Novel BAG3 Variants in African American Patients with Cardiomyopathy: Reduced β-Adrenergic Responsiveness in Excitation-Contraction.

Abstract

Background.

We reported 3 novel non-synonymous single-nucleotide variants of Bcl2-associated athanogene 3 (BAG3) in African Americans with heart failure (HF) which are associated with 2-fold increase in cardiac events (HF hospitalization, heart transplant or death).

Methods.

We expressed BAG3 variants (P63A, P380S and A479V) via adenovirus-mediated gene transfer in adult left ventricular myocytes isolated from either wild-type (WT) or cardiac-specific BAG3 haplo-insufficient (cBAG3^+/−) mice: the latter to simulate the clinical situation in which BAG3 variants are only found on one allele.

Results.

Compared to WT myocytes, cBAG3^+/− myocytes expressed ~50% of endogenous BAG3 levels and exhibited decreased [Ca²⁺]_i and contraction amplitudes after isoproterenol due to reduced L-type Ca²⁺ current. BAG3 repletion with WT BAG3 but not P380S, A479V or P63A/P380S variants restored contraction amplitudes in cBAG3^+/− myocytes to those measured in WT myocytes, suggesting excitation-contraction (EC) abnormalities partly account for HF in patients harboring these mutants. Since P63A is near the WW domain (residues 21–55) and A479V is in the BAG domain (residues 420–499), we expressed BAG3 deletion mutants (Δ1–61 and Δ421–575) in WT myocytes and demonstrated that the BAG but not the WW domain was involved in enhancement of EC by isoproterenol.

Conclusion.

The BAG3 variants contribute to heart failure in African American patients partly by reducing myocyte excitation-contraction under stress, and that both the BAG and PXXP domains are involved in mediating β-adrenergic responsiveness in myocytes.

Introduction

B cell lymphoma 2 (Bcl2)-associated anthogene 3 (BAG3) is an evolutionarily conserved, stress-activated 575 amino acid protein predominantly expressed in the heart, skeletal muscle and many cancers¹. BAG3 contains multiple protein-protein binding domains (Fig. 1) which confer pleotropic effects on the cell. In the heart, BAG3 regulates protein quality control (autophagy), apoptosis^{2, 3} and mitophagy⁴, maintains nuclear envelope integrity⁵ and sarcomere stability⁶, plays a key role in electrical impulse generation and propagation⁷, and facilitates adrenergic enhancement of excitation-contraction coupling⁸.

Figure 1. Consensus protein-protein binding domains in BAG3.

BAG3 serves as a co-chaperone for both the constituitively expressed and the inducible heat shock protein 70 (Hsc70/Hsp70): its BAG domain binding with the ATPase binding site of Hsc70/Hsp70. Hsp70-BAG3 interactions mediate chaperone assisted autophagy by shuttling client proteins along the cellular micro-tubular apparatus that are dedicated to eliminating mis-folded proteins³³. The BAG domain also couples with the anti-apoptotic protein Bcl-2 resulting in inhibition of apoptosis³². Two highly conserved IPV (Ile-Pro-Val) motifs allow BAG3 to interact with the small heat shock proteins HspB8 and HspB6 to support macroautophagy, a process in which mis-folded or damaged proteins and organelles (e.g. mitochondria) are sequestered in autophagosomes and degraded³⁹ whereas the second IPV domain binds to αβ-crystallin with the subsequent inhibition of protein aggregation⁴⁰. The WW domain plays a pivotal role in chaperone-assisted selective autophagy or CASA, a tension-induced autophagy pathway that is essential for mechano-transduction in muscle and immune cells^{34, 35}. The WW region also couples with the PXXP region to modify the three-dimensional structure of the protein. The proline-rich PXXP region also facilitates binding to the SH3 (Src homology 3)-containing protein phospholipase C-γ (PLC-γ) with subsequent stimulation of invasion, adhesion and migration of cancer cells^{36, 37} and the retrograde transport of mis-folded proteins to peri-nuclear aggresomes³⁸.

BAG3 protein levels are reduced in swine hearts 12 weeks post-myocardial infarction (MI)¹, in mouse hearts 18 weeks after transverse aortic constriction¹ or after ischemia-reperfusion (I/R)³ or 11 weeks post-MI⁹, and in human hearts with severe left ventricular (LV) dysfunction secondary to idiopathic dilated cardiomyopathy (IDC) or ischemic heart disease^{10, 11}. In addition, lowering BAG3 in wild-type (WT) mouse myocytes by shRNA-BAG3 (~54% reduction)⁸ or by engineering cardiac-specific BAG3 haplo-insufficient (cBAG3^+/−) mouse (~51% reduction)² resulted in a significant decrease in myocyte contraction amplitude (by 15.3% and 10.1%, respectively) in the presence of isoproterenol. When compared to WT littermates, cBAG3^+/− mice maintained a normal ejection fraction (EF) at 8 weeks but suffered ~29% decline in EF by 10 weeks of age². By contrast, repletion of the depleted BAG3 by virus-mediated gene transfer in mice post-MI⁹ or post-I/R³ dramatically improved the globally depressed myocardial function. These observations in animal models of heart failure (HF), human HF and molecular approaches to manipulate BAG3 levels are all consistent with the hypothesis that reduction in BAG3 levels plays an important pathogenetic role in contractile abnormalities in HF, and that restoring BAG3 levels in failing hearts improves contractile function. Whether the protein-protein binding domains of BAG3 are involved in the regulation of contractility is unknown.

Recently, by analyzing genomic DNA from 509 African American patients with dilated cardiomyopathy enrolled in 3 independent US studies [Genetic Risk Assessment of African Americans with Heart Failure (GRAHF), Intervention in Myocarditis and Acute Cardiomyopathy Trial-2 (IMAC-2), and Genetic Risk Assessment of Cardiac Events (GRACE)], we identified 3 novel non-synonymous single-nucleotide BAG3 variants (P380S, A479V or P63A) in African Americans (but not in patients of European ancestry) with HF¹¹. The presence of any 1 of these BAG3 variants was associated with a worse outcome, as reflected by the combined outcome variable of HF hospitalization, heart transplant or death¹¹. Two important observations are that the BAG3 variants were found on one allele and patients who harbored the P63A variant also carried the P380S variant. Since P63A is near the WW domain (residues 21–55), P380S is in the PXXP domain (residues 302–412) and A479V is in the BAG domain (residues 420–499)(Fig. 1), we took advantage of our cBAG3^+/− mouse (to simulate WT BAG3 on one allele) and expressed either WT BAG3 or the BAG3 variants (P380S, P479V or P63A/P380S) to test if the BAG3 variants behaved as loss-of-function mutants: the readout being reduced β-adrenergic enhancement of EC coupling. We also expressed BAG3 mutants in which either the WW or the BAG domain was deleted to corroborate our findings with non-synonymous single-nucleotide BAG3 variants.

Methods

Generation of cBAG^+/− mouse and animal care.

Cardiac-specific haplo-insufficient BAG3 mice were generated as described previously². Adult cBAG3^+/− mice used in this study were 8 weeks old at which time they had normal EF when compared to their WT littermates². Mice were housed and fed on a 12:12h light-dark cycle at the Temple University Animal Facility supervised by full-time veterinarian staff members. Standard care was provided to all mice used for experiments.

Isolation, adenoviral infection, and culture of adult murine cardiac myocytes.

Cardiac myocytes were isolated from the septum and LV free wall of WT and cBAG3^+/− mice (8 wks old) according to the protocol of Zhou et al.¹², and plated on laminin-coated glass coverslips¹³. Myocytes were used either on the same day (within 2–6 hr after isolation) or after 24 hr of culture.

To express exogenous genes via adenovirus (Adv)-mediated gene transfer, two hours after isolation, myocytes were infected with replication-deficient Adv expressing either green fluorescent protein (GFP)(Adv-GFP, 6.0 × 10⁶ pfu/ml), WT BAG3 (Adv-BAG3, 4.0 × 10⁶ pfu/ml), P380S (Adv-380, 4.0 × 10⁶ pfu/ml), A479V (Adv-479, 4.2 × 10⁶ pfu/ml), P63A/P380S (Adv-63/380, 4.8 × 10⁶ pfu/ml), WW domain deletion (residues 62–575; abbreviated as Adv-575, 3.75 × 10⁶ pfu/ml) or BAG domain deletion (residues 1–420; abbreviated as Adv-420, 3.7 × 10⁶ pfu/ml) mutants in 1 ml of fetal bovine serum-free Eagle minimal essential medium (MEM) containing 0.2% bovine serum albumin (BSA), creatine (5 mM), carnitine (2 mM), taurine (5 mM), NaHCO3 (4.2 mM), penicillin (30 mg/L), gentamicin (4 mg/L), insulin-transferrin-selenium supplement, and 2,3-butanedione monoxime (BDM, 10 mM) for 3 hours. An additional 1 ml of MEM (with same supplements) was then added, and myocytes were cultured for 24h before experiments^{14, 15}. Before measurements of myocyte function, cells were bathed with MEM without BDM and returned to the incubator (37°C) for 30 min. We have previously demonstrated that under our culture conditions, adult mouse LV myocytes cultured for up to 48h maintained rod-shape morphology, transverse-tubule organization, and normal contractile function; and that infection with Adv-GFP had no effect on single myocyte contractility when compared to uninfected myocytes¹⁴.

Myocyte shortening measurements.

LV myocytes (freshly isolated or cultured) adherent to laminin-coated glass coverslips were bathed in 0.6 ml of air- and temperature equilibrated (37°C), HEPES-buffered (20 mM, pH 7.4) medium 199 containing 1.8 mM extracellular Ca²⁺ concentration ([Ca²⁺]_o). Myocytes were electrically paced to contract at 2 Hz and myocyte motion was captured by charge-coupled device video camera (Ionoptix; Milton, MA). Isoproterenol (Iso), when indicated, was present at 1 μM. Myocyte contraction dynamics was analyzed off-line with edge-detection algorithm as previously described^13–15.

Measurement of intracellular Ca²⁺ concentration ([Ca²⁺]_i) in cardiac myocytes.

Fura-2 loaded (0.67 μM fura-2 AM, Molecular Probes, Eugene, OR; 15 min, 37°C) LV myocytes attached to laminin-coated coverslips were incubated in medium 199 containing 1.8 mM [Ca²⁺]_o and field-stimulated to contract (2 Hz, 37°C). Myocytes were exposed to excitation light (360 and 380 nm) only during data acquisition. Epifluorescence (510 nm) was measured in steady-state twitches both before after addition of Iso (1 μM). Daily calibration of fura-2 signals and [Ca²⁺]_i analyses were performed as previously described^{13, 14, 16–19}.

L-type Ca²⁺ current (I_Ca) measurements.

Whole cell patch-clamp recordings were performed at 30°C as described previously^{13, 20}. Pipette diameter was 4 – 6 μm and pipette resistance was 0.8 to 1.4 MΩ when filled with standard internal solution. For I_Ca measurements, pipette solution contained (in mM): CsCl 110, TEA.Cl 20, HEPES 10, MgATP 5, and EGTA 10; pH 7.2. Extracellular bathing solution contained (in mM): N-methyl-D-glucamine 137, CsCl 5.4, CaCl₂ 2, MgSO₄ 1.3, HEPES 20, 4-aminopyridine 4, and glucose 15; pH 7.4. Holding potential was at −90 mV²¹. To ensure steady-state sarcoplasmic reticulum (SR) Ca²⁺ loading, 6 conditioning pulses (from −70 to 0 mV, 100 ms, 2 Hz) were delivered to the myocyte before the arrival of each test pulse (from −90 to +50 mV, 10 mV increments, 60 ms)²¹. Leak-subtracted inward currents were used in analysis for I_Ca amplitudes. Inward currents obtained under these conditions were blocked by 1 μM verapamil^{8, 21}. I_Ca was normalized to membrane capacitance (C_m) before comparison between WT and cBAG3^+/− myocytes, and among Adv-GFP, Adv-BAG3, Adv-420 and Adv-575 myocytes.

Statistics.

All results are expressed as means ± SE. For analysis of contraction amplitudes as a function of mouse (WT vs. cBAG3^+/−), single nucleotide BAG3 variant and Iso, three-way ANOVA was used. For analysis of [Ca²⁺]_i transient, contraction and maximal I_Ca (at 0 mV) amplitudes as a function of BAG3 deletion mutant and Iso or as a function of mouse (WT vs. cBAG3^+/−) and Iso, two-way ANOVA was used. A commercially available software package (JMP version 12; SAS Institute, Cary, NC) was used. In all primary analyses, p≤0.05 was taken to be statistically significant. In subgroup analyses, the Bonferroni adjustment (statistical significance at p≤0.05/n, where n=number of subgroup comparisons) was applied.

Results

[Ca²⁺]_i transient, contraction and maximal I_Ca amplitudes are lower in cBAG3^+/− myocytes after isoproterenol.

We have previously demonstrated that shRNA-mediated reduction in BAG3 (by ~54%) resulted in lower [Ca²⁺]_i transient, contraction and I_Ca amplitudes after Iso stimulation⁸. Similarly, in cBAG3^+/− myocytes in which BAG3 levels are reduced by ~51%², baseline [Ca²⁺]_i transient and contraction amplitudes were similar but were significantly lower after Iso (Fig. 2; for [Ca²⁺]_i transient: p=0.0077, group × Iso interaction effect; for contraction: p=0.0003, group × Iso interaction effect). Reduced maximal [Ca²⁺]_i transient amplitude in cBAG3^+/− myocytes was due to lower systolic [Ca²⁺]_i (p=0.0028, group × Iso interaction effect) since diastolic [Ca²⁺]_i (p=0.9728, group × Iso interaction effect) were similar (Fig. 2G). The t_1/2 of [Ca²⁺]_i transient decline, a measure of SR Ca²⁺ uptake^{22, 23}, was significantly (p=0.0425; group effect) prolonged in cBAG3^+/− when compared to WT myocytes (Fig. 2H), indicating that SR Ca²⁺-ATPase activity was lower in cBAG3^+/− myocytes. Lower systolic [Ca²⁺]_i and [Ca²⁺]_i transient amplitudes after Iso in cBAG3^+/− myocytes were largely due to reduced I_Ca amplitudes (Fig. 3; p=0.0021, group × Iso interaction effect).

Figure 2. Abnormal Ca²⁺ homeostasis and reduced contraction in cardiac-specific haplo-insufficient BAG3 murine myocytes.

[Ca²⁺]_i and contraction were measured in field-stimulated (2 Hz, 37°C) LV myocytes isolated from 8-wk old WT and cBAG3^+/− mice (Methods). Representative traces of contraction at baseline (A) and after 1 μM isoproterenol (Iso)(B), and [Ca²⁺]_i transients at baseline (C) and after Iso (D) of WT (solid lines) and cBAG3^+/− (dashed lines) myocytes are shown. E, G and H: Means ± SE of [Ca²⁺]_i transient amplitudes (E), systolic and diastolic [Ca²⁺]_i (G) and t_1/2 of [Ca²⁺]_i transient decline (H) from 11 WT and 13 cBAG3^+/− myocytes (2 mice each) at baseline and 15 WT and 19 cBAG3^+/− myocytes (2 mice each) after 1 μM Iso. For systolic and diastolic [Ca²⁺]_i, error bars are not shown if they fall within the boundaries of the symbols. F: Means ± SE of contraction amplitudes from 21 WT and 16 cBAG3^+/− myocytes (3 WT and 2 cBAG3^+/− mice) at baseline and 19 WT and 12 cBAG3^+/− myocytes after Iso. For E, F, G: *p<0.008, group (WT vs. cBAG3^+/−) × Iso interaction effect. For H: *p=0.0425, WT vs. cBAG3^+/−.

Figure 3. Reduced L-type Ca²⁺ current (I_Ca) in cardiac-specific haplo-insufficient BAG3 myocytes.

I_Ca was measured in myocytes isolated from 8-wk old WT and cBAG3^+/− myocytes (Methods). A and B: Representative traces of I_Ca (0 mV) from WT and cBAG3^+/− myocytes before (A) and after (B) 1 μM Iso. C: IV relationship of I_Ca in WT (▄, baseline; □, after Iso; n=7 cells from 3 mice) and cBAG3^+/− (●, baseline; ○, after Iso; n=5 cells from 3 mice) myocytes. 3-way ANOVA indicates significant group (WT vs. cBAG3^+/−, p<0.0001), voltage (p<0.0001), Iso (p<0.0001), group × voltage (p=0.0425) and group × Iso (p=0.0021) interaction effects.

WT BAG3, but not P380S, A479V or P63A/P380S BAG3 mutant restores maximal contraction amplitude in cBAG3^+/− myocytes.

To simulate the clinical observation that the 3 non-synonymous single-nucleotide BAG3 variants detected in African Americans with heart failure were heterozygous, we expressed GFP, WT BAG3 or its variants in either cBAG3^+/− or WT myocytes to test if myocyte excitation-contraction in response to β-adrenergic stimulation was affected differently. In cBAG3^+/− myocytes expressing GFP, contraction amplitudes were similar to those observed in WT-GFP myocytes under basal conditions (Fig. 4A). After Iso, contraction amplitude in cBAG3^+/−-GFP myocytes was significantly lower than that measured in WT-GFP myocytes (Fig. 4B; p=0.0090, group × Iso interaction effect). The observation that contraction amplitudes were similar under basal conditions but lower in cBAG3^+/−-GFP myocytes after Iso closely mimicked those measured in uninfected WT and cBAG3^+/− myocytes (Fig. 2), indicating adenovirus infection followed by short-term culture did not affect the intrinsic differences between WT and cBAG3^+/− myocytes.

Figure 4. Non-synonymous, single nucleotide BAG3 variants in African-American patients with heart failure do not enhance β-adrenergic responsiveness compared to WT BAG3.

LV myocytes isolated from 8-wk old WT (open bars) and cBAG3^+/− (hatched bars) mice were infected with adenovirus (Adv) expressing GFP, WT BAG3, P380S, A479V or P63A/P380S BAG3 variants and cultured for 24h before contraction (2 Hz, 37°C) measurements, both before and after 1 μM Iso (Methods). A. Means ± SE of baseline contraction amplitudes [% resting cell length (RCL)] measured in WT and cBAG3^+/− myocytes. B. Means ± SE of contraction amplitudes after Iso. Note the difference in Y-axis scales. For WT-GFP, there are 20 myocytes at baseline and 15 myocytes after Iso (n=6 mice); for cBAG3^+/−-GFP, there are 19 myocytes at baseline and 17 myocytes after Iso (n=6 mice); for WT-BAG3, there are 14 myocytes at baseline and 14 myocytes after Iso (n=5 mice); for cBAG3^+/−-BAG3, there are 16 myocytes at baseline and 21 myocytes after Iso (n=5 mice); for WT-P380S, there are 7 myocytes at baseline and 8 myocytes after Iso (n=3 mice); for cBAG3^+/−-P380S, there are 8 myocytes at baseline and 8 myocytes after Iso (n=2 mice); for WT-A479V, there are 17 myocytes at baseline and 22 myocytes after Iso (n=3 mice); for cBAG3^+/−-A479V, there are 18 myocytes at baseline and 19 myocytes after Iso (n=3 mice); for WT-P63A/P380S, there are 7 myocytes at baseline and 7 myocytes after Iso (n= 1 mouse); and for cBAG3^+/−-P63A/P380S, there are 5 myocytes at baseline and 6 myocytes after Iso (n=1 mouse). 3-way ANOVA indicate significant group (WT vs. cBAG3^+/−; p=0.0002), Adv (p<0.0001), Iso (p<0.0001) and group × Iso interaction (p=0.0003) effects. *p<0.002, BAG3 vs. GFP, P380S, A479V or P63A/P380S by subgroup analysis. #p=0.0003, group (WT vs. cBAG3^+/−) × Iso interaction effects.

To analyze contraction amplitude data (Fig. 4) while taking into account the effects of WT and cBAG^+/− mice (group), adenovirus expressing GFP, BAG3 or its variants (Adv), and isoproterenol (Iso), we used 3-way ANOVA. The primary analysis on 20 sets of data (WT-GFP ± Iso, cBAG3^+/−-GFP ± Iso, etc.) indicates significant group (p=0.0002), Adv (p<0.0001), Iso (p<0.0001) and group × Iso interaction (p=0.0003) effects; indicating contraction amplitudes were lower in cBAG3^+/− myocytes in the presence of Iso, and that expressing GFP, BAG3 or its variants had different effects on contraction amplitudes. To further dissect whether the effects of expressing WT BAG3 on contraction amplitude were different from GFP or BAG3 variants, we performed subgroup analysis (BAG3 vs. GFP, BAG3 vs. P380S, BAG3 vs. P63A/P380S, and BAG3 vs. A479V) with p≤0.0125 taken to be statistically significant (Bonferroni adjustment). The analysis indicates that the effects of expressing WT BAG3 on contraction are significantly different than those of expressing GFP (p<0.0001), P380S (p<0.0001), P63A/P380S (p=0.0012) or A479V (p=0.0007). Stated differently, the 3 BAG3 variants tested behaved like loss-of-function mutants in terms of enhancing excitation-contraction in the presence of isoproterenol.

BAG but not WW domain is required to endow β-adrenergic responsiveness in adult mouse LV myocytes.

The results from non-synonymous single-nucleotide BAG3 variants implicate WW (P63A), PXXP (P380S) and BAG (A479V) domains as important in regulation of excitation-contraction by BAG3. To test this hypothesis, we engineered 2 BAG3 mutants in which either the WW (Adv-575) or BAG (Adv-420) domain was deleted and expressed them in WT myocytes. Since there are 5 subgroup comparisons (GFP vs. BAG3, GFP vs. 420, GFP vs. 575, BAG3 vs. 420 and BAG3 vs. 575), p≤0.01 was taken to be statistically significant. Compared to WT-GFP myocytes, WT-BAG3 myocytes contracted more vigorously in the presence of Iso (Fig. 5; p=0.0022, Adv × Iso interaction effect). Enhanced contraction observed in Iso-treated WT-BAG3 myocytes as compared to WT-GFP myocytes was due to increased systolic [Ca²⁺]_i (Fig. 6; p=0.0052, Adv × Iso interaction effect), resulting in larger [Ca²⁺]_i transient amplitudes (Fig. 6; p=0.0009, Adv × Iso interaction effect). Augmented [Ca²⁺]_i transient amplitudes in Iso-stimulated WT-BAG3 myocytes were primarily due to larger I_Ca (Fig. 7; p=0.01, Adv × Iso interaction effect), resulting in increased Ca²⁺ influx during the action potential and higher systolic [Ca²⁺]_i. Compared to WT-BAG3 myocytes, deletion of the BAG (Adv-420) domain abolished the enhanced β-adrenergic response in terms of contraction (Fig. 5; p=0.0008, Adv × Iso interaction effect), [Ca²⁺]_i transient (Fig. 6; p=0.0071, Adv × Iso interaction effect) and I_Ca (Fig. 7; p=0.0013, Adv × Iso interaction effect) amplitudes; indicating the BAG domain is intimately involved in Ca²⁺ homeostasis in adult cardiac myocytes. By contrast, deletion of the WW domain (Adv-575) did not negatively impact enhanced β-adrenergic responsiveness in terms of contraction (Fig. 5; p=0.9022, Adv × Iso interaction effect), [Ca²⁺]_i transient (Fig. 6; p=0.2907, Adv × Iso interaction effect) and I_Ca (Fig. 7; p=0.0321, Adv × Iso interaction effect) amplitudes.

Figure 5. BAG3 but not WW domain is required for enhanced contraction amplitudes after isoproterenol.

LV myocytes isolated from 8 wk-old WT mice were infected with adenovirus (Adv) expressing GFP, WT BAG3, WW domain (residues 62 to 575 and abbreviated as 575) or BAG domain (residues 1 to 420 and abbreviated as 420) deletion mutants and cultured for 24h before contraction (2 Hz, 37°C) measurements (Methods). A. Representative traces of contraction for GFP, BAG3, 420 and 575 myocytes, both before (dotted lines) and after (solid lines) 1 μM isoproterenol (Iso). B. Means ± SE of contraction amplitudes [% resting cell length (RCL)] before (open bars) and after Iso (solid bars). For GFP, there are 11 myocytes at baseline and 13 myocytes after Iso (n=4 mice); for BAG3, there are 14 myocytes at baseline and 12 myocytes after Iso (n=5 mice); for 420, there are 10 myocytes at baseline and 14 myocytes after Iso (n=2 mice); and for 575, there are 9 myocytes at baseline and 12 myocytes after Iso (n=3 mice). 2-way ANOVA of all 4 groups indicate significant Adv (p=0.0021), Iso (p<0.0001) and Adv × Iso (p=0.0011) interaction effects. *p<0.02, GFP vs. BAG3 or GFP vs. 575 by subgroup analysis (p≤0.01 is statistically significant after Bonferroni adjustment); #p<0.001, BAG3 vs. 420 by subgroup analysis.

Figure 6. BAG3 but not WW domain is required for higher systolic [Ca²⁺]_i and larger [Ca²⁺]_i transient amplitudes after isoproterenol.

LV myocytes isolated from 8 wk-old WT mice were infected with adenovirus (Adv) expressing GFP, WT BAG3, WW domain (575) or BAG domain (420) deletion mutants and cultured for 24h before [Ca²⁺]_i (2 Hz, 37°C) measurements (Methods). A. Representative traces of [Ca²⁺]_i transients for GFP, BAG3, 420 and 575 myocytes, both at baseline (dotted lines) and after 1 μM Iso (solid lines). B. Means ± SE of systolic (●) and diastolic (○) [Ca²⁺]_i at baseline and after Iso (I). Error bars are not shown if they fall within the boundaries of the symbols. For GFP, there are 13 myocytes at baseline and 10 myocytes after Iso (n=2 mice); for BAG3, there are 9 myocytes at baseline and 14 myocytes after Iso (n=3 mice); for 420, there are 9 myocytes at baseline and 9 myocytes after Iso (n=2 mice); and for 575, there are 12 myocytes at baseline and 13 myocytes after Iso (n=3 mice). 2-way ANOVA of systolic [Ca²⁺]_i from all 4 groups indicate significant Adv (p=0.0001), Iso (p<0.0001) and Adv × Iso (p=0.0006) interaction effects. *p≤0.005, GFP vs. BAG3 or GFP vs. 575 by subgroup analysis (p≤0.01 is statistically significant after Bonferroni adjustment). C. Means ± SE of [Ca²⁺]_i transient amplitudes (% increase in Fura2 signal) at baseline (open bars) and after Iso (solid bars). 2-way ANOVA of all 4 groups indicate significant Adv (p=0.0002), Iso (p<0.0001) and Adv × Iso (p=0.0003) interaction effects. *p≤0.0009, GFP vs. BAG3 or GFP vs. 575 by subgroup analysis (p≤0.01 is statistically significant after Bonferroni adjustment).

Figure 7. BAG3 but not WW domain is required for enhanced L-type Ca²⁺ current (I_Ca) after isoproterenol.

LV myocytes isolated from 8 wk-old WT mice were infected with adenovirus (Adv) expressing GFP, WT BAG3, WW domain (575) or BAG domain (420) deletion mutants and cultured for 24h before I_Ca measurements (Methods). A. Representative I_Ca traces at 0 mV for GFP, BAG3, 420 and 575 myocytes, both at baseline (black) and after 1 μM Iso (red). B. Means ± SE of maximal I_Ca amplitudes (0 mV) before (open bars) and after Iso (hatched bars). For GFP, there are 7 myocytes each at baseline and after Iso (n=5 mice); for BAG3, there are 4 myocytes each at baseline and after Iso (n=3 mice); for 420, there are 6 myocytes each at baseline and after Iso (n=3 mice); and for 575, there are 6 myocytes each at baseline and after Iso (n=3 mice). 2-way ANOVA of all 4 groups indicate significant Adv (p=0.0158), Iso (p<0.0001) and Adv × Iso (p=0.0078) interaction effects. *p<0.01, BAG3 vs. GFP or BAG3 vs. 420 by subgroup analysis (p≤0.01 is statistically significant after Bonferroni adjustment).

Discussion

Recent studies have demonstrated the increasing importance of BAG3 in maintaining cardiac health and function. Specifically, in the heart BAG3 regulates autophagy, apoptosis^{2, 3}, mitophagy⁴ and [Ca²⁺]_i homeostasis⁸, maintains sarcomere stability⁶ and nuclear membrane integrity⁵, and facilitates electrical signal propagation⁷. The pathophysiological significance of BAG3 in the heart is highlighted by the observation that patients with severe cardiac failure from dilated cardiomyopathy (both ischemic and non-ischemic causes) have marked BAG3 depletion 10, 11. Multiple animal models of heart failure corroborate the clinical observation that BAG3 levels were severely reduced^{1, 3, 9}. Mechanistically, downregulation of BAG3 in cardiac myocytes either by shRNA^{3, 7, 8} or by heterozygous gene deletion² resulted in decreased autophagy flux, increased apoptosis, reduced myocyte L-type Ca²⁺ current, [Ca²⁺]_i transient and contraction amplitudes, prolonged action potential duration, decreased electrical impulse propagation, and global LV dysfunction. The clinical observations and experimental results support the exciting hypothesis that BAG3 is a novel and rational target for therapeutic intervention in patients with heart failure.

In 2009, Selcen was the first to describe 3 children harboring a non-synonymous single nucleotide BAG3 variant (P209L) with severe myofibrillar degeneration, skeletal muscle weakness, giant axon disease and a hypertrophic/restrictive cardiomyopathy²⁴. Subsequently, cardiomyopathies due to BAG3 deletion, truncation and other non-synonymous single nucleotide polymorphism (SNP) mutants (which can lead to BAG3 haplo-insufficiency) have been reported^{10,11,25–30}. Indeed, BAG3 has been proposed to be a major cardiomyopathy locus²⁷ and may be efficacious in gene rescue in hereditary BAG3 cardiomyopathy.

The first major finding of our current study is that BAG3 haplo-insufficiency resulted in decreased β-adrenergic responsiveness in terms of lower systolic [Ca²⁺]_i due to reduced I_Ca in isolated cardiac myocytes, leading to decreased myocyte contraction amplitudes. This observation mirrors the observations in myocytes in which BAG3 was reduced by shRNA⁸. We have previously demonstrated that in BAG3-deficient cardiac myocytes, addition of either forskolin (10 μM; directly activates adenylyl cyclase) or dibutyrl-cAMP (5 mM; directly activates protein kinase A) restores maximal I_Ca amplitudes to those measured in BAG3-replete myocytes⁸. These observations suggest that the defects in β-adrenergic signaling in BAG3-deficient myocytes occur proximal to adenylyl cyclase. In addition, we demonstrated that BAG3, β1-adrenergic receptor (β1AR), α1c-subunit of the L-type Ca²⁺ channel (Ca_v1.2) and phospholemman (PLM) co-immunoprecipitate together as a macromolecular complex.⁸ We speculated that there is a specific pool of β1AR that complexes with BAG3 and Ca_v1.2 and facilitates β1-adrenergic signaling to increase I_Ca. In this light, recent elegant studies by Marx’s group indicate that the protein-protein interaction between the β subunit of Ca_v1.2 and the α-interaction domain (AID) of the α1c-subunit of Ca_v1.2, while not affecting basal I_Ca, is absolutely required for the β-adrenergic modulation of I_Ca³¹. Therefore mechanistically, when BAG3 levels are low (either genetically or in heart failure), there is less β-subunit interaction with the α1c-subunit of Ca_v1.2, leading to normal baseline I_Ca but reduced β-adrenergic induced I_Ca increase. An alternative hypothesis is that since PLM regulates I_Ca in cardiac myocytes²¹, and since PLM, BAG3, β1AR and Cav1.2 form a macromolecular complex⁸, reduced BAG3 levels may alter interaction of PLM with Ca_v1.2, thereby affecting I_Ca amplitudes.

Our most recent contribution to the role of BAG3 in dilated cardiomyopathy is the identification of 3 novel non-synonymous BAG3 SNPs in African American patients with heart failure but not found in patients of European ancestry¹¹. Thus the second major finding of our current study is that these novel BAG3 SNPs (P380S, A479V and P63A/P380S) behave like loss-of-function mutants in terms of enhancing myocyte contraction amplitudes after β-adrenergic stimulation. In HF patients in which circulating catecholamine levels are elevated, the inability to respond properly to β-adrenergic signals in African American HF patients harboring these BAG3 SNPs may partly account for the LV contractile dysfunction.. An important inference from these loss-of-function BAG3 SNPs is that the WW (P63A), PXXP (P380S) and BAG (A479V) domains may be involved in mediating β-adrenergic response in terms of augmenting I_Ca.

When viewed purely in the context of Ca²⁺ homeostasis, reduced β-adrenergic responsiveness in myocytes that are either BAG3-deficient or harboring one of these BAG3 variants may very well represent a protective mechanism to minimize Ca²⁺ overload under adrenergic stress. Indeed, this is one of the mechanisms by which β-adrenergic blockers exert their beneficial effects in HF patients. In addition, other than modulating excitation-contraction coupling, BAG3 also regulates autophagy, apoptosis^{2, 3}, mitophagy⁴, maintains sarcomere stability⁶ and nuclear membrane integrity⁵ and facilitates electrical signal propagation⁷, the worse clinical outcomes observed in African American patients harboring these BAG3 variants may relate to a combination of these pleotropic effects of BAG3¹¹.

The third major finding is that the BAG but not WW domain is absolutely required for the β-adrenergic receptor induced increase in I_Ca, systolic [Ca²⁺]_i and myocyte contraction amplitudes. The BAG domain is known to inhibit apoptosis by binding to Bcl-2³² and mediate autophagy³³(Fig. 1). The WW domain is involved in chaperone-assisted selective autophagy or CASA^{34, 35} and interacts with the PXXP region (Fig. 1). The PXXP domain binds to phospholipase C-γ and promotes metastasis of cancer cells^{36, 37} and the retrograde transport of mis-folded proteins to peri-nuclear aggresomes³⁸. We speculate that the PXXP (P380S) and the BAG (A479V and deletion mutant) domains may be involved in the interaction of BAG3 to the L-type Ca²⁺ channel to implement β-adrenergic increases in I_Ca.

In summary, low BAG3 levels observed in acquired and some forms of hereditary heart failure results in reduced L-type Ca²⁺ current, lower systolic [Ca²⁺]_i and decreased contraction amplitudes in cardiac myocytes under adrenergic stress, thereby further aggravating left ventricular dysfunction. In addition, the three novel non-synonymous single nucleotide BAG3 variants detected in African American patients with heart failure behave like loss-of-function mutants in terms of augmentation of myocyte contraction after β-adrenergic stimulation. Finally, the PXXP and BAG but not WW domains of BAG3 are required to implement β-adrenergic increases in L-type Ca²⁺ current, systolic [Ca²⁺]_i and contraction amplitudes. We speculate that BAG3 may be an exciting new target for therapeutic intervention for heart failure.