenediaminetetraacetic acid disodium salt dihydrate)	Sigma-Aldrich	E5134
EGTA (ethylene glycol-bis (β-aminoethylether)-N,N,N,N-tetraacetic acid)	Sigma-Aldrich	E3889
Glycerol	Fisher Scientific	BP1605
KOH (potassium hydroxide)	RICCA Chemical	6252.90 32
BSA (bovine serum albumin)	Fisher Scientific	BP1605
Triton X-100	Sigma-Aldrich	X100PC
DMSO (dimethyl sulfoxide)	Sigma-Aldrich	276855
MgCl2 (magnesium chloride)	Millipore Sigma	M1028
ATP (adenosine 5′-triphosphate disodium salt trihydrate)	Roche	10519987001
DTT (dithiothreitol)	GoldBio	DTT100
Odyssey Blocking Buffer	LI-COR	927-40000
Tris base	Sigma-Aldrich	T6066
Glycine	Sigma-Aldrich	G7126
NaCl	Sigma-Aldrich	S3014
Tween 20	Sigma-Aldrich	P9416
Methanol	Thermo Scientific	423950040
Blotting-grade blocker (non-fat dry milk)	Bio-Rad	1706404
Key instruments
Synergy H4 multimode microplate reader (or updated version)	Agilent	Note: Recommended accessories include BioStack and barcode reader
Janus Automated Workstation	PerkinElmer	Note: Use with 384 MDT head
Odyssey Sa imaging system	LI-COR	Note: Use with Image Studio software
Software
GraphPad Prism (version 7 or higher)	GraphPad by Dotmatics	https://www.graphpad.com/
LI-COR Image Studio	LI-COR Biosciences	https://www.licor.com/bio/image-studio/
Useful websites
SciFindern	Chemical Abstracts Service (American Chemical Society)	https://scifinder-n.cas.org/
Pub Chem	National Institutes of Health	https://pubchem.ncbi.nlm.nih.gov/

3. Identifying small molecules that bind to the ERK2 DRS by fluorescence anisotropy assay

The fluorescence anisotropy assay is first optimized on a small scale as the basis for the subsequent high-throughput screening. The stages of assay optimization include the production of ERK2 protein, optimization of ERK2 and fluorescent ligand concentrations, and application of mathematical binding models.

3.1. Purification of recombinant tagless ERK2

We use inactive ERK2 protein produced in our laboratory for the FA assay. The choice of inactive protein is due to our goal of detecting inhibitors that disrupt ligand binding at the DRS, as this site is essential for ERK2 activation in cells (Tanoue, Adachi, Moriguchi, & Nishida, 2000). For this chapter, we consider the details of the expression and purification of ERK2 to be beyond the scope of high-throughput screening development. Furthermore, the protocol for producing ERK2 is well-established (Kaoud, Devkota et al., 2011; Sammons, Perry et al., 2019). Therefore, we provide a brief description of the process in this section.

Tagless ERK2 is expressed and purified essentially as described in our previous works (Kaoud, Devkota et al., 2011). Erk2 DNA (NM_053842) is ligated into a pet28a (+) vector modified to include a TEV cleavage site for His-tag removal (Yan, Kaoud, Lee, Dalby, & Ren, 2011). Note that the mutant ERK2-C159S used in this chapter is expressed and purified like ERK2 (Kaoud, Devkota et al., 2011).

We use electroporation to transform the vector into BL21(DE3) electrocompetent E. coli cells. Cells are plated on agar containing kanamycin and incubated at 37 °C overnight, and colony selection is performed the following day. Selected colonies are grown in lysogeny broth (LB) media at 37 °C until optical density reaches 0.6. Then protein expression is induced with 0.5 mM isopropyl β-d-1 thyogalactopyranoside (IPTG), and the cells in the media are incubated at 30 °C for 4 h. Media is collected and centrifuged at 7000 RPM (8781 × g) for 15 min, and the cell pellet is isolated and frozen with liquid nitrogen to store at −80 °C for purification.

To purify the ERK2 protein, we first lyse the frozen cell pellet and purify the lysate on a Ni-NTA (Ni²⁺ ion coupled to nitrilotriacetic acid) agarose bead column. The resulting eluate from the Ni-NTA column is purified by fast protein liquid chromatography (FPLC) using a Mono Q anion exchange column (Cytiva). The presence of ERK2 protein is confirmed by running sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) for the peak fractions.

The His-tag is cleaved from the purified protein using tobacco etch virus (TEV) protease [expressed and purified as previously reported (Yan et al., 2011)] at 1.5% (w/w) of ERK2 concentration. The mixture is incubated at ambient temperature for 3–5 h, and cleaved ERK2 is purified from the mixture by FPLC using a Mono Q anion exchange column. Cleavage of the His-tag is verified by SDS-PAGE. The tagless protein is dialyzed into a storage buffer containing 10% (v/v) glycerol and frozen with liquid nitrogen for storage at −80 °C.

To activate ERK2, MKK1G7B is expressed and purified in the same manner as ERK2 (Waas & Dalby, 2002). MKK1G7B is a constitutively active form of mitogen-activated protein kinase kinase 1 (MKK1) that can dually phosphorylate and activate ERK2. A mixture of 5 μM inactive ERK2, 0.5 μM MKK1G7B, 4 mM adenosine 5′-triphosphate (ATP), and 14 mM magnesium chloride is incubated at 28 °C for 5–8 h, then run through a desalting column on the FPLC to remove ATP. The active form of ERK2 is purified using the Mono Q anion exchange column as with inactive ERK2. Active ERK2 is dialyzed and stored as described for inactive ERK2. Dual phosphorylation of ERK2 is confirmed via mass spectrometry.

3.2. Fluorescence anisotropy assay development

3.2.1. Materials

3.2.1.1. Reagents

1. N-2-hydroxyethylpiperazine-N’-2-ethanesulfonic acid (HEPES) (Sigma-Aldrich, #H3375)

2. Potassium hydroxide (KOH) (RICCA Chemical, #6252.90 32) for adjusting buffer pH

3. Potassium chloride (KCl) (Sigma-Aldrich, #P9541)

4. Ethylenediaminetetraacetic acid disodium salt dihydrate (EDTA) (Sigma-Aldrich, #E5134)

5. Ethylene glycol-bis(β-aminoethylether)-N,N,N’,N’-tetraacetic acid (EGTA) (Sigma-Aldrich, #E3889)

6. Glycerol (Fisher Scientific, #BP229)

7. Bovine serum albumin (BSA) (Fisher Scientific, #BP1605)

8. Triton X-100 (Sigma-Aldrich, #X100PC)

9. Recombinant ERK2 recombinant protein (see Section 3.1)

10. Fluorescent probe FITC-X-Lig-D (with the sequence FITC-X-FQRKTLQRRNLKGLNLNL, where X is the linker 6-aminohexanoic acid, and FITC is a fluorescein isothiocyanate label) is required. We recommend purchasing this custom-labeled peptide or labeling the peptide with fluorophore in-house. We synthesize FITC-labeled Lig-D (FITC-X-Lig-D) in the same manner as the DRS ligand Lig-D (Lee et al., 2011), with the addition of labeling the C-terminus of Lig-D with FITC via a 6-aminohexanoic acid linker on a cysteine residue (Kaoud, Mitra et al., 2011).

11. A positive control, such as a small molecule inhibitor or non-fluorescent peptide, is required to test the displacement of the fluorescent probe from ERK2. In our assay, we use an unlabeled variant of the Lig-D peptide, Lig-D(Dap), with the sequence Ac-FQRKTLQRRNLK(Dap) LNLNL-NH₂, where Ac is an N-terminal acetyl group, Dap is 2,3-diaminopropionic acid, and the C-terminus is amidated. The control peptide Lig-D(Dap) is synthesized in the same manner as the DRS ligand Lig-D (Lee et al., 2011), with the exception that diaminopropionic acid (Dap, sold as Fmoc-Dap(MTT)-OH from Chem-Impex International) was substituted for Gly-13 in the peptide sequence.

3.2.1.2. Consumables

1. 384-well black polystyrene plates (e.g., Nunc, #262260)

2. Aluminum plate seals (Nunc, #232699)

3. Aluminum foil

4. Various sizes of pipettes and pipette tips

5. Filters (0.2–0.45 μm, sterile)

6. Various plastic tubes and buffer storage containers as needed

3.2.1.3. Tools and Instruments

1. Single channel, multichannel, and/or multi-dispense pipettes for volumes of 1–1000 μL

2. Motorized pipettor for volumes of 1–25 mL

3. A microplate reader with polarizing filters capable of measuring fluorescence polarization/anisotropy for excitation/emission wavelengths of 485/530 nm is required. Here, we used the Synergy H4 multimode microplate reader (Agilent).

4. Microplate centrifuge, e.g., Benchtop Centrifuge 5810 (Eppendorf)

3.2.1.4. Software

Data analysis and fitting software, e.g., GraphPad Prism (we recommend version 7 or higher)

3.2.2. Methods

We must (1) conserve ERK2 protein since it is a limiting resource and (2) ensure sufficient FA signal for the bound probe, leading to the need for the protein to be in excess of the probe for shifting the equilibrium to the bound state. We must minimize the concentrations of probe and enzyme to conserve resources while ensuring that the FA signal-to-background is satisfactory. This is done by measuring FA for fixed fluorescent ligand concentration at varied enzyme concentrations and vice versa and selecting the pair of concentrations that yield an optimal assay signal window. Note that FA is a unitless number, and all data in this chapter are shown as FA × 10³.

3.2.2.1. Fluorescent probe and ERK2 concentration optimization

1. 1× buffer must first be prepared, consisting of 25 mM HEPES–KOH, 50 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, and 1.3% (v/v) glycerol, pH 7.5. It is critical to filter the buffer through a 0.2–0.45-micron filter to remove contaminants and degas it by bubbling with an available inert gas, such as nitrogen or argon, for at least 15 min.

2. Fresh buffer additives BSA and Triton X-100 are added before the experiment begins. BSA and Triton X-100 prevent proteins and peptides from aggregating or sticking to pipette tips and other plastics. These additive concentrations can be optimized similarly to Section 3.3.2.1, which describes assay tolerance assessments. Stock solutions of BSA and Triton X-100 are prepared in 1× buffer from step 1 to concentrations of 10 mg/mL and 10% (w/v), respectively. The full FA buffer is 25 mM HEPES–KOH, 50 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1.3% (v/v) glycerol, pH 7.5, 10μg/mL BSA, and 0.01% (w/v) Triton X-100. In this chapter, FA(−) buffer is the full buffer without Triton X-100 and BSA, and FA(+) buffer is the full buffer with 1.11x BSA and Triton X-100 (11.1 μg/mL and 0.011% (w/v), respectively).

3. FITC-X-Lig-D is dissolved to a concentration of 1 mM in water in a conical vial or tube and should be protected from light by wrapping the tube in aluminum foil. Before the experiment begins, it should be diluted to 500 nM in FA buffer.

4. ERK2 is prepared according to Section 3.1. All ERK2 used in this chapter should come from the same original stock. In our case, this stock is typically 50–70 μM. Before beginning the experiment, it should be diluted to 10 μM in FA buffer and kept on ice.

5. To determine optimal ERK2 and FITC-X-Lig-D concentrations, we must vary the ERK2 concentration at different fixed FITC-X-Lig-D concentrations. Here, FITC-X-Lig-D is held at 10, 60, or 100 nM, and ERK2 (inactive, His-tag-cleaved) is varied from 0 to 12 μM in the FA buffer. The assay volume in each well is 50 μL.

6. At ambient temperature (25 °C), incubate samples in a 384-well black polystyrene assay plate for approximately 30 min to reach binding equilibrium.

7. Read each sample’s fluorescence anisotropy at ambient temperature using a plate reader. If the reader records the data as fluorescence polarization $\left(F_p\right)$, transform them to fluorescence anisotropy (r) using Eq. (1) (Lakowicz, 2006). Fluorescence anisotropy and fluorescence polarization both measure the parallel $\left(I_{\parallel }\right)$ and perpendicular $\left(I_{\perp }\right)$ components of fluorescence emission intensity, but FA is normalized by total fluorescence intensity $\left(I_{\parallel }+2I_{\perp }\right)$ (Eq. (2)), so it simplifies data processing.

   $$r=\frac{2F_p}{3-F_p}$$ (1)

   $$r=\frac{I_{\parallel }-I_{\perp }}{I_{\parallel }+2I_{\perp }}$$ (2)

8. Plot FA as a function of ERK2 concentration at different FITC-X-Lig-D concentrations (Fig. 1).

9. Use the data from step 8 to set the optimal concentration of FITC-X-Lig-D. From Fig. 1, FITC-X-Lig-D can be minimized to 10 nM without affecting the assay signal.

10. Fit the data from Fig. 1 at 10 nM FITC-X-Lig-D to Eq. (3) to determine the EC₅₀, which is the concentration of ERK2 that produces a half-maximal FA signal. In this equation, x is the concentration of ERK2, y is FA, and y_hottom and y_top are the minimum and maximum FA values, respectively. The FA data in Fig. 1 are fit to Eq. (3) and the EC₅₀ is calculated to be 0.8 ± 0.2 μM.

    $$\gamma ={\gamma }_{bottom}+\frac{x\left({\gamma }_{top}-{\gamma }_{bottom}\right)}{x+E{C}_{50}}$$ (3)

11. The optimal ERK2 concentration is set by rounding the EC₅₀ to the nearest 1 μM.

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.

3.2.2.2. Determining K_D

1. We must determine the binding affinity (K_D) for fluorescent ligand binding to ERK2 using the data obtained in Section 3.2.2.1. This parameter is necessary for evaluating inhibitors and hit compounds from the screen in Sections 3.2.2.4 and 3.4.2. We assume a tight-binding model since the total enzyme concentration significantly exceeds the fluorescent ligand concentration (Callaway, Abramczyk, Martin, & Dalby, 2007; Callaway, Rainey, & Dalby, 2005). The following scheme represents the binding equilibrium, where E is ERK2, S is FITC-X-Lig-D, and ES is the reversibly-bound complex.

   $$E+S\rightleftharpoons ES$$

   For 1:1 binding between the fluorescent species and protein, the fraction of bound fluorescent species (Θ) in terms of the fluorescence anisotropy signal is shown in Eq. (4) (Lakowicz, 2006).

   $$\theta =\frac{\left(r-r_F\right)}{\left(r-r_F\right)+R\left(r_B-r\right)}$$ (4)

   Here, r is the fluorescence anisotropy of the sample, $r_F$ is the fluorescence anisotropy of the free ligand in solution when ERK2 is absent, $r_B$ is the extrapolated maximum fluorescence anisotropy of the ligand bound to ERK2 (as ERK2 concentration approaches $\infty$). Note that $r_B={\gamma }_{top}$ and $r_F={\gamma }_{bottom}$ from Eq. (3).

   R is a correction factor for intensities of free vs. bound ligand (Lakowicz, 2006). If the fluorescence intensities of free and bound ligand are equal, $R=1$. Otherwise, R can be evaluated as the ratio of intensities of the bound and free forms of FITC-X-Lig-D ($f_B$ and $f_F$, respectively) as shown in Eq. (5). In our assay, we measured R as 1.8.

   $$R=f_B/f_F$$ (5)

   Using the principles of equilibrium binding, we can extract more information from the data in Fig. 1. The dissociation constant $K_D$ is given by Eq. (6).

   $$K_D=\frac{\left[E_F\right]\left[S_F\right]}{\left[ES\right]}$$ (6)

   In Eq. (6), $\left[E_F\right]$ is the free ERK2 concentration, $\left[S_F\right]$ is the free FITC-X-Lig-D concentration, and $\left[ES\right]$ is the protein–ligand complex concentration at equilibrium. Using mass balances, total enzyme $\left[E_T\right]$ and total substrate $\left[S_T\right]$ are conserved according to Eqs. (7) and (8).

   $$\left[E_T\right]=\left[E_F\right]+\left[ES\right]$$ (7)

   $$\left[S_T\right]=\left[S_F\right]+\left[ES\right]$$ (8)

   Using Eqs. (7) and (8), Eq. (4) can be re-written as Eq. (9).

   $$r=\frac{\frac{\left[ES\right]}{\left[S_T\right]}\left(r_{B}R-r_F\right)+r_F}{1+\frac{\left[ES\right]}{\left[S_T\right]}\left(R-1\right)}$$ (9)

   Solving Eq. (6) for [ES] using Eqs. (7) and (8) yields the quadratic Eq. (10).

   $$\left[ES\right]=\frac{K_D+\left[S_T\right]+\left[E_T\right]-\sqrt{{\left(-K_D-\left[S_T\right]-\left[E_T\right]\right)}^2-4\left[E_T\right]\left[S_T\right]}}{2}$$ (10)

   Combining Eqs. (9) and (10) yields fluorescence anisotropy as a function of enzyme concentration (Eq. (11)).

   $$r=\frac{\frac{K_D+\left[S_T\right]+\left[E_T\right]-\sqrt{{\left(-K_D-\left[S_T\right]-\left[E_T\right]\right)}^2-4\left[E_T\right]\left[S_T\right]}}{2\left[S_T\right]}\left(r_{B}R-r_F\right)+r_F}{\frac{K_D+\left[S_T\right]+\left[E_T\right]-\sqrt{{\left(-K_D-\left[S_T\right]-\left[E_T\right]\right)}^2-4\left[E_T\right]\left[S_T\right]}}{2\left[S_T\right]}\left(R-1\right)+1}$$ (11)

2. Fitting the data to Eq. (11) in a curve-fitting software program of your choice can be used to calculate the binding affinity of FITC-X-Lig-D for ERK2. In GraphPad Prism, Eq. (11) can be entered as a custom nonlinear fit equation using the following text format:

   $$\begin{gathered} \text{Y}=\left(\left(\text{Kd}+\text{S}+\text{X}-\left({\left({\left(-\text{Kd}-\text{S}-\text{X}\right)}^{^}2-4*\text{X}*\text{S}\right)}^{^}\left(1/2\right)\right)\right)/2\left(2*\text{S}\right)*\left(\text{rb}*\text{R}-\text{rf}\right)+\text{rf}\right)/· \\ \left(\left(\text{Kd}+\text{S}+\text{X}-\left({\left({\left(-\text{Kd}-\text{S}-\text{X}\right)}^{^}2-4*\text{X}*\text{S}\right)}^{^}\left(1/2\right)\right)\right)/2\left(2*\text{S}\right)*\left(\text{R}-1\right)+1\right) \end{gathered}$$

   Fluorescence anisotropy is the dependent variable (Y) and ERK2 concentration is the independent variable (X). Table 2 shows inputs into Prism for the other parameters, as well as the results of the curve fit. Note that Prism does not accept symbols or subscript formats, so the parameters are redefined in Table 2.

3. Using the data from Fig. 1 at 10 nM FITC-X-Lig-D and fitting it to Eq. (11) in Prism yields the results shown in Fig. 2. From the fit, K_D is 1.5 ± 0.3 μM. This parameter is used in Sections 3.2.2.4 and 3.4.2 to evaluate molecules that inhibit FITC-X-Lig-D from binding to ERK2.

Table 2.

Table showing the inputs for fitting Eq. (11) in GraphPad Prism software. Columns include parameter definitions, constraints on the fit parameters, initial value predictions for each parameter, and the results of the fit. FA = fluorescence anisotropy. Numerical values shown in the table are unitless unless otherwise noted as micromolar concentrations.

Parameter	Description	Constraint	Initial value	Fit result
Kd	The dissociation constant	Must be greater than 0	0.8 (μM) (EC50 from Eq. (3))	1.5 ± 0.33 (μM)
S	Total ligand concentration	=0.010 (μM)	Fixed	0.010 (μM)
rb	FA signal for maximally bound complex	Must be greater than 0	“YMAX” (maximum FA value)	164 ± 4
rf	FA signal for 100% free ligand	Must be greater than 0	“YMIN” (minimum FA value)	60 ± 6
R	Fluorescence intensity correction factor	=1.8	Fixed (R from Eq. (5))	1.8

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.

3.2.2.3. Positive control test

1. As a positive control for the assay, we must test a known small molecule, peptide, or protein-binding partner that can compete with the fluorescent probe for the same binding site (or otherwise disrupt the binding of ERK2 and FITC-X-Lig-D in a dose-dependent manner). Here, we use the peptide Lig-D(Dap), a non-fluorescent analog of FITC-X-Lig-D with diaminopropionic acid at the C-terminal lysine position.

2. Incubate varied concentrations of Lig-D(Dap) (0–200 μM) with a mixture of 10 nM FITC-X-Lig-D and 1 μM ERK2 in FA buffer for 30 min at ambient temperature in a black 384-well plate. The assay volume in each well is 50 μL.

3. Read the FA of the samples at ambient temperature using a plate reader.

4. Plot FA as a function of Lig-D(Dap) concentrations (Fig. 3).

5. The observed $I{C}_{50}$ from Eq. (12) is used to approximate the concentration of Lig-D(Dap) that displaces 50% of the fluorescent ligand from ERK2. In this equation, x is the concentration of Lig-D(Dap), y is FA, and ${\gamma }_{bottom}$ and ${\gamma }_{top}$ are the minimum and maximum FA values, respectively. The FA data in Fig. 3 were fit to Eq. (12) (dashed curve) and the $I{C}_{50}$ was calculated as approximately 1.3 ± 0.4 μM. At 50 μM, Lig-D (Dap) almost completely displaced the fluorescent probe from ERK2, so this concentration can be selected as a positive control for the screening.

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).

$$\gamma ={\gamma }_{bottom}+\frac{\left({\gamma }_{top}-{\gamma }_{bottom}\right)}{\left(1+\frac{x}{I{C}_{50}}\right)}$$ (12)

Note that if you want to fit y vs log[x], replace x in Eq. (12) with ${10}^{\log \left[x\right]}$ such that ${10}^{\log \left[x\right]}={10}^{x^{\prime }}$ where $x^{\prime }$ is the transformed independent variable. Approximate the $x=0$ μM data point as $x={10}^{-10}$ μM to accommodate the logarithmic transformation. The resulting parameters from the fit will not be significantly different regardless of method.

3.2.2.4. Determining K_I for inhibitors and positive control molecules

1. Because the FITC-X-Lig-D substrate concentration is much less than the concentration of ERK2, molecules that displace it must be carefully evaluated for their binding equilibrium. This is essential because the majority of inhibitor molecules binding to ERK2 will not change the FA signal since the fluorescent probe is bound to only a small fraction of the enzyme in the assay. The $I{C}_{50}$ for the FA signal reduction (Eq. (12)) will greatly underestimate the inhibitor’s potency. We therefore used the following equilibrium model to determine the binding affinity (K_I) of Lig-D(Dap) for ERK2. This model will also be used for evaluating hits from the high-throughput screening in Section 3.4.2.

   FITC-X-Lig-D (S) binds to ERK2 (E) to create the ES complex:

   $$E+S\rightleftharpoons ES$$

   Inhibitor (B) binds to ERK2 (E) to create the EB complex:

   $$E+B\rightleftharpoons EB$$

   The equilibrium model can be used to write the following series of mass-balance equations, where brackets around the species symbols indicate units of concentration.

   $$\begin{gathered} \left[E_T\right]=\left[E_F\right]+\left[ES\right]+\left[EB\right]\approx \left[E_F\right]+\left[E_T\right]\text{since} \\ \left[S_T\right]\ll \left[E_T\right] \end{gathered}$$ (13)

   $$\left[B_T\right]=\left[B_F\right]+\left[EB\right]$$ (14)

   The dissociation constants for EB and ES yield Eqs. (15) and (16). Dissociation constants for the inhibitor and the fluorophore are indicated by $K_I$ and $K_D$, respectively.

   $$\left[ES\right]=\frac{\left[E_F\right]\left[S_T\right]}{K_D+\left[E_F\right]}$$ (15)

   $$\left[EB\right]=\frac{\left[E_F\right]\left[B_F\right]}{K_I}$$ (16)

   Substituting the mass balances (Eqs. (8), (13) and (14)) into Eqs. (15) and (16) and algebraically rearranging them leads to the following equations.

   $$\left[ES\right]=\frac{\left[S_T\right]\left(\left[E_T\right]-\left[EB\right]\right)}{K_D+\left[E_T\right]-\left[EB\right]}$$ (17)

   $$\left[EB\right]=\frac{1}{2}\left[\left(\left[E_T\right]+\left[B_T\right]+K_I\right)-\sqrt{{\left(\left[E_T\right]+\left[B_T\right]+K_I\right)}^2-4\left[E_T\right]\left[B_T\right]}\right]$$ (18)

   Eqs. (9), (17), and (18) form a system of three equations that allow the evaluation of K_I using r as the independent variable and [B_T] as the dependent variable. The values of [E_T], [S_T], R, and importantly, K_D from the protocol in Section 3.2.2.2, are known.

   Fitting the data points in Fig. 3 to the system of equations Eqs. (9), (17), and (18) in a curve-fitting software of your choice can be used to calculate the binding affinity of Lig-D(Dap) for ERK2. In GraphPad Prism, the three equations can be entered as a custom nonlinear fit equation using the following text format. Prism does not accept symbols or subscript formats, so the parameters are redefined in Table 3.

   $$\text{EB}=\left(\left(\text{Et}+\text{X}+\text{Ki}\right)-\text{sqrt}\left({\left(\text{Et}+\text{X}+\text{Ki}\right)}^{^}2-4*\text{Et}*\text{X}\right)\right)/2$$

   $$\text{ES}=\text{St}*\left(\text{Et}-\text{EB}\right)/\left(\text{Kd}+\text{Et}-\text{EB}\right)$$

   $$\text{Y}=\left(\left(\text{ES}/\text{St}*\left(\text{rb}*\text{R}-\text{rf}\right)+\text{rf}\right)/\left(1+\left(\text{ES}/\text{St}*\left(\text{R}-1\right)\right)\right)\right)$$

   The dependent variable (Y) is fluorescence anisotropy as a function of ERK2 concentration (X). Table 3 shows inputs into Prism for the other parameters, as well as the results of the curve fit.

2. Plotting the data to fit to Eqs. (9), (17), and (18) yields the results shown in Fig. 3 (solid curve). From the fit, $K_I$ is 0.18 ± 0.04 μM. As expected, this affinity is much more potent than the value indicated by the $I{C}_{50}$ in Eq. (12). Note that if you want to fit FA vs $\log \left[B_T\right]$, replace $\left[B_T\right]$ in Eq. (18) with ${10}^{\log \left[B_T\right]}$ such that in Prism, ${10}^{\log \left[X\right]}={10}^{X^{\prime }}$ where $X^{\prime }$ is the transformed independent variable. Approximate the $\left[B_T\right]=0$ μM data point as $\left[B_T\right]={10}^{-10}$ μM to accommodate the logarithmic transformation. The resulting parameters from the fit will not be significantly different regardless of method.

Table 3.

Table showing the inputs for fitting the set of Eqs. (9), (17), and (18) in GraphPad Prism software. Columns include parameter definitions, constraints on the fit parameters, initial value predictions for each parameter, and the results of the fit. FA = fluorescence anisotropy. Numerical values shown in the table are unitless unless otherwise noted as micromolar concentrations.

Parameter	Description	Constraint	Initial value	Fit result
Kd	The dissociation constant for FITC-X-Lig-D binding	=1.5 (μM)	Fixed (KD from Eq. (11))	1.5 (μM)
St	Total ligand concentration	=0.010 (μM)	Fixed	0.010 (μM)
rb	FA signal for maximally bound complex	Must be greater than 0	“YMAX” (maximum FA value)	226 ± 6
rf	FA signal for 100% free ligand	Must be greater than 0	“YMIN” (minimum FA value)	55 ± 1
R	Fluorescence intensity correction factor	=1.8	Fixed (R from Eq. (5))	1.8
Et	Total ERK2 concentration	=1.0 (μM)	Fixed	1.0 (μM)
Ki	The dissociation constant for inhibitor binding	Must be greater than 0	1.3 (μM) ($I{C}_{50}$ from Eq. (12))	0.18 ± 0.04 (μM)

3.2.3. Expected outcomes

This procedure should result in a set of optimized experimental conditions that can be adapted to high-throughput screening format. These conditions include assay buffer, concentrations of protein and ligand, data collection steps, and data processing. Positive controls (minimum FA signal) and negative controls (maximum FA signal) are established for the assay. Protein and ligand concentrations should be reasonable for scaling up the assay, considering availability and budget for these resources.

3.2.4. Quantification and statistical analysis

See Sections 3.2.2.2 and 3.2.2.4 for calculating parameters for ERK2 binding to FITC-X-Lig-D and ERK2 binding to inhibitors, respectively.

3.2.5. Advantages

Typical kinase inhibition assays require active enzyme and a substrate that interacts with the active site and can be phosphorylated. Fluorescence anisotropy measures equilibrium binding, so the enzyme can have a variety of activation states, posttranslational modifications, and conformations. Also, ligands can be designed to bind to other sites on proteins aside from the active site, which is important here because ERK2 has a variety of functions that are independent of its enzyme activity. FA measurements of equilibrium binding are mainly dependent on the reagent stability under assay conditions, so timing of reagent additions and data collection is flexible and signal is stable for long periods of time in the case of ERK2. This allows many plates in the high-throughput screening to be read per day with minimal steps, as more samples take more time to read and the number is therefore limited by signal stability over time.

3.2.6. Limitations

This FA assay cannot characterize time-dependent enzyme interactions like rates of phosphorylation of ERK2 substrates. It measures two states of the fluorescent ligand: bound and unbound to ERK2. The binding affinities of inhibitors should only be interpreted relative to positive and negative controls, as these affinity values apply under the conditions of this assay only.

3.2.7. Optimization and trouble-shooting

Assay optimization is described in Section 3.2.2.1. If a fluorescence signal is not observed, check fluorometer settings and the FITC fluorescence emission spectrum of FITC-X-Lig-D alone in assay buffer. If binding is not observed, check the accuracy of the protein concentration and ensure that BSA and detergent are present in the buffer at all times, as proteins and ligands can stick to pipette tips and the sides of plate wells in the absence of these additives.

3.2.8. Safety considerations and standards

Use basic personal protective equipment (PPE) including gloves, eye protection, and lab coats.

3.2.9. Alternative methods and procedures

Other biochemical and biophysical methods to evaluate binding between protein and ligand/inhibitor include surface plasmon resonance, isothermal titration calorimetry, thermal denaturation shifts, and absorbance or fluorescence intensity measurements if these change upon binding.

3.3. High-throughput screening

All FA measurements are collected at ambient temperature on a Synergy H4 plate reader using a 480 nm excitation filter and a 512 nm emission filter. Assays are performed at ambient temperature in a final composition of 1× FA buffer. Samples contain 10 nM FITC-X-Lig-D and 1 μM ERK2 (inactive, His-tag-cleaved), with or without library compounds. Compounds are diluted from 10 mM DMSO stocks to reach a final assay concentration of 50 μM and 0.5% (v/v) DMSO. Positive controls include 50 μM Lig-D(Dap) with 10 nM FITC-X-Lig-D and 1 μM ERK2, and 10 nM FITC-X-Lig-D alone. Compounds or DMSO controls are incubated with the enzyme-probe mixture for 1 h before data collection.

3.3.1. Materials

Materials include the same reagents, consumables, and instruments listed in

Section 3.2.1, in addition to the following:

3.3.1.1. Reagents

1. Selected libraries of compounds to test in the screen (see Section 3.3.2.3 for more information)

2. Library solvent, e.g., DMSO (dimethyl sulfoxide) (Sigma-Aldrich, #276855)

3. Inert gas for compound storage

3.3.1.2. Consumables

1. 384-well dilution plates (Greiner, #781280)

2. Drierite (desiccant for library plate storage, W.A. Hammond Drierite Company, #23001)

3. Plate barcode labels

3.3.1.3. Tools and instruments

If you plan to screen a large number of compounds, consider making the following accommodations for automation:

1. Use a liquid handler capable of dispensing varied volumes to 384-well plates. The choice of liquid handler should depend on the screening scale and instrument access. We use the Janus Automated Workstation (PerkinElmer) equipped with a 384 MDT head.

2. A Synergy H4 microplate reader equipped with a BioStack multiplate stacker that transfers sequential plates between a stacked plate column and the reader and a barcode reader that automatically records plate labels as they enter the Synergy H4 microplate reader (Agilent)

3. Automated plate sealer, such as the PlateLoc thermal microplate sealer (Agilent)

3.3.2. Methods

To adapt the FA assay to a high-throughput screening format, we must take several measures to scale up the number of samples and assess the suitability of the assay for screening. These steps include assessing the tolerance of the assay signal to additives, performing a full-plate validation, selecting the compound library that will be screened, and designing the screening plate format. Here we describe these steps, the primary screening process, quality control methods, and hit selection.

3.3.2.1. Determine the assay tolerance to additives

Protein−ligand binding can be affected by the presence of buffer additives, so it is important to evaluate the assay tolerance to each one. This protocol assesses assay tolerance to DMSO as an example since it is the solvent used for all library compounds.

1. Prepare FA buffer, ERK2, and FITC-X-Lig-D as described in Section 3.2.2.1.

2. To determine the effects of DMSO on the FA signal, dilute DMSO to final concentrations of 0%, 0.2%, 0.5%, 1%, 2%, and 4% (v/v) in the presence of 10 nM FITC-X-Lig-D and 1 μM ERK2 in FA buffer. These samples should be 50 μL each and contained in a black 384-well plate.

3. Measure the FA of the samples after approximately 30 min of incubation at ambient temperature to establish equilibrium.

4. Normalize FA values as a percentage of the control: the FA signal of the samples containing 0% (v/v) DMSO.

5. Plot the assay signal (% of max) as a function of DMSO % (v/v) (Fig. 4). In this case, the assay is relatively insensitive to DMSO concentration up to 4% (v/v).

6. This process should be performed for every assay additive, including Triton X-100, BSA, and any other agent introduced into the original assay buffer. Effects on the assay signal should be evaluated to ensure quality control requirements are met for the chosen assay conditions (see Section 3.3.2.2).

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.

3.3.2.2. Perform a full plate validation

A full plate validation should be performed before beginning the compound screen. The plate should include only positive and negative controls and should be performed in the same manner as the screen, including the same instruments, assay plates, liquid transfer methods, and reagent stocks.

The liquid dispensing steps will include 5 μL DMSO (5% (v/v)) followed by 45 μL of positive or negative control mixture. Each well will contain a final volume of 50 μL.

1. Design the 384-well assay plate layout so that half of the plate contains positive controls and the other half contains negative controls. For example, positive and negative controls can be placed in alternating columns.

2. Prepare the master mix that will be used for negative controls. Negative controls will have final concentrations of 10 nM FITC-X-Lig-D and 1 μM ERK2 in 50 μL assay volume. Before use, thaw ERK2 protein on ice. Dilute ERK2 and FITC-X-Lig-D in the FA(+) buffer such that the master mix contains 1.11 μM ERK2 and 11.11 nM FITC-X-Lig-D.

3. Prepare the master mix for positive controls. Positive controls will have a final concentration of 10 nM FITC-X-Lig-D in the FA(+) buffer without ERK2. Positive controls may also include 50 μM Lig-D(Dap) or a similar agent determined from Section 3.2.2.3.

4. For the entire validation plate, the required volume (V) of each master mix can be determined by V = 45 × 384/2 + D where D is dead volume, or the volume of liquid required to prime the liquid dispenser and occupy residual space in liquid reservoirs, for your chosen liquid dispensing method.

5. Using the pipetting or liquid handling method of your choice, dispense 5 μL of 5% (v/v) DMSO in FA(−) buffer to each well of the plate.

6. Centrifuge the plate at ambient temperature for 1 min at 800 RPM (127 × g).

7. Dispense 45 μL of the positive control mixture into odd-numbered plate columns.

8. Dispense 45 μL of the negative control mixture into even-numbered plate columns.

9. Centrifuge the plates at ambient temperature for 2 min at 800 RPM (127 × g).

10. Seal the plate manually with an aluminum seal, and incubate at ambient temperature for 1 h.

11. Remove the seals from the plate and read the FA of the samples on the microplate reader.

12. Calculate the observed $Z^{\prime }$ factor for the plate. The $Z^{\prime }$ factor measures assay quality and robustness, and can confirm readiness for a high- throughput format (Zhang, Chung, & Oldenburg, 1999). It effectively measures the separation of positive and negative control values, and indicates the signal window of the assay (Eq. (19)).

    $$Z^{\prime }=1-\frac{3\left({\sigma }_p+{\sigma }_n\right)}{\left|{\mu }_p-{\mu }_n\right|}$$ (19)

    Here, ${\sigma }_p$ and ${\sigma }_n$ are the population standard deviation of the positive and negative control sample signals, respectively. Similarly, ${\mu }_p$ and ${\mu }_n$ are the population mean values of the positive and negative controls. We approximate $Z^{\prime }$ using the means and standard deviations of our observed data $\left({Z^{\prime }}_{OBS}\right)$ in place of the population values in Eq. (19). In most of our applications, ${Z^{\prime }}_{OBS}>0.5$ is considered satisfactory and ideally, ${Z^{\prime }}_{OBS}>0.7$. An example set of statistics for a validation plate is shown in Table 4. Using Eq. (19), the ${Z^{\prime }}_{OBS}$ was calculated as 0.84, well above the acceptable threshold for high-throughput screening.

13. If the controls are homogenous and the assay statistics are acceptable, the assay is ready to be applied to the full screening.

14. Assess if the assay volume can be reduced to conserve reagents by repeating the validation and calculating ${Z^{\prime }}_{OBS}$ at lower volumes. For our screen, we reduce the assay volume to 30 μL without loss of signal integrity.

Table 4.

Table showing the mean and standard deviation for the positive (+) and negative (−) controls in the screening validation plate.

	Mean fluorescence anisotropy	Standard deviation
(+) control	106.6	1.7
(−) control	44.8	1.5

3.3.2.3. Screen the compound library

Select a library of compounds to screen. Important considerations include the number of compounds, their format (solvent, purity, mixtures), preparation and storage, and structural and chemical attributes. Library selection can pre-screen for key properties desired in a hit molecule, such as bioavailability. Library selection also determines the diversity of chemical moieties and structures. We have performed the screen using two different approaches to library selection.

• We used a commercial library of over 30,000 individual compounds in one screen. The library includes 674 NIH clinical collection compounds (Evotec, San Francisco, CA), 2000 Spectrum collection compounds (Microsource Discovery, Gaylordsville, CT), 13,440 Fsp3-enriched diversity compounds (Life Chemical, Niagara-on-the-Lake, ON, Canada), 1280 Natural-like compounds (Life Chemical), 12,900 Diversity set compounds (Chemdiv, San Diego, CA), 1597 NexT Diversity set V compounds (NCI), 821 NexT Mechanistic set III compounds (NCI), 419 NexT Natural products IV (NCI), 129 NexT Oncology set compounds (NCI), and 1280 Lopac compounds (Sigma- Aldrich, St Loise, MO). This approach aimed to maximize diversity and selectivity of potential hits.

• In another case, we used a more tailored approach to library selection to screen a combinatorial library (Sammons, Perry et al., 2019). In this example, we screened peptide-like molecules that reversibly bind to ERK2 in collaboration with Torrey Pines Institute for Molecular Studies (Port St. Lucie, Florida). This library format utilizes structurally organized mixtures, allowing high-throughput screening of millions of compounds in only several hundred samples.

Most library plates are formatted in 384-well source plates at 10 mM compounds in 100% DMSO. For a 384-well format, columns 3–22 contain compounds while columns 1–2 and 23–24 contain DMSO for use as controls. Library plates are kept at −40 °C in the presence of a desiccant (Drierite), thawed in a desiccator before use, and purged with argon/sealed immediately after use to minimize compounds’ contact with air.

When screening a large library of compounds, a pilot screen consisting of approximately four 384-well library plates is recommended for an initial assessment. The pilot screen should include selected representative plates from each type of library that will be screened. This allows initial assessment of hit rates, which can signify if the compound concentration is too low or too high and needs to be adjusted. The typical hit rate for a high-throughput screen is less than 1% (Shun, Lazo, Sharlow, & Johnston, 2011). If the pilot shows acceptable results, carry out the full high-throughput screening.

1. Prepare the master mix that will be used for compound samples and negative controls. The required volume (V) can be determined by V = 27 × n × p + D, where n is the number of wells per plate, p is the number of plates, and D is the dead volume for your chosen liquid dispensing method. Each well will contain a total volume of 30 μL. Before use, thaw ERK2 protein on ice. Dilute ERK2 and FITC-X-Lig-D in FA(+) buffer. The final master mix contains 1.11 μM ERK2 and 11.1 nM FITC-X-Lig-D.

2. Prepare positive control solution in the same way as the master mix, except without ERK2.

3. Design the assay plate layout based on the source plate layout described above. In each plate, columns 1–2 contain negative controls, 23–24 contain positive controls, and the rest (columns 3–22) contain test compounds. The compound locations are mapped 1:1 directly from each source plate to each assay plate well. This layout is shown in Fig. 5.

4. Label assay plates with barcodes referencing their corresponding library source plates.

5. Prepare compound dilution plates such that compounds from the source plates at 10 mM are diluted to 0.5 mM in FA(−) buffer.

6. Using the pipetting or liquid handling method of your choice, dispense 3 μL of 0.5 mM library compounds in 5% (v/v) DMSO (or 5% (v/v) DMSO for control columns) to each well of the plate (Fig. 5).

7. Dispense 27 μL of the master mix per well into columns 1–22 of each plate and 27 μL of the positive control solution into columns 23–24 (Fig. 5).

8. Centrifuge the plates at ambient temperature for 2 min at 800 RPM (127 × g).

9. Seal the plates using the automated plate sealer and incubate them at ambient temperature for 1 h.

10. Remove the plate seals and read the FA of the samples on the microplate reader. The resulting data can be processed using the software program of your choice.

11. Calculate the average and standard deviation of signals for both positive and negative controls for each plate. Use these statistics to calculate ${Z^{\prime }}_{OBS}$ (Eq. (19) in Section 3.3.2.2) for every plate. If ${Z^{\prime }}_{OBS}$ is lower than 0.5 for a plate, it must be assessed for outliers and errors and potentially re-screened.

12. For each well, calculate the percent signal change (P) relative to the controls in the same assay plate using Eq. (20). In this equation, ${\mu }_p$ and ${\mu }_n$ are the mean values of the positive and negative controls, and r is the FA of the test sample well.

    $$P=100\left(\frac{{\mu }_n-r}{{\mu }_n-{\mu }_p}\right)$$ (20)

13. Select a threshold for P to define a hit compound. For example, $P\ge 50$ would denote hits as all compounds that induced at least a 50% reduction in FA signal relative to controls. A reasonable hit rate should also be considered. Alternatively, samples can be ranked by P, and the top N number of compounds can be selected for further evaluation.

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.

In our example, we tested 23,534 compounds at 50 μM and selected molecules that caused over 50% reduction in FA signal from the negative controls as hits. This produced 204 hit compounds, indicating a hit rate of 0.87% with an overall average ${Z^{\prime }}_{OBS}$ of 0.86. A randomly selected sample of the screening results is shown in Fig. 6 to illustrate ${Z^{\prime }}_{OBS}$ and P.

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.

3.3.2.4. Confirm the hit molecules

1. Repeat the FA assay using 2–3 concentrations of each selected top compound from the screen to confirm they meet the criteria for a hit. In our confirmation, we test final concentrations of 5.5, 16.7, and 50 μM of each of the 204 compounds. We prepare 10× compounds by diluting 10 mM stocks to 0.5 mM of in FA(−) buffer, and diluting 3-fold twice.

2. We then repeat the FA assay by combining 3 μL of each compound dilution in FA(−) buffer and 27 μL of 1.11 μM ERK2 and 11.1 nM FITC-X-Lig-D in FA(+) buffer.

3. In parallel, add 3 μL of each compound dilution and 27 μL of FA(+) buffer in separate plates. These samples of compound alone in FA buffer will be used to evaluate intrinsic fluorescence of the compounds that can interfere with the FA measurements.

4. Include positive and negative controls as described in the high-throughput screening (Section 3.3.2.3).

5. Seal the plates manually with an aluminum seal and incubate at ambient temperature for 1 h. Then read the FA values.

6. Evaluate the hit compounds for potential false positives and false negatives. If it is initially assumed that the FITC-X-Lig-D binding state is the only influence on FA and the binding state does not affect FITC emission intensity, a library compound inducing a 50% signal reduction indicates that 50% ofbound FITC-X-Lig-D has been displaced from the DRS. However, factors aside from the relative amounts of free and bound FITC-X-Lig-D can affect FA. Intrinsic fluorescence of compounds at the emission wavelength of FITC-X-Lig-D and scattered light from aggregated compounds can generate false positives and negatives. The screening data’s most obvious false positives and negatives were compounds that generated percent signal changes over 100% or under 0%.

7. Eliminate molecules that cause false positives via their intrinsic fluorescence.

8. From the confirmed results of this validation, rank and select the top 10–20 compounds and obtain fresh purified stock solutions of each. These compounds should also be pre-screened for other criteria, such as commercial availability and/or ease of synthesis, toxicity, structural similarity/diversity, and other known biological functions. We use SciFinderⁿ (American Chemical Society), Pub Chem (National Institutes of Health), and literature searches to find this information.

3.3.3. Expected outcomes

The screen should result in a list of hit compounds corresponding to a hit rate of approximately 1% of total samples. The overall average ${Z^{\prime }}_{OBS}$ should ideally be greater than 0.7 and the average signals for the positive and negative controls should not significantly vary from day to day. There should be viable hit compounds remaining after ruling out false positives.

3.3.4. Quantification and statistical analysis

Use Eqs. (19) and (20) to analyze the high-throughput screening data.

3.3.5. Advantages

High-throughput format allows a large number of sample compounds to be tested with reduced labor, especially if liquid dispensing and plate-reading steps can be automated. The ability to use very small sample volumes in microplates allows for conservation of materials.

3.3.6. Limitations

The speed of the HTS is limited by the amount of time it takes to read the FA of each plate, the amount of time it takes to dispense reagents to each plate, and the stability of the reagents at room temperature. FA requires a relatively large concentration of protein per well, so ease of protein expression and purification or costs can restrict the screening size. Library size can also be cost- or time-limited.

3.3.7. Optimization and trouble-shooting

The optimization of the FA assay is described in Section 3.2. If the pilot screening results in too few or too many hit compounds, adjust the compound concentration up or down, respectively, to achieve a reasonable hit rate. If ${Z^{\prime }}_{OBS}$ for a given plate is less than 0.7, there is likely an issue with volume of liquid dispensed, and equipment should be assessed and the assay plate should be repeated. If the compounds do not pass the FA test of different doses described in Section 3.3.2.4, it is likely that the hit is an error due to sample contamination or liquid dispensing inconsistencies.

3.3.8. Safety considerations and standards

Use basic personal protective equipment (PPE) including gloves, eye protection, and lab coats. When operating automated liquid handlers, use caution around moving parts.

3.3.9. Alternative methods and procedures

Other types of compound libraries can be used. For example, see Section 3.3.2.3 for descriptions of two different approaches to library selection and formatting. Other additives can be included in the assay buffer that preferentially identify hits with certain characteristics: for example, a reducing agent such as dithiothreitol (DTT) can be included to eliminate thiol-reactive compounds.

3.4. Dose-dependent evaluation of hit compounds by fluorescence anisotropy

Hit compounds are evaluated for dose-dependent inhibition of FA signal for the cases of inactive ERK2 and the ERK2 mutant C159S. ERK2-C159S represents a point mutation that can potentially disrupt inhibitor binding to the DRS, assuming the inhibitor interacts with Cys-159. It is important to note that other mutations within the DRS (if possible) and further structural studies such as solution NMR, cysteine footprinting, and X-ray crystallography (Abramczyk, Rainey, Barnes, Martin, & Dalby, 2007; Piserchio et al., 2011; Sammons, Perry et al., 2019) should be used to confirm DRS binding.

3.4.1. Materials

Materials include the same reagents, consumables, instruments, and other tools listed in Section 3.2.1, in addition to the following reagents:

3.4.1.1. Reagents

1. Purchase pure inhibitors resulting from Section 3.3.2.4. In this section, we use the compound NSC 194308 obtained from National Cancer Institute (NCI) as an example.

2. DMSO (dimethyl sulfoxide) (Sigma-Aldrich, #276855)

3. Inactive, tagless ERK2 (Section 3.1)

4. Inactive ERK2 with the point mutation C159S (Section 3.1)

3.4.2. Methods

The top confirmed compounds from the screen in Section 3.3.2.4 should be assessed for potency by dose—response FA experiments like Lig-D(Dap) in Section 3.2.2.3. We quantified the potency of the compounds against two variants of ERK2: WT ERK2 (the same used in the screen) and ERK2 with the single point mutation C159S. Cys-159 is located in the DRS and is susceptible to covalent modification by ERK2 inhibitors (Kaoud et al., 2019). In this example, we predicted that hit compounds do not bind to the C159S mutant.

3.4.2.1. Dose–response assay using inactive ERK2

1. Prepare each hit compound by dissolving the solid form in DMSO to make a 10 mM stock. In this protocol, we use NSC 194308 as a representative hit compound.

2. Incubate varied concentrations of NSC 194308 (0–200 μM, 2-fold dilutions, 8 total concentrations) with 10 nM FITC-X-Lig-D and 1 μM ERK2 in FA buffer for 30 min at ambient temperature in a microplate. For controls, use the same concentrations of NSC 194308 (1) with buffer only and (2) with FITC-X-Lig-D in buffer only.

3. Read the FA of the samples at ambient temperature using a plate reader.

4. Plot FA as a function of NSC 194308 concentrations (Fig. 7).

5. Calculate the binding affinity K_I for NSC 194308 using the protocol described in Section 3.2.2.4 using the binding affinity of ERK2 and FITC-X-Lig-D from Section 3.2.2.2. The results of this analysis are shown in Fig. 7, where K_I for NSC 194308 is 0.73 ± 0.26 μM.

6. This process must be repeated for all hit compounds to compare their potencies and select which molecules should be studied further.

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.

3.4.2.2. Dose–response assay using C159S mutant inactive ERK2

1. Prepare ERK2 with a single point mutation of C159S (ERK2-C159S) in the same manner as inactive ERK2 in Section 3.1.

2. ERK2-C159S must be assessed for differences in binding to the FITC-X-Lig-D peptide, so re-optimizing the FA assay should be done in the same manner as Section 3.2.2.1. For the FITC-X-Lig-D concentration, aim to match the signal window of the ERK2 FA assay while conserving ligand and maintaining the tight-binding model. In our example, the optimal concentration of FITC-X-Lig-D is first fixed at 10 nM to match the ERK2 assay. ERK2-C159S concentrations are varied from 0 to 12 μM, and the FA measurements were recorded after incubating samples for approximately 30 min at ambient temperature.

3. Plot the FA as a function of ERK2-C159S concentration (Fig. 8).

4. Fitting the data to Eq. (3) yields an EC₅₀ value of 4.7 ± 0.9 μM. (This fit is not shown in Fig. 8).

5. ERK2 concentration is optimized to be 5 μM from rounding the EC₅₀ value.

6. Fitting the data as described in Section 3.2.2.2, the K_D is 8.4 ± 1.6 μM (Fig. 8). This indicates weaker binding between FITC-X-Lig-D and ERK2-C159S than observed for ERK2, which makes sense since the mutated residue is located in the binding groove for FITC-X-Lig-D.

7. To maintain the same ratio of ERK2-C159S to FITC-X-Lig-D as the original FA assay, we increased the FITC-X-Lig-D concentration to 50 nM. We know from Fig. 1 that the concentration range of 10–100 nM FITC-X-Lig-D does not significantly produce variation in the assay results.

8. For a positive control, follow the protocol described in Section 3.2.2.3 using Lig-D(Dap). Fix ERK2-C159S at 5 μM and FITC-X-Lig-D at 50 nM, vary the concentration of Lig-D(Dap) (0–200 μM, 2-fold dilutions, 8 total concentrations), and carry out the FA assay in the same manner as for ERK2 in Section 3.4.2.1.

9. Plot FA as a function of Lig-D(Dap) concentration, as shown in Fig. 9A. Fitting these data by the process described in Section 3.2.2.4, the K_I is 17.5 ± 3.7 μM.

10. Fix ERK2-C159S at 5 μM and FITC-X-Lig-D at 50 nM, vary the concentration of NSC 194308 (0–200 μM, 2-fold dilutions, 8 total concentrations), and carry out the FA assay in the same manner as for ERK2 in Section 3.4.2.1.

11. Plot FA as a function of NSC 194308 concentration, as shown in Fig. 9B. Unlike Lig-D(Dap), NSC 194308 fails to displace FITC-X-Lig-D from ERK2-C159S over the concentration range of 0–200 μM. This suggests the Cys-159 residue is critical for NSC 194308 interaction with ERK2.

12. Compare the results for all lead molecules from the screening. Note that any observed inhibition of ERK2-C159S does not rule out DRS binding. Other point mutations or structural studies should be used to determine if the molecules bind at other locations within the DRS.

13. This process can be repeated for other DRS mutants, assuming that the mutations do not prevent the binding of FITC-X-Lig-D to ERK2.

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.

3.4.3. Expected outcomes

Compounds should have inhibition potencies against ERK2 that align with their ranking from the screen. Hit compounds are predicted to bind to the DRS but they may or may not interact with C159 since it is a point mutation within a larger binding site. Compounds that displace FITC-X-Lig-D from ERK2-C159S may interact with other residues in the DRS and should be further evaluated. Cysteine-reactive inhibitors are expected to covalently modify C159, so they are predicted to not inhibit ERK2- C159S binding to FITC-X-Lig-D.

3.4.4. Quantification and statistical analysis

The binding affinity of FITC-X-Lig-D for each ERK2 mutant should be evaluated as in Section 3.2.2.2. The binding affinities of the inhibitors for each form of ERK2 should be quantified in the same manner as Lig-D (Dap) in Section 3.2.2.4.

3.4.5. Advantages

An advantage of using this method to evaluate binding affinity is the FP assay is already optimized by following the previous sections in this manuscript, and therefore these dose–response experiments can be used to validate the screening results. The FP assay is also easy to adapt to different ERK2 mutants, and allows meaningful comparison of binding affinities across different forms of ERK2.

3.4.6. Limitations

Preparing mutants of ERK2, especially if expressing and purifying them inhouse, can be time-consuming and will not necessarily yield protein that can bind strongly enough to FITC-X-Lig-D to produce a good FA signal. Choice of mutations should be considered carefully, using the predicted binding mode of FITC-X-Lig-D to the FRS to avoid disrupting key interactions and ensuring the mutations do not significantly alter the conformation of ERK2 or cause it to denature.

3.4.7. Optimization and trouble-shooting

Hit compounds may exhibit dose-dependent effects on the FA signal that are independent of their desired binding to the DRS, either through their intrinsic optical properties or interactions with FITC-X-Lig-D. These effects may not be significantly detectable at the original screening concentration or the confirmation tests described in Section 3.3.2.4, but could appear as the compound concentrations increase to the higher doses used in this procedure. Controls that test the background dose-dependent effects of the compounds on FITC-X-Lig-D and the intrinsic signal of compounds alone in assay buffer can be used to assess and potentially correct for these undesired FA signal influences. Another potential issue is that the pure purchased compounds might give different results compared to previous experiments if the samples in library plates are less pure. The library samples can be evaluated by NMR to check for impurities, and potentially these impurities may be unexpected inhibitors that can be further investigated. Alternatively, the compounds may be eliminated from further studies at this stage.

3.4.8. Safety considerations and standards

Use basic personal protective equipment (PPE) including gloves, eye protection, and lab coats.

3.4.9. Alternative methods and procedures

See Section 3.2.9 for examples of other types of assays to measure binding of ligands or inhibitors to proteins.

4. Evaluating hit small molecules for inhibition of ERK2 functions

There are a multitude of methods that can be used to assess the effects of small molecules binding to the DRS. Here, we describe two examples: a fluorescence-based ERK2 activity assay to measure inhibition of substrate phosphorylation and an assay to measure inhibition of ERK2 activation by MKK1G7B using a Western blot detection method.

4.1. Fluorescence-based ERK2 activity assay

Up to this point, our evaluation of lead compounds from the screen has dealt with binding to ERK2. ERK2 phosphorylates many substrates in cells, so it is also crucial to evaluate the effect of these compounds on this function of ERK2. To this end, we utilize a peptide substrate of ERK2 labeled with the sulfonamido-oxine (Sox) fluorophore, which produces a fluorescence signal upon phosphorylation by ERK2. The Sox moiety and the phosphorylated residue produce fluorescence signal upon magnesium ion chelation. In this section, we describe how to evaluate the Michaelis–Menten kinetics (Srinivasan, 2022) of Sox-labeled substrate phosphorylation by ERK2 and use this information to evaluate the $I{C}_{50}$ and K_I values for inhibitors of this enzymatic reaction.

4.1.1. Materials

4.1.1.1. Reagents

1. Same buffer components as in Section 3.2.1.1 (HEPES, KCl, EDTA, EGTA)

2. Same buffer additives as in Section 3.2.1.1 (Triton X-100 and BSA)

3. Adenosine 5′-triphosphate disodium salt trihydrate (ATP) (Roche, #10519987001)

4. 1 M Magnesium chloride (MgCl2) (Millipore Sigma, #M1028)

5. Pure inhibitors resulting from Section 3.3.2.4 (we use the compound NSC 194308 obtained from National Cancer Institute as an example)

6. Recombinant ERK2 protein, active (Section 3.1)

7. The Sox-STE7 peptide with the sequence FQRKTLQRRNLKGL-NLNL-XXX-TGPLSP-C(Sox)-PF is required. Sox-labeled STE7 was synthesized by alkylation of Sox-Br (2-bromomethyl-8-tertbutyldi-phenylsilyloxy-5- (N, N-dimethyl) sulfonamide quinolone) to the cysteine residue of the peptide, which was purchased on resin from Anaspec (PB# 1570512). The protocol for this labeling procedure is described in a previous study (Zamora-Olivares et al., 2014). Sox-Br was synthesized and provided by William H. Johnson at the University of Texas at Austin. Sox-STE7 shares most of the sequence of FITC-X-Lig-D, except it is extended to include a phosphorylation consensus sequence and is labeled with the Sox fluorophore rather than FITC, so phosphorylation of the peptide can be monitored.

4.1.1.2. Consumables

Consumables are the same as listed in Section 3.2.1.2.

4.1.1.3. Tools and Instruments

Tools and instruments are the same as in Section 3.2.1.3. Additionally, the fluorometer must have monochromators or filters capable of excitation and emission wavelengths of 360 nm and 482 nm, respectively. In our case, the Synergy H4 microplate reader meets this requirement.

4.1.1.4. Software

Software is the same as Section 3.2.1.4, e.g., GraphPad Prism (recommended version 7 or higher).

4.1.2. Methods

4.1.2.1. Evaluate the Michaelis–Menten kinetics for Sox-STE7 phosphorylation by ERK2

1. Assay buffer must first be prepared using 25mM HEPES–KOH, 50mM KCl, 0.1mM EDTA, and 0.1mM EGTA, pH 7.5. It is critical to filter the buffer through a 0.2–0.45-micron filter to remove contaminants and degas by bubbling it with an available inert gas, such as nitrogen or argon, for at least 15 min.

2. Before the experiment begins, fresh buffer additives BSA, MgCl₂, and Triton X-100 are added to the assay buffer. They are initially prepared to concentrations of 1M MgCl₂, 10 mg/mL BSA and 10% w/v Triton X-100. The final additive concentrations in 1× activity assay buffer are 10mM MgCl₂, 10 μg/mL BSA, and 0.01% Triton X-100.

3. In all samples, prepare MgCl₂ and ATP in a 1:1 ratio to form the MgATP complex, such that the final concentration is 1mM MgATP in 35 μL. Note that MgCl₂ is also present in the buffer at 10mM (excess) for chelation of Mg²⁺ with the Sox fluorophore. Add Sox-STE7 such that the final concentrations of Sox-STE7 in 35 μL assay volumes range from 0 to 50 μM.

4. Reactions are initiated by adding ERK2 to a final concentration of 2 nM in the 35 μL assay volume.

5. Read the fluorescence signal at ambient temperature for 1 h with a kinetic interval time of 10 s. Excitation and emission wavelengths are 360 nm and 482 nM, respectively.

6. Initial reaction rates occur during the duration of time where the fluorescence vs. time plot at all Sox-STE7 concentrations are linear to meet the experimental requirements for the Michaelis–Menten kinetics model. In the example of Fig. 10A, the initial rate period is 4 min.

7. Initial reaction rates are calculated as the slopes of the data from step 6 fit by simple linear regression (Fig. 10A). The slopes of fluorescence vs. time under the initial rate conditions are directly proportional to the rate of reaction, since the fluorescence signal is directly proportional to the concentration of phosphorylated Sox-STE7 (Devkota, Kaoud, Warthaka, & Dalby, 2010).

8. The initial rates (v) as a function of Sox-STE7 concentration ([S]) are then fit to the Michaelis–Menten equation (Eq. (21)) to determine the maximal rate of reaction $\left(v_{\max }\right)$ and the Michaelis constant $\left(K_m\right)$. The fit is shown in Fig. 10B where $v_{\max }$ is 20.5 ± 1.0 RFU/s and $K_m$ is 4.3 ± 0.7 μM.

   $$v=\frac{v_{\max }\left[S\right]}{K_m+\left[S\right]}$$ (21)

9. The Michaelis constant $\left(K_m\right)$ will be used in the following experiment in Section 4.1.2.2 to approximate the binding affinity of ERK2 inhibitors from the screen.

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.

4.1.2.2. Evaluate the $I{C}_{50}$ and $K_I$ values for inhibitors Sox-STE7 phosphorylation by ERK2

1. Compounds are pre-incubated with ERK2 for 20 min at ambient temperature such that the final concentrations of compounds and ERK2 in the assay volume of 35 μL are 0–100 μM and 2 nM, respectively.

2. Reactions are initiated by adding MgATP and Sox-STE7 to final concentrations of 1mM and 2 μM, respectively, with a final assay volume of 35 μL (Zamora-Olivares et al., 2014).

3. Read the fluorescence signal at ambient temperature for 1 h with a kinetic interval time of 10 s. Excitation and emission wavelengths are 360 nm and 482 nM, respectively.

4. Initial reaction rates (RFU/s) were set as the first 4 min and calculated by linear regression, as explained in Section 4.1.2.1.

5. Convert the rates (RFU/s) at each inhibitor concentration to percent inhibition by normalizing each rate to the rate at 0 μM inhibitor (100% activity control).

6. Plot the percent inhibition as a function of inhibitor concentration. We show data for NSC 194308 in Fig. 11 as an example.

7. Fit the data from step 6 to Eq. (12) to evaluate the $I{C}_{50}$. From fitting our example, the $I{C}_{50}$ is 1.6 ± 0.5 μM.

8. Compare the $I{C}_{50}$ values for all lead molecules to evaluate their relative effects on ERK2 activity and select the most potent molecules for further study.

9. From the $I{C}_{50}$ we can also approximate the affinity of the inhibitor for ERK2 using the Cheng–Prusoff equation (Eq. (22)).

   $$K_I=\frac{I{C}_{50}}{1+\frac{\left[S\right]}{K_m}}$$ (22)

   In our example of NSC 194308, the $I{C}_{50}$ is 1.7 μM and the Sox-STE7 concentration ([S]) is 2 μM. If we use the $K_m$ value obtained in Section 4.1.2.1 of 4.3 μM we can calculate that $K_I$ is approximately 1.1 μM.

10. Compare the K_I from this method to the K_I determined in Section 3.4.2.1 (0.73 ± 0.26 μM). Both methods should yield a similar value.

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.

4.1.3. Expected outcomes

Inhibitors that block the binding of FITC-X-Lig-D to ERK2 should also prevent Sox-STE7 phosphorylation. Relative binding affinities among the inhibitors are expected to align with the FA dose–response results because the ligand and substrate share the same DRS-binding sequence.

4.1.4. Quantification and statistical analysis

Quantify the Michaelis–Menten kinetics of Sox-STE7 by ERK2 using the analysis in Section 4.1.2.1. Evaluate the inhibitor potencies as directed in Section 4.1.2.2.

4.1.5. Advantages

This method allows for the real-time measurements of substrate phosphorylation and does not require quenching of the reactions at varied time points to evaluate reaction rates. Since the fluorescence signal is linearly proportional to the concentration of phosphorylated substrate, data analysis is straightforward. Since the substrate peptide is similar to the FA ligand, it allows a relevant comparison between the two types of methods and assessment of how docking site inhibition affects ERK2 activity. A conventional method of evaluating reaction rates involves using radiolabeled ATP and measuring the radioactivity of the phosphorylated substrate at different time points. An example is described in our previous work (Kaoud et al., 2019). Using Sox-STE7 and fluorescence detection avoids safety issues of working with radioactive material.

4.1.6. Limitations

Since Sox-STE7 is a designed peptide, it is not a true substrate of ERK2. If possible, ERK2 phosphorylation of full protein substrates should be assessed as well. The excitation wavelength of Sox is in the UV range, at which biomolecules and compounds may be optically active. Compounds may also undergo UV degradation. These issues can exclude certain hit compounds from this procedure.

4.1.7. Optimization and trouble-shooting

Some small molecules are intrinsically fluorescent or absorb at the excitation and emission wavelengths of Sox, so they should be tested alone in assay buffer to rule out interference. Depending on the specific activity of your ERK2, its concentration can be adjusted to ensure the linear initial reaction phase is long enough to collect sufficient data. If the ERK2 concentration needs to be changed, then the Michaelis Menten kinetics experiment should be repeated at the new concentration of ERK2 and the parameters applied in Eq. (22). If no fluorescence is observed, ensure that magnesium chloride is present in the buffer at 10mM excess, as it is critical for Sox fluorescence.

4.1.8. Safety considerations and standards

Use basic personal protective equipment (PPE) including gloves, eye protection, and lab coats.

4.1.9. Alternative methods and procedures

A standard alternative method for measuring ERK2 substrate phosphorylation is referenced in Section 4.1.5.

4.2. Detecting activation of ERK2 by MKK1G7B in an in-vitro phosphorylation assay

The DRS is a known interaction site of MKK1 in the process of ERK2 activation (Tanoue et al., 2000), so we also assessed lead compounds for their ability to block ERK2 phosphorylation by the constitutively active mutant of MKK1, MKK1G7B.

4.2.1. Materials

4.2.1.1. Reagents

1. Assay buffer components and additives from Section 4.1.1.1 (HEPES, KCl, EDTA, EGTA, ATP, MgCl₂, Triton X-100, BSA)

2. Dithiothreitol (DTT) (GoldBio, #DTT100)

3. Proteins: inactive ERK2 and MKK1G7B (Section 3.1)

4. Inhibitors (example here is compound 2507-1 from our previous work (Sammons, Perry et al., 2019) synthesized by Torrey Pines Institute for Molecular Studies)

5. Running buffers, 5× loading buffer, protein ladders, and 10% (w/v) SDS gels for running SDS-PAGE are required. This protocol assumes prior knowledge of SDS-PAGE and that these components are standard lab materials.

6. Odyssey Blocking Buffer (LI-COR, #927-40000)

7. Tris base (Sigma-Aldrich, #T6066)

8. Glycine (Sigma-Aldrich, #G7126)

9. Sodium Chloride (NaCl) (Sigma-Aldrich, #S3014)

10. Tween 20 (Sigma-Aldrich, #P9416)

11. Methanol (Thermo Scientific, #423950040)

12. Blotting-grade blocker (non-fat dry milk) (Bio-Rad, #1706404)

13. Antiphospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) (E10) mouse mAb (Cell Signaling Technology, #9106)

14. Anti-p44/42 MAPK (Erk1/2) (137F5) rabbit mAb (Cell Signaling Technology, #4695)

15. IRDye 800CW Goat (polyclonal) anti-rabbit IgG (LI-COR, #926-32211)

16. IRDye 680RD Goat (polyclonal) anti-mouse IgG (LI-COR, #926-68070)

4.2.1.2. Consumables

1. Pipettes and pipette tips

2. Filters (0.2–0.45 μm, sterile)

3. Various sizes of plastic tubes and liquid containers as needed

4. Immobilon-FL PVDF membranes (Millipore, #IPFL00010)

5. 3 MW Blotting paper (0.34 mm) (MIDSCI, #3 MW-1417)

6. Black Western blot incubation boxes (e.g., LI-COR 929-97305)

4.2.1.3. Tools and Instruments

1. Equipment for running SDS-PAGE and blot transfers, e.g., Mini-PROTEAN Tetra Cell, Mini Trans-Blot Module, and PowerPac HC Power Supply (Bio-Rad, #1658035)

2. Odyssey Sa imaging system (LI-COR)

3. Orbital shaker

4.2.1.4. Software

1. Data analysis software, e.g., GraphPad Prism (recommended version 7 or higher)

2. LI-COR Image Studio image processing software and/or ImageJ

4.2.2. Methods

4.2.2.1. Assay for inhibition of ERK2 phosphorylation by MKK1G7B

1. Prepare assay buffer as described in Section 4.1.2.1: 25mM HEPES–KOH, 50mM KCl, 0.1mM EDTA, and 0.1mM EGTA, pH 7.5. Add additives fresh additives to concentrations of 2mM DTT, 0.5mM MgCl₂, 10μg/mL BSA, and 0.01% (v/v) Triton X-100.

2. In assay buffer, incubate 0, 50, 100, and 200 μM compound 2507-1 (final 2% DMSO) and 1 μM inactive ERK2 (concentrations in final assay volume). The incubation is carried out for 10 min at 28 °C. The final assay volume is 70 μL per reaction.

3. In all reactions, MgCl₂ and ATP are combined in a 1:1 ratio to form the MgATP substrate. Initiate the reactions by adding MKK1G7B and MgATP to 20 nM and 10mM final concentrations, respectively. Note that 0.5mM MgCl₂ is also present in the assay buffer (step 1) so Mg²⁺ is in excess. The final assay volume is 70 μL per reaction.

4. Quench each reaction at a series of time points ranging from 0 to 30 min by adding 5 μL of the reaction mixture into 4 μL SDS-PAGE loading buffer (5× concentrated) and 11 μL water.

5. Dilute 5 μL of each quenched solution into 15 μL SDS-PAGE loading buffer. Final protein amounts are approximately 50 ng per sample.

6. Prepare control samples for 0% phosphorylated (inactive) ERK2 and 100% dually phosphorylated (active) ERK2 in 1× SDS-PAGE loading buffer to amounts of 50 ng per sample.

7. Heat the samples in loading buffer at 95 °C for 5 min

8. Load the samples into 10% SDS gels and run SDS-PAGE.

9. Prepare Tris–Glycine buffer: 3.03 g Tris base, 14.4 g glycine, 200 mL methanol, and 800 mL water.

10. Transfer the gels to Immobilon-FL membranes at 4 °C in Tris–Glycine buffer. Use 3 MW blotting paper to sandwich the membrane and gel in the blot transfer apparatus.

11. Prepare a 10× Tris-buffered saline (TBS) buffer of 500mM Tris pH 7.4 and 1.5M NaCl. Dilute this 10-fold to make 1× TBS. Similarly, 1× TBST is prepared by making the same dilution and including 0.1% (v/v) Tween 20.

12. In Western blot incubation boxes, block the membranes using 4× diluted Odyssey blocking buffer in TBS for 1 h at ambient temperature with gentle shaking on an orbital shaker. Keep the blocking solution protected from light.

13. BSA-TBST is prepared by combining 5 g BSA per 100 mL TBST. Similarly, NFDM-TBST is prepared by combining 5 g blotting-grade blocker (non-fat dry milk) per 100 mL TBST.

14. Membranes are then incubated in NFDM-TBST with 2000-fold diluted antiphospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) (E10) mouse mAb, or BSA-TBST with 1000-fold diluted anti-p44/42 MAPK (Erk1/2) (137F5) rabbit mAb for 1 h at ambient temperature. The total ERK protein antibody is used as a loading control. The incubation is performed in Western blot incubation boxes with gentle shaking on an orbital shaker.

15. The membranes are washed with TBST 3–5 times, and then incubated with appropriate secondary antibodies (15,000-fold diluted IRDye 800CW Goat (polyclonal) anti-rabbit IgG or IRDye 680RD Goat (polyclonal) anti-mouse IgG) for 30 min at ambient temperature in BSA-TBST. The incubation and washing steps are performed in Western blot incubation boxes with gentle shaking on an orbital shaker. The secondary antibody solutions and membranes should be protected from light with aluminum foil and by using black opaque Western blot incubation boxes.

16. After washing again with TBST 3–5 times, followed by rinsing with TBS, fluorescence images of the membrane are collected on an Odyssey Sa imaging system. The resulting images of phosphorylated ERK2 can be seen in Fig. 12A.

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).

4.2.2.2. Data analysis

1. Using the LI-COR image studio software, quantify the intensity of the protein bands.

2. Normalize the phosphorylated band intensities using the active and inactive ERK2 controls as 100% and 0% phosphorylation, respectively. Plot these values as a function of quenching time. These results are shown in Fig. 12B.

3. Calculate the reaction rates (% phosphorylation/min) as described in Section 4.1.2.2.

4. Plot these rates as a function of inhibitor concentration, and fit the resulting data to the $I{C}_{50}$ equation (Eq. (12)). From the fit, the calculated $I{C}_{50}$ is 9.9 ± 1.9 μM.

5. Repeat the protocol for other lead compounds from the screen and compare the $I{C}_{50}$ values to evaluate their relative effects on ERK2 activation. Select the most potent molecules for further study.

4.2.3. Expected outcomes

Since MKK1G7B docks at the DRS, compounds that bind to the DRS should prevent MKK1G7B from binding and subsequently activating ERK2.

4.2.4. Quantification and statistical analysis

Data analysis is described in Section 4.2.2.2.

4.2.5. Advantages

This is a quick and simple method to evaluate protein phosphorylation that only requires knowledge of SDS-PAGE and Western blot protocols. Furthermore, it only requires the recombinant proteins that are used in Section 3.1 so no additional protein purification or purchasing is required for this method. Antibodies and materials can also be used for cell-based evaluation of inhibitors.

4.2.6. Limitations

Since this method uses recombinant proteins and is not a cell-based experiment, these results do not necessarily translate to cells where the environment and the ERK signaling pathway are much more complex. Hit compounds may not enter cells, may be rapidly metabolized, or may target other cellular proteins with higher potency.

4.2.7. Optimization and trouble-shooting

If signal is poor, you may have to experiment and optimize incubation times of antibodies with the membranes or utilize different antibodies altogether. If background noise is high, there is likely an issue with the membrane blocking step or handling of the membranes. When the membrane is read by the instrument, ensure there are no bubbles present as they can interfere with the signal. Do not let the membranes dry out at any point. If the SDS-PAGE or the transfer from gel to membrane is unsuccessful, check that the positive and negative terminals of the power supply have not been reversed. When setting up the Western blot transfer, ensure there are no bubbles between the membrane and the gel.

4.2.8. Safety considerations and standards

Use basic personal protective equipment (PPE) including gloves, eye protection, and lab coats.

4.2.9. Alternative methods and procedures

Instead of using an antibody for phosphorylated ERK, antibodies for detection of threonine/tyrosine phosphorylation can be used (e.g., Cell Signaling #9381) or Phos-tag SDS-PAGE that separates proteins based on phosphorylation state (e.g., Thermo Scientific, #NC0232095). Other methods can evaluate phosphorylation of proteins, for example, autoradiography using radioactive ATP, immunoprecipitation of phosphory-lated proteins, and mass spectrometry. An alternative option to measure MKK1 activation of ERK2 in the presence of inhibitor is to use cell-based assays. In such a case, the cell line of choice would be incubated with doses of the inhibitor, the ERK2 pathway would be stimulated by an extracellular ligand like EGF (if necessary), and the cells collected and lysed at different time points. Phosphorylated ERK could then be quantified by Western blot.

5. Summary

In this chapter, we describe the steps for optimizing and executing a high-throughput screening for identifying inhibitors of ERK2 that target the DRS. This screening assay uses fluorescence anisotropy to detect displacement of the fluorescent peptide FITC-X-Lig-D from the DRS of ERK2. Lead compounds selected from the screen are validated by a FA dose–response assay using ERK2 and the mutant ERK2-C159S.

Lead compounds are further validated in an ERK2 activity assay that detects the fluorescence of a peptide substrate, Sox-STE7, which is proportional to the concentration of the phosphorylated peptide. This indicates that inhibitors that block binding at the DRS can also block substrate phosphorylation if substrates utilize the DRS for docking interactions.

Lead compounds are also evaluated for their ability to block ERK2 activation by MKK1G7B, a constitutively active MKK1 Mutant. MKK1 interacts with ERK2 at the DRS, so inhibitors that target the DRS can disrupt ERK2 phosphorylation.

Our protocols in this chapter focus on biochemical assays and data analysis methods, which can be used to identify DRS inhibitors as a starting point for further studies. It is important to note that these assays should be supplemented by additional validation experiments, such as structural studies to confirm inhibitor binding at the DRS, further mutational studies, cell experiments, and in vivo models. Further examples of DRS inhibitor characterization methods can be found in our previous work (Kaoud et al., 2019; Sammons, Perry et al., 2019). The methods in this chapter can potentially produce inhibitors of ERK2 that can block ERK pathway signaling in cancer for a therapeutic effect, as we have seen in the case of BI-78D3 (Kaoud et al., 2019).

Abbreviations

ATP

adenosine 5′-triphosphate

BRAF

v-raf murine sarcoma viral oncogene homolog B1 gene, a gene that encodes an isoform of Raf protein, B-Raf

BSA

bovine serum albumin

CD

common docking

Dap

2,3-diaminopropionic acid

DMSO

dimethyl sulfoxide

DNA

deoxyribonucleic acid

ERK1/2

extracellular signal-regulated kinases 1 and 2

DRS

D-recruitment site

EC ₅₀

half-maximal effective concentration

EDTA

ethylenediaminetetraacetic acid

EGTA

ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid

FA

fluorescence anisotropy

FITC

fluorescein isothiocyanate

FPLC

fast protein liquid chromatography

FRS

F-recruitment site

g

gravity

HEPES

N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid

$I{C}_{50}$

half-maximal inhibitory concentration

IgG

immunoglobulin G

IPTG

isopropyl β-d-1 thyogalactopyranoside

K_D

dissociation constant describing the binding affinity between protein and ligand

K_I

dissociation constant describing the binding affinity between enzyme and inhibiton

K_m

Michaelis constant, or the concentration of substrate required for an enzyme to reach half-maximal reaction rate

LB

lysogeny broth

mAb

monoclonal antibody

MEK1/2

mitogen-activated protein kinase kinase 1 and 2, synonymous with MKK1/2

MKK1/2

mitogen-activated protein kinase kinase 1 and 2, synonymous with MEK1/2

MgATP

1-to-1 molar ratio of magnesium chloride to ATP

NFDM

non-fat dry milk

Ni-NTA

Ni²⁺ ion coupled to nitrilotriacetic acid

NMR

nuclear magnetic resonance spectroscopy

Raf

rapidly accelerating fibrosarcoma protein family

RPM

revolutions per minute

SDS-PAGE

sodium dodecyl-sulfate polyacrylamide gel electrophoresis

Sox

sulfonamido-oxine

TBS

Tris-buffered saline

TBST

Tris-buffered saline with 0.1% (v/v) Tween 20

TEV

tobacco etch virus