Phosphomimetic cardiac myosin-binding protein C partially rescues a cardiomyopathy phenotype in murine engineered heart tissue

Phosphorylation of cardiac myosin-binding protein C (cMyBP-C), encoded by MYBPC3, increases the availability of myosin heads for interaction with actin thus enhancing contraction. cMyBP-C phosphorylation level is lower in septal myectomies of patients with hypertrophic cardiomyopathy (HCM) than in non-failing hearts. Here we compared the effect of phosphomimetic (D282) and wild-type (S282) cMyBP-C gene transfer on the HCM phenotype of engineered heart tissues (EHTs) generated from a mouse model carrying a Mybpc3 mutation (KI). KI EHTs showed lower levels of mutant Mybpc3 mRNA and protein, and altered gene expression compared with wild-type (WT) EHTs. Furthermore, KI EHTs exhibited faster spontaneous contractions and higher maximal force and sensitivity to external [Ca2+] under pacing. Adeno-associated virus-mediated gene transfer of D282 and S282 similarly restored Mybpc3 mRNA and protein levels and suppressed mutant Mybpc3 transcripts. Moreover, both exogenous cMyBP-C proteins were properly incorporated in the sarcomere. KI EHTs hypercontractility was similarly prevented by both treatments, but S282 had a stronger effect than D282 to normalize the force-Ca2+-relationship and the expression of dysregulated genes. These findings in an in vitro model indicate that S282 is a better choice than D282 to restore the HCM EHT phenotype. To which extent the results apply to human HCM remains to be seen.


RT-PCR and RT-qPCR.
Total RNA was extracted from single EHTs with the TRIzol reagent (Invitrogen, 300 μl/EHT) following the manufacturers' instructions and further processed as described before 17,25 . RT-qPCR was performed using either SYBR Green or TaqMan probes and the TaqMan ABI Prism 7900HT sequence detection system (Applied Biosystems). Messenger RNA levels were measured in triplicates and normalized to Gnas. Quantification of mRNA levels was done using the 2 −ΔΔCt method and related to WT values. Primers are given in Suppl. Table S2.
Gene expression analysis. Gene expression analysis was performed with a customized nanoString's nCounter Element Tag-set panel of 84 genes coding for proteins involved in hypertrophy, Ca 2+ , K + and Na + handling and constituents of the sarcomere or other intermediate filaments. Same quantity of RNAs of three EHTs in each group were pooled, and 50 ng were used for the run on the nCounter Sprint Profiler following manufacturer's instructions. Raw data were analyzed with the nSolver software using 4 housekeeping genes for normalization (Abcf1, Cltc, Gapdh and Pgk1). Data were expressed as fold-change over WT-EHT levels, only genes that were lower than 0.7-fold or higher than 1.3-fold expressed in non-transduced KI EHTs were selected. Heatmaps were generated with the freeware Morpheus (https://software.broadinstitute.org/morpheus/). No statistical analysis was performed since n = 1 pool in each group.
Immunofluorescence analysis of EHTs. Whole EHTs were washed in PBS and fixed with Roti-Histofix (Carl Roth GmbH, P087) at 4 °C overnight. After removal from the posts, EHTs were incubated in blocking solution at 4 °C overnight (10% FCS, 1% BSA, 0.5% Triton X-100 in PBS). After permeabilization, EHTs were stained at 4 °C overnight in blocking solution without FCS with primary antibodies directed against the FLAG epitope (1:800; Sigma #F3165), cMyBP-C (1:200; M-motif gift from C. Witt, Heidelberg) and α-actinin (1:200; Sigma #A7811). Anti-mouse IgG Alexa 488-conjugated (1:800; Life Technologies #A11029) and anti-rabbit IgG Alexa 546-conjugated (1:800; Life Technologies #A11035) were used as secondary antibodies. Nuclei were stained with DRAQ5 (1:1,000; BioStatus). EHTs were finally mounted with Mowiol-488 between object slide and coverslips. Confocal images were acquired with a Zeiss LSM 710 system. Statistical analysis. All data were expressed as mean ± SEM and analyzed with the software GraphPad Prism 8. Statistical analysis comparing the effect of isoprenaline in the same group was done using the paired Student's t-test. More than two groups were statistically compared by a one-way ANOVA analysis followed by Dunnett's multiple comparison test. Force-calcium relationship data were analyzed using the Hill equation, with EC 50 as the free Ca 2+ concentration, which yields 50% of the maximal force and nH representing the Hill coefficient. Curves were fitted to the data points and force-pCa relationship comparison was done by using extra sum-of-squares F-test 48 . A value of P < 0.05 was considered statistically significant.

D282 and S282 cMyBP-C gene transfer prevents accumulation of mutant mRNAs and restores protein levels.
To evaluate the molecular impact of phosphomimetic (D282) and wild-type (S282) Mybpc3 AAV-mediated gene transfer, engineered heart tissues (EHTs) were derived from cardiac cells isolated from neonatal Mybpc3-targeted KI mice and transduced with a multiplicity of infection (MOI) of 1,000 vg/cell. After 14 to 16 days of culture, WT, non-transduced KI (KI-NT) and KI EHTs transduced with either S282 (KI-S282) or D282 (KI-D282) AAVs were collected for characterization of the molecular phenotype on mRNA and protein levels ( Fig. 1). Qualitative RT-PCR of pooled RNAs showed only one band (=WT) in the WT sample and the expected three bands, corresponding to the different Mybpc3 mutant transcripts, in the KI-NT sample 15 . In contrast, only the upper band corresponding to WT was still visible in KI EHTs transduced with S282 or D282. A specific band matching FLAG-Mybpc3 transcript was amplified only in transduced KI EHTs (Fig. 1a). Quantification of total Mybpc3 mRNA level by RT-qPCR revealed a lower level in KI-NT than in WT (only 25% of WT level), whereas the total Mybpc3 mRNA level was 6-and 3.5-fold higher in KI-S282 and KI-D282 than in WT, respectively (Fig. 1b). To evaluate the amount of Mybpc3 WT and mutant mRNAs, a RT-qPCR with specific Taqman probes www.nature.com/scientificreports www.nature.com/scientificreports/ for WT, missense or frameshift Mybpc3 mRNAs was performed as previously described 15 . The WT Mybpc3 transcript was identified in WT as well as in S282-or D282-transduced KI samples but at higher level in the latter two. While no WT Mybpc3 mRNA was detected in KI-NT samples, the missense Mybpc3 mRNA was amplified only in KI-NT, but not in Mybpc3-transduced KI samples (Fig. 1c,d). Frameshift Mybpc3 mRNAs were barely detectable in KI-S282 and KI-D282 (Fig. 1e). Expression levels of genes coding for proteins modulating hypertrophy, Ca 2+ / K + /Na + -handling and constituents of the sarcomere were predominantly higher in KI-NT EHTs, whereas S282 and D282 gene transfer partially restored them to WT levels ( Fig. 1f, Suppl. Table S4). In most cases, normalization of gene expression levels was more pronounced after gene transfer with S282 than D282, as in the case of RyR2 encoding cardiac ryanodine receptor 2.
By Western blotting exogenous FLAG-tagged cMyBP-C protein was detected only in transduced KI EHTs (Fig. 1g). Total cMyBP-C level was significantly lower in KI-NT than in WT EHTs (20% of WT level), corresponding mainly to full-length missense cMyBP-C, whereas no significant difference to WT level was seen in both KI-S282 and KI-D282 EHTs, although high variability were observed between samples (Fig. 1g,h).
To assess the effect of permanent non-reversible phosphorylation-like state of Ser-282 on the phosphorylation level of the other serine residues in the M-motif of cMyBP-C, EHTs were stimulated with 100 nM isoprenaline for 15 min prior to harvesting. Western blots with specific antibodies directed against the three phosphorylated serine residues (pSer-273; pSer-282; pSer-302) showed stronger signals after treatment at all phosphorylation sites especially in the WT and in KI-S282 EHTs. Overall cMyBP-C phosphorylation levels, both at baseline and after treatment, were very low in KI-NT. In KI-D282 EHTs no phosphorylation at Ser-282 neither at baseline nor after isoprenaline stimulation was detected confirming the amino acid substitution and disruption of the pSer-282 epitope. Phosphorylation at Ser-273 was barely detected in KI-NT and transduced samples at baseline and only a faint band was visible after isoprenaline-mediated PKA activation. Ser-302 phosphorylation was similarly increased by isoprenaline stimulation in all groups (Fig. 1i).
Phosphomimetic cMyBP-C is properly incorporated into the sarcomere. The localization of exogenous cMyBP-C proteins was explored by immunofluorescence analyses. The signal intensity of endogenous cMyBP-C was markedly lower in KI-NT than in all other groups, reflecting the lower amount of protein as determined by Western blot (Fig. 2a). Signal intensity did not differ between WT and AAV-transduced KI EHTs. Moreover, almost all α-actinin positive cells were also FLAG-positive, suggesting a nearly complete transduction of EHTs at this MOI. In transduced KI EHTs, the exogenous cMyBP-C protein was well incorporated in the sarcomeres showing doublets at the A-band in regular alternation with α-actinin at the Z-discs (Fig. 2b). Simultaneous staining of FLAG-cMyBP-C and total cMyBP-C showed co-localization (Suppl. Fig. 1).

Spontaneous contractions are faster in KI EHTs and are normalized with S282 and D282
cMyBP-C gene transfer. We then evaluated the impact of D282 and S282 Mybpc3 AAV-mediated gene transfer on contractile parameters of EHTs. EHTs generated from Mybpc3 WT and KI mouse cardiac cells showed a similar development after viral transduction. Beating areas within the murine EHTs were visible already 3 days after casting and coherent contractions of the whole EHTs with deflection of the posts resulting in generation of measurable forces at day 5. Generally, maximal force development was reached between days 12-16 and continuously declined thereafter in every group independently of the genotype or viral transduction (data not shown). Therefore, video-optical recording of contractile parameters was performed between days 13 and 15 ( Fig. 3a,b, Table 1). As quality control, only EHTs beating with a frequency of ≥50 bpm and producing forces ≥15 µN were taken into account for the analysis. Spontaneous beating activity in each group was measured in EHT culture medium containing 1.8 mM external [Ca 2+ ]. In this condition, KI-NT developed a tendency towards higher force than WT EHTs, although non-significant with one-way ANOVA. However, with the Student's t-test force was higher in KI-NT than in WT EHTs (P < 0.05). Transduction of KI EHTs with either S282 or D282 AAV did not affect maximal force development ( Fig. 3c; Table 1). The beating frequency on the day of maximal force development was recorded in all groups and showed a Gaussian distribution. KI-NT beat at significantly higher frequencies than WT EHTs (Table 1), whereas viral transduction with either S282 or D282 construct in KI EHTs normalized beating rate to WT EHTs ( Fig. 3d; Table 1). Contraction time (T1 20% ) and relaxation time (T2 20% ) were significantly shorter in KI-NT than in WT (−17% and −16%, respectively) and did not differ between transduced KI and WT EHTs (Fig. 3e,f; Table 1). Even though contraction velocity (CV) and relaxation velocity (RV) showed higher values for KI-NT, they were not significantly different between the groups (Suppl. Fig. 2a,b).
Electrical pacing reveals hypercontractility in KI EHTs, which is similarly prevented by S282 and D282 cMyBP-C gene transfer. Since in this study the known hypercontractile phenotype (i.e. higher force, accelerated velocities of contraction and relaxation) of KI-NT EHTs was blunted under spontaneous beating due to the increased beating rate (negative force-frequency relationship), EHTs of all genotypes were subjected to electrical pacing at a murine physiological frequency of 6 Hz 17,28 . At 1.8 mM external [Ca 2+ ], KI-NT showed 1.4-fold higher maximal force than WT EHTs ( Fig. 4a; Table 2). Maximal force of contraction was lower in KI-S282 than in KI-NT EHTs (P < 0.05 with Student's t-test) and did not significantly differ from WT EHTs. On the other hand, although the mean maximal force developed by KI-D282 EHTs was halfway between KI-NT and WT EHTs, there was a high variability among samples, excluding significant difference from the other groups ( Fig. 4a; Table 2). No major differences in contraction (T1 20% ) and relaxation (T2 20% ) times were measured in the different groups (Fig. 4b,c; Table 2). Accordingly, contraction and relaxation velocities matched the measured forces (Suppl. Fig. 2c,d; Table 2). In addition to contractile parameters, the sensitivity of the different groups toward external [Ca 2+ ] was investigated. KI-NT EHTs exhibited a left shift of the force-calcium curve, indicating higher [Ca 2+ ] sensitivity than the WT (EC 50   www.nature.com/scientificreports www.nature.com/scientificreports/ measurements showed a positive inotropic effect in WT EHTs (2.8-fold increased force vs. baseline), whereas the response was blunted in KI-NT EHTs (1.2-fold increased force vs. baseline; Fig. 4e). Gene transfer with S282 and D282 partially restored the isoprenaline-response of KI-NT to WT level (1.7-fold increased force vs. baseline). In addition, WT, KI-S282 and KI-D282 EHTs showed a shorter relaxation time (T2 20% ) fitting well with the positive lusitropic effect of isoprenaline (Suppl. Fig. 3). However, during the treatment many EHTs, independently of the genotype, became arrhythmic and were excluded from the analysis.

Discussion
It has been shown previously that substitution of the cMyBP-C serine phosphorylation sites with aspartic acid residues (mimicking activated protein) protects against ischemia-reperfusion injury 21 , whereas substitution with alanine residues (mimicking deactivated protein) results in cardiac hypertrophy with myofibrillar disarray and fibrosis in mice 33 . Overall cMyBP-C phosphorylation levels are reduced in HCM patients and animal models 34-36 , thus increasing cMyBP-C phosphorylation could be a therapeutic option. The goal of this study was to evaluate www.nature.com/scientificreports www.nature.com/scientificreports/ whether gene therapy with a phosphomimetic cMyBP-C protein (D282) would be more beneficial than wild-type cMyBP-C protein (S282) to prevent the molecular and functional disease phenotype in a well-established EHT model derived from Mybpc3-targeted KI mice. The main findings of this work are as follows: i) AAV6-mediated transduction with D282 or S282 similarly suppressed accumulation of mutant Mybpc3 transcripts and restored cMyBP-C protein levels. ii) Both exogenous cMyBP-C proteins were correctly incorporated into the sarcomere in KI EHTs. iii) Spontaneous contractions were faster in KI EHTs and were normalized by S282 and D282 cMyBP-C gene transfer. iv) Electrical stimulation at murine physiological frequency revealed hypercontractility in KI EHTs, which was prevented at least by S282 cMyBP-C gene transfer. v) The effect of S282 on normalization of the force-Ca 2+ -relationship and expression of genes encoding proteins involved in hypertrophy, Ca 2+ -handling, K + / Na + -regulation and sarcomere components was stronger than with D282.
Phosphorylation is a reversible post-translational modification. To investigate its role in protein function a new protein can be engineered carrying an aspartic acid (D) instead of serine (S; here D282 vs S282 cMyBP-C). This substitution introduces a negative charge similarly to the phosphoryl group, however D is smaller and have a different geometry therefore the engineered protein should not be addressed as constitutively phosphorylated. The resulting protein might be considered as a gain of function being a constitutively active protein. AAV6-mediated gene transfer of KI-NT EHTs with either D282-or S282-Mybpc3 resulted in higher amounts of Mybpc3 mRNAs than in WT EHT, but same protein levels. In addition, both exogenous proteins were correctly incorporated in the sarcomere. That sarcomeric proteins are maintained in a strict stoichiometric balance is well known 37 . It has been shown that mRNA transcripts, translation and protein degradation of sarcomeric components occur directly on site at the sarcomere and, by working in unison, these mechanisms ensure correct spatial localization and overcome the large variability in transcription of sarcomeric proteins 38 . Therefore, the stoichiometric maintenance is regulated at all steps from gene transcription to protein degradation. Our molecular data are in agreement with previous findings obtained in KI EHT 25 and in EHTs derived from Mybpc3-targeted knock-out mice (KO EHTs 17 ). We also confirmed in the present study that gene transfer with a full length Mybcp3 sequence prevents the accumulation of mutant transcripts in KI EHT and showed that it did not differ between the two types of construct. Therefore, both treatments have the same effect on haploinsufficiency and poison polypeptides, the two pathomechanisms involved in Mybpc3-related cardiomyopathy.
Cardiac MyBP-C, located transversally in doublets in the C-zone of the A-bands, acts as a "brake", a structural barrier for the cross-bridge cycling preventing the binding of myosin heads to the thin filaments 39 . Thus, cMyBP-C stabilizes a population of myosin heads in the inactive super-relaxed state (SRX), keeping them closely bound to the thick filament and unavailable to generate force 40 . Another molecular model is that the N-terminal domains of cMyBP-C modulates the thin filament activity by displacing tropomyosin and therefore enhancing myosin binding to actin 41,42 . Beside reducing the affinity of the N-terminal domains for the thin filaments 43 , phosphorylation of cMyBP-C at Ser-282 plays a primary role in SRX regulation, releasing myosin heads from the inhibitory state 44 . Low level or absence of cMyBP-C is expected to untether the myosin heads and relieves them from the SRX, thereby enhancing contraction. In KI cardiomyocytes, the level of cMyBP-C (corresponding mainly to missense protein) represents only 10% of its level in WT. Compared to the normal situation, more myosin heads can bind to actin resulting in more force generation. It has been reported before that EHTs derived from Mybpc3-targeted KI and KO mice exhibit higher maximal force development, accelerated rates of both contraction and relaxation and higher sensitivity to external [Ca 2+ ] compared to WT EHTs (=hypercontractile phenotype), both under spontaneous contraction and electrical stimulation 17,25,28,45 . In the present work, one-way ANOVA comparison did not reveal higher force, but shorter contraction and relaxation times and higher bpm in KI-NT EHTs under spontaneous contraction ( Fig. 3; Table 1). However, significant higher force in KI-NT was revealed by applying a Student's t-test between KI-NT and WT EHTs. Pacing unmasked the hypercontractile phenotype (higher force of contraction) of KI EHTs without change in T1 20% and T2 20% (Table 2). Therefore, this model revealed either hypercontractility under pacing or faster kinetics of contraction and relaxation under spontaneous contraction. The faster relaxation was not expected since we previously reported prolonged relaxation in intact adult KI cardiomyocytes and diastolic dysfunction in vivo 35 . On the other hand, our data are in line with accelerated kinetics of contraction and/or higher force development in Mybpc3 KO EHTs 28,45,46 and in Mybpc3 KI EHTs 17 . The discrepancy between the data obtained in EHT and in cardiac myocytes or in vivo in KI mice could be related to the different experimental conditions and/or age of the mice. Indeed, in EHT, cardiac cells are derived from neonatal mice and the muscle strips develop force against silicon posts in an auxotonic manner. On the contrary, intact cardiomyocytes derived from adult heart contract in an isotonic manner. Groups were compared with the one-way ANOVA, followed by Dunnet's multiple comparison test vs WT. *P < 0.05 and ** P < 0.01 vs WT EHTs.  www.nature.com/scientificreports www.nature.com/scientificreports/ Moreover, EHTs as an in vitro model do not present remodeling of the myocytes or of the entire tissue such as in vivo in the whole heart. The prevention of the EHT disease phenotype after Mybpc3 D282 and S282 gene transfer was expected in both conditions since the amount of cMyBP-C protein was completely restored, and we recently showed that ≥73% of wild-type cMyBP-C in a Mybpc3 null background is enough to overcome the disease phenotype 17 . Increased myofilament [Ca 2+ ] sensitivity is a well-accepted hallmark of HCM as seen in human samples 11,36,47,48 and in HCM animal models 15,17,35,[49][50][51][52][53] . Accordingly, KI-NT showed higher sensitivity to external [Ca 2+ ] than WT EHTs. AAV-mediated Mybpc3 gene transfer reduced the sensitivity to external [Ca 2+ ] towards WT values. However, the effect of D282 was not as pronounced as with S282. A nonspecific effect of transduction per se was ruled out with an empty virus (only promoter, no transgene; data not shown).
Isoprenaline cannot give rise to phosphorylation of the D282 but the S282-cMyBP-C. Ser-282 is a highly conserved amino acid in cMyBP-C and located in the vicinity of three more serine residues (Ser-273, Ser-302 and Ser-307) in the M-motif of the protein. These residues are targets of several protein kinases for phosphorylation following β-adrenergic stimulation 6 . Several groups have tried to decipher the role of each single cMyBP-C phosphorylation site and their interplay using transgenic mice generated ad hoc 22,23,54,55 . While it is well accepted that phosphorylation of Ser-282 plays a leading role, its effect on phosphorylation of the other residues in the motif is not completely clear 22,23 . In our study, while isoprenaline stimulation induced a similar, but strong phosphorylation at Ser-302 in both gene therapy approaches, the effect on Ser-273 was clearly smaller, suggesting that beside PKA other kinases might phosphorylate this residue. Isoprenaline showed a positive inotropic effect in all groups except KI-NT EHTs. The blunted response to the drug in the latter is in line with previous findings 28 , and may be due to the fact that most of the cMyBP-C-mediated brake is absent in KI and basal force is already higher. Phosphorylation of cMyBP-C induces bending of the M-motif and structural changes of the N-terminus of the protein, which reveals a stretch of amino acid residues that could be part of a yet-unknown protein-protein interaction site 43 . The PKA-mediated phosphorylation of cMyBP-C could result in increased myosin binding to the thin filament or stronger activation of thin filaments, in both cases ultimately leading to a stronger positive inotropic effect 56 .
Interestingly, the main difference between the two treatments concerned gene expression. Up-regulation of several genes involved in hypertrophic signaling pathway (e.g. Myh7), Ca 2+ -handling (e.g. Cacna1c and Ryr2), K + /Na + -regulation (e.g. Kcnip2) and sarcomere function (e.g. Des) present in KI-NT was prevented by both treatments, but to a lower extent with D282. This only partial correction of gene expression after D282 gene transfer could contribute to the smaller impact on the reduction of the hypercontractile KI-NT EHT phenotype of D282 compared to S282 EHTs, since more L-type calcium channels and ryanodine receptors would result in more intracellular [Ca 2+ ] and therefore contraction.

Limitations of the Study
The number of paced EHTs included in the analyses was low, particularly after isoprenaline treatment, making impossible any powerful statistical conclusion. Nonetheless, our data are consistent with previous findings showing a lower positive inotropic response in KI-EHTs than in WT-EHTs after isoprenaline treatment 28 . A second limitation is the phenotype of KI EHTs per se. This in vitro three-dimensional model derived from cardiac cells of KI mice showed higher force at baseline compared to WT. In contrast, in vivo echocardiographic analysis of KI mice showed impaired cardiac function, as revealed by lower fractional area shortening and E'/A' ratio 15,35 , resembling more the cardiac phenotype seen in HCM patients. In addition, it should be noted that whereas the majority of MYBPC3-related HCM patients carries the mutation at the heterozygous state, the KI-EHT used in the present study are derived from mice carrying the Mybpc3 mutation at the homozygous state. As a matter of fact the heterozygous Mybpc3 knock-in mice, which would reflect the human genetic state, do not develop left ventricular hypertrophy 15,35 . conclusion KI EHTs exhibited hypercontractility, which was similarly prevented by AAV-mediated transduction of wild-type S282 or phosphomimetic D282 cMyBP-C. However, the impact of D282 on normalization of the force-Ca 2+ relationship and gene expression profile of proteins modulating hypertrophy, Ca 2+ -, K + /Na + -handling and constituents of the sarcomere was not as pronounced as for S282, even though protein steady-state levels and localization did not differ. Future work has to explore whether the advantage of S282 compared to D282 in preventing the HCM phenotype extends to in vivo situations.