Dysregulated Autophagy and Sarcomere Dysfunction in Patients With Heart Failure With Co‐Occurrence of P63A and P380S BAG3 Variants

Ethics Statement

All research involving human subjects was performed with strict adherence to the ethical guidelines established by the 1975 Declaration of Helsinki. These studies were performed after receiving written informed consent from patients or their legal representatives and with institutional review board approval of each of the participating centers: University of Pittsburgh, University of Colorado, and Loyola University Chicago.

Tissue Procurement

Human samples used were obtained from the Loyola Cardiovascular Research Institute, University of Pittsburgh, and University of Colorado biorepositories. The nonfailing control tissue was obtained from donor hearts rejected due to size, age, or other incompatibility. Idiopathic DCM and BAG3 ^63/380 mutation samples were obtained from hearts explanted during left ventricular assist device placement or transplant procedures.

Myofilament and Soluble Protein Enrichment

Frozen human left ventricular tissue was homogenized using a mechanical homogenizer in 1 mL of a standard rigor buffer (120 mmol/L CH₃CO₂K, 50 mmol/L HEPES, 4 mmol/L MgCl₂, 5 mmol/L EGTA, pH 7.0) with 1% Triton X‐100 and protease and phosphatase inhibitors. The homogenate was left on ice for 20 minutes. After this incubation, the myofilament fraction was pelleted by centrifugation at 4 °C and 1800g for 2 minutes. The supernatant (containing the soluble fraction) was collected and stored at −80 °C. The pellet was washed with standard rigor buffer without Triton twice before being solubilized by resuspension in 8M urea/0.2% SDS. The resuspended pellet (containing the myofilament fraction) was sonicated and stored at −80 °C.

Western Blotting

Protein concentration was determined by bicinchoninic acid assay (Pierce). Samples were heated at 95 °C for 10 minutes in sample buffer (3:1 vol/vol SDS Tris‐Glycine Buffer [Life Technologies] and Bolt Reducing Agent [Fisher Scientific]) and loaded on 4% to 12% Tris‐Glycine gels (Invitrogen). After separation by electrophoresis, protein was transferred onto nitrocellulose membrane. Equal loading was assessed by Revert Total Protein Stain. After staining, the blot was blocked with Tris‐buffered saline and Intercept Blocking Buffer (LI‐COR Biosciences) (1:1 dilution) for 1 hour at room temperature. The blot was incubated overnight at 4 °C with agitation in blocking solution with primary antibody: BAG3 (Proteintech, 10 599‐1‐AP 1:4000), HSP70 (Proteintech, 10 995‐1‐AP, 1:4000), HSPB8 (Proteintech, 15 287‐1‐AP, 1:1000), Alpha B Crystallin (Proteintech, 15 808‐1‐AP, 1:2500), SYNPO2 (synaptopodin‐2; Proteintech, 25 453‐1‐AP 1:2500), Ubiquitin (Cytoskeleton Inc., AUB01, 1:500–1000), LC3 (CST, 2775S, 1:500–750), P62 (Proteintech, 18 420‐1‐AP, 1:2500), Troponin I (IPOC, MA‐1040, 1:2000), pS23/24 Troponin I (CST, 1:1000), α‐Actin (Sigma, A2172, 1:5000). The membrane was washed 3 times using Tris‐buffered saline with Tween before a final incubation of 1:1 blocking solution with tween and a near‐IR secondary antibody (LI‐COR Biosciences, 1:10 000) for 1 hour. The blot was washed 3 times using Tris‐buffered saline and imaged using an Azure c600. Densitometry analysis was performed using the LI‐COR Image Studio software with protein of interest signal normalized to total protein signal.

Myofilament Functional Assessment

Left ventricular tissue was mechanically homogenized in Isolation Buffer (8.91 mM KOH, 2 mmol/L EGTA, 7.11 mmol/L MgCl₂, 10 mmol/L imidazole, 108.01 mmol/L KCl, 5.8 mmol/L adenosine triphosphate) containing 0.5% Triton X‐100 with 3 one‐second pulses at 7000 RPM and left on ice for 20 minutes to demembranate the myocytes. Myocytes were passed through a 70‐μm filter and pelleted by centrifugation at 120g and resuspended in Isolation Buffer without triton. Myocytes were attached by UV‐curing glue (NOA61, Thor Labs) to 2 pins, 1 attached to a calibrated force transducer (Kronex, AE801) and the other to a Piezo length controller. Calcium‐activated force measurements were performed at 6 different calcium concentrations (0.79, 0.97, 1.28, 2.41, 3.84, and 46.8 μmol/L), and the force produced was normalized to the myocyte cross‐sectional area, as previously described. ²¹ Data were fitted to a Hill Curve to determine the myocyte maximum force and EC₅₀ (calcium concentration required to elicit half‐maximal force). All experiments were performed at room temperature and a sarcomere length of 2.1 μm as measured by fast Fourier transform.

Mass Spectrometry

The differentially expressed proteins from samples of nonfailing and failing human hearts were identified by using a modified in stage tip method and mass spectrometry analysis as described in detail previously. ²² , ²³ Briefly, the hearts were homogenized in M‐PER buffer (Mammalian Protein Extraction Reagent, Thermo Fisher Scientific) containing protease and phosphatase inhibitors. Guanidinium hydrochloride buffer (6M) was then added, and the homogenate was heated for 5 minutes at 95 °C. Lys‐C was then added and the lysate was diluted 5‐fold with 25 mmol/L Tris, 10% acetonitrile buffer. The proteins were digested for 4 hours at 37 °C, and a second extensive digestion was achieved by overnight incubation at 37 °C with trypsin. The digestion was stopped by acidification with 3% trifluoroacetic acid followed by centrifugation at 2000 RCF for 5 minutes. The supernatant was collected and loaded onto an activated in‐house‐made cation stage tip, and the peptides were eluted into 6 fractions. The desalted tryptic peptide samples were then loaded onto an Acclaim PepMap 100 precolumn (C18, 75 μm×2 cm, Thermo Scientific) and separated by Easy‐Spray PepMap RSLC C18 column with an emitter (2 μm, 100Å, 50 μm×15 cm, Thermo Scientific) by an Easy nLC system with Easy Spray Source (Thermo Scientific). To elute the peptides, a mobile‐phase gradient was run using increasing concentration of acetonitrile. Peptides were loaded in buffer A (0.1% formic acid) and eluted with a nonlinear 145‐minute gradient as follows: 0% to 25% buffer B (0.1% formic acid, 85% acetonitrile) for 80 minutes, 25% to 40% B for 20 minutes, 40% to 60% B for 20 minutes and 60% to 100% B for 10 minutes. All percentages were volume/volume. The column was then washed with 100% buffer B for 5 minutes, 50% buffer B for 5 minutes, and reequilibrated with buffer A for 4 minutes. The flow rate was maintained at 300 nL/min.

Electron spray ionization was delivered at a spray voltage of −2000 V. Tandem mass spectrometry fragmentation was performed on the 5 most abundant ions in each spectrum using collision‐induced dissociation with dynamic exclusion (excluded for 10.0 s after 1 spectrum), with automatic switching between mass spectrometry (MS) and tandem MS modes. The complete system was entirely controlled by Xcalibur software (Thermo Fisher Scientific). Mass spectra processing was performed using Proteome Discoverer version 2.4 (Thermo Fisher Scientific). The generated deisotoped peak list was submitted to 3 separate databases: Mascot, MS Amanda 2.0, and Sequest HT. All 3 databases’ search parameters were set as follows: species, homo sapiens; enzyme, trypsin with maximal 2 missed cleavages; minimum peptide length, 4; maximum peptide length, 144; static modification, carbamidomethyl/+57.021 Da(C); 10 ppm precursor mass tolerance and 0.6 Da fragment mass tolerance for tandem MS fragment ions. Gene ontology analyses of the differentially expressed proteins were performed using the DAVID Bioinformatics platform. ²⁴ , ²⁵

Statistical Analysis

Data analysis and graphical representation were performed with Prism 9 (GraphPad software), unless otherwise noted. Comparisons of 2 groups were performed by 2‐tailed Student's t‐test. Comparisons of 3 groups were performed by 1‐way ANOVA. When a significant interaction was identified, multiple comparisons were performed using Tukey's post hoc test. For all Western blot data, due to small sample size, analysis was performed by 1‐way permutational AVOVA using RStudio (R Foundation for Statistical Computing). Experimental group blinding was not done.

Dysregulated Autophagy With BAG3 ^{63 /380} Mutations

Earlier work with the BAG3 P63A and P380S variants found that they were associated with a worse prognosis in DCM compared with DCM from nongenetic causes. ¹⁹ The mechanism of dysfunction remains poorly defined, but neither variant is in an HSP‐binding domain of BAG3 (Figure 1A). The mechanisms of dysfunction with the 63/380 variants are critical to understand, as there are multiple other disease‐associated BAG3 variants that are also located outside of these domains. Thus, with institutional review board approval, we procured left ventricular tissue from explanted hearts or pretransplant endomyocardial biopsies collected from patients with nonfailing (n=4; rejected donor hearts), idiopathic DCM (n=5; explanted hearts), and BAG3 ^63/380 (n=2; left ventricular assist device cores) from patients at the University of Pittsburgh, University of Colorado, and Loyola University Chicago (Table) and examined the molecular impact of these variants to determine broadly how BAG3 variants outside of the HSP‐binding domains may be pathogenic. To begin, we performed an unbiased quantitative proteomics screen of the left ventricular proteome in BAG3 ^63/380 compared with nonfailing. Label‐free quantification MS showed 78 proteins with significantly altered abundance in the BAG3^63/380 hearts compared with nonfailing (Figure 1B). Gene ontology biological process analysis revealed that these proteins were primarily involved in protein translation, muscle contraction, and autophagy (Figure 1C).

Because BAG3 is an established regulator of protein degradation by autophagy, ²⁶ we next examined whether autophagy was altered in the tissue of patients with BAG3 ^63/380. During maturation of autophagosomes, the early phagophore‐associated LC3‐I protein is converted to LC3‐II; therefore, relative expression of these 2 forms of LC3 can provide insight into autophagic activity. ²⁷ By Western blot, we found that while LC3‐I protein expression was consistent among the 3 groups, LC3‐II levels were increased in the BAG3 ^63/380 samples, suggesting altered autophagy (Figure 2A through 2C). We next assessed the expression of another autophagy marker, P62, which is necessary for connecting ubiquitinated protein substrates to the autophagosome to facilitate their removal. We found that P62 levels increased by 37% and 23% in the patients with BAG3 ^63/380 compared with nonfailing DCM and DCM, respectively, although statistical significance was not reached (Figure 2D). Protein ubiquitination levels, however, were found to be significantly increased with BAG3 ^63/380 compared with both the nonfailing and DCM groups (Figure 2E).

Altered Expression of BAG3‐Binding Proteins in the Left Ventricle of Patients With BAG3 ^{63 /380}

The involvement of BAG3 in autophagy stems from mediating the formation and regulation of a multichaperone protein assembly termed the chaperone‐assisted selective autophagy complex. Through its 2 IPV motifs, BAG3 associates with the adenosine triphosphate–independent small HSPs, which bind to denatured proteins and prevent their aggregation. The interaction of HSPBs with BAG3 allows for the connection of their cargo with HSP70—bound to BAG3's C‐terminal BAG domain—and subsequent ubiquitination and trafficking to an autophagosome. SYNPO2 assists with autophagosome maturation and engulfment of the chaperone‐assisted selective autophagy complex with its associated substrates (Figure 1A). ⁴ , ²⁸ Due to our observation of disrupted autophagy in the BAG3 ^63/380 tissue, we hypothesized that this could result from a mutation‐dependent effect on the expression/localization of chaperone‐assisted selective autophagy complex proteins.

We used Western blot to assess expression of BAG3, HSP70, SYNPO2, HSPB8, and HSPB5 (also known as CRYAB) in the triton‐soluble/cytosolic protein fraction of the nonfailing, DCM, and BAG3 ^63/380 samples (Figure 3A and 3B). BAG3 protein expression was reduced in DCM as reported previously ¹⁹ , ²⁹ , ³⁰ ; however, no change was observed for the patients with BAG3 ^63/380. HSP70 and CRYAB protein expression were similar between nonfailing and BAG3 ^63/380. However, HSPB8 and SYNPO2 levels were each reduced by ≈50% in the BAG3 ^63/380 tissue (Figure 3C through 3G). A similar decrease was found in the DCM group for HSPB8, suggesting a mutation‐independent change in expression in DCM, which we have previously shown is due to the reduction in BAG3 levels. ² , ¹⁸ However, for SYNPO2, the DCM group displayed a ≈60% increase, suggesting the observed decrease in SYNPO2 in patients with BAG3 ^63/380 was a mutation‐specific effect.

Our previous work showed that BAG3 and its binding partners localized to the sarcomere under proteotoxic stress conditions and that this localization was partially BAG3 dependent. ² Therefore, we next investigated expression of these proteins in the myofilament protein fraction to determine if BAG3 ^63/380 disrupts their sarcomeric localization. Using Western blot on myofilament‐enriched protein lysates, we found similar expression in the nonfailing, DCM, and BAG3 ^63/380 samples for BAG3, HSPB8, SYNPO2, and HSP70 (Figure 4A through 4F). However, while CRYAB levels were similar between nonfailing and DCM, myofilament CRYAB was completely lost in patients with BAG3 ^63/380 (Figure 4G).

BAG3 ^{63 /380} Cardiomyocytes Display Sarcomere Contractile Dysfunction and Increased Myofilament Protein Ubiquitination

Because several previous studies identified BAG3 as an important factor for maintaining sarcomere structure and function, ² , ³ , ¹² , ¹⁸ , ³¹ , ³² we next assessed myofilament function in cardiomyocytes from the nonfailing, DCM, and BAG3 ^63/380 samples using the skinned myocyte force‐calcium assay. Maximum calcium‐activated force (F_max) was significantly reduced in the DCM group compared with nonfailing (15.27±0.71 versus 24.1±1.62). Notably, F_max was also considerably reduced in the BAG3 ^63/380 cardiomyocytes (11.41±1.44) and trended toward statistical significance when compared with the DCM group (P=0.0897) (Figure 5A and 5B). Cardiomyocyte calcium sensitivity was also increased in DCM compared with nonfailing, as indicated by reduced EC₅₀. However, the EC₅₀ in the BAG3 ^63/380 cardiomyocytes was not significantly different from nonfailing (Figure 5C).

Altered myofilament calcium sensitivity in heart failure is often reflective of changes in the phosphorylation status of troponin I. Troponin I phosphorylation at S23/S24 by protein kinase A regulates myofilament calcium sensitivity. ³³ In heart failure, troponin I phosphorylation decreases, thus increasing the calcium sensitivity. ³⁴ , ³⁵ To determine if the differences in calcium sensitivity observed in our study could be explained by troponin I phosphorylation, we used Western blot for total troponin I and S23/24 phosphorylated troponin I in myofilament protein lysates. Similar qualitative decreases in troponin I phosphorylation were observed in the DCM and BAG3 ^63/380 groups compared with nonfailing (Figure 5D and 5E). However, protein expression of the lowly expressed slow skeletal muscle troponin I isoform were qualitatively lower in the BAG3 ^63/380 samples (Figure 5F).

BAG3 had previously been shown to mediate removal of misfolded sarcomeric proteins by autophagy, and decreased expression/mutation to BAG3 was associated with increased myofilament protein ubiquitination. ² We therefore next assessed total ubiquitin levels in the myofilament protein fraction by Western blot. BAG3 ^63/380 samples displayed a significant increase in myofilament ubiquitin compared with nonfailing (Figure 5G and 5H). This increase was consistent with that observed in the DCM group.

BAG3 P63A Variant Is Predicted to Alter Local Protein Secondary Structure

The P63A and P380S BAG3 variants are caused by missense mutations in exons 2 and 4 of BAG3's 4 coding exons (Figure 1A). Within the protein structure, P63 is slightly outside of a tryptophan–tryptophan domain (AA 21–55), which facilitates interactions with proline‐rich domains of many BAG3‐binding partners including SYNPO2 (Figure 1A). The P380S variant is located in BAG3's own proline‐rich domain, which interacts with the motor protein dynein to transport its misfolded protein substrates to degradation pathways (Figure 1A). ³⁶ , ³⁷

One previous study of a pathogenic BAG3 variant found that replacing proline with leucine at amino acid 209 was predicted to alter local protein secondary structure by increasing beta‐sheet propensity. ³⁸ This resulted in increased exposure of hydrophobic residues on BAG3, making the protein more prone to aggregation. Because the mutations of interest in our 2 patients were also at proline residues, we hypothesized that a similar effect on secondary structure might be observed. We used PepFold ³⁹ to predict whether replacing the prolines at 63 and 380 to alanine and serine, respectively, altered the predicted secondary structure. The P63A mutation greatly increased the predicted β‐sheet propensity and decreased disorder (Figure 6A). No overt differences were predicted by replacing the proline at residue 380 with serine (Figure 6B).

Study Limitations

The study is limited by the low sample number in the BAG3 ^63/380 group. However, we were unable to find additional samples, as genetic cardiomyopathy tissue from the same mutations is rare. The observations made with this sample size should inform future studies in cultured cardiomyocytes. Future work should focus on generating induced pluripotent stem cell–derived cardiomyocytes and isogenic controls from patients with the P63A and P380S variants. We were unable to directly measure autophagy flux in human tissue, as this would require pharmacological inhibition of autophagy before tissue collection. However, our previous in vitro studies of these variants in AC16 cells also identified disrupted autophagy. ¹⁹ Previous work indicated that both reduced BAG3 protein levels and BAG3 mutations alter sarcomere structure, which likely underlies the frequently observed reduced contractile function. ¹¹ , ¹⁴ , ³³ Unfortunately, due to limited tissue amount, ultrastructural assessment was not possible in this study. In addition, the quantitative proteomic assessment of protein expression does not include a DCM group, which makes it difficult to conclude how DCM with the BAG3 variants differs from idiopathic DCM at the global proteome level. Finally, the patients with DCM were not subjected to genetic testing to rule out other possible contributing variants, as there were no indications of familial disease.