A high-throughput fluorescence lifetime-based assay to detect binding of myosin-binding protein C to F-actin

Actin-binding proteins (ABPs) play key roles in skeletal and cardiac muscle (dos Remedios and Thomas, 2001). In cardiac muscle, ABPs are involved in the assembly, organization, and maintenance of the sarcomeres for normal contraction. Examples of muscle ABPs include the motor protein myosin, which interacts with actin to produce force; the tropomyosin (Tm)–troponin complex, which regulates actin’s exposure to myosin to regulate force production; and nebulin and leiomodin/tropomodulin proteins, which regulate thin filament length. Additionally, a thick filament–associated sarcomeric protein, cardiac myosin-binding protein C (cMyBP-C), is an ABP that is anchored to myosin and titin at its C-terminal end and extends toward actin at its N terminus to modulate the strength and speed of myosin cross-bridge cycling (for review, see van Dijk et al., 2014; Moss et al., 2015). Mutations in cMyBP-C, myosin, Tm, and troponin have been shown to cause cardiomyopathy in humans (Ingles et al., 2019). Mutations in cMyBP-C are a leading cause of hypertrophic cardiomyopathy (HCM; Ingles et al., 2019), including truncation mutations that lead to haploinsufficiency and missense mutations that alter binding function. cMyBP-C function is regulated by PKA-mediated phosphorylation, which reduces N-terminal cMyBP-C binding to actin (Shaffer et al., 2009) and myosin (Gruen et al., 1999). Decreases in phosphorylation of cMyBP-C have been observed in heart failure of human patients and animal models (El-Armouche et al., 2007), and phosphorylation is cardioprotective in mouse models (Sadayappan et al., 2006). Thus, increases in phosphorylation or drugs that mimic phosphorylation of cMyBP-C could be a therapeutic option for heart failure. As there are no therapies targeting cMyBP-C or its interactions with actin (or myosin), developing screening technology to identify drugs that modulate actin binding and/or mimic phosphorylation of cMyBP-C is the focus of this work.

The binding of filamentous actin (F-actin) to ABPs, including cMyBP-C, can be studied using a wide variety of technologies, including cosedimentation, affinity chromatography, electron microscopy, immunofluorescence imaging, and transient phosphorescence anisotropy. These methods are useful for studying details of ABP–actin interactions but are labor intensive and not scalable to the degree required for use in screening libraries of chemicals for potential therapeutic drugs. They also limit the number of variables that can be tested in the binding assay.

The development of multiwell plate readers with the capacity to precisely measure fluorescence lifetimes (time-resolved fluorescence; TR-F) in the nanosecond time scale gives rise to the potential of using changes in lifetime to monitor binding of a labeled protein, such as actin, to an unlabeled binding partner. The lifetime of a fluorescent molecule (e.g., 5-((((2-iodoacetyl)amino)ethyl)amino)naphthalene-1-sulfonic acid; IAEDANS) is the time it remains in the excited state before returning to the ground state. Following an excitation pulse (e.g., from a microchip laser or LED source), excited molecules return to the ground state over a few nanoseconds. TR-F is capable of observing the fluorescence emission levels over this time. In addition to radiative decay, the emission of photons as a portion of the excited fluorescent molecules return to the ground state, excited molecules can return to the ground state by nonradiative decay. Nonradiative decay occurs by energy transfer from a subpopulation of the excited molecules to the surrounding medium (buffer or, in the case of a probe attached to a protein, side chains of residues that are close to the probe). Changes in the environment of a fluorescent probe attached to actin due to interactions with another protein (an ABP) in the vicinity of that probe can increase or decrease the population of excited molecules that experience nonradiative decay. Changes in the nonradiative decay population influence the excited molecules available for radiative decay (fluorescence) and consequently can change the fluorescence lifetime. Changes in lifetime of a fluorescent probe on actin can therefore reflect binding of an ABP to actin if the interaction site is in close proximity to the label. Quantitatively, the degree of lifetime change reflects the number of actin–ABP complexes and can be used to generate binding curves. Changes in lifetime may be small (<10%), but the high precision with which they can be measured is such that even a 1% change can be reliably measured and used for screening chemical libraries for potential drugs that alter binding (Gruber et al., 2014) or testing of many physiological variables/biochemical conditions for mechanistic understanding. Changes in fluorescence indicative of actin binding, in a multiwell plate reader format, represent a potentially powerful screening tool to identify drugs that modulate ABPs and a high-throughput tool for scientific discovery.

To develop a simple, quantitative F-actin–binding assay in a 384-well plate format, we have screened potential fluorescent probes that when covalently attached to actin at position Cys-374 report binding, via change in lifetime. Actin’s Cys-374 is naturally occurring and solvent exposed. This is in contrast to the other four cysteines in actin that are buried and not labeled by the reactive dyes under our labeling conditions (Otterbein et al., 2001; Bunch et al., 2019). All the probes we have tested showed a decrease in lifetime upon binding to cMyBP-C N-terminal domains C0 through C2 (C0–C2, whereas full-length cMyBP-C includes domains C0–C10). IAEDANS-labeled F-actin, in the plate reader format, allowed for rapid and easy optimization of pH and ionic strength. Using our optimized conditions, we tested this system’s ability to distinguish between binding mediated by phosphorylated and unphosphorylated cMyBP-C as well as HCM mutants or functional mutations in the Tm-binding region of cMyBP-C. The TR-F assay was able to detect the predicted binding changes to both actin and actin–Tm for variations in cMyBP-C. This finding suggests that it will be valuable in drug screening, assaying binding conditions of ABPs (e.g., PKA phosphorylation, buffer, pH, and Ca²⁺), and evaluation of mutations with unknown effects on actin binding. In our analysis of the HCM mutations in cMyBP-C, we also identified defects in phosphorylation and subsequent actin binding as a potential cause of the mutant phenotype in the cMyBP-C R282W mutation.

Actin was prepared from rabbit skeletal muscle by extracting acetone powder in cold water as described in Bunch et al. (2018). The day before actin-binding experiments (cosedimentation or TR-F), globular actin (G-actin) was polymerized by the addition of MgCl₂ to a final concentration of 3 mM for 1 h at 23°C. F-actin was collected by centrifugation at 4°C, 100,000 rpm (350,000 ×g) in a Beckman TLA-120.2 rotor, and the pellet was resuspended in MOPS actin-binding buffer (M-ABB; 100 mM KCl, 10 mM MOPS, pH 6.8, 2 mM MgCl₂, 0.2 mM CaCl₂, 0.2 mM ATP, 1 mM dithiothreitol [DTT], and 1 mM sodium azide). Any bundled actin was removed by centrifugation at 4°C, 15,000 rpm (21,000 ×g) for 10 min in an Eppendorf 5424R benchtop microfuge. The resulting F-actin at ∼30 µM was stabilized by the addition of an equimolar amount of phalloidin. After 10 min at room temperature, unbound phalloidin was removed by centrifugation at 4°C, 15,000 rpm (21,000 ×g) for 10 min in an Eppendorf 5424R benchtop microfuge. F-actin was adjusted to 10 µM with M-ABB. For actin–Tm binding, Tm was added to a ratio of 3.5:1 (actin/Tm) and allowed to incubate overnight.

For fluorescence experiments (TR-F), actin was labeled at Cys-374 with IAEDANS (Thermo Fisher Scientific). Labeling was done on F-actin. To 50 µM of G-actin in G-buffer (10 mM Tris, pH 7.5, 0.2 mM CaCl₂, and 0.2 mM ATP), Tris pH 7.5 was added to a final concentration of 20 mM and then polymerized by the addition of 3 M KCl (to a final concentration of 100 mM) and 0.5 M MgCl₂ (to a final concentration of 2 mM), followed by incubation at 23°C for 1 h. IAEDANS was added to a final concentration of 1 mM (from a 20 mM stock in dimethylformamide). Labeling was done for 3 h at 23°C and then overnight at 4°C. Labeling of actin with 2-(4′-(iodoacetamido)aniline)naphthalene-6-sulfonic acid (IAANS), N-((2-(iodoacetoxy)ethyl)-N-methyl)amino-7-nitrobenz-2-oxa-1,3-diazole (IANBD), and 7-diethylamino-3-(4′-maleimidylphenyl)-4-methylcoumarin (CPM) were done the same as IAEDANS, except with 250 µM IAANS and IANBD and 500 µM CPM for 1 h at 23°C. Labeling of actin with fluorescein-5-maleimide (FMAL) was done as for IAEDANS, except labeling was for 5 h at 23°C. Labeling was terminated by the addition of DTT (to a final concentration of 5 mM). Labeled F-actin was collected by centrifugation for 30 min at 4°C, 100,000 rpm (350,000 ×g) in a Beckman TLA-120.2 rotor. The pellet was rinsed three times and then resuspended in labeling G-buffer (5 mM Tris pH 7.5, 0.2 mM CaCl₂, and 0.5 mM ATP). G-actin was clarified by centrifugation for 10 min, 4°C, 90,000 rpm (290,000 ×g) in a Beckman TLA-120.2 rotor. G-actin was then repolymerized by the addition of MgCl₂ to 3 mM and incubation at 23°C for 1 h. Labeled F-actin was collected by centrifugation for 30 min at 4°C, 100,000 rpm (350,000 ×g) in a Beckman TLA-120.2 rotor. The pellet was washed three times and then resuspended in M-ABB. Bundled actin was removed, and the labeled F-actin was stabilized with phalloidin and then complexed with Tm as described for unlabeled actin. Labeling efficiency was determined by measuring dye absorbance by UV-vis spectrophotometry, and protein concentration, measured with a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) using unmodified actin as a standard. The extent of labeling (dye/mol actin) was ∼1.0 (IAEDANS), 1.0 (IAANS), 0.75 (IANBD), 0.65 (CPM), and 0.3 (FMAL).

The gene encoding human α-Tm was inserted into a pET3d vector and provided by J. C. Tardiff (University of Arizona, Tucson, AZ). Tm was expressed with two additional N-terminal amino acids (Ala-Ser) needed to ensure proper actin-binding and polymerization function, as bacterially expressed Tm is not otherwise functional due to lack of acetylation of the starting Met (Monteiro et al., 1994; McConnell et al., 2017). Protein production in Escherichia coli BL21(DE3)-competent cells (New England Biolabs) was done in ZYP broth (1% tryptone, 0.5% yeast extract, 0.5% glycerol, 0.05% glucose, and 0.2% lactose) and pelleted by centrifugation in a Beckman JA-10 rotor at 1,800 ×g for 20 min at 4°C. Pellets were resuspended in double distilled H₂O, brought to 1 M saline with solid NaCl, and homogenized using the Emulsiflex C3 (Avestin). Cell debris of homogenized pellets was then pelleted by centrifugation in a Beckman JA-17 rotor at 28,950 ×g for 10 min at 4°C, and the supernatant containing expressed Tm was decanted into a glass beaker. The supernatant was boiled for 6 min and allowed to cool for 45 min at room temperature. The denatured lysate containing Tm was purified from other bacterial proteins by three total cycles of acid/base cuts (or until a ratio of absorbance (A) at 260 nm and 280 nm [A₂₆₀/A₂₈₀] of <0.8 was reached): (1) precipitation of Tm by addition of 1 M HCl to pH ∼4.5, (2) centrifugation at 28,950 ×g for 10 min at 4°C, (3) resuspension of the pellets in 1 M KCl with addition of 1 M KOH to pH ∼7, (4) centrifugation of the dissolved Tm at 28,950 ×g for 10 min at 4°C, and (5) collection of the supernatant. Approximtaely 1 g of α-Tm dimer was purified from 2 liters of culture. Aliquots were stored at −80°C until use. For preparation of fluorescent Tm–actin, Tm (in 1 M KCl) was diluted 10-fold with M-ABB lacking KCl (bringing the final concentration of KCl to 0.1 M) and reconstituted overnight in a 3.5:1, 5:1, or 7:1 molar ratio of actin/Tm. In preliminary experiments, we tested cosedimentation of actin and Tm in 7:1 and 3.5:1 mixtures. As expected for 7:1 actin/Tm, almost all of the Tm was bound to actin and observed in the F-actin pellet, and very low levels of Tm were in the supernatant. At the 3.5:1 actin/Tm ratio ∼50% of the Tm pelleted with F-actin and 50% remained unbound in the supernatant (see supplementalResults and Discussion at bottom of the PDF).

pET45b vectors encoding E. coli–optimized codons for the C0–C1 or C0–C2 portion of human cMyBP-C with N-terminal 6x His tag and TEV protease cleavage site were obtained from GenScript. In addition, C0–C2 mutants were generated with HCM mutations (R282W, E334K, and L352P) or mutations in a positively charged loop (Arg–Ala–Ser–Lys [RASK]; residues 215–218) in the C1 domain that interacts with Tm (Risi et al., 2018). These Tm-binding mutations introduced additional positive charges (A216R/S217K) or reversed charges (R215E/K218E; Risi et al., 2018). Mutations were engineered in the human cMyBP-C fragments using a Q5 Site-Directed Mutagenesis Kit (New England Biolabs). All sequences were confirmed by DNA sequencing (Eton Biosciences). Protein production in E. coli BL21DE3-competent cells (New England Biolabs) and purification of C0–C1 and C0–C2 fusion proteins using His60 Ni Superflow resin were done as described previously (Bunch et al., 2018). C0–C2 and C0–C1 (with the His tag removed by TEV protease digestion) was then concentrated, dialyzed to 50/50 buffer (50 mM NaCl and 50 mM Tris, pH 7.5), and stored at 4°C. Initial characterization of conditions for TR-F was done with this protein, which was ∼75% the correct molecular weight and 25% breakdown products (due to proteolysis within the motif between C1 and C2). For all of the TR-F binding curves and cosedimentation experiments, we further purified the C0–C2 using size exclusion chromatography to achieve >90% intact C0–C2. For size exclusion chromatography, C0–C2 (5–20 mg/ml, 1.7–2.5 ml) was applied in running buffer (150 mM NaCl, 50 mM NaPO₄, and 1 mM DTT, pH 6.7) to a HiPrep Sephacryl S-100 column. Flow rate was 0.7 ml/min. Purified C0–C2 was checked by SDS-PAGE for purity and then dialyzed into the appropriate buffer (usually M-ABB). The HCM mutant E334K was distinct from all other C0–C2 proteins in being highly destabilized in E. coli with many more breakdown products present during the purification process. These breakdown products were removed by size exclusion chromatography in the purification. We therefore measured only the effects of full-length E334K C0–C2. Proteins were typically used for experiments within 2 wk. For longer storage periods, C0–C2 and C0–C1 proteins were stored at −20°C in 50/50 buffer containing 50% glycerol, 1 mM DTT, protease inhibitors (88265, 1 tablet/50 ml; Pierce), and 1 mM sodium azide.

For the determination of phosphorylated Ser residues, phosphorylation was monitored by in-gel staining of proteins with Pro-Q Diamond (Thermo Fisher Scientific) and staining total protein with SYPRO Ruby (Thermo Fisher Scientific), according to the supplier’s instructions. Maximal phosphorylation was shown to plateau at ∼2.5 ng PKA/µg C0–C2 (Bunch et al., 2019). C0–C2 was typically treated with 7.5 ng PKA/µg C0–C2 unless otherwise noted. For liquid chromatography-tandem mass spectrometry (LC-MS/MS), C0–C2 WT and mutant R282W were treated with 65, 2.6, and 0.1 ng PKA/µg C0–C2. Phosphorylated proteins were separated by SDS-PAGE. The C0–C2 bands were excised as described previously (Bunch et al., 2018). Following LC-MS/MS, a probability-based approach for high-throughput protein phosphorylation analysis and site localization (Beausoleil et al., 2006) was used, as described previously for PKA treatment of WT C0–C2 (Bunch et al., 2019).

Actin binding by cMyBP-C fragments (C0–C2 or C0–C1) was determined by cosedimentation and analyzed as previously described in detail (Bunch et al., 2018). Binding levels were determined at 25°C in M-ABB using 1 µM F-actin or 1 µM F-actin containing 0.29 µM Tm incubated with increasing amounts of C0–C2 or C0–C1. All cosedimentation curves were generated from at least two separate purifications of C0–C2 (and actin; n > 4 [typically n = 6] for all data points). All binding mixtures were made separately and not prepared in a single batch.

The maximum molar binding ratio (B_max) and K_d values for C0–C2 binding to actin were determined by fitting the data to a quadratic model (Michaelis–Menten function) using Origin Pro 2019 computer software package through a nonlinear least-squares minimization (Levenberg–Marquardt algorithm). χ² values of quadratic fits for most binding experiments were <0.005. The exceptions to this are C0–C1, and the R215E/K218E mutant of C0–C2 when phosphorylated. In these cases, poor fits were due to much reduced, nonsaturable, binding and they instead fit to a linear function. These apparent K_d and B_max values are used as comparative indicators of binding characteristics for C0–C2 binding to actin under different conditions (− and + phosphorylation or in the presence of mutations). They represent the apparent dissociation constants (C0–C2 concentration required for half-maximal binding) and maximal binding ratios of the cMyBP-C fragments to actin (C0–C2/actin) in cosedimentation experiments where bound cMyBP-C and total actin monomers are directly measured. In TR-F generated curves, the K_d values again represent the apparent dissociation constants (C0–C2 concentration required for half-maximal binding), but the B_max in this case is the maximal change in lifetime of IAEDANS when C0–C2 binding is maximal. K_d values, expressed as μM C0–C2 in both assays, can be compared within and between the two assays. The B_max values, having different units in the two assays, can only be compared within the same type of assay (cosedimentation or TR-F). These values derived from fitting the data of the binding curves to a quadratic model (Michaelis–Menten function) are the standard values used to describe binding of cMyBP-C fragments to F-actin.

50 µl of sample aliquots were loaded manually with a multichannel pipette in 384-well black polypropylene microplates (#781209; Greiner Bio-One). Plates were spun for 1 min at 1,000 rpm (200 ×g) in an Eppendorf rotor (5810R A-4-81) to remove air bubbles. Fluorescence lifetime measurements were acquired using a high-precision fluorescence lifetime plate reader (FLTPR; Fluorescence Innovations; Bunch et al., 2018; Schaaf et al., 2017). Dye-labeled F-actin (alone or mixed with C0–C2) was excited with either a 355-nm microchip laser (Teem Photonics) for IAEDANS, IAANS, and CPM dyes or a 473-nm microchip laser (Bright Solutions) for FMAL and IANBD dyes. Emission was filtered with 409-nm long-pass and 470/20-nm band-pass filters for the 355-nm laser (Bunch et al., 2018) or 488-nm long-pass and 517/20-nm band-pass filters (Semrock) for the 473-nm laser (Gruber et al., 2014). The FLTPR allows for high-throughput fluorescence lifetime detection at high precision by using unique direct waveform-recording technology (Muretta et al., 2010).

Sample means are from four or more independent experiments. Each experiment, following the optimization of binding conditions, was performed using at least two independent protein preparations. Average data are provided as mean ± SE, except for Table 3 and Fig. 2, which used ± SD for Z′ scores. Statistical significance is evaluated by use of an unpaired t test. P values <0.05 were taken as indicating significant differences and are given in the figure and table legends as well as the Results section.

Actin TR-F–binding curves for mutants not included in the main text figures, comparison of the binding of specific substoichiometric levels of C0–C2 not included in the main text, comparisons of multiple or compounded variables (± PKA, Tm, mutation), and mass spectrometry details of phosphorylated C0–C2 proteins are included in the supplemental text. Testing of the four probes other than IAEDANS and further details of optimization of the experimental conditions for pH, salt, buffer, and Tm are also presented in the supplemental Results and Discussion sections. Specifically, Fig. S1 shows lifetime changes of five fluorescent dyes attached to actin at Cys-374 upon binding to C0–C2. Fig. S2 shows buffer conditions optimization of the TR-F assay for C0–C2 binding to IAEDANS–actin and IAEDANS–actin–Tm. Fig. S3 shows actin/Tm ratios. Fig. S4 shows TR-F IAEDANS–actin–Tm and IAEDANS–actin–binding curves for C0–C2 mutants. Fig. S5 shows effects of R282W HCM mutant on phosphorylation-modulated binding to actin–Tm at submaximal phosphorylation by PKA. Fig. S6 shows effects of Tm-binding mutants on phosphorylation-dependent binding to actin–Tm and actin. Fig. S7 shows cosedimentation assays for C0–C1 binding to actin–Tm and actin. Fig. S8 shows a comparison of TR-F and cosedimentation actin-binding assays for C0–C2 mutants. Table S1 lists TR-F and cosedimentation binding parameters of WT and mutant MyBP-C/actin and actin–Tm. Table S2 lists change in IAEDANS lifetime for IAEDANS–actin and IAEDANS–actin–Tm upon substoichiometric binding of C0–C2 and C0–C1. Table S3 lists WT and R282W C0–C2 phosphorylated peptides. Table S4 lists peak intensity ratios for WT and R282W PKA phosphorylated/unphosphorylated peptides.

Our goal was to develop a TR-F–based actin-binding assay able to detect changes in actin binding of cMyBP-C brought about by different experimental conditions. A TR-F actin–binding assay in a 384-well plate format made possible the testing of variables such as pH, ionic strength, phosphorylation of cMyBP-C, and the presence of mutations in cMyBP-C. The results indicate that the assay will be useful in identifying therapeutic drugs that can modify cMyBP-C actin-binding properties to the same extent as phosphorylation. This same approach should be suitable, with appropriate modifications to the study of other ABPs.

Detailed optimization of probes, ionic strength and pH for the TR-F binding assay is presented in detail in the supplemental material. Using our current TR-F–binding assay, we monitored the lifetime of the thiol-reactive dye IAEDANS that is covalently attached to F-actin’s readily labeled cysteine (Cys-374). The buffer for actin-binding experiments was MOPS-Actin Binding Buffer (M-ABB; 100 mM KCl, 10 mM MOPS, pH 6.8, 2 mM MgCl₂, 0.2 mM CaCl₂, 0.2 mM ATP, 1 mM DTT, and 1 mM sodium azide). These conditions allowed us to readily distinguish between binding of C0–C2 that was unphosphorylated or phosphorylated (by PKA; Fig. 1 A) to actin or actin–Tm (Fig. 1 B). The addition of C0–C2 decreased the lifetime of IAEDANS–actin (with and without Tm) in a concentration-dependent manner. Upon phosphorylation of C0–C2 with PKA the decrease in IAEDANS–actin lifetime was much less, at submaximal binding concentrations of C0–C2 (Fig. 1 and Table 1). This indicates that PKA treatment reduced cMyBP-C binding to actin and actin–Tm (Fig. 1, A and B, dashed versus solid lines). These conditions also showed the presence of Tm on actin increases C0–C2 binding (Fig. 1 C, thick versus thin lines). At substoichiometric levels of binding (1.25 µM C0–C2) Tm increased binding by 41% (P < 0.005; Table 2). We compared these data to results derived from cosedimentation experiments done under the same conditions (Fig. 1, D–F). Over the same concentration range, TR-F and cosedimentation assays provided similar binding profiles for unphosphorylated C0–C2, and these were linearly correlated (for actin and unphosphorylated C0–C2, R² = 0.96, adjusted-R² = 0.95; Fig. 1 G; and for actin–Tm, R² = 0.87, adjusted-R² = 0.84; Fig. 1 H). However, subtle differences in K_d and B_max are observed when fit to a quadratic equation for Michaelis–Menten binding kinetics (Table 1 or Fig. 1 G). Cosedimentation showed a less dramatic effect of PKA on binding actin compared with the TR-F assay (Fig. 1 D versus Fig. 1 A, and Table 1), while the effects were similar on actin–Tm (Fig. 1 E versus Fig. 1 B, and Table 1). The effect of PKA phosphorylation of C0–C2 is optimally seen when C0–C2 levels are below saturation (substoichiometric) at 2.5 µM C0–C2 for actin (thin arrows in Fig. 1, A, C, D, and F) and 1.25 µM C0–C2 for actin–Tm (thick arrows in Fig. 1, B, C, E, and F). This is the case for both TR-F and cosedimentation assays. For the TR-F assay, these concentrations showed the largest absolute difference in the change in lifetime between minus and plus PKA of 1.7% and 2.0% (Table 2). These equate to PKA treatment of C0–C2 reducing the change in lifetime of IAEDANS on actin by 66% and 74% compared with the change in lifetime observed when C0–C2 was not phosphorylated (Table 2). These concentrations (2.5 µM C0–C2 for actin and 1.25 µM C0–C2 for actin–Tm) also result in the highest levels of statistical significance when comparing ±PKA (P = 2.9 × 10⁻¹¹ for actin and P = 1.6 × 10⁻¹³ for actin–Tm).

We used Z′ scores to determine if our actin‒cMyBP-C TR-F assay has sufficient sensitivity for employment in HTS. We are interested in screening chemical libraries to identify drugs that affect binding of cMyBP-C or that mimic the effects of phosphorylation of cMyBP-C on actin binding. Z′ scores (see Eq. 2 in Materials and methods) allow for the determination of whether the signal window (Δτ) between two states (bound versus not bound, or phosphorylated binding versus not phosphorylated binding) and the variance (SD) of the two states indicate a worthless HTS (Z′ score <0), a doable HTS (Z′ score 0–0.5) or an excellent HTS assay quality (Z′ score 0.5–1.0; see also Fig. 2, D and E). Using either the IAEDANS-actin or IAEDANS-actin–Tm sensor, we performed a simulated HTS by loading ∼60 wells of each condition in 384-well plates. The conditions tested were: actin alone, actin plus C0–C2, and actin plus PKA-treated C0–C2 as well as actin–Tm alone, actin–Tm plus C0–C2, and actin–Tm plus PKA-treated C0–C2. Means and three SDs (3× SDs) of the lifetimes for each condition were used to determine the Z′ scores (see Fig. 2). The Z′ score for the screen shown in Fig. 2 was calculated as Z′ = 0.56 for IAEDANS–actin alone versus IAEDANS–actin plus C0–C2 and Z′ = 0.27 for IAEDANS–actin plus C0–C2 versus IAEDANS–actin plus PKA-treated C0–C2. The Z′ score determination was performed in triplicate with three different IAEDANS–actin (and IAEDANS-actin–Tm) and C0–C2 preparations. All tests gave Z′ scores that indicated good (doable) or excellent screening quality. Actin–Tm showed higher scores (average Z′ = 0.46) than actin alone (average Z′ = 0.29; Table 3). These results validate the robustness of TR-F for HTS.

The ability of the TR-F binding assay to quickly and reproducibly distinguish between two states of cMyBP-C suggests that it will be useful in probing the effects of mutations in cMyBP-C (and potentially other ABPs) on actin binding. Importantly, the ability to detect effects of mutations on actin binding would serve as a proof of principle that the assay can detect other factors, such as therapeutic drugs, that modulate binding. For these reasons, we tested the sensitivity of the TR-F assay to detect changes in C0–C2 binding due to three HCM mutations predicted from previous work to affect actin binding in very different manners. Previous work suggested L352P increases actin binding and E334K decreases binding. We predicted that R282W, based on sequence analysis, alters phosphorylation and would thereby alter actin binding. We also tested two mutations in a proposed Tm-binding sequence in C1. The location of these mutations is diagramed in Fig. 3. The key comparisons that validate the assay’s usefulness to detect a variety of changes in binding due to mutations are presented in Figs. 4, 5, and 6 and Table 2. As the assay makes possible the comparison of combinations of multiple factors (minus and plus HCM or Tm-binding mutations, minus and plus phosphorylation by PKA, minus and plus Tm) this was done as well and is presented in the supplemental material.

The M-domain and C2 provide binding sites for actin (Howarth et al., 2012; Bezold et al., 2013) and their removal, in C0–C1, dramatically reduces actin binding and this is seen in the TR-F assay as well (Fig. 4 and Table 2; >90% reduction in binding at 2.5 µM, P < 0.0001).

The L352P mutation has been demonstrated to increase actin (Bezold et al., 2013) and thin filament (Mun et al., 2016) binding. The TR-F assay readily detected the increase in binding of L352P to actin (Fig. 4 and Table 2). Increases of 39% were seen at 2.5 µM L352P C0–C2 (P < 0.0005).

The HCM mutant, E334K, was tested because it significantly reduced actin binding when present in mouse C1–C2 (Bezold et al., 2013) and the homologous mutant (E330K) in mouse hearts shortened the duration of ejection (van Dijk et al., 2018). E334K in human C0–C2 displayed binding that was similar to WT measured by TR-F (Fig. 4 and Table 2).

R282W changes the PKA target sequence (281–284) from RRIS to RWIS (Fig. 3), and this change is predicted to eliminate PKA phosphorylation of Ser-284. Ser-284 is one of four serines in the M-domain phosphorylated by PKA (Jia et al., 2010; Ponnam et al., 2019; O’Leary et al., 2019). We predicted that R282W would eliminate PKA phosphorylation of this serine. Under our standard conditions, R282W showed subtle, but not significant, differences in binding when compared with WT in either the phosphorylated or unphosphorylated state at 2.5 µM. At 1.25 µM, unphosphorylated R282W did display a significant decrease in binding (36% decrease, P < 0.01; Fig. 5 and Table 2).

Examination by mass spectrometry of the four serines (Ser-275, Ser-284, Ser-304, and Ser-311) in the motif that are phosphorylated by PKA showed that at PKA levels required for maximal phosphorylation (2.5 ng PKA/µg C0–C2) or 25 times those levels, all four serines were phosphorylated in WT C0–C2. In contrast, in the R282W mutant, phosphorylation was almost undetectable at Ser284. Phosphorylation was detected in the mutant at Ser-275, Ser-304, and Ser-311 (Table S3 and Table S4). The initial test of PKA effects on actin binding used PKA levels three times that required for maximal phosphorylation. From these results, we concluded that phosphorylation in some combination of Ser-275, Ser-304, and Ser-311 in R282W is sufficient to reduce interactions with actin.

In vivo, the PKA target serines are not found completely phosphorylated (Copeland et al., 2010; Jacques et al., 2008). We therefore tested a range of phosphorylation levels of C0–C2 for their effects on binding to actin. As for testing other conditions, use of the TR-F assay in a 384-well plate format made this relatively easy. We reduced the PKA levels used to phosphorylate WT and R282W C0–C2 for actin-binding studies. The effect of submaximal phosphorylation on the change in TR-F upon binding to actin was measured. Immediately following determination of the TR-F, a portion of the samples was removed with a multichannel pipette and placed in protein sample buffer, run on SDS-PAGE gels, and stained with Pro-Q Diamond to determine phosphorylation levels. A wide range of phosphorylation levels was obtained by varying PKA levels, and the R282W mutant was less phosphorylated than WT at all PKA concentrations (Fig. 5, B and C). At levels of phosphorylation mediated by PKA at 1.5 ng PKA/µg C0–C2, binding of WT C0–C2 remained low, at the same level as that seen when either it or R282W was treated with 5 ng PKA/µg C0–C2. R282W, at this concentration (1.5 ng PKA/µg C0–C2), showed a 65% increase in binding to actin (Fig. 5 D; P < 0.00001). As PKA levels were reduced further to 0.5 ng PKA/µg C0–C2, the difference between WT and the mutant remained significant (Fig. 5 D; 27%, P = 0.0009), though both now exhibited increased binding.

Binding to actin by C0–C2 is increased when actin is decorated with Tm (Fig. 1, C and F; and Table 1). To specifically investigate the ability of the TR-F screen to detect changes in cMyBP-C interactions with Tm, we made functional mutations in the C1 domain in a region (amino acids 215–218; RASK) that interacts with Tm (Risi et al., 2018). Mutating this sequence to EASE (R215E/K218E; changing the positively charged R and K to the negatively charged E) resulted in 82% lower levels of binding to actin–Tm at 1.25 µM when compared with WT (P < 1 × 10⁻¹³). This mutant, EASE, displayed 37% lower levels of binding at 1.25 µM C0–C2 to actin–Tm than to actin alone (P = 0.026; Fig. 6, A and B; and Table 2). This is the only mutation tested that resulted in less binding to actin–Tm than to actin. In contrast, WT C0–C2 shows a 41% greater TR-F effect on actin–Tm than on actin alone (P = 0.003; Fig. 1 and repeated in Fig. 6 and Table 2). Addition of positively charged amino acids to the Tm-binding region, RRKK (A216R/S217K), enhanced binding at 1.25 µM C0–C2 compared with WT to actin–Tm by 10% (P = 0.021). At lower concentrations of 0.625 and 0.313 µM, these relative differences were even greater (58% and 275%, respectively; P < 1 × 10⁻⁶; Fig. 6 D and Table 2). The results with Tm-binding region mutants are consistent with effects of the same mutants tested in C0–C1 for effects on thin filament activation of myosin ATPase (Risi et al., 2018). On actin alone, at lower concentrations (1.25 and 0.625 µM C0–C2), the RRKK increased binding by 37% and 160% compared with WT (P < 0.006; Fig. 6 D and Table 2). The negative (EASE) mutation had negative effects on binding to bare actin as well. Binding was reduced by 60% (P < 0.001) at 1.25 µM C0–C2 when compared with WT on actin, but these effects were reduced compared with those with actin–Tm (Fig. 6 B and Table 2).

We found that TR-F can be used to monitor the binding of the N-terminal domain of cMyBP-C (C0–C2) to actin and actin–Tm. The 384-well plate format allowed us to quickly and easily optimize conditions for TR-F monitoring of binding. We optimized the conditions to detect both differences between the binding of phosphorylated and unphosphorylated C0–C2 and increased binding provided by the presence of Tm on the actin filament. Using these conditions, we characterized the assay’s ability to detect differences in binding due to phosphorylation, Tm, or mutations in C0–C2. These tests all indicate that this assay will be suitable for screening drug libraries to identify molecules capable of modulating cMyBP-C’s ability to bind actin and actin–Tm, potentially through mimicking phosphorylation of the M-domain. Z′ score analysis adds additional confidence to the appropriateness of this new method for HTS.

All five fluorescent probes we tested reported a reduction in lifetime upon binding (Fig. S1). The finding that all the fluorescent probes showed reduced lifetimes indicates that the results we are seeing are due to C0–C2 interactions with actin and not with a specific probe attached to actin’s Cys-374. Cryoelectron microscopy of thin filaments decorated with C0–C1 (Risi et al., 2018) found C1 binds predominantly in what they termed the C1-S structural class. This class of binding is characterized by, among other contacts, residues 194–206 of C1 interacting with residues 365–367 of actin (Harris et al., 2016). Actin’s Cys-374 is labeled by our probes. This site is in close proximity to this C1–actin interaction region. Though we do not yet have a structure for C0–C2 binding to thin filaments, this C1–actin interaction in close proximity to Cys-374, may explain the responsiveness of all the probes to C0–C2 binding. We selected IAEDANS–actin for optimization as it displayed the largest change in lifetime upon binding C0–C2 and had low variability. IAEDANS–actin readily distinguished between the binding observed with C0–C2 in its phosphorylated versus unphosphorylated state.

Lifetimes for IAEDANS attached to actin ranged from 16.5 to 18 ns with 0.3- to 1.1-ns changes upon binding to C0–C2. These lifetime changes appear small (∼2.7%), but on a biophysical scale they are very significant, with SEs on the order of ∼5–50 ps (or 0.01–0.05%).

The TR-F assay in the 384-well format makes testing a wide range of conditions relatively easy. Numerous variables can be rapidly tested for scientific discovery and optimization of assay conditions. Our testing of conditions for optimal binding (see Fig. S2) suggested that the assay can be used in wide range of ionic strength and pH conditions. Subtle differences were observed. Lower pH (6.8) showed larger differences between binding of unphosphorylated and phosphorylated C0–C2 compared with that seen at pH 7.5 or pH 8.0 (Fig. S2 A). The higher pH conditions displayed larger differences between actin and actin–Tm binding. Replacing the KCl in our standard buffer with NaCl increased the difference between unphosphorylated and phosphorylated binding but reduced the effects of Tm. Having the ability to test all of these conditions (Fig. S2, A–C; and Fig. S3) allowed us to select buffer conditions that balanced phosphorylation and Tm effects. We would not have attempted this optimization using standard cosedimentation assays. The buffer we opted for is M-ABB (100 mM KCl, 10 mM MOPS, pH 6.8, 2 mM MgCl₂, 0.2 mM CaCl₂, 0.2 mM ATP, 1 mM DTT, and 1 mM sodium azide).

Using the TR-F–optimized conditions, we retested binding by cosedimentation assays. The new conditions led to an increase in binding affinity, as measured by cosedimentation over our previous conditions (Bunch et al., 2019). Previously, using 1 µM phalloidin-stabilized F-actin in cosedimentation assays, we observed apparent K_d values for C0–C2 binding to actin of 11 ± 2 µM (unphosphorylated) and 10 ± 4 µM (phosphorylated). Using our new conditions, we find apparent K_d values, measured by cosedimentation, to be 4.3 ± 0.3 µM (unphosphorylated) and 6.5 ± 0.3 µM (phosphorylated). The K_d values also indicate that in addition to decreased K_d values (increased affinities), we now observe larger differences in binding, measured by cosedimentation, between unphosphorylated and phosphorylated C0–C2 (Table 1 and Fig. 1 D). We conclude that the TR-F assay, in the 384-well plate format, can be used to probe physiological and chemical variables of actin binding, and these findings can be transferable to different complementary assays.

While K_d and B_max values are useful for comparing binding curves, it is questionable how relevant the numbers are to in vivo binding. Binding is likely to be complex under the conditions required to generate in vitro binding curves, as each actin monomer in the F-actin possesses multiple interaction sites for C0–C2. Likewise, C0–C2 possesses multiple binding sites for multiple actin regions (Shaffer et al., 2009; Orlova et al., 2011; Risi et al., 2018). At elevated C0–C2 concentrations, competition between sites, cooperative interactions, and steric hindrance are all likely to be significant. In vivo, high ratios of C0–C2/actin (greater than 1:7) are not present, and therefore, neither competition nor steric hindrance is a factor. Due to these complexities, we show the binding curves (Figs. 1, 4, 5, 6, S4, S5, S6 and S7) as well as report apparent K_d and B_max values (Tables 1 and Table S1). Visual inspection of each curve confirms that when lower K_d values are observed, more binding is observed at low (substoichiometric) C0–C2/actin levels that are more physiologically relevant (in muscle, cMyBP-C is present at approximately one out of every seven actin monomers). For this reason, we have extracted, from the curves, the binding levels observed at substoichiometric concentrations of C0–C2 for direct comparison in Tables 2 and Table S2.

After establishing that IAEDANS labeling of actin at Cys-374 is suitable as a biosensor to detect cMyBP-C binding, we aimed to increase the physiological relevance of the bound complex by including cardiac Tm in the assay. For studies of cardiac and skeletal muscle actin, Tm is a key regulatory protein of contraction that blocks the myosin-binding sites on actin. cMyBP-C has been proposed to interact with Tm. These interactions directly modulate Tm’s position on actin independent of Ca²⁺ activation in cardiac thin filaments (Risi et al., 2018). The TR-F assay worked well using IAEDANS–actin–Tm to detect cMyBP-C binding. The presence of Tm enhanced binding at low C0–C2 concentrations. This suggests that Tm facilitates binding, consistent with cMyBP-C’s proposed interactions with both F-actin (Orlova et al., 2011; Harris et al., 2016) and Tm (Mun et al., 2014; Risi et al., 2018). We conclude that the TR-F assay is suited for detecting binding with either actin or actin–Tm.

These findings set the stage for screening for drugs that modulate cMyBP-C (in either the phosphorylated or unphosphorylated state) binding to actin–Tm. Screening using IAEDANS–actin–Tm offers potential benefits over using IAEDANS–actin. First, due to increased binding detected by TR-F, the presence of Tm reduces the concentration of C0–C2 required to observe the greatest difference between the phosphorylated and unphosphorylated forms (down to 1.25 µM C0–C2 from the 2.5 µM required when Tm is absent). Second, the difference between binding of the phosphorylated and unphosphorylated C0–C2 is greater and the SDs of the lifetime changes are smaller. These two changes increase the robustness of screens as exemplified by the increases in Z′ scores (Fig. 2 and Table 3). Finally, incorporating Tm into the initial screens is relatively simple and can potentially reduce false positives, including drugs that affect binding to bare actin, but not when actin is decorated with Tm, as it would be in vivo. Using this technology, in the presence of Tm, identified drugs can be further tested to determine if their effect is through actin binding alone (by subsequently testing on bare actin) or if it is dependent on Tm. Also, this screening method sets the stage for future work using reconstituted thin filaments containing the Ca²⁺-regulated troponin complex.

Both TR-F and cosedimentation assays offer quantitative evaluation of actin binding. In addition to being high throughput and low manual effort, the TR-F assay has the added advantage of having a higher sensitivity to detect changes in binding due to phosphorylation. Differences in binding between phosphorylated and unphosphorylated C0–C2 are magnified in TR-F measurements as compared with cosedimentation (Table 1). On bare actin, the difference in K_d’s is 8× for TR-F and 1.5× for cosedimentation. On actin–Tm, the changes in K_d’s are 2.6× for TR-F and 2× for cosedimentation. Differences in the two assays may result from there being multiple binding sites on C0–C2 for multiple binding sites on actin (Orlova et al., 2011; Harris et al., 2016; Risi et al., 2018). PKA may affect a particular binding interaction that has a smaller consequence on C0–C2 remaining bound to actin than it has on C0–C2’s influence on the lifetime of IAEDANS attached to actin’s Cys-374. Additionally, we have previously demonstrated large phosphorylation effects on actin’s structural dynamics that accompany small or no changes in binding as measured by cosedimentation (Bunch et al., 2019). These changes in dynamics likely arise from changes in binding sites, but an additional possibility is that actin’s structural dynamics, in the presence of bound phosphorylated C0–C2, contributes to the lifetime changes of IAEDANS attached to actin’s Cys-374. Finally, using C1–C2 fragments of rat cMyBP-C, changes in the activation state of native thin filaments due to phosphorylation of cMyBP-C were found to be greater than the effects on binding as measured by cosedimentation (Ponnam et al., 2019). Changes in binding sites and changes in dynamics or activation status are likely linked and either, or both, may explain the magnification of the differences we observe by TR-F. Whatever the mechanism explaining the increased difference measured by TR-F, it offers a different and complementary binding assay to be used in conjunction with cosedimentation and an improvement for screening purposes and scientific discovery.

We further validated the TR-F assay by testing its ability to probe cMyBP-C’s actin-binding capacity when mutated to contain previously described HCM and Tm-binding mutants and deletion of the M-domain and C2.

Complete removal of the M-domain and C2, in the C0–C1 protein, showed dramatic reduction in actin binding when tested using TR-F (Fig. 4). This is consistent with earlier cosedimentation results by us (Bunch et al., 2019) and others (Shaffer et al., 2009) and confirmed with cosedimentation under our specific binding conditions (Fig. S7 and Table S1).

L352P displayed increased binding (Fig. 4), as has been previously demonstrated in mouse C1–C2 (Bezold et al., 2013).

The HCM mutant E334K displayed binding that was similar to WT measured by TR-F and slightly increased binding when measured by cosedimentation (Figs. 4 and S8). This result was surprising, as the homologous mutant in mouse (E330K) showed greatly reduced binding to actin when tested in a C1–C2 fragment (Bezold et al., 2013). Also, E330K in mouse hearts shortened the duration of ejection (van Dijk et al., 2018). Differences in buffer conditions may explain different results, though our testing of a wide range of buffers made only subtle differences in the binding of WT C0–C2. The presence of the C0 and Pro/Ala-rich linker in our proteins and/or species-specific sequence differences may be responsible for the different results we observed. That binding is different with C0 is supported by Risi et al. (2018), who found that the presence of C0 influences the mode of actin binding of C0–C1. Mutation of the homologous residue, E248K, in slow skeletal MyBP-C (MYBPC1) was found to increase interactions with myosin (Stavusis et al., 2019). Increased binding to myosin by mouse E330K in cMyBP-C might explain the shortened duration of ejection seen by van Dijk et al. (2018). It will be interesting to determine if this mutation in cMyBP-C alters interactions with myosin or if its phenotype is related to impairing the ubiquitin-proteasome system (Bahrudin et al., 2008).

The HCM mutation R282W inhibits phosphorylation of Ser284 and in conditions of submaximal phosphorylation results in increased actin binding by C0–C2. R282W is located in a PKA recognition sequence RRIS in the M-domain (Fig. 3). We predicted that it would reduce or eliminate phosphorylation of Ser284 by PKA. Mass spectrometry data suggest this is the case. When relatively high levels of PKA were used to maximally phosphorylate C0–C2, R282W had no significant effect on binding of phosphorylated C0–C2 to actin when compared with phosphorylated WT C0–C2 (Fig. 5 A and Table 2). Looking at the details of binding (Fig. 5 A), there is the suggestion that R282W reduces the effect of maximal phosphorylation. Unphosphorylated R282W appears to bind actin slightly less that WT, while phosphorylated R282W binds slightly more. This might indicate a reduced PKA effect (increased actin binding), as was predicted. As cMyBP-C is found in a range of PKA phosphorylation states, we used the TR-F assay to test a range of phosphorylation levels in WT and the R282W mutant. At intermediate levels of phosphorylation, we found a very significant difference in binding (Fig. 5; see also Fig. S5 for actin–Tm). C0–C2 binding was more inhibited by PKA treatment in WT than was R282W, which behaved more like unphosphorylated protein. We conclude that the remaining three serines, when phosphorylated at high levels, are sufficient to reduce actin binding to a similar extent as when all four serines are phosphorylated. However, at reduced phosphorylation levels that we expect to be found in cardiac tissue, and especially in HCM (El-Armouche et al., 2007), there is a difference that results in R282W binding more strongly to actin. The TR-F multiwell plate assay greatly facilitated this testing of multiple phosphorylation levels to reveal a potential explanation of how R282W initiates or responds to HCM.

The Tm-binding mutants changed a proposed Tm-binding region in C1 from RASK to either RRKK (adding positively charged amino acids) or EASE (reversing the charges from positive to negative). These mutants displayed, at physiologically relevant substoichiometric levels, increases and decreases in binding, respectively, to actin–Tm (Fig. 6 and Table 2). This is consistent with their effects on thin filament activation of myosin ATPase (Risi et al., 2018). Interestingly, they also show smaller, but significant, differences on bare actin (Fig. 6 and Table 2).

Testing of the TR-F assay with these mutations increases confidence in its ability to measure physiologically relevant changes in binding.

Cys-374–IAEDANS–actin lifetime changes will likely be specific for the binding of a subset of ABPs. Though lifetime was affected by the binding of C0–C2 and by the phosphorylation or mutation status of C0–C2, not all ABPs will alter the lifetime of IAEDANS attached to Cys-374. A good example, provided here, is Tm. Tm complexed with actin had no detectable effect on IAEDANS lifetime. Lifetimes for both were in a range of 17.52–17.68 ns over three different preparations (see Table 3). For the TR-F-actin–binding assay to be useful for ABPs that do not interact with Cys-374, it will be necessary to label actin on different locations on its surface. Given the detailed understanding of actin’s structure, and the ease of monitoring binding effects in the 384-well plate format, this is an attractive strategy to modify the current assay for other ABPs. Probes can be used to label actin at alternative sites that have been previously identified using muscle (Iwane et al., 2009) or yeast actin (Feng et al., 1997; Durer et al., 2012). Other strategies for labeling actin are available as well (Chen et al., 2012; Riedl et al., 2008).

The TR-F assay based on IAEDANS-labeled F-actin, in the 384-well plate format, has proven conducive to quick and easy optimization of conditions for the binding of cMyBP-C C0–C2 to actin and actin–Tm. Testing the binding of multiple mutant versions of C0–C2 and distinguishing between binding by phosphorylated and unphosphorylated states indicate it can be used as a high-throughput complementary assay to current cosedimentation assays that are much more labor intensive. We expect this assay to increase the rate of scientific discovery in understanding ABPs in health and disease. Z′ score analysis indicates the assay is suitable for screening for drugs that modulate cMyBP-C binding to actin or those that mimic the effects of phosphorylation on binding. In the absence of other screens with these specific capacities at the high-throughput level, the TR-F assay described here is a major advancement toward drug development based on cMyBP-C activity.

Richard L. Moss served as guest editor.

This work was supported by National Institutes of Health grants R00 HL122397 and R01 HL141564 (to B.A. Colson) and in part by a University of Arizona Sarver Heart Center Investigator Award (to B.A. Colson).

B.A. Colson filed a PCT patent application based on this work (patent pending, serial no. PCT/US21/14142). The other authors declare no competing financial interests.

Author contributions: B.A. Colson and T.A. Bunch designed the study and wrote the paper. T.A. Bunch, V.C. Lepak, and K.M. Bortz designed and purified recombinant proteins and DNA. T.A. Bunch and V.C. Lepak undertook purification of actin filaments for binding and TR-F assays. V.C. Lepak conducted sedimentation assays, and T.A. Bunch conducted TR-F assays. T.A. Bunch and B.A. Colson analyzed cosedimentation assays and fluorescence lifetime data. V.C. Lepak assisted with critical evaluation of the results and edited the manuscript. All authors critically evaluated and approved the final version of the manuscript.