# CONNECTING SARCOMERE PROTEIN MUTATIONS TO PATHOGENESIS IN MYOPATHIES

EDITED BY : Jose Renato Pinto and P. Bryant Chase PUBLISHED IN : Frontiers in Physiology

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-921-2 DOI 10.3389/978-2-88963-921-2

### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# CONNECTING SARCOMERE PROTEIN MUTATIONS TO PATHOGENESIS IN MYOPATHIES

Topic Editors: Jose Renato Pinto, Florida State University, United States P. Bryant Chase, Florida State University, United States

Citation: Pinto, J. R., Chase, P. B., eds. (2020). Connecting Sarcomere Protein Mutations to Pathogenesis in Myopathies. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-921-2

# Table of Contents

*06 Molecular and Functional Effects of a Splice Site Mutation in the* MYL2 *Gene Associated With Cardioskeletal Myopathy and Early Cardiac Death in Infants*

Zhiqun Zhou, Wenrui Huang, Jingsheng Liang and Danuta Szczesna-Cordary

*20 Increased Titin Compliance Reduced Length-Dependent Contraction and Slowed Cross-Bridge Kinetics in Skinned Myocardial Strips From*  Rbm20DRRM *Mice*

Hannah C. Pulcastro, Peter O. Awinda, Mei Methawasin, Henk Granzier, Wenji Dong and Bertrand C. W. Tanner

*32 The Role of Leucine-Rich Repeat Containing Protein 10 (LRRC10) in Dilated Cardiomyopathy*

Matthew J. Brody and Youngsook Lee


Janelle Geist and Aikaterini Kontrogianni-Konstantopoulos


Lauren A. Cole, Jonathan H. Dennis and P. Bryant Chase


Sampath K. Gollapudi and Murali Chandra

*117 The Effects of Disease Models of Nuclear Actin Polymerization on the Nucleus*

Leonid A. Serebryannyy, Michaela Yuen, Megan Parilla, Sandra T. Cooper and Primal de Lanerolle

*128 Protein Structure-Function Relationship at Work: Learning From Myopathy Mutations of the Slow Skeletal Muscle Isoform of Troponin T* Anupom Mondal and J.-P. Jin

*139 Heart Failure Induced by Perinatal Ablation of Cardiac Myosin Light Chain Kinase*

Yasmin F. K. Islam, Ryan Joseph, Rajib R. Chowdhury, Robert H. Anderson and Hideko Kasahara

*146 Predicting Effects of Tropomyosin Mutations on Cardiac Muscle Contraction Through Myofilament Modeling*

Lorenzo R. Sewanan, Jeffrey R. Moore, William Lehman and Stuart G. Campbell

*159 Recent Advances in the Molecular Genetics of Familial Hypertrophic Cardiomyopathy in South Asian Descendants*

Jessica Kraker, Shiv Kumar Viswanathan, Ralph Knöll and Sakthivel Sadayappan

*173 The Importance of Intrinsically Disordered Segments of Cardiac Troponin in Modulating Function by Phosphorylation and Disease-Causing Mutations*

Maria Papadaki and Steven B. Marston

*180 Amino Acid Changes at Arginine 204 of Troponin I Result in Increased Calcium Sensitivity of Force Development*

Susan Nguyen, Rylie Siu, Shannamar Dewey, Ziyou Cui and Aldrin V. Gomes


Felix W. Friedrich, Frederik Flenner, Mahtab Nasib, Thomas Eschenhagen and Lucie Carrier

*206 Myofilament Calcium Sensitivity: Mechanistic Insight Into TnI Ser-23/24 and Ser-150 Phosphorylation Integration*

Hussam E. Salhi, Nathan C. Hassel, Jalal K. Siddiqui, Elizabeth A. Brundage, Mark T. Ziolo, Paul M. L. Janssen, Jonathan P. Davis and Brandon J. Biesiadecki

*219 Carbonic Anhydrase III is Expressed in Mouse Skeletal Muscles Independent of Fiber Type-Specific Myofilament Protein Isoforms and Plays a Role in Fatigue Resistance*

Han-Zhong Feng and J.-P. Jin

*235 Myofilament Calcium Sensitivity: Role in Regulation of* In vivo *Cardiac Contraction and Relaxation*

Jae-Hoon Chung, Brandon J. Biesiadecki, Mark T. Ziolo, Jonathan P. Davis and Paul M. L. Janssen

*244 Restrictive Cardiomyopathy Caused by Troponin Mutations: Application of Disease Animal Models in Translational Studies*

Xiaoyan Liu, Lei Zhang, Daniel Pacciulli, Jianquan Zhao, Changlong Nan, Wen Shen, Junjun Quan, Jie Tian and Xupei Huang

*250 Myofilament Calcium Sensitivity: Consequences of the Effective Concentration of Troponin I*

Jalal K. Siddiqui, Svetlana B. Tikunova, Shane D. Walton, Bin Liu, Meredith Meyer, Pieter P. de Tombe, Nathan Neilson, Peter M. Kekenes-Huskey, Hussam E. Salhi, Paul M. L. Janssen, Brandon J. Biesiadecki and Jonathan P. Davis

### *264 Modulating Beta-Cardiac Myosin Function at the Molecular and Tissue Levels*

Wanjian Tang, Cheavar A. Blair, Shane D. Walton, András Málnási-Csizmadia, Kenneth S. Campbell and Christopher M. Yengo


Sabrina Stücker, Nico Kresin, Lucie Carrier and Felix W. Friedrich

*340 Burst-Like Transcription of Mutant and Wildtype* MYH7*-Alleles as Possible Origin of Cell-to-Cell Contractile Imbalance in Hypertrophic Cardiomyopathy*

Judith Montag, Kathrin Kowalski, Mirza Makul, Pia Ernstberger, Ante Radocaj, Julia Beck, Edgar Becker, Snigdha Tripathi, Britta Keyser, Christian Mühlfeld, Kirsten Wissel, Andreas Pich, Jolanda van der Velden, Cristobal G. dos Remedios, Andreas Perrot, Antonio Francino, Francesco Navarro-López, Bernhard Brenner and Theresia Kraft

*355 Classifying Cardiac Actin Mutations Associated With Hypertrophic Cardiomyopathy*

Evan A. Despond and John F. Dawson

*361 Familial Dilated Cardiomyopathy Associated With a Novel Combination of Compound Heterozygous* TNNC1 *Variants*

Maicon Landim-Vieira, Jamie R. Johnston, Weizhen Ji, Emily K. Mis, Joshua Tijerino, Michele Spencer-Manzon, Lauren Jeffries, E. Kevin Hall, David Panisello-Manterola, Mustafa K. Khokha, Engin Deniz, P. Bryant Chase, Saquib A. Lakhani and Jose Renato Pinto

# Molecular and Functional Effects of a Splice Site Mutation in the MYL2 Gene Associated with Cardioskeletal Myopathy and Early Cardiac Death in Infants

### Zhiqun Zhou, Wenrui Huang, Jingsheng Liang and Danuta Szczesna-Cordary \*

*Department of Molecular and Cellular Pharmacology, University of Miami Leonard M. Miller School of Medicine, Miami, FL, USA*

#### Edited by:

*Jose Renato Pinto, Florida State University, USA*

### Reviewed by:

*Theresia Kraft, Hannover Medical School, Germany DeWayne Townsend, University of Minnesota, USA HIdeko Kasahara, University of Florida, USA*

#### \*Correspondence:

*Danuta Szczesna-Cordary dszczesna@med.miami.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *28 March 2016* Accepted: *03 June 2016* Published: *17 June 2016*

#### Citation:

*Zhou Z, Huang W, Liang J and Szczesna-Cordary D (2016) Molecular and Functional Effects of a Splice Site Mutation in the MYL2 Gene Associated with Cardioskeletal Myopathy and Early Cardiac Death in Infants. Front. Physiol. 7:240. doi: 10.3389/fphys.2016.00240* The homozygous appearance of the intronic mutation (IVS6-1) in the *MYL2* gene encoding for myosin ventricular/slow-twitch skeletal regulatory light chain (RLC) was recently linked to the development of slow skeletal muscle fiber type I hypotrophy and early cardiac death. The IVS6-1 (c403-1G>C) mutation resulted from a cryptic splice site in *MYL2* causing a frameshift and replacement of the last 32 codons by 19 different amino acids in the RLC mutant protein. Infants who were IVS6-1+/+-positive died between 4 and 6 months of age due to cardiomyopathy and heart failure. In this report we have investigated the molecular mechanism and functional consequences associated with the IVS6-1 mutation using recombinant human cardiac IVS6-1 and wild-type (WT) RLC proteins. Recombinant proteins were reconstituted into RLC-depleted porcine cardiac muscle preparations and subjected to enzymatic and functional assays. IVS6-1-RLC showed decreased binding to the myosin heavy chain (MHC) compared with WT, and IVS6-1-reconstituted myosin displayed reduced binding to actin in rigor. The IVS6-1 myosin demonstrated a significantly lower Vmax of the actin-activated myosin ATPase activity compared with WT. In stopped-flow experiments, IVS6-1 myosin showed slower kinetics of the ATP induced dissociation of the acto-myosin complex and a significantly reduced slope of the kobs-[MgATP] relationship compared to WT. In skinned porcine cardiac muscles, RLC-depleted and IVS6-1 reconstituted muscle strips displayed a significant decrease in maximal contractile force and a significantly increased Ca2<sup>+</sup> sensitivity, both hallmarks of hypertrophic cardiomyopathy-associated mutations in *MYL2*. Our results showed that the amino-acid changes in IVS6-1 were sufficient to impose significant conformational alterations in the RLC protein and trigger a series of abnormal protein-protein interactions in the cardiac muscle sarcomere. Notably, the mutation disrupted the RLC-MHC interaction and the steady-state and kinetics of the acto-myosin interaction. Specifically, slower myosin cross-bridge turnover rates and slower second-order MgATP binding rates of acto-myosin interactions were observed

**6**

in IVS6-1 vs. WT reconstituted cardiac preparations. Our *in vitro* results suggest that when placed *in vivo*, IVS6-1 may lead to cardiomyopathy and early death of homozygous infants by severely compromising the ability of myosin to develop contractile force and maintain normal systolic and diastolic cardiac function.

Keywords: cardioskeletal myopathy, actin-myosin interaction, fluorescence measurements, myosin ATPase, muscle contraction

### INTRODUCTION

A new skeletal muscle fiber type-I myopathy with progressive cardiomyopathy and the early death of infants due to cardiac failure was reported in three unrelated Dutch families by Barth et al. (1998). It was not until recently that the genetic cause of this cardioskeletal disorder was identified by Weterman et al. (2013), and related to mutations in the MYL2 gene encoding for the ventricular and slow-twitch skeletal myosin regulatory light chain (RLC). To date, about 16 single amino acid mutations in MYL2 have been linked to various forms of cardiomyopathy (Poetter et al., 1996; Flavigny et al., 1998; Andersen et al., 2001, 2009; Kabaeva et al., 2002; Richard et al., 2003; Olivotto et al., 2008; Garcia-Pavia et al., 2011; Claes et al., 2015; Huang et al., 2015). The IVS6-1 (c403-1G>C) mutation, associated with the slow-skeletal and cardiac muscle myopathy, resulted from a cryptic splice site upstream of the last exon of MYL2 causing a frameshift and replacement of the last 32 codons by 19 different codons (Weterman et al., 2013). As a consequence, the C-tail of the RLC protein was truncated and contained a completely altered C-terminal amino acid sequence compared with wild-type (WT) RLC (NCBI accession # P10916) (**Figure 1A**). Immunohistochemical staining of skeletal muscle tissue of the Dutch patients homozygous for IVS6-1 showed a diffuse and weak expression of the mutant protein without clear fiber specificity, while the normal RLC protein was absent (Weterman et al., 2013). Therefore, in this report we aimed at elucidating the potential molecular mechanism by which the IVS6-1 mutation may exert its effects on cardiac muscle contraction. This process is highly dependent upon the integrity of myosin, including its two heavy chains (MHC) and the regulatory and essential (ELC) light chains (Holmes and Geeves, 2000), and such drastic amino acid changes in the sequence of RLC due to IVS6-1 are likely to affect the interaction of myosin with actin, force production and lead to cardiac dysfunction responsible for infantile death of IVS6-1 homozygous patients (Barth et al., 1998; Weterman et al., 2013).

Previous studies from our lab demonstrated that single amino acid mutations in the RLC protein shown to be associated with hypertrophic cardiomyopathy (HCM), were able to cause significant changes in the secondary structure of RLC, as well as in the Ca2<sup>+</sup> binding and phosphorylation properties (Szczesna et al., 2001). They also adversely affected the function of mutated myosin and its ability to interact with actin and produce contractile force (Szczesna-Cordary et al., 2004; Greenberg et al., 2010; Farman et al., 2014; Muthu et al., 2014; Karabina et al., 2015). In this report we have examined, for the first time, the effect of the IVS6-1 mutation on the molecular rearrangements in the RLC and the function of mutant myosin in vitro. The interactions of IVS6-1 with the MHC, and the mutant myosin with actin were investigated using recombinant human cardiac RLC proteins, IVS6-1 vs. WT, that could be reconstituted into porcine cardiac muscle preparations, myosin and skinned muscle strips. Prior to reconstitution, the preparations were stripped of endogenous porcine cardiac RLC. We show that in a way similar to that of other HCM causing mutations in myosin RLC, the IVS6-1 mutation induced significant changes in the RLC structure and the function of mutant myosin compromising its interaction with actin and ultimately leading to dysregulated cardiac muscle contraction.

### MATERIALS AND METHODS

### Cloning, Expression and Purification of Wild-Type (WT) Human Cardiac RLC and the IVS6-1 Mutant

The RLC WT was cloned, expressed and purified as previously described (Szczesna et al., 2001). The cDNA of IVS6-1 was synthesized and inserted in the pCR2.1 vector by Eurofins MWG OperonTM. The plasmid was amplified in Subcloning EfficiencyTM DH5α TM competent cells (Invitrogen) and was transformed into BL21(DE3) competent cells (Agilent Technologies) for expression. Similar to WT, the IVS6-1 mutant protein was purified using an S-Sepharose column followed by a Q-Sepharose column chromatography (GE Healthcare Life Science). The S-Sepharose column was equilibrated with 6 M urea, 20 mM Citrate, 0.1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM dithiothreitol (DTT), 0.02% NaN3, pH 6.0. Proteins were eluted using 800 ml salt gradient of 0−450 mM NaCl. For Q-Sepharose purification the following buffer was used: 3 M urea, 25 mM Tris-HCl, pH 7.5, 0.1 mM PMSF, 1 mM DTT, and 0.02% NaN3, and proteins were eluted with a 1000 ml salt gradient of 0–450 mM KCl. The final purity of the proteins was assessed by 15% SDS-PAGE. The N-terminal peptide RLC antibody (NT-1 RLC, anti-rabbit, aa 7–21) and the C-terminal peptide RLC antibody (CT-1, anti-rabbit, aa 143–156), both produced in this laboratory (Wang et al., 2006), were used to verify the quality and purity of IVS6-1 and WT proteins. Proteins were stored in Q-Sepharose buffer at −80◦C until used in experiments.

**Abbreviations:** CD, Circular Dichroism; DCM, dilated cardiomyopathy; ELC, essential light chain of myosin; HCM, hypertrophic cardiomyopathy; IVS6-1, splice site intronic mutation in MYL2; MHC, myosin heavy chain; MYL2, gene encoding for human ventricular RLC; RLC, regulatory light chain of myosin; Tm, tropomyosin; Tn, troponin; WT, wild-type.

compared with RLC-WT, detected with both, CT-1 and NT-1 antibodies.

### Myosin Light Chain Kinase (MLCK)- Dependent Phosphorylation of IVS6-1 vs. WT RLCs

RLC WT and IVS6-1 proteins were first dialyzed into phosphorylation buffer containing 30 mM KCl, 20 mM PO4, pH 8.0, and adjusted to a concentration of 1.5 mg/ml. Skeletal muscle MLCK was prepared as described previously (Greenberg et al., 2009). Phosphorylation reaction was performed with 5.5µM MLCK, 5µM calmodulin (CaM), 0.1 mM CaCl2, 12.5 mM MgCl2, and 5 mM ATP for 30 min at room temperature. The reaction was terminated by adding 8 M urea. The degree of phosphorylation was assessed by 8% PAGE. The samples were prepared by mixing 100µl of protein with 70 mg of ultrapure urea, 10µl β-ME (β-mercaptoethanol), and 5µl of Bromophenol Blue. 8 M urea–containing gels were run at 100 Volts for 180 min. Gels were scanned and analyzed using Image J software.

### Preparation of Porcine Cardiac (PC) Myosin

Left ventricular (LV) muscles of pig hearts obtained postmortem from a slaughterhouse, chilled on ice and washed clear of blood with ice-cold H2O were isolated and minced. The muscle mince was rinsed with ice-cold H2O until clear followed by extraction of myosin using Edsall–Weber solution (0.012 M Na2CO3, 0.04 M NaHCO3, and 0.6 M KCl, pH 9.0; 300 ml/100 g of muscle) on ice with stirring for 1.5 h, as described earlier in Pant et al. (2009). The homogenate was then centrifuged at 13,000 g for 20 min, and the supernatant was precipitated with 13 vol of ice-cold water containing 1 mM EDTA (ethylenediaminetetraacetic acid) and 1 mM DTT, followed by centrifugation at 13,000 g for 10 min. The pellet was resuspended in buffer containing 0.5 M KCl, 20 mM MOPS (pH 7.0), 1 mM DTT, and 10 mM MgATP and centrifuged at 186,000 g for 1.5 h. Supernatant containing native PC myosin was precipitated with 14 vol of ice-cold H2O containing 1 mM DTT and centrifuged at 8000 g for 10 min. The pellet (kept on ice overnight) was resuspended in 0.5 M KCl, 20 mM MOPS (pH 7.0), 1 mM DTT, and 10 mM MgATP and centrifuged at 186,000 g for 1.5 h. The supernatant containing PC myosin was tested for purity by SDS-PAGE, mixed with glycerol (1:1 vol/vol), and stored at −20◦C until needed.

### Depletion of Endogenous RLC from PC Myosin and Reconstitution of Myosin with RLC WT and IVS6-1

About 1.5 mg/ml of PC myosin dissolved in 0.5 M KCl and 10 mM Potassium Phosphate (pH 8.5) was incubated in buffer containing 1% Triton X-100 and 5 mM CDTA (1,2-cyclohexylenedinitrilotetraacetic) for 30 min at room temperature to extract endogenous PC-RLC. Then the mixture was precipitated with 13 vol of ice-cold water containing 1 mM DTT for 30 min on ice and centrifuged at 8000 g for 10 min. The pellet containing RLC-depleted PC myosin was re-suspended in reconstitution buffer (0.4 M KCl, 50 mM MOPS, pH 7.0, 2 mM MgCl2, and 1 mM DTT) to ∼2.8µM concentration and titrated with increasing concentrations of human cardiac WT or IVS6-1 RLCs (from 0.1 to 14µM). The molar ratio of RLC to depleted myosin ranged from 0.1 to 5.0. Titrations were performed in the presence of BSA to prevent nonspecific RLC binding. The mixtures were incubated for 30 min at room temperature and then precipitated with 13 vol of ice-cold water containing 1 mM DTT for 30 min on ice, and centrifuged at 8000 g for 10 min at 4◦C. The pellets containing WT or IVS6-1 -reconstituted myosins were dissolved in small volumes of 3 M KCl to reach a final concentration of 0.5 M KCl, and clarified by ultracentrifugation at 200,000 g for 45 min at 4◦C. Resulting samples were examined by SDS-PAGE. Gel bands were scanned and quantified using Image J software. The degree of reconstitution was calculated based upon the RLC/ELC band intensity ratio of native, RLC-depleted and WT/IVS6- 1-reconstituted PC myosin with ELC bands used as loading controls (Pant et al., 2009). To account for the amount of porcine RLC remaining in RLC-depleted myosin, the RLC/ELC ratio of RLC-depleted myosin was subtracted from the RLC/ELC ratio of WT-reconstituted myosin due to the similar migration pattern of the endogenous porcine RLC and exogenous human RLC. For IVS6-1, the band of IVS6-1 migrates independently of porcine RLC because of different molecular weights of the two proteins, and as such, could be assessed directly. The resultant RLC/ELC ratio in WT/IVS6-1 reconstituted myosin was then divided by the RLC/ELC ratio measured in native PC myosin. The binding isotherms were fitted to the ligand binding equation:

$$f = \wp\_0 + \text{ Bmax} \ast \text{ x/(Kd + x)}\tag{1}$$

where "Bmax" depicts maximal RLC binding and K<sup>d</sup> is apparent dissociation constant.

Preparation of WT or IVS6-1 reconstituted myosins for the in vitro steady state and kinetics experiments—The RLC-depleted porcine myosin, obtained as described above, was incubated with a 3 molar excess of human recombinant WT/IVS6-1 RLCs and the mixtures were dialyzed for 2 h at 4◦C against the reconstitution buffer containing 0.4 M KCl, 50 mM MOPS, pH 7.0, 2 mM MgCl2, and 1 mM DTT. The protein complexes were centrifuged at 8000 g for 10 min and then dialyzed overnight at 4◦C against 5 mM DTT to precipitate the RLC-reconstituted myosin. The samples were then centrifuged at 8000 g for 10 min to collect the reconstituted myosins and the pellets resuspended in 0.4 M KCl, 10 mM MOPS, pH 7.0, 1 mM DTT. This procedure yielded fully reconstituted IVS6-1- and WT-myosins.

### Preparation of Actin and Labeling with Pyrene Iodoacetamide

Rabbit skeletal acetone powder was extracted with a G-actin buffer consisting of 2 mM Tris-HCl (pH 8), 0.2 mM Na2ATP, 0.5 mM β-ME, 0.2 mM CaCl2, and 0.0005% NaN<sup>3</sup> at a ratio of 20 ml g−<sup>1</sup> for 30 min with stirring on ice (Pardee and Spudich, 1982). The extract was clarified by filtration through several layers of cheesecloth, then centrifuged at 11,000 g at 4◦C for 1 h and the tissue pellet was discarded. The supernatant was adjusted to a final concentration of 50 mM KCl, 2 mM MgCl2, and 1 mM Na2ATP (pH 8.0) and the F-actin was allowed to polymerize for 2 h at 4◦C. The KCl concentration was then increased very slowly to a final concentration of 0.6 M and the mixture was stirred slowly on ice for 30 min to remove possible traces of tropomyosin–troponin (Tm–Tn). F-actin pellet was then collected by ultracentrifugation at 160,000 g at 4◦C for 1.5 h. F-actin pellet was re-dissolved in a buffer containing of 10 mM MOPS (pH 7.0) and 40 mM KCl for pyrene labeling. F-actin at a concentration of 20–40µM was incubated at room temperature in the dark, for 16 h, with a 10 molar excess of Pyrene Iodoacetamide (PIA) (Invitrogen/Molecular Probes) in F-actin buffer containing 10 mM MOPS (pH 7.0) and 40 mm KCl as previously described by Cooper et al. (1983), Kazmierczak et al. (2009). The reaction was quenched by adding 1 mM DTT and the preparation was centrifuged at 1000 g for 1 h to clarify the F-actin solution and remove precipitated PIA. F-actin was then dialyzed against 2 mM Tris–HCl, pH 8.0, 0.2 mM CaCl2, 0.2 mM ATP, and 1 mM DTT overnight to depolymerize F-actin and remove excess PIA. G-actin was then polymerized into F-actin overnight at 4◦C by dialysis in 40 mM KCl, 1 mM MgCl2, and 10 mM MOPS, pH 7. Pyrene-labeled F-actin was tested spectroscopically to determine efficiency of labeling using the molar extinction coefficient, e344(pyrene) = 22,000 M−<sup>1</sup> cm−<sup>1</sup> . The usual molar ratio of pyrene/F-actin was ∼0.8 (Kazmierczak et al., 2012).

### Fluorescence Based Actin-Myosin Binding Assays

RLC WT/IVS6-1 reconstituted myosin was added at 0.05µM increments to pyrene labeled F-actin (0.5µM) until reaching ∼2 fold molar excess over the concentration of actin. Fluorescence measurements were carried using a JASCO 6500 Spectrofluorometer. PIA was excited at 340 nm and fluorescence was collected at 407 nm. The titration data were fitted to the following quadratic equation to obtain the binding constant (Kd) and stoichiometry (n):

$$f = m\_1 - m\_2 \left( K\_d + n \ast a + \mathbf{x} - \frac{\mathbf{x}}{\sqrt{(K\_d + n \ast a + \mathbf{x})^2 - 4 \ast n \ast a \ast \mathbf{x}}} \right) / (2 \ast n \ast a) \tag{2}$$

Where m<sup>1</sup> = initial signal, m<sup>2</sup> = maximal amplitude (decrease in fluorescence intensity on myosin binding to pyrene–actin), n = stoichiometry of myosin-actin binding, a = concentration of actin and x = total concentration of added myosin.

### Stopped-Flow Kinetic Measurements

Reconstituted myosin at a concentration of 0.25µM were mixed with 0.25µM pyrene labeled F-actin (stabilized by 0.25µM phalloidin) in rigor buffer containing 0.4 M KCl, 1 mM DTT and 10 mM MOPS, pH 7.0. The complexes were mixed in a 1:1 (vol/vol) ratio with increasing concentrations of MgATP (10–150µM) dissolved in the same buffer in the stopped flow apparatus. The time course of the change in pyrene fluorescence on MgATP-dependent myosin dissociation from actin was monitored. Measurements were performed using a BioLogic (Claix, France) model SFM-20 stopped-flow instrument outfitted with a Berger ball mixer and an FC-8 observation cuvette. The data were collected and digitized using a JASCO 6500 Fluorometer. The estimated dead time was 3.5 ms. The pyrene-F actin was excited at 347 nm and emission was monitored at 404 nm using monochromators set to 20-nm bandwidths. Typically, 8–12 stopped-flow records were averaged and fit to an exponential equation to obtain the rate at a given MgATP concentration. A plot of the observed myosin dissociation rates as a function of [MgATP] was linear and the slope corresponded to the rate constant expressed in M−<sup>1</sup> ∗ s −1 .

### Myosin ATPase Activity Assay

Actin-activated myosin ATPase activity assays were performed in a 120µl reaction volume in a buffer containing 25 mM imidazole, pH 7.0, 4 mM MgCl2, 1 mM EGTA, and 1 mM DTT and the final KCl concentration of 107 mM, as described in Kazmierczak et al. (2012). Briefly, ∼1.9µM myosin dissolved in 0.4 M KCl (in monomeric form) was added to the 96-well microplate containing increasing concentrations of F-actin (in µM): 0.1, 1, 2.5, 5, 7.5, 10, 15, 20, and 25. Protein mixtures were first incubated on ice for 10 min and then for another 10 min at 30◦C. The reactions (run in triplicate) were initiated with the addition of 2.5 mM ATP with mixing in a Jitterbug incubator shaker (Boekel), allowed to proceed for 20 min at 30◦C and then terminated by the addition of 30µl 20% trichloroacetic acid (TCA). Precipitated proteins were cleared by centrifugation at 4000 g for 15 min and the inorganic phosphate was determined using the Fiske Subbarow method (Fiske and Subbarow, 1925). Data were analyzed using the Michaelis– Menten equation yielding the Vmax and K<sup>m</sup> parameters (Trybus, 2000).

### CDTA-Extraction of Endogenous RLC from Skinned Porcine Papillary Muscle Strips and Reconstitution with WT and IVS6-1

Freshly isolated porcine hearts were placed in oxygenated physiological salt solution of 140 mM NaCl, 4 mM KCl, 1.8 mM CaCl2, 1.0 mM MgCl2, 1.8 mM NaH2PO4, 5.5 mM glucose, and 50 mM Hepes buffer, pH 7.4. The papillary muscles of the left ventricles were isolated, dissected into muscle bundles of about 20 mm (length) × 3 mm (diameter), and chemically skinned in a 50% glycerol, 50% pCa 8 buffer (10−<sup>8</sup> M [Ca2+], 1 mM free [Mg2+] (total MgPr -propionate = 3.88 mM), 7 mM EGTA, 2.5 mM [Mg-ATP2−], 20 mM MOPS, pH 7.0, 15 mM creatine phosphate and 15 units/ml of phosphocreatine kinase, ionic strength = 150 mM adjusted with KPr containing 1% Triton X-100 for 24 h at 4◦C (Muthu et al., 2014). Then the strips were transferred to the same solution without Triton X-100 and stored at −20◦C. Depletion of endogenous RLC from porcine cardiac muscle preparations was achieved in strips about 1.4 mm long and 100µm wide, isolated from glycerinated papillary muscle bundles in buffer containing 5 mM CDTA, 40 mM Tris, 50 mM KCl, 1µg/ml pepstatin A, 0.6 mM NaN3, 0.2 mM PMSF, and 1% Triton X-100, pH 8.4 and protease inhibitor cocktail for 65 min at room temparature. After depletion, the fibers were washed in pCa 8 solution, incubated with 40µM RLC-WT or IVS6-1 protein and 2 mM DTT for total 40 min with fresh proteins added after 20 min. Due to a potential partial loss of TnC that can occur during extraction of RLC and to assure fibers' functionality, both the RLC and TnC were added to the fiber for another 20 min of incubation. Reconstituted strips were then washed in pCa 8 solution and subjected to force measurements. Efficiency of depletion and RLC reconstitution was tested by SDS-PAGE.

### The Ca2<sup>+</sup> Dependence of Force Development

Small porcine heart ventricular muscle strips of approximately 1.4 mm in length and 100µm in diameter were attached by tweezer clips to a force transducer (Muthu et al., 2014). The strips were placed in a 1 ml cuvette and freshly skinned in 1% Triton X-100 dissolved in pCa 8 buffer (as mentioned above) for 30 min. They were rinsed 3 times × 5 min in pCa 8 buffer and their length adjusted to remove the slack. This procedure resulted in sarcomere length of ∼2.1µm as judged by the first order optical diffraction pattern as described in Wang et al. (2013a), Muthu et al. (2014). Then the strips were tested for maximal steady state force development in pCa 4 solution (composition is the same as pCa 8 buffer except the [Ca2+] = 10−<sup>4</sup> M). Maximal tension readings at pCa 4 were taken before and after the force-pCa curve, averaged and expressed in kN/m<sup>2</sup> . The cross sectional area of the muscle strip was assumed to be circular. After the initial steady state force was determined, muscle strips were relaxed in pCa 8 buffer and exposed to solutions of increasing Ca2<sup>+</sup> concentrations from pCa 8 to pCa 4. The level of force was measured in each "pCa" solution. Data were analyzed using the Hill equation (Hill et al., 1980):

$$f = \wp\_0 + \left( a \left( 10^{-\chi} \right)^b \right) / \left( \left( 10^{-c} \right)^b + \left( 10^{-\chi} \right)^b \right) \tag{3}$$

where b = n<sup>H</sup> is the Hill coefficient, c = [Ca2+]<sup>50</sup> or pCa50, is the free Ca2<sup>+</sup> concentration which produces 50% of the maximal force. The pCa<sup>50</sup> represents the measure of Ca2<sup>+</sup> sensitivity of force and the n<sup>H</sup> is the measure of myofilament cooperativity.

### Secondary Structure Prediction of WT and IVS6-1 RLCs

The secondary structure prediction was conducted with I-TASSER (online server from Zhanglab, University of Michigan): http://zhanglab.ccmb.med.umich.edu/I-TASSER/ as described earlier (Huang et al., 2015; Yuan et al., 2015). The amino acid sequence of IVS6-1-RLC was compared against template proteins selected from the PDB library of similar structures. The full length protein was assembled from the excised fragments and simulated into the lowest energy model using specific algorithms. The confidence of each predicted model structure was presented as C-score, ranging from −5 to 2. The quality of prediction was proportional to the value of C-score (Zhang, 2008; Roy et al., 2010, 2012). The predicted structures were then modeled using PyMOL molecular visualization system (Huang et al., 2015; Yuan et al., 2015).

### Statistical Analysis

All values are shown as means ±SD (standard deviation) for n (number of independent experiments) ≤ 5 or ±SEM (standard error of the mean) for n ≥ 6. Statistically significant differences between two groups (WT and IVS6-1) were determined using an unpaired Student'st-test (Sigma Plot 11; Systat Software, San Jose, CA), with significance defined as P < 0.05.

## RESULTS

### Molecular Effects of IVS6-1RLC Mutation

The IVS6-1 mutation originates from a frameshift within the MYL2 gene and results in a replacement of the last 32 amino acids by 19 different amino acids, severely altering the Cterminus of the human cardiac RLC protein resulting in a shorter protein sequence (153 aa for IVS6-1 vs. 166 aa for WT) (**Figure 1A**). The purity of the recombinant RLC WT and IVS6-1 proteins was tested by Western blotting with antibodies against the C-terminus of RLC (CT-1) and its N-terminus (NT-1) (**Figure 1B**), both produced in this laboratory (Szczesna-Cordary et al., 2005; Wang et al., 2006). Due to the amino acid changes and the C-terminal truncation of IVS6-1, the mutant lost its C-terminal epitope and could only be detected with NT-1 (**Figure 1B**).

To examine the effect of C-terminal truncation mutation on the secondary structure of the RLC, the I-TASSER computing program was used and the RLC-like protein templates extracted from the Protein Data Bank, as previously described (Huang et al., 2015). Structures with high similarity to the structure of RLC were used: PDB ID 4i2yA (chain A, crystal structure of the genetically encoded calcium indicator Rgeco1), PDB ID 3jvtB [chain B, calcium-bound scallop myosin regulatory domain (lever arm) with reconstituted complete light chains], PDB IF 3j04B (chain B, EM structure of the heavy meromyosin subfragment of Chick smooth muscle myosin with regulatory light chain in phosphorylated state), 1prwA (chain A, crystal structure of bovine brain Ca2<sup>+</sup> calmodulin in a compact form), PDB ID 4ik1A (chain A, high-resolution structure of Gcampj at pH 8.5), PDB ID 2mysA (chain A, myosin subfragment 1) and PDB ID 2w4ab (chain B, isometrically contracting insect asynchronous flight muscle). The resulting modeled structures of RLC WT and IVS6-1 structures (Model with lowest C-score in I-TASSER) are presented in **Figures 2A,B**. **Figure 2C** shows the superimposed structures of WT and IVS6-1. The results show that the majority of changes occur in the C-terminal region of the RLC, leaving the structure of the N-terminus and the region linking the two RLC lobes unchanged (**Figure 2**). In addition, I-TASSER modeling data suggested that the RLC phosphorylation site at Ser15 is not affected by structural rearrangements of the Cterminus of RLC (**Figure 2**). We then pursued the investigation of the ability of IVS6-1 to become phosphorylated in vitro with the Ca2+-CaM activated MLCK (**Figure 3**). The slower band migration of IVS6-1 (MW∼17220 Da) vs. WT (MW∼18789 Da) was observed, and this was because of increased pI of the mutant (pI∼5.50) compared to WT (pI∼4.89). Likewise, phosphorylated forms of both WT and IVS6-1 migrated faster than their non-phosphorylated counterparts (Szczesna et al., 2001; **Figure 3**). The results indicated that Ser15 of IVS6-1- RLC could be phosphorylated by Ca2+-CaM MLCK as easily as WT RLC. However, these in vitro solution data may not directly translate to the in situ measures, when IVS6- 1 is incorporated into the myosin lever arm in the thick filaments.

### The Effect of IVS6-1 on RLC-MHC Interaction

To gain insight into the effect of IVS6-1 on the assembly of the RLC into the lever arm domain of MHC, we have studied the binding profiles of the WT and IVS6-1 proteins to the RLC-depleted porcine cardiac myosin. The CDTA/Tritonbased treatment yielded >80% RLC-free myosin and the level of RLC remaining in RLC-depleted myosin was assessed by comparing the ratio of RLC/ELC bands in RLC-depleted to RLC/ELC of native myosin (**Figure 4A**). Titration experiments of RLC-depleted PC myosin (2.8µM) incubated with increasing concentrations of human recombinant WT or IVS6-1 (from 0.1µM to 14µM) (**Figure 4B**) produced the binding isotherms (**Figure 4C**) and the Kd values of binding using Equation (1). We observed a significant reduction in Kd and the maximal level of reconstitution for IVS6-1 (Kd = 4.41 ± 0.79 (SD) µM and 57 ± 2%, n = 3) compared with WT (K<sup>d</sup> = 1.42 ± 0.21 (SD) µM and 77 ± 2%, n = 4) (P < 0.05). These results suggested that the IVS6-1 truncation mutation was sufficient to impose severe conformational changes in the RLC structure that prevented the mutant to stoichiometrically bind to the MHC and structurally support the lever arm of myosin. These altered protein-protein interactions might be due to the mutant-induced changes in the tertiary structure of the RLC (**Figure 2B**) that ultimately trigger pathologic cardiac remodeling in the IVS6-1-mutated myocardium.

### Binding of RLC-IVS6-1 Reconstituted Myosin to Pyrene Labeled F-Actin

Fluorescence steady-state binding of IVS6-1 mutant vs. WT reconstituted PC myosin to pyrene labeled F-actin was investigated under rigor conditions (**Figure 5**). Titration profiles of pyrene-actin with native PC myosin or WT-reconstituted myosin were not different while those of IVS6-1-reconstituted

myosin showed impaired binding to actin. Titration data were fitted to Equation (2) to obtain the apparent dissociation constants (Kd) and stoichiometry n of binding (**Figure 5**). The binding of PC myosin or WT/IVS6-1 reconstituted myosins to actin was strong (in nM range), but the mutant showed a lower binding affinity compared with WT or PC native myosin. The data for IVS6-1 showed: K<sup>d</sup> = 16.7 ± 2.8 (SD) nM; stoichiometry n = 0.59 ± 0.0.01 (n = 3), and for WT: K<sup>d</sup> = 4.6 ± 1.3 (SD) nM; stoichiometry n = 0.51 ± 0.14 actin (n = 3). The binding affinity of native myosin, used as a control, for actin was K<sup>d</sup> = 2.9 ± 0.9 (SD) nM with stoichiometry n = 0.47 ± 0.04 (n = 4). Therefore, IVS6-1 reduced the affinity of myosin for actin by ∼4-fold compared to WT and by ∼6-fold compared with native PC myosin (P < 0.01). There was no statistically significant difference in K<sup>d</sup> between WT and native PC myosin (P = 0.103; **Figure 5**).

### Stopped Flow Measurements

is higher than that of WT (predicted pI∼4.89).

Fluorescence stopped-flow kinetic experiments were carried out on myosin reconstituted with RLC-WT or –IVS6-1 and pyrene labeled F-actin to further examine the effects of IVS6-1 on the interaction of myosin with actin. The time course of the recovery in the pyrene fluorescence was monitored as a function of Mg-ATP concentrations. The Mg-ATP-dependent transition of the strongly bound acto-myosin complex (M•A) to the weakly bound state (M•A•ATP) was measured by mixing actin-myosin complexes in a 1:1 vol/vol ratio with increasing concentrations of Mg-ATP (10–150µM) in a stopped flow apparatus. An increase in the fluorescence intensity on the addition of MgATP was monitored as a function of time (not shown) as the myosin heads dissociated from pyrene-F-actin on the addition of MgATP. The observed actin-myosin dissociation rate constant (k1) for the MA

in reconstitution; lane 5, WT-RLC used in reconstitution; lanes 6 and 11, IVS6-1 reconstituted PC myosin; lanes 7 and 10, WT reconstituted myosin; lane 12, mixture

*(Continued)*

#### FIGURE 4 | Continued

of WT and IVS6-1 proteins. (B) Titration experiments using RLC-depleted porcine cardiac myosin with increasing concentrations of WT or IVS6-1 RLCs. ELC (which remained intact during the depletion/reconstitution procedure) was used as the loading control. Numbers on the top indicate the molar ratio of RLC protein used for reconstitution to RLC-depleted PC myosin. (C) Binding isotherms of WT or IVS6-1 to RLC-depleted myosin. The data points were average of *n* = 4 experiments ± SD for WT, and *n* = 3 for IVS6-1. The data were fitted to the ligand- binding model Equation (1). Compared to WT, the maximal level of RLC reconstitution was significantly decreased in IVS6-1 and a significant decrease in the binding affinity to the MHC was observed for IVS6-1, \**P* < 0.05.

to MAATP transition was derived from the averaged fluorescence traces and fitted with a single exponential dependence. The values of k<sup>1</sup> ± SD for each MgATP concentration are presented in **Table 1**. The results revealed significant differences in k<sup>1</sup> between IVS6-1- and WT-reconstituted myosins for 80, 125, and 150µM Mg-ATP concentration, indicating slower dissociation rates in IVS6-1 compared with WT. A plot of the observed transition rates (k1) as a function of [MgATP] is presented in **Figure 6**, which showed linear-type of dependence with the slope "a" value corresponding to the effective second-order Mg-ATP binding rates. Significantly altered binding rates were observed for IVS6- 1 compared with WT/native myosins (n = 3–5 experiments per group) with binding rates (in M−<sup>1</sup> s −1 ): 4.3 ± 0.01 × 10<sup>5</sup> (IVS6-1) vs. 5.8 ± 0.02 × 10<sup>5</sup> (WT) vs. 5.5 ± 0.02 × 10<sup>5</sup> (native) (P < 0.01). No statistically significant differences were observed between native myosin and WT-reconstituted myosin.

### Actin Activated Myosin ATPase Activity

To identify the IVS6-1-induced changes in actin-myosin interaction, actin-activated myosin ATPase activity assays were carried out as a function of actin concentration (in µM) using porcine myosin reconstituted with WT or IVS6-1 RLCs. The acto-myosin ATPase profiles of IVS6-1 mutant, WT and native myosin control are shown in **Figure 7**. The data are the average of n = 3–4 individual experiments ±SD and analyzed using the Michaelis–Menten equation yielding the Vmax and K<sup>m</sup> parameters (Trybus, 2000). The Vmax represents the rate constant of the detachment step and the transition from the weakly (A·M·ATP←→A·M·ADP·Pi) to strongly (A·M·ADP←→A·M) bound cross-bridges (Kazmierczak et al., 2012). The IVS6-1 demonstrated significantly decreased Vmax = 0.15 ± 0.01 s−<sup>1</sup> (n = 3) compared with WT: Vmax = 0.25 ± 0.01 s−<sup>1</sup> (n = 3) and native myosin: Vmax = 0.24 ± 0.01 (n = 4) s−<sup>1</sup> (**Figure 7**, P < 0.05). No statistically significant differences were observed between native PC and WT-reconstituted myosins. The results suggested that IVS6-1 mutation may slow down the ATPase cycle or may decrease the number of cycling cross-bridges during muscle contraction. The K<sup>m</sup> (in µM) ±SD values were 1.93 ± 0.06, 3.67 ± 0.23, and 3.33 ± 0.57 for native PC, WT-, and IVS6- 1 reconstituted myosins. No statistical significance was noted between WT and IVS6-1 (P > 0.05).

### Steady-State Force and Force-pCa Relationship in WT- and IVS6-1- Reconstituted Porcine Papillary Muscle Strip

To further assess the effects of IVS6-1 on cardiac muscle contraction, the RLC WT or IVS6-1 -reconstituted skinned porcine papillary muscle strips were subjected to force-pCa measurements. A significant decrease in maximal isometric force was observed in fibers reconstituted with IVS6-1 compared with WT (**Figure 8A**). The values of force per cross-sectional area of muscle (in kN/m<sup>2</sup> ± SEM) were: IVS6-1, 27 ± 1.0 (n = 6) vs. WT, 35 ± 2.0 (n = 8). The average diameter of muscle strips (inµm) was 89 ± 5 for IVS6-1 and 101 ± 5 for WT. The data of force-pCa measurements were plotted and fitted using the Hill equation (Equation 3). There was a statistically significant increase in pCa<sup>50</sup> of the force-pCa dependence: pCa<sup>50</sup> = 5.66 ± 0.01 observed for IVS6-1 compared with 5.48 ± 0.01 for WT (**Figure 8B**, P < 0.05). The IVS6-1 mutation also affected the Hill coefficient, and 2.90 ± 0.15 was observed for IVS6-1 and 2.20 ± 0.14 for WT (**Figure 8B**). The efficiency of RLC-depletion and reconstitution with RLC/IVS6-1 proteins was tested by SDS-PAGE and is shown in **Figure 8C**. On average, more than 80% of RLC-depletion, and near 100% fiber reconstitution was observed for both WT and IVS6-1 proteins. The results from functional studies indicated that IVS6-1 was able to bind to the lever arm domain of myosin cross-bridge and impose significant alterations in the force-pCa dependence and in the ability of myosin to develop maximal isometric force.

experiments per group were performed.


TABLE 1 | Stopped-flow kinetics of MgATP induced actin-myosin dissociation.

*Pyrene-labeled F-actin was complexed with IVS6-1 or WT –reconstituted myosin. The dissociation rates k*<sup>1</sup> *were in s*−*<sup>1</sup> , n* = *5–10. <sup>a</sup>P* < *0.05 compared with WT.*

### DISCUSSION

Myosin regulatory and essential light chains bind to the myosin heavy chain at the lever arm domain (Rayment et al., 1993b; Geeves, 2002) and structurally support this region of the myosin head contributing to its stiffness and cross-bridge compliance (Muthu et al., 2011; Wang et al., 2013a,b). They also actively participate in the ATP-powered myosin cross-bridge cycle and muscle contraction (Rayment et al., 1993a; Geeves and Holmes, 2005). The cardiac myosin RLC belongs to the superfamily of the EF-hand Ca2+-binding proteins and its N-terminal tail contains one Ca2+-Mg2<sup>+</sup> binding site (Lowey and Risby, 1971; Alexis and Gratzer, 1978). It has been postulated that during cardiac muscle contraction, this RLC site may work as a delayed Ca2<sup>+</sup> buffer helping SERCA2a pump Ca2<sup>+</sup> back to the SR (sarcoplasmic reticulum) during diastole (Wang et al., 2006; Szczesna-Cordary et al., 2007). The N-terminus of cardiac RLC also contains the Ca <sup>2</sup>+/CaM-MLCK dependent phosphorylation site at Ser15, which can become phosphorylated when Ca2<sup>+</sup> is released from the SR and activates the myosin light chain kinase (Kamm and

Stull, 2001). Our previous research showed that the properties of both of these two functional sites (Ca2+-binding site and phosphorylation site) of the RLC can be significantly altered in the presence of cardiomyopathy-associated mutations in MYL2 (Szczesna et al., 2001; Szczesna-Cordary et al., 2004).

In this report we describe a novel MYL2 mutation, which was recently identified to be responsible for a long known cardioskeletal myopathy observed in Dutch and Italian families (Barth et al., 1998; Weterman et al., 2013) The genetic cause for this hereditary disorder resulting in skeletal and cardiac muscle myopathy was found to be due to mutations in the gene encoding for the ventricular and slow twitch skeletal isoforms of the RLC, MYL2 (Weterman et al., 2013). The most severe of all was the splice site mutation located in the intron 6 of MYL2 (IVS6-1) that ultimately resulted in the C-terminal truncation of the RLC protein changing its amino acid sequence at the C-terminal tail of RLC (Weterman et al., 2013). The homozygous appearance of IVS6-1 led to the early death of infants (4–6 months of age) due to dilated (DCM), hypertrophic (HCM) or non-compaction cardiomyopathy, while no obvious phenotype

### reconstitution in porcine papillary muscle fibers. Note that IVS6-1 imposed a significant reduction in maximal level of tension in IVS6-1 reconstituted fibers (*n* = 6) compared with WT reconstituted fibers (*n* = 8) (\**P* < 0.05). The average diameter of muscle strips (inµm) was 89 ± 5 for IVS6-1 and 101 ± 5 for WT. (B) IVS6-1- induced increase in Ca2<sup>+</sup> sensitivity of force. The number of experiments as in A. There was a significant difference between pCa50 of *(Continued)*

#### FIGURE 8 | Continued

IVS6-1 vs. WT reconstituted fibers (\**P* < 0.05). (C) Representative 15% SDS-PAGE of CDTA depleted and RLC/TnC reconstituted porcine papillary muscle strips. Lane 1, native PC fiber; lane 2, recombinant human cardiac WT-RLC protein used for fiber reconstitution; lane 3, recombinant human cardiac IVS6-1-RLC protein used for fiber reconstitution; lanes 4, 5, 7, and 8, IVS6-1 reconstituted fibers; lane 6, RLC-depleted fiber; lanes 9 and 10, WT-reconstituted fibers.

was noted in family members heterozygous for IVS6-1 (Barth et al., 1998; Weterman et al., 2013). The homozygous patients demonstrated dual cardiac and skeletal muscle myopathy with morphological features of muscle type I hypotrophy and the skeletal and cardiac myofibril disorganization.

Here, we have examined the molecular and functional consequences of IVS6-1 in vitro using recombinant IVS6-1 and wild-type RLC proteins that could be reconstituted into RLCdepleted porcine cardiac muscle preparations (PC myosin, actomyosin complex and skinned papillary muscle strips). As the IVS6-1 mutation arose from a frameshift in the MYL2 gene, the resultant protein demonstrated a replacement of the last 32 amino acids by 19 different amino acids. The molecular analysis of the mutation-induced conformational changes in the RLC molecule using I-TASSER computation (Huang et al., 2015; Yuan et al., 2015) clearly showed the C-terminal RLC molecular rearrangements due to IVS6-1. The observed differences between WT RLC and the mutant with the latter being 13-amino acid shorter were observed in the newly formed C-terminus of IVS6-1.

Interestingly, IVS6-1 did not eliminate the ability for MLCKinduced phosphorylation of the RLC in vitro, which was not surprising given that the accessibility of Ser15 was not observed to be obstructed by the C-terminal RLC truncation in IVS6- 1. In vivo, the RLC phosphorylation was shown to be a significant modulatory mechanism of myosin activation and muscle contraction, and a severely decreased level of RLC phosphorylation was observed in the hearts of HCM and/or heart failure patients (Van Der Velden et al., 2003a,b,c) and in the animal models of HCM (Abraham et al., 2009; Muthu et al., 2010, 2012; Yuan et al., 2015). Future studies will have to be executed to examine the effect of IVS6-1 on myosin RLC phosphorylation in vivo.

Since the C-terminal region of the RLC is involved in its interaction with the myosin heavy chain (Rayment et al., 1993b), we proceeded to examine the effect of IVS6-1 truncation mutation on the incorporation of the mutant RLC into the myosin lever arm domain. The data revealed that the mutation was sufficient to disrupt the RLC-MHC interaction and reduce the K<sup>d</sup> of binding. This altered RLC-MHC interaction in the mutant was most likely responsible for the significant changes that we observed in the interaction of the mutant-reconstituted myosin and actin. While no differences in the binding profile to pyrene-labeled F-actin between the native or WT-reconstituted PC myosin were observed, the affinity of IVS6-1 -reconstituted myosin for pyrene-actin was ∼4-fold lower compared with WT. Therefore, the truncation mutation of the RLC and the changes in the amino acid sequence of its new C-terminus resulted in a significantly reduced affinity of IVS6-1-mutant myosin to Factin. Likewise, the stopped-flow kinetics of the myosin–actin interaction were significantly reduced with decreased slope of the kobs-[MgATP] relationship for IVS6-1-reconstituted myosin compared with WT. These results suggest that IVS6-1 mutation may not only lower the affinity of the myosin-actin binding, but also reduces the kinetics of the ATP-induced dissociation of IVS6-1 myosin from actin. Results from actin-activated myosin ATPase activity assays and significantly decreased Vmax (by 1.7 fold compared with WT) are in agreement with stopped-flow data. These in vitro results may explain what has been observed in skinned muscle fiber strips where the calcium sensitivity of tension was significantly increased and the maximal level of tension was significantly reduced in IVS6-1-recnstituted fibers compared with WT. Interestingly, the effects on contractile force and myofilament calcium sensitivity observed in this study were similar to previously reported effects of other HCM-associated mutations in myosin RLC, investigated in transgenic RLC mice (Abraham et al., 2009; Kerrick et al., 2009; Yuan et al., 2015). Our collective results suggest that the IVS6-1 mutation may lead to cardiac dysfunction by disrupting the RLC-MHC and acto-myosin interactions (steady-state and kinetics) ultimately leading to compromised ability of the mutant myosin to develop contractile force and sensitizing myofilaments to calcium, effects that are hallmarks of HCM disease.

### LIMITATIONS, CONCLUDING REMARKS AND FUTURE DIRECTIONS

An important issue which was not addressed experimentally here is the effect of IVS6-1 mutation in heterozygous state with 50:50 ratio of WT and IVS6-1 proteins. Heterozygous patients for IVS6-1± have no cardiomyopathy or slow-twitch skeletal myopathy symptoms; however, no data on protein expression were presented in the heterozygous parents of IVS6- 1 <sup>+</sup>/<sup>+</sup> toddlers (Weterman et al., 2013). Higher than 80% IVS6-1

### REFERENCES


reconstitution, achieved in porcine cardiac muscle preparations, most likely resembles the homozygous state. Our results suggest that when placed in vivo IVS6-1+/<sup>+</sup> may lead to diastolic and systolic dysfunction by delaying muscle relaxation, increasing calcium sensitivity of contraction and reducing maximal force generation. These speculations are supported by measurements of the acto-myosin kinetics and the observation of slower myosin cross-bridge turnover rates and slower second-order MgATP binding rates in IVS6-1 vs. WT reconstituted PC cardiac myosin. Our previous studies of the D166V RLC mutation, located at the last amino acid residue of the human cardiac RLC had shown similar effects on force generation in skinned papillary muscle fibers from the hearts of transgenic mice, i.e., significantly increased the Ca2+-sensitivity of contraction, diminished maximal tension and delayed muscle relaxation (Kerrick et al., 2009). These observations in skinned papillary muscle fibers from D166V mice were further confirmed by echocardiography and invasive hemodynamics showing systolic and diastolic dysfunction in D166 mice (Yuan et al., 2015). Future studies on IVS6-1 animal models are necessary to associate this truncation mutation in the RLC with cardiac dysfunction causing the early death of IVS6-1+/<sup>+</sup> infants.

### AUTHOR CONTRIBUTIONS

ZZ, DSC conceived and designed research, analyzed and interpreted data, prepared figures and drafted the manuscript; ZZ, WH, JL conducted experiments; WH cloned IVS6-1-MYL2 and performed I-TASSER computation. DSC prepared the manuscript for publication and approved the revised version of the manuscript.

### ACKNOWLEDGMENTS

This work was supported by NIH grants R01-HL123255 and HL108343 to DSC, and AHA-15POST25080302 to ZZ.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Zhou, Huang, Liang and Szczesna-Cordary. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Increased Titin Compliance Reduced Length-Dependent Contraction and Slowed Cross-Bridge Kinetics in Skinned Myocardial Strips from Rbm20∆RRM Mice

Hannah C. Pulcastro<sup>1</sup> , Peter O. Awinda<sup>1</sup> , Mei Methawasin<sup>2</sup> , Henk Granzier <sup>2</sup> , Wenji Dong1, 3 and Bertrand C. W. Tanner <sup>1</sup> \*

*<sup>1</sup> Department of Integrative Physiology and Neuroscience, Washington State University, Pullman, WA, USA, <sup>2</sup> Department of Cellular and Molecular Medicine, University of Arizona, Tucson, AZ, USA, <sup>3</sup> Voiland School of Chemical Engineering and Bioengineering, Washington State University, Pullman, WA, USA*

#### Edited by:

*P. Bryant Chase, Florida State University, USA*

#### Reviewed by:

*Marion Lewis Greaser, University of Wisconsin-Madison, USA Norio Fukuda, Jikei University School of Medicine, Japan*

#### \*Correspondence:

*Bertrand C. W. Tanner btanner@vetmed.wsu.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

> Received: *10 May 2016* Accepted: *14 July 2016* Published: *29 July 2016*

#### Citation:

*Pulcastro HC, Awinda PO, Methawasin M, Granzier H, Dong W and Tanner BCW (2016) Increased Titin Compliance Reduced Length-Dependent Contraction and Slowed Cross-Bridge Kinetics in Skinned Myocardial Strips from Rbm20*∆*RRM Mice. Front. Physiol. 7:322. doi: 10.3389/fphys.2016.00322* Titin is a giant protein spanning from the Z-disk to the M-band of the cardiac sarcomere. In the I-band titin acts as a molecular spring, contributing to passive mechanical characteristics of the myocardium throughout a heartbeat. RNA Binding Motif Protein 20 (RBM20) is required for normal titin splicing, and its absence or altered function leads to greater expression of a very large, more compliant N2BA titin isoform in *Rbm20* homozygous mice (*Rbm20*∆*RRM*) compared to wild-type mice (WT) that almost exclusively express the stiffer N2B titin isoform. Prior studies using *Rbm20*∆*RRM* animals have shown that increased titin compliance compromises muscle ultrastructure and attenuates the Frank-Starling relationship. Although previous computational simulations of muscle contraction suggested that increasing compliance of the sarcomere slows the rate of tension development and prolongs cross-bridge attachment, none of the reported effects of *Rbm20*∆*RRM* on myocardial function have been attributed to changes in cross-bridge cycling kinetics. To test the relationship between increased sarcomere compliance and cross-bridge kinetics, we used stochastic length-perturbation analysis in Ca2+-activated, skinned papillary muscle strips from *Rbm20*∆*RRM* and WT mice. We found increasing titin compliance depressed maximal tension, decreased Ca2+-sensitivity of the tension-pCa relationship, and slowed myosin detachment rate in myocardium from *Rbm20*∆*RRM* vs. WT mice. As sarcomere length increased from 1.9 to 2.2 µm, length-dependent activation of contraction was eliminated in the *Rbm20*∆*RRM* myocardium, even though myosin MgADP release rate decreased ∼20% to prolong strong cross-bridge binding at longer sarcomere length. These data suggest that increasing N2BA expression may alter cardiac performance in a length-dependent manner, showing greater deficits in tension production and slower cross-bridge kinetics at longer sarcomere length. This study also supports the idea that passive mechanical characteristics of the myocardium influence ensemble cross-bridge behavior and maintenance of tension generation throughout the sarcomere.

Keywords: cross-bridge kinetics, titin compliance, length-dependent activation, Frank-Starling relationship, cardiac muscle contraction

## INTRODUCTION

Titin is the largest protein that has been identified, spanning from the Z-disk to the M-band of the cardiac sarcomere (LeWinter et al., 2007). Acting as a molecular spring in the I-band, titin contributes to passive tension as sarcomeres are stretched and influences diastolic suction, or elastic recoil at short sarcomere lengths (Granzier and Irving, 1995; Helmes et al., 1996; Wu et al., 2000). Titin compliance is primarily dependent upon differential splicing, resulting in isoforms of different lengths (Labeit and Kolmerer, 1995; Freiburg and Gautel, 1996; Wu et al., 2000). RNA Binding Motif Protein 20 (RBM20) suppresses differential titin splicing such that wild-type mice (WT) predominantly express the stiffer N2B titin isoform and homozygous Rbm20∆RRM mice express a very large, more compliant N2BA titin isoform (Guo et al., 2013; Li et al., 2013; Methawasin et al., 2014).

Actin-myosin cross-bridge behavior is regulated by intracellular [Ca2+] and sarcomere length, both of which are constantly changing throughout the heartbeat (for reviews see Tobacman, 1996; Cooke, 1997; Gordon et al., 2000; Kobirumaki-Shimozawa et al., 2014). Previous studies have shown that increased N2BA expression reduces passive tension (Fukuda et al., 2003; Makarenko et al., 2004; Nagueh et al., 2004; Hanft et al., 2014) which can compromise maximal Ca2+-activated tension production and reduce Ca2+-sensitivity of the tension-pCa relationship (Fukuda et al., 2001, 2003; Hanft et al., 2014; Methawasin et al., 2014). Increased myocardial compliance in Rbm20∆RRM mice and rats also demonstrated an attenuated Frank-Starling response (Methawasin et al., 2014; Ait-Mou et al., 2016). We have recently shown that cross-bridge cycling kinetics slowed at longer sarcomere length due to slowing of MgATP binding and MgADP release (Tanner et al., 2015). This led to the hypothesis that increased sarcomeric compliance in Rbm20∆RRM hearts could affect cross-bridge cycling kinetics differently at short vs. long sarcomere lengths, which may provide an explanation for compromised myocardial function in Rbm20∆RRM vs. WT myocardium.

To test this hypothesis we measured tension-pCa relationships, and cross-bridge kinetics at 1.9 and 2.2 µm sarcomere length in skinned papillary muscle strips from WT and Rbm20∆RRM mice. We found increased titin compliance in the Rbm20∆RRM strips resulted in decreased maximal tension, depressed Ca2+-sensitivity of the tension-pCa relationship, and slowed MgADP release compared to WT strips at each sarcomere length. As sarcomere length increased from 1.9 to 2.2 µm sarcomere length, Rbm20∆RRM strips showed a minimal increase in maximal tension Ca2+-sensitivity of the tension-pCa relationship, while WT strips demonstrated a robust increase in tension and Ca2+-sensitivity of the tension pCa relationship. These findings suggest that titin compliance influences sarcomere-length dependent activation of contraction and cross-bridge nucleotide handling rates, influencing myocardial function more greatly at longer sarcomere length.

## MATERIALS AND METHODS

### Animal Models

All procedures were approved by the Institutional Animal Care and Use Committee at the University of Arizona and followed the U.S. National Institute of Health's "Using Animals in Intramural Research" guidelines for animal use. All mice were adult males, 25–32 weeks old. Wild-type (WT) mice were C57BL/6 strain. As previously characterized, exons 6 and 7 were deleted from the Rbm20 mouse gene to cause an in-frame deletion of the RNA Recognition Motif (RRM) that produced the Rbm20∆RRM genotype (Methawasin et al., 2014).

### Solutions for Skinned Myocardial Strips

Muscle mechanics solution concentrations were formulated by solving equations describing ionic equilibria according to Godt and Lindley (1982), and all concentrations are listed in mM unless otherwise noted. Dissecting solution: 133.5 NaCl, 5 KCl, 1.2 NaH2PO4, 1.2 MgSO4, 30 2,3-butanedione monoxime (=BDM), 10 4-(2-Hydroxyethyl)piperazine-1-ethanesulfonic acid, N-(2-Hydroxyethyl)piperazine-N′ -(2-ethanesulfonic acid; =HEPES; Methawasin et al., 2014). Skinning solution: 40 N,N-Bis(2-hydroxyethyl)-2-aminoethanesulfonic acid, N,N-Bis(2-hydroxyethyl)taurine (=BES), 10 Ethylene glycolbis(2-aminoethylether)-N,N,N′ ,N′ -tetraacetic acid (=EGTA), 6.56 MgCl2, 5.88 ATP, 1 1,4-dithiothreitol (=DTT), 46.35 K propionate, 15 phosphocreatine, 0.4 Leupeptin, 0.1 trans-Epoxysuccinyl-L-leucylamido(4-guanidino)butane (=E-64), 0.5 Phenylmethanesulfonyl fluoride (=PMSF), 1% Triton X-100, pH 7.0 (Methawasin et al., 2014). Storage solution: 50 BES, 30.83 K propionate, 10 Na-azide, 20 EGTA, 6.29 ATP, 1 DTT, 20 BDM, 50 µM Leupeptin, 275 µM Pefabloc, and 1 µM E-64 with 50% glycerol wt/vol. Relaxing solution: pCa 8.0, 5 EGTA, 5 MgATP, 1 Mg2+, 0.3 P<sup>i</sup> , 20 BES, 35 phosphocreatine, 300 U/mL creatine kinase, 200 ionic strength adjusted with Na methanesulfonate, pH 7.0. Adding 0.3 mM P<sup>i</sup> matches estimates for cardiac muscle (Wu et al., 2008; Weiss et al., 2015), though others use higher [Pi] (Wang et al., 2014). Activating solution: Same as relaxing with pCa 4.8. Rigor solution: same as activating solution without MgATP.

### Skinned Myocardial Strips

Left ventricular papillary muscles were dissected from the hearts of four WT mice and four Rbm20∆RRM mice (∼180 µm in diameter and 700 µm long). Muscle strips were skinned in skinning solution overnight at 4◦C, and stored at −20◦C in storage solution for up to 1 week. Aluminum T-clips were attached to the end of each strip and strips were mounted between a piezoelectric motor (P841.40, Physik Instrumente, Auburn, MA) and a strain gauge (AE801, Kronex, Walnut Creek, CA), lowered into a 30 µL droplet of relaxing solution maintained at 17◦C, and stretched to 1.9 or 2.2 µm sarcomere length measured by digital Fourier Transform (IonOptix Corp, Milton, MA).

### Dynamic Mechanical Analysis

Stochastic length perturbations were applied for a period of 60 s as previously described (Tanner et al., 2011, 2015), using an amplitude distribution with a standard deviation of 0.05% muscle lengths over the frequency range 0.5–250 Hz. Elastic and viscous moduli, E(ω) and V(ω), were measured as a function of angular frequency (ω) from the in-phase and out-of-phase portions of the tension response to the stochastic length perturbation. The complex modulus, <sup>Y</sup>(ω), was defined as <sup>E</sup>(ω) <sup>+</sup> iV(ω), where <sup>i</sup> <sup>=</sup> <sup>√</sup> −1. Fitting Equation 1 to the entire frequency range of moduli values provided estimates of six model parameters (A, k, B, 2πb, C, 2πc).

$$Y(\omega) = A(i\omega)^k - B\left(\frac{i\alpha}{2\pi\,b + i\alpha}\right) + C\left(\frac{i\alpha}{2\pi\,c + i\alpha}\right). \tag{1}$$

The A-term in Equation (1) reflects the viscoelastic mechanical response of passive, structural elements in the muscle and holds no enzymatic dependence. The parameter A represents the combined mechanical stress of the fiber, while the parameter k describes the viscoelasticity of these passive elements, where k = 0 represents a purely elastic response and k = 1 is a purely viscous response (Mulieri et al., 2002; Palmer et al., 2013). The B- and C-terms in Equation (1) reflect enzymatic cross-bridge cycling behavior that produce frequency-dependent shifts in the viscoelastic mechanical response during Ca2+ activated contraction. These B- and C-processes characterize work-producing (cross-bridge attachment or recruitment) and work-absorbing (cross-bridge detachment) muscle responses, respectively (Kawai and Halvorson, 1991; Zhao and Kawai, 1993; Campbell et al., 2004; Palmer et al., 2007). The parameters B and C represent the mechanical stress from the cross-bridges (i.e., number of cross-bridges formed × their mean stiffness), and the rate parameters 2πb and 2πc reflect cross-bridge kinetics that are sensitive to biochemical perturbations affecting enzymatic activity, such as [MgATP], [MgADP], or [Pi] (Lymn and Taylor, 1971). Molecular processes contributing to crossbridge attachment or tension generation underlie the crossbridge attachment rate, 2πb. Similarly, processes contributing to cross-bridge detachment or tension decay underlie the crossbridge detachment rate, 2πc.

Stochastic system analysis provides a portrait of cross-bridge kinetics as a function of [MgATP]. Assuming that the myosin attachment events include time spent in the MgADP state and in the rigor state, the cross-bridge detachment rate can be described by:

$$2\pi c = \frac{k\_{-ADP}[MgATP]}{\frac{k\_{-ADP}}{k\_{+ATP}} + [MgATP]}.\tag{2}$$

As explained in detail by Tyska and Warshaw (2002) and implemented in our previous publications (Wang et al., 2013; Tanner et al., 2015), fitting the 2πc-[MgATP] relationship to Equation 2 allows a calculation of (i) k−ADP, which represents cross-bridge MgADP release rate and the asymptotic, maximal myosin detachment rate in s−<sup>1</sup> at saturating [MgATP]; and (ii) k+ATP, which represents the second-order cross-bridge MgATP binding rate per myosin concentration in M−<sup>1</sup> s −1 .

### Statistical Analysis

All values are shown as mean ± SEM. Constrained non-linear least squares fitting of Equations (1, 2) to moduli was performed using sequential quadratic programming methods in Matlab (v 7.9.0, The Mathworks, Natick MA). All statistical tests were performed using SPSS (IBM Statistics, Chicago, IL). A two-way ANOVA was used to assess effects of genotype and sarcomere length for parameter estimates from (i) the 3-parameter Hill fits to the tension-pCa relationships and (ii) the parameter estimates from fits to Equation (2) for the nucleotide handling rates. All other relationships were analyzed using linear mixed models with pCa, frequency, or MgATP as a repeated measure, followed by a least significant difference post-hoc comparison of the means between genotype or sarcomere length. Statistical significance is reported at p < 0.05.

### RESULTS

There were no obvious differences in sarcomere organization or monitored sarcomere length in skinned papillary muscle strips from WT and Rbm20∆RRM mice (**Figure 1A**). As skinned myocardial strips were Ca2+-activated from pCa 8.0 to pCa 4.8, steady-state, isometric tension developed in a sigmoidal manner that was fit to a 3-parameter Hill equation (**Figures 1B–F**, **Table 1**). These tension-pCa relationships are shown two different ways, where: (i) where absolute tension values (=measured force values normalized to cross-sectional area of each myocardial strip; **Figures 1C,D**) illustrate the total tension produced by the strip (i.e., both the passive tension value at pCa 8.0 plus the Ca2+-activated active tension values), and (ii) developed tension values illustrate the Ca2+-activated tension produced by the strip (i.e., absolute tension minus the passive, relaxed tension value at pCa 8.0; **Figures 1E,F**).

Under maximally activated conditions, myocardial strips with both WT and Rbm20∆RRM genotypes displayed greater absolute tension at 2.2 vs. 1.9 µm sarcomere length (**Figures 1C,D**). Relaxed tension values (pCa 8.0) were also greater at the longer sarcomere length in both genotypes (**Table 1**). Developed tension was greater at 2.2 vs. 1.9 µm sarcomere length from pCa 5.8– 4.8 in myocardial strips from WT mice (**Figure 1E**). However, in myocardial strips from the Rbm20∆RRM mice, developed tension was only greater at 2.2 µm sarcomere length at pCa 5.5 and 5.4 (**Figure 1F**). Thus, Ca2+-sensitivity of the tensionpCa relationship increased with sarcomere length in the WT strips (by ∼0.08 pCa units), but this sarcomere length-dependent increase in Ca2+-sensitivity of tension was lost in Rbm20∆RRM strips (**Table 1**). At 2.2 µm sarcomere length, WT strips also displayed greater Ca2+-sensitivity of tension than Rbm20∆RRM strips (by ∼0.07 pCa units). In WT strips, the Hill coefficient (nH) for the tension-pCa relationship was smaller at 2.2 vs. 1.9 µm sarcomere length, indicating reduced cooperativity at longer sarcomere length (**Table 1**). At 2.2 µm sarcomere length, n<sup>H</sup> was smaller for WT strips than Rbm20∆RRM strips; there were no differences in n<sup>H</sup> between genotypes at 1.9 µm sarcomere length.

In both genotypes under relaxed conditions (pCa 8.0), elastic moduli values were greater at 2.2 vs. 1.9 µm sarcomere length

changes throughout the time course of an experiment. Absolute tension-pCa relationships for (C) WT, and (D) *Rbm20*∆*RRM* mice and developed tension-pCa relationships for (E) WT and (F) *Rbm20*∆*RRM* mice at 1.9 and 2.2 µm sarcomere length. Solid lines represent 3-parameter Hill fits to the tension-pCa data, with the dashed lines representing the 1.9 µm sarcomere length fit for *Rbm20*∆*RRM* replotted in panel (D,F). \**p* < 0.05 between sarcomere length within a genotype.

for all frequencies >1.5 Hz (**Figures 2A,C**). Viscous moduli values were also greater at longer sarcomere length at frequencies >51 Hz in WT strips and frequencies >54 Hz in Rbm20∆RRM strips (**Figures 2B,C**, respectively). Under activated conditions (pCa 4.8, 5 mM MgATP), elastic moduli values were greater at longer sarcomere length for frequencies above 145 Hz in WT (**Figure 3A**), and frequencies >22 Hz in Rbm20∆RRM strips (**Figure 3C**). In addition to these moduli differences, there was a consistent shift toward lower frequencies for the overall elastic moduli-frequency relationship at longer sarcomere length; this shift toward lower frequencies was larger for Rbm20∆RRM strips vs. WT strips. Under activated conditions, viscous moduli were not different at any particular sarcomere length in the WT strips (**Figure 3B**), and viscous moduli were greater at 2.2 vs. 1.9 µm sarcomere length at frequencies between 9.5 and 54 Hz in the Rbm20∆RRM strips (**Figure 3D**). There was also a

TABLE 1 | Characteristics of tension-pCa relationships in mouse myocardium at 1.9 and 2.2 µm sarcomere lengths, with and without Rbm20∆RRM mutation (mean <sup>±</sup> SEM).


*Tmin, absolute tension value at pCa 8.0.*

*Tmax , absolute tension value at pCa 4.8.*

*Tdev, Ca2*+*-activated, developed tension (Tmax–Tmin).*

*Maxfit, pCa50, and n<sup>H</sup> represent fit parameters to a 3-parameter Hill equation for the Tdev-pCa relationship: Tdev*(*pCa*) = *Maxfit 1*+*10nH*(*pCa*−*pCa50*) *.*

*† p* < *0.05 effect of mutation at same sarcomere length.*

\**p* < *0.05 effect of sarcomere length within a mutation/genotype.*

consistent shift toward lower frequencies for the overall viscous moduli-frequency relationship at longer sarcomere length; this shift toward lower frequencies was larger for Rbm20∆RRM strips vs. WT strips. Altogether these data indicate greater myocardial viscoelasticity at longer sarcomere length under relaxed and activated conditions, although the influence of titin compliance was minimal as there were no significant effects of genotype in the moduli-frequency relationships (**Figures 2**, **3**). The length-dependent shifts toward lower frequencies in the moduli-frequency relationships at pCa 4.8 indicate slower cross-bridge cycling as sarcomere length increased for both genotypes, although this slowing was greater for Rbm20∆RRM.

Moduli values were fit to Equation (1) to extract model parameters related to viscoelasticity, cross-bridge binding, and cross-bridge kinetics as the skinned strips were titrated toward rigor (5.0–0.05 mM MgATP, pCa 4.8). These model parameters are plotted against [MgATP] in **Figure 4**, with p-values listed in the left panel for each parameter that demonstrated significant main effects or interactions from the mixed-model analysis. As [MgATP] was titrated toward rigor, A values increased and k values decreased for both genotypes, suggesting increased viscoelastic myocardial stiffness that became more elastic (vs. viscous) due to greater cross-bridge binding as MgATP decreased (**Figures 4A–D**). In both genotypes, A values were greater and k values were smaller at 2.2 vs. 1.9 µm sarcomere length, which represents greater myocardial viscoelasticity due to a combination of: (i) passive elements of the sarcomere being stretched or extended more at 2.2 vs. 1.9 µm sarcomere length and (ii) greater binding of slower-cycling cross-bridges at 2.2 vs. 1.9 µm sarcomere length. For both genotypes, the values for B and C increased as [MgATP] was titrated toward rigor and the magnitudes for C increased at 2.2 vs. 1.9 µm sarcomere length (**Figures 4E,F**), also suggesting greater cross-bridge binding at longer sarcomere length.

As [MgATP] decreased, cross-bridge attachment rate (2πb, **Figures 4I,J**) slowed in both genotypes. The significant MgATP × genotype interaction suggests cross-bridge attachment rate was more sensitive to [MgATP] in WT than in Rbm20∆RRM strips, although cross-bridge attachment rates were not different at 2.2 vs. 1.9 µm sarcomere length. Similarly, as [MgATP] decreased toward rigor, cross-bridge detachment rate (2πc, **Figures 4K,L**) slowed in both genotypes. Cross-bridge detachment rates were also slower at 2.2 vs. 1.9 µm sarcomere length for both genotypes. The significant genotype effect on cross-bridge detachment rate suggests that Rbm20∆RRM strips displayed a slower cross-bridge detachment rate than WT strips across the entire [MgATP] range, although this statistic was primarily driven by the slowest detachment rates occurring for Rbm20∆RRM strips at 2.2 µm sarcomere length. Again the significant MgATP × genotype interaction suggests that cross-bridge detachment rate was more sensitive to [MgATP] in WT than in Rbm20∆RRM strips.

Fitting the 2πc-MgATP relationship to Equation (2) (solid lines in **Figures 4K,L**) provides an estimate of the cross-bridge rates of MgADP release (k−ADP) and MgATP binding (k+ATP). The MgADP release rate slowed with increased titin compliance for the Rbm20∆RRM fibers, and there was a length-dependent slowing of k−ADP at longer sarcomere length for both genotypes (**Table 2**). For WT fibers, increasing sarcomere length from 1.9 to 2.2 µm slowed MgADP release by 12% (p = 0.015 using a t-test). For Rbm20∆RRM fibers k−ADP slowed 22% as sarcomere length increased from 1.9 to 2.2 µm, showing about twice as much length-dependent slowing of k−ADP for Rbm20∆RRM than WT. However, increased titin compliance in the Rbm20∆RRM strips led to slower rates of MgADP release at both sarcomere lengths (13 and 23% slower at short and long sarcomere length, respectively), compared to WT k−ADP values. The cross-bridge rate of MgATP binding did not differ with genotype or with sarcomere length (**Table 2**). These finding suggest that increased compliance of the myofilament lattice slows cross-bridge cycling kinetics, primarily due to slower MgADP dissociation from cross-bridges.

### DISCUSSION

Computational simulations of muscle contraction have demonstrated that mechanical characteristics of the sarcomere (i.e., filament, cross-bridge, and titin compliance; compliance = stiffness−<sup>1</sup> ) influence the dynamics of cross-bridge binding and tension generation in a muscle fiber (Daniel et al., 1998; Martyn et al., 2002; Chase et al., 2004; Campbell, 2006, 2009, 2016; Sheikh et al., 2012; Tanner et al., 2012a, 2014). These mathematical models predict that increasing sarcomeric compliance diminishes steady-state tension, slows the apparent rate of tension development, slows cross-bridge cycling rates, and can impact the rate of tension relaxation as well. As RMB20∆RRM mice express more of the compliant N2BA titin isoform than the WT (Guo et al., 2013; Methawasin et al., 2014), these transgenic animals represent a useful model system to test some of these model predictions and directly assess the role of titin compliance in length-dependent tension production and ensemble crossbridge behavior in skinned myocardial strips. In this study we observed that increased titin compliance in Rbm20∆RRM fibers diminished steady-state tension, reduced Ca2+-sensitivity of the

tension-pCa relationship, and slowed cross-bridge detachment rate due to slowed MgADP dissociation from strongly-bound cross-bridges. The effects of titin compliance were sarcomere length-dependent, showing almost no length-dependent tension response in Rbm20∆RRM strips, in contrast to the robust lengthdependent increase in maximal tension and Ca2+-sensitivity of the tension-pCa relationships between 1.9 and 2.2 µm sarcomere length in WT strips. This length-dependent activation response was eliminated in Rbm20∆RRM strips despite a slowed cross-bridge detachment rate as sarcomere length increased, which would be expected to enhance thin-filament activation at 2.2 µm sarcomere length due to strong cross-bridge binding (Bremel and Weber, 1972; Wang and Fuchs, 1994; Metzger, 1995; Fitzsimons and Moss, 1998; Smith et al., 2009; Terui et al., 2010; Li et al., 2014). Empirical findings in this study support previous computational simulations predicting the important role that sarcomeric compliance plays in muscle contraction and further suggests that titin mechanics affect length dependent activation of contraction, perhaps by altering how tension propagates throughout the sarcomere to influence thin-filament activation.

Our observations that increased titin compliance in the Rbm20∆RRM strips reduced maximal tension values and decreased Ca2<sup>+</sup> sensitivity of tension agree with previous findings that suggest greater N2BA titin isoform expression depresses maximum tension production (Makarenko et al., 2004; Lewinter et al., 2010; Patel et al., 2012; Hanft et al., 2014; Methawasin et al., 2014). Our measurements also show that effects of titin compliance on Ca2+-activated tension are sarcomere length-dependent, supporting previous studies showing that increased titin compliance depresses tension more significantly at longer sarcomere length (Fukuda et al., 2003; Methawasin et al., 2014). This is most evident by the similar tension-pCa relationships at 1.9 µm sarcomere length among both genotypes (**Figure 1**; **Table 1**), with a robust

length-dependent increase in Ca2+-activated tension production as sarcomere length increased to 2.2 um for WT strips that did not occur for Rbm20∆RRM strips. These data suggest that length-dependent activation of contraction and the slope of the ascending limb of the sarcomere-length vs. Ca2+-activated tension relationship may depend upon mechanical characteristics of titin. This implies that dynamic processes related to crossbridge cycling kinetics, thin-filament activation, and tension development within the sarcomere may be influenced by the mechanical characteristics of titin.

Cross-bridge detachment rates slowed as sarcomere length increased from 1.9 to 2.2 µm among both genotypes, but the slowing was more pronounced for the Rbm20∆RRM strips. For WT strips, slower myosin detachment at 2.2 µm sarcomere length effectively enhances cross-bridge contributions to thinfilament activation to augment tension production and Ca2+ sensitivity of the tension-pCa relationship. Previous studies have linked greater Ca2+-affinity of troponin C and greater opening of the N-terminus of troponin C with increases in strong cross-bridge binding (Hofmann and Fuchs, 1987; Wang and Fuchs, 1994; Terui et al., 2008, 2010; Smith et al., 2009; Li et al., 2014), and our current findings in WT strips and rat papillary muscle strips (Tanner et al., 2015; Pulcastro et al., 2016) imply this cooperative activation pathway becomes stronger at longer sarcomere lengths. The MgADP release rate (k−ADP) was 13% slower at 1.9 µm and 23% slower at 2.2

#### FIGURE 4 | Continued

indicates an expected increase in cross-bridge binding as [MgATP] was titrated toward rigor. The rate of cross-bridge attachment, 2πb (I,J), and the rate of cross-bridge detachment, 2πc (K,L), decreased as [MgATP] decreased, which indicates the expected slowing of cross-bridge cycling kinetics as [MgATP] was titrated toward rigor. Dashed lines representing the *Rbm20*∆*RRM*, 1.9 µm sarcomere length data were replotted in the left set of panels. *P*-values listed within the left panel show significant (<0.05) main effects of [MgATP], genotype, sarcomere length (SL), and any interactions between these effects among all four sets of data, resulting from mixed models analysis of each parameter-[MgATP] relationship. \**p* < 0.05 between sarcomere lengths within a genotype.

TABLE 2 | Estimates of myosin cross-bridge kinetics from fits of the cross-bridge detachment rate (2πc) vs. MgATP relationships to Equation (2) for 1.9 and 2.2 µm sarcomere lengths (mean ± SEM).


*k*−*ADP, cross-bridge MgADP release rate.*

*k*+*ATP, cross-bridge MgATP binding rate.*

*† p* < *0.05, ‡ p* < *0.1 effect of mutation at same sarcomere lengths.*

\**p* < *0.05 effect of sarcomere length under similar treatment conditions.*

µm sarcomere length in Rbm20∆RRM strips, compared to WT strips, which would be expected to slow cross-bridge detachment and stabilize, or amplify thin-filament activation more greatly in Rbm20∆RRM strips. However, slower cross-bridge detachment rates did not enhance tension nor length-dependent activation of contraction in Rbm20∆RRM strips with greater titin compliance. Thus, cross-bridge contributions to thin-filament activation and increased Ca2+-affinity of troponin C may require titin interacting with the thin-filament or titin transmitting tension between the thick and thin-filament. Increased titin compliance in Rbm20∆RRM strips may compromise this titin interaction or tension transmission pathway, thereby depressing Ca2+ activated tension production and length-dependent activation of contraction.

Some muscle mechanics studies use large amplitude releaserestretch protocols (∼15% muscle length) to assess the crossbridge rate of tension redevelopment (ktr), in comparison to the low amplitude strains used for stochastic length perturbation analysis (<0.15% muscle length). Skinned myocardial strips from WT and Rbm20∆RRM strips mice showed no differences in sarcomere length-dependent ktr under maximally Ca2+-activated conditions (Methawasin et al., 2014). Previous studies using skinned myocardium from rats expressing the more compliant N2BA titin isoform have shown mixed reports of slower and faster ktr values as sarcomere length increased (Patel et al., 2012; Hanft et al., 2014), compared to wild-type controls that predominantly express the stiffer N2B titin isoform. Herein we measured cross-bridge kinetics as [MgATP] varied, which allowed us to estimate cross-bridge rates of MgADP release (k−ADP) or MgATP binding (k+ATP; **Table 2**). As the rate of MgADP release limits cross-bridge detachment in a muscle fiber (Siemankowski et al., 1985), the ∼12% slowing in k−ADP from 1.9 to 2.2 µm sarcomere length drives the length-dependent slowing of cross-bridge detachment in WT strips. However, the length-dependent slowing of k−ADP was nearly twice as great in Rbm20∆RRM strips (∼22%) and k−ADP was also slower at each sarcomere length when titin compliance increased. There was not a significant difference between cross-bridge MgATP binding rates at 1.9 vs. 2.2 µm sarcomere lengths for either genotype. These data support our previous observations that slowed MgADP release rate is the predominate step of the cross-bridge cycle that is responsible for the length-dependent slowing of cross-bridge kinetics (Tanner et al., 2015; Pulcastro et al., 2016). We do not think these slowed nucleotide handling kinetics in Rbm20∆RRM strips stem from any α-to-β myosin heavy chain isoform shift, because Methawasin et al. (2014) reported solely α-myosin heavy chain expression in both of these mouse lines. Moreover, these data also demonstrate that titin compliance influences sarcomere length-dependent cross-bridge nucleotide handling rates, and the effects of titin on cross-bridge kinetics become greater as sarcomere length increases.

Under relaxed conditions, both viscoelastic mechanical stiffness (**Figure 2**) and steady-state tension values (**Table 1**) were greater at longer sarcomere length, without any differences between the two genotypes. These differences stem from passive elements of the sarcomere being stretched or extended more greatly at 2.2 vs. 1.9 µm sarcomere length [i.e., titin and collagen (Granzier and Irving, 1995)]. We had anticipated that lengthdependent increases in relaxed stiffness and tension would be greater for WT vs. Rbm20∆RRM, similar to previous observations using skinned myocytes (Methawasin et al., 2014). However, Methawasin et al. (2014) also showed greater collagen expression in Rbm20∆RRM vs. heterozygous Rbm20 knockout mice, which could be a compensatory mechanism to increase myocardial stiffness as titin compliance decreased in the homozygous mice. Given that skinned myocardial strips encompass some component of passive stiffness due to collagen that isn't present in isolated myocytes, it is possible that the mechanical characteristics of collagen, rather than titin, are dominating our relaxed muscle mechanics measurements. While previous studies suggested that greater passive tension values are correlated with greater Ca2+-activated tension production and length-dependent activation of contraction (Fukuda et al., 2001, 2003), our measurements do not support this mechanism driving lengthdependent activation because relaxed stiffness and tension values were similar at each sarcomere length for both genotypes.

Thick-to-thin-filament spacing consistently decreases as sarcomere length increases in skinned and intact muscle preparations (Matsubara and Millman, 1974; Irving et al., 2000; Konhilas et al., 2002; Smith et al., 2009). Mechanical characteristics of titin influence this lattice spacing vs. sarcomere length relationship, showing that increased titin compliance can increase myofilament lattice spacing and affect the relationship between lattice spacing and sarcomere length (both increasing and decreasing the slope of this relationship; Cazorla et al., 2001; Fukuda et al., 2001, 2003, 2005; Irving et al., 2011). In addition, recent measurements show smaller myofilament lattice spacing values in Rbm20 knockout rat myocardium at both short and long sarcomere length, compared to wild-type controls (Ait-Mou et al., 2016). Cross-bridge cycling rates have been shown to slow as thick-to-thin-filament spacing decreased in vertebrate and invertebrate muscle fibers that were osmotically compressed with Dextran (Krasner and Maughan, 1984; Kawai and Schulman, 1985; Smith et al., 2009; Tanner et al., 2012b) and with increases in sarcomere length in skinned (Adhikari and Wang, 2004; Tanner et al., 2015; Pulcastro et al., 2016), and intact (Milani-Nejad et al., 2013) cardiac muscle preparations. Therefore, increases in sarcomere length will accompany decreases in thick-to-thinfilament spacing, which could contribute to slower cross-bridge detachment at longer sarcomere length for both genotypes.

While reduced lattice spacing may slow cross-bridge cycling, this does not translate into increased length-dependent activation of contraction in Rbm20∆RRM strips (Hanft et al., 2014; Methawasin et al., 2014). Therefore, slowed cross-bridge cycling kinetics may not be the primary mechanism responsible for increasing Ca2+-activated tension at long sarcomere length (Patel et al., 2012), particularly when titin compliance increases from normal. Perhaps, titin interacts with the thin-filament to influence thin-filament activation and length-dependent

### REFERENCES


activation of contraction, either directly or by influencing load (or strain) borne by thin-filament proteins (Terui et al., 2008; Hanft et al., 2014). This titin-thin-filament activation pathway may be suppressed with the more compliant titin in Rbm20∆RRM fibers, because titin is less taut and cannot effectively transmit tension between the M-band and Z-disks to maintain tension throughout the sarcomere. Thus, increases in cross-bridge duty ratio due to slowed detachment kinetics in Rbm20∆RRM fibers do not necessarily translate into the greater tension production due to a compromised capacity to generate tension or distribute tension throughout a more compliant sarcomere. Altogether, this would diminish ventricular function, and may scale with the expression ratio between the more compliant N2BA titin isoform and the stiffer N2B titin isoform. These impaired mechanisms of thin-filament activation and tension production may contribute to cardiac dysfunction and the associated cardiomyopathies in humans, rats, and mice bearing RBM20 mutations that influence titin splicing (Makarenko et al., 2004; Nagueh et al., 2004; Guo et al., 2012; Methawasin et al., 2014).

### AUTHOR CONTRIBUTIONS

HP, PA, MM, and BT participated in performing the experiments and data collection. BT, WD, and HG conceived and designed the experiments. HP, PA, and BT analyzed the data. All authors helped interpret the data, write, and revise the manuscript, and have approved the final version of this manuscript.

### ACKNOWLEDGMENTS

This research was supported by Beginning Grant in Aide 14BGIA20380385 from the Western States Affiliate of the American Heart Association (BT), a New-faculty Seed Grant from the College of Veterinary Medicine (BT), and National Institutes of Health Grants R01HL118524 (HG), R01HL80186 (WD), and R21HL109693 (WD).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Pulcastro, Awinda, Methawasin, Granzier, Dong and Tanner. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Role of Leucine-Rich Repeat Containing Protein 10 (LRRC10) in Dilated Cardiomyopathy

Matthew J. Brody <sup>1</sup> and Youngsook Lee<sup>2</sup> \*

*<sup>1</sup> Department of Pediatrics, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA, <sup>2</sup> Department of Cell and Regenerative Biology, University of Wisconsin-Madison, Madison, WI, USA*

Leucine-rich repeat containing protein 10 (LRRC10) is a cardiomyocyte-specific member of the Leucine-rich repeat containing (LRRC) protein superfamily with critical roles in cardiac function and disease pathogenesis. Recent studies have identified *LRRC10* mutations in human idiopathic dilated cardiomyopathy (DCM) and *Lrrc10* homozygous knockout mice develop DCM, strongly linking LRRC10 to the molecular etiology of DCM. LRRC10 localizes to the dyad region in cardiomyocytes where it can interact with actin and α-actinin at the Z-disc and associate with T-tubule components. Indeed, this region is becoming increasingly recognized as a signaling center in cardiomyocytes, not only for calcium cycling, excitation-contraction coupling, and calcium-sensitive hypertrophic signaling, but also as a nodal signaling hub where the myocyte can sense and respond to mechanical stress. Disruption of a wide range of critical structural and signaling molecules in cardiomyocytes confers susceptibility to cardiomyopathies in addition to the more classically studied mutations in sarcomeric proteins. However, the molecular mechanisms underlying DCM remain unclear. Here, we review what is known about the cardiomyocyte functions of LRRC10, lessons learned about LRRC10 and DCM from the *Lrrc10* knockout mouse model, and discuss ongoing efforts to elucidate molecular mechanisms whereby mutation or absence of LRRC10 mediates cardiac disease.

### Edited by:

*P. Bryant Chase, Florida State University, USA*

### Reviewed by:

*Brenda Schoffstall, Barry University, USA Ranganath Mamidi, Case Western Reserve University, USA Miklos Kellermayer, Semmelweis University, Hungary*

### \*Correspondence:

*Youngsook Lee youngsooklee@wisc.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *07 June 2016* Accepted: *20 July 2016* Published: *03 August 2016*

### Citation:

*Brody MJ and Lee Y (2016) The Role of Leucine-Rich Repeat Containing Protein 10 (LRRC10) in Dilated Cardiomyopathy. Front. Physiol. 7:337. doi: 10.3389/fphys.2016.00337* Keywords: LRRC10, leucine-rich repeat, dilated cardiomyopathy, cardiomyopathy, eccentric hypertrophy

## INTRODUCTION

Dilated cardiomyopathy (DCM) and hypertrophic cardiomyopathy (HCM) are the most common primary myocardial diseases with a prevalence of at least 1 in 2500 and 1 in 500 individuals, respectively, (McNally et al., 2013, 2015; Kimura, 2016). DCM is characterized by eccentric cardiac growth resulting in ventricular dilation and reduced cardiac function without an increase in ventricular wall thickness (Maillet et al., 2013; van Berlo et al., 2013; Bang, 2016; Kimura, 2016). In contrast, in HCM, the heart undergoes concentric growth that results in ventricular wall thickening and reduced ventricular inner diameter, ultimately resulting in reduced cardiac output (Maillet et al., 2013; van Berlo et al., 2013; Bang, 2016). Both HCM and DCM can progress to congestive heart failure and are associated with an increased risk of sudden death (Maillet et al., 2013; van Berlo et al., 2013). While a number of mutations of genes encoding sarcomeric proteins are known to cause HCM, the genetic etiology of DCM is much more heterogeneous, including mutation of genes encoding proteins of the Z-disc, costamere, cytoskeleton, sarcolemma, sarcomere, and nuclear lamina (Cheng et al., 2010; McNally et al., 2013, 2015; Bang, 2016). The most prevalent DCM causing mutations are truncations of the sarcomeric protein titin (Herman et al., 2012; Hinson et al., 2015). DCM can also occur in response to myocardial infarction or ischemic damage, which accounts for about half of all cases, while the majority of remaining cases are idiopathic, underscoring the need to identify more genetic mutations that underlie DCM (McNally et al., 2013). The genetic determinants and molecular etiology underlying DCM in a majority of patients remain unclear.

Leucine-rich repeat containing protein 10 (LRRC10) was identified based on its cardiac-specific expression pattern (Nakane et al., 2004; Adameyko et al., 2005; Kim et al., 2007b). LRRC10 is highly conserved (Kim et al., 2007a) and exclusively expressed in cardiomyocytes (Kim et al., 2007b; Brody et al., 2013),suggesting critical cardiac functions for LRRC10. Recently, studies in Lrrc10 knockout mice (Brody et al., 2012, 2016) and the identification of LRRC10 mutations in human DCM (Qu et al., 2015) have sparked interest in the underlying molecular mechanisms that mediate cardiac disease when LRRC10 is absent or mutated. LRRC10 belongs to the diverse LRRC protein superfamily, which is comprised of many proteins that have in common their leucine-rich repeat (LRR) domains that function as protein interaction motifs (Kobe and Deisenhofer, 1994, 1995; Kobe and Kajava, 2001). LRRs are sequences of 20–30 amino acids rich in leucine and other aliphatic amino acids. LRRCs contain two or more LRRs aligned in tandem to form a curved non-globular, solenoid-shaped structure that is ideal for mediating protein:protein interactions (Kobe and Deisenhofer, 1994, 1995; Bella et al., 2008). LRRC10 is about 32 kDa, and has no known functional domains except its seven LRRs (Kim et al., 2007a,b), suggesting that its molecular functions rely on protein interactions.

LRRC10 is expressed in the developing heart and upregulated at birth with elevated protein levels maintained in adulthood (Brody et al., 2013). Cardiomyocyte-specific expression of LRRC10 is tightly controlled by the cardiac transcription factors Nkx2-5, GATA4, and serum response factor (SRF) via conserved regulatory elements near the LRRC10 promoter region (Fan et al., 2011; Brody et al., 2013). Investigation of Lrrc10 homozygous knockout (Lrrc10−/−) mice (Manuylov et al., 2008) has led to recent discoveries linking LRRC10 to the molecular etiology of DCM (Brody et al., 2012; Qu et al., 2015). Here, we review recent findings in the Lrrc10−/<sup>−</sup> mouse and human idiopathic DCM patients that implicate LRRC10 in the pathogenesis of DCM, discuss molecular alterations in the Lrrc10−/<sup>−</sup> heart that may contribute to cardiomyopathy, and preface ongoing work investigating the molecular function of LRRC10.

### LRRC10 DELETION CAUSES DILATED CARDIOMYOPATHY IN MICE

Pioneering studies in zebrafish demonstrated that Lrrc10 is required for normal cardiac function in vertebrates (Kim et al., 2007a). Knockdown of Lrrc10 in zebrafish causes cardiac developmental defects, reduced cardiac function, and lethality (Kim et al., 2007a). To investigate LRRC10 function in the mammalian heart, Lrrc10−/<sup>−</sup> mice were generated (Manuylov et al., 2008). Lrrc10−/<sup>−</sup> mice exhibit perinatal cardiomyopathy and progressive DCM in adulthood (Brody et al., 2012). Lrrc10−/<sup>−</sup> mice have reduced cardiac function prior to birth that progresses to eccentric cardiac growth, ventricular dilation, and further deterioration of cardiac function in adult mice (Brody et al., 2012; **Figure 1**). These studies established the Lrrc10−/<sup>−</sup> mouse as a novel model of pediatric cardiomyopathy and implicated LRRC10 as a candidate DCM gene in humans. Moreover, Lrrc10−/<sup>−</sup> mice exhibit greatly reduced cardiac contractility and exacerbated remodeling in response to pressure overload induced by transverse aortic constriction (Brody et al., 2016). The accelerated progression of DCM observed in Lrrc10−/<sup>−</sup> mice after pressure overload indicates that deletion of LRRC10 renders the heart sensitive to disease pathogenesis during hypertensive remodeling, suggesting that human patients with mutations in the LRRC10 gene may be prone to more fulminant disease and DCM under conditions of pressure overload, such as aortic stenosis or elevated blood pressure.

Lrrc10−/<sup>−</sup> mice exhibit an uncommon form of cardiac remodeling characterized by direct progression to ventricular dilation without compensatory concentric hypertrophic growth, and cardiac functional impairment in the absence of an increase in myocyte death or cardiac fibrosis (Brody et al., 2012). In response to pressure overload Lrrc10−/<sup>−</sup> mice are capable of mounting an appropriate concentric cardiac hypertrophy response, which is accompanied by further eccentric cardiac growth and dilation with dramatically reduced cardiac functional performance but similar fibrotic remodeling and cardiomyocyte death compared to controls (Brody et al., 2016). Thus, LRRC10 appears to be required to maintain cardiac contractile function and its absence causes dilative cardiac remodeling.

Analyses of adult Lrrc10−/<sup>−</sup> hearts identified transcriptional and molecular alterations during the progression of DCM.

Pathway analysis of gene expression profiling in adult Lrrc10−/<sup>−</sup> hearts revealed upregulation of genes involved in oxidative phosphorylation and myofilament contraction as the most prominent transcriptional alterations, including cytochrome c oxidase, ATP synthase, and NADH dehydrogenase genes, and Tnni3, Tnnt1, Tpm1, and Mybpc3 (Brody et al., 2012; **Figure 1**). Despite increased transcript levels, myofilament protein abundance was not increased in Lrrc10−/<sup>−</sup> hearts (Brody et al., 2012), likely because the ordered myofilament lattice contains stoichiometric quantities of sarcomeric proteins (Michele et al., 1999) and thus cannot accommodate additional myofilament proteins beyond the normal rate of turnover, even if transcript levels are elevated. Upregulation of transcripts involved in oxidative phosphorylation and myofilament contraction in the Lrrc10−/<sup>−</sup> heart are likely a compensatory attempt to bolster cardiac bioenergetics and sarcomeric proteins, respectively, to cope with diminished contractile function.

Various signaling pathways play important roles in regulating cardiac function and the progression to cardiomyopathy and heart failure (McNally et al., 2013; van Berlo et al., 2013). However, precise roles of specific signaling pathways in the progression of DCM have not been fully elucidated. Lrrc10−/<sup>−</sup> hearts activate protein kinase C ε (PKCε) and Akt signaling (Brody et al., 2012; **Figure 1**). PKCε is activated downstream of GPCR agonist or mechanical stimulation and is thought to be cardioprotective in large part due to its augmentation of mitochondrial function (Inagaki et al., 2003; Iwata et al., 2005; McCarthy et al., 2005; Budas and Mochly-Rosen, 2007). Activation of PKCε reduces ventricular dilation and hypertrophy but does not rescue contractile dysfunction in the cTnTR141W transgenic mouse model of DCM (Lu et al., 2014), suggesting that PKCε may partially ameliorate pathological cardiac remodeling in some forms of DCM. In this regard, activation of PKCε potentially prevents the progression to congestive heart failure in Lrrc10−/<sup>−</sup> mice.

Akt is a serine/threonine kinase that is protective in the heart predominantly due to its antiapoptotic effects on myocytes (Miyamoto et al., 2009). Adult Lrrc10−/<sup>−</sup> hearts exhibit elevated levels of active Akt phosphorylated at Ser-473 (Brody et al., 2012), which may limit cardiomyocyte apoptosis. Activation of Akt and upregulation of oxidative phosphorylation and myofilament contraction genes observed in Lrrc10−/<sup>−</sup> hearts have also been reported in human DCM and heart failure (Haq et al., 2001; Barrans et al., 2002; Grzeskowiak et al., 2003; Asakura and Kitakaze, 2009; Colak et al., 2009; Sopko et al., 2011). Therefore, these molecular alterations may represent compensatory protective pathways associated with the progression of DCM. However, it remains unknown how these signaling pathways are upregulated and what the downstream effects are in the Lrrc10−/<sup>−</sup> heart. It is plausible that PKCε or Akt has novel roles regulating downstream signaling pathways that function in the setting of cardiac dysfunction and remodeling caused by LRRC10 deletion. Thus, it would be interesting to determine whether upregulation of these pathways is indeed cardioprotective in certain forms of DCM, which would aid in the identification of therapeutic targets for the treatment of heart failure patients.

To investigate potential pathogenic mechanisms underlying DCM caused by LRRC10 deficiency, molecular alterations were identified in embryonic Lrrc10−/<sup>−</sup> hearts (Brody et al., 2012). Gene expression profiling of Lrrc10−/<sup>−</sup> hearts at embryonic day (E) 15.5, prior to the development of cardiac dysfunction at E17.5 or eccentric ventricular remodeling that occurs in adulthood, revealed upregulation of the actin cytoskeleton and focal adhesion gene pathways to be the most significantly dysregulated gene networks (Brody et al., 2012; **Figure 1**). Embryonic Lrrc10−/<sup>−</sup> hearts have elevated transcripts for integrin β, integrin-linked kinase (ILK), and parvin-α, which are all induced in the heart in response to biomechanical stress (Babbitt et al., 2002; Zemljic-Harpf et al., 2007; Sopko et al., 2011). Upregulation of integrin β1, vinculin, and talin was observed at the protein level in embryonic Lrrc10−/<sup>−</sup> hearts (Brody et al., 2012), suggesting alterations in focal adhesion complexes and a potential mechanosensing role for LRRC10 (See Discussion section below).

### MOLECULAR FUNCTIONS OF LRRC10

LRRC10 was shown to interact with α-actinin and α-sarcomeric actin at the Z-disc in cardiomyocytes and directly bind all actin isoforms (Brody et al., 2012). Interaction of LRRC10 with actin thin filaments appears to be dynamic as this interaction is reduced in response to pressure overload (Brody et al., 2016). Thus, LRRC10 may localize to the Z-disc, T-tubule, or other locations within the cardiomyocyte in a stimulus-dependent manner or in response to certain mechanical or molecular signals, potentially using its interaction with actin at the Z-disc or cytoskeleton as a docking station.

Transmission electron microscopy analyses demonstrated that LRRC10 localizes to the dyad region in cardiomyocytes (Kim et al., 2007b), where the T-tubule comes into close juxtaposition to the Z-disc and sarcoplasmic reticulum (SR). The dyad serves as a critical link between the T-tubule network, SR, Z-disc and cytoskeletal proteins, with roles in regulating and anchoring ion channels, contractile and structural proteins, and signaling molecules. Localization of LRRC10 near the Zdisc positions it at an optimal subcellular location to mediate signaling responses to mechanical stress (Frank and Frey, 2011). The Z-disc contains mechanical scaffolding and signaling molecules that structurally and functionally link the myofilament to the costamere, cytoskeleton, and extracellular matrix (Ervasti, 2003; Luther, 2009; Frank and Frey, 2011). Genetic deletion or mutation of many genes encoding Z-disc and cytoskeletal proteins results in DCM in mice and humans, including Cypher (Vatta et al., 2003; Zheng et al., 2009), muscle LIM protein (MLP) (Arber et al., 1997; Knoll et al., 2002), integrin-linked kinase (ILK) (White et al., 2006; Knoll et al., 2007), vinculin (Zemljic-Harpf et al., 2007), and desmin (Li et al., 1999). Thus, defects in focal adhesion complex, cytoskeletal, or Z-disc components can participate in the pathobiology of DCM. MLP anchors calcineurin at the Z-disc to mediate activation of downstream NFAT-dependent prohypertrophic gene expression in response to myocardial infarction (Heineke et al., 2005) and also shuttles between the Z-disc and nucleus in response to certain stimuli (Ecarnot-Laubriet et al., 2000; Boateng et al., 2007, 2009). Therefore, LRRC10 may bind actin to properly localize, displace, or regulate the function of an interacting factor at the Z-disc or may dissociate from the Z-disc to other subcellular locations to perform regulatory functions.

Other LRRCs bind actin, suggesting that actin binding may be a general mechanism for LRRCs to localize or dock at specific cellular locations. For example, LRRC67 binds all actin isoforms (Wang et al., 2010) and the LRR domain of tropomodulin-1 mediates its binding to sarcomeric actin in cardiomyocytes (Tsukada et al., 2011). LRRCs have also been previously reported to have mechanosensing functions. The striated-muscle-specific protein, LRRC39, localizes to the M-line in cardiomyocytes and regulates SRF-dependent transcription (Will et al., 2010), suggesting that LRRCs may serve as local mechanosensors in cardiomyocytes.

Lrrc10−/<sup>−</sup> mice exhibit defective cardiac contractility prior to ventricular remodeling (Brody et al., 2012), and severely compromised cardiac function in response to pressure overload (Brody et al., 2016), indicating that LRRC10 is necessary to maintain cardiac contractile function. Nonetheless, LRRC10 does not have a role in directly regulating cardiomyocyte contraction at the level of the myofilament (Brody et al., 2012, 2016). No alterations in force development or myofilament calcium sensitivity were detected in skinned myocardium from Lrrc10−/<sup>−</sup> hearts (Brody et al., 2012). Although single cell contractility is not altered at baseline in isolated Lrrc10−/<sup>−</sup> myocytes, the contractile response to β-adrenergic stimulation is blunted (Brody et al., 2016). These data indicate that LRRC10 does not directly regulate myofilament contraction or cross-bridge cycling. This is consistent with a mechanosensing function for LRRC10. In the loaded, intact heart where cardiomyocytes must sense mechanical stress and preload, LRRC10 is required for cardiomyocyte contractile function. In contrast, in unloaded isolated myocytes where mechanical strain cannot be sensed by the myocyte, LRRC10 is dispensable for cardiomyocyte contractility.

These data are also consistent with a role for LRRC10 in excitation-contraction coupling such that when LRRC10 is absent or mutated, calcium cycling is defective resulting in aberrant levels of calcium available at the myofilament to stimulate contraction (Luo and Anderson, 2013). Although the reduced contractile response of isolated Lrrc10−/<sup>−</sup> myocytes to β-adrenergic stimulation (Brody et al., 2016) could be a consequence of DCM in Lrrc10−/<sup>−</sup> mice, it could also be explained by a primary function for LRRC10 in coupling adrenergic stimulation to myofilament contraction by facilitating excitation-contraction coupling.

It is intriguing that LRRCs have been shown to regulate ion channel activity. Activation of the large-conductance, calciumand voltage-activated potassium (BK) channel (encoded by the Slo1 gene) is regulated by LRRC26, which serves as an auxiliary γ-subunit to regulate BK channel activity (Braun, 2010; Yan and Aldrich, 2010; Evanson et al., 2014). Moreover, LRRC52 acts as a testis-specific γ-subunit regulator of the alkalization-activated Slo3 potassium channel (Yang et al., 2011), indicating not only that some LRRCs may function as regulators of the Slo family of potassium channels (Zhang and Yan, 2014), but also that LRRCs may provide tissue specificity to the regulation of ion channel activity. Thus, LRRC10 may serve as a cardiomyocytespecific auxiliary protein and regulator of ion channel activity to mediate appropriate excitation-contraction coupling and resultant contractility in cardiomyocytes.

### ASSOCIATION OF LRRC10 MUTATIONS WITH DILATED CARDIOMYOPATHY IN HUMANS

Two heterozygous mutations in LRRC10 were identified in human patients with idiopathic DCM (Qu et al., 2015). These studies identified p.L41V and p.L163I missense mutations in LRRC10 in two unrelated families with DCM. Both mutations were inherited in an autosomal dominant manner and cosegregated with DCM with complete penetrance (Qu et al., 2015). Residues L41 and L163 are highly conserved, suggesting they are critical for the structure and/or function of the LRRC10 protein. These mutants may not localize properly to the Z-disc or T-tubule in the dyad region. Alternatively, LRRC10 DCM-associated mutants may lose their ability to physically or functionally interact with cofactors. The amino acid substitutions in these LRRC10 mutants (L41V and L163I) are relatively moderate biochemical alterations and not predicted to drastically alter the overall three dimensional conformation of the LRRC10 protein. Thus, it remains unknown if the L41V or L163I mutants function as a dominant negative or a loss- or gain-of-function mutation. Further investigation is necessary to determine precisely how these mutations alter the molecular structure of LRRC10 and how this mechanistically perturbs LRRC10 function to cause disease.

Future sequencing for LRRC10 in human idiopathic DCM will be very informative on the prevalence of LRRC10 mutations in DCM and the potential for additional mutations to contribute to human cardiac disease. Identification of additional pathogenic mutations in the LRRC10 protein coupled with molecular studies of recently identified LRRC10 disease-associated mutants will shed light on how these mutations disrupt LRRC10 function and cause DCM.

### PERSPECTIVES

Much insight has been gained into the role of LRRC10 in DCM and molecular mechanisms of DCM pathogenesis in recent years. The identification of two novel mutations in LRRC10 that are associated with human idiopathic DCM has opened the door for investigation into the roles of LRRC10 in human cardiac disease and the underlying molecular mechanisms that cause DCM in response to mutation or genetic ablation of LRRC10. The Lrrc10−/<sup>−</sup> mouse has provided a valuable animal model to investigate the molecular function of LRRC10 that can translate to understanding of human disease. Molecular evidence and investigation of Lrrc10−/<sup>−</sup> mice thus far has pointed to roles for LRRC10 in mechanosensing and/or excitation-contraction coupling. Future studies will

investigate calcium cycling in Lrrc10−/<sup>−</sup> cardiomyocytes to determine if LRRC10 has a fundamental role in regulation of excitation-contraction coupling. Identification of proteins that functionally interact with LRRC10 will be crucial to determine the molecular functions of LRRC10 and mechanistic basis of DCM. Further sequencing of LRRC10 in human idiopathic DCM will reveal the prevalence of LRRC10 mutations in DCM and potentially identify novel mutations associated with cardiac disease in humans. Generation of knock-in mice for human DCM associated LRRC10 mutations will provide valuable models to investigate molecular mechanisms of DCM. Moreover, generation of cardiomyocytes from patient-derived induced pluripotent stem cells (iPSCs) or human pluripotent stem cells (Sharma et al., 2013) containing DCM-linked LRRC10 mutations will serve as powerful tools to investigate human cardiomyopathy. These studies will provide models to test therapeutic strategies to treat DCM and aid in the

### REFERENCES


discovery of pathogenic mechanisms underlying DCM in human patients.

### AUTHOR CONTRIBUTIONS

MB and YL made substantial contributions to the conception and design of the manuscript, drafted and critically revised the manuscript, approved the final version of the manuscript, and agree to be accountable for all aspects of the work.

### FUNDING

This work was supported by National Institutes of Health (NIH) Grant HL-067050 and American Heart Association (AHA) Grant 12GRNT12070021 (to YL), AHA Predoctoral Fellowship 11PRE5580012 (to MB), and NIH Molecular and Environmental Toxicology Training Grant T32ES007015.


insufficiency as a cause of dilated cardiomyopathy. Science 349, 982–986. doi: 10.1126/science.aaa5458


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Brody and Lee. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Commentary: Effect of Skeletal Muscle Native Tropomyosin on the Interaction of Amoeba Actin with Heavy Meromyosin

Joseph M. Chalovich<sup>1</sup> \* and Dylan Johnson<sup>2</sup>

*<sup>1</sup> Department of Biochemistry, Brody School of Medicine at East Carolina University, Greenville, NC, USA, <sup>2</sup> Department of Chemistry, East Carolina University, Greenville, NC, USA*

Keywords: troponin, tropomyosin, cardiomyopathy, troponin T, mutations

#### **A commentary on**

### **Effect of Skeletal Muscle Native Tropomyosin on the Interaction of Amoeba Actin with Heavy Meromyosin**

by Eisenberg, E., and Weihing, R. R. (1970). Nature 228, 1092–1093. doi: 10.1038/2281092a0

#### Edited by:

*Jose Renato Pinto, Florida State University, USA*

#### Reviewed by:

*Aldrin V. Gomes, University of California, Davis, USA Darshan Trivedi, Stanford University, USA*

> \*Correspondence: *Joseph M. Chalovich chalovichj@ecu.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *28 June 2016* Accepted: *15 August 2016* Published: *31 August 2016*

#### Citation:

*Chalovich JM and Johnson D (2016) Commentary: Effect of Skeletal Muscle Native Tropomyosin on the Interaction of Amoeba Actin with Heavy Meromyosin. Front. Physiol. 7:377. doi: 10.3389/fphys.2016.00377* Troponin-tropomyosin inhibits skeletal and cardiac muscle contraction at low Ca2+. Binding of rigor-type myosin S1 to actin-tropomyosin-troponin, particularly at saturating Ca2+, produces activation of myosin ATPase activity in excess of that seen in the absence of the regulatory proteins. The binding energy of S1 can overcome the inhibitory activity of troponin (Bremel et al., 1972) and may allow tropomyosin to move deep into the groove of actin. That particular arrangement of actin, tropomyosin, and troponin is a much better activator of ATP hydrolysis than actin alone. That active configuration of actin was called state 2 in the Hill model (Hill et al., 1980) and later named the M state because of its requirement for tight myosin binding.

Eisenberg and Weihing found evidence that troponin itself can stabilize the active state of actin in the absence of high affinity S1 binding (Eisenberg and Weihing, 1970). They showed that troponin-tropomyosin enhanced the ability of amoeba actin to activate myosin S1 ATPase activity at high Ca2+. That observation is often overlooked but may be an important clue to managing some muscle disorders. Actin filaments containing the hypertrophic cardiomyopathy associated 114 mutation of TnT also enhanced S1 ATPase rates 2-3-fold higher than actin filaments without bound regulatory proteins (Gafurov et al., 2004). Because small changes in the structure of actin or troponin allow this increased activation to occur, the troponin complex must have a latent ability to enhance actin activation of myosin ATPase activity. The 14 C-terminal residues of TnT attenuate the ability of troponin to enhance actin activation. Troponin containing 114 TnT might act by stabilizing tropomyosin in the M state position of the actin groove under saturating Ca2<sup>+</sup> conditions.

The inactive state of actin-tropomyosin-troponin (state 1 or the B state) occurs at low free Ca2<sup>+</sup> when the inhibitory region of TnI is bound to actin. Because of associations among the regulatory proteins, tropomyosin is stabilized outside of the actin groove and there is little stimulation of myosin ATPase activity. Removal of the 14 C-terminal residues of TnT prevents formation of the B state. Compared with wild type actin filaments in EGTA, those containing 114 TnT exhibit

**38**

TABLE 1 | C-terminal troponin T sequence comparison.


less cooperativity in equilibrium binding of myosin S1 (Gafurov et al., 2004), and they do not exhibit the acrylodan tropomyosin fluorescence increase under conditions favoring the inactive state (Borrego-Diaz and Chalovich, 2010; Franklin et al., 2012).

Ca2<sup>+</sup> binding to TnC opens a hydrophobic patch to which the switch region of TnI can bind (Herzberg et al., 1986). Under this condition, TnI is detached from actin and tropomyosin is situated in the actin groove. Several lines of evidence indicate that the major state formed with Ca2<sup>+</sup> is a second inactive state with tropomyosin partially in the actin groove (Trybus and Taylor, 1980; McKillop and Geeves, 1991; Lehman et al., 2001; Kimura et al., 2002; Pirani et al., 2005; Poole et al., 2006). Full movement into the groove to form the active M state requires rigor S1 binding or a structural change in troponin. In the Hill model of regulation, Ca2<sup>+</sup> binding to troponin was thought to create an inactive state 1 with bound Ca2+. State 1 with bound Ca2<sup>+</sup> may be equivalent to the state intermediate between the B and M states that is called the C state (because of its link to Ca2+). The level of activation of ATPase activity at saturating Ca2<sup>+</sup> is determined by the amount of M state formed in its equilibrium with the C state. The major state formed with 114 TnT containing actin filaments at low Ca2<sup>+</sup> is likely to be the C state as the B state cannot form. The C state is also stabilized by a hypertrophic cardiomyopathy causing mutation, R146G TnI. The R146G mutation in TnI gives relative stabilization to the C state at low Ca2<sup>+</sup> and the C state

### REFERENCES


is highly stabilized at saturating Ca2+(Mathur et al., 2009). An analysis of ATPase rates of R146G TnI containing actin filaments supported the idea that the C state is ineffective in stimulating myosin ATPase activity.

Studying natural mutations and modifications of troponin has given muscle researchers insights into the regulation of contraction. Long term deviations from the normal distribution of B, C, and M states of regulated actin seem to lead to progressive cardiac dysfunction. The last 14 residues of human cardiac TnT are critical for controlling the equilibria among the B, C, and M states of regulated actin; they stabilize the B state at low Ca2<sup>+</sup> and destabilize the M state at saturating Ca2+. **Table 1** compares the C-terminal sequences of several forms of troponin T. Note the conservation of the four terminal residues and the pattern of basic residues (bold). The regularly spaced basic residues suggest the possibility of acidic target sites for controlling both the B and M states.

The C-terminal region of TnT might function by directly affecting movement of tropomyosin on the actin surface. The Cterminal region could destabilize the active M state at saturating Ca2<sup>+</sup> by interfering with tropomyosin movement into the actin groove. At low Ca2+, the C-terminal region of TnT could participate in holding tropomyosin away from the actin groove. The C-terminal 14 residues of TnT could also potentially alter the pathway of transmission of information from Ca2<sup>+</sup> binding to TnC through the events leading to tropomyosin repositioning. Deciphering the mechanisms of action of the C-terminal region of TnT may lead to new therapies for cardiac disorders.

### AUTHOR CONTRIBUTIONS

All authors listed, have made substantial, direct and intellectual contribution to the work, and approved it for publication.

### FUNDING

Funded by NIH grant AR44504 and the Brody Brothers Grant to JC.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Chalovich and Johnson. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Contribution of Post-translational Phosphorylation to Sarcomere-Linked Cardiomyopathy Phenotypes

### Margaret V. Westfall\*

*Department of Cardiac Surgery, University of Michigan, Ann Arbor, MI, USA*

Secondary shifts develop in post-translational phosphorylation of sarcomeric proteins in multiple animal models of inherited cardiomyopathy. These signaling alterations together with the primary mutation are predicted to contribute to the overall cardiac phenotype. As a result, identification and integration of post-translational myofilament signaling responses are identified as priorities for gaining insights into sarcomeric cardiomyopathies. However, significant questions remain about the nature and contribution of post-translational phosphorylation to structural remodeling and cardiac dysfunction in animal models and human patients. This perspective essay discusses specific goals for filling critical gaps about post-translational signaling in response to these inherited mutations, especially within sarcomeric proteins. The discussion focuses primarily on pre-clinical analysis of animal models and defines challenges and future directions in this field.

#### Edited by:

*P. Bryant Chase, Florida State University, USA*

#### Reviewed by:

*Aldrin V. Gomes, University of California, Davis, USA Yuanhua Cheng, University of Washington, USA*

> \*Correspondence: *Margaret V. Westfall wfall@med.umich.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *19 July 2016* Accepted: *30 August 2016* Published: *14 September 2016*

#### Citation:

*Westfall MV (2016) Contribution of Post-translational Phosphorylation to Sarcomere-Linked Cardiomyopathy Phenotypes. Front. Physiol. 7:407. doi: 10.3389/fphys.2016.00407* Keywords: myofilament, post-translational modification, signaling, cardiomyopathy, phosphorylation

### INTRODUCTION

More than 3800 gene mutations are linked to inherited cardiomyopathies (ICs) and identification of underlying gene mutations continues to expand (https://www.ncbi.nlm.nih.gov/clinvar/). Animal models expressing individual mutations have provided insight into the human disease and a better understanding of myofilament force transduction mechanisms (Tardiff, 2005, 2011). In these models, the pathophysiological response is often linked to a specific disease progression such as hypertrophic, dilated, restrictive, left ventricular noncompaction, and/or arrhythmogenic right ventricular phenotypes (Fatkin et al., 2014). However, understanding how a specific mutation leads to the cardiac phenotype remains a persistent question (van der Velden et al., 2015), and the factors contributing to disease variability in patients are only partially understood. One area which may provide insight into these issues, and therefore deserves further consideration, is dynamic local myofilament signaling and its impact on downstream networks and/or global signaling within cardiac myocytes. This Perspective focuses on the possibility that IC-linked mutations alter local myofilament signaling and contribute to downstream remodeling and disease progression. Our current understanding of dynamic post-translational myofilament signaling also is briefly summarized to lay the foundation for future work aimed at investigating relationships between IC-linked mutations and myofilament modulation.

First, it is important to point out that previous work in humans and animal models indicate IClinked heart disease is complex. In patients, morbidity and mortality are often not easily explained

**41**

by an identified mutation acting as a primary physiological insult or substrate (Ho et al., 2015). Instead, a temporal and spatial network of factors contributes to progressive cardiac structural and functional remodeling in IC patients, and can ultimately evolve into end stage heart failure. In addition to a primary mutation, factors known or suspected to increase the risk for disease include second hit and epigenetic mutations, polymorphisms, and other genetic modifiers, which include genes linked to cardiac remodeling, and environmental factors such as sex, aerobic activity levels, and risk factors such as hypertension (McNally et al., 2013; Månsson, 2014; Ho et al., 2015; van der Velden et al., 2015). In addition, cardiac remodeling and dysfunction is progressive on many levels and includes alterations in cellular morphology, signaling, and function, cell-cell architecture, plus organ-level electrical, and pump dysfunction. Signaling modulation is predicted to be an important focus for future work in recent reviews (Yar et al., 2014; van der Velden et al., 2015). **Figure 1** illustrates the variety of signaling pathways known to phosphorylate myofilament proteins, and therefore, could contribute to modulation by targeting myofilament proteins for phosphorylation. The emphasis on myofilament modulation in this Perspective is based on the possibility that myofilament phosphorylation may be an early secondary response to IC-linked mutations, and therefore present prior to significant structural and functional remodeling within myocytes. Myofilament signaling may continue to contribute to adaptive functional responses and/or initiate one or more later compensatory behaviors associated with cardiac remodeling and disease, such as alterations in excitation-contraction (E-C) coupling, myocyte Ca2<sup>+</sup> handling, and metabolism (Ashrafian et al., 2011).

In animal models expressing IC-linked mutations, E-C coupling and Ca2<sup>+</sup> handling network alterations are often detected in parallel with in vivo evidence of cardiac performance compensation and/or dysfunction, and prior to end-stage heart failure (Ashrafian et al., 2011). These changes in Ca2<sup>+</sup> increase the risk for developing arrhythmia and sudden cardiac death (Ashrafian et al., 2011; Yar et al., 2014), and the events responsible for initiating and/or causing remodeling of the Ca2<sup>+</sup> signal may be critical for understanding IC-linked disease progression. Interventions to prevent or delay disease progression prior to the onset of Ca2<sup>+</sup> remodeling would be desirable in high risk families and/or patients. However, little is known about the process or mechanism(s) responsible for the initiation of Ca2<sup>+</sup> remodeling in these patients. IC-linked mutations may initiate changes in local myofilament signaling network(s) and the myofilament post-translational modification (PTM) pattern helps to maintain cardiac performance prior to any changes in Ca2<sup>+</sup> handling. Evidence is accumulating that myofilament residues targeted by signaling pathways can initiate additional "secondary" or "adaptive" changes in the phosphorylation of other myofilament residues to modulate function (Montgomery et al., 2002; Scruggs et al., 2006; Lang et al., 2015), which appears to maintain steady state contractile function in the short term (Lang et al., 2015). Chronic activation of this secondary signaling within the myofilament may become inadequate and/or serve as a direct trigger for later structural and functional remodeling such as the IC-associated alterations in Ca2<sup>+</sup> handling and E-C coupling described above. Although critical studies are still needed to prove this idea, future support for a direct role of local myofilament signaling in response to IC-linked mutations could lead to early diagnostic tests and/or therapeutic strategies to prevent or minimize IC disease progression in high risk patients.

### IC-LINKED MUTATIONS AND A ROLE FOR LOCAL MYOFILAMENT SIGNALING

There are some general observations which are consistent with a role for local myofilament signaling responses in IClinked structural and functional remodeling. First, a causative mutation is usually not identified in new probands until cardiac dysfunction develops, which is often during adolescence or later (Cirino and Ho, 2008). The known impact of an IC-linked mutation on myofilament function and/or Ca2<sup>+</sup> remodeling also may not predict the cardiac phenotype in animal models or patients, especially at early time points (Jacques et al., 2008; Jensen et al., 2013). A recent developmental study also demonstrated that inhibition of IC-linked gene expression during the first 6 weeks of life markedly reduced cardiac remodeling at 40 weeks in an α–MHCR403Q mouse model, while more modest improvements developed if mutant protein expression was inhibited after 6 weeks of age (Cannon et al., 2015). Secondary modulatory mechanisms in the myofilament are consistent with these observations and could contribute to developmental lags and/or unexpected phenotypes. Myofilament modulatory networks also may undergo developmental transitions over the same perinatal period observed for many contractile proteins (Cummins, 1982; Lyons et al., 1990; Reiser et al., 1994, 2001; Suurmeijer et al., 2003). Impaired or altered myofilament signaling development could result in permanently sub-optimal myofilament modulation in adults with IC-linked mutations. Alternatively, this local signaling network modulation may be hard-wired to respond to myofilament perturbations such as IClinked mutants, and either directly or indirectly trigger further adverse structural and functional remodeling of myocardium.

The local myofilament signaling concept also is supported by reported changes in the phosphorylation of multiple myofilament protein residues in response to IC-linked mutation expression, and alterations in additional phosphorylated residues linked to heart failure (**Table 1**). Altered myofilament phosphorylation develops in at least one IC-linked mutation for each contractile protein, and there is significant potential for myofilament phosphorylation to modulate contractile function based on the myofilament residues already identified as phosphorylation targets (see **Table 1** and references). However, it is not known whether a given contractile protein or mutations clustered into a specific cardiomyopathy produce common spatial and/or temporal phosphorylation patterns. Thus, to test whether myofilament modulatory phosphorylation makes an early contribution to IC-linked phenotypes requires rigorous experimental testing in the future. As part of these studies, it is important to identify the dose-dependent spatial and temporal impact of each IC-linked mutation on myofilament

(CamKII), sterile 20-like kinase 1 (Mst1), p21-activated kinase (PAK), protein kinase A (PKA), protein kinase C (PKC), protein kinase D (PKD), protein kinase G (PKG), and Rho-associated protein kinase (ROCK). Tyrosine kinases which target myofilament troponin I include non-receptor activated Src and the Lck/Yes novel (Lyn) kinase (Salhi et al., 2014). Phosphatases known to target myofilament proteins include myosin light chain phosphatase (MLCP), protein phosphatase I (PPI), protein phosphatase 2A (PP2A) (Solaro and Kobayashi, 2011). The contractile proteins shown in this illustration are slow/cardiac troponin C (s/cTnC), cardiac troponin I (cTnI), cardiac troponin T (cTnT), alpha-tropomyosin (Tm), and actin in the thin filament plus the myosin heavy chain (MHC), myosin light chains 1 and 2 (MLC1, MLC2, respectively), and cardiac myosin binding protein C (cMyBP-C). The proteins identified as phosphorylation targets include cTnI, cTnT, Tm, MLC1, MLC2, and cMyBP-C (indicated by red P in the legend). For further information see the following references: (He et al., 2003; Barefield and Sadayappan, 2010; Solaro and Kobayashi, 2011; Streng et al., 2013; Westfall, 2014; Huang and Szczesna-Cordary, 2015).

phosphorylation and understand the modulatory impact of each phosphorylated contractile protein residue. Although not included here, phosphorylation of additional sarcomeric proteins, such as titin, also may contribute to this modulation. A few representative studies on cardiac troponin I (cTnI) mutations and phosphorylation are briefly presented below to illustrate our current understanding and the rationale for future directions on myofilament phosphorylation in response to IClinked mutations.

### IC-LINKED MUTATIONS AND β-AR SIGNALING IN MYOFILAMENTS

Previous work on β-adrenergic receptor (β-AR) signaling provides direct support for a role of local myofilament signaling in IC-linked changes in cardiac function. Several IC-linked mutations directly influence myofilament phosphorylation and/or β-AR signal transduction, as illustrated by representative cTnI mutations. Protein kinase A (PKA)-induced cTnI-S23/24 phosphorylation significantly contributes to the positive β-AR-induced lusitropic response (Takimoto et al., 2004; Yasuda et al., 2007; note that residue numbering is based on Uniprot human protein accession numbers, see **Table 1**). Uncoupling between the β-AR receptor and this response often develops in IC-linked animal models (reviewed by Messer and Marston, 2014). Poor outcomes are associated with myofilament β-AR uncoupling in other types of human heart failure, and the ability of IC-linked mutations to cause this uncoupling is proposed to be a prognostic indicator in patients with IC-linked mutations (Messer and Marston, 2014).



\**Uniprot number for human protein; numbering includes Met1 residue.* \*\**Bold font indicates direct evidence; Regular font indicates indirect evidence; Changes in IC-linked sites are also detected during HF.*

Several mechanisms can produce β-AR uncoupling in response to IC-linked mutations. Some IC-linked mutations directly disrupt post-translational cTnI-S23/24 phosphorylation. For example, the IC-linked cTnI-R21C mutation directly blocks PKA-induced phosphorylation of the adjacent S23/24 residues (Gomes et al., 2005; Wang et al., 2012; Dweck et al., 2014; Cheng et al., 2015). IC-linked mutations in more distant proteins also modify this PKA-targeted cTnI phosphorylation (Najafi et al., 2016). Alternatively, PKA continues to phosphorylate myofilament targets, such as cTnI-S23/24, in the presence of other IC-linked mutations. Representative mutations such as cTnI-R145G and -P82S, disrupt signal transduction within cTnI to cause β-AR uncoupling (Deng et al., 2001; Messer and Marston, 2014; Ramirez-Correa et al., 2015; Cheng et al., 2016). The cTnI-P82S mutation is noteworthy because the diastolic dysfunction and late-onset of disease in humans associated with this mutation is postulated to be a long-term consequence of secondary alterations in PKA-related myofilament signaling (Nimura et al., 2002; Mogensen et al., 2004; Frazier et al., 2008; Ramirez-Correa et al., 2015). In addition, IC-linked mutations may indirectly cause β-AR uncoupling due to changes in the overall myofilament phosphorylation status (Kooij et al., 2010), which could result from differences in other myofilament associated kinase and phosphatase activities (**Figure 1**).

### IC-LINKED MUTATIONS AND ADDITIONAL MYOFILAMENT SIGNALING PATHWAYS

While β-AR induced PKA modulation is among the most studied signaling pathways targeting myofilaments, a number of additional signaling pathways also modulate myofilament function (**Figure 1**). A few studies also indicate that IClinked mutations modify both kinase and phosphatase signaling pathways and downstream target residues other than β-AR/PKA-targeted sites. Mutation-related alterations associated with the protein kinase C (PKC) second messenger serve as a representative example. First, progressive increases in cardiac PKC expression and increased PKC affinity for sarcomeric proteins are associated with IC-linked mutations (Arimura et al., 2004; Sfichi-Duke et al., 2010). Modification of endtargets, such as PKC phosphorylation of cTnI-S42/44 provides some initial support. Myofilament function is similarly modified by either PKC-induced phosphorylation or phospho-mimetic cTnI-S42/44 substitutions (Noland et al., 1996; Burkart et al., 2003b). Interestingly, the myofilament response to phosphomimetic cTnI-S42/44 is significantly greater in myofilaments expressing IC-linked tropomyosin (Tm)-E180G compared to controls (Burkart et al., 2003a).

Multiple neurohormones activate receptor-induced PKC signaling in myocytes, such as angiotensin II (AgII), endothelin, and catecholamine activation of α-adrenergic receptors (Dorn and Force, 2005). Accelerated and/or exaggerated cardiac remodeling and dysfunction develop in response to one or more of these neurohormones in mice with IC-linked mutations (Maass et al., 2004; Gramlich et al., 2009). This severe response has been interpreted as a stress response, but myofilament-associated PKC activity may already be modified in myofilaments expressing IC-linked mutations independent from receptor activation. Thus, further neurohormone stimulation of PKC may accelerate additional remodeling and produce progressive deterioration in cardiac function. This interpretation is supported by evidence of more severe remodeling in endothelin-treated, cardiac-derived stem cells from patients with IC-linked mutations (Tanaka et al., 2014). Pro-left ventricular polymorphisms present in the renin-angiotensin-aldosterone (RAA) axis also are associated with higher morbidity and mortality in patients with IC-linked mutations (Ortlepp et al., 2002; Kaufman et al., 2007). In addition, environmental stressors known to activate the RAA axis, such as pressure overload, further exacerbate IC-associated contractile dysfunction (Chen et al., 2013). While angiotensin receptor inhibitors failed to reverse fibrosis (Axelsson et al., 2015), some tangential evidence in Duchenne's muscular dystrophy (DMD) patients indicates earlier treatment with these types of inhibitors may be beneficial in treating DMD patients with cardiomyopathy (Duboc et al., 2007; Kamdar and Garry, 2016). DMD is caused by mutations in dystrophin, a crucial component of the costamere, which anchors sarcomeres to the sarcolemma.

Many of the potential signaling networks associated with myofilaments have the ability to produce a range of outputs via multi-layer signaling cascades capable of targeting both kinases and phosphatases, multiple sarcomeric protein targets, and multiple residues targeted within a single myofilament protein (**Figure 1**). The complexity of local myofilament signaling contributes to difficulties in defining the modulatory role for a given signal in myofilament function (Angeli et al., 2004). As a result, these signaling networks may not be easily recognized as contributors to IC-linked remodeling and/or disease. However, these types of pathways also are noteworthy because they are either predicted or known to act as oscillators capable of flexible outputs. Oscillatory signals have the potential to provide highly dynamic modulation to maintain steady state structure/function (Angeli et al., 2004). An IC-linked mutation could disrupt or alter one or more signals in an oscillatory pathway to produce subtle shifts in phosphorylation turnover at multiple target residues. These types of pathways may have little initial impact, but lead to bi- or multi-phasic temporal alterations in one or more target PTMs (Angeli et al., 2004). An IC-linked mutation which chronically induces secondary myofilament signaling to modulate function may either become inadequate to maintain myofilament structure and function, or directly trigger further adaptations beyond the myofilament, such as the myocyte Ca2<sup>+</sup> handling modifications described earlier.

While a secondary reduction in myofilament phosphorylation can coincide with cardiac dysfunction (Bayliss et al., 2013; Alves et al., 2014), no published reports prove bifurcative/ oscillatory signaling and/or altered PTM levels develop in multiple myofilament proteins prior to detectable morphological and/or functional remodeling in animal models with IC-linked mutations. Short-term expression of cardiac troponin T (cTnT)- R92Q in bigenic mice provides some indirect evidence. These mice develop early alterations in a range of signaling pathways associated with structural remodeling, which returned to baseline after turning off mutant expression (Lutucuta et al., 2004). There is also some indirect support for IC-induced oscillatory changes based on changes in Ca2<sup>+</sup> wave frequencies in cardiomyocytes expressing gain-of-function SHP-2/PTPN11 mutations, which are linked to Noonan's syndrome (Uhlén et al., 2006). In addition, myofilament PTMs and functional responses observed in adult myocytes after PKC gene transfer are consistent with bifurcative myofilament signaling (Hwang et al., 2013). These data alone do not provide adequate proof that local myofilament modulation contributes to IC-linked remodeling, but suggest that local myofilament signaling in IC-linked mice is worthy of analysis. Specifically, studies are needed to determine whether local myofilament modulation precedes and/or works in parallel with other adaptive responses associated with structural and functional remodeling observed in IC animal models.

### DYNAMIC SIGNALING MODULATION IN CARDIAC MYOFILAMENTS

IC-linked mutations also may trigger a secondary signaling response, which may arise from structural changes imposed by a mutation. This secondary response could involve one or more signaling pathways known to target myofilament proteins (**Figure 1**). However, in contrast to the typical receptor-based signaling activation discussed earlier, only signaling networks localized to the myofilament undergo changes in activity. This localized signaling also may undergo dynamic changes in response to structural alterations produced during contraction and relaxation. Signaling studies utilizing phospho-mimetic and -null substitutions at myofilament target residues provide some initial support for the presence and role of dynamic, local myofilament signaling modulation (Lang et al., 2013, 2015). Specifically, alterations in the phosphorylation of other myofilament residues develops in myocytes expressing phospho-mimetic or non-phosphorylatable substitutions at one or more kinase-targeted myofilament residues (Montgomery et al., 2002; Scruggs et al., 2006, 2009; Lang et al., 2013; Nixon et al., 2014; Lang et al., 2015). This secondary phosphorylation also is associated with altered functional responses (Lang et al., 2013, 2015). Secondary adaptations in myofilament phosphorylation are reported in both thick and thin filaments using a variety of approaches. For example, a cTnI-S150 phospho-mimetic blunts the β-AR/PKA myofilament response (Nixon et al., 2014) and elevated myofilament phosphorylation develops in mice expressing non-phosphorylatable ventricular myosin light chain (MLC; Scruggs et al., 2009). Other posttranslational modifications also trigger local myofilament phospho-modulation, as indicated by alterations in cardiac myosin binding protein C (cMyBP-C) phosphorylation after S-glutathiolation increases during heart failure (Stathopoulou et al., 2016). Taken together, these results are consistent with local myofilament modulatory signaling changes during sustained structural or functional perturbations in the sarcomere.

Other approaches, such as proteomic analysis of myofilament proteins during heart failure, also hint at dynamic PTM modulation within myofilaments. Heart failure is associated with altered phosphorylation residues in several contractile proteins (see **Table 1**; Dubois et al., 2011; Zhang et al., 2012; Kooij et al., 2013; Walker et al., 2013). Secondary signaling also is reported in some, but not all myofilament phosphomimetic and -null animal models. These local signaling changes may contribute to phenotypic differences among these models, such as the significant differences reported in cTnI models with phosho-substitutions at PKC-targeted sites (Pi et al., 2002; Sakthivel et al., 2005; Bilchick et al., 2007; Kirk et al., 2009). While differences in genetic approach, mouse strain, age, and mutant expressivity may factor into these differences, highly organized signaling network(s) which locally modulate myofilament structure and function also may contribute to divergent phenotypes.

### FUTURE DIRECTIONS

Future work needs to establish the circuitry, physiological functions, and temporal response of local myofilament modulatory signaling, and test whether this local modulation is an early or longer-term contributor to IC-linked remodeling and/or dysfunction. A parallel, translational goal for this work is the development of diagnostic tools, improved clinical management, and therapies to prevent and/or delay disease progression in IC patients. The integration of

### REFERENCES


computational modeling, myofilament, cellular, and in vivo genetic model work is critical for achieving these goals. As a result, significant advancements are likely to depend on an unusually high level of cooperativity and resource sharing among investigators.

### AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and approved it for publication.


cardiovascular phenotypes. Heart Fail. Rev. 10, 237–248. doi: 10.1007/s10741- 005-5253-5


**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Westfall. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# MYBPC1, an Emerging Myopathic Gene: What We Know and What We Need to Learn

### Janelle Geist and Aikaterini Kontrogianni-Konstantopoulos \*

*Department of Biochemistry and Molecular Biology, University of Maryland School of Medicine, Baltimore, MD, USA*

Myosin Binding Protein-C (MyBP-C) comprises a family of accessory proteins that includes the cardiac, slow skeletal, and fast skeletal isoforms. The three isoforms share structural and sequence homology, and localize at the C-zone of the sarcomeric A-band where they interact with thick and thin filaments to regulate the cycling of actomyosin crossbridges. The cardiac isoform, encoded by *MYBPC3,* has been extensively studied over the last several decades due to its high mutational rate in congenital hypertrophic and dilated cardiomyopathy. It is only recently, however, that the *MYBPC1* gene encoding the slow skeletal isoform (sMyBP-C) has gained attention. Accordingly, during the last 5 years it has been shown that *MYBPC1* undergoes extensive exon shuffling resulting in the generation of multiple slow variants, which are co-expressed in different combinations and amounts in both slow and fast skeletal muscles. The sMyBP-C variants are subjected to PKA- and PKC-mediated phosphorylation in constitutive and alternatively spliced sites. More importantly, missense, and nonsense mutations in *MYBPC1* have been directly linked with the development of severe and lethal forms of distal arthrogryposis myopathy and muscle tremors. Currently, there is no mammalian animal model of sMyBP-C, but new technologies including CRISPR/Cas9 and xenografting of human biopsies into immunodeficient mice could provide unique ways to study the regulation and roles of sMyBP-C in health and disease.

Keywords: MyBP-C slow, MYBPC1, actomyosin crossbridges, phosphorylation, distal arthrogryposis myopathy

## INTRODUCTION

Myosin Binding Protein-C (MyBP-C) comprises a family of accessory proteins expressed in striated muscles, and constitutes 2–4% of the myofibrillar protein mass (Okagaki et al., 1993; Moss et al., 2015) There are three MyBP-C isoforms encoded by different genes; slow (s) skeletal MyBP-C is encoded by MYBPC1 present in human chromosome 12, fast (f) skeletal MyBP-C is encoded by MYBPC2 present in human chromosome 19, and cardiac (c) MyBP-C is encoded by MYBPC3 present in human chromosome 11 (Weber et al., 1993). While cMyBP-C is selectively expressed in cardiac muscle, fMyBP-C and sMyBP-C co-exist in fast and slow twitch muscles at varying amounts (Ackermann and Kontrogianni-Konstantopoulos, 2011b). The three isoforms share structural and sequence homology, primarily consisting of immunoglobulin (Ig), and fibronectin-III (Fn-III) domains, referred from the NH2-terminus to the COOH-terminus as C1-C10; notably, the cardiac isoform contains an additional Ig domain, termed C0 (**Figure 1**; Ackermann and Kontrogianni-Konstantopoulos, 2010).

#### Edited by:

*Jose Renato Pinto, Florida State University, USA*

#### Reviewed by:

*Stuart Campbell, Yale University, USA Ranganath Mamidi, Case Western Reserve University, USA*

\*Correspondence:

*Aikaterini Kontrogianni-Konstantopoulos akontrogianni@som.umaryland.edu*

### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *25 July 2016* Accepted: *31 August 2016* Published: *14 September 2016*

#### Citation:

*Geist J and Kontrogianni-Konstantopoulos A (2016) MYBPC1, an Emerging Myopathic Gene: What We Know and What We Need to Learn. Front. Physiol. 7:410. doi: 10.3389/fphys.2016.00410*

**49**

MyBP-C interacts directly with both thick and thin filaments via its NH2- and COOH-termini. The NH2-terminal domains C1-M-C2 bind to myosin subfragment 2 (S2) (Gruen et al., 1999), while the COOH-terminal C10 domain binds to light meromyosin (LMM) (Okagaki et al., 1993; Harris et al., 2011). The latter interaction is enhanced by binding of domains C8-C10 to titin that is likely instrumental in the periodic arrangement of MyBP-C in the C-zone (Craig and Offer, 1976; Freiburg and Gautel, 1996; Luther et al., 2008; Luther and Craig, 2011). In addition to binding to S2, the NH2-terminal C0-C1-M-C2 region contains (relatively weak) binding sites for actin (Kulikovskaya et al., 2003; Whitten et al., 2008; Ackermann et al., 2009; Shaffer et al., 2009; Orlova et al., 2011; Bhuiyan et al., 2012), although recently domains C6-C10 were suggested to mediate high-affinity binding to actin (Rybakova et al., 2011). The interactions at the NH2-terminus of MyBP-C are highly dynamic and regulated via phosphorylation (**Figure 1**; Gruen et al., 1999; Sadayappan et al., 2005; Shaffer et al., 2009; Barefield and Sadayappan, 2010). Accordingly, phosphorylation of cMyBP-C within the M-motif accelerates contraction by disrupting binding to myosin, thereby increasing the probability of binding to actin and thus the rate of force development (Sadayappan and de Tombe, 2012; Walcott et al., 2015; Colson et al., 2016; Moss, 2016; Previs et al., 2016).

Although several questions remain unanswered regarding the (patho)physiology of MyBP-C, early and recent work (primarily on the cardiac isoform) has postulated that through its direct binding to both actin and myosin filaments, it contributes to their assembly and stabilization, modulates the cycling of actomyosin crossbridges, and regulates the ATPase activity of myosin (Martyn, 2004; McClellan et al., 2004; de Tombe, 2006; Oakley et al., 2007; Ackermann and Kontrogianni-Konstantopoulos, 2010, 2013; James and Robbins, 2011; Rybakova et al., 2011; Ackermann et al., 2013). Below, we summarize old and new information on slow skeletal MyBP-C, highlight its direct involvement in disease pathogenesis and provide a perspective on its roles and regulation.

### sMyBP-C: A COMPLEX SUB-FAMILY OF PROTEINS REGULATED BY PHOSPHORYLATION

MyBP-C proteins are highly modular consisting of tandem immunoglobulin (Ig) and fibronectin-III (Fn-III) modules interspersed with unique short amino acid segments (**Figure 1**; Winegrad, 1999). Ig domain C1 is flanked by a ∼50 amino acids long Proline/Alanine-rich motif (Pro/Ala-rich motif) and a ∼100 amino acids long MyBP-C specific motif, referred to as M-motif (Craig et al., 2014). Single transcripts have been identified for the mammalian cardiac and fast isoforms, which result in proteins of ∼140 and ∼130 kDa, respectively (Yasuda et al., 1995). sMyBP-C is unique, however, as there are several mammalian variants that have been characterized ranging in size from 126 to 131 kDa (Ackermann et al., 2015b). This size variability results from extensive splicing of small amino acid segments within the Pro/Ala-rich motif, the M-motif, Fn-III domain C7, and the extreme COOH-terminus (Ackermann and Kontrogianni-Konstantopoulos, 2011b). The different sMyBP-C variants are co-expressed in different amounts and combinations in both slow- and fast-twitch skeletal muscles were they co-exist with fMyBP-C (Ackermann and Kontrogianni-Konstantopoulos, 2011b, 2013).

FIGURE 1 | Schematic representation of the three MyBP-C isoforms. Dark and light gray lines correspond to the Pro/Ala-rich region and the M-motif, respectively, while white and gray rectangles represent Ig and Fn-III domains, respectively. Vertical colored boxes in the Pro/Ala-rich region, the M-motif, Fn-III domain C7 and the extreme COOH-terminus of sMyBP-C indicate alternative spliced segments. fMyBP-C and cMyBP-C share a conserved linker region between C4 and C5, denoted in light green. cMyBP-C contains an additional Ig domain, C0, and an isoform-specific insertion in C5 shown in light blue. Phosphorylation sites in the Pro/Ala-rich region and the M-motif of sMyBP-C are indicated in black, and myopathic mutations in sMyBP-C and fMyBP-C in the M-motif and Ig domains C2 and C8 are shown in red.

Phosphorylation of cMyBP-C contributes significantly to contractile regulation (Sadayappan et al., 2005; Stelzer et al., 2006; Gresham et al., 2014; Gupta and Robbins, 2014; Gresham and Stelzer, 2016; Mamidi et al., 2016; Moss, 2016; Previs et al., 2016). Contrary to early studies suggesting that sMyBP-C is not subjected to phosphorylation (Gruen et al., 1999), work from our group demonstrated that similar to its cardiac counterpart, sMyBP-C also undergoes phosphorylation (Ackermann and Kontrogianni-Konstantopoulos, 2011a). Interestingly, while phosphorylation of cMyBP-C is restricted to the M-motif, phosphorylation of sMyBP-C takes place primarily in the Pro/Ala-rich motif and to a lesser extent in the M-motif (Ackermann and Kontrogianni-Konstantopoulos, 2011a). In particular, proteomics studies confirmed by the use of phospho-specific antibodies demonstrated that in the Pro/Ala-rich motif Ser-59 and Ser-62 are substrates of PKA, and Thr-84 is substrate of PKC, while in the M-motif Ser-204 is substrate of both PKA and PKC (**Figure 1**). Ser-59 and Ser-204 reside in alternatively spliced exons 5 and 10, respectively, and are therefore present in select slow variants (Ackermann and Kontrogianni-Konstantopoulos, 2013). Consistent with a purported important role of phosphorylation in the regulation of MyBP-C, the phosphorylation levels of sMyBP-C are differentially altered in relation to different (patho)physiological stressors. Accordingly, the phosphorylation levels of sMyBP-C are significantly reduced in fast-twitch Flexor Digitorum Brevis (FDB) muscle as a function of aging and dystrophy (Ackermann et al., 2015b). Similarly, the phosphorylation levels of sMyBP-C are notably decreased in slow-twitch soleus muscle as a result of aging and dystrophy, but increased in response to fatigue (Ackermann et al., 2015a). Although these observations are interesting, a detailed examination of the effects of individual or combinatorial phosphorylation events in the modulation of the structural and regulatory activities of sMyBP-C is currently lacking. It is therefore expected that future endeavors combining sophisticated in vitro and in vivo approaches will shed light on the role of phosphorylation in the modulation of sMyBP-C.

### MYBPC1: A RECENT MYOPATHIC GENE

MYBPC3 has garnered much attention over the past several decades due to its prevalent mutational rate leading to congenital hypertrophic and dilated cardiomyopathy (Harris et al., 2011; Santos et al., 2012; Kuster and Sadayappan, 2014; Lynch et al., 2015). It is only recently, however, that mutations in MYBPC1 have been directly associated with inherited myopathies, and specifically with severe and lethal forms of distal arthrogryposis myopathy (**Table 1**) (Markus et al., 2012; Ha et al., 2013; Li et al., 2015). Contrary to MYBPC3 mutations that mainly result in truncated proteins and function via haploinsufficiency (Marston et al., 2012; Kuster and Sadayappan, 2014; Barefield et al., 2015; Carrier et al., 2015), the currently known MYBPC1 mutations have been suggested to result in poisonous proteins and manifest in a dominant negative manner after incorporation into sarcomeres (Markus et al., 2012; Ha et al., 2013; Li et al., 2015).

Arthrogryposis, also known as arthrogryposis multiplex congenita, is clinically defined by congenital joint contractures or movement restriction in multiple body areas (Bayram et al., 2016). Generally, arthrogryposis occurs as a secondary effect of decreased fetal joint mobility, which can result from abnormalities of the central nervous system, the neuromuscular system, the skeletal system, or connective and cartilage tissue disturbances. (Filges and Hall, 2013; Haliloglu and Topaloglu, 2013; Hall, 2014; Bayram et al., 2016). Distal arthrogryposis (DA) myopathies are a group of autosomal dominant arthrogryposis disorders that mainly involve the distal parts of the limbs (Bamshad et al., 2009). Ten different types of DA have been described to date that share common general features, including a consistent pattern of hand and foot defects, limited involvement of proximal joints and variable expressivity (Bamshad et al., 1996; Krakowiak et al., 1998; Stevenson et al., 2006).

DA type-1 (DA-1) is the most common DA myopathy that affects approximately 1 in 10,000 individuals, and results in contractures limited to the distal muscles of the hands and feet. These include clubfoot, vertical talus, camptodactyly, overriding fingers and ulnar deviations of the fingers with no additional anomalies (Hall, 1985; Klemp and Hall, 1995; Gurnett et al., 2010). DA type-2 (DA-2) is a more severe form of DA, also characterized by contractures of the hands and feet, that is often accompanied by mild to severe craniofacial anomalies and/or scoliosis (Kulkarni et al., 2008; Bamshad et al., 2009). There are two subtypes of DA-2, including DA-2A (Freeman-Sheldon syndrome) and DA-2B (Sheldon-Hall syndrome) (Kulkarni et al., 2008; Bamshad et al., 2009; Li et al., 2015). While individuals with DA-2B Sheldon-Hall syndrome display mild to moderate facial contractures, individuals with DA-2A Freeman-Sheldon syndrome have moderate to severe facial contractures (Beck et al., 2013; Li et al., 2015).

In the last 5 years, dominant missense mutations in MYBPC1 have been linked to the development of both DA-1 and DA-2 (Gurnett et al., 2010; Li et al., 2015). Specifically, missense mutations, W236R and Y856H, located in the M-motif and Ig domain C8, respectively, have been linked to DA-1 (**Figure 1**; Gurnett et al., 2010). Both of these substitutions are present in constitutively expressed exons and thus are contained in all slow variants (Gurnett et al., 2010; Ackermann et al., 2015b). ATPase staining of biopsies obtained from the distal Abductor Hallucis (AH) foot muscle of DA-1 patients carrying either mutation revealed severe type-I fiber atrophy, although localization of the mutant proteins was unaltered (Gurnett et al., 2010). In vitro binding and actin sliding assays demonstrated that the presence of the W236R and Y856H mutations markedly diminished the ability of the NH<sup>2</sup> and COOH termini of sMyBP-C, respectively, to bind actin and myosin, and regulate the formation of actomyosin crossbridges (Ackermann et al., 2013). Examination of the expression levels of mutant sMyBP-C that contained the Y856H or the W236R mutation in human biopsies of AH or gastrocnemius muscles, respectively, revealed that the total amounts of the protein were significantly reduced in AH (∼25%), but not gastrocnemius, muscle compared to controls (Ackermann et al., 2015b). Although puzzling, since both DA-1 mutations reside in constitutive exons, this finding



is in agreement with the selective effects of DA-1 on distal muscles, and the lack of a myopathic phenotype in proximal muscles. Interestingly, the phosphorylation profile of mutant sMyBP-C containing the Y856H mutation was also altered in the affected AH muscle, whereas the phosphorylation profile of sMyBP-C carrying the W236R mutation was unchanged in gastrocnemius muscle (Ackermann et al., 2015b). Accordingly, use of a panel of phospho-specific antibodies and phos-tag gel electrophoresis revealed that mutant sMyBP-C harboring the Y856H mutation in AH muscle was phosphorylated at all four known residues, however the extent of phosphorylation was decreased by 30–70% for individual phospho-sites, compared to control tissue (Ackermann et al., 2015b).

Recently, two novel autosomal dominant missense mutations in Ig domain C2 of sMyBP-C, P319L and E359K, were linked to the development of DA-2 (**Figure 1**) (Li et al., 2015). Although a mechanistic understanding of the effect(s) of these mutations is still lacking, it is tempting to speculate that they may affect binding to the S2 portion of myosin and/or actin via induction of an unfavorable conformation (P319L) or altered electrostatic interactions (E359K). Future studies using a combination of biochemical, structural, biophysical and in vivo approaches will address these hypotheses.

In addition to DA-1 and DA-2, MYBPC1 has been directly linked to the development of a neonatal lethal form of arthrogryposis myopathy, referred to as Lethal Congenital Contracture Syndrome type-4 (LCCS-4; Markus et al., 2012). Specifically, an autosomal recessive nonsense mutation in Ig domain C2 of sMyBP-C results in a premature stop codon at amino acid 318 (R318Stop; **Figure 1**). Given the recessive inheritance of LCCS-4, along with the absence of any phenotypic or functional abnormalities in the heterozygous carriers, it is highly likely that the R318Stop mutation results in loss of sMyBP-C rather than a poisonous truncated protein. Nevertheless, if the mutant protein is indeed expressed, it will lack domains C3-C10 downstream of Ig C2, which contain binding sites for LMM, titin and obscurin (Okagaki et al., 1993; Freiburg and Gautel, 1996; Ackermann et al., 2009).

In addition to mutations in the human MYBPC1 gene that are associated with severe and lethal forms of DA, a new mutation in the bull MYBPC1 gene was recently identified, too (Wiedemar et al., 2015). In particular, a 2-week old female calf presented with muscle tremors since birth, standing difficulty and reduced spinal reflexes (Wiedemar et al., 2015). Whole genome sequencing analysis revealed a de novo missense mutation, L295R, localized in the M-motif following the Ig domain C1, similar to the human W236R mutation. Although the phenotypic manifestation of the L295R mutation is reminiscent of DA-1, it is further accompanied by muscle tremors, which is indicative of a more complex and/or severe myopathy. At this time, a mechanistic understanding of the effects of the L295R mutation is lacking.

As novel mutations in MYBPC1 are being identified in the mammalian genome underscoring its role in skeletal muscle (patho)physiology, it is worth mentioning that recently MYBPC2, encoding fMyBP-C, was also linked to an unclassified, neonatal lethal DA in the form of a compound heterozygote (Bayram et al., 2016). Specifically, a patient presenting with narrow thorax, polyhydramnios during fetal development, and neonatal death was found to possess two missense mutations in MYBPC2, T236I and S255T, located in the M-motif. The same patient also contained an R7Stop homozygous mutation in the GPR126 gene, which encodes a G-protein coupled receptor that regulates neural, cardiac, and ear development (Patra et al., 2014; Bayram et al., 2016). Although mutations in GPR126 have been linked with isolated arthrogryposis multiplex congenital (Ravenscroft et al., 2015), it is likely that the additional mutations in MYBPC2 contributed to the postnatal lethality of the carrier due to accumulating anomalies in motor neurons and muscle structure and function (Bayram et al., 2016).

Although at the current time limited, the above studies clearly indicate that mutations in the genes that encode the skeletal MyBP-C proteins (and especially the slow isoform) are intimately associated with the development of severe and lethal myopathies. Obviously, the challenge now lies in deciphering the cell processes that are altered or compromised due to individual mutations using sophisticated and high resolution in vitro approaches and appropriate in vivo models.

### IN VIVO MODELS OF MYBPC1: PERSPECTIVES AND ENDEAVORS

Over the last four decades, a tremendous emphasis has been placed on the regulation and roles of cMyBP-C due to its direct involvement in congenital heart disease resulting in the generation of multiple animal models (Harris et al., 2002; Sadayappan et al., 2005; Carrier et al., 2015). This is not the case for sMyBP-C (or fMyBP-C). Remarkably though, the direct association of MYBPC1 with the development of severe and lethal forms of DA has tunneled the interest of the scientific community toward the molecular and functional characterization of MYBPC1, too.

A recent study used the zebrafish model and antisense morpholinos to down-regulate the expression of MYBPC1 (Ha et al., 2013). Knock-down zebrafish exhibited severe ventral body curvature, decreased mobility, and early lethality, along with impaired sarcomeric development and reduced number of myofibrils (Ha et al., 2013). Moreover, overexpression of mutant sMyBP-C proteins carrying either of the DA-1 mutations, W236R or Y856H, in zebrafish demonstrated that both mutant proteins exerted a dominant negative effect, resulting in embryos with mild bent body curvature, impaired mobility, and muscles with less tightly compacted fibers compared to controls (Ha et al., 2013).

Apart from the zebrafish model, no mammalian MYBPC1 animal models have been generated yet. Our group has been systematically working on sMyBP-C for the last 5 to 6 years focusing on its molecular characterization, regulation and functional evaluation. Given the unique complexity of MYBPC1 (Ackermann and Kontrogianni-Konstantopoulos, 2010, 2011b, 2013), its early expression during fetal development preceding that of MYBPC2 (Gautel et al., 1998; Kurasawa et al., 1999), and the neonatal lethality of LCCS-4 patients, who most likely lack sMyBP-C, we predict that a constitutive MYBPC1 null model would be postnatal lethal. If this is the case, such a model is still worth generating, since it will highlight the non-redundant roles of sMyBP-C and fMyBP-C, and will allow the study of the structural and regulatory roles of sMyBP-C in myofibrillar assembly and contractility during embryogenesis and (early) postnatal life. Obviously, conditional null models would circumvent the potential neonatal lethality of a constitutive MYBPC1 knock-out allowing the detailed investigation of the roles of sMyBP-C in modulating actomyosin contractility in mature muscles. Along the same lines, knock-in models carrying the DA-1, DA-2, or LSSC-4 mutations are also lacking, limiting our understanding of the effects of the respective mutations to in vitro studies, which although informative, need to be accompanied by in vivo data.

Type II bacterial Clustered Regularly Interspaced Short Palindromic Repeats-associated protein Cas9 (CRISPR-Cas9) mediated genome editing has emerged as a powerful tool for genetic manipulation. Unlike small interfering RNAs (siRNAs) or short hairpin RNAs (shRNAs), the CRISPR-Cas9 system is able to knockout individual gene expression at the genomic level with minimal off-target effects (Zhang et al., 2014; Humphrey and Kasinski, 2015). Conversely, CRISPR/Cas9 technology is also being refined for generating knock-in models to recapitulate disease development (Chu et al., 2015; Tu et al., 2015). As the technology becomes increasingly popular, the generation of constitutive or conditional MYBPC1 knock-out and knock-in mouse models should be feasible in a fairly short amount of time.

In vivo gene transfer (IVGT) followed by electroporation is also an efficient, non-viral method for gene delivery that has been successfully used by several groups (Gehl, 2003; Spanggaard et al., 2012; Hu et al., 2014). Although the mechanisms underlying DNA electrotransfer are not yet fully elucidated, it has been suggested that permeabilization of the cell membrane as well as electrophoretic migration of the DNA are involved (Mir et al., 1999; Bureau et al., 2000; Golzio et al., 2002; Satkauskas et al., 2002). Muscle tissue is a favorable target for gene electrotransfer as it is easily accessible, allowing high-level, longterm transgenic expression (Mir et al., 1999; Lucas and Heller, 2001; Spanggaard et al., 2012). Such an approach could therefore be employed to knock-down (via shRNA technology) or knockout (via CRISPR-Cas9 technology) MYBPC1 in a muscle-specific manner, circumventing the potential neonatal lethality of a constitutive null model, and enabling the functional examination of different slow variants given that they are expressed in distinct combinations among skeletal muscles. Moreover, IVGT combined with electroporation could be used to overexpress myopathic forms of sMyBP-C. Although the stoichiometry of endogenous to exogenous proteins is an important issue to consider, it could potentially be alleviated by inducible, titratable expression systems. Notably, IVGT experiments are particularly beneficial as the effects of gene knock-down, knock-out or knockin experiments can be analyzed from the single fiber to the whole animal level. For instance, many groups have taken advantage of permeabilized muscle fiber preparations of human biopsies or pre-clinical mouse models to elucidate the (patho)physiology of cMyBP-C (Harris et al., 2002; Stelzer et al., 2006; James and Robbins, 2011; Wang et al., 2016).

An alternative to generating animal models of muscle disease has emerged in the last few years, entailing the development and propagation of grafts of myopathic or dystrophic human muscle tissue in mice (Riederer et al., 2012; Meng et al., 2014; Sakellariou et al., 2015). This approach has tremendous benefits, given that animal models often fail to replicate the features of human muscle disease. Along these lines, a recent study reported the generation of xenografts from human bicep muscle biopsies of facioscapulohumeral muscular dystrophy (FSHD) patients that were transplanted into the hindlimbs of immunodeficient NOD-Rag1nullIL2rynull mice (Sakellariou et al., 2015). The engrafted human muscle was efficiently regenerated and innervated, and displayed normal contractile properties (Sakellariou et al., 2015). While the xenografting model approach is still being perfected, the largest hurdle is the unavailability of fresh muscle biopsies and the lack of organized biobanks; obviously, this is a major issue that applies to MYBPC1 related myopathies, as well. Nevertheless, the xenografting model could prove to be an extremely useful tool for studying the effects of human myopathies in vivo, since it may recapitulate the course of disease development more faithfully compared to engineered models of C. elegans, zebrafish, or mouse that are commonly used to date.

### CONCLUSIONS

Slow skeletal MyBP-C has recently attracted considerable interest primarily due to its direct involvement in the development of severe and lethal forms of distal arthrogryposis myopathy. Contrary to the fast and cardiac isoforms, sMyBP-C comprises

### REFERENCES


a subfamily of proteins with possibly distinct structural and regulatory roles, which are modulated by constitutive and variant-specific phosphorylation events. Given the recent involvement of MYBPC1 in severe and lethal myopathies, we predict that a comprehensive, multidisciplinary evaluation of its regulation and roles in health and disease is in order.

### AUTHOR CONTRIBUTIONS

JG and AK-K drafted, revised and approved the final version of the manuscript.

### FUNDING

This work was supported by NIH/NIAMS (Training Program in Muscle Biology, T32 AR007592-17 to JG), and the Muscular Dystrophy Association (Research Grant 313579 to AK-K).


familial hypertrophic cardiomyopathy. Eur. J. Biochem. 235, 317–323. doi: 10.1111/j.1432-1033.1996.00317.x


Freeman-Sheldon syndrome. Pediatrics 117, 754–762. doi: 10.1542/peds.2005- 1219


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Geist and Kontrogianni-Konstantopoulos. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Modeling Ca2+-Bound Troponin in Excitation Contraction Coupling

Henry G. Zot <sup>1</sup> \* and Javier E. Hasbun<sup>2</sup>

*<sup>1</sup> Department of Biology, University of West Georgia, Carrollton, GA, USA, <sup>2</sup> Department of Physics, University of West Georgia, Carrollton, GA, USA*

To explain disparate decay rates of cytosolic Ca2<sup>+</sup> and structural changes in the thin filaments during a twitch, we model the time course of Ca2+-bound troponin (Tn) resulting from the free Ca2<sup>+</sup> transient of fast skeletal muscle. In fibers stretched beyond overlap, the decay of Ca2<sup>+</sup> as measured by a change in fluo-3 fluorescence is significantly slower than the intensity decay of the meridional 1/38.5 nm−<sup>1</sup> reflection of Tn; this is not simply explained by considering only the Ca2<sup>+</sup> binding properties of Tn alone (Matsuo et al., 2010). We apply a comprehensive model that includes the known Ca2<sup>+</sup> binding properties of Tn in the context of the thin filament with and without cycling crossbridges. Calculations based on the model predict that the transient of Ca2+-bound Tn correlates with either the fluo-3 time course in muscle with overlapping thin and thick filaments or the intensity of the meridional 1/38.5 nm−<sup>1</sup> reflection in overstretched muscle. Hence, cycling crossbridges delay the dissociation of Ca2<sup>+</sup> from Tn. Correlation with the fluo-3 fluorescence change is not causal given that the transient of Ca2+-bound Tn depends on sarcomere length, whereas the fluo-3 fluorescence change does not. Transient positions of tropomyosin calculated from the time course of Ca2+-bound Tn are in reasonable agreement with the transient of measured perturbations of the Tn repeat in overlap and non-overlap muscle preparations.

Keywords: contraction, calcium, troponin, excitation, muscle, EC-coupling, model, kinetics

## INTRODUCTION

During a twitch of striated muscle, the intracellular fluorescence probe fluo-3 reveals two types of calcium transients (Minta et al., 1989). Owing to its high affinity for Ca2+, fluo-3 detects and contributes to a cytosolic pool of Ca2<sup>+</sup> that rises rapidly and persists longer than 150 ms after stimulation of frog striated muscle at 16◦C (Harkins et al., 1993; Caputo et al., 1994; Matsuo et al., 2010). The long decay time can be explained by Ca2<sup>+</sup> exchange with binding molecules in the cytosol, which may be immobilized molecules such as troponin (Tn) or diffusive molecules such as ATP and parvalbumin, during sequestration of Ca2<sup>+</sup> by the sarcoplasmic reticulum (Baylor and Hollingworth, 2011). A transient of free Ca2+, well described by the low-affinity probe furaptra (Hollingworth et al., 2009), can also be calculated from the fluo-3 record (Caputo et al., 1994). The transient of free Ca2<sup>+</sup> rises to a peak in 5–7 ms and decays to baseline in about 50 ms at 16◦C (Konishi et al., 1991; Hollingworth and Baylor, 2013). This brief pulse of Ca2<sup>+</sup> produced by Ca <sup>2</sup><sup>+</sup> sparks (Cannell et al., 1995) constitutes the intracellular excitation signal for myofilament contraction (Baylor et al., 2002; Baylor and Hollingworth, 2007).

### Edited by:

*P. Bryant Chase, Florida State University, USA*

#### Reviewed by:

*Corrado Poggesi, University of Florence, Italy John Jeshurun Michael, Cornell University, USA*

### \*Correspondence:

*Henry G. Zot hzot@westga.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *25 June 2016* Accepted: *30 August 2016* Published: *21 September 2016*

#### Citation:

*Zot HG and Hasbun JE (2016) Modeling Ca2*+*-Bound Troponin in Excitation Contraction Coupling. Front. Physiol. 7:406. doi: 10.3389/fphys.2016.00406*

**57**

The Ca2<sup>+</sup> regulatory sites of Tn (Potter and Gergely, 1975) mediate excitation contraction coupling (Robertson et al., 1981). Based on the binding characteristics of purified Tn (Baylor and Hollingworth, 1998), the decay of Ca2+-bound Tn is expected to follow the long process of fluo-3 decay (Matsuo and Yagi, 2008). However, a simple Ca2<sup>+</sup> dissociation rate of Tn is difficult to reconcile with the modeling of the furaptra transient (Baylor et al., 2002). In muscle preparations stretched to average sarcomere lengths of 2.8 and 4.0µm (overlap and non-overlap preparations, respectively), decays of fluo-3 signals are remarkably similar, but, in the non-overlap preparation, the intensity of the meridional 1/38.5 nm−<sup>1</sup> reflection corresponding to the repeat of Tn in the thin filament decays significantly faster than the fluo-3 signal (Matsuo and Yagi, 2008). If Ca2+-bound Tn is a function of the pool of Ca2<sup>+</sup> represented by the fluo-3 signal then a significant fraction of Tn remains in the Ca2+ bound state after the Tn-related structure fully relaxes (Matsuo and Yagi, 2008). Our aim is to provide a theoretical framework for the alternative hypothesis, namely, the structure represented by the meridional 1/38.5 nm−<sup>1</sup> reflection has a direct relationship with Ca2+-bound Tn.

The properties of Ca2<sup>+</sup> binding to the regulatory sites of the C-subunit of Tn (TnC) depend on interactions of Tn with actin in a 7:1:1 molar complex of actin, tropomyosin (Tm), and Tn (regulated actin). Studies using both <sup>45</sup>Ca2<sup>+</sup> and fluorescence change techniques with native and covalently modified preparations, respectively, consistently demonstrate that the Ca2<sup>+</sup> affinity of regulated actin is substantially lower than the Ca2<sup>+</sup> affinity of isolated Tn (Wnuk et al., 1984; Rosenfeld and Taylor, 1987; Zot, H. G. and Potter, J. D., 1987). Kinetic measurements of Ca2+-dependent fluorescence changes show slow and fast rates of Ca2<sup>+</sup> dissociation from regulated actin; the slow rate correlates with the Ca2<sup>+</sup> dissociation rate of isolated Tn, while the other rate is about 10-fold faster (Rosenfeld and Taylor, 1987). Rigor myosin shifts the affinity of regulated actin (myosin:actin:Tm:Tn in a 7:7:1:1 complex with no ATP) to the higher Ca2<sup>+</sup> affinity of isolated Tn and reduces the kinetic measurement to one rate, which matches the slow rate of Ca2<sup>+</sup> dissociation (Rosenfeld and Taylor, 1987). Tropomyosin can occupy three different positions relative to actin: blocking (B), central (C), and myosin dependent (M) positions. Tn in association with Tm can interact with actin only when Tm is in position B (Lehman et al., 2000). A competition between the open conformation of TnC and actin for the same internal structure of Tn in position B (Gagné et al., 1995; Takeda et al., 2003) could lower the apparent Ca2+affinity and increase the Ca2<sup>+</sup> off rate of Tn in position B by energy coupling. By the same energetic principle, when Tm is in either position C or M and Tn cannot interact with actin, the regulatory sites of TnC should have the higher Ca2<sup>+</sup> affinity and slower Ca2<sup>+</sup> off rate of isolated Tn.

Cooperative changes associated with Ca2<sup>+</sup> binding to TnC depend on not only the context of regulated actin but also the context of rigor and steady-state conditions. Although, some preparations of fluorescently modified TnC display cooperative Ca2+-dependent fluorescence changes (Grabarek et al., 1983; Zot, H. G. and Potter, J. D., 1987; Davis et al., 2002), only a single class of non-interacting Ca2+-binding sites is found for the regulatory sites of native and fluorescently modified TnC in regulated actin by techniques using <sup>45</sup>Ca2<sup>+</sup> and fluorescence change, respectively (Wnuk et al., 1984; Rosenfeld and Taylor, 1987; Zot, H. G. and Potter, J. D., 1987). Likewise, a noncooperative fluorescence change in response to Ca2<sup>+</sup> is observed for regulated actin saturated with rigor myosin (Rosenfeld and Taylor, 1987). However, in the presence of ATP, muscle fibers and myofibrils reconstituted with fluorescent TnC display steeply cooperative Ca2+-dependent activation and fluorescence changes (Zot et al., 1986; Zot, A. S. and Potter, J. D., 1987; Brandt and Poggesi, 2014). Hence, cooperative Ca2<sup>+</sup> binding requires steady-state crossbridges.

Here we link the well-described transient of free Ca2<sup>+</sup> to a comprehensive model of contraction (Zot et al., 2009). This model accounts for Ca2+-bound Tn in association with Tm in the three principle structural states of the thin filament (Lehman et al., 2000). As with regulated actin, the muscle fiber is expected to display both slow and fast Ca2<sup>+</sup> dissociation rates, which should be evident in the decay rates of structural changes related to Tn and also depend on cycling crossbridges. We apply the model to transient changes in the fluo-3 fluorescence and meridional 1/38.5 nm−<sup>1</sup> reflection intensities measured in preparations of frog skeletal muscle at 16◦C, with the sarcomere length maintained at overlap or non-overlap of myofilaments (Matsuo et al., 2010), which promotes or prohibits cycling crossbridges, respectively. The model presented here predicts that Ca2+-bound Tn follows the slow decays of fluo-3 fluorescence and meridional 1/38.5 nm−<sup>1</sup> reflection intensities of the overlap preparation and only the faster decay of meridional 1/38.5 nm−<sup>1</sup> reflection intensity of the non-overlap preparation. The pool of Ca2<sup>+</sup> represented by the fluo-3 fluorescence intensity and Ca2+-bound Tn lack a predictable relationship.

### MATERIALS AND METHODS

### Description of Model

The model we employ accounts for the relative distributions of thin filament states (**Figure 1**). The B, C, and M states of Tm refer to Tm's interactions with actin in these respective positions (Lehman et al., 2000). State C is the equilibrium position (Phillips et al., 1986; Lehman et al., 2000), and states B and M are modeled as competing for Tm in state C. To stabilize the non-equilibrium positions B and M, Tm-bound Tn forms an interaction with actin (Greaser and Gergely, 1973) in position B, and Tm forms a ternary complex with crossbridges and actin in position M (Eaton, 1976; Tobacman and Butters, 2000). The interaction of Tn in state B accounts for the states of Tn that are energetically coupled to the states of Tm. Movement of Tm away from B energetically uncouples Tn from possible interactions with actin (**Figure 1**).

Coupled and uncoupled states of Tn are designated B and T, respectively (**Figure 1**). Calcium-dependent states of Tn are designated B<sup>i</sup> and T<sup>i</sup> , where i represents 1 (Ca2<sup>+</sup> free), 2 (singly bound), or 3 (doubly bound). Affinity for actin is progressively reduced as i increases. Based on conservation of the mass for two

**Abbreviations:** Troponin, Tn; tropomyosin, Tm; B, state of Tm in the blocking position; C, state of Tm in the central position; M, state of Tm in the myosindependent position.

Ca2+-bound (*i* = 2; black), and doubly Ca2+-bound (*i* = 3; black). The diagram shows how Tn is isolated from a site of interaction with actin (gray oval) by the movement of Tm from position *B* to positions *C* and *M*. All constants depicted in the figure represent the ratio of component rate constants (Table 1) used to calculate the relative abundance or probability of each state as a function of a transient change in calcium concentration.

partially overlapping subsystems (Tm and Tn), B + C + M = 1 for Tm and B + T = 1 for Tn.

In the system we describe, M can arise by either rigor or cycling crossbridges, but cooperativity derives solely from cycling crossbridges, as seen in the records of Ca2<sup>+</sup> binding measurements. With overlapping thin and thick filaments and ATP, M is a state in constant flux (steady-state) rather than at equilibrium; this is readily observed by Ca2+-dependent in vitro sliding of regulated actin filaments. Steady-state cooperativity is achieved if crossbridge turn-over generates additional opportunities (second chances) for M formation (Zot et al., 2009, 2012). By analogy, the M state operates like a man whose feet are bound to a ceiling by adhesion: changing positions quickly relative to the rate of deadhesion improves the odds of remaining bound by re-establishing the initial binding conditions. A statistical treatment of a second chance mechanism applied to data from biological systems is available (Zot et al., 2016a). In practice, a second chance mechanism is given in the following rate equation

$$dM/dt = K\_0^{'}k\_{-0}C(1 + (\alpha - 1)M)^n - k\_{-0}M\tag{1}$$

where parameters K ′ 0 , k−0, α, and n are derived elsewhere (Zot et al., 2009, 2012). The parameter α expresses second chance opportunities for reestablishing equilibrium before the decay of M. As steady-state or equilibrium approach, dM/dt → 0. Because Equation (1) acts on the C-M transition (**Figure 1**), cooperativity does not directly involve Ca2<sup>+</sup> binding to Tn. Although, Equation (1) performs adequately, any logistic function operating on C can be compatible with our model.

As applied to transient and steady-state striated muscle regulation, the parameters of Equation (1) may have the following interpretations. The equilibrium potential of M as a function of the population of strong binding myosin at any moment of steady-state is expressed by K ′ 0 . The forward rate of ensemble M formation is the product K ′ 0 k−0. Ensemble size, n, expresses the number of Tm subunits acting in concert to form a ternary complex as described above. The orchestrating event could be a lateral stretch imposed on contiguous Tm subunits by an axial force acting on the thin filament (Zot et al., 2009). The parameter α is an expression of the crossbridges poised to replace crossbridges disrupted by internal chemomechanical forces or by active sliding. The value of α may be related to the average number of crossbridges in a target zone (Tregear et al., 2004), as has been described (Zot et al., 2009). We give unit value to K ′ 0 for simplicity and recycle previously discussed values for α, and n (Zot et al., 2009; **Table 1**) for consistency.

### Computational Methods

Transitions other than C-M are spontaneous processes governed by simple mass action (**Figure 1**). Although the model has eight states, only six are independent. If we choose to calculate states C and T<sup>1</sup> by mass conservation (see above), the relative abundances or probabilities of the other six states (**Figure 1**) as a function of an independent calcium transient are calculated by solving a system of six ordinary differential equations (ODE), i.e., in addition to Equation (1), we have

$$\begin{aligned} dB\_1/dt &= K\_1k\_{-1}CT\_1 + k\_{-4}B\_2 - \left(k\_{-1} + K\_4k\_{-4}[Ca^{2+}]\right)B\_1; \\ dB\_2/dt &= K\_3k\_{-3}CT\_2 + k\_{-4}B\_3 + K\_4k\_{-4}[Ca^{2+}]B\_1 \\ &- \left(k\_{-3} + k\_{-4} + K\_4k\_{-4}[Ca^{2+}]\right)B\_2; \\ dB\_3/dt &= K\_5k\_{-5}CT\_3 + K\_4k\_{-4}[Ca^{2+}]B\_2 - \left(k\_{-4} + k\_{-5}\right)B\_3; \\ dT\_2/dt &= K\_2k\_{-2}[Ca^{2+}]T\_1 + k\_{-3}B\_2 + k\_{-2}T\_3 \\ &- \left(k\_{-2} + K\_2k\_{-2}[Ca^{2+}] + K\_3k\_{-3}C\right)T\_2; \\ dT\_3/dt &= K\_2k\_{-2}[Ca^{2+}]T\_2 + k\_{-5}B\_3 - \left(k\_{-2} + K\_5k\_{-5}C\right)T\_3. \end{aligned}$$



*<sup>a</sup>Shown previously to fit steady-state Ca2*+*-dependent tension of skinned fibers of fast twitch (Zot et al., 2009) and slow twitch (Zot et al., 2016b) muscle.*

This system of ODE is solved for each free Ca2<sup>+</sup> concentration of a given transient. The free Ca2<sup>+</sup> transient of a muscle fiber (Matsuo et al., 2010) is reproduced by a linear rise from the origin to the peak (time to peak is 0.005 s), followed by an exponential decay (rate constant is 100 s−<sup>1</sup> ; **Figure 2**). We use the same Ca2<sup>+</sup> transient for all calculations as a control. Hence, the model varies only the contribution of crossbridges in fitting data from overlap and non-overlap preparations. A Matlab program is provided to reproduce calculations presented here (see Supplementary Materials).

Standard conditions refer to a set of constants governing steady-state potentials (upper case, "K"), α, and n that we hold constant (**Table 1**). Thermodynamic principles dictate that K1/K<sup>3</sup> = K3/K<sup>5</sup> = K2/K4, which allows K ′ 0 , K1, K2, and K<sup>4</sup> to be selected independently. The values used here for these four parameters, α, and n (**Table 1**) are the same shown elsewhere to fit steady-state activation of skinned fibers of fast muscle (Zot et al., 2009) and cardiac muscle (Zot et al., 2016b) by Ca2+. Component rate constants (lower case, "k") are varied to fit transient data. Slow and fast dissociation rates of Ca2<sup>+</sup> from Tn (k−<sup>2</sup> and k−4; **Table 1**), are taken from Rosenfeld and Taylor (1987).

### RESULTS

### Steady-State and Equilibrium Behavior of the Model

Steady-state or equilibrium calculations are produced for standard conditions (**Table 1**) by nullifying the system of ODE (cf. Matlab program in Zot et al., 2016b). Calculated Ca2+ dependent activation curves on absolute and normalized scales (inset, **Figure 2**) reproduce fits of steady-state tension and ATPase data of diverse native and mutant protein preparations of fast skeletal and cardiac muscles (Zot et al., 2009, 2016b). To verify that the model reproduces established Ca2<sup>+</sup> binding properties of Tn, Ca2+-bound Tn (B<sup>2</sup> + B<sup>3</sup> + T<sup>2</sup> + T3) is calculated as a function of constant concentrations of Ca2+. With cycling crossbridges, the mathematical solution is a cooperative function of Ca2<sup>+</sup> (inset, **Figure 2**), which reproduces the distinctive binding-activation relationship observed previously with fluorescently labeled Tn in reconstituted myofibrils (Zot et al., 1986; Zot, A. S. and Potter, J. D., 1987; Brandt and Poggesi, 2014). To model equilibrium achieved by non-overlap or rigor, K ′ 0 is set either to zero or to a relatively large value (10<sup>6</sup> ), respectively. Both solutions predict simple mass action (non-cooperative) calcium binding (inset, **Figure 2**), and both simulations reproduce published simple mass action Ca2<sup>+</sup> binding curves for regulated actin and regulated actin with rigor myosin binding, respectively (Rosenfeld and Taylor, 1987). Hence, the model is able to reproduce cooperative and noncooperative Ca2<sup>+</sup> binding measurements of both steady-state and equilibrium preparations, respectively.

### Transient Response with Overlap

The transient of calcium-bound troponin, which is modeled as the sum of B2, B3, T2, and T3, is compared with the transient change measured by fluo-3 fluorescence (Matsuo et al., 2010) in overlap muscle preparations. Rather than choosing parameters to fit the fluo-3 data, we fit tension data (**Figure 3**) from the same preparation with calculated value of M (Equation 1) by adjusting the rate constants (**Table 2**) that comprise standard conditions (**Table 1**). Decreasing K ′ 0 k−<sup>0</sup> shifts the calculated tension transient rightward, and k−<sup>0</sup> is decreased in tandem to maintain constant K ′ 0 . The same constraint is used throughout to maintain standard conditions. Although adjusting either K ′ 0 k−<sup>0</sup> or K1k−<sup>1</sup> changes the tension transient equivalently, using K ′ 0 k−<sup>0</sup> for lateral adjustments and K1k−<sup>1</sup> for vertical adjustments yields the best shape of the tension transient relative to the data. Given the optimum fit of the tension data, the predicted Ca2+-bound Tn transient is seen to fit most of the fluo-3 fluorescent data (**Figure 3**). If we accept a slightly faster time to peak tension, the model predicts slower decay of Ca2+-bound Tn, which may better capture the entire trend of fluo-3 data (see Supplementary Materials).

The dissociation of Ca2<sup>+</sup> from the Ca2<sup>+</sup> regulatory sites of Tn is not uniform over time. Owing to a faster off-rate, Ca2+ bound B states (B<sup>2</sup> + B3) release Ca2<sup>+</sup> faster than Ca2+-bound T states (T<sup>2</sup> + T3; **Figure 3**). Hence, a significant fraction of Tn has released Ca2<sup>+</sup> before tension reaches a peak. A protracted dissociation of residual Ca2<sup>+</sup> comes mainly from T states, which represent a pool of Tn molecules held away from interaction with actin in the B position owing to crossbridges (crossbridgedependent Ca2+-bound Tn).

### Transient Response with Non-overlap

The relationship between calculated Ca2+-bound Tn and fluo-3 data differs dramatically in muscle fibers stretched beyond overlap. To simulate no overlap, we nullify K ′ 0 k−<sup>0</sup> in Equation (1), but otherwise preserve the rates established for the overlap condition (**Table 1**). Absent crossbridges, the decays of calculated Ca2+-bound troponin, whether expressed as a total or separated

FIGURE 3 | Time courses of principle states of the thin filament with overlap. Plotted as a function of a pulse of free calcium (green) are calculated transients of states *<sup>B</sup>* (dark blue), *<sup>C</sup>* (red), and *<sup>M</sup>* (black). Total Ca2+-bound Tn (*B*2+*B*3+*T*2+*T*3; gold) is broken down into components, i.e., fast (*B*2+*B*3; blue) and slow (*T*2+*T*3; gray). Plotted on the same scale are measured tension (circles) and fluo-3 fluorescence (squares) transients of muscle fibers of average sarcomere length of 2.8µm and stimulated by a single impulse (reproduced from Matsuo et al., 2010). Adjustments in standard conditions (Table 1) alter the width (1*x*M), height (1*y*M), and center of the *M* peak (O) as described in Table 2.

into components B and T, are much faster than the decay of fluo-3 fluorescence (**Figure 4**). No combinations of rate constants will allow the model to fit the tension data of the overlap preparation and the fluo-3 data of overlap and non-overlap preparations. By contrast, the decay of fluo-3 fluorescence is constant for overlap and non-overlap conditions (**Figures 3**, **5**). Therefore, there is not a causal relationship between the calculated Ca2+-bound state of troponin and fluo-3 fluorescence.

Matsuo et al. (2010) show that in a non-overlap sarcomere preparation, the fluo-3 fluorescence decay is slower than the

TABLE 2 | Response to adjustments of standard conditions.


decay of meridional 1/38.5 nm−<sup>1</sup> reflection intensity, which corresponds to the troponin repeat in the thin filament (**Figure 5**). We find that the calculated time courses of Ca2+ bound Tn and the C position of Tm correlate with the transient of meridional 1/38.5 nm−<sup>1</sup> reflection intensity (**Figure 5**). A systematic sensitivity test of all rate constants shows that the decay of Ca2+-bound Tn is limited only by k−<sup>4</sup> of the model (**Figure 5**, **Table 2**). This is the faster of two dissociation rates determined for Ca2<sup>+</sup> from regulated actin (Rosenfeld and Taylor, 1987). Increasing k−<sup>4</sup> by a factor of 1.33, which is within the range of measured values (Rosenfeld and Taylor, 1987), and maintaining the same steady-state conditions (inset, **Figure 2**) bring the decay rate of Ca2+-bound Tn closer in alignment with the decay of meridional 1/38.5 nm−<sup>1</sup> reflection intensity (see Supplementary Materials). Furthermore, the decays of Ca2+-bound Tn and state C of Tm align more closely by increasing K1k−<sup>1</sup> (**Figure 5**, **Table 2**; see Supplementary Materials). Hence, the model predicts that the decay rate of meridional 1/38.5 nm−<sup>1</sup> reflection intensity in the non-overlap preparation gives an in vivo measure of the rate of Ca2<sup>+</sup> dissociation from Tn. A characteristic intensity increase in response to Ca2<sup>+</sup> represents Tn-dependent structural changes of the thin filament (Yagi, 2003). Modeling suggests that Ca2+ bound Tn regulates completely the structural changes related to the movement of Tm to position C in the non-overlap preparation.

### Transient Structural Changes with Overlap

In the overlap preparation, the structural changes related to meridional 1/38.5 nm−<sup>1</sup> reflection intensity are more complex (Matsuo et al., 2010), showing intensity changes with positive and negative slopes over the time course of a twitch (**Figure 6**). The early rise in reflection intensity correlates with a brief period at the beginning of the Ca2<sup>+</sup> pulse in which the model predicts a rise in Ca2+-bound Tn and state C of Tm before the transition to state M of Tm begins. The large negative change in reflection intensity has roughly the same time course as the calculated fraction of Tm in the M position (**Figure 6**). The minimum reflection intensity comes at a time after the Ca2<sup>+</sup> transient has decayed and the calculated M state has peaked.

The calculated decay rate of Ca2+-bound Tn can be made more rapid than the measured decay of fluo-3 fluorescence by increasing rate constants, K<sup>1</sup> k−1, at fixed K<sup>1</sup> (**Figure 6**, **Table 2**). However, both absolute rates and the competition between states M and B for Tm in state C (**Figure 1**) must be made more extreme to hold constant the calculated tension transient (**Figure 3**). Thus, a balance of competing factors explains the correlation between the calculated Ca2+-bound state of Tn and the fluo-3 fluorescence transient in the overlap preparation (**Figure 6**).

### Predicted Time Course of Crossbridge-Dependent Ca2+-Bound Tn

Crossbridge-dependent Ca2+-bound Tn is the difference between Ca2+-bound Tn calculated for overlap and nonoverlap conditions, holding the transient of free Ca2<sup>+</sup> constant (**Figure 7**). The residue is a pool of crossbridge-dependent Ca2+-bound Tn, which represents about 30% of the total area under the curve. The rise in the pool of crossbridge-dependent Ca2+-bound Tn begins near the end of the free Ca2<sup>+</sup> transient and continues during the decay of the tension transient. Peaking at ∼60 ms, the time course of rising crossbridge-dependent Ca2+-bound Tn correlates most closely with the time course of the declining phase in meridional 1/38.5 nm−<sup>1</sup> reflection intensity, which reaches a minimum at ∼70 ms.

### DISCUSSION

A comprehensive model of thin filament regulation presented here supports the hypothesis that Ca2+-bound Tn has a causal relationship with the structure of Tn in the thin filament and not with the pool of Ca2<sup>+</sup> represented by the intensity of fluo-3 fluorescence. Rather than recapitulating the decay of the fluo-3 fluorescence change in the non-overlap preparation of frog muscle, the model presented here generates transients for Ca2+ bound Tn and position C of Tm-Tn that match transient changes in intensity of the meridional 1/38.5 nm−<sup>1</sup> reflection in response to Ca2<sup>+</sup> stimulus. The model predicts similar rates of Ca2<sup>+</sup> dissociation from the regulatory sites of Tn in the first 50 ms following stimulation of both preparations, overlap and nonoverlap. After this period, the model predicts that decays of Ca2+ bound Tn in the two preparations diverge, hence demonstrating a fraction of Ca2+-bound Tn that persists owing to crossbridge interaction. Given the same free Ca2<sup>+</sup> transient for both overlap and nonoverlap conditions, calculations of Ca2+-bound Tn, and crossbridge-dependent Ca2+-bound Tn correlate with the time courses of positive and negative changes in intensity of the meridional 1/38.5 nm−<sup>1</sup> reflection, respectively, suggestive of a causal relationship. By reproducing the twitch of a well-studied physiologic system given a highly reproducible experimental Ca2<sup>+</sup> transient, we achieve a proof of concept for the model presented here.

Fluo-3 may be responding to an exchangeable pool of Ca2<sup>+</sup> bound to Ca2<sup>+</sup> buffers in the sarcoplasm (Cannell and Allen, 1984; Baylor and Hollingworth, 1998). Small diffusible Ca2<sup>+</sup> binding molecules such as ATP and parvalbumin in the myofilaments can facilitate the diffusion of Ca2<sup>+</sup> (Feher, 1984) and thereby possibly reduce random non-uniform reactivation events in relaxing myofibrils. However, a fixed Ca2<sup>+</sup> buffer also

prolongs elevated Ca2<sup>+</sup> in the cytosol at all sarcomere lengths, regardless of the status of Ca2+-bound Tn. The decay of the fluo-3 fluorescence may represent a compromise based on the Ca2<sup>+</sup> requirements of the working and relaxed muscle.

We assume that the Ca2<sup>+</sup> sequestration apparatus has the capacity to prevent a significant rise in free Ca2<sup>+</sup> resulting from a pool of crossbridge-dependent Ca2+-bound Tn during a twitch. Aside from equilibrium binding of Ca2<sup>+</sup> to buffering agents represented by the fluo-3 fluorescence intensity, Ca2<sup>+</sup> is actively transported from the sarcoplasm by the Ca2<sup>+</sup> pump of the sarcoplasmic reticulum (SR Ca2+-ATPase). Previous results from amphibian fast skeletal muscle using inhibitors of SR

Tn (gold) under standard conditions (Table 1). Adjustments in standard conditions (Table 2) alter the decay rate of Ca2+-bound Tn (1*x*CaB).

Ca2+-ATPase have shown that the decay and not the peak in free Ca2<sup>+</sup> depends on active Ca2<sup>+</sup> transport (Jiang et al., 1996; Westerblad and Allen, 1996; Même et al., 1998; Caputo et al., 1999). The delay in onset argues against a significant contribution to the rise and decay process of free Ca2<sup>+</sup> sparks by Ca2<sup>+</sup> dissociating from the pool of crossbridge-dependent Ca2+ bound Tn. We suggest the rate and load of Ca2<sup>+</sup> dissociating from crossbridge-dependent Ca2+-bound Tn is within the capacity of Ca2<sup>+</sup> buffers and SR Ca2+-ATPase to move into the SR without effecting a significant increase in sarcoplasmic free Ca2+.

One assumption of the model is that cooperative activation depends exclusively on cycling crossbridges. Although, Equation (1) has the distinction of being a more general form of the Hill equation (Zot et al., 2012), the resulting system of ODE we derive here are mathematically compatible with any logistic function. Hence, the model presented here is limited to a mechanism of regulation in which a steady-state process fully accounts for cooperative transitions between C and M states.

A second assumption of the model is that activated state, M, is proportional to the fraction of maximum tension. This simplification is consistent with the proposal that tension bearing crossbridges are excluded from thin filament states C and B (Lehman et al., 2000).

A third assumption of the model is that Ca2<sup>+</sup> binding to the regulatory sites of TnC is uncoupled from the process of activation when the complex of Tm-Tn is in positions C and M. Consequently, we model the C position as favored at equilibrium. Tm occupies either C or B positions in reconstructions of skeletal and cardiac filaments, respectively, but the complex of Tm-Tn is positioned exclusively in C with Ca2<sup>+</sup> present (Lehman et al., 2000). There is general agreement that Ca2<sup>+</sup> is required to release

Tm-Tn from the B position. A possible, albeit more complicated, mechanism is for Ca2<sup>+</sup> to have a second, independent action, namely, to perturb the equilibrium position of the Tm-Tn complex. This latter possibility is difficult to reconcile with an uncoupling mechanism.

Although, we do not address a specific myopathy, we suggest that results presented here can be extrapolated to mechanisms underlying disease. By employing a model that accounts for the Ca2<sup>+</sup> regulatory properties of Tn, we provide a satisfying explanation for the events of contraction arising from a transient of free Ca2+. The model presented here is consistent with previous experimental results showing non-cooperative Ca2<sup>+</sup> binding to regulated actin in the presence or absence of rigor crossbridges and recapitulates the complex cooperative relationship between Ca2+-binding and force in the steadystate (**Figure 2**). Of the eight adjustable parameters (**Table 2**), we have consistently published results in which K<sup>1</sup> alone is freely adjusted and K<sup>3</sup> and K<sup>5</sup> vary in a prescribed manner. A recent study shows that the model presented here can inform experiments that explain how a mutation in TnC alters the Ca2<sup>+</sup> sensitivity of cardiac myofilaments associated with the hypertrophic state of the heart (Zot et al., 2016b). A previous study shows that the model presented here can fully explain the depressing effect of Ca2+-insensitive mutant TnC on cooperative activation of skeletal muscle fibers (Zot et al., 2009). Hence, a growing body of experimental results in cardiac and skeletal muscle, mutant and wild type preparations, reconstituted and intact systems, and steady-state and transient conditions are explained by the same model and highly constrained parameters

### REFERENCES


of this model. As a robust and reliable predictor of transient and steady-state changes in thin filament structure related to Ca2+-bound Tn, the model presented here is capable of guiding future experiments to uncover mechanisms by which mutations in excitation-contraction coupling lead to pathological conditions.

### AUTHOR CONTRIBUTIONS

Each of the authors contributed significantly to the conception and design of the work, drafting and revision of the manuscript, and final approval of the version to be published. Both authors agree to be accountable for all aspects of the work.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2016.00406


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Zot and Hasbun. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Commentary: Epigenetic Regulation of Phosphodiesterases 2A and 3A Underlies Compromised β-Adrenergic Signaling in an iPSC Model of Dilated Cardiomyopathy

Lauren A. Cole, Jonathan H. Dennis and P. Bryant Chase\*

*Department of Biological Science, Florida State University, Tallahassee, FL, USA*

Keywords: heart, dilated cardiomyopathy, troponin T, induced pluripotent stem cell, nucleus, histone methylation, epigenetic gene regulation

#### **A commentary on**

Edited by: *Li Zuo, Ohio State University, USA*

#### Reviewed by:

*Han-Zhong Feng, Wayne State University School of Medicine, USA Brandon Biesiadecki, Ohio State University, USA Jop Van Berlo, University of Minnesota, USA*

> \*Correspondence: *P. Bryant Chase chase@bio.fsu.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *21 July 2016* Accepted: *05 September 2016* Published: *23 September 2016*

#### Citation:

*Cole LA, Dennis JH and Chase PB (2016) Commentary: Epigenetic Regulation of Phosphodiesterases 2A and 3A Underlies Compromised* β*-Adrenergic Signaling in an iPSC Model of Dilated Cardiomyopathy. Front. Physiol. 7:418. doi: 10.3389/fphys.2016.00418* **Epigenetic Regulation of Phosphodiesterases 2A and 3A Underlies Compromised** β**-Adrenergic Signaling in an iPSC Model of Dilated Cardiomyopathy**

by Wu, H., Lee, J., Vincent, L. G., Wang, Q., Gu, M., Lan, F., et al. (2015). Cell Stem Cell 17. 89–100. doi: 10.1016/j.stem.2015.04.020

Wu et al. (2015) describe pioneering work that utilizes patient-derived induced pluripotent stem cells (iPSCs) from dilated cardiomyopathy (DCM) patients (Sun et al., 2012) and matched non-DCM relatives to study cellular mechanisms of DCM pathogenesis. They find that iPSC cardiomyocytes have proper β-adrenergic signaling while iPSCs from DCM patients exhibit impaired response to β-adrenergic agonist isoproterenol (ISO), which, physiologically, would be expected to compound the mechanical deficit associated with a mutation in troponin T (TnT). Surprisingly, Wu et al. (2015) find that the mechanisms of altered β-adrenergic signaling involve a direct role for TnT in epigenetic control of phosphodiesterase (PDE) expression, and that the mutation affects TnT function not only in the myofilament lattice, but also in the nucleus. This foundational work demonstrates the utility of iPSC-CMs for direct comparison of healthy vs. diseased tissues by providing a platform for identifying previously unrecognized molecular and cellular mechanisms in the progression of DCM.

The mutation studied by Wu et al. (2015) is a point mutation in the gene for the cardiac isoform of TnT, resulting in a single amino acid change (TNNT2 R173W) in or adjacent to TnT's tropomyosin-binding region. Many DCM mutations in myofilament proteins affect muscle function by decreasing Ca2+-sensitivity (e.g., when assaying Ca2+-dependent myofibrillar MgATPase activity, sliding speed of reconstituted thin filaments in motility assays, or force generation by permeabilized muscle preparations; Willott et al., 2010; Watkins et al., 2011); in other words, more cytoplasmic Ca2<sup>+</sup> would be required to achieve the same functional response. This is indeed the case for the TNNT2 R173W mutation which shifts Ca2+sensitivity of myosin S1 MgATPase activity rightward (toward higher [Ca2+]) by almost 0.1 pCa units, with little or no effect on the maximum MgATPase activity or the maximum sliding speed of thin filaments in motility assays (Sommese et al., 2013). This altered Ca2+-responsiveness of the myofilaments almost certainly results directly in reduced mechanical function of the heart during systole, to the detriment of the DCM patient. Remodeling of the DCM heart, however, depends in part on changes in gene expression. Mechanisms of altered gene regulation in cardiomyopathies have typically

**67**

focused on changes in Ca2+-signaling, mechanosensing, and/or energy metabolism (Frey et al., 2004; Ahmad et al., 2005; Kataoka et al., 2007; Lakdawala et al., 2012; Moore et al., 2012; LeWinter and Granzier, 2014). Wu et al. (2015) invoke a novel and more direct role of TnT in gene regulation.

Wu et al. (2015) found that TnT was present in one-third of nuclei from iPSCs derived from DCM patients with the TNNT2 R173W mutation, compared to ∼5% of nuclei of iPSCs derived from normal individuals. TnT is an abundant myofilament protein present in the sarcomere, responsible for attachment of the troponin complex to tropomyosin and transmission of the Ca2<sup>+</sup> signal that activates systolic cardiac contraction (**Figure 1**). Although TnT contains a strong nuclear localization signal (NLS), its functional role in the nucleus of striated muscle myocytes is poorly understood (Bergmann et al., 2009; Zhang et al., 2015, 2016). Identification of TnT interacting proteins in the nucleus is critical to understanding its function.

Wu et al. (2015) performed co-immunoprecipitation studies in cardiomyocyte nuclear extracts to identify TnT interacting proteins. They found that TnT is associated with histone demethylases KDM1A and KDM5A, as well as histone H3. Furthermore, they characterized chromatin patterns of the PDE 2A and 3A genes, where the authors found significant increases of activation marks (H3K4me3) and decreased repressive marks (H3K27me3) in sequences defined by the authors as regions 1 and 2. Assuming high specificity for the various antibodies used throughout their assays, these results suggest that TnT normally plays a role in the epigenetic

FIGURE 1 | The R173W mutation is associated with increased nuclear TnT in DCM patients. Wu et al. (2015) show nuclear TnT is associated with demethylases, and catalog an altered epigenetic landscape of phosphodiesterase (PDE) genes in DCM iPSCs (purple lollipops represent H3K4me3 and green lollipops represent H3K27me3), which may lead to increased transcription of PDE genes in DCM patients.

regulation of at least these PDE genes. Their study furthermore demonstrates that a TnT mutation not only affects sarcomeric function, but also contributes to the improper regulation of both nuclear localization of TnT and PDE gene expression in DCM patients (**Figure 1**). Precise epigenetic regulation of cardiomyocyte differentiation as well as regulation of expression in a cell-type-specific manner has been recently documented, demonstrating this layer of information is critical for understanding cardiomyocyte (dys)function (Paige et al., 2012; Wamstad et al., 2012; O'Meara and Lee, 2015; Preissl et al., 2015). An improper epigenetic landscape likely contributes to inappropriate regulation of many genes, and it will be important for future work to explore other known DCM mutations in the context of genome architecture.

Wu et al. (2015) demonstrate the use of iPSCs to study a prevalent heart disease and determine a novel role of epigenetic regulation in pathogenesis of DCM. This finding demonstrates that mutations in mechanical proteins that lead to DCM pathogenesis via sarcomere dysfunction can also be exacerbated by regulation of epigenomic state. Nuclear localization of cardiac troponin I (TnI), cardiac troponin C (TnC), and cardiac TnT has been shown in rat neonatal ventricular cardiomyocytes, but their relationship with one another, and presumably tropomyosin and actin, in the nucleus has yet to be clearly established (Asumda and

### REFERENCES


Chase, 2012). Interestingly, co-IP data from Wu et al. (2015) did not identify TnI or tropomyosin as interacting partners of nuclear TnT. It may be the case that these partners in thin filament regulation have independent roles in the nucleus. Because these proteins are often mutated in DCM patients, further studies are necessary to not only delineate the function of these proteins in the nucleus in normal individuals, but to determine whether the unique mechanisms identified by Wu et al. (2015) (i.e., unexpected changes in nuclear localization and unexpected interactions with other molecules, which in this instance affect epigenetic regulation of physiologically important genes) are specific only to the R173W mutation in TNNT2, or if they are more commonly associated with other myofilament protein mutations and other mutations that cause cardiomyopathies (Schoffstall et al., 2011; Chase et al., 2013; Hershberger et al., 2013; Teekakirikul et al., 2013; Brunet et al., 2014; Ho et al., 2015; Teo et al., 2015). The involvement of nuclear mechanical proteins in regulation of chromatin, and thus expression, is a new and important aspect of DCM pathogenesis.

### AUTHOR CONTRIBUTIONS

LC and PC wrote and edited the commentary, JD edited the commentary.

of sarcomere mutations in dilated cardiomyopathy. Circ. Cardiovasc. Genet. 5, 503–510. doi: 10.1161/CIRCGENETICS.112.962761


Zhang, T., Taylor, J., Jiang, Y., Pereyra, A. S., Messi, M. L., Wang, Z. M., et al. (2015). Troponin T3 regulates nuclear localization of the calcium channel Cavβ1a subunit in skeletal muscle. Exp. Cell Res. 336, 276–286. doi: 10.1016/j.yexcr.2015.05.005

**Conflict of Interest Statement:** The authors declare that the commentary was written in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer BB and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2016 Cole, Dennis and Chase. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Cardiac Troponin and Tropomyosin: Structural and Cellular Perspectives to Unveil the Hypertrophic Cardiomyopathy Phenotype

### Mayra de A. Marques and Guilherme A. P. de Oliveira\*

Programa de Biologia Estrutural, Centro Nacional de Ressonância Magnética Nuclear Jiri Jonas, Instituto de Bioquímica Médica Leopoldo de Meis, Instituto Nacional de Biologia Estrutural e Bioimagem, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil

#### Edited by:

P. Bryant Chase, Florida State University, USA

#### Reviewed by:

Brenda Schoffstall, Barry University, USA Adriano S. Martins, Icahn School of Medicine at Mount Sinai, USA

\*Correspondence: Guilherme A. P. de Oliveira gaugusto@bioqmed.ufrj.br

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 27 July 2016 Accepted: 09 September 2016 Published: 23 September 2016

#### Citation:

Marques MdA and de Oliveira GAP (2016) Cardiac Troponin and Tropomyosin: Structural and Cellular Perspectives to Unveil the Hypertrophic Cardiomyopathy Phenotype. Front. Physiol. 7:429. doi: 10.3389/fphys.2016.00429 Inherited myopathies affect both skeletal and cardiac muscle and are commonly associated with genetic dysfunctions, leading to the production of anomalous proteins. In cardiomyopathies, mutations frequently occur in sarcomeric genes, but the cause-effect scenario between genetic alterations and pathological processes remains elusive. Hypertrophic cardiomyopathy (HCM) was the first cardiac disease associated with a genetic background. Since the discovery of the first mutation in the β-myosin heavy chain, more than 1400 new mutations in 11 sarcomeric genes have been reported, awarding HCM the title of the "disease of the sarcomere." The most common macroscopic phenotypes are left ventricle and interventricular septal thickening, but because the clinical profile of this disease is quite heterogeneous, these phenotypes are not suitable for an accurate diagnosis. The development of genomic approaches for clinical investigation allows for diagnostic progress and understanding at the molecular level. Meanwhile, the lack of accurate in vivo models to better comprehend the cellular events triggered by this pathology has become a challenge. Notwithstanding, the imbalance of Ca2<sup>+</sup> concentrations, altered signaling pathways, induction of apoptotic factors, and heart remodeling leading to abnormal anatomy have already been reported. Of note, a misbalance of signaling biomolecules, such as kinases and tumor suppressors (e.g., Akt and p53), seems to participate in apoptotic and fibrotic events. In HCM, structural and cellular information about defective sarcomeric proteins and their altered interactome is emerging but still represents a bottleneck for developing new concepts in basic research and for future therapeutic interventions. This review focuses on the structural and cellular alterations triggered by HCM-causing mutations in troponin and tropomyosin proteins and how structural biology can aid in the discovery of new platforms for therapeutics. We highlight the importance of a better understanding of allosteric communications within these thin-filament proteins to decipher the HCM pathological state.

Keywords: hypertrophic cardiomyopathy, protein dynamics, sarcomeric mutations, thin filament, allostery

## INTRODUCTION

Cardiomyopathies represent a collection of disorders that originate in the heart muscle itself or as a side effect of some other systemic conditions, leading to heart damage and electrical function impairment (Maron et al., 2006). According to the American Heart Association (AHA), cardiomyopathies are classified into two major groups: primary, referring to those predominantly affecting the heart muscle, and secondary, referring to those with the pathological involvement of the heart in a large number of systemic diseases (Maron et al., 2006). Hypertrophic cardiomyopathy (HCM) is a primary muscle disease and the most common cause of sudden cardiovascular death in young athletes; however, cardiovascular death caused by HCM is 8 times more frequent when considering not only young athletes. The same pattern is true for the incidence of HCM, which is 3 times higher in the young population (Maron et al., 2016a). Of note, HCM affects 1 in 500 individuals in the general population, but this ratio may be underestimated due to the lack of information regarding familial cases or asymptomatic subjects (Mozaffarian et al., 2016). HCM was the first cardiac disease associated with a genetic background and presents an autosomal dominant pattern of inheritance. Sequencing efforts have allowed the discovery of the first HCM-causing mutation in the β-myosin heavy chain (MHY7 gene, R403Q) (Geisterfer-Lowrance et al., 1990). Since then, more than 1400 mutations in 11 sarcomeric genes have been unveiled, and due to this pattern of affected genes, HCM is also called the disease of the sarcomere (Watkins et al., 1992, 1995; Thierfelder et al., 1994; Seidman and Seidman, 2001; Konno et al., 2010).

The clinical profile of HCM is quite heterogeneous. While some patients exhibit severe to mild manifestations, others are completely unaware of having the disease. The initial suspicious of HCM come from a heart murmur during physical activity, family history, or an abnormal echocardiogram (ECG) pattern (Marian, 2010; Maron et al., 2012; Maron and Maron, 2013). Its diagnosis is based on two-dimensional echocardiography, which permits the detection of an asymmetric hypertrophied left ventricle chamber. Of note, the HCM diagnose should be taken in the absence of other diseases with similar clinical profiles (e.g., aortic stenosis or hypertension) (Maron et al., 2014, 2016b). Moreover, other HCM clinical manifestations include left ventricular hypercontractility, cardiac insufficiency, ventricular fibrillation, syncope and arrhythmias. Regarding its morphological and histological features, left ventricle wall and ventricular septum thickening typically occurs (Teare, 1958; Maron et al., 1979; Varnava et al., 2001). The architecture of the hypertrophic myocardial fibers differs in shape and angle arrangement, leading to a chaotic environment (Maron et al., 1981). In combination with cellular disarray, fibrosis with an abnormal collagen matrix is also observed (St. John Sutton et al., 1980; Shirani et al., 2000; Kwon et al., 2009; Nakamura et al., 2016). Indeed, a possible clinical correlation between these findings and HCM pathology impairs the proper relaxation of the heart, preventing it from filling correctly. Damage to the electrical signal conduction may also occur, leading to arrhythmia, tachycardia and ventricular fibrillation, which may ultimately contribute to the development of secondary pathologies, e.g., ischemia or hypotension (Kon-No et al., 2001; Christiaans et al., 2009; Lan et al., 2013; Crocini et al., 2016). Altered ion channels including at least six susceptible genes, e.g., KVLQT1, HERG, SCN5A, minK, MiRP1, and RyR2 play critical steps during the development of arrhythmia phenotypes (Keating and Sanguinetti, 2001). Of note, the ryanodine channel (RyR2) triggers the release of Ca2<sup>+</sup> from the sarcoplasmic reticulum to start contraction. Mutations in RyR2 lead to aberrant intracellular Ca2<sup>+</sup> metabolism and Ca2<sup>+</sup> overload that may have an involvement in arrhythmias (Keating and Sanguinetti, 2001). Additionally, during the phase 0 depolarization of the cardiac action potential, the binding of calmodulin to the C-terminal region of the hH1 Na<sup>+</sup> channel occurs in a Ca2+-dependent manner and impact the slow inactivation gating process with implications to cardiac arrhythmias (Tan et al., 2002). Because the HCM clinical phenotype ranges from asymptomatic subjects to patients who require surgery or transplant, it is reasonable to use both clinical data and imaging tests during initial screening, but this may not be the most effective approach for the diagnosis probands carrying a silent disease. Genetic tests are available for molecular diagnosis, to identify HCMcausing mutations of the proband and for family screening (Ho et al., 2015). These trials were conducted at the bench and were breakthroughs, promoting a fast and reliable diagnosis (Maron et al., 2012). However, despite all efforts in the molecular biology field, the association between mutational profile and disease phenotype remains elusive. For instance, one interesting question is why some mutations trigger a pathogenic status while others lead to a benign course. The most parsimonious explanation is that mutations not only affect protein function but may also cause changes in folding, dynamics and interactomes. In this review, we focused on recent discussions of the structural basis for the effects and cellular consequences of HCM mutations.

## FROM CODE TO MESSAGE

Almost 50 years ago, Francis Crick published a noteworthy article entitled the "Central dogma of molecular biology" (Crick, 1970). At that time, it was postulated that in the cellular environment, the flow of information initiates through the deoxyribonucleic acid (DNA) molecule that contains all of the elements for protein production. The intermediate molecule responsible for carrying a copy of the target protein is called ribonucleic acid (RNA). Finally, this template requires a complex cellular machinery for protein synthesis.

Drawing a parallel between the central dogma of molecular biology and the theory of communication (**Figure 1**), five key elements should be considered for successful communication. (i) The source of information is important for producing the message. An analogy of the source could be a sequence of letters organized in a coherent way to form a word or sentence. (ii) The transmitter is responsible for sending this message through a channel. (iii) The channel links the transmitter to the receiver. (iv) The receiver makes the message intelligible when reaching its

destination. (v) The destination is where the message arrives and can then perform its function (Shannon, 1948).

Considering the polypeptide chain of a hypothetical protein as a message to be sent, the specific DNA region that carries the correct nucleotide sequence for this protein (i.e., the gene) serves as the source of information. Equally important, the RNA molecule plays the transmitter role, allowing the correct message to reach the channel. The nuclear pore complex recognizes mature RNA molecules and sends them to the receiver, thus serving as a selective channel. Next, the complex machinery of the ribosome acts as the receiver, transforming the genetic information through the synthesis of a correct polypeptide sequence. In this analogy, additional receivers would allow proper polypeptide folding (e.g., the endoplasmatic reticulum and chaperone molecules) and the incorporation of important modifications, such as phosphorylation and glycosylation, among others (e.g., kinase proteins and the Golgi apparatus). Finally, the destination of this functional protein will generate the expected response inside or outside the cell. This simplistic analogy illustrates how any substitution occurring at the message level may significantly affect the correct communication and cause either a misunderstanding or a new understanding at the destination (**Figure 1**).

Correct folding is intimately linked to the function or, analogously, the message of a biomolecule. A remarkable study in the 1950s showed that the three-dimensional assembly of a protein is guided by its specific amino acid sequence. The physicochemical properties of specific amino acid side chains lead to a hydrophobic collapse event; thus, proper folding is expected to occur independently of biological cellular machinery (Anfinsen et al., 1961; Anfinsen, 1973). The energy landscape is an accepted theory to explain the protein folding phenomenon (Onuchic et al., 1997). Proteins normally experience a wide range of conformational changes to reach their low-energy native state. In terms of free energy, folding pathways are commonly summarized using schematic funnels, in which high-energy protein stages (i.e., denatured polypeptides) are guided through preferential intramolecular contacts to achieve the lowest free energy and conformational entropy (i.e., native polypeptides). The folding of a single polypeptide chain, therefore, occurs through the assembly of partially folded intermediates due to dynamics and, together with water solvation, plays an essential role in this process (Bai and Englander, 1996; Dill and Chan, 1997; Onuchic, 1997; Cheung et al., 2002; Onuchic and Wolynes, 2004; Dill et al., 2007).

The assembly of tightly bound or transiently bound molecular complexes in the perspective of energetic landscapes requires an orchestrated and hierarchic environment in which a range of molecular motions take place (**Figure 2**). Polypeptide chains are intrinsically dynamic entities sampling different conformations from pico- to millisecond timescales during folding and upon activation or molecular recognition. Rapid fluctuations (i.e., thermal motions on the order of 10−12– 10−<sup>9</sup> s) commonly occur in native globular proteins and result in structurally similar conformers with implications for molecular recognition (**Figure 2A**). At the opposite side of the energetic landscape, intrinsically disordered proteins (IDPs) are very flexible molecular motors, sampling heterogeneous conformations with high free energy and conformational entropy (**Figure 2B**). In both scenarios, we may exemplify slower timescale dynamics (i.e., on the order of 10−6– 10−<sup>3</sup> s) in which a small number of high-energy and shortlived conformers are populated (**Figure 2C**). This condition is normally observed in more complex proteins with intrinsically disordered regions (IDRs), such as flexible linkers or hingeconnecting domains, and that participate in relevant biological processes, including molecular assemblies, catalysis and ioncoordination. In the context of multi-domain proteins (i.e., globular domains linked by IDRs) and the broad spectrum of motions that arise from these complexes (**Figure 2D**), an understanding of how hierarchical motions trigger the formation of tightly or transiently bound molecular assemblies and

the physiological and pathological consequences is just now emerging.

The consequences of alterations in the DNA can be compared to "heaven or hell." Evolutionary events are commonly beneficial in which adaptive species and precise protein functionalities are selected but may also be catastrophic when loss-of-function (LoF) or gain-of-function (GoF) events occur. Upon mutation, LoF and GoF events are commonly observed in different proteins, including tumor suppressors and sarcomeric proteins, with implications in cancer and cardiac disorders (Marston et al., 2013; Silva et al., 2014). Several organisms have developed specialized repair machineries to address DNA variations and avoid disease, but this is not an infallible process. In addition, the evolution of polypeptide chains does not always result in the correct alphabet sequence (Goldschmidt et al., 2010; Eichner and Radford, 2011). The consequence is that protein misfolding is attributed to more than 50 diseases (Chiti and Dobson, 2006). DNA alterations can change the phenotype with positive, negative or neutral consequences to the adaptability of individuals. The production of faulty proteins causes functional impairment, which may alter contacts with molecular partners, leading to dramatic cellular responses. Among the main variations observed in the DNA, one type can be highlighted in the context of cardiomyopathies: single point mutations. Approximately 90% of these pathogenic mutations are missense, in which one amino acid is changed to another, leading to abnormal molecular, functional and physical properties of the heart.

Pathogenic mutations have at least two important features: (i) they lead to alterations in protein structure and function and (ii) changes frequently occur in highly conserved amino acids throughout evolution (Richards et al., 2008; Maron et al., 2012). In HCM, these two features are observed, and one mutation affecting a conserved region can disrupt pivotal proteinprotein interactions, leading to structural chaos. Interestingly, because only key amino acids contribute to the free energy of binding, interaction sites frequently display an asymmetrically energy distribution along the surface. The presence of highly conserved amino acids surrounding the interaction interface plays a key role in stabilizing molecular contacts. Accordingly, complementary pockets comprising bulky side chains tightly stabilize the interaction in a geometrically and energetically favorable manner (Keskin et al., 2005; Metz et al., 2012). When dealing with protein-protein interactions, in addition to size and chemical complementarities, conformational fluctuations in different timescales also take place for proper binding and signaling (Henzler-Wildman and Kern, 2007; Zen et al., 2010). Mutations in highly conserved regions frequently result in steric impairment and conformational changes with LoF or GoF events. Thus, protein dynamics are highly relevant when considering the investigation of protein-protein interactions in physiological and pathological processes.

With regard to RNA modifications impacting cardiomyopathy phenotypes, much attention are being taken to non-conding RNAs (e.g., microRNAs—miRNA) and the process of alternative cleavage and polyadenylation (APA) of mRNAs as regulators of gene expression. For instance, downregulation of a specific miRNA locus in stressed cardiomyocytes is sufficient to attenuate the increase of cell size (Clark et al., 2016) and the levels of miR-499 are increased in failing and hypertrophied hearts with consequences to the levels of target mRNAs. Proteomic analysis linked to miR-499 identified changes in kinase and phosphatase signaling, supporting the key role of these non-conding RNAs in the pathological regulation of cardiomyopathies (Matkovich et al., 2012). Of note, the secretion of miR-29a in the plasma of HCM subjects is associated to both hypertrophy and fibrosis and may serve as a potential biomarker in HCM supporting a direct role of miRNAs in the HCM pathogenesis (Roncarati et al., 2014). Additionally, the APA process of mRNAs is particularly important to generate RNA isoforms and modulate the levels of protein expression in specific genes. In dilated cardiomyopathy a group of genes involved in RNA and actin binding and structural proteins of the cytoskeleton revealed a different profile of mRNA cleavage and polyadenylation and it may account for an additional level of regulation in failing hearts (Creemers et al., 2016).

The bottleneck in our current understanding is the assessment of dynamic changes arising from mutational events in wellorganized assemblies, such as the sarcomere, and defects generating pathogenic profiles. More interesting, different mutations in diverse targets can culminate in the same disease but sometimes with distinct phenotypes.

### THE SARCOMERE

To better understand the complexity of the sarcomere, it is important to address some important structural features about this system (Clark et al., 2002; Gautel and Djinovic-Carugo, ´ 2016). The sarcomere is a basic contractile unit that repeats regularly throughout myofibrils (Huxley and Niedergerke, 1954) being the responsible for the transformation of chemical energy into mechanical energy, thereby triggering contraction (Bers, 2001). To execute this task, the architecture of the sarcomere is finely orchestrated. Through electron microscopy visualization, several elements can be identified including the Z-discs, M lines, A bands, and H zones. The borders of a sarcomere unit are defined by Z-discs and are the places where the thin filaments, titin and nebulin are anchored. These discs are also involved in mechanosensitivity and nuclear signaling, which contribute to the maintenance of muscle homeostasis (Clark et al., 2002). Of note, the giant titin protein has received much attention with regard to its potential role in the passive and residual force enhancement (Herzog and Leonard, 2002). Stretching active and passive myofibrils to a length that avoided any force contributions from actinmyosin cross-bridges revealed greater force generation in actively when comparing to passively stretched myofibrils, supporting the involvement of titin molecules on this force generation mechanism (Leonard and Herzog, 2010). A three-filament model of force production emerged from these findings with the participation of Ca2<sup>+</sup> and actin binding to titin molecules (Herzog et al., 2016 and references therein). For more in-depth information on the advantages and limitations of the actinmyosin-titin communication for force generation (three-filament model) with respect to the classical view of cross-bridge, please refer to specialized literature (Herzog et al., 2016; Li et al., 2016).

The M line is the transverse structure located in the center of the sarcomere and has been proposed to be the anchor site of the thick filament through the formation of cross-bridges (Obermann et al., 1996; Agarkova et al., 2003). Toward the center of the sarcomere is the A band composed of thick filaments and associated proteins, e.g., the myosin-binding protein C. This protein plays an important role in regulating myosin polymerization and aligning the thick filaments within the A band. Within this region, a whiteness segment called H is also observed and consists of thin filaments that do not overlap into thick filaments (Clark et al., 2002). The thick filaments are mainly composed of myosin which has three functional domains: the head, the neck, and the tail (Sellers, 2000). The head is the motor domain binding ATP and actin, while the neck region binds to its light chains or calmodulin. Finally, the myosin tail anchors and moves the motor domain toward an efficient interaction with actin (Saez et al., 1987; Sellers, 2000). Six subunits promote a hexameric three-dimensional architecture that includes two heavy chains (myosin heavy chain, MHC) and four light chains (myosin light chain, MLC). The MLC is divided in two domains with regulatory functions and another two domains with structural functions. Together, they finely adjust the motor activity of myosin and the versatility of its kinetics (Milligan, 1996).

The thin filament is the major Ca2<sup>+</sup> regulation site and comprises actin, tropomyosin (Tm) and the troponin complex (Tn). Actin is a ubiquitously expressed protein and participates in various cellular events, such as motility, cytokinesis and contraction. Although some actin mutations are involved in the HCM (Bai et al., 2015), we will focus this review on the Tn and Tm mutations. Along the length of monomeric actin resides Tm, a "coiled-coil" dimer that interacts in a "headtail" manner to form a substantial and almost uninterrupted structure around the actin helices. One "coiled" motif interacts with seven actin monomers via saline ionic interactions or through Mg2+, and its main function is to inhibit myosin ATPase activity in the absence of Ca2<sup>+</sup> (Zot and Potter, 1987). The cardiac Tn (cTn) complex is composed of three subunits (C, I, and T) with different three-dimensional structures and functions. Together, these subunits perform the important role of Ca2<sup>+</sup> and contraction regulation. Cardiac troponin C (cTnC) is the direct Ca2<sup>+</sup> sensor in the myofilament. The conformational changes triggered by Ca2<sup>+</sup> binding at specific cTnC sites control the allosteric signaling cascade along the entire complex. Cardiac troponin I (cTnI) performs the classical role of ATPase activity inhibition. In addition, its tridimensional structure plays an important regulatory role in protein-protein interactions. The cardiac troponin T (cTnT) is the "molecular glue" that anchors the Tn members to the thin filament, playing an important role in Ca2<sup>+</sup> transduction structural signaling. Together, the Tn complex and Tm represent the regulatory proteins of the thin filament (Zot and Potter, 1987).

### THE THIN FILAMENT REGULATORS

The cardiac contraction-relaxation cycle is a physiological event controlled by electric and neurohormonal factors. At the molecular level, this process requires sophisticated protein machinery, i.e., the sarcomere, to manage both chemical and mechanical processes. Particularly interesting, this exquisite machinery exhibits an extensive intra- and intermolecular network. In this context, fine-tuned protein-protein interactions are essential for the correct contraction-relaxation operation. The Tn complex and Tm are the key macromolecules responsible for the modulation of both dynamics and structural signaling along the myofilament. As will be discussed further, these proteins ultimately regulate the exposure of the myosin-actin binding sites.

### TROPONIN I

cTnI is the inhibitory unit of the Tn complex and plays an important role as a structural regulator of actomyosin ATPase activity (Leavis and Gergely, 1984). During a heartbeat, cTnI participates in the systolic/diastolic cycle upon changes in the intracellular Ca2<sup>+</sup> concentration. Using a series of Cterminal mutations, the regulatory segments of TnI involved in the anchoring of TnC to the thin filament start to be addressed (Ramos, 1999). Currently, cTnI can be divided into six different functional regions (Li et al., 2004): (i) an N-terminal extension region (residues 1–30) containing two PKA-dependent phosphorylation sites (Ser23 and Ser24) (Robertson et al., 1982; Chandra et al., 1997); (ii) an N-terminal region (residues 34–71) that interacts with the cTnC C-domain, playing a structural role (Gasmi-Seabrook et al., 1999; Mercier et al., 2000); (iii) a region that binds to cTnT (residues 80–136) as part of the IT-arm; (iv) an inhibitory region (TnI128−147) containing a highly conserved amino acid sequence among TnI isoforms; (v) a switch region (cTnI147−163) that experiences a disordered-ordered transition upon binding to the N-terminal domain of cTnC-triggering contraction (Li et al., 1999); and (vi) the C-terminal region (residues 164–210) serving as a second consensus site that binds to actin and Tm (Solaro, 2010; **Figure 3**). Upon β-adrenergic stimulus, two PKA-dependent phosphorylation sites (Ser23/24) within the N-terminal segment of cTnI become phosphorylated and play important roles in the Ca2<sup>+</sup> desensitization activity of the myofilament, culminating in heart relaxation (Zhang et al., 1995; Solaro et al., 2013). Additionally, other residues in cTnI, including Ser43/45 and Thr143, were shown to be phosphorylated by PKC (Noland et al., 1995), with physiological and pathological implications (Solaro et al., 2013). These Nterminal residues of cardiac TnI are absent in fast and slow TnI isoforms, providing an additional key regulatory segment for the cardiac tissue. At the central region of cTnI, the inhibitory segment (residues 128–147) was shown to attach to actin in the absence of Ca2<sup>+</sup> but shifts toward an interaction with cTnC upon Ca2<sup>+</sup> addition (Potter and Gergely, 1974; Kobayashi et al., 1999).

In fact, cTnI is an important molecular switch during the systolic/diastolic cycle. During diastole, low levels of Ca2<sup>+</sup> stabilize a cTnI conformation that suppresses the power stroke and thus prevents actin-myosin interactions, mostly because the cTnI inhibitory region is sitting on actin (Solaro, 2010). During β-adrenergic stimulation, the intracellular Ca2<sup>+</sup> concentration increases from a diastolic level of 100 nmol/L to a systolic level of 1 mmol/L, enabling contraction (Bers, 2000). The affinity of cTnI to actin is reduced upon the binding of Ca2<sup>+</sup> to the N-terminal domain of cTnC (cNTnC), releasing actin inhibition (Solaro and Rarick, 1998). The cTnI inhibitory region now interacts in close proximity to the D/E linker in the cTnC (Lindhout and Sykes, 2003), and the cTnI switch region moves toward the exposed hydrophobic patch in the cNTnC, stabilizing its open (and active) conformation. This mechanism causes cTnI to serve as a molecular latch mediating actin exposure in a Ca2+-dependent manner to trigger or inhibit contraction.

Due to the important regulatory role of cTnI in the thin filament, mutations that affect cTn structural cooperativity can lead to the development of diseases. Because several studies have attempted to characterize and better understand the functional defects of cTn mutations in the HCM phenotype, our main focus here is to provide an overview of groundbreaking works and recent literature; unfortunately, we cannot provide a complete review of this vast and enthusiastic literature. Please, for additional studies, see the references cited therein.

The majority of cTnI mutations associated with the HCM phenotype are located in the C-terminal region (**Figure 3**). Kimura and coworkers reported the first six mutations (R145G, R145Q, R162W, 1K183, G203S, and K206Q), and almost 30 variations have been reported so far (Kimura et al., 1997; Willott et al., 2010). Interestingly, almost 60% of the HCM-causing cTnI mutations occur through the substitution of a positively or negatively charged amino acid for a neutral or hydrophobic one. Of note, arginine replacement occurs ∼40% of the time (Willott et al., 2010). Most of the functional defects triggered by these mutations were explored using skinned cardiac muscle fibers. For example, all of these mutants except for G203S had increased Ca2<sup>+</sup> sensitivity of myofibrillar ATPase activity and force generation (Takahashi-Yanaga et al., 2001a). As expected, both mutations at position 145 comprising the inhibitory consensus site presented decreased inhibitory cTnI activity, in contrast to R162W and 1K183, which presented decreased affinity for the cTnI-actin interaction. With the exception of R162W, none of the other mutations perturbed the cTnIcTnC interaction (Takahashi-Yanaga et al., 2001a). Using surface plasmon resonance, R162W revealed higher affinity for cTnC in the presence of Ca2<sup>+</sup> (Elliott et al., 2000). Further investigations have explained how R145G impacts the cTnI inhibitory activity. Using skinned cardiac fibers, the authors revealed that this mutation impairs force development and muscle relaxation, possibly explaining some of the clinical features of HCM (Lang et al., 2002). Interestingly, cTnI mutations located at the second actin-Tm-binding site (D190H and R192H) did not increase the levels of ATPase activity, in contrast to the previous R145G

mutation located at the cTnI inhibitory segment (Kobayashi and Solaro, 2006). Pathogenic mutations occur at different regulatory segments of cTnI (for the cTnI mutation involved in cardiomyopathies, please Willott et al., 2010). This observation provides insights into the potential repertoire of functional defects that cTnI would be involved. Because cTnI is involved in a series of complex interactions with different biological partners in the thin filament, the expectation is that different cTnI mutation sites will trigger distinct functional defects, as clearly observed in the literature. Although the literature is well designed and ongoing, the explanation of how each of these mutations reflects the clinical HCM phenotypes seems to be challenging.

The development of transgenic animal models represents a serendipitous way to associate mutational defects with disease phenotypes. For example, transgenic mice carrying cTnI R146G (R145G in humans) presented cardiomyocyte disarray, fibrosis, and hypercontractility with diastolic dysfunction, symptoms that align with an increased sensitivity to Ca2<sup>+</sup> and the HCM phenotype (James et al., 2000). In contrast, a slight decrease

in the Ca2<sup>+</sup> sensitivity of force development was also reported for these cTnI R146G mice (Kruger et al., 2005). In an elegant study, Wen and coauthors were able to study using R245G transgenic mice concomitant measurements of force and actomyosin ATPase activity in skinned papillary fibers to extract the rate of cross-bridge turnover and energy cost. Compared to fibers from a human cTnI wt mice, the fibers from the R245G mice decreased the average force per cross-bridge and revealed a higher energy expenditure, suggesting that in the HCM phenotype, compensatory mechanisms may take place in the heart of R145G mice (Wen et al., 2008). More recently, molecular dynamic (MD) simulations using the whole cTn complex incorporated with cTnI-wt, cTnI-R145G and the cTnI-R145G/S23D/S24D phosphomimetics provided atomistic details to explain new interactions between cNTnC (residues 1–89) and cNTnI (residues 1–41) and the effects of R145G on these interactions (Lindert et al., 2015). The incorporation of aspartic acid substitutions at the PKA-dependent phosphorylation sites of cTnI was validated to recapitulate the same contractile properties and Tn function as does PKA (Lindert et al., 2015 and references therein). The MD simulations revealed that in the cTn complex incorporated with the wt cTnI containing the S23D/S24D phosphomimetics, the loss of contact between cNTnI region 1–41 and the A and B helices of cNTnC occur because of the repositioning of the cNTnI region to interact with the cTnI inhibitory peptide due to phosphorylation. Accordingly, this disordered N-terminal segment of cTnI was previously shown to interact with the cNTnC in the unphosphorylated state, thereby stabilizing for cNTnC a rigid and open orientation (Ferrières et al., 2000). Additionally, this interaction was disrupted by the phosphorylation of Ser23/24 during the βadrenergic stimulus to induce a lusitropic condition. More interesting, when the R145G cTnI inhibitory peptide mutant was incorporated for the MD simulations, the cNTnI1–41-cTnI128–<sup>147</sup> interaction did not occur, and this extreme N-terminal region of cTnI maintained contact with the cNTnC region regardless of the presence of phosphomimetic mutations. In the same line, the R146G (R145G in human) and R21C cTnI mutants altered the PKA-dependent effects on weakening cTnC-cTnI interactions and accelerating myofibril relaxation (Cheng et al., 2015). These observations explain how R145G reduces the modulation of the cTn complex by S23/24 phosphorylation upon β-adrenergic stimulus and adds evidence for intramolecular contact in cTnI triggered by phosphorylation (Lindert et al., 2015).

The C-terminal end segment of cTnI is the most conserved structure among the cTnI isoform and species and is the site of the G203S and K206Q mutations. This region presents several charged amino acids and a highly flexible structure and seems to interact with Tm, playing a role in the Ca2<sup>+</sup> switch of the thin filament (Sheng and Jin, 2014). Both mutants affect the backbone structure of cTnI; in particular, K206Q increases the maximum levels of ATPase activity and the filament sliding velocity (Deng et al., 2003; Köhler et al., 2003). Moreover, G203S disrupts the interaction between cTnT and cTnC, resulting in Ca2<sup>+</sup> deregulation. Transgenic mice expressing the cTnI mutant G203S revealed a faster inactivation rate of the L-type Ca2<sup>+</sup>

channels and a greater increase in the mitochondrial membrane potential and metabolic activity upon activation compared to wt myocytes (Tsoutsman et al., 2006; Viola et al., 2016).

Remarkably, there is only one identified mutation in the N-terminal region of cTnI so far: the R21C mutant (Gomes et al., 2005). This mutation is located within the consensus phosphorylation site of PKA. In vitro studies demonstrated that R21C increases the Ca2<sup>+</sup> sensitivity but reduced the phosphorylation levels when incubated with PKA. The physiological effect of PKA on decreasing the Ca2<sup>+</sup> sensitivity of force development was diminished in cTnI R21C (Gomes et al., 2005). Additionally, the generation of R21C knock-in mice revealed the interesting behavior of this intriguing mutant. Top-down electron capture dissociation mass spectrometry revealed that R21C indeed depleted the phosphorylation status of S23/S24 cTnI in R21C homozygous (+/+) mice and decreased it by 8% in R21C heterozygous (+/−) mice compared to wt mice, supporting R21 as a crucial residue for PKA recognition and subsequent phosphorylation (Wang et al., 2012). Additionally, heterozygous R21C mice incorporated ∼25% of R21C to the thin filament. The development of an HCM phenotype with cardiac hypertrophy, fibrosis, and the activation of the fetal gene program in both +/+ and +/− R21C mice supports a negative-dominant effect of this pathogenic mutant to the wt cTnI that is intensified to diastolic dysfunction and excitation-contraction uncoupling upon long-term ablation of cTnI phosphorylation (Dweck et al., 2014). Of note, these mice (+/+) also presented distinct contractile forces when comparing left and right ventricles (Liang et al., 2015). Curiously, the use of post-mortem heart tissues revealed ∼56 and 1% cTnI phosphorylation when comparing normal and affected patients, respectively. Thus, mapping cTnI phosphorylation levels presents a promising biomarker for the early detection of hypertrophy (Zhang et al., 2011).

### TROPONIN T

cTnT is the subunit responsible for anchoring the cTnC and the cTnI to the thin filament and serves as an important communication switch in transferring the conformational changes induced by Ca2<sup>+</sup> over the cTn complex and Tm (Leavis and Gergely, 1984; Tobacman, 1996). Due to genetic shuffling, several TnT isoforms are expressed across species, cell types and within the cellular environment (Anderson et al., 1991; Perry, 1998). The main structural difference between cardiac and skeletal TnT is the length of the N-terminal domain segment. This region carries the major structural multiplicities that are generated by genetic shuffling and is known as the hypervariable region. During heart development, an alternative splicing of exons 4 and 5 gives rise to different variants in size, physicochemical features and, thus, functions and Ca2<sup>+</sup> responsiveness (Anderson et al., 1991; McAuliffe and Robbins, 1991). Four TnT isoforms can be detected in the heart: TnT1 (all exons), TnT2 (exon 4 is spliced out), TnT3 (exon 5 is spliced out), and TnT4 (exons 4 and 5 are spliced out) (Gomes et al., 2002). Due to this arrangement, the TnT molecular weight varies from 31 to 36 kDa, presenting 250–300 amino acid residues (Perry, 1998). In addition, altered patterns of cTnT expression were observed in heart failure and hypertrophy. Protein levels are directly associated with disease severity, suggesting an important role in the pathological state (Anderson et al., 1991, 1995; Townsend et al., 1995; Saba et al., 1996). The TnT is classically divided in two domains, T1 and T2, based on observations by Ohtsuki (1979). The proteolytic cleavage of skeletal TnT generates two different fragments, both binding to Tm. The T1 segment corresponds to the N-terminal region and comprises the hypervariable and central regions (Potter et al., 1995; Oliveira et al., 2000; Gollapudi et al., 2013). The hypervariable region plays a regulatory role in the Tm-binding affinity of TnT site 1 (Amarasinghe and Jin, 2015). The central region is highly conserved across species, playing a key role in anchoring the Tn complex to the thin filament through strong interaction with Tm and performing multiple functions (Heeley et al., 1987; Lehrer and Geeves, 1998; Palm et al., 2001; Regnier et al., 2002; Tobacman et al., 2002; Hinkle and Tobacman, 2003). The C-terminal T2 region interacts with TnC, TnI, actin and a second Tm binding site (Perry, 1998; Jin and Chong, 2010). In vitro studies have reported that TnT presents several phosphorylation sites, including Thr197, Ser201, Thr206, and Thr287 (mouse sequence) (Noland et al., 1989; Wei and Jin, 2011). Upon phosphorylation, these regions seem to be negative regulators of maximal force and Ca2<sup>+</sup> sensitivity. In particular, Thr206 phosphorylation is critical for the functional properties of the thin filament (Sumandea et al., 2003, 2009). Although phosphorylation seems to play an important regulatory role in cTnT, the same pattern of phosphorylation was not observed in vivo. Of note, cTnT from mouse, rat and human tissues appears to be either monophosphorylated in Ser2 or unphosphorylated (Perry, 1998; Sancho Solis et al., 2008; Zhang et al., 2011; Streng et al., 2013). One reasonable explanation for these discrepant phosphorylation results is the high degree of Nterminal region conservation compared to other phosphorylation sites. In addition, the tridimensional arrangement of cTnT in the thin filament could occlude the other phosphorylation sites. Indeed, the specific role of Ser2 phosphorylation remains unclear (Solaro and Kobayashi, 2011). Monasky and coworkers revealed in a mouse model that the p21 kinase regulates cTnT phosphorylation in global myocardial ischemia and reperfusion injury. This result suggests that cTnT phosphorylation regulates cardiac homeostasis (Monasky et al., 2012; Streng et al., 2013).

Inherited cardiomyopathies caused by cTnT mutations account for ∼15–30% of all HCM reports. In this context, the two cTnT-Tm anchoring regions are particularly interesting, harboring the vast majority of mutations observed so far (Willott et al., 2010; **Figure 3**). Once again, we would like to stress that the maintenance of key intermolecular interactions in this complex system is particularly important for the homeostasis of the thin filament function. Thierfelder and coauthors reported the first cTnT variations associated with HCM (Thierfelder et al., 1994). cTnT mutations spread in an autosomal dominant manner and appear to develop a malignant effect with a high incidence of sudden death (Watkins et al., 1995). Several mutations increase the Ca <sup>2</sup><sup>+</sup> sensitivity of force development and the force-pCa relation, e.g., I79N, R92Q, R92L, R92W, R94L, and A104V. However, the maximum force generation, ATPase activity and Ca2<sup>+</sup> cooperativity are maintained. On the other hand, R278C decreases the Ca2<sup>+</sup> cooperativity of force generation in skinned fibers, in addition to the Ca2+-sensitizing effect (Morimoto et al., 1998, 1999; Morimoto, 2007). These defects illustrate that cTnT mutations may alter not only the Ca2<sup>+</sup> affinity for the myofilament but also intermolecular contacts. Mutations in position 92 show different effects on the folding of the cTnT tail domain (Hinkle and Tobacman, 2003). Additionally, interactions of R92Q, R92W, R92L, and R94L with Tm-dependent functions were reported to be impaired (Palm et al., 2001). An interesting atomistic model has been proposed to identify how these mutations affect allosteric modulations through the thin filament (Manning et al., 2012). More interesting, R92L, R92W, and R94L are still able to induce the muscle generation of force in a Ca2+ dependent manner, even under an acidic pH. The resistance to pH also suggests a role in the poor prognosis of HCM (Harada and Potter, 2004). A contraction event at low pH (e.g., ischemia) would decrease the intracellular levels of ATP, causing an upregulation of the cytokines involved in the activation of apoptotic pathways (Morimoto et al., 1998). Structurally, these mutations promote an increase in cTnT helical stability, suggesting a more rigid structure (Palm et al., 2001). Moreover, the measure of Tm affinity decreased in all cTnT mutants, strongly suggesting that the supposed disordered Tm anchoring region plays a role in Tm-cTnT intermolecular contacts (Palm et al., 2001; Manning et al., 2012). cTnT mutations tend to be clustered in a conserved region comprising residues 92 (R92Q, R92L, R92W) and 160– 163 (D160E, E163R, and E163K). Positions 160–163 are located within the conserved, highly charged region (158-RREEEENRR-166) and due to their flexibility are believed to play an important role in the regulation of the thin filament. An interesting study coupling in vitro and in vivo studies revealed that this region is exposed to the solvent. Mutations in this cluster alter critical electrostatic interactions for proper allosteric communication that leads to the transition from the blocked to the closed state (Moore et al., 2013, 2014). When expressed in mouse hearts, structural changes induced by R92Q were able to increase ATP consumption in the intact beating heart. However, using activating Ca2<sup>+</sup> concentrations, R92Q decreases the energydriven force, leading to a failure in the contractile performance (Tian and Ingwall, 1996; Chandra et al., 2001; Javadpour et al., 2003; Schwartz and Mercadier, 2003; Jimenez and Tardiff, 2011). This evidence suggests that this mutation is able to disrupt the myofilament Ca2<sup>+</sup> sensibility probably due to impaired Tm-cTnT interactions (Takahashi-Yanaga et al., 2001b).

Another arginine replacement was reported in position 278 located on the C-terminal end of cTnT (Yanaga et al., 1999). When reconstituted in rabbit cardiac myofibrils (Sirenko et al., 2006) or skinned cardiac muscle fibers (Yanaga et al., 1999), R278C shows an increase in the Ca2<sup>+</sup> sensitivity of ATPase activity; however, the maximum force cooperativity decreases. Moreover, R278C is able to disorder the α-helical structure of the wt cTn in addition to modifying the interface between the cTn core and the rest of the thin filament (Sirenko et al., 2006). In transgenic mice carrying R278C, the effect of Ca2<sup>+</sup> sensitivity was not observed; however, the decrease in the maximal force corroborates the results of previous studies (Hernandez et al., 2005). cTnT-R278C impairs cardiac relaxation and diastolic function, which may be related not only to alterations in the cross-bridge cycling and/or detachment but also to alterations in Ca2<sup>+</sup> sensitivities, contributing to the pathogenic effects of this mutant. Interestingly, Brunet and coworkers proposed a model to explain the structural and molecular role of both cTnT R278C and cTnI R145G. These mutations appear to affect the sliding event, indicating possible molecular explanations for the observed diastolic dysfunction (Brunet et al., 2014). Additionally, some cTnT mutations (e.g., R92Q and K280N) appear to be insensitive to the regulatory role of cTnI phosphorylation (Messer et al., 2016).

### TROPONIN C

Different TnC isoforms are expressed in human cardiac/slow skeletal and fast skeletal muscle cells and are encoded by the TNNC1 and TNNC2 genes, respectively. cTnC presents ∼70% identity with the skeletal form. During evolution, the threedimensional arrangements were widely conserved between both isoforms but diverged in their ability to bind Ca2+. cTnC plays an important role in the regulation of muscle contraction and relaxation due to the binding of Ca2+. TnC belongs to the superfamily of EF-hand proteins, comprising two globular domains connected by a central alpha helix (Herzberg and James, 1985; Sundaralingam et al., 1985). A canonical EF-hand motif is composed of two alpha-helices surrounding a loop segment that is responsible for divalent ion coordination. EFloops are flexible and enriched with negatively charged amino acids, such as aspartic and glutamic acid. The basic coordination geometry comprises a pentagonal bipyramidal arrangement, in which seven chelating groups are responsible for ion connection. Importantly, this is the same arrangement observed in solution. The six chelating residues are classified first based on the linear position and second by aligning the geometric axis of a pentagonal bipyramid: 1 (+X), 3 (+Y), 5 (+Z), 7 (−Y), 9 (−X), and 12 (−Z). Of note, the carboxylate group in the side chain of the residue at the twelfth position (glutamic acid in ∼92% of cases) provides a bidentate bond (Gifford et al., 2007). The C-terminal domain of TnC, also called the structural domain, is essential for the interaction with thin filaments and is able to bind Ca2<sup>+</sup> with high affinity (∼10<sup>7</sup> M−<sup>1</sup> ) and competitively bind Mg2<sup>+</sup> (∼10<sup>3</sup> M−<sup>1</sup> ). Although the cNTnC has two Ca2<sup>+</sup> binding sites, site I is inactive due to several loop substitutions that impair Ca2<sup>+</sup> coordination, for example, the inclusion of hydrophobic residues instead of charged residues in the +X and +Y positions (Gillis et al., 2007). Thus, contraction starts when Ca2<sup>+</sup> binds to site II, conferring upon the N-terminal domain a regulatory role (Kobayashi and Solaro, 2005). It is known that the amount of hydrophobic exposure driven by Ca2<sup>+</sup> binding at the N-terminal domain directly influences the strength of the Ca2<sup>+</sup> signal transmitted through the thin filament (Li et al., 1999). Under physiological conditions, the binding of Ca2<sup>+</sup> at the N-terminal domain is not sufficient to trigger an open cTnC state. To achieve an active state, two factors are important: (i) the binding of Ca2<sup>+</sup> at site II and (ii) the interaction between cTnC and cTnI leading to changes in the number of hydrophobic patches and to the further stabilization of the open active state. Both factors are essential for contraction.

cTnC mutations associated with the HCM phenotype (**Figure 3**) alter two key mechanisms: the affinity for Ca2<sup>+</sup> and the cTnC cellular partner interactions. Currently, the frequency of cTnC mutations is comparable to that of other targets, such as α-Tm and actin (Van Driest et al., 2003). The first described mutation was reported in a 60-year-old male patient presenting clinical signs of atrial fibrillation and hypertrophy of the left ventricle walls (Hoffmann et al., 2001). DNA sequencing revealed a T-to-A substitution at position 112, hence causing a leucine-to-glutamine exchange at amino acid 29. This mutation corresponds to the +X position within inactive site I; however, L29Q can change the Ca2<sup>+</sup> sensitivity (Schmidtmann et al., 2005; Dweck et al., 2008; Liang et al., 2008; Neulen et al., 2009; Gollapudi and Chandra, 2012). Although divergent data exist with regard to whether this mutant leads to a decrease or increase in the cTnC Ca2<sup>+</sup> affinity, the presence of Gln at position 29 may destabilize the A helix (Liang et al., 2008), thereby disturbing the Ca2<sup>+</sup> binding properties at site II. Additionally, the A helix plays a role in the opening of cNTnC, an important stepwise mechanism for contraction (Li et al., 2000). Compared to wt, the L29Q mutant was insensitive to the interaction with a cTnI N-terminal region regardless of the presence of cTnI phosphorylation, suggesting desensitization to important biological contacts (Baryshnikova et al., 2008a; Li et al., 2013; Messer and Marston, 2014). Furthermore, using nuclear magnetic resonance to measure the N-terminal domain backbone dynamics of salmonid orthologous cTnC (ScNTnC), site I was more flexible than site II. ScNTnC displays a Gln at position 29 and results in a more open structure and a larger solventaccessible area (Blumenschein et al., 2004). Finally, structural and functional assays revealed that the overall structure of L29Q has not changed, however small conformational dynamics were observed (Robertson et al., 2015). Based on the physiological mechanism of cTnC, these features help to understand the alterations in Ca2<sup>+</sup> sensibility and protein-protein interactions. Another mutation located within site I was reported in a 5-yearold boy (Parvatiyar et al., 2012). The alanine exchange at position 31 to a serine was reported as a de novo mutation and introduces a polar amino acid into the Y position. This alanine is a highly conserved residue among different species, and the impact of this mutation results in severe alterations in the Ca2<sup>+</sup> binding properties. Functional studies have revealed that A31S increases the Ca2<sup>+</sup> affinity in isolated cTnC or in thin filaments. Moreover, A31S increases actomyosin ATPase activation and enhances thin filament activation (Parvatiyar et al., 2012). Mutagenesis and three-dimensional visualization showed that a hydroxyl group promotes an additional hydrogen bond with D33, thus providing rigidity to site I in a similar conformation as that for the skeletal TnC after Ca2<sup>+</sup> binding (Parvatiyar et al., 2012).

Four other mutations associated with HCM phenotype were described, i.e., A8V, C84Y, E134D, and D145E (Landstrom et al., 2008). Functional studies have revealed that the E134D mutation has very similar Ca2+-binding affinities and force development as those of the wt (Pinto et al., 2009). Thus, this variant does not appear to be pathogenic, at least regarding these studied parameters. However, C84Y showed an increase in the Ca2<sup>+</sup> sensitivity of force development and force recovery, in addition a sensitized ATPase activity of reconstituted myofilaments (Pinto et al., 2009). The A8V mutant is located in the N-helix, a region that plays an important role in the Ca2<sup>+</sup> affinity of the Nterminal domain and interacts with helices A and D through hydrophobic and electrostatic interactions in the presence of Ca2<sup>+</sup> (Herzberg and James, 1988; Gagné et al., 1995; Slupsky and Sykes, 1995; Houdusse et al., 1997; Strynadka et al., 1997). This N-helix mutant showed an increased Ca2<sup>+</sup> sensitivity of force development and force recovery but did not appear to affect the intrinsic Ca2<sup>+</sup> binding property (Landstrom et al., 2008). However, in the presence of the thin filament, A8V revealed an increase in the Ca2<sup>+</sup> sensitivity of the reconstituted myofilament (Pinto et al., 2009, 2011a). Accordingly, A8V significantly increased the sensitivity of actomyosin ATPase regardless of the phosphorylation status of cTnI. In addition, A8V mutation led to a slower rate of Ca2<sup>+</sup> dissociation in the presence or absence of phosphorylated TnI (Albury et al., 2012). Thus, it is reasonable that A8V affects cTnI-cTnC interactions or promotes an imbalance in cross-bridges that may further increase the Ca2<sup>+</sup> affinity. Martins and coworkers reported the first animal model carrying the A8V heterozygous mutation. In agreement with in vitro studies, A8V revealed a Ca2<sup>+</sup> sensitizer effect in mice. A8V mice developed right ventricular hypertrophy, hyperdynamic systolic function, atrial enlargement, fibrosis, and myofibrillar disarray, consistent with an HCM phenotype. Additionally, Ca2<sup>+</sup> mishandling contributed to an altered contraction, leading to a severe heart remodeling. A reduction of phosphorylated cTnI (cTnI-P) levels was observed in A8V heart samples. This finding agrees with recent investigations suggesting that in humans, sarcomeric mutations have decreased the levels of cTnI-P (Sequeira et al., 2013; Martins et al., 2015). Finally, a molecular mechanism based on altered interactions between cTnC and cTnI has been proposed as the primary source of functional changes observed for myofilaments carrying the A8V mutation (Zot et al., 2016). The D145E mutant also revealed Ca2<sup>+</sup> sensitizer effects in the myofilament. Interestingly, D145E is located at the +Z position within site IV and is the only mutation presented in the C-terminal domain that displays Ca2+-binding disarray. Steadystate fluorescence studies using IAANS revealed that the Ndomain of D145E has a higher affinity for Ca2<sup>+</sup> than does the wt in an isolated system or within the troponin complex (Pinto et al., 2011a). However, the ability to bind Ca2<sup>+</sup> in the C-terminal domain appears to be drastically reduced (Swindle and Tikunova, 2010).

### TROPOMYOSIN

Tm constitutes a diverse family of proteins that are ubiquitously expressed in eukaryotic cells. Four genes (TPM1, 2, 3, and 4) integrate this multigene family in vertebrates, and each gene can produce different splicing isoforms in specific tissues. Of note, α-Tm is the predominant isoform expressed in the heart and comprises 284 residues. Gene shuffling leads to the expression of at least 40 Tm isoforms, and this multiplicity seems to play a pivotal role in cell maintenance (Gunning et al., 2005). Accordingly, using transgenic mice overexpressing β-Tm in the heart, Palmiter and coauthors showed that the replacement of α-Tm with β-Tm alters its structural and functional properties, leading to abnormal thin filament activation (Palmiter et al., 1996). Moreover, α-Tm plays a central role in regulating actinmyosin interactions that is indirectly controlled by the levels of Ca2+. A three-state model (blocked, closed, and open) has been proposed in an attempt to explain this dynamic and allosteric actomyosin regulation. Under resting conditions and low Ca2<sup>+</sup> levels, the regulatory site of TnC is empty, characterizing the blocked state. Additionally, the interaction between actin and TnI is stabilized, and Tm locks actin-myosin interactions. The closed state is characterized by the binding of Ca2<sup>+</sup> to the regulatory domain of cTnC and the exposure of a hydrophobic patch that further interacts with the C-terminal domain of cTnI. Moreover, the inhibitory peptide of cTnI moves away from actin. This transition state is characterized by the partial exposure of the myosin binding sites triggered by an azimuthal motion of Tm over the actin filament, generating weakly bound cross-bridges (Geeves and Lehrer, 1994; Lehrer, 1994). Finally, the open state is achieved in the presence of myosin heads, leading to the attachment of strong cross-bridges, allowing Tm to shift further on the actin filament and potentiate thin filament activation. Altogether, a tightly allosteric regulation through dynamic interactions is required to accommodate Tm along with the thin filament (Gordon et al., 2000; Kobayashi and Solaro, 2005; Solaro, 2010).

The human α-Tm is a coiled-coil dimer rolling over seven continuously actin monomers that provides actin filament support and the anchoring of the troponin complex. The α-Tm primary sequence consists of a short range of seven-residue pseudo repeats called a "heptad." These residues are categorized as the form "a-b-c-d-e-f-g," in which the a and d positions are often occupied by non-polar residues. These residues are responsible for the coiled-coil interface of the interaction and are essential for α-Tm stability. In contrast, the e and g positions present hydrophilic amino acids responsible for inter-helical salt bridges. The b, c, and f positions are exposed to the surface of the coiled-coil domain and are mainly filled with negatively charged residues that interact with the positively charged groove of F-actin (Wolska and Wieczorek, 2003; Barua, 2013; von der Ecken et al., 2015).

Similar to other sarcomeric genes, TPM gene mutations are also correlated with hypertrophic and dilated cardiomyopathy phenotypes (Thierfelder et al., 1994; Olson et al., 2001) but occur at very low frequencies (around less than 1%), as reported by large-scale studies (Richard et al., 2003; Van Driest et al., 2003). The incorporation of specific α-Tm mutants (A63V, K70T, D175N, and E180G) into adult cardiomyocytes revealed different isometric force measurements at submaximal Ca2<sup>+</sup> concentrations, suggesting that HCM-related α-Tm mutants would predict clinical severity (Michele et al., 1999). Mutant transgenic mice were generated to investigate the pathological alterations triggered by the α-Tm D175N mutant, resulting in a severe impairment of heart contractility, relaxation and increased thin filament activation (Muthuchamy et al., 1999). Most of the HCM-related mutations in α-Tm increase thin filament Ca2<sup>+</sup> sensitivity of force generation (Redwood and Robinson, 2013). However, different Ca2<sup>+</sup> sensitivity measurements were obtained for D175N when considering transfected cardiomyocytes (mutant with a similar behavior to that of wt) and skinned fibers (higher Ca2<sup>+</sup> sensitivity for the mutant). Interestingly, an evaluation of biopsies from two patients carrying the HCM-related α-Tm mutant D175N revealed the in vivo incorporation of the mutation within the vastus lateralis muscle, resulting in an altered contractile function (Bottinelli et al., 1998). The mutant and wt forms were shown to be equally expressed in this biopsies, demonstrating a negative-dominant profile for α-Tm in human muscle cells (Bottinelli et al., 1998). Following this study, different studies using reconstituted thin filaments with a wt/mutant mixture or Tm heterodimers were conducted with different results (Lakdawala et al., 2010; Janco et al., 2012). Of note, the in vitro production of α-Tm heterodimers carrying wt/D175N and wt/E180G revealed that mutations have little effect on dimer assembly and actin affinity compared to wt homodimers (wt/wt), but mutant homodimers have a slightly slower affinity compared to wt (Janco et al., 2012). More interesting, the D175N mutation was recently characterized using cardiomyocytes derived from patient-specific human-induced pluripotent stem cells (hiPSCs), in which D175N-hiPSCs revealed abnormal Ca2<sup>+</sup> transients and prolonged action potentials compared to hiPSCs carrying the myosin-binding protein C Q1061X mutation (Ojala et al., 2016).

Recent studies have focused on the biochemical and biophysical characterization of HCM- and DCM-associated α-Tm mutations as a strategy to better understand the primary effects and consequences triggered by mutations in the long-range communication of the thin filament and specific phenotypes (Chang et al., 2014; Gupte et al., 2015). Additionally, the phosphorylation status of α-Tm and its effects on hypertrophic hearts were recently explored (Schulz et al., 2012, 2013; Schulz and Wieczorek, 2013). Although α-Tm phosphorylation dates back to the eighties (Ribolow and Barany, 1977), the protective link to hypertrophic phenotypes is just now emerging. Transgenic mice carrying the S283A α-Tm mutation that abrogates the α-Tm phosphorylation site exhibit a hypertrophic phenotype and increased protein levels and activity of the Ca2<sup>+</sup> ATPase 2a (Serca) but, surprisingly, no changes in the myofilament Ca2<sup>+</sup> sensitivity or the response to β-adrenergic challenges (Schulz et al., 2012). Accordingly, in a doublemutant transgenic mouse carrying the HCM-associated α-Tm mutant E180G together with S283A, the pathogenic phenotype of hypertrophic hearts was abrogated (Schulz et al., 2013).

Although several studies have had significant contributions, the impact of α-Tm mutations and phosphorylation on the mechanism developed by the protein during excitationcontraction coupling and its correlation with the hypertrophic phenotype remain speculative. Because this coiled-coil complex influences thin and thick filament interactions, we believe from a simplistic viewpoint that alterations in α-Tm would transmit structural changes by allostery on both sides of these contractile units. Therefore, further studies should explore the long-term effects of α-Tm mutations for a clear-cut correlation between the mutagenic profile and the hypertrophic phenotype.

### ALLOSTERIC COMMUNICATION DEFINES THE HIERARCHY OF FUNCTIONAL AND PATHOLOGICAL STATES

More than 80% of eukaryotic proteins share in their primary sequences intrinsically disordered regions (IDRs) flanked by packed domains. The complexity of these multi-domain architectures was in the past attributed only to their folded modules, but recent investigations and emerging techniques in structural biology have shown the key participation of IDRs as dynamic elements triggering signaling hierarchy and tuning protein functionalities. This new "apple of the eyes" has the potential to dissect hidden molecular mechanisms participating in physiological and pathological phenotypes. The premise that function follows structure has been left behind. A particular example is subunit I of the troponin complex (Hoffman and Sykes, 2008; Julien et al., 2010). TnI has IDRs with degrees of conformational flexibility that directly impact biological effects, mostly by changing the conformation and function of its partner, the TnC. The degree of disorder in the cTn complex was recently linked to HCM and DCM-causing mutations. Mutations mostly cause decrease in the disorder of cTnI and cTnT instead of an increase (Na et al., 2016). The structure and dynamics of the N-terminal region of cTnI were also explored, revealing the multiplicity of structural profiles assumed upon binding to the cNTnC (Hwang et al., 2014). Additionally, clear evidence of allostery inward of the troponin complex was revealed using pathogenic troponin T mutations. A mechanism in which changes in one protein indirectly affect a third through dynamic changes in a second protein reflects the allosteric transfer of information that culminates in pathogenic phenotypes (Williams et al., 2016). Moreover, the recovery of thin filament sliding speed of the double mutant cTnI-R145G-cTnT-R278C in comparison with cTnI-R145G alone provide further evidence of allosteric transmission within the Tn complex (Brunet et al., 2014). These allosteric mechanisms may represent an interesting strategy in future pipelines for therapeutic intervention of cardiomyopathies based on distant drugable sites inward of Tn subunits or even in other components of the sarcomere. The binding of different protein modules, recently characterized as supra-domain units (Papaleo et al., 2016), with different times, cellular conditions, and environments provides a new molecular repertoire for tuning communication transfer and is just now emerging.

The assembly of macromolecular complexes, such as the thin and thick filaments involved in muscle contraction, requires a well-tuned hierarchy of events for proper function. Regardless of the importance of this molecular motor in triggering human mobility, power stroke and heartbeat, correct communication among its individual elements is defined by allostery. Noteworthy is the fact that the switch-on and -off of this complex machinery is triggered by the influx and binding of Ca2<sup>+</sup> to the C subunit of

the troponin complex. The linker connecting the TnC structural and regulatory domains was shown to communicate both regions in a synergistic way (Grabarek et al., 1986; Moncrieffe et al., 1999; de Oliveira et al., 2013). The identification and functional characterization of HCM and DCM mutations in cTnC (Hoffmann et al., 2001; Landstrom et al., 2008; Willott et al., 2010) and other sarcomeric proteins (Geisterfer-Lowrance et al., 1990; Thierfelder et al., 1994; Redwood and Robinson, 2013; Chang et al., 2014), the alterations in Ca2<sup>+</sup> sensitivity of force development and ATPase activity (Pinto et al., 2009, 2011a,b; Parvatiyar et al., 2012), and the appearance of the disease phenotype point toward a multidirectional change of allosteric pathways throughout the entire filament. Of note, an HCM-causing mutant at the Ca2+-binding site IV of cTnC (D145E) leads to an increased affinity of the Ca2+-binding site II (Pinto et al., 2009), providing clear evidence for allosteric communications between both domains. The mechanism under which this mutation affects a distal region of cTnC is still unclear but probably altered dynamics at the structural cTnC domain due to ion impairment at site IV, and communication transfer through the N-/C- linker might occur (Swindle and Tikunova, 2010). In agreement with this hypothesis, an H/D exchange analysis of wt cTnC revealed unprotected behavior in residues linking both domains, suggesting a higher mobility of this segment (Kowlessur and Tobacman, 2010). A short-term intra allosteric communication of D145E would impose molecular recognition changes in cTnC biological partners that ultimately lead to the HCM phenotype. Extensively dynamic propagation to cTnI inhibitory regions may also occur through the release of Ca2+, thus revealing dynamic adjustments throughout the entire troponin complex (Kowlessur and Tobacman, 2012).

Personalized structural and functional studies of hypertrophic and dilated cardiomyopathy-related mutants provide a framework to start assessing the plethora of short- and long-term allosteric pathways and to unveil the mechanisms behind their pathogenic effects. For example, the L29Q cTnC mutation involved in HCM does not drastically affect protein dynamics but reveals a slight increase in backbone flexibility at the cTnC regulatory domain (Robertson et al., 2015). Because intramolecular dynamics do not explain L29Q effects, changes in the long-term molecular recognition of its biological partners should justify its pathogenic behavior. Indeed, L29Q abolishes the effect of force-generation myosin cross-bridges (Robertson et al., 2015). In the case of A8V, a more open N-terminal domain conformation was observed compared to the wt in the apo and holo states, as revealed by paramagnetic relaxation enhancement (Cordina et al., 2013). In the dilated cardiomyopathy G159D mutation, no abrupt changes in backbone dynamics were observed by T1 and T2 relaxation rates. However, a weak anchoring of cTnI to this mutant was revealed by NMR chemical shifts and NOE connectivity patterns (Baryshnikova et al., 2008b), providing insight to explain the disease phenotype. This weaker interaction probably results in increased levels of acto-myosin inhibition and reduced ATPase activity (Mirza et al., 2005). Changes in allosteric communication are not exclusively related to disease-associated mutations but rather are also related to post-translational modifications. The N-terminal region of cTnI is sensitive to phosphorylation by PKA (Chandra et al., 1997) at specific serine residues (Ser23/24), which results in changes in the Ca2<sup>+</sup> sensitivity of the cNTnC through cTnI itself and through interactions with cTnT (Wattanapermpool et al., 1995; Schmidtmann et al., 2002). The kinase promiscuity of Ser23/24 in cTnI (Solaro et al., 2013) reveals the convergence of multiple signaling pathways and the crucial role of cTnI as a central hub for communication transfer. The phosphorylation of cTnI by PKA enhances the rate of closing the cTnC Nterminal domain induced by Ca2<sup>+</sup> dissociation compared to non-phosphorylated TnI, but this enhancement is abolished by both L29Q and G159D mutations (Dong et al., 2008), revealing a complex scenario of allosteric modulators. Regardless of intra- or inter allosteric changes triggered by cardiomyopathy mutations, the communication within this complex machinery is a highway containing several affluents, and an understanding of the short- and long-term dynamic maps would help to decipher the heterogeneous phenotype of cardiomyopathies and pinpoint new platforms for drug discovery.

### STRUCTURAL BIOLOGY EFFORTS AND TREATMENT IN THE SARCOMERE

Structural biology is an exciting field that focuses on the elucidation of the three-dimensional architecture of biomolecules. This field dates back to the fifties with the breakthrough DNA model of Watson and Crick (Watson and Crick, 1953). The thin filament structure was first reported by Ebashi (1972). Groundwork using X-ray diffraction and electron microscopy provided the basis for Tn and Tm regulation in thin filaments (Huxley, 1972; Spudich et al., 1972; Parry and Squire, 1973). At the end of the nineties, the location of tropomyosin on F-actin filaments was determined from negatively stained electron micrographs (Lehman et al., 2000). Because Tn is repeated every seventh actin molecules, the density distribution of Tn is spread out during EM reconstructions. Thus, through a shorter Tn symmetry in thin filaments using an engineered internal deletion mutant of Tm, in which three of the seven actin-binding pseudo repeats were depleted, the Tn complex could be visualized by EM for three-dimensional reconstruction (Lehman et al., 2001). Considering Tn alone, some valuable efforts were made using X-ray diffraction (Takeda et al., 2003; Vinogradova et al., 2005), but because of limitations due to the flexibility and hydrophobicity of some terminal segments of TnI and TnT molecules, the entire Tn complex (∼80 kD) could not be fully crystallized. Of note, half of the inhibitory segment of TnI (TnI127–148) is not observed in current crystal structures (PDB codes 1J1E and 4Y99), limiting the structural and functional characterization of this inhibitory segment for peptide assays (Lindhout and Sykes, 2003). Crystal structures of different constituents of the thin filament to align EM images were used to better represent this molecular assembly (Pirani et al., 2006; Poole et al., 2006). A single-particle analysis can help to better understand the structural complexity of the thin filament (Paul et al., 2009; Yang et al., 2014). Of note, cryo-EM structures were used to explain possible mechanisms during transition from the close to the open state of the actin:tropomyosin complex (Sousa et al., 2013). The correct orientation of Tn in the acto-Tm complex is a matter of debate. However, a recent application of single-particle procedures for molecular reconstructions has revealed the orientation of troponin on native relaxed cardiac muscle at a resolution of 25 Å (Yang et al., 2014). Regarding the question of thin and thick filament interactions, a recent electron cryomicroscopy structure provided a greater understanding of the human actomyosin complex at 3.9 Å, in which the Factin myosin interface is stabilized by hydrophobic interactions throughout most of the interface (von der Ecken et al., 2016). The use of crystal structure comparisons and combined alignments provided structural details on actin-myosin conformational changes that discriminated the weak and strong myosin-binding states in F-actin. The authors declared that the lack information among states is due to the absence of intermediate conformations of myosin bound to F-actin, making interpretations of whether Pi is released before or after the powerstroke inconclusive (von der Ecken et al., 2015, 2016).

In the case of NMR, one tremendous advantage is the possibility to assess protein dynamics. Using <sup>1</sup>H NMR spectra as a function of temperature, the groundwork from Wagner and Wüthrich revealed that aromatic side chains located inside a hydrophobic protein core may have rotational motions (Wagner and Wüthrich, 1978). The broad research community now accepts that molecular function is intimately related to dynamics. Several biological processes require dynamic transitional states, such as enzyme catalysis, in which key residues should be correctly positioned to coordinate the substrate at the active site, and ligand binding, which requires the entry of small molecules or ions to non-exposed clefts. Furthermore, in molecular recognition events and allostery, protein dynamics act as "short-term memories" to pass information. In addition, intra-motions and intermolecular motions that are transmitted to distal sites, also participate in the transfer of these "memories" for proper protein operation. NMR provides a unique opportunity to assess different biological motions ranging from picoseconds to seconds. For example, the use of the T1 and T2 relaxation rates is suitable for measuring fast dynamics on the order of a pico- to nanosecond timescale, while Carr-Purcell-Meiboom-Gill (CPMG) relaxation dispersion experiments are sensitive to millisecond motions (Carr and Purcell, 1954). Notwithstanding, CEST and DEST experiments are pushing toward the quantification of slow chemical exchange (Vallurupalli et al., 2012). Regardless of the size limitation for NMR studies, several of these experiments have been used to not only provide high-resolution NMR models but also to characterize the dynamics of cardiomyopathy-related mutants of isolated sarcomeric proteins. These data provide a unique strategy to start unveiling the communication transfer and allostery within the sarcomere.

NMR spectroscopy is also a very powerful technique for smallmolecule screenings (Valente et al., 2006). Regarding this issue and focusing on sarcomeric proteins, TnC, the Ca2<sup>+</sup> sensor of the sarcomere, has gained much attention. In addition to playing a crucial role triggering muscle contraction, TnC has become a feasible target for drug discovery because it is easily handled, and upon Ca2<sup>+</sup> binding, TnC exposes "drugable" hydrophobic sites for small-molecule tests. The group of Dr. Brian Sykes has greatly contributed to identifying small-molecule candidates that bind to TnC (Hwang and Sykes, 2015). Both synthetic and natural molecules were studied by NMR and were shown to bind the regulatory and structural domains of TnC, e.g., trifluoperazine (Kleerekoper et al., 1998), bepridil (Li et al., 2000; Wang et al., 2002), levosimendan (Robertson et al., 2008), a W7 inhibitor (Hoffman et al., 2005; Hoffman and Sykes, 2009), EMD57033 (Wang et al., 2001), the flavonoid epigallocatechin gallate (Robertson et al., 2009), and the polyphenol resveratrol (Pineda-Sanabria et al., 2011). Most of these TnC binders act as Ca <sup>2</sup><sup>+</sup> sensitizers and bind to the TnC-TnI148–<sup>163</sup> interface. Recently, a TnC-TnIchimera containing the switch segment of TnI (TnI148–163) was validated as a tool for producing isotopically labeled Tn peptides for NMR structural and drug-screening tests (Pineda-Sanabria et al., 2014). NMR spectroscopy was also used to characterize and validate a bifunctional rhodamine probe attached to cysteines of the skeletal TnC as a strategy for the in situ measurement of the orientation and motions of TnC (Mercier et al., 2003; Julien et al., 2008). Finally, the rational design of Ca2+-sensitizing mutants in the regulatory domain of cTnC is a good strategy for pinpointing key residues that would mimic the characteristic effects of cardiomyopathies and also to understand the properties of ion coordination and their structure-activity relationships (Tikunova and Davis, 2004; Parvatiyar et al., 2010).

While structural biology provides a mean to the understanding of how biomolecules behave in solution and has been used to uncover important molecular mechanisms within the sarcomere, cellular therapy, specifically the use of human pluripotent stem cell (hiPSCs), has the potential to reproduce physiological and pathological phenotypes in vitro (Han et al., 2014; Matsa et al., 2016). Viral transduction of combined transcription factors was able to reprogram fibroblats to pluripotent stem cells, opening up a tremendous advance for personalized medicine care (Takahashi and Yamanaka, 2006). The use of patient-specific cells for reprogramming technologies may work as platforms for the generation of a plethora of new differentiated cells, oriented drug discovery, toxicological studies, and new research-based investigations that mimic disease (Matsa et al., 2014; Guo et al., 2015). In HCM, for example, hiPSC-derived cardiomyocytes revealed hypertrophy, disorganization of the sarcomere and increased expression of genes, e.g., NFAT and calcineurin. More interesting, the use of calcineurin inhibitors have shown to reduce the hypertrophic phenotype (Lan et al., 2013).

### SIGNALING IN HYPERTROPHIC CARDIOMYOPATHY

Adult cardiomyocytes are differentiated and complex cells playing an unambiguous role in cardiac excitation-contraction coupling. This is achieved through a fine-tuned communication signaling of hundreds of molecules organized in weakly and tightly bound molecular assemblies, including cytosolic and transmembrane proteins, ion channels, transcriptional factors, adaptors, and second messengers (**Figure 4**). Unfortunately, this fundamental phenomenon is sometimes challenged by a repertoire of intrinsic and extrinsic insults or by the effects of sarcomeric mutations, the latter being the most common but not exclusive cause of cardiomyopathies. Due to disease complexity, the discussion of signaling changes will focus on HCM, where cardiac dysfunction is mainly credited to heart mass increase

FIGURE 4 | Signaling in hypertrophic cardiomyopathy. Schematic representation of the signaling pathways involved in hypertrophic cardiomyopathy. Green, red and yellow arrows represent promoting activity, inhibitory activity and signaling through Ca2+, respectively. "P" in yellow spheres and "Ub" in red spheres mean phosphorylation and ubiquitination. Molecules with abbreviated names are as follows: Cito C, cytochrome C; Casp, cysteine-aspartic acid protease; Apaf, apoptotic protease activating factor; NFAT, nuclear factor of activated T cells; BNP, brain natriuretic peptide; β-MHC, β-myosin heavy chain; CaM, calmodulin; MLP, muscle LIM protein; T-cap, telethonin; GSK3β, glycogen synthase kinase 3 beta; PDK, 3-phosphoinositide dependent protein kinase 1; PI3K, phosphoinositide 3-kinase; FasR, Fas receptor; TNFα, tumor necrosis factor α; GPCR, G protein-coupled receptor; β-AR, β-adrenergic; AngII, angiotensin II; ET-1, endothelin; AC, adenylyl cyclase; PKA, protein kinase A; PLC, phospholipase C; DAG, diacylglycerol; IP3, inositol 3-phosphate; IP3R, inositol 3-phosphate receptor; SR, sarcoplasmic reticulum; Tn, troponin complex; PLB, phospholamban; Serca, sarco/endoplasmic reticulum Ca2+-ATPase; RyR, ryanodine receptor; and ECM, extracellular matrix. LoF and GoF represent the loss-of-function or gain-of-function phenotype for p53.

(hypertrophy), interventricular septal thickening, and fibrosis. The molecular events encompassing the HCM phenotype are increased protein synthesis, activation of genes expressed during embryogenesis, shifts in myosin isoforms, altered metabolism, and altered organization of the sarcomeric architecture.

Regardless of the sarcomeric repertoire of mutations causing HCM and heart tolerability to mechanical stretch upon heartbeat, the pathways discriminating physiological and pathological mechanosensitivities remain scarce. The correct tuning of biomechanical stress and how molecules sense and trigger hypertrophy is a matter of debate, but proteins located at the Z disc and at the costamere complex may play a role in this mechanotransduction process. The Z disc is responsible for transferring the tension between the sarcomeres through the interaction between titin and α-actinin (Luther and Squire, 2002). Other proteins were also identified in this environment, e.g., the muscle LIM protein (MLP) and telethonin (T-cap), in which mutations were already related to some cardiac diseases, including dilated cardiomyopathy (Arber et al., 1997). Costameres are complex protein assemblies connecting the sarcomere to the extracellular matrix and, together with the Z disc, act as a central station for sensing mechanostress and transmit information to alter contractile properties and transcriptional regulation to avoid heart failure. The mechanotransduction signaling triggered by these different elements aims to decrease blood volume, add sarcomeres to increase contractile capacity and change metabolism to favor energy production. Several lines of evidence link the mechanical signaling response to hypertrophic events, but for more in-depth information, please refer to more specialized literature (Sadoshima and Izumo, 1997 and references therein).

Heart hypertrophy is a compensatory event that is not exclusively related to pathology but rather is also related to the adaptation of the heart to altered contractility and biomechanical stress. Concentric hypertrophy mainly occurs due to pressure overload and is characterized by cardiomyocytes growing laterally with the parallel addition of sarcomeres. From the other side, excentric hypertrophy is due to volume overload, leading to longitudinal cellular growth and the "in-line" addition of sarcomeres (Dorn et al., 2003). Concentric hypertrophy in HCM is related to changes in cardiomyocyte alignment, while excentric hypertrophy is related to advanced HCM (Seidman and Seidman, 2001). One of the signaling pathways tuning the physiological (adaptation) or pathological response to hypertrophy is the phosphoinositide 3-kinase (PI3K)/serinethreonine kinase (Akt) cascade. Depending on the duration of Akt activation, different cellular responses take place, as observed from mice carrying ubiquitously activated Akt. The chronic activation of this kinase results in different phenotypes, ranging from moderate cardiac hypertrophy to massive cardiac dilation and sudden death (Matsui et al., 2002). In addition, Akt inhibits the downstream kinase glycogen synthase kinase 3 beta (GSK-3β) through phosphorylation. Once in the unphosphorylated state, active GSK-3β negatively regulates heart size from pathological insults (Haq et al., 2000; Michael et al., 2004). Transgenic mice expressing a TnT mutation involved in familiar HCM revealed that Akt activation and GSK-3β inactivation impact cardiac size and disease phenotype (Luckey et al., 2009).

Another important signaling triggered by pathological insults is the G-protein-coupled receptor (GPCR). These receptors are activated by factors released upon increased pressure or mechanical stretch, such as angiotensin II and endothelin (ET1), and play a role in hypertrophic responses. Additionally, GPCRdependent PLCβ activation can lead to Ca2<sup>+</sup> efflux from the sarcoplasmic reticulum and the activation of calmodulin. This downstream signaling activates the serine/threonine phosphatase calcineurin that dephosphorylates the nuclear factor of activated T cells (NFAT). Thus, NFAT translocates to the nucleus and reactivates the fetal gene program, which includes the expression of the brain natriuretic peptide (BNP), α-skeletal actin and β-myosin heavy chain (β-MHC) (Molkentin et al., 1998). The reactivation of cardiac fetal genes normally occurs in cardiomyopathies (Kuwahara et al., 2003), and these genes have been shown to be up-regulated in calcineurin transgenic mice (Molkentin et al., 1998), suggesting hypertrophic signaling as an alternative to overcome pathological effects. The increase in the β-MHC isoform and the concomitant decrease in the α-MHC isoform are great biomarkers for early cardiomyopathy (Lowes et al., 1997; Miyata et al., 2000).

The production and deposit of collagen type I and III by fibroblasts are other alterations observed in early cardiomyopathy characterizing fibrosis. Increased deposits may impair the excitation-contraction coupling, leading to severe changes in heart contractility (Menon et al., 2009). Hypertrophic cardiomyopathic hearts expressing gene mutations in α-myosin heavy chain were shown to increase the expression of TGF-β and stimulate non-myocyte cells to proliferate and express profibrotic molecules that ultimately lead to myocyte death, contributing to diastolic dysfunction in HCM hearts (Teekakirikul et al., 2010). The energy supply of cardiomyopathic hearts is also altered. Several changes occur to provide more ATP during the initial stages of cardiomyopathy, with fatty acid oxidation but shifts to glucose at advanced stages of the failing heard (Neubauer, 2007).

Several lines of evidence also link the activation of apoptotic signaling in hypertrophic hearts. An investigation of explanted failing hearts from transplanted patients with cardiomyopathy revealed increased levels of cytochrome C associated with the activation of the pro-apoptotic cysteine-aspartic acid protease 3, caspase-3 (Narula et al., 1999). Apoptotic activation was also observed in myocytes of transgenic mice overexpressing the Gs α-subunit coupled with GPC receptors upon the enhanced activation of β-adrenergic stimulus (Geng et al., 1999). Accordingly, upon norepinephrine stimulation, adult rat cardiac myocytes revealed apoptosis via PKA and voltagedependent calcium channels (Communal et al., 1998). The increased mRNA and protein levels of the proinflammatory cytokine tumor necrosis factor-α (TNFα) in explanted hearts from dilated cardiomyopathic and ischemic patients and the TNFα-induced apoptosis in rat cardiomyocytes demonstrate the clear involvement of this programmed cell death mechanism in cardiomyopathy (Krown et al., 1996; Torre-Amione et al., 1996). Tumor suppressor p53 is also a key cell regulator and responds to DNA damage, inducing cell cycle arrest, senescence, DNA repair and apoptosis (de Oliveira et al., 2015). Interestingly, pacinginduced heart failure in dogs revealed increased protein levels of p53 with the downregulation of Bcl-2 and the upregulation of Bax in myocytes, consistent with an apoptotic scenario (Leri et al., 1998). A recent immunohistochemical study in HCM patients also linked increased levels of p53 and S100A4 proteins with an increased content of collagen fibers, thus, the modulation of these targets may ameliorate myocardial interstitial fibrosis (Qi et al., 2016). With regard to the growing literature on the involvement of aggregated forms of p53 in different cancer cells lines and tumor tissues (Levy et al., 2011; Ano Bom et al., 2012; Lasagna-Reeves et al., 2013; Silva et al., 2014; Yang-Hartwich et al., 2015) and given that p53 is marginally stable inside the cell and that aggregation depends on protein concentration, it is reasonable to argue that misfolded and aggregated forms of p53 might also occur in other unrelated diseases. Indeed, increased levels of endogenous p53 have already been reported in control and stressed myocytes, as well as during p53 overexpression (Miyashita and Reed, 1995; Pierzchalski et al., 1997; Leri et al., 1998). In addition, the short-lived p53 is targeted to the proteasome system due to the E3-ubiquitin ligase activity of its negative regulator, MDM-2. Of note, the decreased proteasome activity reported in HCM and failing human hearts (Predmore et al., 2010) supports the hypothesis of misfolded p53 accumulation and the formation of higher-order oligomers of this tumor suppressor. Furthermore, transgenic mice expressing a myosin-binding protein C mutation that were treated with a proteasome inhibitor were not able to regress the HCM phenotype but rather slightly improved cardiac function (Schlossarek et al., 2014). The lossof-function and gain-of-function phenotypes of oligomeric and aggregated p53 and their involvement in the development of HCM are still matters of speculation and require further exploration.

### REFERENCES


## CONCLUDING REMARKS

The combination of multidisciplinary expertise in structural biology, biochemistry, physiology, and cellular biology represent a sine qua non condition for better understanding the thin/thick filament interactions and ultimately the muscle contraction phenomena under normal and pathological conditions. The impact of mutations and post-translational modifications in sarcomeric proteins and their effects on the generation of nonadaptive hypertrophy are clarified from a series of well-designed studies. Because this review was written by structural biologists, the contribution that should come from this paper is the provision of a point-by-point investigation of the short- and long-term allosteric changes in different pathogenic mutations on isolated proteins and within the context of their biological partners. Because HCM presents a heterogeneous profile of genetic inherence, deciphering the altered motions and dynamic pathways through personalized mutagenic studies will certainly provide insight for new drugable sites that would have the potential for future therapeutics to hopefully ameliorate the hypertrophic phenotype from a broader range of pathogenic mutants.

### AUTHOR CONTRIBUTIONS

All authors listed, have made substantial, direct and intellectual contribution to the work, and approved it for publication.

### ACKNOWLEDGMENTS

This work was supported by grants from the Carlos Chagas Filho Foundation for Research Support in the State of Rio de Janeiro (FAPERJ - E-26/201.817/2015) and Cancer Foundation (to GAPdO). MdAM is the recipient of a doctoral fellowship from the National Council of Technological and Scientific Development.


of PKA phosphorylation and involvement in cardiomyopathies. J. Mol. Biol. 375, 735–751. doi: 10.1016/j.jmb.2007.10.062


of the regulatory N-domain of isolated cardiac troponin C. Biochemistry 52, 1950–1962. doi: 10.1021/bi4000172


L29Q affects Ca2<sup>+</sup> binding and myofilament contractility. Physiol. Genomics 33, 257–266. doi: 10.1152/physiolgenomics.00154.2007


gain of function in skeletal muscle myopathy patients. Hum. Mol. Genet. 22, 4978–4987. doi: 10.1093/hmg/ddt345


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Marques and de Oliveira. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Why Is there a Limit to the Changes in Myofilament Ca2+-Sensitivity Associated with Myopathy Causing Mutations?

### Steven B. Marston\*

*National Heart & Lung Institute, Imperial College London, London, UK*

#### Edited by:

*P. Bryant Chase, Florida State University, USA*

#### Reviewed by:

*Murali Chandra, Washington State University, USA Michelle Parvatiyar, University of California, Los Angeles, USA Nicolas Brunet, SUNY Downstate Medical Center Brooklyn, USA*

> \*Correspondence: *Steven B. Marston s.marston@imperial.ac.uk*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *06 July 2016* Accepted: *05 September 2016* Published: *26 September 2016*

#### Citation:

*Marston SB (2016) Why Is there a Limit to the Changes in Myofilament Ca2*+*-Sensitivity Associated with Myopathy Causing Mutations? Front. Physiol. 7:415. doi: 10.3389/fphys.2016.00415* Mutations in striated muscle contractile proteins have been found to be the cause of a number of inherited muscle diseases; in most cases the mechanism proposed for causing the disease is derangement of the thin filament-based Ca2+-regulatory system of the muscle. When considering the results of experiments reported over the last 15 years, one feature has been frequently noted, but rarely discussed: the magnitude of changes in myofilament Ca2+-sensitivity due to myopathy-causing mutations in skeletal or heart muscle seems to be always in the range 1.5–3x EC50. Such consistency suggests it may be related to a fundamental property of muscle regulation; in this article we will investigate whether this observation is true and consider why this should be so. A literature search found 71 independent measurements of HCM mutation-induced change of EC<sup>50</sup> ranging from 1.15 to 3.8-fold with a mean of 1.87 ± 0.07 (sem). We also found 11 independent measurements of increased Ca2+-sensitivity due to mutations in skeletal muscle proteins ranging from 1.19 to 2.7-fold with a mean of 2.00 ± 0.16. Investigation of dilated cardiomyopathy-related mutations found 42 independent determinations with a range of EC<sup>50</sup> wt/mutant from 0.3 to 2.3. In addition we found 14 measurements of Ca2+-sensitivity changes due skeletal muscle myopathy mutations ranging from 0.39 to 0.63. Thus, our extensive literature search, although not necessarily complete, found that, indeed, the changes in myofilament Ca2+-sensitivity due to disease-causing mutations have a bimodal distribution and that the overall changes in Ca2+-sensitivity are quite small and do not extend beyond a three-fold increase or decrease in Ca2+-sensitivity. We discuss two mechanism that are not necessarily mutually exclusive. Firstly, it could be that the limit is set by the capabilities of the excitation-contraction machinery that supplies activating Ca2<sup>+</sup> and that striated muscle cannot work in a way compatible with life outside these limits; or it may be due to a fundamental property of the troponin system and the permitted conformational transitions compatible with efficient regulation.

#### Keywords: muscle regulation, Ca2+-sensitivity, troponin C, HCM, DCM, myopathy, mutation

**Abbreviations:** HCM, hypertrophic cardiomyopathy; RCM, Restrictive cardiomyopathy; DCM, dilated cardiomyopathy; EC50, Ca2<sup>+</sup> concentration that gives 50% maximal activation; pCa50, –log EC50.

Mutations in striated muscle contractile proteins have been found to be the cause of a number of inherited muscle diseases; in most cases the mechanism proposed for causing the disease is derangement of the thin filament-based Ca2+ regulatory system of the muscle. Hypertrophic cardiomyopathy and hypercontractile diseases of skeletal muscle, such as distal arthrogryposis and "stiff child syndrome," have been linked to a higher myofilament Ca2+-sensitivity (Marston, 2011; Donkervoort et al., 2015). In contrast dilated cardiomyopathy mutations are commonly, but not exclusively, linked to decreased Ca2+-sensitivity. Mutations in contractile proteins that are linked to nemaline myopathy and related skeletal muscle myopathies have also been found to be associated with reduced Ca2<sup>+</sup> sensitivity (Marttila et al., 2012, 2014). The causative connection between myofilament Ca2+-sensitivity and muscle dysfunction is a field of intensive research that is too complex to consider in this account. However, when considering the results of such experiments reported over the last 15 years, one feature has been frequently noted, but rarely discussed. The magnitude of changes in myofilament Ca2+-sensitivity due to myopathycausing mutations in skeletal or heart muscle seems to be always in the range 1.5–3x EC50. Such consistency suggests it may be related to a fundamental property of muscle regulation; in this article we will investigate whether this observation is true and consider why this should be so.

Most investigations have found increased Ca2+-sensitivity in muscle with hypertrophic cardiomyopathy (HCM) and restrictive cardiomyopathy (RCM)-causing mutations. Our literature search found 71 independent measurements of the mutation-induced change of EC<sup>50</sup> ranging from 1.15 to 3.8-fold with a mean of 1.87 ± 0.07 (sem) (**Table 1**). We also found 11 independent measurements of increased Ca2+-sensitivity due to mutations in skeletal muscle proteins ranging from 1.19 to 2.7-fold with a mean of 2.00 ± 0.16 (**Table 2**).

Dilated cardiomyopathy-causing mutations were initially found to decrease Ca2+-sensitivity but more recent studies have indicated the situation is more complex. DCM-linked mutations can both increase and decrease Ca2+-sensitivity depending on the individual mutations, moreover the direction of change can be different with a single mutation measured in different systems (Marston, 2011; Memo et al., 2013). This is illustrated in **Table 3** where 42 independent determinations show a range of EC<sup>50</sup> wt/mutant from 0.3 to 2.3. In addition we found 14 measurements of Ca2+-sensitivity changes due skeletal muscle myopathy mutations ranging from 0.39 to 0.63 (**Table 4**).

Thus, our extensive literature search, although not necessarily complete, found that, indeed, the changes in myofilament Ca2+ sensitivity due to disease-causing mutations have a bimodal distribution and that the overall changes in Ca2+-sensitivity are quite small and do not extend beyond a 3–4-fold increase or decrease in Ca2+-sensitivity. Indeed when all the findings are plotted as a histogram one finds that increases in Ca2+-sensitivity on a log scale have an approximately normal distribution with mean increase in Ca2+-sensitivity (EC<sup>50</sup> wt/mutant) of 1.86-fold (corresponding to 1pCa<sup>50</sup> = 0.255 ± 0.015), whilst the decreases in Ca2<sup>+</sup> sensitivity have a mean EC<sup>50</sup> wt/mutant of 0.54-fold (corresponding to 1pCa<sup>50</sup> of –0.286 ± 0.01; **Figure 1A**). It

#### TABLE 1 | Effect of HCM-associated mutations on myofilament Ca2+-sensitivity.


*(Continued)*

#### TABLE 1 | Continued


*The criteria for inclusion in the table are (1) that a missense mutation has been convincingly linked to the myopathy phenotype and (2) that only direct Ca*2+*-sensitivity comparisons of mutant and "normal" are included. Seventy-one independent measurements of the HCM mutation-induced change of EC*<sup>50</sup> *shown as EC*<sup>50</sup> *WT/mutant. Values range from 1.15 to 3.8-fold with a mean of 1.87* ± *0.07 (sem). Shading indicates gene studied.*

*Gene names: ACTC, cardiac alpha actin; TNNI3, cardiac troponin I; TNNT2, cardiac troponin T (T3 isoform); TNNC2 cardiac troponin C; MYL2, ventricular regulatory myosin light chain; MYH7, beta myosin heavy chain; MYBPC3, cardiac myosin binding protein C; TPM1, alpha tropomyosin, Tpm1.1.*

*Measurement methods: IVMA, in vitro motility assay; Fibers TG, skinned fibers from transgenic or knock-in mouse heart; Myofibrils TG, single myofibrils from transgenic or knock-in mouse heart; Fibers X, skinned fibers with mutation protein exchanged in Human fibers, skinned fibers from human heat muscle; ATPase, reconstituted thin filament activation of myosin ATPase activity.*

is also worth noting that this small Ca2+-sensitivity shift is observed independent of the measurement method **Figure 1B** compares the 1pCa<sup>50</sup> distribution measured by unloaded assays (actomyosin ATPase or in vitro motility) and by loaded assays (force measurements in skinned muscles, cell, and isolated myofibrils). The mean magnitude of the Ca2+-sensitivity change is about 20% less when measured in loaded assays.



*The mean change is 1.65*± *0.16-fold (range 1.19–2.70).*

*GENE NAMES: ACTA1, skeletal muscle alpha actin; TPM2, beta tropomyosin, Tpm2.2; TPM3, Tpm3.12, "gamma tropomyosin."*

*Shading indicates gene studied.*

What could be the underlying reason for this consistent and small effect of mutations on EC50? We will consider two possible mechanisms that are not necessarily mutually exclusive. Firstly, it could be that the limit is set by the capacity of the EC coupling system that supplies activating Ca2<sup>+</sup> and that striated muscle cannot work in a way compatible with life outside these limits; alternatively it may be due to a fundamental property of the troponin system and the permitted conformational transitions compatible with efficient regulation.

Before attempting to discuss these mechanisms it is worthwhile considering some additional evidence on Ca2+ sensitivity shifts. Perhaps the most puzzling observation is that there appears to be no correlation between the Ca2+-sensitivity shift and disease severity. Skeletal myopathy mutations that cause life-threating muscle weakness from birth and often require mechanical assistance in breathing (Ravenscroft et al., 2015), have the same Ca2+-sensitivity shifts as dilated cardiomyopathy mutations which are considerably less lethal (Hershberger et al., 2013). Whilst heart muscle has compensatory strategies not available in skeletal muscle to account for this difference, the small change in Ca2+-sensitivity even in the most severe skeletal muscle disease might be indicative of a fundamental structure-based limit on changes in EC50.

Consideration of the Ca2+-sensitivity shifts in cardiomyopathies (**Tables 1**, **3**) do not indicate any correlation with disease severity. Any relationship that may exist is masked by the extreme variability of Ca2+-sensitivity shift measurements. For instance, the "severe" TNNI3 R145G HCM/RCM-linked mutation features at both extremes of the Ca2+-sensitivity range (1.15x and 3.65x); for the 6 assays in the table the mean is 1.84, close to the mean of all 71 HCM measurements (1.87). The same variability can be seen with other mutations where multiple values are available: ACTC E99K, n = 5, 1.24–2.45 mean 1.85; TPM1 E180G, n = 4, 1.30–2.75, mean 1.78. The second relevant observation is that the physiological modulation of cardiac muscle myofilament Ca2+-sensitivity due to phosphorylation



*Forty-two independent measurements of the mutation-induced change of EC*<sup>50</sup> *shown as EC*<sup>50</sup> *WT/mutant.*

*Shading indicates gene studied.*

of troponin I by protein kinase A has been known to be a 2–3-fold shift for many years (Solaro et al., 2008). **Table 5** lists a number of recent determinations of this Ca2+-sensitivity shift TABLE 4 | Skeletal myopathy mutations causing a loss of function.


*Fourteen independent measurements of the mutation-induced change of EC*<sup>50</sup> *shown as EC*<sup>50</sup> *WT/mutant. The mean change is 0.49* ± *0.02-fold (range 0.36–0.63). Shading indicates gene studied.*

in several species and measured by both loaded and unloaded assays illustrating its small range. **Figure 1C** shows how the magnitude and distribution of measured changes is similar to the changes induced by disease-causing mutations. It would be logical to conclude that this represents the range of achievable Ca2<sup>+</sup> sensitivity shifts in cardiac muscle due to the limitations of the EC coupling system.

In principle, it should be possible to go beyond the Ca2+ sensitivity limits set by EC coupling in an in vitro system where Ca2<sup>+</sup> binding affinity can be much greater or much less than the native troponin. Cardiac troponin C presents extreme examples in a single molecule. Only site II binds Ca2<sup>+</sup> in the physiologically relevant range (2.5 × 10<sup>5</sup> M−<sup>1</sup> ) and so is solely responsible for Ca2+-regulation (Holroyde et al., 1980). A few amino acid changes in the EF-hand motifs results in sites that do not bind Ca2<sup>+</sup> (Site I) or sites that bind Ca2<sup>+</sup> 200x tighter (sites III and IV) and are permanently occupied by Ca2<sup>+</sup> or Mg2<sup>+</sup> (Li and Hwang, 2015). Thus, it would seem that neither a very high Ca2<sup>+</sup> sensitivity nor a very low one are able to participate in regulation. How much deviation of Ca2<sup>+</sup> affinity from the norm is compatible with muscle regulation?

It is known that for mutations, the small Ca2+-sensitivity changes correlate with Ca2<sup>+</sup> binding affinity to thin filaments (Robinson et al., 2007). In a study of mutations induced in skeletal muscle troponin C, Davis et al. achieved a 243-fold range of Ca2<sup>+</sup> binding affinities for troponin C. However, this did not translate into such a great range when Ca2+-binding was measured in the presence of TnI (96-148) and caused a still smaller shift in the Ca2+-sensitivity of force production (Davis et al., 2004). Thus, the most extreme Ca2+-sensitizing mutation, V45Q increased TnC Ca2<sup>+</sup> binding affinity 19-fold, but the increase was only 3.1-fold when measured in the presence of the TnI peptide and Ca2+-sensitivity in skinned fibers was just 2.3-fold more than wild-type. This is within the same

Ca2+-sensitivity due to mutations and phosphorylation. The X-axis is pCa50(mutant-WT, 1pCa50) or EC50 (WT/mutant), log scale. (A) All 149 values from Tables 1–4 are plotted. The plot is bimodal. Mean of decreased Ca2+-sensitivity (1pCa<sup>50</sup> <sup>&</sup>lt; 0) <sup>=</sup> –0.286 <sup>±</sup> 0.016, Mean of increased Ca2<sup>+</sup> *(Continued)*

#### FIGURE 1 | Continued

sensitivity (1pCa<sup>50</sup> >0) = 0.255 ± 0.015. (B) Distribution of change in Ca2+-sensitivity is compared for loaded (pale blue) and unloaded (dark blue) assays of cardiac muscle regulation (data from Tables 1, 3). Unloaded assays are IVMA and ATPase, loaded assays are Fibers TG, Myofibrils TG, Fibers X, Human fibers, For decreased Ca2<sup>+</sup> sensitivity mean unloaded 1pCa50 is –0.27 ± 0.02 and mean loaded is –0.21 ± 0.03, *p* = 0.05. For increased Ca2+-sensitivity mean unloaded 1pCa50 is 0.26 ± 0.02 and mean loaded is 0.021 ± 0.02, *p* = 0.04. (C) Distribution of change in Ca2+-sensitivity due to troponin I phosphorylation (EC50 unphosphorylated/EC50 phosphorylated). Data from Table 5. The mean change is 0.50 ± 0.06-fold (*n* = 9), 1pCa50 = −0.30.

TABLE 5 | Ca2<sup>+</sup> sensitivity change due to troponin I phosphorylation 8 independent measurements of the phosphorylation-induced change of EC50 shown as ratio of EC50 unphosphorylated/phosphorylated (uP/P).


*Measurements were made with troponin (IVMA) or skinned muscle from human (donor) or mouse heart. The mean change is 0.50* ± *0.06-fold (range 0.32–0.74).*

range of many HCM-causing mutations (**Table 1**). A similar picture emerges from Cardiac troponin C where the single regulatory Ca2+-binding site simplifies the argument: V44Q increases Ca2+-binding affinity to TnC 6.5-fold but increases myocyte Ca2+-sensitivity by just 3.4-fold (Parvatiyar et al., 2010). Thus, it seems that the structure of troponin and its interactions with the rest of the thin filament does limit the consequences of a modification that increases Ca2<sup>+</sup> binding affinity.

A slightly different situation arises when Ca2<sup>+</sup> binding affinity is less than wild-type. Davis et al., noted that the mutations that decreased Ca2<sup>+</sup> binding affinity the most (F26Q, 63-fold, I37Q, 24-fold and I62Q, 10-fold) could not properly regulate force in skinned fibers since they only produced about 13% of the maximal force of wild-type muscle at saturating Ca2<sup>+</sup> concentrations. On the other hand, two less extreme mutations, M81Q and F78Q decreased Ca2+-sensitivity whilst retaining the same maximum force production as wild type. In these cases, again, the increased Ca2<sup>+</sup> binding affinity for TnC was substantially greater than the increased Ca2+-sensitivity of skinned fibers (5.9x vs. 1.8x for M81Q and 8.4x vs. 4.2x for F78Q). Thus, thin filament structure seems to limit the possible effects of changes in Ca2+-binding affinity.

It is self-evident that changing myofilament Ca2<sup>+</sup> sensitivity will affect contractile output in muscle. It is well-established that EC<sup>50</sup> for skinned muscle fibers is about 1 µM and

comparison.

that Ca2+-activation of contraction is highly cooperative. Most measurements suggest a five-fold range in free Ca2<sup>+</sup> concentration during a cardiac muscle contraction. Peak Ca2<sup>+</sup> concentration is about 600 nM at rest and can be substantially higher during adrenergic stimulation, thus normally muscle is only partially activated (Negretti et al., 1995; Dibb et al., 2007).

**Figure 2** shows a real life example: in a mouse model of HCM (ACTC E99K) we measured both the Ca2+-activation curve for myofibrils and the contractility of intact papillary muscle as well as the Ca2+-transient (Song et al., 2013). Under the conditions of this experiment the Ca2<sup>+</sup> transient was the same in Wild-type and ACTC E99K muscle, Ca2<sup>+</sup> sensitivity was 0.8 µM for wildtype and 0.34 µM for ACTC E99K with a Hill coefficient of about 4. The increase in Ca2+-sensitivity due to the ACTC E99K HCM mutation corresponds to an approximately four-fold increase in twitch force in the absence of a change in the Ca2+-transient that was actually observed.

We can use this model to consider what would happen if Ca2+-sensitivity changed beyond the normal range. If myofilament Ca2+-sensitivity was 4 times normal, maximum force would reach close to 100%, leaving no range for it to be

### REFERENCES

Baudenbacher, F., Schober, T., Pinto, J. R., Sidorov, V. Y., Hilliard, F., Solaro, R. J., et al. (2008). Myofilament Ca2<sup>+</sup> sensitization causes susceptibility to cardiac arrhythmia in mice. J. Clin. Invest. 118, 3893–3903. doi: 10.1172/jci 36642

modulated by adrenergic agents. Moreover, it is likely that the muscle would not fully relax, since, based on the five-fold range of the Ca2<sup>+</sup> transient even at the lowest Ca2<sup>+</sup> level force would be 5– 10%, a substantial fraction of the peak force of wild-type muscle, thus the hypercontractile phenotype would impose a major defect in relaxation, much more severe than the diastolic dysfunction associated with HCM mutations with only a 1.8-fold average Ca2<sup>+</sup> sensitivity increase.

If myofilament Ca2+-sensitivity were decreased to half the normal, contractility would be very low indeed. The fact that mutations that decrease Ca2+-sensitivity are not lethal and indeed in transgenic mice, may exhibit little phenotype, is probably due to a compensatory increase in the Ca2+-transient (Du et al., 2007). However, this compensation may not be enough to support normal contraction in the long term, leading to DCM, the phenotype commonly associated with reduced Ca2+ sensitivity.

### CONCLUSION

The objective of this article was to confirm that Ca2+ sensitivity of contractility only varies within an narrow range of three-fold above and below the normal EC<sup>50</sup> at rest and to investigate why this should be. The high cooperativity of muscle activation by Ca2<sup>+</sup> means there is a narrow [Ca2+] range between relaxed and active muscle. It would appear that the excitation-contraction coupling machinery of the cell has limited ability to change the amplitude of the Ca2+-transient or baseline [Ca2+] to compensate for changes in EC50; thus increased Ca2+-sensitivity would be limited by inability to relax and reduced Ca2+-sensitivity would be limited by inability to contract. It is intriguing that the Ca2+-sensitivity range of the thin filament itself is independently limited. Mutations that change Ca2+-binding affinity to TnC by a large amount nevertheless only produce a small change in EC<sup>50</sup> for activation of loaded or unloaded contractility in vitro. Whether this property is an evolutionary adaptation that limits the deleterious effects of mutations in thin filaments or simply fortuitous in unknown.

### AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and approved it for publication.

### FUNDING

SM's research is funded by British Heart Foundation programme grant RG/11/20/29266.


induced by troponin I Ser23/24 phosphorylation. Circ. Res. 100, 1486–1493. doi: 10.1161/01.RES.0000267744.92677.7f


ATPase activity and the calcium sensitivity of force development. J. Biol. Chem. 280, 30909–30915. doi: 10.1074/jbc.M500287200


cardiomyopathy susceptibility gene. Circ. Cardiovasc. Genet. 8, 653–664. doi: 10.1161/CIRCGENETICS.114.000957


of contractile proteins. Cardiovasc. Res. 57, 37–47. doi: 10.1016/S0008- 6363(02)00606-5


prevents development of hypertrophic cardiomyopathy in mice. Proc. Natl. Acad. Sci. U.S.A. 112, E4138–E4146. doi: 10.1073/pnas.1505819112

Yuen, M., Cooper, S. T., Marston, S. B., Nowak, K. J., McNamara, E., Mokbel, N., et al. (2015). Muscle weakness in TPM3-myopathy is due to reduced Ca2+-sensitivity and impaired acto-myosin cross-bridge cycling in slow fibres. Hum. Mol. Genet. 24, 6278–6292. doi: 10.1093/hmg/ ddv334

**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Marston. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Dilated Cardiomyopathy Mutation (R134W) in Mouse Cardiac Troponin T Induces Greater Contractile Deficits against α-Myosin Heavy Chain than against β-Myosin Heavy Chain

Sampath K. Gollapudi and Murali Chandra\*

*Department of Integrative Physiology and Neuroscience, Washington State University, Pullman, WA, USA*

Many studies have demonstrated that depressed myofilament Ca2<sup>+</sup> sensitivity is common to dilated cardiomyopathy (DCM) in humans. However, it remains unclear whether a single determinant—such as myofilament Ca2<sup>+</sup> sensitivity—is sufficient to characterize all cases of DCM because the severity of disease varies widely with a given mutation. Because dynamic features dominate in the heart muscle, alterations in dynamic contractile parameters may offer better insight on the molecular mechanisms that underlie disparate effects of DCM mutations on cardiac phenotypes. Dynamic features are dominated by myofilament cooperativity that stem from different sources. One such source is the strong tropomyosin binding region in troponin T (TnT), which is known to modulate crossbridge (XB) recruitment dynamics in a myosin heavy chain (MHC)-dependent manner. Therefore, we hypothesized that the effects of DCM-linked mutations in TnT on contractile dynamics would be differently modulated by α- and β-MHC. After reconstitution with the mouse TnT equivalent (TnTR134W) of the human DCM mutation (R131W), we measured dynamic contractile parameters in detergent-skinned cardiac muscle fiber bundles from normal (α-MHC) and transgenic mice (β-MHC). TnTR134W significantly attenuated the rate constants of tension redevelopment, XB recruitment dynamics, XB distortion dynamics, and the magnitude of length-mediated XB recruitment only in α-MHC fiber bundles. TnTR134W decreased myofilament Ca2<sup>+</sup> sensitivity to a greater extent in α-MHC (0.14 pCa units) than in β-MHC fiber bundles (0.08 pCa units). Thus, our data demonstrate that TnTR134W induces a more severe DCM-like contractile phenotype against α-MHC than against β-MHC background.

Keywords: dilated cardiomyopathy, Troponin T, myosin heavy chain, contractile dynamics, thin filament function, myofilament cooperativity

### INTRODUCTION

Dilated Cardiomyopathy (DCM), a disease caused by mutations in many sarcomeric proteins, is characterized by systolic dysfunction and ventricular dilatation (Kushner et al., 2006; Hershberger et al., 2009; Willott et al., 2010; Marston, 2011; Lu et al., 2013). In vitro studies of DCM-causing mutations in cardiac Troponin T (TnT) generally correlate depressed myofilament Ca2<sup>+</sup> sensitivity to systolic dysfunction (Kushner et al., 2006; Hershberger et al., 2009), with some exceptions

#### Edited by:

*P. Bryant Chase, Florida State University, USA*

#### Reviewed by:

*Julien Ochala, King's College London, UK Douglas Root, University of North Texas, USA*

> \*Correspondence: *Murali Chandra murali@vetmed.wsu.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *01 August 2016* Accepted: *20 September 2016* Published: *04 October 2016*

#### Citation:

*Gollapudi SK and Chandra M (2016) Dilated Cardiomyopathy Mutation (R134W) in Mouse Cardiac Troponin T Induces Greater Contractile Deficits against* α*-Myosin Heavy Chain than against* β*-Myosin Heavy Chain. Front. Physiol. 7:443. doi: 10.3389/fphys.2016.00443*

**105**

(Mirza et al., 2005). Therefore, it remains unclear whether a single determinant, such as myofilament Ca2<sup>+</sup> sensitivity, is sufficient to characterize all cases of DCM because the severity of disease varies widely with a given mutation. Proper pumping actions of the heart dictates that—in addition to normal Ca2<sup>+</sup> dynamics—both the magnitude and speed of contraction are not only sustained but adjusted properly on a beat-to-beat basis. Because contractile dynamics are strongly dependent on thin filament cooperativity, and such cooperativity is modulated by the central region (CR) of TnT (Schaertl et al., 1995; Tobacman et al., 2002; Gollapudi et al., 2013), mutations in the CR of TnT are expected to affect myofilament activation by modifying dynamic features of cardiac contractile activation. A better assessment of disparate cardiac phenotypes is made possible when studies account for the mutation-mediated effect on dynamic contractile function because dynamic aspects dominate heart function under physiological conditions.

The focus of this study is the DCM-related mutation, R131W (Mogensen et al., 2004), which lies within the CR (residues 80–180) of human TnT. We previously demonstrated that the CR of TnT plays an important role in tuning the dynamics of crossbridge (XB) recruitment in cardiac muscle by modulating cooperative mechanisms within thin filaments (Gollapudi et al., 2013). Such actions likely involve strong CR-Tropomyosin (Tm) interactions that take place near the overlap junction of adjacent Tm dimers (Jackson et al., 1975; Pearlstone and Smillie, 1977; Palm et al., 2001; Hinkle and Tobacman, 2003; Gollapudi et al., 2013). Therefore, the R131W mutation in TnT may perturb CR-Tm interactions to modulate XB recruitment dynamics. Given that the TnT-mediated function is dependent on the myosin heavy chain (MHC) isoform (Ford et al., 2012; Chandra et al., 2015; Gollapudi et al., 2015; Gollapudi and Chandra, 2016), we hypothesized that the effects of DCM-linked mutations in TnT on contractile dynamics would be differently modulated by α- and β-MHC. To better understand the molecular mechanisms that lead to contractile dysfunction, especially in relevance to humans, it is important to consider the differential impact of α- and β-MHC isoforms on contractile dynamics. This is because previous studies have demonstrated that the effects of cardiomyopathy mutations in TnT on steady-state and/or dynamic contractile features are differently modulated by α- and β-MHC isoforms (Ford et al., 2012; Chandra et al., 2015; Gollapudi et al., 2015; Gollapudi and Chandra, 2016).

To test our hypothesis, we generated a recombinant mouse TnT equivalent (TnTR134W) of the human DCM mutation, R131W. Various indices of steady-state and dynamic contractile function were measured in normal (α-MHC) and transgenic mouse (β-MHC) cardiac muscle fiber bundles reconstituted with wild-type (WT) TnT (TnTWT) or TnTR134W. Dynamic contractile features mediated by TnTR134W were altered only in α-MHC fiber bundles, despite desensitization of myofilaments to Ca2<sup>+</sup> to a different degree in α- and β-MHC fiber bundles. For instance, TnTR134W attenuated rate constants of tension redevelopment, XB recruitment dynamics, XB distortion dynamics and the magnitude of length-mediated XB recruitment only in α-MHC fiber bundles. We will discuss the correlation between altered contractile dynamics and a more severe DCM-like contractile phenotype against α-MHC than against β-MHC background.

### MATERIALS AND METHODS

### Animal Treatment Protocols

3–4 month-old male mice were used in this study. WT C57BL/6N-strain (non-TG, NTG) mice were acquired from Simonsen's laboratories (Gilroy, CA). β-MHC TG mice were a generous gift from Dr. Jil Tardiff, University of Arizona, Tucson, AZ. The generation and characterization of β-MHC TG mice was as previously described (Krenz et al., 2003, 2007; Krenz and Robbins, 2004; Gollapudi et al., 2015). Mice were carefully handled to minimize pain and suffering, as per the established guidelines of the National Academy of Sciences Guide for the Care and Use of Laboratory Animals. All procedures used for the treatment of mice were approved by the board of Washington State University Institutional Animal Care and Use Committee.

### Purification of Recombinant Mouse Cardiac Tn Subunits

c-myc tagged mouse TnTWT and TnTR134W genes were synthesized (GenScript USA Inc., Piscataway, NJ) after codon optimization for enhanced protein expression. c-myc tagged TnTWT served as the control in our study. Recombinant mouse TnT (TnTWT and TnTR134W), mouse TnI, and mouse TnC were generated and purified, as described previously (Guo et al., 1994; Pan and Johnson, 1996; Chandra et al., 1999; Gollapudi and Chandra, 2012; Mamidi et al., 2013). Recombinant proteins were cloned into the T7 promoter-based pSBETa vector and expressed in BL21∗DE3 cells (Novagen, Madison, WI) for protein synthesis. All proteins were purified using ion-exchange chromatography techniques. TnTWT and TnTR134W were purified by anionexchange chromatography on a DEAE fast Sepharose column (Chandra et al., 1999; Gollapudi and Chandra, 2012; Mamidi et al., 2013; Gollapudi et al., 2015), TnI was purified by cationexchange chromatography on a CM Sepharose column (Guo et al., 1994; Gollapudi and Chandra, 2012; Mamidi et al., 2013; Gollapudi et al., 2015), and TnC was purified by anion-exchange chromatography on a DE-52 column (Pan and Johnson, 1996; Gollapudi and Chandra, 2012; Mamidi et al., 2013; Gollapudi et al., 2015). All eluted fractions containing pure proteins were pooled and dialyzed extensively against deionized water containing 15 mM β-mercaptoethanol, lyophilized, and stored at −80◦C for long-term use.

### Preparation of Detergent-Skinned Cardiac Muscle Fiber Bundles and Tn Reconstitution

Left ventricular papillary muscle bundles were isolated from deeply-anesthetized (Isoflurane) mice, and further dissected into smaller muscle fiber bundles (∼0.15 mm in cross-section

**Abbreviations:** ANOVA, analysis of variance; CR, central region; DCM, dilated cardiomyopathy; HR, high relaxing; LSD, least significant differences; MHC, myosin heavy chain; ML, muscle length; NLRD, nonlinear recruitment distortion; NTG, non-transgenic; RU, regulatory unit; SL, sarcomere length; TG, transgenic; Tm, tropomyosin; Tn, troponin; TnC, troponin C; TnI, troponin I; TnT, troponin T; WT, wild-type; XB, crossbridge.

and 2.0–2.5 mm in length) in high-relaxing (HR) solution (Chandra et al., 2006, 2007; Gollapudi et al., 2015). The HR solution contained 20 mM 2,3-butanedione monoxime (BDM), 50 mM N,N-bis (2-hydroxyethyl)-2-amino-ethane-sulfonic acid (BES), 20 mM ethylene glycol tetra-acetic acid (EGTA), 6.29 mM magnesium chloride (MgCl2), 6.09 mM disodium hydrate salt of adenosine triphosphate (Na2ATP), 30.83 mM potassium propionate (K-Prop), 10 mM sodium azide (NaN3), 1.0 mM dithiothreitol (DTT), and 4 mM benzamidine hydrochloric acid (Benz HCl). Fresh protease inhibitors [in µM: 5 bestatin, 2 E-64, 10 leupeptin, 1 pepstatin, and 200 phenylmethylsulfonyl fluoride (PMSF)] were also included in the HR solution. The pH of the HR solution was adjusted to 7.0 and the ionic strength to 180 mM. The smaller muscle fiber bundles were detergentskinned overnight at 4◦C in HR solution containing 1% Triton X-100.

Recombinant Tn subunits were reconstituted into detergentskinned muscle fiber bundles, as described elsewhere (Chandra et al., 2006, 2007; Mamidi et al., 2013; Gollapudi et al., 2015). Briefly, TnTWT or TnTR134W (0.9 mg/ml, W/V) and TnI (1.0 mg/ml, W/V) were solubilized in an extraction buffer (buffer 1) containing the following (in M): 0.05 Tris-HCl (pH 8.0), 6.0 Urea, 1.0 KCl. High salt and urea in the extraction buffer were removed by successive dialysis against buffers 2-4, whose compositions are listed below.

Buffer 2 (in M): 0.050 Tris-HCl, 4 urea, 0.7 KCl (pH 8.0 at 4◦C) Buffer 3 (in M): 0.050 Tris-HCl, 2 urea, 0.5 KCl (pH 8.0 at 4◦C) Buffer 4 (in mM): 50 BES, 200 KCl, 10 BDM, 6.27 MgCl2, 5 EGTA (pH 7.0 at 20◦C)

All buffers (1-4) included several protease inhibitors (0.2 mM PMSF, 2 mM Benz HCl, 1 mM DTT, and 0.01% NaN3). Any undissolved protein in the extraction buffer was removed by spinning it at 3000 rpm for 15 min. Detergentskinned fiber bundles were treated with the extraction buffer containing TnTWT+TnI or TnTR134W+TnI for ∼3 h at room temperature (22◦C) with gentle stirring. Muscle fiber bundles were then washed twice (10 min each) using buffer 4 and incubated overnight at 4◦C in HR solution containing TnC (3.0 mg/ml, W/V).

### Western Blot Analysis

Reconstituted muscle fiber bundles were solubilized in a muscle protein extraction buffer containing the following: 2.5% SDS, 10% glycerol, 50 mM tris base (pH 6.8 at 4◦C), 1 mM DTT, 1 mM PMSF, 4 mM Benz HCl, and a fresh cocktail of phosphatase (PhosSTOP) and protease inhibitors (E 64, Leupeptin, and Bestatin). The final concentration of all solubilized protein samples was adjusted to 2 mg/ml using the protein loading dye (125 mM Tris-HCl (pH 6.8), 20% glycerol, 2% SDS, 0.01% bromophenol blue, and 50 mM β-mercaptoethanol). 5 µg of each protein sample was loaded and run on an 8% SDS-gel for optimal separation ofc-myctagged recombinant and endogenous TnT (Gollapudi et al., 2012, 2013; Mamidi et al., 2013). Proteins were then transferred to a polyvinylidene difluoride membrane and TnT was probed using a monoclonal anti-TnT primary antibody (M401134, Fitzgerald Industries Int, Concord, MA), followed by HRP-labeled anti-mouse secondary antibody (RPN 2132, Amersham Biosciences, Piscataway, NJ). The percentage incorporation of the exogenous Tn was determined by the densitometric analysis of the TnT band profiles on the Western blot using ImageJ software (acquired from NIH at http://rsbweb. nih.gov/ij/).

### PCA Solutions and Their Compositions

For tension measurements, muscle fiber bundles were exposed to various solutions with pCa (= −log of [Ca2+]free) ranging from 9.0 to 4.3. The compositions of pCa 9.0 and 4.3 solutions were based on the program by Fabiato and Fabiato (1979), and are listed below.

pCa 9.0 (in mM): 50 BES, 5 NaN3, 10 phosphoenol pyruvate (PEP), 10 EGTA, 0.024 calcium chloride (CaCl2), 6.87 MgCl2, 5.83 Na2ATP, and 51.14 K-Prop

pCa 4.3 (in mM): 50 BES, 5 NaN3, 10 PEP, 10 EGTA, 10.11 CaCl2, 6.61 MgCl2, 5.95 Na2ATP, and 31 K-Prop

In addition, the pCa 9.0 and 4.3 solutions contained 0.5 mg/ml pyruvate kinase, 0.05 mg/ml lactate dehydrogenase, along with fresh protease inhibitors [(in µM): 10 leupeptin, 1000 pepstatin, 100 PMSF, 20 diadenosine penta-phosphate, 10 oligomycin]. The pH and ionic strength of pCa 9.0 and 4.3 solutions were adjusted to 7.0 and 180 mM, respectively.

### Measurements of Steady-State Isometric Tension and ATPase Activity

Simultaneous measurements of steady-state isometric tension and ATPase activity were made, as described previously (de Tombe and Stienen, 1995; Stienen et al., 1995; Chandra et al., 2006, 2007; Gollapudi et al., 2015). T-shaped aluminum clips were used to attach muscle fiber bundles between a motor arm (322C, Aurora Scientific Inc., Ontario, Canada) and a force transducer (AE 801, Sensor One Technologies Corp., Sausalito, CA). The sarcomere length (SL) of the muscle fiber bundles was set to 2.3 µm in HR solution by laser diffraction. Each fiber bundle was subjected to two cycles of maximal activation (pCa 4.3) and relaxation (pCa 9.0), and the SL was re-adjusted to 2.3 µm if necessary. The cross-sectional area (CSA) and the initial muscle length (ML) corresponding to the SL of 2.3 µm were measured for each preparation. Muscle fiber bundles were then bathed in various solutions with pCa ranging from 9.0 to 4.3, one at a time, in a constantly-stirred chamber. The fiberelicited responses in steady-state force and ATPase activity were recorded on a computer at a sampling frequency of 1 kHz. All measurements were made at 20◦C.

Measurements of steady-state ATPase activity under isometric conditions were based on an enzymatically coupled assay, as described previously (de Tombe and Stienen, 1995; Stienen et al., 1995; Chandra et al., 2006, 2007; Gollapudi et al., 2015). Tension cost was determined as the slope of the linear relationship between steady-state tension and ATPase activity at various pCa (de Tombe and Stienen, 1995; Stienen et al., 1995).

### Mechano-Dynamic Studies

Fully activated muscle fiber bundles were subjected to various amplitude stretch/release perturbations. First, we wanted to test if force scaled linearly with muscle length changes; therefore, we subjected the muscle fiber bundles to varying amplitudes of length changes (±0.1% to ±2.0%). Experimental data showing the averaged relationship between force changes (1T) and ML changes (1L) in **Figure 1** clearly indicates that force scales linearly with ML. We used the previously established protocol (Ford et al., 2010) to record force responses to varying amplitude length changes (±0.5, ±1.0, ±1.5, and ±2.0% of ML). Force and length data were sampled at 2 kHz. A nonlinear recruitmentdistortion (NLRD) model was fitted to this family of force responses to estimate the following four model parameters: the magnitude of instantaneous increase in stiffness caused by a sudden increase in ML (ED); the rate by which the sudden MLinduced increase in stiffness decays to a minimum (c); the rate by which a new steady-state force is attained due to the recruitment of new force-bearing XBs, following an increase in ML (b); and the magnitude of increase in the steady-state stiffness caused by the ML-mediated increase in the number of newly-recruited force-bearing XBs (ER). More details on step perturbation protocol and the physiological significance of NLRD model parameters are provided in our previously published works (Ford et al., 2010; Gollapudi et al., 2012; Chandra et al., 2015).

### Measurement of Rate of Tension Redevelopment (Ktr)

The measurement of ktr was based upon the force response to a slightly modified version of the large slack-restretch ML maneuver, originally designed by Brenner and Eisenberg (1986). In brief, the muscle fiber in the steady-state of maximal Ca2<sup>+</sup>

activation (pCa 4.3) was first subjected to a rapid (1 ms) release by 10% of its ML using a high speed length-control device (322C, Aurora Scientific Inc., Ontario, Canada). After holding the fiber at the decreased length for 25 ms, it is quickly stretched past its ML by 10%, following which it was rapidly brought back to its ML and allowed to redevelop force. ktr was estimated by fitting the following mono-exponential function to the rising phase of the resulting force (F) response:

$$F(t) = (F\_{\rm ss} - F\_{\rm res})(1 - e^{-k\_{\rm tr}t}) + F\_{\rm res}$$

where Fss is steady-state force and Fres is residual force.

### Data Analysis

Normalized pCa-tension relationships were fitted to the Hill equation to derive pCa<sup>50</sup> (an index of myofilament Ca2<sup>+</sup> sensitivity) and n<sup>H</sup> (an index of myofilament cooperativity). We used a two-way ANOVA to analyze the contractile function parameters because our experimental model involved two factors, TnT (TnTWT and TnTR134W) and MHC (α-MHC and β-MHC). First, we assessed if the MHC-TnT interaction effect on a given contractile parameter was significant. A significant MHC-TnT interaction effect does not suggest a direct interaction between MHC and TnT but it demonstrates that the effects of TnTR134W on a parameter are dissimilar in α- and β-MHC fiber bundles. When the MHC-TnT interaction effect was not significant, we interpreted the main effect of TnT. To probe the cause for a significant MHC-TnT interaction effect or a main effect of TnT, multiple post-hoc t-tests were carried out using uncorrected Fisher's Least Significant Difference (LSD) method. Statistical significance was set at P < 0.05. Data are expressed as mean ± standard error of the mean (SEM).

### RESULTS

### Incorporation Levels of Recombinant TnT in α- and β-MHC Fiber Bundles

We have previously demonstrated that the expression level of β-MHC in TG mouse hearts was ∼70% of the total MHC (Gollapudi et al., 2015). This overexpression of β-MHC had no impact on either the stoichiometry of other sarcomeric proteins or the phosphorylation levels of contractile regulatory proteins (Gollapudi et al., 2015). We used the Western blot to quantify the extent of recombinant Tn incorporation into muscle fiber bundles. The addition of an 11-amino acid c-myc tag at the Nterminus of recombinant TnT proteins (TnTWT or TnTR134W) allowed us to separate the recombinant and endogenous TnT on an SDS gel, and to assess the extent of recombinant Tn incorporation in muscle fiber bundles. The inclusion of cmyc epitope had no impact on the TnT-mediated function in cardiac muscle (Tardiff et al., 1998; Montgomery et al., 2001). A representative Western blot showing the incorporation levels of recombinant TnT in α- and β-MHC fiber bundles is presented in **Figure 2**. Densitometric analysis revealed that the incorporation levels of TnTWT and TnTR134W in α-MHC fiber bundles were 93% and 74%, while those in β-MHC fiber bundles were 90% and 72%, respectively. Similar incorporation levels of TnTR134W

in both α- and β-MHC fiber bundles provided a good model to probe the interplay between TnTR134W- and MHC-mediated effects on various contractile parameters.

### TnTR134W-Mediated Impact on Ca2+-Activated Maximal Tension and E<sup>D</sup> in α- and β-MHC Fiber Bundles

We assessed whether TnTR134W altered maximal activation in an MHC-dependent manner by analyzing the steady-state tension measurements at pCa 4.3. Two-way ANOVA of maximal tension did not reveal a significant MHC-TnT interaction effect (P = 0.35) or a main effect of TnT (P = 0.50). This is because TnTR134W showed no impact on maximal tension in either α- or β-MHC fiber bundles. The mean ± SEM values of maximal tension (in mN·mm−<sup>2</sup> ) in α-MHC+TnTWT and α-MHC+TnTR134W fiber bundles were 46.82 ± 1.03 (n = 13) and 46.11 ± 1.60 (n = 12), while those in β-MHC+TnTWT and β-MHC+TnTR134W fiber bundles were 45.50 ± 1.16 (n = 14) and 49.58 ± 1.24 (n = 14), respectively.

Previously, we have demonstrated that maximal tension is correlated to E<sup>D</sup> (Campbell et al., 2004; Mamidi et al., 2013; Chandra et al., 2015). Therefore, to support our observations in maximal tension, we assessed ED. E<sup>D</sup> is an approximate measure of the number of force-bearing XBs in the isometric steady-state prior to ML change (Campbell et al., 2004; Ford et al., 2010). Two-way ANOVA did not show a significant MHC-TnT interaction effect (P = 0.37) on E<sup>D</sup> or a main effect of TnT (P = 0.14). Thus, TnTR134W did not alter E<sup>D</sup> in either αor β-MHC fiber bundles. The mean ± SEM values of E<sup>D</sup> (in mN·mm−<sup>3</sup> ) in α-MHC+TnTWT and α-MHC+TnTR134W fiber bundles were 1041 ± 42 (n = 13) and 895 ± 49 (n = 12), while those in β-MHC+TnTWT and β-MHC+TnTR134W fiber bundles were 980 ± 59 (n = 14) and 951 ± 29 (n = 14), respectively. Similar observations in both maximal tension and E<sup>D</sup> substantiate that TnTR134W did not affect maximal activation regardless of the MHC isoform.

### TnTR134W-Mediated Impact on the pCa-Tension Relationship in α- and β-MHC Fiber Bundles

A comparison of pCa-tension relationships showed that TnTR134W induced a larger rightward shift in the pCa-tension relationship in α-MHC fiber bundles (**Figure 3A**) than in β-MHC fiber bundles (**Figure 3B**). A closer examination of the pCa-tension relationships also revealed that TnTR134W did not alter the steepness of the pCa-tension relationship in α-MHC fiber bundles (**Figure 3A**) but it decreased the steepness in β-MHC fiber bundles (**Figure 3B**). To quantify the magnitude of such effects in α- and β-MHC fiber bundles, we analyzed the Hill model-derived parameters, pCa<sup>50</sup> (myofilament Ca2<sup>+</sup> sensitivity) and n<sup>H</sup> (myofilament cooperativity). Two-way ANOVA of pCa<sup>50</sup> did not show a significant MHC-TnT interaction effect (P = 0.17) but showed a significant main effect of TnT (P < 0.001). Post-hoc analysis revealed that TnTR134W significantly decreased pCa<sup>50</sup> in both α- and β-MHC fiber bundles; however, the magnitude of attenuation was different (**Figure 3C**). For example, TnTR134W significantly attenuated pCa<sup>50</sup> by 0.14 pCa units (P < 0.001) in α-MHC fiber bundles and by 0.08 pCa units (P < 0.001) in β-MHC fiber bundles. These observations suggest that TnTR134W decreases myofilament Ca2<sup>+</sup> sensitivity to a greater extent in the presence of α-MHC than in the presence of β-MHC. To quantify such changes in pCa<sup>50</sup> in terms of tension, we also compared the steady-state tension data at submaximal Ca2<sup>+</sup> activation (pCa 5.5) among groups. Our analysis showed that TnTR134W significantly attenuated tension at pCa 5.5 by 46% (P < 0.001) in α-MHC fiber bundles and by 26% (P < 0.01) in β-MHC fiber bundles. These observations substantiate that, at submaximal Ca2<sup>+</sup> levels, the attenuating effect of TnTR134W on thin filament activation is stronger in α-MHC than in β-MHC fiber bundles.

Two-way ANOVA of n<sup>H</sup> showed a significant MHC-TnT interaction effect (P < 0.05), suggesting that the TnTR134W-mediated impact on n<sup>H</sup> was dissimilar in αand β-MHC fiber bundles. Post-hoc analysis revealed that TnTR134W showed no effect (P = 0.30; **Figure 3D**) on n<sup>H</sup> in α-MHC fiber bundles, but it significantly decreased n<sup>H</sup> by 26% (P < 0.001; **Figure 3D**) in β-MHC fiber bundles. These observations suggest that TnTR134W does not affect myofilament cooperativity in the presence of α-MHC, but attenuates myofilament cooperativity in the presence of β-MHC. Collectively, these observations demonstrate that α- and β-MHC isoforms differently modulate the TnTR134W-mediated impact on thin filaments at submaximal Ca2<sup>+</sup> activation.

not significant). The number of fiber bundles measured is as follows: 13 for α-MHC+TnTWT, 12 for α-MHC+TnTR134W, 14 for β-MHC+TnTWT, and 14 for β-MHC+TnTR134W. Data are presented as mean ± SE. Standard error bars are smaller than symbols in some cases.

### TnTR134W-Mediated Impact on XB Detachment Kinetics in α- and β-MHC Fiber Bundles

To determine whether TnTR134W affected the XB detachment rate (g) in an MHC-dependent manner, we assessed tension cost and c. Tension cost was estimated as the slope of the linear relationship between tension and ATPase data at various pCa (de Tombe and Stienen, 1995; Stienen et al., 1995; Ford and Chandra, 2013). Within the context of a two-state XB model (Huxley, 1957), the ratio of steady-state ATPase activity (fg/(f + g)) and tension (f/(f + g)) is proportional to g; thus, tension cost is an approximate measure of g. c, which is the rate constant of the immediate force decay, following a sudden change in ML (Ford et al., 2010), is also a measure of g because it is positively correlated to tension cost (Campbell et al., 2004).

A comparison showed that TnTR134W induced a downward shift in the tension-ATPase plot in α-MHC fiber bundles (**Figure 4A**), which suggested a decrease in the slope of this relationship. On the other hand, TnTR134W showed no effect on the tension-ATPase plot in β-MHC fiber bundles (**Figure 4B**). These disimilar effects of TnTR134W on tension cost in αand β-MHC fiber bundles gave rise to a significant MHC-TnT interaction effect (P < 0.001). Post-hoc t-tests showed that TnTR134W significantly decreased tension cost by 17% (P < 0.001; **Figure 4C**) in α-MHC fiber bundles, while it showed no effect (P = 0.39; **Figure 4D**) in β-MHC fiber bundles. Observed effects in tension cost were also validated by our findings in c. A comparison of force responses to 2% stretch showed that TnTR134W induced a rightward shift in the immediate force decay phase in α-MHC fiber bundles (**Figure 5A**), which suggested a slower c. However, TnTR134W showed no effect on the immediate force response in β-MHC fiber bundles (**Figure 5B**). Two-way ANOVA showed a significant MHC-TnT interaction effect (P < 0.01) on c, which suggested that the effect of TnTR134W on c was different in α- and β-MHC fiber bundles. Post-hoc t-tests showed that TnTR134W significantly decreased c by 15% (P < 0.01; **Figure 5C**) in α-MHC fiber bundles, while it showed no effect (P = 0.21; **Figure 5D**) in β-MHC fiber bundles. Similar effects in tension cost and c suggest that TnTR134Winduced changes in thin filaments interact differently with those induced by α- and β-MHC isoforms to differently modulate the effect on g.

presented as mean ± SE.

### TnTR134W-Mediated Impact on XB Turnover Rate in α- and β-MHC Fiber Bundles

To determine whether TnTR134W differently altered XB turnover rate in α- and β-MHC fiber bundles, we assessed two independent rate parameters, ktr and b. While ktr represents the rate of force redevelopment following a large release-restretch length maneuver (Brenner and Eisenberg, 1986), b describes the rate of delayed force rise following a sudden stretch in ML (Ford et al., 2010). Both ktr and b have been previously shown to be approximate measures of XB turnover rate (Ford et al., 2010; Gollapudi et al., 2013, 2015; Chandra et al., 2015).

A representative comparison of force responses to a large release-restretch length meneuver showed that, in α-MHC fiber bundles (**Figure 6A**), TnTR134W induced a rightward shift in the rising force phase, which suggested a slower rate of force rise. On the contrary, TnTR134W showed no effect on the force response in β-MHC fiber bundles (**Figure 6B**). Therefore, two-way ANOVA showed a significant MHC-TnT interaction effect on ktr (P < 0.05), which suggested that the effects of TnTR134W on ktr against α- and β-MHC were dissimilar. Post-hoc analysis confirmed that TnTR134W attenuated ktr by 14% (P < 0.01; **Figure 6C**) in α-MHC fiber bundles, while it showed no effect (P = 0.57; **Figure 6C**) in β-MHC fiber bundles. Our analysis of b also revealed similar findings. A comparison of force responses to a 2% stretch showed that TnTR134W induced a rightward shift in the delayed force rise phase in α-MHC fiber bundles (**Figure 5A**), which suggested attenuation of b. However, TnTR134W displayed no effect on the delayed force rise phase in β-MHC fiber bundles (**Figure 5B**). These differential effects of TnTR134W on b in α- and β-MHC fiber bundles gave rise to a significant MHC-TnT interaction effect (P < 0.01). Post-hoc analysis confirmed that TnTR134W attenuated b by 17% (P < 0.001; **Figure 6D**) in α-MHC fiber bundles, while it showed no effect (P = 0.86; **Figure 6D**) in β-MHC fiber bundles. Therefore, similar effects in ktr and b demonstrate that the XB turnover rate is attenuated by TnTR134W only in the presence of α-MHC.

### TnTR134W-Mediated Impact on E<sup>R</sup> in α- and β-MHC Fiber Bundles

To investigate whether TnTR134W differentially altered the magnitude of stretch activation in α- and β-MHC fiber bundles, we assessed estimates of E<sup>R</sup> at maximal Ca2<sup>+</sup> activation (pCa 4.3). E<sup>R</sup> represents the magnitude of muscle length-mediated recruitment of new force-bearing XBs (ER) and is equivalent to the magnitude of stretch activation (Campbell and Chandra, 2006; Stelzer et al., 2006b, 2007; Ford et al., 2010). E<sup>R</sup> is derived

as the slope of the linear regression between (Fnss − Fss) and 1L (see **Figures 5A,B**), where Fnss is the force corresponding to the new-steady state attained after the change in ML, Fss is the steady-state isometric force prior to the change in ML, and 1L is the imposed ML change. Thus, E<sup>R</sup> increases when Fnss increases and vice versa. Comparison of force responses to a 2% stretch showed that TnTR134W attenuated Fnss in α-MHC fiber bundles (**Figure 5A**), which suggested a decrease in ER. On the other hand, TnTR134W showed no effect on Fnss in β-MHC fiber bundles (**Figure 5B**). These different effects of TnTR134W on E<sup>R</sup> in α- and β-MHC fiber bundles gave rise to a significant MHC-TnT interaction effect (P < 0.05). Post-hoc analysis showed that TnTR134W significantly decreased E<sup>R</sup> by 24% (P < 0.01; **Figure 7**) in α-MHC fiber bundles, while it showed no effect (P = 0.92; **Figure 7**) in β-MHC fiber bundles. These observations demonstrate that the magnitude of stretch activation mediated by TnTR134W is differently altered by α- and β-MHC.

### DISCUSSION

The severity of DCM phenotypes in humans varies so widely that a commonly attributed change in a steady-state contractile parameter, such as a modest decrease in myofilament Ca2<sup>+</sup> sensitivity, precludes us from explaining disparate cardiac phenotypes. Because dynamic aspects of cardiac contraction dominate under conditions in which the heart muscle normally operates, dynamic contractile indices may provide more meaningful clues to link disparate phenotypes to different mutations. Our extensive steady-state and dynamic contractile data demonstrate that alterations in contractile dynamics (both rate and magnitude), in addition to the differential impact of αand β-MHC on myofilament Ca2<sup>+</sup> sensitivity, allow us to expand our view on how some mutations in TnT affect heart function and cardiac phenotypes.

### TnTR134W Attenuates Myofilament Ca2<sup>+</sup> Sensitivity to a Greater Extent in α-MHC than in β-MHC Fiber Bundles

A greater magnitude of decrease in myofilament Ca2<sup>+</sup> sensitivity in α-MHC+TnTR134W than in β-MHC+TnTR134W fiber bundles (**Figure 3C**) raises two questions: (1) how does TnTR134W attenuate thin filament activation?; and (2) why is the effect of TnTR134W on thin filaments minimized in the presence of β-MHC? Previous studies have associated residues 112–136

α-MHC and (B) β-MHC fiber bundles. Force data were normalized by the isometric steady-state value following the length perturbation. TnTR134W-mediated effect on (C) *k*tr and (D) *b* in α- and β-MHC fiber bundles. Statistical differences were analyzed by two-way ANOVA and subsequent *post-hoc t-*tests using Fishers LSD method. \*\**P* < 0.01 and \*\*\**P* < 0.001 indicate significant results compared to TnTWT (NS, not significant). The number of fiber bundles measured is as follows: 13 for α-MHC+TnTWT, 12 for α-MHC+TnTR134W, 14 for β-MHC+TnTWT, and 14 for β-MHC+TnTR134W. Data are presented as mean ± SE.

of human TnT in the strong interaction of CR of TnT with Tm at the Tm-Tm overlap junction (Hinkle and Tobacman, 2003). This CR-Tm interaction acts as a gateway not only for the activation of regulatory units (RU; Tn-Tm complex) but also for cooperative interactions between near-neighbor RUs and between near-neighbor RUs and XBs (Schaertl et al., 1995; Razumova et al., 2000; Tobacman et al., 2002; Moss et al., 2004). There is evidence to suggest that the R131W mutation in TnT decreases RU activation by increasing the rate of dissociation of Ca2<sup>+</sup> from Tn (Liu et al., 2012). While such attenuation of RU activation may primarily involve altered CR-Tm interaction, another study suggests that the R131W-induced structural changes in the CR may also spread to the T2 region of TnT to modify Ca2+-sensitive interactions between TnT and TnI/TnC (Mogensen et al., 2004). Based on these findings, we posit that TnTR134W alters allosteric/cooperative mechanisms that underlie RU activation. At submaximal [Ca2+], such actions of TnTR134W increase the amount of Ca2<sup>+</sup> required to attain the magnitude of RU activation that is normally observed in TnTWT, leading to attenuation of pCa<sup>50</sup> in both α- and β-MHC fiber bundles. However, the magnitude of the impact on pCa<sup>50</sup> is lower in β-MHC+TnTR134W than in α-MHC+TnTR134W fiber bundles, suggesting that β-MHC partially counters the effect of TnTR134W on RU activation. Given that enhanced XB-RU, but not XB-XB, cooperativity increases pCa<sup>50</sup> (Razumova et al., 2000), the ability of β-MHC to counter the influence of TnTR134W on pCa<sup>50</sup> appears to arise from greater XB-RU cooperativity.

How differences in XB cycling kinetics permit α- and β-MHC (Rundell et al., 2005; Stelzer et al., 2007; Ford and Chandra, 2013) to differently modify XB-RU cooperativity may be gleaned by considering the initial conditions of thin filaments. At submaximal [Ca2+]free, the TnTR134W-induced attenuation of RU activation leaves behind a larger than normal pool of RUs in the off state, thereby increasing the scope for strong XBs to cooperatively influence XB-RU interactions. Therefore, the slow cycling β-MHC may exert a positive effect on thin filaments by amplifying XB-RU cooperativity. This enhanced XB-RU cooperativity by β-MHC is expected to facilitate the transition

of RUs from the off to the on state, thereby resulting in an increase in RU activation and a subsequent increase in the number of force-bearing XBs. Indeed, tension is augmented at submaximal activation in β-MHC+TnTR134W fiber bundles when compared to α-MHC+TnTR134W fiber bundles. This explains why β-MHC is able to partially counter the negative influence of TnTR134W on pCa50. Such β-MHC-mediated increase in XB-RU cooperativity may have exhausted the pool of RUs from which RU-RU cooperativity could recruit (Razumova et al., 2000), thereby decreasing the contributions of RU-RU cooperativity to n<sup>H</sup> in β-MHC+TnTR134W fiber bundles. Because RU-RU cooperativity has the greatest influence on n<sup>H</sup> (Razumova et al., 2000), a decrease in RU-RU cooperativity may likely explain the decrease in n<sup>H</sup> in β-MHC+TnTR134W fiber bundles.

### β-MHC Neutralizes the Attenuating Effect of TnTR134W on XB Turnover Rate and XB Detachment Rate

Our observations on two contractile rate parameters, ktr and b (**Figure 6**), confirm that TnTR134W attenuates XB turnover rate in α-MHC fiber bundles, but shows no effect in β-MHC fiber bundles. In our previous studies (Campbell, 1997; Campbell et al., 2004; Ford et al., 2010; Gollapudi and Chandra, 2012), we have shown that attenuation of b may be brought about by the following: (1) attenuation of RU on/off kinetics; (2) attenuation of XB cycling kinetics, f and g; (3) augmentation of XB-based cooperativity; or (4) a combinatorial effect of 1, 2, and 3. However, unaltered maximal tension and E<sup>D</sup> in α-MHC+TnTR134W fiber bundles suggest that the impact of TnTR134W on RU activation is minimized. Under these conditions, the available pool of RUs and XBs from which XBbased cooperativity may recruit is expected to be similar in both α-MHC+TnTWT and α-MHC+TnTR134W fiber bundles. Thus, it is unlikely that enhanced XB-based cooperativity is responsible for slower b in α-MHC+TnTR134W fiber bundles. Therefore, our observations suggest that attenuation of b may result from the slowing effect of TnTR134W on XB cycling kinetics. Because ktr = f + g, as per the two-state XB model (Huxley, 1957; Brenner, 1988; de Tombe and Stienen, 2007), a significant attenuation of ktr in α-MHC+TnTR134W fiber bundles also substantiates our assertion that attenuation of b is due to a slowing effect on f and/or g.

Evidence to substantiate that attenuation of g may underlie the slowed XB turnover rate (ktr and b; **Figure 6**) in α-MHC+TnTR134W fiber bundles comes from our observations on tension cost (**Figure 4**) and c (**Figure 5**). Furthermore, other observations suggest that the attenuation of XB turnover rate in α-MHC+TnTR134W fiber bundles may also involve a slowing effect on f ; this is because steady-state isometric force is proportional to f/(f + g) (Huxley, 1957). Therefore, a decrease in g alone should increase force produced in α-MHC+TnTR134W fiber bundles. However, both maximal tension and E<sup>D</sup> are unaltered in α-MHC+TnTR134W fiber bundles, which suggests that TnTR134W does not impact f/(f + g) in the presence of α-MHC at maximal activation. Thus, this conjectural evidence may indicate that f decreases in proportion to g in α-MHC+TnTR134W fiber bundles. In contrast, a lack of effect on g in β-MHC+TnTR134W fiber bundles (**Figures 4**, **5**) suggests that β-MHC negates the attenuating effect of TnTR134W on g. Our observation on g, in conjunction with unaltered ktr, b, and maximal tension in β-MHC+TnTR134W fiber bundles, suggests that TnTR134W does not alter f in the presence of β-MHC. These observations demonstrate that the interplay between the TnTand MHC-mediated effects on the thin filament modulate XB cycling kinetics.

### β-MHC Neutralizes the Attenuating Effect of TnTR134W on the Magnitude of Stretch Activation

Another notable finding from our study is that TnTR134W attenuates the magnitude of stretch activation (ER) in α-MHC but shows no effect in β-MHC fiber bundles (**Figure 7**). The magnitude of E<sup>R</sup> is dependent on the ML-related XB recruitment mechanisms that operate within thin filaments. For example, XB-based (XB-RU/XB-XB) cooperativity, which is mediated through thin filaments, strongly influences E<sup>R</sup> such that a decrease in XB-based cooperativity decreases E<sup>R</sup> and vice versa (Campbell et al., 2004; Campbell and Chandra, 2006; Stelzer et al., 2006b). Our data suggest that different outcomes on E<sup>R</sup> in α-MHC+TnTR134W and β-MHC+TnTR134W fiber bundles may be closely linked to differential effects on XB-RU cooperativity. To clarify, a decrease in XB-RU cooperativity may be responsible for the attenuation of E<sup>R</sup> in TnTR134W +α-MHC fiber bundles, while unaltered XB-RU cooperativity explains why E<sup>R</sup> is unaffected in β-MHC+TnTR134W fiber bundles. One source of this difference between α- and β-MHC fiber bundles may be related to our earlier assertion that the negative effect of TnTR134W on RU activation remains more prominent in α-MHC than in β-MHC fiber bundles. Therefore, E<sup>R</sup> is attenuated in α- but not in β-MHC fiber bundles.

### Implications of Our Findings for Heart Function in Mice and Humans

TnTR134W attenuated XB turnover and detachment rates in α-MHC fiber bundles, but not in β-MHC fiber bundles. When extrapolated to the whole heart level, these observations suggest slower rise and slower fall of ventricular pressure in α-MHC containing hearts but not in β-MHC containing hearts. In addition, slower rates of XB turnover and detachment may also slow dynamics of ejection in α-MHC-expressing hearts. Previous studies also implicate mechanisms such as stretch activation in maintaining the ventricular force output during the late phase of ejection (Stelzer et al., 2006a,c). Thus, attenuation of E<sup>R</sup> in α-MHC+TnTR134W fiber bundles, in conjunction with slowed XB turnover and detachment rates, suggests that the ejection phase may be prematurely terminated in α-MHC-expressing hearts. Inferences drawn from dynamic studies demonstrate that the magnitude of cardiac contractile impairment, induced by TnTR134W, differ significantly in αand β-MHC expressing fiber bundles. The effect on pCa<sup>50</sup> also shows that the severity of contractile deficits induced by TnTR134W is different in α- and β-MHC fiber bundles; for instance, TnTR134W decreases pCa<sup>50</sup> to a greater extent in α-MHC fiber bundles (**Figure 3C**). Although the attenuation

### REFERENCES


of pCa<sup>50</sup> alone may suggest DCM in both α- and β-MHC background, the severity of cardiac phenotype is expected to be greater in mouse hearts because various indices of contractile dynamics were attenuated only in α-MHC fiber bundles.

### AUTHOR CONTRIBUTIONS

Contribution of SG: Conception and design, acquisition of data, analysis and interpretation of data, drafting and revising the manuscript. Contribution of MC: Conception and design, interpretation of data, drafting and revising the manuscript.

### FUNDING

This work was supported, in part, by National Institutes of Health Grant No. HL-075643 (to MC) and a Poncin grant supported by the Autzen foundation.

### ACKNOWLEDGMENTS

The authors thank Sherif M. Reda and Alexis V. Mickelson for proof reading the manuscript.


tropomyosin-binding region. Biophys. J. 81, 2827–2837. doi: 10.1016/S0006- 3495(01)75924-3


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Gollapudi and Chandra. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Effects of Disease Models of Nuclear Actin Polymerization on the Nucleus

Leonid A. Serebryannyy <sup>1</sup> , Michaela Yuen2, 3, Megan Parilla<sup>1</sup> , Sandra T. Cooper 2, 3 and Primal de Lanerolle<sup>1</sup> \*

*<sup>1</sup> Department of Physiology and Biophysics, University of Illinois at Chicago, Chicago, IL, USA, <sup>2</sup> Institute for Neuroscience and Muscle Research, Kids Research Institute, The Children's Hospital at Westmead, Sydney, NSW, Australia, <sup>3</sup> Faculty of Medicine, Discipline of Pediatrics and Child Health, University of Sydney, Sydney, NSW, Australia*

Actin plays a crucial role in regulating multiple processes within the nucleus, including transcription and chromatin organization. However, the polymerization state of nuclear actin remains controversial, and there is no evidence for persistent actin filaments in a normal interphase nucleus. Further, several disease pathologies are characterized by polymerization of nuclear actin into stable filaments or rods. These include filaments that stain with phalloidin, resulting from point mutations in skeletal α-actin, detected in the human skeletal disease intranuclear rod myopathy, and cofilin/actin rods that form in response to cellular stressors like heatshock. To further elucidate the effects of these pathological actin structures, we examined the nucleus in both cell culture models as well as isolated human tissues. We find these actin structures alter the distribution of both RNA polymerase II and chromatin. Our data suggest that nuclear actin filaments result in disruption of nuclear organization, which may contribute to the disease pathology.

#### Edited by:

*Jose Renato Pinto, Florida State University, USA*

#### Reviewed by:

*Morayma Reyes, Montefiore Medical Center, USA Tina Tootle, University of Iowa, USA*

> \*Correspondence: *Primal de Lanerolle primal@uic.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *30 July 2016* Accepted: *21 September 2016* Published: *0 October 2016 7*

#### Citation:

*Serebryannyy LA, Yuen M, Parilla M, Cooper ST and de Lanerolle P (2016) The Effects of Disease Models of Nuclear Actin Polymerization on the Nucleus. Front. Physiol. 7:454. doi: 10.3389/fphys.2016.00454* Keywords: nuclear actin, nuclear filaments, skeletal actin mutations, intranuclear rod myopathy, cofilin rods, RNA polymerase II, chromatin, nuclear structure

## INTRODUCTION

Actin is necessary for maintaining cell structure and driving cell movement and contraction (Dominguez and Holmes, 2011). The capacity of actin to be both a structural protein and generate force comes from actin's ability to dynamically polymerize and depolymerize into polymers and filaments of different lengths. Actin also translocates into the nucleus (de Lanerolle and Serebryannyy, 2011; Dopie et al., 2012). Yet, unlike cytoskeletal actin, actin in the nucleus of somatic cells does not polymerize into persistent filaments. Instead, studies suggest nuclear actin exists as monomers and highly dynamic polymers (McDonald et al., 2006). Nevertheless, nuclear actin has been implicated in a variety of processes including regulation of transcription factors (Rajakylä and Vartiainen, 2014), the nuclear lamina (Simon and Wilson, 2011), DNA damage repair (Andrin et al., 2012; Belin et al., 2015), transcription by RNA polymerase (RNAP) I, II, and III (Hofmann et al., 2004; Hu et al., 2004; Philimonenko et al., 2004), hnRNP binding (Percipalle et al., 2009), and is a necessary component of multiple chromatin remodeling complexes (Bremer et al., 1981; Kapoor and Shen, 2014; Serebryannyy et al., 2016a). Intriguingly, instances of long-range nuclear movement of chromatin and nuclear compartments have been reported to be dependent on nuclear actin polymerization (Forest et al., 2005; Chuang et al., 2006; Wang et al., 2006; Dundr et al., 2007; Hu et al., 2008; Mehta et al., 2010; Chang et al., 2011; Khanna et al., 2014). Collectively, nuclear actin seems to be important in regulating gene accessibility, transcription, and post-transcriptional regulation (de Lanerolle and Serebryannyy, 2011).

Despite multiple studies suggesting nuclear actin is able to polymerize (Amankwah and De Boni, 1994; reviewed in de Lanerolle and Serebryannyy, 2011; de Lanerolle, 2012), there are few documented instances of physiological actin filaments in the nucleus. While in Xenopus oocytes nuclear actin filaments form a structural network (Clark and Rosenbaum, 1979; Feric and Brangwynne, 2013; Samwer et al., 2013), there is little direct evidence for a comparable actin-based structure in the normal mammalian nucleus. Fluorescence recovery after photobleaching experiments and actin probes using cloned actin binding domains have indirectly visualized a population of short nuclear actin polymers that exist throughout the nucleus. Additionally, recent reports have indicated that transient nuclear actin filaments are able to form in response to serum stimulation and cell spreading via a formin-mediated mechanism (Baarlink et al., 2013; Plessner et al., 2015). These reports suggest physiological nuclear actin polymerization occurs transiently and their formation of nuclear actin filaments is tightly regulated. However, the role of nuclear actin polymers and transient filaments are still largely a mystery.

Two distinct nuclear actin structures are found in a number of disease states: phalloidin-stainable nuclear actin filaments and cofilin/actin rods. Phalloidin reportedly binds a site accessible in right-handed actin filaments (Dancker et al., 1975; Barden et al., 1987; Drubin et al., 1993; Visegrády et al., 2005). The affinity and specificity of phalloidin binding to actin has made it a widely used marker for actin filaments. Cofilin is commonly considered an actin depolymerizing factor; however, cofilin binding to actin changes its conformation (McGough et al., 1997; Ghosh et al., 2004; Andrianantoandro and Pollard, 2006) and has been shown to form unconventional actin filaments (Nishida et al., 1987). These left-handed cofilin/actin rods are generally not recognized by phalloidin, thus the labeling of actin filaments with phalloidin or incorporation of cofilin indicate different actin structures (Yonezawa et al., 1988; McGough et al., 1997; Kudryashov et al., 2006).

Cofilin/actin rods, which are usually reversible (Nishida et al., 1987; Iida et al., 1992), have been well documented in both the cytoplasm and the nucleus (Bamburg and Wiggan, 2002; Hofmann, 2009). Formation of nuclear cofilin/actin rods has been noted during a variety of cellular stresses including heatshock (Welch and Suhan, 1985; Iida et al., 1986; Nishida et al., 1987), changes in salt concentrations (Iida and Yahara, 1986), overactive cAMP production (Osborn and Weber, 1984), DMSO treatment (Fukui and Katsumaru, 1980; Sanger et al., 1980; Nishida et al., 1987), ATP depletion (Pendleton et al., 2003), actin depolymerization with actin drugs (Yahara et al., 1982), trifluoperazine treatment (Osborn and Weber, 1980), electrical stimulation (Seïte et al., 1973), and adenosquamous cell carcinoma of the axillary sweat glands (Fukuda et al., 1987). Intriguingly, nuclear cofilin/actin rod formation has been implicated in Huntington's disease, whereby abnormal huntingtin protein expression can stabilize the formation of these structures (Munsie et al., 2011). There is also evidence for nuclear microfilament formation in aging neurons, among other diseases (Fiori, 1987; Bamburg et al., 2010). Nevertheless, it remains unclear why these rods form and what their function may be. Individually, nuclear actin and cofilin have both been documented to regulate transcription (Hofmann et al., 2004; Obrdlik and Percipalle, 2011), yet the effects of nuclear cofilin/actin rods have not been studied in the context of transcription or chromatin organization.

Like cofilin/actin rods, stable, phalloidin-stainable actin filaments are also not present in normal mammalian nuclei, however exceptions include baculovirus infection, antisynthase syndrome, and intranuclear rod myopathy (Charlton and Volkman, 1991; Goley et al., 2006; Domazetovska et al., 2007a; Stenzel et al., 2015). Furthermore, expression of certain actin bundling proteins such as α-catenin (Daugherty et al., 2014; McCrea and Gottardi, 2016), supervillin (Wulfkuhle et al., 1999), myopodin (Weins et al., 2001), c-Abl (Aoyama et al., 2013), mutant espins (Loomis et al., 2006), and expressing nuclear targeted actin all form stable, phalloidin-stainable actin filaments (Kokai et al., 2014). Similarly, knockdown of the actin depolymerization factor MICAL-2 (Lundquist et al., 2014) or the nuclear actin export factor, exportin-6 (Dopie et al., 2012) induce formation of phalloidin-stainable nuclear actin filaments. While formin-dependent nuclear actin filaments that form in response to serum stimulation and cell spreading are reported to be highly transient (Baarlink et al., 2013; Plessner et al., 2015), the phalloidin-stainable nuclear actin filaments that form under the above conditions are highly stable and may be detrimental to the cell as is the case in patients with the skeletal muscle disease, intranuclear rod myopathy (Vandebrouck et al., 2010; Serebryannyy et al., 2016b).

Over 180 different mutations occur in the skeletal α-actin gene (Laing et al., 2009; Ravenscroft et al., 2011); most are dominant and de novo. Skeletal myopathies are usually severe, affect many actin functions, and can lead to premature death. Previous experiments have established that certain mutant actins translocate to the nucleus and form stable, phalloidin-stainable actin filaments in the nucleus (Costa et al., 2004; Ilkovski et al., 2004). The formation of nuclear actin filaments in skeletal muscle causes the human disease intranuclear rod myopathy (Jenis et al., 1969; Goebel and Warlo, 1997). Patients with intranuclear rod myopathy usually exhibit a severe clinical phenotype with fatalities caused by diaphragmatic weakness and respiratory failure (Goebel and Warlo, 1997; North et al., 1997; Laing et al., 2009). These intranuclear actin aggregates stain with phalloidin and α-actinin, but rarely with cofilin (Vandebrouck et al., 2010). The actin mutatants that cause intranuclear rod myopathy also demonstrate decreased incorporation into muscle thin filaments (Costa et al., 2004; Ilkovski et al., 2004; Domazetovska et al., 2007a,b). Despite maintaining contractile function, patients continue to experience severe muscle weakness, and cell culture models exhibit a decrease in mitotic index (Domazetovska et al., 2007a,b). Further, mutations in α-actin that cause intranuclear rod myopathy have been suggested to alter serum response factor signaling (Visegrády and Machesky, 2010). These studies implicate an alternate role for actin and suggest a pathogenesis aside from decreased contractility.

Nuclear actin structures have been observed repeatedly in a variety of circumstances. Yet, the function of nuclear actin polymerization and the different nuclear actin structures described above is unknown. Therefore, we sought to determine what effects the formation of nuclear actin structures has on the nucleus. Using cofilin/actin rods and models of intranuclear rod myopathy, we find nuclear actin structures are able to change the localization of RNAPII and chromatin, potentially contributing to their pathogenesis.

### RESULTS

### Organization of RNAPII and Chromatin in Cells with Cofilin/Actin Rods

To investigate the consequences of nuclear actin polymerization on chromatin and RNAPII, we first examined how different acute cellular stresses, which have been documented to form nuclear cofilin/actin rods, affect a mouse neuronal striatal cell line (STHdh) stably expressing cofilin YFP (Munsie et al., 2011). Under normal culture conditions, STHdh cells do not exhibit phalloidin- nor cofilin-labeled actin filaments within the nucleus (**Figure 1A**). However, with certain stimuli such as heatshock, forskolin treatment, or treatment with the actin depolymerization drug, latrunculin B, a population of STHdh cells formed cofilin-labeled rods which did not stain with phalloidin (**Figure 1B**). Latrunculin B treatment resulted in some co-localization between cofilin/rods and phalloidin labeling in large puncta within the nucleus. Additionally, we found heatshock performed in Dulbecco's phosphate-buffered saline (PBS) lacking fetal bovine serum and supplemented with CaCl<sup>2</sup> (0.9 mM) and MgCl<sup>2</sup> (0.5 mM) resulted in a population of cells with nuclear filaments that stained with phalloidin and incorporated cofilin (**Figure 2A**) as well as neighboring populations that exhibited only cofilin/actin rods or phalloidinstainable filaments but not both (**Figure 2B**). This suggests that the extracellular environment and intrinsic properties of each cell may modulate the type of nuclear actin structure formed.

To determine if cofilin/actin rods have an effect on the organization of the nucleus, we induced rod formation in STHdh cells, fixed, then stained these cells with RNAPII antibodies (**Figure 3A**). Superresolution structured illumination microscopy (SIM) showed that RNAPII is redistributed away from areas with cofilin/actin rods. Furthermore, confocal microscopy of cofilin/actin rods in forskolin-treated STHdh cells similarly showed cofilin/actin rods had redistributed DAPI staining, suggesting a change in chromatin distribution (**Figure 3B**). Therefore, we conclude nuclear cofilin/actin rods cause a change in nuclear topography.

### Nuclear Organization in a Cell Model of Intranuclear Rod Myopathy

We have recently shown the formation of stable, phalloidinstainable nuclear actin filaments alters RNAPII localization using several cell culture models (Serebryannyy et al., 2016b). This included the V163M mutation in skeletal α-actin that causes intranuclear rod myopathy (Ilkovski et al., 2004; Domazetovska et al., 2007a,b). We found that nuclear actin polymerization reduced the monomeric actin pool and correlated with decreased proliferation and transcription in culture and in vitro (Serebryannyy et al., 2016b). Consistent with previous work (Domazetovska et al., 2007a; Serebryannyy et al., 2016b), we find expression of V163M α-actin GFP in COS7 cells causes formation of actin filaments in the nucleus that grow with time but are restricted by the nuclear periphery (**Video S1**). Using confocal microscopy, we find that the presence of nuclear actin filaments induced by V163M α-actin GFP correlates with changes in RNAPII localization into clusters as well as changes in chromatin organization as marked by DAPI staining (**Figure 4A**). This is consistent with the recent finding that monomeric nuclear actin can inhibit nuclear histone deacetylase activity (Serebryannyy et al., 2016a). Nuclear actin polymerization reverses this inhibition and correlates with more condensed chromatin. Furthermore, immunostaining for lamin A/C showed that cells with nuclear actin filaments exhibited defects in nuclear structure and had misshapen nuclei (**Figure 4B**).

### Organization of RNAPII and Chromatin in Human Intranuclear Rod Myopathy Tissue

The geometrical constraints and 3D environment of cells grown on culture dishes are different than cells within tissues (Pampaloni et al., 2007). Therefore, we sought to examine the nuclear topography of human tissues with nuclear actin filaments. Frozen tissues from patients diagnosed with intranuclear rod myopathy due to V163M or V163L mutations in α-actin and age-matched control subjects were examined for changes in RNAPII and chromatin. Tissue was stained with phalloidin to identify intranuclear rods, RNAPII, and DAPI to label chromatin (**Figure 5A**). Strikingly, we see nuclear actin filaments in intranuclear rod myopathy patient tissues are more prominent than those seen in COS7 cells. Furthermore, the formation of these filaments displaced both RNAPII and chromatin (**Figures 5A,B**), likely influencing transcriptional regulation in these cells. Because nuclear compaction in these tissues was already high, it was difficult to document more subtle effects such as changes in nuclear shape or RNAPII aggregation as seen in our cell culture model (**Figure 4**). Nevertheless, it was clear that nuclei with actin filaments had obvious defects in their topology, suggesting this effect may contribute to the pathology of intranuclear rod myopathy.

### DISCUSSION

Previous studies have shown nuclear actin regulates multiple transcription signaling pathways, chromatin remodelers, and general transcription by all three RNAPs (de Lanerolle and Serebryannyy, 2011; Grosse and Vartiainen, 2013; Kapoor and Shen, 2014). This raises the question of the impact of nuclear actin rods or filaments on transcription. While cellular stressors, such as heatshock, can induce the formation of cofilin/actin rods, α-actin mutations implicated in intranuclear rod myopathy induce the formation of phalloidin-stainable filaments in the nucleus. Not only is the composition of these

structures different (i.e., the presence of cofilin vs. phalloidin staining), but they can also alter the nucleus to different extents (i.e., RNAPII/chromatin displacement vs. aggregation), and the properties of these structures vary with cell and tissue type. Despite this heterogeneity, we find that the formation of actin structures within the nucleus leads to alterations in the organization of the nucleus.

It is easy to envision how obtrusive nuclear actin structures could impede proper genetic regulation; however, it is unclear if these structures serve a functional purpose or are a consequence of the disease. Notably, stressors such as heatshock have been shown to cause changes in the cell's transcription program (Morimoto, 1998) and alter RNAPII clustering (Cisse et al., 2013). Our study suggests nuclear actin may play a role in these responses. It has been hypothesized that the purpose of cofilin/actin rods are protective and may form to regulate ATP levels by limiting actin treadmilling (Whiteman et al., 2009; Munsie and Truant, 2012). If these rods also impair transcription, it may be another method of resource conservation during stress. Interestingly, wildtype α-actin also forms nuclear actin filaments in response to hypoxia and treatment with actin depolymerizing drugs (Domazetovska et al., 2007a). Careful cell manipulations and new age techniques such as laser dissection of the nucleus (Paz et al., 2013) may help answer what role these nuclear actin structures play, if certain nuclear proteins or regions of chromatin have preferential association with these structures, and ultimately, how these nuclear actin structures may affect gene expression.

The different stimuli and cell types used in this study also highlight the phenotypic heterogeneity of nuclear actin polymerization. Why different media conditions can induce the formation of cofilin/actin rods or phalloidin-stainable filaments is curious (**Figures 1**, **2**). Additionally, why neighboring cells exhibit different nuclear actin structures when confronted with the same stress (**Figure 2**) remains an outstanding question. Heterogenic responses exist not only between the types of structures that occur, but also between cell types. It is unclear why neurons seem to be the most sensitive cell type to form cofilin/actin rods both in the nucleus and in the cytoplasm, as seen in Alzheimer's and other neuronal pathologies (Minamide et al., 2000; Bamburg and Wiggan, 2002; Huang et al., 2008; Whiteman et al., 2009; Bamburg et al., 2010; Munsie et al., 2011). These differences could depend on cytoskeleton composition, the balance of actin polymerizing and depolymerizing factors in the nucleus, differences in import and export rates, or perhaps be a consequence of the post-mitotic nature of neurons and myocytes. Notably, modulators of cofilin phosphorylation have been implicated in mediating nuclear actin translocation (Dopie et al., 2015) and may therefore be a point of regulation. Future studies to delineate the roles of actin binding proteins in the nucleus may offer new insights into why different actin structures form in the nucleus in response to different stimuli, what regulates the size and composition of these structures, and what functions nuclear actin structures may serve.

Similarly, it is perplexing that certain cell types such as cardiomyocytes resist nuclear actin polymerization. Intranuclear rods do not appear to form in cardiac tissue (North et al., 1997), despite expressing skeletal α-actin during development (Ruzicka and Schwartz, 1988). Further, of the relatively few mutations documented to occur in cardiac α-actin, none have been shown to cause nuclear actin polymerization. Understanding the pathological consequences of these different nuclear actin structures (e.g., chromatin landscape changes or nuclear actin polymerization) in different tissues may have implications in understanding the role of actin in the nucleus, how cells respond to stress, and why pathological nuclear actin filaments are detrimental.

### MATERIALS AND METHODS

Skeletal muscle tissue from one patient with an ACTA1 V163L mutation, two patients with ACTA1 V163M mutations, and age-matched controls with unrelated disorders were obtained during routine diagnostic procedures (Ilkovski et al., 2004; Domazetovska et al., 2007b). This research was approved by the

Human Research Ethics Committee of the Children's Hospital at Westmead, Australia (10/CHW/45) and performed with patient consent.

### Cell Culture and Treatments

COS7 were cultured in Dulbecco's modified minimum essential medium (DMEM; Corning) with 4.5 g/L glucose and Lglutamine, without sodium pyruvate, supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich), penicillin and streptomycin (Gibco) in a humidified atmosphere containing 5% CO<sup>2</sup> at 37◦C. STHdh and STHdh cells stably expressing cofilin YFP (Ray Truant, McMaster University) were cultured in DMEM with 10% FBS, 1% penicillin, and streptomycin as well as 400 µg/mL G418 (Enzo Life Sciences) at 33◦C. To heatshock, STHdh cells in DMEM or Dulbecco's Phosphate-Buffered Saline with calcium and magnesium (PBS; Corning) were paraffin wrapped and submerged into a waterbath held at 42◦C for 1 h then immediately fixed in 4% paraformaldehyde. For forskolin and latrunculin B (Calbiochem) treatment, cells were plated in 12-well plates and treated with 10 µM forskolin overnight or 1 µg/mL latrunculin B for 1 h at 33◦C in DMEM. Cell transfections were carried out using Polyjet transfection reagent

(SignaGen) according to the manufacturer's protocol. V163M α-actin GFP was a gift from Dr. Kathryn North (University of Sydney).

### Immunostaining and Microscopy

Frozen skeletal muscle tissue was cut in 8 µm slices and fixed with 3% PFA in PBS for 10 min followed by permeabilization in 0.5% triton X for 8 min and three washes in PBS. Slides where then incubated in blocking solution (2% BSA) for 15 min. Primary antibody (1:50 dilution of RNA polymerase II antibody, 4H8, ab5408, Abcam) was incubated in blocking solution for 2 h at room temperature. After primary antibody incubation, the tissue sections were washed three times in PBS for 10 min each. Secondary Alexa555-conjugated anti-mouse antibody (Life Techology) diluted 1:200 in blocking solution, Alexa488 conjugated Phalloidin (1:40, Life Technology) and DAPI were applied for 2 h at room temperature. Finally, sections were washed as above, mounted in Vectashield and imaged using an SP5 confocal microscope. Results were consistent among the three intranuclear rod myopathy patients shown (at least 4 images were obtained for each patient).

For immunocytochemistry, cells were plated on glass coverslips at least 24 h before fixation or transfection. V163M α-actin GFP transfection efficiency in COS7 cells at 48 h was 50.4 ± 15.2%. Of the transfected cells, 52.1 ± 6.7% exhibited nuclear actin filaments, and of the population with nuclear actin filaments 65.3 ± 1.9% exhibited changes in RNAPII and chromatin distribution as scored in two experiments (total of 144 and 72 cells; mean ± standard deviation). Cells were fixed in 4% PFA for 10 m then permeabilized with 0.3% Triton X100 (Sigma-Aldrich) in PBS for 7 m. After permeabilization, cells were washed with PBS and incubated in 2% BSA in PBS for 1 h at room temperature. Cells were stained using a humidity chamber. Primary antibody was added for 1 h at room temperature or overnight at 4◦C. Cells were then washed with PBS and secondary antibody was added for 1 h at room temperature. Cells were washed a final time and mounted using Vectashield containing DAPI. Primary antibodies include RNA polymerase II (4H8, Abcam; 1:200), actin (C4, EMD Millipore; 1:100), cofilin (Sigma-Aldrich; 1:10,000), lamin A/C (Santa-Cruz; 1:100) and phalloidin (Invitrogen; 1:400).

Immunocytochemistry samples were examined using a Zeiss 710 Laser Scanning confocal microscope and single confocal slices were taken through the nucleus or using a Nikon N-SIM microscope (Northwestern University Nikon Imaging Center). The SIM fluorescence intensity plot was generated using Nikon NIS-Elements software where fluorescence signal intensity was plotted along the selected region (relative distance). Live cell imaging was performed using an Olympus VivaView fluorescence microscope. Image files were processed with Zen software or Image J.

To quantify nuclear circularity, cells were left untransfected or transfected for 48 h with V163M α-actin GFP, fixed, and stained with lamin A/C antibody as described above. Confocal images were acquired through the nucleus. To quantify nuclear circularity, Image J was used to threshold the images and circularity was calculated as the ratio of the lengths of Y axis:X axis with a value of 1 representing a perfect circle.

### REFERENCES


### AUTHOR CONTRIBUTIONS

LAS, MY, STC, and PdL: Designed the experiments; LAS, MY, and MP: Performed the experiments. LAS and PdL: Wrote the manuscript.

### ACKNOWLEDGMENTS

We thank Dr. Ray Truant (McMaster University) for generously providing the STHdh Cofilin YFP cells. This work was supported by an award from the UIC Center for Cardiovascular Research (to PdL), the Chicago Biomedical Consortium with support from the Searle Funds at The Chicago Community Trust (to PdL, Co-PI) and a pre-doctoral fellowship from the American Heart Association (AHA13PRE17050060), a Chicago Biomedical Consortium Scholar award, and a UIC Dean's Scholar award (to LAS).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2016.00454

Video S1 | Formation of V163M α-actin GFP filaments. COS7 cells

transfected with V163M α-actin GFP were imaged over 22 h at 30 minute intervals.

at nuclear speckles to enhance export of viral late mRNA. Proc. Natl. Acad. Sci. U.S.A. 108, E136–E144. doi: 10.1073/pnas.1103411108


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Serebryannyy, Yuen, Parilla, Cooper and de Lanerolle. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Protein Structure-Function Relationship at Work: Learning from Myopathy Mutations of the Slow Skeletal Muscle Isoform of Troponin T

Anupom Mondal and J.-P. Jin\*

*Department of Physiology, Wayne State University School of Medicine, Detroit, MI, USA*

Troponin T (TnT) is the sarcomeric thin filament anchoring subunit of the troponin complex in striated muscles. A nonsense mutation in exon 11 of the slow skeletal muscle isoform of TnT (ssTnT) gene (*TNNT1*) was found in the Amish populations in Pennsylvania and Ohio. This single nucleotide substitution causes a truncation of the ssTnT protein at Glu<sup>180</sup> and the loss of the C-terminal tropomyosin (Tm)-binding site 2. As a consequence, it abolishes the myofilament integration of ssTnT and the loss of function causes an autosomal recessive nemaline myopathy (NM). More *TNNT1* mutations have recently been reported in non-Amish ethnic groups with similar recessive NM phenotypes. A nonsense mutation in exon 9 truncates ssTnT at Ser108, deleting Tm-binding site 2 and a part of the middle region Tm-binding site 1. Two splicing site mutations result in truncation of ssTnT at Leu<sup>203</sup> or deletion of the exon 14-encoded C-terminal end segment. Another splicing mutation causes an internal deletion of the 39 amino acids encoded by exon 8, partially damaging Tm-binding site 1. The three splicing mutations of *TNNT1* all preserve the high affinity Tm-binding site 2 but still present recessive NM phenotypes. The molecular mechanisms for these mutations to cause myopathy provide interesting models to study and understand the structure-function relationship of TnT. This focused review summarizes the current knowledge of TnT isoform regulation, structure-function relationship of TnT and how various ssTnT mutations cause recessive NM, in order to promote in depth studies for further understanding the pathogenesis and pathophysiology of *TNNT1* myopathies toward the development of effective treatments.

Keywords: troponin isoform, skeletal muscle, slow twitch fiber, TNNT1 myopathies, recessive mutation

### TROPONIN IN VERTEBRATE STRIATED MUSCLES AND FIBER TYPE-SPECIFIC ISOFORMS

Vertebrates have two types of striated muscles, i.e., skeletal muscle and cardiac muscle. The basic contractile apparatus of vertebrate striated muscle is the sarcomeres that are in tandem repeats in the myofibrils. The sarcomeres consist of overlapping myosin thick filaments and actin thin myofilaments (Tobacman, 1996; Gordon et al., 2000). Muscle contraction is powered by actin-activated myosin ATPase (Cooke, 1986), which is regulated by intracellular Ca2<sup>+</sup> through the

Edited by:

*P. Bryant Chase, Florida State University, USA*

#### Reviewed by:

*Kunihiro Sakuma, Toyohashi University of Technology, Japan Marion Lewis Greaser, University of Wisconsin-Madison, USA Yin-Biao Sun, King's College London, UK*

> \*Correspondence: *J.-P. Jin jjin@med.wayne.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *10 August 2016* Accepted: *20 September 2016* Published: *13 October 2016*

#### Citation:

*Mondal A and Jin J-P (2016) Protein Structure-Function Relationship at Work: Learning from Myopathy Mutations of the Slow Skeletal Muscle Isoform of Troponin T. Front. Physiol. 7:449. doi: 10.3389/fphys.2016.00449*

**128**

troponin complex associated with the thin filament (Gordon et al., 2000). The binding of Ca2<sup>+</sup> to troponin induces a series of allosteric changes in the thin filament, allowing the myosin head to form a strong cross-bridge with F-actin to activate myosin ATPase and initiate contraction (Leavis and Gergely, 1984).

Vertebrate skeletal muscle contains slow twitch and fast twitch types of fibers (Eddinger et al., 1985; Sosnicki et al., 1989). Correspondingly, muscle myosin and troponin have both evolved into slow and fast fiber type-specific isoforms. Slow and fast skeletal muscle fibers express type I and type II myosin, respectively, and these myosin isoenzymes differ in their ATPase activity (Bárány, 1967). Previous studies in multiple laboratories have demonstrated the contribution of four skeletal muscle myosin heavy chain (MHC) isoforms (type I, IIa, IIb, and IIx) to the magnitude and velocity of contraction of different types of muscle fibers (Ruff and Whittlesey, 1991; Johnson et al., 1994).

The troponin complex is at the center of the Ca2+-regulation of muscle contraction (Leavis and Gergely, 1984). Troponin consists of three protein subunits: The Ca2+-binding subunit TnC, the inhibitory subunit TnI, and the tropomyosin-binding subunit TnT (Greaser and Gergely, 1973). To convert the cellular signal of rising cytosolic Ca2<sup>+</sup> originated from sarcolemmal electrical activity to myofilament movements, troponin functions through cooperative interactions among the three subunits and with tropomyosin and the actin thin filament (Tobacman, 1996; Gordon et al., 2000).

Among the three subunits of troponin, TnC belongs to a family of Ca2<sup>+</sup> signaling proteins including calmodulin and myosin light chains (Collins, 1991). A fast isoform of TnC is found in fast twitch skeletal muscle fibers (Gahlmann and Kedes, 1990), whereas slow twitch skeletal muscle and cardiac muscle share another isoform of TnC (Parmacek and Leiden, 1989). In contrast, TnI and TnT are striated muscle-specific proteins, and each has diverged into three homologous isoforms corresponding to the cardiac, slow skeletal and fast skeletal types of muscle fibers (Hastings, 1997; Perry, 1998).

The three TnI and three TnT isoform genes are closely linked in three pairs in the chromosomal genome of vertebrates. The fast TnI and fast TnT genes are linked in one pair (Barton et al., 1997), which is consistent with their linked functions in adult skeletal muscles. However, the cardiac TnI gene is linked to the slow TnT gene (Huang and Jin, 1999) and the slow TnI gene is linked to the cardiac TnT gene (Tiso et al., 1997), which are different from their linked fiber type-specific expressions in adult slow skeletal muscle or heart. While such scrambled linkages of the two pairs of TnI and TnT genes indicate that the TnT and TnI isoform gene expression is regulated by the cellular environment rather than by genomic organization, slow TnI and cardiac TnT express and function together in embryonic heart (Jin, 1996). Embryonic cardiac muscle expresses solely slow skeletal muscle TnI that is replaced by cardiac TnI during late embryonic and early postnatal development (Saggin et al., 1989; Jin, 1996). Therefore, they are originally a functional pair of linked genes, whereas the cardiac TnI and slow TnT genes emerged later as the newest pair (Chong and Jin, 2009).

Mutations in the three TnT isoform genes TNNT1, TNNT2, and TNNT3 encoding slow skeletal muscle TnT, cardiac TnT, and fast skeletal muscle TnT, respectively, have been reported to cause cardiac and skeletal myopathies. During the last two decades, numerous mutations in TNNT2 gene have been found to cause various types of cardiomyopathies (Knollmann and Potter, 2001; Sheng and Jin, 2014). In contrast, rather few mutations in skeletal muscle TnT isoform genes have been reported in skeletal muscle diseases (Wei and Jin, 2016), among which five mutationsin the TNNT1 gene encoding slow TnT cause nemaline myopathies (NM). A nonsense mutation in TNNT1 was first identified in the Old Order Amish (Johnston et al., 2000). The identification and mechanistic studies of the Amish NM (ANM) (Jin et al., 2003; Wang et al., 2005) have raised clinical awareness and the inclusion of testing for TNNT1 mutations in the diagnosis of myopathies. As a result, four more truncation or internal deletion mutations in TNNT1 gene have recently been reported in multiple other ethnic groups around the world to cause myopathies similar to that of ANM (van der Pol et al., 2014; Marra et al., 2015; Abdulhaq et al., 2016). No effective treatment is currently available for TNNT1 NM.

Nemaline myopathies are neuromuscular disorders characterized by muscle weakness and rod-shaped or "nemaline" inclusions in skeletal muscle fibers (Wallgren-Pettersson et al., 2011; Nance et al., 2012). The different TNNT1 NM mutations all have recessively inherited lethal phenotypes, indicating that the various truncations or internal deletion of slow TnT all result in the loss of function. This notion and the critical importance of slow skeletal muscle fibers have been confirmed for the case of ANM mutation (Jin et al., 2003; Wang et al., 2005). To understand the molecular basis for the pathogenesis and pathophysiology of TNNT1 NM will provide insights into the structure-function relationship of TnT and the mechanism of muscle contraction. The present review is thus focusing on the myopathy mutations of slow skeletal muscle isoform of TnT. It is our hope that the in depth discussions will help to stimulate further research leading to the development of targeted treatment of these lethal skeletal muscle diseases.

### STRUCTURE-FUNCTION RELATIONSHIP OF TROPONIN T

TnT is a 30–35-kDa protein (Greaser and Gergely, 1973). Based on available sequence information, the length of vertebrate TnT polypeptide chain ranges from 223 to 305 amino acids. This large size variation of TnT isoforms across species is almost entirely due to the variable length of the N-terminal variable region, from nearly absent in some fish fast skeletal muscle TnT to more than 70 amino acids long in avian and mammalian cardiac TnT (Jin et al., 2008; Wei and Jin, 2011; Jin, 2016). Primary structural data showed that while the N-terminal region is hypervariable in length and amino acid sequences, the amino acid sequences of the middle and C-terminal regions of TnT are highly conserved among the three muscle-type isoforms and across vertebrate species (Jin et al., 2008; Jin, 2016; Wei and Jin, 2016).

Electron microscopic studies showed that the TnT molecule has an extended conformation (Cabral-Lilly et al., 1997; Wendt et al., 1997). High-resolution X-ray crystallographic structures have been obtained for the core region of human cardiac troponin complex (Takeda et al., 2003) and chicken fast skeletal muscle troponin complex (Vinogradova et al., 2005). These solved high-resolution structures of the troponin complex contained entire TnC and most regions of TnI, but only a small C-terminal portion of TnT. The data showed that TnT interfaces with TnI in a coiled-coil structure (i.e., the I-T arm) formed by the segments of L224-V<sup>274</sup> of cardiac TnT and F90-R<sup>136</sup> of cardiac TnI in human cardiac troponin or E199-Q<sup>245</sup> of fast TnT and G <sup>55</sup>-L<sup>102</sup> of fast TnI in chicken fast skeletal muscle troponin. The C-terminal portion of the I-T arm also interacts with TnC. The observation that the N-terminal and the middle region as well as the very C-terminal end of TnT were not resolved in the high resolution crystallographic structures implicates a flexibility of these regions, likely reflecting their allosteric functions in the troponin regulation of muscle contraction.

Complementary to the crystallographic structure, the functional sites and the structure-function relationship of TnT have been extensively investigated in protein binding studies using TnT fragments generated from limited chymotryptic and CNBr digestions. The structural and functional domains of TnT are summarized in **Figure 1**. The classic chymotryptic fragments T1 and T2 of rabbit fast skeletal TnT (Tanokura et al., 1981) were studied for their bindings with other regulatory proteins in muscle thin filament. The ∼100 amino acids Cterminal chymotryptic fragment T2 interacts with TnI and TnC and binds to the middle region of tropomyosin (Heeley et al., 1987; Schachat et al., 1995). The chymotryptic fragment T1 that contains both the N-terminal variable region and the middle conserved region of TnT binds the head-tail junction of tropomyosins in the actin thin filament (Heeley et al., 1987; Schachat et al., 1995). The tropomyosin-binding activity of the T1 fragment resides in the 81 amino acids CNBr fragment CB2 of rabbit fast skeletal muscle TnT, which represents a largely α-helical structure (Pearlstone et al., 1976, 1977) in the middle conserved region of TnT. The CNBr fragment CB3 of rabbit fast skeletal muscle TnT (amino acids 2-50) representing the Nterminal variable region does not bind TnI, TnC, or tropomyosin (Pearlstone and Smillie, 1982; Ohtsuki et al., 1984; Heeley et al., 1987; Perry, 1998).

More recent studies using genetically engineered TnT fragments and mapping with site-specific monoclonal antibody probes showed that the T1 region tropomyosin-binding site 1 of TnT included a 39-amino acid segment at the N-terminal portion of the middle conserved region of TnT (Jin and Chong, 2010). Its downstream boundary was further extended to beyond S<sup>108</sup> , the site of a TNNT1 NM nonsense mutation that causes partial destruction of the tropomyosin binding site 1 (Amarasinghe et al., 2016). The tropomyosin binding site 2 in the T2 fragment was mapped to a 25-amino acid segment at the beginning of the T2 fragment (Jin and Chong, 2010). Amino acid sequences of these segments containing the two tropomyosin binding sites are highly conserved in the three muscle-type specific TnT isoforms and across vertebrate species (Jin et al., 2008).

Although, the N-terminal region of TnT does not bind any known myofilament proteins, its structure is regulated by alternative splicing during late embryonic and early postnatal development of the heart (Jin and Lin, 1988) and skeletal muscles (Wang and Jin, 1997), and in pathologic adaptation (Larsson et al., 2008). These developmental and adaptive regulations of the N-terminal variable region of TnT suggested functional significances. The entire N-terminal variable region of cardiac TnT can also be selectively removed during cardiac adaptation to acute energetic crisis by restrictive proteolysis (Zhang et al., 2006; Feng et al., 2008). Similar modification can also be produced in fast skeletal muscle TnT (Zhang et al., 2006).

In vitro studies have demonstrated the role of the N-terminal region in altering the molecular conformation of TnT in the middle and C-terminal regions and the interactions with TnI, TnC, and tropomyosin (Wang and Jin, 1998; Jin and Root, 2000; Jin et al., 2000; Biesiadecki et al., 2007). The physiological and pathological significances of the regulatory effects of the Nterminal variable region of TnT have also been demonstrated in ex vivo working heart and cardiomyocyte studies using transgenic mice expressing N-terminal modified TnT in the heart (Pan et al., 1991; Chandra et al., 1999; Biesiadecki et al., 2002; Feng et al., 2008; Wei et al., 2010; Wei and Jin, 2015).

Protein binding studies further demonstrated that the Nterminal variable region of TnT remotely modulates the binding affinity of TnT for tropomyosin by reducing the affinities of both site 1 in the middle region (Amarasinghe and Jin, 2015) and site 2 in the C-terminal region (Amarasinghe et al., 2016). This inhibitory regulation has been most clearly demonstrated in the case of the TNNT1 exon 8 deletion NM mutant, where removal of the N-terminal segment very effectively restored tropomyosin binding affinity diminished by the mutation (Amarasinghe et al., 2016). These data further suggest that the conserved structures in the middle and Cterminal regions of TnT confer a baseline state of troponin function that is similar for cardiac, slow and fast skeletal muscle isoforms. The diverged N-terminal structure of the muscle type-specific TnT isoforms provides a regulatory mechanism to fine tune the function of troponin adapted to the contractility requirement in different muscle types and in physiological and pathophysiological adaptations.

The 9 amino acids at the very C-terminal end of TnT are highly conserved among the three muscle type isoforms and across vertebrate species (Jin et al., 2008). The functional significance of this segment has been an interest of experimental research. There is no direct evidence for binding of this C-terminal segment of TnT with any other myofilament proteins. This segment was not resolved in the high-resolution crystal structure of either cardiac or fast skeletal muscle troponin complex (Takeda et al., 2003; Vinogradova et al., 2005), implicating its possible nature as a flexible and allosteric structure. The recent finding of a splicing site mutation in TNNT1 gene, which deletes the exon 14-encoded segment of the C-terminal 14 amino acids and causes NM (van der Pol et al., 2014), supports a critical role of the conserved C-terminal segment of TnT.

Consistently with this notion, mutations of single amino acid substitutions (R278C or R286C) in the C-terminal end segment, partial deletion (W287ter) or error-splice out of exon 17 encoding this segment in cardiac TnT have been found

to cause cardiomyopathy (Thierfelder et al., 1994; Watkins et al., 1995; Richard et al., 2003). Biochemical and biophysical studies have demonstrated that the R278C mutant of cardiac TnT produces a slightly increased Ca2<sup>+</sup> sensitivity with a significant elevation of sub-half-maximal force (Morimoto et al., 1999). Deletion of the C-terminal 14 amino acids of cardiac TnT resulted in lower level activation of myofilament ATPase with reduced effectiveness of Ca2+-troponin to switch the thin filament from the off to the on state (Mukherjea et al., 1999) and also caused detectable ATPase activation in the absence of Ca2<sup>+</sup> showing hindered ability of regulated actin filament in conferring the inactive state (Franklin et al., 2012).

### GENES ENCODING TROPONIN T ISOFORMS

Three homologous genes have evolved in vertebrate species encoding the cardiac (TNNT2), slow skeletal muscle (TNNT1) and fast skeletal muscle (TNNT3) isoforms of TnT (Cooper and Ordahl, 1985; Breitbart and Nadal-Ginard, 1986; Jin et al., 1992; Farza et al., 1998; Huang et al., 1999; Hirao et al., 2004). It has been shown in avian and mammalian species that TNNT1 and TNNT3 genes specifically express in the slow and fast twitch skeletal muscle fibers, respectively. In contrast, TNNT2 gene expresses in embryonic and adult cardiac muscle as well as transiently expresses in embryonic and neonatal skeletal muscles, including both slow and fast fiber dominant muscles (Toyota and Shimada, 1983; Cooper and Ordahl, 1985; Jin, 1996).

The functional diversity of TnT isoforms has physiological significances. An interesting example is that the cardiac muscle of toad (Bufo) expresses exclusively slow skeletal muscle TnT together with cardiac forms of TnI and myosin (Feng et al., 2012). This is a unique case since all vertebrate species studied to date from fish to human including the closely related genus frog (Rana) express only cardiac TnT in the cardiac muscle. Analysis of cardiac function demonstrated that toad hearts generated lower maximum stroke volume but significantly higher resistance to the increase of afterload than that of frog hearts (Feng et al., 2012). This feature is consistent with the unique functional requirement for the toad heart to work under drastically fluctuation of blood volumes and regulation via vasoconstrictions. This finding demonstrates a fitness selection value of the evolutionary adaptation of utilizing slow skeletal muscle TnT in toad cardiac muscle. The specific structure(s) of slow TnT in altering the contractility of toad cardiac muscle is worth investigating in order to better understand the critical role of slow TnT in skeletal muscle function as well as the development of a way targeting cardiac TnT to treat heart failure.

The evolutionary linage of the three TnT isoform genes have been thoroughly investigated by sequence analysis and protein epitope studies (Chong and Jin, 2009). Using monoclonal antibodies as site-specific epitope probes, a method was developed to detect evolutionarily suppressed molecular conformation by removing the suppressor structures, such as the evolutionarily added N-terminal variable region. The results demonstrated three-dimensional structure evidence for the evolutionary relationships between TnI and TnT and among their muscle type-specific isoforms (Chong and Jin, 2009). The data further demonstrate a novel mode of protein evolution by allosterically suppressing the ancestral molecular conformation with the evolutionary addition of a modulatory structure, in which the present-day form of TnT isoforms with diverged primary and folded structures have the potential of restoring ancestral conformations after removing the evolutionarily added repressor structure (Chong and Jin, 2009).

The adult heart and skeletal muscles express the three TnT isoform in a muscle fiber type-specific manner (Jin, 2016). Knockout of the TNNT2 gene encoding cardiac TnT resulted in embryonic lethality (Nishii et al., 2008). Consistent with the differentiated role of slow muscle fibers critical to the mobility of animals (Rome et al., 1988), the loss of ssTnT results in severe NMs (Johnston et al., 2000; Jin et al., 2003; van der Pol et al., 2014; Marra et al., 2015; Abdulhaq et al., 2016). Therefore, the three muscle type TnT isoforms play non-redundantly roles in the functions of the three types of striated muscle.

In addition to the sequence and protein conformation lineage data, the undifferentiated utilization of the same TnC isoform in cardiac and slow skeletal muscles also supports the hypothesis that the emergence of the cardiac and slow TnI-TnT gene pairs was a relatively recent event of evolutionary divergence (Chong and Jin, 2009). A further support to this notion that among the three TnI-TnT gene pairs, cardiac TnI-slow TnT genes form the newest pair is the presence of a unique N-terminal extension in cardiac TnI, an additional structure that is absent in fast and slow skeletal muscle TnI isoforms (Parmacek and Solaro, 2004). The latest emergence of the slow TnT gene may be a landmark of vertebrate evolution and its functional significance requires more investigation.

### ALTERNATIVE SPLICING

Expression of the three TnT isoform genes is regulated at the transcriptional level as well as via alternative RNA splicing (Jin et al., 2008; Wei and Jin, 2011, 2016). The splicing variants add to the diversity of TnT structure for fine tuning of muscle contractility during development and in adaptation to physiological stress and pathological conditions.

### Cardiac TnT

The mammalian cardiac TnT gene (TNNT2) contains 14 constitutively expressed exons and three alternatively spliced exons (Jin et al., 1992, 1996; Farza et al., 1998). Exon 5 of cardiac TnT gene, which encodes 9 or 10 amino acids in the N-terminal variable region, is included in embryonic but not adult cardiac TnT (Jin and Lin, 1989). Exon 4 of cardiac TnT gene is alternatively spliced independent of developmental state (Jin et al., 1996). The avian cardiac TnT gene contains 16 constitutively spliced exons and only one alternative exon (the embryonic exon 5) (Cooper and Ordahl, 1985). Correspondingly, four mammalian and two avian cardiac TnT N-terminal alternative splicing variants have been found in normal cardiac muscle.

The inclusion or exclusion of exon 5 generates an embryonic to adult cTnT isoform switching during development (Cooper and Ordahl, 1985; Jin and Lin, 1988; Jin et al., 1996). When TNNT2 gene is transiently expressed in embryonic and neonatal skeletal muscles, the alternative splicing pattern is synchronized to that in the heart (Jin, 1996). The timing of the switching of TNNT2 alternative splicing varies in different species, indicating regulation by a systemic clock, rather than adaptation to changes in contractile function (Jin, 1996).

Splice out of exon 4 that encodes 4-5 amino acids in the N-terminal variable region of cardiac TnT increases in failing human hearts (Anderson et al., 1995; Mesnard-Rouiller et al., 1997), diabetic (Akella et al., 1995) and hypertrophic (McConnell et al., 1998) rat hearts. Aberrant splice out of N-terminal coding exons of cardiac TnT (exon 7 in dogs equivalent to exon 8 in turkey) is found in dilated cardiomyopathy (Biesiadecki et al., 2002; Biesiadecki and Jin, 2002).

There is another alternatively spliced exon (exon 13) encoding a short segment of 2 or 3 amino acids between the T1 and T2 regions of mammalian cardiac TnT (Jin et al., 1992, 1996). The functional significance of this variable region is unknown.

### Fast Skeletal Muscle TnT

Mammalian fast skeletal muscle TnT gene contains 19 exons, of which exons 4, 5, 6, 7, 8, and a fetal exon encoding segments in the N-terminal variable region are alternatively spliced (Breitbart and Nadal-Ginard, 1986; Briggs and Schachat, 1993; Wang and Jin, 1997). Additional alternative N-terminal coding exons are present in avian TNNT3 gene (Smillie et al., 1988; Ogut and Jin, 1998; Miyazaki et al., 1999; Jin and Samanez, 2001). Seven P exons are located between exon 5 and 6 in the N-terminal variable region of avian fsTnT encode a unique Tx segment (Smillie et al., 1988; Jin and Smillie, 1994; Miyazaki et al., 1999; Jin and Samanez, 2001). A w exon and a y exon are found between exons 4-5 and 7-8, respectively (Schachat et al., 1995).

The alternative splicing of two mutually exclusive C-terminal exons (16 and 17) each encoding a segment of 14 amino acids also occurs in TNNT3 expression (Wang and Jin, 1997). This alternatively spliced segment of fast TnT is in the interface with TnI and TnC (Wei and Jin, 2016). Incorporation of exon 17-encoded segment weakened binding of TnT to TnC and tropomyosin (Wu et al., 1995). This region also shows diversity between mammalian and avian cardiac TnT, where the avian cardiac TnT gene contains an additional exon encoding two amino acids (Cooper and Ordahl, 1985).

Like that of cardiac TnT, expression of TNNT3 gene undergoes a high to low molecular weight, acidic to basic isoelectric point splice form switch during development due to alternative inclusions of N-terminal exons (Jin et al., 2008; Wei et al., 2014). The alternative splicing of TNNT3 pre-mRNA is regulated independently of skeletal muscle fiber types as deficiency of slow skeletal TnT did not affect the developmental switch of fast TnT splice forms (Wei et al., 2014).

### Slow Skeletal Muscle TnT

The slow skeletal muscle TnT gene TNNT1 has a simpler structure and fewer alternative-splicing variants than that of the fast and cardiac TnT genes. There are only 14 exons in the TNNT1 gene with one alternatively spliced. With exon-intron organizations same as that of the mammalian slow TnT gene (Huang and Jin, 1999), chicken slow TnT gene is significantly smaller (∼3-kb versus ∼11-kb) due to shorter introns (Hirao et al., 2004). Alternative splicing of exon 5 in the N-terminal region generates two variants of slow TnT (Gahlmann et al., 1987; Jin et al., 1998; Huang and Jin, 1999). Splicing at alternative acceptor sites in intron 5 of mouse slow TnT gene produces a single amino acid variation in the exon 6-encoded segment (Huang and Jin, 1999). The same pattern was found for the splicing of intron 4-exon 5 of chicken slow TnT gene (Hirao et al., 2004). Abnormal inclusion of 48 bases of the 3′ -region of intron 11 was reported in a cloned human slow TnT cDNA (Gahlmann et al., 1987). However, no corresponding high molecular weight slow TnT protein was detectable (Jin et al., 1998), implicating a circumstantial splicing error.

Alternative splicing of slow TnT shows no apparent developmental regulation but may play a role in modulating muscle contractility in physiological and pathophysiological adaptations. While the high molecular weight splice form including the exon 5-encoded segment is the major slow TnT expressed in normal muscles, the low molecular weight slow TnT became predominant in overused prior polio muscle and significantly up-regulated in type 1 (demyelination), but not type 2, Charcot-Marie-Tooth disease (Larsson et al., 2008). Interestingly, the expression of slow skeletal muscle TnT in the toads hearts as an evolutionarily selected cardiac adaptation to the drastic changes in blood volume is solely the low molecular weight splice form (Feng et al., 2012). These observations indicate differentiated functionalities of the alternative spliced variants of slow TnT.

Aberrant splicing of slow TnT causes NM (van der Pol et al., 2014; Abdulhaq et al., 2016). Error splice out of the exon 8 encoded segment in the middle region of slow TnT drastically alters the molecular conformation and function in the C-terminal region and diminishes the binding affinity for tropomyosin (Amarasinghe et al., 2016).

### NEMALINE MYOPATHY MUTATIONS IN TNNT1 GENE

Multiple mutations in TNNT1 gene, located at 19q13.42 in the human genome, have been identified to cause autosomal recessively inherited nemaline myopathies (**Table 1** and **Figure 2**). The first one was identified in the Old Order Amish in Lancaster County, Pennsylvania. Known as the "Chicken Breast Disease" in the Amish community, this "Amish Nemaline Myopathy" (ANM) is a severe myopathy disease with infantile lethality (Johnston et al., 2000). ANM infants exhibit tremors and muscle weakness, followed by the development of contractures and progressive chest deformation due to weakness of the respiratory muscles. Death from respiratory insufficiency usually occurs in the second year. ANM has an incidence of 1 in ∼500 births in the Amish communities in Pennsylvania and Ohio



*NM, Nemaline Myopathy.*

(Johnston et al., 2000). No effective treatment is currently available.

The genetic causes of ANM is a nonsense mutation in exon 11 of TNNT1 gene converting the codon Glu<sup>180</sup> to a premature stop codon to truncate the slow TnT polypeptide chain by deleting the C-terminal 83 amino acids (Johnston et al., 2000). The deletion of the C-terminal segment of slow skeletal muscle TnT by the E180ter mutation causes a loss of the binding sites for TnI, TnC and the tropomyosin-binding site 2 in the T2 region (Jin et al., 2003; **Figure 2**). Although, the middle region tropomyosin binding site 1 remains intact, the truncated slow TnT is not able to form troponin complex or incorporate into the myofilament (Wang et al., 2005). This phenotype demonstrates the necessity of the two-site anchoring of troponin on the thin filament in the assembly and function of the thin filament regulatory system (Jin and Chong, 2010).

The truncated ANM slow TnT fragment is not detectable in the patient muscle (Jin et al., 2003), indicating a rapid degradation of non-myofilament associated TnT protein and fragments in muscle cells (Wang et al., 2005). This effective removal of mutant or damaged TnT from the myocytes when they are not integrated in the myofibrils is an important protective mechanism to avoid cytotoxic effect (Jeong et al., 2009). This mechanism also explains how the various TNNT1 mutations reported to date all present as recessively inherited diseases (Johnston et al., 2000; van der Pol et al., 2014; Marra et al., 2015; Abdulhaq et al., 2016). On the other hand, this mechanism converts a potentially dominant negative mutation into a recessive mutation, which calls for more extensive genetic screening of TnT mutations in the clinical diagnosis of recessive myopathies.

Based on the structural and functional defect of ANM slow TnT mutant, the molecular basis of the pathogenesis and pathophysiology of ANM is the complete loss of slow TnT protein in slow muscle fibers (Jin et al., 2003; Wang et al., 2005). The loss of slow TnT causes atrophy and degeneration of slow twitch muscle fibers that are essential for many vital physiological activities (Jin et al., 2003). In a transgenic mouse models of ANM, slow TnT deficiency caused significant decreases in the contents of type I slow fibers in diaphragm and soleus muscles (Feng et al., 2009; Wei et al., 2014). Although, the slow TnT deficient slow fibers had active regeneration and hypertrophic growth of type II fast fibers, the muscles showed significantly decreased fatigue resistance (Feng et al., 2009; Wei et al., 2014), consistent with the pathophysiological phenotype of posture muscle weakness

or causes an internal deletion are illustrated on a linear map of slow TnT protein with the segment encoded by exons 2-14 outlined. The filled box indicates alternatively spliced exon 5. All of the five *TNNT1* mutations cause recessively inherited nemaline myopathies. The known binding sites for TnI, TnC and tropomyosin are outlined.

and respiratory muscle failure in ANM patients (Johnston et al., 2000).

The identification of ANM and subsequent mechanistic studies have promoted clinical awareness of TNNT1 myopathy and its testing in the clinical diagnosis of myopathies. As results, several recent reports have identified four more TNNT1 mutations in non-Amish ethnic groups, which cause nemaline myopathies clinically similar to ANM (**Table 1**). TNNT1 myopathies are, therefore, no longer considered as an isolated disease of the Amish, but are of increasing medical importance. The increasing application of genetic screening is anticipated to identify more TNNT1 myopathy mutations.

A nonsense mutation in the exon 9 of TNNT1 gene at codon Ser<sup>108</sup> was found in a Hispanic patient in New York City with clinical and histological features were very much like that of ANM, including severe respiratory muscle weakness, type I fiber atrophy and compensatory hypertrophy of type II fibers (Marra et al., 2015). The TNNT1 S108ter mutation is predicted to result in a truncated slow TnT protein missing the C-terminal 155 amino acids. Therefore, the similar recessive phenotypes of the ANM E180ter and S108ter mutations are based on their loss of the T2 region TnI and TnC binding sites as well as the tropomyosin-binding site 2 (**Figures 1**, **2**; Jin and Chong, 2010). Recent biochemical characterization further demonstrated that the Ser<sup>108</sup> truncation of slow TnT also partially damages the middle region tropomyosin binding site 1 (Amarasinghe et al., 2016), which makes it more unlikely to incorporate into the myofibrils.

A genomic DNA rearrangement in TNNT1 gene (c.574\_577 delins TAGTGCTGT) was reported in 9 Palestinian patients from 7 unrelated families with recessively inherited NM (Abdulhaq et al., 2016). This mutation leads to aberrant splicing to truncate the slow TnT polypeptide at Leu<sup>203</sup> (**Figures 1**, **2**). The patients presented with recessive NM phenotypes very similar to that of ANM (Abdulhaq et al., 2016). Biochemical studies demonstrated that although slow TnT truncated at Leu<sup>203</sup> retains both tropomyosin-binding sites 1 and 2, the inability of forming troponin complex due to the loss of TnI and TnC binding sites decreased the binding affinity for tropomyosin, especially at high calcium (Amarasinghe et al., 2016). This loss of function and the loss of the highly conserved C-terminal segment may be responsible for the recessive myopathy phenotype of the Leu<sup>203</sup> truncation.

Another case report of a Dutch patient of inherited nemaline myopathy described two new TNNT1 NM mutations (van der Pol et al., 2014). The patient also presented with phenotypes of severe slow skeletal muscle atrophy and weakness similar to that of ANM. Molecular diagnosis identified that the patient is a compound heterozygote of a mutation in intron 8 of the TNNT1 gene that causes aberrant exclusion of exon 8-encoded sequence and another mutation that causes exclusion of the exon 14-encoded segment (**Figure 2**) (van der Pol et al., 2014). The deletion of exon 8 segment partially destroys the T1 region tropomyosin-binding site 1 (**Figure 1**) but preserves the highaffinity binding site 2 (Jin and Chong, 2010), whereas deletion of the exon 14-encoded C-terminal end segment would not directly affect either of the tropomyosin-binding sites, nor the binding sites for TnI and TnC.

A recent study found that slow TnT with the internal deletion of the exon 8-encoded segment has drastically decreased binding affinity for tropomyosin, which is much lower than that of tropomyosin binding site 2 alone. Deletion of the N-terminal variable region partially restored the binding affinity of exon 8 deleted slow TnT (Amarasinghe et al., 2016). These observations indicate that deletion of the exon 8-encoded segment not only directly damages the middle region tropomyosin-binding site 1 but also augments the effect of the N-terminal region on reducing the binding affinity of site 2. Therefore, the N-terminal variable region provides a potential target for the treatment of the myopathy caused by slow TnT exon 8 deletion.

The molecular mechanism for slow TnT exon 14 truncation to cause recessive myopathy remains to be investigated. As described in Section Structure-Function Relationship of Troponin T, there are evidence that the C-terminal end segment of TnT may contribute to the inhibitory regulation of myofilament ATPase (Morimoto et al., 1999; Mukherjea et al., 1999; Franklin et al., 2012). More investigation along this line would help to understand the pathogenic mechanism of this aberrant splicing mutation of TNNT1 gene.

### PERSPECTIVES: WHAT HAVE BEEN LEARNED FROM THE PATHOGENIC MUTATIONS OF SLOW SKELETAL MUSCLE TNT

Through isoform gene regulation, alternative RNA splicing and posttranslational modifications, structural and functional variations of TnT modulate striated muscle contractility. The fact that the loss of only the slow isoform of TnT causes lethal myopathy regardless of the mixed composition and expression of slow and fast TnTs in human skeletal muscles demonstrates the functional divergence and necessity of the fiber type-specific TnT isoforms. The critical role of slow skeletal muscle fibers and the structural-function relationship of slow muscle TnT demonstrated by the lethal myopathic mutations and mechanistic studies summarized in this review provide many novel insights into the structurefunction relationship of TnT and troponin regulation of striated muscle contraction.

An intriguing feature of the 5 myopathic TNNT1 mutations reported to date is that they all presented as recessively inherited diseases (Johnston et al., 2000; van der Pol et al., 2014; Marra et al., 2015; Abdulhaq et al., 2016). To fully understand the molecular basis of these different structural defects of slow TnT for causing complete loss of function from inability of incorporating into the myofilament will help to further understand the structurefunction relationship of TnT and troponin regulation of muscle contraction.

Genetically modified mice with a knockdown of the expression of slow TnT exhibited decreased muscle resistance to fatigue (Feng et al., 2009). However, carriers of ANM (and other TNNT1 NM) did not report notable clinical symptom (Johnston et al., 1997). Therefore, it would be worth further investigating whether the haploid insufficiency in carriers of these recessive TNNT1 myopathies may cause conditional slow TnT deficiency with symptoms such as experiencing conditional fatigue intolerance and other slow muscle-related dysfunctions.

The potential cytotoxicity of non-myofilament associated slow TnT fragments (Jeong et al., 2009) in the muscle cells of both homozygote patients and heterozygote carriers is also worth investigating. Exhaustive work load or muscle wasting conditions may produce peaks of myofilament decay to add to the existing pool of the mutant slow TnT fragment, which may overwhelm the capacity of protein turnover system in muscle cells to cause apoptosis and inflammatory damage, although such dominant negative phenotype might only be a transient state in the muscle of carriers.

The transient expression of cardiac TnT and embryonic splice forms of fast skeletal muscle TnT in embryonic and neonatal skeletal muscle may explain the postnatal onset of ANM (Jin et al., 2003). This observation suggests a potential compensation of the fetal forms of TnT for the loss of slow TnT in ANM neonatal skeletal muscles, which may be explored as a therapeutic target. This approach would require an activation of cardiac TnT expression or embryonic alternative splicing pathways of fast TnT in adult slow skeletal muscle fibers.

Soleus muscle of slow TnT knockout mouse maintains a slow fiber cellular environment and exhibits signs of active regeneration (Wei et al., 2014). This is a plausible observation suggesting that a restoration of slow TnT in slow muscle fibers of TNNT1 myopathy patients should be able to readily rescue muscle growth and functions. The maintained slow fiber cellular environment and active regeneration also indicate that translational read-through of the nonsense stop codon in the muscles of ANM and S108ter patients may effectively restore muscle function and growth.

Since TNNT1 myopathies is no longer considered as a isolated disease of the Amish, and the power of genetic testing is anticipated to identify more myopathic mutations of the gene, researchers are urged to add joint effort in TnT gene expression and structural-function relationship studies, including the utilization of genetically modified mouse models of the human diseases, toward the development of effective targeted treatment of these lethal muscle diseases.

In conclusion, TNNT1 myopathies demonstrate an excellent example for what we can learn from pathogenic mutations of a myofilament protein as well as how knowledges learned from protein structure-function relationship research can help us to understand the pathogenesis and pathophysiology of genetic diseases. Therefore, we hope this focused review will benefit readers with the vision beyond TNNT1 myopathy studies.

## AUTHOR CONTRIBUTIONS

AM: Drafting and revising the text, making figures, approval submission; JJ: Deciding the topic and contents, drafting and revising the text, making figures, approval submission.

### ACKNOWLEDGMENTS

Our troponin studies have been supported by National Institutes of Health grants AR048816, HL078773, HL098945, and HL127691 to JJ.

### REFERENCES


of the human cardiac troponin T gene. J. Mol. Cell. Cardiol. 30, 1247–1253. doi: 10.1006/jmcc.1998.0698


cardiomyopathy troponin T mutants. Biochemistry 38, 13296–13301. doi: 10.1021/bi9906120


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Mondal and Jin. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Heart Failure Induced by Perinatal Ablation of Cardiac Myosin Light Chain Kinase

Yasmin F. K. Islam<sup>1</sup> , Ryan Joseph<sup>1</sup> , Rajib R. Chowdhury <sup>1</sup> , Robert H. Anderson<sup>2</sup> and Hideko Kasahara<sup>1</sup> \*

<sup>1</sup> Department of Physiology and Functional Genomics, University of Florida, Gainesville, FL, USA, <sup>2</sup> Institute of Genetic Medicine, Newcastle University, Newcastle, UK

Background: Germline knockout mice are invaluable in understanding the function of the targeted genes. Sometimes, however, unexpected phenotypes are encountered, due in part to the activation of compensatory mechanisms. Germline ablation of cardiac myosin light chain kinase (cMLCK) causes mild cardiac dysfunction with cardiomyocyte hypertrophy, whereas ablation in adult hearts results in acute heart failure with cardiomyocyte atrophy. We hypothesized that compensation after ablation of cMLCK is dependent on developmental staging and perinatal-onset of cMLCK ablation will result in more evident heart failure than germline ablation, but less profound when

#### Edited by:

P. Bryant Chase, Florida State University, USA

#### Reviewed by:

Martina Krüger, University of Düsseldorf, Germany Margaret Westfall, University of Michigan, USA Paul M. L. Janssen, Ohio State University, USA

> \*Correspondence: Hideko Kasahara hkasahar@ufl.edu

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 19 July 2016 Accepted: 05 October 2016 Published: 26 October 2016

#### Citation:

Islam YFK, Joseph R, Chowdhury RR, Anderson RH and Kasahara H (2016) Heart Failure Induced by Perinatal Ablation of Cardiac Myosin Light Chain Kinase. Front. Physiol. 7:480. doi: 10.3389/fphys.2016.00480 compared to adult-onset ablation.

Methods and Results: The floxed-Mylk3 gene was ablated at the beginning of the perinatal stage using a single intra-peritoneal tamoxifen injection of 50 mg/kg into pregnant mice on the 19th day of gestation, this being the final day of gestation. The level of cMLCK protein level could no longer be detected 3 days after the injection, with these mice hereafter denoted as the perinatal Mylk3-KO. At postnatal day 19, shortly before weaning age, these mice showed reduced cardiac contractility with a fractional shortening 22.8 ± 1.0% (n = 7) as opposed to 31.4 ± 1.0% (n = 11) in controls. The ratio of the heart weight relative to body weight was significantly increased at 6.68 ± 0.28 mg/g (n = 12) relative to the two control groups, 5.90 ± 0.16 (flox/flox, n = 11) and 5.81 ± 0.33 (wild/wild/Cre, n = 5), accompanied by reduced body weight. Furthermore, their cardiomyocytes were elongated without thickening, with a long-axis of 101.8 ± 2.4µm (n = 320) as opposed to 87.1 ± 1.6µm (n = 360) in the controls.

Conclusion: Perinatal ablation of cMLCK produces an increase of heart weight/body weight ratio, a reduction of contractility, and an increase in the expression of fetal genes. The perinatal Mylk3-KO cardiomyocytes were elongated in the absence of thickening, differing from the compensatory hypertrophy shown in the germline knockout, and the cardomyocyte thinning shown in adult-inducible knockout.

Keywords: heart failure, perinatal, knockout, kinase, Myosin light chain kinase

### INTRODUCTION

The incidence of congestive heart failure in childhood ranges from 2.95 to 23.2 in each 1000 patients, of which just over 80% have congenital heart disease, followed by the population with cardiomyopathies (7%) and arrhythmias (2%) (Schmaltz, 2015). The signs and symptoms of heart failure include growth retardation, respiratory distress, and exercise intolerance. The failure in the volume-overloaded heart occurs in the setting of left-to-right shunts, such as ventricular septal defects, persistent patency of the arterial duct, or atrioventricular septal defects. On the other hand, aortic stenosis is the most common cause of pressure-overload heart failure. Complex malformations can induce both volume- and pressure-overloaded heart failure (Hsu and Pearson, 2009a,b; Schmaltz, 2015). As in adult patients, the brain natriuretic peptide (BNP)/N-terminal prohormone BNP is used as a biomarker heart failure in childhood, despite the knowledge that there is an age-dependent variation in normal individuals from ∼3000 pg/ml in 0–2 days of age, which drops to ∼100 pg/ml between 1 month and 1-year of age (Nir et al., 2009; Schmaltz, 2015).

In clinical settings, a reduction in phosphorylation of myosin light chain 2 (MLC2) has been demonstrated in adult patients with heart failure (Sanbe et al., 1999; Davis et al., 2001; Moss and Fitzsimons, 2006; Stelzer et al., 2006; Scruggs and Solaro, 2011; Sheikh et al., 2012, 2014). It is not yet known, to the best of our knowledge, whether phosphorylation of MLC2, or its responsive kinase, cardiac myosin light chain kinase (cMLCK), is reduced in children with heart failure. When, in genetically modified mice, cMLCK encoded by Mylk3 gene, was ablated either from the germline or inducibly in adulthood, both populations of cMLCK-deficient mice exhibited heart failure (Warren et al., 2012; Massengill et al., 2016). Knockout in the germline, however, leads to compensatory cardiac hypertrophy, while induction of the knockout in adult mice results in cardiomyocyte atrophy. These studies showed the potential that embryonic hearts have a higher ability to adapt in the absence of cMLCK compared to adult hearts. In this study, we queried if cMLCK ablation during the perinatal stage leads to an intermediate phenotype between those encountered in the setting of embryonic- or adultonset ablation. We found that perinatal knockout, beginning at embryonic day 19, did indeed produce mice exhibiting moderate heart failure accompanied by elongated cardiomyocytes, but in the absence of any hypertrophy.

### METHODS

### Mouse Models

A conditional null allele of Mylk3 was generated by introducing loxP sites spanning exon 5, which was done through homologous recombination in ES cells as described previously (Warren et al., 2012). Floxed-Mylk3 homozygous mice (Mylk3flox/flox) (Massengill et al., 2016) were bred with transgenic mice carrying the Cre-ERTM gene under CMV promoter, Tg(CAGGS-Cre-ERTM) (Hayashi and McMahon, 2002). Subsequent matings between offspring generated Mylk3flox/flox/Tg-CAGGS-Cre-ERTM, Mylk3wild/wild/Tg-CAGGS-Cre-ERTM, and Mylk3flox/flox on a mixed genetic background, mainly C57BL/6J. For perinatal deletion of the floxed-Mylk3 genes, a single injection of tamoxifen at 50 mg/kg body weight, was administered intra-peritoneally to pregnant mice on the 19th day of gestation, which is the final day of gestation in the mouse. All animal experiments were performed using protocols reviewed and approved by the University of Florida Institutional Animal Care and Use Committee.

### Echocardiogram

Mice were anesthetized with 1.5–2% isoflurane for M-mode ultrasound imaging of the left ventricles using Vevo700 as described previously (Briggs et al., 2008; Takeda et al., 2009; Warren et al., 2012; Massengill et al., 2016).

### Measurements of Cardiomyocyte Size

Cardiomyocytes were isolated by retrograde perfusion of collagenase as described previously (Takeda et al., 2009; Warren et al., 2012; Massengill et al., 2016). Briefly, the hearts were perfused by Langendorff system with the collagenase digestion buffer (type 2 collagenase 2 mg/ml, in MEM with 10 mM taurine, 3.8 mM creatine, 30 mM 2,3-Butanedione 2-Monoxime, and 20 units of insulin, pH 7.3) at 37◦C for 30 min. Then the hearts were minced into 8–10 pieces, and further digested with the second collagenase digestion (type 2 collagenase 2 mg/ml, in MEM with 10 mM taurine, 3.8 mM creatine, 30 mM 2,3-Butanedione 2- Monoxime, 20 units of insulin, 0.5% bovine serum albumin, and 0.3 mM CaCl2, pH 7.3) at 37◦C for 10 min. The cells were spun down at 46 × g for 2 min, resuspended in the second digestion buffer without collagenase and loaded onto a 36% Percoll gradient (GE Healthcare). After centrifugation at 290 × g for 10 min, cardiomyocytes were pelleted and non-myocytes were floated on top. The cardiomyocytes were washed twice with the second digestion buffer without collagenase, suspended and plated on the laminin-coated coverglass for 15 min. The cells were immediately fixed with 4% PFA and analyzed for measurement of cell size as described previously (Takeda et al., 2009; Warren et al., 2012; Massengill et al., 2016). All the reagents except for Percoll were obtained from Sigma.

### Western Blotting and Histological Analyses

Western blot analyses and immunostaining were performed with GAPDH (MAB374, EMD Millipore), phospho-MLC2v (gift from Dr. N. Epstein, NIH) (Davis et al., 2001), MLC2 (F109/3E1. ALX-BC-1150-S-L005, Enzo Life Science), and cMLCK (Chan et al., 2008) antibodies. The rabbit affinity-purified cMLCK antibody production has been described previously (Chan et al., 2008). Briefly, the antigen was purified from GST-cMLCK(aa 28-463) following cleavage of GST-tag using thrombin and SDS-PAGE separation followed by electro-elution. Rabbits were injected with the antigen and adjuvants 4 times, and the serum from these animals were affinity-purified using GST-cMLCK(aa 28- 463) covalently-coupled beads. 1:4000-diluted antibody was used for Western blotting.

Histological sections (5µm) were analyzed by hemotoxylin and eosin staining, as well as for fibrosis after Picrosirius red staining, which was performed by heating the tissue sections at 60◦C for 45 min before deparaffinization and then immersing them in a combination of 0.1% direct red 80 and 0.1% fast green FCF in 1.2% picric acid for 1 h. TUNEL staining was performed using the In Situ Cell Death Detection Kit (Roche 11684795910).

### Real-Time RT-PCR

Real-time RT-PCR was performed using inventoried Taqman Gene Expression Assays (Thermo Fisher Scientific): Atrial natriuretic factor (ANF) Mm01255748, brain natriuretic peptide

(BNP) Mm00435304, and β-myosin heavy chain (βMHC) Mm00600555. All were normalized to β-actin expression (No. 4352933E). Duplicated experiments were averaged.

### Statistical Analyses

Data presented are expressed as mean values ± S.E.M. Results were compared using T-test or ANOVA with Fisher's post-hoctest (SPSS version 24). p < 0.05 was considered significant.

### RESULTS

### There is an Increase of Heart Weight Relative to the Body Weight, and Overall Weight Loss in Perinatal Mylk3-KO Mice

To examine the effects of cMLCK in perinatal cardiac function, we generated tamoxifen-inducible Mylk3 knockout mice that carry homozygous floxed-Mylk3 alleles and heterozygous Cre-ERTM transgene under the control of the CMV enhancer and the chicken β-globulin promoter (CAGGS-Cre-ERTM) (**Figure 1A**). Deletion of floxed-exon 5 resulted in elimination of the first coding exon of the catalytic domain, and a frame-shift of the subsequent downstream exons. In addition, deletion of exon 5 resulted in reduction of cMLCK mRNA (Warren et al., 2012), attributable to a non-sense-mediated mRNA decay (Conti and Izaurralde, 2005), with targeted cMLCK mRNA containing a premature termination codon. Hereafter, conditional deletion of Mylk3 gene in Mylk3flox/flox/CAGGS-Cre-ERTM mice with tamoxifen injection at embryonic day 19, will be described as perinatal Mylk3-KO. Of note, this model produces a general knockout of the Mylk3 gene, although the expression of cMLCK is known to be restricted to the myocardium (Chan et al., 2008).

A single tamoxifen injection into pregnant mice on gestational day 19, at 50 mg/kg, given intraperitoneally, reduced the expression of cMLCK protein and MLC2 phosphorylation by postnatal day 3 (P3), such that it could no longer be detected (**Figures 1B,C**). During the initial experiment, we noticed that perinatal Mylk3-KO mice appeared unhealthy with growth retardation on P19, 2 days before weaning, and there was concern that these mice would die after weaning. Indeed, the ratio of heart weight/body weight (HW/BW) was increased in the perinatal Mylk3-KO mice compared to the two control groups, represented by the Mylk3flox/flox and Mylk3wild/wild/Tg-CAGGS-Cre-ERTM mice, mainly due to a decrease in body weight (**Figure 1D**). The remaining groups of mice, therefore, were analyzed using the same experimental timeline. In **Figure 1E**, we show representative hearts dissected from the control and perinatal Mylk3-KO mice. Expression of fetal genes, often observed in failing hearts, was increased in perinatal Mylk3-KO mice, including atrial natriuretic factor (ANF), BNP, and β-myosin

Frontiers in Physiology | www.frontiersin.org

wall.

heavy chain (βMHC) (**Figure 1F**). These results indicate that the perinatal Mylk3-KO induced heart failure, as shown by the increase of HW/BW ratio, increased fetal gene expression, and the loss of body weight as compared to the controls.

### Contractile Dysfunction and Cardiomyocyte Elongation in the Perinatal Mylk3-KO Mice

Echocardiography performed on the 19th postnatal day demonstrated a significant reduction in cardiac contractility in perinatal Mylk3-KO mice, as measured by %fractional shortening (%FS), and increased systolic dimensions of the left ventricular cavity (**Figures 2A,B**). Cardiomyocytes isolated from these mice were significantly longer than their controls, albeit without any change in the diameter of the short-axis (**Figures 2C,D**). Histological analyses, however, failed to show any interstitial fibrosis, nor an increase in TUNEL-positive cells (**Figures 3A,B**).

### Comparison of the Germline, Perinatal, and Adult-Inducible Knockout Models

Together with our previous studies (Warren et al., 2012; Massengill et al., 2016), we have now analyzed three different Mylk3 knockout mouse models under similar experimental conditions, with the knockout induced in the germline, in the perinatal period, or at adulthood. The populations show both phenotypic similarities and differences, as we have summarized in **Table 1**. All three models demonstrate a reduction in cardiac contractility, accompanied with elongation of the cardiomyocytes. On the other hand, the findings regarding the fetal gene expression, interstitial fibrosis, and thickening of the cardiomyocytes were different in the three models. When the knockout was introduced in the germline, it failed to induce expression of fetal genes and interstitial fibrosis, but the cardiomyocytes demonstrated thickening. Knockout in the perinatal period, in contrast, did produce an increase in the expression of fetal genes, but again in the absence of fibrosis, and without producing any change in the short-axis dimensions of cardiomyocytes. Knockout in adult mice produced expression of fetal genes, fibrosis, and cardiomyocyte atrophy. These results showed that, when the Mylk3 gene is eliminated in early stages of development, it is most effectively compensated. Compensation, in contrast, is ineffective when the Mylk3 gene is knockedout in adult mice. Notably, we included two controls in the inducible knockout studies, namely flox/flox and wild/wild/Cre mice with tamoxifen injection in the present and previous studies (Massengill et al., 2016), so as to eliminate any effects of the potential toxicity of tamoxifen and/or Cre transgene for cardiac contractility. Neither of the control populations showed any significant differences in HW/BW ratio or cardiac contractility.

### DISCUSSION

We have now shown that elimination of cMLCK in mice in the perinatal period results in an increase of heart weight/body weight ratio with decreased body weight, reduction of contractility accompanied by elongated cardiomyocytes, and

an increase in the expression of fetal genes. Under sustained pressure overload, adaptive cardiomyocyte hypertrophy transits to maladaptive elongation of cardiomyocytes, and persistent reductions in contractile force. Such cardiac remodeling is thought to involve an initial adaptive addition of sarcomeres in



parallel, thus thickening the cardiomyocytes, and a subsequent addition of sarcomere in series to produce elongatation (Diwan and Dorn, 2007; Balasubramanian et al., 2009; Russell et al., 2010; Rosca et al., 2013).

In adult cardiomyocytes, with their semi-crystalline architecture, and under the continuous production of myocardial force, the exact processes of cardiac remodeling remain unknown. In adult mice, nonetheless, expression of cMLCK has been shown to be reduced after pressure overload by ∼80% within 1 week. This effect persists, coinciding with the transition from compensated to decompensated hypertrophic heart failure (Warren et al., 2012). Furthermore, our recent study showed that elimination of cMLCK in adult mice with age of 10–12 weeks led to acute heart failure accompanied by cardiomyocyte atrophy with elongated and thinner cardiomyocytes within 1 week after tamoxifen injection. The heart failure persisted with slight further progression in cardiac contractility demonstrated in echocardiogram 2 weeks after tamoxifen injection. Furthermore, adult Mylk3-KO mice knockout mice fail to demonstrate adaptive pressure overloaded cardiomyocyte thickening (Massengill et al., 2016).

We could not examine the effects of pressure overload in neonatal hearts, such as seen in aortic stenosis in children, due to technical burdens. To best of our knowledge, it is unknown whether heart failure as seen in childhood also involves reduction of phosphorylation of MLC and cMLCK. We noted, nonetheless, that 19 days after tamoxifen injection into perinatal Mylk3-KO mice, there was a heart failure with increase of the heart weight/body weight ratio, reduction of contractility with increase of both systolic and diastolic left ventricular cavity with elongation but without thinning of cardiomyocytes, and increase of fetal gene expression compared to controls. We cannot explain why the heart weight was not increased despite the increase of surface area size and elongation of cardiomyocytes, but it might be possible that individual cardiomyocytes from perinatal Mylk3-KO mice are less heavy than those from controls.

In the perinatal Mylk3-KO mice, we were unable to detect interstitial fibrosis in the ventricles. We have previously analyzed mice with perinatal knockout of transcription factor, Nkx2-5, using floxed-Nkx2-5 mice. Nkx2-5 regulates multiple downstream targets including cMLCK (Chan et al., 2008), and perinatal Nkx2-5-KO mice demonstrate more profound heart failure compared to perinatal Mylk3-KO mice, with a 1.44 fold increase of HW/BW ratio at P12 (Briggs et al., 2008). The perinatal Nkx2-5 knockout mice did not show any apparent interstitial fibrosis (Briggs et al., 2008). To best of our knowledge, there are no additional reports describing geneticallyinduced perinatal models of heart failure in mice. Absence of fibrosis, however, has also been observed in neonatal mice following cardiac injury, such as myocardial infarction or partial ventricular resection, partly due to cardiomyocyte regeneration persisting in neonatal hearts within a few days after birth, and difference in extracellular matrix composition and immune responses (Porrello and Olson, 2014).

In mice, cMLCK is expressed from an early embryonic stage, at least beyond E10.5, and is restricted to the hearts when examined in major organs using Northern blotting (Chan et al., 2008). Because of this, we presume that the reduction in body weight observed in the perinatal Mylk3-KO mice is due to heart failure, but we cannot rule out the possibility that cMLCK may directly regulate body weight by unknown mechanisms. Additional experiments are now needed further to address and explore the mechanisms underscoring the growth retardation noted in these mice.

To mitigate the possibility that a high dose of tamoxifen, such as 80 mg/kg/day given over 5 days, would result in heart failure (Koitabashi et al., 2009), we utilized a substantially lower dose of tamoxifen, namely 50 mg/kg/day given but once. We also included control wildtype mice with CAGGS-Cre-ERTM mice, which showed no change in HW/BW ratio compared to another control, flox/flox mice, after tamoxifen injection, thus eliminating the possibility that the heart failure was due to presence of the CAGGS-Cre-ERTM transgene.

We recognize the limitations in our current study. The mechanisms underlying the loss of cMLCK leading to cardiomyocyte elongation are currently unknown, and future experiments are required to elucidate these processes. We have not examined perinatal Mylk3-KO beyond 19 days of age because these mice were noted to be unhealthy for weaning on day 21. The remodeling and functional differences between the germline and perinatal Mylk3-KO models may be due to the developmental age studied. Further experiments comparing remodeling and cardiac function in these two knockout models at the same ages, therefore, are highly desirable. Based on the profound heart failure seen in adult-inducible Mylk3-KO mice, nonetheless, we speculate that their cardiac function is unlikely to improve in later life.

In summary, we have demonstrated that the cMLCK plays essential roles in cardiac function during the perinatal period. Our experiments point to the need for further clinical studies in children with heart failure, exploring the phosphorylation of MLC and reduction of cMLCK.

### AUTHOR CONTRIBUTIONS

Experiments were designed and performed by YI, RJ, RC and HK. The manuscript was prepared by YI, RA, and HK.

### FUNDING

This work was supported by American Heart Association (14GRNAT20380822 to HK), University of Florida (T35 HL007489 to YI, and Opportunity Fund to HK).

### REFERENCES


### ACKNOWLEDGMENTS

We gratefully acknowledge valuable suggestions and technical support provided by D. Smith, E. Chan and K. Fortin for their valuable suggestions and technical support.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Islam, Joseph, Chowdhury, Anderson and Kasahara. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Predicting Effects of Tropomyosin Mutations on Cardiac Muscle Contraction through Myofilament Modeling

Lorenzo R. Sewanan1, <sup>2</sup> , Jeffrey R. Moore<sup>3</sup> , William Lehman<sup>4</sup> and Stuart G. Campbell 1, 5 \*

*<sup>1</sup> Department of Biomedical Engineering, Yale University, New Haven, CT, USA, <sup>2</sup> Yale School of Medicine, Yale University, New Haven, CT, USA, <sup>3</sup> Department of Biological Sciences, University of Massachusetts Lowell, Lowell, MA, USA, <sup>4</sup> Department of Physiology and Biophysics, Boston University School of Medicine, Boston, MA, USA, <sup>5</sup> Department of Cellular and Molecular Physiology, Yale School of Medicine, New Haven, CT, USA*

Edited by:

*P. Bryant Chase, Florida State University, USA*

#### Reviewed by:

*Brandon Biesiadecki, Ohio State University, USA Daniel Michele, University of Michigan, USA Fan Bai, University of Iowa, USA*

\*Correspondence: *Stuart G. Campbell stuart.campbell@yale.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *07 August 2016* Accepted: *03 October 2016* Published: *26 October 2016*

#### Citation:

*Sewanan LR, Moore JR, Lehman W and Campbell SG (2016) Predicting Effects of Tropomyosin Mutations on Cardiac Muscle Contraction through Myofilament Modeling. Front. Physiol. 7:473. doi: 10.3389/fphys.2016.00473* Point mutations to the human gene TPM1 have been implicated in the development of both hypertrophic and dilated cardiomyopathies. Such observations have led to studies investigating the link between single residue changes and the biophysical behavior of the tropomyosin molecule. However, the degree to which these molecular perturbations explain the performance of intact sarcomeres containing mutant tropomyosin remains uncertain. Here, we present a modeling approach that integrates various aspects of tropomyosin's molecular properties into a cohesive paradigm representing their impact on muscle function. In particular, we considered the effects of tropomyosin mutations on (1) persistence length, (2) equilibrium between thin filament blocked and closed regulatory states, and (3) the crossbridge duty cycle. After demonstrating the ability of the new model to capture Ca-dependent myofilament responses during both dynamic and steady-state activation, we used it to capture the effects of hypertrophic cardiomyopathy (HCM) related E180G and D175N mutations on skinned myofiber mechanics. Our analysis indicates that the fiber-level effects of the two mutations can be accurately described by a combination of changes to the three tropomyosin properties represented in the model. Subsequently, we used the model to predict mutation effects on muscle twitch. Both mutations led to increased twitch contractility as a consequence of diminished cooperative inhibition between thin filament regulatory units. Overall, simulations suggest that a common twitch phenotype for HCM-linked tropomyosin mutations includes both increased contractility and elevated diastolic tension.

Keywords: tropomyosin, tropomyosin-actin interactions, tropomyosin stiffness, cooperativity, hypertrophic cardiomyopathy, computational modeling, diastolic dysfunction

## INTRODUCTION

Point mutations in alpha tropomyosin (TPM1) are associated with inherited cardiomyopathies, most notably hypertrophic cardiomyopathy (HCM) and dilated cardiomyopathy (DCM) (Redwood and Robinson, 2013). While there is no consistent pattern predicting whether amino acid substitutions will lead to HCM versus DCM, mutations are known to alter the interactions within the tropomyosin coiled-coil itself and between tropomyosin and its binding partner actin (Bai et al., 2013; Redwood and Robinson, 2013). Studies on the classic HCM-related tropomyosin mutants E180G and D175N (Thierfelder et al., 1994) using electron microscopy, atomic force microscopy, and molecular dynamics on isolated tropomyosin have shown increased local and global flexibility (Li et al., 2010, 2012; Loong et al., 2012b). Changes in tropomyosin flexibility are widely believed to affect thin filament cooperativity (Loong et al., 2012a; Moore et al., 2016). However, recent work has shown that point mutations in tropomyosin can also affect the energy landscape of tropomyosin interactions with actin, providing another potential route for mutations to affect thin filament calcium regulation (Orzechowski et al., 2014a,b). Altered actin binding by mutant tropomyosin is also supported by numerous experimental and computational studies (Boussouf et al., 2007; Janco et al., 2012; Zheng et al., 2016). The surface interactions of tropomyosin and actin may ultimately affect myosin crossbridge formation, as suggested by length perturbation analysis studies (Bai et al., 2011). Based on these observations, we posit that emergent physiological effects of tropomyosin mutations can be described by considering the effects of residue changes on both the chain-like properties of the tropomyosin molecule and interaction of tropomyosin with the actin surface as it fluctuates between regulatory states.

In order to construct a paradigm for exploring the effects of tropomyosin properties on muscle activation, we employed a computational approach. Beginning with a previous model of thin filament calcium activation (Aboelkassem et al., 2015), we reconsidered the formulation of tropomyosin neighbor interactions, bringing it into closer agreement with structural measurements of tropomyosin movement on actin. The completed model includes parameters that unambiguously relate to three potential mutation-dependent molecular properties of tropomyosin, namely tropomyosin chain stiffness, blockedclosed equilibrium, and the crossbridge duty cycle. Making use of published data, we found that the model was capable of capturing steady-state behavior of cardiac muscle preparations containing wild-type and mutant tropomyosins. Furthermore, the simulations suggest that HCM-related tropomyosin mutations produce hypercontractile twitch phenotypes with diastolic dysfunction.

### METHODS

### Model Formulation

Consider an actin regulatory unit (RU) having a current state Z ∈ {B, C, M} after the three state model established by McKillop and Geeves (1993) (**Figures 1A,B**). In order to transition to a different state, Z ∗ , the RU must overcome an activation energy which we denote 1G ref ZZ<sup>∗</sup> .

Because direct transitions between B and M are prohibited (**Figure 1A**), this yields a total of four reference activation energies, 1G ref BC, 1G ref CB, 1G ref CM, and 1G ref MC (**Figure 1D**). These energies correspond to theoretical reference conditions in which the RU is not connected to its nearest neighbors. From these, it is possible to construct a baseline free energy landscape for RU transitions (depicted conceptually as the gray line in **Figure 1D**).

Having established a reference free energy landscape, the effects of nearest neighbors can be evaluated. Consider a thin filament formed of discrete RUs connected in series (Campbell et al., 2010). Each RU has structural links with its neighbors through tropomyosin-tropomyosin overlap along the thin filament (McLachlan and Stewart, 1975). We include the effect of Tm-Tm overlap by determining the potential energy stored in the entire Tm chain itself when adjacent RUs do not occupy the same state (**Figures 1B,E**). The potential energy is assumed to arise from distortion of the Tm chain according to the difference in azimuthal angle between adjacent Tm. In preliminary calculations, we determined the energy stored in an elastically jointed chain cantilevered at one end and loaded (orthogonal to the un-deformed chain) at the other (**Figure 1B**). We found that the total chain energy was proportional to the square of the azimuthal angle change subtended by the bending chain. In terms of stored energy, this result is equivalent to a virtual spring element connecting RUs azimuthally. Accordingly, rather than represent a continuous chain in the model, we adopted a mechanical analog in which each Tm was considered attached to its neighbors via a linear elastic element (**Figure 1C**). When adjacent RUs occupy the same state, the distortion in their linking spring and its developed elastic force is zero. When they occupy different states, the force is proportional to the azimuthal angle (φ) that separates them. The constant of proportionality is the effective Tm chain stiffness γ . Hence, the potential energy contributed to an RU by its left neighbor (having state X ∈ {B,C,M}) is:

$$
\Delta U\_Z^X = \frac{1}{2}\gamma (\phi\_Z - \phi\_X)^2 \tag{1}
$$

The same equation applies analogously to the right neighbor, having state Y ∈ {B,C,M}:

$$
\Delta U\_Z^Y = \frac{1}{2}\chi (\phi\_Z - \phi\_Y)^2 \tag{2}
$$

The azimuthal angles of tropomyosin for each state are taken from structural data (Vibert et al., 1997; Poole et al., 2006), namely φ<sup>B</sup> = 0 ◦ , φ<sup>C</sup> = 25◦ , and φ<sup>M</sup> = 35◦ . We further assume that the peak free energy barrier between states coincides with an azimuthal angle precisely halfway between the angles associated with each state, that is:

and

$$
\phi\_{CM} = \phi\_{MC} = \frac{1}{2} \left( \phi\_M - \phi\_C \right) + \phi\_C
$$

1 2

φBC = φCB =

The potential energy due to Tm chain distortion at these intermediate angles, accounting for the left-neighbor state X is:

$$
\Delta U\_{ZZ^\*}^X = \frac{1}{2} \gamma (\phi\_{ZZ^\*} - \phi\_X)^2,\tag{3}
$$

(φ<sup>C</sup> − φB) + φ<sup>B</sup>

considers the free energy barrier from the blocked state to closed state to open state. (E) Magnitude of strain energy contribution for a given neighbor X depends on the intermediate angles between a transition for the state of a given regulatory unit and the state of the neighbor. (F) The overall energy barriers for the transitions of a

where Z is the current state of the RU, and Z ∗ is the state into which the RU is transitioning. Similarly, the potential energy contribution of the right neighbor (having state Y) is

$$
\Delta U\_{ZZ^\*}^Y = \frac{1}{2}\gamma (\phi\_{ZZ^\*} - \phi\_Y)^2 \tag{4}
$$

given regulatory unit depends on the contribution of the strain energy with respect to the states of its neighbors X and Y.

The final activation energy for any given transition ZZ<sup>∗</sup> with neighboring RU states X and Y is obtained by altering the reference activation energy as follows:

$$
\Delta G\_{ZZ^\*}^{XY} = \Delta G\_{ZZ^\*}^{ref} - \Delta U\_Z^X + \Delta U\_{ZZ^\*}^X - \Delta U\_Z^Y + \Delta U\_{ZZ^\*}^Y \tag{5}
$$

In other words, distortion-based increases in potential energy raise energy wells, hence the two negative terms in the above equation that diminish the reference activation energy. On the other hand, distortion increases the height of the peaks in between states, adding to the activation energy. This explains the positive potential energy terms.

Once activation energies are defined, neighbor-dependent kinetic rates can be obtained via the Eyring equation:

$$k\_{ZZ^\*}^{XY} = \frac{k\_b T}{h} e^{-\Delta G\_{ZZ^\*}^{XY}/RT} \tag{6}$$

For convenience, we define a reference kinetic rate based on the reference energy change 1G ref ZZ<sup>∗</sup> :

$$k\_{ZZ^\*}^{ref} = \frac{k\_b T}{h} e^{-\Delta G\_{ZZ^\*}^{ref}/RT}$$

Using Equation (5), k ref ZZ<sup>∗</sup> can be substituted into Equation (6):

$$k\_{ZZ^\*}^{XY} = k\_{ZZ^\*}^{ref} e^{\left(-\Delta U\_Z^X + \Delta U\_{ZZ^\*}^X - \Delta U\_Z^Y + \Delta U\_{ZZ^\*}^Y\right)} \tag{7}$$

Equation (7) casts tropomyosin transition rates as a reference kinetic rate scaled by a term that reflects that status of nearest neighbors as well as the apparent tropomyosin chain stiffness (Equations 1–4). In order to conform to nomenclature of previous models, C→M and M→C transition rates are referred to as f and, g as these transitions are driven by crossbridge binding and unbinding, respectively:

$$\begin{aligned} f^{XY} &= k\_{CM}^{XY} \\ g^{XY} &= k\_{MC}^{XY} \end{aligned}$$

In addition to tracking the tropomyosin state of each RU, we also represent Ca2<sup>+</sup> binding to TnC. As introduced in Aboelkassem et al. (2015) we permit all three tropomyosin RU states to be either Ca2<sup>+</sup> bound or Ca2<sup>+</sup> free (denoted by subscripts 1 or 0, respectively). This produces a six-state RU model (**Figure 1A**). In order to reflect the fact that transition away from the B state seldom occurs without Ca2<sup>+</sup> first binding to TnC, the kinetic rate k XY BC is scaled by the factor λ for the transition B0→C0. The Ca2<sup>+</sup> dissociation rate k d Ca is also scaled by λ for C1→C<sup>0</sup> and M1→M<sup>0</sup> transitions in order to satisfy microscopic reversibility (Aboelkassem et al., 2015). Ca2<sup>+</sup> association with the RU is governed by the second order rate constant kCa+.

### Model Implementation

A total of 24 RUs, each obeying the six-state scheme described above, were connected in series to form a virtual thin filament. Dummy RUs, one on each end of the filament, were added and permanently fixed in the B<sup>0</sup> position to define boundary conditions. Simulations were performed on the system using a Markov chain Monte Carlo algorithm described previously (Aboelkassem et al., 2015). Force produced by the model was also computed as before by averaging the output of many repeated simulations. The time step size for simulations was automatically determined for each parameter set such that the maximum cumulative probability of transition for any combination of nearest neighbor states never exceeded 0.7. To guarantee the accuracy of this heuristic threshold, simulations were tested for convergence at other time step values. The 0.7 transition threshold proved more than sufficient to guarantee temporal convergence in each case.

Simulation protocols included both activations at a single, constant Ca2<sup>+</sup> concentration and under time-varying intracellular Ca2<sup>+</sup> concentration (for twitch). Data for fitting steady-state and twitch activation (Janssen and de Tombe, 1997; Dobesh et al., 2002; Bai et al., 2011) were digitized using a custom script that loaded a bitmap of each figure into MATLAB and located the pixel corresponding to the centroid of each data markers. Digitized data points were then overlaid on the original images to check for fidelity. RUs were always set to an initial state of B<sup>0</sup> at the beginning of each simulation. In order to obtain steady-state force and rates of tension recovery (ktr), the model was run for a 7500 ms interval (achieving steady force) and then all RUs in M states were instantaneously transitioned into C states. The rate at which force recovered to the steady-state value was used to determine ktr (**Figure 2A**). The model was implemented in CUDA C++ and executed on an Nvidia Tesla K40 graphic processing card. Custom MATLAB scripts were created to manage simulation setup and data flow.

### RESULTS

Implementation of the new formulation (**Figure 1**) of tropomyosin nearest neighbor interaction constrained by biophysical measurements on tropomyosin azimuthal change (Poole et al., 2006) decreases the number of free parameters of the model by four. This constitutes a drastic reduction in potentially adjustable parameters and in model dimensionality compared to the previous formulation (Aboelkassem et al., 2015). To determine whether the model could still lead to behavior consistent with realistic calcium activation of muscle, we used constrained particle swarm optimization to find a set of parameters (**Table 1**, Set 1) that accurately matched calcium-dependent steady-state force production reported in skinned rat trabeculae (**Figure 2**; Dobesh et al., 2002). The dataset of Dobesh et al. (obtained at a sarcomere length of 2.25µm) was selected because of the meticulous attention paid to sarcomere length stability in that study. The simulation reproduced not only the steady state forcepCa relationship as reported (**Figure 2B**) with roughly the same Hill coefficient and pCa<sup>50</sup> over the experimental pCa range (**Table 2**) but also showed a strong dependence of the rate of force redevelopment on pCa (**Figure 2A**), consistent with experimental observation (Fitzsimons et al., 2001). The simulation furthermore manifested asymmetry in the lower half of the force-pCa relationship compared to the upper half such that a higher Hill coefficient can be calculated at the lower half than the upper half (**Figure 2B**). While an asymmetric cooperativity leads to deviation of the simulation curve from an idealized Hill curve with the same apparent cooperativity, real muscle including the skinned trabeculae modeled here defies uniform cooperativity in its force-pCa relationship (Dobesh et al., 2002).

Having obtained a parameter set for skinned cardiac muscle, we perturbed each of the model parameters that could potentially be affected by tropomyosin mutations (Li et al., 2012; Bai et al., 2013; Orzechowski et al., 2014a) in order to understand their separate effects on the steady state force-pCa relationship (**Figure 3**). The same relative changes in tropomyosin stiffness γ , BC equilibrium constant KBC, and myosin duty cycle δ were used to generate force-pCa curves (**Figures 3A–C**) with their properties quantified as maximum force, fraction of maximum force present at diastolic (low) calcium, calcium sensitivity, and Hill coefficient. The three troponin-related parameters (λ, kCa+, kCa−) were not perturbed in this analysis. The forward rates of BC transition and of myosin binding were also excluded


#### TABLE 1 | Parameters used for fitting data and running simulations.

#### TABLE 2 | Tropomyosin wild-type fit simulation and data properties.


because the absolute rates have little effect on the steady state force-pCa curves (data not shown). The maximum steady state force production was strongly dependent on the myosin duty cycle whereas tropomyosin stiffness and BC equilibrium had only small effects on the maximum force (**Figure 3D**). Examining the effect closely of stiffness and BC equilibrium revealed that they had primarily fine tuning but opposing effects on maximum force production, with higher stiffness slightly decreasing force production and higher BC equilibrium slightly increasing force production. The force present at low calcium (**Figure 3E**) was most sensitive to tropomyosin stiffness. Decreasing the stiffness parameter causes a drastic, non-linear increase in force at low calcium. Proportional increases in BC equilibrium and duty cycle increased force at low calcium, but these changes were modest by comparison (**Figure 3E**). While increases in BC equilibrium constant and duty cycle tended to increase the calcium sensitivity of the thin filament in these simulations (**Figure 3F**), increased stiffness had the opposite effect. BC equilibrium had a larger effect on calcium sensitivity than duty cycle. Cooperativity (as assessed by the Hill coefficient) was affected by all three parameters, but more particularly myosin duty cycle and tropomyosin stiffness (**Figure 3G**). Specifically, a higher BC equilibrium resulted in just a modest increase in cooperativity but ultimately reached a plateau. Of the three parameters, tropomyosin stiffness displayed the most potent modulation of cooperativity.

We next sought to determine whether the effects of tropomyosin mutations on thin filament activation could be captured by altering tropomyosin-related model parameters. To do this, we analyzed a steady-state force-pCa dataset (Bai et al., 2011) measured in bovine cardiac muscle in which thin filament proteins were extracted and reconstituted with 100% human recombinant wild-type or mutant alpha tropomyosin. Importantly, the tension of the reconstituted filaments in that study was carefully measured with respect to a reference tension at 0◦C in the relaxing solution, allowing analysis of not only the calcium dependent activation but also the calcium independent activation of the muscle. We began by fitting the wild-type data using a constrained particle swarm optimization and found a set of parameters (**Table 2**, Set 2) that produced a high fidelity simulation of the reconstituted fibers with wild-type tropomyosin (**Figures 4A,E**); the Hill coefficient and pCa<sup>50</sup> of the wild-type simulation curve was almost identical to that of the experimental curve (**Table 2**). This parameter set was similar to the parameter set for skinned cardiac muscle (Set 1) but included a lower tropomyosin stiffness and non-zero lambda, which permits loose coupling of troponin and tropomyosin (Aboelkassem et al., 2015). These differences seem reasonable in light of the reconstitution process as well as species differences (rat vs. human/bovine). Previous electron microscopy studies and molecular dynamics simulations have established that E180G and D175N mutations increase the mean deviation angle of the tropomyosin coiled-coil from a wild-type value of 22.0 degrees to 27.6 degrees for E180G and 30.5 degrees for D175N (Li et al., 2012), representing a 25 and 40% decrease in stiffness respectively. We ran simulations by altering only the stiffness of the WT fit by these proportions (**Figures 4B,F**) and found that these changes alone were insufficient to reproduce the differences between the mutants and the wild-type data. This suggests that mutation effects are not confined to stiffness alone and that other molecular mechanisms should be considered. Indeed, tropomyosin mutants have previously been determined to alter the interactions between tropomyosin and the actin surface (Orzechowski et al., 2014a; Zheng et al., 2016), which may in turn affect both the blocked-to-closed and closed-toopen (myosin-induced) transitions of tropomyosin across actin. We therefore entertained the possibility that introducing changes to the BC equilibrium constant and duty cycle in addition to the assumed stiffness changes could produce a reasonable fit to mutant data. In order to be thorough, a two-dimensional parameter space for KBC and δ was explored for each mutation (**Figures 4C,D**). In both cases, we found a global minimum in the error landscape, indicating constrained solutions for fitting

FIGURE 4 | Fitting of wild-type and mutant tropomyosin reconstituted myocardium steady state force production data. (A) Baseline parameters were found that allowed the model to fit the wild-type data, with D175N tropomyosin mutation data shown for comparison. (B) In order to fit the D175N data, tropomyosin stiffness was first reduced by 40% and was not sufficient to fit the data. (C) Duty cycle and BC equilibrium were varied systematically while root mean square error of the model output and the D175N data was calculated, with an error minimum which maintained the baseline duty cycle but decreased the BC equilibrium constant by 15%. (D) The D175N simulation from a prescribed stiffness change and an optimized BC equilibrium decrease show a qualitative fit with the D175N mutant experiment data. (E) Wild-type data and the model fit shown, with E180G tropomyosin mutation data shown for comparison. (F) In order to fit the E180G data, tropomyosin stiffness was first reduced by 25% which was not sufficient to fit the data. (G) The error minimum for the E180G heatmap after applying the 25% stiffness reduction was found to include both a 10% increase in BC equilibrium and a 15% increase in duty cycle. (H) After applying the stiffness change, BC equilibrium change, and duty cycle change, the E180G simulation shows a qualitative fit with the experimental data for the E180G mutation.

the mutant force-pCa curves. For D175N, the optimal simulation demanded a decrease in the BC equilibrium by 15%, favoring the blocked state, in addition to the measured 40% decrease in tropomyosin stiffness (**Figure 4D**). For E180G, the optimal simulation required both an increase in the BC equilibrium by 10% and an increase in the myosin duty cycle by 15%, in addition to the reported 25% decrease in tropomyosin stiffness (**Figure 4H**). The model therefore suggests that the molecular consequences of a tropomyosin mutation can include changes in flexibility, the blocked-to-closed tropomyosin transition, and the crossbridge-mediated closed-to-open tropomyosin transition but varies relatively between mutations.

Understanding how mutation-driven changes in skinned fiber behavior might relate to the twitch characteristics of intact muscle is far from straightforward. The model provides a means of doing this, by extending the parameter changes determined in skinned fibers to predict intact muscle contraction. We first established a baseline set of intact muscle parameters (**Table 1**, Set 3) by fitting an isometric rat papillary muscle twitch record in response to its measured calcium transient (Janssen and de Tombe, 1997). The new model reproduced this dynamic calciumactivated response just as well as our previous (higher order) models (**Figure 5A**). In order to gauge the effects of tropomyosin stiffness on dynamic muscle activation, we ran twitch simulations with the same baseline parameter set and calcium transient while varying γ (**Figure 5B**). A striking result was the effect of stiffness on maximum activation of the muscle (**Figure 5C**). Increasing tropomyosin stiffness stunted force production potently above γ values of ∼70 kJ/mol. At the same time, lowering stiffness resulted in higher fraction of the maximum force being present at diastolic calcium levels (**Figure 5C**). The kinetics of muscle activation as measured by time to peak force (TTP) and time from peak force to 50% relaxation of muscle (RT50) were also impacted (**Figure 5C**). Lowering stiffness tended to shorten TTP and increase RT50. Altogether, a more flexible tropomyosin resulted in increased maximum systolic force, modestly increased diastolic force, shortened TTP, and prolonged RT50.

Using model parameters (**Table 1**, Set 3) fitted to mutant tropomyosin fiber data (**Figure 4**), we predicted D175N and E180G effects on twitch force dynamics. For the mutant D175N, we applied a 40% decrease in stiffness and a 15% decrease in BC equilibrium constant (**Table 1**, Set 4) and found that the twitch was stronger overall with a small increase in diastolic force fraction (**Figures 6A,E,F**). Twitch kinetics were also altered, with a shorter TTP and longer RT50 (**Figures 6C,D**). For E180G, we applied a 25% increase in tropomyosin flexibility, 10% increase in BC equilibrium constant, and 15% increase in duty cycle (**Table 1**, Set 5). This yielded a twitch that had a much higher maximum force production and diastolic fraction. It also had a much shorter TTP and a much longer RT50 (**Figures 6B–F**). While the two twitches produced different properties, they both embodied faster activation, slower relaxation, and increased systolic and diastolic force compared to the wild-type.

### DISCUSSION

We have considered the potential impacts of tropomyosin mutations on cardiac muscle function using a new structurally motivated thin filament model. Accounting for experimentally

determined azimuthal shifts in tropomyosin (Poole et al., 2006) allowed us to reduce the dimensionality of our previous model (Aboelkassem et al., 2015) from 12 free parameters to just eight. In effect, four parameters that were considered independent are constrained by structural data and thermodynamic principles in the new formulation. Along with the reduced dimensionality, these constraints have made the model more realistic. It recapitulates observed features of cardiac muscle activation, including distinct Hill coefficients for upper and lower halves of the steady-state force-pCa curve (**Figure 2B**; Dobesh et al., 2002), strong dependence of the rate of force redevelopment on calcium (**Figure 2A**; Brenner, 1988; Fitzsimons et al., 2001) and realistic calcium-activated twitch transients (**Figure 5A**). As the model's parameters also directly relate to the molecular properties of tropomyosin, it provided a means for systematic investigation of HCM-linked tropomyosin mutations. Simulation-based analysis of skinned fiber data suggests that functional consequences of TPM1 mutations may be explained by three fundamental properties, namely tropomyosin chain stiffness, BC equilibrium, and tropomyosin-mediated changes in the crossbridge duty cycle.

Our results suggest that each tropomyosin mutation causes a unique combination of changes to these three molecular properties, and that specific changes can be estimated by model analysis when the right data are available. At first examination, the ability of skinned fiber exchange experiments to discriminate between tropomyosin molecular property changes seemed limited. For instance, the pCa value for half-activation of force (pCa50) is predicted to be highly sensitive to all three properties (**Figure 3F**). However, if the mutation effect on tropomyosin stiffness is known a priori, the model is able to predict its concomitant impact on pCa50, Hill coefficient, and maximum/minimum forces (**Figures 4B,F**). The remaining discrepancies with mutant data can then be reasonably ascribed to changes in actin-tropomyosin surface interactions, appearing as changes to the BC and CM equilibria. Although these two properties have similar effects on the steady-state forcepCa curve, only the CM equilibrium (i.e., crossbridge duty cycle) is predicted to significantly impact maximum calciumactivated tension. As a consequence, a simple two-parameter search was able to minimize error between the model and the mutant fiber data and yield unique parameter values for KBC and δ (**Figures 4C,G**). Our approach was facilitated by recognizing that kinetic rate pairs in the model reduce to equilibrium coefficients when fitting steady-state measurements. The subsequent assumption that tropomyosin mutations do not directly affect the Ca2<sup>+</sup> affinity of troponin C left only KBC and δ as free parameters.

It is worth noting that the precision with which Bai et al. (2011) performed their skinned fiber studies with wild-type and mutant tropomyosins was critical to our approach. The calciumfree force produced in skinned preparations is often not reported due to technical challenges. Instead, force values at very low calcium concentrations are frequently assumed to be equivalent to zero. However, as our analysis demonstrates, force-pCa curves that are made to start at zero do not allow clear differentiation between competing biophysical changes such as tropomyosin stiffness and the apparent crossbridge duty cycle. Hence, while D175N and E180G mutations have been studied by others in various experimental systems (Bottinelli et al., 1998; Michele et al., 1999; Muthuchamy et al., 1999; Evans et al., 2000; Wang et al., 2011; Rysev et al., 2012), we selected the data of Bai et al. for the analysis here because it is the only applicable data set of which we are aware that reports absolute force measurements under very low calcium conditions.

Although tropomyosin stiffness can be directly studied via electron microscopy as well as atomic force microscopy (Li et al., 2010, 2012; Loong et al., 2012b) and simulated using molecular dynamics, no comparable approaches yet exist for obtaining estimates of the intrinsic BC and CM state equilibria. Hence, direct validation of our predictions for the E180G and D175N mutation effects on KBC and δ are not possible at present. Instead, we must rely on indirect evidence to corroborate our estimates. For instance, in previous work, the electrostatic interactions was computed between tropomyosin and actin residues as tropomyosin is moved axially and azimuthally over the actin surface (Orzechowski et al., 2014a). This allows construction of energy landscapes for wild-type and mutant tropomyosins. In

diastolic force production though E180G had a larger effect.

the absence of a reliable way to calculate entropic changes in this milieu, a direct estimate of the free energy changes between B and C states cannot be accurately made. However, qualitative insight is possible. We predicted from fitting fiber data that the E180G mutation would cause a 10% increase in KBC, indicating a relative increase in the free energy associated with the B state. This agrees well with the prediction of Orzechowski et al. (2014a) that the E180G substitution would cause a ∼40 kJ/mol increase in the electrostatic potential of the B state. Corresponding analysis of the actin surface interactions of D175N tropomyosin showed the changes in that case were quite small. Although we found a global fit for D175N suggesting a 15% decrease in KBC (**Figure 4C**), the slope of the error surface is sufficiently shallow that reasonable fits can also be obtained by assuming no change in KBC and a slight decrease in δ. In light of the predicted absence of actin surface interaction changes and the shallowness of the error landscape (**Figure 4C**), we conclude that the majority of D175N effects are explained by reduced tropomyosin chain stiffness (**Figure 4B**). Placing less importance on equilibrium effects also seems prudent in light of at least one study (Evans et al., 2000) that reports increased myofilament Ca2<sup>+</sup> sensitivity in mouse myocardium expressing D175N tropomyosin, as opposed to the unchanged pCa<sup>50</sup> seen in the data of Bai et al. (2011). Increased Ca2<sup>+</sup> sensitivity is consistent with the model-predicted effects of a pure decrease in tropomyosin stiffness (**Figure 4B**).

Another way of indirectly validating our predictions is to look at how E180G affects intact muscle twitches. Our model generally predicts outcomes that are consistent with those data. When predicting the effect of HCM-related mutations E180G and D175N on intact muscle, we found that E180G is more severe than D175N (starting from a rat twitch background and calcium transient). This aligns with observations that transgenic mice expressing E180G show a severe phenotype and often die within 6 months of age, while those expressing D175N display a milder phenotype (Redwood and Robinson, 2013). At the same time, both mutations were predicted in the model to cause increased diastolic and systolic tension. Similar behavior has been consistently demonstrated in studies of myocardium containing mutant tropomyosin. Isometric force measurements of permeabilized myocytes with both adenovirally expressed D175N and E180G showed high force production at low calcium (Michele et al., 1999). In terms of twitch kinetics, only E180G has been extensively characterized, but these results largely agree with our predictions. Papillary muscles isolated from transgenic E180G mice showed a slightly longer time to peak and a severe slowing of relaxation by almost 60% compared to control (Sheehan et al., 2011). Adult cardiomyocytes isolated from these animals showed increased magnitude of unloaded shortening of approximately 300% (Sheehan et al., 2011), which also agrees with our predictions.

One puzzling component to understanding pathogenicity of D175N tropomyosin is that is displays a reduced calcium sensitivity compared to wild-type, and no increase in maximally activated steady-state force (Bai et al., 2011). This seemingly contradicts the notion that HCM mutations are generally hypercontractile in nature. However, we find it interesting that our simulated D175N twitch is in fact hypercontractile (**Figure 6A**), with a predicted 28% increase in peak twitch force. According to the model, cooperativity-linked kinetic effects explain how D175N can increase twitch tension even while having a reduced steady-state calcium sensitivity. The large drop in tropomyosin stiffness associated with D175N means that nearest-neighbor RUs are less strongly coupled. At sub-maximal calcium levels (such as those occurring in a transiently-activated twitch event), weaker RU-RU coupling actually allows tension to develop more rapidly. Hence, although D175N ultimately limits the steady-state force, it develops force more quickly and thus reaches a higher level than wild-type during a short twitch event. This modeling result illustrates the limitations of using steady-state force-pCa curves and their properties as the sole means of understanding mutation pathogenicity.

In addition to the primary sarcomeric effects, HCM is characterized by myocardial remodeling, specifically concentric hypertrophy, fibrosis, and myocardial disarray (Teerkaririkul et al., 2012). Excess diastolic tension (Bai et al., 2013) as well as hypercontractility, may represent a common pathway for how tropomyosin mutations act to initiate HCM-type ventricular remodeling. Recent work suggests that increased myofilament tension during the twitch cycle, calculated as the scalar tension-time integral, induces ERK1/2 signaling leading to HCM-like hypertrophic remodeling (Davis et al., 2016). Our twitch predictions for E180G and D175N both achieve higher peak tension, and have tension-time integrals that exceed wild-type. Even though the final mutant twitches appear quite different from each other, the tension-time integral could explain why both are capable of driving concentric cardiomyocyte hypertrophy. Reducing myofilament activation in E180G mice by either increasing calcium uptake through knocking out phospholamban (Gaffin et al., 2011) or by co-expression of pseudophosphorylated TnI (Alves et al., 2014) was previously shown to decrease ERK1/2 signaling and prevent pathological hypertrophy. Furthermore, our model suggests than D175N would appear milder than E180G when it comes to these metrics, which is consistent with evidence that E180G has much more severe remodeling and disease than D175N (Redwood and Robinson, 2013).

While the twitch predictions show fairly extreme changes in the presence of mutations, particularly for E180G, it should be noted that these are the simulated effects of 100% replacement of wild-type tropomyosin with mutant tropomyosin. The clinical scenario is certainly different. In studies of HCM patients, monoallelic D175N is expressed to a similar amount as wild-type (Bottinelli et al., 1998) which would lead to coiled-coils consisting of a presumed mixture of 25% wild-type homodimers, 25% mutant homodimers, and 50% heterodimers. Recent biochemical studies with E180G and D175N show that the 1:2:1 ratio of dimers is formed in solution and that there was only a very slight preference for wild-type versus mutant tropomyosin when it came to binding of actin (Janco et al., 2012). However, for other properties of heterodimers such as calcium sensitivity, there was not necessarily a consistent trend of heterodimers mimicking homodimers (Janco et al., 2012). Therefore, our model utilizing complete replacement of wild-type homodimers with mutant homodimers would predict the worst case scenario. Furthermore, our model does not take into account any cellular adaptation that may occur as a result of a sarcomeric perturbation, such as tropomyosin phosphorylation, calcium transient changes through regulation such as phospholamban, and other myofilament post-translational changes (Teerkaririkul et al., 2012; Redwood and Robinson, 2013). Presumably, some of these adaptations may act to alleviate the functional phenotype and lead to milder dysfunction than predicted.

Although the model provides valuable insights, it is important to consider some of its limitations. Like our previous models, we employed an RU-level discretization of the thin filament. This means that any effects arising from intra-RU phenomena are absent and that effectively only one myosin crossbridge is represented per RU (see discussion in Campbell et al., 2010). Also, at present the model is only capable of simulating isometric contraction. We did not explicitly represent tropomyosintroponin T interactions in the model, even though recent evidence has emerged suggesting their potential modulation by tropomyosin mutations (Orzechowski et al., 2015) and possible effects on tropomyosin chain stiffness (Sousa et al., 2010). Such interactions are still of unclear significance with regards to thin filament regulation, but could explain some shortcomings of the present model. Specifically, the model deviates from the measured E180G exchange data at low Ca2<sup>+</sup> concentrations (**Figure 4H**). The only way of reconciling this deviation is to assume that the drop in overall tropomyosin chain stiffness is not as large as that measured for individual E180G tropomyosin dimers. It is possible that in the presence of troponin T the effect of E180G on stiffness is reduced. With the appropriate structural information, tropomyosin-troponin T interactions could be included in future modeling efforts.

Of possible consequence to the present study, the discretized thin filament requires us to translate proportional tropomyosin stiffness into energy via a simple unidimensional deformation. As such, we do not account for some of the more complex motions of tropomyosin geometry as considered in continuous flexible chain type models (Smith and Geeves, 2003; Smith et al., 2003; Mijailovich et al., 2012; Metalnikova and Tsaturyan, 2013; Land and Niederer, 2015). One possible consequence of leaving out this complexity is that the simplified model might not behave in a realistic matter. Given the model's ability to produce several fundamental muscle phenomena, any such limitations seem likely to be second order in nature. Another possibility is that the simplified model yields realistic behavior but can only do so when assuming unrealistic parameter values. To examine this, we compared our estimate of tropomyosin chain bending stiffness per unit length (obtained by fitting cardiac muscle force-pCa data) to the estimate used by Smith et al. (2003) and Smith and Geeves (2003). They estimated a value of 2.5×10−<sup>44</sup> Nm<sup>4</sup> for the bending stiffness per unit length; while using our fitted value for the apparent chain stiffness (**Table 1**), we would back calculate a value of 1.3 × 10−<sup>44</sup> Nm<sup>4</sup> . Considering the simplifications involved, the agreement of these values to well within an order of magnitude seems reasonable.

A few additional details surrounding the exchange experiments of Bai et al. (2011) should be considered. They used recombinant human tropomyosin with an added Gly-Ser sequence at the N-terminus to approximate acetylation (Monteiro et al., 1994). N-terminal acetylation of tropomyosin stabilizes its coiled-coil structure and hence could be an important determinant of overall tropomyosin chain stiffness (Greenfield et al., 1994). Although the work of Monteiro et al. (1994) that the dipeptide extension is functionally equivalent to acetylation, there may still be differences between native and recombinant tropomyosins. This may explain why values for the effective tropomyosin chain stiffness (γ ) fit to skinned rat cardiac tissue data differ from those fit to bovine myocardium

### REFERENCES


reconstituted with recombinant human tropomyosin. Another factor to contemplate is that the various experimental data sets were acquired at different temperatures. In order to ascribe the effects of tropomyosin mutations on γ and other parameters as uncovered by fitting to the data of Bai et al. (2011) we computed their proportional effect and then applied the same proportional changes to the baseline parameters fitted to intact rat cardiac twitch data.

Although significant progress has been made already toward modeling the thin filament and tropomyosin at different scales, we present here the first attempt of which we are aware to model tropomyosin mutations in an integrative manner and predict their effects on intact muscle function. In the future, the spatially discretized thin filament approach will also allow simulations with stochastic assembly of wild-type and mutant homodimers and heterodimers which may have important implications for pathogenesis (Janco et al., 2012). We are also continuing work toward a discretized coarse-grain model of tropomyosin and incorporation of sarcomere length dependence, which will improve the ability of the model to translate biophysical alterations of tropomyosin into functional simulations.

### AUTHOR CONTRIBUTIONS

LS: Derived the model, implemented the model, designed and ran simulations, and wrote the paper; JM: Assisted with model derivation and writing of the paper; WL: Assisted with model derivation and writing of the paper; SC: Derived the model, designed simulations, and wrote the paper.

### ACKNOWLEDGMENTS

We would like to thank Jordan Bonilla for help with programming in CUDA C++. This work was supported by NIH Grant HL126025 and NSF Grant 1562587 (both to SC), NIH Grant R37-HL036153 (to WL), and NIH Grants HL077280 and HL123774 (both to JM). LS was supported by a NIH/NIGMS Medical Scientist Training Program Grant (T32GM007205) and by a NASA CT Space Grant, PTE Federal Award No. NNX15AI12H.


in adult cardiac myocytes. Nat. Med. 5, 1413–1417. doi: 10.1038/ 70990


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Sewanan, Moore, Lehman and Campbell. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Recent Advances in the Molecular Genetics of Familial Hypertrophic Cardiomyopathy in South Asian Descendants

#### Jessica Kraker <sup>1</sup> , Shiv Kumar Viswanathan<sup>1</sup> , Ralph Knöll 2, 3 and Sakthivel Sadayappan<sup>1</sup> \*

<sup>1</sup> Department of Internal Medicine, Heart, Lung and Vascular Institute, Division of Cardiovascular Health and Sciences, University of Cincinnati College of Medicine, Cincinnati, OH, USA, <sup>2</sup> AstraZeneca R&D Mölndal, Innovative Medicines and Early Development, Cardiovascular and Metabolic Diseases iMed, Mölndal, Sweden, <sup>3</sup> Integrated Cardio Metabolic Centre, Karolinska Institutet, Myocardial Genetics, Karolinska University Hospital in Huddinge, Huddinge, Sweden

#### Edited by:

P. Bryant Chase, Florida State University, USA

#### Reviewed by:

Brenda Schoffstall, Barry University, USA Kerry S. McDonald, University of Missouri, USA Charles K. Thodeti, Northeast Ohio Medical University, USA

> \*Correspondence: Sakthivel Sadayappan

sadayasl@ucmail.uc.edu

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 30 August 2016 Accepted: 12 October 2016 Published: 28 October 2016

#### Citation:

Kraker J, Viswanathan SK, Knöll R and Sadayappan S (2016) Recent Advances in the Molecular Genetics of Familial Hypertrophic Cardiomyopathy in South Asian Descendants. Front. Physiol. 7:499. doi: 10.3389/fphys.2016.00499 The South Asian population, numbered at 1.8 billion, is estimated to comprise around 20% of the global population and 1% of the American population, and has one of the highest rates of cardiovascular disease. While South Asians show increased classical risk factors for developing heart failure, the role of population-specific genetic risk factors has not yet been examined for this group. Hypertrophic cardiomyopathy (HCM) is one of the major cardiac genetic disorders among South Asians, leading to contractile dysfunction, heart failure, and sudden cardiac death. This disease displays autosomal dominant inheritance, and it is associated with a large number of variants in both sarcomeric and non-sarcomeric proteins. The South Asians, a population with large ethnic diversity, potentially carries region-specific polymorphisms. There is high variability in disease penetrance and phenotypic expression of variants associated with HCM. Thus, extensive studies are required to decipher pathogenicity and the physiological mechanisms of these variants, as well as the contribution of modifier genes and environmental factors to disease phenotypes. Conducting genotype-phenotype correlation studies will lead to improved understanding of HCM and, consequently, improved treatment options for this high-risk population. The objective of this review is to report the history of cardiovascular disease and HCM in South Asians, present previously published pathogenic variants, and introduce current efforts to study HCM using induced pluripotent stem cell-derived cardiomyocytes, next-generation sequencing, and gene editing technologies. The authors ultimately hope that this review will stimulate further research, drive novel discoveries, and contribute to the development of personalized medicine with the aim of expanding therapeutic strategies for HCM.

Keywords: hypertrophic cardiomyopathy, South Asians, β-myosin heavy chain, MYH7, cardiac myosin binding protein-C, MYPBC3

### THE CARDIOVASCULAR DISEASE DILEMMA IN SOUTH ASIAN AMERICANS

Although most cardiovascular disease (CVD) is preventable (Yusuf et al., 2004), it is the leading cause of death worldwide. The burden of CVD mortality is greatest in lower-income countries, yet is responsible for 1 in 4 deaths in the United States (Gupta et al., 2016). The South Asian subcontinent contains 20% of the world's population, yet accounts for about 60% of CVD worldwide. A map of this region, which includes the modern-day countries of Afghanistan, Bangladesh, Bhutan, India, Nepal, the Maldives, Pakistan, and Sri Lanka (Hajra et al., 2013), is shown in **Figure 1**. An estimated 1.8 billion people live in the South Asian region, and comprise one-fifth of the world's population. South Asians (SA) have a significantly increased risk of CVD when compared to their European counterparts (Anand et al., 2000; Yusuf et al., 2004; Fernando et al., 2015). Numerous studies have tried to account for this, assessing both traditional and non-traditional risk factors with inconsistent results (Chaturvedi, 2003; Palaniappan et al., 2010; Fernando et al., 2015). When considered together, risk factors alone do not account for the fact that SA immigrant populations are 3–5 times more likely to die of CVD than other ethnic groups (Gupta et al., 2006).

To date, the 2006 National Health Interview Survey is the only nationally representative data for obesity, CVD, and diabetes in Asian Americans (Barnes et al., 2008). However, the survey is limited in scope by self-reported collection measures, small sample size, and lack of long-term data (Mohanty et al., 2005;

FIGURE 1 | Map of South Asia. South Asia, sometimes referred to as the Indian subcontinent, includes the modern day countries of Afghanistan, Bangladesh, Bhutan, India, Nepal, the Maldives, Pakistan, and Sri Lanka. Myanmar, shown in the figure, is usually not included in South Asia except for population studies by the United Nations. Map modified from wikimedia commons South\_Asia\_(ed)update.PNG. An estimated 1.8 billion people live in this region, comprising one-fifth of the world's population and three-fifths of the global cardiovascular disease burden.

Narayan et al., 2010; Holland et al., 2011). Another glaring problem, especially when considering the existing interethnic comparative CVD studies, is the grouping of this extremely diverse population into the broad category of "Asian & Pacific Islander." Genetic, environmental, and behavioral differences exist among ethnic subpopulations of this group, and such differences contribute to differential health outcomes (Graham et al., 2006). More accurate epidemiology data and risk models could be generated if these groups were disaggregated (Mohanty et al., 2005; Narayan et al., 2010; Holland et al., 2011; Gopal and Usher-Smith, 2016).

Within the last 40 years, changes in immigration policies have paved the way for dramatic migration of SA into the United States. SA were the fastest growing ethnic group in the United States between the years 2000 and 2010 (Gezmu et al., 2014). Currently, there are an estimated 3.5 million SA living in America, meaning that they constitute around 1% of the American population (Tang et al., 2012). Since its inception in 2010, the Mediators of Atherosclerosis in South Asians Living in America (**MASALA**) Study has examined cardiometabolic risk and CVD outcomes in South Asian Americans on a longitudinal basis (Kanaya et al., 2013). This groundbreaking study will undoubtedly generate comprehensive, population-specific data for this at-risk population. A pilot study completed in 2015, called the South Asian Heart Lifestyle Intervention (**SAHELI**) Study, initiated efforts to deliver culturally sensitive educational material to underserved SA at risk for CVD. However, despite these recent efforts health disparities continue to exist for this American subpopulation (Fernando et al., 2015). The paucity of cardiac disease data for Asian Americans is a barrier to addressing one of the most concerning public health issues of our time.

### HYPERTROPHIC CARDIOMYOPATHY, A TREATABLE FORM OF CVD

### The Cardiomyopathies

Cardiomyopathies are a heterogeneous group of diseases of the myocardium associated with mechanical and/or electrical dysfunction. They usually exhibit inappropriate ventricular hypertrophy or dilatation and are due to a variety of causes that are commonly genetic (Elliott et al., 2008). Cardiomyopathies either are confined to the heart or are part of generalized systemic disorders, often leading to cardiovascular death or progressive heart failure-related disability (Maron et al., 2006). Cardiomyopathies are present in all populations, with ethnicity, age, and gender affecting disease severity and expression (McNally et al., 2015). The three main types of cardiomyopathy affecting the left ventricle are hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM), and restrictive cardiomyopathy (RCM) (Elliott et al., 2008). Prototypical cases of HCM show abnormally large and misaligned myocytes localized to the interventricular septum and increased fibrosis (Gersh et al., 2011). The thickened and stiff ventricle reduces the compliance of the heart muscle, decreases preload, and contributes to diastolic heart failure (Jacoby et al., 2013; Hensley et al., 2015). On the other end of the spectrum, typical DCM cases show chamber volume dilatation and thin walls, which reduces contractile force, and causes systolic heart failure. In RCM, which is the least common of the cardiomyopathies, patients typically have normal wall thickness (Elliott et al., 2008; Gersh et al., 2011). A diagnosis of RCM is made when left ventricular pressure is pathologically increased at normal or reduced chamber volumes due to excessive fibrotic accumulation. The structural differences between a normal heart and those affected by DCM, HCM, or RCM are illustrated in **Figure 2**. Although the typifying cases of cardiomyopathies are clearly differentiated, it should be noted that the clinical presentation of end stage cardiomyopathy can overlap significantly.

### Overview of Familial HCM

Familial HCM is the most common inherited cardiac disease and affects 1 in every 500 people worldwide (Maron et al., 2014). Since its first description in the 1950s, much progress has been made in elucidating the extremely heterogeneous genetic, morphogenic, and clinical profile of the disease (Elliott and McKenna, 2009). Symptoms of HCM are variable in severity and overlap with those of general CVD, including chest pain, shortness of breath, lightheadedness, palpitations, fatigue, and inability to perform vigorous exercise (Gersh et al., 2011). Another devastating manifestation of HCM is sudden cardiac death (SCD). Extensive left ventricular hypertrophy (LVH) in the absence of another cause can confirm a diagnosis of HCM (Brouwer et al., 2011). Non-parallel cell arrangement results in a characteristic "whirling" pattern and likely contributes to arrhythmogenicity in HCM patients. These histological changes are typically accompanied by increased collagen deposition and hyperplasia of microvasculature, pathologically narrowing the lumen and leading to symptoms related to myocardial ischemia (Shirani et al., 2000). HCM is typically classified as obstructive or non-obstructive depending on chamber anatomy, which determines the area through which blood can flow out of the heart. Obstructive HCM is normally associated with a worse prognosis due to the combined effects of systolic and diastolic dysfunction (Maron et al., 2003). Left ventricular outflow tract (LVOT) obstruction can result from a severely hypertrophied interventricular septum and/or abnormal mitral valve morphology (Nagueh and Mahmarian, 2006), as seen in **Figure 2**. The dynamic nature of LVOT obstruction aids in differentiation of HCM from other conditions in which LVH is the result of chronic and fixed pressure overload (Jacoby et al., 2013). LVOT obstruction, as well as other diagnostic hallmarks of HCM, can be detected via non-invasive imaging techniques such as echocardiography and cardiac MRI (Hensley et al., 2015) and in some cases provoked by altering loading conditions of the heart.

### Treatment of HCM

As is typical for many forms of CVD, many current therapeutic strategies for HCM try to alleviate symptoms and prevent complications. Although once considered rare and terminal, HCM has now emerged as a very treatable form of heart disease (Maron et al., 2014). Due to the variety of available surgical, pharmacological, electrical treatment options, HCM mortality rates have dropped to 0.5% per year (Maron et al., 2015). Beta-blockers and calcium channel blockers are used to improve diastolic function in patients with HCM (Brouwer et al., 2011; Hensley et al., 2015). Implantable cardiac defibrillators in conjunction with anticoagulants control potentially lifethreatening arrhythmias and avoid SCD. Two common surgical procedures performed in about 3% of obstructive HCM patients are septal myectomy and alcohol septal ablation (Roma-Rodrigues and Fernandes, 2014). Both techniques are very successful at relieving LVOT obstruction and heart failure symptoms and have excellent long-term prognoses (Gersh et al., 2011; Jacoby et al., 2013; Maron et al., 2014). Early diagnosis and treatment of HCM is key to preventing long-term chamber remodeling and reducing the number of patients that advance to end-stage heart failure.

### POPULATION GENETICS OF HCM IN SOUTH ASIANS

### Genetic Susceptibility of South Asians

Modern humans first evolved in Africa 200,000 years ago and dispersed about 100,000 years thereafter (Majumder, 2010). Although details of the initial waves of people leaving Africa have been debated, archeological and genetic evidence points to a single South Asian dispersal route (Kivisild et al., 2003; Mellars, 2006; Majumder, 2010). From there, people populated the rest of Eurasia and then the world. Thus, using European or East Asian populations, who are genetic descendants of South Asians, to predict risk for their genetic ancestors is, by definition, an inappropriate model. The strong cultural ties of SA in America, and the tradition of consanguineous marriage offer unique opportunities to study genetics and allele frequencies of this population (Waldmüller et al., 2003). In particular, India's history of founder events predicts a high rate of recessive disease (Reich et al., 2009) and could act as a model for studying how genetic variation affects disease expression. Prioritization of South Asian Americans for screening of such recessive disease is a largely unexplored opportunity for researchers and clinicians alike. This also requires that such studies be carried out specifically in SA genomes, and not in comparison with Caucasian or East Asian genomes that share very few of these region-specific variants.

### Genetic Heterogeneity of HCM

Recent advances in genetic sequencing techniques and the potential of therapeutic intervention in families with inherited cardiomyopathies have garnered this group of disorders much attention in the scientific community. Commercially available genetic screens are available to supplement a clinical diagnosis, and they have a mutation detection rate of 50–60% (Golbus et al., 2012). Genetic testing is recommended for at-risk family members who may also harbor a "private" familial variant (Gersh et al., 2011; Das et al., 2014). Once a variant is found, however, interpretation of its pathogenicity can be complicated. For example, distinguishing pathogenic variants from rare non-pathogenic variants or variants of unknown significance has proven difficult (Roma-Rodrigues and Fernandes, 2014). Currently, researchers have more variants of unknown significance than they can examine (Das et al., 2014). It would be impractical to study rare variants clinically, but recent

which is signified by darker shading.

large-scale sequencing projects have allowed statistical analysis of these variants (Lopes et al., 2013b). Typically, pathogenicity is estimated via in silico predictive algorithms that estimate a mutation's effect on protein structure and function and assess it in terms of evolutionary conservation (Ritchie and Flicek, 2014; Richards et al., 2015). While the use of multiple software programs to assess pathogenicity or a bioinformatics "pipeline" is recommended (Rehm et al., 2013), such programs still lack clinical data as input, making them insufficient for diagnostic purposes. Continued re-evaluation of all variants, even those previously considered benign, is necessary as predictive testing techniques continue to advance (Das et al., 2014; Richards et al., 2015). Family segregation studies and cellular/animal models are also options for studying pathogenicity of rare variants, but they are time-consuming and expensive.

### Necessity of Population-Specific Genetic Data for HCM

Apart from rare private familial mutations, ethnic "background variations" also affect penetrance and expressivity of genetic variants, further complicating genotype-phenotype predictions (Pan et al., 2012; McNally et al., 2015). Population-specific data have been reported in online databases of sequence variants, including those found in the NCBI, 1000 Genomes Project, and Exome Aggregation Consortium (ExAC). Allele frequencies of sarcomeric genes encoding β-myosin heavy chain, myosin binding protein C, and titin have been shown to differ among distinct ethnic groups (Golbus et al., 2012). Importantly, this study also reported that pathogenic variation in these genes was significantly higher than expected in the 1000 Genomes Database. This underscores the need for more comprehensive sequencing and screening of at-risk populations.

Population genetics studies on SA have been scarce, despite a well-documented increased risk of heart disease (Anand et al., 2000; Yusuf et al., 2004; Fernando et al., 2015). Southeast Asian genetic diversity is significantly underrepresented in the 1000 Genomes Project (Lu and Xu, 2013). Chambers et al. provided the first comprehensive genetic study of SA, performing wholegenome sequencing of 168 subjects (Chambers et al., 2014). Prior to the publication of this study in 2014, a mere two genomes had been used as references for the entire South Asian population, which complicated predictions of disease susceptibility (Kitzman et al., 2011; Gupta et al., 2012). In a recent study, the Exome Aggregation Consortium cited the underrepresentation of certain populations in previous genomic databases, notably Latinos and SA (Lek et al., 2016). Certain South Asian specific polymorphisms have been studied, but are of limited use due to small sample sizes that have very little statistical power (Dodani et al., 2012; Yadav et al., 2013). Extraction of meaningful information from raw datasets in online databases and association with clinical presentation is undoubtedly a necessary next step in the development of personalized medicine (Lek et al., 2016). It seems fitting that more genetic studies should be geared toward SA in order to study inherited treatable diseases like HCM. GenomeAsia 100k is a non-profit consortium aiming to generate Asian-specific genomic data. Importantly, the initial phase includes sequencing 10,000 reference genomes for all major ethnic groups within the Asian umbrella. As the name suggests, their ultimate goal is to generate genetic, microbiomic, clinical, and phenotypic data for 100,000 Asian genomes. If successfully completed, this genetic data will hugely contribute to future research and clinical efforts.

## MOLECULAR GENETICS OF HCM

HCM is inherited in an autosomal dominant manner, and most of the 1400+ variants associated with HCM encode sarcomeric proteins (Brouwer et al., 2011; Schlossarek et al., 2011). Nine sarcomeric genes carry the majority of HCM-related mutations and encode the proteins: β-myosin heavy chain (MYH7), cardiac myosin binding protein C (MYBPC3), cardiac troponin T (TNNT2), cardiac troponin I (TNNI3), α-tropomyosin (TPM1), regulatory myosin light chain (MYL2), essential myosin light chain (MYL3), cardiac α-actin (ACTC), and cardiac troponin C (TNNC1). The arrangement and interaction of proteins in the cardiac sarcomere is illustrated in **Figure 3**. Another group of mutations can be found in the genes encoding sarcomeric Z-disc proteins such as muscle LIM protein, αactinin, or telethonin (Knöll et al., 2010), but these are outside the scope of this review. The number of studies focusing on HCMassociated mutations in SA is disproportionately small when considering the size of this population and collective increased risk of CVD. To date, only 21 of the published variants associated with HCM have been documented in SA. These variants are listed in **Table 1**. In following with the Pareto principle (also known as the 80–20 rule), the majority of adverse HCM events comes from only a few central causes. In the case of HCM, over 80% of known HCM-associated mutations occur in MYBPC3 and MYH7 alone, while an additional 10% come from TNNT2 and TNNI3 (McNally et al., 2015). Thus, at least 90% of known HCM cases originate from four sarcomeric genes. The involvement of these four genes in the development of HCM is discussed below.

### THICK FILAMENT MUTATIONS ASSOCIATED WITH HCM

The thick filament of the sarcomere consists of bundles of myosin molecules, which form a cylindrical backbone as seen in **Figure 3**. The bipolar head domains project radially to interact with the thin filament during contraction (Spudich, 2014). Although attractive, the idea that myosin mutations causing hypercontractility lead to HCM and those causing hypocontractility lead to DCM has not been definitively established (Spudich, 2014). However, changes in calcium sensitivity (higher in HCM, decreased in DCM) have been discussed, and a direct link between calcium sensitivity, which induces heart growth via tension generation, and HCM has recently been established (Davis et al., 2016).

### MYH7 Mutations

MYH7 encodes cardiac β-myosin heavy chain (β-MHC), which is the major part of the thick filament of the sarcomere. The intrinsic ATPase activity of the catalytic domains of β-MHC provides the energy for sarcomeric contraction. Missense mutations in the globular head domain of β-MHC tend to produce HCM because they prevent interaction with actin, which is necessary for proper sarcomeric contraction (Volkmann et al., 2007). The lever-like myosin converter domain and globular head region show significant genetic constraint, and mutations in these regions are associated with severe forms of HCM (Homburger et al., 2016). The combination of normal and mutant or "poison peptide" causes detrimental structural and functional effects in the sarcomere (Brouwer et al., 2011). To date, 289 mutations have been found in MYH7 that are thought to produce HCM, and this gene comprises 40% of the genetic profile of the disease (Morimoto, 2008; Stenson et al., 2014). Only eight of these variants have been published in SA. Mutations in MYH7 are associated with early-onset and extensive LVH (Roma-Rodrigues and Fernandes, 2014), and clinically are associated with an increased risk of atrial fibrillation, SCD and heart failure (Wang et al., 2008; Lopes et al., 2013a). In the Indian population, it has been predicted that the frequency of MYH7 mutations may be lower than their MYBPC3 counterpart, although validation in a large sample is required for confirmation (Morimoto, 2008).

Due to its clinical severity, the MYH7-R403Q variant was the first HCM-causing mutation to be discovered (Geisterfer-Lowrance et al., 1990; Nag et al., 2015). This mutation, which affects the "cardiomyopathy loop" at the interface of actin and myosin (Marian, 2002; Volkmann et al., 2007) is associated with severe LVH and nearly 100% penetrance of SCD at a young age (Lopes et al., 2013a). Studies on this variant have been particularly helpful in studying both loss-of-function and gainof-function mutations associated with HCM. Loss-of-function mutations result in reduced contractile force (Nag et al., 2015), while gain-of-function mutations generate supraphysiological force (Seidman and Seidman, 2001; Moore et al., 2012). The diverse manifestations of the same mutation on sarcomere biomechanics emphasizes the need to fully characterize the signaling pathways involved in disease development and develop specialized therapies.

### MYBPC3 Mutations

MYBPC3 encodes the cardiac isoform of myosin binding protein C (cMyBP-C). cMyBP-C is a thick filament-associated protein that acts as a tether for the myosin head domain and also associates with actin and titin. cMyBP-C is found in the C-zone of the sarcomere and has both structural and regulatory roles in sarcomere assembly. Mutations in MYBPC3 typically take longer to manifest than those in MYH7, presenting in middle age or later and after the typical reproductive age (Wang et al., 2008). To date, around 346 MYPBC3 mutations have been found that are associated with HCM, and comprise at least 40% of the genetic profile of the disease (Morimoto, 2008; Stenson et al., 2014). Of these 346, a mere 5 have been published in South Asian subjects. As opposed to MYH7 mutations, the majority of those affecting MYBPC3 result in abnormal truncation of the C-terminus of cMyBP-C (Kuster and Sadayappan, 2014). The degradation of these aberrant expression products is thought to contribute to HCM via haploinsufficiency, i.e., inadequate amounts of the normal protein product (Brouwer et al., 2011).

Perhaps the most well-cited example of a region-specific polymorphism in SA is a 25-bp deletion in MYBPC3 (MYBPC3∆Int32) (Waldmüller et al., 2003; Dhandapany et al., 2009). The genetic heterogeneity of HCM, background genetic noise, and cost of sequencing make screening impractical for the general population (Kapplinger et al., 2014). However, the prevalence of the MYBPC3∆Int<sup>32</sup> variant is at least 4% in SA but virtually non-existent in Caucasians. The exclusivity of this allele to South Asian descendants makes genetic screening effective and practical for this population (Kuster and Sadayappan, 2014). The 25-bp deletion from intron 32 of MYBPC3 causes a reading

(TnI), troponin C (TnC), tropomyosin, and actin comprise the thin filament of the sarcomere. Titin is a giant protein that connects the thick filament to the edge of the sarcomere and creates passive tension in resting muscle.

frameshift in translation, and the aberrant protein product cMyBP-CC10mut lacks exon 33 and includes abnormal parts of exon 34 and part of the 3′ untranslated region (Waldmüller et al., 2003; Dhandapany et al., 2009). The altered C10 domain cannot normally link cMyBP-C to the myosin heavy chain and leads to contractile dysfunction (Kuster et al., 2015). Double heterozygotes, or patients carrying both the MYBPC3∆Int<sup>32</sup> mutant allele and another mutant allele in MYH7-E927del, seem to have a more severe HCM phenotype suggesting an additive effect of these mutations (Waldmüller et al., 2003). Systematic studies are currently underway at various laboratories to determine if the MYBPC3∆Int<sup>32</sup> variant is sufficient to cause HCM.

### THIN FILAMENT MUTATIONS ASSOCIATED WITH HCM

α-tropomyosin, along with the troponins T, I, and C form the troponin-tropomyosin complex that regulates contraction in striated muscle. As seen in **Figure 3**, this complex is interdigitated with actin and winds together to form the thin filament of the sarcomere. This regulatory complex prevents strong association between the thick and thin filament in low Ca2<sup>+</sup> and ATP conditions (McNally et al., 2015). Thin filament mutations typically increase Ca2<sup>+</sup> sensitivity of tension development (Ashrafian et al., 2011) or uncouple Ca2<sup>+</sup> sensitivity from phosphorylation (Papadaki et al., 2015).

### TNNT2 Mutations

The connection between TNNT2, the gene encoding the cardiac isoform of troponin T (TnT), and HCM was first reported in 1994 (Thierfelder et al., 1994). TNNT2 variants account for about 5% of HCM cases (Morimoto, 2008). TnT mutants have been linked to increased risk of SCD at a young age (Marian, 2002; Lopes et al., 2013a) but typically patients carrying these variants exhibit little to no hypertrophy (Chandra et al., 2005). TnT mutants are thought to alter cross-bridge kinetics, thereby limiting shortening velocity at maximal Ca2<sup>+</sup> activation and increasing the energetic cost of contraction (Sweeney et al., 1998). In mice, degree of Ca2<sup>+</sup> sensitization from mutations in TnT were found to directly correspond to arrhythmic risk and were reduced with the Ca2<sup>+</sup> desensitizer Blebbistatin (Baudenbacher et al., 2008). To date, 43 HCM-related mutations in TNNT2 have been documented in the general population (Morimoto, 2008; Stenson et al., 2014). In SA, 3 TNNT2 variants have been published that are associated with HCM, as seen in **Table 1**.

### TNNI3 Mutations

TNNI3, which encodes troponin I, was first linked to HCM in 1997 (Kimura et al., 1997). TNNI3 mutations make up about 5% of the genetic profile of HCM (Morimoto, 2008). 38 HCMrelated mutations in TNNI3 are known, and 4 of these have been documented in SA (Stenson et al., 2014) as can be seen in **Table 1**. Recently, mutations in the highly conserved inhibitory peptide region of troponin I were found to alter myofilament Ca2<sup>+</sup> sensitivity (Westfall et al., 2002). Mutations in this important


PubMed published variants have been listed with Human Genome Variation Society (HGVS) nomenclature (noted as NM\_XXXX), Single Nucleotide Polymorphism Database (dbSNP) Reference SNP cluster ID (noted as rsXXXX), and ClinVar pathogenicity status where possible. For published variants without a ClinVar reference, N/A is listed under pathogenicity status. Intronic variants and those classified as benign or likely benign have not been included in the table. None of these mutations were reported to be coexistent with other mutations that could confound pathogenicity status of the listed variant.

regulatory region are thought to produce increased tension at physiological Ca2<sup>+</sup> levels and prevent relaxation.

### NON-SARCOMERIC MUTATIONS ASSOCIATED WITH HCM

Non-sarcomeric genes are also involved in HCM, and some researchers have proposed that the disease should be reclassified into sarcomeric and non-sarcomeric because pathophysiology and prognosis differ significantly between these groupings (Olivotto et al., 2011; McNally et al., 2015). The variable penetrance and expressivity in HCM means that the same mutation may cause severe disease in one person, but a completely normal phenotype in another (Tanjore et al., 2010). Genotypically positive but phenotypically negative patients have perplexed researchers, and suggest a role of modifier genes in clinical outcomes (García-Honrubia et al., 2016).

This review will focus on a well-documented angiotensin-1 converting enzyme (ACE) polymorphism in SA that contributes to the HCM phenotype in a dose-dependent manner (Rai et al., 2008). The insertion allele (I) corresponds to the presence of a 287-bp transposable Alu element repeat, and the deletion allele (D) corresponds to the absence of this sequence from the ACE gene (Marian, 2002). LVH was most severe in patients with the DD genotype, which corresponds to higher plasma and tissue ACE levels. LVH was less pronounced in heterozygous ID individuals and least severe in II individuals, who were found to have less circulating ACE (Marian, 2002). Long-term activation of components in the renin-angiotensinaldosterone system such as ACE is thought to lead to vascular remodeling and hypertrophy seen in HCM. However, significant inconsistencies have been reported regarding ACE I/D polymorphisms in HCM patients (Yang et al., 2013). To avoid limitations due to genetic heterogeneity and small sample size, large-scale studies or of ACE I/D polymorphisms in genetically distinct groups are necessary (Kolder et al., 2012; Luo et al., 2013).

### SECONDARY HCM CAN BE A RESULT OF EXTRACARDIAC DISEASE

Patients with extracardiac disease can also present with LVH similar to that seen in HCM, and these are listed in **Table 2**. Many of these diseases present with other obvious systemic findings, allowing early differential diagnosis (Roma-Rodrigues and Fernandes, 2014). A notable exception is milder forms of Fabry disease, which has often been inaccurately diagnosed as HCM with the same prevalence as some sarcomeric mutations (Sachdev et al., 2002). This is significant because Fabry disease is treatable with enzyme replacement therapy, while the chamber remodeling associated with HCM is currently irreversible. For this reason, mutations in the genes causing Fabry disease and Danon disease are often included in genetic test panels for HCM. Of these systemic diseases, only 6 have been documented in SA, as shown in **Table 2**. This further underscores the need to note subjects' ancestry when performing genetic testing and the relevance of this information to the scientific community. The number of mitochondrial disorders causing phenocopies of HCM is too extensive for this review and have not been included in **Table 2**.

### MOLECULAR MECHANISMS UNDERLYING THE DEVELOPMENT OF HCM

On the sarcomeric level, several mechanisms are thought to link the genotype to the clinical phenotype of HCM. Increased myofilament Ca2<sup>+</sup> sensitivity is believed to be proarrhythmogenic, although the exact molecular mechanism(s) involved remain(s) unclear. Altered Ca2<sup>+</sup> handling, inefficient energy utilization, and increased mechanical stretch may play a role in the development of anatomical and functional changes observed in hearts with HCM (Huke and Knollmann, 2010; Ashrafian et al., 2011). Stabilization and prolongation of the Ca2<sup>+</sup>

transient would pathologically activate intracellular signaling pathways and diminish relaxation, contributing to electrical conduction abnormalities and diastolic dysfunction (Roma-Rodrigues and Fernandes, 2014). A predicted consequence of altered Ca <sup>2</sup><sup>+</sup> sensitivity is inefficient energy usage, including increased cross-bridge turnover, increased ATPase activity, and mitochondrial energetic abnormalities causing oxidative stress (Ashrafian et al., 2011; Brouwer et al., 2011). Energy and calcium are intrinsically linked to muscle mechanics due to excitation-contraction coupling; thus, force production of myofilaments will be abnormally increased as a result of Ca2<sup>+</sup> sensitization. Therapeutic strategies are currently being explored which reinstate normal Ca2<sup>+</sup> homeostasis, lower Ca2<sup>+</sup> sensitivity, and increase efficiency of energy utilization (Brouwer et al., 2011). A recent breakthrough showed that increased Ca2<sup>+</sup> sensitivity of troponin C causes prolonged myofilament tension, concentric sarcomeric addition, and hypertrophy consistent with HCM (Davis et al., 2016). DCM was linked to decreased Ca2<sup>+</sup> sensitivity in the same study.

### CURRENT EFFORTS TOWARD PRECISION MEDICINE FOR HCM

The current technologies outlined in this section are still in their infancy, and require much optimization in order to be developed into clinical therapies. However, the existence of HCMassociated variants unique to SA represents a largely unexplored opportunity to develop molecular models for cardiac disease. The vast size of this population makes application of these models to clinical treatments a logical and potentially lucrative next step.

### Reprogramming of Cardiomyocytes

Adult somatic cells can be reprogrammed using sets of transcription factors to generate human induced pluripotent stem cells (iPSCs) and then functional cardiomyocytes (Kamdar et al., 2015), offering researchers the opportunity to study the molecular mechanisms of CVD and myocardial tissue regeneration following ischemic injury. iPSC-derived cardiomyocytes have been successfully used to study systemic diseases that present with HCM, including Pompe disease (Huang et al., 2011) and LEOPARD syndrome (Carvajal-Vergara et al., 2010).

The pathogenic effects of MYH7 and MYBPC3 mutations have been demonstrated using iPSC-derived cardiomyocytes (Ross et al., 2016). Lan and colleagues administered calcium blockers to iPSC-derived cardiomyocytes from HCM patients harboring the MYH7-R663H mutation (Lan et al., 2013). These drugs seemed to prevent structural and functional abnormalities typical of HCM, suggesting that their use in prophylaxis warrants investigation. Han and colleagues similarly examined the MYH7- R442G mutation (Han et al., 2014). They performed wholetranscriptome sequencing and documented an upregulation in gene expression in various cell proliferation signaling pathways including Wnt/β-catenin, Calcineurin-NFAT, and FGF that



Secondary HCM refers to hypertrophic cardiomyopathy that originates from another disease, the origin of which is outside of the heart. References in bold typeface correspond to PubMed published cases documented in South Asians.

are thought to contribute to the disease phenotype. Tanaka and colleagues investigated how Endothelin-1, a circulating vasoconstrictive peptide, induced the HCM phenotype in iPSCderived cardiomyocytes containing the MYBPC3-G999-Q1004del mutation (Tanaka et al., 2014). Lan et al. specified their proband's ethnicity as African-American, but the two studies assessing MYH7 variants lack ethnicity information. In vivo direct cardiac reprogramming of somatic cells into cardiomyocytes is a potential offshoot of current reprogramming techniques but has not yet been tested in humans (Sadahiro et al., 2015). For HCM in particular, the possibility of converting cardiac fibroblasts into functional cardiomyocytes could theoretically ameliorate hypertrophy and improve diastolic function. As reprogramming technology advances, these techniques could offer a renewable source of cardiomyocytes and deliver medicine individually tailored to each patient (Chen and Qian, 2015; Tanaka et al., 2015).

### Variant Databases

Next generation sequencing (NGS) protocols for screening cardiomyopathies have been successfully developed (Gómez et al., 2014), and they will continue to uncover new variants. Improved methods of distinguishing between benign and pathogenic mutations will enhance the accuracy of risk prediction of genetic screens for HCM (Kapplinger et al., 2014). Further, combination of gene expression analysis and clinical phenotypic assessments are necessary to uncover unknown modifier genes in HCM patients (Han et al., 2014). The development of online databases for HCM variants could aid physicians in the timing and implementation of electrical or surgical treatment options (McNally and Puckelwartz, 2015). As mentioned in the preceding sections, inclusion of ethnicity data in such databases has important genetic consequences and thus is clinically relevant. The future of personalized medicine will rely heavily on such collaborative efforts.

### Gene Editing Technology

Of the cardiovascular diseases, primary inherited cardiomyopathies such as HCM are perhaps the strongest candidates for gene editing technologies (Strong and Musunuru, 2016). Directed nucleases such as Transcription Activator-Like Effector Nucleases, Zinc Finger Nucleases (TALENs), and Clustered, Regularly Interspaced, Short Palindromic Repeats (CRISPR) and CRISPR-associated 9 (Cas9) systems allow for specific editing of individual gene mutations. This CRISPR/Cas9 system is the current frontrunner of these gene modification technologies and has the potential to facilitate causation studies, by both creating and correcting a specific mutation and documenting the associated phenotype (Han et al., 2014). This technology promises to provide researchers with more accurate model for studying cardiomyopathies, circumventing the inherent ethical issues when studying human subjects (Waddington et al., 2016). To date, the CRISPR/Cas9 system has been used to successfully engineer cardiomyopathy into in zebrafish and mice models, and is currently being applied to larger animals such as pigs and non-human primates (Duncker et al., 2015).

The CRISPR/Cas9 system has been used to edit the genomes of mice with Duchenne muscular dystrophy, an X-linked genetic disorder that presents with cardiomyopathy. The normal muscle phenotype in mice embryos was rescued following correction of the mutant dystrophin (Dmd) gene (Long et al., 2014). Considerable shortcomings of this technology include the variable correction efficiency and the introduction of mutagenesis at an incorrect but sequentially similar location (Strong and Musunuru, 2016). Currently there are no established techniques that can identify off-target mutagenesis. However, despite these concerns, CRISPR/Cas9-mediated genome editing is an exciting frontier in the genetic realm. In theory, it could eliminate inherited genetic disorders such as the cardiomyopathies altogether.

### Small-Molecule Inhibitors

The recently developed small myosin ATPase inhibitor MYK-461 was found to reduce sarcomeric contractility, LVH, myocyte disarray, and fibrosis in mice with HCM (Green et al., 2016). Although MYK-461 has huge potential as a therapeutic agent, it is not effective at reversing myocyte disarray and fibrosis once substantial hypertrophy has developed (Green et al., 2016). This underscores the need for expanded genetic screening in at-risk populations in order to administer earliest such preventative treatments. Pharmacologic treatments typically relieve heart failure symptoms without altering disease progression (Green et al., 2016). However, in a recent study, the calcium channel blocker Diltiazem used to treat hypertension and chest pain was shown to prevent reduction in left ventricular chamber size in HCM patients carrying MYBPC3 mutations (Ho et al., 2015). This observed attenuation of HCM development is promising, and the drug could be used to treat at-risk family members before they notice the functional heart failure symptoms associated reduced cardiac output. Continued refinement of small-molecule inhibitors to arrest or even reverse hypertrophy, myocyte disarray, and fibrosis will likely aid in untangling the complex cellular mechanisms involved HCM development (Frey et al., 2012).

## CONCLUSIONS

The failure of risk factors to explain the increased prevalence of CVD in SA suggests a strong genetic causation that needs active characterization. For HCM in particular, it is imperative that SA receive higher priority for systematic genome-wide studies. Such studies will uncover population specific variants contributing to susceptibility, and could lead to better diagnosis and management of HCM. By taking advantage of novel technologies such as NGS, human iPSC derived cardiomyocytes, and gene editing nucleases, it is hoped that we will be able to significantly improve the lives of patients affected by genetic heart failure. It is essential that novel genomics technologies are used to reduce health disparities among marginalized and minority groups rather than perpetuate them (Rotimi and Jorde, 2010). Efforts such as those led by GenomeAsia100k will provide the necessary genetic data for future discovery.

Genetic variants, molecular mechanisms, and clinical phenotypes of HCM vary on a patient-by-patient basis. The Precision Medicine Initiative announced by President Obama in 2015 aims to personalize the current "one-size-fits-all" model of healthcare from the level of diagnosis all the way to delivery. In this article, we underscore the relevance of ethnicity data to genetic studies in particular, and stress that future studies specifically address the very large yet largely neglected South Asian population. The Sarcomeric Human Cardiomyopathy Registry (**SHaRe**) is an exciting interdisciplinary model that brings together geneticists and cardiologists to share and develop longitudinal, personalized data for patients with HCM and DCM. Although these efforts have mostly been undertaken in developed countries (Fox, 2015), this review has emphasized the need to address highly populated and underserved regions such as South Asia. Considering the genetic and clinical heterogeneity of the disease and its potential treatability, established cardiovascular centers of excellence provide the best possible treatment for HCM patients (Gersh et al., 2011; Maron et al., 2014). The Amrita Institute of Medical Sciences and Research in India has undertaken the task of becoming such a center, and if successful is expected to have dramatic impacts on HCM healthcare delivery in the South Asian subcontinent (Maron, 2015).

## AUTHOR CONTRIBUTIONS

JK, SV, RK, and SS designed and coordinated the proposed review article, analyzed literature data, and wrote the article. All authors discussed the work and commented on the manuscript.

## ACKNOWLEDGMENTS

SS was supported by National Institutes of Health grants R01HL130356, R01HL105826 and K02HL114749 and American Heart Association, Cardiovascular Genome-Phenome Study (15CVGPSD27020012). RK is supported by the Fondation LeDucq and Deutsche Forschungsgemeinschaft (DFG).

### REFERENCES


MYBPC3 mutation common in populations of South Asian descent causes contractile dysfunction. J. Biol. Chem. 290, 5855–5867. doi: 10.1074/jbc.M114.6 07911


five South African families from various ethnic origins. Am. J. Hum. Genet. 44, 787–793.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Kraker, Viswanathan, Knöll and Sadayappan. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Importance of Intrinsically Disordered Segments of Cardiac Troponin in Modulating Function by Phosphorylation and Disease-Causing Mutations

Maria Papadaki <sup>1</sup> \* and Steven B. Marston<sup>2</sup>

*<sup>1</sup> Department of Cell and Molecular Physiology, Loyola University of Chicago, Maywood, IL, USA, <sup>2</sup> Myocardial Function, National Heart and Lung Institute, Imperial College London, London, UK*

Troponin plays a central role in regulation of muscle contraction. It is the Ca2<sup>+</sup> switch of striated muscles including the heart and in the cardiac muscle it is physiologically modulated by PKA-dependent phosphorylation at Ser22 and 23. Many cardiomyopathy-related mutations affect Ca2<sup>+</sup> regulation and/or disrupt the relationship between Ca2<sup>+</sup> binding and phosphorylation. Unlike the mechanism of heart activation, the modulation of Ca2+-sensitivity by phosphorylation of the cardiac specific N-terminal segment of TnI (1–30) is structurally subtle and has proven hard to investigate. The crystal structure of cardiac troponin describes only the relatively stable core of the molecule and the crucial mobile parts of the molecule are missing including TnI C-terminal region, TnI (1–30), TnI (134–149) ("inhibitory" peptide) and the C-terminal 28 amino acids of TnT that are intrinsically disordered. Recent studies have been performed to answer this matter by building structural models of cardiac troponin in phosphorylated and dephosphorylated states based on peptide NMR studies. Now these have been updated by more recent concepts derived from molecular dynamic simulations treating troponin as a dynamic structure. The emerging model confirms the stable core structure of troponin and the mobile structure of the intrinsically disordered segments. We will discuss how we can describe these segments in terms of dynamic transitions between a small number of states, with the probability distributions being altered by phosphorylation and by HCM or DCM-related mutations that can explain how Ca2+-sensitivity is modulated by phosphorylation and the effects of mutations.

Keywords: cardiac troponin, molecular dynamics, intrinsic disorder, phosphorylation, cardiomyopathy

### INTRODUCTION

Since its discovery by Ebashi and Kodama (1965), troponin has been a molecule that never ceases to be studied, due to its central role in muscle contraction and regulation. In heart muscle, troponin has a dual function, both to switch contraction on and off in response to Ca2<sup>+</sup> and to modulate Ca2+-sensitivity and the rate of relaxation in response to adrenaline, a process that is disrupted in cardiomyopathy (Solaro et al., 2008; Messer and Marston, 2014). To properly

#### Edited by:

*Jose Renato Pinto, Florida State University, USA*

#### Reviewed by:

*Tharin Blumenschein, University of East Anglia, UK Natosha Finley, Miami University, USA*

> \*Correspondence: *Maria Papadaki mpapadaki@luc.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *01 July 2016* Accepted: *17 October 2016* Published: *02 November 2016*

#### Citation:

*Papadaki M and Marston SB (2016) The Importance of Intrinsically Disordered Segments of Cardiac Troponin in Modulating Function by Phosphorylation and Disease-Causing Mutations. Front. Physiol. 7:508. doi: 10.3389/fphys.2016.00508*

**173**

understand these functions of troponin and the mechanisms behind them, it is important to establish structure-function relationships. Recently, cardiac troponin has been recognized as a "dynamic" molecule, containing intrinsically disordered regions (Colson et al., 2012; Hwang et al., 2014). We are particularly interested in these regions, their relationship to troponin's function and how they change due to phosphorylation and mutations. Ultimately, this information can be used for the treatment of cardiomyopathy, by designing drugs that will target these dynamic transitions.

### THE REGULATION OF CONTRACTION BY TROPONIN

The early studies on troponin used biochemical and biophysical methods to measure troponin's activity switching and identify protein-protein interactions involved. It soon became apparent that troponin-regulated thin filaments are a very complex system. The thin filament is a multiprotein complex with a repeating unit comprised of 24 proteins [14 actin, 2 tropomyosin dimers, 2 troponin I (TnI), 2 troponin T (TnT), and 2 troponin C (TnC)]. The troponin molecule complex is shown in **Figure 1A**. The thin filament can contain up to 30 such units in a system that allows Ca2<sup>+</sup> binding to TnC to control myosin binding to actin, allowing contraction (Gordon et al., 2000). As expected for a switch, the whole system is dynamic and the properties of the switch are modified by cellular signaling systems, notably phosphorylation of TnI by PKA. Whilst thin filaments can be described in terms of the protein interactions involved (Farah and Reinach, 1995), or the transitions between states (Maytum et al., 1999), a complete structure of the thin filament remains elusive.

The structure of the actin filament has been solved to high resolution (Oda et al., 2009), as well as tropomyosin by X-ray diffraction (Brown et al., 2005) and the complex of tropomyosin on actin has been accurately deduced from electron micrographs (Moore et al., 2016). Troponin was the last component to be solved by crystallography, reflecting both its heterotrimeric nature and its inherent flexibility (Takeda et al., 2003; Vinogradova et al., 2005). The troponin crystal structure is incomplete, however, and does not include some of the most functionally interesting parts of the complex.

These structural studies suffer from a critical defect: they only provide structures that are frozen in a particular state that may or may not correspond to one of the functional states defined by biochemistry, yet troponin is a highly dynamic molecule that easily flips between activity states. Current structures do not resolve several segments of troponin that are potentially involved in regulation including the inhibitory (138–147) and switch (148–154) regions of TnI as well as the C-terminal mobile domain (164–210) and the C-terminal 18 amino acids of TnT. Thus, for understanding troponin function we need to define the structure of the whole molecule and take into account its dynamic nature. Recent concepts describe that order-disorder transitions in the dynamic regions of troponin are important for muscle contraction regulation (Metskas and Rhoades, 2016).

FIGURE 1 | (A) Ribbon diagram of troponin and its different subunits. Blue represents TnC (1–161), red represents TnI (1–171) and green represents TnT (212–298). Catalytic and structural Ca2<sup>+</sup> are shown as gray spheres, and the inhibitory and the switch peptide of TnI are indicated. (B) Snapshot of troponin MD simulation, showing troponin surface rendering in a solvent box. Blue represents TnC (1–161), red represents TnI (1–171) and green represents TnT (212–298). The image was adapted from Zamora et al. (2016a) with permission from the PCCP Owner Societies.

The C-terminal mobile peptide of skeletal TnI was the first one to be identified as a dynamic peptide on troponin and its dynamic structure gives kinetic activity advantage for binding to actin (Blumenschein et al., 2006; Hoffman et al., 2006). The dynamic segments of cardiac troponin were explicitly defined by Tobacman using isotope exchange methods (Kowlessur and Tobacman, 2012), whereas fluorescent anisotropy experiments on cardiac TnI provided experimental evidence on the existence of these dynamic domains (Zhou et al., 2012). The dynamic regions of TnI vary between troponin isoforms, with skeletal troponin having a higher level of disorder. In addition, cardiac TnI contains a disordered region absent in the skeletal isoform, the N-terminal extension containing the phosphorylatable Ser22 and 23 (Hoffman and Sykes, 2008). Phosphorylation at Ser22 and 23 causes a 2-3-fold change in Ca2+-sensitivity, which has a major physiological impact on cardiac function.

### RECENT STUDIES OF CARDIAC TROPONIN REGULATION

Baryshnikova demonstrated that phosphorylation of TnI Ser22 and 23 modulates the overall affinity of the N-terminal lobe of TnC to TnI (147–163) rather than the Ca2+-affinity (Baryshnikova et al., 2008). A study of the interaction of the native and phosphorylated TnI (1–30) in isolation led to a model for the differential docking of this peptide on TnC in native and phosphorylated states (Howarth et al., 2007), but more recently the question was addressed in a model-free study using whole TnC and TnI (1–73) (Hwang et al., 2014). TnI (1–37) segment, described as intrinsically disordered, interacts electrostatically with the N-terminal lobe of TnC, whilst TnI (41–67) forms a helix that interacts with a hydrophobic patch in the C-terminal TnC lobe, as found in the crystal structure. This study proposed that interactions of TnI (1–30) with the TnC N-terminal lobe stabilized the position of the N-terminal lobe relative to the rest of troponin and that this positioning indirectly affects Ca2<sup>+</sup> and TnI (148–158) affinity. Phosphorylation disrupts the interactions and the loss of positioning leads to changes in Ca2+-sensitivity. Disorder in proteins is currently studied using molecular dynamics (MD) simulations in combination with structural studies, such as NMR.

Molecular dynamics (MD) is a computational technique whereby proteins can be viewed as moving objects rather than stable peptides. Thus, intrinsically disordered peptides can be analyzed in terms of their dynamic, as well as their static properties. The idea behind MD is to calculate the energy of a system as a function of the position of its atoms. MD simulations can help us understand biological functions of proteins, such as conformational changes caused by interactions with other proteins or ligands. Additional properties can be measured, for example flexibility of proteins traditionally measured in nanosecond (ns) timescales, although biological events happen within microseconds (µs).

Recent technical developments have applied this computational method to cardiac troponin. A seminal MD study simulated Ca2<sup>+</sup> binding to the TnC N-terminal domain over a µs timescale, much longer that commonly used (Lindert et al., 2012). This simulation showed the dynamics of opening and closing of the hydrophobic patch upon Ca2<sup>+</sup> binding predicted by NMR studies, whilst also emphasizing the need for extended simulations to allow time for all possible conformations to be explored (Lindert et al., 2012). This highlights a central challenge of MD studies: the immense amount of time and computing power needed for an extended simulation must be balanced by the need to obtain results in a meaningful timescale. Moreover the protein investigated must be placed in a virtual solvent box big enough to avoid any contacts with the box sides. The larger the system studied the greater the problems become, however MD simulations of troponin (about 500,000 atoms including solvent) have now been carried out by two groups.

Molecular dynamics (MD) simulations of the cardiac troponin were performed in a study by Cheng et al. (2014). In that study, the troponin complex structure [TnI (1–172), TnC (1–161), TnT (236–285)] was modeled and the phosphomimetic mutations S23D/S24D were introduced to study the effect of phosphorylation and also the effect of disease-related mutations. The length of simulation was limited to 3 × 150 ns (Cheng et al., 2015). The more recent MD simulation study by Zamora et al provides an even more complete model of human cardiac troponin sequence (**Figure 1B**), including the C-terminal peptide of TnT not present in previous studies. Multiple 750 ns simulations, totaling over 10µs, were performed and the size of the box simulated has been increased to 25Å to avoid virtual self-association (Zamora et al., 2016a).

### INSIGHTS FOR TROPONIN REGULATORY MECHANISM

Recent studies have come to the conclusion that cardiac troponin contains intrinsically disordered regions, and one of them is the TnI (1–30), which contains the phosphorylatable Ser22 and 23. In fact, being disordered is significant for the phosphorylation itself and it has been found that the majority of proteins that get phosphorylated in nature have intrinsically disordered domains close to the phosphorylation site (Iakoucheva et al., 2004). Using MD simulations to study the effects of mutations or phosphorylation is therefore a suitable approach, because the disordered regions are included in the model. MD provides the opportunity to look at the overall stability of the protein by calculating the Root Mean Square Fluctuation (RMSF) and any interaction between the subunits can be measured.

In that respect, it is important to establish which interactions are relevant to troponin's function. Both Zamora's and Cheng's MD studies agree that the order-disorder transition associated with phosphorylation proposed by Hwang et al is unlikely since both states are disordered. Distances between Ca2<sup>+</sup> and coordinating residues within the Ca2<sup>+</sup> binding loop are almost always measured, because direct comparisons with biochemical measurements can be made. The MD studies of Cheng and Zamora indicate that measuring the distances between the TnC Ser69 gamma Oxygen (GO) and Ca2<sup>+</sup> would give information about the Ca2+-sensitivity change. Ser69 GO was found in two different conformations: at a high distance and a low distance. Cheng et al observed the low distance conformation only 10% of the time, while Zamora et al found the opposite; the Ser69 GO is located within the coordination sphere (∼3Å) most of the time but flips to a longer distance (∼6Å) outside the Ca2<sup>+</sup> coordination sphere occasionally, where it can stay for ∼5% of the time. Phosphorylation doubles the time spent at the long distance, which may explain the Ca2+-sensitivity change upon phosphorylation. The difference is ascribed by Zamora et al to the shorter simulation times in the Cheng et al study that have only sampled the longer distance state. In addition, it is thought that the small box size used may result in periodic artifacts mostly due to the L shaped configuration of the starting molecule and the hinge motion that elongates the molecule, increasing the probability of self-association. The study of Zamora et al used real phosphorylation in Ser22 and 23 whereas the study of Cheng et al used phosphomimetic mutations S23/S24D, although there are no structural or physiological differences between the effects of phosphorylation and phosphomimetic mutations in the myofilament (Finley et al., 1999; Mamidi et al., 2012; Rao et al., 2014). Using S23/24D allowed Cheng et al to make direct comparisons between MD simulations and physiological data.

The contacts between the intrinsically disordered segments and the troponin core are interesting since they are involved in the mechanism of regulation by troponin (Cheng and Regnier, 2016). In order for the opening of the hydrophobic cleft upon Ca2<sup>+</sup> binding to occur, TnC must interact with the TnI switch peptide (148–154). The N-terminal TnC-TnI (1–30) interaction is also crucial for the phosphorylation signal to modulate Ca2+ sensitivity. Finally, the disordered C-terminus of TnT (212–298), which has not previously been simulated, maybe involved in these interactions.

In their MD simulations, Cheng et al observed that phosphorylation decreased the overall stability of troponin as determined by RMSF plots and changed the interactions between TnC and TnI. More specifically, they found increased intrasubunit interactions between the N-terminal and the inhibitory TnI peptide, which are not present in the unphosphorylated structure. In contradiction to this study, the study by Zamora et al showed that phosphorylation does not affect overall stability of any part of the troponin complex and that phosphorylation changes the interactions between the intrinsically disordered segments and TnC. Moreover, the C-terminus of TnT has proven to make a significant difference to the analysis since it interacts with both TnC N-terminal domain and the TnI N-terminal peptide.

In essence, Zamora's analysis indicates that phosphorylation never induces new interactions between the subunits but results in a number of subtle changes in the dynamics of existing interactions. A recent study supports this finding, as the TnI mutation R145W causing restrictive cardiomyopathy reduces the interaction frequencies between TnC and TnI leading to blunting of the adrenergic response (Dvornikov et al., 2016). However, contradictory results using peptides found that major conformational changes occur upon phosphorylation (Heller et al., 2003; Howarth et al., 2007). The fact that phosphorylation causes only subtle structural changes is in accordance with the fact that the Ca2+-sensitivity change upon phosphorylation is only 2-fold. It is interesting that despite these subtle changes, the physiological effect of phosphorylation is great.

Another exciting point arising from all these studies on troponin is its potential to determine how mutations affect troponin's structure, Ca2<sup>+</sup> binding and also the effect of phosphorylation. Over 100 troponin mutations in cardiac TnT, TnC, or TnI have been linked so far with genetic cardiomyopathies. Different troponin mutations have different effects on protein-protein interactions, crossbridge cycle, myosin ATPase activity and phosphorylation levels of thin filament proteins, but all mutations affect troponin Ca2+-sensitivity (Marston, 2011; Lu et al., 2013). In a recent study it was observed that mutations in troponin causing cardiomyopathies lead to a decrease in the disorder score, reducing troponin's flexibility (Na et al., 2016).

Previous MD studies have been performed on how loss-offunction and gain-of-function mutations affect Ca2<sup>+</sup> binding to troponin (Kekenes-Huskey et al., 2012; Lindert et al., 2012). Cheng et al. tried to explain how mutations affect phosphorylation changes using MD simulations in a study comparing non-phosphorylated and S23/24D phosphomimetic troponin containing the HCM mutations TnI R146G and R21C. Both mutations increased Ca2+-sensitivity and blunted the effect of phosphorylation, confirmed using biochemical methods. MD simulations showed that both mutations increased the Ca2<sup>+</sup> binding affinity, as measured by distance between Ca2<sup>+</sup> and Ser69 GO and inhibited formation of intrasubunit interactions between TnI N-terminus and inhibitory peptide in the phosphomimetic mutants, which are normally seen in S23/S24D WT troponin (Cheng et al., 2015). P83S mutation in the IT arm of TnI had similar effects, although the blunting effect was "weaker" in MD simulations (Cheng et al., 2016).

The same group studied the structural mechanism of TnI R145G HCM mutation that also blunts TnI phosphorylation (Regnier et al., 2014; Lindert et al., 2015). R145G increased the interaction between Ca2<sup>+</sup> and Ser69 GO explaining the increased Ca2<sup>+</sup> binding. On the other hand, these interactions were not observed in Zamora's analysis and the most evident effect of the mutations TnI R145G (HCM), TnI K36Q, and TnC G159D (both DCM) is that they trap Ser69 GO in the short distance configuration (Sheehan et al., 2016; Zamora et al., 2016b).

The changes caused by mutations can be reversible, as we recently discovered that Epigallocatechin-3-gallate (EGCG) and related compounds could restore the Ca2+-sensitivity change upon phosphorylation in DCM or HCM mutant thin filaments that were originally uncoupled (Papadaki et al., 2015). The molecular mechanism of action of EGCG is still to be elucidated, although docking studies indicate EGCG is likely to bind at the interface between N-terminus TnC and TnI (1–30). Current MD studies provide a basis for the investigation of this recoupling process (Hwang, 2016; Marston et al., 2016).

To address how troponin regulates muscle contractility it is necessary to extend studies from isolated troponin to the thin filament. Yang et al have made the most comprehensive reconstruction of actin-tropomyosin-troponin in the absence of Ca2<sup>+</sup> at a resolution of 25Å and have located troponin on

the previously determined actin-tropomyosin, showing details never observed before (Yang et al., 2014). Most strikingly, the C-terminal inhibitory domain of TnI is orientated across the filament, binding to actin close to the N-terminal TnT of the troponin on the opposite side of the filament. The density envelope corresponding to the core domain of troponin is apparent and an attempt has been made to fit troponin into this; the mobile parts of the molecule are, as expected, not seen but the orientation of troponin core indicates that both the regulatory Ca2<sup>+</sup> binding site and the phosphorylatable serines of TnI would be accessible to the solution (**Figure 2**).

Of course this is a static model of just one state; a fully dynamic model of the thin filament is not yet accessible, although progress is being made on several fronts. The dynamics of the actin filament have been simulated using course-grained methods (Fan et al., 2012; Saunders and Voth, 2012) and the structure of tropomyosin alone and bound to actin has been simulated by MD (Zheng et al., 2013). Manning et al have developed the EM structures of the thin filament into a fully atomistic model and derived structures for the whole filament by energy minimization (Manning et al., 2011). This method seems capable of reproducing the features of the Ca2<sup>+</sup> switch and even to explain how mutations like TnT R92Q that are remote from

### REFERENCES


troponin can change Ca2+-sensitivity by using MD simulations of the whole thin filament (Williams et al., 2016). However, current MD simulations are run for less than 5 ns and so do not even start to explore the range of conformations possible in real life.

### CONCLUSIONS

In conclusion, troponin has a rather dynamic structure, owing to the intrinsically disordered domains. This brings difficulties in elucidating its exact structure and thus the mechanisms of its regulation. Determining the structure of troponin mobile regions will not only lead to understanding the changes that occur upon its regulation, but also to predicting the effect of a mutation or a pharmacological agent. For example, just by knowing the structure of tropomyosin and its exact interactions with actin it has been possible to predict the effect of diseasecausing mutations (Marston et al., 2013; Donkervoort et al., 2015). With troponin it is not possible to make such predictions at the moment, as the structure is still unresolved. However, MD simulations may be the best way to study troponin, since the molecule is treated as a moving object and not as a rigid structure. Technical and mathematical advances give hope that in the near future it will be possible to simulate an ensemble of several million atoms for µs and thereby finally get a full description of the dynamic regulated thin filament.

### AUTHOR CONTRIBUTIONS

MP wrote the manuscript and SM edited it.

### FUNDING

This work has been supported by the British Heart Foundation (RG/11/20/29266 and FS/12/24/29568).

### ACKNOWLEDGMENTS

We would like to acknowledge Dr William Lehman (Boston University, MA, USA) for providing us the thin filament 3D reconstruction structure and Mr Juan Eiros Zamora for all the MD troponin structures. We would like to thank Dr Ian Gould, Mr Juan Eiros Zamora (Imperial College London department of Chemistry, London, UK), Miss Alice Sheehan and Dr Andrew Messer (Imperial College London NHLI, London, UK) for their very insightful discussions about MD and also Dr Jonathan Kirk (Loyola University of Chicago, IL, USA) for helping us edit the manuscript.

complex in solution. Biophys. J. 90, 2436–2444. doi: 10.1529/biophysj.105. 076216


mutation on cardiac troponin structural dynamics. Biophys. J. 107, 1675–1685. doi: 10.1016/j.bpj.2014.08.008


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Papadaki and Marston. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Amino Acid Changes at Arginine 204 of Troponin I Result in Increased Calcium Sensitivity of Force Development

Susan Nguyen<sup>1</sup> , Rylie Siu<sup>1</sup> , Shannamar Dewey <sup>1</sup> , Ziyou Cui <sup>1</sup> and Aldrin V. Gomes 1, 2 \*

*<sup>1</sup> Department of Neurobiology, Physiology, and Behavior, University of California, Davis, Davis, CA, USA, <sup>2</sup> Department of Physiology and Membrane Biology, University of California, Davis, Davis, CA, USA*

Mutations in human cardiac troponin I (cTnI) have been associated with restrictive, dilated, and hypertrophic cardiomyopathies. The most commonly occurring residue on cTnI associated with familial hypertrophic cardiomyopathy (FHC) is arginine (R), which is also the most common residue at which multiple mutations occur. Two FHC mutations are known to occur at cTnI arginine 204, R204C and R204H, and both are associated with poor clinical prognosis. The R204H mutation has also been associated with restrictive cardiomyopathy (RCM). To characterize the effects of different mutations at the same residue (R204) on the physiological function of cTnI, six mutations at R204 (C, G, H, P, Q, W) were investigated in skinned fiber studies. Skinned fiber studies showed that all tested mutations at R204 caused significant increases in Ca2<sup>+</sup> sensitivity of force development (1pCa<sup>50</sup> = 0.22–0.35) when compared to wild-type (WT) cTnI. Investigation of the interactions between the cTnI mutants and WT cardiac troponin C (cTnC) or WT cardiac troponin T (cTnT) showed that all the mutations investigated, except R204G, affected either or both cTnI:cTnT and cTnI:cTnC interactions. The R204H mutation affected both cTnI:cTnT and cTnI:cTnC interactions while the R204C mutation affected only the cTnI:cTnC interaction. These results suggest that different mutations at the same site on cTnI could have varying effects on thin filament interactions. A mutation in fast skeletal TnI (R174Q, homologous to cTnI R204Q) also significantly increased Ca2<sup>+</sup> sensitivity of force development (1pCa<sup>50</sup> = 0.16). Our studies indicate that known cTnI mutations associated with poor prognosis (R204C and R204H) exhibit large increases in Ca2<sup>+</sup> sensitivity of force development. Therefore, other R204 mutations that cause similar increases in Ca2<sup>+</sup> sensitivity are also likely to have poor prognoses.

Keywords: troponin I, familial hypertrophic cardiomyopathy, distal arthrogryposis, Ca2<sup>+</sup> sensitivity, mammalian two-hybrid

### INTRODUCTION

Familial hypertrophic cardiomyopathy (FHC) is a genetically heterogeneous autosomal dominant inherited heart disease of the myocardium that results in a high incidence of sudden cardiac death and has been linked to mutations in genes coding for at least 10 sarcomeric proteins. These mutations have been found in the genes for: α-myosin heavy chain

### Edited by:

*Jose Renato Pinto, Florida State University, USA*

#### Reviewed by:

*Henry G. Zot, University of West Georgia, USA Hyun Seok Hwang, Florida State University, USA*

> \*Correspondence: *Aldrin V. Gomes avgomes@ucdavis.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *01 September 2016* Accepted: *20 October 2016* Published: *15 November 2016*

#### Citation:

*Nguyen S, Siu R, Dewey S, Cui Z and Gomes AV (2016) Amino Acid Changes at Arginine 204 of Troponin I Result in Increased Calcium Sensitivity of Force Development. Front. Physiol. 7:520. doi: 10.3389/fphys.2016.00520*

**Abbreviations:** FHC, familial hypertrophic cardiomyopathy; HcTnI, human cardiac troponin I; R204, arginine residue 204 on cardiac troponin I; TM, tropomyosin; Tn, troponin; TnC, troponin C; TnI, troponin I; TnT, troponin T; MOPS, 3-[N-morpholino]propanesulonic acid; WT, wild-type.

(Geisterfer-Lowrance et al., 1990), cardiac myosin essential light chain, and cardiac myosin regulatory light chain (Poetter et al., 1996), α-tropomyosin (Thierfelder et al., 1994), cardiac troponin T (TnT) (Thierfelder et al., 1994), cardiac myosin binding protein C (Bonne et al., 1995; Watkins et al., 1995), cardiac troponin I (TnI) (Kimura et al., 1997), α-actin (Mogensen et al., 1999), troponin C (TnC) (Hoffmann et al., 2001), and titin (Satoh et al., 1999). FHC is the most commonly identified cause of sudden death in young adults. Symptoms of FHC vary between individuals, and most patients are asymptomatic (or only exhibit mild symptoms) before sudden cardiac death (O'Mahony and Elliott, 2014).

TnI is a basic protein that readily interacts with the acidic TnC and binds with another acidic protein, TnT, to form the Troponin (Tn) complex (Zot and Potter, 1987; Tobacman, 1996; Westfall and Metzger, 2001). TnI is also capable of binding actin and inhibiting Mg2+-activated actomyosin ATPase and is commonly referred to as the inhibitory subunit of Tn. The Tn complex and tropomyosin are critical regulatory components of the muscle contraction apparatus (Xu et al., 2010). Muscle contraction involves cyclic binding and release of cross-bridges (myosin heads of the thick filament interacting with and releasing actin of the thin filament) with energy derived from the hydrolysis of ATP.

Kimura et al. reported five missense mutations in TnI (R145G, R145Q, R162W, G203S, and K206Q) that are linked to FHC (Kimura et al., 1997). Since this initial report, several other TnI FHC mutants [examples include R21C, P82S, K183 deletion (1K183), I195M, D195N, S199N, R204C, R204H and an exon 8 deletion mutant encompassing the stop codon of the cTnI gene] have also been reported (Kokado et al., 2000; Morner et al., 2000; Niimura et al., 2002; Doolan et al., 2005). FHC mutations in TnI and other myofilament proteins have been previously shown by many groups to generally result in increased Ca2+-sensitivity of force development (Gomes and Potter, 2004). Since R204 had two known mutations (R204C and R204H), and arginine is the amino acid most commonly associated with FHC in TnI or TnT (Xu et al., 2010), it was hypothesized that any mutations at R204 would cause similar increases in Ca2+-sensitivity of force development. The R204 residue is an important amino acid to investigate since families with the cardiac TnI (cTnI) R204H mutation have a high incidence of sudden death (Doolan et al., 2005). Doolan et al. (2005), investigated R162G, R162P, L198P, R204H mutations and hypothesized that the FHC phenotype of missense mutations of arginine residues of TnI are due to disrupted functional interactions with TnC and TnT, which is tested in the present work using multiple residue replacements of R204.

The cTnI R204 residue is well conserved in different animals suggesting that this residue is evolutionarily important. A comparison of the primary amino acid sequence of cTnI and fast skeletal Troponin I (fsTnI) shows that R204 residue corresponds to the R174 residue of fsTnI. It is hypothesized that changes at the fsTnI R174 residue will show similar increases in Ca2+-sensitivity of force development as observed for the cTnI R204 mutants. Since a known missense mutation (R174Q) in fsTnI was shown to be associated with distal arthrogryposis type 2B (DA2B, also known as arthrogryposis multiplex congenita, distal, type 2B) (Westfall and Metzger, 2001), this mutation was also investigated in skeletal muscle to determine if this mutation showed similar in vitro physiological properties as the cTnI R204Q mutation. Distal arthrogryposis (DA) refers to a group of disorders characterized by multiple congenital contractures of the hands/wrists and feet/ankles, and affected individuals typically have a triangular shaped face, a small mouth, a prominent chin and positional foot deformities (calcaneovalgus and/or clubfoot).

Investigation of six cTnI R204 residues as well as the fsTnI R174Q mutation showed that all the mutations investigated increased the Ca2<sup>+</sup> sensitivity of force development. The different effects of these mutants on the interaction between Tn subunits and the maximal force suggest that the change in Ca2<sup>+</sup> sensitivity of force development may be a major determinant for the poor prognosis of patients with the cTnI R204C, R204H, and fsTnI R174Q mutations.

### EXPERIMENTAL PROCEDURES

### Mutation, Expression, and Purification of HcTnI and HcTnI Mutants

The HcTnI FHC mutants were formed by overlapping PCR using HcTnI cDNA obtained from Dr. J.D. Potter (Zhang et al., 1995). Mutation, expression, and purification of cTnI and TnT mutants were carried out as previously described (Szczesna et al., 2000; Gomes et al., 2005b). The PCR products obtained were digested with NcoI and BamH1 and then ligated to pET-3d vector. The sequence of the TnI mutants was verified by sequencing prior to expression and purification. HcTnI and HcTnI R204 mutants were purified via conventional methods. Briefly, crude bacterial supernatants were purified by column chromatography on an S-Sepharose column at 4◦C and eluted with a linear KCl gradient of 0–0.5 M in a Tris-HCl buffer containing 6M urea. Semipure HcTnI and HcTnI mutantswere dialyzed against a solution containing 50 mM Tris-HCl, pH 7.5, 1 M KCl, 1 M urea, 1 mM DTT, and 2 mM CaCl<sup>2</sup> and loaded onto an affinity column having covalently bound HcTnC. Pure HcTnI and HcTnImutants were eluted with a gradient of 0–3 mM EDTA and 1–6 M urea. The purity of the TnI proteins was determined by SDS-PAGE as previously described (Gilda et al., 2016).

### Mutation and Expression of Wild-Type and Mutant Fast Skeletal TnI

The cDNA encoding rabbit fsTnI was previously cloned by reverse transcriptase-polymerase chain reaction using a template of total RNA from rabbit skeletal muscle and oligonucleotide primers specific for the 5′ and 3′ regions of the respective coding sequences (Sheng et al., 1992). Additionally, the fsTnI R174Q mutant was made using a sequential overlapping polymerase chain reaction-based method as previously described (Gomes et al., 2004, 2005a). Clones were sequenced to verify the correct sequences prior to expression and purification of the respective proteins.

### Protein Purification of Rabbit Fast Skeletal TnT, TnI, and TnC

The purification of fsTnI was similar to previously described (Pan and Potter, 1992) with the exception that the final purification step was with TnC-agarose affinity chromatography. Briefly, after S-Sepharose cation exchange chromatography in the presence of 6M, fsTnI was eluted with a continuous ionic strength gradient of 0 –0.3 M NaCl. The fsTnI containing fractions were pooled and dialyzed in steps (to remove urea) with 6M urea, 4 M urea, 2 M urea in TnC affinity buffer to TnC affinity buffer with no urea (50 mM Tris, pH 7.5 containing 2 mM CaCl2, 0.5M NaCl, and 1 mM DTT). A column with cardiac TnC immobilized onto CNBractivated Sepharose-4B (2 mg TnC/ml agarose) was prepared as described by Pharmacia LKB Biotechnology. The dialyzed TnI was loaded onto the TnC-agarose column equilibrated with TnC affinity buffer. The bound TnI was eluted with TnC elution buffer containing 6 M urea and 3 mM EDTA (50 mM Tris, 0.5 mM NaCl, 1 mM DTT, 3 mM EDTA, 6 M urea, pH 7.5). The eluates were run on SDS gels and purified TnI pooled.

### Formation of the Troponin Complex

Formation of the human cardiac troponin and rabbit skeletal troponin complexes containing recombinant TnT, TnC, and TnI was carried out as recently described by Szczesna et al. (2000), (Szczesna et al., 2000; Gomes et al., 2005b). Proper stoichiometry was verified by SDS-PAGE (Gilda et al., 2016). Although not done routinely, gel filtration of these formed Tn complexes showed that this reconstitution method resulted in a single species.

### Preparation of Porcine Skinned Cardiac Muscle Bundles

Cardiac skinned muscle fibers were prepared following a common laboratory procedure published by Zhang et al. (1995) (Zhang et al., 1995; Gomes et al., 2005a,b). Freshly isolated porcine hearts were incubated in an O2-saturated solution containing 140 mM NaCl, 4 mM KCl, 1.8 mM CaCl2, 1 mM MgCl2, 1.8 mM NaHPO4, 5.5 mM glucose, and 5 mM HEPES, pH 7.4. Cardiac muscle bundles were dissected from the left ventricle of the porcine hearts and were chemically skinned by incubating with 50% glycerol and 1% Triton X-100 in the relaxing solution (pCa 8.0) containing 10−<sup>8</sup> M Ca2+, 5 mM Mg2+, 7 mM EGTA, 20 mM MgATP, 20 mM creatine phosphate, and 15 U/mL creatine phosphokinase, pH 7.0, at an ionic strength of 150 mM at 4◦ C for 24 h. These skinned muscle preparations were dissected into small bundles (1–2 cm in length, 2–3 mm in diameter) and were stored at –20◦ C in the same solution without triton X-100 before use.

### Steady-State Force and Ca2+-Sensitivity of Force Development of Tn Complexes Containing HcTnI Mutants

The skinned fiber preparation was mounted with stainless steel clips on a force transducer and was immersed in the contracting solution to measure initial force before treatment. The contraction solution (pCa 4) had the same composition as the relaxation buffer except for the increased Ca2<sup>+</sup> concentration (10−<sup>4</sup> M). To determine the Ca2<sup>+</sup> dependence of force development, the contraction of the skinned fibers was tested in solutions of intermediate concentrations of Ca2+. The Ca2<sup>+</sup> dependence of force was determined before and after performing the displacement and reconstitution protocols that are described below. To remove the endogenous Tn complex, the TnT displacement method was used (Hatakenaka and Ohtsuki, 1992; Szczesna et al., 2000; Gomes et al., 2005a). The cardiac fiber was incubated with HCTnT (0.8 mg/mL) for a total of 2.5 h at room temperature with an intermediate buffer change containing fresh HCTnT (0.8 mg/mL). After displacement of the endogenous complex by HCTnT, the level of unregulated force development was observed by measuring the level of force reached by skinned fibers in both the pCa 8 solution and the pCa 4 solutions. Restoration of the Ca2<sup>+</sup> regulation of force development was performed using troponin I:troponin C complexes (30 µm) in pCa 8 solution for ∼1.5 h at room temperature. To determine the Ca2<sup>+</sup> dependence of force development, the contraction of skinned fibers was tested in solutions with intermediate concentrations of Ca2<sup>+</sup> (from pCa 8 to pCa 4). The Ca2<sup>+</sup> dependence was determined before and after treatment of the skinned fibers with displacement and reconstitution solutions. The Ca2<sup>+</sup> dependence data were analyzed using the Hill equation (Sigmaplot, Jandel Scientific): relative force (%) = [Ca2+] n /([Ca2+] <sup>n</sup>+[pCa50] n ), where pCa50 is the pCa of a solution in which 50% of the change is produced, and n is the Hill coefficient.

### Steady-State Force and the Ca2+-Sensitivity of Force Development of Tn Complexes Containing Wild-Type and FsTnIR174Q

Skeletal muscle fiber bundles from rabbit psoas muscle were mounted on a force transducer and treated with the pCa 8 relaxing solution containing 1% Triton X-100 for ∼ 1 h. The composition of the pCa 8 solution was 10−<sup>8</sup> M [Ca2+], 1 mM [Mg2+], 7 mM EGTA, 5 mM [MgATp2+], 20 mM imidazole, pH 7.0, 20 mM creatinine phosphate, and 15 units/ml creatinine phosphokinase, ionic strength = 150 mM. To determine the Ca2<sup>+</sup> sensitivity of force development, the fibers were gradually exposed to the solutions of increasing Ca2<sup>+</sup> concentrations, from pCa 8 to pCa 4. To displace the endogenous Tn complex from the fibers they were incubated in a solution containing 250 mM KCI, 20 mM MOPS, pH 6.2, 5 mM MgCl2, 5 mM EGTA, 0.5 mM dithiothreitol, and 1.6–1.8 mg/ml fsTnT, for 1 h at room temperature. A fresh fsTnT protein was applied to the fibers for another 1 h incubation. This was to increase the efficiency of the endogenous Tn displacement from the fibers. Displaced fibers were then washed with the same solution without the protein (10 min at room temperature) and tested for Ca2+-unregulated force that developed due to the absence of the endogenous TnI and TnC. The Ca2<sup>+</sup> regulation of steady-state force was restored with a preformed fsTnI·fsTnC complex. The reconstitution with the fsTnI:fsTnC complex (25 µM) was performed in the pCa 8 solution for ∼1.5 h at room temperature, or long enough for the force to reach a stable level. Control fibers were run in parallel and treated with the same solutions minus the proteins. The final Ca <sup>2</sup>+-sensitivity of force development was determined after fsTnI·fsTnC reconstitution and the data were analyzed with the Hill equation.

### Mammalian Two-Hybrid Studies

Protein:protein interactions were measured using the Checkmate Mammalian Two-Hybrid System (Promega) as previously described (Gilda et al., 2016). cTnI WT and cTnI deletion mutants as well as WT cTnT, and cTnC were subcloned into pACT and pBIND vectors. CV1 cells were transfected with pACT and pBind DNA constructs as well as the pG5luc Vector. TransIT-LT1 transfection reagent (Mirus) was used for all experiments and cells were incubated for 36–48 h following transfection with no media change. Cells were lysed and analyzed using the dual luciferase reporter assay (Promega). To verify the experiments were working correctly, control experiments using various combinations of empty vectors, cTnI, cTnC, or cTnT subcloned vectors were also carried out as previously described (Gilda et al., 2016). To control for potential differential protein expression in the two-hybrid assay, protein levels of the wild-type, and TnI mutants (cloned into the mammalian two hybrid plasmids) in CV1 cells were initially checked by Western blotting to determine if any of the mutants investigated showed significantly different expression levels. All TnI mutants were found to show similar expression levels after 48 h.

### Statistical Methods

All data are presented as mean ± S.D. Comparisons of mammalian-two hybrid data and maximal force data were carried out using one way ANOVA. Unpaired Student's ttest was used to determine the significance of differences in 1pCa (changes in half-maximal activating pCa in different skinned fibers). Values of P < 0.05 were considered statistically significant.

## RESULTS

**Figure 1A** shows the primary structure of the C-terminal region of cTnI with the cTnI FHC mutations that occur within this region indicated. **Figure 1B** shows the primary sequence alignment for HcTnI and human fsTnI (HfsTnI) indicating that the position of the R204 mutation in HcTnI corresponds to R174 in HfsTnI.

### Force Development and the Ca2<sup>+</sup> Dependence of Force Development for cTnI Mutations

To determine the effect of these mutations (R204C and R204H), as well as other potential cTnI mutations, R204G, R204P, R204Q, and R204W, cardiac skinned fiber calcium-force measurements were carried out. We employed a well-established method in our lab (Szczesna et al., 2000; Gomes et al., 2002; Lang et al., 2002) to displace the endogenous Tn complex from skinned porcine cardiac muscle preparations. After determining the level of unregulated force in skinned fibers, they were incubated with either wild-type or mutant HcTnI·HcTnC complexes in low calcium buffer (pCa 8). This allowed us to determine whether the TnI proteins could fully inhibit Ca2<sup>+</sup> unregulated force established after treatment with HcTnT and also to determine if the proteins were able to fully reconstitute the skinned fibers by forming a functional Tn complex. Wild-type HcTnI·HcTnC complex resulted in complete inhibition of Ca2<sup>+</sup> unregulated force.

All six mutations showed significant increases in calcium sensitivity of force development ranging from 1pCa<sup>50</sup> 0.22 (R204W) to 0.37 (R204G) (**Figure 2**, **Table 1**). The mutations associated with FHC, R204C, and R204H, had 1pCa50 values of 0.27 and 0.28 respectively (**Figure 2**, **Table 1**). Recovered force is equivalent to the level of force developed in fibers after reconstituting the fibers with the appropriate Tn complex. The maximal force (force at pCa 4.0) obtained from the skinned fibers was increased for cTnI R204W, R204C, and R204H mutations relative to wild-type cTnI (**Figure 3**). It is interesting that both known FHC mutations, R204C and R204H, increase maximal force.

### Mammalian Two-Hybrid

To determine the effect of the R204 mutations on cTnI: cTnT and cTnI: cTnC interactions, the mammalian two-hybrid luciferase assay was utilized. In the CheckMateTM mammalian two-hybrid system utilized, the pACT vector contains the herpes simplex virus VP16 activation domain upstream of the cloning region, while the pBIND vector contains the yeast GAL4 DNA-binding domain upstream of the cloning region. The pBIND Vector also expresses Renilla reniformis luciferase under the control of the SV40 promoter, which allows the normalization for differences in transfection efficiency. The troponin subunits were cloned into pBIND and pACT Vectors to generate fusion proteins with the DNA-binding domain of GAL4 and the activation domain of VP16, respectively. Association of the DNA-binding domain and the transcriptional activation domain results in transcriptional activation of the firefly luciferase reporter gene and an increase in firefly luciferase expression when compared to the negative controls. Mutations in interacting proteins that disrupt or significantly reduce the interactions between the two proteins being investigated would result in reduced association between the DNA-binding domain and the transcriptional activation domain resulting in less firefly luciferase expression.

The cTnI R204P and R204H mutations showed the weakest interactions with cTnT when compared to wild-type cTnI (**Figure 4**). The other mutations, including R204C, showed no significant impairment when compared to wild-type cTnI. These results suggest differences in the interactions between R204C and R204H with cTnT. Four of the six mutations investigated (R204Q, R204W, R204C, R204H) showed weaker interactions with cTnC (**Figure 5**). The cTnI R204P mutation which showed weaker interaction with cTnT, interacted with cTnC similarly to wild-type cTnI (**Figure 5**). The cTnI R204G mutant also did not show any significant impairment in cTnC binding when compared to wild-type cTnI. The disruption in binding between cTnI and its binding partners may be due to conformation changes in cTnI caused by the mutations.

Some of the cTnI FHC mutations that occur within this region are indicated. The location of the R204 mutation is indicated. M, mouse; RB, rabbit; H, human; R, rat; B, bovine. (B) Alignment of the primary sequences of Human cardiac Troponin I (HCTNI) and Human fast skeletal Troponin I (HFSTNI). Human cardiac Troponin I (Swiss-Prot accession number: P19429). Human fast skeletal Troponin I (Swiss-Prot accession number: P48788). \*Indicates identical amino acid.: indicates homology.

FIGURE 2 | Effect of mutations in Troponin I at R204 on the calcium sensitivity of force development. Each skinned muscle preparation was treated with HcTnT to displace the endogenous Tn complex and subsequently reconstituted with either HcTnI·HcTnC, or HcTnI R204 mutant. (A) Comparison of HcTnI·HcTnC and HcTnI R204Q·HCTnC. (B) Comparison of HcTnI·HcTnC and HcTnI R204C·HCTnC. (C) Comparison of HcTnI·HcTnC with HcTnI R204W·HcTnC or HcTnI R204P·HcTnC. (D) Comparison of HcTnI·HcTnC with HcTnI R204H·HcTnC or HcTnI R 204G·HcTnC. The Ca2<sup>+</sup> dependence of force was measured in each preparation after reconstituting troponin. Each point is the average of 3 experiments and represents the mean ± S.D. \**p* < 0.05.

TABLE 1 | Effect of wild-type Human cTnI and mutants on the Ca2<sup>+</sup> sensitivity of force development (pCa50) and the Hill coefficient (nH) in skinned porcine cardiac muscle fibers.


*The pCa<sup>50</sup> and n<sup>H</sup> values are the average of 3 independent fiber experiments, and the errors are the standard deviation (S.D.) values.* \* *Indicates that the pCa<sup>50</sup> values and Hill coefficient for the respective TnI mutants are significantly different from wild-type HcTnI (p* < *0.05).*

### Force Development and the Ca2<sup>+</sup> Dependence of Force Development for fsTnI

Rabbit fsTnI was studied in place of HfsTnI because it is the best characterized fast skeletal muscle system and the C-terminal half of rabbit fsTnI shows 100% homology with HfsTnI. The homology between rabbit fsTnI and HfsTnI is 98.9%. Under the conditions utilized, essentially all of the endogenous Tn complex was displaced following treatment with fsTnT. The amount of force development in the presence of very low concentration of Ca2<sup>+</sup> (pCa 8.0) after fsTnT displacement is a measure of the

extent of displacement of endogenous fsTnI (referred to as the Ca2<sup>+</sup> unregulated force). In the absence of fsTnI, fibers were unable to relax, as the inhibitory activity of cTnI is deficient. This measurement of Ca2<sup>+</sup> unregulated force was utilized to ensure that all the skinned fibers are displaced to the same extent. All the fibers selected for these studies had a Ca2<sup>+</sup> unregulated force of >95%.

After displacement and reconstitution, the maximal force obtained for each fiber was measured. This force was measured relative to the initial force of the skinned fibers before TnT displacement. An R174Q fsTnI mutation (associated with DA) increased the Ca2<sup>+</sup> sensitivity of force development (**Figure 6**, **Table 2**). The results suggest that mutations associated with

FIGURE 5 | Effect of mutations in Troponin I at R204 on its interaction with Troponin C. Mammalian two-hybrid was utilized to determine disruptions in the interactions between different cTnI's and wild-type cTnC. \**P* < 0.05 (*n* = 4–5).

significantly increased Ca2<sup>+</sup> sensitivity of force development may be associated with HCM in cardiac tissue or DA in skeletal muscle. Unlike some of the cardiac R204 mutations which showed increased maximal force, the fsTnI R174Q mutant showed decreased maximal force relative to wild-type fsTnI (**Figure 7**).

average of 3–5 experiments and represents the mean ±*S.D*.

### DISCUSSION

Arginine residues in cTnI are associated with multiple mutations resulting in potentially different clinical phenotypes (Xu et al.,


TABLE 2 | Effect of wild-type Rabbit Fast Skeletal TnI and TnI R174Q on the Ca2+-sensitivity of force development (pCa50) and the Hill coefficient (nH) in skinned fast skeletal muscle fibers.

\**Indicates that the pCa<sup>50</sup> values for the respective TnI mutants are significantly different from wild-type HcTnI (P* < *0.05). The pCa<sup>50</sup> and nH values are the average of 3-5 independent fiber experiments, and the errors are the standard deviation (S.D.) values. The average wild-type TnI displacement was 97.6* ± *2.4%.*

2010). Since two mutations (R204C and R204H) at residue 204 of cTnI are associated with FHC and most mutations associated with FHC show increased Ca2+-sensitivity of force development (Xu et al., 2010), we hypothesized that any mutations at cTnI R204 would cause significant increases in Ca2+-sensitivity of force development. To characterize the effects that different arginine mutations at the same residue would have on the physiological function of cTnI, six mutations at R204 (C, G, H, P, Q, W) were investigated in skinned fiber studies. All of the cTnI R204 mutations investigated displayed increased Ca2<sup>+</sup> sensitivity of force development when compared to WT cTnI (1pCa<sup>50</sup> = 0.22–0.37), similar to what is observed for most FHC mutations. These results suggest that the R204 residue is important in cTnI function and that increased Ca2<sup>+</sup> sensitivity may be a major factor in HCM. The increase in myofilament sensitivity to Ca2<sup>+</sup> is important since reduction in the Ca2<sup>+</sup> sensitivity was found to prevent the development of HCM (Alves et al., 2014). Others have suggested that increased Ca2+-sensitivity is associated with HCM (Wei and Jin, 2015).

Certain regions of cTnI seem to be more important functionally than other regions (Mogensen et al., 2015; Wei and Jin, 2015; Meyer and Chase, 2016; Sheng and Jin, 2016). It may be that any or most amino acid changes at certain TnI residues that occur in regions functionally important for regulating Ca2+-sensitivity would all be associated with increased Ca2+-sensitivity of force development. Transgenic mice expressing 9–17% of a C-terminal truncated human cardiac TnI (residues 194–210 deleted, corresponding to residues 166– 182 of fsTnI), developed a phenotype of stunned myocardium (Murphy et al., 2000). These results suggest that the 194– 210 region of cTnI is important for cTnI function. The FHC mutations R204C and R204H both showed similar increases in Ca2<sup>+</sup> sensitivity (1pCa50 values of 0.28 and 0.29 respectively). These two mutations also showed increased maximal force compared to WT cTnI. R204W also showed increased maximal force. The R204H mutation was found to be associated with HCM in an Australian family and is associated with a poor prognosis (Doolan et al., 2005).

To determine how R204 mutations might disrupt proteinprotein interactions, a mammalian two-hybrid luciferase assay was used. Mutation of arginine to proline resulted in a significant reduction in functional interactions with both cTnC and cTnT. The results suggest that binding of cTnI mutants (except R204G) to its binding partners is disrupted pending measurement of actual affinity constants. Interestingly, while the R204C mutation weakened the interaction of TnI with TnC, it did not significantly affect the interaction of TnI with TnT. However, the cTnI R204H mutation disrupted the interaction between cTnI and cTnT as well as between cTnI and cTnC, which is consistent with a previous report using a similar mammalian two hybrid assay (Doolan et al., 2005). Doolan et al. found evidence for altered interactions between TnI and either TnT, TnC, or both, for cTnI R162G, R162P, L194P, and R204H mutations. The previous finding that the cTnI R162G mutation reduced cTnI and cTnT interaction while another mutation at the same cTnI residue, R162P, did not affect cTnI-cTnT interaction (Doolan et al., 2005) is consistent with our results which suggest that not all cTnI HCM mutations affect cTnI-cTnT mutations. Using the data obtained from the four mutations they investigated, Doolan et al. concluded that altered interactions in the Tn complex "increased severity of the disease." The replacement of cTnI arginine residue 204, a relatively large and charged amino acid, by glycine, the smallest amino acid, resulted in the largest increase in Ca2+-sensitivity of force development. The severity of the disease is likely to be more complicated due to altered interactions in the Tn complex since our R204G results suggest increased Ca2+-sensitivity of force development independent of any change in interactions with its Tn binding partners.

Replacement of arginine with proline also resulted in a large increase in Ca2<sup>+</sup> sensitivity of force development, which was significantly larger than the increase in Ca2<sup>+</sup> sensitivity of force development observed for the R204W mutation (p < 0.05). This difference in Ca2<sup>+</sup> sensitivity changes may be due to the size of the tryptophan residue, which is the largest amino acid and closer to the size of arginine than proline or glycine (Morris et al., 1992). Proline is also typically considered a structural disruptor of the protein secondary structure and its side chains are conformationally rigid, unlike glycine which can easily adopt many main chain conformations.

To extend the hypothesis that any mutation at cTnI R204 would cause significant increases in Ca2<sup>+</sup> sensitivity of force development, we hypothesized that mutations at fsTnI R174, which occurs at an equivalent location to the cTnI R204 residue, would also cause significant increases in Ca2+-sensitivity of force development. We investigated the R174Q missense mutation associated with DA. The WT rabbit fsTnI consists of 182 residues, and the C-terminus region of fsTnI, which is affected by the R174Q mutation, is known to interact with TnC and actin. The fsTnI R174Q DA mutant showed increased Ca2+-sensitivity of force development, similar to what was observed for the cTnI R204Q mutation. Hence, the location of the R174Q DA mutation is likely to be physiologically important for the function of TnI. The R174Q was previously investigated by another group that showed a greater increase in Ca2+-sensitivity of force development (1pCa<sup>50</sup> = 0.36) and no change in maximal force (Robinson et al., 2007). The differences from what we observed may be due to species of cTnI utilized, as Robinson et al. utilized human fsTnI in muscle fibers in rabbit skeletal muscle fibers while we utilized rabbit fsTnI in rabbit skeletal muscle fibers (Robinson et al., 2007). However, in both cases the increase in Ca2+-sensitivity of force development relative to WT fsTnI was significant.

DA occurs in about 1 in 3000 children, and other fsTnI mutations have been found to be associated with DA2B (fsTnI R156ter, K175del, and K176del mutations) (Kimber et al., 2012). Patients with DA2B have contractures which are present at birth, suggesting impaired relaxation of muscle fibers. A likely possibility is that the higher Ca2+-sensitivity of force development in fibers containing DA2B mutations affects muscle contracture. These results, as well as two other reports that showed increased Ca2+-sensitivity with DA2B mutations (Robinson et al., 2007; Mokbel et al., 2013), all suggest that joint and muscle contractures may be caused by prolonged muscle hypercontraction due to increased myofibrillar calcium sensitivity and that treatments that reduce skeletal muscle Ca2<sup>+</sup> sensitivity may be beneficial to patients with DA2B.

Overall, these results suggest that different amino acids at the same site on cTnI could affect thin filament interactions differentially. While significant impairment in the interactions of cTnI with TnT or TnC may be enough to cause significant changes in Ca2+-sensitivity of force development, impairment in these interactions is not a requirement for altering the Ca2+ sensitivity as our study showed that R204G had the largest increase in Ca2+-sensitivity yet demonstrated maximal tension and cTnI:TnT/TnC interactions comparable to wild-type levels. While R204G did not show any altered interactions between cTnI and TnC or TnT it is possible that altered interactions between TnI R204G and actin may occur. Unfortunately, we were unable to get cTnI:actin two hybrid studies to work well. It may also be possible that the R204 mutations are affecting thin and thick filament interactions by disturbing tropomyosin displacement. If the large increase in Ca2+-sensitivity of force development observed with these mutations is associated with the poor prognosis, then many other R204 mutations that may be discovered in the population are likely to have poor prognoses.

### ETHICS STATEMENT

This study was exempt from this requirement because the tissues were obtained from the University of California, Davis Veterinary School Slaughterhouse. The Veterinary School Slaughterhouse routinely sacrifices animals for teaching purposes and sells the meat from these animals to the public. A qualified researcher can pay a small fee (\$10) and get heart tissue or some skeletal muscle for research from these animals that have already been sacrificed.

### AUTHOR CONTRIBUTIONS

AG: design of work, experiments, acquisition, interpretation of data, drafting of work, final approval, agreement to be accountable for all aspects of work. SN: acquisition, interpretation of data, revising, final approval, agreement to be accountable for all aspects of work. RS: acquisition, revising, final approval, agreement to be accountable for all aspects of work. SD: acquisition, revising, final approval, agreement to be accountable for all aspects of work. ZC: interpretation of data, revising, final approval, agreement to be accountable for all aspects of work.

### FUNDING

This research was supported by a UC Davis funds and a Hellman Fellowship (AG).

### ACKNOWLEDGMENTS

We would like to thank Dr. Jingsheng Liang (University of Miami) for help with the fiber studies.

### REFERENCES


Zot, A. S., and Potter, J. D. (1987). Structural aspects of troponintropomyosin regulation of skeletal muscle contraction. Annu. Rev. Biophys. Biophys. Chem. 16, 535–559. doi: 10.1146/annurev.bb.16.060187. 002535

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer HSH and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2016 Nguyen, Siu, Dewey, Cui and Gomes. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Connecting Sarcomere Protein Mutations to Pathogenesis in Cardiomyopathies: The Development of "Disease in a Dish" Models

Rebecca Zaunbrecher <sup>1</sup> and Michael Regnier 1, 2, 3 \*

*<sup>1</sup> Department of Bioengineering, University of Washington, Seattle, WA, USA, <sup>2</sup> Center for Cardiovascular Biology, Seattle, WA, USA, <sup>3</sup> Institute for Stem Cell and Regenerative Medicine, University of Washington, Seattle, WA, USA*

Recent technological and protocol developments have greatly increased the ability to utilize stem cells transformed into cardiomyocytes as models to study human heart muscle development and how this is affected by disease associated mutations in a variety of sarcomere proteins. In this perspective we provide an overview of these emerging technologies and how they are being used to create better models of "disease in a dish" for both research and screening assays. We also consider the value of these assays as models to explore the seminal processes in initiation of the disease development and the possibility of early interventions.

### Edited by:

*P. Bryant Chase, Florida State University, USA*

#### Reviewed by:

*Jonathan P. Davis, Ohio State University, USA Jop Van Berlo, University of Minnesota, USA*

> \*Correspondence: *Michael Regnier mregnier@uw.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *08 September 2016* Accepted: *07 November 2016* Published: *22 November 2016*

#### Citation:

*Zaunbrecher R and Regnier M (2016) Connecting Sarcomere Protein Mutations to Pathogenesis in Cardiomyopathies: The Development of "Disease in a Dish" Models. Front. Physiol. 7:566. doi: 10.3389/fphys.2016.00566* Keywords: stem cell derived cardiomyocytes, gene editing, cardiomyopathies, contraction, methods development

It has now been over 35 years since the first report of a genetic linkage between a sarcomere protein mutation (or variant) with a disease phenotype (Geisterfer-Lowrance et al., 1990). Since then well over 1000 genetic variants of sarcomere proteins have been reported as associated with diseases such as hypertrophic, dilated, or restrictive cardiomyopathy, though many fewer have been confirmed as having causative roles (Seidman and Seidman, 2011; Haas et al., 2015). The vast majority of research to understand the phenotypic consequences of these genetic variants has been done using postnatal animals, cell culture, and recombinant protein models. Thus, much has been learned about later stages of the disease process, likely after multiple compensatory processes have been invoked and often when hearts are in failure. However, there is growing consensus that to find effective treatments for these familial diseases it is important to understand the role of these mutations in earlier stages in disease progression, before clinical signs manifest and perhaps at the earliest stages of human development.

The development of technologies to derive human pluripotent stem cells and differentiate them into cardiomyocytes has provided a model system in which to study these crucial early stages of development. Beginning with the derivation of the first human embryonic stem cell line in 1998 (Thomson et al., 1998) and continuing with the derivation of human induced pluripotent stem cells (hiPSCs) in 2007 (Takahashi and Yamanaka, 2006), it has been theoretically possible to generate any tissue in the lab. However, early differentiation protocols for cardiomyocytes depended upon the spontaneous formation of embryoid bodies and resulted in low yields of cardiomyocytes (∼1–5%; Kehat et al., 2001). Better understanding of pathways involved in heart development in vivo has since been leveraged to develop more efficient differentiation protocols and it is now possible to achieve >90% pure cardiomyocyte populations using multiple methods (Spater et al., 2014). Additionally, these technologies in reprogramming and directed differentiation have made it possible to generate cell lines and large numbers of cardiomyocytes from patient samples and have laid the foundation for hiPSC-based cardiac disease modeling.

A series of technological advances in genomic engineering have also advanced the ability to model cardiomyopathies with hiPSC-derived cardiomyocytes (hiPSC-CMs). Until recently, cell lines have largely been generated using samples from patients who both (1) carry a known or suspected pathogenic mutation and (2) clinically present with cardiomyopathy. A major benefit of this approach is that the resulting cell line will carry both the mutation of interest as well as any undefined or unknown genetic modifiers that may be necessary for presentation of the disease phenotype. This enhances the likelihood that a phenotype will emerge in vitro. However, it can often be difficult to establish appropriate controls for patient-derived cell lines. Even control cell lines established from healthy close relatives can have significant genetic variation compared to the cardiomyopathy line, and when relatives' samples are unavailable often completely unrelated wildtype cell lines are used (Siu et al., 2012; Lin et al., 2015). Nevertheless, this has been used to successfully study both hypertrophic (Lan et al., 2013) and dilated (Sun et al., 2012; Wu et al., 2015) cardiomyopathy and gain mechanistic insights into the pathogenesis of these mutations.

Breakthroughs in genome engineering technologies, notably the development of the CRISPR/Cas9 system for use in mammalian systems, currently allow for the generation of isogenic control lines in hiPSC-CM modeling. Using a 20-base pair single guide RNA, the CRISPR/Cas9 system can create double-stranded breaks at nearly any location in the genome. This allows for the straightforward generation of random mutations using non-homologous end joining (NHEJ), as well as specific base pair changes at lower efficiencies by supplying a template for homology-directed repair (HDR; Ran et al., 2013). More recent advances have focused on improving the efficiency of HDR (Chu et al., 2015; Yu et al., 2015) and increasing the target range of CRISPR systems by mutating the commonly used Cas9 nuclease (Kleinstiver et al., 2015a) and deriving nucleases from different species of bacteria (Kleinstiver et al., 2015b).

Two general approaches are available for using genome engineering in hiPSC-CM modeling. In the first, mutations can be specifically engineered into a healthy, wildtype hiPSC line. A significant benefit to this approach is that many different mutations can be tested on the same genetic background, allowing for a rigorous comparison. Particularly for non-sense mutations, this can be done in a relatively high-throughput manner through the use of NHEJ. Alternatively, mutations in a patient-derived line can be corrected using HDR. However, this is often technically more problematic, and it can be difficult to discern if performance returns to healthy, wildtype levels. Although these genome engineering technologies are still relatively new, it is clear that controls generated using these techniques are more accurate than cell lines created from unaffected relatives or unrelated individuals. In the future, isogenic controls should be considered a standard in hiPSC-CM disease modeling.

There are numerous technologies for phenotyping hiPSC-CMs. Of particular interest in many cardiomyopathies are mechanical function measurements of hiPSC-CMs, which can be acquired on single cells or in the context of a multicellular tissue engineered system. Current assays allow for highly sensitive measurements of both force production and kinetics of hiPSC-CM contraction and relaxation, and these measurements have been successfully used to characterize the phenotype of several hiPSC-CM cardiomyopathy models (Sun et al., 2012; Hinson et al., 2015). Recent reviews provide comprehensive coverage of both single cell measurement systems (Polacheck and Chen, 2016) and multicellular tissue engineering approaches (Tzatzalos et al., 2016). Of note, a method has recently been developed to mature hiPSC-CMs sufficiently to harvest isolated myofibrils for mechanics measurements to determine how sarcomeric mutations directly affect the organelle in which they are located (Pioner et al., 2016).

With the addition of this assay, it is now possible to characterize the functional properties of hiPSC-CMs on subcellular, cellular, and multicellular levels, a crucial set of tools for cardiomyopathy disease modeling. Studying the effects of disease-associated mutations at each level of organization provides insight into distinct aspects of cardiomyocyte function and how they are affected by the mutation of interest. Myofibril mechanics measurements assess the contractile ability of the hiPSC-CMs independent of the influences of the cell's Ca2<sup>+</sup> handling properties or intracellular signaling pathways. Single cell measurements assess the function of myofibrils in the context of whole-cell function, but without the influence of cell-cell communication. Finally, multicellular tissue constructs provide insight into the effects of cellular junctions and environmental cues from extracellular matrix, and is currently the most physiological setting in which to study hiPSC-CMs in vitro. By performing these multiscale analyses in an in vitro setting, where the internal and external environments can be manipulated, it is now possible to study how molecular level changes in myofilament protein structure and function (with mutations) affect contractile fibrils and how this may influence coupled systems such as the calcium handling, energetic production, and protein expression systems in cells and tissue. In turn this should provide new insight into which levels of structural organization are most affected during the initiation and propagation of disease phenotype, and whether the disease phenotype requires neurohumeral input.

A blessing and a curse central to the use of hiPSC-CMs as models for familial cardiomyopathies is the maturity of the cells. By most electrophysiological, morphological, metabolic, and mechanical measures, these cells are far from an adult phenotype (Yang et al., 2014). Although it can be difficult to match hiPSC-CMs to an exact gestational age in vivo, a recent study suggests when cells are matured on nanopatterned surfaces for out to 80–100 days in culture they have adult-like cell size and morphological characteristics. These cells express the adult form of cardiac myosin (β-myosin from MYH7), although isoform studies of troponin I suggest a fetal stage of expression (Bedada et al., 2014). Additionally, their myofibril mechanical properties and sarcomere ultrastructure match those of myofibrils from 75 day fetal heart tissue (Pioner et al., 2016). Thus, caution should be used in judging the developmental state of these hiPSC-CMs, based on appearance at the light microscope level.

Phenotyping cells at such an immature state is also a doubleedged sword for studying genetic cardiomyopathies (summarized in **Table 1**). Many genetic cardiomyopathies clinically present in adulthood, and there are concerns about whether in vitro



Using isogenic controls allows for rigorous tests of causality and comparisons of mutations

modeling will demonstrate an appropriate disease phenotype that recapitulates what occurs in vivo. However, in these familial-based diseases there are likely changes occurring in the myocardium well before many cardiomyopathies are detected clinically that can be identified using in vitro assays. For example, a recent study using hiPSC-CMs to study titin truncating mutations as a basis for dilated cardiomyopathy noted a disease phenotype in several assay systems, despite the fact that DCM associated with these mutations is usually detected well into adulthood (Hinson et al., 2015).

In fact, the immaturity of hiPSC-CMs can be beneficial for studying genetic cardiomyopathies. Studies that utilize animal models of cardiomyopathy often focus on characterizing end-stage phenotypes, contributing data mostly to enhance understanding of the final presentation of the disease. Alternatively, the fetal-like properties of hiPSC-CMs can provide mechanistic insight into early differences in function and structure present in cardiomyopathy that may be difficult or impossible to tease out in an in vivo system. For mutations in proteins that are not expressed initially in cardiomyocyte differentiation, but later in the timeline of development, it should be possible to determine the seminal event that leads to disease development and associated compensatory mechanisms. This developmental view of disease progression is particularly important as genetic testing grows in prevalence and robustness, and the possibility of pre-emptively treating genetic diseases before they present clinically becomes more likely. Using hiPSC-CMs as a model of genetic cardiomyopathies can allow for these crucial studies to understand how they develop, so that the progression or development of them can be prevented.

Looking forward, additional new tools are being developed that, combined with current technology, will allow for more and better models of disease in a dish. One example of this is the development of hiPSC lines that express proteins from the endogenous loci with a linked fluoro-tag that allow for studies of structural development with repeated live cell imaging (Ding et al., 2013). Another tool is optogenetics, which has been used

### REFERENCES


primarily by neuroscientists to date (Steinbeck et al., 2015). The ability to express channels that can provide temporal and spatial control of stimulation or inhibition of depolarization could be quite useful for developing culture-based models of arrhythmiagenesis and long Q-T syndrome, but may also provide insight to how sarcomere protein mutations disrupt the balance between the contractile apparatus and Ca2+ cycling dynamics.

There still remain many limitations of in vitro models and some aspects of disease will be challenging or impossible to study using hiPSC-CMs. Examples include the complex temporal and spatial influence of neuro-hormonal, paracrine, and endocrine factors and how these are influenced by the structural and functional changes resulting from mutant protein expression. On the other hand, these new and emerging approaches will allow for studies that are impossible to perform with other platforms (e.g., early development, studying mutations that would be lethal) and inform on design of animal models that can probe those aspects. Ultimately these tools will be most useful when used in conjunction with in vivo models.

In summary, a new generation of biophysical, gene editing, and bioengineering approaches that are emerging holds great promise and potential as tools to improve our understanding of cardiac muscle development and the initiation and early-stage progression of disease development. These, in turn, will allow for better models used for drug and small molecule screening and the next generation of targeted therapies for heart failure.

### AUTHOR CONTRIBUTIONS

RZ and MR contributed equally in the ideas and writing of the manuscript.

### ACKNOWLEDGMENTS

Funding for this works was provided by NIH HL111197 (MR) and NIH T32EB001650 (RZ). MR is an Established Investigator of the American Heart Association.

for CRISPR-Cas9-induced precise gene editing in mammalian cells. Nat. Biotechnol. 33, 543–548. doi: 10.1038/nbt.3198


cardiomyopathy: a β cardiac myosin heavy chain gene missense mutation. Cell 62, 999–1006. doi: 10.1016/0092-8674(90)90274-I


specific induced pluripotent stem cells. Aging (Albany NY) 4, 803–822. doi: 10. 18632/aging.100503


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Zaunbrecher and Regnier. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Epigallocatechin-3-Gallate Accelerates Relaxation and Ca2<sup>+</sup> Transient Decay and Desensitizes Myofilaments in Healthy and Mybpc3-Targeted Knock-in Cardiomyopathic Mice

#### Edited by:

P. Bryant Chase, Florida State University, USA

#### Reviewed by:

Vincent Jacquemond, Centre national de la recherche scientifique, France Gustavo Brum, Universidad de la República, Uruguay Mohammad T. Elnakish, Ohio State University, USA

#### \*Correspondence:

Felix W. Friedrich f.friedrich@uke.de Lucie Carrier l.carrier@uke.de

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 27 September 2016 Accepted: 22 November 2016 Published: 05 December 2016

#### Citation:

Friedrich FW, Flenner F, Nasib M, Eschenhagen T and Carrier L (2016) Epigallocatechin-3-Gallate Accelerates Relaxation and Ca2<sup>+</sup> Transient Decay and Desensitizes Myofilaments in Healthy and Mybpc3-Targeted Knock-in Cardiomyopathic Mice. Front. Physiol. 7:607. doi: 10.3389/fphys.2016.00607 Felix W. Friedrich1, 2 \*, Frederik Flenner 1, 2, Mahtab Nasib1, 2, Thomas Eschenhagen1, 2 and Lucie Carrier 1, 2 \*

<sup>1</sup> Cardiovascular Research Center, Department of Experimental Pharmacology and Toxicology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany, <sup>2</sup> German Centre for Cardiovascular Research (DZHK), Hamburg, Germany

Background: Hypertrophic cardiomyopathy (HCM) is the most common inherited cardiac muscle disease with left ventricular hypertrophy, interstitial fibrosis and diastolic dysfunction. Increased myofilament Ca2<sup>+</sup> sensitivity could be the underlying cause of diastolic dysfunction. Epigallocatechin-3-gallate (EGCg), a catechin found in green tea, has been reported to decrease myofilament Ca2<sup>+</sup> sensitivity in HCM models with troponin mutations. However, whether this is also the case for HCM-associated thick filament mutations is not known. Therefore, we evaluated whether EGCg affects the behavior of cardiomyocytes and myofilaments of an HCM mouse model carrying a gene mutation in cardiac myosin-binding protein C and exhibiting both increased myofilament Ca2<sup>+</sup> sensitivity and diastolic dysfunction.

Methods and Results: Acute effects of EGCg were tested on fractional sarcomere shortening and Ca2<sup>+</sup> transients in intact ventricular myocytes and on force-Ca2<sup>+</sup> relationship of skinned ventricular muscle strips isolated from Mybpc3-targeted knockin (KI) and wild-type (WT) mice. Fractional sarcomere shortening and Ca2<sup>+</sup> transients were analyzed at 37◦C under 1-Hz pacing in the absence or presence of EGCg (1.8µM). At baseline and in the absence of Fura-2, KI cardiomyocytes displayed lower diastolic sarcomere length, higher fractional sarcomere shortening, longer time to peak shortening and time to 50% relengthening than WT cardiomyocytes. In WT and KI neither diastolic sarcomere length nor fractional sarcomere shortening were influenced by EGCg treatment, but relaxation time was reduced, to a greater extent in KI cells. EGCg shortened time to peak Ca2<sup>+</sup> and Ca2<sup>+</sup> transient decay in Fura-2-loaded WT and KI cardiomyocytes. EGCg did not influence phosphorylation of phospholamban. In skinned cardiac muscle strips, EGCg (30µM) decreased Ca2<sup>+</sup> sensitivity in both groups.

Conclusion: EGCg hastened relaxation and Ca2<sup>+</sup> transient decay to a larger extent in KI than in WT cardiomyocytes. This effect could be partially explained by myofilament Ca2<sup>+</sup> desensitization.

Keywords: epigallocatechin-3-gallate, hypertrophic cardiomyopathy, Mybpc3, myofilament Ca2<sup>+</sup> sensitivity, relaxation, Ca2<sup>+</sup> transient

### INTRODUCTION

Hypertrophic cardiomyopathy (HCM) is the most common cardiac genetic disease, with more than 1400 different mutations in genes encoding primarily sarcomeric proteins (Friedrich and Carrier, 2012; Maron et al., 2014; Ho et al., 2015). The most frequently mutated genes are MYH7 (encoding β-myosin-heavy chain) and MYBPC3 (encoding cardiac myosin-binding protein C), which constitute about 80% of known mutations. Besides a typical hypertrophy of the left ventricle, patients often present a normal or increased ejection fraction, but a compromised diastolic function with an incomplete relaxation and increased filling pressures (Elliott et al., 2014). Diastolic dysfunction may result in left atrial enlargement and is associated with exercise intolerance and bad prognosis in HCM, primarily due to supraventricular arrhythmias (Yang et al., 2009). Tissue Doppler measurements have revealed that a reduction in systolic and diastolic velocities is prominent even before the development of left ventricular hypertrophy (Charron et al., 1997).

Increased myofilament Ca2<sup>+</sup> sensitivity, as observed in three Mybpc3 cardiomyopathy mouse models (Mybpc3 KO and KI) developed by us and others (Cazorla et al., 2006; Pohlmann et al., 2007; Vignier et al., 2009; Fraysse et al., 2012; Barefield et al., 2014), and in other animal models of HCM (Knollmann et al., 2001; Robinson et al., 2007; Iorga et al., 2008), could be an underlying cause of diastolic dysfunction. This observation has also been made in human HCM samples (Jacques et al., 2008; van Dijk et al., 2009, 2012) and could explain the incomplete relaxation in diastole in MYBPC3-associated HCM (and probably other cases associated with an increased Ca2<sup>+</sup> sensitivity). Additionally, myofilaments with increased sensitivity to Ca2<sup>+</sup> may act as Ca2<sup>+</sup> buffers, prolonging the export of Ca2<sup>+</sup> and relaxation time which could contribute to diastolic dysfunction and arrhythmias (Morimoto et al., 1998; Baudenbacher et al., 2008). These findings support the hypothesis that interventions decreasing myofilament Ca2<sup>+</sup> sensitivity could reverse the phenotype of HCM and have therapeutic value (Jagatheesan et al., 2007; Alves et al., 2014; Tardiff et al., 2015).

Epigallocatechin-3-gallate (EGCg), a major component of green tea, has been suggested to be effective against cardiovascular diseases. Proposed mechanisms were antioxidative, anti-inflammatory, vasorelaxant, and positive inotropic effects (Chyu et al., 2004; Lorenz et al., 2004; Ludwig et al., 2004). Furthermore, it was shown that EGCg lowered myofilament Ca2<sup>+</sup> sensitivity in a transgenic HCM mouse model expressing a human cardiac troponin T (TNNT2, cTnT) mutant (Tadano et al., 2010) and in HCM-associated human cardiac troponin I (TNNI3, cTnI) and cTnT mutants in a reconstituted acto-myosin system (Warren et al., 2015; Messer et al., 2016). However, the effects of EGCg were not evaluated in other HCM models associated with mutations in the thick filament of the sarcomere. Since MYBPC3 is the major disease gene constituting 45% of genetically diagnosed HCM cases (Ho et al., 2015), we used a representative mouse model carrying the human c.772G> A MYBPC3 mutation (Vignier et al., 2009). This mutation was found in 14% unrelated HCM patients in Tuscany and is associated with a bad prognosis (Richard et al., 2003; Girolami et al., 2006; Ho et al., 2015). These mice exhibit, in addition to left ventricular hypertrophy and decreased fractional area shortening, increased myofilament Ca2<sup>+</sup> sensitivity, and diastolic dysfunction (Fraysse et al., 2012). We evaluated the acute effects of EGCg on sarcomere shortening and Ca2<sup>+</sup> transient in intact ventricular myocytes and on force-Ca2<sup>+</sup> relationship of skinned cardiac muscle strips isolated from KI and wild-type (WT) mice.

### MATERIALS AND METHODS

### Animals

The Mybpc3 KI cardiomyopathy mouse model was generated by the targeted insertion of a G > A transition on the last nucleotide of exon 6 and maintained on the Black Swiss background (Vignier et al., 2009; Fraysse et al., 2012; Schlossarek et al., 2012, 2014; Gedicke-Hornung et al., 2013; Mearini et al., 2013, 2014; Stöhr et al., 2013; Friedrich et al., 2014; Najafi et al., 2015; Thottakara et al., 2015; Flenner et al., 2016). This study was carried out in accordance with the recommendations of the guide for the care and use of laboratory animals published by the NIH (Publication No. 85–23, revised 2011 published by National Research Council). All experimental procedures were in harmony with the German Law for the Protection of Animals and the protocol was approved by the Ministry of Science and Public Health of the City State of Hamburg, Germany (Org 653).

### Ventricular Myocyte Preparation

Cardiomyocytes were isolated from WT and KI mouse heart ventricles as previously described (El-Armouche et al., 2007; Pohlmann et al., 2007; Flenner et al., 2016). Mice were anesthetized with CO<sup>2</sup> and sacrificed by cervical dislocation. Hearts were excised, cannulated via the aorta and installed on a temperature-controlled (37◦C) perfusion system. After retrograde perfusion with Ca2+-free buffer solution (113 mM NaCl, 4.7 mM KCl, 0.6 mM KH2PO4, 0.6 mM Na2HPO4, 1.2 mM MgSO4, 12 mM NaHCO3, 10 mM KHCO3, 30 mM taurine, 5.55 mM glucose, 10 mM 2,3-butanedione monoxime 10 mM HEPES, pH 7.46) for 6.5 min, hearts were digested with 0.075 mg/ml Liberase TM (Roche Diagnostics, Mannheim, Germany) dissolved in buffer solution containing 12.5µM CaCl<sup>2</sup> for 7–8 min. Ventricles were disconnected from the atria and minced with forceps to dissociate single cardiomyocytes. Afterwards Ca2<sup>+</sup> was introduced stepwise up to a concentration of 1 mM.

### Sarcomere Shortening and Ca2<sup>+</sup> Transient Measurements in Intact Ventricular Myocytes

For contractile analysis only rod-shaped myocytes without membrane blebs, hypercontractile zones, and spontaneous activity showing a stable contraction amplitude and rhythm at 1-Hz pacing frequency (4 ms long 10 V pulses) and 37◦C were recorded. Sarcomere shortening and Ca2<sup>+</sup> transients were recorded using a video-based sarcomere detection system and analyzed with the appendant software (IonWizard; IonOptix; Milton, MA) as described (Flenner et al., 2016). For Ca2<sup>+</sup> recordings, cells were loaded with 0.6µM Fura-2-AM and excited at 340 and 380 nm while the emitted light at 510 nm was recorded with a photon multiplier tube. Measurements of contraction and Ca2<sup>+</sup> transients were first performed by perfusion of the cells in basal buffer (135 mM NaCl, 4.7 mM KCl, 0.6 mM KH2PO4, 0.6 mM Na2HPO4, 1.2 mM MgSO4, 1.5 mM CaCl2, 20 mM glucose, 10 mM HEPES, pH 7.46). When the cells showed stable contraction amplitude, contractile function was recorded. Subsequently, the perfusion was switched to buffer containing different EGCg concentrations (Sigma-Aldrich, 10 nM, 100 nM, 1µM, 1.8µM, 3µM, 10µM, 30µM, 100µM for the concentration-response curve; 1.8µM for the definite measurements in KI and WT cells) and contractile function was recorded again.

### Skinned Ventricular Trabeculae Force Measurements

For the determination of force-Ca2<sup>+</sup> relationships, trabeculae were prepared from the left ventricular endocardial surface of WT and KI mice as reported before (Flenner et al., 2016). The Ca <sup>2</sup>+-sensitivity of skinned EHT strips was evaluated using a permeabilized fiber test system (1400A; Aurora Scientific). Triton X-100 permeabilized strips of the left ventricle of WT and KI mouse hearts were mounted between a force transducer and a length controller. Trabeculae were stretched above slack length until they developed force in activating solution (pCa 4.5) at 15◦C. Subsequently they were exposed to increasing Ca2<sup>+</sup> concentrations from pCa 9 to pCa 4.5 in EGTA-buffer. Force development was measured in each pCa solution. Measurements were repeated in the presence of 30 µM EGCg after 5 min preincubation in relaxing solution (Flenner et al., 2016). In every second measurement, EGCg was tested first and a control measurement was performed 5 min after EGCg washout to exclude time-dependent loss of force. Data were analyzed using the Hill equation (Hill et al., 1980), with pCa<sup>50</sup> as the free Ca2<sup>+</sup> concentration which yields 50% of the maximal force and nH representing the Hill coefficient. The pCa<sup>50</sup> represents the measure of myofilament Ca2<sup>+</sup> sensitivity.

### Statistical Analysis

Data were expressed as mean±SEM. Comparisons were performed by paired or unpaired Student's t-test (effects in intact cardiomyocytes in the absence or presence of EGCg), and with one-way ANOVA, followed by Bonferroni's post-test as indicated in the figure legends (analysis of total, Ser16- and Thr17 phosphorylated phospholamban levels in isolated cells), as indicated in the figure legends. Concentration response curves were fitted to the data points and force-pCa relationship comparison was done by using extra sum-of-squares F-test (GraphPad, Prism 6). A value of P < 0.05 was considered statistically significant.

### RESULTS

### EGCg (1.8 µM) has no Effect on Diastolic Sarcomere Length, But Shortens Relaxation Time in Isolated Cardiomyocytes

EGCg has been reported to concentration-dependently increase contractile function in rodents' cardiac myocytes and hearts (Lorenz et al., 2008; Tadano et al., 2010). HCM patients typically present with a normal or increased ejection fraction, but a diminished diastolic function and incomplete relaxation (Elliott et al., 2014). This is mimicked in cardiac myocytes from Mybpc3 KI mice, which showed lower diastolic sarcomere length and higher twitch amplitude than WT cardiomyocytes (Fraysse et al., 2012). We aimed at using an EGCG concentration that would not increase contraction amplitude. Therefore, we performed paired concentration-response curves with sarcomere shortening as the readout on isolated cardiac myocytes of Mybpc3 WT mice with increasing EGCg concentrations ranging from 10−<sup>8</sup> to 10−<sup>4</sup> M (**Figure 1A**). EGCg increased sarcomere shortening in a concentration-dependent manner (curve fit r <sup>2</sup> = 0.85). The positive inotropic effect of EGCg occurred within 5 min of exposure and was reversible by washout (loss of effect after 5 min). The highest concentration of EGCg that did not alter myocyte contractions was 1.8 µM (=10−5.74 M; **Figures 1A,B**), while EGCg concentrations above ≥3µM (=10−5.52 M) increased sarcomere shortening (**Figures 1A,C**).

We therefore tested the acute effects of 1.8 µM on isolated cardiac WT and KI myocytes. At baseline and in the absence of Fura-2, KI cardiomyocytes exhibited lower diastolic sarcomere length and longer contraction and relaxation times than WT (**Figure 2**), recapitulating a relaxation deficit seen in human patients. Application of 1.8 µM EGCg did not affect fractional sarcomere shortening and contraction time (**Figures 2B,C**). It did not further decrease the pathological diastolic sarcomere length (**Figure 2D**) but accelerated relaxation, i.e., it lowered relaxation time in both genotypes (**Figure 2E**). This effect was stronger in KI cells (**Figure 2F**).

### EGCg (1.8 µM) Increases Diastolic Ca2<sup>+</sup> and Accelerates Ca2<sup>+</sup> Transient Kinetics in Isolated Cardiomyocytes

We then investigated whether EGCg influences Ca2<sup>+</sup> homeostasis and performed Ca2<sup>+</sup> transient analysis using Fura-2 AM. Contractile parameters of Fura-2-loaded cells were also measured and evaluated, but not represented here, as Fura-2 has a substantial Ca2<sup>+</sup> buffering effect and therefore interferes with contractile processes. At baseline, KI cardiomyocytes exhibited no difference in Ca2<sup>+</sup> peak height, diastolic Ca2+, time to peak Ca2+, and time to 50% Ca2<sup>+</sup> decay compared to WT cells (**Figure 3**). Stimulation with EGCg had no influence on Ca2<sup>+</sup> peak height, but slightly increased diastolic Ca2+, and markedly reduced time to peak Ca2<sup>+</sup> and time to 50% Ca2<sup>+</sup> decay in both groups. Even though the time to 50% Ca2<sup>+</sup> decay was longer in KI cells in the presence of EGCg, the delta was not different to WT (WT −0.05 ± 0.008 s vs. KI −0.04 ± 0.0098 s, P = 0.33, Student's t-test). Plotting the sarcomere length against the F340/380 ratio of Fura-2 in WT cells showed that with 1.8µM EGCg, the descending phase of the relation between the F340/380 ratio and the sarcomere length was shifted to the right, and the relaxed state of the sarcomere length in diastole was reached at higher F340/380 ratios, resulting in smaller loops (**Figure 3F**). Opposite results, a left-shift of the loop, had been

n = 17–22, N = 5.

baseline, paired Student's t-test, n = 20–26, N = 5. For loops: n = 9.

proceed in a counter-clockwise direction. \*\*P < 0.01 vs. WT in the same condition, unpaired Student's t-test; #P < 0.05, ##P < 0.01 and ###P < 0.001 vs.

reported with the Ca2<sup>+</sup> sensitizer CGP-48506 (Wolska et al., 1996).

### EGCg (30 µM) Decreases Myofilament Ca2<sup>+</sup> Sensitivity to a Greater Extent in KI than in WT Skinned Ventricular Trabeculae

EGCg has been reported to decrease myofilament Ca2<sup>+</sup> sensitivity in three HCM models expressing either a TNNT2 or TNNI3 mutation (Tadano et al., 2010; Warren et al., 2015; Messer et al., 2016). To assess whether the EGCg effects in intact cells resulted from a decrease in myofilament Ca2<sup>+</sup> sensitivity, we measured force-pCa relationships in skinned ventricular trabeculae from WT and KI mice. At baseline and as observed before (Fraysse et al., 2012; Flenner et al., 2016), skinned KI trabeculae showed a higher pCa50than WT trabeculae, indicating higher myofilament Ca2<sup>+</sup> sensitivity (**Figures 4A–C**). We first tested a concentration of 10µM on Mybpc3 KI muscle strips but did not observe any effect (data not shown). Since other groups had reported that EGCg concentrations below 30µM had no effect on Ca2<sup>+</sup> sensitivity we also used 30µM (Tadano et al., 2010; Robinson et al., 2016). Incubation with 30µM EGCg shifted the force-Ca2<sup>+</sup> relationship to the right resulting in a lower pCa<sup>50</sup> in both genotypes (**Figures 4A–C**), which indicates myofilament Ca2<sup>+</sup> desensitization. As observed in the myocyte experiments, the effect of EGCg was stronger in skinned KI than WT muscle strips (**Figure 4D**). The nHill coefficient did not differ between the genotypes neither with nor without EGCg (**Figure 4E**).

## DISCUSSION

One major feature of HCM is a compromised diastolic function with an incomplete relaxation (Elliott et al., 2014). Increased myofilament Ca <sup>2</sup><sup>+</sup> sensitivity as seen in HCM patients and in several mouse models of HCM (Morimoto et al., 1998; Robinson et al., 2007; Huke and Knollmann, 2010; Kimura, 2010; Fraysse et al., 2012; Moore et al., 2012; van Dijk et al., 2012; Barefield et al., 2014; Flenner et al., 2016) could contribute to diastolic dysfunction. Recent findings advocate a potential therapeutic role for EGCg in HCM since it lowered the increased myofilament Ca2<sup>+</sup> sensitivity in a transgenic HCM mouse model expressing a human cardiac troponin T (TNNT2, cTnT) mutant (Tadano et al., 2010) and in HCM-associated human cardiac troponin I (TNNI3, cTnI) and cTnT mutants in a reconstituted acto-myosin system (Warren et al., 2015; Messer et al., 2016). The aim of this study was to evaluate whether EGCg has beneficial effects in another thick filament model of HCM, carrying a human mutation in the thick filament protein gene MYBPC3. The main findings of this study were: 1. At baseline and in the absence of Fura-2, KI cardiomyocytes exhibited higher fractional sarcomere shortening, lower diastolic sarcomere length, longer contraction and relaxation times than WT cells, without differences in Ca2<sup>+</sup> transient amplitude and kinetics. 2. EGCg had no effect on sarcomere shortening or diastolic sarcomere length, but it accelerated relaxation and Ca2<sup>+</sup> transient decay in Mybpc3 WT and KI cardiomyocytes. 3. EGCg induced myofilament Ca2<sup>+</sup> desensitization in permeabilized left ventricular trabeculae

isolated from Mybpc3 WT and KI mouse hearts. 4. EGCg effects on relaxation time and myofilament Ca2<sup>+</sup> sensitivity were more pronounced in KI cells and muscle strips, respectively.

EGCg is a major component of green tea and has been reported to have beneficial effects in a variety of diseases (Peng et al., 2011; Brückner et al., 2012; Ortsäter et al., 2012). Suggested mechanisms in the context of cardiovascular diseases are antioxidative, anti-inflammatory, vasorelaxant and positive inotropic effects (Chyu et al., 2004; Lorenz et al., 2004; Ludwig et al., 2004). Since HCM patients typically present with a normal or even increased ejection fraction, but with a diminished diastolic function (Elliott et al., 2014), mimicked in Mybpc3 KI mice (Fraysse et al., 2012), we intended to apply the highest concentration that would not increase sarcomere shortening (1.8µM) to evaluate its effects on the impaired relaxation in KI myocytes. Similar to Lorenz et al. and Tadano et al. (Lorenz et al., 2008; Tadano et al., 2010), who reported a positive inotropic effect between 2.5 and 5µM in mouse hearts and rat myocytes, we observed an increase in fractional sarcomere shortening at concentrations above 3µM. Although we observed no effect on diastolic sarcomere length in WT cells at 1.8µM EGCg, we speculated that this concentration could have an effect in KI cells. This was not the case. Nevertheless, 1.8 µM EGCg reduced relaxation time, and this effect was more prominent in KI cells, whereas the effects on diastolic Ca2<sup>+</sup> and Ca2<sup>+</sup> kinetics did not differ between the genotypes. The increase in diastolic Ca2<sup>+</sup> could be explained by an inhibiting effect on the Na+/Ca2<sup>+</sup> exchanger (NCX), as reported with 10 nM (Feng et al., 2012) and 2.5µM (Lorenz et al., 2008). The faster time to peak Ca2<sup>+</sup> is probably due to EGCg effects on the ryanodine receptor. Indeed, it has been shown that EGCg activates the ryanodine receptor at 10 nM in sarcoplasmic reticulum (SR) vesicles isolated from rabbit left ventricles (Feng et al., 2012) and in the range of 1 nM–20µM in junctional SR vesicles isolated from skeletal muscle (Najafi et al., 2015). At a concentration of 1.8µM, EGCg had no effect on PLB Ser16/Thr17 phosphorylation (Supplemental Figure 1). This supports findings of Lorenz et al., who neither observed an effect on PLB Ser16/Thr17 phosphorylation with EGCg (4µM) nor an influence on EGCg actions on contractility after β1-adrenoceptor inhibition with 3µM metoprolol (Lorenz et al., 2008). We thus exclude that the acceleration of relaxation and Ca <sup>2</sup><sup>+</sup> kinetics are mediated via the β1-adrenergic pathway. In contrast to Lorenz et al. we did not observe an increased Ca2<sup>+</sup> peak height in the presence of 1.8µM EGCg (Lorenz et al., 2008). We therefore assume that at this concentration EGCg does not increase sarcoplasmic reticulum (SR) Ca2<sup>+</sup> load. Data from canine SR vesicles and HEK293 cell microsomes show that EGCg concentrations > 4.8µM directly inhibit SR calcium ATPase (Kargacin et al., 2011). Additionally, Feng et al. reported no effect on SERCA in rabbit cardiac and skeletal SR membranes with 1–2µM EGCg (Feng et al., 2012; Najafi et al., 2015). We therefore rule out that the acceleration of relaxation kinetics in the presence of 1.8µM EGCg is due to an increased SERCA activity. A plausible contributing reason could be an EGCg-mediated decrease in myofilament Ca2<sup>+</sup> sensitivity.

Indeed, it has been reported that EGCg lowered the increased myofilament Ca2<sup>+</sup> sensitivity and improved the diastolic

function of isolated working heart preparations from a transgenic HCM mouse model expressing a human TNNT2 mutation (Tadano et al., 2010). EGCg also restored the increased Ca2<sup>+</sup> sensitivity of HCM-associated human cTnI and cTnT mutants in a reconstituted acto-myosin system (Warren et al., 2015; Messer et al., 2016). Similar to these previous data, 30 µM EGCg decreased Ca2<sup>+</sup> sensitivity in our thick myofilament mouse model that carries a frequent HCM mutation in the most frequently mutated gene (Vignier et al., 2009; Fraysse et al., 2012; Ho et al., 2015). It has been suggested that EGCg binding to the C-terminal region of cardiac troponin C (cTnC) alters the interaction between cTnC and cTnI and therefore the sensitivity of the myofilaments to Ca2<sup>+</sup> (Liou et al., 2008; Robertson et al., 2009). In both intact myocytes and skinned trabeculae, EGCg had a more profound effect on cells and strips of the KI genotype. This could also be related to the longer baseline relaxation time and elevated myofilament Ca2<sup>+</sup> sensitivity in KI mice. This is similar to a recent study in which we showed that the myofilament Ca2+-desensitizing effect of ranolazine was only present in KI, but not in WT muscle strips (Flenner et al., 2016). Since EGCg is not a pure Ca2<sup>+</sup> desensitizer (Stangl et al., 2007), side effects such as arrhythmias or blood pressure lowering reported in in vivo applications would also be expected in WT mice (Alvarez et al., 2006; Bao et al., 2015).

The study has two limitations. (1) The Mybpc3 KI model shows many HCM features only at the homozygous state. Additionally, Mybpc3 KI mice present a reduced ejection fraction. These two points are in contrast to the more common findings in HCM patients who present left ventricular hypertrophy, interstitial fibrosis, and diastolic dysfunction with heterozygous mutation states and normal or even supra-normal ejection fraction. (2) The Ca2+-desensitizing effect of EGCg on the permeabilized trabeculae was only detected at 30µM, which is higher than the concentrations used in the intact myocyte experiments. This discrepancy has also been observed before (Tadano et al., 2010; Robinson et al., 2016). It should be kept in mind that in permeabilized cardiac muscle fibers the effective concentrations of the Ca2+-sensitizers pimobendan and EMD57033 were also reported to be much higher than those estimated in vivo (Fujino et al., 1988; Solaro et al., 1993; Chu et al., 1999), proposing that drugs could have lower potency in permeabilized cardiac muscle preparations than in vivo, since important components enhancing drug uptake or function such as membrane transporters or the SR could be disturbed in their function or missing after muscle strip skinning.

### CONCLUSION

EGCg accelerated relaxation and Ca2<sup>+</sup> transient decay in Mybpc3 WT and KI cardiomyocytes, which seems to be partly due to Ca2<sup>+</sup> desensitization of the myofilaments. We show for the first time that EGCg is also effective in a thick filament mutation mouse model. In support of other studies (Tadano et al., 2010; Warren et al., 2015; Messer et al., 2016), this confirms that EGCg belongs to a new class of Ca2<sup>+</sup> antagonists which have

### REFERENCES


a very favorable functional profile acting directly on the Ca2<sup>+</sup> regulatory system of cTnC. Further efforts should be made to develop a new generation of Ca2<sup>+</sup> desensitizers based on the green tea catechin EGCg with even more potency and specificity to avoid potential side effects such as arrhythmias or lowering of blood pressure that could also affect healthy individuals.

### AUTHOR CONTRIBUTIONS

FWF: conception and design of research, management of the mouse cohorts, execution of experiments, analysis, and interpretation of data, figure preparation, drafting of the manuscript. FF: isolation and treatment of cardiac myocytes, execution of experiments, interpretation of data, figure preparation, discussion of the manuscript draft. MN: isolation and treatment of cardiac myocytes, execution of experiments. TE: interpretation of data, discussion of the manuscript draft. LC: conception and design of research, analysis, and interpretation of data, drafting of the manuscript. All authors critically discussed the results, and reviewed and approved the manuscript before submission.

### FUNDING

This work was supported by the DZHK (German Centre for Cardiovascular Research) and the Deutsche Stiftung für Herzforschung (F/28/12).

### ACKNOWLEDGMENTS

We thank Giulia Mearini and Marc Hirt for discussion.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2016.00607/full#supplementary-material


versus established atherosclerosis in apolipoprotein E-null mice. Circulation 109, 2448–2453. doi: 10.1161/01.CIR.0000128034.70732.C2


reticulum/endoplasmic reticulum Ca2+ ATPase. J. Muscle Res. Cell Motil. 32, 89–98. doi: 10.1007/s10974-011-9256-7


of mutations, and implications for a molecular diagnosis strategy. Circulation 107, 2227–2232. doi: 10.1161/01.CIR.0000066323.15244.54


downregulated in a mouse model of hypertrophic cardiomyopathy and targets desmin for proteasomal degradation. J. Mol. Cell. Cardiol. 87, 214–224. doi: 10.1016/j.yjmcc.2015.08.020


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Friedrich, Flenner, Nasib, Eschenhagen and Carrier. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Myofilament Calcium Sensitivity: Mechanistic Insight into TnI Ser-23/24 and Ser-150 Phosphorylation Integration

Hussam E. Salhi † , Nathan C. Hassel † , Jalal K. Siddiqui, Elizabeth A. Brundage, Mark T. Ziolo, Paul M. L. Janssen, Jonathan P. Davis and Brandon J. Biesiadecki\*

*Department of Physiology and Cell Biology and Davis Heart and Lung Research Institute, Ohio State University, Columbus, OH, USA*

#### Edited by:

*P. Bryant Chase, Florida State University, USA*

#### Reviewed by:

*John Peter Konhilas, University of Arizona, USA Margaret Westfall, University of Michigan, USA Jennifer Davis, University of Washington, USA*

> \*Correspondence: *Brandon J. Biesiadecki biesiadecki.1@osu.edu*

*† These authors have contributed equally to this work.*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *12 August 2016* Accepted: *08 November 2016* Published: *15 December 2016*

#### Citation:

*Salhi HE, Hassel NC, Siddiqui JK, Brundage EA, Ziolo MT, Janssen PML, Davis JP and Biesiadecki BJ (2016) Myofilament Calcium Sensitivity: Mechanistic Insight into TnI Ser-23/24 and Ser-150 Phosphorylation Integration. Front. Physiol. 7:567. doi: 10.3389/fphys.2016.00567* Troponin I (TnI) is a major regulator of cardiac muscle contraction and relaxation. During physiological and pathological stress, TnI is differentially phosphorylated at multiple residues through different signaling pathways to match cardiac function to demand. The combination of these TnI phosphorylations can exhibit an expected or unexpected functional integration, whereby the function of two phosphorylations are different than that predicted from the combined function of each individual phosphorylation alone. We have shown that TnI Ser-23/24 and Ser-150 phosphorylation exhibit functional integration and are simultaneously increased in response to cardiac stress. In the current study, we investigated the functional integration of TnI Ser-23/24 and Ser-150 to alter cardiac contraction. We hypothesized that Ser-23/24 and Ser-150 phosphorylation each utilize distinct molecular mechanisms to alter the TnI binding affinity within the thin filament. Mathematical modeling predicts that Ser-23/24 and Ser-150 phosphorylation affect different TnI affinities within the thin filament to distinctly alter the Ca2+-binding properties of troponin. Protein binding experiments validate this assertion by demonstrating pseudo-phosphorylated Ser-150 decreases the affinity of isolated TnI for actin, whereas Ser-23/24 pseudo-phosphorylation is not different from unphosphorylated. Thus, our data supports that TnI Ser-23/24 affects TnI-TnC binding, while Ser-150 phosphorylation alters TnI-actin binding. By measuring force development in troponin-exchanged skinned myocytes, we demonstrate that the Ca2<sup>+</sup> sensitivity of force is directly related to the amount of phosphate present on TnI. Furthermore, we demonstrate that Ser-150 pseudo-phosphorylation blunts Ser-23/24-mediated decreased Ca2+-sensitive force development whether on the same or different TnI molecule. Therefore, TnI phosphorylations can integrate across troponins along the myofilament. These data demonstrate that TnI Ser-23/24 and Ser-150 phosphorylation regulates muscle contraction in part by modulating different TnI interactions in the thin filament and it is the combination of these differential mechanisms that provides understanding of their functional integration.

Keywords: cardiac troponin I, phosphorylation, calcium sensitivity, functional integration

## INTRODUCTION

Contraction and relaxation of the heart is fundamentally dependent on the conversion of the dynamic rise and fall in intracellular cytosolic Ca2<sup>+</sup> into force production (Kobayashi and Solaro, 2005). The binding of Ca2<sup>+</sup> to troponin (Tn) activates the myofilament allowing the interaction of myosin with actin, the generation of force and contraction. Calcium regulation of the myofilament through this binding event is not a simple "on and off " switch (Solzin et al., 2007; Davis and Tikunova, 2008; Biesiadecki and Davis, 2014; Davis et al., 2016). Rather, activation and deactivation are both active processes that involve a number of dynamic and complex protein-protein interactions (Manning et al., 2011). Each of these dynamic interactions have the potential to modulate similar and/or different myofilament contraction and relaxation functions (Biesiadecki et al., 2014; Chung et al., 2016; Janssen et al., 2016). Thus, the myofilament response to Ca2<sup>+</sup> can be intricately manipulated through Tn to regulate cardiac function and improve resulting in an improved outcome in cardiac disease (Li et al., 2010; Alves et al., 2014; Shettigar et al., 2016).

The phosphorylation of troponin I (TnI) represents a key mechanism in modulating the myofilament response to Ca2<sup>+</sup> (Solaro et al., 1976; Solaro and Kobayashi, 2011; Liu et al., 2014; Nixon et al., 2014). One of the most physiologically and pathologically relevant Tn phosphorylations is the adrenergic mediated protein kinase A phosphorylation of TnI at serines 23 and 24 (Ser-23/24). TnI Ser-23/24 phosphorylation contributes significantly to Ca2<sup>+</sup> regulation of the myofilament, resulting in decreased Ca2+-sensitive force production and accelerated myofilament relaxation (Kranias and Solaro, 1982; de Tombe and Stienen, 1995; Herron et al., 2001; Kentish et al., 2001; Sakthivel et al., 2005; Biesiadecki et al., 2007; Ramirez-Correa et al., 2010). In addition to Ser-23/24, TnI can undergo phosphorylation on at least 12 additional residues as the end result of different signaling pathways (Zhang et al., 2012). A large subset of literature has described the effects many of these phosphorylations impart on cardiac function through altering the Ca2<sup>+</sup> sensitive regulation of force production in isolation. Although a single phosphorylation is sufficient to alter cardiac function, the heart simultaneously contains multiple TnI phosphorylations that can be independently altered in response to physiological and pathological stress (Pi et al., 2003; Nixon et al., 2014; Lang et al., 2015). Therefore, it is not the isolated effect of a single TnI phosphorylation that is responsible for overall cardiac regulation, but rather the simultaneous presence of multiple TnI phosphorylations that combine to exhibit expected or unexpected functional integration. The significance of these multiple phosphorylations results in a functional integration as demonstrated by the finding that the contractile effects of TnI Ser-23/24 phosphorylation are dependent on the myofilament phosphorylation background (Biesiadecki et al., 2007; Kooij et al., 2010; Nixon et al., 2012; Salhi et al., 2014; Lang et al., 2015). The mechanisms underlying how combined TnI phosphorylations integrate to regulate the myofilament response to Ca2<sup>+</sup> remain poorly understood.

The integrated function of multiple TnI phosphorylations is significant in cardiac disease (Zhang et al., 2012). We have demonstrated that the phosphorylation of TnI at Ser-23/24 and Ser-150 are both significantly increased in response to ischemic conditions following a myocardial infarction (Nixon et al., 2014). At first glance this seems paradoxical in that Ser-23/24 phosphorylation decreases while Ser-150 phosphorylation increases Ca2+-sensitivity (Nixon et al., 2012; Oliveira et al., 2012). However, the combination of Ser-23/24 and Ser-150 phosphorylation exhibit functional integration by differentially regulating cardiac function from that of either phosphorylation alone, retaining contractile force (normal Ca2<sup>+</sup> sensitivity) but accelerating relaxation (accelerated Ca2<sup>+</sup> dissociation) (Nixon et al., 2012, 2014). This and others' findings demonstrate the significance of TnI phosphorylation integration to cardiac function during disease (Kooij et al., 2010; Boontje et al., 2011; Taglieri et al., 2011; Salhi et al., 2014). Ultimately, elucidating the different molecular mechanisms utilized during simultaneous TnI phosphorylation will provide insight into the regulatory protein interactions that can be therapeutically targeted, such as by altered phosphorylation, to modulate contraction and relaxation in cardiac disease.

In the current study, we sought to investigate the molecular mechanisms utilized by Ser-23/24 and Ser-150 phosphorylation to modulate contraction both individually and when integrated. Based on our previous work, we hypothesized that Ser-23/24 and Ser-150 differentially alter the binding affinities of TnI within the thin filament to elicit functional integration. By employing mathematical modeling and protein-protein binding experiments, we identified that Ser-23/24 or Ser-150 pseudophosphorylation each modulate different TnI binding affinities in the thin filament. TnI Ser-150 pseudo-phosphorylation decreases the affinity of TnI for actin, which is unaffected by Ser-23/24 pseudo-phosphorylation. Through force-Ca2<sup>+</sup> measurements in Tn-exchanged skinned myocytes we demonstrate that the magnitude of pseudo-phosphorylation-dependent Ca2<sup>+</sup> sensitive force development is directly related to the amount of TnI pseudo-phosphorylation present at Ser-23/24 or Ser-150. We further demonstrate that these pseudo-phosphorylations function similarly regardless of whether they are present on the same or different TnI molecules. These data demonstrate that TnI Ser-23/24 and Ser-150 phosphorylations affect distinct TnI interactions in the thin filament to result in differential regulation of the steady-state and kinetics of contraction.

### MATERIALS AND METHODS

### Mathematical Model

The thin filament interactions responsible for the steady-state TnC Ca2<sup>+</sup> binding and kinetic Ca2<sup>+</sup> dissociation from TnC for each phosphorylated TnI were determined by solving the mathematical model described by Siddiqui et al. (2016). Briefly, this model describes Ca2<sup>+</sup> mediated thin filament regulation based on six biochemical states of TnC. The TnC states are described by a series of first order differential equations. Reaction

**Abbreviations:** Calcium, Ca2+; Tn, troponin; TnI, troponin I; WT, wild-type.

rate constants for each TnI phosphorylation were solved using Scilab computation to approximate the previously determined Tn and thin filament biochemical determinants (Ca2+-sensitivity and Ca2<sup>+</sup> dissociation Nixon et al., 2012, 2014) for each TnI phosphorylation.

### cDNA Constructs

All cardiac TnI amino acid residue numbers presented in this manuscript are given according to the human sequence including the first methionine. Site-directed mutagenesis (QuickChange Lightning, Agilent, Santa Clara, CA) was conducted according to manufacturer's instructions to generate cDNA constructs encoding human TnI pseudo-phosphorylations: Ser-150 to Asp (S150D), Ser-23/24 to Asp (S23/24D), Ser-23/24 to Asp with Ser-150 to Asp (S23/24/150D). All resultant constructs were verified by DNA sequencing.

### Protein Expression and Purification

Plasmids encoding the individual recombinant human cardiac Tn subunits were transformed and expressed in Escherichia coli and purified to homogeneity as previously described (Sumandea et al., 2003; Kobayashi et al., 2005; Kobayashi and Solaro, 2006; Nixon et al., 2012). Cardiac Tn complexes were prepared and reconstituted by sequential dialysis as previously described (Kobayashi and Solaro, 2006; Biesiadecki et al., 2007; Nixon et al., 2014).

### Solid-Phase Protein Binding

ELISA-based solid-phase protein binding assays were conducted as previously described to determine the effect of phosphorylation to alter TnI binding to actin compared to that of non-phosphorylated TnI (Biesiadecki and Jin, 2011). Briefly, 2µM F-actin dissolved in Buffer A (in mM: 150 KCl, 3 MgCl2, 10 MOPS, pH 7.0) was used to coat a 96-well microtiter plate in 100µL/well at 4◦C overnight. Unbound actin was washed with Buffer T (Buffer A containing 0.1% Tween-20). Following washes, the wells were blocked with Buffer A containing 1% BSA. Following removal of blocking solution, serial dilutions of TnI WT or phosphomimetics (S23/24D, S150D, or S23/24/150D) were incubated in 100µL/well for 2 h at room temperature. The wells were then washed and bound TnI was quantified by ELISA using a mouse anti-cardiac TnI antibody (Fitzgerald; clone C5) and appropriate HRP-conjugated secondary antibody. Following the addition of H2O2-ABTS (2,2′ -azino-bis(3 ethylbenzthiazoline-6-sulphonic acid) substrate solution, the absorbance at 405 nm was monitored over the linear course of color development. Protein binding assays were conducted in triplicate wells and repeated. Each assay contained wells that were coated with F-actin but incubated with buffer in the absence of TnI as a negative control.

### Myocyte Force Production

All animal protocols and procedures were performed in accordance with National Institutes of Health guidelines and approved by the Institutional Laboratory Animal Care and Use Committee at The Ohio State University. Steady-state Ca2+ sensitive force development was measured in Tn-exchange permeabilized rat myocytes as described previously (Salhi et al., 2014). Briefly, following mechanical isolation from frozen rat ventricular tissue, cardiac myocyte preparations were skinned by resuspension in relaxing solution (in mM; 97.92 KOH, 6.24 ATP, 10 EGTA, 10 Na2CrP, 47.58 Kprop, 6.54 MgCl2, 100 BES, pH 7.0) containing 1% peroxide-free Triton X-100 (Anapoe-X-100, Anatrace, Maumee, OH) with incubation at room temperature for 10 min rocking. Following skinning, myocytes were centrifuged and immediately resuspended for Tn exchange.

Exchange of exogenous recombinant human cardiac Tn into skinned rat myocytes was performed as described previously by incubating myocytes in exchange buffer (in mM; 200 KCl, 5 MgCl2, 1 DTT, 5 EGTA, 20 MOPS, pH 6.5) containing 13 uM column purified Tn overnight at 4◦C (Sumandea et al., 2003; Biesiadecki et al., 2007; Nixon et al., 2012). Exchange with Tn WT, Tn S23/24D, Tn S150D, and Tn S23/24/150D groups was conducted by incubating skinned myocytes overnight in exchange solution containing the single purified Tn complex indicated. Exchange with Tn WT+S23/24D, Tn WT+S150D, and Tn S23/24D+S150D groups was conducted by incubating skinned myocytes overnight in exchange solution containing an equal molar ratio of the two purified indicated Tn complexes to a final total Tn of 13 uM. Calcium regulated force development in skinned Tn exchanged rat ventricular myocyte preparations was performed similar to that previously described (Salhi et al., 2014). Briefly, Tn exchanged myocytes were attached to two micro-needles using aquarium sealant (Marineland, Noblesville, IN) and sarcomere length was adjusted to 2.2 um by visualization on a calibrated monitor. The perfusion pipette of a constant perfusion control system (VC-8M Eight Channel Mini-Valve Perfusion System, Warner Instruments, Hamden, CT) was then placed close to the myocyte such that the outflow perfused the entire myocyte. Experiments were performed by flowing a series of activating mixtures consisting of relaxing and activating solution over the myocyte. Activating solution was identical in composition to relaxing solution but containing varied free Ca2<sup>+</sup> concentration (pCa 10.0 to 4.5) (Fabiato and Fabiato, 1979). Developed force was measured at each activating perfusion followed by perfusion with relaxing solution. The activating developed force was subtracted from the subsequent relax perfused force measure. Time-dependent force rundown was determined by comparison of the first developed maximal force to the force developed upon final maximal activation at the end of the experiment. Any cell exhibiting greater than 20% force rundown was discarded. Cell cross-sectional area was calculated after the force experiment on a calibrated monitor as previously described (Salhi et al., 2014). Force-pCa curves were fit using a modified Hill-equation (Biesiadecki et al., 2007; Nixon et al., 2012). Experiments were conducted at room temperature. Data were acquired with custom-made LabView software and analyzed using Igor Pro.

### Protein Electrophoresis and Western Blot

Skinned myocytes were solubilized in denaturing buffer (2% SDS, 0.1% bromophenol blue, 10% glycerol and 50 mM Tris-HCl, pH 6.8), heated for 5 min at 80◦C and clarified by centrifugation for 5 min. SDS-PAGE and Western blot were carried similar to that previously described (Biesiadecki et al., 2010; Liu et al., 2012; Nixon et al., 2012). Briefly, proteins were separated on 12% (29:1) Laemmli gel and transferred to PVDF. Following blocking with 1% BSA in TBS resultant membranes were incubated with a mouse anticardiac TnI antibody (Fitzgerald; clone C5). Following washes, membranes were incubated with a Dylight labeled fluorescent secondary antibody (Jackson ImmunoResearch Laboratories, Inc, West Grove, PA) and visualized on a Typhoon 9410 imager (GE Healthcare, Piscataway, NJ). Differential migration of endogenous mouse TnI versus exogenous exchanged human TnI allowed for quantification by densitometry analysis. In groups where two different Tn complexes were exchanged the amount of total exogenous incorporation was determined and the exchange of the two complexes considered similar.

### Data Processing and Statistical Analysis

Data are presented as mean ± the standard error of the mean. Protein binding and developed force was plotted against TnI concentration or Ca2+, respectively, and fit with a logistic sigmoid function mathematically equivalent to the Hill equation to determine 50% maximal binding or force as previously described. Results of Tn exchange and force development experiments were compared by One-way ANOVA with Tukey's post-hoc test. p < 0.05 was considered statistically significant.

### RESULTS

### The Functional Integration of TnI Ser-23/24 and Ser-150 Phosphorylations Is Determined by Their Differential Protein Interactions within the Thin Filament

Our previous studies demonstrate functional integration of TnI phosphorylations exhibit different contractile function than predicted from the effect of each phosphorylation in isolation (Nixon et al., 2012, 2014; Salhi et al., 2014). Functional integration of Ser-150 with Ser-23/24 phosphorylation restores WT-like Ca2+-sensitivity (normal force) while maintaining accelerated thin filament Ca2<sup>+</sup> dissociation (accelerated relaxation) (Nixon et al., 2012, 2014). We first investigated if altered thin filament protein interactions within a single regulatory unit were sufficient to explain this functional integration. Toward this end we employed a mathematical model based on several biochemical states of TnC (apo or bound to Ca2+, Mg2+, and/or TnI) recently developed by the Davis laboratory to identify mechanisms involved in TnI phosphorylation integration (Siddiqui et al., 2016). Changes in the hill coefficient and cooperativity are not incorporated into the model. This model identified that Ser-23/24 and Ser-150 phosphorylations differentially modulate regulation of contraction by altering the affinity of different TnI thin filament interactions. By simply accelerating the dissociation of TnI from TnC (**Figure 1A**), the model recapitulates the biochemical effects of TnI Ser-23/24 phosphorylation on steady-state Ca2<sup>+</sup> binding and Ca2<sup>+</sup> dissociation in thin filament and the isolated Tn complex (**Figures 1B,C**). Our model further asserts that Ser-150 phosphorylation utilizes a separate mechanism by slowing the dissociation of Ca2<sup>+</sup> from TnC-TnI and increasing the availability of TnI for TnC (**Figure 1A**). Making only these two alterations are sufficient to accurately model the increase in Ca2<sup>+</sup> sensitivity and decrease in Ca2<sup>+</sup> dissociation induced by Ser-150 phosphorylation in both the Tn and thin filament state (**Figures 1B,C**). Based upon these results we hypothesized that modulation of these separate TnI binding affinities is sufficient to reconcile TnI Ser-23/24 and Ser-150 phosphorylation functional integration. Indeed, the model demonstrates that combination of these two phosphorylation mechanisms is sufficient to recapitulate the Ser-150 induced blunting of Ser-23/24 desensitization while maintaining accelerated thin filament Ca2<sup>+</sup> dissociation in both isolated Tn and the thin filament (**Figures 1B,C**), similar to our previously published data for that of Tn S23/24/150D (Nixon et al., 2014, 2012).

TnI availability in the model is dependent on several interactions TnI may have in the thin filament and allows us to predict the molecular mechanisms responsible for Tn mediated modification alteration of thin filament regulation. To mechanistically understand the bases of TnI phosphorylation effects predicted by the model, we employed a solid-phase protein binding assay to measure the affinity of cardiac TnI for actin under different pseudo-phosphorylated states. Protein binding experiments demonstrate that while purified cardiac Ser-23/24 pseudo-phosphorylated TnI does not alter the actin binding affinity, Ser-150 pseudo-phosphorylation decreases TnI-actin binding by 3.3-fold (TnI concentration required for 50% maximum binding to actin: TnI WT = 16.0 ± 0.5 nM, TnI S23/24D = 15.0 ± 0.4 nM, TnI S150D = 52.9 ± 2 nM; TnI WT vs. TnI S150D = p < 0.05) (**Figure 2**). According to the model, we hypothesized that the Ser-150-mediated decrease in TnI actin affinity would be maintained upon integration with Ser-23/24 phosphorylation. Our protein binding data confirms that TnI S23/24/150D still exhibits a 2.6-fold decrease in actin binding affinity compared to WT TnI (TnI concentration required for 50% maximum binding to actin: TnI S23/24/150D = 41.2 ± 0.7 nM; TnI WT vs. TnI S23/24/150D = p < 0.05) (**Figure 2**). This decreased affinity of TnI Ser-150 pseudophosphorylation, but not Ser-23/24 pseudo-phosphorylation for actin demonstrates that Ser-23/24 and Ser-150 phosphorylation can differentially modulate the binding of TnI to actin, validating model predictions that Ser-23/24/150 functional integration is at least in part based on different molecular mechanisms of the two phosphorylations to modulate TnI interactions within the thin filament.

### Functional Integration of TnI Ser-23/24 and Ser-150 Pseudo-Phosphorylation in Rat Myocytes

Several studies have demonstrated that Ser-23/24 phosphorylation decreases Ca2+-sensitivity of skinned cardiac muscle preparations (Robertson et al., 1982; Zhang et al., 1995; Biesiadecki et al., 2007). Previous work from our lab demonstrated that Ser-150 phosphorylation increases Ca2+-sensitivity of force development in isolation and its

functional integration blunts Ser-23/24 phosphorylation mediated desensitization following exchange into skinned cardiac trabeculae (Nixon et al., 2012, 2014). Previously we demonstrated the pseudo-phosphorylation of TnI at both Ser-23/24 and Ser-150 by mutation to Asp alters Ca2<sup>+</sup> dependent force identical to that of actual phosphate at these sites (Biesiadecki et al., 2007; Nixon et al., 2012). To investigate the functional integration of Ser-23/24 and Ser-150 phosphorylation in the isolated skinned rat myocyte system, we measured Ca2<sup>+</sup> dependent force development in left ventricular rat skinned myocytes exchanged with human cardiac Tn containing either WT TnI (Tn WT), S23/24 pseudo-phosphorylated TnI (Tn S23/24D), S150 pseudo-phosphorylated TnI (Tn S150D) or S23/24/150 pseudo-phosphorylated TnI (Tn S23/24/150D). Following exchange, the percent of exogenous Tn incorporated into all measured myocyte preparations was determined by Western blot, as the exogenous human cardiac TnI migrates faster by SDS-PAGE than endogenous rat cardiac TnI. Exchange quantification demonstrated an average of 57% exogenous Tn incorporation that was not different for any of the different Tn's exchanged (% exchange: rat Tn WT = 60.5 ± 7.5%, human WT Tn = 55.3 ± 1.8%, Tn S23/24D = 57.0 ± 5.0%, Tn S150D = 58.0 ± 4.5%, Tn S23/24/150D = 62.0 ± 3.8%; p > 0.05) (**Figure 3**). Furthermore, the exchange of human Tn WT did not alter Ca2+-sensitivity compared to cells exchanged with Tn containing rat WT cardiac TnI, rat cardiac Myc tagged WT TnT

and rat WT TnC (RcTn WT) (**Figure 4** and **Table 1**). Consistent with previous findings, exchange with Tn S23/24D decreased Ca2+-sensitivity of force development, while exchange with Tn S150D increased Ca2+-sensitivity compared to Tn WT (**Figure 4** and **Table 1**) (Nixon et al., 2012). The Ca2+-sensitivity of cells exchanged with Tn S23/24/150D was not different from Tn WT, consistent with previous findings that S150D blunts Ser-23/24 mediated desensitization (**Figure 4** and **Table 1**). These data demonstrate that Ser-23/24 and Ser-150 pseudo-phosphorylation impart different effects on Ca2+-sensitive force development in the skinned rat myocyte system, consistent with previous in vitro and trabeculae data (Nixon et al., 2012, 2014). Additionally, Ser-150 pseudo-phosphorylation is able to blunt the Ser-23/24 mediated desensitization when these phosphorylations occur on the same TnI molecule in the skinned rat myocyte.

### Ca2<sup>+</sup> Sensitive Force Development Is Dependent On the Amount of TnI Phosphorylation Present

The functional effects of integrated TnI phosphorylations may be complexly related to the amount of TnI phosphorylation present. To determine if the relationship between Ca2+-sensitivity and the amount of phosphorylation present is linear for both TnI Ser-23/24 and Ser-150, we measured the Ca2+-dependent force development in left ventricular rat skinned myocytes. Myocytes

conducted in triplicate wells and repeated.

were exchanged with a mixture of 50% human cardiac Tn WT and either 50% Tn S23/24D (Tn WT+S23/24D) or 50% Tn S150D (Tn WT+S150D). Western blot for cardiac TnI demonstrated the incorporation of 51.7 ± 3.3% WT+S23/24D and 56.5 ± 6.9% WT+S150D exogenous Tn (p > 0.05) (**Figure 3**). While we cannot directly determine the percent exchange of each exogenous Tn complex in the mixture, we assume similar exchange as we observed for Tn S23/24D and Tn S150D exchange (**Figure 3**). Following exchange with Tn WT+S23/24D we observed half the decrease in pCa<sup>50</sup> observed for Tn S23/24D exchange compared to Tn WT exchange (change in pCa<sup>50</sup> from Tn WT: Tn S23/24D = 0.11, WT+S23/24D = 0.05) (**Figure 5** and **Table 1)**. We observed similar results following exchange with Tn WT+S150D, in which the myofilament was only half as sensitized compared to exchange with Tn S150D alone (change in pCa<sup>50</sup> from Tn WT: Tn S150D = 0.17, Tn WT+S150D = 0.1; p < 0.05) (**Figure 5** and **Table 1**). These data demonstrate that the magnitude of phosphorylation dependent change in Ca2+-sensitive force development is related to the amount of TnI phosphate present at the TnI Ser-23/24 or Ser-150 residues.

### TnI Ser-23/24 and Ser-150 Pseudo-Phosphorylation Exhibit Functional Integration When On Different TnI Molecules

The overwhelming majority of studies to date have primarily determined the combined function of multiple phosphorylations on the same TnI molecule. We investigated if the different molecular mechanisms utilized by Ser-23/24 and Ser-150 phosphorylation integrate when occurring on different Tn molecules. To this end, we exchanged cells with Tn containing 50% TnI S23/24D and 50% TnI S150D (Tn S23/24D+S150D) that resulted in an average of 56.8 ± 7.5 percent total exogenous Tn exchange. Thus, the Tn S23/24D+S150D exchange group is expected to result in cells containing ∼25% TnI S23/24D, ∼25% S150D and ∼50% endogenous WT Tn in the myofilament. Calcium sensitivity of cells exchanged with Tn S23/24D+S150D was not different from those exchanged with Tn S23/24/150D, even though the pseudo-phosphorylations occurred on separate Tn molecules (**Figure 6** and **Table 1**). Thus, these data demonstrate that pseudo-phosphorylation of TnI molecules at Ser-150 integrate to blunt the Ca2<sup>+</sup> desensitizing effects of Ser-23/24 pseudo-phosphorylation even when these two phosphorylated residues are present on different TnI molecules. Therefore, the molecular mechanisms responsible for phosphorylation functional integration to differentially affect heart function can integrate across different Tn's along the myofilament.

### DISCUSSION

The phosphorylation of different TnI residues is altered by physiological and pathological stress and therefore the combined function of multiple TnI phosphorylations plays a critical role to differentially modulate cardiac function in both the normal and diseases states. Our current study aimed to elucidate underlying mechanisms utilized by Ser-23/24 and Ser-150 phosphorylation. Our findings demonstrate that: (1) The integrated function of TnI Ser-23/24 and Ser-150 phosphorylation is dependent in part upon their regulation of different TnI protein-protein interactions within the thin filament. TnI Ser-23/24 phosphorylation alters the dissociation of the TnI C-terminus from TnC, while TnI Ser-150 phosphorylation alters binding of the TnI Cterminus to actin and slows the dissociation of Ca2<sup>+</sup> from TnC in the presence of TnI. (2) TnI Ser-23/24 or Ser-150 phosphorylation occupancy linearly alters Ca2+-sensitive force development. (3) The mechanisms utilized by Ser-23/24 and Ser-150 phosphorylation integration to affect Ca2+-sensitive force development occur both when on the same and separate Tn molecules along the myofilament.

### Regulatory Mechanisms of TnI Ser-23/24 and Ser-150 Phosphorylation

TnI Ser-23/24 phosphorylation decreases Ca2+-sensitive force production (**Figure 4**) (Solaro et al., 1976; Kranias and Solaro, 1982; Zhang et al., 1995; Nixon et al., 2012). The cardiac TnI N-terminal region containing the Ser-23/24 residues interacts with the N-lobe of TnC (Abbott et al., 2001). The structural basis for Ser-23/24 phosphorylation to decrease Ca2+-sensitivity is proposed to be through alteration of this interaction between the TnI N-terminus and TnC. Peptide binding studies have supported this hypothesis by demonstrating a decreased binding affinity of TnI N-terminal peptides that contain phosphorylated Ser-23/24 to TnC (Ferrières et al., 2000). NMR studies also complement these findings, demonstrating the TnI N-terminus makes contacts with the N-terminal domain of TnC near the regulatory Ca2+-binding site (Finley et al., 1999; Gaponenko et al., 1999; Abbott et al., 2000). These contacts are altered

when Ser-23/24 are substituted with phosphomimetics (Finley et al., 1999; Ward et al., 2004; Hwang et al., 2014). The mechanisms by which alteration of TnI N-terminus binding to TnC is propagated through the Tn complex to affect Ca2+ sensitivity remain unclear. Current data supports that the Ser-23/24 phosphorylation can act either by directly destabilizing Ca2<sup>+</sup> binding to the TnC regulatory site or indirectly by modulating binding affinities of the TnI C-terminal region to TnC or actin (switch peptide or inhibitory peptide binding, respectively). Data from our model and binding studies suggest that Ser-23/24 phosphorylation acts, at least in part, by altering the TnI-TnC interaction to increase the dissociation of TnI from TnC-Ca2<sup>+</sup> (**Figure 1**). Altering the dissociation of TnI from TnC-Ca2<sup>+</sup> is sufficient to recapitulate Ser-23/24 phosphorylation biochemical data and is consistent with a proposed intramolecular interaction between TnI subdomains, whereby Ser-23/24 phosphorylation places the N-terminus in a conformation to interact with the basic residues on the C-terminal TnI regulatory region near the switch peptide (Howarth et al., 2007; Warren et al., 2009).

TnI Ser-150 phosphorylation increases Ca2+-sensitive force production (Buscemi et al., 2002; Nixon et al., 2012; Oliveira et al., 2012). This Ser-150-mediated increase in myofilament Ca2+-sensitivity has been proposed to occur by modulating the affinity of the TnI inhibitory peptide binding to actin and/or switch peptide binding to the TnC N-lobe (Ouyang et al., 2010; Nixon et al., 2012). The location of Ser-150 immediately adjacent to the TnI inhibitory peptide has potential


*Values are mean* ± *SEM. pCa50,* −*log[Ca2*+*] at 50% force development; Hill, Hill coefficient; Fmax , maximal force in* µ*N/mm<sup>2</sup> ; n, number of single myocytes measured from at least three different exchanges per group. One-way ANOVA demonstrates a significant interaction. p* < *0.05.*

*<sup>a</sup>Significantly different vs. Tn WT.*

*<sup>b</sup>Significantly different vs. RcTn WT.*

*<sup>c</sup>Significantly different vs. Tn S23/24/150D.*

*<sup>d</sup>Significantly different vs. Tn S23/24D.*

*<sup>e</sup>Significantly different vs. Tn S150D.*

for phosphate incorporation to repel the TnI C-terminus from actin and/or promote the binding of the switch peptide to TnC. Enhanced TnC-TnI switch peptide binding is supported by FRET experiments indicating that Ser-150 pseudo-phosphorylation shortens inter-site distances between TnI and TnC (Ouyang

et al., 2010). In addition, data from our lab demonstrating that Ser-150 phosphomimetic substitution increases Ca2+-sensitivity in isolated Tn (Nixon et al., 2014), supporting that Ser-150 phosphorylation enhances the binding of the TnI switch peptide to TnC in the absence of interactions with the actin filament.

Enhanced binding of the switch peptide to TnC would contribute to a stabilization of the TnC-Ca2<sup>+</sup> affinity and a decrease in Ca2<sup>+</sup> dissociation, as suggest by our model (**Figure 1**). Current findings from our model further suggest that Ser-150 phosphorylation destabilizes the interaction of the TnI C-terminus with actin to promote thin filament activation. This mechanism is validated by our solid-phase protein binding experiments in which Ser-150 pseudo-phosphorylation decreases the binding affinity of isolated TnI to actin (**Figure 2**). Thus, we propose that the prominent TnI affinities acted upon by Ser-23/24 and Ser-150 phosphorylation differ in that phosphorylated Ser-23/24 alters the affinity of the switch peptide for TnC by accelerating TnI dissociation from TnC-Ca2+. In contrast, while Ser-150 phosphorylation may alter other TnI affinities in the thin filament, we demonstrate that the decreased affinity of pseudo-phosphorylated TnI at Ser-150 for actin is sufficient to elicit the observed increased calcium sensitivity.

Although the structural basis for the integrated effects of TnI of Ser-150 and Ser-23/24 phosphorylation remains unclear, our data provides significant mechanistic insight into this functional integration. Previous work indicates the TnI C-terminus is located in close proximity to the TnI N-terminus in the activated Tn complex (Howarth et al., 2007; Warren et al., 2009). This proximity of the TnI N-terminus containing Ser-23/24 to the C-terminus containing Ser-150 provides a mechanism by which Ser-150 and Ser-23/24 phosphorylation may interact to produce an integrated effect on Ca2+-sensitivity when on the same TnI molecule (Nixon et al., 2012). Our current data, however, demonstrates that phosphorylation of Ser-150 and Ser-23/24 on the same TnI molecule is not necessary to produce the Ca2+ sensitivity blunting effect (**Figure 5**). Although a direct TnI N-C terminal interaction may occur, we demonstrate Ser-23/24 and Ser-150 phosphorylation functional integration also utilizes separate, distinct molecular mechanisms that modulate TnI protein-protein interactions within the thin filament to regulate heart function. These data present a model in which a cumulative effect of each Tn phosphorylation along the thin filament must be considered to contribute to the overall observed Ca2+-sensitivity and contractile effects.

### TnI Phosphorylation Functional Integration Effects On Myofilament Regulation

A large subset of the literature has described how isolated Tn modifications affect Ca2+-mediated force development. The phosphorylation of TnI is significant considering that altered Ca2<sup>+</sup> sensitivity has been shown to underlie physiological modulation of cardiac output as well as potential adaptive and maladaptive responses in cardiac dysfunction (Zakhary et al., 1999; Kobayashi and Solaro, 2005; Zhang et al., 2012;

Lang et al., 2015). While Ser-150 has been shown to be phosphorylated by p-21 activated kinase and AMP-activated protein kinase (AMPK) in vitro, AMPK seems to be the likely relevant kinase in vivo (Buscemi et al., 2002; Sheehan et al., 2009; Oliveira et al., 2012). Previous studies have demonstrated AMPK associates with TnI and phosphorylates Ser-150, and isolated hearts perfused with an AMPK activator increase Ser-150 phosphorylation (Nixon et al., 2012; Oliveira et al., 2012). Several reports have demonstrated the importance of AMPK signaling in cardiac metabolism and contractility. Knockout or inhibition of AMPK has been shown to lead to cardiomyopathy and/or altered contraction concurrent with decreases in TnI Ser-150 phosphorylation (Chen et al., 2014; Sung et al., 2015). It remains to be resolved if the pathological effects of AMPK inactivation result from metabolic dysregulation, contractile deficiencies or both. We have demonstrated the effect of Ser-150 phosphorylation to increase myofilament calcium sensitivity of force development in skinned trabeculae, increase steadystate calcium binding to isolated troponin, and slow calcium dissociation from TnC on the thin filament (Nixon et al., 2012, 2014). Additionally, significant evidence has pointed to a role for Ser-150 phosphorylation under pathophysiologic conditions. Ser-150 phosphorylation is increased following acute ischemia in the murine heart and the calcium sensitization induced by Ser-150 phosphorylation was shown to blunt myofilament calcium desensitization caused by acidic pH (Nixon et al., 2014). As such, Ser-150 phosphorylation may play an important role to increase calcium sensitivity and force development resulting in preserving myocardial function under pathological stress during ischemia. The functional integration between Ser-23/24 and Ser-150 phosphorylation also presents a potentially important interplay in vivo. Previous work has demonstrated chronic beta adrenergic stimulation results in increased TnI Ser-150 phosphorylation (Taglieri et al., 2011). Similarly, both Ser-23/24 and Ser-150 phosphorylation are increased during myocardial ischemia (Nixon et al., 2014). Together these findings suggest cross-talk between beta adrenergic and AMPK mediated signaling to increase cardiac output by accelerating thin filament deactivation/muscle relaxation while maintaining normal force development. Future work is needed to investigate the signaling cross-talk between B-adrenergic and AMPK signaling, functional integration of Ser-23/24 and Ser-150 phosphorylation and their effects on in vivo cardiac contractility and energy metabolism.

Whether these isolated functional effects of Ser-150 phosphorylation are maintained in the integrated network alongside other phosphorylation events remains understudied. Our previous work demonstrated functional integration of TnI Ser-150 with Ser-23/24 phosphorylation blunted the myofilament desensitization induced by TnI Ser-23/24 phosphorylation alone (**Figure 4**) (Nixon et al., 2012, 2014). In the current study we assessed the contribution of TnI Ser-23/24 and Ser-150 integrated TnI phosphorylation by analyzing their effects on Ca2+-sensitive force development. We demonstrate that the overall observed Ca2+-sensitivity is proportional to the amount of phosphorylated TnI present at up to ∼ 57% exogenous Tn exchange (**Figure 3**). Although we did not achieve 100% exchange of exogenous Tn into our single myocyte system, previous studies have suggested 50–60% exchange is sufficient to elicit maximal effects on Ca2+ sensitivity for Ser-23/24 phosphorylation (Wijnker et al., 2013). Thus, it is possible that the proportional functional effects of TnI phosphorylation exist only within a linear range of exchange below 60%. In addition, while we detected a degradation fragment of TnI in our exchanged cells, the presence of this fragment was variable amongst TnI exchange groups and did not appear to correlate with changes in pCa50 nor maximal tension values. The appearance of a TnI cleaved product is consistent with previous studies demonstrating that approximately 13% of the total TnI exists as this TnI fragment in the normal rat heart (Yu et al., 2001; Barbato et al., 2005). Nevertheless, because these changes in Ca2+-sensitive force development are linear with the amount of phosphorylation present, our data suggests that this is not a mechanism contributing the different functional effects of Ser-23/24 and Ser-150 phosphorylation upon integration. As such, quantification of multiple TnI phosphorylations is necessary to infer the effect of TnI phosphorylation on the overall observed Ca2+-sensitivity in its relation to force production.

Previously we demonstrated functional integration of TnI Ser-23/24 and Ser-150 pseudo-phosphorylation occurred when both phosphorylations were on the same TnI molecule (Nixon et al., 2012, 2014). We now demonstrate that TnI Ser-23/24 and Ser-150 pseudo-phosphorylation integration similarly affects steady-state Ca2+-sensitive force development whether these phosphorylations are on the same or different TnI molecules (**Figure 6**). This suggests that the TnI Ser-23/24 and Ser-150 phosphorylation integration of Ca2+-sensitive force development function occurs through a cumulative effect of separate Tn contributions along the thin filament. Our current data further demonstrates that these phosphorylations exhibit complex functional integration both when on the same TnI molecule as well as when on different TnI molecules. These relationships may not be universal for other measurements of myofilament function, such as maximal force, Ca2<sup>+</sup> dissociation, length dependent activation, etc. and may be further dependent upon the specific TnI phosphorylation. In fact, we previously demonstrated TnI Ser-23/24 and Ser-150 phosphorylation integration functions to differently alter Ca2+-sensitivity and Ca2<sup>+</sup> dissociation. That is, while the integration of Ser-23/24 and Ser-150 exhibit expected Ca2+-sensitivity, Ca2<sup>+</sup> dissociation remains unexpectedly accelerated (Nixon et al., 2012, 2014). Likewise, we have also demonstrated the functional integration of TnI Ser-23/24 with Tyr-26 phosphorylation, where the combination of these phosphorylations does not exhibit an expected effect on Ca2+-sensitivity but does exhibit the expected further accelerated Ca2<sup>+</sup> dissociation (Salhi et al., 2014). Further experiments are ultimately needed to define the integrated effects of other TnI phosphorylations on these different functional measurements.

In all, our current study demonstrates that the apparent magnitude and rates of myofilament activation and deactivation are dependent on multiple mechanisms functioning within Tn that comprise intricate protein-protein interactions. We demonstrate different TnI phosphorylations can affect varied mechanisms within Tn that differentially alter specific contractile functions (Ca2<sup>+</sup> binding, kinetics, force development, etc.). It is therefore the combination of each TnI phosphorylation's affect on different mechanisms within Tn to alter myofilament activation that imparts functional integration. The integrative

### REFERENCES


role of multiple TnI phosphorylations thus allows for the complex fine-tuning of cardiac function through Tn.

### CONCLUSION

Ultimately, TnI regulation of cardiac function is dependent upon the homeostatic balance between the mechanistic proteinprotein interactions of each phosphorylation on Tn units along the myofilament. Our findings demonstrate that the different effects of Ser-23/24 and Ser-150 phosphorylation upon functional integration combination are the result of distinct Tn mediated mechanisms that alter different Ca2<sup>+</sup> regulated TnI interactions within the thin filament. We further demonstrate that the blunting effect of TnI Ser-150 phosphorylation integration on Ser-23/24 dependent desensitization is recapitulated in the single myocyte system of Ca2+-sensitive force development and this effect of phosphorylation on Ca2+-sensitivity is proportional to the amount of TnI phosphorylation present. Finally, our data demonstrates that TnI phosphorylations with divergent properties can functionally interact while on separate Tn molecules along the myofilament to produce a complex modulation of overall observed Ca2+-sensitive force production. To fully understand the contractile outcome of multiple TnI phosphorylations it is therefore important to investigate TnI phosphorylation alterations in their integrated state as they occur in the heart.

### AUTHOR CONTRIBUTIONS

HS: Conducted experiments, analyzed data and wrote the manuscript; NH: Conducted experiments and analyzed data; JS: Conducted experiments and edited the manuscript; EB: Conducted experiments and edited the manuscript; MZ: Contributed to design and edited the manuscript; PJ: Contributed to design and edited the manuscript; JD: Contributed to design and edited the manuscript; BB: Contributed to design, oversaw experiments, wrote and edited the manuscript.

### FUNDING

Support for this work was obtained from NIH grants HL114940 (to BB), HL114940-S1 (to HS), HL091986 (to JD), HL113084 (to PJ) and American Heart Association grant GRNT27760114 (To MZ).

myofilaments to Ca2<sup>+</sup> as a therapeutic target for hypertrophic cardiomyopathy with mutations in thin filament proteins. Circ. Cardiovasc. Genet. 7, 132–143. doi: 10.1161/CIRCGENETICS.113. 000324


isoproterenol-induced cardiac hypertrophy in association with activation of Erk1/2 and inhibition of protein phosphatase 2A. J. Mol. Cell. Cardiol. 51, 988–996. doi: 10.1016/j.yjmcc.2011.09.016


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Salhi, Hassel, Siddiqui, Brundage, Ziolo, Janssen, Davis and Biesiadecki. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Carbonic Anhydrase III Is Expressed in Mouse Skeletal Muscles Independent of Fiber Type-Specific Myofilament Protein Isoforms and Plays a Role in Fatigue Resistance

### Han-Zhong Feng and J.-P. Jin\*

*Department of Physiology, Wayne State University School of Medicine, Detroit, MI, USA*

#### Edited by:

*P. Bryant Chase, Florida State University, USA*

#### Reviewed by:

*Robert W. Wiseman, Michigan State University, USA Stephen T. Kinsey, University of North Carolina at Wilmington, USA Christina Karatzaferi, University of St Mark and St John, UK Susan V. Brooks, University of Michigan, USA*

> \*Correspondence: *J.-P. Jin jjin@med.wayne.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *17 August 2016* Accepted: *16 November 2016* Published: *15 December 2016*

#### Citation:

*Feng H-Z and Jin J-P (2016) Carbonic Anhydrase III Is Expressed in Mouse Skeletal Muscles Independent of Fiber Type-Specific Myofilament Protein Isoforms and Plays a Role in Fatigue Resistance. Front. Physiol. 7:597. doi: 10.3389/fphys.2016.00597* Carbonic anhydrase III (CAIII) is a metabolic enzyme and a regulator for intracellular pH. CAIII has been reported with high level expression in slow twitch skeletal muscles. Here we demonstrate that CAIII is expressed in multiple slow and fast twitch muscles of adult mouse independent of the expression of myosin isoforms. Expressing similar fast type of myofilament proteins, CAIII-positive tibial anterior (TA) muscle exhibits higher tolerance to fatigue than that of CAIII-negative fast twitch extensor digitorum longus (EDL) muscle in *in situ* contractility studies. We further studied the muscles of CAIII knockout (*Car3*-KO) mice. The loss of CAIII in soleus and TA muscles in *Car3*-KO mice did not change muscle mass, sarcomere protein isoform contents, and the baseline twitch and tetanic contractility as compared with age-matched wild type (WT) controls. On the other hand, *Car3*-KO TA muscle showed faster force reduction at the beginning but higher resistance at the end during a fatigue test, followed by slower post fatigue recovery than that of WT TA muscle. Superfused *Car3*-KO soleus muscle also had faster total force reduction during fatigue test than that of WT soleus. However, it showed a less elevation of resting tension followed by a better post fatigue recovery under acidotic stress. CAIII was detected in neonatal TA and EDL muscle, downregulated during development, and then re-expressed in adult TA but not EDL muscles. The expression of CAIII in *Tnnt1*-KO myopathy mouse soleus muscle that has diminished slow fiber contents due to the loss of slow troponin T remained high. *Car3*-KO EDL, TA, and soleus muscles showed no change in the expression of mitochondria biomarker proteins. The data suggest a fiber type independent expression of CAIII with a role in the regulation of intracellular pH in skeletal muscle and may be explored as a target for improving fatigue resistance and for the treatment of *TNNT1* myopathies.

Keywords: carbonic anhydrase III, skeletal muscle fatigue, fast and slow twitch muscle fibers, myofilament protein isoforms, TNNT1 myopathy

## INTRODUCTION

Carbonic anhydrases (CA) catalyze the reversible hydration of CO<sup>2</sup> to H2CO3. At least 16 CA isozymes have been identified in mammals with different tissue distribution and catalytic activity (Imtaiyaz Hassan et al., 2013). CAIII is an ∼30-kDa cytosolic protein (Carter et al., 1978) present at high levels in liver, adipocytes, and skeletal muscles (Sly and Hu, 1995). It is a low activity enzyme among CA isozymes (Koester et al., 1977, 1981) but is resistant to most sulfonamide inhibitors (Sanyal et al., 1982). The physiological function of CAIII is controversial. CAIII expression is negligible in preadipocytes and becomes abundant after differentiation (Lynch et al., 1993), implicating a role in fatty acid metabolism (Lyons et al., 1991). CAIII may facilitate rapid conversion of glycolytic intermediates to oxaloacetate and citrate and stimulate their incorporation into fatty acids. However, adipocyte CAIII expression in obese mice is lower than that in lean mice (Lynch et al., 1992).

CAIII expression in skeletal muscle was observed as fiber type-specific, mainly reported in type I slow-twitch muscle fibers (Shima, 1984; Vaananen et al., 1985; Frémont et al., 1988; Zheng et al., 1992; Sly and Hu, 1995). In mouse, CAIII transcripts are first detected in the myotomes of somites in embryos between 9.5 and 10.5 days post coitum, and gradually increase in all skeletal muscles during the next 4 days of development (Lyons et al., 1991). After birth, CAIII mRNAs are expressed at high level in mature slow muscle fibers. The expression of CAIII during early muscle development suggests a correlation with skeletal muscle differentiation. However, Car3 gene knock-out (Car3-KO) in mice did not affect normal development, growth and life span with minimum phenotype in soleus muscle that has a high slow fiber content (Kim et al., 2004). Studies of the gastrocnemius muscle of Car3-KO mice using in situ <sup>31</sup>P magnetic resonance detected reductions of phosphocreatine and ATP, elevations of ADP and inorganic phosphate, and a decrease of pH during 2 min of fatigue contractions, which were at significantly higher degrees than that of wild type (WT) controls (Liu et al., 2007).

A large number of sarcomeric protein mutations have been found to cause inherited cardiomyopathies (Watkins et al., 2011), including many in the gene encoding cardiac troponin T (TnT) (Willott et al., 2010; Sheng and Jin, 2014). In contrast, very few myopathic mutations are identified in skeletal muscle isoforms of TnT. The most investigated skeletal muscle TnT mutations are five nemaline myopathies alleles in TNNT1 gene encoding the slow skeletal muscle isoform of TnT. TNNT1 myopathies are featured by loss of slow twitch muscle fibers and presented with severe muscle atrophy, weakness and failure of respiratory muscle (Johnston et al., 2000; Jin et al., 2003; Amarasinghe et al., 2016). Mouse models of TNNT1 myopathy reproduced the slow muscle atrophy and degeneration phenotypes and showed a significant loss of fatigue resistance of soleus and diaphragm muscles (Feng et al., 2009; Wei et al., 2014).

To investigate the potential function of CAIII in skeletal muscle and in adaptation to the loss of slow fibers in TNNT1 myopathy, here we demonstrated that CAIII is expressed in multiple slow and fast twitch muscles of adult mouse independent of the expression of myosin isoforms. Expressing similar myofilament protein contents, tibial anterior (TA) expressing a high level of CAIII exhibits higher resistance to fatigue than that of CAIII-negative extensor digitorum longus (EDL) muscle. Car3-KO TA muscle showed faster force reduction at the beginning but higher resistance at the end, followed by a slower post fatigue recovery than that of WT TA muscle. Car3-KO soleus muscle also had faster total force reductions during fatigue but less elevation of resting tension followed by a better post fatigue recovery under acidotic stress. The expression of CAIII in Tnnt1-KO myopathy mouse soleus muscle that has diminished slow fibers remained high. These data suggest a fiber type independent expression of CAIII with a role in the regulation of intracellular pH and fatigue resistance in skeletal muscle cells.

### MATERIALS AND METHODS

### Animal Models

Car3-KO mice (Kim et al., 2004) and matching strain (129SEVE) wild type mice were purchased from Jackson Lab. The development of Tnnt1-KO mice in C57BL/6 strain was reported previously (Wei et al., 2014). The genotypes of the mice were verified by PCR. Mice were housed in the animal facility on a 12:12-h light-dark cycle (6:00 AM/6:00 PM) and fed a standard pellet diet and water. Two to three month old mice were used for functional studies. Previous studies of Car3-KO mice did not find gender-generated differences (Kim et al., 2004). The comparison of male and female data in our present study did not indicate statistical significances in muscle weight, force production and fatigability. Therefore, pooled data from male and female mice were used for soleus muscle studies. The TA and EDL muscle studies were from male mice. All animal protocols are approved by the Institutional Animal Care and Use Committees of Wayne State University. The expression or lack of CAIII and slow skeletal muscle TnT in the muscles studied were confirmed by Western blot as described below.

### SDS-PAGE and Western Blotting

Fresh or frozen muscle tissues were rapidly homogenized in SDS-PAGE sample buffer containing 2% SDS and 1% β-mercaptoethanol, pH8.8, using a high speed mechanical homogenizer (Pro 250, Pro Scientific Inc.) to extract total proteins. After heating at 80◦C for 5 min, the samples were clarified by centrifugation at 14,000 × g in a microcentrifuge for 5 min. Protein samples were resolved on 14% SDS-gel with acrylamide:bisacrylamide ratio of 180:1 or 12% SDS-gel with acrylamide:bisacrylamide ratio of 29:1 in a modified Laemmli buffer system in which both stacking and resolving gels were at pH 8.8. The protein bands in the gel were visualized by staining with Coomassie Blue R 250. Total protein in each lane was quantified by ImageJ software for normalizing the amount of sample loading.

Copies of the SDS-gels were transferred to nitrocellulose membrane using a Bio-Rad semidry electrotransfer device at 5 mA/cm<sup>2</sup> for 15 min. The blotted membranes were blocked in 1% bovine serum albumin (BSA) in Tris-buffered saline (TBS, 150 mM NaCl, 50 mM Tris, pH 7.5) with shaking at room temperature for 30 min. The blocked membrane was probed with an anti-CAIII monoclonal antibody (mAb) CP3 (Jin et al., 1996), an anti-TnI mAb TnI-1 (Jin et al., 2001), an anti-TnT mAb 2C8 (Jin and Chong, 2010), or antibodies against peroxisome proliferator-activated receptor coactivator 1 (PGC-1) (ab54481, Abcam) and voltage-dependent anion channel (VDAC) (4866S; Cell Signaling Technology, Beverly, MA), diluted in TBS containing 0.1% BSA, with gentle rocking at 4◦C overnight. The membranes were then washed three times with TBS containing 0.5% Triton X-100 and 0.05% SDS, incubated with alkaline phosphatase-labeled goat anti-mouse IgG second antibody (Santa Cruz Biotechnology), washed again as above, and developed in 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium substrate solution to visualize the protein bands detected.

### Glycerol-SDS-PAGE

Myosin heavy chain (MHC) isoforms expressed in muscle tissues were examined using glycerol-SDS-PAGE (Feng et al., 2011). Briefly, SDS-PAGE samples equivalent to 5 µg of muscle tissue (wet weight) were resolved on 8% polyacrylamide gel with acrylamide:bis-acrylamide ratio of 50:1, prepared in 200 mM Tris base, 100 mM glycine, pH 8.8, containing 0.4% SDS and 30% glycerol. The stacking gel contained 4% polyacrylamide with acrylamide:bis-acrylamide ratio of 50:1, 70 mM Tris-HCl (pH 6.7), 4 mM EDTA, 0.4% SDS, and 30% glycerol. The upper cathode running buffer consists of 100 mM Tris base, 150 mM glycine, 0.1% SDS, and 10 mM β-mercaptoethanol. The lower anode running buffer was 50% dilution of the upper running buffer without β-mercaptoethanol. The 0.75-mm-thick Bio-Rad minigels were run at 100 V in an icebox for 24 h. The resolved protein bands were visualized after staining with Coomassie blue R250.

### In situ Measurement of Muscle Contractile Functions

Due to the large size of TA muscle, ex vivo superfusion may generate hypoxia in the center of the muscle due to limited diffusion of oxygen. Therefore, muscle contractility was measured in situ with physiological blood supply to compare TA and EDL muscle functions.

Mice were anesthetized by inhalation of 3.5% isoflurane for induction and 2% isoflurane for maintenance using a small animal anesthesia system (SomnoSuite, Kent Scientific Corp). On a temperature controlled platform (Aurora Scientific, Aurora, Ontario, Canada) and under a heating lamp to maintain the body temperature at 37 ± 0.5◦C using PhysioSuite system (Kent Scientific Corp.), hair was removed from the leg area, the distal tendon of TA or EDL muscle was exposed surgically and made partially free for mounting to a force transducer (300C-LR, Aurora Scientific Corp) through a stainless steel wire hook and a serrated clip that was able to hold the tendon tightly. As adult mouse TA muscle generates more than 100 g force which is out of range for the force transducer, the hook was connected to a short point of the lever arm to expand the range of measurements. The actual force was then calibrated to correct for the shorter length of the lever arm. The proximal end of tibial bone was mounted on the platform with a pair of pointed screws. The foot was taped a position that the muscle was aligned with the force transducer.

The sciatic nerve was exposed and freed carefully avoiding injury. A pair of custom-made platinum wire electrodes was placed around the nerve for applying stimulations using an electrical stimulator (Aurora Scientific Corp). Continuing dripping of warm Kreb's buffer bubbled with 95% O2, 5% CO<sup>2</sup> at 37◦C around the exposed muscle tendon was used to prevent tissue drying. The exposed sciatic nerve was also covered by a filter paper wetted with dripping Kreb's buffer. Biphasic square pules of 0.1 ms duration at a voltage 50% above the threshold were applied to stimulate twitch contractions of the muscle. After 20 min equilibration at 0.05 Hz twitch contractions, the resting length of muscle was slowly increased to reach an optimal resting length that generated maximal isometric force. Then the muscle was kept at the optimal resting length for tetanic contraction, fatigue, and recovery studies.

After 20 min equilibration with tetanic contractions at 300 Hz for 300 ms in every 1 min, a series of frequencies (200–380 Hz) were tested to identify the optimum frequency that produced maximum tetanic force, in which EDL and TA muscles showed the same optimal frequency of 300 Hz to produce maximum force. Stimulation of 300 ms duration in every 1500 ms at the optimal frequency was applied for 300 repeats to induce muscle fatigue. One minute after the fatigue contractions, 20 min of 300 ms tetanic contractions every 1 min was recorded for the recovery of muscle contractility.

### Histology

After in situ contractility measurement, the TA, EDL, and soleus muscles of the other leg was exposed. Two 30-G needles were inserted into the proximal end of the muscle and the distal tendon. The distance between two needles was measured when the ankle was at 90◦ angle. The muscle was then removed with the needles attached and placed in optimal cutting temperature compound (O.C.T.) in a cryostat tissue holder with the tendons pinned down onto a cork at the in situ muscle length between the two needles. The muscle tissue in O.C.T. was dipped in pre-cooled isopentane at −160◦C and freeze for 1 min before submerged in liquid nitrogen. This protocol eliminates ice crystal formation inside muscle fibers. Five or ten-micron cryo-sections were cut, processed for hematoxylin and eosin staining, and imaged using a Zeiss Axio Observer A1 microscope with an attached digital camera.

### Immunohistochemical Staining of Muscle Sections

As previously reported (Wei et al., 2014), cross sections of mouse soleus muscle were blocked in phosphate-buffered saline (PBS) containing 0.05% Tween-20 (PBS-T) and 1% BSA at room temperature for 30 min. Endogenous peroxidase was inactivated by incubation with 1% H2O<sup>2</sup> in PBS-T at room temperature for 10 min. After a wash with PBS-T, the muscle sections were incubated with an anti-MHC I mAb FA2 (Jin et al., 1990) or SP2/0 myeloma ascites control in PBS-T containing 0.1% BSA at 4◦C overnight. After washes with PBS-T to remove excess primary antibodies, the sections were incubated with horseradish peroxidase conjugated anti-mouse second antibody in PBS-T containing 0.1% BSA at room temperature for 1 h. After washes with PBS-T to remove excess second antibody, MHC I expression in type I fibers was visualized via 3,3-diaminobenzidine-H2O<sup>2</sup> substrate reaction after developing in a dark box for 30 s. The reaction was terminated by washes with 20 mM Tris-HCl, pH 7.6. Nuclei were then counterstained with Haematoxylin for 5 min followed by washes with distilled water. The muscle sections were mounted in PBS containing 50% glycerol, sealed using Cytoseal, and photographed using a Zeiss Observer 125 microscope.

### Measurement of Contractile Function of Superfused Soleus Muscle at Normal and Acidotic PH

Acidosis generated with lowering the pH of perfusion buffer was employed to investigate the mechanism for CAIII to alter fatigue resistance under stress conditions. Ex vivo contractility measurement of intact soleus muscle was performed with superfusing the muscle under the carbogen equilibration with different levels of CO2. Using a protocol modified from our previous studies (Feng et al., 2011), intact soleus muscle was carefully isolated with both tendons avoiding stretching damage and mounted vertically to a dual-mode lever arm force transducer (300B, Aurora Scientific) in an organ bath containing 100 mL modified Kreb's solution (118 mM NaCl, 25 mM NaHCO3, 4.7 mM KCl, 1.2 mM KH2PO4, 2.25 mM MgSO4, 2.25 mM CaCl2, and 11 mM D-glucose, continuously gassed with 95% O2, 5% CO2, pH 7.4). Maintained at 25 ± 0.5◦C in the bath with thermos-controlled circulating water jacket, contractions were elicited with bipolar pulse field electrical stimulation using a stimulator (701B, Aurora Scientific). Twitch contractions were elicited with supramaximal pulses (0.1 ms, 28 V/cm), unless specified otherwise. Tetanic contractions were elicited with a train of the same pulses at 100 Hz for 0.7 s. Isometric force data were collected via a digital controller A/D interface (604C, Aurora Scientific) and recorded using Chart software (ASI, Aurora Scientific). Developed twitch and tetanic forces were determined at the optimal muscle length that gave the highest twitch force and calculated by subtracting the resting tension from the total force.

After 20-min equilibration with 0.7 s tetanic contractions per minute, various stimulation frequencies were tested to determine the optimal frequency that produced maximum tetanic force. A 300 s fatigue protocol was performed with intermittent tetani of 700 ms every second. One minute after the end of fatigue, recovery of muscles contractility was recorded for 20 min with 0.7 s tetanic contractions every 1 min. After the baseline study, the muscle was re-equilibrated, and the perfusion buffer was switched to 70% O2, 30% CO<sup>2</sup> to apply acidosis and repeat the fatigue and recovery protocol.

### Data Analysis and Statistics

Densitometry analysis of SDS-gel and Western blots was done using ImageJ software on images scanned at 600 dpi resolution. The force of EDL and TA muscles was normalized to muscle weight that represents the total volume of contractile units, i.e., the sarcomeres, instead of muscle cross sectional area since the organization for muscle fibers in EDL and TA is rather different.

Quantitative data are presented as mean ± SE, and statistical analysis was performed using student's t-test or two-way ANOVA as noted in the figure legends. A Bonferroni post hoc follow-up test was conducted to compare the mean values between WT and Car3-KO groups.

### RESULTS

### Broad Expression of CAIII in Mouse Skeletal Muscles

The CP3 mAb that we previously developed by immunization using chicken smooth muscle calponin 1 (Jin et al., 1996) strongly reacts with purified bovine CAIII included in the Sigma molecular weight marker set L70 (**Figure 1**). Further testing showed that CP3 specifically recognizes mouse CAIII in liver, white fat and slow type muscle such as the soleus but not EDL. As expected, no CAIII was detected by CP3 in mouse heart and bladder smooth muscle (**Figure 1**). While the specific epitope structure shared by calponin 1 and CAIII merits a future study, mAb CP3 provides a useful tool to study the expression of CAIII in skeletal muscles.

Multiple adult mouse skeletal muscles were examined with Western blotting using mAb CP3 (**Figure 2**). The accompanying Western blots of mAb TnI-1 that recognizes all three isoforms of TnI demonstrated the relative fast and slow fiber contents in these muscles (**Figure 2**). The results in **Figure 2A** demonstrated that all slow fiber-rich muscles where slow TnI is detected at significant levels and most of the pure fast fiber mouse muscles in which only fast TnI is present expressed high or significant amounts of CAIII. Only a few fast fiber skeletal muscles are negative, including EDL, masseter, tongue and the upper portion of esophagus (as well as the heart). The glycerol-SDS-gel data shown in **Figure 2B** further demonstrate that the expression of myosin isoforms in the fast fiber skeletal muscles has no definitive correlation to the expression of CAIII. The observation that CAIII expression in skeletal muscle is independent of the fiber type-specific myofilament proteins and its expression in many but a few fast twitch muscles intrigued our investigation on the physiological significance.

### Distinguishable Contractility and Fatigue Resistance of Mouse TA and EDL Muscles in situ

The finding that CAIII can be either totally absent or expressed at a significant level in some pure fast fiber mouse muscles (**Figure 2A**) provided us with informative experimental

(B) Glycerol-SDS-PAGE showed various MHC isoform contents in the fast fiber type mouse skeletal muscles studied without clear correlation with the expression of CAIII. (C) CAIII positive fast muscle TA and CAIII negative fast muscle EDL were chosen for functional comparisons.

and TA muscles. (A) The time parameters of twitch contraction and relaxation (TP50, time to develop 50% peak force; TPT, time to develop peak force; TR50, time for 50% relaxation; and TR75, time to 75% relaxation) were not significantly different in EDL and TA muscles. (B) Force normalized to the pre-fatigue maximum force in *in situ* fatigue study showed that CAIII-positive TA muscle had a slower decrease in contractility than that of CAIII-negative EDL muscle reflecting a higher fatigue resistance. (C) TA muscle also showed a better post fatigue recovery than that of EDL. The data are presented as mean ± SE. *N* = 4 mice each in TA and EDL groups. Statistical analysis was performed using Student's *t-*test. \**P* < 0.05 vs. EDL using Student's *t-*test. \*\*\**P* < 0.001 vs. EDL using two-way ANOVA test with Bonferroni adjustment (for TA vs. EDL: *DF* = 1, *F* = 27.61, *P* = 1.56 × 10−<sup>6</sup> ).

systems for functional investigation. TA and EDL muscles are anatomically adjacent fast twitch limb muscles, of which a significant level of CAIII is present in TA but completely absent in EDL (**Figure 2C**), indicating a representative pair to study the function of CAIII in TA muscle. While the difference in fiber orientations makes their muscle weight-normalized forces noncomparable, in situ muscle contractility studies demonstrated that TA and EDL muscles had no difference in the time parameters of twitch contraction and relaxation (**Figure 3A**). The 300 s in situ muscle fatigue protocol then revealed a slower force drop in TA muscle than that of EDL during the first 100 s (**Figure 3B**). TA muscle also exhibited a better post fatigue recovery than that of EDL (**Figure 3C**). These results suggest that the presence of CAIII in TA muscle may have contributed to the higher resistance to fatigue than that of the CAIII-negative EDL muscle.

### Car3-KO Decreases the Fatigue Resistance of Mouse TA Muscle

To further investigate the function of CAIII in TA muscle, Car3-KO mouse muscles were studied. mAb CP3 Western blot confirmed the absence of CAIII in Car3-KO mouse TA muscle and liver (**Figure 4**). Body weight of 2-month old Car3-KO mice

CAIII in *Car3*-KO mouse soleus and TA muscles and liver.

showed no difference from WT control (**Supplement Figure 1**). Normalized to body weight, the weights of TA and EDL muscles were similar in Car3-KO and WT mice (**Supplement Figure 1**). Histology examination found no signs of inflammation or fiber degeneration in the TA muscle of young Car3-KO mice as compared with WT control (**Supplement Figure 1**). The cross sectional area of TA muscles of Car3-KO and WT mice are also similar (**Supplement Figure 1**). These data demonstrate that the loss of CAIII did not cause atrophy or degenerative defect in the TA muscle of Car3-KO mice. The EDL and soleus muscles of Car3-KO mice also did not show atrophy or degenerative changes (data not shown).

The SDS-PAGE gel and Western blots in **Figure 5A** showed that adult mouse TA and EDL muscles had very similar overall protein profiles except for two visible protein bands in EDL but not TA. The functional significance of these unknown proteins needs further investigation. The Western blots in **Figure 5A** showed similar isoform expression patterns of tropomyosin, fast TnI and splice forms of fast TnT in TA and EDL muscles. No significant change in tropomyosin, TnT, TnI and MHC isoforms in Car3-KO TA muscle as compared with that of WT TA. Glycerol-SDS-gel and densitometry quantification in **Figures 5B,C** showed that MHC isoforms IIx and IIb are expressed in these two muscles. The only detectable difference her is that the level of MHC IIx is higher and MHC IIb lower in TA than that in EDL muscles, a trend remained in the Car3-KO muscles.

In situ contractility studies showed that TA muscles of Car3-KO and WT mice had similar twitch and tetanic forces normalized to muscle weight (**Figure 6A**) and similar contractile time parameters (**Figure 6B**), demonstrating similar baseline functions. In situ fatigue protocol revealed that Car3-KO TA muscle exhibited a faster initial decrease of contractile force although the remaining contractile force at the end of the fatigue protocol was higher as compared to that of WT TA muscle (**Figure 6C**). Consistent with the possible contribution of CAIII to higher resistance to fatigue of TA muscle (**Figure 3C**), Car3- KO TA muscle showed a significantly slower recovery than that of WT control (**Figure 6D**). The results support the role of CAIII in the fatigue resistance of TA muscle.

### Car3-KO Increased the Resistance of Mouse Soleus Muscle to Acidosis

To investigate the role of CAIII in the fatigue resistance of skeletal muscle via regulating the intracellular pH environment, we compared contractility of superfused soleus muscles of Car3- KO and WT mice under normal and acidotic conditions. SDS-gel and Western blot analysis showed that Car3-KO soleus muscle had no change in TnI, TnT and MHC isoform contents as compared with that of WT soleus (**Figure 7A**). Normalized to body weight, the soleus muscle masses of Car3-KO, and WT mice were also similar (**Figure 7B**). The maintained muscle fiber type and mass in Car3-KO soleus muscle justifies its use in the study of CAIII's role in pH regulation and fatigue resistance.

Superfused at normal pH (7.4) using Kreb's buffer gassed with 5% CO2, Car3-KO and WT soleus muscles showed no difference

in twitch force (**Figure 7C**), and time parameters of contractile and relaxation (**Figure 7D**). At this physiological condition, the tetanic contractile force of Car3-KO and WT soleus muscles were

significant change in the levels of MHC IIx and IIb in TA or in EDL muscles. The data are presented as mean ± SE. *N* = 3 mice in each groups. Statistical

analysis was performed using Student's *t*-test.

similar (**Figure 8A**). When using Kreb's buffer equilibrated with 30% CO<sup>2</sup> to lower the extracellular pH to 6.5, tetanic contraction force of Car3-KO and WT soleus muscles decreased in similar degrees (**Figure 8A**). Whereas, the relaxation time of tetanic contraction was prolonged in both Car3-KO and WT soleus muscles at the acidotic condition, it was significantly less severe in Car3-KO soleus (**Figure 8B**).

It was interesting to note that during the fatigue test, the resting tension of soleus muscle rose in the early phase. This occurred in both WT and Car3-KO soleus muscles and at higher levels at the acidotic condition (**Figure 9A**). Consistent with the effect of deleting CAIII on minimizing the impairment of relaxation time in acidosis, Car3-KO soleus muscle exhibited much less increase in resting tension as compared with that of WT soleus, especially at the acidotic condition (**Figure 9A**). Correspondingly, the development force calculated by subtracting resting tension from the total force in each contraction was better maintained in Car3-KO soleus muscle than WT control during the fatigue test (**Figure 9B**). Supporting the observation that this may be a specific effect in acidosis, in situ contractility and fatigue experiments showed that under physiological in vivo pH, Car3-KO, and WT mouse TA and EDL muscles both had no increase in resting tension (**Supplement Figure 2**).

This finding suggests that the role of CAIII in increasing fatigue resistance is compromised under acidotic conditions. Therefore, CAIII may function in fatigue resistance of skeletal muscle via a bi-phasic regulation of intracellular pH. Under normal pH, it increases resistance to fatigue whereas in acidosis it has an opposite effect. The bi-phasic fatigue responses of Car3- KO TA muscle during in vivo fatigue test shown in **Figure 6C** support this hypothesis by the beneficial effect in the later phase of fatigue contractions when local acidosis may have occurred.

### Postnatal Down-Regulation and Re-Expression of CAIII in Adult Mouse TA Muscle

Glycerol-SDS-gel electrophoresis showed no difference in the developmental switching of MHC isoforms between Car3-KO and WT TA and EDL muscles (**Figure 10A**). Western blots showed that CAIII was detectable in neonatal soleus, EDL and TA muscles and continuously expressed in soleus muscle (**Figure 10B**). The expression of CAIII in EDL muscle ceased

after 7 days postnatal. The expression of CAIII in TA muscle showed similarly postnatal down-regulation but was re-activated to a significant level in adult as shown in the 2.5-month-old sample (**Figure 10B**). The developmental isoform switches of thin filament regulatory proteins TnI and TnT were not different in Car3-KO and WT TA muscles. This observation implicates that the postnatal expression of CAIII in fast fiber skeletal muscles may be an adaptation to specific functions.

### Maintained Expression of CAIII in Tnnt1 Myopathy Mouse Soleus Muscle

The expression of CAIII was examined in soleus muscle of WT and Tnnt1-KO myopathy mice (Johnston et al., 2000; Jin et al., 2003; Wei et al., 2014). Western blot and glycerol-SDS-gel demonstrate that the loss of slow TnT in Tnnt1-KO mice caused a drastic loss of slow type I fibers in soleus muscle indicated by the diminished levels of slow TnI as compared to WT controls (**Figures 11A,C**), reflecting a switch to more fast fiber content. Despite the diminished slow fiber contents, the level of CAIII was maintained in Tnnt1-KO soleus muscle (**Figures 11A,D**). Together with the increased content of fast fibers (**Figures 11B,E**), the results support a notion that the expression of CAIII is not restricted to slow fibers but corresponding to the contractile features of the soleus muscle.

## No Significant Change in Mitochondria Function in CAIII KO Mouse Muscles

Western blots for the levels of PGC-1α that participates in the regulation of multiple mitochondrial genes (Hock and Kralli, 2009) and VDAC that represents the total mitochondrial content (Raghavan et al., 2012) showed no significant difference between soleus, EDL and TA muscles of Car3-KO and WT mice (**Figure 12**). The results indicate that the deletion of CAIII did not alter the function of mitochondria in skeletal muscles.

## DISCUSSION

CAIII is present in many skeletal muscles at significant levels but its physiological function remains unclear. Although CAIII is a low activity enzyme among carbonic anhydrases, it is abundant in muscle cells. A previous study has reported unchanged baseline phenotype of Car3-KO mice (Kim et al., 2004) which we also used in the present study. The lack of drastic consequences after systemic knockout of the Car3 gene implicated that CAIII is a nonessential enzyme in the mouse. Since CAIII is a metabolic enzyme that regulates intracellular pH, our study focused on its potential role in muscle fatigue that involves pH changes in the myocytes. Our data present several interesting findings.

### The Expression of CAIII in Skeletal Muscles is Not Fiber Type Specific and Independent of the Expression Myofilament Protein Isoforms

CAIII was thought to have a specific expression in slow fiber rich muscles. However, our data showed significant expression of CAIII in not only slow type but also most of the fast fiber skeletal muscles of adult mouse (**Figure 2**). Only a few muscles lack CAIII expression, including EDL, masseter and tongue. Therefore, the expression of CAIII in mouse skeletal muscles is not slow fiber specific and independent of the expression myofilament protein isoforms. CAIII has been reported to be under thyroid control (Salvatore et al., 2014). Although thyroid hormone affects muscle fiber type and MHC isoform expression, mice used in our study are of normal thyroid function and the expression of CAIII in mouse muscle is independent of MHC isoforms (**Figure 2**). Therefore, the possible hormonal influences on the conclusions of our present study merits future investigation.

A hypothesis is that the expression of CAIII in skeletal muscle may correspond to the functional features of the muscle. For

example, both TA and EDL are classified as typical fast type muscles but only TA expresses CAIII (**Figure 2C**). Consistent with a previous report that the expression of Car3 mRNA starts in all skeletal muscles at early embryonic stage before concentrated in slow fiber muscles (Lyons et al., 1991), our developmental studies showed that CAIII expression was maintained in soleus muscle from neonatal to adult, but ceased 7 days after birth in TA and EDL muscles (**Figure 10B**). Accordingly, the expression of CAIII in early developmental myoblasts has been suggested as a diagnostic marker for muscle disease (Shima et al., 1983), allowing detection in human fetal plasma samples for diseases such as Duchenne muscular dystrophy (Carter et al., 1982).

An interesting finding is that CAIII is re-expressed in adult TA muscle but not EDL (**Figure 10B**), suggesting a secondary adaptation that activates Car3 expression in adult TA muscle in response to a specific functional demand. The expression of CAIII in TA but not EDL muscles was also reported in rat (Shiels et al., 1984). The mechanism of Car3 gene reactivation requires further investigation. Denervation and innervation affected the level of CAIII in TA and soleus muscles (Milot et al., 1991), supporting a role of the functional feature of muscle in the regulation of Car3 expression. As a counteractive muscle, TA muscle's activity is related to that of the slow fiber-containing soleus and gastrocnemius muscles. More studies are merited to investigate the role of CAIII in this type of coordinated function of skeletal muscles.

### CAIII Increases Muscle Resistance to Fatigue under Physiological Conditions

Our results showed that CAIII-positive fast muscle TA is more resistant to fatigue than that of CAIII-negative fast muscle EDL in in situ contractility studies (**Figure 3**) and Car3-KO TA muscle is less tolerate to fatigue than that of WT TA muscle in in situ contractility studies (**Figure 6**). In the meantime, Car3-KO did not change muscle mass and contractile protein isoform contents or baseline twitch and tetanic contractions (**Figures 5**, **6**). Therefore, CAIII does not directly impact on the basic contractility of skeletal muscle but functions in modulating the intracellular environment of muscle cells. The observation that CAIII positive skeletal muscles have higher resistance to fatigue under physiological conditions is a novel finding and indicates a potential mechanism to improve muscle function and durability.

The finding that the loss of slow fiber contents in Tnnt1-KO mouse soleus muscle did not cause decrease in CAIII expression (**Figure 11**) further supports the fiber type independent and functional demand determined expression of CAIII. The maintained high level expression of CAIII may contribute a partial sustention of fatigue resistance in the weight bearing soleus muscle of Tnnt1-KO mice. The observation that only a few mouse fast fiber muscles are lack of CAIII, such as EDL, tongue and masseter (**Figure 2**), suggests that an increase in the level of CAIII may be explored as an anti-fatigue treatment for TNNT1 myopathies.

Muscle fatigue is known as a decline of muscle performance in repeated and intense muscle contractions. Among multiple processes from excitation-contraction signaling to intracellular ionic equilibrium to changes in metabolites and action of contractile proteins (Allen et al., 2008; Kent-Braun et al., 2012), the fundamental changes during muscle fatigue are the inevitable accumulation of intracellular inorganic phosphate and hydrogen ions which cause impaired functions of Ca2<sup>+</sup> handling system and contractile proteins. Decreased intracellular pH is one of the factors that produce muscle fatigue during intensive contractions. Intensive contractions also increase metabolic production of CO2. By catalyzing the hydration of CO<sup>2</sup> to H2CO3, CAIII may play a role in adjusting intracellular pH to counter acidosis in muscle fatigue.

### CAIII Increases the Sensitivity of Muscle to Fatigue in Acidosis

In contrast to the anti-fatigue function of CAIII in vivo under physiological conditions, Car3-KO soleus muscle exhibited not

300 s tetanic contractions, there was a rise of resting tension in the first 100 s. Acidotic extracellular pH (30%CO2) significantly augmented this rise in WT soleus muscle, which was minimized in *Car3*-KO muscle. (B) Corresponding with the rise of resting tension, active tetanic tension development dropped rapidly during the first 100 s of fatigue contractions, most severe in WT soleus muscle at acidotic pH, which was minimized in *Car3*-KO muscle. The data are presented as mean ± SE. *<sup>N</sup>* <sup>=</sup> 6 mice each in WT and *Car3*-KO groups. \**<sup>P</sup>* <sup>&</sup>lt; 0.05 vs. WT. #*<sup>P</sup>* <sup>&</sup>lt; 0.05 vs. 30%CO2. Statistical test was performed using two-way ANOVA and adjusted mean comparison of Bonferroni test between the four groups (for WT and *Car*3-Ko with 5% and 30% CO2, *DF* = 3, *F* = 165.4 and P was close to 0).

decreased but increased resistance to fatigue as compared with that of WT soleus in ex vivo fatigue test especially under acidosis condition (**Figure 8**). Consistent with the anti-fatigue function of CAIII under physiological conditions, Car3-KO TA muscle is less tolerating to fatigue than that of WT soleus in the early phase of in situ fatigue test (**Figure 6C**). However, Car3-KO TA muscle maintained higher force production in the later phase of in vivo fatigue test where local acidosis might have occurred (**Figure 6C**). This observation suggests that the anti-fatigue function of CAIII is restricted to the physiological range of intracellular pH. A hypothesis is that it functions to decrease the CO<sup>2</sup> produced in the acute phase of intensive contractions (loss of CAIII causes CO<sup>2</sup> accumulation and faster dropping of force as shown in **Figure 6C**). The accumulation of HCO<sup>−</sup> 3 produced by CAIII would then decrease intracellular pH and the acidosis would worsen muscle fatigue. Therefore, deletion of CAIII in Car3- KO TA muscle actually produces higher remaining force in the later phase of fatigue contractions than that of WT control (**Figure 6C**). Further experimental studies directly measuring intracellular CO<sup>2</sup> and pH are required to test this hypothesis.

It is also interesting that Car3-KO soleus muscle had a significantly less rise of resting tension in the early phase of fatigue contractions (**Figure 9A**) corresponding to better sustained force development (**Figure 9B**). This phenomenon was more predominant at the acidotic condition produced with 30% CO2-gassed perfusion buffer. The elevated resting tension may be due to the slower relaxation during muscle fatigue (Westerblad and Lannergren, 1991). Such elevation of resting tension was absent in TA and EDL muscle in vivo under physiological conditions with complete relaxation to the baseline during fatigue test (**Supplement Figure 2**). The mechanism for CAIII to raise resting tension of muscle in fatigue and acidosis remains to be investigated. The rate of sarcoplasmic reticulum Ca2<sup>+</sup> ATPase (SERCA) pump to uptake Ca2<sup>+</sup> decreases in muscle fatigue (Allen et al., 1995), which may contribute to elevation of resting tension and lower force development since

neonatal soleus, TA and EDL muscles. The expression slightly lowered in soleus and ceased in EDL muscles during postnatal development. The expression of CAIIl in TA muscle diminished during postnatal development but re-activated in young adult. The expression of slow vs. fast troponin isoforms during postnatal development was examined using mAbs CT3 and TnI-1 Western blots and no difference was found between *Car3*-KO and WT muscles.

less Ca2<sup>+</sup> uptake causes less release during contraction. Low pH inhibits the SERCA pump (Allen et al., 1995; Wolosker et al., 1997). Therefore, our data suggest that the deletion of CAIII in soleus muscle may sustain SERCA activity during soleus muscle fatigue by reducing the productions of HCO<sup>−</sup> 3 and H<sup>+</sup> under preexisting acidosis (**Figure 9A**). A previous NMR study showed more decreases in phosphocreatine, ATP and pH in Car3-KO mouse gastrocnemius muscle following a short period (2 min) of intense contractions than that in WT controls (Liu et al., 2007), further supporting the hypothesis that the functions of CAIII in regulating intracellular pH is dependent on the range of intracellular pH.

### Maintained Expression of CAIII in TNNT1-KO Soleus Muscle Despite the Loss of Slow Fiber Contents

Revising the notion that CAIII expression in skeletal muscle was specific to type I slow-twitch muscle fibers (Shima, 1984; Vaananen et al., 1985; Zheng et al., 1992; Sly and Hu, 1995), we demonstrated that its expression in soleus muscle remains when the slow fiber content was diminished due to the loss of slow TnT in a transgenic mouse model of TNNT1 nemaline myopathy (**Figure 11**). Mimicking a recessively inherited lethal disease originally found in the Amish caused by a nonsense

expression was remained. (B) Immunohistochemistry staining using anti-MHC I mAb FA2 of cross sections of *Tnnt1*-KO soleus muscle showed atrophy of type I slow fibers and hypertrophy of type II fast fibers as compared with WT control. (C) Densitometry quantification of the TnI-1 mAb Western blots confirmed the significantly decreased ratio of ssTnI/fsTnI in *Tnnt1*-KO soleus muscle in comparison with that of WT soleus muscle. (D) Densitometry quantification of Western blot normalized to the actin band in SDS-gel confirmed no decrease of CAIII in *Tnnt1*-KO soleus muscle vs. the WT control. (E) Quantification of the percentage cross sectional areas of type I and type II fibers in soleus muscles confirmed the atrophy of type I fibers and hypertrophy of type II fibers in *Tnnt1*-KO mice vs. that in WT mice. The data are shown as mean ± SE. *N* = 3 mice each in WT and *Tnnt1*-KO groups. \*\*\**P* < 0.001 as compared with WT using Student's *t*-test.

mutation at codon Glu<sup>180</sup> (Johnston et al., 2000; Jin et al., 2003; Wei et al., 2014), the loss of slow TnT in Tnnt1-KO mice caused a drastic loss of type I slow fiber contents and a switch to more fast fiber contents in soleus muscle with significantly decreased resistance to fatigue (Wei et al., 2014). The maintained level of CAIII in Tnnt1-KO soleus muscle not only demonstrates that the expression of CAIII is not restricted to slow fibers in the soleus muscle but also suggests an adaptation to sustain fatigue tolerance.

The up-regulation of CAIII in fast fibers suggests a novel approach to compensate for the lost function in a sarcomeric myopathy. It is worth noting that the slow isoform of TnI remains

detectable in Tnnt1-KO soleus muscle (**Figure 11A**), indicating a maintained slow muscle tissue environment that may play a role in sustaining CAIII expression and a plausible foundation for exploring new treatment of TNNT1 myopathies.

In summary, our present study demonstrates that CAIII functions in skeletal muscle involving a regulation of CO<sup>2</sup> metabolism and intracellular pH environment, which contributes to the resistance to fatigue. In the meantime, CAIII expression is not solely dependent on muscle fiber types or developmental stages, but may be determined by functional features of the muscle such as slow muscles and functionally related fast muscles. Representing a novel mechanism to reduce muscle fatigue in physiological conditions and compensate for muscle function in diseases that lose slow fibers, the function and regulation of CAIII in skeletal muscle merits further investigation.

### AUTHOR CONTRIBUTIONS

HF: Performed experiments, designed and modified protocol, drafted the paper, edited text, and figures, approved submission. JJ: Designed research, drafted the paper, edited text, and figures, approved submission.

### ACKNOWLEDGMENTS

Research reported in this publication was supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases of the National Institutes of Health under Award Number AR048816 (to JJ) The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2016.00597/full#supplementary-material

Supplement Figure 1 | Young adult Car3-KO mice had normal mass and fiber size in TA and EDL muscles. (A) 2-month old male *Car3*-KO and WT mice showed similar body weight. (B) Normalized to body weight, the weight of *Car3*-KO and WT TA and EDL muscles are nearly identical. (C) H&E stained cross sections showed similar fiber cross-sectional areas in WT and *Car3*-KO mouse TA muscles with no signs of degeneration in *Car3*-KO TA muscle. (D) Quantification by normalizing to the body weight showed that the cross-sectional areas of *Car3*-KO and WT TA muscles had significant difference. The data are presented as mean ± SE. For (A,B), *n* = 8 mice in WT TA group and *n* = 7 mice in *Car3*-KO

TA group; *N* = 5 mice in WT EDL and *n* = 7 mice in *Car3*-KO EDL groups. For (C,D), *n* = 4 mice each in WT and *Car3*-KO groups. Statistical analysis was performed using Student's *t-*test.

Supplement Figure 2 | In situ fatigability test of mouse TA and EDL muscles showed no increase of resting tension. The resting tension of WT

### REFERENCES


and *Car3*-KO mouse TA (A) and EDL (B) muscles during in situ fatigue contractions showed that under *in vivo* physiological systemic pH environment, *Car3*-KO and WT mouse TA and EDL muscles all had no increase in resting tension during fatigue contractions. *N* = 4 in WT TA and WT EDL groups. *N* = 3 in *Car3*-KO TA and 5 in *Car3*-KO EDL groups. Statistical analysis was performed using two-way ANOVA with adjusted mean comparison of Bonferroni test.

development, and life span. Mol. Cell. Biol. 24, 9942–9947. doi: 10.1128/MCB. 24.22.9942-9947.2004


types of human psoas muscle. Histochemistry 97, 77–81. doi: 10.1007/BF002 71284

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Feng and Jin. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Myofilament Calcium Sensitivity: Role in Regulation of In vivo Cardiac Contraction and Relaxation

Jae-Hoon Chung1, 2, <sup>3</sup> , Brandon J. Biesiadecki 1, 2, Mark T. Ziolo1, 2, Jonathan P. Davis 1, 2 and Paul M. L. Janssen1, 2, 4 \*

*<sup>1</sup> Department of Physiology and Cell Biology, The Ohio State University Wexner Medical Center, Columbus, OH, USA, <sup>2</sup> Dorothy M. Davis Heart and Lung Research Institute, The Ohio State University Wexner Medical Center, Columbus, OH, USA, <sup>3</sup> Medical Scientist Training Program and Biomedical Sciences Graduate Program, The Ohio State University Wexner Medical Center, Columbus, OH, USA, <sup>4</sup> Department of Internal Medicine, The Ohio State University Wexner Medical Center, Columbus, OH, USA*

#### Edited by:

*P. Bryant Chase, Florida State University, USA*

### Reviewed by:

*Beata M. Wolska, University of Illinois at Chicago, USA David Grant Allen, University of Sydney, Australia Rosana A. Bassani, University of Campinas, Brazil Bertrand C. W. Tanner, Washington State University, USA*

> \*Correspondence: *Paul M. L. Janssen janssen.10@osu.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *07 September 2016* Accepted: *07 November 2016* Published: *16 December 2016*

#### Citation:

*Chung J-H, Biesiadecki BJ, Ziolo MT, Davis JP and Janssen PML (2016) Myofilament Calcium Sensitivity: Role in Regulation of In vivo Cardiac Contraction and Relaxation. Front. Physiol. 7:562. doi: 10.3389/fphys.2016.00562* Myofilament calcium sensitivity is an often-used indicator of cardiac muscle function, often assessed in disease states such as hypertrophic cardiomyopathy (HCM) and dilated cardiomyopathy (DCM). While assessment of calcium sensitivity provides important insights into the mechanical force-generating capability of a muscle at steady-state, the dynamic behavior of the muscle cannot be sufficiently assessed with a force-pCa curve alone. The equilibrium dissociation constant (Kd) of the force-pCa curve depends on the ratio of the apparent calcium association rate constant (kon) and apparent calcium dissociation rate constant (koff) of calcium on TnC and as a stand-alone parameter cannot provide an accurate description of the dynamic contraction and relaxation behavior without the additional quantification of kon or koff, or actually measuring dynamic twitch kinetic parameters in an intact muscle. In this review, we examine the effect of length, frequency, and beta-adrenergic stimulation on myofilament calcium sensitivity and dynamic contraction in the myocardium, the effect of membrane permeabilization/mechanical- or chemical skinning on calcium sensitivity, and the dynamic consequences of various myofilament protein mutations with potential implications in contractile and relaxation behavior.

Keywords: muscle, twitch, kinetics, desensitize, sensitize

### INTRODUCTION

In this review, we discuss the three major mechanisms (Frank-Starling mechanism, heart rate/frequency-dependent contraction, and beta-adrenergic stimulation) that govern cardiac output as well as affect calcium sensitivity, mainly at the level of troponin C, compare calcium sensitivity measurements in skinned/permeabilized and intact muscle preparations, in order to shed light on myofilament protein mutations that have the potential to be translated to further our understanding of cardiac physiology in vivo. We recognize that drugs, metabolites, pH, etc., also critically impact on calcium sensitivity (and potentially cardiac output), but that these are typically not primarily resulting from sarcomeric mutations, and were deemed beyond the scope of this review.

At the beginning of the cardiac contraction cycle, calcium ions enter cardiomyocytes via voltageactivated L-type calcium channels, leading to calcium-induced calcium release (CICR) from the sarcoplasmic reticulum (SR). The release of calcium from the SR increases free calcium ion

**235**

concentration from approximately 100 nM to 1 µM, making more calcium available for binding to troponin C (TnC), a subunit in the troponin complex. Calcium ions binding to TnC initiates a cascade of events leading to force generation via interaction between thin and thick filaments, i.e., by the cycling of cross-bridges. In order to relax, calcium must come off TnC to cease activation, and to allow dissociation of thin and thick filaments to occur and relax the muscle. Calcium ions are recycled back into the SR via SR Ca2<sup>+</sup> ATPase (SERCA) or extruded out of the cell via Na+/Ca2<sup>+</sup> exchanger (NCX).

Myofilament calcium sensitivity is a concept used by researchers to simplify the complex, dynamic process of cardiac contraction, and relaxation into a relationship between the concentration of free calcium ions available for binding to TnC and the amount of force generated by the muscle. In failing myocardium, calcium sensitivity has been reported to either increase or decrease depending on the etiology of the disease (Willott et al., 2010). As the calcium sensitivity increases, the contractility of the muscle typically increases, but this also means that relaxation may be often impaired if calcium dissociates from TnC more slowly. The vast majority of previous studies have utilized mechanically and chemically skinned muscle preparations for measuring calcium sensitivity because they are able to reduce a complex system into one that only contains two variables: free calcium ions and force of contraction by the myofilaments. This reductionist approach has revealed important mechanical properties of the thin and thick filaments but does not sufficiently translate into the contracting heart. The forcegenerating capacity of cardiac muscle in vivo takes into account not only the simple association and dissociation rate of calcium from TnC but the entire intracellular environment that includes various kinases and phosphatases, for example. From previous studies, it is clear that the myofilaments play an integral role in cardiac muscle contraction and relaxation. Therefore, the myofilaments are an important target in treatment of heart failure, which continues to afflict millions of lives today with limited treatment options. It is imperative that we utilize the invaluable knowledge the cardiac muscle physiology field has already generated regarding calcium sensitivity and produce new data to not only further our understanding of the physiology of a dynamically contracting heart in vivo but also more effectively translate our findings to the clinic.

### CALCIUM SENSITIVITY AND DYNAMIC BEHAVIOR OF A MUSCLE

A typical approach to assess myofilament calcium sensitivity is via construction of a force-pCa curve and determining a potential left- or right-ward shift of the curve (**Figure 1**). A leftward shift indicates an increased calcium sensitivity, as a given steady-state force can be attained using a lower concentration of free calcium. On the other hand, a right-ward shift indicates a decreased calcium sensitivity, as a muscle requires a higher concentration of free calcium to generate a given steady-state force. A deeper insight into this steady-state model reveals that, while a change in myofilament calcium sensitivity can reflect

altered dynamic behavior, one must also know at least one additional parameter to do so. The equilibrium dissociation constant, Kd, of TnC is a ratio between the calcium association rate constant to TnC (kon) and the calcium dissociation rate constant from TnC (koff) (**Figure 1**). TnC however does not work in isolation (Davis and Tikunova, 2008; Biesiadecki et al., 2014). There are many factors that collaboratively change the sensitivity of the myofilament activation and deactivation by calcium. No current models fully explain the complex integration of all components on the governing of thin-filament calcium binding (see Siddiqui et al., 2016). Thus, for the remainder of this review, we will discuss on on-rate (kon) and off rate (koff), as the apparent on- and off-rates of the myofilament system, reflecting the effective on- and off-rates of myofilament activation and deactivation not necessarily reflecting solely Ca2<sup>+</sup> binding to TnC.

Myofilament calcium sensitivity increases when the kon increases relative to the koff, resulting in an overall decrease in Kd. In other words, the kon does not necessarily have to increase to increase TnC's calcium sensitivity. As long as the koff decreases by a larger percentage compared to the kon, one would observe an increase in calcium sensitivity. This is an important distinction because having an absolute increase in the kon would lead to increased activation of the myofilament and thus increased force generation in our model of cardiac muscle twitch (**Figure 2**). Our model is written in Labview (National Instruments) and uses a simple mathematical calcium transient: [Ca2+]<sup>i</sup> = Amplitude<sup>∗</sup> time<sup>∗</sup> e <sup>∧</sup>(-Downamplitude<sup>∗</sup> time/τ). This calcium transient (light blue trace in both **Figures 2**, **3**) with kinetic parameters that reflect literature values drives, via on and off rate, the thin filament activation level (reflecting TnC-Ca2<sup>+</sup> binding). This thin filament activation allows cross-bridge formation using the simple 2 state model, governed by an onrate (f), and an off-rate (g). The model incorporates cross-bridge attachment and detachment rates and thin filament activation levels to generate twitches in real time. Our program allows us to change various parameters such as temperature, calcium transient relaxation constant, cross-bridge attachment rate, and cross-bridge detachment rate. In all the used simulations in this review (**Figures 2**, **3**), all parameters other than the calcium-TnC kon and koff rates were kept constant. Our model reports various twitch kinetic parameters such as time-to-peak (TTP), and

FIGURE 2 | A set of hypothetical twitches generated using a Labview (National Instruments) program demonstrating the possible effects of altered calcium sensitivity on twitch kinetics. If the decrease in calcium sensitivity (increase in Kd) is primarily due to decreased kon, one would observe lower developed force (pink). If the decrease in calcium sensitivity is primarily due to increased koff, one would observe lower developed force and faster relaxation kinetics (gray). If the increase in calcium sensitivity (decrease in Kd) is primarily due to decreased koff, one would observe increased developed force and slower relaxation kinetics (dark blue). If the increase in calcium sensitivity (increase in Kd) is primarily due to increased kon, one would observe increased developed force and faster relaxation compared to the case where koff is decreased (green). Calcium transient (light blue) and original twitch (red) are also included in the figure.

relaxation to 50% (RT50) in real time. We generated our cardiac twitches by initially changing these parameters to best mimic the typical cardiac muscle twitch kinetics we have observed in intact human trabeculae (Milani-Nejad et al., 2015). It is believed that the contraction kinetics of a muscle are much slower than the kon, which has traditionally been believed to be diffusionlimited, and that changes in the kon do not affect the contraction kinetics (Bers, 2001; Davis and Tikunova, 2008). Therefore, an increase in the kon would result in an increase in developed tension but not necessarily in faster contraction kinetics. On the other hand, having an absolute decrease in the koff would result in slowed relaxation (**Figure 3**). The rate-limiting steps of relaxation kinetics are complex and involve several distinct processes that at least partially overlap in time. As reviewed in previous literature, the main processes involved are thought to be the decline of the intracellular calcium concentration, transient calcium coming off TnC (koff), and cross-bridge cycling kinetics (Biesiadecki et al., 2014). Therefore, a decrease in the koff of TnC could slow down relaxation kinetics. An increase in the kon or a decrease in the koff would culminate in an increased calcium sensitivity but would have drastically different effects on dynamic twitch kinetics.

Despite the unchanged calcium sensitivity, significant consequences for dynamic contraction can occur when the kon and koff increase or decrease by the same factor (**Figure 3**). This would result in no apparent change in steady-state calcium sensitivity, however, if the kon and koff both increase, the muscle would develop a higher force and relax faster. If the kon and koff both decrease, the muscle would develop a lower force and relax

FIGURE 3 | A set of hypothetical twitches generated using a Labview (National Instruments) program with the same Kd demonstrating the effect of modulating kon and koff. When kon and koff are both increased by the same factor to yield the same Kd, the contraction and relaxation kinetics speed up (dark blue) and begins to more closely resemble the kinetics of the calcium transient (light blue). When kon and koff are both decreased by the same factor to yield the same Kd, the muscle cannot relax completely at steady-state and has a lower developed force (gray). The original twitch (red) has the same Kd as the other two tracings (dark blue and gray).

slower. Again, this type of hypothetical analysis reveals that a second important reason that simply measuring the myofilament calcium sensitivity as a stand-alone measurement is not sufficient for translation into dynamic contraction and relaxation.

Another way to determine the dynamic contraction and relaxation behavior of a muscle is by measuring the twitch kinetic parameters of an intact muscle. This information, combined with assessments of force-pCa (intact or skinned), will provide supplemental information on the developed tension as well as the contraction and relaxation kinetics, which can also allow, albeit indirect, more insight into the relative contribution of kon and koff, as changes in kon would more heavily influence the developed tension while koff the relaxation kinetics. In summary, the assessment of a force-pCa curve of a muscle at steady-state does not contain sufficient information to make an unambiguous translation into dynamic behavior of a muscle. When dynamic twitch kinetic parameters are not available, but kon and/or koff are assessed in addition to assessment of steady-state myofilament sensitivity, the translation of skinned fiber data toward potential dynamic behavior is greatly enhanced.

To our knowledge, it is virtually impossible to assess the kon and koff in intact muscles, as researchers have typically used in vitro fluorimetry to measure changes in fluorescence in isolated TnC at steady-state, but the knowledge of the two rate constants can yield important insights into the contraction and relaxation kinetics of a working muscle (Tikunova and Davis, 2004). Since the dynamic behavior is critically important in various disease states, assessment of kinetics, as well as kinetic reserve is critically important to further understand cardiac malfunction in disease (Janssen et al., 2016).

### REGULATION OF CALCIUM SENSITIVITY VIA LENGTH, FREQUENCY, AND BETA-ADRENERGIC ACTIVATION

Cardiac output is heavily regulated, and this regulation is mainly governed by three mechanisms: length-dependent activation, frequency-dependent activation, and beta-adrenergic activation (Janssen, 2010). These three factors all encompass modulation of myofilament calcium sensitivity, which has important implications for the dynamic behavior of cardiac muscle, and we will discuss each of these factors below in more detail.

### Length-Dependent Activation

The Frank-Starling mechanism is an inherent property of the heart that allows an increase in stroke volume as ventricular volume increases during diastole (Frank, 1895; Knowlton and Starling, 1912). At the level of cardiac muscle, as a muscle is stretched, its developed force per cross-sectional area increases, and this phenomenon is known as length-dependent activation. This effect is beneficial in a cardiac cycle because the ventricular walls are most stretched at the end of diastole, i.e., at the end of the filling phase. The Frank-Starling effect allows the heart to pump the blood to both the lungs and the body with an increased amount of pressure when it is most needed (when the ventricles are more filled with blood). It has been shown by many groups over the past decades that increased muscle length leads to increased overall myofilament calcium sensitivity (Hibberd and Jewell, 1982; Harrison et al., 1988; Dobesh et al., 2002; Herron et al., 2006; Edes et al., 2007). In intact rat and human trabeculae, increased muscle length resulted in increased developed force, as expected, but also results in a slower contraction and relaxation kinetics, exhibiting increased TTP, decreased +dF/dt/F, increased time from peak tension to 50% relaxation (RT50), and decreased −dF/dt/F (Milani-Nejad et al., 2013, 2015). It has been noted that there is no significant increase in intracellular calcium concentration during the fast phase after a stretch, but there is a slow increase in intracellular calcium concentration and developed force during the slow phase, which could account for increased developed tension (Allen and Kurihara, 1982). On the myofilament side, a decrease in lattice spacing (which occurs as a muscle is stretched) has been reported to result in increased calcium sensitivity (Wang and Fuchs, 1995). However, another study has found no increase in calcium sensitivity due to decreased lattice spacing (Konhilas et al., 2002). The length dependence of calcium sensitivity may also involve the number of attached cross-bridges, as it has been noted that the length-dependence of calcium sensitivity disappears when vanadate was used to prevent actinmyosin interaction (Hofmann and Fuchs, 1987). Interestingly, Allen and Kentish noted that calcium sensitivity continues to increase even when a muscle is stretched beyond optimal length, which should result in a decrease in the number of attached cross-bridges (Allen and Kentish, 1985). The phosphorylation status of myofilament proteins such as troponin I or myosin binding protein C (cMyBP-C) can play a role in lengthdependent activation (Wijnker et al., 2014; Mamidi et al., 2016). Furthermore, the protein titin is thought to play a significant role in the process of length-dependent activation (Ait-Mou et al., 2016). The overall increase in developed tension and slower relaxation kinetics at greater muscle lengths suggest that the length-dependent increase in calcium sensitivity is probably due to an increase in the kon of calcium as well as a larger decrease in the koff.

### Frequency-Dependent Activation

The frequency of contraction itself has an effect on the amount of developed force in the heart. Increased stimulation frequency results in modification of developed tension, and this is known as the Bowditch effect (Bowditch, 1871). Typically, large animals such as rabbits and humans exhibit a positive force frequency relationship (FFR) (Endoh, 2004). In addition, researchers have noted an accelerated rate of relaxation at increased stimulation, also known as frequency-dependent acceleration of relaxation (FDAR) (Kassiri et al., 2000; DeSantiago et al., 2002). FDAR is required in muscles because the cardiac muscle must return to its relaxed state faster at high heart rates, as it spends less time in diastole. Varian and co-workers have found in intact rabbit trabeculae that increased stimulation frequency leads to FDAR as well as decreased calcium sensitivity (Varian and Janssen, 2007). At least in larger mammals, the decreased calcium sensitivity is accompanied by increased developed tension (due to increased intracellular calcium concentration) and increased rate of relaxation. This suggests that the koff is increased to result in calcium desensitization. Varian et al. has found that troponin I (TnI) and myosin light chain-2 phosphorylation are significantly increased as stimulation frequency is increased from 1 to 4 Hz in intact rabbit trabeculae (Varian and Janssen, 2007). Although Varian et al. did not investigate the phosphorylation of specific amino acid residues in TnI, the phosphorylation status of TnI stimulated at 4 Hz was not significantly different from that stimulated at 1 Hz with isoproterenol, which suggests that the increased TnI phosphorylation at 4 Hz may be primarily due to activation of the protein kinase A pathway (Varian and Janssen, 2007). Serine 23/24 are the most extensively characterized phosphorylation sites of TnI. Phosphorylation of serine 23/24 is known to desensitize the myofilament, which makes these two sites potential phosphorylated sites in the context of increased stimulation frequency (Layland et al., 2005). However, one study on intact rat trabeculae actually found no difference in myosin light chain 2 as well as TnI phosphorylation status at high stimulation frequency (9 Hz) compared to low stimulation frequency (1 Hz) (Lamberts et al., 2007). The effect of increased stimulation frequency on the phosphorylation status of myofilament proteins still remains unclear, and there is ample room for further investigation. The increase in developed tension and faster twitch kinetics at increased stimulation frequencies suggest that the kon and koff are both increased. The overall desensitization of the myofilament at increased stimulation frequencies is likely due to a greater increase in the koff than in the kon to result in overall increase in Kd.

### Beta-Adrenergic Activation

When our body is under stress, the adrenal gland releases hormones such as epinephrine and norepinephrine to cope with the stress. One of the effects of these hormones is activation of the beta-adrenergic pathway in the cardiomyocytes, predominantly via the beta1 receptor. This is a useful mechanism for the heart to increase its contractile force, heart rate, and contraction and relaxation kinetics as the demand for oxygen is increased in the body. It is generally accepted in the literature that calcium sensitivity is decreased in response to beta-adrenergic activation (Herzig and Rüegg, 1980; Strang et al., 1994; de Tombe and Stienen, 1995). This is primarily due to the phosphorylation of serine 23/24 in TnI (Layland et al., 2005). Myosin binding protein C is reported to be involved as well, as Cazorla et al. has reported that cMyBP-C knock-out mice had blunted PKAdependent desensitization (Cazorla et al., 2006). The reduction in calcium sensitivity is accompanied by increased developed force as well as faster contraction and relaxation kinetics, which may suggest that the kon and koff may both increase but the koff has a proportionately larger increase (Zhang et al., 1995; Milani-Nejad et al., 2015). However, the increase in developed force is believed to be predominantly due to an increase in intracellular calcium concentration, rather than the change in myofilament calcium sensitivity (Roof et al., 2011). Robertson et al. have explored the effect of phosphorylation of TnI on the kon and koff and saw that the koff significantly increased upon phosphorylation of TnI but the kon remained the same (Robertson et al., 1982).

### CALCIUM SENSITIVITY IN SKINNED vs. INTACT CARDIAC MUSCLE

Most previous studies that assessed myofilament calcium sensitivity have utilized skinned muscle preparations, reporting approximate half-max force at pCa of 6 (EC50). However, intact muscles have been reported to exhibit higher calcium sensitivity compared to skinned muscles (Gao et al., 1994; Varian et al., 2006; Monasky et al., 2010). Later studies, in intact muscle at physiological temperature (Varian et al., 2006; Monasky et al., 2010) confirmed a high sensitivity for calcium in intact muscle compared to published values in skinned/permeabilized muscle. After the muscle skinning process, many intracellular components excluding the myofilament are lost. This naturally leads one to wonder what sensitizing intracellular components are lost during the process of muscle skinning. One possibility might be various kinases that increase the calcium sensitivity of myofilaments. For example, protein kinase D (PKD) has been reported to increase calcium sensitivity via phosphorylation of Ser<sup>315</sup> cardiac myosin binding protein C (cMyBP-C) (Dirkx et al., 2012). However, PKD can also reduce calcium sensitivity via phosphorylation of troponin I (TnI) Ser23/<sup>24</sup> (Cuello et al., 2007). It may be possible that there are "natural" calcium sensitizers other than kinases that are (partially) lost upon skinning. Carnosine-like compounds and taurine are examples of cytosolic compounds that have been shown to alter myofilament calcium sensitivity that may be lost during permeabilization (Steele et al., 1990; Lamont and Miller, 1992). Recently, S-glutathionylation of cMyBP-C as well as phosphorylation of TnI by adenosine monophosphate (AMP) kinase have been shown to increase calcium sensitivity (Nixon et al., 2012; Patel et al., 2013). Phosphatase 2A (PP2A), which is associated with various calcium handling and myofilament proteins such as the L-type calcium channel and myosin light chain 2 (MLC-2), also increases calcium sensitivity (Wijnker et al., 2011).

However, it is important to note that there are many desensitizing cytosolic components as well. For example, other kinases such as protein kinase A (PKA) and protein kinase C (PKC) have both been reported to decrease calcium sensitivity, not increase it (Herzig and Rüegg, 1980; Strang et al., 1994; de Tombe and Stienen, 1995; van der Velden et al., 2006). It has been well-documented that protein kinase A (PKA) phosphorylation of TnI 23/24 results in desensitization of myofilament (Layland et al., 2005). Protein kinase C (PKC) and protein kinase G (PKG) can phosphorylate myofilament proteins such as cMyBP-C to reduce calcium sensitivity (Pfitzer et al., 1982; van der Velden et al., 2006). O-linked N-acetyl-D-glucosaminylation of cardiac myofilament also decreases calcium sensitivity (Ramirez-Correa et al., 2008).

Another aspect to consider with the skinning procedure is that sarcomeric lattice spacing increases due to the procedure (Irving et al., 2000). A number of studies have reported that decreased lattice spacing (analogous to increased muscle length) leads to increased calcium sensitivity (McDonald and Moss, 1995; Wang and Fuchs, 1995). The increase in sarcomeric lattice spacing in skinned muscle preparations may lead to decreased calcium sensitivity. However, Konhilas et al. (2002) have reported that osmotic compression of the lattice spacing does not affect the length-calcium sensitivity relationship.

It is not clear at the moment whether desensitizing or sensitizing cytosolic components play a bigger role in intact muscle preparations. In addition, most of the investigations on these cytosolic components were performed on skinned muscle in vitro, which makes it difficult for one to predict their actual roles in vivo. However, one must carefully consider the implication of the loss of cytosolic signaling molecules and the changes in the myofilament geometry on their direct or indirect effect on calcium sensitivity during the permeabilization process.

### MODIFICATION OF CALCIUM SENSITIVITY AND DYNAMIC BEHAVIOR OF A MUSCLE VIA MYOFILAMENT PROTEIN MUTATIONS

Myofilament proteins work in conjunction to allow the cardiac muscle to contract and relax in response to changes in intracellular free calcium ion concentration. Therefore, it is not surprising that genetic mutations in many of the myofilament proteins impact calcium sensitivity. However, only a few mutations have been characterized to show their translation into dynamic behavior of a muscle. For the purpose of characterization of dynamic contraction of a muscle in vivo, it is necessary but not sufficient to show changes in calcium sensitivity. One must also either report biochemical changes in the kon or koff or report twitch force development and kinetics in an intact muscle to show what exact contributing factors changed calcium sensitivity. In this review, we highlight a few mutations that have originally been found in patients with dilated cardiomyopathy (DCM) or hypertrophic cardiomyopathy (HCM) and discuss how these mutations affect myofilament calcium sensitivity and dynamic behavior.

### Troponin C

Troponin C (TnC) is the "calcium sensor" of the myofilament that directly binds calcium at its N-terminus domain to cause a cascade of conformational shifts of myofilament proteins to generate force. TnC L29Q mutation is the first TnC mutation found in a HCM patient, and Liang et al. (2008) have shown that the mutation increases calcium sensitivity in recombinant skinned mouse cardiomyocytes. In addition, the investigators reported an increase in kon but no change in koff, which suggests that developed force would be increased in but relaxation kinetics would not be affected in a dynamically contracting muscle (Liang et al., 2008). Interestingly, a later study found no changes in calcium sensitivity in mouse papillary muscles reconstituted with TnC L29Q, and another study actually reported a decrease in calcium sensitivity (Neulen et al., 2009; Gollapudi and Chandra, 2012).

### Troponin I

Troponin I (TnI) is a myofilament protein that inhibits actin and myosin binding by binding to actin in the absence of calcium binding to TnC. Upon binding of calcium to TnC, it releases actin and binds to the hydrophobic patch in the Nterminal domain of TnC to allow interaction between myosin and actin. It has potential implications in the development of HCM, as it has been found in 7% familial HCM (Richard et al., 2003). The TnI R145G HCM mutation increases calcium sensitivity, which was attributed either to increased cross-bridge cycling kinetics or to decreased calcium koff from TnC (Wen et al., 2008). They found that cross-bridge cycling kinetics did not change in the transgenic mice and therefore concluded that decreased koff was the main reason for the increase in calcium sensitivity. Wen et al. also reported reduced maximal force in the TnI R145G mice, which suggests that the kon was probably reduced and at least not increased. This leaves the decreased koff as the main contributor for the increased calcium sensitivity. In addition, the relaxation kinetics were slower in the TnI R145G mice papillary muscle, further supporting the investigators' notion that the koff was decreased (Wen et al., 2008). Since TnI has multiple active phosphorylation sites, cross-talk between different phosphorylation sites adds an additional layer of regulation (also see Salhi et al., 2016).

### Troponin T

Troponin T (TnT) interacts with TnI, TnC, Tm, and actin and therefore can regulate the activity of many myofilament proteins (Gordon et al., 2000). Considering its central position in the troponin complex, it is not surprising that 15% of familial HCM patients exhibit mutations in TnT (Sheng and Jin, 2014). The work by Sommese et al. has revealed that R141W and R173W DCM mutations lead to decreased calcium sensitivity, increased Kd, and increased koff (Sommese et al., 2013). This suggests that the decrease in calcium sensitivity is due to increased koff. Deletion of TnT K210, a mutation found in DCM patients, results in desensitization of myofilament in a knock-in mouse model (Du et al., 2007). The investigators also used intact left ventricular papillary muscle to assess its twitch kinetics and reported no change in developed force but faster relaxation (Du et al., 2007). Although the kon and koff were not determined, one can infer from the twitch kinetics data that the decreased calcium sensitivity and faster relaxation kinetics can probably be attributed to the increased koff.

### Myosin Heavy Chain

Myosin heavy chain (MHC) is the force-generating myofilament protein that undergoes power strokes due to its conformational shift. It has two isoforms: α-myosin, the faster isoform, and the βmyosin, the slower isoform. Large mammals such as rabbits and humans express predominantly β-myosin, and small mammals such as mice and rats predominantly express α-myosin (Hoh et al., 1978). Approximately 41% of familial HCM patients have a mutation in the β-myosin heavy chain gene, MYH7 (Richard et al., 2003). A study by Blanchard et al. reported increased calcium sensitivity due to familial HCM mutation R403Q in mouse papillary muscle (Blanchard et al., 1999). However, another study by Palmer et al. did not find any significant changes in calcium sensitivity (Palmer et al., 2008). Chuan et al. measured twitch kinetic parameters at the single cardiomyocyte level and found that the developed force did not change but relaxation kinetics such as RT50 were significantly slower (Chuan et al., 2012). Another study by Kim et al. reported slower contraction and relaxation kinetics in mouse cardiomyocytes, but the cells were unloaded and therefore could not yield information regarding isometric force production (Kim et al., 1999). These results together suggest that the increased calcium sensitivity reported by Blanchard et al. is likely due to decreased koff.

### Myosin Regulatory Light Chain

Myosin regulatory light chain (RLC), also known as myosin light chain-2 (MLC-2), is part of the myosin protein that modulates cardiac contraction. Phosphorylation of MLC-2 by myosin light chain kinase (MLCK) is the mechanism via which MLC-2 can affect force development and cross-bridge cycling (Moss and Fitzsimons, 2006). However, mutations in MLC-2 can also influence cardiac contraction and myofilament calcium sensitivity. For example, E22K mutation is one of the first mutations found in familial HCM that culminates in increased calcium sensitivity in skinned glycerinated mouse left ventricular papillary muscle (Szczesna-Cordary et al., 2005). A subsequent study by the same group in freshly skinned mouse left ventricular papillary muscle actually found no significant change in calcium sensitivity due to the MLC-2 E22K mutation (Szczesna-Cordary et al., 2007). This study also used intact papillary muscle and reported decreased force development and faster relaxation kinetics (Szczesna-Cordary et al., 2007). Based on the twitch kinetics data in the 2007 paper, one would predict that kon to decrease and koff to increase to result in an increase in K<sup>d</sup> and a decrease in calcium sensitivity. However, 2005 and 2007 papers reported either an increase in calcium sensitivity or no change. Further investigation is needed to clearly determine the effect

### REFERENCES


of the E22K on the dynamic twitch kinetics of a cardiac muscle.

### CONCLUSION

The myofilament is crucial in the regulation of contractile and relaxation behavior of the cardiac muscle, especially in the pathophysiology of heart failure. Force-pCa curves generated from skinned muscle preparations are able to reduce the complex environment of a muscle into a much simpler relationship between isometric force and free calcium ions. While such data are necessary and important, as a stand-alone assessment however, the knowledge of myofilament calcium sensitivity alone is not sufficient for the extrapolation to dynamic behavior of a muscle representative of that in vivo. Heart failure continues to be one of the leading causes of death in the U.S., and standardof-care treatment of the disease has been largely limited to beta blockers, diuretics, angiotensin converting enzyme inhibitors, and calcium channel blockers, which have been in use for decades. More studies on calcium sensitivity that incorporate either assessment of kon or koff, or assess dynamic behavior are needed in the field of cardiac physiology to improve interpretation of the impact of myofilament mutations, and for strategizing of novel treatment for the patients who continue to suffer from the disease.

### AUTHOR CONTRIBUTIONS

PJ: concept of review, concept of illustrations, edited final draft. JC: wrote initial draft, made illustrations. MZ, BB, JD: helped discuss concept, edited final draft.

### ACKNOWLEDGMENTS

The authors are funded by grants from the National Institutes of Health R01HL113084 (to PJ), R56HL091986 (to JD), R01HL114940 (to BB), and from the American Heart Association 16GRNT27760114 (to MZ).


Frank, O. (1895). Zur Dynamik des Hersmuskels. Zeitschrift Biologie 32, 370–447.


regulator of myocardial contractile function. Circ. Res. 103, 1354–1358. doi: 10. 1161/CIRCRESAHA.108.184978


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Chung, Biesiadecki, Ziolo, Davis and Janssen. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Restrictive Cardiomyopathy Caused by Troponin Mutations: Application of Disease Animal Models in Translational Studies

Xiaoyan Liu<sup>1</sup> , Lei Zhang<sup>1</sup> , Daniel Pacciulli <sup>2</sup> , Jianquan Zhao<sup>3</sup> , Changlong Nan<sup>2</sup> , Wen Shen<sup>2</sup> , Junjun Quan<sup>1</sup> , Jie Tian<sup>1</sup> \* and Xupei Huang<sup>2</sup> \*

<sup>1</sup> Cardiovascular Research Laboratory, Division of Cardiology, Chongqing Medical University Children's Hospital, Chongqing, China, <sup>2</sup> Department of Biomedical Science, Charles E. Schmidt College of Medicine, Florida Atlantic University, Boca Raton, FL, USA, <sup>3</sup> Department of Cardiology, Bayannaoer City Hospital, Bayannaoer, China

#### Edited by:

P. Bryant Chase, Florida State University, USA

#### Reviewed by:

Aldrin V. Gomes, University of California, Davis, USA Kristina Bezold Kooiker, Stanford University, USA

#### \*Correspondence:

Jie Tian jietian@cqmu.edu.cn Xupei Huang xhaung@fau.edu

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 31 July 2016 Accepted: 02 December 2016 Published: 19 December 2016

#### Citation:

Liu X, Zhang L, Pacciulli D, Zhao J, Nan C, Shen W, Quan J, Tian J and Huang X (2016) Restrictive Cardiomyopathy Caused by Troponin Mutations: Application of Disease Animal Models in Translational Studies. Front. Physiol. 7:629. doi: 10.3389/fphys.2016.00629 Cardiac troponin I (cTnI) plays a critical role in regulation of cardiac function. Studies have shown that the deficiency of cTnI or mutations in cTnI (particularly in the C-terminus of cTnI) results in diastolic dysfunction (impaired relaxation) due to an increased myofibril sensitivity to calcium. The first clinical study revealing the association between restrictive cardiomyopathy (RCM) with cardiac troponin mutations was reported in 2003. In order to illustrate the mechanisms underlying the cTnI mutation caused cardiomyopathy, we have generated a cTnI gene knockout mouse model and transgenic mouse lines with the reported point mutations in cTnI C-terminus. In this paper, we summarize our studies using these animal models from our laboratory and the other in vitro studies using reconstituted filament and cultured cells. The potential mechanisms underlying diastolic dysfunction and heart failure caused by these cTnI C-terminal mutations are discussed as well. Furthermore, calcium desensitizing in correction of impaired relaxation in myocardial cells due to cTnI mutations is discussed. Finally, we describe a model of translational study, i.e., from bedside to bench and from bench to bedside. These studies may enrich our understanding of the mechanism underlying inherited cardiomyopathies and provide the clues to search for target-oriented medication aiming at the treatment of diastolic dysfunction and heart failure.

Keywords: myofibrils, troponin, mutation, cardiomyopathy, diastolic dysfunction, animal models

### INTRODUCTION

Cardiac cells (myocardium) consist of two filaments: Thick filament and thin filament. The former contains mainly myosin and myosin bind C-protein and the latter contains actin and troponintropomyosin complex. The so called cross-bridge formation between the thin filament actin and the thick filament myosin determines the filament movement, i.e., muscle contraction and relaxation. Whereas, the troponin complex plays an important role in regulation of the filament movement. Troponin complex consists of three subunit proteins: Troponin C (TnC), troponin T (TnT), and troponin I (TnI). Among them, TnC is a Ca2<sup>+</sup> binding protein, TnT binds with tropomyosin whereas TnI is an inhibitory subunit that can bind to actin-tropomyosin and prevent muscle contraction by inhibition of actin-tropomyosin- activated myosin (actomyosin) ATPase activity (Greaser et al., 1972; Ohtsuki and Shiraishi, 2002). Many studies have demonstrated that TnI has an important function in the regulation of striated muscle contraction and relaxation (Kranias and Solaro, 1982; El-Saleh et al., 1986; Zot and Potter, 1987; Solaro and Troponin, 1999; Konhilas et al., 2003).

It is well-known that cardiac muscle movement, i.e., contraction and relaxation, is regulated mainly by intracellular calcium. An increase of intracellular Ca2<sup>+</sup> concentration results in an enhanced cardiac contractility whereas the decreased Ca2<sup>+</sup> concentration can reduce the cardiac contractility. The concentration of intracellular Ca2<sup>+</sup> is regulated by various calcium handling proteins in myocardium, such as Ca2<sup>+</sup> channel receptors, the SERCA2a calcium ATPase pump, phospholamban, etc. (Gordon et al., 2000). Recently, a body of studies has demonstrated that cardiac TnI (cTnI) has unique functions in control of cardiac muscle contraction and relaxation, especially in diastolic function (Yasuda et al., 2007; Solaro et al., 2008). PKAmediated cTnI phosphorylation causes a decrease of myofibril sensitivity to Ca2<sup>+</sup> or a desensitization of the contractile apparatus to activation by Ca2<sup>+</sup> (Solaro et al., 1976; Rorbertson et al., 1982; Zhang et al., 1995; Chandra et al., 1997; Solaro, 2001; Metzger and Westfall, 2004; Periasamy and Janssen, 2008). Using a cTnI gene knockout mouse model that generated by Huang et al in 1999, we have demonstrated that impaired relaxation occurs in myocardial cells with a deficiency of TnI and sarcomere length from these cells is shortened due to an increased tension even in the absence of calcium, suggesting that cardiac TnI is critical for cardiac relaxation (Huang et al., 1999).

Physiologically, TnI plays such a critical role in regulation of cardiac function, especially the muscle relaxation. In the following sections, we will discuss the cardiac dysfunction caused by cTnI C-terminal structural changes, i.e., cTnI mutations, in the heart under pathological conditions. In addition, the potential mechanisms underlying the diastolic dysfunction are discussed as well.

### IN VITRO ASSAYS MEASURING THE EFFECTS OF cTnI MUTATIONS ON MYOFIBRIL FUNCTION

Many studies have confirmed that the C-terminal half of cTnI is more conserved than the N-terminal region of the protein (Wilkinson and Grand, 1978). The C-terminal part of cTnI contains specific regions that are crucial for the normal activity of the protein, in particular, cardiac relaxation. In a part of cTnI, there is an inhibitory region that is the minimum sequence necessary for inhibition of actomyosin ATPase activity. This domain includes residues from 147 to 163 and binds strongly to actin and the N terminal domain of TnC regulating the binding of Ca2<sup>+</sup> to TnC (Rieck and Dong, 2014). The region between the residues from 168 to 188 in cTnI is a second actin-binding site that binds specifically to the actin-tropomyosin filament and is known to contribute to the inhibitory activity of cTnI (Tripet et al., 1997). The remaining C-terminal domain, 192 to 210 is not fully characterized, however, some studies indicate that this part of cTnI plays a role in the stabilization of tropomyosin in the actin filament upon Ca2<sup>+</sup> activation (Galinska et al., 2010 ´ ).

The integrity of the cTnI molecule is essential for proper conformation of the troponin complex in the myofilament and the inhibition of actomyosin ATPase activity. It is of great importance, both scientifically and clinically, to elucidate the cellular mechanisms underlying RCM caused by cTnI mutations in order to identify the cause of cardiomyopathies and heart failure. The data from analyzing in vitro reconstituted thin filaments showed that the RCM cTnI mutations had high Ca2+ sensitizing effects on cardiac muscle force generation (Gomes et al., 2005; Kobayashi and Solaro, 2006). The reconstituted filament assays have the advantage of easily obtaining the mutated proteins and quickly testing the myofilament force generations. Very recently, this technique has been applied to explore the role of cardiac troponin I C-terminal mobile domain and linker sequence in regulating cardiac contraction (Meyer and Chase, 2016). Some research groups investigated the role of the mutated troponin in intact cells. Using an acute genetic engineering technique, Davis et al. transferred the mutant cTnI genes into cultured rat myocardial cells and found that the myofibril sensitivity to calcium was increased (Davis et al., 2007, 2008). Numerous mutations in the carboxyl half of the protein are associated with the development of cardiomyopathies further confirming the importance of the C terminal domains of cTnI for proper regulation of cardiac contraction (Chang et al., 2008; Tachampa et al., 2008). Drastic Ca2<sup>+</sup> sensitivity change has been reported in myofilament with cTnI K178E mutation (Yumoto et al., 2005). However, most of the RCM mutations in cTnI have not been incorporated into transgenic models and they have been just characterized in functional in vitro studies.

### CARDIOMYOPATHIES CAUSED BY cTnI MUTATIONS: TRANSLATIONAL STUDIES

Cardiomyopathies have been considered to represent diseases that primarily affect cardiac muscle. Based on their morphology and pathophysiology, three major types of cardiomyopathies are most prevalent: Hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM), and restrictive cardiomyopathy (RCM) (Rivenes et al., 2000). HCM is characterized by a hypertrophic heart and DCM is characterized by a dilated ventricle, which are relatively easier to be recognized clinically. However, RCM, unlike HCM and DCM, manifests itself as a restricted ventricle that prevents or reduces the blood return to the heart because of a stiffened ventricle (Rivenes et al., 2000). Among the three major types of cardiomyopathies, RCM cases are not as common as HCM or DCM, but the prognosis is poor and some RCM patients die in their childhood (Rivenes et al., 2000; Palka et al., 2003). The clinical features of RCM are described as biatrial dilation, along with normal left ventricular internal dimension characterized on echocardiography. A marked elevation of left ventricular end-diastolic pressure with a restricted left ventricular filling and decreased cardiac output are often observed in RCM patients (Ligi et al., 2003; Palka et al., 2003). In the past, most cardiomyopathy cases were described as idiopathic cardiomyopathies, i.e., etiology is unknown (Ammash et al., 2000; Ligi et al., 2003). Recently with the advancement of genetic and molecular biological techniques, we know that most cardiomyopathy cases are heritable and caused by a single gene mutation (Braunwald, 2008).

The first report on cTnI C-terminal mutations associated human restrictive cardiomyopathy (RCM) was in 2003 (Mogensen et al., 2003). In that study, six cTnI mutations (L144Q, R145W, A171T, K178E, D190G, and R192H) have been found to be associated with RCM. Among them, the two mutations K178E and R192H have the worst clinical phenotype (Mogensen et al., 2003).

Our laboratory has participated in the studies to define the effect of the troponin mutations on the development of diastolic dysfunction. We have generated transgenic (TG) mice (cTnI193His) modeling human RCM mutation cTnI R192H (cTnI R193H in mouse sequence) in the heart. In addition, our laboratory has created another TG mouse line containing the RCM cTnI K178E mutation reported by Mogensen et al. (2003). The transgenic animals (cTnI K179E in the mouse genome) presented drastic bi-atrial enlargement in the absence of ventricular hypertrophy and dilation. They presented similar hemodynamic characteristics to the cTnI193His animals in our laboratory confirming the development of RCM as a consequence of cTnI mutation. The cardiac dysfunction was severe in the animals as most of them died prematurely (Jean-Charles et al., 2008). The drastic hypersensitivity to Ca2<sup>+</sup> observed in myocardium from our transgenic animal models is very similar to that reported from the in vitro studies (Yumoto et al., 2005).

We have tried to understand the mechanisms underlying the development of RCM due to cTnI mutations using the transgenic mice (cTnI193His) expressing human RCM mutation cTnI R192H (cTnI R193H in mouse sequence) in the heart. Histological examination confirms that cTnI193His mice do not show cardiac hypertrophy or ventricular dilation. The general morphology of the ventricles from these mice is similar to that of a wild type heart. However, the enlargement of bi-atria, both right and left atria, is very dramatic, which is similar to that in human RCM patients carrying cTnI R192H mutation. Functional measurements on these mice indicate a diastolic dysfunction in the early stage and a diastolic heart failure in the late stage (Du et al., 2006). We have demonstrated that impaired relaxation is a main manifestation in the RCM cTnI transgenic mice (Du et al., 2008) and cTnI mutation caused myofibril Ca2<sup>+</sup> hypersensitivity is a key factor resulting in a delayed calcium dissociation from the myofilaments and a delayed relaxation time (Li et al., 2010).

Using this animal model of disease, we have performed a series of cell-based experiments to determine diastolic dysfunction and calcium dynamics at a single myocardial cell level. Meanwhile, we have tried to reveal the cellular mechanisms of myofilament dysfunction in myocardial cells isolated from RCM mouse heart with cTnI mutations. Furthermore, we have measured left ventricular pressure using a Millar catheter in RCM mice to demonstrate that the increased pressure in restricted ventricles is due to increased internal tension in the wall of the ventricles caused by the myofibril hypersensitivity to Ca2<sup>+</sup> (Zhang et al., 2015; Wang et al., 2016). Once we recognized that Ca2<sup>+</sup>

hypersensitivity was an important factor that is associated with impaired relaxation in myofibril cells resulting in a diastolic dysfunction in RCM mice with cTnI mutations, we have tried to reduce the hypersensitivity to calcium and hoped to reverse the phenotype in RCM mice. By crossing our cTnI193His RCM mice with another transgenic mouse line (cTnI-ND) that expresses the cTnI with N-terminal deleted in the heart, we discovered that the hyposensitivity caused by cTnI-ND favored a general balance of myofibril sensitivity to calcium in the heart and reversed the diastolic dysfunction and rescued RCM phenotype (Li et al., 2010). Our study has demonstrated that desensitization of myofibrils to calcium can be a therapeutic target for restrictive cardiomyopathy with diastolic dysfunction. Later, another study using a different mouse line also confirmed that reduction of myofibril sensitivity to calcium was able to correct diastolic dysfunction in mice suffering from HCM (Alves et al., 2014).

Another similar example of cTnI C-terminal mutationassociated diastolic dysfunction and hypersensitivity to Ca2<sup>+</sup> is cTnI R145W mutation. The mutation of cTnI R145W associated human RCM is first reported by Mogensen (Mogensen et al., 2003). Transgenic mice modeling human cTnI R145W was generated. Characterization of these cTnI R145W transgenic mice (Tg-R145W) has shown that the Tg-R145W myofibers have a large increase in the Ca2<sup>+</sup> sensitivity of both force development and ATPase (Wen et al., 2009). Recent study using the recombinant human cardiac sarcomeres containing cTnI R145W mutation confirms that cTnI R145W mutation induces an increase in myofilament Ca2<sup>+</sup> sensitivity by reducing the interaction between Helix-C of cTnC and cTnI (Dvornikov et al., 2016).

Increased Ca2<sup>+</sup> sensitivity in myofilaments with cTnI C-terminal mutations is a key feature in cardiac muscle pathology. Therefore, it is urgent and necessary to search and find Ca2<sup>+</sup> desensitizers that primarily affect myofilament sensitivity to Ca2+. So far, compounds with such properties are very scarce. Myosin inhibitors such as blebbistatin and 2, 3-butanedione monoxime (BDM) may alter myofilament sensitivity to Ca2<sup>+</sup> via their inhibitory effect on actomyosin cross-bridge formation (Gwathmey et al., 1991; Kettlewell et al., 2004). These myosin ATPase inhibitors, while useful in functional studies in vitro and ex vivo, are too toxic for therapeutic use in live experimental animals or humans (Gwathmey et al., 1991; Kettlewell et al., 2004; Dou et al., 2007). There is a great need to develop or find small molecules and chemical Ca2<sup>+</sup> desensitizers that can be used to alter myofibril sensitivity for Ca2+. Biological agents should be non-toxicity and have good bioavailability, but there are few Ca2<sup>+</sup> desensitizers possessing such qualities. The catechin, (-)-epigallocatechin-3 gallate (EGCg) has Ca2<sup>+</sup> desensitizing abilities via its interaction with cTnC (Liou et al., 2008; Robertson et al., 2009). This compound is the most abundant catechin in green tea and is credited for the numerous health benefits attributed to green tea consumption (Robertson et al., 2009). EGCg desensitizes thin filaments to Ca2<sup>+</sup> by forming a ternary complex with the C-terminal domain of troponin C and the anchoring region of cTnI (Liou et al., 2008). The affinity of TnC for Ca2<sup>+</sup> is

reduced as a result which facilitates cardiac relaxation (Liou et al., 2008). The ability of EGCg to correct myofilament Ca2<sup>+</sup> hypersensitivity and diastolic dysfunction has been demonstrated in a HCM mouse model confirming the therapeutic potential of that compound for diastolic dysfunction (Tadano et al., 2010).

In our recent study, we have reported that diastolic dysfunction is corrected in RCM mice after the treatment of EGCg for 3months, suggesting that desensitizer catechin extracted from green tea is helpful in correcting impaired relaxation caused by calcium hypersensitivity in cTnI193His RCM mice (Li et al., 2010). After our study, another group reported that green tea catechin could normalize the enhanced calcium sensitivity of myofilaments regulated by a HCM-associated mutation in human patient (Warren et al., 2015). These data confirm that desensitizing green extract catechin is able to reduce the hypersensitivity caused by cTnI mutations and correct the diastolic dysfunction. So far, we have received these cellbased and organ-based data from our studies using transgenic mouse models. It is difficult to obtain these data from human patient studies. This is a good model of translational study from bedside to bench as illustrated in **Figure 1**. The idea is that the physicians receive the disease information from the patients and the basic researchers use the information to create animal models of disease to confirm the clinical discovery and use the animal models to further explore the cellular and molecular mechanisms underlying the disorders. The data from basic research can provide information back to clinical studies and the treatment of the disease. For example, the data we have obtained so far could provide us with some clues for future clinical studies and disease treatment. Recently, we have collected samples from RCM patients in an outpatient department at a Children's Hospital in China. The data from genetic tests confirm that among five RCM patients, two patients carry 192 point mutation in cTnI gene and two carry a point mutation in myosin gene, and one patient with no detectable myofibril protein mutation. More experimental treatment data will be collected in this study from more RCM patients.

## CONCLUSIONS

Cardiac troponin plays a critical role in cardiac contraction and relaxation. cTnI deficiency or mutations are associated with RCM characterized with a diastolic dysfunction. The discovery of sarcomeric protein mutations responsible for the development of the disease helps in identifying the etiology of RCM and allows for the screening of potential RCM patients. These measures may facilitate early diagnostic of the disease and proper monitoring and management of RCM patients. It also paves the way for the development of transgenic animals with the RCM phenotype which will contribute greatly to a better understanding and characterization of the disease. In fact, RCM transgenic animals may provide a link in the translational study which is from bedside to bench and from bench to bedside. They will also be very useful for the trial of potential drugs or devices designed to correct the diastolic dysfunction associated with RCM. The lack of effective treatments and the unavailability of drugs that selectively correct the diastolic dysfunction of the restricted heart, make the development of new pharmacological agents an urgent necessity. Desensitizing green tea extract catechin has been proved to be useful in correcting hypersensitivity and reversing diastolic dysfunction both in RCM animal studies and in reconstitute myofilament assays using a cTnI with a point mutation from HCM patient. It seems promising to apply desensitizing green tea extract catechin in correcting impaired relaxation in RCM patients caused by troponin mutations.

### AUTHOR CONTRIBUTIONS

XL, DP, JZ, CN, WS, and XH participated in paper writing and LZ, JQ, and JT participated in patient sample collection and genotyping measurements.

### ACKNOWLEDGMENTS

This work was supported partly by grants from the NIH (GM-073621 and HL 112130-01), the AHA Southeast Affiliate (09GRNT2400138) and NSFC (31271218) to XH.

### REFERENCES


important role for troponin I in regulating relaxation in cardiac myocytes. Circ. Res. 101, 377–386. doi: 10.1161/CIRCRESAHA.106.145557


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Liu, Zhang, Pacciulli, Zhao, Nan, Shen, Quan, Tian and Huang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Myofilament Calcium Sensitivity: Consequences of the Effective Concentration of Troponin I

Jalal K. Siddiqui <sup>1</sup> , Svetlana B. Tikunova<sup>1</sup> , Shane D. Walton<sup>1</sup> , Bin Liu<sup>1</sup> , Meredith Meyer <sup>1</sup> , Pieter P. de Tombe<sup>2</sup> , Nathan Neilson<sup>1</sup> , Peter M. Kekenes-Huskey <sup>3</sup> , Hussam E. Salhi <sup>1</sup> , Paul M. L. Janssen<sup>1</sup> , Brandon J. Biesiadecki <sup>1</sup> and Jonathan P. Davis <sup>1</sup> \*

*<sup>1</sup> Department of Physiology and Cell Biology and the Davis Heart and Lung Research Institute, The Ohio State University, Columbus, OH, USA, <sup>2</sup> Cell and Molecular Physiology, Loyola University Chicago, Maywood, IL, USA, <sup>3</sup> Department of Chemistry, University of Kentucky, Lexington, KY, USA*

#### Edited by:

*P. Bryant Chase, Florida State University, USA*

#### Reviewed by:

*Charles Redwood, University of Oxford, UK Douglas Root, University of North Texas, USA D. Brian Foster, Johns Hopkins School of Medicine, USA*

> \*Correspondence: *Jonathan P. Davis davis.812@osu.edu*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *05 October 2016* Accepted: *05 December 2016* Published: *21 December 2016*

#### Citation:

*Siddiqui JK, Tikunova SB, Walton SD, Liu B, Meyer M, de Tombe PP, Neilson N, Kekenes-Huskey PM, Salhi HE, Janssen PML, Biesiadecki BJ and Davis JP (2016) Myofilament Calcium Sensitivity: Consequences of the Effective Concentration of Troponin I. Front. Physiol. 7:632. doi: 10.3389/fphys.2016.00632* Control of calcium binding to and dissociation from cardiac troponin C (TnC) is essential to healthy cardiac muscle contraction/relaxation. There are numerous aberrant post-translational modifications and mutations within a plethora of contractile, and even non-contractile, proteins that appear to imbalance this delicate relationship. The direction and extent of the resulting change in calcium sensitivity is thought to drive the heart toward one type of disease or another. There are a number of molecular mechanisms that may be responsible for the altered calcium binding properties of TnC, potentially the most significant being the ability of the regulatory domain of TnC to bind the switch peptide region of TnI. Considering TnI is essentially tethered to TnC and cannot diffuse away in the absence of calcium, we suggest that the apparent calcium binding properties of TnC are highly dependent upon an "effective concentration" of TnI available to bind TnC. Based on our previous work, TnI peptide binding studies and the calcium binding properties of chimeric TnC-TnI fusion constructs, and building upon the concept of effective concentration, we have developed a mathematical model that can simulate the steady-state and kinetic calcium binding properties of a wide assortment of disease-related and post-translational protein modifications in the isolated troponin complex and reconstituted thin filament. We predict that several TnI and TnT modifications do not alter any of the intrinsic calcium or TnI binding constants of TnC, but rather alter the ability of TnC to "find" TnI in the presence of calcium. These studies demonstrate the apparent consequences of the effective TnI concentration in modulating the calcium binding properties of TnC.

Keywords: troponin C, troponin I, effective concentration, thin filament, mathematical model

### INTRODUCTION

The Ca2<sup>+</sup> sensitivity of cardiac muscle contraction is compromised by many genetic and acquired cardiomyopathies (Hamdani et al., 2008; Tardiff, 2011; Liu et al., 2012a,b). This is of major significance considering it has been suggested that any chronic change in the Ca2<sup>+</sup> sensitivity of cardiac muscle will eventually lead to a cardiomyopathy (Willott et al., 2010; Davis J. et al., 2016). On the other hand, we have recently demonstrated that chronic Ca2<sup>+</sup> sensitization via gene therapy does not lead to disease and can be utilized to protect and therapeutically aid the heart in a murine model of myocardial infarction (Shettigar et al., 2016). Consistent with these findings, not all disease associated mutations that alter the Ca2<sup>+</sup> sensitivity of cardiac muscle have complete penetrance (Tardiff, 2011; Ploski et al., 2016). Thus, either altering Ca2<sup>+</sup> sensitivity does not always lead to disease or certain alterations in Ca2<sup>+</sup> sensitivity are more or less easier to compensate. In any regard, with a deeper understanding of the molecular mechanism(s) that control the Ca2<sup>+</sup> sensitivity of cardiac muscle, even better design strategies can be employed to correct or compensate for aberrant Ca2<sup>+</sup> binding to improve cardiac function and performance (Davis J. P. et al., 2016).

Ultimately, the steady-state measurement of Ca2<sup>+</sup> sensitivity is dictated by an equilibrium established by the rates of Ca2<sup>+</sup> binding and release (Davis and Tikunova, 2008; Chung et al., 2016). Unfortunately, it is only in the most experimentally reduced systems that Ca2<sup>+</sup> binding to TnC can be directly measured (i.e., isolated troponin C (TnC) and the Tn complex; Tikunova and Davis, 2004; Davis et al., 2007; Tikunova et al., 2010). Interestingly, the vast majority of disease-associated mutations and even several of the phosphomimetics in TnI and TnT do not have much of an impact on the Ca2<sup>+</sup> binding properties of the isolated Tn complex, but do so on the thin filament (Nixon et al., 2012; Liu et al., 2012b, 2014). As the biochemical systems build and become more physiological, technical limitations necessitate having to follow other experimental outputs, which are merely transformations of the actual Ca2<sup>+</sup> binding event (i.e., fluorescence, actomyosin ATPase activity, motility or force; Davis et al., 2007; Tikunova et al., 2010; Sommese et al., 2013; Meyer and Chase, 2016). In these cases, it is difficult to discern intrinsic from extrinsic influences on the apparent Ca2<sup>+</sup> binding properties of TnC to even know what to "fix" or target (Davis J. P. et al., 2016).

Considering TnC has only a single regulatory Ca2+-binding site, it has been difficult to envision how normal or aberrant alterations in so many proteins lead to changes in the Ca2<sup>+</sup> sensitivity and kinetics of this single site. It is unclear whether the aberrant Ca2<sup>+</sup> sensitivity shift is caused by a direct change in the intrinsic Ca2<sup>+</sup> binding properties of TnC (as would be assumed in a simple two-state switch-like mechanism), and/or merely apparently shifting the Ca2<sup>+</sup> sensitivity by altering subsequent downstream events in how TnC interacts with, or is influenced by, its other regulatory subunits (such as TnI and TnT, and all their interacting proteins). Further complicating molecular insight, the disease-associated protein modifications can also impact the myocyte's innate ability to tune Ca2<sup>+</sup> sensitivity via phosphorylation, so that only under certain conditions might the apparent Ca2<sup>+</sup> sensitivity even appear altered (Biesiadecki et al., 2007; Messer and Marston, 2014). Depending on the system being studied, the apparent Ca2<sup>+</sup> sensitivity of both the biochemical and physiological systems can vary by over an order of magnitude (Biesiadecki et al., 2014). We propose a large proportion of the variability in apparent Ca2<sup>+</sup> sensitivities in these different systems is subtly controlled by Mg2<sup>+</sup> competition and strongly influenced by TnI availability to TnC.

With respect to TnI availability, there are fascinating phenomena that occur at the boundaries of how we think about the mechanical and chemical world. One such behavior, effective concentration, emerges when reactants are restricted to interact in very confined spaces, as occurs when two reactants are physically tethered together (Van Valen et al., 2009). TnC and TnI can be both artificially tethered in chimeras (Tiroli et al., 2005; Pineda-Sanabria et al., 2014) and more naturally in the Tn complex (Pineda-Sanabria et al., 2014; Davis J. P. et al., 2016). Due to being tethered (Van Valen et al., 2009), TnC has the potential to experience extremely high effective concentrations of TnI to extremely low (in times when TnI mobility becomes restricted). By comparing the Ca2<sup>+</sup> binding properties of TnC-TnI chimeras to that of their freed counterparts, we suggest the tethering of TnC to TnI can explain: (1) why the Tn complex has such high apparent Ca2<sup>+</sup> sensitivity and slow Ca dissociation (compared to the isolated protein) that is drastically reduced and accelerated, respectively, when the Tn complex is incorporated onto the thin filament; (2) why a large proportion of Tn modifications seem to have no effect on the apparent Ca2<sup>+</sup> binding properties of the isolated Tn complex, yet differences emerge when placed in the context of the thin filament; (3) that there are several different molecular mechanisms within and outside of the Tn complex that influence the intrinsic and/or apparent Ca2<sup>+</sup> binding properties of TnC; and (4) at least four states of TnC are required to simulate the apparent Ca2<sup>+</sup> binding properties of TnC in different experimental and diseased conditions.

### METHODS

### Biochemical Studies Proteins Utilized

The TnI128−<sup>180</sup> peptide was synthesized by The Ohio Peptide, LLC (Powel, OH). We generated two TnC-TnI chimeras consisting of the N-terminal domain of human cardiac TnC (residues 1–89) with the C-terminal domain of human cardiac TnI (residues 128–211) connected by a flexible and cleavable linker containing a site for the Tobacco Etch Virus protease, which contained the sequence GGAGGENLYFQG. For the F27W chimera, the endogenous Cys residues within TnC were converted into Ser and Phe 27 was converted to Trp, resulting in the following protein sequence: MDDIYKAAVEQLTEEQKN EFKAAFDIWVLGAEDGSISTKELGKVMRMLGQNPTPEELQ EMIDEVDEDGSGTVDFDEFLVMMVRSMKDDSGGAGGEN LYFQGLTQKIFDLRGKFKRPTLRRVRISADAMMQALLGAR AKESLDLRAHLKQVKKEDTEKENREVGDWRKNIDALSGM EGRKKKFES. For the T53C-IAANS chimera, the endogenous Cys residues within TnC were converted into Ser, Thr 53 was converted to Cys (resulting in the following protein sequence: MDDIYKAAVEQLTEEQKNEFKAAFDIFVLGAEDGSISTKELG KVMRMLGQNPCPEELQEMIDEVDEDGSGTVDFDEFLVMM VRSMKDDSGGAGGENLYFQGLTQKIFDLRGKFKRPTLRRV RISADAMMQALLGARAKESLDLRAHLKQVKKEDTEKENRE VGDWRKNIDALSGMEGRKKKFES) and labeled with IAANS as previously described (Davis et al., 2007).

**251**

### Chimera Expression and Purification

Pet17b vectors containing the chimeras were transformed into Rosetta 2 BL21 De3 bacteria and expressed after induction with 1 mM IPTG for 4 h. The bacteria were sonicated and the resulting solution was centrifuged at 19,000 RPM at 4◦C for 30 min and the supernatant was collected. Ammonium sulfate was added at 20% saturation to remove some of the contaminating proteins. The solution was centrifuged again at 19,000 RPM at 4◦C for 30 min and the supernatant was collected. Ammonium sulfate was then added to 60% saturation to precipitate the chimera. The solution was centrifuged at 19,000 RPM at 4◦C for 30 min with the supernatant removed. The pellet was resuspended in 30 mL Buffer A (20 mM Tris, 2 mM EDTA, 6 M Urea, 0.5 mM DTT, at pH 8.0) and dialyzed at least four times against 1 L of the same buffer. The solution was then loaded onto an SQ-15 column equilibrated with buffer A. After an initial washing with Buffer A, a gradient was applied with 0–25% of buffer B (buffer A with 1 M NaCl). Fractions were collected and then dialyzed against 4 L of 10 mM MOPS, 150 mM KCl, at pH 7.0 at least four times.

### Steady-State Fluorescence Measurements

All steady-state fluorescence measurements were performed using a Perkin-Elmer LS55 spectrofluorimeter at 15◦C. Trp fluorescence was excited at 295 nm and monitored at 320 nm as microliter amounts of CaCl<sup>2</sup> were added to 2 ml of titration buffer (200 mM MOPS; to prevent pH changes upon addition of Ca2+; 150 mM KCl, 2 mM EGTA, at pH 7.0) with constant stirring. The [Ca2+] free was calculated using the computer program EGCA02 developed by Robertson and Potter as previously described (Davis et al., 2007). The Ca2<sup>+</sup> sensitivities were reported as a dissociation constant Kd, representing a mean of at least three separate titrations ± S.E.M. The data were fit with a logistic sigmoid function (mathematically equivalent to the Hill equation). 0.5µM human cardiac TnCF27W was titrated with Ca2<sup>+</sup> in the absence or presence of up to 10µM TnI128−180. The F27W chimera was also titrated with Ca2<sup>+</sup> in the absence or presence of 3 mM Mg2+.

### Stopped-Flow Fluorescent Measurements

Ca2<sup>+</sup> dissociation rates were characterized using an Applied Photophysics model SX.20 stopped-flow instrument with a dead time of 1.4 ms at 15◦C. IAANS fluorescence was excited at 330 nm with emission monitored through a 420–470 nm bandpass interference filter (Oriel, Stratford, CT). Data traces (an average of at least five individual traces) were fit with a single exponential equation to calculate the kinetic rates. The working buffer used for the kinetic measurements was 10 mM MOPS, 150 mM KCl, at pH 7.0. Ten millimeters EGTA was utilized to remove saturating Ca2<sup>+</sup> from 1 µM of the human cardiac T53C-IAANS TnC (in the absence or presence of increasing concentrations of TnI128−180), uncleaved 0.5µM T53C-IAANS chimera, or cleaved T53C-IAANS chimera (in the presence or absence of increasing concentrations of TnI128−180). The chimera was cleaved overnight at 4◦C by the addition of one part TEV protease for every five parts of chimera in 10 mM MOPS, 150 mM KCl, at pH 7.0. Since we do not complex as much EGTA with Ca2<sup>+</sup> as will occur during the titration experiments, we do not need to use as much MOPS to buffer the pH. Control experiments confirmed that using buffer containing 200 mM MOPS instead of 10 mM MOPS did not affect the apparent rate of Ca2<sup>+</sup> dissociation from the Tn complex following T53C-IAANS TnC fluorescence (data not shown).

### Simulations and Estimations

Using Scilab, an open source numerical computational package, we solved the differential equations to obtain the time-dependent concentrations of each species given a set of rate constants and initial concentrations (described below). For the two state simulations, we plotted the TnC-Ca species as the fluorescent state and for all other simulations we plotted the TnC-Ca-TnI species as the fluorescent state. For the steady-state and transient occupancy studies, the concentration of these species were subsequently converted to a percentage of the total TnC concentration and simply overlaid onto the actual experimental data. For the Ca2<sup>+</sup> dissociation and association rate studies, we normalized the change in the concentration of the species overtime and simply overlaid the simulations onto the actual experimental data.

### Estimation of the Effective Concentration of TnI in the Troponin Complex

While the effective concentration of TnI for TnC cannot be directly measured, estimations are possible. Based on the structure of the Tn complex, it appears that the tether connecting the switch peptide of TnI to the Tn complex is along residues 134–147. If we assume a 3.4 Angstrom distance (maximal extension of the residues) and assume it to be the radius of a sphere, we obtain a total volume of 1.25 × 10−<sup>22</sup> m3. By calculating the molarity of 1 TnI in this space using Avogradro's number we are able to estimate an effective concentration of ∼13 mM. If we include more residues, 134–155, we estimate a lower limit of ∼3 mM. This is very much in line with previous estimates of the effective concentration of a chimeric TnC-TnI protein (Pineda-Sanabria et al., 2014).

### Steady-State Ca2<sup>+</sup> Binding

To simulate steady-state Ca2<sup>+</sup> binding when Mg2<sup>+</sup> was explicitly considered, we ran initial simulations with 1µM TnC and 3 mM MgCl<sup>2</sup> to equilibrium. This was then followed by running a loop where different levels of calcium from 0.0362 to 1000µM were inputted and run to equilibrium. For the loop, each simulation was run to a time span of at least 0.5 s for each inputted Ca2<sup>+</sup> concentration to reach equilibrium. A resulting plot of pCa vs. activated TnC ([TnC-Ca-TnI]) was developed.

### Ca2<sup>+</sup> Dissociation Kinetics

As with steady-state calcium binding, we initially began the simulation with 1µM TnC and 3 mM MgCl<sup>2</sup> and initially ran a simulation (at least 0.5 s time span) to determine the equilibrium concentrations of species resulting from Mg2<sup>+</sup> binding. This was followed by inputting [Ca2+] of 200µM. After running a simulation for at least 0.5 s, we inputted [EGTA] of 10 mM and ran a simulation. A plot of time vs. [TnC-Ca-TnI] was developed and outputted to a file.

### Transient Occupancy Studies/Calcium Input Studies

We initially began the transient occupancy simulations with 1µM TnC, 3 mM MgCl2, and 600µM EGTA. After running a simulation to equilibration, we inputted [Ca2+] levels of 12.5, 25, 50, and 1000 into the simulation. The simulations were normalized to the highest [Ca2+] level and a time vs. [TnC-Ca-TnI] plot was developed. We also performed studies without EGTA simulating the response of thin filaments to different [Ca2+] levels: 2.5–20 µM.

### RESULTS

One of the striking and consistent findings across the literature is that the apparent Ca2<sup>+</sup> binding properties of cardiac TnC vary substantially when studied in different systems (ranging from isolated TnC to muscle; Davis et al., 2007; Davis J. P. et al., 2016). Based on data from our work over the years (performed under as similar conditions as possible in simplified biochemical systems), **Figure 1A** demonstrates that the apparent Ca2<sup>+</sup> sensitivity of TnC falls into three general Ca2<sup>+</sup> sensitivity ranges. For instance, the apparent Ca2<sup>+</sup> sensitivity is the lowest (highest Kd) when only the isolated TnC is investigated, intermediate when the Tn complex is reconstituted onto the thin filament and highest in

the isolated Tn complex or when the thin filament is bound by rigor myosin heads either in reconstituted thin filaments or Tn exchanged myofibrils. Similarly, the apparent rate of Ca2<sup>+</sup> dissociation from TnC in these different systems somewhat scale proportionately to the change in apparent affinity (**Figure 1B**), giving the impression that the K<sup>d</sup> changes are modulated primarily by dissociation rate changes. Thus, the same single EF-hand in the context of different systems can have drastically different apparent Ca2<sup>+</sup> binding properties.

One of the obvious differences between the simplest systems is the presence or absence of TnI. It is well-documented that the binding of the C-terminal domain of TnI to the regulatory domain of TnC increases the apparent Ca2<sup>+</sup> sensitivity and slows the rate of Ca2<sup>+</sup> dissociation from TnC (**Figure 2**; Davis and Tikunova, 2008). As can be observed in **Figure 2A**, increasing the concentration of TnI128−<sup>180</sup> added to TnCF27W increases the apparent Ca2<sup>+</sup> sensitivity up to a limit. This limit approaches the apparent Ca2<sup>+</sup> sensitivity of a chimeric protein in which the first 89 N-terminal residues of TnCF27W were physically tethered by a short peptide linker to the C-terminal domain of human cardiac TnI (residues 128–211). Likewise, **Figure 2B** demonstrates that the apparent rate of Ca2<sup>+</sup> dissociation from TnC (T53C-IAANS) slows with increasing concentration of TnI128−180down to a limit. This limit too approaches the apparent Ca2<sup>+</sup> dissociation rate from a chimeric protein in which the first 89 N-terminal residues of TnC (T53C-IAANS) were physically tethered by a short peptide linker to the C-terminal domain of human cardiac TnI (residues 128−211; **Figure 2C**). Strikingly, the Ca2<sup>+</sup> binding properties of the uncut chimeras and the two fluorescent TnCs in the presence of saturating TnI128−<sup>180</sup> are similar to that of the troponin complex (compare to **Figure 1**). Thus, it takes over an order of magnitude more isolated TnI128−<sup>180</sup> to sensitize isolated TnC to Ca2<sup>+</sup> compared to the chimeras and troponin complex in which TnC and TnI are physically tethered together (high effective concentration) at a stoichiometric ratio of one to one.

Using the chimera, we can demonstrate the principle of effective concentration in yet another way. Within the peptide linker of the T53C-IAANS chimera we engineered a tobacco etch virus protease (TEV) site that can be specifically cleaved by TEV (**Figure 2D**). As can be observed in **Figure 2D**, the TEV efficiently cleaves both chimeras resulting in two bands (a TnI fragment and a TnC fragment). After TEV cleavage of the chimera the rate of Ca2<sup>+</sup> dissociation is no longer observable (**Figure 2E**, black trace). Either the resulting rate is too fast to observe or the freed T53C-IAANS TnC N-domain's IAANS fluorescence is no longer sensitive to changes in Ca2+. In any regard, although we do not know whether the peptide ratios of the free N-domain of TnC to the free C-domain of TnI are at exactly one to one in the cleaved solution, a rate becomes observable again once excess TnI128−<sup>180</sup> is added back to the mixture of the TEV cleaved chimera (**Figure 2E**, blue trace). These experiments highlight and support that there appears to be a significantly higher effective concentration of TnI within the intact chimeras and potentially the troponin complex that can drastically influence the behavior and apparent Ca2<sup>+</sup> binding properties of TnC.

Similar to the chimeras where TnI was artificially tethered to TnC, the proper formation of the troponin complex also physically tethers TnI-TnC (**Figure 3**). Thus, in the troponin complex, the C-terminal domain of TnI is restricted within a small volume of space potentially orbiting (or whipping) around the N-terminal, regulatory domain of TnC (**Figure 3A**). If we assume this volume to be defined by a sphere with a radius equal to the length of TnI that extends from the IT arm up through the switch peptide (also assuming this stretch of amino acids to be linearly and maximally extended), the switch peptide of TnI is restricted within a maximum volume of ∼1.7 × 10<sup>6</sup> Å 3 . If we then place a single switch peptide into this volume we can calculate what the "effective concentration" of this peptide would be for a TnC that shares this volume space, ∼3000 µM. This calculated value is similar to that estimated for another TnC-TnI chimeric protein by NMR (Hwang et al., 2014; Pineda-Sanabria et al., 2014). If we also assume that there are regions of this volume that TnC does not share, then it is possible

to potentially trap, or at least temporarily restrict, TnI away from TnC, drastically plummeting the effective concentration of TnI that TnC "observes." Such an occurrence is not difficult to imagine when the Tn complex is docked onto the thin filament, since the C-terminal domain of TnI can bind both TnC and actin (Tripet et al., 1997; **Figure 3B**). It may be that several proteins, TnT, Tm and myosin can influence the effective concentration of TnI that TnC observes by influencing TnI's ability to bind actin rather than by directly altering TnC's intrinsic Ca2<sup>+</sup> binding properties.

The question now arises as to how many states are required to simulate, or model, the apparent Ca2<sup>+</sup> binding properties of TnC and what molecular mechanisms can explain the transition between the states? Fascinatingly, a simple two-state system seems to capture quite well both the steady-state and kinetic properties of any one single system such as isolated TnC, the Tn complex, reconstituted thin filament as well as rigor myosin bound to the thin filament (**Figures 4A–G**). However, these simulations assume that the intrinsic Ca2<sup>+</sup> association rate and dissociation rate from TnC must change in order to explain observed differences between the systems or modifications performed within a particular system (**Figure 5A**). A logical physical implication of this assumption is that all modifications or system changes that alter Ca2<sup>+</sup> binding somehow directly influence the structure of TnC's EF-hand, Ca2<sup>+</sup> coordination, or how TnC directly interacts with TnI and/or TnT. In contrast, we hypothesize that a change in the apparent Ca2<sup>+</sup> binding properties of TnC can occur upon a particular perturbation without any structural change in the resulting Ca2<sup>+</sup> bound structure.

This is not a trivial matter considering much effort is being put into searching for the atomic molecular mechanism(s) behind disease mutations, which are computationally forced to focus on very specific areas of interest, such as the precise coordination geometry of the Ca2<sup>+</sup> ion itself (Lindert et al., 2012; Williams et al., 2016). Thus, it is critical to understand what structure or protein interactions are behind the apparent changes in Ca2<sup>+</sup> binding and exchange, which may not be reflected in the Ca2<sup>+</sup> bound structure itself. A prime example of this phenomenon occurs when Mg2<sup>+</sup> competitively competes for Ca2<sup>+</sup> binding to an EF-hand protein (Kucharski et al., 2016). Although a new structure must occur to explain Mg2<sup>+</sup> binding, the Ca2<sup>+</sup> bound structure will be the same regardless whether Mg2<sup>+</sup> is present or not (**Figure 5B**). Similar to our previous data demonstrating the N-domain of cardiac TnC has a physiologically relevant and competitive Mg2<sup>+</sup> affinity (Tikunova and Davis, 2004; Liang et al., 2008), the addition of Mg2<sup>+</sup> to the TnCF27W chimera desensitizes the apparent Ca2<sup>+</sup> sensitivity of the chimera ∼2.5 fold, leading to an apparent Mg2<sup>+</sup> affinity of 2.1 ± 0.7 mM (**Figure 5C**). Since Mg2<sup>+</sup> must be present in order to maintain the structural integrity of the Tn complex, we suggest Mg2<sup>+</sup> competition should be taken into consideration when simulating TnC. Although subtle, only the apparent Ca2<sup>+</sup> association rate utilized to approximate each system's data is dependent on whether the effects of Mg2<sup>+</sup> competition are considered, not the Ca2<sup>+</sup> dissociation rate (**Figure 5A**). In essence, Ca2<sup>+</sup> cannot bind until Mg2<sup>+</sup> dissociates and once Ca2<sup>+</sup> binds, Ca2<sup>+</sup> will dissociate at its intrinsic rate. The data in **Figures 4B–F** can be equally well-simulated by both a simple two state model (ignoring competitive Mg2<sup>+</sup> binding) or a three state model (where TnC binds competitively Mg2<sup>+</sup> or Ca2+; **Figure 5B**). However, the "intrinsic" Ca2<sup>+</sup> association rate for each individual system using a simple two state model is significantly slower than what the rate must be if Mg2<sup>+</sup> competition is also considered (**Figure 5A**). Thus, the effects of Mg2<sup>+</sup> can be neglected, or lost, within a model by modulating what would appear to be the intrinsic Ca2<sup>+</sup> association rate of the system.

Much more pronounced and generally accepted, as can be observed in **Figures 1**, **2**, the binding of TnI to isolated TnC has a monumental impact on the apparent Ca2<sup>+</sup> association and dissociation rates from TnC (Kobayashi and Solaro, 2005; Davis and Tikunova, 2008). Considering in this case two different species can bind TnC (Ca2<sup>+</sup> and TnI—ignoring Mg2+), it is reasonable to assume that it is necessary to at least have a three state model to predict TnC behavior in muscle. In fact, a three state model has been utilized quite successfully to simulate the apparent Ca2<sup>+</sup> binding properties of cardiac myofibrils (Solzin et al., 2007). **Figure 6** demonstrates how a classical three state

All measurements were made following the change in fluorescence of TnC T53C-IAANS in increasingly complex biochemical systems.

model would predict the experimental outcomes of **Figures 1**, **2**. In a three-state model, TnC first binds Ca2<sup>+</sup> with kinetics that can be described simply for the isolated TnC (either considering Mg2<sup>+</sup> competition or not; **Figure 6A**). Although we do not know how fast TnI can actually bind Ca2<sup>+</sup> bound TnC, FRET studies suggest that TnI dissociates from TnC at least as fast as ∼110/s (keep in mind this rate too may be an apparent rate rather than an intrinsic rate; Ouyang et al., 2010). We estimated the TnI association rate based on a TnI affinity to TnC of 1.2 µM, which is surprisingly close to the apparent TnI128−<sup>180</sup> affinity extracted from **Figure 2B** of ∼3 µM. **Figures 6B,C** demonstrate a three state model incorporating TnI binding to TnC can predict the apparent Ca2<sup>+</sup> sensitivities and Ca2<sup>+</sup> dissociation rates of the Tn complex and thin filament

by solely altering the effective concentration of TnI without having to alter any intrinsic rates of Ca2<sup>+</sup> or TnI binding to/dissociation from TnC. However, unlike what was observed in **Figures 2A,B**, this three state model suggests that rather than there being a limit to the Ca2<sup>+</sup> sensitivity and Ca2<sup>+</sup> dissociation rate as the TnI concentration is increased, the pCa50 of this model continuously increases (approaching infinity), whereas the Ca2<sup>+</sup> dissociation rate approaches an asymptote of zero. Thus, a three state model that does not allow for Ca2<sup>+</sup> to dissociate from TnC until TnI dissociates predicts that the apparent Ca2<sup>+</sup> sensitivity would approach infinity and the Ca2<sup>+</sup> dissociation rate would approach zero as the TnI concentration is continually increased. Clearly this is not the case based on our peptide and chimera studies demonstrated in **Figure 2**. Therefore, we suggest that in addition to TnI dissociation, the TnC- Ca2+-TnI complex can also dissociate by releasing Ca2<sup>+</sup> as well (**Figure 6D**). Once TnC- Ca2+-TnI is allowed to also dissociate via loss of Ca2<sup>+</sup> as well as TnI, as the concentration of TnI is increased, the apparent Ca2<sup>+</sup> sensitivity and rate of Ca2<sup>+</sup> dissociation approach that of the apparent Ca2<sup>+</sup> binding properties of the Tn complex (which we earlier suggested senses a very high effective concentration of TnI; **Figures 6E,F**).

Thus, we suggest at minimum a four state model of TnC Ca2<sup>+</sup> binding and dissociation is required to accurately predict the Ca2<sup>+</sup> binding properties of TnC (**Figure 6D**). Similar to the three state model, we first assume the initiating event is Ca2<sup>+</sup> binding to the isolated regulatory domain of TnC that possesses very rapid Ca2<sup>+</sup> association and dissociation rates as we have previously demonstrated to approximate the steady-state and kinetic behavior of isolated TnC. Next, TnI is allowed to bind TnC- Ca2+, which is governed by the product of its intrinsic on rate to TnC and the effective concentration of TnI that TnC "sees" or "senses." Once the TnC- Ca2+-TnI species forms, it has two options to decay. The first pathway is via TnI dissociation, which is common to the three state model and is given a rate of 110/s consistent with FRET studies (Ouyang et al., 2010). The second pathway is via Ca2<sup>+</sup> dissociation, which we have set at the rate at which Ca2<sup>+</sup> has been observed to dissociate from the intact Tn complex (∼40/s) with an intrinsic Ca2<sup>+</sup> association rate similar to that predicted by a two-state model for the isolated Tn complex (see **Figure 5A**). Now we have a new state, TnC-TnI, which we assume dissociates at least as fast as TnI can dissociate from TnC-Ca2+-TnI complex at ∼110/s. We also assume that the regulatory TnC-TnI state cannot form in the absence of Ca2+, consistent with experimental data (hence this reaction is not reversible).

Using our four state model and assuming a very high effective concentration of TnI (of at least 850 µM—that could actually be much higher based on the effective concentration calculations and predictions), **Figures 7A,B** demonstrate we can simulate the apparent Ca2<sup>+</sup> binding properties of the isolated Tn complex. By only lowering the effective concentration of TnI to ∼8 µM (without any intrinsic rate alterations) our four state model also predicts the apparent Ca2<sup>+</sup> binding properties of the reconstituted thin filament's: (1) steady-state Ca2<sup>+</sup> binding affinity (**Figure 7C**); (2) Ca2<sup>+</sup> dissociation rate (**Figure 7D**); (3) Ca2<sup>+</sup> association rates (**Figure 7E**); and (4) response to artificial Ca2<sup>+</sup> transients (**Figure 7F**). Currently the 8µM value for effective TnI concentration required to simulate the entire set of thin filament data is empirical. As the field learns more about the intrinsic, as well as effective concentration, of TnI that actin "sees/senses" (**Figure 3B**), one could begin to explicitly simulate this important interaction with TnI too.

Now that we have demonstrated that changing only the effective concentration of TnI connects TnC's Ca2<sup>+</sup> binding properties between different systems, we were curious if this mechanism might also be used to explain how certain disease associated mutations alter the apparent Ca2<sup>+</sup> sensitivity of TnC. Excitingly, **Figures 8A,B** demonstrates that nearly half of all the disease associated modifications in TnI and TnT that our laboratory has studied at both the steady-state and kinetic level can be approximated by only altering the effective concentration of TnI that TnC "sees/senses." Thus, we suggest the ability of cardiac muscle to tune the effective concentration of TnI may be a powerful mechanism to alter the apparent Ca2<sup>+</sup> binding properties of the thin filament without altering any intrinsic Ca2<sup>+</sup> or TnI binding properties of TnC. However, the remaining Tn modifications we have studied require altering the intrinsic

extended model with increasing effective TnI concentration.

properties of TnC as well, in order to simulate the Ca2<sup>+</sup> binding properties of the Tn complex and thin filament (manuscript in preparation). Collecting both steady-state and kinetic data from isolated Tn and the thin filament is essential to helping to elucidate whether a change in Ca2<sup>+</sup> binding properties is due to intrinsic and/or extrinsic factors.

A major modulator of thin filament Ca2<sup>+</sup> binding as well as the speed of muscle mechanics is phosphorylation of cardiac TnI at serine residues 23 and 24 (Biesiadecki et al., 2014; Janssen et al., 2016). Both my, Dr. Biesiadecki's and other laboratories have independently measured the apparent Ca2<sup>+</sup> binding properties of the reconstituted thin filament under similar experimental conditions using the phosphomimetic TnI in which Ser 23/24 have been mutated to Asp (Albury et al., 2012; Liu et al., 2014; Nixon et al., 2014). Not surprisingly, both of our laboratories obtain nearly identical apparent rates of Ca2<sup>+</sup> dissociation from our wild type and phosphomimetic reconstituted thin filaments, suggesting our protein systems behave identically (**Figure 9A**). However, an apparent discrepancy arose in comparing our steady-state Ca2<sup>+</sup> binding behavior of the thin filaments. **Figure 9B** demonstrates that my laboratory's wild type and phosphomimetic thin filaments were both substantially more

published (Davis et al., 2007). Panel (E) in black shows the rates of Ca2<sup>+</sup> association to the reconstituted thin filaments as previously published (Liu et al., 2012b). Panel (F) (noisy colored curves) shows the response of the thin filaments to artificial Ca2<sup>+</sup> transients as previously published (Liu et al., 2012b). For Panel (A–E), the smooth red curves show the simulated output of the four state model. For Panel (F), all the smooth curves represent the simulated output of the four state model.

desensitized to Ca2<sup>+</sup> compared to the results obtained from the Biesiadecki laboratory. Unlike the kinetic measurements where saturating Ca2<sup>+</sup> and EGTA are used to measure the Ca2<sup>+</sup> dissociation rates and subtle differences in each concentration will have no bearing on the results, the steadystate measurements are highly dependent upon the precise concentrations of both EGTA and Ca2+. Excitingly, using our four state model and assuming only the concentration of EGTA was different between the laboratories (by only 10% out of 2 mM, which can easily be accounted for via differences in pipettes, equipment or H2O content of the EGTA powder) we are able to reconcile our experimental findings. Once we were able to correct for the Ca2<sup>+</sup> buffering differences in our experimental systems, we were able to simulate the behavior of TnI S23/24D in both our data sets by making only a single intrinsic rate change of accelerating the TnI dissociation from TnC of 110–460/s (maintaining the effective concentration of TnI at 8µM similar to the wild type condition).

Interestingly, **Figures 9C,D** demonstrate that unlike the reconstituted thin filament, the apparent steady-state Ca2<sup>+</sup> binding properties and Ca2<sup>+</sup> dissociation rate of the isolated Tn complex containing TnI S23/24D were nearly identical to

that of the wild type Tn complex. This result can be simply modeled by again assuming the isolated troponin complex has a very high effective TnI concentration (>1000µM in this case). Thus, a high effective TnI concentration can overcome the Ca2<sup>+</sup> desensitizing effects of TnI S23/24D. On the other hand, TnI S150D, a phosphomimetic that models AMP kinase phosphorylation of TnI, sensitizes both the Tn complex and thin filaments to Ca2+, as well as slows both systems' Ca2<sup>+</sup> dissociation rates (**Figures 9C,D**; Nixon et al., 2012, 2014). In order to simulate these results we have assumed two alterations: (1) the rate of Ca2<sup>+</sup> dissociation from the TnC- Ca2+-TnI complex is slowed by 40% (setting a new limiting value as the effective TnI concentration is elevated) and (2) the effective concentration of TnI for the reconstituted thin filament was raised from 8 to 25µM. Thus, our model predicts that unlike TnI S23/24D, TnI S150D increases the effective concentration of TnI that TnC "sees" when incorporated onto the thin filament. One possibility for this observation and a rationale for having to increase the effective concentration of TnI for TnC on the thin filament could be a weakened binding affinity of TnI S150D for actin. Consistent with this prediction, Salhi et al. has demonstrated TnI S150D has a weaker affinity for actin, compared to TnI S23/24D, which had a similar actin affinity as compared to wild type TnI (Salhi et al., 2016). This observation suggests mechanisms that alter the ability of TnI to interact with actin can have a profound effect of modulating the apparent Ca2<sup>+</sup> binding properties of TnC potentially through the effective concentration of TnI that TnC "sees/senses."

### DISCUSSION

There has been great interest in developing mathematical models that can recapitulate the electrical, Ca2<sup>+</sup> and mechanical responses of the heart. Although there are very good models for elements of each of these processes, no model has been able to unify all the essential steps of contraction/relaxation into a cohesive predictive or diagnostic tool. We argue that one of the problems arises in the need to oversimplify the systems, of which Ca2<sup>+</sup> binding to TnC is a prime example. Based on a wealth of biochemical and physiological data it is clear that the apparent Ca2<sup>+</sup> binding properties of TnC can influence both the extent and speed of cardiac muscle contraction, relaxation, and power (McDonald and Herron, 2002; Biesiadecki et al., 2014; Davis J. P. et al., 2016). By more fully understanding the underlying mechanisms that control the intrinsic and apparent Ca2<sup>+</sup> binding properties of cardiac muscle it may be possible to engineer the response of the thin filament to Ca2<sup>+</sup> to eventually treat various cardiovascular diseases, as we have been attempting to achieve (Davis J. P. et al., 2016; Shettigar et al., 2016).

Although each individual state of TnC (isolated TnC, Tn complex, thin filament, thin filament bound by strong crossbridges, myofibrils, etc.) can be simulated with a simple two state system (see **Figure 4**), by no means does this suggest Ca2<sup>+</sup> binding to TnC in muscle is simple. We argue that TnC transitions through multiple states even during a single heartbeat, which can then influence the mechanics of that heartbeat when the thin filament is altered naturally by phosphorylation or even disease. Thus, we have set out to try and understand what might be the major influences on Ca2<sup>+</sup> binding to TnC as it transitions from state to state. Clearly the ability of TnC to bind TnI is one major factor that influences TnC's Ca2<sup>+</sup> binding properties (Kobayashi and Solaro, 2005; Biesiadecki et al., 2014). However, even this step in the process may be more complicated than just understanding the affinity or intrinsic rate of TnI association and dissociation from TnC. We suggest there are mechanisms beyond the overall "affinity" of TnC for TnI that can feed back on TnC without altering any intrinsic parameters of TnC, such as the effective concentration of TnI that TnC "sees/senses."

The concept of effective concentration is not a new idea (Van Valen et al., 2009), even as it applies to TnC (Hwang et al., 2014), although we are the first to try and apply the concept

to mathematically describing state transitions of TnC. Based on our initial studies we speculate that nearly half of all the disease associated modifications of Tn that we have studied can be explained solely by altering the effective concentration of TnI that TnC "sees." Other Tn modifications such as TnI S23/24D may have no effect on TnI's effective concentration, whereas others such as TnI S150D require modulating both intrinsic TnC properties as well as the effective concentration of TnI (Salhi et al., 2016). As we work through trying to model all our biochemical data, we are finding there are many ways to alter the apparent Ca2<sup>+</sup> binding properties of TnC. Not all of the mechanisms are as straightforward as simply accelerating or slowing the intrinsic Ca2<sup>+</sup> binding properties of TnC. In most cases, the potential mechanism behind a Tn modification is not evident until the protein modification is studied in multiple biochemical systems (at minimum the troponin complex and the reconstituted thin filament) both in the steady-state and kinetically. It is clear that Ca2<sup>+</sup> triggers extensive dynamic changes in Tn that can be altered by disease associated mutations (Kowlessur and Tobacman, 2012; Liu et al., 2012b). As the model is developed to move from Ca2<sup>+</sup>

binding to force, we will need to also consider cooperative unit communications (Gordon et al., 2000). As one moves beyond this TnC centric view, one will also need to consider tropomyosin positions as well as TnT behavior when determining the ability of Ca2<sup>+</sup> to modulate myosin binding to actin (Mijailovich et al., 2012).

We suggest the tethering of TnC to TnI and our model can help to explain: **(1)** why the Tn complex has such high apparent Ca2<sup>+</sup> sensitivity and slow Ca2<sup>+</sup> dissociation (compared to the isolated protein) that is drastically reduced and accelerated respectively, when the Tn complex is incorporated onto the thin filament. **Figure 2** clearly shows that the apparent Ca2<sup>+</sup> sensitivity and dissociation rate of isolated TnC is modulated by the concentration of TnI. Furthermore, at saturating TnI concentrations, both the apparent Ca2<sup>+</sup> sensitivity and dissociation rate are very similar to the intact Tn complex as well as the chimeras. **Figure 3** illustrates that the proximal confinement of TnI to TnC results in a high effective TnI concentration in the Tn complex. **Figures 6**, **7** use our model to show that at a high effective TnI concentration, we can accurately Siddiqui et al. Modeling TnC-TnI Interactions

simulate Tn behavior and by solely lowering the effective TnI concentration we can simulate thin filament behavior. **(2)** Why a large proportion of Tn modifications seem to have no effect on the apparent Ca2<sup>+</sup> binding properties of the isolated Tn complex, yet differences emerge when placed in the context of the thin filament. **Figure 8** shows that the apparent Ca2<sup>+</sup> binding properties of several Tn modifications can be modeled solely by changing the effective concentration of TnI on the thin filament. Furthermore, at a saturating TnI concentration, our model predicts for each of these Tn modifications that the apparent Ca2<sup>+</sup> binding properties of the Tn complex are identical to that of the wild type Tn complex. **Figure 9** shows an example of a modification that requires changing an intrinsic rate parameter in order to model thin filament behavior. However, raising the effective concentration of TnI to saturating levels restores the Tn like behavior. In this case, an accelerated TnI dissociation rate from TnC will be overpowered by the high effective TnI concentration. **(3)** There are several different molecular mechanisms within and outside of the Tn complex that influence the intrinsic and/or apparent Ca2<sup>+</sup> binding properties of TnC. Our study focused on how altering the effective concentration of TnI can regulate the apparent Ca2<sup>+</sup> binding properties of TnC. As pointed out in the manuscript and our model, we do not exclude that there are mutations/modifications in the Tn complex (or a plethora of additional proteins) that can alter the apparent Ca2<sup>+</sup> binding properties of TnC though alternate mechanisms, such as altering the intrinsic rates of Ca2+, Mg2+, and/or TnI binding/dissociation from TnC. In fact, this is the point of a companion paper to be published in this thematic issue by the Biesiadecki lab (Salhi et al., 2016). Thus, all the intrinsic binding and dissociation rates (Ca2+, Mg2+, and TnI) set by the compilation of proteins in the thin and thick filaments, as well as the effective concentration of TnI, work together to set the overall Ca2<sup>+</sup> sensitivity of TnC. **(4)** At least four

### REFERENCES


states of TnC are required to simulate the apparent Ca2<sup>+</sup> binding properties of TnC in different experimental and diseased conditions. **Figures 6**–**9** demonstrate at minimum a four-state model is needed to successfully simulate not only the steadystate, but also kinetic behavior of a wide assortment of Tn modifications.

In conclusion, we have generated a strikingly powerful mathematical model that can simulate several different states of TnC in the presence of different Tn modifications. We have utilized the model to make predictions regarding protein interactions that fall outside of direct TnC interactions, such as TnI-actin binding (Salhi et al., 2016). Currently all that is needed to connect and simulate our data is a four state model to recapitulate TnC's apparent Ca2<sup>+</sup> binding properties as well as evoking the concept of effective concentration. Considering nearly all of the myofilament protein-protein interactions that influence muscle contraction occur within confined and restricted spaces, effective concentration concepts may need to be invoked to simulate several other key reaction steps as well, especially myosin with actin (Fuchs and Smith, 2001).

### AUTHOR CONTRIBUTIONS

Performed experiments: JS, ST, SW, MM, NN, HS, BL. Designed Study: JS, ST, Pd, BB, JD. Analyzed Data: JS, ST, PJ, PK, JD.

### FUNDING

Supported by NIH Grants HL091986 (JD), AG051913 (JD), HL117034 (ST), HL113084 (PJ), HL1114940 (BB), HL62426 (Pd).

### ACKNOWLEDGMENTS

The authors thank Dr. Jianchao Zhang for advice on protein purification and expression of the chimeras.

binding and exchange with cardiac troponin C. Biophys. J. 92, 3195–3206. doi: 10.1529/biophysj.106.095406


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Siddiqui, Tikunova, Walton, Liu, Meyer, de Tombe, Neilson, Kekenes-Huskey, Salhi, Janssen, Biesiadecki and Davis. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Modulating Beta-Cardiac Myosin Function at the Molecular and Tissue Levels

Wanjian Tang<sup>1</sup> , Cheavar A. Blair <sup>2</sup> , Shane D. Walton<sup>1</sup> , András Málnási-Csizmadia<sup>3</sup> , Kenneth S. Campbell 2, 4 and Christopher M. Yengo<sup>1</sup> \*

<sup>1</sup> Department of Cellular and Molecular Physiology, Pennsylvania State University College of Medicine, Hershey, PA, USA, <sup>2</sup> Department of Physiology, University of Kentucky, Lexington, KY, USA, <sup>3</sup> Department of Biochemistry, Eötvös Loránd University, Budapest, Hungary, <sup>4</sup> Division of Cardiovascular Medicine, University of Kentucky, Lexington, KY, USA

Inherited cardiomyopathies are a common form of heart disease that are caused by mutations in sarcomeric proteins with beta cardiac myosin (MYH7) being one of the most frequently affected genes. Since the discovery of the first cardiomyopathy associated mutation in beta-cardiac myosin, a major goal has been to correlate the in vitro myosin motor properties with the contractile performance of cardiac muscle. There has been substantial progress in developing assays to measure the force and velocity properties of purified cardiac muscle myosin but it is still challenging to correlate results from molecular and tissue-level experiments. Mutations that cause hypertrophic cardiomyopathy are more common than mutations that lead to dilated cardiomyopathy and are also often associated with increased isometric force and hyper-contractility. Therefore, the development of drugs designed to decrease isometric force by reducing the duty ratio (the proportion of time myosin spends bound to actin during its ATPase cycle) has been proposed for the treatment of hypertrophic cardiomyopathy. Para-Nitroblebbistatin is a small molecule drug proposed to decrease the duty ratio of class II myosins. We examined the impact of this drug on human beta cardiac myosin using purified myosin motor assays and studies of permeabilized muscle fiber mechanics. We find that with purified human beta-cardiac myosin para-Nitroblebbistatin slows actin-activated ATPase and in vitro motility without altering the ADP release rate constant. In permeabilized human myocardium, para-Nitroblebbistatin reduces isometric force, power, and calcium sensitivity while not changing shortening velocity or the rate of force development (ktr). Therefore, designing a drug that reduces the myosin duty ratio by inhibiting strong attachment to actin while not changing detachment can cause a reduction in force without changing shortening velocity or relaxation.

Keywords: myosin, actin, muscle contraction, molecular motors, cardiomyopathy

### INTRODUCTION

Inherited cardiomyopathies caused by mutations in sarcomere protein-coding genes are a significant cause of cardiovascular diseases in people of all ages (Morimoto, 2007; Watkins et al., 2011). Hypertrophic cardiomyopathy (HCM) is the most common form of inherited cardiomyopathy, and the primary cause of sudden cardiac death in young adults (Maron, 2004; Efthimiadis et al., 2014; Maron et al., 2014). The latest revised HCM prevalence is about 1 in

#### Edited by:

P. Bryant Chase, Florida State University, USA

#### Reviewed by:

Douglas Swank, Rensselaer Polytechnic Institute, USA Gabriella Piazzesi, University of Florence, Italy Donald A. Winkelmann, Rutgers University, USA

> \*Correspondence: Christopher M. Yengo cmy11@psu.edu

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 01 September 2016 Accepted: 15 December 2016 Published: 09 January 2017

#### Citation:

Tang W, Blair CA, Walton SD, Málnási-Csizmadia A, Campbell KS and Yengo CM (2017) Modulating Beta-Cardiac Myosin Function at the Molecular and Tissue Levels. Front. Physiol. 7:659. doi: 10.3389/fphys.2016.00659

**264**

200 of the general population including mutation carriers at risk for developing a phenotype (Semsarian et al., 2015). HCM manifests as left ventricle hypertrophy featuring cardiomyocyte disarray and fibrosis, a thickening of the left ventricular wall and decreased coronary artery blood flow during diastole (Maron, 2002; Maron et al., 2006; Watkins et al., 2011; Vakrou and Abraham, 2014). Dilated cardiomyopathy (DCM) has an estimated prevalence of 1 in 2500 individuals, and the cases with a genetics etiology account for ∼50% (Taylor et al., 2006; Towbin, 2014). DCM is characterized by the thinning of one or both ventricular walls, an enlarged left ventricular chamber, and insufficient systolic contraction (Luk et al., 2009; Hershberger et al., 2010; McNally et al., 2013). Restrictive, arrhythmogenic right ventricular, left ventricular non-compaction, and other types of cardiomyopathies have been classified as well, but are less prevalent in the general population (Elliott et al., 2008; Watkins et al., 2011; Towbin, 2014).

Cardiomyopathy mutations are commonly found in the myosin heavy chain 7 gene (MYH7) encoding human β-cardiac myosin heavy chain (M2β) (Xu et al., 2010), which is the motor that drives contraction of the ventricular myocardium. Single point mutations in M2β are capable of disrupting motor function. Identification of disease mutations has raised expectations for disease prediction and novel therapeutic strategies. More than 300 pathogenic mutations in M2β are distributed throughout the whole myosin molecule, and there is no consensus about the detailed mechanisms behind the impact of these mutations (Moore et al., 2012). The mechanisms responsible for altering motor functions are varied, and are likely dependent on the locations of the mutation (Moore et al., 2012; Colegrave and Peckham, 2014; Homburger et al., 2016).

### Myosin Structure-Function

Many years of research has established that myosin is the motor protein that converts chemical energy into mechanical work and drives the shortening of muscle and other forms of actomyosinbased force generation. The main components of the muscle sarcomere are thick and thin filaments. The thick filaments are composed of myosin molecules that form cross-bridges that interact with thin filaments composed of actin. Myosin consists of two heavy chains, each with two associated light chains, an N-terminal motor domain, and C-terminal coiledcoil tail that allows dimerization and incorporation into the thick filaments (**Figure 1**). There are several proteins associated with the thick filaments (e.g., myosin binding protein C, Titin) involved in contractile regulation. Thin filaments contain actin and regulatory proteins (tropomyosin and troponin complex), which are important for mediating the Ca2+-induced activation of the thin filaments.

### Myosin ATPase Cycle

Myosin is an ATP-dependent molecular motor that cyclically interacts with actin filaments with weak and strong actin-binding states. **Figure 2** describes the key steps in the catalytic cycle and the proposed structural changes that occur in each step. ATP binding to myosin causes a conformational change in the actin binding region resulting in weak actin affinity, and formation of the pre-power stroke state of the lever arm (recovery stroke). ATP is hydrolyzed while myosin is dissociated from actin in a weak actin-binding state. Myosin binding to actin with the hydrolyzed products accelerates the release of phosphate and then ADP, which results in force generation (power stroke). Recent work in the Yengo lab on myosin V has shown that the lever arm swing occurs in two steps, a fast step that gates phosphate release and a slow step coupled to ADP release (Trivedi et al., 2015). Studies with skeletal muscle myosin also demonstrate a rapid movement of the lever arm prior to phosphate release (Muretta et al., 2015). Alternatively, evidence from x-ray crystallography suggests that the movement of phosphate from the active site into the phosphate release tunnel is required for the movement of the lever arm, while release of phosphate from the tunnel into solution occurs after the lever arm swing (Houdusse and Sweeney, 2016). Muscle fiber studies have provided evidence that phosphate release occurs after force generation (Dantzig et al., 1992) or is orthogonal to the power stroke (Caremani et al., 2013, 2015), while correlating the biochemical, structural, and muscle fiber experiments remains a challenge.

### Force-Velocity Properties of Muscle

The chemomechanical ATPase cycle contains two fundamental parts, weak actin-binding states (M.ATP and M.ADP.Pi) and strong actin-binding states (A.M.ADP.Pi, AM.ADP and A.M). Force generation occurs in the strong binding states, during which the myosin power stroke generates a displacement (step size = 5–10 nm in muscle myosins) of the actin filament (Spudich, 2014). The duty ratio is the fraction of ATPase cycle time myosin is in the strong binding states, which determines the number of strongly bound myosin heads interacting with the thin filaments at any time. Each myosin head is an independent motor and produces its own intrinsic force (f). The overall isometric force (F) is the intrinsic force (f) multiplied by the number of force-generating myosin heads, which can be expressed as the following equation (Spudich, 2014):

$$\mathbf{F} = \mathbf{f} \times \mathbf{N}\_{\text{total}} \times \text{ duty ratio}$$

where Ntotal is the number of heads that are potentially able to bind to the thin filaments. The maximum shortening velocity is thought to depend on the myosin step size (unitary displacement—duni) and the period of time myosin is attached to actin (ton). Thus, the following equation is often used to describe the maximum shortening velocity (Warshaw, 2004):

$$\mathbf{v}\_{\text{max}} = \mathbf{d}\_{\text{uni}} / \mathbf{t}\_{\text{on}}$$

Since these parameters can be measured with isolated myosin, it is possible to correlate the individual properties of myosin with the contraction parameters in muscle. However, the ton is altered by the presence of load, as established in muscle fiber studies (Piazzesi et al., 2002; Reconditi et al., 2004) and further explored in single molecule mechanics studies (Sung et al., 2015; Greenberg et al., 2016). In addition, the factors that limit Vmax are controversial with some studies demonstrating detachment rate (1/ton) correlates well with Vmax (Siemankowski et al.,

FIGURE 1 | Diagram of the muscle sarcomere and the myosin molecule. A simplified diagram of the sarcomere (top panel) demonstrates the location of the myosin thick filaments (A-band) and the actin thin filaments (I-band). A diagram of the myosin molecule (lower panel) demonstrates its overall structure. The myosin heavy chains, light chains (essential light chain, ELC, regulatory light chain, RLC), subfragment 1 (S1) (utilized in the current study), subfragment 2 (S2), heavy meromyosin (HMM), and light meromyosin (LMM) are labeled.

1985; Nyitrai et al., 2006; Yengo et al., 2012) and other studies demonstrating attachment rate limits Vmax (Haldeman et al., 2014; Brizendine et al., 2015).

The in vitro motility assay is commonly used to examine the force generating properties of purified myosin (Kron et al., 1991). In this assay myosin is adhered to a microscope cover slip and the sliding velocity of fluorescently labeled actin is monitored in the presence of ATP. The sliding velocity generated by an ensemble of myosin motors is thought to correlate to the shortening velocity measured in muscle (Howard, 2001). In order to examine duni, f, and ton, the single molecule laser trap motility assay is often used (Simmons and Finer, 1994; Sivaramakrishnan et al., 2009). In this assay a single actin filament is strung between two beads that are each trapped with laser tweezers and when a single myosin molecule is brought close to the actin filament individual displacements (duni) are measured. The single molecule laser trap studies are typically performed at low ATP concentrations which can create uncertainty in determining ton and correlating it with muscle fiber studies (Tyska and Warshaw, 2002). The stiffness of the laser trap can allow determination of the force generated by a single myosin head (f), but due to the large compliance of the laser trap the force can be underestimated (Spudich et al., 2011).

### The Impact of Mutations in Human β-Cardiac Myosin

Humans predominantly express the slow β-cardiac myosin isoform in ventricles but most studies examining the impact of mutations have been performed in mice which express α-cardiac myosin, a faster cardiac myosin isoform (Deacon et al., 2012). This has complicated the interpretation of the experimental data because mutations in α-cardiac myosin have different effects than mutations in β-cardiac myosin (Lowey et al., 2008; Palmer et al., 2008; Witjas-Paalberends et al., 2014; Nag et al., 2015). Other studies have examined human muscle fibers purified from skeletal muscle biopsies or from ventricular samples obtained from patients who had cardiac surgeries (Köhler et al., 2002; Seebohm et al., 2009; Brenner et al., 2012; Kraft et al., 2013; Witjas-Paalberends et al., 2014). Measurements on human recombinant β-cardiac myosin are just beginning to be reported and are promising for examining large numbers of different mutations to establish structure-function relationships. Recent studies have demonstrated that some mutations have a relatively small impact on the key parameters mentioned above (f, V, ton, duni) (Alpert et al., 2005; Moore et al., 2012; Nag et al., 2015). Thus, it is still unclear how the point mutations lead to impaired cardiac muscle function and hypertrophy.

### Current Treatments

Despite the lack of a clear understanding of the molecular mechanisms of cardiomyopathies, symptom-based inotropic drugs are still the conventional clinical pharmacological therapy (Maron, 2002; Spirito and Autore, 2006; Vakrou and Abraham, 2014; Tardiff et al., 2015). β-adrenergic antagonists (e.g., Metoprolol and Nebivolol), Ca2<sup>+</sup> channel blockers (e.g., Verapamil and Diltiazem), Na<sup>+</sup> channel blockers (e.g., Disopyramide), antiarrhythmic agents (e.g., Amiodarone), and angiotensin II receptor antagonists (e.g., Losartan) are currently used in the clinic to alleviate the symptoms of HCM (Vakrou and Abraham, 2014; Tardiff et al., 2015). For DCM patients, angiotensin-converting enzyme inhibitors, β-adrenergic blockers, aldosterone inhibitors, and angiotensin receptor blockers have been used clinically (Elliott, 2000; Taylor et al., 2006; Luk et al., 2009). An implantable cardioverter-defibrillator has been shown as the only effective way to prevent sudden cardiac death, and heart transplantations are usually needed for cardiomyopathy patients with end-state heart failure (Elliott and McKenna, 2004; Efthimiadis et al., 2014).

Additionally, inotropic drugs which directly target sarcomeric proteins are under investigation (Malik et al., 2011; Tardiff et al., 2015; Green et al., 2016). The thin filament has been suggested as an ideal target site to treat cardiomyopathies via altering Ca2<sup>+</sup> sensitivity. Thick filaments are also being pursued as drug targets. By changing the kinetics of individual steps in the myosin ATPase cycle, small molecule drugs are proposed to change the duty ratio and thus the number of forge-generating myosin heads capable of interacting with the thin filaments. Utilization of drugs that directly target contractile proteins in cardiac muscle is still in its early stages and will require detailed pre-clinical studies that can examine their specific mechanisms of action and off-target effects.

Blebbistatin (Bleb) is a well-established inhibitor of class II myosins and understanding its mechanism of action has been an important step in developing novel inhibitors of myosin based force generation. Bleb was first identified as a muscle and non-muscle myosin II specific inhibitor with a mechanism of binding to the ATPase intermediate with ADP and phosphate and slowing down phosphate release by trapping myosin in a weak actin-binding conformation (Straight et al., 2003; Kovács et al., 2004; Ramamurthy et al., 2004; Farman et al., 2008). Additionally, Bleb has been shown to inhibit striated and smooth muscle myosins but with no effect on unconventional class I, V, and X myosins (Limouze et al., 2004; Dou et al., 2007; Eddinger et al., 2007). Studies of Bleb in rodent cardiac muscle found that Bleb decreased the twitch force of isolated cardiac trabeculae and the shortening velocity of cardiac myocytes in a dose-dependent manner (Dou et al., 2007; Farman et al., 2008). Since Bleb binds near the actin binding region and traps the myosin heads in a weak actin affinity state, it is also proposed to reduce the myosin binding-induced activation of the thin filaments (Ramamurthy et al., 2004; Allingham et al., 2005; Dou et al., 2007). Bleb has also been found to stabilize the helical ordering of myosin heads, a conformation in which myosin heads interact with each other but not with actin (Zhao et al., 2008; Xu et al., 2009). This state has been referred to as the super relaxed state (SRX) and Bleb has been shown to stabilize the SRX by unknown mechanisms (Wilson et al., 2014). The use of Bleb was hindered by its blue light sensitivity, phototoxicity, and poor solubility (Sakamoto et al., 2005; Mikulich et al., 2012), but this has been addressed by the discoveries of highly soluble, non-phototoxic Bleb derivatives [para-Nitroblebbistatin (pN-Bleb), and amino-blebbistatin; Képiró et al., 2014; Várkuti et al., 2016].

In the current study we examined the impact of pN-Bleb on human β-cardiac myosin in both expressed/purified myosin in vitro motor assays and in human myocardium fiber mechanics studies. We hypothesized that the less phototoxic pN-Bleb would be able to inhibit the in vitro motor properties of human cardiac myosin at the molecular and tissue levels. We proposed that investigating the impact of this drug on human myocardium would lead to insight into strategies for designing cardiac myosin specific drugs. Our results provide evidence of the mechanism of action of pN-Bleb on human β-cardiac myosin and suggest important considerations in designing novel drugs that impact the force and shortening velocity properties of cardiac muscle.

## METHODS

### Reagents

ATP was prepared from powder (De La Cruz and Ostap, 2009). 2 ′ -deoxy-ADP labeled with N-Methylanthraniloy at the 3′ -ribose position (mantADP) was purchased from Jenna Biosciences. pN-Bleb was obtained from András Málnási-Csizmadia and dissolved in DMSO. All motility experiments with M2β were performed in motility buffer with pCa value of 4.5 (7 mM EGTA, 20 mM Imidazole, 51 mM KCl, 7 mM CaCl2, 5.22 mM MgCl2, pH 7.0) and other experiments were performed in MOPS 20 buffer (10 mM MOPS, 20 mM KCl, 1 mM MgCl2, 1 mM EGTA, 1 mM DTT, pH 7.0). The final concentrations of pN-Bleb are described for each experiment and the final concentration of DMSO was 1% for the motility, ATPase, and ADP release experiments. Further details about the solutions for the muscle mechanics studies are given below. All concentrations listed are final unless stated otherwise.

### Construction of Expression Plasmids

The human cardiac myosin cDNA (AAA5187.1) was purchased from Thermo Scientific. PCR amplification was used to subclone the M2β subfragment 1 (M2β-S1) construct (amino acids 1–843) into the pshuttle vector (a gift from Dr. Don Winkelmann). M2β-S1 was engineered to contain an N-terminal FLAG tag sequence and C-terminal Avi tag sequence.

### Recombinant Adenovirus Based Expression and Purification of M2β-S1 in C2C<sup>12</sup> Cells

The production of high titer adenovirus was performed by a method developed in the Winkelmann laboratory (Srikakulam and Winkelmann, 2004; Winkelmann et al., 2015). Homologous recombination was used to produce pAdEasy recombinant adenovirus DNA (pAd.M2β-S1) by transforming the pshuttle.M2β-S1 into E. coli BJ5183 cells. The pAd.M2β -S1 was transformed into XL-10 Gold cells for amplification and the pAd.M2β-S1 DNA was digested with Pac1 and transfected into Ad293 cells to allow for virus packaging and amplification. The Ad293 cells were grown in DMEM media supplemented with 10% fetal bovine serum. The large scale virus preparation was performed by infecting 60 plates (145 mm diameter). The virus was harvested with freeze thaw cycles followed by CsCl density sedimentation. The final virus titers were typically 1010–10<sup>11</sup> plaque forming units (PFU) per ml.

C2C<sup>12</sup> cells grown to 90% confluence in DMEM supplemented with 10% fetal bovine serum (typically 20–30, 145 mm diameter plates) were differentiated by changing the media to DMEM supplemented with 10% horse serum and 1% fetal bovine serum. The C2C<sup>12</sup> cells were infected with recombinant adenovirus (5 × 10<sup>8</sup> PFU/ml) diluted into differentiation media. The media was changed after 2 days and cells were harvested on day 7. The cells were lysed with a 50 ml dounce in lysis buffer (50 mM Tris, pH 7.0, 200 mM KCl, 2 mM ATP, 1 mM ATP, 0.5% Tween20, 0.01 mg/ml aprotenin, 0.01 mg/ml leupetin, 1 mM PMSF) and spun 2 × 15 min at 25 K in a Ti50 rotor at 4◦C. The supernatant was added to a 1 ml anti-FLAG M2 resin column, washed with wash buffer (10 mM Tris, pH 7.5, 200 mM KCl, 1 mM EGTA, 1 mM EDTA, 2 mM MgCl2, 2 mM ATP, 1 mM DTT, 0.01 mg/ml aprotenin, 0.01 mg/ml leupetin, 1 mM PMSF), and eluted with wash buffer containing FLAG peptide (0.167 mg/ml). The eluted M2β-S1 was subsequently ammonium sulfate precipitated and dialyzed into MOPS 20 buffer overnight at 4◦C. M2β-S1 was biotinylated for in vitro motility studies by incubating M2β-S1 with BirA (10µg/ml) for 1 h at 25–30◦C, and subsequently ammonium sulfate precipitated and dialyzed into MOPS 20 buffer overnight at 4◦C (Lin et al., 2005).

M2β-S1 purity was assessed by coomassie stained SDSpolyacrylamide gels and protein concentration was determined by Bradford assay using BSA as a standard. Similar results were obtained by measuring the absorbance and using the predicted extinction coefficient (ε<sup>280</sup> = 1.38 × 10<sup>5</sup> M−<sup>1</sup> ·cm−<sup>1</sup> ). Skeletal muscle heavy meromyosin (Sk HMM) was prepared from rabbit psoas muscle as described (Swenson et al., 2014). Actin was purified from rabbit skeletal muscle using an acetone powder method (Pardee and Spudich, 1982). The actin concentration was determined by absorbance at 290 nm (ε<sup>290</sup> = 2.66 × 10<sup>4</sup> M−<sup>1</sup> ·cm−<sup>1</sup> ). A molar equivalent of phalloidin was added to stabilize F-actin.

### In vitro Motility Assay

We performed in vitro motility assays (Kron et al., 1991) using the recombinantly expressed/purified M2β-S1 and purified Sk HMM. The M2β-S1 experiments were performed in conditions (buffer and temperature) that were similar to the muscle mechanic studies described below. The actin filament sliding assay was performed as previously described (Trivedi et al., 2013; Swenson et al., 2014) except for the method of adhering the myosin to the surface in the case of M2β-S1. Microscope cover slips were coated with 1% nitrocellulose in amyl acetate (Ladd Research). The surface was coated with streptavidin (0.1 mg/ml) and blocked with BSA (1 mg/ml) before the addition of biotinylated M2β-S1 (loading concentration was 0.48µM). Unlabeled sheared actin (2µM) followed by an ATP (2 mM) wash was used to prevent interactions with dead heads. Actin labeled with ALEXA (GFP filter; excitation/emission: 500/535 nm) was visualized by fluorescence microscopy. An activation buffer with 1% DMSO or pN-Bleb (0.1, 1, 5, 10, 20, 50µM) was added to the flow cell to initiate motility. Activation buffer contained the following: 0.35% methylcellulose, 2.5 mM phosphoenolpyruvate, 20 units·ml−<sup>1</sup> pyruvate kinase, 0.1 mg·ml−<sup>1</sup> glucose oxidase, 5 mg·ml−<sup>1</sup> glucose, 0.018 mg·ml−<sup>1</sup> catalase, and 4.8 mM ATP. The slide was promptly viewed using a NIKON TE2000 microscope equipped with a 60×/1.4 NA phase objective and a Perfect Focus System. Images were acquired at intervals (appropriate for each condition) for periods of time (3–15 min) using a shutter controlled Coolsnap HQ2 cooled CCD digital camera (Photometrics) binned 2 × 2. Temperature was maintained at 22–24◦C and monitored using a thermocouple meter (Stable Systems International). Image stacks were transferred to ImageJ for analysis via MTrackJ (Meijering et al., 2012). The average velocity was determined by tracking actin filaments manually for each condition using ImageJ.

### In vitro Steady-State ATPase Activity

Steady-state ATP hydrolysis by M2β-S1 or Sk HMM (100 nM) in the presence of actin (40µM) was examined by using the nicotinamide adenine dinucleotide (NADH)-linked assay (De La Cruz et al., 2000; Dosé et al., 2007, 2008; Quintero et al., 2010) in MOPS 20 Buffer with a final MgATP concentration of 1 mM. The assay was performed in an Applied Photophysics stoppedflow (Surrey, UK) in which the NADH absorbance at 340 nm was monitored continuously for 200 s. The data at each actin concentration represents an average of 2 protein preparations.

### Determination of IC50

We plotted the relative ATPase or sliding velocity data as a function of pN-Bleb concentration which allowed us to determine the IC50 by fitting the data to the following equation: Relative activity = 1/{(1+[pN-Bleb]/IC50)}.

### Transient Kinetic Measurements of ADP-Release

We examined the ADP release rate constant of M2β-S1 in the presence of actin. A complex of M2β-S1, actin, and mantADP (0.375, 1, and 10µM, respectively) was mixed with saturating ATP (1 mM) and the mant fluorescence (excitation 290 nm/emission 395 nm long pass filter) was monitored in the stopped-flow. The fluorescence transients were fit with custom software provided with the instrument or Graphpad Prism.

### Human Tissue

Myocardial samples were obtained at the University of Kentucky from patients who had end-stage heart failure using the protocol described by Blair et al. (2016). Briefly, through-wall sections of the distal anterior region of the left ventricle were obtained from explanted hearts and dissected transmurally (sub-epicardial, mid-myocardial, sub-endocardial). The experiments described in this manuscript were performed using a total of 24 subendocardial samples from 4 patients. All procedures were approved by the University of Kentucky Institutional Review Board and patients gave informed consent.

### In situ Preparations and Experimental Set-Up

Permeabilized multicellular preparations were obtained using the mechanical digest protocol described by Haynes et al. (2014). Multicellular preparations with a mean length 1047 ± 232µm were attached between a force transducer (resonant frequency, 600 Hz; model 403, Aurora Scientific, Aurora, Ontario, Canada) and a motor (step time 0.6 ms; model 312B, Aurora Scientific) and stretched to a sarcomere length of 2.24µm in a solution with a pCa (= −log10[Ca2+]) of 9.0. The cross-sectional area was 5.07 ± 2.47 × 10−<sup>8</sup> m<sup>2</sup> (estimated assuming a circular profile). Experiments were conducted at 22◦C using SLControl software (Campbell and Moss, 2003).

### Para-Nitroblebbistatin Preparation and Incubation of Samples

Separate sets of solutions with pCa values ranging from 9.0 to 4.5 and pN-Bleb concentrations of 0, 1, 10, or 50µM were generated. The final percentage of DMSO in every experimental solution was 1.33%. Half of the preparations were used to assess tension-pCa relationships. Each of these preparations was initially tested in control solutions (0 pN-Bleb) with pCa values ranging from 9.0 to 4.5. The preparation was then immersed for 5 min in a pCa 9.0 solution containing 1, 10, or 50µM pN-Bleb. Additional measurements were then performed using solutions containing the chosen pN-Bleb concentration and pCa values ranging from 9.0 to 4.5. The other half of the preparations were used to assess force-velocity relationships. These samples were only tested in pCa 4.5 solutions with 0 pN-Bleb (control) and then a chosen experimental pN-Bleb concentration. These experimental designs ensured that each preparation could act as its own control and minimized the progressive decline in contractile force (experimental run-down) that occurs when permeabilized preparations are subjected to repeated activations.

### In situ Mechanical Measurements

Multicellular preparations were activated in solutions with pCa values ranging from 9.0 to 4.5. Once tension reached steady-state, the preparations were rapidly shortened by 20%, held for 20 ms, and then re-stretched to their original length. All experiments were performed at a sarcomere length of 2.25µm. The rate of tension recovery (ktr) was then calculated by fitting the portion of the force record immediately after the re-stretch with a single exponential function of the form F(t) = A + B (1-exp(-ktrt)), where F(t) is the force at time t, and A and B are constants.

Ca2<sup>+</sup> sensitivity (pCa50) values were calculated by fitting the steady-state force data to a modified Hill equation of the form F = Fpas + FCa ([Ca2+] n /([Ca2+] <sup>n</sup> + [Ca2<sup>+</sup> <sup>50</sup>] n )). In this equation, Fpas is the force measured in pCa 9.0 solution, FCa is Ca2<sup>+</sup> activated force, n is the Hill coefficient, and [Ca2<sup>+</sup> <sup>50</sup>] n is the free Ca2<sup>+</sup> concentration required to develop half the maximum Ca2+-dependent force.

To measure shortening velocity and power, the multicellular preparations were allowed to shorten for 80 ms against preset loads that ranged from 0 to 100% of the maximum tension measured in pCa 4.5 solution. The shortening velocity in each trial was calculated from the slope of a straight line fitted to a plot of fiber length against time during the final 50 ms of the force clamp. The mean force was also determined during this time. The resulting data were then fitted using a hyperbolic equation of the form (F+a) (V+b) = (F0+a) b, where F is the force developed at a shortening velocity of V, F<sup>0</sup> is the isometric force and a and b are constants with dimensions of force and velocity respectively. Vmax was determined by extrapolating the forcevelocity curve to zero load. Power values (P) were calculated as the product of force and velocity. Power-force curves were calculated by fitting the individual data points with a curve of the form P = F b (((F0+a)/(F+a))−1). Maximum power was defined as the maximum value of this curve.

### Statistics for In situ Muscle Mechanics on Human Samples

Data were analyzed using linear mixed models. These are statistical hypothesis tests that are similar to ANOVA procedures but which allow for the fact that multiple samples were analyzed from each heart (Haynes et al., 2014). This increases the statistical power of the hypothesis test in this type of experimental design. Compound symmetry was assumed for the covariance structure and post-hoc analyses were performed using Tukey–Kramer corrections. P < 0.05 were considered significant. Data are reported as mean ± SEM.

### RESULTS

We have examined the impact of pN-Bleb on the motor properties of recombinantly expressed human β-cardiac myosin subfragment 1 (M2β-S1) and the force and velocity properties of human myocardium. When possible, we performed the motor function assays and muscle mechanics studies under very similar conditions (temperature and buffer) to allow comparison of the impact of the drug on muscle fiber mechanics and isolated myosin motor performance. We also examined the impact of pN-Bleb on the heavy meromyosin fragment of chicken skeletal muscle myosin (Sk HMM) with in vitro motility and actin-activated ATPase assays, which allowed a comparison of the specificity of pN-Bleb for these two myosin isoforms.

### In vitro Motility of M2β-S1 and Sk HMM

The in vitro motility assay was utilized to examine the impact of pN-Bleb on the motile properties of purified M2β-S1 and Sk HMM. The sliding velocity produced by M2β-S1 in the in vitro motility assay (motility buffer at 22◦C) was determined in the presence of varying concentrations of pN-Bleb or 1% DMSO by examining 2 separate protein preparations at a loading concentration of 0.48µM (**Figure 3**). Our previous densitydependent in vitro motility studies with M2β-S1 demonstrated that this motor density (0.48µM loading) was saturating (Swenson et al., 2016). The presence of 1% DMSO had a minor impact on in vitro motility (the average velocity was 1398 ± 19 and 1261 ± 22 nm/s in the absence and presence of 1% DMSO, respectively). The data from 2 preps was pooled together (60 filaments) to determine the average sliding velocity at each pN-Bleb concentration. There was an 85% inhibition of the sliding velocity in the presence of 50µM pN-Bleb (**Figures 3A,B**) and the IC50 (13.3 ± 0.14µM) was estimated from the concentration dependence (**Figure 3C**). The in vitro motility of Sk HMM was performed in MOPS 20 buffer at 24◦C, since it was difficult to obtain results in the higher ionic strength motility buffer that was utilized with M2β-S1. We found that the IC50 for Sk HMM (1.6 ± 0.3µM) was indicative of a higher specificity of the drug for Sk HMM compared to M2β-S1.

### Actin-Activated ATPase Activity of M2β-S1 and Sk HMM

We examined the impact of pN-Bleb on the actin-activated ATPase of purified M2β-S1 and Sk HMM. We examined the ATPase activity in MOPS 20 buffer, since the higher ionic strength of the motility buffer was not feasible for examining actin-activated ATPase. The ATPase assay with M2β-S1 was performed at 22◦C in the presence of 40µM actin and

demonstrated that pN-Bleb inhibits actin-activated ATPase in a dose-dependent manner (**Figure 4A**). The determined IC50 was similar to that determined in the in vitro motility assay (12.3 ± 1.8µM). We also performed ATPase assay experiments with Sk HMM in similar conditions (MOPS 20 buffer and 25◦C) and found the IC50 (0.4 ± 0.1µM) indicated a higher specificity for Sk HMM compared to M2β-S1.

for M2β-S1 (13.3 ± 0.1µM) and Sk HMM (1.6 ± 0.2µM).

## ADP Release Rate Constant of M2β-S1

The ADP release rate constant is thought to be an important determinant of the time period that myosin is attached to actin during the ATPase cycle (Siemankowski and White, 1984; Siemankowski et al., 1985). Therefore, we utilized mant labeled ADP to monitor the release of ADP from acto-M2β-S1 in MOPS 20 buffer at 22◦C, which was identical to the conditions of the actin-activated ATPase assay. The fluorescence transients were fit to a single exponential function which allowed us to determine the ADP release rate constant (**Figure 4B**). The results demonstrate that the ADP release rate constant measured with mantADP is very similar in the presence and absence of 50µM pN-Bleb (208.9 ± 5.1 and 228.6 ± 6.5 sec−<sup>1</sup> , respectively).

### Muscle Mechanics of Human Myocardium

We performed a series of mechanical tests to determine how pN-Bleb impacted the Ca2+-dependence of contractile force and tension-recovery kinetics, and the shortening velocity and power output measured at maximum Ca2+activation. **Figure 5** shows representative experimental records (top 2 rows) for the force-velocity/force-power measurements and curves calculated from these records (bottom 2 rows; Experimental details are provided in Section Methods and in the Figure legend). These measurements yielded data quantifying isometric force (**Figure 6A**), maximum power (**Figure 6B**) and maximum shortening velocity (**Figure 6C**). Summary data for ktr, the rate of tension recovery, are shown in **Figure 6D**. As described in Section Methods, these values were obtained by measuring how quickly force recovered toward steady-state after a large shortening/re-stretch perturbation (raw traces not shown). The statistical hypothesis tests showed that 50 µM pN-Bleb reduced both isometric force (**Figure 6A**) and maximum power (**Figure 6B**) by ∼50% but did not produce significant changes in either maximum shortening velocity (**Figure 6C**) or ktr (**Figure 6D**). Isometric force normalized to cross-sectional area is lower for chemically permeabilized human myocardial samples than it is for some other types of muscle preparations, which we have demonstrated previously (Haynes et al., 2014).

Tension-pCa curves were generated in additional experiments and are plotted in **Figure 7A**. As shown during the forcevelocity measurements, pN-Bleb reduced isometric force in a dose-dependent manner. pN-Bleb also reduced the Ca2+ sensitivity (pCa<sup>50</sup> values; **Figure 7B**) and the Hill coefficient (**Figure 7C**). However, these effects were only significant at the 50µM concentration which suggests that effects of pN-bleb on Ca2<sup>+</sup> activation are relatively modest.

## DISCUSSION

Directly targeting human cardiac myosin with small molecule allosteric regulators has been proposed as a therapeutic strategy for several forms of heart failure (Malik et al., 2011; Tardiff et al., 2015; Green et al., 2016). We demonstrate the impact of a myosin inhibitor, which is a modified version of the wellstudied Bleb, on human cardiac muscle myosin at the molecular and tissue levels. Although, this drug is not specific for cardiac myosin since it has been demonstrated to inhibit several other muscle and non-muscle myosins, it still serves as a model to examine mechanistically how inhibition of cardiac myosin can be accomplished and how this will impact muscle performance. We find that pN-Bleb reduces the in vitro motility of cardiac myosin likely because it decreases the myosin duty ratio by inhibiting the transition into the strongly bound states. In muscle mechanic studies we find that pN-Bleb has no impact on shortening velocity or the rate of force development while the decrease in steady-state force, Ca2<sup>+</sup> sensitivity, and power are also indicative of a reduced duty ratio.

### The Motor Properties of M2β in the Presence of pN-Bleb

The in vitro motility results in the current study demonstrate a pN-Bleb concentration-dependent reduction in sliding velocity. We also observed a similar concentration-dependent reduction in the actin-activated ATPase, which suggests the drug inhibits a similar step in the ATPase cycle in both assays. In light of

representative preparation measured under control conditions (left column, 0 pN-Bleb) and in the presence of 50µM pN-Bleb (right column). The symbols showing force, power, and shortening velocity are drawn in the same color as the raw traces from which they were calculated. As described in the Section Methods and by Haynes et al. (2014), these data were obtained by first activating the preparation in pCa 4.5 solution and then allowing it to shorten against loads ranging from 0 to 100% isometric force in successive trials. The shortening velocity was calculated for each trial from the slope of the muscle length against time trace. Similarly, the mean force during shortening was calculated from the force record. Each single trial thus yielded a single data point on the force-velocity plot. Power values were calculated as the product of force and velocity.

the previous studies on Bleb, it is likely that pN-Bleb traps cardiac myosin in a weakly bound state that reduces the rate of actin-activated phosphate-release. We find that pN-Bleb does not alter the ADP-release rate constant which typically correlates with the time myosin is attached to actin and is an important determinant of maximum velocity and duty ratio. Thus, pN-Bleb acts by stabilizing the weakly bound conformation and in the in vitro motility assay the reduction in myosin heads that productively attach to actin and produce force creates a situation similar to what is observed at low motor densities. When the number of force generating heads in the motility assay decreases, it is proposed that the period of time between myosin attachments becomes rate-limiting (Uyeda et al., 1990; Harris and Warshaw, 1993). Interestingly, we did not observe a reduction in shortening velocity in the human cardiac muscle mechanics studies. These results may reflect the structural organization in muscle which has many myosin heads in close proximity to the actin thin filament and thus is not as sensitive to this type of inhibition. The original theory of muscle contraction outlined by Huxley (1957) proposed that unloaded shortening velocity was independent of the number of cycling myosin crossbridges. Furthermore, it has been demonstrated that only 1–4 myosins per thick filament are required to sustain maximum velocity (Fusi et al., 2016). The reduction in steady-state force and power is consistent with the proposed mechanism of reducing the myosin duty ratio and therefore the number of myosin heads available to generate force. Thus, the mechanism of inhibition utilized by this drug is advantageous because at moderate doses it does not change the kinetics of contraction while it does effectively reduce steady-state force and power. In patients that are hyper-contractile this mechanism may work well since it could normalize the force velocity relationship and power without altering the systolic contraction time and relaxation kinetics. It is also important to consider the impact of this type of inhibition on shortening velocity in the presence of load since this is the more physiologically relevant situation in the heart. From the force-velocity experiments (**Figure 5**) it is clear that pN-Bleb alters shortening velocity in the presence of load and thus the systolic contraction time could be impacted. We did not directly measure the impact of pN-Bleb on the phosphate release rate constant and this measurement as well a detailed examination of all of the transient kinetics steps in the M2β-S1 ATPase cycle will be important to examine in future studies.

Interestingly, the specificity of pN-Bleb for skeletal muscle myosin was nearly 10-fold higher than human cardiac myosin based on the measured IC50 in the motility assay and 30 fold higher based on ATPase assays. The ATPase IC50 value we determined for skeletal muscle myosin was similar to that reported in the literature (Képiró et al., 2014). These results demonstrate that the binding affinity of pN-Bleb for cardiac myosin may be weaker than skeletal myosin or that the structural state that favors pN-Bleb binding is more significantly populated in skeletal. Limouze et al. (2004) determined the specificity of Bleb for many different muscle and non-muscle myosins and found considerable variability. Further high resolution structural studies are necessary to evaluate the structural details of the Bleb binding pocket which may be a useful site for rationally designing myosin inhibitors.

### Impact of pN-Bleb on Human Myocardium

The muscle mechanics data clearly demonstrate that pN-Bleb reduced isometric force in a dose-dependent manner. However, the effects of pN-Bleb on Ca2<sup>+</sup> sensitivity need to be interpreted with care. Although, the pCa<sup>50</sup> values and Hill coefficients were

significantly reduced by a pN–Bleb concentration of 50µM, the lower concentrations of pN-Bleb did not produce marked effects. It's also unclear whether isometric force was completely saturated in a pCa 4.5 solution in the presence of 50µM pN-Bleb (note that the tension-pCa curve did not reach a flat plateau in **Figure 7A**). These data could indicate that a high concentration of pN-Bleb desensitizes the thin filaments by reducing the myosin duty ratio. However, the low Hill coefficients and decreased pCa<sup>50</sup> values measured in the presence of 50µM pN-Bleb could also be explained by a progressive reduction in force development during the experiments. Dou et al. (2007) showed that the effects of Bleb on force development in mouse papillary muscle and trabeculae were time-sensitive and that force took almost 30 min to stabilize after application of the drug. Similar time-dependent effects in human myocardium might produce tension-pCa data similar to those shown in **Figure 7A**.

## Considerations of Thin Filament Regulation

Our results are consistent with the hypothesis that there is a correlation between the myosin duty ratio and Ca2<sup>+</sup> sensitivity and that this relationship can be tuned with small molecule drugs that alter the myosin duty ratio. It is well established that strong-binding cross-bridges have a positive feedback regulation on thin filament Ca2<sup>+</sup> sensitivity, which stabilizes the Ca2<sup>+</sup> bound state of troponin C (Kobayashi et al., 2008). Therefore, changes in the number of strong binding cross-bridges on thin filaments, determined by duty ratio, could impact the thin filament Ca2<sup>+</sup> sensitivity. Thus, changes in duty ratio could explain the disrupted myofilament Ca2<sup>+</sup> sensitivity observed in studies of M2β mutations. Mutations in M2β can increase or decrease duty ratio (proposed to occur in HCM and DCM, respectively) by altering the kinetics of individual steps in the ATPase cycle and perturbing isometric force generation and thin filament Ca2<sup>+</sup> sensitivity. While it seems clear that HCM and DCM mutations in tropomyosin and the troponin complex are Ca2<sup>+</sup> sensitizing and desensitizing, respectively (Sommese et al., 2013; Spudich et al., 2016), it remains controversial how M2β cardiomyopathy mutations impact thin filament Ca2<sup>+</sup> sensitivity. The reduced Ca2<sup>+</sup> sensitivity and Hill coefficient of the human myocardial samples was observed in the presence of pN-Bleb,

but the difference is only significant at 50µM concentration. Therefore, our results suggest the strategy of altering the Ca2<sup>+</sup> sensitivity by altering the myosin duty ratio may be feasible.

## Considerations of Thick Filament Regulation

Regulation at the level of the thick filament can occur by formation of the SRX and drugs that alter the stabilization of this state could be utilized to enhance or depress force generation in cardiac muscle. The SRX has been identified as a state in striated muscle in which myosin heads are folded back on the backbone of the myosin thick filament (Hooijman et al., 2011). In addition, cryo-EM (Wendt et al., 2001; Craig and Woodhead, 2006; Zoghbi et al., 2008), and X-ray diffraction (Linari et al., 2015) studies have also demonstrated the presence of the folded back state of myosin in various muscle types from different species. The SRX provides a protective mechanism for maintaining a pool of quiescent myosin heads that have slow ATP hydrolysis (Fusi et al., 2015). Regulatory light chain phosphorylation, ablation of myosin binding protein C, and mechanical stress may impede the formation of the SRX (Linari et al., 2015; Kampourakis et al., 2016; McNamara et al., 2016). It has been proposed that cardiomyopathy associated mutations in myosin and thick filament associated proteins could disrupt contractile properties by altering the formation of the SRX, which ultimately impacts the number of cross-bridges capable of generating force (Kampourakis et al., 2016). Since the previous work on the parent drug suggests that Bleb may stabilize the SRX (Zhao et al., 2008; Xu et al., 2009; Wilson et al., 2014), the decrease in steady-state force and Ca2<sup>+</sup> sensitivity in the presence of pN-Bleb in the current study could be at least partially attributed to stabilization of the SRX.

### Comparison to Other Small Molecule Regulators

Currently, other drugs are being pursued that directly enhance or depress the activity of human M2β. Omecamtiv Mecarbil (OM) is a cardiac myosin allosteric modulator that is currently in Phase II clinical trials to treat systolic heart failure (Cleland et al., 2011; Greenberg et al., 2015; Teerlink et al., 2016). OM is specific to cardiac myosin with no effect on smooth or skeletal muscle myosin (Malik et al., 2011). While many studies have been done to investigate the impact of OM on muscle fibers in different animal models, the molecular mechanisms of how OM impacts cardiac myosin still remain unclear (Malik et al., 2011; Mamidi et al., 2015; Nagy et al., 2015; Utter et al., 2015). Steady state and transient kinetics have been examined to investigate the detailed mechanism of the impact of OM on purified porcine cardiac myosin (Liu et al., 2015). The kinetic analysis demonstrated that OM shifts the ATP hydrolysis equilibrium constant toward products and favors phosphate release, while the ADP release rate constant is unchanged. These changes translate to an increase in the number of force-generating cross-bridges bound to the thin filament in the presence of OM (Liu et al., 2015), which is consistent with the enhanced force production observed in muscle mechanic studies. The increase in the number of strongly bound force-generating heads in the presence of OM may exert an internal drag on the thin filaments that decreases sliding velocity. The internal drag could also slow the ADP release rate constant by a strain-dependent mechanism which would slow the detachment rate and thus the sliding velocity. Consistent with this hypothesis, the presence of OM dramatically inhibited the sliding velocity of porcine HMM (15- to 20-fold decrease) as measured in the in vitro motility assay in several studies (Wang et al., 2014; Liu et al., 2015; Winkelmann et al., 2015). OM has been found to slow force development as well as activation and relaxation kinetics but increase myofilament Ca2<sup>+</sup> sensitivity in isolated cardiomyocytes from rodent models (Mamidi et al., 2015; Nagy et al., 2015; Utter et al., 2015). Further study is necessary to determine the mechanism of how OM slows filament sliding in muscle mechanics and in vitro motility studies as well as how the impact on motor properties influences the contractile performance of the heart.

Another recent study identified a novel cardiac myosin inhibitor, MYK-461, which is proposed to suppress cardiac myosin motor function by decreasing duty ratio. MYK-461 reduces the overall ATPase activity of cardiac myosin in a dose dependent-manner (with a 90% maximal inhibition; Green et al., 2016). Transient kinetic experiments suggested that MYK-461 slows down the phosphate release step without changing the ADP release rate constant. Myofibril studies showed that the presence of 1 µM MYK-461 reduced maximal tension by 70%. Oral administration of MYK-461 decreased fractional cardiomyocyte shortening in wild-type and HCM-mutant mice, but importantly prevented the development of an HCM phenotype in the mutant mice. Therefore, MYK-461 normalizes the hyper-contractile properties of cardiac muscle by decreasing the power output, and suppresses the development of ventricular hypertrophy in mice carrying heterozygous human mutations (R403Q, R453C, R719W) in M2β (Green et al., 2016).

## CONCLUSIONS

We find that a M2β-S1 inhibitor (pN-Bleb) that acts by reducing strong actin binding without altering detachment kinetics may be advantageous for reducing the myosin duty ratio. The impact of the drug on muscle fiber studies demonstrates that this type of inhibition reduces steady-state force, power, and Ca2<sup>+</sup> sensitivity which may help treat hyper-contractile patients. The

## REFERENCES


maximum shortening velocity is not very sensitive to this type of inhibition in a muscle fiber while it is quite sensitive when examined in the motility assay with purified M2β-S1. Thus, it is important to consider the unique structural organization of muscle and how this may cause differences when comparing the in vitro motility and muscle mechanic studies. The ability to develop motility assays that better mimic a muscle fiber and retain the structural organization of the thick and thin filaments as well as the key regulatory proteins will be extremely helpful in future studies. The exciting new drug, MYK-461, which appears to act in a similar fashion to pN-Bleb, was successfully used to treat HCM in a mouse model. Further studies are necessary to determine if treatment of HCM with drugs that reduce the myosin duty ratio will be successful for a variety of HCM mutations. In addition, it will be interesting to determine if drugs that increase the myosin duty ratio, such as OM, can be used to treat patients with mutations that cause hypo-contractility.

## ETHICS STATEMENT

Study involved animal subjects; Approved by University of Kentucky Institutional Review Board, protocol 08-0338; Patients gave informed consent for sample donation before undergoing cardiac surgeries. Only specimens that were removed as part of normal clinical care and that would otherwise have been discarded were used in this study. No vulnerable populations.

### AUTHOR CONTRIBUTIONS

CY and KC designed research; AM-C provided critical reagents; WT, CB, SW, and CY performed research; WT, CB, KC, and CY analyzed the data; WT, CB, KC, and CY wrote paper. All authors approved the final version of the manuscript.

### ACKNOWLEDGMENTS

We thank William Unrath for outstanding technical assistance. The authors acknowledge support from American Heart Association 14GRNT20380068 and National Institutes of Health R01HL127699 grants to CY, American Heart Association 15GRNT25460003 grant to KC, National Institutes of Health grant UL1 TR000117, a Lyman T. Johnson Fellowship to CB and National Science Foundation grant No. 1538754.


under zero load in skeletal muscle. J. Physiol. doi: 10.1113/jp273299. [Epub ahead of print].


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Tang, Blair, Walton, Málnási-Csizmadia, Campbell and Yengo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Diaphragm Abnormalities in Patients with End-Stage Heart Failure: NADPH Oxidase Upregulation and Protein Oxidation

Bumsoo Ahn1 † , Philip D. Coblentz 1 †, Adam W. Beharry <sup>2</sup> , Nikhil Patel <sup>1</sup> , Andrew R. Judge<sup>2</sup> , Jennifer. S. Moylan<sup>3</sup> , Charles W. Hoopes <sup>4</sup> , Mark R. Bonnell <sup>5</sup> and Leonardo F. Ferreira<sup>1</sup> \*

<sup>1</sup> Department of Applied Physiology and Kinesiology, University of Florida, Gainesville, FL, USA, <sup>2</sup> Department of Physical Therapy, University of Florida, Gainesville, FL, USA, <sup>3</sup> Department of Physiology, University of Kentucky, Lexington, KY, USA, <sup>4</sup> Division of Cardiothoracic Surgery, University of Alabama at Birmingham, Birmingham, AL, USA, <sup>5</sup> Division of Cardiothoracic Surgery, University of Toledo Medical Center, Toledo, OH, USA

#### Edited by:

Jose Renato Pinto, Florida State University, USA

#### Reviewed by:

Brandon Biesiadecki, Ohio State University, USA Vasco Sequeira, VU University Medical Center, Netherlands

> \*Correspondence: Leonardo F. Ferreira ferreira@hhp.ufl.edu

† These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 30 September 2016 Accepted: 23 December 2016 Published: 09 January 2017

#### Citation:

Ahn B, Coblentz PD, Beharry AW, Patel N, Judge AR, Moylan JS, Hoopes CW, Bonnell MR and Ferreira LF (2017) Diaphragm Abnormalities in Patients with End-Stage Heart Failure: NADPH Oxidase Upregulation and Protein Oxidation. Front. Physiol. 7:686. doi: 10.3389/fphys.2016.00686 Patients with heart failure (HF) have diaphragm abnormalities that contribute to disease morbidity and mortality. Studies in animals suggest that reactive oxygen species (ROS) cause diaphragm abnormalities in HF. However, the effects of HF on ROS sources, antioxidant enzymes, and protein oxidation in the diaphragm of humans is unknown. NAD(P)H oxidase, especially the Nox2 isoform, is an important source of ROS in the diaphragm. Our main hypothesis was that diaphragm from patients with HF have heightened Nox2 expression and p47phox phosphorylation (marker of enzyme activation) that is associated with elevated protein oxidation. We collected diaphragm biopsies from patients with HF and brain-dead organ donors (controls). Diaphragm mRNA levels of Nox2 subunits were increased 2.5–4.6-fold over controls (p < 0.05). Patients also had increased protein levels of Nox2 subunits (p47phox, p22phox, and p67phox) and total p47phox phosphorylation, while phospho-to-total p47phox levels were unchanged. The antioxidant enzyme catalase was increased in patients, whereas glutathione peroxidase and superoxide dismutases were unchanged. Among markers of protein oxidation, carbonyls were increased by ∼40% (p < 0.05) and 4-hydroxynonenal and 3-nitrotyrosines were unchanged in patients with HF. Overall, our findings suggest that Nox2 is an important source of ROS in the diaphragm of patients with HF and increases in levels of antioxidant enzymes are not sufficient to maintain normal redox homeostasis. The net outcome is elevated diaphragm protein oxidation that has been shown to cause weakness in animals.

Keywords: diaphragm, carbonyls, weakness, inspiratory muscles, NOX2

## INTRODUCTION

Patients with end-stage heart failure (HF) have respiratory complications that contribute to disease morbidity and mortality. The incidence of pneumonia in end-stage HF patients who receive a heart transplant is 15–20% and 5% of patients have prolonged respiratory failure post-heart transplant (Lenner et al., 2001). The cause of pneumonia and respiratory failure in end-stage HF patients is multifactorial. In this context, inspiratory muscle abnormalities in patients with end-stage HF can play a critical role in respiratory complications (Kelley and Ferreira, 2016). For instance,

**279**

pre-operative inspiratory muscle strength training reduces the incidence of post-operative pulmonary complications in "highrisk" patients, including those with HF, undergoing coronary artery bypass graft (Hulzebos et al., 2006).

The primary inspiratory muscle is the diaphragm, which is necessary for normal ventilatory and expulsive behaviors that promote airway clearance (Sieck and Fournier, 1989; Mantilla and Sieck, 2013). Diaphragm dysfunction has been documented in animal models of HF, being characterized by atrophy and contractile impairments that diminish force and power capabilities (Howell et al., 1995; Stassijns et al., 1998; van Hees et al., 2007; Ahn et al., 2015; Kelley and Ferreira, 2016). The diaphragm of end-stage HF patients also shows ultra-structural and myofibrillar protein alterations that suggest metabolic and contractile abnormalities (Lindsay et al., 1996). However, less is known about the underlying mechanisms of diaphragm dysfunction in end-stage HF.

Excess reactive oxygen species (ROS) and redox imbalance play a causal role in diaphragm dysfunction in animal models of HF (Supinski and Callahan, 2005; Ahn et al., 2015). An important source of ROS in the diaphragm is the Nox2 isoform of NAD(P)H oxidase (Pal et al., 2013; Loehr et al., 2014; Bost et al., 2015). A functionally assembled Nox2 enzyme complex consists of several subunits (Nox2, p22phox, p67phox, p40phox , Rac1/2, and p47phox), and enzyme activation requires p47phox phosphorylation (Javesghani et al., 2002; Lassègue et al., 2012; Pal et al., 2013; Ferreira and Laitano, 2016). Diaphragm p47phox phosphorylation is increased in mice with HF, and genetic deletion of p47phox prevents excess ROS release and contractile dysfunction in diaphragm of mice with HF (Ahn et al., 2015). These findings in mice suggest a pivotal role for Nox2-derived ROS on diaphragm dysfunction in HF.

Diaphragm antioxidant enzymes are unchanged (Ahn et al., 2015) or elevated in animal models of HF (Bowen et al., 2015b; Mangner et al., 2015). The increase in antioxidant enzymes may reflect a compensatory response to scavenge excess ROS. However, these compensatory responses appear insufficient to maintain cellular redox homeostasis as markers of protein oxidation are elevated in diaphragm of HF animals (acute Bowen et al., 2015a and chronic models Supinski and Callahan, 2005; Coirault et al., 2007). This is relevant because heightened ROS and protein oxidation promote diaphragm atrophy and impair contractile function (Ferreira and Reid, 2008; Powers et al., 2011).

Despite advances in understanding the causes and role of redox imbalance on diaphragm dysfunction in animal models of HF, very little is known about ROS sources, antioxidant enzymes, and protein oxidation in the diaphragm of patients. We tested the hypotheses that diaphragm from patients with end-stage HF show heightened Nox2 subunit levels and p47phox phosphorylation with unchanged protein levels of antioxidant enzymes which results in elevated markers of protein oxidation.

### METHODS

### Human Subjects

We obtained diaphragm biopsies from patients with HF with reduced ejection fraction who underwent surgery for heart transplant or placement of left ventricular assist device. Patients were informed of the nature and purpose of the study and signed a written consent in accordance with the Declaration of Helsinki. Control subjects were brain-dead organ donors whose family consented to the diaphragm biopsies. The protocol and sample analyses were approved by the Institutional Review Boards of the University of Kentucky or University of Florida.

### Tissue Collection

The cardiothoracic surgeons (MRB and CWH) obtained diaphragm biopsies and placed them in ice- cold sterile saline. The samples were rapidly processed in the operating room to clear any visible connective tissue and excess blood, then frozen in liquid nitrogen and stored at −80◦C for further processing as described below. All procedures for tissue collection and analyses were approved by the Institutional Review Boards of the University of Kentucky or University of Florida.

### qPCR

We isolated total RNA from human diaphragm tissue with Trizol reagent. We then used Ambion RETROscript First Strand Synthesis Kit (Life Technologies, Carlsbad, CA, USA) to generate cDNA from 1 µg of RNA. The cDNA was then used as template for qRT-PCR (7300 real-time PCR system, Applied Biosystems, Austin, TX). We used TaqMan <sup>R</sup> PCR assay primers from Life Technologies targeting the following genes and NCBI Reference Sequence numbers: Nox2 (CYBB, NM\_000397.3), p47phox (NCF1 NM\_000265.5), Rac1 (RAC1, NM\_006908.4), p22phox (CYBA, NM\_000101.3), p67phox (NCF2, NM\_000433.3), and p40phox (NCF4, NM\_000631.4). Gene expression quantification was performed using the relative standard-curve method, and all data were normalized to the gene expression of 18S (GeneBank NM\_X03205.1) and reported relative to the control group.

### Immunoblotting

We loaded ∼10–50 µg of protein into 4–20% stain-free TGX gels (Bio-Rad Laboratories) and performed electrophoresis at 230 V for 40 min on ice. We scanned the gel to quantify total proteins (Gel DocTM EZ Imager, Bio-Rad Laboratories) and then transferred the proteins to a nitrocellulose membrane at 100 mA overnight at 4◦C. We blocked the membrane using Li-COR Blocking Buffer for 1 h at room temp and subsequently probed with primary antibodies. As markers of protein oxidation, we measured protein carbonyls (OxySelectTM Protein Carbonyl Immunoblot kit, Cell Biolabs), 4-hydroxynonenal (4-HNE, Ab46545, AbCam) adducts, and 3-nitrotyrosines (3-NT, 189542, Cayman Chemical). To probe for sources of ROS, we used primary antibodies targeting Nox2 (CYBB, 1:500 dilution, sc-5827, Santa Cruz), p22phox (CYBA, 1:50 dilution, FL-195, Santa Cruz), p67phox (NCF2, 1:50 dilution, sc-7663, Santa Cruz), Rac1 (RAC1, 1:1000 dilution 05-389, Millipore), p47phox (NCF1, 1:1000 diltuion, SAB2500674, Sigma-Aldrich), and phosphorylated p47phox at serine residues 345 (orb126026, Biobyrt), 370 (A1171, Assay Biotech), 359 (A1172, Assay Biotech), 328 (A1161, Assay Biotech), and 304 (A1160, Assay Biotech). The dilution for antibodies targeting serine residues was 1:1000. For antioxidant enzymes, we used antibodies targeting superoxide dismutase isoform 1 (SOD1; 1:500 dilution, FL-154, Santa Cruz), SOD2 (1:500 dilution, FL-122, Santa Cruz), catalase (1:1000 dilution, Ab16731, Abcam), and glutathione peroxidase (1:1000 dilution, Ab108427, Abcam). We diluted the primary antibodies in LiCor Blocking Buffer, incubated the membranes for 72 h at 4◦C or 1 h at room temperature, and washed in TBS-T (Tris-buffered saline with 0.1% Tween) 4 × 5 min each. We then incubated the membranes in secondary antibodies (IR Dye, LI-COR) in Li-COR Blocking Buffer for 1 h at room temp, followed by 3 × 5 min washes in TBS-T and a 5 min rinse in TBS. We dried the membranes in an incubation chamber at ∼37◦C for 15 min and scanned the fluorescence signal using an Odyssey Infrared Imaging system (LI-COR, Lincoln, NE). We quantified the immunoblot signal using Image Studio Lite <sup>R</sup> (Li-COR) and the abundance of total protein in each gel lane using ImageLab (Bio-Rad Laboratories). The immunoblot signal of each target protein was normalized to the total protein measured in corresponding gel lanes. These procedures are consistent with recent recommendations for data analysis of Western blots using fluorescence methods and stain- free gels (Eaton et al., 2013; Murphy and Lamb, 2013).

### Statistical Analysis

We performed statistical analysis using SigmaPlot v.12.5 (Systat Software, San Jose, CA). For specific comparisons, we used t-test or Mann-Whitney rank sum test for data that failed normality (Shapiro-Wilk test). Non-parametric data are presented as median ± interquartile range and shown in box and whisker plots. We declared statistical significance when P < 0.05.

### RESULTS

Patient characteristics are detailed in **Table 1**. In summary, patients exhibited HF caused by ischemic (n = 5) and nonischemic cardiomyopathy (n = 6).

Diaphragm mRNA levels of Nox2, p22phox, p47phox, p67phox , and p40phox were increased with median values ranging from 2.5- to 4.6-fold over controls (**Figure 1**). The protein levels of Nox2, p47phox, p22phox, and p67phox were also increased in diaphragm of HF patients, while protein levels of Rac1 was not significantly changed (**Figure 1**). We were not able to detect p40phox via immunoblot in the diaphragm, which is consistent with a previous study (Javesghani et al., 2002).



HF, heart failure. Data are mean ± SE. Ejection fraction is shown as median (interquartile range) from 3 controls and 6 patients. \*P < 0.05 compared to control.

Phosphorylation of p47phox promotes activation of Nox2 (El-Benna et al., 2009; Lassègue et al., 2012), thus we examined the phosphorylation status of p47phox using antibodies against specific phosphorylated serine residues. We found that phosphorylation at Ser328, Ser345, Ser359, and Ser370 were increased in diaphragm of HF patients compared to controls when normalized to total protein (**Figure 2**). When we normalized the phosphorylated signal from each serine residue to the total p47phox signal, there was no difference in phosphoto-total p47phox between control and HF patients (**Figure 2**). This suggests that the total abundance of p47phox protein in the phosphorylated state was elevated, whereas the "percentage" of phosphorylated p47phox was unchanged.

We measured the protein level of key cytosolic and mitochondrial antioxidant enzymes (**Figure 3**). Catalase levels were increased 2.2-fold over controls (P < 0.05), whereas the levels of glutathione peroxidase (P = 0.15), SOD1 (P = 0.48), and SOD2 (P = 0.22) were unchanged in the diaphragm of HF patients.

Despite increased levels of catalase, redox imbalance in diaphragm from patients was manifested by increased (∼40%) protein carbonyls, while 4-HNE and 3-NT were unchanged in the diaphragm of end- stage HF patients (**Figure 4**).

### DISCUSSION

Our study shows that diaphragm of patients with end-stage HF have elevated mRNA and protein levels of Nox2 subunits that is accompanied by increased p47phox phosphorylation, which is consistent with Nox2 activation. The antioxidant enzyme catalase was also increased in diaphragm of patients, while superoxide dismutases and glutathione peroxidase were unchanged. These findings suggest disrupted redox homeostasis in the diaphragm of patients with end-stage HF, which are confirmed by elevated levels of protein carbonyls.

The enzyme Nox2 is emerging as an important source of oxidants that cause diaphragm abnormalities in animal models of diseases, including muscular dystrophy (Whitehead et al., 2010; Pal et al., 2014; Henriquez-Olguin et al., 2015) and HF (Ahn et al., 2015). The functionally assembled Nox2 complex includes several subunits (Nox2, p47phox, p22phox, p40phox , p67phox, and Rac1 Bedard and Krause, 2007; Lassègue et al., 2012). We observed that mRNA levels of several Nox2 subunits were increased in the diaphragm of end-stage HF patients, with p47phox subunit having the highest elevation (**Figure 1**).

Increased p47phox mRNA was translated into higher protein abundance compared to control (**Figure 1**). Similarly, we have found heightened protein levels of p47phox in the diaphragm of mice with HF (Ahn et al., 2015). It is unclear whether elevated p47phox is sufficient to heighten Nox2 activity in skeletal muscle cells. Overexpression of p47phox increases Nox2 activity in glial cells (Lavigne et al., 2001). Thus, it is possible that skeletal muscle cells have a constitutive, p47phox-dependent Nox2 activity.

The canonical pathway for Nox2 activation involves p47phox phosphorylation at serine residues that releases auto-inhibition

of membrane- and subunit-binding domains (El-Benna et al., 2009; Drummond et al., 2011; Lassègue et al., 2012). The Cterminal domain of p47phox contains 11 serine residues between amino acids 303–379 that encompasses the auto-inhibitory region. Point mutations have revealed six serine residues of p47phox that are required for full activation of Nox2: Ser303, Ser304, Ser328, Ser359, Ser370, and Ser379 (reviewed in El-Benna et al., 2009). We found that total levels of phosphorylated Ser328, Ser345, Ser359, and Ser370 were elevated in diaphragm of patients with end- stage HF (**Figure 2**). When we calculated the phospho-to-total p47phox ratio, differences between HF and controls were not statistically significant (**Figure 2C**). These findings suggest that increased "absolute" levels of phosphorylated p47phox accompanied the heightened expression of total p47phox. However, the "percentage" (or relative levels) of p47phox protein in the phosphorylated state was unchanged. Typically, an increase in relative levels/percentage of p47phox phosphorylation (i.e., elevated phospho-to-total p47phox) is considered an indicator of Nox2 activation (Isabelle et al., 2005). In the context of elevated levels of total p47phox, unchanged phospho-to-total p47phox ratio should also heighten Nox2 activation because phosphorylated p47phox proteins are more abundant. Biologically, the absolute amount of phosphorylated p47phox would dictate Nox2 activity. Therefore, we consider that Nox2 activity is likely increased in diaphragm of end-stage HF patients. However, the unchanged relative levels/percentage of phospho-p47phox has implications regarding mechanisms of Nox2 activation. Our data suggest that the activity of kinases that phosphorylate p47phox is not necessarily elevated in diaphragm of patients with end-stage HF. It is possible that other pathways of p47phox signaling (e.g., arachidonic acid) or overexpression per se mediate Nox2 activation in the diaphragm (Ferreira and Laitano, 2016). Alternatively, enhanced diaphragm p47phox phosphorylation may be involved in the pathophysiology of diaphragm dysfunction at earlier stages of the disease.

We have not tested Nox2 activity in our study because our tissue collection method (flash freezing) does not lend the sample suitable for reliable measurements of activity, as per recent recommendations (Rezende et al., 2016). We have found, in diaphragm of mice with HF, increases in total p47phox and phospho-to-total p47phox that are consistent with increases in Nox2 activity (Ahn et al., 2015). Indeed, knockout of p47phox prevented excess diaphragm ROS emission suggesting Nox2 as a major source of pathological diaphragm oxidants in HF (Ahn et al., 2015). Overall, our data in humans and animals suggest elevated Nox2 activity in patients with HF.

A decrease in protein levels or intrinsic activity of antioxidant enzymes will contribute to ROS accumulation that disrupts cellular redox balance. Major intracellular antioxidant enzymes include catalase, glutathione peroxidase, and superoxide dismutases (SOD1 and SOD2). Patients had increased levels of catalase, whereas there was no statistical difference in the

levels of GPX, SOD1, and SOD2 between patients and controls (**Figure 3**). This outcome is likely due to our limited sample size and large variability in the human diaphragm data. In animal models of HF with reduced ejection fraction, diaphragm levels of SOD1 or SOD2 were either unchanged (Ahn et al., 2015; Laitano et al., 2016) or elevated (Mangner et al., 2015), whereas the activity of GPX was increased (Mangner et al., 2015) and catalase was unchanged (Mangner et al., 2015). Nonetheless, diaphragm catalase activity was increased in a model of HF with preserved ejection fraction (Bowen et al., 2015b). In general, our data in patients and studies in animals suggest that heightened protein oxidation in the diaphragm induced by HF cannot be explained by a decrease in the protein levels or activity of antioxidant enzymes. Heightened diaphragm antioxidant enzyme levels in HF might be a compensatory adaptation aimed to maintain redox balance when ROS production is increased. However, our findings suggest that any compensatory response is insufficient to maintain normal protein oxidation levels in patients with end-stage HF.

Protein carbonyls, a marker of oxidation, were elevated in the diaphragm of end-stage HF patients (**Figure 4**). These results corroborate previous findings in animals with severe HF induced by aortic stenosis (Coirault et al., 2007) or in the early stages postmyocardial infarction (Supinski and Callahan, 2005; Bowen et al., 2015a), but disagrees with data from our group in rats and mice with moderate HF in the later stages post-myocardial infarction (Ahn et al., 2015; Laitano et al., 2016). We speculate that the increase in diaphragm protein carbonyls in chronic HF is related to disease severity. This concept is supported by progressive increases in systemic markers of oxidation in groups of patients going from NYHA Class I to IV (Belch et al., 1991; Nishiyama et al., 1998).

Excess oxidation can contribute to diaphragm abnormalities due to protein degradation as well as impaired function of excitation-contraction coupling and sarcomeric proteins. Oxidation enhances protein degradation by calpain, caspase-3, and the proteasome (Grune et al., 2003; Moylan and Reid, 2007; Smuder et al., 2010). Calpain and proteasome activity are elevated in diaphragm of HF rats (Dominguez and Howell, 2003; van Hees et al., 2008a), and proteasome inhibition prevents myofibrillar protein degradation and attenuates diaphragm weakness in HF rats (van Hees et al., 2008a). Regarding protein function, carbonylation of sarcomeric proteins in vitro impairs actomyosin cross- bridge kinetics and, in general, exposure to oxidants depresses diaphragm force and shortening velocity (Perkins et al., 1997; Callahan et al., 2001; Coirault et al., 2007). These functional outcomes are consistent with impaired diaphragm contractile

(A) Representative immunoblots. (B) Grouped data from control (n = 3) and all patients with HF (n = 11). (C) Individual data from patients with ischemic and non-ischemic cardiomyopathy. Representative protein gels are similar to that shown in Figure 1. \*P < 0.05 vs. control by Mann-Whitney test.

properties in HF animals (Lecarpentier et al., 1998; Coirault et al., 2007; van Hees et al., 2007, 2008b; Empinado et al., 2014; Ahn et al., 2015), which are prevented by pharmacologic antioxidants (Supinski and Callahan, 2005; Laitano et al., 2016) or knockout of p47phox (Ahn et al., 2015).

### Limitations

There are several limitations in our study that must be considered for data interpretation. These limitations include: (A) age and sex of controls and patients: Patients in the HF group were mostly males with both ischemic and non-ischemic cardiomyopathy, while controls were all females and younger than HF patients. Inspiratory muscle weakness and diaphragm abnormalities are relevant for patients with ischemic and non-ischemic HF (Ambrosino et al., 1994; Lindsay et al., 1996; Tikunov et al., 1997; Daganou et al., 1999; Filusch et al., 2011) as well as male and female patients (Ambrosino et al., 1994; Lindsay et al., 1996; Dall'Ago et al., 2006). We did not have sufficient number of patients with ischemic and non-ischemic cardiomyopathy to resolve potential statistical differences due to etiology of disease. Inspection of data from patients in each group suggest that overall the changes were consistent for both ischemic and non-ischemic cardiomyopathy. Nonetheless, in this data set there is a general trend for exacerbated effects in patients with

non-ischemic cardiomyopathy. The data from the female HF patient were consistent with those from males. For instance, p47phox levels from the female HF patient corresponded to 9.5-fold (mRNA) and 4.5-fold (protein) of the control mean. The lack of age- and sex-matched data in humans reflects the nature of our study and focus on the diaphragm that presents difficulty for obtaining biopsies, especially from control subjects. (B) Mechanical ventilation: Brain-dead organ donors undergo mechanical ventilation, which heightens diaphragm protein oxidation (Betters et al., 2004). We do not have information on the duration of mechanical ventilation in our control subjects. It is possible that controls underwent longer periods of mechanical ventilation than our HF patients experienced during surgery. However, this would minimize rather than accentuate differences in the variables that we studied. (C) Inability to establish cause-and-effect: we cannot establish a causal relationship between Nox2 levels/activity, protein oxidation, and diaphragm abnormalities in HF patients. Instead, our data should serve as an impetus for clinical trials testing selective Nox2 inhibitors or pharmacological antioxidants to treat diaphragm abnormalities and its associated complications in end-stage HF patients.

It is worth noting that important studies relying on diaphragm biopsies of end-stage HF patients and controls also had an unbalanced distribution of age or sex and included brain-dead controls or patients with HF due to several causes (Lindsay et al., 1996; Tikunov et al., 1996, 1997). Finally, we cannot attribute our findings to proteins within diaphragm muscle cells per se. In addition to muscle fibers, several other cell types within the diaphragm express Nox2 subunits, e.g., endothelium, smooth muscle, and macrophages.

### REFERENCES


### CONCLUSION

Diaphragm of patients with end-stage HF shows upregulation of Nox2 subunits, increased total but unchanged relative levels of phosphorylated p47phox, and elevated abundance of catalase. These changes in ROS-producing and scavenging enzymes culminated in elevated diaphragm protein oxidation. Overall, our findings suggest that Nox2 is an important source of ROS in the diaphragm of patients with end-stage HF and increases in catalase levels are not sufficient to maintain cellular redox homeostasis.

### AUTHOR CONTRIBUTIONS

Sample collection and processing: JM, CH, MB, and LF. Experiments: BA, PC, AB, JM, NP, and LF. Data analysis and interpretation: BA, PC, AB, JM, NP, MB, CH, AJ, and LF. Manuscript writing and editing: BA, PC, AJ, and LF. Patients assessment and surgeries: CH and MB.

### FUNDING

This study was funded by NIH grant R00-HL098453 to LF. AJ was funded by NIH grant R01 AR060209.

### ACKNOWLEDGMENTS

We would like to thank Drs. Kenneth S. Campbell, Premi Haynes, Mihail Mitov, Stuart Campbell, and Shawn Stasko for assistance with administrative and technical procedures for human tissue collection.


as therapeutic targets. Nat. Rev. Drug Discov. 10, 453–471. doi: 10.1038/ nrd3403


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Ahn, Coblentz, Beharry, Patel, Judge, Moylan, Hoopes, Bonnell and Ferreira. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Insights and Challenges of Multi-Scale Modeling of Sarcomere Mechanics in cTn and Tm DCM Mutants—Genotype to Cellular Phenotype

Sukriti Dewan1 †, Kimberly J. McCabe1 †, Michael Regnier <sup>2</sup> and Andrew D. McCulloch<sup>1</sup> \*

*<sup>1</sup> Departments of Bioengineering and Medicine, University of California, San Diego, La Jolla, CA, USA, <sup>2</sup> Departments of Bioengineering and Medicine, University of Washington, Seattle, WA, USA*

### Edited by:

*P. Bryant Chase, Florida State University, USA*

### Reviewed by:

*Kenneth S. Campbell, University of Kentucky, USA Jose Renato Pinto, Florida State University, USA John Jeremy Rice, Functional Genomics and Systems Biology, USA*

\*Correspondence:

*Andrew D. McCulloch amcculloch@eng.ucsd.edu †Co-first authors.*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *02 November 2016* Accepted: *24 February 2017* Published: *14 March 2017*

#### Citation:

*Dewan S, McCabe KJ, Regnier M and McCulloch AD (2017) Insights and Challenges of Multi-Scale Modeling of Sarcomere Mechanics in cTn and Tm DCM Mutants—Genotype to Cellular Phenotype. Front. Physiol. 8:151. doi: 10.3389/fphys.2017.00151* Dilated Cardiomyopathy (DCM) is a leading cause of sudden cardiac death characterized by impaired pump function and dilatation of cardiac ventricles. In this review we discuss various *in silico* approaches to elucidating the mechanisms of genetic mutations leading to DCM. The approaches covered in this review focus on bridging the spatial and temporal gaps that exist between molecular and cellular processes. Mutations in sarcomeric regulatory thin filament proteins such as the troponin complex (cTn) and Tropomyosin (Tm) have been associated with DCM. Despite the experimentally-observed myofilament measures of contractility in the case of these mutations, the mechanisms by which the underlying molecular changes and protein interactions scale up to organ failure by these mutations remains elusive. The review highlights multi-scale modeling approaches and their applicability to study the effects of sarcomeric gene mutations *in-silico*. We discuss some of the insights that can be gained from computational models of cardiac biomechanics when scaling from molecular states to cellular level.

Keywords: dilated cardiomyopathy, troponin, Tropomyosin, cross-bridge, myofilament, modeling, mechanics, Markov

## INTRODUCTION

Dilated Cardiomyopathy (DCM) is one of the four classified forms of cardiomyopathy besides hypertrophic cardiomyopathy (HCM), restrictive cardiomyopathy (RCM), and arrythmogenic right ventricular dysplasia/cardiomyopathy (ARVD/C) (Richardson et al., 1996; Seidman and Seidman, 2001). DCM, one of the major causes of cardiac death, is characterized by impaired systolic function and dilatation of one or both ventricles (Richardson et al., 1996; Kärkkäinen and Peuhkurinen, 2007). Hemodynamically, contractility is depressed and Pressure-Volume (PV) loops are right-shifted in DCM. In 30–50% of cases DCM is linked to familial etiology, including mutations in the regulatory thick and thin myofilament proteins - myosin, actin, the Troponin (Tn) complex, Tropomyosin (Tm), and Titin (Ttn) (Kärkkäinen and Peuhkurinen, 2007; Lu et al., 2013). Many times DCM presents with conduction defects and sequelae of other cardiac defects. This is more common with cytoskeletal and z-disc mutations (Seidman and Seidman, 2001; Chang and Potter, 2005). Genome Wide Association Studies (GWAS) have helped identify many of the sarcomeric mutations associated with the phenotype of DCM (Kamisago et al., 2000; Li et al., 2001; Olson et al., 2001; Mogensen et al., 2004; Murphy et al., 2004; Lakdawala et al., 2010, 2012a; Branishte et al., 2013; Pérez-Serra et al., 2016). Despite the identification of sarcomeric mutations associated with DCM, it is still difficult to predict the exact functional consequences of the mutation at the cellular level based on its molecular structure and function. Identified sarcomeric mutations may represent a gain of function or loss of function at the cellular level, as assessed by myofilament mechanics (Kamisago et al., 2000; Spudich and Rock, 2002; McNally et al., 2013). Furthermore, a spatial and temporal translation of the cellular level mechanical phenotype to an organ level DCM phenotype has not been seamlessly achieved. These patho-physiological translative events from genetic to molecular to cellular to organ level, underscore the need for investigation at and across all biological scales to fully comprehend the development of DCM (Spudich, 2014).

Previous experimental studies, incorporating structural data from x-ray crystallography and nuclear magnetic resonance (NMR) into molecular dynamic (MD) models and in-vitro molecular assays, have reported changes in molecular properties of various DCM associated sarcomeric mutants. Findings include altered calcium binding affinity (Robinson et al., 2007; Kekenes-Huskey et al., 2012), rate of cTnC hydrophobic patch opening (Dewan et al., 2016), acto-myosin affinity (Moore et al., 2012), altered surface charge distribution on coiled-coil region of Tm (Olson et al., 2001), Titin to z-disc protein Tcap/Telethonin affinity (Thirumal Kumar et al., 2016) and cTnCcTnI interactions (Mogensen et al., 2004; Dewan et al., 2016). The impact of molecular alteration to cellular phenotype can be qualitatively or semi-quantitatively intuitive in some cases. For example a decrease in calcium binding affinity of cTnC will lead to decreased thin filament activation and force development. However, the exact quantification of such an effect is still to be reported for various mutants and unanswered questions remain. Will a 25% decrease in calcium binding affinity of cTnC lead to a 50% decrease in thin filament activation which would result in 50% decrease in maximal force response or will it trigger compensatory molecular mechanisms and result in a 10% decrease in thin filament activation and force generation? Or does a 10% decrease in thin filament activation imply a 20% decrease in calcium binding affinity of cTnC? As an example, a recent study reported that the severity of DCM is determined by the ratio of mutant to wildtype TnnT2 gene transcript in 1K210 cTnT transgenic mice, as absence of one allele of TnnT2 does not lead to a protein deficit (Ahmad et al., 2008). The quantitative translation of molecular perturbation to cellular events and vice versa warrants more studies. Nonetheless, it has been reported that identified genetic mutations in contractile active force-generating sarcomeric proteins (excluding Titin) exhibit a very specific DCM phenotype without any other associated cardiac phenotype such as hypertrophy or conduction defects with a high prevalence in young people (Mogensen et al., 2004; Memo et al., 2013). This would suggest that diverse molecular abnormalities converge at the cellular phenotype, which then trigger the cardiac remodeling leading to DCM. Indeed depressed contractile function is a common cellular phenotype in DCM despite variable molecular mechanisms (Mirza et al., 2005).

Experimental studies, using the skinned muscle preparation or intact cardiomyocytes from gene-targeted mouse models (Du et al., 2007), adenoviral-mediated transfection (Morimoto et al., 2002; Lu et al., 2003; Mirza et al., 2005; Lim et al., 2008), in-vitro protein exchange experiments (Dweck et al., 2010) for in-vitro protein motility assays (Mirza et al., 2005; Memo et al., 2013), mammalian two-hybrid luciferase assay system (Mogensen et al., 2004; Murphy et al., 2004), steady-state forcecalcium assays (Morimoto et al., 2002; Lu et al., 2003), and contractility assays (Biesiadecki et al., 2007; Lim et al., 2008; Dweck et al., 2010), have reported alterations in key mechanical properties of myofilaments in DCM. These include changes in calcium sensitivity (Robinson et al., 2002; Mirza et al., 2005; Du et al., 2007; Lim et al., 2008; Dweck et al., 2010; Memo et al., 2013; Kalyva et al., 2014), thick and thin myofilament cooperativity and cross-bridge (XB) cycling rates (Moore et al., 2012), as a functional outcome of many identified sarcomeric DCM mutations. Most studies have reported a decrease in calcium sensitivity of myofilaments as a consistent functional phenotype for DCM identified sarcomere mutations (Murphy et al., 2004; Mirza et al., 2005; Du et al., 2007; Robinson et al., 2007; Memo et al., 2013). Interestingly, I61Q cTnC mutant in mice (neither found nor associated with DCM clinically), that has decreased calcium myofilament binding affinity, recapitulates DCM phenotype cellularly and in vivo (Davis et al., 2016). However, exceptions to this have been reported wherein both an increase and no change in calcium sensitivity were reported in DCM (Dweck et al., 2008; Memo et al., 2013). Dweck et al. reported that there is a decrease in calcium sensitivity of tension development in G159D cTnC mutant only when it is incorporated in regulated actomyosin filaments and not in isolated cTnC (Dweck et al., 2008), highlighting the effect of protein interactions as the hierarchy of structural organization becomes more physiological and complex in the contractile apparatus. Furthermore, Memo et al. reported a decrease in calcium sensitivity of myofilaments for four DCM associated mutants (K36Q TnI, R141W TnT, 1K210 TnT, E40K Tm), no change in calcium sensitivity of one mutant (E54K Tm) and increase in calcium sensitivity in another (D230N Tm) (Memo et al., 2013). These observations suggest that decreased calcium sensitivity is a dominant stimulus sufficient to cause DCM, but is neither necessary, nor the only cellular mechanism triggering the remodeling observed in DCM. A recent study postulated that the blunting of the relationship between calcium sensitivity of myofilaments and PKA mediated beta-adrenergic stimulation via cTnI phosphorylation (Memo et al., 2013) in sarcomeric DCM mutants might be the defining cellular phenotype for DCM regardless of the directional shift in calcium sensitivity. It should be noted that aside from DCM mutations in contractile sarcomere proteins that lead to diminished force production, mutations in other key cytoskeletal and sarcomere proteins like the z-disc proteins and Titin can lead to disruptions in transmission of force, sensing of force and mechanotransduction which are also causative toward DCM (Chang and Potter, 2005). These observations suggest that molecular interactions and effects of various DCM mutants converge to a depressed contractile phenotype at the cellular level due to alterations in (a) calcium sensitivity of myofilaments, (b) thinfilament activation, (c) maximal ATPase activity, (d) in-vitro motility (e) calcium affinity of Tn and (f) mechano-transduction, thereby triggering the signaling mechanism leading to DCM (Chang and Potter, 2005; Lakdawala et al., 2012b). Further studies are warranted to establish the key converging cellular mechanism/s and signaling pathway/s in DCM.

It is interesting to note that both HCM and DCM can be caused by different mutations within the same contractile protein gene. HCM is a more common familial disease with a contrasting phenotype of hypertrophied ventricle and preserved systolic function. These observations imply that there are either two different signaling pathways distinguishing these phenotypes or a graded response within the same pathway. An earlier study in transgenic mice with a truncation allele of Myosin Binding protein C (MyBP-C), known to cause HCM in man, exhibited a graded response such that heterozygotes resulted in HCM and homozygotes in DCM (McConnell et al., 1999). In contrast, Ahmad et al. and Ramratnam et al. reported that the ratio of mutated to wildtype transcript of TnT is critical in determining severity of DCM and not haploinsufficiency (Ahmad et al., 2008; Ramratnam et al., 2016). Nonetheless, given that most studies report a contrasting cellular phenotype to HCM with directionally opposite shifts in calcium sensitivity, two separate pathway theory is strongly supported. This is corroborated by a recent study where the authors propose that modeling the tension integral of cardiomyocytes can help distinguish between DCM and HCM (Davis et al., 2016). Notably, a recent RNAseq study profiling molecular signaling in DCM and HCM reported that profibrotic and metabolic networks can distinguish between the two phenotypes (Burke et al., 2016). Multi-scale studies quantifying genotype to functional cellular phenotype will help address these theories. The convergence of various molecular defects to fewer cellular phenotypes to a singular DCM phenotype via cardiac remodeling is notable.

Clinically, mutations are prevalent from birth in familial DCM. However, the temporal transition to DCM is not yet understood (Tardiff, 2012). A recent study using tissue Doppler and strain echocardiography, showed that even early on there are subtle indications (Lakdawala et al., 2012b). In subclinical DCM mutation carriers, reduced systolic myocardial velocity, strain and strain-rate were reported despite normal LV geometry, ejection fraction and diastolic function (Lakdawala et al., 2012b). Early diagnosis in such cases can provide with much needed time (Lakdawala et al., 2012b). It is also important to note that some HCM patients develop dilated ventricles at later stages for example in patients with mutations E180V in Tm and R92W in cTnT (Chang and Potter, 2005). Studies have shown this to be distinct from DCM phenotype (Ohba et al., 2007). So in familial DCM at the whole heart level there is subtle manifestation of DCM from birth and transition to late stage DCM eventually. Mechanical and pathway driven changes lead to this cardiac remodeling (Burke et al., 2016; Davis et al., 2016). Investigations on the effects of DCM mutations quantitatively and qualitatively, in animal model studies, linking genotype to muscle phenotype i.e., from molecular level changes to myofilament level changes to intact tissue level changes to whole heart level changes are scarce (Ahmad et al., 2008; Ramratnam et al., 2016). This is due to the technical challenges and expense of carrying out such an expansive study to experimentally measure data on a single DCM mutation. Additionally, as studies report cellular data, isolating the singular effect of the genetic mutation becomes difficult due to downstream effects of the mutation that also contribute to the systemic perturbations leading to the phenotype.

A potential way to study the mechanical effects of genetic mutations comprehensively is by multi-scale computational modeling. Cardiac muscle biomechanics has been experimentally and computationally investigated in molecular machineries of sarcomeric protein complexes (Varguhese and Li, 2011; Lindert et al., 2012b, 2015), in thick and thin myofilaments (Rice and de Tombe, 2004; Campbell et al., 2010; Dewan et al., 2016), in isolated cardiomyocytes under steady-state and dynamic conditions (Hussan et al., 2006; Rice et al., 2008), in myocardium, and in the whole heart (Göktepe et al., 2010; Campbell and McCulloch, 2011; Trayanova, 2011; Kerckhoffs et al., 2012; Zhang et al., 2016). An integrative mathematical formulation to scale from protein level changes to the cardiomyocyte function is yet to be implemented. Parameterization of computational models requires collating data from various studies for a single mutation at various spatial and temporal scales. This results in data from variable experimental conditions and requires standardization in terms of (a) temperature, (b) species, and (c) bridging of spatial and temporal scales, in order to input model parameters (Tøndel et al., 2015; Dewan et al., 2016). In this review we discuss various in silico approaches and the challenges thereof, by examining previously identified and experimentally studied DCM mutations in thin filament proteins Tn and Tm, with the goal of elucidating the mechanical effects of sarcomeric mutations from molecular level to the cellular scales.

### CARDIAC MUSCLE CONTRACTION AND EFFECTS OF DCM MUTANTS IN REGULATORY THIN FILAMENT PROTEINS cTn AND TM

Cardiac muscle contraction is triggered following calcium induced calcium release (CICR) from the sarcoplasmic reticulum (SR) (Fabiato, 1983). In 1954, two groundbreaking studies proposed the sliding filament theory of muscle contraction describing the molecular basis of muscle contraction (Huxley and Niedergerke, 1954; Huxley and Hanson, 1954). According to the sliding filament theory, thin (actin) and thick (myosin) filaments slide past each other, while maintaining absolute lengths, during contraction to generate contractile force. Cardiac TnC (cTnC), a 18 kDa thin filament protein within the Tn complex, is a key regulator of muscle contraction. Calcium binding to cTnC triggers the biomechanical cascade of contraction events within the sarcomere (Gordon et al., 2000). Calcium binding to cTnC induces a conformational change within the Tn complex, thereby displacing Tropomyosin (Tm) from actin filaments to expose

myosin binding sites and increasing the probability of crossbridges cycling. In the resting state of the myofilaments when Ca2<sup>+</sup> is not bound to cTnC, cTn complex anchors Tm in a "Blocked" position, thereby, sterically hindering access to sites on the thin filament where Myosin S1 heads can bind to actin to form XBs (Gordon et al., 2000). Upon Ca2<sup>+</sup> binding cTnC undergoes a conformational change, which exposes a hydrophobic patch within cTnC and allows cTnC to bind to the switch peptide subunit of cardiac Troponin I (cTnI). Once the cTnC has bound to the cTnI switch peptide, cTn complex releases the anchoring Tm, allowing the Tm molecule to slide around the actin filament. Tm molecules overlap, which leads to cooperative interactions between nearest-neighbor thick-thin myofilament proteins that can be affected by the stiffness of the Tm molecule. Tm moves from the Blocked state to a Closed state in which myosin binding sites are partially exposed, and then to an Open state where myosin such that actin binding sites on the thin filament are exposed, thus initiating XB cycling (Vibert et al., 1997; Gordon et al., 2000; de Tombe et al., 2010).

Mutations in regulatory thin filament sarcomeric proteins Tn and α-Tm have been associated with DCM (**Table 1**) (**Figure 1**). A clinical study of idiopathic DCM patients found mutations in the Tn complex in 7% of patients, with severe prognosis for patients with mutations in cTnC (Chang and Potter, 2005; Lu et al., 2013). cTnC D75Y/E59D missense mutation was detected in an adult male who suffered sudden cardiac death as a result of idiopathic DCM (Lim et al., 2008). D75Y and E59D are point mutations located in the low affinity Ca2<sup>+</sup> binding site on cTnC near the N-terminus. Studies in skinned and intact cardiomyocytes have reported a marked decrease in calcium sensitivity, cell shortening and force production for the double mutant (D75Y/E59D) and D75Y alone in spite of the fact that the mutations do not influence the intracellular calcium homeostasis (Lim et al., 2008; Dweck et al., 2010). However, both D75Y and E59D are required to reduce the actomyosin ATPase activity and maximal force in muscle fibers, indicating that E59D enhances the effects of D75Y (Dweck et al., 2010). In addition to D75Y and E59D, the G159D cTnC mutation has been found in human DCM patients and has been shown to impair cTnC-cTnI interaction and decrease Ca2<sup>+</sup> binding affinity (Mogensen et al., 2004; Biesiadecki et al., 2007; Robinson et al., 2007; Baryshnikova et al., 2008). G159D cTnC mutant exhibits reduced opening and closing rates of N-terminus of cTnC post-calcium binding and dissociating respectively (Dong et al., 2008). Additionally, G159D cTnC mutant abolishes the accelerated closing rate of the N-terminus of cTnC triggered by PKA mediated phosphorylation of cTnI (Dong et al., 2008). Four rare clinical variates (Y5H, M103I, D145E, and I148V) of TnnC1 have been reported in association with DCM (Hershberger et al., 2010), of which Y5H cTnC mutation was reported in a pediatric patient with idiopathic DCM concomitant with a mutation in Myosin (Rampersaud et al., 2011). Three of these (Y5H, M103I, and I148V) showed decreased calcium sensitivity of myofilaments and impaired response of the myofilament to undergo Ca2<sup>+</sup> desensitization upon PKA phosphorylation (Pinto et al., 2011). The fourth variant D145E presented with a MyBP-C rare variant. Given that D145E mutation shows increased calcium sensitivity and is associated with HCM, it is quite plausible that the concomitantly present MyBP-C mutation mediated the observed DCM response (Landstrom et al., 2008; Pinto et al., 2011). Additionally, cTnC mutant Q50R has been identified in a DCM family with a member with the rare disease of peripartum dilated cardiomyopathy (van Spaendonck-Zwarts et al., 2010).

A DCM associated mutation in cTnI, A2V, was found to hinder cTnC-cTnI interaction via mammalian two-hybrid luciferase assay (Murphy et al., 2004). Two other cTnI mutants, K36Q, and N185K, were reported in 2009 and found to decrease calcium sensitivity of actin-myosin S1 ATPase, maximal ATPase activity and reduce calcium binding affinity of cTnC (Carballo et al., 2009; Lu et al., 2013). Two more cTnI mutants P16T and D180G were reported recently (Murakami et al., 2010; Rampersaud et al., 2011).

An earlier study employing two-hybrid assays to investigate cTnT mutations R131W, R205L, and D270N, all of which have been found in human DCM patients, impaired cTnCcTnI/cTnC-cTnT interaction (Mogensen et al., 2004). A contractility study in rabbit muscle fibers reported that these mutations and TnT R141W decrease Ca2<sup>+</sup> sensitivity of force generation, maximal ATPase activity and myofilament sliding speed (Mirza et al., 2005). Additionally, Robinson et al. reported a decrease in calcium sensitivity of these cTnT mutants as well with the exception of cTnT mutation D270N. D270N cTnT led to a decrease cooperativity of Ca2<sup>+</sup> binding but not overall Ca2<sup>+</sup> affinity of cTn as measured by pCa50 in reconstituted thin filaments (Robinson et al., 2007). The 1K210 cTnT mutation has been shown to decrease Ca2<sup>+</sup> sensitivity of force generation and hinder cTnT-cTnI interaction without affecting maximum force generation (Morimoto et al., 2002; Mogensen et al., 2004; Du et al., 2007; Robinson et al., 2007). There are conflicting reports on the effect of 1K210 on cooperativity of force generation (Venkatraman et al., 2003; Du et al., 2007; Robinson et al., 2007). Four missense mutations, R134G, R151C, R159Q, and R205W, in cTnT were identified in probands with familial DCM (Hershberger et al., 2009), of which two (R134G, and R205W) were also reported in pediatric patients with familial DCM (Rampersaud et al., 2011). Functional analysis of these mutations in reconstituted myocytes showed decreased calcium sensitivity of force development (Hershberger et al., 2009). Additionally, a HCM associated mutant E244D, was identified in a DCM associated proband (Hershberger et al., 2009) and in a pediatric patient with familial DCM (Rampersaud et al., 2011). Further, a cTnT mutation R139H showed decreased calcium sensitivity and was reported in late onset DCM in a 70 year old woman (Morales et al., 2010). This is interesting as usually cTnT mutations are associated with early onset and aggressive form of DCM. Lastly, E96K cTnT mutation was reported in a 5 month patient with idiopathic DCM (Rampersaud et al., 2011).

Tm mutations E40K and E54K were identified in a GWAS, and when reconstituted in rabbit muscle fibers were found to decrease Ca2<sup>+</sup> sensitivity (Olson et al., 2001; Mirza et al., 2005). Interestingly, the E40K mutation was found to decrease myofilament sliding speed while the E54K mutation had no effect on sliding speed (Mirza et al., 2005). Both mutations caused


#### TABLE 1 | DCM associated cTn and Tm mutants and their known molecular (M) effects, cellular (C) effects, and organ level (O) effects.

*(Continued)*

#### TABLE 1 | Continued


localized destabilization of the Tm dimers and affect interactions with actin which would then directly affect ATPase activity (Chang et al., 2014). In addition, D230N Tm mutation showed decreased calcium sensitivity of myofilaments, and dissociation between calcium sensitivity and PKA mediated beta adrenergic response to TnI phosphorylation (Lakdawala et al., 2010; Memo et al., 2013). Lastly, multiple Tm mutations (K15N, I92T, A277V) have been reported in pediatric cases with idiopathic or familial DCM (Rampersaud et al., 2011).

The above-mentioned familial mutations in the regulatory proteins cTn and α-Tm, that have been identified in human patients over the course of years, display some wide-ranging molecular effects that mainly converge to few cellular mechanisms such as altered calcium sensitivity and decreased ATPase activity ultimately leading to depressed contractile force observed in DCM. A recent innovative motility assay study proposed that the varying experimental findings of DCM-associated thin filament mutations in α-Tm and cTn can be explained by a decoupling of Ca2<sup>+</sup> sensitivity from cTnI phosphorylation by PKA (Memo et al., 2013). This finding, along with the wide variety of mutations in thin filament proteins that have been connected to the development of DCM, underscores the urgency of studying the complicated cascade of contraction events as a whole in order to create a cohesive picture of DCM causes and progression. Genetically engineered animal models provide an opportunity to understand the sequelae of molecular and cellular events as they translate to the whole heart level. Notably, genetically engineered mice with sarcomeric mutations 1K210 (Du et al., 2007; Ahmad et al., 2008) and R141W (Ramratnam et al., 2016) in cTnT have been generated. These studies showed a gene-dosage effect on the cardiac phenotype and recapitulated the human phenotype. Despite the key genotype-to-phenotype insights gleaned from animal model studies, most experimental studies have so far been conducted in reconstituted in vitro systems. Majority of experimental studies have measured the effects of thin filament protein mutations on cell-level function, but have not looked more closely into molecular mechanisms or expression of dysfunction on the whole heart level. Multi-scale computational modeling offers a complementary set of tools that can help us understand the spatial and temporal transition to clinical DCM.

### INSIGHTS AND CHALLENGES FROM MOLECULAR MODELING OF REGULATORY THIN FILAMENT PROTEIN MUTATIONS

Structural determination from x-ray crystallography and NMR studies has helped provide key insights into molecular basis of cTnC function (Li and Hwang, 2015). Previously published experimental and theoretical studies have used these structural data to probe rapid, nanosecond and microsecond timescale conformational dynamics, by using Brownian Dynamics (BD) and Molecular Dynamics (MD) simulations, that are correlated with calcium binding (Kekenes-Huskey et al., 2012; Kalyva et al., 2014). MD and BD simulations are useful in-silico approaches that have been employed to understand the molecular basis of altered structural and functional dynamics of regulatory myofilament proteins such as cTnC in varied states. For example, Varughese and Li investigated, with MD, changes in the structural dynamics of cardiac Tn, including TnC, upon binding bepridil, a known inotropic agent (Varguhese and Li, 2011; Lindert et al., 2012b) combined long time-scale MD simulations and BD simulations to understand the dynamics of wild-type TnC in its

calcium-free, calcium-bound, and TnI-bound states, as well as V44Q (Lindert et al., 2012a).

Intra-molecular dynamic changes in cTn can cause alterations in: (a) the calcium binding affinity of cTnC; (b) the rate of calcium dissociation from cTnC; (c) the forward rate of cTnC conformation transition; (d) the reverse rate of cTnC conformation transition; and (e) the structure of cTnC such that there are differences in charge within the exposed hydrophobic patch (Dewan et al., 2016). Additionally, cTn mutations can affect interactions between cTnC, cTnI, and cTnT as well as downstream contractile proteins such as Tm and actin (Mogensen et al., 2004; Robinson et al., 2007). A recent in silico study (Dewan et al., 2016) employed MD and BD simulations to postulate the decreased calcium binding affinity of cTnC and the altered rate of hydrophobic patch opening within cTnC as the molecular basis of the reported changes in calcium sensitivity and force production in the D75Y cTnC DCM mutant. An MD study on residues 70–110 of cTnT found that familial HCM linked mutations R92L and R92W, located near the Tm binding domain, lead to increased hinge movement downstream on the cTnT molecule as well as decreased helical stability (Ertz-Berger et al., 2005). Importantly, the study found that divergent phenotypes emerge in a live mouse model of these mutations, which demonstrates a limitation to modeling isolated proteins (Ertz-Berger et al., 2005).

MD simulations have the potential to uncover biophysical effects of DCM associated mutations in cardiac Tm. Recent MD studies have explored the properties of Tm in healthy cases as well as patho-physiological cases, noting that Tm stiffness may impact downstream contractility events greatly (Li et al., 2010a,b; Loong et al., 2012; Lehman et al., 2015). A 2012 study used MD to study familial HCM associated Tm mutations D175N and E180G and found that both mutations lead to increased flexibility of Tm and therefore, decreased persistence length of the molecule (Li et al., 2012). Another MD study has also been performed on HCM Tm mutations E62Q, A63V, K70T, V95A, D175N, E180G, L185R, E192K in order to explore the effects of point mutations on Tm flexibility and Tm-actin interactions (Zheng et al., 2016). In the case of DCM, a time-independent electrostatic snapshot of the DCM associated Tm mutation showed that E54K and E40K mutations alter the surface charge of Tm, which may affect Tm-actin interaction (Olson et al., 2001; Chang et al., 2014). A time-dependent MD study was also performed on the E54K Tm mutant in combination with 7 actin monomers and showed that this mutation causes increased stiffness and decreased curvature in the Tm molecule overall while stabilizing and destabilizing the coiled coil structure in different regions of the molecule and greatly weakening Tm-actin binding (Zheng et al., 2016). More MD studies are needed in the area of Tm DCM mutations in order to visualize effects of other mutations, such as E40K, that may affect Tm-actin and Tm-cTn interactions.

In-silico investigation of sarcomere protein structure-function dynamics by MD and BD simulations provides key insights toward understanding molecular effects of genetic mutations and post-translational modifications (Lindert et al., 2015; Dewan et al., 2016). However, these structure-function dynamics can change significantly when myofilament proteins interact with one another in an integrated physiological system. It has been reported that kinetic rates of a given state transition vary in isolated molecular states and integrated myofilament states (Davis and Tikunova, 2008). This was well demonstrated in a previous study where the off-rates of calcium binding were studied in depth from isolated cTnC molecule to a structurally integrated myofilament preparation (Davis and Tikunova, 2008). The whole cTn complex has been modeled using MD, both with and without Ca2<sup>+</sup> bound (Varughese et al., 2010; Jayasundar et al., 2014). A 2014 study proved the value of modeling the complex as a whole, because removing Ca2<sup>+</sup> from the regulatory binding pocket on cTnC affected cTnC hydrophobic patch opening as well as the folding and flexibility of the cTnI switch region (Jayasundar et al., 2014). Moving further up in scale, an integrated MD study of a whole thin filament including cTn, 14 actin monomers, and two overlapping Tm molecules illustrates the importance of modeling inter-protein interactions along the thin filament (Manning et al., 2011). The MD simulation performed was only 1 ns in length, but was able to accurately capture the rotation of the I-T arm in the Tn complex as a direct consequence of Ca2<sup>+</sup> binding (Manning et al., 2011). The model was used to study cTnT mutations R92W and R92L associated with familial HCM and found that both mutations decrease bending forces in the hinge region of cTnT which affects Tm interaction, a downstream effect that may not have been captured if the study were performed on isolated cTnT (Manning et al., 2012).

Integration of molecular level state-transition kinetic data into the sarcomere level is a key challenge yet to be solved for using insilico methods. At the molecular level it takes about 5–10 µs for calcium to bind to cTnC (Lindert et al., 2012a,b). With the stateof-the-art supercomputers we are now able to simulate molecular events such as calcium-binding events and conformational statetransitions within cTnC (a relatively small protein–18 kDa), events that occur in the microseconds range. At the myofilament level it takes about 700 ms for a contraction-relaxation cycle to take place in a sarcomere (Kerckhoffs et al., 2010; Tøndel et al., 2015). Recording the kinetics of all inter-molecular and intra-molecular state transitions for every myofilament protein during one contraction-relaxation cycle in a sarcomere is a tremendous task, as illustrated by the 1 ns upper limit of an MD simulation incorporating the full thin filament (Manning et al., 2011). This is due to the enormous computational power and speed that would be needed to solve for longer time scale simulations and bulky proteins, such as Titin (3.9 MDa), Myosin (220 kDa), and Tropomyosin (37 kDa), that form the backbone of the sarcomeric thin and thick myofilament protein complexes. Nevertheless, key insights into molecular behavior of bulky proteins like Titin that are commonly known to be mutated in DCM have been achieved by MD simulations (Herman et al., 2012). An earlier MD study of Titin wherein, single Ig domains of Titin were stretched reported sequential unfolding of Ig domains corroborating experiments (Lu et al., 1998, 2000; Gao et al., 2002). A recent study of Titin examined the hydrophobic core region of the protein associated with a DCM mutation V54M and reported destabilization of transition from bend to coil in secondary structure of Titin and reduced affinity to Z-disc protein T-cap/telethonin (Thirumal Kumar et al., 2016). Given that Titin mutations are commonly associated with DCM, molecular modeling of Titin domains associated with DCM is mandated. While we have much to gain from MD simulation studies, gaps in our experimental knowledge toward understanding dynamics of molecular level interactions and kinetics of state transitions between thick and thin myofilament proteins compound the challenges in standardizing conditions and validating results from the molecular simulation studies.

### INSIGHTS AND CHALLENGES FROM IN-SILICO TRANSLATION OF MOLECULAR LEVEL CHANGES TO THIN FILAMENT MECHANICS

The mathematical formulation of cardiac myofilament models that explicitly incorporate spatio-temporal acto-myosin interactions and stochastic XB formation to compute contractile force have lagged behind electrophysiological models of the heart (Rice and de Tombe, 2004; Zhang et al., 2016). This is largely due to the (a) paucity of kinetic data on thick-thin myofilament interactions and molecular state-transitions, (b) requirement of partial-differential equations (PDEs) to solve for explicit spatio-temporal acto-myosin interactions, (c) lack of complete understanding of translation of steady-state contractile force into a length and load-dependent dynamic contractile force response via XB cycling, (d) difficulty in solving for computationally expensive stochastic interactions, (e) partial understanding of cooperative mechanisms involved in myofilament activation, and (f) technical gaps in our knowledge due to species differences and varied experimental conditions in scientific studies. Nonetheless, ordinary differential equation (ODE) and Monte Carlo Markov models of regulated co-operative myofilament activation with nearest neighbor interactions, wherein some molecular states are lumped together empirically and model parameters are optimized such that the best-fit of the base model to the experimentally measured steady-state force-calcium data-sets is achieved, have been formulated (Noble, 2002; Rice et al., 2003, 2008; Puglisi et al., 2004; Rice and de Tombe, 2004; Hussan et al., 2006; Campbell et al., 2010; Aboelkassem et al., 2015; Sewanan et al., 2016). These relatively simplified mean-field Markov models of cooperative myofilament activation and contraction have helped provide deeper insights as to how changes in inter-molecular and intra-molecular interactions in myofilament proteins can alter steady-state myofilament properties (myofilament calcium sensitivity and cooperativity of myofilament activation) and function (maximal force generation), which can then translate to cardiac pump dysfunction as reported in DCM.

Arguably the most logical type of sarcomere-level model to begin with, when testing the effects of point mutations on thin filament proteins, is a Markov model of thin filament activation. One such model, originally developed in 2010 and since expanded for a variety of applications, consists of 26 spatially explicit regulatory units (RUs), each RU including 7 actin monomers, one Tm molecule, and cTn (Campbell et al., 2010). The model captures cooperativity of thin filament activation by relying on the Tm position (blocked, closed, or open) of each RU's nearest neighbor in order to determine rates governing state transitions. Using this model as a starting point, further studies have been able to test a variety of contractility protein mutations at a larger spatial and temporal scale than MD simulation can reasonably accomplish. A relevant example is a recent study that simulated Tm mutations E180G and D175N, which have been found in HCM patients, using a Monte Carlo framework extension of the Campbell model (Sewanan et al., 2016). MD simulations indicated that both Tm mutation increase Tm flexibility while lowering persistence length of the molecule (Li et al., 2012). The thin filament model was able to simulate experimentally gathered wildtype and mutant contraction data by altering Tm persistence length, transition rates between the blocked and closed Tm states, and percentage of the XB cycle spent in the attached force-producing state (Sewanan et al., 2016). MD simulation data of isolated Tm can offer information on persistence length but not on Tm-actin interaction or XB cycling, which were discovered to be possible downstream effects of these mutations.

A previously published multi-scale modeling study from our group was one of the first studies to directly incorporate the molecular changes computed from BD and MD simulations, in the DCM mutant D75YcTnC, into a six-state Markov model of steady-state myofilament contraction (Dewan et al., 2016) (**Figure 2**). The key results from the study reported that the intra-molecular changes (decreased calcium binding affinity of cTnC and depressed rate of hydrophobic patch opening during cTnC conformational change) in D75Y cTnC mutant are sufficient to explain the observed decrease in myofilament calcium sensitivity under steady-state experimental conditions in skinned cardiomyocytes. Additionally, the study highlighted how DCM mutants like E59D cTnC which appear to have healthy myofilament response in skinned myofilament experiments, can in fact be harboring molecular modifications which can potentially turn deleterious under stressful conditions for the myocardium. A weakness of this study was that they did not model the effects of the double mutant D75Y/E59D. This would have been more appropriate as both mutations, D75Y, and E59D are needed for a reduction in maximum myofibrillar ATPase and maximum force of contraction in skinned fibers. Additionally, The D75Y mutant is important for the effect seen in calcium binding affinity, however the "cellular" phenotype may not be explained solely by a single mutation based on the in vitro data. This study was instrumental in its scope, as it bridged the genotypic defects at sarcomeric protein level to myofilament phenotype in silico, by directly incorporating molecular parameters from BD and MD simulations into appropriate markov states of cTnC activation. The study again displays the numerous challenges for the mathematical modeler in scaling from molecular level to myofilament levels to compute steady-state contractile force, as MD data was insufficient to fully capture cellular mutant contractility changes which may be due to cTnC-cTnI interaction or other protein interactions in the sarcomere as shown in case of G159D cTnC mutant. It is important to note that prediction of steady-state force-calcium data is not predictive of cellular phenotype by itself. For example, a reduction in calcium sensitivity of myofilaments can be compensated for by an increase in calcium transient. In such a case there will be no apparent change in contractile function at the cellular level. Similarly, an increase in calcium sensitivity of myofilaments can be offset by defects in mechano-sensing. In fact these compensatory changes are integral to the process of cardiac remodeling leading to DCM. Further, as is evident in many DCM mutants (see **Table 1**), change in calcium sensitivity is not the only predisposing factor in DCM and the acto-myosin ATPase activity should also be modeled in. Therefore, it is important to factor in the contractile phenotype from an intact cell and incorporate dynamic coupling in a model between thin-thick myofilaments, passive tension, calcium homeostasis, and mechanotransduction, at the least to be considered a cellular level model.

Given the gaps in our knowledge of kinetic rates during molecular transitions and variations in experimental data collated from numerous studies, the mathematical modeler is required to carefully choose explicit Markov states within an appropriate framework, and to standardize technical variates of species, temperature, and molecular state in order to glean meaningful insights from any multi-scale computational study. It is known that various rodent species (mouse, rat, guinea pig) express variable isoforms of key myofilament proteins under normal conditions, for e.g., mouse and rat express αmyosin heavy chain (MHC) isoform (Rundell et al., 2005), whereas, guinea pigs express β-MHC under basal conditions (van der Velden et al., 1998). Given that α-MHC is at least three times faster than β-MHC, this difference in basal conditions in rodents is sufficient to alter the rate of XB

cycling significantly (Rundell et al., 2005). Isoform switches in myofilament proteins are known to occur in pathophysiological conditions as well (Nakao et al., 1997). Similar changes in protein isoform expression and kinetics have been reported for other myofilament proteins (de Tombe and Solaro, 2000). These changes impact the myofilament properties of calcium sensitivity, cooperativity and maximal force output significantly. Additionally, increased temperature is known to have a significant effect on increasing calcium sensitivity of myofilaments and increasing the maximal developed contractile force under steady-state conditions in intact and skinned ventricular tissue/cells from various species (Harrison and Bers, 1989, 1990; de Tombe and Stienen, 2007). This is likely due to alterations in molecular kinetics of state transitions of myofilament proteins with temperature. Also, kinetics of statetransitions can vary significantly in isolated molecular states vs. integrated systems, as elucidated in an earlier study where the off-rates of calcium binding were studied in depth from isolated cTnC molecule to a structurally integrated myofilament preparation (Davis and Tikunova, 2008). These biophysical variates must be accounted for during longitudinal scaling from molecular states to myofilament states, as failing to do so can potentially exaggerate or mask the magnitude of, and can even potentially alter the direction of, steady-state myofilament properties as reported for a mutation. It is imperative to note that this is not always possible due to paucity of experimental data, still, careful consideration must be given to these variates in order to arrive at meaningful analysis from in-silico studies and to achieve good quantitative agreement with experimental data.

### INSIGHTS AND CHALLENGES WHILE SCALING FROM THIN FILAMENT MECHANICS TO THICK FILAMENT MECHANICS

Results of Markov models of cooperative myofilament activation with nearest neighbor interactions just described suggest that a meaningful prediction toward patho-physiological outcome of genetic mutations is possible in-silico (Dewan et al., 2016; Sewanan et al., 2016). While this, of course, does not mean that the models are entirely correct in their predictions, as the whole system is not explicitly represented; it does suggest that the key myofilament properties are adequately represented to recapitulate fairly complex phenomena. These Markov models adequately describe the process of thin filament activation, such that the myofilament contractile response is mediated by a prescribed length-clamp and/or calcium-clamp at any given point of time. However, under physiological conditions, myofilament activation and XB cycling are temporally modulated by cyclical variation in length, calcium, and loading conditions during a contraction-relaxation cycle (ter Keurs et al., 1988). Modeling thick-thin myofilament effects into a dynamic model of contraction, wherein length-dependence and load-dependence of the myofilaments have been incorporated, will allow us to further gain quantitative insights in the shortening response of the contractile machinery and the work done by myofilaments. This is imperative to scaling the effects of the identified DCM mutant from genotype to phenotype.

The length-dependence of myofilament activation and the load-dependence of the rate of myofilament shortening, are tightly regulated and complex properties of the healthy myocardium that are highly dependent on the spatio-temporal arrangement of the myofilaments. In a simplistic view of our understanding of cardiac biomechanics, length dependence of myofilament activation is modulated by the degree of overlap between thick and thin filaments and calcium sensitivity of myofilaments (Huxley and Hanson, 1954; Huxley and Niedergerke, 1954; Dobesh et al., 2002; de Tombe et al., 2010). On the other hand, load-dependence of rate of myofilament shortening is modulated by number of XBs in the strongly bound state, distance between the thick and thin filaments, myosin isoform, degree of thick-thin filament cooperativity, rate of XB cycling, and stiffness of XBs (Edman, 1979; Hunter et al., 1998; McDonald et al., 1998; Herron et al., 2001; Janssen et al., 2002; Konhilas et al., 2002). Ultimately, the twitch dynamics of the myofilaments are the key determinants of the cardiac pump output during various phases of the cardiac cycle (Moss and Buck, 2011). In reality, we are yet to fully comprehend the molecular basis of these dynamic emergent myofilament properties and their modulators. An explicit PDE model that simulates the myofilament contraction-relaxation cycle while recapitulating length-dependent myofilament activation and force-velocity relationships is yet to be described to the best of our knowledge. However, phenomenological models based on empirical relationships have been described to quantitatively recapitulate these myofilament properties (Hunter et al., 1998; Lumens et al., 2009; Kerckhoffs et al., 2010). Phenomenological vs. explicit Huxley-type muscle models of XB cycling have been discussed before (Winters and Stark, 1987).

Computational models of XB cycling dynamics are key toward scaling-up and integrating length dependence and dynamic muscle activation into studies of thin filament DCM proteins. For example, the Rice ODE model of XB cycling accurately mimics length dependence of active/passive forces and Ca2<sup>+</sup> binding, while simplifying a complex myofilament geometry by keeping track of the overlap fraction between thick and thin filaments (Rice et al., 2008). This model has been expanded in more recent studies to include more complex geometries and metabolic intermediates, and has been incorporated into organscale finite element models through solving the sarcomere model at different points in a finite element mesh (Gurev et al., 2010; Tran et al., 2010; Sugiura et al., 2012; Washio et al., 2012). Although, the model traditionally calculates force from a half sarcomere and extrapolates to cell-level force values, it may be possible to explicitly model a full cell with 32 sarcomeres using a Blue Gene supercomputer in order to capture cooperativity of the whole cell (Hussan et al., 2006). Another XB modeling approach more explicitly models the sarcomere geometry by including 3 half thick filaments surrounded hexagonally by 13 half thin filaments, mathematically modeled by a 3D spring array using finite element analysis and governed by Monte Carlo processes (Chase et al., 2004). This model importantly includes filament compliance in its calculations, creating a more biophysically accurate window into the individual proteins making up the myofilament. The Chase model has been used to study specific familial HCM-associated cTnI mutations by converting experimentally gathered Ca2<sup>+</sup> transient information into altered probability of XB activation (Kataoka et al., 2007). However, this study assumes that altered Ca2<sup>+</sup> binding kinetics are the sole contributor to altered contractility without considering the interaction of mutated cTnI with other thin filament proteins. When searching for XB models that will be able to accurately model DCM sarcomere mutations, mechanistic models are more promising as they allow for direct input of protein-level changes into a larger scale model.

Phenomenological computational models of twitch tension, which do not explicitly translate the molecular behavior of myofilament proteins to the myofilament properties, can still provide invaluable insight in the workings of the myocardium. Therefore, it is critical to underscore the key considerations and challenges to take into account when formulating dynamic models of cardiac twitch. Temperature dependence of myofilament dynamics, loading conditions (preload and afterload), frequency of electrical pacing which determines the calcium load, and biological species, are biophysical experimental variates that significantly alter twitch dynamics (Stull et al., 2002). Previous studies have reported frequency and temperature dependent changes in contraction-relaxation dynamics and maximal twitch force, based on phase-plane analysis of dynamic force-calcium relationships from intact rat myocardium (Janssen et al., 2002). Briefly, increasing frequency of stimulation resulted in a increased developed force, enhanced myofilament responsiveness to calcium and abbreviated twitch duration; increasing temperature from (22 to 37◦C), while maintaining stimulation frequency and preload, resulted in biphasic effects - between 22 and 30◦C, developed force did not decrease despite abbreviated twitch duration; between 30 and 37◦C, a steep decline in developed twitch force was observed. This would suggest independent temperature dependence on myofilament activation and relaxation kinetics. An earlier study elegantly measured the relationship between cardiac work output and external load in single rat cardiomyocytes (Nishimura et al., 2004). They reported work output of cardiomyocytes under a range of loading conditions ranging from isometric, unloaded and physiologically loaded conditions. Alterations in loading conditions affect XB cycling rates and force-velocity relations. Factoring in the effect of loading conditions, imposed on the myocyte during experimental study from which parameters are being collated, is crucial to correct interpretation of modeling results and for scaling the results to the organ level. Lastly, different experimental studies report data from different biological species, each of which expresses different myofilament

protein isoforms that can independently alter twitch dynamics (Clark et al., 1982; Tøndel et al., 2015). It is important that careful consideration be given to these variables as models are parameterized.

### SUMMARY, PERSPECTIVES AND CONCLUSIONS

This review is focused toward demonstrating the feasibility and applicability of multi-scale modeling to gain mechanistic insights into role of known genetic DCM mutations in contractile proteins (**Figure 3**). We discuss some of the insights that can be gained from computational models of cardiac biomechanics when scaling from molecular states to cellular level states. We have described the considerations and challenges that need to be accounted for, when computationally modeling the mechanical effects of sarcomeric mutations of DCM from genotype to cellular phenotype. In particular, we discussed the difficulties in standardizing parameters for a multi-scale modeling study from varied experimental data sets. An important aspect that requires careful consideration and work in modeling studies is coupling

of calcium homeostasis and mechano-transduction mechanisms to myofilament mechanics. Sarcomere generated force cannot predict the organ level phenotype in entirety if upstream and downstream regulatory mechanisms in the cell are not coupled to it. In order to comprehensively translate to the cellular scale, coupling of models of active tension to models of (a) passive tension mediated by Titin, (b) upstream calcium homeostasis mechanisms mediating CICR, (c) cytoskeletal proteins and zdisc proteins that sense and transmit force signals, (d) ATP, ROS and energy regulation by mitochondria, (e) insulin signaling, (f) beta-adrenergic signaling pathways, and (g) signaling networks mediating nuclear transcription response toward sarcomere addition should be considered. However, with added complexity in models the degree of ambiguity in results also increases which may prove futile. It is important to note that DCM is a disease characterized by an organ level phenotype, so in order to understand the complex cascade that can cause a point mutation in a contractile protein to lead to growth and remodeling of the heart as a whole it is necessary to scale current molecular and cellular level models to finite element models of the heart with realistic geometries. For more information on whole heart modeling and its potential, detailed reviews and modeling frameworks exist (Campbell and McCulloch, 2011; Trayanova, 2011; Sugiura et al., 2012; Zhang et al., 2016). Of note is the recent study reporting phenomenological modeling of the tension integral of the cardiomyocyte twitch as the predictor for DCM and HCM phenotype in cardiomyopathies (Davis et al., 2016). Furthermore, given that familial DCM manifests clinically over-time it is pertinent to simulate pathological cardiac remodeling leading to DCM. Growth modeling frameworks

### REFERENCES


utilizing stress or strain as the growth stimulus for sarcomere addition at the whole-heart level coupled with cardiac mechanics have been recently reviewed and present with a promising avenue to scale-up (Göktepe et al., 2010; Kerckhoffs et al., 2012). In addition, mechanistic frameworks for modeling biochemical signaling networks that have previously been formulated (Zeigler et al., 2016), such as for PKA mediated beta-adrenergic signaling (Ryall et al., 2012), will be useful tools as cardiac signaling in DCM is being studied based on RNAseq and phosphoprotemics (Burke et al., 2016; Kuzmanov et al., 2016). Formulating explicit models of molecular behavior and scaling them longitudinally to cellular levels is an extremely challenging task and perhaps not without its own pitfalls. Nonetheless, in-silico modeling of cardiac biomechanics, in parallel with the experimental studies, is a powerful set of tools that can be applied toward integrative and quantitative understanding of DCM.

### AUTHOR CONTRIBUTIONS

SD took a lead on the project, conceptualized the idea and wrote the article. KM contributed toward conceptualizing and writing the article. MR and AM supervised the project.

### ACKNOWLEDGMENTS

This work was supported by the NIH through the National Biomedical Computation Resource (P41 GM103426, Amaro and AM) and NHLBI U01 grant HL122199 (Bassingthwaighte and AM).

cooperative activation in a markov model of the cardiac thin filament. Biophys. J. 98, 2254–2264. doi: 10.1016/j.bpj.2010.02.010


hypertrophic versus dilated cardiomyopathy. Cell 165, 1147–1159. doi: 10.1016/j.cell.2016.04.002


**Disclosure:** AM is a co-founder, scientific advisor and equity-holders of Insilicomed, Inc., a licensee of UC San Diego software that was not used in this research. Insilicomed, Inc. had no involvement at all in design, performance, analysis or funding of the present study. This relationship has been disclosed to, reviewed, and approved by the University of California San Diego in accordance with its conflict of interest policies. The other authors have no relationships to disclose.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer JRP and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2017 Dewan, McCabe, Regnier and McCulloch. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Hypertrophic Cardiomyopathy Cardiac Troponin C Mutations Differentially Affect Slow Skeletal and Cardiac Muscle Regulation

Tiago Veltri <sup>1</sup> , Maicon Landim-Vieira<sup>1</sup> , Michelle S. Parvatiyar <sup>2</sup> , David Gonzalez-Martinez <sup>1</sup> , Karissa M. Dieseldorff Jones <sup>1</sup> , Clara A. Michell <sup>1</sup> , David Dweck <sup>1</sup> , Andrew P. Landstrom<sup>3</sup> , P. Bryant Chase<sup>4</sup> and Jose R. Pinto<sup>1</sup> \*

<sup>1</sup> Department of Biomedical Sciences, Florida State University College of Medicine, Tallahassee, FL, USA, <sup>2</sup> Department of Molecular and Cellular Pharmacology, University of Miami Miller School of Medicine, Miami, FL, USA, <sup>3</sup> Section of Pediatric Cardiology, Department of Pediatrics, Baylor College of Medicine, Houston, TX, USA, <sup>4</sup> Department of Biological Science, Florida State University, Tallahassee, FL, USA

#### Edited by:

Coen Ottenheijm, VU University Medical Center, Netherlands

#### Reviewed by:

Lori A. Walker, University of Colorado Denver, USA Diederik Wouter Dimitri Kuster, VU University Medical Center, Netherlands

> \*Correspondence: José R. Pinto jose.pinto@med.fsu.edu

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 20 January 2017 Accepted: 27 March 2017 Published: 20 April 2017

#### Citation:

Veltri T, Landim-Vieira M, Parvatiyar MS, Gonzalez-Martinez D, Dieseldorff Jones KM, Michell CA, Dweck D, Landstrom AP, Chase PB and Pinto JR (2017) Hypertrophic Cardiomyopathy Cardiac Troponin C Mutations Differentially Affect Slow Skeletal and Cardiac Muscle Regulation. Front. Physiol. 8:221. doi: 10.3389/fphys.2017.00221 Mutations in TNNC1—the gene encoding cardiac troponin C (cTnC)—that have been associated with hypertrophic cardiomyopathy (HCM) and cardiac dysfunction may also affect Ca2+-regulation and function of slow skeletal muscle since the same gene is expressed in both cardiac and slow skeletal muscle. Therefore, we reconstituted rabbit soleus fibers and bovine masseter myofibrils with mutant cTnCs (A8V, C84Y, E134D, and D145E) associated with HCM to investigate their effects on contractile force and ATPase rates, respectively. Previously, we showed that these HCM cTnC mutants, except for E134D, increased the Ca2<sup>+</sup> sensitivity of force development in cardiac preparations. In the current study, an increase in Ca2<sup>+</sup> sensitivity of isometric force was only observed for the C84Y mutant when reconstituted in soleus fibers. Incorporation of cTnC C84Y in bovine masseter myofibrils reduced the ATPase activity at saturating [Ca2+], whereas, incorporation of cTnC D145E increased the ATPase activity at inhibiting and saturating [Ca2+]. We also tested whether reconstitution of cardiac fibers with troponin complexes containing the cTnC mutants and slow skeletal troponin I (ssTnI) could emulate the slow skeletal functional phenotype. Reconstitution of cardiac fibers with troponin complexes containing ssTnI attenuated the Ca2<sup>+</sup> sensitization of isometric force when cTnC A8V and D145E were present; however, it was enhanced for C84Y. In summary, although the A8V and D145E mutants are present in both muscle types, their functional phenotype is more prominent in cardiac muscle than in slow skeletal muscle, which has implications for the protein-protein interactions within the troponin complex. The C84Y mutant warrants further investigation since it drastically alters the properties of both muscle types and may account for the earlier clinical onset in the proband.

Keywords: cardiac troponin C, hypertrophic cardiomyopathy, slow skeletal muscle, skinned fibers, myofibrillar ATPase

### INTRODUCTION

Hypertrophic cardiomyopathy (HCM) is a cardiac disease with relatively high prevalence (1:200) in the general population, which promotes morphological changes of the heart such as left ventricular thickening (Semsarian et al., 2015). The first case of HCM was described in a patient from the 1950's (Teare, 1958) in which the diagnosis of HCM development was solely based on electrocardiogram analyses associated with angiographic/hemodynamic studies. After 1990, the development of DNA-based sequencing methodologies provided a valuable resource which accelerated a new era of HCM diagnosis with the discovery of the first mutation in the ß-myosin heavy chain gene, MYH7 (Geisterfer-Lowrance et al., 1990). Currently, it is widely accepted that mutations in sarcomeric genes lead to intracellular alterations that manifest as cardiac remodeling and pathologic hypertrophy (Force et al., 2010; Marian, 2010; Harvey and Leinwand, 2011; Seidman and Seidman, 2011; Maron and Maron, 2013).

After two decades of extensive genetic investigation and over 1,400 mutations identified, many genes encoding thin and thick filament, and cytoskeletal proteins have been established as the etiological agents of HCM (Maron et al., 2012). The most common mutations are located in thick-filament encoding genes (MYH7 and MBPC3), representing ∼50–70% of all HCM patients that exhibit positive genetic test results. Meanwhile, mutations in genes that encode thin-filament proteins, such as the troponin (Tn) complex, have also been shown to occur with relatively high incidence compared to other canonical sarcomeric HCM genes (Willott et al., 2010; Brouwer et al., 2011; Maron et al., 2012). HCM-associated mutations lead to several alterations in cardiac contraction e.g., changes in myosin ATPase activity, actomyosin crossbridge interaction, cooperativity of muscle activation, Ca2<sup>+</sup> binding to the thin filament, and a host of other HCM predisposing factors (Willott et al., 2010). In general, mutations in the regulatory proteins (tropomyosin and Tn) that are linked to HCM are associated with an increase in myofilament Ca2<sup>+</sup> sensitivity; thus promoting changes in Ca2<sup>+</sup> homeostasis and triggering development of cardiac dysfunction and arrhythmias seen in HCM patients (Willott et al., 2010; Brouwer et al., 2011; Landstrom and Ackerman, 2012).

Limited studies of HCM patients' skeletal muscle have identified abnormalities in electromyographic (EMG) and/or histologic analyses (Hootsmans and Meerschwam, 1971; Smith et al., 1976; Lochner et al., 1981; Przybojewski et al., 1981). In 1989, Caforio et al. (1989) showed that skeletal muscle dysfunction occurred in 37% of HCM patients analyzed and found that this dysfunction arose from myogenic origins. This finding was independently verified in skeletal muscle biopsies. The authors also observed selective atrophy of type I muscle fibers in these patients (Caforio et al., 1989). Moreover, a recent clinical study which examined 46 HCM patients by EMG showed evidence of subclinical skeletal muscle myopathy in 13 of the patients (28%) (Karandreas et al., 2000).

Many genes for cardiac sarcomeric proteins are also expressed in some skeletal muscles; these include ß-myosin heavy chain (MYH7), regulatory myosin light chain-2 (MYL2), desmin (DES), and troponin C (TNNC1) (Li et al., 1989; Matsuoka et al., 1989; Macera et al., 1992; Song et al., 1996). Mutations in the MYH7 gene can produce simultaneous abnormalities in cardiac and slow-twitch skeletal muscle function that can culminate as cardiomyopathy and/or myopathy (Hedera et al., 2003; Tajsharghi et al., 2003, 2007; Meredith et al., 2004; Mastaglia et al., 2005; Lamont et al., 2006, 2014; Darin et al., 2007; Overeem et al., 2007; Homayoun et al., 2011). Similarly, the MYL2 gene is expressed in both skeletal and cardiac muscle; mutations in myosin light chain appear to affect the function and morphology of both tissue types (Weterman et al., 2013). Desmin, a sarcomeric protein and a primary component of most intermediate filaments, is also a target of muscle disease. Desmin is expressed in cardiac, skeletal, and smooth muscle, and many studies have shown that mutations in the human DES gene promote adult-onset skeletal myopathy and, depending on which mutation, one of the three types of familial cardiomyopathies, i.e., dilated, hypertrophic, and restrictive (Goldfarb and Dalakas, 2009).

Troponin C (TnC) is a key regulatory protein in striated muscle contraction where its function is to bind Ca2+, which subsequently triggers actomyosin interactions and initiates muscle contraction (Farah and Reinach, 1995). The TNNC1 gene expresses TnC in both cardiac and slow skeletal muscle (Song et al., 1996). Seven mutations in TNNC1 (A8V, L29Q, A31S, C84Y, E134D, D145E, and Q122AfsX30) have been identified to date in HCM and restrictive cardiomyopathy patients (Hoffmann et al., 2001; Landstrom et al., 2008; Chung et al., 2011; Parvatiyar et al., 2012; Jaafar et al., 2015; Ploski et al., 2016). The A8V mutation was genetically knocked in and led to HCM in mice (Martins et al., 2015). Unfortunately, no clinical data regarding skeletal muscle abnormalities is available for these patients. With the exception of the E134D mutant, in vitro studies indicate that the above mentioned cTnC mutants alter contractile parameters known to underlie the development of diastolic dysfunction in these patients (Landstrom et al., 2008; Pinto et al., 2009, 2011a; Albury et al., 2012; Parvatiyar et al., 2012; Zot et al., 2016). Despite their extensive characterization in cardiac muscle, however, there are no reports on the effects of these HCM-associated TNNC1 mutations on slow skeletal muscle regulation and function.

The purpose of these in vitro studies was to identify potential pathogenic alterations in slow skeletal muscle regulation arising from the cTnC mutants. Therefore, we evaluated the effects of cTnC A8V, C84Y, E134D, and D145E mutants associated with HCM on the Ca2<sup>+</sup> sensitivity and maximal force when reconstituted into skinned fibers from rabbit soleus, and maximal and minimal ATPase activities when reconstituted into bovine masseter muscle myofibrils. These muscle types were chosen for this study because they are type I fibers that only express myosin heavy chain I (MHC I) and show sufficient resilience after isolation that is suitable for each type of assay presented. Human TnC was utilized to accentuate translational value of the study. The primary sequence of cTnC is the same among human, porcine and bovine homologs, and rabbit cTnC differs by only one amino acid. Other Tn subunits including slow skeletal troponin I (ssTnI) homologs are largely similar (close to 99% homologous) among human, bovine, porcine, and rabbit. However, it should be noted that rabbit slow skeletal troponin T (ssTnT) isoforms are less homologous to human, porcine, and bovine homologs. Utilization of proteins from different but overall largely homologous species has been widely accepted, although small functional differences cannot be discounted. Overall, we found that these cTnC mutants induce mutationdependent effects in slow skeletal muscle that were distinct from their effects in cardiac muscle; at least part of that distinction was found to depend on the TnI isoform. Our results are consistent with findings from other major muscle proteins encoded by the same gene in both cardiac and skeletal muscles. The data presented in this manuscript will be discussed in the context of the patients' clinical presentation.

### MATERIALS AND METHODS

### SDS-PAGE for Myosin Heavy Chain (MHC) Separation

Skeletal muscle MHC composition was analyzed by glycerolcontaining SDS-PAGE. Eight percent acrylamide gels were prepared and run according to Talmadge and Roy (1993). Protein concentration was measured using the PierceTMCoomassie Plus (Bradford) Assay Kit (Bradford, 1976). Myofibrils and skeletal myosin (0.5–1 µg) were boiled in sample buffer (Laemmli, 1970) for 2 min and loaded into glycerol-containing gels. 2 mercaptoethanol was added to the upper electrode buffer at a final concentration of 10 mM. Gels were run at constant voltage (75 V) in a Bio-Rad Mini-PROTEAN Tetra System for 27 h at 8 ◦C. After, gels were stained with silver nitrite.

### Protein Expression and Purification

Human cTnC (WT and mutants), ssTnI, and cardiac troponin T (cTnT) were expressed in E. coli and purified as described previously (Landstrom et al., 2008; Pinto et al., 2008a).

### Formation of Binary Complex ssTnI.cTnC

Five different binary complexes (ssTnI.cTnC-WT, ssTnI.cTnC-A8V, ssTnI.cTnC-C84Y, ssTnI.cTnC-E134D, and ssTnI.cTnC-D145E) were assembled essentially as described previously (Pinto et al., 2008a) for use in displacement assays. After expression and purification, the individual human troponin subunits of ssTnI or cTnC (WT or mutants) were first dialyzed against a buffer containing high urea (3M) and KCl (1M), then against two changes of buffer with the same salt composition but without urea. After this step, the protein concentration was determined using the PierceTM Coomassie Plus (Bradford) Assay Kit. Then ssTnI and individual cTnC (WT, A8V, C84Y, E134D, or D145E) proteins were mixed in a molar ratio of 1.3:1 ssTnI:cTnC. The protein mixtures were dialyzed against buffers of decreasing KCl concentration until binary complexes were formed (ssTnIcTnC). Next, the proteins were dialyzed into relaxing solution (pCa 8.0 = 10−<sup>8</sup> M free [Ca2+], 1 mM free [Mg2+], 7 mM EGTA, 2.5 mM MgATP2−, 20 mM MOPS, pH 7.0, 20 mM creatine phosphate, ionic strength was adjusted to I = 0.15 M using potassium propionate at 21◦C). Protein aggregates were removed by centrifugation and the binary complexes were stored at −80◦C until use.

### Skinned Fibers

Slow skeletal skinned fibers were isolated from rabbit soleus muscle. New Zealand White rabbits were sacrificed and muscles were isolated in accordance with NIH guidelines and animal care protocol approved by the Animal Care and Use Committees of University of Miami and Florida State University. Soleus muscles were tied to toothpicks and excised at the resting, in vivo length from the experimental animals. After this, the slow skeletal fibers were immersed in relaxing low Ca2<sup>+</sup> solution containing EGTA and 1% V/V Protease Inhibitor Cocktail (Sigma-Aldrich, P8340) (for details, see Pinto et al., 2008b).

For skinned cardiac preparations, left ventricular papillary muscles were dissected from fresh porcine hearts that were obtained from a local abattoir. Small muscle bundles were dissected and incubated overnight in relaxing low Ca2<sup>+</sup> solution containing 1% Triton X-100 at 4◦C to remove membranes.

Both slow skeletal and cardiac preparations were stored at −20◦C in relaxing solution (pCa 8.0) plus glycerol (52% V/V). For mechanical measurements, isolated slow skeletal fibers or cardiac muscle preparations were attached to a force transducer on one end and a microtranslator on the other end to adjust the length of the muscle preparation, and immersed initially in relaxing solution (pCa 8.0 with 15 units/ml creatine phosphokinase added). Muscle preparations were stretched at pCa 8.0 by 20% over slack length. Initial, maximal Ca2+-activated force of slow skeletal fibers and cardiac preparations was tested in saturating Ca2<sup>+</sup> conditions (pCa 4.0 = 10−<sup>4</sup> M free [Ca2+]) where the solution composition was otherwise the same as described above for relaxing solution.

### Extraction of Endogenous TnC from Slow Skeletal Skinned Fibers and Reconstitution with WT or HCM Mutant cTnC

After the determination of initial, maximal, active force (P0), the endogenous TnC was depleted by exposing slow skeletal skinned fibers to a CDTA extraction solution for fibers (5 mM 1,2 cyclohexylenedinitrilotetraacetic acid (CDTA), pH 8.4 adjusted with dried Tris powder) for 1.5–2 h, as described previously for extraction of TnC from skinned cardiac preparations (Landstrom et al., 2008). After successive washings with relaxing solution to remove CDTA, the residual active force at pCa 4.0 was measured to determine the efficacy of TnC extraction. The average residual force values after extraction were 22.4 ± 1.6, 25.3 ± 0.8, 25.1 ± 0.8, 23.7 ± 1.8, 26.0 ± 0.6% of P<sup>0</sup> for fibers that were to be reconstituted with WT, A8V, C84Y, E134D, or D145E, respectively. The residual force after extraction was not statistically different (ANOVA and t-test) among the five groups of fibers. For reconstitution with cTnC, the soleus fibers were successively exposed to drops of solution containing 28 µM cTnC, either WT or one of the mutants, in pCa 8.0 solution. TnC extraction and reconstitution were carried out at room temperature. To test the efficacy of TnC reconstitution, maximal force was tested again at pCa 4.0 for comparison with P0.

#### Veltri et al. HCM-cTnCs and Slow-Skeletal Muscle Regulation

### Displacement of Endogenous cTn Complex from Cardiac Preparations and Reconstitution with Binary Complex ssTnI.cTnC-WT or ssTnI.cTnC-HCM mutant

After measuring P0, the skinned cardiac preparations were equilibrated for 10 min in pre-incubation buffer (250 mM KCl, 20 mM MOPS, 5 mM MgCl2, 5 mM EGTA, 1 mM DTT, pH 6.2) (Pinto et al., 2008a) without cTnT. The fibers were then treated for 2.5 h with ∼0.8 mg/ml cTnT in the same buffer. The cTnT-treated fibers were subsequently washed in pre-incubation buffer without added cTnT to remove unbound, endogenous Tn complex and excess exogenous cTnT. The amount of endogenous Tn complex displaced by the excess cTnT was measured as the percentage of Ca2+-unregulated force, %UF = ([tension at pCa 8.0/tension at pCa 4.0]∗100) (Pinto et al., 2008a). Subsequently, the fibers were incubated in pCa 8.0 buffer containing pre-formed binary complex ssTnI.cTnC WT or mutant) for 1 h until the Ca2+-regulated force was restored (i.e., the fibers relaxed) and had reached steady state. All procedures were carried out at room temperature.

### Ca2<sup>+</sup> Dependence of Force Development and Cooperativity of Slow Skeletal Fibers Containing cTnC mutants or Cardiac Preparations Containing ssTnI.cTnC mutants

Measurement of Ca2<sup>+</sup> sensitivity and cooperativity of thin filament activation of steady-state, isometric force were performed using slow skeletal skinned fibers (reconstituted with WT or mutant cTnC) or skinned cardiac preparations (reconstituted with ssTnI.cTnC binary complexes containing WT or HCM mutants). The reconstituted slow skeletal or cardiac preparations were incubated in sequentially increasing [Ca2+] solutions, ranging from pCa 8.0 to 4.0 (decreasing pCa). The standard buffer for Ca2<sup>+</sup> sensitivity curves was 1 mM free [Mg2+], 7 mM EGTA, 2.5 mM MgATP2−, 20 mM MOPS (pH 7.0), 20 mM creatine phosphate, and 15 units/ml creatine phosphokinase, I = 150 mM and free [Ca2+] range from 10−<sup>8</sup> to 10−<sup>4</sup> M (pCa 8 to pCa 4). A pCa calculator program by Dweck et al. (2005) was used to calculate free [Ca2+] in the pCa solutions. Force measurements were conducted at room temperature (∼21◦C). Steady-state, isometric force data were normalized and fit to the 4-parameter Hill equation (Hill, 1910):

$$P = \frac{P\_0 \Big[Ca^{2+}\Big]^{n\_{H\text{ill}}}}{\left[Ca^{2+}\right]^{n\_{H\text{ill}}} + \left[Ca\_{50}^{2+}\right]^{n\_{H\text{ill}}}} + P\_{min}$$

where P<sup>0</sup> is the maximal force, P is the normalized force, "[Ca2<sup>+</sup> <sup>50</sup> ]" is the free [Ca2+] that produces 50% force and the exponent "nHill" is the Hill coefficient, which is related to the steepness of the relationship around [Ca2<sup>+</sup> <sup>50</sup> ] and is an indicator of apparent cooperativity for Ca2<sup>+</sup> activation of steady-state force. The [Ca2<sup>+</sup> <sup>50</sup> ] parameter estimates obtained from regression are reported as pCa<sup>50</sup> (i.e., −log[Ca2<sup>+</sup> <sup>50</sup> ].

### Myofibril Preparation

Rabbit soleus and diaphragm myofibrils used in **Figure 1** were prepared as described previously (Solaro et al., 1971) and stored in 50% glycerol at –20◦C. Rabbit back muscles myosin used in **Figure 1** was prepared as described (Margossian and Lowey, 1982) and stored in 50% glycerol at −20◦C. Myofibrils were prepared from porcine cardiac (used in **Figure 3**) and bovine masseter (used in **Figures 1**, **3**) muscle, obtained from a local abattoir, as described previously (Solaro et al., 1971) and stored in 50% glycerol at −20◦C. For SDS-PAGE analyses of MHC composition, rabbit soleus, rabbit diaphragm, bovine masseter myofibrils, and rabbit back muscles myosin from glycerol stocks were washed with MF buffer (30 mM Imidazole, 60 mM KCl, 2 mM MgCl2, pH 7.0, 1V myofibril or myosin/14V MF buffer). For myofibrillar ATPase activity assays, glycerol was removed by re-suspending and centrifuging the myofibrils using myofibril wash buffer containing 10 mM MOPS, 10 mM KCl, 2 mM dithiothreitol (DTT), pH 7.0 (1V myofibril/5V wash buffer x 3).

### Myofibrillar TnC Extraction and Reconstitution

Native TnC was depleted from myofibrils by incubating in CDTA extraction buffer for myofibrils (5 mM CDTA, 5 mM DTT, pH 8.4 adjusted with dried Tris powder) for ∼2.5 h at room temperature. Every 30 min, the sample was centrifuged and the supernatant

was discarded. At the end, myofibrils were washed three times in wash buffer to remove CDTA. Myofibrils were then reconstituted with cTnC–WT or—HCM mutants for 1 h at 4◦C. In order to minimize non-specific binding of cTnC, the reconstituted samples were washed several times with myofibril wash buffer. Quantification of myofibril protein concentration was performed using the PierceTMCoomassie Plus (Bradford) Assay Kit. Native, TnC-extracted, and TnC reconstituted myofibrils at 0.26 mg/ml were boiled in sample buffer (Laemmli, 1970) for 2 min and loaded into a 15% SDS-polyacrylamide gel.

### Myofibrillar ATPase Activity

Myofibrillar ATPase assays were performed using a buffer containing 2 mM EGTA, 3 mM nitrilotriacetic acid (NTA), 20 mM MOPS, 1 mM free Mg+<sup>2</sup> , ∼2.5 mM MgATP2−, pH = 7.0 with a fixed I = 80 mM and constant myofibril concentration of 0.4 mg/ml at 25◦C. Solutions were either pCa 8.0 or pCa 5.0. A pCa calculator program by Dweck et al. (2005) was used to calculate free [Ca2+] in the pCa solutions. The reaction was initiated by adding 2.94 mM ATP and quenched after 7 min by adding 4.6% trichloroacetic acid. After the assay reaction was terminated, precipitated proteins were removed by centrifugation for 7 min, at 3,095 × g and 4◦C. The concentration of inorganic phosphate in the supernatant, released by ATP hydrolysis, was measured using a colorimetric method according to the method of Fiske and Subbarow (1925).

### Statistical Analyses

Data were tested for significant differences using Student's ttest (unpaired) and ANOVA with post-hoc Tukey's Honest Significant Difference (HSD) test. For Student's t-test, mutant protein data were always compared to WT within the same experimental condition. For ANOVA, a multiple comparison approach was used. Differences for all statistical tests were considered significant when p < 0.05. When ANOVA yielded significance, Tukey's HSD test was then used to identify pairwise differences, and we report those differences found between mutant proteins vs. WT. Formally, multiple comparison using ANOVA is the statistically appropriate starting point, to be followed by a post-hoc test where significance was found by ANOVA. We have also included the results of pairwise comparisons using Student's t-tests where significance was found for consistency with prior work on cardiac muscle where only t-test analyses were performed. Our general interpretation is that ANOVA/Tukey's HSD is the more stringent test that identifies the strongest significant differences (which were also identified by ttests), while those identified by t-test (alone) provide guidance for changes that may still prove to be relevant. The data were expressed as mean ± S.E.M. Non-linear least squares regressions and Student's t-test analyses were performed using SigmaPlot software (version 12.0). ANOVA analyses and post-hoc Tukey's HSD tests were performed using R (version 3.3.2).

## RESULTS

The effects of HCM-associated cTnC mutants on the regulation of cardiac myofilaments have been extensively characterized. In **Figure 1A**, the schematic indicates the location of which muscle sections were excised from rabbit soleus muscle in order to assess whether different MHC isoforms are expressed within the sections. **Figure 1B** shows glycerol SDS-PAGE analysis of MHC isoforms in myofibrils isolated from rabbit soleus muscles, illustrating that only the slow-skeletal myosin isoform (i.e., MHC I) was uniformly detectable throughout these muscles. Glycerol SDS-PAGE analysis in **Figure 1C** shows, similarly, that only MHC I was detectable in bovine masseter muscle myofibrils.

We previously showed that cTnC A8V, C84Y, and D145E mutants increased Ca2<sup>+</sup> sensitivity of steady-state, isometric force development when reconstituted into skinned cardiac preparations, while the cTnC E134D mutant did not affect Ca2<sup>+</sup> sensitivity (**Table 1**) (Landstrom et al., 2008). Here, we determined the effects of the HCM cTnC mutants on the regulation of slow skeletal fibers. To accomplish this, TnC-depleted skinned fibers from rabbit soleus muscle were reconstituted with one of the exogenous cTnC proteins (WT, A8V, C84Y, E134D, or D145E). Only the C84Y mutant displayed a significant increase in the Ca2<sup>+</sup> sensitivity of force development and a decrease of the nHill (t-test and ANOVA/Tukey HSD, **Figure 2A** and **Table 1**). The pCa<sup>50</sup> (i.e., -log free [Ca2+] which yields a half-maximal response, where [Ca2+] is in molar units) obtained for cTnC C84Y reconstituted slow skeletal fibers was 6.33 ± 0.07; whereas, the pCa<sup>50</sup> of fibers reconstituted with WT cTnC was 6.00 ± 0.03. Moreover, when comparing the Ca2+ sensitization associated with the cTnC C84Y mutant between reconstituted slow skeletal and cardiac fibers, the slow skeletal muscle fibers further accentuated this effect. This difference is reflected in the 1pCa<sup>50</sup> (difference between the pCa<sup>50</sup> of fibers reconstituted with cTnC C84Y and WT) of slow skeletal muscle (+0.33 log units) and cardiac muscle (+0.27 log units; Landstrom et al., 2008) (**Table 1**). The nHill for slow skeletal fibers containing cTnC C84Y was significantly lower than WT, 1.26 ± 0.05 vs. 1.61 ± 0.07 (**Figure 2A** and **Table 1**).

Although reconstitution of the cTnC A8V or D145E mutants into skinned cardiac fibers led to an increase in the Ca2<sup>+</sup> sensitivity of contraction (Landstrom et al., 2008), this was not observed in the slow skeletal fibers (compared to WT cTnC reconstituted soleus fibers; **Figure 2A** and **Table 1**), suggesting that the presence of slow skeletal proteins could play a crucial role in normalizing the Ca2<sup>+</sup> sensitivity of contraction. The nHill values of soleus muscle fibers reconstituted with cTnC A8V or D145E were not statistically different from the WT control, indicating that in slow skeletal muscle, these cTnC mutants do not alter cooperativity (**Table 1**).

Reconstitution of the cTnC E134D mutant did not alter the Ca2<sup>+</sup> sensitivity or nHill of either skinned cardiac or slow skeletal preparations (**Figure 2A** and **Table 1**). Additionally, no significant differences in the maximal force were observed among slow skeletal fibers that were reconstituted with the WT or any of the four HCM cTnC mutants (**Figure 2B** and **Table 1**).

To assess the effects of the cTnC mutants on myofibrillar ATPase activity, we performed assays comparing the ATPase activity of both cardiac and slow skeletal myofibrils reconstituted with recombinant WT cTnC or mutants. Supplementary Figure 1 shows the SDS-PAGE analysis of cardiac and slow skeletal

TABLE 1 | Parameter summary for Ca2<sup>+</sup> dependence of steady-state isometric force in skinned porcine cardiac and rabbit soleus muscle preparations reconstituted with WT cTnC or cTnC HCM mutants.


<sup>a</sup>Values from Landstrom et al. (2008).

1pCa<sup>50</sup> = pCa<sup>50</sup> of mutant TnC–pCa<sup>50</sup> of WT TnC.

\*p < 0.05 HCM mutant vs. WT tested with Student's t-test.

#p < 0.05 from ANOVA and p < 0.05 from post-hoc Tukey's HSD test for HCM mutant vs. WT.

myofibrils, demonstrating TnC extraction from native myofibrils followed by cTnC reconstitution. At inhibiting (sub-diastolic) Ca2<sup>+</sup> concentration (pCa 8.0), cardiac myofibrils reconstituted with cTnC D145E (t-test and ANOVA/Tukey's HSD), or C84Y and D145E (t-tests) displayed significantly higher activities than cardiac myofibrils reconstituted with the WT control (**Figure 3A** and **Table 2**). This suggests that the cTnC D145E mutant, and possibly also the C84Y mutant, modifies inhibitory processes within cardiac muscle at low levels of Ca2+. At maximal activating Ca2<sup>+</sup> (pCa 5.0), cardiac myofibrils reconstituted with cTnC A8V, C84Y, or D145E exhibited higher levels of myofibrillar ATPase activity when individually compared to the WT control (t-tests), although no significant differences were identified by ANOVA (**Figure 3A** and **Table 2**). Differences in muscle regulation by the cTnC mutants were especially evident for the C84Y and D145E mutants in slow skeletal myofibrils. In the slow skeletal muscle background, cTnC C84Y decreased the myofibrillar ATPase activity at maximally activating Ca2<sup>+</sup> (pCa 5.0) when compared to the WT control (t-test and ANOVA/Tukey HSD; **Figure 3B** and **Table 2**). The effect of reconstituting the D145E mutant in slow skeletal myofibrils showed similar trends in ATPase activity to those observed in cardiac myofibrils: the D145E mutant increased ATPase activity at both low and high Ca2<sup>+</sup> concentrations when compared to the WT control (t-test and ANOVA/Tukey HSD; **Figure 3B** and **Table 2**).

Although cTnC A8V and D145E are capable of altering the physiological properties of cardiac muscle contraction (e.g., 1pCa<sup>50</sup> of +0.36 and +0.24, respectively), these mutants did not substantially alter the corresponding properties of slow skeletal muscle (**Table 1**). The presence of slow skeletal thin filament proteins may have counteracted the enhancement of apparent Ca2<sup>+</sup> affinity at cTnC's N-terminal regulatory Ca2<sup>+</sup> binding site which was observed in the cardiac preparations reconstituted

FIGURE 3 | Myofibrillar ATPase activity of (A) porcine cardiac and (B) bovine masseter muscle myofibrils that were depleted of TnC and reconstituted with WT cTnC or cTnC HCM mutants. (\*) Indicates significant differences (p < 0.05, Student's t-test) for mutant cTnCs compared to WT in the same muscle type and pCa. (#) Indicates significant differences (p < 0.05, ANOVA/Tukey's HSD) for mutant cTnCs compared to WT in the same muscle type and pCa. ATPase values for these experiments are summarized in Table 2. Data are shown as mean ± S.E.M.

TABLE 2 | Summary of Ca2<sup>+</sup> dependent ATPase activities from porcine cardiac and bovine masseter myofibrils reconstituted with WT cTnC or cTnC HCM mutants.


Values are shown as nmol Pi x mg−<sup>1</sup> x min−<sup>1</sup> .

Min = minimum and Max = maximum

Native cardiac myofibrils ATPase activity: 16.80 ± 0.58 and 67.60 ± 1.43 at pCa 8.0 and 5.0, respectively. TnC-extracted cardiac myofibrils ATPase activity 7.20 ± 2.01 and 8.80 ± 2.43 at pCa 8.0 and 5.0, respectively.

Native slow skeletal myofibrils ATPase activity: 46.75 ± 1.25 and 82.50 ± 1.18 at pCa 8.0 and 5.0, respectively. TnC-extracted cardiac myofrils ATPase activity 13.40 ± 0.92 and 33.40 ± 2.31 at pCa 8.0 and 5.0, respectively.

\*p < 0.05 HCM mutant vs. WT tested with Student's t-test.

#p < 0.05 from ANOVA and p < 0.05 from post-hoc Tukey's HSD test for HCM mutant vs. WT.

Cardiac myofibrils, n = 5–7.

Slow skeletal myofibrils, n = 8–14.

with cTnC A8V or D145E. To better understand the influence of the different sarcomeric isoforms (i.e., cardiac vs. slow skeletal) on these phenomena, we tested whether these effects are partially or totally attributed to the presence of ssTnI. Therefore, we exchanged the endogenous cTn complex of skinned cardiac fibers with exogenous chimeric Tn complexes comprised of cTnC, ssTnI, and cTnT. To assess the amount of displaced endogenous cTn achieved by the cTnT displacement method, we recorded the average values of the cardiac preparation's Ca2+-unregulated force (i.e., force generated by preparations that have displaced cTnC and cTnI by incubation with cTnT), which were: 88.5 ± 3.4, 82.1 ± 4.2, 80.2 ± 7.2, 86.7 ± 8.4, and 91.7 ± 3.8% for preparations that were to be reconstituted with cTnC WT, A8V, C84Y, E134D, or D145E, respectively. After cTn displacement, cardiac preparations were incubated with the exogenous binary complex containing ssTnI and WT or mutant cTnC. Finally, the recovered maximal Ca2<sup>+</sup> regulated force was recorded. The maximal force values were: 81.1 ± 4.7, 82.7 ± 4.6, 91.8 ± 2.4, 92.1 ± 5.0, and 84.4 ± 4.1% of the P<sup>0</sup> for WT, A8V, C84Y, E134D, and D145E, respectively. No significant differences in the Ca2+ unregulated force, as well as, the Ca2+-regulated maximal force among the five experimental groups existed (t-test and ANOVA; **Table 3**).

**Figure 4** indicates that incorporation of ssTnI into cardiac preparations partially attenuated the effects of cTnC A8V and D145E on the Ca2<sup>+</sup> sensitivity of force development. The 1pCa<sup>50</sup> values for cardiac preparations reconstituted with cTnC A8V and D145E together with ssTnI were +0.09 and +0.18, respectively (**Figure 4** and **Table 3**); whereas, in skinned cardiac preparations reconstituted with cTnC A8V and D145E alone (i.e., in the presence of endogenous cTnI), the ∆pCa<sup>50</sup> values were +0.36 and +0.24, respectively (**Table 1**) (Landstrom et al., 2008). In contrast, in skinned cardiac preparations reconstituted with binary complex containing ssTnI and cTnC C84Y, the presence of

FIGURE 4 | Ca2<sup>+</sup> dependence of steady-state isometric force development in porcine cardiac preparations reconstituted with slow skeletal TnI together with WT cTnC or cTnC HCM mutants. Comparisons of Ca2<sup>+</sup> sensitivity of force development obtained for cTnC WT and HCM mutants, in (A) WT vs. A8V, (B) WT vs. C84Y, (C) WT vs. E134D, and (D) WT vs. D145E. Regression parameter estimates of pCa50 and nHill, along with 1pCa50 and relative maximal force values for these experiments are summarized in Table 3. Data are shown as mean ± S.E.M.

TABLE 3 | Parameter summary for Ca2<sup>+</sup> dependence of steady-state isometric force in skinned porcine cardiac preparations reconstituted with ssTnI together with WT cTnC or cTnC HCM mutants.


The Ca2<sup>+</sup> unregulated force following TnT incubation (to displace endogenous troponin) was calculated according to: (PpCa8/PpCa4) X 100, where PpCa<sup>8</sup> and PpCa<sup>4</sup> are the forces generated at pCa 8.0 and pCa 4.0, respectively.

1pCa<sup>50</sup> = pCa<sup>50</sup> of mutant TnC–pCa<sup>50</sup> of WT TnC.

\*p < 0.05 HCM mutant vs. WT tested with Student's t-test.

#p < 0.05 from ANOVA and p < 0.05 from post-hoc Tukey's HSD test for HCM mutant vs. WT.

ssTnI did not attenuate the myofilament Ca2<sup>+</sup> sensitization that was observed in the presence of endogenous cTnI. These cardiac preparations exhibited an even higher Ca2<sup>+</sup> sensitivity of force development than what was observed in cardiac preparations containing the cardiac isoform of TnI (1pCa<sup>50</sup> +0.41 or +0.27, respectively; **Tables 1, 3**). A comparison of nHill regression parameter estimates obtained from skinned cardiac preparations reconstituted with ssTnI and HCM cTnC mutants indicates that the C84Y (t-test and ANOVA/Tukey's HSD) and D145E (t-test) mutants significantly reduced apparent cooperativity in comparison to the WT control (1.28 ± 0.04, 1.57 ± 0.03, and 1.74 ± 0.04 for C84Y, D145E, and WT, respectively; **Table 3**). The cardiac preparations reconstituted with exogenous ssTnI and cTnC E134D showed no significant changes in relation to the pCa<sup>50</sup> and nHill (t-test and ANOVA/Tukey's HSD) when compared with WT control (**Figure 4** and **Table 3**).

### DISCUSSION

Patients that exhibit cardiomyopathy and possess a positive family history are routinely screened for mutations in a number of thin and thick filament encoding genes. Obvious clinical symptoms such as dyspnea, syncope, chest pain, and arrhythmia—in addition to characteristic cardiac remodeling revealed through imaging—are frequently documented complaints. Many sarcomeric mutations that are identified on the basis of cardiac dysfunction, however, are also expressed in skeletal muscle. With the clinical focus on the heart, more subtle skeletal muscle abnormalities may go undetected (Towbin, 2014). We are interested in whether these mutations may have pathogenic consequences in skeletal muscle that go unreported and/or undetected. Here, we queried whether mutations in TNNC1 that are linked to HCM and expressed in slow skeletal muscle can manifest as a dysfunction or affect the contractile properties of slow skeletal muscle.

While the cardiac effects of TNNC1 mutants were first examined in the context of HCM pathogenic properties and disease causation, they have not been assessed in slow skeletal muscle. This study is the first to examine the effects of mutations present in the TNNC1 gene on slow skeletal muscle regulation (Song et al., 1996). We found that for most of the HCM cTnC mutants examined in this study, their incorporation into slow skeletal muscle ablated the increased Ca2<sup>+</sup> sensitivity of force development that is seen in cardiac preparations. Only cTnC C84Y maintained its Ca2<sup>+</sup> sensitizing effects in the slow skeletal muscle fibers. Surprisingly, the effects of this mutation were amplified when incorporated into the slow skeletal myofilament. This suggests that this mutant exerts distinct effects on a common element that exist within both muscle types. Since skeletal muscle comprises a large percentage of a person's mass, and slow type I muscle fibers are found throughout human musculature, it is reasonable to expect that slow muscle dysfunction could impart a substantial burden on the affected individual.

Differences in the protein structure of slow skeletal and cardiac TnI and TnT underlie the functional properties of these muscle types, especially when coupled with the distinctive properties of each mutant cTnC in this study. The overall structures of cardiac and skeletal muscle troponins are similar, at least in troponin's core domain (Takeda et al., 2003; Vinogradova et al., 2005). However, three major differences between skeletal and cardiac troponin structures may contribute to the differential effects: (1) structure of the TnI inhibitory segment is ordered in the skeletal muscle isoform but flexible and disordered in the cardiac isoform (Takeda et al., 2003; Vinogradova et al., 2005); (2) the position and conformation of the TnT2 N-terminal helix (Takeda et al., 2003; Vinogradova et al., 2005); and (3) cTnI contains an N-terminal extension not present in the skeletal isoforms—which harbors two serine residues capable of undergoing phosphorylation by protein kinase A (Zhang et al., 1995). Based upon the above, the experiments involving ssTnI would not be influenced by phosphorylation of the cardiac-specific, N-terminal extention of cTnI.

The myofibrillar ATPase assays further explored the mechanism underlying the Ca2<sup>+</sup> sensitization induced by the C84Y mutant in both types of muscle fibers. At saturating Ca2<sup>+</sup> (pCa 5.0), slow skeletal ATPase activity was significantly reduced, while cardiac ATPase activity trended higher. In cardiac muscle, the residue C84 contacts the cTnI switch region when the N-terminus of cTnC is in the open conformation and the BC loop of cTnC when it is in the closed conformation. Therefore, the C84Y mutant may stabilize the open conformation when bound to cTnI (Li and Hwang, 2015) and affect the critically important positioning of the cTnC N-domain (Hwang et al., 2014). Upon binding of divalent cations, the global structure of cTnC becomes more compact (Jayasundar et al., 2014; Badr et al., 2016), which could likely be influenced by mutations in the cTnC linker region. These alterations in the cTnC N-domain may increase the affinity between cTnC-C84Y and the cTnI switch region and ultimately impact initiation of cardiac contraction. The switch region of cTnI (148–158) is immediately flanked by the inhibitory (136–147) and regulatory (161–209) regions, which anchors to actin during diastole (Tripet et al., 1997) and, in doing so, is largely responsible for inhibition of contractility at diastolic Ca2<sup>+</sup> levels (Meyer and Chase, 2016). Increased affinity between cTnC-C84Y and the cTnI switch region would therefore be expected to reduce interactions between the cTnI inhibitory and regulatory regions during diastole, permitting actomyosin ATPase activity at subthreshold Ca2<sup>+</sup> concentrations, thereby promoting greater ATPase activation at higher Ca2<sup>+</sup> concentrations.

Based upon the study by Westfall et al. (1997) on ssTnIoverexpressing cardiomyocytes, the reduction in Ca2<sup>+</sup> activation threshold in slow skeletal fibers containing cTnC-C84Y (**Figure 2** and **Table 1**), may arise from altered interactions within individual regulatory units of the thin filament (7 actin monomers, 1 tropomyosin, and 1 cTn complex). Weaker interactions between ssTnI and cTnC-C84Y may lower the threshold for Ca2+-activated development of tension and reduce cooperativity in cardiac muscle reconstituted with ssTnI. In addition, it is also possible that mutations in troponin differentially modulate the ATPase activity of myosin (Schoffstall et al., 2011; Chalovich, 2012) on top of their effects on Ca2<sup>+</sup> activation of the thin filament. While Ca2+-activation of the cardiac thin filament is more dependent on cross-bridges than skeletal muscle (Gillis et al., 2007), cycling crossbridges at submaximally activating Ca2<sup>+</sup> concentrations could result in reduced cooperativity in the presence of Ca2+-sensitizing mutations (Westfall et al., 1997). The less flexible TnI switch region in skeletal muscle appears to differently affect Ca2<sup>+</sup> regulation of myofibrillar ATPase activity in the presence of the C84Y mutation in cTnC. The differential effect of C84Y on myofibrillar ATPase activity between tissue types suggests that the C84Y mutation in cTnC could increase Ca2<sup>+</sup> sensitivity of contraction in both muscles by distinct mechanisms. In cardiac muscle, it may increase the overall number of cycling crossbridges; while in the slow skeletal muscle, it may decrease the overall number of cycling cross-bridges and enhance the population of strongly bound cross-bridges.

The other HCM cTnC mutants examined in this study are located in distinct structural regions of cTnC. A8V is present in the cTnC N-helix, which plays a key role in defining the orientation of the cTnC N-domain relative to the troponin complex. As described above, alterations in cTnC Ndomain orientation can alter the affinity between the cTnC N-domain and the cTnI switch peptide; these changes are likely responsible for increasing the cardiac myofibrillar ATPase activity at saturated Ca2<sup>+</sup> concentration in this (**Figure 3** and **Table 2**) and prior studies (Zot et al., 2016). However, cTnC-A8V does not affect slow skeletal myofibrillar ATPase measurements or alter cooperativity in cardiac preparations reconstituted with cTnC.ssTnI.cTnT. Functionally, the D145E mutant is distinct; it is located in the C-domain of cTnC and has been shown to increase the Ca2<sup>+</sup> affinity of site II of isolated cTnC, cTn complex and the thin filament (Pinto et al., 2009). The D145E mutant has also been shown to increase cTnC binding affinity to the cardiac thin filament (Marques et al., 2017). The affinity between the TnI switch region and cTnC may be increased regardless of which TnI isoform is present, as indicated by increased activity of myofibrillar ATPase in both the slow skeletal and cardiac myofibrillar ATPase at both low and high Ca2<sup>+</sup> concentrations (**Figure 3** and **Table 2**). The D145E mutant did not affect the Ca2<sup>+</sup> sensitivity of skinned slow skeletal fibers (**Figure 2** and **Table 1**). Therefore, it could be speculated that new interactions/altered affinity between the cTnC D145E C-domain and ssTnT could potentially normalize the sensitizing effects of the mutant, which could be confirmed by peptide binding studies. The E134D mutant did not affect myofibrillar ATPase activity (**Figure 3** and **Table 2**) or Ca2+-sensitivity of force development or cooperativity of thin filament activation (**Figure 2** and **Table 1**) in either skeletal or cardiac preparations, further confirming the likelihood that it is a non-pathogenic variant (Landstrom et al., 2008).

The incorporation of ssTnI into cardiac muscle attenuated the dysfunction of myofilaments containing the HCM-associated cTnC mutants A8V and D145E. This suggests that ssTnI plays a protective role in regard to its ability to mitigate effects of disease-causing cTnC mutants in the heart. However, the protective role of ssTnI in cardiac muscle contraction is not a novel topic. We have previously shown that ssTnI has the ability to attenuate the functional consequences of two mutations in cTnT that are associated with restrictive cardiomyopathy (Pinto et al., 2008a, 2011b). Others have shown that the expression of ssTnI in the murine heart has a protective effect on skinned cardiac preparations and left ventricular function during stress conditions, such as acidosis (Wolska et al., 2001; Urboniene et al., 2005). As mentioned above, the presence of ssTnI in cardiac muscle preparation was not able to completely abolish the effect of the cTnC mutants A8V and D145E, as seen in **Figure 4** in cardiac preparations reconstituted with ssTnI. This suggests that the presence of ssTnT in slow skeletal muscle may also play a role in rescuing the cTnC mutants' functional phenotype. Major differences exist between the structure of ssTnT and cTnT isoforms, where a large portion of the N-terminus present in cTnT is absent in ssTnT (Pinto et al., 2012). This region of cTnT has been shown to have important roles in regulating Ca2<sup>+</sup> sensitization and activation of the skinned fibers due to effects on the dynamics of strong TnT-Tm interactions (Gomes et al., 2002; Gollapudi et al., 2012). Therefore, we speculate that some of the protective effect observed in slow skeletal muscle could be arising from the N-terminus of ssTnT.

While the crossover of mutation effects in both tissue types may lead to pathological consequences in both slow skeletal and cardiac muscle, manifestations may present themselves over distinctive time courses and with varying severities. When comparing the age of diagnosis and the symptoms of HCM patients in the cohort study that identified four variants in the TNNC1 gene, it is interesting to note that the patient bearing the C84Y mutant was diagnosed at the earliest age (8.4 years) and exhibited syncope on exertion (Landstrom et al., 2008). Although syncope is a common event, it can signal more ominous underlying conditions (including HCM and arrhythmogenic right ventricular dysplasia) and outcomes such as sudden death (O'Connor et al., 1999). Alterations in skeletal muscle function may lead to additional downstream consequences that remain unknown. Function may be altered further by globally affecting signaling pathways that influence sarcomeric protein isoform switching and energy consumption. These changes could have profound consequences if the aberrant function arises in multiple muscle types, e.g., the cTnC mutants in slow skeletal muscle—despite slow muscle's fatigue-resistant properties that are associated with differential sensitivity to metabolite levels (Chase and Kushmerick, 1988, 1995) in normal individuals—could also be synergistic with regard to patient fatigue or muscle weakness. The true impact of skeletal and cardiac muscle dysfunction as a result of HCM mutations in the TNNC1 gene will bear out after examination of more cases within the clinic. Individuals affected by mutations in genes expressed more systemically have the potential to influence the time course and severity of disease; therefore, it is of interest to study the broader effects of mutations that may modify disease outcomes when manifesting in multiple muscle types.

### AUTHOR NOTE

Part of this work was presented in the Biophysical Society 56th Annual Meeting in San Diego, California, USA (February, 2012).

### AUTHOR CONTRIBUTIONS

Performed experiments: TV, ML, DG, KJ, CM, and DD. Designed Study: TV, MP, DD, AL, PC, and JP. Analyzed Data: TV, ML, CM, DD, and JP. Wrote/Edited Manuscript: TV, ML, MP, DD, AL, PC, and JP.

### ACKNOWLEDGMENTS

The authors would like to thank Jingsheng Liang for his technical assistance during this project. In addition, the authors thank the funding agencies that helped fund this work: National Institute of Health grants HL103840 and HL128683 (JP); American Heart Association postdoctoral fellowship #09POST2300030 (MP); James and Esther King Foundation 1KD03-33923 (DD); National Center of Science and Technology for Structural Biology and Bioimaging, CENABIO, Brazil and Brazilian National Research Council, CNPq Science without Borders 229922/2013-9 (ML); fellowship from CNPq (TV). NIH-L40 (HL129273-01), the Pediatric and Congenital

### REFERENCES


Electrophysiology Society Paul C. Gillette Award, and pilot grant funding from the Baylor College of Medicine Department of Pediatrics (AL).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2017.00221/full#supplementary-material


C, TNNC1-A31S, Predisposes to hypertrophic cardiomyopathy and ventricular fibrillation. J. Biol. Chem. 287, 31845–31855. doi: 10.1074/jbc.M112.377713


can we learn about thin filament function? J. Mol. Cell. Cardiol. 48, 882–892. doi: 10.1016/j.yjmcc.2009.10.031


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer DWDK and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2017 Veltri, Landim-Vieira, Parvatiyar, Gonzalez-Martinez, Dieseldorff Jones, Michell, Dweck, Landstrom, Chase and Pinto. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Increased Postnatal Cardiac Hyperplasia Precedes Cardiomyocyte Hypertrophy in a Model of Hypertrophic Cardiomyopathy

Emily T. Farrell 1 †, Adrian C. Grimes 2 †, Willem J. de Lange<sup>1</sup> , Annie E. Armstrong<sup>1</sup> and J. Carter Ralphe<sup>1</sup> \*

*<sup>1</sup> Department of Pediatrics, University of Wisconsin School of Medicine and Public Health, Madison, WI, United States, <sup>2</sup> Department of Medicine, University of Wisconsin School of Medicine and Public Health, Madison, WI, United States*

#### Edited by:

*P. Bryant Chase, Florida State University, United States*

#### Reviewed by:

*Sakthivel Sadayappan, University of Cincinnati, United States Aaron Olson, University of Washington, United States*

#### \*Correspondence:

*J. Carter Ralphe jcralphe@pediatrics.wisc.edu † These authors have contributed equally to this work.*

#### Specialty section:

*This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology*

Received: *27 April 2017* Accepted: *30 May 2017* Published: *14 June 2017*

#### Citation:

*Farrell ET, Grimes AC, de Lange WJ, Armstrong AE and Ralphe JC (2017) Increased Postnatal Cardiac Hyperplasia Precedes Cardiomyocyte Hypertrophy in a Model of Hypertrophic Cardiomyopathy. Front. Physiol. 8:414. doi: 10.3389/fphys.2017.00414* Rationale: Hypertrophic cardiomyopathy (HCM) occurs in ∼0.5% of the population and is a leading cause of sudden cardiac death (SCD) in young adults. Cardiomyocyte hypertrophy has been the accepted mechanism for cardiac enlargement in HCM, but the early signaling responsible for initiating hypertrophy is poorly understood. Mutations in cardiac myosin binding protein C (*MYBPC3*) are among the most common HCM-causing mutations. Ablation of *Mybpc3* in an HCM mouse model (cMyBP-C−/−) rapidly leads to cardiomegaly by postnatal day (PND) 9, though hearts are indistinguishable from wildtype (WT) at birth. This model provides a unique opportunity to explore early processes involved in the dramatic postnatal transition to hypertrophy.

Methods and Results: We performed microarray analysis, echocardiography, qPCR, immunohistochemistry (IHC), and isolated cardiomyocyte measurements to characterize the perinatal cMyBP-C−/<sup>−</sup> phenotype before and after overt hypertrophy. cMyBP-C−/<sup>−</sup> hearts showed elevated cell cycling at PND1 that transitioned to hypertrophy by PND9. An expanded time course revealed that increased cardiomyocyte cycling was associated with elevated heart weight to body weight ratios prior to cellular hypertrophy, suggesting that cell cycling resulted in cardiomyocyte proliferation. Animals heterozygous for the cMyBP-C deletion trended in the direction of the homozygous null, yet did not show increased heart size by PND9.

Conclusions: Results indicate that altered regulation of the cell cycling pathway and elevated proliferation precedes hypertrophy in the cMyBP-C−/<sup>−</sup> HCM model, and suggests that increased cardiomyocyte number contributes to increased heart size in cMyBP-C−/<sup>−</sup> mice. This pre-hypertrophic period may reflect a unique time during which the commitment to HCM is determined and disease severity is influenced.

Keywords: hypertrophic cardiomyopathy, HCM, myosin binding protein C, MyBP-C, hyperplasia, proliferation, cell cycling

**318**

### INTRODUCTION

Hypertrophic cardiomyopathy (HCM) is a prevalent cause of heart failure in adults and a leading cause of sudden cardiac death (SCD) in apparently healthy young individuals (Fananapazir and Epstein, 1991). It is a primary cardiac disease inherited in autosomal dominant fashion and estimated to affect 1 in 200 individuals (Spirito et al., 1997; Semsarian et al., 2015). The features of HCM include left ventricular hypertrophy, myocardial disarray, fibrosis, diastolic dysfunction, and an increased risk of both SCD and congestive heart failure (Fananapazir and Epstein, 1994; Davies and McKenna, 1995). HCM has long been regarded as a disease of late adolescence and adulthood. Involvement in neonates and young children is relatively rare, but early-onset HCM is typically quite severe (Van Driest et al., 2004; Xin et al., 2007; Zahka et al., 2008). At least 14 HCM-causative genes have been identified, with the majority of genotype-positive cases of familial HCM resulting from mutations in the genes for either β-myosin heavy chain (MYH7) or MYBPC3, which encodes cardiac myosin binding protein C (cMyBP-C; Richard et al., 2003; Olivotto et al., 2008; Alfares et al., 2015). cMyBP-C is a thick filament regulatory protein that binds myosin (Okagaki et al., 1993; Gruen and Gautel, 1999), titin (Labeit et al., 1992), and actin (Razumova et al., 2006), and functions as an important determinant of cardiac contractile reserve (Tong et al., 2008; de Lange et al., 2013). Nearly 200 disease-associated mutations in cMyBP-C are described and include missense mutations and early truncations (Harris et al., 2011). While it is recognized that the majority of humans with genotype-positive HCM are heterozygotes, homozygosity, and compound heterozygosity do exist and are associated with early-onset and typically more severe HCM (Xin et al., 2007; Zahka et al., 2008; Ripoll Vera et al., 2010).

We and others previously reported a mouse model of HCM in which cMyBP-C is genetically ablated (Harris et al., 2002; Korte et al., 2003; Stelzer et al., 2006; de Lange et al., 2011, 2013). This model recapitulates several important aspects of human HCM with both sarcomere dysfunction and a pronounced and fully penetrant phenotype described as hypertrophic (Harris et al., 2002; Korte et al., 2003; Stelzer et al., 2006; Moss et al., 2015). However, a remarkable characteristic of this germline ablation of an important structural and regulatory protein is that it appears to have negligible impact on embryonic cardiac development, while producing a severe hypertrophic phenotype over a very rapid time frame after birth (de Lange et al., 2013). The heart of the postnatal day (PND) 1 cMyBP-C−/<sup>−</sup> mouse isindistinguishable from WT, but significantly and rapidly enlarges over the first 9 days of life—as demonstrated by histology and heart weight (HW) to body weight (BW) ratios (Harris et al., 2002; de Lange et al., 2011). The mechanisms involved in the rapid development of HCM during the perinatal transition have not been previously explored in the cMyBP-C−/<sup>−</sup> mouse. Our initial hypothesis was that hypertrophic signaling pathways would be differentially upregulated in the cMyBP-C−/<sup>−</sup> myocardium at PND1, prior to the observed gross morphologic changes seen by PND9. Full characterization and improved understanding of the molecular events during this period could identify potential therapeutic targets to modify disease course.

In this study, we performed a microarray analysis comparing WT and cMyBP-C−/<sup>−</sup> hearts at PND1 and PND9 to define differential gene regulation that may precede and contribute to the hypertrophic phenotype. Based on these results we then further examined the status of the perinatal heart in embryonic (E) day 18.5 to PND9 mice deficient in cMyBP-C to define the relative contribution of hyperplasia and/or hypertrophy in the development of the observed early cardiac enlargement. A recent study performed by Jiang et al. (2015) using a mouse model with a truncation mutation in MYBPC3 (cMyBP-Ct/<sup>t</sup> mouse; McConnell et al., 1999; Sadayappan et al., 2006; Jiang et al., 2015), identified elevated cardiomyocyte proliferation extending through PND17 (Jiang et al., 2015). Unlike the cMyBP-C−/<sup>−</sup> model, however, no cardiomyocyte hypertrophy was observed in the cMyBP-Ct/<sup>t</sup> mice through 5 weeks of age. Importantly, the cMyBP-C−/<sup>−</sup> and cMyBP-Ct/<sup>t</sup> mice are fundamentally different, as the cMyBP-Ct/<sup>t</sup> mouse has been shown to produce protein, albeit at a low level (McConnell et al., 1999), whereas the cMyBP-C <sup>−</sup>/<sup>−</sup> mouse is a true knockout with no protein expression. A comparison of the data from Jiang et al. with data presented here suggests that the two models develop cardiomyopathies through distinct cellular pathways, potentially highlighting a mutationspecific divergence in phenotype.

### METHODS

### Animals

Heterozygous (cMyBP-C+/−) adults were derived from backcrosses of WT E129X1/SvJ mice (Taconic, Hudson, NY) with homozygous cMyBP-C−/<sup>−</sup> animals previously generated on the E129X1/SvJ background (Harris et al., 2002). Subsequent breedings of cMyBP-C+/<sup>−</sup> adults resulted in expected Mendelian frequencies (1:2:1) of WT (cMyBP-C+/+), heterozygous (cMyBP-C <sup>+</sup>/−) and homozygous knockout (cMyBP-C−/−) animals. Mice in this study were produced from MyBP-C heterozygote (MyBP-C <sup>+</sup>/−) crossings, with subsequent postmortem genotyping of pups. Littermate comparisons among genotypes were performed whenever possible.

Pups from these breedings were decapitated and intact hearts were harvested, rinsed in 1X PBS, and the blood ejected using blunt forceps. A post-mortem tail snip was taken from each mouse to allow subsequent genotyping and sex determination by PCR. Live postnatal day (PND) 1 and PND9 WT and

**Abbreviations:** AMP, adenosine monophosphate; APC, anaphase-promoting complex; BW, body weight; Ccna2, cyclin A2; Cdc25c, cell division cycle 25c; Cdk, cyclin dependent kinase; cMyBP-C, cardiac myosin binding protein C (protein); DCM, dilated cardiomyopathy; E, embryonic day; Erk 1/2, extracellular-signalregulated kinases; GC-A, guanylyl cyclase A; HCM, hypertrophic cardiomyopathy; HW, heart weight; IPA, Ingenuity Pathway Analysis; IVRT, isovolumetric relaxation time; KEGG, The Kyoto Encyclopedia of Genes and Genomes; LVAW, left ventricular anterior wall thickness; LVPW<sup>s</sup> , left ventricular posterior wall thickness (during systole); MHC, myosin heavy chain; MYBPC3, human cardiac myosin binding protein C (gene); Mybpc3, mouse cardiac myosin binding protein C (gene); Myh6, beta myosin heavy chain (gene); Myh7, alpha myosin heavy chain (gene); Nppa, atrial natriuretic peptide A; PND, postnatal day; Prkag2, protein kinase AMP-activated non-catalytic subunit gamma 2; SCD, sudden cardiac death; WGA, wheat germ agglutinin; WT, wild-type.

cMyBP-C−/<sup>−</sup> pups used for echocardiographic measurements were derived from cMyBP-C−/<sup>−</sup> × cMyBP-C−/<sup>−</sup> and WT × WT crosses. PND0 is defined here as the day of birth. This study was approved by the Animal Care and Use Committee of the School of Medicine and Public Health at the University of Wisconsin Madison in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health publication no. 85-23, revised 1985).

### Generation of Embryonic Day (E)18.5 pups and Extraction of Hearts

cMyBP-C+/<sup>−</sup> dams were bred to cMyBP-C+/<sup>−</sup> males for 24 h, after which the males were removed from the cages to establish the timing of pregnancy. Pregnant dams were anesthetized using inhaled isoflurane on day 19 after the breeding date. The abdomens of the dams were opened and the uterine horns containing the embryos were placed in a petri dish with 1X PBS. Embryos were removed from the uterine horn and embryonic sacs, and weighed prior to isolation of each heart. Hearts were rinsed in 1X PBS, weighed and snap frozen in liquid nitrogen.

### Routine Histology

PND1 and PND9 mouse hearts harvested as described were fixed and embedded in paraffin according to standard histological procedures detailed in Supplementary Methods. Sections were then stained with hematoxylin and eosin (H&E) using SelecTec reagents and protocols (Surgipath) for general assessment of tissue structure, or by commercially available Masson's Trichrome or Elastic Trichrome stains (Sigma) to highlight elastic fibers and collagen, or by picrosirius red stain (Polysciences, Inc.) to highlight interstitial fibrosis.

### Transthoracic Echocardiography on WT and cMyBPC−/<sup>−</sup> Mice

Hearts of conscious PND1 and anesthetized PND9 mice were imaged using a Vevo 770 ultrasonograph (Visual Sonics) with a 40-MHz transducer (Agilent Technologies). Details regarding anesthesia and animal preparation are available in Supplementary Methods. Two-dimensionally guided M-mode images were acquired at the tip of papillary muscles, and transmitral and aortic velocities measured using Doppler pulse wave imaging. End diastolic and systolic left ventricular (LV) diameter, as well as anterior wall (AW) and posterior wall (PW) thickness, were measured on-line from M-mode images using the leading edge to leading edge convention. All parameters were measured over at least three consecutive cardiac cycles and averaged. Heart rate was determined from at least three consecutive intervals from pulse wave Doppler tracings of the LV outflow tract. Ejection fraction was calculated by 100 × (LV volume diastole - LV volume systole)/LV volume diastole. Isovolumic relaxation time was measured as the time between closing of the aortic valve and the opening of the mitral value from pulse wave Doppler tracings of the LV outflow tract and mitral inflow region.

### Microarray Hybridization and Data Analysis

The left ventricular free wall (LVFW) was dissected from the hearts of PND1 and PND9 pups generated from the breeding of heterozygous cMyBP-C+/<sup>−</sup> adults, and snap frozen in liquid nitrogen. After PCR determination of genotype and sex, the LVFW of seven males from each genotype and age was used for extraction of total RNA as described in Supplementary Methods. Quantification, quality and integrity of RNA were determined with the NanoDrop 2000 (Thermo Scientific), Agilent BioAnalyzer and Agilent RNA Nano Chip (Agilent Technologies). Four hundred nanograms of RNA (and 400 ng of poly-A RNA control) was used for overnight invitro labeling with the Ambion GeneChip WT Expression kit following manufacturer's protocols. Following labeling, samples were purified/quantified on the NanoDrop 2000 and 10µg of purified cRNA used for template input to generate single strand cDNA, 5.5µg of which was then fragmented and subjected to end-terminus labeling using the Affymetrix WT terminal Labeling Kit (Affymetrix). Samples were hybridized to Affymetrix Mouse Gene 1.0 ST whole genome arrays at 45◦C for 16 h following procedures outlined in the GeneChip WT Terminal Labeling and Hybridization user manual. Arrays were postprocessed on the Affymetrix GeneChip Fluidics Station 450, scanned on the GeneChip 3000 7G and the data extracted and processed using Affymetrix Command Console (version 3.1.1.1229) according to manufacturer's protocols. Resulting GeneChip.cel files were uploaded to iReport (Ingenuity Systems) or to Genesifter (Geospiza), and normalized via Robust Multiarray Average (RMA) using the default options of each system. Genes differentially expressed up or down >1.5-fold at p = 0.05) were identified using pairwise analysis (comparing WT to cMyBPC−/<sup>−</sup> and/or comparing PND1 to PND9) or using twoway ANOVA (for multi-factoral analysis). T-test statistics were performed using a Benjamini and Hochberg correction.

### RT-qPCR

Whole hearts were dissected from E18.5, PND0, PND1, PND2, and PND9 pups generated from the breeding of heterozygous cMyBP-C+/<sup>−</sup> adults, and snap-frozen in liquid nitrogen. RNA isolation and qRT-PCR were performed on hearts of each age and genotype according to standard methods, as described in Supplementary Methods.

### Immunohistochemistry for Cell Cycling Marker on Paraffin-Embedded Hearts

IHC on paraffin-embedded hearts were performed according to standard methods as detailed in Supplementary Methods, using the following antibodies: 1:20 anti-α-tropomyosin (CH-1; Developmental Studies Hybridoma Bank, University of Iowa, IA) to label myocardium, and 1:200 anti-Ki67 rabbit polyclonal antibody (Abcam, ab15580) to label cells actively participating in the cell cycle (**Figure 2**). For secondary detection, sections were incubated in 1:200 AlexaFluor 568 goat anti-mouse IgG<sup>1</sup> and Alexafluor 488 goat anti-rabbit IgG(H+L) secondary antibodies (Molecular Probes, Eugene, OR, USA). In each case, as a control, similar sections were incubated with secondary antibodies only. Sections were coverslipped using warmed ProLong Gold Antifade Reagent (Invitrogen) with 4′ ,6 diamidino-2-phenylindole (DAPI) to label nuclei.

### Analysis of Cell Cycling by IHC

The number of cycling cardiomyocytes was determined in images of LV and septum from coronal sections labeled with antibodies against α-tropomyosin and the Ki-67 antigen, and counterstained with DAPI. Ki-67- and α-tropomyosin-labeled cells were counted in representative sections of at least three hearts of each genotype and age, and the mean calculated. Cells labeled with Ki-67 but not α-tropomyosin were designated as non-cardiomyocytes and thus excluded from counts. All calculations were subjected to statistical analysis using standard Students' t-test.

### Cryogenic Preservation of Hearts, Cryosectioning, and IHC

Hearts were cryogenically preserved and cryosectioned according to standard procedures detailed in Supplementary Methods. Sections were mounted on glass slides and IHC was performed using 1:1,000 anti-α-actinin mouse monocolonal antibody (Sigma, A7811) to label cardiomyocytes, followed by AlexaFluor 568 goat anti-mouse IgG<sup>1</sup> (Molecular Probes) at 1:250. Following IHC, sections were incubated for 10 min with 5µg/ml wheat germ agglutinin (WGA)—Alexa 647 [Molecular Probes; for the determination of cross-sectional area (CSA)] in 1X PBS.

### Measurements of Hypertrophy in Cryogenic Heart Sections

For analysis of cardiomyocyte hypertrophy from cryogenic heart sections, measurements were taken from three hearts per genotype/age, with sampling from three cryosections per heart and three fields per section. The investigator performing the measurements and counts was blind to the genotype of all heart sections.

Hypertrophy was determined by measuring cardiomyocyte CSA on images of transversely cut heart cryosections labeled with anti-α-actinin, WGA-Alexa 647, and DAPI. Only images from LV and septum were imaged and quantified. Cells used in the analysis were confirmed as cardiomyocytes by the presence of α-actinin labeling. Manual tracing of the cardiomyocyte cross-sections was performed using the NIS-Elements 4.0 software which then calculated the CSA of individual cardiomyocytes. Ninety cardiomyocyte CSAs were measured per genotype/age. CSAs were averaged for each individual heart, and means were compared between genotypes within each age. All calculations were analyzed using Students' t-test.

### Measurements of Hypertrophy in Fixed, Dissociated Cardiomyocytes

In a method adapted from Mollova et al. (2013) with some modifications, hearts freshly isolated from PND2 and PND9 mice were rinsed in 1X PBS with 30 mM 2,3-butanedione monoxime (BDM, Sigma). Left ventricles (LV) + septums were isolated, minced and fixed in 2% formaldehyde for at least 45 min. After fixation, minced LV + septum were washed 3 × 5 min in 1X PBS, then incubated in 250µl (PND2) or 500µl (PND9) Type II collagenase (Gibco) at 37◦C for 1–4 h for digestion. After digestion, cells were gently triturated with a plastic transfer pipet to disperse cells. Fetal bovine serum (FBS, Hyclone) was added to the cell suspension to reach a final concentration of 10% FBS to quench collagenase activity. Sodium azide (Spectrum) was added for a final concentration of 0.03% to inhibit bacterial growth. A drop of cell suspension was placed on a labeled glass slide and covered with a coverslip that was then sealed with nail polish. Cells were then imaged in bright field at 40X as described in Supplementary Methods.

### Data Analyses

Values are reported as means ± SE. Statistical analyses were performed using SPSS (IBM) or SAS software version 9.3. For comparisons in which there were only two groups (WT vs. cMyBP-C−/−), Students' t-test was performed within each age group. For comparisons of three groups (WT, cMyBP-C+/−, cMyBP-C−/−), one-way ANOVAs were conducted with a posthoc Bonferroni correction to compare between genotypes at each time point. For analysis of cell size on fixed, isolated cardiomyocytes, a compound symmetry correlation structure was used to account for repeated measures from one heart (n = 50–100 cells/heart) and to account for any litter effect. Significance is reported at p < 0.05. All graphs were generated using GraphPad Prism.

### RESULTS

As described in the Methods Section, all comparisons were made using littermate offspring from heterozygous (cMyBP-C <sup>+</sup>/−) breedings to achieve more tightly controlled experimental conditions and reduce variance.

### cMyBPC-Null Hearts Demonstrate a Rapid Increase in Size and Loss of Function Over First 10 Days of Life

There was no evidence of gross morphological differences at PND1 (1 day after birth) between hearts from cMyBP-C <sup>−</sup>/−mice and those of WT littermates (**Figures 1A,B**). In sharp contrast, by PND9 there was a pronounced difference in the thickness of the interventricular septum and LVFWin cMyBP-C−/<sup>−</sup> hearts compared to the hearts of WT littermates (**Figures 1C,D**). The HW to BW ratio was significantly elevated in PND9 cMyBP-C−/<sup>−</sup> pups vs. WT [(%) 0.624 ± 0.015 vs. 0.501 ± 0.0062, respectively, ± SE, p < 0.001, n = 13 (cMyBP-C−/−), 32 (WT)] but was not different at PND1 [(%) 0.602 ± 0.007 vs. 0.619 ± 0.01, ± SE, p = 0.833, n = 18 (cMyBP-C−/−), 13 (WT)], which has been previously observed (de Lange et al., 2013). These morphologic changes at PND9 are consistent with the described hypertrophic phenotype of this mouse model (Harris et al., 2002; de Lange et al., 2013).

The gross morphology was supported by echocardiography, which revealed no difference at PND1 between WT and

FIGURE 1 | Morphologic and functional change in cMyBP-C−/<sup>−</sup> between PND 1 and PND9: Hematoxylin and Eosin (H&E) staining of representative sections from hearts of WT (A, C) and cMyBP-C−/<sup>−</sup> (B, D) mice at PND1 (A,B) and PND9 (C,D). Scale bar = 1 mm. (E–H) Representative M-mode echocardiographic images of WT (*n* = 6; E,G) and cMyBP-C−/<sup>−</sup> (*n* = 7; F,H) mice at PND1 (E,F) and PND9 (G,H). Note different scales for PND1 vs. PND9. Measurements listed in Table 1. PND1 images show increased echogenicity within LV cavity due to nucleated red blood cells that are absent by PND9.

cMyBP-C−/<sup>−</sup> hearts with regard to wall thickness and chamber diameter (**Figures 1E–H** and **Table 1**), except for a minimal but significantly lower left ventricular posterior wall thickness during systole (LVPWs) at PND1 in cMyBP-C−/<sup>−</sup> vs. WT hearts (**Table 1**). This difference was not present during diastole or in the anterior wall. At a functional level, cMyBP-C <sup>−</sup>/<sup>−</sup> mice showed no difference in ejection fraction at the same age. However, while ejection time appeared to be similar between WT and cMyBP-C−/<sup>−</sup> mice at PND1, isovolumic relaxation time (IVRT) was longer in cMyBP-C <sup>−</sup>/<sup>−</sup> pups, suggesting an early subtle functional deficit (**Table 1**) which was explored previously (de Lange et al., 2013).

In sharp contrast to the virtual absence of morphological differences at PND1, by PND9 the cMyBP-C−/<sup>−</sup> ventricle walls were considerably thicker than WT (**Figures 1E–H** and **Table 1**), consistent with the gross morphology. Measured during systole, the cMyBP-C−/<sup>−</sup> anterior and posterior left ventricular wall thicknesses (LVAW and LVPW) were increased 32% and 29%, respectively, compared to WT hearts. The left ventricular internal diameter during diastole (LVIDd) and systole (LVIDs) was also significantly larger in the cMyBP-C <sup>−</sup>/<sup>−</sup> mice at PND9 vs. WT, suggesting that dilation exists at this early stage of the disease. Furthermore, a functional deficit virtually absent at PND1 was readily apparent in the PND9 cMyBP-C−/<sup>−</sup> mice. The ejection fraction was considerably lower (43% of WT), and the isovolumic relaxation time significantly prolonged (188% of WT), with a shortened ejection time (69% of WT) in the PND9 cMyBP-C−/<sup>−</sup> hearts (**Table 1**).

### At PND1, Microarray Analysis of cMyBPC−/<sup>−</sup> Mice Reveals Upregulation of Genes Involved in Cell Cycling and Proliferation

Pairwise comparison using analysis software (iReport and Genesifter) of microarray data at PND1 identified many genes in the cell cycle pathway that were upregulated >1.5-fold (p < 0.05) in the cMyBP-C−/<sup>−</sup> hearts vs. (vs.) WT, indicating either an increase of cell cycling or a delay/persistence of developmental processes, or a combination of both. Genesifter analysis revealed that the highest scoring pathway in KEGG (The Kyoto Encyclopedia of Genes and Genomes) was the "cell cycle" pathway, which includes the differential expression of genes involved in several pathways linked to cell division, replication, and proliferation (**Table 2**).

Data from pairwise analysis of PND1, WT vs. cMyBP-C−/<sup>−</sup> hearts were further analyzed using Ingenuity Pathway Analysis (IPA, Ingenuity Systems). This analysis identified additional cell cycle pathway genes showing differential regulation, including but not limited to cyclin dependent kinase (Cdk), the anaphase-promoting complex (APC), importin α and β, and the extracellular-signal-regulated kinases (Erk1/2; Supplementary Figure 1, Supplementary Table 1). Taken together, the significant upregulation of genes involved in cell cycle, replication, and proliferation in cMyBP-C−/<sup>−</sup> mice compared to WT at PND1 suggests an early postnatal increase in cardiomyocyte cell cycling that precedes the hypertrophic phenotype.

### Microarray Analysis Reveals a Switch to Hypertrophic Signaling in cMyBP-C−/<sup>−</sup> vs. WT by PND9

Pairwise analysis of microarray data comparing WT and cMyBP-C <sup>−</sup>/<sup>−</sup> hearts at PND9 revealed that the KEGG "cell cycle" pathway was no longer upregulated in cMyBP-C−/<sup>−</sup> hearts. In addition, none of the genes implicated in cell division, replication, and proliferation that were identified at PND1 were differentially expressed at PND9. In the PND9 hearts, the top scoring KEGG pathway was "hypertrophic cardiomyopathy," which included eight upregulated genes and 1 (other than cMyBP-C itself) that was downregulated >1.5-fold at p = 0.05 (Supplementary Table 2). Supplementary Table 3 lists the most differentially regulated KEGG pathways in addition to "hypertrophic cardiomyopathy," between WT and cMyBP-C <sup>−</sup>/<sup>−</sup> mice at PND9, along with their corresponding genes, as identified by Genesifter analysis. Genes listed in pathways outside of "hypertrophic cardiomyopathy" are nonetheless directly implicated in HCM or other cardiovascular diseases (Weizmann Institute of Science MalaCards Human Malady Compendium and/or the University of Copenhagen DISEASES database). Hypertrophic pathways were neither activated nor


*Values are means* ±*SD. LVPW, left ventricular posterior wall thickness; d, diastole; mm, millimeters; s, systole; LVAW, left ventricular anterior wall thickness; LVID, left ventricular interior diameter; IVS, intraventricular septum; mg, milligrams; IVRT, isovolumetric relaxation time; ms, milliseconds; bpm, beats per minute. n = 6 (WT), n = 7 (cMyBP-C*−/−*).* \**p* < *0.05 vs. age-matched WT. PND9, but not PND1, mice were anesthetized with isoflurane.*

TABLE 2 | "Cell Cycle" is the Most Differentially Regulated Pathway Between cMyBP-C−/<sup>−</sup> and WT Hearts at PND1.


\**Ratio represents the gene expression of cMyBP-C*−/<sup>−</sup> *divided by gene expression of WT. Thus, a ratio of 1.88, as in Cyclin B1, is equivalent to an 88% increase in RNA expression in the cMyBP-C*−/<sup>−</sup> *hearts vs. WT.*

differentially regulated between WT and cMyBP-C−/<sup>−</sup> mice at PND1.

### Cardiomyocyte Cell Cycling is Increased in Cardiomyocytes from PND1 cMyBP-C−/<sup>−</sup> Mice

To confirm the upregulation of cell cycling noted at the RNA level in the microarray in PND1 cMyBP-C−/<sup>−</sup> mice, we performed IHC on heart sections from WT and cMyBP-C−/<sup>−</sup> mice. Coronal, paraffin-embedded heart sections were labeled with an anti-α-tropomyosin antibody, identifying cardiac, and skeletal muscle cells. Co-labeling using the Ki-67 antibody (**Figures 2A–D**) revealed a significantly higher number of cardiomyocytes actively participating in the cell cycle in the PND1 cMyBP-C−/<sup>−</sup> hearts compared to that of WT littermates (59 ±4 vs. 25 ±4 Ki67-positive cardiomyocytes per section, n = 3 hearts per group, ±SE, p < 0.05, **Figure 2I**), in agreement with the microarray data. At PND9, further corroborating the microarray data, there was no statistical difference in the number of Ki67-positive cardiomyocytes in cMyBP-C−/<sup>−</sup> vs. WT hearts (23 ±4 vs. 17 ±5 cells per section, ± SE, p = 0.38, n = 3 hearts per group, **Figure 2I**). These data support elevated cell cycling of early postnatal cardiomyocytes in cMyBP-C−/<sup>−</sup> hearts that is absent by PND9.

### Cardiomyocyte Hypertrophy Evident in PND9 But Not PND1 cMyBP-C−/<sup>−</sup> Hearts

To quantify the degree of hypertrophy of the cardiomyocytes at both PND1 and PND9, we determined cardiomyocyte size using cryopreserved WT and cMyBP-C−/<sup>−</sup> hearts. Hearts were cut in transverse sections and immunolabeled with WGA attached to the fluorophore Alexa 647 to visualize cell borders, antiα-actinin to confirm cardiomyocyte cell type, and DAPI to visualize nuclei (**Figures 2E–H**). We measured cardiomyocyte cross-sectional area in 90 α-actinin-positive cells per heart from

Ki-67-positive cardiomyocytes (I) and cardiomyocyte cross-sectional area (CSA; J) for WT and cMyBP-C−/<sup>−</sup> (-/-) at PND1 and PND9 in representative heart sections, as shown in (A-H). Means ±SE are reported, \**p* < 0.05.

multiple individual sections from WT (n = 3 hearts) and cMyBP-C <sup>−</sup>/<sup>−</sup> littermates (n = 3 hearts). Averaged cross-sectional areas of individual myocytes are depicted in **Figure 2J**. There was no indication of hypertrophy at PND1 in cMyBP-C−/<sup>−</sup> hearts, evidenced by similar cardiomyocyte mean cross-sectional area vs. WT (31.30 ± 3.19 vs. 31.76 ± 3.60µm<sup>2</sup> , respectively, ±SE, p = 0.93). In contrast, at PND9 cardiomyocytes from cMyBP-C−/<sup>−</sup> hearts were significantly hypertrophied compared to those from WT hearts, with a 30% higher mean cardiomyocyte cross-sectional area (110.30 ± 7.50 vs. 85.04 ± 4.35µm<sup>2</sup> , ±SE, p < 0.05, **Figure 2J**).

Since cardiac fibrosis is often found in HCM and might contribute to heart mass and size, we stained sections of PND1 and PND9 hearts using Masson's Trichrome and Elastic Trichrome (Supplementary Figure 2). No difference in interstitial fibrosis or disarray was observed between genotypes at either age.

### Timeline of Cell Cycling and Onset of Hypertrophy

The microarray and IHC data revealed that elevated cell cycling preceded the hypertrophic response, and that the enhanced cell cycling had normalized by PND9. These data raised the question of if differential hyperplasia precedes birth, and also how long it persists after PND1. The time course duration for cell cycling could have profound implications on the final cardiomyocyte number present in the cMyBP-C−/<sup>−</sup> hearts. Furthermore, the temporal relationship of potential hyperplastic vs. hypertrophic cellular signaling could lend insight as to a potential interaction between these two processes. Therefore, we expanded our study to include the investigation of cardiac morphology and gene expression at the added time points of embryonic day 18.5 (E18.5), PND0, and PND2 as well as at the previously established time points of PND1 and PND9, on an additional set of animals. Furthermore, we investigated heterozygote (cMyBP-C+/−) mice to determine whether any early, mild changes were present in these animals.

### Timeline of Gross Hypertrophy

To assess gross cardiac hypertrophy, animals and hearts were weighed at their respective time points and HW to BW ratios were calculated, as previously. There were no differences in pup weights among WT, cMyBP-C+/<sup>−</sup> or cMyBP-C−/<sup>−</sup> mice at any time point, suggesting that the cMyBP-C−/<sup>−</sup> mice are not globally growth retarded (**Figure 3**, top). HW to BW ratios of cMyBP-C+/<sup>−</sup> mice did not differ from WT at any time point. cMyBP-C−/<sup>−</sup> HW to BW ratios were not different than WT at E18.5, PND0, or PND1. However, cMyBP-C−/<sup>−</sup> pups had higher HW to BW ratios than their time-matched WT littermates at the remaining time points of PND2 (%, 0.61 ±0.01 vs. 0.57 ±0.01, ±SE, respectively) and PND9 (%, 0.64 ±0.02 vs. 0.51 ±0.01, respectively, ±SE, P < 0.001; **Figure 3**, bottom), suggesting that gross cardiac hypertrophy exists at the organ level starting at PND2 in the homozygote null animals (See Supplementary Table 4 for numbers of mice per genotype/day point).

### Timeline of Cellular Hypertrophy Using Genetic Markers

Since gross cardiac enlargement may or may not be the result of cellular hypertrophy, we evaluated the timing of the onset of cardiomyoctye hypertrophy by assessing expression of two genes widely recognized as hypertrophic markers, Nppa (atrial natriuretic peptide) and Myh7 (expressed as the

ratio of Myh7 [β-myosin heavy chain]/Myh6 [α-myosin heavy chain]), at the added time points starting before birth. **Figure 4** shows expression of Nppa (A) and Myh7/Myh6 (B) relative to the housekeeping genes Gapdh and ß-actin, as well as when normalized to WT levels (C, D), measured through targeted qPCR (Separated Myh6 and Myh7 expression is shown in Supplementary Figure 3). Expression of Nppa was higher in cMyBP-C−/<sup>−</sup> and cMyBP-C+/<sup>−</sup> vs. WT hearts at PND0, prior to the onset of gross cardiac hypertrophy in cMyBP-C <sup>−</sup>/<sup>−</sup> mice (see **Figure 3**, bottom), and was also higher at PND2 and PND9 in cMyBP-C−/<sup>−</sup> hearts (**Figures 4A,C**). Although there were trends toward elevated Nppa both prior to birth (E18.5) and at PND1 in cMyBP-C−/<sup>−</sup> hearts, they were not statistically significant. The early elevation of Nppa, prior to gross hypertrophy in the cMyBP-C−/<sup>−</sup> mouse, and in the absence of gross hypertrophy in the cMyBP-C <sup>+</sup>/<sup>−</sup> mouse, raises the question of whether Nppa is a true marker for hypertrophy or rather a non-specific response to stress.

cMyBP-C+/<sup>−</sup> hearts showed elevated Myh7/Myh6 expression at E18.5 but at no other time point. Expression of Myh7/Myh6 in cMyBP-C−/<sup>−</sup> hearts was only elevated at PND9 (**Figures 4B,D**), although HW to BW ratio was increased at PND2 (see **Figure 3**, bottom). The lack of an elevated Myh7/Myh6 ratio at PND2 suggests that the increase in HW to BW ratio at PND2 may not be due to cellular hypertrophy but could be the result of cardiomyocyte hyperplasia, resulting from the observed increase in cell cycling.

### Timeline of Cardiomyocyte Cell Cycling

We performed targeted qPCR at the expanded time points starting before birth to determine whether the enhanced postnatal cardiomyocyte cell cycling in cMyBP-C−/<sup>−</sup> hearts is a prolongation of the highly proliferative embryonic state or represents an independent postnatal increase. RNA expression was quantified for two cell cycling genes, cyclin A2 (Ccna2) and cell division cycle 25c (Cdc25c), both of which showed higher levels of expression in cMyBP-C−/<sup>−</sup> hearts at PND1 (**Table 2**) but no difference at PND9 in the microarray analysis. Targeted qPCR revealed that RNA expression levels of Ccna2 and Cdc25c were not different at E18.5 or PND0 between cMyBP-C−/<sup>−</sup> and WT hearts (**Figure 5**). However, at PND1, levels of expression of Ccna2 and Cdc25c were maintained at a higher level in cMyBP-C <sup>−</sup>/<sup>−</sup> hearts vs. WT (elevated 2.38 ± 0.39- and 2.23 ± 0.28-fold vs. WT, respectively). This elevated RNA expression continued through PND2 for Ccna2 (**Figures 5A,C**). Although the gene expression of Cdc25c trended toward being higher in cMyBP-C <sup>−</sup>/<sup>−</sup> vs. WT hearts at PND2, the difference was not significant at this time point. This elevated cell cycling in cMyBP-C−/<sup>−</sup> vs. WT hearts was also supported by increased IHC detection of Ki-67 in cMyBP-C−/<sup>−</sup> mice (Supplementary Figures 4A,B,E).

The cMyBP-C+/<sup>−</sup> hearts showed no difference in Ccna2 expression at any time point, and showed elevated Cdc25c expression compared to WT only at PND2. However, there was a trend toward upregulation for Cdc25c at PND1 and PND2, and for Ccna2 at PND1. At PND9, RNA expression in the cMyBP-C <sup>−</sup>/<sup>−</sup> and cMyBP-C+/<sup>−</sup> hearts was not different from WT for either of the cell cycling genes. Thus, elevated expression of cell cycling genes in cMyBP-C−/<sup>−</sup> hearts appears to reflect a prolongation of physiologic neonatal cell cycling, while cMyBP-C <sup>+</sup>/<sup>−</sup> hearts show a trend, though mostly non-significant, in the same direction as cMyBP-C−/<sup>−</sup> hearts.

Increased cell cycling in the cMyBP-C−/<sup>−</sup> cardiomyocytes does not necessarily indicate increased cardiomyocyte proliferation. However, unless cellular hypertrophy is present at PND2, the increased HW to BW ratio at PND2 would only be reasonably explained by increased cardiomyocyte proliferation. The absence of an elevated Myh7/Myh6 expression at PND2 raises doubt as to if cellular hypertrophy exists at PND2 in the cMyBP-C−/<sup>−</sup> heart.

### Assessment of Cardiomyocyte Size at PND2 and PND9

To confirm whether or not cardiomyocyte hypertrophy is present in the PND2 cMyBP-C−/<sup>−</sup> LV to contribute to the increased HW to BW ratio, we directly assessed cardiomyocyte size using a method adapted with modifications from Mollova et al. (2013). Freshly excised hearts were fixed in situ with formaldehyde prior to collagenase digestion, to yield striated, rod-shaped cardiomyocytes (**Figure 6A**). Cells were then imaged and measured for length and width. Cardiomyocyte area was calculated as the product of length × width. At PND2, cMyBP-C+/<sup>−</sup> cardiomyocytes were slightly longer in length compared to the cMyBP-C−/<sup>−</sup> cardiomyocytes, but were not different from WT (**Figure 6B** and Supplementary Table 5).

Cardiomyocytes from cMyBP-C−/<sup>−</sup> LVs were not different from WT cardiomyocytes in length, width, or area, demonstrating that cardiomyocyte hypertrophy is not responsible for the increased HW to BW ratio in cMyBP-C−/<sup>−</sup> mice at PND2. The lack of cardiomyocyte hypertrophy at PND2 was further supported by measurements of cardiomyocyte CSA in IHC (Supplementary Figures 4C,D,F).

At PND9, cardiomyocytes from cMyBP-C−/<sup>−</sup> LVs were longer than WT and cMyBP-C+/<sup>−</sup> cardiomyocytes, leading to an increased cell area compared to WT (**Figure 6B**, Supplementary Table 5). No changes in cell width were observed. Thus, at PND9, cardiomyocyte hypertrophy is at least partly responsible for the increased HW to BW ratio in cMyBP-C−/<sup>−</sup> hearts, as expected.

These data indirectly suggest that the increased cell cycling observed by microarray (**Table 2**, Supplementary Figure 1) RNA expression (**Figure 5**), and IHC (**Figure 2**, Supplementary Figures 4A,B,E) does in fact lead to cardiomyocyte proliferation, as further suggested by increased expression of genes involved in the cytokinetic processes of contractile ring assembly or in abcission (Supplementary Table 6). This proliferative response is substantial enough to be wholly responsible for the increased HW to BW ratio observed at PND2 in the cMyBP-C−/<sup>−</sup> cardiomyocytes, as cardiomyocyte hypertrophy is not present at PND2. By PND9, cardiomyocyte hypertrophy is present in the cMyBP-C−/<sup>−</sup> but not cMyBP-C+/<sup>−</sup> LVs, as indicated by cardiomyocyte area (**Figure 6**).

### DISCUSSION

We conclude from the data presented here that in an HCM model in which mice lack cMyBP-C there is a prolongation of postnatal cell cycling which results in increased cardiomyocyte proliferation, prior to activation of the hypertrophic response. These data further reveal that cellular signaling in the heterozygote (cMyBP-C+/−) heart trends in the same direction as the cMyBP-C−/<sup>−</sup> heart, but fails to elicit a hypertrophic phenotype by PND9. At 6 months of age, however, the cMyBP-C+/<sup>−</sup> mice show evidence of diastolic dysfunction, altered contractile kinetics, and pro-arrhythmic changes in their action potential, while remaining free of overt hypertrophy (Cheng et al., 2013). It is unknown whether the cMyBP-C+/<sup>−</sup> mice develop hypertrophy at a more advanced age.

We previously reported that newborn cMyBP-C−/<sup>−</sup> mice are grossly indistinguishable from their WT littermates, but by PND9 the cMyBP-C−/<sup>−</sup> hearts are significantly enlarged (de Lange et al., 2013). This response has been assumed to be purely hypertrophic. In performing a transcriptome comparison of newborn and PND9 mice from cMyBP-C−/<sup>−</sup> and WT pups, we expected to identify early hypertrophic signaling events. Unexpectedly, the data presented here, while confirming that germline ablation of cMyBP-C−/<sup>−</sup> does not cause any apparent overt pathology during cardiogenesis, does however indicate a prolonged phase of increased cell cycling and proliferation postnatally that precedes the significant cellular hypertrophic response apparent by PND9. While we recognize that counting cardiomyocyte number is the most direct method to assess proliferation, it is technically unfeasible in our experimental design. The pups in this study are obtained from heterozygote × heterozygote breedings, such that all litter mates are processed in parallel, with genotypes unknown at the time of harvest. Additionally, recovery and counting of cardiomyocytes from PND2 hearts proved highly unreliable and irreproducible. Therefore, we relied upon several indirect measures to differentiate a proliferative vs. hypertrophic response, including increased HW to BW ratio in the PND2 cMyBP-C−/<sup>−</sup> pups (n = 25) vs. WT (n = 24) in the absence of cardiomyocyte hypertrophy, increased cardiomyocyte Ki-67 positivity, and direct measure of cardiomyocyte areas.

### The Potential Contribution of Cardiomyocyte Proliferation to Heart Size

Under normal processes of mammalian development, the fetal heart increases its size primarily through cell division, but transitions shortly after birth to physiologic hypertrophic growth to correlate heart size with somatic growth (Walsh et al., 2010; Porrello et al., 2011). In the context of HCM, elevated and persistent proliferation may have a significant impact on eventual disease severity. Enhanced early cardiomyocyte cell cycling at PND1 and PND2 in the cMyBP-C−/<sup>−</sup> hearts that is significant enough to lead to an increase in HW to BW ratio in the absence of cardiomyocyte hypertrophy at PND2, as observed here, could provide an additional pool of myocytes that later participate in the hypertrophic response. The impact of prolonged cell division on total cardiomyocyte number may be substantial. Between PND1 and PND4 in normal hearts of WT mice, cardiomyocyte population increases by as much as 40% (Naqvi et al., 2014). Any additional increase and/or prolongation of this exponential proliferative capacity in the neonate would be expected to profoundly influence total cardiomyocyte number. Whether or not the early proliferative responses observed in our murine model occur in human myocardium to exert an as yet unappreciated influence in disease manifestation remains to be determined. However, there are data suggesting that the presence and importance of cardiomyocyte proliferation early in life extends to humans, and that cardiomyocyte proliferation before 20 years of age contributes to heart size (Mollova et al., 2013). The question of if the degree of proliferation is altered in HCM in humans remains unanswered.

### The Perinatal Period May Provide a Unique Opportunity for Modifying Disease Course

Evidence derived from other models of HCM further support the potential importance of this early pre-hypertrophic period in determining the course toward overt HCM (Jiang et al., 2013; Cannon et al., 2015). Silencing an HCM-causing Arg403Gln Myh6 mutation prior to hypertrophy prevented the development of hypertrophy, while silencing the mutation after hypertrophy was already apparent did not lead to a reversal of the phenotype (Jiang et al., 2013). When the same mutation is suppressed from conception through 6 weeks of age, the degree of hypertrophy and fibrosis at 40 weeks was significantly attenuated compared to non-suppressed controls. When suppression occurred after 6 weeks of age, the mice developed severe hypertrophy (Cannon et al., 2015). Additionally, cardiac-specific ablation of cMyBP-C during adulthood results in significantly less cardiac hypertrophy compared to mice with germline ablation of cMyBP-C (Chen et al., 2012). These data suggest that the proliferative capacity of the neonatal cardiomyocyte provides a unique, potentially modifiable mechanism that contributes to HCM in a manner that is unparalleled later in life.

### Cell Proliferation and Cell Hypertrophy: Distinct or Interrelated Pathways?

The transition that we observe in the cMyBP-C−/<sup>−</sup> hearts from an early elevation in cardiomyocyte proliferation, associated with an increased HW to BW ratio, to a later cardiomyocyte hypertrophy raises the question of whether the proliferative and hypertrophic pathways are distinct from each other or interrelated. The phenomenon of increased cardiomyocyte proliferation prior to cellular hypertrophy was recently reported to also occur in non-sarcomere etiologies of pathologic cardiac hypertrophy (Kim et al., 2014; Schipke et al., 2015). Enhanced cardiomyocyte proliferation in the first 2 weeks of life was found in both Prkag2 cardiomyopathy, in which an adenosine monophosphate (AMP)-activated protein kinase mutation causes abnormal glycogen storage (Kim et al., 2014), as well as in mice with a deletion of guanylyl cyclase A (GC-A) that proceed to develop cardiac hypertrophy (Schipke et al., 2015). In both of these models, mice exhibited a prolonged phase of postnatal cardiomyocyte proliferation that occurred prior to the ensuing cellular hypertrophy, but coincident with an increased HW to BW ratio, echoing the results we obtained with the cMyBP-C−/<sup>−</sup> mice. The authors concluded, as do we, that the postnatal elevation in proliferation contributed to the hypertrophic phenotype. The study presented here extends the observation of enhanced hyperplasia preceding cardiomyocyte hypertrophy to the disease of HCM and proposes that communication between the two mechanisms facilitates HCM progression. Together, these studies support a more generalizable role for a cardiac hyperplastic response to gene mutations associated with the later development of hypertrophy. These data imply that variations in the increased pool of cells available to participate in the hypertrophic process may influence the eventual severity of the phenotype, a process further modified by the specific HCM mutation and/or the genetic background of the individual.

While the relevance of an interaction between hyperplastic and hypertrophic cellular response pathways remains to be fully explored, the implication that underlying pathologic conditions can influence the timing of exit of cardiomyocytes from the cell cycle is intriguing. In chondroblasts, the hypertrophic phase of growth is unable to initiate until completion of the proliferative phase, suggesting either cross-talk between the pathways or an absolute requirement for exit from the cell cycle before hypertrophy can proceed (Omelyanenko et al., 2013). The overlap between many of the proteins involved in both of these processes implies additional nuance in their regulation.

### Heterogeneity in HCM

Recently, Jiang et al. (2015) reported the early postnatal phenotype of mice harboring a mutation that prematurely truncates cMyBP-C at amino acid 1,064 of 1,270 (cMyBP-Ct/<sup>t</sup> ). Although both the cMyBP-C−/<sup>−</sup> and cMyBP-Ct/<sup>t</sup> exhibit an early postnatal hyperplastic response, resulting in a HW to BW ratio increase independent of cellular hypertrophy, the two mouse models also show some important divergent findings. While the cMyBP-C−/<sup>−</sup> mouse in our study showed cardiomyocyte hypertrophy following the early elevated proliferation, the cMyBP-Ct/<sup>t</sup> mouse exhibited a purely hyperplastic response without any subsequent cardiomyocyte hypertrophy. The resulting phenotype is a dilated cardiomyopathy, as opposed to HCM. Furthermore, Jiang et al. suggest that in contrast to the homozygote cMyBP-Ct/<sup>t</sup> mouse, the hearts of individual human patients heterozygous for cMyBP-C mutations are purely hypertrophic, and the difference in allelic expression between the cMyBP-Ct/<sup>t</sup> and heterozygous patients results in divergent cellular pathways both leading to cardiomyopathy (Jiang et al., 2015). In contrast, mice in our study heterozygous for the null mutation (cMyBP-C+/−) appear to have an early intermediate expression between WT and cMyBP-C−/<sup>−</sup> for genes involved in cell cycling and stress response, which then return to normal by PND9. These heterozygous mice ultimately fail to show a cardiomyopathic phenotype by PND9, but the trend in gene expression in the same direction as the cMyBP-C <sup>−</sup>/<sup>−</sup> mice suggests similar, not divergent cellular signaling between cMyBP-C−/<sup>−</sup> and cMyBP-C+/−. We propose that the difference in the cMyBP-C−/<sup>−</sup> vs. cMyBP-Ct/<sup>t</sup> phenotypes may be a product of differences in expressed protein. Although the cMyBP-Ct/<sup>t</sup> mouse has been regarded as equivalent to a cMyBP-C null animal (Sadayappan et al., 2006; Tanner et al., 2014; Jiang et al., 2015), initial reports of this model show expressed protein in myofibrillar preparations, albeit at low levels (McConnell et al., 1999). We hypothesize that small amounts of expressed, truncated cMyBP-C might trigger different cellular signaling than a pure absence of the protein. We also postulate that heterogeneity of disease severity observed in the human population might be influenced by differences in mutations which produce variably reduced amounts of abnormal protein vs. pure protein reduction. Understanding the pathways to HCM manifestation, and how the cardiomyocyte transition from hyperplasia to hypertrophy influences this course may provide novel insight into this enigmatic disease and allow development of more effective preventative treatments for families known to harbor pathologic mutations.

### CONCLUSION

A murine HCM model in which cMyBP-C is genetically ablated presents shortly after birth with elevated cardiomyocyte hyperplasia before onset of pathologic cellular hypertrophy evident by PND9. This work implicates cell cycle pathway dysregulation and cardiomyocyte proliferation as early responses that precede the cellular hypertrophic response, and may have implications for the severity of hypertrophy. These data and those from other model systems suggest a relationship between cell

### REFERENCES


cycle and hypertrophic signaling pathways occurring during a critical time period in HCM disease pathogenesis. Awareness of the potential for interplay between cell cycling/proliferation and the hypertrophic signaling pathways provides a new opportunity to expand our understanding of cardiac physiology in both health and disease.

### AUTHOR CONTRIBUTIONS

EF was involved in study design, data collection and analysis, interpretation, writing, and editing. AG was involved in study design, data collection and analysis, writing, interpretation, and editing. AA was involved in data collection and analysis, interpretation, and editing. WdL was involved in study design, data collection and analysis, interpretation, writing and editing. JR was involved in study design, interpretation, writing, and editing.

### FUNDING

This work was supported by R01HL107367-01 (JR).

### ACKNOWLEDGMENTS

The authors would like to thank Tim Hacker, director of the Cardiovascular Physiology Core Facility at the University of Wisconsin, for echocardiography services, as well as Ella Ward and Joseph Hardin at the UW Carbone Cancer Center, Madison, WI, for preparing cryosections (Ella Ward) and for expert advice on IHC (Joe Hardin).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2017.00414/full#supplementary-material


cardiac myosin binding protein-C knockout mice. Circ. Res. 90, 594–601. doi: 10.1161/01.RES.0000012222.70819.64


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Farrell, Grimes, de Lange, Armstrong and Ralphe. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Nebivolol Desensitizes Myofilaments of a Hypertrophic Cardiomyopathy Mouse Model

Sabrina Stücker 1, 2, Nico Kresin1, 2, Lucie Carrier 1, 2 and Felix W. Friedrich1, 2 \*

<sup>1</sup> Department of Experimental Pharmacology and Toxicology, Cardiovascular Research Center, University Medical Center Hamburg-Eppendorf, Hamburg, Germany, <sup>2</sup> German Centre for Cardiovascular Research (DZHK), Hamburg, Germany

Background: Hypertrophic cardiomyopathy (HCM) patients often present with diastolic dysfunction and a normal to supranormal systolic function. To counteract this hypercontractility, guideline therapies advocate treatment with beta-adrenoceptor and Ca2<sup>+</sup> channel blockers. One well established pathomechanism for the hypercontractile phenotype frequently observed in HCM patients and several HCM mouse models is an increased myofilament Ca2<sup>+</sup> sensitivity. Nebivolol, a commonly used beta-adrenoceptor antagonist, has been reported to lower maximal force development and myofilament Ca2<sup>+</sup> sensitivity in rabbit and human heart tissues. The aim of this study was to evaluate the effect of nebivolol in cardiac muscle strips of an established HCM Mybpc3 mouse model. Furthermore, we investigated actions of nebivolol and epigallocatechin-gallate, which has been shown to desensitize myofilaments for Ca2<sup>+</sup> in mouse and human HCM models, in cardiac strips of HCM patients with a mutation in the most frequently mutated HCM gene MYBPC3.

#### Edited by:

P. Bryant Chase, Florida State University, United States

#### Reviewed by:

Paul M. L. Janssen, Ohio State University Columbus, United States Adriano Martins, Maimonides Medical Center, United States

> \*Correspondence: Felix W. Friedrich f.friedrich@uke.de

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 04 May 2017 Accepted: 17 July 2017 Published: 02 August 2017

#### Citation:

Stücker S, Kresin N, Carrier L and Friedrich FW (2017) Nebivolol Desensitizes Myofilaments of a Hypertrophic Cardiomyopathy Mouse Model. Front. Physiol. 8:558. doi: 10.3389/fphys.2017.00558 Methods and Results: Nebivolol effects were tested on contractile parameters and force-Ca2<sup>+</sup> relationship of skinned ventricular muscle strips isolated from Mybpc3-targeted knock-in (KI), wild-type (WT) mice and cardiac strips of three HCM patients with MYBPC3 mutations. At baseline, KI strips showed no difference in maximal force development compared to WT mouse heart strips. Neither 1 nor 10 µM nebivolol had an effect on maximal force development in both genotypes. 10 µM nebivolol induced myofilament Ca2<sup>+</sup> desensitization in WT strips and to a greater extent in KI strips. Neither 1 nor 10 µM nebivolol had an effect on Ca2<sup>+</sup> sensitivity in cardiac muscle strips of three HCM patients with MYBPC3 mutations, whereas epigallocatechin-gallate induced a right shift in the force-Ca2<sup>+</sup> curve.

Conclusion: Nebivolol induced a myofilament Ca2<sup>+</sup> desensitization in both WT and KI strips, which was more pronounced in KI muscle strips. In human cardiac muscle strips of three HCM patients nebivolol had no effect on myofilament Ca2<sup>+</sup> sensitivity.

Keywords: nebivolol, myofilament, Ca2<sup>+</sup> sensitivity, hypertrophic cardiomyopathy, Mybpc3, mouse, human, epigallocatechin-3-gallate

## INTRODUCTION

Hypertrophic cardiomyopathy (HCM) is the most frequent cardiac genetic disease primarily caused by mutations in sarcomeric protein genes (Friedrich and Carrier, 2012; Maron et al., 2014; Ho et al., 2015). The most commonly mutated genes are MYBPC3 (encoding cardiac myosin-binding protein C) and MYH7 (encoding β-myosin-heavy chain) (Walsh et al., 2017). HCM is principally characterized by asymmetric left ventricular hypertrophy, diastolic dysfunction and myocardial disarray (Elliott et al., 2008). Current pharmacological treatment of HCM mainly relies on beta-adrenoceptor (AR) and Ca2<sup>+</sup> channel blockers, which improve clinical symptoms, partially prevent arrhythmias and improve diastolic dysfunction by prolonging left ventricular (LV) filling time and reducing outflow tract obstruction (Maron et al., 2003; Gersh et al., 2011; Spoladore et al., 2012; Hamada et al., 2014; Tardiff et al., 2015). Increased Ca2<sup>+</sup> sensitivity seems to be a common factor in HCM as seen in animal HCM models (Tardiff et al., 1999; Cazorla et al., 2006; Pohlmann et al., 2007; Vignier et al., 2009; Fraysse et al., 2012; Barefield et al., 2014; Wijnker et al., 2016), and human HCM samples (Jacques et al., 2008; van Dijk et al., 2009, 2012). The increased Ca2<sup>+</sup> response may contribute to diastolic dysfunction and arrhythmias (Morimoto et al., 1998; Baudenbacher et al., 2008). Even though the mechanisms accountable for increased myofilament Ca <sup>2</sup><sup>+</sup> sensitivity remain unclear, targeting this pathomechanism by interventions decreasing myofilament Ca2<sup>+</sup> sensitivity may be an attractive alternative for the treatment of HCM and improvement in symptoms (Jagatheesan et al., 2007; Alves et al., 2014; Tardiff et al., 2015). Among beta-AR blockers that are commonly used in the treatment of cardiovascular diseases, nebivolol has been reported to lower maximal force development and to desensitize rabbit and human cardiac myofilaments to Ca2<sup>+</sup> (Zeitz et al., 2000; Janssen et al., 2001). However, the effects of nebivolol were never evaluated in HCM models with increased myofilament Ca2<sup>+</sup> sensitivity. An established HCM mouse model carrying the human c.772G>A MYBPC3 mutation is the Mybpc3 KI mouse model (Vignier et al., 2009). This mutation was frequently found in unrelated HCM patients in Tuscany and is associated with a bad prognosis (Richard et al., 2003; Girolami et al., 2006; Ho et al., 2015). At the homozygous state, this mouse model exhibits HCMlike features such as left ventricular hypertrophy, diastolic dysfunction and increased myofilament Ca2<sup>+</sup> sensitivity (Vignier et al., 2009; Fraysse et al., 2012). We recently showed that epigallocatechin-3-gallate (EGCg), a major component of green tea, hastened relaxation and Ca2<sup>+</sup> transient in KI cardiomyocytes and decreased Ca2<sup>+</sup> sensitivity of KI myofilaments (Friedrich et al., 2016). In this study, we investigated nebivolol effects on myofilament Ca <sup>2</sup><sup>+</sup> sensitivity in Mybpc3 KI cardiac muscle strips. We furthermore assessed nebivolol and EGCg effects in cardiac strips of three HCM patients with MYBPC3 mutations.

## MATERIALS AND METHODS

### Human Samples

Human myocardial samples were obtained from three HCM patients carrying heterozygous MYBPC3 mutations (c.1960C>T, c.2308G>A, c.2234A>G) who underwent septal myectomy due to outflow tract obstruction. The material was taken with written informed consent of the donor and with written approval of the local ethical boards. The study has been carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki).

### Animals

The Mybpc3 KI cardiomyopathy mouse model was generated by the targeted insertion of a G>A transition on the last nucleotide of exon 6 (Vignier et al., 2009; Fraysse et al., 2012; Schlossarek et al., 2012, 2014; Gedicke-Hornung et al., 2013; Mearini et al., 2013, 2014; Stohr et al., 2013; Friedrich et al., 2014; Najafi et al., 2015; Thottakara et al., 2015; Flenner et al., 2016, 2017). Mice were maintained on the C57 background. As controls, Mybpc3 WT mice of the same background were used. The study was exerted in accordance with the recommendations of the guide for the care and use of laboratory animals published by the NIH (Publication No. 85–23, revised 2011 published by National Research Council) and comply with the ARRIVE guidelines (http://www.nc3rs.org.uk/arrive-animal-research-reporting-

vivo-experiments). All experimental procedures were in accordance with the German Law for the Protection of Animals and the protocol was approved by the Ministry of Science and Public Health of the City State of Hamburg, Germany (Org 653).

### Skinned Ventricular Trabeculae Force Measurements

For the determination of force-Ca2<sup>+</sup> relationships, trabeculae were prepared from ventricular endocardial surface of WT and KI mice or human myocardium of a septal myectomy (Flenner et al., 2016; Friedrich et al., 2016). Dimensions of strips were 2.91 ± 0.14 mm in length, 0.36 ± 0.01 mm in width and 0.11 ± 0.01 mm<sup>2</sup> in cross-sectional area (CSA), calculated by 2πr 2 assuming a circular shape, nWT = 17, nKI = 18, nhuman = 57. Strips were permeabilized in relaxing solution (pCa 9) in EGTAbuffer (5.89 mM Na2ATP, 14.5 mM CrP, 6.48 mM MgCl2, 40.76 mM Kprop, 100 mM BES and 7 mM EGTA, pH 7.1) (Kooij et al., 2010; Stoehr et al., 2014) containing 1% Triton X-100 at 4 ◦C for 18 h. The next day strips were either directly used for measurements or stored at −20◦C in a 50% glycerol/relaxing solution containing protease inhibitors (EDTA-free, complete tablets, mini, Roche). The Ca2+-sensitivity of permeabilized cardiac strips was evaluated using a fiber test system (1400A; Aurora Scientific) by mounting them between a force transducer and a length controller. Strips were stretched above slack length until they developed force in activating solution (pCa 4.5) at 15◦C. For contraction-relaxation cycles strips were kept in pCa 9 to achieve full relaxation. Then they were moved to pCa 4.5 until maximal force development was reached. Maximal force was related to cross-sectional area (mN/mm<sup>2</sup> ). For force-Ca2+-curves they were exposed to increasing Ca2<sup>+</sup> concentrations from pCa 9 to pCa 4.5 in EGTA-buffer. Force development was measured in each pCa solution. Measurements were repeated in the presence of 1 or 10 µM nebivolol (nebivolol hydrochloride, Sigma Life Sciences) or 30 µM epigallocatechingallate (Sigma Life Sciences) after 5 min preincubation in relaxing solution (Flenner et al., 2016; Friedrich et al., 2016). In every second measurement, nebivolol was tested first and a control measurement was performed 5 min after nebivolol washout to exclude time-dependent force rundown. Each strip was measured in a pairwise manner (paired analysis baseline vs. intervention) serving as its own control. Data were analyzed using the Hill equation (Hill et al., 1980), with pCa<sup>50</sup> as the free Ca2<sup>+</sup> concentration which yields 50% of the maximal force and nH representing the Hill coefficient. The pCa<sup>50</sup> represents the measure of myofilament Ca2<sup>+</sup> sensitivity.

### Statistical Analysis

Data were expressed as mean ± SEM. Comparisons were performed by paired or unpaired Student's t-test and with oneway ANOVA, followed by Bonferroni's post-test as indicated in the figure legends. Concentration response curves were fitted to the data points and force-pCa relationship comparison was done by using extra sum-of-squares F-test (GraphPad, Prism 6). A value of P < 0.05 was considered statistically significant.

### RESULTS

### Nebivolol (1 and 10 µM) Has No Effect on Maximal Force Development in Permeabilized Cardiac Strips of Mybpc3 WT and KI Mice

The hypercontractile phenotype observed in HCM patients could be attributed to an increased myofilament Ca2<sup>+</sup> sensitivity. Since nebivolol has been reported to lower maximal force development and to desensitize rabbit and human cardiac myofilaments (Zeitz et al., 2000; Janssen et al., 2001), we investigated nebivolol effects on myofilament Ca2<sup>+</sup> sensitivity in cardiac muscle strips of Mybpc3 KI mice with an increased myofilament Ca2<sup>+</sup> sensitivity (Vignier et al., 2009; Fraysse et al., 2012; Flenner et al., 2016; Friedrich et al., 2016). There are conflicting reports concerning the effects of nebivolol on maximal force development and myofilament Ca2<sup>+</sup> sensitivity in cardiac muscle strips. Whereas Zeitz et al. reported that 1 µM nebivolol lowered maximal force development and myofilament Ca2<sup>+</sup> sensitivity in skinned trabeculae (Zeitz et al., 2000), Bundkirchen and colleagues did not observe such an effect at 10 µM (Bundkirchen et al., 2001). We therefore used 1 and 10 µM for our experiments. To investigate nebivolol effects on force development we measured contraction-relaxation cycles in skinned myofilaments (**Figure 1A**). Analysis showed that maximal force development in baseline conditions did not significantly differ between WT and KI muscle strips (**Figure 1B**). Neither 1 nor 10 µM nebivolol had an effect on maximal force development (**Figure 1B**).

### Nebivolol Decreases Myofilament Ca2<sup>+</sup> Sensitivity to a Greater Extent in KI than in WT Skinned Ventricular Trabeculae

Nebivolol has been reported to decrease Ca2<sup>+</sup> sensitivity in rabbit and human cardiac myofilaments (Zeitz et al., 2000; Janssen et al., 2001). To assess whether this is also the case in myofilaments of Mybpc3 KI mouse hearts, we measured force-pCa relationships in skinned ventricular trabeculae from WT and KI mice. In analogy to the experiments on maximal force development, we performed force-pCa relationships in the absence and presence of 1 and 10 µM nebivolol, respectively. As observed before (Fraysse et al., 2012; Flenner et al., 2016; Friedrich et al., 2016), skinned KI trabeculae showed a higher pCa<sup>50</sup> than WT trabeculae in baseline conditions, representing higher myofilament Ca2<sup>+</sup> sensitivity (**Figures 2A,B**). In WT

half-maximal activation ±nebivolol 1 and 10 µM. (C) Delta of pCa<sup>50</sup> ± nebivolol 1 and 10 µM. (D) nHill coefficient (Hill slope) ± nebivolol 1 and 10 µM. \*\*P < 0.01 vs. WT in the same condition and \$p < 0.05 vs. KI 1 µM, unpaired Student's t-test; ##P < 0.01 and ###P < 0.001 vs. baseline, paired Student's t-test, concentration response curves were fitted to the data points and curve comparison was done by using extra sum-of-squares F-test; number of strips is indicated in the bars.

strips, only incubation with 10 µM (by extra sum-of-squares F-test) shifted the force-Ca2<sup>+</sup> relationship to the right resulting in a lower pCa50, whereas in KI strips both 1 and 10 µM lowered pCa<sup>50</sup> (**Figures 2A,B**) indicating myofilament Ca2<sup>+</sup> desensitization. This effect was concentration-dependent since incubation with 10 µM nebivolol induced a stronger shift (1 pCa50) to the right (**Figure 2C**). The nHill coefficient (Hill slope) as a an index for myofilament co-operativity did not differ between the genotypes neither with nor without nebivolol (**Figure 2D**).

### Nebivolol Does Not Impact on Maximal Force Development or Myofilament Ca2<sup>+</sup> Sensitivity in Muscle Strips Derived from Cardiac Tissue of HCM Patients with MYBPC3 Mutations

Since nebivolol has been reported to desensitize human myofilaments for Ca2<sup>+</sup> and since both 1 and 10 µM nebivolol had induced a right-ward shift of the force-pCa curves in Mybpc3 KI cardiac muscle strips we sought to investigate whether it would also affect myofilament Ca2<sup>+</sup> sensitivity in human HCM tissue. Similar to the experiments performed with mouse cardiac strips we investigated the effects of 1 and 10 µM nebivolol on contraction-relaxation cycles in muscle strips of three HCM patients carrying different MYBPC3 mutations. Neither 1 nor 10 µM nebivolol had an influence on Fmax (**Figure 3A**). Furthermore, no shift in Ca2<sup>+</sup>

sensitivity nor change in Hill slope was observed for the Ca2+-dependent force development (**Figures 3B–D**). In contrast and as reported before by us in Mybpc3 WT and KI strips (Friedrich et al., 2016), incubation with a positive control compound (30 µM epigallocatechin-gallate; EGCg) shifted the force-Ca2+-relationship to the right in strips from the same three HCM patients (**Figures 3E,F**).

relationship ±nebivolol 1 and 10 µM. (C) pCa<sup>50</sup> representing the negative logarithm of the calcium concentration needed for half-maximal activation ±nebivolol 1 and <sup>10</sup> <sup>µ</sup>M. (D) nHill coefficient (Hill slope) <sup>±</sup>nebivolol 1 and 10 <sup>µ</sup>M. (E) Force-Ca2<sup>+</sup> relationship <sup>±</sup>EGCg 30 <sup>µ</sup>M. (F) pCa<sup>50</sup> <sup>±</sup>EGCg 30 <sup>µ</sup>M. \*\*\*<sup>P</sup> <sup>&</sup>lt; 0.001 vs. baseline, paired Student's t-test. Concentration response curves were fitted to the data points and curve comparison was done by using extra sum-of-squares F-test; number of strips is indicated in the bars.

### DISCUSSION

HCM patients often present with a normal to supranormal systolic function and diastolic dysfunction. To counteract this hypercontractility, guideline therapies advocate treatment with beta-AR and Ca2<sup>+</sup> channel blockers. One well established pathomechanism for the hypercontractile phenotype frequently observed in HCM patients and several HCM mouse models is an increased myofilament Ca2<sup>+</sup> sensitivity (Morimoto et al., 1998; Robinson et al., 2007; Huke and Knollmann, 2010; Kimura, 2010; Fraysse et al., 2012; Moore et al., 2012; van Dijk et al., 2012; Barefield et al., 2014; Elliott et al., 2014; Flenner et al., 2016; Friedrich et al., 2016). Nebivolol, a commonly used beta-AR antagonist, has been reported to lower maximal force development and myofilament Ca2<sup>+</sup> sensitivity in rabbit and human heart tissues (Zeitz et al., 2000; Janssen et al., 2001). Given the hypercontractile phenotype mentioned above, these pleiotropic actions predestine it for HCM treatment. The aim of this study was to evaluate whether nebivolol would exert similar effects in permeabilized myofilaments of an Mybpc3 HCM mouse model and of HCM patients with mutations in the most frequently mutated gene MYBPC3. The main findings of this study were: (1) At baseline, permeabilized left ventricular trabeculae isolated from Mybpc3 KI mouse hearts showed no difference in maximal force development compared to WT mouse heart strips. (2) Neither 1 nor 10 µM nebivolol had an effect on maximal force development in both genotypes. (3) 10 µM nebivolol induced myofilament Ca2<sup>+</sup> desensitization in both WT and KI strips and this effect was more pronounced in KI muscle strips, respectively. (4) Nebivolol had no effect on Ca2<sup>+</sup> sensitivity in cardiac muscle strips of three HCM patients with MYBPC3 mutations, whereas 30 µM of EGCg induced a right shift in the force-Ca2<sup>+</sup> curve.

In mice, nebivolol did not influence maximal force development. On the other hand, it affected myofilament Ca2<sup>+</sup> sensitivity in mouse strips. The mechanism behind this is unknown so far. In analogy to the mouse results, 1 and 10 µM nebivolol had no effect on maximal force development in human tissues. In contrast to the observations made in mouse strips, it did not impact on myofilament Ca2<sup>+</sup> sensitivity.

The reason why nebivolol exerted a myofilament Ca2<sup>+</sup> desensitizing effect in KI strips at both 1 and 10 µM, whereas in WT strips only 10 µM had an effect and no effect at all in human HCM tissues remains unclear. As mentioned before, discrepancies concerning the effects of nebivolol on maximal force development and myofilament Ca2<sup>+</sup> sensitivity in cardiac muscle strips have been previously described. Whereas Zeitz et al. reported that 1 µM nebivolol lowered maximal force development and myofilament Ca2<sup>+</sup> sensitivity in skinned rabbit and human trabeculae (Zeitz et al., 2000), Bundkirchen and colleagues did not observe such an effect at 10 µM in human tissue (Bundkirchen et al., 2001). The findings are contradictory but could be explained by either differences (i) between species (rabbit vs. mouse vs. human), (ii) in experimental setups, (iii) the functional status of the tissues (non-failing vs. failing), or (iv) a combination of it. Interspecies- or setup-dependent differences have been reported in permeabilized strip experiments with other drug interventions (Lues et al., 1988; Edes et al., 1995). Zeitz et al. used non-failing cardiac tissue from rabbit and explanted human tissue from end-stage failing myocardium from patients undergoing heart transplantation and saw an effect in tissues of both species. In another study the same group did not observe any effect on maximal force development in human explanted heart muscle preparations (Janssen et al., 2001). We observed no effect on maximal force development in neither mouse nor human tissue, but on myofilament Ca2<sup>+</sup> sensitivity in mouse tissue, which was more pronounced in the KI strips. Similar to this study we observed in a previous study with skinned trabeculae that EGCg, another compound with Ca2+ desensitizing properties had a more profound effect on strips of the KI than the WT genotype (Friedrich et al., 2016). This is also compatible with results of a study in which the Ca2+ desensitizing effect of ranolazine was only present in KI, but not in WT muscle strips (Flenner et al., 2016). The reason for the difference between KI and WT is unclear but could be related to the higher baseline myofilament Ca2<sup>+</sup> sensitivity in KI or to the proposed antioxidative activity of ranolazine, which might be important in a potentially hyperoxidized KI tissues (Lovelock et al., 2012; Flenner et al., 2016). In analogy to the study of Bundkirchen et al., in which nebivolol had no effect in explanted left ventricular tissue of patients with dilated cardiomyopathy, we did not observe any effect on Ca2<sup>+</sup> sensitivity at 1 or 10 µM in the HCM samples. In contrast, 30 µM of the positive control compound EGCg, which has been suggested to alter the interaction between cardiac troponin C and I and therefore the sensitivity of the myofilaments to Ca2<sup>+</sup> (Liou et al., 2008; Robertson et al., 2009), induced a rightward shift in the force-Ca2<sup>+</sup> curve in human HCM strips. EGCg has been shown to lower the myofilament Ca2<sup>+</sup> sensitivity in a transgenic HCM mouse model expressing a human cardiac troponin T mutant (Tadano et al., 2010) and in HCM-associated human cardiac troponin I and T mutants (Tadano et al., 2010; Warren et al., 2015; Messer et al., 2016). Similarly, we reported that 30 µM EGCg decreased Ca2<sup>+</sup> sensitivity in our Mybpc3 KI mouse model that carries a frequent Mybpc3 HCM mutation (Friedrich et al., 2016). EGCg action on myofilament Ca2<sup>+</sup> sensitivity in cardiac muscle strips of patients carrying a heterozygous MYBPC3 mutation indicates that the human strips can be desensitized for Ca2+.

Yet the precise mechanism of Ca2+-desensitization of nebivolol in mouse heart tissue remains unaddressed. Nebivolol is a third-generation beta-AR antagonist that exhibits vasodilating properties, most likely due to stimulation of nitric oxide synthase (Cockcroft et al., 1995). Since it was shown to attenuate hydroxyl radical-induced myocardial damage which has been associated with altered intracellular calcium handling and calcium overload of the myocytes (Josephson et al., 1991; Janssen et al., 1999; Piccini et al., 2012), it was proposed that nebivolol has direct free-radical scavenging properties (Janssen et al., 1999). Whether such an indirect effect is the main reason for a decrease in myofilament Ca2<sup>+</sup> sensitivity or another direct mechanism on the moyfilaments, such as binding to the C-terminal region of cardiac troponin C altering the interaction between cTnC and cTnI as in the case of the positive control compound EGCg (Liou et al., 2008; Robertson et al., 2009), exists, remains to be shown.

Clinically, beta-AR-antagonists are the mainstay of HCM therapy (Elliott et al., 2014). They are thought to potentially improve diastolic filling by a negative chronotropic effect. Some studies support the use of beta-AR-antagonists in HF patients with preserved ejection fraction (EF) but impaired relaxation similar to diastolic dysfunction seen in HCM patients (Lund et al., 2014). Recent data suggest that the effect of nebivolol is similar in HF patients with reduced and preserved EF (van Veldhuisen et al., 2009). This initiated the design of a still ongoing trial (https://clinicaltrials.gov/ct2/show/NCT02619526) investigating the effects of nebivolol and carvedilol on diastolic function of the left ventricle in older HF patients with preserved EF (Park and Park, 2016).

### LIMITATIONS OF THE STUDY

(1) The Mybpc3 KI model shows many HCM characteristics only at the homozygous state. Moreover, Mybpc3 KI mice present a reduced EF. These two points are in contrast to the more common findings in HCM patients who present left ventricular hypertrophy, interstitial fibrosis and diastolic dysfunction with heterozygous mutation states and normal or even supra-normal EF. (2) Our study does not explain the precise mechanism of Ca2+-desensitization of nebivolol in mouse heart tissue. (3) The Ca2<sup>+</sup> desensitizing effect of nebivolol in mouse tissues occurred at concentrations which were above the plasma concentrations (0.8–3.7 nM) observed in humans (Stoschitzky et al., 2004; Prisant, 2008). (4) Even though we did not observe any nebivolol effect in the human strips in this study, this observation cannot be generalized to all HCM patients since the number of tissues was low and they were derived from HCM patients carrying only mutations in MYBPC3.

### REFERENCES


### CONCLUSION

Nebivolol had no effect on maximal force, but induced a myofilament Ca2<sup>+</sup> desensitization in both WT and KI mouse cardiac muscle strips, which was more pronounced in KI muscle strips. In human cardiac muscle strips, nebivolol had no effect on force development and myofilament Ca2<sup>+</sup> sensitivity. Further studies should investigate the exact target and mechanism for Ca2<sup>+</sup> desensitization in mouse cardiac tissues in order to be able to develop modified compounds with even more potency and specificity for use in human tissue.

### AUTHOR CONTRIBUTIONS

SS and NK: Isolation and treatment of cardiac muscle strips and execution of experiments. LC: Analysis and interpretation of data, and correction of the manuscript. FF: Conception and design of research, execution of experiments, analysis and interpretation of data, figure preparation, and drafting of the manuscript. All authors critically discussed the results, and reviewed and approved the manuscript before submission.

### FUNDING

This work was supported by the DZHK (German Centre for Cardiovascular Research).

### ACKNOWLEDGMENTS

We thank the patient for donating the cardiac tissue for scientific research. We thank Julia Münch and Monica Patten (University Heart Center Hamburg, Hamburg, Germany) for patients' recruitment, Elisabeth Krämer, Giulia Mearini, and Frederik Flenner (UKE-Pharmacology, Hamburg, Germany) for help in preservation of human septal myectomies and database maintenance.

myosin binding protein C-deficient mice. Cardiovasc. Res. 69, 370–380. doi: 10.1016/j.cardiores.2005.11.009


radical-induced cellular injury and calcium overload in cardiac myocytes. J. Biol. Chem. 266, 2354–2361.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Stücker, Kresin, Carrier and Friedrich. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Burst-Like Transcription of Mutant and Wildtype MYH7-Alleles as Possible Origin of Cell-to-Cell Contractile Imbalance in Hypertrophic Cardiomyopathy

#### Edited by:

P. Bryant Chase, Florida State University, United States

#### Reviewed by:

Christopher Martin Yengo, Penn State Milton S. Hershey Medical Center, United States Kenneth S. Campbell, University of Kentucky, United States Corrado Poggesi, Università degli Studi di Firenze, Italy

#### \*Correspondence:

Theresia Kraft kraft.theresia@mh-hannover.de

†These authors have contributed equally to this work.

‡Deceased.

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 23 January 2018 Accepted: 22 March 2018 Published: 09 April 2018

#### Citation:

Montag J, Kowalski K, Makul M, Ernstberger P, Radocaj A, Beck J, Becker E, Tripathi S, Keyser B, Mühlfeld C, Wissel K, Pich A, van der Velden J, dos Remedios CG, Perrot A, Francino A, Navarro-López F, Brenner B and Kraft T (2018) Burst-Like Transcription of Mutant and Wildtype MYH7-Alleles as Possible Origin of Cell-to-Cell Contractile Imbalance in Hypertrophic Cardiomyopathy. Front. Physiol. 9:359. doi: 10.3389/fphys.2018.00359 Judith Montag<sup>1</sup> , Kathrin Kowalski <sup>1</sup> , Mirza Makul <sup>1</sup> , Pia Ernstberger <sup>1</sup> , Ante Radocaj <sup>1</sup> , Julia Beck <sup>1</sup> , Edgar Becker <sup>1</sup> , Snigdha Tripathi <sup>1</sup> , Britta Keyser <sup>2</sup> , Christian Mühlfeld<sup>3</sup> , Kirsten Wissel <sup>4</sup> , Andreas Pich<sup>5</sup> , Jolanda van der Velden<sup>6</sup> , Cristobal G. dos Remedios <sup>7</sup> , Andreas Perrot <sup>8</sup> , Antonio Francino<sup>9</sup> , Francesco Navarro-López <sup>9</sup> , Bernhard Brenner 1†‡ and Theresia Kraft 1† \*

<sup>1</sup> Hannover Medical School, Institute of Molecular and Cell Physiology, Hannover, Germany, <sup>2</sup> Hannover Medical School, Institute of Human Genetics, Hannover, Germany, <sup>3</sup> Hannover Medical School, Institute of Functional and Applied Anatomy, Hannover, Germany, <sup>4</sup> Clinic for Laryngology, Rhinology and Otology, Hannover Medical School, Hannover, Germany, <sup>5</sup> Hannover Medical School, Institute of Toxicology, Hannover, Germany, <sup>6</sup> Department of Physiology, Institute for Cardiovascular Research, VU University, Amsterdam, Netherlands, <sup>7</sup> Department of Anatomy, Bosch Institute, University of Sydney, Sydney, NSW, Australia, <sup>8</sup> Cardiovascular Genetics, Experimental and Clinical Research Center, Charité-Universitätsmedizin Berlin, Berlin, Germany, <sup>9</sup> Hospital Clinic/IDIBAPS, University of Barcelona, Barcelona, Spain

Hypertrophic Cardiomyopathy (HCM) has been related to many different mutations in more than 20 different, mostly sarcomeric proteins. While development of the HCM-phenotype is thought to be triggered by the different mutations, a common mechanism remains elusive. Studying missense-mutations in the ventricular beta-myosin heavy chain (β-MyHC, MYH7) we hypothesized that significant contractile heterogeneity exists among individual cardiomyocytes of HCM-patients that results from cell-to-cell variation in relative expression of mutated vs. wildtype β-MyHC. To test this hypothesis, we measured force-calcium-relationships of cardiomyocytes isolated from myocardium of heterozygous HCM-patients with either β-MyHC-mutation Arg723Gly or Arg200Val, and from healthy controls. From the myocardial samples of the HCM-patients we also obtained cryo-sections, and laser-microdissected single cardiomyocytes for quantification of mutated vs. wildtype MYH7-mRNA using a single cell RT-qPCR and restriction digest approach. We characterized gene transcription by visualizing active transcription sites by fluorescence in situ hybridization of intronic and exonic sequences of MYH7-pre-mRNA. For both mutations, cardiomyocytes showed large cell-to-cell variation in Ca++-sensitivity. Interestingly, some cardiomyocytes were essentially indistinguishable from controls what might indicate that they had no mutant β-MyHC while others had highly reduced Ca++-sensitivity suggesting substantial fractions of mutant β-MyHC. Single-cell MYH7-mRNA-quantification in cardiomyocytes of the same patients revealed high cell-to-cell variability of mutated vs. wildtype mRNA, ranging from essentially pure mutant to essentially pure wildtype MYH7-mRNA. We found 27% of nuclei without active transcription sites which is inconsistent with continuous gene

**340**

transcription but suggests burst-like transcription of MYH7. Model simulations indicated that burst-like, stochastic on/off-switching of MYH7 transcription, which is independent for mutant and wildtype alleles, could generate the observed cell-to-cell variation in the fraction of mutant vs. wildtype MYH7-mRNA, a similar variation in β-MyHC-protein, and highly heterogeneous Ca++-sensitivity of individual cardiomyocytes. In the long run, such contractile imbalance in the myocardium may well induce progressive structural distortions like cellular and myofibrillar disarray and interstitial fibrosis, as they are typically observed in HCM.

Keywords: hypertrophic cardiomyopathy, human cardiomyocytes, single-cell allelic imbalance, ventricular myosin heavy chain, myosin mutations, burst-like transcription

### INTRODUCTION

Hypertrophic Cardiomyopathy (HCM) is the most frequent inherited cardiac disease with a prevalence of about 1:500 (Maron et al., 1995). It is characterized by asymmetric hypertrophy of the left ventricle in the absence of other causes for hypertrophy (Maron and Maron, 2013). HCM can vary from essentially asymptomatic to highly malignant up to end-stage heart failure, or cause life threatening arrhythmias with sudden cardiac death particularly in young adults (Richardson et al., 1996). Two different presentations of HCM can be found in patients, the obstructive form (HOCM), where patients suffer from an obstruction of the left ventricular outflow tract, and the non-obstructive form (HNCM). Cardiomyocyte disarray with interstitial fibrosis and hypertrophy are hallmarks of HCM (Varnava et al., 2000). The degree of myocardial disarray correlates with risk factors for sudden cardiac death (Varnava et al., 2000), and it was suggested that myocyte disarray directly results from functional changes induced by the HCM related mutations at the sarcomeric level (Ashrafian et al., 2011).

In most familial HCM cases, heterozygous mutations in sarcomeric proteins such as the β-myosin heavy chain (β-MyHC), cardiac myosin-binding protein C (cMyBPC), cardiac troponin-T (cTnT), and cardiac troponin-I (cTnI) have been identified. Very few mutations were found in non-sarcomeric proteins (Ho et al., 2015a). About one third of the patients are heterozygous for mutations in MYH7 (encoding β-MyHC) and MYBPC3 (encoding cMyBPC) (Richard et al., 2003; Ho et al., 2015a).

It was proposed that the different HCM mutations lead to the HCM phenotype by enhancing contractility, and increasing calcium-sensitivity and ATPase activity of the cardiomyocytes (Ashrafian et al., 2011), while the opposite changes were thought to result in dilated cardiomyopathy (DCM) (Robinson et al., 2002; Hoskins et al., 2010; Davis et al., 2016). In studies on expressed human β-myosin with HCM-mutations also evidence for a hypercontractile state was found (Sommese et al., 2013; Bloemink et al., 2014). As mechanism for the postulated hypercontractility in HCM it was recently suggested that the mutations increase the availability of myosin heads for force production by altering the putative folded back state of the myosin heads (Kawana et al., 2017; Nag et al., 2017). Results from our group and others, however, are inconsistent with a generally enhanced contractility in HCM. Instead, contractility and calcium sensitivity can be enhanced or decreased in HCM (Venkatraman et al., 2003; Kirschner et al., 2005; Mirza et al., 2005; van Dijk et al., 2012; Kraft et al., 2013). Hence, the pathomechanism of HCM development is still unclear and a common trigger of HCM has yet to be identified.

In our work, we focused on HCM related mutations in β-MyHC which in humans is also expressed in slow twitch skeletal muscle fibers of e.g., M. soleus. For several missense mutations in the β-MyHC, we found reduced Ca++-sensitivity (Kirschner et al., 2005; Kraft et al., 2013). To our surprise, however, some muscle fibers showed a Ca++-sensitivity quite similar to fibers of control individuals while other fibers of the same patient had highly reduced Ca++-sensitivity, yielding a spectrum of different Ca++-sensitivities from normal to highly reduced within the same tissue sample (Kirschner et al., 2005). Based on these findings we hypothesized that significant functional heterogeneity also exists among individual cardiomyocytes of HCM-patients, and that this may result from cell-to-cell variation in expression of mutated β-MyHC. Within the myocardial cellular network, such unequal force generation among adjacent cardiomyocytes will result in distortions of cardiomyocytes and non-myocyte cells. Some of these will be overstretched or distorted by cardiomyocytes that overcontract. Such distortions could not only initiate cardiomyocyte and myofibrillar disarray but could also trigger stretch-induced signaling, e.g., TGF-β-signaling (Teekakirikul et al., 2010), that results in development of interstitial fibrosis and hypertrophy. Thus, cell-to-cell functional variance may initiate hallmarks of the HCM phenotype (Brenner et al., 2014).

In the present work, we tested our hypothesis by studying contractile properties of individual cardiomyocytes of HCMpatients' myocardium and expression of mutated MYH7 mRNA at the single cardiomyocyte level in the same tissue samples. We found significant functional heterogeneity in Ca++-sensitivity among individual cardiomyocytes of affected HCM patients, including cardiomyocytes with Ca++-sensitivity essentially indistinguishable from control cells. This may result from cell-to-cell variation in the fractions of expressed mutant and wildtype protein, including cardiomyocytes expressing low or almost no fraction of mutant protein. To directly test for cell-to-cell variability in expression of the mutant vs. wildtype alleles, we quantified mutant vs. wildtype transcript in individual cardiomyocytes isolated from the same tissue samples. We found substantial cell-to-cell variability ranging from essentially pure wildtype to essentially pure mutant MYH7-mRNA expression in the very same patients. Data from counting active transcription sites and model calculations suggest that stochastic, burst-like transcription of MYH7, which is independent for the mutant and the wildtype allele, could generate the large cell-to-cell variation in mutant vs. wildtype MYH7-mRNA and β-MyHC-protein, resulting in substantial functional heterogeneity.

### MATERIALS AND METHODS

For detailed methods, additional figures, and references see Supplementary Material.

### Patients and Controls

This study was carried out in accordance with the recommendations of the Ethics Committee of Hannover Medical School with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki (WMA, 1997). The study on anonymized human tissue was approved by the Ethics Committee of Hannover Medical School (No. 2276-2014). Samples of left ventricular free wall and interventricular septum with β-MyHC-mutation p.R723G were from myocardium of two male HCM patients who received a heart transplant (patient II-5, family 26 and patient III-1, family 157) (Enjuto et al., 2000). A sample of the interventricular septum was obtained during myectomy from a severely affected female with Hypertrophic Obstructive Cardiomyopathy (HOCM) with the β-MyHCmutation p.A200V. Control heart tissue from the left-ventricular free wall and interventricular septum was from non-transplanted donor hearts (n = 5). Cardiac tissue was flash frozen in liquid nitrogen immediately after excision.

### Single Cardiomyocyte Function

Cardiomyocytes were mechanically isolated and contraction parameters were measured at different Ca++-concentrations (pCa-values) from relaxing (pCa 9.0) to saturating Ca++ concentration (pCa 4.63) as previously described (Kraft et al., 2013), and (for mutation A200V) as described in Supplementary Material (Figure S1). To adjust PKA-dependent phosphorylation which has been shown previously to be higher in donor cardiac tissue compared to HCM-patient's cardiac tissue (Kraft et al., 2013), all cardiomyocytes of donors and patients were incubated with protein kinase A (PKA) prior to functional assessment. It has been shown that PKA treatment of donor cardiomyocytes induced only a small shift of the force-pCa-relation to higher calcium-concentrations, while for cardiomyocytes from patients with heart failure or HCM-patients the shift was significant (van der Velden et al., 2003; Kraft et al., 2013).

### Relative Quantification of Mutant vs. Wildtype MYH7-mRNA in Individual Cardiomyocytes

Cryosections (thickness 5µm) from frozen left-ventricular cardiac tissue were generated and sections of individual cardiomyocytes were isolated by laser capture microdissection after anti-cadherin staining of desmosomes (Figure S2). Individual cardiomyocytes were identified by the bright-field image clearly showing the striation pattern and were dissected when overlaying the bright-field image and the fluorescent image of intercalated discs labeled by an anti-cadherin-antibody. The cells were marked, laser-dissected, and catapulted from the tissue section into nuclease free water in the lid of a PCR-tube and lysed. Quantitative single cell RT-PCR was performed (for conditions and primers see Table S1 and Supplementary Material). The lysates were subjected to reverse transcription reaction mix and incubated on a custom-made micro-mixer for 10 min to improve uniform distribution of the low amount of mRNA molecules (Boon et al., 2011) and subsequently split into duplicates that were analyzed in parallel. The micro-mixer was also used to optimize successive cDNA synthesis (Boon et al., 2011). For relative quantification of the MYH7 alleles in single cardiomyocytes, nested PCR was applied, followed by a reconditioning PCR to avoid heteroduplex-formation (Thompson et al., 2002). For allele specific restriction digest, R723G- or A200V-PCR-products were treated with MboI or Hpy4CHI, respectively, yielding an allele-specific band pattern. Quantification of mutant vs. wildtype MYH7-mRNA occurred densitometrically after testing for linearity using standard plasmids of wildtype, R723G, or A200V sequence of the PCRamplicons as described (Tripathi et al., 2011) (Supplementary Material, Figure S3).

### Multi-Aliquot Control

A section of a sample from the left ventricular free wall of R723G-myocardium was lysed, diluted serially and subjected to quantitative single cell RT-PCR. The diluted lysate with a normalized IOD comparable to that of single cardiomyocytes was then divided into several aliquots for parallel quantification as described (Supplementary Material).

### MYH7-mRNA Copy Number in Individual Cardiomyocytes

Standard-RNA was generated by in vitro transcription using MYH7-cDNA. Single cardiomyocytes were microdissected and total MYH7-mRNA was determined by absolute quantification using real-time PCR and serial dilutions of the standard-RNA.

### Quantification of Mutated β-Myosin Protein With Mutation A200V in Tissue Samples

As described (Becker et al., 2007) (Supplementary Material; Figure S4), sarcomere-bound myosin was extracted from A200V-myocardium and digested by trypsin. Specific mutant and wildtype peptides were quantified by HPLC and mass spectrometry using corresponding synthetic stable-isotopelabeled internal standard peptides.

### Visualizing Active Transcription Sites

Active transcription sites were visualized by fluorescence in situ hybridization (FISH) using sets of 48 20-mer oligonucleotides (Stellaris <sup>R</sup> -probes; LGC Biosearch Technologies, Petaluma, CA, USA). One set was designed to hybridize with intronic sequences of MYH7-pre-mRNA and each oligonucleotide was labeled with one Cy5-like fluorophore (Quasar 670, LGC Biosearch Technologies). The other set was designed to hybridize with exonic sequences of MYH7-mRNA and labeled with a Cy3 like fluorophore (Quasar 570; LGC Biosearch Technologies). Both probe sets were custom made (Stellaris <sup>R</sup> Probe Designer). Following hybridization, active transcription sites were taken as bright spots inside nuclei of cardiomyocytes showing both fluorescence signals. Further details are described in Supplementary Material.

### Modeling of Independent, Burst-Like Transcription of Mutant, and Wildtype MYH7-Alleles

Model calculations were based on the concept of stochastic, burst-like transcription (Raj et al., 2006) including independent transcription and translation of mutant and wildtype MYH7 alleles to account for our experimentally observed mutant vs. wildtype transcript levels and function of individual cardiomyocytes from the heterozygous R723G-patient II-5. Modeling included the stochastic opening and closing of the transcription sites, synthesis of pre-mRNA, splicing to mRNA, degradation of mRNA, and synthesis and degradation of protein, each for mutant and wildtype, respectively. The only adjustable parameters were the rate constants for activation/inactivation of transcription of the two alleles and the splicing rate constant; all other rate constants were taken from the literature. From the measured fraction of mutated β-MyHC-protein in R723Gpatient myocardium and the mean pCa50-values of controls and R723G-patient cardiomyocytes we could also simulate a distribution of pCa50-values. For details on modeling constraints, additional results and references see Supplementary Material.

### Statistical Analysis

Data are presented as mean ± SD or ± 1.96 SD (range in which 95% of data points are expected; see Supplementary Material). To assure normal distribution (Shapiro-Wilk test) of normalized data, logit transformation was performed for statistics (Ashton, W. D., 1972). Student's t-test was used to determine significance levels. Equality of variances was examined by F-test. For p < 0.05 significance was assumed.

### RESULTS

### Large Cell-to-Cell Functional Heterogeneity Among Individual Cardiomyocytes From HCM Patients Force-pCa Relations

Cell-to-cell functional heterogeneity among individual cardiomyocytes of HCM patients was investigated by recording force-pCa relations of cardiomyocytes isolated from myocardial samples with β-MyHC-mutations R723G (Enjuto et al., 2000) and A200V, respectively, and of healthy controls for comparison (**Figure 1**). Force data of each cardiomyocyte were normalized to the maximum force at saturating Ca++-concentration (pCa 4.5). Interestingly, force-pCa relations of some cardiomyocytes with mutation R723G were similar to that of donor cells, while for others a clear shift to higher Ca++-concentrations was observed (**Figure 1A**). The mean shift to reduced Ca++sensitivity with mutation R723G (Figure S5A) was similar to the change we had previously found in fibers of M. soleus muscle and in myocardium of this and other patients with mutation R723G (Kirschner et al., 2005; Kraft et al., 2013). For cardiomyocytes with mutation A200V we found a similar pattern with some force-pCa relations comparable to donor cells and others with markedly reduced Ca++-sensitivity (**Figure 1B**). On average, the force-pCa curve for A200V was slightly shifted to higher Ca++-concentrations, however, due to the larger cell-to-cell variability the shift is not statistically significant (Figure S5B).

For both mutations, the individual force-pCa relations reveal a larger cell-to-cell variability in the position of the force-pCa relation along the abscissa, i.e., a larger variability in Ca++ sensitivity than in control cardiomyocytes (**Figures 1A,B**). This is particularly prominent for mutation A200V for which forcepCa relations vary from the range of the control myocytes up to positions shifted by about 0.25 pCa units to the right (red solid lines in **Figure 1B**). Thus, for both mutations some cardiomyocytes have a Ca++-sensitivity indistinguishable from that of control cells while other cells have a substantially lower Ca++-sensitivity. This is further illustrated by the pCa50 values of the individual cardiomyocytes obtained from fitting the Hill equation to the force data of the individual cells. In **Figures 1C,D** the pCa50-values of the individual cardiomyocytes are shown together with the related mean value and the range in which 95% of data points are expected (mean ± 1.96 SD). Overall, the pCa50-values of individual R723G and A200V cardiomyocytes are significantly lower than those of the related control cells, respectively, and the variances of the pCa50-values are larger than those for controls. For the A200V cells, the variance is significantly larger than for controls (p = 0.02; F-test).

### Forces at Partial Activation

Functional consequences of heterogeneities in the force-pCa relations of individual cardiomyocytes are illustrated when the forces generated by individual R723G- and controlcardiomyocytes are compared at partial activation level (pCa 5.55–5.6; boxed areas in **Figures 1A,B**). The pCa-values 5.5–5.6 represent physiological intracellular Ca++-concentrations in a twitch (Fabiato, 1981).

For both mutations, the mean relative forces at pCa 5.55–5.6 are significantly lower compared to controls (pR723G < 0.001, pA200V = 0.025; t-test). For cardiomyocytes of both patients, variances of forces at partial activation are significantly higher than that of respective controls (R723G p = 0.003; A200V p = 0.002; F-test). In fact, at pCa 5.55–5.6 forces generated by the "weakest" and "strongest" HCM cardiomyocytes were 10 fold (R723G) or even 20-fold (A200V) different (boxed areas in **Figures 1A,B**). In contrast, at the same partial activation, forces of weakest and strongest control cardiomyocytes differed at most 1.5-fold (**Figures 1A,B**). Since normalized forces are restricted to values between 0 and 1, statistical analysis was

lines delineate range in which 95% of all data points (mean ± 1.96 SD) are expected. pCa50-values of all groups show normal distribution (Shapiro-Wilk test). SD of

pCa50-values of mutated cardiomyocytes larger than that of controls (for A200V-cardiomyocytes statistically significant, p = 0.02, F-test).

done after logit transformation (Ashton, W. D., 1972). For details see Supplementary Material. The variance of the control cardiomyocytes reflects experimental error, while the much larger functional variance among individual cardiomyocytes of both HCM patients very likely results from an additional, large

intrinsic cell-to-cell heterogeneity of their calcium-sensitivity. This large intrinsic heterogeneity among individual cardiomyocytes of HCM patients, where some cells were very similar to controls and others had very much reduced Ca++-sensitivity, raised the question whether this reflects variation from small to rather high fractions of mutated vs. wildtype protein in these cardiomyocytes, respectively.

### Highly Heterogeneous Fraction of Mutant MYH7-mRNA From Cell-to-Cell

Since protein quantification of mutant vs. wildtype β-MyHC with missense mutations in individual cardiomyocytes is beyond the sensitivity of our mass-spectrometry approach (Becker et al., 2007; Tripathi et al., 2011), we quantified relative expression of mutant and wildtype MYH7-alleles in individual cardiomyocytes at the mRNA level. We expect this to reflect expression at the protein level, since in our previous work on several β-MyHCmutations including several patients with mutation R723G we always found a nearly 1:1 relation between fraction of mutant protein and mutant mRNA at the tissue level (Figure S7B) (Tripathi et al., 2011; Montag et al., 2017). Such a 1:1 relation between MYH7-mRNA and β-MyHC-protein was also found here and previously for mutation A200V (see below and Supplementary Material, Figure S7).

As the myocardial tissue samples had been flash-frozen immediately after surgery, we could not enzymatically isolate intact individual cardiomyocytes to determine the fraction of mutant MYH7-mRNA. Instead, individual cardiomyocytes were isolated from cardiac tissue sections by laser capture microdissection (Figure S2). Quantification of each cell was performed in duplicates. A specific micro-mixing method was



\*\*\*p < 0.001; \*p = 0.044; n indicates number of cardiomyocytes.

used to optimize cDNA synthesis (Boon et al., 2011). To quantify the relative abundance of mutant vs. wildtype MYH7-mRNA in cryosections of individual cardiomyocytes, we adapted the RT-PCR/restriction digest approach that we had previously used for quantification of MYH7-mRNA in tissue samples (Tripathi et al., 2011).

Three representative restriction analyses of individual A200Vcardiomyocytes in **Figure 2A** show that the fraction of mutant β-cardiac mRNA varies among individual cardiomyocytes from quite low mutant (cell 1) to almost pure mutant MYH7-mRNA (cell 3). **Figure 2B** schematically shows the fragments generated by the A200V-specific restriction digest. Note, the highly similar band pattern of the two aliquots of each individual cardiomyocyte indicates that the large cell-to-cell variability in the abundance of mutant MYH7-mRNA is not due to experimental error. For a similar sample gel with R723G-cardiomyocytes see Figure S6 and previous work for the other patient (Kraft et al., 2016).

Analysis of in total 35 individual R723G-cardiomyocytes showed that the fraction of mutant MYH7-mRNA varies from essentially pure mutant to almost pure wildtype MYH7 mRNA (**Figure 2C**). This was confirmed for another patient with the same mutation from another family (Figure S6). To obtain an estimate for the experimental error in our mRNA quantification procedure, we generated a "tissue averaged" cDNA sample that could be divided into multiple aliquots for parallel analysis following the same procedure as for the single cardiomyocytes. A whole cryosection of R723Gtissue was lysed and cDNA was synthesized. This cDNA sample was diluted and then divided into aliquots such that each aliquot contained a similar amount of cDNA as we obtained from individual microdissected cardiomyocytes (for details see Supplementary Material). Thus, these aliquots yielded band intensities of the restriction digest that were similar to those observed with microdissected individual cardiomyocytes. The fractions of R723G-MYH7-mRNA in each of the 13 aliquots ranged only between 0.5 and 0.85 (**Figure 2D**).

Since the fraction of mutant MYH7-mRNA (Fmut\_mRNA) is restricted to the range between 0 and 1, for statistical analysis Fmut\_mRNA was transformed into logits = ln((Fmut\_mRNA)/(1- Fmut\_mRNA)). For the 35 microdissected cardiomyocytes, Fmut\_mRNA is the average of the fraction observed in the two aliquots. Data of both groups, the 35 individual microdissected R723G-cardiomyocytes and the 13 control samples, showed normal distribution after logit transformation (Shapiro-Wilk test). Importantly, the variance in the 13 aliquots of the large cDNA sample, representing the experimental error, is significantly smaller than the variance in the fraction of R723G-MYH7-mRNA among the 35 individual microdissected cardiomyocytes (**Figures 2C,D**; p < 0.0001, F-test). In contrast, the mean fraction of mutated MYH7-mRNA of the aliquots with 0.69 was comparable to the mean of the individual cardiomyocytes with 0.70.

Cell-to-cell variance in the fraction of mutant A200V-MYH7-mRNA is shown in **Figure 2E**. Isolation of individual cardiomyocytes and quantitative single cell RT-PCR were performed as described for R723G, with primers and endonuclease adapted for A200V. Quantitative analysis of 21 microdissected cardiomyocytes again revealed a much larger cell-to-cell variation in the fraction of mutant mRNA (**Figure 2E**) than seen in the 13-aliquot control. Together, R723G- and A200V-data reveal a large variability among individual cardiomyocytes in the expression of the mutated allele relative to the wildtype allele, i.e., a large cell-to-cell variance of mutant transcript levels in the myocardium of the HCM patients. The range of variability for mutant A200V is somewhat smaller than for mutant R723G (**Figure 2C**). We did not see very high fractions of mutated A200V-mRNA which could be due to the fact that the mean fraction of mutated mRNA in A200V cells and tissue is lower than for R723G (dash-dot lines in **Figures 2C,E**, Figure S7).

Analysis of the average fraction of mutant mRNA in all analyzed cardiomyocytes together yielded a fraction of 0.70 for the R723G-patient (dashed-dot-line in **Figure 2C**, Figure S7A). This is very similar to the value determined earlier in whole tissue samples of several patients with R723G at the mRNA and protein level (four M. soleus and one other cardiac sample) and reflects the allelic imbalance at the tissue level described previously (Tripathi et al., 2011) (Figure S7). The higher abundance of mutant MYH7-mRNA compared to wildtype MYH7-mRNA may be the result of higher stability (longer life-time) of the mutant MYH7-mRNA (Tripathi et al., 2011).

The average fraction of A200V-mRNA of all analyzed cardiomyocytes was 0.53 (**Figure 2E**, Figure S7A), which is very close to the mean value determined in three whole cryosections of A200V-myocardium of 0.47 ± 0.05 (Figure S7A) and from larger tissue samples of the same myocardium of 0.48 ± 0.02 (Montag et al., 2017). We also determined the fraction of mutated protein in the myectomy sample with mutation A200V. The analysis was performed by mass spectrometry as described previously (Becker et al., 2007) (for peptides and enzyme see Supplementary Material) and yielded a fraction of 0.54 of A200V-β-myosin (Figure S7), which is similar to the previously determined value of 0.49 ± 0.01 (Montag et al., 2017). Thus, also for mutation A200V the mean fraction of mutated MYH7-mRNA and β-MyHCprotein in myocardial tissue samples is essentially the same, as found for several other β-MyHC-mutations before (Tripathi et al., 2011).

MYH7-mRNA-fraction is very similar to average of individual cardiomyocytes. (E) Fraction of A200V-mRNA (dark gray bars) and of wildtype-mRNA (light gray bars) in 21 cardiomyocytes of A200V myectomy-sample. Fraction varies from ≈10 to 90%. Statistics as in (C). Dash-dot line and dashed lines, means ± 1.96 SD. Normal distribution of individual cells and multi-aliquot data (Shapiro-Wilk test). Cell-to-cell variance in fraction of mutated MYH7-mRNA for individual cardiomyocytes significantly larger than variance of multi-aliquot control (experimental error); for R723G, p < 0.0001; for A200V, p < 0.001 (F-test).

(see text). (D) 13 Aliquots from cDNA of R723G-myocardium (left ventricular free wall) analyzed in parallel to test for experimental scatter. Average

### Independent, Stochastic On-Off Switching of Transcription of Mutant, and Wildtype MYH7-Alleles as Possible Cause for Cell-to-Cell mRNA and Functional Heterogeneity

Large variability in mRNA expression with functional heterogeneity among genetically identical cells was previously proposed to be the result of burst-like, stochastic transcription (Raj et al., 2006).

To test if this proposed mechanism of transcription also applies to MYH7 and transcription of MYH7 is indeed discontinuous, interrupted and burst-like, we visualized active transcription sites in the nuclei of cardiomyocytes. For this we used fluorescence in situ hybridization of cryosections of R723G-cardiac tissue samples with two fluorescently labeled 20-mer oligonucleotide probe sets. In active transcription sites within nuclei pre-mRNA contains both intronic and exonic sequences. One of the two probe sets was designed to hybridize with 48 intronic sequences of the MYH7-premRNA. The other probe set targeted 48 exonic sequences thus labeling both MYH7-pre-mRNA as well as MYH7-mRNA. Active transcription sites were identified as bright spots inside the nuclei of cardiomyocytes showing co-localization of both probe sets (**Figure 3A**). Cardiomyocytes were identified by the abundant presence of cytoplasmic spots of the exonic probe set indicating presence of MYH7-mRNA molecules. We counted active transcription sites in 122 nuclei of cardiomyocytes in R723G 16 µm-cryosections (13 cryosections; two R723G cardiac tissue samples) by analyzing 3D-stacks recorded by epifluorescence microscopy (**Figure 3A**) (Bahar Halpern et al., 2015). Of these 122 nuclei 27% had no active transcription sites but cytoplasmic MYH7-mRNA. Only nuclei fully embedded in the tissue sections were included in the analysis. In a control sample (5 cryosections) of a non-transplanted heart we found 32% of 240 nuclei in cardiomyocytes without active transcription sites. Absence of active transcription sites is inconsistent with continuous transcription of the two MYH7-alleles but is expected for discontinuous, stochastic, burst-like transcription.

This finding is supported by the distribution of the MYH7 mRNA copy number in individual cells (**Figure 3B,C**). We quantified the total copy number of MYH7-mRNA in 31 individual cardiomyocytes microdissected from cryosections of R723G cardiac tissue, thereby finding copy numbers per cardiomyocyte-slice from <10 up to 4.660. **Figures 3B,C** show this distribution on a linear and log scale. The distribution on the log scale is consistent with a normal distribution (Shapiro-Wilk test); the solid line represents a Gaussian fit to the decimal logarithm of the MYH7-mRNA copy number per cell. This lognormal distribution suggests a stochastic, burst-like transcription mechanism (Raj et al., 2006; Raj and van Oudenaarden, 2008; Bahar Halpern et al., 2015), while continuous transcription of the alleles is expected to generate a Poisson-distribution on the linear scale which is not consistent with our data (**Figure 3C**). Since here slices were 5µm thick, not the full cell volume was included in the analysis. From the average diameter of a cardiomyocyte (16.9 ± 1.3µm; Olivetti et al., 1996) the mRNA-copy number per cell should be about 3-times larger than determined in the slices.

### Numerical Model Simulation of Cell-to-Cell Variation of Mutant mRNA, Mutant Protein, and Functional Variability Based on Stochastic On/Off Switching of Transcription Sites

Next we asked whether our experimental observations could be accounted for by stochastic, burst-like transcription of the MYH7-alleles, which is independent for the mutant and wildtype allele. These experimental observations include the large cell-tocell variation in fraction of mutant vs. wildtype MYH7-mRNA (**Figure 2**), the cell-to-cell functional imbalance (**Figure 1**), and the 27% of R723G-cardiomyocyte nuclei having no active transcription sites (**Figure 3**). To address the above question we set up a numerical simulation for mutation R723G in which the mutant and wildtype MYH7-alleles were switched on and off stochastically and independently of each other. In addition, production of pre-mRNA, splicing to mRNA, and mRNA decay were also included in the model as was translation to and decay of β-MyHC protein. For details of the simulations see Supplementary Material. The rate constants for synthesis of premRNA, for degradation of mRNA, as well as for synthesis and degradation of protein were all taken from the literature, and are listed together with the respective references in Table S3. The only parameters we could use to fit the response of our simulations to the experimentally observed data were the rate constants for the stochastic switching on and off of transcription of the two alleles, and the rate constant of splicing of pre-mRNA to mRNA that were not available from the literature. These three parameters were adjusted to produce the best fit to all our above stated experimental findings.

Since a substantial fraction of cardiomyocytes can be polyploid, particularly in hypertrophied myocardium (Brodsky et al., 1994), we determined the ploidy of the R723G cardiomyocytes and included the observed distribution of di-, tetra-, octo- 16-, and 32-ploid nuclei in our simulation (for details see Supplementary Material).

A sample time course obtained from the model simulation of the transcription bursts, the mutant and wildtype MYH7-premRNA, the mutant and wildtype MYH7-mRNA, the fraction of mutant MYH7-mRNA, and the fraction of mutant protein are shown for mutation R723G in Figure S8 for a diploid and tetraploid cardiomyocyte. To account for a tissue-wide average fraction of around 0.67 for both R723G MYH7-mRNA and R723G β-MyHC (Tripathi et al., 2011; Figure S7), we adjusted the mRNA decay rate of the R723G MYH7-mRNA to half of that of wildtype MYH7-mRNA.

To be able to directly compare the results of 35 individual cardiomyocytes flash frozen and microdissected from myocardial tissue to the outcome of the simulation (fractions of mutant MYH7-mRNA and mutant β-MyHC protein), we randomly picked 35 points of a very long simulation run (see Supplementary Material). A large separation between the randomly picked points was used to assure 35 uncorrelated, independent points, just like in individual cardiomyocytes at the time of freezing of the myocardial tissue. Comparing the frequency distribution of the observed fractions of mutant mRNA in our experimentally analyzed cardiomyocytes (**Figure 4A**) with the outcome of our model (**Figure 4B**) shows that the distributions are very similar. Likewise, the values for the total number of MYH7-mRNA molecules and their frequency distribution obtained from the simulation at the randomly picked 35 points (**Figure 4C**) are similar to the experimental data (cf. **Figure 3C**). Thereby we have to consider that in 5µm cell slices the mRNA copy number is only about 1/3 of that in whole cells with a mean diameter about 16–17µm (Olivetti et al., 1996). Also, the predicted fractions of mutant mRNA from the 35 randomly picked points (**Figure 4D**) match the experimentally determined values for individual cardiomyocytes (cf. **Figure 2C**).

We next aimed to predict the fraction of mutant β-MyHCprotein per individual cell and its effect on β-MyHC function. Intriguingly, our model predicts that at the 35 randomly picked points the value for the fraction of mutant β-MyHC-protein ranges from 0.25 to 0.87, thus suggesting a marked variation from cell to cell (**Figure 4E**). This is in accordance with our hypothesis that different fractions of mutant β-MyHC may cause the observed functional differences among individual cardiomyocytes. We finally used the predicted fractions of mutant protein to calculate the expected shift in pCa<sup>50</sup> for each cell (**Figure 4F**, for calculation see Supplementary Material). Note, the mean value of the predicted pCa<sup>50</sup> is essentially identical with the experimentally determined one (**Figure 1C**). However, the range in which 95% of all data points are expected is somewhat narrower in the model than seen experimentally (dashed lines in **Figure 4F** vs. dashed lines in **Figure 1C**). This is because in the modeling only intrinsic variance of the pCa<sup>50</sup> arising from different expression of mutant β-MyHC-protein among individual R723G-sample cardiomyocytes is considered. If experimental error (represented by the variance in pCa<sup>50</sup> of control cardiomyocytes, cf. **Figure 1C**, lower panel) is also taken into account, this results in a 95% range that is very similar to the one in the experimental data (dotted lines in **Figure 4F** and dashed lines in **Figure 1C**).

Finally, with the best fit to all our experimental results a fraction of 24% nuclei without active transcription sites was predicted, which is close to the 27% seen experimentally. This fraction is strongly dependent on the two rate constants for the on/off-switching of the mutant and wildtype MYH7-alleles, and

Fluorescence in situ hybridization (FISH) to visualize active transcription sites (aTS) in 16µm thick sections of R723G cardiac tissue samples. Arrows pointing at active transcription sites with co-localization of intronic pre-mRNA signal (red) and exonic mRNA signal (orange). Orange spots in the cytoplasm indicate individual MYH7-mRNA molecules (exonic sequence only), identifying cells as cardiomyocytes. Shown is an example of a nucleus without aTS (upper row), a nucleus with one aTS (2nd row; arrow head indicates non-specific fluorescent spot visible in all channels), and a nucleus with two aTS (3rd and 4th row). Note that the two aTS are located at somewhat different z-levels, as indicated by the two slices of the z-stack shown here. In this analysis, we included only nuclei that were fully embedded in the tissue sections. DAPI (nuclear stain) in blue; test for non-specific fluorescence signals in green (clearly visible in leftmost images only, except bright spot of non-specific fluorescence labeled by arrow head in 2nd row). Scale bar, 5µm. (B,C) Total MYH7-mRNA copy numbers of 31 cardiomyocytes microdissected from sections (thickness 5µm) of R723G cardiac tissue were quantified. Note that the distribution on a linear scale in (B) is not a normal distribution. Distribution of log10 of total MYH7-mRNA copy number in (C) can well be fit by a normal distribution (solid line), which is characteristic for burst-like transcription (Raj and van Oudenaarden, 2009).

is just slightly modulated by the splicing rate constant from premRNA to mRNA. The on-times of mutant and wildtype MYH7 alleles generate "spikes" of pre-mRNA. These are substantially shorter than the spikes of mRNA which are longer due to the slower mRNA decay rate. Similarly, lifetime of the β-MyHC protein is substantially longer than of MYH7-mRNA, resulting in smaller fluctuations of the fraction of mutant β-MyHC protein than of mutant MYH7-mRNA (Figure S8).

points of a very long model simulation of transcription with independent, stochastic on/off switching of mutant and wildtype alleles. (C) Log-normal distribution of total copy number of MYH7-mRNA obtained from the model simulation at the 35 points, simulating cardiomyocytes microdissected from 5µm thick cryo-sections. Simulated numbers were divided by 3 to account for 5µm thickness of patient's tissue sections that represents only about 1/3 of full cardiomyocyte volume. Distribution slightly narrower than for experimental quantification (cf. Figure 3C). (D) Predicted fraction of mutant and wildtype MYH7-mRNA (cf. Figure 2C) and (E) predicted fraction of mutant and wildtype β-MyHC (protein) at the 35 randomly picked points; dash-dot line and dashed lines, means ±1.96 SD. In (D,E) dark gray bars indicate fraction of mutant MYH7-mRNA (D) and mutant protein (E) (y-axis on the left), and light gray areas indicate fraction of wildtype MYH7-mRNA (D) and wildtype protein (E) (y-axis on the right). (F) Predicted distribution of pCa50-values (filled circles) as expected from an experimentally determined mean fraction of mutant β-MyHC-protein of 0.67 (Tripathi et al., 2011) that shifts pCa50 by 0.15 pCa-units to the right (Figure S5A) without taking experimental error into account (dash-dot lines, mean; dashed lines, ±1.96 SD). Dotted lines, 95% range when taking experimental error into account (for details see Supplementary Material). Note the very similar 95% range as in Figure 1C for R723G cardiomyocytes.

### DISCUSSION

In this study, we tested the hypothesis that a functional heterogeneity exists among individual cardiomyocytes in myocardium of HCM patients that results from cell-to-cell variation in the fraction of mutant β-MyHC. Functional heterogeneity with imbalance in force generation during twitches among neighboring cardiomyocytes may well contribute to development of disarray, hypertrophy and fibrosis in HCM in the long run.

The main findings of our present study are (i) a quite substantial heterogeneity in Ca++-sensitivity among individual cardiomyocytes with two different HCM related β-MyHCmutations, mutation R723G and mutation A200V, respectively (**Figure 1**), (ii) large cell-to-cell variation in the fraction of mutant MYH7-mRNA among individual cardiomyocytes isolated from the same tissue samples (**Figure 2**), and (iii) absence of active transcription sites in 27% of nuclei in R723G cardiomyocytes indicating discontinuous, stochastic, burst-like transcription of the two MYH7-alleles (**Figure 3**). (iv) To test for an underlying mechanism that may provide a link between the experimental findings, numerical simulations were set up. They suggest that stochastic on/off switching of transcription of MYH7, which is independent for the mutant and the wildtype allele, can result in the observed cell-to-cell variation in mutant MYH7-mRNA and protein. Since the mutation reduces calcium-sensitivity, the variation in mutant myosin may also cause highly heterogeneous Ca++-sensitivity among individual cardiomyocytes (**Figure 4**). Thus, independent, stochastic on/off switching of transcription could reproduce the experimentally observed significant functional imbalance among individual cardiomyocytes isolated from the same piece of myocardial tissue as well as the observed fraction of nuclei without active transcription sites.

Our data and modeling suggest that the observed cell-to-cell variability in the fraction of mutated and wildtype MYH7-mRNA corresponds to a similar variability in the fraction of mutated and wildtype β-MyHC at the protein level, although the extent of this variability is smoothened by the lifetime of the protein (cf. Figure S8). Thus, since the observed fraction of MYH7-mRNA ranged from near zero (0.05) to essentially pure (1.0) mutant mRNA, the predicted protein fractions will presumably range from about 0.25 to 0.90, which represents a remarkable difference between the individual cells. Although protein quantification was not possible for individual cardiomyocytes, already the functional data corroborate this prediction. The calcium sensitivity of some cardiomyocytes was indistinguishable from controls (**Figure 1**). This suggests that these cardiomyocytes had little mutant myosin but mostly wildtype myosin in their sarcomeres at the time of freezing of the sample. On the other hand, cardiomyocytes with the largest shift of the force-pCa-curve most likely contained quite a large fraction of mutant myosin.

The correspondence of the mean fraction of mutated protein to the mean fraction of mutated MYH7-mRNA is further supported by our previous work on tissues of several HCM patients with β-MyHC-mutations including cardiac samples with mutation R723G (Tripathi et al., 2011), and by protein quantification of tissue samples with mutation A200V (Figure S7) (Montag et al., 2017). In all cases a close correlation between the mean fractions of mutant MYH7-mRNA and mutant protein was found.

Importantly, the populations of individual cardiomyocytes from which we determined the fraction of mutant MYH7 mRNA are representative for the cardiomyocyte populations in larger tissue samples. This is evident from the average fractions of mutant MYH7-mRNA of the individual cardiomyocytes, 0.70 for R723G mutation and 0.53 for A200V cardiomyocytes (**Figures 2C,E**), which are very close to the fractions determined in larger tissue samples, 0.70 for R723G and 0.47 for A200V (Figure S7) (Montag et al., 2017).

### Possible Mechanism for Cell-to-Cell Variation in Mutant MYH7-mRNA and β-MyHC-Protein

Continuous expression of both alleles could only account for the observed large variation in the fraction of mutant MYH7-mRNA among individual cardiomyocytes if only a low copy number (<50) of the MYH7-mRNA was generated per cardiomyocyte (Raj and van Oudenaarden, 2009). Also, continuous expression is expected to generate a Poisson distribution of the mRNA copy number per cell (Raj and van Oudenaarden, 2009). Both points, however, are inconsistent with our data. Absolute quantification revealed a log-normal distribution with a mean of 416 MYH7 mRNA molecules per cell in the 5µm tissue sections (**Figure 3**), yielding an expected median of about 1,200 mRNA molecules per cell.

Another possibility to account for the large cell-to-cell variation in mutant/wildtype MYH7-mRNA is random, monoallelic expression of the MYH7-gene. In fact, for fractions of mutant MYH7-mRNA ≥ 0.9 and ≤ 0.1 we cannot rule out that the small amount of wildtype or mutant mRNA, respectively, results from cross-contamination in cryosectioning and microdissection. That a large number of cardiomyocytes, however, have fractions of mutant MYH7-mRNA between 0.1 and 0.9 makes monoallelic expression rather unlikely. In addition, we also determined cells with more than one active transcription site, which seems incompatible with monoallelic expression. The functional data also argue against monoallelic expression as we do not see two clearly separate groups of force-pCa relations, one like controls, the other clustering around a lower pCa<sup>50</sup> value. We rather see a continuum of curves (**Figure 1**).

Therefore, the mechanism responsible for the large variance in the fraction of mutant MYH7-mRNA should account for both, cardiomyocytes with mixed mutant and wild-type MYH7 mRNA expression, and for essentially monoallelic expression. Using model simulations we show that stochastic, burst-like transcription of the mutant, and wildtype MYH7 alleles, due to stochastic opening and closing of chromatin (Janicki et al., 2004), can account for our experimental data. Burst-like transcription has been described in eukaryotic (Chubb et al., 2006) and mammalian cells (Raj et al., 2006), as well as in intact mammalian tissue (Raj et al., 2006; Bahar Halpern et al., 2015). This mechanism has been proposed as the basis for large cell-tocell variation in the number of specific protein molecules in genetically identical cells. In these earlier studies, however, no distinction was made between transcription of the two alleles of a particular gene like mutant and wildtype as done in our study.

Yet, because of the heterozygous genotype, we can distinguish between mutant and wildtype alleles and clearly see heterogeneous expression of the two alleles among cardiomyocytes. Therefore, we set up a numerical model based on the concept of burst-like transcription, which is independent for the two alleles, mutant and wildtype. This modeling revealed a surprisingly close match with our experimental data, including cell-to-cell variation in mutant MYH7-mRNA and in pCa50, as well as the distribution of the total MYH7-mRNA copy number among individual cardiomyocytes (**Figure 4**; Figure S8). Interestingly, with stochastic, burst-like transcription of mutant and wildtype alleles, a cardiomyocyte is expected to change at the mRNA- and protein-level, once in a while, from an almost wildtype cell via a mixed cell to a cell expressing almost pure mutant protein, and vice versa (cf. Figure S8).

### How Could Cell-to-Cell Functional Imbalance Affect Myocardial Function and Contribute to HCM Development?

The mutations studied here, like most HCM mutations, alter force generation and Ca++-sensitivity of the sarcomere, consistent with the "poison peptide" mechanism (Ashrafian et al., 2011). Therefore, it seems likely that sarcomeres with high or low amounts of mutant β-MyHC generate different force levels during each twitch/heartbeat, as suggested by the heterogeneity in force generation at partial activation (**Figure 1**). Cardiomyocytes with low force would get distorted and stronger cardiomyocytes may over-contract. According to our model calculations, this will change with the expression of mutant β-MyHC over time (Figure S8), affecting neighboring cardiomyocytes randomly (Brenner et al., 2014). Since the myocardium is a network, where adjacent strands of serially arranged cardiomyocytes are interconnected by branched cardiomyocytes, such a functional mosaic may well contribute to the severe myocyte disarray typical for HCM (Varnava et al., 2000). Therapeutic reduction in force generation, e.g., by β-adrenergic blockers or calcium antagonists, results in a milder phenotype, as may specific small molecule inhibitors of cardiomyocyte force generation, since functional imbalance and cellular distortions will be smaller. Along these lines, diltiazem treatment of pre-clinical HCM mutation carriers suggested a delay in early left ventricular remodeling (Ho et al., 2015b).

Even myofibrillar disarray within individual cardiomyocytes may develop since myofibrils, dependent on the time of their formation, may have quite different abundance of the mutant protein, generating subcellular functional imbalance among individual myofibrils. Electron microscopy revealed extensive disarray and loss of myofibrils in myocardium of the affected HCM patients in this (**Figure 5**) and previous studies (Ferrans et al., 1972; Kraft et al., 2013; Witjas-Paalberends et al., 2013). This is consistent with more disordered sarcomere structure of mutated cardiomyocytes in light micrographs (**Figures 5A,B**) and the observed reduced force generation per cross-sectional area of cardiomyocytes of affected patients compared to controls (Figure S5).

It should be noted that the observed heterogeneity among individual cardiomyocytes will most likely not be the only trigger of HCM-development. The globally increased or decreased myofilament calcium sensitivity caused by the mutations may well lead directly to remodeling and alterations of whole heart function, while cell-to-cell heterogeneity could exacerbate the phenotype. This is supported by findings in a severely affected homozygous HCM-patient, who developed cardiomyocyte hypertrophy and abnormal myofibrillar orientation (Nishi et al., 1994). Yet, it cannot be ruled out that an undetected additional compound heterozygous mutation may have contributed to this patient's phenotype. Overall, homozygous patients are very rare but they often show a severe and early onset phenotype that differs from that of heterozygous patients (Nishi et al., 1994; Richard et al., 2003). Similarly, homozygous mice with α-myosin- mutation Arg403Gln die within 7 days after birth, while the heterozygous animals apparently have normal life span where they develop several of the major characteristics of human HCM, including myocyte disarray (Geisterfer-Lowrance et al., 1996).

Importantly, our contractile imbalance hypothesis and the poison peptide principle do not exclude each other. Instead, the functional alterations caused by the mutated β-MyHC which is incorporated into the sarcomeres (poison peptide effect) are an essential starting point for development of the contractile imbalance. We suggest, based on our data that the fractions of functionally different mutated and wildtype β-MyHC in the sarcomeres vary from cell-to-cell due to stochastic burst-like expression of the two alleles. The resulting cell-to-cell contractile imbalance may particularly enhance progressive development of specific HCM features like structural distortions leading to cellular and myofibrillar disarray and interstitial fibrosis (Brenner et al., 2014).

### LIMITATIONS OF THE STUDY

One limitation of the work presented here is that fractions of mutant vs. wildtype mRNA and contractile function cannot be studied on the same cardiomyocytes, thus precluding a direct causal relation between cell-to-cell contractile imbalance, burstlike transcription, and variation in expression of mutated β-MyHC among individual cardiomyocytes. Our functional studies require chemically permeabilized cardiomyocytes where the cytoplasm is replaced by physiological solutions with defined calcium concentrations. This precludes subsequent mRNAisolation and analysis so that we have to rely on data from two groups of cardiomyocytes which, however, were isolated from the same pieces of cardiac tissue. In addition, the fractions of mutant vs. wildtype myosin protein cannot be studied at the single cell level since this requires a quantitative mass spectrometry approach with single cell sensitivity which is not available. Instead we suggest that the large functional variability (e.g., of pCa curves) of the cardiomyocytes is an indirect measure for cell-tocell variability of mutated protein in the sarcomeres.

We also cannot exclude that post-translational modifications of contractile proteins which may even vary from cell-tocell contribute to the observed decrease in Ca++-sensitivity and the functional heterogeneity among individual mutated cardiomyocytes. However, the increased heterogeneity in HCMpatients as compared to donors was detected even though at least PKA-dependent phosphorylation levels were adjusted for all cardiomyocytes by treatment with PKA.

control (C) and patient with mutation R723G (D). Different from control tissue, EM images of R723G tissue show longitudinally (l), obliquely (o), and cross sectioned (c) myofibrils right next to each other within individual cell (d, right panel) as indication of myofibrillar disarray. Z-disks (Z), M-lines (M), mitochondria (Mi), T-tubuli (T), a nucleus (Nu), and lipofuscin granula (Li) are indicated.

The data presented here from three HCM-patients provide experimental evidence for highly variable expression of β-MyHC-mutations in two male patients at end stage of the disease (R723G) and in a female patient (A200V) undergoing myectomy at the age of 19 years (see Supplementary Material). Despite the limitation to three patients the data indicate, however, that the variable allelic expression from cell-to-cell is not associated with the end stage situation of the HCM-heart. Yet, whether the proposed concept of independent, stochastic, burst-like transcription of the two alleles as trigger for HCM does also apply to MYH7-mutations that cause an increase in calcium-sensitivity and to HCM-mutations in other sarcomeric proteins remains to be elucidated in further studies.

### CONCLUSION

Our data show a significant cell-to-cell variation in mutated vs. wildtype MYH7-mRNA-expression and contractile properties among individual cardiomyocytes isolated from the same tissue samples of HCM patients with two different β-MyHCmutations. This large variability at the mRNA-level is most likely due to stochastic, burst-like MYH7-transcription, which is independent for the mutant and the wildtype allele, leading to a similarly large cell-to-cell variability in mutated vs. wildtype protein. The resulting large variation in contractile properties among individual cardiomyocytes may contribute to cellular and myofibrillar disarray and, by distortion of non-cardiomyocytes, trigger stretch sensitive signaling paths that lead to development of interstitial fibrosis. We hypothesize that such stochastic, burstlike transcription of the related alleles may also induce similar changes for other heterozygous HCM mutations.

### REFERENCES


### AUTHOR CONTRIBUTIONS

BB, TK, JM, and KK designed experiments. TK, JM, KK, MM, PE, JB, EB, ST, BK, and CM performed experiments. AR and BB set up and performed model simulations. KW, APi, JvdV, CdR, APe, AF and FN-L provided experimental setups or acquired tissue samples. BB, TK, JM, and AR wrote the manuscript with contributions from all authors.

### FUNDING

This work was supported by a HiLF-Grant (Hochschulinterne Leistungsförderung) of Hannover Medical School to JM, by funding through the Cluster of Excellence Rebirth at the MHH to TK and CM, and grants of the Deutsche Forschungsgemeinschaft to TK (KR1187/19-1 and KR1187/22-1).

### ACKNOWLEDGMENTS

This work stands in memoriam of BB who died in June 2017. His brilliant and innovative mind will keep inspiring us.

The authors thank Birgit Piep, Torsten Beier, and Alexander Lingk, Molecular and Cell Physiology, Hannover Medical School, for excellent technical assistance.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00359/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Montag, Kowalski, Makul, Ernstberger, Radocaj, Beck, Becker, Tripathi, Keyser, Mühlfeld, Wissel, Pich, van der Velden, dos Remedios, Perrot, Francino, Navarro-López, Brenner and Kraft. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Classifying Cardiac Actin Mutations Associated With Hypertrophic Cardiomyopathy

### Evan A. Despond and John F. Dawson\*

Department of Molecular and Cellular Biology, Centre for Cardiovascular Investigations, University of Guelph, Guelph, ON, Canada

Mutations in the cardiac actin gene (ACTC1) are associated with the development of hypertrophic cardiomyopathy (HCM). To date, 12 different ACTC1 mutations have been discovered in patients with HCM. Given the high degree of sequence conservation of actin proteins and the range of protein–protein interactions actin participates in, mutations in cardiac actin leading to HCM are particularly interesting. Here, we suggest the classification of ACTC1 mutations based on the location of the resulting amino acid change in actin into three main groups: (1) those affecting only the binding site of the myosin molecular motor, termed M-class mutations, (2) those affecting only the binding site of the tropomyosin (Tm) regulatory protein, designated T-class mutations, and (3) those affecting both the myosin- and Tm-binding sites, called MT-class mutations. To understand the precise pathogenesis of cardiac actin mutations and develop treatments specific to the molecular cause of disease, we need to integrate rapidly growing structural information with studies of regulated actomyosin systems.

### Edited by:

Julian Stelzer, Case Western Reserve University, United States

### Reviewed by:

Margaret Westfall, University of Michigan, United States Charles Redwood, University of Oxford, United Kingdom Beata M. Wolska, University of Illinois at Chicago, United States

> \*Correspondence: John F. Dawson

jdawso01@uoguelph.ca

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 02 February 2018 Accepted: 04 April 2018 Published: 17 April 2018

### Citation:

Despond EA and Dawson JF (2018) Classifying Cardiac Actin Mutations Associated With Hypertrophic Cardiomyopathy. Front. Physiol. 9:405. doi: 10.3389/fphys.2018.00405 Keywords: actin, cardiovascular disease, hypertrophic cardiomyopathy, mutations, myosin, sarcomere, tropomyosin

### INTRODUCTION

Cardiovascular disease is the leading cause of death worldwide, and puts a strain on the global economy, specifically costing the Canadian economy >\$22 billion annually (Gaziano, 2007; Smith, 2009). The end result of many cardiovascular diseases is heart failure and this can be influenced by a variety of factors, including genetics.

One such cardiovascular disease is hypertrophic cardiomyopathy (HCM), a group of related diseases characterized by hypertrophy of the ventricular myocardium, thought to be the result of increased calcium sensitivity. This disease can exhibit variable phenotypes, leading to difficulties in its clinical diagnosis (Baxi et al., 2016). Genes with mutations linked to HCM include myosin-binding protein-C (MYBP3), myosin heavy chain (MYH7), cardiac troponin T (TNNT2), and α-cardiac actin (ACTC1) (Seidman and Seidman, 2001; Walsh et al., 2017). Current research estimates that 1 in 200 individuals possesses an HCM-linked mutation (Semsarian et al., 2015).

A particularly interesting group of mutations is in the α-cardiac actin gene ACTC1. Due to the highly conserved nature of the actin sequence, the presence of mutations in ACTC1 in patients with HCM is of note. To date, 12 ACTC1 mutations have been identified in individuals with HCM, and 4 others in those with dilated CM (Olson, 1998; Mogensen et al., 1999, 2004; Olson et al., 2000; Van Driest et al., 2003; Morita et al., 2008; Olivotto et al., 2008; Kaski et al., 2009; Lakdawala et al., 2012).

**355**

The variety of genes linked to HCM encode proteins related to the sarcomere, the fundamental contractile unit of the heart, formed from interacting filaments of thin α-cardiac actin (ACTC) and thick β-myosin. These interactions are regulated principally by tropomyosin (Tm) and the troponin complex, which cooperate to reveal myosin-binding sites on the surface of actin filaments in the presence of calcium (Seidman and Seidman, 2001). The Mckillop and Geeves (1993) model of Tm action describes three states of Tm on F-actin in muscle (Lehman, 2017): (1) the blocked state, in which myosin-binding sites on actin are sterically blocked by Tm, (2) the closed state, where weak binding of myosin to actin is possible, and (3) the open state with strongly bound myosin present. Recent electron microscopy (EM) analyses largely agree with the states proposed by Mckillop and Geeves (1993) and reveal molecular interactions of the different states (Behrmann et al., 2012; von der Ecken et al., 2015, 2016; Gurel et al., 2017; Risi et al., 2017). With a growing understanding of the molecular interactions responsible for cardiac muscle contraction, rational explanations can be proposed for the impact of specific HCM-associated mutations in sarcomeric genes.

In this mini-review, we propose classification of the HCM-linked ACTC1 mutations based on the location of the amino acid changes in ACTC, and hence their proposed protein interactions. We classify mutations that alter direct and exclusive interactions with the myosin motor protein as myosin- or M-class mutations, including the extensively studied E99K-ACTC variant, as well as H88Y, R95C, F901, and S271F variants (**Figure 1**). Mutations that alter regulation of actin thin filaments by potentially disrupting interactions with Tm alone are termed Tm- or T-class mutations, including the A230V and R312C ACTC variants. Finally, mutations found in the Tm blocked state binding site that overlap with myosin binding are called MT-class mutations.

To date, the most thoroughly characterized mutation in any class is E99K-ACTC, while the remaining M-class mutations are

largely uncharacterized (**Table 1**). Data exist for some T- and MT-class mutations, but further testing is required to provide a more comprehensive understanding of the molecular cause of HCM arising from ACTC1 mutations.

### M-CLASS MUTATIONS

M-class mutations in ACTC are exclusive to sites of interactions with myosin, observed with loop 3 of the lower 50 kDa domain of myosin in actomyosin complexes as shown in recent EM work (Behrmann et al., 2012; von der Ecken et al., 2016; Banerjee et al., 2017; Fujii and Namba, 2017; Mentes et al., 2018). We also provisionally include the S271F-ACTC variant in the M-class based on its involvement in the high resolution structure of actomyosin structures (Gurel et al., 2017; Mentes et al., 2018).

### E99K

Originally described in a paper by Olson et al. (2000), the mutation encoding E99K-ACTC was found in several members of a family that exhibited HCM or similar cardiovascular conditions. The mutation was also found in 46 of 94 members of an unrelated family in Spain (Monserrat et al., 2007). E99K-ACTC displayed a significantly higher melting

#### TABLE 1 | Summary of published research on M-, T-, and MT-class ACTC variants.


Bold type denotes the original discovery of the mutation/variant.

temperature and critical concentration (Cc) compared to wild-type recombinant (WTrec) ACTC (Mundia et al., 2012). With unregulated filaments, the steady-state actin-activated myosin ATPase activity with E99K-ACTC protein showed a higher K<sup>m</sup> compared to WTrec-ACTC, but no change in the Vmax (Bookwalter and Trybus, 2006; Dahari and Dawson, 2015); there was also a decrease in the sliding velocity as measured by in vitro motility (IVM) assays (Bookwalter and Trybus, 2006; Debold et al., 2010; Dahari and Dawson, 2015; Liu et al., 2017). The binding affinity of E99K-ACTC for a short construct of myosin-binding protein-C (C0C2) was shown to be similar to WTrec-ACTC protein (Chow et al., 2014).

Further experiments examined E99K together with the regulatory proteins Tm and the troponin complex, forming regulated thin filaments (RTFs); these showed a significant decrease in maximum velocity, but no change in the concentration of calcium required to elicit a half-maximal response (pCa<sup>50</sup> value) using IVM (Debold et al., 2010). E99K thin filaments isolated from patients or reconstituted in vitro showed greater calcium sensitivity with higher activation at lower levels of calcium (Bai et al., 2015), as well as impaired relaxation of fibers and a decrease in the number of motile filaments in IVM (Song et al., 2013). At the whole organism level, mice expressing E99K-ACTC had a higher mortality rate, ECG abnormalities, and generally mirrored the disease phenotype seen in humans (Song et al., 2011). Overall, the literature describing E99K-ACTC agrees with the hypothesis that an increase in calcium sensitivity and filament activation gives rise to the phenotype seen in both mice and humans.

### H88Y, F901, and R95C

Three of the remaining M-class variants interact with loop 3 of myosin, and include H88Y, F901, and R95C; these have only been examined as unregulated filaments to date. H88Y and R95C were both identified through genetic screening of pediatric patients with idiopathic cardiac hypertrophy (Morita et al., 2008). H88Y-ACTC produced subtle differences in myosin ATPase activity and sliding velocity in vitro, but no significant change from WTrec-ACTC; conversely, R95C-ACTC resulted in a significant decrease in myosin ATPase Vmax, but no difference in IVM measurements (Liu et al., 2017). F901 was first identified through genetic screening of 79 pediatric patients with HCM (Kaski et al., 2009), and had a significantly lower myosin ATPase Vmax and faster IVM speeds compared to WTrec-ACTC (Liu et al., 2017). Further research with R95C, H88Y, and F901 RTFs is needed to determine if these variants have an impact on calcium sensitivity.

### S271F

The S271F-ACTC variant was discovered in a patient study examining myofilament-positive and -negative HCM (Olivotto et al., 2008). The S271 residue is distal to direct myosin and Tm-binding sites and is part of the hydrophobic plug of actin. To date, S271F-ACTC has not been characterized biochemically; however, recent high resolution structures of different myosin isoforms bound to F-actin (Gurel et al., 2017; Mentes et al., 2018) reveal the movement of the hydrophobic plug and interactions of the neighboring E270 that might be part of actomyosin function. For this reason, we provisionally assign the S271F-ACTC change to the M-class; whether S271F-ACTC modifies actomyosin interactions in cardiac muscle requires further experimentation.

### T-CLASS MUTATIONS

Recent high resolution structures of F-actin with Tm show ACTC changes that are specific to the Tm-binding site, with others involved in both Tm and myosin binding (Behrmann et al., 2012; von der Ecken et al., 2015, 2016; Fujii and Namba, 2017; Risi et al., 2017; Mentes et al., 2018). T-class mutations include those changes on the ACTC protein that likely interfere exclusively with the binding of Tm.

### A230V

A230V was first identified in a study of 389 unrelated HCM patients (Van Driest et al., 2003). This change is removed from the myosin-binding site and is closer to the open state Tm site.

The A230V variant has a lower melting temperature, and a higher C<sup>c</sup> (Mundia et al., 2012); there was no change in C0C2 binding compared to WTrec-ACTC (Chow et al., 2014). There were no significant changes in actomyosin interactions with the A230V-ACTC variant (Dahari and Dawson, 2015) in agreement with the distance of the change from the actomyosin-binding site. However, in reconstituted thin filaments, A230V-ACTC had decreased cross-bridge kinetics and pCa<sup>50</sup> (Bai et al., 2015), showing increased calcium sensitivity and leading to hypercontractile sarcomeres. Together, these data suggest a generally less stable actin variant with impaired function in regulated systems.

### R312C

The R312C HCM-related variant is the most recent to be discovered, with an analysis of 79 unrelated pediatric patients revealing the mutation (Kaski et al., 2009). Direct interactions between Tm and R312 have been observed in high resolution structures (Risi et al., 2017). Changes at the R312 residue might alter local interactions seen with R288 of the Dictyostelium myosin-IE motor domain (Behrmann et al., 2012). Since no such interactions are observed in structures with myosin-II seen in striated muscle (Fujii and Namba, 2017), we include R312C as a T-class ACTC mutation.

To date, no primary research has been generated regarding R312C-ACTC. Interestingly, the R312H variant linked to dilated CM has been studied before (Olson, 1998; Wong et al., 2001; Debold et al., 2010; Mundia et al., 2012; Chow et al., 2014; Dahari and Dawson, 2015). The R312H actin variant displays consistent protein stability issues and, interestingly, results in a lower pCa<sup>50</sup> of myosin activity with RTFs, with reduced maximum activation. The pCa<sup>50</sup> curve for R312H actin in IVM assays (Debold et al., 2010) is similar to that of seen with the A230V variant in reconstituted sarcomeres (Bai et al., 2015). The resulting hypercontractility is thought to cause HCM, whereas the R312H variant is associated with DCM. Future work is needed

to determine if the R312C variant exhibits similar effects on contraction.

### MT-CLASS MUTATIONS

ACTC changes in the Tm-binding site, particularly the blocked state, and myosin-binding site make up the MT-class of ACTC mutations. All of the MT-class ACTC changes are located closer to Tm in the blocked state than in the open state. There may be synergistic negative impacts of MT-mutations as they affect both the regulation and development of force.

### Y166C

The Y166C-ACTC variant was discovered through mutational analysis of 206 unrelated HCM patients (Mogensen et al., 2004). The amino acid at position 166 is 14.4 Å from Tm in the blocked state and located in the hydrophobic pocket between subdomains 1 and 3 of actin. This region also binds sections of the lower 50 kDa domain of myosin through hydrophobic interactions (Fujii and Namba, 2017); however, Y166 participates primarily with the docking of the DNase-I-loop of the next actin subunit in F-actin to form stable filaments.

Y166C-ACTC has slightly more efficient folding than WTrec-ACTC (Vang et al., 2005), an increased polymerization rate, and a decreased G-actin ATPase rate (Müller et al., 2012). Müller et al. (2012) found that Y166C-ACTC had no difference in filament formation, a decrease in Cc, and a decrease in actin-activated myosin ATPase rate and Vmax; our laboratory, however, showed a decrease in filament formation with Y166C-ACTC, an increase in Cc, and no difference in myosin ATPase rate (Mundia et al., 2012; Dahari and Dawson, 2015). Since the same expression system was used for both sets of data with recombinant Y166C-ACTC protein, it is unclear why functional differences were observed. Y166C-ACTC was also shown to have a decreased affinity for C0C2 (Chow et al., 2014), but no altered interactions with other binding partners (Vang et al., 2005; Müller et al., 2012). While there appears to be some impact on F-actin stability, no major impact on actomyosin interactions has been reported with the Y166C variant (Müller et al., 2012; Dahari and Dawson, 2015) and research including regulatory proteins needs to be conducted.

### P164A

P164A was discovered alongside the E99K-ACTC variant and was found in a single patient identified with HCM (Olson et al., 2000). Given the proximity of P164 to Y166, the change from proline to alanine at this position might alter the local conformation of the actin protein, thereby influencing the interactions discussed for Y166. As a result, we place the P164A variant in the MT-class of mutations.

A study of the P164A variant in yeast actin (Wong et al., 2001) showed no significant change in intrinsic actin properties or interactions with myosin. Conversely, characterization of P164A-ACTC produced with in vitro translation suggested some structural instability with the protein (Vang et al., 2005). However, no characterization of P164A-ACTC with myosin alone or in regulated systems has been completed.

### A295S

A295S was the first ACTC variant to be found through clinical research and demonstrated that mutations in α-cardiac actin were linked to HCM. The mutation was discovered in 13 of 18 family members with familial HCM (Mogensen et al., 1999). The A295S position is among the closest to the Tm molecule in high-resolution structures (Behrmann et al., 2012; Risi et al., 2017), approximately 10 Å away. In addition, the A295 position on ACTC packs beside K328 of actin that forms electrostatic interactions with loop 4 of myosin in the actomyosin complex (Behrmann et al., 2012; Fujii and Namba, 2017).

The A295S variant exhibits no difference compared to WTrec-ACTC in several properties, including binding interactions, myosin ATPase and IVM activity, and filament incorporation (Vang et al., 2005; Dahari and Dawson, 2015). Viswanathan et al. (2017) created the only in vivo model of A295S-ACTC to date, using Drosophila melanogaster to express the variant in heart and flight muscle. Flies with cardiac expression of A295S-ACTC showed decreased relaxation, increased pCa50, and tension-generating periods, which led to hypercontractile muscle. When expressed in indirect flight muscle, flies had impaired flight due to physical tearing of the muscle as a result of destructive hypercontractility (Viswanathan et al., 2017). Therefore, the impact of the A295S change appears to be primarily at the level of contractile regulation. The association with loop 4 of myosin at this location is either not interrupted or of lesser significance to the overall binding of myosin.

### A331P

The A331P variant was discovered in one patient from the same study that identified E99K and P164A (Olson et al., 2000). A331 of ACTC interacts either directly with myosin (Behrmann et al., 2012) or through association with the neighboring P333, forming part of a hydrophobic interaction with the CM loop of myosin (Fujii and Namba, 2017). In addition, A331 lies within the blocked Tm-binding site (Risi et al., 2017), suggesting that part of the regulatory action of Tm is to inhibit the interaction of the CM loop of myosin with actin.

Some research suggests that the A331P change affects F-actin characteristics (Wong et al., 2001), while others find no significant impact (Vang et al., 2005; Mundia et al., 2012), perhaps as a result of different isoforms and expression systems. The A331P-ACTC variant has a decreased affinity for the C-terminal fragment of cardiac myosin-binding protein (C0C2) (Chow et al., 2014), but no change in interactions with other binding partners, including myosin S1 (Wong et al., 2001; Vang et al., 2005). There was also no difference in the actin-activated myosin ATPase activity (Dahari and Dawson, 2015), sliding speed, or the number of moving filaments compared to WTrec-ACTC by IVM (Wong et al., 2001; Dahari and Dawson, 2015). These data suggest that the A331P change does not significantly alter local interactions with the CM loop of myosin.

A331P RTFs show markedly reduced calcium sensitivity, contractility, and cross-bridge force, but no change in cross-bridge kinetics (Bai et al., 2014). Decreased calcium sensitivity indicates that the A331P change keeps Tm in the blocked state and is thought to be indicative of DCM; therefore, it is of interest that an HCM-associated change produces the opposite effect on calcium sensitivity.

The differences between WTrec- and A331P-ACTC in RTFs do not translate to an in vivo system, as a transgenic mouse expressing cardiac muscle A331P-ACTC failed to develop a HCM phenotype (Toko et al., 2010), but this may be due to the presence of an epitope tag or WT-ACTC protein during ectopic expression. There is a wealth of information regarding the A331P variant, but a lack of clear connection between the change and the development of HCM.

### M305L

M305L was discovered in one individual during the same clinical investigation that identified Y166C (Mogensen et al., 2004). Similar to A331 above, the M305 residue packs against P333 of ACTC; M305 is the closest of all HCM-associated ACTC variants to Tm, being less than 9 Å apart in the blocked state. Moreover, the M305 residue is part of the nucleotide-binding site of actin forming an essential structural component of the functional protein.

The association with the nucleotide-binding site of actin is reflected in changes in the intrinsic actin properties of the M305L variant, with increased Pi release and polymerization rates (Müller et al., 2012; Mundia et al., 2012). Conversely, it shows no difference in protein-binding interactions (Vang et al., 2005; Müller et al., 2012; Chow et al., 2014) or myosin ATPase activity (Dahari and Dawson, 2015). Given the proximity of M305 to Tm in the blocked state, it will be important to examine regulated systems to determine the mechanism of the M305L variant in HCM development.

### REFERENCES


### CONCLUSION

Hypertrophic CM is the most commonly inherited heart disease and contributes to the significant economic and healthcare burden of cardiovascular disease in society. Understanding the mechanistic cause of HCM has become increasingly important in research, with experiments moving beyond basic biochemical properties to investigate altered protein–protein interactions. Presented in this mini-review is a classification system for the identified α-cardiac actin mutations based on their proposed protein interactions (**Table 1**). Overall, all variants except S271F and R312C have been investigated at the biochemical level, but few have been thoroughly studied in higher-order systems. This situation leaves a gap in our current knowledge regarding the implications of ACTC variants in RTFs, and how changes in regulation translate to a disease state such as HCM. Closing this gap is critical for the development of new therapies that target specific protein interaction deficiencies, resulting in fewer side effects and greater quality of life for people living with heart disease.

### AUTHOR CONTRIBUTIONS

JD took part in the research, wrote the first draft of the manuscript, edited and revised it, and formatted the final manuscript. ED revised the original manuscript, researching, restructuring, and reformatting the manuscript.

### FUNDING

This work was funded by a Grant-in-Aid to the Heart and Stroke Foundation of Canada to JD (G-15-0008961).

hypertrophic cardiomyopathy. J. Biol. Chem. 281, 16777–16784. doi: 10.1074/ jbc.M512935200



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Despond and Dawson. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Familial Dilated Cardiomyopathy Associated With a Novel Combination of Compound Heterozygous TNNC1 Variants

### Edited by:

Julian Stelzer, Case Western Reserve University, United States

#### Reviewed by:

Jonathan P. Davis, The Ohio State University, United States Charles Redwood, University of Oxford, United Kingdom Beata M. Wolska, The University of Illinois at Chicago, United States

#### \*Correspondence:

Saquib A. Lakhani saquib.lakhani@yale.edu Jose Renato Pinto jose.pinto@med.fsu.edu

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Striated Muscle Physiology, a section of the journal Frontiers in Physiology

Received: 19 July 2019 Accepted: 23 December 2019 Published: 22 January 2020

#### Citation:

Landim-Vieira M, Johnston JR, Ji W, Mis EK, Tijerino J, Spencer-Manzon M, Jeffries L, Hall EK, Panisello-Manterola D, Khokha MK, Deniz E, Chase PB, Lakhani SA and Pinto JR (2020) Familial Dilated Cardiomyopathy Associated With a Novel Combination of Compound Heterozygous TNNC1 Variants. Front. Physiol. 10:1612. doi: 10.3389/fphys.2019.01612 Maicon Landim-Vieira<sup>1</sup> , Jamie R. Johnston<sup>1</sup> , Weizhen Ji<sup>2</sup> , Emily K. Mis<sup>2</sup> , Joshua Tijerino<sup>1</sup> , Michele Spencer-Manzon2,3, Lauren Jeffries<sup>2</sup> , E. Kevin Hall<sup>4</sup> , David Panisello-Manterola<sup>2</sup> , Mustafa K. Khokha2,3, Engin Deniz<sup>2</sup> , P. Bryant Chase<sup>5</sup> , Saquib A. Lakhani<sup>2</sup> \* † and Jose Renato Pinto<sup>1</sup> \* †

<sup>1</sup> Department of Biomedical Sciences, College of Medicine, Florida State University, Tallahassee, FL, United States, <sup>2</sup> Pediatric Genomics Discovery Program, Department of Pediatrics, Yale School of Medicine, Yale University, New Haven, CT, United States, <sup>3</sup> Department of Genetics, Yale School of Medicine, Yale University, New Haven, CT, United States, <sup>4</sup> Department of Pediatrics, Yale School of Medicine, Yale University, New Haven, CT, United States, <sup>5</sup> Department of Biological Science, Florida State University, Tallahassee, FL, United States

Familial dilated cardiomyopathy (DCM), clinically characterized by enlargement and dysfunction of one or both ventricles of the heart, can be caused by variants in sarcomeric genes including TNNC1 (encoding cardiac troponin C, cTnC). Here, we report the case of two siblings with severe, early onset DCM who were found to have compound heterozygous variants in TNNC1: p.Asp145Glu (D145E) and p.Asp132Asn (D132N), which were inherited from the parents. We began our investigation with CRISPR/Cas9 knockout of TNNC1 in Xenopus tropicalis, which resulted in a cardiac phenotype in tadpoles consistent with DCM. Despite multiple maneuvers, we were unable to rescue the tadpole hearts with either human cTnC wild-type or patient variants to investigate the cardiomyopathy phenotype in vivo. We therefore utilized porcine permeabilized cardiac muscle preparations (CMPs) reconstituted with either wild-type or patient variant forms of cTnC to examine effects of the patient variants on contractile function. Incorporation of 50% WT/50% D145E into CMPs increased Ca2<sup>+</sup> sensitivity of isometric force, consistent with prior studies. In contrast, incorporation of 50% WT/50% D132N, which had not been previously reported, decreased Ca2<sup>+</sup> sensitivity of isometric force. CMPs reconstituted 50–50% with both variants mirrored WT in regard to myofilament Ca2<sup>+</sup> responsiveness. Sinusoidal stiffness (SS) (0.2% peak-to-peak) and the kinetics of tension redevelopment (kTR) at saturating Ca2<sup>+</sup> were similar to WT for all preparations. Modeling of Ca2+-dependence of kTR support the observation from Ca2<sup>+</sup> responsiveness of steady-state isometric force, that the effects on each mutant (50% WT/50% mutant) were greater than the combination of the two mutants (50% D132N/50% D145E). Further studies are needed to ascertain the mechanism(s) of these variants.

Keywords: genetic analysis, cardiac troponin C, missense variant, dilated cardiomyopathy, TNNC1

### INTRODUCTION

fphys-10-01612 January 17, 2020 Time: 16:21 # 2

Pediatric cardiomyopathies, or diseases of the heart muscle in young patients, have an annual incidence of 1.1–1.5 per 100,000 and are the most common indication for heart transplantation in children (Lipshultz et al., 2003; Dipchand et al., 2013; Lee et al., 2017). Dilated cardiomyopathy (DCM), characterized by dilation and dysfunction of one or both ventricles, is the most common type, accounting for more than 50% of all pediatric cardiomyopathies (Lipshultz et al., 2003). Genetic variation in any of a number of proteins that affect cardiomyocyte function is an important cause of DCM (Hershberger et al., 2013; van der Velden and Stienen, 2019). Some of these variants appear to simply predispose toward cardiomyopathy, while others are a primary cause of cardiomyopathy (Burke et al., 2016).

Troponin plays a central role in contractile regulation in striated muscle. This protein complex is comprised of three distinct subunits: troponin I (TnI), troponin T (TnT), and troponin C (TnC) (Farah and Reinach, 1995). Cardiac Troponin C (cTnC), which is encoded by the TNNC1 gene and is abundantly expressed in cardiomyocytes, functions as a myofilament Ca2<sup>+</sup> sensor and plays a critical role in regulating contraction (Li and Hwang, 2015). The tertiary structure of cTnC is dominated by the two globular halves of the molecule, the C- and N-domains, which are connected by a flexible linker (Takeda et al., 2003; Li and Hwang, 2015). The C-domain binds divalent cations (Ca2<sup>+</sup> and/or Mg2+) at two EF-hand motifs – referred to as sites III and IV based on their location in the primary sequence – and has important structural and modulatory roles in the sarcomere (Holroyde et al., 1980; Marques et al., 2017; Veltri et al., 2017a). The N-domain contains an evolutionarily defunct site I and regulatory EF-hand site II that normally triggers contraction upon Ca2<sup>+</sup> binding during systole (Holroyde et al., 1980).

Missense variants in cTnC have been associated with both dilated and hypertrophic cardiomyopathies (Mogensen et al., 2004; Landstrom et al., 2008; Willott et al., 2010; Pinto et al., 2011b; Parvatiyar et al., 2012), and knockout of cTnC in adult zebrafish results in a phenotype consistent with DCM (Ho et al., 2009). Here, we report the case of two siblings diagnosed with severe, early onset pediatric DCM associated with compound heterozygous variants in TNNC1. As a result of our attempts to understand the molecular basis of these pediatric DCM cases, we provide in vivo experimental evidence confirming that TNNC1 is a crucial gene for heart function, and further provide in vitro experimental evidence that altered contractile mechanics – unexpectedly – cannot always explain the pathogenicity of TNNC1 variants.

### EXPERIMENTAL PROCEDURES

### Ethics

The human subjects research in this study was approved by the Institutional Review Board of Yale University School of Medicine. Xenopus were maintained and cared for in an aquatics facility in accordance with Yale University Institutional Animal Care and Use Committee protocols.

### Sequencing Methods

Genomic DNA was isolated from either venous blood or saliva samples. Whole exome sequencing was performed by using IDT xGen capture kit followed by Illumina DNA sequencing (HiSeq 4000). Paired end sequence reads were converted to FASTQ format and were aligned to the reference human genome (hg19). SNVs and indels were called using a GATK pipeline and annotated using AnnoVar.

### Genome Editing in Xenopus tropicalis

CRISPR/Cas9-mediated genome editing in Xenopus tropicalis tadpoles was used as previously described (Bhattacharya et al., 2015). Briefly, two non-overlapping, independent CRISPR sgRNAs targeting tnnc1 were designed to generate knockdowns. CRISPR 1 oligo (targets exon 4):

CTAGCtaatacgactcactataGGTTCTTGGTCATGATGGTCgttt tagagctagaaTAGCAAG

and CRISPR 2 oligo (targets exon 5):

CTAGCtaatacgactcactataGGAGGAACTCATGCGAGATGgtt ttagagctagaaTAGCAAG.

The sgRNAs were synthesized using the EnGen sgRNA synthesis kit (NEB), and underwent subsequent purification and concentration using the RNA Clean & Concentrator-5 kit (Zymo). Individual sgRNAs were injected at 400 pg/embryo with 1.6 ng of Cas9 protein into single celled embryos according to standard methods (Ran et al., 2013). Uninjected control and CRISPR tadpoles were raised in 10 cm dishes until stage 42 to stage 45 of development. To visualize beating hearts, these tadpoles were embedded in low melt agarose in 1/9 X MR and images were obtained using a Thorlabs Telesto 1325 nm spectral domain optical coherence tomography system as previously described (Deniz et al., 2018).

### Cloning, Expression and Purification of Recombinant Proteins

Wild-type (WT) and mutant (either D132N or D145E) human cTnC proteins were cloned, expressed, and purified as previously described (Landstrom et al., 2008; Dweck et al., 2010). Sitedirected mutagenesis by PCR was used to engineer point mutations into the pET3-d vector. Sequences were verified prior to expression and purification of the cTnC mutants.

### Calcium Solutions

Different Ca2<sup>+</sup> solutions ranging from pCa 8.0 (relaxing) to 4.0 (activating) were calculated utilizing a pCa Calculator (Dweck et al., 2005). All solutions contained: 20 mM 3-[Nmorpholino]propanesulfonic acid (MOPS), 7 mM ethylene glycol-bis(2-aminoethylether)-N,N,N<sup>0</sup> ,N0 -tetraacetic acid (EGTA), 15 mM phosphocreatine, 15 units mL−<sup>1</sup> creatine phosphokinase, 2.5 mM MgATP2−, 1 mM free Mg2+,

age demonstrating sinus rhythm and left ventricular enlargement.

ionic strength maintained constant at 150 mM by adding Kpropionate (KPr), varying [Ca2+], pH 7.0. Anion of choice for the pCa solutions was propionate (e.g., CaPr2, MgPr2, and KPr). Solutions were prepared at room temperature (20–21◦C).

### Skinned Cardiac Muscle Preparation

Left ventricular papillary muscles were harvested from porcine hearts obtained from a local abattoir and dissected into small muscle bundles (Landstrom et al., 2008). Dissected cardiac tissue was exposed to pCa 8.0 relaxation solution (10−<sup>8</sup> M free [Ca2+]) containing non-ionic detergent Triton X-100 (1% v:v). After incubating for 4 h at 4◦C, skinned cardiac muscle preparations (CMPs) were transferred to 51% glycerol-relaxing pCa 8.0 solution (v:v), stored at −20◦C, and utilized within 1 month. Strips of muscle, 0.8–1.0 mm in length and 0.15– 0.25 mm in diameter, were mounted using aluminum foil T-clips to a force transducer on one end and a motor on the other end. CMPs were immersed in pCa 8.0 relaxation solution and the initial length (L0) was set at 2.1 µm sarcomere length measured by HeNe laser diffraction. Initial maximal Ca2+-activated tension was measured at pCa 4.0 (10−<sup>4</sup> M free [Ca2+]).

### Extraction of Native cTnC From CMPs and Reconstitution With Wild-Type or Mutant cTnCs

Native (endogenous) cTnC was extracted by incubating CMPs in 5 mM CDTA solution (pH 8.4) for ∼1.5 h at 21◦C as previously described (Landstrom et al., 2008). To evaluate the efficacy of cTnC extraction, residual Ca2+-activated tension was measured by immersing the CMPs in pCa 4. CMPs exhibiting residual force >40% (n = 2) were excluded from analysis. CMPs with residual force ranging from 8.4 to 39.4% were reconstituted by incubation for ∼5 min total (five incubations of ∼1 min each) at 21◦C with one of four conditions: drops of pCa 8.0 solution containing 1.0 mg mL−<sup>1</sup> of either 100% WT cTnC, 50% WT/50% D132N cTnC mix, 50% WT/50% D145E cTnC mix or 50% D132N/50% D145E cTnC mix. Tension recovery was evaluated by comparing the reconstituted maximal tension at pCa 4.0 to the initial maximal Ca2+-activated tension measured prior to cTnC extraction (P/P0) and ranged from 81% to 123%.

### Muscle Mechanics

pCa-Force: Ca2+-dependence of steady-state, isometric force was measured by incubating reconstituted CMPs in a series

(mother is reference, other three are heterozygous for the variant). The right side shows residues around the c.435C>A change that results in D145E in one of the two alleles for TNNC1 for three of four individuals (father is reference, other three are heterozygous for the variant).

of Ca2<sup>+</sup> solutions ranging from pCa 8.0 to 4.0 at 21◦C. Normalized [Ca2+]-force data for each CMP were fit using nonlinear least squares regression to a 2-parameter Hill equation to obtain parameter estimates for pCa<sup>50</sup> and nHill, as described (Veltri et al., 2017a).

Kinetics of Tension Redevelopment (kTR): After force reached steady-state in each pCa solution, measurement of the rate of tension redevelopment was obtained by shortening CMPs by 20% L0, followed by rapid, 25% re-stretch, then release back to L<sup>0</sup> (Gonzalez-Martinez et al., 2018). The apparent rate constant (kTR) was obtained from each tension recovery time course as described previously (Chase et al., 1994; Regnier et al., 1999). Conditions where steady-state, isometric force was below 15% of the maximal force were excluded from the kTR analysis because of the relatively low signal-to-noise at the lowest levels of isometric force. Estimations of the 3-state model parameters (f, g, and kOFF) were computed in MatLab as previously described (Gonzalez-Martinez et al., 2018) with the exceptions that force for each condition was assumed to be 1.0, and also because there are no direct measurements of Ca2<sup>+</sup> binding to or dissocation from the mutant cTnCs, the value for kON for all simulations was assumed to be that derived from measurements on WT, 1.84 × 10<sup>8</sup> (Pinto et al., 2011a). In addition to estimates for parameters f, g, and kOFF, the modeling also provides an estimate for pCa50; the modified least-squares criterion is weighted such that the best fit, predicted pCa<sup>50</sup> is essentially the same (well within experimental error) as that measured experimentally for the same condition.

### Sinusoidal Stiffness

Sinusoidal stiffness (SS) was obtained by oscillating CMP length ∼0.2% L<sup>0</sup> peak-to-peak at 100 Hz, and recording the length and force signals at a sampling rate of 1 kHz. SS measurements were performed and the data were analyzed as described previously (Gonzalez-Martinez et al., 2018).

### Statistical Analyses

Statistical analyses including non-linear regression were performed using SigmaPlot v.12.0. Data were tested for significant statistical differences using one-way ANOVA

with post hoc Student-Newman–Keuls test. Data are shown as mean ± S.E.

### RESULTS

### Clinical Data

The proband was a female delivered full term following an uncomplicated pregnancy that included two normal fetal echocardiograms. She had initially been well without symptoms referable to the cardiovascular system. She demonstrated no feeding intolerance, premature fatigue, irritability, pallor, cyanosis, increased work of breathing, tachypnea, excessive diaphoresis, or lethargy. She had a good appetite and breast fed without difficulty. Screening echocardiography was done at 18 days of age due to an older sibling with neonatal cardiomyopathy of presumed idiopathic etiology. This showed borderline enlargement of the left ventricle with a left ventricular end diastolic diameter (LVEDD) of 2.18 cm (Z score of 0.8) and a low decreased M-mode left ventricular ejection fraction (LVEF) of 48%. Given the family history, she was admitted to the hospital briefly with a diagnosis of DCM for initiation of medical therapy with captopril, furosemide and aspirin. She was followed closely as an outpatient, with electrocardiogram having signs of left ventricular enlargement and serial echocardiograms demonstrating gradually worsening left ventricular function, necessitating a steady escalation of her medical management (**Figure 1**). Notably, right ventricular size and function remained normal throughout. By 7 months of age both dilation and function of her left ventricle had worsened with echocardiography showing LVEDD 3.40 cm (Z score of 4.09) and M-mode LVEF of 32%. At 13 months of age, echocardiography showed LVEDD 4.50 cm (Z score of 8.55) and M-mode LVEF of 33%. She was being managed with enalapril, carvedilol, digoxin, furosemide and aspirin, and had been listed for cardiac transplantation due to her severely depressed heart function. She suffered a cardiac arrest while awaiting a suitable donor and expired at 14 months of age.

Family history was notable for an 11 year-old brother who presented in the neonatal period with poor feeding and growth, and was diagnosed with severe DCM. He was managed

initially with medications then received a cardiac transplantation at 13 months of age. He has done well since, with good function of his transplanted heart, and without any signs of skeletal muscle problems. Both parents, currently in their early 40<sup>0</sup> s, have had normal echocardiograms and do not have any signs of cardiovascular disease. There is no other family history of cardiomyopathy, congenital heart defects or sudden unexpected cardiac death.

### Genetic Testing

Due to the strong family history, the proband underwent genetic testing with whole exome sequencing, which identified two variants of uncertain significance, D132N (c.394G>A, p.Asp132Asn) and D145E (c.435C>A, p.Asp145Glu), in the TNNC1 gene (NM\_003280.2). Subsequent analysis of the family (proband, brother and both parents) revealed that these variants were inherited in trans (**Figures 2A,B**). Both variants are located in the region of divalent cation binding site IV in the C-domain of cTnC (**Figure 3**). Both of these aspartate residues, as well as the surrounding amino acids, are highly conserved among vertebrates (**Figure 3D**). In the Genome Aggregation Database (gnomAD), D145E has a frequency of 1.28 × 10−<sup>4</sup> , whereas D132N has not been previously reported in a healthy population. These were also noted to be the only two rare (<0.5% minor allele frequency in the general population) single gene variants across the entire exome that were shared by both affected siblings, and thus no other rare or pathogenic variants were found in other known cardiomyopathy genes in either of the siblings. The proband was sequenced to a mean depth of 102 independent reads per targeted base and her affected brother and their unaffected parents were sequenced to a mean depth of 44–49X. Greater than 10X coverage was achieved for 98% of the proband's exome and 94% for the rest of family members' exomes. Greater than 20X coverage was achieved for 96% of the proband's exome and 88% for her other family members.

### TNNC1 in Xenopus

To assess the consequences of altered cTnC in an in vivo system, we began by using CRISPR/Cas9-mediated gene editing to knockout TNNC1 in Xenopus. Loss of the TNNC1 gene did not prevent the early stages of development, and resulted in a dramatic cardiac phenotype in tadpoles consistent with DCM (**Figure 4** and **Supplementary Video S1**). Notably, the knockout tadpoles demonstrated ventricular dilation, wall thinning and almost imperceptible cardiac motion. With the ultimate goal of introducing the two mutations separately and together, we then attempted to rescue this cardiomyopathy phenotype by expression of human cTnC in depleted frog embryos. Unpredictably, we were unable to obtain expression of cTnC in the tadpole hearts despite employing various techniques injecting either TNNC1 mRNA or TNNC1 plasmid DNA under control of a CMV promoter (**Table 1**).

### Muscle Mechanics

It was previously shown that cTnC-depleted CMPs reconstituted with cTnC-D145E increased Ca2<sup>+</sup> sensitivity of steady-state isometric force generation (Landstrom et al., 2008; Pinto et al., 2011a; Veltri et al., 2017a). However, to our knowledge, the effects of cTnC-D132N or mixtures of these cTnCs have not been previously reported. Considering that both siblings were found to be compound heterozygous for D145E and D132N in cTnC, we sought to explore their potential functional significance in vitro.

TABLE 1 | Attemps to rescue TNNC1 knockout in Xenopus tropicalis.


50 pg

to target cardiac tissue) 100 pg,

Native, endogenous cTnC was extracted from porcine CMPs using CDTA incubations and reconstituted with recombinant expressed human cTnC-WT (control) or mixtures of cTnCs (section "Experimental Procedures"). The range of average of residual tension post-extraction was 18% ∼ 35%, with individual values ranging from 8.4 to 39.4%, indicating that the majority of native cTnC had been extracted from the myofilaments (**Table 2**, where force values are reported as specific force). Upon reconstitution, the average tension recovery was at least 87%, indicating that the majority of troponin molecules had been functionally reconstituted with exogenous cTnC and that both WT and mutant recombinant cTnCs were competent for Ca2+-activation of contraction.

CMPs reconstituted with a mixture of 50% WT and 50% D132N variant (WT/D132N) displayed significantly reduced myofilament Ca2<sup>+</sup> sensitivity of force generation (pCa<sup>50</sup> = 5.32 ± 0.03), with a rightward shift of 0.11 pCa units compared with CMPs reconstituted with 100% WT cTnC (pCa<sup>50</sup> = 5.43 ± 0.02) (**Table 2** and **Figures 5A,B**). Consistent with previous reports that compared 100% D145E with 100% WT (Landstrom et al., 2008; Pinto et al., 2009; Veltri et al., 2017a), we found that reconstitution with a mixture of 50% WT and 50% D145E (WT/D145E) significantly increased myofilament Ca2<sup>+</sup> sensitivity of tension (pCa<sup>50</sup> = 5.58 ± 0.04), with a leftward shift of 0.15 pCa units (**Table 2** and **Figures 5A,B**). Interestingly, we observed no significant difference in myofilament Ca2<sup>+</sup> sensitivity of tension upon reconstitution of CMPs with a mixture of 50% D132N and 50% D145E (D132N/D145E) (**Table 2** and **Figures 5A,B**). Although the D132N variant in 50%:50% combination with either WT or D145E tended to reduce maximal tension recovery, the differences were not statistically significant (**Table 2** and **Figure 5B**). The Hill coefficient (nHill) is an indicator of cooperativity of thin filament activation. Compared to WT (nHill = 1.70 ± 0.20), only the WT/D132N mixture exhibited a significant change – an increase – in the apparent cooperativity (nHill = 2.60 ± 0.45) (**Table 2**).

Because we observed no significant difference in myofilament Ca2<sup>+</sup> sensitivity of tension for the 50%:50% combination of the two variant cTnCs (D132N/D145E) compared with 100% WT (**Table 2** and **Figures 5A,B**), we investigated two additional indices of contractile function: SS and kinetics of tension redevelopment (kTR). SS is used as a metric of the overall number of cross-bridges, whereas kTR is used a metric of the kinetics of cross-bridge cycling at maximal Ca2+-activation and also informs about thin filament regulatory unit dynamics at submaximal Ca2+-activation. We found no significant difference in maximum SS values when comparing CMPs reconstituted with the variants (WT/D132N; WT/D145E; D132N/D145E) to those reconstituted with 100% WT (**Table 3** and **Figures 6A,B**), suggesting that force per cross-bridge was unchanged. In addition, there was little or no difference in any of the relationships between SS and isometric force (**Figure 6B**).

Maximum kTR values were not significantly different when comparing CMPs reconstituted with the variants (WT/D132N; WT/D145E; D132N/D145E) to those reconstituted with 100% WT (**Table 3** and **Figures 7A,B**). This result informs us that, according to Brenner (1988), the sum of the sum of the rates of cross-bridge attachment and detachment (f +g) is approximately constant for all four conditions. To explore the possibility that the variants may have altered cross-bridge attachment (f) and detachment (g) rates even though the sum remained constant, we used a 3-state model of muscle regulation to evaluate the Ca2+-dependence of kTR (Landesberg and Sideman, 1994; Hancock et al., 1997; Loong et al., 2013; Gonzalez-Martinez et al., 2018). The force-kTR data for CMPs containing the mutants (WT/D132N, WT/D145E or D132N/D145E) appear to be less curvilinear than for WT (**Figure 7B**). The fits of the 3-state model capture the curvature of the data from CMPs containing mutants, but does not fully capture the greater curvature of the WT data (**Figure 7B**). Best fit values for f, g, and kOFF corresponding to the lines in **Figure 7B** are reported in **Table 4**. The 3-state model's four parameters (f, g, kON, and kOFF) are all lumped parameters that primarily inform about cross-bridge cycling (f and g) and thin filament regulatory unit dynamics including Ca2<sup>+</sup> binding to and dissociation from cTnC (kON and kOFF). Computational modeling indicates that kOFF is, as expected, generally faster as Ca2<sup>+</sup> sensitivity decreases (**Table 4**), although this relationship is modulated by additional factors because kOFF reflects more than just Ca2<sup>+</sup> dissociation from cTnC. Furthermore, the modeling indicates that f was slower than WT and g was faster for both WT/D132N and WT/D145E (**Table 4**). For the two mutants together (D132N/D145E), f and g changed in the same directions as for the individual variants (WT/D132N and WT/D145E), but the magnitudes of the changes were less; in other words, the combination of the two variants appeared to partially ameliorate the effects of the individual variants, as was also found for Ca2<sup>+</sup> sensitivity of steady state isometric force (**Table 2** and **Figures 5A,B**).

TABLE 2 | Hill equation parameter estimates for Ca2<sup>+</sup> dependence of steady-state isometric force obtained from porcine CMPs reconstituted with exogenous cTnC WT or variants depicted in Figure 5.


pCa50, pCa needed to reach 50% of the maximal force. 1pCa<sup>50</sup> = cTnC variant pCa<sup>50</sup> – WT pCa50; nHill, cooperativity of thin filament activation. Statistical significance was determined using one-way ANOVA with post hoc Student-Newman–Keuls test. \*p < 0.05 vs. WT.

FIGURE 5 | Ca2<sup>+</sup> dependence of steady-state isometric tension in porcine CMPs reconstituted with exogenous cTnC WT (100% WT) or variants (50% WT/50% D132N, 50% WT/50% D145E, or 50% D132N/50% D145E). (A) Relative steady-state isometric force as a function of pCa. The force values were normalized to the maximal steady-state isometric force in the same preparation. (B) Normalized steady-state isometric force as a function of pCa. The force values were normalized to the maximal steady-state isometric force generated by WT. Data are shown as mean ± S.E. and best fit parameter estimates from non-linear least squares regression on the Hill equation are summarized in Table 2 (n = 4–8).

TABLE 3 | Maximal sinusoidal stiffness (SSmax) and rate of tension redevelopment (kTRmax) measured in porcine CMPs reconstituted with exogenous cTnC WT or variants depicted in Figures 6, 7, respectively.


SSmax, maximum steady-state sinusoidal stiffness; kTR max, maximum rate of tension redevelopment.

### DISCUSSION

Given the central role of cTnC in cardiomyocyte contraction, it is little surprise that a growing number of variants in the TNNC1 gene have been identified in genetic screens for familial cardiomyopathies, both hypertrophic and dilated. Here, we report two siblings with severe, early onset DCM who were found to be compound heterozygous for two TNNC1 variants: D145E inherited from the healthy mother and D132N inherited from the healthy father. Considering that myofilament dysfunction typically correlates with severity of cardiomyopathy, we anticipated that the combination of the two variants (D132N/D145E) would have adverse consequences for Ca2<sup>+</sup> regulation of contractility. However, in contrast to what might be expected from the family pedigree, our functional data indicate that these two variant cTnC proteins when present as a mixture of one variant with WT (as in the mother and father) significantly altered Ca2<sup>+</sup> regulation of steady-state isometric force in opposite directions, while the combination of the two variants (as in the affected siblings) did not.

### Dissonance Between Clinical Outcomes and Mechanical Measurements

Although multiple in silico predictors of variant effects – including SIFT, PolyPhen2 and CADD – predict that each of these variants is likely to be detrimental, they are both classified as variants of uncertain significance in ClinVar. The D132N variant has not been reported in the Genome Aggregation Database (gnomAD). The population frequency of the D145E variant is reported as 0.0001228 in gnomAD, and the Atlas of Cardiac Genetic Variations classifies this uncommon variant as "unlikely to be pathogenic." This classification of D145E is consistent with individuals such as the proband's mother, who is heterozygous and asymptomatic. The available evidence, therefore, suggests that D145E could possibly be a risk factor for cardiomyopathy in some cases, but does not appear to be disease-causing when present in the heterozygous state. The two affected siblings presented here, in contrast to the parents, have D145E in trans with the D132N variant. When viewed in this light, the D145E population frequency is well within an expected range for a recessive disease.

To assess the combined effects of two potentially dysfunctional TNNC1 alleles, we performed knockout experiments in Xenopus, revealing a dramatic DCM phenotype in tadpoles (**Figure 4** and **Supplementary Video S1**) that is consistent with a previous report in zebrafish (Ho et al., 2009). Our multiple attempts to rescue this phenotype with human cTnC protein expression (**Table 1**) were unsuccessful, however, as the protein could be found expressed in other tissues but not in the heart (data not shown). The reason for this remains unclear to us at this time, though the fact that we have extensive experience obtaining wide expression – including in the heart – of multiple other proteins in tadpoles, suggests that this may be an issue specific to TNNC1.

Interestingly, the D145E variant was first described in a patient with hypertrophic cardiomyopathy (HCM) and it was the only variant found in a screen of fifteen HCM-susceptibility genes (Landstrom et al., 2008). In contrast, a subsequently described proband with DCM also had a rare variant in the MYBPC3 gene, which was hypothesized to combine with cTnC D145E to manifest as DCM (Hershberger et al., 2010; Pinto et al., 2011b). Based on these opposing findings, it was suggested that this second patient may have initially developed HCM that rapidly progressed to DCM (Freeman et al., 2001;

FIGURE 6 | Sinusoidal stiffness analysis in porcine CMPs reconstituted with exogenous cTnC WT (100% WT) or variants (50% WT/50% D132N, 50% WT/50% D145E, or 50% D132N/50% D145E). (A) Ca2<sup>+</sup> dependence of steady-state sinusoidal stiffness. (B) Normalized force vs. steady-state sinusoidal stiffness. The force values were normalized to the maximal steady-state isometric force generated by WT. (B) Dashed lines were drawn to connect the points. Data are shown as mean ± S.E. and maximum SS values are summarized in Table 3.

Fujino et al., 2001, 2002; Nanni et al., 2003). The proband described here, however, was followed closely since birth given her brother's diagnosis, and she never had any echocardiographic or other evidence of HCM, suggesting that this combination of cTnC alleles results purely in DCM. Of significance is another example of compound heterozygosity where the D145E variant was again associated with inherited cardiomyopathy (Ploski et al., 2016). In this case, two pediatric patients were diagnosed with autosomal recessive restrictive cardiomyopathy which was fatal during infancy. Sanger sequencing revealed that both of the affected individuals were compound heterozygous for genetic variants in TNNC1 corresponding to D145E and A8V in the cTnC protein.

Variants in TNNC1 can affect skeletal muscle function because cTnC is also expressed in slow skeletal muscle. The D145E mutant was reported to increase slow skeletal muscle ATPase activity at low and high calcium concentrations in reconstituted myofibrils (Veltri et al., 2017b). Although it is known that D145E does not affect myofilament calcium sensitivity in slow skeletal skinned fibers (Veltri et al., 2017b), it is unclear whether D132N has any effect on slow skeletal muscle function. Importantly, no skeletal muscle weakness was reported in the patient's history.

### Implications for cTnC Function in the Cardiac Thin Filament

Both variants are located in the C-domain of cTnC (**Figure 3**). The missense D132N variant is located within the G-helix that is part of EF-hand site IV, and D145E is one of the divalent cation (Ca2<sup>+</sup> or Mg2+) coordinating residues (+Z position) in site IV. While the regulatory site for Ca2<sup>+</sup> binding is site II in the N-domain of cTnC, our prior results indicate that there is communication between the C- and N-domains (Badr et al., 2016) such that changes in the C-domain can influence the regulatory N-domain (Marques et al., 2017; Veltri et al., 2017a). The results of this study demonstrated that the D145E variant (WT/D145E) increases myofilament Ca2<sup>+</sup> sensitivity even when WT is also present, while the D132N mutation (WT/D132N) had the opposite effect of decreasing Ca2<sup>+</sup> sensitivity (**Table 2** and


TABLE 4 | Optimized parameter estimates and predictions from the 3-state model for force-kTR data depicted in Figure 7.

Best fit parameter estimates for three parameters (f, g, and kOFF) were obtained in MatLab using the Simplex method for the force-kTR data in Figure 7, taking into account the Ca2<sup>+</sup> sensitivity (pCa50) from steady-state isometric force measurements (section "Experimental Procedures") as reported in Table 2; values for kON were derived from measurements reported by Pinto et al. (2011a) and maximal isometric force was assumed to be 1.0 for each dataset. These parameter sets were used to illustrate relations between kTR and steady-state, isometric force in Figure 7 (dashed lines in Figure 7B).

**Figures 5A,B**) even though the two affected residues are relatively close in the primary sequence and the 3D structure (**Figure 3**). In general, myofilament incorporation of troponin variants that cause opposing effects on thin filament Ca2<sup>+</sup> responsiveness are expected to normalize the myofilament Ca2<sup>+</sup> sensitivity (Li et al., 2010; Alves et al., 2014, 2017; Dieseldorff Jones et al., 2018), and that was what was found in this study. Compared to WT, the 50%:50% mix of D132N/D145E showed similar myofilament Ca2<sup>+</sup> sensitivity and Hill coefficient (nHill) values, the latter generally reflecting cooperative processes associated with isometric force generation. While incorporation of the two mutants appeared to normalize opposite effects of the two variants on Ca2<sup>+</sup> dependence of isometric force, what is difficult to reconcile is the severe DCM pathology associated with the presence of the two mutations.

A paradigm associating changes in myofilament Ca2<sup>+</sup> sensitivity commonly found in HCM (increased) and DCM (decreased) is generally well-accepted (Willott et al., 2010; van der Velden and Stienen, 2019). This study, however, presents a unique clinical case in which we observed no significant changes in myofilament Ca2<sup>+</sup> sensitivity due to a combination of variants in patients with DCM. How can this finding be explained? After depleting native cTnC from CMPs, each available unfilled cTnC site was presumably reconstituted, depending on the experiment, with either exogenous WT, WT/D132N mix, WT/D145E mix, or D132N/D145E mix in an attempt to mimic what is expected to be found in the patients (D132N/D145E) or the parents (WT/D132N or WT/D145E), along with the control condition (100% WT). Previous studies showed that D145E cTnC dissociates more slowly from the thin filament compared to WT cTnC indicating that D145E has a higher binding affinity for the thin filament (Marques et al., 2017), although the functional data (**Table 2**) do not suggest that this difference in affinity influenced the results.

As with our finding, there has been some variation in prior reports on variants in TNNC1 with regard to the hypothesis that HCM is always associated with increased Ca2<sup>+</sup> sensitivity and DCM with decreased Ca2<sup>+</sup> sensitivity. The first DCM-associated variant identified in TNNC1, G159D (Mogensen et al., 2004), has been explored by several groups with conflicting findings. It was initially reported that G159D did not alter myofilament Ca2<sup>+</sup> sensitivity in Tn-exchanged skinned rat trabeculae (Biesiadecki et al., 2007). In contrast, both cTnC-depleted skinned porcine papillary muscle and in vitro regulated thin filaments reconstituted with cTnC-G159D exhibited reduced myofilament Ca2<sup>+</sup> sensitivity (Robinson et al., 2007; Dweck et al., 2008). Conversely, G159D exhibited increased myofilament Ca2<sup>+</sup> sensitivity in both human skinned ventricular myocytes obtained from a patient bearing G159D, and thin filaments reconstituted with skeletal muscle actin, human cardiac tropomyosin, and troponin complex containing cTnC G159D (Dyer et al., 2009).

Loss of the lusitropic response (cardiac muscle relaxation) is considered a common feature of DCM and HCM (Wang et al., 2012; Dweck et al., 2014; Messer and Marston, 2014). In the case of thin filament mutants, this can be partially attributed to uncoupling of cTnI phosphorylation from modulation of Ca2<sup>+</sup> sensitivity (Memo et al., 2013; Messer and Marston, 2014). Studies using TnC-extracted porcine papillary muscle fibers reconstituted with cTnC DCM-associated mutants (Y5H, M103I or I148V) exhibited diminished or abolished PKA-mediated Ca2<sup>+</sup> desensitization (Pinto et al., 2011b). Although we did not test for PKA-mediated myofilament Ca2<sup>+</sup> desensitization in our current study, we speculate that the presence of both mutant proteins could interfere with cTnI phosphorylation levels and functional consequences. This could cause a loss of downstream β-adrenergic stimulation which is crucial to effect the lusitropic response and diastolic ventricular filling.

Alternative possibilities exist because the two mutations are not present in the same protein. In the affected siblings, the two variant proteins are expected to be distributed randomly in thin filaments, and differences in function might affect communication between adjacent regulatory units when they contain different cTnCs. The lack of change in nHill for D132N/D145E (**Table 2** and **Figures 5A,B**) suggests this might not be a plausible explanation, although the Hill coefficient represents the overall response and does not inform about highly localized effects when two variant cTnCs are present. Mechanical kinetics provides information beyond steady-state force measurement, and cooperative interactions among regulatory units do not appear to be required for fast rates of tension redevelopment, kTR (Chase et al., 1994). Both the force-kTR data (**Figure 7B**) and the

3-state model analyses of those force-kTR relations (**Table 4**) suggest that the presence of either variant (WT/D132N or WT/D145E) shifts the force-kTR relation, and alters contractile kinetics (kTR in **Figure 7B**, and f and g in **Table 4**) in the same direction. However, both the kinetic data (**Figure 7B**) and modeling (**Table 4**) support the conclusion from analysis of the steady-state force-pCa data (**Table 2** and **Figures 5A,B**) that the combination of the two variants, D132N/D145E, in thin filaments may partially ameliorate the effects of either mutant on contractile function. This is because D132N/D145E yields results more similar to 100% WT than either WT/D132N or WT/D145E (**Table 4** and **Figure 7B**). Thus it appears to be necessary to seek alternative explanations, beyond changes in biomechanical function, to explain the deleterious effects of the two variants in combination in the proband and her brother.

### cTnC Function Beyond the Sarcomere

We and others have previously suggested that troponin subunits may participate in cellular functions that extend beyond contractile regulation in the sarcomere (Asumda and Chase, 2012; Wu et al., 2015; Zhang et al., 2016; Pinto et al., 2017). In fact, cTnC has been previously detected in cardiomyocyte nuclei, as well as the nuclei of certain cancer cells (Johnston et al., 2018). In addition, cTnC has been identified in the mitochondrial compartment of some cancer cell lines. Therefore, a potential mechanistic explanation for the severe disease phenotype associated with troponin mutants may be attributed to perturbations in its putative nuclear/mitochondrial function, whereby gene expression or bioenergetics could be altered by cTnC variants such as those reported here.

In summary, we report the association of compound heterozygous variants in TNNC1 in siblings with early onset familial DCM. We provide functional evidence that both variants separately lead to abnormal myofilament Ca2<sup>+</sup> sensitivity in which the variants D132N and D145E are, respectively, associated with reduced or increased myofilament Ca2<sup>+</sup> sensitivity relative to WT, while these effects compensated for no change relative to WT when the two variants were combined. Analyses of activation-dependence of kTR suggest that both variants have similar effects on contractile kinetics, although there was no significant change in maximal kTR, and that the combination of the two variants again partially ameliorated the effects of the individual variants. None of the observed changes in mechanics, however, can explain the clinical outcomes in this family. Further studies are needed to understand precisely how these variants combine to result in a severe DCM phenotype.

### DISCLOSURE

SL is part owner of Qiyas Higher Health, a startup company unrelated to this work.

### DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

### ETHICS STATEMENT

The studies involving human participants were reviewed and approved by the Institutional Review Board of Yale University School of Medicine. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin. The animal study was reviewed and approved by the Yale University Institutional Animal Care and Use Committee.

### AUTHOR CONTRIBUTIONS

JP and SL conceptualized the study and oversaw project administration. ML-V, JJ, WJ, EM, JT, MS-M, LJ, EH, DP-M, MK, and ED contributed to the data curation. PC contributed to computational data analysis and modeling. ML-V, JJ, JT, EM, PC, SL, and JP wrote the manuscript. All authors reviewed and approved the final version of the manuscript.

### FUNDING

JP was supported by the National Institutes of Health (grant no. HL128683). The Yale PGDP was supported by the Yale New Haven Hospital and Sara and Jeffery Buell.

### ACKNOWLEDGMENTS

The authors are grateful to this family for sharing their story. The authors also thank Bradley's Country Store, Tallahassee, FL, United States, for generously supplying porcine hearts, and Monica Konstantino for patient recruitment.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys.2019. 01612/full#supplementary-material

VIDEO S1 | Optical Coherence Tomography imaging showing contractility of tadpole hearts. Control heart (top panel) with good systolic contractility. Two views of TNNC1-F<sup>0</sup> mutant hearts (middle and lower panels) showing minimal motion and lack of effective contractility.

### REFERENCES

fphys-10-01612 January 17, 2020 Time: 16:21 # 12


that causes familial dilated cardiomyopathy, studied in explanted heart muscle. Circ. Heart Fail 2, 456–464. doi: 10.1161/CIRCHEARTFAILURE.108.818237



**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Landim-Vieira, Johnston, Ji, Mis, Tijerino, Spencer-Manzon, Jeffries, Hall, Panisello-Manterola, Khokha, Deniz, Chase, Lakhani and Pinto. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

digital media

of impactful research

article's readership