Research Article

Imbalanced OPA1 processing and mitochondrial fragmentation cause heart failure in mice

See allHide authors and affiliations

Science  04 Dec 2015:
Vol. 350, Issue 6265, aad0116
DOI: 10.1126/science.aad0116

A change of heart (mitochondria)

Mitochondria provide an essential source of energy to drive cellular processes and are particularly important in heart muscle cells (see the Perspective by Gottlieb and Bernstein). After birth, the availability of oxygen and nutrients to organs and tissues changes. This invokes changes in metabolism. Gong et al. studied the developmental transitions in mouse heart mitochondria soon after birth. Mitochondria were replaced wholesale via mitophagy in cardiomyocytes over the first 3 weeks after birth. Preventing this turnover by interfering with parkin-mediated mitophagy specifically in cardiomyocytes prevented the normal metabolic transition and caused heart failure. Thus, the heart has coopted a quality-control pathway to facilitate a major developmental transition after birth. Wai et al. examined the role of mitochondrial fission and fusion in mouse cardiomyocytes. Disruption of these processes led to “middle-aged” death from a form of dilated cardiomyopathy. Mice destined to develop cardiomyopathy were protected by feeding with a high-fat diet, which altered cardiac metabolism.

Science, this issue p. 10.1126/science.aad2459, p. 10.1126/science.aad0116; see also p. 1162

Structured Abstract

INTRODUCTION

Mitochondria are essential organelles whose form and function are inextricably linked. Balanced fusion and fission events shape mitochondria to meet metabolic demands and to ensure removal of damaged organelles. A fragmentation of the mitochondrial network occurs in response to cellular stress and is observed in a wide variety of disease conditions, including heart failure, neurodegenerative disorders, cancer, and obesity. However, the physiological relevance of stress-induced mitochondrial fragmentation remains unclear.

RATIONALE

Proteolytic processing of the dynamin-like guanosine triphosphatase (GTPase) OPA1 in the inner membrane of mitochondria is emerging as a critical regulatory step to balance mitochondrial fusion and fission. Two mitochondrial proteases, OMA1 and the AAA protease YME1L, cleave OPA1 from long (L-OPA1) to short (S-OPA1) forms. L-OPA1 is required for mitochondrial fusion, but S-OPA1 is not, although accumulation of S-OPA1 in excess accelerates fission. In cultured mammalian cells, stress conditions activate OMA1, which cleaves L-OPA1 and inhibits mitochondrial fusion resulting in mitochondrial fragmentation. In this study, we generated conditional mouse models for both YME1L and OMA1 and examined the role of OPA1 processing and mitochondrial fragmentation in the heart, a metabolically demanding organ that depends critically on mitochondrial functions.

RESULTS

Deletion of Yme1l in cardiomyocytes did not grossly affect mitochondrial respiration but induced the proteolytic cleavage of OPA1 by the stress-activated peptidase OMA1 and drove fragmentation of mitochondria in vivo. These mice suffered from dilated cardiomyopathy characterized by well-established features of heart failure that include necrotic cell death, fibrosis and ventricular remodelling, and a metabolic switch away from fatty acid oxidation and toward glucose use. We discovered that additional deletion of Oma1 in cardiomyocytes prevented OPA1 processing altogether and restored normal mitochondrial morphology and cardiac health. On the other hand, mice lacking YME1L in both skeletal muscle and cardiomyocytes exhibited normal cardiac function and life span despite mitochondrial fragmentation in cardiomyocytes. Imbalanced OPA1 processing in skeletal muscle, which is an insulin signaling tissue, induced systemic glucose intolerance and prevented cardiac glucose overload and cardiomyopathy. We observed a similar effect on cardiac metabolism upon feeding mice lacking Yme1l in cardiomyocytes a high-fat diet, which preserved heart function despite mitochondrial fragmentation.

CONCLUSION

Our work highlights the importance of balanced fusion and fission of mitochondria for cardiac function and unravels an intriguing link between mitochondrial dynamics and cardiac metabolism in the adult heart in vivo. Mitochondrial fusion mediated by L-OPA1 preserves cardiac function, whereas its stress-induced processing by OMA1 and mitochondrial fragmentation triggers dilated cardiomyopathy and heart failure. In contrast to previous genetic models of the mitochondrial fusion machinery, mice lacking Yme1l in cardiomyocytes do not show pleiotropic respiratory deficiencies and thus provide a tool to directly assess the physiological importance of mitochondrial dynamics. Preventing mitochondrial fragmentation by deleting Oma1 protects against cell death and heart failure. The identification of OMA1 as a critical regulator of mitochondrial morphology and cardiomyocyte survival holds promise for translational applications in cardiovascular medicine. Mitochondrial fragmentation induces a metabolic switch from fatty acid to glucose utilization in the heart. It turns out that reversing this switch and restoring normal cardiac metabolism is sufficient to preserve heart function despite mitochondrial fragmentation. These findings raise the intriguing possibility that the switch in fuel usage that occurs in the failing adult heart may, in fact, be maladaptive and could contribute to the pathogenesis of heart failure.

Critical role of balanced mitochondrial fusion and fission for cardiac metabolism and heart function.

Induced processing of the dynamin-like GTPase OPA1 in the inner membrane by the stress-activated peptidase OMA1 leads to mitochondrial fragmentation, cardiomyopathy, and heart failure, which is characterized by a switch in fuel utilization. Heart function can be preserved by reversing this metabolic switch without suppressing mitochondrial fragmentation.

Abstract

Mitochondrial morphology is shaped by fusion and division of their membranes. Here, we found that adult myocardial function depends on balanced mitochondrial fusion and fission, maintained by processing of the dynamin-like guanosine triphosphatase OPA1 by the mitochondrial peptidases YME1L and OMA1. Cardiac-specific ablation of Yme1l in mice activated OMA1 and accelerated OPA1 proteolysis, which triggered mitochondrial fragmentation and altered cardiac metabolism. This caused dilated cardiomyopathy and heart failure. Cardiac function and mitochondrial morphology were rescued by Oma1 deletion, which prevented OPA1 cleavage. Feeding mice a high-fat diet or ablating Yme1l in skeletal muscle restored cardiac metabolism and preserved heart function without suppressing mitochondrial fragmentation. Thus, unprocessed OPA1 is sufficient to maintain heart function, OMA1 is a critical regulator of cardiomyocyte survival, and mitochondrial morphology and cardiac metabolism are intimately linked.

The dynamic behavior of mitochondria preserves mitochondrial integrity and distribution and allows mitochondrial shape and function to be adapted to altered physiological demands (1, 2). Disturbed mitochondrial dynamics is associated with a number of neurodegenerative disorders and cardiac hypertrophy in mice (3, 4). Dynamin-like guanosine triphosphatases (GTPases) mediate the fusion and fission of mitochondrial membranes. Mitofusins 1 and 2 (MFN1 and MFN2) orchestrate outer mitochondrial membrane fusion, whereas OPA1 is required for inner mitochondrial membrane fusion. Fission, on the other hand, is executed by dynamin-related protein 1 (DRP1), a cytosolic protein that is recruited to the mitochondrial surface in response to various physiological cues. This complex machinery, including DRP1-specific receptor proteins and cytoskeletal components, assembles at contact sites between the mitochondria and the endoplasmic reticulum, which mark mitochondrial division sites (5, 6).

Fusion and fission of mitochondrial membranes occur in a coordinated manner. Balanced cycles of fusion and fission determine the shape, size, and number of mitochondria, which leads to a large variability in the morphology of mitochondria in different cell types. Although mitochondria form interconnected, tubular networks in cultured fibroblasts, they appear as distinct entities in tissues, such as heart and skeletal muscle, that are characterized by low fusion and fission rates (7). Moreover, coordinated mitochondrial dynamics is critical for the bioenergetic function of mitochondria and is closely linked to metabolism. Changes in mitochondrial ultrastructure and dynamics occur in response to altered metabolic demands (811), and components involved in mitochondrial fusion are central regulators of cellular metabolism (12). Coordinated fusion and fission events are crucial for mitochondrial quality-control. Fusion contributes to mitochondrial maintenance, whereas excessive fission causes mitochondrial fragmentation, which allows removal of irreversibly damaged mitochondria by mitophagy and is associated with cell death (13, 14). Fragmentation of the mitochondrial network is observed in a wide variety of diseases.

The dynamin-like GTPase OPA1 mediates mitochondrial fusion and orchestrates mitochondrial cristae morphogenesis and resistance to apoptosis in response to physiological demands (1517). The processing of OPA1 is emerging as a central regulatory step coordinating fusion and fission of mitochondria (18, 19). Two peptidases in the inner membrane, OMA1 and the i-AAA protease YME1L, convert long OPA1 forms (L-OPA1) into short forms (S-OPA1) (2023). The balanced accumulation of both forms maintains normal mitochondrial morphology: Fusion depends on L-OPA1 only, whereas S-OPA1 is associated with mitochondrial fission (Fig. 1A) (2426). Cellular stress, mitochondrial dysfunction, or genetic interventions (such as deletion of Yme1l) can activate OMA1, which results in the increased conversion of L-OPA1 into S-OPA1 and mitochondrial fragmentation (25, 2730). Loss of Yme1l in cultured fibroblasts does not impair fusion but triggers mitochondrial fragmentation (22, 25, 31), which can be suppressed by deletion of Oma1 (22, 25, 31). Thus, although OPA1 processing is dispensable for mitochondrial fusion per se, an increased oxidative phosphorylation promotes cleavage of OPA1 by YME1L (10). It thus appears that different stimuli modulate OPA1 processing by YME1L or OMA1, which allows the coordination of mitochondrial fusion and division under various physiological conditions.

Fig. 1 YME1L is required in the developing embryo and the adult heart.

(A) The mitochondrial proteases OMA1 and YME1L cleave L-OPA1 (a and b) at S1 and S2, respectively, to yield S-OPA1 forms (c, d, and e). (B) No viable Yme1l–/– mice were recovered from intercrosses of Yme1l+/– mice (0 out of 289 offspring). Chi-squared test, ****P < 0.0001. (C) Postimplantation developmental delay of Yme1l–/– embryos scaled relative to WT. Scale bar, 2 mm. (D) Life span of cardiomyocyte-specific cYKO mice (Myh6-Cre red; median of 46 weeks, n = 69) is reduced relative to WT littermates (green; n = 74). Log-rank (Mantel-Cox) test, ****P < 0.0001. (E) Immunoblots of tissues isolated from 18-week-old WT and cYKO mice. Antibodies directed against succinate dehydrogenase subunit A (SDHA) were used to control for gel loading. (F) Mean body weight (g) of cYKO males (red; n = 30) declines relative to WT (green; n = 30). Multiple t test, *P < 0.05, ****P < 0.0001. Data are means ± SEM.

In agreement with its role for stress-induced OPA1 processing, ablation of Oma1 in mice causes impaired thermogenesis and diet-induced obesity and protects against ischemic kidney injury (29, 32). Here, we generated tissue-specific mouse models for the OPA1-processing peptidases YME1L and OMA1 and examined the role of OPA1 processing in myocardial function.

Results

YME1L is essential for embryonic development

To study the importance of balanced mitochondrial dynamics (Fig. 1A), we generated conditional mouse models of the OPA1-processing peptidases Yme1l and Oma1 (fig. S1, A to D, and table S1). We used a mouse line expressing Cre recombinase under the control of the β-actin promoter to delete Yme1l or Oma1 by Cre/loxP-mediated recombination in all tissues. As expected (29), Oma1–/– mice were born at the expected Mendelian ratio (fig. S1E). Yme1l+/– mice were viable and exhibited no obvious phenotypes, but heterozygous intercrosses did not yield viable null offspring (Fig. 1B). We observed a generalized developmental delay in Yme1l–/– embryos isolated from embryonic day 8.5 (E8.5) to E12.5 (Fig. 1C). Hearts from Yme1l–/– embryos isolated at E9.5 and E10.5 failed to beat properly, and we did not recover any null embryos after E13.5. Thus, YME1L is essential for embryogenesis.

Cardiomyocyte-specific deletion of Yme1l causes dilated cardiomyopathy

We next examined the requirement of YME1L for the function of the heart, a metabolically demanding organ sensitive to disruption of mitochondrial shape (7, 33). We crossed Yme1lLoxP/LoxP mice to mice expressing Cre recombinase specifically in cardiomyocytes (Myh6-Cre; cYKO) (34). cYKO mice were viable but had a significantly shortened life span (median life span: 46 weeks) (Fig. 1, D and E) punctuated by weight loss before their demise (Fig. 1F), which suggested that YME1L is required for normal heart function.

We examined heart function from an early age up to 40 weeks of age in cYKO mice (Fig. 2A). Longitudinal echocardiographic (echo) analyses (fig. S2A) revealed progressive cardiac dysfunction (table S2), which became apparent at 20 weeks and was characterized by hallmarks of dilated cardiomyopathy (DCM): a reduced percentage of LVEF (% LVEF) (Fig. 2B and fig. S2A), a dilated left ventricular chamber (Fig. 2C), and a preserved left ventricular mass (fig. S2B). We observed myocardial fibrosis (Fig. 2D), increased serum levels of cardiac troponin T (Fig. 2E), and evidence of ongoing necrotic cell death (Fig. 2, F and G) in cYKO mice.

Fig. 2 Deletion of Yme1l causes dilated cardiomyopathy and heart failure.

(A to C) Echocardiographic evaluation of cardiac function (by M-mode) of randomized 40-week-old WT (n = 8) and cYKO (n = 8) males reveals DCM characterized by (B) reduced LVEF (****P < 0.0001) and (C) increased left ventricular internal dimension (LVID) [d (mm), **P = 0.0154]. (D) Cardiac fibrosis in cYKO mice monitored by trichrome and sirius red staining of heart sections [40 weeks old (40w), n = 3; scale bar, 200 μm]. (E) Increased serum levels of cardiac troponin (cTNI ng/ml) in 30-week-old cYKO mice (n = 8) relative to WT (n = 19). Mann-Whitney test, **P = 0.0027. (F and G) Cardiomyocyte necrosis analysis of 20-week-old cYKO (n = 4) and WT (n = 4) hearts stained with Evans Blue (EB red), wheat germ agglutinin (green), and DAPI (blue). Individual t test, *P = 0.0286; scale bar, 50 μm. (H) PET-CT of 40-week-old cYKO (n = 8) and WT (n = 7) animals after [18F]FDG injections. Representative images of four cYKO and four WT thoracic scans are shown. (I) Average standardized uptake value (SUV) in WT (n = 7) and cYKO hearts (n = 8); unpaired t test, *P = 0.0140. In graphs, data are means ± SEM.

The failing adult heart is commonly characterized by an altered metabolism where glucose use is increased and β oxidation is decreased (35, 36). We monitored in vivo cardiac uptake of 18fluorodeoxyglucose ([18F]FDG) in cYKO mice by hybrid positron emission tomography–computed tomography (PET-CT). The loss of YME1L in cardiomyocytes caused an increase in in vivo cardiac glucose uptake (Fig. 2, H and I) and in vitro glycolysis rates (fig. S2C). Gas chromatography–mass spectrometry analyses revealed increased endogenous glucose levels and decreased lactate levels in cYKO hearts, whereas the levels of citric acid cycle intermediates were not altered (fig. S2D) nor were the levels and use of pyruvate (fig. S2E). However, we observed a global reduction of total cardiac acylcarnitines (fig. S2F), which indicated reduced β oxidation in YME1L-deficient cardiomyocytes. Thus, loss of YME1L in cardiomyocyte mitochondria can induce a metabolic shift from lipid utilization to carbohydrate utilization that is typically observed in the failing heart (37). In conclusion, cYKO mice develop DCM, which progresses to heart failure and middle-aged death.

Loss of YME1L impairs mitochondrial morphology in cardiomyocytes

To define the molecular basis of DCM in cYKO mice, we first analyzed mitochondrial respiration. Ex vivo cardiac respiration measurements revealed no differences between resting hearts isolated from cYKO mice and controls (Fig. 3A). Specific activities of mitochondrial complexes II, III, and IV were increased, although we observed only moderately impaired adenosine 5′-triphosphate (ATP) synthesis by complex V in cYKO hearts (Fig. 3B) and no significant differences in the assembly of respiratory chain complexes and supercomplexes (fig. S3A). Respiratory deficiencies thus appeared unlikely to be the major cause for DCM in cYKO mice.

Fig. 3 Stress-induced OPA1 processing in cardiomyocytes perturbs mitochondrial morphology.

(A) Ex vivo respiration measured in resting hearts (n = 3 to 5) from 26- and 44-week-old WT (white or pale gray) and cYKO (black or dark gray) mice. Linear oxygen uptake rates are presented as nmol O2/min per mg (weight, heart weight). Differences were not significant. (B) Respiratory chain activity measurements of complex I to V (CI to CV) in 44-week-old hearts of WT (white) and cYKO (black) mice. Individual unpaired t tests (n = 3 to 5); CI (P = 0.1442), CII (***P = 0.0003), CIII (***P = 0.0003), CIV (**P = 0.0020), and CV (*P = 0.0172). (C) TEM of 20-week-old WT and cYKO hearts (thick scale bar, 500 nm; thin scale bar, 100 nm). (D) Mitochondrial size was represented as median surface area, and frequency distributions of mitochondrial surface were calculated from 20-week-old WT (n = 4224) and cYKO (n = 2308) mitochondria imaged by TEM. Kruskal-Wallis test, ****P < 0.0001. Data are median values. (E) Indirect immunocytochemistry with antibodies directed against TOMM20 (a rabbit-specific antibody) in cardiomyocytes isolated from WT (n = 3) and cYKO (n = 3) mice (thin scale bar, 30 μm; thick scale bar, 7.5 μm). 40w, 40 weeks old. (F) Mitochondrial morphology in WT (8w, n = 2; 40w, n = 3) and cYKO (8w, n = 2; 40w, n = 3) cardiomyocytes (****P < 0.0001). Cells (>100) were counted. 8w, 8 weeks old. (G) Immunoblot analysis of cardiomyocytes isolated from 40-week-old mice fed a normal chow diet (NCD) or high-fat diet (HFD). SDHA was used as a loading control. (H) Quantification of OPA1 processing in hearts of WT and cYKO mice (Fig. 3F). (Pairwise t test, *P < 0.05, **P < 0.01) relative to WT-NCD controls. In graphs (B), (F), and (H), data are means ± SEM.

We next examined the morphology of mitochondria and performed transmission electron microscopy (TEM) of cYKO hearts (Fig. 3C). Smaller mitochondria with normal architecture of cristae accumulated in the absence of YME1L, which indicated impaired mitochondrial dynamics (Fig. 3, C and D). Similar results were seen in primary adult cardiomyocytes isolated from 8-week-old (before DCM development) and 40-week-old cYKO hearts (Fig. 3, E and F, and fig. S3B). Consistent with our TEM data, we observed distorted mitochondrial morphology in Yme1l–/– cardiomyocytes, as seen previously in noncardiac cell types lacking YME1L (20, 22, 25). Loss of YME1L in cardiomyocytes abolished formation of S-OPA1 form d and led to the accumulation of S-OPA1 forms c and e, which are generated by OMA1 (Fig. 3, G and H). Moreover, the mitochondrial lipid transfer protein PRELID1, normally degraded by YME1L, accumulated (Fig. 3G) (38). Notably, Yme1l is specifically deleted in cardiomyocytes and was not lost in cardiac fibroblasts isolated from cYKO mice. However, in vitro deletion of Yme1l in adult cardiac fibroblasts did recapitulate fragmentation of the mitochondrial network (fig. S3, C and D) and impaired OPA1 processing (fig. S3E) as observed in Yme1l–/– cardiomyocytes. Thus, YME1L deficiency in cardiomyocytes induces OPA1 processing and mitochondrial fragmentation, which raises the possibility that disturbed mitochondrial morphology could cause heart failure in cYKO mice.

Deletion of Oma1 restores mitochondrial morphology and myocardial function in the absence of YME1L

The accumulation of OPA1 forms c and e in cardiomyocytes lacking YME1L indicated activation of OMA1, as previously observed in YME1L-deficient mouse embryonic fibroblasts (MEFs) in vitro (25). Because additional deletion of Oma1 in Yme1l–/– MEFs restores tubular mitochondria and apoptotic resistance (25), we reasoned that ablation of Oma1 may preserve the mitochondrial network in cardiomyocytes lacking YME1L. To examine the role of disturbed mitochondrial morphology in DCM and heart failure in cYKO mice, we generated double-knockout mice lacking both YME1L and OMA1 specifically in cardiomyocytes (cDKO; Myh6-Cretg/wtYme1lLoxP/LoxP Oma1LoxP/LoxP) and monitored heart function (Fig. 4A). In contrast to cYKO mice, cDKO mice showed normal cardiac function and normal exercise tolerance in treadmill tests (Fig. 4, B to D). Myocardial fibrosis present in cYKO mice (Fig. 2E) was absent in cDKO hearts (Fig. 4E). Moreover, TEM analysis of cDKO hearts revealed that mitochondrial fragmentation was largely suppressed (Fig. 4, F and G). Similarly, mitochondrial morphology was restored in primary adult cardiomyocytes isolated from cDKO mice (Fig. 4, H and I). L-OPA1 processing was prevented in these cells, whereas other YME1L substrates such as PRELID1 continued to accumulate to similar levels as in cardiomyocytes of cYKO mice (Fig. 4J).

Fig. 4 Oma1 ablation restores mitochondrial morphology and protects cYKO mice against DCM and heart failure.

(A) Echocardiographic evaluation of cardiac function (by M-mode) of 22-week-old WT and cDKO mice. (B and C) Percentage LVEF and (C) diastolic LVID of 22-week-old WT (n = 10) and cDKO (n = 9) mice. n.s., not significant. (D) Treadmill endurance of 20-week-old WT (n = 10), cYKO (n = 10), and cDKO (n = 5) mice (5% incline). cYKO mice versus WT, ****P = 0.0003; cYKO versus cDKO, ****P = 0.0001. (E) Suppression of cardiac fibrosis in cDKO mice. Trichrome and sirius red staining of heart sections of 22-week-old WT and cDKO mice (n = 3). Scale bar, 200 μm. (F) TEM of 20-week-old cYKO and cDKO hearts (thick scale bar, 500 nm). (G) Mitochondrial size represented as median surface area and frequency distributions of mitochondrial surface calculated from 20-week-old cYKO (n = 2308) and cDKO (n = 3122) mitochondria imaged by TEM. Kruskal-Wallis test, ****P < 0.0001. Data are median values. (H) Indirect immunocytochemistry with TOMM20-specific antibodies in cardiomyocytes isolated from 22-week-old WT and cDKO mice (thin scale bar, 30 μm; thick scale bar, 7.5 μm). (I) Quantification of mitochondrial morphology from (H) (n = 3, n > 100 cells; *P = 0.0112). In (B) to (D) and (I), data are means ± SEM. (J) Immunoblot analysis of cardiomyocytes from 22-week-old WT (n = 3) and cDKO (n = 3) mice. SDHA was used as a loading control.

Thus, YME1L ablation in cardiomyocytes activates OMA1 and promotes OPA1 processing and mitochondrial fragmentation, which causes DCM and heart failure.

Loss of YME1L in skeletal muscle preserves the function of YME1L-deficient hearts

The results obtained from cardiomyocyte-specific knockout mice establish an essential role of YME1L for normal cardiac function in vivo. We observed, to our surprise, the normal life span of mice lacking YME1L both in cardiomyocytes and skeletal muscle (hmYKO for heart and muscle–specific YME1L knockout; median life span 125 weeks) (Fig. 5A, fig. S5A, and table S1). hmYKO mice were obtained by crossing Yme1lLoxP/LoxP mice to mice expressing Cre recombinase under the control of the muscle creatine kinase (Ckmm) promoter, which is active in cardiomyocytes and additionally in skeletal muscle (39). Note that differences in life span could not be explained by differences in the efficiency of Yme1l deletion: mRNA (fig. S5B) and protein (fig. S5C) levels were profoundly depleted in adult hearts of both cYKO and hmYKO mice. YME1L was lost with similar efficiencies and kinetics upon Cre recombinase–mediated deletion of Yme1l both in postnatal hearts of cYKO and hmYKO mice (fig. S5, D to G). Unlike cYKO mice, however, myocardial activity was preserved in hmYKO mice. We observed normal heart function by echocardiography and normal cardiac uptake of [18F]FDG in these mice (Fig. 5, B to D, and fig. S5, H and I).

Fig. 5 Deletion of Yme1l in skeletal muscle is cardioprotective.

(A) Life span of WT (green, n = 91) and hmYKO (purple, n = 75) mice are not significantly different (n.s.) but significantly increased relative to cYKO mice (red, n = 69; log-rank Mantel-Cox test, ****P < 0.0001). (B and C) Percentage LVEF and diastolic LVID in 46-week-old WT (n = 5) and hmYKO (n = 6) mice analyzed by echocardiography (fig. S5A). (D) PET-CT average standardized cardiac glucose uptake in 46-week-old WT (green, n = 8) and hmYKO (purple, n = 5) mice fed normal chow. (E) TEM of 20-week-old WT and hmYKO hearts (thick scale bar, 500 nm). (F) Mitochondrial size represented as median surface area and frequency distributions of mitochondrial surface calculated from 20-week-old WT (n = 4224) (Fig. 3, C and D) and hmYKO (n = 3246) mitochondria imaged by TEM. Kruskal-Wallis test, ****P < 0.0001. Data are median values. (G) Immunoblot analysis of lysates from heart skeletal muscle and liver of 18-week-old hmYKO mice (n = 3), hmOKO mice (Ckmm-Cre; Oma1LoxP/LoxP, n = 3), and hmDKO mice (Ckmm-Cre; Oma1LoxP/LoxP Yme1lLoxP/LoxP, n = 3) reveals altered OPA1 processing. (H) Intraperitoneal glucose tolerance tests in 18-week-old WT (n = 18) and hmYKO (n = 13) mice. Two-way ANOVA (**P < 0.01, ***P < 0.001, ****P < 0.0001) relative to WT controls. (I) Fasting insulin levels in 20-week-old WT (n = 6) and hmYKO (n = 7) mice. (*P = 0.0357). In (A), (C), (F), (H), and (I), data are means ± SEM.

We analyzed the morphology of mitochondria in hmYKO hearts by TEM. Smaller mitochondria accumulated in hearts of hmYKO mice as observed in cYKO mice, which indicated mitochondrial fragmentation (Fig. 5, E and F). These morphological changes corresponded to defects in OPA1 processing that were similar in cardiomyocytes isolated from hmYKO and cYKO mice (Figs. 3G and 5G). In both models, deletion of Yme1l in cardiomyocytes prevented formation of S-OPA1 form d, whereas S-OPA1 forms c and e accumulated, which was indicative of OMA1 activation (fig. S5F). OPA1 processing was affected similarly upon loss of YME1L in skeletal muscle of hmYKO mice (Fig. 5G) but normal in skeletal muscle of cYKO mice harboring YME1L (Fig. 1E). Thus, loss of YME1L impairs OPA1 processing and induces mitochondrial fragmentation in cardiomyocytes of both cYKO and hmYKO mice. Furthermore, additional deletion of Yme1l in skeletal muscle maintains heart function and life span without restoring mitochondrial morphology defects in cardiomyocytes lacking YME1L.

Mitochondrial dysfunction in skeletal muscle is associated with impaired insulin signaling and glucose intolerance (4042). Thus, possible endocrine effects owing to the loss of YME1L in skeletal muscle may cause metabolic alterations in cardiomyocytes; this preserves heart function downstream of mitochondrial deficiencies. We thus investigated systemic glucose homeostasis and performed intraperitoneal glucose tolerance tests (GTT) in both hmYKO and cYKO mice (Figs. 5H and 6A). We observed glucose intolerance in hmYKO mice but not in cYKO mice (Figs. 5H and 6A), which suggested that deletion of Yme1l in skeletal muscle impaired glucose homeostasis systemically. Additional ablation of Oma1 in hmYKO mice (hmDKO; table S1) prevented OPA1 processing in both heart and skeletal muscle (Fig. 5G) and restored normal glucose tolerance (fig. S5F). Thus, stress-induced OPA1 processing by OMA1 in skeletal muscle impairs glucose homeostasis in hmYKO mice.

Fig. 6 Suppression of DCM and heart failure by dietary intervention.

(A) Intraperitoneal glucose tolerance tests in WT (green, n = 30; blue, n = 13) and cYKO (orange, n = 5; red, n = 14) mice fed HFD or NCD. Two-way ANOVA (*P < 0. 05, **P < 0.01, ***P < 0.001, ****P < 0.0001) relative to WT-NCD controls. (B) PET-CT in 30-week-old WT and cYKO mice treated with HFD starting at 9 weeks of age. Representative images of 2 cYKO and 2 WT mice are shown. (C) Average standardized cardiac glucose uptake in 30-week-old WT (n = 8; blue) and cYKO mice (n = 14; orange) treated with HFD starting at 9 weeks of age. (D) Echocardiographic M-mode images of 30-week-old WT and cYKO mice fed with HFD starting at 9 weeks of age. (E and F) Percentage LVEF and diastolic LVID of WT (n = 14) and cYKO (n = 10) mice. n.s., not significant. (G) Trichrome and sirius red stainings of heart sections of 40-week-old HFD-fed WT (n = 3) and cYKO (n = 3) mice. Scale bar, 200 μm. (H) Indirect immunocytochemistry using antibodies directed against TOMM20 in cardiomyocytes isolated from 30-week-old HFD-fed WT (n = 3) and cYKO (n = 3) mice (thin scale bar, 30 μm; thick scale bar, 7.5 μm). OPA1 processing in these cells is shown in Fig. 3G. (I) Quantification of mitochondrial morphology from (H) (WT, n ≥ 100 cells; cYKO, n = 3; ***P = 0.001). In graphs, data are means ± SEM.

Deletion of Pgc1α in skeletal muscle significantly impairs glucose-stimulated insulin secretion, which suggests a cytokine-mediated cross-talk between skeletal muscle and pancreatic islets (41). In agreement with these findings, we observed lowered fasting insulin concentration in the serum of hmYKO mice (Fig. 5I), although Yme1l was not deleted in the pancreas of these mice. hmYKO mice had normal fasting blood glucose (fig. S5G), normal weight gains (fig. S5H), and body composition as normal lean and fat mass (fig. S5I).

Thus, loss of YME1L in skeletal muscle induces systemic glucose intolerance and lowers insulin levels, which blunts increased cardiac glucose uptake and alters cardiac metabolism. Furthermore, cell nonautonomous metabolic alleviation can preserve cardiac function and prevent DCM and heart failure despite mitochondrial fragmentation in cardiomyocytes.

Feeding a high-fat diet suppresses heart failure and restores the life span of cYKO mice

The protective effect of systemic glucose intolerance on the heart suggests that deleterious effects of mitochondrial fragmentation in cardiomyocytes can be circumvented by metabolic intervention. To provide further support for this notion, we subjected wild-type (WT) and cYKO mice to a high-fat diet. This diet is commonly used to dysregulate systemic glucose homeostasis. It impairs insulin signaling in target tissues and compromises glucose-stimulated insulin secretion by pancreatic β cells triggering obesity. Mice were fed a high-fat diet beginning at 9 weeks of age, at a time when cardiac function was normal. Both WT and cYKO mice fed a high-fat diet gained weight significantly more rapidly than mice fed normal chow (fig. S6A) and exhibited reduced glucose tolerance (Fig. 6A).

To determine whether this dietary intervention could influence cardiac metabolism, we examined the cardiac uptake of [18F]FDG by PET-CT and determined levels of endogenous glucose and acylcarnitine in the hearts of high fat–fed cYKO mice. In contrast to mice fed normal chow, we did not observe significant differences in cardiac glucose uptake (Fig. 6, B and C) nor in the levels of endogenous cardiac glucose or acylcarnitine between high fat–fed WT and cYKO mice (fig. S2, D and F). The adjustment of the levels of these cardiac metabolites was accompanied by restoration of cardiac function in cYKO mice (Fig. 6, D to F, and fig. S5B): left ventricular ejection fraction (LVEF) and left ventricular chamber diameter values were indistinguishable from those of high fat–fed littermate controls. Treatment with the high-fat diet also prevented cardiac fibrosis (Fig. 6G) and suppressed differences in exercise tolerance previously observed between WT and cYKO mice fed normal chow (fig. S6, C and D).

Similar to normally fed cYKO mice (Fig. 3, F and G), cardiomyocytes isolated from high fat–fed cYKO mice still contained distorted mitochondria (Fig. 6, H and I), because the high-fat diet had not rescued the proteolytic activity of YME1L or OMA1-dependent, stress-induced processing of OPA1 (Fig. 3G, H). High-fat feeding did not markedly alter oxygen consumption or the activity of respiratory complexes in cYKO mice (fig. S6, E and F). Thus, the consequences of mitochondrial defects in cYKO mice can be metabolically circumvented to suppress cardiomyopathy.

Discussion

In these experiments, we observed the deleterious effects of stress-induced OPA1 processing and mitochondrial fragmentation on myocardial function, which revealed an unexpected functional link between systemic glucose homeostasis, cardiac metabolism, and mitochondrial dynamics in vivo.

Uncleaved, fusion-active L-OPA1 is sufficient to maintain cardiac activity. Mice lacking both YME1L and OMA1 in cardiomyocytes exhibited normal heart function, which demonstrates that proteolytic cleavage of OPA1 by YME1L and OMA1 is dispensable. The previously described, essential role of OPA1 for normal cardiac functioning can thus be attributed to the loss of L-OPA1 (43, 44). L-OPA1 is sufficient to mediate mitochondrial fusion, which serves a prosurvival function (2426). Mitochondrial fusion protects against mitophagy (4, 45) and is thought to serve a repair function by allowing content mixing and by preventing the accumulation of mitochondrial damage in cultured cells (46). Although mitochondrial fusion occurs infrequently in adult cardiomyocytes (7), it is required for cardiomyocyte differentiation and cardiac development (7, 4751).

Whereas L-OPA1 is sufficient to preserve cardiac function, accumulation of S-OPA1 and unopposed fission is deleterious for the heart. Our results establish cardiomyocyte-specific YME1L-deficient mice as a model for DCM and heart failure, which culminates in middle-aged death (Fig. 7). The loss of YME1L activates OMA1 and triggers stress-induced OPA1 processing, which unbalances fusion and fission of mitochondria and impairs mitochondrial morphology in cardiomyocytes. In the absence of YME1L, we observed the accumulation of smaller mitochondria in the heart and mitochondrial fragmentation in cardiomyocytes and cardiac fibroblasts in vitro. Mitochondrial fragmentation is caused by the loss of L-OPA1 forms mediating fusion and the concomitant accumulation of S-OPA1 forms c and e that are generated by OMA1 and are associated with fission (21, 2326). Deletion of Oma1 restores normal mitochondrial morphology in cardiomyocytes lacking YME1L and myocardial activity in vivo, which demonstrates that accelerated OPA1 processing and mitochondrial fragmentation cause heart failure (Fig. 7). Consistent with a deleterious effect of unopposed fission in the heart (7), pharmacological inhibition of mitochondrial fission protects against ischemia and reperfusion injury (52). Notably, germline deletion of Oma1 does not impair embryogenesis and is not able to suppress postimplantation embryonic lethality of Yme1l–/– embryos (fig. S1F), which indicates that stress-induced OPA1 processing is not deleterious for prenatal organogenesis in these mice.

Fig. 7 Unbalanced mitochondrial dynamics in cardiomyocytes upon loss of YME1L causes DCM and heart failure.

Unregulated OPA1 processing by OMA1 causes metabolic alterations triggering DCM and heart failure in cardiomyocyte-specific Yme1l–/– mice. Heart function is preserved upon restoration of mitochondrial morphology by Oma1 ablation or without suppressing mitochondrial morphology defects by metabolic intervention bypassing deleterious effects of disturbed mitochondrial dynamics on cardiac metabolism.

How does stress-induced OPA1 processing and mitochondrial fragmentation affect cardiac function? Mitochondria are vital for the beating heart, and defects in mitochondrial respiration cause cardiac dysfunction (36, 53, 54). However, we observed only a minor impairment of respiratory activities and ATP synthesis and no accumulation of lactate in YME1L-deficient hearts, unlike other animal models of mitochondrial cardiomyopathy (7, 36, 51, 54). Whereas severe and lethal cardiomyopathies manifest not long after birth in most of these models, cardiac ablation of YME1L causes DCM and death at ~1 year of age.

Unrestrained autophagy can also cause cardiomyocyte loss and heart failure (55), which raises the possibility that stress-induced OPA1 processing and the accumulation of S-OPA1 affects the autophagic disposal of mitochondria. The analysis of heart-specific DRP1 knockout mice indeed pointed to a major role of DRP1 and fission for mitochondrial quality-control and autophagy in the heart (5658). However, autophagic marker proteins such as p62/SQSTM1 or microtubule-associated protein 1 light chain 3 did not accumulate in YME1L-deficient heart nor were myocardial amino acid levels altered (fig. S7, A and B). Thus, increased autophagy is unlikely to cause the loss of cardiomyocytes in this model. In contrast, our results suggest that stress-induced OPA1 processing by OMA1 promotes cardiomyocyte death in the absence of YME1L (fig. S7, C and D). We observed increased serum levels of cardiac troponin, fibrotic remodeling, and necrotic cell death, as well as the dysregulation of genes functionally linked to cell death in YME1L-deficient hearts (fig. S7, C and D). These results identify OMA1 as a critical regulator of cardiomyocyte survival in vivo, consistent with an antiapoptotic effect of OMA1 in cultured cells (21, 25, 29). Note that cardiac mitochondria form normal cristae in the absence of YME1L. Thus, cristae remodeling and facilitated cytochrome c release does not drive cell death in these mice, which instead is triggered by the loss of L-OPA1 and the impairment of mitochondrial fusion. In light of these observations, it is conceivable that the recently reported, protective effect of OPA1 overexpression in various mouse models for mitochondrial disease (59, 60) can be attributed to L-OPA1 alone, which drives fusion and can support cell survival independent of cristae morphogenesis.

Similar to previous models for the failing heart (36, 61), the cardiomyocyte-specific loss of YME1L and disturbed mitochondrial dynamics provoked a downstream metabolic shift from lipid to glucose metabolism in the myocardium. We observed reduced levels of acylcarnitine and increased glucose uptake that indicated reinforced myocardial glucose utilization, which is known to be associated with heart failure when combined with disturbed fatty acid metabolism (37). These alterations occur in the absence of overt respiratory deficiencies, which were previously observed to enhance cardiac glucose metabolism (36, 61), which highlights the regulatory role of mitochondrial dynamics in cardiac metabolism. Increasing evidence indeed supports a close link between energy metabolism and mitochondrial fusion and fission in various tissues, including the central nervous system and brown adipose tissue, and suggests an association between mitochondrial fission, lipid metabolism, and energy expenditure (6264).

Our experiments provide strong evidence that alterations in metabolism cause heart failure in cYKO mice. Additional deletion of Yme1l in skeletal muscle preserved cardiac function and normal life span of mice lacking YME1L in cardiomyocytes without restoring defects in mitochondrial morphology (Fig. 7). This is likely due to an endocrine effect by the skeletal muscle on glucose uptake in the heart. Dysfunction of mitochondria in skeletal muscle was previously reported to affect systemic glucose metabolism by impairing insulin secretion by pancreatic β cells and glucose uptake in target tissues (41). Similarly, we observed lowered fasting-insulin concentrations, systemic glucose intolerance, and normalized heart glucose uptake in mice upon additional deletion of Yme1l in skeletal muscle. Loss of YME1L appears to impair insulin signaling via its effect on mitochondrial dynamics and stress-induced OPA1 processing by OMA1, because we observed normal systemic glucose homeostasis in mice lacking both OMA1 and YME1L in skeletal muscle. These results highlight the physiological importance of tissue cross-talk that must be taken into consideration when analyzing tissue-specific models for mitochondrial disease.

Further support for a critical role of disturbed cardiomyocyte metabolism for heart failure came from the observation that feeding a high-fat diet preserved normal cardiac function of cardiomyocyte-specific YME1L knockout mice (Fig. 7). Similar to ablation of Yme1l in skeletal muscle, this metabolic intervention did not circumvent the primary mitochondrial defects in cardiomyocytes because the catalytic activity of YME1L, OMA1 activation, mitochondrial morphology, and respiratory chain profiles were unaffected by diet. However, it did blunt downstream metabolic disruptions and normalized cardiac glucose uptake in cardiomyocyte-specific YME1L knockout mice, which suppressed cell death and preserved cardiac function. Remarkably, feeding of cardiomyocyte-specific YME1L knockout mice with a high-fat diet prevents cardiomyopathy at least at early stages despite normal weight gain of the mice. Echocardiographic analyses of failing cYKO hearts demonstrate that these mice have reduced contractile function but normal left ventricular mass, which suggests that YME1L ablation does not result in a concentric hypertrophic response, and thus cardiac dysfunction can be improved by metabolic intervention. In this sense, some forms of DCM in humans can be treated by pharmacological means (65). Consistent with previous reports (37), high-fat feeding of WT mice did not incite contractile dysfunction (table S2). In contrast, feeding mice manifesting cardiomyopathy associated with ventricular hypertrophy a high-fat diet exacerbates myocardial dysfunction (66, 67). It is thus an intriguing possibility that this metabolic intervention is beneficial exclusively in the context of DCM. Our results reveal an intimate relation between alterations in mitochondrial morphology and metabolism in the heart, which may underlie myocardial disease in humans. Metabolic interventions can preserve cardiac function even if mitochondrial morphology is disturbed, which opens potential avenues for therapeutic interventions in myocardial disease.

Methods

Echocardiography and PET-CT

Echocardiography and functional examination in mice was performed as previously described (68). All PET-CT studies were performed with a small-animal PET-CT device. Briefly, animals were fasted overnight, and anatomic thorax CT scanning was performed 1 hour after [18F]FDG injections, followed by metabolic PET static acquisition for 15 min. Image analysis was performed in prefused and prereconstructed images with Osirix (Aycam Medical Systems, LLC); we selected myocardium of the whole left ventricle and calculated mean myocardial standardized uptake value (SUV med) for each animal. See supplemental methods for full methods description.

Incline treadmill

Treadmill experiments were conducted using a TSE treadmill and up to six randomized mice assessed simultaneously. Low-intensity incline (5%) experiments began at 0.05 m/s and increasing to 0.1 m/s for 60 s, 0.1 m/s for 600 s, then increasing to 0.28 m/s for 3100 s by 172-s increments. Blinded determination of exercise exhaustion was assessed and plotted as a function of total distance (m).

Histology and immunochemistry

Paraffin-embedded sections (4 μm) were subjected to hematoxylin and eosin, picrosirius red, or Masson’s trichrome staining. Images were acquired using a Leica SCN400 automated slide scanner at 40×.

Generation of primary cardiomyocytes and cardiac fibroblasts

Adult cardiomyocytes were isolated by retrograde Langerdorff perfusion with an enzymatic digestion buffer containing trypsin and Liberase (Roche Applied Science) (68). Primary cardiac fibroblasts obtained during cardiomyocyte isolations were immortalized using a plasmid encoding SV40 large T antigen and deleted for Yme1l in vitro. See supplemental methods for full methods description.

Transmission electron microscopy (TM)

Left ventricle samples of hearts perfused with paraformaldehyde [2% (w/v) in phosphate-buffered saline (PBS)] and glutaraldehyde [2% (w/v) in PBS] hearts were fixed for 3 days in 2% (v/v) glutaraldehyde, 2.5% (w/v) sucrose, 3 mM CaCl2, and 100 mM HEPES-KOH, pH7.4, at 4°C. After washes, samples were fixed using reduced OsO4 [1% (w/v) OsO4, 10 mg/ml potassium ferrocyanide, 1.25% (w/v) sucrose, and 100 mM sodium cacodylate, pH 7.4] for 1 hour on ice. After washes in water, cells were incubated in 2% (w/v) uranyl acetate for 30 min. After dehydration with 50, 70, 90, and 100% ethanol, samples were embedded in epon resin. Samples were observed under a transmission electron microscope (EM902; Carl Zeiss) at an acceleration voltage of 80 kV.

Confocal fluorescence microscopy

To monitor mitochondrial morphology by immunofluorescence microscopy, primary cardiomyocytes were fixed and stained with antibodies directed against TOMM20 (rabbit-specific antibody 1:1000, Santa Cruz Biotechnology). Fluorescently coupled secondary antibody Alexa Fluor 568 (a goat antibody directed against a rabbit secondary antibody) was used at 1:1000 dilution (Invitrogen). Images were acquired using an UltraVIEW VoX spinning disc microscope (CSU-X1; Yokogawa Corporation of America). Quantification of mitochondrial morphology in primary cardiomyocytes was performed by blinded, randomized examination of z-stack images. Cardiac necrosis was assessed by Evans Blue diffusion as previously described (57). Paraffin-embedded cardiac sections were stained with fluorescein isothoicyante–conjugated wheat germ agglutinin (Invitrogen, W834) for 30 min before nuclear counterstaining with 4′,6′-diamidino-2-phenylindole (DAPI). Evans Blue–positive cardiomyocytes fluorescence is in red. Images were acquired using a Leica SP8 confocal microscope at 40×.

Blood glucose and serum analyses

Intraperitoneal GTTs were carried out in mice after they were fasted for 6 hours. After determination of fasted blood glucose levels, an intraperitoneal bolus of 2 g glucose/kg body weight [20% (w/v) glucose]. Blood glucose levels were determined after 15, 30, 60, and 120 min using Contour test strips (Bayer, Germany). Serum insulin levels were measured from blood collected from 6 hours fasted mice, by enzyme-linked immunosorbent assays (ELISAs), according to the manufacturer’s instructions (Mouse/Rat Insulin ELISA, Mouse Leptin ELISA;Crystal Chem Inc.). Serum levels of cardiac troponin assayed by auto-analyzer Dimension RxL Max HM (Siemens).

Oxygen consumption measurements

Oxygen consumption of WT and cYKO hearts, fed normal chow diet (NCD) or high-fat diet (HFD), was measured using a fluorescence-based micro-optode that consisted of an optic fiber equipped with an oxygen-sensitive fluorescent terminal sensor (FireSting O2; Bionef, Paris, France) as described previously (69). Hearts were perfused in buffer (10 mM KH2PO4, 300 mM mannitol, 10 mM KCl, 5 mM MgCl2, 5 mM bovine serum albumin, pH 7.4), dissected, weighted, and assayed at rest, precisely 3 min after mice were killed by cervical dislocation. Oxygen uptake rates (nmol O2/min) were adjusted relative to wet heart weight.

Respiratory chain measurements

Activity of respiratory chain complexes and citrate synthase was spectrophotometrically assayed as previously described (70).

Statistical analysis

Statistical analyses were performed using Prism (GraphPad Software Inc., San Diego, CA). All values are expressed as means ± SEM. Statistical significance was assessed by using a two-tailed unpaired Student’s t test or the Mann-Whitney test for two-group comparisons. Two-way analysis of variance (ANOVA) with Bonferroni post hoc tests (corrected P values are given for comparison between genotypes at specific time points) were used to evaluate multiple pairwise comparisons of groups. Life span survival curves were compared by using a log-rank Mantel-Cox test. Differences were considered statistically significant at a value of P < 0.05. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Supplementary Materials

www.sciencemag.org/content/350/6265/aad0116/suppl/DC1

Additional Materials and Methods

Figs. S1 to S7

Tables S1 and S2

References (71, 72)

REFERENCES AND NOTES

  1. ACKNOWLEDGMENTS: We thank K. Lemke, H. Bank, E. Barth, V. Zorita, and M. Gómez for technical assistance; A. Polykratis and C. Uthoff-Hachenberg for genetic engineering assistance; A. Pun-García and R.Villena-Gutierrez for cardiomyocyte isolations; A. V. Alonso López, L. Flores Ruiz, and J. Jimenez-Borreguero for echocardiography evaluation; I. Bilbao, C. Velasco, and J. Ruiz-Cabello for PET-CT evaluation; A. Ferrarini and D. Dudzik for help in metabolomics; G. Rapl for single-cell sorting of cardiac fibroblasts; P. Frommolt for microarray quality control; and J. Brüning for discussion. This work was supported by a fellowship of the Human Frontiers Science Program to T.W., by grants of the Deutsche Forschungsgemeinschaft and the European Research Council to T.L., and by a grant from the Spanish Ministry of Economy and Competitiveness (MINECO) through the Carlos III Institute of Health—Fondo de Investigación Sanitaria and European Regional Development Fund (ERDF/FEDER) funds (PI13/01979) and Networks for Cooperative Research in Health (RETIC) (RD12/0042/0054) to B.I. The CNIC is supported by the Ministry of Economy and Competitiveness and the Pro-CNIC Foundation. B.I. is Princess of Girona awardee in science. F.J.R. and C.B. acknowledge MINECO CTQ 2014-55279-R. The authors declare no competing or financial interests. The mice used in this study are available under a materials transfer agreement from the authors. The data are included in the main manuscript and the supplementary materials. Author contributions: M.J.B. generated the floxed Oma1 mouse, T.W. generated the floxed Yme1l mouse, hmYKO mice, hmOKO mice, hmDKO mice, cYKO mice, and cDKO mice. C.M. designed targeting constructs. P.B. and P.R. performed oxygen consumption measurements, F.J. R. and C.B. performed the acylcarnitine profiling, and T.W. and J.G.-P. performed all other experiments. T.W., J.G.-P., B.I., and T.L. drafted the manuscript.
View Abstract

Navigate This Article