BAG3 expression and sarcomere localization in the human heart are linked to HSF-1 and are differentially affected by sex and disease

Abstract

Mutations to the sarcomere-localized cochaperone protein Bcl2-associated athanogene 3 (BAG3) are associated with dilated cardiomyopathy (DCM) and display greater penetrance in male patients. Decreased protein expression of BAG3 is also associated with nongenetic heart failure; however, the factors regulating cardiac BAG3 expression are unknown. Using left ventricular (LV) tissue from nonfailing and DCM human samples, we found that whole LV BAG3 expression was not significantly impacted by DCM or sex; however, myofilament localized BAG3 was significantly decreased in males with DCM. Females with DCM displayed no changes in BAG3 compared with nonfailing. This sex difference appears to be estrogen independent, as estrogen treatment in ovariectomized female rats had no impact on BAG3 expression. BAG3 gene expression in noncardiac cells is primarily regulated by the heat shock transcription factor-1 (HSF-1). We show whole LV HSF-1 expression and nuclear localized/active HSF-1 each displayed a striking positive correlation with whole LV BAG3 expression. We further found that HSF-1 localizes to the sarcomere Z-disc in cardiomyocytes and that this myofilament-associated HSF-1 pool decreases in heart failure. The decrease of HSF-1 was more pronounced in male patients and tightly correlated with myofilament BAG3 expression. Together our findings indicate that cardiac BAG3 expression and myofilament localization are differentially impacted by sex and disease and are linked to HSF-1.

NEW & NOTEWORTHY Myofilament BAG3 expression decreases in male patients with nonischemic DCM but is preserved in female patients with DCM. BAG3 expression in the human heart is tightly linked to HSF-1 expression and nuclear translocation. HSF-1 localizes to the sarcomere Z-disc in the human heart. HSF-1 expression in the myofilament fraction decreases in male patients with DCM and positively correlates with myofilament BAG3.

INTRODUCTION

Bcl-2-associated athanogene-3 (BAG3) is a multifunctional cochaperone protein expressed at varying levels in all human tissues. Through various protein interactions BAG3 mediates numerous cellular processes, including cell survival, protein quality control, cytoskeleton maintenance, and development (1). Notably, BAG3 protein expression is highest in cancer cells, where it enhances tumor pathogenesis by inhibiting apoptosis/activating autophagy (2), and in striated muscle, where it mediates turnover of sarcomere proteins via selective macroautophagy (3–5). In cardiac muscle, BAG3 mutations and decreased BAG3 protein expression are associated with increased apoptosis and sarcomere structural disarray (6, 7). Understandably, both inhibiting and inducing BAG3 expression/activity, in cancer and myopathy respectively, have been proposed as therapeutic strategies for treating human diseases (1, 8–11). However, although the factors that regulate BAG3 protein expression in cancer cells are well established, relatively little is known regarding the regulation of BAG3 expression in cardiomyocytes.

BAG3 mutations, which commonly cause dilated cardiomyopathy (DCM), are among the leading genetic causes of heart failure (12) and have higher penetrance in male patients (13). Preclinical models of heart failure provide evidence that altered BAG3 protein expression is also involved in the progression of heart failure and show BAG3 levels decrease in heart failure secondary to pressure overload and myocardial infarction (8, 14, 15). Additionally, diminished BAG3 expression was found in LV tissue from patients with end-stage heart failure compared with nonfailing donor hearts (16), and we recently showed the myofilament-specific pool of BAG3 also decreases in DCM (4). However, these studies were not powered to examine potential sex differences in cardiac BAG3 expression and myofilament localization, such as have been observed with BAG3 mutation penetrance. That the disease pathogenesis of DCM differs between male and female patients is well established (17), and earlier studies in cancer cells suggest that biological sex impacts BAG3 expression (18, 19). An assessment of the impact of sex on BAG3 expression in the heart is thus warranted.

The primary cellular factor regulating BAG3 expression in cancer cells is the heat shock transcription factor-1 (HSF-1) (20). HSF-1 regulates the gene expression of a myriad of chaperone and cochaperone proteins, which are upregulated in response to elevated cell stress conditions (21, 22). A previous study of HSF-1 in the human heart found that HSF-1 activity decreases substantially in the end-stage of heart failure (23), the effects of which likely contribute to the general proteotoxicity observed in heart disease (24). Interestingly, HSF-1 deficiency in mice accelerated the transition from pressure-overload-induced cardiac hypertrophy to heart failure (25), further delineating the importance of the HSF-1-mediated stress response for heart function. Given the similar effects of BAG3 and HSF-1 deficiency on cardiac function, it is plausible that the observed phenotype with HSF-1 deficiency is due at least in part to reduced HSF-1-mediated BAG3 expression. However, the relationship between HSF-1 and BAG3 in cardiomyocytes has not been examined.

In this study, we demonstrate that the previously observed decrease in BAG3 expression in DCM is more exaggerated for the myofilament-specific fraction compared to the whole LV. We further show that the decrease in BAG3 associated with DCM is specific to males, as expression in female patients was unchanged compared to nonfailing. We show this sex difference in expression cannot be explained simply by differences in estrogen signaling. We further demonstrate that cardiac BAG3 expression is tightly linked with HSF-1 expression and nuclear localization, suggesting the regulation of cardiac BAG3 expression by HSF-1 is conserved from other cell types. Finally, we show that HSF-1 localizes to the sarcomere Z-disk and that this myofilament pool decreases in males with DCM and positively correlates with myofilament BAG3. Collectively, our findings support that the decrease in myofilament BAG3 observed in DCM is specific to disease pathogenesis in male patients and suggest that altered HSF-1 expression and localization may be the root cause.

MATERIALS AND METHODS

Human Heart Tissue Procurement

Human LV samples were obtained from the Cleveland Clinic and Loyola University Chicago Cardiovascular Research Institute biorepositories. The nonfailing (NF) samples were obtained from donor hearts without history of coronary artery disease and with normal left ventricular function (n = 19, 52.6% female; age = 56.1 ± 11.6; LV ejection fraction = 62.3 ± 8.5) that were deemed unsuitable for transplant due to blood type, age, or size incompatibility. The failing heart LV tissue was obtained from explanted failing hearts of patients with nonischemic dilated cardiomyopathy (DCM; n = 25, 32.0% female; age = 52.2 ± 12.1; LV ejection fraction = 22.1 ± 13.4). Tissue samples were flash frozen in liquid nitrogen and stored at −80°C. All patients provided written informed consent before tissue procurement, and patient identifiers associated with the tissue were removed at the time of biobanking.

Ethics Statement

All experimentation involving human tissue samples was performed with adherence to the ethical and responsible conduct of research standards established by institutional and national guidelines in accordance with the Helsinki Declaration of 1975. Myocardial tissue used for these experiments was obtained from the tissue biorepositories at the Cleveland Clinic and within the Cardiovascular Research Institute at Loyola University Chicago Health Sciences Division. The patients provided written informed consent before tissue collection, which was procured with permission from the institutional review boards of Loyola University Chicago and the Cleveland Clinic.

Ovariectomy Rat Model

The experimental procedures used in this study were approved by the Loyola University Chicago Health Science Division Institutional Animal Care and Use Committee (IACUC). The Loyola University Chicago Health Science Division is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC).

This experimental paradigm has been previously described in detail (26). In brief, 18-mo-old female Fisher 344 rats were obtained from the National Institute of Aging (NIA) colony (Taconic Biosciences). The rats were pair-housed and given ad libitum access to standard chow and water. To ensure circulating E₂ levels were minimized, 1 wk after arrival the rats were anesthetized and received bilateral ovariectomy (OVX) surgery in which the ovaries and uterine horns were removed (26). One week after OVX surgery, the rats were assigned to receive a subcutaneous injection of either vehicle (safflower oil) or E₂ (2.5 µg/kg) dissolved in vehicle once a day for three consecutive days. Twenty-four hours after the final injection, the rats were euthanized, and the heart and plasma were collected. The hearts were flash frozen in liquid nitrogen and stored at −80°C. The plasma was analyzed to determine the circulating levels of 17β-estradiol using an ELISA kit and were previously reported for this paradigm (26).

Immunofluorescence Microscopy

Glass chamber slides (Nunc Lab Tek) were coated with poly-d-lysine (Millipore Sigma) and left to incubate at room temperature for 30 min. The poly-d-lysine was then aspirated, and the slides were left to air dry.

Approximately 25 mg of frozen human LV tissue was placed in a glass vial containing isolation solution (2 mM EGTA, 8.9 mM KOH, 10 mM imidazole, 7.1 mM MgCl₂, 5.8 mM ATP, 108 mM KCl) supplemented with 0.3% final volume Triton X-100 and protease and phosphatase inhibitor cocktails (Fisher Scientific). The tissue was homogenized at 7,000 RPM with a mechanical homogenizer, passed through a 70-µm filter, and left to incubate on ice. After 20 min, the filtered homogenate was centrifuged at 120 g to pellet the myocytes. The cell pellet was resuspended in isolation solution free of Triton and pipetted onto the now dry poly-d-lysine-coated chamber slide and incubated at room temperature. The solution was aspirated after 1 h and slides were washed with PBS.

Myocytes were fixed with ice-cold methanol for 1 min, followed by cold 4% paraformaldehyde for 3 min. To further permeabilize the cells, 0.5% Triton was added to each chamber and incubated for 20 min at room temperature, followed by two 15-min incubations with 0.1% Triton. A 0.1 M glycine solution (pH 7.4) was used for antigen retrieval by incubating the slides for 30 min at room temperature. The slides were next washed with PBS and incubated in blocking solution [1:1 vol/vol PBS to BSA solution (% BSA, 1% gelatin, 1% Tween-20, 0.001% NaN₃)]. After 1 h, the primary antibodies (1:300 sarcomeric α-actinin, Millipore Sigma A7811; 1:300 BAG3, Proteintech 10599-1AP; 1:150 HSF-1, Proteintech 51034-1AP) were added in blocking solution and slides were incubated for 12–14 h at 4°C.

Following incubation with primary antibody, the slides were washed with PBS and incubated with blocking solution containing secondary antibody (Abcam, AlexaFluor 488 or 568, 1:1,000) for 50 min at room temperature. The slides were mounted with Vectashield containing DAPI (Vector Laboratories), coverslipped, sealed with nail polish, and imaged at ×63 magnification on a Zeiss LSM 880 microscope under constant laser intensity and photomultiplier gain settings.

Whole Tissue Sample Preparation

Frozen left ventricular tissue (human and rat) was homogenized in lysis buffer (9 M Urea, 1% wt/vol CHAPS) supplemented with protease and phosphatase inhibitor cocktails (Fisher Scientific). The samples were then briefly sonicated and centrifuged at 10,000 g for 10 min. The supernatant contained the proteins of interest and protein concentration was determined by BCA assay (Pierce).

Myofilament Enrichment

The protocol for myofilament enrichment was described previously (27). In brief, frozen left ventricular tissue was added to a glass vial containing a standard rigor buffer (SRB) (28), supplemented with 0.5% vol/vol Triton X-100 and protease/phosphatase inhibitors, and homogenized at 7,000 RPM. The homogenates were transferred to Eppendorf tubes and incubated on ice. After 20 min, the samples were centrifuged at 1,800 g for 2 min at 4°C and the supernatant containing the cytosolic and Triton-soluble proteins was discarded. The pellet containing the myofilament proteins was washed in SRB without Triton by resuspension and centrifugation. The pellet was then resuspended in 9 M urea to solubilize the myofilament proteins and sonicated. After sonication the protein solution was centrifuged at 10,000 g for 10 min and the supernatant containing the myofilament fraction was collected. Protein concentration was determined by BCA assay (Pierce).

Nuclear Fractionation

This protocol was adapted from Dimauro et al. (29) with some minor modifications. LV tissue was placed in a glass sample vial containing STM buffer [50 mM Tris·HCl (pH 7.4), 250 mM sucrose, 5 mM MgCl₂, and 1:100 protease/phosphatase inhibitors] and minced with surgical scissors for 2 min. The sample was next placed into the pestle of a 15 mL Dounce homogenizer (Pyrex) and homogenized for ∼1 min, after which it was added to an Eppendorf tube and left on ice for 30 min. The homogenate was then vortexed and centrifuged at 500 g for 15 min at 4°C to pellet the nuclei. The supernatant was saved for later and the nuclear pellet was washed three times by resuspension in STM and centrifugation at 800 g. To isolate the nuclear proteins, the pellet was next resuspended in 1 mL NET buffer [20 mM HEPES (pH 7.9), 1.5 mM MgCl₂, 0.5 M NaCl, 0.2 mM EDTA, 10% glycerol, 1% triton, 1:100 protease/phosphatase inhibitors], vortexed, and left on ice for 30 min. The solution was then sonicated and centrifuged at 10,000 g for 20 min at 4°C. The nuclear fraction was collected in the supernatant. To purify the cytosolic fraction, the initial supernatant set aside earlier was centrifuged at 21,000 g for 15 min 4°C.

Measurement of BAG3 and HSF1 Protein Expression Level

Protein samples (15 µg) from whole tissue lysis and myofilament enrichment were prepared for Western blot in equal volume SDS Tris-Glycine Buffer (Life Technologies) supplemented with Bolt Reducing Buffer (Fisher Scientific) and heated at 95°C for 10 min. The proteins were then separated electrophoretically on 4–12% gradient Tris-glycine gels (Invitrogen) and transferred onto nitrocellulose membranes (Thermo Scientific). The membranes containing the separated proteins were briefly rinsed with ddH₂O and incubated with Revert Total Protein Stain (LI-COR Biosciences) to assess equal loading. Total proteins were imaged at this point on an Azure c600 imager (Azure Biosystems), and the membrane was then blocked for 1 h at room temperature in 1:1 (vol/vol) 1× tris-buffered saline (TBS) and Intercept Blocking Buffer (LI-COR Biosciences). The blots were incubated overnight with primary antibodies in blocking buffer. Primary antibodies and dilutions: BAG3 rabbit polyclonal (Proteintech, 10599-1AP, 1:5,000), HSF-1 rabbit polyclonal (Proteintech, 51034-1AP, 1:2500), GAPDH rabbit monoclonal (Cell Signaling, 2118, 1:2,000), Histone H3 rabbit monoclonal (Cell Signaling, 9715, 1:2,000). For visualization, IRDye 700CW anti-rabbit secondary antibodies (LI-COR) were added in blocking solution without Tween for 1 h at room temperature with agitation. The blots were imaged on an Azure c600 and analyzed with the LI-COR Image Studio software.

Statistical Analysis

Statistical analyses were performed by two-way ANOVA with Tukey’s post hoc test for multiple comparisons in cases where a significant interaction was identified. Comparisons of only two groups were performed with a two-tailed Student’s t test for unpaired samples, as annotated in the figure legends. All analyses were performed using GraphPad Prism 8.0. All data are presented as means ± SE. A P value of <0.05 was considered as statistically significant.

RESULTS

To ensure our myofilament protein fractions were free of cytosolic contaminants, we first performed a Western blot for BAG3, sarcomeric α-actin, and GAPDH in the whole LV, triton-soluble, and myofilament fractions. The result confirmed that BAG3 is present in both the cytosolic and myofilament fraction, with a greater proportion being observed in the myofilament fraction (Fig. 1A). BAG3 had previously been described to localize in a striated pattern in muscle cells, which resembles sarcomere Z-disk patterning (14, 30). To confirm this localization in the human samples analyzed in this study, we used immunofluorescence microscopy on LV cardiomyocytes and found BAG3 colocalized with the Z-disk protein α-actinin (Fig. 1B). Although such striated patterns are also observed for t-tubule and sarcomplasmic reticulum-localized proteins, due to BAG3 being so highly expressed in the myofilament protein fraction, we interpret this localization to represent the sarcomere Z-disk.

Figure 1.

Confirmation of the purity of myofilament enrichment and example of BAG3 localization in human cardiomyocytes. A: Western blot for BAG3, sarcomeric α-actin, and GAPDH in human whole LV lysate, soluble fraction, and myofilament fraction. B: representative immunofluorescence microscopy image of a human LV cardiomyocyte immunostained for BAG3 and the Z-disc protein α-actinin; ×63 magnification, scale bar = 10 µm. BAG3, Bcl2-associated athanogene 3; LV, left ventricular.

Whole LV BAG3 Expression Is Not Significantly Impacted by Dilated Cardiomyopathy or Sex

One earlier study of nine New York Heart Association (NYHA) class IV heart failure patient samples found a significant reduction in LV BAG3 expression compared with nonfailing donors (16). However, due to limited samples, this study was not powered to assess whether sex differences in BAG3 expression exist. We sought to determine the impact of sex and disease on BAG3 expression in an age-matched cohort of 19 nonfailing LV samples and 25 with heart failure stemming from nonischemic DCM (Table 1). Using Western blot to assess protein expression in whole LV tissue lysates, we did not identify a significant difference in BAG3 levels between the nonfailing and DCM hearts (P = 0.092, Fig. 2, A and B). This result was unexpected given the results of the previous study. However, BAG3 was clearly reduced among a subset of the DCM samples we analyzed, and there was considerable variation in both the nonfailing and DCM groups. Our samples represent a spectrum of heart failure severity and systolic dysfunction (22.1 ± 13.4), whereas the earlier study included only NYHA class IV patients with mean LV ejection fraction of ∼13%. Therefore, we next assessed whether decreased BAG3 expression correlated with in vivo cardiac dysfunction. To do so, we compared BAG3 protein levels with LV ejection fraction from these patients by linear regression but found no correlation, indicating that systolic dysfunction does not correlate with BAG3 levels (Fig. 2C). Next, we assessed if our data could be explained by differential expression of BAG3 in males and females but found no significant differences (Fig. 2, D and E).

Table 1.

Nonfailing and dilated cardiomyopathy patient clinical characteristics

Characteristic	Nonfailing	Dilated Cardiomyopathy	P Value
n	19	25
Age, yr	56.1 ± 11.6	52.2 ± 12.1	0.30
Female, n (%)	10 (52.6)	8 (32.0)
Ethnicity, n (%)
Non-Hispanic origin	18 (94.8)	23 (92.0)
Hispanic origin	1 (5.2)	2 (8.0)
LVEF, %	62.3 ± 8.5*	22.1 ± 13.4**	<0.0001

Values are means ± SD or n (%); n, number of patients. *Left ventricular ejection fraction (LVEF) was not determined for five nonfailing patients. **LVEF was not determined for one dilated cardiomyopathy patient.

Figure 2.

Whole LV BAG3 expression is not significantly altered by sex or dilated cardiomyopathy. A: representative Western blot of BAG3 and total protein loading control for human whole LV lysates in DCM and NF samples. B: quantitative densitometry analysis for BAG3 normalized to total protein input and separated by disease; n = 19 NF, 25 DCM; data are means ± SE, two-tailed Student’s t test. C: linear regression analysis of whole LV BAG3 vs. LVEF. D: quantitative densitometry analysis for BAG3 normalized to total protein input and separated by sex; n = 26 males, 18 females; data are means ± SE, two-tailed Student’s t test. E: normalized whole LV BAG3 separated by sex and disease; n = 9 NF male, 17 DCM male, 10 NF female, 8 DCM female; two-way ANOVA (interaction: P = 0.69, sex: P = 0.36, disease: P = 0.13). n = number of individual biological samples; BAG3, Bcl2-associated athanogene 3; DCM, dilated cardiomyopathy; LV, left ventricular; LVEF, left ventricular ejection fraction; NF, nonfailing.

Myofilament BAG3 Expression Decreases in Male Patients with DCM

Since BAG3 localizes to the sarcomere Z-disk and BAG3 expression in the myofilament fraction decreased with DCM in our recent work on a different patient cohort (4), we next sought to determine whether the myofilament pool of BAG3 displays differential regulation by sex and disease compared with the whole LV. As in our earlier study, in the present study we found that myofilament BAG3 was significantly decreased in DCM (P = 0.001, Fig. 3, A and B). When analyzing these samples after separating them by sex and disease, we found BAG3 expression in male patients was impacted more by DCM (P = 0.009) than females. DCM females did not display a significant reduction in myofilament BAG3 compared to nonfailing (P = 0.694, Fig. 3C), suggesting a potential sex difference. Notably, when sex was considered as the only variable, female samples also had significantly higher myofilament BAG3 expression than males (P = 0.028, Fig. 3D). These results indicate the myofilament pool of BAG3 is differentially affected by sex and disease compared with whole LV BAG3 expression.

Figure 3.

Myofilament BAG3 expression decreases in males with dilated cardiomyopathy, whereas female sex is associated with higher myofilament BAG3 levels. A: representative Western blot of BAG3 and total protein loading control for human myofilament fractions in DCM and NF samples. B: quantitative densitometry analysis for BAG3 normalized to total protein input and separated by disease; n = 19 NF, 25 DCM; data are means ± SE, two-tailed Student’s t test. C: normalized myofilament BAG3 expression separated by sex and disease; n = 9 NF male, 17 DCM male, 10 NF female, 8 DCM female; two-way ANOVA (interaction: P = 0.18, sex: P = 0.09, disease: P = 0.005), Tukey post hoc. D: quantitative densitometry analysis for myofilament BAG3 normalized to total protein input and separated by sex; n = 26 males, 18 females; data are means ± SE, two-tailed Student’s t test. n = number of individual biological samples; BAG3, Bcl2-associated athanogene 3; DCM, dilated cardiomyopathy; NF, nonfailing.

Acute Estrogen Treatment in Ovariectomized Female Rats Does Not Impact BAG3 Expression

Since the decrease of myofilament BAG3 in DCM was only observed in the male samples and previous evidence identified elevated BAG3 levels with increased ERα expression in cancer cells (18), we hypothesized that estrogen might regulate BAG3 expression in the heart, thus explaining the observed sex difference. To test the effects of estrogen on cardiac BAG3 expression and myofilament localization, we used a rat ovariectomy model with estrogen replacement therapy.

Eighteen-month-old female rats were ovariectomized (OVX) to remove circulating estrogen. One week after surgery, the rats were randomized to receive either 17β-estradiol (E₂) or vehicle once a day for 3 days. Twenty-four hours after the last dose, the rats were euthanized, and the hearts and blood were collected (Supplemental Fig. S1A; all Supplemental material is available at https://doi.org/10.6084/m9.figshare.13157837.v2). Previous work with this cohort showed circulating estrogen levels were significantly reduced in the OVX-veh group compared to the OVX-E₂ group (26). We used Western blot to assess BAG3 protein expression in whole LV tissue lysates and myofilament fractions from the two groups and found no significant differences (P = 0.20, Supplemental Fig. S1, B–E). We also assessed expression of heat shock factor-1 (HSF-1) but found no change with E₂ treatment (P = 0.21, Supplemental Fig. S1, F and G). These data suggest that E₂ levels do not affect whole LV BAG3 expression or localization to the myofilament in the healthy heart.

Whole LV BAG3 Expression Positively Correlates with Heat Shock Factor-1 Protein Expression and Nuclear Translocation

Although the factors regulating BAG3 expression in the heart are poorly understood, several factors have been identified as regulators of BAG3 expression in noncardiac cells (31, 32). Foremost among these is HSF-1, which in cancer cells localizes to the nucleus under cell stress conditions and upregulates the expression of several chaperones including BAG3 (33, 34). We therefore sought to determine whether HSF-1 protein expression was impacted in DCM and to assess its relationship with BAG3 levels. Using Western blot of whole LV tissue lysates from the human samples, we found that HSF1 expression displayed high variability in both groups and was not significantly different in DCM hearts compared with nonfailing (Fig. 4, A and B). We further studied whether HSF-1 expression was impacted by biological sex in the nonfailing and DCM samples and found no differences (Fig. 4C). When sex was considered as the only variable, there was also no difference in HSF-1 expression (P = 0.139, Fig. 4D).

Figure 4.

Heat shock transcription factor 1 (HSF-1) expression is not impacted by dilated cardiomyopathy or sex but correlates with BAG3 protein expression. A: representative Western blot of HSF-1 and total protein loading control for human whole LV lysates in DCM and NF samples. B: quantitative densitometry analysis for HSF-1 normalized to total protein input and separated by disease; n = 19 NF, 25 DCM; data are mean ± SE, two-tailed Student’s t test. C: normalized HSF-1 expression separated by sex and disease; n = 9 NF male, 17 DCM male, 10 NF female, 8 DCM female; two-way ANOVA (interaction: P = 0.90, sex: P = 0.12, disease: P = 0.59). D: quantitative densitometry analysis for HSF-1 normalized to total protein input and separated by sex; n = 26 males, 18 females; data are means ± SE, two-tailed Student’s t test. E: linear regression analysis of whole LV HSF-1 vs. whole LV BAG3. F: linear regression analysis of whole LV HSF-1 vs. myofilament BAG3. a.u., arbitrary units; n = number of individual biological samples; BAG3, Bcl2-associated athanogene 3; DCM, dilated cardiomyopathy; LV, left ventricular; NF, nonfailing.

To determine if BAG3 expression correlated with whole LV HSF-1 expression in these samples, we performed linear regression analysis and found a strong positive correlation for HSF-1 with whole LV (r² = 0.686, P < 0.0001) and a weakly significant relationship with myofilament BAG3 (r² = 0.228, P = 0.001, Fig. 4, E and F). These findings suggest a link between HSF-1 and BAG3 expression in the heart; however, we could not verify HSF-1 activity by the method used. To determine the impact of active HSF-1 on BAG3 expression, we selected 11 DCM samples with varying BAG3 protein levels (Fig. 5A) and performed subcellular fractionation on these samples to collect the cytosolic and nuclear fractions (Fig. 5B). Using Western blot, we determined HSF-1 expression in the cytosolic and nuclear fractions of these (Fig. 5C). When compared with BAG3 expression, cytosolic and nuclear HSF-1 each displayed a strong positive correlation with whole LV BAG3 levels (Fig. 5, D and E). Myofilament BAG3 did not significantly correlate with HSF-1 levels (Fig. 5, F and G). These results suggest that human cardiac BAG3 expression is linked to HSF-1 expression and activity, whereas the myofilament BAG3 expression likely requires additional factors.

Figure 5.

Whole LV BAG3 expression correlates with HSF-1 nuclear translocation in human dilated cardiomyopathy samples. A: Western blots for BAG3 in whole LV lysate (top) and myofilament fraction (bottom) from 11 DCM samples with varying BAG3 expression. B: Western blot for HSF-1, GAPDH (cytosol control), and histone H3 (nucleus control) from subcellular fractionation. C: Western blot for HSF-1 in cytosol and nuclear fractions from the same 11 samples used in A. D: linear regression analysis of nuclear HSF-1 expression vs. whole LV BAG3. E: linear regression analysis of cytosolic HSF-1 expression vs. whole LV BAG3. F: linear regression analysis of nuclear HSF-1 expression vs. myofilament BAG3 expression. G: linear regression analysis of cytosolic HSF-1 expression vs. myofilament BAG3 expression. BAG3, Bcl2-associated athanogene 3; DCM, dilated cardiomyopathy; HSF-1, heat shock transcription factor 1; LV, left ventricular.

HSF-1 Localizes to the Myofilament Fraction and Correlates with Myofilament BAG3 Expression

While studying HSF-1 localization in these samples, we discovered by immunofluorescence microscopy that although having strong cytosolic and nuclear localization, HSF-1 was also prominently localized to the sarcomere Z-disk as evident from its colocalization with α-actinin (Fig. 6A). To our knowledge, this is the first description of sarcomere-localized HSF-1. Given the Z-disk localization of HSF-1, we next sought to determine whether HSF-1 expression in the myofilament fraction was impacted by disease. Using Western blot in myofilament-enriched LV tissue lysates, we found HSF-1 was significantly decreased in DCM compared with nonfailing samples (P < 0.0001, Fig. 6, B and C). Myofilament HSF-1 did not display any significant difference between male and female samples (P = 0.116, Fig. 6D), but—like BAG3—was more impacted by disease in male DCM patients (P = 0.0004) than in female (P = 0.274, Fig. 6E). Lastly, when compared with myofilament BAG3 by linear regression, we found a strong correlation for myofilament HSF-1 (r² = 0.661, P < 0.0001, Fig. 6F). Together these data show the transcription factor for many sarcomere chaperones localizes to the sarcomere Z-disk, decreases in male patients with DCM, and this myofilament HSF-1 pool is tightly linked to myofilament expression of at least one of its canonical downstream gene targets, BAG3.

Figure 6.

HSF-1 myofilament localization decreases in males with dilated cardiomyopathy and tightly correlates with myofilament BAG3 expression. A: representative immunofluorescence microscopy image of a human LV cardiomyocyte immunostained for BAG3 and the Z-disc protein α-actinin; ×63 magnification, scale bar = 10 µm. B: representative Western blot for HSF-1 in myofilament-enriched LV tissue. C: quantitative densitometry analysis of myofilament HSF-1 expression normalized to total protein loading control; n = 19 NF, 25 DCM; Student’s two-tailed t test. D: normalized HSF-1 expression separated by sex; n = 26 males, 18 females; Student’s two-tailed t test. E: normalized myofilament HSF-1 expression separated by sex and disease; n = 9 NF male, 17 DCM male, 10 NF female, 8 DCM female; two-way ANOVA (interaction: P = 0.13, sex: P = 0.38, disease: P = 0.0001), Tukey post hoc. F: linear regression analysis of myofilament HSF-1 vs. myofilament BAG3. n = number of individual biological samples; BAG3, Bcl2-associated athanogene 3; DCM, dilated cardiomyopathy; HSF-1, heat shock transcription factor 1; LV, left ventricular; NF, nonfailing.

Data supplements can be found here: https://figshare.com/articles/figure/BAG3_AJP_Revision_Supplement/13157837.

DISCUSSION

Dilated cardiomyopathy (DCM) accounts for up to 40% of all heart failure cases and is characterized by systolic dysfunction and dilation/enlargement of one or both ventricles (35). Mutations and decreased expression of the heat shock protein cochaperone protein BAG3 have been linked to DCM pathogenesis through numerous clinical studies (13, 16, 36–38). In recent years, several mechanistic studies have identified diverse roles for BAG3 in the heart, highlighting its fundamental roles in regulating cardiac function. A study by Fang et al. (14), using BAG3 knockout and a DCM-associated BAG3 mutation in mice, identified BAG3 mediates cardiac protein quality control through stabilizing interactions with heat shock proteins. Through a separate pathway, BAG3 also directly modulates cardiomyocyte contractility via interactions with the β1-adrenergic receptor and L-type calcium channel (8, 39). Perhaps the most prominent localization of BAG3 in striated muscle is to the Z-disk of the sarcomere—the molecular unit of muscle contraction. Skeletal myocyte studies identified the Z-disk localized BAG3 was essential to turnover of the actin scaffolding protein filamin C through autophagy (3, 5), and we recently showed that this myofilament-associated BAG3 decreases in cardiomyocytes from human DCM left ventricle samples (4). However, despite the clear importance of BAG3 for maintaining cardiac function, the factors regulating BAG3 expression and myofilament localization in cardiomyocytes have not explored (40). In the present study, we sought to identify factors regulating cardiac BAG3 expression in human cardiomyocytes from nonfailing donor hearts and in patients with DCM.

A previous assessment of BAG3 expression in the end-stage failing heart found BAG3 levels decreased sharply in disease (16). In the present study, with a much larger cohort, we observed a trend toward decreased BAG3 in DCM. However, this difference was not statistically significant. Comparing the two studies, in Feldman et al. whole LV BAG3 decreased by almost threefold in heart failure, whereas we observed only a ∼20% decrease with considerable variability in the DCM group. One possible explanation was that our DCM cohort represented a spectrum of cardiac dysfunction, whereas the earlier study included only samples from NYHA class IV/end-stage heart failure. The NYHA classification system is based on patient symptoms and places patients into four categories based upon their limitations in physical activity, with stage IV being the worst (41). Unfortunately, the NYHA class diagnosis was not available for all the deidentified patient samples used in our study, and we were thus unable to compare BAG3 expression in each class.

A common cardiac functional parameter not included in NYHA classification is the LV ejection fraction, which is a measure of systolic function. Given the apparent influence of disease severity on BAG3 expression levels, we expected BAG3 expression would decrease with progressing systolic dysfunction. However, when we compared whole LV BAG3 expression with LV ejection fraction, we did not find a significant correlation and some samples with the lowest LV ejection fractions (<10%) had BAG3 levels comparable to nonfailing. Therefore, we conclude that decreased whole LV BAG3 expression in the very end-stage of heart failure may be explained by the overall disease severity or the poor response to physical activity but is not directly linked to systolic dysfunction. Of note, another potentially important difference between ours and the previous study is the etiology of heart failure studied. We included only samples with nonischemic DCM, whereas the earlier study included patients with both ischemic and nonischemic DCM. Future studies are needed to determine the impact of different heart failure etiologies on BAG3 expression in humans.

The disease pathogenesis of DCM frequently differs between males and females, due in part to sex differences in the regulation of gene and protein expression (17). Divergence with biological sex is apparent in the incidence, severity, and often the response to treatment for heart failure (42). The results of two earlier studies in cancer cells suggest that BAG3 expression and functions may not be uniform between males and females (18, 19). Sex differences in BAG3 expression in the heart have not been explored; however, a retrospective study of 129 BAG3 mutation carriers found that DCM disease penetrance of BAG3 mutations was significantly worse for male patients (13). This finding suggests that in some cases BAG3 may be dispensable for healthy cardiac function in females. We show in this study that whole LV BAG3 expression is not different between males and females.

We had previously identified that BAG3 localized to the cardiac sarcomere decreased in a separate cohort of human samples with nonischemic DCM (4). In the present study we also assessed expression of myofilament BAG3, given its apparent functional significance. As we previously showed, when myofilament BAG3 was measured in the present study we identified a significant decrease of ∼40% in DCM compared with nonfailing. Although sex differences in myofilament localized BAG3 were not identified in nonfailing samples, this pool of BAG3 decreased significantly in males with DCM but not in female DCM samples. These results suggest a decrease in BAG3-mediated functions may contribute to the disease pathogenesis more frequently for male patients than for females, as does the disease penetrance of BAG3 mutations (13). Notably, when we assessed the impact of 17β-estradiol treatment on BAG3 expression in a cohort of ovariectomized female rats, we did not observe a change in either the whole LV or myofilament expression levels. These results suggest the sex difference is not estrogen mediated, or, if it is, may only present in the disease state. Future studies focused on this question are warranted.

The primary factor regulating bag3 gene expression identified in noncardiac cells is the heat shock transcription factor-1 (HSF-1). Studies in cancer cell lines showed HSF-1 binds to two heat shock response elements in the bag3 gene promoter to modulate its expression (34, 43–45). In the heart, decreased HSF-1 activity has been linked to end-stage human heart failure (23) and is implicated in the transition from compensatory to pathological cardiac remodeling (25). The underlying mechanisms are not well established; however, given its role as a central mediator of cellular protein quality control, it is probable that this cardiac dysfunction stems from reduced HSF-1-mediated proteostasis. In this study, we found whole LV HSF-1 expression was not affected in humans with DCM. However, HSF-1 expression and nuclear localization were positively correlated with whole LV BAG3 expression in both the healthy and failing hearts, suggesting HSF-1 regulates BAG3 expression in the human heart.

While investigating the localization of HSF-1 in human LV cardiomyocytes, we discovered HSF-1 unexpectedly had prominent localization to the sarcomere Z-disk. Sarcomeric localization of HSF-1 has not previously been described to our knowledge and may represent a chaperone transcription factor that is positioned to rapidly respond to stress conditions experienced at the sarcomere specifically, such as those presented by the mechanical strain of contraction. Sarcomere-associated transcription factors have previously been identified in skeletal muscle to mediate sarcomere-nucleus crosstalk (46, 47). However, we are not aware of such a link for sarcomere protein quality control processes being previously identified in the heart. We show in this study that the myofilament pool of HSF-1 is impacted in DCM, where it decreased significantly in male patients compared to nonfailing. Again, like BAG3, such a decrease was not found in DCM females. Importantly, myofilament HSF-1 displayed a striking positive correlation with myofilament BAG3 expression, further supporting BAG3 expression—even in specific cellular compartments—is regulated by HSF-1. Previous studies have identified therapeutic promise for increasing HSF-1 activity to prevent cardiac dysfunction in ischemia- and pressure-overload-induced cardiac hypertrophy (23, 48, 49). Further studies are warranted to characterize the therapeutic potential for increasing BAG3 and other chaperone expression at the sarcomere through HSF-1 activation, particularly in the context of dilated cardiomyopathy.

Study Limitations

There are several limitations in the present study that must be noted. First, due to the limited clinical information at our disposal from the DCM and NF samples, we cannot rule out the impact of different medications or cardiovascular comorbidities on BAG3 expression. Additionally, although we conducted a robust analysis of nonischemic DCM-associated heart failure, future studies are needed to determine the impact of other heart failure etiologies on BAG3 and HSF-1 expression/localization, including ischemic cardiomyopathy (ICM) and hypertrophic cardiomyopathy (HCM). Regarding the impact of estrogen on BAG3 expression, although we did not see an effect in healthy rats, this study did not test whether there is an estrogen-dependent impact on BAG3 in heart failure. With respect to the relationship between transcription factors and BAG3 expression, although HSF-1 is the primary transcription factor for BAG3, our study did not assess the relationship of BAG3 expression with its other identified transcription factors NFκB (32) and JNK (50). Lastly, the nature of the relationship between myofilament localized HSF-1 and BAG3 on a mechanistic level is not clear from the present study.

Conclusions

BAG3 is a multifaceted cochaperone protein. Previous studies have identified that both decreased protein expression of BAG3 and BAG3 mutations are associated with dilated cardiomyopathy and heart failure. In the present study, we show that HSF-1 expression also impacts BAG3 expression, which may have implications in disease where protein quality control mechanisms are frequently impaired. We also assess the expression of BAG3 in the myofilament protein fraction and show that this pool of BAG3 is differentially impacted by heart failure in male and female patients and is also linked to myofilament HSF-1 localization. Understanding the factors that regulate cardiac BAG3 expression is an important consideration for those focused on BAG3 as a therapeutic target for both inherited BAG3 mutation-associated cardiovascular disease and nongenetic heart failure.

GRANTS

This work was supported by National Institutes of Health Grants HL136737 (to J.A.K.) and AG033605 (to T.R.P.) and American Heart Association Predoctoral Fellowship 20PRE35170045 (to T.G.M.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

T.G.M., S.T., T.R.P., and J.A.K. conceived and designed research; T.G.M., S.T., and C.S.M. performed experiments; T.G.M. and S.T. analyzed data; T.G.M., T.R.P., and J.A.K. interpreted results of experiments; T.G.M. prepared figures; T.G.M. drafted manuscript; T.G.M., T.R.P., and J.A.K. edited and revised manuscript; T.G.M., S.T., C.S.M., T.R.P., and J.A.K. approved final version of manuscript.