Deficiencies in tRNA synthetase editing activity cause cardioproteinopathy

Significance

Misfolded proteins are a hallmark of diverse human cardiac disorders including desmin-related cardiomyopathy and systemic amyloidosis. Defects in translational fidelity can cause neurodegeneration, however the consequences of mistranslation in other tissues, including the heart, are unknown. The fidelity of protein synthesis is largely ensured by aminoacyl-tRNA synthetases, and many tRNA synthetases contain editing domains that hydrolyze mischarged tRNAs, preventing incorporation of incorrect amino acids into proteins. Here, we show that diminished editing efficacy of the alanyl-tRNA synthetase causes misfolded protein aggregation and cell death in the mammalian heart. These results illuminate the importance of translational fidelity in cardiac homeostasis and suggest that genetic factors that disrupt the accuracy of translation may contribute to proteinopathies of the heart and other tissues.

Abstract

Misfolded proteins are an emerging hallmark of cardiac diseases. Although some misfolded proteins, such as desmin, are associated with mutations in the genes encoding these disease-associated proteins, little is known regarding more general mechanisms that contribute to the generation of misfolded proteins in the heart. Reduced translational fidelity, caused by a hypomorphic mutation in the editing domain of alanyl-tRNA synthetase (AlaRS), resulted in accumulation of misfolded proteins in specific mouse neurons. By further genetic modulation of the editing activity of AlaRS, we generated mouse models with broader phenotypes, the severity of which was directly related to the degree of compromised editing. Severe disruption of the editing activity of AlaRS caused embryonic lethality, whereas an intermediate reduction in AlaRS editing efficacy resulted in ubiquitinated protein aggregates and mitochondrial defects in cardiomyocytes that were accompanied by progressive cardiac fibrosis and dysfunction. In addition, autophagic vacuoles accumulated in mutant cardiomyocytes, suggesting that autophagy is insufficient to eliminate misfolded proteins. These findings demonstrate that the pathological consequences of diminished tRNA synthetase editing activity, and thus translational infidelity, are dependent on the cell type and the extent of editing disruption, and provide a previously unidentified mechanism underlying cardiac proteinopathy.

Proteins are the building blocks and major signaling molecules of cells, and as such, their synthesis and degradation are tightly regulated. Disruption of protein homeostasis (proteostasis) can result in the accumulation of abnormal proteins that lead to cellular pathogenesis. Like the neuroproteopathies, misfolding of specific proteins in cardiomyocytes can be caused by genetic mutations that alter the primary structure of aggregated proteins. For example, mutations in the desmin gene, which encodes a muscle-specific intermediate filament protein, or in the gene encoding αB-crystallin, a molecular chaperone for desmin, result in aggregation of these proteins and are a primary cause of hereditary cardiomyopathy (1, 2). Systemic amyloidosis, the extracellular accumulation of abnormal proteinaceous fibrils, can also occur in the heart with detrimental effects on cardiac function (3, 4). In addition, misfolded proteins and changes in the ubiquitin proteasome system or autophagy have been widely reported in failing hearts (4, 5), demonstrating the importance of protein quality control systems for cardiomyocyte homeostasis. However, whether such proteostatic changes induce pathological changes, or are consequences of the diseased state, is not clear.

In addition to genetic mutations, defective proteins can be generated by inaccurate translation. Translational fidelity is largely controlled by the precise aminoacylation of tRNAs with their cognate amino acids, a function carried out by the aminoacyl-tRNA synthetases (aaRSs). To improve substrate specificity, an editing site, which hydrolyses misactivated noncognate amino acids or mischarged tRNAs and is distinct from the aminoacylation domain, is found in approximately half of the aaRSs (6, 7). Failure of proofreading by these aaRSs may result in incorporation of the wrong amino acid into the nascent peptide, which could be detrimental to protein folding and function.

Although errors in aminoacylation are relatively low (8, 9), even small decreases in aaRS proofreading have dramatic effects on cell survival (10–12). Bacterial alanyl-tRNA synthetase (AlaRS) can misactivate glycine or serine, and Escherichia coli expressing a severe editing-deficient form (C666A) of AlaRS had increased death when grown in elevated concentrations of these amino acids (11). In mice, a point mutation (A734E) in the AlaRS editing domain in the mutant strain “sticky” caused a twofold increase in Ser-mischarged tRNA^Ala that resulted in formation of ubiquitinated protein aggregates in cerebellar Purkinje cells and degeneration of these neurons (10). However, because cell loss in other regions of the brain or in other tissues was not observed in the sti mutant mouse, the importance of AlaRS editing activity in other mammalian cell types remains unknown.

Here we hypothesized that further decreases in AlaRS editing function could lead to misfolded protein accumulation in additional cell types. Indeed mouse embryos homozygous for a point mutation at C723 (which corresponds to the C666 amino acid in the editing domain of bacterial AlaRS) died by midgestation. Furthermore, decreasing the amount of the sti-associated AlaRS^A734E enzyme or placing the sti mutation and the more severe AlaRS (Aars^C723A) mutation in trans caused extensive aggregation of misfolded proteins in mouse cardiomyocytes, eventually leading to cardiomyopathy. Using this proteome-wide mistranslation system, we demonstrate that different cell types are distinctly sensitive to errors in protein synthesis and that mistranslation can induce cardiac proteinopathy, which leads to cardiac fibrosis and decrements in heart function.

Results

Previous results determined that Cys666 in the editing domain of bacterial AlaRS was essential for AlaRS editing (11). In agreement, mutation of Cys723 (corresponding to bacterial Cys666) to Ala in mammalian AlaRS resulted in a dramatic increase in production of mischarged Ser-tRNA^Ala (Fig. 1A, Left). In contrast to the twofold increase in mischarged Ser-tRNA^Ala generated by AlaRS^A734E, the mutant enzyme found in sti mutant mice (10), in vitro misacylation assays demonstrated that the C723A enzyme produced Ser-tRNA^Ala at initial rates at least 15-fold higher than that produced by the wild-type enzyme.

Fig. 1.

Generation of editing-defective AlaRS mouse models. (A, Left) Misacylation of tRNA^Ala with serine by mouse wild-type (WT) or C723A AlaRS. (Right) Relative level of mischarged Ser-tRNA^Ala in the presence of WT, C723A, or equal molar ratio of WT and C723A AlaRS at 15 min. Assays were run in duplicate, and points (mean ± SD) were averaged from two experiments. (B) To generate the conditional knock-in allele of AlaRS^C723A, a loxP-flanked transcriptional stop cassette was inserted in intron 1, and the TGT codon encoding cysteine at position 723 in exon 15 was replaced with a GCG alanine codon. Aars transcription remains off until the stop cassette is deleted via Cre expression. The Southern blot probe (5′ probe), the size of the left and right “arm” of the targeting vector, and the distance between the neo cassette and exon 15 are indicated. (C) Body weight of WT (n = 13), Aars^sti/sti (sti/sti; n = 8), and Aars^stop/sti (stop/sti; n = 11) mice at postnatal ages (P) 14 and P21. Values represent mean ± SD; ***P < 0.0001 (one-way ANOVA). (D) Photograph of 2-mo-old WT, Aars^sti/sti, and Aars^stop/sti mice. Note the dorsal alopecia characteristic of the Aars^stop/sti mouse.

To determine the in vivo effects of this dramatic loss in AlaRS editing activity, the C723A mutation was introduced into the mouse Aars gene via homologous recombination (Fig. 1B). To achieve conditional expression of the knock-in allele, a loxP-flanked transcriptional stop cassette (13, 14) was also inserted in intron 1 of this gene. Constitutive expression of Aars^C723A was induced by crossing Aars^stop/+ mice to the EIIa-Cre strain to remove the stop cassette (15). The resulting Aars^C723A/+ offspring were viable and indistinguishable from wild-type littermates. We hypothesized that wild-type AlaRS in the Aars^C723/+ mouse could edit in trans the Ser-tRNA^Ala produced by C723A AlaRS. An in vitro mixing of C723A AlaRS with wild-type AlaRS confirmed this expectation (Fig. 1A, Right).

Still, careful examination of brains from aged Aars^C723A/+ mice revealed mild loss of cerebellar Purkinje cells (Fig. S1). This minor neuronal loss could result from the ability of eukaryotic translation elongation factor 1 alpha to capture some of the mischarged tRNA and prevent its clearance by wild-type AlaRS (10). In comparison, Purkinje cell loss is not observed in aged Aars^sti/+ mice, and Aars^sti/sti mice have early onset and severe Purkinje cell degeneration (10). Moreover, no abnormalities were observed in brain regions other than the cerebellum or in other organs (SI Materials and Methods) from 10- to 18-mo-old Aars^C723A/+ mice (Fig. S1). Together these observations suggest that due to the further reduction in AlaRS editing activity, the effects of the Aars^C723A mutation are more detrimental than the Aars^sti mutation in vivo. Consistent with this notion, no Aars^C723A/C723A offspring were obtained from Aars^C723A/+ intercrosses, demonstrating that homozygous expression of this mutation was lethal (Table S1). In addition, no viable E8.5/E9.5 Aars^C723A/C723A embryos were observed, and 26% (n = 9/34) of decidua at these time points contained embryos that were too reabsorbed for genotyping, suggesting that the homozygous embryos died at an earlier stage of development (Table S1). These results, compared with our previous findings of neuron-specific defects in the sti mutant mouse (10), indicate that the Aars^C723A mutation likely causes a higher rate of translational errors in vivo than the sti mutation and that phenotypes caused by deficiencies in AlaRS editing are dependent on the extent of the residual editing activity of the mutant enzyme.

To create a mouse model with an editing deficiency between that of Aars^C723A/C723A and Aars^sti/sti mice, we generated Aars^stop/sti compound heterozygotes. These mice were viable and present at weaning age in the expected Mendelian ratios. Although initially indistinguishable from wild-type littermates, at 3 wk of age Aars^stop/sti mice were smaller than wild-type or Aars^sti/sti mice (Fig. 1C). In addition, larger regions of alopecia were consistently observed in Aars^stop/sti than in Aars^sti/sti mice (Fig. 1D). The gene dosage effect of the Aars^sti mutation reinforces the notion that cells differ in their sensitivity to disruptions in translational fidelity.

Histological analysis of the Aars^stop/sti brain demonstrated that the temporal and spatial pattern of Purkinje cell degeneration was very similar to that observed in Aars^sti/sti mice (Fig. S1). However, gross morphological examination of other tissues from 10-mo-old mice revealed that the heart of Aars^stop/sti mice was smaller than those from age-matched Aars^sti/sti or wild-type mice (Fig. 2 A and B). Although histological analysis of other organs did not reveal abnormalities, myofiber disarray was observed in the ventricular myocardium of hearts from Aars^stop/sti mice. Masson’s trichrome and picrosirius red staining of sections from 4- and 10-mo-old Aars^stop/sti mice revealed extensive fibrosis, which typically forms where cardiomyocytes have degenerated. Both interstitial and perivascular fibrosis was observed in the left and right ventricular walls and in the interventricular septum (IVS) of affected hearts (Fig. 2 C and D and Fig. S2). Moreover, the development of cardiac fibrosis in the Aars^stop/sti mouse appeared to be age-related: At 1 mo of age, no apparent fibrosis was observed in heart sections; at 4 mo of age, ∼2.5% of the heart tissue was fibrotic; and by 10 mo of age, fibrotic areas increased to about 7.5% of the Aars^stop/sti heart (Fig. 2G). In contrast, no fibrotic lesions were observed in 10-mo-old Aars^sti/sti or wild-type mice (Fig. 2 E–G).

Fig. 2.

Cardiac abnormalites in Aars^stop/sti mice. (A) Ratio of the heart weight to femur length (HW/FL) from 10-mo-old WT (n = 11), Aars^sti/sti (sti/sti; n = 6), and Aars^stop/sti (stop/sti; n = 8) mice. Values represent mean ± SD; *P < 0.01 (one-way ANOVA). (B) Hematoxylin and eosin staining of longitudinal sections of hearts from 10-mo-old mice with the indicated genotypes. (Scale bar, 2 mm.) (C–F) Masson’s trichrome stain of heart sections from 4- (C) and 10-mo-old (D) Aars^stop/sti and 10-mo-old Aars^sti/sti (E) and WT (F) mice. Higher magnification photos of a representative area of each image are shown in the Lower panel. Note the high levels of interstitial fibrosis (blue; white arrows) and perivascular fibrosis (blue; black arrows) in Aars^stop/sti but not Aars^sti/sti or WT hearts. (Scale bars, 200 μm and 60 μm for the Upper and Lower panel, respectively.) (G) Quantification of fibrotic areas (as a percentage of total area). Values are mean ± SD; *P < 0.01 (unpaired t test) and **P < 0.001 (one-way ANOVA); n = 4 mice for each data point. (H) Ejection fraction (EF%; Left graph) and fractional shortening (FS%; Right graph) of 10-mo-old WT (n = 15), Aars^sti/sti (n = 6), and Aars^stop/sti (n = 9) mice. Values are mean ± SD; ***P < 0.0001 (one-way ANOVA).

In an alternative effort to diminish the editing activity of the AlaRS enzyme in vivo, we generated Aars^C723A/sti mice. Similar to Aars^stop/sti mice, these mice were smaller than Aars^sti/sti mice, with larger regions of dorsal alopecia. In addition, widespread myocardial fibrosis was present in the hearts of 4-mo-old Aars^C723A/sti mice (Fig. S3), confirming that editing activity in between that of AlaRS^A734E or AlaRS^C723A results in death of cardiomyocytes.

Fibrotic cardiac muscle is stiffer and less compliant than normal heart muscle and is commonly associated with cardiac failure (16). To determine if fibrosis affected the contractile function of the Aars^stop/sti heart, echocardiography was performed on 10-mo-old Aars^stop/sti, Aars^sti/sti, and wild-type mice (Fig. S4 and Table S2). Echocardiogram parameters revealed that the Aars^stop/sti mutant heart exhibited a slight, but significant, reduction in the thickness of the left ventricular posterior wall. Although not significant, the Aars^stop/sti septum also tended to be thinner than that of wild-type mice (IVSs). Consistent with the observed pathology, less systolic thickening of the left ventricle was seen in the Aars^stop/sti heart, indicating myocardial dysfunction. Furthermore, ventricular output, as measured by the ejection fraction and fractional shortening of the left ventricle, was significantly reduced in Aars^stop/sti but not Aars^sti/sti or Aars^C723A/+ mice (Fig. 2H and Fig. S1).

Electrophysiological function of Aars^stop/sti and Aars^sti/sti hearts was further evaluated by electrocardiogram telemetry in 10-mo-old mice. Electrocardiographic characteristics were similar between different genotypes, and ventricular arrhythmias were not observed in any mice (Fig. S5).

Previously we showed that accumulation of misfolded proteins in Aars^sti/sti Purkinje cells accompanied their degeneration (10). To determine whether similar mechanisms occurred in Aars^stop/sti cardiomyocytes, ubiquitin immunofluorescence was performed on heart sections of 10-mo-old mutant and wild-type mice (Fig. 3A). In the wild-type heart, ubiquitin signals were present in the cytoplasm and in the nucleus, as previously reported (17). However, ubiquitin-positive puncta characteristic of protein aggregates were not detected in the wild-type heart. In contrast, ubiquitin puncta were present in ∼2.4% of cardiomyocytes in Aars^stop/sti hearts (n = 4). In addition, large vacuoles and a reduction in fiber diameter were often observed in aggregate-containing cardiomyocytes, consistent with cellular atrophy and cell death (Fig. 3A).

Fig. 3.

Formation of misfolded protein aggregates in Aars^stop/sti mutant cardiomyocytes. (A) Immunofluorescence with antibodies to ubiquitin on heart sections from 10-mo-old WT and Aars^stop/sti mice. Nuclei are counterstained with Hoechst dye (blue). Higher magnification photos of Aars^stop/sti cardiomyocytes are shown in the Lower panel. Arrowheads indicate myofibers with intense ubiquitin signals and reduced fiber diameter. An arrow indicates a cardiomyocyte with elevated ubiquitin and vacuoles. (B–D) High-magnification confocal microscopic images showing colocalization of p62 with ubiquitin (B), LC3 (C), and Hsp70 (D) within the cytoplasm of representative cardiomyocytes of 2-mo-old Aars^stop/sti mice. [Scale bars, (A, Upper and Middle panels) 100 μm, (A, Lower panel) 20 μm, and (B–D) 10 μm.]

To determine the onset of proteinopathy in mutant cardiomyocytes, ubiquitin immunofluorescence was performed on hearts from younger Aars^stop/sti mice. At 1 mo of age, most cardiomyocytes in Aars^stop/sti hearts appeared normal, with cells containing ubiquitinated aggregates only occasionally observed. However, the number of aggregate-containing cells drastically increased by 2 mo of age, constituting ∼1.8% of total cells in the Aars^stop/sti heart. In comparison, cardiomyocytes of Aars^sti/sti mice appeared normal, and ubiquitin-positive inclusions were rarely observed at any given age. Our observations that protein aggregates occurred in prefibrotic Aars^stop/sti hearts and that the number of cardiomyocytes with aggregates increased with age suggest that misfolded proteins induce cardiomyocyte death.

In addition to the ubiquitin–proteasome protein degradation pathway, aggregates of ubiquitinated proteins can be degraded by autophagy. Thus, autophagy is often up-regulated as an adaptive response to the increased production of misfolded proteins. Correspondingly, suppression of autophagy in diseased hearts leads to cardiomyopathy (5). The autophagic adapter protein p62/sequestosome 1 noncovalently interacts with ubiquitinated proteins through its C-terminal domain and facilitates their degradation via autophagy (18, 19). Recent studies demonstrate that p62 is often associated with aggresomes in desmin-associated cardiomyopathies (5, 20). Coimmunofluorescence using antibodies to p62 and ubiquitin in 2-mo-old Aars^stop/sti hearts demonstrated that the majority of ubiquitinated aggregates also contained p62 (Fig. 3B). Ubiquitin and p62-positive puncta were also observed in cardiomyocytes of Aars^C723A/sti mice, further demonstrating the convergence of the two mouse models (Fig. S3).

Immunofluorescence with antibodies to microtubule-associated protein 1 light chain 3 (LC3), a marker of autophagosomes, on sections of Aars^stop/sti hearts also showed colocalization of LC3 and protein aggregates (Fig. 3C). In addition, coimmunofluorescence with antibodies to stress-inducible Hsp70, a molecular chaperone often colocalized with intracellular inclusions (21, 22), revealed colocalization of p62-positive aggregates with Hsp70, further confirming that these structures contain misfolded or unfolded proteins (Fig. 3D). Lastly, because AlaRS editing defects have the potential to cause mistranslation of any protein containing alanine, we determined if protein aggregates in Aars^stop/sti hearts contained desmin, previously shown to form aggregates in the genetic desminopathy disorders (23). In both wild-type and Aars^stop/sti hearts, desmin was detected at the sarcomeric Z-disk (Fig. S6). However, no colocalization of p62 and desmin was observed in the Aars^stop/sti heart, suggesting desmin is not a major component of protein aggregates.

We further performed transmission electron microscopy that revealed multiple pathological changes in Aars^stop/sti cardiomyocytes. Consistent with our immunofluorescence observations, deposition of protein aggregates that disrupt myofibrillar structure was observed in mutant, but not wild-type, cardiomyocytes (Fig. 4A). In addition, many mutant cardiomyocytes had abnormal accumulation of autophagic vacuoles of different stages, indicative of elevated autophagy (Fig. 4 B and C). Multilamellar bodies indicative of impaired lysosomal degradation of autophagic cargos (24) were apparent in mutant, but not wild-type, cells (Fig. 4 D–F), suggesting that autophagic clearance is likely compromised in Aars^stop/sti mutant cells. In addition to accumulation of autophagic structures, we observed abnormal mitochondria and disruption of the sarcoplasmic reticulum in mutant cardiomyocytes (Fig. 4 B, G, and H). These pathological changes may reflect the toxic effects of misfolded proteins and may contribute to degeneration of cardiomyocytes.

Fig. 4.

Ultrastructural changes in Aars^stop/sti cardiomyocytes. (A–I) Representative transmission electron microscope images of hearts from 2-mo- (B–E, G, and H) and 5-mo-old (A and F) Aars^stop/sti mice and 2-mo-old WT (I) mice. (A) Formation of large protein aggregates (arrowheads) that disrupt myofibril (Mf) structures. (B and C) Accumulation of autophagic vacuoles (white arrow, early autophagosome; black arrow, autolysosome) accompanied by loss of Mf structures. (D–F) Multilamellar bodies (MLBs, arrows), products of active autophagy and a hallmark of insufficient lysosomal activity, are also present. (G) Defective mitochondria (Mt) with disrupted cristae (arrowheads). (H) Disrupted sarcoplasmic reticulum (arrowheads) in Aars^stop/sti cardiomyocytes. [Scale bars, (A, B, and D–I) 1 μm and (C) 1 μm.]

Discussion

The effect of mistranslation on the cellular homeostasis of complex higher organisms remains poorly understood. To explore the pathological impact of increased translational errors in vivo, we genetically decreased the editing function of the mouse AlaRS enzyme. This decrease in editing in turn affects global translation by increasing the incorporation of noncognate amino acids into nascent proteins. Our “fine-tuning” of the editing efficiency of AlaRS demonstrated that the severity of the resulting phenotypes directly correlated with the residual editing activity of the enzyme. The originally identified Aars^sti (A734E) mutation is the mildest allele in our series. Homozygosity for this mutation selectively induces protein aggregation and apoptosis in cerebellar Purkinje cells (10). In contrast, embryos homozygous for the Aars^C723A allele, which causes dramatic increases in mischarged Ser-tRNA^Ala, died before midgestation, suggesting that editing activity is essential for mammalian development. Interestingly, mice that were heterozygous for the Aars^C723A mutation were largely phenotypically normal, with only loss of a minor population of Purkinje cells in aged mice. This finding indicates that AlaRS expressed from the wild-type allele is sufficient in trans to hydrolyze the Ser-tRNA^Ala generated by the mutant enzyme. Further genetic regulation of the editing efficacy of the Aars^sti allele by either lowering the level of the enzyme (as in Aars^stop/sti mice) or placing the Aars^sti allele over the Aars^C723A mutation resulted in phenotypes intermediate to those in Aars^sti/sti and Aars^C723A/C723A mice, including loss of Purkinje cells and widespread protein aggregate formation in cardiomyocytes and subsequent death of these cells. These results suggest that the response to mistranslation varies among cells and that protein aggregates may be induced in additional cell types upon further decreases in translational fidelity.

Our results define a common mechanism of protein aggregation in cardiomyocytes and neurons. The underlying sensitivity of these two cell types to mistranslation may lie in the postmitotic nature of these cells, in which protein aggregates are not diluted by cell division (25). Moreover, hydrolysis of misacylated tRNA may be subject to additional regulation. Indeed, Ser-tRNA^Ala can also be hydrolyzed by a free-standing editing domain protein, Aarsd1 (AlaXP) (26, 27), which could compensate for editing deficiencies in AlaRS. Lastly, the differential sensitivity of cells to translation errors may be due to the varying capacity of protein quality control systems in different cell types.

Although decreased translation fidelity causes protein misfolding and aggregation in both cardiomyocytes and Purkinje cells, the downstream pathways that lead to cell death in the two cell types may be quite different. Purkinje cells in sticky mutant mice undergo apoptosis (10), whereas cardiomyocytes in the Aars^stop/sti mouse were negative for cleaved caspase 3 and TUNEL labeling. In addition, abnormal mitochondria with disorganized cristae that may be a contributing factor in cell death were observed in Aars^stop/sti cardiomyocytes. Interestingly, a point mutation in the editing domain of human mitochondrial AlaRS (AARS2) has been associated with infantile mitochondrial cardiomyopathy (28), suggesting increased errors in either cytoplasmic or mitochondrial protein synthesis can lead to cell death in the heart.

In summary, our data show that a global reduction in translational fidelity, rather than disruption of a specific protein, can induce defects in proteostasis in numerous cell types in the mouse. Indeed, fidelity of protein synthesis is not only subjected to genetic changes that modulate translation machinery, it can also be affected by environmental factors such as oxidative stress. The cysteine 182 residue that is critical for editing activity of threonyl-tRNA synthetase in E. coli can be oxidized by reactive oxygen species, and thus, overall translational fidelity is impaired as a result (29). In mammalian cells, various environmental factors that trigger oxidative stress are known to increase Met-misacylation of tRNAs (30). It is thus possible that proteinopathy can occur in the brain, heart, or even other tissues as a combination of genetic and/or environmental factors that compromise translational fidelity.

Materials and Methods

Misacylation Assays.

Misacylation assays were performed with mouse wild-type or mutant C723A AlaRS, 10 μM ³H-l-serine, and 10 μM human tRNA^Ala following methods described in ref. 31. Overexpression and purification of human tRNA^Ala was performed as previously described (10). Details are provided in SI Materials and Methods.

Mice.

All animal protocols were approved by The Jackson Laboratory Animal Care and Use Committee. Homologous recombination of the Aars^stop allele was performed in C57BL/6J-Tyr^c-2J ES cells. Aars^sti/sti and Tg(EIIa-cre)C5379Lmgd/J mice were maintained on a C57BL/6 genetic background (10, 15). Generation of the knock-in mouse and genotyping information are described in SI Materials and Methods.

Echocardiography and Electrocardiography.

Cardiac function, heart dimensions, and electrical activity of the heart were evaluated by echocardiography and electrocardiography on anesthetized mice. Details are provided in SI Materials and Methods.

Histology.

Hematoxylin and eosin, Masson’s trichrome, and picrosirius red staining was performed on 10% (vol/vol) neutral buffered formalin (NBF)-fixed tissue sections. Details are described in SI Materials and Methods.

Immunofluorescence.

The 10% NBF-fixed, paraffin-embedded heart sections were deparaffinated, rehydrated, and microwaved in 0.01 M citrate buffer (pH 6.0) three times for 2 min each. Tissue sections were then blocked with 4% (wt/vol) goat serum in phosphate-buffered saline with Tween-20 (PBST) for 30 min, incubated with primary antibodies overnight at 4 °C, washed in PBST, and incubated with secondary antibodies for 1 h. Details about antibodies are described in SI Materials and Methods.

Electron Microscopy.

Mice were transcardially perfused with 2% (wt/vol) paraformaldehyde and 2% (wt/vol) glutaraldehyde. Tissues were prepared for analysis using standard procedures.

Statistics.

Results are mean ± SD. Statistical significance was determined by unpaired t tests or one-way ANOVA followed by Dunnett’s multiple-comparisons test. P values less than 0.05 were considered significant.