Variants in ACTC1 underlie distal arthrogryposis accompanied by congenital heart defects

Summary Contraction of the human sarcomere is the result of interactions between myosin cross-bridges and actin filaments. Pathogenic variants in genes such as MYH7, TPM1, and TNNI3 that encode parts of the cardiac sarcomere cause muscle diseases that affect the heart, such as dilated cardiomyopathy and hypertrophic cardiomyopathy. In contrast, pathogenic variants in homologous genes such as MYH2, TPM2, and TNNI2 that encode parts of the skeletal muscle sarcomere cause muscle diseases affecting skeletal muscle, such as distal arthrogryposis (DA) syndromes and skeletal myopathies. To date, there have been few reports of genes (e.g., MYH7) encoding sarcomeric proteins in which the same pathogenic variant affects skeletal and cardiac muscle. Moreover, none of the known genes underlying DA have been found to contain pathogenic variants that also cause cardiac abnormalities. We report five families with DA because of heterozygous missense variants in the gene actin, alpha, cardiac muscle 1 (ACTC1). ACTC1 encodes a highly conserved actin that binds to myosin in cardiac and skeletal muscle. Pathogenic variants in ACTC1 have been found previously to underlie atrial septal defect, dilated cardiomyopathy, hypertrophic cardiomyopathy, and left ventricular noncompaction. Our discovery delineates a new DA condition because of variants in ACTC1 and suggests that some functions of ACTC1 are shared in cardiac and skeletal muscle.


Introduction
Sarcomeres are the repeating functional units of muscle cells that are joined end to end to form skeletal and cardiac muscle fibers. 1 Sarcomeres consist of thick myosin filaments and thin actin filaments along with proteins such as troponin and tropomyosin that facilitate and regulate the interactions between the filaments. 2 Contractile force is generated when the myosin and actin filaments bind to form cross-bridges, which is thought to cause the filaments to slide against each other. The key sarcomeric proteins are encoded by highly conserved and homologous genes that typically express an isoform that is predominant in either cardiac or skeletal muscle. For example, TNNI2, MYH3, and ACTA1 encode isoforms of troponin, myosin heavy chain, and a-actin, respectively, that are primarily expressed in skeletal muscle, while TNNI3, MYH6, and ACTC1 encode isoforms that are primarily expressed in cardiac muscle. 3 Pathogenic variants in genes that encode the skeletal sarcomeric proteins tropomyosin (TPM2; MIM: 190990); 4 troponin I2, fast skeletal type (TNNI2; MIM: 191043); 4 troponin T3, fast skeletal type (TNNT3; MIM: 600692); 5 myosin heavy chain 3 (MYH3; MIM:160720); 6 and myosin heavy chain 8 (MYH8; MIM: 160741) 7 account for most cases of distal arthrogryposis (DA), a group of Mendelian conditions characterized by non-progressive congenital contractures of the limbs and, less frequently, contractures of the face, strabismus, neck webbing, pterygia, short stature, and scoliosis. The precise pathogenesis of the contractures is unknown, although it has been proposed that pathogenic variants lead to perturbation of muscle contraction or relaxation, resulting in reduced limb movement in utero. 8,9 Several additional genes, specifically PIEZO2 [MIM: 613629], 10 14 that underlie other forms of DA encode proteins that are less directly involved in sarcomere contraction.
To date, the vast majority of variants in genes encoding homologous components of the cardiac sarcomere have been found to result in conditions in which only cardiac muscle is affected, including cardiomyopathy and structural heart defects. Here, we report five families with DA and congenital heart defects because of heterozygous missense variants in the gene actin, alpha, cardiac muscle 1 (ACTC1). ACTC1 encodes a highly conserved actin that binds to myosin in cardiac and skeletal muscle. We employ molecular dynamics (MD) simulations of wild-type (WT) and mutant cardiac actin to predict the structural and functional consequences of these variants.

Methods
Exome sequencing, annotation, and filtering From a cohort of 463 families (1,582 individuals) with multiple congenital contractures, we selected 172 families in which pathogenic or likely pathogenic variants had not been identified, for exome sequencing (ES). All studies were approved by the institutional review boards of the University of Washington and Seattle Children's Hospital, and informed consent was obtained from each participant or their parents. ES, annotation, and analysis were performed by the University of Washington Center for Mendelian Genomics (now the University of Washington Center for Rare Disease Research) as described previously. 15 Briefly, variants were called by GATK v.3.7 HaplotypeCaller and annotated with Variant Effect Predictor v.95.3. 16 Variants were filtered using GEMINI v.0.30.1 17 for genotype call quality (GQ R 20), read depth (R6), allele frequency in population controls (i.e., maximum frequency in any continental superpopulation in gnomAD 18 v.2.1 and v.3.0 exomes and genomes < 0.005), consistency with the mode of inheritance in each family, and predicted impact on protein-coding sequence (e.g., annotated as missense, nonsense, canonical splice, or coding insertion or deletion [indel]).

MD: Model preparation and simulation
Initial coordinates for cardiac globular (g-actin) structures were generated via homology to an X-ray crystal structure of rabbit skeletal actin (UniProt: P68135) downloaded from the Protein Data Bank 19 (www.rcsb.org; PDB: 3HBT). 20 PDB: 3HBT is a model of g-actin complexed with ATP, Ca 2þ , and SO 4 . The human ACTC1 sequence was downloaded from UniProt (P68032). The human and rabbit sequences were 98.9% identical, as assessed using Clustal Omega, 21 and there were four conservative amino acid substitutions (human amino acid, human residue number, and rabbit amino acid as follows: Asp2Glu, Glu3Asp, Leu301Met, and Ser360Thr). Homology models of the human WT and four mutant (p.Thr68Asn, p.Arg185Trp, p.Gly199Ser, and p.Arg374Ser) structures were generated using Modeller, 22 which introduced the amino acid substitutions and built coordinates for atoms not present in the PDB file (no coordinates were present for D-loop residues 40-50 in PDB: 3HBT). The p.Arg374His variant was not simulated because of the expected overlap with p.Arg374Ser and because the change to His is more conservative than the change to Ser. For p.Gly199Ser, the initial backbone dihedral (4, c) angles for Gly were (152 , À16 ), which are unfavorable for Ser. Consequently, this loop was further refined, and the initial S199 dihedrals were (À177 , À14 ). During modeling, crystallographic waters and the SO 4 were removed, ATP was retained, and Ca 2þ was replaced by Mg 2þ . Initial coordinates for cardiac filamentous (F-actin) pentamer structures were generated using an electron microscopy structure of mouse tropomyosin and rabbit skeletal actin (PDB: 3J8A). 23 The tropomyosin chains were removed, and the F-actin pentamers complexed with ADP and Mg 2þ were used to construct homology models of WT and p.Thr68Asn human cardiac F-actin with Modeller. After homology models were built, hydrogen atoms were modeled onto the initial structure using the tleap module of AMBER, and each protein was solvated with explicit water molecules in a periodic, truncated octahedral box that extended 10 Å beyond any protein atom. Na þ counterions were added to neutralize the systems.
All simulations were performed with the AMBER20 package 24,25 and the ff14SB force field 26 using standard procedures. Water molecules were treated with the TIP3P force field. 27 Metal ions were modeled using the Li and Merz parameter set. [28][29][30] ATP and ADP molecules were treated with parameters from Meagher et al. 31 The SHAKE algorithm was used to constrain the motion of hydrogen-containing bonds. Long-range electrostatic interactions were calculated using the particle mesh Ewald (PME) method. Each system was minimized in 10,000 steps divided across 3 stages in which restraints were placed either on hydrogen atoms, solvent atoms, or all backbone heavy atoms (C a , C, N, O atoms). After minimization, systems were heated to 310 K over 300 ps using the canonical NVT (constant number of particles, volume, and temperature) ensemble. During all heating stages, 25 kcal mol À1 restraints were present on the backbone heavy atoms (C a , C, N, O atoms). After the system temperature reached 310 K, the systems were equilibrated for 5.4 ns over 5 successive stages using the NPT (constant number of particles, pressure, and temperature) ensemble. During equilibration, restraints on backbone atoms were decreased from 25 kcal mol À1 during the first stage to 1 kcal mol À1 during the fourth stage. During the final equilibration stage, the systems were equilibrated in the absence of restraints. Production dynamics for conventional MD (cMD) simulations were then performed using the NVT ensemble using an 8-Å nonbonded cutoff and a 2-fs time step, and coordinates were saved every picosecond. G-actin cMD simulations were run in triplicate; each replicate simulation was 500 ns long. We used an enhanced sampling scheme called Gaussian accelerated MD (GaMD) 32 to explore conformational sampling in the WT and p.Thr68Asn F-actin pentamer models. GaMD production runs were preceded by a 52-ns-long GaMD equilibration period in which boost potentials were added. The upper limits of the SD of the boost potentials were set to 6 kcal mol À1 . Neither the standard 5.2-ns equilibration nor the 52-ns GaMD equilibration contributed to the length of the production dynamics for any simulation. Production dynamics for GaMD simulations were using the NVT ensemble using an 8-Å nonbonded cutoff and a 2-fs time step, and coordinates were saved every picosecond. Single replicates of the WT and p.Thr68Asn were performed, and each simulation was 300 ns long. Unless specified otherwise, simulations were analyzed separately, and the results of replicate simulations were averaged together.

MD: Analysis
The C a root-mean-square deviation (RMSD), C a root-meansquared fluctuation (RMSF), solvent-accessible surface area (SASA), secondary structure content, inter-atomic distances, and inter-residue contacts were calculated with cpptraj. 33 The C a RMSD was calculated after alignment of all C a atoms to the Webbed elbow HP:0009760 - Contractures of wrists HP:0001239 -

Clinical features: Face and neck
Downslanting palpebral fissures Downturned corners of the mouth Cardiomyopathy HP:0001638 - Ventricular septal defect HP:0001629 - (Continued on next page) minimized structure. The C a RMSF was calculated about average MD structures for each simulation. For each time point in the simulation, two residues were considered to be in contact with one another when at least one pair of heavy atoms was within 5 Å of another. Then we recorded the average percent simulation time each residue pair was in contact for each simulation. A Student's t test was used to identify statistically significant (p < 0.05) differences in inter-residue contact times between the WT and mutant simulations. All protein images were prepared using UCSF Chimera. 34,35 Results After variant filtration of the exome data, three families had compelling candidate variants in the same candidate gene, ACTC1 (MIM 102540; GenBank: NM_005159.4) ( Table 1; Figures 1 and S1). Specifically, each family had a heterozygous candidate missense variant that was either de novo or segregated in an autosomal dominant pattern.
In family A, comprised of an affected father and affected daughter with camptodactyly of the fingers, hypoplastic flexion creases, clubfoot, webbed neck, scoliosis, hip contractures, and ventriculoseptal defect, a heterozygous variant in ACTC1 (c.595G>A, p.Gly199Ser) was identified. This family has been described previously (family D in Chong et al. 36 ) as possibly having autosomal dominant multiple pterygium syndrome (MIM: 178110), but no likely pathogenic or pathogenic variants in MYH3 were identified. In family B, an affected mother and affected daughter with knee contractures, clubfoot, limited neck rotation, scoliosis, and hip contractures were heterozygous for c.1120C>A, p.Arg374Ser. The grandmother in family B was described as having similar clinical findings, but no medical records or photographs were available. Family C, the third family, was a simplex family in which a de novo variant, c.203C>A, p.Thr68Asn, was identified in the proband with clubfoot, camptodactyly of the toes and fingers, webbed neck, and an atrial septal defect. Upon follow-up with family C approximately 20 years after they were originally enrolled, the proband was found to have a son who had camptodactyly of the fingers, overlapping toes, elbow contractures, elbow webbing, and an atrial septal defect. Her son was heterozygous for the c.203C>A, p.Thr68Asn variant. No other variants of interest were identified in any of these families in genes known to underlie DA or congenital heart defects. Two additional simplex families (families D and E) subsequently came to our attention via direct referral after clinical testing identified ACTC1 as a candidate gene. Each had a de novo variant: c.553C>T, p.Arg185Trp in family D and c.1121G>A, p.Arg374His in family E. Notably, the variants in families B and E perturbed the same residue. The proband of family D had short stature, clubfoot, knee and elbow contractures, hypoplastic flexion creases, atrial septal defect, and ventricular septal defect. No muscle weakness was noted. An echocardiogram found borderline left ventricular systolic function with an ejection fraction Plus (þ) indicates presence of a finding, and minus (À) indicates absence of a finding. *, described per report; N/D, no data were available; N/A, not applicable; CADD, Combined Annotation Dependent Depletion v.1.6. cDNA positions are named using HGVS notation and GenBank: NM_005159.4. Predicted amino acid changes are shown. US, ultrasound.
of 52% and reduced longitudinal strain in the basal wall segments. Electron microscopy and immunohistochemistry of a skeletal muscle biopsy revealed possible nemaline bodies ( Figure S2) and the presence of rare myofibers with a central core but no ragged red fibers or rod-like inclusions. The proband in family E was stillborn at 38þ3 weeks. At 31 weeks, ultrasound findings included hydrops fetalis, hydrothorax, a small lower jaw, ductus venosus agenesis, and positioning of the extremities consistent with fetal akinesia. No additional clinical information was available.
In summary, at least one affected individual in each family was reported to have a combination of camptodactyly of the fingers or toes, hypoplastic flexion creases, clubfoot, limited neck rotation, scoliosis (excluding family E, for which only fetal ultrasound was available), and hip contractures. Common facial features included microretrognathia, ptosis, downslanting palpebral fissures, lowset ears, and a long nasal bridge ( Figure 1). Ventricular or atrial septal defects were reported in families A, C, and D, while family E had ductus venosus agenesis in utero, but only one affected individual, the proband in D, had cardiomyopathy. The co-occurrence of these congenital heart defects is notable because ACTC1 is well established to underlie isolated cardiac abnormalities, including dilated and hypertrophic cardiomyopathy (MIM: 613424, 612098), atrial septal defects (MIM: 612794), and left ventricular The characteristics shown include webbed neck, bilateral clubfoot, camptodactyly of the fingers, and hypoplastic flexion creases in family A (A; II-1 and III-1); camptodactyly, webbed neck, bilateral clubfoot, camptodactyly of the fingers and toes, and hypoplastic flexion creases in family B (B; II-2 and III-1); webbed neck, bilateral clubfoot, webbed neck, bilateral clubfoot, and camptodactyly of the fingers and toes in family C (C; II-2); and ptosis, webbed neck, camptodactyly of the fingers, and scoliosis in family D (D; II-1). Table 1 contains a detailed description of the clinical findings of each affected individual, and Figure S1 provides a pedigree for each family. noncompaction (MIM: 613424). However, ACTC1 has not been reported to underlie a multiple malformation syndrome that affects multiple organs.
Combined Annotation Dependent Depletion (CADD; v.1.6) 37 scores greater than 20.0 indicate that all five variants are predicted to be pathogenic (Table 1). For all five of these variants, the homologous residues in ACTA1 have been reported [38][39][40][41][42][43] to be perturbed in infants with autosomal dominant severe congenital nemaline myopathy ( Figure 2), leading to death before 1 year of age, an observation that further suggests that these residues play a critical role in sarcomere function. In addition, these variants were either absent or exceedingly rare in gnomAD v.2.1.1 or v.3.1.2. p.Arg185Trp was heterozygous in a single individual in gnomAD and was the only variant that has been reported previously in ClinVar (twice classified as ''likely pathogenic''). One of these ClinVar entries (SCV000742090.2) reports that p.Arg185Trp was found in an individual with ''arthrogryposis multiplex congenita, multiple suture craniosynostosis, high palate, cleft uvula, pulmonary hypoplasia, bronchomalacia, pulmonary arterial hypertension, hydrocephalus, cryptorchidism, penile hypospadias, dysphagia, secundum atrial septal defect, patent foramen ovale, shallow orbits, infra-orbital crease, microretrognathia, webbed neck, short neck.'' These clinical findings suggest that this individual likely has the same condition we describe here.
Pathogenic variant-associated changes in the overall conformation and dynamics of g-actin We first analyzed the conformations sampled by WT and mutant g-actin to determine whether four of the variants we identified ( Figure 3A) led to large-scale conformational changes within actin monomers. We calculated the C a RMSD (a measure of structural similarity) of each frame in the simulation to the minimized structure. The overall conformation of the actin monomers was preserved despite introduction of each variant. In fact, the mutant simulations all had smaller C a RMSD values than the WT simulations, indicating that the variants dampened structural fluctuations in g-actin. All simulated systems had an average C a RMSD (a measure of structural similarity) of less than 2.6 Å to the crystallographic conformation ( . The largestamplitude structural change was a breathing motion in which relative scissoring of SD2 (subdomain 2) and SD4 opened and closed the nucleotide binding pocket, which occurred in all simulations. We next measured the C a RMSFs and compared them with the WT simulations ( Figure S3). The majority of residues in g-actin had small (<1 Å ) C a fluctuations about their average positions. The regions with the greatest fluctuations were the DNase1 binding loop (also known as the D-loop, residues 41-56) and two loops in SD4 (residues 199-204 and 219-224). The flexibilities of most residues were not affected by the variants. However, all four variants led to a decrease in the C a RMSF of residues in the D-loop ( Figure S3), and statistically significant (residues with significant differences denoted in Figure S3) decreases in the C a RMSF of D-loop residues were observed for p.Thr68Asn, p.Arg185Trp, and p.Gly199Ser. Each of the variants also caused low-magnitude (<0.5 Å ) but statistically significant (residues with significant differences denoted in Figure S3) (p value and test) changes in C a RMSF among residues near the variant sites.
p.Thr68Asn modified the structural organization of SD2 and the D-loop The RMSD and RMSF data indicated substantial changes in the dynamics of the D-loop in the presence of all four simulated variants. Therefore, we examined the dynamics in this region in greater detail with an emphasis on the p.Thr68Asn variant because T68 is located within SD2 and closest structurally to the D-loop. Changing Thr to Asn (p.Thr68Asn) is somewhat conservative; both residues have polar side chains, but the Asn side chain is long, whereas Thr branches at C b . The alternate conformations accessible to Asn led to a cascade of changes in amino acid interactions among neighboring residues in SD2 ( Figures 4A-4C). The changes in contacts affected interactions made by D-loop residues as well as a complex salt bridge formed between residues 39, 70, and 83. p.Thr68Asn increased the extent to which residues in the D-loop formed an a helix secondary structure ( Figures 4B  and 4D). p.Thr68Asn decreased the overall SASA of residues in the D-loop relative to the WT simulations ( Figure 4E). The net effect of the structural changes induced by p.Thr68Asn shifted the structure and dynamic behavior of SD2 so that the mutant SD2 adopted a more compact and less flexible conformation relative to the WT.
These effects were most pronounced for p.Thr68Asn but were also observed for the other variants. The greater effect of p.Thr68Asn was likely due to its central position in SD2. Altered structure and dynamics among D loop residues were also observed for the p.Gly199Ser, p.Arg185Trp, and p.Arg374Ser simulations (Figures S4-S6). All variants altered inter-residue interactions formed by D loop residues and other SD2 residues ( Figure S4). All variants increased the extent to which residues in the D-loop formed an a helix secondary structure in the ensemble average (p.Arg374Ser>p.Arg185Trp>p.Gly199Ser>p.Thr68Asn>-WT; Figure S5). However, there was not a statistically meaningful change in the net amount of a helix formed, and this region did form an enduring a helix in one of the WT simulations.  Figure S6). We analyzed statistically significant (Figures 4 and S4; Table S2) changes in residue-residue contact networks to identify structural pathways by which the variants altered SD2 dynamics ( Figure 5; Table S2). Altered residue-residue interactions were only considered in this analysis when there was at least a 10% difference in the average contact time frequency between the WT and mutant simulations. The extent to which the variants altered residue-residue interaction networks was variable. Disruption was greatest for the p.Arg185Trp variant, and the p.Gly199Ser variant was the least impactful. P.Thr68Asn modified local residue-residue interactions to affect changes in SD2 and the D-loop. p.Arg185Trp, p.Gly199Ser, and p.Arg374Ser instead introduced structural changes that propagated  through SD4 and/or SD2 before ultimately altering SD2 structure ( Figure 5; supplemental information; Table S2). Although operating through distinct mechanisms, all ACTC1 variants simulated in this study led to a common change in the structure and dynamics of SD2 and the D-loop.

p.Thr68Asn modified interactions between actin subunits in F-actin
In our g-actin simulations, all variants led to similar structural changes in SD2 and the D-loop. To make predictions about the effects of altered SD2 structure/dynamics in actin filaments, we performed MD simulations of WT and p.Thr68Asn cardiac F-actin pentamers ( Figure 3B). Simulation performance rapidly decreases with the number of atoms; therefore, we simulated pentamers as a simplified proxy for actin filaments and only analyzed the dynamics of the central chain (chain C) to avoid end effects. F-actin filaments are composed of two protofibrils, each of which contains monomers arranged so that SD3 of one monomer is inserted into the cleft between SD2 and SD4 of the succeeding monomer (moving from theto þ end). Two protofilaments twist around one another, and the face of the monomer containing the variant (the front-facing side in Figure 3B) is buried. SD2 is a critical structural component of actin filaments; it forms interactions between actins within a single protofilament (intra-filament) and between monomers of different protofilaments (inter-filament). For example, in the cryoelectron microscopy (cryo-EM) structure, the D-loop of one monomer encircles Y171 of the succeeding monomer in the same protofilament (in our model, the D-loop of chain C loops around Y171 of chain A). Additionally, R41 forms a salt bridge with E272 of a monomer in the opposite protofilament. In GaMD simulations of F-actin, p.Thr68Asn resulted in a shift in the structure and dynamics of SD2 and the D-loop ( Figure 6). The variant resulted in a change in residue-residue interactions made by the D-loop ( Figure 6A) and reduced the number of contacts made between SD2 of chain C and atoms in other actin subunits ( Figure 6B). As observed in the g-actin simulations, p.Thr68Asn promoted a more compact conformation of the D-loop in F-actin ( Figure 6C). The more compact loop conformation reduced interactions between the D-loop of chain C and Y171 of chain A and also eliminated the salt bridge formed between R41 of chain C and E272 of chain B (Figures 6D and 6E).

Discussion
We identified five unrelated families in which a total of eight individuals have heterozygous, rare, pathogenic variants in ACTC1 and share similar phenotypic effects, including multiple congenital contractures, neck pterygia, scoliosis, and congenital heart defects/cardiomyopathy. This pattern of clinical findings appears to represent an autosomal dominant disorder, distinct from previously reported Mendelian conditions due to ACTC1 variants which are characterized   47 ). MD simulations demonstrate that all four variants (p.Thr68Asn, p.Arg185Trp, G19S, and p.Arg374Ser) disrupt the native structure of the regions of actin most associated with protein-protein interactions (SD2 and the D-loop), impeding interactions between actin and its binding partners, including other actins, during thin filament assembly. Additionally, the altered D-loop structure is predicted to increase structural disorder within thin filaments, resulting in ''stretchier'' thin filaments that may contract more slowly, require greater loads to extend, have weakened force production, and/or have slower rates of force production. Structural perturbations to the D-loop are known to affect thin filament stiffness. 48,49 Thus, while the genetic basis of this DA condition is unique compared with other DAs, the underlying mechanisms may be similar if not identical. 9,15,[50][51][52] All of the pathogenic ACTC1 variants (n ¼ 87) reported to date (Table S1), with the exception of p.Arg185Trp, which we also identified in family D, were found in persons noted only to have abnormalities of the heart. While it is possible that congenital contractures have been overlooked in thoe families, this seems like an unlikely explanation for all or even most families. Alternatively, there may be biological explanations for this observation, none of which are mutually exclusive. First, none of the residues perturbed in the families we identified, except for Arg185, which has been found previously in a person with congenital contractures, have been reported previously. So, the distribution of phenotypic effects associated with these genotypes has been, to date, unknown. Second, substitutions of each of these residues in ACTC1 increases disorder of actin SD2 and D-loop interactions, and these perturbations could be a unique consequence of contracture-associated variants. Third, the presence of a pathogenic ACTC1 variant may be necessary but not sufficient for development of congenital contractures. In other words, skeletal muscle might be affected only in the presence of (a) genetic modifier(s). We verified the absence of additional rare coding ACTC1 or ACTA1 variants but could not exclude the presence of structural variants and/or variants in non-coding regulatory elements that might alter expression of ACTC1 or ACTA1.
The observation that rare genotypes in ACTC1 underlie cardiac and skeletal abnormalities is not without precedent. ACTC1 and ACTA1 are highly homologous, differing by only four amino acids (Figure 2), and both actins are expressed in skeletal and cardiac muscle. [53][54][55][56] During fetal development, ACTC1 is the predominant actin, as measured by protein expression, in skeletal and cardiac muscle. 55 It is downregulated starting around 27-28 weeks of fetal development and continues to decline until $6 months of age, when it accounts for $5% of total actin. 55 In adult skeletal muscle, ACTC1 and ACTA1 account for $5% and 95% of actin, respectively, 55 and ACTC1 accounts for $80% of actin in adult cardiac muscle. 53 These differences in spatial and temporal expression are considered explanations for the exclusive association of skeletal muscle abnormalities (i.e., nemaline myopathy  47 ) with pathogenic variants in ACTC1. Of the hundreds of individuals described with ACTA1-associated myopathy, only 12 (9 unique variants) [57][58][59][60][61][62][63][64][65] have been reported to also have a cardiac abnormality (Table S1), either in conjunction with a skeletal myopathy (n ¼ 10) or alone (n ¼ 2). Rare variants in ACTC1 resulting in congenital contractures in a small fraction of persons with ACTC1 variants appears to be the corollary. For each mutant-WT comparison, residue-residue contacts that were present for statistically different percentages of the simulations were mapped onto the reference crystal structure of g-actin. Contacts that were present more frequently in the WT simulations are denoted by black pipes, and contacts present more frequently in the mutant simulations are colored orange, magenta, purple, or blue. The thickness of the pipes corresponds to the difference in percent simulation time that the contact was present between the WT and mutant simulations (larger pipes indicate that a contact was observed more frequently). Although the variants were distributed throughout the structure, they all led to statistically significant (see Table S2 for test statistics) changes in the structure of SD2 (orange ribbons).
Pathogenic variants in ACTC1 result in skeletal muscle contractures even though ACTC1 accounts for only $5% of total actin in adult skeletal muscle. 53 The most likely explanation is that the skeletal muscle contractures originate during fetal development, when ACTC1 is the predominant source of sarcomeric actin, and replacement of most skeletal muscle actin with wildtype ACTA1 during infancy is insufficient to correct the abnormality. However, it is possible that ACTC1 has a previously unknown function in skeletal muscle biology or that mutant ACTC1 protein interferes with the function of actin encoded by ACTA1. Testing this hypothesis will require further functional characterization of these variants.
The MD simulations are limited by several factors. First and foremost, the method used to introduce the pathogenic variants assumes that the mutant constructs can access WTlike conformations, and the timescale along which they transition from a WT-like ensemble to a mutant ensemble is not known. Second, our simulations have probed isolated states of g-actin and F-actin and cannot directly describe effects the variants have on interactions between g-actin and its binding partners nor between F-actin and the rest of the contractile machinery present in sarcomeres. Nevertheless, these simulations provide predictions about the functional consequences of DA-associated variants in ACTC1, generate hypotheses on disease mechanisms, and provide guidance for future studies, such as investigation of whether the variants result in impaired filament assembly or impaired filament mechanics.
In summary, we identified five unrelated families with heterozygous pathogenic variants in ACTC1 resulting in multiple congenital contractures, webbed neck, scoliosis, short stature, and distinctive facial features as well as cardiac abnormalities, including atrial and ventricular septal defects, left ventricular noncompaction, and cardiomyopathy. This appears to be a novel Mendelian condition because of pathogenic variants in a gene known to underlie conditions characterized only by cardiac defects. Our findings suggest that persons with multiple congenital contractures should be tested for pathogenic variants in ACTC1 and that persons with contractures and pathogenic variants in ACTC1 should undergo cardiac evaluation for structural and functional abnormalities.

Data and code availability
Sequence data for family A is in dbGaP under accession number phs000693, and those for families B and C will be available in Figure 6. p.Thr68Asn alters inter-chain interactions made by SD2 in F-actin Residue-residue interactions formed between SD2 of chain C and chains A and B were analyzed in the GaMD simulations of the WT and p.Thr68Asn F-actin. (A) p.Thr68Asn led to statistically significant differences in residue-residue contacts formed by SD2 of chain C (denoted by pipes as in Figure 5). Differences were found in contacts formed between SD2 of chain C and SD1 of chain A as well as in contacts formed between SD2 of chain C and the SD3-SD4 linker of chain B. (B) The total number of atom-atom interactions formed between SD2 of chain C and all atoms in chains A and B were monitored in the WT and p.Thr68Asn GaMD simulations. Relative to the WT simulation (black), the p.Thr68Asn simulation (orange) had fewer interchain contacts involving chain CSD2. (C-E) In the reference cryo-EM structure and WT simulation, the D-loop of SD2 in chain C fits into a pocket formed by SD1 and SD3 of chain A. The p.Thr68Asn simulations instead sampled non-native conformations in which the D-loop exited this binding pocket. In the reference cryo-EM structure and the WT simulation (D), the D-loop of chain C is stabilized via a network of hydrophobic interactions formed with Y171 of chain A as well as a hydrogen bond network involving Arg 41 (chain C), Thr 68 (chain C), and Glu 272 (chain B). These interactions were disrupted in the p.Thr68Asn simulation (E).
the AnVIL under accession number phs003047 pending the first public release of the GREGoR dataset. Please contact the corresponding author, M.J.B., for further information.

Acknowledgments
We thank the families for their participation and support. Sequencing and data analysis were provided by the University of Washington Center for Rare Disease Research (UW-CRDR) with support from NHGRI grants U01 HG011744, UM1 HG006493, and U24 HG011746.    Figure S1. Pedigrees of families with pathogenic variants in ACTC1 resulting in distal arthrogryposis 3. Figure S2. Histochemistry and immunohistochemistry from muscle biopsy of paravertebral muscles in proband of Family D.
7. Figure S6. Pathogenic variant-associated changes in D-loop solvent accessible surface area.
8. Table S1. ACTC1 and ACTA1 missense variants reported as pathogenic in individuals with phenotypes involving cardiac muscle, skeletal muscle, or both. (provided as a separate Excel spreadsheet) 9. Table S2. Reported residue-residue pair interaction frequencies that differed among the WT and mutant simulations. (provided as a separate Excel spreadsheet and in this document)

Summary of structural effects of R185W, G199S, and R374S
R185Y: In the WT simulations, R185 formed transient electrostatic interactions with S16, G17, E74, and D159. Additionally the Y71 side chain formed a large number of atom-atom contacts with R185 by stacking against its guanidino group. Due to these interactions, R185 stabilizes g-actin structure in the vicinity of the ATP pocket and forms a structural linker between all four subdomains. Mutation of R185 to Trp eliminated these stabilizing interactions; instead, W185 was primarily positioned in the cleft between SD2 and SD4 and interacted primarily with other SD4 residues. This resulted in a large-scale reorganization of inter-residue contacts spanning all 4 subdomains and an increase in the SD2-SD4 cleft distance.
G199S: In the WT simulations, G199 adopts backbone dihedral angles that are unfavorable for most amino acids and accommodates a transition from -helix structure in residues 184-198 to coil structure in residues 199 -204. Mutation to S199 resulted in a shift in the backbone conformation of residue 199 to angles within the lower left quadrant of Ramachandran space, a decrease in a-helix structure in residues 197-198, and a shift in the structure of the loop spanning residues 199 to 204. This change resulted in few statistically significant changes in inter-residue interactions but did modify the breathing motion between subdomains 2 and 4.
R374S: In the WT simulations, R374 formed transient salt bridges with E363 and E366. Each of these residues are located on the surface of SD1. These salt bridges were lost in the R374S simulations.
Simultaneously there was a reorganization of inter-residue interactions in SD1 as well as an increase in residueresidue interactions between an -helix of SD1 (residues 81-96) and residues in SD2.      S1. ACTC1 and ACTA1 missense variants reported as pathogenic in individuals with phenotypes involving cardiac muscle, skeletal muscle, or both. This table contains all missense variants in ACTA1 or ACTC1 that were reported as Likely Pathogenic or Pathogenic in ClinVar; Damaging in HGMD, and/or nonsynonymous in the ACTC1 or ACTA1 LOVD Locus Specific Databases. ACTA1 and ACTC1 are both 377 residues long and differ at only 4 sites (4, 5, 301, 360). Each variant is listed alongside the reported phenotype (one row per affected individual/database entry), whether the affected individual had skeletal muscle or cardiac muscle findings, the Pubmed PMID of any associated publication and/or submitter. The residue numbering listed for ACTA1 variants was updated to current (2022) numbering if necessary. Table S1 is provided as an Excel spreadsheet.  This table enumerates the average percent simulation time for which select residue-residue pairs formed inter-residue interactions in the WT and mutant MD simulations. Altered residue-residue interactions are only reported if there was at least a 10% difference in the average contact time frequency between the WT and mutant simulations and a statistically significant difference in the average WT and mutant interaction frequencies.