Differential requirements for mitochondrial electron transport chain components in the adult murine liver

  1. Nicholas P Lesner
  2. Xun Wang
  3. Zhenkang Chen
  4. Anderson Frank
  5. Cameron J Menezes
  6. Sara House
  7. Spencer D Shelton
  8. Andrew Lemoff
  9. David G McFadden
  10. Janaka Wansapura
  11. Ralph J DeBerardinis
  12. Prashant Mishra  Is a corresponding author
  1. Children's Medical Center Research Institute, University of Texas Southwestern Medical Center, United States
  2. Department of Biochemistry, University of Texas Southwestern Medical Center, United States
  3. Harold C. Simmons Comprehensive Cancer Center, University of Texas Southwestern Medical Center, United States
  4. Advanced Imaging Research Center, University of Texas Southwestern Medical Center, United States
  5. Department of Pediatrics, University of Texas Southwestern Medical Center, United States
  6. Howard Hughes Medical Institute, University of Texas Southwestern Medical Center, United States

Abstract

Mitochondrial electron transport chain (ETC) dysfunction due to mutations in the nuclear or mitochondrial genome is a common cause of metabolic disease in humans and displays striking tissue specificity depending on the affected gene. The mechanisms underlying tissue-specific phenotypes are not understood. Complex I (cI) is classically considered the entry point for electrons into the ETC, and in vitro experiments indicate that cI is required for basal respiration and maintenance of the NAD+/NADH ratio, an indicator of cellular redox status. This finding has largely not been tested in vivo. Here, we report that mitochondrial complex I is dispensable for homeostasis of the adult mouse liver; animals with hepatocyte-specific loss of cI function display no overt phenotypes or signs of liver damage, and maintain liver function, redox and oxygen status. Further analysis of cI-deficient livers did not reveal significant proteomic or metabolic changes, indicating little to no compensation is required in the setting of complex I loss. In contrast, complex IV (cIV) dysfunction in adult hepatocytes results in decreased liver function, impaired oxygen handling, steatosis, and liver damage, accompanied by significant metabolomic and proteomic perturbations. Our results support a model whereby complex I loss is tolerated in the mouse liver because hepatocytes use alternative electron donors to fuel the mitochondrial ETC.

Editor's evaluation

The commonly accepted view is that complex I of the mitochondrial respiratory chain is required for oxidative phosphorylation. This is an important paper, which discovers that complex I is not required for maintaining metabolically functional mitochondria in the liver, explaining a lack of liver defects in complex I deficiency patients. The study includes compelling data of proteomics and metabolomic tracer analyses representing a resource for a broad community with interest in biochemistry and metabolism of the cell as well as biomedical research. Based on this data the authors propose an alternative source of electrons for the respiratory chain in the mouse liver – the concept of potentially large impact on our understanding of metabolism.

https://doi.org/10.7554/eLife.80919.sa0

eLife digest

Mitochondria are specialised structures inside cells that help to convert nutrients into energy. They take electrons from nutrients and use them to power biochemical reactions that supply chemical fuel. Previous studies of cells grown in the laboratory have found that electrons enter this process via a large assembly of proteins in mitochondria called complex I.

Understanding the mechanism of energy production is important, as issues with mitochondria can lead to a variety of metabolic diseases. However, it is still unclear how complex I acts in living animals.

Lesner et al. addressed this knowledge gap by genetically removing a key protein from complex I in the liver of mice. Surprisingly, the animals did not develop any detectable symptoms and maintained healthy liver function. Mice did not seem to compensate by making energy in a different way, suggesting that complex I is not normally used by the mouse liver for this process.

This research suggests that biologists should reconsider the mechanism that mitochondria use to power cells in animals. While the role of Complex I in electron transfer is well established in laboratory-grown cells and some organs, like the brain, it cannot be assumed this applies to the whole body. Understanding energy production in specific organs could help researchers to develop nutrient-based therapies for metabolic diseases.

Introduction

Mitochondrial ETC disease in humans exhibits a striking tissue specificity, whereby mutations in distinct ETC components are causative for specific and non-overlapping syndromes (Gorman et al., 2016). As an example, genetic dysfunction of mitochondrial complex I (cI) is associated with neurological disease including Leigh syndrome (characterized by severe loss of motor abilities, necrotizing brain lesions and early death) and Leber Hereditary Optic Neuropathy (characterized by bilateral loss of vision in early adulthood due to degeneration of the optic nerve) (Rodenburg, 2016; Martín et al., 2005; Distelmaier et al., 2009). The mechanisms underlying sensitivity of the nervous system, with sparing of other organ systems, in the setting of complex I loss is not understood. In contrast, mutations in other mitochondrial ETC components have been associated with multi-organ diseases, and the basis for tissue specificity in those diseases is also unclear (Ahmed et al., 2018).

Liver involvement is a common component of mitochondrial diseases, present in approximately 20% of patients (Lee and Sokol, 2007b; Rahman, 2013), as well as being associated with specific syndromes caused by mutations in distinct ETC-related components (Valnot et al., 2000; Casey et al., 2012; Lee and Sokol, 2007a; Morris et al., 1998; Koh et al., 2018). Common findings in affected patients include hepatomegaly, steatosis, cholestasis and additional metabolic abnormalities associated with liver failure (Koh et al., 2018; Molleston et al., 2013; Cloots et al., 2013; Chinnery and DiMauro, 2005). Genetic analysis of patients has revealed mutations in genes which influence multiple components of the ETC (e.g. POLG, LARS, DGUOK) (Lee and Sokol, 2007a). With respect to single-complex diseases, mutations in SCO1, an assembly protein for mitochondrial complex IV (cIV), and mutations in BCS1L, a chaperone protein for mitochondrial complex III, have been associated with liver failure (Valnot et al., 2000; De Meirleir et al., 2003; de Lonlay et al., 2001). Interestingly, single-complex diseases associated with other components of the ETC (complex I and complex II) do not typically present with liver failure, despite their placement directly upstream of cIV.

The ETC functions to transfer electrons (from the oxidation of nutrients) to O2, as a final electron acceptor; this process is coupled to the formation of an electrochemical gradient across the mitochondrial inner membrane and the production of ATP. Mitochondrial complex I is the largest protein complex of the ETC, and is commonly considered the primary entry point for electrons into the ETC: In cultured cells, cI is required for maintenance of cellular redox status (assessed by the NAD+/NADH ratio) via its role in NADH oxidation, and subsequent electron transfer to ubiquinone (CoQ; Sullivan et al., 2015). Complex I is composed of 14 core subunits, plus an additional 31 accessory subunits (Sharma et al., 2009). Published mouse models of complex I deficiency have largely focused on accessory subunits, including deletion of Ndufs4, Ndufs6, and Ndufa5, which result in neurological and cardiac phenotypes associated with partial loss of complex I function (Kruse et al., 2008; Peralta et al., 2014; Ke et al., 2012); no liver phenotypes or metabolic perturbations were reported in these mice (Yang et al., 2020). A recent report conditionally deleted Ndufs3, a core subunit of cI, in skeletal muscle, resulting in progressive myopathy associated with significant reduction of cI activity (Pereira et al., 2020). Thus, to date, mouse models have supported an essential role for cI in vivo in the central nervous system and muscle.

Here, we report that deletion of the core subunit, Ndufa9, results in complete loss of cI function, but is well-tolerated in the adult mouse liver. Further analysis reveals intact liver function and oxygen status in Ndufa9-/- animals, which is not accompanied by significant metabolic alterations or changes in redox status. This is in contrast with hepatocyte-specific deletion of the complex IV assembly factor Cox10, which results in altered redox status and impaired liver function. In complex IV-depleted livers, we observe the accumulation of long chain acyl-carnitine species, without significant alterations in other potential electron donors, suggesting that fatty acids are an electron donor to the hepatic ETC. These results support a model in which cI is not an essential upstream component of the ETC in adult mammalian liver, in contrast to the conventional view of cI function.

Results

Mitochondrial complex I function is dispensable for homeostasis of adult murine hepatocytes

To understand how liver cells respond to loss of complex I, we made use of a conditional allele of the core subunit Ndufa9. Ndufa9 deletion in cultured mouse embryonic fibroblasts results in loss of mitochondrial respiratory function, as well as a significant drop in the NAD+/NADH ratio (Figure 1—figure supplement 1A, B), indicating that Ndufa9 is critical for redox status in these cultured cells. We deleted Ndufa9 in livers of adult mice, making use of adeno-associated virus 8 (AAV8)-mediated Cre recombination driven by the liver specific Serpina7 promoter in Ndufa9flox/flox animals (see Materials and methods). Ndufa9flox/flox adult animals were injected with control AAV8-GFP virus (hereafter ‘Ndufa9f/f’) or AAV8-Cre virus (hereafter ‘Ndufa9-/-’). At 4 weeks post-AAV administration, AAV-Cre injected animals exhibited loss of Ndufa9 mRNA and protein in hepatocytes (Figure 1A and B), as well as loss of complex I activity in mitochondrial lysates (Figure 1C). Thus, our experimental model results in severe inhibition of complex I function.

Figure 1 with 3 supplements see all
Mitochondrial complex I is dispensable in adult murine hepatocytes.

(A) Ndufa9 mRNA levels (relative to β2-microglobulin; normalized) in isolated hepatocytes of the indicated genotype, assessed by qRT-PCR. n=3 animals per group. The same color scheme is used throughout this figure. (B) Ndufa9 protein levels in isolated hepatocytes of the indicated genotype, assessed by Western blot. β2-microglobulin levels are shown as a loading control. MW markers are indicated in kDa. (C) Complex I activity measurements from livers of the indicated genotype. n=3 animals per group. (D) Representative images of gross liver anatomy in animals of the indicated genotype. Scale bar: 0.5 cm. (E) Representative histology (H&E, Oil Red O, PAS staining) images of liver cross sections from animals of the indicated genotype. Scale bar: 50 μm. (F) Liver weight (normalized to body weight) from mice of the indicated genotype. n=4–5 animals per group. (G) Fasting plasma glucose levels in mice of the indicated genotype. n=12–14 animals per group. (H) Circulating plasma markers (total protein, albumin, and total bilirubin) of liver function in mice of the indicated genotype. n=3–5 animals per group. (I) Plasma markers (ALT and AST) of liver damage in mice of the indicated genotype. n=4–5 mice per group. All data were collected at 4 weeks post AAV-administration. Statistical significance was assessed using two-tailed t-test (A,C,G,H,I) or Mann-Whitney (F,H) tests with adjustments for multiple comparisons. All data represent mean +/-standard deviation from biological replicates. Full gel images are provided in Figure 1—source data 1. Numerical data for individual panels are provided in Supplementary file 3.

Figure 1—source data 1

Full gel images for Figure 1.

Full gel images and original image files for western blots in Figure 1.

https://cdn.elifesciences.org/articles/80919/elife-80919-fig1-data1-v2.zip

At 4 weeks post-AAV administration, Ndufa9-/- livers displayed no differences in gross physiology (Figure 1D) or liver/body weight ratio (Figure 1F). Histological analysis of liver sections revealed no significant pathology (H&E), including no evidence of steatosis (Oil Red O staining) or glycogen loss (PAS staining) (Figure 1E). Liver function was maintained in Ndufa9-/- livers as indicated by unaltered levels of fasting blood glucose (Figure 1G) and circulating indicators of liver function and damage (Figure 1H, I). Similar results were observed in mice at 2 weeks post-AAV injection (Figure 1—figure supplement 2) and 8 weeks post-AAV injection (Figure 1—figure supplement 3). Thus, complex I loss is well-tolerated in adult mouse hepatocytes under homeostatic conditions, with no discernable effects on tissue function, or evidence of tissue damage.

Measurement of steady-state ATP levels did not reveal any changes in complex-I-deficient livers (Figure 2A). Complex I inhibition is often associated with elevated reactive oxygen species (Li et al., 2003; Fato et al., 2009). In Ndufa9-/- livers, we observed a small, but not statistically significant, decrease in the reduced to oxidized glutathione (GSH / GSSG) ratio, indicating that oxidative stress is not significantly induced in the setting of cI loss (Figure 2B). Surprisingly, we did not observe altered NAD+/NADH or NADPH/NADP+ ratios in Ndufa9-/- liver (Figure 2C). We additionally examined compartment-specific redox status by calculating pyruvate / lactate and acetoacetate / β-hydroxybutyrate ratios, which are indicative of cytosolic and mitochondrial redox status respectively, neither compartment display evidence of altered redox status (Figure 2D). A recent report suggests that circulating α-hydroxybutyrate levels serves as a marker of hepatic redox stress (Goodman et al., 2020). Ndufa9-/- animals did not display alterations in tissue or plasma α-hydroxybutyrate levels (Figure 2E). Thus, complex I function is not required for maintenance of hepatic redox status, in contrast to its well-established role in cultured cells (Figure 1—figure supplement 1A, B; Sullivan et al., 2015).

Figure 2 with 2 supplements see all
Complex I deficiency in adult liver does not impact redox status.

(A) ATP levels from livers of the indicated genotype. n=3 mice per group. The same color scheme is used throughout this figure. (B) Reduced glutathione to oxidized glutathione ratio (GSH/GSSG) in livers of the indicated genotype, as a proxy of oxidative stress. n=8–9 mice per group. (C) NAD+/NADH and NADPH/NADP+ ratios in livers of the indicated genotype. n=6 mice per group. (D) Normalized cytosolic (pyruvate / lactate) and mitochondrial (acetoacetate / β−hydroxybutyrate) redox ratios in livers of the indicated genotype. n=6 mice per group. (E) Tissue and plasma α−hydroxybutyrate levels in animals of the indicated genotype. n=4–6 mice per group. (F) Unsupervised hierarchical clustering of protein abundances in Ndufa9f/f and Ndufa9-/- livers, depicted as a heatmap of z-scores. (G) Volcano plot of protein abundances changes in Ndufa9-/- vs. Ndufa9f/f livers, based on proteomics analysis. Log2(Fold change) is plotted against the –log10(adjusted p-value) for each protein. Significantly changing proteins (log2(Fold change)>1 or < –1; Adjusted p-val <0.05) are colored black, and mitochondrial proteins are colored red. n=6 mice per group. (H) Volcano plot of abundances for mitochondrial ETC proteins in Ndufa9-/- vs. Ndufa9f/f livers, based on proteomics analysis. (I) Levels of mitochondrial proteins (PC, pyruvate carboxylase, and Tomm20) in livers of the indicated genotype, as assessed by western blot. Actin levels are shown as a loading control. MW markers are indicated in kDa. (J) Relative amounts of Tomm20 and PC in livers of the indicated genotype, relative to actin levels. n=4 animals per group. (K) Mitochondrial genome (mtDNA) to nuclear genome (nDNA) ratios in livers of the indicated genotype. n=8–9 mice per group. (L) Unsupervised hierarchical clustering of liver metabolite abundances in Ndufa9f/f and Ndufa9-/- livers, depicted as a heatmap of z-scores. (M) Volcano plot of metabolite abundance changes in Ndufa9-/- vs. Ndufa9f/f livers, based on metabolomics analysis. Log2(Fold change) is potted against the –log10(adjusted p-value) for each metabolite. Significantly changing metabolites (log2(Fold change)>1 or < –1; Adjusted p-val <0.05) are colored black. TMP, Thiamine monophosphate. n=6 mice per group. (N) Relative abundances of mitochondrial TCA cycle and related metabolites from livers of the indicated genotype. n=6 mice per group. All data were collected at 4 weeks post AAV-administration. Statistical significance was assessed using two-tailed t-test (A,C,D,E,J,K,N) or Mann-Whitney (B) with adjustments for multiple comparisons. All data represent mean +/-standard deviation from biological replicates. Full gel images are provided in Figure 2—source data 1. Numerical data for individual panels are provided in Supplementary file 1 and Supplementary file 3.

Figure 2—source data 1

Full gel images for Figure 2.

Full gel images and original image files for western blots in Figure 2.

https://cdn.elifesciences.org/articles/80919/elife-80919-fig2-data1-v2.zip

To examine potential compensatory mechanisms that may be enabled in the setting of complex I inhibition, we assessed changes in the proteome and metabolome, comparing Ndufa9f/f and Ndufa9-/- whole livers. At 4 weeks post-AAV, our proteomics analysis detected 3001 proteins with high confidence; however, only 11 proteins were significantly upregulated (log2(fold change)>1, p-val <0.05), and 23 proteins were significantly downregulated (log2(fold change) < –1; p-val <0.05) (Supplementary file 1). Unsupervised hierarchical clustering was unable to distinguish proteomes from Ndufa9f/f and Ndufa9-/- livers (Figure 2F). Of the downregulated proteins, the majority were components of complex I (Figure 2G and H; Supplementary file 1), consistent with destabilization of cI in the setting of Ndufa9 subunit loss. We did not observe loss of components from other ETC complexes (Figure 2H). Among upregulated proteins, there were a small number of mitochondrial proteins impacting translation (Mrpl52, Mrpl42), electron transport (Cox6a1), and metabolism (Shmt2, Ldhd, Dlat) (Figure 2G; Supplementary file 1). The small number of abundance changes (outside of complex I components) suggests that the proteome is not drastically changing in response to complex I loss; however, our analysis is only sampling a portion of the total cellular proteome, and it is possible that we are missing some key regulatory proteins in reaction to complex I loss. To further profile the response in Ndufa9-/- livers, we enriched liver lysates for mitochondrial fractions followed by proteomics (Figure 2—figure supplement 1). With this method, we were able to detect 738 mitochondrial proteins, which accounts for ~64% of the total mitochondrial proteome (Supplementary file 1). We observed significant changes (|log2(fold change)|>1, Adjusted p-val <0.05) in only a small fraction of the mitochondrial proteome (46 proteins;~6% of detected mitochondrial proteins) (Figure 2—figure supplement 1A). Complex I subunits tended to be decreased in abundance, and we did not observe significant changes in most subunits of the other complexes (Figure 2—figure supplement 1B). We observed statistically significant upregulation of some select components of mitochondrial protein import (TOMM5, TIMM9, TIMM10B) and transcription/translation (MRPS33, MTRES1; Figure 2—figure supplement 1A). In addition, we did not detect significant changes in total mitochondrial content in Ndufa9-/- livers, as indicated by measurement of mitochondrial genome and protein content or immunofluorescence (Figure 2I–K, Figure 2—figure supplement 2).

To assess whether complex I – deficient livers maintained functionality via metabolic compensatory mechanisms, we quantitated the liver metabolome of fasted mice, using LC/Q-TOF mass spectrometry-based metabolomics to measure steady-state levels of >150 tissue (liver) metabolites at 4 weeks post-AAV (Supplementary file 1). Unsupervised hierarchical clustering was unable to distinguish metabolomes between Ndufa9f/f and Ndufa9-/- livers (Figure 2L), and statistical analysis revealed only a single significant metabolite difference (TMP, thiamine monophosphate; Figure 2M, Supplementary file 1), and no significant changes in mitochondrial TCA cycle metabolites (Figure 2N). The rationale for the observed increase in TMP is currently unclear, but may be related to a reduced need for thiamine pyrophosphate which is a cofactor for NAD(H)-linked dehydrogenases.

In culture, complex I is required for recycling of NADH to NAD+ in order to maintain glycolytic and TCA cycle activity (Lesner et al., 2020). We therefore assessed glucose contribution to glycolytic and TCA cycle metabolites via steady-state euglycemic infusions of [U-13C]glucose (Figure 3A and B). In these experiments, glucose can contribute to TCA cycle metabolites via glycolysis, PDH (pyruvate dehydrogenase), and PC (pyruvate carboxylase) activity; in addition, extrahepatic production of lactate can be metabolized by the liver and contribute to both TCA cycle metabolites and glycolytic intermediates via gluconeogenic pathways (Figure 3A). Infusion of 13C uniformly labeled glucose allows us to assess the contribution of glucose-derived carbons to liver metabolites by measuring the fraction of unlabeled (m+0) metabolite, as well as the fraction of each possible labeled species (m+1, m+2, m+3, …) in which one or more carbons were derived from the labeled glucose molecule. We observed no significant differences in the hepatic enrichment of infused labeled glucose between wild-type and Ndufa9-/- livers (Figure 3C). In addition, contribution of glucose-derived carbons to glycolytic and TCA cycle intermediates were unaltered (Figure 3—figure supplement 1A), suggesting no substantial changes in glucose handling. As labeling from glucose can be significantly scrambled in vivo and livers exhibit low PDH activity (Cappel et al., 2019; Merritt et al., 2011), we also assessed total enrichment (e.g. 1-(m+0)), which revealed no significant differences (Figure 3D).

Figure 3 with 1 supplement see all
Complex I deficiency in adult liver does not impact glucose utilization or oxygen status.

(A) Schematic of potential fates of glucose and lactate within central carbon metabolism. Carbon atoms are indicated with circles for each compound. G1P, glucose-1-phosphate; 3 PG, 3-phosphoglycerate; 2 PG, 2-phosphoglycerate; PEP, phosphoenolpyruvate; pyr, pyruvate; lac, lactate; acCoA, acetyl-CoA; OAA, oxaloacetate; glu, glutamate; αkg, αketoglutarate. (B) Plasma enrichment of m+6 glucose during steady-state infusions of [U-13C]glucose. n=3 animals per group. The same color scheme is used throughout this figure. (C) Steady-state liver enrichment of m+6 glucose in animals of the indicated genotype. n=3 animals per group. (D) Labeled fractions (1-(m+0)) of the indicated metabolites, relative to liver glucose m+6 enrichment. n=3 animals per group. (E) Schematic of MRI experiments. Anesthetized animals were challenged to breath room air, followed by 100% O2, followed by MR imaging to measure perfusion (in room air), and blood and tissue oxygenation in each condition. (F) Perfusion values (reported as institutional units (I.U.)) based on arterial spin labeling in animals of the indicated genotype. n=3–4 animals per group. (G) Relative (%) changes in hepatic T2* following exposure to 100% O2 in animals of the indicated genotype. n=3–4 animals per group. (H) Representative heatmaps of T1 apparent values (T1app) in livers, overlaid on T1-weighted MR images. (I) Representative distribution of T1app values in an animal of each genotype breathing room air, followed by 100% O2. (J) Average hepatic T1app values in animals of the indicated genotype, breathing room air followed by 100% O2. n=3–4 animals per group. (K) Average regional changes in T1 (ΔT1) in animals of the indicated genotype. n=3–4 animals per group. All data were collected at 4 weeks post AAV-administration. Statistical significance was assessed using two-tailed t-test (C,D,F,G,K) or Mann-Whitney (I) with adjustments for multiple comparisons. All data represent mean +/-standard deviation from biological replicates. Numerical data for individual panels are provided in Supplementary file 1 and Supplementary file 3.

In the conventional view of the mitochondrial ETC, electrons from NADH are transferred via complex I to drive coenzyme Q reduction and O2 consumption via complexes III and IV (Figure 3A). In cultured cells, complex I is the major route of electron entry into the ETC, and inhibition of complex I dramatically lowers cellular oxygen consumption (Figure 1—figure supplement 1A). Our above results suggest this scenario may not be true during hepatic homeostasis. We therefore made use of magnetic resonance imaging (MRI) to assess blood and tissue oxygenation in vivo (Figure 3E–K). Oxygen-sensitive MRI measurements provide a non-invasive method to interrogate changes in oxygen levels in vivo in the absence of a reporter agent. Specifically, the paramagnetic properties of dissolved oxygen within tissue accelerate the decay of longitudinal magnetization of water protons, termed the TOLD (Tissue-Oxygenation-Level-Dependent) effect (Hallac et al., 2014). The decay of longitudinal magnetization is characterized by the decay constant (or longitudinal relaxation time) T1. Thus, increased tissue oxygen levels result in decreased T1 values. Similarly, the unpaired electrons of iron in deoxyhemoglobin allow it to act as a strong paramagnetic agent which accelerates the decay of transverse magnetization in the blood (characterized by the transverse relaxation time T2*), termed the BOLD (Blood-Oxygen-Level-Dependent) effect (Wengler et al., 2019). In this manner, relative increases in T2* values are indicative of decreased deoxyhemoglobin concentration (i.e., an increased blood oxygenation state). We reasoned that by following hepatic T1 levels in response to increased blood oxygen delivery, we could assess the mitochondria’s ability to process this excess oxygen. In these experiments, we controlled for differences in tissue perfusion by measuring tissue perfusion using an arterial spin labeling (ASL) method (Kim and Tsekos, 1997). Here, the magnetization of water protons flowing through the slice of interest is labeled by a radio frequency pulse in such a way that they can be distinguished from that of the stationary tissue. The difference in magnetization between the stationary and flowing water protons is related to blood perfusion through a simple two compartment model of tissue, by which the perfusion is quantified (see Materials and methods).

Ndufa9f/f and Ndufa9-/- littermates were challenged with room air, followed by 100% oxygen, and T1 and T2* relaxation parameters were sequentially assessed (Figure 3E). Based on arterial spin labeling (Kim and Tsekos, 1997), we did not observe changes in liver perfusion between the two genotypes (Figure 3F). BOLD (Blood-oxygen-level dependent) measurements (T2*) are dependent on the concentration of deoxyhemoglobin (Wengler et al., 2019; Barash et al., 2007), and we observed significant increases in T2* in both Ndufa9f/f and Ndufa9-/- animals, indicating increased blood oxygen in the setting of 100% O2 (Figure 3G). TOLD (Tissue-oxygen-level dependent) measurements (based on T1 relaxation) are dependent on the dissolved oxygen concentration; with increased tissue oxygen content, T1 parameters are expected to decrease (Hallac et al., 2014). In wild-type livers, we observed no significant decreases in the average hepatic T1 in response to 100% O2 (Figure 3H, I, J), consistent with previous measurements in human livers (Tadamura et al., 1997). This suggests that when challenged with increased blood O2 delivery, wild-type mouse liver is able to compensate and maintain steady state O2 levels, potentially through increased tissue oxygen consumption. Similar to Ndufa9f/f livers, Ndufa9-/- livers also did not exhibit significant decreases in the average T1 parameter, indicating that tissue oxygen levels are maintained in the absence of complex I function (Figure 3H–J). We also assessed T1 changes locally by aligning images (ΔT1); as above, we did not observe significant decreases in the average ΔT1 in either Ndufa9f/f or Ndufa9-/- livers (Figure 3K). Thus, based on tissue oxygen levels in response to a hyperoxic challenge, Ndufa9-/- livers do not have a measurable change in oxygen consumption as compared with wild-type livers. Altogether, the above results indicate that complex I is largely dispensable for oxygen status and homeostatic function in the adult mouse liver, without signs of significant metabolic or proteomic compensation.

Mitochondrial complex IV is required in adult hepatocytes

To further explore a role for the mitochondrial ETC in hepatic function, we examined the requirement for complex IV, making use of a Cox10flox allele (Diaz et al., 2005). Cox10 is an assembly factor of complex IV, and Cox10 deletion in skeletal muscle is associated with diminished cIV activity accompanied by myopathy (Diaz et al., 2005). A previous report utilizing this allele in combination with the liver-specific Albumin-Cre transgenic reporter revealed partially penetrant phenotypes dependent on the transgene dosage (Diaz et al., 2008), suggesting that the mitochondrial ETC is required for liver development. Thus, we made use of AAV-delivery of Cre recombinase to Cox10f/f animals to investigate its role in the adult liver. Similar to above, we injected Cox10f/f animals with AAV-GFP or AAV-Cre and characterized liver physiology at 4 and 8 weeks post-injection. At 4 weeks post-injection, AAV-Cre injected animals (hereafter, ‘Cox10-/-’) demonstrated significant reductions in Cox10 mRNA and protein (Figure 4A and B) and complex IV activity (Figure 4C). By 8 weeks post-AAV administration, Cox10-/- livers were grossly enlarged and pale in color, as compared with wild-type livers (Figure 4D and F). Histological analysis of liver sections revealed signs of lipid accumulation indicating hepatic steatosis (Oil Red O staining), as well as decreased glycogen accumulation (PAS staining) (Figure 4E). Fasting glucose levels were decreased in animals with Cox10-/- livers (Figure 4G). To directly probe gluconeogenesis, fasted mice were challenged with a [U-13C]-lactate/pyruvate tolerance test; animals with Cox10-/- livers were unresponsive as compared with wild-type animals (Figure 4H, I, Figure 4—figure supplement 1A). Additional plasma indicators of liver function were diminished in Cox10-/- animals, including decreased plasma total protein and increased plasma bilirubin (Figure 4J). We also observed significant elevations in ALT and AST, indicators of liver damage (Figure 4K). Thus, in contrast to loss of complex I, the loss of complex IV in adult hepatocytes is associated with hepatic steatosis, diminished liver function and evidence of liver damage.

Figure 4 with 1 supplement see all
Complex IV deficiency in adult hepatocytes induces liver dysfunction.

(A) Cox10 mRNA levels (relative to β2-microglobulin; normalized) in livers of the indicated genotype, assessed by qRT-PCR at 4 weeks post-AAV administration. n=4–5 animals per group. The same color scheme is used throughout this figure. (B) Cox10 protein levels in livers of the indicated genotype, assessed by Western blot at 4 weeks post AAV-administration. β2-microglobulin levels are shown as a loading control. MW markers are indicated in kDa. (C) Complex IV activity (pmol O2 consumed / min) in isolated mitochondria supplemented with ascorbate/TMPD substrates from livers of the indicated genotype, assessed at 4 weeks post AAV-administration. n=5–6 animals per group. (D) Representative images of gross liver anatomy in animals of the indicated genotype at 8 weeks post AAV-administration. Scale bar: 0.5 cm. (E) Representative histology (H&E, Oil Red O, PAS staining) images of liver cross sections from animals of the indicated genotype at 8 weeks post-AAV administration. Scale bar: 50 μm. (F) Liver weight (normalized to body weight) from mice of the indicated genotype at 8 weeks post-AAV administration. n=7 animals per group. (G) Fasting plasma glucose levels in mice of the indicated genotype at 8 weeks post-AAV administration. n=6–7 animals per group. (H) Relative changes in plasma glucose following a lactate/pyruvate tolerance test in mice of the indicated genotype at 8 weeks post-AAV administration. n=4–5 mice per group. (I) Fractional gluconeogenesis (plasma) in animals of the indicated genotype, measured via pyruvate/lactate tolerance test at 8 weeks post-AAV administration. n=4–5 animals per group. (J) Circulating plasma markers (total protein, albumin, and total bilirubin) of liver function in mice of the indicated genotype at 8 weeks post-AAV administration. n=5 animals per group. (K) Plasma markers (ALT and AST) of liver damage in mice of the indicated genotype at 8 weeks post-AAV administration. n=5 mice per group. Statistical significance was assessed using two-way ANOVA (H), t-test (C,F,G,I,J,K) or Mann-Whitney (A,K) tests with adjustments for multiple comparisons. All data represent mean +/-standard deviation from biological replicates. Full gel images are provided in Figure 4—source data 1. Numerical data for individual panels are provided in Supplementary file 3.

Figure 4—source data 1

Full gel images for Figure 4.

Full gel images and original image files for western blots in Figure 4.

https://cdn.elifesciences.org/articles/80919/elife-80919-fig4-data1-v2.zip

Despite the fatty livers and diminished function, Cox10-/- animals exhibited no survival deficits even 8 weeks after AAV administration, and we did not detect changes in steady state ATP levels in liver tissue (Figure 5A). Cox10-/- livers exhibited a small, but not statistically significant, trend towards increased oxidative stress (lowering of the GSH / GSSG ratio) (Figure 5B). In contrast to cI-deficient livers, cIV deficiency significantly impacted the redox status of the cell, as measured by the NAD+ / NADH ratio (Figure 5C). We observed altered ratios of both pyruvate / lactate and acetoacetate / β-hydroxybutyrate, indicating impaired redox status in both the cytosol and mitochondrial compartments (Figure 5D). Interestingly, tissue α-hydroxybutyrate levels were decreased in Cox10-/- livers, and plasma α-hydroxybutyrate levels were not significantly affected (Figure 5E). Thus, impairment of mitochondrial complex IV has both biochemical and functional consequences in the adult liver and is required for maintenance of cellular redox status.

Figure 5 with 8 supplements see all
Complex IV deficiency alters metabolism and cellular redox status in adult liver.

(A) ATP levels in livers of the indicated genotype. n=3 animals per group. The same color scheme is used throughout this figure. (B) Reduced glutathione to oxidized glutathione ratio (GSH:GSSG) in livers of the indicated genotype, as a proxy of oxidative stress. n=5–6 mice per group. (C) NAD+/NADH and NADPH/NADP+ ratios in livers of the indicated genotype. n=5–6 mice per group. (D) Normalized cytosolic (pyruvate / lactate) and mitochondrial (acetoacetate / β−hydroxybutyrate) redox ratios in livers of the indicated genotype. n=5–6 mice per group. (E) Normalized tissue and plasma α−hydroxybutyrate levels in animals of the indicated genotype. n=4–5 mice per group. (F) Unsupervised hierarchical clustering of protein abundances in Cox10f/f and Cox10-/- livers, depicted as a heatmap of z-scores. (G) Volcano plot of protein abundances changes in Cox10-/- vs. Cox10f/f livers, based on proteomics analysis. Log2(Fold change) is plotted against the –log10(adjusted p-value) for each protein. Significantly changing proteins (log2(Fold change)>1 or < –1; Adjusted p-val <0.05) are colored dark gray; of these, mitochondria-localized proteins are colored red. n=5–6 mice per group. (H) Volcano plot of abundances for mitochondrial ETC proteins in Cox10-/- vs. Cox10f/f livers, based on proteomics analysis. (I) Levels of mitochondrial proteins (PC, pyruvate carboxylase, and Tomm20) in livers of the indicated genotype, as assessed by western blot. Actin levels are shown as a loading control. MW markers are indicated in kDa. (J) Relative amounts of Tomm20 and PC in livers of the indicated genotype, relative to actin levels. n=4 animals per group. (K) Mitochondrial genome (mtDNA) to nuclear genome (nDNA) ratios in livers of the indicated genotype. n=9 mice per group. (L) Unsupervised hierarchical clustering of metabolite abundances in Cox10f/f and Cox10-/- livers, depicted as a heatmap of z-scores. (M) Volcano plot of metabolite abundance changes in Cox10-/- vs. Cox10f/f livers, based on metabolomics analysis. Log2(Fold change) is potted against the –log10(adjusted p-value) for each metabolite. Significantly changing metabolites (log2(Fold change)>1 or < –1; Adjusted p-val <0.05) are colored black. n=5–6 mice per group. (N) Relative abundances of mitochondrial TCA cycle and related metabolites from livers of the indicated genotype. n=5–6 mice per group. All data in this figure were collected at 8 weeks post AAV-administration. Statistical significance was assessed using t-test (A,B,C,D,E,J,K,N), or Mann-Whitney (C,E) tests with adjustments for multiple comparisons. All data represent mean +/-standard deviation from biological replicates. Full gel images are provided in Figure 5—source data 1. Numerical data for individual panels are provided in Supplementary file 2 and Supplementary file 3.

Figure 5—source data 1

Full gel images for Figure 5.

Full gel images and original image files for western blots in Figure 5.

https://cdn.elifesciences.org/articles/80919/elife-80919-fig5-data1-v2.zip

To probe the effects of complex IV loss in mammalian liver, we assessed proteome changes by label-free proteomics (Figure 5F, Supplementary file 2). Unsupervised hierarchical clustering was able to separate proteomes between Cox10f/f and Cox10-/- livers (Figure 5F). A large number of proteins were significantly up-regulated (log2(Fold Change)>1; p-value <0.05), and this group was enriched (209 out of 319) for proteins that localize to mitochondria (Figure 5G (red dots), Supplementary file 2). Gene ontology analysis of all upregulated proteins revealed enrichment of a large number of mitochondrial pathways in all three categories (Biological Processes (BP), Cellular Components (CC), and Molecular Functions (MF)) (Supplementary file 2, Figure 5—figure supplement 1A (red arrows)). Consistent with this, there were significant increases in total mitochondrial content in Cox10-/- livers, as measured by both mitochondrial genome and protein content (Figure 5I–K; Figure 5—figure supplement 8). Gene ontology analysis of mitochondrial proteins indicated enrichment of several gene groups, including electron transport, TCA cycle, translation, and ATP synthesis (Supplementary file 2, Figure 5—figure supplement 1B). A separate analysis focused on non-mitochondrial proteins that were upregulated did not reveal many pathways that were significantly enriched (FDR <0.05), with the single exception of the cornified envelope cellular component pathway, which included various genes implicated in cell damage and death processes [e.g. S100A9 (Ghavami et al., 2010; Ghavami et al., 2008), LGALS3 (Mazurek et al., 2012), NCL (Kobayashi et al., 2012; Goldstein et al., 2013; Yang et al., 2002), HSPB1 (Singh et al., 2017)] (Supplementary file 2, Figure 5—figure supplement 1C), and may potentially be related to the hepatic damage observed in this animal model (Figure 4K).

Only a small number of mitochondrial proteins were present among the downregulated proteins (16 of 214), among which we identified a number of complex IV components (Figure 5G and H), consistent with destabilization of intact complex IV due to loss of an assembly subunit. We did not observe significant decreases in components from other ETC complexes, suggesting an isolated effect of Cox10 loss on complex IV (Figure 5H). Consistent with this, gene ontology analysis of all downregulated proteins identified complex IV as the only Cellular Components pathway enriched among the downregulated genes (Supplementary file 2, Figure 5—figure supplement 2A). Several biological processes related to metabolism were enriched among the downregulated genes which included lipid metabolic and biosynthetic pathways (Supplementary file 2, Figure 5—figure supplement 2A (blue arrows)). Analyses focusing specifically on mitochondrial and non-mitochondrial downregulated genes revealed similar findings (i.e. downregulation of mitochondrial complex IV components, as well as metabolic processes) (Supplementary file 2, Figure 5—figure supplement 2B,C). Thus, proteomic changes in response to hepatic complex IV deficiency largely center around increased mitochondrial constituents, downregulation of complex IV, as well as decreased enzyme abundance for a number of metabolic pathways.

To further characterize the compensatory response in the absence of complex IV, we performed label-free proteomics on enriched mitochondrial fractions from Cox10f/f and Cox10-/- livers, which was able to identify 721 mitochondrial proteins (Supplementary file 2). We observed significant changes (|log2(fold change)|>1, Adjusted p-val <0.05) in 99 proteins (~14% of detected mitochondrial proteins) (Figure 5—figure supplement 3A, Supplementary file 2). Complex IV subunits tended to be decreased in abundance, and we did not observe significant changes in most subunits of the other complexes although they tended to be upregulated (Figure 5—figure supplement 3B). We observed statistically significant downregulation of some select components of fatty acid metabolism (FABP1, ACAD11) as well as upregulation of some components regulating ubiquinol homeostasis (DHODH, UQCC1, UQCC2) (Figure 5—figure supplement 3A).

Steady-state metabolomic analysis of hepatic tissue revealed a number of metabolites altered in response to complex IV loss, and unsupervised hierarchical clustering was sufficient to separate Cox10f/f and Cox10-/- livers (Figure 5L and M). Mitochondrial TCA cycle metabolites were largely maintained, although we observed a significant increase in 2-hydroxyglutarate levels, which has been previously observed in the setting of mitochondrial dysfunction (Figure 5N; Hunt et al., 2019). We observed significant upregulation in a number of cholate species, consistent with cholestasis and hepatomegaly in these animals (Figure 5M, Figure 5—figure supplement 4A). Upregulated metabolites were also highly enriched for long-chain acylcarnitine fatty acid species (Figure 5M). We hypothesize that the accumulation of acylcarnitine species is related to a fatty oxidation defect secondary to loss of mitochondrial complex IV; however, this finding may also be related to the proteomic changes in Cox10-/- livers discussed above. We therefore assessed proteomic changes in annotated GO pathways related to fatty acid metabolism (Figure 5—figure supplement 5). A number of enzymes related to fatty acid synthesis exhibited statistically significant down regulation in Cox10-/- livers, including FASN, ACLY, and ACSL1, key fatty acid synthesis enzymes (Figure 5—figure supplement 5A,C). However, we did not observe overall enrichment or depletion of the fatty acid synthesis pathway by gene set enrichment analysis (GSEA) (Figure 5—figure supplement 5A), and we did not detect significant changes in ACACA and ACACB, which catalyze rate-limiting steps for fatty acid synthesis. Conversely, genes involved in fatty acid oxidation tended to be upregulated (Figure 5—figure supplement 5B,C), although the overall enrichment of this pathway did not reach statistical significance, and we did not observe significant changes in the rate-limiting enzyme CPT1A. As a whole, the trends towards downregulation of fatty acid synthesis enzymes and upregulation of fatty acid oxidation enzymes are not consistent with being causal for the large accumulation of acyl-carnitine species, and instead suggest that the proteome responds in a compensatory manner to limit the accumulation of acyl-carnitine species.

We similarly examined proteomic changes in genes involved in bile acid metabolism (Figure 5—figure supplement 6), and their relation to the large increases in levels of cholate species (Figure 5M, Figure 5—figure supplement 4A). Abundance of these family members tended to be decreased in Cox10-/- livers, although the overall pathway depletion did not reach statistical significance (Figure 5—figure supplement 6A). Unfortunately, our proteomic analysis did not detect most bile acid metabolic enzymes, including the rate-limiting enzyme CYP7A1 (Figure 5—figure supplement 6B).

Downregulated metabolites were enriched for several electron carriers, including pyruvate, ascorbate and tetrahydrobiopterin (Figure 5M, Figure 5—figure supplement 4B). Pyruvate can exchange electrons with NAD(H), while ascorbate and THBP are potent antioxidants which can exchange electrons with metal ions and hydroxyradical species (Traber and Stevens, 2011; Kojima et al., 1995). Thus, inhibition of the mitochondrial ETC via Cox10 removal may influence redox balance in multiple species. To assess related proteomic changes, we investigated alterations in genes of the oxidoreductase family, a large family related to transfer of electrons between many potential substrates (GO: 0016491; 787 members). Overall, oxidoreductase enzymes trended to be upregulated in Cox10-/- livers, largely driven by mitochondrial-localized proteins, and we did identify a number of enzymes related to pyruvate, ascorbate and THBP metabolism as significantly altered in response to hepatic complex IV loss (Figure 5—figure supplement 7A). The overall pathway enrichment did not reach statistical significance. Focusing on each electron carrier in isolation did reveal some interesting trends. Pyruvate has a number of potential metabolic fates, and we did observe upregulation of components of the PDH complex (PDHA1, PDHB, PDHX, DLAT, DLD) which could potentially deplete pyruvate via conversion into acetyl-CoA (Figure 5—figure supplement 7B). Similarly, we observed significant upregulation of BCKDHA and BCKDHB, which potentially serve to deplete the electron carrier α-hydroxybutyrate; in addition, the rate-limiting enzyme of this transsulfuration pathway (CBS) was significantly downregulated (Figure 5—figure supplement 7C). The ketone metabolites acetoacetate and β-hydroxybutyrate were significantly depleted in Cox10-/- livers; however, we did not observe significant changes in the ketone body-related enzymes detected in our dataset, including the rate-limiting synthesis enzymes HMGCS1 and HMGCS2 (Figure 5—figure supplement 7D). With respect to ascorbate metabolism, the majority of related enzymes were not detected in our experiment, including the rate-limiting synthesis enzyme GULO (Figure 5—figure supplement 7E). Lastly, we detected a significant decrease in the BH4 (tetrahydrobiopterin) recycling enzyme QDPR, consistent with the downregulation of BH4 levels. The rate-limiting synthesis enzyme, GCH1, was not detected. To summarize, the precise reasons for changes in the levels of these electron carriers is not clear at this stage; however, it is possible that proteome changes in response to complex IV loss may be contributing to these alterations.

Complex IV inhibition blocks electron entry from the electron transfer flavoprotein (ETF) complex

In cultured cells, Complex IV is the major source of oxygen consumption. To examine its role in vivo, we performed magnetic resonance imaging in Cox10f/f and Cox10-/- animals as described above (Figure 3E). Based on arterial spin labeling measurements, we observed no alterations in perfusion of Cox10-/- livers (Figure 6A). In addition, BOLD (T2*) measurements indicated significant increases in blood oxygen content in both genotypes in response to 100% O2, based on increases in the T2* relaxation time (Figure 6B). Similar to above results, Cox10f/f (wild-type) livers did not display increases in tissue oxygen levels, based on TOLD imaging and average T1 parameters (Figure 6C–E). In contrast, Cox10-/- livers displayed significant decreases in average T1 relaxation times in response to a 100% O2 challenge, indicating increased dissolved oxygen levels in affected liver tissue (Figure 6C–E). A regional assessment of changes in T1 (ΔT1) also indicated decreased T1 parameters in Cox10-/-, but not Cox10f/f livers (Figure 6F). Thus, Cox10-/- livers are deficient in maintaining oxygen levels in response to increased blood O2 delivery, consistent with decreased oxygen consumption in the setting of cIV deficiency.

Figure 6 with 1 supplement see all
Complex IV deficiency impairs electron transport chain function and oxygen status in the adult liver.

(A) Perfusion values (reported as institutional units (I.U.)) based on arterial spin labeling in animals of the indicated genotype at 4 weeks post-AAV administration. n=3 animals per group. The same color scheme is used throughout this figure. (B) Relative (%) changes in hepatic T2* following exposure to 100% O2 in animals of the indicated genotype at 4 weeks post-AAV administration. n=3 animals per group. (C) Representative heatmaps of T1 apparent values (T1app) in livers, overlaid on T1-weighted MR images. (D) Representative distribution of T1app values in an animal of each genotype breathing room air, followed by 100% O2. (E) Average hepatic T1app values in animals of the indicated genotype, breathing room air followed by 100% O2. n=3 animals per group; 4 weeks post-AAV administration. (F) Average regional changes in T1 (ΔT1) in animals of the indicated genotype. n=3 animals per group; 4 weeks post-AAV administration. (G) Plasma enrichment of m+6 glucose during steady-state infusions of [U-13C]glucose. n=3 animals per group; 4 weeks post-AAV administration. (H) Steady-state liver enrichment of m+6 glucose in animals of the indicated genotype. n=3 animals per group; 4 weeks post-AAV administration. (I) Labeled fractions (1-(m+0)) of the indicated metabolites, relative to liver glucose m+6 enrichment. n=3 animals per group; 4 weeks post-AAV administration. (J) Schematic of potential electron sources for mitochondrial ETC. CoQ, ubiquinone; GPD2, mitochondrial glycerol-3-phosphate dehydrogenase; DHODH, dihydroorotate dehydrogenase. ETF, electron transfer flavoprotein complex. DHAP, dihydroxyacetone phosphate. (K) Relative abundance of electron donors in livers of the indicated genotype at 8 weeks post-AAV administration. n=4–6 mice per group. (L) Relative abundance of acylcarnitine metabolites in livers of the indicated genotype. n=5–6 mice per group at 8 weeks post-AAV administration. (M) Relative β-hydroxybutyrate levels in livers of the indicated genotype at 8 weeks post-AAV administration. n=4–5 mice per group. (N) Relative glutarate levels in livers of the indicated genotype at 8 weeks post-AAV administration. n=5–6 mice per group. Statistical significance was assessed using t-test (A,B,E,F,H,I,K,L,M,N) or Mann-Whitney (D) tests with adjustments for multiple comparisons. All data represent mean +/-standard deviation from biological replicates. Numerical data for individual panels are provided in Supplementary file 2 and Supplementary file 3.

The contrast in the tissue oxygen status between Ndufa9-/- and Cox10-/- livers suggests that complex I is not a significant electron donor to the hepatic ETC, at least under homeostatic conditions. If complex I was a major electron source, we would expect that depletion of complex IV might result in altered glucose utilization in the mitochondria. We therefore performed steady-state euglycemic infusions of [U-13C]glucose in Cox10f/f and Cox10-/- animals (Figure 6G). Similar to Ndufa9-/- animals, we did not observe altered glucose handling in Cox10-/- animals (Figure 6H, I; Figure 6—figure supplement 1A), suggesting that complex I activity is low and glucose utilization in the mitochondria is not sensitive to ETC function.

In principle, there are several paths of electron entry (Figure 6J) independent of complex I, including complex II (which receives electron from succinate), DHODH (which receives electrons from dihydroorotate), GPD2 (which receives electrons from glycerol 3-phosphate), and the electron transfer flavoprotein (ETF) complex (which receives electrons from FADH2). These alternate routes bypass complex I by donating electrons directly to ubiquinone. Complex IV dysfunction is expected to impact all of these routes; however, we would only expect to see significant metabolic alterations in pathways with high flux. In culture, ETC inhibition is commonly associated with elevated succinate levels; however, we did not observe this result in Cox10-/- livers (Figure 6K). Similarly, we did not observed elevations in glycerol 3-phosphate or dihydroorotate (Figure 6K), suggesting that GPD2 and DHODH activities are not a major electron source in murine livers.

Lastly, we examined whether the ETF complex constitutes a significant electron source to the hepatic ETC. The ETF complex is made up of three proteins (ETFA, ETFB and ETFDH), which transfers electrons from FADH2 to ubiquinone. Significant sources of FADH2 are thought to include fatty acids (via beta-oxidation), as well as amino acid metabolism (Figure 6J). In Cox10-/- livers, we observed significant elevations in medium/long chain acyl-carnitines (Figure 6L), as well as a decrease in β-hydroxybutyrate levels (Figure 6M), consistent with a fatty acid oxidation defect. In addition, we observed a striking increase in glutarate levels (Figure 6N), consistent with a selective defect in amino acid metabolism linked to FADH2. Thus, the metabolic abnormalities observed in Cox10-/- livers are most consistent with a buildup of ETFDH substrates.

Discussion

The dispensability of complex I for homeostatic function in the adult murine liver is not consistent with a key role for cI in electron entry to the ETC in this tissue; in particular, we observed no alterations in mitochondrial or cytosolic redox status in cI-deficient liver, or evidence of significant metabolic, histologic or functional alterations. It is possible that cI-related defects are minor and not detectable by our assays. The mouse liver is known to have relatively low flux through PDH, the major entry point of carbons to the TCA cycle which generates NADH for cI function (Cappel et al., 2019; Merritt et al., 2011); thus, our data is consistent with a model in which the hepatic TCA cycle does not play a prominent role in nutrient oxidation via the generation of mitochondrial NADH under homeostatic conditions. The impact of cIV deficiency on cellular redox status indicates that the mitochondrial ETC is critical for removing electrons via transfer to oxygen, and thus, alternative electron donors are likely to fuel the ETC. However, the lack of a survival deficit in Cox10-/- animals in this study calls into question the absolute requirement for an intact mitochondrial ETC in mouse liver.

A precise measurement of the source of electrons for the ETC is challenging due to an inability to directly track electrons after incorporation into the redox-sensitive cofactors used by the mitochondrial ETC complexes. Our results from complex IV dysfunctional livers reveal a buildup of substrates that normally feed into the ETF complex, including elevations in multiple acyl-carnitine species, as well as glutarate, a byproduct of lysine and tryptophan metabolism. In principle, the buildup of acyl-carnitines could be due to increased rates of fatty acid synthesis (FAS); however, we observed that a number of FAS enzymes were downregulated. In addition, the reduction in β-hydroxybutyrate (a terminal product of fatty acid oxidation) levels combined with the large increases in fatty acid species is consistent with a defect in fatty acid oxidation (FAO), although these steady state measurements do not directly inform on flux through FAO pathways, and experiments to directly trace flux through the ETF complex will be necessary.

In humans, mutations in proteins of the ETF complex are causative for glutaric acidemia type II disease (also known as multiply acyl-CoA dehydrogenase deficiency), which has similar biochemical abnormalities as our Cox10-/- animal model (Frerman and Goodman, 1985; Grünert, 2014). The clinical progression of these patients can be variable; however, patients with hepatic involvement exhibit hepatic steatosis, elevated ALT/AST levels and periodic hypoglycemia (Siano et al., 2021). In addition, mice carrying a knock-in ETFDH mutation displayed diet-dependent phenotypes, including hepatic steatosis and elevated levels of acyl-carnitine species (Xu et al., 2018). Lastly, patients with SCO1 mutations (impacting complex IV assembly) display hepatic steatosis and elevated acyl-carnitines (Valnot et al., 2000; Stiburek et al., 2009), suggesting a clinical overlap between mitochondrial ETC inhibition and ETF complex deficiency.

Our results therefore indicate the complex I is dispensable in the adult murine liver. A limitation of our study is that we have focused on steady state analysis of homeostatic functions in the mouse liver of the C57BL/6 J background, and it is possible that complex I does have a required role in alternative genetic backgrounds, or in the setting of hepatic stressors or alternative diets. Indeed, it has been reported that C57BL/6 J animals contain a mutation which prevents the incorporation of multimeric complex IV into supercomplexes, and so the major contributing sources of electrons for the ETC may be different in this background (Lapuente-Brun et al., 2013; Jha et al., 2016). Thus, it is possible that cI is required in other genetic backgrounds and it will be important to assess these scenarios in future work. Indeed, the relevant and high flux modes of electron entry into the ETC may be different among distinct tissues and backgrounds, and this provides a plausible hypothesis for the non-overlapping syndromes present in mitochondrial ETC diseases affecting different components. As an upstream component of the ETC, cI mutations from the mitochondrial or nuclear genome may manifest in a narrow set of tissues systems which rely on mitochondrial NADH to fuel respiration. In contrast, complex IV, the terminal component of the ETC which feeds electrons to oxygen as a final electron acceptor, might be predicted to have effects in a wider variety of tissues, particularly those which do not have alternative pathways to remove excess electrons. Understanding the key ETC fuel sources in both normal and stressed tissue systems will therefore provide insight into the pathophysiological processes which occur downstream of specific mitochondrial mutations.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or referenceIdentifiersAdditional information
Strain, strain
background
(Mus musculus,
male and female)
Cox10f/fThe Jackson Laboratory024697
Strain, strain
background
(Mus musculus,
male and female)
Ndufa9f/f
OtherpAAV.TBG.PI.eGFP.
WPRE.bGH (AAV-GFP)
Addgene105535Adeno-associated virus
OtherpAAV.TBG.PI.Cre.
rBG (AAV-Cre)
Addgene107787Adeno-associated virus
AntibodyAnti-Tom20
(rabbit polyclonal)
Proteintech11802WB, 1:2000
AntibodyAnti-PC (rabbit polyclonal)Proteintech16588WB, 1:1000
AntibodyAnti-βActin
(mouse monoclonal)
Proteintech66009WB, 1:5000
AntibodyAnti-Ndufa9
(mouse monoclonal)
ThermoScientific459100WB, 1:2000
AntibodyAnti-β2microglobulin
(rabbit monoclonal)
ThermoScientific701250WB,:5000
AntibodyAnti-Cox10
(rabbit polyclonal)
Abcamab84053WB,:1000
Commercial assay or kitComplex I Enzyme Activity
Microplate Assay
Kit (Colorimetric)
Abcamab109721
Commercial assay or kitATP Colorimetric/
Fluorometric Assay Kit
BiovisionK354
Commercial assay or kitDeproteinizing Sample
Preparation Kit
BiovisionK808
Commercial assay or kitLuna Universal One-
Step RT-qPCR kit
New England BiolabsE3005S
Commercial assay or kitDC Protein AssayBiorad5000112
Chemical compound, drug[U-13C]glucoseCambridge Isotopes
Laboratories
CLM-1396
Chemical compound, drug[U-13C]lactateCambridge Isotopes
Laboratories
CLM-1579
Chemical compound, drug[U-13C]pyruvateCambridge Isotopes
Laboratories
CLM-2440

Reagents

Most chemicals and reagents including antimycin A (A8674), carbonyl cyanide 3-chlorophenylhydrazone [CCCP (C2759)], EGTA (E3889), glutamate (49621), glycerol (G5516), hematoxylin (GHS132), HEPES (H4034), magnesium chloride (208337), malate (240176), D-mannitol (M4125), NAD+ (N8285), NADP+ (N8160), NADH (N6785), NADPH (N9960),15N5-AMP (900382), Orotic acid (O2750), L-Dihydroorotic acid (D7128), Oil Red O (O0625), periodic acid (P7875), PEG 400 (91893), potassium chloride (P9541), rotenone (R8875), sodium chloride (S9888), sodium lactate (L7022), sodium phosphate monobasic (RDD007), sodium pyruvate (P2256), sucrose (S0389), N,N,N′,N′-Tetramethyl-p-phenylenediamine [TMPD (T7394)] were purchased from Sigma-Aldrich. HPLC-grade acetonitrile (A955), formic acid (A117), methanol (A456), and water (W6) were purchased from Fisher Scientific. Eosin Y (SE23) and Tris base (BP301) was purchased from Fisher Scientific. GSH (G597951), GSSG (G597970), 13C2;15N GSH (G597952), and 13C4;15N2 GSSG (G597972) were purchased from Toronto Research Chemicals. [U-13C] lactate (CLM-1579), [U-13C] pyruvate (CLM-2440), [U-13C] glucose (CLM-1396) were purchased from Cambridge Isotope Laboratories. Schiff’s Reagent was purchased from Electron Microscopy Sciences (26052–06). pAAV.TBG.PI.eGFP.WPRE.bGH (AAV-GFP) and pAAV.TBG.PI.Cre.rBG (AAV-Cre) were a gift from James M. Wilson (Addgene viral prep #105535-AAV8 and #107787-AAV8). Antibodies used: Tom20 (Proteintech, 11802), Pyruvate carboxylase (Proteintech, 16588), Actin (Proteintech, 66009), Ndufa9 (ThermoScientific, 459100), β2-microgobulin (ThermoScientific, 701250), Cox10 (Abcam, ab84053).

Mice

The Cox10f/f mice (strain 024697) were purchased from The Jackson Laboratory and were maintained on a C57BL6 background. The Ndufa9f/f mice were obtained from David McFadden (UT Southwestern) and were on a C57BL6 background; generation of these mice are described here (Wang et al., 2022). All mice were housed in the Animal Resource Center at University of Texas Southwestern Medical Center under a 12 hr light-dark cycle and fed ad libitum. Prior to all experiments, mice were fasted for 16–18 hr over night (unless otherwise indicated). All mouse experiments were performed according to protocols approved by the Institutional Animal Care and Use Committee (IACUC) at University of Texas Southwestern Medical Center. Both male and female mice were used in all experiments; if sex differences were not present, both male and female mice were analyzed together. For genotyping, the following primers were used:

  • Cox10 Forward: GAGAGGAGTCAAGGGGACCT

  • Cox10 Reverse: GGCCTGCAGCTCAAAGTGTA

  • Ndufa9 Forward: TTGATTGCCTGTGAGCTTTG

  • Ndufa9 Reverse: TGCTAGGAAAGAGGCAGGTC

For AAV injections, virus was prepared in dilution buffer [NaCl (136 mM), KCl (5 mM), MgCl2 (1 mM), Tris (10 mM, pH 7.4), and glycerol (10%)] and injected into the retroorbital vein at 5x1010 GC per mouse.

Mitochondrial ETC activity measurements

Request a detailed protocol

Oxygen consumption rates (OCR) were measured using a Seahorse XFe96 Analyzer (Agilent Technologies) based on a previously published protocol (Rogers et al., 2011). Liver tissue was homogenized with 40 strokes of a dounce homogenizer in mitochondrial isolation buffer [HEPES (5 mM), sucrose (70 mM), mannitol (220 mM), MgCl2 (5 mM), KH2PO4 (10 mM), and EGTA (1 mM), pH 7.2]. Mitochondria were isolated via differential centrifugation: First nuclei and cell debris were removed by isolating supernatant after five spins at 600 g; and enriched mitochondria were then pelleted with a 10,000 g spin. Five μg of mitochondria were plated in an XFe96 well plate on ice, spun at 4700 RPM (Beckman Coulter S6096) for 2 min at 4 °C. Mitochondria were then supplemented with media containing isolation buffer with complex IV substrates to probe respiration. At the times indicated, ADP (final concentration 4 mM), oligomycin (2 µM), CCCP (2 µM), and either antimycin A (4 µM) or sodium azide (40 mM) was injected.

No substrateIsolation buffer (IB)
Complex IVIB with ascorbate (10 mM)+TMPD (100 uM)+antimycin A (2 uM)

For mouse embryonic fibroblast experiments, Ndufa9f/f and Ndufa9-/- MEFS were obtained as previously described (Wang et al., 2022). Briefly, E13.5 embryos were collected from Ndufa9f/f pregnant females, minced and digested with trypsin for 45 min at 37 °C. Digested embryos were cultured in DMEM (Sigma-Aldrich, D6429) supplemented with 10%FBS, 1% penicillin/streptomycin, and immortalized with a lentiCRISPRv2 lentivirus expressing sgRNA targeting Trp53. Immortalized MEFs were subsequently transduced with Ad5-CMV-EGFP or Ad5-CMV-CreEGP adenovirus (University of Iowa Viral Vector Core), followed by culturing in the above media supplemented with 1 mM sodium pyruvate and 100 μg/ml uridine. GFP + cells were sorted by flow-cytometry to obtain purified Ndufa9f/f and Ndufa9-/- cells, which were then verified by genotyping and western blot analysis, as previously reported (Wang et al., 2022). For oxygen consumption measurements, Ndufa9f/f and Ndufa9-/- cells were plated overnight and exchanged into Seahorse assay buffer and measurements started. Mitochondrial oxygen consumption rates (mito-OCR) were calculated by subtracting non-mitochondrial respiration from baseline respiration.

Complex I activity measurements were performed using an immunocapture absorbance assay (ab109721; Abcam) following manufacturer’s instructions. Approximately 100 mg of tissue were isolated from snap frozen mouse livers that had been stored at –80 °C. Tissue was homogenized in 500 μl PBS and protein concentrations were quantitated (DC Protein assay, Biorad). Samples were diluted with PBS to a concentration of 5.5 mg/ml and 90 μl were mixed with 10 μl of detergent (Abcam) and incubated on ice for 30 min. Samples were then spun down at 17,000 x g and 4 °C and the supernatants were collected and diluted to 250 μg/mL of protein. A total of 50 μg of each sample were loaded in duplicate onto the immunocapture plate and incubated for 3 hr. Experimental wells were then washed three times before adding 200 μL of reaction mixture. The OD450 was monitored over 30 min (SpectraMax), and the resulting reaction curves were fit by linear regression (Prism Graphpad) and slopes reflecting Complex I activity are reported by as mOD 450 /μg protein/min.

ATP measurements were carried out with an ATP Colorimetric/Fluorometric assay kit (K-354, BioVision Inc). Approximately 10 mg of tissue were isolated from snap frozen mouse livers that had been stored at –80 °C. Tissues were homogenized in 100 μl of ATP assay buffer and were rapidly deproteinized by PCA precipitation (K-808, BioVision Inc). Ten μl of deproteinized sample were diluted with 40 μl of ATP assay buffer and loaded in duplicate into a clear bottom 96 well plate. A total of 50 μl of developer solution was added to each well and the plate was incubated for 30 min at room temperature protected from light. The endpoint OD570 of sample wells was measured on a Spectramax iD3 spectrometer (SpectraMax, Molecular Devices) and the total amount of ATP was determined from a standard curve constructed with 0–10  nmol ATP. Reported values are normalized by tissue weight.

qRT-PCR and qPCR measurements

Request a detailed protocol

Genomic DNA was isolated by ethanol precipitation and RNA was isolated from tissue using RNeasy Mini kits (Qiagen). Samples were run using the Luna Universal One-Step RT-qPCR Kit (New England Biolabs) on a CFX384 (Bio-Rad).

TargetPrimer Sequence
Cox10F: AGGGTCAGCATCACCAATAC
R: GGAGACACTTACCAGCATCAA
Ndufa9F: GGAGACACTTACCAGCATCAA
R: CCTCCTTTCCCGTGAGGTA
mtDNA
(ND2 mtDNA)
F: CCTATCACCCTTGCCATCAT
R: GAGGCTGTTGCTTGTGTGAC
nDNA
(Pecam1)
F: ATGGAAAGCCTGCCATCATG
R: TCCTTGTTGTTCAGCATCAC

Histology of liver tissue

Request a detailed protocol

Fresh tissue was fixed overnight in 10% neutral buffered formalin, rinsed with 30% sucrose in PBS, and paraffin embedded. Ten micron sections were cut and placed on slides at –20 °C until staining. Sections were stained by either haemotoxylin and eosin, Oil Red O, or periodic acid Schiff base (PAS) and visualized under ×20 magnification on an Olympus IX83 microscope and processed with ImageJ (NIH).

Plasma analysis

Request a detailed protocol

Blood was collected on ice with heparin and centrifuged at 5000xg for 10 min to obtain plasma. Plasma levels of alanine aminotransferase, albumin, aspartate aminotransferase, glucose, lactate, total bilirubin, and total protein were measured using a VITROS 250 Microslide at the UT Southwestern Metabolic Phenotyping Core.

GSH/GSSG and NAD(P)(H) quantitation

Request a detailed protocol

For glutathione quantitation, 20 mg of frozen tissue was treated with 300 μL of 80% methanol with 0.1% formic acid and homogenized on ice. The samples were freeze-thawed, centrifuged to remove protein, and the supernatant was dried. Dried samples were reconstituted in 1 mL water with 0.1% formic acid, diluted, spiked with internal labeled standards, and analyzed as previously described (Ubellacker et al., 2020; Tasdogan et al., 2020).

NAD(P)+/NAD(P)H was measured as previously described (Lu et al., 2018). Briefly, ~20 mg of tissue was homogenized on ice in cold 40:40:20 (acetonitrile: methanol: water) with 0.1 M formic acid buffer. Tissue was vortexed, chilled, quenched with 15% ammonium bicarbonate, and centrifuged to pellet protein. Supernatant was diluted in 10 mM ammonium bicarbonate and 15N5-AMP was added as an internal standard and injected onto a Sciex 6500 Qtrap using a reverse-phase ion pairing method. Pelleted protein was lysed and quantified using the DC protein assay (Bio-Rad).

For MEFs, cells were plated in six-well dishes and allowed to adhere overnight; the following day cells were at ~80% confluence. Media was removed, cells were rinsed with ice cold saline, and metabolites were collected by scraping cells in dry ice-cold 80% methanol / 20% water with an internal standard (series of 13 C, 15N-labeled amino acids). Metabolite extracts were stored at –80 °C overnight and centrifuged the following day to remove debris. Raw AUC values were obtained using MultiQuant and TIC values for normalization were obtained using Analyst (SCIEX software for MS analysis).

Proteomics analysis of liver tissue

Request a detailed protocol

Protein homogenates were prepared from either whole liver tissues or enriched mitochondria. For whole liver tissue, approximately 100 mg of tissue were isolated from snap frozen mouse livers that had been stored at –80 °C. Tissue was homogenized in 500 μl PBS supplemented with protease inhibitors (Thermo Fisher, 78425) and protein concentrations were quantitated (DC Protein assay, Biorad). A total of 50 μg of each sample was solubilized in 1% SDS, 50 mM Tris pH 8.0 at a concentration of 5 mg/ml, run on a 4–20% Mini-PROTEAN TGX precast protein gel (BioRad) into the top of the resolving portion of the gel, stained with Coomassie Blue, and destained. For enriched mitochondria, approximately 100 mg of liver tissue was homogenized with 40 strokes of a dounce homogenizer in mitochondrial isolation buffer [HEPES (5 mM), sucrose (70 mM), mannitol (220 mM), MgCl2 (5 mM), KH2PO4 (10 mM), and EGTA (1 mM), pH 7.2] supplemented with protease inhibitors (ThermoFisher, 78425). Mitochondria were isolated via differential centrifugation: First nuclei and cell debris were removed by isolating supernatant after five spins at 600 g; and enriched mitochondria were then pelleted with a 10,000 g spin. Protein concentrations were quantitated (DC protein assay, BioRad), and 50 μg of each sample was solubilized in 1% SDS, 50 mM Tris pH 8.0 at a concentration of 5 mg/ml. Samples were run on a 4–20% Mini-PROTEAN TGX precast protein gel (BioRad) into the top of the resolving portion of the gel, stained with Coomassie Blue, and destained.

Samples were analyzed at the UT Southwestern Proteomics Core. Gel slices were cut into 1 mm3 cubes, and digested overnight with trypsin (Pierce 90058) following reduction and alkylation with DTT and iodoacetamide (Sigma–Aldrich I6125). The samples then underwent solid-phase extraction cleanup with an Oasis HLB plate (Waters) and the resulting samples were injected onto an Orbitrap Fusion Lumos mass spectrometer coupled to an Ultimate 3000 RSLC-Nano liquid chromatography system. Samples were injected onto a 75 μm i.d., 75 cm long EasySpray column (Thermo) and eluted with a gradient from 0–28% buffer B over 90 min. Buffer A contained 2% (v/v) ACN and 0.1% formic acid in water, and buffer B contained 80% (v/v) ACN, 10% (v/v) trifluoroethanol, and 0.1% formic acid in water. The mass spectrometer operated in positive ion mode with a source voltage of 1.8 kV, ion transfer tube temperature of 275 °C, MS1 AGC target of 400000, MS1 maximum injection time of 50ms, intensity threshold of 5000, MS2 AGC target of 10000, MS2 maximum injection time of 35ms, MS2 isolation window of 1.6 Da. MS1 scans were acquired at 120,000 resolution in the Orbitrap and up to 10 MS/MS spectra were obtained in the ion trap for each full spectrum acquired using higher-energy collisional dissociation (HCD) for ions with charges 2–7. Dynamic exclusion was set for 25 s after an ion was selected for fragmentation.

Raw MS data files were analyzed using Proteome Discoverer v2.4 SP1 (Thermo), with peptide identification performed using Sequest HT searching against the mouse protein database from UniProt (downloaded January 2022; 17,062 sequences) assuming a trypsin digestion (cleavage after Lys and Arg except when immediately followed by Pro). Fragment and precursor tolerances of 10 ppm and 0.6 Da were specified, and three missed cleavages were allowed. Carbamidomethylation of Cys was set as a fixed modification, with oxidation of Met set as a variable modification. The false-discovery rate (FDR) cutoff was 1% for all peptides and all PSMs were validated with the Percoloator node within Proteome Discoverer. PSMs found in only a subset of samples were re-searched to identify peptides based on retention time and mass. Protein abundance was quantitated based on the total ion count for all identified peptides in each sample and only proteins notated with FDR <1% were considered. Protein abundances were normalized after log2 transform according to previously published protocols using Microsoft Excel 2019 (Aguilan et al., 2020): First, abundance values were log2 transformed, then each sample was normalized by the average abundance and data distribution width. Missing values were imputed as random values centered around 2.5 standard deviations below the mean abundance value. For samples with >4 replicates, Shapiro-Wilk tests were performed to test for normal distributions. If samples were not normally distributed, Mann-Whitney tests were used to calculate p-values. For normally distributed data, F-tests were used to assess equal variance, followed by two-tailed homoscedastic (or heteroscedastic) tests to calculate p-values. Unsupervised hierarchical clustering was performed on z-score transformed values in MATLAB (Mathworks, Inc). Mitochondrial-localized proteins were classified based on their presence in MitoCarta3.0 (Calvo et al., 2016). Significantly up and downregulated proteins were identified based on log2(fold change)>1 or<1 (respectively) and p-value <0.05. Gene ontology analysis on up or downregulated proteins was performed using DAVID (https://david-d.ncifcrf.gov/home.jsp); for each analysis, the complete list of detected proteins was used as a background, and pathways from Biological Processes, Cellular Components and Molecular Function (All and Direct) were assessed. Pathways with a FDR <0.05 are reported in Supplementary file 2. Gene set enrichment analysis was performed using GSEA software (Subramanian et al., 2005; Mootha et al., 2003) using the following gene-ontology pathways: GO:0006633 (Fatty Acid Synthesis), GO:0019395 (Fatty Acid Oxidation), GO:0008206 (Bile Acid Metabolic Process), GO:0016491 (Oxidoreductase Activity).

In vivo isotope tracing

Request a detailed protocol

All in vivo isotope tracing experiments were performed in conscious mice after overnight fasting, based on previously published protocols (Faubert et al., 2021; Davidson et al., 2016). On the morning of the experiment, catheters (Braintree Scientific MRE-KIT 025) were surgically implanted into the right external jugular vein of mice. For steady state euglycemic infusion of [U-13C]glucose, a total dose of 8 g/kg body mass (dissolved in 1000 μl saline) was continuously infused at 2.5 μl min–1 (total infusion time of 3 hr). At the end of the infusion, mice were euthanized and tissue was immediately harvested, snap frozen in liquid N2, and stored at –80 °C prior to GC-MS analysis.

Metabolomics analysis

Request a detailed protocol

Flash frozen liver tissue (20 mg) was pulverized with a mortar and pestle on liquid nitrogen (H37260-0100, Bel-Art Products), 80% methanol was added, and samples subjected to three freeze-thaw cycles. For plasma, 20 μL was extracted with 0.2 mL cold acetone. Protein was removed by centrifugation and supernatant dried (SpeedVac, ThermoFisher).

Data acquisition was performed by reverse-phase chromatography on a 1290 UHPLC liquid chromatography (LC) system interfaced to a high-resolution mass spectrometry (HRMS) 6550 iFunnel Q-TOF mass spectrometer (MS) (Agilent Technologies, CA). The MS was operated in both positive and negative (ESI +and ESI-) modes. Analytes were separated on an Acquity UPLC HSS T3 column (1.8 μm, 2.1x150 mm, Waters, MA). The column was kept at room temperature. Mobile phase A composition was 0.1% formic acid in water and mobile phase B composition was 0.1% formic acid in 100% ACN. The LC gradient was 0 min: 1% B; 5 min: 5% B; 15 min: 99% B; 23 min: 99% B; 24 min: 1% B; 25 min: 1% B. The flow rate was 250 μL min-1. The sample injection volume was 5  μL. ESI source conditions were set as follows: dry gas temperature 225  °C and flow 18  L min-1, fragmentor voltage 175  V, sheath gas temperature 350  °C and flow 12  L min-1, nozzle voltage 500 V, and capillary voltage +3500 V in positive mode and −3500 V in negative. The instrument was set to acquire over the full m/z range of 40–1,700 in both modes, with the MS acquisition rate of 1 spectrum s-1 in profile format. Raw data files were processed using Profinder B.08.00 SP3 software (Agilent Technologies, CA) with an in-house database containing retention time and accurate mass information on 600 standards from Mass Spectrometry Metabolite Library (IROA Technologies, MA) which was created under the same analysis conditions. The in-house database matching parameters were mass tolerance 10 ppm; retention time tolerance 0.5  min. Peak integration result was manually curated in Profinder for improved consistency. Metabolite abundances were normalized based on the total ion count for all identified metabolites. Unsupervised hierarchical clustering was performed on z-score transformed values in MATLAB (Mathworks, Inc), and adjusted p-values were calculated using multiple t-test with adjust for multiple comparisons in Graphpad Prism.

GC-MS analysis was used to quantitate α-hydroxybutyrate and α-ketobutyrate, as well as measure mass isotopomer distributions for the lactate/pyruvate tolerance test and [U-13C]glucose infusions. Dried metabolites were derivatized to form methoxime-TBDMS adducts by incubating with 1% methoxyamine hydrochloride (Sigma-Aldrich) in pyridine at 70 °C for 15 min followed by addition of N-tert-Butyldimethylsiyl-N-methyltrifluoroacetamide (MTBSTFA, Sigma-Aldrich) for 1 hr. Derivatized samples were analyzed using an Agilent Technologies 7890B gas chromatographer with a HP-5MS 5% phenyl methyl Silox column (Agilent) coupled to an Agilent Technologies 5977 A mass spectrometer. The observed distributions of mass isotopologues were corrected for natural abundance (Fernandez et al., 1996).

Lactate/pyruvate tolerance test

Request a detailed protocol

Lactate/Pyruvate tolerance tests were performed as previously described (Cappel et al., 2019). Briefly, mice were fasted for 16 hr overnight and given an IP injection of a 1.5 mg/g of a 10:1 sodium lactate: sodium pyruvate solution. 40% of the sodium lactate and sodium pyruvate were [U-13C] sodium lactate and [U-13C] sodium pyruvate. Blood glucose was measured 0, 10, 20, and 30 min post injection using a blood glucometer (Contour Next) and lactate was measured using a lactate meter (Nova Biomedical Lactate Plus). Fractional gluconeogenesis (GNG) was determined as previously described (Kelleher, 1999), using the following formula:

GNG=(M1glucose+M2glucose+M3glucose)/(2M0lactate(M1lactate+M2lactate+M3lactate))

where Mx represents the fractional enrichment of the m+x species for the indicated metabolite.

Magnetic resonance imaging experiments

Experiments were performed on mice at 4 weeks post-AAV administration. MR imaging was performed on a 7T pre-clinical scanner (Bruker Biospec, Germany) using a 40 mm Millipede radio frequency coil for transmitting and receiving. Mice were anesthetized with 1.5–2.5% Isoflurane. Mice breathed medical air (21% O2) delivered via a nose cone, in the baseline acquisition and then the gas was switched to 100% O2 (Airgas, Radnor Township, PA, USA). The gas flow rate was set to 400 ml/min. A pneumonic sensor placed on the abdomen was used for respiratory monitoring and prospective triggering (SA Instruments, Stony Brook, NY). The animal’s ambient temperature was maintained at 28 °C using an MR Compatible Small Rodent Air Heater System (SA Instruments, Stony Brook, NY). For a given animal, the breathing rate was maintained at 40 ± 5 bpm by varying the Isoflurane levels by ± 0.5%. At each breathing challenge, images were acquired to generate T1 and T2* maps of the liver in the coronal plane.

T1 method

Request a detailed protocol

A respiratory triggered, radio frequency spoiled, flow compensated cine gradient echo method (FLASH) was used. To suppress the motion artifacts from cardiac motion, a saturation pulse was applied to the thoracic region. Two cines (number of frames =65) were acquired with two different flip angles (α=5° and 20°). A steady state signal was achieved after 55 frames. Frames 55–65 were averaged to improve the signal-to-noise ratio. Pixel-by-pixel apparent (uncorrected for B1 inhomogeneity) T1 maps were calculated using the averaged signal intensities (S) according to Helms et al., 2008, Sα= -2TRT1Sα+const. Cine FLASH parameters were: TR/TE =10ms/2.3ms, FOV =40 x 40 mm2, matrix =128 x 128, NEX =1, slice thickness =1 mm, number of segments =1, number of frames =65, in-plane resolution =0.3 mm x 0.3 mm.

Using the apparent T1 maps, and noting that any spatial dependencies due to B1 inhomogeneity are canceled out in relative measures, a map of the actual T1 difference between the gas challenges (Air vs. 100 %O2) was calculated as: T1=100*(T1O2-T1air)/T1air . The mean relative T1 of the liver was calculated in a manually segmented liver region. Pixels representing blood vessels were automatically excluded from this region by thresholding to mean of liver T1 tissue ±2 × standard deviations. The mean of liver T1 was calculated from a region of interest that did not include blood vessels.

T2* method

Request a detailed protocol

A respiratory triggered, bipolar-multi gradient echo (MGE) sequence was used. Sixteen echoes from positive gradients were acquired at 2.5ms intervals. The scan plane geometry and the orientation was identical to that of the T1 method. MGE parameters were: TR = 300ms, TE = 2.5 ms – 47.5 ms, NEX = 4. T2* maps were generated by non-linear least square fitting the echo signals (s) to an exponential model: A*e-B*TE+C. The mean relative T2* was calculated as similar to T1.

Perfusion method

Request a detailed protocol

FAIR arterial spin labeling technique was used to calculate the liver perfusion in a single-slice in the axial plane during the session where mice breathed medical air (21% O2). The FAIR sequence was a multi-echo spin-echo with a 13.6ms inversion pulse. A slice select inversion with an inversion slab of 4 mm was followed by a global inversion. The inversion time was set to 2000ms. A reference image was acquired without the inversion pulse. Other imaging parameter were: TR = 10 s, FOV = 30 x 25 mm2, matrix = 128 x 64, NEX = 1, slice thickness = 1 mm, echo train length = 8. A pixel wise perfusion map was constructed as described previously (Kim and Tsekos, 1997).

Statistical analyses

Request a detailed protocol

All data represent mean and standard deviations from biological replicates. No statistical tests were used to predetermine sample size. Animals were randomly allocated into experimental groups. No blinding or masking of samples was performed. Data sets for each group of measurement was tested for normality using the Shapiro-Wilk test. If the data was not normally distributed, the data was log-transformed and retested for normality. For normally-distributed data, groups were compared using the two-tailed Student’s t-test (for two groups), or one-way ANOVA or two-way ANOVA (>2 groups), followed by Tukey’s or Dunnett’s test for multiple comparisons. For data that was not normally distributed, we used non-parametric testing (Mann-Whitney or Kolmogorov-Smirnov tests for two groups and Kruskal-Wallis test for multiple groups), followed by Dunn’s multiple comparisons adjustment. For metabolomics and proteomics data sets, abundances were compared by multiple t-tests, followed by Bonferroni correction of p-values. Proteomics and metabolomics analysis were performed once using biological replicates (individual mice); for other experiments, multiple (two to four) independent experiments with biological replicates (individual mice) were performed for reported data, and the number of biological replicates are indicated in the figures. No data were excluded.

Data availability

Proteomics datasets have been deposited into the PRIDE database (identifier PXD031685), and metabolomics datasets have been deposited into Metabolomics Workbench (identifier PR001484). All other data are provided within the article and supplementary files.

The following data sets were generated
    1. Lesner NP
    2. Mishra P
    (2022) PRIDE
    Mitochondrial Hepatopathy Proteomics – Whole Tissue.
    https://doi.org/10.6019/PXD031685
    1. Lesner NP
    2. Mishra P
    (2022) Metabolomics Workbench
    Differential requirements for mitochondrial electron transport chain components in the adult murine liver.
    https://doi.org/10.21228/M8WT54

References

    1. Ahmed ST
    2. Craven L
    3. Russell OM
    4. Turnbull DM
    5. Vincent AE
    (2018) Diagnosis and treatment of mitochondrial myopathies
    Neurotherapeutics : The Journal of the American Society for Experimental NeuroTherapeutics 15:943–953.
    https://doi.org/10.1007/s13311-018-00674-4

Decision letter

  1. Agnieszka Chacinska
    Reviewing Editor; University of Warsaw, Poland
  2. Vivek Malhotra
    Senior Editor; The Barcelona Institute of Science and Technology, Spain

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Differential requirements for mitochondrial electron transport chain components in the adult murine liver" for consideration at eLife. Your initial submission has been assessed by a Senior Editor in consultation with members of the Board of Reviewing Editors. Although the work is of interest, we regret to inform you that the findings at this stage are too preliminary for further consideration at eLife.

The reviewers find the topic and findings potentially interesting and significant. The reviewers have brought up serious criticism (comments appended below), which precludes invitation to revise at this stage. However, they encourage you to resubmit. and new manuscript to eLife with a more full picture including a mechanistic evidence requested in their comments.

Specifically, the evidence for the involvement of DHODH is not sufficient. Thus, this interesting and important conclusion remains not fully proven. Furthermore, the proteomic and metabolic analyses are of insufficient depth and/or not analysed thoroughly to its maximum potential.

Reviewer #1 (Recommendations for the authors):

In this work, Lesner and colleagues study the effects of mitochondrial electron transport chain dysfunction at the level of mitochondrial complex I in comparison to complex IV in mouse liver. They report no adverse effects on liver functions in a complex I-deficient mouse model, whereas complex IV deficiency results in decreased liver function. To provide further insight into the mechanisms underlying these different phenotypic outcomes, the authors collect experimental data for alterations in the redox status as well as the proteome and metabolome in complex I- and complex IV-deficient mouse liver. These data generally support the notion that complex I activity is dispensable for liver function. The authors explain this finding by providing evidence for the involvement of mitochondrial enzyme dihydroorotate dehydrogenase as an alternative electron donor to maintain respiratory chain activity. These data add to the field of mitochondrial diseases associated with the dysfunction of mitochondrial respiratory chain complexes and tissue-specific phenotypes.

The authors generally present their data in a clear and logical way, which makes the story easy to follow. However, the description of proteomics experiments including LC-MS analysis in Materials and methods is largely incomplete and needs to be fully revised to provide detailed information about sample preparation, protein identification and quantification. In addition, Supplemental Tables for proteomic (and metabolomics) data are insufficient lacking information about gene names, identified peptides per replicate, ratios, calculated p-values etc. Proteomics data should also be deposited to a public database (e.g. PRIDE).

The authors performed label-free proteomics to investigate changes in the proteome of ndufa9-/- and cox10-/- mice vs. control mice. It should be noted that only less than 1/3 of the proteome was covered and thus they likely missed also some further regulated proteins in their analysis. Since they focus on mitochondrial processes, it is suggested to also analyze mitochondria-enriched fractions. Furthermore, the authors should improve their data presentation. For example, it would be informative to prepare additional plots showing the distribution of individual OXPHOS complexes in ndufa9-/- and cox10-/- liver samples. Do they generally shift to more positive or negative log2 values? In addition, the analysis of OXPHOS complexes by blue native PAGE is recommended to reveal changes in supercomplex formation or assembly states.

To provide further insight into compensatory effects in cox10-/- livers, GO term (BP, CC, MP) overrepresentation analysis of both downregulated and upregulated proteins should be performed. More specifically, the authors should provide information about the down-regulated proteins other than only complex IV components and upregulated proteins other than mitochondrial components. They should analyze the proteomic data in more detail, including the annotation of upregulated non-mitochondrial and mitochondrial proteins in Figures 2E and 4E. Furthermore, how do the authors explain the more than 2-fold increase in mitochondrial content in cox10-/- livers? It might be of interest to acquire EM or immunofluorescence images on these mouse liver samples.

Based on GO term enrichment analysis, the authors state that several metabolic pathways were enriched in cox10-/- livers. However, what is missing is a correlation between the observed changes in the proteome and metabolome. Do changes in the levels of specific metabolites result from the up- or down-regulation of proteins of the respective pathways?

In Figure 5B, quantification of Western blots should be performed and further replicates might likely be needed to make conclusions whether the levels of DHODH do not change in ndufa9-/- mice. In the reviewer's point of view, DHODH levels are slightly increased in liver from ndufa9-/- mice.

Reviewer #2 (Recommendations for the authors):

Strengths of the manuscript.

The fundamental data pillars supporting the authors top-line conclusion are solid – i.e. that liver damage is only apparent in complex IV, but not complex I, knockout adult mice.

After complex IV knockout the circulating clinical-biomarkers of liver injury (AST/ALT/Bilirubin) are elevated with statistical significance. In agreement, the complex IV knockout livers exhibits marked damage in the pathology assessment. Furthermore, the authors confirm robust changes in the liver proteome and metabolome following complex IV knockout; these effects are commensurate with mitochondrial dysfunction and adaptive stress signalling. Effects of complex I deletion on liver homeostasis are either minor or not detectable within the parameters assessed here; the difference between the effect of deleting complex IV versus complex I is clear and justified by data presented in the manuscript.

In the context of mitochondrial biology there is general lack of conditional knockout studies in adult mice. Accordingly, this study offers novel and unexpected insights into the relationship between respiratory chain components and hepatic function – the observation that adult mouse liver exhibit greater dependence on complex IV is of notable conceptual advance.

Weakness of the manuscript.

The key weakness of the manuscript is the lack of convincing mechanistic rationale for the differential effects of complex IV and complex I knockout on adult mouse liver.

The authors show that complex I knockout ~fully suppresses respiration (OCR) in freshly isolated hepatocytes (Figure 1C). Because the NADH/NAD+ ratio is unaffected in complex I-knockout livers the authors infer that another enzyme in the respiratory chain (dihydroorotate dehydrogenase (DHODH)) is catalysing NADH oxidation (to stop NADH accumulation/build-up) – however NADH oxidation by DHODH drives respiration (as depicted in Figure 5A) but Figure 1C shows very clearly that respiration rates are negligible in hepatocytes devoid of complex I activity/expression. Therefore, the authors explicitly show in their own data that respiratory activity is not maintained by DHODH in the absence of complex I.

Furthermore the authors do not present respiration/OCR data in hepatocytes isolated from complex IV knockout livers; there we cannot assess if hepatic respiration is differentially affected by complex I versus complex IV knockout- if the respiration rates are equivalently affected this would show that liver damage following complex IV deletion are not ascribable to low respiration rates/activity.

Next, the authors attempt to present data to support their model that DHODH is central to maintaining liver homeostasis in livers devoid of complex I, and treat mice with a small-molecule-inhibitor BRQ (an inhibitor of DHODH). The authors observe changes in the hepatic NADH/NAD+ ratio and interpret this data to support the DHODH model. However, the data is inadequate to support the authors conclusion;

1) There is already scant respiration in the complex I knockout hepatocytes (DHODH is not compensating for complex I deletion); 2) the authors do not show if BRQ effects respiration in control hepatocytes nor the remnant respiration in hepatocytes from complex I or complex IV livers; 3) whilst BRQ suppresses liver NADH /NAD+ its' impossible to ascribe this pharmacological effect to DHODH-specific inhibition ( it could be an alternative enzyme/target). The authors are not in a position to determine specificity of BRQ to DHODH and to make credible robust mechanistic interpretation; complementary genetic approaches would be required (i.e. showing DHODH knockout in the complex I knockout mouse liver recapitulates the complex IV knockout effect); 4) the authors present no in vivo evidence that BRQ treatment recapitulates the complex IV knockout phenotype – neither circulating markers (ALT/AST/bilirubin) nor pathology are presented.

Taken together the current model ascribing the lack of liver damage following complex I deletion to DHODH is not justified with existing data.

Recommendations:

1) For reasons outlined above the model ascribing the lack of liver damage following complex I deletion to DHODH is not justified with existing data. My recommendation is the data should be withdrawn from the manuscript – otherwise many months of experiments are required to support the possible role of DHODH. However, even without mechanistic insight into why complex IV deletion, but not complex I, exerts liver damage the finding in-itself is of sufficient novelty to justify publication.

2) Respiration/OCR data from the complex IV knockout hepatocytes is critical to aid data interpretation and provide mechanistic insight – it must be included and contrasted to the OCR in complex I knockout hepatocytes.

3) Whilst complex IV clearly exerts liver damage (~4x elevation in ALT and AST and histological staining) the level of damage (4-8 weeks after the knockout) is not overly severe and the animals do not undergo acute liver failure. The text and discussion need to better reflect this. Isn't the story here that mouse liver is not absolutely dependent on respiration to maintain function – otherwise we would expect rapid liver failure soon after the complex IV deletion?

4) Do the authors have data on hepatic ATP levels in absence of complex I and complex IV – both from liver tissue as well as the isolated hepatocytes? Are hepatic ATP levels maintained in both knockout models or is glycolysis maintaining ATP to ~homeostatic levels – the authors have shown respiration is essentially abolished upon complex I deletion, so how is ATP production supported? What is the ECAR in the complex I and complex IV knockout hepatocytes (the authors must have the ECAR data for complex I knockout because the OCR data is presented in figure 1). More generally the text doesn't address or discuss the concept of glycolytic adaptive capability in the absence of respiration.

5) In places the text is too assertive and not reflective of the data – more nuance and precision is required as to not over-interpret data, for example;

Page11 – 'The phenotype of cox10-/- livers noted above indicates that the mitochondrial ETC is required for liver homeostasis'.

How is this justified when the authors have already shown in figure 1 that the complex I knockout hepatocytes exhibit little respiration but liver homeostasis is not measurably effected? Furthermore, the authors do not show the respiration/OCR data in complex IV knockout hepatocytes.

'Together, these results indicate that complex I is largely dispensable for homeostatic function in the adult mouse liver, without significant metabolic or proteomic compensation'

Disagree. Respiration is dramatically reduced in complex I knockout hepatocytes – if no major adaption is ongoing how are hepatic ATP levels supported? There is presumably significant adaption to glycolysis and the PPP pathway to preserve ATP levels?

'Together, these results indicate that complex I is largely dispensable for homeostatic function in the adult mouse liver, without significant metabolic or proteomic compensation'

The authors have assessed liver health 4 and 8 weeks after complex I (and complex 4) deletion. The text needs to explicitly reflect these two sampling time points. Perhaps complex I knockout would affect hepatic function over a longer period?

Is 4 weeks knockout the shortest time point the authors have assessed? Maybe there is a transient stress/death response to complex I deletion (liver stress/death (ALT/AST rises)) but the liver is able to adapt to complex I loss and restore homeostasis before the initial 4 week sampling.

Therefore general statements regarding dispensability of complex I in liver need much greater refinement and nuance.

6) Complex I knockout livers (Supp figure 1D_ – 'Histological analysis of liver sections revealed no significant pathology, including no evidence of steatosis (Oil Red O staining) or glycogen loss (PAS staining)'.

They authors say 'significant', would they agree that are some changes in liver pathology (HandE and the Oil red O staining in particular) albeit they are far less severe compared to the cx4 knockout? If they agree the text needs to better reflect this.

7) Metabolomics for complex I knockout in figure 2 – is this data from 4 or 8 weeks? If its 4 weeks can the authors also show the 8 week data?

8) It will be helpful to label the charts for metabolites as' liver' or 'circulating' to make data interpretation clearer for reader.

Reviewer #3 (Recommendations for the authors):

Lesner and colleagues address the tissue specific phenotype seen in mitochondrial complex I deficiency, the most common cause of paediatric mitochondrial disease. Understanding this problem may lead to more personalised treatment strategies for patients with the disease. The authors leverage mouse models featuring liver specific knockout of complex I subunit Ndufa9 and complex IV assembly factor COX10, previously shown to result in loss of complex I and IV activity respectively. Using a range of classical biochemistry, pathology, proteomics and metabolomic approaches, the authors show that while loss of complex IV leads to severe impacts on cellular redox systems and metabolic compensation as expected, the most telling readout being NAD+/NADH ratio, complex I is not required to maintain metabolically normal mitochondria and livers from these animals are phenotypically similar to controls. Based on accumulation of metabolites in the pyrimidine synthesis pathway in complex IV deficient animals, the authors build a hypothesis that liver mitochondria preferentially utilise DHODH as a primary source of electrons for mitochondrial respiration. DHODH is a key enzyme in the pyrimidine synthesis pathway and an alternate source of electrons for the respiratory chain, effectively bypassing complex I by donating electrons directly to coenzyme Q which are then utilised by complex III (and subsequently passed to complex IV). Using the DHODH inhibitor brequinar the authors are able to generate a severe metabolic defect (the primary readout again NAD+/NADH ratio) in complex I deficient liver, but not controls. The main conclusion is that under normal conditions complex I is not the main entry point of electrons into the mitochondrial respiratory chain in liver, but rather pyrimidine synthesis, possibly supplemented by other sources. It is important to note that the main conclusion of preferential utilisation of pyrimidine synthesis derived electrons is drawn from a single experiment measuring the NAD+/NADH ratio. Additional metabolomic tracer studies using glutamine (and possibly other carbon sources aimed at delineating the other electron sources hinted at by the authors in the discussion) could be used to confirm if there is increased flux through pyrimidine synthesis.

The manuscript is well written, succinct and experiments are all of the highest quality. I have no major concerns in respect to the first conclusion that complex I deficient livers are phenotypically normal, and the resulting suggestion that complex I is not the major source of electrons for the ETC in liver. My concern is the conclusion that the main source is DHODH. This is drawn from a single (but apparently well controlled, though I'm not an expert on liver physiology) experiment using brequinar, with the NAD+/NADH ratio being the key readout. Increased flux doesn't always equate to changes in enzyme protein level or metabolite pool size, thus it not surprising that DHODH protein and metabolites in the pyrimidine synthesis pathway are not increased in abundance in ndufa9-/- livers. Indeed, except for dihydroorotate and orotate even in the extreme of BRQ treated animals you don't see accumulation of pyrimidine pathway metabolites. The authors should consider ways to measure flux directly, for example tracer studies using 13C-glutamine (or possibly H13CO3) could be used on primary hepatocytes from both knockouts. Inclusion of 13C-glucose would also go some way to addressing the possible increased flux through other pathways, e.g. complex II/TCA suggested by the authors in the discussion.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Differential requirements for mitochondrial electron transport chain components in the adult murine liver" for further consideration by eLife. Your revised article has been evaluated by Vivek Malhotra (Senior Editor) and a Reviewing Editor.

The reviewers have discussed their reviews with one another, and agree that the manuscript has significantly improved and is on track to publication after you complete a set of requested modifications to the text mostly for clarity of data presentation and its broad availability for the community. They also indicate the importance of one additional experiment, metaproteome of cox10-/- vs cox10f/f, which seems rather straightforward to provide unless rationally argued otherwise.

The Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) In the revised manuscript the authors hypothesize that ETFDH is the major electron donor in the liver. This is mainly based on the elevation of related proteins and metabolites identified in proteomics and metabolomics data. The evidence for ETFDH is thus still circumstantial so conclusions should be cautious. Based on the tone of the authors in the main text I think they agree, however some of the statements could be softened in the abstract and discussion.

2) The authors should carefully describe the underpinnings of some of the less common techniques being used within the text narrative. Eg., this is the first time I've seen MR imaging and I felt I was just thrown into the deep end of what it measures. Likewise, the authors don't provide non-experts with sufficient background to understand tracer metabolomics. I also didn't find the schematic in Figure 3A particularly helpful as the use of dots to describe isotopologues doesn't really help the reader interpret the associated bar plots. This could probably be solved by a more detailed figure legend.

3) Additional points concerning metabolomics:

– Tracer metabolomics data should be included in the supplement.

– All metabolomics data should be made available in a public repository (e.g. Metabolomics workbench).

4) Data has been deposited in PRIDE as requested, however, the methods section in the manuscript requires some further work.

a) The authors wrote "50 ug per sample of protein homogenate from livers or enriched mitochondria (prepared as described above)", yet is it not clear which sample preparation steps described above they refer to. The authors should clearly state the steps they performed before they solubilized proteins in SDS/Tris.

b) The descriptions in the methods section still lack basic information.

Please add the following parameters to the LC-MS/MS data acquisition section:

AGC target for MS1 and MS2

Max IT for MS1 and MS2

Resolution for MS2

Isolation window for MS2

Intensity threshold

c) The authors wrote, "Fragment and precursor tolerances of 10 ppm and 0.6 Da were specified, and three missed cleavages were allowed". Please name the enzyme and state the cleavage rules for the search.

d) The authors wrote: "The false-discovery rate (FDR) cut-off was 1% for all peptides". Please provide the FDR cut-off for proteins as well. Please add information on whether PSMs were validated with Percolator (or another script) applied in the PD workflow.

e) There is clearly a mistake in the following sentence "Significantly up and downregulated proteins were identified based on log2(fold change) <1 or <1 (respectively) and p-value<0.05". Please correct it.

f) The authors wrote, "Protein abundance was quantitated based on the total ion count for all identified peptides in each sample, and normalized after log transform according to previously published protocols (60)". The authors should state in the methods section how the normalization was performed (rather than just referring to a paper).

g) The tables containing proteomics data look better, my additional suggestion would be to:

1) Add a unique peptide count per replicate (this parameter is more informative than peptide count).

2) Add a score parameter for protein identification.

5) Since the proteome coverage in the WCL samples was rather low (2.6-3k proteins, of which only 396-407 are annotated as mitochondrial in MitoCarta3.0), the authors were asked to analyze proteins in isolated mitochondrial fractions. The authors followed the recommendation only for ndufa9-/-, however, they did not do it for cox10-/-.

The additional analysis they performed yielded quantification data for 738 mitochondrial proteins in isolated mitochondria compared to 396 mitochondrial proteins in WCL samples, which clearly shows an improvement. As the proteomic analysis of isolated mitochondria ndufa9-/- vs ndufa9f/f in fact nearly doubled the number of quantified mitochondrial proteins reported, the authors should also include a similar analysis for the isolated mitochondria cox10-/- vs cox10f/f.

6) They were asked to perform GO term enrichment analysis for the cox10-/- vs cox10f/f WCL dataset. The authors performed the GO term enrichment analyses separately for mitochondrial and other than mitochondrial proteins, however, both mitochondrial and non-mitochondrial proteins may be associated with certain GO terms. The authors should add an additional term enrichment analysis for all the proteins upregulated or downregulated in the cox10-/- vs cox10f/f WCL dataset.

7) The authors were asked to provide a correlation between the observed changes in the proteome and metabolome in cox10-/- livers. Specifically, they were asked to examine whether changes in the levels of specific metabolites result from the up- or down-regulation of proteins of the respective pathways.

In response, the authors added Figure 5 —figure supplement 4 and included a short description in the manuscript. However, no specific examples were presented, e.g. up- or down-regulated level of enzyme A possibly contributed to up- or down-regulated level of metabolite X. Rather than (or in addition to) Figure 5 —figure supplement 4, another figure or a table linking specific enzymatic activities with metabolites should be included.

https://doi.org/10.7554/eLife.80919.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1 (Recommendations for the authors):

In this work, Lesner and colleagues study the effects of mitochondrial electron transport chain dysfunction at the level of mitochondrial complex I in comparison to complex IV in mouse liver. They report no adverse effects on liver functions in a complex I-deficient mouse model, whereas complex IV deficiency results in decreased liver function. To provide further insight into the mechanisms underlying these different phenotypic outcomes, the authors collect experimental data for alterations in the redox status as well as the proteome and metabolome in complex I- and complex IV-deficient mouse liver. These data generally support the notion that complex I activity is dispensable for liver function. The authors explain this finding by providing evidence for the involvement of mitochondrial enzyme dihydroorotate dehydrogenase as an alternative electron donor to maintain respiratory chain activity. These data add to the field of mitochondrial diseases associated with the dysfunction of mitochondrial respiratory chain complexes and tissue-specific phenotypes.

The authors generally present their data in a clear and logical way, which makes the story easy to follow. However, the description of proteomics experiments including LC-MS analysis in Materials and methods is largely incomplete and needs to be fully revised to provide detailed information about sample preparation, protein identification and quantification. In addition, Supplemental Tables for proteomic (and metabolomics) data are insufficient lacking information about gene names, identified peptides per replicate, ratios, calculated p-values etc. Proteomics data should also be deposited to a public database (e.g. PRIDE).

Proteomics data is now deposited into PRIDE (identified PXD031716; reviewer account details provided in manuscript). Supplemental Tables have been updated to include the requested information (gene names, peptides per replicate, fold change and p-values). The Methods section has been updated and includes details on sample preparation, details on LCMS instrumentation and peptide identification/quantitation.

The authors performed label-free proteomics to investigate changes in the proteome of ndufa9-/- and cox10-/- mice vs. control mice. It should be noted that only less than 1/3 of the proteome was covered and thus they likely missed also some further regulated proteins in their analysis. Since they focus on mitochondrial processes, it is suggested to also analyze mitochondria-enriched fractions. Furthermore, the authors should improve their data presentation. For example, it would be informative to prepare additional plots showing the distribution of individual OXPHOS complexes in ndufa9-/- and cox10-/- liver samples. Do they generally shift to more positive or negative log2 values? In addition, the analysis of OXPHOS complexes by blue native PAGE is recommended to reveal changes in supercomplex formation or assembly states.

We thank the reviewer for these suggestions, and have accordingly enhanced our analysis and discussion of the proteomics data. We agree that our analysis is not comprehensive of the complete proteome, and it possible some regulated proteins were missed. We have updated the text to discuss this caveat (lines 174-178). Our goal was not to complete characterize the proteome changes in response to complex I or IV loss, but to examine if the genetically modified livers exhibited signs of adaptation to loss of each mitochondrial ETC component.

As suggested, we performed proteomics on enriched mitochondrial samples in ndufa9/- livers to potentially capture additional regulatory proteins; this analysis is presented in figure 2 —figure supplement 1, and Table S1. We identified and quantified 738 mitochondrial proteins, which represents ~64% of the mitochondrial proteome. Similar to our whole cell data, we do not observe a large number of changes, complex I components are downregulated, while other complexes are largely maintained (Figure 2 —figure supplement 1, Table S1).

As suggested, additional plots focusing on individual components of each OXPHOS complex are now provided in Figures 2H, Figure 2 —figure supplement 1B, and Figure 5H. In ndufa9-/- livers, we observe isolated reduction of complex I components, with sparing of protein components from other complexes. In cox10-/- livers, we observe isolated reduction of complex IV components, with a general increase in components from other complexes. Thus, it appears that each individual knockout allele specifically affects the abundance of components in the respective complex, without major declines in other complex components.

Thank you for the suggestion regarding BN-PAGE analysis. The relevance of supercomplex formation in C57BL6 mice is complicated: a mutation in the SCAF1 gene in this genetic background results in reduced assembly of multimeric cIV-containing supercomplexes (PMID 26928661, 23812712). As a result, it has been proposed that there are fewer supercomplexes and/or cIV-containing supercomplexes in this background. We’ve done some preliminary BN-PAGE analysis (Author response image 1) which is consistent with this interpretation: Loss of cox10 or ndufa9 results in loss of the individual respective complexes, loss of the I+III2+IV1 supercomplex; while cox10-/- livers retain a I+III2+IIn supercomplex.

Author response image 1
BN-PAGE analysis to analyze ETC complexes and supercomplexes from murine liver lysate.

(A) BN-PAGE, followed by Coomassie staining, of isolated mitochondria extracted with 5g/g digitonin/protein ratio. 50mg of protein per lane were assessed by BN-PAGE, and bands are annotated based on previously published data (PMID 26928661). (B) Same as (A), except bands were visualized by in-gel complex I activity, which exhibits a purplish color. Bands with complex I activity are indicated. (C) Same as (A), excepts bands were visualized by in-gel complex IV activity, which exhibits a brown color. Bands with complex IV activity are indicated.

The proposed lack of multimeric cIV-containing supercomplexes in this genetic background is perhaps related to our central finding that cI is not required under homeostatic conditions. Specifically, by not participating in supercomplexes, cIV may be “freed” and therefore available to accept electrons from alternative donors (e.g., succinate, fatty acids/ETF, dihydroorotate, etc.). A detailed analysis of cIV-containing supercomplexes in this genetic background and others, combined with an analysis of the requirement for hepatic ndufa9 in the setting of modulating SCAF1 status would be required to fully address this question. We propose that this be addressed in future work. We have amended our discussion to address this caveat (lines 456-467).

To provide further insight into compensatory effects in cox10-/- livers, GO term (BP, CC, MP) overrepresentation analysis of both downregulated and upregulated proteins should be performed. More specifically, the authors should provide information about the down-regulated proteins other than only complex IV components and upregulated proteins other than mitochondrial components. They should analyze the proteomic data in more detail, including the annotation of upregulated non-mitochondrial and mitochondrial proteins in Figures 2E and 4E.

We have updated the manuscript to included additional analysis and discussion of our proteomic data according to these suggestions. All significant non-mitochondrial and mitochondrial proteins are annotated in Figure 2G; and we have annotated a number of nonmitochondrial and mitochondrial proteins in Figure 5G. For compensatory effects in cox10-/- livers, we have performed GO analysis as suggested, and the results are presented in Figure 5 —figure supplement 1,2, as well as Table S2. The results of these analyses are now discussed in the main text (lines 289-301, 307-313).

Furthermore, how do the authors explain the more than 2-fold increase in mitochondrial content in cox10-/- livers? It might be of interest to acquire EM or immunofluorescence images on these mouse liver samples.

Increases in mitochondrial content are common in both mouse models and patients with mitochondrial dysfunction (eg., PMID 17951359), and has been presumed to represent increased organellar biogenesis in response to the organellar defect. EM in cox10-deficient livers has been previously reported (PMID 17951359), which reports an accumulation of lipid droplets and mitochondria without apparent changes in organelle morphology. We have included immunofluorescence analysis by staining for the mitochondrial protein Tomm20 (Figure 2 —figure supplement 2; Figure 5 —figure supplement 5) which did not reveal significant alterations in ndufa9-depeleted livers, but increased mitochondrial content in cox10-depleted livers.

Based on GO term enrichment analysis, the authors state that several metabolic pathways were enriched in cox10-/- livers. However, what is missing is a correlation between the observed changes in the proteome and metabolome. Do changes in the levels of specific metabolites result from the up- or down-regulation of proteins of the respective pathways?

Thank you for this suggestion; we have accordingly updated the manuscript with additional gene set enrichment analysis and discussion (Figure 5 —figure supplement 4). Specifically, we do observe proteomic changes in several genes relating to fatty acid oxidation, bile acid metabolism and electron transfer activity, although overall enrichment/depletion in these pathways does not rise to statistical significance.

In Figure 5B, quantification of Western blots should be performed and further replicates might likely be needed to make conclusions whether the levels of DHODH do not change in ndufa9-/- mice. In the reviewer's point of view, DHODH levels are slightly increased in liver from ndufa9-/- mice.

During revision, we have eliminated this figure from the manuscript; however the requested analysis of DHODH levels is presented in Author response image 2. We do observe increases in DHODH abundance by western blot and proteomics in response to loss of complex I or complex IV, suggesting that this pathway may be compensating for loss of ETC function. However, as discussed in the current manuscript, we do not observe significant alterations in DHODH metabolite levels, suggesting that DHODH activity is not a major electron source for the hepatic ETC.

Author response image 2
Quantitation of DHODH abundance in ETC-deficient livers.

(A) DHODH abundance was assessed by western blot in ndufa9f/f and ndufa9-/- livers, and was quantitated relative to b2microglobulin (as a loading control). (B) Relative DHODH abundance based on proteomic analysis from ndufa9 and cox10 deficient livers.

Reviewer #2 (Recommendations for the authors):

Strengths of the manuscript.

The fundamental data pillars supporting the authors top-line conclusion are solid – i.e. that liver damage is only apparent in complex IV, but not complex I, knockout adult mice.

After complex IV knockout the circulating clinical-biomarkers of liver injury (AST/ALT/Bilirubin) are elevated with statistical significance. In agreement, the complex IV knockout livers exhibits marked damage in the pathology assessment. Furthermore, the authors confirm robust changes in the liver proteome and metabolome following complex IV knockout; these effects are commensurate with mitochondrial dysfunction and adaptive stress signalling. Effects of complex I deletion on liver homeostasis are either minor or not detectable within the parameters assessed here; the difference between the effect of deleting complex IV versus complex I is clear and justified by data presented in the manuscript.

In the context of mitochondrial biology there is general lack of conditional knockout studies in adult mice. Accordingly, this study offers novel and unexpected insights into the relationship between respiratory chain components and hepatic function – the observation that adult mouse liver exhibit greater dependence on complex IV is of notable conceptual advance.

Weakness of the manuscript.

The key weakness of the manuscript is the lack of convincing mechanistic rationale for the differential effects of complex IV and complex I knockout on adult mouse liver.

We agree that our manuscript does not completely reconcile the differential effects between complex I and complex IV, and have expanded our discussion to comment on this point. Complex I serves as a single potential entry point into the ETC, while complex IV is expected to accept electrons from all potential electron donors. Thus, our findings are most consistent with a model in which cI is not the major source of electrons in the hepatic ETC. As suggested below, we’ve simplified our analysis and discussion to reflect this, and removed data reflecting DHODH activity. At this point, our data suggests the hypothesis that ETFDH is a significant electron donor to the hepatic ETC. This is based on the buildup of glutarate and acyl-carnitines, as well as the similarities in phenotypes between animals/patients with cIV defects and animals/patients with ETFDH defects.

The authors show that complex I knockout ~fully suppresses respiration (OCR) in freshly isolated hepatocytes (Figure 1C). Because the NADH/NAD+ ratio is unaffected in complex I-knockout livers the authors infer that another enzyme in the respiratory chain (dihydroorotate dehydrogenase (DHODH)) is catalysing NADH oxidation (to stop NADH accumulation/build-up) – however NADH oxidation by DHODH drives respiration (as depicted in Figure 5A) but Figure 1C shows very clearly that respiration rates are negligible in hepatocytes devoid of complex I activity/expression. Therefore, the authors explicitly show in their own data that respiratory activity is not maintained by DHODH in the absence of complex I.

We apologize for the confusion. In our initial manuscript, the oxygen consumption measurements from in vitro cultured hepatocytes was meant as a corollary for measuring complex I function. We have now replaced this data with direct measurements of complex I activity in liver lysates (Figure 1C). The major conclusion is that removal of ndufa9 does remove complex I activity in vivo.

More generally, when hepatocytes (or most cell lines) are cultured in high glucose media (as in our previous Figure 1C), the observed respiration is primarily driven by complex I. Whether this represents the “in vivo” scenario is the major investigative point of this study. To avoid further confusion, we have removed all primary hepatocyte data from this manuscript and focuse on in vivo investigations.

Assessing hepatic respiratory function in vivo would allow us to assess the relevance of complex I or complex IV to oxygen consumption – however, O2 flux measurements in vivo are not developed. To address this, we have added new experiments and data in which we perform in vivo MR imaging in the livers of mice in order to assess oxygen status. Ndufa9-/- livers are indistinguishable from wild-type livers in these experiments (Figure 3E-K), indicating no significant deficit in oxygen consumption when complex I is removed. In contrast, we observe elevated oxygen levels in complex IV-deficient livers (Figure 6A-F), indicating that oxygen consumption is impaired in cox10-/- livers. Together, these data indicate that the mitochondrial ETC is required to maintaining oxygen levels in vivo; however, complex I is not required, suggesting that an alternative electron donor is feeding the mitochondrial ETC.

Furthermore the authors do not present respiration/OCR data in hepatocytes isolated from complex IV knockout livers; there we cannot assess if hepatic respiration is differentially affected by complex I versus complex IV knockout- if the respiration rates are equivalently affected this would show that liver damage following complex IV deletion are not ascribable to low respiration rates/activity.

See comments above for more detail. Briefly, we have removed cultured hepatocyte data to remove confusion between the in vitro and in vivo situation. We now provide an assessment of oxygen status in cox10-/- livers (Figure 6A-F), to assess the role of complex IV in oxygen consumption in vivo.

Next, the authors attempt to present data to support their model that DHODH is central to maintaining liver homeostasis in livers devoid of complex I, and treat mice with a small-molecule-inhibitor BRQ (an inhibitor of DHODH). The authors observe changes in the hepatic NADH/NAD+ ratio and interpret this data to support the DHODH model. However, the data is inadequate to support the authors conclusion;

1) There is already scant respiration in the complex I knockout hepatocytes (DHODH is not compensating for complex I deletion); 2) the authors do not show if BRQ effects respiration in control hepatocytes nor the remnant respiration in hepatocytes from complex I or complex IV livers; 3) whilst BRQ suppresses liver NADH /NAD+ its' impossible to ascribe this pharmacological effect to DHODH-specific inhibition ( it could be an alternative enzyme/target). The authors are not in a position to determine specificity of BRQ to DHODH and to make credible robust mechanistic interpretation; complementary genetic approaches would be required (i.e. showing DHODH knockout in the complex I knockout mouse liver recapitulates the complex IV knockout effect); 4) the authors present no in vivo evidence that BRQ treatment recapitulates the complex IV knockout phenotype – neither circulating markers (ALT/AST/bilirubin) nor pathology are presented.

Taken together the current model ascribing the lack of liver damage following complex I deletion to DHODH is not justified with existing data.

We completely agree with these comments, and as suggested below, we have revised the manuscript to remove the emphasis on DHODH.

Recommendations:

1) For reasons outlined above the model ascribing the lack of liver damage following complex I deletion to DHODH is not justified with existing data. My recommendation is the data should be withdrawn from the manuscript – otherwise many months of experiments are required to support the possible role of DHODH. However, even without mechanistic insight into why complex IV deletion, but not complex I, exerts liver damage the finding in-itself is of sufficient novelty to justify publication.

We agree, and have removed this data from the manuscript. We now provide a simplified interpretation based on analysis of substrates which accumulate in response to complex IV depletion. Briefly, we specifically observe accumulation of ETFDH-dependent substrates, leading us to hypothesize that ETFDH is a significant electron donor to hepatic ETC under homeostatic conditions.

2) Respiration/OCR data from the complex IV knockout hepatocytes is critical to aid data interpretation and provide mechanistic insight – it must be included and contrasted to the OCR in complex I knockout hepatocytes.

See comments above; we have removed isolated hepatocyte data to avoid confusion between in vitro and in vivo scenarios. We agree that assessment of respiration in cIV-deficient livers is critical, and provide an in vivo assessment in Figure 6A-F.

3) Whilst complex IV clearly exerts liver damage (~4x elevation in ALT and AST and histological staining) the level of damage (4-8 weeks after the knockout) is not overly severe and the animals do not undergo acute liver failure. The text and discussion need to better reflect this. Isn't the story here that mouse liver is not absolutely dependent on respiration to maintain function – otherwise we would expect rapid liver failure soon after the complex IV deletion?

Thank you for this comment. We agree; it is surprising that these mice are not dependent on either complex I or complex IV for survival. We have amended our text to discuss this point (lines 429-431).

4) Do the authors have data on hepatic ATP levels in absence of complex I and complex IV – both from liver tissue as well as the isolated hepatocytes? Are hepatic ATP levels maintained in both knockout models or is glycolysis maintaining ATP to ~homeostatic levels – the authors have shown respiration is essentially abolished upon complex I deletion, so how is ATP production supported? What is the ECAR in the complex I and complex IV knockout hepatocytes (the authors must have the ECAR data for complex I knockout because the OCR data is presented in figure 1). More generally the text doesn't address or discuss the concept of glycolytic adaptive capability in the absence of respiration.

We have examined hepatic ATP levels, and find no significant differences in complex I or complex IV livers (provided in Figure 2A and Figure 5A), suggesting that the mitochondrial ETC is not absolutely required to maintain steady state energy levels. As discussed above, we have removed experiments from cultured hepatocytes to avoid confounding in vitro and in vivo data.

To assess changes in glycolytic capacity, we have performed [U-13C]glucose isotope tracing in conscious mice, which reveals that removal of neither ndufa9 nor cox10 significantly impact glucose handling in the liver (Figure 3A-D; Figure 3 —figure supplement 1; Figure 6G-I; Figure 6 —figure supplement 1). In addition, we do not observe upregulation of the glycolytic pathway in response to ndufa9 or cox10 removal based on gene ontology analysis.

5) In places the text is too assertive and not reflective of the data – more nuance and precision is required as to not over-interpret data, for example;

Page11 – 'The phenotype of cox10-/- livers noted above indicates that the mitochondrial ETC is required for liver homeostasis'.

How is this justified when the authors have already shown in figure 1 that the complex I knockout hepatocytes exhibit little respiration but liver homeostasis is not measurably effected? Furthermore, the authors do not show the respiration/OCR data in complex IV knockout hepatocytes.

See comments above for clarification with respect to hepatocyte data, and we have adjusted the text accordingly to point out caveats in our interpretation. Specifically, we do not observe that complex I is required for oxygen homeostasis in vivo, although it is clearly required under in vitro culturing conditions.

'Together, these results indicate that complex I is largely dispensable for homeostatic function in the adult mouse liver, without significant metabolic or proteomic compensation'

Disagree. Respiration is dramatically reduced in complex I knockout hepatocytes – if no major adaption is ongoing how are hepatic ATP levels supported? There is presumably significant adaption to glycolysis and the PPP pathway to preserve ATP levels?

We apologize, as our previous in vitro data was clearly inconsistent with our in vivo interpretation. We agree that, in vitro, cI is clearly required for respiration.

In contrast to the in vitro scenario, our new data suggests that complex I is dispensable for oxygen homeostasis in vivo, suggesting that it is not significantly contributing to mitochondrial respiration. Based on our proteomic, metabolomic and isotope tracing analysis, we do not observe evidence of compensation, suggesting that complex I is dispensable in the adult murine liver. We have amended the text to discuss the caveats to our interpretation (lines 456-464).

'Together, these results indicate that complex I is largely dispensable for homeostatic function in the adult mouse liver, without significant metabolic or proteomic compensation'

The authors have assessed liver health 4 and 8 weeks after complex I (and complex 4) deletion. The text needs to explicitly reflect these two sampling time points. Perhaps complex I knockout would affect hepatic function over a longer period?

We have modified the text and figure legends to clarify the sampling dates for these experiments. We now also present data from animals at 8 weeks post complex I removal (Figure 1 —figure supplement 3), which indicates no detectable deficits in hepatic function.

Is 4 weeks knockout the shortest time point the authors have assessed? Maybe there is a transient stress/death response to complex I deletion (liver stress/death (ALT/AST rises)) but the liver is able to adapt to complex I loss and restore homeostasis before the initial 4 week sampling.

Thank you for the suggestion; this is an interesting idea. We now present data from animals at an earlier timepoint (2 weeks after complex removal), which indicates no significant deficits in hepatic function (Figure 1 —figure supplement 2).

Therefore general statements regarding dispensability of complex I in liver need much greater refinement and nuance.

6) Complex I knockout livers (Supp figure 1D_ – 'Histological analysis of liver sections revealed no significant pathology, including no evidence of steatosis (Oil Red O staining) or glycogen loss (PAS staining)'

They authors say 'significant', would they agree that are some changes in liver pathology (HandE and the Oil red O staining in particular) albeit they are far less severe compared to the cx4 knockout? If they agree the text needs to better reflect this.

We provide improved quality images in our revised manuscript and blinded analysis has not revealed detectable differences in ndufa9-/- livers. We have amended the discussion to include these caveats, and agree that it is possible there are minor defects preset in complex I-knockout livers (lines 421-422).

7) Metabolomics for complex I knockout in figure 2 – is this data from 4 or 8 weeks? If its 4 weeks can the authors also show the 8 week data?

The metabolomics and proteomics data for ndufa9-/- livers was performed at 4 weeks post complex I removal, and we have amended the figure legends to indicate the sampling timepoints. Unfortunately, we do not have a similar datasets at 8 weeks post-AAV.

8) It will be helpful to label the charts for metabolites as' liver' or 'circulating' to make data interpretation clearer for reader.

We have included the appropriate labels for clarity in the figure panels and supplementary tables.

Reviewer #3 (Recommendations for the authors):

Lesner and colleagues address the tissue specific phenotype seen in mitochondrial complex I deficiency, the most common cause of paediatric mitochondrial disease. Understanding this problem may lead to more personalised treatment strategies for patients with the disease. The authors leverage mouse models featuring liver specific knockout of complex I subunit Ndufa9 and complex IV assembly factor COX10, previously shown to result in loss of complex I and IV activity respectively. Using a range of classical biochemistry, pathology, proteomics and metabolomic approaches, the authors show that while loss of complex IV leads to severe impacts on cellular redox systems and metabolic compensation as expected, the most telling readout being NAD+/NADH ratio, complex I is not required to maintain metabolically normal mitochondria and livers from these animals are phenotypically similar to controls. Based on accumulation of metabolites in the pyrimidine synthesis pathway in complex IV deficient animals, the authors build a hypothesis that liver mitochondria preferentially utilise DHODH as a primary source of electrons for mitochondrial respiration. DHODH is a key enzyme in the pyrimidine synthesis pathway and an alternate source of electrons for the respiratory chain, effectively bypassing complex I by donating electrons directly to coenzyme Q which are then utilised by complex III (and subsequently passed to complex IV). Using the DHODH inhibitor brequinar the authors are able to generate a severe metabolic defect (the primary readout again NAD+/NADH ratio) in complex I deficient liver, but not controls. The main conclusion is that under normal conditions complex I is not the main entry point of electrons into the mitochondrial respiratory chain in liver, but rather pyrimidine synthesis, possibly supplemented by other sources. It is important to note that the main conclusion of preferential utilisation of pyrimidine synthesis derived electrons is drawn from a single experiment measuring the NAD+/NADH ratio. Additional metabolomic tracer studies using glutamine (and possibly other carbon sources aimed at delineating the other electron sources hinted at by the authors in the discussion) could be used to confirm if there is increased flux through pyrimidine synthesis.

The manuscript is well written, succinct and experiments are all of the highest quality. I have no major concerns in respect to the first conclusion that complex I deficient livers are phenotypically normal, and the resulting suggestion that complex I is not the major source of electrons for the ETC in liver. My concern is the conclusion that the main source is DHODH. This is drawn from a single (but apparently well controlled, though I'm not an expert on liver physiology) experiment using brequinar, with the NAD+/NADH ratio being the key readout. Increased flux doesn't always equate to changes in enzyme protein level or metabolite pool size, thus it not surprising that DHODH protein and metabolites in the pyrimidine synthesis pathway are not increased in abundance in ndufa9-/- livers. Indeed, except for dihydroorotate and orotate even in the extreme of BRQ treated animals you don't see accumulation of pyrimidine pathway metabolites. The authors should consider ways to measure flux directly, for example tracer studies using 13C-glutamine (or possibly H13CO3) could be used on primary hepatocytes from both knockouts. Inclusion of 13C-glucose would also go some way to addressing the possible increased flux through other pathways, e.g. complex II/TCA suggested by the authors in the discussion.

We thank the reviewers for this comment and agree that our experiments do not indicate DHODH is the major electron source. Based on comments from all 3 reviewers, we have removed this data, and clarified our interpretation. In particular, we hypothesize that fatty acids and ETFDH may be a major electron source, via contributions from fatty acids and amino acids. This is because in complex IV-deficient livers, we do not observe a buildup of succinate or dihydo-orotate or glycerol-3-phosphate, and we do not observe changes in glucose handling based on [U-13C]glucose tracing (Figure 6G-K). However, we do observe a significant increase in long chain acyl-carnitine species, suggesting a significant defect in fatty acid oxidation. Consistent with this interpretation, we observe decreased α-hydroxybutyrate levels and significant increases in glutarate levels. Interestingly, mice and humans suffering from ETF deficiency exhibit similar metabolic, histological and functional deficits in the liver. Thus, we propose that ETF serves as a major electron source to the mitochondrial ETC, which alleviates a dependence on complex I. Future experiments to trace flux through ETF will be necessary to examine this model in depth, and we have amended our discussion to include this point (lines 437-443).

As discussed above, we have removed experiments with cultured hepatocytes to avoid confusion between the in vivo and in vitro scenarios. Briefly, we find that culturing hepatocytes in high glucose results in their dependence on complex I for respiratory function. To avoid this potential discrepancy between experimental setups, we have added data assessing oxygen consumption by MR imaging (Figure 3E-K, Figure 6A-F) in vivo, as well as in vivo [U-13C]glucose tracing (Figure 3B-D, Figure 6G-I).

[Editors’ note: what follows is the authors’ response to the second round of review.]

Essential revisions:

1) In the revised manuscript the authors hypothesize that ETFDH is the major electron donor in the liver. This is mainly based on the elevation of related proteins and metabolites identified in proteomics and metabolomics data. The evidence for ETFDH is thus still circumstantial so conclusions should be cautious. Based on the tone of the authors in the main text I think they agree, however some of the statements could be softened in the abstract and discussion.

We agree that the role of ETFDH as the major electron donor is still circumstantial, as it is difficult to directly determine the flux of electrons into the ETC in an in vivo context. As suggested, we have modified the text in the abstract and discussion, please see lines 45-46 (sentence removed) and lines 525-526 (sentence removed). In the discussion, we have stated that our steady-state measurements do not directly inform on flux through fatty acid oxidation / ETFDH pathways, and future experiments will be necessary (lines 512-514).

2) The authors should carefully describe the underpinnings of some of the less common techniques being used within the text narrative. Eg., this is the first time I've seen MR imaging and I felt I was just thrown into the deep end of what it measures. Likewise, the authors don't provide non-experts with sufficient background to understand tracer metabolomics. I also didn't find the schematic in Figure 3A particularly helpful as the use of dots to describe isotopologues doesn't really help the reader interpret the associated bar plots. This could probably be solved by a more detailed figure legend.

As suggested, we have modified the text to include an additional description of the MR imaging experiments, geared towards non-experts; please see lines 229-249. In addition, we have updated our description of the tracer experiments to provide additional clarity for non-experts (lines 207-215).

We apologize for the confusion with respect to figure 3A. In our original figure, the colored (red) dots referred to the fate of heavy label atoms from [U-13C]glucose labeling. However, considering the significant amount of scrambling of labels that occurs in vivo, we chose to plot the total labeled fraction in Figure 3D, which sums the contributions from all individual isotopologues with at least one labeled carbon. We have modified the figure 3A to remove indicators of heavy labeled carbons, and it now simply summarizes some of the potential fates of glucose and lactate as they enter a hepatocyte.

3) Additional points concerning metabolomics:

– Tracer metabolomics data should be included in the supplement.

– All metabolomics data should be made available in a public repository (e.g. Metabolomics workbench).

We have updated the Supplementary files 1 and 2 to include tracer metabolomics data.

All metabolomics data has been deposited into Metabolomics workbench; accession codes are provided within the manuscript.

4) Data has been deposited in PRIDE as requested, however, the methods section in the manuscript requires some further work.

a) The authors wrote "50 ug per sample of protein homogenate from livers or enriched mitochondria (prepared as described above)", yet is it not clear which sample preparation steps described above they refer to. The authors should clearly state the steps they performed before they solubilized proteins in SDS/Tris.

We apologize for the omission, and have now included a description of sample preparation (lines 9971015).

b) The descriptions in the methods section still lack basic information.

Please add the following parameters to the LC-MS/MS data acquisition section:

AGC target for MS1 and MS2

Max IT for MS1 and MS2

Resolution for MS2

Isolation window for MS2

Intensity threshold

We have now included these details in the methods section (lines 1025-1032). A resolution is not specified for MS2 as MS2 scans were obtained in the ion trap of the mass spectrometer where mass resolution is not set in order to allow faster scans and identification of more proteins.

c) The authors wrote, "Fragment and precursor tolerances of 10 ppm and 0.6 Da were specified, and three missed cleavages were allowed". Please name the enzyme and state the cleavage rules for the search.

Added (lines 1035-1036).

d) The authors wrote: "The false-discovery rate (FDR) cut-off was 1% for all peptides". Please provide the FDR cut-off for proteins as well. Please add information on whether PSMs were validated with Percolator (or another script) applied in the PD workflow.

Added (lines 1018-1022).

e) There is clearly a mistake in the following sentence "Significantly up and downregulated proteins were identified based on log2(fold change) <1 or <1 (respectively) and p-value<0.05". Please correct it.

Corrected (lines 1042-1044).

f) The authors wrote, "Protein abundance was quantitated based on the total ion count for all identified peptides in each sample, and normalized after log transform according to previously published protocols (60)". The authors should state in the methods section how the normalization was performed (rather than just referring to a paper).

Added (lines 1044-1053).

g) The tables containing proteomics data look better, my additional suggestion would be to:

1) Add a unique peptide count per replicate (this parameter is more informative than peptide count).

Added for all datasets (Supplementary Files 1, 2).

2) Add a score parameter for protein identification.

We have added Sum PEP and Sequest HT scores for all datasets (Supplementary Files 1, 2).

5) Since the proteome coverage in the WCL samples was rather low (2.6-3k proteins, of which only 396-407 are annotated as mitochondrial in MitoCarta3.0), the authors were asked to analyze proteins in isolated mitochondrial fractions. The authors followed the recommendation only for ndufa9-/-, however, they did not do it for cox10-/-.

The additional analysis they performed yielded quantification data for 738 mitochondrial proteins in isolated mitochondria compared to 396 mitochondrial proteins in WCL samples, which clearly shows an improvement. As the proteomic analysis of isolated mitochondria ndufa9-/- vs ndufa9f/f in fact nearly doubled the number of quantified mitochondrial proteins reported, the authors should also include a similar analysis for the isolated mitochondria cox10-/- vs cox10f/f.

Thank you for this suggestion. We have compiled an analogous dataset comparing proteomes from enriched mitochondrial fractions in cox10-/- and cox10f/f livers, which is presented in Supplementary File 2 and is being deposited in Pride. We detected 721 mitochondrial proteins, which was comparable to the enrichment observed in our ndufa9-/- dataset.

We have created an additional supplementary figure (Figure 5 —figure supplement 3) to show this data, and discuss within the text (lines 351-361). We note that several components of complex IV are downregulated in cox10-/- livers, consistent with destabilization of complex IV. Components of other ETC components are spared or slightly unregulated. We note that there are a number of other changes in the mitochondrial proteome of cox10-/- livers, including downregulation of some fatty acid related proteins, and upregulation of proteins related to ubiquinol homeostasis.

6) They were asked to perform GO term enrichment analysis for the cox10-/- vs cox10f/f WCL dataset. The authors performed the GO term enrichment analyses separately for mitochondrial and other than mitochondrial proteins, however, both mitochondrial and non-mitochondrial proteins may be associated with certain GO terms. The authors should add an additional term enrichment analysis for all the proteins upregulated or downregulated in the cox10-/- vs cox10f/f WCL dataset.

We apologize for the confusion. In the previous version of the manuscript, we did perform GO term enrichment analysis for all up or down-regulated proteins, as well as non-mitochondrial proteins separately. We now refer to these analyses as “all” and “non-mitochondrial” for clarity.

We now provide additional analyses of mitochondrial proteins only in Figure 5 —figure supplements 1B and 2B; these new analyses are referred to as “mitochondrial”, and are also listed in Supplementary File 2.

7) The authors were asked to provide a correlation between the observed changes in the proteome and metabolome in cox10-/- livers. Specifically, they were asked to examine whether changes in the levels of specific metabolites result from the up- or down-regulation of proteins of the respective pathways.

In response, the authors added Figure 5 —figure supplement 4 and included a short description in the manuscript. However, no specific examples were presented, e.g. up- or down-regulated level of enzyme A possibly contributed to up- or down-regulated level of metabolite X. Rather than (or in addition to) Figure 5 —figure supplement 4, another figure or a table linking specific enzymatic activities with metabolites should be included.

As suggested, we have supplemented the previous analysis with a pathway specific analysis focusing on metabolic enzymes and the metabolites they produce/consume. These diagrams are provided now in Figure 5 —figure supplement 5,6,7. We have expanded our text to include a discussion of these results (lines 371-427).

To summarize briefly, we find that fatty acid synthesis and fatty acid oxidation enzymes are largely changing in the opposite directions as would be expected for an increase in acyl-carnitine species (Figure 5 —figure supplement 5), and we suspect that the buildup of acyl-carnitine species is instead due to a direct inhibition of fatty acid oxidation secondary to loss of complex IV. For bile acid species, the majority of direct bile-acid metabolizing enzymes were not detectable in our dataset (Figure 5 —figure supplement 6). For electron carriers, we did observe some interesting trends: While the precise reasons for decreases in electron carrier abundance is not clear at this stage, there were some instances were consuming enzymes were upregulated (e.g., for pyruvate, α-hydroxybutyrate), or synthesis / recycling enzymes were downregulated (e.g., for α-hydroxybutyrate, tetrahydrobiopterin).

https://doi.org/10.7554/eLife.80919.sa2

Article and author information

Author details

  1. Nicholas P Lesner

    Children's Medical Center Research Institute, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9734-8828
  2. Xun Wang

    Children's Medical Center Research Institute, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  3. Zhenkang Chen

    Children's Medical Center Research Institute, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7919-5546
  4. Anderson Frank

    Department of Biochemistry, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  5. Cameron J Menezes

    Children's Medical Center Research Institute, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5759-8099
  6. Sara House

    Children's Medical Center Research Institute, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  7. Spencer D Shelton

    Children's Medical Center Research Institute, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1236-5317
  8. Andrew Lemoff

    Department of Biochemistry, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Resources, Formal analysis, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4943-0170
  9. David G McFadden

    1. Department of Biochemistry, University of Texas Southwestern Medical Center, Dallas, United States
    2. Harold C. Simmons Comprehensive Cancer Center, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Resources
    Competing interests
    No competing interests declared
  10. Janaka Wansapura

    Advanced Imaging Research Center, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Resources, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  11. Ralph J DeBerardinis

    1. Children's Medical Center Research Institute, University of Texas Southwestern Medical Center, Dallas, United States
    2. Harold C. Simmons Comprehensive Cancer Center, University of Texas Southwestern Medical Center, Dallas, United States
    3. Department of Pediatrics, University of Texas Southwestern Medical Center, Dallas, United States
    4. Howard Hughes Medical Institute, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Resources, Writing – review and editing
    Competing interests
    is an advisor to Agios Pharmaceuticals
  12. Prashant Mishra

    1. Children's Medical Center Research Institute, University of Texas Southwestern Medical Center, Dallas, United States
    2. Harold C. Simmons Comprehensive Cancer Center, University of Texas Southwestern Medical Center, Dallas, United States
    3. Department of Pediatrics, University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Visualization, Methodology, Writing - original draft, Writing – review and editing
    For correspondence
    prashant.mishra@utsouthwestern.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2223-1742

Funding

United Mitochondrial Disease Foundation

  • Prashant Mishra

National Institutes of Health (1DP2ES030449-01)

  • Prashant Mishra

National Institutes of Health (1R01AR073217-01)

  • Prashant Mishra

National Institutes of Health (1F31-DK122676)

  • Nicholas P Lesner

Moody Medical Research Institute

  • Prashant Mishra

National Science Foundation (GRFP 2019281210)

  • Spencer D Shelton

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the Hao Zhu lab for assistance with liver experiments, the UT Southwestern Metabolic Phenotyping Core for plasma analysis, the UT Southwestern Proteomics Core, the CRI Metabolomics Facility, the Biochemistry Metabolomics Core, the UT Southwestern Small Animal Imaging Resource for help with various experiments and members of the Mishra and DeBerardinis laboratories for helpful discussions and suggestions during this project. Diagrams were created with BioRender.com. Funding: This work was supported by funding from the United Mitochondrial Disease Foundation (Research Grant to PM), the National Institutes of Health (1DP2ES030449-01 from NIEHS to PM, 1R01AR073217-01 from NIAMS to PM, 1F31-DK122676 from NIDDK to NPL), the Moody Medical Research Institute (Research Grant to P.M.), and the National Science Foundation (GRFP 2019281210 award to SDS).

Ethics

All mouse experiments were performed according to protocols approved by the Institutional Animal Care and Use Committee (IACUC) at University of Texas Southwestern Medical Center (protocol 102654).

Senior Editor

  1. Vivek Malhotra, The Barcelona Institute of Science and Technology, Spain

Reviewing Editor

  1. Agnieszka Chacinska, University of Warsaw, Poland

Publication history

  1. Preprint posted: July 15, 2021 (view preprint)
  2. Received: June 9, 2022
  3. Accepted: September 23, 2022
  4. Accepted Manuscript published: September 26, 2022 (version 1)
  5. Version of Record published: November 10, 2022 (version 2)

Copyright

© 2022, Lesner et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,543
    Page views
  • 545
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Nicholas P Lesner
  2. Xun Wang
  3. Zhenkang Chen
  4. Anderson Frank
  5. Cameron J Menezes
  6. Sara House
  7. Spencer D Shelton
  8. Andrew Lemoff
  9. David G McFadden
  10. Janaka Wansapura
  11. Ralph J DeBerardinis
  12. Prashant Mishra
(2022)
Differential requirements for mitochondrial electron transport chain components in the adult murine liver
eLife 11:e80919.
https://doi.org/10.7554/eLife.80919

Further reading

    1. Cell Biology
    2. Chromosomes and Gene Expression
    Zhanwu Zhu, Jingjing Liu ... Bo Cheng
    Research Article

    Dynamic regulation of transcription is crucial for the cellular responses to various environmental or developmental cues. Gdown1 is a ubiquitously expressed, RNA polymerase II (Pol II) interacting protein, essential for the embryonic development of metazoan. It tightly binds Pol II in vitro and competitively blocks the binding of TFIIF and possibly other transcriptional regulatory factors, yet its cellular functions and regulatory circuits remain unclear. Here, we show that human GDOWN1 strictly localizes in the cytoplasm of various types of somatic cells and exhibits a potent resistance to the imposed driving force for its nuclear localization. Combined with the genetic and microscope-based approaches, two types of the functionally coupled and evolutionally conserved localization regulatory motifs are identified, including the CRM1-dependent nucleus export signal (NES) and a novel Cytoplasmic Anchoring Signal (CAS) that mediates its retention outside of the nuclear pore complexes (NPC). Mutagenesis of CAS alleviates GDOWN1’s cytoplasmic retention, thus unlocks its nucleocytoplasmic shuttling properties, and the increased nuclear import and accumulation of GDOWN1 results in a drastic reduction of both Pol II and its associated global transcription levels. Importantly, the nuclear translocation of GDOWN1 occurs in response to the oxidative stresses, and the ablation of GDOWN1 significantly weakens the cellular tolerance. Collectively, our work uncovers the molecular basis of GDOWN1’s subcellular localization and a novel cellular strategy of modulating global transcription and stress-adaptation via controlling the nuclear translocation of GDOWN1.

    1. Cell Biology
    2. Neuroscience
    Arnau Llobet Rosell, Maria Paglione ... Lukas Jakob Neukomm
    Research Article

    Axon degeneration contributes to the disruption of neuronal circuit function in diseased and injured nervous systems. Severed axons degenerate following the activation of an evolutionarily conserved signaling pathway, which culminates in the activation of SARM1 in mammals to execute the pathological depletion of the metabolite NAD+. SARM1 NADase activity is activated by the NAD+ precursor nicotinamide mononucleotide (NMN). In mammals, keeping NMN levels low potently preserves axons after injury. However, it remains unclear whether NMN is also a key mediator of axon degeneration and dSarm activation in flies. Here, we demonstrate that lowering NMN levels in Drosophila through the expression of a newly generated prokaryotic NMN-Deamidase (NMN-D) preserves severed axons for months and keeps them circuit-integrated for weeks. NMN-D alters the NAD+ metabolic flux by lowering NMN, while NAD+ remains unchanged in vivo. Increased NMN synthesis, by the expression of mouse nicotinamide phosphoribosyltransferase (mNAMPT), leads to faster axon degeneration after injury. We also show that NMN-induced activation of dSarm mediates axon degeneration in vivo. Finally, NMN-D delays neurodegeneration caused by loss of the sole NMN-consuming and NAD+-synthesizing enzyme dNmnat. Our results reveal a critical role for NMN in neurodegeneration in the fly, which extends beyond axonal injury. The potent neuroprotection by reducing NMN levels is similar to the interference with other essential mediators of axon degeneration in Drosophila.