Hyperammonaemia‐induced skeletal muscle mitochondrial dysfunction results in cataplerosis and oxidative stress

Abstract

Key points

• Hyperammonaemia occurs in hepatic, cardiac and pulmonary diseases with increased muscle concentration of ammonia.

• We found that ammonia results in reduced skeletal muscle mitochondrial respiration, electron transport chain complex I dysfunction, as well as lower NAD⁺/NADH ratio and ATP content.

• During hyperammonaemia, leak of electrons from complex III results in oxidative modification of proteins and lipids.

• Tricarboxylic acid cycle intermediates are decreased during hyperammonaemia, and providing a cell‐permeable ester of αKG reversed the lower TCA cycle intermediate concentrations and increased ATP content.

• Our observations have high clinical relevance given the potential for novel approaches to reverse skeletal muscle ammonia toxicity by targeting the TCA cycle intermediates and mitochondrial ROS.

Abstract

Ammonia is a cytotoxic metabolite that is removed primarily by hepatic ureagenesis in humans. Hyperammonaemia occurs in advanced hepatic, cardiac and pulmonary disease, and in urea cycle enzyme deficiencies. Increased skeletal muscle ammonia uptake and metabolism are the major mechanism of non‐hepatic ammonia disposal. Non‐hepatic ammonia disposal occurs in the mitochondria via glutamate synthesis from α‐ketoglutarate resulting in cataplerosis. We show skeletal muscle mitochondrial dysfunction during hyperammonaemia in a comprehensive array of human, rodent and cellular models. ATP synthesis, oxygen consumption, generation of reactive oxygen species with oxidative stress, and tricarboxylic acid (TCA) cycle intermediates were quantified. ATP content was lower in the skeletal muscle from cirrhotic patients, hyperammonaemic portacaval anastomosis rat, and C2C12 myotubes compared to appropriate controls. Hyperammonaemia in C2C12 myotubes resulted in impaired intact cell respiration, reduced complex I/NADH oxidase activity and electron leak occurring at complex III of the electron transport chain. Consistently, lower NAD⁺/NADH ratio was observed during hyperammonaemia with reduced TCA cycle intermediates compared to controls. Generation of reactive oxygen species resulted in increased content of skeletal muscle carbonylated proteins and thiobarbituric acid reactive substances during hyperammonaemia. A cell‐permeable ester of α‐ketoglutarate reversed the low TCA cycle intermediates and ATP content in myotubes during hyperammonaemia. However, the mitochondrial antioxidant MitoTEMPO did not reverse the lower ATP content during hyperammonaemia. We provide for the first time evidence that skeletal muscle hyperammonaemia results in mitochondrial dysfunction and oxidative stress. Use of anaplerotic substrates to reverse ammonia‐induced mitochondrial dysfunction is a novel therapeutic approach.

Key points

• Hyperammonaemia occurs in hepatic, cardiac and pulmonary diseases with increased muscle concentration of ammonia.

• We found that ammonia results in reduced skeletal muscle mitochondrial respiration, electron transport chain complex I dysfunction, as well as lower NAD⁺/NADH ratio and ATP content.

• During hyperammonaemia, leak of electrons from complex III results in oxidative modification of proteins and lipids.

• Tricarboxylic acid cycle intermediates are decreased during hyperammonaemia, and providing a cell‐permeable ester of αKG reversed the lower TCA cycle intermediate concentrations and increased ATP content.

• Our observations have high clinical relevance given the potential for novel approaches to reverse skeletal muscle ammonia toxicity by targeting the TCA cycle intermediates and mitochondrial ROS.

Abbreviations

αKG

α‐ketoglutarate (2‐oxoglutarate)

CK

creatine kinase

DAPI

4′6‐diamidino‐2‐phenylindole

DCFDA

2′,7′‐dichlorodihydrofluorescein diacetate

DMAKG

dimethyl α‐ketoglutarate

ETC

electron transport chain

FCCP

carbonyl cyanide‐4‐(trifluoromethoxy) phenylhydrazone

MnSOD

manganese dependent superoxide dismutase

MRS

magnetic resonance spectroscopy

PCA

portacaval anastomosis

PCr

phosphocreatine

P_i

inorganic phosphate

ROS

reactive oxygen species

TBARS

thiobarbituric acid reactive substances

TCA

tricarboxylic acid

Introduction

Ammonia is generated during the catabolism of amino acids (Olde Damink et al. 2009), from gut microbial metabolism and following gastrointestinal bleeding (Rudman et al. 1973 b; Hawkins et al. 1994). The primary mode of ammonia disposal involves rapid hepatic uptake and conversion to urea in the hepatocytes (Rudman et al. 1973 a). However, when the liver is unable to cope with the increased ammonia load in disease (Rudman et al. 1973 a; Haussinger et al. 1992), the skeletal muscle becomes a major alternative organ for ammonia disposal (Ganda & Ruderman, 1976; Lockwood et al. 1979; Chatauret et al. 2006). Consistent with the impaired ammonia disposal in liver disease, patients with cirrhosis have elevated blood ammonia concentrations (Olde Damink et al. 2009). Interestingly, elevated blood ammonia concentrations have also been reported in advanced heart failure and chronic obstructive lung disease (Bessman & Evans, 1955; Calvert et al. 2008; Olde Damink et al. 2009), as well as in malignancies with actively proliferating cancer cells (Chance et al. 1989). Despite recognition of skeletal muscle ammonia uptake and disposal in disease, little is known about the role of hyperammonaemia‐induced mitochondrial dysfunction in skeletal muscle pathophysiology.

During physiological conditions, skeletal muscle ammonia is metabolized by glutamate dehydrogenase (GDH) catalysing the α‐ketoglutarate (αKG) to glutamate conversion in muscle mitochondria (Olde Damink et al. 2009). The low affinity of the bidirectional enzyme GDH for ammonia favours the anaplerotic conversion of glutamate to αKG (Masumi et al. 1999; Zaganas et al. 2013). However, during hyperammonaemia, the αKG–glutamate pathway is altered due to loss of αKG by reductive deamination (Hod et al. 1982) and consequent reduced availability of αKG. Ammonia also inhibits mitochondrial electron transport chain (ETC) components in synaptosomes in neural cells in vitro (Qureshi et al. 1998). Impaired ETC function results in altered NAD⁺/NADH ratio due to defective regeneration of NAD⁺ by oxidation of NADH with reduced ATP synthesis (Zu et al. 2003). Since skeletal muscle uptake of ammonia is increased in cirrhosis (Lockwood et al. 1979; Qiu et al. 2013), metabolic disposal of ammonia by the skeletal muscle in cirrhosis was hypothesized to result in decreased availability of αKG and simultaneously impaired mitochondrial function. Even though there is evidence that skeletal muscle electron transport chain components are decreased in the skeletal muscle of cirrhotic patients, the mechanistic relationship between hyperammonaemia and skeletal muscle mitochondrial bioenergetics in cirrhosis is not known.

Skeletal muscle is a recognized metabolic partner of the liver in ammonia disposal, and tissue specific adaptive responses occur during hyperammonaemia (Ganda & Ruderman, 1976; Chatauret et al. 2006; Frigerio et al. 2008; Qiu et al. 2013). However, it is not known if increased metabolic demand and cellular stress induced by elevated ammonia uptake by the skeletal muscle results in mitochondrial bioenergetic dysfunction. Using a comprehensive array of models, we show that hyperammonaemia results in impaired ATP synthesis, increased electron leak at complex III of the ETC and oxidative stress. Rescuing cataplerosis of αKG with cell‐permeable esters of αKG partially reversed the reduced ATP content in myotubes. These data demonstrate for the first time that ammonia impairs skeletal muscle mitochondrial function via cataplerosis of αKG.

Methods

Ethical approval

The animal studies performed were approved by the Institutional Animal Care and Use Committee at the Cleveland Clinic. Human studies were performed after approval by the Institutional Review Board, and written informed consent was obtained from all subjects. They conformed with the Declaration of Helsinki.

These studies were performed in vivo in patients with cirrhosis and matched controls, the hyperammonaemic portacaval anastomosis (PCA) rat model and in vitro in C2C12 myotubes exposed to ammonium acetate.

Human subjects

Rectus abdominis muscle was obtained from patients with cirrhosis (n = 6), donors for liver transplantation, and control subjects (n = 6) undergoing elective surgery without evidence of chronic diseases. The clinical details are provided in Table 1. All patients were on their usual diet and in their usual state of physical activity. None of the controls were participating in rigorous physical activity or a structured exercise programme. However, patients with cirrhosis are usually less active physically (Dunn et al. 2016), but objective measures of activity were not routinely quantified and therefore could not be compared.

Table 1.

Clinical characteristics of cirrhotic and control subjects

Characteristic	Control	Cirrhosis
Number	6	6
Gender (M:F)	4:2	5:1
Age (years)	62.7 ± 7.1	60.7 ± 6.3
Aetiology of liver disease
Alcohol		2
NASH		3
HCV		1
Serum bilirubin (mg dl–1)	1.0 ± 0.2	19.6 ± 10.1 ***
Serum alanine aminotransferase (IU dl–1)	21.5 ± 2.7	119.7 ± 22.9 ***
Serum aspartate aminotransferase (IU dl–1)	20.3 ± 2.6	148.2.8 ± 32.4 ***
Blood urea nitrogen (mg dl–1)	11.8 ± 2.9	33.3 ± 11.0 ***
Serum creatinine (mg dl–1)	0.8 ± 0.2	1.62 ± 0.4 **
International normalized ratio	1.02 ± 0.1	1.93 ± 0.2 ***
Albumin (g dl–1)	4.35 ± 0.2	2.6 ± 0.5 ***
Plasma ammonia (μmol l–1)	39.5 ± 4.9	229.5 ± 49.8 ***
Muscle ammonia (mm l–1)	0.37 ± 0.1	4.79 ± 0.6 ***

All values are the mean ± SD. HCV, hepatitis C virus infection; NASH, Non alcoholic steatohepatitis. ^** P < 0.01, ^*** P < 0.001.

Animals

Male Sprague‐Dawley rats (weights 250–260g) with an end to side PCA or sham surgery were obtained from Charles River Inc. (Wilmington, MA, USA). Sham operated rats were pair fed with the PCA rats and all animals were evalulated 4 weeks after surgery as described earlier (Dasarathy et al. 2004). ³¹P magnetic resonance spectroscopy (MRS) was performed in the right leg focusing on the gastrocnemius muscle. Animals were lightly sedated with i.p. pentobarbitone, and the right leg was strapped with a non‐magnetic tape. After 5 min of rest, readings were obtained. After these studies, animals were killed using i.p. pentobarbitone, blood was drawn rapidly from the aorta into EDTA‐coated vials, plasma was separated and blood ammonia was quantified using methods previously described (Dasarathy et al. 2011). The gastrocnemius muscle was rapidly harvested, blotted dry of blood, weighed and flash‐frozen in liquid nitrogen before being stored at −80°C for future analyses. The Institutional Animal Care and Use Committee at the Cleveland Clinic approved all experiments.

³¹P magnetic resonance spectroscopy

Animals were sedated as described above and positioned at isocentre within a Bruker Biospec 9.4 T magnetic resonance image scanner (Bruker Corp., Billerica, MA, USA). A dual‐tuned ¹H/³¹P surface coil (20 mm outer diameter) was positioned over the gastrocnemius muscle. Localizer images and manual shimming were performed on the ¹H channel to ensure high quality ³¹P‐MRS data. A voxel‐selective acquisition was used to acquire the ³¹P‐MRS data. Spectra were analysed to determine the areas of the phosphocreatine (PCr), inorganic phosphate (P_i) and the three ATP peaks assuming Gaussian line shape applying a least‐square fitting. The PCr and P_i peak areas were normalized against that of γ‐ATP assumed to represent 5.5 mm ATP (kg wet weight)^–1.

In vitro culture studies

All studies were performed in differentiated C2C12 myotubes as described earlier (Qiu et al. 2012). In brief, C2C12 myoblasts (ATCC, Manassas, VA, USA) were grown to confluence in proliferation medium (Dulbecco's modified Eagle's medium; DMEM) with 10% fetal bovine serum) that was replaced by differentiation medium (DMEM with 2% horse serum) for 48 h. Differentiation was established by increase in creatine kinase activity and fusion index (Fig. 1 A and B) and expression of myosin heavy chain (Fig. 1 C). C2C12 myotubes were treated with 10 mm ammonium acetate for 0–24 h using protocols previously reported (Qiu et al. 2012, 2013). We have previously demonstrated that this concentration of ammonium acetate does not affect extracellular or intracellular pH, and results in an intracellular concentration of ammonia similar to that observed in the skeletal muscle of cirrhotic patients (Qiu et al. 2013). Cell lysates and protein extraction were carried out using protocols standard in our laboratory (Qiu et al. 2012). All experiments were performed in triplicate.

Figure 1. Differentiation of C2C12 myotubes achieved by 48 h.

A, creatine kinase activity assay showed an increase by 24 h of differentiation, and after 48 h the activity was stable for the next 3 days. All experiments in 3 independent samples. ^** P < 0.01 and ^*** P < 0.001 compared to confluent myoblasts. B, fusion index of myotubes in differentiation medium showed that differentiation by 48 h remained stable subsequently. ^*** P < 0.001 compared to confluent myoblasts (CM). Time in hours is duration in differentiation medium. C, representative photomicrographs of time course of confluent myoblasts and myotubes incubated in differentiation medium for different time points stained with MF20 (myosin heavy chain) demonstrating differentiation by 48 h. [Colour figure can be viewed at wileyonlinelibrary.com]

Creatine kinase activity assay

Creatine kinase (CK) activity was quantified because it increases during differentiation and has been used to demonstrate myotube differentiation (Trendelenburg et al. 2009). C2C12 myoblasts at confluence and subsequently in differentiation medium at 24, 48, 72, 96 and 120 h were washed three times with phosphate‐buffered saline (PBS) and then lysed with radioimmunoprecipitation assay (RIPA) buffer and stored until measurement at −80°C. Enzyme activity was measured using the CK (IFCC‐NAC) reagent (Fisher Diagnostics, Middletown, VA, USA). The CK reagent was prepared according to the manufacturer's instructions. Lysates were adjusted to room temperature, CK reagent was added, and absorbance was immediately read at 340 nm for 20 min; the reading interval was 1 min. Standard curves for CK were generated using CK from rabbit muscle (Roche Diagnostics, Indianapolis, IN, USA). CK standards were freshly prepared for each assay. Protein content was determined using a BCA kit (Thermo Scientific, Rockford, IL, USA). There was a rapid increase in CK activity at 24 h of differentiation and at 48 h, after which the activity did not change significantly (Fig. 1 A). These data show that our studies were performed in differentiated myotubes.

Fusion index and myosin heavy chain expression

To visualize myotubes and nuclei, confluent myoblasts and myotubes in differentiation medium following 24, 48, 72 and 96 h were washed with PBS and fixed in 4% paraformaldehyde for 10 min. Cells were then incubated overnight with clone MF20 antibody (Developmental Studies Hybridoma Bank, Iowa City, IA, USA) at 1:200 dilution overnight and fluorescent tagged secondary antibody. Total cell nuclei and nuclei within myotubes were counted using the NIH ImageJ software (Schneider et al. 2012). A muscle cell containing three or more nuclei was considered a myotube. Average number of nuclei per myotube was determined by dividing the number of nuclei in myotubes by the total number of myotubes. The number of nuclei in confluent myoblasts in growth medium and myotubes in differentiation medium was counted. Fusion index was calculated as a percentage of the total number of nuclei (mononucleated and multinucleated cells) in cells containing at least three nuclei and was similar to that previously reported; after 48 h, fusion index stabilizes. For each experimental situation, at least 400 nuclei per dish were counted in eight independent cultures. There was evidence of rapid increase in the fusion index from myoblast to 48 h differentiated myotubes after which the increase was significantly lower (Fig. 1 B).

Reactive oxygen species

Total reactive oxygen species (ROS) was quantified using 2′,7′‐dichlorodihydrofluorescein diacetate (DCFDA), a cell‐permeable dye that diffuses into the cell. DCFDA is deacetylated by intracellular esterases to the non‐fluorescent compound 2′,7′‐dichlorodihydrofluorescein, which is then oxidized by ROS into the fluorescent 2′,7′‐dichlorofluorescein (DCF; excitation 490 nm; emission 520 nm). Flow cytometry was performed for ROS detection in differentiated C2C12 myotubes following incubation with 10 mm ammonium acetate for different time points and 10 μm CM‐H₂DCFDA (Invitrogen, Carlsbad, CA, USA) for 30 min at 37°C prior to trypsinization and analysis by flow cytometry (LSRII from BD Biosciences). Results were analysed with FlowJO software (FlowJo LLC, Ashland, OR, USA). Quantification of ROS was also performed in a 96‐well plate in control and ammonium acetate‐treated C2C12 myotubes. The ROS assay kit (Cell Biolabs, San Diego, CA, USA) was used as per the manufacturer's instructions. It is based on the same principle as that used for the staining of the cells with the conversion of the non‐fluorescent cell‐permeable dye, DCFDA, which is converted in the presence of ROS and other oxidant species into DCF, a highly fluorescent compound. A fluorescence spectrometer with maximum excitation and emission spectra of 495 and 529 nm, respectively, was used to quantify the fluorescence.

Mitochondrial ROS was quantified by incubation of C2C12 myotubes with ammonium acetate and MitoSOX (excitation 540 nm; emission 560 nm). Results were analysed with FlowJO software. ROS was quenched with the simultaneous addition of 20 nm MitoTEMPO for mitochondrial ROS. Both number of cells stained and mean fluorescence intensity were quantified.

Hydrogen peroxide production

The rate of hydrogen peroxide (H₂O₂) production was quantified using the oxidation of the fluorogenic indicator amplex red in the presence of horseradish peroxidase (HRP) as described earlier (Chen et al. 2003) using Amplex Red Hydrogen Peroxide/Peroxidase assay kit (Life Technologies) according to the manufacturer's protocol. Briefly, C2C12 myotubes were incubated with 10 mm ammonium acetate for 6 h, then washed with Krebs–Ringer phosphate (KRP) buffer, and 1.5 × 10⁴ cells were used in each reaction; 50 μm of amplex red reagent and 0.1 U ml⁻¹ HRP in KRP buffer was added to the cells and incubated for 30 min at 37°C and fluorescence was measured (emission 590 nm; excitation 485 nm) using a Perkin Elmer VICTOR3 (Perkin Elmer, Waltham, MA, USA) multilabel reader. Standard curves were obtained by adding known amounts of H₂O₂ to the assay medium in the presence of amplex red and H₂O₂. Background fluorescence was measured in the absence of cells and data presented as background subtracted from the fluorescence of the cells (pmol per 600,000 cells per 30 min). All data were generated from three independent experiments.

NAD⁺/NADH ratio

Cellular NAD⁺, NADH and their ratio were measured using NAD⁺/NADH luciferase assay kit (Promega, Madison, WI, USA) using the manufacturer's instructions. Briefly, ∼1000 differentiated C2C12 myotubes per well in a 384‐well plate were incubated with 10 mm ammonium acetate for different time points with and without blockers of different components of the electron transport chain (rotenone for complex I, dimethyl malonate for complex II, antimycin A for complex III; sodium azide for complex IV). Cells were washed with PBS for NAD⁺ and NADH measurements. All data were generated from three independent experiments.

Thiobarbituric acid reactive substances

Cellular oxidative stress generates unstable peroxides from polyunsaturated fatty acids that form malondialdehyde (MDA), which forms an adduct with thiobarbituric acid (TBA). The MDA–TBA adducts formed under high temperature and acidic conditions were quantified by a fluorometric assay using a modification of a previously described protocol (Ohkawa et al. 1979). In brief, 50 μg of protein was added to 10 μl of 10% sodium dodecyl sulfate (SDS) and 100 μl of the colour reagent (0.53% thiobarbituric acid in acetic acid, 1× sodium hydroxide). The samples were boiled in a water bath for 30 min and then placed at room temperature for 10 min. Standard curves were generated using known concentrations of MDA. Fluorescence was measured with an excitation wavelength of 530 nm and an emission wavelength of 550 nm. The concentration of MDA–TBA adduct was calculated from the standard curve.

Carbonylated proteins

Oxidative modifications of proteins by free radicals results in post‐translational modifications including carbonylation. Carbonyl groups can be detected by immunoblot assays by derivatization of the carbonyl group to the 2,4‐dinitrophenyl (DNP) group by reaction with 2,4‐dinitrophenyl hydrazine (DNPH). Carbonylated protein can be detected by immunoblot assay using a primary antibody specific to the DNP moiety of the protein using modifications of a previously described protocol (Jacko et al. 2010). Protein (∼20 μg) from muscle tissue and cellular lysate was denatured by adding 5 μl of 12% SDS followed by 10 μl of 1× DNPH, and the reaction mixture was incubated for 15 min at room temperature. Samples were neutralized by adding 7.5 μl neutralization solution (2 m Tris pH 7.0, 30% glycerol) to each tube. Proteins were separated by gel electrophoresis followed by electrotransfer to polyvinylidene fluoride (PVDF) membrane and incubated for 3 h in 1:5000 anti‐DNP antibody (Bethyl Laboratories, Montgomery, TX, USA). Immunoreactivity was detected using methods previously described (Dasarathy et al. 2004).

ATP content

Cellular ATP content in human skeletal muscle, the PCA rat model of hyperammonaemia and hyperammonaemic C2C12 myotubes was quantified using a bioluminescence assay with a commercial kit (Molecular Probes, Eugene, OR, USA) following the manufacturer's recommended protocol.

Intact cell respiration using high resolution respirometry

C2C12 myotubes differentiated for 48 h as described above in 10 cm² plates were treated with 10 mm ammonium acetate for 24 h, trypsinized, centrifuged at 1000 g for 5 min, washed in Hanks’ balanced salt solution, resuspended (750,000 cells ml^–1) in mitochondrial respiration medium (0.5 mm EGTA, 3.0 mm MgCl₂·6H₂O, 60 mm potassium lactobionate, 20 mm taurine, 10 mm KH₂PO₄, 20 mm Hepes, 110 mm sucrose, 1 g l⁻¹ bovine serum albumin, pH 7.1) and transferred to an Oxygraph‐2 K high‐resolution respirometer (Oroboros Instruments, Innsbruck, Austria). Control and ammonia‐treated samples were assessed simultaneously as previously described (Ye & Hoppel, 2013). Intact cell respiration was measured in myotubes treated with and without ammonium acetate. ATP synthesis was quantified by inhibition of oxygen consumption by oligomycin (1 μg ml⁻¹) and continued oxygen consumption during oligomycin inhibition was a measure of proton leak (Rolfe & Brand, 1996). Maximum respiratory capacity (uncoupled respiration) was measured in response to carbonyl cyanide‐4‐(trifluoromethoxy) phenylhydrazone (FCCP), 0.5 μm increments. Antimycin A was used to determine residual oxygen consumption, a measure of non‐mitochondrial respiration. Oxygen consumption rates were calculated using accompanying software (DatLab2, Oroboros) as described earlier (Ye & Hoppel, 2013).

Immunoblot analysis

Cellular protein was extracted using methods previously described (Qiu et al. 2012). In brief, protein was extracted from 48 h differentiated C2C12 myotubes treated with 10 mm ammonium acetate for different time points using the RIPA buffer. After quantifying the concentration, protein samples were denatured and run on a 4–12% gradient gel. Following electrophoresis, the proteins were electrotransferred onto PVDF membranes (Bio‐Rad, Hercules, CA, USA) and incubated with 1:1000 MnSOD antibody (Cell Signaling Technologies, Danvers, MA, USA) and mitochondrial content was quantified by immunoblots for citrate synthase in the human skeletal muscle using the protocol described above with citrate synthase antibody (Proteintech Inc., Rosemont, IL, USA) at 1:2000 dilution and appropriate secondary antibody. Membranes were then washed in Tris‐buffered saline with Tween 20 (TBST) followed by incubation with appropriate secondary antibodies. Immunoreactivity was detected using a chemiluminescent HRP substrate (Millipore, Billerica, MA, USA) and densitometry performed using the Image J program.

Metabolic intermediates of the tricarboxylic acid cycle

Components of the tricarboxylic acid (TCA) cycle including citrate, αKG, succinate, fumarate and malate were quantified by gas chromatography/mass spectrometry (GCMS; 7890 System, Agilent Technologies, Santa Clara, CA, USA). The concentrations of TCA intermediates were assayed as previously described using gas GCMS with minor modifications (Kasumov et al. 2009). Briefly, cell lysates were extracted using RIPA buffer and ∼300 μg protein extracts were spiked with 12.5 nmol each of the following standards: [U‐¹³C₆]citrate, [1,2,3,4‐¹³C₄]αKG, [U‐¹³C₄]succinate, [U‐¹³C₄]fumarate and [U‐¹³C₄]malate. Samples were treated with 5 m hydroxylamine hydrochloride for 1 h at room temperature and pH 8.0 to convert keto acids to hydroxymates. Sample pH was adjusted to pH 2.0 and saturated with NaCl. The TCA cycle intermediates were extracted with ethylacetate from biological samples in parallel with calibration curve samples for each analyte. tert‐Butyldimethylsilyl derivatives of the intermediates were then generated using N‐tert‐butyldimethylsilyl‐N‐methyltrifluoroacetamide with 1% tert‐butyldimethylchlorosilane (Sigma‐Aldrich, St Louis, MO, USA). Under electron ionization, ions monitored included 459(M0) and 465(M+6) for citrate, 446(M0) and 450(M+4) for αKG, 289(M0) and 293(M+4) for succinate, 287(M0) and 291(M+4) for fumarate, and 419(M0) and 423(M+4) for malate. All experiments were performed on four independent samples and data expressed as the mean ± SEM.

Reagents

All reagents were obtained from Sigma‐Aldrich unless specifically stated. Labelled metabolic intermediates for the mass spectrometry assays were generously provided by Cambridge Isotope Laboratories Inc. (Tewksbury, MA, USA).

Data analysis

All assays were done at least in triplicate and values reported as mean ± SEM. Statistical analyses were conducted by two‐way analysis of variance with Bonferroni post hoc analysis for multiple group comparisons and Student's t test for two‐group analysis. Differences were considered statistically significant at a P value < 0.05.

Results

Hyperammonaemia decreases skeletal muscle ATP content

To determine if hyperammonaemia is responsible for lower ATP content in skeletal muscle, we used a comprehensive array of models including human patients with cirrhosis who have a high muscle concentration of ammonia (Table 1), the PCA rat and C2C12 myotubes as previously reported (Dasarathy et al. 2011; Qiu et al. 2012). In the present study, mean plasma ammonia concentration in PCA rats (301.8 ± 124.6 μmol l⁻¹) was significantly higher (P < 0.001) compared to the pair‐fed, sham‐operated control group (75.9 ± 27.2 μmol l⁻¹) (n = 5 in each group). Muscle ammonia concentrations in PCA rats (4.72 ± 0.59 mm (g protein)⁻¹) also were significantly higher (P < 0.001) than in the skeletal muscle of control, sham‐operated, pair‐fed rats (0.48 ± 0.1 mm (g protein)⁻¹).

In vivo magnetic resonance spectroscopy (MRS) showed significantly lower phosphocreatine and ATP to P_i ratio in the gastrocnemius muscle of the PCA compared to sham‐operated control rats (Fig. 2 A). Arterial ammonia concentrations were inversely correlated (R ² = 0.77; P < 0.001) with muscle ATP to P_i ratio (Fig. 2 B). Direct chemical quantification showed lower ATP content (P < 0.001) in the gastrocnemius muscle of PCA compared to sham operated control rats (Fig. 2 C). Consistently, patients with cirrhosis also had significantly lower skeletal muscle ATP content compared to controls (Fig. 2 D). The reduction in ATP content was not related to changes in mitochondrial mass in cirrhosis as observed by citrate synthase expression that was similar in cirrhotics and controls (Fig. 2 E). To determine if hyperammonaemia was responsible for the reduction in muscle ATP content, we used differentiated C2C12 myotubes exposed to 10 mm ammonium acetate that resulted in cellular concentrations of ammonia that are similar to those in the muscle of cirrhotic patients (Qiu et al. 2013). Studies in C2C12 myotubes showed a time‐dependent reduction in ATP during hyperammonaemia (Fig. 2 F). Interestingly, replacing the ammonia‐containing medium with fresh medium reversed the reduction in ATP content in C2C12 myotubes (Fig. 2 G). These data demonstrate that hyperammonaemia in skeletal muscle impairs ATP synthesis.

Figure 2. Skeletal muscle ATP content during hyperammonaemia.

A, in vivo ³¹P MR spectroscopy showed significantly lower phosphocreatine (PC) and ATP content normalized to total inorganic phosphate (P_i) in the gastrocnemius muscle of PCA compared to pair fed sham operated control rats (n = 5 animals each). B, arterial ammonia concentration was significantly inversely correlated with skeletal muscle ATP content in vivo in the portacaval anastomosis (PCA) and sham (sham operated, pair fed control) rats (n = 5 each, P < 0.01). C and D, ATP content in the gastrocnemius muscle from PCA compared to control sham‐operated rats and in human rectus abdominis muscle from patients with cirrhosis was significantly lower than that in controls. E, representative immunoblots and densitometry of citrate synthase expression in human cirrhosis and controls (n = 5 each) were similar (P > 0.05). F, in C2C12 myotubes exposed to hyperammonaemia, ATP content was significantly lower than in controls. G, removal of ammonia from the medium reversed hyperammonaemia mediated reduction in ATP content in myotubes. ^* P < 0.05; ^** P < 0.01; ^*** P < 0.001 compared to controls. ^a P < 0.001 compared to AmAc. AmAc, 10 mm ammonium acetate; CIR, cirrhosis; CTL, control.

Mitochondrial respiration

Since skeletal muscle hyperammonaemia results in low ATP, elevated ROS, defects in electron chain complexes and altered redox status of the cell, response of cellular respiration to hyperammonaemia was quantified by a Clark‐type O₂ electrode (Fig. 3 A and B). Hyperammonaemia decreased basal respiration in intact cells compared to controls (Fig. 3 C). Coupled respiration drives oxidative phosphorylation and ATP synthesis and the oxygen consumption is oligomycin sensitive (Rolfe & Brown, 1997). Oxygen consumption that is insensitive to oligomycin results from proton leak across the inner mitochondrial membrane that is not dependent on complex V–ATP synthase coupling (Rolfe & Brand, 1996). Interestingly, hyperammonaemia lowered both coupled (oligomycin sensitive) and proton leak‐dependent (FCCP) respiration (Fig. 3 D and E). The fraction of cellular respiration dedicated to coupled respiration was inhibited to a greater extent than that due to proton leak following hyperammonaemia (Fig. 3 F). These observations show that ammonia had a greater effect on ATP synthesis than on the proton leak into the mitochondrial matrix. Finally, FCCP, a protonophore that results in uncoupling of oxidation from phosphorylation, showed that ammonia decreased maximum respiratory capacity.

Figure 3. Decreased mitochondrial respiration in myotubes during hyperammonaemia.

A and B, representative tracings of high‐resolution respirometry to quantify intact cell respiration of differentiated C2C12 myotubes treated without (A, control) and with 10 mm ammonium acetate (B). After initial stabilization, ATP synthetase inhibitor, oligomycin, was added and oxygen consumption quantified to determine the oligomycin‐sensitive and ‐insensitive respiration. Protonophore (H⁺ ionophore) and uncoupler of oxidative phosphorylation, FCCP, was then added to quantify maximum respiratory capacity. This was followed by rotenone, which inhibited complex I of the electron transport chain, and then antimycin A, which inhibits complex III, was added to determine residual respiration. Coupled, uncoupled and maximum respiratory capacity (FCCP response) was inhibited by hyperammonaemia in myotubes. C, intact cell respiration. D, oligomycin‐sensitive oxygen consumption that reflects ATP synthesis. E, proton leak‐related respiration. F, ratio of oxygen consumption linked to ATP synthesis and proton leak in myotubes during hyperammonaemia. ^** P < 0.01. S^*Mill, Seconds. Million cells for oxygen flow rate. [Colour figure can be viewed at wileyonlinelibrary.com]

Increased muscle mitochondrial reactive oxygen species generated by ammonia

Cellular stress increases the generation of reactive oxygen species (ROS) from multiple sites (Holmstrom & Finkel, 2014), and we tested if hyperammonaemia causes an increase in ROS production in myotubes. Hyperammonaemia increased total ROS in C2C12 myotubes assessed by DCFDA fluorescence using both flow cytometry (Fig. 4 A–C) and a fluorometric assay (Fig. 4 D). Furthermore, use of MitoSOX, a mitochondrial specific fluorogenic dye, showed a persistent increase in mitochondrial ROS in C2C12 myotubes during hyperammonaemia. Consistently, the mitochondrially targeted antioxidant, MitoTEMPO, reversed the ammonia‐induced increase in ROS (Fig. 4 E–G) showing that hyperammonaemia results in increased mitochondrial ROS.

Figure 4. Increased reactive oxygen species (ROS) in C2C12 myotubes during hyperammonaemia.

A, flow cytometry gated fluorescence data in myotubes. Distribution of differentiated myotube DCFDA fluorescence with gating during hyperammonaemia. B and C, flow cytometry analysis (percentage positive cells, B; mean fluorescence intensity, C) with DCFDA as a fluorophore in C2C12 myotubes exposed to hyperammonaemia for different times showing increased ROS generation. D, time course of ROS generated in C2C12 myotubes exposed to 10 mm ammonium acetate using a fluorometric assay. Rapid increase in ROS quantified in response to ammonium acetate compared to controls. The baseline fluorescence of cells was quantified (without DCFDA). E, flow cytometry gated fluorescence data in myotubes. Distribution of differentiated myotube MitoSOX fluorescence with gating during hyperammonaemia. F, percentage of cells stained by MitoSOX in response to hyperammonaemia (mean ± SEM) showed significant increase in mitochondrial ROS with a decrease in response to MitoTEMPO. G, mean fluorescence intensity of cells stained with MitoSOX in response to hyperammonaemia (mean ± SEM). All experiments in triplicate. ^* P < 0.05; ^** P < 0.01; ^*** P < 0.001. AmAc, 10 mm ammonium acetate; DCFDA, 2′,7′‐dichlorodihydrofluorescein diacetate; MT, MitoTEMPO. [Colour figure can be viewed at wileyonlinelibrary.com]

Oxidative damage of skeletal muscle proteins and lipids during hyperammonaemia

Reactive oxygen species (ROS) induce post‐translational protein modifications and lipid peroxidation (Holmstrom & Finkel, 2014). Since hyperammonaemia induces mitochondrial ROS, oxidative damage to proteins and lipids were assessed by protein carbonylation (Barreiro & Hussain, 2010) and thiobarbituric acid reactive substances (TBARS) (Janero, 1990), respectively. Skeletal muscle from cirrhotic patients and PCA rats showed significant increase in carbonylated proteins (Fig. 5 A and B) and TBARS (Fig. 5 D and E) showing that hyperammonaemia results in oxidative modification of muscle proteins and lipids. Consistently, expression of carbonylated proteins and TBARS was significantly higher in C2C12 myotubes exposed to hyperammonaemia (Fig. 5 C and F). Since hyperammonaemia increased mitochondrial ROS and modification of both lipids and proteins during hyperammonaemia, we assessed if the increase in ROS was due to compromised critical mitochondrial matrix enzyme manganese‐dependent superoxide dismutase (MnSOD) (Higuchi et al. 1985). Interestingly, our studies showed a significant increase in skeletal muscle MnSOD expression in all three models of hyperammonaemia (Fig. 5 G–J).

Figure 5. Oxidative modification of skeletal muscle protein and lipids during hyperammonaemia.

A–C, increased expression of skeletal muscle protein carbonylation in human cirrhosis, PCA rat and myotubes exposed to hyperammonaemia. D–F, significantly higher TBARS in the skeletal muscle from cirrhotic patients, hyperammonaemic PCA rats and C2C12 myotubes during hyperammonaemia compared to respective controls. G, representative photomicrographs showing immunofluorescence of MnSOD expression in C2C12 myotubes, showing increased expression during hyperammonaemia (10 mm ammonium acetate) compared with untreated myotubes. H–J, representative immunoblots and densitometry of MnSOD expression in the skeletal muscle from patients with cirrhosis and controls, hyperammonaemic PCA rat and sham‐operated controls and C2C12 myotubes treated for different time points with 10 mm ammonium acetate, showing increased expression during hyperammonaemia. ^* P < 0.05; ^** P < 0.01; ^*** P < 0.001. AmAc, 10 mm ammonium acetate; CIR, human cirrhotic subject; CTL, control human subject; MnSOD, manganese superoxide dismutase; PCA, portacaval anastomosis rat; Sham, sham‐operated, pair‐fed control; TBARS, thiobarbituric acid reactive substances. [Colour figure can be viewed at wileyonlinelibrary.com]

Complex III is the major source of superoxide generated during hyperammonaemia

Classically, mitochondrial complexes III and I have been reported to be major sites for ROS production, which was quantified by measuring hydrogen peroxide (H₂O₂) production (Turrens & Boveris, 1980; Chen et al. 2003). To determine the specific site of electron leak and ROS generation during hyperammonaemia, single and multiple ETC complexes were inhibited and H₂O₂ generated was measured. We observed increased H₂O₂ production during hyperammonaemia (Fig. 6 A). Blocking complex I (NADH oxidase) with rotenone resulted in a significant reduction in electron leak and H₂O₂ in ammonia‐treated cells. In contrast, dimethyl malonate, a cell‐permeable inhibitor of complex II, sodium azide, an inhibitor of complex IV, and myxothiazol, a competitive inhibitor of ubiquinol that binds at the quinol oxidation site of cytochrome c reductase, only partially inhibited the electron leak and H₂O₂ generation. In contrast, blocking complex III with antimycin A caused a significant increase in electron leakage and H₂O₂ production in ammonia‐treated C2C12 myotubes compared to controls. These data show that ammonia increases electron leakage primarily at complex III and are consistent with previous reports of complex III being a major site of electron leak (Chen et al. 2003). To specifically dissect the site of hyperammonaemia‐mediated electron leak, multiple ETC complexes were blocked simultaneously (Fig. 6 B). Importantly, use of the complex I blocker rotenone resulted in a reduction of electron leak independent of other complexes inhibited during hyperammonaemia. This was most evident in the ammonia‐treated cells that were exposed simultaneously to rotenone (complex I inhibitor) and antimycin A (complex III inhibitor). Interestingly, blocking complex II (dimethyl malonate), the second source of electron input into complex III, did not decrease electron leakage by blocking complex III (antimycin A) during hyperammonaemia (dimethyl malonate+antimycin A). These data show that the principal source of electron leak during hyperammonaemia occurs at complex III and the primary influx of electrons into the ETC is at complex I.

Figure 6. Impaired electron flow through complex I and electron leak from complex III results in lower NAD⁺/NADH ratio and increased ROS during hyperammonaemia.

A, increased generation of hydrogen peroxide measured using the amplex red fluorescence assay in myotubes during hyperammonaemia. Blocking complex III had the greatest effect on hydrogen peroxide production. B, to determine the specific source of the superoxide generated in the ETC in the mitochondria, H₂O₂ generated in the presence of blockers of a combination of complexes of the ETC showed that the principal source of electron leak occurs at complex III. The electron flow was mainly through complex I, because inhibition of either complex II or III increased the electron leak whereas inhibition with complex I reduces the leak. C, ratio of NAD⁺/NADH was significantly lower in myotubes during hyperammonaemia. D, ratio of NAD⁺/NADH during hyperammonaemia with and without inhibitors of different complexes in the ETC. ^* P < 0.05; ^** P < 0.01; ^*** P < 0.001 compared to respective control. ^a P < 0.05, ^b P < 0.01, and ^c P < 0.001 compared to 6 h AmAc. αKG, α‐ketoglutarate; AMA, 10 ng ml⁻¹ antimycin A; AmAc, 10 mm ammonium acetate; CIR, cirrhosis; CTL, control; DM, dimethyl α‐ketoglutarate; DMM, 10 mm dimethyl malonate; ETC, electron transport chain; MT, MitoTEMPO; Myxo, 1 μm myxothiazole; NaAZ, 3 mm sodium azide; N.S., not significant; Rot, 10 μm rotenone.

Impaired ETC activity alters the redox state of the matrix via changes in the NAD⁺/NADH ratio. As NADH oxidation is required to regenerate NAD⁺, impaired electron transport during hyperammonaemia is likely to lower the NAD⁺/NADH redox ratio. Consistently, there was a significant reduction in NAD⁺/NADH ratio after 6 h hyperammonaemia (Fig. 6 C). Correspondingly, blocking complexes I and III impaired oxidation of NADH and subsequent regeneration of NAD⁺ leading to a further reduction in NAD⁺/NADH ratio (Fig. 6 D). Consistent with the flow of electrons primarily down complex I, rotenone (which blocks complex I) caused the greatest reduction of NADH oxidation. Blocking complex II with dimethyl malonate resulted in a reduction in the NAD⁺/NADH ratio, but not to the extent observed with rotenone. Finally, blocking complex III with antimycin A resulted in the greatest reduction in the NAD⁺/NADH ratio because electrons from both complex I and II pass though this complex.

TCA cycle intermediates

The primary mechanism of ammonia disposal by the skeletal muscle is via the conversion of glutamate to glutamine (Zaganas et al. 2013). During hyperammonaemia, cataplerosis of α‐ketoglutarate (αKG) contributes to glutamate–glutamine‐mediated ammonia disposal with consequent increase in glutamine release (Holecek, 2014). In conditions when anaplerotic reactions cannot compensate for cataplerosis of αKG, concentrations of TCA cycle intermediates will decrease, potentially explaining the impaired oxygen consumption during hyperammonaemia. Additionally, the change in the NAD⁺/NADH redox ratio is also an important regulator of flux through the TCA cycle (Stanley & Connett, 1991) that can alter the concentrations of the intermediates. Consistently, concentrations of TCA cycle intermediates in C2C12 myotubes were significantly lower during hyperammonaemia (Fig. 7). Interestingly, treatment of C1C12 myotubes with dimethyl αKG, a cell‐permeable ester form of αKG, reversed the hyperammonaemia‐mediated reduction in TCA cycle intermediates (Fig. 8 A) and ATP content (Fig. 8 B). In contrast, MitoTEMPO did not increase ATP content during hyperammonaemia (Fig. 8 B), suggesting that increased ROS was not responsible for the ammonia‐induced reduction in ATP.

Figure 7. Reduced tricarboxylic acid (TCA) cycle intermediates during hyperammonaemia.

All the TCA cycle intermediates were significantly lower in myotubes during hyperammonaemia. ^* P < 0.05; ^** P < 0.01; ^*** P < 0.001 compared to respective control. ^a P < 0.05, ^b P < 0.01 and ^c P < 0.001 compared to 6 h AmAc. αKG, α‐ketoglutarate; AmAc, 10 mm ammonium acetate; CTL, control.

Figure 8. Rescue of cataplerosis during hyperammonaemia by cell‐permeable ester of α‐ketoglutarate.

A, supplementation with 1 mm cell‐permeable dimethyl αKG in control and ammonia‐treated C2C12 myotubes for 24 h showed reversal of ammonia‐induced reduction in TCA cycle intermediates. ^*** P < 0.001 compared to untreated control myotubes. B, use of 1 mm DM αKG (cell‐permeable ester form) for 24 h reversed ammonia‐induced reduction in ATP in myotubes. In contrast, 20 nm of MitoTEMPO, a mitochondrial ROS scavenger, did not reverse the ammonia‐induced reduction in ATP. AmAc, 10 mm ammonium acetate; CTL, control; DM αKG, dimethyl α‐ketoglutarate; N.S., not significant.

Discussion

Our studies in a comprehensive array of models show skeletal muscle ammonia disposal activates a sequence of metabolic and signalling responses that culminate in disordered mitochondrial function. In three distinct models (PCA rat, human cirrhosis and C2C12 myotubes), we show decreased muscle ATP content with increased mitochondrial ROS generation during hyperammonaemia. Mechanistic studies demonstrated that electron leak and ROS were generated primarily at complex III. These were accompanied by a lower ATP synthesis and a reduction in TCA cycle intermediates that were reversed by supplementing with a cell‐permeable ester of αKG.

Lower cellular and tissue ATP content during hyperammonaemia were due to decreased synthesis as quantified by coupled respiration (Rolfe & Brand, 1996; Rolfe & Brown, 1997). Our data are consistent with previous reports of lower skeletal muscle ATP in cirrhosis (Moller et al. 1984; Jacobsen et al. 2001) and in astrocytes during hyperammonaemia (Haussinger et al. 1992), but the mechanisms were not identified. We provide a mechanistic basis for impaired skeletal muscle ATP synthesis during hyperammonaemia via impaired NADH oxidation (complex I) and decreased oxygen consumption due to impaired coupled respiration. The reduction in oligomycin‐sensitive (coupled) respiration shows that ammonia inhibits ATP synthesis. Our observation that the fraction of cellular respiration dedicated to coupled respiration was inhibited to a greater extent than that due to proton leak showed that ammonia had a greater effect on ATP synthesis than on the proton leak into the mitochondrial matrix. Additionally, our systematic studies also provide evidence that increased oxidative damage of muscle protein and lipids was due to increased mitochondrial ROS due to electron leak at complex III, which was consistent with prior reports of the primary site of ROS generation in the ETC (Chen et al. 2003). Lower maximum respiratory capacity during hyperammonaemia also suggests that if there is a greater demand for oxidation of substrates or any additional mitochondrial stress, this will further impair ATP synthesis. Reversal of low ATP content in myotubes following replacement with fresh medium suggests that the lower ATP synthesis is reversible, at least in the short term.

Consistent with our observations on impaired complex I/NADH oxidation on respirometry, the NAD⁺/NADH redox ratio was significantly lower during hyperammonaemia. An alteration in redox ratio impairs the TCA cycle flux (Stanley & Connett, 1991). Another potential mechanism of lower TCA cycle intermediates during hyperammonaemia is due to the cataplerosis of αKG to metabolize ammonia to glutamate and then glutamine (Lai & Cooper, 1986; Zaganas et al. 2013; Holecek, 2014). In keeping with this interpretation, lower αKG and other TCA cycle intermediates were noted in the myotubes during hyperammonaemia. Importantly, cell‐permeable dimethyl α‐ketoglutarate (DMAKG) reversed the lower intermediate concentrations and ATP content indicating that anaplerosis from extracellular αKG compensated for the ammonia‐induced cataplerosis of the TCA cycle. These results are in contrast to previous reports suggesting that DMAKG impairs mechanistic target of rapamycin complex I (mTORC1) signalling and ATP synthesis in C. elegans (Chin et al. 2014). However, in skeletal myotubes DMAKG increased cellular ATP content during hyperammonaemia because during ammonia‐induced cellular stress, there is increased cataplerosis. Our novel data show that DMAKG supplementation during hyperammonaemia‐induced cellular stress with cataplerosis replenished the low αKG and reversed low ATP content in myotubes.

Our studies with DCFDA showed that total ROS was increased during hyperammonaemia, which is consistent with that reported in astrocytes (Murthy et al. 2001; Rama Rao et al. 2005). However, the site of ROS generation during hyperammonaemia has not been identified. Our studies using a mitochondrial ROS fluoroprobe, MitoSox, complemented by a mitochondrial ROS quencher, MitoTempo, demonstrate that mitochondria are the source of the ROS during hyperammonaemia. Mitochondrial ROS have physiological functions (Hamanaka & Chandel, 2010), but injury and oxidative stress are prevented by a series of reactions including the mitochondrial manganese‐dependent superoxide dismutase activity that quenches the ROS (Sugioka et al. 1988). Consistently, we demonstrated an increased skeletal muscle MnSOD expression. Our observations are consistent with reports that increased skeletal muscle mitochondrial ROS, following exercise (Higuchi et al. 1985) and ammonia exposure in fish (Hegazi et al. 2010) induced MnSOD activity. Our observations are, however, in contrast to those reported in the brain and astrocytes where ammonia impairs MnSOD activity (Kosenko et al. 1997; Singh et al. 2008). These observations suggest that modulation of MnSOD by ammonia is context specific. Ammonia‐induced MnSOD may be mediated via transcription factors like p65NFκB, which is activated during hyperammonaemia (Sompol et al. 2006; Qiu et al. 2013) in addition to being an adaptive response to the mitochondrial ROS generated (Higuchi et al. 1985; Li et al. 2010). Despite the increased MnSOD, ammonia‐induced mitochondrial ROS overwhelmed the adaptive response, as noted by our data that show oxidative modifications of proteins and lipids. Increased protein carbonylation in cirrhotic muscle compared to control subjects was less than that observed in our rodent and cellular models. This may be due to either a longer duration of exposure, with resultant effective adaptive responses including autophagy, or activation of antioxidant defence mechanisms. The adaptive mechanisms that are activated by the electron leak and generation of superoxide are an increase in mitochondrial superoxide dismutase and autophagy to clear the oxidatively modified proteins (Higuchi et al. 1985; Warner et al. 1996; Hegazi et al. 2010; Scherz‐Shouval & Elazar, 2011). The present studies, complemented by our previous report of activation of autophagy by ammonia (Eng et al. 2010; Qiu et al. 2012), showed that that the cellular responses are unable to cope with the oxidative modification of cellular components that accumulate during hyperammonaemia. These observations suggest that hyperammonaemia increases ROS generation that overwhelms the adaptive ROS antioxidant mechanism with resultant tissue injury and muscle loss.

During mitochondrial respiration, oxygen is reduced by accepting electrons in a series of steps and converted to water. The partially reduced oxygen intermediates are highly reactive, but cytochrome c retains these intermediates before complete reduction and water is formed. However, a small proportion of oxygen is converted to superoxide ions (O₂ ⁻) at complexes I and III (Cadenas et al. 1977; Turrens & Boveris, 1980; Turrens, 1997). Hyperammonaemia increased hydrogen peroxide production showing an increased ROS. Our studies with blockers of electron transport chain complexes rotenone (complex I), antimycin A (complex III) and a combination of blockers of complexes I and II show that the major site of ROS generation is in complex III of the electron transport chain, consistent with a previous report (Chen et al. 2003). Limiting electron flow through complex III prevents mitochondrial ROS formation during hyperammonaemia.

Our observation of reduced TCA cycle intermediates is consistent with previous reports of reduced plasma αKG in hyperammonaemia (Batshaw et al. 1980). Even though there are strong theoretical considerations for reduced TCA cycle intermediates due to cataplerosis in hyperammonaemia (Holecek, 2014), this is the first report of reduced intermediates in the myotubes during hyperammonaemia. Lower TCA cycle intermediates, specifically αKG, can lower oxidative phosphorylation and ATP synthesis and only small increases are required to meet cellular energy needs (Sahlin et al. 1990; Owen et al. 2002). In our studies, replenishing αKG by a cell‐permeable ester reversed low TCA cycle intermediates and ATP content in myotubes during hyperammonaemia. This observation supports our interpretation that hyperammonaemia is a cataplerotic state and anaplerotic supply of specific TCA intermediates is a potential therapeutic strategy. Other strategies to provide anaplerotic substrates include branched chain amino acids, specifically isoleucine and valine via propionyl CoA while leucine provides acetyl CoA (Wagenmakers, 1998), and potentially other amino acids and glucose (Owen et al. 2002). When amino acids are used for anaplerosis to replenish TCA cycle intermediates, there is net generation of ammonia that adds to the cellular ammonia pool during hyperammonaemia (Holecek, 2014). Even though direct cataplerotic flux of individual TCA cycle intermediates was not quantified in the present studies, αKG was targeted because it is likely to be depleted during hyperammonaemia (Batshaw et al. 1980; Holecek, 2014), and continued cataplerosis of αKG results in lower concentrations of all other intermediates presumably due to inability of physiological anaplerosis to replenish the deficiency. A potential clinical application of this approach will be to use anaplerotic substrates as a novel ammonia lowering strategy by increased ammonia disposal by non‐ureagenic, extrahepatic metabolism similar to ornithine phenylacetate (Ytrebo et al. 2009). Caution must be exerted in such an approach because critical intermediates are not cell permeable and only small changes in TCA cycle intermediates alter energy generation (Sahlin et al. 1990; Stanley & Connett, 1991; Wagenmakers, 1998). Furthermore, αKG supplementation in C. elegans inhibits mTORC1, aggravating the impaired muscle protein synthesis during hyperammonaemia, and increasing TCA cycle flux by providing αKG can potentially increase muscle ROS generation via αKG dehydrogenase (Ambrus et al. 2015).

Mitochondrial mass is affected by physical activity, and patients with cirrhosis have been reported to have lower activity (Higuchi et al. 1985; Figueiredo et al. 2009; Dunn et al. 2016). Therefore, mitochondrial content could have affected our measurements of function including ATP content, ROS generation and MnSOD expression in the skeletal muscle of patients with cirrhosis. Even though routine physical activity was not documented or compared, citrate synthase expression as a measure of mitochondrial mass was similar in cirrhotics and controls suggesting that changes in mitochondrial mass were not responsible for the perturbations observed in the cirrhotic skeletal muscle.

We provide a mechanistic basis for hyperammonaemia‐induced skeletal muscle mitochondrial functional alterations (Fig. 9). We propose that hyperammonaemia‐induced depletion of TCA cycle intermediates via conversion to glutamate and glutamine results in impaired mitochondrial respiration and generation of ROS that causes oxidative modifications of proteins and lipids. Since hyperammonaemia‐induced mitochondrial dysfunction is associated with depletion of TCA cycle intermediates and lower ATP content that are reversed by DMAKG, a cell‐permeable precursor of αKG, ammonia‐induced cataplerosis contributes to mitochondrial dysfunction. These observations provide evidence to support novel approaches to reverse ammonia‐induced mitochondrial dysfunction using anaplerotic substrates including cell‐permeable analogues of αKG. Our studies have broad relevance since hyperammonaemia occurs in multiple chronic diseases with skeletal muscle loss and impaired muscle contractile function.

Figure 9. Schematic representation of mitochondrial dysfunction during hyperammonaemia.

In the skeletal muscle, ammonia enters mitochondria and is converted to glutamate by cataplerosis of αKG and consequent loss of TCA cycle intermediates. Ammonia also causes impaired function of complex I of the electron transport chain with reduced NADH oxidation, and leak of electrons at complex III with oxidative modifications of proteins (carbonylation) and lipid (lipid peroxidation). [Colour figure can be viewed at wileyonlinelibrary.com]

Additional information

Competing interests

The authors have no conflict of interest.

Author contributions

G.D.: conception and design of the work, data acquisition and analysis, interpretation of data, drafting and critically revising the manuscript for important intellectual content. A.A.: design of the work, acquisition, analysis and interpretation of the data and drafting and reviewing the final manuscript. S.T.: conceptualization and design of the work, acquisition and interpretation of the data, and drafting the manuscript. J.R.: design of the work, acquisition, analysis and interpretation of the data, drafting and reviewing the manuscript. D.S.: acquisition, analysis and interpretation of the data, drafting the manuscript, revising the manuscript and critical review of the final submission. A.K.: design of the work, data acquisition, analysis and interpretation, drafting and reviewing the manuscript. Y.S.: design of the work, data acquisition, analysis and interpretation, manuscript drafting and revisions. DVW: design of the work, data analysis, interpretation, drafting the manuscript and revisions. C.F.: design, data acquisition, analysis, interpretation, drafting the manuscript, revision of the manuscript. C.H.: conceptualizing the work, design and troubleshooting experiments, data analysis and interpretation, manuscript drafting and revisions. T.K.: conception and design of the experiments, data acquisition and analysis, interpretation of data, manuscript preparation, and revisions. S.D.: conceptualization of the work, design of experiments, data acquisition, analysis and interpretation of data, drafting the manuscript, and revising the same. Animal studies, human muscle analyses and in vitro cell culture molecular studies were performed in the Department of Pathobiology, Cleveland Clinic. The Cleveland Clinic Imaging Core was used to perform confocal microscopy. The mitochondrial respirometry was performed in the Department of Molecular Cardiology, Cleveland Clinic. Additional respirometry was performed in the Department of Pharmacology, Case Western Reserve University. The in vivo MR spectroscopy was performed in the Case Imaging Center in the Department of Biomedical Engineering, Case Western Reserve University. The quantification of the intermediates was carried out in Department of Chemistry, Cleveland State University and Department of Pharmaceutical Sciences, Northeast Ohio Medical University. All authors approved the final version of the manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported in part by NIH RO1 DK 83414, R21 AA 022742, P50 AA02433‐01‐8236 and Cleveland Clinic Institutional Support Funds to S.D.

Translational perspective

Ammonia is a toxic metabolite generated during endogenous amino acid and nucleotide metabolism as well as from gut bacteria. It is physiologically converted to urea in the liver and excreted by the kidneys. Impaired ammonia disposal and hyperammonaemia have been reported in advanced liver, heart and lung diseases. During hyperammonaemia, the skeletal muscle becomes a major source of ammonia uptake and metabolic disposal via mitochondrial glutamate synthesis followed by cytoplasmic glutamine formation and extracellular transport. Cataplerosis of the critical tricarboxylic acid (TCA) cycle intermediate α‐ketoglutarate (2‐oxoglutarate) has been suggested to provide the carbon skeleton for mitochondrial glutamate synthesis during hyperammonaemia. Our studies in in vitro cell culture of myotubes, and skeletal muscle from the hyperammonaemic portacaval anastomosis rat and human subjects with cirrhosis and controls show mitochondrial functional abnormalities including reduced cellular respiration, impaired electron transport chain function of complex I, lower muscle ATP content and increased reactive oxygen species. Expectedly, TCA cycle intermediates were lower during hyperammonaemia and were reversed by supplementation with a cell‐permeable ester of α‐ketoglutarate. We showed that hyperammonaemia results in disordered cellular energy metabolism. The present studies are of high translational significance because they identify novel therapeutic targets to reverse mitochondrial dysfunction during hyperammonaemia including the use of ammonia‐lowering strategies, supplementary TCA cycle intermediates and reducing mitochondrial reactive oxygen species. These approaches have the potential to prevent or reverse skeletal muscle loss, impaired muscle strength and increased fatigue in patients with cirrhosis, heart failure and lung diseases.