ACTN2 Mutant Causes Proteopathy in Human iPSC-Derived Cardiomyocytes

Genetic variants in α-actinin-2 (ACTN2) are associated with several forms of (cardio)myopathy. We previously reported a heterozygous missense (c.740C>T) ACTN2 gene variant, associated with hypertrophic cardiomyopathy, and characterized by an electro-mechanical phenotype in human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs). Here, we created with CRISPR/Cas9 genetic tools two heterozygous functional knock-out hiPSC lines with a second wild-type (ACTN2wt) and missense ACTN2 (ACTN2mut) allele, respectively. We evaluated their impact on cardiomyocyte structure and function, using a combination of different technologies, including immunofluorescence and live cell imaging, RNA-seq, and mass spectrometry. This study showed that ACTN2mut presents a higher percentage of multinucleation, protein aggregation, hypertrophy, myofibrillar disarray, and activation of both the ubiquitin-proteasome system and the autophagy-lysosomal pathway as compared to ACTN2wt in 2D-cultured hiPSC-CMs. Furthermore, the expression of ACTN2mut was associated with a marked reduction of sarcomere-associated protein levels in 2D-cultured hiPSC-CMs and force impairment in engineered heart tissues. In conclusion, our study highlights the activation of proteolytic systems in ACTN2mut hiPSC-CMs likely to cope with ACTN2 aggregation and therefore directs towards proteopathy as an additional cellular pathology caused by this ACTN2 variant, which may contribute to human ACTN2-associated cardiomyopathies.


Introduction
α-actinin-2 (ACTN2) is a component of the sarcomere in skeletal and cardiac myocytes, which forms anti-parallel homodimers that can anchor and crosslink actin thin filaments to the Z-disk (for reviews, see [1,2]). Additionally, ACTN2 is implicated in assembling large Both ACTN2wt and ACTN2mut lines were derived from the previously described heterozygous ACTN2 HCM hiPSC line (ACTN2het; c.740C>T; [8]) using CRISPR/Cas9 gene editing and homology-directed repair. Initially, both hiPSC lines were intended to produce an isogenic control without the mutation. However, sequencing analysis revealed only one functional wild-type allele in ACTN2wt [19] and only one functional mutant allele in ACTN2mut ( Figure S1A). The other allele of both cell lines contained on-target defects of CRISPR/Cas9 (splice-site mutation in ACTN2wt [19] and large rearrangement in ACTN2mut ( Figure S1A,B)), leading to nonsense mRNAs, which are not visible by RNA-seq ( Figure S2D). Thus, we created two heterozygous functional knock-out lines with a second wild-type and missense ACTN2 allele, respectively. Both ACTN2wt and ACTN2mut hiPSC lines presented a normal karyotype ( Figure S1C) and were therefore differentiated to hiPSC-CMs according to our protocol ( Figure S1D). HiPSC-CMs were produced with high purity (on average >90% cardiac troponin T (TNNT2)-positive cells) and quantity ( Figure S2A).
Both hiPSC-CM lines were evaluated for ACTN2 abundance and localization, myofibrillar disarray, cellular hypertrophy and multinucleation. Immunofluorescence analysis revealed a cross-striated pattern of TNNT2 in both hiPSC-CM lines, indicating the proper formation of sarcomeres ( Figure 1A). However, ACTN2 was less organized and formed aggregates in ACTN2mut when compared to ACTN2wt hiPSC-CMs. Quantification revealed a higher index of myofibrillar disarray and more ACTN2 aggregates in ACTN2mut hiPSC-CMs ( Figure 1B,C). Further, cell area and volume were higher in ACTN2mut hiPSC-CMs ( Figure 1D,E). Finally, the percentage of multinucleated (>1 nucleus) hiPSC-CMs was higher in ACTN2mut than in ACTN2wt ( Figure 1F). The mononucleated-to-multinucleated ratio was 80:20 in ACTN2wt hiPSC-CMs, supporting the previous estimation obtained in human hearts [20], whereas it was 53:47 in ACTN2mut, as reported in other HCM hiPSC-CMs [21]. ACTN2 mRNA level did not differ between the two hiPSC-CM lines  Figure S2C,E), whereas ACTN2 protein level was markedly lower in ACTN2mut hiPSC-CMs ( Figure S2F,G).

Exogenous Mutant ACTN2 Causes Aggregate Formation Leading to Sarcomere Disarray
We then tested whether exogenous mutant ACTN2 could induce protein aggregation in living, 2D-cultured ACTN2wt hiPSC-CMs. Therefore, ACTN2wt hiPSC-CMs were transduced with AAV6 carrying the MUT-ACTN2-HaloTag ® (c.740T) and were compared to ACTN2mut hiPSC-CMs transduced with AAV6 carrying the WT-ACTN2-HaloTag ® (c.740C). After 7 days of culture in 96-well plates, live-cell imaging experiments were performed by staining ACTN2-HaloTag protein using TMR-ligand in combination with Hoechst ( Figure 2A; and examples in Videos S1 and S2). Exogenous MUT-ACTN2 in ACTN2wt and WT-ACTN2 in ACTN2mut reversed the phenotypes, inducing aggregation in about 83% and 17% of hiPSC-CMs, respectively (Figure 2A,B). This indicates that MUT-ACTN2 causes aggregation in ACTN2wt and WT-ACTN2 reverses aggregation in ACTN2mut. Quantification of acquired live cell images for ACTN2 aggregates revealed more and larger aggregates in MUT-ACTN2-transduced ACTN2wt than in WT-ACTN2transduced ACTN2mut ( Figure 2C). Western blot analysis revealed only the endogenous ACTN2 in MUT-ACTN2-transduced ACTN2wt ( Figure 2D,E), suggesting a very low level of correctly folded MUT-ACTN2, which should be visible as a larger molecular weight protein. Conversely, both exogenous and endogenous ACTN2 were detected in WT-ACTN2transduced ACTN2mut hiPSC-CMs, with about 46% of replacement of endogenous by exogenous ACTN2 (Figure 2F,G).
Immunofluorescence analysis of hiPSC-CMs transduced with WT-or MUT-ACTN2 using antibodies directed against the HaloTag ® and total ACTN2 revealed co-localization of WT-ACTN2-HaloTag ® and total ACTN2 staining in ACTN2mut, confirming Z-disk integration of exogenous WT-ACTN2, whereas exogenous MUT-ACTN2 was barely detectable in ACTN2wt and exhibited co-localization with total ACTN2 in some parts ( Figure 2H).

ACTN2mut hiPSC-CMs Exhibit Alterations of Several Canonical Pathways
To understand the molecular changes caused by the ACTN2 mutation, MS was performed in 2D-cultured hiPSC-CMs. Three replicates of each hiPSC-CM line were pooled, and three batches of differentiation were analyzed. Volcano plots depict 481 (250 higher, 231 lower) dysregulated proteins in ACTN2mut vs. ACTN2wt ( Figure 3A). Ingenuity Pathway Analysis (IPA) revealed dysregulation of several canonical pathways, diseases, and biological functions in ACTN2mut hiPSC-CMs ( Figure 3B; Dataset S1). Specifically, mitochondrial function, sirtuin signaling, protein ubiquitination, hereditary myopathy, sliding of myofilaments, and stabilization of mRNA were highly dysregulated in ACTN2mut. A deeper analysis of proteomic data revealed that the protein levels of ACTN2, several other sarcomere-associated proteins and desmosomal proteins were markedly lower in ACTN2mut hiPSC-CMs (Table S1). Some of these proteins (FLNC, MYOZ2, NEBL, SYNPO2, SYNPO2L, TTN) are known to interact directly with ACTN2 [22][23][24]. On the other hand, FHL1 and FHL2, located at the Z-disk of the sarcomere [25,26] were more abundant in ACTN2mut. Several proteins associated with the UPS and/or ALP (e.g., BAG3, CTSC, GBA, HSPA1A, PSMA3, PSMA6, PSMB5, PSME2, TRIM54, UBA1, UBE2O, UBQLN2) were more abundant in ACTN2mut hiPSC-CMs (Table S2). We then performed RNA-seq on 3 pooled replicates of each hiPSC-CM linefrom 3 cardiac differentiation batches. The Volcano plot showed 344 (164 higher, 180 lower) dysregulated mRNAs in ACTN2mut vs. ACTN2wt ( Figure 3C). IPA analysis revealed several different dysregulated canonical pathways, diseases, and biological functions in ACTN2mut when compared to ACTN2wt ( Figure 3D; Dataset S2). Some of the highlighted IPA pathways were metabolism, hypoxia, oxidative and cellular stress, cardiac hypertrophy, and cellular remodeling, whereas signaling of actin cytoskeleton was less pronounced. Specifically, the mRNA levels of sarcomere-associated proteins did not differ between the two groups, except for FHL2 and MYH6, which were lower and higher in ACTN2mut than in ACTN2wt, respectively (Table S1). These data were confirmed by mRNA count analysis using the nanoString nCounter ® Elements technology ( Figure S2E). RNA-seq also revealed dysregulation of several genes, encoding proteins involved in the UPS and ALP in ACTN2mut hiPSC-CMs (Table S2).
Taken together, Omics analysis supported experimental findings for structural sarcomere abnormalities in ACTN2mut hiPSC-CMs and suggested alterations in pathways such as cellular stress response, cell survival/apoptosis or protein homeostasis, which directly point towards proteopathy as an important disease feature.

ACTN2mut hiPSC-CMs Exhibit Higher Activities of the Ubiquitin-Proteasome System and the Autophagy-Lysosomal Pathway
The higher abundance of several UPS-and/or ALP-associated proteins and the presence of ACTN2 aggregates in ACTN2mut hiPSC-CMs suggested an altered proteostasis. Therefore, the activity of both systems was evaluated in 2D-cultured hiPSC-CMs. To evaluate the UPS, cells were treated either with vehicle (0.05% DMSO) or the UPS inhibitor epoxomicin (250 nM; Figure 4A-D). Under basal conditions (DMSO), the levels of (poly)ubiquitinated proteins and of their shuttle protein for autophagy-mediated degradation SQSTM1 did not differ between cell lines, whereas ACTN2 level was lower in ACTN2mut than ACTN2wt, reproducing our findings ( Figure S2F,G). Epoxomicin treatment induced a marked accumulation of (poly)ubiquitinated proteins and SQSTM1 in both hiPSC-CMs ( Figure 4A-C), validating the efficacy of the treatment. In contrast, epoxomicin did not increase the level of ACTN2 in any cell line, indicating that ACTN2 was not degraded by the UPS in this experimental condition ( Figure 4A,D). On the other hand, the chymotrypsin-like activity of the proteasome was markedly higher in ACTN2mut hiPSC-CMs ( Figure 4E), suggesting UPS activation.

ACTN2mut hiPSC-CMs Exhibit Higher Activities of the Ubiquitin-Proteasome System and the Autophagy-Lysosomal Pathway
The higher abundance of several UPS-and/or ALP-associated proteins and the presence of ACTN2 aggregates in ACTN2mut hiPSC-CMs suggested an altered proteostasis. Therefore, the activity of both systems was evaluated in 2D-cultured hiPSC-CMs. To evaluate the UPS, cells were treated either with vehicle (0.05% DMSO) or the UPS inhibitor epoxomicin (250 nM; Figure 4A-D). Under basal conditions (DMSO), the levels of (poly)ubiquitinated proteins and of their shuttle protein for autophagy-mediated degradation SQSTM1 did not differ between cell lines, whereas ACTN2 level was lower in

ACTN2mut hiPSC-CMs Exhibit Force Impairment in Engineered Heart Tissues
The low abundance of several sarcomere-associated proteins in ACTN2mut hiPSC-CMs (Table S1) suggested an impairment of contractile function. Therefore, we assessed force amplitude and kinetics of the ACTN2wt and ACTN2mut in 3D EHTs after 30 days ( Figure 5A,B). Unpaced ACTN2mut EHTs developed significantly lower force starting from day 9 onwards than ACTN2wt EHTs ( Figure 5C). Beating frequency was significantly higher in ACTN2mut than in ACTN2wt EHTs (50 vs. 28 beats per minute from day To evaluate the ALP, the autophagic flux was measured in hiPSC-CMs after treatment with either DMSO (0.05%) or the late-stage ALP inhibitor bafilomycin A1 (50 nM; Figure 4F-H). The level of microtubule-associated protein 1 light chain 3b-II (LC3-II) did not differ between the genotypes in basal conditions. Treatment with bafilomycin A1 markedly increased LC3-II levels in both groups ( Figure 4F,G). The difference in LC3-II level between bafilomycin-treated and DMSO-treated samples, which represents the autophagic flux, was higher in ACTN2mut than in ACTN2wt (difference in log2, ACTN2wt: 1.93, ACTN2mut: 2.09). On the other hand, bafilomycin A1 did not increase ACTN2 levels ( Figure 4H), implying that ACTN2 is not degraded by the ALP. To support the autophagic flux data, we performed a high-content imaging in the hiPSC-CM lines transduced with an AAV6 encoding mTagRFP-mWasabi-hLC3 under the control of the TNNT2 promoter. After 30 days of culture, hiPSC-CMs were fixed and immunostained for TNNT2 and Hoechst to ensure imaging of (solely) cardiomyocytes ( Figure S3A; >80% TNNT2+, data not shown). The number of green and red puncta per well was quantified using an unbiased and statistically powerful method and normalized to the number of hiPSC-CMs per well. The number of green puncta per hiPSC-CM was markedly lower in ACTN2mut ( Figure S3B), whereas the number of red puncta per hiPSC-CM did not differ between the groups ( Figure S3C). The utilization of the LC3-tandem-construct allows to determine autophagosomes (AP) and autolysosomes (AL), since the green fluorescence (mWasabi) is susceptible to low pH and hence quenched within Als. Therefore, green puncta correspond to Aps, red puncta to Aps plus Als, and the difference between red and green puncta (=red minus green puncta) to Als. ACTN2wt hiPSC-CMs exhibited a similar number of Aps and Als per hiPSC-CM (Figure 4I), suggesting a steady-state autophagic flux. In contrast, the AP number per hiPSC-CM was markedly lower and the AL number per hiPSC-CM was higher in ACTN2mut ( Figure 4I). The combination of low AP number and high AL number supports the view of an activation of autophagy, particularly at the step of fusion of Aps with lysosomes to form autolysosomes in ACTN2mut hiPSC-CMs.
Taken together, these data showed higher activities of both protein degradation systems in 2D-cultured ACTN2mut hiPSC-CMs, most likely to eliminate protein aggregates causing proteopathy.

ACTN2mut hiPSC-CMs Exhibit Force Impairment in Engineered Heart Tissues
The low abundance of several sarcomere-associated proteins in ACTN2mut hiPSC-CMs (Table S1) suggested an impairment of contractile function. Therefore, we assessed force amplitude and kinetics of the ACTN2wt and ACTN2mut in 3D EHTs after 30 days ( Figure 5A,B). Unpaced ACTN2mut EHTs developed significantly lower force starting from day 9 onwards than ACTN2wt EHTs ( Figure 5C). Beating frequency was significantly higher in ACTN2mut than in ACTN2wt EHTs (50 vs. 28 beats per minute from day 21 on, respectively ( Figure 5D). To compare functional parameters independent of variable baseline frequencies, EHTs were subjected to electrical pacing at 1 Hz (Videos S3 and S4). Contraction traces of EHTs showed markedly lower force in ACTN2mut than ACTN2wt EHTs ( Figure 5E,F). Normalized averaged force exhibited 19% shorter time to peak (TTP −80% ; Figure 5G,H) and 25% shorter relaxation time (RT 80% ; Figure 5G,I) in ACTN2mut EHTs. Similar results were obtained at 1.5 and 2 Hz (data not shown). Overall, ACTN2mut EHTs exhibited a significant force impairment, which can be explained by the marked deficiency of sarcomere-associated proteins detected in 2D-cultured hiPSC-CMs (Table S1) and EHTs (Table S3). S4). Contraction traces of EHTs showed markedly lower force in ACTN2mut than ACTN2wt EHTs (Figure 5E,F). Normalized averaged force exhibited 19% shorter time to peak (TTP−80%; Figure 5G,H) and 25% shorter relaxation time (RT80%; Figure 5G,I) in ACTN2mut EHTs. Similar results were obtained at 1.5 and 2 Hz (data not shown). Overall, ACTN2mut EHTs exhibited a significant force impairment, which can be explained by the marked deficiency of sarcomere-associated proteins detected in 2D-cultured hiPSC-CMs (Table S1) and EHTs (Table S3).   (Panels F,H,I).

Discussion
This study investigated the cellular and functional impacts of an ACTN2 gene variant (c.740C>T; p.Thr247Met) in hiPSC-CMs. Compared to ACTN2wt, ACTN2mut hiPSC-CMs exhibited (i) cellular hypertrophy, myofibrillar disarray, multinucleation, ACTN2 protein aggregation, and activation of both the UPS and ALP in 2D culture, (ii) a marked reduction in the levels of sarcomere-associated proteins in 2D and EHTs, and (iii) force impairment in EHTs. These findings indicate impaired sarcomerogenesis and proteopathy as typical features in ACTN2mut.
We reproduced previous findings observed in 2D-cultured heterozygous ACTN2 (ACTN2het) hiPSC-CMs [8], such as hypertrophy and myofibrillar disarray in ACTN2mut hiPSC-CMs. Furthermore, diseased cells exhibited dysregulation of pathways involved in sarcomere function and proteostasis, and ACTN2mut EHTs exhibited force impairment, resembling a DCM phenotype [27][28][29][30]. This is in line with the low abundance of several sarcomeric proteins, including ACTN2 in ACTN2mut hiPSC-CMs, leading to a poorly developed sarcomere and possibly to a more immature cardiomyocyte state. In addition, SYNPO2 and SYNPO2L, which contribute to early assembly and stabilization of the Z-disk via interaction with filamin and ACTN2 [22,27], were also less abundant, supporting disruption of the ACTN2 interactome and deficient sarcomere development in ACTN2mut hiPSC-CMs. The reduced Z-disk integration of exogenous MUT-ACTN2 in ACTN2wt by live cell imaging supports the susceptibility of mutant ACTN2 to aggregate. Conversely, exogenous WT-ACTN2 in ACTN2mut ameliorated sarcomere integration and partially replaced endogenous mutant without changing the total level of ACTN2. The inverse correlation between sarcomere incorporation and aggregation suggests that non-incorporated mutant proteins form aggregates and contribute to the low level of ACTN2 protein in ACTN2mut hiPSC-CMs. Previous analysis of the dynamic behavior of two ACTN2 missense variants (p.Ala119Thr and p.Gly111Val), which are also located in the calponin-homology domain, revealed similar phenotypes [31]. Both mutants exhibited reduced binding affinities to F-actin by biochemical assays and alterations of Z-disk localization and dynamic behavior after gene transfer of mEos2-tagged ACTN2 in adult cardiomyocytes.
The higher levels of several proteins involved in proteostasis such as the UPS and ALP in ACTN2mut hiPSC-CMs found in this study are in agreement with previous findings in HCM septal myectomies [32]. This was associated with a higher chymotrypsin-like activity of the proteasome and global activation of the ALP in ACTN2mut hiPSC-CMs. Even though others have shown that WT ACTN2 is degraded by the UPS [33], the low ACTN2 protein level detected by Western blot and proteomic analysis in the ACTN2mut line was unlikely due to degradation by the UPS or the ALP. This suggests that the global activation of both the UPS and ALP are rather compensatory, protective mechanisms against ACTN2 aggregation. The low abundance of ACTN2 and of several other sarcomeric proteins likely reflects a reduced mRNA translation and protein incorporation into myofilament to maintain the overall stoichiometry of the sarcomere [34]. This might explain the poor and reduced formation of sarcomeres in ACTN2mut hiPSC-CMs. Alteration of Z-disk protein turnover combined with subsequent activation of autophagy has been recently reported in hiPSC-CMs carrying the p.Gly1674X or p.Val1668_Gly1674del FLNC variant, resulting in haploinsufficiency and misfolded protein, respectively [35]. Similarly, FLNC protein aggregation and myofibrillar disarray were reported in cardiac muscle specimens of HCM patients carrying the p.Gly2151Ser or p.Ala1539Thr FLNC variant, and was associated with a high risk of sudden cardiac death [36]. These data emphasized the disease-causing role of proteotoxicity in FLNC-related cardiomyopathies and presumed its therapeutic potential.
Functional deficits found in ACTN2mut EHTs are in line with a recent disease modeling study that investigated the p.Arg14del variant in phospholamban (PLN; [29]). The authors showed activation of the unfolded protein response as a compensatory, protective mechanism in the setting of PLN-caused hypocontractility in hiPSC-CMs and EHTs. These findings are further supported by the evidence of PLN protein aggregates in a p.Arg14del mouse model [37]. Interestingly, PLN aggregates and altered protein homeostasis pathways were observed before the onset of functional deficits.
To date, only one study has investigated a homozygous truncating ACTN2 variant (p.Gln860Stop) associated with RCM in mutant carriers [6]. Corresponding hiPSC-CMs displayed hypertrophy, impaired contractility, and myofibrillar disarray. In contrast to our findings, the ACTN2 protein level was not reduced. However, because of the C-terminal truncation, the authors suggested loss of protein-protein interaction as the main cause for disease development. This implies a differing mode of action for truncating and missense ACTN2 variants, further depending on affected functional domains. Nevertheless, these findings are in line with the diminished contractile function in ACTN2mut EHTs, as the patient affected by the homozygous truncating ACTN2 variant (p.Gln860Stop) developed RCM and heart failure (HF) at the age of 23 [6]. Based on the severity of cardiomyopathy phenotypes found in this study combined with the recent evidence that ACTN2 is linked to HF [38], it can be assumed that a patient harboring the homozygous ACTN2 missense variant would develop DCM or RCM leading to HF.
In conclusion, this study revealed an additional cellular pathology for the p.Thr247Met ACTN2 variant, leading to proteopathy. Our data indicate the (compensatory) activation of the proteolytic machinery in ACTN2mut hiPSC-CMs, likely to 'cope' with protein aggregation.

Study Limitations
When working with hiPSCs, it is important to consider possible limitations of the model such as unstable genome integrity, storage of hiPSC lines, immaturity, and reproducibility when using hiPSC-CMs (for reviews, see [39][40][41]). To comply with best cell culture practices, we applied regular karyotyping and genotyping and established master cell banks of each hiPSC line [42]. However, both ACTN2wt and ACTN2mut hiPSC lines were generated from the ACTN2het hiPSC line with CRISPR/Cas9 genetic tools and homology-directed repair (HDR). We found that both hiPSC lines have a correct WT or MUT allele, and that the other recombined allele after HDR contains additional CRISPR/Cas9-mediated on-site defect, i.e., a splicing mutation for ACTN2wt ( [19]) and a large genomic rearrangement for ACTN2mut ( Figure S1A,B), both leading to nonsense mRNAs ( Figure S2D; [19]). Thus, the consequence is the presence of only WT mRNA and protein in the ACTN2wt cardiomyocytes, whereas ACTN2mut cardiomyocytes exhibited only MUT mRNA and protein, making these two cell lines still interesting for comparison. Another limitation is that we cannot decide whether ACTN2 aggregate formation is detrimental, contributing to disease progression, or rather beneficial, to avoid sarcomere incorporation of the mutant protein.

Informed Consent Statement:
The HCM patient carrying the heterozygous ACTN2 (c.740C>T; dbSNP ID: rs755492182) mutation was recruited in the outpatient HCM clinic at the University Heart and Vascular Center Hamburg and provided written informed consent for genetic analysis and the use of skin fibroblasts [8].
Data Availability Statement: Datasets, analysis and study materials will be made available on request to other researchers for purposes of reproducing the results or replicating the procedures. The full description of materials and are provided in the Supplementary Materials. All data of OMICs experiments have been made publicly available. The mass spectrometry data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD034258. The RNA-seq data have been deposited to the European Nucleotide Archive (ENA) at EMBL-EBI under accession number PRJEB52889.