Piperine, an alkaloid inhibiting the super-relaxed state of myosin, binds to the myosin regulatory light chain.

Abstract

Piperine, an alkaloid from black pepper, was found to inhibit the super-relaxed state (SRX) of myosin in fast-twitch skeletal muscle fibers. In this work we report that the piperine molecule binds heavy meromyosin (HMM), whereas it does not interact with the regulatory light chain (RLC)-free subfragment-1 (S1) or with control proteins from the same muscle molecular machinery, G-actin and tropomyosin. To further narrow down the location of piperine binding, we studied interactions between piperine and a fragment of skeletal myosin consisting of the full-length RLC and a fragment of the heavy chain (HCF). The sequence of HCF was designed to bind RLC and to dimerize via formation of a stable coiled coil, thus producing a well-folded isolated fragment of the myosin neck. Both chains were co-expressed in Escherichia coli, the RLC/HCF complex was purified and tested for stability, composition and binding to piperine. RLC and HCF chains formed a stable heterotetrameric complex (RLC/HCF)₂ which was found to bind piperine. The piperine molecule was also found to bind isolated RLC. Piperine binding to RLC in (RLC/HCF)₂ altered the compactness of the complex, suggesting that the mechanism of SRX inhibition by piperine is based on changing conformation of the myosin.

1. Introduction

Thermogenesis in resting skeletal muscles is modulated by conversion of myosin into a super-relaxed state (SRX) with an ATP turnover which is 10% of that observed for a myosin that is in a disordered state, (DRX) [1]. The SRX is characterized by a fixed spatial orientation of myosin heads [2], likely owing to their binding to the thick filament surface in a bent “backwards” conformation, known as the Interacting Heads Motif, IHM [3, 4].

Inhibiting the SRX by disordering its structure would increase the population of the DRX, thus leading to a considerably faster ATP hydrolysis in resting muscle and increased whole body thermogenesis. Understanding the molecular mechanisms that regulate the SRX is of great interest to researchers in the muscle field [5, 6]. More specifically, an approach to inhibit the SRX in skeletal muscles is thought to provide a way to combat obesity and type 2 diabetes [5, 6].

To find a proof-of-concept compound that can serve as a prototype for a therapeutic capable of inhibiting the SRX and increasing muscle thermogenesis, a library of the US Food and Drug Administration approved clinical drugs was screened against skinned rabbit fast skeletal muscle fibers [7]. Fluorescence of a probe attached to the myosin regulatory light chain was used as the screen readout [8]. Piperine, an alkaloid responsible for the pungency of black and long peppers, was identified as one chemical substance with the desirable functionalities, destabilizing the SRX and increasing metabolic rate. Importantly, piperine showed little effect on SRX in slow twitch skeletal fibers and rabbit cardiac tissue [7].

Myosin is an asymmetric molecule consisting of two heavy chains, two essential light chains (ELC) and two regulatory light chains (RLC). The chains are organized into a tail, and two head domains, each head domain comprising of a motor domain (MD) and a neck region (NR). The myosin tail region is formed by C-terminal parts of two heavy chains self-associated into a coiled coil structure. Each MD is formed by the N-terminal of one heavy chain and is responsible for binding to actin and ATPase activity. The NR of myosin connects the MD and the tail, and each NR is formed by one ELC and one RLC bound to a heavy chain [9]. ELC and RLC interact with the heavy chain via specific binding sites.

To find the location of the binding site for piperine on myosin, we tested piperine binding to myosin fragments of decreasing size. Specifically, we tested heavy meromyosin (HMM), RLC-free subfragment-1 (S1), a stable RLC-containing myosin fragment (referred to as myosin neck fragment in further text), and isolated RLC. The fragment of the myosin neck was designed to include a pair of heavy chain fragments (HCF) and two molecules of RLC, all forming a structurally stable complex, (RLC/HCF)₂. We showed that out of the four tested myosin fragments, only S1, which lacks the RLC, does not bind piperine. Additionally, we showed that the interaction of the complex (RLC/HCF)₂ with piperine leads to destabilization and a decrease in compactness of the (RLC/HCF)₂ complex.

2. Materials and Methods

2.1. DNA constructs of myosin heavy chain fragment (HCF) and the regulatory light chain (RLC)

A double strand DNA fragment encoding amino acid residues K810-E934 of the heavy chain of skeletal muscle myosin from Mus musculus (GenBank: AAI50740.1) was synthesized at Life Technologies (Thermo Fisher Scientific, Waltham, MA). The DNA sequence was optimized for Escherichia coli expression using an online GeneOptimizer algorithm [10]. In addition, Cys820 and Cys912 were replaced with Val and Ser, respectively, to avoid undesired disulfide bond formation at high protein concentrations and facilitate spectral studies. The synthetic DNA contained an N-terminal TEV cleavage site. Using a ligation-independent cloning protocol [11], the DNA encoding the myosin heavy chain fragment (HCF) was inserted into the pCDFDuet-1 vector (Novagen/EMD Millipore, Burlington, MA) linearized between the BamHI and EcoRI sites. We will refer to this plasmid as pCDFDuet-his-HCF, and to the protein product as his-HCF. The His-tagged RLC DNA construct was described previously [7, 8]. This plasmid will be referred to as pET28-his-RLC and the protein product as his-RLC.

2.2. Site-directed mutagenesis for the coiled-coil stabilization

The mutant plasmid pCDFDuet-his-HCF[K919I, A933L] encoding his-HCF[K919I, A933L] was created by two sequential site-directed mutagenesis steps. First, a mutant pCDFDuet-his-HCF[K919I] double-nicked plasmid was generated by extending a pair of complementary oligonucleotides with the desired mutation annealed to the “wild type” super-coiled template plasmid pCDFDuet-his-HCF. The sequence of the sense primer was 5’- CGT AGC GAT CAA CTG ATT AAA ACC ATC ATT CAG CTG GAA GC - 3’ (underlined are nucleotides containing the mutation). The DNA polymerase reaction was performed with a high fidelity Pfu Turbo DNA Polymerase (Agilent Technologies). Following the temperature cycling, the original “wild-type” template plasmid was diges ted with DpnI (New England Biolabs, Ipswich, MA). The generated plasmid encoding the mutant his-HCF[K919I] was used to transform NEB® 5-alpha Competent E. coli (High Efficiency) cells (New England Biolabs). Plasmids from several spectinomycin-resistant E. coli clones were purified and submitted for sequencing. The plasmid from a clone with the desired nucleotide substitutions was used as a template for the second site-directed mutagenesis to generate the double mutant pCDFDuet-his-HCF[K919I, A933L]. The sense primer used for the generation of the second mutation was 5’- CAA AGA AGT TAC CGA ACG CTT AGA ATA ATC GAG CTC GGC G-3’. All primers were synthesized and purified by Integrated DNA Technologies (Coralville, IA). Purification of plasmids was performed using the QIAprep® Spin Mini prep Kit (Qiagen, Hilden, Germany). Gene sequencing for all constructs were performed at Genewiz, Inc. (South Plainfield, NJ).

2.3. Creating a plasmid encoding a his-tag-less version of RLC

To create a plasmid encoding a version of RLC without the His-tag (pET28-RLC) we first linearized the pET28-his-RLC plasmid with an NcoI restriction enzyme (New England Biolabs). The NcoI restriction enzyme cuts the pET28-his-RLC plasmid near the start codon and in the middle of the RLC sequence. The 5’ ends of the digested plasmid were de-phosphorylated with Antarctic Phosphatase (New England Biolabs) to prevent self-religation. The linearized plasmid was recombined with a PCR-generated DNA fragment encoding an N-terminus of the RLC flanked with two NcoI recognition sites. As a consequence of recombination Pro3 changed to Ala, and it was reverted to Pro using site-directed mutagenesis. The sequence of the mutagenic sense primer was 5’-GAA GGA GAT ATA CCA TGC CCA AGT GCG CCA AGA G-3’, and the protocol was essentially the same as the one used to alter the his-HCF sequence.

2.4. Expression and purification of recombinant proteins

Expression and purification of the his-RLC/his-HCF, RLC/his-HCF and RLC/his-HCF[K919I, A933L] protein complexes were performed very similarly. A mixture of plasmids pET28-RLC (or pET28-his-RLC) and pCDFDuet-his-HCF (or pCDFDuet-his-HCF[K919I, A933L]) was used to transform competent Rosetta 2 (DE3) E. coli (EMD Millipore) by giving the cells a 30 sec heat shock at 42°C. The transfor med bacteria were grown at 37°C on LB plates with chloramphenicol (34 µg/mL), kanamycin (50 µg/mL) and spectinomycin (25 µg/mL) as selection antibiotics.

A colony of the co-transformed E. coli cells was used to inoculate 50 mL Super LB medium which contained 2% tryptone, 1% yeast extract, and 1% NaCl (w/v). The medium was supplied with chloramphenicol (34 µg/mL), kanamycin (50 µg/mL) and spectinomycin (25µg/mL) and the culture was incubated at 37°C with sh aking at 250 rpm. When OD⁶⁰⁰ reached ~0.35, the culture was transferred into 1 L Super LB supplied with the same antibiotics, and the culture was grown at 37°C until OD ⁶⁰⁰ reached 0.7–0.8. Protein expression was induced with 0.15 mM IPTG and the cells were incubated with continuous shaking at 37°C for 2 hours. The cells were harvested by centrifugation at 4000 g for 30 minutes and frozen at −30°C until further use. Frozen E. coli cells were thawed on ice and resuspended in 10 mL buffer A (50 mM Tris-Cl, pH 8.0, 200 mM NaCl, 2 mM MgCl₂, 5% sucrose, 25 mM imidazole) supplied with 2 mM β-mercaptoethanol, 0.1 mM PMSF and 0.5 tablet of cOmplete, Mini, EDTA-free Protease Inhibitor Cocktail (Roche Life Science, Indianapolis, IN). The cells were disrupted by sonication on slush ice on a Branson Digital Sonifier equipped with a flat tip, at 25% amplitude, for a total of 10 minutes, using 2 second pulses interrupted by 8 second rest periods to prevent overheating of the sample. After the sonication was complete, 10 mL buffer A (50 mM Tris-Cl, pH 8.0, 200 mM NaCl, 2 mM MgCl₂, 5% sucrose, 25 mM imidazole) supplied with 2 mM β-mercaptoethanol and 0.1 mM PMSF was added and cell debris was removed by centrifugation at 16000 rpm (Beckman JA-17 rotor) for 30 minutes at 4°C.

The protein complexes were purified on a Ni affinity column at 4°C. The cleared cell lysate was mixed with ~5 mL of Ni-NTA Superflow resin (Qiagen) and rocked for 30 minutes. The resin was loaded on a column and washed with buffer B (50 mM Tris-Cl, pH 8.0, 200 mM NaCl, 2 mM MgCl₂, 5% sucrose, 50 mM imidazole) supplied with 2 mM β-mercaptoethanol. In preparation of the protein complexes RLC/his-HCF and RLC/his-HCF[K919I, A933L] the excess RLC was removed during the washing step. The stoichiometric protein complex was eluted in buffer C (50 mM Tris-Cl, pH 8.0, 200 mM NaCl, 2 mM MgCl₂, 5% sucrose, 200 mM imidazole) and 1 mM dithiothreitol (DTT) was added to the eluted protein. The presence of approximately 1:1 polypeptide chains in the eluted complex was confirmed by SDS-PAGE for both RLC/his-HCF and RLC/his-HCF[K919I, A933L]. The presence of two species with molecular weight close to predicted was confirmed by mass spectrometry for RLC/his-HCF[K919I, A933L]. In preparation of the protein complex his-RLC/his-HCF, after elution from the Ni-NTA resin, the excess his-RLC was removed by size-exclusion chromatography (Superdex 200 10/300 GL, GE Healthcare, Chicago, IL) in 50 mM MOPS at pH 6.8, 150 mM potassium acetate, 1mM MgCl₂, 1mM EGTA, 0.5 mM TCEP.

His-RLC was prepared and purified by following essentially the same protocol as was used to prepare RLC/his-HCF complexes, except spectinomycin (the antibiotic for pCDFDuet-1 plasmid selection) was omitted from bacterial cultures. To isolate RLC (the His-tag-less version of RLC), the purified RLC/his-HCF[K919I, A933L] complex was loaded on the His60 Ni Superflow Resin (Clontech) and RLC was eluted with 50 mM Tris-Cl, pH 8.0, 200 mM NaCl, 2 mM MgCl₂, 5% sucrose, 8 M urea. The RLC protein was incubated with 40 mM DTT for 1 hour and refolded via dialysis against several changes of 20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 0.5 mM MgCl2, 0.5 mM DTT with decreasing concentrations of urea (6 M, 5 M, 4 M, 3 M, 2M, 1M, 0.5 M, and 0 M).

Recombinant protein concentrations were determined by measuring a difference in 294 nm absorption at pH 7.0 and 12.5 in 6 M guanidine-HCl. The concentrations were calculated using the extinction coefficient of 2.357 cm⁻¹ mM⁻¹ per each tyrosine and 0.830 cm⁻¹ mM⁻¹ per each tryptophane [12].

2.5. Purification of HMM and S1

HMM and RLC-free, chymotryptic S1 were prepared from rabbit skeletal muscle as described in [13] and stored frozen at −20°C until use. The RLC of HMM was dephosphorylated as shown by isoelectric focusing gel electrophoreses. Concentrations of S1 and HMM (per myosin head) were determined using the 280 nm extinction coefficients of 84.7 and 117.3 cm⁻¹ mM⁻¹ respectively.

2.6. Glutaraldehyde crosslinking

RLC/his-HCF[K919I, A933L] (200 µL, at 0.5 mg/mL) was dialyzed against the reaction buffer (5 mM HEPES buffer, pH 7.5) at 4°C. Differen t concentrations of glutaradehyde (final 0.025 to 0.1% v/v) were added to the 0.5 mg/mL RLC/his-HCF[K919I, A933L] in the same buffer and incubated for 15 or 30 minutes at room temperature. The reactions were stopped by adding 250 mM Tris-HCl, pH 8.0, and analyzed on 15% SDS-PAGE.

2.7. Preparation of NMR samples and NMR spectra collection

RLC/his-HCF[K919I, A933L] was dialyzed overnight against 100 mM sodium/potassium phosphate buffer, pH 7.8, at 4°C, with three buffer changes. His-RLC and rat muscle α-tropomyosin Tpm1.1 were dialyzed overnight against 100 mM sodium/potassium phosphate buffer, pH 7.8, 0.1 mM MgCl₂, at 4°C, with two buffer changes in the presence o f 0.5 mM DTT. To remove DTT, the last several hours of dialysis was performed against 100 mM sodium/potassium phosphate buffer, pH 7.8, 0.1 mM MgCl₂, at 4°C. Chicken G-actin was dialyzed overnight against 2 mM potassium phosphate buffer, pH 7.4, 0.1 mM CaCl₂, 0.1 mM ATP, with two buffer changes, and for two hours against 2 mM potassium phosphate buffer, pH 7.4, 0.1 mM CaCl₂, 0.01 mM ATP.

To remove aggregates and potential contamination with the RLC or RLC fragments, S1 was centrifuged at 10,000 g for 10 minutes and passed through the Superdex 75 10/300 GL column (GE Healthcare) at 0.5 mL/min in 100 mM sodium/potassium phosphate buffer, pH 7.8, 1 mM MgCl₂, 0.5 mM DTT, at 4°C. HMM and S1 were dialyzed over night against 100 mM sodium/potassium phosphate buffer, pH 7.8, 1 mM MgCl₂, 0.5 mM dithiothreitol (DTT), at 4°C, with two buffer changes, and for four hours against 100 mM sodium/potassium phosphate buffer, pH 7.8, 1 mM MgCl₂.

All proteins were used in NMR experiments immediately after the dialysis step. To 0.6 mL of the appropriate protein concentration were added 60 µL of D₂O and 1.5 µL of 40 mM piperine solution in deuterated dimethyl sulfoxide (DMSO-d₆). Special care was taken to avoid exposure of piperine solutions to ambient light since piperine is known to undergo light-induced cis-trans isomerization, which is readily observed by NMR [14]. We did not observe any detectable isomerization upon the end of titration experiments. In these and other experiments the concentration of piperine was 90–100 µM to saturate the site and as this is the concentration used in previous work [8].

1H NMR spectra were recorded utilizing PURGE water suppression at 25°C on a Varian Inova 500 spectrometer (500 MHz) equipped with a 5 mm triple resonance probe. Each spectrum was an accumulation of 128 or 256 scans. The spectra were processed, visualized and analyzed with Mnova (Mestrelab Research, Santiago de Compostela, Spain). Errors in peak intensities were estimated on the basis of S/N calculated by Mnova. The piperine proton spectral assignment in CDCl₃ was from the Spectral Database for Organic Compounds (National Institute of Advanced Industrial Science and Technology, Japan), and reconfirmed by DQF-COSY, peak integration and observed patterns of spin-spin splitting.

2.8. Preparation of samples for circular dichroism (CD) experiments and CD measurements

For CD measurements, RLC/his-HCF and RLC/his-HCF[K919I, A933L] were incubated for 1 hour with 10 mM DTT and dialyzed overnight against 20 mM Tris-HCl, pH 7.8, 100 mM NaCl, 0.5 mM MgCl₂, 0.5 mM DTT at 4°C, with three buffer changes. Th e dialyzed sample was diluted before CD measurements to ~0.1 mg/mL with the same buffer. For urea-induced protein denaturation experiments, RLC (0.8 mg/mL) and RLC/his-HCF[K919I, A933L] (2.55 mg/mL) were dialyzed against 20 mM Tris-HCl, pH 7.8, 100 mM NaCl, 0.5 mM MgCl₂ and 0.5 mM DTT. Stock solution of urea (10 M) prepared fresh in the same buffer was added to the protein complex to the desired urea concentration. The final concentrations of RLC and RLC/his-HCF[K919I, A933L] were ~0.2 mg/mL. Piperine was added as a 15 mM methanol solution to the final concentration of 100 µM (2.0 µL piperine solution per 0.3 mL sample). Same volumes of methanol were added to control samples without piperine.

We performed three independent urea denaturation experiments on two independently prepared batches of RLC/his-HCF[K919I, A933L]. For each urea denaturation experiment, the points were measured “side-by-side” in the presence or absence of piperine. All of the experiments led to the same conclusion, which is presented in the paper. Due to technical difficulties in ensuring the sufficiently close protein concentrations between different independent experiments and impossibility of data normalization for urea titrations, the numerical data between the experiments were not averaged. The latest experiment is shown in Figure 9. The errors shown in Figure 9 are the estimated standard deviation of the averaged signal after 5 sec averaging, as provided by the software controlling the Aviv CD spectropolarimeter. The time constant was 0.1 sec.

Figure 9.

Urea-induced protein denaturation of the (RLC/his-HCF[K919I, A933L])₂ complex and RLC (B) at 222 nm in the presence (filled circles) and absence (open circles) of piperine. 0.2 mg/mL protein (2.7 µM (RLC/his-HCF[K919I, A933L ])₂ and 10.6 µM RLC) was used for both titrations. The titrations were performed in 20 mM Tris-Cl, pH 7.8, 100 mM NaCl, 0.5 mM MgCl₂, 0.5 mM DTT and varying concentrations of urea at 4ºC. Errors are shown as vertical bars. The piperine concentration was 100 µM.

Far-UV CD spectra were recorded in quartz cuvettes with 1 mm path length (Hellma Analytics, Plainview, NJ) on an Aviv model 400 spectropolarimeter (Lakewood, NJ) at 0°C and 19°C. In urea-induced protein denaturation experime nts the CD signal at 222 nm was recorded at 4°C .

2.9. Small-angle X-ray Scattering (SAXS) measurements

To remove the N-terminal 6xHis tag from the his-HCF[K919I, A933L] chain prior to SAXS measurements, aliquots of the complex RLC/his-HCF[K919I, A933L] (at ~2 mg/mL, 1 mL volume) were incubated with 50 µg of TEV proteas e for 3–4 hours at room temperature. The completeness of TEV cleavage was verified by SDS-PAGE with 1X MES running buffer. The TEV-treated complex (RLC/HCF[K919I, A933L]) was purified by the size-exclusion chromatography (Superdex 200 10/300 GL, GE Healthcare) in 50 mM MOPS at pH 6.8, 150 mM potassium acetate, 1mM MgCl₂, 1mM EGTA, 0.5 mM TCEP. Monodisperse peak corresponding to the complex was pooled and concentrated with a spin-concentrator (10 KDa MW cutoff, EMD Millipore). X-ray scattering was recorded on an Anton Paar SAXSESS mc² instrument (Graz, Austria). An average of 150 frames (30s/frame exposure) was collected for each sample. The apo sample contained 0.5%(v/v) DMF to match addition of piperine dissolved in DMF. Experiments were done with 40 µM RLC/HCF[K9 19I, A933L] (2.7 mg/mL protein). P(r) error evaluation was done by GNOM (ATSAS package) [15].

The data analysis was performed using custom software written in Python/Fortran (available at https://github.com/delnatan/UCSFsaxs). It implements a Bayesian algorithm for choosing optimal particle dimension (Dmax) and smoothness [16].

3. Results.

3.1. Piperine binds to HMM, but not to RLC-free S1

We hypothesized that to be able to destabilize SRX as observed previously [7] piperine should bind to the myosin head. To test if we can detect piperine binding to the myosin head and narrow down the location of the binding site, we studied interactions of piperine with myosin fragments HMM and chymotryptic S1 using NMR differential line broadening [17, 18] (Figures 1 and 2, respectively). We compared 1D ¹H-NMR piperine spectra in the absence and presence of increasing concentrations of either HMM or S1. Upon addition of HMM, the resonance peaks broadened in a dose dependent manner. The peak broadening is particularly evident from the loss of resolution of peaks at ~6.8 and ~7.1 ppm, and from peak height attenuation. While HMM displayed interaction with piperine, (Figure 1), we did not observe clearly detectable piperine line broadening induced by comparable concentrations of S1 (Figure 2). This suggests binding of piperine to the neck region of myosin in the vicinity of RLC, since chymotryptic S1 is RLC-free [19], or to the coiled-coil subfragment 2 (S2).

Figure 1.

Benzodioxole/alkene region of piperine 1D ¹H-NMR spectra in the absence (A) and presence (B, C, D) of HMM. The spectra were recorded at room temperature, on a 500 MHz Varian NMR instrument. The concentration of piperine was 100 µM. Concentrations of HMM per myosin head were (B) 0.5 µM, (C) 1.0 µM, and (D) 1.6 µM.

Figure 2.

Benzodioxole/alkene region of piperine 1D ¹H-NMR spectra in the absence (A) and presence (B, C, D) of S1. The spectra were recorded at room temperature, on a 500 MHz Varian NMR instrument. The concentration of piperine was 100 µM. Concentrations of S1 were (B) 0.5 µM, (C) 1.0 µM, and (D) 1.5 µM.

As a control experiment for piperine specificity, we performed two similar NMR titrations using G-actin or tropomyosin. Neither G-actin nor tropomyosin had a detectable effect on line widths of piperine proton resonances at similar concentrations of piperine and the protein (Supplementary materials, Figures S1A and S1B).

3.2. Selection of the HCF fragment including a complete binding region for RLC

The data described above suggests that piperine may bind in the vicinity of the RLC. To test if piperine binds to the neck region of myosin in the vicinity of RLC, we designed and prepared a stable protein complex that contains RLC and a fragment of the myosin heavy chain (HCF). In addition to the RLC-binding sequence, HCF included a part of the myosin tail allowing it to dimerize via a coiled-coil structure.

Six PDB structures (3J04, 1I84, 2W4H, 3DTP, 2MYS and 1B7T) were inspected to identify the putative binding interface between RLC and the heavy chain from the mammalian skeletal muscle myosin. Based on sequence homology, we chose the heavy chain sequence K810-S849 as the segment encompassing the entire RLC binding region. Residues from the coiled coil region, A850-E934, were added to ensure the formation of a putative four-chain (RLC/HCF)₂ protein complex. A three-dimensional (3D) model of the (RLC/HCF)₂ complex (Figure 3A) was built on the basis of the 3D CryoEM structure of the chicken smooth muscle HMM (PDB 3J04, [20]) and the CryoEM structure of the relaxed tarantula thick filament, PDB 3DTP [4]. The length of the coiled-coil region was chosen to be sufficient for dimerization, after modification, but short enough to allow crystal formation for future structural studies.

Figure 3.

A: A model of the (RLC/HCF)₂ complex. Two HCF strands forming a coiled coil in their C-terminal region are shown in black, and two RLC strands are shown in gray. B: Sequences of RLC and HCF versions used in this study (capital letters). Coiled-coil stabilizing mutations K919I and A933L in HCF are shown in bold on gray background. Small letters (abcdefg) display the amino acid residue position in the consensus coiled-coil heptad.

The DNA fragment encoding RLC was cloned into a pET-28a vector (Novagen), whereas the DNA fragment encoding HCF was cloned into a pCDFDuet-1 vector (Novagen). These two plasmids have compatible origins of replication allowing co-expression of inserted sequences. Both expressed proteins (his-RLC and his-HCF) contained an N-terminal His-tag. The complete amino acid sequences of the expressed his-RLC and his-HCF proteins are shown in Figure 3B.

3.3. RLC and his-HCF form an equimolar molecular complex

In the initial experiments on his-RLC/his-HCF co-expression we noticed that his-RLC was produced at markedly higher yields than his-HCF in Rosetta 2 (DE3) cells. As a result, upon purification on Ni-NTA agarose we obtained a mixture of the complex and excess his-RLC. An additional gel filtration step of purification was required to separate the complex and unbound his-RLC. To facilitate purification of the complex with the equimolar RLC/HCF ratio, we modified the RLC expression plasmid and eliminated the N-terminal His-tag from the original his-RLC construct. The newly generated protein construct was named RLC (Figure 3B). The separation of the non-covalent complex RLC/his-HCF was performed on a Ni-NTA column viathe His-tag on his-HCF. Since RLC does not have a His-tag itself, it was only retained on the column when bound to his-HCF. A single washing step prior to the elution of the complex removed excess RLC. The RLC/his-HCF complex eluted in the presence of 200 mM imidazole was analyzed by SDS-PAGE, which showed two bands with an approximately 1:1 ratio in the intensity of staining. The bands had very close mobilities but were visibly resolved and matched the calculated molecular weights of RLC and his-HCF (Supplementary materials, Figure S2A).

3.4. HCF mutant (his-HCF[K919I, A933L]) forms a more stable coiled coil than his-HCF

Circular dichroism (CD) analysis of the purified RLC/his-HCF protein complex indicated that his-HCF does not form a sufficiently stable coiled coil structure, which affected its ability to form the desired (RLC/his-HCF)₂ complex. We inspected the HCF sequence and found two amino acid residues in positions a of a characteristic heptad repeat (abcdefg)_n, that appeared to contribute to the destabilization of the coiled coil (Figure 1B). The geometric requirements of tight coiled-coil packing result in a preference for large hydrophobic residues in positions a and d [21, 22]. Therefore, to stabilize the coiled coil while not increasing its length, we replaced Ala933 with a bulky hydrophobic Leu using site-directed mutagenesis (Figure 1B). Additionally, since Lys919 did not have a negatively charged partner to form a stabilizing salt bridge, it was replaced with hydrophobic Ile (Figure 3B).

A comparison between CD spectra of complexes RLC/his-HCF and RLC/his-HCF[K919I, A933L] is shown in Figure 4. A negative band at 222 nm as well as a positive band at 195 nm, both characteristic of α-helical secondary structure, are more pronounced for the RLC/his-HCF[K919I, A933L] protein complex. The intersection of the spectrum with the abscissa shifted to higher wavelengths for the complex with the mutated HCF, which is also indicative of a decrease in disorder. This suggests higher helical content and stabilization of the coiled coil interactions in the his-HCF[K919I, A933L] compared to his-HCF. Consequently, in further studies we used RLC/his-HCF[K919I, A933L].

Figure 4.

Mean residue ellipticity of RLC/his-HCF (gray) and RLC/his-HCF[K919I, A933L] (black). The spectra were recorded at 19°C in 20 m M Tris-HCl, pH 7.8, 100 mM NaCl, 0.5 mM MgCl₂, 0.5 mM DTT.

3.5. RLC/his-HCF[K919I, A933L] complex is a tetramer consisting of two RLC and two his-HCF[K919I, A933L] molecules.

Similarly to the RLC/his-HCF (Supplementary materials, Figure S2A), SDS-PAGE analysis showed that the stoichiometric ratio of the RLC and his-HCF[K919I, A933L] chains in the eluted complex was approximately equal to 1. In addition, the complex eluted as a single major band on size-exclusion chromatography (Supplementary materials, Figure S2B), consistent with forming a stable RLC/his-HCF[K919I, A933L] binding interface shown in (Figure 3A). In comparison with gel filtration globular protein standards, the elution volume of 14 mL that we observed for the complex roughly corresponds to ~70 kDa, while the calculated molecular weight of a heterotetramer consisting of two his-HCF[K919I, A933L] and two RLC chains is ~72.6 kDa. Observation of a comparatively high α-helical content in the RLC/his-HCF[K919I, A933L] by CD also suggests that a heterotetramer kept together by a stabilized coiled coil was formed. To further confirm that the tetramer was formed we performed glutaraldehyde cross-linking followed by SDS-PAGE separation of the covalently bonded polypeptide chains.

Glutaraldehyde is a somewhat indiscriminant cross-linking reagent forming covalent bonds primarily with ε-amino groups of lysines. However, other amino acid residues can also react with glutaraldehyde [23]. The result of a glutaraldehyde reaction with even a single polypeptide chain protein is a collection of cross-linked polypeptide species with slightly different mobilities that will appear as a diffuse band on a SDS-PAGE protein gel. Typically, to obtain a set of cross-linked species that are minimally dispersed and well resolved on a SDS-PAGE gel, it is important to stop the reaction early, before too many covalent bonds with glutaraldehyde are formed [24, 25].

To capture the entire spectrum of possible sets of cross-linked polypeptides in the tetramer complex, we tested several conditions with two reaction times and different glutaraldehyde concentrations (Figure 5). We observed two distinct bands that matched the molecular weight of dimers, consistent with formation of his-HCF[K919I, A933L]-his-HCF[K919I, A933L] homodimers, RLC-RLC homodimers, and/or his-HCF[K919I, A933L]-RLC heterodimers, and faint heavier bands that could potentially represent trimers or tetramers. If a tetramer complex was not forming, we would have observed only one dimer band corresponding to the his-HCF[K919I, A933L]-RLC heterodimer.

Figure 5.

Cross-linking of the RLC/his-HCF[K919I, A933L] complex with glutaraldehyde. Left two lanes are BioRad Precision Plus Unstained Protein Standard and the untreated RLC/his-HCF[K919I, A933L] complex, respectively. For the remaining lanes, percentage of glutaraldehyde in solution is shown at the bottom of the figure, whereas the reaction time is shown at the top. Arrows on the right side of the figure indicate bands corresponding to MW of monomers, dimers, and putative trimers/tetramers. Heavy bands at the top of the lanes corresponding to 0.05% and 0.1% glutaraldehyde are likely to be high molecular weight aggregates forming upon extensive crosslinking.

3.6. Piperine interacts with (RLC/his-HCF[K919I, A933L])₂ and his-RLC.

We compared 1D ¹H-NMR piperine spectra in the absence and presence of increasing concentrations of the RLC/his-HCF[K919I, A933L] heterotetramer, which we will call hereafter (RLC/his-HCF[K919I, A933L])₂ to reflect the stoichiometry of the protein complex (Figure 6) and his-RLC (Figure 7). Upon addition of the proteins, the resonance peaks broadened in a dose dependent manner similarly to HMM (Figure 1). Compared to HMM, we used higher concentrations of the proteins to compensate for the decrease in the molecular weight and therefore slower transverse relaxation of the complexes. This confirms binding of piperine to the complex and to his-RLC alone. We could not test piperine binding to the his-HCF[K919I, A933L] alone, because this protein fragment precipitates if it is not in a complex with the RLC.

Figure 6.

Benzodioxole/alkene region of piperine 1D ¹H-NMR spectra in the absence (A) and presence (B-D) of (RLC/his-HCF[K919I, A933L])₂. The spectra were recorded at room temperature, on a 500 MHz Varian NMR instrument. The concentration of piperine was 90 µM. Concentrations of RLC/his-HCF[K919I, A933L] (per RLC chain) were (B) 2.5 µM, (C) 5.1 µM, and (D) 7.6 µM. This corresponds to (B) 1.25 µM, (C) 2.55 µM, and (D) 3.8 µM of the (RLC/his-HCF[K919I, A933L])₂ complex.

Figure 7.

Benzodioxole/alkene region of piperine 1D ¹H-NMR spectra in the absence (A) and presence (B, C) of his-RLC. The spectra were recorded at room temperature, on a 500 MHz Varian NMR instrument. The concentration of piperine was 90 µM. Concentrations of his-RLC were (B) 1.8 µM, (C) 3.6 µM, and (D) 5.5 µM.

3.7. Piperine binding destabilizes the RLC and the (RLC/his-HCF[K919I, A933L])₂ complex.

Urea-induced protein denaturation of the (RLC/his-HCF[K919I, A933L])₂ complex and RLC in the presence and absence of piperine was studied at 4ºC by measuring the CD signal at 222 nm as a function of urea concentration (Figure 9). Statistically significant CD absorption differences between unfolding of the samples with and without piperine were observed in the urea range of 0–1.75 M for the complex (p<0.02). We also observed an effect (although weaker) of piperine on unfolding of RLC between 0.5 and 1.0 M urea (p~0.05). In both RLC and the (RLC/his-HCF[K919I, A933L])₂ complex, piperine binding caused urea-induced decreases in α- helical content to shift to lower urea concentrations. This indicates a decrease in stability of both the complex and RLC in the presence of piperine. The effect of piperine on denaturation of the complex appears to be biphasic, where the effect of piperine mostly manifests itself in the range of 0–1.75 M urea concentration. The piperine-sensitive change of ellipticity of isolated RLC upon denaturation matches the ellipticity change (~10 millidegrees) and the range of urea concentrations (<2 M urea) in the complex. We assume that the piperine-sensitive part of the complex denaturation curve corresponds to denaturation of the RLC protein component. This behavior is consistent with the NMR titration data showing that RLC directly binds piperine.

3.8. Piperine binding reduces the compactness of the (RLC/HCF[K919I, A933L])₂ complex.

The piperine-induced destabilization of RLC and the (RLC/his-HCF[K919I, A933L])₂ complex in CD experiments was rather small. To confirm that piperine binding causes structural changes in the (RLC/his-HCF[K919I, A933L])₂ complex, we used small-angle X-ray scattering (SAXS). For SAXS experiments the flexible His-tag of the his-HCF[K919I, A933L] chain (Figure 3B) was proteolytically cleaved with TEV protease (recognition site ENLYFQ|S) and the resulting (RLC/HCF[K919I, A933L])₂ complex was purified by size-exclusion chromatography (Supplementary materials, Figure S3). The pair-distance distribution P(r) obtained from SAXS is a histogram of all distances between atoms in the protein. Without piperine, the (RLC/HCF[K919I, A933L])₂ complex appears to be an elongated molecule with a maximum dimension D_max of ~150 Å (radius of gyration of 36.5 Å) (Figure 8), which is consistent with the structural model of the complex (Figure 1A). The addition of piperine causes larger distances to appear in the P(r) distribution function, increasing the D_max to 185 Å (radius of gyration 51.3 Å). The P(r) function broadens at distances larger than 80 Å, while the major peak at ~35 Å is little affected. This suggests that most of the structure remains intact with larger distances coming from changes affecting only parts of the molecule. A similar (~35 Å) increase in D _max was also observed for the complex before the TEV cleavage, except that the D_max value changed from 168 to 202 Å upon the piperine addition. Larger values of D_max as compared to the TEV cleaved complex were from the uncleaved N-terminal polypeptide tag, but the change caused by piperine was essentially the same, therefore excluding the possibility that the increase was due to impurities introduced by TEV cleavage.

Figure 8.

Piperine-induced conformational change in the (RLC/HCF[K919I, A933L])₂ complex as seen by SAXS. Pairwise-distance histogram, P(r), obtained from 40 µM of the (RLC/HCF[K919I, A933L])₂ complex with (open squares) or without 100 µM piperine (apo, 0.5% DMF control, closed circles). Error bars represent one standard deviation. The mode and width of the peak around 35 Å is similar between samples. Addition of piperine causes broadening of the minor shoulder peak around 120 Å and causes larger distances to appear.

3.9. The mode of piperine binding to RLC and the RLC/his-HCF[K919I, A933L] complex is similar to that for HMM.

We compared NMR peak attenuation patterns for isolated proton resonances upon the addition of his-RLC, (RLC/his-HCF[K919I, A933L])₂ and HMM (Figure 10). The plots shown in Figure 10 revealed a consistent pattern of peak attenuation among different complexes, with NMR peak attenuation of methylenedioxy protons being the largest, peak attenuation of benzene ring protons being the smallest, and the alkene protons displaying intermediate peak attenuation. This suggests that piperine binding modes in complexes with RLC, (RLC/HCF)₂, and HMM are somewhat similar.

Figure 10.

Piperine NMR peak attenuation upon the addition of receptor protein complexes. Panels A, B, and C display peak attenuation for titration with his-RLC, (RLC/his-HCF[K919I, A933L])₂, and HMM, respectively. The attenuation is shown for methylenedioxy protons (closed triangles, δ~5.86 ppm), benzene ring atoms (closed and open circles, δ~7.02 ppm and ~6.9 ppm, respectively), and alkene linker protons (closed squares and open triangles, δ~7.13 ppm and ~6.5 ppm, respectively). Error bars are estimated on the basis of S/N values (see Materials and Methods). The concentrations for (RLC/his-HCF[K919I, A933L])₂, and HMM are indicated per RLC chain and per myosin head, respectively.

4. Discussion

For a large multi-domain protein such as myosin, identification of the protein fragments or domains binding to its inhibitors provides the approximate location of functionally important molecular interfaces. This also offers a deeper understanding of how the inhibited modules contribute to the function of the entire protein or protein complex. Piperine has been shown recently to destabilize SRX in fast twitch skeletal muscle fibers [7]. In the structure that has been proposed to account for the low ATPase activity of the SRX, the myosin heads bind to each other, stabilized by a number of interfaces any one of which could be a target for piperine [26, 27]. In this work, to localize the binding site for piperine on myosin, we studied interactions of piperine with myosin fragments of gradually decreasing size. We expected that if piperine interacts with myosin via a specific and comparatively small interface, the formation of a complex with piperine could be detected for any myosin fragment where the binding site is present and not occluded by much stronger intra-protein contacts or not completely denatured. Even if reducing the size of such myosin fragments causes subtle structural distortions in the piperine binding site leading in turn to the increase in the dissociation constant K_d, binding can still be detected by NMR which provides magnetization relaxation-based approaches to detect weak interactions.

The piperine binding to myosin fragments was tested using NMR differential line broadening, which can be used to detect a broad range of relatively weak interactions (K_d~µM-mM) [17, 18]. Ligand NMR line broadening occurs as a consequence of kinetic exchange between the free and receptor-bound ligand states resulting in an increase of the transverse relaxation rates R₂. Binding-induced proton resonance line broadening is a straightforward, well established and a commonly used NMR binding assay. It is observed when the kinetic exchange rate between the protein-bound and unbound ligand species is not too slow or too fast on the chemical shift timescale [28]. In a typical NMR titration experiment where the concentration of the ligand bound to a protein is much smaller than the concentration of an unbound ligand, the population of the unbound species (and therefore the integral intensity of the dominant peak) remains almost unchanged. Since the integral of an NMR resonance peak is proportional to the species concentration and remains approximately constant throughout a titration, the line broadening is accompanied by peak intensity attenuation.

We found that piperine binds to HMM but it does not bind to RLC-free S1, placing the binding site for piperine in the vicinity of RLC. We observed piperine binding to myosin fragment (RLC/HCF)₂, and to RLC alone. Piperine is a hydrophobic molecule and can potentially bind non-specifically to proteins with large exposed hydrophobic surfaces. However, when we tested other components of the muscle molecular machinery (G-actin and tropomyosin) and RLC-free S1, none of them showed clearly detectable binding to piperine. This confirms that piperine binding to RLC and RLC-containing complexes is specific.

Although the NMR line broadening effects are the consequence of kinetic exchange processes and are dictated by dissociation/association rates, we can roughly estimate the K_d upper limit for the piperine/protein complex by making simple assumptions. Since binding induces piperine line broadening but no observable proton resonance shift, we can estimate the two-state dissociation rate constant k_off to be < ~2500 s⁻¹ [18]. For a diffusion-controlled binding mechanism the on-rate constant k_on is expected to be around 10⁹ M⁻¹s⁻¹ [29], which suggests that K_d = k_off/k_on is <2.5 µM. This is consistent with observations made previously evaluating the K_d to be approximately in the range of 1–3 µM [7]. This upper bound estimate is only true for the diffusion-limited association. With slower on-rates the upper bound estimate for the K_d that can be derived from the NMR data will rise proportionally.

Generally, the transverse relaxation rate of the dominant (unbound) resonance line depends on the chemical shifts of the protein-bound and free ligand states, relative populations of the states, kinetics of association/dissociation and also on internal protein dynamics [30–32]. If we do not assume any kinetic model of piperine-protein interaction, we can use resonance peak attenuation as an empirical measure of binding to a protein surface. Comparison of the patterns of peak attenuations in the presence of RLC, (RLC/HCF)₂, and HMM suggests that the modes of binding are similar. If we assume a two-site exchange model, a considerable increase in peak attenuation for the pair of protons in the methylenedioxy functional group is indicative of a larger change in chemical shift caused by the piperine binding [30, 31]. A large complexation-induced change of the chemical shift is typical for protons located in a deep pocket and/or being in direct contact with an aromatic ring [18]. This would make the benzodioxole moiety a good specific molecular “anchor” and a starting point for fragment-based drug design.

The CD study of urea-induced denaturation showed that piperine binding to the (RLC/HCF)₂ tetramer is accompanied by the destabilization of the protein. Comparison of the unfolding curves of the (RLC/HCF)₂ tetramer and RLC indicated that unfolding of RLC is affected by piperine. Interestingly, the piperine-induced difference in unfolding observed for the (RLC/HCF)₂ complex is noticeably higher than that for the isolated RLC.

Typically, ligand binding causes stabilization of receptor proteins, but scenarios with destabilization of the native conformation upon formation of a complex have also been reported. In one of them, destabilization of a protein complex upon ligand binding can be a consequence of the ligand having a higher affinity for an unfolding intermediate [33]. Another possibility is that ligand binding can stabilize one part of a protein while destabilizing another part of the protein [34, 35]. One of the principal interfaces between myosin heads in SRX involves interaction between the two RLC subunits [26], which may also form in the (RLC/HCF)₂ tetramer. Considering a lower effect of piperine binding on the stability of the RLC alone, the increased destabilization of the (RLC/HCF)₂ complex suggests long-range effects of piperine binding, possibly affecting the putative RLC-RLC interactions in the tetramer. This destabilization may alone be the reason for the piperine-induced inhibition of the SRX reported previously [7].

Pairwise interatomic distance distribution obtained from SAXS experiments show an increase in the dimensions of the complex when piperine was added to the (RLC/HCF)₂ tetramer. Since P(r) is mostly affected at distances >80 Å, w hile most of distribution is not influenced, piperine is likely to have a somewhat local rather than global effect. Since data towards higher scattering angles appear too noisy to distinguish Kratky plots in the presence and absence of piperine (not shown), it is not possible to conclude if the appearance of larger interatomic distances is a consequence of the increase in chain flexibility due to complex destabilization, or the change in the complex conformation (e.g. involving a large scale domain movement). These two effects are not mutually exclusive. If piperine directly interferes with the putative RLC/RLC interface in the (RLC/HCF)₂ tetramer, this may lead to the observed destabilization effects and the decrease in the (RLC/HCF)₂ complex compactness once RLC separate from one another.

Establishing the exact mode of binding of piperine to the (RLC/HCF)₂ complex will give valuable information for the next steps in the development of a SRX inhibitor. Solving the structure of the complex with bound piperine would greatly accelerate the search for therapeutic compounds to treat obesity and type 2 diabetes.

5. Conclusions

In conclusion, by studying interactions between piperine, an alkaloid known to destabilize the SRX in fast twitch skeletal muscles, and fragments of skeletal muscle myosin, we identified RLC as a binding site for piperine. Our studies indicated that the mechanism of SRX inhibition by piperine is based on conformational changes in the myosin caused by the ligand binding.

Highlights:

• The SRX is crucial for maintaining the high economy of resting muscle.

• Piperine destabilizes the SRX by a previously unknown mechanism.

• We found that piperine binds to the regulatory light chain (RLC) of myosin.

• Piperine destabilizes and reduces the compactness of the myosin neck.

• This study facilitates the search for therapies for obesity and type 2 diabetes.