The metabolic state of the heart regulates mitochondrial supercomplex abundance in mice

Abstract

Mitochondrial supercomplexes are observed in mammalian tissues with high energy demand and may influence metabolism and redox signaling. Nevertheless, the mechanisms that regulate supercomplex abundance remain unclear. In this study, we examined the composition of supercomplexes derived from murine cardiac mitochondria and determined how their abundance changes with substrate provision or by genetically induced changes to the cardiac glucose-fatty acid cycle. Protein complexes from digitonin-solubilized cardiac mitochondria were resolved by blue-native polyacrylamide gel electrophoresis and were identified by mass spectrometry and immunoblotting to contain constituents of Complexes I, III, IV, and V as well as accessory proteins involved in supercomplex assembly and stability, cristae architecture, carbohydrate and fat oxidation, and oxidant detoxification. Respiratory analysis of high molecular mass supercomplexes confirmed the presence of intact respirasomes, capable of transferring electrons from NADH to O₂. Provision of respiratory substrates to isolated mitochondria augmented supercomplex abundance, with fatty acyl substrate (octanoylcarnitine) promoting higher supercomplex abundance than carbohydrate-derived substrate (pyruvate). Mitochondria isolated from transgenic hearts that express kinase-deficient 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase (Glyco^Lo), which decreases glucose utilization and increases reliance on fatty acid oxidation for energy, had higher mitochondrial supercomplex abundance and activity compared with mitochondria from wild-type or phosphatase-deficient 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase-expressing hearts (Glyco^Hi), the latter of which encourages reliance on glucose catabolism for energy. These findings indicate that high energetic reliance on fatty acid catabolism bolsters levels of mitochondrial supercomplexes, supporting the idea that the energetic state of the heart is regulatory factor in supercomplex assembly or stability.

Graphical abstract

Highlights

• •Cardiac mitochondrial supercomplexes can function as intact respirasomes.
• •Supercomplexes contain accessory proteins associated with diverse cellular functions.
• •Mitochondrial substrate availability influences supercomplex abundance ex vivo.
• •Conditions of high fat oxidation promote higher levels of mitochondrial supercomplexes in vivo.

Abbreviations:

AFG3L2

ATPase family member 3 like matrix AAA peptidase subunit 2

ATP5A

ATP synthase subunit alpha

BN-PAGE

blue native polyacrylamide gel electrophoresis

CN-PAGE

clear native polyacrylamide gel electrophoresis

DDM

dodecyl-β-D-maltoside

DIG

digitonin

LBN-PAGE

light blue native PAGE

MCAD

medium chain acyl CoA dehydrogenase

MIC60

MICOS complex subunit (also called mitofilin)

MTCO1

cytochrome c oxidase subunit 1

NDUFB8

NADH:ubiquinone oxidoreductase subunit 8B

NDUFA9

NADH:ubiquinone oxidoreductase subunit A9

OCM

octanoylcarnitine and malate

OGDC

oxoglutarate dehydrogenase complex

PDHC

pyruvate dehydrogenase complex

PFK1

phosphofructokinase 1

PFK2

phosphofructokinase 2 (also denoted 6-phosphofructo-2-kinase/fructo-2,6-bisphosphatase)

PHB

prohibitin

PM

pyruvate and malate

OCR

oxygen consumption rate

ROS

reactive oxygen species

SAMM50

sorting and assembly machinery component 50

SDHA

succinate dehydrogenase A

SDHB

succinate dehydrogenase subunit B

SOD2

superoxide dismutase 2

STML2

stomatin-like protein 2

TFP

trifunctional protein

TUFM

mitochondrial Tu translation elongation factor

UQCRC2

ubiquinol-cytochrome c reductase core protein 2

VLCAD

very long chain acyl CoA dehydrogenase

XF

extracellular flux

1. Introduction

Observations over the past 40 years provide evidence that metabolic enzymes are not homogenous in cells and that many metabolic pathways function within dynamic enzyme assemblies. These assemblies, termed metabolons, have been hypothesized to provide microenvironments where substrates or electrons are transferred successively between enzymes, which could improve pathway flux and efficiency [1,2]. Nevertheless, it remains unclear how such metabolic enzyme assemblies form and how they regulate metabolism and biology.

Among the most well-known metabolons are mitochondrial supercomplexes. These are large (>1 MDa), non-covalently associated protein complexes typically consisting of Complexes I, III, and/or IV [3,4]. The abundance of mitochondrial supercomplexes is diminished under conditions of heart failure [5], diabetes [6] and aging [7], and is elevated in response to physiological stressors such as exercise [8]. Although it is thought that mitochondrial supercomplexes bestow kinetic advantages through electron channeling [[9], [10], [11]], there remains debate whether supercomplexes enhance catalysis and efficiency [[12], [13], [14]]. Other potential roles for supercomplexes include respiratory chain stability or control over reactive oxygen species (ROS) production [[15], [16], [17]]. Beyond mitochondrial supercomplexes, other metabolons present in cells and tissues include enzymes involved in the Krebs cycle [18,19], branched-chain amino acid metabolism [20,21], fatty acid oxidation [22,23], nucleotide synthesis [24], and glycolysis [[25], [26], [27]], among others. Notwithstanding, how such metabolons form and how they affect tissue form and function remain poorly understood.

In this study, we tested the hypothesis that substrate utilization regulates mitochondrial supercomplex abundance. Guided by findings that suggest that metabolon formation may be regulated by substrate gradients [28], we examined how mitochondrial supercomplex abundance changes under conditions of low or high fatty acid oxidation and carbohydrate utilization in vivo and in vitro. For this, we used isolated mitochondrial preparations treated with different substrates and exploited mice with high or low phosphofructokinase activity, which regulates the cardiac glucose-fatty acid cycle [29]. We find that conditions of high fat oxidation appear to augment mitochondrial supercomplex levels. Collectively, these findings provide new understanding of the composition of mitochondrial supercomplexes in the heart and of the mechanisms that control their abundance.

2. Experimental procedures

Materials and Reagents: All materials and reagents were from Sigma-Aldrich, unless noted otherwise.

Mouse models: All procedures were approved by the University of Louisville Institutional Animal Care and Use Committee and were in accordance with NIH guidelines. Transgenic mice expressing a kinase-deficient (i.e., Glyco^Lo mice) or phosphatase-deficient (i.e., Glyco^Hi mice) form of phosphofructo-2-kinase/Fru-2,6-P₂ bisphosphatase (PFK2) under the control of the α-MHC promoter and wild-type (WT) littermates on the FVB/NJ background were used for this study [[29], [30], [31]]. For the purposes of this study, all mice used were male and between 12 and 20 weeks of age. Food and water were provided ad libitum, and the mice were maintained on a 12:12-h light-dark schedule. Prior to tissue harvest, mice were anaesthetized with sodium pentobarbital (150 mg/kg, i.p.), followed by euthanasia via excision of the heart. The euthanasia procedures were consistent with the AVMA Guidelines for the Euthanasia of Animals.

Mitochondrial isolation: Heart mitochondria were isolated similar to that described previously [29,32,33]. Briefly, whole hearts were washed 5 × with ice cold buffer A (220 mM mannitol, 70 mM sucrose, 5 mM MOPS, 1 mM EGTA; pH 7.2 with KOH) followed by homogenization using a Teflon-coated Glass-Col homogenizer in 2 ml of buffer A containing 0.2% fatty-acid-free BSA. Homogenate was then subjected to centrifugation at 800g for 10 min followed by supernatant collection and centrifugation at 9,000g for 15 min. The pellet containing mitochondria was then resuspended in 1 ml of buffer A (without BSA) and centrifuged at 10,000g, with this step repeated once. For some experiments, the washed mitochondrial pellet was then resuspended in respiration buffer (120 mM KCl, 25 mM sucrose, 10 mM HEPES, 1 mM MgCl₂, 5 mM KH₂PO₄; pH 7.2) and kept on ice.

Blue-native PAGE separation of protein complexes: Multimeric enzyme complexes were separated in their native state using Blue Native PAGE (BN-PAGE), similar to that described [34]. Briefly, isolated mitochondria were solubilized in either n-dodecyl-β-D-maltoside (0.5%) or digitonin (6:1 DIG:mitochondrial protein, w/w) prior to protein separation by BN-PAGE. For experiments using digitonin, we prepared a stock of 5% digitonin in 0.75 M aminocaproic acid in 50 mM Bis-Tris, pH 7.0. The mitochondrial pellets were then solubilized in the digitonin-containing buffer, with addition of 1:100 protease inhibitor cocktail (Sigma), and, after centrifugation at 16,000g for 15 min at 4 °C, the supernatant was collected.

For protein complex separation, BN-PAGE gels were prepared using polyacrylamide (acrylamide:bis-acrylamide, 37.5:1) at either a 5–15% or a 3–12% gradient, similar to that described [34]. The cathode buffers contained 50 mM Tricine, 15 mM Bis-Tris, pH 7.0, and 0.02% Coomassie Blue G250 (high blue buffer) or 0.004% Coomassie Blue (low blue buffer). The anode buffer consisted of 50 mM Bis-Tris pH 7.0. Electrophoresis was performed at 4 °C using high blue buffer at 100 V for 1 h, followed by low blue buffer at 250 V for 1.5 h. For assessing supercomplex abundance in Coomassie-stained gels, we loaded 70 μg mitochondrial protein per lane. For separation in the second dimension, either the entire lane or individual bands were excised from BN-PAGE gels and immersed in Tris buffer containing 1% SDS and 1% 2-mercaptoethanol for 20 min at 37 °C. The gel lane or excised bands were then placed on top of a 12% SDS-PAGE resolving gel, followed by Western blotting with the indicated antibodies.

Mass spectrometric protein identification: Protein excised from BN-PAGE gels were identified by liquid chromatography (LC) ESI MS/MS, after the in-gel trypsin digestion. For in-gel digestion, Coomassie-stained BN-PAGE gel bands were cut into 1-mm³ plugs and incubated in 100 mM triethylammonium bicarbonate (TEA-BC; Sigma) at room temperature for 15 min. Acetonitrile was then added to the TEA-BC solution, and the gel plugs were incubated at room temperature for 15 min with gentle vortexing. The solvent was removed, and the washing process was repeated until the Coomassie Blue stain was no longer visible. Solvent was removed, and the gel plugs were dried in a SpeedVac for 5 min. The dried plugs were incubated in DTT (20 mM DTT, 100 mM TEA-BC) at 56 °C for 45 min, followed by iodoacetamide (55 mM iodoacetamide, 100 mM TEA-BC) at room temperature for 30 min. Iodoacetamide was removed, and gels were washed in 50 mM TEA-BC at room temperature for 15 min, followed by gentle vortexing in acetonitrile for 15 min at room temperature. The gel plugs were again dried for 5 min in a SpeedVac and incubated in digestion buffer [20 ng/μl modified Trypsin (Promega) in 50 mM TEA-BC] for 10 min. Then, 50 mM TEA-BC was added to the plugs, followed by incubation at 37 °C overnight in a shaker.

The digested peptides were resuspended in buffer A (5% ACN and 0.1% FA in water) and loaded onto an Ultimate 3000 nanoLC system (Thermo Fisher Scientific) with a C18 column (75 μm × 35 cm × 1.7 μm, prepared in-house). The peptides were separated via gradient elution at 500 nL/min, starting from 5 to 25% of buffer B (95% ACN, 0.1% FA) in 150 min, raised to 35% in 10 min, then raised to 80% in 5 min. Gradient elution was then maintained at 80% with Buffer B for 5 min, before returning to buffer A. The eluted peptides were directly subjected to tandem mass spectrometry on a Q EXACTIVE HF (Thermo Fisher Scientific) for peptide identification, in which the detection mode was set at data-dependent acquisition and the relevant parameters were programmed as: electrospray voltage, 1.6 kV; precursor scan range, 400–1250 m/z at a resolution of 120,000 in Orbitrap; and MS/MS fragment scan range, >100 m/z at a resolution of 30,000 in higher-energy C-trap dissociation mode. The acquired MS data was analyzed with MaxQuant (version 1.5.8.0), and the MS/MS spectra files were searched with the integrated Andromeda search engine against the mouse protein database (UniProt, 79,207 entries) with the following parameter settings: digestion specificity of trypsin with tolerance at 2 missed cleavages; carbamidomethylation of cysteine; oxidation of methionine; deamidation of asparagine and glutamine; and N-terminal acetylation of glutamine. Peptide spectrum matches and proteins were retained with a false discovery rate (FDR) < 0.01.

Respirasome activity assays: The respiratory activity of protein complexes migrating in Native PAGE gels was assessed by extracellular flux (XF) analysis. Mitochondrial proteins were separated by colorless-native PAGE (CNE) [35]. A parallel CNE gel was stained with Coomassie Blue. Gel plugs from high molecular weight regions of the CNE gel were placed underneath the capture screens of XF24 islet plates and incubated with respiration buffer. After baseline oxygen consumption rate (OCR) recordings, NADH (2 mM), ADP (4 mM), cytochrome c (10 μM), and rotenone (2 μM) were added sequentially to each well, with OCR measurements occurring after addition of each compound.

In-gel mitochondrial complex activity assays: The activity of mitochondrial respiratory complexes was measured using in-gel enzymatic assays, similar to that described [36]. Mitochondrial protein (50 μg) was separated by light BN-PAGE, which uses only 0.004% Coomassie G250 in the cathode buffer. The sample buffer contained Coomassie Blue G250 (0.5%). After electrophoresis, the activity of mitochondrial complexes was assessed in the polyacrylamide gel. Briefly, the gels were washed in 50 mM phosphate buffer (pH 7.40) and then incubated with the following substrates: for complex I activity, 2 mM Tris-HCl (pH 7.40) containing 1 mg/ml NADH and 2.5 mg/ml of nitroblue tetrazolium; for complex IV activity, 45 mM phosphate buffer (pH 7.40) containing 1 mg/ml cytochrome c and 0.5 mg/ml 3,3′-diaminobenzidine; for complex II, 5 mM Tris-HCl (pH 7.40) containing 20 mM sodium succinate and 0.2 mM phenazine methasulphate. The reactions were performed at room temperature and stopped by fixing the gels for 1 h in solution containing 45% methanol and 10% acetic acid. The gels were stored in a 10% acetic acid solution prior to imaging. The specificity of each reaction was demonstrated using inhibitors for each complex, i.e., rotenone for complex I, cyanide for complex IV, and malonate for complex II.

Substrate-dependence of supercomplex abundance: Mitochondria from freshly isolated wild-type mouse hearts were incubated in respiration buffer with either 5 mM pyruvate, 2.5 mM malate and 1 mM ADP or 100 μM octanoyl-l-carnitine, 2.5 mM malate, and 1 mM ADP. Another group was incubated without substrate. The mitochondrial suspension (200 μg mitochondrial protein in 300 μl respiration buffer) was incubated with substrates and protease inhibitor at 37 °C for 10 min, with intermittent mixing every 2 min. The tubes were then placed on ice for 5 min and centrifuged at 10,000g for 10 min. Mitochondrial pellets were solubilized with digitonin (6:1 w/w) and the lysates were subjected to BN-PAGE and BN-PAGE Western blotting.

Visualization of supercomplexes: Protein complexes were visualized by Coomassie staining and Western blotting. For Coomassie staining, BN-PAGE gels were incubated in a solution containing 50% methanol, 10% acetic acid, and 0.25% Coomassie Blue R-250 for 2–4 h. The gels were then destained in solution containing 50% methanol, 10% acetic acid, and 40% deionized water for 10 min. Background staining was removed further by washing the gels in 7% acetic acid.

For BN-PAGE Western blots, we loaded 2–3 μg of mitochondrial protein, and, after electrophoresis, the BN-PAGE gel was immersed in Tris-buffered saline (TBS) containing 1% SDS and 1% 2-mercaptoethanol at 37 °C for 20 min. The gels were then washed twice in TBS, followed by transfer to PVDF membranes. After transfer, the membranes were air-dried, immersed in a mixture of 10% acetic acid and 50% methanol for 1 min, and then washed twice with 100% methanol. Western blotting was then performed as described [29,33]. Antibodies used included: anti-NDUFA9 antibody (1:2000, Abcam), anti-succinate dehydrogenase A antibody (1:1000, Abcam), and the MitoProfile® Total OxPhos Rodent Antibody Cocktail (1:2000, Abcam), anti-Mic60 (1:1000, Abcam).

Statistical analysis: Data are presented as mean ± SD. Multiple groups were compared using One-way ANOVA, followed by Tukey post-tests. A p value ≤ 0.05 was considered significant.

3. Results

Supercomplex detection in mouse heart: To examine supercomplex formation in the heart, we first optimized the conditions for separation of native complexes in polyacrylamide gels. Using blue native (BN)-PAGE, we compared the relative levels of high molecular weight complexes in cardiac mitochondria solubilized with dodecyl-β-D-maltoside (DDM) or digitonin (DIG). As shown in Fig. 1A, few complexes with molecular masses (M_r) higher than 700 kDa were found in DDM-solubilized mitochondria; however, high molecular mass complexes were preserved in DIG-solubilized mitochondria.

Fig. 1.

Standardization of BN-PAGE for examining multimeric complexes in cardiac mitochondria. (A) Representatative Coomassie-stained 1D, non-reducing Blue Native Polyacrylamide Gel Electrophoresis (BN-PAGE) gel of isolated mouse cardiac mitochondria solubilized with β-dodecylmaltoside (DDM) or digitonin (DIG); (B) Representative non-reducing, Coomassie-stained gel (top), and anti-MitoProfile immunoblot of proteins separated in the 2nd dimension (bottom).

To determine if the high molecular weight proteins were constituents of supercomplexes, we further resolved the complexes in the 2nd dimension under reducing and denaturing conditions and immunoblotted for the presence of complexes I–V using the MitoProfile antibody cocktail. Consistent with their inclusion in supercomplexes, immunoreactive regions in high-molecular-mass regions of the BN-PAGE gel were positive for subunits of Complex I (NDUFB8), III (UQCRC2), IV (MTCO1), and complex V (ATP5A) (Fig. 1B). The complex II subunit, succinate dehydrogenase B (SDHB), was absent in high-molecular-weight regions on the gel.

To gain more insight into the individual components of the mitochondrial supercomplexes, we changed the polyacrylamide composition from 5–15% to 3–12%, which allowed further separation of complexes of high molecular mass, and excised the six most abundant bands having an apparent M_r higher than the complex I (C–I) band. The constituent proteins in the supercomplex bands were then identified by mass spectrometry (Fig. 2A). As shown in the Supplementary Material, each of these bands were replete with subunits of Complexes I–V as well as cytochrome c. To construct rudimentary models of the supercomplexes (Fig. 2B), we included proteins that showed ≥10 unique MS/MS spectra. Using this criterion, we identified several proteins that appeared to bind tightly to the supercomplexes, including: components of the pyruvate dehydrogenase complex (PDHC), superoxide dismutase 2 (SOD2), the oxoglutarate dehydrogenase complex (OGDC), trifunctional protein (TFP), MICOS complex subunit (Mic60), sorting and assembly machinery component 50 (Samm50), prohibitin (PHB), and mitochondrial Tu translation elongation factor (TUFM), among others. In addition, some complexes also contained proteins such as voltage-dependent anion channel 1, enolase 3, and sarcoplasmic reticulum calcium ATPase 2a, which suggests confluence of inner membrane supercomplexes with enzymes critical for cytosolic metabolism and calcium dynamics. To further confirm the identity of the supercomplexes, we excised bands 1–6 from separate BN-PAGE gels, separated protein constituents under denaturing and reducing conditions, and subjected them to immunoblotting using the MitoProfile antibody cocktail. As shown in Fig. 2C, each of the six bands contained subunits from Complexes I, III, IV, and V. Components of complex II were generally absent from the complexes.

Fig. 2.

Identification of mitochondrial respiratory supercomplexes in murine heart. (A) Representative image of 1D-BN-PAGE showing bands excised for LC/MS/MS protein identification. (B) Putative respiratory supercomplexes corresponding to bands 1–6 in panel A. Identified proteins having at least 10 unique spectra were used to create the models shown here. Additional proteins that may be associated with the complexes are shown based on the abundance of unique spectra are shown; and (C) In separate experiments, bands corresponding to 1–6 in Panel A were separated in the 2nd dimension under reducing and denaturing conditions and subjected to Western blot analysis using the MitoProfile antibody.

To assess whether some of these complexes are functional respirasomes, we measured oxygen consumption in gel plugs excised from the native polyacrylamide gel. Because the Coomassie stain used to impart negative charge can diminish enzyme function [35], we used a colorless-native PAGE approach (CN-PAGE) to separate protein complexes. This technique relies on intrinsic protein charge to promote migration of proteins to the anode. Although the CN-PAGE gel stained with Coomassie Blue after electrophoresis demonstrated less resolution than typical BN-PAGE gels, excised regions of a parallel unstained gel could then be used to assess the activity of complexes migrating to regions higher than 1240 kDa, complexes migrating to approximately 1200 kDa, complexes migrating to approximately 1050 kDa, and complexes less than 1000 kDa. The gel plugs were placed underneath capture screens in XF24 islet plates and incubated with respiration buffer (Fig. 3A). Upon addition of NADH, we observed increases in oxygen consumption in all gel plugs except from regions below 1000 kDa. As expected, addition of ADP did not further stimulate oxygen consumption; however, addition of cytochrome c augmented respiratory activity, and rotenone fully inhibited oxygen consumption (Fig. 3B and C). These data complement our characterization of supercomplexes and demonstrate that they are capable of respirasome activity. We note that, because the bands cannot be visualized directly with the CN-PAGE approach, this assay is likely to provide only qualitative assessments of respirasome activity.

Fig. 3.

Validation of functional respirasomes. As a measure of respirasome activity, digitonin-solubilized mitochondria were subjected to BN-PAGE in the absence of Coomassie Blue-G250 dye. (A) After native separation, oxygen consumption in regions corresponding to high molecular weight regions of the unstained gel was assessed by XF analysis; (B) O₂ traces after addition of 2 mM NADH, 4 mM ADP, 10 μM cytochrome c, and 2 μM rotenone (Rot) to each well; and (C) O₂ consumption rates of gel plugs corresponding to protein complexes of different molecular masses.

Respiratory substrates regulate supercomplex abundance: To determine whether substrate provision might impact supercomplex abundance, we isolated cardiac mitochondria from wild-type mice and incubated them without substrates or with octanoylcarnitine and malate (OCM) or pyruvate and malate (PM), which simulates fatty acid oxidation or carbohydrate oxidation, respectively. Following a 10-min incubation with each substrate mix, we sedimented the mitochondria and prepared them for separation via BN-PAGE. As shown in Fig. 4A and B, the presence of either substrate mix led to higher mitochondrial supercomplex levels than that observed in mitochondria incubated in the absence of substrate. To address this further, we measured supercomplex abundance by immunoblotting. We found that only OCM, and not PM, increased NDUFA9 immunoreactivity in the supercomplex regions (Fig. 4C and D).

Fig. 4.

Respiratory substrates regulate mitochondrial supercomplex abundance. Isolated cardiac mitochondria from naïve, WT mice were incubated with no substrate (none), 100 μM octanoylcarnitine, 2.5 mM malate and 1 mM ADP (OCM), or 5 mM pyruvate, 2.5 mM malate and 1 mM ADP (PM) for 10 min at 37 °C. The mitochondria were then solubilized with digitonin, and the proteins were separated by BN-PAGE. (A) Representative Coomassie-stained BN-PAGE gel of cardiac mitochondrial proteins. Shown are three representative mice from each treatment; (B) Quantification of supercomplex band density under different substrate conditions. n = 9 mice per group, **p < 0.01, ***p < 0.001; (C) Representative Western blot of BN-PAGE gel: Antibodies against the Complex I subunit, NADH:Ubiquinone Oxidoreductase Subunit A9 (NDUFA9), were used to detect supercomplexes containing Complex I. The blot was reprobed with antibodies against succinate dehydrogenase A (SDHA), which was used as a loading control. Shown are three representative biological replicates; (D) Quantification of NDUFA9-containing SC bands, normalized to SDHA. n = 9 mice per group, ***p < 0.001.

Mitochondria from Glyco^Lohearts have higher levels of respiratory supercomplexes: To determine if differences in cardiac metabolism in vivo might affect mitochondrial supercomplex abundance, we next examined supercomplex abundance in mitochondria isolated from mice that constitutively express kinase- or phosphatase-deficient 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase in the heart; these mice are termed Glyco^Lo or Glyco^Hi mice, respectively, and were shown previously to influence the glucose-fatty acid cycle in the heart [29]. As shown in Fig. 5A, mitochondria from Glyco^Lo hearts appeared in Coomassie-stained BN-PAGE gels to have higher levels of supercomplexes than mitochondria from WT or Glyco^Hi hearts. Quantification of the supercomplex regions indicated significantly higher levels of mitochondrial supercomplexes in Glyco^Lo hearts versus Glyco^Hi hearts (Fig. 5B). To further discern differences, we examined supercomplex abundances by BN-PAGE immunoblotting. As shown in Fig. 5C and D, we observed significantly more immunoreactivity with anti-NDUFA9 antibodies in high-molecular-weight regions of gels from Glyco^Lo-derived mitochondria than from WT- and Glyco^Hi-derived mitochondria.

Fig. 5.

The glucose-fatty acid cycle influences respiratory supercomplex abundance in murine heart. BN-PAGE-mediated detection of changes in mitochondrial supercomplexes from digitonin-solubilized cardiac mitochondria from WT, Glyco^Lo and Glyco^Hi mice: (A) Representative Coomassie-stained, BN-PAGE gel of isolated cardiac mitochondria from WT, Glyco^Lo, and Glyco^Hi mice. Shown are three representative samples from each genotype; (B) Quantification of supercomplex (SC) band density: SC bands were normalized to the Complex III (C-III) band in each lane; the asterisk (*) in panel A represents the C-III band to which the SC region were normalized. n = 9–12 mice per group, *p < 0.05; (C) Representative Western blot of BN-PAGE gel: Antibodies against the Complex I subunit, NADH:Ubiquinone Oxidoreductase Subunit A9 (NDUFA9), were used to detect supercomplexes containing Complex I. The blot was reprobed with antibodies against succinate dehydrogenase A (SDHA), which was used as a loading control. Shown are three representative mice from each genotype; (D) Quantification of NDUFA9-containing SC bands, normalized to SDHA. n = 9–12 mice per group, *p < 0.05, **p < 0.01; (E) In-gel mitochondrial complex activity assays: Gels were incubated with suitable chemical substrates that change color and form a precipitate due to the enzyme activities of respiratory complexes. The activities of Complex I, Complex IV, and Complex II are shown; and (F) Quantification of respiratory complex activity in the SC region of the BN-PAGE gels. The intensity of staining was normalized to the Complex III band of the gel after Coomassie blue staining. n = 3 mice per group, *p < 0.05.

To confirm these changes at the functional level, we separated supercomplexes using a light blue native PAGE approach (LBN-PAGE), which affords improved separation over CN-PAGE approaches and minimizes loss of enzymatic function, and examined respiratory complex activity. As shown in Fig. 5E and F, the supercomplex regions of complex I and complex IV activity gels showed modestly higher staining. As predicted from results in Fig. 2, no complex II activity was observed in the supercomplex regions of the complex II activity gels. Collectively, these findings indicate that mitochondria from Glyco^Lo hearts, which rely more on fatty acid oxidation for energy, have higher levels of mitochondrial supercomplexes than WT or Glyco^Hi hearts.

4. Discussion

Our results provide new understanding of the composition of cardiac mitochondrial supercomplexes and how fuel utilization regulates mitochondrial supercomplexes in the heart. We found that acute provision of fatty acyl substrates to isolated mitochondria is sufficient to increase mitochondrial supercomplex abundance to levels higher than that of carbohydrate-derived substrates, which indicates that supercomplex abundance may be regulated in a substrate-specific manner. Indeed, mitochondria from hearts that rely more on fats for energy (Glyco^Lo hearts) were found to have higher levels of mitochondrial supercomplexes than mitochondria from hearts that rely on glucose oxidation for energy (Glyco^Hi). These findings suggest that conditions of high fat oxidation may be sufficient to increase mitochondrial supercomplex abundance.

Our findings are consistent with studies showing that deletion of critical fatty acid oxidation enzymes, such as medium-chain acyl CoA dehydrogenase (MCAD) and very-long-chain acyl CoA dehydrogenase (VLCAD), disrupts mitochondrial supercomplexes [23,37]. Those studies suggested that loss of fatty oxidation enzymes results in structural destabilization of mitochondrial supercomplexes, implicating direct protein-protein interactions in supercomplex stability and abundance. Consistent with this idea, we find that both the α and β subunits of mitochondrial trifunctional protein (TFP), which catalyze the last three steps of mitochondrial β-oxidation of long chain fatty acids, were present in each supercomplex band. Our studies using genetically modified mice expressing kinase-deficient or phosphatase-deficient point-mutant isoforms of PFK2 enabled testing of the role of metabolic activity in supercomplex formation. The transgenes modulate an allosteric regulator (i.e. fructose 2,6-bisphosphate) of phosphofructokinase 1 and do not directly alter the abundance of enzymes in glycolysis. These mice, termed Glyco^Lo and Glyco^Hi mice, show constitutively lower or higher glucose utilization rates, respectively, and modulate mitochondrial substrate utilization by exploiting the cardiac glucose-fatty acid (Randle) cycle [[29], [30], [31],38,39]. Mitochondria isolated from Glyco^Lo hearts, which are poised to catabolize fat for energy, were found to have higher levels of supercomplexes than mitochondria isolated from Glyco^Hi mice, which rely more on glucose catabolism for energy. These data appear to indicate that fatty acid oxidation is a nidus for supercomplex formation, an idea further supported by our finding that provision of fatty acyl substrate heightens supercomplex levels more than pyruvate, a carbohydrate-derived substrate. Thus, our results support the concept that metabolic pathway activity can modulate the abundance of supercomplexes.

Substrate-driven chemotactic assembly of enzyme cascades is a plausible mechanism by which metabolic activity regulates supercomplex formation. This phenomenon has been shown to occur for other enzymatic cascades, such as those in glycolysis [28]. According to this model, the presence of initial substrate for an enzymatic pathway enhances single-enzyme diffusion as well as results in each enzyme following substrate gradients, which favors protein-protein interactions [28,40,41]. Applying this mechanism to supercomplex formation, it is possible that electron flow and the levels of reducing equivalents produced during substrate catabolism could be factors that initiate and sustain supercomplex formation. Indeed, mitochondria incubated without substrate ex vivo showed much lower levels of supercomplex formation than mitochondria incubated with substrate, and octanoylcarnitine provision promoted higher levels of supercomplex formation than provision of pyruvate. Nevertheless, it remains unclear whether the presence of substrate acutely increases supercomplex assembly or whether it stabilizes already-formed supercomplexes.

Our proteomic characterization of mitochondrial supercomplexes revealed several interesting candidate proteins that could modulate supercomplex formation, stability, or function. Several of these accessory proteins have been found to be associated with supercomplexes. For example, PHB1 or PHB2 was present in all of the supercomplex bands analyzed, and deletion of PHB has been shown to be impair mitochondrial supercomplex formation and lead to increased ROS generation [42]. Similarly, STML2—identified in three of the six supercomplex bands—has been shown to be required for mitochondrial supercomplex formation [43]. All supercomplex bands contained MIC60 and SAMM50, which are regulators of mitochondrial function and cristae structure [[44], [45], [46]] and could regulate the assembly of respiratory complexes [47]. Most of the supercomplex bands also comprised proteins in the pyruvate and oxoglutarate dehydrogenase complexes. Thus, it is possible that these complexes may also integrate functionally with mitochondrial supercomplexes. Other interesting enzymes that were found to be part of supercomplexes include SOD2, catalase, and AFG3L2. These may play roles in local ROS detoxification and ATP-dependent proteolysis.

Although these studies further develop our understanding of metabolic factors that influence supercomplexes, there remain several limitations, the first of which involves general understanding of the biological role of supercomplexes. Although pioneering studies suggest that supercomplexes are determinants of mitochondrial electron transfer efficiency [[9], [10], [11]], recent studies refute the idea that mitochondrial supercomplexes enhance enzyme catalysis [[12], [13], [14]]. Of note, our previous studies show that cardiac mitochondria from both Glyco^Lo and Glyco^Hi hearts have lower mitochondrial respiration rates than mitochondria from WT mice. Thus, the heightened levels of supercomplexes in Glyco^Lo mice may not confer a bioenergetic advantage, at least not in this model. Nevertheless, it remains possible that higher levels of supercomplexes improve general respiratory complex stability [48,49] or regulate production of mitochondrial ROS [15,50]. Another major limitation is the relatively observational nature of the study, and it remains unclear how the availability or relative reliance on particular substrates influences supercomplex abundance. Although our studies highlight the phenomenon that substrate provision or genetically modulated changes in glucose-fatty acid cycle influence supercomplex abundance, the underlying mechanism remains unclear. We posit that substrate gradients [28] could play a role; however, it does not appear that substrate abundance acts in isolation to promote supercomplex formation. Indeed, in Glyco^Hi hearts, acylcarnitines accumulate because of the coordinated decrease in fat oxidation and increase in glucose oxidation [29]; yet, despite endogenously high levels of acylcarnitines, mitochondria from Glyco^Hi hearts have lower levels of supercomplexes than Glyco^Lo hearts. Furthermore, it is unclear whether metabolic states primed for high fat oxidation actively promote supercomplex assembly or whether they simply improve supercomplex stability. It also remains unknown how biological sex and other substrates such as glutamine/glutamate, branched chain amino acids, and ketone bodies influence supercomplex formation. Addressing each of these limitations in future studies would improve our understanding of the significance, regulation, and function of mitochondrial supercomplexes.

Author contributions

Y.Z., project planning, execution of experiments, data analysis and interpretation, writing of manuscript; A.A.G., project planning, data analysis and interpretation, writing of manuscript; S.L. and H.X., execution of experiments; B.G.H., project planning, data analysis and interpretation, writing of manuscript, financial support.

Declaration of competing interest

The authors declare no competing interests.

Appendix A. Supplementary data

The following is the Supplementary data to this article.