BSCL2/Seipin deficiency in hearts causes cardiac energy deficit and dysfunction via inducing excessive lipid catabolism

Abstract Background Heart failure (HF) is one of the leading causes of death worldwide and is associated with cardiac metabolic perturbations. Human Type 2 Berardinelli‐Seip Congenital Lipodystrophy (BSCL2) disease is caused by mutations in the BSCL2 gene. Global lipodystrophic Bscl2−/− mice exhibit hypertrophic cardiomyopathy with reduced cardiac steatosis. Whether BSCL2 plays a direct role in regulating cardiac substrate metabolism and/or contractile function remains unknown. Methods We generated mice with cardiomyocyte‐specific deletion of Bscl2 (Bscl2cKO ) and studied their cardiac substrate utilisation, bioenergetics, lipidomics and contractile function under baseline or after either a treatment regimen using fatty acid oxidation (FAO) inhibitor trimetazidine (TMZ) or a prevention regimen with high‐fat diet (HFD) feeding. Mice with partial ATGL deletion and cardiac‐specific deletion of Bscl2 were also generated followed by cardiac phenotyping. Results Different from hypertrophic cardiomyopathy in Bscl2−/− mice, mice with cardiac‐specific deletion of Bscl2 developed systolic dysfunction with dilation. Myocardial BSCL2 deletion led to elevated ATGL expression and FAO along with reduced cardiac lipid contents. Cardiac dysfunction in Bscl2cKO mice was independent of mitochondrial dysfunction and oxidative stress, but associated with decreased metabolic reserve and ATP levels. Importantly, cardiac dysfunction in Bscl2cKO mice could be partially reversed by FAO inhibitor TMZ, or prevented by genetic abolishment of one ATGL allele or HFD feeding. Lipidomic analysis further identified markedly reduced glycerolipids, glycerophospholipids, NEFA and acylcarnitines in Bscl2cKO hearts, which were partially normalised by TMZ or HFD. Conclusions We identified a new form of cardiac dysfunction with excessive lipid utilisation which ultimately causes cardiac substrate depletion and bioenergetics failure. Our findings also uncover a crucial role of BSCL2 in controlling cardiac lipid catabolism and contractile function and provide novel insights into metabolically treating energy‐starved HF using FAO inhibitor or HFD.

Cardiac deletion of BSCL2 causes ATGL overexpression, excessive fatty acid oxidation (FAO), drastic reduction of cardiac lipidome and ultimately energetic and contractile dysfunction in mice. Inhibiting FAO by partial ATGL deletion or trimetazidine and increasing lipid supply via high fat diet (HFD) feeding alleviate cardiac dysfunction caused by BSCL2 deficiency. BSCL2 is a novel protein indispensable for cardiac lipid metabolism and function. TMZ, or prevented by genetic abolishment of one ATGL allele or HFD feeding. Lipidomic analysis further identified markedly reduced glycerolipids, glycerophospholipids, NEFA and acylcarnitines in Bscl2 cKO hearts, which were partially normalised by TMZ or HFD. Conclusions: We identified a new form of cardiac dysfunction with excessive lipid utilisation which ultimately causes cardiac substrate depletion and bioenergetics failure. Our findings also uncover a crucial role of BSCL2 in controlling cardiac lipid catabolism and contractile function and provide novel insights into metabolically treating energy-starved HF using FAO inhibitor or HFD.

K E Y W O R D S
BSCL2/Seipin, heart failure, lipid metabolism, lipidomics INTRODUCTION Heart failure (HF) is one of the leading causes of morbidity and mortality worldwide. In healthy individuals, the heart exhibits striking metabolic flexibility, being capable of utilising carbohydrate, lipid, amino acids and/or ketone bodies to meet energetic demands, cellular constituent turnover and metabolic signalling; these processes are critical for maintenance of mechanical work. Oxidation of fatty acids (FAs) predominates, accounting for 60-70% of myocardial oxygen consumption. 1 In patients with hypertrophied and failing hearts, derangements of substrate utilisation include an increased reliance on glycolysis concomitant with an overall reduced oxidative metabolism. 2 The severely failing heart usually demonstrates a lower concentration of ATP, supporting the concept that energy starvation contributes significantly to the pathogenesis of HF. 3 Indeed, there is a striking correlation between cardiac energetic status and survival in HF patients. 4 Thus, targeting metabolic processes in the heart may represent a promising way to develop new therapeutic approaches for HF.
Cardiac lipid metabolism is precisely controlled to maintain a balance between FA uptake, triglyceride (TG) synthesis, TG hydrolysis and FA oxidation (FAO). 5 Imbalances in these processes are commonly seen in obese and diabetic patients and animal models, which are associated with cardiac lipotoxicity (i.e., accumulation of toxic lipid intermediates) and contractile dysfunction. 5 Recent preclinical and clinical evidence argues for an important role of adipose triglyceride lipase (ATGL)-mediated cardiac lipolysis in promoting mitochondrial FAO and ATP production thus contractile function. 6 Constitutive Atgl −/− mice develop severe cardiac steatosis and HF, associated with a high mortality. 7 Conversely, cardiomyocyte-specific overexpression of ATGL maintains normal cardiac function in lean mice and reduces cardiac TG content and improves cardiac function during diabetes and obesity. 8,9 The precise mechanisms linking cardiac lipid catabolism with contractile dysfunction remain to be further explored.
The present study highlights a new form of metabolic cardiomyopathy caused by deletion of BSCL2 in hearts. Cardiomyocyte-specific deletion of BSCL2 enhanced ATGL expression and FAO, resulting in markedly reduced cardiac lipid reserve, associated with compromised ATP production and contractile dysfunction. Inhibition of FAO or supplying FAs by high-fat diet (HFD) feeding partially alleviated cardiac energetic stress and augmented contractile function. These findings improve our understanding of how perturbations in lipid utilisation/storage contribute towards HF development and provide preclinical insights into metabolic treatment of HF. Mice were initially anaesthetised with a 3% isoflurane, followed by maintenance at 1-2%. Cardiac gravimetric and histological measurements were performed as described elsewhere. 35 All mice were housed in the central animal facility with room temperature controlled at 21 • C, and an artificial 12 h:12 h light:dark cycle (lights on at 06:00 am). Mice were directly sacrificed by cervical dislocation and hearts were rapidly excised. In ex vivo perfused working heart experiments, hearts were rapidly excised following anaesthesia via intraperitoneal ketamine/xylazine (80/10 mg/kg) injection.
Measures of cardiac metabolism (e.g., oleate and glucose oxidation and oxygen consumption) and function (e.g., cardiac power) were determined. At the end of the perfusion period, hearts were snap-frozen and stored at −80 • C prior to analysis. Data were presented as steady-state values (i.e., the mean of the last two time points during a distinct perfusion condition for each individual heart).

2.3
Lipidomic analysis by high-resolution/accurate mass spectrometry and tandem mass spectrometry Total lipids from frozen ventricles were extracted and resuspended in isopropanol:methanol:chloroform (4:2:1, v:v:v) containing 20 mM ammonium formate followed by untargeted lipidomic analysis. Relative quantification of abundances between samples was performed by normalising target lipid ion peak areas to the PC (14:0/14:0) internal standard followed by normalisation to tissue weights as previously described. 35

FA and glucose oxidation assays
FAO and glucose oxidation reaction assays with LV homogenates were prepared and carried out as detailed previously. 39

Mitochondrial isolation and measurement of mitochondrial respiration
Fresh ventricles were isolated and minced for mitochondrial isolation and measurement of mitochondrial respiration by XFe24 Extracellular Flux Analyzer (Seahorse Bioscience) as previously described. 34 See Supplemental Materials and Methods for details.

2.6
Tissue ATP measurement ATP content was determined by using ATP Bioluminescent Assay Kit (FL-AA; Sigma-Aldrich, Saint Louis, MO, USA) according to the manufacturer's procedure. Briefly, frozen heart tissues were homogenised in cold 10% trichloroacetic acid buffer, centrifuged at 5000×g for 10 min at 4 • C followed by neutralisation with 50 mM Trisacetate (pH 7.8). The ATP content was then determined by a microplate reader with luminescence luminometer (FLUOstar Omega; BMG Labtech). Data were normalised to tissue weight.

RNA isolation and quantitative real-time PCR
Total RNA was extracted with Trizol Reagent (Thermo Fisher) and reverse-transcribed using MLV-V reverse transcriptase using random primers (Invitrogen). Real-time quantitative RT-PCR was performed on the Strategene MX3005 system. Data were normalised to two housekeeping genes (Actb and 36B4) based on Genorm algorithm (medgen.ugent.be/genorm/) and expressed as fold changes. All tissue gene expression studies were performed in non-fasted mice. qRT-PCR primers were listed in Table S1.

Immunoblotting
Protein extraction, determination of protein concentrations, Western blotting and quantification were performed as previously described. 35 Specific antibodies were listed in Supplemental Materials.

Statistical analysis
Quantitative data were presented as means ± SEM. Statistical comparisons were made by using unpaired t-test, one-way ANOVA or two-way ANOVA followed by Tukey's post-hoc tests or multiple t-tests after correction using the Holm-Sidak method using the built-in statistics of Graph-Pad Prism 9 software. The number of independent biological replicates is indicated as n in the figure legends. ANOVA F-test was used to check homogeneity of variance, and Shapiro-Wilk test was used to check normality before unpaired t-test using GraphPad Prism 9 software. A p value of less than .05 was considered statistically significant. Additional methodological details are included in the Supplemental Materials.

Myocardial deletion of BSCL2 induces systolic heart dysfunction with dilation
Previously, we demonstrated lipodystrophic Bscl2 −/− mice develop mild hypertrophic cadiomyopathy. 35 In order to interrogate the specific role of cardiac BSCL2, we generated a mouse model with a cardiomyocyte-specific deletion of BSCL2 (Bscl2 cKO ) via Myh6-Cre. Gene expression analysis confirmed an approximate 75% reduction of Bscl2 specifically in cardiac muscle but not in liver, skeletal muscle and epididymal white adipose tissue (eWAT) of Bscl2 cKO mice ( Figure S1(A)). Cardiac-specific deletion of Bscl2 also resulted in no change in body fat and lean masses ( Figure S1(B)). We were not able to confirm the cardiac deletion of BSCL2 at protein level as no antibody is sensitive enough to detect the low-abundance endogenous BSCL2 protein in murine hearts.
We next determined the impact of cardiomyocytespecific deletion of Bscl2 on cardiac function, in comparison with two distinct control groups [Ctrl and Cre+;Bscl2 w/w ]. No changes were observed in total body weights (Table S2) To further pinpoint a cell-autonomous role of BSCL2 in mediating cardiac function, we also compared echocardiography in 5-month-old mice, an age when the cardiac function of Myh6-Cre+ (Cre+,Bscl2 w/w ) mice is not compromised due to prolonged Cre expression. Not surprisingly, while the wall thicknesses of 5-month-old Bscl2 cKO mice were marginally reduced, these mice still demonstrated significantly reduced contractile function accompanied with increased dilation compared to both Ctrl and ). In addition, we observed a greater induction of brain natriuretic peptide (Nppb) and growth differentiation factor 15 (Gdf15), and reduction of adult Myh6 gene expression, in 6-month-old Bscl2 cKO hearts versus Ctrl hearts (Figure 1(J)). The BSCL2 deletion-induced cardiac dysfunction was not accompanied by abnormal cardiomyocyte morphology and excessive myocardial fibrosis (assessed by trichrome staining) in hearts of Bscl2 cKO mice (Figures S1(D) and S1(E)). Together, these data suggest that cardiomyocyte-specific BSCL2 deletion leads to systolic heart dysfunction independent of the long-term expression of transgene Myh6-Cre.

3.2
Myocardial-specific deletion of BSCL2 causes elevated TG turnover and FAO preceding functional decline It remains to be determined whether loss of BSCL2 in cardiomyocytes directly reduces cardiac TG contents as we have observed in lipodystrophic Bscl2 −/− hearts. 35 Indeed, we found reduced ventricular TG content (by 57%) (Figure 2(A)), increased protein expression of ATGL (≈2.5fold), but not HSL (Figure 2(B)), in 3-month-old Bscl2 cKO mice compared with Ctrl mice. Changes in ATGL protein expression were not attributed to alterations in the mRNA level of Pnpla2, the gene encoding ATGL ( Figure S2(A)). Isolated adult Bscl2 cKO cardiomyocytes displayed similar upregulation of ATGL, but not HSL, compared with Ctrl cells (Figures 2(C) and 2(D)). Ctrl and Bscl2 cKO cardiomyocytes responded similarly to the stimulation of isoproterenol in terms of PKA-mediated phospholamban (PL) phosphorylation ( Figure S2(B)). Analysis of whole heart lysates also revealed comparable basal PKA-mediated substrate phosphorylation between Ctrl and Bscl2 cKO mice ( Figure S2(C)). These data suggest that cardiac-specific deletion of BSCL2 did not affect cAMP/PKA signalling but induced posttranscriptional ATGL upregulation in hearts, different from what we previously observed in BSCL2deleted adipose tissue. [32][33][34] Higher ATGL-mediated lipolysis is coupled with cardiac FAO. 6 When incubating heart homogenates with 14 C-palmitate, 3-month-old Bscl2 cKO mice demonstrated elevated release of 14 C-labelled acid soluble metabolites despite a lack of change in 14 CO 2 production (Figure 2(E)). Ex vivo perfusions in working hearts from 3-month-old male Ctrl and Bscl2 cKO mice identified no differences in heart rates, cardiac oxygen consumption, cardiac efficiency as well as cardiac power between two genotypes ( Figures S2(D) and S2(G)), suggesting maintained cardiac function at this age. However, the Bscl2 cKO hearts demonstrated a 33% increase in oleate oxidation (Figure 2(F)), concomitant with a tendency of lower glucose oxidation (Figure 2(G)) and an 83% increase of baseline glycogen content (Figure 2(H)). In spite of enhanced FAO and glycogen accumulation, Bscl2 cKO hearts demonstrated no changes in the expression of genes involved in FA uptake (e.g., Lpl, Cd36), TG synthesis (e.g., Dgat1, Dgat2), mitochondrial and peroxisomal β-oxidation (e.g., Pparα, Cpt1β, Acadl, Acox1), mitochondrial biogenesis (e.g., Pgc1α, Tfam) and glucose metabolism (e.g., Pdk4, Glut1, Glut4) (Figure 2(I)). Lack of changes in mitochondrial biogenesis was further confirmed by the similar ratios of mitochondrial DNA to nuclear DNA ( Figure S2(H)) and protein levels of each of the electron transport chain (ETC) complexes between heart lysates of 3-month-old Ctrl and Bscl2 cKO mice ( Figure S2(I)). Together, these data clearly suggest that cardiomyocyte-specific BSCL2 deficiency results in higher rates of cardiac TG turnover and FAO independent of transcriptional changes of mitochondrial and extramitochondrial metabolic genes.

Chronic derangements in myocardial FAO leads to massive lipid remodelling and reduced endogenous substrates in Bscl2 cKO hearts
To identify mechanisms underlying the development of HF, we performed untargeted lipidomic analyses of ventricles from 6-month-old Ctrl and Bscl2 cKO mice. Total normalised lipid ion abundances identified in Bscl2 cKO hearts were reduced by about 55% (Figure 3(A)). Heatmap analysis revealed massive reductions in the absolute levels of five broadly classified lipid classes defined by the Lipid MAPS consortium; that is, glycerophospholipids, fatty acyls [mainly non-esterified fatty acids (NEFA)], sphingolipids, sterol lipids and glycerolipids in Bscl2 cKO mice (Figure 3(B)). When comparing the % distributions of these five lipid classes, the proportions of glycerophospholipids, sphingolipids and sterol lipids were significantly higher in Bscl2 cKO hearts compared with Ctrl hearts (Figure 3(C)). The proportions of NEFA were relatively comparable between two genotypes, whereas the proportion of glycerolipids was markedly reduced by 58% in Bscl2 cKO hearts (Figure 3(C)). Analysis of the absolute levels of glycerolipids revealed 78%, 50% and 75% reductions in TG, diacylglycerol (DAG) and monoacylglycerol (MAG), respectively, in Bscl2 cKO hearts (Figure 3(D)). The absolute levels of NEFA and total acylcarnitines (ACs) were significantly lower in Bscl2 cKO hearts (Figure 3(E)). Specifically, the abundances of the major long-chain ACs (AC16:0, AC16:1, AC18:0, AC18:1, AC18:2) were all reduced by approximately 70% (Figure 3(F)). These data suggest myocardial BSCL2 deletion results in a dramatic remodelling of lipid compositions highlighted by reduced levels of energy providing substrates indicative of impairment of cardiac metabolic reserve. Furthermore, only Bscl2 cKO hearts displayed reduced cardiac TG contents ( Figure S3(A)) and increased ATGL expression ( Figure S3(B)), and no differences were observed between Ctrl and Cre+;Bscl2 w/w hearts. The expression of PPARα and its target proteins (CD36, CPT1β) were also similar between three genotypes ( Figure S3(B)). Collectively, these data emphasise a BSCL2specific regulation of lipid remodelling in Bscl2 cKO hearts independent of transcriptional activation of PPARα.

3.4
Cardiomyopathy in aged Bscl2 cKO mice is associated with energy deficiency independent of overt mitochondrial dysfunction and oxidative stress We next examined whether the massive lipid remodelling exerts maladaptive effects on mitochondrial function, leading to the progressive development of HF when Bscl2 cKO mice age. Transmission electron microscopy images revealed a complete lack of LDs in 6-monthold Bscl2 cKO hearts, in support of cardiac TG reduction (Figure 4(A)). Sarcomere arrangement, mitochondrial morphology and sizes, as well as mitochondria distribution were generally preserved in both Ctrl and Bscl2 cKO hearts (Figure 4(A)). Respirational analysis of isolated mitochondria from hearts of Ctrl and Bscl2 cKO mice identified similar basal (in the presence of exogenous succinate and rotenone) and maximal oxygen consumption rates, as well as coupled (ADP-driven, Complex V) and uncoupled respiration (Figures 4(B)-4(C)). Interestingly, the activity of Complex I (CI) was significantly increased, whereas the activities of CII and CIII trended higher in Bscl2 cKO mitochondria compared with those from Ctrl or Cre+;Bscl2 w/w mice (Figure 4(D)). Again, the expression of marker proteins per mitochondrial mass for each mitochondrial ETC complex remained comparable (Figure 4(E)), which was also the case in the whole hearts of 6-month-old Bscl2 cKO mice ( Figure S4(A)). There were no differences in the expressions of PGC1α and mitochondrial stress and uncoupling proteins (Prohibitin, UCP2 and UCP3) in Bscl2 cKO hearts compared to both Ctrl and Cre+;Bscl2 w/w hearts ( Figure S4(A)). Six-month-old Bscl2 cKO hearts still possessed a tendency of higher capacity to oxidise exogenous palmitate relative to Ctrl hearts (Figure 4(F)); while its glucose oxidation rate trended higher (Figure 4(G)), resulting in a tendency toward lower cardiac glycogen content in these mice (Figure 4(H)). Oxidation of 2′, 7′-dichlorofluorescein diacetate was not augmented in heart extracts of 6-month-old Bscl2 cKO mice ( Figure S4(B)). The level of lipid peroxides malondialdehyde was even reduced in Bscl2 cKO hearts compared with Ctrl hearts ( Figure S4(C)). Consistent with lack of oxidative stress, there were no differences in the expression of SOD1, SOD2 and Catalase in Bscl2 cKO hearts relative to both Ctrl and Cre+;Bscl2 w/w hearts ( Figure S4(A)). Furthermore, genes regulating glutathione metabolism (e.g., Gclc, Gsr, Gss) were largely unaltered although GPX3 expression showed a trend toward increase in both Cre+;Bscl2 w/w and Bscl2 cKO hearts compared with Ctrl hearts (Figure S4(D)). These data suggest that mitochondrial dysfunction and oxidative stress are unlikely to play roles in maladaptive cardiac remodelling and progression of HF in Bscl2 cKO mice.
While the ventricular ATP content was maintained in 3month-old Bscl2 cKO mice, it was significantly lower when Bscl2 cKO mice reached 6-month-old (Figure 4(I)), in line with the reduced cardiac metabolic reserve (i.e., lipid and glycogen stores). Furthermore, there were no differences in the plasma concentrations of glucose and lipid substrates (TG, cholesterol, NEFA and glycerol) between 6month-old Ctrl and Bscl2 cKO mice (Table S3). These data suggest that deprivation of myocellular endogenous substrates is likely responsible for the progressive metabolic transition leading to cardiac energetic and contractile failure in Bscl2 cKO mice.

Inhibition of cardiac TG lipolysis or FAO partially rescues cardiac dysfunction in Bscl2 cKO mice
Since cardiac dysfunction in Bscl2 cKO mice is associated with higher ATGL expression and FAO, we first generated mice with partial loss of ATGL in Bscl2 cKO mice (A h B cKO ) ( Figure S5(A)). ATGL haploinsufficiency indeed abolished cardiac ATGL upregulation in B cKO mice, with only 50% myocardial ATGL present compared with Ctrl hearts (Figure S5(B)). A h B cKO mice showed no alterations in body weights, and VW to TL ratios when compared with Ctrl and B cKO mice ( Figure S5(C) and Table S4). Partial ATGL loss neither significantly restored the wall thicknesses nor reduced dilation of B cKO mice ( Figure S5(D) and Table S4). However, the cardiac contractile function was  Figure S5(G)). These data suggest cardiac ATGL upregulation indeed contributes to the development of cardiac dysfunction in Bscl2 cKO mice.
To further dissect whether elevated FAO is involved in the pathogenesis of HF in Bscl2 cKO mice, we treated mice with TMZ, a specific 3-Ketoacyl-CoA thiolase qRT-PCR analysis of lipid and glucose transport genes and cardiac stress genes in ventricles of 6-month-old PBS and TMZ-treated male mice. n = 6 per group. *, p < .05; **, p < .005. ns: not significant. Two-way ANOVA followed by Tukey's post-hoc tests  inhibitor that inhibits FAO, 37 for up to 6 weeks starting at 6-month-old when Bscl2 cKO mice already developed cardiac dysfunction ( Figure 5(A)). TMZ did not alter body weight ( Figure S6(A)) nor liver weight ( Figure S6(B)), but tended to increase white fat mass in both genotypes ( Figure S6(C)). TMZ elevated circulating cholesterol levels in both Ctrl and Bscl2 cKO mice, but there were no dif-ferences in plasma glucose, NEFA, glycerol or TG levels in the experimental groups (Table S5). Ctrl and Bscl2 cKO mice also demonstrated similar ventricle to body weight ( Figure S6(D)) and tibia length ( Figure 5(B)) ratios after TMZ treatment. TMZ produced no changes in cardiac dilation and contractile function of Ctrl mice. However, it was able to significantly improve LVPWs and cardiac contractile function (including increasing fractional shortening by 20.6%) without notably normalising dilatation in Bscl2 cKO hearts (Figures 5(C)-5(E) and Table S5). Attenuation of cardiac dysfunction by TMZ was also evident in female Bscl2 cKO mice ( Figure S6(E)-S6(H)).
We performed another independent set of untargeted lipidomics to examine the effect of TMZ on cardiac lipidome. As indicated in Figure 5(F), PBS-treated Bscl2 cKO hearts recapitulated all changes in cardiac lipid contents as we previously observed in 6-month-old Bscl2 cKO hearts (Figure 3). TMZ unexpectedly reduced the absolute abundances of cardiac lipids in all categories except ACs in Ctrl hearts. In contrast, TMZ dramatically enhanced the accumulation of these lipid classes in Bscl2 cKO hearts relative to PBS-treated Bscl2 cKO hearts. Especially, the absolute levels of phospholipids, sphingolipids, and sterols were even greater in TMZ-treated Bscl2 cKO than PBS-treated Ctrl hearts ( Figure 5(F)). Specifically, TMZ markedly increased TG, NEFAs and ACs in Bscl2 cKO hearts (Figures 5(G) and 5(H)). While it slightly lowered glycogen in Ctrl hearts, it caused more glycogen accumulation in Bscl2 cKO hearts ( Figure 5(I)). Despite a significant upregulation of metabolic reserve, there was only a very minimal but non-significant improvement of ventricular ATP content in TMZ-treated Bscl2 cKO hearts ( Figure 5(J)). As expected, TMZ itself did not abolish ATGL upregulation in Bscl2 cKO hearts, since it acts on FAO downstream of ATGL ( Figure 5(K)). Downregulation of lipid and glycogen contents in TMZ-treated Ctrl heart could not be explained by similar expression levels of Cd36 and Glut4, two key genes governing cardiac lipid and glucose entries. However, we did observe a slight upregulation of these two genes in TMZ-treated compared to PBS-treated Bscl2 cKO hearts, in line with their elevated lipid and glycogen contents ( Figure 5(L)). Furthermore, there was also a trend for TMZ to attenuate the gene expression of cardiac stress markers in Bscl2 cKO hearts ( Figure 5(L)). Together, our data suggest that inhibiting FAO in Bscl2 cKO hearts is able to partially restore cardiac function, potentially through modifying cardiac metabolism.

Preventative treatment of HFD significantly attenuates cardiac bioenergetics deficiency and dysfunction in Bscl2 cKO mice
To gain further insight into the importance of metabolic reserve in energy-starved Bscl2 cKO hearts, we also fed 3month-old Ctrl and Bscl2 cKO male mice a normal chow diet (NCD) or a 60% HFD for a period of 3 months (Figure 6(A)). By 6 months old, Ctrl and Bscl2 cKO mice exhibited similar obese phenotype with comparable weight gains after HFD feeding (Table S3). They exhibited similarly higher levels of plasma glucose, cholesterol, NEFA and glycerol relative to their NCD-fed counterparts, and there were no differences in serum TG concentrations in all four groups (Table S3). HFD increased comparable cardiac hypertrophy as evidenced by ventricle weight to tibia length ratios and a tendency of increase in wall thickness in both genotypes (Figure 6(B) and Table S3). There were no significant changes in dilation and contractile function in Ctrl mice after 3 months of HFD feeding (Figures 6(C) and 6(D) and Table S3). However, HFD was able to increase wall thickness (LVPWs) and improve cardiac contractile function in Bscl2 cKO mice to the levels similar to HFD-fed Ctrl mice despite exerting no effect on dilation (Figures 6(C) and 6(D) and Table S3). Interestingly, in Ctrl hearts, we found HFD reduced the amounts of phospholipids (i.e., PC, PS, PI, PG, CL and PA) and sphingolipids, but preferentially increased the levels of sterols and glycerolipids when compared with NCD ( Figure 6(E)). The amounts of almost all lipid classes were increased in Bscl2 cKO hearts after HFD feeding, albeit still lower than HFD-fed Ctrl hearts (Figure 6(E)). Specifically, TG levels were greatly upregulated in both Ctrl and Bscl2 cKO hearts after HFD feeding, whereas the levels of NEFA and ACs in Bscl2 cKO hearts were increased to a lesser extent by HFD when compared with NCD (Figures 6(F) and 6(G)). Nevertheless, while Bscl2 cKO hearts contained less ATP under NCD, the ventricular ATP contents in HFDfed Bscl2 cKO mice was partially recovered to the levels of HFD-fed Ctrl hearts, suggesting improved cardiac energetics ( Figure 6(H)). Immunoblot analysis revealed similar upregulation of ATGL in Bscl2 cKO relative to Ctrl hearts regardless of diets ( Figure 6(I)). Collectively, HFD could restore cardiac function by improving cardiac energetics via increasing energy supply in Bscl2 cKO mice.

DISCUSSION
In this study, we show that genetic deletion of BSCL2 in cardiomyocytes leads to dramatic cardiac lipid remodelling and systolic heart dysfunction with dilation in mice. Mechanistically, cardiac BSCL2 ablation causes ATGL overexpression, excessive FAO and overt cardiac lipid remodelling. Increased lipid catabolism ultimately exhausts intramyocellular lipid and glycogen reserve and is likely responsible for the energetic and contractile dysfunction in Bscl2 cKO mice. Importantly, inhibiting TG turnover by partial genetic deletion of ATGL or suppressing FAO by promoting substrate switch or HFD feeding via increasing lipid supply can partially mitigate cardiac dysfunction in Bscl2 cKO mice (Figure 7). Our results thus identify a novel and indispensable role of BSCL2 in regulating a BSCL2 deletion enhances cAMP/PKA triggered ATGLmediated lipolysis and FAO in adipose tissue. [32][33][34] Similar to global Bscl2 −/− hearts, 35 Bscl2 cKO hearts exhibited higher ATGL expression associated with reduced TG contents and accelerated FAO, highlighting the cellautonomous role of cardiac BSCL2 in lipid catabolism. Interestingly, lipodystrophic Bscl2 −/− mice develop cardiac hypertrophy due to increased basal IGF1 receptor (IGF1R)-mediated PI3K/AKT signalling secondary to hyperinsulinemia. 35 These mice develop mild cardiomyopathy, which can be attenuated by partially restoring fat mass and/or improving whole-body and cardiac insulin resistance. 35,41 This suggests that insulin resistance largely accounts for the pathophysiology of metabolic cardiomy-opathy in lipodystrophic Bscl2 −/− mice, resembling diabetic hearts. Indeed, in patients with congenital generalised lipodystrophy, leptin treatment, which is known to improve metabolic complications including insulin resistance, has been recently shown to attenuate cardiac hypertrophy and increase contractile function. 42 On the other hand, Bscl2 cKO mice exhibit neither left ventricular hypertrophy nor functional changes at 3 months old despite an increase in FAO. By 6 months old, these mice tend to have thinner left ventricular wall and develop energy deficitinduced systolic dysfunction but not end-stage HF. Such changes occur independent of profound structural remodelling (e.g., hypertrophy and fibrosis) ( Figure 1) and insulin resistance in Bscl2 cKO hearts (data not shown). Notably, a previous report identified no differences in heart weights of 1-year-old mice with cardiac deletion of BSCL2 driven F I G U R E 7 Schematic diagram of the indispensable role of BSCL2 in regulating cardiac lipid metabolism and function. Cardiac deletion of BSCL2 causes ATGL overexpression and excessive fatty acid oxidation (FAO) which exhaust intramyocellular triglyceride (TG) and induce drastic depletion of cardiac lipidome, ultimately resulting in energetic and contractile failure in mice. Partial deletion of ATGL, inhibiting FAO by trimetazidine or increasing lipid supply via high-fat diet (HFD) feeding replenishes cardiac lipidome and alleviates cardiac dysfunction caused by loss of BSCL2 by the same Myh6-Cre. 43 However, cardiac function was not directly assessed and it remains unknown whether BSCL2 directly controls cardiac function in their model. Alleviation of cardiac dysfunction by partial ATGL loss in Bscl2 cKO mice provides solid evidence that an uncontrolled ATGL-mediated lipid catabolism plays a part in impairing cardiac performance ( Figure S5). ATGLmediated fat catabolism has been directly linked to cardiac PGC-1/PPARα expression and FAO rates. 6 While Bscl2 cKO hearts clearly exhibited excessive ATGL-coupled FAO preceding impaired cardiac function, none of PGC-1α and PPARα and their downstream target genes were altered ( Figure 2). However, lean mice with myocardial ATGL overexpression 8,9 or acetyl-CoA carboxylase 2 (ACC2) deletion 44 maintain normal cardiac energetics and performance despite higher TG turnover or FAO rates. Thus, other changes in cellular metabolism independent of ATGL-mediated intramyocardial lipid catabolism may exist to contribute to the cardiac energy deficit and pro-gressive deterioration of cardiac function in Bscl2 cKO mice at baseline, which warrants further investigation.
ATGL is known to be ubiquitinated (Ref. 45 and data not shown) and we have previously demonstrated enhanced ATGL stability in Bscl2 −/− cardiomyocytes and mouse embryonic fibroblasts (MEFs). 35 However, we were not able to pull down endogenous ATGL using current available antibodies, which prevents us from examining ATGL ubiquitination in BSCL2-deleted hearts. Therefore, the molecular events for the posttranscriptional regulation of ATGL in the absence of BSCL2 remain to be identified.
Reliance on FAO in obesity and/or diabetes is correlated with lower cardiac efficiency, impaired mitochondrial respiratory function and increased ROS production. 46 Alteration of the cardiac lipidome may also mediate functional impairment through dampening mitochondrial function. 47 However, none of these abnormalities occurred in our Bscl2 cKO mice despite massive reduction of cardiac lipid contents. In fact, mitochondrial function still trends higher even in BSCL2-deleted hearts with reduced ATP levels. Such finding is seemingly counterintuitive, as it is hard to associate higher myocardial mitochondrial function with reduced ATP level. It is worth to mention that mitochondrial hyperactivation in Bscl2 cKO hearts was derived from in vitro or ex vivo assays performed in the presence of excessive exogenous substrates, whereas this may not be the case in vivo. Higher cardiac FAO is normally associated with increased exogenous lipid import as observed in diabetic hearts or hearts with cardiac-restricted overexpression of PPARα. 48 This was also not true with our Bscl2 cKO mice which displayed similar expression of lipid uptake genes and comparable circulating lipid metabolites. Therefore, imbalance of lipid consumption and supply within Bscl2 cKO myocardium leads to downregulation of the vast amounts of lipid substrates (Figure 3), contributing to the ATP deficit in Bscl2 cKO hearts. In addition, AC levels tightly reflect the FAO rates, and AC profiling has been used to identify FAO dysregulation. 49 Notably, Bscl2 cKO hearts demonstrated no alterations in CPT1 expression, suggesting intact carnitine shuttle ( Figure S3(B)). The almost unanimous reduction of long-chain ACs in Bscl2 cKO hearts highlights the presence of increased FAO which may eventually deplete mitochondrial ACs thus reducing substrates entering TCA cycles and causing energy deficit. While it is plausible that BSCL2 deletion is still causing constant hydrolysis of the remaining TGs and channelling FAs to mitochondria for combustion in a compartmentalised domain in vivo, the energy status of the whole myocardium is ultimately compromised due to the limited availability of substrates. One limitation of our study is that the amount of myocardial ATP was not directly measured by NMR and probably does not reflect the whole myocardium ATP reserve.
Nevertheless, our Bscl2 cKO mice constitute as the first animal model that demonstrates excessive myocardial lipid catabolism associated with deterioration of metabolic reserve and cardiac dysfunction in the absence of elevated lipid uptake.
Treatment with TMZ results in an insignificant increase of skeletal muscle TG content and higher plasma AC in HFD-induced obese mice. 37 However, thus far, no study has examined the effects of TMZ treatment on myocardial lipid and glycogen contents. Surprisingly, we identified TMZ drastically remodels cardiac lipidome by downregulating the lipid contents of almost all lipid classes in normal mouse hearts. Whether this is primary to the cardiac effect of TMZ remains unknown. TMZ improves cardiac function in ischemia/reperfusion injury predominantly by shifting energy production from NEFA to glucose oxidation in the heart. [50][51][52][53] The mechanisms underlying the cardioprotective role of TMZ in our Bscl2 cKO mice could not be simply explained by alleviation of energy deficit, as TMZ failed to significantly improve intracellular ATP levels in Bscl2 cKO hearts. Conversely, TMZ promotes drastic lipid and glycogen accumulation in metabolicstressed Bscl2 cKO hearts. The prominence of glycogen in TMZ-treated Bscl2 cKO hearts may reflect enhanced glucose utilisation as glycogen content has also been shown to commensurate with augmented carbohydrate metabolism such as in GLUT1-overexpressing hearts. 54 Interestingly, aside from glucose utilisation, TMZ has been recently shown to induce cardiac β-hydroxybutyrate flux to attenuate isopropanol-induced rat HF. 55 Whether TMZ indeed induces substrate switch to glucose and ketone bodies in Bscl2 cKO hearts needs to be further dissected by ex vivo heart perfusion experiments. Meanwhile, it is not clear whether TMZ treatment benefits the chronically dysfunctional myocardium by metabolically acting on heart itself or other organs such as liver, skeletal muscle and adipose tissue to ultimately moderate cardiac substrate utilisation. In addition, whether TMZ inhibits cardiac FAO in vivo still remains controversial, 56 and we also cannot exclude the non-metabolic effects of TMZ in preventing left ventricle remodelling independent of its inhibitory activity on FAO. 57,58 Nevertheless, our study underscores that TMZ has potential in ameliorating cardiac function and slows HF progression in a non-ischemic model of HF.
HFD alone has been shown to be cardioprotective especially in alleviating energy-compromised HF. 59 In the present study, we find that decreased cardiac function in Bscl2 cKO mice can be attenuated by HFD feeding. Lipidomics study further confirmed improved cardiac metabolic substrates mainly in the form of TG in HFD-fed Bscl2 cKO mice. This was in line with the notion that HFD feeding provides more coronary circulation of substrates to match up the rate of enhanced lipid utilisation thus atten-uating the myocardial ATP deficit in the energy-deprived failing Bscl2 cKO hearts. In spite of an improvement in cardiac contractile function, we were not able to observe a significant reduction of cardiac stress markers in HFD-treated Bscl2 cKO mice (not shown). Nonetheless, results from our HFD feeding studies support the cardio-protective effect of HFD on the energy-deprived HF.

CONCLUSIONS
Our study highlights an important link between BSCL2 and myocardial energy metabolism and function and advances our understanding of the relationship between TG dynamics, FAO rates and the pathogenesis of HF. It may also provide insights into the therapeutic approaches in the treatment of cardiac disorders related to dysregulated lipid metabolism in general.

A C K N O W L E D G E M E N T S
We thank the Electron Microscopy and Histology Core at Augusta University for technical assistance and electron microscope imaging. This work was supported by National Heart, Lung and Blood Institute (R01HL132182 to W. C.), National Institute of General Medical Sciences (R01GM124108 to J. L.), National Heart, Lung and Blood Institute (R01HL158515 to J. L.) and the American Heart Association Career Development Award (18CDA34080244 to H. Z.).

C O N F L I C T O F I N T E R E S T
None declared.

C O N S E N T F O R P U B L I C AT I O N Not applicable.
O R C I D Weiqin Chen https://orcid.org/0000-0002-3894-5953