Disrupted mechanobiology links the molecular andcellular phenotypes in familial dilated cardiomyopathySarah R. Clippingera,1, Paige E. Cloonana,1, Lina Greenberga, Melanie Ernsta, W. Tom Stumpa,and Michael J. Greenberga,2

aDepartment of Biochemistry and Molecular Biophysics, Washington University School of Medicine, St. Louis, MO 63110

Edited by James A. Spudich, Stanford University School of Medicine, Stanford, CA, and approved July 23, 2019 (received for review June 27, 2019)

Familial dilated cardiomyopathy (DCM) is a leading cause of suddencardiac death and a major indicator for heart transplant. The diseaseis frequently caused by mutations of sarcomeric proteins; however,it is not well understood how these molecular mutations lead toalterations in cellular organization and contractility. To address thiscritical gap in our knowledge, we studied the molecular and cellularconsequences of a DCM mutation in troponin-T, ΔK210. We deter-mined the molecular mechanism of ΔK210 and used computationalmodeling to predict that the mutation should reduce the force persarcomere. In mutant cardiomyocytes, we found that ΔK210 notonly reduces contractility but also causes cellular hypertrophy andimpairs cardiomyocytes’ ability to adapt to changes in substratestiffness (e.g., heart tissue fibrosis that occurs with aging and dis-ease). These results help link the molecular and cellular phenotypesand implicate alterations in mechanosensing as an important factorin the development of DCM.

cardiomyopathy | mechanobiology | troponin | contractility | muscle

Dilated cardiomyopathy (DCM) is a major cause of suddencardiac death in young people, and it is a significant cause of

heart failure (1). DCM is phenotypically characterized by dilation ofthe left ventricular chamber, and it is often accompanied by changesin cellular and tissue organization, including lengthening of indi-vidual myocytes and fibrosis-induced stiffening of the myocardialtissue. Genetic studies have demonstrated that familial DCM can becaused by single point mutations, many of which are in sarcomericproteins responsible for regulating cardiac contractility (1); how-ever, the connection between the initial insult of molecular-basedchanges in contractile proteins and the development of the cellulardisease phenotype is not thoroughly understood. This lack of un-derstanding has hampered efforts to develop novel therapeutics,and there is currently no cure for DCM, with heart transplantationbeing the only long-term treatment. The goal of this study wasto better understand the connection between mutation-inducedchanges in contractility and the cellular phenotype in DCM.Understanding the link between point mutations in sarcomeric

proteins and the development of the disease phenotype in cellshas been challenging for several reasons. One challenge stemsfrom the fact that the clinical presentation and prognosis appearto depend on the specific mutation. In fact, different point mu-tations within the same molecule, and even different substitu-tions at the same amino acid site, can lead to different forms ofcardiomyopathy (e.g., hypertrophic, dilated, restrictive) withdifferent ages of onset (2–5). Another challenge stems fromselecting appropriate model systems of the human disease forthe specific question being posed. In vitro biochemical studiesare excellent for deciphering the molecular consequences of theinitial insult (4); however, they are not clearly predictive of howthe disease will manifest itself in cells (6). Studies using trans-genic mice have significantly advanced our understanding of thedisease; however, these mice do not always recapitulate the hu-man disease phenotype due to differences in cardiac physiologybetween humans and mice (7–13). Patient tissue is excellent forstudying the disease phenotype in humans (14); however, tissuefrom patients is difficult to obtain and usually only available from

patients in the advanced stages of the disease or post mortem,when compensatory mechanisms can mask the initial insult.Recent advances in stem cell (15, 16) and gene editing tech-

nologies (17) have provided a new avenue for modeling DCM inhuman induced pluripotent stem cell-derived cardiomyocytes(hiPSC-CMs) (18, 19). These hiPSC-CMs can faithfully recapit-ulate many aspects of the early human disease (20), makingthem an excellent tool for deciphering how the initial molecularinsult leads to the development of the early disease phenotype.Moreover, these cells can be examined in simplified in vitro sys-tems that mimic various aspects of the environment in the heart todissect how the local environment influences the development ofthe disease phenotype. It has been shown that cardiomyocytes canadapt their structural organization and contractility in response totheir local mechanical and geometric environment (21–23). It ispossible aberrant responses of cardiomyocytes to disease-relatedalterations in the local environment such as fibrosis-induced stiff-ening of the heart tissue could contribute to the development ofthe disease phenotype.To uncover the connection between molecular changes and

the development of the cellular disease phenotype, we deter-mined the molecular and cellular mechanisms of a DCM-causingpoint mutation in troponin-T (TNNT2), deletion of lysine 210(ΔK210) (24) (Fig. 1A). The ΔK210 mutation has been identi-fied in patients as young as 2 y old in at least 4 unrelated families(25–27). Troponin-T is a subunit of the troponin complex, which,together with tropomyosin, regulates the calcium-dependent

Significance

One of the outstanding challenges in understanding familial dilatedcardiomyopathy has been connecting mutation-induced changes insarcomeric protein functionwith the phenotype seen in cells. Manyof the mutation-induced changes in contractility at the molecularscale are subtle, begging the question of how they lead to a dev-astating progressive disease. Here, we show that disease-causingmutations in sarcomeric proteins affect not only contraction butalso how cardiomyocytes sense and respond to changes in theirmechanical environment associated with aging and the diseaseprogression. Our results implicate impaired mechanosensing as anunderappreciated mechanism in the pathogenesis and progressionof dilated cardiomyopathy, and they have important implicationsfor the design of precision therapeutics.

Author contributions: S.R.C., P.E.C., L.G., W.T.S., and M.J.G. designed research; W.T.S.contributed new reagents/analytic tools; S.R.C., P.E.C., L.G., M.E., and M.J.G. performedresearch; S.R.C., P.E.C., L.G., M.E., and M.J.G. analyzed data; and S.R.C. and M.J.G. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1S.R.C. and P.E.C. contributed equally to this work.2To whom correspondence may be addressed. Email: greenberg@wustl.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1910962116/-/DCSupplemental.

Published online August 19, 2019.

www.pnas.org/cgi/doi/10.1073/pnas.1910962116 PNAS | September 3, 2019 | vol. 116 | no. 36 | 17831–17840

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 3

, 202

1

interactions between the force-generating molecular motor my-osin and the thin filament. TNNT2 is one of the most frequentlymutated genes in DCM (28, 29). There have been several ex-cellent model systems developed to better understand aspects ofthe disease caused by this mutation (30–34); however, the con-nection between the molecular mutation and the cellular phe-notype seen in humans remains unclear.Our results demonstrate that ΔK210 causes a reduction in

myosin-based force generation at the molecular and cellularlevels, and we used computational modeling to help link thecontractile phenotypes seen at these scales. Surprisingly, wefound that this disease-causing mutation of a sarcomeric proteinnot only impairs cardiomyocyte contraction but also causes cel-lular hypertrophy and impairs the ability of cardiomyocytes tosense and respond to changes in their mechanical environment.Our results suggest a central role for defects in cardiomyocytemechanosensing in the disease pathogenesis of DCM.

ResultsΔK210 Decreases Calcium Sensitivity in an In Vitro Motility Assay.Weset out to decipher the molecular mechanism of the ΔK210 mu-tation in vitro. The molecular effects of cardiomyopathy mutationsdepend on the myosin heavy chain isoform (7–9, 35–37), andtherefore, we used porcine cardiac ventricular myosin (38). Por-cine ventricular cardiac myosin (MYH7) is 97% identical to hu-man, while murine cardiac myosin (MYH6) is only 92% identical.Unlike murine cardiac myosin, porcine cardiac myosin has verysimilar biophysical properties to human cardiac myosin, includingthe kinetics of the ATPase cycle, step size, and sensitivity to load(38–41), making it an ideal myosin for biophysical studies.Given the role of troponin-T in thin filament regulation, we

first determined whether the ΔK210 mutation affects calcium-based regulation of myosin’s interactions with thin filamentsusing an in vitro motility assay (42). Reconstituted thin filaments,consisting of porcine cardiac actin and recombinantly expressedhuman troponin complex and tropomyosin, were added to a flowcell coated with porcine cardiac myosin in the presence of ATP.The speed of filament translocation was measured as a functionof added calcium. As has been reported previously, the speed ofregulated thin filament translocation increased sigmoidally withincreasing calcium concentration (43) (Fig. 1B). Data were fitwith the Hill equation to obtain the pCa50 (i.e., the concentra-tion of calcium necessary for half-maximal activation). Consis-tent with previous studies using mouse cardiac, rabbit cardiac,and rabbit skeletal muscle fibers (31, 33, 44), ΔK210 shows aright-shifted curve (pCa50 = 5.7 ± 0.1) compared to the wild type(WT) (pCa50 = 6.1 ± 0.1; P < 0.0001), meaning more calcium isneeded for the same level of activation. This suggests that the mutantcould show impaired force production during a calcium transient.

Molecular Mechanism of ΔK210-Induced Changes in Thin FilamentRegulation. The changes in calcium sensitivity seen in the in vitromotility assay could be due to changes in the kinetics of myosinattachment and/or detachment from the thin filament. To testwhether the mutation affects the kinetics of myosin detachmentfrom thin filaments, we measured the rate of ADP release frommyosin, the transition that limits actomyosin dissociation at satu-rating ATP in the absence of load (45). Consistent with previousstudies of other DCM troponin mutations (46), ΔK210 does notchange the kinetics of ADP release from myosin (SI Appendix, Fig.S1). Therefore, the primary effects of the mutation are likely due toalterations in the kinetics of myosin attachment to the thin filament.The kinetics of myosin attachment to the thin filament are

regulated by myosin and the thin filament proteins troponin andtropomyosin. Biochemical (47) and structural biological (48, 49)experiments have shown that thin filament regulation is a com-plicated process where tropomyosin can lie along the thin filamentin 3 positions (blocked, closed, and open), and that this positioningis determined by both calcium and myosin binding (Fig. 2A). Todetermine the molecular mechanism of the observed changes inthe calcium sensitivity of myosin-based motility, we determined theequilibrium constants for the transitions between these states usingan approach pioneered by McKillop and Geeves (47).The equilibrium constant for the transition between the

blocked and closed states, KB, was measured using stopped-flowkinetic techniques (47), where the rate of myosin binding topyrene-labeled regulated thin filaments was measured at high(pCa 4) and low (pCa 9) calcium concentrations. Myosin strongbinding to labeled thin filaments quenches the pyrene fluores-cence, leading to a roughly exponential decrease in fluorescence(Fig. 2 B and C). The ratio of the observed rates for myosinbinding to the thin filament at high and low calcium can be usedto calculate the equilibrium constant, KB (47):

kobs�−Ca2+

�kobsð+Ca2+Þ ≈

KB

1+KB. [1]

The values of KB obtained were 0.36 ± 0.15 for WT (n = 3) and0.36 ± 0.10 for ΔK210 (n = 3), indicating that ΔK210 does notsignificantly affect the equilibrium constant for the transitionbetween the blocked and closed states (P = 0.99).To determine the equilibrium constant for the transition between

the closed and open states, KT, we measured the steady-state bindingof myosin to pyrene-labeled regulated thin filaments over a range ofmyosin concentrations (Fig. 2 D and E). Titration relationshipsobtained for WT and ΔK210 at high (pCa 3) and low (2 mMEGTA) calcium concentrations show 2-state (i.e., no calcium-basedblocking) and 3-state processes, respectively, as has been observedpreviously (47). Five technical replicates were used to define eachcurve. As described in SI Appendix, Materials and Methods, wemodified the fitting and analysis procedure used by McKillopand Geeves to better define the values of the fitted variables(50). We performed an additional titration at an intermediatecalcium concentration (pCa 6.25) and used an annealing algo-rithm to globally fit the binding curves of the fractional change inpyrene fluorescence ðaÞ as a function of myosin concentration forall 3 calcium concentrations (Fig. 2 D and E). The fractionalchange in the pyrene fluorescence is given by the following (47):

a=F0 −FF0 −F∞

=KW½M�Pn−1ðKTð1+KSÞn + 1Þ�KTP

n +Qn + 1KB

�ð1+KSÞn−1

, [2]

where [M] is the concentration of myosin, F is the measuredfluorescence, F0 is the fluorescence in the absence of myosin,F∞ is the fluorescence at saturating myosin concentrations, n isthe size of the cooperative unit, P = 1 + KW[M] (1 + KS), and

pCa

Nor

mal

ized

Spe

ed

0.6

0.2

0.4

0

1.0

0.8

9 8 7 6 5 4

WTK210

BA Troponin-CTroponin-ITroponin-TK210

Fig. 1. (A) Structure of the troponin core complex (Protein Data Bank ID4Y99), showing troponin-C (cyan), troponin-I (white), troponin-T (purple),and K210 of troponin-T (red). (B) Speed of thin filament translocation as afunction of calcium measured in the in vitro motility assay. Error bars showthe SD of n = 3 experiments. Data show a significant shift in thepCa50 toward submaximal calcium activation for ΔK210 (P < 0.0001). TheHill coefficient is not changed between the WT and ΔK210 (P = 0.45).

17832 | www.pnas.org/cgi/doi/10.1073/pnas.1910962116 Clippinger et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 3

, 202

1

Q = 1 + KW[M] (Fig. 2 D and E). For the fitting, KT, KW (i.e., theequilibrium constant between the open and myosin weaklybound states, sometimes denoted K1), and n were fit parameters.KB and KS (the equilibrium constant between the myosin weaklyand strongly bound states, sometimes denoted K2) were fixedbased on the calculated values for KB (Fig. 2 B and C) and pre-viously measured values of KS (51). For the fitting at pCa6.25 and 4, KB was fixed at 20, since the fit is insensitive to theexact value of KB at those calcium concentrations, as long as KBis greater than 5. 95% confidence intervals were calculated from1,000 rounds of bootstrapping simulations (SI Appendix, Mate-rials and Methods).We found that while the measured values of KT for ΔK210 were

consistently lower compared to the WT at all calcium concentra-tions (Fig. 3A), only the difference in KT at pCa 6.25 was statisticallysignificant (P = 0.03). From the measured equilibrium constants, itis straightforward to calculate the fraction of regulatory units ineach state from the partition function (47). At all calcium concen-trations, ΔK210 shows a decrease in the fraction of thin filaments inthe open, weakly bound, and strongly bound states compared to theWT (Fig. 3B and SI Appendix, Fig. S7), and an increase in thefraction of thin filaments in the blocked and closed states. Thesedata give the biochemical mechanism of the reduced activation at

subsaturating calcium seen in the in vitro motility assays (Fig. 1B)and would predict a reduction in the force per sarcomere in mutantmuscle due to a reduction in the fraction of myosin cross-bridges inthe strongly bound state.

Computational Modeling Recapitulates the Shift in Calcium Sensitivityand Predicts a Lower Force per Sarcomere for ΔK210. To predict howthe observed changes in the equilibrium populations of thin fil-ament regulatory units correspond with contractility in a sarcomere,we used a computational model developed by Campbell et al.(52). This model was chosen since it has many similarities to theMcKillop and Geeves biochemical model of muscle regulationused to analyze the biochemical experiments (47), although itshould be noted that there are some important differences(Discussion and SI Appendix, Materials and Methods). TheCampbell et al. model simulates an ensemble of sarcomeric reg-ulatory units that can transition between the blocked, closed,open, and myosin bound states based on the equilibrium constantsbetween these states and a constant describing the coupling be-tween adjacent regulatory units. This model enables the calcula-tion of both the expected steady-state force as a function ofcalcium and the expected force produced per sarcomere in re-sponse to a calcium transient. The details of the implementation

pCa 4 pCa 9

K210

0 42 181614121086 20

1.2

1.0

0.8

0.6

0.4

0.2

0

Nor

mal

ized

Flu

ores

cenc

e

Time (s)Time (s)0 42 181614121086 20

1.2

1.0

0.8

0.6

0.4

0.2

0Nor

mal

ized

Flu

ores

cenc

e WT

pCa 4 pCa 9

pCa 3pCa 6.25

2 mM EGTA0.2

0.4

0.6

0.8

1.0

0.1 1 100.010.1 1 100.010

0.2

0.4

0.6

0.8

1.0

pCa 3pCa 6.25

2 mM EGTA

K210WT

Frac

tion

Bou

nd

Frac

tion

Bou

nd

[Myosin S1] ( M)[Myosin S1] ( M)

B C

D E1.2 1.2

ADP.Pi

Ca+2 Pi

ADP.Pi ADP.Pi ADPCa+2 Ca+2 Ca+2 Ca+2 Ca+2 Ca+2

KB KT KSADP.Pi

Ca+2 Ca+2

KW

FORCE

Blocked Closed Open Weakly Bound Strongly Bound

A

0

Fig. 2. Determination of equilibrium constants governing thin filament regulation for ΔK210. (A) The 3-state model for tropomyosin positioning along thethin filament (47). Activation of the thin filament and subsequent force generation requires both calcium and myosin binding. The equilibrium constant forthe blocked to closed transition is KB, and the equilibrium constant for the closed to open transition is KT. KW and KS are the equilibrium constants for myosinweak and strong binding to the thin filament, respectively. (B and C) Normalized stopped-flow fluorescence traces of myosin binding to regulated pyrene-actin. The quenching of pyrene-actin upon binding of S1 occurs at a faster rate at pCa 4 (green) compared to pCa 9 (pink) due to the activating effect ofcalcium. Traces for (B) WT and (C) ΔK210 are very similar, and the calculated values of KB are not statistically different (P = 0.998). (D and E) Fluorescencetitrations in which increasing amounts of S1 myosin were gradually added to regulated thin filaments to a final concentration of 10 μM S1 in the presence ofADP. Titration curves at 3 different calcium concentrations are shown for both (C) WT and (D) ΔK210. Five repeats were performed for each condition, anderror bars show the SD. At low concentrations of myosin, more myosin is bound to the thin filament at higher calcium concentrations (pCa 6.25 [orange] andpCa 3 [green]) than at low calcium (2 mM EGTA, pink).

Clippinger et al. PNAS | September 3, 2019 | vol. 116 | no. 36 | 17833

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 3

, 202

1

of the model are described in Materials and Methods. Using ourmeasured equilibrium constants, the steady-state normalized forcewas simulated for both WT and ΔK210 over a range of calciumconcentrations (Fig. 4A). Simulation of ΔK210 shows a rightwardshift toward supermaximal calcium activation, consistent with theshift seen in the in vitro motility assay (Fig. 1B). Therefore, theobserved set of changes in the equilibrium constants in the mutantare consistent with the observed changes seen in the in vitromotility assay, validating the approach taken here.To predict the effects of the ΔK210 mutation on the force per

sarcomere, we used the same computational model to simulatethe force per sarcomere in response to a calcium transient. In themodeling, we assumed that the primary effects of the mutationare on thin filament positioning, and that to a first-order ap-proximation, the calcium transient is not significantly changed bythe mutation. The consequences of these assumptions are ex-plored in Discussion. The simulation predicts a smaller maximaltwitch force per sarcomere in response to a calcium transient forΔK210 compared to the WT (Fig. 4B). Taken together, thesesimulations predict that ΔK210 should decrease the force persarcomere in cardiomyocytes.

Generation of ΔK210-Induced Stem Cell-Derived Cardiomyocytes. Toexamine the effects of the ΔK210 mutation in human cells, wegenerated hiPSC-CMs. The hiPSC lines were derived from theBJ foreskin fibroblast line by the Washington University Ge-nome Engineering core (SI Appendix, Materials and Methods).Whole-exome sequencing of the stem cell line demonstrated thatthese cells do not have any genetic variants associated with fa-milial cardiomyopathies (SI Appendix, Fig. S2). Cell lines ho-mozygous for the ΔK210 mutation were generated using theCRISPR/Cas9 system (17) (SI Appendix, Fig. S3 and Materialsand Methods). ΔK210 mutant stem cells have normal karyotypes(SI Appendix, Fig. S4) and are pluripotent, as assessed by im-munofluorescence staining for pluripotency markers (SI Appen-dix, Fig. S5). For all experiments with the mutant hiPSC-CMs,2 separate stem cell lines were used to ensure that the obser-vations are not due to potential off-target cuts introduced duringthe CRISPR/Cas9 editing. We also used hiPSC-CMs from atleast 2 separate differentiations per cell line for both WT andΔK210. hiPSCs were differentiated to hiPSC-CMs using estab-lished protocols (15, 53) involving temporal modulation ofWNT signaling using small molecules (SI Appendix, Fig. S5) andaged at least 30 d before use. Using this procedure, >90%

cardiomyocytes were obtained, as determined by immunofluo-rescence staining for cardiac troponin-T (SI Appendix, Fig. S5).To examine the sarcomeric organization of hiPSC-CMs, cells

cultured on glass were fixed and stained for troponin-I andα-actinin to visualize the thin filament and z-disks, respectively(Fig. 5). Unlike adult human tissue-derived cardiomyocytes,which are rectangular with sarcomeres that align with the longaxis of the cell, WT hiPSC-CMs cultured on glass orient ran-domly and display robust sarcomeric staining that is typically notaligned along a single axis (Fig. 5A). Compared to the WThiPSC-CMs, ΔK210 hiPSC-CMs display more disorganized sar-comeres with patches of punctate staining (Fig. 5B). Similardisorganization and punctate sarcomere structure has been seenwith other hiPSC-CM models of DCM cultured on stiff sub-strates (6, 18, 54), suggesting that it might be a common featureof some forms of DCM.

ΔK210 hiPSC-CMs Show a Pronounced Increase in Size andMechanosensitive Alterations in Sarcomeric Structure. Cardiomyocytesare sensitive to their local physical environment, and it has beenshown that providing physical cues that mimic the environment ofthe heart can improve cardiomyocyte structure and contractility(21, 55). During aging and DCM disease progression, the heartbecomes stiffer, and the myocytes become disordered. Wehypothesized that ΔK210 hiPSC-CMs would show aberrant

ADP.Pi

Ca+2 Pi

ADP.Pi ADP.Pi ADPCa+2 Ca+2 Ca+2

KB KT KSADP.Pi

Ca+2

KW

FORCE

Blocked Closed Open Weakly Bound Strongly Bound

A

-50-40-30-20-100+10pCa >9

pCa 6.25

pCa 3

% C

hang

e C

ompa

red

to W

T

B+5.2%

W/S-66.7%

O-71.2%

C+5.5%

B+25.4%

W/S-57.6%

O-63.3%

C+25.4%

B+16.6%

W/S-16.0%

O-27.3%

C+16.6%

+20

-70-60

B

WT K210 pKB 0.36 (-0.15/+0.15) 0.36 (-0.10/+0.10) 0.99Kw 0.12 (-0.01/+0.02) 0.13 (-0.01/+0.01) 0.09

KT pCa >9 0.09 (-0.08/+0.12) 0.02 (-0.02/+0.03) 0.20KT pCa 6.25 0.14 (-0.09/+0.09) 0.04 (-0.02/+0.03) 0.03

KT pCa 3 0.28 (-0.18/+0.23) 0.17 (-0.09/+0.12) 0.40n 5.23 (-1.50/+3.12) 6.88 (-1.13/+1.54) 0.26

Fig. 3. Effects of ΔK210 on tropomyosin positioning along the thin filament. (A) The 3-state model for tropomyosin positioning along the thin filament (47)and parameter values obtained from the stopped-flow and fluorescence titration experiments. Reported uncertainties are 95% confidence intervals. KT forΔK210 at pCa 6.25 is statistically different from WT (P = 0.03). Error bars are the 95% confidence intervals. (B) Percent change in the occupancy of each statefor the mutant compared to the WT. Reds denote increased population of a state, while blues denote decreased population. ΔK210 causes a decrease in thepopulation of the states where myosin is bound to the thin filament and an increase in the population of the inhibited (i.e., blocked and closed) states. Notethat the fractional change in the population of thin filaments in the blocked and closed states is similar, since KB does not change between the WT and themutant.

Time (ms)

Sim

ulat

ed T

witc

h Fo

rce/

Cal

cium

0 400 800 1200 16004567890

1.0

0.8

0.6

0.4

0.2

pCa

Sim

ulat

ed N

orm

aliz

ed

Forc

e/S

arco

mer

e WTK210

A B

Fig. 4. Computational modeling of the steady-state and transient force persarcomere. (A) Using the model developed by Campbell et al. (52), thesteady-state normalized force per sarcomere as a function of calcium can besimulated for both ΔK210 and WT. The simulation shows a shift in thepCa50 toward supermaximal calcium activation in the mutant. (B) Simulatedtwitch forces per sarcomere in response to a calcium transient. The modelpredicts that ΔK210 should show a lower twitch force per sarcomere.

17834 | www.pnas.org/cgi/doi/10.1073/pnas.1910962116 Clippinger et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 3

, 202

1

responses to changes in their physical environment. To test thishypothesis, we examined the size and structure of cells on glass,which has a high stiffness (approximately gigapascals) and onhydrogel substrates with a stiffness matched to healthy hearttissue (10 kPa).To examine the role of geometric cues in cellular organization,

we micropatterned extracellular matrix for cellular adhesion inrectangular patterns with a 7:1 aspect ratio. It was previouslyshown that micropatterning of hiPSC-CMs improves cell maturityand the 7:1 aspect ratio optimizes force production (21). We ex-amined live cells using bright-field microscopy, and fixed cellsstained for the z-disk marker α-actinin and troponin-I using con-focal microscopy (Fig. 5 C–F). Interestingly, the ΔK210 hiPSC-CMs are significantly larger than the WT in fixed cells on glass(Figs. 5 and 6), fixed cells on 10-kPa hydrogels (Figs. 5 and 6), andin live cells on 10-kPa hydrogels (Fig. 7). We analyzed a large

number of cells and generated cumulative distributions of cellsizes for single cells, since this methodology does not assume aform for the underlying distribution (Figs. 6 and 7). The cumu-lative distribution for area shows the fraction of cells with an arealess than or equal to the value of the x axis. The increase in size inthe mutant is due to increases in both cell width and length (Table1 and Fig. 6). These data demonstrate that ΔK210 hiPSC-CMs arelarger than the WT cells, and that this increase in size is not de-pendent on the mechanical environment.Next, we examined the influence of substrate stiffness on the

sarcomeric organization of ΔK210 hiPSC-CMs. We fixed andstained micropatterned hiPSC-CMs for the sarcomeric markerα-actinin on both glass and 10-kPa substrates. Both WT andΔK210 hiPSC-CM sarcomeres orient preferentially along thelong axis of the cells patterned on glass and on 10-kPa hydrogels(Fig. 5). To quantify the degree of sarcomeric organization, we

A BC

D

FE

WT K210

WT K210

WT

K210

Fig. 5. Immunofluorescence images of sarcomeres in hiPSC-CMs. Troponin-I is red and α-actinin is green. All images are z-projections. (A) WT hiPSC-CM onglass. (B) ΔK210 hiPSC-CM on glass, showing sarcomeric disorganization. (C) WT cell on a rectangular pattern on glass. (D) ΔK210 cell on a rectangular patternon glass. (E) WT cell on a rectangular pattern on a 10-kPa hydrogel. (F) ΔK210 cell on a rectangular pattern on a 10-kPa hydrogel. The sarcomeric organizationis significantly improved on the hydrogel with a physiologically relevant stiffness.

Area ( m2)0 1000

1.0

0.5

0

Gla

ssC

umul

ativ

e Fr

actio

n

1500 2000500

1.0

0.5

010 k

Pa

Hyd

roge

lC

umul

ativ

e Fr

actio

n

Length ( m)0 80 12040

Width ( m)6 14 1810 22

OOP0 0.4 0.60.2

WT(n=120)K210(n=66)

WT(n=33)K210(n=38)

A

B

0

0

Fig. 6. Quantification of immunofluorescence staining for sarcomeres in hiPSC-CMs on rectangular patterns on (A) glass and (B) 10-kPa hydrogels. Data areplotted as cumulative distributions and parameter values can be found in Table 1. ΔK210 cells are significantly larger on both glass and 10-kPa hydrogels (P =0.00002), due to increases in both length and width. The orientation order parameter (OOP) is a measurement of sarcomeric organization. For a hiPSC-CM inwhich all of the sarcomeres are aligned along a single axis, the OOP = 1, and for a hiPSC-CM with no preferred axis, the OOP = 0. On glass, ΔK210 cells havelower OOP values than the WT (P = 0.008). On physiologically stiff hydrogels, there is no significant difference between the OOP of the WT and mutant cells(P = 0.5).

Clippinger et al. PNAS | September 3, 2019 | vol. 116 | no. 36 | 17835

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 3

, 202

1

used a program developed by Pasqualini et al. (55). The programanalyzes fluorescence images with periodic structure, and itcalculates the orientational order parameter (OOP) (i.e., thefraction of z-disks that are oriented along a single axis). For cellswith z-disks all oriented along a single axis, OOP = 1, and forcells with z-disks with no preferred axis, OOP = 0.When patterned on glass (gigapascal stiffness), the ΔK210

mutant hiPSC-CMs show a reduction in the fraction of z-disksaligned along the long axis compared to the WT (P = 0.008) (Fig.6A and Table 1). This is consistent with what we saw withunpatterned cells on glass (Fig. 5 A and B). Strikingly, whenpatterned on hydrogels that mimic the stiffness of healthy hearttissue, the ΔK210 cells show significantly improved z-disk orga-nization (Fig. 6B and Table 1). As quantified using the OOP, theorganization of the ΔK210 z-disks are indistinguishable from theWT on 10-kPa substrates (P = 0.5). Taken together, these resultsdemonstrate that the ΔK210 mutation affects how the hiPSC-CMs respond to their mechanical environment, with ΔK210 cellsshowing normal z-disk organization on physiologically stiff sub-strates, but reduced sarcomeric organization on stiff substrates.

ΔK210 Cells Show Altered Contractility and Features of Hypertrophyon Substrates of Physiological Stiffness. Our biochemical mea-surements (Fig. 3B) and computational modeling predict thatΔK210 cells should show a lower force per sarcomere (Fig. 4B);and therefore, we examined whether the ΔK210 mutation affectsforce production of single hiPSC-CMs. To measure force pro-duction of single hiPSC-CMs, we used traction force microscopy.Spontaneously beating cells were micropatterned onto 10-kPahydrogels prepared with embedded fluorescent microbeads tomonitor beating. Spinning disk confocal imaging was used tomonitor the bead displacement as a function of time, and thetotal force of contraction for each cell was calculated using aMATLAB routine developed by Ribeiro et al. (56) (Fig. 7 B and

C). For each cell, the force per sarcomere was approximated bycalculating the force per area.The average total force per cell measured by traction force

microscopy is the same for both WT [0.27 (−0.03/+0.04) μN] andΔK210 hiPSC-CMs [0.27 (−0.04/+0.05) μN; P = 0.98] (Fig. 7C).These forces are consistent with previous measurements (21).For each cell analyzed by traction force microscopy, we measuredits area from bright-field illumination images (Fig. 7A). Bright-field images of live cells lack the resolution of the immunofluo-rescence images, making it challenging to accurately measure thecross-sectional area of the live cells; however, ΔK210 hiPSC-CMsare significantly larger than the WT cells (Fig. 7A). Moreover,fixed ΔK210 hiPSC-CMs on 10-kPa hydrogels have a z-disk or-ganization that is indistinguishable from the WT (Fig. 6B). Wecalculated the force per area for each cell (Fig. 7B). The ΔK210cells have a lower force per area [0.25 (−0.04/+0.05) μN/μm2]compared to the WT [0.32 (−0.04/+0.04) μN/μm2; P = 0.01]. Thelower force per area for ΔK210 is consistent with our biochemicalmeasurements and the computational modeling, which predicted alower force per sarcomere (Fig. 4B). These data help link themolecular and cellular phenotypes. Moreover, they demonstratethat although the force per sarcomere is reduced in the mutant,the ΔK210 hiPSC-CMs can compensate through an increase insize that results in a total force per cell that is indistinguishablefrom the WT.

DiscussionHere, we determined the molecular mechanism of a mutation introponin-T that causes DCM in humans, ΔK210. We found thatthis mutation affects not only molecular and cellular contractility,but also the ability of the cardiomyocytes to respond to changesin the mechanical environment like stiffening of heart tissueassociated with aging and disease in patients. These results im-plicate defective mechanosensing by cardiomyocytes as an im-portant factor in the pathogenesis of DCM caused by mutations

Area ( m2)200 1400

1.0

0.5

0Cum

ulat

ive

Frac

tion

2000800Force/Area (10-3 N/ m2)

0 0.8 1.20.4Total Force ( N)

0 0.8 1.20.4

WT(n=160)K210(n=94)

A B C

WT (n=160) K210 (n=94) p<0.00002

0.010.98

Area ( m2)Force/area (10-3 N/ m2)

Total Force ( N)

900 (-40/+50) 1200 (-80/+80)0.32 (-0.04/+0.04) 0.25 (-0.04/+0.05)0.27 (-0.03/+0.04) 0.27 (-0.04/+0.05)

Fig. 7. Traction force microscopy of single hiPSC-CMs patterned on 10-kPa polyacrylamide gels. (A) Live ΔK210 hiPSC-CMs on hydrogels are significantlylarger than WT cells. (B) ΔK210 cells have a lower force per area (i.e., lower force per sarcomere), consistent with the molecular studies; however, since theyare larger, (C) their total force production is not significantly different from the WT cells. Reported uncertainties are 95% confidence intervals.

Table 1. Analysis of fixed cell sizes on rectangular patterns (reported uncertainties are 95% confidence intervals)

Patterned on glass Patterned on 10-kPa hydrogels

Parameter WT (n = 120) ΔK210 (n = 66) P WT (n = 33) ΔK210 (n = 38) P

Area, μm2 1,000 (−50/+50) 1,470 (−70/+70) 0.00002 770 (−80/+90) 990 (−60/+60) 0.0002Length, μm 91 (−3/+3) 108 (−3/+3) 0.00002 85 (−7/+7) 96 (−4/+4) 0.007Width, μm 13 (−0.4/+0.4) 17 (−0.5/+0.5) 0.00002 11 (−1/+1) 13 (−0.7/+0.8) 0.06Length/width ratio 6.7 (−0.3/+0.3) 6.4 (−0.2/+0.3) 0.1 7.6 (−0.9/+0.9) 7.6 (−0.6/+0.6) 0.9OOP 0.28 (−0.02/+0.02) 0.24 (−0.02/+0.02) 0.008 0.28 (−0.04/+0.04) 0.27 (−0.03/+0.03) 0.5

17836 | www.pnas.org/cgi/doi/10.1073/pnas.1910962116 Clippinger et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 3

, 202

1

in sarcomeric proteins, and they highlight the importance ofmultiscale studies in understanding heart disease.

Proposed Molecular Mechanism for ΔK210 and Its Relationship toPrevious Studies. The disease presentation of familial cardiomy-opathies and their biophysical manifestation (i.e., contractile andforce-dependent properties) can depend on the myosin isoform(7, 8, 57–60). Our in vitro motility results (Fig. 1B) using porcinecardiac actin and ventricular β-cardiac muscle myosin (MYH7),which closely mimics the biophysical and biochemical propertiesof human ventricular β-cardiac muscle myosin (38, 39), demon-strate a shift toward supermaximal calcium activation, consistentwith previous experiments studying the ΔK210 mutation usingdifferent myosin isoforms (31, 44).We determined that this shift is due to alterations in the

equilibrium constants that govern the positioning of tropomyosinalong the thin filament, leading to a reduction in the populationof force-generating cross-bridges at submaximal calcium con-centrations (Figs. 3 and 4A). The structural basis of thesemutation-induced shifts is not well understood; however, it ispossible that the mutation affects the interaction between thetroponin complex and tropomyosin (61) and/or the allostericcoupling between subunits of the troponin complex (2, 5, 62, 63).These nonexclusive mechanisms could contribute to the ob-served decrease in force-producing states seen in ΔK210 thinfilaments, and future studies should be able to shed light on thestructural basis of these changes.Our biochemical data were analyzed using the formalism de-

veloped by McKillop and Geeves (47). These data clearly dem-onstrate a decrease in the population of strongly bound myosincross-bridges in the mutant (Fig. 3B and SI Appendix, Fig. S7),suggesting that the mutant would show reduced force production.We predicted the effects of the mutations on sarcomeric contrac-tility using a computational model developed by Campbell et al.(52) that is similar, but not identical to the McKillop and Geevesmodel (47). For example, the McKillop and Geeves model has acalcium dependence for KT, while this term is calcium independentin the Campbell et al. model. Although these models are not exactlyequivalent, the computational modeling was able to recapitulate theshift in calcium sensitivity seen in the motility assay (Figs. 1B and4A). Other models of thin filament regulation exist (64–68), butthe McKillop and Geeves model is one of the most frequentlyused. Regardless of the model used to interpret the results, ourbiochemical data clearly show molecular hypocontractility inthe mutant at micromolar calcium concentrations.The computational modeling based on our biochemical studies

predicts a larger reduction in the force per sarcomere than weobserved in hiPSC-CMs (Fig. 7). This is likely due to severalnecessary simplifying assumptions made in the modeling: 1) Weassumed that the mutation does not change the magnitude ortime course of the calcium transient, which is not necessarily truegiven that the knockin ΔK210 mouse shows altered calciumtransients (31). 2) We assumed that the troponin in hiPSC-CMshas the same biochemical properties as the adult cardiac troponincomplex that we examined in our biochemical experiments. This isnot necessarily the case, since 4 troponin-T isoforms can be de-tected in the human heart at different developmental stages (69,70), and hiPSC-CMs primarily express a slow skeletal muscleisoform of troponin-I (71), rather than the cardiac isoform used inthe biochemical studies. While all isoforms of troponin expressedin hiPSC-CMs would have the ΔK210 mutation (SI Appendix, Fig.S3), different isoforms have different calcium sensitivities (69).That said, even with these simplifying assumptions, the simulationswere able to predict the reduction in the force per sarcomere seenin the ΔK210 cardiomyocytes.

Linking the Molecular and Cellular Contractile Phenotypes for ΔK210.Patients with the ΔK210 mutation show early onset of the dis-

ease phenotype and a high incidence of sudden death (26). As-pects of the disease have been replicated in knockin mouse models(31), which have significantly advanced our understanding of thedisease phenotype. However, the ΔK210 mouse model showssignificant differences with respect to the human phenotype, dueto inherent physiological differences between mouse and humanhearts, including differences in gene expression, calcium-handlingmachinery, and ion channel composition (34).We used human hiPSC-CMs to study how the initial insult of a

molecular-based change leads to the early disease pathogenesisin human cells. For these studies, hiPSC-CMs are excellentmodels, since they can capture cellular changes that occur beforemajor compensatory mechanisms (e.g., fibrosis, activation ofneurohormonal pathways, changes in gene expression) mask theinitial molecular phenotype (18–20, 54, 72). Contractile changesassociated with the disease have been observed in cardiomyop-athy patients before adverse remodeling of the heart (73–75).That being said, hiPSC-CMs are developmentally immature,meaning that they do not necessarily recapitulate all of the as-pects of the phenotype in adults or the later stages of the diseaseprogression. All of our molecular and cellular studies modeledthe homozygous ΔK210 mutation, so care should be used whenextrapolating the results here to the phenotype seen in patients.ΔK210 hiPSC-CMs on patterned hydrogels show a reduced

force per area compared to WT cells (Fig. 7B) without any sig-nificant differences in the z-disk organization (Fig. 6B). These dataare consistent with the reduced force per sarcomere predictedfrom our biochemical experiments (Fig. 3B) and computationalmodeling (Fig. 4B), linking the molecular phenotype with cellularcontractility. It should be noted that our studies only examined thez-disk organization and that it is possible that the mutation couldaffect the expression of other sarcomeric proteins.Interestingly, while the force per sarcomere is reduced in the

mutant, the total force per cell is not different from the WT (Fig.7C). This can be explained by the fact that the mutant cells arelarger (Figs. 6 and 7), a result that would not have been predictedfrom the molecular studies alone. This increase in size occurs insingle cells, independent of endocrine or cell–cell signaling. Onepossible explanation for this increase in size is that the cells areundergoing compensatory hypertrophy, which allows the cells tocounteract the reduced force per sarcomere. While we have notinvestigated whether the increase in size seen in our cells is dueto activation of hypertrophic signaling pathways, future studiesshould elucidate the mechanism of the increase in cell size seenhere. Recent work from Molkentin and coworkers (76) hasshown that reduced total tension developed by cardiomyocytes isstrongly correlated with the development of dilated cardiomy-opathy in both hiPSC-CMs and mice; however, it is also possiblethat the increase in cell size could be due to alterations incalcium handling.

Defects in ΔK210 Mechanosensing Contribute to the Disease Phenotype.Fundamentally, troponin serves as a gate that regulates myosin-based tension. As such, one would expect that the primary ef-fects of the mutation would be to reduce force production;however, the ΔK210 cells show several changes beyond alteredcontractility. While healthy cardiomyocytes adapt their con-tractility and structure to changes in their mechanical envi-ronment, such as age or disease-related stiffening of the hearttissue (21), ΔK210 cells do not adapt in the same way as theWT cells. Similar to other hiPSC-CM models of DCM grownon stiff substrates, ΔK210 hiPSC-CMs cultured on glass showimpaired sarcomeric organization (18, 54, 77); however, ΔK210hiPSC-CMs z-disks can organize normally when given mecha-nobiological cues that mimic the healthy heart (Figs. 5 and 6).Therefore, the ΔK210 mutation affects not only contractilitybut also the ability of the hiPSC-CMs to sense and respond totheir environment.

Clippinger et al. PNAS | September 3, 2019 | vol. 116 | no. 36 | 17837

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 3

, 202

1

The connection between the reduction in myosin-generatedtension in the mutant cells and an impaired ability to respond tomechanical forces warrants further investigation; however, a possiblelink between cellular tension and sarcomeric organization in DCMwas revealed by live-cell imaging of sarcomerogenesis in hiPSC-CMs(54). It was shown that pharmacological inhibition of myosin con-tractility or genetic disruption of titin leads to an impaired ability ofcells to generate well-organized sarcomeres. Chopra et al. proposedthat the transduction of myosin-driven tension is necessary for theproper assembly of sarcomeres in hiPSC-CMs. We propose that asimilar mechanism could be relevant in ΔK210 hiPSC-CMs, whichshow reductions in myosin-based contractility and z-disk disorgani-zation when patterned on glass. Consistent with this idea, treat-ment of hiPSC-CMs containing a different DCM mutation withomecamtiv mecarbil, a myosin-based thin filament activator (41, 78,79), leads to an improvement in sarcomeric organization (77). Wespeculate that some of the salutary effects of the drug are due to therestoration of tension necessary for proper mechanosensing.The result that ΔK210 hiPSC-CMs show altered responses to

changes in their mechanical environment has important impli-cations for the mechanism of the disease pathogenesis. Previousstudies of cardiomyocytes have shown that the organization ofsarcomeres varies depending on substrate stiffness (21, 77), withpeak contractility occurring on substrates of physiological stiff-ness. ΔK210 hiPSC-CMs can organize properly and they havenormal force production on substrates that match the stiffness ofthe healthy heart (Figs. 6 and 7); however, heart tissue becomesstiffer with the disease progression and aging. We propose thatas the heart tissue becomes stiffer, ΔK210 cells would becomemore disordered, effectively causing a progressive loss of func-tion, and potentially contributing to myocyte disarray. Thus, theinability of the mutant cells to adapt to changes in their me-chanical environment may contribute to the disease pathogenesisand progression.

The Potential Importance of Mechanosensing in DCM Pathogenesis.Our results suggest that reduced force at the molecular scale dueto a mutation in a sarcomeric protein can lead to impairedmechanosensing in cardiomyocytes. While the disease pre-sentation in DCM depends on the exact mutation (80), wepropose that disruption of mechanosensing in cardiomyocytescould be a common mechanism in the disease pathogenesis ofother sarcomeric and nonsarcomeric DCM mutations. It hasbeen proposed that DCM can be caused by molecular hypo-contractility (4, 76, 81); however, there are also many DCM-causing mutations in nonsarcomeric genes with no known rolesin contractility; suggesting that other mechanisms might be in-volved (1, 3, 82). Most of these genes are located along proposedmechanosensing pathways (83, 84) used to sense and transducemechanical forces. For example, some of these genes link thecytoskeleton to the extracellular matrix (e.g., dystrophin [DMD],vinculin [VCL]), others link the cytoskeleton to the nucleus (e.g.,nesprin-1 [SYNE1], lamin-A [LMNA], centromere protein F[CENP-F]), and others are mechanosensitive transcription fac-tors (e.g., TAZ). Consistent with this idea, altered mechanobi-ology has been identified in DCM cells with mutations in thenonsarcomeric protein lamin A/C (82). Moreover, patients withmutations in these genes, such as those with Duchenne musculardystrophy and Emery–Dreifuss muscular dystrophy, often de-velop DCM. It is intriguing to speculate that impaired mecha-nosensing by cardiomyocytes in the heart could link some ofthese seemingly different diseases and provide a common path-way that could be targeted by therapeutic interventions.

ConclusionsUsing a combination of biochemical, computational, and cell-biological techniques, we reveal several features of the earlydisease pathogenesis of a DCM mutation that would have been

missed by studying the molecular or cellular phenotypes alone.We demonstrate that the ΔK210 mutation in human cardiactroponin-T causes a change in the equilibrium positioning oftropomyosin, leading to alterations in the calcium sensitivity ofthin filament activation and a reduction in the number of stronglybound myosin cross-bridges at all calcium concentrations. Com-putational modeling predicts that these changes should lead to areduction in the force generated per sarcomere in mutant cells,and we demonstrate that this is indeed the case in hiPSC-CMs.We demonstrate that the ΔK210 hiPSC-CMs can increase in sizeto normalize their force production. Moreover, our results dem-onstrate that the ΔK210 mutation affects the structural organi-zation of hiPSC-CMs and alters how they respond to theirmechanical environment. This impaired mechanosensing likelycontributes to the DCM disease progression and the developmentof myocyte disarray as the heart stiffens. Taken together, theseresults demonstrate that disease-causing mutations of sarcomericproteins affect not only contraction but also how cardiomyocytessense and respond to changes in their mechanical environmentassociated with aging and disease. These data also suggest thatdisrupting mechanosensing pathways can contribute to the diseasephenotype in DCM.

Materials and MethodsInvestigation of the Molecular Mechanism Using Recombinant Proteins. Humantroponin and tropomyosin were expressed recombinantly in Escherichia coli,and cardiac actin and myosin were purified from porcine ventricles. In vitromotility assays were conducted as previously described (36). The equilibriumconstants that govern thin filament activation were determined usingstopped-flow and steady-state fluorescence techniques (47, 50). Full detailsare provided in SI Appendix, Materials and Methods.

Computational Simulations. Simulations of force production by sarcomereswere conducted using the model developed by Campbell et al. (52). Fulldetails are provided in SI Appendix, Materials and Methods.

Differentiation of hiPSCs to hiPSC-CMs. The parent stem cell line, BJFF.6, wasgenerated from the human BJ fibroblast line (CRL-2522, ATCC) by the Ge-nome Engineering and iPSC Center atWashington University in St. Louis. Twoindependent stem cell lines homozygous for the ΔK210 deletion weregenerated using the CRISPR/Cas9 system (17, 85). Differentiation to hiPSC-CMs was accomplished using the method of Lian et al. (53). Full details areprovided in SI Appendix, Materials and Methods.

Analysis of Sarcomeric Structure in Cardiomyocytes. hiPSC-CMs were fixed in4% paraformaldehyde (Electron Microscopy Sciences), and sarcomericstructure was visualized via immunofluorescence using confocal microscopy(Washington University Center for Cellular Imaging). Analysis of the orga-nization of the sarcomeres was conducted using software developed byParker and coworkers (55). Full details are provided in SI Appendix,Materialsand Methods.

Analysis of Single Cardiomyocyte Contractility. Single hiPSC-CMs were seededonto rectangular patterns on hydrogels for traction force microscopyaccording to the protocol of Ribeiro et al. (21). All traction force microscopymeasurements were performed on patterned polyacrylamide hydrogels witha stiffness of 10 kPa with hiPSC-CMs that were at least 30 d post-differentiation. Single, spontaneously beating cells were imaged in an en-vironmentally controlled chamber (Tokai Hit) on a Nikon spinning diskconfocal microscope (Washington University Center for Cellular Imaging).Traction force movies were analyzed by generating displacement maps of thefluorescent beads using a MATLAB program developed by the Pruitt and co-workers (56). Full details are provided in SI Appendix, Materials and Methods.

ACKNOWLEDGMENTS. We thank Stuart Campbell and Francesco Pasqualinifor sharing their code for the simulations and sarcomeric structure analysis,respectively. We acknowledge Mike Ostap and Ken Margulies for extremelyhelpful discussions during the early planning phases of this project. We alsoacknowledge Drew Braet for technical assistance and Samantha Barrick forassistance with biochemical experiments. We acknowledge the WashingtonUniversity Institute of Materials Science and Engineering for the use ofmicrofabrication instruments and staff assistance, and we thank the Alvin J.Siteman Cancer Center at Washington University School of Medicine and

17838 | www.pnas.org/cgi/doi/10.1073/pnas.1910962116 Clippinger et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 3

, 202

1

Barnes-Jewish Hospital (St. Louis, MO) for the Genome Engineering andInduced Pluripotent Stem Cell Center, which provided the genome editingservice (National Cancer Institute Cancer Center Support Grant P30 CA091842).Confocal microscopy was performed through the Washington University Cen-ter for Cellular Imaging supported by Washington University School of Med-icine, The Children’s Discovery Institute of Washington University and St. LouisChildren’s Hospital (CDI-CORE-2015-505), and the Foundation for Barnes-

Jewish Hospital (3770). Exome sequencing was performed by the McDonnellGenome Institute. Funding for this project was provided by a pilot grant fromthe Children’s Discovery Institute of Washington University and St. Louis Chil-dren’s Hospital, the Washington University Center for Cellular Imaging (CDI-CORE-2015-505), the National Institutes of Health (R00HL123623 andR01HL141086 [to M.J.G.]; T32EB018266 [to S.R.C.]), and the March ofDimes Foundation (FY18-BOC-430198 [to M.J.G.]).

1. E. M. McNally, L. Mestroni, Dilated cardiomyopathy: Genetic determinants andmechanisms. Circ. Res. 121, 731–748 (2017).

2. M. L. Lynn, S. J. Lehman, J. C. Tardiff, Biophysical derangements in genetic cardio-myopathies. Heart Fail. Clin. 14, 147–159 (2018).

3. E. M. McNally, J. R. Golbus, M. J. Puckelwartz, Genetic mutations and mechanisms indilated cardiomyopathy. J. Clin. Invest. 123, 19–26 (2013).

4. J. A. Spudich, Hypertrophic and dilated cardiomyopathy: Four decades of basic re-search on muscle lead to potential therapeutic approaches to these devastating ge-netic diseases. Biophys. J. 106, 1236–1249 (2014).

5. M. R. Williams, S. J. Lehman, J. C. Tardiff, S. D. Schwartz, Atomic resolution probe forallostery in the regulatory thin filament. Proc. Natl. Acad. Sci. U.S.A. 113, 3257–3262(2016).

6. J. T. Hinson et al., HEART DISEASE. Titin mutations in iPS cells define sarcomere in-sufficiency as a cause of dilated cardiomyopathy. Science 349, 982–986 (2015).

7. S. Lowey, V. Bretton, J. Gulick, J. Robbins, K. M. Trybus, Transgenic mouse α- andβ-cardiac myosins containing the R403Q mutation show isoform-dependent transientkinetic differences. J. Biol. Chem. 288, 14780–14787 (2013).

8. S. J. Ford, R. Mamidi, J. Jimenez, J. C. Tardiff, M. Chandra, Effects of R92 mutations inmouse cardiac troponin T are influenced by changes in myosin heavy chain isoform. J.Mol. Cell. Cardiol. 53, 542–551 (2012).

9. S. Lowey et al., Functional effects of the hypertrophic cardiomyopathy R403Q mu-tation are different in an alpha- or beta-myosin heavy chain backbone. J. Biol. Chem.283, 20579–20589 (2008).

10. H. He, M. M. Javadpour, F. Latif, J. C. Tardiff, J. S. Ingwall, R-92L and R-92Wmutationsin cardiac troponin T lead to distinct energetic phenotypes in intact mouse hearts.Biophys. J. 93, 1834–1844 (2007).

11. J. Roh, N. Houstis, A. Rosenzweig, Why don’t we have proven treatments for HFpEF?Circ. Res. 120, 1243–1245 (2017).

12. G. Conceição, I. Heinonen, A. P. Lourenço, D. J. Duncker, I. Falcão-Pires, Animal modelsof heart failure with preserved ejection fraction. Neth. Heart J. 24, 275–286 (2016).

13. A. Sacco et al., Short telomeres and stem cell exhaustion model Duchenne musculardystrophy in mdx/mTR mice. Cell 143, 1059–1071 (2010).

14. L. Fananapazir, M. C. Dalakas, F. Cyran, G. Cohn, N. D. Epstein, Missense mutations inthe beta-myosin heavy-chain gene cause central core disease in hypertrophic car-diomyopathy. Proc. Natl. Acad. Sci. U.S.A. 90, 3993–3997 (1993).

15. X. Lian et al., Robust cardiomyocyte differentiation from human pluripotent stemcells via temporal modulation of canonical Wnt signaling. Proc. Natl. Acad. Sci. U.S.A.109, E1848–E1857 (2012).

16. I. Kehat et al., Human embryonic stem cells can differentiate into myocytes withstructural and functional properties of cardiomyocytes. J. Clin. Invest. 108, 407–414(2001).

17. M. Jinek et al., A programmable dual-RNA-guided DNA endonuclease in adaptivebacterial immunity. Science 337, 816–821 (2012).

18. N. Sun et al., Patient-specific induced pluripotent stem cells as a model for familialdilated cardiomyopathy. Sci. Transl. Med. 4, 130ra47 (2012).

19. C. K. Lam, J. C. Wu, Disease modelling and drug discovery for hypertrophic cardio-myopathy using pluripotent stem cells: How far have we come? Eur. Heart J. 39, 3893–3895 (2018).

20. K. Musunuru et al.; American Heart Association Council on Functional Genomics andTranslational Biology; Council on Cardiovascular Disease in the Young; and Council onCardiovascular and Stroke Nursing, Induced pluripotent stem cells for cardiovasculardisease modeling and precision medicine: A scientific statement from the AmericanHeart Association. Circ. Genom. Precis. Med. 11, e000043 (2018).

21. A. J. Ribeiro et al., Contractility of single cardiomyocytes differentiated from plurip-otent stem cells depends on physiological shape and substrate stiffness. Proc. Natl.Acad. Sci. U.S.A. 112, 12705–12710 (2015).

22. T. K. Feaster et al., Matrigel mattress: A method for the generation of single con-tracting human-induced pluripotent stem cell-derived cardiomyocytes. Circ. Res. 117,995–1000 (2015).

23. Y. Jiang, P. Park, S. M. Hong, K. Ban, Maturation of cardiomyocytes derived fromhuman pluripotent stem cells: Current strategies and limitations. Mol. Cells 41, 613–621 (2018).

24. R. E. Hershberger et al., Clinical and functional characterization of TNNT2 mutationsidentified in patients with dilated cardiomyopathy. Circ Cardiovasc Genet 2, 306–313(2009).

25. J. Mogensen et al., Severe disease expression of cardiac troponin C and T mutations inpatients with idiopathic dilated cardiomyopathy. J. Am. Coll. Cardiol. 44, 2033–2040(2004).

26. M. Kamisago et al., Mutations in sarcomere protein genes as a cause of dilated car-diomyopathy. N. Engl. J. Med. 343, 1688–1696 (2000).

27. E. L. Hanson et al., Cardiac troponin T lysine 210 deletion in a family with dilatedcardiomyopathy. J. Card. Fail. 8, 28–32 (2002).

28. S. Dewan, K. J. McCabe, M. Regnier, A. D. McCulloch, Insights and challenges of multi-scale modeling of sarcomere mechanics in cTn and Tm DCM mutants-genotype tocellular phenotype. Front. Physiol. 8, 151 (2017).

29. A. C. Garfinkel, J. G. Seidman, C. E. Seidman, Genetic pathogenesis of hypertrophicand dilated cardiomyopathy. Heart Fail. Clin. 14, 139–146 (2018).

30. M. Memo et al., Familial dilated cardiomyopathy mutations uncouple troponin Iphosphorylation from changes in myofibrillar Ca2+ sensitivity. Cardiovasc. Res. 99, 65–73 (2013).

31. C. K. Du et al., Knock-in mouse model of dilated cardiomyopathy caused by troponinmutation. Circ. Res. 101, 185–194 (2007).

32. Q. W. Lu et al., Cardiac troponin T mutation R141W found in dilated cardiomyopathystabilizes the troponin T-tropomyosin interaction and causes a Ca2+ desensitization. J.Mol. Cell. Cardiol. 35, 1421–1427 (2003).

33. S. Morimoto et al., Ca2+-desensitizing effect of a deletion mutation Delta K210 incardiac troponin T that causes familial dilated cardiomyopathy. Proc. Natl. Acad. Sci.U.S.A. 99, 913–918 (2002).

34. F. Ahmad et al., The role of cardiac troponin T quantity and function in cardiac de-velopment and dilated cardiomyopathy. PLoS One 3, e2642 (2008).

35. M. Chandra, M. L. Tschirgi, S. J. Ford, B. K. Slinker, K. B. Campbell, Interaction betweenmyosin heavy chain and troponin isoforms modulate cardiac myofiber contractiledynamics. Am. J. Physiol. Regul. Integr. Comp. Physiol. 293, R1595–R1607 (2007).

36. M. J. Greenberg, K. Kazmierczak, D. Szczesna-Cordary, J. R. Moore, Cardiomyopathy-linked myosin regulatory light chain mutations disrupt myosin strain-dependentbiochemistry. Proc. Natl. Acad. Sci. U.S.A. 107, 17403–17408 (2010).

37. M. J. Greenberg et al., Regulatory light chain mutations associated with cardiomy-opathy affect myosin mechanics and kinetics. J. Mol. Cell. Cardiol. 46, 108–115 (2009).

38. M. J. Greenberg, H. Shuman, E. M. Ostap, Inherent force-dependent properties ofβ-cardiac myosin contribute to the force-velocity relationship of cardiac muscle. Bio-phys. J. 107, L41–L44 (2014).

39. J. Sung et al., Harmonic force spectroscopy measures load-dependent kinetics of in-dividual human β-cardiac myosin molecules. Nat. Commun. 6, 7931 (2015).

40. J. C. Deacon, M. J. Bloemink, H. Rezavandi, M. A. Geeves, L. A. Leinwand, Identifi-cation of functional differences between recombinant human α and β cardiac myosinmotors. Cell. Mol. Life Sci. 69, 2261–2277 (2012). Correction in: Cell. Mol. Life Sci. 69,4239–4255 (2012).

41. M. S. Woody et al., Positive cardiac inotrope omecamtiv mecarbil activates muscledespite suppressing the myosin working stroke. Nat. Commun. 9, 3838 (2018).

42. S. J. Kron, J. A. Spudich, Fluorescent actin filaments move on myosin fixed to a glasssurface. Proc. Natl. Acad. Sci. U.S.A. 83, 6272–6276 (1986).

43. E. Homsher, B. Kim, A. Bobkova, L. S. Tobacman, Calcium regulation of thin filamentmovement in an in vitro motility assay. Biophys. J. 70, 1881–1892 (1996).

44. P. Robinson et al., Alterations in thin filament regulation induced by a human cardiactroponin T mutant that causes dilated cardiomyopathy are distinct from those in-duced by troponin T mutants that cause hypertrophic cardiomyopathy. J. Biol. Chem.277, 40710–40716 (2002).

45. M. Barany, ATPase activity of myosin correlated with speed of muscle shortening. J.Gen. Physiol. 50 (suppl), 197–218 (1967).

46. R. F. Sommese et al., Effects of troponin T cardiomyopathy mutations on the calciumsensitivity of the regulated thin filament and the actomyosin cross-bridge kinetics ofhuman β-cardiac myosin. PLoS One 8, e83403 (2013).

47. D. F. McKillop, M. A. Geeves, Regulation of the interaction between actin and myosinsubfragment 1: Evidence for three states of the thin filament. Biophys. J. 65, 693–701(1993).

48. W. Lehman, R. Craig, P. Vibert, Ca2+-induced tropomyosin movement in Limulus thinfilaments revealed by three-dimensional reconstruction. Nature 368, 65–67 (1994).

49. P. Vibert, R. Craig, W. Lehman, Steric-model for activation of muscle thin filaments. J.Mol. Biol. 266, 8–14 (1997).

50. S. K. Barrick, S. R. Clippinger, L. Greenberg, M. J. Greenberg, Computational tool tostudy perturbations in muscle regulation and its application to heart disease. Biophys.J. 116, 2246–2252 (2019).

51. R. Maytum, B. Westerdorf, K. Jaquet, M. A. Geeves, Differential regulation of theactomyosin interaction by skeletal and cardiac troponin isoforms. J. Biol. Chem. 278,6696–6701 (2003).

52. S. G. Campbell, F. V. Lionetti, K. S. Campbell, A. D. McCulloch, Coupling of adjacenttropomyosins enhances cross-bridge-mediated cooperative activation in a Markovmodel of the cardiac thin filament. Biophys. J. 98, 2254–2264 (2010).

53. X. Lian et al., Directed cardiomyocyte differentiation from human pluripotent stemcells by modulating Wnt/β-catenin signaling under fully defined conditions. Nat.Protoc. 8, 162–175 (2013).

54. A. Chopra et al., Force generation via beta-cardiac myosin, titin, and alpha-actinindrives cardiac sarcomere assembly from cell-matrix adhesions. Dev. Cell 44, 87–96.e5(2018).

55. F. S. Pasqualini, S. P. Sheehy, A. Agarwal, Y. Aratyn-Schaus, K. K. Parker, Structuralphenotyping of stem cell-derived cardiomyocytes. Stem Cell Reports 4, 340–347(2015).

56. A. J. S. Ribeiro et al., Multi-imaging method to assay the contractile mechanicaloutput of micropatterned human iPSC-derived cardiac myocytes. Circ. Res. 120, 1572–1583 (2017).

Clippinger et al. PNAS | September 3, 2019 | vol. 116 | no. 36 | 17839

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 3

, 202

1

57. S. Nag et al., Contractility parameters of human β-cardiac myosin with the hyper-

trophic cardiomyopathy mutation R403Q show loss of motor function. Sci. Adv. 1,

e1500511 (2015).58. M. J. Tyska et al., Single-molecule mechanics of R403Q cardiac myosin isolated from

the mouse model of familial hypertrophic cardiomyopathy. Circ. Res. 86, 737–744

(2000).59. K. A. Palmiter et al., R403Q and L908V mutant beta-cardiac myosin from patients with

familial hypertrophic cardiomyopathy exhibit enhanced mechanical performance at

the single molecule level. J. Muscle Res. Cell Motil. 21, 609–620 (2000).60. M. J. Greenberg, G. Arpag, E. Tüzel, E. M. Ostap, A perspective on the role of myosins

as mechanosensors. Biophys. J. 110, 2568–2576 (2016).61. J. P. Jin, S. M. Chong, Localization of the two tropomyosin-binding sites of troponin T.

Arch. Biochem. Biophys. 500, 144–150 (2010).62. L. Sfichi-Duke et al., Cardiomyopathy-causing deletion K210 in cardiac troponin T

alters phosphorylation propensity of sarcomeric proteins. J. Mol. Cell. Cardiol. 48,

934–942 (2010).63. B. Liu, S. B. Tikunova, K. P. Kline, J. K. Siddiqui, J. P. Davis, Disease-related cardiac

troponins alter thin filament Ca2+ association and dissociation rates. PLoS One 7,

e38259 (2012).64. A. Houmeida, D. H. Heeley, B. Belknap, H. D. White, Mechanism of regulation of

native cardiac muscle thin filaments by rigor cardiac myosin-S1 and calcium. J. Biol.

Chem. 285, 32760–32769 (2010).65. S. M. Mijailovich et al., Resolution and uniqueness of estimated parameters of a

model of thin filament regulation in solution. Comput. Biol. Chem. 34, 19–33 (2010).66. D. A. Smith, M. A. Geeves, J. Sleep, S. M. Mijailovich, Towards a unified theory of

muscle contraction. I: Foundations. Ann. Biomed. Eng. 36, 1624–1640 (2008).67. T. Longyear, S. Walcott, E. P. Debold, The molecular basis of thin filament activation:

From single molecule to muscle. Sci. Rep. 7, 1822 (2017).68. T. L. Hill, E. Eisenberg, L. Greene, Theoretical model for the cooperative equilibrium

binding of myosin subfragment 1 to the actin-troponin-tropomyosin complex. Proc.

Natl. Acad. Sci. U.S.A. 77, 3186–3190 (1980).69. A. V. Gomes, G. Guzman, J. Zhao, J. D. Potter, Cardiac troponin T isoforms affect the

Ca2+ sensitivity and inhibition of force development. Insights into the role of troponin

T isoforms in the heart. J. Biol. Chem. 277, 35341–35349 (2002).70. B. Iorga et al., Differences in contractile function of myofibrils within human em-

bryonic stem cell-derived cardiomyocytes vs. adult ventricular myofibrils are related

to distinct sarcomeric protein isoforms. Front. Physiol. 8, 1111 (2018).

71. S. Kadota, L. Pabon, H. Reinecke, C. E. Murry, In vivo maturation of human inducedpluripotent stem cell-derived cardiomyocytes in neonatal and adult rat hearts. StemCell Reports 8, 278–289 (2017).

72. M. J. Birket et al., Contractile defect caused by mutation in MYBPC3 revealed underconditions optimized for human PSC-cardiomyocyte function. Cell Rep. 13, 733–745(2015).

73. J. Forsey, L. Benson, E. Rozenblyum, M. K. Friedberg, L. Mertens, Early changes inapical rotation in genotype positive children with hypertrophic cardiomyopathymutations without hypertrophic changes on two-dimensional imaging. J. Am. Soc.Echocardiogr. 27, 215–221 (2014).

74. C. Y. Ho et al., Assessment of diastolic function with Doppler tissue imaging to predictgenotype in preclinical hypertrophic cardiomyopathy. Circulation 105, 2992–2997(2002).

75. I. K. Rüssel et al., Increased left ventricular torsion in hypertrophic cardiomyopathymutation carriers with normal wall thickness. J. Cardiovasc. Magn. Reson. 13, 3 (2011).

76. J. Davis et al., A tension-based model distinguishes hypertrophic versus dilated car-diomyopathy. Cell 165, 1147–1159 (2016).

77. K. M. Broughton et al., A myosin activator improves actin assembly and sarcomerefunction of human-induced pluripotent stem cell-derived cardiomyocytes with atroponin T point mutation. Am. J. Physiol. Heart Circ. Physiol. 311, H107–H117 (2016).

78. L. Nagy et al., The novel cardiac myosin activator omecamtiv mecarbil increases thecalcium sensitivity of force production in isolated cardiomyocytes and skeletal musclefibres of the rat. Br. J. Pharmacol. 172, 4506–4518 (2015).

79. F. I. Malik et al., Cardiac myosin activation: A potential therapeutic approach forsystolic heart failure. Science 331, 1439–1443 (2011).

80. M. A. Burke, S. A. Cook, J. G. Seidman, C. E. Seidman, Clinical and mechanistic insightsinto the genetics of cardiomyopathy. J. Am. Coll. Cardiol. 68, 2871–2886 (2016).

81. J. R. Moore, L. Leinwand, D. M. Warshaw, Understanding cardiomyopathy pheno-types based on the functional impact of mutations in the myosin motor. Circ. Res. 111,375–385 (2012).

82. P. Bhattacharjee, D. Dasgupta, K. Sengupta, DCM associated LMNA mutations causedistortions in lamina structure and assembly. Biochim. Biophys. Acta, Gen. Subj. 1861,2598–2608 (2017).

83. C. Uhler, G. V. Shivashankar, Regulation of genome organization and gene expressionby nuclear mechanotransduction. Nat. Rev. Mol. Cell Biol. 18, 717–727 (2017).

84. T. Iskratsch, H. Wolfenson, M. P. Sheetz, Appreciating force and shape—The rise ofmechanotransduction in cell biology. Nat. Rev. Mol. Cell Biol. 15, 825–833 (2014).

85. P. D. Hsu et al., DNA targeting specificity of RNA-guided Cas9 nucleases. Nat. Bio-technol. 31, 827–832 (2013).

17840 | www.pnas.org/cgi/doi/10.1073/pnas.1910962116 Clippinger et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 3

, 202

1