The HCM-linked W792R mutation in cardiac myosin-binding protein C reduces C6 FnIII domain stability

Abstract

Cardiac myosin-binding protein C (cMyBP-C) is a functional sarcomeric protein that regulates contractility in response to contractile demand, and many mutations in cMyBP-C lead to hypertrophic cardiomyopathy (HCM). To gain insight into the effects of disease-causing cMyBP-C missense mutations on contractile function, we expressed the pathogenic W792R mutation (substitution of a highly conserved tryptophan residue by an arginine residue at position 792) in mouse cardiomyocytes lacking endogenous cMyBP-C and studied the functional effects using three-dimensional engineered cardiac tissue constructs (mECTs). Based on complete conservation of tryptophan at this location in fibronectin type II (FnIII) domains, we hypothesized that the W792R mutation affects folding of the C6 FnIII domain, destabilizing the mutant protein. Adenoviral transduction of wild-type (WT) and W792R cDNA achieved equivalent mRNA transcript abundance, but not equivalent protein levels, with W792R compared with WT controls. mECTs expressing W792R demonstrated abnormal contractile kinetics compared with WT mECTs that were nearly identical to cMyBP-C-deficient mECTs. We studied whether common pathways of protein degradation were responsible for the rapid degradation of W792R cMyBP-C. Inhibition of both ubiquitin-proteasome and lysosomal degradation pathways failed to increase full-length mutant protein abundance to WT equivalence, suggesting rapid cytosolic degradation. Bacterial expression of WT and W792R protein fragments demonstrated decreased mutant stability with altered thermal denaturation and increased susceptibility to trypsin digestion. These data suggest that the W792R mutation destabilizes the C6 FnIII domain of cMyBP-C, resulting in decreased full-length protein expression. This study highlights the vulnerability of FnIII-like domains to mutations that alter domain stability and further indicates that missense mutations in cMyBP-C can cause disease through a mechanism of haploinsufficiency.

NEW & NOTEWORTHY This study is one of the first to describe a disease mechanism for a missense mutation in cardiac myosin-binding protein C linked to hypertrophic cardiomyopathy. The mutation decreases stability of the fibronectin type III domain and results in substantially reduced mutant protein expression dissonant to transcript abundance.

INTRODUCTION

Hypertrophic cardiomyopathy (HCM) is the most common inherited cardiovascular disease, reported to affect as many as 1 in 200 people worldwide (32). Clinically heterogeneous, HCM is an autosomal dominant disease characterized by progressive myocardial wall thickening in the absence of left ventricular dilation, primarily caused by mutations in one of more than a dozen genes encoding proteins of the cardiac sarcomere (26). While the severity of the disease is highly variable and incompletely penetrant, it is recognized as the leading cause of sudden death in young athletes (19). With >200 identified mutations linked to disease, the MYBPC3 gene is responsible for the largest proportion of all known cases of HCM (2, 14, 24); MYBPC3 encodes cardiac myosin-binding protein C (cMyBP-C), a key regulator of cardiac contractility within the sarcomere. Localized to the C zone of the cardiac sarcomere, cMyBP-C comprises 11 domains (C0–C10) and a largely unstructured region known as the MyBP-C motif (Fig. 1A) (36). This ~100-residue regulatory motif contains the phosphorylation sites believed to impart a majority of cMyBP-C’s function by modulating its interactions with the thick and thin filaments (13). COOH-terminal domains C8–C10 contain the binding sites for titin and light meromyosin, crucial for thick-filament integration, whereas NH₂-terminal domains C0–C2 interact with the S2 region of myosin and the thin filament (21). Of the 11 domains, 8 domains are immunoglobulin-like (IgI) domains, and only C6, C7, and C9 are fibronectin type III (FnIII) domains.

Fig. 1.

Schematic of cardiac myosin-binding protein C (MyBP-C) and cross-species alignment illustrating the highly conserved C6 domain. A: depiction of cMyBP-C. Ig domains are shown as white hexagons and fibronectin type III (FnIII) domains as black circles; areas of sarcomeric interaction are indicated by solid lines. Residue 792 (W792R) in domain C6 is highlighted. B: sequence alignment of C6 domains of cMyBP-C, with completely conserved residues (asterisks), conservative substitutions (colons), and semiconservative substitutions (periods). Amino acids that are invariant from Homo sapiens C6 in positions that are conserved are highlighted in gray. The arrowhead shows tryptophan at position 792. Protein sequences are obtained from the National Center for Biotechnology Information.

The predominant pathomechanism for HCM-linked mutations in MYBPC3 involves haploinsufficiency due to the many nonsense, insertion/deletion, and splice site mutations that lead to frame shifts and truncation of cMyBP-C (34). Mutations that cause a frame shift or truncation often lack COOH-terminal incorporation domains and are targeted for rapid degradation by nonsense-mediated mRNA decay or through the ubiquitin-proteasome system (UPS) (8). Although mutations that result in incomplete protein products have been well characterized, many HCM-causing mutations result in single amino acid substitutions (missense mutations) that are predicted to produce full-length protein products (16). These missense mutations may exert varied and unpredictable effects on cMyBP-C function, depending on their location within the protein (10).

While NH₂- and COOH-terminal domains of cMyBP-C have clear roles in the protein’s regulatory function or incorporation into the sarcomere, the central four domains of cMyBP-C (C3–C6) have no known specific function. Functional importance is suggested, however, since these domains harbor many of the identified HCM-causing missense mutations (1, 27). Specifically, the C6 FnIII domain is highly conserved (Fig. 1B), with the greatest number of identified disease-linked variants. Substitution of a highly conserved tryptophan residue by an arginine residue at position 792 (W792R) in the C6 domain, first identified in 2004, is one of the most severely pathogenic mutations in cMyBP-C (35). Probands carrying the W792R mutation were identified at a young age (mid-20s), and myectomy samples from patients heterozygous for the W792R mutation display disrupted immunohistochemical staining of cMyBP-C compared with normal tissue (33).

Based on homology models of the C6 domain and due to the phenotypic severity, we hypothesized that the W792R mutation alters domain structure and/or stability, modifying the overall function of the protein. To determine the effects of the W792R mutation on contractile function, adenovirus containing the full-length coding sequence for either wild-type (WT) or W792R mutant human cMyBP-C was introduced into neonatal murine cardiomyocytes lacking endogenous expression of cMyBP-C (cMyBP-C^−/−). Transduced cardiomyocytes were used to generate living three-dimensional murine engineered cardiac tissue constructs (mECTs) for functional assessment. Surprisingly, while adenoviral transduction readily achieved physiological levels of mRNA expression, the W792R protein could not be similarly expressed. The contractile phenotype of W792R mECTs mimicked the contractile abnormalities found in cMyBP-C^−/− mECTs. Expression experiments of the domain variants suggest that decreased stability of the mutant domain that may target the protein for rapid degradation before incorporation into the sarcomere.

MATERIALS AND METHODS

Materials.

All experiments and procedures were conducted following the manufacturers’ recommended protocols unless otherwise stated. The following chemicals were obtained from Sigma-Aldrich: NaCl (catalog no. S98888), d-glucose (catalog no. G8270), KCl (catalog no. P9541), NaHCO₃ (catalog no. S5761), potassium phosphate (catalog no. P5655), CaCl₂ (catalog no. C7902), l-glutamic acid potassium salt monohydrate (catalog no. G1149), HEPES (catalog nos. H3375 and H0887), cytosine β-d-arabinofuranoside (catalog no. C1768), bromophenol blue (catalog no. B8026), glycerol (catalog no. G5516), Trizma hydrochloride (Tris·HCl; catalog no. T3253), collagen from the rat tail (catalog no. C7661), neutral buffered formalin (catalog no. HT501128), isopropyl-β-d-1-thiogalactopyranoside (IPTG; catalog no. I6758), digitonin (catalog no. D141), thiazolyl blue tetrazolium bromide (MTT reagent; catalog no. M5655), bafilomycin A₁ (catalog no. B1793), MG-132 (catalog no. 474787), and lactacystin (catalog no. L6785).

The following reagents were obtained from GIBCO: horse serum (catalog no. 16050122), DMEM, high glucose (catalog no. 11965-092), Ham’s F-12 nutrient mix (catalog no. 11765-054), MEM nonessential amino acids solution (catalog no. 11140-050), sodium pyruvate (catalog no. 11360-070), NaHCO₃ (7.5% solution, catalog no. 25080-094), insulin (catalog no. I9278), gentamicin (catalog no. 15750-060), collagenase type II (catalog no. 17101-015), 0.025% trypsin-EDTA (catalog no. R001100), 10× MEM (catalog no. 11430-030), and HBSS (catalog no. 14025092).

Chemicals and reagents purchased from other manufacturers include FBS (catalog no. SH30071.03, HyClone, GE Healthcare), LB broth (catalog no. BP1426-500, Fisher Scientific), TRIzol (catalog no. 15596026, Ambion), Matrigel (catalog no. 354234, Corning), bacterial protein extraction reagent (B-PER, catalog no. 89822, Thermo Fisher), lactacystin (catalog no. SC-3575, Santa Cruz Biotechnology), and SYPRO orange (catalog no. S6650, ThermoFisher).

Sequence analysis.

All sequences analyzed were extracted from the protein database at the National Center for Biotechnology Information (NCBI) on October 4, 2016. The entries listed here are in the format organism name (NCBI reference sequence) and were selected based on their ecological diversity and sequence length similarity. cMyBP-C entries are as follows: Homo sapiens (NP_000247), Pan paniscus (XP_003815230), Camelus bactrianus (XP_010962856), Mus musculus (AAI54409), Canis lupus familiaris (NP_001041571), Felis catus (XP_019667956.1), Rattus norvegicus (XP_006234565), Bubalus bubalis (XP_006057416), Eptesicus fuscus (XP_008145205), Bos taurus (NP_001070004), Balaenoptera acutorostrata scammoni (XP_007181268), Elephantulus edwardii (XP_006896827), and Ailuropoda melanoleuca (XP_002917865.1). Human fibronectin domain sequences, listed as protein/gene name (GenBank/SwissProt accession number), are as follows: FnIII 10 (P02751), cMyBP-C (ABQ59032), MYBPC1 (AAI43503), MYBPC2 (Q14324), MyBP-H (AAB86737), myomesin (AAI16184), and titin (X90568). Initial sequence alignments were produced using Clustal Omega (EMBL-EBI).

Cardiomyocyte isolation and culture.

This study was approved by the Animal Care and Use Committee of the School of Medicine and Public Health of the University of Wisconsin-Madison in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (8th ed.). All animals were either cMyBP-C knockout (KO) or WT E129X1/SvJ generated by Dr. Richard Moss (University of Wisconsin-Madison) (15). Briefly, neonatal mouse hearts were minced on ice, and cells were dissociated using collagenase type II followed by multiple intervals of 0.025% trypsin in KG buffer [containing 127 mM l-glutamic acid potassium salt monohydrate, 0.1335% (wt/vol) NaHCO₃, 16.5 mM d-glucose, 0.42 mM Na₂HPO₄, and 25 mM HEPES]. After centrifugation at 200 g and resuspension in culture medium [60.3% DMEM, 20% Ham’s F-12 medium, 1 mg/ml gentamicin, 8.75% FBS, 6.25% horse serum, 1% HEPES, 1× MEM nonessential amino acid solution, 3 mM sodium pyruvate, 0.00384% (wt/vol) NaHCO₃, and 1 μg/ml insulin], cells were preplated into 100-mm cell culture dishes (BD) and incubated for 45 min at 37°C to enrich for cardiomyocytes. Cardiac cells remaining in suspension were collected and plated into 12-well cell culture dishes (BD) at ~250,000 cells/well or were prepared for subsequent mECT generation.

Protein and mRNA expression.

To select for cardiomyocytes, cytosine β-d-arabinofuranoside (10 μM) was added to the culture medium for 48 h, preventing proliferation of dividing noncardiomyocytes. Cells were harvested for Western blot analysis using chilled RIPA buffer or TRIzol for mRNA quantification. After protein quantification using the standard BCA protein assay protocol (catalog no. 23225, ThermoFisher), 15 μg of protein in 3× sample buffer [240 mM Tris·HCl (pH 6.8), 30% glycerol, 6% SDS, 1.2% β-mercaptoethanol, and 0.06% bromophenol blue] were loaded into each well of a 12-well TGX gel (Bio-Rad). Electrophoresis was performed for 25 min at 300 V. Protein was transferred to a PVDF-LF membrane using the Bio-Rad transfer system at 15 V overnight. After the membrane was blocked for 60 min at room temperature with Li-Cor blocking buffer, it was probed using the following antibodies: polyclonal anti-cMyBP-C generated by the Moss laboratory (1:10,000 dilution) (15) and monoclonal anti-tropomyosin CH1 (0.2 μg/ml, Developmental Studies Hybridoma Bank). The following secondary antibodies were used for both Western blot analysis and immunofluorescent labeling (Life Technologies): goat anti-rabbit 647 (1:10,000 dilution, catalog no. A21244), goat anti-mouse 647 (1:10,000 dilution, catalog no. A21235), and goat anti-mouse 488 (1:10,000 dilution, catalog no. A11001).

For characterization of transcript expression, mRNA was isolated as previously described (10) by TRIzol-chloroform extraction before column purification using the RNeasy kit (Qiagen). cDNA was generated from the isolated RNA using iScript Supermix (Bio-Rad) and quantified using a spectrophotometer (NanoDrop 2000, ThermoFisher), and transcript expression levels were characterized using quantitative real-time PCR (Stratagene Mx3005P quantitative PCR system, Agilent Technologies) and the following gene assays (Applied Biosystems): mouse Gapdh Mm99999915_g1, mouse β-actin 4352663, and human MYBPC3 Hs00165232_m1.

Immunofluorescence imaging of cardiomyocytes.

Neonatal mouse cardiomyocytes were seeded onto Matrigel-coated circular cover glasses (Assistant, 31 mm) in 6-well culture plates and cultured for 7 days, similar to the 12-well protocol described above. At 72 h posttransduction, cMyBP-C^−/− cells expressing human WT or mutant cMyBP-C were formalin fixed for 15 min at 23°C. After 10 min of permeabilization with digitonin (20 μg/ml), cells were blocked in 5% normal goat serum in Tris-buffered saline, incubated overnight at 4°C with primary antibody and then for 60 min with secondary antibody, mounted using ProLong Gold antifade reagent (Invitrogen) with DAPI, and imaged on a Nikon Eclipse 90i photo microscope. Images were captured at 22°C with a S Fluor ×40 oil differential interference contrast H N2 objective with a numerical aperture of 1.3 using NIS Elements version 4.0 (Nikon) and a Nikon DS-Qi1Mc-U3 12-bit camera. The fluorochromes were Alexa Fluor 647-goat anti-rabbit IgG and Alexa Fluor 488-goat anti-mouse IgG (Invitrogen). Line-scan analysis of the separate channels for average sarcomeric morphology was performed using ImageJ (v1.48, National Institutes of Health).

Adenoviral generation and transduction.

Adenovirus expressing human W792R MYBPC3 was generated using site-directed mutagenesis similar to that previously described (10). The previously generated hemagglutinin-tagged WT pENTR/D-TOPO-MYBPC3 clone and the QuikChange II site-directed mutagenesis kit (Agilent Technologies) along with the following primers were used to introduce the W792R mutation: 5′-CTGCACAGTACAGCGGGAGCCGCCTGC-3′ (forward) and 5′-GCAGGCGGCTCCCGCTGTACTGTGCAG-3′ (reverse). After sequence verification of the entire coding region, W792R cMyBP-C inserts were subcloned into the pAd/CMV/V5-DEST adenoviral shuttle vector (Invitrogen) using the Gateway LR Clonase recombination system (Invitrogen). After adenoviral particle generation using human embryonic kidney-293A cells (Invitrogen), viral particles were harvested and titrated using the 50% tissue culture infectious dose assay, as described by VIRAPUR, for the determination of multiplicity of infection (MOI). Quantified adenoviral aliquots were stored at −80°C in single-use volumes to prevent unnecessary exposure to freeze-thaw cycles.

Cells plated in two-dimensional culture were incubated with adenovirus for 24 h, the medium was replaced, and a period of 48 h was allowed to ensure protein expression and incorporation. Transduction of mECTs occurs for 4 h during the rotational culture. Protein expression levels were verified by Western blot analysis or quantitative RT-PCR at the conclusion of the experiments.

Cell viability assay.

Isolated neonatal cMyBP-C^−/− cardiomyocytes were plated into 96-well plates and transduced as described above. After treatment with the inhibitors for the proteasome [MG-132 (10 µM for 48 h) and lactacystin (25 µM for 48 h)] and lysosome [bafilomycin A₁ (25 nM for 48 h)], culture medium was removed and replaced with 0.5 mg/ml MTT reagent in HBSS and incubated at 37°C for 4 h. After incubation, the medium was removed, and the converted dye was solubilized in 0.1 N HCl in isopropanol. Absorbance was measured at 570 nm with background subtraction at 630–690 nm.

mECT generation.

Neonatal mouse cardiac cells suspended in culture medium were rotated at 43 rpm for 4 h at 37°C to allow aggregation of cardiomyocytes. mECTs were then generated as previously described (11) using FlexCell Tissue Train culture plates; 2.4 × 10⁶ cells were used per construct. mECTs were cultured for 7 days, with media changes every other day, before physiological testing.

Contractile twitch recording of mECTs.

To analyze the contractility of mutant cMyBP-C-expressing mECTs, isometric force was measured in a physiological perfusion chamber as previously described (10). Briefly, individual mECTs were attached to a fixed arm and a force transducer (model 403A, Aurora Scientific) within a small intact fiber test apparatus (model 801C, Aurora Scientific) continuously perfused at 1 ml/min with 37°C Krebs-Henseleit buffer [containing (in mM) 119 NaCl, 12 glucose, 4.6 KCl, 25 NaHCO₃, 1.2 KH₂PO₄, 1.2 MgCl₂, and 1.8 CaCl₂, bubbled with 95% O₂-5% CO₂, pH 7.4] and paced at 4 Hz using field stimulation (2.5 ms, 12.5 V). mECTs were stretched to maximum twitch force using the stretching previously described protocol. After a 15-min equilibration period, mECT twitch traces were recorded in 20-s intervals from 4 to 9 Hz, with 1 min of equilibration allowed at each frequency. Recordings were analyzed using IonWizard 6.0 software (IonOptix), with subsequent calculations for twitch kinetics computed in Microsoft Excel. All data are from twitches stimulated at 6 Hz.

Bacterial expression of cMyBP-C domains.

After directional cloning of the expression vector to include the domain sequence, all expression experiments were carried out using the GST Gene Fusion System (GE Healthcare) following the manufacturer’s recommended protocol. The bacterial strains used for transduction and expression were HI-Control 10G and BL21(DE3) competent cells (Lucigen). Initial expression sequences originated from amplified viral plasmid generated above using the following primers: 5′-ATCGAAGGATCCAGGCAGGAACCTCCCAAG-3′ (forward) and 5′-TGCAGTGCGGCCGCTTACTACCGTTGCAGGATCTCCTG-3′ (reverse) for WT and W792R C5-C6-C7. Cells were grown in a shaking incubator (37°C at 225 rpm) to an optical density of 600 nm = 0.5, and protein expression was then induced for 18 h at 27°C with 0.1 µM IPTG. After centrifugation (3,900 g for 20 min at 4°C), bacterial protein was harvested using B-PER at 5 ml/mg. Samples used for gel electrophoresis were complete cell lysates in 3× SDS sample buffer, with protein visualization using Coomassie brilliant blue R-250 (catalog no. 20278, ThermoFisher). GST-tagged C5-C6-C7 domains were purified using the Pierce GST spin purification kit (catalog no. 16107, Thermo Scientific) according to the manufacturer’s protocol.

Top-down proteomic analysis of WT and W792R C5-C6-C7 purified from Escherichia coli.

WT and W792R cMyBP-C C5-C6-C7 were purified from E. coli using the GST tag-based method as described above, and proteins were filtered through 30-kDa molecular weight cutoff filters for seven cycles with H₂O before top-down proteomic analysis. The cleaned-up protein samples were mixed with 1 volume of acetonitrile, and formic acid was added to a final concentration of 0.1%. Samples were stored at −80°C for later experiments. Proteins were delivered to the 12-T Fourier transform-ion cyclotron resonance (FT-ICR) mass spectrometer (solariX, Bruker Daltonics, Bremen, Germany) via nanoelectrospray ionization using a chip-based source (Triversa NanoMate, Advion Bioscience, Ithaca, NY) with voltage and gas pressure adjusted to achieve stable spray at 100–200 nA. The skimmer 1 voltage was set to 70 V, and mass spectra were acquired at a mass range of 1,157–1,227 m/z at a constant sweep excitation power of 21%. The transient length was ~1.15 s, and the accumulation time was adjusted to produce an initial total ion signal of 2–5 × 10⁸. Mass spectra were analyzed by DataAnalysis v3.2 (Bruker Daltonics) and MASH Suite Pro developed in house (7). Truncated W792R was isolated with a 7-m/z window, and in-cell isolation was performed to clear out other coisolated ions using shots excitation with 0.5% power and 0.2-s pulse time. The gas phase purified ions were subjected to electron capture dissociation, and the tandem mass spectrum was acquired at a mass range of 200–3,000 m/z. The tandem mass spectrum was analyzed by MASH Suite Pro.

Fluorescence thermal stability assay of C5-C6-C7.

Bacterially expressed and purified WT and W792R C5-C6-C7 were concentrated to 3 µM in Dulbecco’s PBS using 30-kDa cutoff Amicon Ultra 4-ml centrifugal filters (Millipore). SYPRO orange dye was diluted to 200× in Dulbecco’s PBS, and 5 µl of dye were added to 45 µl of purified protein in a 96-well plate with optical caps. The Stratagene MX3005P quantitative PCR system (see above) measured emission at 568 nm for every 0.5°C after excitation at 492 nm through a melt ramp (25–95°C). Data from four traces (2 samples from 2 preparations) were averaged, and the baseline was corrected.

Tryptic digestion of C5-C6-C7.

The susceptibility of bacterially expressed WT and W792R C5-C6-C7 to digestion by trypsin was determined by incubation of the purified proteins with trypsin. Trypsin-EDTA solution in PBS (catalog no. R001100, GIBCO) was incubated with the recombinant proteins at a 1:500 trypsin-to-protein ratio at 37°C for the desired time. Reactions were stopped by addition of 3× SDS buffer and incubated at 95°C for 3 min. Proteins were separated by gel electrophoresis and visualized with Coomassie brilliant blue R-250.

Predicted C6 structure modeling.

RaptorX (20) was used to model the WT and W792R C6 domain of cMyBP-C (residues 772–869) using the reference sequence NM_000256 obtained from NCBI. Structural files were loaded into PyMOL for analysis and imaging.

Statistical analysis.

SPSS (IBM) was used for mECT contractile parameter assessment, whereas SigmaPlot (Systat) was used for all other statistical analyses. Outliers were defined as being outside 2 SDs from the mean and removed from further analyses. Student’s t-tests were used for all comparisons; statistical significance was set at P < 0.05.

RESULTS

Adenoviral expression of W792R cMyBP-C in neonatal mouse cardiomyocytes.

Cardiac cells were isolated from neonatal cMyBP-C^−/− mice and transduced with adenovirus encoding human WT cMyBP-C (adWT) or human W792R cMyBP-C (ad792). Transduction of adWT at a MOI of 20 on the cMyBP-C-null background expressed cMyBP-C protein at a level equivalent to endogenously expressed protein in normal WT control hearts (11). Transduction of cMyBP-C^−/− cardiac cells with ad792 across a range of MOIs from 1 to 100 resulted in mRNA expression that incrementally increased and, at higher MOIs, exceeded normal physiological levels (Fig. 2A).

Fig. 2.

Expression of mRNA and protein in cardiomyocytes revealed reduced full-length adenovirus encoding human W792R cardiac myosin-binding protein C (ad792 cMyBP-C) protein expression. A: Western blot showing mRNA transcript levels of adenovirus encoding human wild-type (adWT) and ad792 MYBPC3 in neonatal mouse cMyBP-C^−/− cardiomyocytes after adenoviral transduction at a multiplicity of infection (MOI) of 20 for adWT and 5, 10, 20, 50, and 100 for ad792. A representative image of end-point RT-PCR is shown with β-actin (ACTB) as the loading control. B: quantitative RT-PCR with MYBPC3 expression normalized to ACTB and GAPDH. C: Western blot analysis of adWT and ad792 cMyBP-C and α-tropomyosin (α-TM) expression in cMyBP-C^−/− mouse cardiomyocytes after adenoviral transduction at MOI of 1, 5, 10, 20, 50, and 100. D: densitometric quantification of expression levels. cMyBP-C protein levels were normalized to myocyte-specific α-TM. The dashed line reflects endogenous WT cMyBP-C levels. Values are means ± SE; n ≥ 3 for each MOI. *P < 0.05 vs. adWT at the same MOI (by t-test).

Western blot analysis using antibodies specific for the NH₂-terminal region of cMyBP-C showed that transduction with adWT achieved full-length cMyBP-C levels equivalent to normal WT controls at MOIs of ≥20. Transduction with ad792resulted in a very low level of full-length cMyBP-C expression that was significantly less than adWT transduced at an equivalent MOI. Ad792 protein levels were unable to achieve >30% of the adWT protein level (Fig. 2B). The mutant protein expression dramatically underrepresents transcript levels compared with adWT, a disparity that is suggestive of a less stable protein product for ad792 than adWT (Fig. 2B).

Sarcomeric localization of WT and W792R cMyBP-C.

To determine the localization of adenovirally expressed adWT and ad792 cMyBP-C, immunofluorescent labeling using antibodies against cMyBP-C and cardiac α-actinin was performed. Even though equivalent cMyBP-C protein levels were unattainable, a MOI of 20 for each adenovirus was selected, as it resulted in similar MYBPC3 transcript levels (Fig. 2). More than 95% of cardiomyocytes (α-actinin-positive) transduced with adWT or ad792 also expressed cMyBP-C, indicating high transduction efficiency. Colabeling with antibodies specific for the Z-line protein α-actinin and cMyBP-C identified a doublet pattern of cMyBP-C indicative of localization to the C zone of the cardiac sarcomere (Fig. 3). Additionally, the localization pattern of human cMyBP-C in cardiomyocytes transduced with adWT or ad792 was similar to endogenous mouse cMyBP-C in control WT cardiomyocytes, as previously established (30). This localization pattern, highlighted by the averaged fluorescence intensity profile of sarcomeres, suggests that both mutant and WT cMyBP-C incorporated normally into the sarcomere (Fig. 3). Furthermore, there appears to be no aberrant accumulation of the mutant W792R cMyBP-C outside the C zone, as seen by other HCM-linked mutations (22). Although images of cells transduced with adWT or ad792 were acquired under identical settings (including exposure time, aperture, and gain), the intensity of the cMyBP-C immunofluorescence was dramatically lower in cardiomyocytes transduced with ad792 than in those transduced with adWT.

Fig. 3.

Decreased fluorescence intensity in adenovirus encoding human W792R cardiac myosin-binding protein C (ad792 cMyBP-C) cardiomyocytes reflects the lower amount of cMyBP-C. A−H: immunofluorescent imaging of neonatal cMyBP-C^−/− mouse cardiomyocytes transduced with adenovirus encoding human wild-type (WT) cMyBP-C (adWT; A−D) or ad792 (E−H), each at a multiplicity of infection of 20, with identical image acquisition settings. anti-cMyBP-C antibody is shown in green, anti-cardiac α-actinin antibody is shown in red, and nuclei DAPI is shown in blue. D and H are expanded images of the areas highlighted in C and G, revealing that the low level of full-length ad792 localized to the sarcomere. I: averaged fluorescence intensity profiles of sarcomeres from adWT (n = 26 sarcomeres from 5 cardiomyocytes) and ad792 (n = 33 sarcomeres from 5 cardiomyocytes) illustrating similar localization patterns with reduced intensity in mutant (triangles) versus wild-type cMyBP-C (circles) *P < 0.05, ad792 vs. adWT.

Effects of W792R cMyBP-C on contractile function.

To determine the functional effects of W792R cMyBP-C expression on cardiac contractility, twitch force production was measured in mECTs derived from neonatal cMyBP-C^−/− mouse cardiac cells that were either untransduced (KO) or transduced with adWT or ad792 at a MOI of 20. There was no significant genotype-dependent difference in remodeling or gross morphology of mECTs [diameter = 0.93 ± 0.09 mm (adWT) vs. 0.94 ± 0.03 mm (KO) vs. 1.01 ± 0.09 mm (ad792), P = 0.727]. Ad792 and adWT mECTs showed differential twitch force profiles, demonstrated by the average twitch traces, whereas ad792 and KO mECTs produced twitch traces that are highly comparable (Fig. 4, A and B). The twitch force amplitude produced by KO and ad792 mECTs was significantly higher than that produced by adWT mECTs (Fig. 4C). The increase in twitch force amplitude was accompanied by accelerated contractile kinetics in KO and ad792 mECTs, as evidenced by shortened time from electrical stimulation to peak force (CT₁₀₀) and time from peak force to 50% relaxation (RT₅₀) (Fig. 4, D and E). Late relaxation (RT_50–90) was not significantly different between the groups (Fig. 4F). Analysis of the first-order derivatives of the twitch traces (Fig. 5) confirmed that maximum velocities of contraction and relaxation were significantly higher (Fig. 5, B and C) and that maximum velocities were reached earlier in KO and ad792 than adWT mECTs (Fig. 5, D–F). It is striking that there were no significant differences between KO and ad792 mECTs in any of the contractile parameters measured (Figs. 4 and 5), suggesting that the contractile phenotype observed in ad792 mECT is likely due to diminished cMyBP-C, rather than the mutant protein having a neomorphic gain of function.

Fig. 4.

Adenovirus encoding human W792R (ad792) cardiac myosin-binding protein C (cMyBP-C) has increased twitch amplitude and accelerated contraction time, similar to knockout (KO) murine engineered cardiac tissue constructs (mECTs). A: averaged force traces of adenovirus encoding wild-type (adWT; blue), KO (red), and ad792 (orange). B: normalized average twitch traces. C–F: quantification of average twitch force produced, time from electrical stimulation to peak force (CT₁₀₀), time from maximum contraction to 50% relaxation (RT₅₀), and time from 50% relaxation to 90% relaxation (RT_50–90). All data were acquired from pacing at 6 Hz. Values are means ± SE; n = 11 adWT, 13 KO, and 14 ad792. *P < 0.05 (by one-way ANOVA with Tukey’s post hoc test).

Fig. 5.

Contractile kinetics are identical in knockout (KO) and adenovirus encoding human W792R (ad792) cardiac myosin-binding protein C (cMyBP-C) murine engineered cardiac tissue constructs (mECTs). A: averaged amplitude-normalized first-order derivative traces of adWT (blue), KO (red), and ad792 (orange) mECTs. B–F: quantification of normalized maximum contractile velocity (+dF/dt_max; B), normalized maximum relaxation velocity (−dF/dt_max; C), time to +dF/dt_max (D), time from maximum contractile velocity to maximum twitch force (+dF/dt_max to F_max; E), and time from F_max to −dF/dt_max (F). All data were acquired from pacing at 6 Hz. Values are means ± SE; n = 11 adWT, 13 KO, and 14 ad792. *P < 0.05 (by one-way ANOVA with Tukey’s post hoc test).

Degradation of adenovirally expressed cMyBP-C.

The discordance between mutant W792R transcript and protein expression suggests a less stable protein product. Several studies have established the roles of the lysosome and UPS in the degradation of mutant and WT cMyBP-C (4, 31). Bafilomycin A₁ is a specific H⁺-ATPase inhibitor that prevents vacuolar acidification and degradation of proteins through the lysosome (38). Transduced cardiomyocytes were incubated in bafilomycin A₁ (25 nM for 48 h) to determine the role of the lysosome in the degradation of W792R cMyBP-C (Fig. 6A). Quantification of adWT and ad792 cMyBP-C after inhibition of the lysosome showed an increased full-length protein that only achieved significance for inhibited adWT (Fig. 6A), a finding similar to that previously reported by others (31). The inability to achieve equivalent full-length ad792 expression, as well as the similar increases in full-length expression after bafilomycin A₁ incubation (155% for adWT vs. 160% for ad792), suggests that lysosomal autophagy pathways are not responsible for the low level of mutant protein expression. Validation of lysosomal inhibition using the MTT reagent suggested significantly decreased cardiomyocyte viability after bafilomycin A₁ treatment that was not influenced by genotype (data not shown).

Fig. 6.

Neither lysosome nor proteasome inhibition increased adenovirus encoding human W792R (ad792) cardiac myosin-binding protein C (cMyBP-C) expression to wild-type (adWT) equivalence. A: representative Western blot of cMyBP-C^−/− cardiomyocytes transduced with adWT or ad792 at a multiplicity of infection (MOI) of 20 and exposed to the lysosome inhibitor bafilomycin A₁ (Baf; 25 nM for 48 h) using antibodies against cMyBP-C and α-tropomyosin (α-TM) (top) and quantification of the full-length protein expression (bottom). Inhibition of the lysosome did not increase ad792 expression to WT levels. B: representative Western blot of cMyBP-C^−/− cardiomyocytes transduced with adWT or ad792 at a MOI of 20 and exposed to the proteasome inhibitor lactacystin (L; 25 µM for 48 h) or MG-132 (10 µM for 48 h) using antibodies against cMyBP-C and α-TM (top) and densitometric quantification of full-length cMyBP-C with and without exposure to the respective inhibitor (bottom). C: Western blot of cMyBP-C^−/− cardiomyocytes transduced with ad792 at a MOI of 20 or 100 and exposed to the proteasome inhibitor MG-132 (10 µM for 48 h) using antibodies against cMyBP-C and α-TM (top). Abundance of the 50-kDa fragment is increased in the absence of increased full-length expression, quantified by the full length vs. the 50-kDa fragment (bottom). Values are means ± SE; n = 7. *P < 0.05 vs. untreated adWT (by t-test); #P < 0.05 vs. full length (by t-test).

To determine the role of the UPS in the degradation of W792R cMyBP-C, cMyBP-C^−/− cardiomyocytes transduced with adWT or ad792 were exposed to the proteasome inhibitors lactacystin (25 µM for 48 h) and MG-132 (10 µM for 48 h), and verification of functional inhibition was reported as decreased cell viability (data not shown). Immunoblots were performed using antibody generated against the NH₂-terminal region of cMyBP-C, and quantification was normalized to the cardiac-specific α-tropomyosin (Fig. 6B). Inhibition of the UPS increased full-length adWT cMyBP-C but was unable to increase ad792 expression to equivalent levels. Interestingly, we noted a smaller 50-kDa cMyBP-C fragment in ad792 cardiomyocytes after exposure to MG-132 (Fig. 6C). This fragment increased with an increase in MOI, differing from the full-length protein expression. The increase in expression of the smaller fragment coinciding with an increase in MOI suggests efficient translation of mRNA followed by rapid degradation of the mutant protein.

Increased proteolysis of the bacterially expressed mutant C5-C6-C7 fragment.

To quantify stability differences between WT and W792R C6 domains, E. coli were transformed and induced to express a GST-tagged C5-C6-C7 fragment of cMyBP-C. After induction under identical conditions, Coomassie staining highlights the reduced expression of W792R C5-C6-C7 compared with WT C5-C6-C7 (Fig. 7A). The reduced expression in bacteria was consistent with the reduced full-length W792R expression in cardiomyocytes (Fig. 2B). Interestingly, WT and W792R C5-C6-C7 expression products were susceptible to proteolytic cleavage during harvest, degradation that was not preventable by the presence of protease inhibitors. After isolation of the full-length C5-C6-C7 expression products and NH₂-terminal fragments for WT and W792R, mass spectrometry was used to determine the identity of the different protein species present (Fig. 7B). The spectra obtained highlight the decreased expression of full-length W792R C5-C6-C7 compared with WT, with a greater abundance of the cleavage product from the mutant construct. The specific protease cleavage site was identified to follow Y806 in domain C6 (Fig. 8A).

Fig. 7.

Reduced stability with preferential cleavage of bacterially expressed W792R C5-C6-C7 domains. A: representative Coomassie-stained gel of bacterial lysates induced for 4 h to express GST-tagged wild-type (WT) or W792R C5-C6-C7 domains and subsequent elution of the expressed construct from glutathione-Sepharose. B: mass spectrometry data identifying the WT (top) and W792R (bottom) expressed protein products. Highlighted peaks represent the full-length (FL) product with or without the NH₂-terminal methionine (-Met). The most abundant truncation product (TC) is also highlighted. Peak offset is due to mass difference of the substituted amino acid. Experimentally obtained monoisotopic mass of WT GST-C5-C6-C7 = 63,005.14 Da and that of W792R GST-C5-C6-C7 = 62,975.11 Da. C: fluorescence determination of thermal stability of WT (solid line) and W792R (dashed line) C5-C6-C7 proteins. RFU, relative fluorescence units. D: integration of the melting curves identified peaks that indicate unique thermal states of WT versus mutant proteins. E: limited tryptic digestion of the purified expression products highlights the susceptibility of the mutant domain to proteolysis. Coomassie-stained gel depicts a time course of incubation with trypsin at 37°C. F: quantification of the full-length GST-tagged C5-C6-C7 domains over time following incubation with trypsin at 37°C. Values are means ± SE; n = 3. *P < 0.05 (by t-test).

Fig. 8.

Sequence alignment and structural predictions of the C6 domain against other fibronectin type III (FnIII) domains. A: sequence alignment of FnIII domains. Completely conserved residues are marked with asterisks, conservative substitutions with colons, and semiconservative substitutions with periods. Amino acids that are invariant are also highlighted in gray. Tryptophan at position 792 is indicated by an arrowhead. B: 2-dimensional schematic representation of the canonical FnIII fold, where β-strands are represented by block arrows and the three conserved hydrophobic residues are noted. C: predicted three-dimensional structure of the C6 domain, with the hydrophobic side chains of the three 100% conserved amino acids identified. Tryptophan at position 792 is identified by the black arrow. D: predicted three-dimensional structure of the W792R C6 domain. Note the introduction of the positively charged arginine into the hydrophobic core. Protein sequences were acquired from the National Center for Biotechnology Information. RaptorX was used to predict domain structure, with subsequent modeling in PyMOL.

After enrichment of full-length expression products, the thermal stability of WT and W792R C5-C6-C7 was determined using a thermal denaturation assay (Fig. 7C) (12, 23). SYPRO orange was used to determine the thermal profile of the expression products, where increased fluorescence is indicative of greater protein unfolding. The higher initial fluorescence of W792R C5-C6-C7 than WT suggests greater unfolding of the mutant protein product under baseline conditions (Fig. 7C). The first-order derivative of the melting curves highlights the greater number of folding states of WT than W792R C5-C6-C7 (Fig. 7D).

To further characterize stability differences caused by the W792R mutation, purified WT and W792R C5-C6-C7 underwent limited trypsin digestion (Fig. 7E). The decrease in full-length mutant protein after tryptic digestion is accompanied by increasing abundance of smaller fragments. Greater susceptibility to tryptic cleavage in the mutant protein is indicative of a less stable tertiary structure, as quantified in Fig. 7F.

DISCUSSION

We used expression of full-length mutant cDNA in cMyBP-C^−/− cardiomyocytes and expression of a protein fragment in bacteria to characterize the cMyBP-C W792R HCM-causing missense mutation. We showed that the W792R cMyBP-C mutation results in significantly reduced full-length protein expression in cardiomyocytes. The reduced protein level could not be rescued by inhibition of the proteasome or lysosome. Bacterial expression of a C5-C6-C7 cMyBP-C fragment containing the W792R mutation was also reduced. Analysis of a degradation fragment observed by mass spectrometry revealed increased cleavage at a protease site within the C6 FnIII domain that specifically interacts with W792 within the domain’s hydrophobic core. Expression of mutant W792R cMyBP-C in functional mECTs produced a contractile phenotype virtually identical to cMyBP-C^−/− mECTs, with increased twitch force amplitude and accelerated relaxation kinetics. Furthermore, the significantly reduced level of full-length W792R cMyBP-C appears to localize normally within the sarcomere without evidence of accumulation of truncated protein products. These data indicate that the W792R mutation destabilizes the C6 FnIII domain in cMyBP-C and exerts its pathogenicity through the unexpected mechanism of haploinsufficiency.

HCM-causing mutations in cMyBP-C.

While mutations in cMyBP-C are recognized as a leading cause of HCM, understanding mutation-specific mechanisms of pathogenicity, particularly for missense mutations, is still in its infancy (37). Many of the >200 HCM-causing mutations described in cMyBP-C lead to a truncated protein through nonsense mutations, small insertion/deletion mutations, or splice-site mutations that cause frame shifts, leading to premature stop codons (16). Both nonsense-mediated mRNA decay and rapid degradation of truncated protein through the UPS have been shown to be involved in the clearance of truncated cMyBP-C from the myocardium (3, 4). In contrast to truncation mutations, missense mutations are predicted to produce full-length protein. Therefore, the pathomechanisms of missense mutations are likely heterogeneous, with features unique to specific mutations or regions of the protein. How missense mutations in the central domains cause disease, with no ascribed functional roles, remains unknown.

C6 structure and residue 792.

The W792R HCM-causing missense mutation is caused by substitution of a highly conserved hydrophobic tryptophan residue for a hydrophilic, positively charged arginine residue in the C6 FnIII domain. Only 3 of 11 domains comprising cMyBP-C are FnIII domains, and all 3 domains are found in the distal portion of the protein remote from the regulatory region. The FnIII domain family exhibits a tightly conserved tertiary structure; however, the amino acid sequence is often highly variable (Fig. 8) (28). Sequence alignment of FnIII domains from sarcomeric proteins and the canonical 10th FnIII domain shows 100% conservation of only four residues: P775, W792, Y806, and G853 (Fig. 8A). The side chains of three of these four residues are hydrophobic (proline, tryptophan, and tyrosine) and are positioned within the β-sandwich structure of the FnIII domain directed toward the hydrophobic core (Fig. 8C). W792 is predicted to be the final residue of β-strand B, directly preceding the B-C interstrand loop (Fig. 8B). The side chain of this conserved tryptophan is angled toward the central hydrophobic region of the domain. Substitution with a positively charged arginine would appear to be unfavorable and disrupt proper folding of the domain (Fig. 8D). Predictive modeling data indicate that the two aromatic residues (W792 and Y806) are critical to the stability and structure of all FnIII domains (17). Specifically, the homologous counterpart to W792 in the 10th FnIII domain, W22, has the greatest number of intradomain contacts and is the core residue of the “folding nucleus” required for proper tertiary structure formation (9).

Our data suggest that the W792R mutation in domain C6 destabilizes the protein. The W792R mutation changes the thermal stability of the C6 domain, such that it is predominantly unfolded at room temperature, characterized by the elevated baseline of the mutant protein’s thermal profile (Fig. 7C). Furthermore, the normal two-stage melting profile of the WT protein, as identified by the two distinct peaks of the integrated melt curve, is reduced to one in the W792R protein, suggesting that, under basal conditions, the W792R is already unfolded (Fig. 7C). The decreased thermal stability of the W792R mutant is consistent with the heightened susceptibility of the mutant protein to tryptic digestion (Fig. 7D). Both assays corroborate the findings in cardiomyocytes that the W792R mutation in cMyBP-C reduces the stability and abundance of the protein. Interestingly, the degradation of W792R cMyBP-C is not initiated through the proteasome or the lysosome, since inhibition of these two protein degradation pathways did not increase full-length expression, as seen in adWT cardiomyocytes (Fig. 6A). The increasing abundance of a 50-kDa fragment with increasing MOI indicates that protein translation of the mutant mRNA is occurring (Fig. 6B). This finding suggests a mechanism of proteolytic cleavage before complete degradation by the UPS. Proteolytic cleavage before degradation by the proteasome has been observed for similarly unstable proteins (25). Mass spectrometry data and sequence analysis of the C6 domain identified a putative proteolytic cleavage site at residue Y806, the side chain of which directly interacts with W792 within the hydrophobic core. The major fragment we observed when expressing the C5-C6-C7 domain in bacteria would correspond to a ~90-kDa fragment observed when ad792 cardiomyocytes were exposed to MG-132 (Fig. 6B). We postulate that destabilization of the C6 domain by the W792R mutation increases its susceptibility to cleavage by a cytosolic chymotrypsin. The implications of conserved aromatic residues contributing not only to domain stability, but the stability of the entire protein, may provide insights into the mechanisms through which similar mutations in FnIII domain-containing proteins may cause disease.

Mutations in the C5–C6 region.

While the postulated relative flexibility of the C5–C6 region of cMyBP-C may be essential for the functioning of the protein within the sarcomere, it may also render it vulnerable to disease-causing missense mutations that may further destabilize domain folding. A previous study of the N755K disease-causing missense mutation in the C5 IgI domain shows the domain to be almost completely unfolded in solution when the mutation is present (18). The well-characterized R820W mutation in the C6 domain has been associated with both HCM and left ventricular noncompaction in humans and is highly prevalent and severely pathogenic in the Ragdoll cat (29). In silico modeling suggests that perturbations of the β-strand secondary structure resulting from increased hydrophobicity underlie the R820W disease phenotype (5, 6). We speculate that these mutations may act through a mechanism similar to the W792R mutation, where decreased stability leads to a reduction in total protein levels. Our mECT data characterizing the contractile phenotype of the W792R mutation highlight an undetectable difference in contractility and cell survival between the mutant protein and cMyBP-C^−/− cardiomyocytes (Fig. 4). This suggests that pathogenesis of the W792R mutation is not through a “poison polypeptide” mechanism but, instead, causes disease through haploinsufficiency.

Potential limitations and future directions.

This study used adenovirus to exogenously express WT and W792R cMyBP-C on a cMyBP-C-null background. We have previously demonstrated that physiological levels of cMyBP-C can be expressed and incorporated within the sarcomere and that reintroduction of WT cMyBP-C normalizes contractility (11). Furthermore, we have shown that missense mutations in cMyBP-C can be characterized using adenoviral transduction (10). The W792R mutation, however, introduces the issue of protein instability. Although we cannot completely exclude the possibility that the adenovirally expressed protein may be more susceptible to degradation than the endogenously expressed and trafficked protein, the fact that adenoviral expression of WT cMyBP-C at transcript levels similar to endogenous mouse transcript results in equivalent protein expression strongly argues against this. Mouse or human transgenic approaches will provide additional insights into the pathogenesis of mutations in this domain of cMyBP-C. Further characterization of the cryptic cleavage sites exposed in the presence of the W792R mutation may reveal shared disease pathomechanisms involving FnIII domains and suggest potential approaches to stabilize the mutant protein, particularly as the W792R mutant protein does not detectably cause aberrations to the overall contractile kinetics.

Conclusions.

In summary, we show evidence that the cMyBP-C missense mutation W792R produces a full-length protein that is inherently unstable in both cardiac cells and bacteria. The significantly decreased level of cMyBP-C alters cardiac contractility in a fashion similar to complete loss of the protein. The C6 FnIII domain of cMyBP-C may be particularly susceptible to mutations that further destabilize the intrinsically unstable domain. This indicates a potential shared pathomechanism for other diseases involving mutations in FnIII domains. Finally, our data illustrate that missense mutations in cMyBP-C exert their pathological influence through varied mechanisms, emphasizing the need for mutation-specific disease modeling to better understand this heterogeneous condition.

GRANTS

This work was supported by National Institutes of Health Grants R01-HL-107367 (to J. C. Ralphe), R01-HL-109810, R01-HL-096971, and S10-OD-018475 (to Y. Ge), and F31-HL-131230 and T32-HL-07936 (to D. F. Smelter) and by American Heart Association Grant 17PRE33660224 (to W. Cai).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

D.F.S. and J.C.R. conceived and designed research; D.F.S. and W.C. performed experiments; D.F.S. and W.C. analyzed data; D.F.S., W.J.D.L., and J.C.R. interpreted results of experiments; D.F.S. prepared figures; D.F.S. drafted manuscript; D.F.S., W.J.D.L., W.C., Y.G., and J.C.R. approved final version of manuscript; W.J.D.L., W.C., Y.G., and J.C.R. edited and revised manuscript.