The Obscn-ΔIg58/59 constitutive deletion model was generated and genotyped as previously described (10). Animal care and procedures were conducted under protocols approved by the Institutional Animal Care and Use Committee at the University of Maryland, School of Medicine (UMSOM) and in accordance with the NIH guidelines (Guide for the Care and Use of Laboratory Animals).

Lysates were prepared from flash frozen cardiac tissue and protein expression was evaluated by immunoblotting as previously described (10). Briefly, frozen right and left atria were combined, ground to a powder in a glass homogenizer while immersed in liquid nitrogen, and incubated at −20°C for 20 min. The ground tissue was solubilized in urea-thiourea lysis buffer (8 mol/L urea, 2 mol/L thiourea, 3% SDS, 0.05 mol/L tris–HCl, 0.03% bromophenol blue, 0.075 mol/L dithiothreitol, pH 6.8) and 50% glycerol supplemented with protease and phosphatase inhibitors (Halt Protease and Phosphatase Inhibitor Cocktail, Thermo Fisher Scientific, Waltham, MA, United States) in a 60°C water bath. Homogenates were centrifuged and supernatants were aliquoted and flash frozen in liquid nitrogen. For western blotting, equal amounts of protein lysates were thawed at 55°C for 5 min, separated by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose membrane, and probed with the respective primary antibodies. Alkaline phosphatase (AP)-conjugated or horseradish peroxidase (HRP)-conjugated secondary antibodies and the respective chemiluminescent reagents (NovaBright; AP, or Pierce ECL; HRP) were used to detect immunoreactive bands. Densitometry was performed using ImageJ and each band was normalized to a loading control (glyceraldehyde 3-phosphate dehydrogenase, GAPDH; heat shock protein 90, Hsp90; or α-actinin). At least two technical replicates of at least three different biological samples (i.e., hearts) were quantified per genotype for each protein evaluated. The original representative blots shown in Figures 1, 2 and 8 are included in Supplementary Image 1; please note that in some instances, immunoblots were flipped for ease of presentation.

The following primary antibodies were used for western blotting: rabbit polyclonal antibodies to obscurin Ig58/59 (1 μg/mL) (11), obscurin Ig67 (1 μg/mL) (10), novex-3 (1:1000, a generous gift from Dr. Henk Granzier) (9), PLN-pSer16 (1:1000; 07–052, Millipore, Temecula, CA, United States), PLN-pThr17 (1:2000; A010-13AP, Badrilla, Leeds, United Kingdom), sAnk1 (1 μg/mL) (12), RyR2-pSer2808 (1:2000; ab59225, Abcam, Cambridge, MA, United States), RyR2-pSer2814 (1:500; A010-31AP, Badrilla), sarcolipin (1 μg/mL, a generous gift from Dr. Robert Bloch) (13), ERK2 (1:1000; CST-9108S, Cell Signaling Technology, Danvers, MA, United States), ERK1/2-pThr183/pTyr185 (1:2000; CST-4370 T, Cell Signaling Technology), rabbit monoclonal antibodies to Hsp90 (1:1000; CST-4877, Cell Signaling Technology), and mouse monoclonal antibodies to PLN (1:5000; ab2865, Abcam), SERCA2 (1:1000; MA3-919, Thermo Fisher Scientific, Waltham, MA, United States), RyR2 (1:1000; MA3-925, Thermo Fisher Scientific), GAPDH (1:15000; G8795, Millipore), and α-actinin (1:2500; A7811, Sigma-Aldrich, St. Louis, MO, United States). The following secondary antibodies were used for western blotting: goat anti-mouse IgG (1:3000; A3688, Sigma-Aldrich), goat anti-rabbit IgG (1:3000; AB_2337947, Jackson Immunoresearch, West Grove, PA, United States), goat anti-mouse IgG (1,3,000; CST-7076S, Cell Signaling Technology), and goat anti-rabbit IgG (1,3,000; CST-7074S, Cell Signaling Technology).

Atrial lysates prepared as described above were separated on 16 × 18 cm gels composed of 1% agarose in 1X running buffer (50 mM tris, 0.384 mol/l glycine, 0.1% SDS) and 30% glycerol using the Hoefer SE600 unit system at 4°C for 3 h as described previously (10). Gels were stained with Coomassie Blue and the bands corresponding to giant titin were quantified using ImageJ and normalized to myosin heavy chain (MHC) as a loading control. Any digital adjustments to promote visualization of the bands were applied uniformly across the entire gel. At least two technical replicates of three different biological samples (i.e., hearts) were quantified for each genotype. The original representative titin gels shown in Figure 2 are included in Supplementary Image 1; please note that in some instances, gels were flipped for ease of presentation.

Proteomic experiments were performed in the Mass Spectrometry Center at the University of Maryland School of Pharmacy. Atrial tissues dissected from male wild-type and homozygous Obscn-ΔIg58/59 mice (n = 5 hearts per group) were homogenized in phosphate buffered saline using the Precellys CK14 lysing kit (Bertin Corp., Rockville, MD, United States). Proteins were extracted and purified from tissue lysates by trichloroacetic acid precipitation. Protein concentration was measured by bicinchoninic acid assay as described previously (14). Lysates were reduced, alkylated, and trypsinolyzed on a 10 K filter for shotgun proteomics as in (15). Phospho-peptides were enriched by TiO₂ affinity chromatography (Sigma). Tryptic peptides were separated by a nanoACQUITY UPLC analytical column on a Waters nano-ACQUITY UPLC system and analyzed with a coupled Thermo Scientific Orbitrap Fusion Lumos Tribrid mass spectrometer as previously described (16). In detail, tryptic peptides were separated on a nano-ACQUITY UPLC analytical column (BEH130 C18, 1.7 μm, 75 μm × 200 mm, Waters) over a 165-min linear acetonitrile gradient (3–40%) with 0.1% formic acid on a Waters nano-ACQUITY UPLC system and analyzed on a coupled Thermo Scientific Orbitrap Fusion Lumos Tribrid mass spectrometer as previously reported (17). Full scans were acquired at a resolution of 120,000 and precursors were selected for fragmentation by higher-energy collisional dissociation (normalized collision energy at 30%) for a maximum 3-s cycle. Tandem mass spectra were searched against a UniProt reference Mus musculus proteome using Sequest HT algorithm (18) and MS Amanda algorithm (19) with a maximum precursor mass error tolerance of 10 ppm. Carbamidomethylation of cysteine was treated as static modification. Phosphorylation of serine (Ser), threonine (Thr), and tyrosine (Tyr), and deamidation of asparagine and glutamine were treated as dynamic modifications. Resulting hits were validated at a maximum global false discovery rate (FDR) of 0.01 using a semi-supervised machine learning algorithm Percolator (20). Label-free quantifications were performed using Minora, an aligned AMRT (Accurate Mass and Retention Time) cluster quantification algorithm (Thermo Fisher Scientific, 2017). Label-free quantitation of protein abundances was measured by comparing the MS1 peak volumes of peptide ions, whose identities were confirmed by MS2 sequencing.

The above abundance values were imported into Partek GS software for further statistical and bioinformatic analyses (Transcriptomics and Deep Sequencing Core, Johns Hopkins University). There were two mappings of the mass spectra: (a) to individual proteins, and (b) to unique phospho-peptides, wherein a phospho-peptide represents a unique specific position within its peptide. Both the individual proteins and unique phospho-peptide proteins were annotated with their cognate genes’ approved MGI/NCBI nomenclature. Following log₂ transformation, the abundance values were then quantile normalized for each time point to minimize experimental noise among the lanes that represent replicate samples for the two biological classes (i.e., genotypes), and the Obscn-ΔIg58/59 samples were compared to the wild type with two-tailed one-way t-test ANOVA. Each protein or unique phospho-peptide compared received a relative abundance and statistical value, as a fold-change and value of p, and the log₂ fold changes were analyzed to determine their standard deviation from the mean value of no change. Proteins and phospho-peptides with a value of p of <0.05 and log₂ fold changes that differed by >2 standard deviations were deemed to be significantly different. The precise phospho-site residues that exhibited a probability >75% are specified, whereas ambiguous phospho-site residues that displayed <75% probability are denoted in the associated phospho-peptide when referenced in the text and tables. Phosphorylation sites not previously annotated in PhosphoSitePlus (v. 6.6.0.4.) were denoted as novel. The raw mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (21) partner repository with the dataset identifier PXD028904. A complete list of all proteins and phospho-peptides that were identified at 6- and 12-months are listed with associated statistics in Supplementary Data Sheets S1–S2 and S3–S4, respectively.

The physical and functional associations of the proteins and phospho-proteins deemed to be significantly different between genotypes were plotted using the STRING database (v.11.0b) (22). To be as inclusive as possible, protein–protein interactions with a confidence score of >0.4 (i.e., medium confidence threshold) were plotted for differentially regulated proteins, while those relationships exhibiting high confidence scores (>0.7) were highlighted within the network. Only high confidence interactions (>0.7) were plotted and highlighted for deregulated phospho-proteins. Disconnected nodes and clusters comprised of less than 3 proteins were excluded.

The differentially expressed proteins and phospho-proteins were further analyzed with the QIAGEN Ingenuity Pathway Analysis (IPA) platform to determine their biological significance. Due to the large number of significantly altered pathways identified, the top 10 most significant pathways and cellular functions that are relevant to cardiovascular physiology and associated with at least 3 deregulated proteins/phospho-proteins were included in the text, figures, and tables. A complete list of all significantly altered pathways and cellular functions identified at 6- and 12-months are listed in Supplementary Data Sheets S5–S6 and S7–S8, respectively.

We recently generated a constitutive deletion mouse model, Obscn-ΔIg58/59, that expresses obscurin lacking the Ig58/59 region and comprehensively evaluated the effects of this deletion on cardiac morphology and function through aging. Our studies showed that male Obscn-ΔIg58/59 mice exhibit episodes of severe atrial arrhythmia by 6-months, manifesting as junctional escape and spontaneous loss of regular p-waves (10). By 12-months, the incidence and severity of arrhythmias intensified accompanied by significant atrial enlargement (10). Notably, female Obscn-ΔIg58/59 mice do not exhibit any structural or functional deficiencies through aging and develop only mild arrhythmia that occurs less frequently compared to Obscn-ΔIg58/59 males (10). Therefore, we focus our molecular characterization on Obscn-ΔIg58/59 males only.

To investigate the mechanistic basis for the development of atrial fibrillation and remodeling in aging Obscn-ΔIg58/59 males, we first evaluated the expression levels of obscurin, the binding partners of obscurin-Ig58/59, and a panel of Ca²⁺ regulators that are commonly associated with the development of atrial fibrillation. Earlier studies assessing the expression of obscurin during embryonic development reported reduced obscurin transcript levels in mouse atria compared to ventricles at embryonic day 12 (23). However, comparison of obscurin expression between adult atrial and ventricular tissues has not yet been experimentally determined. We therefore performed immunoblotting experiments using lysates prepared from 12-month-old wild-type hearts but did not observe significant differences in the levels of giant obscurin between the left ventricle and atria (Figures 1A,A’).

We next evaluated the impact of the Ig58/59 deletion on atrial obscurin expression in sedentary aging animals. Immunoblotting experiments utilizing antibodies to obscurin-Ig58/59 confirmed the absence of this region in lysates generated from 6- and 12-month-old Obscn-ΔIg58/59 atria (Figure 1B). Similar to our prior findings in Obscn-ΔIg58/59 left ventricles (10), antibodies to obscurin-Ig67 did not reveal significant differences in prototypical obscurin A expression between wild-type and Obscn-ΔIg58/59 atria at 6- or 12-months (Figures 1C,C’). Intriguingly, the up-regulation of obscurin B (the largest known isoform containing two serine/threonine, Ser/Thr, kinases) previously reported in Obscn-ΔIg58/59 left ventricular tissue (10) was barely detectable in Obscn-ΔIg58/59 atria (Figures 1C,C’). This finding indicated that the Ig58/59 deletion leads to distinct molecular alterations in Obscn-ΔIg58/59 atria compared to the ventricle and suggested that Obscn-ΔIg58/59 atria potentially lack compensatory signaling mechanisms that could be mediated by obscurin-kinase bearing isoforms in Obscn-ΔIg58/59 left ventricles.

We next evaluated the expression and phosphorylation levels of the known binding partners of Ig58/59 in addition to select Ca²⁺ cycling regulators. Interestingly, there were no statistically significant differences in the levels of giant titin, novex-3, PLN or its phosphorylation at Ser16 or Thr17, small ankyrin 1 (sAnk1), sarcolipin (SLN), SERCA2, RyR2 or its phosphorylation at Ser2808 or Ser2814 in Obscn-ΔIg58/59 atria compared to age-matched controls at either 6- or 12-months (Figures 2A,A’). Therefore, the atrial remodeling and arrhythmia in aging Obscn-ΔIg58/59 male mice cannot be explained by changes in the expression levels and/or canonical phosphorylation sites of these proteins, as is the case for Obscn-ΔIg58/59 left ventricles (10). This suggested alternative mechanisms in the atria, perhaps involving additional Ca²⁺ or cytoskeletal regulators and/or less characterized/novel phosphorylation events. Along these lines, many studies have reported the presence of phosphorylation sites on titin (24), RyR2 (25–27), PLN (28), sAnk1 (29), and SERCA2 (30), for which their (patho)physiological impact has not been established.

Given the lack of significant alterations in the levels of obscurin, the binding partners of obscurin-Ig58/59, and canonical Ca²⁺ cycling regulators, we performed proteomic and phospho-proteomic experiments using 6- and 12-month-old male wild-type and Obscn-ΔIg58/59 atrial tissue (n = 5 hearts per group) to obtain a more comprehensive molecular profile of the Obscn-ΔIg58/59 atria. At 6-months, we identified 45 proteins (out of 1700 detected) that exhibited significantly altered expression levels (Figure 3A; Supplementary Table S1; Supplementary Data Sheet 1) and 30 phospho-peptides (out of 1,401 detected) originating from 27 different proteins that displayed altered phosphorylation levels (Figure 3A’; Supplementary Table S2; Supplementary Data Sheet 2). By 12-months, we identified 48 proteins (out of 1708 detected) with altered expression levels (Figure 4A; Supplementary Table S3; Supplementary Data Sheet 3) and 78 phospho-peptides (out of 2,656 detected) corresponding to 67 different proteins that exhibited altered phosphorylation levels (Figure 4A’; Supplementary Table S4; Supplementary Data Sheet 4). Of the 45–48 affected proteins and 27–67 affected phospho-proteins in Obscn-ΔIg58/59 atria, relatively few were commonly deregulated throughout aging (Figure 5).

In order to discover the functional relationships and/or physical associations shared by the deregulated proteins or phospho-proteins in Obscn-ΔIg58/59 atria through aging, we performed a network analysis using the publicly available STRING database (v.11.0b) (22). At 6-months, there was a single network exhibiting a high confidence interaction score (>0.7) among the proteins displaying deregulated expression in Obscn-ΔIg58/59 atria (Figure 3B), which consisted of ubiquitin-specific protease 14 (Usp14) and the proteasome subunit alpha type-2 (Psma2), possibly reflecting deregulated protein degradation pathways. Also connected to this network, albeit with a medium confidence interaction score (>0.4), is the purine biosynthetic enzyme, phosphoribosylaminoimidizole carboxylase/succinocarboxamide synthetase (Paics), possibly implicating altered DNA synthesis, intracellular signaling, and/or metabolic processes (31). On the other hand, STRING analysis of the deregulated phospho-proteins at 6-months revealed a high confidence network (Figure 3B’) comprised of sarcomeric proteins (i.e., titin, Ttn; myosin light chain 7, Myl7) and cytoskeletal proteins localizing to the Z-disc (i.e., myozenin, Myoz2; synaptopodin 2-like, Synpo2l; LIM domain binding protein 3, Ldb3). This network also included β-taxilin, a muscle-specific member of the taxilin family of vesicular trafficking regulators that is proposed to regulate myoblast differentiation (32). At 12-months, we observed a significant clustering of intermediate filament proteins (i.e., keratins, Krt1, Krt5, Krt6a, Krt14, Krt16, Krt17), in addition to a smaller high confidence network of protein homeostasis regulators (i.e., eukaryotic translation initiation factor 2, Eif2s3x; eukaryotic translation initiation factor 4 gamma 2, Eif4g2; proteasome 20s subunit beta 3, Psmb3; 26S proteasome regulatory subunit 4, Psmd4) that all exhibited reduced expression levels in Obscn-ΔIg58/59 atria (Figure 4B). Lastly, STRING analysis of the deregulated phospho-proteins at 12-months revealed an extensive high confidence network comprised of 23 interconnected sarcomeric proteins, cytoskeletal proteins, ion channels, Ca²⁺ regulators, and kinases (Figure 4B’). Of note, titin represents the most prominent node in the phospho-proteomic network at both timepoints, forming functional and/or physical associations with 4/6 (66%) and 10/23 (43%) of the phospho-proteins that were deregulated at 6- or 12-months, respectively. Given that titin is a binding partner of the obscurin Ig58/59 module, this suggests that the disruption of the obscurin/titin complex could be integral to the deregulated phosphorylation events in Obscn-ΔIg58/59 atria.

To more quantitatively delineate the major molecular pathways and cellular functions that were impacted by the Ig58/59 deletion, we performed an enrichment analysis on the proteins that exhibited significantly altered expression or phosphorylation levels (Figures 6A,B, 7A,B; Supplementary Tables S5–S8). At 6-months, proteins exhibiting altered expression were associated with the regulation of inositol phosphate metabolism (Figures 6A,B; Supplementary Table S5), whereas at 12-months, proteins exhibiting altered expression largely belonged to the keratin subfamily of intermediate filaments (Figures 7A,B; Supplementary Table S7). Additionally, proteins displaying altered phosphorylation were primarily associated with the regulation of cellular assembly/organization (i.e., organization of sarcomeres, filaments, and microtubules), ion transport, and cardiac hypertrophy at 6-months (Figures 6A,B; Supplementary Table S6) and various signaling cascades (i.e., protein kinase A; PKA, integrin, apelin, and Ca²⁺ cycling), striated muscle development, formation, morphology, and hypertrophy, as well as cardiomyopathy and familial arrhythmogenic right ventricular dysplasia at 12-months (Figures 7A,B; Supplementary Table S8).

Given the large number of deregulated proteins and phosphorylation events identified in our proteomic screen, we decided to focus on (1) direct alterations to obscurin, and (2) proteins that belong to enriched molecular pathways or cellular functions and represent nodes within high confidence protein networks identified via STRING analysis. Of note, we discuss these deregulated proteins and phospho-proteins in terms of their canonical protein class since many of them belong to multiple affected cellular processes and/or protein interaction networks.

At 12-months of age, we observed a significant, yet modest, reduction (~1.8 fold) in the expression levels of obscurin in Obscn-ΔIg58/59 atria compared to wild-type (Figure 4A; Supplementary Table S3). Of note, the lack of statistical significance in our immunoblotting analysis (Figures 1C,C’) that revealed only a trend toward decreased obscurin expression in Obscn-ΔIg58/59 atria (~1.1 fold; p = 0.3), is most likely due to the reduced sensitivity of the immunoblotting technique compared to proteomics. Nevertheless, complete knockout or down-regulation of obscurin in striated muscles has been linked to major structural defects, including the disorganization of the longitudinal SR (33), disrupted thick filament assembly (34) and lateral alignment of myofibrils (35), loss of dystrophin at costameres, and alterations in the arrangement of the subsarcolemmal microtubule lattice (36). Therefore, this moderate reduction in obscurin expression in aged Obscn-ΔIg58/59 atria could indicate a mild loss in the structural integrity of the myofibril, the cytoskeleton, and/or the SR membranes. Accordingly, obscurin expression levels are also reduced in human cardiac biopsies carrying DCM-linked point mutations in OBSCN (E963K, V2161D, or D5966N) (37), substantiating that obscurin haploinsufficiency is pathogenic in the heart.

In addition to the essential cytoskeletal protein obscurin, a striking portion of the deregulated proteins and phospho-proteins in aging Obscn-ΔIg58/59 atria were canonical cytoskeletal and structural regulators. In particular, the Ig58/59 deletion induced profound abnormalities in the expression and/or phosphorylation of intermediate filaments, sarcomeric proteins, Z-disk-associated and/or actin-linked cytoskeletal modulators, and structural components of the costamere.

Our proteomic and phospho-proteomic analysis also revealed alterations in the expression and phosphorylation of regulatory proteins and signaling mediators in aging Obscn-ΔIg58/59 atria. Specifically, we observed alterations in canonical regulators of Ca²⁺ cycling and major protein kinases that could potentially contribute to the development of atrial fibrillation in Obscn-ΔIg58/59 mice.