Mitofusins 1 and 2 are Essential for Postnatal Metabolic Remodeling in Heart

Abstract

Rationale

At birth, there is a switch from placental to pulmonary circulation and the heart commences its aerobic metabolism. In cardiac myocytes, this transition is marked by increased mitochondrial biogenesis and remodeling of the intracellular architecture. The mechanisms governing the formation of new mitochondria and their expansion within myocytes remain largely unknown. Mitofusins (Mfn-1 and Mfn-2) are known regulators of mitochondrial networks but their role during perinatal maturation of the heart has yet to be examined.

Objective

Determine the significance of mitofusins, during early postnatal cardiac development.

Methods and Results

We genetically inactivated Mfn-1 and Mfn-2 in mid-gestational and postnatal cardiac myocytes using a loxP/Myh6-cre approach. At birth, cardiac morphology and function of double-knockout (DKO) mice are normal. At that time, DKO mitochondria increase in numbers, appear to be spherical and heterogeneous in size but exhibit normal electron-density. By postnatal day 7, the mitochondrial numbers in DKO myocytes remain abnormally expanded and many lose matrix components and membrane organization. In this context, DKO mice develop dilated cardiomyopathy (DCM). This leads to a rapid decline in survival and all DKO mice perish before 16 days of age. Gene expression analysis of DKO hearts shows that mitochondria biogenesis genes are down regulated, the mitochondrial DNA is reduced and so are mitochondrially-encoded transcripts and proteins. Furthermore, mitochondrial turnover pathways are dysregulated.

Conclusions

Our findings establish that Mfn-1 and Mfn-2 are essential in mediating mitochondrial remodeling during postnatal cardiac development, a time of dramatic transitions in the bioenergetics and growth of the heart.

INTRODUCTION

Remodeling a fetal into an adult heart is a complex process and relies on important transitions that are triggered shortly after birth. During this early postnatal stage the heart undergoes a switch in substrate utilization to catabolize fatty acids, and carbohydrates become a secondary source of energy ^{1, 2}. Failure to enact this metabolic switch in mice is marked by cardiac defects and a lifespan of up to two weeks ^{3, 4}. Apart from the metabolic switch, the early postnatal stage is also marked by a switch in the mode of cardiac growth. Cardiac myocytes largely cease proliferation around postnatal day (P) 04 and undergo hypertrophic growth leading to an increase in diameter and mass ⁵. In mice, physiological myocyte hypertrophy between P05 and P14 results in a nearly three-fold increase in heart weight ⁶. Furthermore, mitochondrial density doubles in cardiac myocytes during early post-natal development, and the small, round and tubular mitochondria that are found in fetal hearts are reformed into large ovoid and rectangular mitochondria ^{7, 8}. This change in mitochondrial shape is paralleled by reorientation of mitochondria such that there is larger contact surface and closer proximity between mitochondria and the myofibrils, and between mitochondria and the sarcoplasmic reticulum (SR), a transition that is noted to take place around postnatal day 7 (P07) in mice ⁹. This histological evidence suggests that mitochondrial remodeling is important for the passage of the heart from a fetal to an adult state.

Genetic and functional studies firmly established mitofusins as key regulators of mitochondrial morphology in a wide spectrum of organisms ranging from yeast (single gene, Fzo1p) to mammals (two genes, Mfn-1 and Mfn-2) ^10–13. Murine Mfn1 and Mfn2 have ~65% identity at the amino acid level, and in several contexts they exhibit functional overlap or complementation in mediating outer mitochondrial membrane fusion ^13–16. The molecular characterization of mitofusins has been instrumental in elucidating the significance of mitochondrial dynamics, an enigmatic process that involves the flow of mitochondrial matter through merging, splitting and transposition of mitochondria ^{17, 18}. Mitochondrial dynamics have implications in many aspects of mitochondrial function ¹⁹ and appear to be particularly important in mitochondrial quality control, the pathway that selectively separates-out defective mitochondria for subsequent elimination ^{20, 21}. Importantly, the dynamic behavior of mitochondria is not the same in all cell types and is strongly influenced by the intracellular motility and the cytoskeleton dynamics ^22–24. Along these lines, the striated contractile cells of the adult heart and skeletal muscle have a very rigid cytoskeleton and their mitochondria are stationary ^{25, 26}. Nevertheless, the expression of Mfn-1 and Mfn-2 is very robust in these cells with immotile mitochondria, indicating that mitofusins are still biologically important.

To address the functional significance of mitofusins in striated tissues, several laboratories have pursued the conditional inactivation of Mfn-1 and/or Mfn-2 in the skeletal and cardiac muscle of mice. So far it has been found that both Mfn-1 and Mfn-2 are necessary for the maintenance of the mitochondrial genome in skeletal muscle ²⁷. Furthermore, Mfn-2 in skeletal muscle is important in preserving insulin sensitivity and repressing oxidant stress ²⁸. Regarding the heart, both Mfn-1 and Mfn-2 are shown to be important during early embryonic cardiac development and in adulthood they protect against long-term cardiac dysfunction ²⁹. Our studies with cardiomyocyte-restricted mitofusin knockouts have focused on the identification of common as well as distinct properties of Mfn-1 and Mfn-2. Work with the Mfn-2 deficient strain showed this isoform to be required for suppressing mitochondria from undergoing excessive growth in cardiomyocytes ³⁰. In contrast, analysis of the Mfn-1 deficient strain showed that this isoform operates in the reverse mode, i.e. Mfn-1 protects from excessive shrinkage/fragmentation of the mitochondria ³¹. Another finding from these studies was that Mfn-1 or Mfn-2 deficient cardiomyocytes contain mitochondria that are relatively impervious to injurious stimuli such as Ca²⁺ and ROS overload ^{30, 31}. High doses of these stressors are known to target the mitochondrial boundaries and cause organelle dysfunction by triggering mitochondrial permeability transition (MPT) ³². The resistance to MPT was detectable in both Mfn-1 or Mfn-2 deficient mitochondria, suggesting that there exists a direct relationship between mitochondrial membrane fusion and permeabilization ^{30, 31, 33}. Lastly, these studies showed that the baseline heart function is not severely impacted by the deletion of either Mfn-1 or Mfn-2 in adult animals ^{30, 31}.

In this report, we present work regarding the dual deletion of Mfn-1 and Mfn-2 in cardiac myocytes. We show that permanent inactivation of mitofusins in cardiomyocytes during mid-gestation has profound effects on the function of the early postnatal heart. Our findings also show that cardiac Mfn-1 and Mfn-2 have important roles for survival post natum and provide evidence for the significance of mitochondrial remodeling during this particular developmental stage.

METHODS

Detailed methods are provided in the online supplement.

Genetically engineered mice

Mice have a mixed background containing 129S, Black-Swiss, and C57Bl/6. Littermates with the appropriate genotypes were used in all experiments. The cre-mediated recombination of loxP sites excises exons encoding key residues of the G1 motif of the GTPase domain of Mfn-1 (exon 4) or Mfn-2 (exon 6) ³⁴. Cardiac myocyte-specific cre expression was driven by a constitutive Myh6 promoter (active during embryogenesis), or a modified estrogen receptor (MER) that is inducible (active upon injection with raloxifen). For studies with adult animals, male littermates were used. Animal euthanasia and tissue collection were performed according to protocols approved by the Institutional Animal Care and Use Committee (IACUC) of Boston University School of Medicine.

Physiological studies

Adult animals were lightly anesthetized prior to undergoing echocardiography or hemodynamic analysis as previously described ³⁰. Echocardiography of neonates was performed at the conscious state. The neonatal electrocardiograms were obtained as previously described ³⁵ with the aid of light isoflurane anesthesia delivered through a mouth/nose cone.

RESULTS

Mitofusins have redundant functions in cardiomyocytes that are important for survival post natum

We have previously described mice with conditional deletions of Mfn-1 or Mfn-2 in cardiomyocytes and found that single-knockout mice are viable, fertile and display a normal basal heart function as young adults ^{30, 31, 36}. Given the high similarity between Mfn-1 and Mfn-2 we asked whether they could operate interchangeably in cardiac myocytes and therefore attempted to generate double-knockout animals in both the adult and embryonic heart. Previously, it has been shown that depletion of mitofusins by the tamoxifen-activated protein MerCreMer leads to progressive decreases in fractional shortening (FS) in adult mice ²⁹. In agreement with this study, we find that mitofusin depletion in mice (herein referred to as Mer-DKO) leads to cardiac dysfunction evident at 8 weeks after raloxifen administration and death ensues 3 weeks later (Online Figure I). These observations confirm the notion that Mfn-1 and Mfn-2 operate redundantly and are essential for cardiac function and survival in adulthood.

To examine the roles of cardiomyocyte Mfn-1 and Mfn-2 specifically during late-gestation/early postnatal development of the heart we used the Myh6-cre transgenic line to selectively disrupt the mfn-1 and mfn-2 loci of mice (genotype: Mfn-1^F/F;Mfn-2^F/F;Myh6-cre^+/−, herein DKO). Genotyping of 104 animals from multiple litters at their weaning (P21) did not detect any mice with the desired genotype (Online Table I), suggesting that DKO mice were not viable and that all four mitofusin alleles are needed for survival either during mid-gestation or during the first three weeks of life. In this analysis it was found that even one functional allele (regardless of whether this is from Mfn-1 or Mfn-2) is sufficient to compensate for the absence of the other three alleles. Mice harboring one functional mitofusin allele (Mfn-1^F/F;Mfn-2^+/F;Myh6-cre^+/−, or Mfn-1^+/F;Mfn-2^F/F;Myh6-cre^+/−, collectively referred to as “monoallelic”) survive through adulthood, although their hearts exhibit chamber enlargement, decreased fractional shortening (FS) and alterations in their elastic properties (Online Figure II).

To test the possibility that DKO mice die during gestation, we genotyped mice at birth (P0). The analysis of a total of 134 pups obtained from 15 litters indicated that DKO mice were born at the expected Mendelian ratios (Online Table II). This rules-out the possibility that DKO mice die during gestation that could potentially lead to their resorption in utero. Furthermore, from the 32 DKO pups identified all 32 were alive. Therefore, DKO mice survive through the embryonic and fetal stage and are born alive in the expected Mendelian ratios. However, because none of the DKO animals are recovered at P21, we conclude that death occurs at an early postnatal stage and that even a single mitofusin allele in cardiac myocytes is sufficient for survival through this period.

Cardiomyocyte-produced mitofusin expression is elevated during postnatal heart development

To determine the temporal pattern of mitofusin expression in the developing heart, we examined embryonic and early postnatal hearts (i.e. E15.5 through P07) for the expression of Mfn-1 and Mfn-2. One finding of the transcriptional analysis is that in WT hearts both Mfn-1 and Mfn-2 transcripts are produced at relatively low amounts during gestation (E15.5 and E18.5, Figures 1A and B) but undergo a more than two-fold upregulation around the time of birth (P0 and P01). Furthermore, significant transcriptional upregulation of mitofusins is also evident at P07 (Figures 1A and B). By contrast, in DKO hearts the levels of Mfn-1 and Mfn-2 were found to be down-regulated as early as E15.5 and persisted to these very low levels throughout the remaining gestational and postnatal period examined (Figures 1A and B, open bars). Consistent with the low levels of mitofusin expression already at E15.5 in DKO hearts, previous reports have shown that the Myh6 promoter is activated in early myocytes at E9.5 and that genetic recombination using the Myh6-cre construct can be achieved by E12.5 ^{37, 38}. The residual levels of mitofusin transcripts detected in DKO extracts likely arise from non-myocyte cardiac-resident cell populations that do not express the Myh6-cre transgene; however, this analysis reveals that the bulk of mitofusin expression in the postnatal heart is attributable to the cardiomyocyte population (Figure 1A and B). Taken together, the analysis of mitofusin expression in WT and DKO hearts indicates that gene ablation takes place during mid-gestation and provides evidence for transcriptional activation of mitofusin genes during postnatal heart development.

Figure 1. Mitofusins are transcriptionally activated in postnatal WT but not DKO hearts.

A) Quantitative assessment of cardiac Mfn-1 mRNA in different developmental stages. The quantitation was performed in real time according to the ΔΔCt method where the levels of Mfn-1 in WT E15.5 hearts were used as a reference (value=1). B) Quantitative assessment of cardiac Mfn-2 mRNA in the indicated developmental stages. In both A and B, Rpl30 was used as the housekeeping gene. Two or three hearts per group per developmental stage were used and error bars represent standard error. C) The growth of DKO hearts during the developmental window E15.5-P03 is not hampered. Overall morphology in freshly-dissected WT and DKO hearts at the indicated developmental stages. Hearts were photographed in ice cold PBS.

Myh6-cre mediated ablation of mitofusins spares the heart from gestational developmental abnormalities and dysfunction

To test whether mitofusin ablation in this model is potentially associated with developmental abnormalities, we examined WT and DKO hearts from E15.5 through P03. This analysis showed that the expected developmental increases in cardiac size that occur in WT hearts also take place in DKO hearts (Figure 1C). To assess a potentially deleterious effect in heart function that may have arisen in DKO hearts during mid-gestation or at the time of birth, we employed echocardiographic analysis on newborn mice. The morphometric data were obtained by examining conscious mice on the day of their birth (P0). As shown in Figure 2, the evidence is in agreement with the notion that P0 DKO hearts are functionally equivalent to WT hearts (Figure 2A and B). More specifically, we identified close similarities between WT and DKO mice in terms of their heart rate, their fractional shortening (FS) and a number of other parameters related to the structure and the relaxation/contraction of the heart including left ventricular (LV) wall thickness, LV volume and mitral and pulmonic valve flow velocities (Figure 2C–J). Taken together, the absence of any evidence for heart dysfunction indicates that mid-gestational ablation of mitofusins, selectively in cardiac myocytes, spares the heart from significant developmental abnormalities and dysfunction through P03.

Figure 2. The heart function at birth (P0) is similar in DKO compared to WT mice.

A) Representative M-mode echocardiograms taken from parasternal short axis views of the left ventricle of WT and DKO pups. B) Representative flow patterns at the mitral valve of WT and DKO mice. The bright arcs appearing in pairs above the baseline are the E and A waves. Note that at this age the A wave peaks higher than the E wave. C–J) Values of different echocardiographic parameters in conscious WT and DKO mice at P0 (n=10 per group). FS; fractional shortening, LVIDd; Left ventricle internal diameter in diastole, LVPWs; LV ventricle posterior wall thickness in systole, LV Vol.d; Left ventricle volume in diastole, E/A ratio between the peak velocity of the early (A) and late (D) mitral inflow.

Morphological features of P0 mitochondria

Recently, we have demonstrated that absence of either Mfn-1 or Mfn-2 in adult cardiac myocytes leads to characteristic but disparate mitochondrial morphologies ^{30, 31}. Whereas loss of Mfn-2 creates larger mitochondria, loss of Mfn-1 has the opposite phenotype and creates smaller mitochondria. To determine the impact of combined Mfn-1 and Mfn-2 ablation in cardiomyocyte ultrastructure, we examined heart sections from P0 hearts by electron microscopy (Online Figure III). Wild-type P0 cardiomyocytes contain tubular and ovoid mitochondria and sometimes elongated, vermiform mitochondria (Online Figure III, panel A). The electron-dense WT mitochondria are dispersed among myofibrils and the P0 myofibrils also exhibit irregular orientations inside the cytoplasm (Online Figure III, panel C), and have yet to assume the linear arrangement which is typical for myofibrils in longitudinal sections of fully developed hearts. Lipid droplets were frequently observed in other regions of the cytosol. In P0 DKO cardiomyocytes, the cytoplasm appears to be poorly organized similarly to the WT cytoplasm. However, a large proportion of mitochondria assume a spherical configuration (Online Figure III, panel B). In addition to a transition to a spherical state, DKO mitochondria undergo enlargement and diminution, and big mitochondria are found to be intermixed with small mitochondria (Online Figure III, panels D and E). Tubular mitochondria were also detected but less frequently (Online Figure III, panel D). By analysis of the micrographs, we find that the mitochondrial volume density is significantly elevated in DKO (Online Figure III, panel F) indicating the presence of higher numbers of mitochondria in these hearts. Nevertheless, P0 DKO mitochondria do not exhibit significant defects in their internal structure and they maintain an electron dense appearance. Furthermore, the formation of myofibrillar bundles does not appear to be impaired in DKO hearts at P0. Consistently, the volume density of the myofibrillar compartment in these hearts appears to be normal (Online Figure III, panel G). In summary, ultra-microscopy of the P0 DKO cardiac myocytes revealed mitochondrial alterations in terms of the structure and number of mitochondria but a normal assembly of the contractile elements.

Loss of mitofusins from myocytes leads to heart failure by postnatal day 7

Both Mfn-1 and Mfn-2 genes are markedly up-regulated in heart during the postnatal period between P0 and P07 (Figure 1A). We therefore investigated the cardiac phenotype of DKO mice at P07. As shown in Figure 3, echocardiographic assessment of conscious WT and DKO mice at P07 revealed that the latter underwent a very rapid and severe dilated cardiomyopathy (DCM). The wall motion in DKO hearts was drastically attenuated and the blood flow in the LV chamber was not properly maintained (Figure 3A and B). Consistently, significant alterations in different quantitative parameters were detected at P07. As such, the heart rate was lower in DKO hearts and their fractional shortening was markedly decreased (Figure 3C and D). Furthermore, the internal diameter of the left ventricle was increased while its walls underwent significant thinning (Figure 3E and F). The calculated diastolic volume of the left ventricle underwent a nearly three-fold increase in DKO hearts compared to WT, in agreement with the DCM phenotype (Figure 3G). At the same age, DKO had abnormal electrocardiogram patterns where the QRS complex was evidently altered (Figure 3H). Taken together, these characteristics are indicative of the detrimental remodeling and poor pumping function of the DKO heart. The early temporal manifestation of this phenotype (i.e. between P01 and P07) suggests that the overlapping actions of Mfn-1 and Mfn-2 in cardiomyocytes are essential during the developmental remodeling of the early postnatal heart.

Figure 3. The heart function at P07 is impaired in DKO mice.

A) Motion of the ventricular walls is monitored by echocardiography in WT and DKO mice. Note the poor movement of the walls in the DKO. B) Pattern of the inward (above baseline) and outward (below baseline) flow of blood through the mitral valve in WT and DKO hearts. C–G) Morphometric analysis of WT and DKO hearts at P07. For definitions of acronyms see legend in Fig. 2. The thresholds of significance are indicated for each pairwise comparison. H) Representative ECG patterns (lead II configuration) of WT and DKO mice at P07. A single arrow indicates the QRS complex in WT and a double arrow the QRS complex in DKO mice. ^†indicates the P waves. mV; millivolts, s; seconds

Rapid decline in survival and cardiac defects in mice lacking mitofusins

Based upon the genotype frequencies (Online Table I and Online Table II), it is apparent that although DKO mice are born alive they perish prior to their weaning (P21). To define the profile of this premature lethality in greater detail, we performed genotyping on newborn WT and DKO mice and monitored them until weaning or death, respectively (Figure 4A). From the resulting survival curve, it is evident that DKO mice start to die as early as P06 and none survive beyond P16 (Figure 4A, red trace). The majority of the DKO mice die between P08–P10. This pattern is consistent with the evidence from echocardiography that detected no signs of heart dysfunction at P0, but identified dilated cardiomyopathy by P07 (Figure 2 and 3). Thus, DKO mice likely undergo premature death owing to dilated cardiomyopathy and heart failure. DKO mice typically display a decline in body condition that becomes evident around P06–P07 (Figure 4B). The body weight of DKO mice was found to be significantly lower compared to WT littermates at that age (Figure 4D). On autopsy at P04, some DKO hearts displayed readily evident abnormalities (Figure 4C and F). At that age, the DKO heart was flaccid and the ventricular walls appeared to fold inwardly (Figure 4C, right panel, arrowhead point the right ventricular wall). This was in contrast to the round and well-shaped WT ventricular walls. In addition, the left atrium appeared to undergo hypertrophy and retained blood in the form of a clot (Figure 4C, arrow).

Figure 4. Diminished lifespan and features of cardiac pathology in DKO mice.

A) Survival curves of WT, monoallelic and DKO mice. Pairwise comparisons (Log Rank) reveal P=0.024 in the difference in survival between WT and monoallelic mice. Difference in survival between DKO and WT or DKO and monoallelic is significant for P<0.001. The median survival time of DKO mice is 9.83 days. B) Representative picture of WT and DKO littermates at P06. C) Gross morphology of freshly-dissected WT and DKO hearts at P04. The arrowhead shows thinning of the right ventricular (RV) wall and the arrow thrombus formation in the left atrium. Scale bar is 500 μm. D–E) Metrics of WT and DKO mice at age P06–P07. F) Sagittal sections of WT and DKO hearts at P04 stained with Masson’s trichrome. White arrow indicates focal fibrosis in the DKO heart. Black arrow and arrowhead indicate atrial congestion and interventricular septum thinning respectively.

In cross-section and trichrome staining, the P04 DKO hearts displayed increases in the right and left ventricular cavities as well as substantial thinning of the walls (Figure 4F, arrowhead pointing to the interventricular septum) an observation consistent with the DCM phenotype detected by echocardiography (Figure 3). Furthermore, patches of collagen deposition could be observed in the DKO myocardium, suggestive of ongoing cardiac fibrosis (Figure 4F, white arrow). The histological analysis also confirmed the atrial enlargement and congestion (Figure 4F, dark arrow). Despite these adaptations, the total heart weight did not differ significantly between WT and DKO mice autopsied at P04 (results not shown) and P07 (Figure 4E). Consistent with the onset of cardiopathic stimuli in DKO heart, the transcript levels of myocardial stress indicators Anp and Bnp were markedly elevated, whereas the transcript levels of the different myosin heavy chain (Mhc) isoforms and inhibitory troponin (TnI) isoforms were uniformly downregulated in P07 DKO hearts (Online Table III). Interestingly, there was no evidence for significant activation of apoptosis in P07 DKO hearts (Online Figure IV).

Post-natal structural defects in the mitochondrial and myofibrillar compartments of the Mfn-1/Mfn2 double-deficient cardiomyocytes

Since the absence of mitofusins from cardiac myocytes led to a dramatic decline in heart function within the first week of postnatal life, we examined the ultrastructure of cardiac sections by electron microscopy at P07 to obtain clues as to the mechanisms that underlie this dysfunction. Imaging of WT myocardium revealed that myocytes at this stage exhibit extensive features that are reminiscent to adult cardiac myocytes, i.e. the actin/myosin banding becomes readily evident, the myofibrillar bundles increase in diameter and they assume a linearized orientation and run parallel to the longitudinal axis of the myocyte (Figure 5A). The WT mitochondria at this stage have mostly rectangular or spherical shapes while the tubular mitochondria become less frequent. Coincident with the linear orientation of myofibrils, P07 mitochondria are arranged parallel to, and stay in close proximity to the myofibrils (Figure 5A, C and E). In addition to maintaining close apposition to the contractile elements, P07 mitochondria exhibit tight packing leaving behind little free cytoplasm. Lipid droplets are also present, appearing as grey or white spots within the mitochondrial conglomerate (Figure 5C and E).

Figure 5. Aberrant accumulation of spherical mitochondria in DKO hearts at P07.

A) Low power image showing the typical intracellular structure of WT cardiac myocytes. B) Low power image of DKO cardiac myocytes containing spherical mitochondria of variable size that occupy a large portion of the cell. Z; sarcomeric banding, M; mitochondria, L; lipid droplet. C) Structural arrangement of mitochondria and myofibrils in WT cardiac myocyte. D) Altered proportions between mitochondria and myofibrils in DKO cardiac myocytes. Arrowheads in C and D indicate the distance between myofibrils. E) Morphologies of intermyofibrillar/perinuclear mitochondria in WT cardiac myocytes. F) Abnormally patterned mitochondria in DKO cardiac myocytes. Mitochondria with unusual sizes (small and large) are indicated by arrows. These representative images were taken from two animals per genotype. G–H) Quantitation of volume density of mitochondria and myofibrils. Each quantified field (×6300 magnification) had dimensions 12.7×13.9 μm.

In P07 DKO hearts, the myocyte structure exhibits marked deviations from that in WT cells. A profound alteration is the high frequency of spherical mitochondria (Figure 5B), a feature consistent with previous reports on double-mitofusin null models in striated muscle ^{27, 29}. The formation of spherical mitochondria was also observed in P0 DKO cardiomyocytes. Although an increased mitochondrial number was also evident at P0 (e.g. Online Figure III, panels B, D and E), this did not coincide with major aberrations in the structure and organization of the myofibrils. In P07 DKO samples, the abnormal increase in mitochondrial numbers is still prevalent, but there are regions of cardiomyocytes where the mitochondria appear to replace the myofibrils (Figure 5B). Related to this disorganized mitochondrial spreading, the distances between parallel myofilaments increase in DKO sections (e.g. compare Figure 5C and D, arrowheads). These features were also quantitatively important. Using the grid-method to determine how the cell volume is allocated into different compartments, we detect a significant increase in mitochondrial density that is associated with reduced myofibrillar density in DKO hearts (Figure 5G and H). In addition to the apparent increase in mitochondrial number, a characteristic dysregulation in mitochondrial diameter was observed, whereby unusually large mitochondria coexist with petite mitochondria (Figure 5F, large and small arrows respectively, also compare with WT mitochondria, Figure 5E).

The mitofusin null mitochondria presented varying degrees of cristae defects and as a consequence the structure of the inner membrane was modestly (Figure 6B and B′) or severely (Figure 6C and C′) impaired. This pattern of inner membrane dysregulation, as detected in DKO mitochondria, is in sharp contrast with the dense cristae packing that is observed in WT mitochondria at P07 (Figure 6A) or the WT and DKO mitochondria at P0 (Online Figure III). Furthermore, the loss of cristae and the decondensation of the matrix appeared to occur similarly among the different sizes of mitofusin null-mitochondria, i.e. loss of the inner structure happens regardless of whether the mitochondria have small, intermediate or large diameters (compare Figure 6A with 6C).

Figure 6. Structural abnormalities of P07 DKO mitochondria.

A and A′) Sample image of WT intermyofibrillar mitochondria. The boxed region in A is shown in greater detail in A′. B) Image of DKO intermyofibrillar mitochondria and select region (white box, B′) presenting mitochondria undergoing unpacking of cristae and loss of matrix density. C) Image of DKO mitochondria and select region (white box, C′) presenting extreme degeneration of cristae, mitochondrial swelling and loss of matrix density. D) Sample region of WT mitochondria and select region (white box, D′) presenting the boundary relationship between WT mitochondria. E) Image of DKO mitochondria forming atypical protrusions and select region (white box, E′) showing in detail the local curvature of the mitochondrial membrane (*). F) Sample image from DKO cells. ^‡indicates mitochondria with poor internal organization, arrows indicate mitochondrial voids. G–I) Sample DKO regions containing chains of mitochondria with atypical membrane protrusions (*) and omega-like membrane structures (#).

The mitochondrial boundaries in WT samples are smooth and their shape is defined by subtle changes in membrane curvature (Figure 6D and D′). In DKO samples, however, the boundaries of some mitochondria appear to undergo dramatic changes in curvature and the membranes create ‘finger-like’ projections (Figure 6E, E′, G, H and I). Furthermore, a single DKO mitochondrion is capable of forming one, two or three such projections (more typically just one, e.g. Figure 6G, asterisk), and these projections appear to push into the membranes of an adjacent mitochondrion creating an invagination. Consequently, short chains of mitochondria are created where one mitofusin-null mitochondrion extends a projection towards its neighbor (Figure 6H). In other situations, the surfaces of adjacent mitochondria undergo even more extreme curvature to form Ω (‘omega-like’) membrane configurations (Figure 6G, hash and also Figure 6E). The membrane protrusions appear to preferentially form on DKO mitochondria that exhibit extensive matrix decondensation (Figure 6E, G, H and I, asterisks). Nevertheless, not all mitochondria with apparent structural defects form these protrusions (Figure 6F, dagger), indicating that the abnormalities in the internal structure of mitochondria (reflecting mitochondrial dysfunction) precede the formation of these membrane deformities.

Lastly, DKO sections contained small, vesicle-like organelles that appeared to be ‘voids’ surrounded by a single membrane (Figure 6E and F, arrows). These vacuoles do not represent lipid droplets because they are smaller in size nor exhibit the gray coloration that is typically seen inside the droplet.

Altered expression of mitochondrial biogenesis and mitophagy markers in mitofusin-deficient hearts at P07

Mitochondrial biogenesis is currently viewed as the net result of two opposing processes. On the one hand, forming mitochondria involves the incorporation of de novo synthesized proteins into preexisting mitochondria that replicate their genome and subsequently segregate into daughter mitochondria. On the other hand, formation of mitochondria is counteracted by whole-organelle elimination, through mitochondria-directed autophagy (mitophagy). Thus, we analyzed whether these processes were altered given the severe mitochondrial abnormalities and the appearance of defective mitochondria in DKO hearts at P07. Biosynthesis of new mitochondrial components is controlled by a group of transcription factors (Nrf-1, Err-α and Tfam) and the coactivator PGC-1α ³⁹. As shown in Online Figure V, the transcript levels of these factors in DKO hearts exhibit little or no changes at P0 or P03. Nevertheless, the majority of these genes become significantly suppressed in P07 DKO hearts (Figure 7A). Next, we examined the transcription of genes important in the autophagy pathway. As shown in Online Figure VI, we analyzed genes such as Ulk-1 and -2 (encoding upstream kinases regulating the initiation of autophagy), Atg-5 and Atg-12 (effectors of autophagosome formation through the Atg-12 conjugation system), LC3-b (component of the Atg-8/LC3 conjugation system) and Atg-6/Beclin-1 (component of the lipid kinase complex). At P0 and P03 these genes did not change significantly in DKO hearts, but at P07, their expression (as well as that of Bnip-3, a candidate mediator of mitophagy) was clearly dysregulated (Online Figure VI, panel C). Dysregulation was also evident at the level of protein turnover, because the levels of p62 and LC3-II were found to be significantly elevated in P07 DKO hearts (Figure 7B–E). In contrast, the levels of the lysosomal marker cathepsin-D were unaltered (Figure 7D–E). Taken together, these data suggest that in the present model of heart failure, the activity of the mitochondrial biosynthetic machinery is decreased and at the same time the mitochondrial elimination pathway appears to be halted.

Figure 7. Repression of transcription factors/coactivators of mitochondrial biogenesis and accumulation of early mediators of autophagy in DKO hearts at P07.

A) Relative abundance of target mRNAs at P07. Abbreviations: Nrf-1;Nuclear respirator factor-1, Err-α; Estrogen receptor related-α, Pgc-1α;PPARγ coactivator-1α, Tfam; Transcription factor α mitochondrial B) Western blot analysis of the early autophagic mediator p62. C) Protein levels of p62 in WT and DKO hearts at P07. The arrow on A indicates that only the upper band was quantified in this analysis. D) Protein levels of LC3 (early autophagy marker) and cathepsin-D (late autophagy/lysosomal marker). E) Densitometry-based quantification of LC3-II and cathepsin-D.

Loss of mtDNA and an impaired expression of mitochondrially-encoded genes in the mitofusin deficient heart

The transcriptional repression of mitochondrial and metabolic genes is frequently observed in the context of heart failure⁴⁰. Along these lines, we examined the transcription of genes encoding essential components of the mitochondrial electron transport chain in P0, P03 and P07 DKO hearts. As shown in Online Figure VII, many of the examined genes displayed a progressive downregulation in DKO hearts. Whereas little changes were observed in their expression at P0, an intermediate reduction was evident at P03 and more severe repression was detected at P07. Similarly, a number of genes encoding enzymes important in glycolysis and fatty acid oxidation pathways were transcriptionally repressed in a progressive manner (Online Figure VIII). Notable exceptions in this analysis were Nd-5 and Cyt-b that are encoded by the mitochondrial genome (mtDNA), as they were the only genes to be downregulated already by P0 in DKO hearts (Online Figure VII), thus preceding the onset of severe heart failure. Consistent with these early reductions, Nd-5 and Cyt-b transcripts were robustly downregulated in P07 DKO hearts (Figure 8A and B).

Figure 8. Defects in the maintenance and expression of mtDNA in DKO hearts.

A–B) Transcript levels of genes located in the mtDNA. Nd-5; NADH dehydrogenase subunit- 5, Cyt-b; cytochrome-b, subunit of the bc₁ complex. C) Temporal pattern of mtDNA expansion in WT and DKO hearts. The quantitation was performed in real time according to the ΔΔCt method where the levels of COX-I in WT P03 hearts were used as a reference (value=1). D) Western blot analysis assessing the abundance of COX-I and F1, subunit-α. E) Dual COX/SDH staining in situ in freshly-cut heart sections. Arrows indicate cardiac myocytes where SDH- specific staining predominates. These fields are magnified × 40 and the hearts were from P09 mice. Arrows indicate myocytes with low COX activity and more prevalent SDH activity.

Previous work identified mitofusins to be important determinants of mtDNA maintenance in skeletal muscle ²⁷. In light of the early reductions in the expression of mtDNA genes Nd-5 and Cyt-b, we quantified the levels of mtDNA in heart extracts from WT and DKO mice using real-time PCR. Regarding WT samples, this analyses showed that in the normally developing postnatal heart there is a more than two-fold increase in the levels of mtDNA between P03 and P07 hearts, consistent with ongoing mitochondrial biogenesis (Figure 8C, filled bars). In DKO samples, however, the amount of mtDNA was slightly but significantly decreased at P03 and exhibited substantially more severe reductions by P07 (Figure 8C, empty bars). Thus, the reduction in the mtDNA of postnatal DKO hearts occurs in a progressive manner, and generally coincides with or precedes the onset of heart failure. Consistent with mtDNA instability, the protein levels of COX-I (subunit 1 of the cytochrome c oxidase which is a mitochondrially encoded polypeptide) are severely reduced at P07 (Figure 8D). On the other hand, the protein levels of subunit α, a component of the F₁-ATPase that is encoded by the nuclear genome, displayed little or no reduction in P07 DKO hearts (Figure 8D). Consistent with a reduction in COX-I protein levels, the in situ enzymatic activity of COX was markedly reduced in DKO heart sections, as indicated by the paler staining with DAB in COX/SDH staining assays (Figure 8E).

Mitochondria from WT and DKO hearts at P06–P07 were isolated and assessed for their capacity to produce ATP. In the course of mitochondrial isolation we found that the mitochondrial protein yield is significantly decreased in DKO preparations (Online Figure IX, panel A) despite the observed increase in mitochondrial number by EM (Figure 5). This indicates that the number of extractable mitochondria is markedly lower in P07 DKO hearts, suggesting that the defective mitochondria observed by EM are structurally unstable or that the protein content per mitochondrion is reduced in the DKO. Given this baseline difference, we adjusted the dilution in the different suspensions to contain equal amounts of mitochondrial protein. Assessment of the citrate synthase activity in the extracted mitochondrial samples showed the mitochondrial content to be equal in the different preparations following this adjustment (Online Figure IX, panel B). Next, mitochondria were assessed for their ability to synthesize ATP in the presence of different substrates and after the addition of exogenous ADP. Surprisingly, the maximum rates of ATP synthesis did not differ significantly between WT and DKO mitochondrial preparations (Online Figure IX, panels C–F). In addition, measuring total ATP content in whole hearts did not detect significant decreases in P07 DKO hearts suggesting that the pool of cardiac ATP is maintained even in the context of severe cardiac dysfunction (Online Figure IX, panel G). Taken together, these findings indicate that despite a pervasive mitochondrial deficit, the DKO hearts contain sufficient, functionally-competent mitochondria that allow the heart to maintain normal ATP levels at this time point (P07).

DISCUSSION

The present study highlights the importance of mitofusins (Mfn-1 and Mfn-2) in the early postnatal heart. Mitofusins are mechanoenzymes that support the mitochondrial continuum in cells by catalyzing the integration of the outer mitochondrial membrane, and they are abundantly present in human and other mammalian hearts ^41–44. The particularly high expression of mitofusins in this very oxidative tissue is not surprising because the heart exhibits an extremely high mitochondrial content, e.g. 25% in humans and 38% in mice ⁴⁵. Using a cell-specific gene knockout approach, we find that the combined inactivation of Mfn-1 and Mfn-2 from cardiac myocytes results in a rapidly evolving, early postnatal phenotype that is characterized by major abnormalities in mitochondrial structure, dilated cardiomyopathy and premature death during the second week of life.

The gross morphology of DKO hearts at prenatal developmental stages (e.g. E15.5 and E18.5) does not differ significantly from WT hearts. Furthermore, there are no structural or functional differences between WT and DKO hearts at P0, although changes in mitochondrial morphology are already evident. At this time point, DKO mitochondria assume spherical shapes and are heterogeneous in size. Work with single knockouts showed that cardiac myocytes lacking Mfn-1 have mitochondria with diminished diameters whereas myocytes that lack Mfn-2 have mitochondria with increased diameters ^{30, 31}. In what appears to be a combination of the above phenotypes, we find here that cardiac myocytes lacking Mfn-1 and Mfn-2 contain both enlarged and small mitochondria at P0. Another characteristic of the P0 DKO hearts is the increase in mitochondrial volume density. Increased mitochondrial volume density is also detected in skeletal muscle following dual mitofusin deficiency ²⁷. These findings indicate that the absence of mitofusins favors the expansion of mitochondrial numbers. Along these lines, the increase in mitochondrial numbers may represent an adaptation of the muscle to compensate for dysfunctional mitochondria. However, it is important to note that there are no apparent alterations in the internal structure of P0 DKO mitochondria and no matrix decondensation is observed at this time point. Additionally, myofibril volume density is unaltered, consistent with the finding that heart function of DKO hearts is normal at P0. Thus, it appears that although P0 DKO hearts have preexisting abnormalities in mitochondrial structure and number, they maintain short-term functionality prior to being exposed to the postnatal rigors.

In the developmental window between P0 and P07, the transcripts encoding Mfn-1 and Mfn-2 are markedly upregulated in WT but not in DKO hearts. At the P07 time point, DKO hearts exhibit hallmarks of DCM, including decrease in FS, increase in LV diameters, thinning of ventricular wall and abnormal ECG pattern. These features are also accompanied by poor body condition, atrial congestion, focal collagen deposition, upregulation of the stress markers ANP and BNP, and a decline in survival where no live DKO animals are recovered beyond P16.

The DKO failing hearts present some striking and unique characteristics in cardiac myocyte and mitochondrial ultrastructure. As mentioned above, mitochondria in double-deficient hearts appear to undergo unrestrained expansion and occupy a significantly larger portion of the cellular volume. This is accompanied by defects in the organization of the contractile apparatus and a state of structural disarray specifically in P07 DKO hearts. In addition to appearing as exceedingly small or large spherical organelles, P07 DKO mitochondria have further abnormalities, such as matrix decondensation and loss of cristae structure. In P07 DKO cells, many dilated/swollen mitochondria are evident. These coexist with unusually small round organelles with little or no inner membrane structure. One possibility is that these ‘voids’ result from petite mitochondria that underwent extreme degeneration and cast off their internal or external membrane, or they could represent remnants of late-stage dysfunctional mitochondria that collapse into multiple fragments. Apart from its important role in biochemical pathways and energy conversion, the inner mitochondrial membrane is an integral and quantitatively important part of the myocyte. Therefore, its extreme degeneration/loss that is detected in many P07 DKO mitochondria could be, by itself, an important element contributing to cell and organ dysfunction. Another interesting microscopic observation is that the boundaries between adjacent P07 DKO mitochondria are sometimes found to rearrange into extreme curvatures and form finger-like protrusions. Chains of mitochondria exhibiting these characteristics are uniquely present in DKO myocytes. We currently interpret the above-described projections to reflect abortive fusion attempts of mitofusin null mitochondria. It can be suggested that in the absence of mitofusins other unknown factor(s) are recruited, or become activated as an utmost effort to induce mitochondrial fusion or tethering.

Soon after birth, the heart undergoes rapid increases in mitochondrial content and respiratory function ^{46, 47}. These adaptations are necessary for postnatal cardiac growth and for effective support of the pulmonary and systemic circulation. To coordinate these changes, the heart relies on a specialized transcriptional circuit, into which the co-activator Pgc-1 operates as a central switch ^{48, 49}. Consistent with Pgc-1’s important role in postnatal heart growth, mice with total absence of Pgc-1 (i.e. Pgc-1α^−/− and Pgc-1β^−/−) are born in the expected Mendelian ratios but die shortly after birth ⁴. Similarly, we demonstrate here that mice lacking Mfn-1 and Mfn-2 in cardiac myocytes survive embryonic development and start to die largely during the second week of life. Moreover, we find that the expression of Mfn-1 and Mfn-2 in wild-type hearts is induced from birth and onwards, a period coinciding with Pgc-1’s activation in heart ⁵⁰. The similarities in the temporal pattern of lethality between the two knockout models (i.e. Pgc-1 null vs. mitofusin null) may imply a functional relationship between Pgc-1 and mitofusins. Along these lines, mitofusins are demonstrated to be transcriptionally activated by Pgc-1 in muscle cells ^51–53. Thus, it can be suggested that mitofusins are putative downstream effectors in the mitochondrial biogenic program that is activated by Pgc-1 in postnatal cardiac myocytes.

Apart from the early postnatal phenotype, we also provide evidence for a necessary role of mitofusins in adult mice. This is in agreement with an independent study focusing on the role of mitofusins in the heart ²⁹. Using the MerCreMer, the earlier study reported that the disruption of mitofusins in adult hearts leads to mitochondrial fragmentation and gradual deterioration of heart function ²⁹. Consistent with these findings, using the same Cre expressing strain we show reductions in FS and progressive lethality in mitofusin double-deficient animals (Online Figure I). Chen et al. also reported that cardiac ablation of Mfn-1 and Mfn-2 during embryonic development causes lethality around E10.5 ²⁹. Interestingly, this differs from the early postnatal lethal phenotype that we detect here. A possible explanation for obtaining dissimilar results could be because different promoters were used to induce cre expression in the two studies. The study by Chen et al. used Nkx-2.5cre known to be active in cardiac progenitor cells and therefore has the potential to recombine loxP sites in all cardiac cells including cardiomyocytes, endothelial, smooth muscle and mesenchymal cells of the heart ⁵⁴. By contrast, the expression of the Myh6-cre transgene used in the current study is restricted only to cardiac myocytes ^{37, 54, 55}. Thus, one possibility is that the embryonic lethality observed by the earlier study is due to the inactivation of mitofusins in the non-myocyte compartment of the heart. In this regard, mitofusins have been shown to be important for vascular endothelial cell differentiation, survival and function ⁵⁶. Apart from differences in cell specificity, there are differences in the timing of mitofusin inactivation between the two studies. Nkx-2.5cre begins gene excision at E7.5⁵⁷, whereas Myh6cre begins excision at E9.5. Thus an alternative possibility is that early inactivation of mitofusins by Nkx-2.5cre elicits a severe effect in the developing heart as this period encompasses important transitions such as heart tube formation and looping. On the other hand, the utilization of Myh6-cre is more informative in interpreting the role of mitofusins, specifically in the cardiac myocyte, during mid-gestation and early postnatal development of the heart.

The evidence from this study suggest that mitofusins, and by extension mitochondrial fusion, are important for the formation of functional mitochondria in cardiac myocytes during the P0–P07 period. Conceptually, mitochondrial fusion is beneficial during this period of unprecedented high energetic demand because it allows the diffusion of metabolites in different compartments of the network and promotes more efficient energy conversion ⁵⁸. In this regard, it is demonstrated that fusion is activated during mild intracellular stress or nutrient deprivation to enhance mitochondrial ATP production ^{59, 60}. Membrane fusion has also been suggested to be important for genome exchange between mitochondria and facilitate mtDNA complementation ⁶¹. Consistent with a relationship between mitochondrial fusion and mitochondrial genome maintenance, previous work has established a link between mitofusins and mtDNA copy-number in skeletal muscle ²⁷. In agreement, we find here that loss of mitofusin activity is accompanied by reductions in mtDNA. This reduction in mtDNA appears to preexist cardiac dysfunction, although it is likely to be further exacerbated as the hearts begin to fail.

It is also conceivable that mitofusins are important during the P0–P07 developmental stage through mechanisms that do not involve fusion and content exchange between mitochondria. One possibility is that mitofusins allow actively proliferating mitochondria to establish connections with the cytoskeleton and correctly position themselves into the myofibrillar blueprint during cardiomyocyte postnatal development. Along these lines, molecular defects in the assembly of the cytoskeleton in cardiac myocytes are common causes of dilated cardiomyopathy (DCM) in mice and in humans ^62–64. We find here that the loss of mitofusins is accompanied by rapid and lethal DCM, which may indicate an underlying defect in cytoskeletal function. The possible interaction of mitofusins with the cytoskeleton has been recently addressed in neurons, where Mfn-1 and Mfn-2 were shown to interact with adaptor proteins that tether the OMM to microtubules ⁶⁵. Regarding striated muscle, the interaction of mitochondria with intermediate filaments (IFs) is noted and genetic defects in the assembly of IFs are sometimes accompanied by dramatic changes in mitochondrial morphology in hearts ^{66, 67}. Therefore, there is emerging evidence to support a potential role of mitofusins at the mitochondrion/cytoskeleton interface and perturbations in this association may have contributed to the developmental abnormalities that we identify here.

In both the P0 and P07 DKO hearts there is an overpopulation of mitochondria indicating altered mitochondrial biogenesis. This increase in numbers could reflect a compensatory response to an underlying mitochondrial dysfunction. Nevertheless, there is no significant evidence for the activation of biosynthetic pathways regulating the formation of new mitochondrial components. This is supported by the finding that in DKO hearts the Pgc-1 transcriptional pathway is either unchanged or suppressed, as is the transcription of many genes that encode key mitochondrial proteins. On the other hand, increased mitochondrial numbers in DKO hearts may originate from a defect in the elimination of excessive mitochondria through mitophagy. Recent work indicates that Mfn2 is important for the autophagy pathway ^{68, 69}. However, in the present model, we identify significant evidence of altered autophagy that coincides with the onset of heart failure, but not prior. This evidence includes aberrant gene expression of key autophagy mediators (i.e. Ulk-2, Atg-5, Atg-12, Beclin1, Bnip-3) and accumulation of p62 and LC3-II proteins at P07. Thus, in this DKO model, heart failure at P07 appears to be associated with a diminished ability to synthesize new mitochondrial components as well as with a reduction in the ability to remove mitochondria. Given this reciprocal relationship, it could be suggested that these alterations (i.e. decreased autophagy and a concomitant decrease in de novo biogenesis) may be part of a global mechanism that operates to maintain or prolong cardiac homeostasis in the stressed heart. Along these lines, it is widely appreciated that the heart has a remarkable ability to maintain stable levels of ATP despite varying workloads ⁷⁰. Here, we found that despite the drastic mitochondrial deficit of P07 DKO hearts, isolated mitochondria were capable of producing maximum amounts of ATP and the total levels of ATP in these hearts did not change. Thus, it is tempting to speculate that ATP levels are maintained in the face of mitofusin deficiency by mechanisms that globally adjust rates of mitochondrial biogenesis and mitophagy as the heart begins to fail.

Novelty and Significance.

What is Known?

• Mitofusins are novel regulators of mitochondrial structure and function in the heart.

• The hearts of newly-born mammals exhibit rapid increases in mitochondrial biomass.

• The surge in mitochondrial biogenesis coupled with mitochondrial maturation is thought to be key for the physiological growth of the postnatal heart.

What New Information Does This Article Contribute?

• Mouse hearts exhibit notable upregulation of mitofusin genes at the time of birth and onwards.

• Aberrantly expanding mitochondria lacking mitofusins fail to develop appropriately in postnatal myocytes; they undergo accelerated degeneration, leading to organ dysfunction and premature death.

• Mitofusins are pivotal moderators of acute mitochondrial biogenesis and maturation that is activated in the early postnatal heart.

Birth triggers mitochondrial biogenesis and maturation in heart but the molecular mechanisms that control this developmental stage are poorly understood. In the present study, we addressed whether mitofusins are important during the acute phase of mitochondrial remodeling that occurs in hearts after birth. We found that inactivating mitofusins profoundly alters the number, morphology and distribution patterns of mitochondria in cardiac myocytes. Although mitofusin deficiency is benign for heart function up until birth, the mitofusin-deficient mitochondria undergo marked degeneration and hearts exhibit signs of dysfunction as early as postnatal day four followed by premature death. This study provides a link between mitofusins and the metabolic reprogramming of the newborn heart, and shows that mitofusins function at the interface between deregulated mitochondrial biogenesis and cardiac dysfunction.

Non-standard Abbreviations

Myh

Myosin heavy chain

DCM

Dilated cardiomyopathy

Mfn

Mitofusin

mtDNA

mitochondrial DNA

GTP

Guanosine triphosphate

PGC-1

PPARγ coactivator 1